Variation of reactant concentrations and water content

Different experimental parameters were systematically varied, primarily the concentration of the added CaSO₄ solution and the water content in the final solvent mixture. Results of these experiments are summarized in Figure 7.2, which shows a 3D plot illustrating the amount of bassanite detected in the product (z-axis, in wt% relative to gypsum) as a function of the water content (x-axis, in wt% relative to ethanol in the final solvent mixture after addition of CaSO_4(aq)), and the concentration of the CaSO₄ solution (y-axis, in mM before quenching into ethanol).

146

Figure 7.2: Bassanite content of precipitates formed upon addition of aqueous calcium sulfate solutions (with variable concentrations) into ethanol/water mixtures at different final water contents. Products were isolated from the resulting dispersions by centrifugation and subsequent drying in vacuum at room temperature.

Bassanite yields were obtained by IR analyses as depicted in Figure 7.3.

The yield of bassanite was determined from IR spectra of the obtained solids by using the intensity of the ν₁ band at 1684 cm^-1, which only occurs for gypsum.^326,364 Intensity values were correlated with bassanite/gypsum mass ratios by measuring a series of defined mixtures of the two polymorphs. It was found that the intensity of the ν₁ peak decreases exponentially with increasing bassanite content, so that a corresponding fit of the data was used as calibration curve (see Figure 7.3).

Figure 7.3: (a) IR spectra of samples containing bassanite and gypsum in varying defined mass ratios (obtained by mixing the two polymorphs in dry state and homogenizing the conglomerate by careful grinding in a mortar).

The red arrow marks the peak at 1684 cm^-1, which only occurs for gypsum. (b) Plot of the intensity of the band at 1684 cm^-1 (in units of transmission) as a function of the bassanite content (w_Bassanite) in the mixtures. Squares represent experimental data, while the full line is an exponential fit giving the following “calibration” equation:

w_Bassanite = 46.3∙ln[(1-T₁₆₈₄)/0.16].

147 The error associated with this procedure is estimated to ca. ± 15%. Essentially, the main product obtained for final water contents ranging from 16ⁱ to 33 wt% was bassanite, regardless of the CaSO₄ concentration. In all these cases, the apparent yield of bassanite was

> 90%, which corresponds to a virtually phase-pure material with regard to the limits of experimental error. When the water content was further increased to 41 wt%, the bassanite fraction decreased to ca. 60% for 75 mM CaSO4(aq), and to less than 20 and 10% for 50 and 25 mM, respectively. However, for CaSO4 concentrations of 100 and 150 mM, bassanite was still the main phase obtained, although certain amounts of gypsum crystals could be detected in these samples; moreover, the bassanite nanoparticles formed under these conditions exhibited somewhat different, partially spindle-shaped morphologies instead of regular rod-like habits (see Figure S43 in the Appendix). Starting from 43 wt% water in the final dispersions, the bassanite yield was progressively lowered at all CaSO4 concentrations investigated, although the trend of higher bassanite fractions in samples containing more CaSO4 is still apparent. At 54 wt% water, the measured bassanite/gypsum ratios were below 20% at all CaSO4 contents.

Analyses of isolated solids confirmed that gypsum was the dominant polymorph in these samples, while bassanite was hardly detectable in most cases (see Figure S44 in the Appendix). Taken together, these data show that quenching of aqueous calcium sulfate solutions into ethanol yields pure bassanite as long as the total water content in the final dispersions does not exceed a certain limit. According to the present results, this threshold lies between 33 and 41 wt%. This suggests that ethanol plays a crucial role in stabilizing formed bassanite nanoparticles. On the other hand, as the initial aqueous CaSO₄ solutions were all clear and transparent with no sign of turbidity, one may speculate that ethanol actually induces the precipitation of bassanite from solution,ⁱⁱ while subsequently protecting it against transformation to gypsum. In any case, from a technical point of view, the data indicate that syntheses of bassanite via the developed protocol should be conducted at 150 mM CaSO4(aq)

and 41 wt% water, as this requires the lowest amount of organic solvent and affords the highest possible absolute yield of pure bassanite.

i Note that 16 wt% is the minimum final water content in syntheses where 50 mL CaSO_4(aq) was added to 500 mL EtOH. This is because the used ethanol was of technical grade and thus already contained about 4 wt% water.

