Upload
gedix5
View
19
Download
1
Embed Size (px)
DESCRIPTION
Influence of metal chloride salts on calcium aluinate cement hydration
Citation preview
Advances in Cement Research, 2012, 24(5), 249–262
http://dx.doi.org/10.1680/adcr.11.00012
Paper 1100012
Received 08/03/2011; revised 20/07/2011; accepted 17/08/2011
Thomas Telford Ltd & 2012
Advances in Cement ResearchVolume 24 Issue 5
Influence of metal chloride salts on calciumaluminate cement hydrationUkrainczyk, Vrbos and Sipusic
Influence of metal chloridesalts on calcium aluminatecement hydrationNeven UkrainczykPost-doctoral Researcher, Faculty of Chemical Engineering andTechnology, University of Zagreb, Zagreb, Croatia
Nevenka VrbosAssistant Professor, Faculty of Chemical Engineering and Technology,University of Zagreb, Zagreb, Croatia
Juraj SipusicAssociate Professor, Faculty of Chemical Engineering and Technology,University of Zagreb, Zagreb, Croatia
In this paper the influence of alkali, alkaline earth and transition metals in the form of chlorides salts on the
hydration of commercial iron-rich calcium aluminate cement (CAC) is investigated. The effect on setting time and
mechanical properties is studied. The results of the setting time obtained from the measured temperature
evolution of cement pastes are compared to standard Vicat needle measurements. Addition of alkali metal salts
both accelerates the setting time of CAC and deteriorates long-term strengths in the order of decreasing ionic
radius of the metals. The effect of alkaline earth metal salts on setting behaviour depends on the amount of
addition: low concentrations retard, but high concentrations accelerate setting behaviour. Calcium is found to have
the strongest acceleration effect with the lowest deterioration in strength. An interesting regularity is observed
when plotting the results of both retarded setting times and strengths against atomic number of added transition
metal cation.
NotationA aluminium oxide (Al2O3)
a, b fitting parameters
C calcium oxide (CaO)
CAC calcium aluminate cement
F iron oxide (Fe2O3)
H water (H2O)
K equilibrium constant
m cement mass (g)
M molar mass (g/mol)
PC Portland cement
pK log (K)
R ionic radius
R correlation coefficient
rn nucleation rate (s�1)
S silica (SiO2)
T temperature (8C)
t time (h)
ti time of initial setting obtained by Vicat measurement
tf time of final setting obtained by Vicat measurement
tmax time of peak maxima in calorimetric measurement
t0 time of initial setting obtained by calorimetric
measurement
w mass fraction of added salt relative to weight of
cement (%)
w/c water-to-cement mass ratio
Z atomic number
ª molar concentration of added salt per kg of cement
(mmol/kg)
IntroductionThe duration of the suitable workability of cement-based mater-
ials during placement is defined by setting time. The use of
additives to modify the properties of cement-based materials for
each specific application is a subject of great practical and
financial interest. From practice, it is known that even small
amounts of different organic and inorganic compounds have a
pronounced influence on the setting behaviour and mechanical
properties of cement-based materials. Calcium aluminate cement
(CAC) as a special cement is particularly interesting because of
its versatility in high-performance applications (Bensted, 2002;
George, 1983; Mangabhai, 1990; Mangabhai and Glasser, 2001;
Scrivener et al., 1999), such as those requiring: high early
strength, resistance to chemical attack, resistance to abrasion,
refractory properties and/or low ambient temperature placement.
In comparison to commonly used Portland cement (PC), CACs
show relatively slow setting but rapid hardening characteristics.
These excellent properties make CAC products useful for rapid
repairs of roads, hydraulic dams, industrial floors, pipe linings
and other such purposes (Scrivener et al., 1999).
Hydraulic hardening of CAC is primarily due to the hydration of
CA (cement notation: C ¼ calcium oxide (CaO), A ¼ aluminium
oxide (Al2O3), F ¼ iron oxide (Fe2O3), S ¼ silica (SiO2),
249
H ¼ water (H2O)), but other compounds may also participate in
the hardening process, especially in long-term strength develop-
ment (Ukrainczyk, 2010; Ukrainczyk and Matusinovic, 2010).
The hydration of CAC is highly temperature dependent, yielding
morphologically different hydration products at different tempera-
tures of hydration. At ambient temperature metastabile hydrates,
CAH10 and C2AH8 (Ukrainczyk et al., 2007) transform to the
more stable C3AH6 and AH3 with consequent material porosity,
permeability increase and loss of strength. The transformation is
accelerated by temperature and moisture availability for the
dissolution and re-precipitation processes to take place (Mangab-
hai and Glasser, 2001).
Along with rapid setting, some metal salts result in the rapid
strength development of CACs at very early ages, so they are
advantageously used for rapid repair works of highways and
airport runways (Justnes, 2008). The influence of metal salts on
cement hydration, particularly setting time and compressive
strengths, is also of interest in stabilisation of hazardous waste
containing potentially soluble heavy metals (Jantzen et al., 2010).
Upon water addition to cement a sharp increase of the pH of the
water solution occurs, less for CAC than for PC hydration, which
causes precipitation of highly insoluble metal oxide–hydroxide,
thus opening an attractive route for heavy metal waste stabilisa-
tion. Lastly, the role of foreign ion additions in the nucleation and
growth process of hydration products provides a fundamental
insight into this intricate and enigmatic process. The admixtures
can interfere with hydration of cement in a variety of ways. A
number of different mechanisms have been suggested (e.g.
Currell et al., 1987; Murat and Sadok, 1991; Rodger and Double,
1984) to explain the action of set regulators on cement hydration,
mainly with respect to the dissolution of anhydrous cement, and
the nucleation and precipitation processes for the main hydration
products. The influence of the particular additive on setting
behaviour of CAC is generally not so well understood and
contradictorily statements can be found in the literature, as
detailed below. Part of the controversy could arise primarily from
differences in the cement composition, the measurement tech-
nique employed and the conditions of hydration.
Rodger and Double (1984) studied the influence of alkali,
alkaline earth and transition metal chlorides additions (25 mmol
of salts per kg of cement) on setting of iron-rich CAC pastes
prepared with a water-to-cement ratio (w/c) of 0.5 by measuring
the maximum temperature rise in a semi-adiabatic calorimeter
(T ¼ 22 � 28C, ˜T . 1008C). A similar chemical composition of
CAC is used in the present work (with 10% F (iron oxide)) but
from a different producer, namely Lafarge Ciment Fondu. The
summarised effect on setting time is in the following series:
Liþ ,, Mg2þ , Ba2þ , none , Kþ , Naþ , Sr2þ , Ca2þ:
Furthermore, Rodger and Double reported that all transition
metals chloride salts have a retarding effect on the set of CAC.
Using a semi-adiabatic calorimeter (T ¼ 18 � 28C, ˜T , 208C)
Currell et al. (1987) investigated the setting of iron-rich CAC
(manufactured by Lafarge, with a relatively low iron oxide
content of 6% compared with the CAC used in the present study)
pastes with a w/c ¼ 0.27 in the presence of many additives with
different concentrations. The effects on the setting time may be
summarised as follows: Liþ ,, Naþ , none , Kþ , Caþ ,
Mg2þ , Sr2þ:
Griffiths et al. (1991) studied the setting of iron-rich CAC (Lafarge,
similar chemical composition as in the present work except for
somewhat lower iron oxide content at 10%) pastes (w/c ¼ 0.4) at
different temperatures in the presence of 15 mmol/kg metal chlor-
ides and sulfates, as well as sea water. Using the time corresponding
to the maximal rate of heat generation to compare the setting
behaviour in the presence of chlorides, they reported the following
order at the various temperatures: Mg2þ , Kþ , Naþ , Ba2þ ,
none , Ca2þ at 158C; Mg2þ , Kþ , Naþ , none , Ca2þ ,
Ba2þ at 208C and 308C.
The literature regarding the influence of alkali and alkaline earth
metal chlorides on the hydration of iron-rich CAC is confusing.
