Superplasticizers and Other Chemical Admixtures in Concrete

Superplasticizers and
Other Chemical Admixtures

in Concrete
Proceedings
Eleventh International Conference
Ottawa, ON, Canada
July 2015
V. Mohan Malhotra
Pawan R. Gupta
Terence C. Holland
SP-302
Eleventh International Conference

on Superplasticizers and Other
Chemical Admixtures in Concrete
Editors:
V. Mohan Malhotra
Pawan R. Gupta
Terence C. Holland
SP-302
First printing, June 2015

Discussion is welcomed for all materials published in this issue and will appear ten months from this journals date
if the discussion is received within four months of the papers print publication. Discussion of material received
after specified dates will be considered individually for publication or private response. ACI Standards published
in ACI Journals for public comment have discussion due dates printed with the Standard.
The Institute is not responsible for the statements or opinions expressed in its publications. Institute publications
are not able to, nor intended to, supplant individual training, responsibility, or judgment of the user, or the supplier,
of the information presented.
The papers in this volume have been reviewed under Institute publication procedures by individuals expert in
the subject areas of the papers.
Copyright 2015
AMERICAN CONCRETE INSTITUTE
38800 Country Club Dr.
Farmington Hills, Michigan 48331
All rights reserved, including rights of reproduction and use in any form or by any means, including the making of
copies by any photo process, or by any electronic or mechanical device, printed or written or oral, or recording
for sound or visual reproduction or for use in any knowledge or retrieval system or device, unless permission in
writing is obtained from the copyright proprietors.
Printed in the United States of America

Editorial production:
Carl R. Bischof
Tiesha Elam
Kaitlyn J. Hinman
Ryan M. Jay
Kelli R. Slayden
ISBN-13: 978-1-942727-22-4
Preface
In May 1978, the Canada Center for Mineral and Energy Technology (CANMET), in
association with the American Concrete Institute (ACI) sponsored a three-day conference
in Ottawa, ON, Canada, on the use of superplasticizers in concrete. Selected papers from
the symposium were published as ACI SP-62.
In 1981, CANMET, again in association with the ACI, sponsored a second three-day
international conference in Ottawa on the use of superplasticizers in concrete. Proceedings
of the conference were published as ACI SP-68.
The purpose of the third international conference in Ottawa in 1989 was to review the
progress made since the meetings in 1978 and 1981, and to bring together representatives
of the chemical admixtures, cement, and concrete industries to exchange information and
delineate new areas of needed research. The scope of this conference was expanded to
include chemical admixtures other than superplasticizers. Proceedings of the conference
were published as ACI SP-119.
In October 1994, CANMET, in association with the ACI and several other
organizations, sponsored the fourth conference in Montreal, QC, Canada. The objective of
this conference was to bring attention to new developments in chemical admixtures since
the last conference in 1998. Proceedings of the conference were published as ACI SP-148.
In October 1997, the Committee for the Organization of CANMET/ACI International
Conferences (ACI Council), in association with the ACI and several cement and concrete
organizations in Italy, sponsored the fifth conference in Rome, Italy. The conference
was aimed at transferring technology in the fast-moving field of chemical admixtures.
Proceedings of the conference were published as ACI SP-173.
In October 2000, the Committee of the Organization of CANMET/ACI International
Conferences (ACI Council), in association with several organizations in Canada and France,
sponsored the sixth conference in Nice, France. More than 50 papers from more than 20
countries were received and reviewed by the ACI review panel, and 37 were accepted for
publication in the proceedings of the conference. The proceedings were published as ACI
SP-195.
Conferences (ACI Council), in association with several organizations in Canada and
Germany, sponsored the seventh conference in Berlin, Germany. The conference attracted
more than 275 delegates and proceedings of the conference, consisting of 39 papers, were
published as ACI SP-217.
Conferences (ACI) Council), sponsored the eight conference in Sorrento, Italy. More than
60 papers from more than 25 countries were received and peer reviewed by the CANMET/
ACI review panel in Budapest, and 36 were accepted for publication as ACI SP-239.
In October 2009, the Committee for the Organization of International Conferences
(COIC) (formerly CANMET/ACI International Conferences), sponsored the ninth ACI
International Conference in Seville, Spain. More than 50 papers from more than 20
countries were received and peer reviewed, and 35 were accepted for publication in the
proceedings of the conference. The proceedings were published as ACI SP-262.
In October 28 to 31, 2012, COIC, sponsored the Tenth International Conference on
Superplasticizers and Other Chemical Admixtures in Concrete in Prague, Czech Republic.
More than 70 papers from the world over were peer reviewed, and 33 were accepted for
publication in the proceedings of the conference. The proceedings were published as ACI
SP-288.
In July 10-13, 2015, the COIC, in association with ACI, sponsored the Eleventh
International Conference on Superplasticizers and Other Chemical Admixtures in Concrete
in Ottawa, Canada. More than 60 papers from the world over were peer reviewed, and 28
were accepted for publication in the proceedings of the conference. The proceedings were
published by ACI as SP-302. Also, additional papers were presented at the conference that
were published in the Supplementary Papers Volume.
Thanks are extended to members of the Technical Paper Review Panel that met in
the Bahamas from October 3 to 11, 2014. Without their dedicated effort and hard work,
it would not have been possible to publish the proceedings. Cooperation of the authors
in accepting the reviewers suggestions and in revising the manuscripts accordingly is
appreciated greatly.
The help and assistance of Dr. Pawan R. Gupta and Prabha Gupta are acknowledged
gratefully in the administrative work associated with the conference and processing of the
manuscripts for both the conference proceedings (ACI SP-302) and the Supplementary
Papers Volume.
V. Mohan Malhotra
Pawan R. Gupta
Terence C. Holland
Editors
Eleventh International Conference on Superplasticizers and Other Chemical Admixtures
July 12 to 15, 2015
Ottawa, Canada
Contents
Preface 3
SP-302-01 11
Effect of Clinker Grinding Aids on
Static Yield Stress of Cement Pastes
by Joseph J. Assaad and Salim E.
Asseily
INTRODUCTION
11
RESEARCH SIGNIFICANCE
12
EXPERIMENTAL INVESTIGATION 12
TEST RESULTS AND DISCUSSION 15
CONCLUSIONS
21
SP-302-02 25
A 13C NMR Spectroscopic Study
on the Reparation of Acid and
Ester Groups in MPEG Type
PCEs Prepared via Radical
Copolymerization and Grafting
Techniques
by Julia Pickelmann, Huiqun Li, Robert
Baumann, and Johann Plank
INTRODUCTION
25
26
EXPERIMENTAL PROCEDURES
27
RESULTS AND DISCUSSION
29
CONCLUSION
34
SP-302-03 39
Effects of Particle Volume Fraction
and Size on Polysaccharide
Stabilizing Agents
by Wolfram Schmidt, Sarah Peters,
and Hans-Carsten Khne
INTRODUCTION
39
41
EXPERIMENTAL
RESULTS
AND
DISCUSSION
44
SUMMARY AND CONCLUSIONS 50
SP-302-04 53
New Additive to Enhance the Slump
Retention
by David Platel, Jean-Marc Suau,
Clement Chosson, and Yves Matter
INTRODUCTION
53
54
EXPERIMENTAL AND ANALYTICAL
INVESTIGATIONS 54
EXPERIMENTAL
RESULTS
AND
DISCUSSION
56
FURTHER RESEARCH
59
CONCLUSIONS
60
SP-302-05 63
Synthesis of a Novel
Superplasticizer Prepared from
Brown Coal
by Manuel Ilg and Johann Plank
INTRODUCTION
63
64
EXPERIMENTAL PROCEDURE
64
EXPERIMENTAL
RESULTS
AND
DISCUSSION
68
CONCLUSIONS
74
SP-302-06 77
Influence of Superplasticizers on
the Flocculation Degree of Cement
Suspensions
by Lucia Ferrari and Pascal
Boustingorry
INTRODUCTION
77
78
78
ANALYTICAL INVESTIGATION
80
COMPARISON OF PREDICTIONS AND
EXPERIMENTAL RESULTS
81
EXPERIMENTAL
RESULTS
AND
DISCUSSION
81
FURTHER RESEARCH
87
CONCLUSIONS
88
SP-302-07 93
Influence of Temperature and
Retarder on Superplasticizer
Performance
by Karen Luke and Adrian Torres
INTRODUCTION
93
94
95
EXPERIMENTAL
RESULTS
AND
DISCUSSION
97
SP-302-08 113
Properties of a New Type of
Polycarboxylate Admixture for
Concrete Using High Volume Blast
Furnace Slag Cement
by Shinji Tamaki, Kazuhide Saito,
Kazuhisa Okada, Daiki Atarashi, and
Etsuo Sakai
INTRODUCTION
113
114
114
EXPERIMENTAL
RESULTS
AND
DISCUSSION
116
CONCLUSIONS
123
SP-302-09 125
Synthesis and Properties of High
Solid-Content Polycarboxylate
Superplasticizer
by Yongwei Wang, Liya Wang,
Yongsheng Liu, and Zepeng Chu
INTRODUCTION
125
126
EXPERIMENTAL
RESULTS
AND
DISCUSSION
128
CONCLUSIONS
131
SP-302-10 133
Mastering Flow Loss in
Superplasticized Cementitious
Systems
by S. Mantellato, Q. Mehmeti, L. Ceni,
M. Palacios, and R.J. Flatt
INTRODUCTION
133
134
RESULTS
DISCUSSION
CONCLUSIONS
134
136
140
142
SP-302-11 145
Putting Concrete to Sleep and
Waking It Up with Chemical
Admixtures
by L. Reiter, M. Palacios, T. Wangler,
and R.J. Flatt
INTRODUCTION
145
146
147
DISCUSSION
151
CONCLUSIONS
152
SP-302-12 155
A Simplified Preparation Method for
PCEs Involving Macroradicals
by Lei Lei and Johann Plank
INTRODUCTION
155
158
EXPERIMENTAL
RESULTS
AND
DISCUSSION
162
CONCLUSIONS
165
SP-302-13 169
Impact of Slags Contained in
Blended Cement on Dispersing
Effectiveness of PCEs
by Ahmad Habbaba and Johann Plank
INTRODUCTION
169
170
EXPERIMENTAL
RESULTS
AND
DISCUSSION
173
CONCLUSIONS
178
SP-302-14 183
Preparation and Characterization
of Star-Shaped Polycarboxylate
Superplasticizer
by Xiao Liu, Ziming Wang, Jie Zhu,
Ming Zhao, Wei Liu, and Dongjie Yin
INTRODUCTION
183
184
EXPERIMENTAL
RESULTS
AND
DISCUSSION
187
CONCLUSIONS
195
SP-302-15 199
Influence of Diester Content in
Macromonomers on Performance of
MPEG-Based PCEs
by Johannes Paas, Maike W. Mller,
and Johann Plank
INTRODUCTION
199
202
EXPERIMENTAL
RESULTS
AND
DISCUSSION
203
CONCLUSIONS
209
SP-302-16 211
Evidences about the Interactions
between Grinding Aids and Cement
Particles Surface
by Valerio Antonio Patern, Sara
Ottoboni, Marco Goisis, and Paolo
Gronchi
INTRODUCTION
211
RESEARCH SIGNIFANCE
212
MATERIALS AND METHODS
212
DISCUSSION
217
FURTHER RESEARCH
223
CONCLUSION
223
SP-302-17 227
The Influence of C3A on the
Dissolution Kinetics of Alite/
Gypsum Mixtures in the Presence of
PCEs
by Giorgio Ferrari, Vincenzo Russo,
Massimo Dragoni, Gilberto Artioli,
Maria Chiara Dalconi, Michele Secco,
Leonardo Tamborrino, and Luca
Valentini
INTRODUCTION
227
228
229
231
CONCLUSIONS
238
SP-302-18 243
Preparation and Mechanism Study
of Slow-Release Polycarboxylate
Superplasticizers
by Jinzhi Liu, Jiaping Liu, Yong Yang,
Dongliang Zhou, and Qianping Ran
INTRODUCTION
243
246
CONCLUSIONS
250
SP-302-19 253
Effect of Pore Solution
Composition on Zeta Potential and
Superplasticizer Adsorption
by Dirk Lowke and Christoph Gehlen
INTRODUCTION
253
ELECTRIC DOUBLE LAYER AND ZETA
POTENTIAL
254
256
257
CONCLUSIONS
263
SP-302-20 265
Optimization of the Structural
Parameters and Properties of PCE
Based on the Length of Grafted Side
Chain
by Zi-Ming Wang, Zi-Chen Lu, and
Xiao Liu
INTRODUCTION
265
266
267
CONCLUSIONS
275
SP-302-21 279
Blended Antifreezing Admixture
with Extreme Freezing-Point
by Anatoly I. Vovk
INTRODUCTION
279
280
EXPERIMENTAL
RESULTS
AND
DISCUSSION
281
CONCLUSIONS
286
SP-302-22 289
Effect of Expansive Agents and
Shrinkage Reducing Admixtures on
the Performance of Fiber-Reinforced
Mortars
by M. Collepardi, V Corinaldesi, S.
Monosi, and A. Nardinocchi
INTRODUCTION
289
290
RESULTS
291
CONCLUSIONS
295
SP-302-23 299
Interactions between Cements with
Calcined Clay and Superplasticizers
by Jens Herrmann and Jrg Rickert
INTRODUCTION
299
300
300
EXPERIMENTAL
RESULTS
AND
DISCUSSION
305
SP-302-24 315
Modification of Fresh State
Properties of Portland CementBased Mortars by Guar Gum
Derivatives
by Alexandre Govin, Marie-Claude
Bartholin, and Philippe Grosseau
INTRODUCTION
315
316
316
320
DISCUSSION
324
CONCLUSIONS
328
SP-302-25 333
Interaction of Polycarboxylatebased Superplasticizer/Poly(vinyl
alcohol) with Bentonite and Its
Application in Mortar with Claybearing Aggregates
by Weishan Wang, Zuiliang Deng,
Zhongjun Feng, Lefeng Fu, and Baicun
Zheng
INTRODUCTION
333
MATERIALS AND METODS
334
RESULTS AND DISCUSSIONS

CONCLUSION
336
344
SP-302-26 349
Effect of the Stereochemistry of
Polyols on the Hydration of Cement:
Influence of Aluminate and Sulfate
Phases
by Camille Nalet and Andr Nonat
INTRODUCTION
349
350
350
351
FURTHER RESEARCH
357
CONCLUSIONS
357
SP-302-27 359
A New Accelerator Approach for
Improved Strength Development
by Franz Wombacher, Christian Brge,
Emmanuel Gallucci, Patrick Juilland,
and Gilbert Mder
INTRODUCTION
359
360
360
EXPERIMENTAL
RESULTS
AND
DISCUSSION
362
CONCLUSIONS
367
SP-302-28 371
Sorption Kinetics of Superabsorbent
Polymers in Cement Pastes
Quantified by Neutron Radiography
by Christof Schroefl and Viktor
Mechtcherine
INTRODUCTION
371
372
EXPERIMENTAL
RESULTS
AND
DISCUSSION
376
SP-302-29 387
Study on the Rheology of Fly Ash
Versus Calcined Marl Blended
Cements with PolycarboxylateBased Superplasticizers
by Serina Ng and Harald Justnes
INTRODUCTION
387
388
EXPERIMENTS AND METHODS 388
390
CONCLUSIONS
398
SP-302-30 401
A Study on the Cement
Compatibility of PCE
Superplasticizers
by A. Lange and J. Plank
INTRODUCTION
401
402
EXPERIMENTAL PROCEDURES 403
EXPERIMENTAL
RESULTS
AND
DISCUSSION
407
CONCLUSIONS
412
FURTHER RESEARCH
413
SP-302-31 415
Enhancing Workability Retention of
Concrete Containing Natural Zeolite
by Superplasticizers Combination
by Hessam AzariJafari, Mohammad
Shekarchi, Javad Berenjian, and
Babak Ahmadi
INTRODUCTION
415
416
EXPERIMENTAL
RESULTS
AND
DISCUSSION
418
CONCLUSIONS
421
SP-302-32 425
Fluidity Change of Cement Paste
with Superplasticizer by K2SO4 and
KF
by Kazuki Matsuzawa, Daiki Atarashi,
Masahiro Miyauchi, and Etsuo Sakai
INTRODUCTION
425
427
EXPERIMENTAL
DISCUSSION
CONCLUSIONS
RESULTS
AND
429
432
SP-302-33 437
Cement Recycling System Using
Sodium Gluconate
by Daiki Atarashi, Yutaka Aikawa,
Yuya Yoda, Masahiro Miyauchi, and
Etsuo Sakai
INTRODUCTION
437
438
EXPERIMENTAL
RESULTS
AND
DISCUSSION
440
CONCLUSIONS
445
SP-302-34 449
The Influence of Paste Thixotropy
on the Formwork-Filling Properties
of Concrete
by Lucia Ferrari and Pascal
Boustingorry
INTRODUCTION
449
450
EXPERIMENTAL INVESTIGATION AT
THE GROUT SCALE
450
ANALYTICAL INVESTIGATION 452
COMPARISON OF PREDICTIONS AND
453
EXPERIMENTAL
RESULTS
AND
DISCUSSION
455
FURTHER RESEARCH
460
CONCLUSIONS
461
SP-302-35 463
Interaction of Montmorillonite
with Poly(ethylene Glycol) and
Poly(methacrylic Acid) Polymers.
Consequences on the Influence of
Clays on Superplasticizer Efficiency
by Rachid Ait-Akbour, Christine
Taviot-Guho, Fabrice Leroux, Pascal
Boustingorry, and Frdric Leising
INTRODUCTION
463
464
EXPERIMENTAL PROCEDURES 464
EXPERIMENTAL
DISCUSSION
CONCLUSIONS
RESULTS
AND
466
474
SP-302-36 477
Plasticizing Geopolymer-Type
Auspensions: A Challenge
by L. Nicoleau, M. Pulkin, and T.
Mitkina
INTRODUCTION
477
478
EXPERIMENTAL SECTION
478
EXPERIMENTAL
RESULTS
AND
DISCUSSION
479
CONCLUSIONS
487
SP-302-37 491
Admixture Concepts for the SubSaharan African Environment with
Indigenous Raw Materials
by Wolfram Schmidt, Nsesheye S.
Msinjili, Herbert C. Uzoegbo, and John
K. Makunza
INTRODUCTION
491
494
494
OBSERVATIONS
497
COMPARISON OF THE CONCEPT
WITH DATA FROM PRACTICE
503
FURTHER RESEARCH
503
Index 507
SP-302-01
Effect of Clinker Grinding Aids on Static

Yield Stress of Cement Pastes
by Joseph J. Assaad and Salim E. Asseily
The impact of clinker grinding aids (GAs) based on amine, glycol, or phenol on static yield
stress (0) of cement pastes is not well understood. Results obtained from this project have
shown that GA molecules remain active after the grinding process and provide variations
in cement properties, whether in the fresh or hardened states. Flowability improved and 0
decreased when the cement is ground using increased GA concentrations. This was attributed to the adsorption of these molecules onto the cement grains and saturation of surface
charges, thus creating repulsive forces between neighboring particles. The decrease in
0 was particularly pronounced when phenol-based GA was used, given the presence of
polycarboxylate polymers that help dispersing cement particles upon mixing with water.
Keywords: amine; cement; clinker; glycol; grinding aids; phenol; static yield stress.
INTRODUCTION
Grinding aids (GAs) are incorporated during comminution of clinker to reduce electrostatic forces and minimize agglomeration of cement grains.1-3 Because of their highly
organic polar nature, such additives are preferentially adsorbed on surfaces formed by the
fracture of electrovalent bonds (Ca-O and Si-O). For a given cement fineness, this helps in
improving mill productivity and/or reducing grinding energy consumption. Typical dosage
rates vary from 0.01% to 0.15% of the manufactured cement mass.
After the grinding process, GAs may not preserve their original molecular structures;
however, they do remain adsorbed onto the cement particles to provide variations of cement
properties whether in the fresh or hardened states. The setting and hardening properties of
cement containing GAs are well documented in literature. For instance, the hydration of
C3A was accelerated in the presence of triethanolamine (TEA) due to the rapid formation
of hexagonal aluminate hydrate and its transformation to a cubic form.4 Heren and Olmez
found that the addition of increased ethanolamine concentrations alters cement hydration
and leads to retardation in setting times in the order of TEA > diethanolamine (DEA) >
monoethanolamine (MEA).5 Triisopropanolamine (TIPA), which is an amino-alcohol and
belongs to the group of alkanolamines, was found to change hydration reactions and particularly increase cement strengths. Perez et al. reported that TIPA remains in the interstitial
paste solution (not adsorbed to the cement surface, as the TEA) and form iron complexes
11
12SP-302-01
to accelerate hydration of C3S and C4AF.6 Ichikawa et al. presented evidence that TIPA
promotes hydration of limestone and densifies the interfacial transition zone (ITZ) between
hydrated cement paste and sand or aggregate particles.7
Limited data exits in literature pertaining to the effect of GAs added during clinker
grinding on rheology of cement-based materials. Aiad et al. found that viscosity of cement
pastes is dependent on the type and dosage rate of ethanolamine used, whereby a decrease
in viscosity was noticed following the sequence of TEA > poly-TEA > MEA.8 This was
related to the number of O-H groups in the ethanolamine molecules that are adsorbed on
the surface of cement grains, causing different repulsive forces and leading to variations
in fluidity levels. However, it is important to note that the tests carried out by Aiad et al.8
cannot be conclusive as the ethanolamines were post-added to the cement (i.e., not added
during the grinding process) at concentrations varying from 0.1% to 2% of cement weight
(i.e., substantially higher than in real situations). Anna et al. compared the Z-potential of
clinker containing TEA with others ground with polycarboxylate (PC) or poly-naphthalene sulphonate (PNS) concrete superplasticisers.9 They found that the TEA fluidifying
mechanism for the dry cement system lies between the steric hindrance associated with PC
polymers and electrostatic interaction of PNS with the positive charges of cement grains.9
While characterizing GAs and their impact on cement performance, Katsioti et al. noted
an improvement in workability of cement pastes containing TIPA.10 This was related to the
breaking down of cement agglomerates and balance modification between inter-particle
forces.
The objective of this paper is to assess the effect of GAs on variations in flow and static
yield stress of cement pastes prepared with different water-to-cement ratios (w/c). Grinding
tests were performed by adjusting the specific energy consumption (Ec) in order to maintain similar Blaine fineness. Amine, glycol, and phenol-based GAs were used at various
concentrations. Relevant parameters including flowability, yield stress, water demand,
setting time, and compressive strength were evaluated.
Grinding aids are increasingly used during comminution of clinker to prevent cement
particle attraction and re-agglomeration, thus resulting in clinker and energy savings that
can both lead to reduced carbon dioxide (CO2) emissions. Results presented in this paper
aim at assessing the effect of such additions on flow and yield stress of cement pastes. Such
data can be of particular interest to cement manufacturers and concrete technologists as
well as standardization committees dealing with specifications for GAs.
EXPERIMENTAL INVESTIGATION
Materials
Industrial clinker used for the production of ASTM C150 Type I cement, ground granulated blast furnace slag meeting the requirements of ASTM C989 Grade 80, and gypsum
materials were employed in this study. The C3S, C2S, C3A, and C4AF of clinker were
54.6%, 17.4%, 9.2%, and 13.7%, respectively. The slag activity index with cement at
28 days is 86.4%. The relative hardness of the clinker, slag, and gypsum determined
according to the Mohs hardness scale were around 5.5, 6, and 2, respectively.
Effect of Clinker Grinding Aids on Static Yield Stress of Cement Pastes 13
Figure 1Photo of the grinding mill used for testing.

Two commercially available GAs and a third one specially formulated for this study
were tested. First, the amine-based GA is commonly used as a grinding aid and strength
enhancer in the cement industry. It had 68% active chemicals when determined by the
Karl Fischer method, and specific gravity and pH values of 1.09 and 7.2, respectively.
The second GA is glycol-based composed by diethylene glycol (DEG) and monoethylene
glycol (MEG) chemicals. It is commonly referred to as a grinding aid and pack-set inhibitor
in the cement industry. Its active chemicals, pH, and specific gravity were 72%, 7.8, and
1.107, respectively. Finally, a specially formulated phenol-derivative GA with 70% active
chemicals, pH of 5.8, and specific gravity of 1.09 was used. This latter GA was blended
with 17.5% oleic fatty acids and 30% polycarboxylate ester molecules that are usually used
in concrete admixtures to improve cement dispersion and workability during mixing.
Production of cement used for testingA 50-L (13.25-gal) grinding mill connected to
an electric counter for monitoring Ec was used2,3 (Fig. 1). The mills drum diameter, width,
and rotational speed were 400 mm (15.7 in.), 400 mm (15.7 in.), and 50 rpm, respectively.
It contained a total of 80 kg (176 lb) steel balls among which 36 kg (79.2 lb) have 20-mm
(0.7 in.) diameter and 44 kg (96.8 lb) have 30-mm (1.18 in.) diameter. Prior to grinding,
the clinker, gypsum, and slag materials were crushed and sieved so that all particles were
smaller than 10 mm (0.4 in.).
14SP-302-01
All grinding tests were conducted using 7 kg (15.4 lb) of a mix composed of 90% clinker,
5% gypsum, and 5% slag. First, a mix ground without GA at 42 kWh/ton was tested and
considered in this project as being the control cement; its Blaine fineness was 3460 cm2/g.
Then, GAs were introduced at pre-selected concentrations varying from low to high levels;
the Ec was adjusted accordingly in a way to maintain the fineness equivalent to the control
cement, i.e. Blaine of 3460 100 cm2/g. The high GA levels were determined following
previous studies,2,3 thus ensuring that ASTM C465 requirements for water demand, setting
time, and compressive strength are satisfactorily fulfilled.11 The method for cement production consisted on grinding the materials for a certain elapsed time, stopping the mill, and
sampling around 100 grams to check whether the Blaine was close to the targeted value.
If not, additional grinding was performed. At the end of grinding, the temperature of the
charge was found to increase from ambient (i.e., 23 C (73 F)) to around 37 C (99 F).
Testing equipment and procedures
Tests on powder cementFollowing grinding, the cement fineness was determined using
the Blaine apparatus, as per ASTM C204 Test Method, and by mechanical sieving on 106,
90, and 38 m mesh openings. The R-90 and R-38 values given in this paper refer to the
individual percentages retained on the 90-m and 38-m sieve, respectively. The residues
on the 106-m sieve were in the range of 0.2% to 3.5%, depending mostly on the Ec used
(note that all particles retained on this later sieve were not included in the cement mix used
for subsequent testing).
Tests on cement pastesAll pastes were batched with a laboratory mixer using water
cooled to a temperature of around 23 C (73 F). Water was first introduced in the mixer
followed gradually by the ground cement over 2 minutes. After a rest period of 30 s, the
mixing was resumed for one additional minute. The ambient temperature and relative
humidity during testing were maintained at 23 2 C (73 36 F) and 55 5%, respectively.
The water demand required to achieve normal consistency was determined following
ASTM C187 Test Method. Using the same cement paste, the Vicat initial and final setting
times were then determined as per ASTM C191 Test Method.
The effect of GA types and concentrations on flow and rheological properties was evaluated using cement pastes prepared at 0.48 and 0.42 w/c. These w/c were selected in order to
produce pastes with different consistency levels ranging from highly flowable to relatively
cohesive. The flow was determined as the average diameter of the paste after spreading
on a horizontal surface using an ASTM C230 mini-slump cone. A rotational viscometer
connected to a data logger was used to evaluate static yield stress (0). The vane used
consisted of four blades arranged at equal angles around the main shaft; it measured 24mm
(0.95 in.) in height and 12 mm (0.47 in.) in diameter. Right after mixing, the cement paste
was poured in a cylindrical bowl and allowed to rest for one min prior to measuring the 0
value. The testing protocol consisted on subjecting the paste to a very low rotational speed
of 0.3 rpm and recording the changes in torque as a function of time. The 0 was determined
in accordance to Nguyen and Boger,12 by considering the maximum torque registered that
indicates the initiation of flow. It is important to note that the total elapsed time from the
initial contact of cement with water until the flow and 0 measurements was around 10
minutes.
Table 1Properties when cement is ground with amine-based GA

Ec, kWh/ton
42
40.4
38.6
GA dosage, % of mass
0
0.03
0.06
Blaine, cm2/g
3460
3430
3495
R-90, %
8.6
8.1
8.2
R-38, %
28.3
25.8
22
Water demand, %
27.25
27.3
27.45
Final set, min
235
240
235
w/c = 0.48
Flow, mm
195
190
200
0, Pa
9.2
10
8.8
w/c = 0.42
Flow, mm
150
150
160
0, Pa
30.4
28.8
29.6
7-d compression, MPa
35.6
35.9
38.1
47.2
48.8
49.7
Notes: 1 cm2/g = 6.8 10-5 in2/lb; 1 MPa = 145 psi; 1 in. = 25.4 mm
37.1
0.09
3360
9.3
30.1
27.2
255
195
7.6
155
28
38
49.6
35.7
0.11
3405
9.8
28.6
27.5
275
200
7.2
160
28.8
39
52.1
34.8
0.13
3375
9.5
32.8
27.05
290
205
7.6
165
26.4
36.9
50.3
Table 2Properties when cement is ground with glycol-based GA

Ec, kWh/ton
42
40.7
39.2
38.4
0
0.03
0.06
0.08
Blaine, cm2/g
3460
3415
3475
3520
R-90, %
8.6
7.8
8.4
9
R-38, %
28.3
25.2
24
29.1
Water demand, %
27.25
27.05
27.2
27.1
Final set, min
235
255
265
260
w/c = 0.48
Flow, mm
195
210
210
225
0, Pa
9.2
8.8
8
7.6
w/c = 0.42
Flow, mm
150
145
160
175
0, Pa
30.4
28.8
25.6
21.6
35.6
34.8
37
36.7
47.2
48
47
49.5
37.5
0.1
3515
8.9
34.3
27.3
300
220
7.6
170
22.4
35.3
45
Tests on mortarsThe compressive strength was determined as per ASTM C109 Test
Method. The 50-mm (2-in.) cubes were cured in water until testing at 7 and 28 days.
TEST RESULTS AND DISCUSSION
The various cement properties determined following clinker grinding at fixed Blaine
fineness of 3460 100 cm2/g using either amine, glycol, or phenol-based GA are summarized in Tables 1, 2, and 3, respectively. It is to be noted that several cement mixtures were
ground two to three times to evaluate reproducibility of testing. Acceptable coefficients of
variation (COV) were obtained; as these were equal to 3.8%, 5.1%, 4.6%, 7.4%, and 5.7%
for the Blaine, R-38, water demand, setting time, and compressive strength, respectively.
The COV increased to 8.8% for 0 responses, given the variations in Blaine fineness that
could lead to different restructuring rates during the rest period prior to shearing. Also,
confinement and yielding of paste within the rotating vane impeller could lead to some
16SP-302-01
Table 3Properties when cement is ground with phenol-based GA

Ec, kWh/ton
42
41.1
40.8
39.8
0
0.02
0.03
0.06
Blaine, cm2/g
3460
3390
3475
3405
R-90, %
8.6
8
8.2
8
R-38, %
28.3
24.1
23
27.4
Water demand, %
27.25
27.3
27.1
26.75
Final set, min
235
250
230
215
w/c = 0.48
Flow, mm
195
205
220
235
0, Pa
9.2
8.4
8.2
6.5
w/c = 0.42
Flow, mm
150
160
165
180
0, Pa
30.4
28.9
26.4
21.6
35.6
37
36.8
37.2
47.2
48.1
45.6
47.1
39
0.08
3420
9.1
30.2
26.4
205
250
4.6
190
20.7
34.4
43.5
38.5
0.09
3370
9.3
33
26.1
195
260
3.7
210
16.5
33
42.6
discrepancies in rheological measurements.12 The COV is taken as the ratio between standard deviation and mean values, multiplied by 100.
Effect of GA on Ec values
Regardless of GA type, the addition of increased concentration led to consecutively
reduced Ec values. For example, Ec decreased from 42 kWh/ton for the control cement to
37.5 kWh/ton with the use of 0.1% glycol-based GA, corresponding to 10.7% reduction in
energy consumption. Such decrease reached 34.8 and 38.5 kWh/ton with the use of 0.13%
amine-based GA and 0.09% phenol-based GA, respectively (i.e., 17.1% and 8.3% energy
decrease, respectively). This can normally be related to the GA molecules that are adsorbed
onto newly fractured surface grains, thus attenuating electrostatic forces and improving
cement fineness.2
It is to be noted that the decrease in Ec provided during the grinding process led to an
increase in R-38 and R-90 values, particularly at high GA concentrations (Tables 1, 2, and
3). Hence, R-38 increased from 28.3% for the control cement to 32.8% and 34.3% when
the amine or glycol-based GA were used, respectively, at a rate of 0.13% or 0.1%, respectively. This indicates that the sieve residues that are functions of the maximum particle size
are directly affected by the amount of grinding energy.3 Nevertheless, data provided by
the particle size analysis has shown an increased fraction of particles finer than around 5
m for those later mixtures, which allowed maintaining the Blaine fineness within 3460
100 cm2/g.
For a given GA concentration, higher decreases in Ec were achieved with the use of
amine-based GA. For example, at a rate of 0.06%, the targeted Blaine was ensured at Ec
of 38.6, 39.2, and 39.8 kWh/ton for cement ground with amine, glycol, or phenol-based
GA, respectively.
Effect of GA on variations in water demand and setting time
Water demandThe effect of incorporating various dosages of amine or glycol-based GA
did not lead to remarkable variations in water demand, as compared to the 27.25% value
Figure 2Variations in setting times for cement ground with

various concentrations of amine, glycol, or phenol-based
GA.
determined on the control mixture (i.e., values given in Tables 1 and 2 remained within
27.25% 0.25%). Conversely, water demand necessary to achieve normal consistency
considerably decreased with the use of phenol-based GA; this reduction reached 26.1%
with the addition of 0.09% dosage rate (Table 3). This can essentially be attributed to the
presence of polycarboxylate molecules that improve cement dispersion upon mixing with
water. In other words, this indicates that the GA molecules added during grinding preserve,
to a certain extent, their original functionality to effect variations in cement properties.3,13
Setting timeThe variations in final setting time for cement ground with various concentrations of amine, glycol, or phenol-based GA are plotted in Fig. 2. In the case of amine
and glycol-based GA, the setting time slightly increased at low to relatively moderate GA
dosages, and then increased sharply at higher dosages. For example, the setting increased
from 235 min for the control cement to 240 and 255 min when the amine-based GA was
added at a rate of 0.03% and 0.09%, and then reached 290 min at a rate of 0.13%. Such
increase was equal to 255, 260, and then 300 min when the glycol-based GA was used at
dosages of 0.03%, 0.08%, and 0.1%, respectively. The delay in setting with increased GA
concentrations is in complete agreement with other data published in literature. Teoreanu
and Guslicov suggested the existence of an optimum GA concentration that can be correlated with the achievement of a continuous absorption monolayer onto the micro-fractured
cement particles.14 Beyond such optimum, the adsorbed monolayer may partly blocks the
hydration reactions and leads to retardation in setting times,5,14 just like what happens with
the use of water reducers and superplasticizers in concrete mixtures.
Unlike the variations obtained for cement ground with amine or glycol-based GA, the
incorporation of phenol-based GA accelerated the setting times, as compared to the control
mix (Fig. 2). Hence, a decrease in final setting from 235 to 205 and 195 min was measured
for the control cement containing 0.08% and 0.09% GA, respectively. This can indirectly
be related to the reduced water demand added to the cement paste that was used for testing
normal consistency. The reduced w/c leads to increased cement hydration rates, thereby
18SP-302-01
Figure 3Typical variations in shear stress for 0.48 w/c

cement pastes.
shortening setting times.4 Additionally, phenol molecules are reported to remain in pore
water and form insoluble complexes that react with calcium ions released from cement,15
thus possibly contributing in accelerated hydration reactions and setting times.
Effect of GA on variations in 0 values
Typical plots of shear stress as a function of time determined on various cement pastes are
shown in Fig. 3. All profiles exhibited an elastic linear region whereby the material resists
shearing, until reaching a maximum shear stress indicating breakage of majority of bonds
and yielding of structure. The maximum value required to initiate flow is considered as 0.
The variations of 0 for tested cement pastes containing various GA types and concentrations at w/c of 0.42 and 0.48 are illustrated in Fig. 4 and 5, respectively (variations in
flow are also shown). As can be seen, the 0 appears to decrease when the cement is ground
using increased GA concentrations. For example, at w/c of 0.42, a decrease from 30.4 Pa
for the control cement to 28 and 26.4 Pa is noted when the amine-based GA was used at a
rate of 0.09% and 0.13%, respectively. The corresponding flow improved from 150 to 155
and 165 mm, respectively (5.9 to 6.1 and 6.5 in., respectively) (Fig. 4). Given that Blaine
fineness and water demand were almost similar for cement mixtures ground with aminebased GA, this indicates that the GA molecules remained active after the grinding process
and resulted in certain variations in rheological properties. The 0 decrease in presence of
GA can be due to the adsorption of these molecules onto cement particles and saturation of
surface charges, thus creating repulsive forces between neighboring cement and improving
flowability.10,13
It is interesting to note that the decrease in 0 was more pronounced when the glycol-based
GA was used, as compared to the cement ground using amine-based GA. For example, at a
dosage rate of 0.06%, 0 decreased from 29.6 to 25.6 Pa for mixtures prepared at 0.42 w/c
using cement ground with amine or glycol-based GA, respectively. When cement comes
into contact with water, it is the aluminate phases (C3A and C4AF) that react first to form a
gel based on complex sulfoaluminate hydrates. This gel exerts a barrier effect and governs
the mass flow between the inner part of the cement grain and pore water, thus controlling the
Figure 4Variations in 0 for 0.42 w/c pastes made using

cement ground with various concentrations of amine, glycol,
or phenol-based GA (1 MPa = 145 psi; 1 in. = 25.4 mm).
Figure 5Variations in 0 for 0.48 w/c pastes made using

cement ground with various concentrations of amine, glycol,
or phenol-based GA (1 MPa = 145 psi; 1 in. = 25.4 mm).
rheological behavior and hydration of the silicate phases.9 Given that TEA and TIPA have
been identified to rapidly react with the aluminate phases,4,5 this may increase viscosity of
the interstitial phase and formation of colloidal crystals between connected cement grains.
This therefore explains the relative increase in 0 measurements for pastes prepared with
cement ground with amine-based GA, as compared to those containing glycol-based GA.
The pastes prepared using cement ground with phenol-based GA exhibited the highest
improvements in flowability and decrease in 0. For example, at 0.42 w/c, the flow
increased to 210 mm and 0 decreased to 16.5 Pa with the addition of 0.09% phenol-based
GA (Fig.4). As already explained, this can be related to the polycarboxylate polymers
present in this GA that help dispersing cement particles upon mixing with water.13 There-
20SP-302-01
Figure 6Relationships between 0 and flow for all tested

cement pastes (1 MPa = 145 psi; 1 in. = 25.4 mm).
Figure 7Variations in compressive strength for mortars

made using cement ground with various concentrations of
amine, glycol, or phenol-based GA (1 MPa = 145 psi).
fore, given that the tested pastes were prepared at given w/c of 0.42 or 0.48, this explains
the improvement in flowability and corresponding decrease in 0.
Acceptable correlations exist between 0 and flow values of all tested cement pastes, as
shown in Fig. 6. The correlation coefficients (R2) were equal to 0.79 and 0.83 for pastes
made with w/c of 0.42 and 0.48, respectively. Hence, the higher the flow, the lower 0
responses.
Effect of GA on compressive strength development
The variations in compression strength determined after 7 and 28 days for mortars made
with cement ground with various GA types and concentrations are plotted in Fig. 7. As
can be seen, the increase in strength was much more pronounced for mortars prepared
using cement ground with amine-based GA, as compared to those containing the glycol
or phenol-based GA. For example, the strength increased from 47.2 MPa (6.84 ksi) for
the control mortar to 52.1 MPa (7.55 ksi) after 28 days for the mixture containing cement
ground with 0.11% amine-based GA. For given Blaine, this can mainly be attributed to
the presence of TIPA that strengthens the C-S-H compounds and densifies the interfacial
transition zone between the cement paste and sand particles.3,6,7
At high GA concentrations, the hydration reactions can be altered by the adsorbed GA
molecules,10,14 thus leading to decreased strength (Fig. 7). Such decrease was particularly
pronounced in the case of mortars prepared with cement ground with glycol and phenolbased GA. Hence, at 0.1% rate of glycol-based GA, the compressive strength dropped
to 45 MPa (6.52 ksi) after 28 days. A value of 42.6 MPa (6.17 ksi) was registered for
mortar prepared using cement ground with phenol-based GA at a rate of 0.09%. In literature, the decline in strength for cement containing fatty acids was related to oxidization
with dissolved oxygen of mixing water during the curing period, thus causing microscopic
cracks in the mortar skeleton.16
CONCLUSIONS
Based on the above results, the following conclusions can be warranted:
The incorporation of amine or glycol-based GA did not lead to remarkable variations
in water demand, as compared to control mix. Conversely, water demand necessary to
achieve normal consistency considerably decreased with the use of phenol-based GA,
due to the presence of polycarboxylate molecules that improve cement dispersion.
The setting time slightly increased at low to relatively moderate dosages of amine and
glycol-based GA, and then increased sharply at higher dosages. Such delay in setting
can be correlated to the GA adsorption onto the micro-fractured cement particles, thus
partly blocking the hydration reactions and leading to retardation in setting times.
Conversely, the use of phenol-based GA accelerated setting times, given the reduced
w/c and creation of insoluble complexes that react with calcium ions released from
the cement.
The flowability improved and 0 decreased when the cement is ground using increased
concentrations of amine-based GA. For a given Blaine fineness and water demand,
this was attributed to the adsorption of these molecules onto the cement particles and
saturation of their surface charges, thus creating repulsive forces between neighboring
and improving flowability.
The decrease in 0 was more pronounced when the glycol-based GA was used, as
compared to the cement ground using amine-based GA. This was related to the presence of TEA and TIPA in this later GA that can rapidly react with the cement aluminate
phases, thereby increasing viscosity of the solution and formation of colloidal crystals
between connected cement grains.
The pastes prepared using cement ground with phenol-based GA exhibited the highest
decreases in 0, given the presence of polycarboxylate polymers that help dispersing
cement particles upon mixing with water and improving flowability.
The increase in strength was particularly pronounced for mortars prepared using
cement ground with amine-based GA, as compared to those containing glycol or
phenol-based GA. This was attributed to the presence of TIPA that strengthens the
22SP-302-01
C-S-H compounds and densifies the interfacial transition zone between the cement
paste and sand particles.
AUTHOR BIOS
Joseph Assaad is Professor of civil engineering and R&D Manager at Holderchem
Building Chemicals, Lebanon. He received his Ph.D in 2004 from Sherbrooke University,
Qubec, Canada. His research interests include grinding aids and strength enhancers for
cement, rheology, formwork pressure, specialty concrete, injection grouts, repair systems,
durability, and use of geotechnical equipment for testing cementitious-based materials
modified with chemical and mineral additives.
Salim Asseily is Managing Director at Holderchem Building Chemicals, Lebanon. He
has been actively involved in the development of cement additives, concrete admixtures, ready-to-use mortars, and specialty chemicals. He received his BS in chemical
engineering, MS in operation research, and MBA from Columbia University, New York,
U.S.A.
REFERENCES
1. Engelsen, C. J., Quality improvers in cement making State of the art, SINTEF
Building and Infrastructure, COIN Project Report 2-2008; ISBN 9788253610719,
pp. 1-24.
2. Assaad, J.; Asseily, S.; and Harb, J., Effect of specific energy consumption on cement
fineness incorporating amine and glycol-based grinding aids, Materials and Structures,
V. 42, No. 8, 2009, pp. 1077-1087. doi: 10.1617/s11527-008-9444-0
3. Assaad, J.; Asseily, S.; and Harb, J., Effect of Grinding Aids on the Clinker Factor
and Energy Consumption of Portland Cement, Advances in Cement Research, V. 22, No.
1, 2010, pp. 29-36. doi: 10.1680/adcr.2008.22.1.29
4. Ramachandran, V. S., Hydration of cement Role of triethanolamine, Cement and
Concrete Research, V. 6, No. 5, 1976, pp. 623-631. doi: 10.1016/0008-8846(76)90026-0
5. Heren, Z., and Olmez, H., The influence of ethanolamines on the hydration and
mechanical properties of cement, Cement and Concrete Research, V. 26, No. 5, 1996,
pp. 701-705. doi: 10.1016/S0008-8846(96)85007-1
6. Perez, J.P., Nonat, A., Pourchet, S., Garrault, M., and Canevet, C., Why TIPA leads to
an increase in the mechanical properties of mortars whereas TEA does not, ACI Materials
Journal, SP217-38, Vol. 217, 2003, pp. 583-594.
7. Ichikawa, M.; Kanaya, M.; and Sano, S., Effect of triisopropanolamine on hydration
and strength development of cements with different character, Proceedings of the 10th
International Congress on the Chemistry of Cement, Ed. Amarkai A.B. and Goteborg A.B.,
Gothenburg, Sweden, 1997, 10 p.
8. Aiad, I.; Mohammed, A. A.; and Abo-El-Enein, S. A., Rheological properties of
cement pastes admixed with some alkanolamines, Cement and Concrete Research, V. 33,
No. 1, 2003, pp. 9-13. doi: 10.1016/S0008-8846(02)00911-0
9. Anna, B.; Tiziano, C.; Mariagrazia, G.; and Matteo, M., Grinding aids: A study on
their mechanism of action, http://www.mapei.com/dam/Pdf/ConferencesGrinding.pdf,
pp. 1-5.
10. Katsioti, M.; Tsakiridis, P. E.; Giannatos, P.; Tsibouki, Z.; and Marinos, J., Characterization of various cement grinding aids and their impact on grindability and cement
performance, Construction & Building Materials, V. 23, No. 5, 2009, pp. 1954-1959. doi:
10.1016/j.conbuildmat.2008.09.003
11. ASTM C465, Standard specification for processing additions for use in the manufacture of hydraulic cements, ASTM Int; 2010. Document number ASTM C465-10.
12. Nguyen, D. Q., and Boger, D. V., Direct yield stress measurement with the vane
method, Journal of Rheology, V. 29, No. 3, 1985, pp. 335-347. doi: 10.1122/1.549794
13. Assaad, J., and Asseily, S., Use of water reducers to improve grindability and
performance of portland cement clinker, ACI Materials Journal, V. 108, No. 6, 2001,
pp. 619-627.
14. Teoreanu, I., and Guslicov, G., Mechanisms and effects of additives from the dihydroxy-compound class on cement grinding, Cement and Concrete Research, V. 29, No. 1,
1999, pp. 9-15. doi: 10.1016/S0008-8846(98)00180-X
15. Ervanne, H., and Hakanen, M., Analysis of cement superplasticizers and grinding
aids A literature survey, Working report 2007-15, Posiva, Finland, 2007, 85 p.
16. Albayrak, A. T.; Yasar, M.; Gurkaynak, M.; and Gurgey, I., Investigation of the
effects of fatty acids on the compressive strength of the concrete and the grindability of the
cement, Cement and Concrete Research, V. 35, No. 2, 2005, pp. 400-404. doi: 10.1016/j.
cemconres.2004.07.031
24SP-302-01
SP-302-02
A 13C NMR Spectroscopic Study on the

Reparation of Acid and Ester Groups
in MPEG Type PCEs Prepared via
Radical Copolymerization and Grafting
Techniques
by Julia Pickelmann, Huiqun Li, Robert Baumann, and
Johann Plank
The microstructure of MPEG-type polycarboxylate (PCE) copolymers, i.e. the distribution
of side chains along the main chain was investigated via 13C NMR spectroscopy and the
effect on the interaction with cement was determined. For this purpose, two series of polycarboxylate samples (one series synthesized by radical copolymerization, the other one via
grafting/esterification) at molar ratios of COO to side chain of 2 to 10 were compared.
The 13C NMR spectra suggest that the copolymerized PCEs possess a gradient-like distribution of side chains along the main chain while the grafted PCEs exhibit a statistical
(random) repartition. Owed to those microstructural differences the grafted PCE copolymers show a tendency to adsorb in lower amount on cement. The reason is that in the
copolymerized PCEs, large blocks of COO groups are present which exhibit high affinity
to the surface of cement and therefore promote adsorption.
Keywords: copolymerization; dispersing force; grafting; microstructure;
polycarboxylates.
INTRODUCTION
Superplasticizers possessing PEO side chains which are connected with the backbone via
methacrylic ester groups are known as PCEs of the first generation. They are widely used
to improve the rheology of mortar and concrete or to reduce the water demand resulting
in higher durability of the concrete. Generally, a major advantage of polycarboxylates is
their variability in chemical composition and structure. This feature allows tailoring PCE
molecules to specific needs in highly diverse applications. For example, PCE polymers
exhibiting a long slump retention can be designed.1
25
26SP-302-02
In the past, PCEs have been studied intensively with respect to the effects of their
composition (e.g. side chain length, side chain density, anionic charge, hydrodynamic size
of the polymer, etc.) on their dispersing force and the adsorbed amounts.2-6 However, only
limited information on the correlation between the dispersing power in cement and the
microstructure of PCEs is available. The reason is the complexity of the analysis of the
microstructural composition of the polymer, especially the repartition of the side chains
along the polymer trunk. Borget et al. were the first to propose an analytical method to
elucidate the microstructure of polymethacrylic acid--methoxy polyethylenglycol graft
copolymers using 13C NMR spectroscopic measurements. According to them, the grafting
process results in a copolymer with a random distribution of the side chains along the
main chain.7 Rozzoni and Belotto also used 13C NMR spectroscopy to study the molecular
configuration and tacticity of two copolymerized MPEG-type PCEs. They found a correlation between the tacticity of the copolymer and its macroscopic properties such as e.g. a
lower critical solution temperature (LCST, cloud point).8
Still, the most interesting point is the correlation between the microstructure of the
polymers and their behavior in cement. In a first attempt Pourchet et al. compared two
well-defined polymers possessing a high side chain density but different repartition of the
pendant groups.9 The first PCE sample exhibited a gradient distribution of the side chains
while the second one possessed a random distribution. They found that the different microstructures significantly impacted the adsorption behavior of the polymers and their sulfate
resistance. However, a major drawback of this study is that both copolymers were synthesized via RAFT copolymerization which is not performed in industrial manufacturing of
PCEs. Therefore, at present no information is available which describes the relationship
between the microstructure of commonly produced MPEG PCEs and their interaction with
cement. Furthermore, no data on PCEs possessing low grafting density have ever been
published.
To close this gap, two series of MPEG-type PCEs were prepared employing the common
industrial manufacturing methods of radical copolymerization and grafting (esterification)
of -methoxy polyethylene glycol and a polymethacrylic acid (PMAA) backbone. Within
each series, the molar ratios between the COO and the side chain bearing monomers
were varied between 2 and 10 to represent the common PCE products used in actual applications. Their individual microstructures were determined by 13C NMR spectroscopic
measurements and then related with their adsorption behavior in cement.
The repartition of COO groups present in MPEG type PCEs synthesized via the two
routes practiced in the industry (radical copolymerization and grafting/esterification) was
determined analytically. It was found that copolymerized PCEs exhibit a gradient distribution of the COO groups and therefore generally adsorb in higher amounts on cement than
grafted PCEs. The study reveals that different preparation methods for PCEs may result
in different microstructures which can impact the interaction of those PCEs with cement.
Furthermore, it provides evidence for the existence of different reactivity ratios for methacrylic acid and the macromonomer in radical copolymerization.
A 13C NMR Spectroscopic Study on the Reparation of Acid and Ester

Groups in MPEG Type PCEs Prepared via Radical Copolymerization and
Grafting Techniques 27
Table 1 Phase composition of the CEM I 52.5 N sample, as determined by
Q-XRD analysis.
C3Sm
52.0
C2Sm
27.6
C3Ac
4.4
C3Ao
3.6
C4AFo
4.3
Phase [wt. %]
CaSO4.
CaSO4.
CaSO4 H2O* 2 H2O*
2.1
0.7
0.4
Calcite
3.3
Quartz
0.8
Arcanite
0.5
LOI
0.01
* determined by thermogravimetry
Cement
A CEM I 52.5 N (Milke classic, HeidelbergCement, Geseke plant, Germany) was used.
Its phase composition as determined by Q-XRD employing Rietveld refinement is shown
in Table 1. Furthermore, a mean particle size (d50 value, determined by laser granulometry)
of 10.2 m and a Blaine fineness of 3316 cm2/g (Helium pycnometry) were found for this
sample.
Synthesis of PCE samples
PCE via copolymerization For preparation, two solutions were prepared. Solution
I contained water, the monomers methacrylic acid (MAA, Sigma Aldrich, Steinheim/
Germany) and polyethylene glycol methacrylate ester (PEG-MA 1000, Clariant, Burgkirchen/Germany), and the chain transfer agent mercapto propionic acid (MPA, abcr, Karlsruhe/Germany). Solution II constituted of an aqueous solution (100 mL) of the initiator
sodium persulfate (SPS) which was kept at 0C during the entire synthesis. In preparation,
50 mL of water were placed in a 500 mL five neck round bottom flask equipped with a
reflux condenser and stirrer, flushed for 20 min with N2 and then heated to 85C. Using
two peristaltic pumps, solution I was added to the flask over 4 h and solution II over 5 h.
After cooling the polymer solution to ambient, its pH value was adjusted to ~ 7 by adding
NaOH (30 wt. %). The resulting PCE solutions exhibit a solid content of ~ 30 wt.% and are
yellowish and slightly viscous. The individual quantities and molar ratios of all reactants
used in the different syntheses are shown in Table 2.
PCE via grafting (esterification) Polymethacrylic acid (PMAA, Dow, Walsrode/
Germany) and -methoxy polyethylenglycol (PEG-M 1000, Clariant, Burgkirchen/
Germany) were placed in a 250 mL flask and then homogenized at 95C. Esterification
was initiated by gradually heating to 175C under vacuum (0.1 mbar, 0.75 Torr) while
water was distilled off and collected in a cooling trap. After ~ 6 h of reaction time, the
final product was cooled to ambient, diluted with water to a solid content of ~ 30 wt. %
and adjusted to pH ~ 7 by adding NaOH (30 wt. %). The resulting polymer solutions are
colorless and slightly viscous. The quantities of raw materials and the molar ratios for all
polymers prepared are shown in Table 2.
Characterization of PCE samples
Size exclusion chromatography Molar masses (Mw, Mn), polydispersity index (PDI),
hydrodynamic radius (Rh(z)) and conversion of all PCE samples were determined by size
exclusion chromatography (SEC). PCE solutions (concentration 10 g/L) were injected into
28SP-302-02
Table 2 Quantities and molar ratios of reactants used in the preparation of

copolymerized and grafted PCE samples
molar ratio COO
to side chain
2
3
5
6
10
MAA
10.0 g
116 mmol
15.0 g
174 mmol
25.0 g
290 mmol
30.0 g
348 mmol
50.0 g
581 mmol
PCEs via copolymerization

Macromonomer
MPA
water
58.08 g
1.23 g
42.0 g
58 mmol
12 mmol
58.08 g
1.64 g
45.0 g
58 mmol
15 mmol
58.08 g
2.07 g
51.0 g
58 mmol
23 mmol
58.08 g
2.71 g
54.0 g
58 mmol
23 mmol
58.08 g
4.52 g
66.0 g
58 mmol
43 mmol
PCEs via grafting

SPS
1.38 g
5.8 mmol
1.22 g
5.2 mmol
2.77 g
11.6 mmol
3.23 g
13.6 mmol
5.07 g
21.3 mmol
PMAA
15.5 g
180 mmol
20.66 g
240 mmol
31.0 g
360 mmol
36.16 g
420 mmol
56.8 g
660 mmol
MPEG
60.0 g
60 mmol
60.0 g
60 mmol
60.0 g
60 mmol
60.0 g
60 mmol
60.0 g
60 mmol
a Waters 2695 Separation Module equipped with three UltrahydrogelTM columns (120,
250, 500) and a UltrahydrogelTM guard column (Waters, Eschborn/Germany) and a subsequent 3 angle static light scattering detector (mini Dawn, Wyatt Technology Corp., Santa
Barbara, CA/USA). After separation, the polymer concentrations were monitored with a
differential refractive index detector (RI 2414, Waters, Eschborn/Germany). Aqueous 0.1
M NaNO3 solution adjusted to pH = 12 with NaOH was used as an eluent at a flow rate of
1.0 mL/min. The value of dn/dc used to calculate Mw was 0.135 mL/g (value for polyethylene oxide).10
Specific anionic charge amount The specific anionic charge of the polymers was
determined in alkaline solution (pH = 12) utilizing a particle charge detector PCD 03 pH
(Mtek Analytic, Herrsching, Germany). In this experiment, solutions of 0.2 g/L of the
polymer samples were adjusted to pH = 12 with NaOH and titrated against an aqueous
0.162 g/L solution of polydiallyl dimethyl ammonium chloride (polyDADMAC) until
charge neutralization (zero potential) was reached. From the amount of polyDADMAC
consumed to reach a zero potential, the amount of negative charge per gram of polymer
was calculated.
Adsorbed amounts Adsorption on cement was determined using the depletion
method, i.e. the non-adsorbed amount of polymer remaining in solution at equilibrium was
measured by analyzing the total organic carbon (TOC) content of the solution. In a typical
experiment, 30 g of cement, 9.0 g of DI water and the amount of superplasticizer to be
tested were filled into a 50 mL centrifuge tube, shaken in a wobbler (VWR International,
Darmstadt, Germany) for 2 minutes at 2,400 rpm and then centrifuged for 10 minutes at
8,500 rpm. The supernatant was filtered through a 0.2 m syringe filter and diluted with
0.1 M HCl. The TOC content of the solution was determined by combustion at 890C on
a High TOC II instrument (Elementar Analysensysteme, Hanau, Germany). The difference between the carbon content of the reference polymer and the TOC content of the
supernatant reflects the adsorbed amount of superplasticizer. Measurements were generally
repeated twice and the average was reported as adsorbed amount.
13
C NMR measurements All 13C NMR spectra were recorded in D2O at 25C and a
polymer concentration of 30 mg/mL using a Bruker AV500cryo spectrometer operating at

500 MHz and equipped with a cryo magnet. In each measurement, 512 scans and a relaxation delay of 8 s were performed to obtain quantitative results.
Characterization of synthesized PCE samples
Two different synthesis methods (copolymerization and grafting) were selected to
prepare PCE samples which were then compared with respect to their individual microstructures. Based on some of our previous experiments where representative samples
were pulled during the copolymerization process and analyzed it was expected that the
two methods will produce PCEs with different distribution of the side chains along the
main chain. According to our earlier observations, MAA and MPEG-MA exhibit different
reactivities, with the macromonomer reacting faster than methacrylic acid. This result
differs from earlier findings of Smith and Klier.11 They reported that in D2O, MAA and
PEG-MA show comparable reactivity ratios while in a 50/50 ethanol-D2O mixture, the
macromonomer reacts faster, confirming our observation. Consequently, these discrepancies required clarification.
Based on our conception of different reactivity ratios for the monomers in radical copolymerization it was expected that this method will produce PCEs with a gradient-like
distribution of the COO group whereas from the grafting process, polymers possessing a
statistical (random) distribution of the side chains should result. To investigate this point,
PCE polymers possessing molar ratios of COO to side chain of 2, 3, 5, 6 and 10 were
synthesized using both methods, free radical copolymerization and grafting technique. The
general chemical structures of the two series of PCEs used in this study are shown in
Figure 1.
The synthesized polymer samples were characterized by SEC. The chromatograms (not
shown here) confirmed high purity and uniformity for all polymers. Molar masses (Mw,
Mn), PDI, hydrodynamic radii and conversion of the PCE samples are shown in Table
3. The data show that the grafted PCEs have a slightly more narrow molecular weight
distribution and therefore lower PDIs in comparison to the copolymerized PCE samples.
All copolymerized PCEs exhibit molecular weights in the range of 13,500 to 30,000 g/
mol whereas the molar masses of the grafted PCEs decrease as a result of the decreasing
number of side chains at constant length of the PMAA backbone. For all PCE samples, the
anionic charge amounts increase linearly with increasing MAA content. Interestingly, the
specific anionic charge of grafted PCE samples consistently is slightly lower than those of
the copolymerized samples. The reason behind this difference remained unclear, because
the additional charge contribution provided by the chain transfer agent used in the copolymerization process should be almost negligible.
Microstructural analysis via 13C NMR spectroscopy
To determine the chemical shift of the 13C resonance signals representing the carboxylate
and the ester carbon atoms, the spectra of pure PMMA and of a PCE without carboxylate
groups holding only side chains designated as 23PC0 where captured first. From Figure 2
it is evident that the carbon atoms present in the carboxylate groups (COO) appear as a
sharp signal at a chemical shift of ~187 ppm while the carbon atoms of the ester group
(COOR) occur as a broad signal in the range from ~ 176-181 ppm. According to literature,
30SP-302-02
Figure 1 General structures of the copolymerized (left) and

grafted (right) MPEG type PCE samples prepared for this
study.
Table 3 Molecular weights, polydispersity index, hydrodynamic radius,
anionic charge 226 and conversion of the PCE samples prepared via copolymerization or grafting.
Polymer
sample
23PC2
G-23PC2
23PC3
G-23PC3
23PC5
G-23PC5
23PC6
G-23PC6
23PC10
G-23PC10
PMAA
Mw [g/
mol]
25,000
29,500
26,900
22,400
24,000
18,000
30,000
16,800
26,700
13,500
6,800
Mn [g/
mol]
12,600
14,900
11,900
11,800
12,200
9,700
11,900
8,800
13,100
7,300
5,700
PDI
2.0
2.0
2.3
1.9
2.0
1.9
2.5
1.9
2.0
1.8
1.2
Rh(z)
[nm]
4.3
4.7
4.5
3.9
4.0
3.4
5.8
3.5
4.7
3.0
1.5
anionic charge in NaOH (pH = 12)

[eq/g]
2,200
1,600
2,800
2,200
3,600
3,100
4,400
3,700
11,300
9,700
14,500
conversion
[%]
92
95
93
96
92
97
93
96
92
96
the carbon signals will shift depending on whether electron accepting or electron donating
groups are present in the vicinity of these carbon atoms.8 For example, a high amount of
acid groups in the vicinity of a carboxylate carbon atom causes a shift to a lower field (=
higher value of chemical shift ) while a high amount of electron donating ester groups next
to a carboxylate group leads to a high field shift of the signal. Using this approach it can
be determined whether a PCE molecule holds large blocks of COO groups, or whether
a rather homogeneous (statistical) distribution of side chains along the main chain exists.
As an example the complete 13C NMR spectrum of PCE sample G-23PC5 is presented
in Figure 3. There, the section at = 176-187 ppm which holds the information about the
repartition of COO groups in the PCE copolymer is shown enlarged. Thus, in further

Figure 2 13C NMR spectra of the reference polymers PMAA (top) and 23PC0 (bottom),
measured in D2O
analysis of all PCE samples only the range of chemical shifts between = 176-187 ppm
signifying the carboxylate/ester carbon atoms was looked at.
For analysis of the PCE samples, the carboxylate carbon atom signals were divided into
two segments: carbon atoms mainly surrounded by COO carbons as represented in area
(a) (see Figure 4), and carbon atoms surrounded by COO and ester or mainly ester
carbon atoms in area (b). According to Figure 4 it becomes obvious that copolymerized
PCEs generally contain larger segments of (a) than the grafted PCEs. This signifies that
the copolymerized PCEs contain larger blocks of COO along the main chain, compared
to the gradient-like distribution of comonomers in the copolymerized PCE samples which
apparently is owed to the different reactivities of MAA and the macromonomer. Such
disparaged incorporation of the monomers leads to blocks of methacrylic acid within the
copolymer and therefore to increased intensity of the signals in range (a) (see Figure 4). In
contrast to that, for the grafted polymers G-23PC2 to G-23PC5 this signal is either weak
in intensity or completely missing which confirms a statistical distribution of COO and
ester groups along the polymer trunk.
However, for the PCE samples G-23PC6 and G-23PC10 possessing high charge density
(= low grafting density) the difference between the carbon signals of grafted and copolymerized PCEs becomes smaller. Thus, for the corresponding copolymerized and grafted
polymer samples, both the shape of the carboxylate signal and its chemical shift are comparable. This suggests that at high anionic charge density, the microstructures of copolymer-
32SP-302-02
Figure 3 Typical 13C NMR spectrum of PCE sample G-23PC5, measured in D2O at pH =
7; insert: magnification of the carbonyl region.
ized and grafted PCE samples converge, as is schematically shown in Figure 5. Whereas at
low anionic charge density, the copolymerized PCE samples still possess significant blocks
of polymethacrylic acid and of the PEG polyester. Such structural disparities do not occur
in the grafted PCE samples.
In order to understand whether those microstructural differences translate into a different
interaction with cement, adsorption isotherms of the PCE samples on cement were determined (see Figure 6). Note that under the experimental conditions chosen here the adsorption behavior of the PCE samples is mainly influenced by the type and reactivity of C3A
(i.e. the surface area of the ettringite produced).
For all PCE samples, the tendency as expected was observed: at increased charge density,
the maximum adsorbed amount of polymer increases. When comparing the copolymerized and the grafted PCE polymers, however, it becomes obvious that the grafted copolymers consistently produce lower saturated adsorbed amounts than the copolymerized
ones. Those variations appear to become smaller at increased charge density. For example,
sample G-23PC2 adsorbs only half of the amount of 23PC2 while the maximum adsorbed
amounts of samples G-23PC10 and 23PC10 are similar. This conversion of copolymerized
and grafted copolymers exhibiting high charge density correlates well with the observations made in the 13C NMR spectra and the microstructural analysis (see Figure 4).
To confirm the model that the adsorption tendency of PCE is owed to the amount of
COO groups present, a PCE molecule without carboxylate groups (a homopolymer of the

Figure 4 13C NMR signals of the carboxylate and the ester

carbon atoms of copolymerized (left) and grafted PCEs
(right), recorded in D2O at pH = 7. Spectra of polymethacrylic acid (PMAA) and of a completely esterified polymer
(23PC0) are shown for reference (bottom).
macromonomer) was synthesized via radical copolymerization. When mixed into cement
paste, no adsorption at all was detected thus confirming the strong influence of the COO
blocks on the adsorption behavior of PCEs.
Generally, a polymer with a gradient distribution of carboxylate groups is more likely
to adsorb in a tail conformation, as the blocks of COO groups provide particularly
strong electrostatic attraction to the surface of cement. However, such adsorbed conformation leads to only partial coverage of the surface of cement particles, as is schematically
illustrated in Figure 7. Whereas, polymers exhibiting a statistical repartition of carboxylate
groups can be expected to adsorb in a loop conformation whereby a larger area of the
surface is covered (Figure 7). Consequently, there a lower amount of polymer should be
required to achieve the same surface coverage as from a copolymer with a gradient-like
34SP-302-02
Figure 5 - Schematic illustration of the microstructure of

copolymerized and grafted PCE samples in dependence of
their charge density.
Figure 6 - Adsorption isotherms on cement of PCE samples

synthesized via copolymerization or grafting.
microstructure. This effect can also explain the lower adsorbed amounts for the grafted
PCEs. However, at increasing charge density the gradient structure converges into a more
random one which is similar to that of the grafted PCE (Figure 5). Accordingly, both types
of PCE (copolymerized and grafted) now adsorb in a loop conformation which results in
quite comparable maximum adsorbed amounts.
CONCLUSION
MPEG-type PCE superplasticizers were synthesized via two different methods resulting
in polymers possessing different microstructures. 13C NMR spectroscopic measurements

Figure 7 Idealized illustration of the adsorbed conformations of PCE samples with gradient (left) and statistical
(right) distribution of carboxylate groups along the trunk
chain.
revealed a gradient-like distribution of the side chains for the copolymerized PCEs and
a statistical (more regular) distribution of the side chains for the grafted PCEs. With
increasing charge density those structural differences become smaller which is also
reflected in the interaction between the polymers and cement. At high side chain density,
copolymerized PCEs possessing a gradient structure adsorb in higher amounts on cement
due to enhanced accessibility of the charged parts. However, for PCE polymers exhibiting
high charge density (= low side chain density), such differences in the maximum adsorbed
amounts for the copolymerized and the grafted PCEs disappear.
The study demonstrates that a microstructural analysis of individual PCE polymers is
possible and that differences in the microstructure can affect the behavior of PCE superplasticizers in cement.
AUTHOR BIOS
Julia Pickelmann studied chemistry at TU Mnchen and currently works on her PhD
thesis at the Chair for Construction Chemistry. Her focus is the correlation between the
microstructure of PCEs and their dispersing effectiveness in cement.
Huiqun Li studied Material Science at Beijing University of Technology and currently
is hosted as a guest Ph.D. student at TUMs Center For Advanced PCE Studies. Her
research interests include synthesis and characterization of PCE polymers.
Robert Baumann received a PhD degree in chemistry from Regensburg University in
1988 and then joined Dow Europe in Horgen (Switzerland) as development chemist. In
36SP-302-02
his current role as R&D Fellow for Dow Construction Chemicals he is responsible for the
strategic development of additives for the construction industry.
Johann Plank is full Professor at the Institute of Inorganic Chemistry of Technische
Universitt Mnchen, Germany. Since 2001, he holds the Chair for Construction
Chemistry there. Research interests include cement chemistry, chemical admixtures,
organic-inorganic composite and nano materials, concrete, dry-mix mortars and oil well
cementing.
ACKNOWLEDGEMENTS
Julia Pickelmann wishes to thank Dow Europe for sponsoring her Ph.D. and Li Huiqun
expresses her gratitude to TUM Center For Advanced PCE Studies for supporting her
stay as a guest researcher.
REFERENCES
1. Ramachandran, V. S., and Malhotra, V. M., Superplasticizers, Concrete Admixtures
Handbook, (Second Edition), W. A. Publishing, Park Ridge, NJ, 1996.
2. Yamada, K.; Takahashi, T.; Hanehara, S.; and Matsuhisa, M., Effects of the chemical
structure on the properties of polycarboxylate-type superplasticizer, Cement and Concrete
Research, V. 30, 2000, pp. 197-207.
3. Sakai, E.; Yamada, K.; and Ohta, A., Molecular Structure and Dispersion-Adsorption Mechanisms of Comb-Type Superplasticizers Used in Japan, Journal of Advanced
Concrete Technology, V. 1, 2003, pp. 16-25.
4. Winnefeld, F.; Becker, S.; Pakusch, J.; and Gtz, T., Effects of the molecular architecture of comb-shaped superplasticizers on their performance in cementitious systems,
Cement and Concrete Composites, V. 29, 2007, pp. 251-262.
5. Plank, J.; Pllmann, K.; Zouaoui, N.; Andres, P. R.; and Schaefer, C., Synthesis
and performance of methacrylic ester based polycarboxylate superplasticizers possessing
hydroxy terminated poly(ethylene glycol) side chains, Cement and Concrete Research, V.
38, 2008, pp. 1210-1216.
6. Lange, A.; Hirata, T.; and Plank, J., The Role of Non-Adsorbed PCE Molecules In
Cement Dispersion: Experimental Evidence For a New Dispersion Mechanism, V. M.
Malhotra (Ed.), 10th CANMET/ ACI Conference on Superplasticizers and Other Chemical
Admixtures in Concrete, Prague, 2012, SP-288.30, 435-449.
7. Borget, P.; Galmiche, L.; Le Meins, J.-F.; and Lafuma, F., Microstructural characterisation and behaviour in different salt solutions of sodium polymethacrylate-g-PEO comb
copolymers, Colloids and Surfaces A: Physicochemical and Engineering Aspects, V. 260,
2005, pp. 173-182.
8. Rozzoni, A., and Bellotto, M., Configurational NMR Study of Sodium Polymethacrilate-g PEO Comb Polymers, 14th International Congress on the Chemistry of Cement,
Madrid, 2011
9. Pourchet, S.; Liautaud, S.; Rinaldi, D.; and Pochard, I., Effect of the repartition of the
PEG side chains on the adsorption and dispersion behaviors of PCP in presence of sulfate,
Cement and Concrete Research, V. 42, 2012, pp. 431-439.

10. Kawaguchi, S.; Aikaike, K.; Zhang, Z.-M.; Matsumoto, K.; and Ito, K., Watersoluble bottlebrushes, Int. J. Polym., V. 30, 1998, p. 1004
11. Smith, B. L., and Klier, J., Determination of monomer reactivity ratios for copolymerizations of methacrylic acid with poly (ethylene glycol) monomethacrylate, Journal
of Applied Polymer Science, V. 68, 1998, pp. 1019-1025.
38SP-302-02
SP-302-03
Effects of Particle Volume Fraction and

Size on Polysaccharide Stabilizing Agents
by Wolfram Schmidt, Sarah Peters, and Hans-Carsten
Khne
Polysaccharides modify the rheological properties of cement based systems. Depending
upon their chemistry, molecular architecture, and adsorption tendency, they have different
modes of action. Some polysaccharides like diutan gum have strong effect on the fluid
phase; others like starch strongly interact with particles. This paper presents effects of
diutan gum and starches in presence of polycarboxylates. Rheometric investigations
with varied particle volume fractions and increasing coarse aggregate diameters were
conducted. The results show that starches have stronger influence on the rheology at high
particle volume fractions than diutan gum. At lower particle volume fractions this trend is
inverted. Experiments with aggregates sizes up to 16 mm (0.63 in.) indicate that stabilizing
agent influences on the effects of aggregates on yield stress were small; however up to 1.0
mm (0.04 in.), a significant effect on the plastic viscosity could be observed, which levelled
off at larger diameters.
Keywords: diutan gum; starch; polysaccharides; polycarboxylate ether; volume fraction.
INTRODUCTION
Today, the fresh properties have become a characteristic item to designate various classes
of concrete. For many concrete types, such as self-consolidating concrete (SCC) or ultrahigh performance concrete (UHPC), the rheological properties directly affect their performance at hardened state. The major driving force for innovations in concrete technology
over the last three decades has been brought by the invention and steady enhancement of
superplasticizers, since they allow uncoupling the workability properties from the water
content.1 The main relevant rheological properties for cement based systems are the yield
stress (0) and the plastic viscosity (pl), which describe the minimum shear stress that
needs to be exceeded in order to make a system flow and the resistance of w flowable
system against a shear deformation, respectively.
However, sophisticated rheological properties in complex multi-phase systems like
concrete come along with a high risk of failure as soon as the boundary framework changes.
Since the function of superplasticizers depends upon the adsorption on the early cement
hydration phases AFt and AFm, any influence that accelerates or slows down cement
39
40SP-302-03
Fig. 1Possible effects of diutan gum and starch based STA without and in presence of
PCE.
hydration at a very early point in time after water addition can automatically affect the
rheological properties.2-4 Therefore, unexpected effects that affect the cement hydration at
early age can yield both, unwanted stagnation or segregation.
A stabilizing agent (STA) can help reduce unwanted loss of flowability or segregation.
Therefore, their application has gained increasing interest in the scientific research.5-15
For the stabilization of cement based systems, polysaccharides such as cellulose, starch,
or bio-gums (e.g. sphingans) are typically used. Depending on their molecular weight,
radius of gyration, and ionic charges, they affect the rheology of cement based systems in
different ways. The stabilizing mechanisms can be based on the water binding capacity,5,7
the viscosity modification of the pore solution,8,15 depletion forces,9,14 or effects induced by
the sheer size of the polymers causing interactions with particles,8,9 which can have radii of
gyration as large as 500 nm, or if still agglomerated even larger.
For the understanding of the working mechanism of STAs, it is also of highest importance to take into account that superplasticizers can have a significant influence on their
efficiency. These influences are shown in Fig. 1. It was found that in systems without
superplasticizers, starch had only a tangible effect on the yield stress (0) at higher particle
volume fractions, while diutan gum increased 0 already without the presence of particles. Approximately 20% of the starch consists of amylose, a linear glucose chain, and
about 80% consists of amylopectin, which is a huge branched molecule formed by glucose
chains. The effect of the starch was mainly attributed to the effects of the amylopectin
molecules, which interact with small solid particles like spacers but require interacting
particles for activating their stabilizing effect. The effect of the diutan gum, which contains
anionic groups in the backbone, was attributed to water absorption and thickening of the
fluid phase as well as adsorption on cement and hydration phases. However, in the presence
Effects of Particle Volume Fraction and Size on Polysaccharide Stabilizing

Agents41
Table 1Stabilizing agent modifications used for the investigations and
dosages required to achieve hPl of 3.5 mPas
Diutan gum
Abbreviation
Dosage to achieve
pl = 3.5 mPas in water
[% by mass of water]
Starch with
Starch with
low degree of high degree of
modification modification
Cationic
starch
Anionic starch
DGUM
ST-low
ST-high
ST-cat
ST-an
0.04%
0.84%
1.3%
1.04%
0.44%
of superplasticizer, effects of STA adsorption retreated into the background and both STAs
showed only negligible effects on 0 but maintained a strong effect on the plastic viscosity
pl.9,13
It is hence becoming obvious that for the understanding of STAs, it is necessary to distinguish between systems with and without superplasticizers as well as to take into account
the particle volume fraction P and possible interactions between STAs and particles. This
study aims at understanding these influences for diutan gum and various starch modifications in limestone filler (LSF) and cement based flowable systems as well as to look at
whether the aggregate size and volume fraction may have an effect on the efficiency of
STAs.
Today, the use of superplasticizers has become common practice for flowable concrete.
However, flowable concrete systems often lack stability, and therefore may need to be
amended by stabilizing admixtures, which are typically based on polysaccharides. Their
mode of operation is not very well understood today. Being able to manipulate the rheology
has become of utmost importance for bringing innovative concrete into practice. The
present study investigates interactions of polysaccharides with superplasticizers, cement
hydration, particle volume fraction, and the sand and aggregates. The goal is better understanding of the mode of operation of these materials.
Materials
In order to understand effects of different STA modifications, investigations were
conducted on paste, mortar, and concrete matrices. The observations were focused on
diutan gum and a variety of starches. Two starches were modified at different degrees with
hydroxypropyl groups, another starch was modified with cationic groups, and another with
anionic groups. The STAs that were evaluated are listed in Table 1. Their dosages were
adjusted in a way that a plastic viscosity (pl) of 3.5 mPas in water was achieved. These
required dosages to achieve this property are listed for each STA in Table 1.
Paste investigations were conducted at varied particle volume fractions P with LSF and
with cement, respectively. Their specific gravities can be found in Table 2. For the LSF
pastes, P was varied between 0%, 33%, and 50%. For the cement pastes, P was varied
42SP-302-03
Table 2Specific gravities of mixture constituents and mixture composition

for SCC
Spec.
Gravity [-]
Net weight
[kg/m3]
Volume fraction [%]
Water
Cement
Lime
stone
filler
3.13
2.74
1.00
2.60
2.60
2.60
2.60
2.60
2.60
354
130
177
331
192
192
126
126
702
11.3
4.8
17.7
12.7
7.4
7.4
4.9
4.9
27.0
0.1-0.5
mm
Sand
0.5-1.0 1.0-2.0
mm
mm
2.0-4.0
mm
Aggregate
4.0-8.0
8.0-16
mm
mm
Fig. 2Influence of the particle volume fraction on 0 and

pl of LSF and cement pastes.
between 0%, 25%, and 40%. The reason is the stronger influence of P on the increase of
both yield stress (0) and pl in cement based systems, as can be found in Fig. 2.
The influence of different dosages of polycarboxylate ether superplasticizer (PCE) was
also investigated in LSF and cement pastes. These investigations were conducted at P of
50% and 40% for LSF and cement pastes, respectively. The dosages of PCE were varied
between 0% and 1% by mass of solid PCE related to the mass of solid particles.
Mortar and concrete investigations were conducted on a mixture composition for SCC,
which is shown in Table 2. The aim was to identify influences of P and the maximum
particle diameter. Therefore, rheometric investigations were conducted on paste, mortar
and concrete with varied solid volume contents and maximum grain sizes according to
Table 2. Pure systems without STA were compared with systems incorporating DGUM
and ST-low. The STA dosages were as in Table 1 and the PCE dosage was 0.68% solids by
the cement mass. The PCE in use was a low charge density PCE. The respective dosage
provided a long flow retention at a slump flow value of 650 mm (25.6 in.) at 30 minutes
after water addition.

Agents43
Table 3Measurement regime for rheometric paste investigations.
Profile characteristic
Time
(max.) Shear rate [1/s]
Purpose
Upward ramp
0 s - 30 s
0 to 73.5
-
Plateau
30 s - 60 s
73.5
-
Downward ramp
60 s- 240 s
73.5 to 0
Flow curve determination
Mixing of pastes and mortars

For all mixes, the STAs were always first dissolved in a blender mixer in the water for the
mixture addition. The mixing of the pastes and mortars was conducted in a mortar mixer.
After a 30 s dry mixing phase, water including STA was added and mixed for 2 minutes.
Then (if required) PCE was added and mixing was continued for another minute. In order
to avoid peculiar initial effects induced by the hydration of ettringite and monosulfate, the
experiments were conducted approximately at 10 minutes after the water addition.
Paste rheometry
For the investigations of the rheological properties of pastes, a Couette type viscosimeter (Schleibinger NT) was used in combination with a double gap cell with a network
structured grid as shear body. The cell allows the measurement of pastes and mortar up to
a grain size of 2 mm. Since shear forces affect the cement hydration and the adsorption of
polymers, for cementitious systems it is of utmost importance to keep the measurement
time as short as possible. The applied profile is shown in Table 3. It is considered to be
a reasonable compromise between precision and compactness. For the conversion of the
flow curves, a Bingham approximation was chosen. Though this does not take into account
shear thinning effects that were observed to certain extent in some systems, it is in good
agreement with the results for all P and admixture dosages, and the comparison between
the pastes is facilitated.
Mortar rheometry
For the rheometric mortar investigations in systems containing sand in the size fractions 0.1 mm to 4.0 mm (0.004 - 0.16 in.) the same rheometer was used as for the paste
investigations. However, a mortar cell including a stirrer was used. The cell does not allow
the conversion to fundamental units, but it allows determining an ordinate intercept and
a slope of the measured torque, which can provide relative information on differences in
yield stress and plastic viscosity.
Concrete rheometry
For concrete with aggregate sizes 4.0 mm to 16.0 mm (0.16 - 0.63 in.), a concrete rheometer (Rheometer-4SCC) was used. Like for the mortar equipment used in these investigations there is no reliable method for the conversion of torque and rotational velocity
into shear stresses and shear rates, the results can only provide qualitative information on
changes in yield stress and plastic viscosity. Therefore the measurements are referred to as
G-Yield for qualitative observations of yield stress changes and H-Viscosity for qualitative
changes of the plastic viscosity, respectively.
44SP-302-03
Fig. 3Influence of P on 0 of LSF and cement pastes with different STAs

Paste slump flow tests
In addition to the rheometric investigations, slump flow tests were conducted with varied
dosages of PCE. In order to avoid segregation, they were conducted at a water volume,
which just surrounds the particles and fills the voids in between but without any excess
water. This water demand was determined according to the so called Puntke test.2,16,17 The
slump flow tests were conducted 10, 20, and 30 minutes after the addition of water. Furthermore, paste slump flow tests were conducted at fixed PCE and STA dosages with varied
cement volume fractions varying between 69.8% and 63.5%. Below 63.5% the cement
pastes were no longer free from segregation at the used dosage of PCE.
EXPERIMENTAL RESULTS AND DISCUSSION
Paste rheometry at varied particle volume fraction
The effect of the different STAs on 0 depending upon P can be observed for LSF and
for cement pastes in Fig. 3. Since the differences in 0 were strongly affected by the P, the
figure shows results normalized to reference mixtures without STA. The figure therefore
shows the ratio of 0 of a paste modified with STA in the dosage given in Table 1 and an
identical paste without STA.
Apart from the anionic starch ST-an, all STAs increase the 0 of LSF pastes at P =
33%. ST-an shows lower values for 0 than the reference at 33% and at 50%, indicating
that the anionic charges may interact with the LSF and have a plasticizing effect regardless
of the P. At 50%, the yield stress increasing effect of the other STAs is reduced significantly for all STAs. Only for the ST-low this reduction is less significant than for all other
STAs. ST-cat and ST-high show lower 0 at that solid volume fraction than the reference.
In the cement pastes at low P, all STAs apart from the cationic starch show higher 0
than the reference, while this effect is reduced at higher particle solid volume fraction. An
exception is DGUM. No effect of P on how DGUM increases 0 can be observed, while

Agents45
Fig. 4Influence of P on pl of LSF and cement pastes with different STAs

in LSF paste this drop is significant. The attractive forces in a LSF system are significantly
lower than in a cement system. Therefore, the lower influence of P on the effect of the
DGUM in the cement system may be attributed to higher adsorption compared to the LSF
system. The higher adsorption would withdraw polymers from the solution that would
otherwise increase the solutions viscosity.
For ST-high, a similar drop of the yield stress effect can be observed in LSF and cement
systems; however, the effect of this STA is much stronger in the cement system. For ST-low,
which has a strong effect in limestone paste but not in cement paste, this observation is
inverted. The reason for this effect could be that the higher ionic strength of the cement
pore solution may negatively affect the performance of the slighty modified agent, e.g.
by causing interactions within the huge amylopectin molecule or by linking amylose and
amylopectin. The higher degree of modification of ST-high may more effectively prevent
such interactions.
A prominent observation is that the anionic and the cationic starch exactly show inverted
behavior in LSF and in cement systems. Both show a plasticizing effect at high P, indicating that they induce electrostatic repulsive forces. However, at low P this plasticizing
effect can only be observed for ST-an in limestone and for ST-cat in cement pastes. It is
known that LSF in water attracts anionic polymers.2 It is therefore likely that in LSF pastes
the adsorption of ST-an incorporates repulsive forces, while this adsorption is reduced in
the cement system due to the presence of sulfates.
Fig. 4 shows the effect of P on pl. For the LSF system, it can be observed that apart from
ST-an, all starch based STAs showed reduced effects of the STAs with increasing amount
of P. For DGUM at P = 50%, an increasing effect can be observed, which is similar to
the effect of the ST-an. Both STAs have anionic charges, so that this effect is likely to be
attributed to their adsorption on particles. The strong effect of the ST-an on the viscosity
increase relative to the reference system at P = 33% indicates that the huge amylopectin
molecule may bridge particles with increasing effect on pl. In this context, it is interesting
46SP-302-03
Fig. 5Influence of the PCE dosage on 0 of LSF and cement pastes with different STAs
to observe that at P = 50% both anionic STAs show higher effect on pl, but as shown in
Fig. 3, in terms of 0 only the starch based agent had a plasticizing effect. It is assumable
that this plasticizing effect may be induced by the adsorption of mainly the small amylose
molecules, while the viscosity increase is caused rather by the huge amylopectin molecules
and thus similar to the effect of the DGUM. In cement based systems, it can be observed
that a P of 25% causes reduced effectiveness of the STAs, which further reduces at 40%.
At 40%, the plastic viscosities of all systems were lower than the reference.
Paste rheometry at varied PCE dosage
In order to observe effects of STAs in presence of superplasticizers, LSF and cement
pastes were evaluated at P of 50% and 40% for LSF and cement, respectively, at varied
PCE dosages. Upon addition of PCE, 0 was reduced significantly for all mixtures. Fig. 5
shows the ratios of 0 with STA compared to systems without STA.
For the LSF system, it is interesting to see that the two non-ionic STAs both do not distinguish greatly from the reference paste, regardless of the PCE dosage. The anionic admixtures, ST-an and DGUM, both show a horizontal curve up to a PCE dosage of 0.5% above
which they generate a higher 0 than the reference system. It is assumable that competitive
adsorption between the STAs and the PCEs takes place causing less adsorption of PCE,
which increases 0. For the cationic starch, it can be observed that above a PCE dosage
of 0.1% 0 is always higher than in the reference system. In this case it is most likely that
the cationic starch interacts with the anionic PCEs, which reduces their adsorption, which
automatically increases 0.
In the cement system, up to a dosage of 0.5% the observations are similar to those of
the limestone system with the only difference that the anionic starch ST-an does not have a
different effect on 0 than the other agents. At 1% PCE dosage, the observations are more
complex. What is common to the LSF system is the observation that the influence of the
non-ionic starches is small also at 1% of PCE addition. The effect of the cationic starch

Agents47
Fig. 6Influence of the PCE dosage on pl of LSF and cement pastes with different STAs
is smaller than that of the other charged agents. In case of ST-cat 0 is lower than that of
the reference, which is different to the limestone filler system. Also 0 of the paste with
the anionic starch is lower than the reference. Only for the diutan gum STA a significantly
higher 0 can be observed at 1% PCE. The reason for this can be attributed to a competitive adsorption that reduces the effect of the PCE. However, it is interesting to see that
the anionic starch shows an adverse effect. This may be caused by adsorption mainly of
the amylose molecules, which may be responsible for a supplementary fluidizing effect
as already shown in Fig. 3. The large amylopectin molecule may not be found adsorbed
directly on the surfaces. This would also explain why the anionic and the cationic starch
show similar behavior.
The results for the ratio pl,STA/pl,Ref is shown in Fig. 6. It can be found that in the
LSF system, all STAs maintained pl significantly higher than in the reference system.
The ST-an, ST-cat and the ST-high became particularly effective at high PCE 0.5% PCE
dosage, and this effect was maintained also at 1% PCE.
In the cement system, all pastes with STAs had a lower pl than the reference. However,
with increasing PCE dosage, they became increasingly more efficient in generating higher
plastic viscosities than the references. At PCE dosages above 0.5% all STA modified
systems apart from the system with ST-low showed higher plastic viscosities than the reference with an increasing trend for higher PCE dosages. The least effect could be observed
for ST-low. Only at 1% PCE dosage a slightly viscosity enhancing effect can be observed.
It can be assumed that compared to the adsorption of PCE, adsorption of STAs has
a negligible effect on pl (different from 0). Therefore, the yield stress enhancement
compared to the reference is most likely induced mainly due to the pure presence of large
polymers in the pore solution.
48SP-302-03
Fig. 7Influence of the presence of STA on the slump flow

of cement pastes with different dosages of high charge PCE
at P = 0.6. The dosage of starch was selected to achieve a
similar slump flow reduction at the PCE dosage that yields
the maximum slump flow.
Influence of STA on the effectiveness of PCE
It was shown that the presence of PCE can have a significant influence on how effective
stabilizing admixtures are . In order to better assess this interaction, flow tests on cement
pastes were conducted with increasing PCE dosages. Due to the high powder particle
fraction, the observed pastes do not show any flow without PCE addition. For each PCE
dosage, the water content of the PCE is reduced from the total water content, hence, the
flow is only induced by the presence and dosage of PCE. A more detailed description of
the experiment can be found elsewhere.2 Without the presence of STA, the maximum flow
of about 320 mm (12.6 in.) can be achieved at a PCE dosage of approximately 1% (Fig.
7), beyond which further addition of PCE does no longer contribute to a wider flow diameter. The flow initiation dosage is at approximately 0.2%. In a next step, it was observed
how these values are changed by the addition of 0.1% DGUM, and it can be seen that the
characteristic dosages for the flow initiation and maximum remain similar to the reference
system without PCE. However, the maximum flow is significantly reduced to a value of
approximately 250 mm (9.84 in.). In a further step, the dosage of ST-low was determined
that causes a similar maximum flow as the system with DGUM. Based on the dosages
listed in Table 1, the first attempts were conducted at significantly higher dosages than
for the DGUM system. However, it was found that these systems were not flowable at all.
Eventually, the dosage that showed performance like DGUM was only 0.06% of the starch.
The fact that in water and low P systems, the DGUM was required at significantly lower
dosages than the starches for similar flow properties (Table 1). But since this is obviously
inverted at very high P (Fig. 7), it is recommended to test the influence of the P. Fig. 8
shows results of slump flow values at 1.0% PCE and in the presence of 0.1% ST-low and
DGUM dosage, respectively. This PCE dosage was the dosage that yielded the maximum
flow for the reference systems in the tests shown in Fig. 7.

Agents49
Fig. 8Influence of the particle volume fraction of cement

pastes on the effectiveness of starch based and diutan gum
based STA.
Fig. 9Influence of the maximum particle size diameter

and the particle volume fraction on the effectiveness of
stabilizing agents to affect the 0.
It can be clearly identified that indeed the starch addition causes a smaller slump flow
than the diutan gum at a very high solid volume fraction of 69.8%. This effect is reduced
with decreasing solid volume fraction. At a solid volume fraction of approximately 67%,
both STAs have similar effects on the slump flow, while with decreasing P the diutan gum
becomes more and more efficient. It is hence obvious that P has a very strong effect on
the effectiveness of different STAs in flowable systems.
50SP-302-03
Fig. 10Influence of the maximum particle size diameter

and the particle volume fraction on the effectiveness of
stabilizing agents to affect the pl.
Influence of cumulative addition of sand and aggregate fractions
For the observation of the influence of the addition of aggregates, for the different dimensions of the aggregates different equipment was used. Both rheometers used, cannot derive
adequate fundamental properties for yield stress and plastic viscosity and the results appear
in different units. Therefore, here again, the ratios of G-yield with STA to the G-yield
without (Fig. 9) as well as the ratio of the H-viscosity with STA to the H-viscosity without
(Fig. 10) are shown. This way of representing has the benefit that effects of the aggregate
size and the volume fraction on the mortar and concrete rheometry are neglected and the
direct influence only of the presence of STA can be identified.
For yield stress (Fig. 9) it can be observed that the effect of the starch based agent
compared to the diutan gum was significantly higher. This observation is in contrast to
Fig. 5, where in the presence of PCE the diutan gum showed stronger effects in presence
of LSF and cement at a similar P. Nevertheless, the pure pastes and the paste of the SCC
from LSF and cement (Table 2) were different mixtures. Increasing aggregate volumes and
maximum grain sizes slightly increase the effectivity of both STAs. For both STAs this
effect is not very prominent, but clearly identifiable.
Regarding the plastic viscosity, which is shown in Fig. 10, it can be observed that both
STAs show similar trends. Up to sand size of about 1.0 mm (0.04 in.) it can be observed
that the effect of STA on the plastic viscosity is reduced and then levels off still at higher
value than the reference. This indicates that the maximum grain size and P contribute
significantly to the efficiency of STAs.
SUMMARY AND CONCLUSIONS
For the evaluation of the effectivity of stabilizing agents, the ionic strength of the pore
solution, the presence of superplasticizers, and the P have to be considered. The following
conclusions can be drawn:

Agents51

The performance of STAs in LSF and cementitious systems can vary greatly due to
different ionic strengths.
Charges incorporated in STAs can have a strong effect on their performance.
For starch based STA supplementary effects have to be taken into account due to the
difference between amylose and amylopectin.
In the presence of PCE, STA has only little influence on the yield stress but a significant increase of the plastic viscosity can be achieved.
As a function of P the effectiveness of starch based STA is increased.
Increasing sand and aggregate contents enhance the yield stress effect of STAs but
reduce their viscosity effect.
A strong effect of STAs on the viscosity can be observed up to a grain size of 1.0 mm
(0.04 in.).
As a consequence of the complex interactions, the performance of an STA in a liquid
or paste system cannot be converted to concrete without further considerations.
AUTHOR BIOS
Dr. Dipl.-Ing. Wolfram Schmidt is a researcher at the BAM Federal Institute for Materials Research and Testing in Berlin. He received a Dipl.-Ing. from RWTH Aachen and a
PhD from TU Eindhoven. His research focuses on SCC, admixtures, and rheology. He is
member of the RILEM committee 228-MPS and the fib task group 8.8.
M.Eng. Sarah Peters is a researcher at the BAM Federal Institute for Materials
Research and Testing in Berlin. She received a M.Eng from BHT Berlin in structural
engineering. Her master thesis dealt with the use of stabilizing agents in different pastes.
Her current research focuses on the application of lightweight granules, which are made
from recycled masonry rubble.
Dr.-Ing. H.-C. Khne studied mineral processing at the TU Berlin. He received a
Dr.-Ing. degree from the Technical University Hamburg-Harburg at the chair of building
physics and construction materials. In 2003, he joined the Department for Safety of
Structures at BAM and became head of the division for Technology of Construction
Materials in 2011.
REFERENCES
1. Schmidt, W.; Sonebi, M.; Brouwers, H. J. H.; Khne, H.-C.; and Meng, B.Chemistry
and Materials Research, V. 5, 2013, pp. 115-120.
2. Schmidt, W., Eindhoven University of Technology, 2014.
3. Schmidt, W.; Brouwers, H. J. H.; Khne, H.-C.; and Meng, B., Influences of superplasticizer modification and mixture composition on the performance of self-compacting
concrete at varied ambient temperatures, Cement and Concrete Composites, V. 49, 2014,
pp. 111-126. doi: 10.1016/j.cemconcomp.2013.12.004
4. Plank, J., and Hirsch, C., Impact of zeta potential of early cement hydration phases
on superplasticizer adsorption, Cement and Concrete Research, V. 37, No. 4, 2007, pp.
537-542. doi: 10.1016/j.cemconres.2007.01.007
52SP-302-03
5. Khayat, K. H., Viscosity-enhancing admixtures for cement-based materials An

overview, Cement and Concrete Composites, V. 20, No. 2-3, 1998, pp. 171-188. doi:
10.1016/S0958-9465(98)80006-1
6. Pourchez, J.; Govin, A.; Grosseau, P.; Guyonnet, R.; Guilhot, B.; and Ruot, B., Alkaline stability of cellulose ethers and impact of their degradation products on cement hydration, Cement and Concrete Research, V. 36, No. 7, 2006, pp. 1252-1256. doi: 10.1016/j.
7. Pourchez, J.; Peschard, A.; Grosseau, P.; Guyonnet, R.; Guilhot, B.; and Vallee, F.,
HPMC and HEMC influence on cement hydration, Cement and Concrete Research, V.
36, No. 2, 2006, pp. 288-294. doi: 10.1016/j.cemconres.2005.08.003
8. Marlire, C.; Mabrouk, E.; Faure, P.; Lamblet, M.; and Coussot, P., Advances in
Cement and Concrete Technology in Africa, Johannesburg, South Africa, 2013.
9. W. Schmidt, H. J. H. Brouwers, H.-C. Khne and B. Meng, Applied Rheology, 2013,
23.
10. Phyfferoen, A.; Monty, H.; Skaggs, B.; Sakata, N.; Yanai, S.; and Yoshizaki, M., First
North American Conference on the Design and Use of Self-Consolidating Concrete, 2002.
11. Sonebi, M., Rheological properties of grouts with viscosity modifying agents
as diutan gum and welan gum incorporating pulverised fly ash, Cement and Concrete
Research, V. 36, No. 9, 2006, pp. 1609-1618. doi: 10.1016/j.cemconres.2006.05.016
12. Sonebi, M.; Schmidt, W.; and Khatib, J.Chemistry and Materials Research, V. 5,
2013, pp. 106-111.
13. Schmidt, W.; Brouwers, H. J. H.; Kuehne, H.-C.; and Meng, B., Tenth International
Conference on Superplasticizers and Other Chemical Admixtures in Concrete, Prague,
Czech Republic, 2012.
14. Palacios, M.; Flatt, R.; Puertas, F.; and Sanchez-Herencia, A., 10th International
Conference on Superplasticizers and Other Admixtures in Concrete, Prague, Czech
Republic, 2012.
15. Hot, J., and Roussel, N., Fifth North American Conference on the Design and Use of
Self-Consolidating Concrete, Chicago, USA, 2013.
16. Hunger, M., and Brouwers, H. J. H., Flow analysis of waterpowder mixtures:
Application to specific surface area and shape factor, Cement and Concrete Composites,
V. 31, No. 1, 2009, pp. 39-59. doi: 10.1016/j.cemconcomp.2008.09.010
17. W. Puntke, beton, 2002, 2002, 242-248.
SP-302-04
New Additive to Enhance the Slump

Retention
by David Platel, Jean-Marc Suau, Clement Chosson,
and Yves Matter
For the Ready Mix Concrete mix design, the initial workability depends on the chemical
composition of the Polycarboxylate Ether (PCE). Up to now the use of best PCE can
achieve 2 hours of slump retention. However, a tendency to segregate is observed when
over-dosage is made, due to water reduction capability of PCEs. An approach is to use
also a combination of a water-reducing agent and a retarding agent which has the main
disadvantage to delay the setting time and consequently the early strength of the concrete.
This paper demonstrates the possibility to boost the performance of currently used PCEs.
The new slump retention additive that we developed allows a significant increase of the
slump retention while maintaining the initial fluidity without impacting the water reduction
ability. The homogeneity of the concrete is also controlled by using this additive. On top
of that, the combination between this new product and a standard water-reducting PCE is
made at commonly used dosage.
Keywords: polymer; slump; slump retention; water reduction; fluidity; cement; polycarboxylate ether; segregation.
INTRODUCTION
Since the development of the Polycarboxylate Ether (PCE), numerous studies1-12 have
been carried out which demonstrated that they are essential components of the concrete
admixtures when high fluidity and high strength are requested. For the retention of the
workability for a long period of time, typically more than 120 minutes, several solutions
based on the PCE technology have been developed for the ready-mix concrete industry
such as the formulation of the PCE with a retarding agent, or the PCE including a release
component13-16 or the new hyperbranched PCE using a cross-linked monomer.17 All these
previous technologies have been found to be effective on the control of the slump retention
but they also lead to some problems such as low early strength development or poor slump
retention capacity when the dosage in PCE is limited as in certain concrete mix designs.
In this paper, it is shown that the slump retention of the ready-mix concrete can be
controlled by the use of Polycarboxylate Ethers (PCEs) having low water reduction
capacity due to a poor dispersing ability. By combining macroscopic and microscopic
53
54SP-302-04
Table 1Physical properties and chemical compositions of cement

Chemical
composition
(%)
SiO2
Al2O3
Fe2O3
CaO
MgO
SO3
K2O
Na2O
Na2Oeq
C3S
C2S
C3A
C4AF
Blaine
fineness
(cm2/g)
C1
18.8
6.0
4.1
62.7
1.1
3.5
0.59
0.45
0.84
63.1
15.0
8.5
13.4
3750
Table 2Characteristics of the Polycarboxylate Ether
PCE 1
PCE 2
PCE 3
Function
High Water
Reducing
Slump
Retention
Slump
Retention
Brookfield
viscosity @
25C (mPa.s)
Mw (g/mol)
Ip
rh (nm)
Solid content
(%)
100 000
1.5
40
400
720 000
4.6
24
40
900
1 200 000
11.5
40
40
2 000
analysis techniques, the relationship between the molecular structure of the PCE and the
slump retention control is clarified.
Most of the studies regarding the slump retention properties of the PCE have been
developed one unique chemical structure which controls two key parameters of the fresh
concrete: the dispersibility and the slump retention. The right balance between these two
parameters is very complex. For example a PCE with a high water reduction capability
usually has a poor slump retention ability. This phenomenon is directly linked to the PCE
dosage added in the concrete mix design. This paper demonstrates the possibility to independently control the slump retention of a fresh concrete by combining two PCEs: a high
water-reducing agent and a slump retention agent.
EXPERIMENTAL AND ANALYTICAL INVESTIGATIONS
Cement
The cement used for all tests was an Ordinary Portland Cement (OPC) or CEM I 52.5
N according to the European Standard EN 197-1. The physical properties and chemical
analysis of the cement are listed in Table 1.
Polymers
Three Polycarboxylate Ether (PCE) were produced by free radical polymerization. The
PCE 1 is a high water reducing agent and its main characteristics are described in Table
2. The PCE 2 & PCE 3 differ from the PCE 1 by their low water reducing capacity and by
their high molecular weight (Mw) as described in Table 2.
Gel Permeation Chromatography (GPC)
The GPC analyses were performed using a chromatography system equipped with three
columns (Ultrahydrogel Waters) equilibrated at T = 55C in an aqueous solution (1%
New Additive to Enhance the Slump Retention 55
Table 3Concrete mixture proportions
Mix 1
Mix 2
Mix 3
W/C
0.69
0.54
0.67
Water reduction (%)

0%
22%
3%
Water
241 (406)
189 (318)
234 (394)
kg/m3 (lb/yd3)
Cement
Sand
350 (590)
860 (1449)
351 (590)
861 (1449)
352 (590)
862 (1449)
Gravel
990 (1517)
991 (1517)
992 (1517)
of KNO3) with a multi-detection system: a differential refractometer (Waters) and a

dual detector which combines a viscometer and a static light scattering detector at 90
(Malvern). The molecular weight calibration was performed with only one standard
from Polycal Malvern. Samples were injected at a concentration of (2-3) x 10-3 g/ml,
after filtration through a 0.2 m filter (Millipore).
Samples have been prepared following two different protocols. Concerning the Mw
measured for the PCEs (cf. Table 2), the samples were diluted at a concentration of (2-3) x
10-3 g/ml in an aqueous solution at 1% of KNO3 . This method is refered to as Reference.
For the second protocol, all the samples have been diluted at a concentration of (2-3) x 10-2
g/ml in different aqueous solutions at 1% of KNO3 which already contain a concentration
from 6 to 10% in weight of the following salts: Na2SO4, K2SO4, CaCl2, CaO, NaOH and
KOH. Then a sample of each previous preparation has been injected at a concentration of
(2-3) x 10-3 g/ml after filtration through a 0.2 m filter (Millipore) in order to measure the
evolution of Mw in different salt conditions over time.
Tests on concrete
All the concrete formulations were produced according to the European Standard EN
480-1. Natural sand and coarse aggregates (max diameter 31.5 mm) were used in all the
cases and the concrete mixture proportions are listed in Table 3. All the concrete mixtures
were prepared with the same initial slump values (190 230 mm, measured 5 minutes after
mixing the ingredients). Then slump values were measured at intervals of 45 minutes.
Cylinder specimens of 160 mm in diameter and 320 mm in length were produced and
investigated. All specimens have been prepared according to the European Standard EN
12930-2 in 160x320 mm steel cylinder molds. The strength properties were determined
after 1, 7 and 28 days of curing at 20C.
Cement pastes
For all the cement paste evaluations, the water cement ratio (w/c) was 0.38. It corresponds to a paste with the PCE 1 that has a spread diameter of about 180-200 mm.To
ensure that all experiments can be easily compared, all mixtures were prepared using the
same amount of materials (around 550g as shown in Table 4). Firstly, the PCE and the
water were pre-mixed together then the cement was added within 30s into the solution,
then a periodg of 30 s is given to the cement in order to be fully wet and finally hand mixing
was carried-out for 2 minutes.
Flow tests were performed 15 seconds after mixing on a ceramic tile having a smooth
surface. The cylinder used had a diameter of 60mm and a height of 50mm. The measurements of the initial paste flow and those made every 30 minutes during 120 minutes afterwards allow to plot the flow of the different PCEs over time as shown in Figure 4.
56SP-302-04
Table 4Cement paste formulations
Paste 1
W/C
0.38
PCE dosage (% dry

PCE/cement)
0.16%
g (lb)
Water
150 (0,331)
Cement
400 (0,881)
Table 5Water reduction of the PCE 1 & PCE 2

Concrete Mix Design
Polymer
PCE dosage (% dry
PCE/cement)
W/C
Water Reduction
DENSITY, kg/m3
(lb/yd3)
Initial Slump Flow,
mm (in.)
Initial Air Content, %
Mix 1
None
Mix 2
PCE 1
Mix 2
PCE 2
Mix 3
PCE 2
0,21
0,21
0,21
0,69
0%
0,54
22%
0,54
22%
0,67
3%
2287 (3855)
2301 (3878)
2281 (3845)
2296 (3870)
205 (8.1)
190 (7.5)
45 (1.8)
190 (7.5)
1.7
3.3
2.2
2.4

Water reduction capacity
The data in Table 5 show the water reduction capacity of the PCE 1 and PCE 2 which
have been realized according to the European Standards. The PCE 1 is a high water
reducing agent and in these conditions, its water reduction ability is 22%. In the same
conditions and at the same dosage, the PCE 2 showed clearly a weaker dispersing ability
which is measured by a poor initial slump of 45 mm instead of 190 mm for the PCE 1. The
water reduction of the PCE 2 at this polymer dosage was only 3% which clearly demonstrated that PCE 2 has no water reduction capability. This evaluation has also been carriedout on the PCE 3 and the water reduction was close to 1.5%. The high molecular weights
(Mw) of the PCE 2 and the PCE 3 have a strong impact on their dispersing capabilities by
comparison with the PCE 1 which has a lower Mw of 100 000 g/mol.
Slump retention and early strength development
Even if the PCE 1 has a strong water reduction capacity, data in Figure 1 show also
the weakness of the PCE 1 regarding the slump retention behavior. By using the concrete
mixture proportions (Mix 2) described in Table 3, the PCE 1 has a polymer dosage limited
to 0.21% of dry PCE/cement weight in order to reach an initial slump of 190 mm and a
good homogeneous concrete formulation. Above this polymer dosage, side effects appear
such as over-fluidity, bleeding, segregation, which typically lead to the production of
heterogeneous concrete materials. For example, with 0.24% of PCE1 per cement weight
instead of 0.21%, a severe segregation appears.
In Figure 1, the Mix 2 made with combinations of the PCE 1 and 0.1% of the PCE 2 or
0.1% of the PCE 3 show very good slump retentions. By increasing the polymer dosage
in PCE and by using the combination of a high water reducing PCE and a slump retention
PCE, no bleeding or no segregation have been observed for the two concretes made with
the PCE 2 or with the PCE 3.
Fig. 1Slump retention values of the PCE 1 (0.21%)

combined or not with a Slump Retention PCE.
Fig. 2Compressive strength development of the PCE 1

(0.21%) combined or not with a Slump Retention PCE.
In Figure 2, the measurement of the early age strength after one day and of the compressive strength at 28 days show that there is no retarding effect when using the PCE 2 or the
PCE 3, and there is no retarding effect by increasing the polymer dosage in this concrete
mix design. The incrementation of the slump retention PCE into a concrete mix design
formulated with a high water-reducing polymer allow to control the fluidity over time
without any side effects on the compressive strength development.
Cement pastes
In order to understand the mechanism of such slump retention PCEs, some cement paste
formulations have been studied according to the proportions detailed in Table 4. The use of
cement pastes as models instead of concrete mixture formulations allowed us to assess the
behavior of the PCEs over time at the microscale. In Figure 3, the cement paste prepared
with the PCE 1 alone showed a fluidity loss as it has been already observed on the concrete
58SP-302-04
Fig. 3Paste flow values of the PCE 1 (0.21%) combined or

not with a Slump Retention PCE.
Fig. 4Paste flow values of the PCE 1, PCE 2 & PCE 3.

tests. By adding the PCE 2 or the PCE 3 in combination with the PCE 1, no fluidity loss
has been observed for the two cement paste formulations. The fluidity retention behaviors
are similar for the formulations with the PCE 2 or the PCE 3 on the cement pastes or on the
concrete mixture formulations. All these results confirm that model cement pastes can be
used as a simplified model for the concrete formulation for this study.
Figure 4 shows some very interesting behavior of the PCEs. The PCE 1 has the best
initial fluidity and then a slump loss. The initial fluidity of the PCEs are related to their
molecular weight (Mw), consequently, the higher the initial fluidity, the lower the Mw.
Nevertheless, the paste flow evolutions are completely different between the PCEs which
increase for the PCE 2 or for the PCE 3 and which decreases for the PCE 1.The fluidity of
the PCE 2 is above the PCE 1 before 30 minutes and then is stable for the next 120 minutes.
For the PCE 3, the evolution of the paste flow is fast for 30 minutes like for the PCE 2, and
then increases slowly until 120 minutes. The paste fluidity of the PCE 3 is higher than that
of PCE 1 just before 90 minutes. These observations on the cement pastes demonstrate a
Fig. 5Mw values of the PCE 3 in different salt conditions

after storage at room temperature.
change of dispersing ability of the PCE 2 and the PCE 3 over time which can be related to
a modification of the polymer structure.
Molecular weight evolution
The molecular weight is one of the main characteristics which describs the Polycarboxylate Ether. Several studies11,12,18 have already been carried-out on the interactions between
the PCEs and the salts present in the cement or pore solution composition. Different ions
such as sulfates, calcium, sodium, potassium or the pH are well known to affect the polymer
structure and conformation of the PCEs. In Figure 5, the Mw of the PCE 3 as measured
by Gel Permeation Chromatography in an aqueous solution of KNO3 at 1wt% is around
1 200 000 g/mol and referred to named as Reference. Then the Mw values of PCE 3
measured in different salt solutions such Na2SO4, K2SO4, CaCl2, CaO, NaOH and KOH
show important variations. For Na2SO4, K2SO4 and CaCl2, the Mw values are similar to the
one measured in the standard GPC conditions. In basic pH range, the decrease of the Mw
is very significant during the workability period. These different measurements confirm
the change of polymer architecture of the PCE 3. In Figure 6, the Mw of PCE 1 does not
change in basic pH range on the contrary to the PCE 2 which shows a strong variation of
its Mw from 720 000 to 320 000 g/mol.
FURTHER RESEARCH
The results shown in this study explains the interactions between different kinds of
polymer structures and their behaviors in a cement paste or concrete formulation. All the
experimental tests have been carried-out with an Ordinary Portland Cement. Consequently,
further developments have to be done on blended cements with various ionic concentrations in order to establish if the modifications of the polymer structures are similar.
60SP-302-04
Fig. 6Comparison of the Mw values for the PCE 1, PCE 2

& PCE 3 in standard and basic pH conditions
CONCLUSIONS
According to the different experimental investigations carried-out in this study, the high
molecular weight PCEs are able to control the slump retention from the cement paste to
the concrete mixutre. No increase of fluidity, no retarding effect and no bleeding have been
observed by using a combination of a slump retention PCE and a standard high waterreducing PCE.
AUTHOR BIOS
David Platel is a R&D Manager of the Construction Laboratory at the Coatex SAS,
France. He received his PhD from UPMC in Paris, France. His research interests include
the development of new building chemicals and especially the chemistry & physicochemistry of the polymers in hydraulic binder formulations.
Jean-Mars Suau is a Research Manager and is in charge of the Synthesis and Pilot
Laboratory at Coatex SAS, France. His research interests include the Controlled Radical
Polymerization, the Hydrophobically Alkali Swellable Emulsion, the Hydrophobically
Ethoxylated Urethane, and the Polycarboxylate Ether. All theses polymers are water
soluble or water compatible and they are rheological additives dedicated to industrial
areas such as Construction, Paint, Paper, Cosmetic
Clement Chosson is a R&D Manager of the Analysis and the Quality Control Laboratories at the Coatex SAS, France. He received his MS from Claude Bernard University in
Lyon, France. His research interests include the advanced characterization of polymers.
Yves Matter is a Research Scientist at Coatex SAS. He received his PhD from Karlsruhe
Institute of Technology. His research interests include polymer synthesis and physicochemical characterization. He is in charge of the development of new macromolecules
used as concrete admixures.
ACKNOWLEDGMENTS
The authors wish to express their gratitude and sincere appreciation to Benoit Magny,
R&D Director of Coatex group, for helpful support and discussions.
rh
NOTATION
= hydrodynamic radius
REFERENCES
1. Kirby, G. H., and Lewis, J. A., Comb polymer architecture effects on the rheological
property evolution of concentrated cement suspensions, Journal of the American Ceramic
Society, V. 87, No. 9, 2004, pp. 1643-1652. doi: 10.1111/j.1551-2916.2004.01643.x
2. Yamada, K., and Hanehara, S., Working mechanism of polycarboxylate superplasticizer considering the chemical structure and cement characteristics, Proceedings of the
11th International Congress on the Chemistry of Cement. 2003.
3. Magarotto, R.; Torresan, I.; and Zeminian, N., Influence of the molecular weight
of polycarboxylate ether superplasticizers on the rheological properties of fresh cement
pastes, mortar and concrete, Proceedings of the 11th International Congress on the Chemistry of Cement. 2003.
4. Flatt, R. J., Polymeric dispersants in concrete, Polymers in particulate systems properties and applications, Hackley, V.A., Somansundaran, P., and Lewis, J.A., Editors. 200,
p. 247-294.
5. Yoshioka, K. etal., Adsorption characteristics of superplasticizers on cement component minerals, Cement and Concrete Research, V. 2056, 2002, pp. 1-7.
6. Yamada, K.; Ozu, H.; and Yano, M., Prevention of incompatibility phenomena
between cement and superplasticizer by blending several types of polycarboxylates polymers, Proceedings of 1st fib congress, 2002, 143: p. 16-26.
7. Yamada, K., and Hanehara, S., Interaction mechanism of cement and superplasticizers - The roles of polymer adsorption and ionic conditions of aqueous phase, Concrete
Science and Engineering, V. 3, 2001, pp. 135-145.
8. Platel, D., Impact of polymer architecture on superplasticizer efficiency, Proceedings of 9th ACI Conference. 2009, p 381-394.
9. Yamada, K., A summary of important characteristics of cement and superplasticizer,
Proceedings of 9th ACI Conference. 2009, p 85-96.
10. Flatt, R. J. et al., Polymer physics and superplasticizers, Proceedings of 9th ACI
Conference. 2009, p 113-122.
11. Flatt, R. J. et al., The role of adsorption energy in the sulfate-polycarboxylate
competition, Proceedings of 9th ACI Conference. 2009, p 153-164.
12. Zimmermann, J. et al., Effect of polymer structure on the sulfate-polycarboxylate
competition, Proceedings of 9th ACI Conference. 2009, p 165-175.
13. Gller, F, et al., The relationship between retention stability and chemical structure, Proceedings of 9th ACI Conference. 2009, p 249-260.
14. Izumi, T. et al., A new hybrid type superplasticizer, Proceedings of 7th ACI Conference. 2003 supplementary papers, p 67-81.
15. Cerulli, T. et al., Superplasticizers for extending workability, Proceedings of 8th
ACI Conference. 2006 supplementary papers, p 263-277.
62SP-302-04
16. Hamada, D. et al., Development of slump-loss controlling agent with minimal

setting retardation, Proceedings of 7th ACI Conference. 2003, p 127-142.
17. Hamada, D. et al., Development of new Superplasticizer providing ultimate workability, Proceedings of 8th ACI Conference. 2006, p 31-50.
18. Borget, P.; Galmiche, L.; Le Meins, J.-F.; and Lafuma, F., Microstructural characterisation and behaviour in different salt solutions of sodium polymethacrylate-g-PEO
comb copolymers, Colloids and Surfaces A: Physicochemical and Engineering Aspects,
V. 260, No. 1-3, 2005, pp. 173-182. doi: 10.1016/j.colsurfa.2005.03.008
SP-302-05
Synthesis of a Novel Superplasticizer

Prepared from Brown Coal
by Manuel Ilg and Johann Plank
Polycondensates and polycarboxylates are known to be effective superplasticizers. Here,
the synthesis of a novel, brown coal based superplasticizer by grafting acrylic acid (AA)
and 2-acrylamido-2-tert.butyl sulfonic acid (ATBS) onto the alkali soluble components of
brown coal (lignite) as backbone using free radical copolymerization technique is described.
Furthermore, an ATBS-acrylic acid copolymer was synthesized to investigate the influence
of the graft chains on the performance of the lignite copolymer. Successful grafting was
confirmed by size exclusion chromatography (SEC) and comparison of the adsorbed layer
thicknesses of the brown coal substrate and the grafted product. The dispersing performance of the graft copolymer was probed via mini slump tests and compared with that
of BNS. Additionally, slump flow retention and sulfate tolerance were determined. It was
found that the graft copolymer possesses higher dispersion effectiveness than BNS and
exhibits high sulfate tolerance.
Keywords: admixture; adsorbed layer thickness; brown coal; graft copolymer; lignite;
superplasticizer.
INTRODUCTION
Superplasticizers are an important part of modern concrete technology to improve the
workability of fresh concrete. With the help of these admixtures the water-to-cement ratio
can be reduced, thus obtaining cementitious building materials with higher strength and
durability.1,2 Although polycondensates and polycarboxylates represent very effective
polymers, new structures and concepts are still investigated.3 In recent years, natural polymers became increasingly popular as a starting material for the synthesis of new admixtures.4 A main reason is that natural polymers offer novel structural motifs and often bear a
high diversity of functional groups. In literature, a number of examples already exist where
through functionalization and modification of natural polymers, novel superplasticizers
have been synthesized.5,6 Ordinary brown coal (often referred to as lignite) also presents
a promising candidate for the development of new superplasticizers, because of its global
abundance and its low cost.
Brown coal is a combustible sedimentary rock that was built million years ago from dead
plant material through a process called coalification.7 In flooded areas such as swamps
63
64SP-302-05
the main constituents of plant residues, lignin and cellulose, were decomposed gradually
to humic substances (peat). Burial by sediments and subsequent tectonic pressure over
millions of years led to an increase of the temperature causing compaction of the decomposition products. In this process, peat was successively metamorphosed into brown coal
under release of water and gases such as methane and carbon dioxide.7
In this study, a caustic extract of brown coal containing humic and fulvic acids as
main constituents was utilized.8-10 Such alkali extracts are commercially available and
widely used in different industries including agriculture (as fertilizer)10 and in oil well
drilling as drilling fluid (the so-called lignite mud).11 However, up to date no brown
coal based concrete superplasticizer has been developed. Here, we report on the synthesis
of a novel superplasticizer prepared by grafting acrylic acid and ATBS monomers onto
the alkali soluble components of brown coal. Generally, humic and fulvic acids possess a
large number of functional groups (i.e. aliphatic/aromatic carboxylate, carbonyl, amino,
phenolic and ketone groups) which present a perfect substrate for graft copolymerization
reactions.12,13
The synthesized brown coal-ATBS-acrylic acid graft copolymer was characterized with
respect to its molecular properties (Mw, Mn) and its anionic charge amount. Its dispersing
effectiveness and slump retention behavior was compared with that of an industrial grade
-naphthalene sulfonate (BNS) sample. Additionally, the sulfate tolerance of the graft
copolymer was studied and heat flow calorimetry measurements were performed to assess
its effect on cement hydration. To prove successful grafting, the adsorbed layer thicknesses
of the brown coal substrate and the graft copolymer were measured via dynamic light
scattering utilizing cationic polystyrene nanoparticles as adsorbent.14 Finally, the working
mechanism of the superplasticizer was clarified via zeta potential measurements.
A novel brown coal based superplasticizer has been synthesized using graft copolymerization technique. From a commercial brown coal the alkali soluble fraction (mainly humic
and fulvic acids) was extracted and utilized as backbone. Up to date, coal and natural gas
are the only readily available alternatives to petroleum. Therefore, the different constituents of brown coal could be an alternative feedstock in the production of superplasticizers
to save the more precious resources of petroleum. The overall goal of this study was to
validate new structural motifs and to offer a synthetic route to highly economical, brown
coal based superplasticizers.
Materials
Brown coal The used brown coal was mined in the Lusatia brown coal mining district
near Cottbus, Germany.
Chemicals 2-Acrylamdio-2-tert.butyl sulfonic acid (ATBS), acrylic acid (AA), sodium
hydroxide (NaOH), sodium peroxodisulfate (Na2S2O8), sodium pyrosulfite (Na2S2O5) and
EDTA were used as per obtained.
Industrial superplasticizer sample A spray dried powder BNS superplasticizer was used
as an industrial grade reference sample.
Synthesis of a Novel Superplasticizer Prepared from Brown Coal 65
Table 1 Phase composition of the CEM I 52.5 N sample determined by

Q-XRD using Rietveld refinement.
Phase
C3S, m
C2S, m
C3A, c
C3A, o
C4AF, o
Free Lime (Franke)
Periclase (MgO)
Anhydrite
Hemihydrate*
Dihydrate*
Calcite
Quartz
Arcanite (K2SO4)
wt.%
54.14
26.63
3.28
4.26
2.45
0.10
0.03
2.64
1.21
0.02
3.61
1.16
0.46
* determined by thermogravimetry
Fig. 1 Scheme for extraction of alkali soluble components

from brown coal.
Cement The cement used for this study was an ordinary Portland cement CEM I 52.5 N.
Its phase composition as determined by X-ray diffraction and subsequent Rietveld refinement is illustrated in Table 1. Its specific surface area was 3,583 cm2/g (Blaine method)
and its particle size (d50 value) was 11.5 m [45.3 10-5 in.] (laser granulometer). The
density as obtained by helium pycnometry was 3.15 g/cm3.
Extraction of brown coal
The individual steps performed in the extraction process are schematically presented in
Fig. 1. First, chunks of brown coal were coarsely crushed with a hammer and the residue
was sieved to < 250 m [9.84 10-3 in.] to obtain a brown coal powder exhibiting a narrow
66SP-302-05
particle size distribution and a high specific surface area. For extraction of the alkali soluble
components, 70 g [2.47 oz.] of the brown coal powder were mixed with 700 mL [23.7 fl.
oz.] of 0.5 M NaOH placed in a 1 L [33.8 fl. oz.] round-bottom flask.9 The mixture was
refluxed for three hours at 90 C [194 F] under constant stirring. Next, the dark solution
was cooled to ambient temperature and centrifuged two times for 10 minutes at 8,500 rpm.
The supernatant was separated from the residue and freeze dried, yielding 21.7 g [0.765
oz.] of a black solid (theor. yield 31%). It is noteworthy that many coal producers offer
such extracts in liquid or powder form designated as caustic lignite.
Synthesis of brown coal-ATBS-acrylic acid graft copolymer
The novel brown coal-ATBS-acrylic acid graft copolymer was synthesized by aqueous
free radical copolymerization using sodium peroxodisulfate as initiator. Here, 2-acrylamido-2-tert.butyl sulfonic acid (ATBS) and acrylic acid were grafted onto the extracted
alkali soluble components (i.e. humic and fulvic acid) of brown coal. The molar ratio
between ATBS and acrylic acid was 1: 0.15 and the weight ratio between the brown coal
and the grafted monomers was 20: 80 (wt/wt). In a five necked, 1 L [33.8 fl. oz.] roundbottom flask equipped with stirrer, thermometer, reflux condenser and inlet for N2 gas,
13.2 g [0.466 oz.] of the dry alkali soluble components were dissolved in 206 mL [6.97 fl.
oz.] DI water. Thereafter, 8.8 g [0.31 oz.] NaOH pellets were added to adjust a pH of 12
and then cooled to 18 C [64 F]. Next, 50.0 g [1.76 oz.] ATBS (241 mmol, 1.0 eq) were
dissolved stepwise and the temperature was kept constantly under 25 C [77 F] to avoid
homopolymerization of ATBS. 2.60 g [0.092 oz.] acrylic acid (36.1 mmol, 0.15 eq), 600
mg [0.021 oz.] EDTA and 1.00 g [0.035 oz.] of an organo-modified polysiloxane defoamer
were added and the mixture was purged with N2 for 1 h at room temperature. After heating
to 50 C [122 F] the first portion of sodium peroxodisulfate initiator was added (8 g [0.282
oz.]) and the reaction mixture was stirred for 50 minutes at this temperature. The second
part of initiator was added (8 g [0.282 oz.]) and the polymerization continued for additional
70 minutes at 50 C [122 F]. Then the temperature was increased to 60 C [140 F] and
kept there for 1 h. Finally, the reaction flask was heated to 80 C [176 F], 3.60 g [0.127
oz.] of sodium pyrosulfite were added to quench remaining radicals and stirring continued
for 1 h at 80 C [176 F]. The solution was cooled to room temperature to obtain a viscous,
dark brownish polymer solution with a solid content of 29 wt.% and a pH of 2.5. The
polymer solution was used without any further purification.
Synthesis of ATBS-acrylic acid copolymer
An ATBS-acrylic acid copolymer was synthesized according to the procedure described
above except that no alkali soluble components were present. Here, a pale yellowish, 27.6
wt.% aqueous solution with low viscosity and a pH of 2.5 was obtained. This polymer was
used for comparison.
Characterization of polymers
Size exclusion chromatography Molecular weights (Mw and Mn) and polymer radii
(Rh(z) and Rg(z)) of all synthesized polymers were determined by size exclusion chromatography. The instrument is equipped with a RI detector and an 18 angle dynamic light
scattering detector. Polymer solutions exhibiting a concentration of 2 g/L were prepared
for the SEC analysis. The polymers were separated on a precolumn and two columns using
0.2 M NaNO3 solution (adjusted with NaOH to pH 9) as an eluent at a flow rate of 1.0 mL/
min. The value of dn/dc used to calculate Mw and Mn was 0.218 mL/g (value for lignin).15
Anionic charge amount of the polymers The anionic charge of the polymers was determined via polyelectrolyte titration using a particle charge detector. 0.001 M cationic polydiallyl dimethyl ammoniumchloride (polyDADMAC) solution was employed as titrator.
Polymer solutions with a concentration of 0.1 g/L were prepared in DI water, in 0.1 M
NaOH and in cement pore solution (CPS). Cement pore solution was freshly prepared
by vacuum filtration of neat cement slurries using a water-to-cement ratio of 0.455. In a
typical experiment, 10 mL [0.34 fl. oz.] of the polymer solution were pipetted into a PTFE
cylinder with an oscillating PTFE piston in the center. The dissolved polymers can adsorb
via Van der Waals forces on the surface of the cylinder and the piston. Because of the oscillating movement of the piston, counter ions are removed from the immobilized polymers
and a streaming current results which can be measured by two platinum electrodes located
within the PTFE cylinder. The polyDADMAC solution was titrated until the isoelectronic
point was reached. For every polymer sample the measurement was repeated three times
and the values were averaged. From the consumption of polyDADMAC the amount of
negative charge per gram of polymer was calculated.
Heat flow calorimetry To investigate the influence of the synthesized polymers on
cement hydration, isothermal heat flow calorimetric measurements were carried out. There,
4 g [0.141 oz.] cement were filled into 20 mL [0.68 fl. oz.] glass ampoules and mixed with
the respective amount of aqueous polymer solution to obtain a water-to-cement ratio of
0.455. The ampoules were sealed, homogenized for 1 min in a wobbler and then placed
into the isothermal conduction calorimeter. Data logging was continued until heat evolution from the hydration reaction subsided completely.
Performance of the synthesized polymers
Mini slump test The dispersing effectiveness of the synthesized polymers was assessed
utilizing a mini slump test following in principle DIN EN 1015, but with some modifications. At first, the water-to-cement ratio required to reach a slump flow of 18 0.5 cm [7.1
0.2 in.] was established for the cement paste without polymers. At this specific waterto-cement ratio, the dosage of the polymers was determined to attain a spread flow of 26
0.5 cm [10.2 0.2 in.]. In a typical experiment, the superplasticizer was dissolved in the
required amount of mixing water placed in a porcelain cup. The amount of water contained
in the polymer solution was subtracted from the amount of mixing water. 300 g [10.6 oz.]
of cement were added to the mixing water over a period of 1 min, then rested for 1 min
and subsequently were stirred manually for 2 min with a spoon. Immediately after the end
of stirring, the cement slurry was poured into a Vicat cone (height 40 mm [1.57 in.], top
diameter 70 mm [2.76 in.], bottom diameter 80 mm [3.15 in.]) placed on a glass plate, filled
to the brim and the cone was lifted vertically. The resulting paste spread was measured
twice, the second measurement being perpendicular to the first one and averaged to obtain
the slump flow value.
Time dependent mini slump test Development of dispersing performance over time
was investigated via time dependent mini slump testing. Here, 400 g [14.1 oz.] cement
were mixed with the required amount of water and polymer to achieve an initial slump flow
68SP-302-05
of 26 0.5 cm. The procedure was the same as described above. After each measurement
the cement paste was transferred back into the porcelain cup and covered with a wet towel
to avoid desiccation. Prior to each measurement the cement paste was vigorously stirred
for two minutes. Measurements were conducted every 15 minutes over a total period of
120 minutes.
Adsorbed layer thickness
The adsorbed layer thickness of the polymers was captured by dynamic light scattering.
Here, monodisperse polystyrene nanoparticles exhibiting an average particle size of 75.5
0.5 nm [2.97 10-6 in.] were utilized as adsorbent. Starting from a 150 mg/L stock solution of each polymer in 0.1 M NaOH, different concentrations were prepared by dilution
(diluent 0.1 M NaOH). Prior to the measurements, the solutions were filtered through a 0.2
m [7.9 10-6 in.] filter to remove undesired dust particles that can disturb measurements
because of their high scattering intensity. Next, 50 L [1.7 10-3 fl. oz.] of a suspension
of cationic polystyrene nanoparticles (for preparation see14) were added and sonicated for
5 min. The polymer solution was filled into a glass cuvette and then placed in the instrument. Every measurement was repeated 150 times per sample and the average value was
calculated. Each test consisted of a 10 s light scattering run taken at a temperature of 25
C [77 F]. Measurements were continued at increasing polymer concentrations until a
stable, final value was reached that was regarded as the point of saturated adsorption. The
adsorbed layer thickness was calculated using equation 1.

adsorbedlayerthicknessnm=dadsnm-dpolystyrene[nm]2(1)
where dads represents the particle size of the polystyrene nanoparticle holding adsorbed
polymer, and dpolystyrene represents the particle size of the native polystyrene particle.
Zeta potential measurement
Zeta potential measurements were performed at room temperature on an electro acoustic
spectrometer. This instrument measures a vibration current induced by an acoustic wave
which causes the aqueous phase to move relative to the cement particles. From that, a
potential difference results which can be measured and designated as zeta potential. Immediately after mixing, the freshly prepared cement slurries holding the respective dosage of
the polymer samples required for a slump flow of 26 0.5 cm were filled into the cup of the
instrument and measured under continuous stirring for a total period of 30 min.
Characterization of brown coal-ATBS-acrylic acid graft copolymer
The synthesized graft copolymer and the ATBS-acrylic acid copolymer were characterized using SEC. The molar masses and polymer radii are listed in Table 2. According
to these data, the brown coal-ATBS-acrylic acid graft copolymer exhibits a significantly
higher molar mass (Mw ~ 443,300 g/mol) than the ATBS-acrylic acid copolymer (Mw ~
183,300 g/mol). This result is a first indication that grafting was successful. Furthermore
it was observed that the hydrodynamic and gyration radii increased from 14.5 nm [5.71
10-7 in.] and 23.9 nm [9.41 10-7 in.] respectively for the ATBS-acrylic acid copolymer to
Table 2 Molar masses, polydispersity index (PDI) and polymer radii of the
synthesized polymers and of BNS as reference superplasticizer sample.
Polymer
Brown coal-ATBS-AA
ATBS-AA copolymer
BNS
Mw [g/mol]
443,300
183,300
140,000*
Mn [g/mol]
216,100
93,460
-
PDI
2.0
2.0
-
Rg(z) [nm]
37.2
23.9
-
Rh(z) [nm]
22.1
14.5
-
* = batch measurement
Fig. 2 SEC spectrum of the brown coal-ATBS-AA graft

copolymer.
Table 3 Anionic charge amounts of the polymer samples in DI water, CPS
and 0.1 M NaOH solution.
Polymer
Brown coal-ATBS-AA
ATBS-AA copolymer
BNS
DI water [eq/g]
3,347
4,241
3,643
CPS [eq/g]
3,964
3,974
2,713
0.1 M NaOH [eq/g]

4,765
4,855
3,989
22.1 nm [8.70 10-7 in.] and 37.2 nm [1.46 10-6 in.] resp. after the grafting process. In
comparison, BNS possesses a much lower molar mass (Mw ~ 140,000 g/mol) than the graft
copolymer. In summary, the graft copolymerization process produces a relatively homogeneous graft copolymer that exhibits a quite narrow polydispersity index of ~ 2.0 (see SEC
spectrum of graft copolymer in Fig. 2).
Next, the anionic charge amounts of the polymers were determined in DI water, cement
pore solution (CPS) and in 0.1 M NaOH solution. The results are summarized in Table
3. Generally, the graft copolymer possesses a highly anionic character (charge amount
e.g. in CPS 3,964 eq/g), owed to the carboxylate and sulfonate groups present in the
graft chains. Such high anionic charge promotes adsorption of the graft copolymer onto
positively charged surfaces of cement. The ATBS-acrylic acid copolymer exhibits an even
higher anionic charge while BNS possesses a much lower anionic charge than the brown
coal based graft copolymer and the ATBS-acrylic acid copolymer.
70SP-302-05
Fig. 3 Proposed chemical structure of the brown coal-ATBS-AA graft copolymer.

