The effect of equilibration time on Al uptake in C-S-H Cement and Concrete Research

Calcium silicate hydrate (C-S-H) is the main hydration product of Portland cements. The partial replacement of Portland cements by supplementary cementitious materials can result in C-S-H with lower Ca/Si ratios and more aluminum and alkalis. The effect of equilibration time on Al uptake in C-S-H was investigated using equilibration times from 7 days up to 3 years. Lower Al concentrations were measured in the solution after longer equilibration times. In addition, a higher uptake of Al in C-S-H was observed based on the decrease in the content of secondary phases. Little further decrease in Al concentrations was observed after 2 years and longer at low Ca/Si. At high Ca/Si no significant change in solution concentrations was observed after more than 3 month, while the destabilization of secondary phases continued up to 1 year, indicating that a (meta)stable equilibrium was reached faster at higher Ca/Si ratios.

katoite and Al(OH) 3 has been observed in other studies, which lowered the aluminum concentration in the aqueous solution and thus increased the Al uptake in C-S-H [16,20,28]. Further analysis revealed that the content of katoite, which had precipitated in the first months in addition to C-A-S-H phases, decreased with equilibration time, indicating an increasing Al uptake in C-S-H with time [12] pointing towards a continuing restructuring of C-A-S-H.
The main aim of this study was to investigate the effect of different reaction times on Al uptake in C-S-H. Long-and short-term sorption isotherms were recorded in order to perform a kinetic study at different ages of C-A-S-H samples. C-A-S-H samples were synthesized using different Ca/Si ratios, Al/Si ratios and at different alkali hydroxide concentrations. Samples were filtrated after different equilibration times from 7 days to 56 days in short-term experiments and from 3 months to 3 years in long-term experiments. The elemental concentrations of different ions in the solution were determined by Inductively Coupled Plasma Optical Emission Spectrometry (ICP-OES).

Material and methods
Short-and long-term sorption experiments were performed in order to investigate the effect of different equilibration times on Al uptake in C-S-H. In long-term experiments, the synthesis of samples were performed at room temperature by following the same procedure as detailed in [29] by adding a total of 3.8 g calcium oxide (CaO), silica fume (SiO 2 , Aerosil 200, Evonik) and calcium aluminate (CA: CaO⋅Al 2 O 3 ) into 171 mL of Milli-Q water or NaOH solutions (liquid/ solid = 45 mL/g) to obtain C-A-S-H with different compositions (Table 1). CaO was obtained by burning calcium carbonate (CaCO 3 , Merck, pro analysis) at 1000 • C for 12 h. CA was synthesized by heating the mixture of CaCO 3 and Al 2 O 3 (Sigma Aldrich) at 800 • C for 1 h, at 1000 • C for 4 h and at 1400 • C for 8 h. Then, the mixture was cooled down with a rate of 600 • C/h. The molar ratios of CaO, SiO 2 and CA were varied in order to obtain C-A-S-H with Ca/Si ranging from 0.6 to 1.4, and Al/Si from 0 to 0.1, as indicated in Table 1. In addition, NaOH concentrations of 0, 0.1, 0.5 and 1 M were selected to cover the range of pH values relevant for hydrated cements [20].
In the short-term experiments, C-S-H samples with Ca/Si ratios of 0.8 and 1.2 and 1 M NaOH were synthesized and equilibrated for 1 month. Afterwards, 0.086 and 0.073 g of sodium aluminate (NaAlO 2 ), corresponding to an Al/Si molar ratio of 0.03 and 1000 mM Na, were added to the pre-equilibrated C-S-H samples with Ca/Si ratios of 0.8 and 1.2, respectively. The composition of liquid and solid phases of C-S-H samples containing Al (C-A-S-H) was measured after 7, 14, 28 and 56 days.
All samples were synthesized in a nitrogen filled glovebox to minimize carbonation and were stored in 200 mL PE-HD containers placed on a horizontal shaker moving at 100 rpm and equilibrated at 20 • C. After different equilibration times, the solid and liquid phases were separated by vacuum filtration using nylon filters (pore size: 0.45 μm) and analyzed.

