Synthesis of wollastonite from AlF3-rich silica gel and its hardening in the CO2 atmosphere

This work combines some aspects of eco-friendliness: consumption of toxic waste, cutback of energy consumption during the synthesis of the binding material, reduction of CO2 emission by using less CaCO3 in the raw meal, and consumption of carbon dioxide. In the study, the kinetics of two-step synthesis of wollastonite from CaO and AlF3 production waste, namely, silica gel, its carbonisation process and the mechanical properties of obtained samples were investigated. According to XRD and DSC data, the optimal temperature in the mixture with CaO/(Al2O3 + SiO2) = 1 for the hydrothermal synthesis of the wollastonite precursors is 130 °C: F−–containing compounds were bound into katoite and cuspidine, and portlandite reacted completely within 8 h. The optimal temperature for wollastonite formation is 900 °C, but fluormayenite, cuspidine, and the traces of larnite form as well. During the curing in the CO2 atmosphere, wollastonite and larnite reacted completely and formed calcite, vaterite, and amorphous CaCO3. Cuspidine also participates in the carbonisation process and, in addition to amorphous SiO2, it releases fluorite, which contributes to the total compressive strength of the products. The values of the compressive strength (10–15 MPa) in the wollastonite-sand samples match the requirements for the belite and special low-heat cements.

wollastonite and rankinite are non-hydraulic, they can be activated by CO 2 with the presence of humidity 21,22 . Wollastonite is more preferred for this process because of its better solubility in water, which means better reaction with CO 2 and lower carbon dioxide emission caused by CaO production from CaCO 3 23 . A newly formed matrix after the carbonation is composed of calcite CaCO 3 and amorphous SiO 2 provides similar mechanical properties of hydrated OPC where matrix consists mainly of calcium silicate hydrates and calcium hydroxide 21,24 .
Bukowski and Berger 24 investigated mechanical properties of a binder which contained 50% sand and 50% wollastonite. They determined that the binder reached a compressive strength of 35 MPa within 24 h in 1 bar of CO 2 using dynamic 1.4 l/min gas flow system (from the formation of the sample to the end of carbonisation). Other researchers 21 reached a compressive strength of 70 MPa when the samples contained only wollastonite and were carbonised for 65 h at 60 °C in 100% carbon dioxide. This proves that wollastonite can be an alternative to OPC. Another advantage of wollastonite over ordinary Portland cement is that wollastonite can be synthesised from the same raw materials as the cement clinker and at 250-500 °C lower temperature 19,20,25 .
The aim of this work was to investigate the kinetics of a two-step synthesis of wollastonite from CaO and silica-gel waste, its carbonisation process, and the mechanical properties of the obtained samples. This work combines four aspects of eco-friendliness: (1) consumption of toxic waste, (2) cutback of energy consumption during the synthesis of the binding material, (3) reduction of carbon dioxide emission by using less CaCO 3 in the raw meal, and (4) consumption of carbon dioxide.

Materials and Methods
Materials. AlF 3 production by-product -silica gel from chemical plant SC Lifosa (Lithuania), dried at room temperature until constant mass for 2 weeks was used. The main elements were 39.86 wt% Si, 5.37 wt% Al (determined using X-ray fluorescence analyser Bruker X-ray S8 Tiger WD (Germany)), and 8.76 wt% F − (determined potentiometrically using Mettler Toledo titrator T 70 (USA) with F − selective electrode), mass losses -4.1 wt% at 105 °C and 20.0 wt% at 1000 °C. The silica gel waste was milled for 2.5 min at 950 rpm using a planetary mill Fritsch Pulverisette 9 (Germany) until specific surface area S a = 1537 m 2 /kg by Cilas LD 1090 (France) granulometer and density ρ = 2354 kg/m 3 (gas pycnometer Quantachrome Instruments Ultrapyc 1200e, USA). Calcium oxide was obtained by calcining calcium hydroxide (≥96%, Honeywell, Germany) at 550 °C for 1 h and milling for 0.5 min at 950 rpm (S a = 2076 m 2 /kg, ρ = 2837 kg/m 3 , free CaO -94.22%). CEN standard sand (DIN EN 196-1 (ISO 679)) was used to press wollastonite-sand samples. Figure 1 shows that the silica-gel waste contains several crystalline compounds: aluminium hydroxyfluoride (Al 2 (OH) 3 25,30, and 35% calcined product were prepared. They were homogenised for 1 h at 49 rpm in Turbula Type T2F and mixed with water to obtain water/binding material ratio W/C = 0.35. The samples of 36 × 36 mm were pressed in a cylindrical frame under the pressure of 10, 12.5 and 15 kN at 1 kN/s speed and 20 s exposure at the maximum pressure in a Test Form Mega 10-400-50 (Germany) hydraulic press. The samples of humidified binding material without sand were also formed to determine the changes in the mineral composition during the curing process. The curing in the CO 2 environment was carried out in Parr Instruments 4600 autoclave at 45 °C and 15 bar for 8-24 h. Before curing, the autoclave was purged twice from atmospheric air by letting CO 2 up to 2 bar. The compressive strength was determined right after the curing using a Test Form Mega 10-400-50 hydraulic press at a loading rate of 1.5 kN/s. For the statistical accuracy, at least three samples of each batch were tested and the average value of a compressive strength was used.
