U DC 666.942.7.

017

High-magnesia Portland cements: studies on

cements prepared with reagent-grade
chemicals*
S. S. Rehsi,

MSc(Tech), PhD

and A. J. Majumdar,

PhD

MINISTRY OF PUBLIC BUILDING AND WORKS: BUILDING RESEARCH STATION

SYNOPSIS
Several Portland cement clinkers containing up to
15 %MgO have been prepared with reagent-grade chemicals, and the properties of cements made from them
have been investigated. In particular, the effect of variations in composition, burning temperature and cooling
conditions of the clinkers upon the autoclave expansion
and strength of the corresponding cements has been
examined in detail. It is not possible by controlling these
factors alone to produce 'sound' (i.e. volume-stable)
Portland cements having MgO percentages much above
the present limits recommended by the Standards in
different countries. The addition of a suitable reactive
silicate material in finely divided form, e.g. pulverizedfuel ash, effectively stabilizes high-magnesia cements and
these mixtures pass the ASTM test for soundness of
Portland cements. Data on the autoclave expansion and
mechanical strength of these mixtures of high-magnesia
cement and pfa are presented.

Introduction
Portland cements containing up to 5 % MgO remain
'sound', i.e. volume-stable, in use. With more than 5 %
MgO, a cement exhibits instability, commonly known
as 'unsoundness'. The reason for such behaviour is
that, except for a small amount which may enter solution in the silicate liquid and so become part of the
glass, in a normal, slowly cooled, cement clinker most
of the MgO remains as free, hard-burned, crystalline
MgO (the mineral periclase). This crystalline MgO
hydrates slowly in water at normal temperatures. (1) In
cement paste, the hydration reaction may be retarded
even further because the MgO grains are usually embedded in the matrix. The hydration of MgO to

*Crown copyright.

Mg(OH)2 is accompanied by a volume increase of

118 %(2), and the delayed expansion on hydration of
free MgO may disrupt the mortar or concrete in a
structure long after the cement has hardened. The
damaging effect of delayed expansion due to MgO was
first observed in 1884 on a number of bridges and
viaducts in France and on the Town Hall of Cassel in
Germany(3). It was the investigations which followed
that led to the introduction of limits for the total MgO
content in cement specifications all over the world.
The British Standard (4) restricts it to a maximum of
4 %, whereas the ASTM recommends (5) a level of not
more than 5 % by weight of MgO in Portland cement.
The restriction of a maximum level of MgO in
cement to 5 %means that only limestones of very high
quality can be used in cement manufacture. For countries with unlimited reserves of such limestones, this is
not a serious handicap. On the other hand, there are a
large number of countries (India, for instance) where
the reserve of low-magnesia limestone is limited.
Efforts are, therefore, being made in some of these
countries(6-10) to devise suitable means by which sound
and durable cements with MgO levels significantly
higher than 5 % may be manufactured. In this respect,
reports of research carried out by Rosa(11-13) in
Czechoslovakia have been very encouraging. His results demonstrate that cements containing as much as
15 % MgO can be made sound (i.e. volume-stable) by
mixing with them sufficient quantities of a finely
divided, reactive silicate material such as pulverizedfuel ash (pfa). These mixtures were shown to pass the
autoclave expansion test, both after a short hydration
period and after having been stored under water for
ten years. Bridges and office buildings are reported to
have been built in Bratislava with this type of cement.
An investigation was undertaken at the Building
Research Station two years ago which aimed at examining the possibility of producing Portland cement
67

Magazine o.fConcrete Research: Vol. 21, No. 67 : June 1969

from Indian limestones containing large percentages
of MgC0 3. In the exploratory phase of this work, it
was necessary to study clinkers of widely differing
compositions and to permit variations in the conditions under which they were produced. Such clinkers
were produced by burning reagent-grade chemicals in
large enoug"t quantities for subsequent measurements
to be made of the properties of the cements. Experimental results obtained with these clinkers and cements made in the laboratory are presented in this
paper.

Experimental
PREPARATION AND TESTING OF
SYNTHETIC CLINKERS

Compositions of synthetic clinkers

Compositions of ordinary Portland cements are such
that, if one ignores the minor constituents of the
clinkers, they can be plotted on the phase diagram for
the CaO-Alz03-SiOz system (14). This particular diagram becomes less important as the percentage of
MgO increases in the clinker because the delineation
of the phase boundaries changes quite sharply. SandraDeudon and Cavalier(lS) have recently published phase
diagrams of the CaO-Alz03-SiOz-MgO system at 5, 10
and 15 % MgO levels which confirm this belief. However, it is well known that ordinary Portland cement
compositions lie close to the ternary CaO-3CaO.SiO[
2CaO.SiO z invariant point, and in the persent investigation the base compositions of the clinkers at the 75,
10 and 15 % MgO levels were selected with this fact
in mind. Other considerations which governed the
selection of clinker compositions were:
(I) the range and proportions of 3CaO.SiO z(C 3S),
2CaO.SiO z(C zS), 3CaO.Al zOiC 3A) and 4CaO.
Al z0 3.FePiC4 AF) normally present in ordinary
Portland cement;
(2) specified limits on lime saturation factor and
AIP3/FeP3 ratio;
(3) the compressive strength of the cement so obtained.
The compositions of some of the clinkers prepared
in this study are given in Table 1. It will be noted that,
at each MgO level, several compositions were synthesized. They differed considerably in their Alz03/
Fe Z0 3 ratios. Other variations included burning temperature and the method of quenching. It has been
claimed from time to time by various investigators
that, if clinkers containing large amounts of MgO are
burned at temperatures lower than clinkering temperatures used industrially, cements made from them do
not show excessive volume expansion. Budnikov(16)
recently supported this view. To confirm these observations, a few cements were prepared from clinkers of
68

