Facilitated phase transformation of CaAl₂Si₂O₈ by kaolinite-derived aluminosilicate nanoscrolls

Abstract

Aluminosilicate nanoscrolls, derived from two-dimensional layered materials, were studied for their role in facilitating phase transformations in solid-state materials. This study focused on kaolinite (Al₂Si₂O₅(OH)₄), a layered clay mineral with platy morphology, as a nanoscroll precursor. Rolling-up of kaolinite layers occurs via an inherently incomplete intercalation, producing incompletely rolled-up kaolinite containing nanoscrolls. This morphology serves as an ideal model to investigate the effect of halloysite—a naturally occurring nanoscroll mineral with the same layer structure and composition as kaolinite—on kaolinite solid-state reactions. Notably, kaolin clay, widely used as a raw material in industrial inorganic materials, primarily consists of kaolinite but often contains by-products, including halloysite. In this study, purified Kanpaku kaolinite (Si/Al = 1.01) was mixed in a 1 : 1 molar with calcium carbonate (CaCO₃) and calcined to evaluate phase transformation into metastable and stable CaAl₂Si₂O₈. At 900 °C, where the metastable phase predominates, X-ray diffraction pattern revealed a weak reflection corresponding to anorthite—the stable phase of CaAl₂Si₂O₈—in products containing nanoscrolls. Although this reflection is still weak, it is noticeably stronger than that observed in the specimen prepared using only kaolinite with platy morphology. This result indicates nanoscrolls in platy particles accelerate the transformation from metastable to the stable phases. The results in this study highlight the role of kaolinite-derived nanoscroll morphology in influencing solid-state reactions of kaolinite.

---

Introduction

Purification of chemicals is essential for the syntheses of materials and the elucidation of their mechanisms. Naturally-abundant clay minerals are promising raw materials for industrial ceramic materials. Notably, by-products, particle size, and crystallinity of clay minerals are varied depending on their geographical origin. Additionally, a portion of substitution of elements, such as Al³⁺ to Fe³⁺, can occur in crystal structures of clay minerals.¹ Therefore, the use of clay minerals for preparing ceramic materials requires careful consideration. Moreover, purification of clay minerals is helpful for revealing the mechanisms of their reactions, such as solid-state reactions.

Kaolinite, a layered aluminosilicate with the formula of Al₂Si₂O₅(OH)₄, displays platy particles comprising the stacked aluminosilicate layers, each of which consists of an AlO₂(OH)₄ layer and a SiO₄ layer.^1–4 Kaolinite is classified into 1 : 1 type layered aluminosilicate.^1–4 Additionally, kaolinite features intercalation of small polar molecules and organic salts to form intercalation compounds.^2–6 A kaolinite-urea intercalation compound was used for raw materials of potties from ancient China.⁵ Moreover, solid-state reactions of kaolinite with carbonates are useful for preparing ceramic materials, such as kalsilite (KAlSiO₄),^7–9 cordierite (Mg₂Al₄Si₅O₁₈),^10,11 anorthite (CaAl₂Si₂O₈),^12–19 celsian, and hexacelsian (BaAl₂Si₂O₈: the former is stable phase, and the latter is metastable phase).^18–23 Kaolinite is a specific clay mineral of kaolin.^1–4 Kaolin contains by-products, and the composition differs from the ideal ones of kaolinite.^{1–4,15,20–22} Despite the degree of kaolin by-products and kaolinite purification procedures, kaolinite solid-state reactions are generally conducted in the presence of by-products. Some reports purified kaolinite to the extent that quartz, a common by-product, was absent.^{15–18,24,25} Notably, the purification procedures used in these reports^{15–18,24,25} are highly specialized, based on the optimization of intercalation chemistry, and are not comparable to conventional purification methods. By contrast, many reports used kaolin which contains relatively large amounts of by-products. Halloysite, whose formula and layer structure are the same as those of kaolinite but has a nanoscroll morphology with a SiO₄ exterior and an AlO₂(OH)₄ interior, is a component of kaolin.^1,3,4 Additionally, halloysite can be the specific mineral of the kaolin.^20–22 Such kaolin, e.g., New Zealand kaolin, can be used for preparing industrial ceramics.^20–22 Thus, effect of halloysite on the solid-state reaction of kaolinite warrants investigation to elucidate reaction mechanisms and the properties of solid-state products. To our knowledge, there are no reports of purification and precise hydrothermal syntheses of halloysite.

Notably, intercalation of specific alkylammonium salts into kaolinite leads to exfoliation and rolling-up of the layers to form kaolinite nanoscrolls that is a similar to halloysite.^{1,3,4,26–32} Although rolling-up of kaolinite layers were achieved by grinding and deintercalation procedures,^5,33,34 the highest rate of rolling-up was obtained by intercalation of hexadecyltrimethylammonium chloride (C16TAC) between the layers of Georgia kaolinite that shows relatively higher intercalation capability.²⁶ Notably, kaolinite intercalation capability is influenced by geographical origin, which affects factors such as particle size and crystallinity.^2,15 Despite the highest intercalation capability of Georgia kaolinite, the by-product and the brown coloration due to Fe substituted for Al in kaolinite layers cannot be eliminated by the previously-reported purification procedure.¹⁵ This procedure offers purified-kaolinite without quartz when Kanpaku kaolinite, an originally-white kaolinite, was used. The brown portion was successfully extracted from the white kaolinite. Although the intercalation capability of Kanpaku kaolinite is lower than that of Geogia kaolinite, intercalation of C16TAC into Kanpaku kaolinite is a promising method for obtaining purified kaolinite with some nanoscrolls – i.e., incompletely rolled-up kaolinite, which is an ideal form for combining with purified halloysite.²⁷ Consequently, the incompletely rolled-up kaolinite is the best candidate for studying the effect of halloysite on the solid-state reaction of kaolinite. To our knowledge, the preparation and investigation of synthetic kaolinite and its intercalation behavior have not been reported since 2000s.^35–38

