Structural and ion exchange properties of nanocrystalline Si-doped antimony pyrochlore

Abstract

Antimonic acid Sb₂O₅·4H₂O with the pyrochlore structure is a well known ion-exchanger that is selective for Sr²⁺ in mildly acidic solution. An enhancement of the ion exchange properties of antimony-based pyrochlore materials, with respect to Cs⁺ and Sr²⁺, was previously reported by Möller and co-workers to be achievable if Si⁴⁺ is incorporated during synthesis of these materials (T. Möller, R. Harjula, M. Pillinger, A. Dyer, J. Newton, E. Tusa, S. Amin, M. Webb and A. Araya, J. Mater. Chem., 2001, 11, 1526–1532). The present study aims to shed light on the role of Si⁴⁺ in the pyrochlore structure and its influence on the ion-exchange properties. A series of samples were prepared with increasing concentrations of Si added during preparation. X-Ray powder diffraction and electron microscope observations indicate that the average particle size of the antimony pyrochlores decreases as Si concentration increases. Refinement of X-ray powder data shows a small and abrupt decrease in unit cell volume at an added Si concentration corresponding to about 10 atom%. For undoped samples ²⁹Si solid-state MAS NMR showed only a single sharp resonance at −75 ppm assigned to Si(OH)₄ species located within the hexagonal channels of the pyrochlore structure. The increasing addition of Si to the reactant mixture gave rise to a progressive increase in the intensity of resonances from hydrated amorphous silicate species. Base treatment succeeded in removing only the resonances due to these silicate species, leaving the resonance assigned to structural Si unaffected. These data are taken as confirmation of the assignment of the sharp −75 ppm resonance to isolated species Si(OH)₄ within the pyrochlore tunnels. Neutron powder diffraction of partially dehydrated samples shows the existence of two pyrochlore phases that appear to differ slightly in their unit cell parameters and confirms that incorporation of Si⁴⁺ into the pyrochlore structure and a reduction in water content cause volume contraction of the cubic cell. The slight unit cell contraction resulting from Si incorporation is hypothesised to be responsible for enhanced Cs⁺ selectivity of samples containing Si.

---

Introduction

Materials with the pyrochlore structure have been extensively studied for a range of applications including as adsorbents,^1,2 radioactive waste form materials,^3–5 as fast ion conductors,⁶ as Li-battery electrodes and recently for the photocatalytic splitting of water.^7–10

The ideal defect pyrochlore structure has cubic symmetry (space group Fd3m) and stoichiometry A₂M₂X₆X′ where A is a large, low valent cation (e.g. lanthanide or alkali metal or alkaline earth cation) and M is a smaller cation that can adopt octahedral coordination (e.g. Ti⁴⁺, Zr⁴⁺, W⁶⁺, Sb⁶⁺). Typically X is O²⁻ while X′ may be an anion such as O²⁻, OH⁻ or F⁻. The MX₆ octahedra share corners to form a three-dimensional framework possessing tunnels running down the c-axis in which the A cations are located (Fig. 1). Typically, the M cation is located in the 16c Wyckoff position (0, 0, 0) and the X anion is in 48f sites (x, 1/8, 1/8). The location of the A cation is variable but is usually in the 16d site (0.5, 0.5, 0.5) although other assignments such as 8b (3/8, 3/8, 3/8), or 32e (x, x, x) are possible. The X′ anion is usually ascribed to the 8b site. In hydrated materials, water molecules can also be distributed on the 8b, 16e or 32e positions.

Fig. 1 Graphical representation of the crystal structure of pyrochlore. The A cations are depicted as orange spheres on the 8b position while putative tunnel Si(OH)₄ groups are shown as very small tetrahedra consisting of Si atoms on 8b (3/8, 3/8, 3/8) and water oxygen atoms on 32e (0.421, 0.421, 0.421).

Largely because of this tunnel structure, pyrochlore compounds display numerous interesting properties including rapid ion transport and pressure-induced volume expansion.¹¹ It is also well known that, especially for pyrochlore materials produced using soft chemical methods, and therefore having small particle sizes, the A cations can be exchanged relatively easily from solution by a range of other cations.

It has been known for some time that antimonic acids having the pyrochlore structure are cation exchangers with some unique selectivities.² Of particular interest is their high selectivity for Sr²⁺ from mildly acidic solution. This is a rare property of significance for the removal of ⁹⁰Sr from acidic radioactive waste streams.

Although the basic ion-exchange properties of hydrous antimony oxides (Sb₂O₅·4H₂O) were described early on, the effect of variations in composition had not been extensively investigated until recently. Moller and coworkers, and others,^12–21 have undertaken extensive and detailed studies of the ion-exchange properties of antimony pyrochlore-based materials in which some of the Sb⁵⁺ was replaced by a variety of metals including W⁶⁺, Si⁴⁺, Sn⁵⁺ and Ti⁴⁺.

