Cement and Concrete Research 121 (2019) 58–71

Contents lists available at ScienceDirect

Cement and Concrete Research

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

Synthesis, characterization, and water uptake property of alkali-silica T

reaction products
Zhenguo Shia, , Guoqing Gengb, Andreas Leemanna, Barbara Lothenbacha,c
⁎

a
Laboratory for Concrete & Construction Chemistry, Swiss Federal Laboratories for Materials Science and Technology (Empa), 8600 Dübendorf, Switzerland
b
Laboratory for Waste Management, Paul Scherrer Institut (PSI), 5232 Villigen, Switzerland
c
Department of Structural Engineering, Norwegian University of Science and Technology (NTNU), 7491 Trondheim, Norway

ARTICLE INFO ABSTRACT

Keywords: The reaction mechanism of alkali-silica reaction (ASR) is poorly understood due to the difficulties to directly
Alkali-silica reaction characterize ASR products in concrete. In this study, ASR products with initial Ca/Si of 0.25 and (K + Na)/Si
Shlykovite ratio 0.5 with different K/Na ratios are synthesized at 80 °C. The synthesized ASR products are characterized by
C-S-H different techniques, also the solution chemistry is analyzed and saturation indices are calculated. The results
pH
show that crystalline and nano-crystalline phases are formed in the presence of both alkalis and calcium. No ASR
Saturation index
product is present in the absence of calcium. All synthesized crystalline ASR products highly resembles the
crystal structure of shlykovite, indicating that a substitution of K in shlykovite by Na can occur. Its silicate sheet
structure has strong similarity to the ASR products formed in concrete according to Raman spectra, while some
differences are observed in their morphologies and basal peak of the XRD patterns.

1. Introduction products are usually formed in tiny amounts with very small size (μm
range) co-existing with other silica-rich phases in concrete [11]. This
Alkali-silica reaction (ASR) causes expansion and cracking of con- makes it difficult to characterize them using conventional techniques
crete due to the reaction of reactive silica from aggregates with hy- such as solid-state 29Si magic-angle spinning Nuclear Magnetic Re-
droxide ions from the pore solution of concrete in presence of alkalis sonance (29Si MAS NMR), thermogravimetric analysis (TGA) or X-ray
(Na+ and/or K+) and calcium. Since ASR was first described by Stanton diffraction analysis (XRD). The bulk characterizations of the ASR pro-
in 1940s [1], a large number of concrete structures such as bridges, ducts using these techniques as reported in literature are mainly based
dams and pavements have been affected by ASR leading to substantial on reaction products extruded or present in air voids [14,15]. However,
costs for repair and replacement. Over the past 80 years, ASR has been the structure of these ASR products can be altered by the uptake of
extensively investigated [2–7]. Scanning Electron Microscopy coupled additional Ca from the surrounding cement paste leading to the for-
with Energy Dispersive X-ray Spectroscopy (SEM/EDX) has shown that mation of a low Ca/Si ratio calcium-silicate-hydrate (C-S-H) as shown
ASR products formed in aggregates predominantly consists of silicon, in [8]. As these products are not the same as the ones present in the
alkalis and calcium with characteristic atomic ratios of (Na + K)/ aggregates, this situation may lead to misunderstanding about the ASR
Si ≈ 0.25 and Ca/Si ≈ 0.25 [8–11]. Measures to mitigate ASR expan- products formed in concrete aggregates where the stress leading to
sion including the use of supplementary cementitious materials [3,4] or concrete damage is generated. Several studies have also described the
lithium salts [12,13] have been successfully developed. However, much crystal structure of ASR products based on kanemite [16], okenite [17]
of the knowledge obtained so far about ASR and possible mitigation mountainite and rhodesite [18]. Recent advances in synchrotron-based
strategies is based on empirical approaches, as the mineralogy and micro XRD allowed to directly study the ASR products with a size of
structure of ASR products, and the precise mechanisms of the reaction 15–40 μm formed within aggregates [19]. The results revealed that the
remain poorly understood. Therefore, predictions of the occurrence and ASR product consisted of a sheet silicate which showed some similarity
kinetics of the ASR and the resulting expansion of concrete are still to mountainite and rhodesite. Crystalline and amorphous ASR products
difficult. could also be directly analyzed within the aggregate using micro-
The main challenge limiting the understanding of the fundamentals Raman spectroscopy [20], which showed that the silicate species of ASR
of ASR is the insufficient characterization of ASR products, as ASR products are dominated by Q3 sites indicating again layers of silica

⁎
Corresponding author.
E-mail address: [email protected] (Z. Shi).

https://doi.org/10.1016/j.cemconres.2019.04.009
Received 25 January 2019; Received in revised form 27 March 2019; Accepted 16 April 2019
Available online 28 April 2019
0008-8846/ © 2019 Elsevier Ltd. All rights reserved.
Z. Shi, et al. Cement and Concrete Research 121 (2019) 58–71

tetrahedra sharing three bridging oxygens. However, more work needs (a)

SiO
8.
8.0K-gel
~ 9.0
to be conducted along these tracks to provide full structure information 7.
7.0Na-gel
~ 8.0 0.00

2
of the ASR products. 6.
6.0C-S-H_Na
~ 7.0
1.00
An alternative approach which allows the characterization of the 5.
5.0SKC
~ 6.0or SKC#

ASR products using wider range of techniques is to synthesize larger 4.

4.0SK 0.43N0.07C
~ 5.0

amount of ASR products in laboratory. This approach has been at- 3.

3.0SK 0.35N0.15C
~ 4.0
0.25 Na-gel 7
tempted in several studies [21–24]. Overall, the synthesized products 2.
2.0SK 0.25N0.25C
~ 3.0 K-gel 8 0.75
from these studies also show presence of Q3 polymerization, which 1.0SNC
1. ~ 2.0 6
suggests the formation of layered silicate sheet structure of ASR pro-

Na
H_
1-5
ducts, consistent with the observation from synchrotron-based micro

S-
ASR products

C-
0.50
XRD and Raman spectroscopy on field samples [19,20]. However, these 0.50

5
synthesized ASR products contained usually mainly C-S-H or unreacted

0.

Ca
i=
silica as a result of the following factors: (i) low reactivity of the silica

/S
/S

i=
a)
used, (ii) overestimation of the amount of Ca(OH)2 for the starting

+N

0.
(K

25
mixtures with the idea to assure a constant reservoir of Ca as in actual 0.75
hydrated cement or concrete, and (iii) low reaction temperature and
0.25

short reaction period. Nevertheless, the observation of layered silicate

species within the synthesized reaction products suggests that it is
possible to synthesize ASR products under laboratory conditions. 1.00
In this study, syntheses of different ASR products are attempted by 0.00

O
mixing freshly burnt CaO and highly reactive amorphous silica with 0.00 0.25 0.50 0.75 1.00 M O

Ca
2
alkali solution at target Ca/Si and (Na + K)/Si ratios similar to those
observed under field conditions and with varying K/Na ratios at 80 °C. (b)

K2
8.
8.0K-gel
~ 9.0

O
The obtained products are characterized by several different techniques 7.
7.0Na-gel
~ 8.0 0.00
1.00
such as XRD, 29Si MAS NMR, Raman spectra, Fourier Transformation-
6.
6.0C-S-H_Na
~ 7.0
5.
5.0SKC
~ 6.0 or SKC# 8
Infrared (FTIR) spectra, thermogravimetric analysis (TGA/DTA), dy- 4.
4.0SK
K-gel
0.43N0.07C
~ 5.0
namic vapor sorption (DVS), and SEM. The chemical compositions of 3.
3.0SK 0.35N0.15C
~ 4.0
the solutions are also analyzed using ionic chromatography (IC). The 2.
2.0SK 0.25N0.25C
~ 3.0 0.25

6.1
saturation index of the solution with respect to amorphous silica and C- 0.75

a=
1.0SNC
1. ~ 2.0
S-H are calculated based on the measured concentrations of the solu-

N
K/

3
tions. The information obtained for the synthesized ASR products are

2.
=
expected to add fundamental knowledge to the structures and potential

a
N
K/
0.50
formation mechanism of the ASR products formed in concrete and to 5 0.50
=1
develop thermodynamic data for the ASR products. 4 Na
K/
AS

3
2. Materials and methods
Rp

0.75
rod

2 0.25
2.1. Sample preparation
2
uc
ts
(1-

The alkali-silica reaction products were synthesized by mixing

5)

amorphous silica (hydrophilic silica (SiO2), surface area 200 m2/g, from 1.00 C-S-H_Na
6 1 Na-gel 7
0.00
EVONIK industries) with freshly burnt CaO (obtained by burning cal-
O

0.00 0.25 0.50 0.75 1.00 Na O

Ca

cium carbonate for 12 h at 1000 °C) and NaOH and/or KOH pellet at
2

designated portions. Samples with different K/Na ratios (0, 1.0, 2.3 and Fig. 1. Bulk chemical compositions of the starting materials (empty circle) with
6.1) and Na/K ratio of 0 were prepared with a constant (K + Na)/Si the chemical composition of each sample projected in both (a) CaO-SiO2-M2O
ratio of 0.5 and Ca/Si ratio of 0.25 as shown in Table 1, which is also and (b) CaO -K2O-Na2O ternary diagrams.
illustrated in Fig. 1 with the initial chemical compositions projected in

Table 1
Starting materials and mix proportions for the synthesis of ASR products and reference samples.
No. I.d. SiO2 CaO NaOH KOH H2O w/s Ca/Si (Na + K)/Si K/Na Ca/(Na + K) [Na+ + K+]

g g g g mL g/g Molar ratio mmol/L

1 SNC 6.008 1.402 2.000 – 50 5.31 0.25 0.5 0 0.5 1000

2 SK0.25N0.25C 6.008 1.402 1.000 1.403 50 5.10 0.25 0.5 1.0 0.5 1000
3 SK0.35N0.15C 6.008 1.402 0.600 1.964 50 5.01 0.25 0.5 2.3 0.5 1000
4 SK0.43N0.07C 6.008 1.402 0.280 2.413 50 4.95 0.25 0.5 6.1 0.5 1000
5a SKC or SKC# 6.008 1.402 – 2.805 50 4.89 0.25 0.5 – 0.5 1000
6 C-S-H_Na 4.000 1.120 0.666 – 100 17.3 0.3 0.25 0 1.2 170
7 Na-gel (silica gel) 4.000 – 1.332 – 60 11.3 0 0.5 0 0 560
8 K-gel (silica gel) 4.000 – – 1.868 60 10.2 0 0.5 – 0 330

a
SKC# is a replicate synthesis of SKC, the difference is that SKC# was shaken weekly during synthesis, while SKC was not shaken.

