Cement & Concrete Composites 46 (2014) 99–108

Contents lists available at ScienceDirect

Cement & Concrete Composites

journal homepage: www.elsevier.com/locate/cemconcomp

Influence of limestone and anhydrite on the hydration of Portland

cements
Maciej Zajac a,⇑, Anne Rossberg b, Gwenn Le Saout b, Barbara Lothenbach b
a
HeidelbergCement Technology Center GmbH, Rohrbacher Str. 95, 69181 Leimen, Germany
b
Empa, Laboratory for Concrete & Construction Chemistry, Überlandstrasse 129, CH-8600 Dübendorf, Switzerland

a r t i c l e i n f o a b s t r a c t

Article history: The addition of CaCO3 and CaSO4 to Portland cement clinker influences the hydration and the strength
Received 20 February 2013 development. An increase of the CaSO4 content accelerates alite reaction during the first days and results
Received in revised form 26 July 2013 in the formation of more ettringite, thus in a higher early compressive strength. The late compressive
Accepted 25 November 2013
strength is decreased in Portland cements containing higher quantities of CaSO4. The reduced late com-
Available online 2 December 2013
pressive strength seems to be related to an increase of the S/Si and Ca/Si content in the C–S–H.
The presence of calcite leads to the formation of hemicarbonate and monocarbonate thus indirectly to
Keywords:
more ettringite. Only a relatively small quantity of calcite reacts to form monocarbonate or hemicarbon-
Hydration
Limestone
ate in Portland cement. Although hemicarbonate is thermodynamically less stable than monocarbonate,
Sulfate hemicarbonate formation is kinetically favored. Monocarbonate is present only after 1 week and longer
Modelling independent of the quantity of calcite available and the content of sulphate in the cement.
X-ray diffraction Ó 2013 Elsevier Ltd. All rights reserved.
Compressive strength

1. Introduction
The addition of CaCO3 and CaSO4 to Portland cement clinker
influences the setting behaviour, the progress of hydration and
the strength development during hydration. If limestone is added
in the small quantities, limestone has a positive impact on the
engineering properties of the cement and concretes [1]. The posi-
tive impact of limestone on the cement properties was explained
as a filler effect [2]. More recent studies showed that calcite is also
a reactive cement component [3]. In the presence of limestone,
the formation of hemicarbonate (C3A
0.5CaCO3
12H2O) and
monocarbonate (C3A
CaCO3
11H2O) is observed instead of mono-
sulphate [4], which prevents the decomposition of ettringite. In
contrast, in limestone free cements the further reaction of alumi-
nate phase leads to the destabilization of ettringite and formation
of monosulphate, after all gypsum is consumed [5,6]. The stabilisa-
tion of the voluminous ettringite instead of monosulphate gives
rise to an increase in the total volume of hydration products
[5–7] and can thus result in an increase of the compressive
strength of mortars [8–11].
Thermodynamic calculations indicate that the main stable
phases in hydrated calcite-containing cement are C–S–H, CH,
ettringite, monocarboaluminate, and calcite [5,7], while hemicar-
bonate is calculated to be stable only if no excess calcite is present,
⇑ Corresponding author.
E-mail address: maciej.zajac@htc-gmbh.com (M. Zajac).
i.e. in hydrated Portland cements which contain less than 2–3 wt%
of calcite. In experimental studies generally, the formation of
hemicarbonate is observed initially [5,6,12,13] in contrast to the
thermodynamic calculations. The content of hemicarbonate
decreases over the hydration time as monocarbonate is formed
instead. The reason for the initial formation of hemicarbonate in-
stead of monocarbonate is unclear. It could be related either to
the slow reaction kinetics of limestone at high pH values or to a
faster formation kinetic of hemicarbonate when compared to the
kinetics of monocarbonate formation.
The amount of calcium sulphate originally added to a Portland
cement influences the cement matrix properties. The addition of
sulphate regulates not only the setting but also the composition
and volume of the hydrate assemblage, the remaining porosity
and thus indirectly also compressive strength. In general, more
ettringite and an increase of compressive strength are observed
in samples containing more SO3. However, above 3–4 wt% SO3 con-
tent the compressive strength generally decreases [14–18],
although the volume of the solid hydrates is expected to increase
if more ettringite forms.
In this paper, the hydration of three laboratory limestone
cements containing different amounts of CaSO4 as well as industri-
ally produced cements with different quantities of limestone were
investigated in order to asses the influence of calcium sulphate and
calcium carbonate on hydration and strength development. Ther-
modynamic modelling has been used to predict the composition

0958-9465/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.cemconcomp.2013.11.007
100 M. Zajac et al. / Cement & Concrete Composites 46 (2014) 99–108

of the solid phases. Additionally the dissolution of limestone was Table 2

studied at high pH values. Mineral composition of the cements determined by XRD-Rietveld analysis.

Sample Phase contents (wt%)

