Cs⁺-exchange property of Na-bearing GTS-type titanosilicate and possible distribution of Cs⁺ ions in its Cs⁺-exchanged form

Abstract

Single phase of Na-bearing grace titanosilicate (Na-GTS) powder was prepared hydrothermally. Its Cs⁺-exchanged forms [Na_4(1−x)Cs_4xTi₄O₄(SiO₄)₃·nH₂O] with various compositions up to x = 1.0 were produced by shaking the Na-GTS in the CsCl aqueous solutions with the Cs-concentration (C_Cs) up to 1.0 mol/L at temperatures up to 80 °C for 24 h. The Cs⁺-exchange rates (x) at each temperature increase steeply with increasing C_Cs up to 0.1 mol/L, after which they are almost constant. Meanwhile, up to 60 °C the x values do not depend significantly on treatment temperature, but at a higher temperature of 80 °C they increase dramatically up to unity. Thus, at least above 80 °C, the increase in treatment temperature is considerably effective for promotion of ion-exchange, as well as the increase in C_Cs. TG-DTA measurements show that the exchange of Na⁺ ions for Cs⁺ ions largely decreases H₂O contents, as a consequence of a remarkable reduction in available void-space within the cavities due to a remarkable increase in mean cation size. The comparison between the observed XRD pattern and the simulated ones suggests the cation-distribution model that Cs⁺ ions prefer the 6g site in the GTS cavity in the assumed pseudocubic structure to the 4e site. The unit cell volumes increase largely with increasing x although the H₂O contents reduce largely. This can be because moderate bonding forces acting between a Cs⁺ ion on the 6g site, located at the proximity of the centers of the 8-membered ring windows, and adjacent O atoms forming the windows directly influence the framework structure. Furthermore, these findings are compared with those of the previously reported Er³⁺-exchanged forms, in terms of the difference in valence and size of the extra-framework cations.

INTRODUCTION

The potential of various microporous crystals, such as zeolites, as radioactive-element removers has been investigated through an examination of their ion-exchange functionality. In nuclear accidents, the radioactive elements of particular concern are ¹³¹I, ¹³⁴Cs, ¹³⁷Cs, and ⁹⁰Sr. These radioactive elements are removed from the radioactive waste water of the Fukushima Daiichi Nuclear Power Plant using microporous crystals as adsorbents. Their representative examples are chabazite, clinoptilolite, mordenite, A-type zeolite, crystalline silicotitanate (CST), and so on. In particular, the CST with a one-dimensional tunnel structure is well known to have a high ion-exchange selectivity for Cs⁺ and is currently believed to be the most highly effective microporous crystal for Cs⁺-removal.

Our recent study (Kawata et al., 2024) proposed the following three crystal-chemical factors necessary for a highly efficient Cs⁺-remover from the single-crystal X-ray diffraction study of the Cs⁺-exchanged Ca-chabazite: (1) the framework structure is composed of a three-dimensional multi-branch network channel; (2) there is the crystallographic site having the strong preference for Cs⁺ ions and fixing Cs⁺ ions by moderate bonding forces; (3) the framework structure is rigid. Particularly from the viewpoint of (1), grace titanosilicates (GTS) with a three-dimensional tunnel structure can be expected to be more effective for Cs⁺-adsorption (exchange) rather than the CST with the one-dimensional tunnel structure. The GTS is a structural analogue of pharmacosiderite KFe₄(OH)₄(AsO₄)₃·6H₂O with a cubic symmetry (space group P43m) and its general formula is expressed as M_4/mTi₄O₄(SiO₄)₃·nH₂O (m: valence of extra-framework cation M). As illustrated in Figures 1a and 1b, the GTS framework has a structural feature that a Ti₄O₁₆ cluster produced by four edge-shared TiO₆ octahedra is linked to adjacent clusters through SiO₄ tetrahedra, which results in the three-dimensional tunnel structure described as an interconnected pore system involving cavities of 8-membered ring channels. The cavities are occupied by H₂O molecules and exchangeable extra-framework cations. The negative charge of the framework is compensated by the presence of the latter. The crystallographic symmetry of GTS varies depending on the extra-framework cation species. For example, the (Sr, H)-bearing GTS [(Sr, H)-GTS] with a simplified composition of SrH₂Ti₄O₄(SiO₄)₃·7H₂O (Spiridonova et al., 2011) was reported to crystallize in the cubic space group P43m [a = 7.834(5) Å]. The (Cs, H)-bearing GTS [(Cs, H)-GTS] with a composition of Cs₃HTi₄O₄(SiO₄)₃·4H₂O (Harrison et al., 1995) and the (K, H)-bearing GTS [(K, H)-GTS] with a composition of K₃HTi₄O₄(SiO₄)₃·4H₂O (Behrens et al., 1996) were also reported to crystallize in the same cubic space group [7.8301(9) Å for (Cs, H)-GTS; a = 7.7644(3) Å for (K, H)-GTS]. Meanwhile, Na-bearing GTS (Na-GTS) with a composition of Na₄Ti₄O₄(SiO₄)₃·6H₂O (Dadachov and Harrison, 1997) was reported to crystallize in the trigonal space group R3m [a = 7.8123(6) Å, α = 88.794(9)°]. This unit cell is very close to a cubic cell and the crystal structure of the Na-GTS is often described as a pseudocubic cell. For comparison, the framework structures and extra-framework chemical species of the Na-GTS as an example of trigonal forms and the (Sr, H)-GTS as an example of cubic forms are shown in Figures 1a and 1b, respectively.