These investigations appeared to indicate that these materials exhibited increased selectivity for both ⁹⁰Sr and ¹³⁷Cs, two problematic nuclei for radioactive waste pretreatment, when Si⁴⁺ was incorporated into the antimony pyrochlore structure.¹² In addition it was demonstrated that the selectivity of these antimony silicates for a range of radioisotopes could be modulated by incorporating codopant species such as W. This property is extremely attractive as it means that only one ion-exchange material is required to remove both radioisotopes which dominate the activity in older radioactive waste liquids deriving from spent fuel reprocessing and uranium target irradiations. The generation of a single waste composition simplifies disposal of the spent ion-exchange material. Even though these Si⁴⁺-doped antimony pyrochlore materials have these unique ion-exchange properties little understanding so far exists regarding the role of Si⁴⁺ in the structure as the poorly crystalline nature of these materials precludes detailed analyses aimed at linking structure and function. Indeed it is not clear if Si⁴⁺ is in fact intimately associated with the pyrochlore structure at all. For natural samples Lumpkin and Mariano²² have argued that Si⁴⁺ species exist in framework sites. While six-coordinate Si is not unknown, it is a relatively rare occurrence.

The aim of the present study is to ascertain if Si⁴⁺ is actually present within the antimony pyrochlore structure, and if so, what structural sites silicon may occupy. An additional aim of the study is to elucidate how Si incorporation influences the Cs⁺ and Sr²⁺ selectivity of the materials.

Experimental

The antimony pyrochlores were prepared as previously described.¹² Briefly, for the undoped material 15 mL of Millipore water was added to a mixture of 18 g SbCl₅ in 31.5 g 4 M HCl. The final volume of the mixture was made up to 300 mL using Millipore water. On addition of water, precipitation occurred and the resulting suspension was added to two separate 45 mL Teflon lined autoclaves which were heated to 70 and 150 °C respectively for two days. A series of antimony pyrochlores were prepared containing variable amounts of Si⁴⁺ ranging from 5–50 atom%. In addition to preparing samples at 70 °C (SiSb-70-x; x = Si atom%) as previously described,¹² samples with similar Si⁴⁺ contents were also prepared under hydrothermal conditions at 150 °C (SiSb-150-x) in an attempt to improve crystallinity, and therefore also improve the reliability of structure refinements. The Si⁴⁺-doped samples were prepared in a similar way only sodium silicate solution was combined during the initial water addition.

The ion-exchange behavior of the sorbents were determined on the as-prepared ion-exchangers using the batch contact method with solutions containing 0.75 mmol L⁻¹ of Cs⁺ and Sr²⁺ in 1.0 M HNO₃. The as-prepared exchangers therefore contain only H⁺ in the tunnels sites for the undoped pyrochlore sample and a combination of H⁺ and Na⁺ for the samples containing Si. Since the matrix solution is 1.0 M HNO₃ it is considered unlikely that this variation in initial exchangeable cation content will alter the distribution coefficients in any significant way. Al-Attar and coworkers²³ has shown that the variation in K_d for Cs and Sr is not affected dramatically for low concentrations of Na⁺. Typically, for the equilibrium batch contact data, powdered sorbent (200 mg) and a given aqueous cation solution (20 ml) were contacted at 25 °C for 24 h with constant agitation after which time the supernatant was removed, filtered (0.45 µm), and equilibrium metal cation concentrations analysed by ICPMS. Specific solution concentrations of metal species are given in the captions of the figures. Determination of the metal cation concentrations before and after contact with a given sorbent allows the calculation of a distribution coefficient (K_d) for the given metal cation with respect to the sorbent by the formula:

where C_i = initial cation concentration; C_f = final cation concentration; V = volume of solution contacted with sorbent; m = mass of sorbent employed; F = form factor to normalize for hydration of the given sorbent or in other words the fraction of dried material in the sample. In this work the hydration of the sorbent has not been considered in the determination of K_d because it is more-or-less constant for all samples.

Survey X-ray powder diffraction patterns were recorded on a Scintag X1 diffractometer using Cu K_α radiation and a Peltier detector. Data for unit cell refinements were recorded on a Panalytical X′Pert Pro diffractometer using the same wavelength radiation and employing a solid state detector utilizing real time multiple strip technology.

To carry out neutron diffraction measurements on samples, the water in all the samples was exchanged with deuterium oxide in order to eliminate incoherent scatter and high background. Patterns were recorded with samples sealed in standard 11 mm diameter deep-drawn cylindrical vanadium cans which were not hermetically sealed. The cans were initially evacuated while survey scans were recorded on the ISIS HRPD diffractometer of the Rutherford Appleton Laboratory. On removal from vacuum, the sample cans were sealed with parafilm prior to insertion into the evacuated furnace (10⁻⁵ mbar) used to obtained room temperature and 180 °C data on the Polaris powder diffractometer at ISIS. Prior to insertion into the furnace, the bolts on the sample cans were loosened in order to guarantee that the samples were under vacuum while data were taken.

Powder pattern refinements of X-ray and neutron data were carried out using the Rietveld method as implemented in the program Rietica²⁴ which uses the LHPM engine.