2. Materials and methods C3S C2S C3A C4AF Anhydrite Bassenite Calcite Others
L2-15 51.7 17.3 9.8 1.7 2.8 0 15 1.7
2.1. Materials L3-15 51.1 17.1 9.7 1.6 4.3 0 15 1.2
L4-15 50.5 16.9 9.6 1.6 5.6 0 15 0.8
C3.5-1 65.5 9.0 8.2 9.3 0.5 3.4 1.1 3.0
In order to prepare the laboratory cements, Portland cement C3.5-3 57.6 14.4 8.7 8.8 0.6 3.0 2.8 4.1
clinker, natural anhydrite and natural limestone were used. The ce- C3.5-9 54.2 14.0 8.4 7.8 2.1 1.5 8.7 3.3
ment clinker was ground together with the anhydrite in the labo-
ratory ball mill. The cement clinker was homogenized with 15 wt%
of ground limestone and different quantities of natural anhydrite
(2.5, 3.5 and 4.5 wt%) to give Portland limestone cement. The final
sulphate contents of the cements were: 2.1 wt%, 3.0 wt% and
3.8 wt%. The corresponding cements are labelled: L2-15, L3-15
and L4-15, where ‘‘L’’ indicates laboratory cement; the first number
corresponds to the approximate content of the sulphate and the
second to the content of limestone.
Three commercial cements: CEM I 52.5 R, CEM I 42.5 R and CEM
II/A-LL 42.5 R were investigated as well. The cements contain about
3.5 wt% SO3 and 1, 3 and 9 wt% of limestone, respectively. These
cements are labelled: C3.5-1, C3.5-3 and C3.5-9, where again the
first number corresponds to the content of the sulphate and the
second to the content of the limestone.
The chemical composition determined by XRF and the physical
properties of the materials are given in Table 1. The mineralogical
composition of cement clinker and commercial cement was ob-
tained from XRD – Rietveld analysis (Table 2). The particle size dis-
tributions of laboratory and commercial cements determined by
laser granulometry using a Malvern Mastersizer are given in
Fig. 1. The CaCO3 content of the natural limestone, determined
by XRD – Rietveld analysis, is about 97 wt%. The limestone had a
Blaine fineness of 7000 cm2/g and a d50 of 8 lm.
The fineness of the limestone fraction of the commercial ce-
ments is not known. As the limestone, sulphate and clinker were
co-ground during the production of the cements it is expected that
the limestone is very fine as it is easier to grind than the cement
clinker phases [19,20].
The distribution of the alkalis between sulphates and oxides in
the unhydrated cement was determined based on the measured
concentration of the readily soluble alkalis in double distilled Fig. 1. The particle size distribution of investigated (a) laboratory cements, and (b)
water at w/c (water to cement ratio) of 10 after an equilibration commercial cements determined by laser diffractometry.
time of 5 min.

Table 1
Chemical composition of the used materials.

Parameter Unit L2-15 L3-15 L4-15 C3.5-1 C3.5-3 C3.5-9

SiO2 wt% 18.9 18.5 18.1 19.32 19.35 18.49
Al2O3 wt% 4.7 4.5 4.5 5.32 5.47 5.27
TiO2 wt% 0.2 0.2 0.2 0.26 0.27 0.26
MnO wt% 0.0 0.0 0.0 0.04 0.04 0.03
Fe2O3 wt% 1.6 1.5 1.4 2.88 2.83 2.71
CaO wt% 63.7 63.2 62.7 63.2 62.0 61.2
MgO wt% 1.0 0.9 1.0 2.14 2.3 2.12
K2O wt% 0.7 0.6 0.6 0.78 0.99 0.99
Na2O wt% 0.2 0.2 0.2 0.06 0.06 0.05
SO3 wt% 2.1 3.0 3.9 3.58 3.5 3.45
P2O5 wt% 0.3 0.3 0.3 0.11 0.2 0.22
LOI 950 °C wt% 6.6 6.7 6.7 2.07 2.59 4.55
Total wt% 100.0 99.8 99.7 99. 8 99.6 99.4
CO2(TG) wt% 6.2 6.3 6.2 0.7 1.8 4.3
Readily soluble alkalis
Na2O wt% 0.07 0.07 0.07 0.05 0.05 0.05
K2O wt% 0.5 0.5 0.5 0.5 0.6 0.62
Physical parameters
Blaine surface area cm2/g 4873 4550 4669 6450 4990 5080
Density g/cm3 3.00 3.00 3.00 3.12 3.11 3.07
d50 lm 10.1 11.7 12.0 7.0 9.9 9.5
 M. Zajac et al. / Cement & Concrete Composites 46 (2014) 99–108 101

2.2. Experimental methods decoupling and a 0.2 s relaxation delay was employed. The 27Al
chemical shifts were referenced relative to a 1.0 M AlCl3–6H2O
2.2.1. Mortar samples solution at 0 ppm.
Mortar prisms (40  40  160 mm) with cement–sand–water
proportions of (1/3/0.5) were prepared. The samples were cured 2.2.3. Dissolution tests
in Ca(OH)2 saturated solution at 20 °C. The compressive and flex- The dissolution of the limestone was followed by measuring the
ural strengths were determined on three mortar prisms for each dissolved calcium concentrations after different reaction times in
testing age according to EN 196-1. 0.01, 0.1 and 0.5 M KOH, respectively. The dissolved concentrations
of calcium were determined with a Dionex DP ICS-3000 ion chro-
2.2.2. Cement pastes matograph or by a calcium ion selective electrode (ISE). The error
Calorimetric measurements (Thermometric TAM Air) were car- of the measurements was 10%.
ried out with 6 g of fresh paste.
To study the hydration process thermogravimetric analysis
(TGA), X-ray diffraction (XRD) and scanning electron microscopy 2.3. Thermodynamic modelling
(SEM) were used. Cement paste samples were prepared with water
to binder ratio of 0.45 and stored at 20 °C in 20 ml sealed plastic The composition of the solid phase as a function of added sul-
vials. phate and calcite was calculated based on the geochemical model-
For thermogravimetric analysis, the samples were crushed. The ling program, GEMS [23]. The thermodynamic data from the PSI-
hydration of the laboratory samples was stopped by solvent ex- GEMS database [24,25] was supplemented with cement specific
change using isopropanol during 15 min and flushing with diethyl data [26,27]. For the calculations a reaction degree of the clinkers
ether. The hydration of the industrial cement samples was stopped of 90 wt% was assumed. Sulphate and limestone was allowed to re-
by drying at 40 °C. Thermogravimetric analysis was performed on act freely.
about 50 mg of the resulting powder by monitoring the weight
while heating up from 30 to 1000 °C at 20 °C/min in a Mettler To- 3. Results and discussion
ledo TGA/SDTA851 (laboratory cements) or TGA/DTG NETZSCH
STA 409C/CD (industrial cements). The amounts of bound water 3.1. Influence of calcium sulphate
and portlandite were deduced from the weight losses between
40 and 550 °C and 450–550 °C respectively. The thermodynamic calculations presented in Fig. 2 show that
For the XRD experiments, slices (diameter 3 cm) were cut from the addition of anhydrite is expected to lead to the formation of
the cylinder and placed in the diffractometer for XRD pattern more ettringite and less AFm phases and thus to a greater volume
acquisition. The laboratory cements were measured using a PANa- of hydrated solids, less porosity and potentially higher compres-
lytical X’Pert Pro MPD diffractometer in a h–2h configuration with sive strength. The calculations indicate an increase in the solid vol-
an incident beam monochromator and Cu Ka radiation (k = 1.54 Å) ume of the hydrates of 2.1 cm3/100 g from 56.8 to 58.9 cm3/100 g
and the Rietveld analysis was performed using an external CaF2 of unreacted cement if the CaSO4 content increases from 2.1% to
standard [21]. The industrial cements were measured on a Bruker 3.8% (Fig. 2). These theoretical observations agree with the exper-
D8 Advance and the Rietveld refinements were performed using imental observations, where in general the formation of more
Topas 4 software from Bruker AXS using a corundum standard. ettringite and an increase of compressive strength are observed
AFm phases have generally low crystallinity and variations in com- [14–18]. However, above 3–4 wt% SO3 content a decrease of com-
position that lead to changes in position and intensity reflections in pressive strength is generally observed, eventhough the volume of
the XRD patterns [22]. Their quantification by Rietveld analysis the solid hydrates is expected to increase if more ettringite forms.
might be therefore not reliable. In this paper the peak intensities This indicates that other effects are important for compressive
of the strongest peaks were used as a measure of the evolution strength development besides the degree of space filling.
quantities of hemicarbonate. The results of the Rietveld analysis
gave the sum of phases normalized to 100 wt%. Due to the hydra-
tion reactions, water is bound in the hydrates and the total amount
of solids increases. In order to correct for this dilution effect, the
amount of bound water by TGA was deduced, so that the results al-
ways refer to the mass of anhydrous materials.
Laboratory made cements were additionally investigated by
SEM. Slices of hydrated cement paste samples were cut and imme-
diately immersed in isopropanol for 30 min and subsequently
dried at 40 °C for 24 h. A piece of the slices of the hydrated paste
was impregnated using low viscosity epoxy resin, polished down
to 0.25 lm, coated with carbon and examined using a Philips ESEM
FEG XL 30 scanning electron microscopy (SEM). Energy dispersive
X-ray spectroscopy (EDX) was applied to determine the elemental
compositions of the matrix.
The porosity of the hydrated pastes was determined by mercury
intrusion porosimetry (MIP) and by water intrusion, where the
capillary porosity is calculated from the difference in water content
between the saturated sample and the sample dried at 50 °C. The
total porosity refers to drying at 110 °C.
Bruker ASX 400 spectrometer (9.4 T) at 104.2 MHz was used to
Fig. 2. Effect of addition of anhydrite on the hydrate assemblage of a hydrated
measure 27Al NMR of samples after 1 and 91 days. The spectra Portland cement containing 15 wt% of limestone. The composition of clinker
were recorded at a spinning rate of 20 kHz in 2.5 mm ZrO2 rotors. corresponds to the clinker used for the preparation of the laboratory made samples
Single pulse (p/12) excitation width pulse of 1 ls without 1H as given in Table 2.
102 M. Zajac et al. / Cement & Concrete Composites 46 (2014) 99–108

