Alkali-silica reaction - the influence of calcium on silica dissolution and the
formation of reaction products
A. Leemann*, G. Le Saout, F. Winnefeld, D. Rentsch, B. Lothenbach
Empa, Swiss Federal Laboratories for Materials Science and Technology, 8600 Dübendorf,
In a model system for alkali-silica reaction consisting of micro silica, portlandite (0 to 40
mass-%) and 1M alkaline solutions (NaOH, KOH), the influence of calcium on silica
dissolution and the formation of reaction products are investigated. The reaction and its
products are characterized using calorimetry, X-ray diffraction, thermogravimetry analysis,
nuclear magnetic resonance, desorption experiments and pore solution analysis in
combination with thermodynamic modeling. Silica dissolution proceeds until portlandite is
consumed due to the formation of C-S-H and subsequently saturation of dissolved silica in
the alkaline solution is reached. As a result, the amount of dissolved silica increases with
increasing portlandite content. Depending on the amount of portlandite added, the reaction
products show differences in the relative amounts of Q1, Q2 and Q3 sites formed and in their
average Ca/Si-ratio. The ability of the reactions products to chemically bind water decreases
with decreasing relative amount of Q3 sites and with increasing Ca/Si-ratio. However, the
amount of physically bound water in the reaction products reaches a maximum value at a
Ca/Si-ratio between 0.20 and 0.30.
* corresponding author: email@example.com
A. Leemann, G. Le Saout, F. Winnefeld, D. Rentsch, B. Lothenbach, Alkali-silica reaction - the influence of calcium on
silica dissolution and the formation of reaction products, J. Am. Ceram. Soc., 94  1243–1249 (2011)
Alkali-silica-reaction (ASR) is one of the most important deterioration mechanisms in
concrete leading to substantial damages of structures worldwide.1,2 When amorphous or
crystalline SiO2 in aggregates reacts with the alkaline pore solution of concrete (pH 13.0-
13.5), expansion may result. The generated stress may eventually cause cracks in the concrete
years after the affected structures were built. Since the first cases were reported3, ASR is in
the focus of concrete research. Nevertheless, several questions remain unanswered. This
applies in particular to the role of calcium in regard to silica dissolution, the formation of the
reaction products and the resulting expansion.4-10
It is well known that silica dissolution increases with increasing alkalinity and temperature.11-
14 But the effect of calcium on the dissolution of reactive minerals in concrete aggregates is
not clear. Calcium plays an important role in the formation of the reaction products as it
controls aggregation and gelation processes in silico-alkaline solutions.4,15 However, the
available information on the influence of calcium on expansion due to ASR is not consistent;
some studies show that the presence of calcium decreases expansion while others conclude
the opposite.5,8,16-23 Apart from calcium, the potassium-sodium-ratio in the pore solution may
have an additional effect on ASR with a high sodium content leading to higher expansion.24-28
In this work, a model system is used to study the effect of calcium on silica dissolution and
on the composition of the reaction products. Micro silica (MS) assumes the role of the
reactive SiO2 in the concrete aggregates. The alkaline pore solution in concrete is represented
by 1M NaOH and KOH solutions. Due to the presence of relatively large amounts of
portlandite in the hardened cement paste, the concrete pore solution is saturated in calcium. In
order to imitate this situation and to investigate the effect of calcium, varying amounts of
portlandite are added. As the main components in the model system (silica, alkaline solution
and portlandite) are identical to the components taking part in ASR in mortar and concrete,
similar reaction products can be expected. The chosen ratio of solids to liquid leads only to
partial dissolution of the micro silica, resembling the situation in concrete where reactive
minerals are usually only partially dissolved. In this system, reaction times are relatively
short due to the high surface area of the micro silica. Furthermore, the simplicity of the
system facilitates material characterization. Silica dissolution and the formation of reaction
products are investigated using calorimetry, thermogravimetry analysis (TGA), X-ray
29Si solid state nuclear magnetic resonance (NMR), desorption
experiments and pore solution analysis in combination with thermodynamic modeling.
II. Experimental Procedure
The composition of the commercially available MS is given in Table I. The mass-ratio
between solid and alkaline solution was kept constant at 1:2 for all experiments. Different
amounts of portlandite were added to MS ranging from 2.5 to 40 mass-% (Table II). 1M
solutions of NaOH and KOH were used as reagents.
Isothermal conduction calorimetry measurements (Thermometric TAM Air) started
immediately after mixing the samples for a duration of 144 h. Heat flow and cumulative heat
refer to the total amount of solid consisting of MS and portlandite. The information about the
duration of the reaction was used to define the point in time for further sample analysis.
Samples for XRD, TGA and NMR were prepared in a glove box under N2-atmosphere. After
a reaction for 96 h (MS0-MS10) and 144 h (MS20 and MS40), the samples were immersed in
isopropanol, treated with diethyl ether and ground to a grain size < 63 µm.
X-ray diffraction (XRD) data were collected using a PANalytical X’Pert Pro MPD
diffractometer in a θ−2θ configuration employing CuKα radiation (λ=1.54 Å) with a fixed
divergence slit size of 0.5° and a rotating sample stage. The samples were scanned between
5° and 80° 2θ with an X’celerator detector.
Before conducting TGA (Mettler Toledo TGA/SDTA 851e), the samples were stored in a
desiccator under N2-atmosphere above CaCl2 solution (approximately 25 % relative humidity
(RH)) at 20 °C for four weeks. The water determined with TGA is referred to as “chemically
bound water” and refers to the total amount of anhydrous solids.