Thermogravimetric analyses (TGA) and differential thermal analsyses (DTA) were conducted simultaneously on a Setaram TAG24 (France) instrument.

Transmission electron microscopy (TEM) was conducted using a JEOL 2000FXII instrument operating at 200 keV. TEM specimens were prepared by lightly grinding a small amount of powder in ethanol to form a suspension which was dropped by pipette onto holey carbon coated copper TEM grids which were allowed to dry in air. The chemical compositions of individual grains were determined using an Oxford Instruments LINK ISIS energy dispersive X-ray analysis (EDX) system attached to the microscope.

²³Na and ²⁹Si MAS NMR spectra were acquired at ambient temperatures on a Bruker MSL-400 spectrometer (B_o field of 9.40 T), operating at the ²³Na and ²⁹Si frequencies of 105.808 and 79.484 MHz, respectively. The ²³Na data were acquired using single pulse (Bloch decay) experiments with a Bruker 4 mm double-air-bearing probe from which MAS frequencies of ∼15 kHz were achieved. Non-selective π/2 pulse times of 2.5 µs were measured on a 1.0 M NaCl solution, from which selective pulse times of ∼0.8 µs were employed for data acquisition on all solid samples. Recycle delays of 3 s were typically used, however, the quantification and speciation of these data obtained with these short delays were verified with experiments which had recycle delays extending to 30 s. All ²³Na chemical shifts were referenced to 1.0 M NaCl which was set to δ 0.0 ppm. ²⁹Si MAS NMR data were acquired using a Bruker 7 mm double-air-bearing probe with cross-polarisation (CPMAS) and single pulse (Bloch decay) methods, in which both pulse sequences utilised high-power ¹H decoupling during data acquisition. The MAS frequencies implemented for these measurements were ∼5 kHz. For the ²⁹Si CPMAS experiments a recycle delay of 5 s, a ¹H–²⁹Si Hartmann–Hahn contact period of 5 ms and an initial ¹H π/2 pulse width of 5 µs were common to all CPMAS spectra. For the corresponding ²⁹Si MAS single pulse/high power ¹H decoupling measurements, a ²⁹Si π/4 pulse width of 2.5 µs was employed in conjunction with recycle delays of 30 s. All ²⁹Si MAS and CPMAS chemical shifts were externally referenced to tetramethylsilane (TMS) via a high purity sample of kaolinite, which was also used to establish the ¹H–²⁹Si Hartmann–Hahn condition.

Results and discussion

The SiSb-150-x samples were investigated for their selectivity for Cs⁺ and Sr²⁺ from solutions containing 0.75 mmol L⁻¹ of Cs⁺ and Sr²⁺ in 1.0 M HNO₃. It is apparent from the plot in Fig. 2 that a maximum occurs in the distribution coefficient (K_d) for both Cs⁺ and Sr²⁺ for samples with x = 5 and 10. This observation appears to be consistent with previous assertions of enhanced selectivity as Si⁴⁺ content increases. However, it is to be noted that these apparent increases are not overly significant with respect to the uptake (mmol g⁻¹ basis) of each cation. Lower selectivity increases on Si-doping compared with the reports by Moller et al.¹² might be because in that study only trace level concentrations of Cs⁺ and Sr²⁺ were being investigated.

Fig. 2 Selectivity of Si⁴⁺-doped antimonate pyrochlores prepared at 150 °C (SiSb-150-x; x = 0–15) for Cs⁺ (■) and Sr²⁺ (□) as a function of added atom% of Si⁴⁺. Lines through the points are a guide for the eye.

Thermogravimetric analyses of the SiSb-150-x (x = 0–15) samples are presented in Figs. 3a–c. There is only a slight variation in water content as a function of x within this range. As outlined by Moller and co-workers¹² for similar samples prepared at 70 °C, the materials display a high level of hydration, ca. 16–18%, with the majority of H₂O contained within the tunnels of the structure. Water is eliminated from the SiSb-150-x materials in several distinct steps and is in line with the investigations of Riviere et al. on hydrous antimonic acid (Scheme 1).²⁵ The initial mass loss of about 5% occurs around 100 °C and is observed at all levels of doping. This is probably due to weakly bound structural water, either within the tunnels or bound to particle surfaces. A further 5–10% of more strongly bound water is eliminated between about 100 and 300 °C and is presumably due to water in the tunnel positions. Between about 400 and 600 °C an additional 2% weight loss occurs and this is likely due to the elimination of hydroxyl groups also in tunnel positions.

Fig. 3 TGA (–), DTG (– · –) and DTA (⋯) traces for SiSb-150-x samples with x values of (a) 0, (b) 5, and (c) 15.