The XRD analysis of dry powders was performed using D8 Advance diffractometer (Bruker AXS, Germany) operating at the tube voltage of 40 kV and tube current of 40 mA. The X-ray beam was filtered with 0.02 mm Ni filter to select the CuK α wavelength. The diffraction patterns were recorded in a Bragg-Brentano geometry using a fast counting detector Bruker LynxEye based on a silicon strip technology. The specimens were scanned over the range 2θ = 3-70 at a scanning speed of 6 min −1 using a coupled two theta/theta scan type 26 .
The simultaneous thermal analysis (STA) was employed for the measuring phase transformation of the synthesised products at a heating rate of 10 °C/min, the temperature ranged from 30 to 945 °C in the nitrogen atmosphere, the mass of the sample was 20 mg. The test was carried out on a Linseis PT 1000 (Germany) instrument. The ceramic sample handlers and crucibles of Pt/10 wt%Rh were used 26 .
The amount of free CaO was determined according to standard ASTM C114-11b. The quantitative analysis (QXRD) of the phases in the samples was carried out with the Rietveld method (Topas Software) by using XRD data.
BET analysis was performed by surface area analyser "Autosorb iQ Station 1" (Quantachrome Instruments, USA). The specific surface area of the hydrothermal synthesis products was calculated by the BET Eq. (1) using the data of the lower part of N 2 adsorption isotherm (0.05 < p Since all portlandite was bound completely during the synthesis at 130 °C within 8 h and 1.13 nm tobermorite started to crystallise after 24 h, it was expected that this temperature would be more favourable rather than 95 °C for the hydrothermal synthesis of the precursors for wollastonite.
The admixtures in silica gel waste, i.e. aluminium fluoride and hydroxide, were bound into chemically inert compounds: katoite and cuspidine. The latter one is stable in the temperature range 20-1410 °C.
DSC results (Fig. 3) supplemented and confirmed the XRD data: after 0 h at 130 °C ( Fig. 3 curve 1), the first endothermal effect in the DSC curve (100-130 °C) can be assigned to the loss of crystallisation water from calcium silicate hydrates, the second endothermal effect (320-325 °C) is typical to the decomposition of katoite, and the third endothermal effect (444-446 °C) is common to the dehydration of portlandite. The exothermal effect (832-868 °C) is typical to recrystallization of semi-crystalline C-S-H into wollastonite. The endothermal effect (662-697 °C) belongs to the decarbonisation of calcite. Both DSC and XRD showed that portlandite was bound within 8 h (Fig. 3 curve 3) because there was no endothermal effect at ~445 °C and no peak at 0.2627 nm in the XRD patterns.
The other important difference between the products shows the exothermal effect (830-870 °C) in the DSC curves. With a longer isothermal treatment duration, the effect migrates from the higher (850-870 °C after 0-12 h (Fig. 3, curves 1-4)) to the lower (840-832 °C after 24-48 h (Fig. 3 curves 5-7)) temperature and its enthalpy increases from ~20 to 86 J/g. This can be explained by different types of semi-crystalline C-S-H in the synthesis www.nature.com/scientificreports www.nature.com/scientificreports/ products: a small and "blunt" effect in samples of 0-12 h is common to C-S-H(II) 29 , while a tall and "sharp" one in the samples of 24-48 h belongs to C-S-H(I) 16,29 which has lower basicity and is closer to the stoichiometric composition of the primary mixture.
The samples of hydrothermal synthesis at 130 °C (8, 24 and 48 h) were characterised by BET method (Table 1). It was determined that the surface area depended on the chemical/mineral composition of the formed compounds during hydrothermal synthesis. As it was mentioned, after 8-48 h of synthesis semi-crystalline C-S-H (II) (Fig. 3, curve 3, "blunt" exothermal effect at 868 °C), C-S-H (I) (Fig. 3, curve 5, "sharp" exothermal effect at 840 °C) and 1.13 nm tobermorite (Fig. 2, curve 4) were identified. For this reason S BET values increased from 69.3 (8 h) to 90.8 m 2 /g (24 h). During 48 h of synthesis more 1.13 nm tobermorite (Fig. 2, curve 5) formed and its crystals were larger, therefore S BET values of the sample decreased to 82.7 m 2 /g.