The 29Si NMR spectra were recorded on a Bruker ASX 400 spectrometer (9.4T magnetic
field) at 79.5 MHz and 4.5 kHz spinning rate in a 7 mm ZrO2 CP MAS probe. Single pulse
experiments were carried out by applying 90° pulses of 8.8 µs with 1H decoupling at 31.3
kHz (TPPM15) and a recycle delay of 100 s in order to respect the relaxation times T1 of the
species present in the samples (T1 = 20 s for silica fume determined by a T1 saturation
recovery experiment). The 29Si chemical shifts of the peaks were analyzed using the Qn
classification, where the Si tetrahedron is connected to n Si tetrahedra with n varying from 0
to 4. The 29Si chemical shift was referenced externally relative to tetramethylsilane (TMS) at
0 ppm. The relative proportions of silicon associated with the Qn units were determined by
the deconvolution of the spectra and measurement of the area associated with each peak using
the Dmfit program.29 The amount of unreacted MS was defined based on the quantity of
remaining Q4 sites.
The amount of adsorbed water was measured on samples reacted in 1M NaOH (portlandite
content: 2.5-40 mass-%). Only one sample (MS10) was mixed with 1M KOH. After a
reaction time of 144 h, the samples were stored without any additional preparation in
desiccators with a N2-atmosphere and an initial relative humidity of 97 % RH (adjusted with
saturated K2SO4 solution) at a temperature of 20 °C. After the excess water was evaporated
and the samples reached constant mass, they were moved to 94 % RH (adjusted with
saturated KNO3 solution) and subsequently to 86 % RH (adjusted with saturated KCl
solution) after again reaching constant mass. RH in the desiccators was periodically
controlled using a calibrated hygrometer. These humid climates were chosen as ASR takes
place in concrete exposed to high relative humidity.30,31 As one part of the water is
chemically bound in reaction products and the other part is physically bound, the determined
values are referred to as “total water”. The amount of physically bound water was calculated
by subtracting the chemically bound water from total water content.
Pore solution analysis was conducted on MS10 reacted in 1M NaOH and 0.85M KOH
respectively. The samples (~ 200 g of solid) were mixed for 90 s according to EN 196-3 in a
Hobart mixer. The solutions of the samples with a reaction time shorter than four hours were
collected by vacuum filtration using 0.45 μm nylon filters. Samples with longer reaction
times were filled in 500 ml liter PE-bottles, sealed and stored at 20 °C. Pore fluids of the
hardened samples were extracted using the steel die method and pressures up to 50 MPa
according to the method described in Longuet et al.32. After immediate filtration (0.45 μm
nylon filters), one part was diluted by a factor of 10 with HNO3 (6.5 %) to prevent the
precipitation of solid phases, while the remaining fraction was used for pH measurements.
The pH electrode had been calibrated against KOH solutions of known concentrations. The
total concentrations of the elements analyzed were determined by ion chromatography.
Thermodynamic modeling was carried out using the Gibbs free energy minimization program
GEMS.33 GEMS is a broad-purpose geochemical modeling code which computes equilibrium
phase assemblage and speciation in a complex chemical system from its total bulk elemental
composition. Chemical interactions involving solids, solid solutions, and aqueous electrolytes
are considered simultaneously. The speciation of the dissolved species as well as the kind and
amount of solids precipitated are calculated. The thermodynamic data for aqueous species as
well as for many solids were taken from the PSI-GEMS thermodynamic database.34,35
Solubility products for calcium-silicate-hydrates (C-S-H) were taken from the recent
cemdata07 compilation.34 For C-S-H an ideal solid solution between a jennite and a
tobermorite-type C-S-H was used, but no solid solution between tobermorite and amorphous
The duration of heat release and cumulative heat are increasing with increasing portlandite
content (see Fig. 1). The point of maximum heat flow occurs later with increasing portlandite
content, indicating a faster reaction in the absence of portlandite. The duration of the reaction
increases with increasing portlandite content. There is little difference between the samples
reacted in NaOH and KOH solution, respectively (Table III). The results of calorimetry are
not used quantitatively but qualitatively; they show the duration of the reaction and the
change of kinetics with the addition of micro silica.
The qualitative determination of the mineral phases reveals an increasing amount of reacted
MS with increasing portlandite content as indicated by the decreasing signal at 20° 2θ (Fig.
2). Following the same order, the amount of C-S-H is increasing. Reacted portlandite is only
present in minor amounts in sample MS40. The peak at low angle shows that the C-S-H is
well ordered or has relatively large crystalline size as it is typical for C-S-H with a Ca/Si-ratio
lower than the one in ordinary Portland cement paste of 1.7.36 The peaks at 36° and 42° are
caused by silicon carbide impurities present in MS. The diffractograms for the samples
reacted in KOH (data not shown) are very similar.
The amount of chemically bound water per anhydrous solid (mass loss at 980°C) increases
with increasing portlandite content (Table III). The samples reacted in NaOH generally
contain more water than the ones reacted in KOH. In MS20 and MS40 some minor amounts
(< 0.2 mass-%) of calcite are present. MS40 additionally contains about 0.3 mass-% of
portlandite. This applies for samples reacted in NaOH and KOH.
(4) Water Adsorption
The sorption isotherms show that total water content of the samples increases with increasing
portlandite content (Table IV). The change in total water content between 97 and 94 % RH is
much larger than the change from 94 to 86 % RH. MS10 in NaOH and KOH shows similar
total water contents.
