TRANSPORTATION RESEARCH RECORD 1284 53

Early-Age Cement Hydration Reactions

p AUL WENCIL BROWN

The chemistry of the early reactions that occur during the hydra- BLENDED CEMENTS
tion of the phases in portland cement concrete are considered.
Desirable concrete properties are gained because of hydration of Eight types of blended cement, containing pozzolan or slag,
the cement. If pozzolanic or latent hydraulic admixtures are pres-
ent, additional hydration reactions will contribute to strength along with portland cement are designated in ASTM C 595.
development. The rate and products of cement hydration may Pozzolans are a source of reactive silica and, although they
be influenced by the presence of these or of chemical admixtures. are not typically hydraulic in themselves, pozzolans consume
Although emphasis is placed on portland cement concrete, reac- calcium hydroxide liberated during cement hydration with the
tions in other cements are briefly discussed. formation of additional binder phase. Commonly used poz-
zolanic materials include power plant fly ash, silica fume, and
certain natural pozzolans. Granulated blast furnace slag is
At the time of mixing, fresh concrete contains cement, fine
vitrified by rapid quenching and is a latent hydraulic material.
aggregate, coarse aggregate, and water. In addition, chemical
The highly alkaline environment of concrete pore solution
and mineral admixtures may be present. Chemical admixtures
activates the hydration of slag.
include accelerating, retarding, water-reducing, or air-entraining
admixtures; mineral admixtures include pozzolans, such as fly
ash and silica fume, and blast furnace-slag.
Concrete gains desirable mechanical properties principally HYDRATION REACTIONS IN MODEL SYSTEMS
as a result of the hydration of the cement present. However, SIMULA TING PORTLAND-BASED CEMENTS
if pozzolanic or latent hydraulic admixtures are present, addi-
tional hydration reactions will contribute to strength devel- Very Early Reactions
opment.
The rate of cement hydration, the composition of the prod- Although the consumption of all the anhydrous phases present
ucts of hydration, and the morphologies of these hydration in cement initiates at the time of mixing, some hydration
products may be influenced by the presence of such mineral reactions end very early . This may occur as a result of the
admixtures as well as by the presence of chemical admixtures . rapidity of these reactions , because the reacting anhydrous
The chemistry of the early reactions that occur during the solids are finely ground and immediately available to react
hydration of the phases in portland cement concrete are con- with the liquid phase or because only a minor amount of the
sidered . Early reactions are considered to be those that begin reacting phase is present. Hydration of calcium oxide, reac-
in the first 28 days after mixing. Although emphasis is placed tion of calcium sulfate hemihydrate to form gypsum, and the
on portland cements, reactions in blended cements, regu- formation of syngenite may be categorized in this way.
lated set cements, shrinkage-compensating cements, and high-
alnmina cements are briefly discussed.

Hydration of Free Lime

PORTLAND CEMENTS
Virtually all portland cement contains minor amounts of anhy-
Five types (I-V) of portland cement are designated in ASTM drous calcium oxide, CaO . Calcium oxide is present as a result
C 150 (J) . These contain five major compounds whose pro- of both the proportioning of the raw materials in the manu-
portions vary with cement type. Four of the compounds, which facture of cement and the nature of the pyroprocessing proc-
include impure forms of tricalcium silicate and dicalcium sil- ess. In the manufacture of cement, sources of calcium oxide
icate, an aluminate phase, and a ferrite phase, are formed and silica are finely ground, mixed, and pyroprocessed. Dur-
during the pyroprocessing of cement. A fifth compound, cal- ing the pyroprocessing step, the calcium oxide and silica ini-
cium sulfate, is added during the grinding of portland cement tially react to form dicalcium silicate, leaving excess calcium
clinker. This compound may be present as anhydrite, CaS0 4 ; oxide. Tricalcium silicate forms as a result of the reaction of
as calcium sulfate hemihydrate , CaSO.,· V2H 2 0; or as dihy- dicalcium silicate with calcium oxide. Often this reaction does
drate or gypsum, CaS0 4 ·2H2 0. not completely consume all the calcium oxide available . The
calcium oxide present in cement is called "free lime" and
reacts rapidly with water to form calcium hydroxide according
to
Department of Materials Science and Materials Resea rch Laboratory ,
Pennsylvania State University , University Park, Pa. 16802. (1)
54 TRANSPORTATION RESEARCH RECORD 1284