59
Z. Shi, et al. Cement and Concrete Research 121 (2019) 58–71

both CaO-M2O-SiO2 and CaO-Na2O-K2O ternary diagrams. To distin- frequency range of 110–1560 cm−1 was 3–5 cm−1. About 10–20
guish the K/Na ratios, these samples are noted as SNC, SK0.25N0.25C, spectra were collected per sample. The spectra were background cor-
SK0.35N0.15C, SK0.43N0.07C and SKC. SKC# is a replicate synthesis of rected and normalized. FTIR spectra were recorded on a Bruker Tensor
SKC, the difference is that SKC# was shaken weekly during synthesis. 27 FT-IR spectrometer by transmittance between 600 and 4000 cm−1
Different K/Na ratios were chosen since concrete mixtures produced with a resolution of 4 cm−1 on ~3 mg of powder.
with cements having a similar Na2O-equivalent but different K/Na-ra- Thermogravimetric analysis (TGA/DTA) and dynamic vapor sorp-
tios can have considerably different expansion [25]. A reference tion (DVS) were applied to study the water behavior of the studied
sample, C-S-H_Na, with relatively higher Ca/Si of 0.3 and lower Na/Si samples. TGA/DTA was done with a Mettler Toledo TGA/SDTA 8513 on
ratio of 0.25 was prepared. Two additional reference samples (labeled 10 to 30 mg samples using a heating rate of 20 °C/min from 30 to
Na-gel and K-gel) were prepared in the absence of Ca with Na(K)/Si 980 °C. DVS was performed on approximately 0.3 g samples. At the
ratio of 0.5. The water/solid (w/s) mass ratio was around 5 for synth- beginning, the sample was equilibrated at RH of 35%. Afterwards, the
esis of the ASR products. The slight adjustment of the water content was sample was subjected to a stepwise absorption and then desorption
to ensure that these samples having the same initial alkali (K + Na) process. Each RH step was maintained until moisture equilibrium was
concentration of 1 mol/L. Higher w/s ratios (i.e., 17.3, 10.2, 11.3) were reached with a mass change of 0.002% per 5 min.
applied to the three reference samples to lower the initial solution al- Secondary electron images of the carbon coated samples were col-
kalinity. All samples were prepared in Teflon bottles in a N2 filled glove lected using a scanning electron microscope (SEM) FEI Quanta 650 with
box to minimize CO2 contamination, and then placed in incubator at an Everhart Thornley Detector. An acceleration voltage of 2–10 kV, a
80 °C for reaction for 80 days. Afterwards, roughly 5 mL solution of spot size 2 or 3 and a pressure of 3.0–5.0 × 10−6 Torr were used.
each sample was obtained by filtration of the sample with 20 μm paper
filter. The obtained solution for each sample was then filtered with 2.3. Solution analysis and chemical composition of the obtained solids
0.45 μm syringe filter for pH measurement and solution composition
analysis. After filtration, the solids were rinsed first with approximately The pH measurements were carried out immediately after filtration
50 mL of 1:1 water-ethanol solution and then with 50 mL 94% ethanol using small fraction of the filtrated solution at room temperature
solution in the N2 filled glove box. The solids were vacuum dried for around 23 °C with a Knick pH meter (pH-Meter 766) equipped with a
seven days, and then stored in N2 filled desiccators with CO2 absorbent Knick SE100 electrode. Another part of filtrated solution was diluted in
to prevent carbonation. ratios of 1:10, 1:100 and 1:1000 with MilliQ water immediately after
The field samples were obtained from an ASR affected bridge built filtration and used for ionic chromatography (IC) analysis. The che-
in 1969. More information on the structure, the concrete and the cores mical composition of the obtained solids is calculated by mass balance
taken is given in [10]. Only ASR products present within aggregates based on the chemical composition of the starting materials and the
were used for analysis. chemical composition of the solution at equilibrium, assuming no sig-
nificant change of volume of the aqueous phase throughout the reac-
2.2. Solid phase analysis tion.

The phases in the solids were determined by a conventional X-ray 2.4. Calculation of saturation indices
powder diffraction (XRD, PANalytical X'pert Pro) with CoKα radiation
in a θ–θ configuration. The samples were scanned with a step size of Saturation indices with respect to amorphous silica and C-S-H were
0.017° 2θ between 5 and 90° 2θ with the X'Celerator detector during calculated based on the measured compositions of the solutions using
150 min. In addition to the conventional XRD measurement, the XRD of the Gibbs free energy minimization program, GEMS v3.3 [27,28] to-
SKC# was also measured in beamline 6-ID-D in Advanced Photon gether with the PSI/Nagra general thermodynamic database [29] and
Source to further analyze the structural similarity and difference be- the Cemdata18 database [30] using the CSHQ model for C-S-H [31].
tween the synthesized crystalline ASR product with its natural analog.
The SKC# powder was packed in a 1.5 mm Kapton capillary tube and 3. Results and discussion
placed between the upstream and an image plate detector. The incident
beam energy is ~100 keV. The sample position was calibrated via 3.1. Chemical compositions of the solids
measuring a CeO2 standard, which was later used to calibrate the in-
strument line broadening effect. The data was analyzed using the The bulk chemical compositions of the solids obtained after filtra-
MAUD package [26]. Similar structural analysis was also performed on tion are calculated from mass balance and projected in both CaO-SiO2-
the SNC sample however based on conventional XRD data. The sample M2O and CaO-K2O-Na2O ternary diagrams as shown in Table 2 and in
geometries for the different XRD experiments were taken into account Fig. 2 together with the projection of the starting material proportions
in the MAUD package, so that the angle-dependences of the diffraction
intensity were properly accounted during the refinement. Table 2
The 29Si MAS NMR experiments were recorded on a Bruker Avance Mass balance calculated chemical compositions of the solids based on the
III NMR spectrometer using a 7 mm CP/MAS probe at 79.5 MHz ap- measured dissolved concentrations in the solutions of the mixtures in equili-
plying the following parameters: 4500 Hz sample rotation rate, brium.
minimum of 10,240 scans or more, 30° 1H pulse of 2.5 μs, 20 s relaxa- Ca/Si Na/Si K/Si Total H2O/Si at 35% RH
tion delays, RF field strength of 33.3 kHz during SPINAL64 proton de-
coupling. The 29Si chemical shifts NMR spectra were externally refer- Solid Solid Solid mol/mol

enced to tetramethylsilane (TMS) at 0.0 ppm. 1 SNC 0.29 0.35 – 0.95

Raman spectra and Fourier Transformation-Infrared (FTIR) spectra 2 SK0.25N0.25C 0.32 0.15 0.23 0.73
were recorded on the powdered samples. In case of the bridge, Raman 3 SK0.35N0.15C 0.32 0.10 0.29 –
microscopy was conducted on grinded concrete samples as described in 4 SK0.43N0.07C 0.32 0.05 0.34 0.63
5 SKC 0.30 – 0.41 0.78 ± 0.08 for SKC# (0.87 from
more detail in [20]. Raman spectra were measured with a Bruker
[32])
Senterra instrument equipped with a Peltier-cooled CCD detector and 6 C-S-H_Na 0.35 0.15 – –
operated with the software Opus 6.5. The wave length of the used laser 7 Na-gel (silica gel) – 0.06 – –
was 532 nm operated at 20 mW. Length magnification was 50×. The 8 K-gel (silica gel) – – 0.28 –
spectral acquisition time was 20 s and spectrum resolution in the used

60
Z. Shi, et al. Cement and Concrete Research 121 (2019) 58–71

reference samples (K-gel and Na-gel) in the absence of Ca, the major

SiO
(a)
phase identified is amorphous silica as reflected by the hump observed
ASR in concrete
0.00

2
8.
8.0K-gel
at 26° 2θ. A minor amount of a crystalline phase is observed in the Na-
~ 9.0
1.00
7.
7.0Na-gel
gel sample, which is attributed to the formation of natrite (Na2CO3,
~ 8.0
7 Na-gel
6.
6.0C-S-H_Na
~ 7.0
5.
5.0SKC
~ 6.0or SKC# 8
K-gel PDF# 98-006-8104). No crystalline alkali-silicate hydrates such as e.g.
4.
4.0SK 0.43N0.07C
~ 5.0 7
kanemite (NaHSiO2O5·3(H2O)) are formed. The absence of ASR pro-
0.25
3.
3.0SK 8 0.75 ducts in the absence of Ca underlines that Ca is necessary to form ASR
0.35N0.15C
~ 4.0
2.
2.0SK 6 6 products in agreement with observations in synthetic systems [22] and
0.25N0.25C-1
~ 3.0
1.0SNC
1. ~ 2.0
1-5 with the presence of some calcium observed in ASR products formed in
C-S-H_Na field concrete [11,19,20]. In contrast, a complete reaction of amor-
0.50 ASR products (1-5) phous silica is observed for the samples containing Ca as seen by the
absence of reflection at 26° 2θ from the XRD patterns. In sample C-S-
0.50