For the chemical structure of the graft copolymer, the model as follows was developed:
Humic acid which presents the main component in the alkali extract possesses numerous
functional groups (i. e. phenolic and carboxylate groups) in its structure which provide
perfect docking sites for the grafting reaction. After addition of an initiator (e.g. sodium
peroxodisulfate), macroradicals can form through abstraction of hydrogen from these functional groups. Consequently, ATBS and acrylic acid monomers can be grafted onto these
free radical sites. Simultaneously, side chain propagation can continue. As a result, a graft
copolymer is formed that is composed of a humic acid backbone and grafted ATBS-coacrylic acid side chains. A structural model of the graft copolymer is proposed in Fig. 3.
Note that for humic acid, only model structures exist16 due to complexity of composition and abundant natural variants. Generally, humic acid contains a number of condensed
aromatic rings, and thus attains a relatively stiff, linear conformation. In contrast to this,
the ATBS-co-acrylic acid graft chains exhibit high conformational flexibility and are coiled
in solution. These differences in solution conformation were confirmed by the Burchard
parameters (ratio of Rg(z)/Rh(z)). There, a value of 1.7 was obtained for the graft copolymer
which represents a linear, stretched random coil.17
Cement dispersion
The dispersing performance of the brown coal based graft copolymer was determined
using a mini slump test. Here, the dosages were established to reach a paste flow of 26
0.5 cm. The water-to-cement ratio of the neat cement paste without superplasticizer was
set to produce a slump flow of 18 0.5 cm (w/c ratio of 0.455). This water-to-cement
Fig. 4 Time dependent development of the slump flow

of a cement slurry (w/c = 0.455) containing 0.21% bwoc
of brown coal-ATBS-AA graft copolymer, 0.19% bwoc of
ATBS-AA copolymer and 0.3% bwoc of BNS respectively.
value was applied for all measurements. For the brown coal-ATBS-acrylic acid graft copolymer a dosage of 0.21% bwoc was required to reach the desired slump flow of 26 0.5
cm. Thus, the novel graft copolymer was even more effective than a commercial BNS
reference sample (dosage 0.30% bwoc). Additional mini slump testing carried out for the
ATBS-acrylic acid copolymer revealed that this polymer was slightly better (dosage 0.19%
bwoc) than the graft copolymer itself. The results suggest that the dispersing effect mainly
originates from the ATBS-co-acrylic acid graft chains. This was confirmed further via mini
slump tests, evidencing that the alkaline brown coal extract does not disperse cement.
Furthermore, the time dependent slump loss behavior of a cement paste prepared at a
water-to-cement ratio of 0.455 containing the polymers at dosages required for a slump
flow of 26 0.5 cm was investigated. The results of these measurements are displayed in
Fig. 4. Starting from an initial slump flow of 26 0.5 cm, fluidity was monitored every 15
min for a period of 120 min. According to Fig. 4, the dispersing performance of the brown
coal-ATBS-acrylic acid graft copolymer quickly decreases in the first 30 min (slump flow
21.4 cm [8.43 in.]). Afterwards, the decrease becomes very slow. In contrast, the ATBSacrylic acid copolymer exhibits better slump retention and behaves more similar than the
commercial BNS superplasticizer.
Sulfate tolerance of graft copolymer
The presence of alkali sulfates (i.e. K2SO4, Na2SO4) in cement can have a significant
impact on the dispersing performance of superplasticizers. This phenomenon (the so-called
sulfate effect) has been observed mainly for polycarboxylates. In literature, two mechanisms are discussed for the negative impact of sulfate on PCEs: competitive adsorption
between sulfate ions and PCEs18 and shrinkage of PCE molecules.19 To investigate whether
the brown coal based graft copolymer is also affected by sulfate ions, mini slump tests at
increasing additions of sodium sulfate were carried out. The graft copolymer was applied
at a dosage of 0.21% bwoc (slump flow 26 0.5 cm). The results are illustrated in Fig. 5.
72SP-302-05
Fig. 5 Effect of different Na2SO4 dosages on the slump

flow of cement pastes (w/c = 0.455) holding different
polymer samples.
There, it was observed that the graft copolymer is not much affected by the presence of
different concentrations of sulfate, thus confirming high sulfate tolerance for the grafted
product. In contrast, the ATBS-acrylic acid copolymer shows strong sensitivity to sulfate.
At increasing Na2SO4 additions, its initial slump flow of 26.3 cm [10.4 in.] quickly drops
to 20.9 cm [8.22 in.]. Furthermore, a physical blend of 20 wt.% brown coal extract and 80
wt.% of the ATBS-acrylic acid copolymer (as present in the graft copolymer) was tested.
It was found that this physical mixture exhibits the same poor sulfate tolerance than the
ATBS-acrylic acid copolymer, thus providing further evidence that in fact a chemical reaction had occurred between the brown coal substrate and the monomers. A potential explanation for the different behaviors of the brown coal-ATBS-acrylic acid graft copolymer and
the ATBS-acrylic acid copolymer are the different molar masses of the polymers. The graft
copolymer possesses a higher Mw than the copolymer, so its adsorption is mainly driven
by entropic effects (desorption of a huge amount of ions and water molecules adsorbed
on the surface of cement), thus producing a high Gibbs energy of adsorption.20 Whereas
sulfate ions, whose energy of adsorption results from an enthalpic contribution only and is
comparatively low cannot displace already adsorbed polymers from the surface of cement.
Adsorbed layer thickness
Additional experiments were carried out to prove successful grafting of ATBS and acrylic
acid monomers onto the alkaline extract of brown coal. For this purpose, the adsorbed
layer thicknesses of the graft copolymer, the ATBS-acrylic acid copolymer and the brown
coal extract were measured and compared. Monodisperse, spherical cationic polystyrene nanoparticles were taken as adsorbent and layer thicknesses were determined using
dynamic light scattering. This method allows facile determination of the adsorbed layer
thickness of negatively charged polyelectrolytes at high pH conditions such as in cement
pore solution.14 Layer thicknesses were measured as a function of polymer concentration
until the point of saturated adsorption was reached. The results are displayed in Fig. 6. The
alkaline extract of brown coal reaches the point of saturated adsorption at an adsorbed layer
thickness of ~ 2.5 nm [9.8 10-8 in.] only whereas the brown coal-ATBS-acrylic acid graft
Fig. 6 Concentration-dependent adsorbed layer thicknesses

of the brown coal-ATBS-AA graft copolymer, of ATBS-AA
copolymer, of the alkaline brown coal extract and BNS.
polymer exhibits a substantially higher layer thickness of ~ 6.4 nm [2.5 10-7 in.]. Contrary
to this, the ATBS-acrylic acid copolymer produces an adsorbed layer thickness of ~ 1.7 nm
[6.7 10-8 in.] only. These values signify that grafting of the monomers onto the brown coal
substrate has indeed occurred, and that the graft product possesses pendants of ATBS-coacrylic acid. However, BNS shows a very low adsorbed layer thickness of only ~ 0.3 nm.
Effect on cement hydration
The influence of the superplasticizers on cement hydration was tested by means of
isothermal heat flow calorimetry. The heat evolution from cement hydration was monitored
for cement pastes prepared at a water-to-cement ratio of 0.455 holding 0.25% bwoc of the
brown coal-ATBS-acrylic acid graft copolymer, of ATBS-acrylic acid copolymer and of
BNS respectively. The results are illustrated in Fig. 7. It was observed that the brown coalATBS-acrylic acid graft copolymer causes very minor retardation, apparently caused by
the ATBS-co-acrylic acid pendant groups.
Mechanism of dispersion
Adsorption of superplasticizers on cement particles can be tracked via zeta potential
measurements. Here, the zeta potential of cement slurries holding different polymer
samples was measured. Concentrations of the polymer samples were those required for a
slump flow of 26 0.5 cm. The neat cement paste exhibited a slightly negative zeta potential of 3.3 mV. After addition of the polymer samples, the zeta potentials of all cement
slurries decreased to similar values: brown coal-ATBS-acrylic acid graft copolymer 28.0
mV; ATBS-acrylic acid copolymer 27.3 mV and BNS 29.3 mV. These results signify
that the brown coal-ATBS-acrylic acid graft copolymer adsorbs and also achieves dispersion through an electrostatic repulsion effect.
74SP-302-05
Fig. 7 Isothermal heat flow calorimetry of cement slurries

(w/c = 0.455) holding 0.25% bwoc of brown coal-ATBSAA graft copolymer, ATBS-AA copolymer and BNS
respectively.
CONCLUSIONS
An alkali brown coal extract holding humic and fulvic acids was successfully used as
substrate for the synthesis of a brown coal based superplasticizer. Acrylic acid and ATBS
were successfully grafted onto the extracted lignite, as was confirmed via SEC analysis
and measurements of the adsorbed layer thicknesses of the graft copolymer and an ATBSacrylic acid copolymer prepared under comparable conditions. A structural model for the
novel graft copolymer suggests that humic/fulvic acid backbones hold grafted chains of
ATBS-acrylic acid copolymer. The synthesized brown coal-ATBS-acrylic acid graft copolymer presents an effective cement dispersant which is superior over industrial BNS and
exhibits excellent sulfate tolerance. Its working mechanism relies on a combination of a
strong electrostatic and a minor steric effect.
The study shows that caustic lignite extracts present interesting and low-cost starting
materials for the development of novel superplasticizers holding unique structural motifs.
Future research should focus on the replacement of the ATBS monomer by less expensive
alternatives such as e.g. itaconic, methacrylic, styrene sulfonic, allyloxy hydroxy propyl
sulfonic acid or esters thereof. Due to the chemical variability of brown coal and the number
of natural variants, different brown coal types should be tested as well. We consider our
work as an initial study which may stimulate further ideas to exploit the potential of this
concept, especially in countries where more sophisticated monomers are not available.
AUTHOR BIOS
Manuel Ilg studied chemistry and received his B.Sc. and M.Sc. from Technische Universitt Mnchen. He is currently a Ph.D. student at the Chair for Construction Chemistry in
Garching where he works on new structural concepts for superplasticizers.
Johann Plank is a full Professor at the Institute of Inorganic Chemistry of Technische

Chemistry there. His research interests include cement chemistry, concrete admixtures,
cementing.
ACKNOWLEDGEMENTS
The authors would like to thank Vatenfall Europe Mining AG for providing the brown
coal samples and Lubrizol for the supply of the ATBS monomer.
REFERENCES
1. Ramachandran, V. S.; Malhotra, V. M.; Jolicoeur, C.; and Spiratos, N., Superplasticizers: Properties and applications in concrete, CANMET, Ottawa, Canada, 1998.
2. Spiratos, N., Page, M., Mailvaganam, N. P., Malhotra, V. M., Jolicoeur, C., Superplasticizers for Concrete: Fundamentals, Technology, and Practice, Supplementary Cementing
Materials for Sustainable Development, Ottawa, Canada, 2003.
3. Plank, J.September 2012 , PCE Superplasticizers Chemistry, Applications and
Perspectives, 18. ibausil, Weimar, Germany, V. 1, pp. 91-102.
4. Plank, J.2003 , Applications of Biopolymers in Construction Engineering, Biopolymers, V. 10, pp. 29-95.
5. Lv, S.; Gao, R.; Cao, Q.; Li, D.; and Duan, J.2012 , Preparation and characterization
of poly-carboxymethyl--cyclodextrin superplasticizer, Cement and Concrete Research,
V. 42, No. 10, pp. 1356-1361. doi: 10.1016/j.cemconres.2012.06.006
6. Zhang, D. F.; Ju, B. Z.; Zhang, S. F.; He, L.; and Yang, J. Z.2007 , The study on the
dispersing mechanism of starch sulfonates as a water-reducing agent for cement, Carbohydrate Polymers, V. 70, No. 4, pp. 363-368. doi: 10.1016/j.carbpol.2007.04.024
7. Hatcher, P. G., and Clifford, D. J.1997 , The organic geochemistry of coal: from
plant materials to coal, Organic Geochemistry, V. 27, No. 5/6, pp. 251-274. doi: 10.1016/
S0146-6380(97)00051-X
8. Ashida, R.; Morimoto, M.; Makino, Y.; Umemoto, S.; Nakagawa, H.; Miura, K.; Saito,
K.; and Kato, K.2009 , Fractionation of brown coal by sequential high temperature solvent
extraction, Fuel, V. 88, No. 8, pp. 1485-1490. doi: 10.1016/j.fuel.2008.12.003
9. Pang, L. S. K.; Vassallo, A. M.; and Wilson, M. A.1990 , Chemistry of alkali extraction of brown coals I. Kinetics, characterization and implications to coalification,
Organic Geochemistry, V. 16, No. 4-6, pp. 853-864. doi: 10.1016/0146-6380(90)90122-G
10. Garcia, D.; Cegarra, J.; Abad, M.; and Fornes, F.1993 , Effects of the extractants on
the characteristics of humic fertilizer obtained from lignite, Bioresource Technology, V.
43, No. 3, pp. 221-225. doi: 10.1016/0960-8524(93)90034-9
11. Caenn, R., and Chillingar, G. V.1996 , Drilling fluids: state of the art, Journal of Petroleum Science Engineering, V. 14, No. 3-4, pp. 221-230. doi: 10.1016/0920-4105(95)00051-8
12. Pehlivan, E., and Arslan, G.2006 , Comparison of adsorption capacity of young
brown coals and humic acids prepared from different coal mines in Anatolia, Journal of
Hazardous Materials, V. B138, No. 2, pp. 401-408. doi: 10.1016/j.jhazmat.2006.05.063
13. Krol-Domanska, K., and Smolinska, B.2012 , Advantages of lignite addition in
purification process of soil polluted by heavy metals, Biotechnology and Food Sciences,
V. 76, No. 1, pp. 51-58.
76SP-302-05
14. Tiemeyer, C.; Lange, A.; and Plank, J.2014 , Determination of the adsorbed layer
thickness of functional anionic polymers utilizing chemically modified polystyrene
nanoparticles, Colloids and Surfaces A: Physicochemical and Engineering Aspects, V.
456, pp. 139-145. doi: 10.1016/j.colsurfa.2014.05.014
15. Gupta, P. R., and Goring, D. A. I.1960 , Physicochemical Studies of Alkali Lignins
III: Size and Shape of the Macromolecule, Canadian Journal of Chemistry, V. 38, No. 2,
pp. 270-279. doi: 10.1139/v60-036
16. Stevenson, F. J., Humus Chemistry: Genesis, Composition, Reactions, John Wiley
& Sons, New York, USA, 2nd Edition, 1994.
17. Burchard, W.1983 , Static and dynamic light scattering from branched polymers and
biopolymers, Advances in Polymer Science, V. 48, pp. 1-124. doi: 10.1007/3-540-12030-0_1
18. Han, S., and Plank, J.2013 , Mechanistic study on the effect of sulfate ions on polycarboxylate superplasticizers in cement, Advances in Cement Research, V. 25, No. 4, pp.
200-207. doi: 10.1680/adcr.12.00002
19. Yamada, K.; Ogawa, S.; and Hanehara, S.2001 , Controlling of the adsorption and
dispersing force of polycarboxylate-type superplasticizer by sulfate ion concentration in
aqueous phase, Cement and Concrete Research, V. 31, No. 3, pp. 375-383. doi: 10.1016/
S0008-8846(00)00503-2
20. Plank, J.; Sachsenhauser, B.; and de Reese, J.2010 , Experimental determination of
the thermodynamic parameters affecting the adsorption behaviour and dispersion effectiveness of PCE superplasticizers, Cement and Concrete Research, V. 40, No. 5, pp. 699-709.
doi: 10.1016/j.cemconres.2009.12.002
SP-302-06
Influence of Superplasticizers on
the Flocculation Degree of Cement
Suspensions
by Lucia Ferrari and Pascal Boustingorry
Cementitious suspensions all feature common flow characteristics when their flow curve
is observed. When plotted as shear stress vs shear rate a minimum in stress is observed
towards low shear rates which may be related to the hydration of cement. When plotted as
apparent viscosity versus shear rate a minimum often appears towards high shear rates,
beyond which the suspension enters a shear-thickening regime the origin of which remains
unclear.
In between these two limits of shear rate, the expected shear-thinning behaviour takes
place, where apparent viscosity may be linked to a shear-rate-dependent degree of suspension flocculation.
The present paper aims at shedding some light onto the origins of those features in the
context of mix design and superplasticizer technology.
Keywords: pumping; civil engineering concrete; viscosity; superplasticizer;
phosphonate.
INTRODUCTION
Whereas the rheological behaviour of cementitious suspensions has often been discussed
under the framework of the Bingham model,1-3 thorough observations at the steady-state
and under no-slip conditions indicate more complex behaviour. It has been reported that
high dosages of superplasticizer may lead to a shear-thickening behaviour4-7 when shear
rate increases, which is quite well simulated by the Herschel-Bulkley model, but with
no satisfactory explanation of the underlying mechanism.3,5 Low shear rate regimes have
attracted less attention, but some studies have shown that the ageing nature of a suspension
may lead to a noticeable change in the flow curves with the occurrence of an increasing
stress branch at low strain rates.8,9 Between these two limits, the material features the often
reported shear-thinning behaviour.
The present work aims at showing that cementitious compositions all feature such characteristics, though in various intensities and ranges, depending on the mixture proportions
77
78SP-302-06
Table 1Grout Mix #1 French mix

Component
Type V Cement
0/0.160 mm Siliceous filler
Silica Fume
0/0.315 mm sand
Weight (g)
220,42
117,57
18,25
43,76
Weight (oz)
7,7751 oz
4,1472 oz
0,6437 oz
1,5436 oz
and the superplasticizer technology and dosage. More specifically the differences between
a phosphonate based and a polycarboxylate ether (PCE)-based admixture will be described.
This work points out some peculiar features of the flow of cementitious materials especially at low and high shear rates where, respectively, concrete placing and pumping are
concerned. The influence of PCE and phosphonate superplasticizer technologies on such
properties is studied and the advantages and drawbacks of each are discussed within a
theoretical framework allowing an interpretation in terms of suspension microstructure.
Materials
A CEM I 52.5N PM-ES Portland Cement (approximately equivalent to a Type V cement),
silica fume, a 0/0.160 mm and a 0/0.315 mm sand were used.
A PCE and a phosphonate based superplasticizer were used as aqueous solutions of
30% by weight of polymer in tap water with the addition of a suitable defoamer to prevent
air entrainment. The phosphonate superplasticizer is a linear poly(ethylene oxide) diphosphonate bearing roughly 45 ethylene oxide units. The average structure of the PCE is a
comb-shaped molecule of about 50 total methacrylic units out of which 20 bear 23-unit
poly(ethylene oxide) grafts (40% grafting ratio).
Equivalent grout mix design
The concrete mix-design was scaled down through the use of an approach inspired by
multiscale studies previously published.10 It relies upon applying a cutoff to the concrete
grading curve at an arbitrary particle sizein our case, 315 m. After normalizing to 100%
passing, a target grading curve is obtained which is then matched as closely as possible
by a blend of the binders and fine sands. This methodology may be considered as a way
to simulate the grout surrounding the largest aggregates in the concrete, while allowing
working in a rheometer where the sample is sheared in a very small gap, usually of the
order of several millimeters.
Two mixture proportions were used in the study: the dry composition of Grout Mix #1
is summarized in Table 1 while the different water amounts used are displayed in Table 2.
This grout mixture was computed according to the procedure described above, based upon
a civil engineering concrete composition from a French jobsite.
The proportions of the second Grout Mix #2, with a very low water-to-binder ratio are
provided in Table 3. This exceptional design simulates a civil engineering concrete from
an Asian jobsite, and will illustrate quite extreme flow features.
Influence of Superplasticizers on the Flocculation Degree of Cement

Suspensions79
Table 2Water amounts for the investigated solid volume fractions for Grout
mix #1
Target value
0.555
0.580
0.606
0.655
Water amount (ml)

144.21 ml
129.79 ml
116.81 ml
94.62 ml
Water amount (fl oz)

4.88 fl oz
4.39 fl oz
3.95 fl oz
3.20 fl oz
Table 3-Grout Mix #2 Asian mix

Component
Type V Cement
Fly Ash
Silica Fume
0/0.315 Sand
Total Water
Weight (g)
272.36
44.63
24.85
58.15
71.01
Weight (oz)
9.61 oz
1.57 oz
0.88 oz
2.05 oz
2.51 oz
Procedures
Water and admixtures were weighed in a Krups YY8506FD mixer bowl; the dry powders
were added during the first 30 seconds of mixing at speed 1 with a leaf-shaped blade. The
mixing speed was increased to speed 7 for 1 minute and then stopped for 30 seconds (to
scrape the sides of the bowl) before applying a last mixing stage of 1 minute at speed 7.
The sample was then loaded in the cylinder bob of a Kinexus Pro rheometer (Malvern
Instruments, U.K.) equipped with a vane geometry (4 mm side gap width, 1 mm bottom
gap). For applied shear rate measurements, the procedure started five minutes after the
beginning of mixing with a pre-shear at 200 s-1 during one minute, followed by logarithmic shear rate steps from 200 to 0.1 s-1. Each stress data point was sampled after the
steady state was reached whenever possible in order to build the flow curve. At the same
time, minislump tests at 5 minutes were performed (cone dimensions: upper diameter 18
mm 0.71 in, lower diameter 36 mm-1.42 in, height 54 mm-2.13 in equipped with a pneumatic lifting fork for reproducible results).
For applied stress protocols, the only changes are a pre-shear at 100 Pa, the stress step
program being logarithmic between 100 and 0.1 Pa.
After the flow curve measurement, the structure is reset to zero with an oscillating shear
period during one minute with a strain amplitude of 100% and a 1 Hz frequency. Then a
constant stress of 6.5 Pa is applied to the material in order to observe the structure buildup
close to rest through the increase of viscosity with time. This applied stress value was
chosen to match the stress applied by the weight of the largest aggregate in the system,
according to the following rough calculation:

gd / 2
(1)
10.5
: Specific weight difference between the falling aggregate and the suspending fluidg:
gravity constant, 9.81 m/s2d: particle diameter
80SP-302-06
Eq. 1 yields approximately 6.5 Pa for a diameter of 16 mm, a paste density of 1800 kg/
m3 (112.4 lb/ft3) and an aggregate density of 2600 kg/m3 (162.3 lb/ft3). This last section of
material testing are called ageing stage in the rest of the paper.
Since suspension structure in the flow regime is to be discussed, a model developed
explicitly with structure level and solid volume fraction parameters was chosen. The discussion is based on the works of Quemada11-14 who considered the suspension as a dispersion
of so-called Structural Units (SU), aggregates of grains each containing a number fraction
S of individual particles, S being a function of shear rate and time.
The resulting equation for the apparent viscosity follows a Krieger-Dougherty-like law,
the solid volume fraction being replaced by an effective volume fraction11:
2
eff
= F 1
(2)
m
eff: effective volume fractionm: maximum packing fractionF: suspending fluid viscosity
This effective volume fraction may be defined as a function of S and the compactness C
of the SUs:

eff = (1 + CS)
(3)
: Solid volume fractionC: SU compactness

The last step is to define a behaviour law for S by stating that its value results from a
dynamic equilibrium between aggregation and deflocculation by shear. When the system
reaches a steady state, an expression for S is given:

Seq =
S0 + S
(4)
1+
: function of shear rate which may take several forms depending on the complexity of the
phenomena to model. In its simplest form (shear-thinning), it reads:

= (tc ) p (5)
tc: characteristic time for deflocculation by shear : steady shear ratep: exponent for deflocculation by shear, lower than 1 and shown experimentally to often take values close to .
This theoretical development provides a shear rate dependence of the effective volume
fraction through Eq. 3, Eq. 4 and Eq. 5.
In its simplest form the model requires no less than five individual parameters, namely
C, S0, S, tc and p given that and m are suspension parameters determined by the chosen
water/solid ratio and the dry material particle size distribution. A simpler analysis may

Suspensions81
Fig. 1- Example of the general shape of flow curves

be performed though, by a simple transformation of the data according to the method
described below.
COMPARISON OF PREDICTIONS AND EXPERIMENTAL RESULTS
From flow curve data, apparent viscosity as a function of shear rate may be extracted.
The first level of analysis is quite straightforwardfrom Eq. 1 comes:

eff
m
( )
= 1
1/ 2
(6)
This transformation allows plotting the ratio of the effective volume fraction to the
maximum packing fraction as a function of shear ratethis allows an observation of the
change in relative structure under shear. In the present paper it was chosen not to extend the
analysis into further stages since data provided by Eq. 6 already describes the microstructural evolution of the suspension.
General features of the flow of cementitious suspensions
An illustration of typical experimental flow curves obtained as close to steady state as
possible is provided in Fig. 1. On Fig. 1(a) was plotted the measured stress vs the applied
shear rate; the shape of the curve is counter-intuitive in the sense that the general consensus
stands upon an ever-increasing curve when shear rate increases. In numerous cases during
this study, such U-shaped curves are obtained, the minimum of which defines a critical
stress c and critical shear rate c. The origin of such a shape will be discussed below but
has also been reported in the past8,15 and is discussed more precisely in the paper SP-089 of
the present conference.
82SP-302-06
Fig. 1(b) shows the same data plotted in terms of apparent viscosity vs shear rate. After
the expected shear-thinning behaviour (decreasing viscosity branch), a viscosity minimum
min is also observed, beyond which the suspension enters a shear-thickening regime. The
critical shear rate defining the onset of this behaviour was noted st , the st index standing
for shear thickening.
As already stated, these features are quite ubiquitous among the cement suspensions
studied by the authors of this paper, the only differences from one system to another being
the values obtained for the four critical parameters described above.
Low shear limit: the influence of suspension ageing
As previously stated, Fig. 1 shows an unexpected non-monotonous trend in shear stress
vs shear rate. Some previous work seems to point at the influence of ageing or thixotropy;
an analysis by Roussel et al8 shows that the introduction of a time-dependence component
in a flow curve equation leads to U-shaped flow curves (see also paper SP-089). A more
general theoretical approach developed by Picard et al15 also concludes that non monotonous curves are expected. Both approaches rely upon the coupling of a general flow equation with a differential equation providing a time-dependence of an underlying parameter
(structure level for Roussel et al., fluidity for Picard et al.) showing that structure evolution
through time is involved.
A closer look to the raw data recorded here allows observing such an influence. As a
matter of fact, all data points below c appear to be flagged with a non-steady state alarm
by the rheometer software. This means that the data points were recorded despite the fact
that no steady torque value was achieved for the requested shear rate. Fig. 2 shows the
transient data in terms of steady-state index, a value supposed to reach one when the steady
state is achieved. Fig. 2(b) shows that when the applied shear rate is higher than the critical
shear rate, steady flow is achieved in mere seconds whereas an applied rate lower than the
critical value induces an oscillating behaviour with an increasing amplitude as seen on Fig.
2(a), preventing the rheometer to establish a steady flow. Roussel et al8 explain that below
critical stress or rate no steady flow may indeed be achieved anymore and the suspension
ages (i.e. its structure level increases), though slower than at complete rest. It may be
argued that since no homogeneous or steady flow is possible anymore, the critical stress
acts as an apparent yield stress.
Some further insight about this interesting flow feature may be brought up by a stress
step protocol applied to a freshly mixed batch of the same suspension. Fig. 3 shows the
flow curves obtained by applying a decreasing stress step program, then an increasing
stress step program with a steady-state condition. The obvious observation is that the flow
curves do not have the same shape as the applied shear rate curve.
At relatively high applied stress where flow is established all data sets are consistent. The
decreasing stress stage shows a critical stress below which shear rate suddenly decreases
by almost five decades to reach very low values in the range 10-5-10-4 s-1. The subsequent
increasing stress protocol features the same rate jump, but for a higher stress.
There appears a hysteresis loop around a low value of stress that is a result of the competition between shear and ageing, as supported by the work by Picard et al,15 quoted below:

Suspensions83
Fig. 2- Steady state index during the transient periods right after the setting of a new shear
rate value. (a) below critical shear rate. (b) above critical shear rate.
Fig. 3- Decreasing then increasing stress steps applied to

the suspension near the critical stress. Shear rate jumps over
several decades are observed for different applied stresses
depending on the shear history.
When [stress] is imposed, the flow curve jumps in a hysteretic fashion between two
branches: one that corresponds to no flow but for a wall layer in the vicinity of the walls;
the other corresponds to a fully fluidized situation. When the [global shear rate] is imposed,
shear banding can occur, as well as sometimes a stick-slip like oscillating behaviour at
small shear rate that corresponds to a localized oscillation of the fluidity.
This description based upon the assumption of a time-depending evolution is then totally
consistent with the observations presented in this paper. It was shown previously9 and is
shown in paper SP-089 in the same book that superplasticizer nature and dosage both influence the values of critical stresses and shear rate, with a noticeable effect on concrete flow
during casting.
84SP-302-06
Fig. 4- Influence of solid volume fraction on the shape of flow curves. Grout mix #1. PCE
superplasticizer with a 0.65% dosage.
Here the influence of suspension solid volume fraction was studied by gradually
decreasing the amount of water in the mix. Fig. 4 shows that an increase of solid volume
fraction at a constant PCE superplasticizer dosage of 0.65% by weight of binder yields both
an increase of apparent viscosity throughout the observed shear rate range and an increase
in critical shear rate and stress, a sign of a faster ageing.
Ageing itself was studied as described above under a constant stress of 6.5 Pa. The
results are plotted in Fig. 5, which shows for = 0.555 a rather slow viscosity increase
through time, with a sharper slope towards 30 min corresponding to gelling. For = 0.580,
the flow period before gelling is shorter while the upper two values yield gelling in less
than a minute.
There seems to be a connection between the values of c and cand the ageing kinetics of
the suspension since they feature the same trend. It may then be stated that suspension
ageing is responsible for the flow behaviour at low shear rates, the above parameters
defining a boundary between homogeneous flow and heterogeneous or impossible flow.
This contributes to showing how suspension ageing through hydration is closely connected
to the low-shear-rate flow characteristics.
Steady flow regime: effective volume fraction flow curves, admixture
robustness
As stated above, the flow regime is composed of a shear-thinning domain followed by a
shear-thickening regime beyond a shear rate st .
It seemed interesting to compare flow curves of both superplasticizers at equal workability, which is the purpose of Fig. 6. In terms of viscosity, there appears a crossover
between the diphosphonate and the PCE technology at around 4 s-1, beyond which shear
thickening is more intense for the PCE. Fig. 7 corresponds to the same data transformed
into the ratio of effective volume fraction to maximum packing fraction through the use of

Suspensions85
Fig. 5- Ageing curves in rotational protocol as a function of solid volume fraction. PCE
superplasticizer. 0.65% by weight of total binder.
Fig. 6- Comparison of flow curves at 5 min obtained with both technologies at equivalent
workability
Eq. 6. It may be observed that the phosphonate technology indeed allows a lower structure
level of the suspension, thus a higher amount of deflocculation, beyond 10 s-1, i.e. for shear
rates corresponding to pumping or mixing. Interestingly, shear thickening is here interpreted as an increase of structure with an increase of shear rate, i.e. shear is here considered
as a flocculation mechanism for example through the formation of hydroclusters.12,13 Shear
thickening may interfere with operations at high shear rate, i.e. pumping, and these results
show the advantage of a phosphonate technology over a PCE technology on this property.
The tradeoff is a much higher dosage, thus a higher cost and a possible influence on the
early strength.
86SP-302-06
Fig. 7- Flow curves of Fig. 6 transformed

according to Eq. 6. Grout mix #1. =
0.555.
Another important admixture property is its robustness to the water amount. Any inaccuracy in the weighing of water or the moisture measurement of aggregates may lead
to a variation in initial workability. Water reduction at constant dosage was investigated,
leading to an increase of the solid volume fraction . The data recorded at 5 min, expressed
in terms of relative structure through Eq. 6, are shown in Fig. 8.
There is an obvious difference between the two technologies; while the phosphonate is
able to keep a consistent structure level in the 0.1-10 s-1 range despite the increase in , the
PCE is unable to do the same and the structure degree increases with at all shear rates,
leading to a loss of workability. It may then be concluded that the phosphonate admixture
is more robust to water variations, the tradeoff being once again a much higher dosage.
Another interesting feature of the phosphonate experiments is that the shear thickening
effect seems to intensify when increases, showing that this phenomenon depends on the
solid volume fraction. This feature will be further investigated in the next section.
High shear limit: possible mechanisms at the origin of shear thickening
The first mechanism that may explain shear thickening is the rise of inertial forces16,17
and this was investigated here by following the approach used by Brown and Jaeger16 who
defined a suspension Reynolds number under the form:

Re =
l d 2
(7)
min
l: liquid specific gravityd: particle diametermin: minimum viscosity defined on Fig. 1.

Even by taking the diameter of the largest particle (315 m), at the highest shear rate
investigated (200 s-1), the Reynolds number of the suspension is no higher than 5.10-3. This
shows that inertial effects may be neglected in the system.
The high shear rate limit was investigated on Grout #2 the advantage of which is to
require almost the same dosage for both technologies (2.0% for the PCE, 2.5% for the

Suspensions87
Fig. 8- Relative structure levels of suspensions (Eq. 6) with a variable volume fraction
phosphonate) at equal workability. This difference from Grout #1 comes primarily from
a much lower water amount and the presence of fly ash in Grout #2. The phosphonate
robustness may be stressed out again with a small dosage difference between grouts at the
expense of a high overall dosage with respect to the PCE.
Grout #2 features a much more intense shear thickening regime than Grout #1 as shown
in Fig. 9, though the phosphonate superplasticizer still seems to mitigate shear thickening
when compared to the PCE. Indeed the apparent structure level computed from Eq. 6 on
Fig.9(b) shows obviously that the phosphonate induces a much lower degree of structure
in the range 10 200 s-1, though shear thickening still appears towards 100 s-1.
A final insight may be given about the possible origin of shear thickening. Though the
vane geometry is not properly designed for such a purpose, normal force was measured
during the experiments. Fig. 10 represents the evolution of the normal force exerted by the
suspension onto the vane versus the recorded shear stress. A very sharp increase of normal
force (up to 0.14 N) may be observed at high stress in the case of the PCE superplasticizer
while for the phosphonate the normal force remains lower than 0.025 N in the same stress
range.
It was previously shown16-19 that the rise of contact forces between particles, especially
of frictional nature, are often involved whenever such a coupling between shear and normal
stresses is observed. This means that the occurrence of frictional regimes may be a cause of
shear thickening in the observed suspensions.
If such is the case, it would mean that the phosphonate adsorbed layer is able to mitigate
friction between particles while the PCE adsorbed layer is less able to do so, leading to
a sharper rise of friction, normal stresses and shear viscosity. This would be obviously
linked to the structure of the adsorbed layers, which would be dependent upon the polymer
structure.
FURTHER RESEARCH
If the low shear rate regime may be well understood in the framework of suspension
ageing, there still remains some work to better understand the underlying mechanisms of
88SP-302-06
Fig. 9- Comparison between apparent viscosity and apparent suspension structure on both
technologies. Grout mix #2.
Fig. 10- Normal force applied on the

geometry as a function of stress. Grout
mix #2.
the shear-thickening regime. If inertia may be ruled out, hydrocluster formation or the rise
of contact or frictional forces at high shear rates may still be involved. Superplasticizer
technology seems to have an influence on the phenomenon, but the exact molecular parameters controlling the performance are still to be determined.
CONCLUSIONS
This paper is dedicated to seldom-studied features of the flow of cementitious materials, beyond the often described shear thinning or Bingham behaviours. It was first shown
that low-shear-rate regimes may lead to a discontinuous behaviour with a critical value of
applied rate or stress below which no steady flow may occur. Previous work points at the
influence of the ageing of the suspensions, through cement hydration, that interferes with

Suspensions89
shear. The diphosphonate and the PCE technologies both behave in a similar manner in
this regime.
Beyond the main shear thinning regime, there appears a high shear rate regime of
viscosity increase similar to what some authors define as continuous shear thickening.17,25
This is where the studied technologies differ, the phosphonate superplasticizer producing
much lower viscosities in the 10-200 s-1 shear rate range. Elements from the literature and
some experimental facts allow arguing that the underlying mechanism is related to the way
the adsorbed phosphonate polymers mitigate contact forces between particles in a more
efficient manner than the PCE, at the expense of its water reducing ability, which seems
noticeably lower.
AUTHOR BIOS
Dr Lucia Ferrari is the Physical Chemistry manager in the main research and development laboratory of CHRYSO in France. She received her PhD from the Technische
Universitt Mnchen (Germany) after she completed her PhD work with the EMPA
in Dbendorf (Switzerland) under the supervision of Dr Frank Winnefeld and Pr. Dr.
Johann Plank.
Dr Pascal Boustingorry is the Head of the Interface Physical Chemistry Team in the main
research and development laboratory of CHRYSO in France. He received his PhD from
the INP Grenoble along with the Ecole des Mines in Saint Etienne (France). Their main
research interests are the interaction of organic molecules with cement suspensions and
the links between superplasticizer chemical architecture and flow properties of building
materials.
REFERENCES
1. S, G., S, G. Determination of Bingham Parameters of Fresh Portland Cement Concrete
Using Concrete Shear Box, Bonfring International Journal of Industrial Engineering and
Management Science, V. 2, No. 4, 2012, pp. 84-90. doi: 10.9756/BIJIEMS.1620
2. Wallevik, J. E., Relationship between the Bingham parameters and slump,
Cement and Concrete Research, V. 36, No. 7, 2006, pp. 1214-1221. doi: 10.1016/j.
3. Wallevik, O. H., and Wallevik, J. E., Rheology as a tool in concrete science: The
use of rheographs and workability boxes, Cement and Concrete Research, V. 41, No. 12,
2011, pp. 1279-1288. doi: 10.1016/j.cemconres.2011.01.009
4. Cyr, M.; Legrand, C.; and Mouret, M., Study of the shear thickening effect of superplasticizers on the rheological behaviour of cement pastes containing or not mineral additives, Cement and Concrete Research, V. 30, No. 9, 2000, pp. 1477-1483. doi: 10.1016/
S0008-8846(00)00330-6
5. Feys, D.; Verhoeven, R.; and De Schutter, G., Why is fresh self-compacting concrete
shear thickening?, Cement and Concrete Research, V. 39, No. 6, 2009, pp. 510-523. doi:
10.1016/j.cemconres.2009.03.004
6. Raghavan, S. R., and Khan, S. A., Shear-thickening response of fumed silica suspensions under steady and oscillatory shear, Journal of Colloid and Interface Science, V. 185,
No. 1, 1997, pp. 57-67. doi: 10.1006/jcis.1996.4581
90SP-302-06
7. Toussaint, F.; Roy, C.; and Jzquel, P. H., Reducing shear thickening of cementbased suspensions, Rheologica Acta, V. 48, No. 8, 2009, pp. 883-895. doi: 10.1007/
s00397-009-0362-z
8. Roussel, N.; Le Roy, R.; and Coussot, P., Thixotropy modelling at local and macroscopic scales, Journal of Non-Newtonian Fluid Mechanics, V. 117, No. 2-3, 2004, pp.
85-95. doi: 10.1016/j.jnnfm.2004.01.001
9. Ferrari, L.; Boustingorry, P.; Pineaud, A.; and Bonafous, L., From cement grout to
concrete scale: a study of superplasticizer-design-controlled thixotropy to match SCC
application requirements. in Rheology and processing of Construction Materials 7th
RILEM International Conference on Self-Compacting Concrete and 1st RILEM International Conference on Rheology and Processing of Construction Materials 285292
(RILEM Publications, 2013).
10. Toutou, Z., and Roussel, N., Multi scale experimental study of concrete rheology:
from water scale to gravel scale, Materials and Structures, V. 39, 2006, pp. 167-176.
11. Quemada, D. Rheological modelling of complex fluids. I. The concept of effective
volume fraction revisited. Eur. Phys. J. AP 1, 119127
12. Quemada, D. Rheological modelling of complex fluids: II. Shear thickening behavior
due to shear induced flocculation. Eur. Phys. J. AP 2, 175181
13. Quemada, D. Rheological modelling of complex fluids: III. Dilatant behavior of
stabilized suspensions. Eur. Phys. J. AP 3, 309320
14. Quemada, D. Rheological modelling of complex fluids: IV: Thixotropic and
thixoelasticbehaviour. Start-up and stress relaxation, creep tests and hysteresis cycles.
Eur. Phys. J. AP 5, 191207
15. Picard, G.; Ajdari, A.; Bocquet, L.; and Lequeux, F., Simple model for heterogeneous flows of yield stress fluids, Physical Review E: Statistical, Nonlinear, and Soft
Matter Physics, V. 66, No. 5, 2002, p. 051501 doi: 10.1103/PhysRevE.66.051501
16. Brown, E. & Jaeger, H. M. The role of dilation and confining stresses in shear thickening of dense suspensions. Journal of Rheology (1978-present) 56, 875923 (2012).
17. Brown, E. etal., Generality of shear thickening in dense suspensions, Nature
Materials, V. 9, 2010, pp. 220-224.
18. Seto, R.; Mari, R.; Morris, J. F.; and Denn, M. M., Discontinuous shear thickening
of frictional hard-sphere suspensions, Physical Review Letters, V. 111, No. 21, 2013, p.
218301 doi: 10.1103/PhysRevLett.111.218301
19. Brown, E., Frictions Role in Shear Thickening, Physics, V. 6, 2013, p. 125 doi:
10.1103/Physics.6.125
20. de Gennes, P. G., Conformations of polymers attached to an interface, Macromolecules, V. 13, No. 5, 1980, pp. 1069-1075. doi: 10.1021/ma60077a009
21. Alexander, S., Adsorption of chain molecules with a polar head a scaling
description, Journal of Physics, V. 38, No. 8, 1977, pp. 983-987. doi: 10.1051/
jphys:01977003808098300
22. Fleer, G. J., Polymers at interfaces and in colloidal dispersions, Advances in Colloid
and Interface Science, V. 159, No. 2, 2010, pp. 99-116. doi: 10.1016/j.cis.2010.04.004
23. Fleer, G. J., and Skvortsov, A. M., Reconciling lattice and continuum models for
polymers at interfaces, The Journal of Chemical Physics, V. 136, No. 13, 2012, p. 134707
doi: 10.1063/1.3693515

Suspensions91
24. Lyklema, H.; Stuart, M. C.; and Leermakers, F., Gerard Fleer, Advances in Colloid
and Interface Science, V. 159, No. 2, 2010, pp. 95-98. doi: 10.1016/j.cis.2010.06.008
25. Fall, A., Lematre, A., Bertrand, F., Bonn, D. & Ovarlez, G. Continuous and discontinuous shear thickening in granular suspensions. Physical Review Letters (2010). tables
and figures List of Tables:
92SP-302-06
SP-302-07
Influence of Temperature and Retarder on

Superplasticizer Performance
by Karen Luke and Adrian Torres
Superplasticizers are often used in conjunction with other additives and this can produce
either an adverse or synergistic effect on rheology and setting properties of cementitious
systems. These effects can be enhanced when temperatures are increased due to environmental changes or induced temperature as in hydrothermal curing. This research focuses
on the compatibilities of different types of superplasticizer either sulfonated naphthalene
or polycarboxylate based in combination with a lignosulphonate or hydroxycarboxylic
acid type retarder.
Rheological measurements were made using a rotational viscometer at temperatures
from 25C (77F) to 120C (248F) under pressure, and plastic viscosity and yield point
determined based on the Bingham Plastic model though in almost all cases it was noted
that the Power Law or more so the Herschel-Buckley model gives a better fit. Zeta potential was used to characterize particle surface interactions to understand synergy of additive combinations. Setting properties, investigated using conduction calorimetry, were
observed to be dominated by retarder response.
Keywords: Calorimeter; polycarboxylate; polynaphthalene; sulfonated acetone formaldehyde; rheology; retarder; temperature; zeta-potential.
INTRODUCTION
Superplasticizer technology and utilization has advanced notably since its early development (Hattori, K., 1978) in the late 1960s. There is an abundance of literature on the use
of superplasticizer as a high performance concrete (Aitcin, P.-C., 1998), with compressive strengths nowadays exceeding 100 MPa (14,503 psi), and also on high-workability
concrete or more recently self-compacting concrete (Walraven J.C., 2010). The benefits of
high performance concrete are well known where water to binder ratios as low as 0.2 can
be achieved without reducing the consistency giving low permeability, high compressive
strength, elastic modulus and flexural strength, as well as improved abrasion resistance
and durability. Self-compacting concrete technology has advanced considerably with the
invention of modern superplasticizers based on polycarboxylate polymers and coupled
with advances in the use of stabilizing additives (Schmidt, W., 2014). Superplasticizers are
often used in combination with other additives (Michaux, M., et al., 1986, 1986a) and the
93
94SP-302-07
effects, adverse of synergistic, are notably influenced by temperature either due to environmental factors or hydrothermal conditions as obtained in cementing of oil and gas wells
(Nelson, E.B. et al., 2006). Although publications on superplasticizers in well cementing
are limited they are used in a variety of applications to improve mixability and/or flow
characteristics of the cement slurry (mixture of cement, water and additives).
Superplasticizer addition along with particle packing methodology is used in designing
low density, 1020 kgm-3 (63.7 lb/ft3) to about 1450 kgm-3 (90.5 lb/ft3), slurries that are also
required to provide high strength and low permeability, achieved by using a high solid fraction consisting of a light weight extender. In normal density slurries, 1700 kgm-3 (106.1
lb/ft3) to 1900 kgm-3 (118.6 lb/ft3), superplasticizers are used to improve slurry mixability
and reduce friction pressures, particularly where small annular spaces may be encountered
and high friction pressures may cause fracture to weaker formations. Densities above 2000
kgm-3 (124.9 lb/ft3) where water to solid ratios are less than 0.4, superplasticizers are used
to give flowable slurries. Such slurries are used for abandonment plugs that are required to
have very low permeabilities or for whipstock plugs used for directional drilling where the
cement slurry should have adequate time for placement, develop strength rapidly and be
stronger than the surrounding formation. Superplasticizers are also often used in combination with fluid loss additives, where they improve cement particle dispersion and aids in
optimal fluid loss performance. In addition many fluid loss additives are viscosifying and
the addition of superplasticizer aids with mixability and flowability.
Rheological parameters, obtained on a coaxial rotational viscometer (API RP 10B,
2010), are used to provide predictions of friction pressure and slurry flow properties
through pipes, casing and annulus using appropriate mathematical relationships (Guillot,
D., 2006). Current technology has limited rheological test conditions for cement slurries
to 80C (176F) and atmospheric pressure whereas under well conditions the slurry can
be exposed to temperatures up to 320C (608F) and pressures of 280 MPa (40,610 psi).
In addition, testing under conditions specified by API RP 10B at 60C (140F) to 80C
(176F) the slurry can potentially viscosify due to evaporation of the aqueous water phase
during the conditioning phase. Recent developments in rotational viscometers allows
cement slurry rheological parameters to be measured up to 316C (600F) and 207 MPa
(30,023 psi) and eliminates potential issues of aqueous phase evaporation of the slurry.
This study provides a preliminary investigation using the high temperature, high pressure
(HTHP) rotational viscometer to determine in the first instance the effect of three commercial superplasticizers on the rheological parameters from 25C (77F) to 120C (248F)
and 10.0 MPa (1450 psi) in combination with two different retarder types and an antifoam
additive. Additional testing was also performed using zeta potential measurements and
calorimetry to further elucidate on the effects of the superplasticizer, retarder and antifoam
combinations.
Rheological data obtained to date for determining flow properties and friction pressures
for well applications are typically measured at atmospheric pressure and at temperatures in
the range of 25C (77F) to 80C (176F). At 60 C (140F) to 80C (176F) there is potential for viscosification of the slurry due to evaporation during the conditioning period of
the test. Development of a HTHP rotational viscometer that can measure rheologies at high
Influence of Temperature and Retarder on Superplasticizer Performance 95
temperatures and pressures is believed to give more accurate values on friction pressures
and slurry flow properties of the cement slurry under well conditions. The significance of
this research is in elucidating the rheological properties of different types of superplasticizer when used in combination with retarder and antifoam at temperature and pressure
more realistic of actual well conditions.
Rheological experiments were carried out on three different types of superplasticizer,
a sulfonated naphthalene formaldyehyde condensate, a sulfonated acetone formaldehyde
condensate and a carboxylated polymer, at temperatures from 25C (77F) to 120C
(248F) in a high temperature, high pressure (HPHT) rotational viscometer. Two different
retarders were used, one for low temperature less than 100C (212F) and the second for
high temperature equal to or greater than 100C (212F). Two different retarders were
used as in field applications in this temperature range the reaction rate of cement has an
exponential effect on retarder response. The result is that retarders effective below 100C
(212F) are ineffective above 100C (212F) and effective above 100 C (212F) too
powerful below 100C (212F) even at low concentrations. In addition small changes in
the concentration of HC-R below 100C (212F) can cause large variations in thickening
time. This in field applications can lead to premature set in the casing, or long waiting on
cement to set times that can result in costly rig time and in laboratory testing can cause
variable data and non-reproducible results. An antifoam additive, common in most well
slurries to minimize foaming during mixing (Nelson, E.B. et al. 2006), was included in
all tests. Correlation of data was made in relation to Bingham Plastic, Power Law and
Herschel-Buckley fluid models. Additional measurements were also made using calorimeter and zeta-potential to better define the performance of the different types of superplasticizer, retarder combinations.
Materials
A commercially available Class G cement conforming to API Specifications (API Specification 10A, 2011) was used in this study. Physical, chemical and mineralogical data are
provided in Table 1. Three different chemical categories of superplasticizers were investigated, a sulfonated polynaphthalene formaldehyde condensate (PNS), a sulfonated acetone
formaldehyde condensate (PAS) and a polycarboxylate polymer (PC). The superplasticizers
were in powder form, with the PNS and PAS powders containing about 10% sodium sulfate
non-reacted material from the manufacturing process whereas the PC powder consisted of
20% by weight of polymer and 80% by weight of an added inert material. As such the PNS
and PAS powders contained 90% polymer component while the PC contained only 20%
polymer component. Retarders used were, a lignosulfonate (LS-R) for temperatures less
than 100C (212F) and a hydroxycarboxylic acid (HC-R) for temperatures at and above
100C (212F). A glycol polymer antifoam (PG-A) was also used in the slurry blends in
order to minimize foaming during blending. All the additives were commercially available.
Reversed osmosis (RO) water was used for all testing.
96SP-302-07
Table 1Physical and chemical composition of Class G cement

Class G
Specific gravity
Blaine fineness, cm2/g
Loss of ignition, %
SiO2%
CaO %
Al2O3%
Fe2O3%
MgO %
SO3%
Na2O %
K2O %
Na2Oe %
Phase mineralogy C3S
C2S
C3A
C4AF
3.14
2919
1.14
20.69
62.25
3.54
4.29
2.70
2.97
0.32
0.52
0.67
57.74
15.84
2.12
13.03
Table 2 Slurry compositions

Temperature C
(F)
Slurry Components
20 80
(77-176)
100 120
(212-248)
Blend
3
Cement
Superplasticizer (% bwoc)
LS-R (% bwoc)
PG-A (% bwoc)
100
0.0
0.5
0.2
100
0.2
0.5
0.2
100
0.4
0.5
0.2
100
0.6
0.5
0.2
100
0.8
0.5
0.2
Cement
HC-R (% bwoc)
PG-A (% bwoc)
6
100
0.0
0.5
0.2
7
100
0.04
0.5
0.2
Blend
8
100
0.1
0.5
0.2
9
100
0.2
0.5
0.2
10
100
0.4
0.5
0.2
Test Methods
A dry blend of additives and cement was mixed with water at a water to cement ratio of
0.44, in a Waring blender according to API Recommended Practice (API RP 10B-2, 2010)
to give a 600 mL slurry. Additives were added as a percentage by weight of cement (%
bwoc). The compositions of the blends are provided in Table 2. All slurries had a density
of 1904 kgm-3 (118.6 lb/ft3) and yield of 0.757 m3/t (0.012 ft3/lb). Aliquots of the blended
slurry were then taken and used for either rheological or calorimetric studies.
Rheological studies
The slurry after blending was conditioned in an atmospheric consistometer at 25C (77F)
for 20 minutes. An aliquot of approximately 150 mL was then placed in the measuring cup
of a Chandler 7500 HPHT rotational viscometer. Samples were heated to the required
temperature, 40C (104F), 60 (140F), 80 (176F), 100C (212F), or 120C (248F)
under a pressure of 10 MPa (1,450 psi). Ramp time to reach the required temperature and
pressure was in all cases 45 minutes. At the selected temperature data were recorded during
both a ramp- up and ramp-down at 3, 6, 100, 200, and 300 RPM with the speed held for one
minute at each RPM setting. The total time from blending to the first measurement was an
average time of 70 minutes.
Calorimetric studies
Approximately 2.5 mL of the 600 mL slurry was syringed into a pyrex flat bottomed tube
(6 mm diameter x 5 cm), and placed into a high pressure stainless steel Setaram C80 calorimeter cell supplied by Setaram Instrumentation, France. The cell was sealed and placed
into the calorimeter simultaneously with the reference cell. The reference sample used for
tests at 80C (176F) and below, consisted of a LS-R retarded Class G cement previously
cured at 80C (176F) for 24 hours on the assumption that the thermal capacity matched,
as closely as possible, the actual experiments (Aukett and Bensted, 1992). At temperatures
greater than or equal to 100C (212F), the reference sample was a HC-R retarded Class G
cement cured at 120C (248F) for 24 hours. Nitrogen was applied to each cell at a pressure
of 7.5 MPa (1,087 psi) using a high pressure gas control panel. Heat was applied according
to API schedules (API RP 10B, 2010) for appropriate temperature and then maintained at
temperature for 24 hours. Ramp time for 25C (77F) to 40C (104F) and 25C (77F) to
120C (248F) was 22 minutes, and 60 minutes respectively. The benefit of using a ramp
simulating well conditions versus isothermal conditions has been discussed (Aukett, P.N.
and Bensted, J. 1992).
Zeta Potential
Zeta potential of dilute suspensions, 200 mgL-1, solid material in RO water or pore solution extracted from cement slurry after approximately 70 minutes hydration, was measured
in a Horiba, Nano-particle analyzer SZ 100. The solid material consisted of Class G cement
or Class G cement plus 0.5% bwoc LS-R or HC-R and 0.2% bwoc PG-A consistent with
slurry compositions used for rheology and calorimeter studies. PNS, PAS or PC was added
at a concentration of 0.4% bwoc to the LS-R cement suspensions and 0.1% bwoc to the
HC-R cement suspensions and zeta potential data obtained after approximately 1 hour.
Data analysis and rheological models
Bingham Plastic, Power Law and Herschel-Bulkley mathematical fluid models are
commonly used to define the flow properties of well cement slurries. Of the three models
the Bingham Plastic model as described in equations (1) and (2) has been the most widely
used.

= y + p when > y (1)
= 0 when y (2)
98SP-302-07
where: = shear stressy = yield stressp = plastic viscosity = is the shear rate
However, the shear-stress/shear-rate relationship appears more consistent with the
Power Law model or more the Herschel-Bulkley model as illustrated in equation (3) that
also includes a yield stress component, and is common in concentrated suspensions such
as obtained in cement slurries.

= y + kn when > y (3)
where: = shear stressy = yield stress = is the shear ratek = consistency indexn = Powerlaw index
When n is less than 1, Herschel-Bulkley fluids are shear thinning, while n greater than 1 is
shear thickening. At n equals 1 the Herschel-Bulkley model reduces to the Bingham Plastic
model. At present there is no defined rational on whether to use the Herschel-Bulkley or
Bingham Plastic model other than choosing the model that appears to give the best fit
to the rheological data. Shear stress/shear rate plots were determined for all slurries and
compared with the Bingham Plastic or Herschel-Bulkley models. Figure 1 shows the shear
stress/shear rate plots obtained from the rheological data of selected slurries and illustrates
how the best fit model, Bingham Plastic or Herschel-Bulkley, is selected in the current
study. Figure 1a and b shows the data from the shear stress/shear rate plots for the 0.2%
bwoc PAS slurry at 25C (77F), 10 MPa (1,450 psi) illustrating that the best fit in this
case is the Herschel-Bulkley model. Figure 1c and d, shows data from 0.5% bwoc PNS at
80C (176 F), 10 MPa (1,450 psi) where either the Bingham Plastic or Herschel-Bulkley
model give similar fit. Figure 1e and f, 0.4% bwoc PAS at 40C (104F), 10 MPa (1450
psi), provides an example where n equals 1 and the Herschel-Bulkley model reduces to
the Bingham Plastic model and as would be expected both plots give the same linear slope
and yield stress. The change in the shear stress/shear rate relationship and hence choice of
model for the different slurries can result from shear effects, pressure and/or temperature
effects, additive chemistry and/or additive concentration.
Rheological profiles and effect of superplasticizer at 25-80C (77-176F)
The rheological data obtained for the superplasticizers investigated with the LS-R retarder
in combination with the antifoam PG-A at 25C (77F) to 80C (176F), is provided in
Table 3, Table 4, Table 5, and Table 6. Data shows that slurries with low concentrations
of superplasticizer, 0.2% bwoc or less, follow the Herschel-Bulkley model and notably
have an n value significantly less than 1.0 and is indicated by the shaded sections in the
tables. As concentrations increase above 0.2% bwoc superplasticizer slurries behave either
according to the Bingham Plastic or Hershel-Bulkley models as illustrated in Fig 1 c and
d, and notably as n increases from about 0.6 to around 1.0. In a number of cases such as
for 0.5% bwoc PNS at 40C (104F) a value of n greater than 1, is observed. This is even
more evident at 60C (140 F), not only for the PNS but also PAS indicating shear thickening behavior of the slurry, Figure2. The explanation for this behavior, remains unclear.
It is noted that for the 0.6% bwoc PNS at 60C (104F) slurry with n equal to 1.186 that
both the Bingham Plastic and Herschel-Bulkley models give a good fit to the data, Figure
2 a and b whereas with the 0.5% bwoc PNS at 40C (104F) and n equals 1.274 it is clear
that the Herschel-Bulkley model is a better fit, Figure 2d and e. At values of n above 1.274
Fig 1.Shear stress/shear rate plots showing a)Bingham

Plastic model, b)Herschel-Bulkley model(n = 0.077) for
0.2% bwoc PNS slurry at 25C(77F), c)Bingham Plastic
model, d)Herschel-Bulkley model (n = 0.500) for 0.5%
bwoc PNS at 80C(176F)and e) Bingham Plastic model,
f)Herschel-Bulkley model(n = 1.01) for 0.4% bwoc PAS at
40C(104F).
the fit to the Herschel-Bulkley model was even more evident. Yield stress values for the
slurries having n less than 1, and using the Herschel-Bulkley model, generally tend to be
lower than those calculated from the Bingham Plastic model whereas for n approximately
equal to 1.0, values are similar. It should be noted, however, with complex fluids such as
cement slurries in-homogeneities can occur due to particle settling, slip or time-dependent
behavior that can create conditions for erroneous measurement particularly at low end
shear rates and makes model fitting problematic (Sairam, P.K.S., et al., 2012). Values taken
for rheological studies were an average of the ramp-up and ramp-down readings. These
are in general very close, though it was noted in the present study that values increased
significantly during ramp-down, for all superplasticizers, at 80C (176F) showing an
100SP-302-07
Table 3Rheological properties at 25 C (77F) and 10 MPa (1,450 psi)

PNS
PAS
PC
0.0
0.2 (0.18)*
0.4 (0.36)
0.5 (0.45)
0.6 (0.54)
1.0 (0.90)
0.0
0.2 (0.18)
0.4 (0.36)
0.6 (0.54)
0.8 (0.72)
0.0
0.2 (0.04)
0.4 (0.08)
0.6 (0.12)
0.8 (0.16)
Bingham Plastic
PV (cP)
YP (Pa)
38.0
23.3
19.6
23.0
13.8
12.8
16.4
4.42
16.4
0.83
14.3
0.80
38.0
23.3
19.6
21.8
14.1
1.62
17.2
1.07
17.6
0.24
38.0
23.3
17.8
9.13
15.2
3.02
13.0
1.54
19.1
1.03
n
0.161
0.077
1.116
0.817
0.825
0.850
0.161
0.304
0.730
0.827
0.915
0.161
0.905
0.776
0.818
0.869
Herschel-Bulkley
Yield stress (Pa)
K (Pasn)
12.72
17.0
19.12
15.0
0.006
12.9
0.048
4.15
0.047
0.57
0.035
0.61
12.72
17.0
1.890
17.5
0.072
1.22
0.048
0.80
0.023
0.00
12.72
17.0
0.031
8.99
0.059
2.69
0.039
1.32
0.041
0.82
* () polymer component

PNS
PAS
PC
0.0
0.2 (0.18)*
0.4 (0.36)
0.5 (0.45)
0.6 (0.54)
1.0 (0.90)
0.0
0.2 (0.18)
0.4 (0.36)
0.6 (0.54)
0.8 (0.72)
0.0
0.2 (0.04)
0.4 (0.08)
0.6 (0.12)
0.8 (0.16)
Bingham Plastic
PV (cP)
YP (Pa)
48.6
26.0
40.9
39.2
17.2
23.3
10.0
10.6
10.5
2.74
10.6
1.91
48.6
26.0
18.2
27.6
14.1
4.73
15.0
1.63
16.0
0.57
48.6
26.0
21.7
17.8
11.9
5.77
9.50
2.37
11.9
2.42
n
0.163
0.105
0.412
1.274
0.803
0.613
0.163
0.202
1.010
0.731
0.847
0.163
0.167
0.916
0.744
0.916
Herschel-Bulkley
k
Yield stress Pa)
17.05
18.0
29.69
17.5
0.302
22.2
0.002
10.8
0.035
2.54
0.116
1.38
17.05
18.0
4.050
20.9
0.013
4.75
0.077
1.21
0.40
0.35
17.05
18.0
6.78
7.58
0.019
5.69
0.046
2.07
0.019
2.34
approximately 20% difference at the lower shear rates. It is assumed that this relates to
gelation effect where PNS > PC > PAS as indicated by the apparent plastic viscosity and
yield point values. This was also confirmed by the thickening time (API RP 10B-2 (2010)
Table 5Rheological properties at 60 C (176F)and 10 MPa (1,450 psi)

PNS
PAS
PC
0.0
0.2 (0.18)*
0.4 (0.36)
0.5 (0.45)
0.6 (0.54)
1.0 (0.90)
0.0
0.2 (0.18)
0.4 (0.36)
0.6 (0.54)
0.8 (0.72)
0.0
0.2 (0.04)
0.4 (0.08)
0.6 (0.12)
0.8 (0.16)
Bingham Plastic
PV (cP)
YP (Pa)
83.9
51.8
50.5
48.5
3.8
38.0
2.6
32.2
4.6
14.3
10.2
4.39
83.9
51.8
23.1
32.7
6.7
9.46
9.0
3.47
15.9
2.17
83.9
51.8
33-3
25.6
6.3
10.2
7.4
5.50
7.6
3.60
n
0.149
0.106
1.290
0.880
1.186
1.240
0.149
0.523
1.326
0.778
0.743
0.149
0.121
0.817
1.005
0.979
Herschel-Bulkley
Yield stress (Pa)
k Pasn)
38.83
40.0
36.58
38.0
0.007
38.1
0.006
32.1
0.001
14.4
0.002
4.54
38.83
40.0
0.493
30.2
0.001
9.58
0.035
3.26
0.767
1.73
38.83
40.0
18.77
21.0
0.020
10.0
0.007
5.49
0.008
3.57

PNS
PAS
PC
0.0
0.2 (0.18)*
0.4 (0.36)
0.5 (0.45)
0.6 (0.54)
0.0
0.2 (0.18)
0.4 (0.36)
0.6 (0.54)
0.8 (0.72)
0.0
0.2 (0.04)
0.4 (0.08)
0.6 (0.12)
0.8 (0.16)
Bingham Plastic
PV (cP)
YP (Pa)
117.0
88.9
95.3
66.8
75.5
68.2
24.9
48.6
2.0
18.0
116.4
89.2
39.5
44.4
3.9
14.5
5.0
13.8
5.7
12.4
116.1
89.3
58.6
47.4
17.8
28.3
17.5
17.7
9.6
19.1
n
0.174
0.118
0.106
0.500
1.014
0.174
0.241
0.533
1.322
1.312
0.168
0.113
1.484
1.529
0.477
Herschel-Bulkley
Yield stress (Pa)
k (Pasn)
33.65
70.0
52.23
50.0
51.82
37.5
43.70
46.0
0.002
18.0
33.46
41.3
6.252
33.0
0.073
14.2
0.001
13.9
0.001
12.5
35.50
75.0
35.70
39.0
0.001
28.7
0.0004
18.7
0.276
17.9
plots, Figure3, where at 70 minutes, Bearden units of consistency (Bc) values were, 20 Bc,
15 Bc and 11 Bc for PNS, PC and PAS respectively. These plots also indicated different
profiles for each superplasticizers in building up consistency to 70 Bc where the slurry is
102SP-302-07
Fig 2. Shear stress/shear rate plots showing a)Bingham Plastic model, b)Herschel-Bulkley
model (n = 1.186) for 0.6% bwoc PNS slurry at 60C(140F), and c)Bingham Plastic
model, d)Herschel-Bulkley model (n = 1.274) for 0.5% bwoc PNS at 40C(104F).
considered no longer pumpable. Bc unit is a dimensionless quantity with no direct conversion factor to the more common units of viscosity.
Effect of superplasticizer concentration on rheology
The trends observed on increasing superplasticizer concentration and illustrated in
Figure4 are consistent with that observed previously (Michaux, M., et al. 1986). Notably
there is, in a number of cases, a minimum concentration below which there is only a minimal
effect on plastic viscosity and yield point at 0.2% bwoc or below, suggesting incomplete
interaction with the cement grain surfaces or modification of the surface charge. Optimum
values for dispersions, in all cases, are around 0.3 to 0.4% bwoc for the PAS and PC and
0.5% bwoc for the PNS. At concentrations above this there is little change in PV until
around 0.8% bwoc, where there is slight increase, though YP continues to fall. Increase in
PV can be attributed to too high a concentration of superplasticizer in the aqueous solution
Figure 3 Thickening time curves for LS-R retarded cement

with superplasticizer at 80C(176F)
Figure 4 Concentration effect on a) plastic viscosity and b) yield point

that causes bridging between particles giving rise to particle agglomeration and segregation that is detrimental to cement slurry stability. Although the PC superplasticizer used in
this study shows similar results to the PNS and PAS, it composes only 20% of the polycarboxylate polymer component compared to 90% polymer component for the PNS and PAS.
On this basis the concentration of polycarboxylate polymer if used by itself and not diluted
with inert material would give a value of around 0.04% bwoc as the minimum concentration required, 0.08% bwoc as the optimal concentration and over-dispersion occurring at
0.16% bwoc.
Impact of temperature on the rheology of superplasticized retarded slurries
Rheological data of PNS, PAS and PC superplasticizers on Class G cement slurries
containing LS-R retarder and antifoam GP-A at temperatures up to and including 80C
(176F) have been discussed based of rheological models, Table 3, Table 4, Table 5, and
Table 6. A closer investigation on the rheological data, Figure 5, shows that there are,
104SP-302-07
Figure5Temperature effect on plastic viscosity a) 40C, c) 60C, e) 80C and yield point
b) 40C, d) 60C and f) 80C.
however, notable differences on the behavior of the different superplasticizers. Plastic
viscosity and yield values increase with temperature in the absence of superplasticizer
increases from 38 cP and to 117 cP and 23.3 Pa to 88.9 Pa respectively over the 25C
(77F) to 80C (176F) temperature range. Addition of superplasticizer shows a decrease
in both plastic viscosity, Figure 5a, c and e and yield point, Figure 5b, d and f. The PAS and
PC superplasticizers show similar trends with the PAS showing a greater rate of decrease in
plastic viscosity and yield point compared to PC as concentration increases, with optimal
values being around 0.3 to 0.4% bwoc. The higher yield point, 28.3 Pa of the PC, and lower
plastic viscosity of 17.8 cP may have potential to minimize the effect of thermal thinning
that occurs at high temperature. Although the PC produces trends similar to PAS it is at a
much lower dosage given that the PC has only 20% polymer component compared to the
PAS which has 90% polymer component. The PNS superplasticizer by comparison shows
a somewhat different trend that becomes more evident with increase in temperature. PNS

PNS
PAS
PC
0.0
0.1 (0.09)*
0.2 (0.18)
0.4 (0.36)
0.0
0.06 (0.054)
0.1 (0.09)
0.2 (0.18)
0.4 (0.36)
0.0
0.04 (0.008)
0.1 (0.02)
0.2 (0..04)
0.4 (0.16)
Bingham Plastic
PV (cP)
YP (Pa)
26.6
8.2
19.4
4.2
11.6
1.9
6-6
1.0
26.6
8.2
24.9
5.7
20.0
2.3
12.9
1.6
26.6
28.0
17.4
10.4
12.2
8.2
9.8
4.8
2.3
2.3
n
0.334
0.455
0.959
1.008
0.334
0.454
0.675
0.749
0.334
0.243
0.410
0.825
0.815
Herschel-Bulkley
Yield stress (Pa)
k (Pasn)
2.009
3.6
0.610
2.3
0.014
1.9
0.006
1.0
2.009
3.6
0.791
3.2
0-147
1.5
0.060
1.2
2.009
4.363
0.751
0.030
0.037
3.6
1.8
2.7
2.1
2.1
requires a minimal concentration before there is an obvious decrease in plastic viscosity

or yield point and this value appears to change with temperature being around 0.3% bwoc
at 40C (104F) to 60C (140F) and 0.4% bwoc at 80C (176F) and with minimal or
optimum values obtained at about 0.6%. The effects observed both in rheology and thickening time can be related to an acceleration of the aluminate phase hydration resulting in a
precursor ettringite gel that increases viscosity and in addition retards silicate hydration by
coating on the cement grains (Luke K. and Aitcin, P.-C., 1991). Clearly a higher concentration of PNS is required in comparison to PAS and even more to PC to achieve the same
rheological results with LS-R retarder and PG-A antifoaming agent.
Rheological profiles and effect of superplasticizer at 100 -120C (212
248F)
The data for PNS, PAS and PC on the rheological properties are given in Table 7 and Table
8. Clearly values are much lower than those observed with the LS-R retarded systems and
there appears to be little difference in the rheological performance at either 100C (212F)
or 120C (248F) and irrespective of the superplasticizer used. Trends are however similar
to those obtained at lower temperatures with slight to no effect at lower concentrations, and
over-dispersion at higher concentrations giving rise to particle segregation and free water
formation over time consistent with the low yield stresses observed. Optimum concentration of superplasticizer to decrease the plastic viscosity and yield point in this case occurs
in the range of 0.1 to 0.2% bwoc. However, the low rheological values obtained in the
cement slurry without superplasticizer are more than sufficient to provide good slurry flowability, and is a result predominantly of the HC-R retarder. LS-R and HC-R although used
primarily for retarding cement setting both have secondary effects of acting as dispersant.
As evidenced by the data the HC-R has a much stronger dispersing effect than the LS-R
and in the concentrations required to provide adequate pumping time at 100C (212F)
106SP-302-07

PNS
PAS
PC
0.0
0.1 (0.09)*
0.2 (0.18)
0.4 (0.36)
0.0
0.06 (0.054)
0.1 (0.09)
0.2 (0.18)
0.4 (0.36)
0.0
0.04 (0.008)
0.1 (0.02)
0.2 (0..04)
0.4 (0.16)
Bingham Plastic
PV (cP)
YP (Pa)
25.4
7.1
23.9
6.5
15.6
1.9
11.7
1.5
25.4
7.1
24.9
6.4
20.0
3.4
12.9
2.0
25.4
24.1
19.2
16.3
11.5
7.1
5.7
4.5
3.3
3.4
n
0.584
0.629
0.764
1.208
0.584
0.395
0.534
0.907
0.584
0.602
0.669
0.617
0.581
Herschel-Bulkley
Yield stress (Pa)
k (Pasn)
0.338
5.7
0.236
5.4
0.066
1.5
0.003
1.7
0.338
5.7
0.342
2.9
0.371
1.9
0.025
1.8
0.338
0.287
0.148
0.177
0.157
5.7
4.4
3.7
2.4
2.8
and 120C (248F), 0.5% bwoc, it also dominates rheological properties in relation to the
superplasticizers. This is consistent with previous observations using sodium gluconate in
combination with polycarboxylate superplasticizer which allowed a reduction of approximately 50% of the dosage of the superplasticizer (Plank J. et al. 2009). It is also of note that
in this case the slurries tend to follow the Herschel-Bulkley model more closely than the
Bingham plastic model, at most superplasticizer concentrations used particularly at 120C
(248F) as illustrated for selected samples in Figure 6. Figure 6 a) and b) show the shear
stress/shear rate plots comparing the Bingham Plastic model with the Herschel-Bulkley
model for 0.06% bwoc PC at 120C (248F), and Figure 6 c) and d) show the comparison
at 0.4% bwoc PC at 120C (248F) illustrating the Herschel-Bulkley model as giving the
best fit. Figure 6 e) and f) for 0.2% bwoc PNS at 120C (248F) shows that the HerschelBulkley model gives the better fit with n equals 0.764 although the Bingham Plastic model
is relatively close.
Superplasticizer effect on calorimetric and zeta potential of Class G slurry
Variations in the heat evolution of the LS-R retarded Class G cement with and without
PNS, PAS and CF at 40C (104F) is shown in Figure 7. The retarded Class G cement shows a
heat flow curve typical of cements cured at 25C (77F) where the induction period, second
hydration peak from formation of C-S-H and CH from hydration of the silicate phases and
then a third peak observed as a shoulder occurring after the silicate hydration and attributed
to ettringite (AFt) to monosulfate (AFm) conversion (Ghose, A. and Pratt, P.L., 1981). The
initial peak due to wetting and initial dissolution can be observed as the decline of the peak
since the main reaction has occurred during the mixing procedure that is performed outside
the calorimeter. In addition there is an endothermic peak that occurs during temperature
ramp-up and is attributed to non-equilibrium conditions within the calorimeter (Luke, K.,
2011). The PNS shows a significant increase in the induction period retarding the hydration
Figure 6 - Shear stress/shear rate plots showing a)Bingham Plastic model, b)HerschelBulkley model for 0.06% bwoc PC slurry at 120C(248F), c)Bingham Plastic model,
d)Herschel-Bulkley model for 0.4% bwoc PC at 120C(248F)and e) Bingham Plastic
model, f)Herschel-Bulkley model for 0.2% bwoc PNS at 120C(248F).
108SP-302-07
Figure 7 Heat flow curves showing the effect of superplasticizers at 40C (104F)
from 7.5 hours to 10 hours. The increased heat flow indicates though a faster reaction rate
for the silicate reaction and more notably the conversion of AFt to AFm which appears to
occur at about the same time (bimodal peak). The PAS causes a decrease in the induction
period from 7.5 hours to 6.5 hours and a notable increase in the rate of the reaction though
in this case the AFt to AFm conversion is less pronounced. The PC appears to have the
least effect with perhaps a slight increase in the rate of reaction compared to the retarded
Class G without superplasticizer. At 120C (248F) the curves differ from that obtained at
40C (104F), the induction period is shorter based on the different retarder used and also
shows a different heat flow curve typical of retarded cement slurries at temperatures of
90C (194F) and above (Luke, K., 2011). In this case the decline in the initial hydration
peak is clearly evident for the HC-R retarded cement whereas the superplactized cements
show a peak maxima. However it is not possible to interpret the data at this point as the
calorimeter is in the temperature ramp-up process and is not in equilibrium. Equilibrium
is only established when temperature has been reached and the heat flow value is around 0
mW which coincides with the end of the endothermic peak. This is notably longer at higher
temperatures as ramp-up time is longer. An additional peak observed before the silicate
hydration peak is attributed to aluminate phase hydration as previously indicated by XRD
(Luke, K., 2011). The addition of superplasticizer to the HC-R retarded Class G cement at
a temperature of 120C (248F) shows minimal effect with perhaps a slight acceleration
with PNS and retardation in the induction period with PC, Figure8. The rate of reaction and
the peaks observed are the same with or without the superplasticizer indicating at higher
temperature the effect of the retarder is more dominant.
Zeta potential values of the PNS, PAS, and PC on the retarded cement systems in both
RO water and in pore solution extracted from a Class G cement hydrated for approximately
70 minutes equivalent to time of rheology testing is given in Table 9. The high negative
value of the retarded cement system in RO water without superplasticizer is consistent
with that determined previously, -17 to -15 mV (Ngele, E., 1986) at short hydration times
Figure 8 Heat flow curves showing the effect of superplasticizers at 120C (248F)
indicative of the retardation of reaction by the LS-R and HC-R. Addition of superplasticizer notably decreases the value to even higher negative values consistent with deflocculation of the particles. However, there was no differentiation in zeta potential of the PNS,
PAS or PC. Alkali is known to influence the zeta potential both based on pH value and
on the ions present, K+ or Na+ (Ngele, E., 1986). In order to simulate the solution to that
of the cement in terms of pH and ionic composition the pore solution of the Class G was
extracted after approximately 70 minutes of hydration. Zeta potential values as expected
were significantly less negative than obtained with the RO and is attributed to the fact that
the increased ionic strength compresses the diffuse double layer (Ngele, E., 1987). Values
were close to zero and consistent with those obtained on well cement slurries (Hodne, H.
and Saasen, A., 2000). In this case variations in value were noted with the different superplasticizers where the PNS and PAS give negative values and the PC gives a positive value
and the validity of the data is being further investigated.
This preliminary study investigated the effect of three different types of superplasticizer on the rheological characteristics of two different retarded cement systems that also
included an antifoam additive at 25C (77F) to 120C (248F). The Herschel-Bulkley
model give the best fit for all slurries studied though where the value of n ranged from
about 0.5 to about 1.2 the Bingham Plastic model was also noted to give a good fit to the
shear stress/shear rate plots of the data. This was particularly more notable as n approached
1, consistent with the fact that the Herschel-Bulkley model reduces to the Bingham Plastic
model n = 1. PNS and PAS, in some cases, showed shear thickening effects, n greater
than 1, though there was no correlation to superplasticizer concentration or temperature.
Optimal concentration of superplasticizer was around 0.3 to 0.4% bwoc for the PAS and
PC superplasticizers, though given that the PC is only 20% active the optimal concentration for pure PC is closer to 0.16% bwoc and for PNS the least effective it is 0.5% bwoc.
The PNS was also observed to have a lesser effect in controlling plastic viscosity and yield
110SP-302-07
Table 9 Zeta Potential of PNS, PAS and PC on Retarded Class G slurries

Retarder
Aqueous phase
RO water
LS-R
Pore solution
RO water
HC-R
Pore solution
SP
No SP
PNS
PAS
PC
No SP
PNS
PAS
PC
No SP
PNS
PAS
PC
No SP
PNS
PAS
PC
Zeta Potential (mV)

-16.6
-22.5
-22.5
-23.2
1.9
-2.9
-5.5
1.8
-17.0
-24.0
-18.7
-22.2
-4.5
-0.3
-1.2
1.4
point in comparison to PAS and PC superplasticizers at higher temperature and this may
be attributed to acceleration of the aluminate phase hydration and formation of a precursor
ettringite gel. Calorimetric results showed that the PNS had a significant retarding effect
compared to the PAS which was slightly accelerating and the PC which had little effect
except to slightly increase reaction rate of the silicate phases. At temperatures of 100C
(212F) and 120C (248F) the slurries followed a Herschel-Buckley model irrespective
of the concentration of superplasticizer used. The Bingham Plastic model showed a good
fit to the shear stress/shear rate plot of the data only at n values above about 0.8. Plastic
viscosity and yield point values were notably lower than at temperatures of 20C (77F) to
80C (176F) and was dominated by the HC-R retarder which also shows strong dispersing
properties. This was also confirmed by calorimetry which showed minimal effect of the
superplasticizers. Zeta potential showed minimal differences between superplasticizers in
RO water irrespective of the retarder used though there were some slight differences in
cement pore solution in that PC had a positive value compared the negative value of the
PNS and PAS. These are being further investigated.
AUTHOR BIOS
ACI member Karen Luke is Principal Research Advisor - Cement at the R&D Centre,
Trican Well Service, Calgary, Alberta, Canada. She received her PhD from the University of Aberdeen, Scotland, UK. Her primary research interests are in the chemistry of
cement, cement additive interactions, durability of cement, hydrothermal phase equilibria
of well cements, well cement performance and mechanical properties.
Adrian Torres is research Technical Specialist at the R&D Centre, Trican Well Service,
Calgary, Alberta, Canada. He received his BSc at the University of Calgary, Alberta,
Canada. His current interests are cement additive interactions as well as enhancing the
mechanical properties of oil well cement.
REFERENCES
Aitcin, P.-C., (1998), High Performance Concrete, Ed. Aitcin, P.-C., E & FN Spon
Publisher, 591 pp.
API RP 10B-2 (2010), Recommended Practice for Testing Well Cements, Washington,
DC:API.
Aukett, P. N., and Bensted, J.1992 , Application of Heat Flow Calorimentry to the
Study of Oilwell Cements, Journal of Thermal Analysis, V. 38, No. 4, pp. 701-707. doi:
10.1007/BF01979399
Ghose, A., and Pratt, P. L., (1981), Studies of the Hydration Reactions and Microstructure of Cement-Flyash Pastes, Materials Research Society, Effects of Flyash Incorporation in Cement and Concrete, Proceedings, Symposium N, Annual Meeting Nov 16-19, pp.
82-91.
Guillot, D., (2006), Rheology and Flow of Well Cement Slurries, In Well Cementing,
Eds. E.B. Nelson and D. Guillot, Schlumberger Publisher, pp. 93 - 189.
Hattori, K.1978 , Experiences with Mighty Superplasticizer in Japan, ACI Publication,
V. SP-62, pp. 37-61.
Hodne, H., and Saasen, A.2000 , The Effect of Cement Zeta Potential and Slurry
Conductivity on the Consistency of Oilwell Cement Slurries, Cement and Concrete
Research, V. 30, No. 11, pp. 1767-1772. doi: 10.1016/S0008-8846(00)00417-8
Luke, K., (2011), Conduction Calorimetry and X-Ray Diffraction Investigation of
Cement Retardation at 70 120C, 13th International Congress on the Chemistry of
Cement, Madrid Jul 3-8, pp 6.
Luke, K., and Aitcin, P.-C.1991 , Effect of Superplasticizer on Ettringite Formation,
Ceramic Transactions on Advances in Cementitious Materials, V. 16, pp. 147-166.
Michaux, M., and Defosse, C.1986 , Oilwell Cement Slurries Pt.1: Microstructural
Approach of their Rheology, Cement and Concrete Research, V. 16, No. 1, pp. 23-30. doi:
10.1016/0008-8846(86)90064-5
Michaux, M.; Oberste-Padtberg, R.; and Defosse, C.1986a, Oilwell Cement Slurries
Pt.2: Adsorption Behaviour of Dispersants, Cement and Concrete Research, V. 16, No. 6,
pp. 921-930. doi: 10.1016/0008-8846(86)90016-5
Ngele, E.1986 , The Zeta-Potential of Cement Part II: Effect of pH Value, Cement
and Concrete Research, V. 16, No. 6, pp. 853-863. doi: 10.1016/0008-8846(86)90008-6
Ngele, E.1987 , The Zeta-Potential of Cement Part III: The Non-Equilibrium Double
Layer on Cement, Cement and Concrete Research, V. 17, pp. 573-580.
Nelson, E. B.; Michaux, M.; and Drochon, B., (2006), Cement Additives and Mechanisms of Action, In Well Cementing, Eds. E.B. Nelson and D. Guillot, Schlumberger
Publisher, pp. 71-80.
Plank, J., Schrfl, C. and Gruber, M., (2009), Use of Supplemental Agent to Improve
Flowability of Ultra-High-Performance Concrete, SP-262-1, Superplasticizers and Other
Chemical Admixtures in Concrete, pp. 1-16.
Sairam, P. K. S.; Morgan, R.; and Pangu, G., (2012), A Combined Mixer Design with
Helical Blades to Probe Rheology of Complex Oilfiled Slurries and Pastes, Paper SPE
159112, Presented at SPETT 2012 Energy Conference and Exhibition, Port of Span, Trinidad, June 11-13.
112SP-302-07
Schmidt, W., (2014), Design Concepts for the Robustness Improvement of SelfCompacting Concrete Effects of Admixtures and Mixture Components on the Rheology
and Early Hydration at Varying Temperatures, PhD Thesis, Eindhoven University of Technology, The Netherlands, pp. 308.
Specification, A. P. I., 10A (2011), Specification for Cements and Materials for Well
Cementing, Washington DC:API
Walraven, J. C., (2010), Self Compacting Concrete: Properties, Development and Code
Recommendations, 6th International RILEM Symposium on Self-Compacting Concrete
and 4th North American Conference on the Design and Use of SCC, Montreal, Canada,
pp. 25-44.
SP-302-08
Properties of a New Type of

Polycarboxylate Admixture for Concrete
Using High Volume Blast Furnace Slag
Cement
by Shinji Tamaki, Kazuhide Saito, Kazuhisa Okada,
Daiki Atarashi, and Etsuo Sakai
Several studies have been pursued in Japan on developing concrete using high volume
blast-furnace slag cement for reducing CO2 emissions arising from calcination of
cement. However, when using high volume blast-furnace slag cement, various problems
are encountered, such as decreased fluidity retention ability caused by the reduction of
admixture dosage and decreased strength enhancement. In this paper, the authors focus
on the adsorption properties of polycarboxylate ether superplasticizers and the properties
of hardened concrete that incorporates a component of high volume blast-furnace slag
cement, and discuss the development of a new type of superplasticizer through molecular
design and optimization of the admixture composition. The admixture improved the fluidity
and properties of hardened concrete using slag cement containing more than 60% blastfurnace slag.
Keywords: adsorption; blast-furnace slag; blast-furnace slag cement; chemical admixture; fluidity retention ability.
INTRODUCTION
Presently, 5060 million tons of cement are produced per year in Japan, and CO2 emissions during cement production, including emissions separated and discharged from the
limestone mineral, account for 3%4% of the total CO2 emissions in Japan. Attempts to
reduce CO2 emissions derived from cement production have been ongoing for some time.
Industrial byproducts, such as blast-furnace slag (BFS) and fly ash (FA), have been used
quite effectively, and blended cements comprising mixtures of these byproducts with ordinary Portland cement (OPC) are normalized in Japanese Industrial Standards (JIS). Blended
cement tends to possess low early strength, high drying shrinkage, and high carbonation
rate. Therefore, both BFS and FA cements are typically used in the Type B formulation.
The types of blended cement defined in JIS are shown in Table 1.
113
114SP-302-08
Table 1Standards of blended cement (JIS)
BFS cement
FA cement
Replacement ratio, %
Type B
30 < BFS 60
10 < FA 20
Type A
5 < BFS 30
5 < FA 10
Type C
60 < BFS 70
20 < FA 30
* BFS cement: JIS R 5211 **FA cement: JIS R 5213
Table 2Physical and chemical properties of OPC and BFS

Code
OPC
BFS4000
BFS6000
Binder type
Ordinary
Portland cement
Blast-furnace
slag
Blast-furnace
slag
Blaine
Density, fineness,
g/cm3
cm2/g
SiO2
Chemical composition, %
Al2O3 Fe2O3 CaO MgO SO3 Na2O K2O
3.16
3310
21.56
4.68
2.98
65.63 1.30
1.30
0.33
0.39
2.91
4570
33.60
14.80
0.29
42.30 6.30
0.21
0.35
2.91
6130
33.70
14.50
0.42
42.60 6.10
0.19
0.38
To significantly reduce the CO2 emissions, the authors have been developing a high
volume BFS cement (denoted as Energy CO2 minimum; ECM) containing more than 60%
BFS by weight.1 To improve low early strength and reduce drying shrinkage, we have
adjusted the cement composition; to address the high carbonation rate, we have selected
the concrete structures in addition to adjusting the cement composition.
Furthermore, for practical use, the chemical admixture for concrete derived from ECM
requires a high fluidity retention ability that is caused by a reduction in the admixture dosage
compared with that when using OPC or low-volume-blended cement (for example, BFS
cement Type B). The focus of this paper is the adsorption of polycarboxylate ether (PCEs)
superplasticizers in ECM and the resulting fluidity, and the development of a new type of
superplasticizer through molecular design and optimization of the admixture composition.
The authors report on a new admixture for ECM containing more than 60% blast-furnace
slag. The new admixture ameliorates the problems encountered when using ECM, such as
the concerns associated with the fluidity and properties of the hardened concrete. ECM
greatly reduces CO2 emissions from cement production by contributing to the expansion
of the application range of BFS cements and by furthering efforts toward environmental
preservation.
Materials
Two types of BFS with fineness of 4570 cm2/g (hereafter BFS4000) and 6130 cm2/g
(hereafter BFS6000), were used. In addition to BFS, OPC and ECM were used. The physical and chemical properties of OPC and BFS are listed in Table 2. ECM was blended
in the ratio of OPC: BFS: anhydrite (fineness 3890 cm2/g) = 30: 63: 7. ECM containing
Properties of a New Type of Polycarboxylate Admixture for Concrete Using

High Volume Blast Furnace Slag Cement 115
Fig. 1Molecular structure of superplasticizer

Table 3Mortar mixture proportions
W/B
0.4
Target air, %
2.0
W
300 [10.6]
Weight, g [oz.]
B
1000 [35.3]
S
1500 [52.9]
* Water-to-binder ratio (W/B), Water (W), Binder (B), and Fine aggregate (S)
BFS4000 is denoted as ECM4000, and that containing BFS6000 is denoted as ECM6000.

These blends were designed to improve early strength and reduce drying shrinkage by
increasing the amount of SO3 relative to that used in general-use BFS cement.2,3
The fine aggregate (S) comprised land sand with saturated surface dry (SSD) density of
2.58 g/cm3 and fineness modulus of 2.87. The coarse aggregate (G) was crushed stone with
SSD density of 2.68 g/cm3 and solid volume of 60%.
Chemical admixtures
The superplasticizers used in this study were PCEs with methoxy polyethylene glycol
graft chains. A commercially available product (hereafter PCE) for the OPC concrete was
used as reference. Two types of polymers (M23 and M9) of different side chain lengths and
methacrylic acid (MAA) to polymer ratio were synthesized and used in the development
of the admixtures for ECM. For M23, the side chains comprised 23 mol of ethylene oxide,
and the molar ratio of MAA to polymer was 69%. For M9, the side chains comprised 9 mol
of ethylene oxide, and the molar ratio of MAA to polymer was 40%. In addition, in this
work, the concentration of the superplasticizer added was converted to a solid component.
The molecular structure of the superplasticizer is shown in Fig. 1. The superplasticizers
were characterized by gel permeation chromatography.
Sodium gluconate (C6H11O7Na; GLNa) was used as the slump retention agent.
Mortar test
Table 3 shows the mortar mixture proportions. The binder consisted of OPC and
BFS4000 in different ratios. The mortar test was conducted in accordance with JIS R 5201;
the extent of mortar flow on removal of the cone was measured.
Paste test
Paste specimens were obtained at a water-to-cement ratio (W/C) of 0.32 by mixing for 5
min, allowing the mixture to stand for 4 min, and mixing again for 1 min. The paste flow
116SP-302-08
Table 4Mixture proportions of concrete

W/C
0.5
0.4
Cement
ECM6000
Target slump,
cm [in.]
Target air,
%
182.5
[7.11.0]
4.51.0
Unit weight, kg/m3 [lb/yd3]

W
C
S
G
160 [270] 320 [539] 854 [1439] 951 [1603]
160 [270] 400 [674] 740 [1247] 997 [1680]
* Water (W), Cement (C), Fine aggregate (S), and Coarse aggregate (G)
was measured in accordance with JASS15M-103; the extent of paste flow on removal of
the cone was measured.
Adsorption of superplasticizer
Paste mixed in the same ratio as that for the paste flow measurement was separated
into the solid and liquid phases by a centrifuge operating at 5000 rpm for 10 min. The
amount of organic carbon in the liquid phase was subsequently measured by a total organic
carbon analyzer (TOC5050A, Shimadzu Corporation), and the adsorbed superplasticizer
amount per unit mass was calculated. In addition, the hydration of the paste was stopped
using a large quantity of acetone, and drying under reduced pressure (<5kPa) for 24 h. The
BrunauerEmmettTeller specific surface area of the test specimen was measured using
a surface area analyzer (Gemini V2380, Shimadzu Corporation). These results were then
used to determine the amount of the superplasticizer adsorbed per unit surface area.
Concrete test
Table 4 lists the mixture proportions of the concrete. The binder used was ECM6000.
ECM6000 has been confirmed to result in a higher strength development than ECM4000.
The slump was measured in accordance with JIS A 1150. Changes in the fluidity of fresh
concrete were measured after the concrete was allowed to rest for 30, 60, and 90 min in a
mixing vessel; the concrete was re-mixed prior to each measurement. The setting time was
measured in accordance with JIS A 1147, using penetration resistance measurements on
mortar sieved from the concrete. Compressive strength was determined in accordance with
JIS A 1108, using 200 mm [8.0 in.] 100 mm [4.0 in.] cylindrical specimens. The specimens were demolded 24 h after mixing. Some specimens were tested immediately, whereas
the remaining specimens were tested after water curing at 20 C [68 F] for 7 and 28 days.
Influence of the replacement ratio of BFS on the fluidity of mortar
Fig. 2 shows the relation between the replacement ratio of BFS to OPC and the mortar
flow. The superplasticizer used was PCE and the ratio of the additive to the binder was
0.20% by mass. The fluidity of BFS alone was low; however, the fluidity tended to increase
with increasing BFS4000 substitution. Thus, the amount of admixture added can be significantly reduced when using ECM.
Relation between the polymer structure and the adsorption of polymer
The adsorbed amounts of each powder per 1.0 g [3.5 10-2 oz.] of M23 and M9 for the
ECM admixture are shown in Fig. 3 and 4, respectively.4 For all powders, the adsorbed

Fig. 2Relation between the replacement ratio of BFS4000

and the mortar flow (20 C, [68 F])
Fig. 3Adsorption of M23

amount of M23 increased with increasing additive but saturated at low additive amounts
relative to the OPC in BFS4000 and BFS6000. The amount of M23 adsorbed was higher
in ECM4000 and ECM6000 than in OPC despite the OPC replacement rate being as low
as 30%; this is due to the great increase in the adsorption to the BFS component in ECM.
This phenomenon is presumed to be consistent with the results of the mortar test, where
Ca2+ eluted from the OPC was specifically adsorbed on the BFS surface that becomes the
adsorbing surface of the admixture.
In M9, although adsorption to the binder tended to increase with increasing additive,
the amount adsorbed to the OPC reached saturation at a lower amount of addition, and the
amount adsorbed for BFS4000 and BFS6000 did not change as significantly as in M23.
118SP-302-08
Fig. 4Adsorption of M9
Fig. 5Flow of OPC paste and ECM paste with added M23
However, Fig. 4 shows that the amount of M9 adsorbed in ECM4000 and ECM6000 is
approximately as high as that observed in M23, as shown in Fig. 3. The phenomenon in
M9 is assumed to be the same as that observed in M23, wherein Ca2+ eluted from the OPC
is specifically adsorbed on the BFS surface.
Relation between polymer structure and paste flow
The relation between the dosage of M23 and M9 and the paste flow using OPC and ECM
are shown in Fig. 5 and 6, respectively. In M23, the results of the paste test demonstrated
high fluidity for both OPC and ECM (especially for ECM). Moreover, a high correlation

Fig. 6Flow of OPC paste and ECM paste with added M9

is observed between paste flow and the amount of M23 adsorbed, as shown in Fig. 3,
where both the adsorbed amount of M23 and the fluidity increased with increasing additive
content. In M9, the paste fluidity of both OPC and ECM increased with increasing admixture content; however, the effect on OPC was small. A correlation is observed between
paste flow and the amount of M9 adsorbed, as shown in Fig. 4.
Based on these results, M23 is concluded to adsorb onto both OPC and BFS in ECM
and it enhances the fluidity; the adsorption of M9 to BFS also significantly contributes to
fluidity enhancement.
Properties of concrete
The results for concrete at a W/C of 0.5 and 0.4 are listed in Table 5 and 6, respectively.
The admixtures used were PCE, M23, M9, and M9 with GLNa.
Time Dependent slump loss behavior
The results of the slump test at a W/C of 0.5 are shown in Fig. 7. In the case of PCE or
M23 added to ECM6000, the dosage was decreased and a large slump loss was observed
at 30 min after mixing. However, the dosage could be maintained in M9, and a significant
improvement in retention of up to 60 min was observed. In addition, further improvement
of retention was obtained by adding both M9 and GLNa, which retained sufficient fluidity
even after 90 min. Therefore, fluidity retention can be controlled by adjusting the ratio of
M9 and GLNa.
The results of the slump test at a W/C of 0.4 are shown in Fig. 8. In the case of PCE or
M23 added to ECM6000, the dosage was decreased and a large slump loss was observed
at 30 min, similar to that at a W/C of 0.5. In the case of M9, the improvement in retention
up to 60 min was observed. However, a slight initial growth of slump and a large slump
loss after 90 min was observed. Therefore, M9 and GLNa were combined with M23 for
suppressing the initial growth of slump because the combination demonstrated excellent
120SP-302-08
Table 5Concrete test results

Dosage of
admixture,
C%
Dosage of
GLNa,
C%
PCE
0.11
M23
0.05
M9
0.21
M9
0.20
0.04
PCE
0.10
M23
0.05
M9
0.20
M9+M23
0.13+0.04
0.04
Admixture
W/C
type
0.5
0.4
Slump, cm [in.] (Air content, vol%)

0 min
19.5
[7.7]
(4.0)
18.5
[7.3]
(4.2)
19.0
[7.5]
(4.4)
18.5
[7.3]
(4.1)
19.5
[7.7]
(4.2)
19.0
[7.5]
(4.5)
20.0
[7.9]
(4.2)
18.0
[7.1]
(4.4)
Setting time, h
30 min
60 min
90 min
initial
final
9.5 [3.7]
(3.3)
6.0
11.3
5.0 [2.0]
(3.0)
5.8
11.1
16.0
[6.3]
(4.0)
20.5
[8.1]
(4.1)
12.0
[4.7]
(3.8)
19.5
[7.7]
(4.2)
8.5
13.7
9.3
14.5
6.0 [2.4]
(3.4)
20.5
[8.1]
(4.1)
20.5
[8.1]
(4.2)
17.5
[6.9]
(4.0)
20.5
[8.1]
(4.2)
11.0
[4.3]
(3.3)
18.0
[7.1]
(4.2)
19.5
[7.7]
(4.4)
20.5
[8.1]
(4.0)
11.5
[4.5]
(3.7)
* % by mass of cement (C%)
Table 6Compressive strength

W/C
0.5
0.4
Admixture type
PCE
M23
M9
M9
PCE
M23
M9
M9+M23
Dosage of admixture, C%
0.11
0.05
0.21
0.20
0.10
0.05
0.20
0.13+0.1
Dosage of
GLNa, C%
0.04
0.04
Compressive strength, MPa [psi]

24 h
7 days
28 days
2.7 [390]
30.9 [4480]
43.8 [6350]
2.7 [390]
31.0 [4500]
43.6 [6320]
2.5 [360]
32.8 [4760]
47.3 [6860]
2.0 [290]
33.3 [4830]
47.7 [6920]
4.1 [600]
41.0 [5950]
58.3 [8460]
4.2 [610]
40.8 [5920]
58.2 [8440]
3.4 [490]
42.9 [6220]
60.6 [8790]
3.1 [450]
43.5 [6310]
60.9 [8830]
* % by mass of cement (C%)
initial dispersion power. Thus, fluidity up to 90 min was ensured without the initial growth
of the slump.
These results show that M9 displays better retention ability than PCE or M23. This
is likely due to the maintenance of the adequate dosage and the suppression of the early

Fig. 7Slump retention behavior (W/C of 0.5)
Fig. 8Slump retention behavior (W/C of 0.4)

admixture consumption caused by the early hydration of OPC by the small molar ratio of
MAA and the short side chain length of M9.
GLNa adsorbs to ECM or OPC very easily, and the fact that it is adsorbed competitively is considered to be a reason for the effective functioning of GLNa for ECM, thereby
suppressing the early admixture consumption.
Compressive strength and setting time
As shown in Table 6, M9 and M9 + GLNa, although the compressive strength at 24 h
was slightly reduced, the compressive strength subsequently increased as compared to PCE
or M23. Fig. 9 shows the relation between the compressive strength (at 28 days) and the
setting time (final) at a W/C of 0.5. The observed increase in the compressive strength with
122SP-302-08
Fig. 9Relation between 28-d compressive strength and

setting time (W/C of 0.5)
Fig. 10Relation between 28-d compressive strength and

setting time (W/C of 0.38)
delay in the setting time when using ECM has already been shown, as will be discussed
shortly, and this was probably the effect.
Fig.10 shows the relation between the setting time (final) and the compressive strength
of concrete (W/C of 0.38) that used ECM4000. This is an experimental verification. The
setting time had been adjusted by changing the admixture and the amount of the admixture
(by changing the unit volume of water), and by using retarders. The compressive strength
of the concrete using ECM was shown to improve by delaying the setting time of the
concrete.
Although the definite potential for enhancing compressive strength using ECM has been
verified, further studies considering workability are necessary.

CONCLUSIONS
This study is summarized as follows.
1. M23 and M9, with varying methacrylic acid ratio and side chain lengths, demonstrated different adsorption behavior for OPC.
2. M23, which has a longer side chain length and a larger amount of carboxylic acid than
M9, demonstrated sizable absorbance to both OPC and ECM. In addition, on the basis of
the fluid tendency of the paste, M23 adsorbs to both the OPC and BFS components in ECM
and enhances fluidity.
3. M9, which has a shorter side chain length and a smaller amount of the carboxylic acid
than M23, exhibited a smaller absorbance to OPC, whereas a significant amount absorbed
to ECM. Additionally, on the basis of the fluid tendency of the paste, M9 is found to adsorb
to the BFS component of ECM, thereby enhance fluidity.
4. M23 and M9 impart high fluidity to the concrete made with ECM. In addition, fluidity
retention can be controlled using M23, M9, and GLNa.
5. M9 and GLNa improve the compressive strength of concrete made with ECM.
However, the early strength was found to decrease slightly. Further study is required to
overcome this problem.
Standards cited
Japanese Industrial Standards
JIS R 5201 Physical testing methods for cement
JIS A 1150 Method of test for slump flow of concrete
JIA A 1147 Test method for time of setting of concrete mixtures by penetration resistance
JIS A 1108 Method of test for compressive strength of concrete
JIS R 5211 Portland blast-furnace slag cement
JIS R 5213 Portland fly-ash cement
AUTHOR BIOS
Shinji Tamaki is a Research Engineer of R&D, Construction Chemicals Division,
Takemoto Oil & Fat Co., Ltd., 2-5 Minatomachi, Gamagori, Aichi, Japan. His current
research interests are chemical synthesis of water-soluble functional polymers and materials science.
Kazuhide Saito is a Research Engineer of R&D, Construction Chemicals Division,
research interests include fluidity, strength, and durability development for concrete using
high volume blast-furnace slag cement.
Kazuhisa Okada is a Research Engineer of R&D, Construction Chemicals Division,
research interests are chemical synthesis of water-soluble functional polymers and materials science.
Daiki Atarashi is an assistant professor of metallurgy and ceramics science, Graduate
School of Science and Engineering, Tokyo Institute of Technology, Japan. He received
124SP-302-08
his B.S. in 2001, M.S. in 2003, and Dr. Eng. in 2006 from Tokyo Institute of Technology.
His research interests include the action mechanisms of chemical admixtures, fluidity
of cement paste, and material design of high-recycled-content and low-CO2-emission
cements.
Etsuo Sakai is a professor in metallurgy and ceramics science, Graduate School of
Science and Engineering, Tokyo Institute of Technology. He received his Dr. Eng. from
Tokyo Institute of Technology in 1979. His research interests include construction chemistry, material recycling, and material design for low-carbon cement.
ACKNOWLEDGMENTS
This research was conducted under a grant for Strategic Development of Energy Use
Rationalization Technology / Leading Research and Development of Fundamental Technologies for Efficient Energy Use / Research and Development of Energy CO2, Minimum
(ECM) Cement-Concrete System from the New Energy and Industrial Technology Development Organization (NEDO).
REFERENCES
1. Yonezawa, T.; Sakai, E.; Koibuchi, K.; and Kinoshita, M., High-Slag Cement and
Structures for Substantial Reduction of Energy CO2, Proceedings of the fib Symposium,
Stockholm, Sweden, pp. 463-466 (2012)
2. Nito, N.; Osawa, T.; Koibuchi, K.; and Miyazawa, S., Heat and shrinkage of concrete
by slag gain and SO3 of blast-furnace slag cement (in Japanese), Proceedings of the Japan
Concrete Institute, Japan, Vol. 30, No.2, pp. 121-126(2008)
3. Tsuji, D.; Wachi, M.; Inoue, K.; Mitsui, K.; Yonezawa, T.; and Kanda, T., Properties
of Concrete using High Slag Cement, Proceedings of the First International Conference on
Concrete Sustainability, Tokyo, Japan, pp. 139-144 (2013)
4. Sasabe, T.; Atarashi, D.; Tamaki, S.; and Sakai, E., Adsorption Mechanism of Superplasticizer on High Volume Blast Furnace Slag Cement (in Japanese), Cement Science and
Concrete Technology, Japan Cement Association, Japan, No. 65, pp. 27-32 (2011)
SP-302-09
Synthesis and Properties of High SolidContent Polycarboxylate Superplasticizer

by Yongwei Wang, Liya Wang, Yongsheng Liu, and
Zepeng Chu
In this study, the polycarboxylate superplasticizers (PCs) with solid content up to 80%
were synthetized using special redox initiator at 318K. In the radical polymerization reactions, combining with Fourier Transform Infrared Spectroscopy (FTIR) and Gel Permeation Chromatograph (GPC), the initiator dosing dosage, reaction temperature, reaction
time and the concentration of system in the copolymerization reaction were systematic
investigated through orthogonal design experiments. The performances of new PCs in
cement paste were tested by measuring the fluidity and fluidity retention. The slump and
the compressive strengths of concrete were also determined. Compared with traditional
PC, the new PC has a better advantage in workability of fresh concrete and mechanical
properties of hardened concrete.
Keywords: polycarboxylate; high solid-content; workability.
INTRODUCTION
Polycarboxylates are commonly used as superplasticizers to disperse cement particles
in concrete and mortar. They are usually stored, packed and transported as 20%~40%
aqueous solutions, which increase the costs of transportation.1,2 It is necessary to research
the synthesis and application of high solid-content polycarboxylates. If the concentration of
monomer in traditional synthesis is simply changed, the content of free water is less simple;
because the viscosity in the system becomes higher. Then the self-acceleration in polymerization might be occurred and part of polymer chains increases abnormally, resulting in the
decreasing of water-reducing rate of polycarboxylates.2,3 There are two common methods
to control the self-acceleration in polymerization: one is increasing the temperature of
synthesis system, the other is adding a good solvent.4 Both methods contribute to rearrange
the chain segment of polycarboxylates. Increasing the temperature of synthesis system can
accelerate the molecular thermodynamic movement, and adding a good solvent can reduce
the viscosity of synthesis system. Because the synthesis of polycarboxylates takes place in
water solution, the temperature of synthesis system is no more than 373 K.5 Increasing the
temperature needs changing the medium of the synthesis, so the costs could increase. Most
of good solvents are more or less toxic. They are not only harmful to the builders, but also
125
126SP-302-09
Table 1 The chemical composition and mineral component of China ISO

standard cement(%)
SiO2
21.8
Al2O3
4.5
Fe2O3
2.6
CaO
62.8
MgO
11.0
SO3
2.6
Na2Oeq
0.6
f-CaO
1.5
C3S
56
C3A
7.6
Table 2 The physical and mechanical properties of China ISO standard

cement
specific surface, m2/kg normal consistency,
(ft2/lb)
%
325 (7714)
26.5
rupture strength,
MPa (psi)
3d
28d
5.6 (812)
9.5 (1377)
compressing strength,
MPa (psi)
3d
28d
28.9 (4190)
59.5 (8627)
pollute the environment.6 In this paper, we focused on the simple synthesis route and got a
high solid-content polycarboxylates with high water-reducing rate .
Molecular weight and its distribution are the most basic structural parameters of polymer
materials, which influence the performance of polycarboxylates. They can be measured by
Gel Permeation Chromatography (GPC).7,8 The poly dispersity index (PDI) is calculated
by dividing the weight average molecular weight (Mw) by the number average molecular
weight (Mn).5,7 PDI describes the uniformity of a polymer with respect to molecular mass
distribution. The minimum value of the PDI is 1, which would correspond to only exactly
one molecular length being present in the polymer. In GPC, the unreacted monomers
and polycarboxylates polymers are revealed in the same diagram, and the peak areas are
proportional to their concentrations. So, the yield of reaction mixtures can also be determined by GPC method and calculated by the proportion of peak area.
The factors affecting the synthetic process of high solid-content polycarboxylate superplasticizer were investigated and optimum conditions were obtained. The GPC method was
used to determine the conversion data of reaction mixtures.
Materials
Isopentenyl polyether (TPEG, Zhejiang Huangma Chemical Industry Group Co., LTD),
Acrylic acid (AA, Zibo Xinglu Chemical Industry CO., LTD), Chain transfer agent (TGA,
Sinopharm Chemical Reagent Co., LTD), Caustic soda liquid, (SINOPEC Qilu Company).
Sodium nitrate (chromatographically pure, Sinopharm Chemical Reagent Co., LTD) were
used for the synthesis of the new superplasticizer
We selected a traditional polycarboxylate superplasticizer (LS) as comparison. China
ISO standard cement were used in this experiment, which mineral component, physical
and mechanical properties were shown in Table 1 and 2.
Synthesis
The synthesis process of polycarboxylate superplasticizer is as follows: TPEG and
water(5% by total weight) were mixed in a flask with thermometer and mechanical stirrer,
Synthesis and Properties of High Solid-Content Polycarboxylate

Superplasticizer127
under the combination of reheating temperature 333~338 K and holding time 40~60 min
till all the TPEG dissolved. The initiator was added in the flask and then the catalyst was
added. AA and TGA were mixed and were added in the flask by constant flow pump, the
addition time was 3 hours. After addition, the temperature was kept at 343~353 K for 1
h. after that the mixture was cooled down to room temperature, adjusting pH to 6~7 with
caustic soda liquid. The solid content is 80%.
Gel permeation chromatography (GPC)
Molecular weight (Mw, Mn) and polydispersity index (PDI) of superplasticizer samples
were measured by gel permeation chromatograph (GPC, also known as size exclusion
chromatography) Waters 1525/2414 instrument (Waters, USA) with differential refraction
detector for measurement of molecular weight of the polymer. Sodium nitrate solution
(0.10 mol/L) was utilized as the carrying phase with a elution rate of 1 mL/min (0.26 gal/
min), where large molecules pass the column quicker than small molecules. The measurement was performed at 308 K, using polyethylene glycol as the calibration standards.
Fourier Transformed Infrared Spectroscopy (FTIR)
The IR absorption spectra of the samples were recorded with the use of an FTIR spectrometer (model VERTEX-70, Bruker). The samples for tests were prepared as potassium
bromide (KBr) disks. In order to determine the contents of hydrophobic aliphatic groups
(CH, CH2) and hydrophilic polyoxyethylene groups (CH2-O-CH2), FTIR spectra for superplasticizers were taken over the whole spectral range (4000~400 cm-1[10160~1016 in-1]).
Performances of the Cement Paste and Concrete.
Cement pastes were mixed at 295 K at a water to cement ratio (w/c) of 0.29, using
a Hobart mixer. First, polycarboxylate superplasticizer and 87 g (3.07 qz) water were
weighed into a beaker and mixed well. Then 300 g (10.58 qz) cement was mixed for 1 min
at a low speed and another 2 min at high speed. The amount of polycarboxylate superplasticizer was calculated as a percentage of dry solid with respect to the cement mass. The
fluidity of the cement paste was measured using a mini-slump cone (~60 mm [2.36 in]
high with a top diameter of 36 mm [1.42 in] and a bottom diameter of 60 mm [2.36 in]),
as described in the Chinese National Standard GB/T 80772012, Methods for Testing the
Uniformity of a Concrete Admixture. Mini-slumps was measured just after mixing (t=0)
and after 30 min. The amount of polycarboxylate superplasticizer used was calculated as a
percentage of the dry mass of the cement.
The slump of the fresh concrete was determined immediately after the mixing according
to Chinese National Standard GB/T 50080-2002, and was controlled in the range of 80
mm(3.15 in) to 90 mm (3.54 in), by adjusting water. The air content test was done according
to the Chinese Construction Industry Standard JG/T 246-2009. The compressive strength
of the concrete was determined after 3, 7, 28 days of standard curing.
128SP-302-09
Fig. 1 The GPC of polycarboxylate superplasticizer with

different initiators.
Table 3 the result of GPC and the fluidity of cement paste of PCs at 333 K.
initiator
Mn
Mw
Mw/
Mn
A
B
C
17900
18800
18600
40700
40000
37600
2.27
2.13
2.02
acreage of peak /%
I
II
III
fluidity of cement
paste(0 h), mm(in)
52
76
78
7
6
6
31
18
16
190 (7.48)
255 (10.04)
fluidity of cement
paste(0.5h), mm
(in)
170 (6.69)

The influence of initiator
The suitable initiator of the polycarboxylate superplasticizer is important for the copolymerization. We chose three different kinds of initiators to synthesze polycarboxylate superplasticizer, and then we analyzed them by GPC, shown in Fig. 1.
The curves of GPC were divided into three different areas by the GPC peaks. As shown
in Fig. 1, the area of peak, at 15~23 min of elution time, marked as I, is polycarboxylate
superplasticizer. The peak at 23~25 min was marked as II. The last peak at 25~28 min was
the peak of unreacted isoamyl alcohol polyxyethylene ether, marked as III. The polymerization conversion rate could be shown by the relative of area III. The fluidity of cement
paste with different polycarboxylate superplasticizer is listed in Table 3. As shown in Table
3, the polymerization conversion rate is strongly influenced by different initiators. For
the polycarboxylate superplasticizer synthesized with initiator C, the relative area III is
smallest. This is means that the highest of polymerization conversion rate was obtained
with initiator C. So the initiator C suits the polymerization in this paper.

Superplasticizer129
Fig. 2 The GPC of polycarboxylate superplasticizers

synthetised under different temperature.
The influence of temperature
The temperature of copolymerization influences the free radical generation, so we
synthesized three polycarboxylate superplasticizers at different temperature, other reacted
conditions being the same, as shown in Fig. 2. and Table 4. To get better information on
the influence of the temperature, the curve of GPCs were segmented in four areas. Area
I, the elution time at 15~18 min, was the polycarboxylate superplasticizers with higher
molecular weight. The smaller ones, elution time at 18~23 min, were marked as area II.
Area III, elution time at 23~25 min, and area IV, elution time at 25~28 min, were characteristic of unreacted monomer. Area IV was the peak of isoamyl alcohol polyxyethylene ether.
By increasing the temperature from 323 K to 333 K, the percentage of area II became
larger, while the percentage of area III and IV decreased, indicating that the polymerization conversion rate was higher. By further increasing the temperature up to 343 K, the
percentage of area I increased, but the percentage of area III and IV also raised. The free
radical generated faster when temperature rise to 343 K, so the percentage of polycarboxylate superplasticizers with higher molecular weight raised. The test of cement paste
showed the higher molecular weight of polycarboxylate superplasticizers could decrease
the fluidity. So, the best temperature of copolymerization was 333 K.
The influence of initiator dosage
We synthesized three polycarboxylate superplasticizers with dosage of different initiator,
the other conditions being the same, as shown in Fig. 3 and Table 5. As shown in Fig. 3,
when the initiator dosage raised from 1.0% to 1.4%, the percentage of area I increased, and
the percentage of area II raised. The test of cement paste showed that the initiator dosage
influenced the polymerization conversion rate, and the best initiator dosage was 1.2%.
130SP-302-09
Table 4 the result of GPC and the fluidity of cement paste of PCs with
different temperature (initiator C)
Temperature, K
Mn
Mw
Mw/Mn
323
333
343
18100
18900
25600
41600
41300
61600
2.29
2.18
2.41
acreage of peak /%
I
II
III
IV
25
21
38
55
64
43
6
5
7
14
10
11
fluidity of
fluidity of cement
cement paste(0 paste(0.5 h), mm
h), mm (in)
(in)
230 (9.06)
150 (5.91)
275(10.83)
210 (8.27)
160 (6.30)
--
Fig. 3 The GPC of polycarboxylate superplasticizers

synthetised under different initiator dosage.
The fourier transform infrared transmission spectrum of polycarboxylate
superplasticizers
The FTIR spectrum of polycarboxylate superplasticizers is shown in Fig. 4. It was illustrated that polycarboxylate superplasticizers had stretching vibration band of C-H in 2885
cm-1 (7328 in-1). The peak near 3441 cm-1 (8740 in-1)is assigned to the stretching vibration
band of O-H, and the peak at 1100cm-1 (2794 in-1)is from C-O-C stretching in polyxyethylene ether. The absorption bands at 1363cm-1 (3462 in-1)is the characteristic absorption
peak of carboxyl group of COONa.
Properties of Concrete with PCs
The slump of the concrete was determined according to the Chinese National Standard
GB/T 50080-2002. The water consumption was measured and the water-reducing rate of
PCs was calculated, as shown in Table 6.
The water-reducing rate of PC was 32.5%, little higher than LS, which was 31.5%. The
slump of concrete test showed almost the same properties of concrete with LS and PC.