Herein, special attention has been paid to the solid-state reaction of incompletely rolled-up kaolinite with calcium carbonate (CaCO₃). The solid-state reaction of kaolinite with CaCO₃ with the same molar ratio is a suitable model reaction because the phase transformation of metastable to stable phases of CaAl₂Si₂O₈ occurs at relatively low temperature.^13,15 Notably, previous studies have clearly shown that the stable phase of CaAl₂Si₂O₈ is anorthite.^13,15 This solid-state reaction proceeds as follows: Al₂Si₂O₅(OH)₄ + CaCO₃ → CaAl₂Si₂O₈ + 2H₂O + CO₂.¹³ Therefore, the degree of phase transformation is regarded as an indicator of the reaction progress. Additionally, gehlenite (Ca₂Al₂SiO₇) was generated when the grinding degree of kaolinite and CaCO₃ with the stoichiometric composition of CaAl₂Si₂O₈ was not enough.¹³ Thus, the absence of gehlenite indicates the that the composition of the calcined product reflects the ratio of the component included in the raw materials. Notably, intercalation of C16TAC into kaolinite to obtain kaolinite nanoscrolls requires methoxy-modified kaolinite (MeO-Kaol), an organically-modified kaolinite in which a portion of hydroxyl groups is substituted to methoxy groups,^39,40 as an intermediate.^26,27 Therefore, in the present study, the solid-state reactions of CaCO₃ with incompletely rolled-up kaolinite and MeO-Kaol are compared. Note also that purified MeO-Kaol were obtained by the aforementioned purification procedure (see the previous paragraph). When the calcination of ground raw materials was conducted at temperature to form metastable CaAl₂Si₂O₈, the rate of the transformation into the stable phase, anorthite, of the ground raw material containing incompletely rolled-up kaolinite is slightly higher than that containing MeO-Kaol. The plausible mechanisms are discussed, including the attempted purification of halloysite, the preparation of methoxy-modified halloysite (MeO-Hal) that was mixed with pristine kaolinite, and the solid-state reaction of halloysite with CaCO₃. Additionally, the solid-state reaction of MeO-Kaol and CaCO₃ with the addition of amorphous silicon dioxide (SiO₂) are considered beneficial. Notably, kaolins containing kaolinite and halloysite, respectively, were compared as the main components in cordierite systems;⁴¹ however, the samples contained additional components that were not present in trace amounts. Other comparable studies follow a similar trend.

Experimental

Characterization

The expansion of kaolinite or halloysite layers, the degree of stacking order of kaolinite or halloysite specimens, crystalline phases of solid-state products or by-products of kaolinite or halloysite specimens, and the mixture of pristine kaolinite with MeO-Hal were characterized by the powder X-ray diffraction (XRD) using a diffractometer (Empyrean, PANalyical) operated at 40 mA and 45 kV with monochromated Cu Kα radiation. The step size and scan time were 0.01° (2θ) and 1.0 s, respectively. Regarding the expansion of kaolinite layers, the shift of the diffraction line associated with the basal spacing, observed at lower 2θ region (below 15°), is important. Reflections due to lateral atomic arrangements, which can be used to discuss the degree of stacking order,⁴² appear below 30°. Reflections above 50° are not specific, and some are difficult to interpret due to overlapping.⁴³ Meanwhile, the XRD patterns of pristine kaolinite used in this study, measured over the 1.5–70° range, is shown in Fig. S1† for reference. The percentage of each crystalline phase in the calcined specimens was tentatively estimated by Rietveld refinement using HighScorePlus software. Specimen morphologies were examined using field-emission scanning electron microscopy (FE-SEM: SU8000, HITACHI) at an acceleration voltage of 2.0 kV, with an electron beam energy of 10 µV and a working distance of 8 cm. Prior to observation, the samples were sputter-coated with a 3 nm osmium. For simplicity, FE-SEM image is hereafter referred to as SEM image. The presence of anorthite in the specimens was characterized by Raman spectroscopy (JASCO, NRS-7100) using an excitation wavelength of 532 nm and a resolution of 0.05 cm⁻¹. Thermogravimetric (TG) and differential thermal analysis (DTA) curves were recorded on a Rigaku TG 8120 or a NETZSCH TG-DTA at a heating rate of 10 °C min⁻¹ under air. The surface areas of the samples were determined from the nitrogen (N₂) adsorption isotherms⁴⁴ using Brunauer–Emmett–Teller (BET) method.⁴⁵ The N₂ adsorption/desorption isotherms were measured at −196 °C using a Belsorp MINI instrument (MicrotracBEL Inc.). Before the measurement, the samples were dried at 300 °C under reduced pressure. The pore size distribution was determined from the adsorption isotherm by the Barret–Joyner–Halenda (BJH) method.⁴⁶ The solid-state ²⁹Si and ²⁷Al nuclear magnetic resonance spectra were recorded on a JEOL JNM CMX-400 spectrometer at 104.17 and 179.42 MHz, respectively. The magic angle spinning (MAS) technique was used with a pulse delay of 60 and 1 s for ²⁹Si and ²⁷Al, respectively. The samples were placed into a 5 mm-zirconia rotor with a spinning rate of 8 kHz. The ²⁹Si and ²⁷Al chemical shifts were externally referenced to polydimethylsiloxane at −33.8 ppm and an aluminum nitrate aqueous solution at 0 ppm, respectively. Elemental analyses of kaolinite and purified MeO-Kaol were conducted by X-ray fluorescence (EDX-700HS, Shimadzu), followed by more precise quantification using inductively coupled plasma (ICP) optical emission spectrometer (SPS-3100, Seiko instruments Inc.) and carbon/sulfur analysis (CS844, LECO), which were requested to Tokai Technology Center. Before the measurements, the kaolinite specimens were thermally decomposed in various acids including hydrofluoric acid and boric acid or sulfuric acid, before being dissolved in hydrochloric acid. The Si/Al molar ratios of Kanpaku kaolinite and purified MeO-Kaol were estimated at 1.03 and 1.01, respectively. Furthermore, the pristine kaolinite was found to contain Na (0.46 mol%), Ca (0.21 mol%), K (0.16 mol%), Mg (0.18 mol%), Fe (0.088 mol%), Sr (0.46 mol%), Ti (0.48 mol%), Ba (0.43 mol%), and S (1.1 mol%), while the purified MeO-Kaol contains Na (0.30 mol%), Ca (0.19 mol%), K (0.16 mol%), Mg (0.23 mol%), Fe (0.088 mol%), Sr (0.44 mol%), Ba (0.088 mol%), Ti (0.38 mol%), and S (0.75 mol%), all relatively to the kaolinite composition. These results indicate that the pristine kaolinite inherently contains trace impurities, some of which are reduced by the purification procedure.