H_Na the formation of C-S-H gel is observed at relatively high Ca/Si

ratio and low Na/Si ratio.
0.75 At initial Ca/Si ratio of 0.25 and (K + Na)/Si of 0.5, the formation
0.25 of reaction products with high crystallinity and basal spacing of 13.1 Å
is observed for the SKC samples with their XRD pattern matching well
with the recently investigated natural mineral shlykovite, KCa
[Si4O9(OH)]·3H2O [32], which has a layered silicate structure with its
1.00
SiO44− tetrahedron charge balanced by K+ and Ca2+ in the main layer
0.00
and by H+ in the interlayer [33] as shown in Fig. 4. Moreover, the XRD
O

0.00 0.25 0.50 0.75 1.00 M O

Ca

2
pattern of the SNC sample shows a similar basal spacing of 13.1 Å (7.7°
2θ, Fig. 3). The high similarity of the XRD patterns of SKC and SNC
K2

ASR in concrete (b)

samples suggests that K+ ions in shlykovite may be fully exchanged by
O

8. K-gel 0.00
Na+ ions without altering the basic structure, which will be discussed
7.1 ~ 8.0
1.00
7. Na-gel
6.3 ~ 7.1
6. C-S-H_Na
5.4 ~ 6.3
88 in the following section. Hereafter, we denote the lab-synthesized
5. SKC
4.5 ~ 5.4 or SKC#
K-gel shlykovite-like structure as K-shlykovite and Na-shlykovite when the
4. SK
3.6 0.43N0.07C
~ 4.5 alkali cations are merely K and Na, respectively. K-shlykovite is also
0.25
3. SK
2.8 0.35N0.15C
~ 3.6 0.75 observed in the sample SK0.43N0.07C where K+ is partially replaced by
2. SK
1.9 0.25N0.25C
~ 2.8 Na+ at a K/Na ratio of 6.1.
1.0 ~ 1.9
1. SNC At low K/Na ratios of 1.0 and 2.3, another not yet identified phase,
hereafter denoted as ASR-P1, is consistently observed with two major
0.50 broad reflections observed at 7° and 33° 2θ, respectively. Comparison of
0.50
these reflections with those of C-S-H formed in the C-S-H_Na sample at
5

5
4
relatively higher Ca/Si ratio of 0.3 and lower Na/Si ratio of 0.25 in-
4 3 dicates that the ASR-P1 phase is different from the C-S-H phase. In
0.75 3 addition, formation of the ASR-P1 phase is also observed in SKC and
AS

2 0.25 SK0.43N0.07C samples, however ASR-P1 is not present in sample SNC in

Rp

2
the absence of K. It is worth to mention that synthesis of SKC is repeated
rod
uc

with weekly shaking during the synthesis (noted as SKC#).

ts

Interestingly, K-shlykovite is obtained without formation of ASR-P1

(1-

C-S-H_Na Na-gel
1.00
5)

7
phase as seen from the XRD pattern for the sample SKC#.
6 6 1 1 7
0.00
O

0.00 0.25 0.50 0.75 1.00 Na O

Ca

3.3. Structural characterization of the synthesized reaction products

Fig. 2. Bulk chemical compositions of the starting materials (empty circle) and
the solids obtained after 80 days of reaction at 80 °C (filled circle). The same
chemical composition of each sample is projected in (a) CaO-SiO2-M2O and (b) 3.3.1. Structure refinements
CaO-K2O-Na2O and ternary diagrams. The three star symbol correspond to the The XRD of SKC# measured with synchrotron X-ray and SNC
chemical composition of the ASR products from concrete structure [10]. measured with conventional X-ray are used to refine the crystal-
lographic parameters based on modified shlykovite structure. The Ca/Si
and alkali/Si in the natural shlykovite are both 0.25, whereas in the
used for the synthesis. The results show a shift of the alkali contents to synthesized shlykovites they are in the range of 0.29–0.32 and
lower values as compared to the starting alkali contents, while an in- 0.35–0.41, respectively. The significantly increased alkali/Si ratio may
crease of Ca content in the solids is observed for all samples containing indicate the uptake of K(Na) into the interlayer region of shlykovite, as
CaO in the starting materials. The lowering of (Na + K)/Si ratios is observed in the natural mineral cryptophyllite [33]. Therefore, mod-
expected as the starting ratios except for sample C-S-H_Na are two times ified shlykovite structures of different K(Na)/Si ratio are evaluated in
higher than the target ratio of 0.25 for the ASR products. The chemical the refinement of the XRD patterns for the SKC# and SNC samples as
compositions of the ASR products formed in concrete [10] are also shown in Fig. 5. Initially, two K atoms are placed at fractal coordinates
plotted in Fig. 2 for comparison. Overall, the Ca/Si and (Na + K)/Si (0.5 0 0) and (0.5 0.5 0.5) in one unit cell of shlykovite. To investigate
ratios of the synthesized ASR products are similar to those of ASR the influence of interlayer K(Na) to the diffraction behavior, their oc-
products formed in concrete in particular for samples with high K/Na cupancies are fixed to be 0, 0.5, 1.0 and 1.5, resulting in a interlayer K
ratios. (Na)/Si ratio increasing from 0 to 3/16. The lattice constants, aniso-
tropic crystallite size and the x- and y-positions of the added interlayer
3.2. Phase identification by conventional XRD K are refined. As shown in Fig. 5a, with increasing interlayer K/Si ratio,
the calculated intensity of (002) and (004) diffraction significantly
The powder X-ray diffraction (XRD) patterns for the solids obtained decreases and diverges from the experimental data, while the (102),
after 80 days of reaction at 80 °C are shown in Fig. 3. For the two (106) and peaks near 0.336 Å−1 exhibit increasing intensity from

61
Z. Shi, et al. Cement and Concrete Research 121 (2019) 58–71

phase identified

SNC
Na-shlykovite
SK0.25N0.25C
ASR-P1 ASR-P1
ASR-P1 SK0.35N0.15C
ASR-P1
SK0.43N0.07C
K-shlykovite + ASR-P1
SKC
K-shlykovite + ASR-P1
SKC#
K-shlykovite
C-S-H_Na
Na-gel C-S-H C-S-H
K-gel Silica gel
Silica gel

5 10 15 20 25 30 35 40
Position [°2theta] (Cobalt(CoKα))

0.05 0.10 0.15 0.20 0.25 0.30 0.35

1/d (Å-1)
Fig. 3. XRD diffractograms of the solids obtained after 80 days of reaction at 80 °C. C-S-H: calcium-silicate-hydrate; ASR-P1: an new phase; K-gel and Na-gel: silica
gels in the absence of Ca.

Fig. 4. A schematic structure of shlykovite from literature [33], (a) viewed along [100], (b) the potassium silicate layer viewed along [001].

62
Z. Shi, et al. Cement and Concrete Research 121 (2019) 58–71

Fig. 5. The XRD patterns of (a) SKC# measured with syn-

chrotron X-ray and (b) SNC measured with conventional X-
ray, which are refined with modified shlykovite structures
that have various K or Na content in the interlayer.
Experimental data is plotted in red dashed line together with
each fitting result plotted in black solid line to ease compar-
ison. (For interpretation of the references to colour in this
figure legend, the reader is referred to the web version of this
article.)

calculation. Via refining the occupancy of interlayer K, the best fit was satisfactory fitting to the experimental data. The best fit is obtained
obtained when the occupancy of interlayer K is 0.45, corresponding to when the interlayer Na occupancy is 0, corresponding to a Na/Si of
an interlayer K/Si ratio of 0.9/16, and an overall K/Si ratio of 0.31. The 0.25. The lattice constants were refined to be a = 6.605(2),
corresponding lattice constants were refined to be a = 6.485(9), b = 7.012(1), c = 26.516(4) and β = 94.0°. The Na-shlykovite exhibits
b = 6.990(4), c = 26.642(7) and β = 94.7°, which are very close to a clear increase in lattice parameter a, compared with SKC#.
those of the natural shlykovite. However, using the natural shlykovite
structure (P21/c), two clear diffractions (between 0.242 Å−1 and 3.3.2. 29Si MAS NMR
0.251 Å−1) are predicted but not observed in the experimental data. In The 29Si MAS NMR spectra of the studied samples are shown in
contrast to natural shlykovite, the crystallite size of the synthesized K- Fig. 6. For the C-S-H_Na sample, the spectrum contains resonances
shlykovite was refined to be 1720 nm along a-axis, 330 nm along b-axis centered at −82 ppm (shoulder) and −85 ppm corresponding to brid-
and 600 nm along c-axis, which indicates that synthesized K-shlykovite ging silica tetrahedral (Q2b) and pair silica tetrahedral (Q2p), which is
has larger crystal size along one direction than the other two directions. similar to those reported in literature for C-S-H [21,34]. A broad peak
Similar strategy was also applied in refining the SNC data by re- centered at −111 ppm related to unreacted amorphous silica (Q4) has
placing the K with Na in the shlykovite structure. Note that we did not also been observed. A shoulder at −94 ppm as well as the signals in the
intend to refine the line broadening of diffraction peaks since the in- range of −99 to −103 ppm are assigned to Q3 (Si-OH) sites in hydrated
strument conditions of the conventional XRD were not calibrated. As amorphous silica [34]. Signals in the range of −99 to −103 ppm had
shown in Fig. 5b, the merge of peak (004) and (100) in Na-shlykovite also been reported in systems where amorphous siliceous aggregates
(0.151 Å−1) is readily obtained by refining the lattice constants. Using reacted with alkali solution in the absence of calcium [35].
the natural shlykovite structure (P21/c), the (002) and (102) peaks are The 29Si MAS NMR spectrum of the “pure” K-shlykovite formed in
hardly fitted. Several diffractions are predicted but not observed in the the sample SKC# is dominated by resonances at −90, −92 and
experimental data. We then reduced the symmetry from P21/c to Pc, −94 ppm corresponding to Q3 sites from shlykovite. The peak assign-
and added interlayer Na, followed by a refinement of its x- and y-po- ments are supported by the molecular structure of shlykovite reported
sitions. The Pc symmetry clearly improves the fitting, although some in [33] as shown in Fig. 4, where only four equal numbers of Q3 sites
predicted peaks are still not observed in the experimental data. In- are present with two of them sharing similar chemical environment
creased Na in the interlayer seems to decrease the intensity of (002) and resulting in the major peak observed at −94 ppm. Minor peaks at −85,
(004) & (100) diffraction intensity from calculation, resulting in a less −96, and −109 ppm are also observed in the SKC sample corre-
sponding to minor impurities of Q2, Si-OH group in hydrated amor-
phous silica (Q3) and a trace of well-ordered silica (Q4). In contrast to
the SKC# sample, the 29Si MAS NMR spectrum of SNC sample shows
Na-shlykovite