Figure 1. Framework structure and extra-framework chemical species of (a) Na-GTS (Dadachov and Harrison, 1997) and (b) (Sr, H)-GTS (Spiridonova et al., 2011) projected along [010]. In (a), Na(1a) and Na(3b) represent Na⁺ positions on the 1a site (coordinates: 0.65, 0.65, 0.65) and the 3b site (coordinates: 0.91, 0.44, 0.44) in the trigonal space group R3m, respectively. In (b), Sr(4e) and Sr(6g) represent Sr²⁺ positions on the 4e site (coordinates: 0.63, 0.63, 0.63) and the 6g site (coordinates: 0.96, 0.5, 0.5) in the cubic space group P43m, respectively. The 1a and 3b sites in the trigonal space group R3m are associated with one of four equivalent positions on the 4e site and three of six equivalent positions on the 6g site in the cubic space group P43m, respectively. The squares drawn by the dotted lines represent unit cells.

Investigating ion-exchange properties of GTS for various cations can provide important insights into the search and development of radioactive-element removers (Behrens et al., 1996; Behrens and Clearfield, 1997; Fujiwara et al., 2013). In particular, the search and development of still more highly efficient Cs⁺-exchangers are eagerly desired, because ¹³⁷Cs was released in large quantities by the nuclear accident and its half-life (30.1 years) is considerably long. Investigation of the Cs⁺-exchange properties of the Na-GTS, a representative GTS compound, is thus essential for clarifying its value as a radioactive-element remover. Meanwhile, our previous study (Fujiwara et al., 2020) on Er³⁺-exchanged Na-GTS, Na_4(1−x)Er_4x/3Ti₄O₄(SiO₄)₃·nH₂O (0 ≤ x ≤ 1), reported that the Er³⁺-exchange rate increases up to x = 0.96 with increasing the Er³⁺-concentration and temperature in the ErCl₃ aqueous solutions employed in the Er³⁺-exchange treatments, in their investigated ranges. Moreover, the distribution of Er³⁺ ions in the cavities was discussed from comparison between the observed powder X-ray diffraction (XRD) pattern and the simulated ones. The comparison between Cs⁺- and Er³⁺-exchanged forms is significant for obtaining knowledge of the effects of valence and size of the extra-framework cations on ion-exchange property. For this purpose, we here investigate variation of Cs⁺-exchange rate of Na-GTS with Cs⁺-concentration and temperature in the ion-exchange treatments and examine the distribution of Cs⁺ ions in the GTS cavity from the powder XRD simulation. The obtained results are compared with those of the reported Er³⁺-exchanged forms, in terms of ion-exchange properties, H₂O contents, cation distributions within the cavities and the resulting structural variations.

