Sol–gel obtained silicophosphates as materials to retain caesium at high temperatures

Abstract

Silicate-based and silicophosphate-based compounds containing caesium have been synthesized by a sol–gel process. The structural changes induced by thermal treatment have been studied by XRD, XRF, DRIFTS, thermogravimetric and XPS analyses. The solids crystallise upon heating and the behaviour of the crystallisation process depends on the P-content. A partial loss of caesium is observed from 900 °C in the case of the sample with no or too low P-content, but for the Cs–SiPO solids with higher P-content, no Cs or P loss are observed. It seems that there exists a threshold of P-content, above which the quantity of P is sufficient to retain caesium at least up to 900 °C. This work opens a new, promising way in potential waste confinement matrix research.

---

1 Introduction

Nuclear waste with Very High Activity (C-class) presents a great hazard for human beings and their environment and is nowadays a real problem for society. Fission products are a type of nuclear waste. Even if they are generally less radiotoxic than minor actinides, their mobility caused by lixiviation seems potentially more critical. In the case of caesium, the ¹³⁵Cs isotope should be stored at great depth since it possesses a period of 2,300,000 years. Nevertheless, the stable ¹³³Cs isotope gives rise to ¹³⁵Cs by neutronic irradiation. So, if separation of the different isotopes is not possible, ¹³³Cs could increase the ¹³⁵Cs content. Immobilisation of caesium into crystalline networks must be studied since the separation of its different isotopes is very hard to achieve. Most of the efforts to find a material able to stock and isolate actinides from the biosphere for a long period are based on the discovery of crystallised natural materials with an apatite structure.^1–3 Previous work has shown the remarkable properties of apatite for the confinement of actinides.⁴ However, the presence of actinides in the apatite provokes phenomena like exfoliation of the ceramic, lixiviation, expansion of the structure and amorphisation,⁵ increasing the risk of liberation of the actinides into the geosphere. The confinement of actinides could be performed through the use of composite materials⁶ but inhomogeneities in the composite composition still remains a problem.⁶

Therefore, new solutions must be found. The volatile nature of caesium makes it difficult to synthesize materials with caesium through high temperature synthesis. For this reason, we decided to follow a sol–gel route. Phosphate compounds cannot be considered because of their high lixiviation rate. Aluminosilicate compounds are interesting due to their remarkable intrinsic properties but this composition does not match the composition of apatite. Silicophosphate compounds can prove interesting because many apatites contain Si and P elements.

In this work, we present the synthesis by a sol–gel process and the characterisation of a series of silicophosphate solids containing caesium. The structural changes induced by thermal treatment will be studied and, in particular, the liberation of the Cs-content with heat treatment into these solids will be shown.

2 Experimental section

2.1 Sample preparation

Mixed oxides “Cs–SiO” and “Cs–SiPO” compounds were prepared by a sol–gel process. Tetraethoxysilane (TEOS, Alfa 99%), caesium carbonate (Cs₂CO₃, Alfa 99%) and di-phosphorus pentaoxide (P₂O₅, Panreac purissimum) were used as precursors of Si, Cs and P, respectively. First, the adequate amount of P₂O₅ was dissolved in 4 ml of methanol and the solution was refluxed for 20 h. After cooling to room temperature, 5 ml of TEOS was added dropwise under vigorous stirring for 1 h. Then, the appropriate amount of Cs₂CO₃, previously dissolved in 4 ml of methanol, was added dropwise to the solution. The resultant sol was allowed to gel at room temperature in a vessel covered by paraffin to avoid moisture from the atmosphere. After about one month, a transparent gel is obtained. The theoretical molar composition of the solids prepared are xP₂O₅·(95−x)SiO₂·5Cs₂O (x = 0, 10, 25 and 40). The samples are denoted in the text as Ax. The gels obtained were calcined in air at 150, 300, 500, 700 and 900 °C for 10, 8, 4, 2 and 2 h, respectively.

2.2 Characterisation techniques

X-Ray diffraction. X-Ray powder diffraction (XRD) patterns of the solids were recorded using a Siemens Daco MP diffractometer working with Cu Kα radiation in continuous scan mode from 15° to 55° of 2θ with a 0.05° sampling interval and 3 °C min⁻¹ scan rate.

Infrared spectra. Diffuse reflectance infrared (DRIFTS) spectra were obtained in a Nicolet 510 spectrometer with KBr optics and a deuterated triglycine sulfate (DTGS) detector. Samples were previously diluted in KBr (1/20 wt%) and spectra were obtained by co-adding 200 scans at 4 cm⁻¹ resolution at room temperature. The spectrum of powdered KBr was used as background.

X-Ray fluorescence. Analyses by X-ray fluorescence spectrometry (XRF) were performed in a Siemens SRS 3000 sequential spectrophotometer with a rhodium tube as the source of radiation. XRF measurements were performed onto pressed pellets (sample included in 10 wt% of wax). The XRF apparatus was previously calibrated from references of known compositions of Si, P and Cs, these references being mixtures of SiO₂, CaP₂O₇ and Cs₂CO₃.

Thermogravimetric analyses. A Seiko Exstar 6000 thermal analysis instrument was used for sample weight evolution and differential thermal analysis (DTA). The compounds studied were heated from room temperature (RT) to 900 °C under oxygen flow and with a 5 °C min⁻¹ rate.