This reaction is expansive in that the solid volume of a mole

of calcium hydroxide exceeds that of a mole of calcium oxide
by a factor of about 2. The occurrence of this reaction while
concrete is in a plastic stage is of little consequence in terms 0.8
of the dimensional stability of the structure in question . How-
ever, in improperly pyroprocessed cements containing high
amounts of free lime , undesirable expansion is possible. If all o.e
the free lime in a cement is not immediately available to the .e. s•c
I/lo
mixing water, its hydration cannot occur until the anhydrous • 2s•c
phases surrounding the grains of free lime have been con- 0.4 o 41°C
sumed. Should this occur to a significant extent after the
concrete has hardened, the expansion associated with the
hydration of the free lime could lead to crack development. 0.2

Hydration of Calcium Sulfates

20 40 60 80
One or more forms of calcium sulfate (anhydrous, hemihy- TIME (minut•l
drous, or dihydrous) are interground with cement clinker to
FIGURE 1 Relative ra tes of conversion of
control the rate at which the aluminate phases in portland hemihydrate to gypsum depending on
cement hydrate. The chemical formulas for these compounds temperature.
are as follows:
Type Formula Compound
false set. However, if the mixing process is continued beyond
Anhydrous CaS0 4 Anhydrite the completion of the hemihydrate-to-gypsum conversion
Hemihydrate CaS04Y2H 20 Plaster of Paris process, the viscosity of the mixture decreases. This is called
Di hydrate CaS042H2 0 Gypsum
mixing through false set.
Anhydrous calcium sulfate is classified into two types accord-
ing to its dissolution rate. The terms "soluble anhydrite" and
"insoluble anhydrite" have come to be used to indicate the Syngenite Formation
rate at which a particular anhydrous CaS0 4 dissolves . In some
instances naturally occurring anhydrite or anhydrite formed Syngenite is a potassium sulfate-calcium sulfate double salt
by the dehydration of gypsum or hemihydrate may dissolve having the formula K2 S0 4 ·CaS04 ·H 2 0 . Syngenite formation
very slowly. If the anhydrite dissolution rate is less than that may rapidly occur in cements containing calcium langbeinite ,
necessary to control the rate of aluminate hydration , the anhy- 2CaS0 4 ·K2 S0 4 , according to (3)
drite is considered to be insoluble.
2CaS0 4 ·K2 S0 4 + nH 2 0~
Even if gypsum is interground with cement clinker, the heat
generated during grinding is frequently sufficient to cause the CaK 2 (S0 4 ),-H 2 0 + CaS0 4 ·2H 2 0 (4)
gypsum lo <.lehyuiale lo hemihydrate . In extreme instances
dehydration may also result in anhydrite formation. During and lead to accelerated set. Flash set has, for example, been
the batching of con'c rete, both soluble anhydrite and hemi- observed in cements to which K2 C0 3 has been added (4).
hydrate will react with the mixing water to form gypsum as Syngenite formation is favorable in pore solutions where the
follows : concentration of potassium exceeds 0.35 molar (5). This is
not uncommon in cements produced in the United States,
CaS0 4 + 2H 2 0 ~ CaS0 4 ·2H2 0 (2) especially in those containing alkali sulfates.
The occurrence of set associated with syngenite formation
CaS0 4 • 1/ 2 H 2 0 + l.5H 2 0 ~ CaS0 4 ·2H2 0 (3) is, as in the case of gypsum formation, in part a result of the
morphology of syngenite. Syngenite has a needlelike habit
Figure 1 shows relative rates at which hemihydrate converts and syngenite crystals are capable of causing a loss in plasticity
to gypsum, depending on temperature (2). At the fineness of by bridging water-filled spaces . It is not unreasonable that
hemihydrate used in this experiment , the conversion is vir- syngenite has sometimes been confused for ettringite ; both
tually complete by 30 min at 25 and 41 °C, and by about 1 hr have generally similar morphologies and strong x-ray diffrac-
at 5°C. Because calcium sulfates are relatively soft, they are tion peaks over the same angular ranges (6). Syngenite should
easily ground and tend to occur as fine particles or as layers persist as a stable phase in concrete until the concentration
on the surfaces of the cement particles. Thus, the hydration of sulfate in solution becomes depleted through the formation
of hemihydrate tends to be complete at shorter times than of ettringite.
those typically required for initial set of concrete . In some
instances the conversion of hemihydrate to gypsum can result
in a phenomenon called "false set," which occurs when the Silicate Hydration
gypsum crystals that have formed result in an apparent set of
the mix . Gypsum crystals tend to form as hexagonal needles The silicate phases collectively constitute about 80 weight
and their needlelike shape contributes to the occurrence of percent of portland cement. These phases are impure trical-
Brown 55