EXPERIMENTAL

The powder sample of Na-GTS [Na₄Ti₄O₄(SiO₄)₃·6H₂O] was prepared hydrothermally, following the procedure described in previous studies (Kostov-Kytin et al., 2007; Fujiwara et al., 2010; Fujiwara et al., 2013; Fujiwara et al., 2017). The initial materials employed in the sample preparation were NaOH, amorphous SiO₂ fine powder, TiCl₄ aqueous solution and HCl. These reagents were mixed in a compositional ratio of TiO₂/SiO₂ = 0.320 and Na₂O/TiO₂ = 5.625, and the mixture was subjected to a heat treatment at 100 °C for 24 h in a closed pressure-resistant vessel. The prepared sample was filtered using ultra-pure water and dried at 80 °C for 24 h to obtain Na-GTS powder. The Cs⁺-exchanged forms were prepared by shaking the Na-GTS powder (0.5 g) in CsCl aqueous solutions (25 mL) at temperatures of 25, 40, 60, and 80 °C for 24 h. The concentration of Cs⁺ (C_Cs) in the aqueous solutions was varied between 0.01 and 1.0 mol/L. This range was selected based on the theoretical minimum amount necessary for complete exchange of Na⁺ ions in Na-GTS for Cs⁺ ions, which is 25 mL of 0.1093 mol/L CsCl aqueous solution. The resulting samples were filtered and washed with ultra-pure water. After that, the samples that had undergone the Cs⁺-exchange treatments at 25, 40, and 60 °C were dried at 80 °C for 24 h, and the samples that had undergone it at 80 °C were dried at 40 °C for 48 h. As will be discussed later, this difference in the dry temperatures provides important knowledge of H₂O contents in GTS compounds.

The amounts of Na⁺ eluted into the supernatant solutions from Na-GTS after the ion-exchange treatments were analyzed using atomic absorption spectrometry (AAS) in order to evaluate the ion-exchange rates, which are defined by the variable x in the chemical formula Na_4(1−x)Cs_4xTi₄O₄(SiO₄)₃·nH₂O (0 ≤ x ≤ 1). Practically, the x value was calculated from n(Na)/4n(Na-GTS) based on the ion-exchange reaction 4xCs⁺ + Na₄Ti₄O₄(SiO₄)₃ → Na_4(1−x)Cs_4xTi₄O₄(SiO₄)₃ + 4xNa⁺, where n(Na-GTS) represents the amounts of as-prepared Na-GTS employed in the ion-exchange treatments and n(Na) denotes the amounts of Na⁺ present in the supernatant solutions after the ion-exchange treatments. The phase identification and structural characterization of the obtained samples were conducted using a Rigaku RINT2200 powder X-ray diffractometer. The thermogravimetry (TG) coupled with differential thermal analysis (DTA) was conducted to determine the H₂O contents of the samples and examine the dehydration process. The TG-DTA measurements were conducted with a heating rate of 10 °C/min.

RESULTS AND DISCUSSION

Cs⁺-exchange property

Figure 2 shows the variation of Cs⁺-exchange rates (x) in the Cs⁺-exchanged Na-GTS [Na_4(1−x)Cs_4xTi₄O₄(SiO₄)₃·nH₂O (0 ≤ x ≤ 1)] prepared at each temperature as a function of the Cs⁺-concentration (C_Cs) in CsCl aqueous solution. As shown in the figure, the Cs⁺-exchange rates (x) at each temperature increase steeply with increasing C_Cs up to 0.1 mol/L, after which they are almost constant. Meanwhile, in the range of 25-60 °C the x values do not significantly depend on treatment temperature and reach only at most 0.8, but at a higher temperature of 80 °C they increase dramatically up to 1, corresponding to the complete Cs⁺-exchange. Thus, at least above 80 °C, the increase in treatment temperature is considerably effective for promotion of ion-exchange, as well as the increase in C_Cs.

Figure 2. Variations of Cs⁺-exchange rates (x) in the Cs⁺-exchanged forms prepared at each temperature of T_ex = 25, 40, 60, and 80 °C as a function of C_Cs. T_ex represents temperatures at the Cs⁺-exchange treatments. The upper and lower figures are presented using a linear scale and a logarithmic scale for the horizontal axis, respectively.

Our previous study (Fujiwara et al., 2020) conducted the Er³⁺-exchange treatments of Na-GTS in the range of C_Er ≤ 0.5 mol/L at 25 and 60 °C. The result showed that the increase in C_Er up to 0.5 mol/L increases the x values up to 0.80 at 25 °C and up to 0.96 at 60 °C. On the other hand, the present Cs⁺-exchange treatments showed that the increase in C_Cs up to 0.5 mol/L increases the x values only up to 0.67 at 25 °C and only up to 0.72 at 60 °C in comparison under the same conditions of aqueous-solution concentration and treatment temperature. This shows that the exchange of Na⁺ ions in Na-GTS for Er³⁺ ions is more prominent than that for Cs⁺ ions up to at least 60 °C.