X-Ray photoelectron spectroscopy. XPS spectra were obtained in an ultra high vacuum chamber (UHV) to which an energy electron analyser (VG 100 AX) was fitted (vacuum better than 3.10⁻⁹ torr, Mg Kα radiation, 15 kV, 20 mA). Before analysis, the samples were heated at 300 °C for 48 h under vacuum (10⁻⁷ torr) in the pre-treatment chamber and then introduced in the analysis chamber. A 300 °C evacuation temperature was needed because of the high water content of the gels. After subtraction of a Shirley-type non-linear baseline, the spectra were decomposed according to a commercial fitting program (VGX 900) with a Gaussian/Lorentzian ratio of 85/15. Binding energies are referenced to the spurious C(1s) signal at 284.6 eV. The atomic ratios were calculated from relative intensities corrected by the elemental sensitivity factor of each atom.⁷

3 Results

3.1 Chemical analysis

The chemical compositions of the solids calcined at different temperatures are presented in Table 1. A slight loss of caesium is observed at high temperatures (>900 °C) in the case of A0 and A10 samples, while for the solids with higher P-content (A25 and A40), the caesium content remains constant with the calcination temperature.

Table 1 Chemical compositions of the studied solids from XRF measurements

		Si (wt%)	P (wt%)	Cs (wt%)
A0	Theoretical	66.75	0	33.25
	RT–700 °C	62.1	–	37.9
	900 °C	67.5	–	32.5
	1100 °C	73.2	–	26.8
A10	Theoretical	55.04	14.29	30.66
	RT–700 °C	50.0	22.8	27.1
	900 °C	56.0	22.2	21.8
A25	Theoretical	40.59	31.97	27.44
	RT–900 °C	38.1	28.5	33.2
A40	Theoretical	28.86	46.30	24.83
	300 °C	27.2	48.7	24.1
	700 °C	27.1	50.3	22.6
	900 °C	28.1	47.4	24.4
	Average	27.5	48.8	23.7

---

3.2 Evolution of the crystallographic properties

The structure of the different compounds with the calcination temperature has been studied by powder X-ray diffraction (XRD). No diffraction peaks are observed in the XRD diagrams of A0 (solid without P) up to 700 °C, indicating the amorphous nature of the sample (Fig. 1). At 900 °C, some diffraction peaks appeared, assigned to SiO₂ (cristobalite and tridymite phases). No crystalline phase containing caesium was detected. The evolution of the XRD patterns of A10, A25 and A40 as a function of the temperature reveal that all the silicophosphate-based samples were amorphous at room temperature (RT) (Fig. 1). Heating the sample allows crystallisation. Indeed, at 300 °C, some diffraction peaks assigned to CsH₅(PO₄)₂ appear in the case of A10 while crystallisation begins at lower temperature (150 °C) when the P-content increases. By increasing the temperature, a crystalline Si₅P₆O₂₅ phase starts to appear at 500 °C in the case of A10 and at 300 °C in the case of A25 and A40. Si₅P₆O₂₅ consists of isolated SiO₆ and Si₂O₇ units linked by PO₄ groups⁸ and has been already reported to be formed in SiO₂–P₂O₅ gels after heating at 200 °C.⁹ Thus, the detection of this phase is indicative of the formation of hexacoordinated silicon. At higher temperatures, A10 and A25 crystallise in SiO₂ at 900 °C while only crystalline Si₅P₆O₂₅ can be observed in the case of A40 at 700 °C and 900 °C.

Fig. 1 Evolution of the XRD pattern of the studied solids as a function of the calcination temperature (□: Si₅P₆O₂₅; ○: CsH₅(PO₄)₂; ✦: SiO₂ cristobalite; ●: SiO₂ tridymite).

3.3 Diffuse reflectance infrared Fourier transform spectroscopy

DRIFTS spectra of non-heated A0, A10, A25 and A40 solids are presented in Fig. 2. For comparison, the spectrum of amorphous SiO₂ obtained via a similar sol–gel method is also presented. SiO₂ shows IR bands at 458, 562, 802, 954, 1083, 1160 (shoulder) and 1635 cm⁻¹. According to literature,^10–12 these bands are ascribed to Si–O–Si symmetric bending, Si–O–Si symmetric stretching, Si–OH stretching, TO mode of Si–O–Si asymmetric stretching, LO mode of Si–O–Si asymmetric stretching and H–O–H bending of adsorbed water. The band at 562 cm⁻¹ has already been reported¹³ in the silicate system, although no assignment has been proposed for it.

Fig. 2 DRIFTS spectrum in the 2000–400 cm⁻¹ region of the non-heated solids. For comparison, the DRIFTS spectrum of non-heated sol–gel obtained SiO₂ is also shown.

Besides the bands observed in the SiO₂ spectrum, additional ones are present at 1690 and 1583 cm⁻¹ in the room temperature (RT) spectrum of the A0 solid. According to the literature,¹³ they can correspond to symmetric and asymmetric stretching of COO⁻, which can be due to residues of the precursor of caesium.

In the RT spectra of A10, A25 and A40 (solids with P content), the absence of well-defined bands in the region corresponding to the C=O bond (1500–1700 cm⁻¹) is noticed. The C=O bond originating from carbonate seems to have reacted by opening itself by hydrolysis. The resulting C–O bonds would fall at about 1050–1330 cm⁻¹ but cannot be discerned with clarity in our case. For the series of Cs–SiPO samples, a shift of the position of the principal band to lower wavenumbers is observed as P content increases. The shift of the band at 960 cm⁻¹ to higher wavenumbers can be explained by the presence of P–OH stretching and P–O–P stretching (940–1100 and 900–980 cm⁻¹, respectively) mode vibrations.