Materials

The kaolinite used in this study was a reference clay sample of the Clay Science Society of Japan (JSCC-1101c: Kanpaku kaolinite, as mentioned in the Introduction). The halloysite used in this study was obtained from Sigma-Aldrich (Dragonite). The composition of this halloysite was described elsewhere.⁴⁷N-Methylformamide (NMF), methanol (MeOH), 1 mol L⁻¹ hydrochloric acid (HCl), and ethanol were obtained from Wako Pure Chemical. C16TAC was obtained from Tokyo Chemical Industry. CaCO₃ was obtained from Hayashi Pure Chemical. Amorphous SiO₂ was obtained from Kojundo Chem. Lab.

Sample preparation

Preparation of incompletely rolled-up kaolinite. For simplicity, incompletely rolled-up kaolinite is denoted herein as NS-MeO-Kaol. NS-MeO-Kaol was prepared following the previous study.^15,27 Notably, following the first report of kaolinite layer rolling in 1963,⁵ such methods have been widely reported.^26–34 Among them, Kuroda and co-workers were the first to report a one-step and efficient rolling-up of kaolinite layers,²⁶ whose reproducibility was later confirmed using kaolinite from the same production area.²⁷ Kanpaku kaolinite was soaked in 1 mol L⁻¹ HCl for a day, then centrifuged and treated with 91 vol% NMF aqueous solution for 3 days to form a kaolinite-NMF intercalation compound (Kaol-NMF).⁶ Afterward, Kaol-NMF was subjected to repeated washing with fresh methanol for 7 days to obtain MeO-Kaol.³⁹ The brown portion at the bottom of the centrifuge tube was removed to purify MeO-Kaol. Finally, the purified specimen was stirred in a 1 mol L⁻¹ methanolic solution of C16TAC, then centrifuged and washed with excess ethanol to remove C16TAC—i.e., deintercalation—and to obtain NS-MeO-Kaol.

Attempted purification of halloysite. Purification of halloysite was attempted using the method applied for preparing MeO-Kaol (see the previous paragraph). The product obtained was MeO-Hal, with the top and bottom portions of the centrifuged specimen being white and brown, respectively. Additionally, another purification method, previously applied for kaolinite,²⁵ was attempted for pristine halloysite. After preparing a halloysite-NMF intercalation compound (Hal-NMF), the specimen underwent repeated washing with deionized water. After several washes, the centrifuged specimen displayed white top and brown bottom portions. Notably, MeO-Hal and Hal-NMF were reported previously.^48,49

Solid-state reactions of kaolinite or halloysite specimens with CaCO₃. The experimental condition mostly followed those of the previous reports^15–17 for comparison purposes. To prepare the mixture, the ground raw material – kaolinite or halloysite specimens (516 mg) and CaCO₃ (200 mg) – were wet-milled in MeOH (8 mL) using a planetary ball mill at 250 rpm for 12 h in a resin vessel (12.5 mL) with 120 alumina balls (3 mm in diameter). The resultant solids were centrifuged at 5000 rpm for 5 min and dried at 100 °C for 1 h. After drying, the obtained powders were passed through a 150 µm-mesh to achieve visual uniformity. The kaolinite or halloysite specimens and CaCO₃ were used in a molar ratio of 1 : 1, corresponding to the stoichiometric composition of CaAl₂Si₂O₈. For the solid-state reaction of purified MeO-Kaol and CaCO₃ with the addition of amorphous SiO₂, the latter was added to achieve a molar ratio of Si/Al was set to be 1.05, a value slightly larger than 1.03 of Kanpaku kaolinite. The ground raw materials were calcined at 750, 800, 850, 900, or 1000 °C for 12 or 24 h, respectively. Notably, all the calcined specimens were obtained in powder form as polycrystalline materials, which is generally observed in solid-state reactions. The specimens calcined for 12 h, prepared using purified MeO-Kaol, NS-MeO-Kaol, halloysite, and amorphous SiO₂-added and purified MeO-Kaol, were denoted herein as CAS-MeO-X, CAS-NS-X, CAS-Hal-X, and CAS-Si-MeO-X, respectively (X represents calcination temperature). The specimens calcined for 24 h were denoted herein as CAS-MeO-900-24 and CAS-NS-900-24. For comparison purposes, amorphous silica was calcined at 900 °C.

Results and discussion

Fig. 1 presents XRD patterns of kaolinite specimens. For purified MeO-Kaol (Fig. 1b), a diffraction line with a d value of 0.86 nm, attributed to the basal spacing of MeO-Kaol,^39,40 is observed alongside a diffraction line with a d value of 0.72 nm, attributed to the basal spacing of pristine kaolinite.^2–6 This result is consistent with common feature that kaolinite intercalation reactions do not fully proceed.^2–6 Additionally, the quartz reflection¹⁴ observed in the brown portion of the MeO-Kaol specimen after purification (Fig. 1c) is absent in the purified MeO-Kaol. In NS-MeO-Kaol (Fig. 1d), the intensity of the 0.86 nm diffraction decreases compared to purified MeO-Kaol, indicating disruption of the kaolinite stacking order due to rolling-up, as previously reported.^26–28

Fig. 1 XRD patterns of (a) pristine kaolinite, (b) purified MeO-Kaol, (c) the brown portion of the MeO-Kaol after purification, and (d) NS-MeO-Kaol. Open triangles indicate quartz reflections, and the cross mark represents an unknow phase.

Fig. 2 shows SEM images of the kaolinite specimens. Hexagonal platy particles, typical of kaolinite, are observed in both pristine kaolinite and purified MeO-Kaol.^3–5 (Fig. 2a and b). In contrast, NS-MeO-Kaol shows some nanoscrolls, both aggregated and independent, which are infrequently present and mostly located at the edges or surfaces of the platy particles (Fig. 2c), a result consistent with those reported previously.²⁷

Fig. 2 SEM images of (a) pristine kaolinite, (b) purified MeO-Kaol, (c) NS-MeO-Kaol, and (d) an enlarged view of the yellow rectangle in (c). The yellow arrows and the dotted line in (d) indicate nanoscrolls and their aggregate.