Dehydration behaviour and H₂O content

Figure 3 illustrates the TG and DTA curves of as-prepared Na-GTS and of the Cs⁺-exchanged forms prepared under each condition. The TG curves demonstrate that the samples undergo a rapid decrease in weight up to ∼ 300 °C, followed by a gradual completion of weight-losses by 700 °C. The weight-loss ratios up to 800 °C are in the range of 11-22% for each sample. These weight-losses can be attributed to dehydration from the samples, as evidenced by the presence of endothermic peaks in the DTA curves of each sample up to 300 °C. As shown in Figures 4a-4c, the exchange of Na⁺ ions in the Na-GTS for Cs⁺ ions generally reduces the H₂O contents calculated from the weight-loss ratios. For comparison, the data of Er³⁺-exchanged forms are also shown in Figure 4c. As the previously reported H₂O contents of the Er³⁺-exchanged forms (Fujiwara et al., 2020) were subsequently identified as erroneous, the values recalculated by making the data-processing described later for the corrected data are presented in the figure. It is noteworthy that there is the obvious difference in the x-dependence of the H₂O contents between the present Cs⁺-exchanged forms and the reported Er³⁺-exchanged forms (Fig. 4c).

Figure 3. TG and DTA curves of as-prepared Na-GTS and of its Cs⁺-exchanged forms prepared at each temperature of T_ex = 25, 40, 60, and 80 °C. T_ex is as in Figure 2. T_dry represents temperatures at the dry treatments carried out after the hydrothermal synthesis for the Na-GTS and after the Cs⁺-exchange treatments for the Cs⁺-exchange forms. The vertical dotted lines represent 120 °C, at which the desorption of the excess water (1) (adsorbed water on grain surface or capillary-condensed water between grains) is expected to be completed.

Figure 4. Variations of H₂O contents calculated from the weight-loss ratios in TG as a function of ion-exchange rate (x): (a) the case in the present Cs⁺-exchanged forms calculated from the weight-loss ratios between 20 and 800 °C; (b) the case in the present Cs⁺-exchanged forms calculated from the weight-loss ratios between 120 and 800 °C (calculated by excluding the weight-loss ratios up to 120 °C); (c) comparison of (b) in the present Cs⁺-exchanged forms with variation of H₂O contents in the reported Er³⁺-exchanged forms (Fujiwara et al., 2020) recalculated by excluding the weight-loss ratios up to 120 °C. In (a), the samples Cs⁺-exchanged at T_ex = 80 °C (solid symbols) were dried at T_dry = 40 °C after the sample preparation, and the remaining samples (open symbols) were dried at T_dry = 80 °C after the sample preparation; the solid line and the dotted line are the regression lines for the former samples and the latter samples, respectively. In (b), the solid line is the regression line for all the investigated samples. In (c), the solid line and the dotted line are the regression lines for the present Cs⁺-exchanged forms (solid circles) and the reported Er³⁺-exchanged forms (open circles), respectively; the former line is a re-recording of (b). T_ex and T_dry are as in Figures 2 and 3, respectively.

Dadachov and Harrison (1997) reported from the Rietveld analysis using powder XRD data that in Na-GTS, the number of H₂O molecules coordinating to the extra-framework cation Na⁺ is 6 molecules per formula unit (mpfu). On the other hand, in the present study, the H₂O content of as-prepared Na-GTS powder sample calculated from the weight-loss ratio resulted in a value of 8.64 mpfu, which is much larger than the expected H₂O content of 6 mpfu. A comparable phenomenon was also identified in our previous study (Fujiwara et al., 2010, 2020). Fujiwara et al. (2010) identified the following two distinct types of excess water in a powder sample of GTS-type (K, Na, H)₄Ti₄O₄(SiO₄)₃·nH₂O, from the hydration-heat measurements: (1) adsorbed water on grain surface or capillary-condensed water between grains; (2) adsorbed water to the framework without any coordination to extra-framework cations. The discrepancy between the observed H₂O content (8.64 mpfu) in the present as-prepared Na-GTS and the expected value (6 mpfu) may be attributed to the presence of such excess water. Similarly, it is reasonable to suggest that the excess water is also present in the Cs⁺-exchanged powder samples.