In all these spectra, new vibration bands at 1463, 1403, 850 and 793 cm⁻¹ also appear, the intensity of which increases with the P content. They can be associated with the raw material, TEOS, which has not entirely reacted yet and the condensation process. Indeed, the band at 1463 cm⁻¹ is assigned to –CH₃δ¹³ asymmetric and/or –CH₂δ¹³ groups that can be associated with –SiOC_xH_y units (SiCH₃δ symmetric gives a band at 1410 cm⁻¹), and the bands at 850 and 793 cm⁻¹ can be attributed to Si–(O–(C_xH_y))_z groups. In this sense, Si–O–C groups have been reported to present two bands at about 1000–1100 cm⁻¹ and 800–850 cm⁻¹,¹³ and >Si(CH₃)₂ and –Si(CH₃)₃ species give two IR bands in the range 840–855 and 765–800 cm⁻¹.¹³ It is worth noting that TEOS residues are only detected in the solids with phosphorus, suggesting that when phosphorus is present, the condensation reactions are not complete or that the kinetic of condensation decreases.

A new IR band at 1187 cm⁻¹ is also observed. The higher the phosphorus content of the solid, the higher the intensity of this band, while its position remains constant. This band can be assigned to PO₂⁻ group which has been reported to appear at 1150–1300 cm⁻¹.¹³ In this sense, (RO)₂PO₂⁻ presents two bands at 1120–1285 and 1050–1120 cm⁻¹.¹³ These species must proceed from the unreacted phosphorus precursors. Such assignment could explain the shift observed in the IR principal band with the P content. The principal band can hardly be shifted due to the caesium, since the Cs-content is the same whatever the solid. Moreover, the effect of the cations on the spectra of the phosphates is not very marked, except in the case of silver salts,¹⁴ where a shift is observed in the stretching frequencies, due probably to the tendency to form covalent bonds Ag–O. So, we don't have to expect an important effect due to the caesium.

The DRIFTS spectra of the calcined solids are shown in Fig. 3. In the A0 DRIFTS spectra, only slight changes with temperature are noticed, overall in the 1600 cm⁻¹ region where the bands related to organic groups decrease upon heating until disappearance at 300 °C. The band at 960 cm⁻¹ due to Si–OH groups also decreases upon heating. At 500 °C, a slight shift of the principal band (about 10 cm⁻¹) is observed. This feature is also observed in the case of the pure sol–gel obtained SiO₂. At 700 °C, the bands become larger and at 900 °C new bands assigned to the SiO₂ cristobalite phase¹⁵ emerge.

Fig. 3 DRIFTS spectrum in the 2000–400 cm⁻¹ region of the studied solids: (a) at 25 °C; (b) 150 °C; (c) 300 °C; (d) 500 °C; (e) 700 °C; and (f) 900 °C.

When heating the phosphorus-containing samples, the bands ascribed to TEOS residues decrease and disappear at 300 °C (Fig. 3). Simultaneously, changes in the structure of bands are observed, depending on the phosphorus content. An increase in the relative intensity of bands at around 1000 cm⁻¹ and 890 cm⁻¹ and a decrease in the one observed at 1100 cm⁻¹ is detected with the P content in the spectra of the samples calcined at 150 and 300 °C. The first two bands must be related to the presence of phosphorus. In this sense, a band at around 890 cm⁻¹ has been assigned to P–O–P bonds¹⁴ and to P(OH)₂ groups.¹⁶ The band at about 1000 cm⁻¹ could be associated with terminal PO₃²⁻ groups since such species present IR bands in the 970–1040 cm⁻¹ region.^13,17,18 On the other hand, the band at 1100 cm⁻¹ is due to the overlapping of Si–O–P,^10,19 Si–O²⁰ and P–O bonds.^20,21

At 500 °C, new bands appear at 640, 713 and 1030 cm⁻¹, the intensity of which increases with the phosphorus content. All of them are indicative of the presence of Si^VI. The band at 640 cm⁻¹ is related to six-coordinated Si,^22–25 that at 713 cm⁻¹ is assigned to stretching vibrations involving SiO₆ octahedral units²⁵ and the band at 1030 cm⁻¹ can be assigned to P–O–Si.^19,26 In the case of A40 additional bands are observed that must be related to the crystallisation of the Si₅P₆O₂₅ phase. The bands at 403, 415, 477, 499, 555 and 794 cm⁻¹ are related to a combination of the bending modes of Si–O–P, P–O–P and O–P–O.^22,25 The band at 1111 cm⁻¹ is attributed to P–O stretching vibration in the P–O–P and P–O–Si bonds,^22,25,27 and to P–O⁻ groups,^18,28 while the band at 1173 cm⁻¹ can correspond to the pyrophosphate P₂O₇⁴⁻ group²⁹ and/or the PO₄³⁻ group.²⁹ The vibration that occurs at 1195 cm⁻¹ is attributed to the asymmetric stretching mode of P–O⁻ groups,^18,30 that at 1238 cm⁻¹ is assigned to the stretching mode of P=O in the tetrahedric phosphates^18,28 and the one at 1311 cm⁻¹ corresponds to P=O.

The phosphorus content of the solid influences the formation and stability of the hexacoordinated silicon species. In the sample without P (A0), no evidences of such species exists whatever the calcination temperature. In the case of A10 and A25, they are formed at 500 °C and the relative intensities of their characteristic IR bands grow until 700 °C, when they are removed. For the sample with the highest phosphorus content (A40), the intensity continuously increases with temperature. The bands relative to crystalline Si₅P₆O₂₅ follow the same trend as Si^VI.

At 900 °C a band at 620 cm⁻¹ is observed in the A10 spectrum, characteristic of crystalline SiO₂ cristobalite phase.¹⁵ As a general observation, the proportion of P=O bonds in this material is high, as deduced from the very intense band around 1310 cm⁻¹.¹³

The progressive disappearance of the bands due to Si–OH as the temperature increases can be observed. The raw materials are not entirely consumed at room temperature but no leftovers are noticed up to 300 °C. The shifts we pointed out have been related to the presence of phosphorus. The existence of six-coordinated silicon has been demonstrated but seems to exist only when the Si₅P₆O₂₅ phase crystallises.