The above-stated conflicting results can be summarised as
follows. For alkali chlorides sodium (Na) and potassium (K) are
reported to act as both retarders and accelerators. For alkaline
earth metal chlorides magnesium (Mg) acts as an accelerator,
according to Rodger and Double (1984), and as a retarder,
according to Currell et al. (1987). Only Rodger and Double
(1984) investigated the effect of both strontium (Sr) and barium
(Ba), which were reported to act differently, that is as retarder
and accelerator, respectively.
Although the influence of admixtures on setting behaviour of
iron-rich and refractory Secar 71 CACs, as two different classes
of CACs, has been compared in the literature, care should be
taken because they differ substantially in their properties (princi-
pally chemical and mineralogical composition). Murat and Sadok
(1991) considered the role of foreign cations in the hydration
kinetics of refractory CAC (Secar 71, produced by Lafarge, that
has an aluminium oxide (Al2O3) content of about 70%). They
concluded that all chlorides metal salts (mono-, bi-, tri- and tetra-
valent) have an accelerating effect on the set of Secar 71. This
result differs from most other studies summarised above. Nilfor-
oushan and Sharp (1995) investigated the setting of Secar 71 with
a range of concentrations (12.5–250 mmol/kg) of alkaline earth
metal chloride salts at different temperatures. At 128C all of the
alkaline earth metal chlorides investigated exhibited retardation
of setting, which increased with increasing concentration of the
admixture. At 208C, at low concentrations, they were shown to
accelerate the setting, but they acted as retarders at higher
concentrations. At 288C and 368C, they retarded the setting, the
more so with increasing concentration of the admixture.
In the present paper, the authors have studied the influence of
alkali, alkaline earth and transition metal chlorides additions on
setting behaviour as well as strength development of commercial
iron-rich CAC.
250
Advances in Cement ResearchVolume 24 Issue 5
Influence of metal chloride salts oncalcium aluminate cement hydrationUkrainczyk, Vrbos and Sipusic
Experimental work
Materials
This paper examines the hydration of three samples of commer-
cial CAC ISTRA 40 taken from a regular production of Calucem,
Pula, Croatia. The cement has the oxide mass fraction composi-
tion listed in Table 1. Physical properties of the cement are given
in Table 2. The effect of alkali, alkaline earth and transition metal
chlorides was studied on cement samples A, E and T, respec-
tively. The main compounds are CA and ferrite phase (C4AF–
C6AF2), with mayenite, C12A7, gehlenite, C2AS, and �-C2S as
minor compounds. All specimens were prepared with a water-to-
cement mass ratio (w/c) that corresponds to a water requirement
for the standard consistency (given in Table 2). The salts used as
additives were commercial Analar grade reagents, dissolved in
freshly deionised water before mixing with the cement. Alkali
metal salts, NaCl, KCl, RbCl and CsCl, were mixed with a mass
fraction (relative to weight of cement) of 0.05% and molar
concentration of 3 mmol/kg of cement, and LiCl from 0.00025 to
0.50% (0.06–118 mmol/kg). Alkaline earth metal salts: MgCl2
6H2O, CaCl2, SrCl2 6H2O and i BaCl2 2H2O, were mixed with
mass ratios (relative to weight of cement) of 0.10, 0.50, 1.0, 2.0
and 5.0%. Transition metal salts, MnCl2 H2O, CoCl2 6H2O,
CuCl2 2H2O and ZnCl2, were studied at 10 and 100 mmol/kg of
cement. To prepare the samples of pastes the additives were
weighed into deionised water according to a specified concentra-
tion and then mixed with cement.
Mechanical properties
Cement paste specimens (40 3 40 3 160 mm) were prepared
according to ASTM C 109 (with a standard consistency).
Bending and compression tests at different ages were done
according to EN 1015-11. Three specimens were tested for each
age.
Setting time measurement
The setting time was determined using the standard Vicat needle
method carried out according to EN 196-3. The experiments used
a standard consistency CAC paste, as given in Table 2.
CAC temperature evolution measurement set-up
The specimens were cast in high-density polyethylene containers
with inner diameter 2R ¼ 31 mm, 50 mm high and 0.7 mm thick.
The high-density polyethylene container was fully filled with the
cement paste (with a standard consistency), continuously applying
vibrations in order to minimise air entrapment. The thermocouple
measuring end was placed exactly at the centre of the container
(r ¼ 0 and z ¼ h/2 as shown in Figure 1). This was designed by
fixing a thin (1.3 mm), wooden support at the axis of the
container. Specimens were carefully sealed with lids and poly-
vinyl chloride electrical insulation tape and placed vertically in
the temperature-controlled water bath T ¼ 208C (�0.038C). Be-
fore mixing, the cement and water components were left to reach
thermal equilibrium (overnight) in a sealed container placed in a
thermostat.
Temperature evolution measurements were also performed simul-
taneously on the specimens tested for setting time, using a
standard Vicat apparatus. After the Vicat mould (DIN standar-
dised truncated conical form: 75 mm in diameter at the base,
65 mm at the top and 40 mm high, made up from non-absorbent
plastic with a glass base plate) was filled with the cement paste,
the thermocouple sensing end was placed at the weight centre,
that is through the axis at half-thickness of the specimen. This
was done by immersing a thin (1.3 mm) wooden support through
the axis of the paste in the mould. Specimens were tested at
laboratory conditions of T ¼ 20 � 28C.
The K-type thermocouples used were 0.2 mm thick with
Sample CaO Al2O3 Fe2O3 FeO SiO2 TiO2 MgO SO3 Na2O K2O Sum
A 37.10 38.47 14.39 2.90 4.43 1.05 0.90 0.20 0.14 0.17 99.8
E 40.2 39.0 11.7 4.3 1.9 n.a. n.a. n.a. 0.13 0.16 97.4
T 39.12 37.78 13.17 2.46 4.42 1.80 0.91 0.30 0.18 0.17 100.3
Table 1. Chemical composition of investigated CAC
Sample .90 �m: % ,40 �m: % Blaine, cm2/g Specific
gravity:
g/cm3
Setting time: min Water-to-cement mass
ratio of standard
consistencyInitial Final
A 3.76 80.50 3407 3.19 264 279 0.240
E 3.70 80.10 3421 3.19 300 322 0.240
T 3.71 80.90 3563 3.20 209 236 0.250
Table 2. Physical properties of investigated CAC
251
Advances in Cement ResearchVolume 24 Issue 5
Influence of metal chloride salts oncalcium aluminate cement hydrationUkrainczyk, Vrbos and Sipusic
grounded, twisted-shielded wiring. An eight-channel data logger
with a 20 bit resolution was used to collect the temperature
measurements. This stores the entire set of temperatures once
every second. The experimental data are simultaneously trans-
ferred to a personal computer. The cold junction temperature held
at room temperature was sensed by a precision thermistor in good
thermal contact with the input connectors (on thermal block) of
the measuring instrument. In order to have accurate cold junction
compensation, any change of its temperature was kept as slow as
possible.
Results and discussion
Alkali metals
Measurements of the Vicat initial and final setting times, that is ti
and tf respectively, were obtained as described in the Setting time
measurement section. The effect of addition of lithium chloride
(LiCl) in a range of concentrations from 0 to 118 mmol/kg on the
setting of CAC is presented in Table 3. The data show that
lithium chloride accelerates the setting time of CAC even for
very low additions of only w ¼ 2.5 3 10�4%, which corresponds
to a molar concentration of ª ¼ 0.06 mmol/kg. As the addition of
LiCl increases, both of the setting time parameters continuously
decrease. With the mass fraction of 0.1% the setting time occurs
during the mixing time, whereas with the fraction of 0.5% the
setting of CAC occurs immediately after the addition of lithium
chloride.
The results on the investigated effect of various alkali metal
chlorides on the setting time of CAC are presented in Table 4. It
is obvious that lithium metal shows a drastic effect of accelera-
tion in comparison to other alkali metals. The reason for this is
discussed next.