(5) 29Si NMR
The 29Si spectrum of unreacted MS presents a broad peak Q4 at -110 ppm characteristic of
amorphous SiO2 (Fig. 3). Silica dissolution increases with increasing amount of portlandite
(Fig. 4, Table V). The dissolution leads to the formation of silanol sites with three bridging
oxygen (Q3 OH) at -98 ppm and two bridging oxygen sites at -92 ppm (Q2 OH) as previously
reported in the study of silica gels.37
Under the action of Ca2+ ions, a Q1-Q2 based calcium silicate structure close to that of cement
C-S-H is formed with probably a pure alkali silicate inserted in it (Na-S-H).38,39 For MS40, a
good resolution is observed with narrow resonance lines at -79.3, -82.6, -85.2 and -88.9 ppm.
The positions agree well with those found in a previous study on calcium silicate hydrates40
and main peaks can be assigned to end chain tetrahedra Q1 (-79.3 ppm) and non-bridging
tetrahedra Q2 (-85.2 ppm). For MS40, the small intensity line at -82.6 ppm is attributed to the
bridging tetrahedral Q2,p with one hydroxyl and water molecules in the interlayer space
whereas the small shoulder at -88.9 ppm may be assigned to Q2 sites next to Q3 (Q2,v).40,41
The Q3 peak that should be present at around -92 ppm according to Brunet et al.40 certainly
overlaps the Q2(OH) line. This Q3 site that links two silicate chains of two adjacent layers is a
defect in comparison with the tobermorite structure but has already been observed at low
Ca/Si ratio.40,41 The amount of Q1,Q2-products increases steadily with increasing portlandite
content in MS, while the amount of Q3-products changes little from MS to MS20 but
decreases in MS40. This decrease goes together with the nearly complete dissolution of the
silica. The spectra of the products hydrated in NaOH and KOH are similar although, for
MS40, the Q1/Q2 ratio indicates a more polymerized structure for Na-C-S-H. While the
amount of unreacted silica is lower in MS10 reacted in NaOH compared to KOH, the
opposite applies to MS40.
So, the original MS not exposed to alkaline solutions consists entirely of Q4-sites. Q1,Q2,Q3-
sites are formed due to the hydroxyl attack on MS exposed to the alkaline solutions. There
are remaining Q4-sites in all samples exposed to alkaline solutions as only part of the silica
reacts. In the discussion of the results, “reaction products” are referred to as the sum of
Q1,Q2,Q3-sites including the reacted portlandite determined with TGA. “Unreacted silica”
refers to the quantity of remaining Q4-sites.
(6) Pore Solution
Calcium concentrations in the pore solution stay low during the entire reaction time (Table
VI). Sodium and potassium in the pore solution decrease gradually with time. After two days,
silicon concentrations increase rapidly reaching higher values in NaOH compared to KOH.
At the same time hydroxide concentration decreases significantly. The amount of sodium
remaining in the pore solution at 28 days is higher compared to potassium. This applies as
well for the samples with an initial 1/1-molar-ratio of sodium to potassium.
The amount of dissolved silica is mainly dependent on the portlandite content of the samples
as indicated qualitatively by XRD and confirmed quantitatively by NMR. As soon as silica is
dissolved in NaOH or KOH, it reacts with calcium to form C-S-H. The presence of
portlandite first retards the reaction but later prolongs it leading to a higher degree of silica
dissolution. It has to be taken into account that the ratio between alkali and silica of the
samples is increasing with increasing portlandite content as solid-liquid-ratio was kept
constant (Table II). However, while the NaOH/SiO2-ratio (mass) increases from MS to MS40
by a factor of 1.7, the amount of unreacted silica decreases by a factor of 13.5 at the same
time. Consequently, the portlandite content is the main factor determining the amount of
dissolved silica in the studied system. In spite of the lower pH, silica dissolution in NaOH is
not lower compared to KOH as indicated by total heat flow in calorimetry and NMR.
The process of silica dissolution and precipitation can be shown by calculating the saturation
indices of the involved components using the pore solution data in combination with
thermodynamic modeling (Fig. 5). A positive saturation index (SI) indicates oversaturation;
the precipitation of the respective solid is possible. A negative SI indicates undersaturation
and thus either the absence of a solid or its dissolution.
At the beginning of the reaction of MS10 (Fig. 5), the solutions are undersaturated with
respect to silica, thus the MS present dissolves. The solutions are initially also undersaturated
with respect to portlandite. During the first four hours, when portlandite still is present, its SI
is only moderately negative indicating the fast dissolution of portlandite to form C-S-H. After
16 hours and longer, all portlandite is consumed by C-S-H formation and the solutions are
more undersatured. With the disappearance of portlandite, the SI indicates the formation of a
tobermorite-like C-S-H. At the same time, the SI of silica increases as less silica is removed
from the pore solution by C-S-H formation due to portlandite disappearance. Only when no
additional C-S-H a formed (or formed slowly), the solution will equilibrate with the micro
silica. Consequently, saturation with respect to silica is reached between 2 and 7 days and
thus the dissolution reaction of MS is expected to stop. In addition, the pH of the solution
drops from 13.5 to below 12. This is consistent with the results of calorimetry (Fig. 1), where
a significant reaction of MS10 was observed during the first 4 days.
If no portlandite is added to MS, the reaction will initially produce alkali silica polymers (see
Fig. 3), but the reaction will be stopped soon as the solution becomes saturated with respect
to silica. In MS40, silica continues to react and form C-S-H during the first 6 days as
saturation of silica in the solution is not reached.