Solution phase analysis
ICP-OES was used to measure the elemental concentrations of Na, Ca, Si and Al in the filtrates. The samples were first acidified with HNO 3 to have 1% HNO 3 (using Suprapur HNO 3 , Merck). In order to keep the Na concentrations below 1100 mg/L for the ICP-OES analysis, the samples were diluted with 1% HNO 3 . Multi-standard solutions and blank solution (1% HNO 3 ), containing Al, Ca and Si, were prepared in the range from 0 to 20 mg/L (0, 0.2, 2 and 20) for ICP-OES.
The pH measurements were made at room temperature with a Knick pH meter (pH-Meter 766) equipped with a Knick SE100 electrode to measure the hydroxide concentration (OH − ) on the undiluted samples on the day of the filtration. The pH electrode was calibrated against 0.1, 0.2, 0.5 and 1 M NaOH solutions in order to minimize the alkali error [30].

Solid phase analysis
In order to investigate the effect of equilibration time on the composition of the solid phase, analysis were performed using Thermogravimetric Analysis (TGA), X-Ray Diffraction (XRD) and Fourier-Transform Infrared Spectroscopy (FTIR). Prior to analysis, the filtrated samples were washed first with a 50%-50% (volumetric) water-ethanol solution and then with 94% ethanol inside the glovebox to avoid the precipitation of alkali compounds during drying and to remove the free water, respectively. The samples were then dried in the freeze dryer for one week and then stored until analysis (at least 1 week) in nitrogen filled desiccators in the presence of saturated CaCl 2 .2H 2 O solution, which generates a relative humidity of 30%, following the procedure described by [25,28].
TGA data were acquired using a TGA/SDTA851e Mettler Toledo device. The weight loss of approximately 30 mg of each sample was recorded under N 2 atmosphere from 30 • C up to 980 • C with a heating rate of 20 • C/min. The water loss of strätlingite, Al(OH) 3 , katoite, portlandite and CaCO 3 was quantified using the tangential method and the theoretical weight loss of solids based on the measured weight loss between 150-240, 240-300, 300-350, 350-450 and 600-800 • C, respectively [31], was used to calculate the amount of these solids.
The X-ray powder diffraction patterns were recorded using the PANalycatical X'Pert Pro MDF diffractometer equipped with an X'Celerator detector. The diffraction patterns were collected in increments of 5 • to 55 • 2θ at a conventional step size of 0.026 • 2θ and a step measurement time of 447 s. The presence of different phases was determined with X'Pert HighScore Plus.
FTIR spectra were recorded in the range of 400-4300 cm − 1 on a Bruker Tensor 27 spectrometer with a resolution of 4 cm − 1 by transmittance on small amounts of powder. To make the comparison easier, the spectra were corrected for background and scaled to the maximum signal of Si-O bonds at 1100 cm − 1 . The second derivative of the measured transmittance (T) over the wavelength (W): d 2 T/dW 2 , were also plotted in order to better identify the different bands and differentiate the wavenumbers [32].
The elemental composition in C-A-S-H was calculated by mass balance taking into account the initial quantities, the amount of Al, Ca, Si and Na in secondary phases, if present and the fraction of Al, Ca, Si and Na in solution as detailed in [29].