3.1.1. Influence of anhydrite on the reaction kinetics

The early hydration of cement pastes was followed by calorim-
etry as shown in Fig. 3. The initial evolution of the heat of hydra-
tion is similar for all the investigated laboratory made samples.
The different sulphate levels have no impact on the onset of the
main hydration peak, which is generally associated with the hydra-
tion of alite [28]. The shoulder after the main silicate peak in Fig. 3
between 0.5 and 1 day is associated with the depletion of gypsum
and the renewed reaction of the aluminate phases during which
ettringite and AFm phases are formed [28,29]. The aluminate reac-
tion is delayed when more sulphate is present and is in agreement
with other findings reported in the literature [30–32].
The reaction of the anhydrous phases was monitored by XRD/
Rietveld analysis as shown in Fig. 4. The calorimetric measure-
ments had indicated that more CaSO4 retarded the aluminate reac-
tion during the first 20 h. The XRD measurements indicate that
after 1 day most of the aluminate has reacted and no significant
differences exist in the amount of unreacted aluminate after 1 day.
The presence of more calcium sulphate does not accelerate the
onset of hydration peak (Fig. 3) but it clearly accelerates the alite
reaction at 1 and 2 days, whereas the difference after 7 days and
longer is small (Fig. 4). This increased alite reaction during the first
days in the presence of more calcium sulphate is consistent with
observations on the accelerating effects of sulphate on the reaction
of C3S, alite and belite reported in the literature [18,33–38]. Gunay
et al. [37] suggested that the adsorption of calcium sulphate on
C–S–H surface modifies the nucleation-growth process and in turn
accelerates the alite hydration.

3.1.2. Hydrates formed

More anhydrite in the cement increases the amount of bound
water and portlandite at early ages (Fig. 4). After 7 days and longer
no significant difference in the amount of bound water is observed,
whereas the samples with the more anhydrite form less
portlandite. This decrease of portlandite is observed for the sam-
ples L3-15 and L4-15, both by TGA and XRD techniques, and has
also been reported for cements and for C2S containing more cal-
cium sulphate [39,40].
The addition of more anhydrite leads to a higher S/Si and Ca/Si
atomic ratios in the C–S–H as determined by EDX (Table 3). Thus
the decrease in portlandite in the presence of more anhydrite is re-
lated to higher Ca/Si ratio in the C–S–H. The Al/Si ratio in the C–S–
H equals to 0.06 and is not significantly affected by the amount of
Fig. 4. Influence of anhydrite on (a) the reaction of clinker phases determined by
anhydrite. A similar increase in both the S/Si and Ca/Si ratios has
Rietveld/XRD; C3S – solid line, C2S – dash line, C3A – dash-dot line, error is ±2%. (b)
been observed in Portland cements [35], in C3S [38] and in C–S– The amount of bound water and (c) Portlandite as determined by TGA in wt% per
H–ettringite mixtures [41]. While the increase of SO3 in the mass of cement.
C–S–H is expected, the simultaneous increase of Ca in the C–S–H
indicates a coupled uptake of Ca and SO3 within the C–S–H as
Table 3
Al/Si, Ca/Si and S/Si atomic ratios in the C–S–H after 91 days of reaction.