Superplasticizer131
Table 5 the result of GPC and the fluidity of cement paste of PCs with
different initiator dosage (initiator C, 333 K)
initiator
dosage /%
Mn
Mw
Mw/Mn
1.0
1.2
1.4
20000
18900
18600
44800
41300
38900
2.24
2.18
2.09
acreage of peak /%
I
II
III
IV
26
21
19
58
64
63
5
5
6
11
10
12
fluidity of
cement paste(0
h), mm (in)
245 (9.65)
275 (10.83)
280 (11.02)
fluidity of cement
paste(0.5 h), mm
(in)
140 (5.51)
210 (8.27)
160 (6.30)
Fig. 4 The FTIR spectrum of polycarboxylate

superplasticizers.
CONCLUSIONS
Based on the results of this experimental work, the following conclusions can be drawn:
1. The factors influencing on the synthetic process were investigated and the optimum
conditions were obtained, i.e. the initiator was C, reaction temperature of polymerization
was 333 K, the initiator dosage was 1.2%.
2. In GPC the peak areas are proportional to concentration. So GPC method was used to
determine the conversion data of reaction mixtures.
3. Compared with traditional polycarboxylate superplasticizer LS, the high solid-content
polycarboxylate superplasticizer showed the same properties of concrete.
AUTHOR BIOS
Yongwei Wang is a researcher at the Shandong Provincial Academy of Building
Research, Jinan, China. He received his BS from Chongqing Jianzhu University; MS
from Chongqing University; and PhD from Chongqing University. His research interests
include civil engineering materials and hydraulic structures.
132SP-302-09
Table 6 Water-reducing rate and slump of the concrete

Polymer
LS
PC
water-reducing rate
[%]
31.5
32.5
slump of fresh concrete, mm (in)
slump after1h, mm (in)
80 (3.15)
90 (3.54)
60 (2.36)
60 (2.36)
Liya Wang is a Research Engineer at the Shandong Provincial Academy of Building

Research, Jinan, China. He received his BS from Shandong University; MS from Shandong University. His research interests include surface and interface activity of cementwater dispersion with superplasticizers.
Yongsheng Liu is a researcher at the Shandong Provincial Academy of Building
Research, Jinan, China. He received his BS from Shandong University; MS from Shandong University. His research interests include the synthesis of superplasticizers.
Zepeng Chu is a researcher at the Shandong Provincial Academy of Building Research,
Jinan, China. His research interests include the synthesis of superplasticizers.
REFERENCES
1. Peiwei, G.; Min, D.; and Naiqian, F., The influence of superplasticizer and superfine
mineral powder on the flexibility, strength and durability of HPC, Cement and Concrete
Research, V. 31, No. 5, 2001, pp. 703-706.
2. Yamada, K. etal., Effects of the chemical structure on the properties of polycarboxylate-type superplasticizer, Cement and Concrete Research, V. 30, No. 2, 2000, pp.
197-207.
3. Baskoca, A.; Ozkul, M. H.; and Artirma, S., Effect of Chemical Admixtures on Workability and Strength Properties of Prolonged Agitated Concrete, Cement and Concrete
Research, V. 28, No. 5, 1998, pp. 737-747.
4. Xiuxing, M., Study on Synthesis and Properties of High Solid-content Polycarboxylate Superplasticizer, V.No.2010, pp.
5. Li, C. etal., Effects of polyethlene oxide chains on the performance of polycarboxylate-type water-reducers, Cement and Concrete Research, V. 35, No. 5, 2005, pp.
867-873.
6. Kreppelt, F. etal., Influence of solution chemistry on the hydration of polished clinker
surfacesa study of different types of polycarboxylic acid-based admixtures, Cement
and Concrete Research, V. 32, No. 2, 2002, pp. 187-198.
7. Plank, J. etal., Synthesis and performance of methacrylic ester based polycarboxylate superplasticizers possessing hydroxy terminated poly(ethylene glycol) side chains,
Cement and Concrete Research, V. 38, No. 10, 2008, pp. 1210-1216.
8. Roy, D. M. etal., Application of GPC for the analysis of the oligomer distribution of
naphthalene-based superplasticizers, Cement and Concrete Research, V. 14, No. 3, 1984,
pp. 439-442.
SP-302-10
Mastering Flow Loss in Superplasticized

Cementitious Systems
by S. Mantellato, Q. Mehmeti, L. Ceni, M. Palacios,
and R.J. Flatt
One of the essential problems of superplasticized concrete is the loss of fluidity over time.
To limit this problem one must improve the compatibility of superplasticizers and cement.
This is not a trivial task as cement contains phases with different responses to superplasticizers in the first hours of hydration.
In the present work, the role of the polymer structure on the flow loss over time on
superplasticized cement pastes has been studied. For this, we have correlated the impact
of different molecular structures on the adsorption degree and ionic solution composition
with the rheological properties of fresh cement pastes. The results revealed a high excess
of aluminium in the aqueous solution. This could be due to aluminum complexation by the
polymer or a poisoning of ettringite growth complemented by a stabilization of nano-sized
ettringite particles. In addition, except for one of the studied polymers, the flow loss seems
to decrease abruptly when the concentration of carboxylate ions in solution drops below a
critical value (0.7-1.2 eq/g).
INTRODUCTION
One of the essential problems of superplasticized concrete is the loss of fluidity over
time.1 Consequently, flow loss can lead to water being added on job sites in order to recover
enough workability to place concrete. This compromises both strength and durability. To
avoid such problems one must improve the compatibility of superplasticizers and cement.
This is not a trivial task as cement contains phases with different responses to superplasticizers in the first hours of hydration.
The aluminates (C3A, 5-10% of the clinker mass) are the most reactive part of the clinker.
Both the adsorption of the PCEs on the surface of the hydrates (monoaluminatesulfate,
AFm, and ettringite, AFt) and a complex intercalation of the polymer into the layers of
the hydrates, forming new organo-aluminate composites, reduce the dispersing capability
of the polymer.2-5 It is expected that this can be translated in a change of the specific
surface that probably affects the evolution of the flow. These perturbing phenomena can be
handled to some extent by varying the admixture dosage to reach the initially desired flow.
However, the evolution after that is much more problematic to manage.
133
134SP-302-10
Table 1. Mineralogical composition (%w/w) of Portland cement determined

by Rietveld analysis of the XRD patterns.
C3S
66.3
C2S
6.7
C3A
5.9
C4AF
10.8
Quartz
0.4
Calcite
3.9
Gypsum
4.3
Hemihydrate
1.7
This probably explains why only relatively few people deal with the structure-function
performance relation of superplasticizers with respect to flow loss. To us, this appears
as one of the big open challenges in admixtures science. Moreover, recent progress in
understanding both the dispersion ability of superplasticizer6-9 and early hydration10 represent important motivations to take on this challenging problem. The present study represents a first step in this quest. It deals with a study of the correlation between the amount
of adsorbed polymer on the cement surface and the rheological properties of the cement
pastes over time. To gain insight into these evolutions, we quantify the impact of superplasticizers of varying chemical structures, on the solution composition and adsorption
degree. Results are discussed in the light of how superplasticizers may affect in particular
the nucleation and growth of hydrated aluminate phases, mainly ettringite.
RESEARCH SIGNIFICANCE (8 LINES)
In the next years, complexity of concrete will grow because of the increasing amounts of
supplementary cementitious materials and chemical admixtures used. Consequently, issues
of robustness will also increase. Specifically, it will become a challenge to guarantee a
given maintenance of flow and more robust admixtures combinations need to be developed. The present study has allowed us to gain knowledge concerning the relation between
the molecular structure of PCEs and the flow loss of cement pastes over time.
EXPERIMENTAL PROCEDURE (3 PAGES)
Materials
A commercial Portland cement CEM I 52.5N according to the European standard EN
197-1:2000 was used. The specific surface area measured by nitrogen adsorption (BET
model) was 1.17 m2/g following the procedure described elsewhere.11 Its Blained specific
surface was 4200 cm2/g. Its mineralogical composition was determined by Rietveld analysis of the X-ray diffraction (XRD) patterns and expressed in values normalized to 100%
of crystalline phases (see Table 1).
Three pure non-commercial PCE polymers, provided by SIKA Technology AG (Zrich,
Switzeraland), were used in the present study. Before use, they were ultrafiltrated to
remove the copolymerization residues. Their backbone is based on polymethacrylic acid
(PMA) and their side chains on polyethylene oxide (PEO) with different lengths, 1000 and
5000 Da. The real carboxylate to ester ratio (C/E) was calculated by UPLC (Waters). The
molecular weight of comb polymers was determined by Gel Permeation Chromatography
(GPC) using a Agilent 1260 Infinity equipment with PSS Suprema columns (0.8 30 cm,
particle size 10 m). Na2HPO4 0.067 M was used as eluent and PEO/PEG were used as
calibration standards.
Mastering Flow Loss in Superplasticized Cementitious Systems 135
Table 2: Molecular characteristics of the superplasticizers used.

Name
C/E ratio
Mn (g/mol)
Mw (g/mol)
2.5PMA1000
4.0PMA1000
2.5PMA5000
2.7
4.5
2.3
22600
16400
70700
39500
25800
121800
Polydispersity index (PDI =

Mw/ Mn)
1.75
1.58
1.72
Table 2. Selected dosages for each superplasticizer to reach the selected

common low and high surface coverage
Investigated dosages
Low surface coverage
High surface coverage
2.5 PMA 1000

[mg polymer/g cement]
1.5
4.0
4.0 PMA 1000

1.5
2.0
2.5 PMA 5000

2.5
4.0
Methods
Cement paste preparation Two mixing procedures were used for mixing the cement
paste keeping the water-to-cement ratio of 0.30 constant. Small batches of cement pastes
were prepared mixing 200 grams of cement and 60 g of ultrapure water containing PCE
admixture with an IKA stirrer at 800 rpm for 5 minutes. After mixing, both spread flow (see
below) and filtration of pore solutions were conducted.
A big batch of about 2.5 Kg of cement paste was prepared in a Hobart N50 mixer to
study the flow loss using the same original paste. Right after mixing, the cement paste was
poured in small plastic beakers and covered with parafilm. Then, these small batches were
remixed for 30 seconds at a speed of 200 rpm with the IKA stirrer and left to rest for one
minute before carrying out spread flow measurements (see below). The first flow test was
carried out after 10 minutes, considered t0.
In a first series of experiments, the superplasticizer dosages were tested between 0 and
8 mg superplasticizer (SP) dry per g cement. With this initial set of data, two dosages for
each polymer were selected for further tests (to limit the number of samples of which the
evolution in time is measured). These were selected from spread flow tests giving diameters of 15 cm and 20 cm. As these can be considered to be respectively associated to low
and high surface coverage of the particles by the polymers, we will refer to them as such.
These were chosen so as to provide a low and high surface coverage of identical magnitude and are given in Table 2.
Cement paste spread flow tests After the mixing described previously, the paste was
filled in a cylinder of 50 mm of diameter and 50 mm height. The cylinder was lifted and
the diameter of the resulting cake was measured as a flow value. The yield stress was
calculated from the following equation that interpolates between the analytical asymptotes
determined by Roussel and Coussot12,13:
0 =
225gV 2
225
(128 2 R 5 (1 +
* 3VR 3 )
128
Equation (1)
136SP-302-10
Figure 1. Adsorption isotherms of PCEs on cement pastes.

Diamonds, squares and circles are for 2.5PMA1000,
2.5PMA5000 and 4PMA1000.
where and V are the density and volume of the paste, respectively, g the gravitational
acceleration and R the radius of the spread. The data are reported as normalized yield stress
with respect to the initial point.
Characterization of cement pore solutionsCement pore solutions were obtained by
filtration of cement pastes through a membrane filter Sartorius 0.45 m and immediately
acidified to prevent the precipitation of hydrates with HCl 0.1 M and 2% (w/w) HNO3 solutions for TOC and ICP measurements, respectively. In a further series of samples, filters of
0.2 and 0.1 m were used.
The total organic carbon content in pore solutions was determined by SHIMADZU
TOC-V CSH total organic carbon (TOC) analyser. The amount of adsorbed polymer on
cement was measured by depletion method, considering the difference between the amount
of initially added and the amount in the liquid phase.
Inductively Coupled Plasma Optical Emission Spectroscopy (ICP-OES) Thermo iCAP
6000 was used for quantifying the ionic composition of cement pore solutions. The
elements of interest were Na, Ca, K, S, Fe, Mg, Si and Al. Pore solutions were diluted 200
times to quantify Na (589.5), Ca (315.8 and 318.1), K (766.4 and 769.8), and S (182.6).
The same cement pore solutions were also diluted 10 times to improve the accuracy in the
quantification of the elements whose concentrations lie in the range of ppb (g/L), such as
Fe (259.9), Mg (279.5), Si (251.6), and Al (396.1). The numbers in brackets represent the
spectral lines selected for each element expressed in nm.
RESULTS
Initial impact of superplasticizer (for a several dosages)
Adsorption isothermsThe adsorption isotherms for the three PCE polymers are shown
in Figure 1, where the content of polymer adsorbed is plotted versus the initial PCE dosage,
Figure 2. Concentration of Al (diamond), Si (square), Fe (circle) and Mg (triangle) as a

function of dosage for the three different superplasticizers used. a) 2.5 PMA 1000, b) 2.5
PMA 5000, c) 4.0 PMA 1000.
Table 3. Average concentration of elements for the different superplasticizer
dosages used
2.5PMA1000
4.0PMA1000
2.5PMA5000
Na [mM]
55 2
70 4
69 2
k [mM]
490 17
481 24
464 16
Ca [mM]
19 3
22 5
19 1
S [mM]
202 9
209 11
198 11
expressed as mg of active polymer per gram of cement. At low dosages, the added polymers are almost totally adsorbed on the cement, which is evidenced by the data lining up
on the 1:1 continuous line in Figure 1. The departure from this line has the following order
4.0 PMA 1000 < 2.5 PMA 1000 < 2.5 PMA 5000, which follows the order of decreasing
charge density of these polymers. The adsorption plateau are not clearly determined in
these experiments, but appear to tend all towards similar values, which is consistent with
side chains maintaining a coiled conformation when polymers are adsorbed.9
Aqueous phase compositionThe elemental analysis of the aqueous phases of these
pastes shows constant Na, K, Ca, and S concentrations, regardless the structure and the
dosage of the polymer (Table 3). The amounts of Al, Si, Fe and Mg increase by increasing
138SP-302-10
Figure 3: Normalized yield stress versus time. Partial (top)

and full (bottom) surface coverage. Diamonds, squares and
circles are for 2.5PMA1000, 2.5PMA5000 and 4PMA1000.
the polymer dosage in the cement paste. The polymer addition causes the largest changes
for aluminium and magnesium (Figure 2). Aluminum concentrations reached in particular
very high values, which is explicitly addressed in the discussion section.
Impact of superplasticizer over time (for two selected dosages)
We restricted the study of flow loss to two dosages per superplasticizer. As explained in
the methods section, these were chosen such that they achieved identical common values
in the spread test. Values indicated are given in Table 2 and are referred to as high and low
surface coverages.
Yield stress Spread flow data were converted to yield stress, using equation (1). Each
data series was further normalized by its initial yield stress. This facilitates the comparison
relative changes in yield stress between the data sets at low and high surface coverage.
Results in Figure 3, show that as expected that for each polymer the highest dosage leads
to the highest flow retention. However, the sequence of flow retention among polymers
Figure 4. Concentrations of Al in the aqueous phases over

time. Partial (top) and full (bottom) surface coverage.
Diamonds, squares and circles are for 2.5PMA1000,
2.5PMA5000 and 4.0PMA1000.
is inverted between low and high surface coverage. It can also be observed that, at high
surface coverage, all polymers show a delayed fluidification, the timing and magnitude of
which is strongly polymer dependent.
Aqueous phase evolutionThe concentrations of Na, K, Ca, and S remain constant over
time, regardless the structure and dosage of the polymer and are therefore not reported in
the present paper. The most useful information from the solution analysis in the context of
this paper concerns the aluminium, which shows very significant variations depending on
the type and dosage of the polymer (Figure 5). A particularly striking observation is that,
at low surface coverage, the polymer 2.5PMA5000 is associated with concentrations of
aluminium and magnesium that are one order of magnitude higher than with either of the
other polymers having short side chains.
140SP-302-10
Figure 5. Saturation index of the aqueous phase in suspensions prepared with superplasticizer 2.5 PMA1000 at
different dosages after mixing.
At high surface coverage, the amount of elements in the aqueous phases shows quite
different trends over the time depending on the polymer and its dosage. The most remarkable results concern the concentrations of aluminium, shown in Figure 5 (Concentrations
of other elements will be reported upon separately). Here again it can be observed that the
content of the elements in cement suspension with the polymer 2.5 PMA 1000 and 5000
is higher than with 4.0 PMA 1000. In the pore solution with the polymer 2.5 PMA 1000
the concentration of elements was constant until 200 minutes. At this point the amount of
elements starts to decrease. The high aluminium concentrations reached are discussed in
the next section.
DISCUSSION
Impact of the PCE on aqueous phase composition
The high amount of aluminium that are reached in Figure 5 are unusual and suggest that
these aqueous phases may be supersaturated with respect to ettringite. To gain quantitative
insight into this question we have computed the activity coefficients of all species in these
solutions using GEMS14,15 with the CemData07 database.16 Results in Figure 6 reveal that
solutions are supersaturated not only with respect to ettringite, but also with respect to
other aluminate phases such as monocarboaluminate and monosulfoaluminate.
This supersaturation may be due either to complexation of Ca and/or Al ions by the polymers or to a stabilization of small ettringite particles. In the later case, the polymer would
not only have to poisen their growth but also hinder their agglomeration. As a consequence
nano-particles of ettringite might be passing through the filter, something already reported
for PCEs by Comparet.17 To test this hypothesis, we filtered some solutions through filters
of 0.2 and 0.1 microns. In most cases this did not decrease the aluminium concentration in
Figure 6. Normalized yield stress from Figure 3 replotted

versus polymer remaining in solution (top) and carboxylate groups in solution (bottom). Empty and filled symbols
respectively represent low and high dosages. Diamonds,
squares and circles are for 2.5PMA1000, 2.5PMA5000 and
4.0PMA1000.
solution, contrary to what Comparet observed. We are in the process of using nano-particle
analysis on these solutions to determine whether particles smaller than 100 nm can be
identified in these solutions.
An alternative explanation to the high Al concentration could be a strong complexing
capacity of the SPs for Al. At this stage we have only conducted a qualitative examination of this question. For this we first determined the concentration of carboxylate groups
from the polymer in solution (not adsorbed). Then in absence of detailed knowledge on
the stoichiometry of Al-polymers complexes, we examine the relative number of charges
coming from the polymer in solution versus those from the aluminium. It turns out that
142SP-302-10
the number carboxylates in solutions is much larger than the number of charges from the
aluminium. This suggests that the supersaturation with respect to ettringite may be due to
the presence of complexes between aluminium and the polymers. This is further supported
by the fact that all aluminium containing phases are supersaturated. In contrast, gypsum
and portlandite are not supersaturated, which suggests that the calcium complexation is not
responsible for the calculated supersaturations.
Flow loss
Figure 3 shows that the normalized yield stress is relatively steady over a period of time
and then increases rapidly. The first order explanation for this is that as hydrates are formed
the total system surface increases and excess polymers in solutions adsorb. Once no more
enough polymers are available to maintain a high specific surface, the fluidity would begin
to be lost.18 There are two ways to examine the relevance of this view. The first and most
pragmatic approach consists in re-plotting the data of Figure 3 not versus time, but versus
the amount of polymer remaining in solution. As shown in Figure 7 (top) this does not link
to lead to a collapse of data on a master curve. However, when data are plotted versus the
amount of carboxylates remaining in solution (Figure 7, bottom) most data series show a
similar concentration range in which the normalized yield stress abruptly increases (0.7-1.2
eq/g). Only the high dosage of 2.5PMA1000 leads to a substantially different behaviour.
Otherwise the rather similar common behaviour of the other polymers suggests that its
complexing ability in solution slows down the growth of hydrates and thereby delays the
flow loss.
Towards additional insight on flow loss
The results presented in this paper concerning flow loss can be summarized by the these
two main observations:
1. A high excess of ettringite is measured in the aqueous phase. This may be due to
aluminum complexation by the polymer or a poisoning of ettringite growth complemented
by a stabilization of nano-sized ettringite particles.
2. Apart for one case, the flow loss appears to take off when the concentration of carboxylate ions in solution drops below a critical value.
These observations suggest that flow loss is controlled by the development of hydrate
surfaces and that polymers delay this by hindering the growth of ettringite and/or by
poisoning its nucleation (because of aluminum complexation). Determining which of these
mechanisms is at stake require more extensive characterization, which we are undertaking
but are beyond the scope of this paper.
CONCLUSIONS
Flow loss is a complex phenomena that has only received limited attention in the literature, particularly with regard to the role of the molecular structure of the superplasticizer.
The data presented in this paper suggest a critical role of the polymer in either complexing
aluminum or preventing ettringite growth. This defines specific characterization objectives
to start shedding light on how to relate molecular structure of PCEs to flow loss.
AUTHOR BIOS
Sara Mantellato is Chemist and has great experience on chemical admixtures. She is
currently doing her PhD Thesis on the effect of comb compolymer superplasticizers on the
specific surface are evolution in order to understand the flow loss of cementitious systems.
Qendrim Mehmeti and Lindrit Ceni are civil engineering students from ETH Zrich,
who carried out their Bachelor Thesis on the topic of Flow loss in superplasticized
cementitious binders under the supervision of Sara Mantellato.
Dr. Marta Palacios, Ph.D. in Chemistry, has been working in the area of building materials for more than 10 years. Her research interests include the interaction of chemical
admixtures with Portland and alkaline cements and their effect on the rheological properties and hydration process.
Prof. Dr. Robert Flatt is Professor of Building Materials at ETHZ. Before that he was
Principal Scientist at Sika Technology AG and postdoctoral researcher at the Princeton University. He owns a master in Chemical Engineering and a PhD from EPFL.
He has received various awards among which the RILEM Robert LHermite Medal, the
Ross C. Purdy and the Brunauer awards from the American Ceramic Society, as well an
Outstanding Research Contribution in the Broad Area of Chemical Admixtures presented at
the 10th International Conference on Superplasticizers and Other Chemical Admixtures.
ACKNOWLEDGEMENTS
The authors would like to thank Dr. B. Lothenbach for her advices in the use of GEMS
software. We also thank Dr. F. Caruso for his support during the analysis of the aqueous
solution by ICP and to G. Gelardi, D. Marchon and D. Altermatt for their help in the
characterization of the PCEs. Finally we thank Dr. Lukas Frunz (Sika Technology) for
providing us the polymers.
REFERENCES
1. Flatt, R. J., and Schober, I., Superplasticizers, in Understanding the rheology of
concrete (Ed. Roussel N.), Woodhead publishing (2012) 144-208
2. Flatt, R. J., and Houst, Y., A simplified view on chemical effects perturbing the action
of superplastizers, Cement and Concrete Research, V. 31, No. 8, 2001, pp. 1169-1176.
doi: 10.1016/S0008-8846(01)00534-8
3. Zingg, A.; Winnefeld, F.; Holzer, L.; Pakusch, J.; Becker, S.; Figi, R.; and Gauckler, L.,
Interaction of polycarboxylate-based superplasticizers with cements containing different
C3A amounts, Cement and Concrete Composites, V. 31, No. 3, 2009, pp. 153-162. doi:
10.1016/j.cemconcomp.2009.01.005
4. Giraudeau, C.; DEspinose De Lacaillerie, J.-B.; Souguir, Z.; Nonat, A.; and Flatt, R.
J., Surface and intercalation chemistry of polycarboxylate copolymers in cementitious
systems, Journal of the American Ceramic Society, V. 92, No. 11, 2009, pp. 2471-2488.
doi: 10.1111/j.1551-2916.2009.03413.x
5. Plank, J.; Zhimin, D.; Heller, H.; Hossle, F.; and Seidl, W., Fundamental mechanisms
for polycarboxylate intercalation into C3A hydrate phases and the role of sulfate present
144SP-302-10
in cement, Cement and Concrete Research, V. 40, No. 1, 2010, pp. 45-57. doi: 10.1016/j.
6. Ran, Q.; Somasundaran, P.; Miao, C.; Liu, J.; Wu, S.; and Shen, J., Effect of the
length of the side chains of comb-like copolymer dispersants on dispersion and rheological
properties of concentrated cement suspensions, Journal of Colloid and Interface Science,
V. 336, No. 2, 2009, pp. 624-633. doi: 10.1016/j.jcis.2009.04.057
7. Pourchet, S.; Liautaud, S.; Rinaldi, D.; and Pochard, I., Effect of the repartition of
the PEG side chains on the adsorption and dispersion behaviours of PCP in presence of
sulfate, Cement and Concrete Research, V. 42, No. 2, 2012, pp. 431-439. doi: 10.1016/j.
8. Houst, Y. F.; Bowen, P.; Perche, F.; Kauppi, A.; Borget, P.; Galmiche, L.; Le Meins, J.
F.; Lafuma, F.; Flatt, R.; Schober, I.; Banfill, P. F. G.; Swift, D. S.; Myrvold, B. O.; Petersen,
B. G.; and Reknes, K., Design and Function of Novel Superplasticizers for More Durable
High Performance Concrete (Superplast Project), Cement and Concrete Research, V. 38,
No. 10, 2008, pp. 1197-1209. doi: 10.1016/j.cemconres.2008.04.007
9. Flatt, R. J.; Schober, I.; Raphael, E.; Plassard, C.; and Lesniewska, E., Conformation
of Adsorbed Comb Copolymer Dispersants, Langmuir, V. 25, No. 2, 2009, pp. 845-855.
doi: MPa Materiales de construccion10.1021/la801410e
Cement and Concrete Composites, V. 29, No. 4, 2007, pp. 251-262. doi: 10.1016/j.
cemconcomp.2006.12.006
11. Mantellato, S.; Palacios, M.; and Flatt, R. J., Reliable specific surface area measurements on anhydrous cements, Cement and Concrete Research, (accepted)
12. Roussel, N., and Coussot, P., Fifty-cent rheometer for yield stress measurements:
From slump to spreading flow, Journal of Rheology (New York, N.Y.), V. 49, No. 3, 2005,
pp. 705-718. doi: 10.1122/1.1879041
13. Zimmermann, J.; Hampel, C.; Kurz, C.; Frunz, L.; and Flatt, R. J., Effect of polymer
structure on the sulphate-polycarboxylate copmetition, Proc. 9th ACI Int. Conf. Superplasticizers and Other Chemical Admixtures in Concrete, (editors: Holland, T.C., Gupta,
P.R., Malhotra, V.M.), American Concrete Institute, Detroit, SP-262-12 (2009) pp. 165-176
14. Kulik, D., GEMS-PSI 2.1, PSI-Villigen, Switzerland, 2005. http://les.web.psi.ch/
Software/GEMS-PSI/ available at.
15. Damidot, D.; Lothenbach, B.; Herfort, D.; and Glasser, F. P., Thermodynamics and
cement science, Cement and Concrete Research, V. 41, No. 7, 2011, pp. 679-695. doi:
10.1016/j.cemconres.2011.03.018
16. Lothenbach, B.; Matschei, T.; Mschner, G.; and Glasser, F. P., Thermodynamic modeling of the effect of temperature on the hydration and porosity of Portland
cement, Cement and Concrete Research, V. 38, No. 1, 2008, pp. 1-18. doi: 10.1016/j.
17. Comparet, C., Etude des interactions entre les phases modles reprsentatives dun
ciment Portland et des superplastifiants du bton, PhD thesis, Universit de Bourgogne,
France (2004).
18. Flatt, R. J., Towards a prediction of superplasticized concrete rheology, Materials
and Structures, V. 37, No. 5, 2004, pp. 289-300. doi: 10.1007/BF02481674
SP-302-11
Putting Concrete to Sleep and Waking It

Up with Chemical Admixtures
by L. Reiter, M. Palacios, T. Wangler, and R.J. Flatt
The ability to control the setting of cement can be of use in various applications such as slip
forming, oil well cements, or in normal applications due to variable conditions and time
constraints. Typically cement setting is controlled via set retarders or set accelerators, but
rarely are the two used in combination. The combination of the two, however, can lead to
increased flexibility in construction methods. In this work, we present a system in which the
dormant period of cement is extended with sucrose, and then drastically reduced by adding
calcium hydroxide, a phase that preferentially adsorbs sucrose. We demonstrate that
increasing doses of calcium hydroxide decrease the dormant period of sucrose-retarded
cement, up to a complete cancellation of retardation, which is reached at the plateau of the
sucrose-calcium hydroxide adsorption isotherm.
Keywords: admixtures; adsorption; cement hydration; set accelerators; set retarders;
setting control; setting time.
INTRODUCTION
The ability to control the setting time of cementitious materials is an extremely valuable one to possess, giving much greater flexibility in terms of how, when and under what
conditions concrete can be cast for varied applications. For example, set retarders are used
in hot climates, oil well cementing or when transport requirements demand a longer open
time. Set accelerators are used in cold climates, shotcrete applications, or when operational demands require a faster setting time. A large body of research already exists on the
manipulation of the setting time via the use of set accelerators and set retarders, as well as
their respective mechanisms of action.1-6
While the idea of set-on-demand is not a new one, a new casting application known as
Smart Dynamic Casting has been recently developed that utilizes this concept in a novel
and effective way.7 In this method, a heavily retarded batch of self-compacting mortar
is accelerated bit-by-bit before being placed into a mold. This is similar to slip forming,
with the difference that the mold is slipped at a time when the material is still deformable
and can be shaped. This is done by controlling, via the setting, the shear strength within
a narrow range of acceptable values in a time range of exponential increase of the shear
145
146SP-302-11
Table 1. Mineralogical composition (%w/w) of Portland cement determined

by Rietveld analysis of the XRD patterns.
C3S
64.6
C2S
9.2
C3A
5.2
C4AF
11.6
Gypsum
3.0
Hemihydrate
5.2
Quartz
0.3
Calcite
0.9
strength. The success of this process further depends on preventing the catastrophic flow
out from the mold and also allowing the shaping without creating cracks.7,8
The essence of this developed method is having a heavily retarded batch of sleeping
concrete that can be woken up in a controlled manner, hardening at a specific time and
a specific rate. This can be done by putting the concrete to sleep with a heavy dose of a
retarder, such as sucrose, and having the activation performed by an accelerator of C-S-H
seeds. Alternatively, the action of the retarder can itself be reversed by the preferential
adsorption of the retarder onto another phase, with that phase acting as the activator of the
mix. In this work, we demonstrate the feasibility of such an approach by using sucrose to
retard the hydration of a mortar for several hours and adding a high specific surface area
calcium hydroxide to wake up the mix.
Set-on-demand is an extremely useful ideal to pursue with cementitious materials, that
gives much more freedom in construction with concrete and other cementitious materials.
While the usage of retarder and accelerator admixtures is very common and their individual mechanisms of action an area of very active research, the combination of the two
admixtures has not been systematically studied.
Materials
A commercial portland cement CEM I 52.5R according to the European standard EN
197-1:2000 was used. The specific surface area measured by nitrogen adsorption (BET
model) was 1.17 m2/g following the procedure described elsewhere.9 Its Blaine specific
surface was 4200 cm2/g. Its mineralogical composition was determined by Rietveld
analysis of the X-ray diffraction (XRD) patterns and expressed in values normalized to
100% of crystalline phases (see Table 1). D(+)-Sucrose (99.7% for biochemistry - Acros
Organics) and Ca(OH)2 (purum p.a. Fluka) with a specific surface area of 15 m2/g were
used as admixtures.
In the preparation of mortars, in addition to the cement, sucrose and Ca(OH)2, the
following materials were used: Commercial fly ash according to ASTM C618, undensified
silica fume, commercial polycarboxylate ether superplasticizer used in precast applications, siliceous aggregates of grain size 0-4mm and polyvinyl alcohol (PVA) fibers of 6
and 12mm length.
Preparation of pastes and mortars
PastesThe cement pastes were prepared by mixing for 3 minutes at 500 rpm with
a blade stirrer 100 g of cement with 40 g of water (liquid/solid ratio of 0.4). Dosages of
0, 0.20, 0.41 and 0.82 mg of sucrose/g cement (0, 0.02, 0.41, 0.082% weight sucrose by
cement) were added to the mixing water. For each concentration of sucrose, the effect of
Putting Concrete to Sleep and Waking It Up with Chemical Admixtures 147
Table 2.- Composition for 1L of mortar mix.

Sand
CEM I
Fly ash
Silica fume
637g
837g
141g
79.3g
Water
Super-plasticizer
Sucrose
358g
3.41g
0.69g
PVA 6mm
PVA 12mm
Ca(OH)2
11.2g
11.2g
24.7g
the Ca(OH)2 addition on the hydration of cement was studied (0, 33, 65 and 130 mg of
Ca(OH)2/mg sucrose 0, 33, 65, 130 times weight Ca(OH)2 by sucrose). Ca(OH)2 was
mixed for 1 minute with cement powder at 200rpm with a blade stirrer previous to addition
of the water.
MortarsThe binder consisted of 79.2% portland cement, 13.3% fly ash and 7.5% silica
fume by weight. The mortars were prepared by adding a ternary mixture of water, sucrose
and superplasticizer to a previously dry-mixed mixture consisting of sand, cement, fly ash
and silica fume in a Hobart compulsory mixer at 139 rpm and mixing the ingredients for
4 minutes.
Dosages of 0 and 0.82 mg of sucrose/g cement and 1.22 mg of dry superplasticizer/g
cement were added to the mixing water. PVA fibres and different concentrations of Ca(OH)2
in the range of 0 and 36 mg Ca(OH)2/mg sucrose were added to the mix after 4 minutes of
mixing. The resulting composition was mixed for one additional minute. Table 2 shows the
composition for 1L of a reference mortar containing 0.82 mg sucrose / g cement, 29.5 mg
Ca(OH)2/g cement and 1.22 mg superplasticizer /g cement.
Methods
Adsorption curves of sucrose on Ca(OH)2 suspensionsTo determine the adsorption
isotherms of the sucrose 30 grams of Ca(OH)2 and 60 g of distilled water containing
different amounts of sucrose were mixed and stirred for 3 hours at 25 C (77 F).
The suspensions were subsequently centrifuged at 5000 rpm during 10 minutes using a
Centrifuge Allegra 25R High Speed Benchtop (Beckman Coulter). The aqueous phase was
filtered through a membrane filter Sartorius 0.45 m and the total organic carbon content
was determined on a Shimadzu TOC-VCSH/CSN total organic carbon (TOC) analyser.
The amount of adsorbed sucrose on Ca(OH)2 was measured by depletion method, considering the difference between the initially added amount and the amount in the liquid phase.
Isothermal calorimetryIsothermal calorimetric measurements were carried out in a
I-Cal 8000 calorimetry (Calmetrix) at 23 C (73 F) on cement pastes and mortars. This
technique was used to monitor the kinetics of the hydration reaction of the cementitious
systems under study.
Adsorption isotherm of sucrose on Ca(OH)2Figure 1 shows the adsorption of the
sucrose on Ca(OH)2. This adsorption is linear at concentrations up to 4 mg sucrose/g
Ca(OH)2, reaching a plateau value at 7.4 mg sucrose /g Ca(OH)2.
Effect of the combination of sucrose and Ca(OH)2 on cement hydration in cement
pastesFigure 2 shows an example of the relevant impact of the addition of sucrose and
variable dosages of Ca(OH)2 on the hydration of cement. The addition of the highest dosage
of sucrose (0.82 mg sucrose/g cement) increases the induction period up to 14 hours. This
148SP-302-11
Figure 1. Adsorption of sucrose on Ca(OH)2. a) Adsorbed amounts are reported as a function of the dosage. The continuous line shows the 1:1 behaviour that would be obtained if
all added sucrose was adsorbed. b) Adsorption plotted versus concentration in solution.
retarding effect decreases significantly as the amount of Ca(OH)2 in the mix rises, being
almost cancelled when a dosage of 130 mg Ca(OH)2/mg sucrose is added.
No major changes in the shape of the curves and on the slope of the heat rate in the
acceleration period are observed, suggesting that the presence of both admixtures does
not have a significant impact on the hydration mechanism of cement Although with the
highest dosage of Ca(OH)2, the slope is slightly higher and the main exotherm and the
sulphate depletion peak appear to come together with increasing retardation. Supplementary measurements also showed, that the delayed addition of Ca(OH)2 relative to the mixing
protocol by an offset of 1 hour, did not strongly alter the setting time.
Figure 2.- Rate of heat released by cement pastes containing 0.82mg sucrose/g cement and
different dosages of Ca(OH)2
Figure 2 also shows the effect of Ca(OH)2 on the hydration of cement in the absence of
sucrose. The addition of Ca(OH)2 induces a slight acceleration of the hydration of cement,
probably due to the increase of the content of fine solids.
In order to quantify the effect of the combination of sucrose and Ca(OH)2 on the hydration of cement, the first inflexion point in the acceleration period for all the measured
calorimetric curves was determined. This inflexion point is a reliable way to quantify the
effect of the combination of these admixtures, as the initial and final setting time based on
calorimetry curves is not easy to determine unambiguously.
Figure 3 shows that for the lower dosages of sucrose (0.20 and 0.41 mg sucrose/g
cement), the time of the inflexion point decreases linearly with the increase of Ca(OH)2
content. For the highest dosage of sucrose (0.82 mg sucrose/g cement), a linear relationship
is not observed.
It is worth highlighting that for all cement pastes containing 130 mg Ca(OH)2/ mg
sucrose, the inflexion point appears at the same time for all observed sucrose concentrations and corresponds to the value of the cement pastes without any admixture. A particularly nice aspect of this result is that this dosage of Ca(OH)2 corresponds to the maximum
adsorption of sucrose according to Figure 1 (7.4 mg sucrose/g Ca(OH)2).
Effect of the combination of sucrose and Ca(OH)2 on cement hydration in mortars
Figure 4 shows an example of the impact of the addition of sucrose and various dosages of
Ca(OH)2 on the hydration of mortars. The addition of 0.82 mg sucrose/g cement increases
the induction period and delays the setting according to the inflexion point by 7.5 hours. By
adding 29.5 mg Ca(OH)2/g cement to the mortars the retarding effect of sucrose is reduced
by 2.8 hours. As in the case of the pastes, the addition of Ca(OH)2 does not induce major
changes in the slope of the acceleration period with respect to the reference mortar without
sucrose and Ca(OH)2. Again, it is observed that in absence of sucrose, Ca(OH)2 acceler-
150SP-302-11
Figure 3 Effect of combinations of sucrose and Ca(OH)2 on setting time according to

inflexion point of heat release in the studied cement pastes.
Figure 4 Rate of heat released in mortars containing different dosages of Ca(OH)2 and
sucrose.
ates the hydration of mortar slightly by 0.5 hours. The heat release at the beginning of the
acceleration period is however larger.
Figure 5 shows the relation of setting time according to the inflexion point, for variable amounts of Ca(OH)2 with constant sucrose and superplasticizer contents. The setting
time according to the inflexion point appears to follow a linear decrease with increasing
amounts of Ca(OH)2.
Figure 5 Effect of combination of 0.82 mg sucrose/ g cement and Ca(OH)2 on setting time
according to inflexion point of heat release.
DISCUSSION
It is well known that sucrose is one of the most effective retarders of cement hydration.
Previous studies2,4 have concluded that sucrose increases the dormant period of cement
by adsorbing onto the surface of cement particles and hydration products, preventing the
nucleation and growth of hydrates.
In the present study, we have shown that sucrose adsorbs onto the surface of Ca(OH)2
(see Figure 1), with a maximum adsorption of 7.4 mg sucrose/g Ca(OH)2. By adding
Ca(OH)2 with a higher specific surface area than the anhydrous cement particles, we are
creating new sites in the cement paste for sucrose to be adsorbed, reducing or cancelling its
retarding effect. This is confirmed by the fact that the higher the amount of Ca(OH)2 added
to the mix, the higher the suppression of the retarding effect of sucrose. Figures 2 and 3
confirm additionally that such retarding effects can be partially or totally cancelled by the
addition of different amounts of Ca(OH)2 in cement pastes.
For any of the dosages of sucrose studied, the addition of an amount of Ca(OH)2 equal
to the plateau value of the adsorption isotherm (7.4 mg sucrose/g Ca(OH)2 = 135 mg of
Ca(OH)2/mg sucrose), the retardation of sucrose is almost totally cancelled.
Figure 5 shows that the setting on demand idea proposed in this paper also works in more
complex mixtures such as mortars, which can contain supplementary cementitious materials with large surfaces such as silica fume, or admixtures that can also delay the hydration
of cement such as superplasticizers.5,10 In this study, the hydration of cement pastes is more
strongly delayed than in mortars for the same sucrose content (0.82 mg sucrose/g cement).
The qualitative result of the combination of Ca(OH)2 and sucrose on the setting of cement
pastes and mortars is however similar.
While the use of C-S-H seeds allowed Thomas et al.1 to control the setting of mortars
with a combination of C-S-H seeds and sucrose, the C-S-H seeds increased the maximum
heat rate with increasing C-S-H seed amounts. This effect was explained by the introduc-
152SP-302-11
tion of additional nucleation points in the pore space on which the hydration products can
grow. Figures 2 and 4 show, that the maximum heat release and the slope at the inflexion
point of the heat rate curve is not changing strongly by changing Ca(OH)2 amounts in the
mixture.
This feature of the Ca(OH)2 interaction with sucrose is potentially useful for many applications, such as slip forming, that requires an initially retarded concrete mix, that can be
accelerated on demand without changing the hydration reaction rate and the concomitant
increased rate of the products mechanical bearing capacity. This would allow prediction
of the mechanical strength development of consecutively placed and individually accelerated layers. It would provide an alternative or complement to the addition of C-S-H seeds7
without the inconvenience of including the water and superplasticizers contained in C-S-H
suspensions used for this purpose. Additionally, in contrast to the use of seeds the kinetics
of the acceleration period is not changed, so that the rate of heat release can be better
controlled.
The combined use of sucrose and Ca(OH)2 offers the possibility to define an open time
by choosing the sucrose content and to define a production timeframe given by the choice
of the maximum Ca(OH)2 content.
Due to the possible adsorption of other agents such as superplasticizers and changes
in the solid volume fraction within the composition, the addition of Ca(OH)2 powder is
not exempt from ambiguities, mainly with respect to rheology. Furthermore deeper understanding of the sucrose mechanism is required to use the approach discussed in this paper
reliably, especially in consideration of the variability of sucrose delay reported in literature
especially possible severe retardation depending on the cementitious systems.
CONCLUSIONS
This paper describes the combined use of accelerators and retarders in cementitious
systems, which can be used to control the setting time of a composition, relative to a
composition containing only the retarder, by the choice of the accelerator.
Sucrose is used as retarder to increase the dormant period by adsorption to the cement
surface. It is shown, that high surface Ca(OH)2, used as accelerator in this paper, can adsorb
the sucrose up to 7.4 mg sucrose / g Ca(OH)2 by adsorption isotherm. This approach works
equally well in cement pastes and mortars.
The combined use of sucrose and Ca(OH)2 does not strongly change the shape of the heat
rate curves, as opposed to systems adding C-S-H seeds. This offers a better control of the
evolution of mechanical properties closely for slip forming like applications in the future.
More importantly the dosage of Ca(OH)2 can be directly determined in relation to the
dosage of sucrose in the range of linear hydration delay with variable sucrose content at
low sucrose additions. It follows a very basic consideration about the adsorption plateau
of sucrose on Ca(OH)2. This is to say that the needed dosage can be determined without
the need of extensive trial and error campaigns. Last but not least an excess of Ca(OH)2
does not lead to an additional acceleration of cement hydration. Therefore, there are also
little negative consequences to be expected from situations in which the Ca(OH)2 may be
overdosed.
AUTHOR BIOS
Lex Reiter, Msc. in Civil Engineering. His research interests include setting and hardening of cementitious materials.
Dr. Marta Palacios, Ph.D. in Chemistry, has been working in the area of building materials for more than 10 years. Her research interests include the interaction of chemical
admixtures with Portland and alkaline cements and their effect on the rheological properties and hydration process.
Dr. Timothy Wangler, Ph.D. in Chemical Engineering with research interests in the area
of building materials.
Prof. Dr. Robert Flatt is Professor of Building Materials at ETHZ. Before that he was
Principal Scientist at Sika Technology AG and postdoctoral researcher at the Princeton
University. He owns a master in Chemical Engineering and a PhD from EPFL. In 2003,
he received the RILEM Robert LHermite Medal for his contributions to the understanding of chemical admixtures. In 2007, he received the Ross C. Purdy award from the
basic science division of the American Ceramic Society.
ACKNOWLEDGMENTS
The authors would like to acknowledge Dr. Michael Pltze for his support in the XRD
and Rietveld analysis carried out for this study. This research was supported by the NCCR
Digital Fabrication, funded by the Swiss National Science Foundation. (NCCR Digital
Fabrication, Agreement # 51NF40-141853) and by an ETH Research Grant ETH-13 12-1
under the name Smart Dynamic Casting (SDC)
REFERENCES
1. Thomas, J. J.; Jennings, H. M.; and Chen, J. J., Influence of Nucleation Seeding on
the Hydration Mechanisms of Tricalcium Silicate and Cement, The Journal of Physical
Chemistry C, V. 113, No. 11, 2009, pp. 4327-4334. doi: 10.1021/jp809811w
2. Thomas, N. L., and Birchall, J. D., The retarding action of sugars on cement
hydration, Cement and Concrete Research, V. 13, No. 6, 1983, pp. 830-842. doi:
10.1016/0008-8846(83)90084-4
3. Bruere, G. M., Set-retarding Effects of Sugars in Portland Cement Pastes, Nature,
V. 212, No. 5061, 1966, pp. 502-503. doi: 10.1038/212502a0
4. Bishop, M., and Barron, A. R., Cement Hydration Inhibition with Sucrose, Tartaric
Acid, and Lignosulfonate:Analytical and Spectroscopic Study, Industrial & Engineering
Chemistry Research, V. 45, No. 21, 2006, pp. 7042-7049. doi: 10.1021/ie060806t
5. Cheung, J., Jeknavorian, A., Roberts, L. & Silva, D. Impact of admixtures on the
hydration kinetics of Portland cement. Conf. Spec. Cem. Hydration Kinet. Model. Quebec
City 2009 CONMOD10 Lausanne 2010 41, 12891309 (2011).
6. Smith, B. J.; Rawal, A.; Funkhouser, G. P.; Roberts, L. R.; Gupta, V.; Israelachvili, J. N.;
and Chmelka, B. F., Origins of saccharide-dependent hydration at aluminate, silicate, and
aluminosilicate surfaces, Proceedings of the National Academy of Sciences of the United
States of America, V. 108, No. 22, 2011, pp. 8949-8954. doi: 10.1073/pnas.1104526108
154SP-302-11
7. Lloret, E. etal., Complex concrete structures: Merging existing casting techniques

with digital fabrication, Computer Aided Design, doi: 10.1016/j.cad.2014.02.011
8. Shahab, A. R. et al., Smart dynamic casting or how to exploit the liquid to solid transition in cementitious materials. Proc. CD 1st Int. Conf. Rheol. Process. Constr. Mater. 7th
Int. Conf. Self-Compact. Concr. Paris Fr. (2013).
9. Mantellato, S., Palacios, M. & Flatt, R. J. Reliable specific surface area measurements
on anhydrous cements. Cem. Concr. Res. Accept.
SP-302-12
A Simplified Preparation Method for PCEs

Involving Macroradicals
by Lei Lei and Johann Plank
Polycarboxylate superplasticizers are known to be most powerful admixtures which exhibit
superior dispersing force even at extremely low water-to-cement ratio. In this study, a
simplified one-pot synthesis method for a PCE using only maleic anhydride and methoxy
polyethylene glycol as sole raw materials was developed. Compared to conventional
synthetic routes, the new method constitutes a much simpler process which performs
esterification and grafting in one reactor. Macromonomers are no longer needed for the
synthesis of this PCE. The resulting copolymer was characterized by size exclusion chromatography and anionic charge density measurement. Performance of the polymer in
cement was probed via mini slump test. To detect a potential retarding effect of the copolymer, time-dependent heat evolution was monitored. Finally, a model for the formation of
this PCE is proposed. According to this, maleic anhydride and MPEG maleate monoester
are grafted onto MPEG macroradicals which present the backbone of this PCE.
Keywords: Polycarboxylate; Admixture; Synthesis; Graft copolymer; Polymerization
mechanism; Dispersion.
INTRODUCTION
Polycarboxylate-based superplasticizers (PCEs) have become indispensable admixtures,
especially for the production of high performance concretes such as self-compacting (SCC)
or ultra-high performance concrete (UHPC).1,2 Nowadays, a broad variety of different PCE
products exhibiting diverse chemical compositions are on the market.3 In literature, they
are referred to as -methoxy poly(ethylene glycol) methacrylate ester (MPEG), -allyl-methoxy or -hydroxy poly(ethylene glycol) ether (APEG), poly(ethylene glycol) vinyl
ether (VPEG), -methallyl--methoxy or -hydroxy poly(ethylene glycol) ether (HPEG)
and isoprenyl oxy poly(ethylene glycol) ether (IPEG or TPEG) type PCEs whereby the
acronym indicates the chemical nature of the side chain-bearing unit (for example, MPEG
indicates that this PCE contains a poly(ethylene glycol) methacrylate ester as major
building block). The general composition of these kinds of PCEs is displayed in Fig. 1.
The industrial synthesis of PCEs occurs in different ways. The most common process
includes aqueous free radical copolymerization whereby a macromonomer holding the side
chain is reacted with an acid-bearing monomer (e.g. acrylic, methacrylic, maleic, itaconic
acid etc.) to yield the typical comb structure of PCE.4 Different initiator systems including
peroxodisulfate, redox systems such as Fe2+/H2O2, rongalite/Fe2+/H2O2 or ascorbic acid/
1 sulphonate or mercapto propionic acid) are
H2O2 and chain transfer agents (e.g. methallyl
2SP-302-12
Fig. 1 General composition of major kinds of PCEs used by the industry

utilized to tailor the molecular properties such as molar masses (Mw, Mn) and polydispersity
of the PCE to the individual requirements in specific applications. This process of radical
copolymerization is fairly easy to carry out in an industrial plant, and even polymerization
methods which can be performed at room temperature and do not require heating have
been developed.5 However, this preparation method depends on the local availability of
macromonomers which may not always be at hand.
For (meth)acrylate ester-based PCEs, the so-called grafting process presents an alternative synthesis route (Fig. 2). There, commonly available -methoxy poly(ethylene glycol)
is grafted on to a poly(meth)acrylic acid backbone via esterification reaction, resulting in
the desired comb-structure of PCE. Different acid catalysts including p-toluenesulfonic
acid or hypophosphite are employed in this reaction to speed up esterification and increase
A Simplified Preparation Method for PCEs Involving Macroradicals 3
Fig. 2 Preparation of MPEG PCEs via esterification of

poly(methacrylic acid) with -methoxy poly(ethylene
glycol)
the yield.6 Typically, such esterification reactions are carried out at relatively high temperatures (120 150 C) and application of azeotropic solvents (e.g. benzene, cyclohexane etc.)
or reduced pressure to remove the water helps to accelerate esterification.7 Compared with
free radical copolymerization, the esterification method has the advantage of producing
a comb polymer with statistical (rather regular) distribution of the side chains along the
polymer backbone whereas free radical copolymerization results in a gradient copolymer
with decreasing side chain density along the main chain as copolymerization progresses.
This effect is owed to the different reactivities of the ester and the (meth)acrylic acid.
A severe draw-back of the esterification process is the synthesis of short-chain poly(meth)
acrylic acid which is utilized as backbone. Its Mw should be less than 7,000 g/mol and
ideally even < 4,000 g/mol to avoid extremely long backbones which adsorb only slowly
on cement and are not ideally suitable for application in concrete. Additionally, such high
molecular weight can lead to reduced water solubility and/or highly viscous solutions
which are difficult to handle in the plants of admixture companies. Therefore, inspite of
its several advantages the esterification process suffers from the difficulty of preparing a
suitable poly(meth)acrylic acid backbone, because of the strong tendency of (meth)acrylic
acid to homopolymerize and form long chains. As a result of this difficulty, the esterification process is less common among MPEG PCE producers, compared to free radical
copolymerization.
In order to overcome the problems of both methods we introduce a new preparation
method for PCEs which uses macroradicals to achieve the characteristic comb structure of
PCE. The target of our study is to prepare a PCE of reasonable performance from simple,
easily available raw materials such as maleic anhydride and -methoxy poly(ethylene
glycol) in a process which is feasible for common chemical plants. For this purpose,
different synthesis conditions were evaluated and a suggestion for a typical process is
made. Furthermore, the molecular properties and performance in cement paste of the
synthesized PCE samples were compared utilizing mini slump and slump retention tests
and via heat flow calorimetry. Finally, the reaction mechanism underlying the formation
of these PCE polymers was investigated using gel permeation chromatography (GPC) and
1
H NMR spectroscopy.
4SP-302-12
Table 1 Phase composition of the CEM I 52.5 N sample as determined by

Q-XRD using Rietveld refinement
Phase
C3S, m
C2S, m
C3A, c
C3A, o
C4AF, o
Free lime (Franke)
Periclase (MgO)
Anhydrite
CaSO4-hemihydrate
CaSO4-dihydrate
Calcite
Quartz
Arcanite (K2SO4)
wt.%
54.14
26.63
3.28
4.26
2.45
0.1
0.03
2.64
1.21
0.02
3.61
1.16
0.46
A new preparation method for PCEs including a simplified process which uses macroradicals to achieve the characteristic comb structure of PCE is presented as an alternative
to common free radical copolymerization and esterification processes. The PCE products
synthesized according to the new method exhibit competitive dispersing performance
although the process has not yet been optimized.
Materials
Chemicals maleic anhydride (> 99%, Merck Schuchardt OHG, Hohenbrunn,
Germany), -methoxy poly(ethylene glycol) (Polyglycol M350, Polyglycol M500,
Polyglycol M2000) (> 99% purity, Clariant Produkte Deutschland GmbH, Burgkirchen,
Germany), H2SO4 (98% purity, SIGMA-ALDRICH CHEMIE, Steinheim, Germany), tertbutyl peroxybenzoate (98% purity, SIGMA-ALDRICH CHEMIE, Steinheim, Germany)
were used as is without any further purification.
Reference sample a self-synthesized polycarboxylate superplasticizer possessing
high grafting density, designated 45PC2, was used for comparison. This PCE was prepared
from methacrylic acid and -methoxy poly(ethylene oxide) methacrylate ester via aqueous
free radical copolymerization reaction. The molar ratios of the monomers were 2:1, and the
number of ethylene oxide units in the side chain was 45. The synthesis process is described
in detail in.4
Cement A CEM I 52.5 N (Milkeclassic, HeidelbergCement, Geseke plant) was used.
Its phase composition as determined by XRD is presented in Table 1. The average particle
size (d50 value, determined by laser granulometry) was found at 11.5 m. Its density was
3.153 g/cm3 (Helium pycnometry).
Synthesis of PCE Graft Copolymers

MPEG 350 - graft - MPEG 350 mono maleate/maleic anhydride 7.5 g (76 mmol)
of maleic anhydride, 25.2 g of MPEG 350 (72 mmol) and 0.15 g of 98 wt.% H2SO4 were
placed in a 250 mL three-neck round bottom flask equipped with a stirrer and heated for 3
hours at 140 C under constant stirring. Thereafter, 15 g (153 mmol) of maleic anhydride
were dissolved in 20.125 g (58 mmol) of MPEG 350. Furthermore, 1.2 g (6 mmol) of
tert-butyl peroxybenzoate initiator were also added to this homogeneous mixture. Using
a peristaltic pump, this solution was fed continuously into the vessel over a period of 60
min while maintaining the temperature at 140 C. The viscosity of the mixture increased
during the synthesis procedure. When the addition was complete, the mixture was stirred
for another 60 min at 140 C. After that, the mixture was cooled to ~ 70 C and ~ 65 g of
water were added. Then the pH value was adjusted to 7 by addition of 30 wt.% aqueous
NaOH solution. The final product was designated as MPEG 350 - g - MPEG 350 mono
ester/maleic anhydride. It presented a reddish, 34.3 wt.% aqueous solution which was used
without further purification.
MPEG 500 - graft - MPEG 500 mono maleate/maleic anhydride 7.5 g (76 mmol) of
maleic anhydride, 36 g of MPEG 500 (72 mmol) and 0.15 g of 98 wt.% H2SO4 were placed
in a 250 mL three-neck round bottom flask equipped with a stirrer and heated for 3 hours
at 140 C under constant stirring. Thereafter, 15 g (153 mmol) of maleic anhydride were
dissolved in 28.8 g (58 mmol) of MPEG 500, then 1.2 g (6 mmol) of tert-butyl peroxybenzoate initiator were added to this homogeneous mixture. Using a peristaltic pump,
this solution was fed continuously into the vessel over a period of 60 min while maintaining the temperature at 140 C. Also here, viscosity of the mixture increased during the
synthesis process. All further steps were the same as for the MPEG 350 graft copolymer.
The final product was designated as MPEG 500 - g - MPEG 500 mono ester/maleic anhydride. It presented a reddish, 43.8 wt.% aqueous solution which was used without further
purification.
MPEG 2000 - graft - MPEG 2000 mono maleate/maleic anhydride 1.875 g (19 mmol)
of maleic anhydride, 36 g of MPEG 2000 (18 mmol) and 0.15 g of 98 wt.% H2SO4 were
placed in a 250 mL three-neck round bottom flask equipped with a stirrer and heated for 3
hours at 140 C under constant stirring. Thereafter, 3.75 g (38 mmol) of maleic anhydride
were dissolved in 7.2 g (3.6 mmol) of MPEG 2000. Furthermore, 0.6 g (3 mmol) of tertbutyl peroxybenzoate initiator were added to this homogeneous mixture. Using a peristaltic
pump, this solution was fed continuously into the vessel over a period of 60 min while
maintaining the temperature at 140 C. The rest of the preparation followed the method as
described above. The final product was designated as MPEG 2000 - g - MPEG 2000 mono
ester/maleic anhydride. It presented a reddish, 43.7 wt.% aqueous solution which was used
without further purification.
Characterization of PCE samples
Molar masses (Mw, Mn) and polydispersity index (PDI) of the PCE samples were determined utilizing size exclusion chromatography (SEC). Additionally, the anionic charge
density was measured.
Size exclusion chromatography (SEC) PCE solutions containing 10 g/L of the
polymer were prepared for SEC analysis. Measurement was performed on a Waters 2695
6SP-302-12
Separation Module equipped with three Ultrahydrogel columns (120, 250, 500) and an
Ultrahydrogel guard column from Waters, Eschborn, Germany, and a subsequent 3 angle
static light scattering detector (mini Dawn from Wyatt Technology Corp., Santa Barbara,
CA, USA). The polymer concentration was monitored with a differential refractive index
detector (RI 2414, Waters, Eschborn, Germany). Aqueous 0.1 N NaNO3 solution adjusted
to pH 12 with NaOH was used as an eluent at a flow rate of 1.0 mL/min. From the SEC
measurements, the polydispersity index (PDI) and the molar masses (Mw and Mn) were
obtained. The value of dn/dc used to calculate Mw and Mn was 0.135 mL/g (value for polyethylene oxide).8
Specific anionic charge density The specific anionic charge densities of the polymers were determined employing a particle charge detector PCD 03 pH (Mtek Analytic,
Herrsching, Germany). This instrument allows the experimental determination of the
anionic charge of polymers in solution. Here, 0.2 g/L of the polymers were dissolved in DI
water and titrated against an aqueous 0.34 g/L solution of poly-diallyl dimethyl ammonium
chloride (polyDADMAC) until charge neutralization (zero potential) was reached. From
the amount of polyDADMAC consumed to reach a zero potential, the amount of negative
charge per gram of polymer was calculated.9
Calorimetry To detect a potential retarding effect of the novel PCE samples, timedependent heat evolution during cement hydration was monitored. There, 4 g of cement
were filled into 10 mL glass ampoules, mixed at 21 C with the appropriate amount of
aqueous polymer solution (w/c = 0.3), shaken for 1 min in a wobbler and then placed in
an isothermal heat conduction calorimeter (TAMair, Thermometric, Jrflla, Sweden) to
monitor the heat flow of the hydration reaction. Data logging was continued for 7 days.
Dispersing performance in cement
For determination of the paste flow, a mini slump test modified from DIN EN 1015
was utilized and carried out as follows: First, the water-to-cement (w/c) ratio of the paste
without polymer was set to 0.3. At this w/c ratio, the dosages of the PCE samples required
to reach a spread of 260.5 cm were determined. Generally, the polymer was dissolved in
the required amount of mixing water placed in a porcelain cup. When aqueous polymer
solutions were used, then the amount of water contained in the polymer solution was
subtracted from the amount of mixing water. In preparation, 300 g of cement were added
to the mixing water and agitated manually for 1 minute utilizing a spoon, then rested for 1
minute without stirring and were again stirred for 2 minutes. After the stirring, the cement
paste was immediately poured into a Vicat cone (height 40 mm, top diameter 70 mm,
bottom diameter 80 mm) placed on a glass plate and the cone was removed vertically. The
resulting spread of the paste was measured twice, the second measurement being in a 90
angle to the first and averaged to give the final spread value.
For the time-dependent flow behavior of the paste, 400 g of cement were mixed with 120
mL of DI water as described in the procedure above. After each measurement, the slurry
was transferred back into the cup and covered with a wet towel in order to avoid drying.
Before each subsequent measurement, the paste was stirred again for 2 minutes. Measurements were taken every 15 minutes over a total period of 120 minutes.
Fig. 3 SEC spectra of the three synthesized graft polymers

Properties of synthesized PCE polymers
In this novel synthesis process, maleic anhydride as well as MPEG mono maleate ester
were grafted onto MPEG macroradicals serving as backbone of the PCE polymer. This
way, a comb polymer is formed whereby along an MPEG trunk chain, maleic anhydride
and MPEG mono maleate pendants are randomly arranged, as will be shown in the mechanistic part of this paper.
The SEC spectra of the synthesized PCE polymers are displayed in Fig. 3. According to
them, the conversion of the monomers was 71 - 77%. It should be mentioned here that so
8SP-302-12
Table 2 Molar masses and polydispersity index (PDI) of the synthesized

polymers and of a comparative commercial PCE sample
Polymer
MPEG 350 - g - MPEG 350 mono
ester/maleic anhydride
Industrial PCE
Mw (g/mol)
9,100
Mn (g/mol)
3,349
PDI (Mw/Mn)
2.7
Yield
77.3%
6,848
2,946
2.3
72.6%
16,500
6,163
2.7
71.3%
63,840
28,650
2.2
88.0%
Table 3 Specific anionic charge amount of superplasticizer samples

tested in pH 12.5 aqueous solution
Fluid
System
pH 12.5
aqueous
MPEG 350 - g - MPEG

350 mono ester/maleic
anhydride
3,239
Specific anionic charge amount [eq/g]

MPEG 500 - g - MPEG
MPEG 2000 - g - MPEG
anhydride
anhydride
1,249
1,551
45 PC 2
1,033
far we have not attempted to optimize the yield of polymer. Molar composition of the polymers was confirmed by 1H NMR spectroscopy whereby the peak areas of the individual
protons were integrated (spectra not shown here).
The characteristic molecular properties of the synthesized polymers are presented in
Table 2. According to the SEC data, the synthesized PCE polymers exhibit relatively low
molecular weights (Mw, Mn), compared to a commercial benchmark MPEG PCE product
prepared from MPEG - MAA macromonomer via free radical copolymerization. The polydispersity index of the samples lies at ~ 2.5 which indicates a fairly narrow molecular
weight distribution.
Next, the specific anionic charge amounts of the synthesized polymers and of the industrial polymer were determined in DI water. The results are exhibited in Table 3. From the
data, it is obvious that at increasing side chain length of the PCE polymers, the anionic
charge density decreases as expected.
Cement dispersion
Mini slumptest To probe into the dispersing effectiveness of the newly synthesized
PCE samples, the dosages required to achieve a cement paste spread of 260.5 cm were
determined. For this test, a cement paste prepared at a w/c ratio of 0.3 was employed.
According to Fig. 4, the newly synthesized polymers work as effective cement dispersants
at low w/c ratios. Their effectiveness increases with increased length of the MPEG utilized
in the synthesis.
Time - dependent slump loss behavior Next, time - dependant slump loss behavior
of the superplasticizer samples was measured over a period of two hours. In this test,
cement pastes prepared at a w/c ratio of 0.3 treated with the superplasticizer dosages as
shown in Fig. 4 were employed. The results are displayed in Fig. 5. According to this data,
Fig. 4 Dosages of synthesized graft copolymers required

to achieve a paste spread of 260.5 cm (w/c = 0.3)
Fig. 5 Slump loss behavior of cement pastes (w/c=0.3) containing the synthesized graft
copolymers
the synthesized graft polymers maintain workability of the cement paste for a short time
period only.
Generally, polymer MPEG 2000 - g - MPEG 2000 mono ester/maleic anhydride seems
to retain fluidity slightly better than the polymers prepared from the short chain MPEGs
(MPEG 350 and MPEG 500). This trend is owed to the specific molecular architecture of
the PCE polymers, because PCE molecules with longer side chains are more coiled, adsorb
slower and thus can provide better slump retention over time.
Impact on cement hydration To detect a potential retarding effect of the synthesized graft polymers, time - dependent heat evolvement from a cement paste (w/c ratio =
0.3) holding different polymer samples was monitored. The results are exhibited in Fig.
6. There, it is demonstrated that the synthesized graft polymers retard cement hydration,
possibly due to the excessive amount of maleic anhydride used in the synthesis which
10SP-302-12
Fig. 6 Time-dependent heat evolvement from cement paste

(w/c = 0.3) holding the synthesized graft polymers
Fig. 7 Esterification of maleic anhydride with MPEG to

yield a mono maleate ester
partially did not incorporate into the PCE (see SEC spectra in Fig. 3). Again, the polymer
containing MPEG 2000 performs better than the two polymers with shorter pendants.
Grafting Mechanism via Macroradicals
Stepwise analysis of the cascade of reactions occurring during this novel synthesis
method revealed a mechanism as follows:
In the first step, a mono maleate ester is produced via esterification reaction from
-methoxy polyethylene glycol and maleic anhydride, as is shown in Fig. 7.
In the subsequent grafting process, the peroxide initiator at first undergoes thermal
homolytic cleavage to form radicals RO (see Fig. 8). In a second step, -methoxy polyethylene glycol (MPEG) macroradicals are formed through a chain transfer reaction with
the radicals RO whereby hydrogen is abstracted from MPEG. It should be noted that in
previous literature, formation of such macroradicals from peroxide initiators has been well
documented for polyethylene and polypropylene and successful grafting of maleic anhydride onto these macroradicals is described there as well.10-14
Subsequently, maleic anhydride and the unsaturated MPEG mono maleate ester can graft
onto these macroradicals by forming a comb - shaped PCE structure which again presents
a macroradical, see step (3) in Fig. 8. This reaction is favoured by the strong electron
attracting properties of the double bond of maleic anhydride and the mono maleate ester.
Through continued chain transfer reactions as shown in step (4) of Fig. 8, new MPEG
macroradicals are formed which then can undergo the same sequence of grafting reaction
as before, thus leading to an MPEG trunk chain which holds multiple maleic acid and
MPEG maleate pendant groups. Termination occurs when two -methoxy polyethylene
glycol macroradicals combine into the final product.
The proposed chemical structure of the final graft copolymer derived from this reaction
mechanism is displayed in Fig. 9. It should be noted here that due to the complexity of
the reaction, the final product likely consists of a mixture of several polymers exhibiting
different degrees of grafting (as is also the case in MPEG PCE synthesis via free radical
copolymerization using macromonomers, but this fact is often ignored), thus the structure presented in Fig. 9 represents only one of several possibilities. Another possibility
requiring consideration is the formation of poly(maleic acid) homopolymer. However,
the SEC spectra do not provide evidence of its presence. Also, according to literature the
homopolymerization of maleic anhydride is inhibited by organic electron donors such as in
oxygen or sulfur containing compounds.12,13 Consequently, this side reaction was ruled out.
Another consideration was the formation of an - olefin occurring in the initial esterification step from dehydration of MPEG with sulfuric acid. Such olefin could then undergo
copolymerization. This possibility was checked via 1H NMR spectroscopy of the esterification product. However, no signals characteristic for - olefinic protons could be detected.
CONCLUSIONS
A new preparation method for PCEs including a simplified process is presented as an
alternative to the common free radical copolymerization and esterification processes. Here,
the polymerization mechanism relies on the formation of -methoxy polyethylene glycol
macroradicals which are generated by hydrogen abstraction from the MPEG. Subsequently,
maleic anhydride as well as MPEG - mono maleate ester are grafted onto such macroradicals yielding a comb type PCE polymer with carboxylic and EO pendants arranged
along the MPEG main chain. SEC data show that the synthesized graft polymers exhibit
relatively low molecular weights (Mw, Mn). Performance tests indicate that this new type
of PCE can disperse cement well although it was synthesized from simple raw materials.
Future work should focus on further optimization of molar mass and conversion of raw
materials to capture the full potential of this novel PCE chemistry.
12SP-302-12
Fig. 8 Cascade of reactions involved in the grafting of

maleic anhydride and MPEG mono maleate ester onto a
MPEG macroradical
Fig. 9 Proposed chemical structure for the newly synthesized MPEG - g - MPEG mono ester/maleic anhydride PCE
polymers
AUTHOR BIOS
Lei Lei is a Ph.D. student at the Chair for Construction Chemistry at Technische Universitt Mnchen, Germany. Her research focuses on the impact of different clay minerals
on the performance of polycarboxylate - based superplasticizers and new routes to PCE
synthesis.
Chemistry there. His research interests include cement chemistry, chemical admixtures,
cementing.
ACKNOWLEDGMENTS
L. Lei wishes to thank the Jrgen Manchot Foundation for generously providing a scholarship to finance this research at TU Mnchen.
REFERENCES
1. Petit, J. Y.; Wirquin, E.; Khayat, K. H.; and Vanhove, Y., Coupled effect of temperature and superplasticizer on rheological properties of SCC mortar, 5th International
RILEM Symposium on Self-Compacting Concrete, RILEM Publications SARL, Ghent,
Belgium, 2007, pp. 1099 1104.
2. Plank, J., Schroefl, C., Gruber, M., Lesti, M., Sieber, R., Effectiveness of Polycarboxylate Superplasticizers in Ultra-High Strength Concrete: The Importance of PCE compatibility with Silica Fume, Journal of Advanced Concrete Technology, V. 7, No.1, 5-7, 2009,
pp. 5 - 12.
3. Plank, J., PCE Superplasticizers Chemistry, Applications and Perspectives, 18.
IBAUSIL, Weimar, V. 1, 2012, pp. 91-102.
14SP-302-12
hydroxy terminated poly(ethylene glycol) side chains, Cement and Concrete Research, V.
5. Wang, Z. M.; Xu, Y.; Wu, H.; Liu, X.; Zheng, F. Y.; Li, H. Q.; Cui, S. P.; Lan, M. Z.;
and Wang, Y. L., A Room Temperature Synthesis Method for Polycarboxylate Superplasticizer CN patent 101974135 B, 2013, assigned to Beijing University of Technology.
6. Guicquero, J. P.; Maitrasse, P.; Mosquet, M. A.; and Sers, A., A water soluble or
water dispersible dispersing agent, FR Patent 2,776,285, 1999, assigned to Chryso.
7. Hirata, T.; Yuasa, T.; and Nagare, K., Cement admixture and cement composition,
US patent 6,166,122, 2000, assigned to Nippon Shokubai.
8. Teresa, M.; Laguna, R.; Medrano, R.; Plana, M. P.; and Tarazona, M. P., Polymer
characterization by size-exclusion chromatography with multiple detection, Journal of
Chromatography. A, V. 919, No. 1, 2001, pp. 13-19. doi: 10.1016/S0021-9673(01)00802-0
9. Plank, J., and Sachsenhauser, B., Experimental determination of the effective anionic
charge density of polycarboxylate superplasticizers in cement pore solution, Cement and
Concrete Research, V. 39, No. 1, 2009, pp. 1-5. doi: 10.1016/j.cemconres.2008.09.001
10. Ghaemy, M., and Roohina, S., Grafting of Maleic Anhydride on Polyethylene in a
Homogeneous Medium in the Presence of Radical Initiators, Iranian Polymer Journal, V.
12, 2003, pp. 21-29.
11. Gaylord, N. G.; Mehta, R.; Kumar, V.; and Tazi, M., High density polyethylene-gmaleic anhydride preparation in presence of electron donors, Journal of Applied Polymer
Science, V. 38, No. 2, 1989, pp. 359-371. doi: 10.1002/app.1989.070380217
12. Gaylord, N. G., and Mehta, R., Peroxide-catalyzed grafting of maleic anhydride on
to molten polyethylene in the presence of polar organic compounds, Journal of Polymer
Science. Part A, Polymer Chemistry, V. 26, No. 4, 1988, pp. 1189-1198. doi: 10.1002/
pola.1988.080260419
13. Cheng, Q.; Lu, Z.; and Byrne, H., Synthesis of maleic anhydride grafted polypropylene-butadiene copolymer and its application in PP/OMMT/SBS composite as compatibilizer, Journal of Applied Polymer Science, V. 114, No. 3, 2009, pp. 1820-182. doi:
10.1002/app.30678
14. Yin, J. H.; Shi, D.; and Yang, J. H. etal., Functionalization of isotactic polypropylene
with maleic anhydride by reactive extrusion: mechanism of melt grafting, Polymer, V. 42,
No. 13, 2001, pp. 5549-5557. doi: 10.1016/S0032-3861(01)00069-6
SP-302-13
Impact of Slags Contained in Blended

Cement on Dispersing Effectiveness of
PCEs
by Ahmad Habbaba and Johann Plank
In portland slag cements (PSC), different slag compositions can produce variations in
workability due to the disparity in the surface chemistry of the slags. Here, the surface
chemistry of different PSCs dispersed in water was studied in the absence and presence
of polycarboxylate (PCE) superplasticizers. Six PSCs were prepared by mixing portland
cement with 30 or 70 wt.% of three slags. As PCEs, two copolymers based on methacrylic
acidcomethoxy poly(ethylene glycol) methacrylate ester were employed. It was found
that the slags sequester ions from the pore solution, namely Ca2+ and SO42- ions forming
an electrical double layer on the slag surface. Zeta potential measurements confirmed that
different slags can exhibit different surface charges which can strongly affect PCE adsorption. The differences in the amounts of PCEs adsorbed result in different dosages required
to achieve comparable dispersion. Generally, all slag cements tested required less PCE to
achieve the same fluidity as with neat cement.
Keywords: adsorption; anionic charge amount; polycarboxylate; pore solution; portland
slag cement; superplasticizer; zeta potential.
INTRODUCTION
Cement is considered to present the largest industrially made product in the world. Due
to its calcination process at 1450 C (2,642 F), the manufacturing of cement consumes a
huge amount of energy which is associated with a significant emission of greenhouse gas
(7% of the worlds total CO2 emission).1 To overcome this environmental problem, portland
cement is blended with other supplementary cementitious materials (SCMs) to produce
composite cements (e.g. CEM II and CEM III). Blast furnace slag obtained from the iron
and steel industry belongs to the group of SCMs which are blended with portland cement
clinker and gypsum to produce portland slag cement (PSC).2 Nowadays, PSC is widely
recognized as an eco-friendly binder because it can replace a large amount of clinker (up
to 80%) and thus reduce CO2 emission from cement production. In PSC, slag not only acts
as a mere filler or aggregate, but it can also enhance the properties of hardened concrete by
improving sulfate and chloride resistance. This effect is owed to the low content of C3A in
169
170SP-302-13
PSC as well as a decrease in the permeability against water and different types of ions.3-6
It has also been reported that compared to the non-blended portland cement, dispersants
show higher fluidizing properties in PSC.7
Polycarboxylate (PCE) based superplasticizers are applied in the construction industry
to produce highly flowable concrete possessing a low water-to-cement ratio (w/c). Generally, PCEs are comb-shaped copolymers which consist of a negatively charged backbone
comprised of carboxylate groups, and uncharged graft chains, mainly polyethylene glycols.
The charged backbone of PCE can adsorb onto the surface of hydrated cement particles
in three different possible conformations (train, loop or tail), while the non-adsorbed side
chains freely protrude from the cement surface into the pore solution.8 Therefore, the
anionic charge amount of a PCE polymer presents a main factor guiding its adsorption
behavior and thus its dispersing performance. The higher the anionic charge amount of
a PCE molecule, the higher its adsorbed amount. The mechanism behind the dispersing
effect of a PCE is based on a combination of steric hindrance and electrostatic repulsive
forces between the cementitious particles.9-11
In previous works by the authors, the surface chemistry of slag dispersed in deionized
water and synthetic cement pore solution (SCPS) has been studied extensively.12,13 The
major observation was that in slag suspensions, strong interaction occurs between the
surface of slag and the ions present in the cement pore solution. First, at the high pH value of
the suspension, the surface of slag becomes negatively charged as a result of deprotonation
of silanol groups. This negatively charged surface then adsorbs calcium ions, thus forming
a layer of Ca2+ on the surface of slag which in turn attracts sulfate ions present in the pore
solution, hence forming a second ion layer consisting of sulfate ions. Therefore, when slag is
dispersed in SCPS, the apparent surface charge of slag is negative, because of the high sulfate
content. Further deprotonation of silanol groups present on slag occurs over time, and more
calcium and sulfate ions are adsorbed. These processes alter the zeta potential of slag slurries
after a certain time period (~ 3 h) until a stable state of equilibrium is reached.
Furthermore, interaction between slag and PCE superplasticizers was studied as well.13
It was found that a competitive adsorption between the polymers and sulfate ions for positively charged sites present on the slag surface occurs. Sufficiently anionic PCE molecules
occupy adsorption sites on slag while PCEs of low anionic character cannot compete with
SO42- and hence do not adsorb in large amounts on the surface of slag.
The study here continues the previous work from the authors on pure slags and now
investigates the interaction between PCEs and different portland slag cements (PSCs). At
first, the electrical surface charge of PSC samples dispersed in water was determined using
a zeta potential instrument. Next, the concentrations of Ca2+ and SO42- ions contained in the
pore solutions were determined and compared with those occurring in neat cement pore solutions. Furthermore, the physico-chemical interactions between PCE and PSC were studied
via adsorption and zeta potential measurements. Based on these experimental results, the
mechanism behind the different dispersing effect of two chemically different PCE polymers
added to PSC pastes comprised of slags from different sources will be discussed.
Previous studies on the interaction between slag cements and polycarboxylates have
considered slag as an inert component which has no influence on this interaction. There-
Impact of Slags Contained in Blended Cements on Dispersing Effectiveness

of PCEs 171
Table 1 Oxide compositions and properties of cement and GGBFS
samples studied
Oxide content (wt. %)
SiO2
CaO
Al2O3
MgO
TiO2
K2O
Na2O
Fe2O3
Mn3O4
SO3
SrO
ZrO2
BaO
P2O5
Spec. surface area (Blaine)
[cm2/g (ft2/oz)]
d50 value [m]*
Density [g/cm3 (oz/in3)]
CEM I 52.5N
23.56
67.76
3.58
0.53
0.24
0.73
0.02
1.29
0.04
2.61
0.20
0.00
0.05
0.18
Slag S1
35.9
42.8
11.4
6.44
0.82
0.33
0.27
0.45
0.28
2.40
0.09
0.03
0.13
0.00
Slag S2
36.3
36.4
11.5
11.50
0.78
0.66
0.34
0.26
0.22
2.57
0.09
0.03
0.19
0.00
Slag S3
38.6
38.6
12.4
6.40
0.82
0.53
0.45
0.46
0.26
1.55
0.09
0.03
0.00
0.00
3400 (102)
4000 (120)
3480 (104.4)
4080 (122.4)
11.52
3.16 (1.83)
9.53
2.86 (1.65)
10.19
2.91 (1.68)
9.25
2.91 (1.68)
*conversion factor: 1 m = 3.937 10-6 in
fore, the impact of different types of slag on this interaction was ignored. Various slags can
result in considerable differences in workability when incorporated into cement. This effect
is due to the disparity in the surface chemistry of different slag samples which results from
their different ability to adsorb calcium ions. The anionic charge amount of a PCE polymer
and the thickness of the calcium layer present the key factors for the dispersing effect of
PCE in PSC.
Materials
A commercial portland cement sample (CEM I 52.5 N) was used as base cement to
prepare a total of six blends whereby 30 or 70 wt.% of the cement were replaced with
either slag S1, slag S2 or slag S3. The three ground granulated blast furnace slag (GGBFS)
samples were from different sources in Germany. Table 1 lists the oxide composition
(XRF), specific surface area (Blaine instrument) and particle size distribution (d50 value;
laser granulometer) of the neat cement sample and the three slag samples.
Two PCE superplasticizers (denominated as 45PC1.5 and 45PC6) were synthesized
and used in the tests. Free aqueous radical copolymerization was employed to synthesize
the copolymers using sodium peroxodisulfate as initiator and methallyl sulfonic acid as
chain transfer agent. A detailed description of the synthesis process has been published in
previous work.14 These PCEs are based on a copolymer of methacrylic acid and methoxy
poly(ethylene glycol) methacrylate ester (so-called MPEG type PCE). Their general chemical structure is displayed in Fig. 1. In the designations 45PC1.5 and 45PC6, 45 refers to
172SP-302-13
Fig. 1 Chemical structure of the synthesized PCE samples

Table 2 Characteristic properties of the synthesized PCE polymers
Copolymer
45PC1.5
45PC6
Molar ratio
methacrylic
acid:ester
1.5
6.0
Side
chain
nEO
45
45
MW (g/mol)*
196,300
222,300
Mn (g/mol)*
51,900
52,340
Polydispersity
index
(MW/Mn)
3.8
4.2
Hydrodynamic
radius Rh(avg)
(nm)**
8.7
10.4
*conversion factor: 1 g/mol = 0.035 oz/mol; **conversion factor: 1 nm = 3.937 10-8 in
the number of ethylene oxide units (nEO) present in the side chain, whereas 1.5 and 6
refer to the molar ratio between methacrylic acid and the MPEG methacrylate ester. The
characteristic properties of the PCE polymers are presented in Table 2.
Methods
Rheological properties Cement pastes were prepared and their flow properties were
determined using a mini slump test according to DIN EN 1015. The w/c ratios of the
cement pastes were selected such as to produce a flow value (spread) of 18 0.5 cm (7.1
0.2 in) in the absence of PCE polymer (Table 3). In measurements incorporating PCE,
the polymer was added to the mixing water, and its dosage was adjusted to produce a flow
value of 26 0.5 cm (10.2 0.2 in). The test was carried out as follows: In a porcelain cup,
over 1 min300 g (10.58 oz) of PSC were added to the specific amount of mixing water as
given in Table 3, then left to soak for 1 min which was followed by manual stirring with
a spoon for 2 min. Immediately after the end of stirring, the slurry was poured into a Vicat
cone [height 40 mm (1.57 in), top diameter 70 mm (2.76 in), bottom diameter 80 mm (3.15
in)] placed on a glass plate and filled to the brim. The cone was removed vertically and
the resulting spread of the paste (= diameter of slurry cake) was taken as flow value of the
slurry. The diameter was measured twice perpendicularly, and the two values were aver-

of PCEs 173
Table 3 Water to cement ratios determined via mini slump test to achieve
180.5 cm spread
Sample
w/c ratio
Cal. w/c*
CEM I 52.5 N
0.505
0.505
30% S1
0.57
0.81
30% S2
0.55
0.79
30% S3
0.52
0.74
70% S1
0.60
2.00
70% S2
0.60
2.00
70% S3
0.55
1.83
* calculated as a ratio between the cement part present in PSC and water
aged to give the slump spread value. Each test was repeated three times, and the average
was reported as slump flow value (the margin of error was 3%). Generally, the amount of
water introduced with the PCE solution was subtracted from the amount of mixing water to
maintain comparable w/c ratios. The dosage of PCEs is expressed on a dry mass basis and
is stated in % by weight of cement (% bwoc).
Ion concentrations in pore solutions From these cement pastes, 10 mL (0.338 fl oz)
were taken in 20 min intervals over a total period of 180 min, and then were centrifuged for
10 min (8,500 rpm) and diluted with 0.1N HCl to avoid precipitation of calcium carbonate.
The ion concentrations were obtained from an atomic absorption spectroscope. Additionally, sulfate concentrations were quantified by utilizing ion chromatography.
Electrokinetic properties of blended cements Electrokinetic properties were measured
using Electroacoustic Spectrometer. The highly solids loaded suspensions used in this work
require an electroacoustic instrument to obtain zeta potential values which are representative of the conditions occurring in actual concrete.15 Zeta potential values were measured
during the dropwise addition of aqueous solutions of the copolymers (concentration 10 wt.
%, pH=7) to the cement pastes, and the zeta potentials of the slurries were recorded as a
function of PCE concentration.
PCE adsorption on PSC Polymer adsorption was determined according to the depletion
method. Different dosages of PCE copolymer were added to the individual cement pastes
(w/c ratios as listed in Table 3), stirred for 2 min, centrifuged for 10 min (8,500 rpm) and
then diluted with 0.1 N HCl to remove inorganic carbonates and to prevent dissolution of
carbon dioxide in the alkaline solution. A High TOC II apparatus was employed to determine the total organic carbon content in the supernatants. The adsorbed amount of PCE
was calculated by subtracting the concentration of PCE found in the supernatant from the
initial PCE concentration used prior to contact with PSC.
Zeta potential of PSC pastes
In previous publications where the authors studied neat slag suspensions, it was found
that in alkaline solution the surface of slag is generally negatively charged as a result of
deprotonation of the silanol groups present on the surface.12,13 This charged surface attracts
counter ions from the pore solution which then adsorb and form an electrical double layer
on the surface of slag. The thickness of this layer mainly depends on two parameters: the
type of slag (i.e. the number of silanol groups) and the ionic strength of the pore solution.
Therefore, different slag samples exhibit different zeta potential values. Over time, the zeta
potential values change due to partial hydration of the slag whereby some ions are released
into the pore solution. These results explain the fundamental behavior of slag in cement pore
solution and were taken as basis for the following experiments with the slag cement samples.
174SP-302-13
Fig. 2 Time-dependent zeta potential of CEM I 52.5 N and PSC paste (w/c ratios as
shown in Table 3)
For all 30% PSC samples (i.e. cement containing 30% slag), the initial zeta potential
values were negative (~ - 4.5 mV), as is shown in Fig. 2, while the 70% PSC samples
consistently exhibited positive initial zeta potential values (e.g. +3.5, +1.6 and +1.5 mV
for 70% S1, 70% S2 and 70% S3, respectively). When considering the initial zeta potential value for the neat cement slurry (- 6 mV), it seems that the addition of slag to cement
changes the zeta potential to less negative or even positive values, depending on the amount
of slag added. Furthermore, the PSC samples containing the same amount of slag (30% or
70%) still exhibit slightly different initial zeta potential values, signifying that variations
in the chemical composition of the slags cause different behaviors. For the slag samples
studied here, the zeta potential values of the resulting PSC blends follow the trend S1 >
S2 ~ S3 which infers a higher positive charge to the blend incorporating S1 than S2 or
S3. These results are in agreement with previous results on pure slag.12,13 Over time, the
surface charges of the PSCs increased to more positive values and stabilized after ~ 3 h
(Fig. 2). The results indicate that the zeta potential values of PSCs are strongly related to
the amount and type of slag added.
Ion Concentrations in PSC Pore Solution
The time - dependent evolution of Ca2+ and SO42- concentrations present in the pore
solutions of all PSC slurries was measured. These measurements were performed to determine the impact of each slag on the ionic strength of the pore solutions. For comparison,
neat portland cement slurries prepared at the same w/c ratios as the PSCs (see Table 3)
were also analyzed. Their w/c ratios were calculated assuming that slag constitutes an inert
component in the PSC samples.
When compared with the neat portland cement slurries, lower concentrations of Ca2+
and SO42- ions in the pore solutions were observed for all PSC samples. For the 30% PSC
samples, the slags seem to consume significant amounts of Ca2+ (Fig. 3 top). This trend
was even stronger for PSCs possessing high slag contents (70% PSC) (Fig. 3 bottom).
In all PSC pastes the changes in Ca2+ concentrations over time followed almost the same

of PCEs 175
Fig. 3 Time-dependent evolution of Ca2+ concentrations in CEM I 52.5 N and PSC pore
solutions (conversion factor:1 g/L = 0.001 oz/fl oz)
trend found for neat portland cement pastes. This effect is due to the continuous release of
Ca2+ ions from the cement at the same rate as the slag consumes Ca2+.
On the other hand, the 30% PSCs consume considerable amounts of sulfate ions from
the pore solution, compared to the related neat portland cement samples (Fig. 4 top). The
sulfate ions released from the slag samples were not significant because of the substantially
higher concentration of sulfate present in cement. Contrary to this, the sulfate concentrations occurring in the pore solutions of 70% PSCs are lower or comparable to those from
the neat cement (Fig. 4 bottom). According to these data, slag S2 releases the highest
amount of SO42- ions into the pore solution while slag S3 produces a medium and slag S1
the lowest concentration of sulfate ions in the pore solution. The same trend was observed
before for the pure slag samples.13 Note that the concentrations of SO42- ions in the pore
solutions of the 70% PSCs is lower than that of the 30% PSCs. This signifies that in the
30% PSC samples, the negative sulfate layer adsorbed onto the positive sites of the slag
surface is thicker. It explains why the 30% PSCs exhibit negative zeta potential values
while the 70% PSCs show positive zeta potentials (see Fig. 2).
In general, the presence of slag in blended cements significantly changes the ionic
strength of the pore solution, compared to the neat cement. Furthermore, the pore solution
compositions of PSC samples containing the same amount of different slags can differ
considerably as a result of variations in the chemical composition of the slags. This finding
confirms that in such composite cements, slag is not an inert constituent.
176SP-302-13
Fig. 4 Time-dependent evolution of SO42- concentrations in CEM I 52.5 N and PSC pore
solutions (conversion factor:1 g/L = 0.001 oz/fl oz)
Dispersing effectiveness of PCEs in PSC
Similar to what has been reported before from another research group,7 PCEs show
higher fluidizing effect in PSC compared to the non-blended portland cement. Here, for
the 70% PSC pastes, enhanced flow properties were recorded compared to the 30% PSCs.
However, the dosage of PCEs required to achieve the same paste spread of PSC samples
incorporating the same amount of different slags varied significantly.
Generally, in CEM I 52.5 N as well as PSC pastes significantly lower dosages of PCE
polymer 45PC6 (0.012 0.06% bwos) were required to achieve the same paste spread than
with 45PC1.5 (0.09 0.55% bwos), as is shown in Fig. 5. The reason behind this result
is the high anionic charge amount of polymer 45PC6 [~ 1200 eq/g (34286 eq/oz)].
Obviously, PCE polymer 45PC1.5 which is a polycarboxylate typically used in ready-mix
concrete to provide extended slump life presents a less effective dispersant. To clarify the
mechanism behind this effect, the adsorption of both polymers on the PSCs was compared.
Adsorption of PCE on PSC
For the polymers 45PC1.5 and 45PC6, adsorption isotherms in PSC pastes were produced
using TOC measurement. In general, for both PCEs the adsorbed amounts increase with
dosage until they reach a saturation point (Langmuir type isotherm; see Fig. 6). The saturated adsorbed amounts are the lowest for the cement blended with slag S3 and the highest
for cements which contain slag S1.

of PCEs 177
Fig. 5 Dosages of polycarboxylate superplasticizers

required to obtain a target slump flow of 260.5 cm (10.20.2
in) for PSC pastes (w/c ratios as shown in Table 3)
Comparable adsorbed amounts of PCE 45PC1.5 were found for all 30% PSCs (see Fig.
6 top left). The same trend has been observed before for the pure slags dispersed in SCPS.13
Apparently, adsorption of 45PC1.5 is driven by a gain in entropy and not by electrostatic
attractive forces, as in 30% blended slag systems its adsorption is independent of the different
positive surface charges of the slag.16 Furthermore, the low adsorbed amount explains why
45PC1.5 is not an effective dispersant, as was shown in Fig. 5. In general, for the 70% PSCs
the adsorbed amounts of 45PC6 and 45PC1.5 follow the order: Slag S1 > S2 > S3. This trend
correlates well with the zeta potential results presented in Fig. 2: PSCs containing slag S1
exhibit the most positive surface charge while those blended with slag S3 show the lowest.
The adsorbed amounts of PCE polymer 45PC6 on the blended cements were consistently
lower than those on the neat cement (Fig. 6). This explains why sample CEM I 52.5 N
required a higher dosage of this polymer to achieve the 26 cm (10.2 in) spread compared
to the same cement blended with slag (Fig. 5). The reason behind this effect is that cement
particles adsorb higher amounts of this polymer [3 mg/g (21 gr/lb)] than pure slag [0.4
1.8 mg/g (2.8 12.6 gr/lb)].12,13 Apparently, PCE polymer 45PC6 exhibits a higher affinity
for cement than for slag.
Influence of PCE on Zeta potential of PSC slurry
The evolution of zeta potential during dropwise addition of the PCE solutions to the PSC
pastes is shown in Fig. 7. When PCE is added to the PSC slurries, a shift in zeta potential
values towards the isoelectric point (IEP) occurs which confirms that adsorption of PCE
does indeed take place. The reason behind such shift to the IEP is the steric effect of the
polyethylene oxide (PEO) side chains contained in the PCEs. They move the shear plane of
the zeta potential further away from the particle surface. At such distance, the zeta potential
approaches a value close to zero.17
178SP-302-13
Fig. 6 Adsorption isotherms for PCE polymer 45PC1.5 (top) and 45PC6 (bottom) on
PSC (w/c ratios as shown in Table 3) (conversion factor: 1 mg/g = 7.015 gr/lb)
Fig. 7 Effect of dosage of PCE polymers 45PC1.5 (left) and 45PC6 (right) on zeta potential of PSC pastes (w/c ratios as shown in Table 3)
CONCLUSIONS
When blended with cement, various slags are known to produce different results with
respect to workability of the concrete or mortar. This is due to the disparity in the surface
chemistry of the slag samples which results from their dissimilar ability to sequester
calcium and sulfate ions from the pore solution. Partial replacement of cement by slag
alters the ionic strength of the pore solution, and this change varies depending on the
composition of the slag. In PSC pastes, at first significant amounts of calcium ions are

of PCEs 179
Fig. 8 Schematic illustration of the electrochemical double layer existing in the equilibrium state on the surface of the slag samples in PSC suspension, and the consequences for
the adsorption behavior of PCE polymers on these slags
released from the cement into the pore solution which are partially taken up by the negatively charged slag surface. The amount of calcium consumed by a slag sample depends on
the absolute value of its negative surface charge which can vary. Slags having thick layers
of Ca2+ ions on their surface adsorb higher quantities of PCE and thus require increased
dosages of superplasticizers to achieve the same fluidity as slags possessing a thin Ca2+ ion
layer. This mechanism is schematically summarized in Fig. 8.
This study demonstrates that in blended cements, strong interaction between polycarboxylate superplasticizers and the surfaces of both slag and portland cement occurs. Apparently, slag is not inert towards PCE. In portland slag cements, normally lower polycarboxylate dosages are required compared to the pure cement. Furthermore, when using slags
from different sources, the dosages of superplasticizer required to establish comparable
180SP-302-13
flowability can vary significantly. Accordingly, different qualities of slag may impact the
economies of portland slag cements in different ways.
AUTHOR BIOS
Ahmad Habbaba is a Postdoctoral researcher at the Chair for Construction Chemistry, Technische Universitt Mnchen, Germany. He studied chemistry and received his
BS, MS and Diploma degree from Aleppo University (Syria) and a Ph.D. degree from
TU Mnchen. His research interests include interaction of admixtures with cement and
supplementary cementitious materials (SCMs).
Chemistry there. His research interests include cement chemistry, concrete admixtures,
cementing.
ACKNOWLEDGMENTS
A. Habbaba wishes to express his gratitude and sincere appreciation to the Chair for
Construction Chemistry for financing his postdoctoral researches.
REFERENCES
1. Malhotra, V. M., Sustainability issues and concrete technology Paper presented at
7th International Symposium on Cement and Concrete (ISCC), Jinan (China), May 9 - 12,
2010.
2. Moranville, M., Cement made from Blast furnace slag, in: Leas Chemistry of
Cement and Concrete, 4th edition, Peter Hewlett, Elsevier Science & Technology Books,
2004, pp 637-678.
3. Frearson, J. P. H., Sulphate resistance of combination of Portland cement and ground
granulated blast furnace slag, Proceedings, 2nd International Conference on Fly Ash,
Silica Fume, Slag and Natural Pozzolans in Concrete, Madrid, Spain, ACI, SP-91, Detroit,
MI, USA, 1986, pp. 14951524.
4. Osborne, G. J., Durability of Portland blast-furnace slag cement concrete, Cement and
Concrete Composites, V. 21, No. 1, 1999, pp. 11-21. doi: 10.1016/S0958-9465(98)00032-8
5. Lang, E., Blast furnace cements in: Structure and performance of cements, 2nd ed.
Bensted, J. and Barnes, P., Spon Press, 2002, pp. 310-325.
6. Fu, X.; Hou, W.; Yang, C.; Li, D.; and Wu, X., Studies on high-strength slag and fly
ash compound cement, Cement and Concrete Research, V. 30, No. 8, 2000, pp. 12391243. doi: 10.1016/S0008-8846(00)00312-4
7. Palacios, M.; Puertas, F.; Bowen, P.; and Houst, Y. F., Effect of PCs superplasticizers
on the rheological properties and hydration process of slag-blended cement pastes, Journal
of Materials Science, V. 44, No. 10, 2009, pp. 2714-2723. doi: 10.1007/s10853-009-3356-4
Research, V. 30, No. 2, 2000, pp. 197-207. doi: 10.1016/S0008-8846(99)00230-6

of PCEs 181
9. Ramachandran, V. S., and Malhotra, V. M., Superplasticizers, in: Ramachandran, V.
S., editor, Concrete Admixtures Handbook, 2nd ed, New Jersey, USA, Noyes publications,
1996, pp. 410-517.
10. Sakai, E.; Kang, J. K.; and Daimon, M., Action mechanisms of comb-type superplasticizers containing grafted polyethylene oxide chains, in: Malhotra, V. M. editor,
6th International Conference on Superplasticizers and Other Chemical Admixtures in
Concrete, Nice, France, SP-195, CANMET/ACI, 2000, pp. 75-90.
11. Blask, O., and Honert, D., The electrostatic potential of highly filled cement suspensions containing various superplasticizers, in: Malhotra, V. M., editor, 7th International
Conference on Superplasticizers and Other Chemical Admixtures in Concrete, Berlin,
Germany, SP-217, CANMET/ACI, 2003, pp. 87-101.
12. Habbaba, A., and Plank, J., Interaction Between Polycarboxylate Superplasticizers
and Amorphous Ground Granulated Blast Furnace Slag, Journal of the American Ceramic
Society, V. 93, No. 9, 09 2010, pp. 2857-2863. doi: 10.1111/j.1551-2916.2010.03755.x
13. Habbaba, A., and Plank, J., Surface Chemistry of Ground Granulated Blast Furnace
Slag in Cement Pore Solution and Its Impact on the Effectiveness of Polycarboxylate
Superplasticizers, Journal of the American Ceramic Society, V. 95, No. 2, 2012, pp.
768-775. doi: 10.1111/j.1551-2916.2011.04968.x
hydroxy terminated poly(ethylene glycol) side chain, Cement and Concrete Research, V.
15. Dukhin, A. S., and Goetz, P. J., Acoustic and Electroacoustic Spectroscopy, Langmuir, V. 12, No. 18, 1996, pp. 4336-4344. doi: 10.1021/la951086q
16. Plank, J., and Sachsenhauser, B., Experimental determination of the thermodynamic
parameters affecting the adsorption behaviour and dispersion effectiveness of PCE superplasticizers. V. M. Malhotra (Ed.) 9th CANMET/ACI Conference on superplasticizers
and other chemical admixtures in concrete (supplementary papers), ACI, Seville, 2009,
pp. 87-102.
17. Plank, J., and Sachsenhauser, B., Impact of molecular structure on zeta potential
and adsorption conformation of -allyl--methoxypolyethylene glycol - maleic anhydride superplasticizers, Journal of Advanced Concrete Technology, V. 4, No. 2, 2006, pp.
233-239. doi: 10.3151/jact.4.233
182SP-302-13
SP-302-14
Preparation and Characterization of StarShaped Polycarboxylate Superplasticizer

by Xiao Liu, Ziming Wang, Jie Zhu, Ming Zhao, Wei
Liu, and Dongjie Yin
A polycarboxylate superplasticizer (PCE) with a novel star-shaped structure was prepared
through copolymerization of acrylic acid (AA), isobutenyl polyethylene glycol (IPEG), and
star-shaped polymerizable active center by an esterification between polyol and AA. In
the first esterification step, the esterification rate reached more than 95% with the catalyst/polyol ratio of 0.07:1, inhibitor/AA ratio of 0.04:1 (or 0.011:1), water-carrying agent
dosage of 70g and esterification time of 7 hours. In the second polymerization step, the
highest fluidity of cement paste was achieved at the initiator/AA/IPEG ratio of 0.28: 3.3: 1.
Infrared spectroscopy (IR) and 1H Nuclear magnetic resonance (1H NMR) measurements
were used for structural characterization, and the spectral results confirmed the products
star-shaped structure. Furthermore, this star-shaped PCE exhibited higher energy efficiency than the conventional comb-shaped PCE, indicated by its excellent paste fluidity
and adsorption behavior in cement paste.
Keywords: polycarboxylate superplasticizer; star-shaped; molecular design; adsorption;
fluidity.
INTRODUCTION
In recent years, with the developing technology of construction engineering, the demand
for increased concrete quality has become greater. Polycarboxylate superplasticizer (PCE)
invented in 1981 as a novel class of superplasticizers clearly represented a milestone technology in construction engineering, due to its advantages of low dosage, high slump retention and water-reducing capability. PCE is a key chemical admixture which can be added to
concrete mixtures to improve workability1 and prepare high-performance concrete, and has
rapid development and growth in construction engineering.2,3 The main function of PCE is
to enable a decrease in the water-cement ratio (W/C) without losing fluidity of the cement
paste, which results in higher strength and better durability of concrete.4,5 The properties
of the PCE polymer are closely related to its structure, and thus the structural innovation of PCE is very important to promote the development of new PCE technologies. At
present, the structure of common PCE is a comb-shape structure, and thus it has exhibited
a unique mechanism and excellent performance in cement paste or concrete systems up to
183
184SP-302-14
now. As the modification of new molecular structures becomes a research focus, polymers
with branched structures have received greater attention because of their unique properties
and potential application prospects.6,7 The branched or star-shaped polymer has particular
characteristics including low viscosity, more side chains and functional ends compared to
conventional comb polymers. Therefore, a PCE with this structure is likely to bring higher
adsorption efficiency and wider coverage on the surfaces of cement particles.8
According to this research idea, a star-shaped PCE is synthesized by means of the structure innovation, which relies on the theory of molecular design to achieve some unique
characteristics. This type of PCE may exhibit higher adsorption efficiency and excellent
paste fluidity by virtue of its special structure. Only a few researchers have reported the
synthesis and properties of polymers with branched or star-shaped structures which were
usually synthesized by a complex Atom Transfer Radical Polymerization (ATRP).9,10
However, there were no studies reported on the field of PCE with innovative branched or
star-shaped structure synthesized by a simple synthetic technique for improving cement
paste fluidity.
In this study, a PCE with a star-shaped structure was successfully synthesized by the
simple synthetic reactions without having to resort to the ATRP method. Its detailed reaction conditions were determined by investigating catalyst, inhibitor and water-carrying
agent dosages and esterification time. Additionally the structure and performances of
PCEs with either star-shape or conventional structure were respectively characterized and
compared to confirm their unique structure and excellent fluidity performances. Finally,
the effects of adsorption behaviors of the star-shaped PCE in cement paste system were
evaluated.
To improve the workability, efficiency and other performances of PCE is an important
subject for both PCE researchers and to improve concrete performance. The structure innovation based on the theory of molecular design creates a polymer with novel structure, and
depending on this, the synthesized PCE displays excellent performances to achieve the
improvement of workability efficiency. Furthermore, this research pays researchers attentions to how to design PCE structures to solve application problems. This novel-structural
type of PCE can enrich the relevant theories and widen its application to further provide the
theoretical basis of synthesis of the high-workability PCE. Consequently, this novel PCE is
suitable for the chemical admixture of high-performance concrete.
Materials
Reaction raw materials include: Isobutenyl polyethylene glycol (IPEG, Mn=2348g/
mol, Mw=2484g/mol), acrylic acid (AA), pentaerythritol, ammonium persulfate (APS),
phenothiazine, toluene, concentrated sulfuric acid, P-toluene sulfonic acid, hydroquinone,
sulfamic acid and sodium hydroxide (NaOH). Reference cement P.I.42.5 was supplied by
China Building Materials Research Institute (Beijing, China). The chemical and mineral
compositions of reference cement are illustrated in Table 1.
Preparation and Characterization of Star-Shaped Polycarboxylate

Superplasticizer185
Table 1Chemical and mineral compositions of reference cement
Composition
SiO2, %
Al2O3, %
Fe2O3, %
CaO, %
MgO, %
SO3, %
Na2O, %
Loss, %
f-CaO, %
C3S, %
C2S, %
C3A, %
C4AF, %
Reference cement P. I. 42.5

22.93
4.29
2.89
66.23
1.92
0.35
0.70
1.48
0.64
58.78
21.38
6.49
8.77
Fig. 1Schematic diagram of the synthesis of star-shaped PCE.

Synthesis and calculation
The pentaerythritol was mixed with AA at a molar ratio of 1:5 in a four-neck roundbottom flask with a stirrer, and then the catalyst, inhibitor and water-carrying agent were
added in sequence, followed by heating to the set temperature with stirring. At the end
of reaction which lasted for several hours, the esterification products can be obtained by
vacuum-distillation to remove the water-carrying agent.
The esterification product and IPEG with a molar ratio of 1:20 were diluted by distilled
water in a four-neck round-bottom flask which was placed in a constant temperature bath
at 65C with stirring. When stirred homogeneously, AA and initiator in aqueous solution
was dropwise added to the flask. After reacting at a constant temperature for 5 hours, the
NaOH aqueous solution was used for adjusting the pH value to 6-7 followed by cooling to
room temperature, thus yielding the final product. The schematic diagram of the synthesis
of the star-shaped PCE is shown in Fig. 1. The specimens used for the characterization
were further purified by repeated precipitation into the excess diethyl ether.
The calculation formula of esterification rate (E%) is listed as the following equation:

E% = [(10.8 C NaOH VNaOH 40 m) / 10.8] 100%
(1)
186SP-302-14
Where CNaOH = molar concentration of sodium hydroxide solution, 1 mol/L; VNaOH =

consumed volume of sodium hydroxide solution for neutralization, L; m = mass of catalyst
containing crystal water, g.
Items of investigation
The 1H Nuclear Magnetic Resonance (1H NMR) spectra of PCEs were obtained at room
temperature (25C) with an ARX-400 spectrometer (Bruker Co., Germany) operating at
a frequency of 400MHz, and the chemical shift values were expressed in values (ppm)
relative to tetramethylsilane (TMS) as an internal standard. Samples for 1H NMR were
prepared by dissolving samples in the solvent, i.e., deuterated water (D2O) as internal
reference.
The Fourier Transform infrared (FTIR) spectra of the PCE was prepared by mixing a
fixed mass of solid PCE with KBr, which was then pressed into a disk, and analyzed by
a VERTEX 70 Fourier transform infrared spectrometer (Bruker Co., Germany) at room
temperature (25C). The solid PCE samples were prepared via dissolution and precipitation at least three times, and then dried in vacuum at 60C for 24 hours to constant weight.
The fluidity of cement pastes treated with PCE were measured according to the standard
method (GB 8077-2000 Uniformity Test Method for Concrete Admixture)11 described in
the National Standards of the Peoples Republic of China, at a water-cement ratio (W/C)
of 0.29, and the dosage was based on the weight ratio of solid PCE to cement. To investigate the effect of PCE dosage on the paste fluidity, the fluidities of cement pastes mixing
with PCE at different dosages were examined. Moreover, to study the maintenance of
fluidity, cement pastes mixed with PCE were examined every 60 min within a total period
of 120 min. For each test of cement paste fluidity, the cement paste mixture was poured
into a truncated cone (2.36 in. [60 mm] height, 1.42 in. [36 mm] top diameter, 2.36 in. [60
mm] bottom diameter) on a glass plate, and then the cone was vertically removed. After
30 seconds, the diameter of paste was recorded as the fluidity of paste, and the resulting
spread of the paste was measured twice. The second measurement was perpendicular to the
first measurement, and the average was calculated to yield the spread value. The adsorption amounts of PCEs on the surfaces of cement particles were evaluated according to a
reported method.12 The non-adsorbed portion of polymer remaining in solution at equilibrium condition was determined by analyzing the total organic carbon (TOC) content of the
solution. In a typical experiment, 30 g of cement, 120 g of deionized water and the amount
of PCE (remaining the same liquid concentration as the fluidity testing of cement paste,
namely 0.15:29) were mixed and then the homogeneous slurry was filled into a 50 mL
centrifuge tube to be centrifuged for 5 min at 6000 rpm in a TGL-16C High speed bench
centrifuge produced by ShangHai Anting Scientific Instrument Factory (Shanghai, China).
The supernatant was diluted 100 times with deionized water, and then the total organic
carbon of the solution was determined on a Vario TOC cube instrument (Elementar Analysensysteme GmbH, Germany). The adsorption amounts of PCE were calculated from the
difference between the TOC content of the polymer reference sample in the initial solution
and that of the supernatant of cement paste. The adsorption amount changes with time
were obtained by testing the PCE adsorption after 5 min, 30 min, 60 min, 90 min and 120
min, respectively. Measurements were generally repeated three times and the average was
reported as the adsorbed amount.

Superplasticizer187
Table 2Equilibrium reaction temperature and esterification rate at different
dosages of water-carrying agent
Water carrying agent, g
50
60
70
75
80
85
Equilibrium reaction temperature, C

128
122
119
118
117.8
117
Esterification rate, %
58
57.9
66.5
62
63
62.4

Effect of water-carrying agent dosages on the equilibrium reaction
temperature and esterification rate
The effect of water-carrying agent dosages on the equilibrium reaction temperature and
esterification rate is shown in Table 2.
From Table 2, the esterification rate is the highest when the dosage of water-carrying
agent is 70 g and the corresponding equilibrium reaction temperature is 119C. In the
range of water-carrying agent dosage from 50 to 70g, the esterification rate increases with
the increase of water-carrying agent dosage, which results from the higher water-carrying
efficiency (corresponding to higher dosage of water-carrying agent), leading to higher
esterification rate. However, with continual increase of water-carrying agent dosage, the
esterification rate decreased. This is attributed to the excessive amounts of water-carrying
agent, leading to a low reactant concentration and a low esterification rate. Correspondingly, the equilibrium reaction temperature decreased with the increase of water-carrying
agent dosage, which is also caused by the reflux of a great deal of cooled water-carrying
agent.
Effects of catalysts and their dosages on the esterification rate
The concentrated sulfuric acid and p-toluene sulfonic acid were selected as catalysts.
The ratio of acid to pentaerythritol is 1.25 and the dosage of water-carrying agent is 70 g
in this experiment. The effects of catalysts and their dosages on the esterification rate are
shown in Fig. 2.
From Fig. 2, the esterification rate is the highest when the molar ratio of p-toluene
sulfonic acid to alcohol is 0.07 and the molar ratio of sulfuric acid to alcohol is 0.11.
However, the real value of esterification rate is likely to be affected owing to the total
amount of water containing a part of dehydrated water for the dehydrating property of
concentrated sulfuric acid. Insufficient catalyst dosages can lead to incomplete esterification reactions; similarly, excessive catalyst dosages can also lead to low esterification rate,
which is possibly because the excessive dosage of catalyst leads to the occurrence of side
reactions. Overall, the p-toluene sulfonic acid is chosen as the esterification catalyst, and
its molar ratio to alcohol is 0.07.
Effects of inhibitors and their dosages on the esterification rate
The hydroquinone and phenothiazine were selected as inhibitors in this experiment;
besides, the ratio of acid to pentaerythritol, the water-carrying agent dosage, and the molar
188SP-302-14
Fig. 2Esterification rates at catalysts different molar ratio

to alcohol.
Table 3Effects of inhibitors and their dosage on the esterification rate
Catalyst
Hydroquinone
Phenothiazine
Dosage, %
2
3
4
4.5
5
0.2
0.5
0.8
1.1
1.4
1.7
Esterification rate, %
94.9
95.0
95.3
93.3
93.0
88.7
93.1
92.4
95.6
91.7
94.2
ratio of catalyst to alcohol are 1.25, 70 g and 0.07, respectively. The effects of inhibitors
and their dosages on the esterification rate are shown in Table 3. The dosage is the molar
percentage of inhibitor to AA.
From Table 3, phenothiazine plays an inhibiting effect at the lower dosages; whereas the
hydroquinone can exhibit an effective inhibiting effect only at the relatively higher dosages.
It also can be seen from Table 3 that the esterification rates are the highest when the molar
ratios of hydroquinone and phenothiazine to AA are 4% and 1.1%, respectively. With the
continual increase of inhibitor dosage, the esterification rate decreases. This is probably
because the acrylic acid is apt to homopolymerize with each other rather than esterification
reaction when the dosages of inhibitors are too low, leading to the decrease of esterification
rate. With the continual increase of inhibitor dosage, the free radicals produced from reaction system will be excessively captured, which can interfere with or even eliminate the

Superplasticizer189
Fig. 3Esterification rates for different reaction times.

role of catalyst and leads to the decrease of esterification rate. Therefore, the proper molar
ratios of hydroquinone and phenothiazine to AA are 4% and 1.1%, respectively.
Effect of reaction times on the esterification rate
The esterification rate is also affected by the reaction time, and thus the investigated
reaction times were selected from 4 hours to 8 hours. In this experiment, the ratio of acid to
pentaerythritol, the water-carrying agent dosage, and the molar ratio of catalyst to alcohol
are 1.25, 70 g and 0.07, respectively. The effect of reaction times on the esterification rate
is shown in Fig. 3.
The results show that the esterification rate increases with the increase of reaction time
in the range of 4-6 hours. The esterification rate directly affects the quality of esterification
product, and further correlates to the performance of the final synthesized PCE. Besides,
the esterification reaction time should be theoretically prolonged at a proper esterification
temperature to ensure the esterification rate as high as possible. However, in Fig. 3, the
esterification reaction rate gradually increases after reacting for 6 hours, and it changes
slightly with the extension of reaction time. Based on economical considerations and
energy conservation, the proper esterification time therefore is determined as 7 hours.
Characterization of esterification product
The esterification product in this study was characterized by 1H NMR (Fig. 4).
The esterification reaction occurred between the hydroxyl groups of pentaerythritol and
the carboxyl groups of AA. During the process of the esterification reaction, its methylenes initially linked to the hydroxyl groups change to link to ester groups. As a result, the
electron withdrawing effect of ester structure leads to its peak shifting to the lower field.
From the peak positions in Fig. 4, the primary peaks of esterification product in 1H NMR
spectrum correspond to the three H atoms on the -CH=CH2 bonds (6.154ppm, 6.328ppm,
190SP-302-14
Fig. 41H NMR spectrum of esterification product.

5.963ppm). The structure diagram displayed in Fig.4 clearly indicates the peaks and their
corresponding H atoms in the molecular structure. All of the displayed characteristic peaks
confirm the occurrence of esterification reaction and the indication of a relatively ideal
molecular structure for the esterification product.
Effects of AA/IPEG ratios on fluidity performances of paste mixing with
synthesized PCE
After the confirmation of structure of esterification product which is used for the subsequent polymerization, the relevant polymerization conditions also should be determined.
The effects of AA/IPEG ratios on the fluidity performances of paste mixing with synthesized PCE at a dosage of 0.15% are shown in Fig. 5.
The ratio of AA to IPEG affects PCEs structure and performance in cement paste. From
Fig. 5, the fluidity of cement paste increased slightly with the increase of AA/IPEG ratio
in the range of 2.6-3.3. This is possibly because the effective adsorption sites increase
caused by the increased proportion of carboxyl groups with the increase of AA/IPEG ratio.
Thereafter, the fluidity of cement paste reached maximum at the AA/IPEG ratio of 3.3 and
then decreased with a continual increase of AA/IPEG ratio. This is possibly because the
continual increase of AA/IPEG ratio is equivalent to decrease the density of side chain,
which is a key factor to determine the workability and efficiency of PCE. Too low density
of side chain will lower the steric hindrance effect of side chain (polyoxyethylene), and
thus leads to the decrease of cement paste fluidity.
Effects of initiator/IPEG ratios on fluidity performances of paste mixing with
synthesized PCE
There were better fluidity performances of cement pastes at the initiator/IPEG ratio of
around 0.3 by experiments, and thus the initiator/IPEG ratio was investigated in the range
of 0.26-0.38. The effects of initiator (APS)/IPEG ratios on the fluidity performances of
paste mixing with synthesized PCE at a dosage of 0.15% are shown in Fig. 6.

Superplasticizer191
Fig. 5Fluidities of cement paste mixing with PCE at

different AA dosages.
Fig. 6Fluidities of cement paste mixing with PCE at

different initiator/IPEG ratios.
From Fig. 6, the fluidities of cement paste increased and then decreased with the increase
of APS/IPEG ratio; besides, the cement paste exhibited the best fluidity at the APS/IPEG
ratio of 0.28. The dosage of initiator affects the polymerization rate and molecular weight
in a free radical polymerization. Too low dosage of initiator will produce fewer primary
radicals, and thus cause incomplete polymerization. With the increase of initiator dosage,
the polymerization rate is accelerated and the reaction becomes sufficient; furthermore,
the suitable initiator dosage is helpful to obtain the product with suitable molecular
192SP-302-14
Fig. 71H NMR spectrum of the synthesized polymerization product.

weight, which optimizes its adsorption rate on the surfaces of cement particles and ensures
good fluidity retention. Too high dosage of initiator will reduce the molecular weight of
PCE, which can weaken the steric hindrance effect and further lead to the low fluidity
performance.
Characterizations of polymerization product
The synthesized polymerization product was characterized by 1H NMR to confirm its
star-shaped structure. The 1H NMR spectrum of the synthesized polymerization product is
shown in Fig. 7.
From Fig. 7, the peaks at 4.5-5.0 ppm and 3.3-3.8 ppm were distinctly higher than other
peaks, because they correspond to the H atoms of the polyethylene glycol connected with
the arm structure. Also from Fig. 7, the peaks at 2.314 and 1.645 ppm correspond to the H
atoms of methylenes, whose schematic structure diagram is also displayed in Fig. 7. It is
well-known that the peaks corresponding to H atoms of methylenes will not be displayed
at these positions if the esterification product does not polymerize with other monomers or
forms other products. The total integral area for these two peaks also indicates many starshaped structures have formed. All of these characteristic peaks confirm the polymerization reaction and the expected polymerization product with a star-shaped structure.
The FTIR spectra of the conventional comb-shaped PCE and synthesized star-shaped
PCE are shown in Fig. 8, and these spectra were analyzed by means of other reported
references.13,14
It can be seen from Fig. 8 that the two PCEs both exhibit characteristic peaks at around
1108cm-1 and 1554cm-1. The peak at 1108 cm-1 belongs to the vibration of polyethylene
glycol chain, and the characteristic peak at 2500-3300 cm-1 corresponds to its stretching
vibration. The peak at 1554 cm-1 belongs to the symmetric vibration of C=O bond in
carboxylic acid. By contrast, there is a large peak at around 1772cm-1 in the spectrum
of the star-shaped PCE but no peak at this position in the spectrum of the conventional

Superplasticizer193
Fig. 8FTIR spectra of the synthesized polymerization

product.
comb-shaped PCE. The peak in this position corresponds to the ester groups, and thus, this
result proves the star-shaped structure of this PCE due to the esterification reaction. All of
the analysis indicates that the star-shaped PCE has not only the conventional groups such
as polyethylene glycol and carbonyl groups, but also the characteristic groups, i.e., ester
groups, which confirm the completion of reactions and the achievement of the targeted
structure.
Fluidity performances of the pastes mixing with PCEs
The star-shaped PCE was synthesized according to the above determined conditions;
besides, the conventional PCE as reference was synthesized with the same conditions
as the polymerization-step of star-shaped PCE. The fluidity performances of the cement
pastes mixing with the two PCEs respectively at a dosage of 0.15% are shown in Fig. 9.
From Fig. 9, the saturation dosages of the synthesized star-shaped and conventional
comb-shaped PCEs were about 0.3% and 0.4%, respectively. These results indicate that
excellent fluidity performance can be achieved for lower dosages of star-shaped PCE,
demonstrating its high workability efficiency. Furthermore, the fluidity values of the
synthesized star-shaped PCE were higher than those of conventional comb-shaped PCE
over the range of experimental dosages.
Adsorption mechanism of star-shaped PCE in cement paste
As is well-known, the carboxyl groups of PCE adsorbs on the surfaces of cement
particles when added to the cement paste; besides, the hydrophilic part, i.e., polyethylene
glycols of PCE can retain the fluidity of cement paste. Thus, the PCEs adsorption behavior
on the surfaces of cement particles is an important factor to investigate the mechanism of
star-shaped PCE in cement paste. The adsorption behaviors and the relationship between
194SP-302-14
Fig. 9Fluidities of cement pastes mixing with PCEs at

different dosages (W/C=0.29).
Fig. 10Adsorption behaviors of star-shaped and conventional PCEs in cement paste (PCE dosage=0.15:29).
adsorption amount and fluidity for star-shaped and conventional PCEs in cement pastes are
shown in Fig. 10 and Fig. 11 respectively.
It can be seen from Fig. 10 and Fig. 11, the adsorption amount and rate of star-shaped PCE
are higher than those of conventional comb-shaped PCE, but also the similar results were
displayed in terms of relationship between adsorption amount and fluidity. This is caused
by their different molecular structures. For the star-shaped PCE compared with conventional PCE, its multi-arm structure leads to a stronger steric-hindrance effect; moreover, it

Superplasticizer195
Fig. 11Relationship between adsorption amount and

fluidity for star-shaped and conventional PCEs in cement
paste.
has smaller hydrodynamic volume15 which leads to a higher contents of polar groups per
unit volume and has more adsorption sites which leads to a stronger affinity and a higher
probability of surface adsorption. This is in accordance with the results reported by Kazuo
Yamada.16 Also, with the hydration process, a part of conventional PCE molecules gradually lose the workability for the coverage of hydrate layer by its lower adsorption amount
and more surfaces of cement particles. This well agrees with other researchers study17
investigating the effects of charge density of PCE on the dispersant behavior and showing
that the dispersion effect of PCE is well correlated to the amount of adsorbed polymer and
can be interpreted in terms of surface coverage. Then, we suppose that the star-shaped PCE
still has other arms dispersed in the paste pore solution, and the hydrate layer is still
covered by these free arms, leading to good dispersion retaining of cement particles in
the cement paste. This speculative mechanism is diagramed in Fig. 12.
Based on the above results and analysis, the good fluidity performances of star-shaped
PCE are attributed to its special multi-arm structure, which has higher energy efficiency
to achieve a better workability in cement pastes. This star-shaped PCE is one member of a
large PCE family which has very broad performances. It is meaningful for researchers to
investigate the PCE with novel structure, which can diversify the PCE family.
CONCLUSIONS
Based on the results of this experimental investigation, the following conclusions are
drawn:
1. A star-shaped PCE was successfully synthesized through two-step reactions: the first
esterification between pentaerythritol and AA, and the second free radical polymerization
among the esterification product, IPEG and AA.
196SP-302-14
Fig. 12Adsorption mechanism of star-shaped PCE in

cement paste.
2. The esterification rate can reach above 95% at a water-carrying agent dosage of 70g, a
catalyst/alcohol molar ratio of 0.07:1, an inhibitor (hydroquinone) /AA molar ratio of 0.04:1
(or phenothiazine/AA molar ratio of 0.011:1), and a reaction time of 7 hours; furthermore,
in the second polymerization step, the best fluidity of cement paste was achieved at the
initiator/AA/IPEG ratio of 0.28: 3.3: 1.
3. The structure of esterification product was characterized by its 1H NMR spectrum,
which confirms the occurrence of esterification; moreover, the structure of final synthesized star-shaped PCE was characterized by its 1H NMR and FTIR spectra, which confirm
the completion of subsequent polymerization and the achievement of an ideal star-shaped
structure.
4. The saturation dosages of the synthesized star-shaped PCE and a conventional combshaped PCE are 0.3% and 0.4% respectively, indicating star-shaped PCEs dosage improved
efficiency; besides, the synthesized star-shaped PCE exhibits better fluidity performance
and adsorption behaviors.
5. The improved performances of star-shaped PCE are attributed to its novel branched
multi-arm structure with higher energy efficiency, and thus it can be considered as a newtype PCE to provide the theoretical basis and technological application in cement and
concrete research.
AUTHOR BIOS
Xiao Liu is an associate professor at the College of Materials Science and Engineering
in Beijing University of Technology. He majored in Materials Science and Engineering
and received his PhD from Beijing University of Chemical Technology. His research
interests include functionalized polycarboxylate superplasticizer with high performance
by molecular structure design, as well as the interface adsorption of polycarboxylate
superplasticizer in cement paste system.