Evidence for the presence of the excess water, in particular the excess water (1), can be found in Figures 3, 4a, and 4b. In Figure 3, the endothermic peaks below ∼ 100 °C in DTA curves are clearly observed in the samples dried at 40 °C (the samples Cs⁺-exchanged at 80 °C), whereas they are less clear in the samples dried at 80 °C (the samples Cs⁺-exchanged at 25, 40, and 60 °C). Moreover, in Figure 4a, there is the large difference in the x-dependence of H₂O contents calculated from the weight-loss ratios from room temperature (20 °C) to 800 °C between the samples dried at 40 and 80 °C; the H₂O contents in the former samples are remarkably overestimated than those in the latter samples. Here, it is noteworthy that all the samples prepared for the present study had been stored in enclosed sample-containers at ambient conditions for only a few hours between the end of the drying treatments and the start of the TG-DTA measurements. Fujiwara et al. (2010) reported from the hydration-heat measurements of the above (K, Na, H)-GTS powder sample that the excess water (1), once desorbed, is not immediately re-adsorbed even under a condition of high humidity, in contrast to the excess water (2). These observations and circumstances show that a large amount of the excess water (1) was present in the samples dried at a lower temperature of 40 °C, but indicate that the most of it was desorbed in the samples dried at a higher temperature of 80 °C, without its re-adsorption during the short-term storage of the samples in enclosed sample-containers at ambient conditions. Thus, we may consider that in all the investigated samples, the weight-loss ratios up to 120 °C, corresponding to the approximate end-temperature of the endothermic peaks observed below ∼ 100 °C, are due to the desorption of the excess water (1). Indeed, the comparison of Figures 4a and 4b reveals that the effect of the reduction in H₂O contents caused by excluding the weight-loss ratios up to 120 °C is more pronounced for the samples dried at 40 °C than for those dried at 80 °C. Consequently, the x-dependence of H₂O contents is on the same trend for both (Figs. 4b and 4c). At this time, the H₂O contents in the present samples with x = 0 (Na-GTS) and x =1.0 [Cs-bearing GTS (Cs-GTS)] are reduced from 8.64 to 6.11 mpfu and from 9.00 to 4.81 mpfu, respectively. The reduced H₂O contents for both samples are quite close to the number of H₂O molecules (6 mpfu) coordinating to Na⁺ in Na-GTS with a composition of Na₄Ti₄O₄(SiO₄)₃·6H₂O (Dadachov and Harrison, 1997) and of those (4 mpfu) coordinating to Cs⁺ in (Cs, H)-GTS with a composition of Cs₃HTi₄O₄(SiO₄)₃·4H₂O (Harrison et al., 1995; Behrens et al., 1996) reported by the structure analyses, respectively. It is also noteworthy that the reduced H₂O contents display a near-linear decline with increasing Cs⁺-exchange rate (x) (Figs. 4b and 4c). In the Cs⁺-exchanged forms, the exchange of smaller Na⁺ ions for larger Cs⁺ ions [ionic radii: ^VIIIr(Na⁺) = 1.18 Å and ^VIIIr(Cs⁺) = 1.74 Å; Shannon, 1976] is based on the substitution of Na⁺ → Cs⁺. The substitution does not change the number of cations, and can result in a remarkable reduction in available void-space within the cavities for incorporating H₂O molecules due to a remarkable increase in mean cation size. This can bring about the rapid reduction in H₂O contents with increasing x, in contrast to the case of the Er³⁺-exchanged forms (Fig. 4c).

On the other hand, in the Er³⁺-exchanged forms, the exchange of larger Na⁺ ions for smaller Er³⁺ ions [ionic radius: ^VIIIr(Er³⁺) = 1.004 Å; Shannon, 1976] can remarkably expand available void-space within the cavities because of both effects of the reductions in the number and mean size of extra-framework cations due to the substitution of 3Na⁺ → Er³⁺. However, the H₂O contents are almost unchanged or increase only slightly with increasing x (Fig. 4c), where H₂O contents in the reported Er³⁺-exchanged forms (Fujiwara et al., 2020) recalculated by excluding the weight-loss ratios up to 120 °C, as in the Cs⁺-exchanged forms, are presented in the figure. This suggests that the effect of increasing the amount of H₂O molecules incorporated into the expanded available void-space without coordinating to extra-framework cations is almost offset by the effect of reducing the amount of H₂O molecules coordinating to extra-framework cations, due to the reduced number of extra-framework cations. Thus, the variation of the H₂O content with ion-exchange can be related to the variation of the available void-space depending on valence and size of extra-framework cations.