cium silicate, 3CaO·Si0 2 , called alite, and impure dicalcium

silicate, 2CaO·Si0 2 , called belite. It is the hydration reactions
of these silicate phases that are primarily responsible for the
development of concrete strength. Alite and belite hydrate
to form the same hydration products, calcium silicate hydrate
(C-S-H) and calcium hydroxide. However, both the rates of
reaction and the relative proportions of the hydration prod-
ucts formed are quite different. The rate of alite hydration is
significantly more rapid than that of belite . The relative rates
of hydration of the various phases in portland cement are
qualitatively shown in Figure 2 (7).
The hydration reactions for these phases may be expressed
as follows, once calcium hydroxide has precipitated:

3CaO·Si02 + 5.3H 2 0 ~

1. 7CaO·Si0 2 ·4H2 0 FIGURE 3 Water-rich portion of the ternary system CaO-

Si02-H20.
+ l.3Ca(OH)i (5)
by the equilibria with the solution. Eventually the solution
2CaO·Si0 2 + 4.3H 20 ~
composition will reach Point B. When this invariant point is
l. 7CaO·Si0 2 ·4H2 0 reached, the C-S-H forming will be at equilibrium with a
solution also saturated with respect to Ca(OH) 2 and will have
+ 0.3Ca(OH)i (6) a lime-to-silica ratio of about 1.7.] Dissolution of tricalcium
silicate or dicalcium silicate continues until the solution becomes
During the early stages of alite or belite hydration , the stoi- supersaturated with respect to calcium silicate hydrate. The
chiometry of the calcium silicate hydrate varies as shown in calcium silicate hydrate that initially precipitates has a cal-
Figure 3 (8). [First-formed C-S-H will have a lime-to-silica cium-to-silica ratio near 0.83. This is far lower than the value
ratio near 0.8. Because this ratio is less than 3, the solution of 1.7 shown in Equations 5 and 6. Because the calcium-to-
will become enriched in calcium. C-S-H formed subsequently silica ratio of the calcium silicate hydrate is always less than
will have higher lime-to-silica ratios. These are determined that in the anhydrous dicalcium silicate or tricalcium silicate,
the solution becomes enriched with calcium as hydration pro-
100
ceeds . To maintain equilibrium, the calcium-to-silica ratio in
~ the C-S-H also increases. Eventually, the solution becomes
supersaturated with respect to calcium hydroxide. However,
~ BO
the initial precipitation of this phase does not occur until the
!;;:
a: 60 calcium hydroxide concentration in solution is about 1.5 times
0
> saturation. The calcium-to-silica ratio of the C-S-H responds
J:
u. 40 to the calcium hydroxide supersaturation by reaching a value
0
w
w
possibly as high as 2. Eventually, as calcium hydroxide does
a: 20 precipitate and its concentration in solution approaches sat-
(!)
w uration, the system shown in Figure 3 approaches the invariant
0
Point B, and the calcium-to-silica ratio of the C-S-H reaches
TIME (daysl about 1.7. Thus, the values for the calcium-to-silica ratio given
<al in Equations 5 and 6 assume that this invariant point has been
reached.
100 Using tricalcium silicate as an example, a more general
~ equation describing silicate hydration may be written as fol-
z 80 lows:
0
~
a:
0 60
>
J:
u. 40 (CaO)xSi0 2 ·nH2 0
0
w
w
a: 20
+ (3 - x)Ca(OH) 2 (7)
(!)
w
0 This equation indicates the variability in the composition of
0 20 40 60 80 100 the C-S-H in response to the prevailing solution composition
TIME (days)
and, along with a companion equation for the hydration of
(b)
dicalcium silicate, is more relevant to the hydration of the
FIGURE 2 Degree of hydration versus time for (a) pure silicate phases in cement than are Equations 5 and 6. Variation
cement phases and (b) those phases in portland cement. in solution composition may be expected during cement
56 TRANSPORTATION RESEARCH RECORD 1284