Sample Al/Si Ca/Si S/Si

L2-15 0.05 ± 0.02 1.92 ± 0.1 0.050 ± 0.01
L3-15 0.06 ± 0.02 1.96 ± 0.1 0.057 ± 0.01
L4-15 0.05 ± 0.02 2.04 ± 0.1 0.062 ± 0.01

suggested by Labbez et al. [42]. Copeland, Kantro and co-workers

Fig. 3. Isothermal calorimetry results of investigated cements expressed per mass
of cement.
observed that an increase in the S/Si ratio in the C–S–H resulted
in a decrease of the water content in the C–S–H and a clear de-
crease of the compressive strength of C3S and C2S samples
[35,40,43]. Gunay et al. [37] indicated that the absorption of cal-
cium sulphate on C–S–H results in a decrease of the forces between
C–S–H particles which could be related to the decrease of the com-
pressive strength of hydrated alite pastes.
To investigate the evolution of the hydrates in the investigated
samples XRD, TGA and NMR are used. The main hydration phases
 M. Zajac et al. / Cement & Concrete Composites 46 (2014) 99–108 103

found in all the samples are portlandite, ettringite and poorly-crys-

talline C–S–H, characterized in the XRD spectra by a hump at 28–
33° 2h and in TGA by a broad peak of water loss between 100 and
200 °C.
The main difference in the XRD-patterns can be seen at low an-
gles, where the main peaks of the AFm and AFt phases are found
(Figs. 5 and 6). After 1 day of hydration the peak of ettringite
(2h = 9.1°) is visible for all samples. The remaining aluminates react
with calcium carbonate to form mono- and hemicarbonates (Mc at
11.7° 2h and Hc at 10.8° 2h). In the samples with lower sulphate
content, i.e. L2-15 and L3-15 the hemicarbonate peak is visible
after one day of hydration. In L4-15 Hc appears only after 2 days
of hydration as the presence of more calcium sulphate leads to in-
creased ettringite formation and thus to a retardation of the Hc for-
mation. At 7 days and longer the Hc peak decreases in all Fig. 6. Amount of ettringite (solid lines) and monocarbonate (dash lines) phases
investigated samples, whereas the peak of monocarbonate in- determined by XRD/Rietveld analysis as a function of time in wt% per mass of
cement, error is ±2%.
creases with time. The intensities of the monocarbonate peak, after
90 days of hydration, is correlated with the SO3/Al2O3 weight frac-
tion of the anhydrous cements. The lower the SO3/Al2O3 ratio, the
27
higher the intensities of the monocarbonate peak. Al NMR spectra agree very well with the XRD and TGA
The TGA measurements shown in Fig. 7 confirm the significant observations.
precipitation of the AFt phases and the formation of some AFm
after the first day of hydration in L2-15 and L3-15, whereas in
the L4-15 sample, which contains more calcium sulphate, the for- 3.1.3. Porosity
mation of AFm is observed only after 2 days. After 91 days, more It is expected that a higher calcium sulphate content would in-
ettringite and less AFm phases are present as the calcium sulphate crease the volume of the solid phases in the samples. More sul-
content increases, in agreement with the calculations shown in phate in the mix stabilizes more ettringite and reduces the AFm
Fig. 2. The presence of ettringite and hemicarbonate or monocar- content (see Fig. 2). Ettringite (1.8 g/cm3) has a lower density than
bonate in the limestone containing samples is also confirmed by monocarbonate (2.2 g/cm3) [27] thus occupies more volume. Con-
the EDX analysis. sequently, comparing the sample L4-15 (3.9 wt% SO3) to L2-15
Both ettringite and AFm contain exclusively octahedrally coor- (2.1 wt% SO3) one would expect a volume increase in the hydration
dinated Al and lead to peaks in the 27Al NMR spectra near +13 products by 2.1 cm3/100 g cement as predicted by the thermody-
and near +10 ppm [44,45]. Different AFm phases have almost equal namic modelling (Fig. 2). This should be mirrored by a significant
chemical shift and relative strong quadrupolar interactions in com- decrease in the porosity and in an increase of the compressive
parison to ettringite, thus the distinction between monosulphate, strength. The porosity measurements given in Table 4 show that
hemicarbonate and monocarbonate is not possible. After one day, all investigated samples are characterized by a similar porosity at
the presence of ettringite is observed in all samples (Fig. 8). A longer hydration times although clearly more ettringite is present
shoulder can be also observed that reveals the presence of AFm as shown above in Figs. 6–8. This similar porosity agrees with the
phase in the case of the samples L3-15 and L2-15, whereas in L4- similar amount of bound water at longer reaction times (Fig. 4).
15 which contains more anhydrite, AFm phases are not yet present. This suggests that the additional volume of ettringite in the pres-
After 91 days, the most ettringite and the least AFm phase has ence of more CaSO4 is compensated, possibly by a decrease in
formed in the sample L4-15. The relative intensity of the AFm peak the water content and volume of C–S–H, by the higher S/Si and
increases as the amount of calcium sulphate decreases and the Ca/Si ratios.

Fig. 5. Influence of different quantities of anhydrite on the hydration of laboratory limestone containing Portland cements: (a) L2-15, (b) L3-15, and (c) L4-15. The main
reflexes of ettringite (Et), hemicarbonate (Hc), monocarbonate (Mc) are indicated.
104 M. Zajac et al. / Cement & Concrete Composites 46 (2014) 99–108

Fig. 7. TGA analysis of paste samples hydrated for 1 (a), 2 (b) and 180 (c) days. Et – etrringite, Hc/Mc hemi and monocarbonate.

