Redox structural control of Pd and PdO silica matrices

Abstract

This work shows that superior structural control can be conferred by reducing palladium derived silica structures at room temperature instead of oxidizing at high temperatures. A PdCl₂ precursor was incorporated into silica matrices via a silica sol–gel process. Xerogel samples were reduced very quickly at room temperature, as PdCl₂ decomposed into Pd, thus forming a reduced PdSi xerogel. Under calcination conditions up to 630 °C, small Pd nanoparticles were formed where crystallite sizes remained constant at ∼11 nm. In contrast, in the case of the oxidised PdOSi, the larger crystallite sizes (∼45 nm) of PdCl₂ decomposed into PdO for temperatures above 400 °C in air, also forming smaller crystallite sizes of ∼12 nm. However, the crystallite sizes increased by almost four-fold to ∼41 nm as the oxidation temperature was raised from 500 to 630 °C, thus suggesting PdO nanoparticle agglomeration, made possible by the voids left from the decomposition of the larger PdCl₂. In the case of the reduced PdSi at room temperature, the uncondensed silica structure enveloped the small Pd nanoparticles and hindered metal Pd diffusion at high temperatures, thus delivering a narrower and smaller pore size distribution.

---

1. Introduction

The incorporation of metals or metal oxides into silica during the solution phase of sol–gel synthesis has been pursued as a strategy to homogeneously disperse nanoparticles into silica matrices. This method is very efficient and has found application in the preparation of tailored structures for microporous cobalt oxide silica membranes,^1–5 silica containing cobalt for catalysis,^5–7 hierarchical structures containing Pt nanoparticles,^8–10 metal oxide silica and silicates for adsorption,^11–13 and In₂O₃ nanowires coated in silica as gas sensors.^14–16 Metal oxides have also conferred functionalities otherwise not available in pure silica matrices such as improved structural hydrothermal stability in cobalt silica matrices.^17,18 However, it was recently reported that Co³⁺Si (Co₃O₄) provided superior stability whilst Co²⁺Si (Co²⁺ tetrahedral coordination) was unstable.¹⁹

In view of the versatility of the sol–gel method and metal or metal oxide incorporation in the formation of functional materials, there have been a number of studies in this field for silica matrices containing Ni, Co, Fe, La and Nb. In some instances, it was reported that the Ni/Si molar ratios below or above 0.2 controlled the size of nickel nanoparticles to ∼5 nm or ∼40 nm, respectively, whilst improving hydrogen adsorption.^20,21 In other cases, the oxidation state of Co was tailored via surfactants in the sol–gel method.²² Further, Fe oxides provided improved thermal stability opposing the microporous silica densification at higher temperatures,²³ though La oxides led to the formation of silica mesoporous structures.²⁴ Due to the synthesis of Nb alkoxides in silica sol–gel, there is tendency to form Nb–O–Si bonds and chains,^25,26 instead of nanoparticle domains as it is the case of the metals discussed above.

A great majority of the work on metal and metal oxide silica has focused on the formation of nanoparticles and their properties for catalysis, adsorption and pore formation. The manipulation of nanoparticle size with changing nanoparticle phase has also been investigated by several chemical synthesis methods and thermal treatment. Tsuru and co-workers²⁷ reported the effect of temperature on Pd particle sizes in silica matrix for membrane application, which ranged from 31 to 42 nm under reduction conditions between 300 and 550 °C. In another example, PdO particles in PdOSiO₂ catalyst were hydrothermally aged and thermally aged in air, resulting in PdO sizes of 3 and 14 nm, respectively.²⁸ The use of two different silicon alkoxides and a micro-emulsion, in addition to hydrazine as a reducing agent, led to the formation of smaller Pd particles of 4.2 (±2.0) nm in silica shells.²⁹ Although metal oxide nitrates have been extensively used in the preparation of metal oxide,^30,31 metal chlorides have also shown as a good precursor in controlling particle sizes in silica matrices. Examples include Pt chloride complexation MCM-41 (ref. 32), Pd and Au chlorides complexation with silica dendrimer,³³ Ni and Co chlorides in silica sol–gel.³⁴

Phase change of the metal embedded within the silica matrix plays a major role in further tailoring the properties of silica materials. Miller and co-workers reported a reversible effect whereby the gas permeation increased and decreased as cobalt oxide silica membranes were reduced or oxidized respectively over several cycles.³⁵ This was attributed to changes in the silica pore structure caused by the redox effect on the cobalt particle. Recently, it was reported that in palladium cobalt oxide silica, Pd preferentially reduced instead of Co, thus resulting in a slightly enlargement of the silica pore sizes.³⁶ In another case, lanthanum formed silicates in LaCoSi membranes, and under redox conditions, the gas selectivity increased due to the slight closure of larger pores in the silica derived structure.³⁷ Whilst these studies have shown clear evidence that structural changes in the metal oxide silica matrix occurred as a result of the chemical methods use to prepare the materials, the redox effect adds an extra functionality. The precise nature of the interactions between the reduced/oxidised metal or metal oxide particles and the silica matrix warrants further understanding.