XRD and possible Cs⁺ distribution

Figure 5 shows the observed powder XRD patterns of as-prepared Na-GTS and of the present Cs⁺-exchanged forms prepared under each condition. The unit-cell parameters assuming the rhombohedral cell-setting for the trigonal system were evaluated by a profile fitting method. The variations of the resulting unit-cell parameters (a, α) and unit-cell volume (V) in the present Cs⁺-exchanged forms as a function of x are shown in Figure 6, together with those in the reported Er³⁺-exchanged forms (Fujiwara et al., 2020). In the Er³⁺-exchanged forms, the increase in x decreases the a-axis length, increases the α angle and consequently slightly decreases the unit-cell volume V. This can be due to the reductions in the number and mean size of extra-framework cations (i.e., the reduction in their occupied space within the cavities), in consideration of almost the unchanged H₂O contents seen in Figure 4c. On the other hand, in the present Cs⁺-exchanged forms, the increase in x largely reduces H₂O contents in the cavities (Figs. 4b and 4c), but increases the a-axis length and the α angle and consequently increases the unit-cell volume V (Fig. 6). This suggests that the effect on V is greater in the increase in mean size of extra-framework cations than in the reduction in H₂O contents. The α angles range between 88.82 and 89.53°, exhibiting a proximity to 90°, corresponding to a cubic cell structure. This observation is consistent over the whole range of x, as well as in the Er³⁺-exchanged forms. The diffraction peaks illustrated in Figure 5 are accordingly indexed as a pseudocubic cell for the sake of convenience.

Figure 5. Observed XRD patterns of as-prepared Na-GTS and of its Cs⁺-exchanged forms prepared at each temperature of T_ex = 25, 40, 60, and 80 °C. T_ex is as in Figure 2. The diffraction indices in the XRD patterns are labeled assuming the pseudocubic structure.

Figure 6. Variations of unit-cell parameters (a, α) and unit-cell volumes (V) of the present Cs⁺-exchanged forms and the reported Er³⁺-exchanged forms (Fujiwara et al., 2020) as a function of ion-exchange rate (x).

Spiridonova et al. (2011) conducted the single-crystal XRD analysis of (Sr, H)-GTS [simplified composition: SrH₂Ti₄O₄(SiO₄)₃·7H₂O; cubic space group: P43m], which is a Sr²⁺-exchanged form of a natural ivanyukite-Na-T. The analysis reported that Sr²⁺ ions occupy both the 4e-site (coordinates: 0.63, 0.63, 0.63) and the 6g-site (coordinates: 0.96, 0.5, 0.5) within the GTS cavity. In order to examine the distribution of Cs⁺ in the present Cs⁺-exchanged form (x = 1) with the pseudocubic structure, the XRD patterns were simulated on the three cation-distribution models presented in Table 1. The simulations were based on the assumptions that the present Cs⁺-exchanged form is isostrucural with the (Sr, H)-GTS and the possible occupied sites of its Cs⁺ ions are identical to those of Sr²⁺ ions in the (Sr, H)-GTS. Here, the model (A) assumes that Cs⁺ ions are distributed only to the 4e site, the model (B) that Cs⁺ ions occupy the 6g site to the maximum allowance and the remainder occupy the 4e site, and the model (C) that Cs⁺ ions are distributed on the two sites without any site-preference. Simultaneous incorporation of Cs⁺ ions into an equivalent position (e.g., coordinates: 0.96, 0.5, 0.5) on the 6g site and its nearest-neighbor equivalent position (e.g., the corresponding coordinates: 1.04, 0.5, 0.5) must be ruled out because the separation (∼ 0.6 Å) between the two positions is too short. The maximum allowance for occupancy of the 6g site should thus be 0.5, corresponding to 3 atoms per formula unit (apfu). This concept is applied to the model (B).

Table 1. Cation-distribution models of Cs⁺-exchanged form with x = 1.0

Models	Site occupancy factors of Cs+
	4e site*	6g site*
A	1.00	0
B	0.25	0.50
C	0.40	0.40

^*The Sr²⁺ sites (4e, 6g) in (Sr, H)-GTS with the cubic P43m structure determined from the single-crystal XRD technique (Spiridonova et al., 2011) were assumed as possible occupied sites of Cs⁺ in the present Cs⁺-exchanged form with x = 1.0 (Cs-GTS).

