*3.1. Carbonate Contamination*

The synthesis of carbonate-free HC from metal oxides and hydroxides has not been attempted with this aim in literature before, to our knowledge. However, some authors have used only calcium and aluminium oxides and hydroxides for their HC synthesis which could facilitate the formation of a CaAl-OH-LDH phase [3,23–25]. In most of these studies involving the formation of HC, especially when attempting to form nitrate or chloride forms or studying the thermodynamics, carbonate contamination was limited. When carbonate contamination occurred, it was typically said to have occurred during synthesis, filtering or drying. In the studies where carbonate intercalation was not desired, grea<sup>t</sup> care was taken to carry out the synthesis in an inert environment (glove boxes or covered systems), use freshly precipitated Ca(OH)2 and/or dried materials under nitrogen or in a desiccator to minimise contact with CO2 [3,23–25].

In this study, an inert environment was accomplished by synthesising the LDHs under a slight over-pressure induced by constant nitrogen flow. The materials were filtered in open atmosphere but only required approximately 5 min to filter. XRD and FTIR-ATR analysis was carried out wet and within 0.5 h and 1 h after synthesis, respectively. The only possible CO2 contamination from the atmosphere could have occurred during the 5 min of filtering (after which the filtered samples were sealed from air immediately and only unsealed for the wet XRD analysis and again for wet FTIR-ATR analysis). To put this into perspective, if the carbonate contamination were to come from CO2 in air, at the atmospheric conditions in Pretoria, South Africa, where these experiments were conducted, 6.69 m<sup>3</sup> of air would have needed to pass through the sample and every available mol of CO2 absorbed within the 5 min of filtration time or surface adsorbed during XRD and subsequently intercalated. As this is essentially impossible, the interlayer carbonate, if present to a significant degree, thus had to come from a different source.

For the synthesis of the materials, a fresh bottle of analytical grade Ca(OH)2 was used. Unfortunately, after synthesis and our surprise of this possible CO2 contamination, it was found that the Ca(OH)2 used nevertheless consisted to 3.76% of calcite. This constitutes only 13.36% of the amount of calcite stoichiometrically required to form a fully carbonate intercalated CaAl-LDH, though. Taking into account theoretical requirements of carbonate to form a fully carbonate intercalated CaAl-LDH, one could thus expect that the LDH consisted of 13.36% carbonate in the interlayer. However, almost every material synthesised contained some calcite at the end of the synthesis. In prior studies [25,26,28], CaCO3 was used as a carbonate source during synthesis and most of these studies produced very pure CaAl-CO3-LDHs. Considering the approximately dried material yield, our materials consisted of between 0.5 g to 1 g of calcite after synthesis (approximately 2% as determined from Rietveld refinement). This is approximately as much as initially fed with the Ca(OH)2, indicating that a small amount of carbonate could have been taken up by the CaAl-OH-LDHs, but a large fraction remained as calcite.

Our Rietveld refinement results showed that between 1% and 3% calcite typically remained in the material after synthesis. However, there were some outliers. Some materials (MR1, WS3, T1, T2 and A3) contained no or less than 1% calcite, while others (WS1 and T5) consisted of close to 10% calcite and CO3 even of 20% calcite. MR1, due to the use of less Ca(OH)2 during synthesis to achieve the required molar calcium-to-aluminium ratio would contain less calcite to begin with. This result is thus not entirely surprising. WS3 contained less material in general to achieve the desired water-to-solids ratio. In this case, one could argue, that the calcite could have dissolved better due to a higher shear synthesis induced by the thinner reaction mixture (the same would apply to MR1 and A3). However, FTIR-ATR showed a decreased vibrational strength in the doublet 1414 cm<sup>−</sup>1/1361 cm<sup>−</sup><sup>1</sup> vibration, which in most literature is designated to interlayer carbonate. T1 and T2 both showed no remaining calcite, however, the primary phase present in these materials was found to most likely be a calcium aluminate carbonate hydrate of the form 3 CaO · Al2O3 · 0.5Ca(OH)2 · 0.5CaCO3 · 11.5H2O, which could explain where some of the CaCO3 was used. Interestingly, the calcium aluminate hydrate phases presented by [23] at 20 °C and 30 °C were not observed. The only difference between A3 and its other series materials (A1 and A2) was the use of a highly crystalline gibbsite as Al source. While interlayer carbonate was present in this material as identified through FTIR-ATR analysis, it had a much lower vibrational response at 1414 cm<sup>−</sup>1/1361 cm<sup>−</sup><sup>1</sup> and 663 cm<sup>−</sup>1, three vibrations that are typically linked to interlayer carbonate—especially the vibration at 663 cm<sup>−</sup>1, where this material had by far the least intense vibrational response of all materials.