SNC
broad line shape, which can be explained by the lower symmetry of
SK0.25N0.25C silicate structure as indicated by XRD refinement showing that the a-
axis of Na-shlykovite is stretched. Overall, the NMR peaks are also
SK0.35N0.15C dominated by Q3 sites centered at −94.5 ppm with a shoulder at
ASR-P1

−96 ppm, which overlaps with part of the spectrum of the SKC#
K-shlykovite
K-shlykovite

sample.
SK0.43N0.07C The SK0.25N0.25C and SK0.35N0.15C samples with formation of the
ASR-P1 observed from XRD are also analyzed by 29Si MAS NMR. Their
29
SKC Si NMR spectra show that the silicate species are dominated by Q3
sites at −90.4 ppm. Trace of Q2 site is observed at −84.6 ppm for these
SKC# Q3 two samples, which could also be related to the presence of ASR-P1 or
Q2p ASR products formation of minor amount of C-S-H. Formation of both K-shlykovite
and ASR-P1 phases are also observed for SK0.43N0.07C sample according
Q4
Q2b
Q3
C-S-H
to 29Si MAS NMR spectrum, which is consistent with the observation
from XRD pattern shown in Fig. 3.
C-S-H SiO2

-60 -70 -80 -90 -100 -110 -120

δ 29
Si [ppm] 3.3.3. Raman spectra
The Raman spectra of the studied samples are shown in Fig. 7,
29
Fig. 6. Si MAS NMR spectra of the solids obtained after 80 days of reaction at where intense peaks are observed in the frequency range
80 °C. 200–1200 cm−1 with peak assignments given in Table 3. Each spectrum

63
Z. Shi, et al. Cement and Concrete Research 121 (2019) 58–71

Table 4

shlykovite
Summary of the main characteristic peaks of Raman and FTIR spectra of the
studied samples.
Raman Corresponding FTIR frequency Corresponding

shlykovite
SNC frequency phases (cm−1) phases
(cm−1)

ASR-P1
607 Na-shlykovite (Q3) 629 ASR-P1
595 K-shlykovite (Q3) 652 ASR-P1

ASR-P1
639 ASR-P1 (Q3) 688 Na-shlykovite

ASR-P1
675 ASR-P1 (Q3) 931 C-S-H
SK0.25N0.25C 668 C-S-H (Q2) 958 K-shlykovite
1007 C-S-H (Q2) 967 Na-shlykovite
Q3 1063 ASR-P1 (Q3) 969 C-S-H
SK0.35N0.15C 1087 Na/K-shlykovite 990 Na-shlykovite
(Q3)
1120 ASR-P1 (Q3) 1023 Na/K-shlykovite

SK0.43N0.07C Q3

associated with symmetrical C-O stretching vibrations at 1078 cm−1 is

also observed for this sample as a result of slight carbonation of the C-S-
SKC
H phase.
shlykovite

Q2 For the SKC, SNC and SK0.43N0.07C samples where shlykovite is

present, two major well-resolved bands attributed to symmetrical
C-S-H

C-S-H
bending vibrations and stretching vibrations are observed at 594 and
C-S-H

Q2 1086 cm−1, respectively, as compiled in Table 4. Several studies have

assigned them to Q3 species of silica tetrahedra [36,38,40]. The results
200 400 600 800 1000 1200
are consistent with the NMR observations addressed above for these
Raman shift (cm-1)
samples and with the known structure of shlykovite. For the
Fig. 7. Raman spectra of the solids obtained after 80 days of reaction at 80 °C. SK0.25N0.25C and SK0.35N0.15C samples, two well-resolved peaks attrib-
uted to symmetrical bending vibrations of Si-O-Si is observed at 638
and 675 cm−1, and another two well-resolved peaks attributed to
shows peaks falling into four ranges: (i) high frequency range of Si-O symmetrical stretching vibrations of Si-O are observed at 1065 and
symmetrical stretching vibrations (800–1150 cm−1), (ii) low frequency 1120 cm−1. These bands reflect the characteristics of the structure of
range of Si-O-Si symmetrical bending vibrations (600–700 cm−1), (iii) the ASR-P1 phase, whose structure has not yet been identified. Based on
internal deformation of the Si-O tetrahedra (400–500 cm−1), and (iv) 29
Si MAS NMR spectra of these samples, it can be deduced that these
Ca-O polyhedra and lattice vibrations (200–350 cm−1) [36,37]. In ad- bands are most likely associated with Q3 species of silica tetrahedra.
dition, symmetrical C-O stretching vibrations at about 1070 cm−1 for The assignment of the high frequency bands (i.e., 1065 and 1120 cm−1)
carbonate groups [36,38] may also be identified. As seen in Fig. 7, the to the Q3 species for these samples are in line with those reported for Q3
C-S-H phase shows vibrations at 320, 430, 669, 882, 1011, 1078 cm−1, in C-S-H and alkali silicate glasses [36,38,40]. However, the assignment
similar to those observed by others [38,39]. However, the intensity of of the low frequency bands between 630 and 670 cm−1 are generally
the peak associated with Ca-O at about 320 cm−1 is relatively low as a assigned to Q2 species in C-S-H [36,38,41], which is in contrast to the
29
result of the low Ca/Si ratio of the C-S-H phase synthesized in the Si NMR data in the present study. This observation may suggest a
present study. The intensity of peaks associated with internal de- possible overlapping of Q2 and Q3 species in this range. The divesity of
formation of the Si-O tetrahedra of Q4 tetrahedral sites at 430 cm−1 and the assignment to Q2 and Q3 for the 630–670 cm−1 bands may be ex-
symmetrical stretching vibrations of Si-O tetrahedra at 882 cm−1 are plained by the fact that the frequencies of bands in this region are very
also very low indicating a substantial reaction extent of the amorphous sensitive to the Si-O-Si bond angle as reviewed in [42]. In addition to
silica. The dominant peaks associated with symmetrical bending at these major bands, the Ca-O vibrations and Si-O tetrahedra of Q4 spe-
669 cm−1 and symmetrical vibrations at 1011 cm−1 are both attributed cies are also observed for these samples at 240–350 cm−1 and
to the Q2 tetrahedral sites according to previous studies [36], which is 400–500 cm−1, respectively.
consistent with 29Si MAS NMR results presented in Fig. 6. Minor peak In addition to shlykovite, the ASR-P1 phase is also identified in the

Table 3
Peak assignments of Raman bands from published studies.
Raman frequency (cm−1) Assignment References

200–350 Vibrations involving Ca-O polyhedra [36–38,41]

400–500 Internal deformations of Si-O tetrahedra, O-Si-O bending, Q4 [36,38,41]
600 Symmetrical bending (SB) involving Q3 Si tetrahedra [36,38]
638 SB of Si-O-Si linkages for Q2 or Q3 groups [36], this study
670 SB of Si-O-Si linkages in Q2 or Q3 groups [36,38,41], this study
880 Symmetrical stretching (SS) of Q1 tetrahedra [36,38]
1010 SS of Q2 tetrahedra [36,38]
1020 SS of Q2 tetrahedra [36,38]
1065 SS of Q3 tetrahedra [36,38,40]
1078 Symmetrical C-O stretching in carbonate groups [38]
1086 SS of Q3 tetrahedra [36,40]
1115 Symmetrical stretching (SS) of Si-O tetrahedra in Q3 groups [36,53]

64
Z. Shi, et al. Cement and Concrete Research 121 (2019) 58–71

(a) SNC
ASR products
SK0.25N0.25C
C
7
SK0.35N0.15C N 0. 0
.43
SK0.43N0.07C SK 0
ASR products
SKC C
SN

3580
SKC#
#
C-S-H_Na SKC
Na-gel
C
SK OH group

3030
K-gel

Amorphous silica
loosely bound water
SiO2 Amorphous silica
C-S-H SK0.3 N C
5 0.15