Fig. 9. Influence of anhydrite on the compressive strength of mortar samples.

strength, which agrees with the similar porosity observed for all
samples (Table 4). The slight decrease of strength, instead of ex-
pected increase, may be related to the increased uptake of S/Si
27
Fig. 8. Al MAS NMR spectra of hydrated laboratory made cements. and Ca/Si in the C–S–H [35,37,40], and a corresponding decrease
of water content [40,43]. The negative influence of calcium sul-
phates on the properties of C–S–H seems to be compensated at la-
Table 4
Capillary and total porosity of the paste samples determined by means of water
ter ages by the positive impact of the presence of additional
intrusion (w) and mercury intrusion porosimetry (MIP). ettringite.
Sample 28 Days 91 Days
Capillary Total (MIP) Capillary Total (MIP)
3.2. Influence of limestone on the hydration
(w) (w) (w) (w)
L2-15 13.2 21.5 19.6 9.7 21.4 20.3
The addition of limestone to Portland cements generally accel-
L3-15 12.7 21.6 20.2 9.3 21.1 20.2 erates the early hydration and increases the strength at early ages
L4-15 13.0 21.8 22.5 9.6 21.4 20.0 of the cement as the limestone provides additional nucleation
sites, which accelerate the early reaction [5,6,46,47]. In addition,
monocarbonate and hemicarbonate are formed in the presence of
3.1.4. Effects on compressive strength calcite, which indirectly stabilize ettringite leading to an increased
The results of the compressive strength measurements are sum- volume as indicated in Fig. 10. The higher volume of hydrates is re-
marized in Fig. 9. For all investigated samples the compressive lated to higher strength at low additions of calcite. For higher lime-
strength increases with the hydration time. In the case of the lab- stone additions, however, the dilution of the clinker phases by the
oratory made cements the maximal compressive strength at 1 day limestone leads to a decrease volume and of compressive strength
is measured for samples L3-15 and L4-15, whereas the early [8–11].
strength of L2-15 is lower.
The slightly higher compressive strength in the presence of 3.2.1. Influence of limestone on the reaction kinetics
more anhydrite during the first two days is consistent with the in- The kinetic of reaction of the samples is dominated by their
creased solid volume due to the formation of more ettringite fineness; the finer sample C3.5-1 reacts faster than the two coarser
(Figs. 6–8) and due to the increased reactivity of silicates during cements (Fig. 11). Due to the difference in cement fineness, the
the first week (Fig. 4). At later ages, however, the increased calcium accelerating effect of the limestone on the reaction of the silicates
sulphate content has no positive effect on the compressive was not clearly visible, in contrast to other studies [5,6], where the
 M. Zajac et al. / Cement & Concrete Composites 46 (2014) 99–108
Fig. 10. Effect of calcite (CaCO3) on the hydrate assemblage of a hydrated Portland
cement. The composition of clinker corresponds to the clinker used for the
preparation of the sample C3.5-3, with a SO3 content of 3.5%.
Fig. 11. Isothermal calorimetry results of investigated cements expressed per mass
of cement.
limestone had been blended to the cement instead of interground.
The same dominance of the effect of the fineness is also visible in
the reaction of the clinker phases as shown in Fig. 12.
Fig. 12. Influence of cement fineness and limestone content on the reaction of
clinker phases determined by Rietveld/XRD in wt% per mass of cement error is ±2%,
C3S – solid line, C2S – dash line, C3A – dash-dot line.
105
3.2.2. Hydrates formed
XRD characterization shows that independent of the amount of
limestone present and cement finesses ettringite is observed in
comparable quantities after 1 day of hydration (Figs. 13 and 14).
After 28 days of hydration, the ettringite content starts to decrease
for the low limestone sample C3.5-1.
Hemicarbonate is clearly identified in all samples after 2 days of
hydration. The intensity of the hemicarbonate peak increases up to
7 days, when monocarbonate starts to form. The hemicarbonate
peak intensities are significantly higher for both C3.5-3 and C3.5-
9 than for C3.5-1, which is related to the lower calcite content in
the C3.5-3 sample. In the presence of only 1% limestone (C3.5-1),
the formation of monocarbonate is retarded while a broad peak
at 2h 11.3° is visible between the hemicarbonate and monocar-
bonate peak. This peak probably corresponds to hemicarbonate
containing additional carbonate [48], i.e. with a composition be-
tween hemicarbonate and monocarbonate. The monocarbonate
content increases in all investigated samples over the time, the
slowest rate is found for the C3.5-1 with the lowest content of
limestone.
3.3. Reaction of limestone
Thermodynamic calculations indicate that the monocarbonate
is the thermodynamically stable phase in the limestone – OPC
blends that contain more than 2–3 wt% of limestone [5,7]
(Fig. 10). In contrast, the XRD measurements show that initially
hemicarbonate has formed also in the cements containing 3, 9 or
15 wt% of limestone (Figs. 5 and 13). The hemicarbonate trans-
formed slowly within a few weeks to monocarbonate. Such an ini-
tial formation of hemicarbonate has been reported by several
authors in Portland cements [5,6,12,13] and in C3A, lime, gypsum
and calcite systems [49], while monocarbonate precipitated later.
The observed initial formation of the hemicarbonate instead of
monocarbonate indicates a kinetic hindrance of this reaction. The
cause of the initial formation of hemicarbonate instead of
monocarbonate is unclear and could be related either to a slow
dissolution kinetics of limestone at high pH values or due to a
difference in the formation kinetics of hemi- and monocarbonate.
3.3.1. Reactivity of limestone at high pH values
The dissolution of calcite is relatively fast under neutral to mod-
erately alkaline conditions and slows down significantly above pH
values of 6 [50–52]. Whether the rate increases again above a pH of
12 is unclear. The investigation of the changes in calcium concen-
trations during the dissolution of calcite in 0.1, 0.2 and 0.5 M KOH
solutions (Fig. 15) showed that calcite equilibrated within a few
hours in KOH solutions. Thus, the reactivity of limestone seems
to be also at high pH values sufficiently high to supply sufficient
carbonate into the solution.
Other ions present in the pore solution could influence the cal-
cite dissolution. The presence of silicate has been reported to
slightly decrease the dissolution rate of calcite, while sulphate
slightly increases reaction kinetics [53]. The influence of both ions,
however, is quite small, which makes a significant retardation of
calcite dissolution not probable as also indicated by the investiga-
tions of calcite in the Portland cements detailed below.
3.3.2. Reaction of limestone in Portland cements
The quantification of the calcite reaction in the cement matrix is
experimentally difficult. The experimental methods like XRD and
TGA are not sufficiently sensitive to determine the small changes
of the calcite content in the samples as the error of measurements
in both methods is around 1–2 wt%. A general decrease in the cal-
cite content of 1–3% can be observed during the hydration (Fig. 16),
with exception of the C3.5-1, which contained only 1% of
106 M. Zajac et al. / Cement & Concrete Composites 46 (2014) 99–108
Fig. 13. Influence of different quantities of limestone on the hydration of cements: (a) C3.5-1, (b) C3.5-3, and (c) C3.5-9. The main peaks of ettringite (Et), hemicarbonate (Hc),