hydration depending on the amount of free lime present, the of heat evolution associated with its rapid hydration may be
presence of pozzolans or slag, as well as the concentration of sufficient to cause the water to boil. The rapidity of this reac-
the alkalis, sodium, and potassium in solution. tion leads to a phenomenon in concrete called "flash set."
In addition to the compositional changes that occur during Flash set results because the products of aluminate phase
the hydration of the silicate phases in cement, mechanistic hydration cause a stiffening of the fresh concrete . Premature
and morphological changes occur. Indeed, the latter two stiffening associated with aluminate hydration makes place-
changes are closely related. From a mechanistic standpoint, ment and finishing difficult. The use of gypsum to control
tricalcium silicate hydration occurs in three stages. The first flash set goes back over a century to the studies of Michaelis
of these is responsible for the onset of the induction period in 1870 (13). The mechanism by which the hydration of the
and is the result of the formation of a layer of C-S-H on the aluminate phase in cement is retarded in the presence of
surfaces of the hydrating particles. The formation of this layer gypsum has remained a matter of debate (14) . However, it
causes hydration to become diffusionally controlled, with the is generally recognized that retardation is associated with the
rate of tricalcium silicate hydration depending on the rate of formation of ettringite, 3CaO·Al 2 0 3 ·3CaS04 ·32H2 0. The
diffusional transport through this layer. The rate of hydration reaction of tricalcium aluminate in the presence of gypsum to
during this stage is slow and does not appear to be strongly form ettringite may be written as follows:
dependent on fineness (9). This period of hydration is called
the "induction period ." Eventually this layer becomes unstable 3CaO·Al 2 0 3 + 3CaS04 ·2H 2 0 + 26H 2 0 ~
and a morphological transformation to an acicular, cauliflow-
erlike, C-S-H hydration product occurs (10). The specific rea- 3CaO·Al 2 0 3 ·3CaS0 4 ·32H2 0 (8)
son for this transformation remains a matter of speculation;
however, there is evidence to suggest that it is associated with The amount of calcium sulfate interground with cement
a polymerization reaction in which monomeric silicate tetra- clinker is carefully proportioned to ensure that ettringite for-
hedra in the C-S-H condense to form dimers (11). mation does not continue to any significant extent after about
Once the initial layer of C-S-H has undergone the trans- one day, when the concrete has started to gain strength. The
formation to an acicular product, the anhydrous tricalcium formation of ettringite is an expansive reaction. The expan-
silicate particle surfaces are again exposed to the solution. sion associated with ettringite formation is acceptable when
This results in the onset of a period of rapid hydration, referred concrete is in a plastic state and the formation of ettringite
to as the "acceleratory period." A significant fraction of the needles may indeed contribute to set. It is usually undesirable
alite in cement is consumed during the acceleratory period that an expansive reaction occur in hardening concrete because
with the actual amount consumed depending on the fineness dimensional instability and cracking may result. There are
of the cement. However, the acceleratory period eventually two reasons why ettringite formation is expansive. First, the
comes to an end and the rate of hydration decreases again as solid volume of the ettringite product phase exceeds the sum
the small particles of tricalcium silicate are consumed and as of the solid volumes of the tricalcium aluminate and gypsum
the layers of hydration products aro und the larger particles reactant phases. Second, ettringite needles tend to bridge
thicken. A transition to a "postacceleratory period" is com- pores in the hardening concrete. Once an ettringite needle
plete when the layers of C-S-H surrounding the hydrating has bridged a pore, its continued growth will generate a tensile
particles are sufficiently thick to cause hydration to again stress on the pore walls. These internal tensile stresses may
become diffusionally controllerl. Tt ;ippears that the hydration eventually lead to crack formation.
rate during this period is independent of fineness (12). Thus, A mechanism proposed for the retardation of tricalcium
in the hydration of cement, it appears that the primary dif- aluminate hydration in the presence of gypsum involves the
ference in behavior between a Type I or a Type II and a Type rapid formation of a layer of ettringite around the hydrating
III cement lies in the amount of hydration that occurs during tricalcium aluminate grain. However, the molar volume of
the acceleratory period and that grinding to higher fineness the ettringite is much larger than that of the tricalcium alu-
contributes principally to the amount of hydration occurring minate from which it has formed . This results in stresses that
during the acceleratory period. lead to cracking of the ettringite layer. As the cracks are filled
by additional reaction product, more cracks form. Eventually
all the gypsum is consumed. When this occurs, ettringite reacts
Aluminate Hydration with the remaining unhydrated tricalcium aluminate to form
the monosulfate phase, 3CaO ·Al2 0 3 ·CaS0 4 ·12H 2 0 , accord-
The percentage of the aluminate phase in portland cements ing to
may vary from essentially zero in so-called "zero C3A cements"
to as high as 14 percent or more.
The aluminate phase in cement is generally considered to
be an impure form of tricalcium aluminate. The principal
substitution that affects the stoichiometry of the aluminate
phase is that of sodium for calcium. However, in cements the
composition of the aluminate phase may also depart signifi- Supporting this mechanism is the observation that tricalcium
cantly from that of tricalcium aluminate because of the aluminate hydration is not effectively retarded in the absence
quenching of nonequilibrium compositions during c.:ooling. of ettringite formation (15). On the other hand, the needlelike
Tricalcium aluminate is a highly reactive phase. lftricalcium morphology of ettringite is not in accord with that of a phase
aluminate is mixed with water at room temperature , the rate that forms an effective diffusion barrier and thereby controls
Brown 57