3.4 Thermogravimetric analyses

Fig. 4 presents the weight loss and the DTA curves of the studied samples. A total weight loss of 6.3%, 27%, 36% and 40% is measured for A0, A10, A25 and A40, respectively, and occurs below 450 °C. In all the cases, two DTG peaks are observed, except for A40 for which an additional peak is present (not shown). The first peak, an endothermic one, ranges from 57 to 82 °C, depending on the sample, and is associated with the desorption of weakly bound water.³¹ This peak can also be assigned, in a minor way, to a loss of alcohol.³² A similar peak is present at about 75 °C in the case of pure silica obtained by a similar sol–gel process (not shown).

Fig. 4 Weight loss and DTA curves for the studied solids.

The second DTG peak observed in the range 199–210 °C corresponds to the main weight loss process. The higher the P content, the greater the area of this peak. A similar observation has been reported²⁶ for a silicophosphate system and the detected weight loss was associated with the elimination of organic groups. This peak can be due to the desorption of molecular water strongly adsorbed (hydration water^33,34) to the loss of water proceeding from polycondensation reactions between hydroxyl groups and/or to a release of organic material³² (loss of the solvent trapped in the xerogel, or pyrolysis or combustion reactions). The overlapping of two or more of these processes cannot be discarded. Indeed, the DTA curve associated with this DTG peak present several events, suggesting the coexistence of different phenomena. In this sense, in the Al₂O₃–SiO₂ system, a large DTG endothermic peak was observed between 200 and 400 °C, being assigned to the simultaneous removing of water, ethanol and nitrate.³⁵ Moreover, in the case of phosphosilicate compounds obtained by sol–gel process,²⁶ a weight loss was observed between 200 and 300 °C, and was assigned to the elimination of organic groups; in the Si–Ti–O system,³² a DTG peak has been reported at 309 °C and was assigned to the oxidation to CO₂ of the alkoxy residues present in the solids. In our system, the existence of TEOS residues has been evidenced by DRIFTS at temperatures lower than 300 °C (Fig. 3).

DRIFTS measurements have shown that the more important the P content, the more important the quantity of precursor remaining at RT, suggesting that the process of condensation is not complete. The second DTG peak, the intensity of which increases with P content, could be related to the completion of the condensation process with release of H₂O, alcohol, etc. This feature would explain why, in the case of A0, this second peak can hardly be seen. None of these peaks can be associated with a loss of unreacted precursors (by evaporation, for example) since no loss of phosphorus, silicon or caesium is detected by XRF up to 700 °C.

In the case of A40 (the solid with the highest P content), the presence of a third DTG peak at 386 °C, associated with several DTA events, can be related to the crystallisation of the Si₅P₆O₂₅ phase. In this sense, the XRD pattern of this solid shows well defined diffraction peaks. For the other compounds, no DTG or DTA peaks at this temperature are observable, and large and poorly defined peaks are visible in the XRD diagrams, suggesting poor crystallinity of the phases.³³

Above 400 °C the weight loss is relatively weak, suggesting that there is a non-appreciable loss of Cs or P. The XRF data only show weak loss of phosphorus and caesium in A10 and A0 at 900 °C. Our materials retain phosphorus more effectively than other silicophosphate systems, in which a substantial weight loss associated with phosphorus between 300 °C and 1000 °C has been reported.³⁶

3.5 X-Ray photoelectron spectroscopy

The XPS study was performed on A0, A10 and A40 calcined at 300 °C, 500 °C, 700 °C and 900 °C. The XPS spectra of the 150 °C-heated solids were not recorded since the samples were heated at 300 °C in the XPS pre-treatment chamber. In the case of A0, the measurement of the 300 °C-heated sample was not possible due to a problem of compaction of sample. XPS of 1100 °C-calcinated solid was also recorded since an important loss of Cs is detected for A0 at 900 °C (Table 1). In this sense, SiO₂ (cristobalite and tridymite phases) was detected in its corresponding XRD pattern. For A40, the 700 °C-heated XPS spectrum was not recorded since no difference was found by XRD with respect to the 900 °C-heated solid (the same crystalline phase was detected) and no variation in the XRF elemental analyses was observed between both solids.

Fig. 5 shows the O(1s) and Si(2s) regions of the A0 solid heated at the indicated temperatures. The O(1s) peak can be decomposed into two peaks at about 531.0 and 533.0 eV. At 500 °C, only one component for the Si(2p) peak can be observed at 103.9 eV, but an additional one emerges at lower binding energy (BE) on increasing the temperature. In the Cs(3d) region, only one component is observed (not shown). We will denote further in the text by subscripts A and B the components located at lower and higher BE, respectively.

Fig. 5 Evolution of the Si(2p) and O(1s) XPS spectra of the A0 solid as a function of the calcination temperature.

For the O(1s) peak, we assign the component at lower BE (denoted as O_A) to non-bridging oxygen (NBO) and the component at higher BE (denoted as O_B) to bridging oxygen (BO), in good agreement with the observations of Peters et al.³⁷ who found that BE(NBO) < BE(BO) in their silicophosphate system with an energy difference (ΔE) value of about 2 eV, similar to the one observed in our case. Moreover, the value of 533 eV is consistent with a SiO₂ network.^38,39 For the Si(2p) peak, the only component observed at 500 °C agrees with the position reported for a SiO₂ network.^39–42 The second component that appears upon heating at lower BE is located at around 108 eV (with a ΔE(Si_B−Si_A) value of almost 2 eV). Such low position values have already been reported for the Si(2p) peak in the SiO₂ system containing elements like alkali metals⁴³ or others (Ca, Sr, Pb or Mn).⁴⁴ The Si_B/O_B ratio remains practically constant (0.67 ± 0.02) over all the range of the temperatures studied. In agreement with the DRIFTS results, no evidence of TEOS residues are noticed since Si(2p) of Si(OEt)₄ would be located at about 107.56 eV.³⁹ On the basis of these considerations, we assign the Si_B component to non-terminal and the Si_A one to terminal Si. The Si atoms near Cs should be included in Si_A, since a Si–O–Cs bond implies a break in the Si–O–Si and/or Si–O–P networks. Concerning the Cs(3d) region, only one component can be distinguished which does not shift significantly with temperature. From the quantification analysis (Table 2) we can notice that the caesium content at the surface remains constant up to 700 °C, increases at 900 °C and then reduces drastically at 1100 °C.