This leads to the discussion of the importance of the blue-shifted doublet (from the doublet 1414 cm<sup>−</sup>1/1361 cm<sup>−</sup>1) vibrations of the Al(OH)3-SA and Al(OH)3-ACE sources indicated in the results section. To our amazement, the possibility of carbonate contamination in HC through adsorption of CO2 onto the surface of Al(OH)3 has not been indicated in any previous research, to our knowledge. We were also unable to find any FTIR spectra of Al sources used in previous research for the synthesis of HC. It is, however, well established in the catalytic field that aluminium oxides and hydroxides adsorb carbonate species. CO2 adsorption onto boehmite and other aluminium phases was studied by [38] at different activation temperatures and pressures of CO2. While their findings of the adsorbed species do not match our vibrational responses, the amorphous nature of the Al(OH)3 sources used in this text could still allow for good explanation of the reason why CO2 adsorbed onto the surface of Al(OH)3-SA and Al(OH)3-ACE and not onto the crystalline gibbsite phase. Amorphicity of materials typically goes hand in hand with some degree of disorder in these systems, thus making it more likely to adsorb CO2 onto coordinatively unsaturated Al atoms with octahedral coordination that are very active toward CO2 adsorption. Another explanation could come from the presence of small amounts of sodium present in the amorphous Al(OH)3. While Al does not form carbonates itself, the presence of Na easily leads to the formation of dawsonite (NaAl( OH)2CO3) phases. Prior to the mid-1990s, several authors reported vibrational bands between 1520 cm<sup>−</sup>1–1570 cm<sup>−</sup><sup>1</sup> and/or 1350 cm<sup>−</sup>1–1410 cm cm<sup>−</sup><sup>1</sup> that were linked to CO2 adsorption on *γ*-alumina phases. Ref. [36] linked these to carbonate vibrations of the bidentate form which is enhanced through the reaction of moisture, CO2 and Na to dawsonite-like phases that are vibrationally active in this region. The amorphous Al(OH)3 used in this work, unfortunately, did not have an elemental analysis provided by the manufacturer, but some Na was present in Al(OH)3-M. If small amounts of Na were present, this is a possible reason for these vibrations. The vibrational intensity also scaled with the surface area of the Al sources used. [37] also found these vibrations on their amorphous Al(OH)3 but ascribed them to the monodentate carbonate form. They performed experiments reacting amorphous aluminium hydroxide with sodium bicarbonate solutions of different concentrations. The wet environment could be the reason for the difference in carbonate orientation as later described by [39] for the adsorption of CO2 onto hydrotalcites in the presence of water vapour. Nevertheless, the different amounts of sodium bicarbonate solution used by [37] could give an indication towards the amount of carbonate adsorbed on the amorphous Al(OH)3. The absorbance spectrum of their sample prepared in 0.1 M sodium bicarbonate solution (corresponding to 0.26 g C/kg Al(OH)3 on the surface) fit the spectra obtained for Al(OH)3-SA and Al(OH)3-ACE best. If this was the amount of carbonate adsorbed on the surface of Al(OH)3-SA, this would make the starting materials consist of an additional 1.56% of carbonate. Thus equalling a total carbonate (not calcium carbonate) percentage of approximately 2% supplied with the raw materials during synthesis.

If this was the amount of carbonate present in the system initially, this would not, however, explain the large amounts of calcite remaining in WS1, T5 and CO3. The 20% content of calcite in CO3 is easily explained by the calcite fed to the system to prepare this carbonate-intercalated form. The relatively high percentage of calcite in WS1 can be reasoned through the large amount of starting material present to achieve the desired water-to solids ratio. In T5 the only difference between this

material and that synthesised with S3 is the reaction temperature (90 ◦C). A much higher pH was also recorded during this synthesis, which can easily be explained by the effect of temperature on the dissolution of the solid phases. There exist two possibilities for this increased amount in calcite. Either the high temperature reaction favours the formation of CaCO3 (and insolubility of this phase) instead of a carbonate intercalated LDH at this elevated temperature and pH—thus depicting a close approximation to the "true" amount of carbonate in the system—or there was simply more carbonate present to start with and it is also less soluble at these conditions. CaCO3 solubility is known to decrease at increasing temperatures and also at increasing pHs (at 25 ◦C and 1 bar) [40]. As all Ca(OH)2 and Al(OH)3 was taken from the same bottle, the last explanation thus seems unlikely, albeit not impossible.

While carbonate contamination definitely occurred (due to the variety of reasons mentioned), the phases formed in this work most likely consisted mainly of CaAl-OH-LDH as desired and as can be inferred by considering the difference in FTIR spectra of A3 and A1 in terms of the presence of adsorbed carbonate species on the Al sources and overall carbonate supplied to the system.