Thermodynamic modeling
Thermodynamic modeling was performed with Gibbs Free Energy Minimization (GEM-Selektor) software [33,34], version 3.7. The PSI-Nagra thermodynamic database [35] was used to obtain the thermodynamic data for aqueous species, portlandite and amorphous SiO 2 . The Table 1 The Ca/Si, Al/Si molar ratios and NaOH concentrations used to prepare C-A-S-H at 20 • C. (note: the samples were prepared at each combination of Ca/Si, Al/Si and NaOH concentration with 4 replicates in order to be analyzed after different equilibration times). (*: samples at Ca/Si ratio of 0.8 were analyzed also after 24 and 36 months).  3 , strätlingite, C-S-H and katoite was taken from the Cemdata18 database [36] and the data for the zeolitic phases such as Ca-gismondine (CaAl 2 Si 2 O 8 4.5H 2 O) and OHsodalite (Ca 8 Al 6 Si 6 O 24 (OH) 2 2H 2 O) from [37]. The CSHQ thermodynamic solid solution model was used in order to model the concentrations in the C-S-H system [33]. The activity coefficients of the aqueous species γ i were calculated using the extended Debye-Hückel equation (Eq. (1)) with common ion-size parameter a i = 3.31 Å for NaOH solutions [38] and common third parameter b y according to where z i denotes the charge of species i, I the effective molal ionic strength, b y is a semi-empirical parameter (~0.098 for NaOH electrolyte at 25 • C), and A y and B y are P, T-dependent coefficients. This activity correction is applicable up to ~1 M ionic strength [39]. Solubility products (K so ) for C-A-S-H were calculated from the generalized dissolution reaction shown in Eq. (2): where a, b, c and d are the respective stoichiometric coefficients for CaO, SiO 2 , Al 2 O 3 and H 2 O in C-A-S-H. This reaction implies the following relationship for K so for Ca/Si = 0.8 (Eq. (3)): And for Ca/Si = 1.2 (Eq. (4)): Activities of Ca 2+ The saturation index (SI) of different solids according to SI=log IAP/ K so was calculated using the elemental concentrations of Al, Ca, Si and Na. The ion activity product (IAP) was calculated based on the measured concentrations in solution. A positive saturation index (> 0) indicates that the solution is oversaturated with respect to this solid phase and that this phase could possibly precipitate. A negative value indicates undersaturation. The SI calculation was used to verify the solid phase composition found experimentally.

The effect of equilibration time on secondary phases
The changes in the water content of C-A-S-H and the possible presence of secondary phases with time are followed by TGA. The TGA analysis of C-A-S-H samples after different equilibration times of 7 days up to 3 years are shown in Fig. 1. The water present at the surface and interlayer of C-S-H is lost over a broad temperature range up to 150 • C in agreement with different studies on C-S-H [28,40]. An additional weight loss is observed for the samples after 7, 14, 28 and 56 days at ≈430 • C, although no crystalline phases have been observed to be present by XRD after 3 months [29]. This weight loss has been observed previously for C-S-H in the presence of Na and Al and was tentatively assigned to an unidentified calcium sodium aluminate silicate hydrate (C-N-A-S-H) [25]. This poorly defined weight loss is absent after 2 and 3 years of equilibration and no other solids were detected by TGA and XRD [29].
The physically bound water content of C-S-H is strongly dependent on the drying procedure, duration and the relative humidity [41], resulting in some variation of the total weight loss up to 600 • C as shown in Fig. 1 The presence of Al could also lead to the formation of secondary phases containing Al. Microcrystalline Al(OH) 3 and strätlingite are observed by TGA mainly at low Ca/Si ratios and in the absence of NaOH, katoite mainly at higher Al contents but at all Ca/Si ratios and pH values (Supplementary Information A). In addition, the formation of zeolites or zeolitic precursor like Ca-gismondine and sodalite cannot be excluded at low Ca/Si ratios and high NaOH concentrations, although they are not clearly observed by TGA or XRD, as at Ca/Si = 0.6 some solutions are moderately oversaturated with respect to sodalite and at Ca/Si = 0.8 with respect to gismondine (Supplementary Information C). The TGA signal of the samples without alkali at Ca/Si ratio of 0.8 is also shown in Fig. 2a and b and at Ca/Si = 1.4 in Fig. 2c. While no or very little secondary phases are observed at Al/Si of 0.03 or lower, secondary phases are observed at higher Al/Si ratios in agreement with the observations of L'Hopital et al. [7]. Fig. 2 shows that increasing the equilibration time from 3 months to 3 years for samples with Al/Si = 0.1 and 0.2 leads to a decrease in the amount of secondary phases such as strätlingite, aluminum hydroxide and katoite. At Ca/Si = 0.8 and Al/Si ratio = 0.1, 1.4 wt% katoite is observed after 3 months equilibration which decreases to a non-detectable level after 3 years, while the solution remains undersaturated. Similarly at Al/Si = 0.2, increasing the equilibration time from 3 months to 3 years leads also to a decrease in the amount of Al(OH) 3 (from 1.4 wt% to 0.29 wt%) and katoite (from 2.1 wt% to 0.18 wt%) (Fig. 2b). An increase in equilibration time from 1 year to 3 years at Al/Si = 0.2 decreases the amount of strätlingite from 0.58 wt% to a non-detectable level. Also at Ca/Si ratio of 1.4 (Fig. 2c) the amount of katoite decreases from 1.9 wt% after 3 months to a nondetectable level after 1 year (around 0.28 wt%).
The XRD pattern of C-A-S-H samples at Ca/Si = 0.8 and Al/Si = 0.1 in the absence of NaOH is shown in Fig. 3. A trace of strätlingite and Al (OH) 3 is observed only after 3 months equilibration. In addition, the presence of katoite is observed after 3 months and 1 year equilibration, however, it disappears after 2 and 3 years. The decrease in the quantity of katoite with time is in agreement with TGA results as shown in Fig. 2a. L'Hopital et al. also observed katoite in addition to C-A-S-H phases after 6 months equilibration and suggested that katoite has formed initially from strongly oversaturated solutions and its quantity decreased with time for C-A-S-H at Al/Si = 0.1 as the solution was undersaturated with respect to katoite at the time studied [12]. The SI, in Supplementary  and katoite with time as reported in [7,12] and in the present paper, indicates consistently an increased uptake of aluminum in a thermodynamically more stable C-A-S-H phase with time.

