Professional Documents
Culture Documents
V. Mohan Malhotra
Pawan R. Gupta
Terence C. Holland
SP-302
Editors:
V. Mohan Malhotra
Pawan R. Gupta
Terence C. Holland
SP-302
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.
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.
In October 2003, the Committee for the Organization of CANMET/ACI International
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.
In October 2006, the Committee for the Organization of CANMET/ACI International
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
RESEARCH SIGNIFICANCE
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
RESEARCH SIGNIFICANCE
41
EXPERIMENTAL INVESTIGATION 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
RESEARCH SIGNIFICANCE
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
RESEARCH SIGNIFICANCE
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
RESEARCH SIGNIFICANCE
78
EXPERIMENTAL PROCEDURES
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
RESEARCH SIGNIFICANCE
94
EXPERIMENTAL PROCEDURE
95
EXPERIMENTAL
RESULTS
AND
DISCUSSION
97
SUMMARY AND CONCLUSIONS 109
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
RESEARCH SIGNIFICANCE
114
EXPERIMENTAL PROCEDURE
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
RESEARCH SIGNIFICANCE
126
EXPERIMENTAL INVESTIGATION 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
RESEARCH SIGNIFICANCE
134
EXPERIMENTAL PROCEDURE
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
RESEARCH SIGNIFICANCE
146
EXPERIMENTAL INVESTIGATION 146
EXPERIMENTAL RESULTS
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
RESEARCH SIGNIFICANCE
158
EXPERIMENTAL INVESTIGATION 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
RESEARCH SIGNIFICANCE
170
EXPERIMENTAL INVESTIGATION 171
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
RESEARCH SIGNIFICANCE
184
EXPERIMENTAL INVESTIGATION 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
RESEARCH SIGNIFICANCE
202
EXPERIMENTAL INVESTIGATION 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
RESEARCH SIGNIFICANCE
228
MATERIALS AND METHODS
229
RESULTS AND DISCUSSION
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
EXPERIMENTAL INVESTIGATION 244
RESULTS AND DISCUSSION
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
MATERIALS AND METHODS
256
RESULTS AND DISCUSSION
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
RESEARCH SIGNIFICANCE
266
EXPERIMENTAL INVESTIGATION 266
RESULTS AND DISCUSSION
267
CONCLUSIONS
275
SP-302-21 279
Blended Antifreezing Admixture
with Extreme Freezing-Point
by Anatoly I. Vovk
INTRODUCTION
279
RESEARCH SIGNIFICANCE
280
EXPERIMENTAL INVESTIGATION 281
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
MATERIALS AND METHODS
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
RESEARCH SIGNIFICANCE
300
EXPERIMENTAL PROCEDURE
300
EXPERIMENTAL
RESULTS
AND
DISCUSSION
305
SUMMARY AND CONCLUSIONS 310
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
RESEARCH SIGNIFICANCE
316
MATERIALS AND METHODS
316
EXPERIMENTAL RESULTS
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
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
RESEARCH SIGNIFICANCE
350
EXPERIMENTAL PROCEDURE
350
RESULTS AND DISCUSSIONS
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
RESEARCH SIGNIFICANCE
360
EXPERIMENTAL PROCEDURE
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
RESEARCH SIGNIFICANCE
372
EXPERIMENTAL INVESTIGATION 372
EXPERIMENTAL
RESULTS
AND
DISCUSSION
376
SUMMARY AND CONCLUSIONS 381
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
RESEARCH SIGNIFICANCE
388
EXPERIMENTS AND METHODS 388
RESULTS AND DISCUSSION
390
CONCLUSIONS
398
SP-302-30 401
A Study on the Cement
Compatibility of PCE
Superplasticizers
by A. Lange and J. Plank
INTRODUCTION
401
RESEARCH SIGNIFICANCE
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
RESEARCH SIGNIFICANCE
416
EXPERIMENTAL INVESTIGATION 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
RESEARCH SIGNIFICANCE
427
EXPERIMENTAL INVESTIGATION 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
RESEARCH SIGNIFICANCE
438
EXPERIMENTAL INVESTIGATION 439
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
RESEARCH SIGNIFICANCE
450
EXPERIMENTAL INVESTIGATION AT
THE GROUT SCALE
450
ANALYTICAL INVESTIGATION 452
COMPARISON OF PREDICTIONS AND
EXPERIMENTAL RESULTS
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
RESEARCH SIGNIFICANCE
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
RESEARCH SIGNIFICANCE
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
RESEARCH SIGNIFICANCE
494
EXPERIMENTAL PROCEDURE
494
OBSERVATIONS
497
COMPARISON OF THE CONCEPT
WITH DATA FROM PRACTICE
503
FURTHER RESEARCH
503
SUMMARY AND CONCLUSIONS 503
Index 507
SP-302-01
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.