K-gel
FTIR spectra SiO2 SK
0.25 N
0.25 C
Na -
600 700 800 900 1000 1100 1200 1300 ge l

-1
Wavenumbers (cm )
C-S
-H_
Na

3395
C-H(EtOH)
(b) SNC

SK0.25N0.25C

SK0.35N0.15C FTIR spectra

ASR-P1 SK0.43N0.07C
2400 2600 2800 3000 3200 3400 3600 3800
SKC
-1
SKC# Wavenumbers (cm )
K-shlykovite
Fig. 9. FTIR spectra in the range with respect to stretching vibrations of O-H
C-S-H
groups in H2O or hydroxyls with a wide range of hydrogen-bond strengths.
C-S-H_Na

Na-gel
vibrations [43]. A close examination of the broad band in the range
K-gel 1130–950 cm−1 reveals presence of several distinct Si-O-Si bonds in the
Amorphous silica
reaction products for the samples containing K, although XRD data
shows completely different crystallinity between ASR-P1 and shlykovite
(Fig. 3). For the SKC sample, the following different stretching vibra-
SiO2
2nd derivative of FTIR spectra

600 700 800 900 1000 1100 1200 1300 tions of Si-O-Si bonds are identified at 1132, 1098, 1023, 996,
-1
960 cm−1, which are identical to those observed for the natural shly-
Wavenumbers (cm ) kovite [32]. A band observed at 877 cm−1 for SK0.35N0.15C and SKC is
Fig. 8. (a) FTIR spectra and (b) second derivative of FTIR of the solids obtained
attributed to out-of-plane bending of CO32– [43]. The distinction of
after 80 days of reaction at 80 °C. shlykovite and ASR-P1 can be better visualized by the characteristic
peak of these phases in the second derivative of the FTIR curves as
shown in Fig. 8b. The results show formation of K-shlykovite in the SKC
SKC sample as seen by the bands at 638, 675, 1066 and 1120 cm−1, and SK0.43N0.07C samples, and formation of ASR-P1 phases in the
which is in line with the XRD measurements. For the SK0.25N0.25C and S0.25K0.25NC, SK0.35N0.15C and SK0.43N0.07C samples, which confirms
SK0.35N0.15C samples, mainly the ASR-P1 phase is observed, although a the observations from XRD, 29Si MAS NMR, and Raman spectroscopy.
low quantity of C-S-H might also be present. The bands associated with The features FTIR peaks associated with these phases are also sum-
all the identified phases are summarized in Table 4. marized in Table 4.
In addition to the asymmetric and symmetric stretching vibrations
3.3.4. FTIR of Si-O bonds addressed above, the stretching vibrations of O-H groups
Compared to the Raman spectra, the FTIR spectra in Fig. 8a show in H2O or hydroxyls with a wide range of hydrogen bond strength is
broader signals. The spectra indicate the presence of mainly unreacted also observed from FTIR spectra from the broad band at
amorphous silica in the K-gel and Na-gel samples by the broad band 2400–3800 cm−1 as shown in Fig. 9. The broad band within this range
centered at 1050 cm−1 and an additional band centered at 790 cm−1, for the K-gel sample is associated with the gel water or loosely bound
which shift slightly to lower wavenumbers compared to the bands of water. For the C-S-H_Na sample, two broad bands centered at 3580 and
amorphous silica at 1070 and 800 cm−1 plotted in the same figure. For 3395 cm−1 are observed. The 3580 cm−1 band can be assigned to the
the C-S-H_Na sample, the Si-O stretching vibrations of Q2 species as- stretching vibration of non- or weakly-hydrogen-bonded hydroxyl
sociated with C-S-H phase are observed at 966 cm−1, while the band groups in CaOH linkages in the main layer as also observed in other
attributed to the minor amount of unreacted amorphous silica is ob- study [43]. The 3395 cm−1 band is probably related to O-H stretching
served at 1087 cm−1. Similar characterization bands for C-S-H samples of more strongly hydrogen bonded molecular H2O in the interlayer. The
have also been reported in literature [43–45]. bands associated with the main layer water and interlayer water are
For the samples containing shlykovite and/or ASR-P1, the FT-IR also observed in SNC, SKC and SK0.43N0.07C samples as shown in Fig. 9.
spectra show bands in the range 1130–950 cm−1 originating from The assignments of these peaks are supported by the molecular struc-
asymmetric and symmetric stretching vibrations of Si-O bonds, and ture of the shlykovite, which shows the presence of both OH group and
bands in the range 700–600 cm−1 corresponding to Si-O-Si bending loosely bound water. The narrowness of the band associated with the

65
Z. Shi, et al. Cement and Concrete Research 121 (2019) 58–71

100 6.0
5.0 SNC
SK0.25N0.25C
4.0
SK0.43N0.07C
95 3.0 SKC#
C1.6SH

amorphous
H2O/Si (mol/mol)
TG (%)

8
2.0
90 n
4
ptio

crystalline
2 sor
de
5
(5) 1.5 rption
3 C-S-H abs o
85 7
1

ASR products
6 1.0
80 0.2
bound H2O
loosly bound H2O OH group 1. SNC
0.0 0.5
2. SK0.25N0.25C 30 40 50 60 70 80 90 100
3. SK0.35N0.15C Relative humidity (%)
-0.2

DTG (1/oC)
4. SK0.43N0.07C
Fig. 11. Dynamic vapor sorption results for different synthetic ASR products
5. SKC
-0.4 and C-S-H phase. Crystalline phases: Na-shlykovite in SNC; K-shlykovite in
(5). SKC#
SKC# and SK0.43N0.07C. Amorphous phase: ASR-P1 phase in SK0.25N0.25C and
6. C-S-H_Na
-0.6 SK0.43N0.07C.
7. Na-gel
8. K-gel
-0.8
for samples with low Ca/Si of 0.25 and high alkali/Si ratio of 0.5. K-
-1.0
shlykovite is formed when only K is the alkali source or at high K/Na
0 100 200 300 400 500 600 700 800 900 ratio of 6.1. Na-shlykovite is formed when only Na is the alkali source.
Poorly crystalline ASR-P1 is formed in samples containing K except for
Temperature (oC) the sample SKC# which was shaken weekly during synthesis.
Fig. 10. DTG curves of the solids obtained after 80 days of reaction at 80 °C.
3.4. Volume stability

OH groups, indicates that this types of water is well-defined ordered The most widely reported hypothesis regarding expansion of con-
sites. A small peak at 2970 cm−1 is also observed due to the trace of crete caused by ASR is swelling of the ASR products after absorbing
contamination of the samples by the ethanol used to wash the solid water. In this regard, the water uptake capacity and the volume stabi-
after filtration. In contrast, the samples containing mainly ASR-P1, lity of synthesized ASR products under moist conditions are of im-
show only a broad signal relatively weak signal in this range, indicating portance. The water vapor isotherm measured by dynamic vapor
the absence of well-ordered OH-groups. sorption (DVS) method is shown in Fig. 11. The measurement was
performed on the samples after vacuum drying in desiccator. The data
3.3.5. TGA are normalized by the weight loss determined by TGA at 600 °C, i.e.,
The results obtained from thermogravimetric analysis (TGA) on the 16.15% (SNC), 12.15% (SKC#), 10.73% (SK0.43N0.07C) and 11.9%
studied samples are shown in Fig. 10. For the three reference samples, (SK0.25N0.25C). For comparison, the result for synthesized C-S-H with
i.e., K-gel, Na-gel and C-S-H_Na, only a gradual loss of weight from Ca/Si ratio of 1.6 from a previous study [46] is plotted in the same
30 °C up to 400 °C is observed. For samples SK0.25N0.25C and figure. Overall, gradual water absorption is observed for all samples
SK0.35N0.15C with formation of only ASR-P1 phase according to XRD within the relative humidity (RH) range of 35–75%. Over this range,
results, similar weight loss curves as for silica gel and C-S-H are ob- water molecules are likely absorbed in interlayer space (in the case of C-
served, indicating that these solids contain relatively little water and S-H) as well as on external surfaces of the reaction products, as e.g. also
that this water is poorly confined. The low temperature at which the observed for C-S-H phases [47]. At high relative humidity capillary
weight losses of amorphous silica, C-S-H and ASR-P1 phases occurs is condensation is observed; this is more distinct for the amorphous ASR-
most likely related to the presence of free or loosely bound water at the P1 than for the well crystalline shlykovite. Both shlykovite and ASR-P1
surface or in the C-S-H interlayer. show less water uptake than C-S-H, where water is bound in the in-
Multi-step weight losses are observed for the other samples with terlayer space as well as in the interparticle space as gel water [48].
formation of shlykovite phases, i.e., SKC#, SNC and SK0.43N0.07C. The During desorption, a hysteresis is obtained for all samples, as typical for
weight losses up to 200 °C indicate the presence of some free or loosely C-S-H phase shown in Fig. 11 and [47] and clay minerals [49]. How-
bound water. Additional loss of water between 200 and 400 °C could be ever, the synthesized ASR products in this study have relatively lower
associated with water loss from the OH group of the reaction products. water uptake capacity and smaller hysteresis compared to the C-S-H
The total weight loss observed up to 400 °C for SKC# corresponds to phase.
0.78 ± 0.08 H2O per Si, consistent with the 0.87 H2O per Si suggested As DVS measurement only measure the water uptake within the
for K-shlykovite [32]. The water loss up to 400 °C for SNC sample solids without direct differentiation whether the water is taken up in
corresponds to 0.95 H2O per Si, indicating the presence of slightly more the interlayer or on the surface, the effect of water uptake on the
water. The observation of loosely bound water and water in OH groups structure is also investigated by XRD. The results in Fig. 12 show the
from DTG results confirm the observations from FTIR measurements XRD pattern of the K-shlykovite and Na-shlykovite before and after
(Fig. 9). drying. The results show that no increase of basal d-spacing (13.1 Å or
In summary, the solid phase analyses indicate that amorphous silica 7.7° 2θ) is observed for the synthesized crystalline ASR products, which
remains unreacted in the absence of calcium. At high initial Ca/Si ratio suggest that the uptake of water by the synthesized ASR products ob-
and low alkali/Si ratio, C-S-H is formed, while ASR products are formed served in the DVS results is likely related to the water absorption on the