Scheme 1 Thermal decomposition of hydrous antimonic acid according to Riviere et al.²⁵

X-Ray powder patterns were recorded of the SiSb-70-x and SiSb-150-x samples and these are shown in Fig. 4. The samples prepared at 150 °C display narrower line widths compared with those prepared at 70 °C, particularly for higher Si⁴⁺ contents. Increased line broadening as a function of Si⁴⁺ content is visually apparent from the patterns but is clearly seen in Fig. 5a. Use of the Scherrer equation on patterns of the samples prepared at 150 °C gave average particle diameters of 54, 43, 30 and 22 nm for the samples with x = 5, 10, 20, and 40 respectively. Such values are in line with qualitative estimates based on the TEM images (see Fig. 7 below). For the SiSb-150-0 sample, a typical pyrochlore pattern is also observed. However, a weak shoulder is visible on the low-angle side of the characteristic pyrochlore reflection at 29.9° 2θ (arrow). No such reflection should be observed for the Fd3m space group. The refinement of these data using a single phase model is relatively poor compared to the Si⁴⁺-doped samples and therefore the question arises as to the reasons for this.

Fig. 4 Survey X-ray diffraction patterns of SiSb-70-x pyrochlores with x values of (a) 5, (b) 10, (c) 20, and (d) 40 and SiSb-150-x samples with x values of (e) 0, (f) 5, (g) 10, (h) 20, and (i) 40.

Fig. 5 (a) Line width variations as full width at half maximum (FWHM) for SiSb-70-x (■) and SiSb-150-x (□) and (b) a-dimension for SiSb-150-x samples dried at 70 °C then exposed to ambient air (◆) and SiSb-150-40 sample heated to 400 °C and allowed to rehydrate (◇). Data in (b) are the average of several separate determinations.

The results of Rietveld refinements of the X-ray powder data as a function of x are shown in Fig. 5b. Multiple determinations of the a-parameter were obtained from different data sets for each sample. Attempts to include Si⁴⁺ in the A or M element positions did not result in any significant improvement in the quality of the refinements (Fig. 6). There appears to be an initial slight reduction in a-parameter on going from x = 5 to 10. This indicates that there is some limit to the effect that Si⁴⁺ incorporation has on the lattice dimensions, or in other words, a possible solid solution limit is being reached. The TEM observations to be discussed indicate clearly that not all of the added Si finds its way into the structure when x exceeds 5.

Fig. 6 Rietveld refinement of X-ray pattern of the SiSb-150-5 sample using parameters similar to those in Table 1.

TEM images of the SiSb-150-x (x = 0–15) samples are shown in Fig. 7. The SiSb-150-0 sample was found to be quite homogeneous and consisted of 50 to 150 nm rounded particles that display a degree of faceting (Fig. 7a). This is consistent with the broadened XRD patterns observed for this material. Detailed analyses of the elemental ratios indicated an average composition of 0.4 atom% Si, 23.2 atom% Sb. No sodium was present in this sample because sodium is only incorporated on addition of the sodium silicate solution used for the Si⁴⁺-doping. Charge balance is therefore most likely to be maintained by H₃O⁺. The very small amount of Si found in these samples has been attributed to an impurity.

Fig. 7 Bright field TEM micrographs of SiSb-150-x samples with x values of (a) 0, (b) 5 and (c) 15.

For the SiSb-150-5 sample (Fig. 7b) the observed particles appear grouped into two size regimes. Smaller rounded particles have a size distribution estimated to be between 35 and 210 nm while larger more faceted particles with diameters of about 750 nm are also observed. As the Si content increases, the proportion of non-faceted particles increases. The smaller particles have composition Na_0.21Si_0.10Sb₂ while the larger particles have a lower Na and Si content analysing as Na_0.043Si_0.06Sb₂. No evidence could be found for bulk, secondary Si-rich phases in this sample suggesting that all the available Si is associated with the pyrochlore particles, either structurally or bound to the nanoparticle surfaces.

The SiSb-150-15 sample (Fig. 7c) consisted of small rounded particles with average diameter in the 20–100 nm range and larger particles with average diameter of about 750 nm as observed in the SiSb-150-5 sample. In addition, regions of fine-grained polycrystalline silica could be readily observed. EDS analysis indicated that the composition of the smaller particles was Na_0.42Si_0.23Sb₂ while the larger particles had comparable Na content and lower Si content analysing as Na_0.43Si_0.063Sb₂. This is consistent with the results observed for the SiSb-150-5 sample; the larger particles in the sample are depleted in silicon relative to the smaller particles. For the SiSb-150-20 sample, even larger domains of hydrous amorphous silica are observed.

Silicon-29 solid-state NMR spectra of the SiSb-150-x (x = 0–40) series are shown in Fig. 8. At low reactant Si contents (e.g. SiSb-150-5) only a single sharp resonance is observed with a chemical shift of about −75 ppm (Fig. 8a). As the Si⁴⁺ content increases to x = 10 and above, several additional resonances are observed in the chemical shift range −85 to −120 ppm. These upfield resonances are consistent with assignment to Q¹–Q⁴ silicon atoms as is generally observed for hydrous amorphous silica and is consistent with TEM observations of extraneous bulk silica. Since the narrow low-field resonance is clearly not consistent with hydrous silica, it must be concluded that this Si⁴⁺ species is somehow intimately associated with the pyrochlore particles. This association could take the form of a surface coating or as isolated Si⁴⁺ within the tunnels of the framework.