RESEARCH SIGNIFICANCE
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.
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.
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
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
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
18SP-302-01
20SP-302-01
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
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.
RESEARCH SIGNIFICANCE
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.
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
EXPERIMENTAL PROCEDURES
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
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
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
30SP-302-02
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
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
34SP-302-02
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.
38SP-302-02
SP-302-03
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
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.
RESEARCH SIGNIFICANCE
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.
EXPERIMENTAL INVESTIGATION
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
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
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
44SP-302-03
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
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
50SP-302-03
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
SP-302-04
54SP-302-04
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
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.
RESEARCH SIGNIFICANCE
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%
Mix 1
Mix 2
Mix 3
W/C
0.69
0.54
0.67
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)
56SP-302-04
Paste 1
W/C
0.38
g (lb)
Water
150 (0,331)
Cement
400 (0,881)
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
58SP-302-04
60SP-302-04
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
SP-302-05
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.
RESEARCH SIGNIFICANCE
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.
EXPERIMENTAL PROCEDURE
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.
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
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.
EXPERIMENTAL RESULTS AND DISCUSSION
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
DI water [eq/g]
3,347
4,241
3,643
CPS [eq/g]
3,964
3,974
2,713
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
72SP-302-05
74SP-302-05
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
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.
RESEARCH SIGNIFICANCE
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.
EXPERIMENTAL PROCEDURES
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.
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.
ANALYTICAL INVESTIGATION
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)
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
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.
EXPERIMENTAL RESULTS AND DISCUSSION
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:
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.
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
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
Re =
l d 2
(7)
min
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.
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
92SP-302-06
SP-302-07
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.
RESEARCH SIGNIFICANCE
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
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.
EXPERIMENTAL PROCEDURE
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
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
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
Superplasticizer (% bwoc)
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.
EXPERIMENTAL RESULTS AND DISCUSSION
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.
= 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.
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
100SP-302-07
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
* () polymer component
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)
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
* () polymer component
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
* () polymer component
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
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
* () polymer component
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
* () polymer component
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.
SUMMARY AND CONCLUSIONS
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
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
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
114SP-302-08
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
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.
RESEARCH SIGNIFICANCE
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.
EXPERIMENTAL PROCEDURE
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
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)
116SP-302-08
Cement
ECM6000
Target slump,
cm [in.]
Target air,
%
182.5
[7.11.0]
4.51.0
* 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.
EXPERIMENTAL RESULTS AND DISCUSSION
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
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
120SP-302-08
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
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)
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
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
122SP-302-08
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
126SP-302-09
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
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.
RESEARCH SIGNIFICANCE
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.
EXPERIMENTAL INVESTIGATION
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,
128SP-302-09
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)
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)
--
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)
132SP-302-09
water-reducing rate
[%]
31.5
32.5
80 (3.15)
90 (3.54)
60 (2.36)
60 (2.36)
SP-302-10
133
134SP-302-10
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.
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
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
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
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
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.
cemconres.2009.08.013
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.
cemconres.2011.11.011
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
10. 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, 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.
cemconres.2007.08.017
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
145
146SP-302-11
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.
RESEARCH SIGNIFICANCE
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.
EXPERIMENTAL INVESTIGATION
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
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.
EXPERIMENTAL RESULTS
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 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
SP-302-12
2SP-302-12
4SP-302-12
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
RESEARCH SIGNIFICANCE
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.
EXPERIMENTAL INVESTIGATION
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).
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.
8SP-302-12
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%
Fluid
System
pH 12.5
aqueous
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. 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
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. 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.
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. His research interests include cement chemistry, chemical admixtures,
organic-inorganic composite and nano materials, concrete, dry-mix mortars and oil well
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.