Therefore, this work investigates the redox effect on the particle formation of metal and metal oxides in silica structures. To address this investigation, this work departs from the conventional approach of high temperature oxidation of metal oxide silica xerogels, followed by high temperature reduction. The proposed approach in this work is to use palladium chloride (PdCl₂) as a metal/metal oxide precursor in silica sol–gel and carry out the reduction step at room temperature before calcining the silica structure. The choice of Pd chloride as the salt metal precursor is based on its superior solubility in silica sol–gel as compared to Pd nitrate. In a similar fashion, oxidised samples can also be prepared by calcining the xerogel in under atmospheric conditions. Therefore, the xerogels were studied at varying redox thermal treatments to elucidate the metal phase and structural evolution of Pd or PdO particles in the silica matrix.

2. Experimental

Palladium doped silica materials were synthesised via the acid catalysed sol–gel processing of tetraethyl orthosilicate (TEOS). Initially palladium chloride was dissolved in a solution of concentrated hydrochloric acid. The solution was then diluted with ethanol and cooled to 0 °C in an ice bath. Finally TEOS was added drop wise at a rate of 5 drops per second to the solution at 0 °C which was stirred for 3 h. The drop-wise method coupled with stirring of the solution allowed for homogeneous mixing whilst the mixing at 0 °C avoided the premature partial hydrolysis of TEOS. The final molar ratio of all reagents was 4.0TEOS : 0.24PdCl₂ : 41.5H₂O : 11HCl : 256EtOH. Xerogels were produced by drying the reacted sol in an oven at 60 °C for 10 days. The resultant monoliths were then crushed by mortar and pestle to produce a xerogel powder. The xerogels were subsequently calcined at 630 °C for 2.5 hours with a ramping rate of 1 °C min⁻¹ from room temperature. For the oxidised series (PdOSi) samples were calcined in air and for the reduced series (PdSi) samples were calcined in pure hydrogen, both under a flow rate of 60 mL min⁻¹. It should be noted that the PdOSi and PdSi notation refers to the oxidised and reduced palladium silica particles, respectively, where the particles were embedded into the silica matrix.

Characterisation was carried out on samples which were prepared by finely crushing calcined xerogel powders. The phases of the palladium species were detected by X-ray diffraction (XRD) which was conducted using a Bruker D8 Advance with a graphite monochromator using Cu Kα radiation. The tested range of 2θ was from 30° to 90°. High resolution transmission electron microscopy (HR-TEM) was used to determine the nanoparticle size of the palladium species and to confirm their phase. The xerogel samples were in ethanol and coated on a carbon grid. Measurements were conducted using a JEOL-JEM 2100 transmission electron microscope operating at 200 keV. The porous structure of the xerogels were analysed using a Micromeritics Tristar3020 N₂ adsorption apparatus. Samples were degassed in a Micromeritics VacPrep061 under a vacuum of 2 Pa for 12 hours. The pore size distribution (PSD) was calculated via density functional theory (DFT) software available in the Micromeritics Tristar3020. The calculation assumed cylindrical pores with an oxide surface. A regularization factor of 0.25 was used for curve smoothing purposes.

3. Results and discussion

The XRD patterns in Fig. 1a for the xerogels calcined in air show 2θ peaks at 16.6, 37.6 and 40.0°, which correspond to the PdCl₂ phase.³⁸ As the xerogel was heated under an air atmosphere, the PdCl₂ fully oxidised between 400 °C and 630 °C to palladium oxide.³⁹ These results suggest that the oxidization of Cl⁻ in the PdCl₂ phase. In contrast, when the xerogel was exposed to a hydrogen atmosphere at room temperature (Fig. 1b), PdCl₂ decomposed instantaneously to metallic palladium,⁴⁰ a phase that it retained at all temperatures tested up to 630 °C.

Fig. 1 XRD spectra of palladium silica xerogels after calcination in (a) air (PdOSi) and (b) hydrogen (PdSi) at various temperatures.