The hydration process of CA, as the principal mineral in CAC, is
generally believed to occur through initial dissolution, formation
of a metastable gel and subsequent precipitation (Sorrentino et
al., 1995; Taylor, 1998), principally CAH10: After dissociation of
CA, the metastable gel formed will acquire stability by condensa-
tion of mono-coordinated hydroxyl groups linked to aluminium
to form ox-obridges between two aluminium centres, leading to
the crystalline CAH10: Solid-state nuclear magnetic resonance
Motor
Thermostat
Cooling
T constant�
Pt100
Pt100 datalogger
TC datalogger
PC
Vicatneedle
rh
2R
Figure 1. Schematic diagram of the CAC paste temperature
evolution measurement set-up
w (LiCl): % ª: mmol/kg ti: s tf: s
0 0 15840 16720
0.00025 0.06 13800 15000
0.00050 0.12 10560 11330
0.00075 0.18 9930 10850
0.0010 0.24 9545 10267
0.0025 0.59 4330 5750
0.0050 1.18 3030 3700
0.0075 1.77 1060 1810
0.010 2.36 790 1433
0.025 5.90 338 413
0.050 11.8 270 293
0.10 23.6 F F
0.50 118 I I
Note: F denotes setting of the CAC with lithium chloride during themixing; I denotes setting of the CAC immediately after addition oflithium chloride
Table 3. Comparison of setting time for different quantities of
lithium chloride addition
252
Advances in Cement ResearchVolume 24 Issue 5
Influence of metal chloride salts oncalcium aluminate cement hydrationUkrainczyk, Vrbos and Sipusic
imaging data indicate that aluminium in CA is entirely four-
coordinated, but the hydration products of CA, CAH10 and
C3AH6 contain six-coordinated aluminium (Taylor, 1998), while
C2AH8 also contains four-coordinated aluminium located in the
interlayer of this double-layer hydroxide phase (Ukrainczyk et al.,
2007). Cong and Kirkpatrick (1993) have shown that the hydra-
tion of CAC proceeds by way of conversion of four to six-fold
coordinated aluminium and the occurrence of intermediate
phases, which still need further clarification. The ox-obridge
condensed structure will be affected by alkali metal cations
forming coordination linkage with the hydroxyl groups (Currell
et al., 1987). Of the alkali ions studied, Liþ behaves differently
from the other alkali cations because of its ability to form
tetrahedral symmetry with hydroxyl groups, while Naþ, Kþ, Rbþ
and Csþ will form octahedral symmetry (Wells, 1962). This is
proved by the present experimental results (Table 4), because
lithium ions show a drastic effect, while differences between the
other cations are not great and exhibit a definitive trend. The
setting time of CAC decreases in the following order:
Cs . Rb . K . Na .. Li. This order is in agreement with the
decrease of ionic radius of the metals. The induction period
during the precipitation of CAC hydration products from a
supersaturated solution is a reflection of the nucleation barrier to
the formation of these compounds (Barret et al., 1974). The
accelerating effect of lithium has been attributed to removal of
this barrier, caused by an initially fast precipitation of lithium
hydrometaaluminate (Matusinovic et al., 1994; Rodger and
Double, 1984), which acts as a heterogeneous nucleation sub-
strate, promoting the nucleation of the hydration products.
The influence of alkali metal chloride addition on the flexural and
compressive strength of CAC pastes has been studied. The results
of measurements are given in Figures 2–4. The results of
compressive strength measurement at early ages of CAC pastes
and CAC pastes prepared with different fractions of lithium
chloride are shown in Figure 2. It can be seen that along with
rapid setting, lithium salts cause strength development at early
ages. The precipitation of lithium hydrometaaluminate is respon-
sible for the rise of compressive strengths of CAC samples at
very early ages (t , 10 h) resulting in higher strengths compared
w ¼ 0.05% ª ¼ 3 mmol/kg
ti: s tf: s ti: s tf: s
Lithium chloride 270 293 712 1371
Sodium chloride 8560 8860 10 960 11 293
Potassium chloride 10 727 11 357 12 027 12 927
Rubidium chloride 11 860 12 560 11 965 12 660
Caesium chloride 11 940 12 820 11 940 12 820
Table 4. Comparison of setting times parameters for alkali metal
chlorides additions (0.05% mass fraction and 3 mmol/kg)
LiCl
w 0·050%�w 0·025%�w 0·010%�CAC
60
50
40
30
20
10
00 5 10 15 20
t: h
Com
pres
sive
str
engt
h: M
Pa
Figure 2. Evolution of compressive strength during early ages of
CAC pastes made with different mass fractions of lithium chloride
13
12
11
10
9
8
7
60 5 10 15 20 25 30
t: d
Flex
ural
str
engt
h: M
Pa
CACLiNaKRbCs
w 0·05%�
Figure 3. Effect of alkali metal chlorides on evolution of flexural
strength of CAC pastes
253
Advances in Cement ResearchVolume 24 Issue 5
Influence of metal chloride salts oncalcium aluminate cement hydrationUkrainczyk, Vrbos and Sipusic
to the reference (plain) CAC sample. The increase of the strength
is caused by precipitation of lithium hydrometaaluminate, which
acts as a heterogeneous nucleation substrate and promotes the
nucleation of the calcium aluminate hydrates. After about 15 h
the strength of the specimens with the lithium addition is lower
than the reference strength. A higher mass fraction of LiCl causes
lesser long-term compressive strengths (Figure 2).
Based on the setting time results in Table 3 the mass fraction of
alkali metal chloride addition of w ¼ 0.05% was taken for further
research. The influence of alkali metal chlorides on evolution of
flexural and compressive strength of CAC pastes is presented in
Figures 3 and 4, respectively. It can be seen that all alkali metal
salts decrease compressive and flexural long-term strengths of
CAC, compared to the reference (plain) CAC paste. The extent of
this reduction in strength follows the same order as for the setting
time reduction: Cs , Rb , K , Na ,, Li (which is in agree-
ment with the order of the ionic radius of the metals). It is worth
noting that although the lithium ion exhibits a drastic setting time
reduction (Table 3), the corresponding long-term strength reduc-
tion (Figures 3 and 4) is only somewhat more pronounced in
comparison to other alkali cations. The strength reduction pattern
for lithium has a higher reduction effect, while differences
between the other cations are not great and exhibit a definitive
trend. This is similar, but not as drastic to the setting behaviour.
The strength reduction with alkali addition can be attributed to a
more rapid nucleation and growth of hydration products because
of the smaller crystalline sizes of hydration products and thus
smaller surface contact area, as well as to the higher degree of
the transformation reactions of the metastable hydration products
because of the higher self-heating of the specimens. At ambient
temperature metastable hydrates transform to the more stable
C3AH6 and AH3 with a release of water and a consequent higher
degree of hydrated CA. This transformation is accelerated by
temperature and moist conditions. During the hydration of CAC,
a large quantity of total heat is liberated in the first 24 h of
hydration, which causes a considerable temperature increase of
prepared specimens (Banfill, 1995; Ukrainczyk and Matusinovic,
2010). Such uncontrolled temperature increase of specimens
could also contribute an increase in the amount of the formed
C2AH8 when the lithium ion is added (Matusinovic et al., 1994;
Rodger and Double, 1984).
The results of the present study are in agreement with Griffiths et
al. (1991) who reported that sodium (Na) and potassium (K) act
as accelerators (15 mmol/kg). On the other hand, Rodger and
Double (1984) indicated that sodium and potassium behave as
retarders (25 mmol/kg), while Currell et al. (1987) reported
sodium as an accelerating but potassium as a retarding admixture.