monocarbonate (Mc) and ferrite phase (C4AF) are indicated.
Fig. 14. Evolution of the ettringite (solid lines), and monocarbonate (dash lines)
content in the investigated samples in wt% per mass of cement, error is ±2%.
Fig. 15. Reaction of limestone in 0.01, 0.1 and 0.5 M KOH solution. Ca concentration
determined by ion chromatography and by ion selective electrode (ISE), error
is ±10%.
limestone. Although slightly less limestone seems to have reacted
in the anhydrite rich L4-15 cement than in the L2-15, the differ-
ences are within the measurement error as only a small quantity
of calcite can react in the ordinary Portland cement matrix [5,7].
Therefore, the reaction of calcite in cement matrix was also inves-
tigated indirectly by following the changes in hemicarbonate and
monocarbonate.
Fig. 16. Decrease of the amount of calcite during the hydration deduced by XRD/
Rietveld analysis in the laboratory cements (a) and commercial cements (b) in wt%
per mass of cement, error is ±2%.
In the case of the laboratory made samples, the initial intensi-
ties of the hemicarbonate peak is higher for the samples containing
less sulphate. In all samples, the intensities of hemicarbonate de-
crease after the second day of hydration and the similar intensities
are observed after 90 days of reaction. The intensities of monocar-
bonate peak increase parallel to the decrease of hemicarbonate
reflection for all the samples. The amount of monocarbonate and
thus the intensities of the monocarbonate peak after 90 days of
hydration are inversely correlated with the SO3/Al2O3 weight ratio
of the anhydrous cements. At lower SO3/Al2O3 less ettringite is
formed (Figs. 2 and 6) and thus more monocarbonate. The evolu-
 M. Zajac et al. / Cement & Concrete Composites 46 (2014) 99–108
Fig. 17. The intensity of hemicarbonate (solid lines) and monocarbonate (dash
lines) (Hc at 10.8° and Mc at 11.7° 2h) reflexes over the time in (a) laboratory
cements and (b) commercial cements.
tion of hemicarbonate and monocarbonate intensities with time
shows the same trends in the industrial and laboratory made sam-
ples (Fig. 17). The hemicarbonate peak intensities are significantly
higher for both C3.5-3 and C3.5-9 as compared to the C3.5-1, which
can be related to the lower calcite content in the C3.5-3 sample.
Fig. 17 indicates that the kinetics of hemicarbonate and mono-
carbonate formation (and thus of the calcite reaction) does not de-
pend on the amount of calcite present. The reaction of the calcite is
governed by the amount and kinetics of hemi- and monocarbonate
formation. This is supported by the observation that the same
kinetics of initially hemicarbonate and later monocarbonate for-
mation has been observed in Portland cement blended with calcite
or with dolomite [54]. If the formation of monocarbonate would be
limited by calcite reaction, the formation of monocarbonate would
have been much more retarded in the presence of dolomite only as
the dissolution of dolomite is more than 10-times slower than the
dissolution of the calcite [52].
The results of the experiments suggest that
i. Calcite starts to react only when the gypsum is depleted.
ii. The amount of reacted calcite is limited by the availability of
the alumina for the hemicarbonate and monocarbonate
formation.
iii. That the dissolution of calcite is limited by the rate of hemi-
and monocarbonate formation.
iv. That hemicarbonate formation is faster than monocarbonate
formation, even in the presence of a large surplus of calcite.
4. Conclusion
The present work shows the effect of different calcium sulphate
and limestone contents on the hydration and compressive strength
of the Portland cements.
107
The content of the calcium sulphate in the Portland cement has
a significant impact on the reaction of the clinker phases and lime-
stone as well as on the hydrate assemblage. During the first day,
the presence of more calcium sulphate accelerates the alite reac-
tion and retards aluminate reaction. After 1 day and longer, the
presence of more calcium sulphate results in the presence of more
ettringite and less AFm phases. The early compressive strength is
improved for the samples containing more calcium sulphate. At
longer hydration times, however, the formation of additional
ettringite had no positive influence on porosity or compressive
strength. This slight strength decrease, opposite to the expected in-
crease, seems to be related to an increase of the S/Si and Ca/Si ratio
and the presence of less water in the C–S–H.
The presence of calcite leads to the formation of hemicarbonate
and monocarbonate, thus stabilizing ettringite. The calcite reacts
only when the gypsum is depleted, thus earlier in a cement with
a higher Al2O3/SO3 ratio, where less ettringite is formed. Only a rel-
atively small quantity of calcite reacts to form monocarbonate or
hemicarbonate in Portland cement; somewhat more in a cement
with a high Al2O3/SO3 ratio as less ettringite but more hemi- or
monocarbonate forms as indicated both by thermodynamic model-
ing (Fig. 2) and experimental investigation (Fig. 16).
In contrast to the thermodynamic calculations, the formation of
hemicarbonate instead of monocarbonate is observed experimen-
tally during the first days. This initial formation of hemicarbonate
instead of monocarbonate could be related either to a slow lime-
stone dissolution kinetics at high pH values or to a faster formation
of hemicarbonate rather thanmonocarbonate. The experimental
investigations showed that the dissolution of limestone is suffi-
ciently fast in KOH solutions and that the amount of limestone
present had no influence on the kinetics of monocarbonate forma-
tion, thus indicating that a faster formation kinetic of hemicarbon-
ate is responsible. Monocarbonate is only slightly more stable than
hemicarbonate in Portland cement systems thus the driving force
and also the kinetics of the transformation of hemicarbonate into
the thermodynamically slightly more stable monocarbonate is low.
As reported previously e.g. [1], OPC systems with limestone
might show a better performance as limestone acts as a filler and
reacts with the cement that results in the higher volume of the hy-
drates. Here it is shown that the effect of limestone strongly de-
pends on the sulphate content of the cement and considered
time. The interaction between the limestone, sulphate and cement
clinker needs to be taken into account when optimizing the sul-
phate content in the cement plants. The higher content of sulphate
results in an acceleration of the early reaction of cement clinker
and stabilization of the bigger volume of ettringite. Both bring a
higher early compressive strength. Thus, the filler effect of the
limestone may be less pronounced. At later times, the higher sul-
phate content has a negative effect on the compressive strength
as described above, whereas the limestone improves strength.
The results demonstrate that the main effect of limestone on per-
formance is related to the stabilization of ettringite. It seems that
the slow transformation of hemicarbonate to monocarbonate has
a limited impact on the compressive strength in the case of inves-
tigated samples.
Acknowledgements
The financial support of the HeidelbergCement is gratefully
acknowledged. The authors would like to thank Angela Steffen,
Luigi Brunetti, and Boris Ingold for help with the laboratory exper-
iments, Mohsen Ben Haha for the SEM investigations, D. Rentsch
for his support during NMR experiments as well as Frank Winne-
feld and André Nonat for helpful discussions.
108 M. Zajac et al. / Cement & Concrete Composites 46 (2014) 99–108