There are no significant differences between the reaction of MS10 in NaOH and KOH (not
presented). However, the higher amount of silicon in the NaOH solution (28 days) could
indicate the formation of a relatively high amount of dissolved Na-S-H complexes. This
confirms the results of an extensive study about the solubility of C-S-H (Ca/Si-ratio from 1.0
to 1.9) in alkaline solutions where a greater sorptive capacity of C-S-H for potassium
compared to sodium was observed.42 The composition and type of products formed during
the reaction change with increasing portlandite content. As the amount of Q1,Q2 sites
increases significantly with increasing portlandite content and the amount of Q3 sites changes
little, the relative amount of Q3 sites decreases (Fig. 4). Average Ca/Si-ratio of the reaction
products increases up to 0.62 in NaOH and 0.73 in KOH as calculated from the amount of
dissolved portlandite and silica. Moreover, it can be expected that the negative charge of
Q1,Q2,Q3 sites is more frequently balanced by calcium instead of alkalis with increasing
portlandite content. Such a replacement of alkalis by calcium in the reaction products seems
to take place as well in concrete structures.43 Furthermore, it agrees with the increase in
sodium content observed in C-S-H of decreasing Ca/Si-ratio.42 In case of MS10, the average
sodium and potassium contents of the reaction products can be assessed by the pore solution
data (Table VI); more potassium than sodium is bound leading to a K/Si-ratio of 0.21 and a
Na/Si-ratio of 0.18 in the reaction products.
The amount of chemically bound water in the reaction products decreases with decreasing
relative amount of Q3 sites and increasing Ca/Si-ratio (Fig. 6). However, substantially more
water in the reaction products is bound physically than chemically (Fig. 7). Contrary to the
chemically bound water, physically bound water content of the reaction products at 97 and 94
% RH shows no linear decrease with increasing Ca/Si-ratio but reaches a maximum at a
Ca/Si-ratio of 0.20 and 0.30. It has to be considered that the alkalis in the pore solution can
have an effect on the sorption isotherms like it was shown for cement-based materials.44 In
particular, they may account for a reduction of the water activity of several percent as
estimated by using Raoult’s law.45 However, as the alkali concentration in the pore solution
of the reaction products after the consumption of portlandite (see Table VI) is similar to the
one of mortar and concrete affected by ASR,28,46 the water binding capacity of the reaction
products of MS should be comparable to the water binding capacity of ASR reaction products
in the 86-97% relative humidity range. Additionally, the relative course of the sorption
isotherms from 84 to 97 % relative humidity are consistent with the ones commonly reported
on cementitious materials.47 The mechanism leading to ASR induced expansion in concrete is
not entirely understood. However, it has been shown that synthetic gels can expand and
develop a pressure of several MPa under restrained condition due to the physical adsorption
of water.18,23,48 Based on the high amount of physically adsorbed water, it is expected that the
potential of the reaction products for an expansion may be highest at a Ca/Si-ratio between
0.20 and 0.30. Moreover, if a connection between this model system and ASR in concrete is
made, the change in the amount of physically bound water at different RH is important as the
RH in a concrete is constantly changing due to fluctuations of RH and temperature of the
environment.49,50 The change in the amount of physically bound water in the reaction
products going from a relative humidity of 97 to 94 % and down to 86 % is the highest at a
Ca/Si-ratio of 0.20 and 0.30 (Fig. 8). This further emphasizes the expansion potential of the
reaction products at this Ca/Si-ratio. Reaction products in cracked concrete aggregates have
in fact a Ca/Si-ratio of 0.1 to 0.4.51-53 When the reaction products extrude the aggregates and
are deposited in the cement paste, their Ca/Si-ratio increases to a value of 0.5 to 1.052,53
approaching the Ca/Si-ratio of 1.7 typical for C-S-H formed by cement hydration.36 C-S-H
with Ca/Si-ratio 1.0 and lower is supposed to belong to a different stability field than C-S-H
with a Ca/Si-ratio between 1.2 and 1.9.42 However, due to the calcium up-take of the reaction
products their potential to physically absorb water and cause expansion should be
significantly decreased. The occurrence of nearly calcium-free reaction products collected at
the concrete surface of a dam damaged by AAR, as reported by Tambelli et al.54, seems to be
an exception. Nevertheless, these nearly calcium-free products are mainly composed of Q3
It has to be considered that not only the properties of the reaction products are important but
the amount present as well. A small amount of highly expansive reaction products may cause
less overall expansion than a higher amount of moderately expansive material. The amount of
reaction products that has to be present in a concrete aggregate in order to cause expansion
and cracking likely depends on the type of aggregate. In gneiss aggregates, substantial
expansion of concrete already takes place when 0.5 % of the minerals have reacted.55
It was observed in several studies that water adsorption and expansion of synthetic gels are
reduced by the addition of calcium.18,20,23,26,56 In synthetic gels, the products generally consist
mainly of Q3 sites. Water is supposed to be either incorporated in relatively large interlayers
as it is typical for kanemite20 or between nano-sized silicate particles.57 When calcium is
added, the exchange of alkalis with the higher charged calcium leads to a reduced ability to
swell. Lower repulsive forces and less expansion are generated in the presence of divalent
ions as calcium instead of monovalent alkalis.19,27,58 Additionally, a volume increase of the
reaction products going together with the formation of Q3 sites can occur, when silica is
dissolved in an alkali solution in the absence of calcium.59 Generally, there seems to be an
agreement that the ability of the reaction products to expand is decreasing with the addition
When such systems are compared to mortar and concrete, three important differences have to
be pointed out. Firstly, calcium in the experiments mentioned above is usually added after
silica dissolution and the formation of the initial reaction products. In concrete, portlandite is
already present at the start. Secondly, a high mass ratio between alkaline solution and solid is
used, whereas in mortar and concrete the amount of pore solution is relatively small. Thirdly,
the degree of silica dissolution in such experiments is very high or even complete, leading to
a high amount of reaction products. In mortar and concrete, only a small fraction of
amorphous or crystalline SiO2 in aggregates reacts. These differences could be the reason for
seemingly contradictory results about the influence of calcium on ASR. It was shown
empirically in several studies by systematically increasing or decreasing the availability of
portlandite, that the presence of portlandite is mandatory for the expansion of mortars. 5,8,17 It
can be assumed that in these studies Q3 sites were formed in the reactive aggregates as the
pore solution of the mortars was highly alkaline. This seems to indicate that silica dissolution
and the formation of Q3 sites alone is not sufficient to cause expansion in mortars or concrete.