Higher final water contents were realized by using H₂O/EtOH mixtures instead of pure ethanol for quenching.

ii This notion is further supported by the fact that fairly large amounts of bassanite were obtained after quenching, whereas the number of particles dispersed in the purely aqueous solution before quenching was too low to be detected by scattering techniques.

148 7.4.3 Influence of solvent volume and polarity

In further experiments, the effect of other reaction parameters was investigated, such as the initial volume and type of solvent used for quenching. A decrease in the ethanol volume from 500 to 300 mL already led to the occurrence of significant fractions of gypsum upon addition of 50 mL CaSO4 solution (50 mM; see Figure S45 in the Appendix for corresponding data).

This effect became even more pronounced when the volume was further reduced to 200 and 100 mL. Again, the results suggest that a critical excess of ethanol is required to kinetically produce and stabilize bassanite. Replacing ethanol as the organic solvent by methanol, isopropanol, or acetone did not affect the yield of bassanite at a final water content of 16 wt%

(i.e. when adding 50 mL of 50 mM CaSO₄ solution to 500 mL organic solvent), as confirmed by IR analyses (see Figure S46 in the Appendix). By contrast, when tetrahydrofuran (THF) was used as solvent, the obtained solids contained about 67% gypsum straight after precipitation. This means either that THF is less capable of inhibiting the bassanite-to-gypsum transformation, or that it favors the kinetic formation of bassanite over gypsum to a lower extent than the other solvents studied. At 33 wt% water, still 100% bassanite was obtained with methanol (as for ethanol, cf. Figure 7.2). In the case of isopropanol and acetone, however, the yield of bassanite was markedly reduced under these conditions, leading to measured bassanite fractions of 55% (isopropanol) and 75% (acetone). Interestingly, the observed trends in efficiency of the different solvents to generate and/or stabilize bassanite directly correlates with their polarity, as demonstrated by the following order of the solvents according to their suitability for bassanite synthesis (see Figure 7.4): methanol (ε = 33.0) ≈ ethanol (ε = 25.3) > acetone (ε = 21.0) > isopropanol (ε = 20.8) > THF (ε = 7.5), where ε is the dielectric constant of the solvent³⁷⁰ at 25°C (ε = 80.1 for water). In other words, the better the miscibility of the solvent with water (or respectively the higher its mixing enthalpy with water), the more effective will the solvent be in terms of bassanite formation. This behavior can be understood when considering that bassanite contains less hydration water than gypsum and, hence, that more water has to be removed from any solution (or amorphous solid) precursors to form bassanite. Consequently, solvents that are more prone to withdraw water (i.e. more polar ones) should be more suitable for the preparation of bassanite via the applied methodology, in line with what is observed experimentally.

149

Figure 7.4: The graph illustrates the trend (dashed red line) observed for the correlation between the bassanite fraction (wt%) of the product and the dielectric constant of the solvent. Note, the higher water content in the final isopropanol/water and acetone/water reaction mixtures compared to the other reaction mixtures (see Figure S46) was neglected.

Such a dehydration-based scenario could furthermore explain the porosity of the bassanite nanoparticles, as the voids in the rods may result from rapid removal of water from calcium sulfate species precipitating under kinetically controlled, non-equilibrium conditions. In this regard, it seems as if the primary role of the organic solvents is to favor the crystallization of bassanite over that of gypsum. Interestingly, early studies on the precipitation of CaSO₄ in pure alcohol (MeOH in this case) reported the formation of solvate adducts with a crystal structure akin to that of bassanite, but containing methanol molecules instead of water in the lattice.³⁷¹ This, however, was not observed in the present work even at low final water contents, again suggesting that dehydration of water-rich solute (or solid) CaSO4 precursors is a key step in the formation of bassanite in excess organic solvent.