N₂ adsorption was implemented to determine the effect of the palladium chloride decomposition temperature on the porosity of the silica xerogel. The PdSi and PdOSi isotherms are shown in Fig. 2a. Both xerogels exhibit a type I isotherm, characteristic of microporous materials. It can be seen that the main difference between the isotherms occurs at low relative pressures, with the PdSi xerogel reaching full N₂ adsorption capacity around P/P_o = 0.1, while the PdOSi xerogel achieves this around P/P_o = 0.2. These results suggest the reduced PdSi xerogel has a smaller pore size and narrower pore size distribution (PSD) than the oxidized PdOSi xerogel, attributed to reaching saturation at lower relative pressure. This is confirmed in Fig. 2b, where the PSD exhibits additional porosity centered around 2 nm for the PdOSi xerogel. The PdSi xerogel does not contain porosity in this region, with a maximum pore size around 1.5 nm.

Fig. 2 N₂ adsorption (a) isotherm and (b) pore size distribution for PdSi, PdOSi xerogels calcined at 630 °C.

To further investigate the differences in the crystallite size of the xerogels, the crystallite sizes of the palladium phases were calculated by the Scherrer equation and are displayed in Fig. 3. It can be seen that the crystallite size for the reduced sample remained essentially constant over the calcination procedure. This was not the case for the oxidised xerogel which displayed a significantly smaller crystallite size at 500 °C and corresponded with the decomposition of PdCl₂ to PdO. After complete decomposition at 630 °C, the PdO crystallite returned to a size comparable with the original PdCl₂ crystallite. Fig. 3 also highlights the significant drop in the crystallite diameter upon reduction of the dried xerogel (60 °C). The crystallite diameter of the reduced xerogels was approximately one fourth of the oxidised xerogels. Perhaps coincidently, the specific volume of metallic palladium (8.85 cm³ mol⁻¹) is also one fifth that of PdCl₂ (44.3 cm³ mol⁻¹). It is thus expected that the crystallite volume should decrease by a factor of 4–5 with the conversion of PdCl₂ to Pd.

Fig. 3 Crystallite size for palladium silica xerogels calcined in air (black) and hydrogen (red).

The decomposition of PdCl₂ was monitored by observing the change in colour of the PdSi xerogel. When the atmosphere was switched from air to hydrogen, the xerogel completely changed colour from brown (PdCl₂) to black (Pd) within seconds at room temperature. As smaller crystals were observed in Fig. 1 and 3 for the reduced PdOSi xerogel, it is thus proposed that, due to the rapid rate of PdCl₂ decomposition coupled with the increase in crystallite density, the large PdCl₂ crystallites split into many smaller Pd crystallites upon reduction. This is in line with the shrinking core model, associated with gas reaction with solid particles leading to formation of smaller particles with porous structures.^41,42 This process leaves voids within the silica structure as schematically illustrated across process (a) in Fig. 4.

Fig. 4 Palladium nanoparticle morphology with respect to calcination method.

Similar to the PdCl₂–Pd phase conversion, the PdCl₂–PdO transition results in a reduction in the volume of the Pd nanoparticle phase since the specific volume of PdO (14.7 cm³ mol⁻¹) is one third that of PdCl₂. This is clearly seen in Fig. 3, as the crystallite diameter reduces from ∼43 nm for PdCl₂Si to ∼12 nm for the oxidized PdOSi sample as the temperature increases from 400 to 500 °C. This highlights a significant difference between the PdOSi which oxidized at 500 °C versus the PdSi which reduced very quickly in a matter of seconds under reducing conditions at room temperature. It is interesting to observe that the crystallites match in size for both PdOSi and PdSi at 500 °C. However, further increase in temperature to 630 °C yielded different crystalline formation behavior, as the PdSi remained almost constant whilst the PdOSi increased by fourfold. Therefore the calcination conditions and the temperature at which the PdCl₂ decomposition in silica occurred directly affected the size of the Pd nanoparticle.

Transmission electron microscopy (TEM) was performed on the oxidised and reduced xerogels calcined at 630 °C to justify link between the crystallite size determined by XRD and the nanoparticle size used in the nanoparticle formation mechanism. Fig. 5a and b show the representative micrographs of the PdOSi and PdSi xerogels, respectively, while Fig. 5c and d show high resolution images. The inter-lattice fringes of the PdOSi nanoparticle were measured as 0.307 nm (left) and 0.303 nm (right), while the inter-lattice fringes of the PdSi nanoparticle were measured to be 0.221, 0.226, 0.229 nm. These values match well with literature values for the PdO(100) (0.305 nm) and the Pd(111) (0.225 nm) phases respectively.^43,44 Interestingly, it can be seen that the nanoparticles in the PdOSi xerogel are far larger than those in the PdSi xerogel, which matches the XRD results.

Fig. 5 Representative TEM images of (a) PdOSi and (b) PdSi xerogels. The inter-lattice fringes of the nanoparticles highlighted in (a) and (b) are displayed in (c) and (d) respectively.