Figure 7 presents the comparison between the observed XRD pattern of the present Cs⁺-exchanged form with x = 1.0 and its simulated ones. The simulated XRD pattern from the model (A) exhibits the peak-intensity distribution manifestly disparate from the observed one. Notably, the 200 and 220 peaks are negligible in the simulated XRD pattern, whereas they are observed with the strongest and comparable intensities in the observed one, respectively. The 110 and 400 peaks appear with the strongest and moderate intensities, respectively, in the simulated XRD pattern, whereas they are negligible in the observed one. It is thus concluded that the model (A) should be excluded from further consideration. On the other hand, it is somewhat more challenging to ascertain which the model (B) or (C) the observed XRD pattern is more closely aligned with. However, the discrepancy between the two models can be attributed to the fact that the 110 and 400 peaks appear only with negligibly weak intensities in both the observed XRD pattern and the simulated pattern derived from the model (B), while they are clearly present with a certain degree of intensities in the simulated pattern derived from the model (C). This observation leads to the conclusion that the model (B), assuming that Cs⁺ ions prefer the 6g site in the assumed pseudocubic structure to the 4e site, is the most likely. By comparison, Er³⁺ ions in the Er³⁺-exchanged forms were reported to be distributed to both the 4e and 6g sites (Fujiwara et al., 2020), although the preference of Er³⁺ ions between the two sites was not referred.

Figure 7. Comparison of the observed XRD pattern of the present Cs⁺-exchanged form (x = 1.0: Cs-GTS) with its simulated ones from the cation-distribution models (A), (B), and (C) in Table 1. The left figure presents overall view of the XRD patterns in the measured 2θ range. The enlarged views for 110 and 400 peaks are provided in the middle and right figures, respectively. The simulation of XRD patterns was performed using the program Powder Cell (Kraus and Nolze, 1996). The diffraction indices are labeled assuming the pseudocubic structure.

The 6g site is located at the proximity of the centers of the 8-membered ring windows in the GTS framework. The previous studies (Harrison et al., 1995; Behrens et al., 1996) reported from the structure analyses of (Cs, H)-GTS with a composition of Cs₃HTi₄O₄(SiO₄)₃·4H₂O that the mean bond length between a Cs⁺ ion on the 6g site and its eight adjacent O atoms forming the 8-membered ring window is <Cs-O> = 3.285 Å (Fig. 8) and the number of Cs⁺ ions occupying the 6g site is 3 apfu. On the other hand, although we do not exactly know the coordination number of Cs⁺ ions on the 6g site as positions of H₂O molecules are not determined at the present stage, the average number of H₂O molecules coordinating to one Cs⁺ ion on this site is estimated to be at most 1.5. This is based on the present TG results showing that the H₂O content in the present Cs⁺-exchanged powder sample (x = 1.0) after excluding the above-mentioned excess water (1) is estimated to be 4.5 mpfu from its regression line (Figs. 4b and 4c). Thus, the average coordination number of Cs⁺ ions on the 6g site is estimated to be 9.5 from the sum of the number of coordinating framework O atoms and H₂O molecules. Even if the coordination number of Cs⁺ ions is assumed as 10, higher than the average value of 9.5, the mean observed <Cs-O> bond length on the 6g site of 3.285 Å is close to and, strictly speaking, somewhat longer than the expected ^XCs-O bond length (3.22 Å) from the ionic radii (Shannon, 1976). This suggests that the Cs-O bonds are somewhat weaker than those of typical ionic Cs-oxides. A comparable phenomenon is also identified in the Cs⁺-exchanged Ca-chabazite studied by Kawata et al. (2024). They proposed that such a moderate bond strength, i.e., a moderate window size would be a crucial factor to highly effective Cs⁺-exchange, which requires the equilibrium between the progressive diffusion and fixing of Cs⁺ ions within chabazite crystal during Cs⁺-exchange. In terms of this idea, it follows that in the present Cs⁺-exchanged forms, the large increase in V with increasing x (Fig. 6) despite the large reduction in H₂O contents (Fig. 4) can be because such moderate bonding forces acting between a Cs⁺ ion on the 6g site and adjacent O atoms forming the 8-membered ring window directly influence the framework structure. In contrast, in the Er³⁺-exchanged forms, the bonding forces acting between an Er³⁺ ion on the 6g site and adjacent O atoms forming the 8-membered ring window are inferred to be too weak since the window size is too large for the cation size of Er³⁺. The fact that the V in the Er³⁺-exchanged forms does not depend significantly on x (Fig. 6) can be because such weak Er-O bonding forces, along with almost the unchanged H₂O contents (Fig. 4c), do not significantly influence the framework structure.