In the absence of portlandite, either the amount of reaction products formed is too small to
cause expansion or their properties are not suitable to exercise expansive pressure.
Looking at the effect of the different alkalis, the amount of chemically bound water at a
Ca/Si-ratio of about 0.26 is lower in the sample reacted in KOH compared to NaOH, despite
the higher amount of Q3 sites. The content of physically bound water is higher for the sample
reacted in NaOH as well. The trend of the reaction products with NaOH to bind and adsorb
more water compared to ones with KOH agrees with the findings of Wieker et al..26
A model system for ASR consisting of micro silica, portlandite and alkaline solutions is used
to study the effect of calcium on silica dissolution and on the composition of the reaction
products. The ratio between solids and liquid is chosen so that only partial dissolution of the
micro silica takes place. The system permits a precise characterization of the reaction
products, providing insights that are difficult or impossible to obtain in more complicated
systems like mortar or concrete. Even if differences between the reactions taking place in the
model system and the ones present in mortar or concrete are probable, the basic mechanisms
of ASR can be expected to be similar.
The amount of calcium present in form of portlandite plays a major role in the degree of
silica dissolution and in the type of products formed:
• Silica is dissolved by the alkaline solution until saturation is reached. When
portlandite is present, it reacts with the dissolved silica forming C-S-H with a low
Ca/Si-ratio that contains alkalis. As a consequence, silica is removed from the
solution enabling further silica dissolution. Because this process proceeds until
portlandite is consumed, the amount of portlandite present determines how much
silica can be dissolved.
• Silica dissolution starts with the formation of Q3 sites and the ensuing formation of
Q1,Q2 sites in the presence of calcium as determined by 29Si NMR. The Q1,Q2-based
calcium silicate structures are close to that of C-S-H.
• The total amount of Q3 sites in the different samples is very similar but the amount of
Q1,Q2 sites increases with increasing portlandite content. As a results, the relative
amount of Q3 sites in the reaction products decreases with increasing portlandite
content. At the same time, the average Ca/Si-ratio of the reaction products is
Whereas the amount of dissolved silica increases with increasing portlandite content, the
simultaneous increase in the Ca/Si-ratio of the reaction products affects their ability to bind
• The amount of chemically bound water in the reaction products decreases with
decreasing relative amount of Q3 sites and increasing Ca/Si-ratio.
• The amount of physically bound water is considerably higher than the amount of
chemically bound water. At 97 and 94 % RH, the total water content (chemically and
physically bound water) does not decrease with increasing Ca/Si ratio but shows a
maximum between Ca/Si-ratio of 0.20 and 0.30. The reactions products in this Ca/Si-
range show the largest change in total water content going from a RH of 97 to 94%
and from 94 to 86 %.
• Total water content of the reaction products formed in NaOH and KOH are similar,
but products formed in NaOH contain more chemically bound water.
The authors thank J. Lin for preliminary experiments, L. Brunetti and B. Ingold for sample
preparation, and P. Lura for the critical review of the manuscript.
1 M. A. Bérubé, B. Fournier, and B. Durand (Editors), Alkali-aggregate reaction in
concrete, Proceedings of the 11th International Conference on Alkali-Aggregate
Reaction in Concrete, Québec City, Canada, 2000.
2 M. Tang, and M. Deng (Editors), Alkali-aggregate reaction in concrete, Proceedings of
the 12th International Conference on Alkali-Aggregate Reaction in Concrete,
International Academic Publishers, World Publishing Corporation, Beijing, China, 2004
3 T.E. Stanton, “Expansion of concrete through reaction between cement and aggregate”;
pp. 1781–1811 in Proceedings of the American Society of Civil Engineers, 66, 1940.
4 L. S. D. Glasser, and N. Kataoka, “On the role of calcium in the alkali-aggregate
reaction,” Cem. Concr. Res., 12, 321-31 (1982).
5 H. Wang, and J. E. Gillott, “Mechanism of alkali-silica reaction and the significance of
calcium hydroxide,” Cem. Concr. Res., 21, 647-54 (1991).
6 R. Helmuth, D. Stark, S. Diamond, and M. Moranville-Regourd, Alkali-silica reactivity:
An overview of research, Strategic Highway Research Program, National Research
Council, Washington DC, 1993.
7 S. Diamond, “ASR - another look at mechanisms,” pp. 83-94 in Proceedings of the 8th
International Conference on Alkali-Aggregate Reaction in Concrete. Edited by K.
Okada, S. N. Nishibayashi, and M. Kawamura. Elsevier Applied Science, Kyoto, Japan,
8 M. D. A. Thomas, “The role of calcium in alkali-silica reaction”; pp. 325-335 in
Proceedings of the Sidney Diamond Symposium on Materials Science and Engineering
of Concrete and Cementious based Composites. Edited by M. Cohen, S. Mindess, and J.