The effect of equilibration time on the structure of C-A-S-H
The  [45][46][47][48][49][50]. The signal at 820 cm − 1 is related to Si -O stretching of Q 1 tetrahedra and is more intense at high Ca/Si ratios. The main signal ranging from 900 to 1200 cm − 1 is assigned to the stretching and deformation or bending vibration of the Si-O bands [51]. The band centered in the range 950-970 cm − 1 has often been attributed to Q 2 silica units of C-S-H [50,52,53]. However, a recent publication highlighted that the several bands contribute to that band at around 970 cm − 1 [51]. Two bands at ~930 cm − 1 and ~1060 cm − 1 are associated with Q 2 -silicate species, and a third band at ~1005 cm − 1 is attributed to Q 1 -silicate species [51]. This is in agreement with the observation of Yu et al. [50], where a higher intensity of shoulder above 1000 cm − 1 was established at higher Ca/Si ratios. Bands in the range of 500-750 cm − 1 can be assigned to the stretching vibrations of Al -O bonds of the octahedrally coordinated Al, and bands in the range of 750-900 cm − 1 to the vibrations of Al-O bond in AlO 4 units [54]. The band at about 880 cm − 1 has been associated with the stretching vibration of Al-O-Si (terminal bond) [55]. The symmetric bending of Al-O-H have been reported in the range of 1034-1075 cm − 1 [58][59][60].   vibration of Q 2 sites at 930 cm − 1 increases with increasing Al content, as is visible in Fig. 4a, which could be related to the replacement of silica in the Q 2 sites by Al.
The effect of equilibration time on C-A-S-H composition is represented in Fig. 5 for samples without alkali for Ca/Si ratio of 0.8 and Al/Si ratio of 0.2. The main difference is observed in the intensity of the Al-O-H shoulder at around 1050 cm − 1 which increases with longer equilibration time indicating a rearrangement of the Al over time (Fig. 5a). Moreover, the intensity of Si -O stretching of Q 2 sites at 930 cm − 1 increases with equilibration time, which could indicate longer silica chains with time. No clear changes are observed in the intensities of the other signals. Fig. 6 shows the effect of equilibration time on C-A-S-H in the presence of 1 M NaOH for Ca/Si ratios of a, b) 0.8 and c, d) 1.2. At Ca/Si of 0.8, the intensity of (Si-O) Q 1 peak at 820 cm − 1 is higher than in the alkali free samples (Fig. 5) indicating the presence of more end-of-chain-SiO 2 groups and thus shorter silica chain length in the presence of high alkali hydroxide concentrations, in agreement with Si-Nuclear Magnetic Resonance (NMR) observations of similar samples [20,25,61]. At Ca/Si of 1.2, the intensity of the band at 820 cm − 1 is higher, consistent with the shorter silica chain length expected at higher Ca/Si ratios. For both Ca/Si ratios of 0.8 and 1.2, the intensity of (Si -O) Q 1 peak at 820 cm − 1 increases with an increase in equilibration time. At Ca/Si = 0.8, the signal of Si -O stretching vibration at around 950 cm − 1 shows a splitting into two signals at low equilibration times up to 56 days, while at later times only one signal is observed. Such a splitting of the signal can indicate a low symmetry [62]. When structural sites are present as groups of nonequivalent sites, it results in a splitting of the absorption bands indicating a low symmetry. The disappearance of this signal splitting indicates that environment becomes more symmetric with time, which suggests a structural rearrangement in the C-A-S-H at least up to 90 days. The Si -O stretching vibrations around 950 cm − 1 and 910 cm − 1 move to a lower wavelength with time for Ca/Si ratio of 0.8 and 1.2, respectively. Also, the intensity of the signal at around 665 cm − 1 , which is related to the Si-O-Si vibrations and water librations [50], decreases with time which could indicate a restructuring of the water with time in the C-A-S-H structure. Less clear changes are observed at Ca/Si = 1.2, indicating that less rearrangement occurs at a higher Ca/Si ratio which is consistent with the observations in the liquid phase as discussed in Section 3.2.2.