It has to be noted that Griffiths obtained results from conduction
calorimetry at carefully controlled conditions of temperature,
while Rodger and Double (1984) and Currell et al. (1987)
performed semi-adiabatic calorimetric measurements at 22 � 28C
and 18 � 38C, respectively. Owing to a high sensitivity of setting
time variations with temperature, such poorly controlled tempera-
ture conditions could explain the confusing results in the
literature. Furthermore, in Griffiths et al. (1991) and Rodger and
Double (1984), the authors did the comparison analysis of the
setting behaviour according to the time of maximal heat genera-
tion (or temperature) tmax instead of t0, although the parameter t0
better corresponds to the real initial setting time as obtained by
Vicat measurements. Indeed the effect of salt addition may
decrease the hydration rate and postpone tmax as a result of the
flattening and stretching of the shape of the main hydration peak
(as discussed in the Transition metals section). Moreover, in
different boundary conditions, that is semi-adiabatic (Currell et
al., 1987; Rodger and Double, 1984) or semi-isothermal (Griffiths
et al., 1991), the heat (or temperature) evolution curve is
significantly different, resulting in different tmax values.
The setting time of CAC decreases in the following order:
Cs . Rb . K . Na .. Li, which correlates with the ionic ra-
dius of the metals. To correlate experimental results of the
influence of various cation additions on the CAC (Secar 71)
hydration, Murat and Sadok (1991) employed mathematical
modelling based on a modification of the expression of nucleation
free enthalpy (relying on a hypothesis of homogeneous nuclea-
tion), which in the dilute solution approximation leads to the
following single equation
ln (rn) ¼ aþ b=R1:
where rn is the nucleation rate approximated to (1/t0), where t0 is
the experimentally determined setting time, R is the ionic radius
and a and b are fitting parameters. This model (Equation 1) has
been previously verified on experimental results of calcium
sulfate hydration (anhydrite and hemihydrate).
For the alkali metals we have investigated a correlation according
to Equation 1. The results are shown in Figure 5. The lithium ion
80
75
70
65
60
55
50
Com
pres
sive
str
engt
h: M
Pa
0 5 10 15 20 25 30t: d
CACLiNaKRbCs
w 0·05%�
Figure 4. Effect of alkali metal chlorides (w ¼ 0.05%) on
evolution of compressive strength of CAC pastes
254
Advances in Cement ResearchVolume 24 Issue 5
Influence of metal chloride salts oncalcium aluminate cement hydrationUkrainczyk, Vrbos and Sipusic
is omitted from the regression analysis because of its drastically
different behaviour, as already detailed above. As expected, from
Figure 5 it is obvious that constant mass concentration is not an
adequate parameter for the analysis. For w ¼ 0.05% the corre-
sponding molar concentrations for lithium (Li), potassium (K),
sodium (Na), rubidium (Rb) and caesium (Cs) are 11.8, 8.6, 6.7,
4.1 and 3.0 mmol/kg, respectively. This significant increase of
molar concentrations from the reference concentration of
3.0 mmol/kg results in further/higher increase of the setting time.
For a constant molar concentration (3 mmol/kg) the regression
analyses resulted in a reasonably good correlation coefficient
(R ¼ 0.92927).
Alkaline earth metals
Measurements of the initial and final setting time parameters, that
is ti and tf , respectively, was obtained as described in the Setting
time measurement section. In order to compare the results, a
range of the prepared mass concentrations (fractions relative to
weight of cement) were transformed to molar concentrations, as
shown in Table 5. The effect of a range of molar concentrations
from 4.8 to 525 mmol/kg on the setting time is presented in
Figure 6. It can be observed that lower concentrations (0.1% or up to 10.5 mmol/kg, according to Table 5) resulted in a retarda-
tion of the setting behaviour. Addition of 0.5% of salts for
calcium (Ca) and strontium (Sr) brought about setting times close
to that in the absence of any admixture, while barium (Ba) and
magnesium (Mg) showed significant acceleration. A further in-
crease of admixture concentration results in a dramatic accelera-
tion of the setting time in the presence of any of the four salts.
Furthermore, from Figure 6 it can be seen that calcium chloride
(CaCl2) has the strongest acceleration effect in comparison with
the other alkaline earth metal chlorides, while magnesium
chloride (MgCl2) has the weakest acceleration effect. The highest
concentration (w ¼ 5%) of calcium (Ca) and strontium (Sr)
resulted in a flash set during mixing. As can be seen in Table 5
the addition of a constant mass fraction of salts results in
different molar concentrations that complicate a further detailed
analysis. Unfortunately, the present experiments on the influence
of alkaline earth metal chlorides were conducted with a constant
mass concentration, not fixing the molar concentrations.
The results presented here are compared with the available
literature results obtained on Lafarge iron-rich CACs, focusing on
3·0
2·5
2·0
1·5
1·0
0·5
0
�0·5
�1·0
�1·5
ln(1
/)t i
w 0·05%�γ 3 mmol/kg�Linear fit
ln(1/ ) 1·36 24/t ri � � �
R 0·92927�
Cs� Rb�K�
CAC
Li�
0·006 0·008 0·010 0·012 0·0141/ : pmr �1
Na�
Figure 5. Investigation of the correlation according to Equation 1;
ti is measure in hours (horizontal line: value for plain CAC)
w 0.1% 0.5% 1.0% 2.0% 5.0%
ª: mmol/kg
MgCl2 3 6H2O 10.50 52.52 105.03 210.06 525.15
CaCl2 9.01 45.05 90.11 180.21 450.53
SrCl2 3 6H2O 6.31 31.54 63.08 126.17 315.42
BaCl2 3 2H2O 4.80 24.01 48.02 96.05 240.12
Table 5. Calculated molar concentrations in mmol/kg of CAC for
a corresponding mass fraction
MgCaSrBa
400
300
200
100
00 100 200 300 400 500
γ: mmol/kg CAC
Sett
ing
time:
min
Figure 6. Influence of alkaline earth metal chloride addition on
the initial setting time of CAC pastes (horizontal line: value for
plain CAC)
255
Advances in Cement ResearchVolume 24 Issue 5
Influence of metal chloride salts oncalcium aluminate cement hydrationUkrainczyk, Vrbos and Sipusic
the range of concentrations from 15 to 25 mmol/kg in Figure 6
that correspond to the literature experimental conditions (which
are summarised in the introduction). In agreement with the
finding of the present study, calcium (Ca) is reported by all three
groups of researchers (Currell et al., 1987; Griffiths et al., 1991;
Rodger and Double, 1984) to act as a retarder. The present
study’s results also confirm the different behaviours of strontium
(Sr) and barium (Ba) ions as reported by Rodger and Double
(1984). The differences in the literature regarding the effect of
alkaline earth metal chlorides could be attributed to the change of
the set behaviour influence (i.e. acceleration or retardation) with
the molar concentration, as well as sensitivity to different
chemical compositions of the similar iron-rich cements investi-
gated, temperature of hydration and method of obtaining the
setting parameters.