a particular composition burnt at different temperatures in the range 1300 to 1500C. The composition
of the clinker was deliberately chosen to have a high
C 4 AF content in order to reduce the clinkering temperature. Various properties of these cements were
measured.
The synthetic clinkers were prepared from reagentgrade chemicals CaC0 3 (999 %); crushed quartz, SiOz
(9995 %); AIP3 (9958 %); Fep3 (Mathey 'Specpure')
and MgO obtained by decarbonating MgC0 3 (999 %)
at 1300C. The powdered materials passing a 200 mesh
BS sieve were accurately weighed in required proportions, mixed dry in a shaker and then decarbonated in
platinum boats at 1000C. The decarbonated mix was
passed through 100 mesh sieve, mixed in the shaker
and then pelletized to form cylindrical pellets of about
25 x 25 mm under pressure.
The pellets were burnt in a platinum boat at the
desired temperature for 2 h in a muffle furnace wound
with 08 mm diameter Pt. 80 % Rh. 20 % wire, fitted
with a motorized variac controller and a Kent multipoint temperature recorder. The control of temperature was within 1 Co. One pellet was burnt at a time.
After burning, the pellet was cooled, either by leaving
it out in the laboratory atmosphere (air-cooled) or by
submerging it under water (water quenched).
A batch of 20 g clinker was made in one run.

Analysis of cement clinkers

The free CaO in the clinker was determined by the
Lerch and Bogue method(17) as specified in the ASTM
Designation C114-63(18). Uncombined MgO in clinkers was estimated by quantitative X-ray diffractometry, following a procedure described by us
recently(19) .
Phases present in the cement clinkers were identified
by X-ray powder diffraction. A focusing camera of the
Guinier-Nonius type, having a diameter of 114 mm,
and Ni-filtered CuKoc radiation were used for recording the photographs. Typical exposure time varied
from Ii to 2 h. The powder photographs were compared with similar photographs obtained with 'standard' specimens of substances expected in these
samples. When such a standard specimen was not
available in the pure form, X-ray data published by
Taylor(ZO) and those appearing in the ASTM powder
data file(Zl) were consulted.
Quantitative determination of the major phases in
the cement clinkers was carried out by microscopic
point-counting on clinker specimens following the
procedure described by Midgley and Dharmadhikari (22).
About 1000 to 1200 counts were made on each section.
The counting method gives the percentages by volume;
these were converted to percentage weight by multiplying by the densities of the phases identified. The
densities used were 313, 328 and 340 g/cm 3 for C 3S,
CzS and the interstitial phases respectively.

TAB LEI:

Compositions of high-magnesia cement clinkers.

Oxide composition
(% by weight)

Lime
saturation
factor

Clinker
Ref. No.
CaO
1
l(a)

61268

Si0 2
I

I AlP3 I Fe,03

21807

4824

(CC)

'

750 I

0'8.78

2
2(a)

61920

;~~~

21-450

4660

- - -..~

,
I
1970 I 1000

0'9. 16

I~~
0923

.----+------t--1
,58011

19879

0916

,
1 58 '252

16(b)
16(c)

l7
17(a)
l7(b)
17(c)

i
I'

4150

2960

15'00

r-

air-cooled. w-q

1450

1500
1450

1786,
-

1450
1'213.

II

1500

1450

2-465

1500

1-402

0933

1786
I

19879

4660

1970

1500

006
0.12

320
260

600

000
012
0'00
024

7-10
665
5-45
515

2898
29.17

a-c
w-q
a-c
w-q

5500
5444
55.00
53-95

2000
2044
20'00
20'81

a-c
w-q
a-c
w-q
a-c
w-q
a-c
w-q

5500
5500
5500
5424
5500
5473
54.73
54-49

2000
000
2000
I
000
20'00. 7 . 5 0 8 7 . 5 0
000
2058 I
018
2000 +---I
000
20'21
006
20.21
500 I 1000
006
2039
,
! 012

500

9'00

52-47
17-415223 I 1759 I
'
52-47 I INI
750 I

w-q

0935

2365

-I

1500

!I

1450

17-40 I
I
6 '00
17.21
INO
I

a-c
w-q
a-c

1500
------.l_
water-quenched

1400

44-46
44.21

52-48
52'73
52.48

-=T' - 1:::10927

1450

4-404

Free
MgO

a-c
w-q

w-q
a-c
w-q

1500

19879

Mineralogical analysis
(% by weight)
Free
CaO

I
t
t
-,a-c
5300
1700.