Based on the XRD patterns, SEM images, the hysteresis curve in the N₂ adsorption/desorption isotherm, the 11 nm peak in the BJH pore distribution (Fig. 1, 2, S2b, and S3a†), and the Si/Al ratio of purified MeO-Kaol with 1.01 (see general information), the preparation of NS-MeO-Kaol—a purified kaolinite containing some nanoscrolls—was successful. Previous studies have noted that the lateral size of the SiO₄ layer is slightly larger than that of the AlO₂(OH)₄ layer due to their lattice parameters,^30–32 which supports the observed Si/Al ratio of 1.01 and not the ideal 1.00. The proportion of nanoscrolls in NS-MeO-Kaol was tentatively estimated by preparing a mixture of MeO-Hal with pristine kaolinite. When MeO-Hal and pristine kaolinite were mixed at a weight ratio of 1 : 6, the relative intensity of the 0.86 nm diffraction line to the 0.72 nm diffraction line (Fig. S4b†) closely matches that of NS-MeO-Kaol (Fig. 1d). However, the SEM image of the mixture (Fig. S5c†) revealed that the nanoscrolls in the mixture appear larger than those in the SEM image of NS-MeO-Kaol (Fig. 2d). Notably, nanoscrolls prepared from kaolinite were better defined than halloysite nanoscrolls.²⁶ Additionally, nanoscrolls were observed at edges and surfaces of kaolinite, with aggregated nanoscrolls, which were difficult to separate from platy particles in image analyses. Because kaolinite intercalation reactions do not fully proceed, complete rolling-up of kaolinite is unattainable across kaolinite specimens of various geographical origins. Estimating the proportion of nanoscroll based on the surface area of fully rolled-up kaolinite is impractical. Although halloysite nanoscrolls exhibit a relatively high BET surface area (51 m² g⁻¹), this value is similar to that of NS-MeO-Kaol (52 m² g⁻¹), despite the presence of platy particles (Table 1). This similarity highlights the difficulty of accurately estimating the proportion of nanoscrolls, which remains a challenge in kaolinite chemistry and a subject for future investigation.

Table 1 BET surface areas of kaolinite, halloysite, calcined specimens in this study

	BET surface area (m^2 g^−1)
MeO-Kaol	17
NS-MeO-Kaol	52
Halloysite	51
Ground raw material containing MeO-Kaol	31
Ground raw material containing NS-MeO-Kaol	48
CAS-MeO-750	28
CAS-NS-750	48
CAS-MeO-900	4
CAS-NS-900	4

---

Fig. S4a† shows that MeO-Hal contains by-products despite the application of the purification procedure used for purified MeO-Kaol. Similar results were obtained using an alternative purification procedure (see experimental), as illustrated in Fig. S6,† reported previously.²⁵ Thus, achieving precise purification of natural halloysite remains a significant challenge, particularly within the field of clay mineral engineering, and is beyond the scope of this study.

Based on the results above, incompletely rolled-up and purified kaolinite is the best raw material for studying halloysite's effect on kaolinite solid-state reactions. Although the relative intensity of the (020) diffraction line, which corresponds to kaolinite's lateral atomic arrangement,¹⁵ to the 0.86 nm diffraction line in the XRD pattern of ground raw materials of NS-MeO-Kaol with CaCO₃ is lower than that of the ground raw materials of MeO-Kaol with CaCO₃ (Fig. 3), there are no major differences in the degree or extent of kaolinite layer stacking disorder compared to the previous report.¹⁵ The solid-state reaction of kaolinite was facilitated by layer expansion and subsequent disruption of stacking order due to grinding when the relative intensity of the (020) reflection to the (001) reflection decreased by more than 2 times.¹⁵ In the present study, the relative intensity of ground raw materials containing NS-MeO-Kaol and MeO-Kaol are 0.9 and 1.1, respectively (Fig. 3). Although the particle shape of kaolinite is well-definite and its size distribution is narrow compared with various layered clay minerals,¹ the size distribution remains wide, as seen in the SEM image (Fig. 2a). Additionally, the particle shape of CaCO₃ used in this study is less defined, its size distribution is wide (Fig. S7c†). Therefore, estimating the difference in the particle size distribution of ground raw materials is impractical. In contrast, the appearance of the SEM images of ground raw materials show no significant difference (Fig. S7a and b†). Notably, the grinding condition (see experimental) in this study is milder than those reported previously.^13,50,51 Additionally, a hysteresis curve in the N₂ adsorption/desorption isotherm and an 11 nm peak in the BJH pore distribution are observed in the ground raw material containing nanoscrolls (Fig. S2e and S3c†). Furthermore, the BET surface area is not significantly reduced by the grinding process, with NS-MeO-Kaol at 52 m² g⁻¹ and the corresponding ground raw material at 48 m² g⁻¹ (Table 1). Therefore, the grinding procedure did not significantly affect the disruption of stacking order and particle size of kaolinite in the ground raw materials. This allows us to focus on the morphology of nanoscolls as follows.

Fig. 3 XRD patterns of (a) CaCO₃ used in this study, and ground raw materials containing (b) purified MeO-Kaol, and (c) NS-MeO-Kaol.

Fig. 4 presents the XRD patterns of specimens calcined at 900 °C. All the specimens exhibit reflections attributed to metastable CaAl₂Si₂O₈, with the patterns corresponding to the reference PDF No. 00-062-0853 for calcium aluminosilicate, which has a monoclinic structure (C2).⁵² The XRD patterns of specimens calcined for 12 and 24 hours show no noticeable differences, regardless of whether MeO-Kaol or NS-MeO-Kaol was used as the raw material, indicating that the 12 h calcination was sufficient to reach the reaction equilibrium. Compared with this PDF pattern, the XRD patterns of these specimens exhibit relatively strong and broad reflections in the 27–28° (2θ) range. These reflections correspond to the strongest reflections in the PDF profile of anorthite (PDF No. 41-1486), the stable phase of CaAl₂Si₂O₈. The relative intensity of the reflection corresponding to anorthite is higher in the XRD pattern of CAS-NS-900 compared to CAS-MeO-Kaol. This indicates that, in the solid-state reaction involving platy particles of 1 : 1-type layered aluminosilicate, the presence of nanoscrolls facilitated the reaction, particularly the phase transformation from the metastable into the stable phases. The plausible mechanisms are discussed below. Notably, metastable CaAl₂Si₂O₈ exhibits six polymorphs, some of which have layered structures, as recently revealed by single-crystal XRD analyses of nature samples.⁵³ These samples contain Ba as an impurity, suggesting the presence of the portion of hexacelsian whose layered structure closely resembles that of metastable CaAl₂Si₂O₈.⁵⁴ Thus, precise solid-state reactions or hydrothermal syntheses will be required to prepare purified single-crystal tetrahedral aluminosilicate layered compounds. Meanwhile, the presence of anorthite in CAS-MeO-900 and CAS-NS-900 is further supported by broad Raman spectra (Fig. S8†), which show two relatively well-separated bands near 500 cm⁻¹, characteristic of anorthite.⁵⁴ Previous studies have also demonstrated that the solid-state reaction between kaolinite and CaCO₃ ultimately yields anorthite.^13,15 Additionally, broad bands in the 450–400 cm⁻¹ region indicate that the metastable CaAl₂Si₂O₈ phase in these specimens possesses a relatively disordered structure compared to single crystals.⁵⁴ Notably, a dominant Raman band at 640 cm⁻¹ is observed, which is attributed to Al–O–Al vibrations.^55,56 In general, polycrystalline materials formed by solid-state reactions can exhibit structural disorder. Given the relatively broad XRD profiles and the NMR results discussed later, the specimens in this study are considered to possess such disordered structures. Detailed studies on the structural evolution of polycrystalline materials are a subject for future work and beyond the scope of this study. Furthermore, grain growth and increased crystallinity of calcined products proceed concurrently during solid-state reactions, indicating that the amount of calcined product cannot be simply determined. These points could be addressed in future studies; meanwhile, the anorthite contents of CAS-MeO-900 and CAS-NS-900 were tentatively estimated at 15% and 24%, respectively, by Rietveld refinement. This highlights the greater anorthite content in CAS-NS-900 compared to CAS-MeO-900.