Moreover, it is interesting to compare the present study’s results
to the literature results obtained on a completely different class of
CAC, a pure CAC Secar 71 (produced by Lafarge, that has a high
aluminium oxide (Al2O3) content of about 70%). Nilforoushan
and Sharp (1995) reported an inverted (opposite) trend of the
results in Figure 6, that is contrary to the present findings. They
showed that alkaline earth metal additions in low concentrations
(12.5 mmol/kg and 25 mmol/kg) exhibit acceleration of the
setting, and significant retardation at higher concentrations (from
50 mmol/kg to 250 mmol/kg). Moreover, Murat and Sadok
(1991) reported that all chloride metal salts (mono-, bi-, tri- and
tetra-valent) have an accelerating effect on the setting of Secar
71, which also differs from most other studies on iron-rich CACs
as summarised in the introduction. Although the influence of
additions on setting behaviour of these two different classes of
CAC has been compared in the literature, resting on the hypoth-
esis of the same hydration mechanism of CA mineral, care should
be taken in this interpretation because they differ substantially in
their properties, principally chemical and mineralogical composi-
tion. Refractory Secar 71 CAC, produced by Lafarge, has an
aluminium oxide (Al2O3) content of about 70%, resulting in CA
and CA2 as the main mineral phases with no C12A7, C4AF and
C2S. The results of the current study have shown that the
influence of alkaline earth metal chlorides additions on setting
behaviour of iron-rich CAC and refractory Secar 71 CAC is
completely different, with an opposite effect of salts concentra-
tion on setting behaviour. This suggests that the hydration
mechanism of the two CACs may be quite different. This
difference may be principally explained by the C/A ratio of the
liquid phase during the nucleation period. The dissolution of the
‘pure’ main mineral CA shows the C/A ratio to be in the range
1.05–1.10 (Fujii et al., 1986; Nikushchenko et al., 1973) thereby
inevitably initiating the precipitation of AH3 gel. Even slight
amounts of C12A7 or CA2 that accompany CA in the commercial
cement composition as separated phases induce a variation of the
C/A ratio in the liquid phase. It is known that the presence of
small quantities of C12A7 (in iron-rich CAC) or CA2 (in Secar
71) influence the duration of the induction period: C12A7
accelerates the nucleation, whereas CA2 retards it. This has been
shown by hydrating a series of synthesised CA samples prepared
with C/A ratios ranging from 0.96 to 1.04 (Galtier and Guilhot,
1984). The dissolution of the ‘pure’ C12A7 exhibits a C/A ratio of
exactly 1.71, showing unambiguously the congruency of its
dissolution. Upon hydration of CA2, large amounts of AH3 gel
are produced (Edmonds and Majumdar, 1989) as a first hydration
product. Later, when the lime concentration reaches a sufficient
level, C2AH8 precipitates coinciding with an acceleration of the
hydration reaction rate. Thus, the chemical composition of ce-
ment influences the change of the C/A ratio in the liquid phase.
With an increasing C/A ratio the length of the induction period
diminishes. Because alkaline earth metal cations bear the same
charge they behave in a chemically similar way, thus the addition
of alkaline earth metal salt increases ‘apparent calcium’ concen-
tration, thus modifying C/A ratio in the solution. Increasing the
C/A ratio of iron-rich CAC induces the formation of AFm and
AFt phases that may protect the cement grains surface from
dissolution and results in the deceleration of the hydration.
The mass concentration of w ¼ 1% was chosen to investigate the
influence of alkaline earth metal salt additions on the flexural and
compressive strength of CAC pastes. The results of measurements
are presented in Figures 7 and 8. Alkaline earth metal salts
decrease compressive and flexural long-term strengths of CAC,
compared to the reference (plain) CAC paste. Interestingly,
although CaCl2 (with the second highest molar concentration,
Table 5) results in the strongest acceleration of the initial setting
time (t0 ¼ 69.5 min) the compressive and flexural strengths
exhibit the lowest reduction from the reference ones (with no
admixtures). On the other hand, addition of 1% magnesium
chloride (MgCl2) with the highest corresponding molar concen-
tration (Table 5) results in a much lower acceleration of the
setting time (t0 ¼ 163.0 min) but higher strength reduction than
calcium chloride (CaCl2). Moreover, strontium chloride (SrCl2)
and barium chloride (BaCl2) (with the smallest molar concentra-
tions) show no significant acceleration effect (Figures 7 and 8)
12·512·011·511·010·510·0
9·59·08·58·07·57·06·56·05·5
0 5 10 15 20 25 30
t: days
Flex
ural
str
engt
h: M
Pa
CACMgCaSrBa
w 1%�
Figure 7. Effect of alkaline earth metal chlorides (w ¼ 1%) on
evolution of flexural strength of CAC pastes
256
Advances in Cement ResearchVolume 24 Issue 5
Influence of metal chloride salts oncalcium aluminate cement hydrationUkrainczyk, Vrbos and Sipusic
but exhibit the highest strength reduction (Figures 7 and 8).
These results indicate the role of the foreign cations, which can
replace the calcium in the hydration products and/or even form
novel hydration products that change the rate of hydration
development and influence strength development. Further re-
search is needed to investigate the nature of the hydration
products formed (e.g. on the hydration of the pure phases with
foreign cation additions).
Temperature evolution and setting time measurements
The rapid hydration and the resulting heat evolution allows an
accurate determination of both the induction time t0 and the time
tmax of the maximum temperature (or heat generation) hydration
peak. Setting can be attributed to the interlocking of the hydration
products when their formation takes place to an appreciable
extent. Unlike calorimetry or temperature rise measurements, the
Vicat initial setting time is significantly affected by the water-to-
cement ratio attributable to the consistency of the paste being
tested. This reflects the fact that the setting time is an arbitrarily
defined parameter. In a more fluid paste, a longer time is needed
to reach the required resistance to penetration. As suggested by
Stegemann and Buenfeld (2001) preferable techniques for meas-
uring setting time, although less traditional and indirect, are
isothermal calorimetry, temperature rise and solution conductivity
measurements. In this paper times of initial, t0, and final, tf , sets
for experimental series on transition metals were established from
the obtained time–temperature curves, as illustrated in Figure 9.
Initial set is acquired from the intersection of two straight lines
(Bushnell-Watson and Sharp, 1986): one fitted through the
induction period of the curve and the other fitted through the
inflection point of the rising slope of the main peak. The final set
was approximated as the point of maximal heat generation.
In order to conduct the comparison of the values of setting time
parameters obtained by Vicat method and temperature rise meas-
urements, the temperature of the specimen in a Vicat mould was
measured as described in the CAC temperature evolution mea-
surement set-up section. The experiment on plain cement paste
(without admixtures) was repeated five times. It was found that
the difference between the initial setting time obtained by the two
methods, (ti � t0) was only 1.2 min � 0.6 min. This validates the
further use of the employed alternative method to obtain accurate
values of the setting time parameters. Interestingly, the time of
final set tf was found to be 9 min � 1.5 min below the point of
inflection (Figure 9).
An example of a temperature evolution of a specimen during
simultaneous standard Vicat method measurement is shown in
Figure 9. The influence of the sample mass and heat transfer
boundary conditions can be analysed by comparing Figures 9
and 10 (for the plain CAC). After moulding the cement paste
and fixing of the thermocouple within a few minutes, a decrease
of temperature is occurring owing to the dissipation of the heat
generated by the initial wetting of cement, chemical dissolution
0 5 10 15 20 25 30
t: days
CACMgCaSrBa
w 1%�
Com
pres
sive
str
engt
h: M
Pa
90
85
80
75
70
65
60
55
50
45
Figure 8. Effect of alkaline earth metal chlorides (w ¼ 1 %) on
evolution of compressive strength of CAC pastes
tmax
tf
Inflectionpoint
0 2 4 6 8 10 12 14 16 18 20 22 24t: h
20
22
24
26
28
30
32
34
T:C°
t0
Figure 9. Temperature evolution of specimen during standard
Vicat method measurement
γ(Me ) 10 mmol/kg CAC2� �CACCo Mn
Cu
Zn
CACCoMnCuZn
23
22
21
20
5 10 15 20 25t: h
T:C°
Figure 10. Effect of transition metal chlorides on the temperature
evolution of CAC, ª ¼ 10 mmol/kg of CAC
257
Advances in Cement ResearchVolume 24 Issue 5
Influence of metal chloride salts oncalcium aluminate cement hydrationUkrainczyk, Vrbos and Sipusic
and partial hydration before reaching the induction period with
very small heat generation (Banfill, 1995; Ukrainczyk and
Matusinovic, 2010). The hydration of CAC at temperatures
below ,208C shows the high influence of water-to-cement ratio
on hydration rate and the obtained hydration degree due to the
high stoichiometric water requirement for the formation of
metastable hydration products (Bensted, 2002; Fryda et al.,
2001; Ukrainczyk and Matusinovic, 2010). The cement hydration
is incomplete if there is not enough water for stoichiometric
hydration, so the hydration rate decreases because of the free
water consumption and reaches zero value upon water insuffi-
ciency. In non-isothermal conditions (Figure 9) the temperature
increase is higher due to lower heat loss to the environment. The
‘shoulder’ in Figure 9 before reaching tmax is a consequence of
the significant commencement of the transformation reactions
(Fryda et al., 2001). As the rate of transformation reactions of
metastable products to the stable C3AH6 and AH3 is significantly
increasing with the temperature, the associated free water
liberation allows further reaction of cement and thus increases
the hydration rate (Bensted, 2002; Fryda et al., 2001; Ukrainczyk
and Matusinovic, 2010). This also explains why the time of
maximal temperature rise, tmax, is reached significantly later after
the onset of the acceleration period in Figure 9 than in Figure
10, but with a faster diminution of the hydration rate after the
maximum at tmax:
The high temperature sensitivity of the setting time measurement
of CACs is not adequately addressed in most of the literature.