the rate of tricalcium aluminate hydration. Recent work elu- Ferrite Phase Hydration
cidating the phase relationships in the system CaO-Al 20-CaS0 4 -
H20 may reconcile this (8,16). Formation of ettringite from The relative abundance of the ferrite phase in portl;md cements
tricalcium aluminate results in the existence of local equilibria is in the same range as that of the aluminate phase or slightly
in the interfacial region between the hydrating particle and lower. In the pure system, CaO-Al2 0 3-Fe 20 3 , the ferrite phase
the solution. At the exterior of the interfacial region ettringite is actually a solid solution having a generalized composition
is in equilibrium with the solution. Inside the ettringite layer of Ca 2 Fe( 2 - xiA1x0 5 where x varies between 0 and 0.6. The
is a layer of the monosulfate phase . Inside that is a layer of hydration behavior of the compound tetracalcium alumino-
calcium aluminate hydrates , and inside that, on the anhydrous ferrite, 4CaO·Al 2 0 3 ·Fe2 0 3 , has often been regarded as repre-
surface, is a layer of alumina gel. It is suggested that, in sentative of the behavior of ferrite solid solutions in general.
analogy with the induction period in the hydration of trical- However, in cement, as with the aluminate phase, a variety
cium silicate, it is the transport processes across the alumina of substitutions may occur in the ferrite phase , and this, cou-
gel layer that control the rate of tricalcium aluminate hydra- pled with quenching-in of nonequilibrium compositions, can
tion. Although an ettringite layer exists at the surface of the result in ferrite phase compositions remote from tetracalcium
hydrating particle, it in itself is not a diffusional barrier; rather aluminoferrite.
its existence allows the formation of an alumina gel diffusional Of the four major phases in portland cement, the hydration
barrier at the tricalcium aluminate interface. of the ferrite phase is the least well understood. Investigations
If the hydration of one mole of tricalcium aluminate is (17-19) of the hydration of tetracalcium aluminoferrite in the
carried out in the presence of three moles of gypsum, the presence of gypsum have reported the following hydration
reaction proceeds according to Equation 8 and the final prod- reactions, respectively :
uct is ettringite. If hydration is carried out in the presence of
one mole of gypsum, the reaction proceeds according to Equa-
tion 9 and the final product is monosulfate. At gypsum-to-
tricalcium aluminate mole ratios greater than 1 but less than + mH 2 0~
3, a mixture of ettringite and monosulfate forms. At gypsum- 4(3Ca0·{0.75Al 2 0 3 ·0.25Fe 2 0 3 }·3CaS0 4 ·32H2 0)
to-tricalcium-aluminate mole ratios less than unity, all the
available sulfate in the system will have reacted to form mono- (12)
sulfate according to Equation 9 before the anhydrous trical-
cium aluminate has been consumed . In this case a third set
of hydration reactions involving the formation of calcium alu-
minate hydrates will occur as follows : + mH 2 0~

(3Ca0·{0 .64Al 2 0 3 ·0.36Fe 2 0 3}·3CaS0 4 ·32H 2 0)

2(3Ca0·Al 2 0 3) + 21H 2 0 ~

2Ca0· Al 2 0 3 ·8H2 0

+ 4Ca0· Al 2 0 3 • 13H 2 0 (10) + 0.36Al(OH) 3 (13)

4Ca0·Al 2 0 3 ·Fe 2 0 3 + 6CaS0 4 ·2H2 0

Because all the available sulfate has already been consumed,
the hydration reaction represented by Equation 10 involves + 2Ca(OH) 2 + mH 2 0 ~
only the formation of calcium aluminate hydrates. Although
the formation of the hydrates 2CaO·Al 2 0 3 ·8H 2 0 and 2(3Ca0·{0.5Al2 0 3 ·0.5Fe 2 0 3}·3CaS04 ·32H2 0) (14)
4Ca0· Al 20 3 • 13H 20 is kinetically favored, these hydrates are
metastable with respect to the hydrate 3CaO·Al 2 0 3 ·6H2 0 at These reactions suggest a variable stoichiometry for the ettrin-
above about 15°C. As a consequence , 2CaO·Al 2 0 3 ·8H20 and gite-Iike hydration product. However, a recent investigation
4CaO·Al 20 3 ·13H2 0 will eventually transform to the stable (20) has shown that early hydration of C 4 AF in gypsum and
phase 3CaO·Al 2 0 3 ·6H2 0 according to lime-gypsum solutions results in the formation of iron-free
ettringite and a calcium ferrite gel according to
2CaO·Al 2 0 3 ·8H2 0 + 4CaO·Al 2 0 3 ·13H2 0 ~