Fig. 4 XRD patterns of (a) CAS-MeO-900, (b) CAS-MeO-900-24, (c) CAS-NS-900, and (d) CAS-NS-900-24. The closed triangles indicate anorthite reflections.

Negligible differences are observed in the SEM images (Fig. S9†) and surface areas (Table 1) of CAS-MeO-900 and CAS-NS-900. Attempts to identify differences the stable phase formation between specimens calcined at 1000 °C revealed no detectable changes in their XRD patterns and SEM images (Fig. S10 and S11†). Similarly, the exothermic peaks at around 1000 °C associated with anorthite formation¹³ in the DTA curves of both ground raw materials containing MeO-Kaol and NS-MeO-Kaol (Fig. S12d and e†) show no significant differences. The SEM image of NS-MeO-Kaol indicates a limited presence of nanoscrolls in platy particles (Fig. 2c), making differences in these characteristics unlikely to be detected. However, the exothermic peak associated with the formation of metastable CaAl₂Si₂O₈ in the DTA curve of ground raw material containing NS-MeO-Kaol is 25 °C higher than that containing MeO-Kaol (Fig. S12d and e†), despite the rapid transformation of the metastable phase into the stable phase of CaAl₂Si₂O₈ (Fig. 4). Therefore, differences in the formation of the metastable phase between MeO-Kaol and NS-MeO-Kaol as raw materials warrant further consideration, as discussed below.

A previous report, based on ²⁷Al and ²⁹Si MAS NMR spectra, showed that the crystalline metastable CaAl₂Si₂O₈ formed via a low crystalline phase.^13,14 Both low and high crystalline phases exhibited similar profiles, with the low-crystalline phase generated though 800 °C calcination of the raw materials.^13,14 In the present study, although very weak reflections corresponding to metastable CaAl₂Si₂O₈ are detected at 800 °C—likely due to the rapid solid-state reaction facilitated by kaolinite layer expansion (as previously reported¹⁵)—calcination at 750 °C produced products with halo profiles in both MeO-Kaol and NS-MeO-Kaol systems (Fig. 5). The calcination temperature was chosen based on endothermic peaks near 700 °C in the DTA curves (Fig. S12d and e†), which reflect decarbonation.¹³ The ²⁷Al and ²⁹Si MAS NMR spectra of calcined specimens at 750 °C confirm the formation of metastable CaAl₂Si₂O₈ (Fig. 6b and c). At higher calcination temperature, the NMR signals sharpen, aligning with the increased reflection attributed to metastable CaAl₂Si₂O₈ in the XRD patterns of specimens calcined at 800 and 900 °C (Fig. 4a, c, 5c, and d). Despite halo profiles in XRD patterns (Fig. 5) and solid-state NMR spectra indicative of metastable CaAl₂Si₂O₈ (Fig. 6b–e), rod-like morphology—likely originating from nanoscrolls—appear in the SEM images of the 750 and 800 °C calcined specimens when NS-MeO-kaol was used, compared to those prepared with MeO-Kaol (Fig. 7). Consistent with this, the surface area of the 750 °C specimen using NS-MeO-Kaol is higher than that of its MeO-Kaol counterpart (Table 1). Furthermore, a slight hysteresis curve is observed in the N₂ adsorption/desorption isotherm of this specimen calcined at 750 °C, which results in a very weak and broad peak at 12 nm in the BJH pore distribution (Fig. S2g and S3d†). These results, together with the calcination behavior of kaolinite, the solid-state reaction of halloysite with CaCO₃, and instrument-based characterization of crystalline metastable CaAl₂Si₂O₈, form the basis for further discussion and are explored in detail in the following section.

Fig. 5 XRD patterns of (a) CAS-MeO-750, (b) CAS-NS-750, (c) CAS-MeO-800, and (d) CAS-NS-800.

Fig. 6 ²⁷Al (left) and ²⁹Si (right) MAS NMR spectra of (a) kaolinite, (b) CAS-MeO-750, (c) CAS-NS-750, (d) CAS-MeO-800, (e) CAS-NS-800, (f) CAS-MeO-900, and (g) CAS-NS-900.

Fig. 7 SEM images of (a) CAS-MeO-750, (b) CAS-MeO-800, (c) CAS-NS-750, and (d) CAS-NS-800. The yellow rectangle in (c) is enlarged in (c′).