#### *3.2. Increasing Conversion to HC*

Considering the results presented in this work, it was evident that certain reaction parameters had a larger influence on the reaction outcome than others. Generally speaking, a higher purity HC was obtained by lowering the water-to-solids ratio, increasing the reaction time, having sufficient mixing, using an amorphous Al(OH)3 source with high surface area, using an adequate reaction temperature (80 ◦C for the highest purity within 3 h) and most surprisingly, by using a calcium-to-aluminium ratio stoichiometrically favouring katoite formation. It is possible that this occurred because of an overall slightly lower pH during synthesis. pH sampling results showed that the materials with the highest HC purity (WS1 and MR1, 73.48% and 84.91%, respectively) followed a similar pH behaviour during synthesis, starting with a high pH and ending on a lower pH. We expect that this occurred due to the conversion to the LDH phase. It is possible that the high initial pH facilitates better dissolution of the phases (especially Al(OH)3) for reaction. Unpublished results of the same molar-ratio-experiment (but conducted for a shorter period of time) indicated that the best purity was achieved by using the stoichiometric ratio for hydrocalumite formation, contradicting the results found here [41]. Time, especially in this regard, thus seems a very important factor.

During one of our previous studies, entirely different results were obtained for the temperature series by using a more crystalline Al(OH)3 source and a shorter reaction time [27]. While the Al(OH)3 used in that study was also sourced from ACE Chemicals, it showed to be a highly crystalline material, akin to the Merck source used in this work. Reaction kinetics were thus seemingly very different and the LDH-like phases formed in T1 and T2 remained undetected. The morphology, surface area, crystallinity and even crystal structure of the starting materials most likely remains one of the main contributing factors towards a successful synthesis in a short period of time. As seen by comparison of even our own results, the choice of starting materials can change the reaction outcome completely. Conversion to better crystallised and more morphologically even HCs seems to be achievable by using a higher crystallinity Al source. This is possibly observed due to lower reaction rates, a subsequent slower crystal growth and hence higher crystallinity and larger platelets formed.

In closing the discussion in this section, the presence of the particulate matter in LDHs formed and XRD identified phases deserve some discussion. As shown through the morphological evaluation of the Ca(OH)2 and Al(OH)3 phases used, it was difficult to discern these using SEM. Due to the high amount of portlandite remaining in many samples, it is highly likely that this particulate matter is a mixture of the two phases and that an increase in reaction time would increase the yield of the HC phase. As the cyrstal structure of the amorphous Al(OH)3 could not be determined, it is also possible that slightly too little aluminium was fed to the system, which could have an effect on the final outcome with sufficient reaction time. During XRD, most scans contained very, very small amounts of phases that remained unidentified. It is possible that small amounts of especially the Cl-intercalated

HC form exist in some more samples. However, due to shifted peaks and/or other effects, these phases could not be assigned. Further, any Rietveld refinement results (as mentioned in the text) were only applicable to the crystalline phases formed. The fraction of amorphous material present in each material remains undetermined.

#### *3.3. A Hint Towards the Reaction Mechanism of the Formation of HC in a Hydrothermal Process*

Finally, a short discussion on what could potentially be inferred regarding the reaction mechanism through which HC phases form using hydrothermal synthesis. The temperature experiments showed that, at least at low temperatures, an alternate LDH-like phase was formed. Some research suggests that this compound be a meta-stable phase that forms in cement (although it had not been found in Portland cement at the time) [35]. It is possible that this is a sort of pre-cursor to the HC phase, at least in low temperature systems. This seems likely considering observations in cement literature. Ref. [42] investigated the time dependent formation of calcium carboaluminate phases. They found that calcium hemicarboaluminate transforms into calcium monocarboaluminate given time and reaches high conversion after 100 days of reaction in the cement phase. Only the calcium monocarboaluminate phase is present in well-hydrated fully cured cement. Further, the calcium hemicarboaluminate form occurs early in the hydration process of cement, even if large quantities of calcium carbonate are present. Ref. [43] described how these calcium carboaluminate phases only really constitute a large fraction of the cement phase after about a day of reaction.