Superplasticizer197
Ziming Wang is a professor at the College of Materials Science and Engineering in
Beijing University of Technology. He majored in Materials Science and Engineering and
received his PhD from Beijing University of Technology. He is a vice executive secretary
of Association of Concrete Admixture. His research interests include high-performance
cement-based materials and rheology of cement paste.
Jie Zhu is a Researcher Engineer at Beijing BBMG Cement Energy Technology Co., Ltd.
Ming Zhao is a master student in Beijing University of Technology.
Wei Liu is a master student in Beijing University of Technology.
Dongjie Yin is a master student in Beijing University of Technology.
ACKNOWLEDGMENTS
The authors wish to express their gratitude and sincere appreciation to the National
Natural Science Foundation of China (Grant number: 51208012), Research Fund of
New Teachers for the Doctoral Program of Higher Education of China (Grant number:
3c009011201301), Project of Central Research Institute of Building and Construction Co.,
Ltd (Contract number: CBM2014Ky01-01) and Project of China Railway Engineering
Materials Technology (Anhui) Ltd (Project number: 40009011201409) for financing this
research work.
REFERENCES
1. Bykyac, A.; Tuzcu, G.; and Aras, L., Synthesis of copolymers of methoxy polyethylene glycol acrylate and 2-acrylamido-2-methyl-1-propanesulfonic acid: Its characterization and application as superplasticizer in concrete, Cement and Concrete Research, V.
2. Sakai, E.; Ishida, A.; and Ohta, A., New trends in the development of chemical
admixtures in Japan, Journal of Advanced Concrete Technology, V. 4, No. 2, 2006, pp.
1-13. doi: 10.3151/crt1990.17.2_1
3. Cerulli, T.; Clemente, P.; Decio, M.; Ferrari, G.; Gamba, M.; Salvioni, D.; and Surico,
F., A new superplasticizer for early high-strength development in cold climates, Seventh
CANMET/ACI International Conference on Superplasticizers and Other Chemical Admixtures in Concrete, edited by V, M, Malhotra., American Concrete Institute, Farmington
Hills, 2003, 113 pp.
4. Lu, S., H., Liu, G., Ma, Y, F., and Li, F., Synthesis and application of a new vinyl
copolymer superplasticizer, Journal of Applied Polymer Science, V. 117, No. 1, 2010, pp.
273-280.
5. Plank, J., and Dai, Z., M., Keller, H., Hossle, H., and Seidl, W., Fundamental mechanisms for polycarboxylate intercalation into C3A hydrate phases and the role of sulfate
present in cement, Cement and Concrete Research, V. 40, No. 1, 2010, pp. 45-57. doi:
10.1016/j.cemconres.2009.08.013
198SP-302-14
6. Voit, B., New developments in hyperbranched polymers, Journal of Polymer

Science. Part A, Polymer Chemistry, V. 38, No. 14, 2000, pp. 2505-2525. doi:
10.1002/1099-0518(20000715)38:14<2505::AID-POLA10>3.0.CO;2-8
7. Jikei, M., and Kakimoto, M., Hyperbranched polymers: a promising new class of
materials, Progress in Polymer Science, V. 26, No. 8, 2001, pp. 1233-1285. doi: 10.1016/
S0079-6700(01)00018-1
8. Liu, X., and Wang, Z., M., and Cui, S, P., Preparation method of high-performance
star-shaped polycarboxylate superplasticizer, US patent, 13/910, 785, 2013.
9. Dai, F.; Sun, P.; Liu, Y.; and Liu, W., Redox-cleavable star cationic PDMAEMA by
arm-first approach of ATRP as a nonviral vector for gene delivery, Biomaterials, V. 31,
No. 3, 2010, pp. 559-569. doi: 10.1016/j.biomaterials.2009.09.055
10. Francis, R.; Lepoittevin, B.; Taton, D.; and Gnanou, Y., Toward an easy access to
asymmetric stars and miktoarm stars by atom transfer radical polymerization, Macromolecules, V. 35, No. 24, 2002, pp. 9001-9008. doi: 10.1021/ma020872g
11. GB 8077-2000, Methods for Testing Uniformity of Concrete Admixture, State
Bureau of Quality and Technical Supervision of the Peoples Republic of China, 2000. (in
Chinese).
12. Plank, J.; Sachsenhauser, B.; and Reese, J., D., Experimental determination of the
thermodynamic parameters affecting the adsorption behaviour and dispersion effectiveness of PCE superplasticizers, Cement and Concrete Research, V. 40, No. 5, 2010, pp.
699-709. doi: 10.1016/j.cemconres.2009.12.002
13. Verdonck, B.; Gohy, J.; Khousakoun, E.; Jrme, R.; and Prez, F., D., Association
behavior of thermo-responsive block copolymers based on poly(vinyl ethers), Polymer, V.
46, No. 23, 2005, pp. 9899-9907. doi: 10.1016/j.polymer.2005.07.079
14. Plank, J., and Yu, B., Preparation of hydrocalumite-based nanocomposites using
polycarboxylate comb polymers possessing high grafting density as interlayer spacers,
Applied Clay Science, V. 47, No. 3-4, 2010, pp. 378-383. doi: 10.1016/j.clay.2009.11.057
15. Aggarwal, S., L., Structure and properties of block polymers and multiphase
polymer systems: an overview of present status and future potential, Polymer, V. 17, No.
11, 1976, pp. 938-956. doi: 10.1016/0032-3861(76)90170-1
17. Pourchet, S.; Liautaud, S.; Rinaldi, D.; and Pochard, I., Effect of the repartition
of the PEG side chains on the adsorption and dispersion behaviors of PCP in presence of
sulfate, Cement and Concrete Research, V. 42, No. 2, 2012, pp. 431-439. doi: 10.1016/j.
SP-302-15
Influence of Diester Content in

Macromonomers on Performance of
MPEG-Based PCEs
by Johannes Paas, Maike W. Mller, and Johann
Plank
Macromonomers for MPEG type of PCEs are produced through esterification of methacrylic acid (MAA) with methoxypoly(ethylene glycol) (MPEG) yielding the MPEG-MAA
ester. However, PEG impurities present in MPEG may lead to MAA diester (PEG-di-MAA)
formation. Such diester can cause crosslinking of the PCE polymer which might reduce its
dispersing power. To investigate this effect, MPEG-MAA macromonomers containing 0
20 wt. % of PEG-di-MAA diester were used in PCE synthesis. It was found that when the
PEG-di-MAA content in the macromonomer exceeds 2 wt. %, then dispersing effectiveness
starts to decrease and the solution viscosity of the PCE increases. Surprisingly, incorporation of the diester into the PCE polymer does not occur randomly. Instead, two distinct
species of crosslinked PCE molecules (Mw ~ 300.000 and ~ 3 mio g/mol) are formed within
the first minutes of copolymerization. Apparently, the crosslinked PCE species counteract
the dispersing effect of the main product.
Keywords: cement; crosslinking; diester; dispersing performance; macromonomer; polycarboxylate ether; superplasticizer.
INTRODUCTION
The invention of polycarboxylate-based superplasticizers (PCEs) in the 1980s greatly
broadened the application of superplasticizers in concrete.1,2 These comb-type copolymers
contain polyethylene oxide side chains and show a very high water reducing capability
combined with low dosage of polymer.3,4 In Europe and North America, methacrylate ester
(MPEG)-based polycarboxylates are widely used.5,6 The mechanism behind the excellent
dispersing effect of the PCEs is based on their interaction with cement hydrates.7-9
MPEG-PCEs can be synthesized by three different methods: (1) free radical copolymerization of methacrylic acid with MPEG-MAA macromonomer; (2) by grafting of MPEG
onto a poly(methacrylic acid) backbone; and (3) by transesterification of the methylester of
methacrylic acid with MPEG.
199
200SP-302-15
Fig. 1 Preparation of MPEG-MAA macromonomer

through esterification of methacrylic acid
In industrial manufacturing of MPEG-type PCEs free radical copolymerization involving
an MPEG-MAA macromonomer presents the predominant route for preparation. Typically,
MPEG-MAA macromonomers are produced through the esterification of methacrylic acid
with -methoxy polyethylene glycol (MPEG) using e.g. an acid catalyst.10 Furthermore,
the precursor compound MPEG is obtained via reaction of poly(ethylene glycol) (PEG)
with methanol.11 However, depending on process conditions this reaction might be incomplete and residual PEG is then contained in the MPEG precursor. Thus, in the following
synthesis step where methacrylic acid is reacted with such contaminated MPEG, not only
the desired MPEG-MAA macromonomer (Fig. 1), but also a second macromonomer, PEGdi-MAA diester is obtained as by-product (Fig. 2).
However, such PEG-di-MAA diester can crosslink the PCE molecules formed during the
subsequent copolymerization step. Fig. 3 schematically illustrates how crosslinked PCEs
are formed when the MPEG-MAA macromonomer is contaminated with PEG-di-MAA.
In this work, through free radical copolymerization MPEG-PCEs were prepared from
methacrylic acid and MPEG-MAA macromonomer containing 0 20 wt. % PEG-di-MAA
as impurity at a molar ratio of 3.2: 1. The side chain of the PCEs consisted of 23 EO units.
The synthesized copolymers were characterized with respect to their molecular properties using gel permeation chromatography. Furthermore, their dispersing performance
was tested using mini slump testing. Finally, to understand the crosslinking mechanism,
representative samples were pulled during the copolymerization reaction, and time-dependent formation of the crosslinked polymer fractions was tracked.
From these experiments it was hoped to gain insight into the potentially negative effect
of PEG-di-MAA impurities on the dispersing performance of PCE. Furthermore, as a
Influence of Diester Content in Macromonomers on Performance of MPEGBased PCEs 201
Fig. 2 Formation of PEG-di-MAA diester as by-product in

the macromonomer synthesis from PEG impurities contained
in the MPEG precursor
Fig. 3 Formation of crosslinked MPEG-PCE from MPEG-MAA macromonomer which is

contaminated with minor amounts of PEG-di-MAA diester
202SP-302-15
Table 1 Physical properties and chemical composition of cement CEM I

52.5 N sample
Properties / Oxides
Specific gravity, kg/m3 (Ib/ft3)
Blaine fineness, m2/kg (ft2/Ib)
Loss of ignition, %
SiO2
Al2O3
CaO
MgO
SO3
Na2O
K2O
TiO2
MnO
Fe2O3
wt. %
3150 (197)
368.3 (67,822)
1.90
20.53
5.13
61.47
1.74
3.11
0.05
0.83
0.32
0.05
2.82
guideline for the industry, it was aimed to establish threshold values for the diester content
which should not be exceeded.
It has often been speculated that crosslinking of MPEG-PCEs might occur when the
MPEG-MAA macromonomer is contaminated with PEG-di-MAA diester. However, no
literature exists which provides scientific evidence for this effect. In our study, the impact
of 0 20 wt. % diester contained in the macromonomer on molecular properties and
dispersing performance of the resulting PCE was systematically investigated. The goal was
to understand the consequences of such impurities for PCE performance and ultimately to
develop guidelines for the industry which can help to ensure optimal quality for MPEG
type PCEs.
Materials
Ordinary portland cement (OPC) type CEM I 52.5 N was used. The physical properties
and oxide analysis of the cement sample are listed in Table 1.
For PCE synthesis, as MPEG-MAA macromonomer (nEO = 23) Polyglykol MA 1000
70%, as PEG-di-MAA diester (nEO = 23) NK Ester 23G and methacrylic acid 99% were
used. Na2S2O8 was uses as initiator and 3-mercaptopropionic acid 99% as chain transfer
agent. For neutralization of the synthesized polymers, 30 wt. -% aqueous NaOH was used.
Procedures
Generally, the MPEG PCE samples were synthesized via aqueous free radical copolymerization from methacrylic acid (MAA) and -methoxy polyethylene glycol methacrylate ester (MPEG-MAA) holding 23 ethylene oxide (EO) units. Additionally, PCE samples
incorporating 0 20 wt. % of PEG-di-MAA diester (nEO = 23) were synthesized using
the same procedure by replacing part of the MPEG-MAA macromonomer with PEG-di-
Fig. 4 Gel permeation chromatogram of the PCE copolymer prepared from pure macromonomer (no diester
present)
MAA diester. Details of the synthesis process are described in a previous publication.5
Molecular properties (Mn, Mw) of the synthesized PCE samples were obtained via gel
permeation chromatography (GPC) using separation modul 2596 equipped with refractive index detector 2414 and three-angle static light scattering detector mini Dawn and
QELS. Three UltrahydrogelTM columns (120, 250, and 500) were used for separation of the
polymer fractions. As eluent, a mixture of 0.1 M NaNO3 and 0.1 g/L NaN3 dissolved in DI
water adjusted to pH = 12 was used.
Additionally, time-dependent formation of crosslinked PCEs was tracked by pulling
representative 2 mL samples at defined intervals from the ongoing polymerization batch
holding a total volume of 80 mL. The samples were immediately analyzed by GPC.
Dispersing effectiveness of the synthesized PCE samples in cement paste (w/c = 0.3)
was assessed via mini slump test adapted and modified after DIN EN 1015-3.12 The
tests were carried out as follows: firstly, over one minute 300 g of cement were filled into
a porcelain cup containing 90 mL of water, then left to soak for one minute and stirred for
two minutes with a spoon. Thereafter, the cement paste was poured into Vicat cone (height
4.0 cm (1.57 in.), bottom diameter 8.0 cm (3.15 in.), and top diameter 7.0 cm (2.76 in.)),
placed on a glass plate and filled to the brim. The cone was removed vertically and the
diameter of the resulting cement cake was taken as slump flow (or spread) value. All tests
were performed twice and the avarage value was reported as cement spread.
Kinematic viscosities of the aqueous PCE solutions were obtained on an Ubbelohde
015T viscometer. Furthermore, measurements of shear-dependent dynamic viscosity were
performed on a HAT Synchro-lectric viscometer.
Synthesis and properties of non-crosslinked PCEs
At first, a non-crosslinked PCE was synthesized from pure MPEG-MAA macromonomer
free of PEG-di-MAA diester. Its GPC diagram is shown in Fig. 4. There, the refractive
index detector (dRI) indicates a narrow peak of high intensity between 16.8 and 23.5 mL
204SP-302-15
Fig. 5 Dispersing performance of PCE polymers synthesized from macromonomers holding 0 10 wt. % diester
elution volume which corresponds to the PCE polymer. Its molar masses were found at ~
34,000 g/mol (Mw) and ~ 17,000 g/mol (Mn) respectively, thus signifying a polydispersity
index of 2.0. The peaks at elution volumes of 23.8 to 25 mL are attributable to unreacted
MPEG macromonomer and methacrylic acid. Based on the RI signal, a conversion of
89.1% for the monomers was achieved. The peak at elution volumes from 28.5 to 30.5 mL
can be assigned to salt from the eluent (NaNO3). Furthermore, at elution volumes between
14.5 and 16.8 mL the light scattering signal (LS) detects a very intense peak which can be
attributed to the homopolymer of methacrylic acid (PMAA). Its concentration is extremely
low (< 0.1 wt. %), as indicated by the RI signal which is concentration-dependent while its
molar mass is extremely high, as evidenced by the LS detector which is mass-dependent.
In the next step, additional PCE polymers were synthesized whereby in the MPEG-MAA
macromonomer up to 10 wt. % were replaced with PEG-di-MAA diester. The resulting
polymers were characterized via mini slump test in cement paste.
Dispersing performance of PCE samples
The results obtained from dosage-dependent mini slump tests are presented in Figure
5. It was found that considering the typical error of this test, diester contents of 1 wt. % did
not noticeably affect the dispersing performance of the PCE samples. Whereas, beginning
at 2% addition of the diester and more clearly at 3%, higher PCE dosages are required to
achieve the target spread of 26 0.5 cm (10.26 inch). For example, at 3% diester content,
the increase in PCE dosage required is ~ 14%, compared to the non-modified reference
polymer. This trend continues quite linearly for polymers holding up to 5% diester. There,
the increase in dosage vs. the reference polymer is 21%.
However, at diester contents of > 5%, a considerable jump in dosage occurs whereby
the PCE synthesized from 5.8% diester already requires 63% more dosage than the reference polymer. Interestingly, at even higher diester contents (8.1% and 10%) the dosages
increase only slightly.
Fig. 6 Gel permeation chromatogram of PCE copolymers

prepared with 8.1 wt. % diester present in the macromonomer
The results of the mini slump tests allow to conclude that beginning at 2 wt. % addition
of diester, a minor decrease in PCE performance can be detected. This effect becomes more
pronounced for 3% diester content and is particularly strong at diester additions of > 5%.
Hence, to ensure high quality of PCE, the diester content present in the macromonomer
should be kept below 2 wt. %, and preferably even at less than 1 wt. %. Such values seem
to be feasible in large-scale industries manufacturing of PCE macromonomers.11
To understand the reason behind the negative effect of the diester content on dispersing
performance, the composition of the synthesized PCE samples was analysed via GPC.
Analysis of crosslinked PCEs
As an example for the procedure in the analysis of the GPC spectra, the diagram for
the PCE polymer prepared in the presence of 8.1 wt. % of diester is shown in Fig. 6. The
chromatogram was divided into five parts: (1) elution volumes 14 -16 mL representing a
highly crosslinked PCE; (2) elution volumes 16 - 18 mL referred to crosslinked polymer;
(3) elution volumes 18.5 - 23.5 mL for non-crosslinked polymer; (4) elution volumes 23.5
- 28.3 mL representing unreacted monomers including macromonomer; and (5) elution
volumes 28.3 - 30.5 mL for salt (NaNO3).
Only data for the segments (1) (3) in the diagrams will be discussed, because only they
represent the PCE polymer.
A comparison of the GPC diagrams of the PCE samples prepared with 0 8.1 wt. %
diester content in the macromonomer is shown in Fig. 7.
From this figure it can be recognized that, already at 1% diester addition, a shoulder
develops, which increases in intensity by increasing diester contents. It represents a crosslinked PCE with an Mw of ~ 300,000 g/mol. At even higher diester contents, a third peak
representing another, highly crosslinked PCE (Mw ~ 3,000,000 g/mol), was observed. From
the GPC spectra, the individual contents of non-crosslinked, crosslinked and highly crosslinked PCE polymer fractions were determined. The results are summarized in Table 2.
It was found that already at 1% diester content, the synthesized PCE polymer contains
~ 3.8% of crosslinked and 0.15% of a highly crosslinked fraction. Fortunately, such low
contents of crosslinked PCE do not seem to severely affect the dispersing performance of
206SP-302-15
Fig. 7 Gel permeation chromatograms of PCE copolymers prepared with 0 8.1 wt. %
diester present in the macromonomer
the PCE (see Fig. 5). Whereas at 2% diester content, the total amount of crosslinked and
highly crosslinked molecules reaches ~ 7% of total PCE mass, and this now starts to negatively impact the dispersing capacity (see Fig. 5). Even more, at > 5% diester content the
amount of highly crosslinked PCE (Mw several mio. g/mol) suddenly surges considerably

Table 2 Analysis of polymer fractions contained in PCE samples prepared
with different diester contents in the macromonomer
Content
of diester
in macromonomer
[%]
0.0
1.0
2.0
3.0
4.0
5.0
5.8
8.1
Non-crosslinked
Mw
wt.-%
[g/mol]
[%]
Polymer fractions
Crosslinked
Mw
wt.-%
[g/mol]
[%]
31,600
27,960
28,670
28,590
30,520
31,860
34,030
38,310
298,400
353,300
357,700
387,100
366,000
506,900
480,000
100.00
96.07
93.18
90.35
89.02
86.19
87.25
83.08
3.78
6.59
9.18
10.47
11.51
9.24
12.18
Highly crosslinked
Mw
wt.-%
[g/mol]
[%]
1,121,000
3,496,000
2,287,000
3,376,000
2,607,000
8,017,000
9,033,000
0.15
0.23
0.46
0.52
2.30
3.51
4.73
which explains the very significant increase in the dosage required for those PCEs (see
Fig. 5).
A graphical display of the individual diester content-dependent polymer fractions reveals
that above ~ 3% diester addition, the amount of crosslinked PCE remains fairly constant
whereas further addition of diester only leads to highly crosslinked PCE species (Fig. 8).
The results signify that incorporation of the diester macromonomer into the PCE molecule does not occur statistically (randomly), but, the presence of diester leads to the formation of two distinctly different polymer species with characteristic molecular masses (Mw
~ 300,000 500,000 g/mol and ~ 3 mio g/mol) which are ~ 15 or 200 times higher than
that of the non-crosslinked reference polymer. Consequently, depending on the amount of
diester present, either a bimodal or trimodal polymer composition results.
The amount of the PCE molecules linked together in the crosslinked polymers can be
calculated by dividing Mw of the crosslinked polymer by Mw of the non-crosslinked reference PCE. Using this method it was found that in the crosslinked PCEs, 10 15 individual
PCE molecules are linked together. For the highly crosslinked fraction, this number is ~
30 300 molecules. This result implies that the portion of slightly crosslinked PCE is
much less harmful to PCE performance than the highly crosslinked fraction. Its formation
represents the main cause for deterioration of PCE performance.
Time-dependent evolution of PCEs crosslinking
To investigate the mechanism of crosslinking, time-dependent sampling was performed
during the synthesis of the PCEs with 4.0% diester content in the macromonomer. Aliquot
samples were pulled at 5, 15, 30, 60, 120, 180, 240 and 360 minutes of reaction time. The
samples were then characterized via GPC measurement. Table 3 shows the results of the
analysis.
Surprisingly and most remarkably, the crosslinked PCE forms in relatively high amount
(~ 25%) already at the beginning of the copolymerization. Subsequently, its content drops
to ~ 3% within the next 4 hours of reaction time and then remains constant until the end of
the copolymerization. Here, no highly crosslinked PCE fraction was detected.
208SP-302-15
Fig. 8 Mass fractions of non-crosslinked, crosslinked and

highly crosslinked polymer contained in PCEs prepared
from macromonomer holding 0 8.1% diester
Table 3 Time-dependent evolution of crosslinked PCE occurring during
synthesis of PCE using a macromonomer holding 4% of diester
Reaction
time
, [min]
5
15
30
45
60
120
180
240
300
360
final
neutral
Non-crosslinked
Mw
wt.-%
[g/mol]
[%]
72,950
74.80
63,340
71.73
50,680
75.30
49,940
78.55
46,880
80.10
37,690
82.88
33,630
87.46
31,910
90.43
31,910
90.90
30,900
90.11
30,610
90.41
29,960
90.21
PCE Polymer
Crosslinked
Mw
wt.-%
[g/mol]
[%]
244,800
25.20
303,000
28.27
338,000
24.70
362,000
21.45
356,500
19.90
331,400
17.12
323,600
12.54
331,400
9.57
341,300
9.10
321,700
9.90
326,100
9.59
317,200
8.72
Highly crosslinked
Mw
wt.-%
[g/mol]
[%]
-
Viscosifying effect of crosslinked PCEs

In order to quantify the viscosifying effect of the individual PCE polymers, measurements of the kinematic and dynamic viscosity of their aqueous solutions (solid content
25 wt. %) were performed. The results are displayed in Table 4. It was observed that
starting at ~ 4% diester content, solution viscosity increases considerably. Even more,
diester contents of 10% lead to excessively high solution viscosities. In fact, during the
synthesis of these polymers formation of hydrogels occurred. This observation suggests
that the crosslinked and highly crosslinked polymer fractions counteract the dispersing
effect of the non-crosslinked PCE.

Table 4 Kinematic and dynamic viscosity of synthesized PCE solutions
(solid content 25 wt. %)
Content of diester in
macromonomer
[%]
0.0
1.0
2.0
3.0
4.0
5.0
5.8
8.1
10.0
15.0
19.0
20.0
Solution viscosity
kinematic
[cSt]
19.6
21.7
22.9
25.7
25.0
31.7
31.3
43.0
666.7
-
dynamic
[cP]
39.0
39.7
40.7
41.7
42.7
49.0
54.7
56.0
448.0
1,742.7
2,422.0
1,538.7
CONCLUSIONS
Several PCE polymers were synthesized to investigate a potential crosslinking caused
by impurities of PEG-di-MAA diester contained in the macromonomer. It was found that
starting from 2% diester content, dispersing effectiveness of the synthesized PCE polymers
in cement paste (w/c ratio 0.3; CEM I 52.5 N) was progressively affected, less at lower
diester content and very significantly at higher diester contents ( 5%). Excessive diester
contents ( 10%) lead to the formation of hydrogels which can no longer disperse cement.
Remarkably, incorporation of the diester into MPEG-PCEs does not occur randomly, but
in the copolymerization process and depending on the diester content, one or two distinctly
different species of crosslinked PCE molecules are formed, thus producing a bimodal or
trimodal polymer distribution. Time-dependent analysis of the formation of crosslinked
polymers during the synthesis revealed that the crosslinked species are formed within the
very first minutes of copolymerization whereas by the progress of the reaction, non-crosslinked species become predominant. Our investigation suggests that MPEG macromonomers with diester contents of 1 wt. % or less are safe to obtain a PCE polymer of superior quality. When encountering difficulties with quality, a PCE producer may consider to
determine the diester content present in the macromonomer via HPLC. The 1% threshold
value represents what major macromonomer producers currently supply to the market.
Thus, when using macromonomers of such quality, no negative effect from the diester
content on PCE performance is to be expected.
AUTHOR BIOS
Johannes Paas studied chemistry at Technische Universitt Mnchen, Germany. The
present work was performed during his Master thesis at the Chair for Construction
Chemistry.
210SP-302-15
Maike Mller studied chemistry at Technische Universitt Mnchen, Germany.

Currently, she works on the the subject of structure performance relationship for PCE
superplasticizers within her PhD study at the Chair for Construction Chemistry.
Chemicals there. Research interests include cement chemistry, chemical admixtures,
organic-inorganic composite and nano materials, concrete, dry-mix mortar and oil well
cementing.
REFERENCES
1. Hirata, T.: Cement dispersant JP1984-18338 (1981), assigned to Nippon Shokubai.
2. Ebner, M.; Baumgartner, J.; Ohta, A.: Polycarboxylate based admixtures trend in
Europe, Concr. J. 42 (2), 2004.
3. Ramachandran, V. S., and Malhotra, V. M., Superplasticizers, Concrete Admixtures
Handbook, second edition, V. S. Ramachandran, ed., Noyes Publications, Saddle River,
NJ, 1996, pp. 410-517.
4. Plank, J., Current Developments on Concrete Admixtures in Europe, In: Proceedings
of the Symposium Chemical Admixtures in Concrete, Dalian/China, 2004, pp. 13-27.
5.. Plank, J.; Pllmann, K. Zouaoui, N.; Andres, P.R.; and Schaefer, C., Synthesis and
Performance of Methacrylic Ester-Based Polycarboxylate Superplasticizers Possessing
Hydroxy Therminated Poly(ethylene glycol) Side Chain: Cement and Concrete Research
(38) 2008, pp. 1210-1216.
6. Plank, J. PCE Superplasticizers Cemistry, Applications and Perspectives, 18.
ibausil, 12-15 Sep. 2012, Weimar, pp. 91 102.
10.1016/j.cemconcomp.2009.01.005
V. 336, No. 2, 2009, pp. 624-633. doi: 10.1016/j.jcis.2009.04.057
9. Aitcin, P.-C.; Jolicoeur, C.; and MacGregor, J. G., Superplasticizers: how they work
and why they occasionally dont, Concrete International, V. 16, No. 5, 1994, pp. 45-52.
10. Hirata, T.; Yuasa, T.; Shiote, K.; Nagare, K.; and Syogo, S., Cement dispersant,
method for producing polycarboxylic acid for cement dispersant and cement composition,
US Patent 6, 174, 980, 2001, assigned to Nippon Shokubai Co.
11. Crass, G. Polyglycols as Macromonomers for PCEs from an Industrial Viewpoint, Presentation at Opening Ceremony of TUM Center for Advanced PCE Studies, TU
Mnchen, March 2014.
12. European Norm DIN EN 1015-3 Prfverfahren fr Mrtel fr Mauerwerk Teil 3:
Bestimmung der Konsistenz von Frischmrtel (mit Ausbreittisch), DIN Deutsches Institut
fr Normung e. V.
SP-302-16
Evidences about the Interactions between

Grinding Aids and Cement Particles
Surface
by Valerio Antonio Patern, Sara Ottoboni, Marco
Goisis, and Paolo Gronchi
The measure of wettability of cement particles and the evaluation of the type and the strength
of active surface sites of clinker particles were carried out to score the efficacy of organic
grinding aids. The first analysis, based on the Washburn method, allows measuring the
contact angle of cement with different solvents and is directly related to the surface tension.
The second analysis, based on the Hammett method, helps to classify the sites that may be
responsible of the surface tension. Milled clinker (portland clinker CEM I 52, 5R; 0,025%
w/w DEG and TEA) was investigated. Firstly the wetting rates of powder with 4 different
solvents (ethanol, n-hexane, toluene, and formamide) were detected using a tensiometer.
Then the powder dispersions in a solvent were titrated by acid solutions to get information
on the acid/basic character of the surface sites. Techniques and results are shown.
Keywords: additive; cement; clinker; contact angle; grinding aid; surface tension;
wetting.
INTRODUCTION
Key engineering properties of concrete such as workability, reactivity, strength, and durability depend on the fineness of clinker.1 Unfortunately only a small amount of grinding
energy is used to create new surfaces from the broken products.2-4 As a result cement
grinding is a high energy intensive process consuming about 40% of all electrical energy
spent in a plant.5 Organic grinding aids (GA), such as glycols, amine or poly(alcohols)
are important tools to reduce energy consumption during mill operations. Indeed they
obstacle the re-agglomeration of the fresh ground particles so improving the yield of the
mill. However the knowledge of their action on the surface of cement particles is still
approximate. The scenario is coarsely described and comprehends firstly the formation
of some anisotropic energy distribution on the walls of the cleaved surfaces and then the
interaction of the organic molecules with the active sites. It is generally supposed that they
are rich of unsaturated valence forces which promote the interactions of the functional
groups of the chemical with the mineral surface.6 The need of choosing an efficient organic
211
212SP-302-16
molecule capable of maintaining separated the fresh cleaved surfaces after milling, push
to examine in depth the interfacial situation. The analysis, however, is challenging due
to the difficulty to monitor interfaces and to interpret available data. Many approaches,
including molecular simulation, can be used with the aim to describe type and origin of the
surface energy.7 Taking into account that similar characterization techniques are used in
workings on heterogeneous catalysis, in this paper preliminary results of a study of wettability of cement powders by liquids with different molecular structure, of measurement
of acid/basic strength and of density of active sites by acid/base titrations are reported.
Plain clinker milled powder and clinker ground with diethylene glycol (DEG) and triethanolamine (TEA) as GA have been used. The comparison is focused at understanding the
modification of the surface of cement particles w/wo grinding additives, and consequently
the role of chemicals to maintain particles separated during the grinding process.
RESEARCH SIGNIFANCE
This research is focused at improving the comprehension of the effect of grinding aids
on cement particles applying an original approach taken from the studies on heterogeneous
catalysis to this investigation.
Materials
P-toluenesulfonic acid, (PTSA; Alfa Aesar, 98%), chlorophorm (CHCl3; Sigma Aldrich,
99-99.4%), n-hexane (HEX; Alfa Aesar, 95%+), methanol (MeOH; Sigma Aldrich,
99.9%), ethanol (EtOH; Fluka, 99.8%), xylene (XYL; Alfa Aesar, 98.5%), toluene (TOL;
Alfa Aesar, 99%), N,N-dimethylformamide (DMF; Alfa Aesar, 99%), formamide (FA;
Alfa Aesar, 99%) benzene (BEN; Alfa Aesar, 99%) were used as received.
The plain milled clinker (CLKF) was prepared without gypsum. The chemical composition is reported in Table 1. The phase composition, according to the Rietveld method, is
shown in Table 2.
Methods
Clinker grindingClinker samples were ground by a Ball mill, Bond type, BICO International, rotating at 70 rounds/min, equipped with 285 grinding balls (tot. weight 20.125g)
of five sizes (1.5, 1.25, 1.0, 0.75, 0.625 inches) and charged with 1kg of clinker milled to
4000 Blaine fineness (blank sample name CLKF1) adding 0.025% (w/w) diethylene glycol
(DEG; sample name CLKF2) or triethanolamine (TEA; sample name CLKF3). CLKF1CLKF2- CLKF3 samples particles size distributions are shown in Fig. 1. Size distribution analysis was carried out by Helos KFS equipped with dry dispersion system Rodos.
Data elaboration by Fraunhofer model. Attention was paid to obtain a reproducible PSD
(Particle Size Distribution) in order to avoid different porosity of the bed of the column for
the Washburn tests.
Experimental set-upThe Washburn tests8 and Hammett9 measures were performed
using the usual laboratory apparatus and the instruments described below.
Data were calculated from at the least 10 times repeated tests for each sample to assure a
set of data for statistical evaluation. The reported values are the arithmetical medium value,
while the standard deviation is maintained around the 10%.
Evidences about the Interactions between Grinding Aids and Cement

Particles Surface 213
Table 1CLKF physical and chemical parameters (CLKF)*
CLKF
Oxide
SiO2
Al2O3
Fe2O3
CaO
MgO
SO3
Na2O
K 2O
SrO
Mn2O3
P 2O 5
TiO2
Total
Loss on ignition
Blaine fineness
%
21.31
5.03
3.70
64.55
1.57
1.29
0.33
1.28
0.07
0.01
0.20
0.13
99.88
0.44%
4000 cm2/g**
* Instrument: PANalytical Cubix XRF 200W.
Table 2Phase composition of CLKF according Rietveld method*

CLKF
Determination
C3S
C2S
Cubic C3A
C4AF
MgO
Free CaO
K3Na(SO4)2
K2SO4
Total
%
58.1
19.6
3.02
14.04
0.03
2.00
0.09
1.03
97.91
* Instrument: Bruker D8 Advance.
The Washburn model

The Washburn model (Eq. 1) was used to determine the powder contact angle6:

h2 =
r L cos
t (1)
2
where h2 is the clinker height wetted by the liquid, r is the capillary radius, L is the liquid
surface tension, is the liquid viscosity, the contact angle.
214SP-302-16
Fig. 1 Particle size distribution and cumulative distribution of milled Clinker (CLKF1-CLKF2-CLKF3). The
symbols and the lines of the density distribution are overlapped. (1 kg = 2.204 lb)
The model was modified and used in a suitable form (Eq. 2) for a powder bed. A linear
relation between the mass of the liquid in the porous bed and the time for rising up is
possible10:

m2 =
K 2 L cos
t (2)
where the geometrical factor K (Eq. 3) is function of the porous bed radius Rc, and the bed
porosity .

r R 2
c
K=
2
(3)
It has to be underlined that the linearity of the Eq.2 depends on some critical parameters
concerning the bed, as the packing factor, the void distribution and the capillaries radius,
the porous bed uniformity, the PSD These are important features which can be considered
as constant only after carefully and reiterated tests as done by us packing the samples by
centrifuge.
Liquid selectionThe choice of the appropriate liquid is essential to the application of the
thermodynamic model of van Oss-Chaudury and Good (OCG).10-13 Indeed it is fundamental
to know some basic parameters such as the contact angle of a non-polar liquid for the determination of the LV non-polar component and the contact angles of two polar liquids for the
determination of the acid (L+) and basic (L) components of the surface tensions. The chemical-physical characteristics of the selected liquids are shown in Table 3. The reference liquid

Table 3Chemical-physical proprieties of the liquid used in the Washburn
method
Parameter
(viscosity)
(density)
L (surface
tension)
LLW
(liquid-wall)
LA (A
component)
LB (B
component)
Units
N*s/m2 x103
(psf s)
Toluene
Hexane
Ethanol
0.604 (12.61)
0.326 (6.81)
1.07 (22.35)
Kg/m3 (lb/ft3) 866.9 (54.11) 654.8 (40.87)
789.1 (54.08)
Methanol
0.544
(11.36)
791.8
(49.42)
Formamide
3.3 (68.92)
1130.0
(70.53)
mN/m (lb/ft)
28.5 (1.95)
184 (12.61)
22.4 (1.54)
22.5 (1.54)
58 (3.98)
mN/m (lb/ft)
28.5 (1.95)
184 (12.61)
18.8 (1.29)
18.2 (1.25)
39 (2.97)
mN/m (lb/ft)
0.0191 (0.001)
0.6 (0.04)
2.83 (0.19)
mN/m (lb/ft)
68 (4.66)
77 (5.28)
39.6 (2.71)
Fig. 2Adsorption rate of different liquids on plain milled

clinker (CLKF1). (1 kg = 2.204 lb)
must completely wet the powder surface (so that =0 and cos =1) allowing to simplify
the Eq. 2. Consequently the geometrical factor K (Eq. 3) can be determined from the curve
slope m2/t provided that the other liquid properties (, LV, ) are known. Figure 2 shows
the absorption rate of several liquids and their maximum adsorbed amount.
In order to fulfil the above statement, the reference liquid must not present any polar
site. Moreover several liquids must be tested to calculate the polar acid (L+) and basic (L-)
components of the surface tension (LV).
From the data reported in Table 3, toluene was assumed as reference liquid because the
higher value of the product
216SP-302-16
m2
(from Eq. (2))
t 2
The choice agrees with literature.12

The experimental Washburn testsA known amount of powder equals to 2 g (4 x 10-3
lbs) was placed in a in a glass tube of 5 cm (0.16 ft) height and 1 cm (0.032 ft) diameter,
joined at the top to an arm of the tensiometer TE3 Lauda, and, at the other side, maintained
into contact with a liquid through a porous membrane. Uniformity of the porous bed of the
column was improved by packing with a centrifuge (see after).
A continuous measure of the column weight progressively increasing due to the permeating liquid, was executed. The properties of the test liquids are reported in Table 3.
Powder column packingA reproducible, uniform porous bed of powder was obtained
by centrifugation at 5000 rpm for 5 min.(Thermoscientific SL-16 centrifuge).
The Hammett method
The method enables to determine the basic strength expressed by an Hammett acidity
function and, hence, the acid-basic strength distribution of a solid surface on a common
scale.
To determine the strength of basic sites of the powder suspended in toluene, the colour
swing points of four indicators in the 1-8 pH range, possibly adsorbed on the surface previously treated with para-toluene sulfonic acid (PTSA) have been used. The acid was added
as first component to the powder repeating the operation each time with different amount in
such a manner to individuate the right addition for the clear colour swing. With the PTSA
(Brnsted acid) amount exceeding the saturation of the basic sites (SBS) of the surface,
the addition of the indicator gives a colour corresponding to acidity, while, with PTSA less
than the SBS, the indicator colour corresponds to basicity. As the goal of the test is to quantify the basic sites related to the PTSA salt at neutral condition, the values of the Hammett
parameter at different colour swing indicates the basic strength of the base conjugated to
its salt. The amount of acid on a solid is usually expressed as the number or mmol of acid
sites per unit weight or per unit surface area of the solid.
Relatively to the Brnsted acid-base equilibrium (Eq. 4)

B + H + BH + (4)
The Hammett and Deyrups acidity function Ho17 is:

H 0 = pKa + log
[B]
BH +
(5)
Where [B] and [BH+] are the concentrations of the neutral base and its conjugate acid
respectively, and

f
pKa = pK BH + = log aH + B
f BH +
(6)

Table 4Geometrical coefficient and contact angle. (m =1 kg = 2.204 lb)
Sample
Liquid
Toluene
Ethanol
Hexane
FA
CLKF1
CLKF2
CLKF3
K
(geom. coeff.) (contact angle) (geom. coeff.) (contact angle) (geom. coeff.) (contact angle)
(m5) x 1016
()
(m5) x 1016
()
(m5) x 1016
()
1.45
Reference
1.42
Reference
1.50
Reference
1.36
20
1.25
28.2
1.32
28.3
1.45
84.3
0.85
84.4
1.71
83.5
1.28
27.5*
1.11
38.4
1.19
37.4
If Lewis equilibrium takes place by means of electron pair transfer from the adsorbate to
the surface, Ho is expressed by

H 0 = pKa + log
[ B ] (7)
[ AB ]
Where [AB] is the concentration of the neutral base which reacts with the acid or electron
pair acceptor, A.
During titration, stronger basic sites are neutralized earlier, while weaker ones later and
the latter require stronger acids for neutralization. Therefore, it can be assumed that the
weakest basic sites have been finally neutralized by an acid having an acid strength for
which Ho = pKBH+.
The experimental Hammett tests0.2 g (4 x 10-4 lbs) of milled clinker was dispersed in
10 ml of toluene and a known amount of para-toluenesulfonic acid (PTSA) was added. In
order to clearly evaluate the swing point, the addition of PTSA was repeated up to reaching
the threshold value. Power sonication (40 KHz maximum) was carried out for 30 minutes
to assure full dispersion of the powder; 2 ml (0.067 oz) of indicator were added later.
The 1-8 pH range was covered by the following indicators: thymol blue (1.2-2.8 pH
range), methyl yellow (3.3-4.0 pH range), methyl red (4.4-6.2 pH range) and neutral red
(6.8-8.0 pH range).
DISCUSSION
Surface tension
The results of the Washburn tests are reported in Table 4 and the curves representing
mass of liquid (m2) vs. time (t) are presented in Figure 3 for the polar liquids, FA and EtOH.
The same relationship is shown in Figure 4 for the non-polar liquid, TOL and HEX.
All data have been collected at the same times and up to the equilibrium state.
Contact angle The contact angles of polar and non-polar liquids are very different:
polar liquids, EtOH and FA, applied on samples treated with DEG (CLKF2) and TEA
(CLKF3), give higher values respect to those obtained with the not treated powder, the
increasing percentage being about 40% with both the liquids. On the contrary, the nonpolar liquid (HEX) gives the same value with non-treated and treated powder. The lower
contact angles observed with the polar liquid respect to HEX can be explained in terms of
remarkable interaction with the powder surface able to increase the spreading component
of the surface tension (SV).
218SP-302-16
Fig. 3(a)(b)Washburn plots for polar liquid with different molecular structure. (1 kg =
2.204 lb)
Fig. 4(c)(d)Washburn plots for non-polar liquids, alkane and aromatic structured. (1 kg
= 2.204 lb)
Rate of liquid rising and amount of adsorbed liquidConsidering the apparent curve
slope, that is proportional to the rate of liquid moving up (m2/t; Eq. 2), significant differences are present between the polar and the non-polar liquids for the samples w/wo GA
(Figures 3 and 4).
Table 5 shows the relevant slopes obtained by graphical interpolation of the m2/t
linear relationship.
The gradient values decrease according to the sequence TOL>FA>HEX>EtOH while
for the same liquid, we observe a CLKF3>CLKF1>CLKF2 sequence. From the liquid
sequence we observe that the rising rate can be ascribed to the molecular structure, and
consequently the more pronounced ability of EtOH to form hydrogen bond, could be
responsible of the lowest rate. An increase of the interface tension SL slows the spreading

Table 5m2/t relationship (g2/s x 106; 1g = 2.204 x 10-3 lb)
CLKF1
CLKF2
TOL
5.13
3.72
CLKF3
5.4
FA
5.03
3.42
4.5 initial
2.7 after 30s
HEX
3.55
3.05
EtOH
1.77
1.11
4.42
1.79
rate as appear from the balance condition (SV> LVcos+SV).The literature agrees with
the lowest rising rate with EtOH as confirmed by studies of the adsorption on montmorillonite14 which claims that the observed strong adsorption is due to hydrogen bonding of the
alcohol groups to the oxygen atoms of the surface.15,16
Furthermore EtOH has about the same rate with treated and non-treated powder (Tab. 5).
Only the point CLKF3-FA does not clearly agree with a continuous decreasing sequence.
However in this case the curve m2/t has two linear segments with different gradients:
initially the gradient is higher (4.5x10-6 g2/s) and after 30 s it becomes lower (2.7x10-6 g2/s).
As far as concerns the amount of liquid captured by the powder at the equilibrium (from
fig. 3 and 4, plateaus) and stored in the interparticle spaces, the amount of FA in plain
CLKF1 bed is greater (not determined) than in CLKF3 and in the CLKF2 (Fig. 3b). With
EtOH much smaller differences of the same parameter are observed (Fig. 3a). From Fig. 4
clearly appears that TOL is more adsorbed on the powders than HEX.
The interpretation of the data needs a look at the clinker particle surface-liquid interaction. The conformations of TEA adsorbed on C3S were calculated by R.T. Mishra, et al.17
TEA bonds the silicate surface through hydrogen bonds (HB) and through the oxygen
coordination of the hydroxyl groups with the Ca++ ions that are diffused all over the clinker
surface. Due to the ratio of the hydroxyl groups, it is allowed to suppose that TEA forms
more hydrogen bonds (HB) with silicate (probably in the present case the C3S phase) than
DEG, that, on the contrary, easier coordinates the Ca++ ions in reason of the greater basicity
of its electron pairs on the oxygen of the ether group. Moreover DEG is known to have only
one OH link because the most probable molecular conformation has intramolecular HB of
the second hydroxyl group with the ether central oxygen (Figure 5).
Other bi-dentate (two HB bonds or one HB bond and one coordination bond) and tridentate (two HB bonds and one coordination bond) conformations of adsorbed DEG might
be hypothesized but they appear less probable due to the great chain flexibility.
The rising liquids can interact both with the adsorbed GA and the free surface. Clear
indications may be acquired by the observed polar liquid behaviours considering that FA
acts mainly by the carbonyl dipole creating coordinative bonds, while EtOH links the
surface exclusively with the HB bonds. Similarly both the molecules link a Lewis base at
the sites of the surface.
Due to the great rate of FA liquid on CLKF1 (Tab.5) a prevalence of the coordinative
bonds on strong HB bonds is then supposed on plain clinker. The amount of FA stored
inside the spaces of the powder column seems to confirm this hypothesis.
Now, comparing the FA rising rate and the amounts between TEA and DEG treated
samples, TEA seems able to maintain more sites for coordinative bonds than DEG as both
the values are higher. The observation agrees with the TOL adsorbing behaviour. TOL has
polarizable electrons, the aromatic electron cloud, that is deformed by methyl group induc-
220SP-302-16
Fig. 5Monodentate HB and coordination of DEG.

tive effect. The dipole promotes coordinative bonds causing rising rates greater than with
FA. Similarly the behaviours with EtOH can be interpreted with the formation of strong
HB bonds.
The previously advanced hypotheses confirm that different types of active sites exist on
the surface coordination bonds and second order HB bonds.
The hypothesis agrees also with the chemical and physical data of the liquids: FA has
the highest surface tension among those listed in Table 1 that derives from its greatest LLW
among the used liquids, the greatest acid LA, and the moderate basic component of surface
tension, LB.
In a separate experiment, the clinker CLKF1 was treated with vapour of DEG obtained
from liquid DEG heated at 70C for 92 h and transported to clinker by nitrogen carrier in
order to completely saturate the powder surface.
Examining the Figure 6, the comparison between the 0.025% DEG treated clinker
(CLKF2) and the clinker that has been completely coated by DEG, shows that EtOH is
adsorbed faster on the low DEG (EtOH-CLKF2) treated clinker. The plateau is not reached
by the ETOH-CLKF1+DEG Ads sample, probably due to the different interaction at the
interface (solid-vapour instead of solid-liquid) which may cause a different pore filling.
Inversely, using FA, the absorption rate is high on the completely saturated sample and
no remarkable difference of the amount of the adsorbed liquid is present. The observation agrees with the greater tendency of the DEG coated powder to form HB bonds rather

Fig. 6Comparison between the completely coated surface

and the surface obtained with 0,025% of GA (1 kg = 2.204 lb)
than coordination bonds so decreasing the absorption rate of EtOH and, on the contrary,
increasing that of FA.
Investigation on surface tension originThe results of the Washburn tests were evaluated with the Van Oss thermodynamic theory that, starting from the elaboration of the acidbasic interaction proposed by Fowkes10,15 elucidates the relevance of the electron-acceptor
(Lewis acid) and the electron-donor (Lewis base) sites and their different behaviours. The
free energy at the interface between the solid and the liquid phase is represented by the Eq.
8 where the total free energy is the sum of the non-polar interaction LW (Lifshitz-Van der
Waals) and the polar interaction of acid and base Lewis sites (AB):

LW
GSL = GSL
+ GSLAB (8)
Fowkes demonstrated that in case of non-polar interactions (Eq. 9):

LW
LW
GSL
= 2 LW
(9)
L S
After the revision of vanOss, the free energy for polar interaction is described by the relation (Eq. 10)

GSLAB = 2 +L S 2 L +S (10)
where + is the acid component and - is the basic component of the interfacial polar tension
and A and B indicate the Acid and Basic site. The relation of the total free energy (Eq. 8)
becomes:

LW
GSL = 2 LW
2 +L S 2 L +S (11)
L S
222SP-302-16
Table 6Components of the surface tension S (solid-liquid interface;

mN/m)
Sample
CLKF1
CLKF2
CLKF3
Hex
S LW
5.56
5.54
5.71
FA
S28.49
54.07
58.99
EtOH
S+
5.35
4.98
4.93
total
30.25
38.37
39.80
That can be transformed into Eq. 8 (the Oss-Chaudary-Good equation) using the YoungDupr equation (L(1 + cos) = GSL)12:

L (1 + cos ) = 2
LLW SLW + +L S + L +S (12)
Using the Eq. 8 all the components of the surface tension of the solid, LW, + and -, can be
measured. Three liquids with different surface tension were put in contact with the powder
for the determinations of the solid surface tension. From the contact angles, applying the
OCG theory, the powder surface tension can be calculated. In this study, HEX, FA and
EtOH have been used for the SLW, S-, and acid component S+ respectively. The results are
listed in Table 6.
The data indicate that the basic (polar S-) component of the samples CLKF2 and CLKF3
milled with DEG or TEA respectively, increases respect to the not treated sample CLKF1
following the order CLKF3>CLKF2CLKF1. The non-polar component S LW does not
change sensibly.
Acid Base Strength Distribution
The basicity (mmol/g) at the Ho value (basic site strength) is stated by the number of
basic sites whose basic strength is equal to or greater than the Ho value.19 The basic sites
(mmol/g) at Ho strengths are reported in Table 7.
Firstly it appears an increase of the basic strength when DEG or TEA (GA) is used
respect to the plain milled clinker. The values of the TEA treated samples are however
lower than those of the DEG treated ones.
Remembering that a high Ho=pKa (or pKBH+) value is related to a high basic population,
the greater basicity value in presence of the GA additive may be interpreted as consequence
of the direct reaction of PTSA with the additives.
The proton-donating ability of the solid at the endpoint of titration is considered to be
either due to the conjugate acids which were formed by the proton transfer from Brnsted
acid solution to the solid surface or to the Brnsted acid which was physically adsorbed on
the surface during the titration.
Moreover, as previously reported, the conformations of TEA (and DEG similarly)
adsorbed on the basic sites of the C3S surface,17 indicate the formation of HB between
the OHs of CH2-CH2-OH groups and the oxygen of the silicate together or alternatively
between the coordination bond of the oxygen of the hydroxyl group and the Ca++ of the
surface .

Table 7Basic site strength HO
CLKF1
CLKF2
CLKF3
H0 2.8
20
38
32
Basic sites (mmol/g) at Ho strengths

H0 3.3
H0 4.8
40
0*
46
0*
34
0*
H0 6.8
0*
0*
0*
* H0 5x10e-5
However these types of bonds are weak and are broken easily by acids. The HB equilibrium and the coordination equilibrium

GA-O-H + B GA-OH---B HB bond
(13)
GA-HO + Ca++ GA-HO--- Ca++ coordination bond
(14)
are shifted to the left by the acid addition as a consequence of the acid character of the
hydroxyl group present in the structure of the GA and of the basic strength of the oxygen
in the silicate. A stronger HB means a greater amount of PTSA.
The different behaviour between TEA and DEG might be attributed to their molecular
structure that causes some greater inductive effect on the electron donor/acceptor character of the OH group and to the steric hindrance. The structure is supposed to influence
directly the shift of the equilibria towards left and to increase the number of the basic sites.
FURTHER RESEARCH
Our aim is to extend the examination to many grinding additives to obtain a large number
of data on the interaction between the organic molecules and the clinker surface. The field
of the study is of interest for many applications relating the behaviour of material surface
at nano and micro dimension particularly in the cement sector. The molecular conformation analysis and the analysis of the kind and strength of surface active sites by chemical
and instrumental methods, might give useful information to choice the right GA additive.
CONCLUSION
In this work several methods already present in literature have been employed to study
the surface of milled clinker powder. Samples were obtained by grinding clinker with additives largely used in the industrial practice, such as DEG and TEA, at dosage equal to
0,025% w/w, and the results compared with those of the plain powder. The study is part
of a larger investigation aimed at understanding the mechanism of action of grinding aids
to design new molecules. The methodological approach performed in this study allows
to calculate the total surface tension and its basic/acid components by the measure of the
contact angles and the rate of wetting of the powder. Furthermore the approach allows to
get information on the sites present on the surface. These indications are obtained by using
polar (FA, EtOH) and non-polar (HEX, TOL) solvents, according to the Washburn method,
as the liquids differently interact with the surface, i.e. by H atoms and/or by dipoles/WdW
bonds and/or only by one of these. The results agree with the existence of two types of sites
224SP-302-16
mainly, suitable for hydrogen bond and coordinative bond (dipole-induced dipole) respectively. The elaboration of the Washburn theory by a thermodynamic point of view permits
to highlight the basicity of the sites according to the predominance of the basic component on the total surface tension. It increases in the order TEA>DEG>plain clinker. The
Hammett method puts in evidence that the basic sites suitable for the HB bond (Lewis basic
site) increase when the clinker is milled with DEG or TEA, respect to the plain clinker. In
this framework DEG seems responsible of a greater population of basic sites than TEA.
The explanation is not straight and relies on the hypothesis that grinding aids increase the
amount of basic sites suitable for the HB bond. From these data it could be argued that the
GAs cause an unforeseen and surprising increase of surface tension pushing to a deeper
investigation in order to find possible different origins of their activity. It should be the case
of adsorption energies,17 some shielding effect, or creation of an electric field around the
surface of the particle.
AUTHOR BIOS
Valerio Antonio Patern graduated in Materials Science Engineering at Politecnico di
Milano, (Italy) in the field of hybrid materials and admixtures for cement and concrete.
Sara Ottoboni graduated in Materials Science Engineering at Politecnico di Milano
(Italy), where later occupied a post-degree position. She is focusing her study on Science
and Engineering of surface of materials.
Marco Goisis, graduated (MS) in Chemical Engineering at Politecnico di Milano, is
specialist in additives for grinding cement, admixtures for concrete and cementitious
products at Innovation Department, Italcementi. Since 1989 he has been working with
Italcementi Group companies, in R&D and in the Innovation Departments. His current
research interest includes carbon allotropes modified smart construction materials.
Paolo Gronchi is associate professor at the Chemical, Materials and Engineering Chemistry Dept. of the Politecnico di Milano, Italy. His current research activity, beyond the
industrial chemical research, concerns the structural investigation on polymer adsorbed
on inorganic oxide and comprehends the synthesis of organic compounds.
ACKNOWLEDGMENTS
The authors acknowledge Italcementi Group for the financial and scientific support.
REFERENCES
1. Bentz, D. P.; Garboczi, E. J.; Haecker, C. J.; and Jensen, O. M., Effects of Cement
Particle Size Distribution on Performance Properties of Portland Cement-Based Materials, Cement and Concrete Research, V. 29, No. 10, 1999, pp. 1663-1671. doi: 10.1016/
S0008-8846(99)00163-5
2. Schlanz, J. W., Grinding: An Overview of Operation and Design, (1987) (http://
mrl.ies.ncsu.edu/reports/ 87-31-P_Grinding_ Operations_Design.pdf), 27 February 2015.

3. Worrell, E.; Martin, N.; and Price, L., potential for energy efficiency improvement
in the US cement industry, Energy, V. 25, No. 12, 2000, pp. 1189-1214. doi: 10.1016/
S0360-5442(00)00042-6
4. Madlool, N. A.; Saidur, R.; Hossain, M. S.; and Rahim, N. A., A critical review
on energy use and savings in the cement industries, Renewable & Sustainable Energy
Reviews, V. 15, No. 4, 2011, pp. 2042-2060. doi: 10.1016/j.rser.2011.01.005
5. Schneider, M.; Romer, M.; Tschudin, M.; and Bolio, H., Sustainable, Cement
Production - Present and Future, Cement and Concrete Research, V. 41, No. 7, 2011, pp.
642-650. doi: 10.1016/j.cemconres.2011.03.019
6. Sohoni, S.; Sridhar, R.; and Mandal, G., The Effect of Grinding Aids, on the Fine
Grinding of Limestone, Quartz and Portland Cement, Clinker, Powder Technology, V.
67, No. 3, 1991, pp. 277-286. doi: 10.1016/0032-5910(91)80109-V
7. Mishra, R. T.; Flatt, J.; and Heinz, H., Force field for tricalcium silicate and insight
into nanoscale properties: clevage, initial hydration, and adsorption of organic molecules,
The Journal of Physical Chemistry C, V. 117, No. 20, 2013, pp. 10417-10432. doi: 10.1021/
jp312815g
8. Washburn, E. W., The dynamics of capillaryX, Journal of the American Physical
Society, 2nd Ser. 17, 1921, pp.374-375.
9. Hammett, L. P., and Deyrup, A. J., A series of simple basic indicators. I. The acidity
functions of mixtures of sulfuric and perchloric acids with water, Journal of the American
Chemical Society, V. 54, No. 7, 1932, pp. 2721-2739. doi: 10.1021/ja01346a015
10. vanOss, C. J.; Chaudhury, M. K.; and Good, R. J., Monopolar surfaces, Advances in
Colloid and Interface Science, V. 28, 1987, pp. 35-64. doi: 10.1016/0001-8686(87)80008-8
11. Ahadian, S.; Mohseni, M.; and Moradian, S., Ranking proposed models for
attaining surface free energy of powders using contact angle measurements, International Journal of Adhesion and Adhesives, V. 29, No. 4, 2009, pp. 458-469. doi: 10.1016/j.
ijadhadh.2008.09.004
12. Dang-Vu, T., and Hupka, J., Characterization of porous materials by capillary rise
method, Physicochemical Problems of Mineral Processing, V. 39, 2005, pp. 47-65.
13. Van Oss, C. J.; Giese, R. F.; Li, Z.; Murphy, K.; Norris, J.; Chaudhury, M. K.; and
Good, R. J., Determination of contact angles and pore sizes of porous media by column
and thin layer wicking, Journal of Adhesion Science and Technology, V. 6, No. 4, 1992,
pp. 413-428. doi: 10.1163/156856192X00755
14. Emerson, W. W., and Raupach, M., The reaction of polyvinyl alchohol with montmorillonite, Australian Journal of Soil Research, V. 2, No. 1, 1964, pp. 46-55. doi:
10.1071/SR9640046
15. Nguyen, T. T.; Raupach, I. M.; and Janik Csiro, L. J., Fourier-transform infrared
study of ethylene glycol monoethyl ether adsorbed on montmorillonite: implications for
surface area measurements of clays, Clays and Clay Minerals, V. 35, No. 1, 1987, pp.
60-67. doi: 10.1346/CCMN.1987.0350108
16. Parfitt, R. L., and Greenland, D. J., The adsorption of poly(ethylene glycols)
on clay minerals, Clay Minerals, V. 8, No. 3, 1970, pp. 305-315. doi: 10.1180/
claymin.1970.008.3.08
17. Mishra, R. K., and Heinz, H., Zimmermann., Muller, T., Flatt, R., J., Understanding
the effectiveness of polycarboxylate as grindings aids, Proceedings of Tenth CANMET/
226SP-302-16
ACI International Symposium on Superplasticizer and Other Chemical Admixtures in

Concrete, SP-288.16, V. M. Malhotra, ed. American Concrete Institute, Farmington Hills,
MI, 2012, pp. 235-249.
18. Fowkes, F. M., Additivity of intermolecular forces at interfaces. I. Determination of the contribution to surface and interfacial tensions of dispersion forces in various
liquids, Journal of Physical Chemistry, V. 67, No. 12, 1963, pp. 2538-2541. doi: 10.1021/
j100806a008
19. Wang, K.; Wang, X.; and Li, G., Quantitatively study acid strength distribution on
nanoscale ZSM-5, a, Microporous and Mesoporous Materials, V. 94, No. 1-3, 2006, pp.
325-329. doi: 10.1016/j.micromeso.2006.03.049
SP-302-17
The Influence of C3A on the Dissolution

Kinetics of Alite/Gypsum Mixtures in the
Presence of PCEs
by Giorgio Ferrari, Vincenzo Russo, Massimo Dragoni,
Gilberto Artioli, Maria Chiara Dalconi, Michele Secco,
Leonardo Tamborrino, and Luca Valentini
Portland cement is a multi-phase material, which can be simplified as a two-phase system
with alite and C3A as the main constituents determining early properties. Alite is the
most abundant phase and in a first approximation it is responsible for the development of
mechanical strength during hydration, while C3A mainly affects the plastic behavior before
set. In portland cement, superplasticizers are preferentially adsorbed onto C3A and its
hydrates rather than alite, due to the different interaction with the mineral surfaces. Three
different polycarboxylate superplasticizers (PCEs) were studied, based on copolymers
of methacrylic acid and MPEG-methacrylate and characterized by different side chain
length and different charge density. Their affinity to C3A and alite surfaces was determined
through adsorption measurements on alite/gypsum and alite/gypsum/C3A mixtures. The
results of the adsorption tests indicated that the charge density of PCEs, expressed as
the ratio between carboxylic groups to ester groups, is the main parameter affecting the
adsorption of the PCEs: the lower the charge density, the lower the adsorption on both the
phases. The same parameter affects the induction period of alite phase, as demonstrated
by in situ XRPD dissolution kinetics experiments, both in the presence and in the absence
of C3A. These results can be put in relation with both the hindrance of adsorbed PCE
molecules in the dissolution kinetic of alite and the concentration of PCE molecules in
solution in conditions of saturation.
KEYWORDS: alite; adsorption; C3A; comb copolymers; gel permeation chromatography; hydration; induction period; PCE superplasticizers; X-ray powder diffraction.
INTRODUCTION
Water reducing admixtures and superplasticizers (high-range water reducing agents) are
nowadays extensively used to control rheology and to improve performances of normal
and high performance portland cement-based concrete mixtures1,2. After naphthalene
227
228SP-302-17
and melamine sulfonate based superplasticizers, poly-carboxylate ether (PCEs) became

increasingly popular for their high efficiency in dispersing cement particles at low dosages.
The molecular structure of such polymers basically consists in a polymer chain bearing
anionic groups (backbone) with attached hydrophilic polyoxyethylene based (EO) side
chains. This structure can be easily modified, by properly operating on the reaction conditions (type and ratio of monomers, length of side chains, etc.) and targeted to efficiently
disperse the clinker and hydrating cement particles3-6 in order to reduce W/C, yield stress,
viscosity and, ultimately, to improve final strength and durability of concrete mixtures.
However, the addition of PCE superplasticizers invariably produces a reduction of the
hydration rate of the alite phase, inducing a retardation of the hydration process. The
reasons for the prolonged induction period induced by PCE has been variously attributed
to (i) a reduced diffusion of water and calcium ions at the clinker particle surface caused
by the adsorbed polymer, (ii) the calcium-polymer interaction in solution or (iii) the influence of the adsorbed polymers on the nucleation and growth of the hydration products.7-9
Although it is generally assumed that the Ca-PCE interactions are weak and therefore the
calcium complexation should not largely affect the activity of the calcium ions in solution,8,9 it has been reported10 that, at dosages of PCEs above the saturation dosage, the
nucleation and growth of C-S-H could be significantly affected by the PCEs molecules in
solution, causing a retardation on C3S dissolution rate.
Extensive research work has been performed in the attempt to characterize the interaction of the PCE dispersants with C3S, which is the main constituent of the clinker.11,12
All experimental evidence indicate that adsorption of the PCE polymers on the crystal
surface has a critical inhibiting effect on the dissolution process of C3S (and alite as well).
A number of investigations emphasize the correlation between the polymer molecular
structure (charge density, ratio carboxylate/PEO groups, backbone and side-chains length)
on the delay time, although the exact mechanisms still remain unclear.9,12,13 The observed
drastic reduction in C3S dissolution induced by the polymer adsorption is increased at high
concentrations of Ca(OH)2 in solution12 and the nucleation mechanism of C-S-H and the
activation energy required for the nucleation and growth process are markedly changed
in presence of superplasticizers.9 Some speculation has also been put forward concerning
the possible interaction of the superplasticizer macromolecules and the hydration products, with the formation of intercalate phases, though the issue is still debated.14-16 The
retardation in the dissolution of C3S is substantially modified in cement pastes because of
the combined effect of C3A and calcium sulphates.17 In fact, it is known that superplasticizers are preferentially adsorbed on the surface of the aluminate phases compared to
the silicate phases.18 In C3S-C3A-sulphate mixtures therefore a complex competition takes
place between sulphate availability and adsorption of the polymers on the different clinker
phases. The uptake of PCE by C3A and its hydration products, which potentially encompasses the formation of intercalate phases,19 sensibly reduces the retardation of the C3S
hydration reaction and in extreme cases inhibits the dispersing effect of the plasticizers.20,21
Industrial research on new superplasticizers aims to develop new polymers capable to
impart improved characteristics and performance to concrete mixtures, such as higher
reduction of W/C, workability, workability retention and strength development. In order
The Influence of C3A on the Dissolution Kinetics of Alite/Gypsum Mixtures

in the Presence of PCEs 229
Table 1. Structural characteristics of the different PCEs.
Sample
[P-(N-1)]
PCE 17-2
PCE 17-3
PCE 113-9
Mw
36112
27892
75241
Mn
20148
14060
43335
Mz
62991
50341
135753
N; Rc
2.9
3.9
10.5
P
17
17
113
n
20.3
13
17.3
Degree of
polymn.
58
50
77
where:

Mw = Weight average molecular weight
Mn = Number average molecular weight
Mz = Z-average molecular weight
P = Repeating EO units in side chain
n = Number of segment per macromolecule
N = Number of units per segment
N-1= Number of acidic groups per segment
Rc = Acidic groups/Ester groups
to attain these results, it is very important to study the interaction of superplasticizers with
the main mineral phases of portland cement (alite, C3A) and to clarify their mechanism
of action. As a further step in the understanding of the hydration process of cement in
the presence of PCE superplasticizers and as a possible test of the proposed models, this
study presents the results of in situ XRPD measurements of the hydration reactions and
bulk absorption experiments using different alite/gypsum/C3A/ mixtures and PCE superplasticizers with different molecular structure. In this perspective, this article represents a
contribution for a better comprehension of this complex and exciting matter.
In this section, the characteristics of the different materials (both PCEs and mineral
phases), the compositions of the model cement pastes and the analytical techniques and
procedures for the different tests, are described.
Polycarboxylate superplasticizers (PCEs)
Three different PCEs based on random copolymers of methacrylic acid and polyoxyethylene methacrylate were synthesized in the laboratory, starting from methacrylic acid and
methoxypolyethyleneglycol-methacrylate with different MPEG chain length.
Molecular weight distribution analysis (MWD) was carried out with the Gel Permeation
Chromatography (GPC) apparatus consisting of a Waters 515 pump (1ml/min, 0.06 in3/
min) equipped with a Waters 717 Plus auto-sampler. Separation was made by a set of four
Waters Ultrahydrogel columns (7.8x3000mm, pore size 2x120, 250, 500 ). Detection
was performed by a Waters 410 Differential Refractometer detector (temperature 40C,
104 F). All data were evaluated using Waters Empower Pro software. All samples were
dissolved in the mobile phase (NaNO3 0.04M 80% - Acetonitrile 20%) at 1% concentration. Molecular weight distribution of polymers was determined by using a calibration
curve done with standards polysaccharides (Polymer Laboratories LTD).
The charge density of the different PCEs, expressed as the ratio of carboxylic groups/
ester groups (Rc = CG/EG) in the backbone, was determined by potentiometric titration,
230SP-302-17
Table 2. Composition of model cement pastes

Sample
alite/gyp
alite/gyp/113-9
alite/gyp/17-3
alite/gyp/17-2
alite/5C3A/gyp
alite/5C3A/gyp/113-9
alite/5C3A/gyp/17-3
alite/5C3A/gyp/17-2
alite/10C3A/gyp
Alite
(wt%)
Gypsum
C3A (wt%) (wt%)
Water/solid
W/S
95
0.5
90
0.5
85
10
0.5
Superplasticizer
Dosage
Type
(wt% dry)
None
PCE 113-9
0.15
PCE 17-3
0.15
PCE 17-2
0.15
None
PCE 113-9
0.15
PCE 17-3
0.15
PCE 17-2
0.15
None
PCE 113-9
0.15
PCE 17-3
0.15
PCE 17-2
0.15
carried out with an acid/base titration with HCl 0.1N performed in alkali media with automatic titration system Mettler Toledo T70 with Rondo20 Auto-sampler.
The different PCEs were classified according the general parameters defined by Gay and
Raphal22 for comb polymers in solution. According to this classification, a random comb
copolymer can be represented as the repetition of n segments, each segment made by N
repeating unit and containing one side chain of P repeating units. The characteristics of
PCE comb copolymers used in the present work are shown in Table 1.
Mineral phases and model cement paste composition
Alite (C3S monoclinic polymorph), C3A (cubic polymorph), supplied by MR PRO,
Meyzieu, France and reagent-grade gypsum were used to prepare the model cement pastes.
Specific surface area of alite and C3A was measured by BET analysis with a Coulter instrument mod. 3100SA. Measurements were performed by volumetric He adsorption at low
temperature (liquid nitrogen), after dehydration of the samples at 100 C under vacuum.
BET specific surface area was 1.292 m2/g (51.68 yr2/ounce) for alite and 1.812 m2/g (72.48
yr2/ounce) for C3A. The mean particle size, measured by means of laser diffraction, using
a Malvern Mastersizer 2000, was 16 m (6.310-4 in), 13 m (5.110-4 in) and 70 m
(27.610-4 in), respectively, for C3S, C3A and gypsum.
Model cements were prepared by mixing alite, C3A and gypsum in different proportions.
The proportion of C3A was varied from 0 to 10 per cent on the whole composition mass
and the amount of alite was correspondingly reduced. The amount of gypsum was fixed at
5 per cent of the whole mass, in order to maintain all the mixtures in properly sulphated
conditions.23 The compositions of the model cement pastes are shown in Table 2.
X-Ray Powder Diffraction measurements (XRPD)
The alite/C3A/gypsum dry mixtures for XRPD diffraction were prepared by gently
co-grounding the powders by hand in an agate mortar for 15 minutes. A saturated calcium
hydroxide solution (lime water) containing the selected amount of the different PCEs, was
added to the dry mixtures at a water to solid ratio W/S = 0.5 and then the pastes were

mixed by hand and an orbital mixer for 2 minutes. It has been reported that the use of a
calcium hydroxide-saturated solution induces a decrease in the initial rate of dissolution
of C3S.24 Such slower hydration kinetics allowed a better discrimination of the duration of
the induction period for the different systems. Boron-glass capillaries with 0.5 mm internal
diameter were used as sample holder for in situ XRPD measurements. The capillaries filled
with paste were sealed with wax in order to prevent carbonation and water loss during
the measurments. In situ XRPD data were collected using a PANalytical XPert Pro MPD
diffractometer equipped with a PixCel detector and operating in transmission focusing
geometry with a CuK radiation. Incident and diffracted beam optics included an elliptical
focusing mirror, 0.04 rad Soller slits, divergence and antiscatter slits of aperture. The
hydrating cement pastes were measured continuously for 24 hours collecting diffraction
patterns (6-66 2 interval) of 20 minutes duration. During measurements, the temperature inside the diffractometer case was maintained at 23.0C 0.5C (73.4 F). Each
diffraction pattern was analyzed using the Rietveld method as implemented in Topas v4.1.
Adsorption measurements
30 grams (1.2 ounce) of the model cements, without preliminary co-grinding, were
added to 15 grams (0.6 ounce) of saturated calcium hydroxide solution (lime water) (W/S
= 0.5) containing the selected amount of the different PCEs and mixed for 5 minutes. The
compositions of the mixtures were the same as those of Table 2 for XRPD measurements,
but multiple dosages of the different PCEs were used: 0.05, 0.10, 0.15, 0.20 and 0.30 per
cent of dry PCE by weight of powder. The pastes were then filtered and the filtrates were
neutralized with diluted HCl and analysed by GPC. The amount of adsorbed polymer was
calculated by comparing the areas of the chromatograms of the different PCE polymers
before and after mixing, according to a previously described method.25 Different from the
method based on TOC measurements, which measures the total amount of organic carbon
in the solution after filtering or centrifuging, including unadsorbed polyglycol residues and
unreacted monomers, by this method it is possible to distinguish all the different constituents and to determine only the effectively adsorbed polymer fraction.
In this paper, adsorption data were referred only to the polymer fraction, not considering
the other unadsorbed constituents.
In this section, the results of XRPD and adsorption tests are presented and discussed.
These results are further treated in order to investigate the mechanism of retardation of alite
phase dissolution induced by PCEs at different dosages and in the presence of different
amount of C3A.
X-Ray Powder Diffraction tests (XRPD)
The results of XRPD measurements on the dissolution of alite in the presence of different
PCEs and different alite/gypsum/C3A compositions, are shown in Figures 1A, 1B and 1C.
Each curve represents the relative intensity of the alite diffraction peaks (Rietveld scale
factor) as a function of the hydration time. The sharp decrease of the intensity during hydration corresponds to the end of the induction period and the acceleration of the dissolution
of alite to form C-S-H and CH (hydration). From Figure 1A (no C3A added), it is possible
232SP-302-17
Figure 1. (A) Dissolution curves of alite in the presence of different PCEs (dos. 0.15% of dry polymer
by weight) for mixtures with 0% C3A. (B) Dissolution curves of alite in the presence of different
PCEs (dos. 0.15% of dry polymer by weight) for
mixtures with 5% C3A. (C) Dissolution curves of
alite in the presence of different PCEs (dos. 0.15%
of dry polymer by weight) for mixtures with 10%
C3A.
Table 3. Retardation on alite dissolution induced by different PCEs
(dos.0.15% by weight of dry polymer) in the presence of different amount of
C3A.
Type
None
113-9
17-3
17-2
Type of PCE
Dos. %wt
0.15
0.15
0.15
0% C3A
260
540
980
1280
End of induction period (minutes)

5% C3A
10% C3A
260
260
380
315
640
455
840
495

Figure 2. Curves of precipitation of portlandite in the presence of different PCEs (dos. 0.15% of dry polymer by weight)
for mixtures with 0% C3A.
to recognize that the different PCEs induced different periods of retardation in the alite
dissolution. Particularly, at the same dosage of 0.15% of dry polymer by weight of powder,
PCE 17-2 showed the highest retarding effect, followed by PCE 17-3 and PCE 113-9, the
last showing the lowest retarding effect. The same order of retardation was observed in the
presence of 5% and 10% of C3A (Figures 1B and 1C, respectively), but with a reduction of
the induction period for all the mixtures. The end of the induction period for the different
mixtures containing different PCEs and different amount of C3A, as determined by the
XRPD curves, is reported in Table 3.
These results confirmed that all the polymers induced a retardation in the dissolution of
alite. Particularly, PCE 17-2, characterized by the lowest carboxylic acid to ester ratio and
lowest anionic charge density on the backbone (Rc = 2.8) was the most retarding, followed
by PCE 17-3 (Rc = 3.9) and PCE 113-9 (Rc = 10.5). It is noteworthy that PCE 113-9 has
the longest EO side chain compared to PCE 17-3 and PCE 17-2. The addition of C3A did
not exert any influence on the duration of the induction period of alite in the mixtures
without PCEs and progressively reduced the induction period of mixtures containing
PCEs. Furthermore, the sequence of retardation of the different PCEs was not altered by
the presence of C3A, at both the dosages of C3A. These results could be in relation with
the higher affinity of PCEs towards C3A, as previously reported by different authors.17,18
Similar behaviour for the different polymers was observed by monitoring the precipitation of portlandite, as shown in Figure 2 for the mixtures without C3A. Also in this case,
all the polymers induced a retardation in the precipitation of portlandite, with the same
sequence observed for the alite dissolution. Furthermore, for each polymer mixture, the
occurrence of the precipitation of portlandite was concurrent with the increase of the dissolution of alite. Similar results were obtained for the mixtures with different dosage of C3A,
not reported in this paper.
The results of adsorption tests on the different model cements containing different PCEs
are shown in Tables 4, 5 and 6. Table 4 reports the adsorption data for mixtures without
C3A, Table 5 for mixtures with 5% C3A and Table 6 for mixtures with 10% C3A, respectively. Different from XRPD tests, which were conducted only at the dosage of 0.15% by
weight of dry PCE, adsorption tests were performed over a range of dosages (0.05, 0.10.
0.15. 0.20 and 0.30% by weight of dry PCE by weight of powder). Results of adsorption
234SP-302-17
Table 4. Adsorption data of different PCEs on mixtures alite/gypsum in the

ratio 95/5 in the absence of C3A.
Dosage
%wt
%wt
0.05
100
0.10
100
0.15
100
0.20
64
0.30
41
PCE 113-9
PCE 17-3
PCE 17-2
Adsorption
Adsorption
Adsorption
mg/m2
mmole/m2
mg/m2
mmole/m2
mg/m2
mmole/m2
%wt
%wt
(ounce/yr2) (mmole/yr2)
0.4
8.9E-06
0.4
2.8E-05
0.4
1.9E-05
100
100
(1.8E-06)
(7.4E-06)
(1.8E-06)
(2.3E-05)
(1.8E-06)
(1.6E-05)
0.8
1.8E-05
0.8
5.5E-05
0.6
3.1E-05
100
81
(3.6E-06)
(1.5E-05)
(3.6E-06)
(4.6E-06)
(2.7E-06)
(2.6E-06)
1.2
0.9
6.5E-05
0.7
3.7E-05
2.7E-05
79
64
(5.4E-06)
(4.1E-06)
(5.4E-05)
(3.2E-06)
(3.1E-05)
1.0
2.3E-05
0.8
5.8E-05
0.6
3.1E-05
53
40
(0.5E-06)
(1.9E-05)
(3.6E-06)
(4.8E-05)
(2.7E-06)
(2.6E-05)
1.0
2.2E-05
0.7
4.6E-05
0.5
2.6E-05
28
23
(0.5E-06)
(1.8E-05)
(3.2E-06)
(3.8E-05)
(2.3E-06)
(2.2E-05)
Table 5. Adsorption data of different PCEs on mixtures alite/gypsum/C3A in

the ratio 90/5/5.
Dosage
%wt
%wt
0.05
100
0.10
100
0.15
100
0.20
69
0.30
45
PCE 113-9
PCE 17-3
PCE 17-2
Adsorption
Adsorption
Adsorption
mg/m2
mmole/m2
mg/m2
mmole/m2
mg/m2
mmole/m2
%wt
%wt
2
2
2
2
2
(ounce/yr ) (mmole/yr )
(ounce/yr ) (mmole/yr )
(ounce/yr ) (mmole/yr2)
0.4
0.9E-05
0.4
2.8E-05
0.4
2.0E-05
100
100
(1.8E-06)
(0.8E-05)
(1.8E-06)
(2.3E-05)
(1.8E-06)
(1.7E-05)
0.8
1.8E-05
0.8
5.7E-05
0.6
3.1E-05
100
79
(3.6E-06)
(1.5E-05)
(3.6E-06)
(4.8E-05)
(2.7E-06)
(2.6E-05)
1.2
2.8E-05
1.1
8.1E-05
1.0
4.8E-05
96
82
(5.4E-06)
(2.3E-05)
(5.0E-06)
(6.8E-05)
(4.5E-06)
(4.0E-05)
1.1
2.5E-05
1.4
9.7E-05
0.9
4.7E-05
85
59
(5.0E-06)
(2.1E-05)
(6.3E-06)
(8.1E-05)
(4.1E-06)
(3.9E-05)
1.1
2.5E-05
1.2
8.8E-05
1.0
4.8E-05
52
41
(5.0E-06)
(2.1E-05)
(5.4E-06)
(7.3E-05)
(4.5E-06)
(4.0E-05)
tests were expressed in different ways: a) as percent by weight of PCE adsorbed with
respect to the initial amount (%wt), b) as adsorption of milligrams of PCE by unit surface
area (mg/m2, ounce/yr2) and c) as adsorption of millimole of PCE by unit surface area
(mmole/m2, mmole/yr2). Adsorption in mg/m2 were calculated from the percent adsorption
and the specific surface area of both alite (1.292 m2/g, 51.68 yr3/ounce) and C3A (1.812
m2/g, 72.48 yr3/ounce) phases, while adsorption in mmole/m2 were obtained by dividing
adsorption in mg/m2 by the number average molecular weight (Mn) of the different PCEs.
Results of Table 4 indicate that PCE 113-9 is completely adsorbed on alite up to the
dosage of 0.15% by weight and PCE 17-3 is fully adsorbed up to the dosage of 0.10%
by weight; finally, PCE 17-2 is completely adsorbed only up to the dosage of 0.05% by
weight. In general, a strong interaction between PCE and the substrate is observed and
no partition of the PCE between surfaces and solution is experimentally detected at low
dosages. At higher dosages, the adsorption value, expressed as per cent by weight (%wt),

Table 6. Adsorption data of different PCEs on mixtures alite/gypsum/C3A in
the ratio 85/5/10.
Dosage
%wt
%wt
0.05
100
0.10
100
0.15
100
0.20
100
0.30
73
PCE 113-9
Adsorption
mg/m2
mmole/m2
0.4
0.9E-05
(1.8E-06)
(0.8E-05)
0.8
1.8E-05
(3.6E-06)
(1.5E-05)
1.2
2.7E-05
(5.4E-06)
(2.3E-05)
1.6
3.6E-05
(7.2E-06)
(3.0E-05)
1.7
4.0E-05
(7.7E-06)
(3.3E-05)
%wt
100
100
100
100
62
PCE 17-3
PCE 17-2
Adsorption
Adsorption
mg/m2
mmole/m2
mg/m2
mmole/m2
%wt
0.4
2.8E-05
0.4
1.9E-05
100
(1.8E-06)
(2.3E-05)
(1.8E-06)
(1.6E-05)
0.8
5.6E-05
0.7
3.4E-05
88
(3.6E-06)
(4.7E-05)
(3.2E-06)
(2.8E-05)
1.2
8.3E-05
1.1
5.4E-05
94
(5.4E-06)
(6.9E-05)
(5.0E-06)
(4.5E-05)
1.6
11.1E-05
1.2
6.1E-05
79
(7.2E-06)
(9.3E-05)
(5.4E-06)
(5.1E-05)
1.5
10.4E-05
1.5
7.2E-05
62
(6.8E-06)
(8.7E-05)
(6.8E-06)
(6.0E-05)
Table 7. Adsorption of different PCEs on alite and C3A at saturation.

PCE
113-9
17-3
17-2
Saturation concentration, mmole/m2

(mmole/yr2)
Alite
C 3A
2.2E-05
10.2E-05
(1.8E-05)
(8.5E-05)
5.7E-05
47E-05
(4.8E-05)
(39.1E-05)
3.1E-05
24.6E-05
(2.6E-05)
(20.5E-05)
Selectivity Ratio at saturation

SRC3A/alite
4.6
8.2
7.9
is always below 100%, indicating that unadsorbed polymer molecules remain dissolved in
the mixing solution. This condition is considered representative of the saturation of PCE
on alite phase and the corresponding concentration of adsorbed molecules is considered to
be the saturation concentration of each PCE on alite phase. The saturation concentration
of the different PCEs on alite phase was expressed as the average values of adsorption at
saturation (in mmole/m2, mmole/yr2 of alite) and are reported in Table 7.
Table 5 and Table 6, concerning mixtures containing 5% and 10% C3A, respectively,
indicate that the saturation conditions were attained for higher dosages compared to the
mixtures with only alite. In saturation conditions (adsorption less than 100% by weight),
both alite and C3A particles resulted simultaneously saturated. By subtracting the saturation concentration of the different PCEs on alite phase, it is possible to calculate the saturation concentration for the different PCEs on C3A, at both the dosages of C3A.
As for alite phase, the saturation concentration of the different PCEs on C3A was calculated from the adsorption concentrations at saturation at both the dosages of C3A (5% and
10%) and the average results (in mmole/m2 of C3A) are reported in Table 7. The assumption that the amount of PCE adsorbed on alite remains constant in the presence of C3A
represents a limiting case, such that the figures displayed in Table 7 for the amount of PCE
adsorbed onto C3A are minimum values. These results indicate that the saturation concentration of PCEs for C3A is much higher than for alite phase, confirming the higher selec-
236SP-302-17
Figure 3. Increase of the time until end of induction of alite

dissolution as a function of the concentration of unadsorbed
PCE molecules in the solution.
tivity of PCEs to C3A, as previously reported by other authors.17,18 When C3A is added to
the mixture, it acts as a preferential sink for PCEs adsorption and leaves the surface of alite
phase less hindered by PCE molecules. As a consequence, the induction period and the
retardation of hydration are reduced, as shown from the dissolution kinetic curves (Figures
1A, 1B and 1C) and from the data of Table 3. Table 7 also indicates that the different PCEs
have different affinity for C3A and alite phase. This difference can be expressed by the
Selectivity Ratio SRC3A/alite, defined as the ratio of the saturation concentration on C3A and
on alite phase for each PCE. According to Table 7, PCE 17-3 and PCE 17-2 are the more
selective to C3A compared to PCE 113-9. Such results, derived from adsorption measurements, are in full agreement with the kinetic dissolution data measured by XRPD (Figures
1A, 2A, 3A and Table 3). In fact, the addition of C3A to the alite/gypsum mixtures reduced
the retardation of PCEs 17-3 and 17-2 much more than for PCE 113-9.
These results indicate that the different PCEs showed different affinity for C3A and alite
phases and their hydrates. PCEs with shorter chain (PCE 17-2 and PCE 17-3) showed
higher affinity for C3A and its hydrated phases than PCE 113-9. These findings are consistent with the results of other authors26,27 who demonstrated that PCEs intercalate in the
hydrated aluminates phases and that PCEs with longer side chains showed lower intercalation compared with PCEs with shorter side chains.
From the adsorption data and the Selectivity Ratios of the different PCEs, it is possible
to evaluate the influence of the different PCEs on the retardation in alite dissolution both in
saturated and unsaturated conditions.
Retardation in saturation conditions
The unadsorbed fraction of the different PCEs at the dosage of 0.15% by weight of dry
polymer depends on the amount of C3A and the type of PCE. The concentration of the
unadsorbed PCEs left in the mixing solution (W/S =0.5) can be easily calculated from
the initial dosage of molecules of PCEs and the corresponding adsorption, expressed in
per cent by weight (%wt) (Table 8). These results indicate that with PCE 17-2, a residual
amount of polymer always remained dissolved in the solution, even with the highest

Table 8. Unadsorbed PCEs molecules in solution after mixing with alite
containing different amount of C3A.
Type of PCE
Dosage (% by
weight of dry
polymer)
113-9
0.15
17-3
0.15
17-2
0.15
Initial mmole/g of
powder (mmole/
ounce)
2.7E-05
(9.0E-04)
10.0E-05
(33.3E-04)
6.8E-05
(22.7E-04)
Unadsorbed fraction, mmole/g of solution

(mmole/g ounce of solution)
0% C3A
5% C3A
10% C3A
0
4.2E-05
(0.0011)
4.9E-05
(0.0012)
8.0E-06
(0.0002)
2.5E-05
(0.0006)
0
0
8.2E-06
(0.0002)
amount of C3A (10%). On the other hand, with PCE 17-3, unadsorbed polymer could be
detected only in the mixture without C3A and with 5% of C3A, but not in the mixture with
10% C3A. Finally, no residual polymer in the solution were detected with PCE 113-9 at
any dosage of C3A.
The relationship between the end of induction period of the different mixtures (Table
3), measured by XRPD, and the unadsorbed fractions of Table 8 is shown in the following
Figure 3.
Figure 3 indicates that the induction period is prolonged as the concentration of PCE
polymer molecules in solution increases. These results are in agreement with the previous
work of Sowoidnich and Rssler,10 who reported an increase of the retardation on hydration by increasing the dosage of PCEs beyond the saturation concentration. These authors
attributed these results to the effect of PCEs molecules in the pore solution on the nucleation and growth of C-S-H. This hypothesis is supported by Dragoni et al.,28 who found
remarkable effects of PCEs on both the delay of the nucleation and growth of synthetic
calcium carbonate and the morphology of the newly formed crystals. These authors
hypothesize that similar effects could be produced by PCEs in cement based systems, with
consequences on the delay of setting time. Furthermore, Figure 3 indicates that, at least in
the range of investigated dosage, the retardation seems to be dependent on the number of
the residual molecules in solution rather than the type of PCEs (PCE 17-2 and PCE 17-3).
Retardation in undersaturation conditions
In conditions of undersaturation, all the PCE molecules are adsorbed and no residual
molecules are left in the solution, as in the case of PCE 113-9 for all the mixtures and of
PCE 17-3 with 10% C3A (Table 8). The mmoles of adsorbed PCE molecules on alite, in
conditions of undersaturation (100% of polymer adsorption, expressed as %wt), can be
calculated from the adsorption data by assuming that the selectivity ratios of the different
PCEs to alite and C3A are the same as those calculated in saturated conditions (Table 7).
The relationship between the end of the induction period and the concentration of PCE
molecules adsorbed in conditions of undersaturation is shown in Figure 4.
Figure 4 indicates that the end of induction period is delayed by the increase of the
coverage of alite surface by PCE molecules, confirming that the adsorbed PCE molecules
act as an hindrance for alite dissolution at the interface solid/solution. As for the oversaturated condition at higher dosages, this effect seems to be dependent on the number of
238SP-302-17
Figure 4. Increase of the time until end of induction of alite

dissolution as a function of the coverage by PCE molecules
in condition of undersaturation.
molecules which cover the alite surface rather than the molecular structure of the examined
PCEs.
According to these results, it is possible to assume that the retardation on hydration
induced by PCEs is due to a double contribution: 1) the primary effect of adsorbed molecules onto the alite surface which acts as an hindrance for alite dissolution and 2) the
supplementary effect of unadsorbed molecules in the mixing solution which delays the
nucleation of C-S-H, causing a further reduction on the alite dissolution. This supplementary effect might be even stronger in retarding the alite dissolution than the primary one.
CONCLUSIONS
The results on the simplified model cement alite/gypsum/C3A obtained by using the
coupled complementary XRPD and GPC techniques, indicated the strong influence of C3A
on the dissolution kinetic of alite in the presence of PCEs. The addition of C3A causes the
reduction of the induction period of alite in the presence of PCEs, basically for the higher
selectivity of these polymers to C3A and its hydrates compared to alite. Consequently, the
alite surface is less hindered towards hydration and also the overall adsorption of PCEs is
increased, with a corresponding reduction of unadsorbed PCE molecules in the solution.
Both these effects influence the retardation on alite hydration, and two distinct mechanisms
can be inferred: 1) the primary effect of adsorbed molecules onto the alite surface, which
acts as an hindrance for alite dissolution and 2) the supplementary effect of unadsorbed
molecules in the mixing solution which delays the nucleation of C-S-H and CH, causing
a further reduction on the alite dissolution. In both the cases, for the examined PCEs the
retardation seems to be mainly dependent on the number of polymer molecules and not on
their molecular structure. The number of polymer molecules is related to the total number
of charges, which represent both the anchoring sites for adsorption onto the surface and the
active functions for chelation/complexation of calcium ions by the unadsorbed polymer in
the solution. Nevertheless, the total number of charges introduced with the polymers cannot
fully explain the retardation induced by PCEs. The charge density of PCEs, expressed as
the ratio between carboxylic groups to ester groups, seems to be the main parameter influ-

encing the adsorption of the PCEs and, consequently the retardation: the lower the charge
density, the lower the adsorption to both phases and the higher the supplementary effect
due to unadsorbed PCEs molecules.
Specific experiments will be addressed to clarify the role of the anionic charges of the
adsorbed PCEs polymers on the dissolution rate of alite and on the potential chelation/
complexation of calcium ions in the mixing solution. Future work will be focused also on
real cement systems in order to verify the results obtained on the model cements.
AUTHOR BIOS
Giorgio Ferrari graduated in Chemistry at Padua University in 1977 and presently
is senior researcher in Mapei for concrete admixtures and development of sustainable
concrete technologies. He is author of several scientific publications and international
patents on concrete technology and of two books on environmental issues.
Vincenzo Russo received his B.Sc. degree in chemistry at the University of Messina,
Italy, in 2004 and M. Sc. degree in physical chemistry at the University of Pisa, Italy, in
2006. He received a Master of Arts degree in organic chemistry from Rice University,
Houston, Texas, in 2009. He currently holds a research scientist position at Mapei S.p.A.,
Milan, Italy, where he works in the field of concrete admixtures.
Massimo Dragoni graduated in Industrial Chemistry at the Milan University in 1992.
Presently, he is senior researcher and responsible of the liquid chromatography section of
the Analytical Department in Mapei.
Gilberto Artioli holds a Laurea Degree from the University of Modena and a PhD in
Geophysical Sciences from the University of Chicago. He is presently full professor at
the University of Padua and Director of the CIRCe Centre for the investigation of cement
materials.
Maria Chiara Dalconi received a degree (B.S. and M.S.) in Geological Sciences in 1997
and a Ph.D. in Mineralogy, Petrology and Crystallography in 2001 from the University of
Modena and Reggio Emilia (Italy).
Presently she is assistant professor of Mineralogy in the Department of Geoscience at the
University of Padua and her principal research activities are in mineralogy applied to
cement materials.
Michele Secco is postdoctoral Fellow at the Department of Civil, Environmental and
Architectural Engineering of the University of Padua. He graduated in Science and
Technology for the Archaeological and Artistic Heritage in 2008, and obtained his Ph.D.
degree in Earth Sciences at the Department of Geosciences of the University of Padua
in 2012. His research focuses on the mineral-petrographic, chemical, microstructural
and physical-mechanical characterization of inorganic binders and structural materials
(cement, lime, mortars and plasters, concrete).
240SP-302-17
Leonardo Tamborrino received his master degree in Geological Sciences and Technologies from the University of Modena and Reggio Emilia in 2014, with exchange visits as
guest scientist in the Marine Geology group of MARUM-University of Bremen. He is
presently a post-graduate grant holder at the University of Padua for chemical-physical
investigation on cement materials.
Luca Valentini received a PhD in Earth sciences at the National University of Ireland
Galway in 2009. He is currently a post-doctoral fellow at the Department of Geosciences,
University of Padua. He is the author of several papers focused on computer modelling
and 3D microstructural studies in the field of materials and Earth sciences.
REFERENCES
1. Rixom, R., and Mailavaganam, N. P., Chemical admixtures for concrete, E&F Spon,
London, 1999.
2. Flatt, R. J., and Schober, I., Superplasticizers, in Understanding the rheology of
concrete (Ed. Roussel N.), Woodhead publishing, 1999, pp. 144-208.
3. Uchikawa, H.; Hanehara, S.; and Sawaki, D., The role of steric repulsive force in the
dispersion of cement particles in fresh paste prepared with organic admixture, Cement and
Concrete Research, V. 27, No. 1, 1997, pp. 37-50. doi: 10.1016/S0008-8846(96)00207-4
4. Ohta A., Sugiyama T. and Uomoto T., Study of the dispersing effect of polycarboxylate-based dispersant on fine particles, 6th CANMET, SP 195-14, 2000, pp. 211-227.
5. Flatt, R. J.; Schober, I.; Raphael, E.; Plassard, C.; and Lesniewska, E., Conformation
of adsorbed comb copolymer dispersants, Langmuir, V. 25, No. 2, 2009, pp. 845-855. doi:
10.1021/la801410e
6. Ferrari, L.; Kaufmann, J.; Winnefeld, F.; and Plank, J., Interaction of cement model
systems with superplasticizers investigated by atomic force microscopy, zeta potential,
and adsorption measurements, J. Coll. Interf. Sci., V. 347, No. 1, 2010, pp. 15-24. doi:
10.1016/j.jcis.2010.03.005
7. Mollah, M. Y. A.; Adams, W. J.; Schennach, R.; and Cocke, D. L., A review of
cement-superplasticizer interactions and their models, Advances in Cement Research, V.
12, No. 4, 2000, pp. 153-161. doi: 10.1680/adcr.2000.12.4.153
9. Cheung, J.; Jeknavorian, A.; Roberts, L.; and Silva, D., Impact of admixtures on the
hydration kinetics of Portland cement, Cement and Concrete Research, V. 41, No. 12,
10. Sowoidnich, T., and Rssler, C., The influence of superplasticizers on the dissolution of C3S, Superplasticizers and Other Chemical Admixtures in Concrete, SP-262, T.C.
Holland, P.R. Gupta, V.M. Malhotra ed., American Concrete Institute, Farmington Hills,
Mich., 2009, pp. 335-346
11. Ridi, F.; Dei, L.; Fratini, E.; Chen, S.-H.; and Baglioni, P., Hydration kinetics of tricalcium silicate in the presence of superplasticizers, The Journal of Physical Chemistry
B, V. 107, No. 4, 2003, pp. 1056-1061. doi: 10.1021/jp027346b

12. Pourchet, S.; Comparet, C.; Nicoleau, L.; and Nonat, A., Influence of PC superplasticizers on tricalcium silicate hydration, 12th ICCC, Montreal, Canada, 2007.
13. Eusebio, L.; Goisis, M.; Manganelli, G.; Patern, V. A.; and Gronchi, P., Influence
of comb-polymer structure on C3S phase hydration, Mat. Sci. Appl., V. 4, 2013, pp. 35-44.
14. Matsuyama, H., and Young, J. F., Intercalation of polymers in calcium silicate
hydrate: a new synthetic approach to biocomposites, Chemistry of Materials, V. 11, No. 1,
1999, pp. 16-19. doi: 10.1021/cm980549l
15. Popova, A.; Geoffroy, G.; Renou-Gonnord, M.-F.; Faucon, P.; and Gartner, E., Interaction between polymeric dispersants and calcium silicate hydrates, Journal of the American Ceramic Society, V. 83, No. 10, 2000, pp. 2556-2560. doi: 10.1111/j.1151-2916.2000.
tb01590.x
16. Beaudoin, J. J.; Dram, H.; Raki, L.; and Alizadeh, R., Formation and properties of
C-S-H-PEG nano-structures, Materials and Structures, V. 42, No. 7, 2009, pp. 1003-1014.
doi: 10.1617/s11527-008-9439-x
10.1016/j.cemconcomp.2009.01.005
18. Nawa T. and Eguchi H., Effect of cement characteristics on the fluidity of cement
paste containing organic admixture, 9th ICCC, 1992, pp. 597-603.
19. Plank, J.; Zhimin, D.; Keller, H.; Hossle, F.; and Seidl, W., Fundamental mechanism
for polycarboxylate intercalation into C3A hydrate phases and the role of sulfate present
in cement, Cement and Concrete Research, V. 40, No. 1, 2010, pp. 45-57. doi: 10.1016/j.
20. Flatt, R. J., and Houst, Y. F., A simplified view on chemical effects perturbing the
action of superplasticizers, Cement and Concrete Research, V. 31, No. 8, 2001, pp. 11691176. doi: 10.1016/S0008-8846(01)00534-8
21. Yamada, K.; Ogawa, S.; and Hanehara, S., Controlling of the adsorption and
aqueous phase, Cement and Concrete Research, V. 31, No. 3, 2001, pp. 375-383. doi:
10.1016/S0008-8846(00)00503-2
22. Gay, C., and Raphal, E., Comb-like polymers inside nanoscale pores, Advances
in Colloid and Interface Science, V. 94, No. 1-3, 2001, pp. 229-236. doi: 10.1016/
S0001-8686(01)00062-8
23. Quennoz, A., and Scrivener, K. L., Interactions between alite and C3A-gypsum
hydrations in model cements, Cement and Concrete Research, V. 44, 2013, pp. 46-54.
doi: 10.1016/j.cemconres.2012.10.018
24. Juilland, P.; Gallucci, E.; Flatt, R. J.; and Scrivener, K., Dissolution Theory applied
to the Induction period in Alite Hydration, Cement and Concrete Research, V. 40, No. 6,
25. Ferrari, G.; Cerulli, T.; Clemente, P.; Dragoni, M.; Gamba, M.; and Surico, F., Influence of Carboxylic Acid-Carboxylic Ester Ratio of Carboxylic Acid Ester Superplasticizers
on Characteristics of Cement Mixtures, Superplasticizers and Other Chemical Admixtures in Concrete, SP-195, V.M. Malhotra ed., American Concrete Institute, Farmington
Hills, Mich., 2000, pp. 505-519.
242SP-302-17
26. Flatt, R. J., and Houst, Y. F., A Simplified View on Chemical Effects Perturbing the
Action of Superplasticizers, Cement and Concrete Research, V. 31, No. 8, 2001, pp. 11691176. doi: 10.1016/S0008-8846(01)00534-8
27. Plank, J.; Dai, Z.; Zouaoui, N.; and Vlad, D., Intercalation of Polycarboxylate
Superplasticizers into C3A Hydrate Phases, 8th CANMET/ACI International Conference
on Superplasticizers and Other Chemical Admixtures in Concrete, SP-239, V.M. Malhotra
ed., American Concrete Institute, Farmington Hills, Mich., 2006, pp. 201-214
28. Dragoni, M.; Lo Presti, A.; Cerulli, T.; Biancardi, A.; Moretti, E.; and Salvioni, D.,
Interaction of polycarboxylate-based superplasticizers with cement: synthetic calcium
carbonate as model to investigate the structural evolution of C-S-H, Proceeding of the
34th International Conference in Cement Microscopy, April 1-4, 2012 Halle (Saale),
Saxony-Auhalt, Germany.
SP-302-18
Preparation and Mechanism Study

of Slow-Release Polycarboxylate
Superplasticizers
by Jinzhi Liu, Jiaping Liu, Yong Yang, Dongliang Zhou,
and Qianping Ran
Slump loss of fresh concrete was a common issue in engineering construction, especially under high temperature and long distance transportation conditions. Therefore,
slow-release polycarboxylate superplasticizers (PCEs) have been widely used to reduce
the slump loss in various engineering projects. In this study, three kinds of PCEs with
different proportions of hydroxyl ester groups (HEG) were synthesized and characterized
by 1H-NMR and Gel Permeation Chromatography (GPC). The effects of the HEG content
on dispersion retention, adsorption kinetics and zeta potential of fresh cement suspensions
were systematically investigated to figure out the mechanism.
For PCEs with the same molar ratio of carboxyl group and reactive polyether, the
dispersion retention ability of PCEs is improved with the increasing of HEG ratio. HEG in
PCEs can be slowly converted to carboxyl groups in the alkaline environment of cement
suspension, which could enhance the adsorption of PCE molecules onto the surface of
cement particles. Despite major of the initially adsorbed-PCE molecules might have been
embedded in hydration products, free PCE molecules with released carboxyl groups in
the solution can continuously adsorb onto the surface of cement particles and play a role
in dispersion. This explains why slow-release PCEs have a dispersion retention effect on
cement particles within a certain time.
Keywords: polycarboxylate; slow-release; zeta potential; adsorption; dispersability
retention.
INTRODUCTION
Polycarboxylate superplasticizers (PCEs) are widely used in different industrial fields to
improve the performance of concrete. With the development of the construction industry
and requirements of environmental protection, the proportion of ready-mixed concrete in
building construction is growing. The slump loss of fresh concrete is a major problem,
especially under high temperature and long distance transportation conditions. The traditional methods to solve the fluidity loss, such as changing the mixing concrete process,
243
244SP-302-18
Table 1 Phase composition and physical properties of the cement

Chemical composition (wt %)
SiO2
19.5
Al2O3
6.16
CaO
64.3
MgO
1.04
Fe2O3
4.41
SO3
2.52
K2O
0.20
Na2O
0.48
Loss
1.39
Percent volume
diameters/m
D50
VMD
16.38
20.76
Blain surface
areas/cm2g-1
2965
adding water to reshape concrete, adding water reducing agent repeatedly and set retarder,
might not only increase the production cost, but also have a serious impact on the strength
and durability of concrete.1-4 Therefore, slow-release PCEs have been widely used to
reduce the slump loss in various engineering projects.
Slow-release PCEs could disperse cement particles for a long time. The slump of fresh
concrete mixed with slow-release PCEs could keep the slump for a long period while
its loss is small. This characteristic can maintain good performance of fresh concrete
under high temperature and long distance transportation conditions. Besides, the concrete
obtained the good properties such as compressive strength, freeze-thaw resistance, and
low-carbonation properties and so on.
Robert Flatt5 considered there might be three forms of PCE molecules in early hydration
stage: (1) One form is the adsorbed PCE molecule on the surface of unhydrated cement
particles and cement hydration products. The initial dispersion effect comes from PCE of
this form. (2) The second form is consumed by intercalation,coprecitation or micellization,
which does not play any role in the dispersion process. (3) The last form is the remained
free PCE molecule neither consumed nor adsorbed in the slurry, which might be in equilibrium with PCE molecule of the first form and thus closely associated with the long-time
fluidity.
In order to maintain the slump of concrete, the amount of PCE molecule of the last form
should be increased. Herein hydroxyl ester groups (HEG) were introduced into PCE molecules. This type of PCE contained low content of carboxyl groups and thus low amount of
PCE was adsorbed at initial stage. Based on the molecular design principles, three kinds
of PCEs with different proportions of HEG were synthesized and characterized, and the
mechanism of slow-released PCEs was explored.
Materials
An ordinary portland cement (52.5, Jiangnan Onoda Co., China), which meets the
requirements of GB8076 standard was used throughout all the experiments. The cement
composition was determined by X-ray fluorescence. The particle size analysis using a
Helos-Sucell Laser particle size analyzer (SYMPATEC Instruments, Germany) showed
the volume average mean particle size (VMD) to be about 20.76 m(8.1710-4in), and the
percentage volume diameters d50 to be 16.38 m(6.4510-4in). The Blaine surface area
is 2965 cm-2g-1. The characteristics and compositions of the cement are given in Table 1.
Synthesis of slow-release Superplasticizers
Slow-release PCEs with different proportions of hydroxy ester groups (HEG) were
prepared from methallyl ether of polyethylene glycol (MAPEG, Mw=2000) and hydroxyethyl acrylate (HEA) through a free-radical copolymerization in an aqueous medium at
Preparation and Mechanism Study of Slow-Release Polycarboxylate

Superplasticizers245
Fig. 1 -Chemical structure of the slow-release PCEs comb

polymer dispersants
room temperature under a nitrogen gas atmosphere. Sodium persulfate (Na2S2O8) was used
as the initiator and sodium hydrogen sulfite (NaHSO3) as a reductant and a chain transfer
agent in order to control the molecular weight. The chemical structure of synthesized
polymer samples was shown in Figure 1. Polymer samples with different MAPEG to HEA
ratios (1:2, 1:4, 1:6) and a constant side chain length were synthesized and named respectively MAH2, MAH4 and MAH6. The composition and molecular weight of polymers
were determined by 1H-NMR spectroscopy and a Shimadzu Refractive Index detector
equipped with TSK-PWXL (TOSOH) columns, respectively. The architecture of prepared
PCEs and their molecular characteristics are listed in Table 2.
Dispersion effect of slow-release PCEs
The fresh cement pastes were prepared at 20oC and a water to cement ratio (w/c) of
0.29 using a cement mortar mixing machine according to GB/T8077-2008. Water (87 g)
containing a certain amount of PCEs was added to cement (300 g) and mixed for 2 min at
low speed and further 2 min at high speed. The paste was filled in a cone with a height of
60 mm, an inner diameter at the bottom of 36 mm and at the top of 60 mm, respectively.
The cone was pulled up and the spread diameter (F) of the cement cake was measured. The
spread value (F) was the average of two perpendicularly crossing diameters. The cement
paste was also used for the measuring its fluidity. From the spread (F), the dispersion ability
(relative flow area ratio) was calculated by,6 Q was the area ratio of final spread and the
bottom cone

Q= (F2-602)/602=F2/602-1
246SP-302-18
Table 2 Structure information of slow-release copolymers

Sample
Molar composition/%
MAPEG
HEA
Mna
Mw b
MAH2
33.3
66.7
24100
42900
MAH4
20
80
25900
43600
MAH6
16.7
83.3
24600
45100
Mn: number average molecular weight.
Mw: weight average molecular weight.
Adsorption of slow-release PCEs on cement particles

Adsorption measurements were performed to detect the quantity of superplasticizer
adsorbed on the cement particles by means of a total organic carbon analyzer (TOC),
Multi N/C3100 (analytikjene AG, Germany). After mixing the cement with the solution
containing the polymer, the amount of superplasticizer remaining in the solution can be
measured from the separated solution phase from the suspension. The consumed polymer
is estimated to be the difference in concentration before and after contact with the cement
powder. 120g solution containing 0.2% of PCEs and 60g of cement were mixed. The
sample solution was separated at 1, 5, 15, 30, 60 and 90 min by centrifuging at 10,000 rpm
for 2 minutes. The supernatant was immediately decanted and diluted with deionized water
for TOC.
Zeta potential of cement particles containing slow-release PCEs
The surface charge of colloidal systems is called zeta potential. The zeta potential data
were collected with the electroacoustic-based ZetaProbe analyzer (Colloidal Dynamics
Inc., American), zeta potential represents the potential separations between the dispersion
medium and the stationary layer of water molecules and ions attached to the dispersed
particle. The suspensions were prepared by dispersing 100 g of cement in 200 ml deionized
water containing different dispersants under vigorous agitation, and then zeta potential of
the cement paste was determined from 1 to 90 min.
Slow-release PCEs with different proportions of hydroxyl ester groups (HEG) were
synthesized and characterized by 1H-NMR. The effects of the HEG content on dispersion
retention, the adsorption, and zeta potential of the concentrated cement suspensions were
investigated in details.
Structures of the prepared slow-released PCEs
1
H-NMR spectra of slow-released PCEs copolymer are shown in Fig. 2. The 1H-NMR
spectra of the three copolymers are substantially the same. Peaks are the same position only
different peak intensity. A case study in MAH4, the signals at = 1.0~1.8 ppm (b) and =
1.9~2.5 ppm (c) were attributed to CH2 protons and CHCOOR proton in the polymer backbone, respectively. The peak at =0.82 ppm (a) corresponds to the CH3 protons, and the
peak at = 3.4~3.8 ppm (d) corresponds to the CH2 protons of the ethylene oxide repeating
units in MAPEG. In addition, a small peak at = 4.21~4.28 ppm (f) is attributed to the CH2

Fig. 2- 1H-NMR spectra of slow-release PCEs copolymer of

MAH4 (D2O).
protons that are attached to the ester group, and the peak at = 3.86 ppm (e) corresponds
to the CH2 protons attached to the hydroxyl group. Based on the 1H-NMR spectra, it was
demonstrated that the preparation of slow-release PCEs was successful.
Effect of slow-release PCEs dispersion of cement pastes
The effect of PCEs with different proportions of HEG on dispersion retention was shown
in Fig. 3. The spread of cement pastes containing different proportions of HEG shows the
same behavior. As the proportion of HEG increased, the paste fluidity increased quickly.
This result was in good agreement with the adsorption behavior. The dispersion ability of
MAH6 was far greater than MAH2 after 30 min. So PCE with the higher HEG proportion
had a better dispersion retention effect on cement particles within a certain time.
Adsorption behavior of slow-release PCEs
Adsorption of dispersant on solidliquid interfaces is a crucial step in dispersing
cement,7-9 and adsorption was the one of the most important measurements for illustrating
the interaction between PCEs and cements. The adsorption amounts of PCE in the suspensions with the same dosage were shown in Fig. 4. For the three PCE polymers, the adsorption curves appeared the same trend. The adsorption amount increased slowly within 30
min. with this period, although the surfaces of cement particles contained empty sites, relatively small amount molecules can be adsorbed on the surface of cement particles owning
to few adsorption groups (carboxyl group). At the same time, the remaining parts of PCEs
in the water phase can be slowly converted carboxyl group and gradually adsorbed on the
cement particles. After 30min, the adsorption of polymers on cement particles carried on
continuing to quickly increase. MAH6 had the maximum amount of adsorption among
248SP-302-18
Fig. 3- Influence of different HEG in PCEs on dispersion

ability of cement paste
three polymers, resulting from the larger proportion of adsorption groups (carboxyl group)
which came from the hydrolysis of HEG. PCE molecules adsorbed more easily onto the
surface of cement particles if the HEG ratio in PCE was larger, and with larger proportion
of adsorption.10,11
Effect of slow-release PCEs on zeta potential of cement pastes
The zeta potential of cement suspensions with slow-release PCEs with different HEG
proportions and polyoxyethylene long side chain of the same length was measured against
time (Fig. 5). The behavior of zeta potential development of the cement systems with
MAH2, MAH4 and MAH6 are completely different. Zeta potential of cement with MAH2
underwent a gradual increase with time. Zeta potential of cement with MAH4 appeared
with the same behavior before 72 min and slightly fell in later time. Zeta potential of the
cement with MAH6 seemed similar with MAH4 and changed in larger range.
When adding the PC, cement particles adsorbed carboxyl groups and formed electrical
double layer on the cement particles. As time went on, HEG could be slowly converted to
carboxyl groups (COO-) in the alkaline environment of cement suspension, which would
have a direct influence on the zeta potential of the system, while the carboxyl groups in
the polymer backbone might be wrapped by the long side chains.12 Therefore, different
amounts of HEG per molecular caused different zeta potentials.
Within 30min the zeta potential increase was caused by hydration: (1) More positive
charge generated by the initial hydration of C3A; (2) The adsorbed PCE molecules might
be embedded in hydration product. After 30min, the increase of zeta potential became
slower (MAH2), or the zeta potential began to decrease (MAH4 and MAH6). This could
be attributed to the hydrolysis and continuous adsorption of PCE molecules.

Fig. 4- Zeta potential of cement suspensions with different

PCEs.
The effect of slow-release PCEs on zeta potential and adsorption has been measured
in order to elucidate the mechanisms of slow-release PCEs with different HEG proportions. When PCEs was added in the cement system, low amount of adsorbed PCE caused
a small initial fluidity of cement paste. At the same time, HEG continuously hydrolyzed
and converted to carboxyl group in the alkaline environment of cement. Thus, the proportion of carboxyl groups became larger and larger. As the hydration of cement went on, the
PCE molecules adsorbed on the cement particles would be covered by hydration products gradually and a fluidity loss might be observed. The remaining parts of PCEs in the
water phase can be adsorbed on cement particles and continue to play a role of dispersant.
Therefore,the adsorption amount increased within a certain period. With the increase in the
proportion of HEG, carboxyl group gradually increased which is providing adsorption on
the cement particles. So more PCEs in cement slurry can keep the cement particles fluidity
and adsorption equilibrium, and slump loss of cement fluidity became lower.
PCEs with different carboxyl content have different adsorption capacity and adsorption
rate on cement particles, which affect the zeta potential of cement particles. For MAH2,
due to the low HEG proportion of MAH2, less carboxyl groups could be converted from
HEG compared with MAH4 and MAH6. The amount of carboxyl groups of adsorbed PCE
molecules were not enough to offset the positive charge increase from the released calcium
ions during hydration process,8 the zeta potential increased gradually. But for PCEs with
a high proportion of HEG (MAH4 and MAh6), higher proportion of carboxyl groups was
released. At a certain time, the increasing positive charge of cement particles from the
released Ca2+ might be neutralized by the adsorbed PCEs with higher carboxyl group
proportions. Then the zeta potential began to decrease.
In short, the slump retention mechanism of slow-release PCEs was caused by the hydrolysis of PCEs and continuous adsorption of PCE molecules on the surface of cement particles. The adsorption speed and adsorption equilibrium of PCEs to cement particles affects
the dispersion retention of cement particles within a certain time.
250SP-302-18
Fig. 5-Amount of slow-release PCEs adsorbed on cement

particles versus time
CONCLUSIONS
In summary, this paper developed a slow-release type PCE which could effectively
reduce the slump loss of cement paste. The mechanism of PCEs in dispersion retention
was studied. This type of PCE might be significantly useful for high temperature and long
distance transportation conditions of concrete.
1. Slow-release PCEs with different proportions of HEG were synthesized by radical
polymerization and characterized by 1HNMR and GPC. The polymer structure was consistent with the experimental design.
2. With the increasing of HEG ratio while the same molar ratio of reactive polyether, the
dispersion retention ability of PCEs was improved.
3. Researches on zeta potential and adsorption kinetics show that HEG in PCEs could
be slowly converted to carboxyl groups in the alkaline environment of cement suspension,
which could enhance the adsorption of PCE molecules onto the surface of cement particles.
The adsorption speed and adsorption equilibrium of PCEs to cement particles affects the
dispersion retention of cement particles within a certain time.
AUTHOR BIOS
Jinzhi Liu is a researcher of Jiangsu Research Institute of Building Science, Nanjing,
China. She received her MS in Polymer Chemistry and Physics Zhejiang University.
She has been working in the field of concrete admixtures, focusing on polycarboxylate
superplasticizers. She is author or co-author of over 10 papers and patents on chemical
admixtures.
Jiaping Liu is a Vice-President of Jiangsu Research Institute of Building Science and
General Manager of Jiangsu Bote New Material Ltd. He received his MS in Civil Engineering from Southeast University and PhD in materials science from Nanjing University
of Technology. He has written and presented over 30 papers on his research activities

related to high performance concrete, high performance admixture, deformation and
cracking of concrete.
Yong Yang is a researcher of Jiangsu Research Institute of Building Science, Nanjing,
China. He received his MS in Polymer Chemistry and Physics from Lanzhou University. His research interests include polycarboxylate superplasticizers. He is author or
co-author of over 20 papers and patents on chemical admixtures.
Dongliang Zhou is a researcher of Jiangsu Research Institute of Building Science,
Nanjing, China. He received his MS in Polymer Chemistry and Physics from Sichuan
University. He has been working in the field of polycarboxylate superplasticizers. He is
author or co-author of over 10 papers and patents on chemical admixtures.
Qianping Ran is a Chief scientist at State Key Laboratory of High Performance Civil
Engineering Materials and a Research Engineer (Prof.) in Admixture Product Technologies at Jiangsu Research Institute of Building Science, Nanjing, China. He received his
MS in Material Science from Sichuan University and PhD in Polymer Chemistry and
Physics from Nanjing University. He has been working in the field of concrete admixtures, focusing on the development of new polymeric superplasticizers. He is also author
or co-author of over 60 papers and patents on chemical admixtures.
REFERENCES
1. Schober, I., and Flatt, R. J., Optimizing polycarboxylate polymers. 8th CANMET/ACI
International Conference Superplasticizers and Other Chemical Admixtures in Concrete
Sorrento: ACI, 2006, 16
3. Plank, J., and Winter, C., Competitive adsorption between superplasticizer and
retarder molecules on mineral binder surface, Cement and Concrete Research, V. 38, No.
5, 2008, pp. 599-605. doi: [J]10.1016/j.cemconres.2007.12.003
4. Papayianni, I.; Tsohos, G.; Oikonomou, N.; and Mavria, P., Influence of superplasticizer type and mix design parameters on the performance of them in concrete mixture,
5. Flatt, R.-J., and Houst, Y.-F., A simplified view on chemical effects perturbing the
6. Yamada, K.; Hanehara, S.; and Honma, K., Effects of the chemical structure on the
properties of polycarboxylate-type superplasticizer, Cement and Concrete Research, V.
30, No. 2, 2000, pp. 197-207. doi: 10.1016/S0008-8846(99)00230-6
7. Yoshioka, K.; Tazawa, E.; Kawai, K.; and Enohata, T., Adsorption characteristics of
superplasticizers on cement component minerals, Cement and Concrete Research, V. 32,
No. 10, 2002, pp. 1507-1513. doi: 10.1016/S0008-8846(02)00782-2
252SP-302-18
8. Zingg, A.; Winnefeld, F.; Holzer, L.; Pakusch, J.; Becker, S.; and Gauckler, L., Adsorption of polyelectrolytes and its influence on the rheology, zeta potential, and microstructure
of various cement and hydrate phases, Journal of Colloid and Interface Science, V. 323,
No. 2, 2008, pp. 301-312. doi: 10.1016/j.jcis.2008.04.052
9. Yamada, K.; Ogama, S.; and Hanehara, S., Controlling of the adsorption and
aqueous phase, Cement and Concrete Research, V. 31, No. 3, 2001, pp. 375-383. doi:
10.1016/S0008-8846(00)00503-2
No. 10, 2002, pp. 1507-1513. doi: [J]10.1016/S0008-8846(02)00782-2
11. Plank, J., and Winter, C., winter, Ch. Competitive adsorption between superplasticizer and retarder molecules on mineral binder surface, Cement and Concrete Research,
V. 38, No. 5, 2008, pp. 599-605. doi: 10.1016/j.cemconres.2007.12.003
12. Plank, J., and Hirsch, C., Impact of zeta potential of early cement hydration phases
on superplasticizer adsorption, Cement and Concrete Research, V. 37, No. 4, 2007, pp.
537-542. doi: 10.1016/j.cemconres.2007.01.007
SP-302-19
Effect of Pore Solution Composition

on Zeta Potential and Superplasticizer
Adsorption
by Dirk Lowke and Christoph Gehlen
The effect of pore solution composition on zeta potential and superplasticizer adsorption
has been investigated experimentally. The investigations were conducted on highly concentrated suspensions, containing quartz flour, limestone flour, cement and combinations of
these materials. Furthermore cement-limestone suspensions with different types of cements
and a varying ratio of cement to limestone were investigated.
The results show that the zeta potential is significantly determined by pore solution. In
a pore solution of highly concentrated cement suspensions the zeta potential can be characterized by the ratio of calcium to sulfate concentration. Furthermore it was shown that
the superplasticizer adsorption is affected the zeta potential. At higher, more positive zeta
potentials the superplasticizer molecules are more likely adsorbed onto the solid surfaces.
Moreover, superplasticizer adsorption in limestone-cement suspensions is predominantly
controlled by the composition of pore solution rather than the ratio of cement to limestone flour. If the ion concentration of the pore solution is artificially kept constant, the
polymer adsorption is almost constant independent of the cement to limestone ratio in the
suspension.
Keywords: zeta potential; pore solution; superplasticizer adsorption.
INTRODUCTION
Rheological properties of high performance concrete like SCC and UHPC are affected
by the surface properties of the particles, the properties of the liquid phase and the adsorbed
polymers. It is necessary to consider the interactions of the colloidal particles in the cementbased suspension to understand the effect of these parameters on the rheological properties.
The present contribution focuses on the effect of pore solution composition on zeta potential and superplasticizer adsorption in highly concentrated cement-based suspensions.
The aim of the investigations was to provide a clear understanding of the mechanism of
surface charge and zeta potential in cement suspensions with high solid fractions as well
as the interactions between the composition of pore solution, zeta potential and superplasticizer adsorption onto surfaces of cement and mineral additions.
253
254SP-302-19
For this reason, the experimental program was divided into to three test series:
a) Zeta potential of quartz, limestone and cement in suspensions with high solid fractions
- Effect of pore solution composition
b) Superplasticizer adsorption onto quartz, limestone and cement - Effect of zeta potential
c) Superplasticizer adsorption in limestone-cement suspensions - Effect of cement and
limestone content
ELECTRIC DOUBLE LAYER AND ZETA POTENTIAL
Surface charge and Stern potential of oxides and carbonate minerals
As soon as a cement or mineral addition particle is exposed to water, its surface becomes
electrically charged. The potential 0 resulting from the surface charge is causing a predominant attraction of ions of opposite charge (counterions). As a result, an electrical double
layer is formed between the charged particle surface and the liquid phase. From a historical
point of view, the double layer model originates from the Helmholtz model (1879) and the
Gouy-Chapman model (1910/13). Finally, both models were combined by Stern (1924).1
The Stern model describes the interface as a combination of a rigid layer of adsorbed ions
on the surface, the so called Stern layer, and a diffuse layer of mobile counterions.
The mechanisms causing a surface charge are complex. According to the Surface
Complexation Theory2 surface charges are formed due to chemisorption and dissociation
of water molecules onto unsaturated surface lattice ions. The oxygen atoms of chemisorbed
water molecules fill the vacant cationic surface sites. Simultaneously, the anionic surface
sites are stabilized by the transfer of dissociated protons from the chemisorbed water
molecules. This hydration process leads to a surface composed of hydroxylated cationic
sites (>KatOH0) and protonated anionic sites (>AnH0), where > represents the mineral
lattice and Kat or An is the mineral cation or anion respectively.2 A surface potential 0 is
finally formed by protonation (>KatOH2+) or deprotonation (>KatO, >An) of the hydration surface sites.
For oxide minerals like SiO2, the anionic crystal lattice sites are formed by an oxygen
atom of the mineral. Thus the primary hydration surface sites are described as hydroxylated
surface cations (>SiOH0). By protonation or deprotonation of the hydroxyl groups surface
charges are formed as follows:
Positive surface sites: >SiOH0 + H3O+ >SiOH2+ + H2O
Negative surface sites: >SiOH0 + OH >SiO + H2O
Hence, the surface charge of oxides in water is controlled by the pH-value of the solution. At high pH-values, as usual in cement-based suspensions, a negative surface potential
0 is formed.
For carbonate minerals like CaCO3, the surface consists of hydroxylated cationic sites
(>CaOH0) as well as protonated anionic sites (>CO3H0). By protonation or deprotonation
surface charges are formed, again depending on solution pH.
Positive surface sites: >CaOH0 + H3O+ >CaOH2+ + H2O
Negative surface sites: >CaOH0 + OH >CaO + H2O
>CO3H0 + OH >CO3 + H2O
In addition, the charge of a CaCO3-particle is determined by the adsorption of calcium
ions (Ca2+), hydrogen carbonate ions (HCO ;3 ) and carbonate ions (CO 2 ;3 ). These ions
Effect of Pore Solution Composition on Zeta Potential and Superplasticizer

Adsorption255
originate from carbonic acid (H2CO3) as well as marginal dissolved amounts of calcium
carbonate (14 mg/l at 20C).
Calcium ions: (CaCO3(s) + H2O Ca2+ + HCO3 + OH)
>CO3-Ca+, >CaO-Ca+, >Ca-HCO30
Carbonate ions: (CO2 + 3H2O H2CO3 + 2H2O HCO3 + H3O+ + H2O CO32 +
2H3O+)
>Ca-HCO30, >Ca-CO3
The solubility of these ion species is controlled by the pH-value of the solution. Mainly
hydrogen carbonate ions as well as calcium ion are existent in neutral pH-range. Providing
a sufficient calcium concentration, CaCO3particles dispersed in water in contact with
atmospheric air therefore can have a positive Stern potential. With increasing pH-value, the
number of dissolved carbonate ions increases. Due to adsorption of the divalent carbonate
ions onto cationic surface sites, an increasingly number of negative charges are formed in
the stern layer (>Ca-CO3). Simultaneously, the concentration of dissolved calcium ions is
decreasing, resulting in a reduction of positive sites like >CO3-Ca+ or >CaO-Ca+. Furthermore, negative surface charges are formed by deprotonation of the hydroxylated calcium
sites or carbonate sites of the surface (>CaO, >CO3). Thus, the Stern potential of calcium
carbonate in alkaline milieu becomes progressively negative.
The role of ion adsorption in cement-based suspension
The pore solution of cement-based suspensions is typically characterized by a high pH >
12 as well as a high ionic strength > 100 mmol/l. A considerable difference compared to the
mechanism described before is the high concentration of dissolved ions in pore solution.
Furthermore, besides dissolved ions from the crystal lattice of the particle surface, there
are additional ion species, originated from readily soluble cement components. During
the first minutes after water addition alkali sulfates (Na2SO4, K2SO4), free lime (CaO)
and parts of the setting regulator (CaSO4xH2O) are dissolved. Besides pH-controlling
hydroxide anions, the pore solution contains further anions like sulfate and chloride as
well as cations like calcium, potassium and sodium. Further Ca2+-ions and OH-ions results
from the hydrolysis of tricalcium aluminate and tricalcium silicate.
These ions affect the charge of the dispersed particle significantly by adsorbing to the
stern layer. In this context, the adsorption of divalent calcium cations and sulfate anions are
of particular importance for the formation of the Stern potential. Surface charge equalization increases with increasing adsorption of ions to the Stern layer. Furthermore, in case of
the adsorption of divalent ions a charge reversal is possible.1 For calcium carbonate suspensions, it could be experimentally confirmed that the effect of calcium ion concentration is
of more importance than the effect of pH-value.3 Similarly, a positive charge reversal of an
initially negative surface charge due to adsorption of cations was verified in experiment.4,5
Zeta Potential
While the surface potential 0 in a cement-based suspension is - owing to ion adsorption - experimentally not accessible, the determination of the zeta potential Z allows the
conclusions to the Stern potential S. The zeta potential Z of a particle surface is defined
as the potential at a shear plane that arises from particle motion. The exact location of
this shear plane is a matter of controversial discussions in the literature. While in various
256SP-302-19
Table 1 Characteristics of powder materials

r
[oz/in3]
1.79
As
[ft2/in3]
3.3
r
[lb/yd3]
5225
3.7
5225
1.79
653
3.0
5225
1.79
529
1.56
529
1.62
723
r
[g/cm3]
As
[m2/cm3]
Cement (C(a))
3.1
Cement (C(b))
3.1
Cement (C(c))
3.1
Quartz flour (Q)
2.7
3.0
4550
Limestone flour (L)
2.8
4.1
4720
582
Table 2 Pore solution composition of suspension L.C(a)

Ion species
Concentration [mmol/l]
Ca2+
16.9
Na+
27.8
K+
198.9
Cl
33.6
SO42
88.0
OH
50.9
hypothesizes, the shear plane is within the diffuse layer, investigations done by Lyklema,6
Smith7 or Sprycha8 show that the shear plane is located directly at the interface between
the Stern plane and the bulk electrolyte. Thus, the zeta potential can be used as a direct
measure of the Stern potential depending on the measurement technique, compare.9
According to the manifold of influencing factors, the zeta potential of cement and
mineral additions given in the literature vary over a wide range. A detailed overview is
available in.10 In the subsequent investigations, the effect of pore solution on zeta potential
and superplasticizer adsorption is discussed in detail.
Mixture proportions
Three Portland cements (C(a) - C(c)) as well as quartz flour (Q) and ground limestone
(L) as mineral additions were used in the investigations. The density of the materials was
determined by helium pycnometry and the surface area by nitrogen adsorption, Table 1.
The mixes were prepared with a commercial polycarboxylate ether based superplasticizer
(SP) with 35% solids in aqueous solution and a number average molecular weight of
69.000 g/mol.
Three kinds of pastes were prepared for the investigations.
L, Q, C(a) using 100 l/m3 (0.1 yd3/yd3) of water and 100 l/m3 (0.1 yd3/yd3) of ground
limestone (L), quartz flour (Q) or cement (C)
L*, Q* using 100 l/m3 (0.1 yd3/yd3) of artificial pore solution and 100 l/m3 (0.1 yd3/yd3)
of ground limestone (L) or quartz flour (Q)
L.C(a), Q.C(a) using 100 l/m3 (0.1 yd3/yd3) of water, 50 l/m3 (0.05 yd3/yd3) of ground
limestone (L) or quartz flour (Q) and 50 l/m3 (0.05 yd3/yd3) of cement (C)
The volumetric water-to-powder ratio Vw/Vp of all pastes was 1.0. The superplasticizer
dosage was kept constant at 2.9 mg per cm3 of solids (1.710-3 oz per in3 of solids). The
composition of the artificial pore solution was equivalent to the pore solution of paste
containing 100 l/m3 (0.1 yd3/yd3) of water, 50 l/m3 (0.05 yd3/yd3) of cement C(a) and 50 l/
m3 (0.05 yd3/yd3) of ground limestone (L), Table 2.