2(3Ca0 · Al 2 0 3 ·6H2 0)

+ 9HP (11)
(15)

The situation is somewhat different for portland cement when

silica and ferric oxide are available. In cement , the calcium HYDRATION REACTIONS IN PORTLAND
aluminate hydrate 3Ca0·Al2 0 3 ·6H2 0 may form solid solu- CEMENTS
tions with a calcium ferrite hydrate and silica. The general
formula for this solid solution may be written as The reactions described for the pure phases or their combi-
nations to a large extent also occur in portland cements. There
are , however , important differences both in the rates of reac-
58 TRANSPORTATION RESEARCH RECORD 1284

tion and in the compositions and morphologies of the hydra- reactive silica to produce additional C-S-H. This reaction is
tion products. called the pozzolanic reaction, which can be described by the
None of the major phases in portland cement are pure; all following equation:
are solid solutions. Although the nature of the substitutions
varies, all these solid solutions contain alkalis. As a conse-
quence, except for the very earliest reactions, hydration in
cement occurs at pH values higher than would occur in the It is not possible to predict the stoichiometry of the C-S-H
model systems. This affects both the stabilities and the com- while the pozzolanic reaction is occurring, but compositional
positions of the hydration products that form. For example, analyses of C-S-H in the vicinity of reacting fly ash particles
if a cement pore solution contains alkali hydroxides in suffi- indicate the C/S ratio to be in the range of 1.2 to 1.4. This
cient concentration, the alkali-silica reaction will occur. How- represents an averaged value because the C/S ratio at the ash
ever, hydration of a mixture of pure phases in the presence particle-(C-S-H) interface is probably near 0.83, as Figure 3
of reactive aggregate will not lead to the alkali-silica reaction. would suggest, whereas the C/S ratio of the C-S-H at the
Although the hydration of cement produces significant (C-S-H)-pore solution interface is probably near 1. 7. If, when
amounts of calcium hydroxide, very little is observed in the the pozzolanic reaction has reached completion (i.e., the source
pore solutions extracted from cements. This is because the of reactive silica has been completely consumed) there is cal-
common ion effect with potassium and sodium hydroxides cium hydroxide remaining, the C/S ratio of the C-S-H should
depresses the calcium ion concentration. On the other hand, eventually reach about 1.7. If, on the other hand, the available
there is sufficient calcium present to preclude the formation calcium hydroxide is consumed and there is still unreacted
of the hydrated calcium aluminate phase, 2Ca0· Al 2 0 3 ·8H2 0. silica present, the C/S ratio of the C-S-H formed because of
Rather, the phase 4CaO·Al 2 0 3 ·13H2 0 forms. Other effects the pozzolanic reaction will be less than 1. 7.
associated with elevated alkali hydroxide concentrations in
solution include a decrease in the stability of ettringite and,
as discussed previously, a tendency for syngenite to form. Shrinkage-Compensating and Regulated-Set Cements
Major differences may occur in the hydration of the alu-
minate and ferrite phases when the behavior of the pure phases It was mentioned that the amount of calcium sulfate inter-
and the analogous phases in portland cement are compared. ground with portland cement clinker is proportioned to pre-
Figure 2 suggests that the ferrite phase in cement is rather clude ettringite formation after about the first day of hydra-
slow to react. On the other hand, it has been observed that tion. Although the expansive reaction associated with ettring-
the hydration of the phase tetracalcium aluminoferrite with ite formation can be accommodated while concrete is in a
gypsum occurs at a higher rate than that of tricalcium alu- plastic state, the occurrence of an expansive reaction in hard-
minate with gypsum (P. W. Brown, unpublished data). The ened concrete used in ordinary construction is undesirable.
reason for this difference is that the hydration reaction of the There are, however, exceptions to this and cements have been
ferrite phase is strongly retarded in the presence of calcium formulated based, in part, on the formation of ettringite.
hydroxide, whereas that of the aluminate phase is only mildly These cements fall into two general categories-shrinkage-
retarded. compensating and regulated-set cements.
The hydration of both the ferrite and the aluminate phases Shrinkage-compensating cements are formulated so that an
in cement is affected by the competition of the C-S-H for expansive reaction compensates for the drying shrinkage typ-
sulfate. In cement the molar ratio of calcium sulfate-to- ical in portland cement concretes. In cements of this type,
(aluminate + ferrite) is about 0.7. Thus, monosulfate phases ettringite is formed as a result of the reaction between gypsum
and hydrated calcium aluminate/ferrite phases would be and the compound 4Ca0·3Al 2 0 3 ·S0 3 according to the reac-
expected to be the terminal phases associated with aluminate tion
and ferrite phase hydration. However, a small percentage of
the silicate in C-S-H can be replaced by sulfate (21), and the 4Ca0·3Al 2 0 3 ·S03 + 8CaS0 4 ·2H20 + 6Ca0
formation of a sulfate substituted C-S-H results in the decom-
+ 80H2 0 ~
position of the monosulfate phases in cements. Finally, as
mentioned above, silicate also substitutes in the hydrated cal- 3(3CaO·Al2 0 3 ·3CaS04 ·32H2 0) (17)
cium aluminate/ferrite phases. As a consequence, the terminal
aluminate and ferrite phases in portland cement are hydro- Shrinkage-compensating cements are also referred to as
garnets. expansive cements or self-stressing cements.
Regulated-set cements contain the compounds that react
rapidly with sources of calcium sulfate to form ettringite. The
EARLY HYDRATION REACTIONS IN MODIFIED early strength gain in regulated-set cements is associated with
PORTLAND CEMENTS AND NONPORTLAND ettringite formation. After ettringite formation has reached
CEMENTS virtual completion, reaction of the silicate phases results in
continued strength gain. In one class of regulated-set cements,
Blended Cements 11Ca0·7Al2 0 3 ·CaF2 is added. This compound is rapidly
hydraulic and ettringite is rapidly formed when a source of
The principal difference between the hydration of portland calcium sulfate is present. In a second class, the compound
and blended cements is that the calcium hydroxide liberated 4Ca0·3Al 2 0 3 ·S0 3 is present but in higher proportion than
during silicate hydration reacts with the additives containing that used in shrinkage-compensating cements.
Brown 59