Fig. 8 Si-29 MAS solid-state NMR spectra of Si⁴⁺-doped Sb-pyrochlores prepared from reactant mixtures with (a) 5, (b) 10, (c) 20, and (d) 40 mol% Si in reactant. Spectrum (e) was recorded using cross polarization with a contact time of 2 ms.

Silicon-29 spectra obtained using cross polarization from protons (Fig. 8e) continued to show spectral intensity around −100 ppm employing contact times of 2–15 ms. However, the Q¹ resonances from extraneous hydrous silica together with the sharp resonance at δ = −75 ppm from the putative pyrochlore-bound Si are only weakly excited using this technique. Weak excitation of these resonances using this technique suggests that there are few protons in the vicinity of these Si sites. A possible reason for the weak excitation of Q⁰ (pyrochlore) or Q¹ (silicate) species which must be bound to OH groups could be due to rapid proton exchange.

It is well known that hydrous amorphous silica can be readily dissolved in alkaline solutions. If the resonances between −85 and −120 ppm and the sharp resonance at −75 ppm are indeed from extraneous silica then it would be expected that treatment of samples containing significant amounts of this phase with 1.0 M sodium hydroxide should remove these resonances. The effect of base treating an SiSb-70-50 sample is shown in the ²⁹Si MAS NMR spectra of Fig. 9. What is observed is the removal of resonances assigned to hydrous silicate species but persistence of the sharp low field resonance tentatively assigned to Q⁰ (pyrochlore). In fact, this resistance to dissolution also excludes the possibility that the Si⁴⁺ species responsible for this resonance originates from silicate species coating the surfaces of the nanocrystalline pyrochlore since these would also be expected to be removed by this treatment.

Fig. 9 SiSb-70-50 pyrochlore before (a) and after (b) treatment with 1 mol L⁻¹ NaOH solution.

If the narrow sharp resonance really is due to structural Si⁴⁺ in the pyrochlore as suggested, and if this Si⁴⁺ species is undergoing rapid exchange with protons, then dehydration of the pyrochlore would be expected to have significant impact on the resonance. In Fig. 10a is shown the spectrum of the SiSb-150-40 sample after dehydration at 400 °C recorded in the dried state by using a special air-tight Teflon insert within the NMR rotor which was prepared in the absence of atmospheric moisture. It is apparent that the sharp resonance is not observed in this dehydrated state. When the sample is allowed to rehydrate in air this resonance reappears and eventually reaches almost its full intensity relative to the bulk hydrous silicate resonances, albeit over an extended period of time at ambient humidity (Figs. 10b–d). The slow kinetics of this rehydration process would only be expected if the species were located within the tunnels of the pyrochlore structure as access to these sites would be diffusion limited.

Fig. 10 The effect of rehydration on the ²⁹Si spectrum of sample SiSb-150-40: (a) sample dehydrated at 400 °C, and rehydrated for (b) several days, (c) three weeks and (d) seven weeks.

To ensure that the effects observed in the ²⁹Si NMR spectra of the dehydrated samples are not due to structural transitions induced by heating, the XRD pattern of an SiSb-150-40 sample that had been heated to 400 °C was recorded. The pattern of this dehydrated sample recorded immediately after dehydration could be adequately refined using similar unit cell parameters to the hydrated pyrochlore except that a significant reduction in a-parameter occurs on dehydration. Subsequent rehydration in ambient air for extended periods of time did not cause any increase in a-parameter regardless of the rehydration time (Fig. 5b) and following the reappearance of narrow ²⁹Si resonance at −75 ppm. This leads to the conclusion that both the state of hydration and the Si content of the samples determine the cell volume of the antimony pyrochlores, with the former having by far the greatest impact. Indeed, the fact that there is no appreciable difference in water content as a function of Si content is consistent with this observation. The fact that rehydration restores the −75 ppm resonance but does result in lattice reexpansion suggests that rehydration of the Si site within the tunnels can occur without swelling of the pyrochlore tunnels by water molecules.

The ²³Na NMR spectra (Fig. 11) of the series of SiSb-150-x samples do not show a multiplicity of ²³Na resonances or line width increases as x increases. However, a small monotonic downfield movement of the ²³Na chemical shift by about 3 ppm is observed with increasing x. This downfield ²³Na shift trends with the small monotonic downfield shift of the resonance assigned to Q⁰ (pyrochlore) and the slight contraction of the cell volume. For the ²³Na case, this trend to higher deshielding reflects a reduced electron density at the Na⁺ cation. For this to occur there would need to be increased electron density in the bonds between the Na⁺ cations and either water molecules from the hydration sphere and/or oxygen atoms of the framework. This apparent deshielding of the Na⁺ cations can be attributed to a contraction in tunnel dimensions which we propose is being driven by partial replacement of more bulky hydrated Na⁺ cations by a more compact silicate species, possibly Si(OH)₄. Such a scenario would not result necessarily in any reduction in total water content since both species could be hydrated to similar extents.