4. Plank, J.; Pllmann, K.; Zouaoui, N.; Andres, P. R.; and Schaefer, C., Synthesis
and performance of methacrylic ester based polycarboxylate superplasticizers possessing
14SP-302-12
hydroxy terminated poly(ethylene glycol) side chains, Cement and Concrete Research, V.
38, No. 10, 2008, pp. 1210-1216. doi: 10.1016/j.cemconres.2008.01.007
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
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.
RESEARCH SIGNIFICANCE
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-
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)
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.
EXPERIMENTAL INVESTIGATION
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
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
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-
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.
EXPERIMENTAL RESULTS AND DISCUSSION
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
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.
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
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).
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. His research interests include cement chemistry, concrete admixtures,
organic-inorganic composite and nano materials, concrete, dry-mix mortars and oil well
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
8. 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, No. 2, 2000, pp. 197-207. doi: 10.1016/S0008-8846(99)00230-6
182SP-302-13
SP-302-14
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.
RESEARCH SIGNIFICANCE
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.
EXPERIMENTAL INVESTIGATION
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.
(1)
186SP-302-14
Esterification rate, %
58
57.9
66.5
62
63
62.4
188SP-302-14
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
190SP-302-14
192SP-302-14
194SP-302-14
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
196SP-302-14
198SP-302-14
SP-302-15
199
200SP-302-15
202SP-302-15
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.
RESEARCH SIGNIFICANCE
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.
EXPERIMENTAL INVESTIGATION
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.
EXPERIMENTAL RESULTS AND DISCUSSION
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.
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
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
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]
[%]
-
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
SP-302-16
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 AND METHODS
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%.
%
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**
%
58.1
19.6
3.02
14.04
0.03
2.00
0.09
1.03
97.91
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
Units
N*s/m2 x103
(psf s)
Toluene
Hexane
Ethanol
0.604 (12.61)
0.326 (6.81)
1.07 (22.35)
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)
216SP-302-16
m2
(from Eq. (2))
t 2
B + H + BH + (4)
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)
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
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
LW
GSL = GSL
+ GSLAB (8)
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
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
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 .
CLKF1
CLKF2
CLKF3
H0 2.8
20
38
32
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
(13)
(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.
226SP-302-16
SP-302-17
228SP-302-17
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:
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.
MATERIALS AND METHODS
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
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
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
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
%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)
(ounce/yr2) (mmole/yr2)
(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)
%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),
%wt
0.05
100
0.10
100
0.15
100
0.20
100
0.30
73
PCE 113-9
Adsorption
mg/m2
mmole/m2
(ounce/yr2) (mmole/yr2)
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
(ounce/yr2) (mmole/yr2)
(ounce/yr2) (mmole/yr2)
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)
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
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)
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
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
8. 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, No. 4, 2007, pp. 251-262. doi: 10.1016/j.
cemconcomp.2006.12.006
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,
2011, pp. 1289-1309. doi: 10.1016/j.cemconres.2011.03.005
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
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
244SP-302-18
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.
EXPERIMENTAL INVESTIGATION
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
Q= (F2-602)/602=F2/602-1
246SP-302-18
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
248SP-302-18
250SP-302-18
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
dispersing force of polycarboxylate-type superplasticizer by sulfate ion concentration in
aqueous phase, Cement and Concrete Research, V. 31, No. 3, 2001, pp. 375-383. doi:
10.1016/S0008-8846(00)00503-2
10. 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: [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
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
256SP-302-19
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
2.7
3.0
4550
2.8
4.1
4720
582
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.
MATERIALS AND METHODS
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.
258SP-302-19
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-
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
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.
cemconres.2012.03.003
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
266SP-302-20
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.
RESEARCH SIGNIFICANCE
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.
EXPERIMENTAL INVESTIGATION
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.
Items of investigation
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
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.
RESULTS AND DISCUSSION
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
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
270SP-302-20
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.
272SP-302-20
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
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
y2 = 350.1e0.013n (2)
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
278SP-302-20
SP-302-21
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.
RESEARCH SIGNIFICANCE
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-
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
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
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.
EXPERIMENTAL INVESTIGATION
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.
EXPERIMENTAL RESULTS AND DISCUSSION
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
Setting time
(hour-min)
Oxide composition, %
initial final SiO2 Al2O3 Fe2O3 CaO MgO R2O SO3
Mineralogical composition, %
C3S C2S C3A C4AF
3-00
0.77 2.55
64
15
4.2
14
3-10
61
15
11
4-05
5.0
14.0
2-45
3-30
no
no
no
data data data
no
data
n.a.