Table 2 XPS surface composition of the considered solids. For comparison the bulk compositions (measured at room temperature by XRF) are indicated

		Si (at.%)	P (at.%)	Cs (at.%)	O (at.%)
A0	XRF	31.3	0	4.0	64.6
	500	34.4	0	5.1	60.5
	700	37.9	0	5.3	56.8
	900	32.2	0	8.7	59.0
	1100	38.9	0	1.1	60.0
A10	XRF	21.6	9.0	2.5	66.9
	300	10.5	20.7	2.6	66.2
	500	22.6	10.5	2.8	64.1
	700	18.1	13.6	7.5	60.9
	900	26.5	10.1	2.3	61.1
A40	XRF	11.2	18.1	2.1	68.6
	300	16.3	15.9	0.8	67.0
	500	19.0	18.7	1.1	61.2
	900	25.3	14.7	3.9	56.1

---

The presence of P in the material does not induce apparent changes in the number of components in the O, Si and Cs regions, but affects their relative intensities.

In the Si(2p) region, the component named (Si_A) appears at a higher temperature than in the case of the solid without P (A0) and predominates at 900 °C for the solid with the highest P content (A40). For A10 and A40, some slight shifts in the position of the Si_B component are observed. This shift coincides with the growth of the Si₅P₆O₂₅ phase, as shown by XRD. From here we assign it to the presence of Si–O–P bonds and Si^VI (Si^VI appears to be exclusively included into Si–O–P³⁷). Si_B cannot be related to bonds containing Cs since, in the case of A0, terminal Si atoms were ascribed to Si_A, and in the case of A40, this Si_A component is also observed but only appears at 900 °C. At this temperature, the shift towards lower BE observed in A10 is consistent with its XRD pattern, showing the formation of SiO₂. In particular, Si_B goes back to its original value which corresponds to SiO₂, in good agreement with the positions observed in the case of A0.

In the O(1s) region, the relative intensity of O_A increases with temperature, becoming predominant at 900 °C in the case of the solid with the highest P content. For A10, a relevant feature is the predominance of this O_A component at 300 °C, assigned to NBO. This observation suggests the existence of numerous NBO due to the presence of phosphorus at the surface of the particles since non-terminal Si atoms are detected.

The positions observed for the O(1s) components are consistent with the presence of P=O (or P–O⁻) and P–O–P-like bonds.^45–48 For A40, a shift with the temperature towards higher BE is observed for O_B, which must be assigned to the presence of Si–O–P bonds, in agreement with the XRD observations and the XPS results in the Si(2s) region. Nevertheless, this shift is not marked for A10, suggesting that there are a low number of Si–O–P bonds at the surface of A10, or that the presence of numerous bonds related to P at low temperature hides the shift of the O_B component due to the formation of Si–O–P bonds at the expense of P–O–P ones.

Concerning the P(2p) region, only one component is discerned but there must be a mixture of different contributions (such as formation of Si–O–P bonds, variation of the content of O–P–O, P=O or P–O–Cs bonds, etc.). In this sense, P=O and O–P–O can coexist since these two contributions can hardly be discerned.⁴⁹ Both species have been detected in the bulk by DRIFTS and their existence has been also revealed by XRD patterns. Due to the presence of Si–O–P bonds, shifts are also expected in the P(2p) region. Even if assignation is difficult, we can notice that when the Si_B component shifts, the P(2p) component also shifts.

Comparing the different ratios obtained by XRF and XPS (Table 2), a complex pattern appears for the low temperature samples depending on the phosphorus content. For the A0 sample the XPS and XRF composition are similar, whereas for the A10 sample there is a clear enrichment in phosphorus and for the A40 the surface enrichment is in silicon. This pattern should be associated with the formation of crystalline phases, CsH₅(PO₄)₂, in the A10 sample at 300 °C as detected by XRD.

On increasing the temperature an increase in the surface content of caesium is observed in every case, which should be associated to the migration of caesium to the surface prior to its volatilisation. The temperature and the distance at which this effect occurs depends on the phosphorus content. The surface evolution of Cs follows the same pattern observed for the total caesium content by XRF, Table 1.

For A10 at 500 °C, migrations of Si and Cs are observed, that of Si being more important. XRD results show that a Si₅P₆O₂₅ phase is forming, probably at the expense of a CsH₅(PO₄)₂ one, and this phenomenon can occur also at the surface. For A40, no significant migration of elements is noticed at 300 °C and 500 °C, and the surface composition is almost constant (20Si∶20P∶1Cs). For these two samples, the Si_A component is not visible, indicating that terminal Si atoms are not present in an significant quantity at this stage. From here, Cs is probably not located near Si atoms, meaning that Cs is isolated or incorporated into a silicophosphate or phosphate network. At 700 °C, a migration of P and Cs to the surface is observed for A10. The migration of Cs in A10 occurs at a lower temperature than in A0 (Table 2). Nevertheless, the component Si_A is not observed yet, indicating the absence of terminal Si. At 900 °C, in agreement with the XRF results, a loss of Cs and P is detected in the case of A10, resulting in an enrichment of the surface in Si. The Cs loss is higher than the P one, in good agreement with the XPS ratios, which increase in the order Si/Cs > Si/P > P/Cs. At this temperature, the A10 and A40 Si_A component appears clearly, indicating the existence of terminal Si in appreciable quantity. In the case of A40, the relative proportion of the different components changes and Si_B and O_B are predominant with respect to Si_A and O_A. This observation confirms that Si_A and O_A are related to each other. Since no loss of P, Si or Cs is noticed when using XRF, we can deduce that the A40 surface at 900 °C is richer in Si and Cs than that of 500 °C.