One point is the poor control of the starting temperature (e.g.
Currell et al., 1987; Rodger and Double, 1984), probably the
result of following the Vicat standard procedure, which is written
for measurements on a PC that does not have such pronounced
temperature sensitivity as CAC. A second point is that, owing to
a rapid hydration of CAC with a corresponding high heat
generation effect (about 20 mW/g in isothermal conditions (Ban-
fill, 1995; Ukrainczyk and Matusinovic, 2010)), the temperature
evolution of the material strongly depends on the heat transfer
boundary conditions around the testing specimen. In different
boundary conditions, that is semi-adiabatic as in work by Currell
et al. (1987) or Rodger and Double (1984) or semi-isothermal as
in work by Griffiths et al. (1991) or Murat and Sadok (1991), the
heat (or temperature) evolution curve is significantly different,
which could result in significantly different tmax values. Some
authors (Currell et al., 1987; Griffiths et al., 1991; Rodger and
Double, 1984) did the comparison analysis of the setting behav-
iour according to the time of maximal heat generation (or
temperature) tmax instead of t0, although parameter t0 better
corresponds to the real initial setting time as obtained by Vicat
measurements. For an isoperibol calorimeter, the evolved heat in
time t equals the accumulated heat in the calorimeter (that results
in the increase of the calorimeter temperature) and the heat loss
to the surroundings (a constant-temperature environment). Con-
duction (semi-isothermal) calorimeters have small temperature
rise of the samples, whereas semi-adiabatic calorimeters have
small heat transfer to the surroundings.
Transition metals
To the best of the current authors’ knowledge, Rodger and
Double (1984) are the only researchers who investigated the
effect of transition metal chlorides on iron-rich CAC (manufac-
tured by Lafarge). They reported the time of maximum tempera-
ture rise of CAC samples (w/c ¼ 0.5 with 25 mmol of salt per kg
of CAC) in semi-adiabatic conditions. The present authors
analysed the results of Rodger and Double against the added
metallic atomic number, as depicted in Figure 11. A well-defined
trend can be observed in the influence of transition metals on
setting behaviour. It is well known that various trends are
observed when studying chemical reactions or some physical
properties within the transition metal group. For example,
enthalpy change accompanying solvation of cations by water
molecules or melting point of pure metals, and so on. The cause
of such behaviour is qualitatively explained by the gradual filling
of the 3d-orbital by electrons taking into account Hund’s rule of
maximum multiplicity. Interestingly, the regularity of the setting
times observed in Figure 11 follows the same principle. In order
to clarify this finding, previously not addressed in the literature,
part of this paper is devoted to investigating further the influence
of transition metals, in the form of chloride salts, on the hydration
of similar (same class) CAC from a different manufacturer
(namely, Istra cement, Calucem).
The equilibrium between precipitated and soluble species is in
fact dependent on the acid/base properties of the precipitate. The
present authors argue that the acid/base properties of the
precipitate cause the observed regularity of setting times plotted
against atomic number in Figure 11. The acidity of the precipitate
can be described by the following chemical reaction in base
media
Me OHð Þ2 cr:ð Þ þ OH� aq:ð Þ $ Me OHð Þ3� aq:ð Þ2:
1000
800
600
400
25 26 27 28 29 30Atomic number, Z
t T(
): m
inm
ax
6
5
4
3
2
pKa
CAC
p curveKa
t T( ) curvemax
Mn2�
Fe2�
Co2�
Ni2�
Cu2�
Zn2�
Figure 11. Variation of time of maximum temperature rise of
CAC samples (see text for details) and acidity of crystalline
Me(OH)2 quantified by pKa value, plotted against the added
cation’s atomic number
258
Advances in Cement ResearchVolume 24 Issue 5
Influence of metal chloride salts oncalcium aluminate cement hydrationUkrainczyk, Vrbos and Sipusic
The values of acidity constants, pKa, are obtained from thermo-
dynamic data (Shkol’nikov, 2005) and plotted in Figure 11. These
values for pKa correspond to the fully crystalline hydroxide
precipitate. The observed discrepancy for copper and zinc could
be explained by the fact that freshly precipitated amorphous
hydroxide has higher solubility and is therefore expected to be
more acidic. Aluminium also participates in this type of acid/base
equilibrium according to the following reaction
Al OHð Þ3 cr:ð Þ þ OH� $ Al OHð Þ4� aq:ð Þ3:
The reaction in Equation 3 has pKa ¼ 5.9 (Nordstrom and May,
1989). Calcium hydroxide has no observable acidic properties.
By comparing the trends of the setting time and pKa change
plotted against atomic number (Figure 11) it can be concluded
that they agree only qualitatively. Generally, the discrepancies in
quantitative agreement can be ascribed to the complexity of the
real chemistry behind the studied process of CAC hydration.
Overall, the suggested explanation is that the more the acid/base
properties (real pKa values) of transition metal precipitates
deviate from the aluminum hydroxide; the setting time of the
CAC is more prolonged. This concept of acid/base reactions and
the influence on the setting time of calcium aluminate cement is
also corroborated by Mohmel and Gessner (1997), who showed
that acid/base properties of corundum filler significantly influence
the setting of CAC.
The influence of transition metal chloride salts on hydration was
investigated by measuring the temperature evolution of CAC
pastes (as described in the CAC temperature evolution measure-
ment section). Results of temperature evolution measurements for
pastes prepared with ª(Me2þ) ¼ 10 mmol/kg are shown in Figure
10. No significant temperature rise is observed during the induc-
tion period owing to a small and constant rate of heat generation
(about 0.3 mW/g (Banfill, 1995; Ukrainczyk and Matusinovic,
2010)). The induction period is followed by the onset of the
accelerated stage of reaction due to massive precipitation of
hydrates. The maximal temperature rise is reached in only a few
hours after the onset of the accelerated stage. With the addition of
the transition metal chlorides the main hydration peak of the
temperature evolution curve becomes flatter and longer (with a
lower value of maximal temperature at tmax) in the following
order: Co , Mn , Cu , Zn. This flattening and stretching effect
of the main hydration peak supports the notion of slower growth
and crystallisation of hydration products. The effect of various
transition metal chlorides on the setting time of CAC is investi-
gated for two concentrations, 100 mmol/kg and 10 mmol/kg. The
concentration of 100 mmol/kg showed no setting behaviour within
30 h of hydration. The influence of transition metal chloride
addition (ª ¼ 10 mmol/kg) on the setting time of CAC pastes is
depicted in Figure 12. The results are plotted against the atomic
number. It is obvious that all investigated transition metal
chlorides act as retarders, which is in accord with results obtained
by Rodger and Double (1984). The retardation effect is higher for
higher concentrations. In agreement with the finding in Figure 11,
the experimental data from the present study reproduced the
defined trend of the change of setting times against the added
cation’s atomic number well. A comparison of the obtained results
with the literature (Rodger and Double, 1984) is depicted in Figure
13. The results are plotted against the reciprocal values of ionic
radii (Shannon, 1976) in order to test the validity of Equation 1,
which relies on an assumption of homogeneous nucleation (Murat
and Sadok, 1991). The results of Rodger and Double have lower
(more negative) values of ordinate, which corresponds to the
longer absolute setting time values (compare Figures 10 and 11),
primarily owing to the much lower water-to-cement mass ratio
employed in this investigation (i.e. w/c ¼ 0.25 as opposed to
0.50). The translation of the results shows very good agreement of
the trend between the present results and the results obtained
previously by Rodger and Double (1984). It is clear that the
Initial setting time, ti
tmax
Zn2�
Cu2�
Co2�
Mn2�
25 26 27 28 29 30Atomic number, Z
Sett
ing
time:
min
200
300
400
500
600
700
Figure 12. Influence of transition metal chloride addition
(ª ¼ 10 mmol/kg) on the setting time of CAC pastes (horizontal
lines: values for plain CAC)
ln(1
/)
t 0
�1·2
�1·4
�1·6
�1·8
�2·0
�2·2
�2·4
�2·6
�2·8
�3·00·0135 0·0140 0·0145 0·0150 0·0155 0·0160 0·0165
1/ : pmr �1
Current paperRodger and Double (1984)Zn
Cu
Ni
Mn
CoFe
Figure 13. Comparison of results obtained in the present study
with those presented in the literature; ti is measured in hours
259
Advances in Cement ResearchVolume 24 Issue 5
Influence of metal chloride salts oncalcium aluminate cement hydrationUkrainczyk, Vrbos and Sipusic
experimental data significantly deviate from the model in
Equation 1.