Finally, the work-up procedure was modified in order to assess the potential of the developed protocol towards up-scaling. Instead of removing residual solvent after centrifugation by applying vacuum, the wet solids were dried at an elevated temperature of 60°C for 30 min. As in the experiments described above, synthesis conditions were varied with respect to the water content and the CaSO4 concentration. Exemplary IR, XRD, TEM and SEM data for a sample at 150 mM CaSO4 and 16 wt% water are reproduced in Figure S47 in the Appendix, while Figure S48 shows the “phase diagram” established for this series by means of IR analyses. At low water contents (16 wt%), pure bassanite was again obtained in the experiments, independent of the CaSO₄ concentration used. However, starting from 25 wt% water, there were already significant losses in the yield of bassanite (ca. 80% at 25/50 mM CaSO₄, and

< 50% at higher concentrations; see Figure S48 in the Appendix). Any further increase in the

150 water content led to a drastic reduction of the amount of bassanite obtained. In fact, solids isolated from mixtures containing 41 wt% H₂O consisted almost exclusively of gypsum, while only minor traces of bassanite could be detected in some cases. Thus, drying the samples at elevated temperature seems to promote the transformation of bassanite to gypsum in the wet slurries, especially when a significant amount of gypsum was already formed during the initial precipitation step.

7.4.4 Stability of bassanite and mechanism of bassanite-to-gypsum transformation Given that bassanite is metastable with respect to gypsum,³³² the temporal stability of the phase-pure bassanite nanoparticles obtained at low water contents, both in dry state and in ethanolic dispersion, was tested. In the former case, there was no noticeable transformation into gypsum over a period of 3 weeks, as suggested by XRD data (see Figure S49 in the Appendix). In dispersion, the bassanite particles proved to be stable for about 4 weeks at a CaSO4 concentration of 50 mM and a water content of 16 wt% (see Figure 7.5a) – in the absence of any stabilizing additives and despite continuous stirring. By contrast, progressive transformation of bassanite into gypsum could be observed when the CaSO4 concentration was increased to 150 mM or longer ageing times were chosen. Under these conditions, 25%

conversion to gypsum was detected after an ageing time of 8 days, whereas ca. 85% of the initially present bassanite had transformed after 4 weeks (see Figure 7.5b).

Figure 7.5: Bassanite stability in dispersion: IR patterns of samples drawn after different ageing times from ethanolic dispersions prepared by adding 50 mL of (a) 50 mM and (b) 150 mM CaSO_4(aq) solution to 500 mL ethanol (final water content: 16 wt%). During ageing, the dispersions were stirred continuously and formed particles were isolated at the respective time by centrifugation and drying the resulting sediment in vacuum at room temperature. Note that bassanite was stable for at least 28 days at the lower CaSO₄ concentration (absence of the band at 1684 cm^-1), whereas significant amounts of gypsum (ca. 25%) were present after 8 days at the higher concentration. With time, transformation of bassanite proceeded gradually at 150 mM CaSO4, and about 85% of gypsum were detected after 28 days.

In order to study the mechanism of bassanite-to-gypsum transformation in more detail, the synthesis parameters were tuned such that conversion to gypsum occurred within a reasonable frame of time, allowing for convenient sampling and monitoring of the process. This was

151 found to be possible at a CaSO₄ concentration of 50 mM and a final water content of 33 wt%.

When corresponding dispersions were stirred vigorously, it took approximately 3 h for the transformation to be completed, as indicated by bassanite fractions determined through IR spectroscopy, which show a roughly linear decrease from 100 to ~0 wt% over time (Figure 7.6a).

Figure 7.6: Transformation of bassanite nanoparticles into stable, micron-sized gypsum crystals. (a) Fraction of bassanite detected by IR as a function of ageing time in an ethanolic dispersion with a water content of 33 wt%

(obtained by addition of 50 mL 50 mM CaSO₄ solution to 500 mL ethanol). Note that the dispersion was stirred vigorously during the entire experiment. (b-d) TEM images showing the supposedly crucial step in the transformation process, i.e. oriented attachment of bassanite nanorods leading to crystallographically aligned particle assemblies that later merge to yield crystalline gypsum.