At 900 °C, XRF shows a partial loss of caesium in A0 and A10. XPS data detect a caesium loss in A10, but an enrichment of the surface in Cs for A0. This difference in behaviour must be related to the presence of phosphorus in A10, suggesting that phosphorus affects the kinetics of the caesium migration and the volatilisation process. In A40, where the amount of phosphorus in the network is high, the volatilisation of caesium is prevented at high temperature, as is shown above.

In the case of A40, the O_A/(O_A + O_B) ratio increases at 500 °C, indicating the formation of NBO at the expense of BO. Such increase cannot be ascribed to the increase in the proportion of terminal Si since the Si_A component does not exist. It must be assigned to the formation of P–O–Cs and/or P=O bonds at the surface of the particles (in agreement with DRIFTS results that showed the presence of P=O groups). This result is also observed in the case of A10. Nevertheless, the increase of the O_A/(O_A + O_B) ratio occurs from 500 °C (at 300 °C, the presence of numerous NBO is due to the presence of CsH₅(PO₄)₂ phase). At 900 °C, the O_A/(O_A + O_B) ratio tends to stabilize in the case of A10 while in the case of A40 it goes on growing. These observations are related to the loss of caesium: for A10, volatilisation of the caesium occurs at this temperature while for A40 caesium accumulates on the surface.

4 Discussion

According to XRF measurements, caesium and phosphorus are detected in all the compounds up to 900 °C and no loss of caesium or phosphorus is observed up to 700 °C. When the calcination temperature increases up to 900 °C, a partial loss of caesium for the Cs–SiO compound and for the Cs–SiPO solid with low P-content is noticed. For example, in A0, a loss of caesium is detected between 700 °C and 900 °C, corresponding to about one third of the Cs-content. It is worth noting that no loss of any element is observed upon heating at 900 °C for the Cs–SiPO solids with high P-content (A25 and A40).

The XRD patterns show that heating the samples allows their crystallisation. For the sample with no P-content, A0, crystallisation in SiO₂ occurs at 900 °C. When phosphorus is present in the sample, the process of crystallisation is different, its temperature depending on the P-content: some diffraction peaks begin to appear at 150 °C in the case of A40 (the compound with the highest P-content) and crystallisation only starts at 300 °C for A10 and A25. For these three solids, during the crystallisation process and up to 700 °C, the same crystalline phases appear. First, a CsH₅(PO₄)₄ phase crystallises, and then Si₅P₆O₂₅ appears progressively. At 900 °C, A10 and A25 give crystalline SiO₂ as in the case of A0, while A40 crystallises in Si₅P₆O₂₅. No crystalline phase containing caesium is detected at 900 °C and, for A10 and A25, any crystalline phase containing phosphorus is detected at this temperature. Nevertheless, XRF analyses reveal the presence of caesium and phosphorus in all the solids up to 900 °C. Thus, the existence of one or more phases containing caesium and phosphorus, which are amorphous or below the crystalline detection limit of XRD, must be admitted. Also, the crystallisation of the samples must be related to the presence of caesium, even if the temperature at which the crystalline phases appear and the nature of the final phases are P-dependent. Indeed, for a SiO₂ solid obtained by a similar sol–gel process, no crystallisation is observed at 900 °C (not shown). Moreover, in the SiO₂–P₂O₅ system with a similar Si/P ratio, crystallisation occurs at higher temperature. The same observation can be made in the case of Li–SiPO solids.

The information regarding the structural properties of the silicophosphate gels that can be extracted from the infrared technique is limited due the overlapping of the Si–O and P–O vibration modes.²⁰ DRIFTS results have shown that there is no precursor residue detected up to 300 °C. This result is in agreement with the thermogravimetric analyses that showed that all processes implying organic compounds occur below 450 °C. The shifts of the bands that can pointed out have been related to the presence of phosphorus, the bands related to phosphorus having a higher intensity as a function of the P-content of the sample. Appearance of new bands is also observed with the heat-treatment, in good agreement with the crystallisation process shown by XRD. The presence of six-coordinated silicon has been demonstrated by DRIFTS but such species seem to be related only with the crystallisation of an Si₅P₆O₂₅ phase. Firstly, for polymorphs only detected in SiO₂ at high pressure (like stishiovite⁵⁰), the existence of SiO₆ has been already reported in the SiO₂–P₂O₅ binary system, its content depending essentially on the P₂O₅ content,⁵¹ even if the addition of alkali metals to the system also favours the SiO₆ proportion.^51–55

XPS results showed that the surface of the particles of A0 at 500 °C is mainly composed of a SiO₂ network and Cs is not included yet in the SiO₂ network. The existence of the O_A component is due entirely to the presence of Cs at the surface. When the temperature increases up to 700 °C, terminal Si are forming, probably because Cs is breaking the silica network. At this stage, no surface enrichment in Cs is noticed. At 900 °C, the Cs content increases at the surface, but at this temperature XRF results revealed that partial loss of Cs occurs. These contradictory observations can be interpreted by a migration of Cs from the bulk to the surface, and at the same time a partial loss of Cs occurs. As XPS results indicate that the surface of the solid at this temperature is richer in Cs than at 700 °C, we can suppose that in such conditions, the migration of Cs is faster than its volatilisation. A 1100 °C, the Cs content has reduced drastically, confirming its volatilisation. There is less Cs in the bulk and, as a consequence, insufficient Cs can migrate to the surface to compensate for the volatilisation, or the volatilization is much more effective than the migration. Moreover, the Si_A component practically no longer exists. We can see that the Si_A and Cs components follow the same trend, confirming their relationship even if they do not have a direct one.

