Hydration of sodium phosphate-modified high alumina cement

Weiping Ma and Paul W. Brown

Materials Research Laboratory, The Pennsylvania State University, University Park,
Pennsylvania 16802-4801
(Received 27 August 1992; accepted 6 January 1994)

High strength can be achieved in high alumina cement (HAC) through the incorporation
of phosphate-based additions at levels of 10 and 20 wt. %. In order to establish the
mechanism that results in higher strength, the effects of a variety of condensed sodium
phosphates (NaPO 3 ) n , (NaPO 3 ) n • Na 2 O, Na 5 P 3 O 10 , and (NaPO 3 ) 3 were studied. The
influence of these additions on the kinetics of hydration was studied using isothermal
calorimetry. The phosphatic additions enhanced reactivity, but x-ray diffraction analyses
did not reveal evidence of new crystalline phosphate-containing hydration products.
Microstructural evolution was examined in real time using environmental SEM, and
hydration products exhibiting distinct morphologies were observed. The features exhibited
ranged from amorphic to polygonal shapes, plates, and fibers. These frequently formed
between crystalline calcium aluminate hydrate grains and by doing so appear to provide a
means to enhance the strengths of these cements. In spite of the morphological variations,
companion energy dispersive x-ray analysis showed that the compositions of these
products did not vary widely. Their ranges of compositions are 52-60 wt. % AI2O3,
20-26 wt. % P 2 O 5 , and 20-24 wt. % CaO.

I. INTRODUCTION 200 °C led to the formation of hydroxyapatite as the

It is generally recognized that conventional cements
exhibit useful compressive strengths but low tensile and
flexural strengths.1 The mechanical properties of cement-
based materials are controlled by the compositions of
starting materials, the manner in which these are mixed
with an aqueous media, the chemical bonding of the
reaction products, and the microstructures of the final
assemblage. Strategies to improve the mechanical prop-
erties of cement-based materials have been addressed by
a large number of studies.2
A wide range of compositions have been studied
in the development of phosphate-bonded materials.3"6
Phosphates have a long history of use as binding agents
in ceramic systems.7 Their reactions have resulted in the
formation of materials which have been used in such di-
verse applications as alumina refractories,8 bioceramics,9
and concrete repair.10 The [PO4] group provides a basic
building block that results in a series of compounds that
are a remarkable series of analogues to silicate com-
pounds formed on the basis of the [SiO4] building block.
Various phosphate-based cements, where HAC
was mixed with phosphoric acid and cured at elevated
temperature, were studied by Roy.11 Tensile splitting
strengths reached 27 MPa after 1 day of moist curing
at 90 °C. The chemical reactions at room temperature
between HAC and NH 4 H 2 PO 4 were studied by Sugama
and Carciello.12 They found that ammonium calcium
pyrophosphate is the principal product responsible for
the development of strength. Hydrothermal treatment at
major phase and anorthite as the minor one and con-
tributed to the development of compressive strengths in
excess of 70 MPa. We previously determined the flexu-
ral strengths of high alumina cement modified by sodium
phosphates.13'14 Cements containing 10 wt. % of these
constituents exhibited flexural strengths more than twice
that of corresponding unmodified cement cured under the
same conditions.13 The reactivities of these formulations
were high with rapid setting and strength gain.
High alumina cement differs from portland cement
in composition and exhibits properties that severely lim-
its its use. HAC is a valuable material for repair work
of limited life and in temporary works but no longer
used in structural concrete above or below ground.15 A
factor severely limiting its use is the strength loss that
occurs in applications at room temperature and above
or in humid conditions because of the conversion of
the hexagonal hydrate C 2 AH 8 to the more dense cubic
hydrate C3AH6.16"18 Strength regression also occurs in
phosphate-modified cements. However, our prior studies
have shown that strength regression can be substan-
tially reduced in phosphate-modified formulations. The
strengths of the modified cements always exceed those of
the unmodified cement. These findings suggest the pos-
sibility of formulating high strength HAC-based cements
in which significant strength regression is precluded.
The goal of the present study was to investigate
the kinetics of hydration, and the compositions and
morphologies of the products that form in phosphate-