58-491

1500

'I

---+-I

3(c)
16
16(a)

1450
1500
1450

-I

+-

---

II

'I

I
I

~'370

3
3(a)
3(b)

1048

1 - :
4
I
I
4(a)
61681 21-450 4-404
2-465 I 1000
4(b)
-4(c)
--15,
61279 21'450 3'985
3286
1000
15(a)
15(b)
15(c)
I

*a-c

Fe 2 0 3

MgO

I 4601 I

Conditions I Calculated compound comp:Jsition

of
(% by weight)
cooling
clinker*
C3S
C,S
C 3A I C.AF

Burning
temperature

Al,03

a~~

w-q
a-c
_
wq

II

52-47

115l2~43.
52.68
52.68
52-43

17 -41
IN5
17.26
17.26
IN5

900

900

000
0'12
006
012

rI

0-'1-2

C 3S

5212
53-93 ,
57'55 I
5984
5980
61-41

623
615
535
515
530
495
4'70 I
465 I
i
I

970
955
9'20
905

C,S

Interstitial

2505
2091

19'57
22'44

24'37
2123
23'74
18-43

1098
12.16
11.01
1477

5521
2603 112'53
5640
2320
1425
5612
2571
1282
5804
21'85
1478
5517
2192
1760
5662
1923
19'14
5576 : 2173
1775
5683
1777
2063
I
.
I
'50-43
2395
1592
5161
21-45
1727
5123
2334
16.17
5310
1979
1794
II

012

1020
955
970

I 012

9-40

5523

2125

1400

I 012

1025
985
970
960

5501
5601
5554
5651

23'77
22-45
23-44
1999

1085
11.63
11.26
\378

I 018

750

; 006
6'00~. 006
012

5334
5458
5475

24-45
2215
23-89

1189
13-54
1154

VJ
~

~
:::r;q'
:::-

~::::
~

is'

~
~

is"'

::::

t:).,

~
0"\D

r;;-

Magazine of Concrete Research: Vol. 21, No. 67 : June 1969

The fineness in terms of specific surface of cement

samples was determined by the air permeability method,
the Lea and Nurse apparatus as specified in BS 12: 1958
being used(4).

ments is to be found in the classic study by Gonnerman,

Lerch and Whiteside(27) who demonstrated, with synthetic clinkers made in the laboratory, that cements
containing more than 5 % MgO show large volume
expansion when autoclaved. Results obtained in the
present study confirm this view. (26) The addition of a
suitable pozzolanic material, however, stabilizes these
high-magnesia cements, as first pointed out by Rosa (11).
In the present investigation, confirmatory evidence of
this technologically important observation was sought
by incorporating suitable quantities of a British pulverized-fuel ash (pfa) in the cement mixes. The pfa
fraction passing a 300 mesh BS sieve was used as the
additive and the mixtures were tested for autoclave
expansion and compressive strength.
The chemical analysis, mineralogical composition
and the specific surface of the pfa fraction used in this
study are given in Table 2.

Determination of compressive strength

EXAMINATION OF SET CEMENTS

Compressive strengths of cement samples were

determined by using 13 mm cubes, according to the
method developed by Parker(23) and currently used at
the Building Research Station for testing strengths of
experimental cements. The relationship of these
strength results with those obtained with the 70 mm
cube recommended in the BSI2:1958(4) has been
established by Welch and Gutt(24) to be
Strength of 70 mm cube = 22 x strength of
13 mm cube + 8274
N/mm 2

Autoclaved cements and cement-pfa mixtures were

examined by X-ray diffraction to identify the phases
present.

PREPARATION AND TESTING OF

CEMENTS

Preparation of cements
The clinker samples were first crushed in a percussion mortar to pass a 100 mesh sieve. Four per
cent by weight of gypsum (Analar grade CaS0 4 .2Hp
and passing a 100 mesh sieve) was then added and the
material ground in a small steel ball mill of 50 g
capacity, to a fineness in terms of specific surface of
about 035 m 2/g. Ordinary Portland cements have been
generally found to be of this fineness.

Determination offineness

Determination of soundness
The extent of the potential danger of delayed expansion due to MgO was assessed by the performance of
cements in the autoclave expansion test recommended
by the ASTM(25). In this test, hydration of cement is
substantially accelerated in the presence of superheated steam (at 203 007 MN/m 2) and it is considered that all free MgO is hydrated. The expansion
after the autoclave treatment should not exceed 08 %
for sound cements. In the present study, this standard
test was slightly modified and 13 mm cubes were used
in place of bars measuring 25 x 25 x 285 mm with a
250 mm effective gauge length. The need for such a
modification arose from the fact that experimental
cements were produced in small quantities, whereas
about 500 g of cement are needed to cast only one
specimen of the size recommended by the ASTM
standard procedure(25).
Details of the 13 mm cube method and a comparison
of results obtained with this method and with standard
bar specimens have been reported by us earlier(26).

The use of additives

The experimental basis of the present ASTM restriction on maximum permissible MgO contents in ce70

Results
In Table 1, the oxide compositions of the synthetic
clinkers and their potential phase compositions calculated according to ASTM Designation C 150-65(5) are
given as well as the experimental results of analysis
carried out on them.
In Table 3 are listed the specific surfaces, autoclave
expansions and compressive strengths of highmagnesia cements and of mixtures of these cements

TAB L E 2:

Analysis of pfa.