The products of hydrothermal syntheses were calcined in the temperature range of 850-1050 °C for 1 h to investigate the formation of wollastonite and the changes of its crystallinity.
Even though the amount of portlandite in the products of hydrothermal synthesis at 95 °C decreased from 23.31% (0 h) to 4.73% (48 h), the intensity of wollastonite main peak (PDF No. 42-0550, d = 0.2980 nm) in the XRD patterns of calcined products (Fig. 4a) was in the range of 0-600 cps and did not depend neither on the hydrothermal synthesis duration (from 0 to 48 h) nor the calcination temperature (from 850 to 1050 °C). The amount of wollastonite in the calcined products of the hydrothermal synthesis at 130 °C (Fig. 4b) was much larger: the intensity range of the main peak of wollastonite was from 450 to 1750 cps. The products that contained portlandite (0-4 h) showed similar intensity (450-600 cps) of the wollastonite main peak to those of 95 °C synthesis (Fig. 5, curves 1). The products without portlandite had more intense peaks (600-1750 cps) of wollastonite and depended on both the synthesis duration and the calcination temperature. The latter dependence is more evident for the products of 12 h and longer syntheses (Fig. 5, curves 2 and 3). According to the intensity of the main peak of wollastonite in the XRD patterns, it can be concluded that portlandite had to react completely during the synthesis. As a result, after calcination, the products without portlandite contained more wollastonite, while the products with portlandite yielded this mineral less and contained more larnite Ca 2 SiO 4 (PDF No. 33-0302; d = 0.2780; 0.2745; 0.2788; 0.2735; 0.2620 nm) -a compound where CaO/SiO 2 molar ratio is twice higher than of the primary mixture.
According to the XRD data of the calcined samples, in addition to wollastonite and larnite, cuspidine and fluormayenite Ca 12 Al 14 O 32 F 2 (PDF No. 87-2492; d = 0.4887; 0.2677; 0.2992; 0.2443; 0.2185 nm) were identified. The latter crystalline compounds are non-toxic, chemically and thermally stable up to 1000 °C.
According to XRD and DSC data, the mineralogical composition variation of the products during the hydrothermal synthesis at 130 °C and calcination at 850-1050 °C was summarised in Fig. 6. curing in the co 2 atmosphere. The product of the hydrothermal synthesis at 130 °C for 8 h (Fig. 2, curve 3) and calcining at 900 °C for 1 h (Fig. 7a, curve 1) was used for the production of wollastonite-based mortar samples and their potential for hardening in the CO 2 environment was determined. The mineral composition of the binding material is given in Table 2.
These conditions were chosen from the economic point of view because 8 h was the shortest period of time for the hydrothermal synthesis to bind portlandite completely and the calcination at 900 °C for 1 h is sufficient to produce wollastonite.
It was determined that the compressive strength depended mostly on the amount of binding material in the samples. By increasing the binding material from 20 to 35 wt%, the strength boosted more than twice: from 7.34 to 15.68 MPa (Fig. 8a). With the increment of compaction pressure, when the binder/sand ratio was 1:3 (amount of binder -25 wt%), the compressive strength went up by a third: from 9.51 (10 kN) to 12.15 (15 kN) MPa (Fig. 8b). The curing duration had a low impact on the compressive strength, as the increase of compressive strength was only 14%: from 10.64 (4 h) to 12.13 (24 h) MPa (Fig. 8c).