The transformation of bassanite into gypsum was also followed via in-situ IR, using the same conditions. Three distinct peaks (1680 , 1620 and 1134 cm^-1) were chosen to study the ongoing phase changes while stirring the mixture over a reaction time of ~2 h. Unless normalized, the absorbance of these peaks increases with increasing fraction of gypsum in the formed products (see Figure S50). Indeed, an increase of absorbance over time was observed in the resulting data at all three positions, suggesting that gypsum is continuously formed over the entire studied time period (Figure 7.7). The low signal-to-noise ratio at the beginning of the measurement is probably due to addition of the aqueous calcium sulfate solution and thus, a rapid change of solvent composition. The system was equilibrated after a few minutes.

152 The data obtained by in-situ IR confirm observations from ex-situ IR that an increasing amount of gypsum is formed with increasing time. For a detailed comparison of the data acquired by in-situ and ex-situ IR, an in-situ measurement over a longer time period of 3-4 h is required.

Figure 7.7: Transformation of bassanite into gypsum was followed by in-situ IR experiments. An increase in absorbance at distinct wavenumbers of 1680 cm^-1 and 1620 cm^-1 (a) as well as 1134 cm^-1 (b) suggests the formation of gypsum.

On the other hand, TEM micrographs of samples collected during the transformation process clearly show that the initially present bassanite nanorods convert into gypsum via an aggregation-based process, in which individual particles first assemble and align parallel to the long axes of the rods (Figure 7.6b-d), to subsequently merge in a common crystallographic register and collectively transform into gypsum. In this regard, the data directly confirm the notion that gypsum crystals are formed from bassanite nanorods via particle aggregation and oriented attachment, as observed in previous studies.^62,333 However, while this mechanism appears to be the preferential reaction pathway for the formation of gypsum starting from an existing population of bassanite nanoparticles, it does not necessarily apply for gypsum crystallization from purely aqueous media in general. This is because, in light of the experiments conducted in this work, the initial formation of bassanite seems to be induced, or at least favored, by organic solvents that often are used for ‘quenching’ the aqueous systems.323,326,333,362

7.4.5 Synthesis of phase-pure anhydrite

It was shown that well-defined nanoparticles of bassanite can be synthesized by quenching aqueous solutions of CaSO4 into an excess of an organic solvent like ethanol. The polymorphic composition of the product was found to be essentially determined by the amount of water in the mixtures: pure bassanite was obtained below a certain critical water content (ca. 33 wt%), whereas increasing fractions of gypsum (and ultimately only gypsum)

153 formed as more and more water was added. This behavior was ascribed to the limited availability of H₂O in the medium, leading to the preference for the less hydrated, though actually metastable, phase under these conditions. This methodology was extended to produce anhydrite via a further reduction in the water content. Thus, full control over CaSO4

polymorphism at ambient temperature was gained, allowing to switch between all three crystalline phases simply by adjusting the concentration of water in organic solvent-based syntheses.

In order to successfully prepare pure anhydrite via the aforementioned approach, several modifications had to be made to the protocol developed for the formation of bassanite (see Chapter 7.4.1). First, the reagents (i.e. Ca²⁺ and SO4

2-) were not pre-mixed in water and then quenched into ethanol, but were rather combined directly in the organic medium. However, when dissolving CaCl2 in ethanol and adding a concentrated aqueous solution of Na2SO4 (see Experimental Section for more details), immediate precipitation of crystalline sodium sulfate was observed due to its very low solubility in alcohol (giving peculiar spherical superstructures, as shown by Figure S51 in the Appendix). Therefore, as a second measure, Na₂SO₄ was replaced as the sulfate source by sulfuric acid; this led to the crystallization of bassanite particles with interesting fiber-like morphologies (see Figure 7.8), but still no anhydrite could be detected.