Adsorption257
Zeta Potential
The zeta potential was determined using electro-acoustic spectroscopy (Quantachrome
Dispersion Technology, Electroacoustic Spectrometer, DT 1200) 15 min after water addition. Electro-acoustic spectroscopy allows measurements of undiluted pastes with w/cratios in a range of 0.3 to 0.8.
For the determination of the zeta potential a relative motion between the charged particles and the surrounding fluid is necessary. As soon as the particle or the surrounding fluid
is moving, a part of the diffuse layer is slipping off. The higher the velocity, the larger is the
part of the diffuse layer that is slipped off. As a result, the particle appears no longer electrically neutral. The electrical potential at the shear plane characterizes the zeta potential.
For the electroacoustic zeta potential measurement, a high frequency acoustic wave (
1 MHz) induces an oscillating motion of particles and ions. Due to the lower inertia, there
is a larger movement of the ions in the double layer. This generates dipoles, creating a
macroscopic detectable electrical field. Due to the high ion concentration of the pore solution of cement-based suspensions the measured electroacoustic signal TVI (total vibration
current) contains a signal contribution from the colloidal particles CVI (colloid vibration
current) and a signal contribution from the ionic background IVI (ion vibration current).
Therefore, for the determination of the CVI, two measurements have to be performed: a) a
separate measurement of the extracted pore solution for the determination of the IVI and b)
a measurement of the suspension for the determination of the TVI.
De-ionized water was used for the preparation of suspensions of cement and mineral
additions. Water and solids were manually mixed for 3 min. 15 min after water addition
pore solution was extracted from the pastes by means of a vacuum pump. Subsequently the
IVI of the pore solution (15 ml / 0.5 fl oz) was determined. The determination of the TVI of
a separately prepared suspension (500 ml / 16.7 fl oz) was again carried out 15 min after
water addition. To prevent particle sedimentation the suspension was gently stirred during
the measurement.
Ion concentration of pore solution and Superplasticizer Adsorption
In addition, 500 ml (16.7 fl oz) of suspensions were prepared to determine pH-value and
ion concentration of pore solution as well as superplasticizer adsorption. Pore solution was
extracted from the pastes 15 min after water addition by means of a cylindrical pressure
device, see.1 The pH-value was determined by potentiostatic titration. Furthermore, the
concentration of anions and cations was determined using ion chromatography and inductively coupled plasma optical emission spectrometry (ICP-OES) respectively.
The total organic carbon (TOC) content of the pore solution and the superplasticizer
solution were determined by high-temperature oxidation of the organic ingredients. The
amount of adsorbed superplasticizer was calculated as the difference between the TOC of
the added superplasticizer solution and the pore solution of the mortar.
Zeta potential of quartz, limestone and cement - Effect of pore solution
composition
The zeta potential of mineral additions, cement and mixtures of mineral additions and
cement dispersed in water or artificial pore solution is shown in Figure 1. For quartz in
258SP-302-19
Figure 1 Zeta potential of quartz, limestone and cement

water (Q) the zeta potential is strongly negative at -52.8 mV. As there are almost no ions
dissolved from the SiO2particles, the zeta potential of quartz-water suspensions represents
the charge conditions at the surface. The negative characteristic of the zeta potential is
caused by the deprotonation of the hydroxyl groups >SiOH0 + OH >SiO + H2O.
In contrast, for limestone in water (L) the zeta potential is positive at + 17.3 mV. The
positive zeta potential is primarily attributed to the adsorption of monovalent hydrogen
carbonate anions as well as divalent calcium cations, which are the predominantly dissolved
ion species in pore solution of limestone suspensions at neutral pH-range. While the
adsorption of monovalent hydrogen carbonate ions induces a neutralization of the cationic
calcium sites of the surface (>Ca-HCO30), the adsorption of divalent calcium ions causes
a positive charge reversal of the anionic carbonate surface sites (>CO3-Ca+). In total, this
results in a larger number of positive charges, i.e. a positive zeta potential.
The zeta potential of the investigated cement (C(a)) and cement-addition mixtures
(Q.C(a), L.C(a-c)) varies between -4.6 and +2.1. Thus, the absolute value of the zeta potential of the cement-based suspensions is significantly lower than for quartz or limestone
dispersed in water, Figure 1. The main reason for the decrease is the high concentration of
ions in the pore solution of cement-based suspensions. Immediately after water addition,
ions are dissolved from the readily soluble compounds of the cement. The adsorption of
counterions from the solution to the Stern plane causes a charge equalization, which is
reflected in a significantly reduction of the absolute zeta potential value.

Adsorption259
Figure 2 Effect of Ca/SO4-ratio on zeta potential of cement

and mineral additions
To illustrate the mechanism of counter ion adsorption, the mineral additions were
dispersed in an artificial pore solution. The composition of the artificial pore solution is
equivalent to the pore solution of the limestone-cement suspension L.C(a), Table 2. It was
apparent that the absolute value of zeta potential of quartz and limestone dispersed in
the artificial pore solution (Q* and L*) significantly decreases to -17.5 mV and -6.3 mV,
respectively, Figure 1. Analogous to the cement-based suspensions, the ions dissolved in
the artificial pore solution adsorb to the Stern layer of quartz or limestone, resulting in a
charge equalization or even in a charge reversal, like in the case of limestone. The charge
reversal of the limestone surface is attributed to two mechanisms: a) With increasing
pH-value (increasing OH-concentration) the number of negative surface sites increases
(>CaO, >CO3) and b) the divalent sulfate anions (SO42), which are dissolved in high
concentration in the artificial pore solution, adsorb onto the remaining cationic surface
sites. As a result, the zeta potential of the limestone particles in artificial pore solution is
similar to the zeta potential of cement and cement-addition mixtures.
In the case of quartz flour, again the high pH of the artificial pore solution causes an
increase of negative surface sites (SiO). Additionally, due to the adsorption of the divalent calcium ions, dissolved in the artificial pore solution as well, the number of negative charges decreases, resulting in a substantial decrease of the zeta potential absolute
value. However, owing to the low solubility of calcium in water, the number of calcium
ions dissolved in the artificial pore solution is limited. By contrast, in a real cement-based
260SP-302-19
suspension, calcium ions can be dissolved constantly, causing a further reduction of the
zeta potential, compare Q.C(a).
According to the previous results and discussions, the ion concentration of the pore solution affects the zeta potential significantly. In principle all ion species in pore solution
are able to adsorb to the Stern layer. However, the adsorption capacity is controlled by
sign of charge, charge density (valence and ion size) and the size of the ionic hydration
shell. Various investigations of ion adsorption onto mineral surfaces demonstrate that the
adsorption capacity increases with increasing valence of the ions. Thus, with regard to ions
dissolved in cement-based suspensions, divalent calcium cations as well as divalent sulfate
anions can be classified as potential determining ions.
For this reason the Ca/SO4-ratio was used to describe the effect of ion concentration
on zeta potential quantitatively. The zeta potential of the investigated mineral additions,
cements and addition-cement mixtures depending on the Ca/SO4-ratio of the pore solution
is shown in Figure 2. It is apparent that the zeta potential increases with increasing Ca/
SO4-ratio. A Ca/SO4-ratio of about 0.4 to 0.6 characterizes the point of zero charge (pzc).
Another important conclusion can be derived from the interrelation between Ca/
SO4-ratio and zeta potential z. Interestingly, the mineral additions dispersed in water or
artificial pore solution (Q*, L*, L) as well as the cement-based suspensions (C(a), Q.C(a),
L.C(a,b,c)) are within the same - almost material non-specific - functional correlation. For
this reason, it seems reasonable to assume that all particles dispersed in cement-based
suspensions exhibit comparable low Stern potentials. The sign and value of the Stern
potential is primarily controlled by the adsorption of inorganic ions and can be described
by the Ca/SO4-ratio.
Superplasticizer adsorption onto quartz, limestone and cement Effect of
zeta potential
From the point of view of thermodynamics, the adsorption of superplasticizer polymers
onto a particle surface is associated with a heat release (enthalpy loss) and/or an entropy
increase of the total system. An entropy increase results for example from a release of
adsorbed ions or water molecules from the surface during polymer adsorption. On the other
hand, enthalpic adsorption is caused by attractive electrostatic as well as van der Waals
interactions.
Adsorption due to electrostatic forces requires both a charge at the surface and a polymer
bearing dissociated ionic groups. Polycarboxylate superplasticizers are polyanions
containing carboxylate groups COO. The adsorption of the negatively charged carboxylate groups can occur either directly onto cationic surface sites or indirectly onto cations,
adsorbed to the Stern layer. In the latter case, owing the high charge density, mainly calcium
ions are suitable as adsorption sites for the carboxylate groups. Thus, provided a sufficient
high calcium concentration, superplasticizer adsorption is possible even in domains of
initial negative surface charge.
The electrostatically caused adsorption of superplasticizer polymers should therefore
be directly depending on the charge conditions of the particle surface. For this reason,
the effect of zeta potential on the superplasticizer adsorption was investigated. Therefore
suspensions of water or artificial pore solution, superplasticizer and quartz flour, limestone
flour, cement or mixtures of quartz, limestone and cement were prepared. The superplasti-

Adsorption261
Figure 3 Effect of zeta potential on superplasticizer

adsorption (Note: 1mg/m2 = 3.310-6 oz/ft2)
cizer content of all suspensions was kept constant at 2.9 mgpol/cm3P (in relation to the solid
content = 1.710-3 oz per in3 of solids). The fraction of adsorbed superplasticizer molecules
was determined 15 min after addition of water and superplasticizer. After that time the
cement hydration changes from the dissolution period to the induction period, characterized by a slowdown of the reaction kinetics.10,11 Thus almost equilibrium conditions of
superplasticizer adsorption and desorption can be assumed.
The effect of zeta potential on superplasticizer adsorption is shown in Figure 3. It is
apparent that for strong negative zeta potentials < 15 mV only minor polymer adsorption of about 0.1 mgPol/m2P occurs (Note: 1mg/m2 = 3.310-6 oz/ft2). At higher zeta potentials between -6.3 and +17.3 mV the superplasticizer adsorption increases by trend with
increasing zeta potential from 0.39 to 0.56 mgpol/m2p. However, not every influence on
superplasticizer adsorption is captured by this trend. In spite of a higher zeta potential at
+2.1 mV for the limestone-cement mixture L.C(c) compared to the mixture L.C(b) at -2.3
mV, there is a lower superplasticizer adsorption for the mixture L.C(c). This is mainly
attributed to lower hydration kinetics of cement C during the first 15 min after water
addition. Thus, measurements using a heat flow calorimeter show a significantly lower
maximum heat flow for the suspension L.C(c) at 1.3 mW/g than for the suspension L.C(b)
at 8.1 mW/g, compare.10
262SP-302-19
Figure 4 Effect of cement and limestone content on superplasticizer adsorption (Note: 1mg/m2 = 3.310-6 oz/ft2)
Superplasticizer adsorption in limestone-cement suspensions Effect of
cement and limestone content
The previous results show clearly that superplasticizer polymers adsorb onto cement
as well as onto mineral addition particle surfaces. At this point, the question of selective
polymer adsorption arises: In which amount does the polymer adsorb onto the different
particle surfaces in suspensions containing both cement and mineral addition particles?
In the case of zeta potential, it was shown that, due to the high concentration of divalent
ions in pore solution and adsorption of these ions to the Stern layer, the charge conditions of different materials in a cement-based suspension are equalized. Furthermore it
was discussed that superplasticizer polymers adsorb onto calcium ions of the Stern layer.
Thus, it seems reasonable to assume that despite different material properties of cement
and mineral additions, there is an equalization of the superplasticizer adsorption characteristics, as soon as these materials are present together in a cement-based suspension.
To clarify this question, adsorption measurements of limestone-cement suspensions
L.C(a) were carried out at different cement contents (0,0/12,5/25,0/37,5/50,0 Vol.%).
Owing to the difference in specific surface of cement (Sv = 3,3 m2/cm3 = 582 ft2/in3)
and limestone flour (Sv = 4,1 m2/cm3 = 723 ft2/in3) the total surface area of the solids is
decreasing slightly with increasing cement content from 4.1 m2/cm3 to 3.7 m2/cm3 (723
ft2/in3 to 653 ft2/in3). The volumetric water/powder ratio Vw /Vp of the suspensions was
kept constant at 1.0. Furthermore, the composition of the pore solution was kept constant
at the same level as the pore solution at a cement content of 50 Vol.%. To compensate the
decreasing ion concentration at decreasing cement contents, water was substituted by artificial pore solution according to Table 2 from 0% at a cement content of 50 Vol.% to 100% at

Adsorption263
a cement content of 0 Vol.%. In all cases, the dosage of superplasticizer was constant at 2.9
mgpol/cm3p (mass of polymer mpol in relation to total volume of cement and limestone Vp).
Figure 4 shows the amount of adsorbed superplasticizer polymers in dependence of the
cement content. The results indicate that the adsorbed amount of superplasticizer in relation to the specific surface of the solids is almost constant, independent of the cement
content. Even for the limestone suspension without cement but artificial pore solution is
the amount of adsorbed polymers at the same level as for the suspension containing 50
Vol.% of cement. According to these results, provided a constant composition of the pore
solution, superplasticizer adsorbs in the same order onto cement and limestone. Thus, the
adsorption characteristics of superplasticizer of the investigated limestone-cement suspensions, is preponderantly controlled by the composition of the pore solution rather than the
kind of solids dispersed in the suspension.
CONCLUSIONS
The effect of pore solution composition on zeta potential and superplasticizer adsorption has been investigated experimentally. The investigations were conducted on highly
concentrated suspensions, containing quartz flour, limestone flour, cement and combinations of these materials. Furthermore cement-limestone suspensions with different types of
cements and a varying ratio of cement to limestone were investigated.
The results show that the zeta potential of cement and mineral additions is significantly
determined by the composition of pore solution. It seems reasonable to assume that all
particles dispersed in cement-based suspensions exhibit comparable low Stern potentials.
The sign and value of the potential is primarily controlled by the adsorption of inorganic
ions. Moreover, the zeta potential cement-based suspensions can be characterized by the
ratio of calcium to sulfate dissolved in the pore solution.
Furthermore it was shown that the superplasticizer adsorption is affected by zeta potential. At higher more positive zeta potentials the superplasticizer molecules are more likely
adsorbed onto the solid surfaces. Moreover, the superplasticizer adsorption in limestonecement suspensions is predominantly controlled by the composition of pore solution rather
than the ratio of cement to limestone flour. If the ion concentration of the pore solution
is artificially kept constant the polymer adsorption is almost constant independent of the
cement to limestone ratio in the suspension.
AUTHOR BIOS
Dirk Lowke is Senior Researcher and Head of the Working Group Concrete Technology
at the Centre for Building Materials (cbm), Technische Universitt Mnchen in Munich
(Germany). His major research fields are mixing, workability and rheology of fresh
concrete as well as durability and time dependent deformation properties of hardened
concrete.
Christoph Gehlens work focusses on the description and prediction of the service life
of mineral and metallic building materials in dependence of exposure conditions. He
graduated at RWTH Aachen University. A doctorate followed in 2000. Together with two
partners, he then founded an international engineering consultancy. After accepting an
264SP-302-19
appointment at Stuttgart University, he took up his position at the TUM in autumn 2008.
Hes involved in numerous national and international boards and committees.
REFERENCES
1. Lowke, D., Segregation resistance and robustness of Self-Compacting Concrete. Optimization based on modelling interparticle interactions in cement based suspensions. (in
German). PhD-Thesis, 2013
2. Foxall, T.; Peterson, G. C.; Rendall, H. M.; and Smith, A. L., Charge determination at
calcium salt/aqueous solution interface. Journal of the Chemical Society, Faraday Transactions 1, Physical Chemistry in Condensed Phases, V. 75, 1979, pp. 1034-1039.
3. James, R.O.; Healy, T.W.: Adsorption of hydrolyzable metal ions at the oxide-water
interface. II. Charge reversal of SiO2 and TiO2 colloids by adsorbed Co(II), La(III), and
Th(IV) as model systems. Journal of Colloid and Interface Science 40(1972)1, pp.53-64
4. Koetz, J., and Kosmella, S., Polyelectrolytes and Nanoparticles. Berlin, Heidelberg:
Springer, 2007
5. Kumar, A.; Bishnoi, S.; and Scrivener, K. L., Modelling early age hydration kinetics
of alite, Cement and Concrete Research, V. 42, No. 7, 2012, pp. 903-918. doi: 10.1016/j.
6. Lyklema, J.: Water at Interfaces: A Colloid-Chemical Approach. Journal of Colloid
and Interface Science 58(1977)2, pp.242-250
7. Plank, J.; Sieber, R.; Schrfl, C.; Lesti, M.; and Gruber, M., Interactions between
polycarboxylate superplasticizers, cement and microsilica in ultra-high strength concrete.
In: 8th International Symposium on Utilization of High-Strength and High-Performance
Concrete, Tokyo (Japan), 2008, pp.129-134.
8. Smith, A.L.: Electrokinetics of the Oxide-Solution Interface. Journal of Colloidal and
Interface Science. 55(1976)3, pp.525-530
9. Sprycha, R., and Matijevi, E., Electrokinetics of Uniform Colloidal Dispersions
of Chromium Hydroxide, Langmuir, V. 5, No. 2, 1989, pp. 479-485. doi: 10.1021/
la00086a033
10. Stern, O., Zur Theorie der elektrolytischen Doppelschicht. Zeitschrift fr Elektrochemie und angewandte physikalische Chemie, Z. Elektrochem., V. 30, 1924, p. 508
11. Van Capellen, P.; Charlet, L.; Stumm, W.; and Wersin, P., A surface complexation
model of the carbonate mineral-aqueous solution interface, Geochimica et Cosmochimica
Acta, V. 57, No. 15, 1993, pp. 3505-3518. doi: 10.1016/0016-7037(93)90135-J
SP-302-20
Optimization of the Structural Parameters

and Properties of PCE Based on the
Length of Grafted Side Chain
by Zi-Ming Wang, Zi-Chen Lu, and Xiao Liu
A series of polycarboxylic ether (PCE) superplasticizers were copolymerized through
acrylic acid (AA) and isobutylene polyethylene glycol (IPEG) with different molecular
weight (500, 1000, 1500, 2000, 2400 and 2700). The molecular weight, molecular weight
distribution and reaction ratio were measured by gel permeation chromatography (GPC).
The initial fluidity and flow retaining ability were evaluated through mini slump test at the
same dosage (by mole and by weight). The results indicated that the dispersing capability
of single PCE molecule firstly improved with the increase of backbone length, and then
decreased after reaching a critical value. Both the appropriate backbone length and side
chain density of PCE were affected significantly by the side chain length. In general, PCE
with longer side chain, shorter backbone length and lower side chain density are appropriate to achieve better comprehensive properties. Finally, regression equations were
proposed to predict the suitable backbone length and proper side chain density of PCE
based on the length of grafted side chain.
Keywords: PCE; IPEG; backbone; side chain; structural parameters.
INTRODUCTION
The PCE molecule has comb structure, which contains negative charged backbone and
grafted side chains.1 One outstanding characteristic of comb like PCE is that its chemical
structure could be designed and modified according to diverse requirements of concrete
constructions. PCEs are characterized by different molecular weight and distribution, backbone length, grafted side chain length, type and density of functional groups. Change of
any of the above mentioned structural parameters could alter the final molecular conformation and then affect the properties. Tremendous efforts were paid to study the relationship
between structure and properties of PCE, but many of the conclusions are still incomplete
and limited to the specific experiments. Yamada2 reported that PCE with longer side chain,
lower backbone polymerization and higher contents of sulfonic groups showed higher
dispersing power. Nawa3 studied the effect of chemical structure on steric stabilization of
PCE and found that longer side chain could provide better dispersing ability with small
amount of adsorption. Winnefeld4 concluded that the decreasing density of the side chains
265
266SP-302-20
Table 1 Chemical and mineral compositions of standard cement (%, by

weight)
SiO2
22.93
Al2O3
4.29
Fe2O3
2.89
CaO
66.23
MgO
1.92
SO3
0.35
Na2Oeq
0.70
f-CaO
0.64
C3S
58.78
C2S
21.38
C3A
6.49
C4AF
8.77
enhanced workability while the length of side chain and its molecular weight had only a
minor effect. Renkas5 found that maleic-based superplasticizers with longer backbones and
side chains are more efficient. On the other hand, the acrylic acid derivatives, with shorter
backbones and side chains and without carboxylic groups, appeared less efficient.
All cited work discussed single structural parameter and the induced properties without
considering the coordination and impact of these parameters to each other. It is commonly
believed that all of these structural parameters were actually integrated together to decide
the final property of the PCE. Hence, it is necessary to study how one structural parameter
change with others to obtain the satisfied overall properties.
Materials
For the synthesis of series of PCE acrylic acid (AA) and isobutylene polyethylene glycol
(IPEG) with different molecular weight (500, 1000, 1500, 2000, 2400 and 2700) were used
as copolymerization monomers, mercaptoacetic acid as chain transfer agent and ammonium
persulfate (APS) as initiator were used as received. Ordinary portland cement complying
with the Chinese National Standard GB8076-1997 was used to prepare the cement pastes.
Chemical and mineral compositions of this cement clinker are presented in Table1.
Preparation and characterization of the synthesized PCEsSeries of PCEs with
different side chain and backbone length were prepared by free radical polymerization
in a three necked glass flask equipped with a mechanical stirrer, dosing units for both the
monomers and the initiator. Firstly macromer and de-ionized water were charged into the
flask and heated to 60C[140F],then the mixed solution of mercaptoacetic acid with AA
and the APS solution were separately added dropwise into the flask at a constant rate in 180
minutes. After that the flask was kept at 60C [140F] for 120 minutes to complete the reaction. The obtained polymer solution (40 wt%) was cooled down to the room temperature
and pH adjusted to 6~7 by 0.1 mol/L NaOH solution. Recipes for preparation of PCEs are
shown in Table 2 and Table 3. The molar ratio of AA to macromer ([A/E]) was gradually
increased when the other reactant amount were fixed. Through the mini-slump test, the
[A/E] of synthesized PCE with the best property was ascertained. With [A/E] value fixed,
the main chain length of PCE was changed through gradually increasing the molar ratio of
mercaptoacetic acid to macromer ([T/E]) value. In this case, series of PCEs with different
main chain length and side chain density were synthesized with different macromers.
Gel permeation chromatographyGel permeation chromatography (GPC) equipped with
three different detectors, namely refractive index detector (RI), laser light scattering detector
(LS) and viscosity detector (VIS) was used to measure the molecular weight (Mw), molecular
Optimization of the Structural Parameters and Properties of PCE Based on

the Length of Grafted Side Chain 267
Table 2 Recipes for preparation of PCEs with gradual decreasing side chain
density from different macromers
Macromers
IPEG-500
IPEG-1000
IPEG-1500
IPEG-2000
IPEG-2400
IPEG-2700
1.0
1.0
1.0
1.0
1.0
1.0
1.5
1.5
1.5
1.5
1.5
1.5
2.0
2.0
2.0
2.0
2.0
2.0
2.5
2.5
2.5
2.5
2.5
2.5
A/E (mole/mole)
3.0
3.5
4.2
3.0
3.5
4.2
3.0
3.5
4.2
3.0
3.5
4.2
3.0
3.5
4.2
3.0
3.5
4.2
4.5
4.5
4.5
4.5
4.5
4.5
5.0
5.0
5.0
5.0
5.0
5.0
----6.0
6.0
6.0
6.0
--------7.0
7.0
Table 3 Recipes for preparation of PCE with gradual decreasing main chain
length from different macromers
Macromers
IPEG-500
IPEG-1000
IPEG-1500
IPEG-2000
IPEG-2400
IPEG-2700
0.05
0.05
0.05
0.05
0.10
0.05
0.08
0.09
0.08
0.09
0.17
0.10
T/E (mole/mole)
0.13
0.17
0.13
0.17
0.13
0.17
0.11
0.15
0.25
0.30
0.13
0.20
0.25
--0.25
0.22
0.40
0.25
0.30
--0.30
0.26
0.50
0.30
-------0.60
---
A/E (mole/mole)
1.5
3.0
3.0
3.5
4.2
5.0
weight distribution (MWI) and the reaction ratio of the synthesized PCE.6 For all polymers,
The eluent with 0.05 wt% sodium azide was prepared by dissolving 1g sodium azide in 2kg DI
water. The PCE solutions with concentration 4mg/mL were prepared 1 day earlier to allow the
PCE dissolve completely. After that, the solutions were filtered with 0.45m filtering membrane
before injection. The test time for every sample was 900s, and the flow velocity was 1mL/min.
Mini-slump testThe mini-slump test was conducted according to the Chinese National
Standard GB/T 8077-2000 to evaluate the fluidity of fresh cement pastes (FCP). The
fluidity of the FCP was represented by the spread flow of the mini- slump test. The spread
diameter was recorded as the average of two perpendicularly crossing diameters.
To evaluate dispersing property of the prepared PCE precisely, the unreacted macromers
were calculated based on the analysis of GPC chromatogram and then subtracted from
the dosage. Two dosing methods were employed to better understand the structure-property
relationship of PCE, namely the equivalent weight addition method (EWAM) and equimolar
addition method (EMAM). The dosage of pure PCE was fixed at 0.15% by weight of cement
in the case of EWAM, while the addition of PCE samples with the same side chain length
were calculated by ensuring the same mole (molecule number) in the case of EMAM. The
water to cement ratio of the cement pastes was fixed at 0.29 in both cases. For each cement
paste, the spread diameter was respectively recorded at 5min, 60min and 120 min after mixing
with water. The spread diameter at 5 min was defined as the initial fluidity and the values at
60 min and 120 min were used to evaluate the flow retaining ability of PCE.
PCE structure analysis and macromer reaction ratio
There are two problems affecting the precise evaluation of structure-property relationship. The first one is how to detect the true polymer content as the existence of unre-
268SP-302-20
Table 4 Practical and theoretical molecular weight of synthesized PCEs

[A/E] Value
1.0:1
2.0:1
2.5:1
3.0:1
3.5:1
4.2:1
4.5:1
5.0:1
6.0:1
7.0:1
Theoretical M
153900
105600
91800
81400
73400
64700
61700
57300
50400
45200
Practical Mw
67300
72700
105900
73500
74000
52300
61700
53700
58800
45400
Polydispersity
1.330
1.254
1.397
1.303
1.303
1.168
1.265
1.273
1.215
1.186
Effective polymer content/%

56.0
82.7
95.6
94.8
95.8
97.3
99.7
98.9
99.1
100.0
Fig.1 Cement paste fluidity by EWAM

acted macromer will lead to the unequal adding amount, the other is the difference of
real polymer structure and the supposed one. As we know, the free radical polymerization is hard to precisely control the polymer structure, and it is difficult to determine the
precise structure of polymer presently. GPC provides us one helpful method to identify the
desired structure of synthesized PCEs when only one structure parameter changed regularly while keeping others fixed. The real molecule weight was detected through GPC and
then compared with the theoretical molecular weight to ascertain whether the right structure was obtained as designed. We here consider macromer IPEG-2400 as the example
to discuss. Data are shown in Table 4. When the [A/E] value is over 2.5, the theoretical
molecular weight and the practical molecular weight fitted well, which means the satisfied
structure was obtained. However, the effective polymer content decreased sharply when
the [A/E] was less than 2.5, which meant that lots of macromer were not co-polymerized
into the PCE molecule, resulting in the difference of the theoretical Mw and the real Mw.
Effect of side chain density
The EWAM and EMAM were taken to evaluate the overall property of PCE and the single
PCE molecule property when the side chain density decreased gradually, as shown in Fig.1

Fig.2 Cement paste fluidity by EMAM
Fig.3 Cement paste fluidity loss rate by EWAM

to Fig.4. The fluidity loss rate is the ratio of fluidity loss value within t minutes to the initial
fluidity indicating the flow retaining ability of PCE.
The initial fluidity increased obviously with the decreasing side chain density while
keeping the backbone length constant (Fig.1), which complies the former researcher
work.2,7,8 At the same time the fluidity retaining ability weakened proportionally with the
decreases of side chain density (Fig.3) in the case of EWAM. This phenomenon can be
explained by the adsorption rate of PCE on the cement particles. The increase of anionic
COO- group in the backbone improved the adsorption amount of PCE on the cement particles, which was beneficial to the initial fluidity growth. Nevertheless, as more PCE polymer
was consumed at the beginning, less polymer were left to keep the fluidity. This effect
was more evident in the case of EMAM (Fig.2 and Fig.4), which reflected the dispersing
and fluidity retaining ability of single PCE molecule. Integration of considering the initial
fluidity and retaining ability of PCE comprehensively, the proper [A/E] value is in the
range of 3.0 to 4.2 when using IPEG 2400 as macromer
270SP-302-20
Fig.4 Fluidity loss rate by EMAM

Table 5 Molecular structural parameters of PCE with different backbone length
Macromer
IPEG-500
IPEG-1500
IPEG-2400
[T/E]
0.05
0.10
0.17
0.25
0.30
0.05
0.10
0.17
0.25
0.30
0.05
0.10
0.17
0.25
0.30
Mw
36600
26500
15800
10700
<10000
80300
57600
23000
14000
13200
258800
127700
83700
45300
31600
Polydispersity
1.696
1.825
1.632
1.538
--1.619
1.776
2.225
1.402
1.728
1.603
1.386
1.294
1.128
1.143
Active
polymer
content/%
100
100
100
100
100
86.5
89.9
80.6
75.5
72.2
97.8
93.8
94.7
95.3
88.4
Practical
Acid/ether
ratio
1.5:1
1.5:1
1.5:1
1.5:1
1.5:1
3.5:1
3.3:1
3.7:1
4.0:1
4.2:1
4.3:1
4.5:1
4.4:1
4.4:1
4.7:1
Side chain
density/%*
66.7
66.7
66.7
66.7
66.7
28.6
30.3
27.0
25.0
23.8
23.3
22.2
22.7
22.7
21.3
Number of
C-C bond on
the backbone
302
216
130
90
414
284
122
80
72
1012
516
332
180
132
* the calculation method of side chain density is the ratio of macromer mole number to the mole number of acrylic acid. The
following side chain density is also calculated like this.
Effect of main chain length

Three series PCEs were synthesized by using different macromers (500, 1500 and 2400)
respectively. The structural parameters of these PCE were shown in Table 5.
The results of initial fluidity and fluidity retaining ability were shown in Fig.5 and
Fig.6 by equivalent weight adding method(EWAM). As shown in Fig 5 and Fig.6, the
initial fluidity and retaining ability of PCE changed remarkably with the backbone length
according the length of side chain. In the case of longer side chain (IPEG 2400), a proper
backbone length range existed (T/E = 0.10~0.25). When using the shorter side chain (IPEG

Fig.5 Initial fluidity of cement paste with different

main chain length and different macromer PCEs
Fig.6 Fluidity loss rate in 2h of cement paste

with different main chain length and different
macromer PCEs
500), the initial fluidity and fluidity retaining ability decreased with the shortening of the
backbone length. PCE with medium side chain length also indicated a preferable backbone
length range (T/E = 0.10~0.17). In addition, the longer side chain was helpful to enhance
the fluidity retaining ability.
To further observe the properties of PCE with longer side chain (IPEG-2400), both
EWAM and EMAM were conducted and the results were shown in Fig.7 and Fig.8 The
dosage of the two methods were listed in Table 6. As shown in Fig.7, there was lower
initial fluidity even though the adding amount was more than 21.7410-6 mole, when the
[T/E] value is more than 0.30, which proved that PCE with too shorter backbone length
has little dispersing ability. It is interesting to notice that the dispersing property of PCE
was strongly decreased by controlling the same mole amount(EMAM) when the [T/E]
value were 0.25 and 0.30 (Fig.8), but more PCE molecules could compensated the weaker
272SP-302-20
Fig.7 Cement paste fluidity by EWAM (IPEG2400 PCE)
Fig.8 Cement paste fluidity by EMAM (IPEG2400 PCE)

dispersing ability to an overall good level (Fig.7), which implied the fact that overall property not only depends on the single molecule property, the molecule number can also make
an unneglectable contribution to the final property. This can be further consolidated by
the results of PCE with [T/E] in the range of 0.10 to 0.17, the property of single molecule
was excellent, but as the higher molecule weight lead to the decreasing molecule number
when the EMAM applied, so property is not as good as that of PCE with [T/E] at 0.25 by
EWAM .
Effect of side chain length
Among all these synthesized PCEs, the structure of gradually changed side chain density
and main chain length were obtained with the same side chain length. In order to better
understand the effects of the side chain length on the property of PCE, one PCE sample
with the best property was picked out among the PCEs having the same side chain length.

Table 6 correlation table of different mass and mole amount by EWAM and
EMAM (IPEG 2400)
[T/E] of
IPEG2400
0.05
0.10
0.17
0.25
0.30
0.40
Mass-based Addition
Mass amount /g Mole amount10-6 /mol
0.45
2.79
0.45
4.88
0.45
6.96
0.45
11.21
0.45
16.28
0.45
21.74
Mole-based Addition
Mass amount /g Mole amount10-6 /mol
0.87
5.37
0.49
5.37
0.35
5.37
0.22
5.37
0.15
5.37
0.14
5.37
Table 7 Optimized structural parameters of PCEs synthesized from different

macromers
Samples
M500
M1000
M1500
M2000
M2400
M2700
Concentration
44.1%
36.0%
35.5%
39.1%
38.7%
39.5%
[A/E]
1.5:1
3.0:1
3.0:1
3.5:1
4.2:1
5.0:1
[T/E]
0.05
0.13
0.13
0.15
0.25
0.30
Side chain
density/%
66.7
33.3
33.3
28.6
23.8
20.0
C-C bond
number in
main chain
302
285
234
183
175
173
Mw
36600
43300
50200
46000
45300
44200
Effective
content/%
100.0
100.0
91.5
92.3
95.3
100.0
The initial fluidity and retaining ability of cement pastes with these PCEs were compared
and analyzed. It was found that there were proper backbone length and side chain densities for different macromer to obtain satisfactory overall properties. The optimized structure parameters from different macromers were shown in Table 7. It was clearly shown
that the proper backbone length and side chain density are closely related to the length of
side chain. Furthermore, it was interesting to notice that the Mw of all these samples were
located in the range from 35 000 to 50 000, which implies that the structure parameters
should be correlated and mutually restricted. The relationship of side chain density with
the side chain length (average EO polymerization number) was plotted in Fig.9, and the
trend of proper backbone length (by the average C-C bond number) with the side chain
length was showed in Fig.10. It is obviously indicated that the optimum structure of PCE
is somewhat depending on the side chain length, the PCE with longer side chain should has
lower side chain density and shorter main chain length to achieve excellent overall properties, and vice versa. The schematic diagrams for the PCEs with optimized structure were
illustrated as Fig.11.
The fitting equations concerning the proper side chain density and main chain length
with side chain length were obtained respectively. The side chain density (y1) can be
expressed as

y1 = 244.7n0.594 100%
The main chain length can be expressed as number of C-C bond (y2):
(1)
274SP-302-20
Fig.9 Curve of side chain density to macromer

molecular weight
Fig.10 Curve of main chain C-C bond number to

macromer molecular weight

y2 = 350.1e0.013n (2)
where n is the polymerization degree of the macromer.

Based on the fitting equations, the molecular weight of PCE can be calculated as follow.

M = (a MAA + MPEG) b (3)
MAA is the molecular weight of acrylic acid, MPEG is the molecular weight of the macromer,
a = 1/y1, b = y2/2(a +1) = y2 y1/2(1 + y1).
The schematic diagram of the synthesized PCEs and the above parameters are shown in
Fig.12. So once the side chain length is fixed, the suitable molecular weight of PCE can
be predicted.

Fig.11 Structure diagram of PCE with best property synthesized from IPEG-500 to IPEG2700 (a) IPEG-500; (b) IPEG-1000; (c) IPEG-1500; (d) IPEG-2000; (e) IPEG-2400; (f)
IPEG-2700
CONCLUSIONS
According to the results of initial dispersing property and fluidity retaining ability of
PCEs with different structural parameters by using two different evaluation methods, the
effects of side chain length, side chain density and main chain length on the property of
PCEs were analyzed and discussed. Following conclusions can be drawn.
276SP-302-20
Fig.12 The schematic diagram of the synthesized PCEs

1. GPC is an effective method to identify the desired PCE structure through comparing
the measured molecular weight and the theoretical molecular weight. In addition, higher
side chain density will help to increase the fluidity retaining ability, but sacrifice some
initial fluidity.
2. The main chain length has a remarkable influence on the properties of PCE, but the
influence trends changed differently according the length of side chain. In the case of longer
side chain (IPEG 2400), a proper backbone length range existed owing to the outstanding
single molecule property or the synergistic effect of large number of PCE molecules. When
using the shorter side chain (IPEG 500), the initial fluidity and retaining ability decreased
with the shortening of the backbone length. The longer side chain was helpful to enhancing
the fluidity retaining ability.
3. The Mw of PCE with excellent property coincidently locates in the range from 35000
to 50000. The proper main chain length and side chain density of PCEs to achieve overall
outstanding properties are decided by the length of side chain (molecular weight of
macromer). PCE with longer side chain should has lower side chain density and shorter
main chain length to achieve excellent overall properties The relationship of these three
structural parameters was expressed by two regression equations. The molecular weight of
PCE can be predicted on the basis of the regression equations.
AUTHOR BIOS
Zi-Ming Wang is a professor at the College of Materials Science and Engineering,
Beijing University of Technology. He received his BS from Harbin Institute of Technology,
MS from China Building Materials Academy, PhD from Beijing University of Technology.
His research interests include high performance cementitous materials, chemical admixtures for concrete.

Zi-Chen Lu is a doctoral candidate at the Department of Civil Engineering, Tsinghua
University. He received his BS from Lanzhou Jiaotong University; MS from Beijing
University of Technology. His research interest is the cement and concrete additives.
Dr. Xiao Liu is a lecturer at the College of Materials Science and Engineering, Beijing
University of Technology. He received his BS, MS and PhD from Beijing University of
Chemical Technology. His research interest is the cement and concrete additives.
ACKNOWLEDGMENTS
The authors wish to express their gratitude and sincere appreciation to the National Natural
Science Foundation of China (Grant number: 51208012), and Project of China Railway
Engineering Materials Technology (Anhui) Ltd (Project number: 40009011201409) for
financing this research work
REFERENCES
1. Otha.A, Sugiyama, Uomoto.T. Stuty of dispersing effects of polycarboxylate-based
dispersant on fine particle. American Concrete Institute Special Publication. SP-195,
211-227
3. Nawa, T., Effect of chemical structure on steric stabilization of polycarboxylatebased superplasticizer, Journal of Advanced Concrete Technology, V. 4, No. 2, 2006, pp.
225-232. doi: 10.3151/jact.4.225
4. Winnefeld, F.; Becker, S.; Pakusch, J.; and Gtz, T., Effect of the molecular architecture of comb-shaped superplasticizers on their performance in cementitious systems,
5. E. J. Renkas. The effect of superplasticizers chemical structure on their efficiency in
cement pastes, Construction and Building Materials, V. 38, 2013, pp. 1204-1210.
6. Zichen, L. U., Wang Ziming, Lu Fang, Li Huiqun. Determination of molecular weight
and distribution of polycarboxylate superplasticizer by triple detection GPC. Advanced
Materials Research. 2013
7. Rongguo, Z.; Huiling, G.; and Jiaheng, L. etal., Effect of molecular structure on the
performance of polyacrylic acid superplasticizer. Journal of Wuhan University of Technology, Materials Science, V. 22, No. 2, 2007, pp. 245-249.
V. 336, No. 2, 2009, pp. 624-633. doi: 10.1016/j.jcis.2009.04.057
278SP-302-20
SP-302-21
Blended Antifreezing Admixture with

Extreme Freezing-Point
by Anatoly I. Vovk
In many countries there are regions with very cold winter climate. To provide for yearround building and construction works antifreezing admixtures must be used. When suitable technology and high performance antifreezing admixtures have been used, ready mix
concrete may be delivered and placed at temperature as low as -25 C. The transfer from
summer concrete mix to winter one implies the addition of optimal amount of antifreezing admixture. In principal, both separate superplasticizer and antifreezing admixture
dosing and usage of blended (combined) admixture is possible. In Russia the usage of
blended plasticizing-antifreezing admixtures is predominant historically.
The prime blended admixtures on the base of sodium formate didnt freeze down to -15
o
C and allowed secure concrete hardening with dosages about 10% of commercial product.
At present day such admixtures are unsuitable neither by the dosage level (danger of ASR)
nor by the freezing-point.
The method for elaboration of blended antifreezing admixture (BAFA) with extreme
freezing-point was developed. These admixtures allow building and construction works at
a temperature as low as -25 oC with dosages about 1-1.3% for PCE-based admixtures and
1.8-2.2% for PNS-based admixtures. It is important that these dosages ensure superplasticizing effect. Non-chloride non-plasticizing admixture with freezing-point below -40 oC
was developed also.
The kinetic of concrete strengthening at various temperature regimes has been thoroughly studied.
Keywords: antifreezing admixtures; blended admixtures; extreme freezing-point;
superplasticizer.
INTRODUCTION
Climate in many European countries (or North American regions) is often characterized
by steady low temperatures (below zero) in winter. In some countries (regions) most part
of the year may fall on negative temperatures. In order to continue construction works
and prevent ready-mix concrete from freezing (and concrete properties from deterioration)
antifreezing admixtures are frequently used.
279
280SP-302-21
Originally, inorganic salts were used as antifreezing admixtures at first calcium chloride, then calcium nitrate, sodium nitrate and mixed salts.1 With the evolution of specialists opinions on concrete durability and toughening of regulatory requirements to chemical
admixtures mild deicing salts (such as sodium formate) were introduced and predominantly applied.
However, requirements to the properties and effectiveness of antifreezing admixtures
vary from country to country. In Russia where temperatures below -20C are not uncommon
throughout almost all its territory one of the main requirements to the admixtures is their
non-freezing ability at such low temperatures.
Moreover, for historical reasons the practice of blended admixtures application per
plasticizer + antifreezing component pattern (including lignosulfonates-based) has been
generally accepted in Russia. Separate application of plasticizing and antifreezing admixtures was known for a long time,1 but only at the end of the XX century the first commercially available blended antifreezing admixture was developed. This admixture could be
stored and allowed concrete strengthening under temperatures as low as -15 C without
additional heating. The dosage of the admixture was appr. 3-3.5% by dry substance (appr.
9-10% by commercial product) and it was much lower than previously recommended
dosages of pure antifreezing salts (15-18% of the mass of mixing water1). The higher the
temperature the lower the dosage obviously and under the temperature from 0 to -5 the
admixture dosages were within 1-1.5% by dry substance.
This admixture and its subsequent analogues became so popular that now all Russian
national companies included them into their product range of blended antifreezing admixtures. Moreover, even transnational companies coming into the Russian market began to
offer this kind of products. Similar approach is used in China.2
The main difficulty in developing such blended antifreezing admixtures (BAFA) is
as follows: conventional SNF-based superplasticizers (as well as lignosulfonate-based
admixtures) are characterized by quite high freezing point (-5-6 depending on concentration). In order to impart low freezing point for a blended admixture one shall add much
antifreezing agents (which cryohydric points are reached at rather high concentrations only
(Table 1)). According to a detailed analysis of available BAFAs (Table 2), they either have
a very low concentration of superplasticizers (and thus their dosages into concrete must be
quite high) or are frozen under the temperatures slightly below zero.
The freezing point of BAFA solutions can be decreased by adding ethylene glycol
(this approach was used in2), but we consider this way inadmissible for possible alcohol
leaching from the concrete and vapours volatility at high temperatures. Another classic
path of lowering freezing temperature is the usage of calcium chloride but we rule out
the usage of any chlorides (even in combination with inhibitors) to avoid possible risk of
reinforcement corrosion.
Different types of plasticizing admixtures (SNF, PCE, lignosulfonates) were introduced
in blended antifreezing admixtures with extreme freezing point. New blended admixtures
characterized by low recommended dosages and allow building and construction works at
a temperature as low as -25 oC. Non-chloride non-plasticizing admixture with freezing-
Blended Antifreezing Admixture with Extreme Freezing-Point 281
Table 1 Freezing point for common antifreezing salts

Solution concentration, %
NaCl
-16.5
-19.4
-8.8
-
20
23
25
30
32
40
CaCl2
-17.6
-23.7
-29.0
-50.2
-14
+11.8
Freezing temperature,
NaNO2
K2CO3
-10.8
-8.9
-13.9
-11.2
-15.7
-13.0
-16.5*
-18.7
-14
-21.5
-6
-36.5
Ca(NO3)2
-7.2
-10.1
-11.9
-15.6
-17.1
-23.9
* - minimum temperature -19.6 corresponds to 28% concentration
Table 2 Composition and properties of commercially available blended

antifreezing admixtures
Guaranteed
Concentration
storage
of commercial
temperature,
product, %
Recommended
admixture
dosage, % by
commercial
product for
-20-25
up to10
(for -15 )
Country of
source
SP-base
Russia
SNF
-10 -15
32-35
SNF
-5
32-35
SNF
-15
371
SNF
-15
47
SNF
SMF
Germany
Switzerland
SP content, %
in solid
substance
in commercial product
17
5-6
75
24-26
4-5
15
5-6
6-7
32.5
13
30-38
2-2.5
~70
21-27
20
27
2.5 (for -15 oC)
65
18
SNF
-15
36
8-9
15
~5
SNF
-20
40
5.5
25-30
12-13
point below -40 oC was developed also; this admixture is compatible wholly with all types
of superplastisizers.
Cement. Russian industrial cements (analogs of CEM I 42.5N and CEM II/A-S 32.5R)
were used in the tests. The composition and characteristics thereof are described in Table 3.
Admixtures. All the tested admixtures are commercially available products. Physical and
che-mical properties of SNF and PCE-based blended antifreezing admixtures are described
in Table 4.
According to the fundamental laws of physical chemistry the solution to the problem of
developing BAFAs with extreme freezing point lies in selection of a system of antifreezing
agents with high solubility and compatibility at low temperatures.
282SP-302-21
Table 3 Industrial Portland cement composition and characteristics

(according CC)
Portland
Water
cement
demand,
(plant)
%
CEM I 42.5N
26.8
(Volsk)
CEM I 42. N
26.6
(Tula)
CEM I 42.5N
(Novoros24.5
siysk)
CEM II/A-S
32.5R
27.5
(Voskresensk)
Setting time
(hour-min)
Oxide composition, %
initial final SiO2 Al2O3 Fe2O3 CaO MgO R2O SO3
Mineralogical composition, %
C3S C2S C3A C4AF
3-00
4-18 22.15 4.53
4.56 65.82 1.3
0.77 2.55
64
15
4.2
14
3-10
4-00 21.21 5.07
3.31 65.29 2.36 0.56 2.74
61
15
11
4-05
5-09 21.03 4.37
4.16 64.81 0.78
5.0
14.0
2-45
3-30
no
no
no
data data data
no
data
n.a.
2.8
62.7 15.8
Table 4 Physical and chemical characteristics of the new blended

anifreezing admixtures (BAFA)
Superplasticizer type in BAFA
Appearance
Density, g/cm3
Concentration, %
Freezing temperature,
SNF
brown liquid
1.23-1.24
71
55
-28.7
liquid from yellow to red-brown color

1.20
5.51
52
-31.8
We established a set of criteria for technical effectiveness for the developed admixtures
to meet. Apart from obligatory compliance with a standard for compressive strength level3
these are: freezing point equal to or lower than -25 , no crystallization during long storage
period at temperature around -25 , viscosity suitable for pumping and dosage value.
We shall admit that its quite difficult to fit widespread SNF-type superplasticizers into
this system of criteria and there are 2 reasons for that. Firstly, its high dosage to ensure
superplasticizing effect (2 times higher than with lignosulfonates and 2.5-3 times higher
than with PCE). Thus, either PNS concentration in the blended admixture or admixture
dosage shall be increased. The second reason is sodium sulphate.
Its considered that sodium sulphate content amounting to appr. 3% (by dry substance)
is crystallization-safe for SNF at low temperatures. Indeed, such admixtures can be stored
under temperatures -4 -5 for a long time without precipitation. However if these
superplasticizers are used as a base for BAFAs the last becomes highly viscous at temperature -15 and acquire honey consistency (Table 5) when provided further temperature
decrease.
Ca-PNS-based BAFAs possess maximum permissible viscosity value (Table 5 line 4)
while the usage of chemically modified Na-PNS,4 possessing decreased dosage value, is
optimal in terms of this property. In this case (line 5) despite high sodium sulphate content
in the superplasticizer the viscosity of BAFA even at -25 is lower than at -15 for
Na-PNS-based-analogues with minimum sulphate content.
Table 5 Viscosity of PNS-based blended antifreezing admixtures

No.
1
2
3
4
5
PNS type
Na-PNS
Na-PNS
Na-PNS
-PNS
modified Na-PNS
Content of
Na2SO4, %
1
2
4.9
0
2.6
Viscosity, cP, for temperature

-15
-25
1150
impossible to measure
1490
402
1200
87
188
Table 6 Viscosity of PCE-based blended antifreezing admixtures

No.
1
2
3
PCE sample
-1
-2
-3
Viscosity, cP, for temperature

-15
-25
194
455
180
454
174
371
The main problems we came across in development of PCE-based BAFAs are polymer
compatibility with antifreezing agents (absence of segregation), compatibility of PCE +
antifreezing agent system with defoamer, high slump retention. Since the specified requirements are met, development procedure of this kind of BAFA could be similar to those of
SNF. It must be pointed out that viscosity level of PCE-based admixtures is between the
values of Ca-PNS and chemically modified Na-PNS-based analogues (Table 6).
According to GOST3 effectiveness of BAFAs is determined in concrete mixes belonging
to S3 consistence class (cone slump 10-15 cm (3.9-5.9 in)), however, in practice readymixed concretes are more frequently used as super-fluid mixes (S5 class). Thus, we tested
both admixtures in mixes of medium and high slump rate. Table 7 shows that PNS-based
BAFAs ensure strengthening at temperature -25 for all w/c ratios, however, the required
strength level (30% of the control composition) can be ensured only at w/c=0.43 (initial
S3 consistence class).
The required strength level for PCE-based BAFA is ensured even at higher w/c ratio
corresponding to S4 consistence class (Table 8).
In cases where extremely low freezing point is not required, the suggested compositions
of BAFAs can be used in decreased concentrations (Table 9). The recommended dosages
of these slightly diluted admixtures for application at moderately low negative temperatures remain approximately the same because superplasticizer concentration in the blended
admixture was reduced.
Another challenge is development of an antifreezing admixture with freezing point
around -40 . Its obvious that were not discussing concreting at a temperature that low.
However, new gas and oil fields are situated in regions with extremely severe climate and
its vitally important to ensure admixture non-freezing during transportation to the wells
and storage up to the application time.
Construction chemicals market offers a number of calcium nitrate-based admixtures but
cryohydric point of this salt is only -28,4 . Table 1 shows some compounds which in
the form of solution do not freeze at temperature around -40 but have serious draw-
284SP-302-21
Table 7 Characteristics of concrete with PNS-based blended antifreezing

admixtures
Slump, cm/in
Admixture
dosage, % by
commercial
product
W/C
initial
30 min
Density, kg/
m3 /lb/yd3
0.54
22/8.7
19/7.5
2395 4040
0.51
20/7.9
12/4.7
2405 4060
0.43
15/5.9
12/4.7
2420 4080
Compressive strength, MPa/psi

7d
28 d
moist
moist
chamber
-25
chamber
-25
29.8
31.5
7.2 1040
5.5 800
4320
4570
30.4
33.8
7.6 1100
8.1 1170
4410
4900
41.2
47.5
8.6 1250
9.9 1440
5970
6890
Table 8 Characteristics of concrete with PCE-based blended antifreezing

admixtures
Slump, cm/in
Admixture
dosage, % by
commercial
product
W/C
initial
30 min
Density, kg/
m3 /lb/yd3
1.3
0.52
21/8.3
10/3.9
2415 4070
1.2
0.46
17/6.7
8/3.1
2410 4060
1.3
0.43
18/7.1
12/4.7
2420 4080

7d
28 d
moist
moist
chamber -25 chamber
-25
31.2
35.0
8.8 1280
7.5 1090
4520
5080
39.6
41.7
10.1
7.6 1100
5740
6050
1460
35.2
40.7
10.0
8.6 1250
5100
5900
1450
Table 9Dependence of crystallization point on admixture concentration

Crystallization point,
Concentration, %
52
50
48
46
44
42
40
SNF-based
-25.2
-23.7
-20.9
-18.9
-17.2
-16.0
-14.0
PCE-based
-31.8
-28.9
-26.9
-23.7
-21.3
-19.7
-17.6
backs as concrete admixtures, e.g. calcium chloride causes reinforcement corrosion, potassium carbonate results in rapid concrete setting and porous structure deterioration, sodium
aluminate (is not included into Table 1) - rapid concrete setting, risk of ASR and strength
decrease.
On the ground of accomplished researches the composition on mixed organic and inorganic basis was developed. It is frost and crystallization-resistant at temperatures as low as
-35 and ensures effective concrete hardening at negative temperatures as low as -25 .
Table 10 Characteristics of concrete with non-plasticizing antifreezing

admixture (NPAFA)
Admixture
W/C
Dosage,%
by
commercial
product
Slump,
cm/in
Density,
kg/m3 /lb/
yd3
0.52
1.0
10/3.9
2430/4100
NPAFA
NPAFA+SP
Test
temperature

7d
28 d
moist
chamber
freezing
chamber
41.0 5950 9.2 1330

-25
0.50
1.4
11/4.3
2400/4050
38.0 5510 8.9 1290
0.46
1.0+1.1
13/5.1
2435/4110
46.7 6770 8.3 1200
0.44
1.0+1.1
12/4.7
2410/4060
0.43
1.0+1.1
11/4.3
2420/4080
-30
42.6 6180 9.4 1360

37.4 5420
10.2
1480
moist
freezing
chamber chamber
42.2
9.8 1420
6120
49.2
9.8 1420
7130
55.1
10.3
7990
1490
55.4
10.4
8030
1510
58.0
11.3
8410
1640
This admixture doesnt contain plasticizing components, so its action mechanism is

based on acceleration of hydration and hardening processes. This peculiarity accounts for
sufficient strength increase in comparison to BAFAs with comparable values of w/c ratio
(Table 10). At negative temperatures and w/c=0.52 this admixture ensures strength similar
to the concretes with BAFA with w/c=0.430.46. If it is used in combination with a superplasticizer it was found to be able to ensure the required strength even at -30 .
The basic principles of developing non-freezing blended admixtures being clear enough
(the most important one is decreasing content of water and compounds poorly soluble
at low temperatures), the question why these admixtures with low dosages turn to be so
effective in concrete is still open! When mixing water is poured all active components
become greatly diluted and liquid phase freezing point cannot remain as low as the one of
the admixture solution.
Figure 1 shows temperature curves in the concrete sample with minimum dosage of
BAFA (line 2) and in the reference sample containing only superplasticizer in similar
dosage (line 1). Sample cubes 100100100 mm were simultaneously casted and put into
a freezing chamber set at -18 . Three hours later the reference mix was cooled to 0 and
liquid phase began to freeze. Due to heat release during ice formation temperature decrease
stopped and resumed only in 2.5 hours.
Liquid phase in the sample with BAFA began to freeze below 0 and crystallization process ran at noticeably lower speed. It can be explained by the fact that admixture
concentration in the liquid phase increased after ice has started to form, thus the temperature of further ice formation is decreasing. So ice formed more slowly causing less damage
to the cement matrix.
Difference in temperature kinetics is much more distinct during defrosting (Fig. 2). In
reference sample temperature increase almost at a constant speed and slightly slow only
around 0 (owing to ice melting). Ice melting in the sample with minimum BAFA dosage
started at -17 and became intensive at -5 .
286SP-302-21
Figure 1. Freezing kinetics of concrete with superplasticizer

and BAFA (1 - - - - - SP only; 2 minimum BAFA
dosage; 3 maximum BAFA dosage).
The effects specified above become stronger at maximum admixture dosage (as shown
in Fig. 1-2, line 3). While freezing (Fig. 1), ice started to form only at -8 and process of
further cooling down of a sample was very slow. As far as the conditions of heat transfer
remained unchanged (invariable sample dimensions and weight) so cooling slowdown can
be explained by heat release during cement hydration. Judging by results of defrosting
(thawing) procedure (Fig. 2) the sample with higher admixture dosage contains less
ice (melting temperature ~ -7 ), upon melting hereof rapid heating of the sample was
observed and positive temperatures were reached much earlier than with reference mix.
Therefore, hydration reactions in the sample with antifreezing admixture start earlier and
run more vigorously.
In series of additional experiments, it was determined that some components of BAFA
significantly increase Portland cement hydration. Taking this in consideration, BAFA
efficiency have to be explained by: lowering ice formation temperature and amount of
ice, accelerating of Portland cement hydration and, as consequence, by formation of finer
microstructure at early stage of concrete hardening. Water reduction provided by superplasticizer contribute to early strength rise.
CONCLUSIONS
1. The compositions of blended antifreezing admixtures with freezing point below -25
have been developed.

2. The SNF and PCE-based blended antifreezing admixtures meet the effectiveness
requirements of national standard.
3. The composition of non-chloride antifreezing admixture with freezing point below -35
has been developed.

4. An explanation for effectiveness of low dosages of blended antifreezing admixtures in
concrete at extremely low temperatures has been proposed.
Figure 2. Thawing kinetics of concrete with superplasticizer

and BAFA ((1 - - - - - SP only; 2 minimum BAFA
dosage; 3 maximum BAFA dosage).
AUTHOR BIOS
Anatoly I. Vovk is Director of R&D Centre of Joint venture POLYPLAST, based
in Novomoskovsk, Tula region, Russia. He is a Doctor of Science, graduated and postgraduated Chemical Department of Moscow State University. Untill 1980, he worked as
Senior Research Officer in the Laboratory of Chemical Admixtures in Research, Design
and Technology Institute for Concrete and Reinforced Concrete, Moscow. Since 2006 he
work in POLYPLAST, Novomoskovsk, Russia.
REFERENCES
1. Concrete Admixture Handbook. Properties, Science, and Technology,
V.S.Ramachandran ed. Noyes Publications, Park Ridge, New Jersey, 1984.
2. Wang, Z.-M.; Sun, J.; and Wang, Y., A polycarboxylate based anti-freezing admixture for sub-freezing concreting, Ninth International Conference on Superplasticizers and
Other Chemical Admixtures in Concrete. Seville, Spain, October 2009, Supplementary
papers, American Concrete Institute, Farmington Hill, Michigan, pp. 103-110.
3. Russian Standard GOST 24211. Admixtures for concrete and mortars. General
specifications.
4. Vovk, A. I., Polyfunctional superplasticizers based on copolymer, Fifth International
Conference on Superplasticizer and Other Chemical Admixtures in Concrete. Rome, Italy.
1997. Suppl. vol., American Concrete Institute, Farmington Hill, Michigan, pp. 357-370.
288SP-302-21
SP-302-22
Effect of Expansive Agents and Shrinkage

Reducing Admixtures on the Performance
of Fiber-Reinforced Mortars
by M. Collepardi, V Corinaldesi, S. Monosi, and A.
Nardinocchi
In this research several fiber-reinforced mortars (FRMs) were studied. The effectiveness
of three different kinds of macro-fibers was tested: brass-coated steel fibers (SF), polyvinyl-alcohol fibers (PM and anti-crack HP glass fibers (GF) were separately added to
superplasticized mortar mixtures, at the same rate of about 1,2% by volume of the mortar.
Moreover, special FRMs were also manufactured combining a CaO-based expansive agent
with a shrinkage reducing admixture (SRA) in order to reduce the risk of cracking induced
by drying shrinkage and make more reliable mortars from the durability point of view.
All the mortar mixtures were characterized by the same w/c ratio of 0.45, and the same
sand/cement ratio of 3.0, as well as the same amount of a polycarboxylate-based superplasticizer (0.6% by weight of cement). A control superplasticized mixture (CM) with the
same w/c, the same sand/cement ratio, but without fibers, expansive agent and SRA was
also prepared and studied for comparison purpose. Moreover, an expansive mortar (EM)
without fibers but with expansive agent and SRA was manufactured and studied.
All the mortar mixtures were characterized for the workability in the fresh state (where
they showed approximately the same plastic consistency), and in the hardened state by
measuring compressive and flexural strength, as well as free or restrained length changes.
The results obtained show the effectiveness in the restraint expansion of the combined use
of macro-fibers, expansive agent and SRA when glass fibers and steel fibers were used.
Keywords: expansive agent. SRA. fiber reinforced mortars. steel fibers. polyvinylalcohol fibers. glass fibers. compressive strength. flexural strength. restrained lengthchange. free length-change.
INTRODUCTION
Mortar and concrete are the most widely used construction materials because they
develop high strength, high toughness and long-term durability. However, drying shrinkage
and the related cracking are two important properties which can reduce the durability. It has
289
290SP-302-22
Fig. 1 Restrained length-change in a reinforced expansive

mortar containing CaO or CaO + SRA.
been shown1,2 that incorporating fibers into cementitious materials can effectively improve
their thoughness and ability of resisting cracks.
In order to improve the performance of fiber-reinforced cementitious materials in terms
of higher ability of resisting cracks, expansive agents3,4 or a combination of an expansive
agent with a shrinkage-reducing admixture5 have been studied.
The purpose of the present research was to examine the influence of a CaO-based
expansive agent combined with SRA on the properties of mortars reinforced with three
different types of macro-fiber: metallic (steel), polymer (PVA) and alkali-resistant glass.
The following properties were determined: compressive and flexural strength, restrained
and free-length change.
High-strength portland cement (CEM I 52.5 R according to EN 197-1) was used in
combination with a standard sand according to EN 196-1. The water/cement ratio and the
sand/cement ratio of all the mortars were 0.45 and 3.0, respectively.
A 30% aqueous solution of a polycarboxylate-based superplasticizer was used (2.9 kg/
m3-4.9 lb/yd3) in order to manufacture plastic mortars at about the same slump (110-150
mm/4.4-6 in).
A dead-burnt CaO was used (35 kg/m3-59.1 lb/yd3) as an expansive agent for some
special mortar mixtures (EM, ESFM, EPFM and EGFM). In these mixtures a neopentylglycol- based SRA (4.5 kg/m3-7.6 lb/yd3) was used in order to emphasize the properties of
the length-change by both increasing the expansion of CaO and reducing the subsequent
drying shrinkage as it is shown in Fig.1.6 According to Maltese et al.7 the increase in the
expansion in the presence of SRA is due to the change in the morphology of Ca(OH)2
caused by SRA.
The macro-fibers were used at about the same volume percentage of 1.2% with respect to
the mortar. By taking into account the different specific weight of the fibers, the following
amounts were adopted: 100 kg/m3 (169 lb/yd3) of steel fibers; 10 kg/m3 (16.9 lb/yd3) of
PVA fibers and 30 kg/m3 (50.7 lb/yd3) of anti-crack HP alkali-resistant glass fibers. The
Effect of Expansive Agents and Shrinkage Reducing Admixtures on the

Performance of Fiber-Reinforced Mortars 291
Table 1 - Composition and slump of mortars with a w/c ratio of 0.45 and a
sand/cement ratio of 3
Composition kg/
m3 (lb/yd3)
Water
Cement
Sand
Superplasticizer
CM
EM
Fibers
SRA
CaO
Slump
mm (in)
130
(5.2)
SFM
100
(169)
-
4.5 (7.6)
35
(59.1)
145
110 (4.4)
(5.8)
ESFM
PFM
200 (338)
450 (761)
1350 (2282)
2.9 (4.9)
10
100 (169)
(16.9)
4.5 (7.6)
-
EPFM
GFM
4.5 (7.6)
30
(50.7)
-
10 (16.9)
EGFM
30 (50.7)
4.5 (7.6)
35 (59.1)
35 (59.1)
35 (59.1)
115 (4.6)
130
(5.2)
150 (6.0)
125
(5.0)
125 (5.0)
length of the fibers were:12 mm (0.5 in) for the steel fibers, 15 mm (0.6 in) for the PVA
fibers and 36 mm (1.5 in) for the glass fibers. The aspect ratio (length/diameter) was 75 for
the steel fibers, 48 for the PVA fibers and 67 for the glass fibers.
The following methods were adopted to determine the properties of the mortar mixtures
whose compositions are shown in Table 1:
- slump after mixing the fresh mortars;
- 100 mm cube compressive strength at room temperature with RH of 95%;
- flexural strength at room temperature with RH of 95% measured on 50x50x200 mm
(2x2x8 in) specimens placed on two supports 175 mm (7 in) distant with a 0.25 mm (0.1
in) thick and 25 mm (1 in) deep notch in the middle of the length;
- length-change restrained by a 280 mm (11.2 in) long steel bars of 50x50x250 mm
(2.5x2.5x10in) mortar specimens demolded at 6 hours to measure the initial length of the
steel bar, then wrapped for 2 days by a plastic sheet and finally exposed to RH of 55%
according to the Italian Standard UNI 8147- method B;
- free length- change of the mortar specimens demolded at 6 hours to measure the initial
length, then wrapped for 1 day from by a plastic sheet and finally kept at RH of 50%
according the Italian Standard UNI 6687.
RESULTS
Compressive strength
Figures 2, 3, and 4 show the compressive strength of the mortars as a function of the
curing time (RH = 100%) at 20C (68F).
The results of Fig. 2 show that the addition of steel fibers increases the compressive
strength of the mortar. On the other hand, in the mortars containing the expansive agent
combined with SRA (EM and ESFM) the compressive strength is lower than that of the
control mortar (CM). This effect is due to the presence of SRA which reduces the cement
hydration and consequently the related compressive or flexural strength.8
Figure 3 indicates that the compressive strength of the mortars containing polymer fibers
(PFM and EPFM) are significantly lower than that of the control mix (CM). On the other
292SP-302-22
Fig.2 - Compressive strength as a function of curing time for

steel fibers mixes

polymer fibers mixes
hand is confirmed the negative effect of SRA in the expansive mortar on the compressive
strength: the 28-day compressive strength of the EM mortar is about 30% less than that of
the CM mortar.
Figure 4 indicates that the addition of glass fibers (GFM) does not change the compressive strength with respect to the control mortar (CM). Again, the presence of SRA reduces
the compressive strength in the EM and EGFM mortars with respect to the SRA-free
mortars.
Flexural strength
Figures 5-7 show the flexural strength of the mortars as a function of the curing time (RH
= 100%) at 20C (68 F).


glass fibers mixes
Fig.5 - Flexural strength as a function of curing time for

steel fibers mixes
Figure 5 indicates that the addition of steel fibers increases the flexural strength by 45%
with respect to the control mortar (CM); the influence of steel fibers appears to be more
effective on flexural than on the compressive strength (Fig.2). In the presence of SRA there
is a reduction in the flexural strength. The strength increase from expansive mortar (EM) to
the corresponding steel reinforced mortar (ESFM) is about 30%.
Figure 6 indicates that the flexural strength of the PVA fiber-reinforced mortars is slightly
higher than that of the corresponding mortars without fibers. The presence of SRA does not
affect significantly the flexural strength.
Figure 7 indicates that the flexural strength of the glass fiber-reinforced fibers (GFM and
EGFM) is about 30% higher than that of the corresponding mortars without fibers (CM
and EM).
294SP-302-22

polymer fibers mixes

glass fibers mixes
Length change
Figures 8 and 9 show the restrained and the free length change respectively.
The restrained length change (Fig.8) occurs in mortars 250 mm (10 in) long reinforced
by a 280 mm (11.2 in) long steel bar and completely wrapped for 2 days by a plastic sheet
in order to keep the specimens in a humid environment; then after 2 days the plastic sheet
is removed and the restrained shrinkage can occurs at a RH of 55%. In all the mortars
containing the expansive agent CaO combined with SRA there is an initial expansion of
about 70010-6 . Then it takes about 50 days of permanent exposure to the dry environment
(RH = 55%) for completely losing the expansion state. In the absence of expansive agent
there is a slow drying shrinkage which reach about 50010-6 at 1 month of exposure at a
RH of 55%.
The free expansion in the absence of the reinforcing steel bar is significantly higher in
the control mix (Fig.9) with respect to the restraint expansion of same mortar (Fig.8). The

Fig.8 - Restrained length-change as a function of time (RH

= 55% after 2 days)
Fig.9 - Free length-change as a function of time (RH = 55%

after 1 day)
addition of fibers reduces the free expansion due to some restrain caused by the fibers. In
the mortars without expansive agent there is a drying shrinkage after the first day which
is significantly lower in the mortars containing steel or glass fibers (120010-6 at 50 days)
with respect to the control mix or the mortar containing PVA fibers (250010-6 at 50 days).
CONCLUSIONS
Due to the retarding effect of SRA on the portland cement hydration, the compressive
and the flexural strength of the fiber reinforced mortars in the presence of the CaO-based
expansive agent combined with SRA are reduced.
The restrained expansion of all the fiber-reinforced mortars is the same as that of the
control mortar without fibers; also the restrained shrinkage in the absence of the expansive
agent of the control mortar is not changed by the presence of fibers.
296SP-302-22
The free expansion of the fiber-reinforced mortars is reduced with respect to the control
mortar without fibers; the free shrinkage of the mortar containing steel or glass fibers is
lower than that of the control mortar, whereas the free shrinkage of the mortar containing
the PVA fibers is the same as that of the control mortar.
AUTHOR BIOS
Mario Collepardi is ACI Honorary Member. He used to be Professor of construction materials in the Italian University of Cagliari, Rome, Ancona, Venice and Milan.
He authored or co-authored about 400 papers. He is also author or co-author of some
textbooks on the science and technology of concrete in Italian, English and Chinese. His
research has resulted in five European patents in a variety of fields including chemical
admixtures, expansive agents, silica fume and special cements.
Valeria Corinaldesi is BS and MS in Civil Engineering, PhD in Materials Engineering.
She is Assistant Professor at the Universit Politecnica delle Marche, Ancona, Italy. She
is author and co-author of more than 160 publications mainly on peer-reviewed international journals and conference proceedings, and of 2 national patents. She works in
cement and concrete technology. Main research interests are development of new cementbased, multi-functional and sustainable composites.
Saveria Monosi is Associate Professor of Materials Science and Technology at the
Universit Politecnica delle Marche, Ancona, Italy. She is author or co-author of
numerous papers on international journals and conference proceedings in the area of
cement and concrete properties, chemical and mineral admixtures, and durability of
building materials.
Alessandro Nardinocchi is BS and MS in Civil Engineering, PhD student at the Universit Politecnica delle Marche, Ancona, Italy. He is co-author of some papers on international conference proceedings, as well as author of a national patent. Main research
interest are fiber reinforced concrete and ultra-high performance concrete.
REFERENCES
1. Swamy, R. N., ed., Fiber Reinforced Cement and Concrete, 4th RILEM International Symposium, Chapman and Hall, Sheffield, UK, 1992.
2. Cheyrezy, M.; Deniel, J. I.; Krenciel, H.; Mihashi, H.; Pera, J.; Rossi, P.; and Xi, Y.,
Specific Production and manufacturing issue in: A.E. Naaman, H.W. Reinhardt (Editors),
High Performance fiber Reinforced Cement Composite Volume 2 (HPFRCC2), Proceedings of the 2nd International Workshop, E&FN SPON, London, pp.25-41, 1995.
3. Sun, W.; Chen, H.; Luo, X.; and Qian, H., The effect of hybrid fibers and expansive agent on the shrinkage and permeability of high-performance concrete, Cement and
Concrete Research, V. 31, No. 4, 2001, pp. 595-601. doi: 10.1016/S0008-8846(00)00479-8
4. Aiguo, W.; Min, D.; Daosheng, S.; Liwu, M.; June, W.; and Mingshu, T., Effect of
Combination of Steel Fibers and MgO-type Expansive Agent on Properties of Concrete,
Journal of Wuhan University of Technology-Mater, Science Editor, V. 26, 2011, pp.
786-790.

5. Park, J.-J.; Yoo, D.-Y.; Kim, S.-W.; and Yoon, Y.-S., Drying shrinkage cracking
characteristics of ultra-high-performance fibre reinforced concrete with expansive and
shrinkage reducing agents, Magazine of Concrete Research, V. 65, No. 4, 2013, pp.
248-256. doi: 10.1680/macr.12.00069
6. Collepardi, M., The New Concrete, Tintoretto, Villorba, pp.195-197, 2010.
7. Maltese, C.; Pistolesi, C.; Lolli, A.; Bravo, A.; Cerulli, T.; and Salvioni, D.,
Combined effect of expansive and shrinkage reducing admixtures to obtain stable and
durable mortars, Cement and Concrete Research, V. 35, No. 12, 2005, pp. 2244-2251. doi:
10.1016/j.cemconres.2004.11.021
8. Gettu, R., and Roncero, J., On the long term response of concrete with a shrinkage
reducing admixture, Proceedings of the Conference Admixtures-Enhancing Concrete
Performance, Editors R. K. Dhir, P.C. Hewlett, M. D. Newlands, pp.209-216, 2005.
298SP-302-22
SP-302-23
Interactions between Cements with

Calcined Clay and Superplasticizers
by Jens Herrmann and Jrg Rickert
Interactions between cements with calcined clay and superplasticizers based on polycarboxylates were investigated. All cements contained the identical clinker and sulfate
carriers as well as a systematic variation of the type and proportion of calcined clay.
Investigations were carried out on fresh cement paste as well as on fresh concrete by means
of pore solution analyses, zeta potential experiments and rheological as well as consistency measurements. Depending on the mineralogical composition of the clay and on the
calcining conditions, the specific surface area of the calcined clay varied significantly.
With an increasing proportion of calcined clay in the cement, the fresh cement pastes pore
solution composition changed and, as a result, so did the zeta potential. Depending on the
type and proportion of calcined clay in the cement, the specific sorption behavior of the
superplasticizers diminished and consequently, their particular plasticizing effects.
Keywords: adsorption; blended cements; calcined clay; pore solution; rheology; superplasticizer; supplementary cementitious material; zeta potential; workability.
INTRODUCTION
To enhance its strength and durability, concrete is produced with reduced w/c ratios
and, for workability reasons, with added superplasticizers. The active substance of most
of these concrete admixtures is polycarboxylate ether (PCE).1 The production of efficient
blended cements and their application in technically demanding and sustainable concrete
constructions is state-of-the-art. Besides numerous technical advantages, these cements
contribute to saving natural resources and energy as well as reducing CO2 emissions due
to their reduced portland cement clinker content. As a substitute for clinker, most of these
cements contain ground limestone, natural pozzolanas, pozzolanic fly ash, and / or latent
hydraulic ground granulated blastfurnace slag.2 Unfortunately, the availability of industrial
pozzolanas and latent hydraulic supplementary cementitious materials is limited either in
proportion in cement or in local availabilities. An innovative and currently the most promising clinker substitute is calcined clay.3 Clays of suitable quality for use in the cement
industry are available almost world-wide in sufficient quantity and often near to cement
plants. Meeting the requirements of the European cement standard EN 197-1 ( 25 mass
299
300SP-302-23
% reactive SiO2), it is already permitted to use calcined clay as a substitute for clinker in
cement up to 55 mass %.
Action modes of superplasticizers and their interactions with portland cement and
its initial hydration products have been investigated extensively and are widely known
today.4-16 Interactions of superplasticizers with cements blended with limestone, fly ash,
and/or slag have been increasingly moved into the spotlight of research in recent years.17-21
If at all, interactions of superplasticizers with clay minerals have been reported in technical
and scientific literature22-25 but there has not yet been much knowledge about the interactions between calcined clays and superplasticizers.
The aim of comprehensive investigations at VDZ was to determine influences of the type
and proportion of calcined clay as supplementary cementitious material in cement on the
sorption behavior and plasticizing effect of polycarboxylate-based superplasticizers.
Investigations were performed with priority on fresh cement pastes using pore solution analyses, zeta potential experiments, and rheological measurements. Cements were
composed of identical portland cement clinker and sulfate carriers. The type and proportion of calcined clay in the cements were varied systematically to determine influences on
the sorption behavior, plasticizing effect and duration of plastification of different structured polycarboxylates. The results obtained were verified by concrete consistency tests.
Calcined clay is an innovative and one of the most promising supplementary cementitious materials. A literature study revealed that extensive knowledge on interactions
between cements with calcined clay and polycarboxylates, which are used wildly in
concrete production, did not yet exist. Varying the type and proportion of calcined clay in
the cement, their influences on the performance of polycarboxylates with different structures were determined systematically. This and the methodical evaluation of the results
of pore solution analyses, zeta potential experiments, and rheological measurements are
enhancing the understanding of the precise interactions between these superplasticizers
and cements with calcined clay.
Constituent materials
Calcined claysa chloritic-illitic and an illitic-kaolinitic clay were used. The calcining
conditions needed to purposefully activate their pozzolanic properties were determined and
the suitability of the calcined clays as cement main constituent according to EN 197-1 were
proven in a previous project.26 The calcined clays (chloritic-illitic = Q11; illitic-kaolinitic
= Q21) were ground to the same fineness in a laboratory ball mill. Besides their mineralogical composition, they differed in their specific BET surface areas (Q11 >> Q21). Their
chemical compositions as well as their granulometrical and physical characteristics are
shown in Table 1. The specific BET surface area was determined using nitrogen adsorption
in accordance with ISO 9277. The particle size distribution was analyzed using laser granulometry. The fineness (x) and the slope (n) of the particle size distribution were calculated
according to Rosin, Rammler, Sperling and Bennet (RRSB). Their mineralogical compositions prior to and after the calcination are shown in Figure 1.
Interactions between Cements with Calcined Clay and Superplasticizers 301
Table 1Chemical compositions, granulometrical, and physical characteristics of the calcined clays
Parameter
SiO2
Al2O3
Fe2O3
MgO
CaO
SO3
Na2O-Eq.
react. SiO2, EN 196-2 1)
density, EN 196-6
surface area, BET (N2)
fineness x, RRSB
slope n, RRSB
water demand, EN 196-3
Unit
mass %
(XRF)
mass %
mass %
g/cm3
cm2/g
m
mass %
Q11
(chloritic-illitic)
58.8
23.1
8.1
1.8
2.4
0.1
3.0
30.2
2.73
50,501
13.4
0.60
30.5
Q21
(illitic-kaolinitic)
71.8
21.8
1.6
0.6
0.0
0.0
2.1
33.4
2.63
23,043
12.5
0.71
27.0
1) requirement of EN 197-1: 25.0 mass %
Figure 1X-ray diffraction pattern of the clays prior to and

after the calcination.
Cementsportland cement CEM I 42,5 R (CEM I) in accordance with EN 197-1 was
used as a reference. Its chemical and mineralogical compositions as well as its granulometrical, mechanical, and physical characteristics are presented in Table 2. Laboratory-made
cements were produced by intensive mixing of the CEM I with the particular calcined
clay. The proportions of calcined clay in the cements were 20 mass % (portland-pozzolan
cement CEM II/A-Q), 35 mass % (portland-pozzolan cement CEM II/B-Q) and 55 mass %
(pozzolan cement CEM IV/B (Q)), respectively. The cements met the requirements of EN
197-1. Analyses data for the cements are reported in Table 3.
302SP-302-23
Table 2Chemical-mineralogical composition, granulometrical, mechanical,

and physical characteristics of the portland cement CEM I 42,5 R
Parameter
loss on ignition
SiO2
Al2O3
Fe2O3
MgO
CaO
SO3
Na2O-Eq.
strength 1), EN 196-1
initial setting, EN 196-3
water demand, EN
196-3
Unit
mass %
MPa
min
CEM I
2.4
20.5
5.9
2.0
1.0
64.9
2.6
0.75
32/50/59
155
Parameter
alite / belite
C3Acubic/orthorhombic
ferrite
anhydrite
hemihydrate
gypsum
free lime/calcite
surface area, BET (N2)
fineness x, RRSB
slope n, RRSB
mass %
27.5
fineness, EN 196-6
mass %
(XRF)
Unit
cm2/g
m
-
CEM I
59 / 12
11/3
3
1.1
2.9
1.5
0.5/4.2
11,438
19.7
0.85
cm2/g
3,650
mass %
(XRD/
RV)
1) 2 / 7 / 28 d (1 MPa 0.145 ksi); RV: Rietveld analysis
Table 3Granulometrical, mechanical, and physical characteristics of the

laboratory-made cements
cement
type
CEM I 42,5
R
CEM II/A-Q
CEM II/A-Q
CEM II/B-Q
CEM II/B-Q
CEM IV/B
(Q)
CEM IV/B
(Q)
abbrev.
CEM I
Z20Q11
Z20Q21
Z35Q11
Z35Q21
Z55Q11
Z55Q21
calcined
type of
clay
calcined
proportion
clay
0 mass %
20 mass
%
20 mass
%
35 mass
%
35 mass
%
55 mass
%
55 mass
%
Q11
Q21
Q11
Q21
Q11
Q21
fineness
x
m
slope n
-
19.7
0.85
17.6
0.77
15.5
0.74
15.8
0.74
15.2
0.74
15.5
0.66
13.2
0.71
BET
surface
cm2/g
11,438
WD
mass%
strength 1)
MPa
27,5
32/50/59
18,125
n. d.
n. d.
13,973
n. d.
n. d.
24,855
28.0
20/35/48
14,345
27.5
18/34/47
30,158
27.0
11/23/36
15,963
27.0
11/20/34
WD: water demand; 1): compressive strength 2, 7, 28 d (1 MPa 0.145 ksi); n. d.: not determined
Superplasticizerstwo commercially available superplasticizers in accordance with EN

934-2 based on polycarboxylates, denoted as PCE11 and PCE22, were used. PCE11 was
recommended by the manufacturer for the use in ready-mixed concrete, and PCE22 for
precast concrete. Chemical and physical characteristics are given in Table 4. The effective
charge of the superplasticizers in deionized water, in solution of potassium hydroxide (KOH)
at pH 12.6, and in the pore solution of different cement pastes (w/c = 0.35) were determined
using a particle charge detector titrating polydiallyldimethylammoniumchloride.
Table 4Chemical and physical characteristics of the commercial

superplasticizers
Parameter
density, aerometer
solid content, EN 480-8
TOC, EN 1484
molar mass Mw
molar mass Mn
SEC
hydrodyn. radius
gyration radius
deion. Water
KOH (pH 12.6)
effective
CEM I
charge in
Z35
pore solution
Z55
Unit
g/cm3
mass %
g/dm3
Da
nm
eq/g
PCE11
1.066
30.3
153
81,000
22,000
6.7
27.2
-1,256
-4,763
-68
n. d.
-50
PCE22
1.062
29.1
161
95,000
41,000
6.5
16.5
-1,124
-4,217
-274
-95
-245
SEC: Size Exclusion Chromatography with static and dynamic light scattering; n. d.: not determined
Mixing waterdeionized water was used for the production of fresh cement paste and
tap water for that of fresh concrete.
AggregatesRhenish sand and gravel of the grading curve B in accordance with the
German Standard DIN 1045-2, Annex L, with a maximum grain size of 16 mm (0.63 in.),
were used.
Proportion, mixing procedure, storage of fresh cement paste/fresh concrete,
and separation of pore solution
The constituent materials were stored at approximately 20C (68F). Fresh cement paste
with a water to cement ratio of 0.35 (w/c = 0.35) as well as suspensions of deionized water
and calcined clay (w/s = 0.35) were produced in a mortar mixer in accordance with EN
196-1 at approximately 20C (68F). Cement and water were mixed for 30 s at approximately 140 rpm. A break of approximately 30 s was taken to reincorporate the material that
had adhered at the bottom and the walls of the mixing bowl into the fresh cement paste.
Mixing was re-started for another 30 s at approximately 280 rpm. The superplasticizer
was added approximately 90 s after the water addition and mixed for approximately 60 s
at 280 rpm. Quantities of superplasticizer added always refer to the dry solid content of
the admixture (Table 4) and with respect to the mass of cement. One cubic meter (1 yd3)
of fresh concrete contained 355 kg (599 lbs) cement, 188 kg (317 lbs) water (w/c = 0.53),
and approx. 1770 kg (2987 lbs) of aggregates. Concrete was mixed in accordance with EN
480-1 at approximately 20C (68F). For each dosage of superplasticizer, a fresh cement
paste or concrete was produced to maintain uniform influences on the hydration and sorption processes. The quantity of water within the superplasticizer was always taken into
account for the quantity of mixing water added.
Samples were stored at approximately 20C (68F) in a plastic bowl (paste) or bucket
(concrete), both covered with a damp cloth to avoid evaporation. Prior to testing, the
samples were homogenized with a spoon or a scoop. Tests were conducted at approximately 20C (68F). Approximately 5, 10, 15, 30, 60, 90, and 120 min after the water
304SP-302-23
addition, the particular fresh cement paste or suspension was vacuum-filtrated for approximately 60 s using a Buchner funnel with a blue-ribbon filter. Separated pore solution was
filtered by a 0.45 m PTFE syringe filter and stored flushed with argon in sealed tubes cold
and dark until the analyses were conducted within the following week.
Test methods
Analyses of pore solutionthe conductivity was measured with a conductometer on
the original sample. The other analyses were done on subsamples. The concentration of
OH- was determined by dynamic equivalence point titration with hydrochloric acid. By ion
chromatography, the concentrations of Cl- and SO42- as well as Na+, K+, Ca2+, and Mg2+
were determined (cations: methane sulfonic acid, anions: mixture of sodium carbonate and
sodium bicarbonate). Prior to the determination of the cations, the particular sub-sample
was acidified with hydrochloric acid. The concentration of aluminum ions was determined
in accordance with ISO 10566, that of silicon dioxide in accordance with the German
Standard DIN 38405-21. The TOC content was determined in accordance with EN 1484.
Zeta potentialthe zeta potential of fresh cement paste as well as that of the suspensions
with calcined clay and pore solution or deionized water was determined approximately
15 min after the addition to the respective liquid using an electro-acoustic method.27 The
electro-acoustic background, resulting from dissolved ions in the particular solution (ion
vibration current), was always taken into account for the calculation of the zeta potential.
Sorption of superplasticizerthe sorbed quantity of superplasticizer (i.e. intercalated
and/or adsorbed)28 was the difference between the TOC content determined in the pore
solution and the quantity of TOC added with the addition of the superplasticizer. The TOC
content of the particular pore solution without adding superplasticizer was always taken
into account.
Plastification and saturation dosage of fresh cement pastethe plasticizing effect of
the polycarboxylates in fresh cement paste was investigated using a rotational rheometer
for building material suspensions. Immediately after the end of the mixing procedure, the
sample was filled into the rheometers vessel and a paddle was inserted into it. The vessel
rotated at 60, 80, 100, 80, and 60 rpm for 5 min each and the torque was measured via
the paddle. The torque at 80 rpm after approximately 20 min was the shear resistance.
If the shear resistance could not be significantly reduced further by a larger dosage of
superplasticizer, the rheological saturation of the particular combination of cement and
superplasticizer at the testing conditions used was reached. This dosage was the saturation
dosage (SD). Generally, dosages beyond the saturation dosage can cause segregation and
retardation.11,29
Duration of plastification of fresh cement pastethis was determined with the mini
slump-flow test on the basis of EN 12350-8 using a mini-slump cone in accordance with
EN 1015-3 approximately 5, 10, 15, 30, 60, 90, and 120 min after adding water. The quantity of superplasticizer added was 90% of the particular saturation dosage.
Consistency and saturation dosage of fresh concretethe consistency of fresh concrete
was determined by measuring the spread of the concrete after 15 shocks with the flow table
test in accordance with EN 12350-5 at approximately 10 min after water addition. For
the concrete composition and test conditions used, the saturation dosage of the particular
cement and superplasticizer combination was reached if the concretes flow table spread
Figure 2Conductivity and ion concentration of pore solution in dependence of the proportion of Q21 in the cement
(left) and normalized to the portland cement proportion, pcp
(right).
could not be significantly increased further by a larger admixture dosage and a paste ring
started to form during the spread, indicating segregation. Retention of the concretes
consistency was determined with an added superplasticizer quantity of 90% of the particular saturation dosage at the same testing ages as for the fresh cement paste.
Influences of the type and proportion of calcined clay in the cement on the
composition of the pore solution and the zeta potential
With an increasing proportion of calcined clay Q21 in the cement the pore solutions
composition varied significantly (Figure 2, left side). The conductivity as well as the
concentrations of Na+, K+, SO42-, and Cl- decreased in direct proportion to approximately
zero (suspension of calcined clay and deionized water). Also, the OH- concentration
decreased in almost direct proportion. These influences on the pore solution compositions are attributed to the substitution of the clinker and sulfate carriers, i.e. readily soluble
alkalis and sulfates.20,21 With decreasing alkalinity of the pore solution the solubility of
Ca2+ increases.30 Thus, the Ca2+ concentrations remained almost at the same level. The
concentrations of Mg2+ were always in the range of the detection limit. The concentrations
of Al3+ and Si4+ were in the range of mol/L. During the testing period of approximately
120 min after adding water, the particular ion concentration remained almost unchanged.
Increasing proportions of calcined clay Q11 in the cement influenced the composition of
the pore solution to a comparable degree (not shown). Generally, changes in the pore solutions conductivity, as a measure of its ionic strength, as well as in the ionic composition,
especially OH-, SO42- and Ca2+, can influence the zeta potential as well as the constitution
of polycarboxylates, depending on the particular polymer structure, and so their respective
sorption behavior and performance potential.20,21,30,31
On the right side of Figure 2 the concentrations of the ions in the pore solution are
normalized by the proportion of clinker and sulfate carriers in the particular cement. The
concentrations of Na+, K+, SO42-, and Cl- remained unchanged emphasizing that these ions
were released by the clinker and sulfate carriers. The increased concentration of Ca2+ was
electrochemically neutralized by the increased concentration of OH-.
306SP-302-23
Figure 3Zeta potential of fresh cement paste in dependence of the type and proportion of calcined clay in the
cement (left) and zeta potential of the constituents for their
particular proportion in the respective cement (right); CEM
I suspended in deionized water and calcined clays in the
pore solution of the respective fresh cement paste.
The zeta potential of the particular fresh cement paste in dependence of the type and
proportion of calcined clay in the cement is shown in Figure 3 on the left side. The zeta
potential of the paste with CEM I was slightly negative (approximately -6 mV). Increasing
proportions of calcined clay in the cement shifted the zeta potential towards the iso-electrical point (IEP). This was slightly more pronounced using Q21. In general, the more the
zeta potential of a suspension is near the IEP, the more this suspension is agglomerated and
the demand for a dispersing admixture increases and vice versa. Here, agglomeration and
the respective shift of the fresh cement pastes zeta potential towards the IEP was mainly
induced by the changes in compositions of the pore solutions (cf. Figure 2) as well as due
to the different zeta potentials of the particular cement constituents (Figure 3, right side).
Portland cement was suspended in deionized water and the calcined clays in the particular
pore solution (cf. Figure 2) to mimic the conditions for calcined clay within the cement
paste.
Sorption and plasticizing effect
The sorption of the polycarboxylates and their particular plasticizing effect in dependence of the quantity of active substance added as well as of the cement type are shown
in Figure 4. In combination with CEM I, the sorbed quantity of PCE11 increased slightly
with the superplasticizer addition (Figure 4, upper left side). This is probably due to the
low anionic charge of PCE11 (cf. Table 4) as well as the high concentration of SO42- in the
pore solution of the paste with CEM I (cf. Figure 2, left side) and its negative zeta potential
(cf. Figure 3, left side). Thus, the sorption was moderate and the majority of added PCE11
remained in the pore solution (depot effect). With an increasing proportion of Q21 in the
cement, the sorption of PCE11 increased. This may be due to the decreasing SO42- concentration, the less negative zeta potential of the particular cement paste and Q21, respectively,
and also to the large calcined clay surface. This also applies for increasing proportions of
Q11, but was more pronounced due its specific surface area which was at least twice as
large as that of Q21 (cf. Tables 1 and 3).
Figure 4Sorption and plasticizing effect of PCE11 (left)

and PCE22 (right) in dependence of the quantity added as
well as the type of cement.
On the lower left side of Figure 4, the particular plasticizing effect of PCE11 is shown.
Due to its moderate sorption in combination with CEM I, also its plasticizing effect was
moderate. The saturation dosage was approximately 0.18 mass % of CEM I. Compared to
the fresh cement paste with CEM I, the paste with cement with 35 mass % Q11 (Z35Q11)
was plasticized to a greater extent with lower quantities of PCE11 due to the higher sorption. The saturation dosage decreased to 0.12 mass % of Z35Q11. The proportion of 55
mass % of Q11 in the cement (Z55Q11) enhanced its plasticizing effect further, but approximately the same dosage was necessary to reach saturation (0.13 mass % of Z55Q11). This
is due to the significant increase in the specific surface area of the cement with increasing
content of Q11. The particle size distribution and the particle packing also play a role and
these change with the calcined clay content. The increasing proportions of Q21 led also to
an increase in the plasticizing effect of PCE11 but to a significant decrease in the saturation dosage (0.07 mass% of Z55Q21). This was due to the increased sorption of PCE11
with decreasing clinker proportions and sulfate concentrations, as well as due to the lower
specific surface area of Q21.
On the right side of Figure 4, the results for PCE22 are shown. The sorption of PCE22
was higher in combination with CEM I compared to that of PCE11 (Figure 4, upper right
side). This can be explained by its higher anionic charge (cf. Table 4) that enables a stronger
sorption even at the high SO42- concentration in the pore solution of the fresh cement paste
with CEM I. The sorbed quantities of PCE22 decreased with a decreasing proportion of
clinker in the cement. This indicates a strong affinity of PCE22 to clinker. At 55 mass % of
calcined clay in the cement, the sorption of PCE22 and PCE11 were comparable. As with
308SP-302-23
Figure 5Shear resistance of fresh cement paste in dependence of the sorbed quantity of active substance (left:
PCE11, right: PCE22). The shear resistance is normalized
to that of the respective paste without superplasticizer.
PCE11, the sorbed quantity of PCE22 in combination with the cements containing Q11 was
higher than in combination with cements containing Q21 due to their specific surface areas.
On the lower right side of Figure 4, the particular plasticizing effect of PCE22 is
shown. The strong sorption resulted in the strong plasticizing effect. The saturation dosage
was approximately 0.12 mass % of CEM I as well as of the cements with Q11. With the
increasing proportions of Q21 in the cement, the saturation dosage decreased slightly.
Figure 5 shows the shear resistance of fresh cement paste in dependence of the sorbed
quantity of active substance (left: PCE11, right: PCE22). The shear resistance is normalized to that of the respective paste without superplasticizer. On the left side of Figure 5 by
the steeper slopes, it is clearly visible that the plasticizing effect of PCE11 increased with
the proportion of calcined clay in the cement as well as with the specific surface area of the
substitute. For PCE22, this was not such pronounced because of its stronger sorption even
in combination with CEM I.
Sorption and duration of plastification
The sorption of the polycarboxylates and their particular duration of plastification in
dependence of the cement type are shown in Figure 6. The dosage was always 90% of the
particular saturation dosages (cf. Figure 4). The quantity of PCE11 sorbed in combination
with CEM I was approximately 22 mass % to 30 mass % of the added quantity during the
first 30 min after water addition (Figure 6, upper left side). After 120 min the quantity
sorbed was approximately 45 mass %, only. Due to the depot effect, PCE11 remained in the
pore solution to a large extent caused mainly by the high concentration of SO42- in the pore
solution of the paste with CEM I (cf. Figure 2, left side) and its negative zeta potential (cf.
Figure 3, left side) as well as benefiting from the low effective anionic charge of PCE11
(cf. Table 4). In combination with Z35Q21, the time-dependent sorption of PCE11 was
comparable. In combination with Z55Q21, the quantity sorbed was approximately 40 mass
% to 45 mass % during the first 30 min and approximately 55 mass % after 120 min due to
the lowered concentration of SO42- and the increase in Ca2+ as well as the more positively
charged surface (cf. left sides of Figure 2 and Figure 3). The substitution of the portland
cement component with Q11 with larger specific surface area led to a further increase in
the time-dependent sorption of PCE11. On the lower left side of Figure 6, the particular
Figure 6Sorption and duration of the plasticizing effect of

PCE11 (left) and PCE22 (right) in dependence of the type
of cement.
duration of plastification of PCE11 is shown. In combination with all tested cements, the
duration of the plasticizing effect was given within the test period of 120 min after water
addition.
On the right side of Figure 6, the results for PCE22 are shown. The TOC contents of the
pore solutions increased in course of the testing period (not shown), indicating desorption
of PCE22 (upper right side of Figure 6). Either PCE22 desorbed14 or more probably, as
the SO42- concentration remained almost unchanged, PCE22 lost the grafted side chains
by hydrolysis,32 both leading to an increased TOC content in the pore solution. On the
lower right side of Figure 6, the particular duration of plastification of PCE22 is shown.
As a result of the desorption process or rather the loss of side chains, the effect of the
steric hindrance decreased considerably and the fresh cement pastes exhibited significant
stiffening.
Performance of fresh concrete
The results obtained by the investigations on fresh cement pastes were verified by
concrete tests. Therefore, CEM I (circle) and Z55Q21 (triangle) were used in combination with PCE11 (filled symbols) and PCE22 (open symbols). The results are presented
in Figure 7. The results of concrete tests confirmed the results of the cement paste trials.
Without the addition of superplasticizer, the concrete with Z55Q21 exhibited a softer
consistency than the concrete with CEM I (Figure 7, left side, grey shaded symbols). This
was largely due to the improved particle size distribution (cf. slope n, Table 2) and thus the
improved packing density of the cement with calcined clay as well as the reduced binding
of mixing water into initial hydration products. Hence, more free water was available
to fluidize the fresh cement paste and the fresh concrete, respectively. The slightly larger
310SP-302-23
Figure 7Concretes consistency increase and consistency

retention in dependence of the cement-PCE-combination
used.
cement paste volume caused by the slightly lower specific density of the cement with
calcined clay (cf. Tables 1 and 3) also accounted for this.
The expected performance of both polycarboxylates observed in combination with CEM
I was scarcely present in combination Z55Q21. The fresh concrete with Z55Q21 could be
plasticized with lower dosages of superplasticizer to greater extent (Figure 7, left side).
Both polycarboxylates lost their characteristic performance and their plasticizing effect
was almost identical. This was particularly the case for PCE11 (ready-mixed concrete).
The reasons were explained previously. The consistency class F4 with a flow table spread
up to 550 mm ( 21 in.) was already reached by approximately 0.09 mass % of PCE11 and
PCE22, respectively, for the concrete with Z55Q21 compared to approximately 0.15 mass
% of PCE22 or approximately 0.20 mass % of PCE11 needed for the fresh concrete with
CEM I.
On the right side of Figure 7, the particular consistency retention is shown. In combination with CEM I, both polycarboxylates exhibited their distinct performance. PCE11
retained concretes consistency and PCE22 showed a significant initial plasticizing effect
combined with pronounced stiffening. In combination with Z55Q21, the specific performance of both polycarboxylates was diminished (cf. arrows on the right side of Figure 7).
Using pore solution analyses, zeta potential experiments, as well as rheological and
consistency measurements on fresh cement paste and fresh concrete influences of the type
and proportion of calcined clay in the cement on the sorption behavior and plasticizing
effect of polycarboxylate-based superplasticizers were determined. Based on the results
presented herein, the following conclusions can be drawn:
1. With an increasing substitution of the portland cement component (readily soluble
alkalis and sulfates) with calcined clay, the fresh cement pastes pore solution contained
less dissolved alkali, hydroxide, and sulfate ions as well as more calcium ions. The ionic
strength decreased. The zeta potential of the particular fresh cement paste was less negative. Changes in the ionic composition and ionic strength of the pore solution and the zeta
potential of the fresh cement paste can influence the constitution, the sorption behavior
and the performance potential of polycarboxylates depending on their particular polymer
structure.
Interactions between Cements with Calcined Clay and

2. Depending on the mineralogical composition of the clay and the calcining conditions,
the specific surface area of the calcined clay and thus the quantity of superplasticizer necessary to reach the maximum flowability of fresh cement paste varied.
3. Decreasing concentrations of SO42- in the pore solution and less negative zeta potentials of fresh cement paste as well as an increased specific surface area depending on the
type and proportion of calcined clay in the cement led to a stronger sorption and consequently to a stronger plasticizing effect of the less anionic superplasticizer recommended
for the use in ready-mixed concrete (PCE11) and hence to a reduced consistency retention
of fresh cement paste and fresh concrete, respectively.
4. The more anionic superplasticizer for the use in precast concrete (PCE22) sorbed to a
greater extent even at high SO42- concentrations inducing always a strong initial plastification in combination with significant stiffening of fresh cement paste and fresh concrete,
respectively.
5. The workability of fresh concrete containing cement with a clinker substitution of
up to 55 mass % by calcined clay could be improved compared to that of the reference
concrete with portland cement due to the higher volume of fresh cement paste as well as the
rheological benefit of more free water caused by an increased cements packing density
and less hydration products formed initially. Less superplasticizer was necessary for plastification and the specific performance of the polycarboxylates diminished increasing
proportion of calcined clay in the cement.
6. For an ideal plastification of concrete, the type and dosage of the superplasticizer
always have to be adjusted to concrete technological parameters such as the cement type
and content, the w/c ratio, the quantity of fresh cement paste, the mixing process, the
temperature, and the time until the concrete is placed.
AUTHOR BIOS
Civil engineer Jens Herrmann was awarded a scholarship from the Science Foundation
of the German Cement Industry, Gerd Wischers Foundation, at VDZs Concrete Technology Department. Since 2012 he has been working as a Scientific Assistant at VDZs
Cement Chemistry Department. He is a Ph.D student in material sciences at Clausthal
University of Technology, Germany. His research interests include the workability of
mortar and concrete, the performance of blended cements and their interactions with
fluidizing and water-retaining admixtures.
Jrg Rickert is a civil engineer and has been working at VDZ since 1996. He started his
career working in the Concrete Technology Department and obtained a Ph.D in materials science in 2003. His topics of research have been cements and admixtures and their
interactions in concretes. Jrg Rickert has been Head of the Cement Chemistry Department since 2009. He is a member of both national and international standardization
committees.
ACKNOWLEDGMENTS
The IGF project 16726 N of VDZ gGmbH was sponsored by the Federal Ministry of
Economics and Energy through the Federation of Industrial Cooperative Research Associations as part of the program to promote the Joint Industrial Research (IGF) based on a
decision of the German Bundestag.
312SP-302-23
REFERENCES
1. Schrter, N. and Fischer, P., Entwicklungen und Trends bei Betonzusatzmitteln,
beton, V. 60, No. 6, 2010, pp. 226-231.
2. CEMBUREAU, ed., CEMBUREAU statistics, 2012.
3. Scrivener, K. L., Options for the future of cement, Indian Concrete Journal, V. 88,
No. 7, 2014, pp. 11-21.
4. Aitcin, P.-C.; Jolicoeur, C.; and MacGregor, J. G., Superplasticizers: How they work
and why they occasionally dont, Concrete International, V. 16, No. 5, May 1994, pp.
45-52.
5. Andersen, P. J.; Roy, D. M.; and Gaidis, J. M.W.R. Grace & Co.. , The effects of
adsorption of superplasticizers on the surface of cement, Cement and Concrete Research,
V. 17, No. 5, 1987, pp. 805-813. doi: 10.1016/0008-8846(87)90043-3
6. Ferrari, L.; Kaufmann, J.; Winnefeld, F.; and Plank, J., Multi-method approach
to study influence of superplasticizers on cement suspensions, Cement and Concrete
7. Flatt, R. J., Interparticle Forces and Superplasticizers in Cement Suspensions,
Ph.D theses, cole Polytechnique Fdrale de Lausanne, EPFL, Dpartment de Matriaux,
Lausanne, 1999, 301 pp.
8. Hirsch, C. M., Untersuchungen zur Wechselwirkung zwischen polymeren Fliemitteln und Zementen bzw. Mineralphasen der frhen Zementhydratation, Ph.D theses, TU
Mnchen, Mnchen, 2005, pp. 261.
9. Jolicoeur, C., and Simard, M.-A., Chemical admixture-cement interactions: Phenomenology and physio-chemical concepts, Cement and Concrete Composites, V. 20, No.
2+3, 1998, pp. 87-101. doi: 10.1016/S0958-9465(97)00062-0
10. Regnaud, L.; Rossino, C.; Alfani, R.; and Vichot, A., Effect of comb type superplasticizers on hydration kinetics of industrial Portland cements, in: Palomo, A., Zaragoza, A.,
and Lpez, A. (eds.), 13th International Congress on the Chemistry of Cement (Madrid,
04.-08.07.2011), Abstracts and Proceedings, 7 pp.
11. Spanka, G., Grube, H., and Thielen, G., Wirkungsmechanismen verflssigender
Betonzusatzmittel, beton, V. 45, No. 11/12, Nov./Dec. 1995, pp. 802-808/876-881.
12. Uchikawa, H.; Hanehara, S.; Shirasaka, T.; and Sawaki, D., Effect of admixture on
hydration of cement, adsorptive behaviour of admixture and fluidity and setting of fresh
cement paste, Cement and Concrete Research, V. 22, No. 6, 1992, pp. 1115-1129. doi:
10.1016/0008-8846(92)90041-S
13. Uchikawa, H.; Hanehara, S.; and Sawaki, D., Effect of electrostatic and steric repulsive forces of organic admixtures on the dispersion of cement particles in fresh cement
paste, Cement and Concrete Research, V. 27, No. 1, 1997, pp. 37-50. doi: 10.1016/
S0008-8846(96)00207-4
14. Yamada, K., and Hanehara, S., Interaction mechanism of cement and superplasticizers The roles of polymer adsorption and ionic conditions of aqueous phase, Concrete
Science and Engineering, V. 3, No. 11, Sep. 2001, pp. 135-145.
15. Yamada, K., and Hanehara, S., Working mechanism of polycarboxylate superplasticizer considering the chemical structure and cement characteristics, in: Grieve, G.
and Owens, G. (eds.), 11h International Congress on the Chemistry of Cement (Durban,
11.-16.05.2003), Durban: Halfway House, 2003.
Interactions between Cements with Calcined Clay and

16. Flatt, R. J., and Schober, I., Superplasticizers and the rheology of concrete, in:
Roussel, N. (ed.), Understanding the rheology of concrete, Woodhead publishing, 2012,
pp. 144-208.
17. Sakai, et al., Influence of molecular structure of comb-type and inorganic electrolytes on the dispersion mechanisms of limestone power, in: Malhotra, V.M. (ed.), 7th
(Berlin, 20.-23.10.2003), Farmington Hills, MI, USA: American Concrete Institute, ACI,
2003, Special Publication SP 217, pp. 381-392.
18. Palacios, M. et al., Compatibility of PC superplasticizers with slag-blended cements,
in: Holland, T.C., Gupta, P., and Malhotra, V.M. (eds.), 9th International Conference on
Superplasticizers and Other Chemical Admixtures in Concrete (Seville, 12.-15.10.2009),
Farmington Hills, MI, USA: American Concrete Institute, ACI, 2009, ACI Special Publication SP 262, pp. 97-112.
19. Regnaud, L. et al., Interactions between comb-type superplasticizers and slag
cement pastes, in: Holland, T.C., Gupta, P., and Malhotra, V.M. (eds.), 9th International
Conference on Superplasticizers and Other Chemical Admixtures in Concrete (Seville,
12.-15.10.2009), Farmington Hills, MI, USA: American Concrete Institute, ACI, 2009,
Special Publication SP 262, pp. 139-151.
20. Herrmann, J., and Rickert, J., Influences of Slag or Limestone on the Performance
of Superplasticizers, in: Malhotra, V.M. (ed.), 10th International Conference on Superplasticizers and Other Chemical Admixtures in Concrete (Prague, 29.-31.10.2012), Farmington Hills, MI, USA: American Concrete Institute, ACI, 2012, ACI Publication SP-288,
pp. 317-328.
21. Herrmann, J., and Rickert, J., Influences of slag or fly ash on cement-superplasticizer-interactions, in: Gesellschaft Deutscher Chemiker, GDCh, (ed.), 1st International
Conference on the Chemistry of Construction Materials (Berlin, 07.-09.10.2013), Frankfurt / M.: GDCh, GDCh-Monographie Bd. 46, pp. 457-460.
22. Ng, S., and Plank, J., Interaction mechanisms between Na-montmorillonite clay and
MPEG-based polycarboxylate superplasticizers, Cement and Concrete Research, V. 42,
23. Lei, L., and Plank, J., Synthesis, properties and evaluation of a more clay tolerant
polycarboxylate possessing hydroxy alkyl graft chains, in: Malhotra, V.M. (ed.), 10th
(Prague, 29.-31.10.2012), Farmington Hills, MI, USA: American Concrete Institute, ACI,
2012, Supplementary Papers, pp. 1-20.
24. Lei, L., and Plank, J., A study on the impact of different clay minerals on the
dispersing force of conventional and modified vinyl ether based polycarboxylate superplasticizers, Cement and Concrete Research, V. 60, No. 6, 2014, pp. 1-10. doi: 10.1016/j.
25. Nehdi, M. L., Clay in cement-based materials: Critical overview of state-ofthe-art, Construction & Building Materials, V. 51, 2014, pp. 372-382. doi: 10.1016/j.
conbuildmat.2013.10.059
26. Schulze, S. E., and Rickert, J., Pozzolanic Activity of calcined clays, in: Malhotra,
V.M. (ed.), 12th International Conference on Recent Advances in Concrete Technology and
314SP-302-23
Sustainability Issues (Prague, 30.10.-02.11.2012). Farmington Hills, MI, USA: American

Concrete Institute, ACI, 2012 (ACI Publication SP-289), pp. 277-287.
27. Dukhin, A. S., and Goetz, P. J., Ultrasound for characterizing colloids - Particle
sizing, zeta potential, rheology, in: Mbius, D. and Miller, R. (eds.), Studies in Interface
Sciences Vol. 15. Elsevier, Amsterdam, 2002, 372 pp.
28. Flatt, R. J., and Houst, Y. F., A simplified view on chemical effects perturbing the
29. Aitcin, P.-C., Superplasticizers, in: HBRC (ed.) 1st International Conference on
New Cements and their Effects on Concrete Performance (Cairo, 16-18.12.2008), Cairo:
Helwan University, 2008, 14 pp.
30. Diamond, S., The Status of Calcium in Pore Solutions of Mature Hardened Portland
Cement Paste, Il cemento, V. 74, No. 4, 1977, pp. 149-156.
31. Sachsenhauser, B. K., Kolloidchemische und thermodynamische Untersuchungen
zur Wechselwirkung von Alpha-Allyl-Omega-methoxypolyethylenglykol-Maleinsureanhydrid-Copolymeren mit CaCO3 und Portlandzement, Dissertation, Technische Universitt Mnchen, Mnchen, 2009, 272 pp.
32. Flatt, R. J., and Schober, I., Superplasticizers and the rheology of concrete, in:
Roussel, N. (ed.), Understanding the rheology of concrete, Woodhead publishing, 2011,
pp. 144-208.
SP-302-24
Modification of Fresh State Properties of

Portland Cement-Based Mortars by Guar
Gum Derivatives
by Alexandre Govin, Marie-Claude Bartholin, and
Philippe Grosseau
Viscosity-modifying admixtures (VMA) are often introduced in the formulation of modern
factory-made mortars in order to prevent segregation and to improve the homogeneity and
workability of cement-based system. Among VMAs, organic admixtures, and more especially polysaccharides such as cellulose ethers (CE), are widely used, since they improve
both rheological property and water retention capacity of the mortars.
The present study examines the influence of chemical composition and structure of guar
gum derivatives on water retention capacity (WR) and rheological behavior of fresh state
Portland-based mortars. The investigation was also completed by adsorption isotherms.
For this, original guar gum, HydroxyProplyl Guars (HPG) and hydrophobically modified
HPGs were selected. The effect of the molar substitution (MSHP) and the degree of substitution (DSAC) was investigated. The results highlight that chemical composition of HPGs
has a remarkable effect on fresh state properties of mortars. The original guar gum does
not impact both WR and rheological behavior. Increasing MSHP leads to an improvement
of the WR and the stability of mortars while the hydrophobic units further enhance WR
and lead to a decrease in the yield stress and an increase in the resistance to the flow of
admixed mortars.
Keywords: cement; HydroxyPropyl Guar; mortar; rheology; water retention.
INTRODUCTION
Modern factory-made mortars are complex materials, in which several kinds of admixtures are added in order to obtain specific properties, from the fresh state to the hardened
material. Indeed, since many years, concretes, mortars or cement grouts with high fluidity
have been developed, since their use implies many economical and technical advantages.
However, the use of highly flowable mixtures may lead to segregation or excessive bleeding
and subsequently, durability issues. In order to overcome this problem by enhancing the
315
316SP-302-24
sedimentation resistance while maintaining high fluidity, viscosity-enhancing admixtures

(VEA) are frequently introduced within the formulation.1-4 Among these admixtures,
natural polysaccharides or their derivatives (such as welan gum, starch derivatives or cellulose ethers) are the most widely used. Moreover, the incorporation of these VEAs in shotcrete or render mortar is useful to ensure sagging resistance for thick application on vertical
support, and to allow sufficient fluidity for normal pumpability by supplying shear thinning
rheological behavior.5 Indeed, these admixtures provide, generally, high yield stress and
apparent viscosity at low shear rate but low resistance to flow at high shear rate.6 However,
their mode of action is not fully understood, since results are sometimes contradictory.
Water retention (WR) is another essential property of monolayer render at fresh state.
Indeed, high water retention improves the cement hydration and limits the absorption of the
mixing water by a substrate and thus provides good mechanical and adhesive properties to
the mortar.7,8 Among admixtures enhancing water retention capacity of the freshly-mixed
mortars, cellulose ethers (CE) are the most widely used. Nevertheless, hydroxypropyl guar
(HPG) are now also well-established in the construction industry as water retention agent
for mortars.9-12 Moreover, HPGs are already widely used in various industrial fields, such
as textile printing, hydraulic fracturing process, oil production or paper manufacturing, due
to their thickening effect.13,14 Consequently, since HPGs improve the two main properties
of mortar, they appear as suitable admixtures to be used in render formulation.
The aim of this study is to provide an understanding of the effect of chemical composition and structure of HPGs and its dosage on macroscopic properties of mortars. For this
purpose, an original guar gum and five HPGs with specific chemical modifications, such
as increase in MSHP or substitutions by hydrophobic units, were selected. The impact of
admixtures on the water retention capacity and on the rheological behavior of mortars was
investigated.
Polysaccharides are commonly used in cement-based materials, but most studies focus
on cellulose ethers or welan gum. Studies about hydroxypropyl guars are still scarce in
the technical literature despite the fact that these molecules provide interesting properties
comparable to those obtained with cellulose ethers. The aim of this study is to highlight
the role of the chemical structure of hydroxypropyl guars on water retention properties and
rheological properties.
Mineral products
Mineral products used in this study consist in blend Portland cement (Holcim), lime
(Holcim), calcium carbonate (Calcitec V60, Mineraria Sacilese S.p.A.) and dolomite
(Bombardieri and Leidi 0.1-0.4mm). The mineral compositions of the commercial Portland cement, CEM II/B-LL 32.5 R according to the European standard EN 197-115, used
are given in Table 1.
The phase composition was determined by Rietveld refinement method (Siroquant V2.5
software) after XRD analysis (D5000, Siemens) and the oxide composition was quantified
by means of X-ray fluorescence spectroscopy. The median particle diameters by volume
(d50%), determined by means of laser diffractometry with dry powder disperser, (Master-
Modification of Fresh State Properties of Portland Cement-Based Mortars

by Guar Gum Derivatives 317
Table 1 Mineral composition (%, weight) of the investigated cement determined by XRF and XRD-Rietveld refinement
Chemical composition (% wt)
Phase composition (% wt)
Oxides
CaO
XRF
57.87
Oxides
SO3
XRF
3.95
Phases
C3S
XRD
(Rietveld)
54.3
Phases
Calcite
XRD
(Rietveld)
28.9
SiO2
12.31
Na2O
0.99
C2S
3.5
Gypsum
3.0
Al2O3
5.25
K2O
1.66
C3A
4.7
Quartz
0.9
MgO
1.19
TiO2
0.16
C4AF
4.6
Free CaO
0.8
Fe2O3
4.05
LOI
13.7
Fig. 1 Particle size distribution of raw materials constituting the mortar

Table 2 Median particle diameters by volume (d50) and specific surface
area of the mineral phases
d50 (m)
BET specific surface
area (m2/g)
CEM II/B-LL
15
Lime
5
Dolomite
Leidi
300
Dolomite
Bombardieri
630
Calcite
20
2.40
5,67
0.43
0.34
1.44
(Note: 1 m = 0.0000394 in, 1 m2/g = 0.00488 ft2/lbm)
sizer 2000 and Scirocco dispersing unit, Malvern), are 630 m (248 10-4 in), 300 m (118
10-4 in), 20 m (7.9 10-4 in), 15 m (5.9 10-4 in) and 5 m (1.9 10-4 in) for the dolomite
from Bombardieri, dolomite from Leidi, calcium carbonate, cement and lime, respectively.
The particle size distribution and the specific surface area (determined by BET) are given
in Fig. 1 and Table 2.
318SP-302-24
Fig. 2 Molecular structure of original guar gum (a) and HydroxyPropyl Guar (b)
Table 3 Qualitative description of the HPG used
MS
DS
Additional substitution
HPG 1
Low
HPG 2
Medium
HPG 3
High
HPG 4
High
Short alkyl chain
HPG 5
HPG 6
High
-
Higher DS than HPG 4

-
Short alkyl chain

-
Organic admixtures
Guar gum is a natural polysaccharide extracted from the seed endosperm of Cyamopsis
tetragonolobus. This polymer consists in a (1-4)-linked D-mannopyranose backbone with
random branchpoints of galactose via an (16) linkage (Fig.2(a)). Hydroxypropyl guars
(HPGs) are obtained from the original guar gum via an irreversible nucleophilic substitution, using propylene oxide in the presence of an alkaline catalyst (Fig.2(b)). The manufacture of HPGs has the advantage of having a more reduced impact on the environment
than cellulose derivatives. Indeed, guar gum is extracted by simple thermo-mechanical
process, exhibits a higher chemical reactivity and is soluble in cold water thanks to its
branched-chain structure with a lot of hydroxyl groups. Thus, the chemical modification
of the original guar gum requires normal reaction conditions of temperature and pressure,
does not generate large quantity of by-products, and requires minimal purification procedure.9 In this paper, five HPGs and an original guar gum provided by Lamberti S.p.A were
studied. They exhibit roughly the same molecular weight, around 2.106g.mol-1 since they
are all from the same original guar gum (noted HPG 6 in the paper).16 Table 3 provides
a qualitative description of the polymers used. The qualitative substitution degrees are
provided by the manufacturers. The molar substitution ratio (MSHP) represents the number
of hydroxypropyl units per anhydroglucose unit and is less than 3 for the investigated
HPGs. The degree of substitution (DSAC) represents the amount of alkyl chain per anhydroglucose unit. The only difference between HPGs 1, 2 and 3 is the molar substitution ratio,
which increases, while HPGs 4 and 5 exhibit an additional substitution (short alkyl chains).
The DSAC of HPG 5 is higher than that of HPG 4.