High-Alumina Cement 5. F. E. Jones. The Quinary System Ca0-Al 2 0 3 -CaSO.-K 2 0-H 2 0

at 25°C. Journal of Physical Chemistry, Vol. 44, 1944, pp. 356-
378.
High-alumina cement is a nonportland cement composed
6. P. W. Brown et al. The Hydration of Tricalcium Aluminate and
essentially of monocalcium aluminate, CaO·Al 2 0 3 . The Tetracalcium Aluminoferrite in the Presence of Calcium Sulfate.
hydration reaction in this cement is Materials and Structures, Vol. 19, No. 110, 1986, pp. 137-147.
7. S. Mindne s and J. F. Young. Concrete. Prentice-Hall, Engle-
wood liJ(s, N.J., 1981.
(18)
8. P. W. Brown. The Implications of Phase Equilibria On Hydration
in the Tricalcium ilicate-W11ter and the Tricalcium A luminate-
High-alumina cement offers the desirable features of high Gypsum-Water Systems. Presented at 8th International Congress
early strength and resistance to chemical attack. The source on the Chemistry of Cement, Vol. III, 1986, pp. 231-238 .
of resistance to chemical attack is the fact that the calcium 9. P. W. Brown, E. ranz, G. Frohnsdorff, nnd H. F. W . Taylor.
Analysis of the Aqueous Phase During Early Tricalcium Silicate
aluminate hydrate formed is of low basicity and is not attacked Hydration. Cement and Concrete Research, Vol. 14, 1984, pp.
by mild acids. 257- 262.
High-alumina cements perform adequately providing the 10. J. Ings and P. W. Brown. Early Hydration of Single Crystals of
exposure condition does not promote the conversion of the Tricalcium Silicate. Cement and Concrete Research, Vol. 13, 1983,
pp. 843-848.
monocalcium aluminate hydrate to tricalcium aluminate hexa-
11. S. A. Rodger et al. A Study of Tricalcium Silicate Hydration
hydrate. This conversion reaction may be written as follows: from Very Early to Very Late Stages. In Microstructural Devel-
opment During Cement Hydration (L. Struble and P. W. Brown,
3(CaO·Al2 0 3 ·10H2 0)----> 3Ca0·Al2 0 3 ·6H 2 0 eds.), Materials Research Society, Pittsburgh, Pa. (in press), pp.
13-20.
+ 2Al 2 0 3 ·3H 2 0 + l8H 20 (19) 12. P. W. Brown. The Influence of Particle Size Distribution on the
Kinetics of Tricalcium Silicate Hydration . Journal of the Amer-
ican Ceramic Society, Vol. 72, No. 10, 1989, pp. 1829-1832.
The monocalcium aluminate hydrate phase becomes increas- 13. W. Michaelis. Cited in H. Kanilre. Optimum Sulfate in Portland
ingly unstable with respect to tricalcium aluminate hexahy- Cement. In Cements Research l'rogress-19 4 (J. F. Young, ed.),
drate as the temperature is increased above room tempera- American Ceramic ociety, We Lerville , Ohio, 1985.
ture. This conversion reaction leads to a loss of strength that 14. P. W. Brown et al. The Hydration of Tricalcium Aluminate and
Tetracalcium Aluminoferrite in the Presence of Calcium Sulfate.
has resulted in structural failures (22). Materials and Structures, Vol. 19, No. 110, 1986, pp. 137-47.
15. M. Collepardi et al. Tricalcium Aluminate Formation in the Pres-