Calcinating kaolinite at 400–800 °C range generates metakaolinite, an amorphous layered aluminosilicate (Al₂O₃·2SiO₂).^57–60 As the temperature increases, the original 6-fold Al environment in kaolinite undergoes dehydroxylation, giving rise to 4- and 5-fold Al sites.^54–56 Correspondingly, ²⁷Al MAS NMR signals for these 4- and 5-fold sites appear downfield from the 0 ppm signal of 6-fold Al.^57–59 In the ²⁹Si MAS NMR spectra, a broad Q⁴ signal appears upfield from the Q³ doublets at −90.3 and −90.7 ppm, which are characteristic of kaolinite.^57–59 When kaolinite reacts with CaCO₃, the ²⁷Al MAS NMR spectrum of metastable CaAl₂Si₂O₈ displays a 58 ppm signal attributed to 4-fold Al in its layered calcium aluminosilicate framework (alternating stacked tetrahedral aluminosilicate layers and interlayer Ca ions) along with a 13 ppm signal.^13,14 However, the exact origin of this 13 ppm signal remains uncertain. Only three prior studies^13,14,23 have noted this signal without assigning a specific cause, highlighting the need for more detailed ²⁷Al MAS NMR investigations. However, in zeolitic studies, the 13 ppm signal is attributed to distorted Al octahedral sites.^61,62 Likewise, the ²⁹Si MAS NMR spectrum of metastable CaAl₂Si₂O₈ shows a −86 ppm signal for tetrahedral Si, accompanied by broader signals in the upfield region indicative of unreacted Si.^13,14,23 These results further support the presence of relatively disordered structures in CAS-MeO-900 and CAS-NS-900, as suggested by the XRD and Raman profiles (Fig. 4 and S8†). In the present study, the 58 ppm signal in CAS-NS-750 is slightly sharper than in CAS-MeO-750, and the 0 ppm signal corresponding to the 6-fold Al environment of kaolinite is absent from both specimens (Fig. 6b and c, left). Moreover, in the ²⁹Si MAS NMR spectra, the relative intensity of the −86 ppm signal to the broad upfield signal is higher in CAS-NS-750 than in CAS-MeO-Kaol (Fig. 6b and c, right). These observations indicate that, the solid-state reaction at 750 °C proceeded in both NS-MeO-Kaol and MeO-Kaol systems, but it is more pronounced in the former than in the latter.

The halloysite used in this study has a relatively high quartz content as a by-product (Fig. S6†), but its solid-state reaction with CaCO₃ remains noteworthy. The XRD patterns of CAS-Hal-750, -800, and -850 show hallo profiles with weak reflections attributed to quartz (Fig. 8). SEM images reveal rod-like particles, likely originating from halloysite nanoscrolls, in CAS-Hal-750, -800, and -850 (Fig. 9). Additionally, the ²⁷Al MAS NMR spectrum of CAS-Hal-750 (Fig. 10) closely resembles that of CAS-NS-900 rather than CAS-MeO-900 (Fig. 6g and f). The DTA curve of ground raw materials containing halloysite shows an endothermic peak due to decarbonation at 700 °C, corresponding to decarbonation,¹⁰ and an exothermic peak at 900 °C, indicating the formation of metastable CaAl₂Si₂O₈ (Fig. S13†). The latter peak occurs at a higher temperature compared to the 885 °C exothermic peak in the DTA curve of ground raw materials containing NS-MeO-Kaol (Fig. S12e and S13†). Notably, a previous study demonstrated that halloysite, under similar reaction conditions (900 °C for 4.5 h), produced anorthite as the main product, accompanied by metastable CaAl₂Si₂O₈.¹⁵ In the present study, the XRD patterns of CAS-Si-MeO-900 show stronger anorthite reflections compared to CAS-MeO-900 (Fig. S14d† and 4a). However, the metastable CaAl₂Si₂O₈ phase remains dominant, and no significant changes are observed in the XRD patterns of amorphous silica before and after calcination at 900 °C or in the patterns of ground raw materials containing MeO-Kaol and amorphous silica (Fig. S14†). Thus, the slightly Si-rich composition facilitates the phase transformation, consistent with a previous study showing that Si-enrich hexacelsian generates defects near Ba ions upon heat treatment, which in turn promotes the transformation of hexacelsian to celsian.⁶³ Notably, metastable CaAl₂Si₂O₈ belongs to the RAl₂Si₂O₈ series (where R represents alkaline earth ion) and consisted of alternately stacked tetrahedral aluminosilicate layers and R ions.⁶⁴ Thus, hexacelsian exhibits a similar layered structure to metastable CaAl₂Si₂O₈, further supporting their structural analogy in their phase transformation behavior. In contrast, celsian does not have a layered structure.⁶⁵

Fig. 8 XRD patterns (a) CAS-Hal-750, (b) CAS-Hal-800, and (c) CAS-Hal-850. Open triangles indicate quartz reflections.

Fig. 9 SEM images of (a) CAS-Hal-750, (b) CAS-Hal-800, and (c) CAS-Hal-850.

Fig. 10 ²⁷Al MAS NMR spectrum of CAS-Hal-750.

The BET surface area of CAS-NS-900 (4 m² g⁻¹) is the same as that of CAS-MeO-900, but it is five to ten times lower than those of CAS-MeO-750 and CAS-NS-750 (28 and 48 m² g⁻¹, respectively; Table 1). SEM images of CAS-NS-750 and CAS-NS-800 show rod-like morphologies, which are absent in CAS-NS-900 (Fig. 9 and S9†). Particles in the SEM image of CAS-NS-900 exhibit an ambiguous platy morphology (Fig. S9b†). These observations indicate that rod-like particles disappeared due to the formation of crystalline metastable CaAl₂Si₂O₈, a layered compound.

Based on the results, the proposed mechanisms comparing the platy and nanoscroll morphologies of kaolinite are as follows; (1) aggregated nanoscrolls observed in SEM images (Fig. 2c and S7b†) and a higher temperature shift for the exothermic peak during metastable CaAl₂Si₂O₈ formation (Fig. S12d, e†) indicate that the surface area of kaolinite in contact with CaCO₃ is reduced, because the inner parts and interiors of aggregated nanoscrolls are less accessible to CaCO₃ compared to the surfaces of platy particles; (2) the nano-sized hollow shape of nanoscrolls promotes ion diffusion relative to the mainly submicron- or micrometer-sized platy particles (Fig. 2 and S7†), thereby accelerating the formation of the metastable phase and its subsequent transformation into the stable phase once the reaction initiates; (3) according to the previous study discussing defects generated by increased Si content in tetrahedral aluminosilicate layers,⁶³ which promotes the transformation of metastable to stable phases, increased silica exteriors in contact with CaCO₃, resulting from nanoscroll formation, enhances the relative Si exposure compared to Al. Based on these proposed mechanisms, with the acceleration of the phase transformation by nano-sized hollow morphology, the solid-state reaction between kaolinite used in this study (Al_2.00Si_2.02O_5.00(OH)_4,00; see Experimental) and CaCO₃ appears to be enriched in Si due to the increased Si exteriors. Therefore, the purification of kaolinite specimens and the formation of nanoscrolls revealed the influence of halloysite-like morphologies on kaolinite's solid-state reactions. Additionally, low-crystalline metastable CaAl₂Si₂O₈ retains the raw material's morphology, potentially increasing its surface area.^66,67 The preserved morphology of raw materials likely contributes to the higher surface area of low-crystalline metastable CaAl₂Si₂O₈, which has potential applications in selective heavy metal ion uptake.¹⁴ Consequently, this study not only sheds light on the mechanisms of kaolinite's solid-state reactions but also advances the design of solid-state products by improving raw material purity and morphologies.^65–67