On the other hand, on the Secar 71 as a different class of CAC,
Murat and Sadok (1991) reported that transition metal chlorides
have an accelerating effect. Again, as discussed in the Earth
alkali metals section, the influence of transition metal chloride
additions on setting behaviour of iron-rich CAC and refractory
Secar 71 CAC is shown to be completely different. This suggests
that the hydration mechanisms of these two classes of CAC are
significantly different (as discussed in the Earth alkali metals
section). Murat and Sadok (1991) have explained the acceleration
effect by an hypothesis of a heterogeneous nucleation process,
the nuclei formed being grains of metallic hydroxide (or hydrated
aluminates for Liþ and Ni2þ ions in the form of double-layer
hydroxide phases) that lead to an increase of the nucleation rate.
In the present case of iron-rich CAC, the retardation effect may
be explained by the massive precipitation of protective gel
(amorphous hydroxides and/or double-layer hydroxide phases)
coatings on the phase boundary surface of cement grains (Rashid
and Turrillas, 1997), which block hydrolysis and inhibit growth
of the crystalline hydration products. The effect of flattening and
stretching of the main hydration peaks observed in Figure 10
could thus be explained by a formation of larger quantities of,
and/or more impermeable, gel coatings that hinder crystallisation
and growth of the hydration products.
The influence of transition metal salt additions on the flexural
and compressive strength of CAC pastes was studied. The results
of measurements of flexural and compressive strength at 1, 7 and
28 days are presented in Figures 14 and 15, respectively.
Horizontal lines represent experimental values of strength for the
plain CAC, which are increasing with age. Generally, the addition
of transition metal salts decreases compressive and flexural
strengths of CAC, compared to the reference (plain) CAC paste.
When comparing Figures 14 and 15 with Figure 12 it can be
concluded that the magnitude of strength reduction for all tested
ages follows the order observed for the setting time retardation.
For example, the cobalt (Co) sample, which has the setting time
the closest to the plain CAC, achieves the smallest reduction in
strength. On the contrary, the zinc (Zn) addition results in the
highest retardation of the setting time and exhibits the highest
reduction in strength.
Integration of the temperature change evolution curve (curves
shown in Figure 10 were subtracted by 208C before integration)
gives an approximation of a relative amount of achieved degree
of hydration. It was observed that the relative degree of hydration
after 24 h has a tendency to lower in the following order:
none . Mn . Co . Cu . Zn. This order is in agreement with
the magnitude of the strength reduction in Figures 14 and 15.
Thus, the strength reduction may be primarily attributed to the
lower hydration degree achieved owing to the retardation of the
hydration reaction rate (Figure 10).
ConclusionsAddition of alkali metal salts accelerates the setting time of CAC
in the following order: Cs , Rb , K , Na ,, Li. This order is
in agreement with the increase of ionic radius of the metals. A
reasonable correlation was observed between setting time and
the ionic radius of alkali metals, excluding lithium (Li). The
accelerating effect of alkali metals has been attributed to the
removal of the nucleation barrier, through an initially fast
precipitation of alkali metal hydration products (e.g. lithium
hydrometaaluminate) that act as a heterogeneous nucleation
substrate, promoting the nucleation of the CAC hydration pro-
ducts. The lithium has a much greater effect on the setting time
than other alkali cations because of its ability to form tetrahedral
symmetry while other alkali cations form octahedral symmetry.
Along with rapid setting, small quantities of lithium salts cause
Flex
ural
str
engt
h: M
Pa
14
13
12
11
10
9
8
7
6
525 26 27 28 29 30
Atomic number, Z
CAC
Mn2�
Co2�
Cu2�
Zn2�
28 days7 days1 day
Figure 14. Effect of transition metal chlorides (ª ¼ 10 mmol/kg)
on evolution of flexural strength of CAC pastes (horizontal lines:
values for plain CAC).
85
80
75
70
65
60
55
50
45
Com
pres
sive
str
engt
h: M
Pa
Mn2�
Co2�
Cu2�
Zn2�
28 days7 days1 day
25 26 27 28 29 30Atomic number, Z
CAC
Figure 15. Effect of transition metal chlorides (ª ¼ 10 mmol/kg)
on evolution of compressive strength of CAC pastes (horizontal
lines: values for plain CAC)
260
Advances in Cement ResearchVolume 24 Issue 5
Influence of metal chloride salts oncalcium aluminate cement hydrationUkrainczyk, Vrbos and Sipusic
strength development of CAC at early ages, so they can be used
for preparation of rapid hardening cement-based materials – for
example, in rapid repair mortars. Alkali metal salts deteriorate/
reduce compressive and flexural long-term strengths of CAC. The
extent of strength reduction followed the same order as for the
setting time reduction.
Effect of alkaline earth metal salts on setting behaviour depends
on the amount of addition. Low concentrations (0.1%) resulted in
a retardation of the setting behaviour. A further increase of
admixture concentration results in a dramatic acceleration of the
setting time. Calcium addition results in the strongest acceleration
of the initial setting time, while the compressive and flexural
strengths exhibit the lowest reduction from the reference material.
Thus the change in the nucleation rate and the corresponding
deterioration of the strengths can be attributed to the role of the
foreign cations, which can replace the calcium in the hydration
products and/or form novel hydration products.
A simple method of analysing CAC hydration based on the
measurement of temperature rise of cement pastes was described.
Setting times were established from the time–temperature curves
obtained. The results of the setting time obtained from the
measured temperature evolution of cement pastes were compared
to the standard Vicat needle measurements. The difference
between the initial setting time obtained by the two methods is
only 1.2 min � 0.6 min. This validates the further use of the
employed alternative method to obtain accurate values for the
initial setting time parameter. The effects of the sample mass and
heat transfer boundary conditions during setting time measure-
ments were analysed.
With the addition of the transition metal chlorides the main
hydration peak of the temperature evolution curve becomes flatter
and longer, and the setting is retarded in the following order:
Co , Mn , Cu , Zn. The retardation effect is higher for higher
concentrations. A well-defined trend, which was observed when
plotting the results of setting times against added cation’s atomic
number, was related to the change in the acid/base properties of
the precipitate. The retardation effect may be explained by the
precipitation of protective gel (amorphous) coatings on the phase
boundary surface of cement grains, which block hydrolysis and
inhibit growth of the crystalline hydration products. Magnitude of
strength reduction for all tested ages follows the order observed
for the setting time retardation. The strength reduction may be
primarily attributed to the lower hydration degree achieved due to
the retardation of the hydration reaction rate.