J. Mater. Res., Vol. 9, No. 5, May 1994 1994 Materials Research Society 1291
 W. Ma and P.W. Brown: Hydration of sodium phosphate-modified high alumina cement

modified HAC's. Isothermal calorimetry was used to ex- 35.0

plore the rates of hydration; heat evolution during early a

A—Secar 71
hydration was measured for both phosphate modified and B-20% (NaPO3)n
unmodified HAC. The hydration products were identified C-20% (NaPO3)xNa2O
and characterized by x-ray diffractometry. Morphologi- D-20% Na5P30io
E-20% (NaPO3)3
cal development in real time was studied by hydration in
an environmental scanning electron microscope, and hy-
dration product compositions were determined by energy
dispersive x-ray analysis (EDX).

II. EXPERIMENTAL
The high alumina cement used in this investigation
is Lafarge Secar 71. Typical chemical analysis showed
71.3 wt.% A12O3, 26.8 wt.% CaO, 0.68 wt. % 12.0 18.0 24.0 30.0
Hours
SiO 2 , 0.46 wt.% MgO, 0.37 wt.% Na 2 O, 0.09 wt. %
Fe 2 O 3 , 0.04 wt.% SO 3 , and 0.02 wt. % K 2 O. X-ray 170.0 • 1 ' T ' 1 '
powder diffraction indicated that the cement principally A-Secar 71 b .
contains CA and CA2 as the hydraulic phases. B-20% (NaPO 3 ) n
Four condensed sodium phosphates, sodium hexa- 136.0 - C-20% (NaPO3)xNa2O
D-20% Na5P30io
metaphosphate (NaPO 3 ) n , sodium metaphosphate E-20% (NaPO3)3
(NaPO 3 ) n • Na 2 O, sodium tripolyphosphate Na 5 P 3 O 10 ,
I 102.0- B.,—-—'"'^___
and sodium trimetaphosphate (NaPO3)3, were used as
the phosphate-based materials.
Isothermal calorimetry was performed at 25 °C and | 68.0 -
w/s = 0.67 to compare the reactivities of the phosphate-
modified cements. In the method used, a known amount
of cement (typically 3 g) was placed in a gold-plated 34.0

copper calorimetric sample holder and then thermally

equilibrated to 25 °C. The water was separately equi-
librated and, once achieved, injected onto the cement,
and the rate of heat evolution was measured. X-ray
diffraction analyses were performed on a Scintag Auto-
mated x-ray diffractometer using CuK a i radiation. Both
environmental scanning electron microscopy (ESEM)
and conventional SEM were used in this study. ESEM
allowed microstructural evolution during hydration of
uncoated cements to be observed in real time, while the
chemical compositions of the hydrates were determined
by using conventional SEM-EDX analyses.
III. RESULTS AND DISCUSSION
A. Isothermal calorimetry
The heat evolved during the first 30 h of hydra-
tion was determined. Phosphate additions had a major
effect on the hydration of HAC, and the calorimetric
curves are shown in Fig. l(a). The figure shows that the
phosphate modification accelerated the initial hydration
rate regardless of phosphate composition. When the
sodium phosphates were present, the onset of the major
hydration peak occurred immediately after mixing. The
onset of this peak was delayed for about 16 h in the
absence of the phosphates. The total heat evolution
curves are shown in Fig. l(b). Significantly more heat
0.0
6.0 12.0 180 24.0 30.0
Hours
FIG. 1. (a) Comparison of the rate of heat evolved during the first
30 h of hydration for unmodified HAC and four phosphate-modified
cements, (b) Total heat for the unmodified cement and four phos-
phate-modified cements during the first 30 h of hydration.
was produced during the first several hours when the
sodium phosphates were present. However, after 30 h of
hydration, the cumulative heats evolved when HAC was
modified by the various phosphates were lower than that
from unmodified HAC. Strength measurements13 carried
out on the phosphate-modified cements after 24 h of
hydration indicated that the strengths achieved correlated
with the amount of heat evolved. Thus, the phosphate
modifiers have their greatest effects at the onset of
hydration. The nature of the effects of phosphate modifi-
cation were further studied by observing microstructural
evolution during this period of rapid hydration.
B. Microstructural analysis by ESEM
1. Unmodified HAC
The morphologies of hydration products of Secar 71
were first observed by ESEM. A relatively high water/