66
Z. Shi, et al. Cement and Concrete Research 121 (2019) 58–71

K-shlykovite wet

dry

wet
Na-shlykovite
dry

5 10 15 20 25 30 35 40
Position [°2Theta] (Cobalt (CoKα))
Fig. 12. Effect of wet and drying conditions on the basal d-spacing of the K-shlykovite formed in sample SKC# and Na-shlykovite formed in sample SNC.

surface and interstice of the particles of shlykovite. The above ob- crystallization pressure. The high RH known to play a significant role in
servations are further confirmed by the DTG measurements where the ASR may be related to the chemical reaction itself or to the transport of
SKC# sample with formation of only K-shlykovite was exposed to 100% the alkali and Ca ions into the relative sites within the aggregates.
relatively humidity (RH) or mixed with smaller amount of water before
TGA measurement. The DTG results shown in Fig. 13 confirm that the 3.5. pH and measured concentration of the solution
shlykovite structure remains unchanged as reflected by the unaltered
amount of loosely bound water and OH groups. This phenomenon is The measured concentrations of Si, Ca, Na and K of the solutions in
different from most of the layer silicate minerals, such as clay minerals the supernatants together with calculated OH– ion concentrations from
and C-S-H, where water can also be absorbed in their interlayer space. measured pH are shown in Table 5. High silicon concentrations are
The results suggest that ASR expansion may not be caused by swelling observed for the Na-gel and K-gel samples in the absence of calcium,
of the ASR products but by alternative mechanism such as e.g. indicating that silica was partially dissolved by the alkali solution but

240 240
mixed with 100% H2O
100%H2O (a) (b)
50%H2O
mixed with 50% H2O SKC wet sample SNC
200 200 dry sample
at100%RH
100% RH
at30%RH
30% RH
TG (%)
TG (%)

160 160

120 120

80 80
0.0 0.0
OH group
-0.3 OH group -0.3
loosely bound H2O
DTG (1/oC)

DTG (1/oC)

-0.6 loosely bound H2O -0.6

-0.9 -0.9

free H2O -1.2 free H2O -1.2

-1.5 -1.5

-1.8 -1.8
0 100 200 300 400 500 600 700 800 900 0 100 200 300 400 500 600 700 800 900

Temperature (oC) Temperature (oC)

Fig. 13. Effect of drying conditions of (a) SKC sample and (b) SNC sample on the loosely bound H2O and OH group in K-shlykovite and Na-shlykovite, respectively.
The relative weight losses of the samples at different drying conditions are normalized by their relative weights at 400 °C.

67
Z. Shi, et al. Cement and Concrete Research 121 (2019) 58–71

Table 5
Measured dissolved concentrations in the solutions of the mixtures in equilibrium and calculated saturated indices.
Si Na K Ca OH− pHcal (23 °C) Saturation indices

mM mM mM mM mM SiO2 C-S-H

1 SNC 302 402 – 0.025 48.5 12.64 −2.1 −0.9

2 SK0.25N0.25C 442 269 135 0.026 7.85 11.92 −1.3 −0.9
3 SK0.35N0.15C 450 146 243 0.029 7.53 11.90 −1.3 −0.9
4 SK0.43N0.07C 417 57.5 317 0.018 10.7 11.99 −1.4 −1.0
5 SKC 315 – 303 0.012 14.3 12.12 −1.5 −1.1
6 C-S-H_Na 99.2 82.0 – 0.321 1.48 11.17 −0.6 −0.4
7 Na-gel (silica gel) 904 543 – 0.036a 1.66 11.19 −0.2 −4.7
8 K-gel (silica gel) 902 – 496 0.025a 2.2 11.30 −0.4 −4.6

a
The trace of Ca measured for these two samples is attributed to the impurity of the raw materials. Similar amount of Ca measured for samples 1–5 suggests that
all the freshly burnt CaO introduced in the beginning has reacted to form solid phases.

1000 3.6. Saturation indices

Na
K The saturation indices calculated by GEMs software for all samples
based on the measured concentrations are summarized in Table 5. As
Equilibrium concentration (mM)

g
in

800
nd

expected, the saturation indices with respect to C-S-H are negative for
bi
li
ka

the K-gel and Na-gel samples, due to the absence of Ca. However, also
al
no

the saturation indices with respect to amorphous silica are slightly

600 negative, although unreacted silica had been clearly seen by XRD in
Na-gel
Fig. 3. Similarly the saturation indices for the C-S-H_Na sample with
K-gel respect to amorphous silica and C-S-H are both slightly negative, al-
400 g SNC though presence of unreacted silica and formation of C-S-H are ob-
din
bin served from other measurements such as XRD and 29Si MAS NMR.
Na
K0.43N0.07 These apparent discrepancies are related to high silicon concentrations
and relatively high temperature of 80 °C. The thermodynamic databases
K0.25N0.25 SKC
K0.35N0.15
200 [29,30] used, i) describes only poorly the aqueous silica complexes at
a
_N ding
-H K bin
C
-S K0.35N0.15 K0.25N0.25 such high silica concentration of up to 0.9 M as observed for the Na-gel
K0.43N0.07
and K-gel samples, ii) misses high temperature parameters for poly-
0 nuclear silica species, and iii) has poorly defined temperature behavior
0 200 400 600 800 1000 of amorphous silica and C-S-H, which makes calculation of saturation
Initial concentration (mM) indices less accurate. This indicates the need for more advanced
amorphous SiO2 and C-S-H models valid at higher temperature as well
Fig. 14. Relationships between the equilibrium and initial concentration of Na as advances in the description of aqueous silica species at high con-
and K in the solutions of different mixtures. centrations and temperatures to be able to better take into account the
effect of temperature and very high Si concentrations. Independently of
these restrictions, the calculated saturation indices results show that the
no or little precipitation of new solids. When CaO is present in the saturation indices with respect to C-S-H and amorphous silica are more
starting materials, the silicon concentration at equilibrium is sub- negative for the samples with formation of ASR products, i.e., Na(K)-
stantially decreased, indicating the reaction of calcium with silicon to shlykovite and/or ASR-P1, consistent with the other measurements that
solid phases. In the case of C-S-H_Na sample where C-S-H is formed, no major C-S-H and amorphous silica are formed in these samples.
highest calcium concentration at equilibrium is observed. However, the
calcium concentrations of the other Ca-containing samples, where 3.7. Comparison with ASR products formed in concrete
shlykovite and/or ASR-P1 are formed, are low and even comparable to
those of Na-gel and K-gel samples where the measured trace of calcium The possibility to thoroughly characterize synthesized ASR products
is attributed to the impurity of the raw materials. The results indicate is expected to further our understanding of ASR products formed in
that nearly all calcium in these samples has reacted to form shlykovite field samples. Thus the synthesized ASR products are compared with
and/or ASR-P1 phase. Comparing the K and Na concentrations between ASR products formed in concrete. The SEM imagines shown in Fig. 15
the equilibrium solution (Table 5) and the starting solution (Table 1), a demonstrate the morphologies of the different synthesized ASR pro-
substantial decrease of alkali ion concentration is observed for all ducts. Nano-crystalline texture is observed for the ASR-P1 phase formed
samples. The reaction of alkalis with silica is also observed when both in the SK0.25N0.25C and SK0.43N0.07C samples. In contrast, plate-
comparing their concentrations between initial and equilibrium solu- like and fiber-like morphologies are observed for Na-shlykovite and K-
tions as shown in Fig. 14. Overall, a strong reduction of alkali con- shlykovite, respectively. The observations of nano-crystalline and
centration is observed for all Ca-containing samples, indicating the crystalline products from SEM measurements for the corresponding
uptake of alkalis in the solid phases. Linear correlations between the samples confirm the structural information obtained from the other
initial and equilibrium concentrations of K and Na are observed. The techniques presented in previous sections. The morphologies of the
slightly lower slope for K binding curve than Na binding curve suggest synthesized K-shlykovite and ASR-P1 are somewhat different to the
that K is more favorably bound in the solids than Na for the studied crystalline ASR products formed in concrete aggregates, which usually
samples. In contrast to the Ca-containing samples, negligible uptake of have plate-like morphology often forming typical rosettes independent
alkalis in solids is observed for the Na-gel and K-gel samples. These of concrete mixtures and binder types [8,11,50,51]. The platelets of Na-
observations suggest that almost all of the reacted alkalis in the Na-gel shlykovite strongly resemble the product formed in concrete.
and K-gel samples remained in solution. Some limited differences in the crystal structure is further confirmed

68
Z. Shi, et al. Cement and Concrete Research 121 (2019) 58–71

(a) (b)

Na-shlykovite ASR-P1

(c) (d)
ASR-P1

ASR product
K-shlykovite in field concrete

Fig. 15. SEM morphologies of (a) plate-like Na-shlykovite formed in the sample SNC, (b) nanocrystalline ASR-P1 phase formed in the sample SK0.25N0.25C, (c) fiber-
like K-shlykovite and nanocrystalline ASR-P1 phase formed in the sample SK0.43N0.07C, and (d) plate-like ASR product formed in the concrete from a 40 year old
bridge in Mels, Switzerland.