Skalny. Materials Science of Concrete, American Ceramic Society, Westerville, OH,
9 M. C. G. Juenger, and C. P. Ostertag, “Alkali–silica reactivity of large silica fume-
derived particles,” Cem. Concr. Res., 34, 1389-402 (2004).
10 X. Hou, L. J. Struble, and R. J. Kirkpatrick, “Formation of ASR gel and the roles of C-S-
H and portlandite,” Cem. Concr. Res., 34, 1683-96 (2004).
11 R. K. Iller, The chemistry of silica, John Wiley & Sons, New York, 1979.
12 W. L. Marshall, “Amorphous silica solubilities I. Behaviour in aqueous sodium nitrate
solutions: 25-300°C, 0-6 molal,” Geochim. Cosmochim. Acta, 44, 907-13 (1980).
13 K. G. Knauss, and T. J. Wolery, “The dissolution kinetics of quartz as a function of pH
and time at 70°C,” Geochim. Cosmochim. Acta, 52, 43-53 (1988).
14 J. P. Icenhower, and P. M. Dove. “The dissolution kinetics of amorphous silica into
sodium chloride solutions: effects of temperature and ionic strength,” Geochim.
Cosmochim. Acta, 64, 4193-203 (2000).
15 F. Gaboriaud, A. Nonat, and D. Chaumont, “Aggregation and gel formation in basic
silico-calco-alkaline solutions studied: a SAXS, SANS, and ELS study,” J. Phys. Chem.
B, 103, 5775-81 (1999).
16 T. C. Powers, and H. H. Steinmour, “An interpretation of some published researches on
the alkali-aggregate reaction. Part I: The chemical reaction and mechanism of
expansion,” J. Am. Concr. Inst., 26, 497-516 (1955).
17 S. Chatterji, “Role of Ca(OH)2 in the breakdown of Portland cement concrete due to
alkali-silica reaction,” Cem. Concr. Res., 9, 185-8 (1979).
18 L. J. Struble, and S. Diamond, “Swelling Properties of Synthetic Alkali Silica Gels,” J.
Am. Ceram. Soc., 64, 652-5 (1981).
19 M. Prezzi, P. J. M. Monteiro, and G. Sposito, “The alkali-silica reaction, part I: use of the
double layer theory to explain the behaviour of reaction-product gels,” ACI Mater. J., 94,
20 W. Wieker, C. Hubert, D. Heidemann, and R. Ebert, “Alkali-aggregate reaction - A
problem of the insufficient fundamental knowledge of its chemical base”; pp. 395–408 in
Proceedings of the Sidney Diamond Symposium on Materials Science and Engineering
of Concrete and Cementious based Composites. Edited by M. Cohen, S. Mindess, J.
Skalny. Materials Science of Concrete, American Ceramic Society, Westerville, OH,
21 K. E. Kurtis, C. L. Collins, and P. J. M. Monteiro, “The surface chemistry of the alkali-
silica reaction: a critical evaluation and x-ray microscopy,” Concr. Sci. Eng., 4, 2-11
22 M. Kawamura, and K. Iwahori, “ASR gel composition and expansive pressure in mortars
under restrained conditions,” Cem. Concr. Compos., 26, 47-56 (2004).
23 T. Mansfeld, Das Quellverhalten von Alkalisilikatgelen unter Beachtung ihrer
Zusammensetzung (Expansion behaviour of alkali silicate gels considering their
composition), Doctoral Thesis, Bauhaus-Universität Weimar, Germany, 2008.
24 S. Chatterji, N. Thaulow, and A. D. Jensen, “Studies of alkali-silica reaction. Part 4.
Effect of different alkali salt solutions on expansion,” Cem. Concr. Res., 17, 777-83
25 P. W .J. G. Wijnen, T. P. M. Beelen, J. W. de Haan, C. P. J. Rummens, L. J. M. van de
Ven, and R. A. van Santen, “Silica gel dissolution in aqueous alkali metal hydroxides
studied by 29Si NMR,” J. Non-Cryst. Solids, 109, 85-94 (1989).
26 W. Wieker, C. Hübert, D. Heidemann, and R. Ebert, “Some experiences in chemical
modelling of the alkali-silica reaction”; pp. 119-128 in Proceedings of the 11th
International Conference on Alkali-Aggregate Reaction in Concrete. Edited by M. A.
Bérubé, B. Fournier, and B. Durand. Québec, Canada, 2000.
27 F. A. Rodrigues, P. J. M. Monteiro, and G. Sposito, “The alkali-silica reaction. The effect
of monovalent and bivalent cations on the surface charge of opal,” Cem. Concr. Res., 31,
28 A. Leemann, and B. Lothenbach, “The influence of potassium–sodium ratio in cement on
concrete expansion due to alkali-aggregate reaction,” Cem. Concr. Res., 38, 1162-8
29 D. Massiot, F. Fayon, M. Capron, I. King, S. Le Calvé, B. Alonso, J. O. Durand, B.
Bujoli, Z. Gan, and G. Hoatson, “Modelling one-and two-dimensional solid-state NMR
spectra,” Magn. Reson. Chem., 40, 70-6 (2002).
30 H. Olafson, “The effect of relative humidity and temperature an alkali expansion of
mortar bars”; pp. 461-465 in Proceedings of the 7th International Conference on Alkali-
Aggregate Reaction in Concrete. Edited by E. Grattan-Bellew. Noyes Publications,
Ottawa, Canada, 1986.
31 T. Kurihara, and K. Katawaki, “Effects of moisture control and inhibition on alkali silica
reaction”; pp. 881-885 in Proceedings of the 8th International Conference on Alkali-
Aggregate Reaction in Concrete. Edited by K. Okada, S. N. Nishibayashi, and M.