1292 J. Mater. Res., Vol. 9, No. 5, May 1994

 W. Ma and P.W. Brown: Hydration of sodium phosphate-modified high alumina cement
cement ratio (0.5) was used in order to keep the samples
moist for an extended period of time. The samples were
hand mixed and then immediately placed in the ESEM.
A crystalline calcium aluminate hydrate grows within
the first 3 h of reaction, but no further changes were ob-
served during the next 15 h of hydration. Figure 2 shows
the typical features which are polygonal crystallites of
various sizes. As will be discussed, this morphology is
very different from that of the modified cements.
2. HAC modified by (NaPO3)3
Figure 3 shows the microstructural evolution during
the hydration of HAC modified by (NaPO 3 ) 3 using
a phosphate/cement ratio of 0 : 1 . The first noticeable
change was the occurrence of microcracks after 11 min
of hydration [Fig. 3(a)]. The width of the cracks was
about 2 fim. A significant change was observed after
about 1 h of hydration, associated with the appearance
of a fibrous phase. This can be seen by comparing
Fig. 3(a) with Fig. 3(b) which are micrographs of the
same area. The fibers had diameters of 3-8 /j,m and
attained various lengths. The cracks shown in Fig. 3(a)
enlarged slightly to about 5 ^um during the next 3 h.
The experiments were carried out at a water vapor
pressure of 2.5 Torr, which is below the saturation
vapor pressure for water in vacuum. Although water is
present throughout the hydration process, it is likely that
crack formation was due to shrinkage during specimen
desiccation. In addition, since there is a high initial rate
of heat evolution, the sample heating may aggravate loss
of moisture.
During this 3 h interval, a new phase formed in
the cracks [Fig. 3(c)]. The morphology, shown at higher
magnification in Fig. 3(d), was very similar to that
of hydroxyapatite.19 Therefore, two new phases were
produced during early hydration in the HAC-(NaPO 3 ) 3
FIG. 2. Microstiucture of HAC after hydration for 20 h.
J. Mater. Res., Vol. 9, No. 5, May 1994
system, and they exhibit distinct morphologies. Signifi-
cantly, the formation of a fibrous phase may provide
a mechanism by which a composite microstructure, in
which microstructurally distinct phases form from a
single precursor, may be created.
3. HAC modified by Na5P3O10
The sample modified by Na 5 P 3 O 10 exhibited a mor-
phological development which was different from that
caused by Na(PO 3 ) 3 . Figure 4 shows the evolution in
morphology in the same area between 8 and 25 min
of hydration. It can be seen that morphology changed
significantly along with the crack during this period.
After 25 min, a distinct phase can be observed embedded
in a structureless matrix.
4. HAC modified by (NaPO3)n
The effect of the polymer (NaPO 3 ) n was also
studied. The changes in microstructure were small during
the first 20 h of hydration with small differences in the
micrographs. The first noticeable change observed was
the enlargement of cracks and the separation of grains
after 30 min of hydration. Small bubble-like features
then appeared on the grain surfaces. These events are
most likely the result of the slow evaporation of water
from the specimen. No further changes were observed
during the subsequent 20 h of hydration. Considering
both microstructural observations and heat evolution
characteristics together, the phosphate samples that
derived the most heat did not produce the greatest
microstructural changes. The presence of Na(PO 3 ) n
resulted in minimal microstructural evolution over the
first 20 h, but caused the greatest amount of heat to
be evolved. Our previous work13-14 has shown that
Na(PO 3 ) n -modified HAC also attained the highest initial
strength, but showed the greatest loss of strength due to
the conversion of metastable hydrates to cubic hydrate.
This suggests that the formation of the hydrates, C 2 AH 8
and CAH 10 , may not be responsible for the early strength
development. Therefore, phases that formed during early
hydration were determined.
C. Phase identification
1. X-ray diffraction
The phases that formed during the hydration of
phosphate-modified HAC were analyzed by x-ray dif-
fraction (XRD) and energy dispersive x-ray analy-
sis (EDX). Sequential x-ray diffraction patterns of
HAC-10% Na(PO 3 ) n hydrated between 4 min and 2 h
indicated the absence of the formation of crystalline
hydration products, although heat evolution during this
period was significant (Fig. 1). Figure 5 compares the
1293
 W. Ma and P.W. Brown: Hydration of sodium phosphate-modified high alumina cement

t .v 1 ^ it

(a) (b)
* </
# • • >

(c) (d)
FIG. 3. ESEM images of a (NaPO3)3-modified cement (a) after 10 min of hydration, (b) after 1 h and 9 min of hydration, (c) after 3 h and
13 min of hydration, and (d) higher magnification of the phase that has formed in the cracks.