Methodology
Mortars were prepared according to the following mixture proportions: 12% of cement,
3% of lime, 18% of calcium carbonate, 43% of dolomite Bombardieri and 24% of dolomite
Leidi (by weight). The admixtures (0.05, 0.075, 0.1, 0.125 and 0.15%) were in addition to
the total dry mixture (i.e. cement, lime, calcium carbonate and dolomite) and are expressed
in weight percent by weight of binder (% bwob). Dry mixture was blended in a shaker
(Wab, Turbula, Germany) for 10 min. Deionised water was added in order to obtained a
liquid-to-solid ratio L/S = 0.22. The mixing procedure was in accordance with EN 196-1.17
The experimental methodology consisted in dividing the freshly mixed mortar into three
parts in order to characterize several properties from the same mixing. A first part was
used to characterize the rheological behavior of the mortar, the water retention study was
performed on the second part and the third part of the freshly mixed mortar was centrifuged
in order to determine the adsorption isotherms and the polymer concentration within the
pore solution following a procedure described later.
All tests were carried out, at least, in triplicate and at a controlled temperature because
water retention, rheological behavior of the mortar and adsorption isotherm are temperature-dependent. A control test was also performed with a mortar without admixture.
Water retention measurements
The water retention capacity of freshly-mixed mortar can be assessed using different
tests where the removed water after suction or depression is measured.18 In this study, the
standard method used to estimate the water retention capacity of a mortar, was the test
described in ASTM C1506-09.19 It had to be performed 15 min after mixing to measure
the water loss of a mortar under depression. The standardized apparatus was submitted to a
vacuum of 50 mm of mercury (6.6 103 Pa) for 15 min. Then, the water retention capacity,
WR, was calculated using the following equation:

WR(%) =
W0 W1
100 (1)
W0
where W0 represents the initial mass of mixing water; W1 is the loss of water mass after
aspiration.
All the experiments were carried out at 23 C (73.4 F). Three classes of water retention
(measured by ASTM method) of a fresh mortar can be specified according to the DTU
26.1.20 The first class (low water retention category) contains mortars that exhibit a water
retention lower than 86%. The second class (intermediate) corresponds to values ranging
from 86% to 94%. The last one (strong) is defined by water retention higher than 94%,
corresponding to the required values in the field of rendering application.
Rheological behavior
The rheological measurements were performed with Rheometer MCR 302 (Anton-Paar),
thermostated at 20 C (68 F). The rheological properties of fresh mortars were investigated with vane-cylinder geometry since this system is suitable for granular pastes like
mortars.21,22 The gap thickness, distance between the periphery of the vane tool and the
outer cylinder, was set at 8.5 mm (0.33 in), in order to be less sensitive to the heterogeneity
320SP-302-24
of the mortar. Using a Couette analogy, the shear stress and shear rate were calculated from
the torque and the applied rotational velocity respectively, after calibration with glycerol.23
The mortar was introduced into the measurement system at the end of the mixing cycle. At
10 min, the mortar was pre-sheared for 30 s at 100 s-1 in order to re-homogenize the sample
and to eliminate its shear history because of thixotropic character of cementitious materials.24,25 After a period of rest of 5 min, the rheological measurements were started (total
time = 15 min). At this time, the hydration rate is low enough which allows overcoming the
irreversible effect of cement hydration on rheological behavior, especially at low shear
rate.24 The imposed shear rate was decreased by step from 300 to 0.06 s-1 (16 steps). At
each shear rate, the measuring time was adjusted in order to obtain a steady state whatever
the formulation. The samples were systematically submitted to high shear rate (100 s-1) for
30 s before each imposed shear rate in order to resuspend particles of mortar within the
mortar mixtures. The results were expressed as shear stress according to shear rate and
the Herschel-Bulkley (HB) model was applied to fit the experimental data and used to
describe mortars rheological behavior26:

= 0 + K n (2)
where 0 correspond to the yield stress, K the consistency coefficient and n the fluidity
index which characterizes shear-thinning behavior of mortar.
Adsorption curves of HPGs on binder
The adsorption isotherms were determined using the depletion method. The nonadsorbed polymer remaining within the pore solution was quantified by means of Total
Organic Carbon (TOC) measurements. Prior to analysis, the pore solution was extracted
from admixed or non-admixed mortar. The extraction was performed by means of two
centrifugation steps. The first step consisted in the centrifugation of around 150 g (0.30
lbm) of mortar at 5000 rpm for 5 min. The supernatant was, afterward, centrifuged again at
14500 rpm for 10 min in order to avoid the presence of mineral particles within the solution. The supernatant was diluted with hydrochloric acid solution at 0.1 mol.L-1 (0.378 mol.
gal-1). The total organic carbon was determined by combustion at 850 C (1562 F) with
a Vario-TOC Cube (Elementar). The adsorbed amount of polysaccharides was calculated
from the difference of TOC content of the HPG reference solution and the TOC content of
the supernatant.
Impact of HPGs on the water retention property of fresh mortars
Fig. 3 represents the evolution of the water retention capacity of fresh admixed mortars,
according to the polymer dosage. The non-admixed mortar exhibits a low water retention
capacity of about 72% 0.3%. Then, as expected, the water retention increases with the
use of HPGs and with increasing polymer dosage, until reaching a plateau with very high
WR values (>97%). In the range of polymer dosage studied, the WR values reached for
HPGs 2, 3, 4 and 5, are greater than 94% and therefore belong to the strong WR class. One

Fig. 3 Impact of polymer dosage on water retention

capacity of fresh admixed mortars
can also clearly notice the very limited impact of the original guar gum and, to a lesser
extent, the one of HPG 1, regardless of dosage. These results suggest that the substitution of hydroxyl units from original guar gum by hydroxypropyl units increases the WR
of mortars. Furthermore, the increase in the MSHP (from HPG 1 to 3) improves the WR
capacity of mortar, since HPG 3 provides the higher WR despite lower dosage, followed
by HPG 2 and then by HPG 1.
The results highlight moreover the positive impact of additional alkyl chain on WR.
Indeed, the highest WR are obtained with HPGs 4 and 5 for the lowest polymer dosages.
Moreover, concerning the shape of WR curves, an abrupt change in slope can be noticed
for mortars admixed with HPGs 2 to 5. This occurs for a decreasing polymer dosage from
HPG 2 to HPG 3 and from HPG 3 to HPG 5.
Adsorption curves of HPGs on binder
Fig. 4 shows the adsorption isotherms of the hydroxypropyl guar and the original guar
gum on Portland based-mortars. The results confirm the adsorption of original guar gum
and HPGs on cementitious materials. It has been shown that the adsorption mechanism
of galactomannose polysaccharides at solidliquid interfaces involves strong hydrogen
bonding.27 In the range of polymer dosage used in the present study, no plateau is reached,
and this, whatever the admixture. For the original guar gum (HPG 6), the adsorption is
totally linear (Fig. 4(a)). Moreover, its adsorption is the highest of all the tested polymers
and corresponds to a total adsorption higher than 98.5% of the introduced polymer.
The presence of hydroxypropyl substitutions on the guar leads to a decrease in the affinity
of the polymer with the binder since the amount of HPG 1 adsorbed is lower by 35% than
the original guar gum. Moreover, the adsorption is further reduced by the increasing values
of MSHP by 46% and 64% for HPG 2 and HPG 3, respectively, with respect to HPG 6
(Fig. 4(a)). This tendency is consistent with previous studies on HPGs and cellulose ethers
(CE).10,28
Fig. 4(b) highlights the effect of the additional alkyl chain on the adsorption. It appears
that the hydrophobic side chains slightly intensify the adsorption of the hydrophobically
322SP-302-24
Fig. 4 Adsorption isotherms of HPGs 1 to 3 and original

guar gum (HPG 6) (a) and HPGs 3 to 5 (b) on binder
modified HPGs on surface of grains with respect to HPG 3. However, the adsorption of
HPGs 4 and 5 is lower than that of HPG 2. For polymer dosages up to 0.1% bwob, the effect
of the DSAC is negligible since the experimental data superimpose. Nevertheless, when
HPG dosage is higher than 0.1% bwob, the affinity of HPG 5 with the binder becomes
higher than that of HPG 4.
From the TOC measurements, the real polymer concentration within the extracted pore
solution was determined. Fig. 5 shows the evolution of this concentration versus the introduced polymer dosage. Excepted HPG 6, the amount of non-adsorbed polymer increases
with increasing polymer dosage. According to the HPG, the concentration rises following
this order: HPG 6 < HPG 1 < HPG 2 < HPG 5 < HPG 4 < HPG 3.
Impact of HPGs on the rheological properties of fresh mortars
The rheological results (not shown here) suggest that the thixotropy of the mortars
does not affect the rheological measurements. Indeed, in the range of tested shear rates
and thanks to the experimental procedure, the shear stress obtained by the increasing or
decreasing shear rate ramps are superimposed which justifies the choice to consider only
the decreasing ramps for all the rheological study.

Fig. 5 Concentration of HPGs in extracted mortar pore

solution
Fig. 6 shows the evolution of the yield stress, extracted from Herschel-Bulkley model,
for all the studied mortars with and without admixture. The mortar without admixture
exhibits a yield stress value of around 45 Pa. From the presented results, three different
classes of HPG, inducing different evolution of the yield stress with the polymer dosage,
can be highlighted for admixed mortars. The first category is only composed of the original
guar gum (HPG 6), which induces a quasi linear decrease in the yield stress of mortar
when HPG dosage increases. On the contrary, HPGs 1, 2 and 3 lead to a continuous rise of
the yield stress of mortars from 50-60 Pa to around 120 Pa with the increase in the HPG
dosage from 0.05% to 0.15%. Finally, HPG 4 and 5 constitute the third class of admixture.
The use of these admixtures leads an improvement of the yield stress compared to the nonadmixed mortar, whatever the dosages tested in the study. However, the improvement is
not proportional to the admixture dosage. Indeed, our first dosage (0.05% bwob) leads to
an increase in the yield stress. Beyond this dosage, increasing the dosage provides a slow
and low decrease, before reaching a plateau. The value of the yield stress reached on the
plateau is still higher than that of the mortar without admixture.
The evolution of the consistency coefficient (K from Herschel-Bulkley equation) during
the increase of polymer dosage is presented in Fig. 7. As in the case of the yield stress, the
results can be divided into three classes of polymer. The first class is only composed of the
original guar gum which provides a very low or negligible modification of the consistency
coefficient with increasing polymer dosage compared to non-admixed mortar. HPGs 1 to 3,
constituting the second group, induce first an increase followed by a plateau in the consistency coefficient. Finally, HPGs 4 and 5 lead to a continuous increase in the consistency
coefficient of admixed mortars.
Fig. 8 shows the evolution of the fluidity index (n) versus the polymer dosage for all the
studied mortars. It is worth to note that whatever the mortars (non-admixed and admixed),
the values of the fluidity index are lower than 1, meaning that they are all shear thinning.
Due to the high standard deviation, the value of the fluidity index of mortars admixed with
HPGs 1 to 3 and HPG 6 seem to be unchanged as the dosage of HPGs increase. However,
HPGs 4 and 5 leads to a low increase followed by a continuous decrease in the fluidity
324SP-302-24
Fig. 6 Impact of polymer dosage on yield stress of fresh

admixed mortars (HPGs 1 to 3 and original guar gum (HPG
6) (a) and HPGs 3 to 5 (b))
index until reaching values around 0.5. It means that the shear thinning behavior of mortars
becomes more and more pronounced.
DISCUSSION
The effect of the original guar gum (HPG 6) on the studied macroscopic properties is
negligible. This result is coherent with adsorption of the polymer from the Water Retention
point of view. Indeed, since the adsorption is higher than 98.5% of the initial amount of
polymer, very few molecules are still in the pore solution. The composition of pore solution
is thus very close to that of the non-admixed mortar, leading to similar WR.
Concerning the hydroxypropyl guar, the results from WR experiments are consistent with those of previous studies performed with HPGs or CEs and with the proposed
mechanism.11,29,30 Indeed, the WR of admixed mortars is mainly governed by the ability
of polysaccharidic admixtures to form a hydrocolloidal associated polymer molecules
network and to induce overlapping of polymer coils within the pore solution.11,29,30 When
the concentration of polymer increases in solution, the isolated polymer coils, existing at
low polymer concentration, begin to come into contact with one another. This concentra-

Fig. 7 Impact of polymer dosage on consistency coefficient

of fresh admixed mortars (HPGs 1 to 3 and original guar
gum (HPG 6) (a) and HPGs 3 to 5 (b))
tion is defined as the coil-overlap concentration (noted C*). Above this critical concentration, the polysaccharide aggregates stop the water flow by plugging the porous network
of a thin polysaccharide-enriched filter cake at the interface mortar-substrate resulting in
a sudden and sharp rise in WR curves.11 As previously mentioned, the abrupt change in
slope is reached for a decreasing polymer dosage from HPG 1 to HPG 3. The only difference between these HPGs is the increasing substitution degree. According to literature, the
increase in MSHP does not lead to a change in the C*.31 However, the increasing substitution degree leads to a decrease in polymer adsorption on mortar components (Fig. 4) and
hence an increase in polymer amount in pore solution (Fig. 5). Consequently, the coil
overlapping occurs at lower dosage. The results highlight furthermore the positive impact
of additional alkyl chain on WR. The presence of additional alkyl chains (HPG 4 and 5),
despite slightly higher adsorption than HPG 3, leads to the formation of polymer associates at lower polymer dosage. Indeed, the interconnection between alkyl chains creates
intramolecular and intermolecular interactions through specific hydrophobic interactions
which cause a decrease in the coil-overlapping concentration.32-34 Consequently, the abrupt
change in slope is reached for a lower polymer dosage of HPG 4 and 5 than HPG 3.
326SP-302-24
Fig. 8 Impact of polymer dosage on fluidity index of fresh

admixed mortars (HPGs 1 to 3 and original guar gum (HPG
6) (a) and HPGs 3 to 5 (b))
However, an increase in the DSAC (from HPG 4 to 5) can lead to a conversion of some
intermolecular associations to intramolecular associations and hence an increase in the
polymer dosage necessary to reach coil overlap.31
The rheological results (Fig. 6, Fig. 7 and Fig. 8) highlight that HPGs 1 to 3, HPGs 4 to
5 and HPG 6 behave quite differently. Indeed, HPGs 1-2-3 lead to a continuous increase
in the yield stress, while HPGs 4-5 modify mainly the consistency coefficient and the
fluidity index. This means that HPGs 1-3 increase the stability of mortars while HPGs 4-5
increase the resistance to the flow of admixed mortars. HPGs 1 to 3 affect the rheological
behavior of the admixed mortars in the same way, i.e. an increase in the yield stress, a low
increase followed by a plateau in the consistency coefficient and a negligible modification
of the fluidity index when the polymer dosage rises. Fig 4 shows that HPGs adsorb onto
particles constituting the mortar. Prima facie, this adsorption could be responsible for the
increase in the yield stress because of bridging flocculation.28 However, despite a strong
drop of the adsorption (50%) with the increase in the MSHP (from HPG 1 to 3) the yield
stress also increases. This suggests that the non-adsorbed polymer may be responsible for
the yield stress increase. The potential loss of bridging can be compensated by an increase

in the pore solution viscosity induced by the rise in the polymer concentration and/or by the
depletion flocculation induced by the non-adsorbed coils (Fig. 5).35
Moreover, the presence of HPG coils within the pore solution leads to an increase in
the consistency coefficient (K) compared to non-admixed mortars. However, the expected
increase in K due to the rise of pore solution viscosity with the polymer dosage can be
compensated by steric hindrance, leading to a plateau for K.
Since the adsorption of HPG 6 onto the surface of the binder is higher than 98.5%,
one expects to detect a very strong increase in the yield stress compared to non-admixed
mortar. However, the rheological behavior of the admixed mortar with the original guar
gum is very close to that of the non-admixed mortar. This result suggests that the entire
molecule of the original guar gum could be mainly adsorbed onto the surface of only one
particle, limiting therefore the bridging flocculation. The very high concentration of free
hydroxide groups on the backbone of the guar could be responsible of this mechanism.
Moreover, the adsorption of the guar molecule onto a single particle leads to an increase
in the steric hindrance and in the dispersion and lubrication effects, leading to a low but
continuous decrease in the yield stress. For dosages higher than 0.1% bwob, the concentration in polymer coils (HPG 6) into the pore solution begins to increase slightly (Fig. 5),
leading to the beginning of the increase in the consistency coefficient and of the decrease
in the fluidity index.
The additional alkyl chain also modifies the rheological properties of mortars. Contrary
to HPG 3, HPGs 4 and 5 lead to a strong and continuous increase in the consistency coefficient and a decrease in the fluidity index. These results highlight that mortars become
more and more shear-thinning since the fluidity index decreases from 0.8 to 0.5. This rheological behavior gets more pronounced as the HPG dosage increases. These results are
consistent with the fact that the hydrophobically modified HPGs leads to the formation of
coil overlapping at lower HPG dosage (0.05% in this study) since the presence of additional alkyl chains enhances the entanglement. Above this dosage, entanglement occurs
between polymer coils, inducing a shear thinning behavior to the solution. At low shear
rate, the entanglement of polymer coils leads to a higher pore solution viscosity and thus
higher mortar viscosity. When the shear rate increases, the polysaccharide chains align in
the direction of the flow resulting in less and less effect on mortar fluidity. The shear thinning behavior of the solution, and thus of the mortars, amplify with the increasing polymer
dosage. The yield stress is also impacted by the additional alkyl chain. Indeed, 0, of mortar
admixed with HPGs 4 and 5, increases for a dosage equal to 0.05% bwob then slowly
decreases for dosages ranging from 0.05% to 0.075% bwob, before reaching a plateau for
higher dosages (0 reached is still higher than that of the non-admixed mortar). This result
could be explained by a change in the HPG conformation due to the presence of additional alkyl chains. Indeed, as previously mentioned, alkyl chain creates intramolecular
and intermolecular interactions through specific hydrophobic interactions. Intramolecular
association of hydrophobic units tends to force the polymer chain into a more compact
conformation.34,36 Moreover, the conformation of hydrophobically modified polymers in
aqueous solution involves the presence of alkyl chains mainly inside the coils in order
to limit contacts between hydrophobic chains and water.34 Consequently, the hydrophilic
groups, such as hydroxyl and hydroxypropyl, are preferentially on the outskirts of the
coils, promoting the adsorption onto the surface of particles (Fig. 4 (b)) and therefore steric
328SP-302-24
hindrance which implies a prevention of direct contacts between particles. Moreover, due
to a more compact conformation and avoidance between water and hydrophobic units,
the bridging ability of hydrophobically modified HPGs should decrease. All these points
should lead to a decrease in the yield stress.
CONCLUSIONS
In this paper, we studied the effect of several guar gum derivatives on water retention
property and rheological behavior of mortars. Based upon the results, it was found that
the original guar gum was totally adsorbed onto particle surface, leading to a negligible
modification of WR and rheological behavior with respect to the non-admixed mortar.
Depending of the chemical structure of HPGs, it is possible to promote the water retention according to two different ways. First, by increasing the MSHP of HPGs, the amount
of adsorbed polymer drops, which leads to an increase in the HPG concentration within
the pore solution and therefore to lower HPG dosage necessary to reach coil overlapping.
Second, by enhancing overlapping, the hydrophobically modified HPGs improve the effectiveness of WR agent at low dosage. HPGs also modify the rheological behavior of the
mortars. As in the case of WR, the hydrophobic characteristic of HPGs is the preponderant
parameter. Indeed, it was shown that additional alkyl chain mainly leads to a more shear
thinning behavior of the mortar and to a rise in the consistency coefficient, while classical
HPGs strongly increases the yield stress.
AUTHOR BIOS
Alexandre Govin: Assistant Professor at the cole Nationale Suprieure des Mines de
Saint-tienne, SPIN-EMSE, CNRS: UMR 5307, LGF, 158 Cours Fauriel, CS 62362,
42023 Saint-tienne Cedex 2, France, E-mail: govin@emse.fr
Marie-Claude Bartholin: Laboratory technician at the cole Nationale Suprieure des
Mines de Saint-tienne, SPIN-EMSE, CNRS: UMR 5307, LGF, 158 Cours Fauriel, CS
62362, 42023 Saint-tienne Cedex 2, France, E-mail: bartholin@emse.fr
Philippe Grosseau: Professor at the cole Nationale Suprieure des Mines de Sainttienne, SPIN-EMSE, CNRS: UMR 5307, LGF, 158 Cours Fauriel, CS 62362, 42023
Saint-tienne Cedex 2, France, E-mail: grosseau@emse.fr
ACKNOWLEDGMENTS
The authors would like to acknowledge Lamberti S.p.A. for the provided products and
for their technical and financial support.
REFERENCES
1. Rols, S.; Ambroise, J.; and Pra, J., Effects of different viscosity agents on the properties of self-leveling concrete, Cement and Concrete Research, V. 29, No. 2, 1999, pp.
261-266. doi: 10.1016/S0008-8846(98)00095-7
2. Khayat, K. H., and Yahia, A., Effect of welan gum-high-range water reducer combinations on rheology of cement grout, ACI Materials Journal, V. 94, No. 5, 1997, pp.
365-372.

3. Lachemi, M.; Hossain, K. M. A.; Lambros, V.; Nkinamubanzi, P.-C.; and Bouzouba,
N., Self-consolidating concrete incorporating new viscosity modifying admixtures,
Cement and Concrete Research, V. 34, No. 6, 2004, pp. 917-926. doi: 10.1016/j.
4. Sonebi, M., Rheological properties of grouts with viscosity modifying agents
as diutan gum and welan gum incorporating pulverised fly ash, Cement and Concrete
5. Paiva, H.; Silva, L. M.; Labrincha, J. A.; and Ferreira, V. M., Effects of a water-retaining
agent on the rheological behaviour of a single-coat render mortar, Cement and Concrete
6. Khayat, K. H., Viscosity-enhancing admixtures for cement-based materials an
overview, Cement and Concrete Composites, V. 20, No. 23, 1998, pp. 171-188. doi:
10.1016/S0958-9465(98)80006-1
7. Bertrand, L.; Maximilien, S.; and Guyonnet, R., Wedge Splitting Test: a test to
measure the polysaccharide influence on adhesion of mortar on its substrate, International
Congress on Polymers in Concrete, Berlin, 2004.
8. Jenni, A.; Holzer, L.; Zurbriggen, R.; and Herwegh, M., Influence of polymers on
microstructure and adhesive strength of cementitious tile adhesive mortars, Cement and
9. Biasotti, B.; Giudici, M.; Langella, V.; and Pfeiffer, U., Highly substituted hydroxypropylguar: a strong contribution to construction chemistry, International Dry mix Mortar
Conference, Nrnberg, 2011.
10. Poinot, T.; Govin, A.; and Grosseau, P., Impact of hydroxypropylguars on the early
age hydration of Portland cement, Cement and Concrete Research, V. 44, 2013, pp. 69-76.
doi: 10.1016/j.cemconres.2012.10.010
11. Poinot, T.; Govin, A.; and Grosseau, P., Importance of coil-overlapping for the effectiveness of hydroxypropylguars as water retention agent in cement-based mortars, Cement
and Concrete Research, V. 56, 2014, pp. 61-68. doi: 10.1016/j.cemconres.2013.11.005
12. Cappellari, M.; Daubresse, A.; and Chaouche, M.; Influence of organic thickening admixtures on the rheological properties of mortars: Relationship with waterretention, Construction & Building Materials, V. 38, 2013, pp. 950-961. doi: 10.1016/j.
13. Risica, D.; Barbetta, A.; Vischetti, L.; Cametti, C.; and Dentini, M.; Rheological
properties of guar and its methyl, hydroxypropyl and hydroxypropyl-methyl derivatives
in semidilute and concentrated aqueous solutions, Polymer, V. 51, No. 9, 2010, pp. 19721982. doi: 10.1016/j.polymer.2010.02.041
14. Lapasin, R.; De Lorenzi, L.; Pricl, S.; and Torriano, G.; Flow properties of hydroxypropyl guar gum and its long-chain hydrophobic derivatives, Carbohydrate Polymers, V.
28, No. 3, 1995, pp. 195-202. doi: 10.1016/0144-8617(95)00134-4
15. Standard EN 197-1, Cement Part 1: Composition, specifications and conformity
criteria for common cements, 2012.
16. Poinot, T.; Benyahia, K.; Govin, A.; Jeanmaire, T.; and Grosseau, P.; Use of ultrasonic degradation to study the molecular weight influence of polymeric admixtures for
mortars, Construction & Building Materials, V. 47, 2013, pp. 1046-1052. doi: 10.1016/j.
330SP-302-24
17. Standard EN 196-1, Methods of testing cement Part 1: Determination of strength,

2006.
18. Patural, L.; Marchal, P.; Govin, A.; Grosseau, P.; Ruot, B.; and Devs, O.; Cellulose
ethers influence on water retention and consistency in cement-based mortars, Cement and
19. Standard C1506-09, Standard test Method for Water Retention of Hydraulic
Cement-Based Mortars and Plasters. American Society for Testing and Material, 2009.
20. NF DTU 26.1, Travaux denduits de mortiers , 2008.
21. Bouras, R.; Kaci, A.; and Chaouche, M.; Influence of viscosity modifying admixtures on the rheological behavior of cement and mortar pastes, Korea-Australia Rheology
Journal, V. 24, No. 1, 2012, pp. 35-44. doi: 10.1007/s13367-012-0004-3
22. Barnes, H. A., and Nguyen, Q. D., Rotating vane rheometry - a review, Journal
of Non-Newtonian Fluid Mechanics, V. 98, No. 1, 2001, pp. 1-14. doi: 10.1016/
S0377-0257(01)00095-7
23. Ait-Kadi, A.; Marchal, P.; Choplin, L.; Chrissemant, A. S.; and Bousmina, M., Quantitative analysis of mixer-type rheometers using the Couette analogy, Canadian Journal of
Chemical Engineering, V. 80, No. 6, 2002, pp. 1166-1174. doi: 10.1002/cjce.5450800618
24. Phan, T. H.; Chaouche, M.; and Moranville, M., Influence of organic admixtures on
the rheological behaviour of cement pastes, Cement and Concrete Research, V. 36, No.
10, 2006, pp. 1807-1813. doi: 10.1016/j.cemconres.2006.05.028
25. Roussel, N.; Ovarlez, G.; Garrault, S.; and Brumaud, C., The origins of thixotropy
of fresh cement pastes, Cement and Concrete Research, V. 42, No. 1, 2012, pp. 148-157.
doi: 10.1016/j.cemconres.2011.09.004
26. Herschel, W. M., and Bulkley, R., Measurements of consistency as applied to
rubberbenzene solutions, Proceedings of the American Society for the Testing of Materials, V. 26, 1926, pp. 621-633.
27. Wang, J.; Somasundaran, P.; and Nagaraj, D. R., Adsorption mechanism of guar
gum at solidliquid interfaces, Minerals Engineering, V. 18, No. 1, 2005, pp. 77-81. doi:
10.1016/j.mineng.2004.05.013
28. Brumaud, C.; Baumann, R.; Schmitz, M.; Radler, M.; and Roussel, N., Cellulose
ethers and yield stress of cement pastes, Cement and Concrete Research, V. 55, 2014, pp.
14-21. doi: 10.1016/j.cemconres.2013.06.013
29. Blichen, D.; Kainz, J.; and Plank, J., Working mechanism of methyl hydroxyethyl
cellulose (MHEC) as water retention agent, Cement and Concrete Research, V. 42, No. 7,
30. Marliere, C.; Mabrouk, E.; Lamblet, M.; and Coussot, P., How water retention in
porous media with cellulose ethers works, Cement and Concrete Research, V. 42, No. 11,
31. Volpert, E.; Selb, J.; and Candau, F., Influence of the Hydrophobe Structure on
Composition, Microstructure, and Rheology in Associating Polyacrylamides Prepared by
Micellar Copolymerization, Macromolecules, V. 29, No. 5, 1996, pp. 1452-1463. doi:
10.1021/ma951178m
32. Cheng, Y.; Brown, K. M.; and Prudhomme, R. K., Characterization and Intermolecular Interactions of Hydroxypropyl Guar Solutions, Biomacromolecules, V. 3, No. 3,
2002, pp. 456-461. doi: 10.1021/bm0156227

33. Semenov, A. N.; Joanny, J.-F.; and Khokhlov, A. R., Associating polymers: equilibrium and linear viscoelasticity, Macromolecules, V. 28, No. 4, 1995, pp. 1066-1075. doi:
10.1021/ma00108a038
34. Simon, S.; Dugast, J.; Le Cerf, D.; Picton, L.; and Muller, G., Amphiphilic polysaccharides. Evidence for a competition between intra and intermolecular associations in dilute system, Polymer, V. 44, No. 26, 2003, pp. 7917-7924. doi: 10.1016/j.
polymer.2003.10.054
35. Palacios, M.; Flatt, R. J.; Puertas, F.; and Sanchez-Herencia, A., Compatibility
between Polycarboxylate and Viscosity-Modifying Admixtures in Cement Pastes,
Proceedings of the 10th International Conference on Superplasticizers and Other Chemical Admixtures in Concrete, Prague, 2012, pp. 29-42.
36. Aubry, T., and Moan, M., Rheological behavior of a hydrophobically associating
water soluble polymer, Journal of Rheology, V. 38, No. 6, 1994, pp. 1681-1692. doi:
10.1122/1.550566
332SP-302-24
SP-302-25
Interaction of Polycarboxylate-based
Superplasticizer/Poly(vinyl alcohol) with
Bentonite and Its Application in Mortar
with Clay-bearing Aggregates
by Weishan Wang, Zuiliang Deng, Zhongjun Feng,
Lefeng Fu, and Baicun Zheng
Interaction mechanisms of polycarboxylate-based superplasticizer (PCE) and poly(vinyl
alcohol) (PVA) with bentonite were systematically investigated. The adsorption of PCE
onto bentonite in aqueous solution was carried out, and changes in the surfaces and microstructures of the resultant PCE/bentonite complex were characterized by FTIR, XRD, TGA
and HRTEM. The results indicated a large adsorption amount of PCE onto bentonite
ranging from 157 mg/g to 230 mg/g. The interlayers of bentonite were intercalated by PCE
molecules with some surface adsorption. PVA adsorbed onto bentonite competitively with
PCE which decreased the adsorption amount of PCE drastically. Cement mortar experimental data showed ether-based PCE had better clay tolerance than ester-based PCE.
PVA as sacrifice agent can enhance the dispersibility of PCE for cement with clay.
Keywords: adsorption; bentonite; clay; intercalation; manufactured sand; mortar; polycarboxylate superplasticizer; poly(vinyl alcohol).
INTRODUCTION
During the last few decades, there has been a growing dependency on manufactured
sands caused by increasing regulatory pressure coupled with diminishing availability of
naturally local occurring sands as fine aggregates for concrete and mortar. Generally, the
presence of fracture microfines from manufactured sands can have different effects on the
performance of concrete or mortar. Non-clay powders such as limestone,1 slag2,3 and fly
ash.3,4 have been used successfully in concrete. However, the presence of clay as impurities
is always considered harmful in concrete or mortar, such as increasing the water demand to
provide a concrete or mortar of given workability,5 compromising the rheological property
of concrete or mortar after absorbing water and swelling,6 weakening the bond between
the cement paste and aggregates,7 and especially diminishing the dosage efficiency of
superplasticizers.8 These attributes have been ascribed to the presence of microfines such
333
334SP-302-25
Table 1-Main chemical composition of bentonite

SiO2
72.99
Al2O3
16.33
MgO
4.67
Chemical composition (wt%)

CaO
Fe2O3
TiO2
3.18
1.78
0.35
K2O
0.17
Na2O
0.04
P2O5
0.05
LOI
(wt%)
11.7
LOI: Loss of ignition at 1273 K.
as limestone9,10 and clay powder11-13 in manufactured sand, which can obstruct superplasticizers from dispersing cement particles. As is known from prior research,11-13 the
microfines in manufactured sands consume a large amount of superplasticizer which can
decrease the dispersing efficiency of polycarboxylate-based superplasticizer. Bentonitebased clay influences concrete workability most seriously.
The objective of the present investigation was to study the interaction mechanism of
bentonite with polycarboxylate superplasticizer (PCE) by adsorption isotherms and characterization methods such as FTIR, XRD, TGA and TEM. PCEs with different carboxyl
density and side chain length were synthesized to reduce the sensitivity to clay. In addition,
competitive adsorption by PVA with PCE was carried out to elucidate the possibility to find
a sacrificial agent to weaken the adverse effect caused by clay.
MATERIALS AND METODS
Materials
Natural bentonite was crushed, ground, sieved through a 200-mesh sieve and dried
at 105oC (221oF) in an oven for 2h prior to use. Chemical analysis of bentonite was
performed by sequential X-Ray fluorescence spectrometer (XRF-1800, Shimadzu Corporation) and the results are shown in Table 1. The specific surface area of natural bentonite
was found to be 38.6 m2/g by BET method using nitrogen as an adsorbent (ASAP 2010N,
Macromeritics).
A commercial polycarboxylate-based superplasticizer labeled PCE-1 was selected as
the adsorbate without further purification, which are comb-like copolymers based from
a sodium polymethacrylate backbone and grafted side chains of polyethylene oxides (see
Fig.1).The PCE architectures and its molecular characteristics are listed in Table 2. All
other chemicals were of analytical grade.
PVA with a degree of polymerization of 300 and degree of hydrolysis value of 87-89%
was obtained from Kuraray Co., Ltd., Japan.
Methods
Adsorption equilibrium studies-Adsorption experiments were carried out by using a
batch technique on a horizontal thermostated shaker (SPH-103B, Shanghai Shiping Laboratory Equipment Co., Ltd.) operated at 300 rpm. A stock solution of 5 g/L was prepared
by dissolving a weighed amount of PC in deionized water. The experimental solution was
prepared by diluting the stock solution with deionized water when necessary. A series of
PC solutions with initial concentration range of 0.1-2.5 g/L were prepared by diluting
the stock solution above. The bentonite suspension, 1g/L, was also prepared before 1 h.
Known volumes of solutions and suspensions (50mL (3.05 cu in.), respectively) were
Interaction of Polycarboxylate-based Superplasticizer/Poly(vinyl alcohol)

with Bentonite and Its Application in Mortar w/ Clay-bearing Aggregates 335
Fig.1-Chemical structure of polycarboxylatebased superplasticizer.

Table 2-Characterization of the polycarboxylate-based superplasticizer
Sample
PCE-1
a
Length of
side chain n
n45:n22=1:2.5
Density of side
chains b:a
2:1
Mna
(g/mol)
18643
M wb
(g/mol)
34822
PDIc
(Mw/Mn)
1.87
Solid content
(wt%)
40
Mn: number-average molecular weight.
Mw: mass-average molecular weight.
PDI: Mw/Mn = polydispersity index.
transferred into 100 mL (6.10 cu in.) erlenmeyers with glass stoppers. Then the dispersions
were shaken for 30 min, which was found to be sufficient to reach adsorption equilibrium.
After the adsorption was over, the dispersions were rapidly centrifuged using a laboratory centrifuge (TG16-WS, Shanghai Lu Xiangyi Centrifuge Instrument Co., Ltd.) at 12000
rpm (14800g) (32.63lb) for 10 min. Then residual PC concentrations of the supernatant
solutions were determined by the standard calibration curve. The adsorption capacity of PC
was calculated by the eq. (1):

qe =
(C0 Ce )V
(1)
m
where qe is the adsorption amount of PC (mg/g) at equilibrium, C0 is the initial concentration of PC in solution (g/L), Ce is the equilibrium concentration of PC in solution (g/L), m
is the mass of adsorbent used (g) and V is the volume of the dispersion (L). The reproducibility during concentration measurements was ensured by repeating the experiments at
least two times under identical conditions and the average values were reported.
336SP-302-25
Competitive adsorption of PVA with PCE was carried out by testing the adsorption
amount of PCE with different concentration of PVA solution (0.025g/L, 0.05g/L, 0.10g/L,
0.15g/L, 0.40g/L and 0.60g/L).
Characterization methods-FTIR spectra of samples were acquired by a 6700 Fourier
transform infrared spectrometer (Nicolet) using KBr pressed disk technique. Natural
bentonite and KBr were weighed and then were ground in an agate mortar prior to pellet
making. The spectra were obtained by accumulating 32 scans at a resolution of 2 cm-1 in
the range of 4000-400 cm-1.
TG and DTG (differential thermogravimetric) curves were obtained simultaneously by
using a STD Q600 thermal analyzer (TA). The measurements were conducted in highpurity flowing nitrogen atmosphere (100 mL/min) and the test temperature was from 25 oC
(77 oF) to 800 oC (1472 oF) with a heating rate of 20 oC/min. Approximately 20 mg sample
was heated in an oven alumina crucible.
The X-ray studies were performed using the powder diffraction technique. The analysis
was conducted using a D/MAX 2550 VB/PC diffractometer (Rigaku). The source of X-ray
radiation was a sealed tube with a copper anode and nickel filter supplied by the generator
(40kV, 100mA). Diffraction measurements were conducted with the 2 angle of 2-80o at
the scanning rate of 0.02 o/min. The basal spacing was calculated by using Braggs equation (Eq.(2)).

n = 2dsin
(2)
where n is an integer (n=1), is the wavelength of incident wave (=0.15418 nm), d is the
spacing between the layers in the bentonite lattice and is the angle between the incident
ray and the scattering planes.
JEM-2100 high-resolution TEM (JEOL) was used to investigate the microstructure
of natural and PCE/bentonite complex at an accelerating voltage of 200 kV. Bentonite
samples were dispersed in ethanol ultrasonically for 30 min and dropped on Cu mesh grids
coated carbon, then dried in a hot air oven at 50 C (122 F) for 10 min.
Mortar tests-Cement mortars were made by mixing cement, fine aggregates, bentonite
(2.0% bwoc) and polymer solutions. The mixing procedure was as follows: (1) cement
and fine aggregate were added into the bowl and mixed in a Hobart mixer at a low speed
for 1 min, (2) the polymer solution was added into the mixture and mixed at low speed for
1.5 min, (3) stopping the mixer, quickly scraping down into the batch any mortar adhered
on the side of the bowl, then starting the mixer and keep on mixing for another 1 min at
medium speed. The fluidity was measured at 20 oC (68 F) by pulling out spread of the
mortar from a cone of top diameter of 50 mm (2.165 in.), bottom diameter of 100 mm
(3.937 in.) and height of 150 mm (5.906 in.). The spread was the average of two perpendicularly crossing diameters.
Adsorption isotherms of PCE onto bentonite
The adsorption capacity of PCE onto bentonite at equilibrium was affected by temperature from 303 to 333K (85.73 to 139.73 oF) (shown in Fig.2). As depicted in Fig.2, the
PCE uptake increases as the temperature increases, indicating favorable adsorption occurs

Fig.2-Adsorption isotherms of PCE onto bentonite.

at higher temperatures. The phenomenon may be due to the acceleration of originally slow
adsorption or the creation of some new active sites on the adsorbent surface.14 In general,
before the equilibrium time, the increase in PCE adsorption resulting from the increase of
temperature shows a kinetically controlling process. After the equilibrium was achieved,
the uptake decreased with increasing temperature indicating that the adsorption of PC onto
bentonite is controlled by an endothermic process.
Competitive adsorption of PCE with PVA onto bentonite
The competitive adsorption isotherms of PCE with different concentrations of PVA onto
bentonite were shown in Fig.3. The adsorption amount of PCE decreased with the increment in PVA concentration, indicating competitive adsorption behavior between PVA and
PCE molecules. The variation trend in adsorption amount of isotherms with certain PVA
concentration were different. When PVA concentration was less than 0.15 g/L, PCE adsorption amount increased initially, and then slowed down gradually until it reached a platform.
When PVA concentration was more than 0.40 g/L, PCE adsorption amount increased at the
beginning and then decreased without reaching equilibrium, which indicates that competitive adsorption behavior appears to involve exchanging of PVA molecules with PCE molecules adsorbed on bentonite.15,16
FTIR spectra
The FTIR spectra of bentonite and PCE/bentonite complex are shown in Fig.4, depicting
the major changes of bentonite before and after the adsorption of PCE. After the adsorption
of PCE, the PCE/bentonite complex not only has characteristic bentonite bands, but also
exhibited some new characteristic bands.
As can be seen from Fig.4, the adsorption peak at 3620 cm-1 was assigned to the -OH
stretching vibration of water molecules within the bentonite interlayer and weakly bonded
338SP-302-25
Fig.3-Adsorption isotherms of PCE with different concentrations of PVA onto bentonite.
Fig.4-FTIR spectra of natural bentonite and PCE/bentonite

complex.
to the Si-O surface, and the broad peak centered at 3470 cm-1 was due to the -OH stretching
vibration of adsorbed water. The peak at around 1640 cm-1 corresponded to the -OH deformation of water in both pristine bentonite and PCE/bentonite, but the peak intensity of
PCE/bentonite was lower than natural bentonite. This might suggest the increased hydrophobicity of the bentonite surface because of the adsorption. Compared to natural bentonite,
the spectra of PCE/bentonite showed two additional peaks at 2923 and 2879 cm-1, which
were attributed to the asymmetric and symmetric stretching vibrations of the methyl and

Fig.5-The XRD patterns of natural bentonite and PCE/

bentonite complex
methylene groups. This observation also indicated the presence of PCE molecules on the
surface of PCE/bentonite complex.
XRD diffraction
The most widely used method for the study of intercalation surfactants in the galleries of
phyllosilicates is XRD, which provides information on the interlayer structure of surfactant.17 The XRD patterns of bentonite adsorbed different concentrations of PCE are shown
in Fig.5. An intense reflection at 2=7.14o (corresponding interlayer distance was 1.24
nm) was observed for natural bentonite, which was attributed to the d001 plane of montmorillonite. In comparison with natural bentonite, the d001 peak of 0.1 g/L PCE/bentonite
complex shifted towards lowder angle 6.18o, corresponding to a basal spacing of 1.43 nm,
indicating the expansion of the interlayer space due to the intercalation of PCE molecules.
When PCE concentration increased from 0.5 g/L to 2.5g/L, the diffraction peak increased
from 5.00o to 5.08o, and the corresponding interlayer distance were 1.76nm and 1.74nm,
respectively. That means PCE molecules intercalated into bentonite interlayers during
adsorption process, enlarging the interlayer space.
TG analysis
The TG and DTG curves of natural bentonite (see Fig.6) showed two main stages of
mass losses. In the first stage, the evolution of physically adsorbed water on the surface of
natural bentonite in the region of 22-175 oC (71.6-347 oF) gave rise to a peak maximum of
71 oC (159.8 oF) on the DTG curve. The peak at 134 oC (273.2 oF) which was accompanied
by a mass loss of 0.30% was ascribed to the elimination of the water species coordinated
to the interlayer cations. In addition, the second mass loss occurred at temperature ranging
from 450-750 oC (842-1382 oF), where the TG curve displayed a step weight loss (about
5.88%) related to the release of structural OH of natural bentonite. The dehydroxylation
340SP-302-25
Fig.6-TG/DTG curves of natural bentonite.
Fig.7-TG/DTG curves of PCE/bentonite.

temperature of about 668 oC (1234.4 oF) is in agreement with the classical range of dehydroxylation temperature (600-700 oC) (1112-1292 oF) observed by other authors for cisvacant montmorillonites.18
For the TG and DTG curves of PCE/bentonite complex (shown in Fig.7), there arises two
extra stages, which were about 5.37% from 300-450 oC (572-842 oF) and about 2.30% from
450- 570 oC (842-1058 oF) besides the two stages of mass losses corresponding to natural
bentonite. This phenomenon is interpreted to suggest that two different types of association take place between the bentonite and the polymer. The first mass loss can be from
thermal decompostion of the polymer adsorbed on the surface of bentonite, and the second
mass loss associated with the molecules intercalated in the interlayers of bentonite.19 More-

Fig.8-TEM micrographs of natural bentonite (a) and PCE/

bentonite complex(b).
over, Tmax (the temperature when the rate of weight loss reaches a maximum) of natural
bentonite was observed to be at approximately 668 oC (1234.4 oF), in contrast, Tmax of PCE/
bentonite is 657 oC (1214.6 oF) and the intensity of the peak also decreased, which means
the dehydroxylation process was altered by changes in the bonding nature of the interlayer
water, possibly caused by the intercalation of PCE molecules. This was supported by the
results obtained from FTIR and XRD studies.
TEM observation
In order to find further evidence of the basal spacing enlarging, the microstructure
of natural bentonite and PCE/bentonite complex were evaluated by using TEM, which
permits the direct observation of microstructural features of clays.
In comparison with natural bentonite, it proved to be difficult to obtain detailed TEM
micrographs since the high vacuum of TEM and the high-energy beam can remove the
water or surfactant molecules that makes the layer structures collapse and prohibits the
structures from being readily observed. Therefore, one must take photographs as soon as
possible to obtain clear, accurate images during the period of TEM viewing, especially at
high magnifications. In the PCE/bentonite complex, layer spacing of 1.42 nm was in good
agreement with the XRD results, but in some areas, the layer spacing remained about 1.24
nm similar to natural bentonite. This suggests that not all the interlayers were intercalated by copolymers, resulting from irregular intercalation or structure collapse mentioned
above. Meanwhile, a characteristic swelling of bentonite containing termination as shown
by the arrow in Fig.8b also was observed. This suggested that the swelling of silicate layers
maybe augmented by defect in the clay structure.20
Adsorption amount of PCE with different molecular structure onto bentonite
Based on the reported intercalation of PCE molecules into bentonite interlayers, PCEs
with different carboxyl density and side chain length were synthesized to investigate if it is
possible to decrease the sensitivity of PCE to bentonite by changing its molecular structure.
The molecular structure of PCE was shown in Table 3.
The variation in adsorption amount of PCE is shown in Fig.9. For ester-based PCE
(Fig.9a), adsorption amount of PCE decreased with increasing carboxyl density, and when
nMAA/nMPEG increased to 0.96 (M400-3), 2.40 (M1000-3), respectively, the decreasing trend
342SP-302-25
Table 3-Molecular structure of PCE
Sample/Item
mMAA/ (mMAA+mMPEG)
M400
nMAA/nMPEG
M1000
Sample/Item
mAA/ (mAA+mTPEG)
T1200
nAA/nTPEG
T2400
T4000
Mxxx-1
6.44%
0.32
0.80
Txxx-1
4.58%
0.80
1.60
2.67
Mxxx-2
12.10%
0.64
1.60
Txxx-2
6.72%
1.20
2.40
4.00
Mxxx-3
17.11%
0.96
2.40
Txxx-3
8.76%
1.60
3.20
5.33
Mxxx-4
21.58%
1.28
3.20
Txxx-4
10.71%
2.00
4.00
6.67
Mxxx-5
25.60%
1.6
4.00
Txxx-5
12.59%
2.40
4.80
8.00
Note: Mxxx-y is ester-based PCE, and Txxx-y is ether-based PCE.
Fig.9-Comparison of adsorption amount of polycarboxylate

with different molecular structure onto bentonite (a:esterbased, b-ether-based)
Mxxx-6
29.22%
1.92
4.80
Txxx-6
14.38%
2.80
5.60
9.33

of adsorption amount slowed down. The affinity between anionic bentonite and anionic
PCE was mainly realized by intercalation of EO side chains into bentonite interlayers.
When the carboxyl density of PCE main chain increased, the repelling force between
adsorbate molecules and adsorbent particles increased correspondingly. The intercalation
of side chains into bentonite interlayer became more difficult and the adsorption amount
decreased. When molecular weight of side chain increased from 400 to 1000 with a certain
carboxyl density, the adsorption amount had a little reduction (shown in Fig.9a). That was
because the enlargement of side chain size was unfavorable for the intercalation.
For ether-based PCE (Fig.9b), the variation in adsorption amount of PCE with carboxyl
density and side chain increasing was similar to ester-based PCE. However, adsorption
amount of ether-based PCE onto bentonite was much less than that of ester-based PCE.
The adsorption amount of ether-based PCE ranged from 15 to 30 mg/g with comparison to
75-120 mg/g for ester-based PCE in our experiment. The ester-based PCE was synthesized
with methoxy polyethylene glycol methacrylate (MPEG) as macromonomer and methacrylic acid as comonomer, and the ether-based PCE was synthesized with isoamyl alcohol
polyoxyethylene ethers (TPEG) as macromonomer and methacrylic acid as comonomer.
The methyl groups in ester-based PCE has strong shielding effect on negatively charged
carboxyl groups. Otherwise, the carboxyl groups in ether-based PCE will play effective
electrostatic repulsion between the polymer and bentonite particles, resulting in the reduction in adsorption amount.
Mortar test
The variation in cement mortar with 2%bwoc bentonite with PCE above was shown
in Fig.10 (water/cement ratio-0.49, sand/cement ratio-2.07, PCE dosage-0.40%bwoc).
For ester-based PCE, the fluidity increased initially, and then decreased with increasing
carboxyl density. However, the effect of side chain size with same carboxyl density on
fluidity didnt show regular trend. For chain Mw of 400, maximum flow observed with
Mxxx-5, whereas for chain Mw of 1000, Mxxx-3 exhibited maximum flow. For etherbased PCE, mortar fluidity increased with increasing carboxyl density with same side
chain length. That is because the increase of carboxyl density lowered the harmful loss
caused by bentonite and cement particles were dispersed well. In addition, mortar fluidity
increased with increasing side chain length with same carboxyl density. With side chain
length increasing, adsorption amount of PCE onto bentonite decreased, meanwhile, the
steric hindrance of PCE molecules on cement particle surface decreased the agglomeration
trend and improved the fluidity of cement mortar.
Effect of PVA dosage on the fluidity of cement mortar with 2%bwoc was shown in Fig.11
(water/cement ratio-0.36, sand/cement ratio-2.07, PCE dosage-0.50% bwoc). Without
PVA, mortar almost had no fluidity, which is ascribed to the large amount of consumption of PCE by bentonite resulting in inadequate dispersion of cement particles. When
PVA dosage increased from 0.25% bwoc to 1.00% bwoc, the mortar fluidity increased
from 130 mm (5.118 in.) to 310 mm (12.205 in.). PVA adsorbed onto bentonite with PCE
competitively, and that decreased the PCE adsorption amount, weakening the consumption
of PCE caused by bentonite. As a result, the insensitivity of cement dispersion with PCE
was enhanced.
344SP-302-25
Fig.10-Effect of PCE on the fluidity of cement mortar with

bentonite.
CONCLUSION
(1) A large amount of PCE was found to be adsorbed onto bentonite ranging from 157
mg/g to 230 mg/g. The interlayers of bentonite are intercalated by PCE molecules with
surface adsorption to a certain extent indicated by characterizing PCE/bentonite complex.
(2) PVA adsorbs onto bentonite competitively with PCE which decreases the adsorption
amount of PCE drastically.
(3) Cement mortar experimental data indicates ether-based PCE has better clay tolerance
than ester-based PCE. Increasing side chain length and carboxyl density in main chain are
favorable for weakening the sensibility of PCE to bentonite. PVA as sacrificial agent can
enhance the dispersibility of PCE for cement with clay.

Fig.11-Effect of PVA dosage on the fluidity of cement mortar

with bentonite.
AUTHOR BIOS
Weishan Wang is a R&D Engineer at the Technology Center, Shanghai Engineering
Research Center of Construction Admixtures (CAERC). He received his PhD from East
China University of Science and Technology (ECUST). His research interests include
surface/interface chemistry, rheology and compatibility of PCE with concrete raw
materials.
Zuiliang Deng is a PhD Candidate at school of resources and environment engineering
of ECUST, and also a R&D Engineer at the Technology Center, CAERC. She received her
MS from ECUST. Her research interests include applications of PCE in all kinds of high
performance concrete.
Zhongjun Feng is a R&D Engineer at the Technology Center, CAERC. He received his
MS from ECUST. His research interests include molecular design of PCE and interaction
of PCE with cement.
Lefeng Fu is a senior R&D Engineer at the Technology Center, CAERC. He received his
PhD from ECUST. His research interests include R&D of key common issues in concrete
and PCE industry.
Baicun Zheng is a professor at the Technology Center, CAERC. He received his PhD
from ECUST. His research interests include chemical engineering, surface/interface
chemistry, and nano materials.
346SP-302-25
ACKNOWLEDGEMENTS
The financial support (14YF1414700) from Science and Technology Commission of
Shanghai Municipality was gratefully acknowledged.
REFERENCES
1. Zhu, W., and Gibbs, J. C., Use of different limestone and chalk powders in selfcompacting concrete, Cement and Concrete Research, V. 35, No. 8, 2005, pp. 1457-1462.
doi: 10.1016/j.cemconres.2004.07.001
2. Al-Jabri, K. S.; Hisada, M.; Al-Oraimi, S. K.; and Al-Saidy, A. H., Copper slag as
sand replacement for high performance concrete, Cement and Concrete Composites, V.
31, No. 7, 2009, pp. 483-488. doi: 10.1016/j.cemconcomp.2009.04.007
3. Yazc, H.; Yiiter, H.; Karabulut, A. .; and Baradan, B., Utilization of fly ash and
ground granulated blast furnace slag as an alternative silica source in reactive powder
concrete, Fuel, V. 87, No. 12, 2008, pp. 2401-2407. doi: 10.1016/j.fuel.2008.03.005
4. Jaturapitakkul, C.; Kiattikomol, K.; Sata, V.; and Leekeeratikul, T., Use of ground
coarse fly ash as a replacement of condensed silica fume in producting high-strength
concrete, Cement and Concrete Research, V. 34, No. 4, 2004, pp. 549-555. doi: 10.1016/
S0008-8846(03)00150-9
5. Yool, A. I. G.; Lees, T. P.; and Fried, A., Improvements to the methylene blue dye test
for harmful clay in aggregates for concrete and mortar, Cement and Concrete Research,
V. 28, No. 10, 1998, pp. 1417-1428. doi: 10.1016/S0008-8846(98)00114-8
6. Norvell, J. K.; Stewart, J. G.; Juenger, M. C. G.; and Fowler, D. W., Influence of clays
and clay-sized on concrete performance, Journal of Materials in Civil Engineering, V. 19,
No. 12, 2007, pp. 1053-1059. doi: 10.1061/(ASCE)0899-1561(2007)19:12(1053)
7. Topu, . B., and Uurlu, A., Effect of the use of mineral filler on the properties
of concrete, Cement and Concrete Research, V. 33, No. 7, 2003, pp. 1071-1075. doi:
10.1016/S0008-8846(03)00015-2
8. Leslie A. Jardine, Koyata, Kevin J. Folliard, et al., Admixtures and method for optimizing addition of EO/PO superplasticizer to concrete containing smectite clay-containing
aggregates, US 6352952 B1, 2002.3.5.
9. Li, B. X.; Wang, J. L.; and Zhou, M. K., C60 high performance concrete prepared
from manufactured sand with a high content of mirofines, Key Engineering Materials, V.
405-406, 2009, pp. 204-211. doi: [J]10.4028/www.scientific.net/KEM.405-406.204
10. Li, B.; Wang, J.; and Zhou, M., Effect of limestone fines content in manufactured
sand on durability of low- and high-strength concretes, Construction & Building Materials, V. 23, No. 8, 2009, pp. 2846-2850. doi: [J]10.1016/j.conbuildmat.2009.02.033
11. Sakai, E.; Atarashi, D.; and Damaon, M., Interaction between superplasticizers and
clay minerals [C]. Proceedings of the 6th International Symposium on Cement & Concrete.
Japan: Japan Cement Association. 2006, 2: 1560-1566
12. Plank, J.; Liu, C.; and Ng, S., Interaction between clays and polycarboxylate superplasticizers in cementitious systems [C]. 9th CANMET/ACI International Conference in
Superplaticizers and Other Chemical Admixtures (supplementary papers). ACI Special
Publication, 2009, 279-298
13. Jeknavorian, A. A.; Jardine, L.; and Ou, C. C., Interaction of superplasticizers with
clay-bearing aggregates [C]. 7th CANMET/ACI International Conference on Superplasti-

cizers and Other Chemical Admixtures in Concrete. America: American Concrete Institute.
2003, 143-159
14. Daifullah, A. A. M.; Girgis, B. S.; and Gad, H. M. H., A study of the factors affecting
the removal of humic acid by activated carbon prepared from biomass material, Colloids
and Surfaces A: Physicochemical and Engineering Aspects, V. 235, No. 1-3, 2004, pp.
1-10. doi: 10.1016/j.colsurfa.2003.12.020
15. Bae, J. A. E.-H. Y. U. N.; Song, D. O. N. G.-I. K.; and Jeon, Y. O. U. N. G.-W. O. O.
N. G.Bae Jae-Hyun. , Song Dong-Ik, Jeon Young-Woong. Adsorption of anionic dye and
surfactant from water onto organomontmorillonite, Separation Science and Technology,
V. 35, No. 3, 2000, pp. 353-365. doi: [J]10.1081/SS-100100161
16. Al-Degs, Y., and Khraisheh, M. A. M., Effect of carbon surface chemistry on the
removal of reactive dyes from textile effluent, Water Research, V. 34, No. 3, 2000, pp.
927-935. doi: [J]10.1016/S0043-1354(99)00200-6
17. Li, Y., and Ishida, H., Concentration-dependent conformation of alkyl tail in the
nanoconfined space: Hexadecylamine in the silicate galleries, Langmuir, V. 19, No. 6,
2003, pp. 2479-2484. doi: 10.1021/la026481c
18. Drits, V. A.; Besson, G.; and Muller, F., An improved model for structural transformations of heat-treated aluminous dioctahedral 2:1 layer silicates, Clays and Clay
Minerals, V. 43, No. 6, 1995, pp. 718-731. doi: 10.1346/CCMN.1995.0430608
19. Zhou, Q.; Frost, R. L.; He, H.; and Xi, Y., Changes in the surfaces of adsorbed paranitrophenol on HDTMA organoclayThe XRD and TG study, Journal of Colloid and
Interface Science, V. 307, No. 1, 2007, pp. 50-55. doi: 10.1016/j.jcis.2006.11.016
20. Lee, S. Y., and Kim, S. J., Expansion characteristics of organoclay as a precursor to
nanocomposites, Colloids and Surfaces A: Physicochemical and Engineering Aspects, V.
211, No. 1, 2002, pp. 19-26. doi: 10.1016/S0927-7757(02)00215-7
348SP-302-25
SP-302-26
Effect of the Stereochemistry of Polyols

on the Hydration of Cement: Influence of
Aluminate and Sulfate Phases
by Camille Nalet and Andr Nonat
The difference in the retarding effects induced by sugar alcohols on the hydration of pure
tricalcium silicate and white cement pastes is investigated. The polyols studied which are
stereoisomers (D-glucitol, D-galactitol and D-mannitol) generate a lower retarding effect
on the hydration of white cement than on the hydration of pure tricalcium silicate. The
presence of aluminate and sulfate phases in white cement pastes is shown to reduce the
retarding effect induced by the molecules. Moreover, these alditols strongly complex aluminate in solution and adsorb on hydrating cement. The interactions of polyols with the
anhydrous and/or hydrated aluminate phases and their effects on the hydration kinetics of
white cement are discussed.
Keywords: adsorption; aluminate; cement; complexation; hydration; kinetics; polyols;
reactivity; stereochemistry.
INTRODUCTION
Portland cement is composed of silicate (C3S, C2S), aluminate (C3A, C4AF) and sulfate
(CaSO4, xH2O) phases which interact between each other during hydration and make
its study complicated. The generalized use of chemical admixtures confers to concrete
specific properties either related to its fresh state (rheology modifiers, set accelerators
or retarders) or longer term properties (strength enhancers, durability improvers...).1
Although used since decades those additives are often considered from a performance point
of view2,3 but their interactions with mineral surfaces are hardly resolved and understood.
Hence, this study is focused on the understanding of the interactions of hexitols with
white cement paste and their effects on its kinetic of hydration. Here, we compare the effect
of the conformation of D-glucitol, D-galactitol and D-mannitol on the hydration kinetics
of pure tricalcium silicate (C3S) and of white cement. Then, the interactions of the hexitols
with the anhydrous and/or hydrated aluminate phases present during cement hydration are
assessed.
349
350SP-302-26
Table 1-Physical and chemical compositions of the cement phases.

White cement Tricalcium silicate
Specific surface area
(m2/g)
Mono. Alite (%)
Tric. Alite (%)
Belite (%)
Ferrite (%)
Cub. Aluminate (%)
Ortho. Aluminate (%)
Lime (%)
Gypsum (%)
Hemihydrate (%)
Anhydrite (%)
Calcite (%)
Portlandite (%)
Quartz (%)
Tricalcium
aluminate-Gypsum-Hemihydrate
0.40 a
0.49 b
0.34 b
66.10
0.00
24.20
0.40
2.00
0.70
0.30
0.20
0.70
1.00
2.40
1.80
0.10
100.00
-
80.00
5.00
15.00
-
= not measured items.

a
Blaine method,
Calculated from particle size distribution assuming that the density of particles is homogeneous with the size, that the particles
are spherical and considering the different densities (C3S: 3210 kg/m2 and C3A= 3030 kg/m2).
b
With the diversity of existing cements and the increasing levels of Supplementary
Cementitious Materials of Portland cement in modern concrete, the current limited knowledge on the interactions between chemical admixtures and mineral phases and their influence on the hydration mechanisms of cement phases represents a real limitation in the
development of new products with improved properties. This study intends to contribute
to fill this knowledge gap by focusing on the effects and interactions of hexitols with
hydrating cement phases.
Materials
A batch of white cement, a C3A-gypsum-hemihydrate mixture and triclinic C3S were
used, Table 1. The molecules studied were D-glucitol ( 98%), D-galactitol ( 99%) and
D-mannitol ( 98%) which were in a powder form. The preparation of all pastes, suspensions and solutions was made with water which was both distilled and deionised. Calcium
oxide used in different experiments was obtained after decarbonation of calcium carbonate
(98.5-100%) at 1000 C [1830 F] for 24 h. All the saturated lime solutions were obtained
by equilibrating water at 25 C [77 F] with an excess of calcium oxide.
Methods
Isothermal calorimetry - The kinetics of hydration of different cement phases in presence
of sugar alcohols were monitored by isothermal calorimetry at 23 C [73 F]. The mixing
Effect of the Stereochemistry of Polyols on the Hydration of Cement:

Influence of Aluminate and Sulfate Phases 351
Table 2-Chemical structure of the hexitols studied.
D-glucitol
D-galactitol
D-mannitol
protocol was as follows: 1 g of mineral phases was put in a plastic ampoule and mixed with
0.4 mL of aqueous sugar solutions (0-27 mmol/L) at 3200 rpm for 2 min with a stirrer.
After mixing, the ampoules were capped and inserted in the calorimeter.
Adsorption - The adsorption of polyols on hydrating white cement (L/S=5) was measured
during the induction period. Samples of the cement suspension were collected over time.
The different samples were centrifuged at 9000 rpm for 5 min. Finally, the supernatants
were filtered with a 0.20 m [7.87 in] syringe filter (PTFE). The adsorption of the organic
molecules was determined by using the depletion method. The non adsorbed portion of
molecules remaining in solution was measured by analyzing the Total Organic Carbon
(TOC) of this solution with a TOC analyzer.
Different C3A-gypsum-hemihydrate mixture contents in white cement - The C3A
percentage of white cement was increased from 2.7% to 15% by adding different amount
of C3A-gypsum-hemihydrate to the initial white cement.
Ionic concentrations of cement pore solutions - The hydration of the aluminate phase
of white cement in presence of the hexitols in saturated lime solutions (L/S=100) was
followed by continuously measuring the concentration of aluminium in solution over time
by Inductively Coupled Plasma Atomic Emission Spectroscopy, ICP-AES.
Effect of the stereochemistry of polyols on the hydration kinetics of pure
C3S and of white cement
In order to identify the effect of the stereoisomers D-glucitol, D-galactitol and D-mannitol
(structures in Table 2) on the kinetics of hydration of C3S and white cement, calorimetric
measurements were carried out during the hydration of these phases in presence of the
molecules at different dosages. In the range of concentration studied (0-27mmol/L), these
polyols do not change the slope of the calorimetric curves during the acceleration period
as can be seen in presence of D-glucitol for example, Fig 1. Hence, the retardation is
calculated as the difference in time between the maximum peaks of heat flow of the polyol
samples and the reference without additive.
Even though D-glucitol, D-galactitol and D-mannitol have the same chemical formula
C6H14O6, they induce different effects on the hydration kinetics of pure C3S pastes and of
white cement pastes (Fig 2 and Fig 3). This effect was also shown by Zhang4 who studied
the setting behavior of cement and C3S pastes in presence of D-glucitol and D-mannitol.
It is noted that D-glucitol is the most retarding polyol for all the systems studied, see Fig
2 and Fig 3. It is followed by D-galactitol and then D-mannitol when hydrating pure C3S
(Fig 2) whereas in cement both molecules induce approximately the same delay in the
hydration of C3S (Fig 3). The results suggest that white cement is less discriminating than
pure C3S.
352SP-302-26
Fig 1-Calorimetric curves followed during the hydration of

C3S and white cement pastes in presence of a sugar alcohol,
L/S=0.4.
Fig 2-Retardation of C3S hydration depending on the concentration of different stereoisomers (D-glucitol, D-galactitol,
D-mannitol), L/S =0.4.
In order to highlight the role of the aluminate phase present in white cement on the
retarding effect induced by the presence of the hexitols, the retarding effects of each molecule on the hydration of white cement and of pure C3S were compared. In a first approximation, the retardation of the hydration of both white cement and of pure C3S is forced
to fit a linear function of the concentration of molecules with the equation of the form
y(x)= ax. Rr is defined as the ratio of the slope concerning the retardation of white cement
to the slope concerning the retardation of pure C3S, both depending on the concentration of molecules. The Rr ratios induced by the different alditols are below 1 (Fig 4): the
retardation of the hydration of white cement is lower than the retardation of pure C3S.
Moreover, the molecules generate different Rr ratios between 0 and 1. This means that the
molecules interact differently with the other components present in white cement paste,
i. e. the aluminate (and/or its hydrates) and/or the sulfate phases. Given that the pKa of
the polyols are above 135 and that the pH of the pore solution is around 12.5 during the
hydration of cement, the amount of polyols with deprotonated alcohol groups is assumed

Fig 3-Retardation of white cement pastes hydration

depending on the concentration of different sugar alcohols,
L/S =0.4.
Fig 4-Comparison of Rr (ratio of the slope concerning the

retardation of white cement pastes to the slope concerning
the retardation of pure C3S pastes) in presence of different
molecules.
to be low. Since the interaction of the polyols in the pastes is not of electrostatic nature,
sugar alcohols are then not supposed to compete with sulfate. Hence, the difference in the
Rr ratios identified above should come from different affinities of the molecules with the
anhydrous or/and hydrated aluminate phases. D-glucitol is supposed to have the higher and
D-mannitol the lower affinity with the aluminate phase and/or its hydrates.
Effect of the C3A-gypsum-hemihydrate mixture content of white cement on
its kinetics of hydration in presence of alditols
In order to investigate possible interactions of the polyols with anhydrous and/or
hydrated aluminates, we focused on the kinetics of hydration of white cement with different
C3A-gypsum-hemihydrate mixture contents. It was observed that varying the content of the
C3A-gypsum-hemihydrate mixture from 2.7% to 15% in cement at constant concentration
of alditols (20.25 mmol/L), the slopes of the calorimetric curves during the acceleration
354SP-302-26
Fig 5-Calorimetric curves followed during the hydration

of white cement with different C3A-gypsum-hemihydrate
mixture contents with and without the presence of a sugar
alcohol, L/S=0.4.
Fig 6-Comparison of the Tei (time ending the induction

period from the calorimetric curve) as a function of the
percentage of C3A in white cement pastes in presence of
three different sugar alcohols (20.25 mmol/L), L/S=0.4.
period were different, Fig 5. In this case, instead of considering the maximum heat flow
which ends the acceleration period, the time ending the different induction periods (Tei)
was compared. This time is defined at the intersection between the slope of the calorimetric
curve during the acceleration period and the x-axis.
In presence of the three polyols, Tei is reduced when increasing the amount of
C3A-gypsum-hemihydrate mixture in cement and gets closer to the Tei of the references,
Fig 6. By extrapolation, it can be argued that increasing the C3A-gypsum-hemihydrate
mixture content in cement would inhibit the retarding effect of the alditols on the hydration of white cement. Thus, there is an interaction of the sugar alcohols with the aluminate
part and/or its hydrates during the hydration of white cement which reduces their retarding
effect.

Fig 7-Pseudo adsorption isotherm of sugar alcohols on

hydrating white cement in saturated lime solution during the
induction period, L/S=5.
Adsorption and retarding effect of the hexitols on hydrating white cement
The stereoisomers affect differently the retardation of the hydration of white cement.
Hence, in order to understand why sugar alcohols retard the hydration of white cement,
it makes the investigation of a possible link between the adsorption of polyols and their
retarding effects on cement hydration a priority.
Fig 7 shows the pseudo adsorption isotherm of the different sugar alcohols in saturated
lime solution on hydrating white cement pastes. The term pseudo adsorption isotherm is
used to highlight the fact that equilibrium is not reached and that the value of the adsorption corresponds to the amount of molecules adsorbed during the induction period to the
amount of dry cement instead of the surface at this time of measurement as a function of
the concentration of the remaining molecules in solution. The pseudo adsorption isotherm
shows that there are different adsorption behaviors depending on the conformation of
the molecules, Fig 7. D-galactitol and D-mannitol reach a saturation plateau whereas
D-glucitol does not reach a plateau in the range of concentrations studied (0-54 mmol/L).
In fact, for D-glucitol, the pseudo adsorption isotherm even diverges suggesting a massive
precipitation at some point. In the initial linear part of the adsorption isotherm, for a fix
concentration in solution, D-glucitol has the higher adsorption followed by D-galactitol
and then D-mannitol. Thus, D-glucitol has the most important affinity with the anhydrous
and/or hydrated component(s) of white cement compared to the other polyols.
The evolution of the retardation of the hydration of C3S in white cement pastes as a
function of the adsorption of polyols is shown in Fig 8. The retardation of the hydration
of white cement pastes was measured by using the calorimetric curves obtained with and
without additive (L/S=0.4). The last retardation value for each polyol (initial concentration
54 mmol/L) was estimated by extrapolation. The adsorption of hexitols was measured in
white cement suspensions as described in the paragraph just above. For all the polyols,
the retardation of the hydration of white cement increases as a function of the adsorption
but the trends are different, Fig 8. We observe a low retardation of the hydration of white
cement in presence of D-glucitol despite a high adsorption whereas for D-galactitol and
356SP-302-26
Fig 8-Retardation of the hydration of white cement pastes

induced by hexitols depending on their adsorption on
hydrating cement during the induction period.
Fig 9-Aluminium concentration during the hydration of

white cement in saturated lime solution with sugar alcohols
(20.25 mmol/L), L/S=100.
D-mannitol, it is the contrary. These results allow us to suppose that D-glucitol particularly
adsorbs on C3A and/or its hydrates.
Effect of polyols on the concentration of aluminate present in the pore
solution during white cement hydration
The hydration of C3A of white cement suspensions in presence of the alditols in saturated
lime solutions have been followed by measuring the aluminium concentration in the pore
solution over time by using ICP-AES, Fig 9. Results show that the presence of polyols
in solution increases more than 10 times the aluminium concentration of pore solution.
This result indicates that the hexitols complex aluminium in the pore solution. In 1996,
using NMR and Raman spectroscopy, Smith6 already detected the existence of complexes
between hexitols and aluminium in alkaline solutions.

FURTHER RESEARCH
The role played by the surface adsorption of alditols and/or of their aluminate complex
and the one played by the aluminate complexation in solution on the hydration of C3A
should be further investigated. Moreover, the impact of these interactions between polyols
and hydrating C3A on the hydration of C3S present in cement should be examined in depth.
CONCLUSIONS
The retarding effects generated by polyols (D-glucitol, D-galactitol and D-mannnitol)
are lower on the hydration of white cement than on the hydration of pure C3S and differ
depending on the stereochemistry of their hydroxyl groups. Moreover, the increasing
content of the C3A-gypsum-hemihydrate mixture shortens the induction period induced
by the sugar alcohols during the hydration of white cement. Hence, the molecules are
assumed to interact with anhydrous and/or hydrated aluminate phases. The polyols are
shown to complex with aluminate in solution and to adsorb on hydrating white cement.
D-glucitol induces a low retarding effect on the hydration of cement compare to its important adsorption on hydrating cement. For D-mannitol, it is the contrary. Then, D-glucitol
is supposed to have the highest and D-mannitol the lowest affinity with anhydrous and/or
hydrated aluminate phases. This way, the high amount of adsorbed D-glucitol would not
be free any more to disturb the hydration of C3S in white cement and would induce a lower
retardation. As for D-galactitol and D-mannitol are concerned, the amount of non adsorbed
molecules after saturation on the surface of anhydrous and/or hydrated aluminate phases
would particularly delay the hydration of C3S in white cement.
AUTHOR BIOS
Camille Nalet is PhD student at the University of Burgundy in the Physical Chemistry of Cementitious and Colloidal Media laboratory group (Dijon, France). She has
focused her work on the influence of structural and chemical parameters of small organic
molecules on the reactivity and hydration kinetics of cement phases.
Andr Nonat is senior researcher at the CNRS/University of Burgundy in the Physical
Chemistry of Cementitious and Colloidal Media laboratory group (Dijon, France). His
research is based on the reactivity of cement with special interest in studying the mechanism of hydration reactions, the thermodynamics and microstructure of hydrates and the
mechanism of the setting.
ACKNOWLEDGMENTS
The financial support of Nanocem, the European Consortium of Academic and Industrial
partners is gratefully acknowledged. We are also thankful to the working group for advice
and fruitful discussions.
NOTATION
C
S
A
F
= CaO,
= SiO2,
= Al2O3,
= Fe2O3.
358SP-302-26
REFERENCES
1. Kovler, K. and Roussel, N., Properties of fresh and hardened concrete, Cement and
Concrete Research 41, V. 41, 2011, pp. 775792.
3. Flatt, R. J., and Schober, I., Superplasticizers and the rheology of concrete, Understanding the rheology of concrete, Woodhead Publishing, 2012, pp. 144-208.
4. Zhang, L.; Catalan, L. J. J.; Balec, R. J.; Larsen, A. C.; Esmaeili, H. H.; and Kinrade,
S. D., Effect of saccharide set retarders on the Hydration of Ordinary Portland Cement
and Pure Tricalcium Silicate, Journal of the American Ceramic Society, V. 93, No. 1,
2010, pp. 279-287. doi: 10.1111/j.1551-2916.2009.03378.x
5. Gaidamauskas, E.; Norkus, E.; Vaiciuniene, J.; Crans, D. C.; Vuorinen, T.; Jaciauskiene,
J.; and Baltrunas, G., Evidence of two-step deprotonation of D-mannitol in aqueous
solution, Carbohydrate Research, V. 340, No. 8, 2005, pp. 1553-1556. doi: 10.1016/j.
carres.2005.03.006
6. Smith, P.; Watling, H.; and Crew, P., The effects of model organic compounds on
gibbsite crystallization from alkaline aluminate solutions: polyols, Colloids and Surfaces
A: Physicochemical and Engineering Aspects, V. 111, No. 1-2, 1996, pp. 119-130. doi:
10.1016/0927-7757(95)03488-9
SP-302-27
A New Accelerator Approach for Improved

Strength Development
by Franz Wombacher, Christian Brge, Emmanuel
Gallucci, Patrick Juilland, and Gilbert Mder
One of the main problems associated with supplementary cementitious materials (SCMs),
which are used as clinker or cement replacement, is the slow strength development
compared to pure OPC. This is especially evident in the early stages of cement hydration
and may cause significant problems for the customers. Therefore, the demand for new and
powerful accelerators only having a marginal influence on the workability of the concrete,
is rising. These types of accelerators may find their application in normal ready-mixed
concrete but, much more evident, in precast applications.
In this paper, a new accelerator is presented, which can significantly improve the
early strength (up to 2 days) of concrete. In addition to this, the components of this new
approach do not bear any potential risk of corrosion for steel, be it normal reinforcement
or prestressed steel.
Keywords: hardening accelerator; calcium oxide; admixture; booster; early compressive
strength; fly ash; limestone filler.
INTRODUCTION
The demand for effective hardening accelerators which have only little influence on the
workability of concrete and mortar is increasing. This is based on different facts:
On one hand, the use of supplementary cementitious materials (SCM) is increasing due
to clinker and cement replacement in order to reduce carbon dioxide emissions.1,2 These
binders are either produced as blended cements by the cement producer or mixed out of
OPC and SCM at the batching plant. They often lack in sufficient strength development,
which is especially evident at early ages, up to 2 days.
On the other hand, precast concrete as a fast and reliable production method would
be very efficient, however, steam curing is often necessary. As a result, a large amount
of energy is consumed and the turnover of forms is low which can lead to a reduction in
productivity.3
An increase in early compressive strength of concrete, either based on pure OPC or on
blended systems can be achieved either by addition of hardening accelerators, either classical types or new ones based on calcium silicate hydrate (CSH) suspensions.4,5
359
360SP-302-27
Classical hardening accelerators are often based on amino alcohols and soluble inorganic salts like nitrates and thiocyanates6,7 or even chlorides,8 whereas CSH-suspensions
are produced from soluble calcium-salts (like nitrate) and a silicate source.4,5
In both cases, a potential risk of corrosion cannot be excluded.9 The discussion about
this potential risk of these soluble inorganic salts is ongoing since long time10 and is not
yet concluded. Especially the use of higher dosages of these new types of accelerators and,
therefore, the salts might intensify the discussion again.
Classical hardening accelerators and CSH-suspensions which promote the seeding are
today used in several applications to especially improve the early strength development of
cementitious systems, either in pure OPC or in blended systems.
This paper deals with the use of a combination of calcium oxide and a new type of
booster component for the improvement of the early compressive strength (up to 2 days)
of cementitious materials.
Mortar and concrete mixes were investigated in the laboratory. The variables include
different types of cements, mixes with fly ash (FA) and variations in the content of the
investigated accelerator composition at 20C (68F). Flow table spread (FTS) and compressive strength were determined after varying time periods.
Isothermal conduction calorimetry measurements were performed at 23C (73.4F) both
on pastes at a water-cement ratio of 0.35 using a Thermometrics TAM AIR calorimeter
and on standard mortar according to DIN EN-196-1 at a water-cement ratio of 0.4 using a
Calmetrix I-Cal device.
Materials
Superplasticizers: High range water reducers HRWR 1 (PCE for precast concrete) and
HRWR 2 (PCE for ready-mix concrete)
Booster: an organic ester of phosphoric acid
Classical hardening accelerator: ACC 1 (based on amino alcohol and inorganic
compounds such as sodium thiocyanate and sodium nitrate)
Cements: CEM I 42.5 N and CEM I 52.5 R, both from Holcim AG, Siggenthal,
Switzerland.
Fly ash (FA) was obtained from SAFA Saarfilterasche-Vertriebs GmbH Co. GK,
Baden-Baden/Germany.
Calcium oxide (CaO) was obtained from Kalkfabrik Netstal AG, Switzerland.
CEM I 42.5 N has a Blaine fineness of 3060 cm2/g, CEM I 52.5 R a Blaine fineness of
4120 cm2/g, FA a Blaine fineness of 3930 cm2/g and CaO a Blaine fineness of 4360 cm2/g.
The physical properties and chemical analysis of the OPCs, FA, limestone filler, quartz
flour and calcium oxide are shown in Table 1.
Aggregates: river sands 0/1 (0-1 mm, 0-0.04 in), 1/4 (1-4 mm, 0.04-0.16 in), 4/8 (4-8
mm, 0.16-0.31 in), stones 8/16 (8-16 mm, 0.31-0.63 in) and 16/32 (16-32 mm, 0.63-1.26
in) were all obtained from Carlo Bernasconi AG, Zurich, Switzerland, limestone filler
A New Accelerator Approach for Improved Strength Development 361
Table 1 Physical and chemical compositions of OPC, FA, limestone filler,

Calcium oxide and quartz flour
FA
2.30
3930
CaCO3
Limestone
filler
2.97
4840
CaO
3.31
4360
a)
a)
a)
a)
9000
2.6
1.9
3.8
2.9
0.21
SiO2 [%]
Al2O3 [%]
CaO [%]
MgO [%]
SO3 [%]
Na2O [%]
19.9
4.6
62.8
2.0
3.1
0.2
20.3
4.7
63.1
2.0
3.6
0.2
0.9
1.5
2.4
0.1
0.07
95.9
0.7
0.2
-
98.8
0.8
0.02
0.01
0.01
K2O [%]
1.0
1.0
0.01
TiO2 [%]
MnO [%]
Fe2O3 [%]
CaCO3 [%]
MgCO3 [%]
3.0
-
3.0
-
0.1
95.5
1.7
0.03
n.d.
-
0.03
0.02
-
CEM I
42.5 N
3.13
3060
CEM I 52.5 R
3.13
4120
a)
Loss on ignition [%]
Density [g/cm3]
Blaine fineness [cm2/g]
BET [cm2/g]
a)
SiO2
Quartz flour
2.63
a)
not determined
Table 2 - Mortar and concrete mixture proportions
Binder
Fines (limestone or quartz flour)
Sand 0/1
Sand 1/4
Aggregate 4/8
Aggregate 8/16
Aggregate 16/32
Mortar
Lab mix, g (lb)
750 (1.65)
141 (0.31)
738 (1.63)
1107 (2.44)
1154 (2.54)
-
Concrete
Lab mix, kg (lb)
Kg/m3 (lb/yd3)
8 (17.64)
320 (539.4)
2.17 (4.78)
86.8 (146.3)
8.7 (19.18)
348 (586.8)
9.15 (20.17)
366 (616.9)
5.65 (12.46)
226 (380.9)
6.5 (14.33)
260 (438.2)
11.3 (24.91)
452 (761.9)
(CaCO3) from Kalkfabrik Netstal AG, Switzerland and quartz flour (SiO2) was obtained
from Dorfner GmbH, Germany.
Test specimens
Mortars were prepared from sand and aggregates up to 8 mm (0.31 in) and concrete up
to 32 mm (1.26 in). The exact mixture proportions are depicted in Table 2.
The w/b was chosen corresponding to the added HRWR. All admixture dosages are
based on the weight of binder. The experiments were carried out at 20C (68F). Mortar
prisms of the size 160 x 40 x 40 mm (1.57 x 1.57 x 6.30 in) were casted at 20C (68F)
according to EN 196-1 for the determination of compressive strength and for the measure-
362SP-302-27
Fig.1Compressive strength of mortars: Influence of CaO,

Booster and ACC1 on CEM I 52.5 R with quartz flour and
HRWR1 at 0.5% by weight of binder (w/b 0.42)
ment of the expansion. For the expansion measurement the prisms were demoulded after
24 hours and cured under water at 20C (68F).
The flow table spread (FTS) and air content of the mortar and concrete mixes were
determined according to EN 196-1 and EN 206 (concrete), respectively. The compressive
strength of mortar prism and concrete cube specimens was tested at various ages according
to EN 196-1 for mortar and EN 206 for concrete. The shrinkage or expansion of the mortar
specimens were measured according to EN 196-1 as well.
General
Classical hardening accelerators such as ACC 1 are very efficient strength improvers.6
They are optimized to increase the compressive strength from early times up to later ages.
The results of compressive strength tests in a classical precast application are shown in
Fig. 1. The strength increase with ACC1 is significant, starting from 6 hours on. Calcium
oxide at a dosage of 3% leads to a remarkable increase in compressive strength as well.
The combination of calcium oxide with a classical hardening accelerator further improves
the early strength development. The booster under investigation shows no or only limited
impact when used alone. This seems to be reasonable, as it is a phosphate compound, which
are normally used as retarding components in admixture formulations.11 However, when
this booster component is used in combination with calcium oxide the most significant
strength increases are obtained in this series. This is especially interesting as the amount
of booster used is very low compared to the addition of the classical hardening accelerator.
It was also noted that the initial workability of the mortars was drastically reduced when
the booster component was added to the mix. This can be either due to a very fast initial
reaction or due to a competitive adsorption between HRWR and booster.
The effect of the booster with and without calcium oxide was also investigated via calorimetric experiments as shown in Fig. 2 and 3. It can be seen, that both plasticizer and
Fig. 2- Calorimetry curves of CEM I 42.5 N with HRWR2

and booster without CaO
Fig.3 Calorimetry curves of CEM I 42.5 N plus 3% CaO

with HRWR2 and booster
booster induce a delay in the hydration kinetics, when no calcium oxide is added. When
calcium oxide is present, the induced retardation by the plasticizer can be completely
recovered thanks to the booster. This probably accounts for a specific interaction between
the booster and calcium oxide in the early stages of the reaction.
364SP-302-27
Table 3 - Influence of CaO, Booster and ACC1 on CEM I 52.5 R with HRWR1
at 0.5% by weight of binder at w/b 0.42 with limestone as filler
Admixture
Dosage
[%]
Reference
Booster
CaO
CaO
Booster
ACC1
ACC1
CaO
0.075
3
3.0
0.075
2.0
2.0
3.0
Flow table spread

at 3 min, mm (in)
4h
Compressive strength, MPa (psi)

6h
8h
175 (6.9)
3.0 (435)
6.2 (899)
170 (6.7)
165 (6.5)
0.9 (131)
1.3 (189)
2.6 (377)
3.9 (566)
5.4 (783)
7.9 (1146)
132 (5.2)
1.4 (203)
4.2 (609)
8.2 (1189)
154 (6.1)
1.2 (174)
4.0 (580)
10.2 (1479)
154 (6.1)
2.5 (363)
8.5 (1233)
15.7 (2277)
Trials with limestone

In the trials shown in Table 3 the filler component was exchanged and limestone was
used instead of quartz flour. Apart from that the experiments carried out are the same as
the ones shown in Fig. 1. The strength of the reference mortar is definitely higher (100%)
compared to the reference with quartz flour. This might indicate an accelerating effect of
the fine calcium carbonate which is in line with the data from earlier studies.12,13 ACC1
yields similar results as before, alone and in combination with calcium oxide, proving the
robustness of this type of accelerator. The use of calcium oxide alone gives strength values
on a similar level as with quartz flour as well. However, the results with the booster component are not similar anymore. The booster alone even results in slightly lower values than
the reference mortar. When the booster component is used in combination with calcium
oxide, which has resulted in the highest strength gain before, basically no beneficial effect
can be noted. The strength levels in this combination are similar as with the pure calcium
oxide.
This indicates that the type of filler has an influence on the performance of this booster
component. To further evaluate this influence of the type of fines on the performance of the
different systems, mortars with different ratios quartz flour / limestone filler were tested.
The results thereof are shown in Table 4.
The resulting strength values confirm the trends and results as obtained in the experiments before. The mortars with high range water reducer alone again show an increasing
strength performance with an increasing amount of limestone powder. The trials with a
classical hardening accelerator ACC1 yield similar high strength values irrespective of the
type of fines, as do the mortars with the addition of calcium oxide only. The results with the
combination calcium oxide / booster component show a clear dependence on the content
of limestone addition. The higher the limestone content the lower the accelerating effect
of this combination. However, the strength values with 75% limestone filler are still on the
level of the results obtained with the addition of a classical hardening accelerator ACC1.
Regarding the flowability of the mortars two things seem to be evident as well:
The obtained values are generally lower when quartz flour is used, which might be a
result of the higher fineness of this filler material. The values are drastically reduced when
the booster component is used together with calcium oxide, especially when there is a
significant accelerating effect.
Table 4 Influence of the fines on the performance in CEM I 52.5 R with

HRWR1 at 0.5% by weight of binder at w/b 0.42
Admixture
Dosage
[%]
Reference
Reference
Reference
Type of fines
75% limestone
20% quartz flour
50% limestone
50% quartz flour
Flow table spread

at 3 min, mm (in)
100% quartz flour
CaO
3.0
CaO
3.0
CaO
CaO
Booster
CaO
Booster
CaO
Booster
3.0
3.0
0.075
3.0
0.075
3.0
0.075
ACC1
2.0
ACC1
2.0
75% limestone
25% quartz flour
50% limestone
50% quartz flour
100% quartz flour
75% limestone
25% quartz flour
50% limestone
50% quartz flour
100% quartz flour
50% limestone
50% quartz flour
100% quartz flour

4h
6h
8h
170 (6.7)
1.6 (232)
4.9 (711)
160 (6.3)
1.9 (275)
4.5 (653)
168 (6.6)
1.5 (218)
3.9 (566)
165 (6.5)
1.0 (145)
2.4 (348)
6.4 (928)
155 (6.1)
1.1 (159)
3.1 (449)
6.8 (986)
152 (6.0)
1.1 (159)
2.9 (421)
7.7 (1117)
125 (4.9)
2.0 (290)
6.4 (928)
11.2 (1624)
120 (4.7)
3.1 (449)
8.5 (1233)
14.2 (2060)
118 (4.6)
3.2 (464)
9.8 (1421)
15.5 (2248)
158 (6.2)
1.1 (159)
4.4 (638)
9.3 (1349)
158 (6.2)
1.1 (159)
4.4 (638)
10.9 (1581)
Table 5 - Increase of Booster on CEM I 52.5 R with 3% CaO with quartz flour
as filler with HRWR1 at 0.5% by weight of binder at w/b 0.42
Admixture
Dosage
[%]
Reference
Booster
Booster
Booster
Booster
0.05
0.1
0.15
0.2

6h
8h
Flow table spread at

3 min, mm (in)
4h
157 (6.2)
1.1 (159)
2.9 (420)
7.6 (1102)
125 (4.9)
120 (4.7)
112 (4.4)
110 (4.3)
0.9 (131)
1.6 (232)
2.4 (348)
2.1 (305)
3.4 (493)
6.0 (870)
8.9 (1291)
8.6 (1247)
8.3 (1204)
12.1 (1755)
14.8 (2146)
14.5 (2103)
The dependence of the booster effect on the type fines would of course be a limiting
factor as in real life the type of fines cannot be controlled or exchanged.
The effect of the booster component was therefore further investigated and a series of
mortar strength tests was performed with variable dosage of the phosphoric acid ester,
keeping the calcium oxide content constant at 3%.
The results of these tests are shown in Table 5 and 6, in mortars with limestone and
quartz flour as filler component.
It can be seen that an increase of the dosage of the booster component leads to higher
strength at all ages tested, irrespective of the filler component used.
In the case of quartz flour as filler this acceleration effect is already remarkable at an
addition of 0.05% of the booster and seems to reach an optimum at 0.15%, at least in the
dosage range tested.
366SP-302-27
Table 6 - Increase of Booster on CEM I 52.5 R with 3% CaO with limestone

as filler with HRWR1 at 0.5% by weight of binder at w/b 0.42
Admixture
Dosage
[%]
Reference
Booster
Booster
Booster
Booster
0.05
0.1
0.15
0.2
Flow table spread at

3 min, mm (in)
Compressive strength [MPa] psi c)

4h
6h
8h
150 (5.9)
1.1 (159)
3.4 (493)
7.8 (1131)
130 (5.1)
130 (5.1)
128 (5.0)
126 (4.9)
0.8 (116)
0.9 (131)
1.3 (189)
1.6 (232)
3.0 (435)
3.9 (566)
5.5 (798)
6.6 (957)
7.6 (1102)
9.1 (1320)
11.1 (1610)
12.7 (1842)
Fig.4 Compressive strength of concrete tests with CEM I

52.5 R, w/b 0.45
When limestone is used as the filler component the strength values are lower than with
quartz flour at the same dosage of the booster compound. This is as expected from the
earlier experiments. An acceptable performance is obtained at 0.15% and 0.2%. However,
compared to quartz flour, an increase in the dosage of the booster component leads to a
continuous increase in strength over the whole dosage range tested. It might therefore be
assumed that a further dosage increase would lead to an even more pronounced acceleration; however, this would have to be verified.
It is still evident that the workability of the mortars is drastically reduced in all cases and
such a performance would not be realistically acceptable.
Therefore, the authors decided to check the performance of the system in concrete and
adjusted the initial plasticity of the tests to a comparable level via an increased dosage of
the high range water reducer. The results of this test series are shown in Fig 4.
The tests with booster need significantly more plasticizer (0.7% and 0.9% vs. 0.5% for
the reference) to reach the targeted initial workability. The results in concrete reflect the
behavior in the mortar systems tested before.
As the concrete system contains limestone as filler component, the dosage of the booster
was chosen as 0.15% according to the evaluation in the mortar system. The strength
Table 7 - Performance in blended system CEM I 42.5 N/FA (3:1) at w/b 0.42
and use of HRWR1 at 0.5%
Admixture
Dosage
[%]
Reference
CaO
CaO
Booster
3
3
0.15

12 h
16 h
20 h
Flow table spread

at 3 min, mm (in)
9h
232 (9.1)
1.2 (174)
5.2 (754)
238 (9.4)
3.4 (493)
7.5 (1088) 15.2 (2205) 21.8 (3162) 26.2 (3800)
170 (6.7)
4.2 (609)
9.3 (1349) 18.2 (2640) 24.5 (3553) 29.0 (4206)
7.2 (1044)
24 h
14.1 (2045) 18.1 (2625)
increase of the booster system with CaO is high at 6 h and 8 h (300% vs. reference) and
still significant after 1 day (20% higher than the reference).
Due to the increased HRWR dosage, the negative impact of the system on the initial
plasticity could be overcome as well.
It is clear that such a powerful accelerator system can have its application in normal
precast systems based on ordinary portland cements as was shown in the experiments up to
now, but it might also be a very interesting approach in blended systems. The experiments
in Table 7 show such an application in a system with 25% of fly ash as cement replacement.
The results show that the booster component can also lead to a significant strength
increase in this system. The effect of the calcium oxide is significant in this fly ash system,
most probably due to an activation effect of the puzzolanic reaction. The combination of
the booster and the calcium oxide leads to an impressive strength increase compared to the
reference mortar starting from 8 h on.
These results are also supported by calorimetric data. In Fig. 5 the time to reach the main
peak of hydration is shown. It can be seen that the increase of the dosage of the booster
component in a mix CEM I 42.5 N/FA (1:1) leads to a significant shift of the main peak of
hydration to earlier times, which would mean a higher compressive strength at early age.
One of the drawbacks of CaO can be its expansive reaction. This is desirable in some
cases and it is, therefore, also used in some expansive systems. However an expansion at
later ages might lead to a disintegration of the structure.
Therefore, the effect of calcium oxide in the normal dosage of 3%, but also at 6% and
9% was investigated and the length change upon storage under water was measured. The
behavior over time (91 days) is illustrated in Fig. 6. The calcium oxide under investigation
only shows a higher expansion at a very high dosage of 9%. However, even this level of
expansion should not bear any negative consequences.14 At the normal dosage of 3% only
a negligible expansion is measured.
CONCLUSIONS
With the booster component a totally new accelerator could be found. From its chemical
nature, being a phosphoric acid ester, it is supposed to be a retarding admixture and the
experimental data also show that it delays the hydration and the strength development when
it is used on its own. In combination with calcium oxide, however, this booster component
is able to accelerate the strength development of ordinary portland cement systems as well
as blended cements as the results of combinations with fly ash are indicating. The early
compressive strengths obtained are higher than the ones obtained with classical types of
accelerators containing inorganic salts. These inorganic salts bear the risk of promoting
368SP-302-27
Fig. 5 Time to reach main peak of hydration for CEM I /

FA (50: 50) with 3% CaO, 0. 16% HRWR1 and increasing
amount of booster (Calorimetry measurements)
Fig.6 Length change of mortars (CEM I 52.5 R, HRWR1

0.7%, 0.15% booster, w/b 0.45), stored under water, with
different amounts of CaO
steel corrosion which could be even more pronounced as these accelerators have to be used
at rather high dosage. The booster component can be used at very low dosages and is, from
its chemical nature, not corrosive for steel. On the contrary, phosphates are often used as
components in corrosion protection or corrosion inhibitor systems.
Calcium oxide as main part of the accelerator system might even be beneficial in corrosion protection, being able to increase the pH of blended cement systems. The risk of
delayed expansion, as it might occur with calcium oxide, seems to be limited in this case.
The question about the chemical mechanism of the booster effect could not be answered
yet. It seems to be clear that the combination calcium oxide/booster is relevant for a very
fast reaction as the workability drops dramatically in this case and has to be compensated
with additional HRWR. However, even though additional HRWR might delay the hydration, the early strength is still extremely promoted.
It seems that the booster component is less effective in cases where a lot of calcium
carbonate is present. However, this reduced effectiveness can be compensated by increasing
the dosage of the booster. This might indicate that the booster is either absorbed by or
adsorbed on the calcium carbonate. The real reason for this reduced effectiveness is not
known yet and therefore possible assumptions are pure speculation.
The conclusion that the system calcium oxide / booster can be a new and powerful accelerator system in pure and blended cement systems with only little side effects, however,
seems to be valid.
AUTHOR BIOS
Franz Wombacher is Head of the Technology Center of Sika Technology AG, Construction Chemicals & Mortars in Zurich/Switzerland. He received his M.S. in Chemistry and
the Ph.D. in Inorganic Chemistry from the Swiss Federal Institute of Technology (ETH),
Zurich/Switzerland. He is author of several papers and patents in the field of concrete
admixtures, shotcrete accelerators and corrosion protection of concrete structures.
Christian Brge is Department Manager Admixtures at Sika Technology AG, Construction Chemicals & Mortars in Zurich/Switzerland. He received his B.S. in Chemical Engineering from the University of Applied Sciences, Winterthur/Switzerland. He is author of
several papers and patents in the field of concrete admixtures.
Emmanuel Gallucci is Head of the Central Research department on Construction
Materials of Sika Technology AG. He received a PhD in Physical Chemistry from the
University Lyon/France. He is author of several papers in the field of cement chemistry
and admixtures.
Patrick Juilland is Research Scientist at Sika Technology AG. He received his M.S. and
PhD in Material Science form the Swiss Federal Institute (EPF) Lausanne/Switzerland.
He is author of several papers in the field of cement chemistry, admixtures and shotcrete
accelerators.
Gilbert Mder is Laboratory Manager at Sika Technology AG, Construction Chemicals
& Mortars in Zurich/Switzerland.
REFERENCES
1. Scrivener, K. L., Options for the future of cements, Indian Concrete Journal, V. 88,
No. 7, 2014, pp. 11-21.
2. Bellmann, F. et al., Reduction of CO2 emissions during cement clinker burning, part
1, Cement International, V. 10, (1/2013), pp. 62 71.
3. Min, T. B.; Cho, I. S.; Park, W. J.; Choi, H. K.; and Lee, H. S., Experimental study on
the development of compressive strength of early concrete using calcium-based hardening
370SP-302-27
accelerator and high early strength cement, Construction & Building Materials, V. 64,
2014, pp. 208-214. doi: 10.1016/j.conbuildmat.2014.04.053
4. Thomas, J. J.; Jennings, H. M.; and Chen, J. J., Influence of Nucleation Seeding on
the Hydration Mechanisms of Tricalcium Silicate and Cement, The Journal of Physical
Chemistry C, V. 113, No. 11, 2009, pp. 4327-4334. doi: 10.1021/jp809811w
5. Nicoleau, L., Accelerated growth of calcium silicate hydrates: Experiments and
simulation, Cement and Concrete Research, V. 41, No. 12, 2011, pp. 1339-1348. doi:
10.1016/j.cemconres.2011.04.012
6. Wombacher, F.; Brge, C.; Kurz, C.; and Marazzani, B., Hardening accelerators for
blended cements New approaches to improve strength development, Supplementary
papers Tenth International Conference on Superplasticizers and Other Chemical Admixtures in Concrete, Prague, Czech Republic, Oct. 28 -31, 2012, pp. 49 59.
7. Schaefer, S., Sodium Thiocyanate: The alternative to non-chloride accelerators,
Supplementary papers Eighth International Conference on Superplasticizers and Other
Chemical Admixtures in Concrete, Sorrento (Italy), Oct. 29 Nov.1, 2006, pp. 381 387.
8. Riding, K.; Silva, D. A.; and Scrivener, K., Early Age Strength Enhancement of
Blended Cement systems by CaCl2 and Diethanol-Isopropanolamine, Cement and
9. Eichler, W. R., and Manns, W., Zur korrosionsfrdernden Wirkung von thiocyanathaltigen Betonzusatzmitteln, Betonwerk + Fertigteil-Technik, V. 82, No. 3, 1982, pp.
154-162.
10. Corbo, J. M., and Nmai, C. K., Sodium Thiocyanate and the Corrosion Potential
of Steel in Concrete and Mortar, Concrete International, V. 11, No. 11, 1989, pp. 59-67.
11. Oppliger, M., and Tsohos, G., An Introduction to Concrete Admixtures, 2007,
University Studio Press, Thessaloniki, Greece.
12. Lothenbach, B.; Le Saout, G.; Gallucci, E.; and Scrivener, K., Influence of limestone on the hydration of Portland cements, Cement and Concrete Research, V. 38, No. 6,
13. Ramachandran, V. S., Thermal analyses of cement components hydrated in the presence of calcium carbonate, Thermochimica Acta, V. 106, September, 1988, pp. 273-282.
14. Neville, A. M., Properties of Concrete, 4th edition, 1995, Pearson Educ. Ltd, Edinburg Gate, Harlow, England.
SP-302-28
Sorption Kinetics of Superabsorbent

Polymers in Cement Pastes Quantified by
Neutron Radiography
by Christof Schroefl and Viktor Mechtcherine
Water desorption from superabsorbent polymers (SAP) into cement-based pastes was
characterized by neutron radiography imaging to promote the understanding of the mechanisms behind internal curing of concrete. Two anionic SAP samples were used which
differed in their inherent sorption kinetics in cement pore solution (SAP 1: self-releasing;
SAP 2: retentive). Portland cement pastes with W/C of 0.25 and 0.50 and a paste additionally containing silica fume (W/C = 0.42, SF/C = 1/10) were investigated.
Desorption from SAP 1 initiated immediately. SAP 2 released water into all the matrices
as well, even in the cement paste with the high W/C of 0.50. In the other two pastes,
which require internal curing by principle, SAP 2 retained its stored liquid for as long
as the dormant period of cement hydration. Intense desorption then set in and continued
throughout the acceleration period and even beyond.
These findings explain the pronouncedly higher efficiency of SAP 2 as an internal curing
admixture when compared to SAP 1.
Keywords: cement hydration; cement paste; internal curing; neutron radiography
imaging; portland cement; silica fume; superabsorbent polymer.
INTRODUCTION
Superabsorbent polymers (SAP) have evolved as multifunctional admixtures in highperformance concretes due to their ability to quickly absorb water and release it slowly
into the hydrating, cement-based matrix. They were found to be especially efficient as
admixtures for internal curing to mitigate autogenous shrinkage in portland cement-based
construction materials with low water-cement ratios (W/C, water by weight of cement).
They significantly reduce the potential of crack formation caused by self-desiccation.
By now, their functioning related to this effect of internal curing (IC) has well been
described in principle.1-4 The majority of studies have been performance-oriented. Only
few has been published on mechanisms and kinetics of water migration within the cementbased material in detail, including, for example migrating amounts of water from SAP into
the surrounding matrix or the onset and duration of water release.4-6 Wyrzykowski et al.4
371
372SP-302-28
have described a so-called demand-supply mechanism based on the findings with their
individual SAP sample. They found that liquid is extracted from swollen SAP at that point
of time which equals a first significant drop of internal relative humidity. Utilizing neutron
tomography, Trtik et al.6 visualized and quantified the water release from one SAP particle
into a cement paste. They linked these kinetics to the progress of cement hydration and
concluded that the onset of water release from the SAP particle coincides with the transition from the dormant to the acceleration period.
None of these manuscripts disclosed chemical details of the SAP particles. Comparative
studies regarding differently composed SAPs and focusing on their molecular sorption
mechanisms in cement-based materials have been very rare. One approach showed how
different SAPs perform in a cement mortar depending on their cross-linking density and
their relative density of anionic functional groups.7-10
Subsequently, it will be of essential importance to deepen the mechanistic knowledge
based on experimental data within the cement-based pastes and link the characteristic
behaviors of SAPs both to their chemical composition and the progress of cement hydration.
Neutron radiography imaging has evolved as a powerful method in observing water
migration inside hardening and hardened cement-based specimens, for example.6,11-14
The work at hand used neutron radiography imaging to quantify water migration from
two individual, chemically different acrylate-based SAP samples into pastes of normal
portland cement (W/C = 0.25 and W/C = 0.50) and one made of normal portland cement
and silica fume (W/C = 0.42, silica fume 10 wt-% by weight of cement [bwoc]).
Macroscopic, performance-based results on the effects of two superabsorbent polymer
(SAP) samples are explained on behalf the sorption kinetics of these SAPs directly within
cement mortars. Free sorption tests in extracted cement pore solution had up to date been
the method of choice. It is now demonstrated that neutron radiography imaging is the most
meaningful technique to quantitatively characterize and distinguish the sorption kinetics of
differently composed SAP samples. Both rheological aspects of mortars and efficiencies in
mitigating autogenous shrinkage could be explained on the basis of the presented results.
Materials and specimen preparation
Pastes of cement, SAP, water, and where applicable high-range water-reducing admixture or silica fume were prepared right before beginning neutron radiography imaging and
calorimetry, respectively. The cement was a rapid hardening, normal portland cement type
I with no extra defined properties, CEM I 42.5 R, according to DIN EN 197-1.15 Its composition is provided in Table 1. The silica fume was a dry powder with an average particle
size of 85 nm (3.3 10-6 in.) and a purity of 97 wt-% amorphous SiO2. As high-range waterreducing admixture a polycarboxylate-based commercial product was used as obtained as
aqueous solution.
Two superabsorbent polymers were used, both of which were synthesized by means of
block polymerization and crushed into small pieces after synthesis (R&D samples provided
by an external partner). SAP 1 had the maximum relative anionicity; it was produced of
acrylic acid as the only main monomer. Its cross-linking density is qualitatively rather high,
Sorption Kinetics of Superabsorbent Polymers in Cement Pastes Quantified

by Neutron Radiography 373
Table 1 Chemical and mineralogical composition of the portland cement
Chemical composition [wt-%]
SiO2
19.4
Al2O3
4.9
Fe2O3
3.1
CaO
62.8
MgO
3.3
SO3
3.2
K2O
0.9
Na2O
0.2
CO2
1.7
H2O
0.6
Loss on ignition
2.4
Mineralogical composition (p-XRD, Rietveld refinement) [wt-%]

C3S
54.2
C2S
17.0
C3A cubic
7.1
C3A orthorhombic
1.1
C4AF
7.1
Periclase
2.6
Quartz
0.1
Arcanite
0.9
Calcite
3.0
Anhydrite
3.9
Bassanite
1.4
Gypsum
0.1
Lime
0.7
Portlandite
0.8
Table 2 Particle size distributions of SAP 1 and SAP 2 in the dry state
d10 [m] (in.)
d50 [m] (in.)
d90 [m] (in.)
SAP 1
224 (0.0088)
586 (0.0231)
1029 (0.0405)
SAP 2
386 (0.0152)
903 (0.0356)
1446 (0.0569)
and especially it is higher than that of SAP 2. The relative anionicity of SAP 2 is below that
of SAP 1. SAP 2 is a cross-linked copolymer composed of the two main monomers acrylic
acid and acryl amide. These SAP samples had been used in previous studies: in,7,8 present
SAP 1 was denominated as SAP B and SAP 2 as SAP D. The present SAP 1 was called
SAP 2 in9 and SAP 2 is SAP 4 in that study, respectively. SAP-B in10 was present
SAP 1 and SAP-DS, -DC, -DN were three gradings of SAP 2.
Their performance in cement-based mortar has been discussed in7-10 with respect to their
correlation between structure/composition, impact on rheology and efficiency in mitigating
autogenous shrinkage. Table 2 summarizes the characteristic values of their particle size
distributions in the dry state (laser granulometry; dispersive liquid: propan-2-ol).
Sorptivities of SAP 1 and SAP 2 in the filtrate of the cement used were quantified by the
so-called tea-bag method as described in.7 Approximately 0.2 to 0.3 g (0.44 10-3 to 0.66
10-3 lbs.) of SAP particles with the exact mass being m1 were inserted in a tea-bag which
had the tare weight m2 after pre-wetting. The tea-bag containing the polymer sample was
completely immersed in a beaker filled with the excess of approximately 250 mL (0.33 10-3
yd3) of the cement pore solution. The beaker was sealed tightly with a plastic foil, which
was only briefly released to conduct the weighing. In a time-resolved series, the tea-bag
with the polymer was taken out and weighed, providing the soaked mass m3. It was carefully wiped with two dry cloths for a short time to remove surplus and weakly bound solution. Some minor amount of capillary water might remain between the individual particles,
which was considered as to be irrelevant for the overall result. On the other hand, the
374SP-302-28
Fig. 1 Sorptivity of the SAP samples in filtrate of cement

suspension for 24 hours as determined by the tea-bag method
(CEM I 42.5 R, W/C = 4.3, cf.7)
tea-bag was not to be squeezed so as to not interfere with the free sorption capability of the
moistured SAP particles. The ratio of intaken liquid related to the initial dry mass of the
polymer was calculated according to Eq. 1.

m absorbed = (m3 m 2 m1 ) / m1
(Eq. 1)
In the study at hand, the time was extended from three hours utilized in7 to 24 hours to
cover the entire period of neutron radiography imaging (Fig. 1).
Three kinds of pastes were regarded in the present study (Table 3). The SAP dosage was
0.3 wt-% bwoc in any case. This proportion accords to the common dosage for efficient
internal curing of high-performance mortar.7-9
All mixtures were prepared in a small casserole by first thoroughly preblending all dry
substances (cement, SAP and where applicable silica fume) with a spoon. The mixing
water was added within 10 seconds. In the case of W/C = 0.25, the high-range waterreducing admixture had completely been pre-dissolved in this water. The pastes were
manually stirred with the spoon for two minutes until a homogenous paste had formed.
For neutron radiography imaging an appropriate portion of the material was filled into an
aluminum container in such a way that no voids remained. All pastes were sufficiently flowable and this placement could be conducted without any necessity of further compaction.
The paste volumes put into the containers were not controlled but an experimentally suitable portion was used only. This portion was extracted arbitrarily from the entire prepared
amount. With the samples Cem-SF-042 the containers had the dimensions of width/height/
depth = 80/30/10 mm/mm/mm (3.14/1.18/0.39 in./in./in.), while smaller boxes were used
for the other two types of paste. Their dimensions were width/height/depth = 30/30/5 mm/
mm/mm (1.18/1.18/0.20 in./in./in.). The top was closed to airtightness with a self-adhesive
aluminum tape. Special attention was paid to a quick handling and transfer of the samples
into the neutron beam and start of image recording. Naturally, the relative water contents
are not detectable in the time between paste preparation and this earliest image recorded.

Table 3 Paste compositions for neutron radiography imaging
Material, property
Cement
Tap water
W/C
Silica fume (dry powder)
SAP (dry powder)
PCE high-range water reducing admixture
(aqueous solution, solids content 35 wt-%)
Unit, dimension
g
g
g/g
g
g
Cem-025
70.00
17.25
0.25
0
0.21
Cem-050
70.00
35.00
0.50
0
0.21
Cem-SF-042
70.00
29.40
0.42
7.00
0.21
0.70
Thus, the time between paste preparation and the start of image recording should be as
short as experimentally feasible. With the pastes of W/C = 0.25 this time span was 22
minutes and for those of W/C = 0.50 it was 26 minutes. For the paste containing SAP 1,
silica fume and featuring a W/C of 0.42 it was seven and for the respective one with SAP
2 it was eight minutes. Consequently, all representations and interpretation of sorption
kinetics start at these respective times of delay after paste preparation.
Neutron radiography imaging and image processing
Neutron radiography imaging was conducted at the neutron imaging and activation
group of Paul Scherrer Institut (PSI) in Villigen/Aargau, Switzerland, with their beam
line NEUTRA (for details, cf.11,12). The neutrons stem from the spallation source SINQ,
in which protons from the Swiss ring cyclotron impinge a lead target that releases the
neutrons as a steady beam. Within the beam line, the samples were placed on position 2.
After passing through the sample, the remaining neutron beam intensity is transformed
into visible light by a 6LiF/ZnS scintillation screen, which is directed to a CCD camera via
mirrors.
Open beam and dark current images were acquired before each time series. The open
beam (or flat field) image represents the spatial distribution of the neutron beam intensity, in which the full neutron flux directly impacts the scintillation screen without passing
through any solid. The dark current image was recorded to measure the bias introduced
by the background noise level of the CCD camera without any neutrons approaching (all
shutters in the beam line closed).
Pictures were processed and quantitatively evaluated in an established style of timeresolved differential image series for water migration as described earlier, for example
in.13,14 The time series images were processed with the open source image-processing tool
ImageJ using built-in functions and a plugin for image normalization developed by the
Neutron Imaging and Activation Group at PSI.16
Quantification of relative water contents in time
The target of quantitative image analysis was to elucidate the evolution of the water
content in one spot of the sample with respect to the earliest state recorded. A relative water
content with respect to the initial state was calculated via referencing each image of the
time sequence to the earliest image recorded. The results are normalized pictures, in which
the grey value of each pixel directly represents the relative water content with respect to
the reference image.
376SP-302-28
Fig. 2 Relative water content inside the SAP particles and

the cement matrix in the specimens with W/C = 0.25
Every spot of several pixels in size which gets continuously brighter in the course of
time unambiguously shows one SAP particle. The SAP particles are the only components
in the paste which are able to lower their water content in the course of cement hydration
and become visible in the selected technology of measurement. As a consequence of the
careful sample preparation procedure and the well-flowable consistency of all pastes, these
spots are not to be ascribed to compaction pores. Each of these SAP spots was individually
evaluated by measuring the relative grey value of a centered region of 25 to 150 pixels.
In between these spots, plain matrix remained in a visually intermediate shading
throughout the time of image recording. Quantitative results, however, disclosed a slight
but still significant increase in the relative water content in the course of time.
For self-control, the grey shading of the entire specimens was evaluated over time. This
should identify whether exchange of moisture with the surrounding occurred to a detectable extent, which might indicate that the sealing was not hermetic. It was found that
this overall water content did not change throughout the entire period of image recording.
Hence, autogenous conditions were preserved all the time.
Sorption kinetics of SAP 1 and SAP 2 in cement paste with W/C = 0.25
containing a polycarboxylate-based high-range water-reducing admixture
The relative water contents inside the hydrogel particles and the SAP-free matrix in the
specimens prepared with W/C = 0.25 shows Fig. 2. Any spot of significant continuous
change in water content over time and at least 25 pixels in size was interpreted as one SAP
particle. This way, 54 individuals of SAP 1 and 46 of SAP 2, respectively, were identified.
The error bars indicate the standard deviations based on the average values in each time of
image recording.
The water content in the matrices increased slightly in both pastes. This indicates
internal curing, that is water from the hydrogels relocates in the cement matrix.