ence of Lime, Gypsum, and Sodium Sulfate. Cement and Con-
Miscellaneous Cements crete Research, Vol. 8, 1978, pp. 571-580, 1978.
16. P. W. Brown. Phase Equilibria and Cement Hydration. In The
Materials Science of Concrete, American Ceramic Society , Wes-
A variety of cements used for specialized applications, such terville, Ohio, 1989.
as white cements, oil well cements, alinite cements, and so 17. M. Fukuhara et al. Mechanism and Kinetics of C4AF Hydration
forth have not been considered in this paper. In addition, with Gypsum. Cement and Concrete Research, Vol. 11, 1981, pp.
the above discussion of blended, shrinkage-compensating, 407-414.
18. E. Andreeva and R. Sanzhaasuren. Investigation of the Processes
regulated-set, and high-alumina cements is not intended to
of Chemical Interaction on Aqueous Suspensions ofTetracalcium
be comprehensive and reference may be made to a variety of Aluminoferrite in the Presence of Gypsum Di hydrate (English
sources regarding these types of cements (19,23-26). translation). Colloid Jo1mwl USSR, Vol. 39, 1977, 197-202.
19. J. Bensted. The Hydration of Portland Cement. In Advances in
Cement Technology (S. D. Ghosh ed.), Pergamon Press, New
York City, 1983, pp. 307-347.
ACKNOWLEDGMENT 20. P. W. Brown. Early Hydration of Tetracalcium Aluminoferrite
in Gypsum and Lime-Gypsum Solutions. Journal of the American
The author gratefully acknowledges the support of the Air Ceramic Society, Vol. 70, 1987, pp. 493-496.
Force Office of Scientific Research, Electronic and Materials 21. D. Menetrier, I. Jawed, and J. Skalny . Effect of Gypsum on C3S
Hydration. Cement and Concrete Research, Vol. 10, 1980, pp.
Science Directorate.
697-702.
22. High Alumina Cement Concrete in Buildings. Current Paper CP34/
74. Building Research Establishment, Garston, England, April
REFERENCES 1975.
23. Expansive Cement Concretes-Present State of Knowledge.
1. Cement; Lime; Gypsum. Annual Book of ASTM Siandards, Vol. Journal of the American Concrete Institute, ACI Communication
04.01, 1986. 223, Vol. 67, 1970, pp. 583-610.
2. P. W. Brown and P. LaCroix. The Kinetics of Ettringite For- 24. F. Lea. The Chemistry of Cement and Concrete. Chemical Pub-
mation. Cement and Concrete Research, Vol. 19, 1989, pp. 879- lishing Company, New York City, 1971.
884. 25. A. Neville and P. J. Wainwright. High Alumina Cement Con-
3. Z. T. Jugovic and J. L. Gilliam. Early Hydration Reactions in crete . Wiley, New York City, 1975.
Abnormal Setting Cements. Journal of Materials, Vol. 3, 1968, 26. S. Mindness and J. F. Young. Concrete. Prentice-Hall, Engle-
pp. 517-537. wood Cliffs, N.J., 1981.
4. E. M. M. G. Nial. The influence of Alkali-Carbonate on the
Hydration of Cement. Presented at 5th International Congress Publication of this paper sponsored by Committee on Mechanical
on the Chemistry of Cement , Vol. II, 1968, pp. 472-486. Properties of Concrete.