The influence of alkaline earth and transition metal chlorides
additions on setting behaviour of iron-rich CAC and refractory
Secar 71 CAC is shown to be opposite in nature. This suggests
that the hydration mechanisms of these two classes of CAC are
significantly different.
It is recommended that for further investigations of the admix-
tures a constant molar concentration (represented in added mols
per mass of cement) should be used. This will enable a better
comparison of the results among researchers, even when mixtures
are prepared with different water-to-cement ratios.
AcknowledgementThe authors acknowledge support from the Croatian Ministry of
Science, Education and Sports under projects no. 125-1252970-
2983 ‘Development of hydration process model’.
REFERENCES
Banfill PFG (1995) Superplasticizers for Ciment Fondu. Part 2:
Effect of temperature on the hydration reaction. Advances in
Cement Research 7(28): 151–157.
Barret P, Menetrier D and Bertrandie D (1974) Contribution to
the study of the kinetic mechanisms of aluminous cement
setting. Cement and Concrete Research 4(4): 545–556.
Bensted J (2002) Calcium Aluminate Cements, in Structure and
Performance of Cements, 2nd edn (Bensted J and Barnes P
(eds)). Spon Press, London, UK, pp. 114–138.
Bushnell-Watson SM and Sharp JH (1986) The effect of
temperature on the setting behaviour of refractory calcium
aluminate cement. Cement and Concrete Research 16(6):
875–884.
Cong X and Kirkpatrick RJ (1993) Hydration of calcium
aluminate cements: a solid-state 27Al NMR study. Journal of
the American Ceramic Society 76(2): 409–416.
Currell BR, Grzeskowlak R, Midgley HG and Parsonage JR (1987)
The acceleration and retardation of set high alumina cement
by additives. Cement and Concrete Research 17(3): 420–432.
Edmonds RN and Majumdar AJ (1989) The hydration of Secar 71
aluminous cement at different temperatures. Cement Concrete
Research 19(2): 289–294.
Fryda H, Scrivener KL and Chanvillard G (2001) Relevance of
laboratory tests to field applications of calcium aluminate
cement concretes. In Calcium Aluminate Cements 2001
(Mangabhai RJ and Glasser FP (eds)). IOM Communications,
London, UK, pp. 227–246.
Fujii K, Kondo W and Ueno H (1986) Kinetics of hydration of
monocalcium aluminate. Journal of the American Ceramic
Society 69(4): 361–364.
Galtier P and Guilhot B (1984) Conductimetrie, hydration, and
reactivite de l’aluminate monoalcique. Cement and Concrete
Research 14(5): 679–685.
George CM (1983) Industrial aluminous cements. In Structure
and Performance of Cements (Barnes P (ed.)). Applied
Science, London, UK, pp. 415–470.
Griffiths DL, Al-Qaser ANF and Mangabhai RJ (1991)
Calorimetric studies on high alumina cement in the presence
of chloride, sulphate and seawater solutions. In Proceedings
of International Conference on CAC, Calcium Aluminate
Cements (Mangabhai RJ (ed.)). Chapman and Hall, London,
UK, pp. 167–178.
Jantzen C, Johnson A, Read D and Stegemann JA (2010)
261
Advances in Cement ResearchVolume 24 Issue 5
Influence of metal chloride salts oncalcium aluminate cement hydrationUkrainczyk, Vrbos and Sipusic
Cements in waste management. Advances in Cement
Research 22(4): 225–231.
Justnes H (2008) Rapid repair of airfield in cold weather using
CAC mortar. In Proceedings of the Centenary Conference,
Calcium Aluminate Cements, Avignon, France (Fentiman CH,
Scrivener KL and Mangabhai RJ (eds.)). IHS BRE Press,
Avignon, France, pp. 365–373.
Murat M and Sadok EH (1991) Role of foreign cations in
solution in the hydration kinetics of high alumina cement. In
Proceedings of International Conference on CAC, Calcium
Aluminate Cements (Mangabhai RJ (ed.)). Chapman and Hall,
London, UK, pp. 155–166.
Matusinovic T, Vrbos N and Curlin D (1994) Lithium salts in
rapid setting high-alumina cement materials. Industrial and
Engineering Chemistry Research 33(11): 2795–2800.
Mangabhai RJ (ed.) (1990) Proceedings of International
Conference on CAC, Calcium Aluminate Cements. Chapman
and Hall, London, UK.
Mangabhai RJ and Glasser FP (eds) (2001) Proceedings of
International Conference on CAC, Calcium Aluminate
Cements. IOM Communications, London, UK.
Mohmel S and Gessner W (1997) The influence of alumina
reactivity on the hydration behaviour of mono calcium
aluminate. Solid State Ionics 101(103): 937–943.
Nikushchenko VM, Khotimchenko VS, Rumyantsen PF and
Kalinin AI (1973) Determination of the standard free energies
of formation of calcium hydroaluminates. Cement and
Concrete Research 3(5): 625–632.
Nilforoushan MR and Sharp JH (1995) The effect of additions of
alkaline-earth metal chlorides on the setting behavior of a
refractory calcium aluminate cement. Cement Concrete
Research, 25(7): 1523–1534.
Nordstrom DK and May HM (1989) Aqueous equilibrium data for
mononuclear aluminum species. In The Environmental
Chemistry of Aluminum (Sposito G (ed.)). CRC Press, Boca
Raton, Florida, USA, pp. 29–53.
Rashid S and Turrillas X (1997) Hydration kinetics of CaA12O4
using synchrotron energy-dispersive diffraction.
Thermochimica Acta 302(1–2): 25–34
Rodger SA and Double DD (1984) The chemistry of high alumina
cement in the presence of accelerating and retarding
admixtures. Cement and Concrete Research 14(1): 73–82.
Scrivener KL, Cabiron JL and Letourneux R (1999) High-
performance concretes from calcium aluminate cements.
Cement and Concrete Research 29(8): 1215–1223.
Shannon RD (1976) Revised effective ionic radii and systematic
studies of interatomic distances in halides and chalcogenides.
Acta Crystallographica A 32(5): 751–767.
Shkol’nikov EV (2005) Thermodynamic characteristics of the
amphoterism of M(OH)2 hydroxides in aqueous media.
Russian Journal of Applied Chemistry 78(11): 1786–1790.
Sorrentino D, Sorrentino F and George M (1995) Mechanisms
of hydration of calcium aluminate cements. In Materials
Science of Concrete IV (Skalny J and Mindess S (eds)).
The American Ceramic Society, Westerville, Ohio, USA,
pp. 41–90.
Stegemann JA and Buenfeld NR (2001) Neural network
prediction of setting of calcium aluminate cements containing
additions. In Proceedings of International Conference on
CAC, Edinburgh, UK (Mangabhai RJ and Glasser FP (eds)).
IOM Communications, London, UK, pp. 267–279.
Taylor HFW (1998) Cement Chemistry, 2nd edn. Thomas Telford,
London, UK, Ch. 10.
Ukrainczyk N (2010) Kinetic modeling of calcium aluminate
cement hydration. Chemical Engineering Science 65(20):
5605–5614.
Ukrainczyk N and Matusinovic T (2010) Thermal properties of
hydrating calcium aluminate cement pastes. Cement and
Concrete Research. 40(1): 128–136.
Ukrainczyk N, Matusinovic T, Kurajica S, Zimmermann B and
Sipusic J (2007) Dehydration of a layered double hydroxide –
C2AH8: Thermochimica Acta 464(1–2): 7–15.
Wells AF (1962) Structural Inorganic Chemistry. Oxford Press,
London, UK.
WHAT DO YOU THINK?
To discuss this paper, please submit up to 500 words to
the editor at www.editorialmanager.com/acr by 1 Febru-
ary 2013. Your contribution will be forwarded to the
author(s) for a reply and, if considered appropriate by
the editorial panel, will be published as a discussion in a
future issue of the journal.
262
Advances in Cement ResearchVolume 24 Issue 5
Influence of metal chloride salts oncalcium aluminate cement hydrationUkrainczyk, Vrbos and Sipusic