CHEMICAL ANALYSIS

Oxide

Percentage by weight

Si0 2
AI 2 0,
Fe 20,

498
249
92

CaO
MgO
NazO
KzO
TiO z
Acid-soluble SO,
Loss on ignition
SPECIFIC SURFACE

37
21
14
33
11
0'8

41

= 3642 cm 2 Jg

MINERALOGICAL COMPOSITION

(I) Glassy phase, observed under microscope to be present as

small spheres
(2) Crystalline phases, based on X-ray powder diffraction data:
Quartz (Si0 2 )
Mullite (2Si0 2 .3AI 20,)
Haematite (FeZ03)
Magnetite (Fe304)

Studies of high-magnesia PorI/and cements

with variou proportion of pfa. It will be noted that
the compre ive trength f the cement-pfa mixture
are Ii ted only for tho mi ture which howed autoclave expan ion below the ASTM pecified limit f
O %. To facilitate a ready compari on, the compre ive trength of an ordinary P rtland cement determined on 13 mm a well a 70 mm cube are given at
the bottom of Table 3.
igure 1 and 2 are photograph which h w the
relative ize of the aut laved pecimen of highmagne ia cement ha ing varying MgO content with
and without addition of pfa. ome of the re ult on
autoclave expan ion Ii ted in Table 3 are h wn in
Figure 3 to 6.
The compo ition of the mi u ed to prepare clinker
at lower temperature i gi en in Table 4. Thi Table
a l 0 contain data on free CaO ana ly e and computed

p tential pha e compo ition for the e clinker. The

la t two column of thi Table gi e the figure for
pecific urface of cement made from the e c1inke
and the autoclave e pan ion alue of them.

Di cu ion of result
Data pre ented in able 3 c nfirm Rosa' b er ation (ll) that P rtland cement ha ing more than 5 to
6 % MgO are not' ound' on their own (ee igure I)
but their volume e pan ion can be brought under
control by addition of uitable quantitie of finely
divided pfa . Although present re utt apply to aut cia e condition only, Ro a ha hown that the e
re ull are valid at ambient temperature al 0 over a
period of to 10 years. Furthermore. the cement-pfa
mi tllre have adequate trength to pa the tandard
pecificati n .

Figurt J: Condilion 01
auloc/avtd J3 mm Clllx> sptcimtns
ol ..arious umtnl wilhoul pia.
Tht umtnl ...trt prodllctd Irom
high-magn ia umtnl clinktrs
containing (from righl 10 11'11)
O. I , 2. 3, 4. 5.5'5.6, 7'5. 10
and 15% MgO rtspl'cli\tly.

Figurt 2: COlldilions 01
aUlac/avl'd 13 mm Cllbt
sptcimtn 01 aml'nts wilh and
wilhoul pIa. Tht umtnt wt,t
prodllud Irom high-maglll'Sia
umtnl dinkt'r
as. J4 a1ld 16
conlaining 10 and 15 % MgO
rtsPt'cli~tly.

UpPt'r row: Ctmtnl prodllud

Irom high-l1lQgnt ia alii nl
clinktr o. 16 (f'O/1/ Itft (0 righ/)
(i) ""ilhoul pia. (ii) wilh 25%
pia, (iii) wilh 30 % pia, (iv) wilh
35 % pia and (v) wilh 40 % pia.
Middlt: Ordinary Porllolld
amenl (comlll rcial, MgO
('onltnl 104 "/.) wilhoUI pIa.
Lcwtr row: Ctmtnl produced
Irom high-magnesia ctmtnl
dinktr Q. /4 (from lell IQ righl)
(I) wilhour pIa, (ii) 25 %pIa,
(iii) wilh 30% pia, (il') wilh 35
% pia and (v) with 40% pia.

71

Magazine of Concrete Research: Vol. 21, No. 67 : June 1969

TAB L E 3:

Autoclave expansions and compressive strengths of high-magnesia cements without pfa and containing
different amounts of pfa by weight.
Clinker
Ref. No.

Content
of pfa
(% by weight)

Specific surface
of cement
(cm 2 /g)

Autoclave
expansion

0
10
15
20
0
10
15
20
0
20
25
30
35
0
20
25
30
35
0
25
30
35
0
25
30
35
0
25
30
35
0
25
30
35
0
25
30
35
0
25
30
35
0
25
30
35
0
25
30
35
0
25
30
35
0
25
30
35

3555
3564
3568
3572
3529
3540
3546
3552

12640
s.c.*
3550
0445
11-400
s.c.
1180
0265
17090
s.c.
s.c.
0'750
0457

l(a)

2(a)

2(b)

2(c)

14

14(a)

14(b)

14(c)

---~-.

15

15(a)

15(b)

15(e)

* s.c. = specimen cracked

72

3533
3555
3560
3566
3571
3498
3527
3534
3541
3548
3482
3522
3530
3537
3513
3545
3552
3558
3513
3545
3552
3558
3572
3589
3593
3596
3577
3593
3596
3599
3491
3529
3536
3544
3543
3568
3573
3578
3503
3538
3545
3552
3501
3536
3543
3550
3537
3563
3568
3574

(%)

15145
s.c.
0518
0-410
0264
16'960
s.c.
0548
0'426
l3-780
0568
0-421
0286
16-465
s.c.
0597
0-479
15300
0628
0-406
0313
14050
1117
0-493
0392
l3-21O
0558
0-403
0'278
16032
4015
0573
0'397
----15112
0'538
0379
0284
13990
1102
0-476
0374
l3-240
0583
0332
0277

Compressive strength (N/mm') at

3 days
7 days
28 days
-

_.