References [28] Bullard JW. Mechanism of cement hydration. Cem Concr Res
2011;41(12):1208–23.
[29] Jansen D, Goetz-Neunhoeffer F, Lothenbach B, Neubauer J. The early hydration
[1] Hawkins P, Tennis P, Detwiler R. The use of limestone in Portland cement: a
of ordinary Portland cement (OPC): an approach comparing measured heat
state-of-the-art review. Skokie, Illinois, USA: Portland Cement Association;
flow with calculated heat flow from QXRD. Cem Concr Res 2012;42(1):134–8.
2003.
[30] Minard H, Garrault S, Regnaud L, Nonat A. Mechanisms and parameters
[2] Sprung S, Siebel E. Assessment of the suitability of limestone for producing
controlling the tricalcium aluminate reactivity in the presence of gypsum. Cem
Portland limestone cement (PKZ). Zement Kalk Gips 1991;44(1):1–11.
Concr Res 2007;37(10):1418–26.
[3] Barker AP, Cory HP. The early hydration of limestone-filled cements. In:
[31] Pourchet S, Regnaud L, Perez JP, Nonat A. Early C3A hydration in the presence
Swamy RN, editor. Proc blended cements in construction. Sheffield,
of different kinds of calcium sulfate. Cem Concr Res 2009;39(11):989–96.
UK: Elsevier; 1991. p. 107–24.
[32] Quennoz A, Scrivener K. Interactions between alite and C3A-gypsum
[4] Kakali G, Tsivilis S, Aggeli E, Bati M. Hydration products of C3A, C3S and
hydrations in model cements. Cem Concr Res 2013;44(1):46–54.
Portland cement in the presence of CaCO3. Cem Concr Res 2000;30(7):1073–7.
[33] Minard H. Etude intégrée des processus d’hydratation, de coagulation, de
[5] Lothenbach B, Le Saout G, Gallucci E, Scrivener K. Influence of limestone on the
rigidification et de prise pour un système C3S–C3A–sulfates–alcalins. PhD
hydration of Portland cements. Cem Concr Res 2008;38(6):848–60.
thesis. Université de Bourgogne, Dijon; 2003.
[6] De Weerdt K, Ben Haha M, Le Saout G, Kjellsen KO, Justnes H, Lothenbach B.
[34] Garrault S, Nachbaur L, Sauvaget C. Influence of alkali sulphate on the
Hydration mechanisms of ternary Portland cements containing limestone
rheological properties of tricalcium silicate pastes. Annales de Chimie: Sci des
powder and fly ash. Cem Concr Res 2011;41(3):279–91.
materiaux 2003;28:S43–50.
[7] Matschei T, Lothenbach B, Glasser FP. The role of calcium carbonate in cement
[35] Copeland LE, Kantro DL. Hydration of Portland cement. In: 5th Int Cong Chem
hydration. Cem Concr Res 2007;37(4):551–8.
Cem. Tokyo; 1968, vol. 2, p. 387–420.
[8] De Weerdt K, Kjellsen KO, Sellevold E, Justness H. Synergy between fly ash and
[36] Häcker CJ, Bentz DP. Einfluss des Sulfatträgers (Menge und Modifikation) auf
limestone powder in ternary cements. Cem Concr Compos 2011;33(1):30–8.
die Hydratation von Portlandzement: experimentelle Untersuchungen und
[9] Moesgaard M, Herfort D, Steenberg M, Frank Kirkegaard L, Yue Y. Physical
Computersimulationen. In: Proc, 14 Int Baustofftagung (ibausil). Germany:
performances of blended cements containing calcium aluminosilicate glass
Weimar; 2000. p. 1.0475–1.0483.
powder and limestone. Cem Concr Res 2011;41(3):359–64.
[37] Gunay S, Garrault S, Nonat A, Termkhajornkit P. Influence of calcium sulphate
[10] Carrasco MF, Menéndez G, Bonavetti V, Irassar EF. Strength optimization of
on hydration and mechanical strength of tricalcium silicate. In: Proc 13th ICCC,
‘‘tailor-made cement’’ with limestone filler and blast furnace slag. Cem Concr
Madrid, Spain; 2011.
Res 2005;35(7):1324–31.
[38] Bentur A. Effect of gypsum on the hydration and strength of C3S pastes. J Am
[11] Damidot D, Lothenbach B, Herfort D, Glasser FP. Thermodynamics and cement
Ceram Soc 1976;59(5–6):210–3.
science. Cem Concr Res 2011;41(7):679–95.
[39] Tang FJ. Optimization of sulfate form and content. Portland Cement
[12] Ipavec A, Gabrovgek R, Vuk T, Kaucic V, Macek J, Meden A. Carboaluminate
association. Research and development bulletin RD105T; 1992. p. 13–14.
phases formation during the hydration of calcite-containing Portland cement. J
[40] Kantro DL, Weise CJ. Hydration of various beta-calcium silicate preparations. J
Am Ceram Soc 2010;94(4):1238–42.
Am Ceram Soc 1979;62:621–6.
[13] Antoni M, Rossen J, Martirena F, Scrivener K. Cement substitution by a
[41] Barbarulo R, Peycelon H, Leclercq S. Chemical equilibria between C–S–H and
combination of metakaolin and limestone. Cem Concr Res
ettringite, at 20 and 85 °C. Cem Concr Res 2007;37(8):1176–81.
2012;42(12):1579–89.
[42] Labbez C, Pochard I, Nonat A, Jönsson B. Colloidal behavior of C–S–H
[14] Jelenic I, Panovic A, Halle R, Gasesa T. Effect of gypsum on the hydration and
nanohydrates in cement paste. In: Proc CONMOD 2010. Switzerland:
strength development of commercial Portland cements containing alkalis