With the Cs–SiPO samples, for the low P-content material, A10, the surface is richer in P than expected at 300 °C. This observation can be related with XRD results that detected the presence of CsH₅(PO₄)₂ phase, which may exist at the surface of the particles. A more P-rich surface was expected for A40 (since the P-content brought by the synthesis is higher for A40 than for A10) but the XRF measurements reveals that P is more present at the surface of the A10 particles than that of the A40 ones. Moreover, the relevant feature is the poor content of caesium at the surface at 300 °C: the caesium seems to be inserted deeper in the bulk for A40 than for A10.

For A10, a high migration of caesium to the surface occurs at lower temperature than in the case of A0 but without the XRF-detected loss of caesium. However, in A10, a loss of caesium in the bulk and in the surface is envisaged, confirming its volatilization at 900 °C. In this sample, the too low P-content cannot have an efficient effect in retaining caesium. Like A0, the matrix seems to be able to retain the caesium up to a certain temperature but for at too high a temperature caesium volatilization becomes favourable and leaves the solid. This is not the case for the most charged in phosphorus solid, A40, for which no P and Cs losses are detected up to 900 °C. At this temperature, a greater proportion of terminal Si is detected by XPS, contrary to A10 and A0 at high temperatures. One explanation could be the migration of Cs to the surface, breaking the network and leading to an increase of the terminal Si content (in this case, the intensities of the Si and O components assigned to terminal Si and NBO must increase respectively). Nevertheless, this phenomenon is hidden in the case of A0 and A10 since, at the same time, volatilization of the caesium occurs, leading to a decrease in terminal Si and NBO. Another explanation could be that in A40, caesium integrates into the network and, as a consequence, is retained at the surface of the particles. In this case, volatilization is avoided. The presence of the caesium causes the break in the network, which explains the high content of terminal Si and NBO for 900 °C heated A40. The presence of the Si_A component seems not to be related to a leaving of caesium. Indeed, in the case of A0, the Si_A component is observed from 700 °C but no loss of Cs is observed by XRF.

On the surface of the particles of the Cs–SiPO solids, the existence of phases different from those detected by XRD must be suggested. Indeed, for A40 at 900 °C, the XRD patterns have unambiguously identified the presence of pure Si₅P₆O₂₅ crystalline phase (Si/P ratio of 0.8) but the calculated XPS surface composition at 900 °C does not correspond to a Si₅P₆O₂₅ formulation, suggesting the presence of a Si-rich phase. This observation is in agreement with the XRF results: in A40 compound where only crystalline Si₅P₆O₂₅ phase is observed at high temperature, we observe a P/Si relation of 1.77 by XRF. For A10 heated at 900 °C, a P and Cs rich phase must inferred since these two elements are detected by EDS and XRF analyses. These phases must be amorphous or only exist at the surface of the particle in too low a quantity to be detected by XRD.

5 Conclusions

We have synthesized silicate-based and silicophosphate-based compounds containing caesium by a sol–gel process. The solids crystallise upon heating and the behaviour of the crystallisation process depends on the P content. A partial loss of caesium is observed from 900 °C in the case of the sample with no or too low P content, but for the Cs–SiPO solids with higher P content, no Cs nor P loss is observed. It seems that there exists a threshold of the P content, above which the quantity of P is efficient to retain caesium at least up to 900 °C. At this stage of the study it is not possible to say if it is the P/Si ratio or the P/Cs ratio which determines the P threshold. The sol–gel synthesis process we propose is a simple and mild-conditions one and seems efficient in retaining volatile caesium, opening a new and promising way for the safe waste confinement of elements of high volatility.

Acknowledgement

The authors are grateful to J. M. Blanes for carrying out the XRF measurements.