The effect of short and long equilibration times on C-A-S-H
The effect of equilibration time on Al uptake in C-S-H for Ca/Si ratios of 0.8 and 1.2 for samples containing 1 M NaOH is shown in Fig. 7. For both Ca/Si ratio of 0.8 and 1.2 the Al concentrations seem quite constant between 7 and 28 days, but decrease clearly after 56 days, i.e. at the time when the FTIR signals indicate a clear structural rearrangement as shown in Fig. 6. This indicates an increase in Al uptake in C-S-H over time at both Ca/Si ratios. Little further change in Al concentration is observed at Ca/Si = 1.2 between 3 months and 1 year equilibration time indicating that a (meta)stable equilibrium might have been reached. In contrast, for Ca/Si = 0.8 the Al concentrations in solution continue to decrease, albeit more slowly, up to 3 years equilibration, indicating that equilibrium is reached slower at low Ca/Si than at high Ca/Si ratios.
This very slow ripening at low Ca/Si is visible in the presence of 0.1 M NaOH as shown in Fig. 8, where the Al uptake in C-S-H for the Ca/Si ratio of 0.8 is plotted from 3 months to 3 years (data for 0.5 and 1 M NaOH are shown in Supplementary Information D). In agreement with results discussed above, the decrease of Al concentrations and hence the uptake of Al in C-S-H accelerate over time. Increasing the equilibration time from 3 months to 1 year leads to a more distinct lowering of Al concentrations compared to the change from 1 year to 2 years. In fact, little changes in the Al concentrations are observed between 2 and 3 years equilibration indicating that the reaction of Al with C-S-H is approaching a metastable equilibrium after 2 years. Fig. 9 shows the effect of equilibration time on the Al uptake in C-S-H at 5 different Ca/Si ratios of a) 0.6; b) 0.8; c) 1.0; d) 1.2 and e) 1.4. It can be seen (Fig. 9a-c) that at Ca/Si ratios of 0.6, 0.8 and 1.0 the aluminum concentrations decrease significantly with increasing the equilibration time from 3 months to 1 year. However, at higher Ca/Si ratios of 1.2 and 1.4 (Fig. 9d and e) the Al concentrations in solution do not change significantly with time.

The effect of Ca/Si, NaOH and time on Al uptake in C-S-H
Moreover, the change in Al concentrations over time is more distinct at lower NaOH concentrations compared to higher concentrations. In fact, for an increase in equilibration time from 3 months to 1 year the relative decrease in Al concentrations is higher for no alkali and 0.1 M NaOH, however, it is less in the presence of 0.