No study of the hydrothermal formation mechanism of HC exists (to our knowledge), especially at elevated temperatures. At elevated temperatures, these calcium hemi/monocarboaluminate phases could not be identified, but there seems to be a strong relationship to the formation/depletion of katoite during synthesis. In our previous results [27], these and those of [26], katoite was present to a large degree at a synthesis temperature of 90 °C. Katoite was also present in larger amounts at lower temperatures than 80 ◦C. The higher katoite content at 90 °C could indicate that this is the more stable phase past 80 °C. No elevated temperatures are required, however, to form katoite, which made up a large fraction of the phases formed even at 20 °C. The largest fraction of katoite was present at 40 °C and 90 °C, though, possibly indicating that HC is the favoured phase between these two temperatures. The time experiments also revealed a grea<sup>t</sup> amount regarding the progression of the phase contents with time. At 6 h, only approximately a quarter of the katoite present after 1 h was observed. This seems to be a strong suggestion that, given enough time, katoite is converted into HC during synthesis and could thus be a precursor to the HC phase.

### **4. Materials and Methods**

Chemically pure Ca(OH)2 and Al(OH)3 were sourced from ACE Chemicals (analytical reagen<sup>t</sup> grade) and Sigma Aldrich (SA) (reagent grade). Two other Al(OH)3 sources were used in the experiments. They were sourced from ACE Chemicals (chemical purity grade) and from Merck Chemicals (95% purity). Distilled and dissolved gas free (boiled prior to use and cooled/heated to the desired temperature under nitrogen flow) water was used in all experiments.

Experiments were performed using a bench top reactor set up as shown in Figure 25. The figure also depicts all variables that were investigated (time: t, temperature: T, mixing: M, molar ratio: MR, aluminium source: A and water-to-solids ratio: WS). An inert N2 environment was maintained in all experiments at slight over-pressure. The dry reactant powders were added to preheated water at the desired temperature under constant stirring. The mixture was kept in suspension by magnetic stirring at the desired speed and reacted for the desired time at the desired temperature. pH measurements were taken intermittently. The samples were filtered using vacuum filtration, immediately sealed and analysed as a wet paste within 0.5 h with XRD and 1 h with ATR-FTIR.

**Figure 25.** Setup used for the synthesis of hydrocalumite.

The synthesis of HC using Ca(OH)2, Al(OH)3 and H2O stands in competition with the formation of katoite Ca3Al2(OH)12 (a member of the hydrogrossular family) as shown Equation (2).

$$\begin{aligned} \text{\textit{aCaCa(OH)}}\_{2} + b\text{Al(OH)}\_{2} + c\text{H}\_{2}\text{O} &\longrightarrow & d\text{Ca}\_{2}\text{Al}\_{2}(\text{OH})\_{12} + c\text{H}\_{2}\text{O} \\ &\longrightarrow & \text{\textit{cCaAl}}\_{2}\text{Al}\_{2}(\text{OH})\_{12}\text{A}\_{2/\text{n}} \cdot \text{xH}\_{2}\text{O} + (\text{c} - \text{x})\text{H}\_{2}\text{O} \end{aligned} \tag{2}$$

All lower case letters represent stoichiometric coefficients. *A* denotes the desired intercalated anion. Table 7 shows the experimental conditions used for each experiment group.

**Table 7.** Experimental conditions used for each of the experiments performed. The experiment IDs and colour codes defined in the table were used for each experiment in the text. Purple: molar Ca:Al ratio, red: temperature, green: time, blue: water : solids ratio, brown: Al source, grey: mixing. Note: S1, S2, S3 = MR2, T4, t2, WS2, A1 and M2. \*Ca(OH)2 used in CO3 was partially substituted with CaCO3 as described in the text to achieve stoichiometric carbonate intercalation.


XRD measurements were performed on a Panalytical X'Pert PRO X-ray diffractometer in *θ* − *θ* configuration, using Fe filtered Co-K*α* radiation (1.789 ), an X'Celerator detector and variable divergence- and fixed receiving slits. The data were collected in the angular range of 5 ◦ ≤ 2*θ* ≤ 90 ◦ with a step size and time of 0.008 ◦ 2*θ* and 13 s, respectively. Phases were identified using X'Pert Highscore plus software. Molar fractions of phases present were determined using Rietveld refinement. The samples were analysed wet in order to minimise carbonate contamination and changes in the crystal structure due to drying effects. This process was chosen to closely simulate in-situ XRD.

ATR-FTIR spectra were obtained using a Perkin Elmer 100 Spectrophotometer. Samples were pressed in place with a force arm. Spectra were obtained in the range of 550–4000 cm<sup>−</sup><sup>1</sup> each with 32 scans at a resolution of 2 cm<sup>−</sup>1.

SEM micrographs were obtained using a Zeiss Ultra PLUS FEG SEM at 1 keV. Samples were coated with 1.4 nm carbon prior to analysis. The LDH samples were dried in a desiccator prior to study with SEM.

The BET surface areas of the materials were determined using isotherms recorded at 77.35 K with a Micromeritics TriStar II 3020. The samples were degassed at 80 ◦C for 1.5 h prior to the analysis.