diffraction patterns of Secar 71 and those of four modi-
fied samples with different phosphate after 30 months
of hydration along with that of anhydrous Secar 71. Al-
though there are significant differences in the mechani-
cal properties among these samples,13'14 the differences
in the x-ray diffraction patterns are negligible.
Two possible phenomena may explain this. One is
that the new products are x-ray amorphous. The other
is that phosphate-based additions only accelerate hydra-
tion, but that significant amounts of sodium or phosphate
do not become incorporated into the hydration products.
However, the calorimetric data shown in Fig. 2 indi-
cate that less total heat was evolved when phosphates
are present, and that the molecular weights of com-
positionally similar phosphates, Na(PO 3 ) 3 vs Na(PO 3 ) n ,
have a significant effect of heat evolution. Therefore,
although no x-ray detectable phase is produced, it is
likely that phosphates are incorporated into the hydration
products.
2. Hydration product composition
EDX was used to determine the compositions of the
hydration products. Figure 6 shows SEM micrographs
of HAC modified by 10 wt. % (NaPO 3 ) n [Fig. 6(a)]
and by 10 wt. % (NaPO 3 ) n • Na 2 O [Fig. 6(b)], and the
corresponding EDX spectra of selected hydration prod-
uct regions. The spectra show the presence of alu-
minum, phosphorus, and calcium in an amorphous phase.
Figure 6(c) shows an SEM micrograph and EDX spec-
tra of HAC modified by 10 wt. % Na 5 P 3 Oi 0 . Hexagonal
crystallites, which tended to grow in the pores and are

1294 J. Mater. Res., Vol. 9, No. 5, May 1994

 W. Ma and P.W. Brown: Hydration of sodium phosphate-modified high alumina cement
(a)
FIG. 4. ESEM images of a Na5P3O10-modified cement (a) after
8 min of hydration and (b) after 25 min of hydration.
surrounded by other hydration products, are present. The
EDX spectrum of the crystallite, marked "X", shows it
to consist principally of calcium and aluminum. Only
minor amounts of phosphorus and sodium are present,
suggesting the phase to be a hexagonal calcium alu-
minate hydrate, most likely C 2 AH 8 . The composition
of the matrix, which contains aluminum, phosphorus,
and calcium, is similar to that observed in the other
analyses.
The relative proportions of the constituents in the hy-
dration products are very similar for all four phosphate-
modified cements (Table I). These data indicate that
an amorphous calcium aluminate phosphate hydrate
(C-A-P-H) gel was produced during hydration. The
microstructure of this phase varied with the phosphate
present. Although the C-A-P-H gels formed from
J. Mater. Res., Vol. 9, No. 5, May 1994
(NaPO 3 ) n
-v—^_
(NaPO 3 ) n Na 2 O
Secar 71
29
FIG. 5. X-ray diffraction patterns of anhydrous HAC and HAC and
four phosphate-modified compositions hydrated for 30 months at
room temperature.
the four sodium phosphates investigated are not
compositionally identical, their components fall within a
narrow range (Table I). The gel appears to bond both the
anhydrous HAC grains as well as the calcium aluminate
hydrates. Compared to the compressive strength obtained
in unmodified HAC, 1314 the formation of C-A-P-H gel
improved structural integrity. Although there was still
a regression in the strength as the metastable calcium
aluminate hydrates undergo conversion to the cubic
hydrate, the losses in strength appeared to be mitigated
by the presence of this phosphate-containing hydration
product.
IV. CONCLUSIONS
Hydration of HAC is substantially accelerated when
sodium phosphates are present. However, the total heat
evolved in the hydration of unmodified HAC is ac-
tually greater. The phosphate-based additions do not
tend to promote the crystallization of calcium aluminate
hydrates during the high alumina cement hydration.
Although the hydration products are compositionally
quite similar, ESEM shows that the hydration of HAC
modified by different phosphates can result in mor-
phologically distinct hydration products. The phosphate-
containing products formed are x-ray amorphous. This
amorphous C-A-P-H phase serves to interlink the other
constituents that are present. Thus, the formation of
this phase provides enhanced structural integrity to high
alumina cement and reduces the strength regression
which occurs on conversion.
ACKNOWLEDGMENT
The authors gratefully acknowledge the support of
the Air Force Office of Scientific Research, Aerospace
Science Directorate.
1295
 W. Ma and P.W. Brown: Hydration of sodium phosphate-modified high alumina cement