2.8
62.7 15.8
Concentration, %
Freezing temperature,
SNF
brown liquid
1.23-1.24
71
55
-28.7
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.
PNS type
Na-PNS
Na-PNS
Na-PNS
-PNS
modified Na-PNS
Content of
Na2SO4, %
1
2
4.9
0
2.6
PCE sample
-1
-2
-3
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
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
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
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 .
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
freezing
chamber
0.50
1.4
11/4.3
2400/4050
0.46
1.0+1.1
13/5.1
2435/4110
0.44
1.0+1.1
12/4.7
2410/4060
0.43
1.0+1.1
11/4.3
2420/4080
-30
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
286SP-302-21
288SP-302-21
SP-302-22
289
290SP-302-22
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
294SP-302-22
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.
298SP-302-22
SP-302-23
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.
RESEARCH SIGNIFICANCE
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.
EXPERIMENTAL PROCEDURE
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.
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
302SP-302-23
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)
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
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.
EXPERIMENTAL RESULTS AND DISCUSSION
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).
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
310SP-302-23
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
Research, V. 41, No. 10, 2011, pp. 1058-1066. doi: 10.1016/j.cemconres.2011.06.010
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.
314SP-302-23
SP-302-24
316SP-302-24
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
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
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
HPG 5
HPG 6
High
-
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.
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.
EXPERIMENTAL RESULTS
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
322SP-302-24
324SP-302-24
326SP-302-24
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.
330SP-302-24
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
Al2O3
16.33
MgO
4.67
K2O
0.17
Na2O
0.04
P2O5
0.05
LOI
(wt%)
11.7
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
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
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.
RESULTS AND DISCUSSIONS
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
338SP-302-25
340SP-302-25
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
Mxxx-6
29.22%
1.92
4.80
Txxx-6
14.38%
2.80
5.60
9.33
344SP-302-25
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-
348SP-302-25
SP-302-26
349
350SP-302-26
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
-
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
RESEARCH SIGNIFICANCE
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.
EXPERIMENTAL PROCEDURE
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
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.
RESULTS AND DISCUSSIONS
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 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
354SP-302-26
356SP-302-26
= 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.
2. 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,
2011, pp. 1289-1309. doi: 10.1016/j.cemconres.2011.03.005
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
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.
RESEARCH SIGNIFICANCE
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.
EXPERIMENTAL PROCEDURE
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
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)
Density [g/cm3]
Blaine fineness [cm2/g]
BET [cm2/g]
a)
SiO2
Quartz flour
2.63
a)
not determined
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
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
4h
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)
Dosage
[%]
Reference
Reference
Reference
Type of fines
75% limestone
20% quartz flour
50% limestone
50% 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
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
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
Dosage
[%]
Reference
Booster
Booster
Booster
Booster
0.05
0.1
0.15
0.2
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)
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
9h
232 (9.1)
1.2 (174)
5.2 (754)
238 (9.4)
3.4 (493)
170 (6.7)
4.2 (609)
7.2 (1044)
24 h
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
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
Concrete Research, V. 40, No. 6, 2010, pp. 935-946. doi: 10.1016/j.cemconres.2010.01.008
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,
2008, pp. 848-860. doi: 10.1016/j.cemconres.2008.01.002
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
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]).
RESEARCH SIGNIFICANCE
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.
EXPERIMENTAL INVESTIGATION
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,
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
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.
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
378SP-302-28
380SP-302-28
382SP-302-28
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
386SP-302-28
SP-302-29
388SP-302-29
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:
(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.
RESULTS AND DISCUSSION
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
Study on the Rheology of Fly Ash Versus Calcined Marl Blended Cements
with Polycarboxylate-Based Superplasticizers 391
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
Study on the Rheology of Fly Ash Versus Calcined Marl Blended Cements
with Polycarboxylate-Based Superplasticizers 393
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
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.
Study on the Rheology of Fly Ash Versus Calcined Marl Blended Cements
with Polycarboxylate-Based Superplasticizers 395
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
Study on the Rheology of Fly Ash Versus Calcined Marl Blended Cements
with Polycarboxylate-Based Superplasticizers 397
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).