Figure 7.8: Calcium sulfate particles produced upon addition of concentrated sulfuric acid to an excess of ethanol containing dissolved calcium chloride (H2SO4 and CaCl2 combined in equimolar amounts; final water content: ca. 0.2 wt%; Ca:H₂O = 1:5.7). (a-b) TEM images of the precipitates, showing extended networks of fibers as main morphology, with typical widths of ca. 40-50 nm and lengths of several microns per fiber. (c) Corresponding ED pattern, verifying the presence of bassanite.

The third and decisive modification was to use anhydrous calcium chloride (instead of the dihydrate) and, most importantly, to switch from ethanol to dry methanol as the solvent. In this way, the effective water content in the medium could be decreased to below 0.1 wt%, ultimately enabling the spontaneous formation of anhydrite.

154 Thus, in a typical experiment, CaCl₂ (10 mmol) was dissolved in absolute methanol (50 mL) and an equimolar portion of H₂SO₄ was added by dosing aliquots of concentrated sulfuric acid (assumed to contain 5 vol% H₂O); the amount of water in the final mixture was about 1.6 mmol (~0.07 wt%) and hence, there was a more than six-fold molar excess of CaSO4 over H2O in the reaction volume (note that the CaSO4:H2O molar ratios used for the preparation of bassanite were much lower, usually in the range of 1:2000). Upon addition of H2SO4, the initially transparent solution turned cloudy immediately, indicating rapid formation of calcium sulfate particles that remained dispersed in the resulting viscous slurry. The product was isolated by centrifugation, washed and dried, and subsequently analyzed by means of TEM, (Figure 7.9), IR spectroscopy (Figure 7.10), and XRD (Figure 7.11). TEM images show that the precipitates consist of ca. 10-20 nm primary particles, either spherical or elongated, which have aggregated and partially fused into larger, less defined structures (Figure 7.9a).

Figure 7.9: TEM micrographs (left) and ED patterns (right) of particles formed after addition of concentrated H₂SO₄ to CaCl₂ solutions in (a-b) dry methanol (containing about 0.07 wt% H₂O, CaSO₄:H₂O molar ratio of 1:0.16) and (c-d) methanol with a total water content of 4.14 wt% (CaSO₄:H₂O = 1:10). Precipitation in pure methanol yields aggregates of spherical nanoparticles (a), which consist of pure anhydrite (b). In the presence of more water, rod-shaped particles are obtained (c), which were confirmed to be bassanite (d).

Electron diffraction (ED) patterns of such agglomerates demonstrate that they are composed of crystalline anhydrite (Figure 7.9b). In some cases, reflections corresponding to calcium oxide were detected (see Figure S52 in the Appendix), likely due to rapid transformation of

155 anhydrite into CaO under the influence of the electron beam (since no signs for CaO could be found in XRD patterns, see below). IR spectra confirm the selective formation of anhydrite, as is evident from the position (and splitting) of the ν₄ anti-symmetric bending modes of the sulfate ion (Figure 7.10): anhydrite shows three distinct bands at 594, 617 and 673 cm^-1 in this region whereas, due to higher symmetry, there are only two peaks at different wavenumbers in the case of bassanite (601 and 659 cm^-1, cf. Figure 7.10b) as well as gypsum (594 and 667 cm^-1, see Figure S53 in the Appendix).^326,372

Figure 7.10: Infrared spectra of (a) pure anhydrite, formed by precipitation from methanolic solutions at a water content of 0.07 wt% (CaSO₄:H₂O ratio of 1:0.16), and (b) anhydrite, bassanite and anhydrite-bassanite mixtures obtained in the presence of increasing amounts of water, namely (from top to bottom, corresponding CaSO₄:H₂O ratios in brackets) 0.07 (1:0.16, pure anhydrite, A), 0.54 (1:1.25), 0.60 (1:1.4), 1.49 (1:3.5), and 3.14 wt% (1:7.5, pure bassanite, B). Relative percentages of the two polymorphs, as determined by calibration with a series of defined anhydrite-bassanite mixtures (cf. Figure 7.12) are indicated.

This notion is further supported by the absence of defined water vibrations at about 1600 and

This notion is further supported by the absence of defined water vibrations at about 1600 and