I5KU 4.96KX 5U 8835 MA181

10Wt.%(NaPO3)n.Na2O

ENERGY (KeV) ENERGY (KeV)

(a) (b)

lOWt.%Na5P30lo

ENERGY (KeV) ENERGY (KeV)

i5KU S.83KX 5U 0020 MA2i 612

(c)
FIG. 6. Morphologies and the associated EDX spectra for (a) (NaPO3)n-modified HAC, (b) (NaPO 3 ) n • Na2O-modified HAC, and
(c) Na s P 3 Oi 0 -modified HAC.

1296 J. Mater. Res., Vol. 9, No. 5, May 1994

 W. Ma and P.W. Brown: Hydration of sodium phosphate-modified high alumina cement

TABLE I. Percentage of the oxides of phosphate-modified cements. 8. J. E. Cassidy, Ceram. Bull. 56, 640-643 (1977).
9. P.W. Brown and M. Fulmer, J. Am. Ceram. Soc. 74, 934-940
Oxide percentage (1991).
Admixtures 10. A. K. Sarkar, Ceram. Bull. 69, 234-238 (1990).
(P/C = 0 . 1 ) A12O3 P2O5 CaO 11. D. M. Roy, in Advanced Cements and Chemically Bonded Ceram-
ics, edited by M. Daimon, S. Somiya, G. Sudoh, and K. Takemoto
(NaPO 2 ) n • Na 2 O 60 20 20 (Mater. Res. Soc. Symp. Int. Proc. 13, Pittsburgh, PA, 1989),
(NaPO 3 ) n 59 21 20 pp. 213-227.
(NaPO 3 ) 3 54 22 24 12. T. Sugama and N.R. Carciello, J. Am. Ceram. Soc. 74 (5),
Na 5 P 3 O 10 52 26 22 1023-1030 (1991).
13. W. Ma and P.W. Brown, Cem. Concr. Res. 22, 1192-1200
(1992).
14. W. Ma and P. W. Brown, Proc. 9th Int. Congress on the Chemistry
REFERENCES of Cement, Delhi (1992), Vol. IV, pp. 424-429.
15. A. M. Neville, Properties of Concrete (Pitman Publishing Inc.,
1. H. F. M. Taylor, Cement Chemistry (Academic Press, Inc., London, 1981), pp. 92-99.
New York, 1990), pp. 316-330. 16. R. N. Edmonds and A. J. Majumdar, Cem. Concr. Res. 19,
2. D.M. Roy, Science 235, 651-658 (1987). 289-294 (1989).
3. J. Hu, D.K. Agrawal, and R. Roy, J. Mater. Res. 3, 772-780 17. R. N. Edmonds and A.J. Majumdar, Cem. Concr. Res. 18,
(1988). 311-320 (1988).
4. R. A. Steinke, M. R. Silsbee, D. K. Agrawal, R. Roy, and D. M. 18. S.M. Bushnell-Watson and J.H. Sharp, Cem. Concr. Res. 16,
Roy, Cem. Concr. Res. 21, 66-72 (1991). 875-884 (1986).
5. T. Sugama and L.E. Kukacka, Cem. Concr. Res. 13, 407-416 19. Y. Fang, D. K. Agrawal, D. M. Roy, R. Roy, and P. W. Brown,
(1983). J. Mater. Res. 7, 2294-2298 (1992).
6. P. W. Brown, J. Am. Ceram. Soc. 75, 17-22 (1992).
7. W. D. Kingery, J. Am. Ceram. Soc. 33, 239-250 (1950).

J. Mater. Res., Vol. 9, No. 5, May 1994 1297