Fig. 11 ²³Na MAS solid-state NMR of SiSb-150-x with x values of (a) 5, (b) 10, (c) 15, (c) 20.

Since the scattering factor of silicon and oxygen for neutrons is much higher than it is for X-rays, it is reasonable to expect that neutron diffraction patterns of the present samples would show greater sensitivity to changes in the tunnel composition. Neutron diffraction patterns of samples that had been D₂O-exchanged are shown in Fig. 12. There is a clear splitting of many of the reflections in the pattern of the SiSb-150-0 (not shown) and SiSb-150-5 (Fig. 12a) samples. This splitting is removed when the Si⁴⁺ content increases beyond 5 atom%. For example, the pattern of the SiSb-150-15 (Fig. 12d) sample is consistent with the presence of a single crystalline pyrochlore phase. While the TEM of the SiSb-150-0 sample indicated a single morphology and composition, those of the SiSb-150-5 and SiSb-150-10 samples showed morphological and compositional variations in the antimonate phases. In particular, the SiSb-150-10 sample, for which two compositionally distinct morphologies (Na_0.42Si_0.23Sb₂ and Na_0.39Si_0.062Sb₂) were observed by TEM, appeared as a single phase by neutron diffraction. It can be concluded from this that the structural variations that are observed have a greater dependence on variations in water content than composition.

Fig. 12 (a) Single phase neutron refinement of room temperature data of unheated SiSb-150-5, (b) two-phase refinements of data from (a), (c) single-phase refinement of neutron powder data for SiSb-150-5 at 180 °C, (d) single-phase neutron refinement of SiSb-150-15, (e) single phase X-ray refinement of SiSb-150-15. There appear to be two cubic pyrochlore structures with a-dimensions around 10.29 and 10.377 Å evident in the SiSb-150-5 neutron pattern. The pattern with the largest a-parameter disappears at 180 °C.

In an attempt to shed light on the origins of the additional reflections observed in the patterns of SiSb-150-0 and SiSb-150-5, neutron diffraction patterns were recorded at room temperature and during heating at 180 °C for both the SiSb-150-5 and SiSb-150-15 samples. As shown by the thermogravimetric analyses, heating at 180 °C in air is sufficient to remove only the more loosely bound physisorbed water. The present neutron diffraction patterns, however, were acquired in vacuum (experimental requirement) for both unheated and heated samples. Such conditions probably assist in the removal of some of the tunnel water even when no heating is involved. Nevertheless, it was anticipated that heating might remove any differences caused by inhomogenous hydration of the sample. Indeed, the neutron data of the SiSb-150-5 sample heated at 180 °C (Fig. 12c) shows that more complete removal of tunnel water also eliminates additional reflections from the second phase pattern with the larger a-parameter.

The neutron diffraction pattern of the hydrated SiSb-150-15 sample recorded at room temperature (Fig. 12d) was refined using as a starting point the parameters derived from the X-ray refinement. The quality of the refinement is reasonable, but there exists a small degree of misfit on the low-angle side of the three lowest angle reflections. The goodness-of-fit can be improved somewhat if water is included in the 32e sites (Table 1). The refinement was relatively insensitive to the inclusion and location of Si⁴⁺ although a slightly better fit was obtained when Si⁴⁺ was included in the 8b site. This siting implies 4-fold coordination of Si⁴⁺ to four tunnel water molecules as in Fig. 1. For the SiSb-150-5 sample (Fig. 12a), significant residual intensity is observed on the low-angle side of almost all of the reflections which indeed suggests the possible presence of an additional cubic pyrochlore phase rather than a change of symmetry. Inclusion of an additional cubic pyrochlore phase (Fd3m) in the refinement results in a significant improvement in the fit but is far from acceptable. The a-parameters for these two phases are 10.453 and 10.341 Å, respectively. The larger a-parameter is close to the values obtained from both the X-ray and neutron patterns of the hydrated SiSb-150-5 sample. However, the pattern remains quite difficult to refine and it is clear that the model is not totally adequate. The most likely interpretation of the multiphase nature of the undoped and unheated samples is that partial dehydration provided by the vacuum gives rise to particles with varying water contents and therefore varying cell parameters. When such a situation prevails satisfactory refinement is virtually impossible and therefore only qualitative conclusions can be drawn.

Table 1 Best fit model for the neutron diffraction pattern of SiSb-150-15 recorded at room temperature

	x	y	z	Wyckoff
Na	1/2	1/2	1/2	16d
Sb	0	0	0	16c
O	0.3196	1/8	1/8	48f
Si	3/8	3/8	3/8	8b
O (H2O)	0.4213	0.4213	0.4213	32e
R p	2.043
R wp	3.275

---

The cell with the larger a-parameter that is observed in the neutron pattern of the hydrated undoped and low-doped samples is eliminated either on complete dehydration under vacuum at 180 °C or on increased Si⁴⁺-doping to give only the cell with the smaller a-parameter. It would therefore seem that removal of water from the more hydrated phase causes unit cell contraction and this is within expectations. The fact that volume contraction is also observed in the neutron pattern on Si⁴⁺-doping is consistent with the X-ray data which we propose is due to replacement of some of the solvated Na⁺ by small silicate species such as Si(OH)₄. In aqueous solutions such silicic acid species are not inherently stable and undergo condensation to form oligomeric species. However, in the constrained environment afforded by the pyrochlore tunnels, and given the fact that the tunnel occupancy is low, it is quite conceivable that the silicic acid species is stable. Such a species would also have fairly dissociated protons which would explain the difficulty in cross polarization of its ²⁹Si NMR resonance.