References

1. R. Boudu, H. Bouzigues, N. and J. P. Pfiffelmann, C. R. Seances Acad. Sci., 1972, 275.
2. J. Carpéna, J. R. Kienast, K. Ouzegane and C. Jehanno, Geol. Soc. Am. Bull., 1998, 100, 1237.
3. M. R. Hugson and J. E. Sen Gupta, Am. Mineral., 1991, 49, 937.
4. See, for example, J. Carpéna and J. L. Lacout, French patent 93 08676 1993.
5. S. Soulet, PhD thesis, Université de Paris-Orsay, Paris, France, 2000.
6. S. Titeux, PhD thesis Université de Rennes I, Rennes, France, 2000.
7. C. D. Wagner, W. N. Riggs, L. E. Davies, J. F. Moulder and G. E. Muilenberg in Handbook of X-ray Photoelectron SpectroscopyPerkin Elmer Co., Physical Electronic Division, Eden Prairie, 1978.
8. T. L. Weeding, B. H. W. S. de Jong, W. S. Veeman and B. G. Aitken, Nature (London), 1982, 318, 352.
9. S. P. Szu, L. C. Klein and M. Greenblatt, J. Non-Cryst. Solids, 1992, 143, 21.
10. M. Stan, A. Vasdilescu, S. Moscu and M. Zaharescu, Rev. Roum. Chim., 1998, 43, 425.
11. N. Shibata, M. Horigudhi and T. Edahiro, J. Non-Cryst. Solid, 1981, 45, 115.
12. A. Bertoluzza, C. Fagnano, M. A. Morelli, V. Gottardi and M. Gugliemi, J. Non-Cryst. Solids, 1982, 48, 117.
13. E. Pretsch, T. Clerc, J. Seibl and W. Simon, in Tablas Para la Elucidación Estructural de Compuestos Orgánicos por Métodos Espectroscópicos, Vertix, Alhambra , 2nd edn., 1985.
14. D. E. C. Corbridge and E. J. Lowe, J. Chem. Soc., 1954, 493.
15. H. Moenke, Silikattechnik, 1962, 13, 287.
16. A. C. Chapman and L. E. Thirlwell, Spectrochim. Acta, 1964, 20, 937.
17. D. E. C. Corbridge, J. Appl. Chem., 1956, 6, 456.
18. H. S. Liu, T. S. Chin and S. W. Yung, Mat. Chem. Phys., 1997, 50, 1.
19. Y. S. Kim and R. E. Tressler, J. Mat. Science, 1994, 29, 2531.
20. E. A. Hayri and M. Greenblatt, J. Non-Cryst. Solids, 1987, 94, 387.
21. P. P. Tsai and M. Greenblatt, J. Non-Cryst. Solids, 1988, 103, 101.
22. M. W. G. Lockyer, D. Holland and R. Dupree, Phys. Chem. Glasses, 1995, 36, 22.
23. H. Yamashita, K. Nagata, H. Yoshino, K. Ono and T. Maekawa, J. Non-Cryst. Solids, 1999, 248, 115.
24. R. Dupree, D. Holland, M. G. Mortuza, J. A. Collins and M. W. G. Lockyer, J. Non-Cryst. Solids, 1989, 112, 111.
25. I. N. Chakraborty and R. A. Condrate, Phys. Chem. Glasses, 1985, 26, 68.
26. Z. Cao, B. I. Lee, W. D. Samuels, L.-Q. Wang and G. J. Exarhos, J. Mater. Res., 1998, 13, 1553.
27. J. Wong, J. Non-Cryst. Solids, 1976, 20, 83.
28. R. F. Bartholomew, J. Non-Cryst. Solids, 1972, 7, 221.
29. K. B. Müller, Glastechn. Ver., 1969, 42, 83.
30. J. O. Byun, B. H. Kim, K. S. Hong, H. J. Jung, S. W. Lee and A. A. Izyneev, J. Non-Cryst. Solids, 1995, 190, 288.
31. F. Orgaz and H. Rawson, J. Non-Cryst. Solids, 1986, 82, 57.
32. B. E. Handy, M. Maciejewski, A. Baiker and A. Wokaun, J. Mater. Chem., 1992, 2, 833.
33. Y. Guo, P. Woznicki and A. Barkatt, J. Mater. Res., 1996, 3, 639.
34. L. Bonhomme-Coury, F. Babonneau and J. Livage, J. Chem. Mater., 1993, 5, 323.
35. D. W. Hoffman, R. Roy and S. Komarneni, J. Am. Ceram. Soc., 1984, 67, 468.
36. D. M. Poojary, R. B. Borade and A. Clearfield, Inorg. Chim. Acta, 1993, 208, 23.
37. J. Peters, H. Jain, O. Kanert, R. Kuchler and V. Blache, J. Non-Cryst. Solids, 1997, 222, 113.
38. T. L. Barr, J. Phys. Chem., 1990, 10, 760.
39. C. D. Wagner, W. M. Riggs, L. E. Davis and J. F. Moulder, in Handbook of X-ray Photoelectron Spectroscopy, Perkin-Elmer Corporation, Physical Electronics Division, 1978.
40. E. C. Onyiriuka, Appl. Spectrosc., 1993, 47, 1.
41. C. D. Wagner, J. Vac. Sci. Technol., 1978, 15, 518.
42. C. D. Wagner, D. E. Passoja, H. F. Hillery, T. G. Kinisky, H. A. Six, W. T. Jansen and J. A. Taylor, J. Vac. Sci. Technol., 1982, 21, 933.
43. S. Matsumoto, T. Nanba and Y. Miura, J. Ceram. Soc. Jap., 1998, 106, 415.
44. Y. Kanedo and Y. Suginohara, Graduate School, Tohoku University, 1977, 285.
45. P. Y. Shih, S. W. Yung and T. S. Chin, J. Non-Cryst. Solids, 1998, 224, 143.
46. G. D. Khattak, M. A. Salim, A. S. Al-Harthi, D. J. Thompson and L. E. Wenger, J. Non-Cryst. Solids, 1997, 212, 180.
47. R. Gresch, W. Muller-Warmuth and H. Dutz, J. Non-Cryst. Solids, 1979, 34, 127.
48. J. F. Ducel, J. J. Videau, D. Gonbeau and G. Pfilster-Guillouzo, Phys. Chem. Glasses, 1995, 36, 247.
49. E. E. Khawaja, S. M. A. Durrani, F. F. Al-Adel, M. A. Salim and M. S. Hussain, J. Mater. Sci., 1995, 30, 225.
50. W. H. Baur and A. A. Khan, Acta Crystallogr. B, 1977, 27, 2133.
51. D. Li, G. M. Bancroft and M. E. Fleet, Am. Miner., 1996, 81, 111.
52. R. Dupree, D. Holland and M. G. Mortuza, Nature, 1987, 328, 416.
53. R. Dupree, D. Holland, M. G. Mortuza, J. A. Collins and M. W. G. Lockyer, J. Non-Cryst. Solids, 1988, 106, 403.
54. D. Li, M. E. Fleet, G. M. Bancroft, M. Kasrai and Y. Pan, J. Non-Cryst. Solids, 1995, 188, 181.
55. S. Muthupari and M. E. Fleet, J. Non-Cryst. Solids, 1998, 238, 259.

---

Footnote

† On leave from: ISMANS, 44, avenue F.A. Bartholdi, 72000 Le Mans, France.

---