398SP-302-29
Study on the Rheology of Fly Ash Versus Calcined Marl Blended Cements
with Polycarboxylate-Based Superplasticizers 399
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).
5. 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,
2011, pp. 1289-1309. doi: 10.1016/j.cemconres.2011.03.005
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.
35, No. 1, 2013, pp. 151-162. doi: 10.1016/j.cemconcomp.2012.08.020
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
MPEG-based polycarboxylate superplasticizers, Cement and Concrete Research, V. 42,
No. 6, 2012, pp. 847-854. doi: 10.1016/j.cemconres.2012.03.005
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
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.
RESEARCH SIGNIFICANCE
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.
EXPERIMENTAL PROCEDURES
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).
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
408SP-302-30
410SP-302-30
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
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
SP-302-31
416SP-302-31
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
RESEARCH SIGNIFICANCE
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.
EXPERIMENTAL INVESTIGATION
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
418SP-302-31
Specific
Gravity
1.14
pH
6.3
1.20
7.2
50
< 0.01
1.17
6.1
50
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.]).
EXPERIMENTAL RESULTS AND DISCUSSION
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
420SP-302-31
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.
conbuildmat.2013.08.086
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,
2010, pp. 1704-1713. doi: 10.1016/j.cemconres.2010.08.012
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
424SP-302-31
SP-302-32
426SP-302-32
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)
RESEARCH SIGNIFICANCE
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.
EXPERIMENTAL INVESTIGATION
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
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
2CaOSiO2
16
3CaOAl2O3
10
4CaOAl2O3Fe2O3
9
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).
EXPERIMENTAL RESULTS AND DISCUSSION
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)
+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
432SP-302-32
+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)
+KF
0.128 mol/kg
(0.0582 mol/lb)
0.928 mg/g
(0.928 mlb/lb)
No KF
0.913 m2/g
(0.645 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:
434SP-302-32
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
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.
RESEARCH SIGNIFICANCE
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 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.
EXPERIMENTAL INVESTIGATION
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
OPC
SiO2
21.6
C3S
64.4
Cl
0.11
SO3
1.8
Density
(g/cm3)
3.17
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).
EXPERIMENTAL RESULTS AND DISCUSSION
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.
442SP-302-33
Fig. 5 Relationship between residual ratio of C3S and relative heat liberation in 24 h
444SP-302-33
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
450SP-302-34
Weight % Na2Oeq
0.18
Weight % Na2Oeq
0.87
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.
RESEARCH SIGNIFICANCE
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
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.
ANALYTICAL INVESTIGATION
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
eq =
4.
n
= 1 + ( )
5.
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 =
( 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.
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.
EXPERIMENTAL RESULTS AND DISCUSSION
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
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
460SP-302-34
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.
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,
2011, pp. 1279-1288. doi: 10.1016/j.cemconres.2011.01.009
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,
No. 3, 2004, pp. 363-371. doi: 10.1016/j.cemconres.2003.08.005
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
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
RESEARCH SIGNIFICANCE
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.
EXPERIMENTAL PROCEDURES
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
466SP-302-35
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)
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
470SP-302-35
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. 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
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
MPEG-based polycarboxylate superplasticizers, Cement and Concrete Research, V. 42,
No. 6, 2012, pp. 847-854. doi: 10.1016/j.cemconres.2012.03.005
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.
RESEARCH SIGNIFICANCE
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
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.
EXPERIMENTAL RESULTS AND DISCUSSION
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
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)
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 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 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+.
484SP-302-36
486SP-302-36
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
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
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
494SP-302-37
Fillers
Superplasticizer
Stabilizing agent
Polycarboxylate based
Cellulose, potato starch, sphingans
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
RESEARCH SIGNIFICANCE
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.
EXPERIMENTAL PROCEDURE
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
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
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.
24.5
60
21.4
30
18.4
30
15.3
30
12.2
30
9.2
30
6.1
30
3.1
30
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. 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
500SP-302-37
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
502SP-302-37
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.
SUMMARY AND CONCLUSIONS
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.
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
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
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
kinetics 349
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
temperature 93
thixotropy 449
viscosity 77
volume fraction 39
W
water reduction 53
water retention 315
wetting 211
workability 125, 299
workability loss 415
9 781942
727224