Kawamura. Elsevier Applied Science, Kyoto, Japan, 1989.
32 P. Longuet, L. Burglen, and A. Zelwer, “La phase liquide du ciment hydraté,” Rev.
Matér. Constr., 676, 35-41 (1973).
33 D. Kulik, GEMS-PSI 2.2, available at http://gems.web.psi.ch/, PSI-Villigen, Switzerland,
34 T. Thoenen, and D. Kulik, Nagra/PSI chemical thermodynamic database 01/01 for the
GEM-Selektor (V.2-PSI) geochemical modeling code, PSI, Villigen; available at
35 W. Hummel, U. Berner, E. Curti, F. J. Pearson, and T. Thoenen, Nagra/PSI Chemical
Thermodynamic Data Base 01/01, Universal Publishers/uPUBLISH.com, USA, 2002.
36 H. F. W. Taylor, Cement chemistry, Thomas Telford Publishing, 2nd edition, London,
37 G. E. Maciel, and D. W. Sindorf, “Silicon-29 Nuclear Magnetic Resonance Study of the
Surface of Silica Gel by Cross Polarization and Magic-Angle-Spinning,” J. Am. Chem.
Soc., 102, 7606-7 (1980).
38 X. D. Cong, R. J. Kirkpatrick, and S. Diamond, “29Si MAS NMR spectroscopic
investigation of alkali silica reaction product gels,” Cem. Concr. Res., 23, 811-23 (1993).
39 H. Zanni, L. Fernandez, R. Couty, P. Barret, A. Nonnat, and D. Bertrandie, “NMR study
of concrete alkali-aggregates reaction”; pp. 263-276 in Application of NMR
spectroscopy to cement science. Edited by P. Colombet, and A. R. Grimmer. Gordon and
Breach, Amsterdam, Netherlands, 1994.
40 F. Brunet, P. Bertani, T. Charpentier, A. Nonnat, and J. Virlet, “Application of 29Si
Homonuclear and 1H-29Si Heteronuclear NMR Correlation to Structural Studies of
Calcium Silicate Hydrates,” J. Phys. Chem. B, 108, 15494-502 (2004).
41 I. Klur, B. Pollet, J. Virlet, and A. Nonat, “C-S-H structure evolution with calcium
content by multinuclear NMR”; pp. 119-141 in Nuclear Magnetic Resonance
Spectroscopy of Cement-Based Materials. Edited by P. Colombet, A. R. Grimmer, H.
Zanni, and P. Soozzani. Springer, Berlin, Germany, 1998.
42 D. E. Macphee, K. Luke, F.P. Glasser, and E.E. Lachowski, “Solubility and aging of
calcium silicate hydrates in alkaline solutions at 25°C,” J. Am. Ceram. Soc., 72, 646-54
43 M. D. A. Thomas, “The role of calcium hydroxide in alkali recycling in concrete”, pp.
225–236 in Proceedings of the Workshop on the Role of Calcium Hydroxide in
Concrete. Edited by J. Skalny. Materials Science of Concrete, American Ceramic
Society, Westerville, OH, 2001.
44 T. C. Powers, and T.L. Brownyard, ‘‘Studies of the physical properties of hardened
Portland cement paste,’’ J. Am. Concr. Inst. 43 (nine parts: Oct. 1946 - April 1947),
Bulletin 22, Research Laboratories of the Portland Cement Association, Chicago, 1948.
45 P. Lura, O. M. Jensen, and K. van Breugel, “Autogenous shrinkage in high-performance
cement paste: An evaluation of basic mechanisms,” Cem. Conc. Res., 33, 223-32 (2003).
46 A. Leemann, B. Lothenbach, and C. Thalmann, “Influence of superplasticizers on pore
solution composition and on expansion of concrete due to alkali-silica reaction,” Constr.
Build. Mater., doi:10.1016/j.conbuildmat.2010.06.019.
47 R. M. Espinosa, and L. Franke, “Influence of the age and drying process on pore
structure and sorption isotherms of hardened cement paste,” Cem. Conc. Res., 36, 1969-
48 H. E. Vivian, “The reaction products of alkalies and opal”; pp. 60-81 in Studies in
Cement-Aggregate reaction. CSIRO Bulletin No. 256. Commonwealth Scientific and
Industrial Research Organization, Australia, 1950.
49 L. O. Nilsson, “Interaction between microclimate and concrete - a prerequisite for
deterioration,” Constr. Build. Mater., 10, 301-8 (1996).
50 C. Andrade, J. Sarría, and C. Alonso, “Relative humidity in the interior of concrete
exposed to natural and artificial weathering,” Cem. Concr. Res., 29, 1249-59 (1999).
51 T. Knudson, and N. Taulow, “Quantitative microanalyses of alkali-silica gel in
concrete,” Cem. Concr. Res., 5, 443-54 (1975).
52 N. Thaulow, U. H. Jakobson, and B. Clark, “Composition of alkali silica gel and
ettringite in concrete railroad ties: SEM-EDX and X-ray diffraction analysis,” Cem.
Concr. Res., 26, 309-18 (1996).
53 T. Katayama, “ASR gel in concrete subject to freeze-thaw cycles - comparison between
laboratory and field concretes from Newfoundland, Canada”; In Proceedings of the 13th
International Conference on Alkali-Aggregate Reaction in Concrete, Trondheim,
54 C. E. Tambelli, J. F. Schneider, N. P. Hasparyk, and P. J. M. Monteiro, “Study of the
structure of alkali–silica reaction gel by high-resolution NMR spectroscopy,” J. Non-
Cryst. Solids, 352, 3429-36 (2006).