ASR in aggregate
from a bridge
Dähn et al 2016 (Leemann 2017)

ASR in aggregate
Mels_03 from CPT
(Leemann 2016)

SKC#
K-shlykovite

SNC
Na-shlykovite

0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40

-1
200 400 600 800 1000 1200
1/d (Å )
Raman shift (cm-1)
Fig. 16. Comparison of XRD patterns of the K-shlykovite and Na-shlykovite
formed in the SKC# and SNC samples respectively with XRD pattern obtained Fig. 17. Comparison of the Raman spectra of K-shlykovite (i.e., crystalline part
by synchrotron-based micro- X-ray diffraction (micro-XRD) for the ASR product of the sample SK0.43N0.07C), Na-shlykovite (i.e., SNC) synthesized at 80 °C, the
formed in aggregate of a 40 year old bridge in Mels, Switzerland [19] and for crystalline ASR product formed in aggregate from concrete prism test (CPT),
ASR product formed in an air void of a concrete from the same bridge (labeled where the sample had been subjected to 60 °C for 20 weeks [52], and the ASR
as Mels_03). product formed in aggregate from a 40 year old bridge in Mels, Switzerland
[20].

69
Z. Shi, et al. Cement and Concrete Research 121 (2019) 58–71

by the comparison of XRD patterns of the K-shlykovite and Na-shly- products formed in concrete uptake the water into interlayer or not
kovite formed in the SKC# and SNC samples with XRD pattern obtained at elevated RH, as these data are not available so far.
by synchrotron-based micro-X-ray diffraction (micro-XRD) for the
crystalline ASR products formed in an air void of a concrete collected Acknowledgement
from a 40 year old bridge in Mels, Switzerland, (labeled as Mels_03). As
seen in Fig. 16, strong similarities are observed at reflections between The authors would like to thank the following two funding resources
0.146 Å−1 and 0.382 Å−1, although the field sample has much broader, for this research: (a) SNF Sinergia: Alkali-silica reaction in concrete
less well defined peaks. However, a lower basal spacing is observed for (ASR), grant number CRSII5_17108; (b) European Union's Horizon
the Mels sample than for the lab-synthesized ASR products. This dif- 2020 research and innovation programme under the Marie Skłodowska-
ference is even bigger compared with another ASR product formed in Curie grant agreement number 754364 and number 701647. Luigi
an aggregate (Dähn et al. [19]) of the same bridge where sample Mels- Brunetti is acknowledged for measurement of the ion concentration of
03 was obtained as shown in Fig. 16. the solution, and Daniel Rentsch for acquiring all the 29Si MAS NMR
Despite the differences in basal spacing observed by XRD, the spectra. Frank Winnefeld is acknowledged for help and discussion with
Raman spectra in Fig. 17 show high similarity of the molecular struc- XRD measurement and results, Mateusz Wyrzykowski for help with DVS
ture in short-range order among K-shlykovite, Na-shlykovite and ASR measurement, Douglas Robinson for help with the XRD measurement in
products formed in aggregates during the concrete prism test [52], and beamline 6-ID-D at Advanced Photon Source, and Jørgen Skibsted for
in aggregate of the bridge [20]. In particular the Raman spectrum for K- helpful discussions on the 29Si MAS NMR spectra.
shlykovite is almost identical to that of ASR product from the concrete
prism test (where the samples had been subjected to 60 °C for References
20 weeks), indicating similar local environment of the Si species and
also of the Ca environment as reflected by the band at about 320 cm−1 [1] T.E. Stanton, Influence of cement and aggregate on concrete expansion, Eng. News-
observed for both K-shlykovite and the ASR product from the concrete Record. 1 (1940) 59–61.
[2] M. Thomas, B. Fournier, K. Folliard, J. Ideker, M. Shehata, Test methods for eval-
prism test. The field sample which has not been exposed to such high uating preventive measures for controlling expansion due to alkali–silica reaction in
temperatures [20], shows similar features, but the positions of the concrete, Cem. Concr. Res. 36 (2006) 1842–1856.
peaks at 600 and 1100 cm−1 is somewhat shifted. The results tenta- [3] R.N. Swamy, Role and effectiveness of mineral admixtures in relation to alkali-silica
reaction, Alkali-Silica React. Concr, CRC Press, 2002, pp. 112–137.
tively indicate that the obtained K-shlykovite and Na-shlykovite could [4] M. Thomas, The effect of supplementary cementing materials on alkali-silica reac-
in fact be a synthetic proxy for some of the ASR products observed in tion: a review, Cem. Concr. Res. 41 (2011) 1224–1231.
the field. [5] J. Lindgård, Ö. Andiç-Çakır, I. Fernandes, T.F. Rønning, M.D.A. Thomas,
Alkali–silica reactions (ASR): literature review on parameters influencing labora-
tory performance testing, Cem. Concr. Res. 42 (2012) 223–243.
4. Conclusions [6] J.W. Pan, Y.T. Feng, J.T. Wang, Q.C. Sun, C.H. Zhang, D.R.J. Owen, Modeling of
alkali-silica reaction in concrete: a review, Front. Struct. Civ. Eng. 6 (2012) 1–18.
In this study, ASR products with initial Ca/Si ratio of 0.25, [7] F. Rajabipour, E. Giannini, C. Dunant, J.H. Ideker, M.D.A. Thomas, Alkali–silica
reaction: current understanding of the reaction mechanisms and the knowledge
(K + Na)/Si ratio of 0.5 and different K/Na ratios have been synthe- gaps, Cem. Concr. Res. 76 (2015) 130–146.
sized at 80 °C. The observed products had Ca/Si ≈ 0.3 and alkali/ [8] T. Katayama, Petrographic Study of the Alkali-aggregate Reactions in Concrete,
Si ≈ 0.4. This range of composition corresponds to the ASR products Grad. Sch. Sci. Univ. Tokyo, Dep. Earth Planet. Sci, 2012.
[9] A. Leemann, P. Lura, E-modulus of the alkali–silica-reaction product determined by
formed in aggregates of ASR-affected concrete structures. Different micro-indentation, Constr. Build. Mater. 44 (2013) 221–227.
techniques were applied to characterize the reaction products. The [10] A. Leemann, C. Merz, An attempt to validate the ultra-accelerated microbar and the
solution chemistry and saturated indices were also evaluated. Based on concrete performance test with the degree of AAR-induced damage observed in
concrete structures, Cem. Concr. Res. 49 (2013) 29–37.
the results and discussion, the following conclusions can be drawn: [11] A. Leemann, T. Katayama, I. Fernandes, M.A.T.M. Broekmans, Types of alkali–ag-
gregate reactions and the products formed, Proc. Inst. Civ. Eng. Mater. 169 (2016)
(i) In addition to reactive silica, the formation of ASR products re- 128–135.
[12] C.L. Collins, J.H. Ideker, G.S. Willis, K.E. Kurtis, Examination of the effects of LiOH,
quires the presence of both alkalis and calcium. Without calcium, LiCl, and LiNO3 on alkali–silica reaction, Cem. Concr. Res. 34 (2004) 1403–1415.
the main reaction products are amorphous silica; at high Ca/Si [13] X. Feng, M.D.A. Thomas, T.W. Bremner, B.J. Balcom, K.J. Folliard, Studies on li-
ratio and low alkali/Si ratio, C-S-H phase is formed, while ASR thium salts to mitigate ASR-induced expansion in new concrete: a critical review,
Cem. Concr. Res. 35 (2005) 1789–1796.
products formed at low calcium concentrations and high alkali
[14] W.F. Cole, C.J. Lancucki, M.J. Sandy, Products formed in an aged concrete, Cem.
concentrations. Na-shlykovite is formed in presence of Ca, when Concr. Res. 11 (1981) 443–454.
Na is the only alkali. K-shlykovite is formed in presence of Ca when [15] C.J. Benmore, P.J.M. Monteiro, The structure of alkali silicate gel by total scattering
K is the only alkali or at high K/Na ratio of 6.1. A nano-crystalline methods, Cem. Concr. Res. 40 (2010) 892–897.
[16] W. Wieker, C. Hübert, D. Heidemann, R. Ebert, Alkali-aggregate reaction–a problem
product labeled ASR-P1 phase is observed in the presence of both K of the insufficient fundamental knowledge of its chemical base, in: M. Cohen,
and Na, preferably at lower K/Na ratios 1.0 and 2.3. S. Mindess, J. Skalny (Eds.), Proc. Sidney Diam. Symp. Mater. Sci. Eng. Concr. Cem.
(ii) In contrast to C-S-H, the silica structures of both shlykovite and Based Compos. Am. Ceram. Soc. Westerville, OH, 1998, pp. 395–408.
[17] W.F. Cole, C. Lancucki, Products formed in an aged concrete the occurrence of
ASR-P1 phase are dominated by Q3 species indicating a silicate okenite, Cem. Concr. Res. 13 (1983) 611–618.
sheet structure, similar to ASR products from field concrete. In [18] L. De Ceukelaire, The determination of the most common crystalline alkali-silica
particular, Raman spectra of K-shlykovite are identical to that of reaction product, Mater. Struct. 24 (1991) 169–171.
[19] R. Dähn, A. Arakcheeva, P. Schaub, P. Pattison, G. Chapuis, D. Grolimund,
ASR product formed in aggregates during the concrete prism test. E. Wieland, A. Leemann, Application of micro X-ray diffraction to investigate the
The general similarity between synthesized ASR products (shly- reaction products formed by the alkali–silica reaction in concrete structures, Cem.
kovite) and crystalline ASR products observed in field samples Concr. Res. 79 (2016) 49–56.
[20] A. Leemann, Raman microscopy of alkali-silica reaction (ASR) products formed in
suggests that the structure information obtained for the K-shlyko- concrete, Cem. Concr. Res. 102 (2017) 41–47.
vite and Na-shlykovite can be used to better understand the [21] X.-D. Cong, R.J. Kirkpatrick, S. Diamond, 29Si MAS NMR spectroscopic investiga-
structure and formation conditions of ASR products in concrete. tion of alkali silica reaction product gels, Cem. Concr. Res. 23 (1993) 811–823.
[22] A. Leemann, G. Le Saout, F. Winnefeld, D. Rentsch, B. Lothenbach, Alkali–silica
(iii) The experimental results show that the water uptake capacity and
reaction: the influence of calcium on silica dissolution and the formation of reaction
hysteresis effect during the wet-dry cycle are smaller or compar- products, J. Am. Ceram. Soc. 94 (2011) 1243–1249.
able to C-S-H phases. The XRD and DTG results suggest that the [23] X. Hou, L.J. Struble, R.J. Kirkpatrick, Formation of ASR gel and the roles of CSH and
structure of the synthesized ASR products remained unchanged portlandite, Cem. Concr. Res. 34 (2004) 1683–1696.
[24] X. Hou, R.J. Kirkpatrick, L.J. Struble, P.J.M. Monteiro, Structural investigations of
after absorption of water, suggesting that they are not swelling. alkali silicate gels, J. Am. Ceram. Soc. 88 (2005) 943–949.
However, more work needs to be done to validate whether ASR [25] A. Leemann, B. Lothenbach, The influence of potassium–sodium ratio in cement on