The effect of equilibration time on Ca and Si concentrations in solution
The results presented above, in particular, the decrease of secondary Al-containing phases with time and the decrease of Al concentrations in solution with time in the absence of secondary phases indicate the slow formation of thermodynamically more stable C-A-S-H phases with time. The effect of equilibration time on the concentrations of Ca and Si in solution at Ca/Si ratios of a) 0.8 and b) 1.2 is shown in Fig. 10 (data for the other Ca/Si ratios are shown in Supplementary Information E). The data measured at different Ca/Si ratios and at 0 to 0.1 Al/Si ratios are plotted as symbols as a function of pH and compared to the expected solubility of C-S-H indicated by lines. The measured values deviate to some extent from the modelled data, since the CSHQ model was developed based on data for alkali free C-S-H only [33,34] and needs to be further extended to describe the changes at high pH values and the uptake of alkalis in C-S-H as discussed in L'Hopital et al. [25].
In the absence of alkali hydroxide the Si and Ca concentrations do not vary significantly with time. At low Ca/Si ratios of 0.6 ( Fig. E1 and 2 in Supplementary Information E) and 0.8 ( Fig. 10a and b), no significant changes are observed in the Ca and Si concentrations in the absence of alkali hydroxide. The saturation indexes (SI) calculated from the measured concentrations indicate that the solution is near to saturation with respect to C-S-H in all cases. The solution is generally under saturated with respect to amorphous SiO 2 (with the exception of Ca/Si = 0.6 without NaOH), microcrystalline Al(OH) 3 , katoite and strätlingite indicating that these phases will dissolve if precipitated initially. The SI values for different phases are presented in Supplementary Information C. At high alkali concentrations (12 < pH < 14), however, the Si concentrations slightly decrease in particular from 3 months to 1 year (Fig. 10b). Little variation with time is observed for Ca, with the exception of the outlier values after 1 year for Ca/Si = 0.8 at 0.1, 0.5 and 1 M NaOH, where might some contaminations prior to the Ca measurements had occurred). The observed changes in Si concentrations at high alkali concentrations can be associated to the disappearance of an unidentified calcium sodium aluminate silicate hydrate (C-N-A-S-H) phase with time, which is only observed at high alkali concentrations [25]. At

The effect of equilibration time on C-A-S-H solubility
The solubility of C-A-S-H depends on the composition of the solid, i.e. the amount of Ca, Si and Al in the C-S-H structure, and is often described with solid solution models using different end-members [33,63,64]. The development of such solid solution models is generally based on a large number of experimental data and is a complex and challenging task, which is not within the focus of the present paper. However, the solubility of a single composition in equilibrium with the solution can easily be calculated based on the measured elemental concentrations.
The C-A-S-H solubility with Ca/Si ratios of 0.8 and 1.2 at Al/Si = 0.03 and 1 M NaOH is calculated as detailed in Eq. (3) and (4), assuming a constant composition of the solid C-A-S-H and is shown in Fig. 11 after  [63]. The absence of any big changes in the solubility product with time despite the decrease in Al concentrations is an important result, which will help for the development of thermodynamic models for C-A-S-H.

Conclusions
The Al uptake in C-S-H was investigated after different equilibration times between 7 days to 3 years using sorption isotherm experiments. The experiments indicated that Al uptake in C-S-H increases with equilibration time and this change over time is more distinct at low Ca/ Si ratios than at high Ca/Si ratios indicating that equilibrium is reached faster at high Ca/Si ratios. At low Ca/Si ratios, changes in Al and Si concentrations in solution were observed up to 3 years, although the changes were less important for the aged samples.
The solubility of C-A-S-H was calculated with thermodynamic  modeling using the concentrations measured after different equilibration times for C-A-S-H with Ca/Si ratios of 0.8 and 1.2. The calculation indicated that the solubility of C-A-S-H with Ca/Si = 0.8 and 1.2 does not change significantly with equilibration time.
In addition to the C-A-S-H phase, also secondary phases such as strätlingite, Al(OH) 3 and katoite formed initially, in particular at higher Al contents. Their content decreased in all cases after longer equilibration times indicating the dissolution of such secondary phases and an increase in uptake of Al in C-S-H. The FTIR analysis showed a restructuring of the C-A-S-H phase with time and this change is more significant at lower Ca/Si ratios.
CRediT authorship contribution statement S. Barzgar

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.