916

13-48

1906
-

951

14'18

19'77

1029

1542

2137

1051

1544

2168

.-

10'45

1404

2165
-

.-

j--------

10'37

1503

2264

1034
-

1580

2033

.~--

j--._----

1007

1393

22'17

--

1076
-

1073
-

990

1015
-

1009

1453
- - c--

2099

13-90

2176

.-

13-65

1867

- - c---------- r------------

2162

1429

13-82

1980

10'15
-

13-71
-

---------

1958

Studies of high-magnesia Portland cements

~-

Clinker
Ref. No.

Content
ofpfa
(~~ by weight)

Specific surface
of cement
(cm'/g)

Autoclave
expansion

0
25
30
35
40

3470
3513
3522
3530
3539

21395
s.c.
0787
0-497
0388

0
25
30
35
40
0
25
30
35
40
0
25
30
35
40
0
25
30
35
40
0
25
30
35
40
0
25
30
35
40
0
25
30
35
40

20080
s.c.
0633
0-411
0314
- -1 - - - - - - - 20165
s.c.
0645
0-417
0329
19925
s.c.
0631
0-408
0323
22375
s.c.
1044
0695
0-491
20590
3594
3606
s.c.
3608
0980
0540
3610
3613
0-425
3542
20750
3567
s.c.
3572
0995
0612
3577
3582
0-423
20175
3508
3541
s.c.
0761
3548
3555
0508
3562
0-413
22990
3483
3523
s.c.
3531
1520
3539
0786
3547
0610
3527 ------- - - - - - - 21820
3556
s.c.
3561
1265
3567
0690
3573
0525
-3514
21525
3546
s.c.
3552
1315
0740
3559
3565
0578
3539
21 170
3565
s.c.
3570
0985
3575
0628
3580
0-478

( '~~)

Compressive strength (N/mm') at

28 days
7 days
3 days
-

1186

932
-

1829

--~

3(a)

3(b)

3(c)

16

16(a)
---

16(b)

16(c)

0
25
30
35
40

17

0
25
30
35
40

17(a)

0
25
30
35
40
0
25
30
35
40

17(b)

17(c)

3515
3547
3553
3559
3566
3553
3575
3580
3584
3589
3587
3601
3603
3606
3609
3551
3574
3578
3583
3587

929
-

907

1848

1856

-~

--

~---

13-13

921

1806

---

924

-~-

----~--

1175

1806

---~

0-401

00

3440
_ _ _ _ _ _ L _ _ 3440
___

1316

9-43

935

__

1671
"--------

1175

1732

- - - -[-----

957

-~---

1183

1724

- -

830

1183

1591
-

~----

--

-------

830
-

-----

11-42
-

~.---

--

------

1624
-

8-41

1128

----

-~

833
-

1210

-----

1172
-

----ORDINARY PORTLAND CEMENT

13 mm cube
70 mm cube

1255

- ----

16-44
- -- -

1167

1619
-

13-76 ---]----20-:s2---

0_-4_0_1_~_ _2_6_61_ _- L - . _4_0_-4_7__

__~~ ____ _

73

MaKazine of Concrete Research: Vol. 21, No. 67 : June 1969

~ air-cooled clinker
~

water-quenched clinker

Key to Figures 3 to 6

10

15145 1709

08

1530 16465

15112 16032

--

--

06

06

Q.

X
w
w

05

04

1324 1399

>

08

1321 1405

1378 1696

10

X
w
w

05

:5

04

>

f-

f-

:>

:>

02

02

@
25

30

@
35

25

30

35

25

30

35

25

30

PFA CONTENT-% by weight

35

25

30

35

25

30

35

PFA CONTENT-% by weIght

(a) C 3 A,C4 AF = 15

(a) C 3 A;C.AF = 15

(b) C 3A!C,AF = 10

(b) C3 AC,AF = 10

(c) C3 A!C4 AF = 0'5

(c) C3AiC.AF = 0'5

Figure 3: Autoclave expansion of cement with and without

different amounts of pfa. The cements were produced from
high-magnesia cement clinker Nos. 2, 2(a), 14, 14(a), 15 and
15(a).

Figure 4: Autoclave expansion of cements with and wiThout

different amounts of pfa. The cements were produced from
high-magnesia cement clinker Nos. 2(b), 2(c), 14(b), 14(c),
15(b) and 15(c).

TAB L E 4: Composition of clinker and the autoclave expansions of the cements produced by burning it at
temperatures from 1300 to 1500C.

Composition
of the mix

Burning
temperature

Conditions
of
cooling
clinker

Free
CaO
(%)

C,S

C,S

air-cooled
water-quenched

1020
924

12-45
1635

4885
4591

3618
3592

s.d.*
s.d.*

1350

air-cooled
water-quenched

480
5-40

34-43
3198

32'27
3412

3521
3606

1300
1259

1400

air-cooled
water-quenched

036
018

5250
5323

1864
1809

1450

air-cooled
water-quenched

012
012

53-47
53-47

1791
1791

1500

air-cooled
water-quenched

012
006

53-47
5372

1791
1772

eC)

CaO
SiO,
AI,O,
Fe,O ..
MgO

=
=
=
=
=

61844 %
20316%
5-411%
4929%
7500%

Lime saturation
factor = 0929
AI,O,;,Fe,O, = 1098

* s. d.
74

specimen disintegrated

1300

Calculated compound composition

(% by weight)
C,A

CAF

Specific
surface
of cement
(cm','g)

Autoclave
expansion
(%)

-- -----

...