55 M. Ben Haha, E. Gallucci, A. Guidoum, and K. L. Scrivener, “Relation of expansion due
to alkali silica reaction to the degree of reaction measured by SEM image analysis,”
Cem. Concr. Res., 37, 1206-14 (2007).
56 H. Krogh, “Examination of synthetic alkali-silica gels”; pp. 133-163 in Symposium on
Alkali-Aggregate Reaction. Icelandic Building Research Institute. Reykjavik, Iceland,
57 R. J. Kirkpatrick, A. G. Kalinichev, X. Hou, and L. Struble, “Experimental and
molecular dynamics modeling studies of interlayer swelling: water incorporation in
kanemite and ASR gel,” Mater. Struct., 38, 449-58 (2005).
58 R. Kjellander, S. Marcelja, R. M. Pashley, and J. P. Quirk, “Double-layer ion correlation
forces restrict calcium-clay swelling,” J. Phys. Chem., 92, 6489-92 (1988).
59 E. Garcia-Diaz, J. Riche, D. Bulteel, and C. Vernet, “Mechanism of damage for the
alkali–silica reaction,” Cem. Concr. Res., 36, 395-400 (2006).
Fig. 1: Heat flow of MS with different portlandite contents reacted in 1M NaOH.
Fig. 2: XRD patterns of MS with different portlandite contents after a reaction time of 96 h
(MS0-MS10) and 144 h (MS20 and MS40) in 1M NaOH (CH = portlandite, CC = calcite,
SiC = silicon carbide). “MS (original)” refers to the micro silica not exposed to alkaline
Fig. 3: 29Si NMR spectra of MS with different portlandite contents after a reaction time of 96
h (MS0-MS10) and 144 h (MS20 and MS40) in 1M NaOH (plain lines) and KOH (dotted
lines). “MS (original)” refers to the micro silica not exposed to alkaline solutions.
Figure 4: Amount of unreacted silica (Q4-sites) and reaction products (Q1,Q2,Q3-sites) as a
function of the portlandite content.
0 1020 30 40
Q4 (unreacted MS)
MS10 in 1M NaOH
Effective saturation index
Fig. 5: Saturation indices of MS10 (10 mass-% portlandite) in 1M NaOH calculated from the
composition of the pore solution (see Table VI).
Fig. 6: Content of chemically bound water in the reaction products as a function of their
average molar Ca/Si-ratio.
Physically bound water
Fig. 7: Content of physically bound water in the reaction products at different RH as a
function of their average molar Ca/Si-ratio.
0.1 0.30.5 0.7
Difference in physically bound
97 to 94% RH
94 to 86% RH
Fig. 8: Change of physically bound water content in the reaction products between different
RH as a function of their average molar Ca/Si-ratio.
Table I: Composition of MS (L.O.I. = loss on ignition)
SiO2 Al2O3 Fe2O3 CaO MgO
92.8 0.94 0.13 0.36 0.74
Table II: Portlandite content of the MS samples and their molar Ca/Si and (Na+K)/Si ratios.
Samples MS0 MS2.5 MS5 MS10 MS20 MS40
Portlandite content [mass-%] 0 2.5 5
Total molar Ca/Si-ratio 0.00 0.02 0.05
Total molar (Na+K)/Si-ratio 0.129 0.132 0.136 0.143
Table III: Total heat flow determined with calorimetry and water loss at 980 °C determined
with TGA of MS with different portlandite contents reacted in NaOH or KOH for 144 hours
(n.a. = not analyzed).
Sample MS0 MS2.5 MS5 MS10 MS20
Cumulative heat per solid [J/g]
1M NaOH 78 86 131 152 186
1M KOH 86 n.a. 148 157 195
Water loss per anhydrous solid at 980°C [mass-%]
1M NaOH 7.9 10.3 10.7 12.4 16.3
1M KOH 7.4 n.a. 10.0 10.2 14.8
Table IV: Total water content of MS with different amounts of portlandite reacted for 144
hours in 1M alkaline solution.
MS2.5 MS5 MS10
Humidity change RH [%] 1M NaOH
Total water per anhydrous solid [mass-%]
97 50 72 104
Desorption 94 26 34 49
86 22 23 26
MS20 MS40 MS10
Table V: Amount of Q4 (not reacted silica), Q3 and the sum of (Q1,Q2) in MS with different Download full-text
portlandite contents reacted in NaOH and KOH. Data evaluated by simulation of Q1-Q4-sites
of 29Si NMR spectra.
[mol-%] [mol-%] [mol-%]
None MS (original) 100 0
MS0 83 11
MS2.5 78 11
MS5 77 11
MS10 67 14
MS20 52 15
MS40 6 5
MS10 63 14
MS40 20 9
Table VI: Pore solution composition of MS10 exposed to different alkaline solutions. Na, K,
Ca and Si refer to total concentrations; OH- to free concentrations. The lower OH-
concentration in the 1M NaOH than in the 0.85 M KOH solution is consistent with the higher
fraction of NaOH0 complex formation, which lowers the amount of free OH- (n.a. = not
1M NaOH 0.85M KOH
Time Na K Ca Si OH- Na K
0.4 4 454 9 811
20 min 955
0.1 3 438 9 798
0.4 3 438 10 803
0.4 3 438 10 749
0.1 45 393 8 585
0.4 418 172 9 586
0.6 910 7 10 369
0.6 921 3 8 264
270 0.30 813
Si OH- Na Si OH-