Figure 8. Coordination environment of a Cs⁺ ion on the 6g site projected along [010] and its Cs-O bond lengths (Å) in (Cs, H)-GTS reported by Harrison et al. (1995).

ACKNOWLEDGMENTS

This work was supported by JSPS KAKENHI Grant Numbers JP24561005, JP16K06927, JP19K05349, and JP20H02683.

REFERENCES

•  Behrens,  E.A.,  Poojary,  D.M. and  Clearfield,  A. (1996) Syntheses, crystal structures, and ion-exchange properties of porous titanosilicates, HM₃Ti₄O₄(SiO₄)₃·4H₂O (M = H⁺, K⁺, Cs⁺), structural analogues of the mineral pharmacosiderite. Chemistry of Materials, 8, 1236-1244.
•  Behrens,  E.A. and  Clearfield,  A. (1997) Titanium silicates, M₃HTi₄O₄(SiO₄)₃·4H₂O (M = Na⁺, K⁺), with three-dimensional tunnel structures for the selective removal of strontium and cesium from wastewater solutions. Microporous Materials, 11, 65-75.
•  Dadachov,  M.S. and  Harrison,  W.T.A. (1997) Synthesis and crystal structure of Na₄[(TiO)₄(SiO₄)₃]·6H₂O, a rhombohedrally distorted sodium titanium silicate pharmacosiderite analogue. Journal of Solid State Chemistry, 134, 409-415.
•  Fujiwara,  K.,  Shiode,  T.,  Sugimoto,  H.,  Nakatsuka,  A., et al (2010) Hydration state of GTS-type titanosilicate (K, Na, H)₄Ti₄Si₃O₁₆.nH₂O fine particles. Topics in Chemistry and Material Science: Advanced Micro- and Mesoporous Materials-09, 4, 184-190.
•  Fujiwara,  K.,  Tamaki,  T.,  Kishimori,  C.,  Titorenkova,  R., et al (2013) Crystal structures of cobalt-exchanged sodium GTS-type titanosilicates and the elution test by acid solution. Transactions of the Materials Research Society of Japan, 38, 455-458.
•  Fujiwara,  K.,  Shinohara,  K. and  Nakatsuka,  A. (2017) Hydrothermal synthesis and Er³⁺ ion exchange of sodium GTS-type titanosilicate. Journal of Physics: Conference Series, 950, 042048.
•  Fujiwara,  K.,  Kawata,  N. and  Nakatsuka,  A. (2020) Preparation and crystal structural properties of Er³⁺-exchanged GTS-type sodium titanosilicate. Journal of Mineralogical and Petrological Sciences, 115, 59-64.
•  Harrison,  W.T.A.,  Gier,  T.E. and  Stucky,  G.D. (1995) Single-crystal structure of Cs₃HTi₄O₄(SiO₄)₃·4H₂O, a titanosilicate pharmacosiderite analog. Zeolites, 15, 408-412.
•  Kawata,  N.,  Furuhashi,  R.,  Fujiwara,  K.,  Ohkawa,  M. and  Nakatsuka,  A. (2024) Single-crystal X-ray diffraction study of a largely Cs⁺-exchanged natural Ca-chabazite: crystal-chemical factors for its excellent Cs⁺-exchange ability. Microporous and Mesoporous Materials, 379, 113262.
•  Kostov-Kytin,  V.,  Ferdov,  S.,  Kalvachev,  Yu.,  Mihailova,  B. and  Petrov,  O. (2007) Hydrothermal synthesis of microporous titanosilicates. Microporous and Mesoporous Materials, 105, 232-238.
•  Kraus,  W. and  Nolze,  G. (1996) POWDER CELL — a program for the representation and manipulation of crystal structures and calculation of the resulting X-ray powder patterns. Journal of Applied Crystallography, 29, 301-303.
•  Shannon,  R.D. (1976) Revised effective ionic radii and systematic studies of interatomic distances in halides and chalcogenides. Acta Crystallographica, A32, 751-767.
•  Spiridonova,  D.V.,  Krivovichev,  S.V.,  Yakovenchuk,  V.N. and  Pakhomovsky,  Ya.A. (2011) Crystal structures of the Rb- and Sr-exchanged forms of ivanyukite-Na-T. Geology of Ore Deposits, 53, 670-677.