Lausanne; 2010.
sulfates. Cem Concr Res 1977;7(3):239–46.
[43] Copeland LE, Bodor E, Chang TN, Weise CJ. Reactions of tobermorite gel with
[15] Nakagawa K, Isozoki K, Watanabe Y. Hydration and strength of normal
aluminates, ferrites, and sulfates. J PCA Res Develop Lab 1967;9:61–74.
Portland cement admixed with anhydrous calcium sulphate. In: Proc 7th Int
[44] Skibsted J, Jakobsen HJ. Characterization of the calcium silicate and aluminate
Cong Chem Cem. Paris; 1980, vol. II, p. 192–7.
phases in anhydrous and hydrated Portland cements. In: Colombet P, Grimmer
[16] Le Jean Y. Clinker grinding method and secondary components. Their effect
AR, Zanni H, Soozzani P (editors). Nuclear magnetic resonance spectroscopy of
upon cement reactivity and rheology’. In: Proc 7th Int Cong Chem Cem. Paris;
cement-based materials. Berlin; 1998. p. 3–45.
1980, vol. II, p. 252–8 [in French].
[45] Faucon P, Charpentier T, Bertrandie D, Nonat A, Virlet J, Petit JC.
[17] Soroka I, Abayneh M. Effect of gypsum on properties and internal structure of
Characterization of calcium aluminate hydrates and related hydrates of
PC paste. Cem Concr Res 1986;16(4):495–504.
cement pastes by 27Al MQ-MAS NMR. Inorgan Chem 1998;37(15):3726–33.
[18] Matschei T, Costoya M. A contribution to an improved understanding of the
[46] Poppe AM, De Schutter G. Cement hydration in the presence of high filler
hydration kinetics of OPC. In: Proc 18 Int Baustofftagung (ibausil). Germany:
contents. Cem Concr Res 2005;35(12):2290–9.
Weimar; 2012, V1.21.
[47] Cyr M, Lawrence P, Ringot E. Efficiency of mineral admixtures in mortars:
[19] Tsivilis S, Voglis N, Photou J. Technical note: a study on intergrading of clinker
quantification of the physical and chemical effects of fine admixtures in
and limestone. Miner Eng 1999;12(7):837–40.
relation with compressive strength. Cem Concr Res 2006;36(2):264–77.
[20] Schiller B, Ellerbrock HG. Mahlung und Eigenschaften von Zementen mit
[48] Runcevski T, Dinnebier RE, Magdysyuk OV, Pöllmann H. Crystal structures of
mehreren Hauptbestandteilen. Zement Kalk Gips 1992;45(9):1951–6.
calcium hemicarboaluminate and carbonated calcium hemicarboaluminate
[21] Le Saout G, Füllmann T, Kocaba V, Scrivener K. Quantitative study of
from synchrotron powder diffraction data. Acta Crystallogr Sect B – Struct Sci
cementitous materials by X-ray diffraction/Rietveld analysis using an
2012;68(5):493–500.
external standard. In: Proc 12th Int Cong Chem Cem. Montréal, Canada; 2007.
[49] Kuzel HJ, Pöllmann H. Hydration of C3A in the presence of Ca(OH)2,
[22] Matschei T, Lothenbach B, Glasser FP. The AFm phase in Portland cement. Cem
CaSO4
2H2O and CaCO3. Cem Concr Res 1991;21(5):885–95.
Concr Res 2007;37(2):118–30.
[50] Chou L, Garrels RM, Wollast R. Comparative study of the kinetics and
[23] Kulik D, Wagner T, Dmytrieva SV, Kosakowski G, Hingerl F, Chudnenko KV,
mechanisms of dissolution of carbonate minerals. Chem Geol 1989;78(3–
et al. GEM-Selektor geochemical modeling package: revised algorithm and
4):269–82.
GEMS3K numerical kernel for coupled simulation codes. Comput Geochem
[51] Inskeep WP, Bloom PR. An evaluation of rate equations for calcite precipitation
2013;17(1):1–24.
kinetics at pCO2 less than 0.01 atm and pH greater than 8. Geochim
[24] Thoenen T, Kulik D. Nagra/PSI chemical thermodynamic database 01/01 for
Cosmochim Acta 1985;49(10):2165–80.
GEMS-selektor (V.2-PSI) geochemical modeling code. <http://gems.web.psi.ch/
[52] Morse JW, Arvidson RS. The dissolution kinetics of major sedimentary
doc/pdf/TM-44-03-04-web.pdf PSI>, Villingen; 2003.
carbonate minerals. Earth-Sci Rev 2002;58(1):51–84.
[25] Hummel W, Berner U, Curti E, Pearson FJ, Thoenen T. Nagra/PSI chemical
[53] Pokrovsky OS, Golubev SV, Jordan G. Effect of organic and inorganic ligands on
thermodynamic data base 01/01, Universal Publishers/uPUBLISH.com, USA
calcite and magnesite dissolution rates at 60 °C and 30 atm pCO2. Chem Geol
also published as Nagra Technical Report NTB 02-16. Switzerland: Wettingen;
2009;265(1–2):33–43.
2002.
[54] Zajac M, Dienemann W, Bolte G. Comparative experimental and virtual
[26] Matschei T, Lothenbach B, Glasser FP. Thermodynamic properties of Portland
investigations of the influence of calcium and magnesium carbonates on the
cement hydrates in the system CaO–Al2O3–SiO2–CaSO4–CaCO3–H2O. Cem
reacting cement. In: Proc 13th ICCC. Madrid, Spain; 2011.
Concr Res 2007;37(10):1379–410.
[27] Lothenbach B, Matschei T, Möschner G, Glasser FP. Thermodynamic modelling
of the effect of temperature on the hydration and porosity of Portland cement.
Cem Concr Res 2008;38(1):1–18.