Both SAPs released rather similar amounts of water as the overlapping error bars show.
However, it is worth noting that the average values feature individual curve progressions
for several hours. Right from the start of imaging, which is 22 minutes after paste preparation, SAP 1 released water. Contrarily, SAP 2 did not lower its water content for up to
approximately 1.5 hours but kept it rather constant or potentially showed even some minor
delayed absorption. However, SAP 2 initiated desorption after 1.5 hours.
SAP 1 had disclosed itself as an incontinent hydrogel with respect to cement pore solution in the tea-bag test (Fig. 1). This fact reasons the desorption from SAP 1 right from the
start in the cement paste as well.
However, SAP 2 had been found to not give away absorbed extracted cement pore solution individually but to retain it for at least 24 hours (Fig. 1). Thus, desorption from SAP
2 has to be triggered by the hydrating cement paste. The progress of cement hydration at
this very low W/C of 0.25 causes a rather quick lack of freely available water which finally
results in internal self-desiccation2,3. Wyrzykowski et al.4 formulated a demand-supply
mechanism to reason desorption from their individual SAP sample into the cement-based
structure (W/C = 0.25). Their approach can be applied to present SAP 2 and explain its
desorption.
Sorption kinetics of SAP 1 and SAP 2 in cement paste with W/C = 0.50
Fig. 3 displays the relative amounts of water in the cement pastes prepared with a W/C
of 0.50. Similar to the pastes with W/C = 0.25, the matrix is enriched in water in the
course of time, while the SAPs are releasing intensely. Only 18 spots could undoubtedly
be assigned to hydrogel particles in the specimen containing SAP 1. This amount seems to
be rather low when compared to any other of the pastes. However, due to the overall rather
high water content and even higher ones in the centers of the SAPs, sample-scattering and
detector-backscattering of neutrons may impede quantification.17,18 In the present mode
of evaluation, the 18 established spots were considered only and no further mathematical
treatment was applied.
The behavior of SAP 1 well corresponds to its inherent sorption kinetics as obtained by
the tea-bag method (Fig. 1). However, the considerable error bars indicate a pronounced
non-uniform behavior of the 18 individual SAP 1 particles, especially from four hours
onwards. To account for this, each spot was individually re-considered. It turned out that a
group of seven (A) had a similar and very high extent of release while eleven of them had
similarly uniform extent of desorption (B) with a pronouncedly lower amount of released
water (Fig. 4). As a trivial approach of interpretation, it may be claimed that at least two
hydrogel particles were aligned in the direction of projection. However, due to the geometry of the sample container with a depth of 5 mm (0.2 in.) only, such an alignment may
be considered as rather improbable. Additionally, if this were the case, the release may be
expected to be double or three times right from the start and the splitting of the intensities
not initiate at a delayed point of time. Positively speaking, one can conclude that the less
active portion of particles (SAP 1 B) ran out of desorbable water as early as two to three
hours after paste preparation whereas the other particles (SAP 1 A) continued. Local in
homogeneities in the surrounding matrix or slightly differing individual polymer characteristics may be further reasons.
378SP-302-28

the cement matrix in the specimens with W/C = 0.50; all
particles of SAP 1 in average

the cement matrix in the specimens with W/C = 0.50; two
portions of SAP 1 respected
The pronouncedly enhanced desorbed amount of water in comparison to that in the
paste featuring W/C = 0.25 (Fig. 2) may be explained as follows. In cement paste with
a higher W/C a clearly bigger amount of Ca2+ ions dissolves into the pore solution.19,20
This enhanced Ca2+ concentration intensifies a screening of the water retention ability
of the hydrogel and results in a significantly enhanced desorbed amount of water on the
long term.21 Interestingly, there was no correlation between the size of a SAP spot and its
released portion of water.
While desorption from SAP 1 in the paste with the rather high W/C of 0.50 is reasonable in conjunction to its individual sorption kinetics as derived from the tea-bag test, the
desorption from SAP 2 in this paste is not quite plausible at first glance. In the cement paste


the cement matrix in the specimens with W/C = 0.42 and
containing 10 wt-% silica fume bwoc
featuring W/C = 0.50 no internal curing is required in principle2,3 and a demand-supply
mechanism based on the drop of relative humidity and capillary pressure buildup4,22 is
most unlikely to apply. Hence, these approaches should be withdrawn in this case. As well,
during the relevant times of shorter than seven hours after paste preparation, the ionic
conditions in the liquid phase remain almost constant.19,20 The concentration of dissolved
Ca2+ does not pronouncedly change during that time and should, hence, not come into
question to trigger the release from SAP 2. It is worth noting that Figs. 3 and 4 clearly
indicate water loss from SAP 2 right from the begin of image evaluation. Most likely, the
concentration of Ca2+ by itself can be as high as to screen the inherent retentiveness of SAP
2 and have it release rather individually.21 However, the question of why SAP 2 released
stored liquid in this kind of cement paste will have to be elucidated in future studies.
Sorption kinetics of SAP 1 and SAP 2 in paste of cement and silica fume
with W/C = 0.42
In contrast to the pastes discussed above, the third pair of specimens contained silica
fume as a further fine and pozzolanic compound. The water contents in the paste portions
as well as in the hydrogels SAP 1 and SAP 2, respectively, shows Fig. 5. 87 individual
particles of SAP 1 and 63 of SAP 2 were found and evaluated in the corresponding pastes.
These particle numbers are much higher than those in the other pastes but the sample
volume has been significantly bigger as well. In the specimen containing SAP 1 there were
two individual spots which released fairly well the double amount of all the residual ones
throughout the entire time of neutron radiography imaging. This finding indicates that two
particles were aligned along the direction of projection at these positions. As these were
two instances only, they remained unconsidered in all further data evaluation.
Similar to the plain cement-based pastes the water content in the SAP-free portions of
the present specimens increased visibly. Water given away from the hydrogels redistributed into the matrices, which is the fundamental mode of internal curing1. SAP 1 released
water right from the start of imaging, which was seven minutes after paste preparation.
380SP-302-28
Fig. 6 Isothermal calorimetry of the specimens with W/C =

0.42, containing 10 wt-% silica fume bwoc, heat flux
The reason is its inherent sorption characteristic with respect to extracted cement pore
solution (Fig. 1). Contrarily, SAP 2 kept its water content for as long as 2.5 to three hours
and it may even be assumed a slight delayed absorption. After that time, intense release
initiated. With respect to the sorption kinetics as measured by the tea-bag method, release
from SAP 2 has to be directly driven by priming forces arising in the progress of cement
hydration. Compared to the paste featuring W/C = 0.25 (Fig. 2), SAP 2 kept its initial level
of humidity for a clearly longer time in the present paste. This finding indicates that the
sucking forces arise much earlier in the paste with the lower W/C.
On the long term, SAP 2 continued its desorption as long as the end of neutron radiography imaging while SAP 1 apparently levelled off after approximately 15 hours. Obviously, no further extractable water was left behind in SAP 1 at this instance whereas SAP 2
further on provided water to the hydrating matrix at a significant pace.
Link of the findings from neutron radiography imaging to cement hydration
by the example of the pastes made of cement and silica fume (W/C = 0.42)
The progress of cement hydration was characterized by isothermal calorimetry as well
(Figs. 6 and 7). The dormant period finalized by approximately 2.5 to three hours after
paste preparation, no matter whether SAP were present or not. Interestingly, no pronounced
effect on the main peak of hydration was observed. These findings are in contrast to a delay
in time and a decrease in intensity as published by Justs et al.23 However, their system
consisted of white portland cement and the W/C was varied between 0.20 and 0.30 whereas
the present pastes featured W/C = 0.38 and 0.42, normal portland cement and silica fume.
Because no chemical information about the polymer sample was provided in23 and both
W/C and the mineralogy of the solid components differed, the apparent discrepancy of the
experimental results cannot be elucidated mechanistically in this place.
Conclusively, SAP 1 released its stored water starting amidst the dormant period and
hence provided water at a stage when it cannot be used up by the paste in a chemically reactive way. On the other hand, SAP 2 did not release water during the first three hours after
mixing to a significant extent (Fig. 5). In other words, SAP 2 did not spill its uptaken

Fig. 7 Isothermal calorimetry of the specimens with W/C

= 0.42, containing 10 wt-% silica fume bwoc, released heat
water during the dormant period. By the start of the acceleration period, SAP 2 began
providing water to the matrix. During the consecutive hours, SAP 2 accomplished a most
intense supply of water that continued at the slowing down of the acceleration period and
even beyond the turning point to the deceleration period. Hence, SAP 2 delivered its water
coevally with the most intense requirement whereas SAP 1 released it prematurely and
came to rest quite too early.
The equivalent mechanism of liquid extraction from SAP 2 applies in the cement paste
with the W/C of 0.25.
Desorption of water from two superabsorbent polymers (SAP) into cement-based pastes
was characterized by neutron radiography imaging to promote the understanding of the
mechanisms behind internal curing of concrete. The hydrogels differed in their chemical composition and structure and hence, their inherent sorption kinetics with respect
to extracted cement pore solution. SAP 1 was a free releasing polymer according to the
so-called tea-bag test, which desorbed quasi all of its initially uptaken liquid by three to
five hours, whereas SAP 2 was retentive for as long as the period of measurement of 24
hours.
Quantification of the referenced images recorded over time revealed that desorption
from SAP 1 initiated very early after paste preparation in any of the pastes. This way,
the results from the tea-bag test were confirmed. The release process from this kind of
hydrogel is governed by its individual non-retentiveness with respect to the saline cement
pore solution.
Contrarily, SAP 2 was found to release water into all the matrices in spite of its retentiveness with respect to the extracted, filtered cement pore solution. In the pastes which require
internal curing in principle (W/C = 0.25; W/C = 0.42 with silica fume) SAP 2 was expected
to release its intaken liquid in the course of time because macroscopic measurements to
assess internal curing had previously shown a high efficiency in mitigation of autogenous
shrinkage. In fact, SAP 2 kept its internal water content constant for approximately three
382SP-302-28
hours but afterwards it desorbed intensely. As was characterized by isothermal calorimetry

the onset of desorption from SAP 2 was coeval to the start of the acceleration period of
cement hydration. Hence, release from SAP 2 matched the period of most intense requirement of water. It may be claimed that the water is actively sucked out of the hydrogel
particles by chemical and physical forces, which arise as a consequence of the structural
build-up of a rigid skeleton. Interestingly, as well in the cement paste prepared with the
high W/C of 0.50 water was released from the SAP 2 particles. This W/C is clearly too high
for the demand of classical internal curing and hence, reasoning of this experimental result
has to be postponed at the time being.
It is worth noting that in any specimen the relative content of water in the SAP-free
districts of matrix increased. This indicates that the water released from the hydrogel particles redistributes into the matrix and will there be available for internal curing.
Regarding the application of SAPs as a chemical admixture in concrete technology
the findings of the present paper explain their impact on the rheological properties in
fresh cement-based pastes10 as well as the distinct efficiencies in mitigating autogenous
shrinkage.7-9 The rheological characteristics of cement-based mortars were elucidated for
up to 90 minutes after mixing.10 Plastic viscosity as well as yield stress developments in
time had indicated that SAP 1 handed over water, which was absorbed during mixing, back
to the mortar during this period. Neutron radiography imaging now disclosed that SAP
1 indeed desorbed a considerable amount of intaken water during this time, even in the
resting paste without agitation by a mixer or a rheometric test equipment. Contrarily, three
size fractions of SAP 2 used in10 were interpreted as to keep their stored water for that time
or to potentially even absorb some further water after the end of mixing. The present study
provided the proof of these statements as derived from rheometry. SAP 2 kept the amount
of intaken water constant after paste preparation until the onset of the acceleration period
of cement hydration.
With respect to mitigation of autogenous shrinkage of a high-strength mortar,7-9 had
demonstrated that the retentive sample of SAP 2 was pronouncedly more efficient than
the incontinent SAP 1 as they had been classified by so-called tea-bag test. These
macroscopic results can now be interpreted based on neutron radiography imaging and
calorimetry. SAP 1 releases its water as early as during the dormant period, during which
no chemical or physical requirement with respect to internal curing arises. Later on, SAP 1
tends to slow down its desorption intensity during the acceleration period. Obviously, this
desorption behavior in time is not very appropriate for optimum mitigation of autogenous
shrinkage. On the other hand, SAP 2 dispensed water into the matrix in a focused way
during the acceleration period and even continued desorption after the turning point to the
deceleration period. In chemical as well as physical terms, internal curing water from SAP
2 is most beneficially used by the cement-based matrix during these stages. Hence, the
proper performance of SAP 2 in mitigating autogenous shrinkage of a high-strength mortar
is a direct result of the well-timed release of water.
AUTHOR BIOS
Christof Schroefl is head of the research group for morphological characterization and
analytics at the Institute of Construction Materials at the TU Dresden (Germany). He
earned his Dr. rer.nat. degree in chemistry at TU Mnchen (Germany) on PCE-based

superplasticizers for UHPC in 2010. His research topics include the working mechanisms of chemical admixtures in and durability issues of high-performance cement-based
materials.
Viktor Mechtcherine is Full Professor and head of the Institute of Construction Materials at the TU Dresden since 2006. His scientific interests include the development, characterization and modeling of high-performance cement-based construction materials.
Since 2007, he has chaired the RILEM TC 225-SAP Application of Superabsorbent
Polymers in Concrete Construction and has initiated the RILEM TC RSC Recommendations for Use of Superabsorbent Polymers in Concrete Construction in 2014.
ACKNOWLEDGMENTS
Part of this research project (several of the experiments performed at the Swiss spallation
neutron source SINQ, Paul Scherrer Institut, Villigen/Switzerland) has been supported by
the European Commission under the 7th Framework Programme through the Research
Infrastructures action of the Capacities Programme, Contract No: CP-CSA_INFRA2008-1.1.1 Number 226507-NMI3. The support of the present study by this funding is
gratefully acknowledged. The authors are obliged to Dr. Eberhard Lehmann, Dr. Anders
Kaestner, Jan Hovind and Peter Vontobel (PSI) for experimental support and for teaching
the fundamentals of image analyses and differential quantification. The authors gratefully
thank SNF Floerger (Andrzieux Cedex/France) for the generous supply with superabsorbent polymers and for the permission to publish the data in the paper at hand. The research
group of Prof. Dr. Johann Plank (TU Mnchen) is thanked for performing calorimetry.
REFERENCES
1. Mechtcherine, V., and Reinhardt, H. W., eds., 2012, Application of superabsorbent
polymers in concrete construction. State-of-the-art report of the RILEM TC 225-SAP,
Springer, Heidelberg (Germany), 164 pp.
2. Jensen, O. M., and Hansen, P. F., Water-entrained cement-based materials: I. Principles and theoretical background, Cement and Concrete Research, V. 31, No. 4, 2001, pp.
647-654. doi: 10.1016/S0008-8846(01)00463-X
3. Jensen, O. M., and Hansen, P. F., Water-entrained cement-based materials: II. Experimental observations, Cement and Concrete Research, V. 32, No. 6, 2002, pp. 973-978.
doi: 10.1016/S0008-8846(02)00737-8
4. Wyrzykowski, M.; Lura, P.; Pesavento, F.; and Gawin, D., Modeling of internal
curing in maturing mortar, Cement and Concrete Research, V. 41, No. 12, 2011, pp. 13491356. doi: 10.1016/j.cemconres.2011.04.013
5. Nestle, N.; Khn, A.; Friedemann, K.; Horch, C.; Stallmach, F.; and Herth, G., Water
balance and pore structure development in cementitious materials in internal curing with
modified superabsorbent polymers studied by NMR, Microporous and Mesoporous Materials, V. 125, No. 1-2, 2009, pp. 51-57. doi: 10.1016/j.micromeso.2009.02.024
6. Trtik, P.; Mnch, B.; Weiss, W. J.; Herth, G.; Kstner, A.; Lehmann, E.; and Lura, P.,
2010, Neutron tomography measurements of water release from superabsorbent polymers
in cement paste, in: W. Brameshuber (ed.), International RILEM conference on material
384SP-302-28
science, volume III: Additions improving properties of concrete, RILEM Proceedings PRO
77, RILEM Publications S.A.R.L., Bagneux (France), pp. 175-185
7. Schrfl, C.; Mechtcherine, V.; and Gorges, M., Relation between the molecular
structure and the efficiency of superabsorbent polymers (SAP) as concrete admixtures to
mitigate autogenous shrinkage, Cement and Concrete Research, V. 42, No. 6, 2012, pp.
865-873. doi: 10.1016/j.cemconres.2012.03.011
8. Schrfl, C., and Mechtcherine, V., 2012, Superabsorbent polymers to mitigate autogenous shrinkage of cement mortar, in: V. M. Malhotra (ed.), 10th International conference
on superplasticizers and other chemical admixtures in concrete (Prague, Czech Republic),
SP-288-18, ACI, Farmington Hills (MI/USA), pp. 265-278
9. Mechtcherine, V.; Schroefl, C.; and Gorges, M., 2013, Effectiveness of various superabsorbent polymers (SAP) in mitigating autogenous shrinkage of cement-based materials,
in: F.-J. Ulm, H. M. Jennings, and R. Pellenq (eds.), Mechanics and physics of creep,
shrinkage, and durability of concrete, ASCE, Reston (VA/USA), pp. 324-331
10. Mechtcherine, V.; Secrieru, E.; and Schrfl, C., Effect of superabsorbent polymers
(SAPs) on rheological properties of fresh cement-based mortars Development of yield
stress and plastic viscosity over time, Cement and Concrete Research, V. 67, 2015, pp.
52-65. doi: 10.1016/j.cemconres.2014.07.003
11. Lehmann, E.; Pleinert, H.; and Wiezel, L., Design of a neutron radiography facility
at the spallation source SINQ, Nuclear Instruments & Methods in Physics Research.
Section A, Accelerators, Spectrometers, Detectors and Associated Equipment, V. 377, No.
1, 1996, pp. 11-15. doi: 10.1016/0168-9002(96)00106-4
12. Lehmann, E. H.; Vontobel, P.; and Wiezel, L., Properties of the radiography facility
NEUTRA at SINQ and its potential for use as European reference facility, Nondestructive
Testing and Evaluation, V. 16, No. 2-6, 2001, pp. 191-202. doi: 10.1080/10589750108953075
13. Zhang, P.; Wittmann, F. H.; Zhao, T.; and Lehmann, E. H., Neutron imaging of
water penetration into cracked steel reinforced concrete, Physica B, Condensed Matter,
V. 405, No. 7, 2010, pp. 1866-1871. doi: 10.1016/j.physb.2010.01.065
14. Lieboldt, M., and Mechtcherine, V., Capillary transport of water through textile-reinforced concrete applied in repairing and/or strengthening cracked RC structures, Cement
and Concrete Research, V. 52, 2013, pp. 53-62. doi: 10.1016/j.cemconres.2013.05.012
15. DIN EN 197-1, 11/2011, Cement Part 1: Composition, specifications and conformity criteria for common cements; German version
16. Rasband, W. S., 1997-2012, ImageJ, http://imagej.nih.gov/ij downloaded 2013,
incl. plugins for image referencing developed and provided by members of Paul Scherrer
Institut, Villigen/Aargau, Switzerland
17. Kardjilov, N.; de Beer, F.; Hassanein, R.; Lehmann, E.; and Vontobel, P., Scattering
corrections in neutron radiography using point scattered functions, Nuclear Instruments &
Methods in Physics Research. Section A, Accelerators, Spectrometers, Detectors and Associated Equipment, V. 542, No. 1-3, 2005, pp. 336-341. doi: 10.1016/j.nima.2005.01.159
18. Hassanein, R.; Lehmann, E.; and Vontobel, P., Methods of scattering corrections for
quantitative neutron radiography, Nuclear Instruments & Methods in Physics Research.
Section A, Accelerators, Spectrometers, Detectors and Associated Equipment, V. 542, No.
1-3, 2005, pp. 353-360. doi: 10.1016/j.nima.2005.01.161

19. Lothenbach, B., and Winnefeld, F., Thermodynamic modelling of the hydration of
Portland cement, Cement and Concrete Research, V. 36, No. 2, 2006, pp. 209-226. doi:
10.1016/j.cemconres.2005.03.001
20. Lothenbach, B.; Winnefeld, F.; Alder, C.; Wieland, E.; and Lunk, P., Effect of
temperature on the pore solution, microstructure and hydration products of Portland cement
paste, Cement and Concrete Research, V. 37, No. 4, 2007, pp. 483-491. doi: 10.1016/j.
21. Richter, A.; Paschew, G.; Klatt, S.; Lienig, J.; Arndt, K.-F.; and Adler, H.-J. P.,
Review on hydrogel-based pH sensors and microsensors, Sensors (Basel, Switzerland),
V. 8, No. 1, 2008, pp. 561-581. doi: 10.3390/s8010561
22. Slowik, V.; Schmidt, M.; and Fritzsch, R., Capillary pressure in fresh cement-based
materials and identification of the air entry value, Cement and Concrete Composites, V.
23. Justs, J.; Wyrzykowski, M.; Winnefeld, F.; Bajare, D.; and Lura, P., Influence of
superabsorbent polymers on hydration of cement pastes with low water-to-binder ratios,
Journal of Thermal Analysis and Calorimetry, V. 115, No. 1, 2014, pp. 425-432. doi:
10.1007/s10973-013-3359-x
386SP-302-28
SP-302-29
Study on the Rheology of Fly Ash Versus

Calcined Marl Blended Cements with
Polycarboxylate-Based Superplasticizers
by Serina Ng and Harald Justnes
The dispersing effectiveness of three polycarboxylate-based superplasticizers (PCE) was
investigated in two blended cement systems containing entirely different SCMs; fly ash
(FA) and calcined marl (CM) at replacement percentages of 20% and 60%. The methods
of investigation employed include rheological studies, hydration profiling up to 24h, and
packing density analysis. Generally, replacing clinker phases by FA decreased the dynamic
yield stress and delayed hydration of the pastes due to increased PCE to clinker ratios,
regardless of PCE type. Little variation except for cement with 60% FA replacement (FA60)
was observed on the Bingham viscosity. On the other hand, CM competed with clinkers
not only for water, but also for PCEs even in CM20, reducing the fluidity of the paste but
maintaining a similar initial rate of hydration of the pastes. PCE possessing intermediate
side chain lengths proved to be more effective for CM systems than PCEs possessing long
side chains.
Keywords: plasticizers; PCE; fly ash; calcined marl; blended cements.
INTRODUCTION
Green building and construction is the current trend in the building sector. Concrete,
largely composed of cement (OPC), is one of the major materials employed. Cement had
an annual production of about 3.5 billion tons in 2011.1,2 During its production, one ton of
OPC clinker emits an equivalence of about one ton of CO2 when pulverized coal is used
as a fuel, contributing to ~6% CO2 emission annually worldwide. This makes the cement
industry the third largest CO2 emission source after housing and transport.3,4 Therefore,
there is still a strong desire to reduce this CO2 emission from cement production. A most
direct method to aid in this process is to find suitable substitutions for cement clinker.
Greener and more environmentally friendly binders are thus sought after, and employing
supplementary cementitious materials (SCMs) as replacement for clinker phases in blended
cements is one of the main approaches.
Incorporation of SCMs into cements often can result in a change in the properties of the
blended cements, and thus altering the compatibility with other additions. One such chal387
388SP-302-29
lenge is in the application of superplasticizers. Superplasticizers are commonly employed

to modify the rheology of cement pastes to improve handling on site and/or increase and
enable enhanced final properties and aesthetic of the composite or buildings. These polymers are commonly designed to work with pure OPC systems, where rapid adsorption onto
the clinker, particularly C3A surfaces aids in the dispersion of the cementitious systems.5
In presence of SCMs, which can possess different chemical compositions and thus surface
chemistry, etc, the dispersing ability of these polymers may be altered leading to different
rheological behaviour arising from deviations in the interactions between these admixtures
and binder particles. The fundamentals of the colloidal interface of the polymer-inorganic
materials have been assessed in both isolated SCM systems6,7 and selected blended cement
systems.8 Despite these extensive studies, an actual understanding of the applied rheological effectiveness of these superplasticizers is required to ensure optimal usage of the
admixtures in such system. Additionally, the influence of these admixtures on blended
cements, possessing higher replacement amounts of SCMs, is expected to create a greener
construction industry.
For this purpose, three different commercial polycarboxylate ether plasticizers (PCEs)
were selected to test for their dispersing effectiveness with two different blended cements
based on fly ash and calcined marl respectively. Fly ash was selected due to its prevalent utilization in commercial products, whereas calcined marl based on its abundance.
The three different PCEs contain polymers with different PEO side chain lengths. Two
cement replacement levels were selected for this investigation: 20% and 60% replacement
by weight. Rheological tests were performed employing a parallel plate rheometer, while
the packing density was measured by the centrifugal consolidation method.9 Finally, the
heat of hydration of the cement slurries was investigated by isothermal calorimetry; and a
correlation between the heat evolution profile and the rheological properties of the cement
pastes was attempted.
With the need for cheaper, greener and better alternatives to OPCs, high replacement
levels by SCMs are predicted to be an important approach for the cement industry. This
can lead to compatibility issues with other materials, such as superplasticizers, particularly
PCEs, which have been commonly employed in the concrete industry construction. The
authors believe that this study dealing with workability of superplasticizers with blended
cements having high loading of SCMs, will prove to be highly useful.
EXPERIMENTS AND METHODS
Materials
An ordinary Portland cement (OPC), a class F fly ash (FA) and a calcined marl (CM)
calcined at 850 C were employed in this investigation. The OPC and FA were supplied by
Norcem A.S. Brevik, Norway, while the CM was from Saint-Gobain Weber, Oslo, Norway.
Table 1 displays the chemical compositions of the OPC, FA and CM respectively. The
calcined marl contains mainly smectite (>50%) and calcite (~25%). Minor amounts of
kaolinite (~8%), quartz (~4%), siderite (~3%) and pyrite (~1%) are present. The specific
Blaine surface areas of the OPC and FA were 382 and 357 m2/kg respectively. Whereas
for CM, the specific surface area as measured by BET was 15.1 m2/g. Blaine and BET
Study on the Rheology of Fly Ash Versus Calcined Marl Blended Cements
with Polycarboxylate-Based Superplasticizers 389
Table 1Chemical compositions of OPC, FA and CM
SiO2
Al2O3
Fe2O3
CaO
MgO
P 2O 5
K 2O
Na2O
SO3
Alkali
Total
OPC
20.8
4.6
3.5
61.6
2.4
0.2
1.0
0.5
3.5
1.1
99.2
FA
50.0
23.9
6.0
6.3
2.1
1.1
1.4
0.6
0.4
1.6
93.4
CM
49.6
18.1
10.6
14.1
2.9
0.2
2.4
0.7
98.6
values cannot be compared directly as Blaine is related to the outer surfaces of particles
(i.e. voids between them), whereas BET measures both outer and inner surface of nonconnected pores and cracks.
Three commercial polycarboxylate based superplasticizers (PCEs) were investigated.
These PCEs were supplied by Mapei AS, Sagstua, Norway and were denoted as PC1,
PC2 and PC3 in this investigation. PC1 and PC2 are both based on polymethacrylate type
polycarboxylates. Commercially, PC1 is commonly employed in ready-mix systems and
is comprised of a PCE with long side chains with low charge density. PC2 on the other
hand, possesses intermediate workability and plasticizing effect, and is comprised of a
polymethacrylate-based PCE with short side chains and high charge density. PC3 is made
up of a combination of one polymethacrylate based PCE and a polyacrylate based PCE in
the ratio of 1:9. The polymethacrylate based PCE here possesses long side chains and low
charge density and is similar to that in PC1. The polyacrylate based PCE possesses very
long side chains and very low charge density. The main differences between the polyacrylate based PCE and polymethacrylate based PCE are in their backbone structure, which
can be explained by the fact that for the grafting of longer side chains onto PCE, a polyacrylate backbone is normally required. Therefore, the actual main difference between these
two polymers was in the length of the side chains. PC3 is commonly used in the precast/
prestress industry and generally generates pastes with the shortest workability.
All materials were utilised as received. For preparation of the blended cements, the OPC
and FA/CM were manually mixed in the ratio of 4:1 and 2:3 to produce blended cements
with CM contents of 20 and 60%, respectively. These blended cements were denoted as
FA20, FA60, CM20 and CM60, respectively. Two different dry polymer dosages were
employed; 0.2 and 0.4% dry polymer bwob (by weight of binder), respectively.
Experimental procedures
All cement pastes were prepared at a w/b of 0.36 to amplify the performance of the
PCEs. When PCEs were added, the required dosages were homogenized in the water
before adding to the dry powder mix over 30s. The mixture was blended under high shear
for 1min utilizing a high shear mixer (Philips, 600W, capacity = 200mL), let stand for 5min
and a final high shear mixing of 1min to avoid false setting. The high shear mixing was to
390SP-302-29
mimic the high shear energy in a concrete partly imposed by coarse aggregate.10 In each
mix, the amount of pastes prepared was ~205g to minimise weighing errors, and they were
employed for the rheological, calorimetric and packing density analyses.
Rheological measurements About 2ml of the prepared pastes were smeasured within
the first 12 min after first contact of dry powder with water. A Physica MCR 300 rheometer
(Anton Paar, Graz/Austria) equipped with parallel plate geometry was utilised. The upper
(rotor) and lower (stator) plates were serrated to a depth of 150m, the gap distance was
1mm and measurements were conducted at 20C. Before starting the measurements, the
paste was sheared at 100s-1 for 1min. The up and down flow curves was measured over the
range of shear rates from 2 to 150s-1 over a period of 6min.
For analysis, the down flow curve was fitted with a linear regression at the high shear
rates (threshold limits of >50s-1). The Bingham viscosity (2) can be obtained from the
gradient of the linear regression, while the dynamic yield point (d) of the cement paste was
derived from the ordinate intercept of the linear regression line. All measurements were set
to end and jump to next test segment when dQ/dt>10. The dynamic yield points present an
indication of the amount of force needed to start shearing the cement slurry.
Calorimetric investigations About 8 g of the prepared pastes were weighed accurately
into a glass vial, sealed with a lid and placed in an isothermal TAM Air calorimeter (TA
Instrument, New Castle/USA). Measurements were performed up to 24 h from the point of
first contact between dry powder and water against a calibrated reference of inert alumina
powder of similar mass. The time of placement was recorded and all subsequent hydration profiles were calculated and tabulated after 1 h to exclude any excessive heat transfer
arising from initial preparation.
Packing densities The packing densities of the cement pastes were determined by a
modification of the centrifugal consolidation method, proposed by Miller.11 About 80 g
of the paste was added into a 50 mL falcon tube and subjected to centrifugation at 4,000
rpm for 5 min. The supernatant was carefully removed with a pipette and the weights of
the cement samples (before and after compaction/removal of liquid) were determined. The
packing density of the cement slurry was calculated at this applied compaction energy as
follows:

Packing density = Volbinder / Vol(binder+water) after centrifugation
(1)
where the volume of binder and water were calculated from the mass remaining in the
residue after centrifugation utilizing the densities of cement (3.15kg/m3), fly ash (2.35kg/
m3), calcined marl (2.65kg/m3) and water (1.0kg/m3) respectively. For this calculation, it
was assumed that only water was extracted during the centrifugal process.
Rheological values quantify the fluidity of the cement pastes, which are very significant
when the performance of PCE superplasticizers is needed. In particular, the ability of these
polymers to deflocculate and disperse particles in solution or to reduce the water uptake
for hydration, etc can be given a value. Coupled with calorimetry and packing density
measurements, the driving force behind the performance of the PCE superplasticizers can
be derived. The results here will be divided into four sections. First, the properties of neat
Fig. 1(a) Viscosity versus dynamic yield stress of OPC, FA20, FA60, FA100 and CM20;
(b) Cumulative heat evolved per binder mass of pastes containing 0 to 60% replacement of
FA/CM at 3h, 6h, 12h and 24h (w/b = 0.36).
cement pastes, both OPC and blended will be discussed. Thereafter, the interaction of OPC
with PCE will be highlighted, presenting the normal effective behaviour of PCEs. In the
third section, the properties of pastes with low replacement levels of SCMs in presence of
PCEs will be clarified. Finally, the properties of pastes possessing high replacement level
of SCMs (60%) with added PCEs will be discussed.
Properties of neat OPC, FA and CM blended cements
The properties of the neat OPC, FA20, FA60, CM20 and CM60 were first investigated
(Fig. 1). At a w/b 0.36, the OPC paste exhibited a Bingham viscosity of 0.28 Pas and
dynamic yield stress of ~230 Pa. With increasing FA replacement, a proportional decrease
(R2 = 0.9317) in the rheological values of FA blended cements was observed. Replacement
of clinker by FA appeared to decrease the dynamic yield stress greater than viscosity. This
signifies that increasing FA content improves the flow of neat pastes, driven greatly by the
reduction in dynamic yield stress of the pastes, possibly due to the spherical nature of the
FA particles. On the other hand, when neat CM-cement pastes were tested, only CM20
was sufficiently fluid for initial rheological measurements (2 = 0.49 Pas, d = 370 Pa)
reflecting the high consumption (i.e. by adsorption and/or absorption) of water by CM,
which can impair the early age hydration of OPC.
The cumulative heat evolved from the hydration of OPC increased gradually up to 6h,
which accelerated after that to attain a total heat of 12.1 J/g by 24h of hydration (Fig.
1b). When SCMs were introduced, a decrease in the cumulative heat of hydration with
increasing SCM replacement was observed, regardless of the type of SCM employed. This
observation indicated that the amount of OPC present dominated the hydration heat of the
pastes.
Two additional observations could be made. Firstly, regardless of replacement contents,
the decrease in heat evolved relative to the neat OPC paste was larger for FA blended
cements than the CM blended ones. Secondly, especially when analysing the heat evolved
within the first 6h, it can be observed that FA caused a greater delay in early age hydration
than CM right from the early age (% heat evolved relative to OPC by 6h: FA60 = 11%,
392SP-302-29
Fig. 2Packing densities of blended cements as

a function of SCM loading.
CM60 = 58%), indicating that these two SCMs contributed very differently to the hydration profiles of the pastes. When coupled with the cumulative heat evolved by 24h, it can
be observed that FA acted mainly through filler effect, whereby hydration of FA20 and
FA60 gave rise to 84% and 40% of the total heat evolved by a neat OPC (12.1 J/g). In the
case of CM, hydration of CM20 and CM60 displayed relative heat of 90% and 49% respectively, indicating that CM contributed to the overall hydration of the paste. The decrease in
proportional amount of heat evolved from CM20 to CM60 can be attributed to the decrease
in OPC content and also the availability of water for hydration as a result of rapid initial
consumption by CM.
Then, the packing densities of the cement pastes were investigated. The packing density
of the neat OPC was 0.496 at the compaction energy employed. When OPC was replaced
by the SCMs, the packing densities increased as a function of SCM replacement (Fig. 2).
It could be deduced at this point that similar mechanisms were at play for both SCMs.
However, it is important to take into account that the densities of both FA and CM were
lower than that of OPC (2.35kg/m3, 2.65kg/m3 and 3.15kg/m3 respectively), thus the variation in their packing densities were based on a sum of different factors. For FA systems,
FA100 displayed a high packing density of 0.580 due to its low hydrating ability. Therefore, much of the water was released in the supernatant during centrifugation. The packing
densities of FA20 and FA60 thus layed between that of FA100 and OPC. In the case of
CM, the increase in packing density can be accounted for by the decrease in density of the
binder when OPC was replaced with CM. In terms of affinity for water, CM consumed
water rapidly and the saturation point was not reached at a w/CM of 0.36.
Here, it is of interest to note that FA and CM are two very different SCMs, whereby FA
improved the fluidity of the pastes while CM reduced the flow. In the next section, the
impact of increasing SCM contents, comparing behavior of FA/CM on the PCEs, will be
presented accordingly.
Properties of OPC (FA0/CM0) with added PCEs
First, the pastes without any SCMs will be discussed. In presence of 0.2%bwob of
PCEs, the viscosity of the pastes remained in the same order as that of neat OPC, whereas
Fig. 3(a) Viscosity versus dynamic yield stress of OPC, FA20 and FA60 with added
0.2%bwob PCEs; (b) Viscosity versus dynamic yield stress of OPC, CM20 and CM60
with added 0.2%bwob PC, w/b = 0.36. For CM60, 0.4%bwob of PCEs were investigated
as only 0.2%bwob of PC1 was effective. Rheology of pastes is represented by unshaded
markers corresponding to that of the neat pastes, i.e. neat OPC is shaded circle, whereas
OPC + PCE are displayed as unshaded circles.
Fig. 4(a) Cumulative heat evolved from the hydration of OPC pastes in absence and presence of 0.2%bwob of varying PCEs as a function of time (w/b = 0.36); (b) Packing densities of OPC pastes in the absence of presence of 0.2%bwob of varying PCEs.
dynamic yield stress of all pastes decreased by at least a factor of 2.3 (PC2, Figure 3). With
PC1 and PC3, the decrease was much greater with factors of >10, indicating that these
PCEs dispersed the OPC effectively by reducing the dynamic yield stress (i.e. preventing
flocculation or gel strength build-up). The variation in rheological impact can be attributed
to the length of side chains (SC) present in the PCE, whereby PC3 with the longest side
chain dispersed the paste best, whereas PC2 was the least effective as a result of the short
side pendants present in this PCE.
The impact of PCEs on the cumulative heat of hydration of OPC pastes can be observed
from Fig. 4. During early age of hydration (up to 3h), a delay in hydration was observed
for all pastes containing PCEs, potentially due to decreased dissolution and hydration of
the clinker phases as PCEs adsorb topochemically onto the reactive C3A and alite phases.
394SP-302-29
Table 2Packing densities of OPC, FA pastes with PCEs after consolidating

centrifugation
OPC
FA20
CM20
FA60
CM60*
None
PC1
PC2
NRGPC3
0.496
0.514
0.502
0.548
0.510
0.480
0.497
0.492
0.632
0.507
0.485
0.501
0.493
0.561
0.509
0.481
0.504
0.493
0.578
0.508
* 0.4%bwob of PCEs were investigated too. Results showed similar packing density as when 0.2%bwob of superplasticizers
were employed.
PC3 retarded the least as it possessed the lowest coverage on the clinkers phases as a result
of its low effective molecular coverage per clinker surface due to its high Mw (long side
chains). Further hydration of the pastes showed deviations in the retarding effects of PCEs
on OPC hydrations. By 12h of hydration, PC3 displayed a synergetic effect on the hydration of OPC paste, showing an 18% increment in cumulative heat evolved as compared to
the neat OPC paste. This indicates that when PC3 is employed, a higher 1 day strength than
that of neat OPC paste is to be expected. By 24h, all PCEs filled pastes possessed higher
cumulative hydration heat relative to the neat OPC paste despite the prolongation of the
induction period due to dispersion of cement agglomerates which resulted in more free
surfaces for subsequent hydration.
The packing density of the neat OPC was 0.496 at the compaction energy employed
(Fig. 2). In the presence of PCEs, a decrease in packing densities was observed (Table 2).
This resulted from an increase in the retention of water by the polymers which in effect
contributed to the improved rheological properties of the cement pastes. In general, the
packing density was the lowest when PC2 was added, followed by PC3 and PC1 respectively. No direct correlation to the side chain lengths could be drawn here.
Properties of FA20/CM20 pastes with added 0.2% bwob of PCEs
The impacts of PCEs on the rheological, heat evolution and packing densities of pastes
with 20% SCM replacements were next investigated. 20% SCM replacement was selected
as it represents the average normal amount of SCMs present in commercial cements
currently available on the market.12 As observed in Fig. 3, addition of PCEs to the 20%
SCM replaced pastes resulted in the decrease of both Bingham viscosities and dynamic
yield stresses for all pastes. Similar to the OPC system, the dynamic yield stress of the
pastes was more affected than the Bingham viscosity.
More specifically, for the FA20 pastes, the trend in effectiveness of the PCEs was PC2
> PC1 > PC3, similar to that observed for OPC pastes. Additionally, the relative improvements in rheologies of the pastes were greater for FA20 than OPC, comparing the neat
pastes to pastes with added PCEs. These findings confirmed that addition of FA to OPC not
only diluted the system (higher effective OPC/PCE ratio), but also FA appeared to have no
significant impact on the performance of PCEs at this level of replacement. In other words,
independent of the nature of the PCEs, these polymers interacted and dispersed mainly
with the OPC present in the system.
Fig. 5Cumulative heat evolved from the hydration of (a) FA20 and (b) CM20 pastes in
absence and presence of 0.2%bwob of varying PCEs as a function of time (w/b = 0.36).
In the CM20 pastes, despite the improvement in rheological behaviours, the trend in
performance of PCEs differed from that of OPC, whereby here, PC1 performed the best,
followed by PC3 and finally PC2. This indicates that other than mere water consumption,
CM may interfere with the performance of PCEs, particularly in the case of PC3.
For the cumulative heat of hydration, a similar trend was observed here for FA20
pastes as compared to that in rheological measurements. Figure 5 presents the cumulative heat evolved from the hydration of FA20 and CM20 pastes in absence and presence
of 0.2%bwob of varying PCEs as a function of time (w/b = 0.36). The retarding effect of
PCEs on the hydration of FA20 was as follow: PC1 > PC2 > PC3, similar to OPC pastes.
The variation in the degree of hydration for FA20 with and without PCEs was similar to
that of OPC, confirming that PCEs mainly affect the hydration of OPC. In the case of CM
replacement, PCEs addition displayed less influence on the overall heat of hydration of the
paste, particularly after 24h. The apparent little impact of PCEs on the hydration of CM20,
coupled with the significant influence on the rheological behaviour of the paste, indicates
that PCEs may be prevented from excessive surface interaction with OPC for dispersion
by other mechanisms than surface adsorption by the CM. This could also be accounted for
by the unavailability of effective plasticizers to adsorb onto the silicate phases of the OPC,
rendering little or no retardation in the hydration of this phase. Additional reaction between
OPC clinker and CM may prevent the inhibition of hydration by the PCEs. Further elaborations will be discussed in high SCM loading system (CM60).
The trend in the packing densities of the FA20 pastes was similar to that for OPC pastes
as a result of the apparent inertness of the FA particles. The packing densities of CM20 with
added PCE, on the other hand, showed clear deviation from that in OPC systems. Instead
of measuring varying packing density due to different PCEs addition, all three CM20 with
different PCEs registered a consistent packing density. This indicated that the PCEs were
not available to interact with water molecules, thus confirming that they are not surface
bound on the CM particles.
396SP-302-29
Properties of 60% FA/CM blended cements with added PCEs

Next, the impact of high SCM replacement of up to 60% was investigated (Fig. 3). As
observed before, FA60 was highly fluid (d = 80 Pa versus 230 Pa for neat OPC pastes),
whereas no rheological values were measured for CM60 as it was no longer flowing. When
0.2%bwob of PCEs were added to these blended cements, drastic reduction in Bingham
viscosity and dynamic yield stress were observed as expected for FA60, whereby the latter
falls to almost 0Pa regardless of PCE employed. In the presence of 0.2%bwob of PC3,
sedimentation of FA60 occurred, significantly over dispersing the binder system since the
effective PCE/OPC ratio was increased due to insignificant FA interaction.
In the case of CM60 pastes, addition of 0.2%bwob of PCE showed little flow, except
when PC1 was added. This implied two things. Firstly, CM possessed a strong affinity for
water and PCEs. Secondly, PC1 which possess intermediate side chain appeared to be the
best dispersing PCEs for CM systems as CM appear to have a lower affintiy (thus increased
dispersing power). No possible explanation could be given at this moment in time for the
low Bingham viscosity of the CM60 paste with only 0.2%bwob of PC1.
For closer analysis, 0.4%bwob of PCEs were dosed to the CM60 paste. Immediately,
all CM60 pastes displayed fluidity, whereby Bingham viscosity was the greatest for
PC1 > PC2 > PC3 (0.52, 0.47 and 0.43 Pas) respectively. The dynamic yield stress was
highest for PC3, followed by PC2 and finally PC1 (300, 200 and 150 Pa), confirming that
PCEs behave very differently in CM blended cements than in OPC. A likely explanation for this behavior could be attributed to the method PCEs being consumed by CM,
as discussed earlier on in this paper. For the CM employed in this investigation, some
uncalcined portions may remain (or partly calcined retaining layered structures) and these
can be susceptible to absorb PCEs via their hydrophilic PEO side chains.13,14 In this way,
CM has a higher capacity for PC3 which possessed very long side chains, thus excess PC3
(0.4%bwob) was needed to exert their dispersing effectiveness. On the other hand, PC1
could function better than PC3 at 0.2%bwob, but were also consumed by CM at higher
CM replacement, normalising the performance of PC1 and PC3 to be similar to that in the
OPC system. In case of PC2, the performance of this PCE was less affected as it possessed
shorter side chains that were less vulnerable to intercalation in CM layers. However, due to
the inherent dispersing power of PC2, it remained to be inferior to the other two PCEs in
initial dispersion regardless of dosage.
Further analysis on the cumulative heat of hydration revealed that at high replacement
level of FA (60%), the FA60 pastes followed a similar trend as the OPC or FA20 pastes up
to 6h hydration. Therefore, due to the high PCE/OPC ratio, the excessive amount of PCE
blocked any active sites on the OPC particles, particularly alite phases, thus prolongating
the induction period, and probably greatly reducing the early strength of concretes during
the first 24 h. PC2 appeared to affect the hydration least, potentially due to the amount of
free surfaces it can generate relative to the other 2 PCEs as discussed before.
In the case of CM (60%), the cumulative heat of hydration of CM60 was comparable
with and without PCE addition, contrary to that observed in OPC or FA blended cement
systems (Fig. 6). This can be attributed to the decreased effectiveness of PCE to delay
hydration as they are consumed within the CM layers, while at the same time, the hydration
of CM blended cement underwent a different mechanism route, which was dominated by
the hydration of CM (Fig. 7). It appeared that a threshold amount of CM must be present in
Fig. 6Cumulative heat evolved from the hydration of (a) FA60 and (b) CM60 pastes in
absence and presence of 0.2%bwob of varying PCEs as a function of time (w/b = 0.36).
Fig. 7Rate of heat of hydration per mass

of binder of CM20 and CM60 with 0.2 and
0.4%bwob of PC2, w/b = 0.36 (Solid lines
represent 0.4%bwob, whereas dotted line represent 0.2%bwob of PC2 respectively).
the system before alternation of the mechanism occur. Nonetheless, the early heat of hydration (and thereby strength of concretes prepared with CM in presence of PCE) was thus not
affected by the presence of these polymers. This finding is in line with that observed in the
rheological measurements, where consumption of PCEs by CM caused a decrease in their
dispersing ability. When PCEs were consumed by the CM through possible intercalation
of the PEO side chains, the PCEs were no longer available to block the silicate surfaces on
OPC and thus hydration was not retarded.
The packing densities of blended cements with 60% SCM replacement can be observed
in Table 2. In present of PCEs, the packing density of FA60 tended towards the characteristics of a pure FA paste. The high packing density of the FA60 and 0.2%bwob PC1 (0.632)
can be accounted for by the high quantity of suspended FA particles in the supernatant,
giving a false high packing density. Therefore, a threshold balance between the amount
398SP-302-29
of FA and plasticizer must be present to ensure a homogeneous and non-bleeding system.

Packing densities of CM60, on the other hand, remained similar to that of CM20, whereby
all pastes possessed similar packing densities regardless of PCE type added. This remained
relatively constant even when the dosages of PCE were doubled to 0.4%bwob, showing the
high affinity of PCEs by CM.
CONCLUSIONS
The effect of PCE superplasticizers on the rheological and hydration properties of FA and
CM blended cements at 20% and 60% replacement was investigated and quantified. FA and
CM are two very different materials which affected the rheological and early age hydration
properties of the blended cements and the effectiveness of PCEs in totally different manner.
In general, FA improved the rheology of cements, with or without PCEs while CM, on the
contrary, decreased the flow of the cement pastes, and was particularly detrimental to the
performance of PCEs possessing long side chains.
FA exerted its presence in the blended cement through dilution effect. It was generally
inert to hydration and its affinity to PCE is strongly outweighed by that from OPC. Therefore, the main influencing factor from FA in blended cement, regardless of replacement
level was an increase in PCE/OPC ratio, allowing a reduction of PCE dosage needed in the
dispersion of FA blended cements.
On the other hand, CM was found to affect the rheological properties of the cement by
first possessing a high demand of water, and also through an active role in hydration reactions with OPC. Upon addition of PCEs, PC3 failed to perform due to its rapid consumption by CM through the intercalation of PEO side chains of these PCEs between the incompletely calcined layers of the CM, rendering them ineffective as plasticizers. Thus, PC1
which possess only PEO side chains with intermediate length performed better in this
investigation.
A close look of these two SCM materials revealed that a new ternary binder could be
possible, where OPC, FA and CM are added together to form a complementary system,
where the different components act to compensate for the performance deficiency of one
another, even in the presence of superplasticizers such as PCEs.
AUTHOR BIOS
Serina Ng is a Research Scientist at SINTEF Building and Infrastructure, Department of
Materials and structures, Concrete group, Trondheim, Norway, since January 2013. She
graduated with a PhD in construction chemistry at the Technical University of Munich,
Germany. Her field of research includes construction materials, blended cements, admixtures and additives, insulation; ranging from the understanding of their fundamental
properties to their applications.
Harald Justnes is Chief Scientist at SINTEF Building and Infrastructure, Department of Materials and structures, Concrete group, Trondheim, Norway, where he has
worked since 1985. His research covers the chemistry of cement, concrete, admixtures
and additives from production, through reactivity, to durability. He graduated with PhD
in inorganic chemistry at the Institute of Materials Technology, Inorganic Chemistry,
NTNU, Trondheim, Norway, and has been Adjunct Professor in Cement and Concrete
Chemistry at the same institute since 2000.
ACKNOWLEDGMENTS
The authors wish to express their gratitude to COIN consortium for the funding of this
project.
REFERENCES
1. World Cement Production, (2011). European Cement Association (2012).
2. Oss, H. G. v., Cement Annual Publication 2012, In: Mineral Commodity Summaries,
Cement Statistics and Information, USGS U.S. Geological Survey (2012).
3. Cement Industry Energy and CO2 Performance Getting the numbers right, World
Business Council for Sustainable Development, The cement sustainability initiative (2011).
4. Roadmap, C. T., 2009 Carbon emissions reduction up to 2050, World Business
Council for Sustainable Development (2009).
6. Lesti, M.; Ng, S.; and Plank, J., Ca2+ Ion mediated interaction between
microsilica and polycarboxylate comb polymers in model cement pore solution,
Journal of the American Ceramic Society, V. 93, No. 10, 2010, pp. 3493-3498. doi:
10.1111/j.1551-2916.2010.03901.x
7. Habbaba, A., and Plank, J., Surface Chemistry of Ground Granulated Blast Furnace
Slag in Cement Pore solution and Its impact on the effectiveness of polycarboxylate Superplasticizers, Journal of the American Ceramic Society, V. 95, No. 2, 2012, pp. 768-775.
doi: 10.1111/j.1551-2916.2011.04968.x
8. Alonso, M. M.; Palacios, M.; and Puertas, F., Compatibility between polycarboxylate-based admixtures and blended-cement pastes, Cement and Concrete Composites, V.
9. Miller, K. T.; Melant, R. M.; and Zukoski, C. F., Comparison of the compressive
yield response of aggregate suspensions: pressure filtration, centrifugation, and osmotic
consolidation, Journal of the American Ceramic Society, V. 79, No. 10, 1996, pp. 25452556. doi: 10.1111/j.1151-2916.1996.tb09014.x
10. Williams, D. A.; Saak, A. W.; and Jennings, H. M., The influence of mixing on the
rheology of fresh cement paste, Cement and Concrete Research, V. 29, No. 9, 1999, pp.
1491-1496. doi: 10.1016/S0008-8846(99)00124-6
11. Miller, K. T.; Melant, R. M.; and Zukoski, C. F., Comparison of the compressive
yield response of aggregate suspensions: pressure filtration, centrifugation, and osmotic
consolidation, Journal of the American Ceramic Society, V. 79, No. 10, 1996, pp. 25452556. doi: 10.1111/j.1151-2916.1996.tb09014.x
12. Thomas, M.; Optimizing the Use of Fly Ash in Concrete, Concrete, Portland
Cement Association, (2007) pp. 24 IS548. http://www.cement.org/docs/default-source/
fc_concrete_technology/is548-optimizing-the-use-of-fly-ash-concrete.pdf?sfvrsn=4.
400SP-302-29
13. Ng, S., and Plank, J., Interaction mechanisms between Na Montmorillonite clay and
14. Svensson, P. D., and Hansen, S., Intercalation of smectite with liquid ethylene glycol
resolved in time and space by synchrotron X-ray diffraction, Applied Clay Science, V.
48, No. 3, 2010, pp. 358-367. doi: 10.1016/j.clay.2010.01.006
SP-302-30
A Study on the Cement Compatibility of

PCE Superplasticizers
by A. Lange and J. Plank
It is well established among concrete producers that specific cements seem to be incompatible with most PCE products, thus causing excessive PCE dosages or even a total failure
of the PCE. This effect is commonly referred to as cement incompatibility of PCE. The
study here investigates the reasons for such incompatibility. First, it was found that only
cements which upon contact with water instantaneously form large amounts of ettringite
exhibit such incompatibility phenomenon. Their characteristics are elevated C3A content
(> 7 wt.-%) and high initial heat of hydration. Second, it was observed that PCEs strongly
influence early ettringite crystallization by acting as morphology modifying agent. Most
PCEs transform common micro meter-sized ettringite into nano-sized crystals which bring
about a huge surface area and thus require abnormal dosages of PCE to achieve dispersion. Such nano-sized particles can be separated from the cement paste by centrifugation where it appears as a viscous, gel-like top layer. From five chemically different PCE
polymers tested, one (a modified APEG type) was identified as extremely compatible with
all cement samples, whereas three other ones (two conventional MPEG and one APEG
type) exhibited pronounced incompatibility with C3A rich cements. An IPEG PCE showed
moderate cement compatibility. The phenomenon of cement incompatibility occurs only
when the PCE is present in the mixing water, and disappears when PCE is added in delayed
mode. Finally, a simple and quick test to identify cementPCE incompatibility is proposed.
Keywords: admixture; adsorption; cement dispersion; ettringite; high range water
reducer; morphology; nano-size; polycarboxylate.
INTRODUCTION
Polycarboxylate based superplasticizers (PCEs) are widely used in concrete manufacturing, especially for high performance concrete such as ultra-high strength concrete or
self-compacting concrete.1-3 PCEs were invented more than 30 years ago in Japan4 and
since then spread all over the world with highly increasing production rates, especially
in China. The first PCE products represented copolymers of sodium methacrylate and
methoxypolyethylene glycol methacrylate macromonomers (the so-called MPEG type),
but nowadays many other PCE products are on the market including allyl ether (APEG),
methallyl ether (HPEG) and isoprenyl ether (IPEG or TPEG) based PCEs.5
401
402SP-302-30
PCEs possess an anionic polymer trunk with polyglycol pendants. Their dispersing force
originates from the ability to adsorb on sites of the cement particles exhibiting positive
surface charge, i.e. mainly on the aluminate and ferrite phases (C3A and C4AF) and their
initial hydration products such as e.g. ettringite or monosulfo aluminate.6 Cement clinker
is composed of several mineral phases with different surface charges which causes strong
agglomeration.7 The dispersing effectiveness of PCEs first is owed to charge neutralization of the positively charged domains and second to the steric hindrance provided by the
polyglycol side chains.8 Without superplasticizer, a significant part of the mixing water is
entrapped between the cement agglomerates, thus causing a stiff consistency of the mortar
or concrete. PCE superplasticizers efficiently break down these agglomerates and free the
entrapped water, thus inducing high flowability of the mixture.
Compared to other superplasticizers including polycondensates, the key feature of PCEs
is their outstanding dispersing performance at low water-to-cement ratios. However, applicators sometimes have experienced an incompatibility phenomenon of PCE with certain
cements, whereby the admixture dosage became unreasonably high. In some cases it was
totally impossible to disperse such cements with PCEs at all.9 It is well known that clay
impurities can impede the performance of PCE superplasticizers,10 but even in clay-free
concretes, high dosages were observed when cements containing a high amount of C3A
were used. Early investigations indicated that such negative effect was owed to interaction of the PCE with early cement hydration products.11,12 Especially in ultra-high strength
concrete (UHSC) where the PCE commonly is added to the mixing water to reduce mixing
time, incompatibility between PCEs and cement was frequently observed13.
In this study, compatibility of several structurally different PCE superplasticizers with
nine specifically selected ordinary Portland cements (OPCs) was analyzed with respect to
the dispersing performance of the PCEs in cement paste, initial heat of hydration within
the very first minutes of cement hydration and via extraction of the initial hydration products from the cement pastes. Furthermore, the effect of point of addition time of PCE was
tested to clarify how fast these initial hydration products are formed in the cement paste.
The initial hydration products were identified using X-ray diffraction, elemental analysis
and thermogravimetry. From this data, interaction between the PCE superplasticizers and
the initial hydration products was elucidated and an explanation for the incompatibility
phenomenon was sought. Finally, the impact of PCE superplasticizers on the morphology
of ettringite synthesized from solution was studied to confirm the results obtained in
cement.
The incompatibility phenomenon of PCEs with certain cements is addressed in this study.
For optimum cost-effectiveness, PCE dosages should be as low as possible, but it is applicators experience that sometimes unusually high dosages are needed in some systems. This
phenomenon is analyzed with respect to the influence of PCEs on initial cement hydration
products such as ettringite. It was found that almost all PCEs act as morphology modifying
agent for ettringite, resulting in nano-sized ettringite particles which possess a huge surface
area and high PCE uptake capacity.
A Study on the Cement Compatibility of PCE Superplasticizers 403
PCE Samples A total of five PCE samples was studied. Two of them (samples
MPEG-7 and MPEG-25) were methacrylic acid-co--methoxy polyethylene glycol
(MPEG) methacrylate ester polymers with side chains holding 7 and 25 EO units respectively. The molar ratio of methacrylic acid and MPEG-MA ester was adjusted in such a way
that optimal dispersing performance (i.e. lowest dosage) was achieved. For MPEG-7, the
MA: MPEG-MA ratio was 1.2:1 while for MPEG-25 it was 3:1.
Furthermore, two PCE samples (denominated as APEG-34 and APEG-34AM) were allyl
ether (APEG) based polycarboxylates with a side chain holding 34 EO units. APEG-34AM
contained allyl maleate as additional monomer. The molar ratios between allyl ether,
maleic anhydride and allyl maleate were 1:1:0 (for APEG-34) or 1:1:1 (for APEG-34AM)
respectively.
The fifth PCE sample (IPEG-25) was a copolymer of acrylic acid and -hydroxy polyethylene glycol isoprenyl ether (IPEG) with 25 EO units in the side chain. Again, the molar
ratio was adjusted for maximum dispersing performance and was 2.7:1 (acrylic acid: IPEG
ether).
All polymers were self-synthesized. For preparation of the MPEG PCEs, two solutions
were prepared, one containing the MPEG-MA ester, methacrylic acid and mercaptopropionic acid as chain transfer agent (solution I, 50 wt.-% in DI water). The other solution
contained the radical initiator sodium persulfate (3 mol.-% relative to the monomers, 1
wt.-% in DI water, solution II). The reactor was equipped with stirrer, nitrogen inlet and
thermometer and charged with a small amount (~30 mL, 1 fl. oz.) of DI water. The vessel
was then heated to 80 C (176 F). Both solutions were fed continuously into the reactor,
solution I within 3 hours and solution II within 4 hours. After completion of the addition,
the mixture was stirred for one more hour, cooled and neutralized with 30 wt.-% NaOH
solution.
In preparation of the APEG-PCEs, all monomers were placed into a round bottom flask
equipped with stirrer, heated to 90 C (194 F) and polymerized in bulk. Benzoyl peroxide
was used as radical initiator. Initiator dosage was 4 mol.-% with respect to the monomers.
The radical initiator was added uniformly as powder over 1.5 hours. When the reaction
was finished, DI water was added to yield an aqueous solution with a solid content of ~50
wt.-%. After cooling to room temperature, the samples were neutralized with 30 wt.-%
NaOH solution.
The IPEG PCE was synthesized by placing the IPEG ether macromonomer as 50 wt.-%
solution into the reactor as above and heated to 80 C (176 F). Again, acrylic acid was fed
into the reactor within 3 hours and the initiator solution (same amount as for MPEG PCEs)
within 4 hours. The final procedure was identical to that for the MPEG PCEs.
The chemical structures of all synthesized PCE polymers are shown in Figure 1.
All polymer samples were used without further purification. The polymers were characterized by gel permeation chromatography using Waters 2695 separation module equipped
with 2414 RI detector (Waters) and a Dawn EOS 3 angle light scattering detector (Wyatt
Technology). A dn/dc of 0.135 mL/g was used to calculate molar masses relative to polyethylene oxide.14 The analytical results are shown in Table 1.
Cement Nine different cements were selected for this study. All of them were ordinary
Portland cements (CEM I) ranging from 32.5 N to 52.5 R. Additionally, an API Class G oil
404SP-302-30
Figure 1Chemical structures of the polycarboxylate samples synthesized for this study
Table 1Analytical properties of the synthesized PCE samples
Polymer
MPEG-7
MPEG-25
APEG-34
APEG-34AM
IPEG-25
Mw
[g/mol]
44,300
67,600
63,100
78,400
93,800
Mn
[g/mol]
23,300
28,200
22,500
24,500
36,100
PDI
[Mw/Mn]
1.9
2.4
2.8
3.2
2.6
nEO
[mol]
7
25
34
34
25
Table 2Oxide analysis of cement samples as determined via X-ray fluorescence analysis
Cement
sample
CEM I 32.5 N
CEM I 32.5 R
CEM I 42.5 N
CEM I 42.5 R
CEM I 42.5 R HS
CEM I 52.5 N
CEM I 52.5 R
CEM I 52.5 R HS
API Glass G
CaO
wt.-%
SiO2
wt.-%
Al2O3
wt.-%
Fe2O3
wt.-%
MgO
wt.-%
Na2O
wt.-%
K2O
wt.-%
SO3
wt.-%
TiO2
+MnO
+P2O5
wt.-%
67.1
62.8
64.7
65.7
65.8
67.1
66.4
66.3
62.8
20.3
19.1
19.7
24.2
22.4
23.5
20.5
21.0
21.8
3.77
5.10
3.81
3.95
3.51
3.98
5.26
3.36
4.27
4.18
3.18
2.37
1.55
4.08
1.25
2.80
5.43
5.30
0.64
2.94
1.38
0.59
0.76
0.58
1.19
0.74
0.69
0.14
0.27
0.15
0.09
0.15
0.01
0.04
0.17
0.14
0.64
0.90
0.76
0.66
0.63
0.77
0.54
0.35
0.56
2.76
3.26
2.73
2.47
2.33
2.66
2.98
2.08
2.34
0.47
2.45
4.40
0.79
0.34
0.15
0.29
0.57
2.10
Table 3C3A and CaO contents of cements, initial hydration energies and
amount of ettringite gel formed in the presence of 1.0% of PCE sample
MPEG-25 (1 J/g = 0.43 Btu/lb)
Cement
sample
CEM I 32.5 N
CEM I 32.5 R
CEM I 42.5 N
CEM I 42.5 R
CEM I 42.5 R HS
CEM I 52.5 N
CEM I 52.5 R
CEM I 52.5 R HS
API Glass G
C3A
wt.-%
1.90
9.85
6.00
7.57
1.77
8.04
8.90
1.56
1.20
CaO
(free)
wt.-%
0.11
0.97
0.15
0.04
0.06
0.03
0.10
0.27
0.10
Spec.
surf. area
(Blaine)
[cm2/g]
3,319
3,589
3,889
4,972
3,280
3,299
4,803
4,332
2,998
5 min
hydration
energy
[J/g]
1.71
14.83
11.66
22.18
0.88
5.49
25.74
0.22
1.99
100 min
hydration Amount of ettringite
energy
gel
[J/g]
2.86
no gel
23.88
very large
13.65
large
23.95
very large
3.11
no gel
9.67
medium
31.28
very large
5.65
no gel
2.75
no gel
well cement was tested. The X-Ray fluorescence (Oxide) analysis of the cements is exhibited in Table 2. Their phase compositions were analyzed by quantitative XRD (Rietveld)
and their free lime content was assessed using the Franke method.15 The cement samples
were selected such as to cover a broad range of C3A contents from 1.2 to 9.85 wt.-% (see
Table 3).
PCE performance test The dispersing force of the PCE samples was determined
using a mini slump test carried out as follows: First, a constant water-to-cement ratio
(w/c) of 0.3 was chosen. At this w/c ratio, the dosages of the polymers required to reach a
spread of 26 0.5 cm (10.2 inch) were determined. Generally, the polymers were dissolved
in the required amount of mixing water placed in a porcelain cup. The amount of water
contained in the PCE solution was subtracted from the amount of mixing water. Next,
within 5 seconds, 350 g (12.35 oz.) of cement were added to the mixing water and thoroughly agitated manually for a total period of 4 minutes. The cement paste was then poured
406SP-302-30
into a Vicat cone (height 40 mm (1.57 inch), top diameter 70 mm (2.76 inch), bottom
diameter 80 mm (3.15 inch)) placed on a glass plate and the cone was lifted vertically. The
resulting spread of the paste was measured twice, the second measurement being in a 90
angle to the first and averaged to give the spread value.
When PCE was added in a delayed mode, then the cement was first mixed with water
only, and the PCE solution was added after 0.5, 1 or 2 minutes respectively. Total mixing
time was kept constant at 4 minutes.
Initial hydration energyThe initial hydration energy was determined utilizing an
isothermal calorimeter (TAM Air Thermometric, Jrflla, Sweden). For this purpose, 4 g
(0.141 oz) of cement and 2 g (0.07 oz.) of mixing water were placed into the calorimeter
in two separate compartments and were left to equilibrate until the calorimeter showed a
constant base line. Using the titration cell, the mixing water was added to the cement placed
in the vial and the resulting cement paste was carefully mixed for 1 minute. The measurement proceeded over 100 minutes and the hydration energy released during this period
was captured. All measurements were then repeated under adiabatic conditions using a
Dewar flask (inner height: 9 cm (3.54 inch); inner diameter: 7 cm (2.76 inch)) in which the
temperature increase caused by the hydrating cement paste was determined. In this case
the measurement was stopped after 5 minutes. Both methods yielded very similar trends.
Separation of initial ettringite gel It was found that the initial hydration products
(mainly ettringite) can be separated from the cement via centrifugation (20 minutes at
10.000 g). There, they form an aqueous top layer above the cement residue. For the separation, the same cement paste like in the mini slump test was prepared, except that here
the w/c ratio was increased to 0.5 and PCE dosages were set constant at 1% by weight of
cement. At this high w/c ratio and PCE dosage, the cement pastes exhibit strong bleeding.
After mixing, the cement paste rested for 5 minutes. Then the bleeding water (which is
where a significant portion the nano-sized ettringite accumulates) was taken up using a
pipette and centrifuged as described above. A gel-like top layer with ~ 85 wt.-% solids
content was obtained. The gel layer was carefully removed and dried at 30 C under atmospheric pressure. It is highly important to mildly dry the product, as ettringite easily can
dehydrate to metaettringite which is X-ray amorphous.16 The resulting colorless powder
was analyzed using X-ray diffraction (Bruker AXS D8 Advance, Karlsruhe, Germany),
elemental analysis (Elementar vario EL, Hanau, Germany) and thermogravimetry (Netzsch
STA 409 TG-MS, Selb, Germany).
Synthetic ettringiteEttringite was precipitated from solution by dissolving 0.255 g
(0.009 oz.) of Al2(SO4)3 18 H2O in 10 mL (0.338 fl. oz.) of water. To this solution, 100 mL
(3.38 fl. oz.) of a saturated Ca(OH)2 solution (~1.6 g/L) were rapidly added under vigorous
stirring. Instantaneous precipitation of ettringite was evident as the solution turned turbid.17
Influence of the PCE samples on ettringite crystallization was analyzed by dissolving the
PCE in the aluminum sulfate solution. The amount of PCE was kept constant at 0.35 g
(0.0123 oz.) (~ 10 wt.-% of the amount of ettringite formed). One half of the resulting
dispersion was centrifuged for 20 minutes at 10.000 g and carefully dried under atmospheric pressure. The resulting colorless powder was analyzed using X-ray diffraction. The
other part of the dispersion was immediately analyzed for particle size distribution using
dynamic light scattering (ZetaSizer Nano ZS, Malvern Instruments, Worcestershire, United
Kingdom).
Table 4PCE dosages required in cement paste to achieve a 26 cm (10.2

inch) slump flow (w/c ratio = 0.30)
Cement
sample
CEM I 32.5 N
CEM I 32.5 R
CEM I 42.5 N
CEM I 42.5 R
CEM I 42.5 R HS
CEM I 52.5 N
CEM I 52.5 R
CEM I 52.5 R HS
API Glass G
MPEG-7
0.25
> 1.00
0.35
> 1.00
0.15
0.22
> 1.00
0.31
0.10
MPEG-25
APEG-34
APEG-34AM
[PCE dosages in % by weight of cement]
0.11
0.21
0.10
> 1.00
> 1.00
0.25
0.16
0.29
0.14
> 1.00
> 1.00
0.22
0.09
0.14
0.08
0.12
0.20
0.12
0.90
0.45
0.31
0.12
0.30
0.12
0.06
0.11
0.05
IPEG-25
0.10
0.51
0.13
0.35
0.08
0.10
0.26
0.11
0.05

Cement dispersion effectiveness of PCEsThe dispersing force of the PCE samples
was determined in a mini slump test using cement paste. In this test, a paste was prepared
at a w/c ratio of 0.3 and the PCE dosage was adjusted to yield a slump flow of 26 0.5
cm (10.2 inch). The PCE was dissolved in the mixing water prior to cement addition. This
test was carried out for the nine different cement samples and for all five PCE polymers.
Table 4 shows significant variations in PCE dosages required for the different cements.
Dosages of > 1% suggest that this PCE is not able to effectively disperse the cement. It
is also apparent that PCE sample APEG-34AM exhibits the most consistent performance
(i.e. best cement compatibility), followed by IPEG-25. For example, for APEG-34AM
the dosages vary between 0.05 and 0.31%, and for IPEG-25 between 0.05 and 0.51%
by weight of cement (bwoc), thus indicating relatively stable performance with different
cement samples. Opposite to this, the three other PCE samples exhibit much higher variations in their dosages, and they cannot fluidize at least two cement samples at all.
The data from Table 4 also clearly indicate that cement samples CEM I 32.5 R and CEM
I 42.5 R seem to be most difficult to disperse. Both cements exhibit relatively high C3A
contents (9.85% and 7.51% respectively). Accordingly, for the PCE compatibility of these
cements the order as follows was established (C3A content in wt.-%): API Class G (1.20%)
> CEM I 42.5 R HS (1.77%) > CEM I 52.5 N (8.04%) > CEM I 32.5 N (1.40%) > CEM
I 52.5 R HS (1.56%) > CEM I 42.5 N (6.00%) > CEM I 52.5 R (8.90%) > CEM I 42.5 R
(7.57%) > CEM I 32.5 R (9.85%). Yet, the C3A content alone does not seem to represent
the only criteria for incompatibility, because one cement sample (CEM I 52.5 N) possesses
a high amount of C3A (8.04%), but still is easy to disperse.
Furthermore, in additional tests the dispersing performance was assessed at delayed addition of the PCEs. As cement, the difficult sample CEM I 42.5 R (C3A content 7.57%) was
used. In the following, results for PCE sample MPEG-25, a PCE with low cement tolerance,
will be described. Its dosage was fixed at 0.20% bowc. At first, the PCE was dissolved in
the mixing water, as described above. In the other 3 tests, the PCE was added in a delayed
mode after 0.5, 1.0 and 2.0 minutes respectively. Figure 2 shows the development of slump
flow with increasing delay of PCE addition. It became apparent that the phenomenon of
cement incompatibility disappears completely when the PCE sample is added only 1 min
408SP-302-30
Figure 2Slump flow of CEM I 42.5 R (w/c = 0.3) depending

on the point of addition of 0.20% bwoc of MPEG-25 PCE
after the cement has been mixed with water. Similar results were obtained when the other
incompatible PCE samples were tested with the difficult cements.
These findings allow to conclude that the phenomenon of incompatibility between
specific cements and PCE polymers can be avoided when PCE is added in a delayed mode.
However, especially when formulating mortars or concretes possessing low w/c ratios, the
preferred mode of addition includes placing the PCE in the mixing water to achieve faster
wetting of the cement and aggregate particles, thus to reduce mixing time. Consequently,
still a need exists for PCE polymers which work well with a broad diversity of cement
samples, even when the PCE is placed in the mixing water.
Another conclusion from this test is that the incompatibility phenomenon seems to be
linked to processes occurring during the very first seconds of cement hydration. This gave
reason to study the initial heat release from hydration of the cement samples.
Initial heat of hydration of cementsThe initial hydration energy released by the
cements was determined in an adiabatic calorimeter for the first 5 minutes of hydration
and in an isothermal calorimeter for the first 100 minutes of hydration. The results are
presented in Table 3. The amounts of heat measured using the isothermal calorimeter are
generally slightly higher than those obtained under adiabatic conditions as hydration time
was longer in the isothermal test. For both instruments, a similar trend was found. The heat
flow curves obtained from the isothermal test are presented in Figure 3. Note that there,
the curves for cement samples CEM I 32.5 N, CEM I 42.5 R HS and API Class G oil well
cement are almost identical and therefore cannot be distinguished. In the first 5 minutes of
cement hydration, the initial hydration energies vary greatly between the cements tested.
They range from almost 0 J/g (0 Btu/lb) for CEM I 52.5 R HS to more than 25 J/g (10.75
Btu/lb) for sample CEM I 52.5 R. Furthermore, it became evident that the amount of initial
hydration energy correlates with PCE dosages: Cement samples which release substantial
amounts of heat are more difficult to disperse and require higher PCE dosages. Generally,
the initial hydration energies of cements possessing a very low C3A content (< 2% in CEM
I 32.5 N, CEM I 52.5 R HS, CEM I 42.5 R HS and API Class G) are extremely low (0.22
1.99 J/g (0.09 0.86 Btu/lb)), suggesting that the amount of heat released is linked to
Figure 3Heat flow curves of neat cements (w/c = 0.5)

hydrated over 100 minutes in an isothermal calorimeter.
the C3A content. However, one cement exhibiting a high C3A content of 8.04% (sample
CEM I 52.5 N) exhibits a rather low initial hydration energy of 5.48 J/g (2.34 Btu/lb). This
indicates that the amount of immediately soluble sulfate or the modification of C3A present
plays an additional role here. Additionally, the impact of the specific surface area a cement
sample on the amount of ettringite gel was considered. However, the fact that cement
sample CEM I 32.5 R possesses a relatively low specific surface area (see Table 3), but still
produces copious amounts of ettringite, suggests that the surface area of a cement sample
plays an insignificant role here, if any. Further investigations will be necessary to clarify
this point. The measurements on the initial hydration energy were repeated in the presence
of the PCE samples to determine whether any of them affects the amount of heat released.
However, no significant differences could be observed. This finding led the authors to
conclude that the incompatibility phenomenon is linked to hydration products which crystallize almost immediately after the cement comes into contact with water.
Analysis of colloidal gelBy coincidence it was observed that pastes prepared from
difficult cements and PCEs form a white, gelous top layer when centrifuged for 20
minutes at 10.000 g (Figure 4). Especially PCEs exhibiting poor cement compatibility such
as MPEG-25 or APEG-34 produced a significant volume of this layer. Also, the amount of
gel was especially high for the samples CEM I 32.5 R, CEM I 42.5 R and CEM I 52.5 R
which before were identified as those which are difficult to disperse (Table 3). The amount
of gel produced by PCE MPEG-25 and the cement samples are exhibited in Table 3. There,
it is evident that cements which are difficult to disperse produce large amounts of this gel,
and vice versa. For example, from 300 g (10.59 oz) of difficult cement sample CEM I
32.5 R ~ 6 g (0.21 oz) of a transparent, waxy gel with a solids content of ~ 85 wt.-% is
obtained. Figure 4 displays how the amount of gel can vary with cement composition (i.e.
the C3A content) when PCE polymer MPEG-25, a PCE with low cement tolerance, is used.
Relative to the different PCE samples, the amount of gel was very similar among the
MPEG-type PCEs and for APEG-34 (= the less cement compatible PCEs). Whereas,
IPEG-25 produced slightly less gel and for APEG-34AM, hardly any gel formation was
detectable. Thus, based on the amounts of gel produced, the order as follows with respect to
cement compatibility was established: APEG-34AM > IPEG-25 >> APEG-34 > MPEG-25
410SP-302-30
Figure 4Comparison of the amounts of ettringite gel

obtained as top layer after centrifugation of cement paste
(w/c = 0.5) prepared in the presence of 1% bwoc of PCE
sample MPEG-25 using (from left to right): CEM I 52.5 R
HS (no gel), CEM I 52.5 N (medium amount) and CEM I
32.5 R (large amount).
Figure 5XRD pattern of the gel layer formed in the presence of PCE sample MPEG-25 atop CEM I 32.5 R paste
after centrifugation.
> MPEG-7. This order suggests that even within the same chemical group of PCE, different
cement compatibilities can occur.
XRD analysis revealed that this gel contains pure ettringite (Figure 5), as was confirmed
also by elemental analysis of the dried powder. However, in the presence of 1% bwoc of
PCE sample MPEG-25, a carbon content of 7.73 wt.-% was detected in the gel, thus indicating significant adsorption of PCE on the surface of colloidal ettringite. Furthermore,
TG-MS analysis of the ettringite gel containing PCE was performed (Figure 6). Dehydration of ettringite occurs in the temperature range between 100 C and 180 C (212 356
Figure 6TG-MS analysis of ettringite gel produced from

CEM I 32.5 R hydrated in the presence of 1 wt.-% PCE
sample MPEG-25.
F) while the PCE begins to decompose above 200 C (392 F) as evidenced by the mass
signal of m/z = 44 corresponding to the release of CO2. The opaque appearance of the gel
indicates that the ettringite crystals must be extremely small and thus can adsorb large
amounts of PCE.
Generally, an ettringite gel was not observed when no PCE was present in the cement
paste. Apparently, the PCE polymers either hinder ettringite formation substantially or
more likely act as morphology modifying agent which reduce the size of the early ettringite crystals to colloidal or even nano scale.
Furthermore, formation of the ettringite gel was studied at delayed addition of PCE
sample MPEG-25. However, no ettringite gel was obtained even when the PCE was added
only 30 seconds after the cement paste had been mixed. This again instigates that the
incompatibility phenomenon is derived from an instantaneous interaction between cement
und PCE which occurs as soon as water is added.
To summarize, the amounts of gel formed correlate well with the PCE dosages required
in the performance test and also with the initial hydration energies. In other words, cements
which are difficult to disperse are characterized by high initial hydration energy, large
amounts of colloidal ettringite gel and abnormal PCE dosages. The experiments also
suggest that specific PCE molecules can strongly impact ettringite crystallization. They
seem to act as morphology modifying agent which greatly reduce the aspect ratio of the
early ettringite crystals. To investigate this aspect, the crystallization of pure synthetic ettringite was studied in the presence and absence of the PCE polymers. Pure ettringite crystals
were preferred over an ettringite gel obtained from cement to eliminate any effects derived
from cement impurities.
For applicators of PCEs who want to find out whether poor performance of a PCE
product is owed to cement incompatibility, the simple test as follows is recommended:
mix your cement sample at w/c = 0.5 with water containing 1 wt.-% of your PCE polymer,
then centrifuge the bleeding water of the paste for 20 min at 10.000 g. If a significant
412SP-302-30
amount of opaque gel appears at the top of the centrifugate, then incompatibility between
this particular cement and the admixture is confirmed. This test allows assessing whether
mal-performance of a PCE product is owed to incompatibility with cement, or derives from
other factors such as e.g. the presence of clay impurities etc.
Impact of PCE on ettringite crystallization Following the method of Struble,17
synthetic ettringite was prepared via precipitation from aluminum sulfate solution combined
with a saturated calcium hydroxide solution. Note that this synthesis route for ettringite
is quite different compared to the formation mechanism in cement, still valuable results
relating to the effect of PCE superplasticizers on ettringite morphology can be retrieved.
Ettringite precipitates were collected from both without PCE as well as in the presence of
PCE. XRD analysis confirmed that pure ettringite was formed in all cases. Visual inspection of the samples already revealed substantial differences in the crystal sizes of the ettringite. The product obtained in the absence of PCE settled quickly and appeared as milky
white suspension, whereas samples produced in the presence of PCEs, especially of the
MPEG PCEs and of APEG-34, were stable, did not settle, exhibited only slight turbidity
or were almost transparent. Concentration of these samples via centrifugation yielded the
same viscous gel as found before when using cement pastes.
Next, the suspensions holding the precipitated ettringite were characterized by dynamic
light scattering. For the sample where no PCE was present, using laser granulometry
particle sizes of ~10 m (3.910-4 inch) (d50 value) were detected. Opposite to this, the sizes
of the ettringite samples obtained in the presence of PCEs were much smaller: MPEG-7
718 14 nm (2.810-5 inch); MPEG-25 613 11 nm (2.3910-5 inch); APEG-34 1,536
38 nm (5.9910-5 inch); IPEG-25 1,720 93 nm (6.7110-5 inch) and APEG-34AM 2,050
65 nm (7.9910-5 inch). Remarkably, except for MPEG-7 the sizes of the ettringite crystals
correlate very well with the PCE dosages required for difficult cements, the amounts of gel
formed and the initial hydration energies. More specifically, PCEs which fail with difficult cements produce large amounts of a gel holding particularly small colloidal ettringite
crystals. MPEG-7 does not fit well into this order because per se it presents a relatively
poor dispersant, as a result of its short side chain. This polymer always requires substantially higher dosages than the other PCE samples tested here.
The experiments on precipitation of pure ettringite from solution also revealed that ettringite formation and precipitation is instantaneous, i.e. turbidity immediately occurs when
aluminum sulfate and calcium hydroxide come into contact. Even more, the precipitation was complete within ~ 10 seconds. This observation is consistent with the results
on delayed PCE addition to cement paste (Figure 2). There, the negative effect on PCE
performance was observed only within the first seconds. After 1 min of hydration, the
negative impact from the difficult cement sample had vanished.
CONCLUSIONS
The study suggests that the incompatibility phenomenon occurring between specific PCE
polymers and certain cements relies on modification of the ettringite crystal morphology.
Based on these findings, incompatibility between a PCE and cement sample is likely to
occur when the following conditions exist:
a) The cement sample exhibits an elevated C3A content (> 7 wt.-%) and contains significant amounts of immediately soluble sulfates (alkali sulfates, CaSO4 hemihydrate) to
produce large amounts of ettringite.
b) The PCE sample strongly impacts ettringite morphology and produces nano-sized
instead of the common meso-sized, large crystals. The nano-sized ettringite possesses a
multiple of the surface area of conventional ettringite formed in the absence of PCE, and
therefore consumes large amounts of superplasticizer via adsorption.
FURTHER RESEARCH
The authors acknowledge that this study only represents a first step in understanding
the complexity of this incompatibility phenomenon. For example, it will be most useful
to investigate why specific PCE molecules interact differently with the different surfaces
of ettringite crystals. Molecular simulations evidencing the interaction energies might
be useful to establish a correlation between PCE molecular structure and ettringite
morphology. Additionally, the nano-sized ettringite crystals need to be characterized using
high-resolution electron microscopy. Attempts to achieve this in the present study failed
because under the impact of the vacuum and electron beam, the ettringite crystals immediately dehydrated to amorphous meta ettringite.
AUTHOR BIOS
M. Sc. Alex Lange studied chemistry at Technische Universitt Mnchen, Germany.
Currently, he is a Ph.D. student at the Chair for Construction Chemicals at the same
university. His research focuses on the synthesis and working mechanism of polycarboxylate-based superplasticizers and the effect of molecular architecture on PCE
performance.
Prof. Johann Plank is full Professor at the Institute of Inorganic Chemistry at Technische Universitt Mnchen, Germany. Since 2001, he holds the Chair for Construction
Chemicals there. His research interests include cement chemistry, chemical admixtures,
organic inorganic composite and nano materials, concrete, dry mortar and oil well
cementing.
ACKNOWLEDGMENT
The authors greatly thank Clariant and Nippon Oil and Fats Company for generously
supplying monomers for the PCE synthesis. A. Lange also wishes to thank the TUM
Center for Advanced PCE Studies for financing the final part of his Ph.D study.
REFERENCES
1. , . Okamura, H., Ouchi, M., Self-Compacting Concrete, Journal of Advanced
Concrete Technology, V. 1, 2003, pp. 5-15.
2. Sakai, E.; Kakinuma, Y.; Yamamoto, K.; and Daimon, M., Relation between the
Shape of Silica Fume and the Fluidity of Cement Paste at Low Water to Powder Ratio,
Journal of Advanced Concrete Technology, V. 7, No. 1, 2009, pp. 13-20. doi: 10.3151/
jact.7.13
3. Plank, J.; Schrfl, C.; Gruber, M.; Lesti, M.; and Sieber, R., Effectiveness of Polycarboxylate Superplasticizers in Ultra-High Strength Concrete: the Importance of PCE
414SP-302-30
Compatibility with Microsilica, Journal of Advanced Concrete Technology, V. 7, No. 1,

2009, pp. 5-12. doi: 10.3151/jact.7.5
4
. T. Hirata, Cement dispersant, JP1984-18338, 1981.
5. Plank, J., PCE Superplasticizers Chemistry, Application and Perspectives 18.
ibausil, Weimar, V. 1, Sep. 2012, pp. 91-102.
6. Bonen, D., and Shondeep, S., The Superplasticizer Adsorption Capacity of Cement
Pastes, Pore Solution Composition, and Parameters Affecting Flow Loss, Cement and
Concrete Research, V. 25, No. 7, 1995, pp. 1423-1434. doi: 10.1016/0008-8846(95)00137-2
No. 10, 2002, pp. 1507-1513. doi: 10.1016/S0008-8846(02)00782-2
8. Yoshioka, K.; Sakai, E.; Daimon, M.; and Kitahara, A., Role of Steric Hindrance
in the Performance of Superplasticizers for Concrete, Journal of the American Ceramic
Society, V. 80, No. 10, 2005, pp. 2667-2671. doi: 10.1111/j.1151-2916.1997.tb03169.x
9. Agarwal, S. K.; Masood, I.; and Malhotra, S. K., Compatibility of Superplasticizers
with different cements, Construction & Building Materials, V. 14, No. 5, 2000, pp.
253-259. doi: 10.1016/S0950-0618(00)00025-8
10. Lei, L., and Plank, J., A Study on the Impact of Different Clay Minerals on the
Dispersing Force of Conventional and Modified Vinyl Ether Based Polycarboxylate
Superplasticizers, Cement and Concrete Research, V. 60, 2014, pp. 1-10. doi: 10.1016/j.
11. Prince, W.; Espagne, M.; and Atcin, P.-C., Ettringite formation: A crucial step in
cement superplasticizer compatibility, Cement and Concrete Research, V. 33, No. 5, 2003,
pp. 635-641. doi: 10.1016/S0008-8846(02)01042-6
12. Aitcin, P.-C.; Jolicoeur, C.; and MacGregor, J. G., Superplasticizers: How they work
and why they occasionally dont, Concrete International, V. 16, 1994, pp. 45-52.
13. .] Schrfl, C., Gruber, M., Plank, J., Structure-performance relationship of polycarboxylate superplasticizers based on methacrylic acid esters in ultra high performance
concrete, in: Fehling, E.; Schmidt, M.; and Strwald, S., Ultra High Performance Concrete
(UHPC) Second International Symposium on Ultra High Performance Concrete, Kassel,
2008, 383-390.
14. Teresa, M.; Laguna, R.; Medrano, R.; Plana, M. P.; and Tarazona, M. P., Polymer
characterization by size-exclusion chromatography with multiple detection, Journal of
Chromatography. A, V. 919, No. 1, 2001, pp. 13-19. doi: 10.1016/S0021-9673(01)00802-0
15. Franke, B., Bestimmung von Calciumoxyd und Calciumhydroxyd neben wasserfreiem und wasserhaltigem Calciumsilikat, Zeitschrift fur Anorganische und Allgemeine
Chemie, V. 247, No. 1-2, 1941, pp. 180-184. doi: 10.1002/zaac.19412470115
16. Zhou, Q.; Lachowski, E. E.; and Glasser, F. P., Metaettringite, a decomposition
product of ettringite, Cement and Concrete Research, V. 34, No. 4, 2004, pp. 703-710.
doi: 10.1016/j.cemconres.2003.10.027
17. Struble, L. J., Synthesis and Characterization of Ettringite and Related Phases,
Proceedings of the 8th International Congress on the Chemistry of Cement, Rio de Janeiro,
Brazil, 6, 1986, 582-588.
SP-302-31
Enhancing Workability Retention of

Concrete Containing Natural Zeolite by
Superplasticizers Combination
by Hessam AzariJafari, Mohammad Shekarchi, Javad
Berenjian, and Babak Ahmadi
Use of pozzolanic materials such as natural zeolite as portland cement replacement helps
to reduce amount of CO2 emission due to clinker production. Natural zeolite also improves
mechanical and durability properties of concrete. It is common to use natural zeolite as a
rheological modifying admixture in flowing concrete. However, many cases were reported
that zeolite blended cements showed severe workability loss. The object of the analysis is to
investigate compatibility of different chemical-based superplasticizers and effect of superplasticizers combination on workability retention of concrete made with zeolite blended
cement. The results show that combination of lignosulfonate admixture with naphthalene
and polycarboxylate based admixture not only reduces the superplasticizers demand to
achieve certain workability retention, but also helps to reduce slump loss.
Keywords: natural zeolite; superplasticizers combination; superplasticizers demand;
workability loss.
INTRODUCTION
Since ancient times, natural pozzolans such as zeolite have been broadly used in limebased mixtures. Nowadays, zeolite is consumed as a supplementary cementitious material
(SCM) in many countries that have enough resources of zeolitic tuffs. Natural Zeolite
(NZ) is a three dimensional structured pozzolan containing alumino-silicate mineral. NZ
can enhance microstructural properties of cement paste by decreasing the size and amount
of C-S-H crystals.1,2 Earlier studies have shown that Portland cement replacement by NZ
(5% to 55%, by mass) contributes to consumption of calcium hydroxide which is formed
during hydration of portland cement.3-6 Pozzolanic activity and effects of zeolite powder
on the hardened properties of concrete have been widely investigated by researchers1,2,7-11
On the fresh properties of zeolite blended cements, Ramezanianpour et al.8 reported
that high NZ replacement level (30%) promotes segregation resistance of low portland
cement SCC mixtures, which was measured by sieve segregation test. A study by Ranjbar
et al.9 showed that use of NZ (up to 20%) contributes to satisfying the SF2 criteria for SCC
415
416SP-302-31
Table 1 Physical Properties of cementitious materials.

Portland cement
SiO2
Al2O3
Fe2O3
CaO
MgO
SO3
Na2O
K2O
L.O.I
Specific Gravity (kg/m3)
Blaine (m2/kg)
Chemical composition (%)

21.25
3.38
3.56
63.1
1.96
1.71
0.216
0.56
1.87
3140
320
NZ
67.79
13.66
1.44
1.68
1.20
0.50
2.04
1.42
10.23
2200
320
mixtures (slump flow diameter of 660 to 750 mm [26 to 29.52 in.]). Criterion of SF2 is
satisfactory for normal applications. Similar results were reported by Cioffi et al.12 at 40%
NZ replacement level.
Still, influence of NZ inclusion on the fresh state behavior of cementitious materials has
been investigated only to a basic level. Some observations regarding the decreased workability, higher viscosity, and accelerated workability loss (in the case of NZ incorporation)
are reported by some researchers.1,6,8,9,13
While the compatibility issue has been investigated for various pozzolan-admixture
combinations, it has not been carried out for NZ pozzolan. In fact, some observations with
respect to the higher workability loss of NZ included materials are often reported, but
detailed investigation on the involved parameters has been neglected by the researchers.
In addition, effect of lignosulfonate- based admixture (LS) as partial replacement of naphthalene sulfonate formaldehyde (NSF) and Polycarboxylate ether (PC) on workability loss
and dosage on plain concrete and zeolite-incorporated concrete were studied.
Based on the earlier research works on Iranian NZ, 10% was selected as the cement
replacement level in zeolite-incorporated concrete (NZ concrete).9,10,13 Also, each of the
chemical admixtures was employed within the concentration range recommended by the
producer.
Materials
The cement used in this study was ASTM C150 Type II portland cement. The clinoptilolite type NZ was from the quarries in the north of Semnan, central region of Iran. The
physical and chemical compositions of the cementitious materials are listed in Table 1.
The concrete aggregates were composed of two fractions: 0-6 mm [0-0.23 in.] and 6-19.5
mm [0.23- 0.77 in.]. The lower fraction (0-6 mm [0-0.23 in.]) which was used in concrete
mixtures had specific gravity of 2.67, water absorption of 2.8% and fineness modulus of
Enhancing Workability Retention of Concrete Containing Natural Zeolite by

Superplasticizers Combination 417
Fig. 1 - Sieve analysis of combined aggregates

3.06. The coarse aggregate is crushed type. Specific gravity and water absorption of the
coarse aggregate (6-19.5 mm [0.23- 0.77 in.]) were 2.75 and 1.8%, respectively. The sieve
analysis of total aggregates of the mixtures is presented in Fig. 1.
In order to investigate influence of superplasticizers on zeolite blended concrete, Ligno
sulfonate (LS), naphthalene sulfonate formaldehyde (NSF), and Polycarboxylate ether
(PC) were selected. The superplasticizers were the most widespread chemical admixtures
in 3 different generations. As they performance are more effective in different w/c range,
it might be a suitable choice for studying the effect of NZ workability retention. The properties of these three commercially available admixtures are presented in Table 2. Each
admixture was used single and combined. The combined admixtures included LS and NSF
or PC at different ratios. The ratios of superplasticizers combination were 30-70, 50-50,
and 70-30.
Mixture proportions of concrete including different dosages and combinations of chemical admixtures are listed in Table 3. Mixtures including no zeolite (Plain concrete) as
control mixtures were made for comparison purposes. These mixtures involve only portland cement as binder.
The initial slump for the mixtures was kept constant at 17520 mm [6.890.78 in.].
It should be noted that the slump was measured according to ASTM C14314 at the end
of mixing procedure and also at 15, 30, 45, and 60 minutes (after initial water-cement
contact). The fresh concrete was placed in slump cone. The cone was lifted and the height
of fresh concrete cone was measured at two locations using a caliper. The average value
was used as the slump. Depending on the water reduction ability of the superplasticizing
418SP-302-31
Table 2 Properties of chemical admixtures

Superplasticizer Type
poly-carboxylate Ether (PC)
Naphthalene Sulfonate Formaldehyde (NSF)
Lignosulfonate (LS)
Specific
Gravity
1.14
pH
6.3
Solids Content (%)

40
Chloride Content (%)
1.20
7.2
50
< 0.01
1.17
6.1
50
Table 3- Mixture proportions of concrete mixtures
Mixture type
Non-zeolite
mixtures(Plain
concrete)
Zeolite
mixtures(NZ
concrete)
Stone Aggregate
0-6 mm(Kg/ 6-19.5 mm(Kg/
m3)
m3)
Portland cement(Kg/
m3)
Natural zeolite(Kg/
m3)
W/C
Ratio
350
0.42
889
892
315
35
0.42
883
885
agents, one w/cm ratio was selected for each of the three admixtures, and adequate dosages
were added to obtain the desired initial slump (17520 mm [6.890.78 in.]).
Required dosage of single and combined superplasticizers
The effect of LS combination with NSF and PC with different ratios of 30-70, 50-50, and
70-30 on dosage was investigated. Fig. 2 and 3 demonstrate the effect of superplasticizers
combination on dosage of the admixtures for plain and NZ concrete, respectively. It should
be noted that the measured slump before addition of the chemical admixtures was 755
mm [2.950.2 in.] for plain concrete. At w/c 0.42, the measured slump for NZ mixtures was
205 mm [0.790.2 in.]. This reduction of slump is usually justified by the honeycomb like
structure of NZ crystals which have extremely small pores and channels, varying in size
from 31044104 m. These characteristics enable NZ to absorb water by over 30% of its
dry weight.1,15 Comparing SP dosage reveals that incorporation of NZ has increased superplasticizers demand. The most increase in SP demand belongs to PC which is nowadays
the most common type of superplasticizer for use in special concretes such as SCC. The PC
dosage has been increased by 0.5% at NZ concrete compared to plain concrete. Increasing
the rate of superplasticizers demand might be related to high specific area, which is existed
in NZ. Therefore, a high portion of superplasticizers might be absorbed by NZ. As a result,
the superplasticizer demand increased. Dispersing mechanisms in PC superplasticizer are
based on steric hindrance and electrostatic repulsion of copolymers included in PC.16 The
electrical charge induced on cementitious materials particles with PCs is lower than for
NSF and LS.3 As it was shown in Fig. 2 and 3, inclusion of LS as partial replacement of
NSF and PC reduced the required dosage in NZ concrete. The combination of 70% PC and
30% LS could result in the lowest superplasticizer demand in NZ mixture. Compared to
PC, combination of 70% PC and 30% LS have resulted 50% reduction in dosage. It should

Fig. 2 Required dosage for obtaining Initial slump of

17.51 cm in plain concrete
Fig. 3 Required dosage for obtaining Initial slump of

17.51 cm in NZ concrete
be noted that PC and NSF have stronger ability of water reduction. It can be inferred the
single use of PC and NSF lead to high amount of adsorption on the cementitious materials
particles. The presence of LS could reduce the adsorption of the other superplasticizers by
NZ. Therefore, combination of the admixture could result in less superplasticizer demand
to achieve a certain workability level. Similar results were reported by Tingshu et al. and
Zidong in plain concrete.17,18
Workability retention of single and combined superplasticizers
Workability retention of plain concrete and NZ concrete in 1 hour are presented in Fig. 4
and 5, respectively. As it is demonstrated, presence of LS in plain concrete and NZ concrete
leads to higher workability retention rather than single use of superplasticizers. Single use
of NSF led to higher slump loss, while the workability loss was less severe for PC incorporated mixture. Therefore, single use of NSF in NZ concrete is not recommended in long
time agitation and hot environments. These observations could be attributed to the fact
420SP-302-31
Fig. 4 Total Slump loss in plain concrete after one hour

from batching
Fig. 5 Total Slump loss in NZ concrete after one hour from

batching
that NZ decreases the concentration of alkaline ions (Na+, K+, Ca2+) in the pore solution
through ion exchange and adsorption.19 So, the amount of alkalinity in pore solution will
be decreased. Reports have shown that NSF show more severe slump loss in low-alkali
aqueous solutions due to higher adsorption level.20 Therefore, it might be the reason for
severe slump loss of NZ concrete incorporating NSF. Workability retention of combined
superplasticizers was also investigated. Combination of LS with the other superplasticizers
result a significant reduction in slump loss. According to Fig. 4, combination of 30% PC
with 70% LS could reduce slump loss 10 cm in one hour in plain concrete (compared to
single use of PC). Blending 30% NSF and 70% LS could enhance slump retention for 75
mm [2.95 in.] compared to single use of NSF. Regarding to Fig. 5, blending LS with the
other superplasticizers could result in higher slump retention compared to single use of
NSF and PC. The observed performance might be related to dual mechanism of LS and
PC or NSF. Some authors reported that LS can delay hydration of cement particles significantly.21,22 Retarding effect of LS is a supporting hand for electro static repulsion and steric
hindrance of NSF and PC. In NZ concrete, single use of LS reached to the lowest slump
loss. High dosage of LS in the mixture (1.2% of cementitious materials weight) provide

higher amount of dispersing materials. In constant amount of cementitious materials, higher
dosage of superplasticizers leads to higher amount of superplasticizers in pore solution.
Therefore, the higher amount of the superplasticizer at longer time interval can replace
with consumed adsorbed molecules. Consequently, the mixture can maintain its workability for longer time. However, there is limitation in use of LS in concrete. Researchers
have shown that water reduction ability of LS is restricted.21 As a result, it might not be a
suitable choice for NZ incorporated mixtures with lower w/cm ratios. Blending 30% PC
and 70% LS can be the second choice for compensating slump loss in NZ concrete.
CONCLUSIONS
Based on the results of this experimental investigation of single and combined use of
three different chemical- based superplasticizers (lignosulfonate, naphthalene sulfonate
formaldehyde, and polycarboxylate ether) on workability retention of plain and natural
zeolite blended concrete, the following conclusions are drawn:
1. Incorporation of natural zeolite in concrete mixtures led to a substantial reduction on
initial slump of the concrete due high capacity of NZ water absorption.
2. The obtained results indicated that simultaneous inclusion NSF admixture was not
efficient for NZ blended cement. This observation could be related to the high alkali
adsorption capacity of natural zeolite particles.
3. Inclusion of LS was an efficient measure to mitigate the workability loss of NZ
blended concrete made with NSF and PC admixture. (Especially at incorporation of 70%
LS with 30% NSF or PC).
4. Lignosulfonate-based admixture shows the best performance in slump retention of
natural zeolite incorporated concrete. It might be due to sufficient amount of free LS in the
pore solution to compensate the consumed LS as well as its retarding effect.
AUTHOR BIOS
ACI member Hessam AzariJafari is a PhD candidate at Civil Engineering Department,
University of Sherbrooke, Sherbrooke, Quebec, Canada. He is a member of ACI Committees 212 (Chemical Admixtures) and 237 (Self-Consolidating Concrete). His research
interests include chemical admixtures, supplementary cementitious materials, and life
cycle assessment of building materials.
Mohammad Shekarchi is a professor at construction materials institute (CMI) at
University of Tehran, Tehran, Iran. His research interests include durability properties of
concrete, fresh properties of concrete and dimensional changes.
Javad Berenjian is an assistant professor at Tabari Institute of Higher Education, Babol,
Iran. His research interests include self-consolidating concrete and effect of chemical
admixtures in concrete.
Babak Ahmadi is a PhD candidate at Civil and Environmental Engineering Department,
Amirkabir University of Technology. He is also a research assistant at construction materials institute (CMI) at University of Tehran. His research interests include supplementary cementitious materials and durability properties of concrete in harsh environment.
422SP-302-31
REFERENCES
1. Ahmadi, B., and Shekarchi, M., Use of natural zeolite as a supplementary cementitious material, Cement and Concrete Composites, V. 32, No. 2, 2010, pp. 134-141. doi:
10.1016/j.cemconcomp.2009.10.006
2. Dousti, A.; Rashetnia, R.; Ahmadi, B.; and Shekarchi, M., Influence of exposure temperature on chloride diffusion in concretes incorporating silica fume or natural
zeolite, Construction & Building Materials, V. 49, 2013, pp. 393-399. doi: 10.1016/j.
3. Burgos-Montes, O.; Palacios, M.; Rivilla, P.; and Puertas, F., Compatibility between
superplasticizer admixtures and cements with mineral additions, Construction & Building
Materials, V. 31, 2012, pp. 300-309. doi: 10.1016/j.conbuildmat.2011.12.092
4. Kocak, Y.; Tasc, E.; and Kaya, U., The effect of using natural zeolite on the properties and hydration characteristics of blended cements, Construction & Building Materials,
V. 47, 2013, pp. 720-727. doi: 10.1016/j.conbuildmat.2013.05.033
5. Snellings, R.; Mertens, G.; Cizer, .; and Elsen, J., Early age hydration and pozzolanic reaction in natural zeolite blended cements: Reaction kinetics and products by in situ
synchrotron X-ray powder diffraction, Cement and Concrete Research, V. 40, No. 12,
6. Uzal, B., and Turanl, L., Blended cements containing high volume of natural zeolites:
Properties, hydration and paste microstructure, Cement and Concrete Composites, V. 34,
No. 1, 2012, pp. 101-109. doi: 10.1016/j.cemconcomp.2011.08.009
7. Narasimhulu, K.; Gettu, R.; and Babu, K., Beneficiation of Natural Zeolite through
Flash Calcination for Its Use as a Mineral Admixture in Concrete, Journal of Materials
in Civil Engineering, 2013
8. Ramezanianpour, A. A.; Kazemian, A.; Sarvari, M.; and Ahmadi, B., Use of Natural
Zeolite to Produce Self-Consolidating Concrete with Low Portland Cement Content
and High Durability, Journal of Materials in Civil Engineering, V. 25, No. 5, 2013, pp.
589-596. doi: 10.1061/(ASCE)MT.1943-5533.0000621
9. Ranjbar, M. M.; Madandoust, R.; Mousavi, S. Y.; and Yosefi, S., Mousavi, Yasin,
S., Youosefi, S., Effects of natural zeolite on the fresh and hardened properties of selfcompacted concrete, Construction & Building Materials, V. 47, 2013, pp. 806-813. doi:
10.1016/j.conbuildmat.2013.05.097
10. Valipour, M.; Pargar, F.; Shekarchi, M.; and Khani, S., Comparing a natural
pozzolan, zeolite, to metakaolin and silica fume in terms of their effect on the durability
characteristics of concrete: A laboratory study, Construction & Building Materials, V. 41,
2013, pp. 879-888. doi: 10.1016/j.conbuildmat.2012.11.054
11. Valipour, M.; Pargar, F.; Shekarchi, M.; Khani, S.; and Moradian, M., In situ study
of chloride ingress in concretes containing natural zeolite, metakaolin and silica fume
exposed to various exposure conditions in a harsh marine environment, Construction &
Building Materials, V. 46, 2013, pp. 63-70. doi: 10.1016/j.conbuildmat.2013.03.026
12. Cioffi, R.; Colangelo, F.; Caputo, D.; and Liguorim, B., Influence of High Volumes
of Ultra-Fine Additions on Self-Compacting Concrete, ACI Publication, vol. SP-239-9,
2006.
13. Sabet, F. A.; Libre, N. A.; and Shekarchi, M., Mechanical and durability properties
of self consolidating high performance concrete incorporating natural zeolite, silica fume

and fly ash, Construction & Building Materials, V. 44, 2013, pp. 175-184. doi: 10.1016/j.
14. ASTM, ASTM C143-12: Standard Test Method for Slump of Hydraulic-Cement
Concrete 2012.
15. M. FA, Mineralogy and geology of natural zeolites. Newyork: Reprint of Mineralogical Society of Americas Reviews in Mineralogy, 1993.
16. H. Liu, Pang, H., Ou, J., Zhang, L., Dai, Y., Liao, B., Effect of cross-linked polycarboxylate-type superplasticizers on the properties in cementitious system, Journal of
Applied Polymer Science, vol. 131, 2014.
17. S. F. HE Tingshu, WANG Fuchuan,WANG Huian, INFLUENCES OF COMBINATION OF SUPERPLASTICIZERS AND RETARDERS ON PROCESS OF CEMENT
HYDRATION, Journal of the Chinese Ceramic Society, vol. 06, 2007.
18. Zidong, L., Reasearch on polycarboxylates water-reducing admixture combined
with modified lignosulfonate, Ready-Mixed Concrete, vol. 01, 2012.
19. Feng, N., and Hao, T., Mechanism of natural zeolite powder in preventing alkalisilica reaction in concrete, Advances in Cement Research, V. 10, No. 3, 1998, pp. 101-108.
doi: 10.1680/adcr.1998.10.3.101
20. Kim, B.-G.; Jiang, S.; Jolicoeur, C.; and A[UNKNOWN ENTITY &idie;]tcin, P.-C.,
The adsorption behavior of PNS superplasticizer and its relation to fluidity of cement
paste, Cement and Concrete Research, V. 30, No. 6, 2000, pp. 887-893. doi: 10.1016/
S0008-8846(00)00256-8
21. V. S. Ramachandran, Effect of Sugar-free Lignosulphonates on Cement Hydration,
1979.
22. Zhen Ping Sun, L. Z., Xiong Rong Huang, Hui Yang, Liang Liang Shui, Polycarboxylate-Lignosulfonate Copolymerized High Performance Water Reducing Agent: Preparation and Application, Advanced Materials Research, V. 687, 2012, pp. 435-442.
424SP-302-31
SP-302-32
Fluidity Change of Cement Paste with

Superplasticizer by K2SO4 and KF
by Kazuki Matsuzawa, Daiki Atarashi, Masahiro
Miyauchi, and Etsuo Sakai
The calcination temperature in cement manufacturing can be reduced by the addition of
sulphate and fluoride containing compounds and it is possible that sulphate and fluoride
ions elute to the suspension after calcination. This paper describes the influence of sulphate
and fluoride ions on the action of polycarboxylate based superplasticizer in cement paste.
When the amount of K2SO4 or KF was increased, the viscosity of the cement paste with
superplasticizer increased. The amount of adsorbed superplasticizer was decreased by
K2SO4 addition but increased by KF addition. The fluidity with polycarboxylate based
superplasticizer containing more functional groups was less affected by K2SO4 addition.
In contrast to the case of K2SO4 addition, the increase in the degree of viscosity by KF
addition was not dependent on the amount of functional groups. The specific surface area
increased with K2SO4 or KF addition.
Keywords: polycarboxylate based superplasticizer; cement paste; fluidity; sulphate ion;
fluoride ion; adsorption of superplasticizer.
INTRODUCTION
High fluidity concrete is necessary to save labor and reduce the time spent in the construction of concrete. In addition, reduction of the water ratio is required to increase the strength
and the durability of the concrete, but sufficient fluidity is required for construction. For
consistent fluidity and strength, polycarboxylate based superplasticizers are used as the
air-entraining and high range water-reducing admixture. Polycarboxylate based superplasticizers are comb-type polymers that adsorb to the surface of cement particles by their
carboxyl functional groups.1 Polycarboxylate based superplasticizer stabilizes the dispersion of particles by the steric hindrance effect of the graft chains, and this type of superplasticizer can produce high fluidity with less dosage than other types of superplasticizer. In
addition, the molecular structure of polycarboxylate based superplasticizers can be easily
modified by changing the monomers.
High temperatures (>1450C: >1720 K) are necessary for alite production in cement
manufacturing, and a large amount of energy is consumed in this calcination process. The
amount of energy for the calcination is equal to 80% of all energy consumed in cement
425
426SP-302-32
manufacturing.2 Furthermore, high temperature calcination results in deterioration of

the furnace and NOx emission. Because of these problems, decrease of the calcination
temperature and reduction of energy consumption by the addition of sulphate and fluoride compounds have been investigated.3,4 Compounds that can decrease the calcination
temperature are separated into two groups based on the mechanism of the temperature
decrease. Fluxes enable decrease of the calcination temperature by affecting the liquid
phase diagram, amount of liquid, viscosity of the liquid and surface energy. Sulphate
containing compounds are potential fluxes. On the other hand, mineralizers enable a
decrease of the calcination temperature by affecting the solid phase diagram. Mineralizers
widen the temperature range in which the alite can exist to include lower temperatures.
Fluorides are potential mineralizers. There have been many studies of the use of fluxes
and mineralizers. For example, Raina and Janakiraman added CaSO4 and CaF2 to the raw
materials of cement, and evaluated the degree of temperature decrease by the amount of
free-lime after calcination.5 In their research, it has been revealed that fluxes and mineralizers bring about a 220C temperature decrease at most by the addition of 12 mass%
CaSO4/CaF2 to the raw materials. The reduction of the energy consumption ratio in the
manufacturing process is about 2.2% for every 100C temperature decrease.6 Therefore,
the energy reduction ratio is calculated to be about 4.5% for the addition of CaSO4/CaF2 at
most. However, when sulphate and fluoride are added to the raw material, it is possible that
the sulphate and fluoride ions become soluble by calcination and elute to the suspension.
Authors have investigated limestone powder (CaCO3) without hydration reactions as a
model system, and it was revealed that the fluidity of the CaCO3 paste with polycarboxylate based superplasticizer considerably decreased by the addition of sulphate and fluoride
ions.7 It has been revealed that ions that can form an insoluble salt with the calcium ions,
such as sulphate and fluoride ions, decrease calcium ions on the surface of CaCO3 particles.
It is thought that polycarboxylate based superplasticizers adsorb by complexing the calcium
ions on the surface of particles by their negatively charged functional groups. Therefore,
the adsorption of these superplasticizers will be hindered by the decrease of calcium ions
on the surface of the CaCO3 particles. This adsorption hindrance effect of inorganic ions
decreases the fluidity of the CaCO3 paste with superplasticizers. Although there are many
previous studies on CaCO3 paste, the influence of fluoride ions on the fluidity of the cement
paste with polycarboxylate based superplasticizer has not been investigated. Moreover,
regarding the influence of the sulphate ion, there are only studies of the suspension with
a smaller amount of sulphate ions than is needed in low-temperature calcination technology. To perform the low-temperature calcination by mineral salt addition in industry,
more effective superplasticizers are desirable and detailed analysis is required about the
mechanism of the fluidity change of cement paste by inorganic ions.
In this study, authors discuss the influence of K2SO4 and KF addition on the fluidity,
amount of adsorbed superplasticizer, specific surface area, and hydration reaction of
cement paste with polycarboxylate based superplasticizer. Additionally, authors investigate the mechanism for the fluidity decrease of cement paste with superplasticizer by the
sulphate and fluoride ions.
Fluidity Change of Cement Paste with Superplasticizer by K2SO4 and KF 427
Fig. 1 Molecular structure of P-n

Table 1 Polymerization ratio and mean molecular weight of P-n
Mean number of graft chains in a

molecule
P-10
1.0
0.005
47.5
P-34
1.0
0.014
13.8
Mw [g/mol] ([lb/mol])
29100
(64.2)
23100
(50.9)
This research evaluates the fluidity decrease of cement paste with superplasticizers
by inorganic ions from a viewpoint of low-temperature calcination technology and this
research can contribute to the performance of low-temperature calcination. There are no
other studies of the influence of fluoride ions on the fluidity of the cement paste with
polycarboxylate based superplasticizer except for the authors studies.8,9 This research
discusses the fluidity decrease of cement paste by inorganic ions with the adsorption action
mechanism of superplasticizer, and can provide some countermeasures against the fluidity
decrease by low-temperature calcination of cement by the addition of inorganic salts.
Materials
The polycarboxylate based superplasticizers used in this study (P-10 and P-34) were
-allyl--methoxypolyethylenemaleic anhydrite copolymers with graft chains of polyethylene oxide. The number in the superplasticizers name indicates the mean polymerization
degree (n) of the graft chains. P-10 has shorter graft chains and contains more functional
groups per unit mass than P-34. The comb-type polymers may also contain impurities, such
as low molecular weight non-grafted polyethylene oxide oligomers. Since these impurities
do not act as dispersing agents, their concentration was not included in the total superplasticizer concentration. Figure 1 shows the molecular structure of the polycarboxylate based
superplasticizers (P-n). Table 1 shows the monomer polymerization ratio of P-n (: : ),
the mean number of the graft chains in one molecule (length of the main chain), and the
mean molecular weight of P-n measured by gel permeation chromatography analysis.
Table 2 shows the chemical composition ratio of the ordinary Portland cement (OPC)
used in this study. Table 3 shows the mineral composition ratio of OPC calculated by
Bogues equation. For sulphate and fluoride ions addition, K2SO4 (>99.0%, Kanto Chem-
428SP-302-32
Table 2 Composition of OPC [mass%]

CaO
64.96
Na2O
0.32
SiO2
20.94
K2O
0.48
Al2O3
5.45
TiO2
0.27
Fe2O3
2.83
P2O5
0.31
MgO
1.54
MnO
0.08
SO3
2.05
Table 3 Mineral composition of OPC [mass%]

3CaOSiO2
59
2CaOSiO2
16
3CaOAl2O3
10
4CaOAl2O3Fe2O3
9
Annotation: These values are calculated by Bogues equation.
ical Co., Inc., Japan) and KF2H2O (>95.0%, Wako Pure Chemical Industries, Ltd., Japan)
were selected because the influence of potassium on the hydration of cement is small.
Specimens
The cement paste for the measurements was made by adding superplasticizer solution,
K2SO4 solution and KF solution to OPC. The mass ratio of water to OPC (W/C) was fixed
at 0.32. After mixing by hands with a stainless steel spoon in a rubber cup for 5 min, the
cement paste was used as the specimens for the experiments.
Methods
The fluidity of the ordinary Portland cement paste was measured by a rotational
cylinder viscometer (Haake MARS III Z41-TI, Thermo Fisher Scientific K.K, Japan)
at 20C (293 K), and the shear stress was changed linearly from 0.542000.54 Pa
(0.0000780.02900.000078 lbf/in2) in 240 s. The fluidity was evaluated by the apparent
viscosity at a shear stress of 200 Pa (0.0290 lbf/in2). In the measurement of the amount
of superplasticizer adsorbed to the solid, the adsorption time was fixed at 5 min and the
temperature was 20C (293 K). The liquid phase in the paste was separated by centrifuging
at 8200 m/s2 (26900 ft/s2) for 10 min. After separation, particles of which diameter is larger
than 0.20 m was removed by filtering from the liquid and the unadsorbed superplasticizer
concentration in the liquid was measured using a total organic carbon analyser (TOC-L
CSH/CSN, Shimadzu Corporation, Japan). The amount of adsorbed superplasticizer was
calculated from the superplasticizer concentration in the initial solution and the liquid
phase of the paste. To determine the specific surface area of the solid in the cement paste,
the hydration time was fixed at 5 min and the temperature was 20C (293 K). The hydration
was stopped by acetone, and the solid phase was separated from the liquid by centrifuging
at 8200 m/s2 (26900 ft/s2) for 10 min. After separation, the solid phase was dried under low
pressure (0.01 MPa = 1.5 lbf/in2) for 24 hours at 20C (293 K), and the surface area of the
solid was measured by the BrunauerEmmettTeller (BET) method with N2 adsorption
using and a surface area analyser (Gemini V2380, Micromeritics, USA). The degassing
of specimens before the specific surface area measurement was done in N2 gas flow for
3 hours at 40C (313 K). Also the solid phase was analyzed by X-ray diffraction (XRD)
using the wave length of CuK1 = 0.15406 nm (6.0654 ninch). To determine the hydration
reaction of the cement paste, the heat liberation rate of the cement paste was measured
Fig. 2 Relationship between the amount of added a) K2SO4, b) KF and the apparent
viscosity
using a sandwich-type conduction calorimeter (SCM-12L, Tokyo Riko, Japan). The heat
liberation rate was measured from 3 to 90 hours after hydration start at 20C (293 K).
The range of the amount of added K2SO4 was determined as 0-2.18 mass% of OPC and
the amount of KF was determined as 0-0.745 mass% of OPC. The amount of sulphate and
fluoride ions contained in 2.18 mass% K2SO4 and 0.745 mass% KF of OPC is equal to the
amount of sulphate and fluoride ions in 1 mass% SO3 and 0.5 mass% CaF2 of OPC. At this
addition ratio, the amount of K2SO4 molecules is 0.125 mol per 1 kg of OPC (0.0567 mol/
lb) and the amount of KF molecules is 0.128 mol per 1 kg of OPC (0.0582 mol/lb).
Influence of Sulphate Addition
Influence of K2SO4 Addition on FluidityFigure 2 a) shows the relationship between
the amount of added K2SO4 and the apparent viscosity of the cement paste with polycarboxylate based superplasticizer (P-n). The dosage of P-n was fixed at 0.192 mass% of OPC.
As previously reported,8,9 when the amount of added K2SO4 was increased, the apparent
viscosity of the cement paste with P-n increases. Comparing P-10 with P-34, the increase in
the degree of apparent viscosity with P-10 by K2SO4 addition was smaller than with P-34.
Therefore, the fluidity of cement paste with the polycarboxylate based superplasticizer
with more functional groups (P-10) was less susceptible to K2SO4 addition. This result
is the same as the result of CaCO3, where it has been reported that the fluidity of CaCO3
paste containing polycarboxylate based superplasticizer with more functional groups is
less susceptible to inorganic ion addition.1,10
Influence of K2SO4 Addition on the Amount of Adsorbed SuperplasticizerTable 4
shows the relationship between the amount of added K2SO4 and the amount of adsorbed
P-n to the solid phase in cement paste. The dosage of P-n was fixed at 0.192 mass% of
OPC.
Compared with the case of no K2SO4 addition, when a small amount of K2SO4 (0.0626
mol/kg=0.0284 mol/lb) was added, the amount of adsorbed P-n per unit mass solid
increased and the amount of adsorbed P-n per unit area decreased. At this concentration,
it can be considered that the fluidity decrease was brought by the adsorption hindrance
mechanism mainly. However, compared with the case of 1.09 mass% K2SO4 addition,
430SP-302-32
Table 4 Relationship between the amount of added K2SO4 and the amount
of adsorbed P-n per unit mass solid or unit area in cement paste
P-10 0.192 mass%
+K2SO4
+K2SO4
0.0626 mol/kg
0.125 mol/kg
No K2SO4
(0.0284 mol/lb) (0.0567 mol/lb)
0.487 mg/g
0.566 mg/g
0.731 mg/g
(0.487 mlb/lb) (0.566 mlb/lb) (0.731 mlb/lb)
0.533 mg/m2
0.468 mg/m2
0.490 mg/m2
2
2
(1.82 mlb/in ) (1.60 mlb/in )
(1.67 mlb/in2)
No K2SO4
0.303 mg/g
(0.303 mlb/lb)
0.376 mg/m2
(1.29 mlb/in2)
P-34 0.192 mass%

+K2SO4
0.0626 mol/kg
(0.0284 mol/lb)
0.383 mg/g
(0.383 mlb/lb)
0.306 mg/m2
(1.05 mlb/in2)
+K2SO4
0.125 mol/kg
(0.0567 mol/lb)
0.515 mg/g
(0.515 mlb/lb)
0.319 mg/m2
(1.09 mlb/in2)
Table 5 Relationship between the amount of added K2SO4 and the specific
surface area of the solid in the paste
P-10 0.192 mass%
P-34 0.192 mass%
+K2SO4
+K2SO4
+K2SO4
+K2SO4
0.0626 mol/kg
0.125 mol/kg
0.0626 mol/kg
0.125 mol/kg
No K2SO4
(0.0284 mol/lb) (0.0567 mol/lb)
No K2SO4
(0.0284 mol/lb) (0.0567 mol/lb)
0.913 m2/g
1.21 m2/g
1.49 m2/g
0.806 m2/g
1.25 m2/g
1.61 m2/g
(0.645 Min2/lb) (0.855 Min2/lb)
(1.05 Min2/lb) (0.570 Min2/lb) (0.885 Min2/lb)
(1.14 Min2/lb)
when a large amount of K2SO4 (0.125 mol/kg=0.0567 mol/lb) was added, the amount of
adsorbed P-n per unit area increased. Such a phenomenon when the amount of adsorbed
superplasticizer increased with the addition of a large amount of sulphate ions has not been
previously reported. Although the amount of adsorbed P-n increased with a large amount
of K2SO4, the fluidity of the paste with P-n did not increase with K2SO4 addition, as shown
in Figure 2 a). Therefore, the decrease in the fluidity with much K2SO4 addition in cement
paste cannot be solely explained by the adsorption hindrance mechanism.
Comparing P-10 and P-34, the amount of adsorbed P-10 was larger than the amount
of adsorbed P-34. It is supposed that the difference of the adsorbed amount relates to the
number of functional groups in the superplasticizer. It can be considered that P-10 has
stronger adsorption ability than P-34 because P-10 has more functional groups than P-34.
Influence of K2SO4 Addition on the BET Specific Surface AreaTable 5 shows the relationship between the amount of added K2SO4 and the BET specific surface area of the solid
in the paste with P-n. Figure 3 shows the XRD pattern of the solid phase in cement paste
with 0.125 mol/kg (0.0567 mol/lb) K2SO4 addition. The dosage of P-n was fixed at 0.192
mass% of OPC.
When the amount of added K2SO4 was increased, the specific surface area increased. Since
Ettringite was found in the paste with K2SO4 addition, it is supposed that the cause of the
increase of specific surface area is hydrated products generation by the reaction between
calcium aluminate and added sulphate ions. It can be considered that the increase in the
amount of adsorbed superplasticizers per unit mass solid as shown in Table 4 is related
to the cause of the increasing of specific surface area, resulting in the increasing of the
number of adsorption site for superplasticizers. The decrease in fluidity and the increase in
the amount of adsorbed superplasticizers on the solid with large amount of K2SO4 addition
suggests that superplasticizers adsorbed to the hydrated products, resulting in the concentration of superplasticizers decrease and the amount of adsorbed superplasticizers to the
Fig. 3 X-ray diffraction pattern of the solid in cement

paste with K2SO4 addition
unhydrated cement particles decrease. Therefore, it is supposed that the influence of the
hydrated products generation on the fluidity should be considered in the case of much
sulphate addition.
Influence of Fluoride Addition
Influence of KF Addition on FluidityFigure 2 b) shows the relationship between the
amount of added KF and the apparent viscosity of the cement paste with superplasticizers.
The dosage of P-n was fixed at 0.192 mass% of OPC.
As previously reported,8,9 when the amount of added KF was increased, the apparent
viscosity of the cement paste with P-n increased. In contrast to the case of K2SO4 addition,
the increase in the degree of apparent viscosity with P-10 by KF addition was almost equal
to the degree with P-34. This result is also different from the result of CaCO3, where it has
been reported that the fluidity of CaCO3 paste containing polycarboxylate based superplasticizer with more functional groups is less susceptible to inorganic ion addition.1,10
Influence of KF Addition on the Amount of Adsorbed SuperplasticizerTable 6 shows
the relationship between the amount of added KF and the amount of P-n adsorbed to the
solid phase in cement paste. The dosage of P-n was fixed at 0.192 mass% of OPC.
When the amount of added KF was increased, the amount of adsorbed P-n increased.
In particular, when the amount of added KF was 0.745 mass% of OPC, the amount of
adsorbed P-10 was 74% of the dosage of P-10. Although the amount of adsorbed P-n
increased with KF addition, the fluidity of the paste with P-n decreased with KF addition,
as shown in Figure 2 b). This result is also different from the results of CaCO3, where the
amount of adsorbed superplasticizer decreased by the adsorption hindrance of inorganic
ion addition.11 The adsorption hindrance mechanism cannot explain the decrease in fluidity
with KF addition to cement paste with P-10. The results of the fluidity and the amount of
adsorbed superplasticizer suggest other mechanisms influence the fluidity.
432SP-302-32
Table 6 Relationship between the amount of added KF and the amount of

adsorbed P-n per unit mass solid or unit area in cement paste
P-10 0.192 mass%
+KF
0.0641 mol/kg
No KF
(0.0291 mol/lb)
0.487 mg/g
0.858 mg/g
(0.487 mlb/lb) (0.858 mlb/lb)
0.533 mg/m2
0.593 mg/m2
2
(1.82 mlb/in )
(2.03 mlb/in2)
+KF
0.128 mol/kg
(0.0582 mol/lb)
1.43 mg/g
(1.43 mlb/lb)
0.773 mg/m2
(2.64 mlb/in2)
No KF
0.303 mg/g
(0.303 mlb/lb)
P-34 0.192 mass%

+KF
0.0641 mol/kg
(0.0291 mol/lb)
0.645 mg/g
(0.645 mlb/lb)
+KF
0.128 mol/kg
(0.0582 mol/lb)
0.928 mg/g
(0.928 mlb/lb)
Table 7 Relationship between the amount of added KF and the specific

surface area of the solid in the paste
No KF
0.913 m2/g
(0.645 Min2/lb)
P-10 0.192 mass%

+KF
0.0641 mol/kg
(0.0291 mol/lb)
1.45 m2/g
(1.02 Min2/lb)
+KF
0.128 mol/kg
(0.0582 mol/lb)
1.84 m2/g
(1.30 Min2/lb)
Influence of KF Addition on the BET Specific Surface AreaTable 7 shows the relationship between the amount of added KF and the BET specific surface area of the solid in the
paste with P-10. The dosage of P-10 was fixed at 0.192 mass% of OPC.
When the amount of added KF was increased, the specific surface area increased. It
is supposed that the cause of the increase of specific surface area is some types of fine
particles generation by the reaction regarding the added fluoride ions. It can be considered
that the increase in the amount of adsorbed superplasticizers is related to the cause of the
increasing of specific surface area. The decrease in fluidity and the increase in the amount
of adsorbed superplasticizers on the solid with KF addition suggests that superplasticizers
preferentially adsorbed to the fine particles, resulting in the amount of adsorbed superplasticizers on the cement particles decrease and the fluidity of cement paste decrease.
Influence of P-10 Addition on the Hydration of Cement Paste
Figure 4 shows the relationship between the dosage of P-10 and the time in which the
heat liberation rate of cement paste was largest (T1).
In the case of no KF addition, when the dosage of P-10 was increased, T1 increased and
the hydration reaction was retarded. In contrast, in the case of 0.745 mass% KF addition
to OPC, when the dosage of P-10 was increased, the hydration reaction was not retarded.
These results also suggest that P-10 preferentially adsorbed to the fine particles generated
by KF addition and the amount of P-10 adsorbed to cement particles decreased with KF
addition.
CONCLUSIONS
Based on the results of this experimental investigation, the following conclusions are
drawn:
Fig. 4 Relation between the dosage of

P-10 and T1
1. Compared with no K2SO4 addition, when a small amount of K2SO4 (0.0626 mol/kg =
0.0284 mol/lb) was added, the apparent viscosity of the cement paste with polycarboxylate
based superplasticizer increased. The amount of adsorbed polycarboxylate based superplasticizer to the unit mass solid phase in the cement paste increased by K2SO4 addition
but the amount of adsorbed polycarboxylate based superplasticizer per unit area decreased.
In contrast, compared with a small amount of K2SO4 addition, when a large amount of
K2SO4 (0.125 mol/kg=0.0567 mol/lb) was added, the amount of adsorbed superplasticizer
increased and the viscosity of the cement paste increased. The fluidity of the cement paste
with the polycarboxylate-based superplasticizer with more functional groups was less
susceptive to K2SO4 addition than the polycarboxylate-based superplasticizer with less
functional groups.
2. It is suggested that the causes of the fluidity decrease with K2SO4 addition are not
only the adsorption hindrance mechanism but also the adsorption of superplasticizer to the
hydrated compounds generated by sulphate ions and calcium aluminate.
3. On the other hand, when KF was added to the cement paste with polycarboxylate
based superplasticizers, the apparent viscosity of the cement paste and the amount of
adsorbed superplasticizer to the solid in the paste increased. This suggests that the adsorption hindrance mechanism by ions does not explain the viscosity increase with KF addition
to cement paste. In contrast to the case of K2SO4 addition, the influence of the number of
functional groups on fluidity change by KF addition was small.
4. It is suggested that the cause of the fluidity decrease with KF addition was the preferential adsorption of superplasticizer to some types of fine particles generated by fluoride
ion addition.
AUTHOR BIOS
Kazuki Matsuzawa is a doctoral candidate of Metallurgy and Ceramics Science,
Graduate School of Science and Engineering, Tokyo Institute of Technology, Japan. He
received his BS in 2013 from the Tokyo Institute of Technology. His research interests are
the action mechanisms of chemical admixtures, the fluidity of cement paste, and chemical
reactions in cement.
434SP-302-32
Daiki Atarashi is an Assistant Professor of Metallurgy and Ceramics Science, Graduate

School of Science and Engineering, Tokyo Institute of Technology, Japan. He received
his BS in 2001, MS in 2003 and Dr. Eng. in 2006 from the Tokyo Institute of Technology.
His research interests are the action mechanisms of chemical admixtures, the fluidity of
cement paste and the material design of high-recycled-content and reduced-CO2 emission
cement.
Masahiro Miyauchi is an Associate Professor in Metallurgy and Ceramics Science,
received his PhD. in 2002 from the University of Tokyo. He worked at TOTO Ltd. from
1995 to 2006 and at the National Institute of Advanced Industrial Science and Technology from 2006 to 2011. His research interests include photo-electrochemistry, photocatalysis, solar cells, wet chemical synthesis of nanoparticles, and quantum dots.
Etsuo Sakai is a Professor in Metallurgy and Ceramics Science, Graduate School of
Science and Engineering, Tokyo Institute of Technology, Japan. He received his Dr. Eng.
from the Tokyo Institute of Technology in 1979. His research interests include construction chemistry, material recycling and material design of high-recycled-content and
reduced-CO2 emission cement.
ACKNOWLEDGMENTS
The authors would like to thank NOF Corporation for the synthesis of the polycarboxylate based superplasticizers used in this research.
REFERENCES
1. Sakai, E.; Atarashi, D.; Kawakami, A.; and Daimon, M., Influence of Molecular
structure of Comb-Type Superplasticizers and Inorganic Electrolytes on the Dispersion
Mechanisms of Limestone powder, SP-217, 2003, pp. 381-392.
2. Qian, Z., CO2 abatement in the cement industry, IEA CLEAN COAL CENTRE,
CCC/184 ISBN 978-92-9029-504-4, 78 (2011)
3. Yoshikawa, T., Low-temperature Sintering Technology for Cement Clinker Using
Mineralizers and Fluxes, JOURNAL OF RESEARCH of the TAIHEIYO CEMENT
CORPORATION, No., V. 161, 2011, pp. 66-73. (in Japanese)
4. Klemm, W. A.; Jawed, I.; and Holub, K. J., Effects of calcium fluoride mineralization on silicates and melt formation in portland cement clinker, Cement and Concrete
Research, V. 9, No. 4, 1979, pp. 489-496. doi: 10.1016/0008-8846(79)90046-2
5. Raina, K., and Janakiraman, L. K., Use of mineralizer in black meal process for
improved clinkerization and conservation of energy, Cement and Concrete Research, V.
28, No. 8, 1998, pp. 1093-1099. doi: 10.1016/S0008-8846(98)00082-9
6. Gardeik, H. O., Effect of the clinkering temperature on the specific energy consumption in cement clinker burning, Zement Kalk Gips, V. 34, 1981, pp. 169-174.
7. Sakai, E.; Kawakami, A.; Hamamoto, H.; Honda, S.; Itoh, A.; and Daimon, M.,
Influence of Various Types of Inorganic Salts on Dispersion Mechanisms of Comb-Type
Polymer Containing Graft Polyethylene Oxides Chains, Journal of the Cement Society of
Japan, V. 108, No. 10, 2000, pp. 904-908. doi: (in Japanese)10.2109/jcersj.108.1262_904
8. Matsuzawa, K., Atarashi, D., Miyauchi, M., and Sakai, E., Influence of Sulphate Ion
and Fluoride Ion on the Fluidity of Cement Paste with Polycarboxylate Based Superplasticizer Having Different Molecular Structure, Cement Science and Concrete Technology
Vol. 66, JAPAN CEMENT ASSOCIATION pp.59-64 (2014) (in Japanease)
9. Matsuzawa, K.; Atarashi, D.; Miyauchi, M.; and Sakai, E., Influence of Potassium
Sulphate and Potassium Fluoride on the Fluidity of Cement Paste with Polycarboxylate
Based Superplasticizers Having Different Molecular Structure, The 8th International
Symposium on Cement & Concrete Proceedings, Nanjing/China (2013) (CD-ROM)
10. Atarashi, D., Sakai, E., Honda, S., Itoh, A., and Daimon, M., Adsorption and Dispersion Mechanisms of Comb-type Superplasticizer Containing Grafted Polyethylene Oxide
Chains, Journal of the Ceramic Society of Japan, Supplement 112-1, PacRim5 Special
Issue, 112 [5], S1304-S1307 (2004)
11. Sakai, E.; Kawakami, A.; and Daimon, M., Dispersion mechanisms of comb-type superplasticizers containing grafted poly(ethylene oxide)
chains, Macromolecular Symposia, V. 175, No. 1, 2001, pp. 367-376. doi:
10.1002/1521-3900(200110)175:1<367::AID-MASY367>3.0.CO;2-9
436SP-302-32
SP-302-33
Cement Recycling System Using Sodium

Gluconate
by Daiki Atarashi, Yutaka Aikawa, Yuya Yoda, Masahiro
Miyauchi, and Etsuo Sakai
The objective of this study was to establish a recycling system for cement sludge by hydration control and an evaluation of the residual cement content of sludge. The adsorption
behavior of sodium gluconate on cementitious materials was clarified in this study. The
saturated absorption behavior between sodium gluconate and alite follows a Langmuir
adsorption isotherm.
The delayed hydration time depends on the concentration of residual sodium gluconate. When the residual concentration of sodium gluconate was over 0.0180.020 mass%,
cement hydration did not proceed. The relationship between the heat liberation after 1 day
and the amount of non-hydrated alite as calculated by X-ray diffractometry is linear. The
amount of non-hydrated alite can be estimated by using calorimetric data. Finally, the
amount of non-hydrated alite in sludge water can be estimated by using calorimetric data,
magnesium hydrate hexahydrate (Mg(NO3)26H2O).
Keywords: sludge water; sodium gluconate; set retarder; rate of heat liberation; heat
liberation.
INTRODUCTION
In todays society there is a need to implement low carbon recycling systems and therefore, the reduction and reuse of waste sludge water from the production of ready-mixed
concrete at factories is an important issue. Data from 2006 indicate that ~1 million tons of
sludge are generated annually in Japan.
Sludge is a mixture of cement and very fine aggregate that occurs as a waste at construction sites, or forms when residual recycled concrete is sieved to remove aggregates.
Cleaning water that contains a sludge component, which is referred to as sludge water, is
formed when water is used to clean equipment such as ready-mixed-concrete mixers or
agitators.
To reduce the quantity of sludge formation at ready-mixed-concrete factories, Japanese
Industrial Standard A 5308 Ready-mixed Concrete prescribes a method of using mortar
that adheres to the inside of truck agitator drums. This allows for a solid sludge component
of up to 3% (the mass of the solid sludge component as a percentage of the unit cement
437
438SP-302-33
quantity in the concrete mix) to exist in sludge water when it is used as concrete mixing
water.
However, if sludge water is added, concrete fluidity is reduced and the quantity of water
required to produce the concrete increases. This method is not used widely because of
its high cost and complicated management of sludge water. The number of ready-mixedconcrete factories that use sludge water as concrete mixing water is low and the amount of
industrial waste that they generate after the dewatering process is large.
Sludge water contains fine particles that originate from cement that has already been
hydrated, and from non-hydrated cement and aggregates. By using a set retarder such as
sodium gluconate (GLNa), the hydration reaction of the unhydrated cement in the sludge
water can be controlled (Atarashi et al. 2012). The sludge water can then be reused effectively to replace a portion of the new cement when producing ready-mixed concrete. This
could reduce the required cement production quantity, which would contribute significantly to reductions in CO2 emissions.
A set retarder can delay the hydration of cement within the sludge water. However, few
studies exist of the mechanism of action of the effect of set retarders on cement hydration.
To achieve more widespread recycling of sludge using set retarders, the authors need to
clarify the mechanism of cement hydration control and establish a method to evaluate the
residual cement content of sludge.
It is necessary to establish a method to determine the amount of non-hydrated cement in
sludge water so that this non-hydrated cement can be used as cement. To date, the authors
have found that there is a correlation between the percentage of reacted alite (C3S), which
contributes to the generation of early strength in cement, and the liberated heat of hydration as obtained from a conduction calorimeter (Atarashi et al. 2012, 2013). A prototype
conduction calorimeter that is capable of measuring the heat of hydration of cement has
been produced jointly with Tokyo Riko, a calorimeter-manufacturing company (Sakai et
al. 1997). It is suggested that the measurement of heat using a calorimeter can be used
effectively in cement quality control and inspection (Sakai et al. 2010).
In this paper, the authors discuss the adsorption of GLNa and its set-retarding mechanisms. The recycling of cement sludge is discussed, especially hydration control by GLNa
and a method to evaluate the residual cement content of sludge.
The objective of this research was to formulate a method for the rapid evaluation of
the quantity of cement remaining in sludge water whose hydration reaction has been
suppressed using GLNa. This could enable the effective reuse of residual cement in sludge
water. In the method, the amount of non-reacted cement remaining in the sludge water is
determined using a conduction calorimeter. Furthermore, a method for the rapid evaluation
of the quantity of non-reacted cement in sludge water, for which the hydration has been
suppressed, was developed using a metal salt. The objective was to integrate these research
results to establish a new cement recycling system to use the cement contained in sludge
water effectively.
Our research could enable the effective reuse of residual cement in sludge water. The
amount of unreacted cement remaining in the sludge water is determined using a conduction calorimeter. Furthermore, a method for the rapid evaluation of the quantity of unreacted
Cement Recycling System Using Sodium Gluconate 439
cement in sludge water, for which the hydration has been suppressed, has been developed
using a metal salt. Integration of these research results would make it possible to establish
a new cement recycling system to use the cement contained in the sludge water effectively.
Sample preparationThe water to powder (synthesized C3S and ordinary Portland
cement, OPC) ratio was 0.5 and the mixing time was 10 min. The GLNa dosage was 0, 0.1,
0.2, 0.4, 1.0, and 2.0%.
Hydration reaction of cementThe hydration ratio of C3S or C3A was calculated using
an X-ray diffractometry (XRD) internal standard method.
Adsorption and residual concentration of GLNa in the liquid phaseThe liquid phase
in the paste was obtained by centrifugal suspension. The residual concentration of GLNa
in the liquid phase was measured using a total organic carbon analyzer (TOC-5050A,
Shimadzu). The adsorbed amounts of GLNa were calculated from the concentration of
GLNa in the initial solution and the liquid phase after adsorption testing. A 10-min-adsorption time was used to reach equilibrium.
Measurement of heat liberationThe cumulative heat liberation and rate of heat liberation of C3S and OPC with or without GLNa were measured using a multi-channel conduction (sandwich-type) calorimeter (Tokyo Riko Co. Ltd). The time required for the maximum
heat liberation rate (T1) was estimated and used to characterize the cement hydration.
Measurement of BET specific surface areaAfter 10 min, hydration was stopped by
adding a large amount of acetone and the samples were dried on an aspirator (1.0104 Pa).
The specific surface area of the hydrated cement with various concentrations of GLNa was
measured using N2 gas adsorption.
Determining the residual quantity of cement in sludge waterSimulated sludge was
produced with a water cement ratio (W/C) of 4.0 without the addition of GLNa and magnesium hexahydrate (Mg(NO3)26H2O). It was mixed manually for 3 min at 20C, and then
left for 4 h at 20C. After allowing the hydration reaction to proceed for 1 to 24 h at 20C,
the hydration was stopped, and the quantity of residual cement (C3S) was determined by
comparing the areas under the peaks obtained from XRD.
Sample hydration was stopped for samples that had been hydrated between 0 and 24 h,
and recommenced for 24 h at 20C. The integrated quantity of the heat of hydration liberated from the cement paste was measured using a twin conduction calorimeter (Tokyo
Riko, Tokyo, Japan). By comparing the XRD quantitative method and the calorimetric
data, the validity of the method for determining the quantity of residual C3S in the residual
cement was determined.
Rapid evaluation of the quantity of residual cement in sludge water using metal salt and
a conduction calorimeterThe cement paste (produced using OPC for research purposes)
had a W/C of 4.0, and a GLNa content of 00.2 mass%. Mixing was carried out manually
for 3 min at 20C. GLNa was added after hydration for 1 h at 20C to produce simulated
sludge water in which the hydration was suppressed. Mg(NO3)26H2O was added at 08.0
mass% with respect to the cement, and the mixture was sampled. The heat of hydration
properties of this simulated sludge water were determined using a conduction calorimeter.
440SP-302-33
Table 1 Chemical and mineral compositions and properties of OPC
OPC
SiO2
21.6
C3S
64.4
Chemical compositions (mass %)