It was determined that the main compounds in the binding material before curing (Fig. 7a, curve 1) were wollastonite CaSiO 3 (Fig. 7a, curve 2), the main components were two polymorphs of calcium carbonate CaCO 3     www.nature.com/scientificreports www.nature.com/scientificreports/ According to the STA data (Fig. 7b), the first endothermal effect in the DSC curve (107 °C) is typical to the loss of adsorbed water, and the second double endothermal effect (640-760 °C) is typical to the decomposition of carbonates: the first one (645 °C) -decarbonisation of vaterite and the second one (756 °C) -calcite. This suggests that very low onset temperature (~500 °C) may be related to the decomposition of amorphous calcium carbonate so these results are directly in line with previous findings 22,30 . To describe the origin of the exothermal effect at 872 °C, the sample was calcined at 820 and 900 °C. According to the XRD patterns, the effect belongs to the formation of wollastonite from calcium oxide, a product of calcination of CaCO 3 , and amorphous SiO 2 , which was released during carbonation process.   Table 2. Mineral composition of the binding material according to QXRD analysis.  investigation of co 2 curing of cuspidine. According to the XRD data, the products of calcination along with wollastonite always contained cuspidine Ca 4 Si 2 O 7 F 2 which forms from admixtures of silica-gel waste and calcium oxide. However, the literature sources lack data describing the role of cuspidine in the CO 2 -curing process. To investigate what chemical changes the mineral undergoes during the curing in the CO 2 atmosphere, cuspidine was synthesised from reagent grade materials using the hydrothermal synthesis at 130 °C for 24 h and calcination at 900 °C for 1 h) (Fig. 9a, curve 1). Sample preparation (water/solid ratio = 0.35, compaction pressure -12.5 kN) and curing (for 24 h in 15 bar CO 2 at 45 °C) conditions were the same as in the wollastonite case. According to the XRD data, the cured sample no longer contained cuspidine, however, there were fluorite and two types of calcium carbonate -calcite and vaterite, instead (Fig. 9a, curve 2). The STA results confirmed the XRD data (Fig. 9b). The DSC curve showed four endothermal effects: the first one at 103 °C is typical to loss of adsorbed water, and the other three occurred at 609, 692 and 739 °C. To determine the latter effects, the samples of cured cuspidine were calcined at corresponding temperatures. According to the XRD data (Fig. 9a, curve 3), the effect at 609 °C means the decomposition of vaterite and the effects at 692 www.nature.com/scientificreports www.nature.com/scientificreports/ and 739 °C -double decomposition of calcite (Fig. 9a, curves 4 and 5). The very low onset temperature (~500 °C) may be related to the decomposition of amorphous calcium carbonate similarly to 22,30 . The exothermal effect at 838 °C can be attributed to the formation of calcium silicates. In conclusion, the processes during the calcination of cured wollastonite and cuspidine are similar.
According to the obtained data, the chemical equation of carbonisation of cuspidine may be written by formula (5): Therefore, cuspidine, analogically to wollastonite, also reacts with carbon dioxide and forms calcite, which contributes to the total compressive strength of the cured products. However, in addition to calcium carbonate, inert compounds (in this system 33 ), namely, amorphous silica and fluorite are released. According to the Mohs' scale, the latter mineral is harder than calcite 34 .
To conclude the investigation of cuspidine curing in the elevated CO 2 pressure, it should be pointed out that silica-gel waste which contains fluoride ions can be used as a source of SiO 2 for the wollastonite synthesis and as a binder for CO 2 -cured products, because the F − ions are bound into cuspidine which participates in the carbonation process in the same way as calcium silicates. conclusions 1. The study showed that AlF 3 production waste, silica gel, is suitable for the wollastonite binder synthesis and, subsequently, could be used for the production of carbonated construction materials. This process is an environmentally-friendly approach reducing carbon footprint as well as the consumption of toxic waste. Furthermore, this could be an economically viable option as wollastonite binder production requires 550 °C lower temperature than Ordinary Portland cement. 2. It was determined that, in the CaO/SiO 2 mixture, the optimal temperature for the hydrothermal synthesis of the precursors of wollastonite is 130 °C: the admixtures in silica gel (aluminium fluoride and hydroxide) were bound into chemically non-aggressive katoite and cuspidine and portlandite reacted completely within 8 h. 3. The precursors for the sintering of wollastonite must be Ca(OH) 2 -free, because the products with portlandite after the calcination contained less wollastonite and more larnite. The intensity of wollastonite peaks from precursors without portlandite was higher (600-1750 cps) and it augments with prolonging the hydrothermal synthesis duration and increment of calcination temperature. Fluormayenite, cuspidine and the traces of larnite forms as well. 4. During the curing calcined products in the CO 2 atmosphere, wollastonite, larnite, and cuspidine reacted completely and formed several polymorphs of calcium carbonate: calcite, vaterite and amorphous CaCO 3 . It was determined that, during cuspidine carbonisation, in addition to amorphous silica, it releases fluorite. Fluormayenite remained inert during the curing process. 5. Wollastonite-sand samples gain relatively high compressive strength (10-15 MPa) during curing in the CO 2 atmosphere. It was determined that with increment of the amount of binding material, compaction pressure, as well as the exposure duration in the CO 2 environment, sample compressive strength is highly increased. These values meet the requirements for the belite and special low-heat cements.