An alternative location for the Si⁴⁺ species giving rise to the sharp NMR signal at −75 ppm is in structural sites substituting for the Sb⁵⁺ as has been suggested previously.²² The ionic radius for Sb⁵⁺ in six-fold coordination is 0.60 Å while for Si⁴⁺ in four- and six-fold coordination the ionic radii are 0.26 and 0.40 Å. This difference in ionic radii corresponds to a longer Sb–O bond length in undoped antimony pyrochlores (1.98 Å) compared to the maximum Si–O bond length observed for six-coordinate Si in silicates which range from about 1.75 to about 1.800 Å for oxygen coordination numbers ranging from two to about four.²⁶ This together with the large difference in ionic radii between Sb⁵⁺ and Si⁴⁺ would tend to suggest that if some Si⁴⁺ replaces Sb⁵⁺ in the pyrochlore lattice then significant volume contraction might be expected which is not what is observed experimentally. On the other hand the ionic radii of Na⁺ in four- and six-fold coordination are 0.99 and 1.02 Å, respectively, so that if Si(OH)₄ were to replace some of the solvated Na⁺ in the tunnels then this would also result in volume contraction. It is to be noted, however, that our data points to the actual Si-doping levels achieved using the present procedures being very low and so the effect of Si incorporation within the tunnels should not be large. At the very least, substitution of significant amounts of Si⁴⁺ for Sb⁵⁺ would be expected to result in measurable structural distortion and volume contraction and this was not observed. On the basis of these arguments therefore, substitution of Si⁴⁺ for Sb⁵⁺ would seem untenable.

Perhaps more compelling evidence against Si⁴⁺ for Sb⁵⁺ substitution comes from the ²⁹Si NMR data. Solid state ²⁹Si NMR chemical shifts for relatively rare six-fold coordinated Si are well established and range from −140 to −220 ppm.²⁷ The crystalline mineral thaumasite (Ca₃Si(OH)₆(CO₃)(SO₄)·12H₂O) provides a pertinent example as the existence of six-coordinate silicon in this compound is unequivocal. Silicon in thaumasite is present as isolated [Si(OH)₆]²⁻ octahedra and thaumasite is the only mineral in which this type of coordination is stable at (or near) ambient pressures and temperatures.²⁸ The Si octahedron contains three equivalent Si–O bonds of 1.776 Å and three bonds of 1.790 Å and its chemical shift occurs at −180 ppm. This is significantly shifted upfield compared with isolated ^IVQ⁰ silicon species which resonate between ∼−75 and −85 ppm. By analogy with ^IVQⁿ, a progressive upfield shift of the ^VIQⁿ species to higher frequency would be expected with increasing n or the degree of condensation. Therefore the sharp resonance observed at about −75 ppm in the spectra of SiSb-150-x samples is definitely not due to six-coordinate Si in either the pyrochlore framework, tunnels or surfaces. Its most probable assignment is therefore to four-coordinate and isolated Si(OH)₄ within the pyrochlore tunnels.

The volume contraction resulting from the incorporation of putative Si(OH)₄ species into the pyrochlore tunnels seems also to be the likely reason for the initial increase in the selectivity of the pyrochlore framework for large ions such as Cs⁺ and Sr²⁺. Of these two ions however, it is the affinity for Cs⁺ that is influenced the most by the incorporation of Si⁴⁺. Even though the presence of tunnel species such as Si(OH)₄ might be viewed as an impediment to ion exchange, small concentrations of such species would not so much affect the selectivity as the capacity.

The nanocrystalline nature of the present samples results in XRD patterns and NMR resonances that are broad. This has the result that XRD and neutron powder diffraction data are not of sufficient quality to be able to distinguish between more complex structural models incorporating Si⁴⁺ and water molecules in various positions within the tunnels. It has recently been demonstrated that volume expansion can occur in hydrated pyrochlores containing large A cations (e.g. NH₄NbWO₆ and Rb₄NbWO₆) when pressure is applied due to movement of the A cations and water molecules within the tunnels.¹¹ In the present situation it is clear that volume contraction is mostly a result of the expulsion of water from the channels and that this process is not reversible for samples that are heated to relatively high temperatures (e.g. 400 °C).