Al2O3
Fe2O3
CaO
MgO
Na2O
K2O
5.3
2.3
64.5
2.11
0.25
0.55
Mineral compositions* (mass %)
Blaine
(cm2/g)
C2S
C3A
C4AF
11.6
9.5
8.0
3090
Cl
0.11
SO3
1.8
Density
(g/cm3)
3.17
Calculated by Bogues equation
Materials
C3S was synthesized in the laboratory from industrial raw materials and reagents using
an electric furnace.
The cement used in this study was OPC with properties as listed in Table 1. GLNa
was used as the set retarder, and Mg(NO3)26H2O, which has an accelerating effect on the
retarded cement, was used as the metal salt (Harada 1995, 1996).
Hydration of cement minerals
Fig. 1 shows the hydration reaction ratio of C3S and C3A in the initial hydrated cement.
The hydration ratio of C3S is ~68% at 1 h and no significant change was observed from 1
to 4 h. This phenomenon is explained by the induced mechanism of C3S. C3A has a higher
hydration activity and the reaction ratio of C3A was ~30% at 1 h.
After 4 h of hydration, the hydration reaction ratios of C3A and C3S were ~30 and 10%,
respectively. In other words, 70% of C3A and 90% of C3S still remain as hydration active
cement. The authors could reuse this cement effectively to reduce the environmental
loading.
Adsorption of GLNa on C3S
Fig. 2 shows the isotherm of GLNa on C3S. The amount of adsorbed GLNa on C3S (V)
increased gradually and the adsorption became saturated at ~0.8 mass% residual concentration of GLNa. The relationship between P/V and residual GLNa concentration (P) is
linear (figure omitted, (Atarashi et al. 2012)). The adsorption of GLNa on C3S is thought to
be a Langmuir-type adsorption and the cross-sectional area of a GLNa molecule as calculated from the experimental results was 0.303 nm2 under saturation.
Hydration control of C3S with GLNa
Fig. 3 shows the relationship between the concentration of residual GLNa and the time
of maximum heat liberation rate (T1). The retarded hydration time is also dependent on
the residual GLNa concentration. This result indicates that the retardation time can be
controlled by controlling the residual GLNa concentration. Hydration was retarded for 220
h with a 0.05% residual concentration of GLNa in the liquid phase. The hydration of C3S
was retarded for 40 h with a 0.019% residual concentration of GLNa in the liquid phase.
Fig. 1 Reaction ratio of C3S and C3A during initial cement

hydration
Fig. 2 Adsorption isotherm of GLNa on C3S

Hydration control of OPC with GLNa
The relationship between residual GLNa concentration and the specific surface area of
the hydrated cement after 4 h hydration is shown in Fig. 4. The specific surface area of the
442SP-302-33
Fig. 3 Comparison of T1 of GLNa on C3S

OPC did not change when the residual GLNa in the liquid phase was increased. When the
residual concentration of GLNa was greater than 0.0180.020 mass%, the cement hydration did not proceed.
Development of method to determine the quantity of residual cement in
sludge water using calorimetric measurement
An investigation was carried out to establish a method to determine quantitatively the
amount of residual cement in sludge water using a conduction calorimeter.
Fig. 5 shows the relationship between the quantity of residual (Non-reacted) cement (C3S)
in the sludge water and the cumulative heat of hydration liberated during the 24-h-period
after recommencing the hydration. The cumulative heat of hydration of the non-hydrated
cement is expressed as a relative heat of hydration with the cumulative heat of hydration
in 24 h being 100.
As the quantity of residual cement in the sludge water increased, the relative heat of
hydration in 24 h increased, and a correlation was found between the two.
The conventional quantitative determination of the amount of non-hydrated cement
remaining in cement must be measured using an XRD internal reference method, which
requires skill in measurement and time. The measurement of the heat of hydration as
proposed in this research enables the quantity of residual cement in sludge water to be
estimated simply and in ~24 h.
Fig. 4 Relationship between concentration of residual

GLNa and specific surface area of hydrated cement after 4 h
Fig. 5 Relationship between residual ratio of C3S and relative heat liberation in 24 h
444SP-302-33
Fig. 6 Effect of GLNa and Mg(NO3)26H2O on the heat

liberation of cement hydration
Investigation of rapid method to determine the quantity of cement in sludge
water using magnesium nitrate
If it were possible to estimate the quantity of residual cement in sludge water rapidly, it
may be possible to use the cement in the sludge water from ready-mixed-concrete factories
effectively and thereby reduce the environmental load.
Fig. 6 indicates the example data of effect of GLNa and Mg(NO3)26H2O on the heat
liberation of cement paste.
Hydration is retarded by adding of GLNa. And by using Mg(NO3)26H2O, retardation of
cement hydration is disappeared.
Fig. 7 shows the effect of Mg(NO3)26H2O on the heat of hydration properties of sludge
water to which GLNa has been added.
The time T1, which indicates the maximum rate of heat liberation of plain paste, is ~11
h as shown by the dashed line on the graph. The T1 for plain paste increases significantly
with the addition of GLNa. As reported previously (Song et al. 2004, 2008, Atarashi et al.
2012), this occurs because GLNa is absorbed onto the cement surface thereby suppressing
hydration and making it possible to control the reaction of the cement by controlling the
GLNa dosage. For example, when this dosage is 0.05 and 0.1 mass%, T1 is 14 and 29 h,
respectively, and at 0.15 and 0.20 mass%, T1 is more than 50 h. With Mg(NO3)26H2O
addition, T1 decreases as the dosage increases. For example at a GLNa dosage of 0.1
mass%, when 1.0 and 2.0 mass% Mg(NO3)26H2O are added, the T1 is 15 and 11 h, respectively. The latter value is almost the same as the T1 value of the plain paste without the
addition of GLNa and Mg(NO3)26H2O.
It has therefore been shown that by sampling a portion of sludge water in which hydration has been prevented by GLNa and Mg(NO3)26H2O addition, it is possible to eliminate
the suppression of cement hydration. This rapid method to estimate the amount of nonreacted C3S is required for practical use of sludge water.
Fig. 7 Influence of Mg(NO3)26H2O on T1

CONCLUSIONS
In this paper, the authors discuss the adsorption of GLNa and its set-retarding mechanisms. The authors also discuss the recycling of cement sludge, especially hydration
control by GLNa and a method for the evaluation of the residual cement content of sludge.
A method has been established for the rapid evaluation of the quantity of cement
remaining in sludge water whose hydration reaction has been suppressed using GLNa.
The results are as follows.
The adsorption behavior of GLNa on cementitious materials was clarified. The saturated
absorption behavior between the GLNa and C3S follows a Langmuir adsorption isotherm.
The delayed time of hydration depended on the concentration of residual GLNa. When
the residual concentration of GLNa was over 0.0180.020 mass%, cement hydration did
not proceed.
The relationship between the heat liberation after 1 day and the non-hydrated amount
of C3S as calculated by the XRD method is linear. The authors can estimate the amount of
non-hydrated C3S by using calorimetric data.
Finally, the authors can estimate the amount of non-hydrated C3S rapidly, by using calorimetric data and Mg(NO3)26H2O.
AUTHOR BIOS
Daiki Atarashi is an Assistant Professor in Metallurgy and Ceramics Science at the
received his B.S. in 2001, M.S. in 2003, and Dr. Eng. in 2006 from the Tokyo Institute of
Technology, Tokyo, Japan. His research interests are the action mechanisms of chemical
446SP-302-33
admixtures, the fluidity of cement paste, and the material design of high recycled content
cement and low CO2 emission cement.
Yutaka Aikawa is a Researcher of Metallurgy and Ceramics Science at the Graduate
School of Science and Engineering, Tokyo Institute of Technology, Japan. He received his
Dr. Sci. in 1993 when he worked at Taiyo Yuden Co., Ltd. His research interests include
the theory of void fractions in particle systems and theoretical cement hydration.
Yuya Yoda is a Researcher at Shimizu Co., Ltd., Shimizu Institute of Technology, Center
for Structural and Production Engineering. He received his B.S. in 2009 and M.S. from
Tokyo Institute of Technology, Japan in 2011. His research interests are the relationship
between the heat of hydration and the reaction rate of high-recycled-content cement and
low-CO2-emission cement.
Masahiro Miyauchi is an Associate Professor in Metallurgy and Ceramics Science
at the Graduate School of Science and Engineering, Tokyo Institute of Technology. He
received his B.S. in 1993 and M.S. in 1995 from Tokyo Institute of Technology, Tokyo,
Japan. He received his Ph. D. from the University of Tokyo, Tokyo, Japan in 2002. His
research interests include photo-electrochemistry based on semiconductor nanomaterials
for the reduction of environmental load.
Etsuo Sakai is a Professor in Metallurgy and Ceramics Science at the Graduate School
of Science and Engineering, Tokyo Institute of Technology. He received his Dr. Eng. from
the Tokyo Institute of Technology, Tokyo, Japan in 1979. His research interests include
construction chemistry, material recycling and material design of low-carbon cement.
REFERENCES
Atarashi, D.; Kamio, T.; Aikawa, Y.; Miyauchi, M.; and Sakai, E.2014 , Method for
Estimating Quantity of Non-Hydrated Cement in a Cement Recycling System, Journal of
Advanced Concrete Technology, under review.
Atarashi, D.; Song, Y.; Nishimura, T.; and Sakai, E., Control of Cement Hydration by
Sodium Gluconate in Recycling System, The 10th CANMET/ACI International Conference Superplasticizers and other chemical admixtures in concrete Supplementary Papers,
pp.197-208(2012)
Japan Society of Civil Engineering: The report of Properties and performance estimation
of concrete using admixtures (2007) in Japanese
Japanese Industrial Standard (JIS) A5308 2009: Ready-mixed concrete
J.Nakamoto et.al: Properties of Concrete with Sludge (in Japanese), Cement Sci. and
Concrete Tech., No.53, pp.318-323 (1999)
Sakai, E.; Atarashi, D.; Kawakami, A.; and Daimon, M.2003 , Influence of Molecular
structure of Comb-Type Superplasticizers and Inorganic Electrolytes on the Dispersion
Mechanisms of Limestone powder, ACI, V. SP-217, pp. 381-392.
E. Sakai, E. Maruya, S. Hagiwara and M. Daimon: Material design of cement for
increased waste usage and quality control systems of cement by using of various types of
calorimeter. Cement & Concrete, 756, 48-52. (2010) in Japanese
Sakai, E.; Tsutsumi, K.; and Daimon, M.1997 , Measurement of cement hydration by
means of sandwich type calorimeter, Cement Science & Concrete Technology, V. 51, pp.
68-71. in Japanese
Young, J. F.1972 , A Review of the Mechanisms of Set-Retardation in Portland Cement
Pastes Containing Organic Admixtures, Cement and Concrete Research, V. 2, No. 4, pp.
415-434. doi: 10.1016/0008-8846(72)90057-9
448SP-302-33
SP-302-34
The Influence of Paste Thixotropy on the

Formwork-Filling Properties of Concrete
by Lucia Ferrari and Pascal Boustingorry
Self compacting concretes for precast applications were scaled-down to concrete-equivalent grouts where only the surrounding paste around bigger particles was studied in a
rheometer. Low shear rate steady-state flow curves feature a non-monotonous variation of
shear stress versus shear rate, with a minimum stress obtained for a critical rate. The thixotropy description initially developed by Roussel, Le Roy and Coussot1 was successfully
applied to the data in order to model this behaviour.
Extending the study over different superplasticizers showed that their influence depends
on the molecular architecture.
Specific concrete tests were developed in order to assess workability and formworkfilling retention while applying as little energy as possible to the material. The trends
observed at the rheometer scale were confirmed showing that structure build-up kinetics
has a major influence on concrete placing and that superplasticizers may help control it
up to some extent.
Keywords: thixotropy; self-compacting concrete; superplasticizer; formwork filling.
INTRODUCTION
In the last decade, self compacting concretes (SCCs) have spread widely in the precast
market owing to the ability to cast without vibration. This aspect enabled to reduce human
operations and exposition to noise and other health-related risks. Superplasticizers are
essential to the production of SCCs for which very low yield stresses are desired2,3 but they
should also provide more complex flow characteristics, such as speeding up the filling of
intricate geometries or dense rebar networks. Such features require a thorough control of
rheological properties, including yield stress and plastic viscosity,4 and beyond.
Early on during this research it appeared that common laboratory tools such as Abrams
slump cone, V-Funnel flow time, J-Ring flow or even L-Box tests failed to distinguish
subtle differences between concretes. A rheological approach was then developed for a
more precise insight on the flow properties and specific tools were developed to further
differentiate concrete properties at lab scale, according to customer feedback.
A close look at casting operations indicated that the behaviour at low shear rates was
most significant. Low shear rate flow was shown to be largely influenced by suspension
449
450SP-302-34
Table 1-Cement analysis

Blaine specific surface
4445 cm2/g
Weight % Na2Oeq
0.18
Weight % Na2Oeq
0.87
Table 2Grout composition

Cement g (oz)
196.1 (6.9)
Limestone Filler g
(oz)
89.6 (3.2)
Sand 0/0.160mm
g (oz)
97.8 (3.5)
Sand 0/0.315mm
20.7 (0.7)
Water mL (US
fl oz)
93.6 (3.2)
structure build-up kinetics, often summed-up under the term thixotropy5. This aspect
is the key topic of the present paper, from the scale of a concrete-equivalent grout to full
concrete scale evaluations, both in the laboratory and in the field.
This paper focuses on the often overlooked influence of ageing on the rheological behaviour through the shape of flow curves and on the formwork-filling properties of SCCs. The
influence of superplasticizer nature and dosage is also discussed, and some evidence is
shown that polycarboxylate molecular design allows controlling such properties.
EXPERIMENTAL INVESTIGATION AT THE GROUT SCALE
Materials
The materials used are Portland cement from Port La Nouvelle, Lafarge, the basic characteristics of which are shown in Table 1. Its high specific surface makes it suitable for
precast applications where rapid setting is needed. Limestone filler was supplied from the
FACO company, Vaiges quarry (France). The Millisil C4 0/0.16 mm sand was supplied
by the Sifraco Company (France) and the 0/0.315 mm sand was supplied from Sablires
Palvadeau (France). All admixtures were used as aqueous solution of roughly 20% by
weight of polymer in tap water with the addition of a suitable defoamer.
Equivalent grout mix proportioning
The concrete mix proportioning was scaled down through the use of an approach
inspired by multiscale studies previously published.6 It relies upon applying a cutoff to the
concrete grading curve at an arbitrary particle size in this study, 315 m. After normalizing to 100% passing, a target grading curve is obtained which is then matched as closely
as possible by a blend of the binders and fine sands.
This methodology may be considered as a way to simulate the grout surrounding the
largest aggregates in the concrete, while allowing working in a rheometer where the sample
is sheared in a very small gap, of the order of several millimeters.
The composition of the grout mixture is described in Table 2.
Experimental procedure grout rheology
Water and admixtures were weighed in a Krups YY8506FD mixer bowl, the dry powders
were added during the first 30 seconds of mixing at speed 1 with a leaf-shaped blade. The
The Influence of Paste Thixotropy on the Formwork-Filling Properties of

Concrete451
Table 3-Prestressed SCC mix proportioning
CEM I 52,5R
Cement
kg/m3 (lb/yd3)
350 (590)
Limestone filler
kg/m3 (lb/yd3)
160 (270)
0-4 mm sand
kg/m3 (lb/yd3)
642 (1082)
4-10mm crushed
stone
kg/m3 (lb/yd3)
859 (1448)
8-16 mm
Riverbed stone
kg/m3 (lb/yd3)
114 (192)
Total Water
L/m3 (gal/yd3)
176 (35.5)
mixing speed was increased to speed 7 for 1 minute and then stopped for 30 seconds (to
scrape the sides of the bowl) before applying a last mixing stage of 1 minute at speed 7.
The sample was loaded on the lower plate of a Kinexus Pro rheometer (Malvern Instruments, U.K.) equipped with a serrated parallel plate geometry (1 mm gap). The procedure
started five minutes after the beginning of mixing with a pre-shear at 200 s-1 during one
minute, followed by logarithmic shear rate steps from 200 to 0.1 s-1. Each stress data point
is sampled after the steady state is reached whenever possible in order to build the flow
curve. At the same time minislump tests are performed (cone dimensions: upper diameter
18 mm 0.71 in, lower diameter 36 mm-1.42 in, height 54 mm-2.13 in) using a pneumatic
lifting fork for reproducible results.
After the flow curve measurement, the structure is reset to zero with an oscillating shear
period during one minute with a strain amplitude of 100% and a 1 Hz frequency. Then a
constant stress of 4 Pa is applied to the material in order to observe the structure buildup
close to rest through the increase of viscosity with time. This applied stress value was
chosen to match the stress applied by the weight of the largest aggregate in the system,
according to the following rough calculation7:

gd / 2

10.5
1.
: Specific weight difference between the falling object and the suspended fluid, kg/m3.g:
gravity constant, 9.81 m/s2.d: particle diameter, m
Eq. 1 yields approximately 4 Pa for a diameter of 10 mm, a paste density of 1800 kg/m3
(112.4 lb/ft3) and an aggregate density of 2600 kg/m3 (162.3 lb/ft3).
This section of material testing is called ageing in the rest of the paper.
Experimental procedure Concrete flow
The concrete mix proportioning tested in this study represents a typical composition
of a self compacting concrete for prestressed applications (see Table 3). The cement and
limestone filler are the same as the ones used in the grout-scale study. The aggregates were
supplied from relevant sources across the French territory.
Using a laboratory SKAKO horizontal mixer the sand and the gravel were blended and
pre-soaked with a fraction of the batching water for 5 minutes. The binders were then
added in the dry state. The admixtures were dosed in a pail with the remaining water,
and then the whole was added into the mixer under stirring for 3 minutes and 30 seconds.
In order to keep the water-to-cement ratio constant, the amount of water brought by the
admixtures was substracted from the total amount of water.
General flow properties were assessed with an Abrams cone and a V-shaped funnel for
the slump flow and flow time properties respectively.
452SP-302-34
Given the need for differentiating thixotropy-induced flowability variations at low shear
rate, a robust and simple tool was designed to evaluate such behaviour on concrete. A
proprietary device named below Double-box was built under the form of a rectangular
container separated in two compartments of equal size by a removable gate. One compartment (side 1) is filled with concrete right after the end of mixing and the inner wall is then
lifted 2, 6 or 10 minutes later. The time taken by the concrete to reach the opposite end (side
2) of the box is measured for each resting time and gives a practical evaluation of structure
build-up at rest during the considered time interval. This device differs from the U-box or
U-tube sometimes used for SCC in that its height is much smaller so that a lesser amount
of concrete is used and a lower kinetic energy is imparted to the material.
Roussel et al1,8 proposed a model for taking thixotropy into account in the flow curves of
a suspension. It is based on the assumption that apparent viscosity is linked to a parameter
which measures the structure degree of the suspension as expressed in Eq. 2:

= (1 + n)
2.
: apparent viscosity (Pa.s): viscosity Newtonian plateau at very high shear rates (Pa.s)
: structure coefficient (unitless)n: exponent which quantifies the influence of on the
viscosity (unitless).
The structure parameter is a result of the competition between a time-dependent increase
and a shear rate-dependent decrease and thus obeys a relaxation law expressed by Eq. 3:

d 1
=
dt
3.
: time constant for the structure build-up rate (s).: coefficient describing the efficiency of
shear in breaking up the structure (unitless).
Eq. 3 shows how structure level increases with an assumed constant rate through time if
no shear is applied, whereas shear induces a slow down of increase, or even a decrease of
structure if the applied shear rate is high enough.
At steady state under a constant shear rate Eq. 2 may be rewritten under the form of a
steady state flow curve by simply stating that
d
1
= 0 = eq which yields the steady
dt
state value for the structure degree:

eq =
4.
Eq. 2 then becomes:

n
= 1 + ( )
5.

Concrete453
Fig. 1-Example flow curves from an experiment. (a) Shear stress vs. shear rate, (b) Apparent
viscosity vs. shear rate
Or, as expressed in terms of stress:
n
= = + () 1 n
6.
Eq. 5 or Eq. 6 may be considered as flow curve equations since they describe a unique
relationship between apparent viscosity (resp. stress) and shear rate. They may be used
as models for fitting to experimental data, provided the considered data are obtained at a
steady state.
COMPARISON OF PREDICTIONS AND EXPERIMENTAL RESULTS
Fig. 1a shows an example flow curve from a typical experiment of our study. It shows
that the stress vs. shear rate curve features a non-monotonous shape which departs from the
expected result of a monotonous decrease in stress with a decrease in shear rate.
Quite interestingly, provided that n > 1, Eq. 6 predicts that flow curves plotted as stress
vs. shear rate should feature a non-monotonous trend with a minimum occurring for a
critical shear rate c defined by:

d
d
= 0 = 1 + ()
= c
(1 n) c n )
7.
c: critical shear rate (s-1).

After solving, Eq. 7 eventually yields:

c =
( n 1)1/ n
8.
454SP-302-34
Fig. 2-Transients recorded by the rheometer for two applied shear rates (a) below and (b)
above the critical shear rate
The curve on Fig. 1 presents a critical shear rate around 3 s-1.
Inputting the result of Eq. 8 into Eq. 6 gives a value for the corresponding critical shear
stress:
1
n ( n 1) n
c =

9.
Eq. 8 and Eq. 9 show that the existence of such a critical stress is only compatible with
an exponent n strictly greater than one. If a stress below this critical value is applied to
the material then no steady homogeneous flow may be achieved.1 In this sense the critical
stress may be considered as a form of dynamic yield stress reached at a shear rate which in
practice may be far from zero. Further insight about this question may be found in paper
SP-017 The influence of Superplasticizers on the Flocculation Degree of Cement Suspensions in the present Conference Proceedings.
Correspondingly, trying to force the material into flowing at a shear rate below c may
not result into a homogeneous and steady flow. As a matter of fact simply observing the
transient signals recorded by the rheometer shows that with an applied shear rate below the
critical value the regulation loop is unable to set a steady flow (Fig. 2a) whereas beyond
the critical value a steady state is reached within 5 seconds (Fig. 2b).
As shown in Fig. 1b, a shear-thickening behaviour often appears in our experiments
beyond a shear rate of the order of 10 s-1. As a result the apparent viscosity vs shear rate
curve features a minimum min for a critical shear rate st .This led us to modify Eq. 5 by
adding a dissipation term , following the approach by Hot and Roussel9:

n
= 1 + ( ) +
10.

Concrete455
Fig. 3 - Grout flow curves as a function of superplasticizer dosage. (a) Apparent viscosity
as a function of shear rate, (b) Stress as a function of shear rate.
Eq. 10 was then used for fitting to flow curve data and extracting the parameter values.
Though Eq. 10 slightly departs from Eq. 6 the above discussion about critical stress and
shear rate remains qualitatively valid.
Grout scale - Material response to a change of superplasticizer dosage
In order to check the model response and determine the most relevant parameters, Eq.
10 was fitted to experimental data obtained with an increasing dosage of a superplasticizer.
Fig. 3 shows flow curves obtained for increasing dosages of a common polycarboxylate
superplasticizer. As expected, apparent viscosity and yield stress decreases at all shear rates
when dosage increases.
Fig. 3a stresses out the need of at least an inertial term in the model in order to better
simulate the increase in apparent viscosity beyond a critical shear rate. Fig. 3b shows the
minimum shear stresses occurring at critical shear rates below which no steady-state may
be achieved; consequently Eq. 10 was not fitted to these data according to the approach
described in the paper by Roussel et al.1
Fig. 4 displays the parameter values obtained by the modelling of the flow curves
according to Eq. 10. Increasing the dosage leads to an increase in the product which
means that either structure build-up is slower ( increases) or shear break-up efficiency is
higher ( increases). decreases when superplasticizer dosage increases which implies
that the high shear rate state of the suspension is more deflocculated. The parameter n does
not seem to have a meaningful dependence on the superplasticizer content. increases
with dosage meaning that the inertial dissipation increases according to the deflocculation
degree.
These conclusions allow interpreting the role of a superplasticizer as a means to slow
down thixotropy, increase the efficiency of shear for breaking up the suspension structure,
and reduce the overall aggregation degree.
456SP-302-34
Fig. 4 - Model parameters as a function of superplasticizer dosage after a fit of Eq. 10 to

the data
By fitting the model to a well-known superplasticizer property, this preliminary study
provided a better understanding of the physical meaning of the different parameters and
showed that and are most relevant for the description of low-shear rate flow. For a
lesser influence of thixotropy on flow, a higher and a lower are preferred.
For a further insight into thixotropy, the ageing curves will now be discussed. Fig. 5
shows the time evolution of the apparent viscosity under a constant applied stress of 4 Pa.
The grout response strongly depends on the dosage.
At the lowest dosage the apparent viscosity sharply increases over several orders of
magnitude in less than one minute, as a sign of a very fast stiffening at rest. When dosage
increases, the rate of viscosity increase slows down with an apparent sudden transition
between 0.6 and 0.7% beyond which the grout enters a creep regime. Quite interestingly,
this regime change seems to occur when the ratio between the applied stress 0 and the
critical stress c is higher than one. This is another possible interpretation of the critical
stress: any applied stress above this value will induce some flow whereas the material will
keep stiffening if the applied stress is smaller.
To conclude, the increase in superplasticizer dosage seems to have further effects on the
considered system than merely decreasing yield stress. The observations discussed above

Concrete457
Fig. 5 Influence of superplasticizer dosage on the ageing of grout samples under a

constant probe stress of 0 = 4 Pa.
show that ageing is slowed down by the addition of superplasticizers and that the influence
of ageing onto the shape of flow curves is somewhat mitigated.
Grout scale - Comparison of different superplasticizer formulations at equal
mini-slump flow
The method described above was used to select the most efficient superplasticizers
before testing at the concrete scale. Dosages were adapted in order to obtain a target
mini-slump flow of 105 5 mm; this target value was determined from the slump flow of
the corresponding concrete. Three examples from the study are displayed in Fig. 6, SP1
being a common precast superplasticizer and SP2 and SP3 being two products specifically
designed to reduce thixotropy.
Despite the grouts being prepared at equal mini-slump flow, slight critical stress differences may be observed in Fig. 6b. This was quite unexpected since the general idea is that
flow spread is directly linked to yield stress.10,11 Here, it may show the influence of thixotropy on the slump test itself, which is possibly magnified by a high paste volume fraction
compared to a concrete mixture. Model parameters fitted according to Eq. 10 are shown
in Fig. 7.
SP2 and SP3 seem to induce higher values of , which corresponds to a slower structure
build-up or a higher shear efficiency, both inducing better flow properties. They also have a
similar influence on n while SP1 yields a slightly higher n value. shows the same trend
as the critical stress and SP3 seems to be the most efficient superplasticizer to decrease
this parameter, which confirms its deflocculating efficiency. The hydrodynamic dissipation
term seems to be equivalent for the three measurements. It seems then that SP2 and SP3
are more efficient to prevent thixotropy from interfering with flow.
458SP-302-34
Fig. 6 - Grout flow curves for three different formulations at equal mini-slump flow. (a)
Apparent viscosity as a function of shear rate, (b) Stress as a function of shear rate.
Fig. 7 - Model parameters for the three selected superplasticizers after a fit of Eq. 10 to
the data

Concrete459
Fig. 8 Influence of three different formulations on the

ageing of grout samples under a constant probe stress of 0
= 4 Pa.
Fig. 8 shows the ageing behaviour induced by superplasticizers SP1, SP2 and SP3 right
after the pre-shear stage following the flow curve measurement. SP1 provides a very fast
stiffening, with the apparent viscosity increasing by several orders of magnitude in a matter
of seconds. SP2 and SP3 prevent this sharp increase by keeping the apparent viscosity
below 10 Pa.s during almost 25 minutes before the eventual increase due to hydration. SP3
seems to be able to set a lower viscosity than SP2 despite a slightly lower dosage.
Once again, the influence of the ratio between the applied stress and the critical stress
is shown; With SP1, as 0/c < 1, the grout suddenly stiffens in less than one minute. SP2
and SP3 are able to lower the critical stress below 4 Pa hence 0/c > 1 and the grout
enters a slow-viscosity-increase regime. This is another sign of their ability to slow down
thixotropy.
Concrete scale tests of SP1, SP2 and SP3 presented in the next section perfectly illustrate
the relevance of the above rheological study.
Concrete scale Flow properties after different resting times
SCCs were prepared with the three superplasticizers described above. The dosages were
determined in order to achieve a 700-720mm slump flow with the Abrams cone. Doublebox filling times are displayed in Fig. 9. For a 2-minute resting time, all three products
seem to perform equally. For a 6-minute resting time, SP1 seems less efficient in slowing
down the structure build-up at rest, with results in a slightly longer filling time, while SP2
and SP3 have a similar performance. For a 10-minute resting time SP3 allows achieving
a significantly faster filling after gate lifting, which confirms it is the best admixture for
limiting thixotropic effects.
Compressive strengths were assessed on 15x15x15 cm3 cubes cured at 10C (50F)
during 16h and at 20C (68F) for 24h. Fig. 10 shows that the flow improvement brought
by SP3 does not induce a decrease in strength, even in the harsh conditions of 16h curing
at 10C. In the same conditions SP2 induces a small strength decrease due to its slightly
higher dosage.
460SP-302-34
Fig. 9- Double-box filling times after periods of rest of 2, 6

and 10 min
Fig. 10- Compressive strength at 16h for a curing cycle at 10C and 24h for a curing cycle
at 20C
The concrete-scale study thus confirms the observations at the grout scale, SP3 being the
most efficient superplasticizer when it comes to improve the concrete flow properties, even
after periods of rest.
FURTHER RESEARCH
Given the influence of hydration on the ageing thus the rheology of concrete it may be
inferred that all cases where hydration is accelerated (either by temperature or the use of an
admixture) will be concerned by the notions introduced in the present paper.
Some further investigation beyond the scope of the present study already showed how
civil engineering concretes, where water to cement ratios are quite lower than traditional
ready mix concretes, also feature a certain amount of thixotropy somewhat limiting the
placing of concrete in complex formworks or dense rebar conditions. The application of the
present approach may then spread way beyond precast applications.

Concrete461
It was shown that some superplasticizers are able to slow down the ageing due to cement
hydration while no retardation was observed, at least on early strength in cold conditions.
The fundamental mechanisms involved remain unclear and should be investigated in the
future.
CONCLUSIONS
The present study shows how a scaled-down grout-like material may allow a successful
simulation of the behaviour of full-scale concrete with the beneficial wealth of information
brought by rheometric experiments.
Cement hydration was observed as an ageing mechanism having a noticeable influence
of the rheological behaviour, namely the shape steady state flow curves, which feature a
minimum for a characteristic pair of values of critical stress and critical shear rate.
A theoretical framework designed for the description of thixotropic materials was
successfully applied to the data, allowing a better understanding of the role of superplasticizers in this interplay between ageing and flow. They may be considered not only as
yield-stress-decreasing admixtures, but also as agents able to slow down the ageing due to
hydration, increase the efficiency of shear thus leading to a steeper shear-thinning behaviour and decrease the overall structure level of the cement paste, thus improving the flow
properties of concrete.
The grout-scale observations were confirmed at the concrete scale in laboratory trials. It
is worth noting that these results were extended an applied on numerous jobsites in France,
Spain, Poland and India, leading to the launching of a whole product range.
AUTHOR BIOS
Lucia Ferrari is the Physical Chemistry manager in the main research and development
laboratory of CHRYSO in France. She received her PhD from the Technische Universitt
Mnchen (Germany) after she completed her research work with the EMPA in Dbendorf
(Switzerland) under the supervision of Dr Frank Winnefeld and Pr. Dr. Johann Plank.
Pascal Boustingorry is the Head Manager of the Interface Physical Chemistry Team in
the main research and development laboratory of CHRYSO in France. He received his
PhD from the INP Grenoble and the School of Mines in Saint Etienne (France). Their
main research interests are the interaction of organic molecules with cement suspensions
and the links between superplasticizer chemical architecture and the flow properties of
building materials.
REFERENCES
1. Roussel, N.; Le Roy, R.; and Coussot, P., Thixotropy modelling at local and macroscopic scales, Journal of Non-Newtonian Fluid Mechanics, V. 117, No. 2-3, 2004, pp.
85-95. doi: 10.1016/j.jnnfm.2004.01.001
2. Banfill, P. F. G., The rheology of fresh cement and concrete-a review, Rheology
Reviews, V. 2006, 2006, pp. 61-130.
3. Kauppi, A. et al., Improved superplasticizers for high performance concrete. in Proc.
11th ICCC, Durban 1116 (2003).
462SP-302-34
4. Wallevik, O. H., and Wallevik, J. E., Rheology as a tool in concrete science: The
use of rheographs and workability boxes, Cement and Concrete Research, V. 41, No. 12,
5. Roussel, N.; Ovarlez, G.; Garrault, S.; and Brumaud, C., The origins of thixotropy
of fresh cement pastes, Cement and Concrete Research, V. 42, No. 1, 2011, pp. 148-157.
doi: 10.1016/j.cemconres.2011.09.004
6. Toutou, Z., and Roussel, N., Multi scale experimental study of concrete rheology:
from water scale to gravel scale, Materials and Structures, V. 39, 2006, pp. 167-176.
7. Ovarlez, G. & Coussot, P. Sdimentation dans les fluides seuil en coulement. in
(2011).
8. Roussel, N., A thixotropy model for fresh fluid concretes: theory, validation and
applications, Cement and Concrete Research, V. 36, No. 10, 2006, pp. 1797-1806. doi:
10.1016/j.cemconres.2006.05.025
9. Hot, J., and Roussel, N., Influence of adsorbing polymers on the macroscopic
viscosity of concentrated cement pastes. in Proceedings of the 10th International Conference on Superplasticizers and Other Admixtures in Concrete SP-288, 223233 (American
Concrete Institute, 2012).
10. Saak, A. W.; Jennings, H. M.; and Shah, S. P., A generalized approach for the determination of yield stress by slump and slump flow, Cement and Concrete Research, V. 34,
11. Roussel, N., and Coussot, P., Fifty-cent rheometer for yield stress measurements:
From slump to spreading flow, Journal of Rheology, V. 49, No. 3, 2005, pp. 705-178. doi:
10.1122/1.1879041
SP-302-35
Interaction of Montmorillonite
with Poly(ethylene Glycol) and
Poly(methacrylic Acid) Polymers.
Consequences on the Influence of Clays
on Superplasticizer Efficiency
by Rachid Ait-Akbour, Christine Taviot-Guho, Fabrice
Leroux, Pascal Boustingorry, and Frdric Leising
The interaction of methoxy-capped poly(ethylene glycol) polymers (MPEG) and a
poly(methacrylic acid) anionic polymer (PMA) from water onto sodium Montmorillonite
(Na-Mmt) particles untreated or treated by calcium chloride was studied at 20C. In the
absence of Ca2+, MPEGs are able to intercalate by displacing the water molecules present
in the interlayer space, as shown by XRD and TGA analyses. In contrast, the adsorbed
amount of PMA remains low. The saturation of Mmt with Ca2+ prevents MPEG intercalation through replacing sodium by a stronger water coordinator in the interlayer space, but
slightly increases PMA adsorption possibly through a calcium bonding mechanism.
This was confirmed with PCE superplasticizers and Na- and Ca-saturated Mmt clays.
Whatever the PCE, a larger amount was consumed on Na-Mmt than on Ca-Mmt. This
confirms the occurrence of two consumption mechanisms: (i) a superficial adsorption via
cation bonding of the carboxylate groups with anionic sites on clay surfaces, (ii) intercalation of ether units of the grafts in the interlayer space by displacement of water molecules
coordinated to the exchangeable cations.
Keywords: superplasticizer; clay interaction; interlayer space; exchangeable cations;
coordinated water; polymer intercalation.
INTRODUCTION
Concrete performance is affected by the presence of clay-minerals in aggregates. Clays
are a part of the phyllosilicate family i.e. a family of minerals structured as stacks of silicate
sheets or platelets separated by interlayer gaps. Due to a similar ionic size, silicium ions
may be easily replaced by aluminum ions in the platelets, leading to an electrical charge imbalance which is compensated by the insertion of cations in the interlayer spaces.1 A variety of clay
463
464SP-302-35
Table 1-PCE polymer structures used in the study

Polymer name
PCE-A
PCE-B
PCE-C
Polymer average graft length

(number of monomers)
45
31
114
Polymer grafting ratio

(acid: graft)
4:1
3:2
4:1
Most abundant molecular

weight (Mp) (g/mol)
27000
35000
51000
structures is very well known today, and molecular models were established that describe this
equilibrium state of aluminosilicate sheets separated by cation and water-filled gaps.2-4
These stacks make up particles that possess properties of ion exchange (where the native
cations in the interlayer are prone to be replaced by foreign cations from the surrounding
solution5) or polymer adsorption/intercalation properties, opening way to a range of interesting hybrid materials.6-8
It may be inferred that they may interact with the superplasticizer present in concrete
formulation therefore leading to a reduction in their dispersion efficiency. Solving this
problem requires to unravel the interactions between the superplasticizer molecules and
clay particles. This study focuses on the influence of Sodium-Motmorillonite (Na-Mmt), a
clay which was shown to have an intense detrimental influence on polycarboxylate ether
(PCE) superplasticizers.1,9
The increasing use of low-grade aggregates in the manufacturing of concrete leads to a
more frequent occurrence of clay presence in the mixtures. Clays induce higher superplasticizer dosages, thus an increase in cost, without the usual benefit of longer workability
retention times. This work sheds some light on the interaction mechanisms and a possible
cure is proposed.
Materials
Montmorillonite (Mmt) particles (KSF), specific surface area of 36 m2.g-1 and median
particle size of d50=50 m, were used as received from Sigma Aldrich.
The model molecules used were poly(methacrylic acid) (Mw < 10000 g.mol-1) hereafter called PMA and methoxy-poly(ethylene glycol) series with weight average molecular weights of Mw=750, 2000 and 5000 g.mol-1 and called MPEG750, MPEG2000 and
MPEG5000, respectively.
In a second part, three different PCEs were used denoted hereafter as PCE-A, B and C.
They featured very different molecular structures in terms of number of carboxylate groups,
number and length of ether groups, as shown in Table 1. These PCE were supplied by
CHRYSO (France) and their average molecular weights Mw were determined by Size Exclusion Chromatography (SEC) with a Malvern Viscotek TDA 305 chromatographer (eluent
NaNO3 0,1N with pH buffered at 7, flow 1 ml/min, column temperature 30C) equipped with
a triple detector (Refractive index, Light Scattering and Viscometer). The average molecular
masses ranged between 27000 and 51000g.mol-1 as shown also in Table 1.
Aqueous Mmt solutions were prepared using fresh deionized water and chemical
reagents of p.a. quality. Two sets of experiments were prepared: the ionic strength was
Interaction of Montmorillonite with Poly(ethylene Glycol) and

Poly(methacrylic Acid) Polymers. Consequences on the Influence of Clays
on Superplasticizer Efficiency 465
imposed using either sodium chloride (NaCl (10-2M)) and the pH was kept constant by
adding a small amount of either sodium hydroxide (NaOH) or calcium hydroxide Ca(OH)2
(Carlo Erba [Ca2+]0=2.10-2M and pH0 > 12).
Methods
Adsorption experimentsVarious series of Montmorillonite polymer aqueous dispersions were prepared in order to elucidate the effect of Ca2+ ions on the adsorption isotherms
of model polymers on Montmorillonite. The same experiments were conducted with real
case molecules i.e. PCE-A, B and C
In the first case (no Ca2+ ions), a dispersion of Montmorillonite at 8.5g/l in 10-2 M NaCl
solution at pH=12 was prepared. Appropriate small volumes of polymer stock solution
(20g/l) were added in order to obtain final concentrations ranging from 0 to 1g/l.
In the second set of experiments (with Ca2+ ions), Montmorillonite particles were equilibrated with increasing quantities of calcium hydroxide Ca(OH)2 (initial calcium concentration, [Ca(OH)2]0=2.10-2 M, pH>12) during 24h under gentle magnetic stirring. Subsequently appropriate volumes of polymer stock solution (20g/l) were added in order to
obtain the same range of concentrations (0 to 1 g/l) as in the first set. In all cases dispersions
were stirred by a magnetic stirrer for 20h at 20C (except if indicated).
After sorption, the mixture was centrifuged for 30min at a speed of 10.000 rpm and the
supernatant solution was analysed with a TOC analyzer (TOC 5050A from Shimadzu). The
adsorbed amount of polymer on the mineral particles was calculated from the difference
between the amount of polymer in the liquid before and after contact with the clay.
X-ray diffractionThe basal spacing of the modified Montmorillonites was analyzed by
using CuK (=1.5418) radiation from an automated X-ray diffractometer (XRD) (XPert
Pro Philips). The equipment was operated under a 40kV tension and a 30mA current in a
continuous scan mode. The scanning speed was 0.02/s. X-ray diffraction measurements
were performed over a 2 range of 2.5 -70.
Thermogravimetric analysisThermogravimetric analysis of the modified montmorillonite was carried out using a Setaram Instrumentation equipment (Setsys Evolution) operating at a ramp of 5C/min from room temperature to 1100C under a flowing air atmosphere.
Mortar testingMortar compositions are summarized in Table 2. Two mix designs are
described, with #2 being the same as #1 but with the addition of Na-Mmt. The mixing
procedure was the following.
1. Both sands (Fulchiron and AFNOR) were dry-blended in a Perrier-type planetary
mixer during 30 seconds then 2/3 of the water was added and the wet mixture underwent
another 30-second mixing.
2. The pre-wet sands were left at rest for 4 minutes after which cement and filler were
added and mixed at low speed for one minute.
3. The remaining water with superplasticizer was then added in 30 seconds under low
speed mixing and the wet mixture was homogenized during 90 seconds at the same speed.
4. A 30 second stop allowed scraping the bowl in order to recover the whole material spread
over by the mixer and then an ultimate mixing stage of one minute at high speed was applied.
When Na-Mmt was present in the blend, it was added at the pre-wetting stage of the
sands (stage 1).
466SP-302-35
Table 2-Mortar compositions

Component
Type I Cement (Lafarge Saint Pierre
La Cour, France)
Limestone Filler
0/0.5 mm sand
0/2 mm AFNOR standard sand
0/4 mm unwashed riverbed sand
(Lafarge Aggregates Lillion)
Montmorillonite
Total water
Mix design #1 in g
(oz)
Mix design #2 in g
(oz)
Mix design #3 in g
(oz)
624.9
(22.04)
624.9
(22.04)
624.9
(22.04)
412.1
587.7
1350.0
(14.54)
(20.73)
(47.62)
412.1
587.7
1350.0
(14.54)
(20.73)
(47.62)
412.1
-
(14.54)
(20.73)
-
1946.5
(68.7)
375.1
(13.23)
19.4
375.1
(0.68)
(13.23)
375.1
(13.23)
Fig. 1Polymer consumption isotherms onto Na-Mmt clay

Interaction of MPEGs and PMA with Na-Montmorillonites
Fig. 1 represents adsorption isotherms of MPEG750, 2000, 5000 and PMA polymers
from water onto Na-Mmt particles at ambient temperature, at fixed pH and ionic strength
values (pH=12, NaCl=10-2M).
As for PMA, a non monotonous trend with low adsorption values is observed. A possible
cause is a change in adsorption regime: polycarboxylates are known to provide H-bonding
with clay surfaces10 but beyond a given adsorbed amount electrostatic repulsion between
molecules may take over and establishes a lower adsorbed amount. Complementary
measurements (not included here for lack of space) show that the zeta potential of clay
dramatically decreases towards more negative values when PMA dosage increases.
For all MPEGs, polymer consumption increases with an increase of the polymer initial
concentration and then reaches a maximum value. The maximum adsorption seems to obey
a decreasing logarithmic law with the MPEG molecular weight as shown in Fig. 2, which
shows that the longer the MPEG the fewer the molecules able to interact with this clay.

Fig. 2Relationship between MPEG uptake

saturation onto Na-Mmt and MPEG molecular
weight
Moreover, the affinity of the poly(ethylene glycol) polymers for Mmt surface, as indicated
by the initial slope of the isotherm, increases with molecular weight. The consumption of
those polymers by Mmt particles may be explained by the combination of three mechanisms:
(i) The hydrophobic interaction (between CH2-CH2- groups and siloxane surface). In
fact, the neutral siloxane (-Si-O-Si-) surfaces located on clays with no isomorphous substitution have an overall hydrophobic character.11
(ii) The hydrogen bonding interactions between ether oxygen atoms (Lewis base) and
OH groups (Bronsted acid) associated with aluminol and silanol groups of Mmt.9,12
(iii) A direct interaction between ethylene oxide units and sodium cations in the interlayer space. Some previous work claims the ethylene oxide segments may indeed coordinate around the sodium ions in a crown-ether type chelation.13,14
This means that MPEG750, 2000 and 5000 polymer adsorption may occur partly
through intercalation of poly(ethylene glycols) chains in the interlayer space of Mmt,
promoted by Na+ interlayer cations. This intercalation would cause the displacement and/or
replacement of water molecules forming the hydration shell of Na+ exchangeable cations.
In order to assess polymer intercalation and water replacement by ethylene oxide groups
treated, Mmt particles were analyzed by XRD and TGA. X-ray diffractograms are shown
in Fig. 3 where a noticeable shift of the peak attributed to the (001) plane is observed in
the 4 to 8 2-theta range. This is a result of intercalation of polymers in the interlayer space
and the conversion into distances yields an increase of basal spacing from 1.29 nm to 1.38,
1.59 and 1.86 nm for MPEG molecular weights of 750, 2000 and 5000 g/mol respectively.
A rough volume balance may be computed by taking a specific weight of 2.45g/cm3 (thus
2.45.10-21 g/nm3) for Na-Mmt as claimed on the product datasheet from Sigma Aldrich and
considering that the 4% interlayer water mass replacement (thus 9.8.10-23 g/nm3) measured
above accounts for a proportional volume loss of around 10% (with the specific gravity of
interlayer water taken as 1 g/cm3) whatever the MPEG molecular weight. The respective
net volume increases from XRD are estimated at 7, 23 and 44% yielding a true volume
468SP-302-35
Fig. 3X-ray diffractograms of dried Na-Mmt samples treated with polymer solutions.
Arrows show the peaks attributed to the main [001] basal spacing of the clay
increase by polymer intercalation of 17, 33 and 54% for MPEG molecular weights of 750,
2000 and 5000 respectively.
No noticeable shift was observed with PMA which shows that this polymer only adsorbs
on the surface of clay, if it ever adsorbs at all.
TGA curves are shown in Fig. 4 which shows that reference Na-Mmt undergoes two
main weight loss events. The first event occurs below 150C (302 F) and is related to the
departure of water from the interlayer space, roughly accounting for an 8% mass loss. In
the 400-600C range (752-1112F), transformations of the mineral phase, namely surface
dehydroxylation, lead to a secondary weight loss.
In the presence of MPEGs, a third intense event is observed in the 300-400C range (572752 F), attributed to the calcination of organic matter (interlayer MPEG). Below 150C
(302F), the weight loss is less than 4% which is a proof that MPEG-treated Na-Mmt
contains half less water in the interlayer space than the reference.
The PMA-treated Mmt curve lies very close to the reference curve, proving that no water
substitution in the interlayer space occurs, a confirmation that this polymer only interacts
with clay platelet surface through external adsorption.
To sum up the results, it is expected that only the lateral MPEG grafts of a PCE would
interact by intercalation in the interlayer space, expelling water molecules and replacing
them by ethylene oxide units around sodium cations.
It was previously shown that at least 5 to 6 ethylene oxide units are necessary to promote
interlayer insertion of MPEG, due to the crown-ether-like conformation taken by such
polymers.14 This led to a recent proposition to decrease the length of the lateral grafts
below this value in order to obtain PCEs less sensitive to clays.9

Fig. 4Thermogravimetric analysis of Na-Mmt samples treated with polymer solutions

A different way was chosen here, where sodium ions were exchanged by other cations
such as calcium, the water-binding and basal layer-binding energies of which are much
stronger than those of sodium.
Interaction of MPEGs and PMA with calcium-exchanged Montmorillonite
The measured polymer sorbed amounts onto Mmt with sodium and with calcium are
plotted on Fig. 5. Almost no change in PMA sorbed amount was observed whereas a decrease
in MPEG consumption occurred in the presence of calcium, whatever the MPEG molecular
weight. This decrease is larger for the highest molecular weights i.e. 2000 and 5000 g/mol.
No noticeable increase of the basal spacing was observed, as shown on Fig. 6. Indeed
the peaks of the (001) plane lie at the same 2-theta angle regardless of the polymer weight
which shows that almost no intercalation occurs anymore in the presence of calcium.
TGA curves plotted on Fig. 7 show that Ca-treated Mmt features a more intense weight
loss in the interlayer-related temperature range (below 150C/302F) while a second
weight loss around 600C (1112 F) may be related to dehydroxylation as shown in the
work of Koster van Groos and Guggenheim.15
In the presence of MPEGs, the curves are remarkably close and parallel to the Ca-Mmt
reference with only a slight extra weight loss in the 300400C (572-752F) range coming
from the combustion of polymer. The same combustion, slightly more intense, may be
observed on the PMA-curve in the 400-500C range (752-932 F). However, no decrease
in the interlayer water amount was observed below 200C (392F) which proves that the
above explanation for Na-Mmt intercalation does not stand anymore for polymer sorption
by Ca-Mmt. Surface adsorption may then be involved.
This study shows then that provided the Na-Mmt is exchanged with calcium ions prior
to or even simultaneously with polymer contact, MPEG intercalation is prevented and only
surface adsorption seems to occur.
470SP-302-35
Fig. 5Compared sorbed amounts of polymers on Na-Mmt and Ca-exchanged Mmt
Fig. 6 X-ray diffractograms of dried Ca-Mmt samples treated with polymer solutions.

Fig. 7Thermogravimetric analysis of Na-Mmt and Ca-Mmt samples treated with polymer
solutions
The consequences of this calcium-dependent behaviour were investigated with PCE
comb copolymers, as described in the next section.
Interaction of PCE comb polymers with Montmorillonites
Three PCEs the structures of which are described in Table 1 were tested according to
the protocols above. Polymer sorption onto Na-Mmt is represented on Fig. 8. It shows that
polymer consumption decreases when graft ratio or length increases. Indeed, the lowest
amount was measured for PCE-C bearing the longest grafts whereas PCE-B with short
grafts and a higher grafting ratio features intermediate uptake values. PCE-A with the same
grafting ratio as PCE-C but with shorter grafts, shows the strongest affinity for Na-Mmt.
This may be explained by the combined contribution of simple adsorption onto the platelet
surfaces and graft intercalation. This is quite consistent with the results obtained on the
grafts alone and discussed in the previous section, where MPEG consumption saturation
value decreases with an increase of the molecular weight.
The influence of calcium ions is illustrated in Fig. 9 for all three PCEs where a general
decrease in polymer consumption is observed, which again is consistent with the results
obtained on the grafts. An intense decrease was observed for PCE-C which was expected
since it bears the longest grafts, and these are the most sensitive to the presence of
calcium. X-ray diffractograms on Fig. 10 again bring evidence that intercalation seems to
be prevented by calcium exchange in the interlayer space, leading to the conclusion that
polymer uptake by Ca-Mmt only occurs through surface adsorption.
Investigation of a possible cure protocol
According to the above results, any pre-treatment of a clay-containing aggregates with
a calcium-containing solution should have a beneficial effect on superplasticizer behav-
472SP-302-35
Fig. 8Polymer consumption isotherms of three PCEs onto

Na-Mmt
Fig. 9Influence of calcium presence on the sorbed amounts of PCEs onto Mmt
iour. This hypothesis was checked through mortar trials following the procedure described
above, the mortar mix proportioning being recalled in Table 2.
A solution of calcium acetate was prepared by adding acetic acid to a suspension of
calcium hydroxide until complete dissolution. A subsequent filtration ensured that no
particle remained in the solution the final concentration of which is 20% by weight of
calcium acetate.

Fig. 10X-ray diffractograms of Mmt samples treated with PCE. (a) Na-Mmt, (b) Ca-Mmt
Fig. 11Mortar test results showing the influence of a calcium acetate salt solution on the
performance of PCE-B superplasticizer. The reference mortar was prepared according to
Mix design #1 and the others correspond to Mix design #2.
It was chosen to use this solution either as a co-admixture i.e. by adding it with the superplasticizer in the batching water, or as a pre-treating additive for the aggregate, by spraying
it onto the sand during the pre-wetting stage.
The influence of this calcium solution is demonstrated in Fig. 11 where a constant dosage
of 1,4% PCE-B by weight of total binder was used. The addition of 1% Na-Mmt to sand
had a tremendous effect on PCE efficiency, the initial flow dropping from 315 mm (12.4
474SP-302-35
Fig. 12Mortar test results showing the influence of a calcium acetate salt solution on the
performance of PCE-B superplasticizer. All mortars correspond to Mix design #3.
in) to 205 mm (8.1 in). The subsequent flow measurements could not be carried out beyond
60 minutes due to a sudden decrease in workability.
The addition procedure of the calcium solution (0.2% by weight of total sand) allowed
to increase the initial slump flow almost back to the reference measurement: 270 mm(10.6
in) instead of 315 mm(12.4 in) ; it also retrieves the workability retention curve of the
reference mix.
In another series of experiments, Mix design #3 was used for which the sands were
replaced by a riverbed unwashed sand provided by Lafarge Granulats France. In that case,
the superplasticizer dosage was determined in order to reach a target initial flow. Fig. 12
shows that a superplasticizer dosage decrease from 2.20% to 1.75% was achieved whatever
the addition procedure of the calcium solution. The results are then still valid on natural
sand.
CONCLUSIONS
It was shown that the interaction mechanism between a PCE superplasticizer and sodium
Montmorillonite clay relies on both adsorption onto the platelet surface and poly(ethylene
oxide) graft intercalation between the aluminosilicate sheets. The latter phenomenon is
promoted by a relatively strong interaction between sodium ions and ethylene oxide units
which probably replace water molecules initially present in the coordination layer of the
cations.
Exchanging sodium by calcium ions allowed populating the interlayer space with cations
featuring a stronger interaction energy with water and inducing a much lower, if ever at all,
intercalation of the PCE grafts.
Those results were transposed at the mortar scale, showing that either an improvement of
workability at the same superplasticizer dosage or a decrease in dosage at constant workability may be achieved by adding a calcium acetate solution to the admixture system. This

protocol was efficient either as a pre-treatment of the aggregates minutes before mixing
or as a co-admixture in the batch water. The assumed mechanism was interlayer cation
exchange as well as an adsorption of acetate ions onto the platelet surface competing with
PCE adsorption, which should be checked in any subsequent work.
AUTHOR BIOS
Dr Rachid At-Akbour received his PhD in the Universit Ibnou Zohr in Agadir
(Morocco). He was a post-doctoral student in the Inorganic Materials Team in the
Universit Blaise Pascal in Clermont-Ferrand (France) at the time of the present study.
Pr Christine Taviot-Guho is a Researcher in the Inorganic Materials Team in the
Universit Blaise Pascal in Clermont-Ferrand (France). Dr Fabrice Leroux has a CNRS
senior research position in the Chemical Institute of Clermont-Ferrand and is the Head
of the Inorganic Materials Research Team in the same university. Their professional
interests revolve around Layered Double Hydroxides (LDH) and their interaction with
organics and minerals for the design of new materials with original properties.
Dr Pascal Boustingorry is the Head of the Interface Physical Chemistry Team in the main
research and development laboratory of CHRYSO in France. He received his PhD from
the INP Grenoble along with the Ecole des Mines in Saint Etienne (France). Dr Frdric
Leising received a PhD from the Universit Louis Pasteur in Strasbourg (France) and
led a career in the R&D Departments of Rhne-Poulenc then Rhodia in the fields of
latices and silicon chemistry. He is now an Associated Researcher in the main research
and development laboratory of CHRYSO in France. Their main research interests are the
interaction of organic molecules with cement suspensions and the links between superplasticizer polymer architecture and the flow properties of building materials.
REFERENCES
1. Ng, S., Interactions of Polycarboxylate based Superplasticizers with Montmorillonite
Clay in Portland Cement and with Calcium Aluminate Cement. (Technische Universitt
Mnchen, 2012).
2. Cygan, R. T.; Liang, J. J.; and Kalinichev, A. G., Molecular models of hydroxide,
oxyhydroxide, and clay phases and the development of a general force field, The Journal
of Physical Chemistry B, V. 108, No. 4, 2004, pp. 1255-1266. doi: 10.1021/jp0363287
3. Cygan, R. T.; Greathouse, J. A.; Heinz, H.; and Kalinichev, A. G., Molecular models
and simulations of layered materials, Journal of Materials Chemistry, V. 19, No. 17, 2009,
pp. 2470-2481. doi: 10.1039/b819076c
4. Kalinichev, A. G.; Kirkpatrick, R. J.; and Cygan, R. T., Molecular modeling of
the structure and dynamics of the interlayer and surface species of mixed-metal layered
hydroxides; chloride and water in hydrocalumite (Friedels salt), The American Mineralogist, V. 85, 2000, pp. 1046-1052.
5. Bansal, O. P., and Bansal, V., Influence of time, pH, temperature, organic matter and
exchangeable cations on the adsorption of oxamyl on illites and kaolinites, The Journal of
Agricultural Science, V. 94, No. 03, 1980, pp. 557-563. doi: 10.1017/S0021859600028562
476SP-302-35
6. Plank, J.; Keller, H.; Andres, P. R.; and Dai, Z., Novel organo-mineral phases
obtained by intercalation of maleic anhydrideallyl ether copolymers into layered calcium
aluminum hydrates, Inorganica Chimica Acta, V. 359, No. 15, 2006, pp. 4901-4908. doi:
10.1016/j.ica.2006.08.038
7. Burchill, S. etal., Smectite-polymer interactions in aqueous systems, Clay Minerals,
V. 18, No. 4, 1983, pp. 373-397. doi: 10.1180/claymin.1983.018.4.04
8. Merlin, F.; Lombois, H.; Joly, S.; Lequeux, N.; Halary, J.-L.; and Van Damme, H.,
Cement-polymer and clay-polymer nano- and meso-composites: spotting the differenceBasis of a presentation given at Materials Discussion No. 5, 22???25 September 2002,
Madrid, Spain, Journal of Materials Chemistry, V. 12, No. 11, 2002, pp. 3308-3315. doi:
10.1039/b205279m
9. Ng, S., and Plank, J., Interaction mechanisms between Na montmorillonite clay and
10. Mpofu, P.; Addai-Mensah, J.; and Ralston, J., Investigation of the effect of polymer
structure type on flocculation, rheology and dewatering behaviour of kaolinite dispersions, International Journal of Mineral Processing, V. 71, No. 1-4, 2003, pp. 247-268.
doi: 10.1016/S0301-7516(03)00062-0
11. Charnay, C.; Lagerge, S.; and Partyka, S., Assessment of the surface heterogeneity of talc materials, Journal of Colloid and Interface Science, V. 233, No. 2, 2001, pp.
250-258. doi: 10.1006/jcis.2000.7259
12. Su, C.-C., and Shen, Y.-H., Adsorption of poly (ethylene oxide) on smectite: Effect
of layer charge, Journal of Colloid and Interface Science, V. 332, No. 1, 2009, pp. 11-15.
doi: 10.1016/j.jcis.2008.12.024
13. Aranda, P., and Ruiz-Hitzky, E., Poly (ethylene oxide)-silicate intercalation materials, Chemistry of Materials, V. 4, No. 6, 1992, pp. 1395-1403. doi: 10.1021/cm00024a048
14. Lu, Y.; Kong, S.-T.; Deiseroth, H.-J.; and Mormann, W., Structural Requirements
for the Intercalation of Polyether Polyols into SodiumMontmorillonite: The Role of
Oxyethylene Sequences, Macromolecular Materials and Engineering, V. 293, No. 11,
2008, pp. 900-906. doi: 10.1002/mame.200800155
15. Van Groos, A. K., and Guggenheim, S., Dehydroxylation of Ca-and Mg-exchanged
montmorillonite, The American Mineralogist, V. 74, 1989, pp. 627-636.
SP-302-36
Plasticizing Geopolymer-Type
Auspensions: A Challenge
by L. Nicoleau, M. Pulkin, and T. Mitkina
The composition of the aqueous solution in alkali-activated binders, i.e., the high alkalinity and the high ionic strength challenge chemists to design molecules exhibiting the
same plasticizing effects as in cementitious materials. The highest difficulty probably lies
in alkali-silicate activated systems due to the presence of multivalent silicate oligomers in
solution. Reported here are new insights about the adsorption of polymers in presence of
various concentrated electrolyte solutions in order to mimic the harsh conditions present
in geopolymer pastes. In order to eliminate the problem of the reactivity of such systems,
TiO2 nanoparticles were used as a model substrate. The adsorption of polymer molecules
as well as the specific adsorption of monovalent and divalent ions is revealed. Those results
are compared to the rheological characteristics of alkali-hydroxide or alkali-silicate activated geopolymers. The conclusions which can be drawn from the model system fit qualitatively very well with the classical slump tests done on real systems.
Keywords: geopolymer; high-alkalinity; silicate adsorption.
INTRODUCTION
Geopolymers are alkali-activated alumino-silicates and an interesting alternative to portland cement because of their lower CO2 footprint, their excellent acid and heat resistance
and their low pore-connectivity.1 In spite of these attractive properties, geopolymer-based
materials are still only marginally present in the construction market due to three main
drawbacks, which seriously hamper their use when compared to more traditional concretes.
The first one is the lack of durability investigations, the second one is shrinkage and the last
one, being the subject of this paper, is the poor rheological properties.
Unlike cement, the development of highly efficient plasticizers for geopolymers has
revealed strong difficulties. In this regard, we have recently developed new highly charged
polycondensates which outperform the typical polycarboxylate ethers used in cement
mixtures and which significantly enhance the flowability of alkali-activated geopolymer
systems. Nevertheless, the adsorption of these molecules, i.e. their efficiency as plasticizers, strongly depends on solution conditions. A study of the main relevant parameters in
solution for the adsorption of these polymers is presented in this paper. It is questionable to
try to produce adsorption isotherms on reactive materials like aluminosilicates, hence the
477
478SP-302-36
decision was made to use an unreactive model substrate, such as titanium dioxide (TiO2), in
order to mimic the adsorption on aluminosilicates. TiO2 also offers the advantage that the
calcium, hydroxide and silicate concentrations can be independently varied. The polymer
adsorption, the ion adsorption and the rheological properties of suspensions are discussed,
which lead to a better understanding of the parameters controlling the efficiency of anionic
plasticizers in typical geopolymer solutions.
To the authors best knowledge, no study has been reported so far that investigates in
detail the adsorption behavior of plasticizers on systems representative of geopolymer materials, and the mechanism leading to this adsorption. With this paper, the authors present a
rationale about the factors limiting the adsorption of anionic polymers in such systems, and
why the fluidification of geopolymers represents a challenge for the construction industry.
EXPERIMENTAL SECTION
Materials
A new polycondensation polymer (called as PGP) was used for this study. This polymer
consists of a comb structure with phosphate and carboxylate anchor groups likely to adsorb
on inorganic surfaces. The negative charge density borne by these groups measured by
acid-base titration is 0.003 e-/g (0.085 e-/oz); this charge corresponds to the complete
deprotonation. The effect of this new polymer has been also compared to a regular and
commercial Polycarboxylate Ether provided by BASF and marketed under the name
MasterGlenium 51.
In order to achieve adsorption isotherms, an unreactive model substrate, i.e. which does
not significantly dissolve, has to be used. At the same time, it has to exhibit a representative surface similar to the one of geopolymer raw materials. To ensure these requirements,
aluminosilicates or silica should be avoided since they are too reactive in alkaline solutions. The authors decided to work with TiO2 and in particular with the anatase-rich (90%
anatase-10% rutile) powder P25 (Evonik Industries AG). This powder presents the advantage to be very fine, and thus features a large interface area with the solution. The specific
surface area obtained by the BET method is 60.2 m2/g (18400 ft2/oz). An analysis by ICP
spectrometry reveals that a suspension of this powder does not dissolve any significant
amounts of impurities and in particular negligible amounts of sodium, calcium, potassium,
aluminum or silicon have been detected. As it will be explained below in detail, the surface
charging of TiO2 is similar to that of other alumino-silicate materials, i.e. the zeta potential
evolution over pH is comparable and the overcompensation in presence of calcium ions is
evident.
As the adsorption of charged polymers onto inorganic substrates at high pH is mainly
driven by electrostatics, the first important physical property to be determined is the surface
charge density of the substrate. Upon the increase of pH, the surface of minerals is getting
charged. In the particular case of TiO2, due to the deprotonation of the Ti-OH groups, the
surface is getting negatively charged. These protons can be titrated with NaOH. Without
any additional salt, the charge density is 0.3 e-/nm2 (2.77E16 e-/ft2) and in presence of a
background 1:1 electrolyte (NaCl) around 1.1 e-/nm2 (1.019E17 e-/ft2). The presence of
Plasticizing Geopolymer-Type Auspensions: A Challenge 479
Ca2+ ions favors the deprotonation of the surface and the charge density reaches the, a
priori, maximum value of 2.1 e-/nm2 (1.94E17 e-/ft2).
Methods
All adsorption experiments were performed on the same type of suspension. 0.25g
(0.0088 oz) of TiO2 powder was mixed with 50g (1.76 oz) of solution (CTiO2 = 5 g/L
(0.026 oz/in3)). Samples were prepared with increasing amount of adsorptive species and
stirred for 24 hours at controlled room temperature (23C (73.4F)). Then, suspensions
were ultra-centrifuged and the supernatant was collected to be filtered and analyzed. When
the adsorptive species are ions, the analysis was always done by ICP spectrometry (Ciros
Vision, SPECTRO Analytical Instruments GmbH). In case of the polymer, the analysis
was preferentially done by TOC analysis (TOC II, Elementar Analysensysteme GmbH).
However, this method was not always possible since the solutions are not always stable
after acidification (a necessary step before the analysis). In this case, the determination of
the polymer content in solution was done by ICP by measuring the C content. The calibration was made over a series of standards containing a known quantity of the same polymer,
leading to a minimization of the matrix effects. Nevertheless, it has to be noted that ICP
spectrometry on carbon was not as accurate as the TOC method. In some suspensions,
portlandite was present or precipitated. A series of adsorption experiments on portlandite
have been performed, and it was found that the PGP polymer does not significantly adsorb
on portlandite, thus not causing an interference on the adsorption results on TiO2 (Data not
reported in this paper).
The zeta potential was measured in solutions containing 1g/L (0.000527 oz/in3) of
TiO2 powder with a Zetasizer Nano ZS apparatus (Malvern Instruments Ltd). In order
to vary pH, concentrated NaOH or HCl solutions were titrated. In some cases, different
TiO2 suspensions were prepared over the desired pH-range with additions of Na2SiO3 or
Ca(OH)2 solutions, in order to determine for example the effect of calcium or silicate ions
on surface charging.
Rheology of geopolymer suspensions.
From this investigation, the rheological behavior of different suspensions is reported.
These suspensions consist of two types of raw materials, either blast furnace slag (Table
1a) or fly-ash (Table 1b), and are activated with two types of solutions, either with a KOH
solution or with a metasilicate solution. Though the polymer studied in this paper demonstrates a much higher efficiency than classical polycarboxylate ethers, its plasticizing efficiency varies according to (1) the composition of the geopolymer raw material and (2) the
composition of the activator. Indeed, the plasticizing efficiency decreases when the binder
is a class F fly-ash instead of a slag, and, drops dramatically when an alkali-silicate activator is used instead of KOH.
Adsorption of polymer on TiO2
The adsorption isotherms of PGP are reported in this section. In order to qualitatively
evaluate the efficiency of the molecule as a plasticizer from the adsorption results, a crosshatched zone, corresponding to the required adsorption level for a valuable plasticizing
480SP-302-36
Table 1a - Mortar spread values measured according to the norm DIN EN

1015-3 on blast furnace slag-based mortar activated with KOH or Na2SiO3
solutions. 1% of polymer is added to the mixture. Dosages of activator/
polymer are expressed in weight of activator/dried polymer by weight of
binder. The water to binder ratio is 0.5.
Admixture
none
Polycarboxylate Ether
PGP
7% KOH
19 cm (7.5 in)
20 cm (7.9 in)
>30 cm (>11.8 in)
7% Na2SiO3
15 cm (5.9 in)
15 cm (5.9 in)
16 cm (6.3 in)
Table 1b - Mortar spread measured on low calcium fly ash-based mortars

(class F in European classification). Same proportions are used as in Tab.
1a.
Admixture
none
Polycarboxylate Ether
PGP
7% KOH
19 cm (7.5 in)
19 cm (7.5 in)
27 cm (10.6 in)
7% Na2SiO3
17 cm (6.7 in)
17 cm (6.7 in)
17 cm (6.7 in)
effect, is drawn. This zone is estimated from adsorption curves realized with typical polycarboxylate ethers in cementitious systems and by comparison of the respective surface
areas between TiO2 and cements. In addition, a dotted straight-line is given and represents
a degree of adsorption equal to 50%. First, the results relative to the adsorption of PGP in
different alkaline conditions are presented in Fig. 1. There is a very low adsorption of PGP
at low pH (water) and at relatively low ionic strength (100 mM NaOH). The adsorption is
slightly increased when 400 mM of NaCl is added to 100 mM NaOH or when the particles
are mixed in 500 mM NaOH. Even if the critical level of adsorption may be reached, it
requires substantially more polymer than typically for cement mixtures and the adsorption
degree is only 10% which reveals a low affinity for the TiO2 surface in these conditions.
Similarly to the addition of 400 mM NaCl in 100 mM NaOH (Fig. 1), the addition of low
amounts of calcium chloride favors the adsorption of PGP (Fig. 2a). Under condition of a
saturated solution with calcium ions, which can be realized for instance using a reservoir
of portlandite (Ca(OH)2), the adsorption is drastically enhanced (Fig. 2b). It turns out that
calcium ions likely mediate the adsorption of PGP as it has been already reported in many
works related to calcium-rich systems (like cements).2 The difference of effect between
the CaCl2 and the Ca(OH)2(sat) experiments is the saturation in calcium. Indeed, assuming
also a complex formed with calcium and polymer molecules, the concentration of free
calcium significantly decreases upon the addition of polymer. It thus leads to the defection
of calcium close to TiO2 surface and calcium ions cannot mediate the adsorption anymore.
Such an effect was already shown elsewhere.2,3 Further support for this hypothesis is found
looking at the curve representing the addition of 5 mM CaCl2 (Fig. 2a). At the concentration of PGP above 4500 mg/L (0.002372 oz/in3), the adsorption slightly decreased as
indicated by the bending of the curve (marked with an arrow). The charge density of this
polymer is about 3 meq/g (85 meq/oz), 4500 mg (0.158 oz) or 1L of 4500 mg/L solution
of this polymer can therefore complex about 6.75 mmol of Ca2+ assuming that 100% of
the polymer charges are counter-balanced only by Ca2+. It is typically in the same order
Figure 1 - Adsorption isotherm of PGP onto TiO2 in presence of different salts.
Figure 2 - (a) Adsorption isotherm of PGP onto TiO2 in equilibrium with 100mM NaOH
and in presence of different concentrations of calcium chloride. (b) Adsorption isotherm of
PGP onto TiO2 in equilibrium with Portlandite and in presence of different salts.
of magnitude as the initial calcium concentration (5 mM) used in this case. A competition
between the complexation of calcium by the polymer in solution and the adsorption of
calcium onto the mineral surface can be assumed. The addition of NaCl does not alter the
adsorption of polymer in saturated conditions with respect to portlandite, but the high pH
conditions (500 mM NaOH) do. In 500 mM NaOH, the solubility of portlandite is about 1
mmol/L of Ca2+ which is insufficient to saturate the polymer molecules and the surface of
TiO2 with calcium. Nevertheless, even at such a low concentration of Ca2+, it seems to be
possible to reach the critical level of adsorption with high polymer concentration.
Unfortunately, the measurement of adsorption isotherms in solutions combining calcium
and silicate ions is not reliable because of the possible precipitation of calcium silicate hydrates, especially after 24 hours equilibrium time. Also, based on the adsorption
isotherms performed so far, the main conclusion is that without calcium ions the adsorption of polymer is difficult. Calcium ions seem to be the main driver for the adsorption.
482SP-302-36
Figure 3 - Adsorption of calcium ions on TiO2 at different

pHs. pHs are maintained constant at pH=11 and pH=12 and
considered as constant in 100 mM and 500 mM of NaOH,
the pH is 12.91 and 13.48 respectively.
Without calcium ions, no significant difference is seen between polymer adsorptions in
alkali-hydroxide or in alkali-silicate activated solutions.
Adsorption of anions and cations on TiO2
1) Ca2+Inspired by their key-role in cementitious systems and in the adsorption of
PGP, the adsorption of Ca2+ onto TiO2 particles was investigated and is reported in the
following section. Fig. 3 shows the adsorption of Ca2+ at different pHs. The vertical adsorption slope at very low Ca concentrations reveals that the first calcium ions added to the
solution are immediately adsorbed. At pH=11, all Ca ions are adsorbed up to the coverage
of 1 Ca2+ per nm2 of TiO2. Above pH=11, the coverage of strongly adsorbed Ca2+ ions
increases up to 1.5 Ca2+ per nm2 and over the whole alkaline pH range, the maximum
Ca2+ coverage is about 2 Ca2+ per nm2 which is achieved when [Ca2+] = 8 mM at pH=11
and [Ca2+] = 1.5 mM at pH=13. It means that one calcium ion is complexed by one TiO2
negative site, which should lead to an overcompensation value equal to the absolute charge
density. In turn, the electroneutrality is ensured by hydroxyl ions, the only anions present
here. It is worthwhile to point out the significant variation of the calcium concentration
next to the surface at low calcium concentration range.
The concentration of calcium present in the solution can be a prime performance parameter. Actually, in such concentrated salt solutions, representative of geopolymer systems,
one should not refer to the concentration but rather to ion activity. Other ions and especially the anions strongly modify the activity of calcium. In order to estimate the possible
influence of these anions, different concentrations of sodium sulfate were added in the
equilibrium solution and the calcium adsorption isotherms produced. The sulfate ions are
divalent anions and form ion pairs with Ca2+ decreasing therefore the activity of Ca2+. It
would have been difficult to use another polyvalent anion since all of them precipitate with
calcium already at very low calcium concentration and it would have hinder the adsorption
Figure 4 - Adsorption isotherms of calcium ions on TiO2 in 100 mM NaOH and in presence of different concentration of sodium sulfate, in function of (a) the calcium concentration in solution and (b) in function of the activity in Ca2+.
Figure 5 - Adsorption isotherms of silicate ions on TiO2 at

different pHs.
study on a meaningful calcium concentration range. The results with sulfates are shown in
Fig. 4a. The presence of SO42- ions does not drastically change the adsorption of calcium,
except at very low calcium concentration where a slight decrease is observed. This effect
has almost vanished when the activity of Ca2+ is considered (Fig. 4b)
2) H3SiO4-/H2SiO42-Other important anions in geopolymer systems are the silicate ions
as concentrated silicate solutions (>1 M), which can also be used for the activation of
aluminosilicate binders. In high alkaline conditions, the silicate ions are present as monomeric species once and twice deprotonated at low concentration of Si and as oligomers
at higher concentrations of Si. It is assumed in this study that the adsorption of monomer
species is representative of the adsorption of all silicate oligomers. The adsorption of silicate ions at different pHs is presented in Fig. 5. Silicate ions, negatively charged, adsorb on
TiO2 also negatively charged in this pH range. This adsorption seems to be little depending
484SP-302-36
Figure 6 - Zeta-potential evolution of TiO2 particles over pH

in presence of different amounts of NaCl. The pH is given by
addition of concentrated NaOH or HNO3 solutions.
of the alkaline pH conditions. It would indicate a weak influence of electrostatics in this
case.
In silicate activated geopolymers, the concentration of silicate in solution is often higher
than 1 M, i.e. higher than the concentration used for the adsorption isotherms in this study.
Presuming that polymers and silicates can adsorb on the same TiO2 surface sites, the
adsorption results above indicate a possible competition between both. This behavior is
akin to the role of sulfate in cement paste with respect to the adsorption of polycarboxylate
ethers.
Electrokinetic measurements
In order to corroborate the adsorption results, electrokinetic measurements have been
carried out. These ones have the advantage to be easier and faster than the adsorption
studies but have a disadvantage of stronger artifacts related to the method and the calculus
of zeta-potential. In addition, it has to be pointed out that the technique measures the potential resulting from the surface charge of the particles plus the ions closely adsorb next to the
particle. The definition of the ion cloud next to the surface which moves with the particle is
not uniquely defined, and is a source of possible different interpretations.
1) Influence of ionic strengthA first series was carried out in different salt concentrations (Fig. 6) and illustrates well the purpose and the difficulty to get unique interpretation of the results. In the salt-free solution, the isoelectric point of TiO2 is found to be at
pH=6 which appears to be a reasonable value4 for the anatase polymorph. The absolute
zeta plateau values at low pH and high pH are quite similar. This indicates that, without
additional salt, the number of sites being extra protonated and the number of sites being
once deprotonated are similar and most probably the same considering the TiO2 surface.
It is also clear that the addition of NaCl decreases the absolute zeta values at high pHs.
This can be explained by two phenomena: (1) the specific adsorption of sodium (Na+) on
the surface and (2) the so-called breaking forces depending on the ionic strength of the
Figure 7 - Zeta-potential evolution of TiO2 particles over pH,

when the pH is given by addition of NaOH or by addition of
Ca(OH)2 or by addition of NaOH in a solution containing
2mM of CaCl2.
solution which limits the displacement of the particles. Though quantitative description of
both phenomena is not possible from solely electrokinetic measurements, some details of
the curve help us to suggest qualitative statements. The isoelectric point is shifted to the
higher pH with the addition of NaCl which indicates that sodium adsorbs. However, the
shift is not proportional to the addition of salt. The absolute value of the pseudo plateau at
high pH is decreasing with the increasing salt concentration, which can be attributed to the
screening effect or to the adsorption of sodium ions. This variation is lower at very acidic
pHs, i.e. the addition of salt influences only a little the positive zeta value due probably to
the sole screening effect. This indicates that the variation at high pH is mostly due to the
adsorption of sodium ions. At a given salt concentration, it seems also that the absolute
zeta value is decreasing at very high pH (>12). It could be associated to the increase of
ionic strength, i.e. a salt screening effect, but, it even leads sometime to a reversal of the
zeta values (zeta>0) which cannot be attributed to a screening effect. The conclusion of
this simple examination is that alkali ions adsorb but no qualitative information is given
(values are not proportional to the charge next to the surface because of the screening
effect).
2) Addition of Ca ionsAs previously highlighted, calcium ions strongly adsorb on
TiO2 at high pH. This phenomenon can be also revealed by electrokinetic measurements.
Instead of an addition of NaOH to increase the pH, a solution saturated with respect to
the portlandite is used ([Ca] ~ 22 mM at 25C) and this way the pH and the calcium
concentration simultaneously increase without adding any alkali. It leads to a reversal of
the zeta potential at pH=10 as the curve with squares in Fig. 7 shows. Another manner is
to add certain amount of Ca salt and to increase step-wise the pH with NaOH additions. At
pH=10-11, the addition of 2 mM CaCl2 is sufficient to fully reverse the zeta potential, i.e.
~ -30mV without salt and ~ 30 mV with 2 mM CaCl2. It thus indicates that about one
Ca2+ is bound to one negative site of TiO2. When pH further increases to pH = 12-13 the
zeta potential decreases. Such a phenomenon is not observed with Ca(OH)2 solutions, and
486SP-302-36
Figure 8 - Zeta-potential evolution of TiO2 particles over

pH when the pH of the solution results of addition of NaOH
or Na2SiO3.
is therefore due to the presence of sodium ions added with NaOH. Sodium ions compete
with calcium for the site adsorption, if one Na+ replaces one Ca2+ the positive charge next
to the surface decreases.
3) Influence of silicate ionsThe same approach has been applied to silicates ions. The
results presented in Fig. 8 reveal undoubtedly that the silicate ions are adsorbed on the
surface of TiO2. The zeta potential is even more negative when pH is given by the step-wise
addition of a metasilicate solution (curve with squares) than by the usual addition of NaOH
(curve with triangles). The authors recall that even more sodium ions are added with the
silicate than with NaOH in order to achieve a same pH. With the addition of NaOH in a
solution containing only 1mM Na2SiO3, the zeta values are also more negative, indicating
also an adsorption of silicate ions even at low concentrations of silicate.
The influence of calcium ions on the silicate adsorption was also investigated. The
precipitation of C-S-H has to be avoided and it forced us to work at low concentration of
calcium. Fig. 9 shows that the presence of silicate ions clearly influences the zeta potential
leading to negative values even if calcium ions are in solution. From these sole results, it is
difficult to conclude whether the silicate ions yield desorption of calcium or whether they
adsorb.
Thermodynamic calculations
Calcium ions play a key role in the adsorption of anionic polymers. Whereas their
influence in alkali-hydroxide solutions is readily measurable, the same measurements in
alkali-silicate solution are more prone to interference since the formation of calcium silicate hydrates is possible and can thus produce erroneous results. In typical alkali-silicate
activated geopolymers, the silicate concentration in solution is very high (>1M) and the
supersaturation with respect to calcium silicate hydrates is rapidly reached as soon as
some calcium is released by the dissolution of geopolymer raw materials. In addition,
although poorly defined and never experimentally evidenced, the formation of CaH2SiO4,
CaH3SiO4+ and CaHSiO4- ion pairs can be assumed.5 Both factors decrease the activity of
Figure 9 - Zeta-potential evolution of TiO2 particles over

pH in presence of calcium chloride. The pH is given with
Na2SiO3 additions.
Ca2+ which is the adsorption mediating agent for anionic polymers. In order to estimate
this decrease, thermodynamic calculations with the software Phreeqc using the DebyeHckels approximation for the calculus of activities were carried out. Alkali-hydroxide
or alkali-silicate solutions containing a fixed calcium concentration are put in equilibrium
with calcium silicate hydrates. Equilibrium constants of calcium-silicate ion pairs are not
known and therefore not considered in the calculus. It also means that this calculus maximizes the activity of Ca2+ since any other complexes involving calcium would further
decrease the Ca2+ activity. A starting calcium concentration of 2 mM is chosen and the
results are reported in Table 2. It can be concluded that, either for a same pH or for a same
concentration of activator, the activity of Ca2+ is drastically reduced when a metasilicate
activator is used instead of a pure alkali-hydroxide activator. This simple calculus indicates
that the activity of calcium in solution and thus its adsorption on any mineral surface has
to severely decrease in presence of silicates which should clearly disfavor the adsorption
of anionic polymers.
CONCLUSIONS
The experimental observations related to the adsorption of PGP are summarized below:
1) The surface of TiO2 is highly negatively charged and at high pH, up to one calcium ion
is adsorbed per surface charge, which leads to a large overcompensation next to the surface
if hydroxide ions are the co-anions.
2) In the presence of only Na+ ions, a low adsorption of PGP is exhibited, but nevertheless, the higher the concentration of alkali ions, the greater the adsorption of polymer. The
adsorption of PGP is getting higher by addition of calcium, but it is strongly depending of
the concentration of free Ca2+ ions. With a reservoir of calcium, the adsorption of polymer
is high.
3) The electrokinetic measurements indicate that in presence of silicate ions, the charge
next to the surface is negatively charged, even in presence of calcium. There are two
possible explanations: either the calcium desorbed in presence of silicate ions, or the silicate ions are adsorbed on top of calcium ions.
488SP-302-36
Table 2 - Evolution of the activity of OH- and Ca2+ in different activator solutions when 2 mmol/L of calcium is added to a solution in equilibrium with
respect to C-S-H.
NaOH
NaOH
Na2SiO3
Na2SiO3
Concentration
[mmol/L]
500
1000
500
1300
[Ca2+]
[mmol/L]
2
2
2
2
pH
13.46
13.71
13.04
13.46
(OH-)
2.87E-01
4.97E-01
1.06E-01
2.71E-01
(Ca2+)
1.94E-04
3.07E-05
6.01E-08
3.81E-09
4) The activity of calcium in solution drastically decreases in the presence of solubilized

silicates, either due to the precipitation of calcium silicate hydrates and/or to the likely
formation of calcium silicate ions.
The results presented in this paper elucidate the difficulty to obtain a sufficient polymer
adsorption, and thus a reasonable fluidification, in geopolymer systems. All factors influencing the calcium concentration at the interface between the mineral and polymer molecules are of prime importance. First, the dissolution of the various raw materials release
more or less calcium according to their composition: class F fly-ashes deliver less calcium
than blast furnace slags and are potentially less favorable substrates for anionic polymers.
Second, the activity of Ca2+ decreases with the concentration of activator, and strongly
drops in presence of silicates: silicate solutions prevent the adsorption compared to pure
soda or potash solutions. Third, silicate ions, and in particular divalent H2SiO42- ions,
adsorb on minerals, meaning that there is a potential competition between silicate and
polymer which still makes the polymer adsorption in alkali-silicate activated systems more
difficult. Taking into account this competition and the low concentration of Ca2+ in solution, polymers bearing a lot of negative charges, and especially anchor groups having a
high affinity for calcium, are preferential candidates as anionic plasticizers for alkali-activated geopolymer systems.
AUTHOR BIOS
Luc Nicoleau is Senior Scientist within BASF Advanced Materials & Systems Research
division, working at the global research center for construction of BASF. He received
his PhD in 2004 from the University of Dijon. His research interests include the interactions cement-admixtures, the hydration of cement and the alkali-silica reaction. He was
the main developer of a new accelerator technology based on suspended C-S-H nuclei:
Master XSEED100.
Maxim Pulkin is a researcher within BASF Advanced Materials & Systems Research
division. He received his PhD in 2011 from the University of Bremen (Germany). His
research interests are in area of materials science and currently include the interactions
between inorganic particles and polymers as well as the chemistry and properties of
alternative (non-OPC) inorganic binders.
Tatiana Mitkina is a researcher within BASF Advanced Materials & Systems Research
division. She received her PhD in 2008 from the Nikolaev Institute of Inorganic Chem-
istry SB RAS (Novosibirsk, Russia). Her current research interests are focused on development of admixtures for alternative inorganic binders, in particular for alkali activated
binders, and, on additives for mining applications.
REFERENCES
1. Geopolymers Structure, processing, properties and industrial applications, Edited by
J.L. Provis & J.S.J. van Deventer, Woodhead Publishing Limited and CRC Press LLC
Cambridge 2009
2. Turesson, M.; Labbez, C.; and Nonat, A., Calcium mediated polyelectrolyte adsorption on like-charged surfaces, Langmuir, V. 27, No. 22, 2011, pp. 13572-13581. doi:
10.1021/la2030846
3. Flood, C.; Cosgrove, T.; Espidel, Y.; Howell, I.; and Revell, P., Sodium polyacrylate
onto anionic and cationic silica in the presence of salts, Langmuir, V. 23, No. 11, 2007,
pp. 6191-6197. doi: 10.1021/la070047z
4. Bourikas, K.; Hiemstra, T.; and van Riemsdijk, W. H., Ion Pair Formation and
Primary Charging Behavior of Titanium Oxide (Anatase and Rutile), Langmuir, V. 17,
No. 3, 2001, pp. 749-756. doi: 10.1021/la000806c
5. Taylor, H. F. W., Cement Chemistry 2nd Edition, Thomas Telford edition, 1997
490SP-302-36
SP-302-37
Admixture Concepts for the Sub-Saharan

African Environment with Indigenous Raw
Materials
by Wolfram Schmidt, Nsesheye S. Msinjili, Herbert C.
Uzoegbo, and John K. Makunza
The economic use of chemical admixtures depends on supply chains. Therefore, in most
regions ins sub-Saharan Africa (SSA), the use of admixtures is not common practice. This
amplifies the unfavorable framework for concrete construction such as fragmentary supply
chains, high local cement prices, and unfavorable construction site facilities in this region
significantly. The use of superplasticizer (SP) and stabilizing agents (STA) can enhance the
concrete technology in SSA, since they can disassociate the concrete quality from external
boundary influences. After providing a general overview of the peculiarities of the SSA
boundary framework, economic concepts are provided, how existing material solutions can
be significantly improved by the use of SPs and STAs based on locally available materials
such as lignosulphonates and cassava starch. Finally a three step optimization process is
described that helps developing flowable concrete based on materials that can be accessed
in most locations in SSA.
Keywords: admixtures; lignosulphonate; polycarboxylate ether; robustness; selfcompacting concrete; starch ether; sub-Saharan Africa.
INTRODUCTION
The sub-Saharan African environment for concrete casting varies greatly from what can
be found in most industrialized countries of the northern hemisphere. For instance, the
price of cement in SSA ranges between 8 - 16 USD per 50 kg (110 lbs) bag, which is the
standard delivery. Regionally, the sale prices can be relatively higher depending mainly
upon the logistic effort. If these prices are related to the gross national income, identical
volumes of cement can be several hundred times more expensive in SSA than in Europe or
Northern America.1,2 Considering that labor costs are extremely low in SSA, it is obvious
that the basis for constructional cost calculation is exactly the opposite of Europe, Northern
America, or Japan.3 Hence, it is worthwhile for engineers in SSA not to follow the existing
trend of concrete regulations in Europe or Northern America but to develop customized
concepts.
491
492SP-302-37
The current situation for casting concrete in the majority of SSA is done on-site. Pre-cast
concrete or ready-mix concrete is uncommon outside of South Africa. The disadvantageous
consequences are often the lack of quality control and homogeneity of the final product.
This is often accompanied by unsteady supply chains and expensive thus often unaffordable rental fees for adequate equipment, which often does not belong to the contractor.
Most processes (especially concrete mixing) are conducted manually, and often the only
possibility to adjust an appropriate consistency and workability retention is by addition
of water on haptic and visual assessment. Nevertheless, the advantage of having on-site
concrete casting in SSA is the availability of cheap labor. Therefore, the solution is not to
further incorporate pre-cast or ready-mix concrete, but rather to take advantage of what is
readily available in the market, i.e. local available materials and cheap labor.
Solutions for the casting of concrete in sub-Saharan Africa
The existing situation in SSA as described before adversely affects concrete performance
and durability. In order to enhance the situation, it is desirable to exclude as many factors
as possible that result in a negative impact on concrete casting. A similar consideration
was the trigger for the invention of self-compacting concrete (SCC) in the 1980s in Japan.
Some leading scientists observed a correlation between the decreasing skills of laborers
and decreasing structural durability.4-6 They found that the only solution to counteract this
trend was a durable concrete that performs reliably regardless of the laborers skills.
Although Japan and SSA have greatly differing characteristic in terms of their geographic
and economic conditions, the situation that existed in Japan before the concept of SCC,
is somewhat comparable to the current situation in SSA. The difference is that beyond
the laborers skill level SSA exhibits additional disadvantageous influencing parameters
on construction sites. Furthermore, SSA exhibits a clear necessity to save costs, and the
construction needs in SSA are more related to low-cost housing, multi-story buildings and
infrastructure rather than to skyscraper towers. Therefore, if SCC is opted as a solution in
SSA, it is important to discuss about tailored concepts for the local boundary framework.
Flowable concrete solutions for high temperatures
Due to its accelerating effect on the hydration, hot climate conditions cause rapid loss
of workability.7,8 Short workability retention is disadvantageous for construction sites in
SSA due to the unsteady casting logistics caused by unaffordable equipment such as pumps
and machinery. The major influencing factors of flowable concrete that determine the flow
retention are the charge density of the SP and the water to powder ratio (w/p).7,8 Apart from
that the use of a set retarder can significantly enhance the workability retention.
Realistically, a SP-based solution to customize flowable concrete for high temperatures
can only be achieved by using polycarboxylate ether (PCE) due to the high versatility. In
this case, it is beneficial to use low charge agents, since they provide better workability
retention due to their lower adsorption tendency on cement hydration phases. Higher
charge SPs are adsorbed on surfaces rapidly, but in return become inefficient rapidly, due
to morphologic changes of the particles caused by the high temperature induced rapid
hydration reaction. Low charge SPs adsorb at a lesser degree initially, causing a higher
polymer content in the solution which allows for ongoing adsorption upon growth of new
hydrates with time (Fig. 1).7-9
Admixture Concepts for the Sub-Saharan African Environment with

Indigenous Raw Materials 493
Fig. 1Approaches to obtain workability retention for concrete at high temperatures.

Since PCE is hard to purchase outside of the country of South Africa, a temperature
optimization based on the polymeric architecture is a less feasible solution for the moment.
Therefore, flowable concrete concepts need higher w/p values. As illustrated in Fig. 1,
low w/p mixtures stiffen quickly due to rapidly occurring morphological changes on
the surfaces that cause lower mobility of particles.10 Consequently, SCC optimized for
hot weather conditions should not be designed for ultimately low w/p values (w/p ratios
between 0.5 and 0.65 might be realistic).8 In this case, good flow properties can also be
achieved with SPs based on lignosulfonates (LS) or poly naphthalene sulphonates (PNS).
However, the higher w/p may also induce a tendency to segregation, which may need to
be compensated by STA, which in return adds a sophisticated component into the mixture
composition.
Applicable solutions for flowable concrete based on local raw materials
Major concerns about flowable concrete mixtures in SSA are the affordability, the availability of construction chemicals, as well as how to bring them into practice in an environment, which is not optimized for robust concrete proportioning and casting.
Since cement is the most cost-driving factor in SSA at the moment, any method that helps
saving cement helps reducing material costs. Appropriate mixture composition with optimized particle packing help significantly in reducing the costs. Furthermore an economically reasonable implementation can only be conducted by incorporating locally available
materials such as rice husk or bagasse ashes, limestone fillers or natural pozzolans.
494SP-302-37
Table 1Raw materials and admixture solutions for sub-Saharan African

concrete
Constituent
Cement extenders
Typically used in SCC

Limestone filler, groung granulated blast
furnace slag, fly ash
Fillers
Limestone filler, fly ash, silica fume
Superplasticizer
Stabilizing agent
Polycarboxylate based
Cellulose, potato starch, sphingans
Locally available alternatives in SSC

Natural pozzolans, limestone filler, ricehusk ash, bagasse ash
Natural pozzolans, limestone filler, ricehusk ash, bagasse ash
Lignosulfonates, naphthalene sulfonates
Cassava/maize/potato starch, cellulose
Since the demand for daily concrete practice in SSA is mainly normal concrete, it is
worthwhile taking into consideration SPs such as LS and PNS, which are cheaper and
readily available in SSA. Particularly LS seems to be an attractive solution, since it does
not need sophisticated processing and occur as a waste of the cellulose production, which
exists all over Africa. However, since LS is available at a low price, the calorific value of
LS also makes it attractive as a source of energy in areas of high energy prices. This is the
reason, why some admixture companies project better market chances for PNS products.
In conjunction with a moderately high w/p, LS and PNS can function rather effectively as
well as for self-compacting or very flowable concrete. However, a moderately high w/p
may cause segregation. This can be counteracted by STAs, e.g. based on Cassava, which
grows nearly everywhere on the continent. The starch of Cassava is very similar to the
starch of potatoes, which are used effectively as STAs for concrete.
Using materials from local sources, some of which are listed in Table 1, can help in
developing economically efficient SCCs for the SSA markets. To overcome the problem
field of unaffordable equipment and supply chains for proper proportioning and casting,
a concept as illustrated in Fig. 2 can be adopted. Well-proportioned and pre-homogenized
dry compounds including cement, fillers, SPs, STAs, and sand (e.g. up to 2 mm/0.08 in.)
can be delivered to the construction sites, as opposed to the individual constituents delivered separately. Once on-site, such pre-homogenized dry compounds solely need to be
amended by water and coarse aggregates in order to provide reproducible flow performance. The feasibility of such a concept has previously been proven for SCC.11
Markets in SSA are on the rise. This exhibits a high pressure on the local cement and
concrete industry to build up state of the art houses and infrastructure. However, the
boundary framework shows a significant number of drawbacks in casting of concrete.
Flowable concrete or SSC can significantly contribute in enhancing the durability and
safety of concrete structures in SSA. However, in order for it to be feasible in SSA, it needs
to be customized for the local boundary framework and supplies. A study of how this can
be practically done is presented in this paper.
The aim of the study was to prove that a pre-mixed binder compound can obtain
highly flowable fresh concrete properties and function well despite the lack of sophisticated raw materials. Therefore preferably materials from SSA were used. Other materials

Fig. 2Comparison of state of the art in SSA and a pre-mixed mortar concept.
were replaced by materials from Germany, which could be obtained in SSA with similar
characteristics.
Materials under investigation
The materials taken into account for the development of a pre-mixed dry binder
compound are listed in Table 2. The admixtures were commercial powder type products
for the use in dry mortar technology. They are listed in Table 3. Since the focus of the study
was on the entire concrete concept, the admixtures were not further specified.
Water demand of binder pastes and binder-sand mixtures
In order to reduce the cement content, the water demand of cement with different
replacement ratios of powders was determined using the Puntke-method. The lower the
water demand, the higher the packing density becomes. The principle also functions for
finding the optimum adjustment of binder paste and sand to obtain the highest possible
packing density. For the test, which is described in detail elsewhere,8,12,13 water is added
gradually to a powder under constant stirring and agitation until the powder or powder
compound tends to settle with a remaining surface texture and a light shimmer, but without
bleeding water. The water demand can be derived from equation (1). For the investigations
496SP-302-37
Table 2Mineral binder materials, sand and aggregate properties

Material
Ordinary portland cement
Limestone filler
Rice husk ash
Fly ash
Natural pozzolan
Sand
Aggregate
Provenience
Germany
Germany
Tanzania
Tanzania
Uganda
Germany
Germany
Specification
CEM I 42.5 R
Burnt under random conditions
Burnt under random conditions
Ground at BAM laboratory
Fractions 0-0.5/0.5-1.0/1.0-2.0
Fractions 2.0-4.0/4.0-8.0/8.0-16.0
Specific gravity
3.12
2.74
2.07
2.06
2.50
2.60
2.60
Table 3Commercially available admixtures used in the experiments

Admixture type
Polycarboxylate ether
Lignosulfonate
Naphthalene sulfonate
Cassava starch
Provenience
Germany
Germany
Germany
Nigeria
Specification
Powder type
Powder type
Powder type
Powder type, cold-water soluble
in sand-binder mixtures, the sand fractions 0-0.5, 0.5-1.0, 1.0-2.0 were mixed in a ratio of
25:25:50.
.nW = VW / (VW+VP)
(1)
where: nW = water demand [-]; VW = water volume [ml]; VP = volume of solids [ml]
Determination of the optimum admixture dosage
For the determination of the necessary admixture dosage a method was used, which was
recommended by Schmidt8 in order to determine the adsorption properties of PCE without
sophisticated equipment. Spread flow tests were conducted with increasing SP dosages at
powder water mixtures determined according to equation (1). This low water dosage is
necessary to avoid segregation at high SP dosages and to make sure that the paste definitively does not flow without SP addition. The resulting curve provides information about
the dosage of a SP, which is required to first induce flow. It also shows the dosage above
which no further yield stress reduction can be obtained.8,14,15
After 0.5 minutes of dry mixing, the water was added and the wet compound was mixed
for 2 minutes at a high rotational speed. After that, the mixture was left to rest for seven
minutes before the SP was added, and it was mixed again for one minute at a low rotational
speed. The reason for the resting period was to avoid intercalations and influences of the
formation of ettringite and monosulfate during the first couple of minutes.
Rheometric investigations of the temperature sensitivity
Rheometric investigations on mortars were conducted with a Couette type viscosimeter
(Schleibinger Viskomat NT) using a basket cell.8,16 The measurement setup is shown in
Table 4. The investigated mortars were tested at 5, 20, 30, and 40 C. The measurements
were conducted at 5 and 20 minutes following the respective mixing.

Table 4Test protocol for the rheometric investigations of mortars at varied
temperature
Shear rate [s-1]
Duration [s]
24.5
60
21.4
30
18.4
30
15.3
30
12.2
30
9.2
30
6.1
30
3.1
30
Table 5Test protocol for the rheometric investigations of concrete

Shear rate [s-1]
Duration [s]
0.47
10
0.4
5
0.33
5
0.27
5
0.20
5
0.13
5
0.7
5
These investigations were only conducted with the PCE type SP and the compound
including natural pozzolan. Since the aim of the study was the development of a dry powder
compound, for these investigations, the powders including the powder type SP were mixed
for one minute before water addition, thereafter mixing was continued for another four
minutes at a low rotational speed.
Determination of the maximum aggregate volume
After the development of a mortar compound rheometric investigations were conducted
with a concrete rheometer (Rheometer 4-SCC) on mortar with increased aggregate content.
The aggregates fractioned 2-4 mm (0.08-0.16 in.), 4-8 mm (0.16-3.15 in.), and 8-60 mm
(0.31-0.63 in.), were blended in the proportion 30:40:30, respectively. A reference mortar
mixture of 10 l with a maximum aggregate size of 2 mm (0.08 in.) was amended by aggregates in increments of 2 kg (4.4 Lbs), which is equal to a supplementary volume of 0.77 l.
The measurement setup can be taken from Table 5.
OBSERVATIONS
Powder optimisation and binder-sand mixtures
As long as the powders do not significantly vary in their water absorption, the water
demand of a powder mixture is a good indicator for the packing density of powders. If two
powders are blended, the blend exhibiting the lowest water demand could thus be considered as the mixture with the highest packing density. Fig. 3 shows the water demand of
cement (at 0% replacement) and the varied fillers (at 100% replacement) as well as their
blends in different replacement ratios. It can be found that the water demands of pure limestone filler and pure natural pozzolan were lower than the water demands of pure cement,
and that with increasing replacement of cement by these fillers the water demand can be
reduced. The water demand of the fly ash (FA) as well as the rice husk ash (RHA) was
significantly higher than the water demand of cement. These materials were burnt under
uncontrolled conditions in non-industrialized processes. The LOI for both materials was
very high and the water absorption was also very high due to the porous structure. Above a
replacement ratio of 20% the water demand increased steadily. However, it could be found
that up to a replacement ratio of about 20% the water demand in case of RHA maintained at
a similar level as for pure cement and was even lower in case of FA. For further investigations a volumetric replacement ratio of 20% was chosen.
Fig. 4 shows the respective compressive strength at varied replacement ratios and at
water content as determined from the water demand experiments. After 7 days there
498SP-302-37
Fig. 3Water demand of cement filler mixtures at varied

replacement ratio.
Fig. 4Compressive strength for different replacement ratios after 7 and 28 days.
was only a negligible effect of the replacement of cement with limestone filler, and the
replacement with RHA showed an increased strength. The replacement of cement with
FA and natural pozzolan had negative effects on the compressive strength. Nevertheless at
20% replacement the natural pozzolan showed a similar behavior as the limestone filler.
Whether the increase in strength between 15% and 20% replacement with natural pozzolan

Fig. 5Water demand of binder (80% cement, 20% filler)

sand (max grain size 2 mm) mixtures at varied replacement
ratio.
can be explained by the higher packing density, a beneficial chemical interaction between
the binder components, a modified curing process or a testing error cannot be derived from
the available data.
It is interesting to observe that both, the FA and the RHA have a distinct negative effect
after 28 days, while the performance loss of mixtures incorporating natural pozzolan or
limestone filler show reduced strength properties in the order of magnitude of the replacement rate. Both ashes were derived from an incomplete combustion process with a high
loss on ignition, containing a high content of sulfates and chlorides. Therefore it is assumable that the chemical composition of the ashes may have had a compromising effect on the
mechanical performance of the specimens.
Subsequently, mortars were generated with different replacement ratios of binder, and
the water demand was similarly determined according to equation (1). The results are
presented in Fig. 5. It can be seen that for all fillers, the minimum of the water demand
occurs between 15% and 30% binder volume. In order to support the self-compacting
properties, for further investigations a volumetric powder fraction of 30% was chosen for
all powders.
Admixture adsorption
Based on the investigations presented in Fig. 5 the mixture compositions presented in
Table 6 were derived. These mortar mixtures were further investigated on their interactions
with the SP types as listed in Table 3. For these investigations the mortars in Table 6 were
mixed with different SP dosages and spread flow tests were conducted. The aim was to
identify the SP dosage at which maximum spread flow occurs.
It was found that the mixtures containing FA and RHA hardly showed any flowability
regardless of the SP type or dosage. By increasing the water content, flowability could
be obtained, most effectively by using polycarboxylate ether superplasticizer. However,
at the same time, the systems with these two filler materials showed distinct segregation.
500SP-302-37
Table 6Mortar mixture compositions derived from binder-sand

optimizations
Water
Binder
Sand
demand
Vol.-% of Vol.-% of Vol.-% of
Filler type
solids
solids
total
Limestone filler
30%
70%
22%
Rice husk ash
30%
70%
30%
Fly ash
30%
70%
24.5%
Natural pozzolan
30%
70%
20.5%
Cement
[kg/m3]
(lb/ft3)
580(36.2)
521(32.5)
562(35.1)
591(36.9)
Filler
[kg/m3]
(lb/ft3)
128(7.8)
87(5.4)
93(5.8)
119(7.4)
Water
Sand
w/b
[kg/m3] [kg/m3] (lb/ by
(lb/ft3)
ft3)
mass
220(13.7) 1420(88.6) 0.31
300(18.7) 1274(79.5) 0.49
245(15.3) 1374(85.8) 0.37
205(12.8) 1447(90.3) 0.29
Fig. 6Slump flow with increased dosage of PCE, PNS, LS

in powder mixtures containing limestone filler and natural
pozzolan.
The reason for the negative influence of the FA and RHA used can be found in the incomplete and uncontrolled incineration conditions. They exhibited a high loss on ignition, very
porous surfaces and high water absorption. Furthermore their high gypsum content is likely
to cause interactions with adsorbing SPs, since SPs and sulfate ions adsorb on hydration
phases like ettringite and monosulfate competitively.17 Based on these observations, it can
be concluded that without further processing and without a controlled burning process,
these raw materials cannot be considered to be feasible materials for flowable concrete.
The slump flow evolution of the limestone filler mixture at w/b = 0.31 (Table 6) can be
seen in Fig. 6. PCE improved the flow properties significantly, while PNS and LS were
required at higher dosages and could not achieve the same slump flow level. If the test was
repeated for the natural pozzolan at w/b = 0.29 (Table 6) it was found that all SPs showed
poor performance. Therefore, the water was increased stepwise at a PCE dosage of 1.6%
solids related to cement (Fig. 7). The PCE dosage is equal to the dosage that yielded the
widest slump flow diameter with limestone filler. It can be observed that increasing water
demands enhance the slump flow value as expected. During the test, up to a w/c of 0.5, no
serious segregation could be observed, indicating that the material has high water binding
capacity.

Fig. 7Increased slump flow with increasing w/p for the

mixture containing natural pozzolan and a PCE solid dosage
of 1.6% of the cement.
Therefore, the slump flow tests with increased SP dosages were repeated at a w/b of 0.5
for the natural pozzolan. These results can also be seen in Fig. 6. While for PNS dosages
above 0.4% the paste showed distinct segregation, with LS, the dosage could be increased
and a maximum flow was observed at a dosage of 1.2% with slight signs of segregation.
The flow performance compares to the limestone filler paste at w/b = 0.31 with PCE.
The experimental results clearly indicate that the highest efficiency can be achieved by
using PCE superplasticizers. In addition the formerly discussed versatility of PCE and the
possibility to customize the performance according to the ambient temperature influences
clearly advocate PCE for ultimate performance. However, in the sub-Saharan African environment the majority of construction needs are in the field of housing and infrastructure
where the challenge is more related to cost issues rather than ultimate performance. The
supply chains for PCE in SSA are fragmentary. Hence it is worthwhile looking at concepts
that can be implemented on the short term. Therefore, the aim of the investigation was to
obtain SCC without sophisticated or expensive constituents like PCE and without the need
for ultimate performance. As a consequence, the final mortar with a moderate w/b of 0.5
was chosen, as well as the lignosulphonate plasticizer. In order to avoid segregation, the
mixture was amended by adding 1.0% STA based on cassava starch related to the water
or 0.58% related to the cement. Cassava starch is easily available at a low price in many
countries in Africa.
Temperature behavior
The obtained mortar was investigated in a rheometer at varied temperatures. The results
for yield stress and plastic viscosity after 5 and 20 minutes can be seen in Fig. 8. There
is no significant influence of the temperature on the yield stress of the mortar, and the
values maintain stable over the first 20 minutes. There is similarly no significant influence of the temperature on the plastic viscosity, but the viscosity increases at all temperatures between 5 and 20 minutes. Since yield stress is mainly affected by the SP adsorption
502SP-302-37
Fig. 8Yield stress and plastic viscosity of the derived mortar

at varied temperature.
Fig. 9Torque gradients and ordinate intercepts at varied

aggregate contents.
and plastic viscosity predominantly by the solid volume fraction, this indicates that the
natural pozzolan does not withdraw active SPs over the course of time, but rather obviously
absorbs water, thus increasing the plastic viscosity.
Aggregate dosage
The aim of the study was to develop a pre-mixed dry mortar compound to be amended
by coarse aggregates and water, to obtain concrete with good flow properties. Therefore,
in the final step it was evaluated, which volume of aggregates can be added to the mortar
composite. Fig. 9 shows that increasing additions of aggregates mainly increase both yield
stress and plastic viscosity. At an addition of 6 kg, which is equal to the total coarse aggregate volume of 18.8%, the yield stress was too high to obtain good self compacting proper-

Table 7Mixture composition of the derived self-compacting concrete
Homogenized, pre-mixed dry mortar compound
Resulting SCC
Cement
385 kg/m3
24.0 lb/ft3
Natural
pozzolan
78 kg/m3
4.8 lb/ft3
Sand
0-2 mm
944 kg/m3
58.9 lb/ft3
Ligno
sulfonate
1.78% of
cement
Cassava
starch
0.58% of
cement
To be added on-site
Aggregate2-16
Water
mm
232 kg/m3 650 kg/m3
14.5 lb/ft3 40.6 lb/ft3
ties. Therefore, the dosage of LS was increased, which signifiantly reduced the yield stress
but also increased the plastic viscosity. At an aggregate addition of 8 kg, which is equal to
a total coarse aggregate volume of 23.5% after further addition of LS, a good SSC could be
obtained with a sufficiently low yield stress and high plastic viscosity. A further increase of
the aggregate was considered to be critical.
COMPARISON OF THE CONCEPT WITH DATA FROM PRACTICE
Based on the stepwise development, a pre-mixed dry mortar compound could be developed which can be amended by coarse aggregates and water according to the mixture
composition in Table 7. In experiments with the aggregates specified in Table 2, a slump
flow of 635 mm (25 in.) could be obtained. The 28-d compressive strength was 56.2 MPa,
and the 90-d compressive strength was 68.4 MPa. The authors found a high number of
construction sites in SSA, where the concrete strength was lower than 20 MPa despite high
cement contents of more than 385 kg/m3 (24.0 lb/ft3) and w/c below 0.5. Furthermore, the
scatter of the performance was significant. The pre-mixed binder compound can therefore
significantly enhance the casting situation on construction sites in SSA, without the need
to change the existing technology.
FURTHER RESEARCH
It was found that the observed RHA and the FA were not feasible for concrete. This,
however, is not related to the material itself, but rather the processing technology. Further
studies will focus on how the performance of these ashes can be enhanced through
tempering or adequate processing.
The cement content in the developed mixture was 385 kg/m3 (24.0 lb/ft3) and the strength
after 28 days was 56.2 MPa. Since for most daily concrete works in SSA lower strength
values are sufficient and the need to save cement is high, further studies will focus on
reducing the cement by replacement with inert materials.
The study could show that under laboratory conditions a functioning SCC could be
derived with admixtures and additions of low level of sophistication. However, further
studies under practical conditions need to be conducted to show the robustness of the
concept as well as the robustness of the used admixtures.
The framework for casting concrete in SSA shows a number of peculiarities with negative effect on the safety and durability of structures. SSC or flowable concrete can be a
feasible way to uncouple the concrete quality from the unfavorable boundary framework.
504SP-302-37
SSC can be achieved from resources that are available in SSA and without sophisticated
admixtures. The mortar compound presented in this study consists of ordinary portland
cement, natural pozzolan, sand, lignosulphonate and cassava starch. In order to generate
SSC with these materials, the w/p has to be moderately high, but the strength results
show that the mechanical performance is much higher than the normal design strength for
concrete in SSA.
AUTHOR BIOS
Dr. Wolfram Schmidt is a researcher at the BAM Federal Institute for Materials
Research and Testing in Berlin. He received a Dipl.-Ing. from RWTH Aachen and a
PHD from TU Eindhoven. His research focuses on SCC, admixtures, and rheology. He is
member of the RILEM committee 228-MPS and the fib task group 8.8.
MSc. Nsesheye S Msinjili is a researcher at BAM Federal Institute for Materials
Research and Testing in Berlin. She received a MSc. in Structural & Fire Safety Engineering at University of Edinburgh and a BSc. (Hons.) in Civil & Structural Engineering
at University of Dar es Salaam. She is a Structural Engineer with experience in concrete
design in Africa.
Prof. Herbert C Uzoegbo of the University of the Witwatersrand in South Africa graduated as Dipl. Ing. in Civil Engineering at the University of Bucharest. He also obtained
MSc and DIC in concrete structures from Imperial College, University of London and a
PhD in concrete structures from Kings College, University of London. He is a fellow of
the South African Institution of Civil Engineers and of the International Masonry Society,
UK.
Dr. John K Makunza is senior lecturer at the University of Dar es Salaam in the
Department of Structural and Construction Engineering. He obtained his PhD from the
Technical University of Dortmund. He is a Civil and Structural Engineer expert with
local and international experience in his field. He has worked with UN-WFP as a Consultant Civil Engineer for both civil and structural works.
REFERENCES
1. Schmidt, W.; Hirya, N. N. M.; Bjegovic, D.; Uzoegbo, H. C.; and Kumaran, S.
G.American Ceramic Society Bulletin, V. 91, 2012, p. 82
2. Schmidt, W.; Radliska, A.; Nmai, C.; Buregyeya, A.; Lai, W. L.; and Kou, S., International Conference on Advances in Cement and Concrete Technology in Africa, Johannesburg, South Africa, 2013.
3. Schmidt, W.Concrete Trends, V. 16, 2013, p. 18
4. Okamura, H., and Ouchi, M., 1st International RILEM Symposium on Self-Compacting
Concrete, Stockholm, Sweden, 1999.
5. Ozawa, K.; Maekawa, K.; Kunishima, M.; and Okamura, H., The second East-Asia
and Pacific Conference on Structural Engineering and Construction, 1989.
6. Nagataki, S., International Symposium on Mineral and Chemical Admixtures in
Concrete, Toronto, Canada, 1998.

7. Schmidt, W.; Brouwers, H. J. H.; Khne, H.-C.; and Meng, B., Influences of superplasticizer modification and mixture composition on the performance of self-compacting
concrete at varied ambient temperatures, Cement and Concrete Composites, V. 49, 2014,
pp. 111-126. doi: 10.1016/j.cemconcomp.2013.12.004
8. Schmidt, W., Eindhoven University of Technology, 2014.
9. Schmidt, W.; Brouwers, H. J. H.; Khne, H.-C.; and Meng, B., in Design, Production and Placement of Self-Consolidating Concrete - Proceedings of SCC2010, Montreal,
Canada, September 26-29, 2010, eds. K. H. Khayat and D. Feys, Springer, 2010, pp. 65-77.
10. Schmidt, W.; Msinjili, N. S.; and Khne, H.-C., International Conference on Advances
in Cement and Concrete Technology in Africa, Johannesburg, South Africa, 2013.
11. W. Schmidt, H.-C. Kuehne, D. Rosignoli and B. Meng, Concrete Plant + Precast
Technology, 2008, 4-11.
12. Hunger, M., Eindhoven University of Technology, 2010.
13. W. Puntke, beton, 2002, 2002, 242-248.
14. Schmidt, W., 22. Workshop und Kolloquium Rheologische Messungen an Baustoffen, Regensburg, Germany, 2013.
15. Schmidt, W.; Brouwers, H. J. H.; Khne, H.-C.; and Meng, B., in 7th International
RILEM Symposium on Self-Compacting Concrete in Conjunction with the 1st International
RILEM Conference on Rheology and Processing of Construction Materials - Supplementary Papers, ed. N. Roussel, Paris, France, 2013.
16. R. Vogel, Concrete Plant + Precast Technology, 2008, 124-126.
17. Plank, J., and Vlad, D.Zement-Kalk-Gips International, V. 59, 2006, pp. 28-39.
506SP-302-37
Index
A
additive 211
admixture 63, 155, 359, 401
admixtures 145, 491
adsorbed layer thickness 63
adsorption 113, 145, 169, 183, 227, 243,
299, 333, 349, 401
adsorption of superplasticizer 425
alite 227
aluminate 349
amine 11
anionic charge amount 169
antifreezing admixtures 279
compressive strength 289

contact angle 211
coordinated water 463
copolymerization 25
crosslinking 199
D
diester 199
dispersability retention 243
dispersing force 25
dispersing performance 199
dispersion 155
diutan gum 39
backbone 265
bentonite 333
blast-furnace slag 113
blast-furnace slag cement 113
blended admixtures 279
blended cements 299, 387
booster 359
brown coal 63
early compressive strength 359

ettringite 401
exchangeable cations 463
expansive agent 289
extreme freezing-point 279
C
C3A 227
calcined clay 299
calcined marl 387
calcium oxide 359
Calorimeter 93
cement 11, 53, 199, 211, 315, 349
cement dispersion 401
cement hydration 145, 371
cement paste 371, 425
chemical admixture 113
civil engineering concrete 77
clay 333
clay interaction 463
clinker 11, 211
comb copolymers 227
complexation 349
F
fiber reinforced mortars 289
flexural strength 289
fluidity 53, 183, 425
fluidity retention ability 113
fluoride ion 425
fly ash 359, 387
formwork filling 449
free length-change 289
G
gel permeation chromatography 227
geopolymer 477
glass fibers 289
glycol 11
graft copolymer 63, 155
grafting 25
grinding aid 211
grinding aids 11
induction period 227

intercalation 333
interlayer space 463
internal curing 371
IPEG 265
polycarboxylate based superplasticizer

425
polycarboxylate ether 39, 53, 199, 491
polycarboxylate superplasticizer 183, 333
polymer 53
polymer intercalation 463
polymerization mechanism 155
polynaphthalene 93
polyols 349
polysaccharides 39
poly(vinyl alcohol) 333
polyvinyl-alcohol fibers 289
pore solution 169, 253, 299
portland cement 371
portland slag cement 169
pumping 77
kinetics 349
rate of heat liberation 437

reactivity 349
restrained length-change 289
retarder 93
rheology 93, 299, 315
robustness 491
H
hardening accelerator 359
heat liberation 437
high-alkalinity 477
high range water reducer 401
high solid-content 125
hydration 227, 349
HydroxyPropyl Guar 315
L
lignite 63
lignosulphonate 491
limestone filler 359
M
macromonomer 199
manufactured sand 333
microstructure 25
molecular design 183
morphology 401
mortar 315, 333
N
nano-size 401
natural zeolite 415
neutron radiography imaging 371
P
PCE 265, 387
PCE superplasticizers 227
phenol 11
phosphonate 77
plasticizers 387
polycarboxylate 25, 93, 125, 155, 169,
243, 401
S
segregation 53
self-compacting concrete 449, 491
set accelerators 145
set retarder 145, 437
setting control 145
setting time 145
side chain 265
silica fume 371
silicate adsorption 477
slow-release 243
sludge water 437
slump 53
slump retention 53
sodium gluconate 437
SRA 289
starch 39
starch ether 491
star-shaped 183
static yield stress 11
steel fibers 289
stereochemistry 349
structural parameters 265

sub-Saharan Africa 491
sulfonated acetone formaldehyde 93
sulphate ion 425
superabsorbent polymer 371
superplasticizer 63, 77, 169, 199, 279,
299, 449, 463
superplasticizer adsorption 253
superplasticizers combination 415
superplasticizers demand 415
supplementary cementitious material 299
surface tension 211
synthesis 155
X-ray powder diffraction 227
temperature 93
thixotropy 449
viscosity 77
volume fraction 39
W
water reduction 53
water retention 315
wetting 211
workability 125, 299
workability loss 415
zeta potential 93, 169, 243, 253, 299
9 781942
727224

Superplasticizers and Other Chemical Admixtures in Concrete

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Superplasticizers and Other Chemical Admixtures in Concrete

Uploaded by

Copyright:

Available Formats

Superplasticizers and

Other Chemical Admixtures

Eleventh International Conference

First printing, June 2015

Printed in the United States of America

RESULTS AND DISCUSSIONS

Effect of Clinker Grinding Aids on Static

Effect of Clinker Grinding Aids on Static Yield Stress of Cement Pastes 13

Figure 1Photo of the grinding mill used for testing.

Effect of Clinker Grinding Aids on Static Yield Stress of Cement Pastes 15

Table 1Properties when cement is ground with amine-based GA

Table 2Properties when cement is ground with glycol-based GA

Table 3Properties when cement is ground with phenol-based GA

Effect of Clinker Grinding Aids on Static Yield Stress of Cement Pastes 17

Figure 2Variations in setting times for cement ground with

Figure 3Typical variations in shear stress for 0.48 w/c

Effect of Clinker Grinding Aids on Static Yield Stress of Cement Pastes 19

Figure 4Variations in 0 for 0.42 w/c pastes made using

Figure 5Variations in 0 for 0.48 w/c pastes made using

Figure 6Relationships between 0 and flow for all tested

Figure 7Variations in compressive strength for mortars

Effect of Clinker Grinding Aids on Static Yield Stress of Cement Pastes 21

Effect of Clinker Grinding Aids on Static Yield Stress of Cement Pastes 23

A 13C NMR Spectroscopic Study on the

A 13C NMR Spectroscopic Study on the Reparation of Acid and Ester

Table 2 Quantities and molar ratios of reactants used in the preparation of

PCEs via copolymerization

PCEs via grafting

A 13C NMR Spectroscopic Study on the Reparation of Acid and Ester

Figure 1 General structures of the copolymerized (left) and

anionic charge in NaOH (pH = 12)

A 13C NMR Spectroscopic Study on the Reparation of Acid and Ester

A 13C NMR Spectroscopic Study on the Reparation of Acid and Ester

Figure 4 13C NMR signals of the carboxylate and the ester

Figure 5 - Schematic illustration of the microstructure of

Figure 6 - Adsorption isotherms on cement of PCE samples

A 13C NMR Spectroscopic Study on the Reparation of Acid and Ester

A 13C NMR Spectroscopic Study on the Reparation of Acid and Ester

Effects of Particle Volume Fraction and

Effects of Particle Volume Fraction and Size on Polysaccharide Stabilizing

Table 2Specific gravities of mixture constituents and mixture composition

Fig. 2Influence of the particle volume fraction on 0 and

Effects of Particle Volume Fraction and Size on Polysaccharide Stabilizing

Mixing of pastes and mortars

Fig. 3Influence of P on 0 of LSF and cement pastes with different STAs

Effects of Particle Volume Fraction and Size on Polysaccharide Stabilizing

Fig. 4Influence of P on pl of LSF and cement pastes with different STAs

Effects of Particle Volume Fraction and Size on Polysaccharide Stabilizing

Fig. 7Influence of the presence of STA on the slump flow

Effects of Particle Volume Fraction and Size on Polysaccharide Stabilizing

Fig. 8Influence of the particle volume fraction of cement

Fig. 9Influence of the maximum particle size diameter

Fig. 10Influence of the maximum particle size diameter

Effects of Particle Volume Fraction and Size on Polysaccharide Stabilizing

5. Khayat, K. H., Viscosity-enhancing admixtures for cement-based materials An

New Additive to Enhance the Slump

Table 1Physical properties and chemical compositions of cement

Table 2Characteristics of the Polycarboxylate Ether

New Additive to Enhance the Slump Retention 55

Table 3Concrete mixture proportions

Water reduction (%)

of KNO3) with a multi-detection system: a differential refractometer (Waters) and a

Table 4Cement paste formulations

PCE dosage (% dry