Conclusions

This study demonstrates the solid-state reaction of incompletely rolled-up kaolinite with CaCO₃. By purifying kaolinite to remove quartz by-products and achieved a Si/Al molar ratio of 1.01, the effect of halloysite nanoscrolls compared to kaolinite platy particles on the solid-state reaction was clarified as follows; (1) nanoscrolls delayed the initiation of the reaction; (2) nanoscrolls facilitated the phase transformation of metastable into stable phases once the reaction started; (3) the nanoscroll morphology was retained in the amorphous metastable phase of the product. The results in this study pave the way for a deeper understanding of kaolinite solid-state reactions^8–25 and offer insight into designing advanced solid-state products by optimizing kaolinite intercalation chemistry.⁵

Author contributions

Shingo Machida: conceptualization, data curation, investigation, writing – original draft, project administration, supervision. Hajime Okawa: investigation. Masaya Suzuki: investigation. Toshimichi Shibue: investigation, writing – review and Editing.

Data availability

The data supporting this article have been included within the tables and images presented in this paper and ESI.†

Conflicts of interest

The authors declare no conflict of interest.

Acknowledgements

Special thanks to Mr Toshiaki Ogasawara and Ms Mikiko Kuroda, staffs in Tokai Technol. Center, for elemental analyses and their sample preparations. Our sincere thanks to Mr Yasuo Nagano, a researcher in Japan Fine Ceramics Center (JFCC), for the TG-DTA experiment of halloysite specimen. This research was supported by JFCC Grant for Advanced Technology Development Research (T24A4560). The solid-state MAS NMR experiments were conducted in the Materials Characterization Central Laboratory of Waseda University, which was shared in the MEXT Project for promoting public utilization of advanced research infrastructure (program for supporting the construction of core facilities, Grant No. JPMXS0440500021).