600 .

1500

---.-.~-

----------

12-41
1239
.. _.._-- - - - 1231
3579
1238
3511
3651
3585

---

---.--~----

3637
3598

--.~---

1232
1227

Studies of high-magnesia Portland cements

2059
22375

16

2008

16
21-395

1-4

14

12

12

"xw

10

xw

08

08

-- ---

>

>

--'
u

:J

19925
20165

10

f-

20175
20 75

24525

21-17

f-

06

:J

06

05

------

05

04

04

02

02

30

35

40

30

35

40

30

35

40

30

35

40

PFA CONTENT-% by weight

30

35

40

40

PFA CONTENT-% by weight

(a) C3 A/C.,AF = /5

(a) C 3 A/CAF = 1'5

(b) C 3 A/C,AF = /'0

(b) C 3 A/CAF = 1'0

(c) C3 A/C,AF = 067

(c) C 3 A/C,AF = 067

Figure 5: Autoclave expansion of cements with and without

different amounts of pfa. The cements were produced from
high-magnesia cement clinker Nos. 3, 3(a), 16(a) /7 and /7(a).

Figure 6: Autoclave expansion of cements with and without

different amounts of pfa. The cements were produced ji-om
high-magnesia cement clinker Nos. 3(b), 3(c), J6(b), /6(c).
17(b) and 17(e).

The amount ofpfa necessary to stabilize a particular

cement depends, as is to be expected, upon the free
MgO content of the cement. However, for each cement,
there is a critical percentage of pfa below which the
mixture does not have any rigidity (see Figure 2).
Menzel(28) had pointed out a long time ago that there
is a minimum in the relationship between strength and
CaO/SiO z ratio for cement and silica mixtures. Present
work suggests that a similar relationship may exist in
the case of mixtures of high-magnesia cement and pfa
also.
Results presented in Tables I, 3 and 4 show that it
is not possible to produce sound Portland cements
having more than 5 to 6 % MgO by altering the conditions of burning and cooling of the clinker or by changing its composition. When clinkers were prepared at a
temperature as low as l300C, it was noted that these
'dusted' extensively. For clinkers burnt at 1350C, the
dusting was less vigorous and it was not observed to

occur when clinkers were prepared at 1400C and

above. As is well known, dusting of a clinker occurs
due to inversion of fJ-C 2S to y-C 2 S. The latter modification of C 2S does not possess any hydraulic properties and is deliberately avoided in cements. Budnikov l161
surprisingly makes no mention of dusting of clinkers
prepared by him at 1350C or below. Autoclave expal,1sion results in Table 4, however, show all clinkers
made from the same mix but at different temperatures
in the range 1300 to 1500C to be unsound.
Examination of the data presented in Table I shows
that variations in the AI 2 0 J /Fe 20 1 ratios in the clinker
compositions as reflected in the variations in ClA/C 4A F
ratios influence the amounts of uncombined MgO in
these clinkers. With increase in Fe 20 3 content in clinker composition. the amounts of uncombined MgO
decrease. A rise in the burning temperature of clinkers
from 1450C to 1500C also results in some decrease
in the quantity of free MgO present in the clinker.
75

Magazine of Concrete Research: Vol. 21, No. 67 : June 1969

Rapid cooling of the c1in ker has a similar effect. These
observations confirm the belief(Z9) that the glassy phase
in the clinker does incorporate a substantial portion of
the MgO present in the mix composition. Conditions
which favour the formation of this glass produce less
uncombined MgO in the clinker. These clinkers were
found relatively easier to grind.
Results presented in Tables 1 and 3 show quite
clearly that there is a direct correspondence between
the amount of uncombined MgO present in the clinker
and the autoclave expansion of the cement made from
it, variables such as burning temperature, cooling conditions and Al z0 3jFe z0 3 affecting this.
All clinkers (listed in Table I) gave identical X-ray
powder patterns in respect of the phases present. Intensity variations were noticed, however, in the case of a
large number of powder lines. This is to be expected
because the clinkers differed widely in their potential
phase analyses. The principal phases identified were
alite, (3-C zS, C 3A and a ferrite phase belonging to the
CzF-'CzA' solid solution series. These are the principal
phases of an ordinary Portland cement c1inker(30). In
addition, there is conclusive evidence of the presence
of a large amount of uncombined MgO. This is also
expected, because the total MgO in the clinkers studied
was equal to or greater than 75 % and only a small
portion of this enters the glassy phase (produced in the
burning process) and other crystalline phases. There
was no evidence to suggest that y-CzS was present.
The amounts of MgO, Fe Z0 3 and Al z0 3 as well as the
glassy phase present in the clinker were presumably
adequate for the stabilization of (3-C zS.
Photomicrographs taken in the course of this study
indicate that water quenching of the clinker produces
smaller periclase crystals than air cooling. Results of
quantitative microscopic examination on high-magnesia cement clinkers, presented in Table I, indicate
that the proportions of C 3S and the interstitial phase
increase and that of the CzS phase decreases when a
clinker is water-quenched instead of air-cooled. An
increase in the Fe Z0 3 content in the clinker compositions reduces the percentages of C 3S and CzS phases
and increases that of the interstitial.
In photomicrographs of autoclaved samples of cements which were stabilized by the addition of suitable
quantities of pfa, periclase crystals were not positively
identified. It must be pointed out, however, that microscopic examination of hydrated cement is rendered
particularly difficult by the presence of extremely finegrained matrix material of hydrate compounds.
All autoclaved high-magnesia cements gave an identical X-ray pattern. The phases which could be identified positively in the photographs were Ca(OH)z,
Mg(OH)z, C 3S-hydrate and anhydrous cement minerals, mainly ferrite. The presence of small quantities
of xonotlite (6CaO.6SiO z.HP), hillebrandite (2CaO.
SiOz.HzO), MgO and the hydrogarnet phases cannot
be ruled out. No evidence was seen, however, of the
76