Conclusion

In this work we have sought to furnish an understanding of the nature of the Si environment in nanocrystalline antimony pyrochlore samples in which Si has been included during preparation with the aim of improving adsorption properties. Using a combination of techniques that probe the local and longer range structure, it has been possible to definitively assign Si to the structure of the antimonate rather than on surfaces. Furthermore it has been possible to locate with some certainty the Si within the hexagonal tunnels defined by the antimonate framework. The positive influence that Si incorporation has on the selectivity of the antimony pyrochlore materials for Cs⁺ appears to be due to the resulting slight decrease in tunnel dimensions.

Acknowledgements

The authors are indebted to Dr Ron Smith of Rutherford Appleton Laboratories and Dr C. J. Howard of ANSTO for assistance with acquisition of the neutron data. Support for travel by CJH from the Access to Major Research Facilities Program is gratefully acknowledged.

References

1. L. G. Nagy, G. Torok and G. Foti, Proc. Int. Conf. Colloid Surf. Sci., 1975, 33–40.
2. M. Abe, T. Kotani and S. Awano, in Advances in Ion Exchange for Industry and Research, ed. P. A. Williams and A. Dyer, Royal Society of Chemistry, Cambridge, Special Publication, 1999, vol. 239, pp. 199–211.
3. R. C. Ewing, W. J. Weber and J. Lian, J. Appl. Phys., 2004, 95, 5949–5971.
4. R. C. Ewing, W. J. Weber and W. Lutze, NATO ASI Series 1: Disarmament Technologies, 1996, 4, 65–83.
5. R. C. Ewing, Can. Mineral., 2001, 39, 697–715.
6. W. A. England, M. G. Cross, A. Hamnett, P. J. Wiseman and J. B. Goodenough, Solid State Ionics, 1980, 1, 231–49.
7. J. Wang, Z. Zou and J. Ye, Mater. Sci. Forum, 2003, 423–425, 485–490.
8. Z. G. Zou and H. Arakawa, J. Photochem. Photobiol., A, 2003, 158, 145–162.
9. Z. Zou, J. Ye and H. Arakawa, J. Mol. Catal. A: Chem., 2001, 168, 289–297.
10. Z. Zou, J. Ye and H. Arakawa, Int. J. Hydrogen Energy, 2003, 28, 663–669.
11. P. W. Barnes, P. M. Woodward, Y. Lee, T. Vogt and J. A. Hriljac, J. Am. Chem. Soc., 2003, 125, 4572–4579.
12. T. Möller, R. Harjula, M. Pillinger, A. Dyer, J. Newton, E. Tusa, S. Amin, M. Webb and A. Araya, J. Mater. Chem., 2001, 11, 1526–1532.
13. P. Karhu, T. Möller, R. Harjula and J. Lehto, Ion Exchange at the Millennium, Proceedings of IEX 2000, 8th, Cambridge, United Kingdom, July 16–21, 2000, pp. 109–115.
14. T. Möller, R. Harjula and J. Lehto, Sep. Purif. Technol., 2002, 28, 13–23.
15. R. Harjula, P. Kelokaski, A. Paajanen, T. Möller and J. Lehto, Proceedings of the International Conference on Radioactive Waste Management and Environmental Remediation, 8th, Bruges, Belgium, Sept. 30–Oct. 4, 2001, pp. 509–512.
16. R. Harjula, J. T. Möller, S. Amin, A. Dyer, M. N. Pillinger, J. A. Newton, E. H. Tusa and M. Webb, WO Patent 9959161.
17. T. Möller, A. Clearfield and R. Harjula, Microporous Mesoporous Mater., 2002, 54, 187–199.
18. T. Möller and R. Harjula, ‘Uptake of radionuclides by metallosilicate ion exchangers’, Spec. Publ. – R. Soc. Chem., 1999, 239, 276–287.
19. R. Koivula, R. Harjula and J. Lehto, Microporous Mesoporous Mater., 2002, 55, 231–238.
20. R. Koivula, R. Harjula and J. Lehto, J. Mater. Chem., 2002, 12, 3615–3619.
21. U. Y. Koudsi and A. Dyer, J. Radioanal. Nucl. Chem., 2001, 247, 209–219.
22. G. R. Lumpkin and A. N. Mariano, Mater. Res. Soc. Symp. Proc., 1996, 412, 831–8.
23. L. Al-Attar, A. Dyer and R. Harjula, J. Radioanal. Nucl. Chem., 2004, 260, 199–203.
24. C. J. Howard and B. A. Hunter, A Computer Program for Rietveld Analysis of X-Ray and Neutron Powder Diffraction Patterns (NSW, Australia: Lucas Heights Research Laboratories), 1998, pp. 1–27.
25. M. Riviere, J. L. Fourquet, J. Grins and M. Nygren, Mater. Res. Bull., 1988, 23, 965–75.
26. F. Liebau, Structural Chemistry of Silicates: Structure, Bonding, and Classification, Springer-Verlag, Berlin, 1985, p. 347.
27. A. R. Grimmer, F. Von Lampe and M. Magi, Chem. Phys. Lett., 1986, 132, 549–53.
28. S. D. Jacobsen, J. R. Smyth and R. J. Swope, Phys. Chem. Miner., 2003, 30, 321–329.

---