70
Z. Shi, et al. Cement and Concrete Research 121 (2019) 58–71

concrete expansion due to alkali-aggregate reaction, Cem. Concr. Res. 38 (2008) study. Part II: carbonated phases, J. Am. Ceram. Soc. 90 (2007) 908–917.
1162–1168. [40] G.M. Bancroft, H.W. Nesbitt, G.S. Henderson, C. O'Shaughnessy, A.C. Withers,
[26] L. Lutterotti, S. Matthies, H.R. Wenk, MAUD: a friendly Java program for material D.R. Neuville, Lorentzian dominated lineshapes and linewidths for Raman sym-
analysis using diffraction, IUCr Newsl. CPD 21 (1999). metric stretch peaks (800–1200 cm−1) in Qn (n = 1–3) species of alkali silicate
[27] D.A. Kulik, T. Wagner, S.V. Dmytrieva, G. Kosakowski, F.F. Hingerl, glasses/melts, J. Non-Cryst. Solids 484 (2018) 72–83.
K.V. Chudnenko, U.R. Berner, GEM-Selektor geochemical modeling package: re- [41] B. Lothenbach, D. Nied, G. Achiedo, A. Dauzères, Magnesium and calcium silicate
vised algorithm and GEMS3K numerical kernel for coupled simulation codes, hydrates, Cem. Concr. Res. 77 (2015) 60–68.
Comput. Geosci. 17 (2013) 1–24. [42] P. McMillan, Structural studies of silicate glasses and melts-applications and lim-
[28] T. Wagner, D.A. Kulik, F.F. Hingerl, S.V. Dmytrieva, GEM-Selektor geochemical itations of Raman spectroscopy, Am. Mineral. 69 (1984) 622–644.
modeling package: TSolMod library and data interface for multicomponent phase [43] P. Yu, R.J. Kirkpatrick, B. Poe, P.F. McMillan, X. Cong, Structure of calcium silicate
models, Can. Mineral. 50 (2012) 1173–1195. hydrate (C-S-H): near-, mid-, and far-infrared spectroscopy, J. Am. Ceram. Soc. 82
[29] T. Thoenen, W. Hummel, U. Berner, E. Curti, The PSI/Nagra Chemical (1999) 742–748.
Thermodynamic Database 12/07, PSI Report 14-04, Villigen PSI, Switzerland, [44] I. García-Lodeiro, A. Fernández-Jiménez, M.T. Blanco, A. Palomo, FTIR study of the
2014. sol–gel synthesis of cementitious gels: C–S–H and N–A–S–H, J. Sol-Gel Sci. Technol.
[30] B. Lothenbach, D.A. Kulik, T. Matschei, M. Balonis, L. Baquerizo, B. Dilnesa, 45 (2008) 63–72.
G.D. Miron, R.J. Myers, Cemdata18: a chemical thermodynamic database for hy- [45] I.G. Lodeiro, D.E. Macphee, A. Palomo, A. Fernández-Jiménez, Effect of alkalis on
drated Portland cements and alkali-activated materials, Cem. Concr. Res. 115 fresh C–S–H gels. FTIR analysis, Cem. Concr. Res. 39 (2009) 147–153.
(2019) 472–506. [46] E. L'Hôpital, B. Lothenbach, D.A. Kulik, K. Scrivener, Influence of calcium to silica
[31] D.A. Kulik, Improving the structural consistency of CSH solid solution thermo- ratio on aluminium uptake in calcium silicate hydrate, Cem. Concr. Res. 85 (2016)
dynamic models, Cem. Concr. Res. 41 (2011) 477–495. 111–121.
[32] I.V. Pekov, N.V. Zubkova, Y.E. Filinchuk, N.V. Chukanov, A.E. Zadov, [47] C. Roosz, S. Gaboreau, S. Grangeon, D. Prêt, V. Montouillout, N. Maubec, S. Ory,
D.Y. Pushcharovsky, E.R. Gobechiya, Shlykovite KCa [Si4O9(OH)]·3H2O and cryp- P. Blanc, P. Vieillard, P. Henocq, Distribution of water in synthetic calcium silicate
tophyllite K2Ca[Si4O10]·5H2O, new mineral species from the Khibiny alkaline hydrates, Langmuir 32 (2016) 6794–6805.
pluton, Kola peninsula, Russia, Geol. Ore Deposits 52 (2010) 767–777. [48] A.C.A. Muller, K.L. Scrivener, A.M. Gajewicz, P.J. McDonald, Use of bench-top NMR
[33] N.V. Zubkova, Y.E. Filinchuk, I.V. Pekov, D.Y. Pushcharovsky, E.R. Gobechiya, to measure the density, composition and desorption isotherm of C–S–H in cement
Crystal structures of shlykovite and cryptophyllite: comparative crystal chemistry of paste, Microporous Mesoporous Mater. 178 (2013) 99–103.
phyllosilicate minerals of the mountainite family, Eur. J. Mineral. 22 (2010) [49] L. Michot, I.A. Maslon, F. Thomas, F. Vandeuvre, Mechanism of adsorption and
547–555. desorption of water vapor by homoionic montmorillonites: 2. The Li+, Na+, K+,
[34] T.F. Sevelsted, J. Skibsted, Carbonation of C–S–H and C–A–S–H samples studied by Rb+ and Cs+-exchanged forms, Clay Clay Miner. 43 (1995) 324–336.
13
C, 27Al and 29Si MAS NMR spectroscopy, Cem. Concr. Res. 71 (2015) 56–65. [50] Z. Shi, C. Shi, S. Wan, Z. Zhang, Effects of alkali dosage and silicate modulus on
[35] X. Cong, R.J. Kirkpatrick, 29Si MAS NMR study of the structure of calcium silicate alkali-silica reaction in alkali-activated slag mortars, Cem. Concr. Res. 111 (2018)
hydrate, Adv. Cem. Based Mater. 3 (1996) 144–156. 104–115, https://doi.org/10.1016/j.cemconres.2018.06.005.
[36] R.J. Kirkpatrick, J.L. Yarger, P.F. McMillan, Y. Ping, X. Cong, Raman spectroscopy [51] Z. Shi, C. Shi, J. Zhang, S. Wan, Z. Zhang, Z. Ou, Alkali-silica reaction in waterglass-
of CSH, tobermorite, and jennite, Adv. Cem. Based Mater. 5 (1997) 93–99. activated slag mortars incorporating fly ash and metakaolin, Cem. Concr. Res. 108
[37] D. Müller, R. Hochleitner, K.T. Fehr, Raman spectroscopic investigations of natural (2018) 10–19.
jennite from Maroldsweisach, Bavaria, Germany, J. Raman Spectrosc. 47 (2016) [52] A. Leemann, The influence of lithium on the structure of ASR products in concrete,
602–606. 15th Int. Conf. Alkali-aggregate React, 2016.
[38] S. Ortaboy, J. Li, G. Geng, R.J. Myers, P.J.M. Monteiro, R. Maboudian, C. Carraro, [53] L. Robinet, C. Coupry, K. Eremin, C. Hall, The use of Raman spectrometry to predict
Effects of CO2 and temperature on the structure and chemistry of C–(A–) S–H in- the stability of historic glasses, J. Raman Spectrosc. 37 (2006) 789–797. An Int. J.
vestigated by Raman spectroscopy, RSC Adv. 7 (2017) 48925–48933. Orig. Work All Asp. Raman Spectrosc. Incl. High. Order Process. Also Brillouin
[39] L. Black, C. Breen, J. Yarwood, K. Garbev, P. Stemmermann, B. Gasharova, Rayleigh Scatt.
Structural features of C–S–H (I) and its carbonation in air—a Raman spectroscopic

71