presence of II A tobt;rmorite (5CaO.6SiO z.5H zO) or

738 A serpentine (3MgO.2SiO z.2HP) in these photographs.
When adequate quantities of pfa are added to these
cements and the mixtures autoclaved, nearly all
Ca(OH)z is used up and C 3S.H gradually disappears
and crystalline 11 A tobermorite makes its appearance.
This is due to the fact that the CaOjSiO z ratio of the
mix changes markedly with additions of pfa and
appropriate hydrate phases are formed. Present knowledge about the dependence of the stability of these
phases upon the CaOjSiO z ratio of the mix under
autoclave condition has been summarized by Taylor(3!).
Present experimental results are in substantial agreement with Taylor's assessment. The amounts of
Mg(OH)z also progressively decrease, partly owing to
the dilution effect of adding pfa but also probably
owing to the formation of hydrates containing magnesium ion in their crystal structures. Periclase is not
completely hydrated to brucite in the autoclave treatment. Among other hydrate phases which are probably
present in small quantities, xonotlite, hillebrandite,
IX-CzS.H and the hydrogarnet phase may be mentioned. A certain proportion of the cement and the
pfa remain un hydrated even after the autoclave treatment. This explains the presence of the powder lines
corresponding to the ferrite phase in the X-ray patterns
of the autoclaved cement, and ferrite, quartz, iron
oxide and mullite in the X-ray patterns of the autoc1aved cement containing pfa.
Although no conclusive evidence of the formation
of any magnesium silicate hydrate was seen in these
photographs, a very diffuse band was seen at 152 A
and sometimes at 245 A in a large number of photographs. The most abundant calcium silicate hydrate
formed was C 3S.H in autoclaved high-magnesia cement and II A tobermorite in autoclaved cement-pfa
mixes. Products of long-term room temperature (I r
1C) hydration under distilled water in the case of
one cement when analysed by X-rays show that even
after three months, no Mg(OH)z or tobermorite was
formed; only traces of CSH gel were seen. Hydration
at room temperature, of course, leads to the formation
of ettringite (3CaO.Al z0 3.3CaS0 4 .32HP), which is
not stable under autoclave conditions.

Conclusions
It is fair to conclude from this work that highmagnesia Portland cements containing as much as
15 % MgO can be stabilized by the addition of suitable
quantities ofpfa. The cement-pfa mixtures which pass
the ASTM autoclave expansion test also possess adequate strength to pass the standard recommendations
for ordinary Portland cement. Although it has not

Studies of high-magnesia Portland cements

been possible in this work to collect data on volume
stability of these mixtures at room temperature, Rosa's
work shows the autoclave results to be valid at ambient
temperatures also. Mixtures of high-magnesia cement
and pfa can therefore be used as substitutes for
ordinary Portland cement. Present work indicates that
these mixtures could be used in precast products which
are to be steam-cured subsequently.

12.

13.

14.

15.

16.

ACKNOWLEDGEMENT
The work forms a part of the training in high-temperature silicate chemistry and cement technology of one of
us (SSR) on attachment to the Building Research
Station under the Colombo Plan Scheme.

17.

18.

19.

REFERENCES
1.

2.

3.

4.
5.

6.

7.

8.

9.

10.

11.

BOGUE, R. H. The chemistry of portland cement. Second

edition. New York, Reinhold Publishing Corporation,
1955. p. 141.
BROWN, L. s. and SAWAYZE, M. A. Autoclave tests and the
microscope. Rock Products. Vol. 41, No.6. June 1938.
pp.65-69.
LEA, F. M. The chemistry of cement and concrete. Second
edition. London, Edward Arnold (Publishers) Ltd, 1956.
p.315.
BRITISH STANDARDS INSTITUTION. BSI2:1958. Portland cement (ordinary and rapid-hardening). London. p. 40.
AMERICAN SOCIETY FOR TESTING AND MATERIALS. ASTM
Designation: CI50-65. Standard specification for Portland
cement. 1965 Book of ASTM Standards: Part 10. October
1965. pp. 98-102.
DOLEZSAl. K. High-magnesia Portland cement. Proceedings
of the Seventh Conference on the Silicate Industry, Budapest,
1963. Budapest, Akademiai Kiad6, 1965. pp. 633-649.
BUDNIKOV, P. P. Physico-chemical properties, microstructure and phase composition of high-magnesia cements.
Proceedings of the Seventh Conference on the Silicate
Industry, Budapest, 1963. Budapest, Akademiai Kiad6,
1965. pp. 651-659.
GILLE, F. Untersuchungen tiber das Magnesia-Treiben von
Portland cement. (Investigations on expansion tendencies
in Portland cement due to MgO.) Zement-Kalk-Gips. Vol.
5, No.5. May 1952. pp. 142-150.
GUYE, F. Ciments a haute teneur en magnesie stables a
l'essai a l'autoclave. (High-magnesia cements stable under
autoclave tests.) Revue des Materiaux de Construction et de
Travaux Publics. No. 567. 1962. pp. 333-348. Translated
from the French by V. McEachran. Garston, Building
Research Station, May 1965. pp. 16. Library Communication No. 1291.
BUDNIKOV, P. P. and VOROBE'V, KH. s. Properties of Portland
cement with high MgO contents. Tsement. Vol. 26, No.1.
1960. pp. 14-21. (In Russian.) Chemical Abstracts. Vol. 54.
1960. p. 11430.
ROSA, J. Raumbestandige Zemente mit hohem MgOGehalt. (Sound cements with high MgO content.) ZementKalk-Gips. Vol. 18, No.9. September 1965. pp. 460-470.