References

1. M. F. Brigatti, E. Galán and B. K. G. Theng, in Developments in Clay Science Volume 5A Handbook of Clay Science, ed. F. Bergaya and G. Lagaly, Elsevier, Oxford, UK, 2nd edn, 2013, pp. 21–29.
2. G. Lagaly, M. Ogawa and I. Dekány, in Developments in Clay Science Volume 5A Handbook of Clay Science, ed. F. Bergaya and G. Lagaly, Elsevier, Oxford, UK, 2nd edn, 2013, pp. 435–506.
3. C. Detellier and R. A. Schoonheydt, Elements, 2014, 10, 201–206.
4. C. Detellier, Chem. Rec., 2018, 18, 868–877.
5. A. Weiss, Angew. Chem., Int. Ed. Engl., 1963, 12, 697–703.
6. S. Olejnik, A. M. Posner and J. P. Quirk, Clays Clay Miner., 1970, 8, 421–434.
7. K. Okada, M. Nemoto, Y. Kameshima and A. Yasumori, J. Ceram. Soc. Jpn., 2000, 108, 898–903.
8. D. Novembre and D. Gimeno, Clay Clay Miner., 2017, 65, 355–366.
9. S. Machida, T. Shibue, K. Katsumata and A. Yasumori, Appl. Clay Sci., 2024, 247, 107220.
10. K. Sumi, Y. Kobayashi and K. Etsuro, J. Ceram. Soc. Jpn., 1998, 106, 89–93.
11. S. Machida, Int. J. Appl. Ceram. Technol., 2024, 21, 3229–3236.
12. Y. Kobayashi and E. Kato, J. Am. Ceram. Soc., 1994, 77, 833–834.
13. K. Okada, N. Watanabe, K. V. Jha, Y. Kameshima, A. Yasumori and K. J. D. MacKenzie, Appl. Clay Sci., 2003, 23, 329–336.
14. K. Okada, N. Watanabe, V. K. Jha, Y. Kameshima, A. Yasumori and K. J. D. MacKenzie, J. Mater. Chem., 2003, 13, 550–556.
15. S. Machida, K. Katsumata and A. Yasumori, RSC Adv., 2021, 11, 38473–38477.
16. S. Machida, K. Katsumata and A. Yasumori, RSC Adv., 2022, 12, 15435–15439.
17. S. Machida, K. Katsumata and A. Yasumori, Int. J. Ceram. Eng. Sci., 2023, 5, e10165.
18. S. Machida, K. Katsumata and A. Yasumori, Separate Studies of Effects of Mixing and Grinding on Solid-State Reaction of Kaolinite with Calcium Carbonate, J. Ceram. Soc. Jpn., 2003, 131, 160–165.
19. S. Machida, N. Emori, K. Katsumata, K. Maeda and A. Yasumori, RSC Adv., 2022, 12, 31577–31584.
20. Y. Kobayashi, K. Sumi, T. Asai and E. Kato, J. Ceram. Soc. Jpn., 1999, 107, 657–661.
21. Y. Kobayashi, Ceram. Int., 2001, 27, 179–184.
22. Y. Kobayashi, J. Pee, Y. Murabayashi and M. Katayama, J. Ceram. Soc. Jpn., 2016, 124, 881–885.
23. K. Okada, H. Arai, Y. Kameshima, A. Yasumori and K. J. D. MacKenzie, Mater. Lett., 2003, 57, 3554–3559.
24. S. Machida, K. Katsumata and A. Yasumori, J. Solid State Chem., 2023, 322, 123945.
25. S. Machida, K. Katsumata and A. Yasumori, J. Solid State Chem., 2023, 326, 124218.
26. Y. Kuroda, K. Ito, K. Itabashi and K. Kuroda, Langmuir, 2011, 27, 2028–2035.
27. S. Machida, RSC Adv., 2023, 13, 26430–26434.
28. S. Machida, K. Katsumata and A. Yasumori, Materials, 2022, 15, 588.
29. É. Makó, D. István, P. Pekker, M. Pósfai, A. Kovács, Z. Ható and T. Kristóf, Appl. Clay Sci., 2020, 198, 105800.
30. X. Li, Q. Liu, H. Cheng, S. Zhang and R. L. Frost, J. Colloid Interface Sci., 2015, 444, 74–88.
31. Q. Liu, X. Li and H. Cheng, Appl. Clay Sci., 2016, 124–125, 175–182.
32. S. Zhang, Q. Liu, J. Luo, N. Yang, Y. Zhnag, Y. Liu and L. Sun, Inorg. Chem., 2023, 62, 13205–13211.
33. B. Singh and I. D. R. Mackinnon, Clays Clay Miner., 1996, 44, 825–834.
34. J. E. F. C. Gardolinski and G. Lagaly, Clay Miner., 2005, 40, 547–556.
35. R. Miyawaki, S. Tomura, K. Inukai, Y. Shibasaki, M. Okazaki, S. Samejima and S. Satokawa, Clay Sci., 1992, 8, 272–284.
36. S. Satokawa, Y. Osaki, S. Samejima, R. Miyawaki, S. Tomura, Y. Shibasaki and Y. Sugahara, Clays Clay Miner., 1994, 42, 288–297.
37. S. Satokawa, R. Miyasaki, Y. Osaki, S. Tomura and Y. Shibasaki, Clays Clay Miner., 1996, 44, 417–423.
38. S. Satokawa, R. Miyawaki, S. Tomura and Y. Shibasaki, Clay Sci., 1997, 10, 231–239.
39. J. J. Tunney and C. Detellier, J. Mater. Chem., 1996, 6, 1679–1685.
40. Y. Komori, H. Enoto, R. Takenawa, S. Hayashi, Y. Sugahara and K. Kuroda, Langmuir, 2000, 16, 5506–5508.
41. M. Katayama, J. Nakakuki, J.-H. Pee and Y. Kobayashi, J. Ceram. Soc. Jpn., 2013, 121, 934–939.
42. D. N. Hinckley, Clays Clay Miner., 1963, 11, 229–235.
43. R. Miyawaki, T. Sano, F. Ohashi, M. Suzuki, T. Kogure, T. Okumura, J. Kameda, T. Umezome, T. Sato, D. Chino, K. Hiroyama, H. Yamada, K. Tamura, K. Morimoto, S. Uehara and T. Hatta, Nendo Kagaku, 2010, 48, 158–198.
44. K. S. W. Sing, D. H. Everett, R. A. W. Haul, L. Moscou, R. A. Pierotti, J. Rouquerol and T. Siemieniewska, Pure Appl. Chem., 1985, 57, 602–619.
45. S. Brunauer, P. T. Emmett and E. Teller, J. Am. Chem. Soc., 1938, 60, 309–319.
46. E. P. Barrett, L. G. Joyner and P. P. Halenda, J. Am. Chem. Soc., 1938, 407, 309–380.
47. A. J. Koivisto, A. B. Bluhme, K. I. Kling, A. S. Fonseca, E. Redant, F. Andrade, K. S. Hougaard, M. Krepker, O. S. Prinz, E. Segal, A. Holländer, K. A. Jensen, U. Vogel and I. K. Koponen, NanoImpact, 2018, 10, 153–160.
48. D. Tan, P. Yuan, F. Annabi-Bergaya, F. Dong, D. Liu and H. He, Appl. Clay Sci., 2015, 114, 190–196.
49. Z.-L. Cheng, B.-C. Cao and Z. Liu, Micro Nano Lett., 2019, 14, 585–589.
50. P. B. Malla and S. Devisetti, US Pat., US6478865B1, 2002.
51. P. B. Malla and S. Devisetti, US Pat., US8382016B2, 2013.
52. K. Akatsuka, A. Yasumori and K. Maeda, Mater. Lett., 2019, 242, 163–165.
53. L. A. Gorelova, O. S. Vereshchagin, V. N. Bocharov, S. V. Krivovichev, A. A. Zolotarev and M. A. Rassomakhin, Geosci. Front., 2023, 14, 101458.
54. L. Gorelova, S. Britvin, O. Vereshchagin, D. Pankin, V. Bocharov, O. Silyukov and A. Kasatkin, J. Eur. Ceram. Soc., 2024, 44, 6402.
55. Z. Song, K. Song, B. Liu, P. Zheng, H. B. Bafrooei, W. Su, H. Lin, F. Shi, D. Wang and I. M. Reaney, Int. J. Appl. Ceram. Technol., 2020, 17, 771–777.
56. A. R. Allu, S. Balaji, D. U. Tulyaganov, G. C. Mather, F. Margit, M. J. Pascual, R. Siegel, W. Milius, J. Senker, D. A. Agarkov, V. V. Kharton and J. M. F. Ferreira, ASC Omega, 2017, 2, 6233–6243.
57. J. Rocha and J. Klinowski, Phys. Chem. Mater., 1990, 17, 179–186.
58. J. Rocha, J. Klinowski and J. M. Adams, J. Mater. Chem., 1991, 26, 3009–3018.
59. S. Machida, K. Katsumata and A. Yasumori, RSC Adv., 2021, 11, 23090–23094.
60. F. Bergaya, P. Dion, J.-F. Alcover, C. Clinard and D. Tchoubar, J. Mater. Chem., 1996, 31, 5609–5075.
61. H.-M. Kao and Y.-C. Chen, J. Phys. Chem. B, 2003, 107, 3367–3375.
62. I. A. Beta, B. Hunger, W. Böhlmann and H. Jobic, Microporous Mesoporous Mater., 2005, 79, 69–78.
63. K. T. Lee and P. B. Aswath, Mater. Chem. Phys., 2001, 71, 47–52.
64. J. Töpel-Schadt, W. F. Müller and H. Pentinghaus, J. Mater. Sci., 1978, 13, 1809–1816.
65. A. Táborosi, R. K. Szilagyi, B. Zsirka, O. Fónagy, E. Horváth and J. Kristóf, Inorg. Chem., 2018, 57, 7151–7167.
66. S. Machida, T. Gotoh, K. Katsumata and A. Yasumori, Appl. Clay Sci., 2021, 214, 106295.
67. M. Hashimoto, S. Takahashi, K. Kawahara, T. Ogawa and M. Kawashita, Adv. Powder Technol., 2022, 33, 1034112.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d5dt00598a

---