20.
21.

22.

23.

24.

25.

26.

27.

28.

29.

ROSA, J. High-magnesia cements with steady volume

changes and their hydration. Tsement. Vol. 31, No.6.
November-December 1965. pp. 6--8. (In Russian.)
ROSA, J. The mechanism of the stabilisation of highmagnesia Portland cement. Proceedings of the Eighth Conference on the Silicate Industry, Budapest, 1966. Budapest,
Akademiai Kiad6, 1966. pp. 263-273.
LEA, F. M. The chemistry of cement and concrete. Second
edition. London, Edward Arnold (Publishers) Ltd, 1956.
p.57.
CAVALIER, G. and SANDRER-DEUDON, M. Laitiers quaternaries CaO-MgO-Al z0 3 -SiO z. (Quaternary system clinkers
CaO-MgO-AlzOrSiO z.) Revue de Mhallurgie. Vol. 57, No.
12. December 1960. pp. 1143-1157.
BUDNIKOV, P. P. Physico-chemical properties, microstructure and phase composition of high-magnesia cements.
Proceedings of the Seventh Conference on the Silicate
Industry, Budapest, 1963. Budapest, Akademiai Kiad6,
1965. pp. 651-659.
BOGUE, R. H. The chemistry of portland cement. Second
edition. New York, Reinhold Publishing Corporation,
1955. p. 98.
AMERICAN SOCIETY FOR TESTING AND MATERIALS. ASTM
Designation: CI14-63. Standard methods of chemical
analysis of Portland cement. 1964 Book of ASTM Standards: Part 9. January 1964. pp. 88-135.
REHSI, s. and MAJUMDAR, A. J. Quantitative determination
of uncombined MgO in Portland cement clinker by X-ray
diffractometry. Journal of Applied Chemistry. Vol. 18, No.
10. October 1968, pp. 297-300.
TAYLOR, H. F. w. The chemistry of cements. London,
Academic Press, 1964. Vol. 2. pp. xi, 460.
AMERICAN SOCIETY FOR TESTING AND MATERIALS. Index to
the X-ray powder data file. 1958. Special Technical Publication No. 48-G.
MIDGLEY, H. G. and DHARMADHIKARI, P. v. The point counting microscopic method for the quantitutive determination
of the silicate phases in Portland cement clinkers. Garston,
Building Research Station. Current Papers, Research
Series No. 13, pp. 1-7.
PARKER, T. w. Influence of the heat treatment of Portland
cement clinker on the properties of cement. Journal of the
Society of Chemical Industry. Transactions and Communications. Vol. 58. June 1939. p. 203.
WELCH, J. H. and GUTT, w. The effect of minor components
on the hydraulicity of the calcium silicates. Proceedings of
the Fourth international Symposium on Chemistry of Cement,
Washington, D.C., 1960. Washington D.C., National
Bureau of Standards, 1962. NBS Monograph No. 43. Vol.
1. pp. 59-67.
AMERICAN SOCIETY FOR TESTING AND MATERIALS. ASTM
Designation: CI51-63. Standard method of test for autoclave expansion of Portland cement. 1964 Book of ASTM
Standards: Part 9. January 1964. pp. 163-169.
REHSI, s. s. and MAJUMDAR, A. J. The use of small specimens
for measuring autoclave expansion of cements. Magazine
of Concrete: Research. Vol. 19. No. 61. December 1967.
pp. 243-246.
GONNERMAN, H. F., LERCH, W. and WHITESIDE, T. M. Inl'estigations of the hydration expansion characteristics of Portland
cements. Chicago, Portland Cement Association, June 1953.
pp. 168. Research Department Bulletin 45.
MENZEL, c. A. Strength and volume changes of steam-curved
portland cement mortar and concrete. Journal of the
American Concrete Institute. Proceedings Vol. 31. November-December 1934. pp. 125-148.
BOGUE, R. H. The chemistry of Portland cement. Second
edition. New York, Reinhold Publishing Corporation,
1955. p. 361.

77

Magazine of Concrete Research: Vol. 21, No. 67 : June 1969

30.

The formation and phase composition of

Portland cement clinker. The chemistry of cements edited
by H. F. W. TAYLOR. London, Academic Press, 1964. Vol. 1.
p.90.

MIDGLEY, H. G.

31.

w. The steam curing of Portland cement

products. The chemistry of cements. London, Academic
Press, 1964. Vol. 1. pp. 417-432.
TAYLOR, H. F.

Contributions discussing the above paper should be in the hands

of the Editor not later than 31 December 1969.

78