To explain these unusual results, the conversion of the PdCl₂ phase was viewed within the context of the condensation state of the surrounding silica matrix. In the case of the PdSi xerogel, the phase conversion occurred at room temperature (and the silica had only been exposed to 60 °C during drying). Under these dry conditions, gelation of the sol has already taken place⁴⁵ resulting in the crosslinking of the silica network.^46,47 The silica structure at this initial stage is relatively uncondensed at these low temperature conditions.^48,49 Upon heating, the lack of rigidity in the silica structure causes it to collapse around the Pd nanoparticles as seen over process (b) in Fig. 4. As the calcination further progresses the silica structure continues to condense^50–52 in close proximity to the Pd nanoparticles. As such, the silica structure provided a hindrance for metal Pd nanoparticles to diffuse towards each other to coalesce at higher temperatures. This results in the maintenance of a constant Pd crystallite size throughout the entire calcination process as seen in Fig. 3, supported by small Pd nanoparticles (Fig. 5), and represented schematically across processes (b) and (c) in Fig. 4. The rapid reduction of PdCl₂ to Pd at room temperature, coupled with the encapsulation by uncondensed silica, results in the formation of Pd nanoparticles of ∼11 nm, a far smaller size than those of ∼31 to 44 nm produced in studies by altering the reduction conditions at different temperatures.²⁷

In the case of the PdOSi xerogel, the PdCl₂ decomposition to metal oxide occurs around 450 °C (refer to Fig. 3). This process led to a reduction in the nanoparticle size, as shown in process (d) in Fig. 4, and resulted in the formation of porous space surrounding the nanoparticle. Since the silica continues condensation and densification at these temperatures,^53,54 the new PdO phase is able to diffuse through the weak silica branches and agglomerate in the voids left from the larger PdCl₂ phase. As the calcination temperature continued to rise, the ongoing re-arrangement of the silica matrix resulted in the closing of the remaining large voids, as shown in processes (d) and (e) in Fig. 4, coupled with the formation of large particles supported by the TEM micrographs in Fig. 5.

The structural bulk properties determined by N₂ adsorption in Fig. 2 also supports the schematic of the redox effect proposed in Fig. 4. The agglomeration of PdO forming large particles caused structural re-arrangements in the silica matrix leading to changes in pore size distribution, as evidenced by a broader pore size distribution in Fig. 2b with the addition of an extra pore size peak at 2 nm. In the case of the PdSi xerogels, the reduction at room temperature allowed for a fine control to produce small pore sizes and narrower pore size distribution, as verified by N₂ adsorption. Therefore, the reduction process conferred superior structural tailorability, by enveloping small Pd nanoparticles in the silica matrix which remained undisturbed though the calcination process.

4. Conclusions

This work shows that the PdSi xerogels reduced at room temperature formed small nanoparticles with small crystallite sizes, small pore sizes and narrow pore size distribution. Contrary to these results, the oxidized PdOSi xerogels resulted in larger values for these parameters. For instance, crystallite size was fourfold larger than the reduced PdSi. Therefore, the reduction of Pd at room temperature conferred a superior fine structural control otherwise not available on the oxidized samples. This final control was attributed to the enveloping of the Pd nanoparticles by the silica matrix which hindered their coalescence at high temperatures. In the case of PdOSi, the PdO phase was formed only above 400 °C due to the degradation of the larger PdCl₂ particle. The void left by the latter particle after degradation allowed for further rearrangement of the silica matrix and PdO agglomeration, thus forming larger PdO nanoparticles, as the calcination temperature increased to 630 °C.

Acknowledgements

The authors acknowledge financial support provided by the Australian Research Council (ARC) through Discovery Project Grant DP110101185. The authors would like to thank the facilities and the scientific and technical assistance of the Australian Microscopy & Microanalysis Research Facility at the Centre for Microscopy and Microanalysis at The University of Queensland. J. C. Diniz da Costa gratefully thanks the support given by the ARC Future Fellowship Program (FT130100405).

References

1. C. Yacou, S. Smart and J. C. Diniz da Costa, Energy Environ. Sci., 2011, 5, 5820–5832.
2. H. Qi, J. Han, N. Xu and H. J. M. Bouwmeester, ChemSusChem, 2010, 3, 1375–1378.
3. M. Kanezashi, M. Sano, T. Yoshioka and T. Tsuru, Chem. Commun., 2010, 46, 6171–6173.
4. M. Elma, D. K. Wang, C. Yacou, J. Motuzas and J. C. Diniz da Costa, Desalination, 2015, 365, 308–315.
5. M. Kanezashi and M. Asaeda, J. Membr. Sci., 2006, 271, 86–93.
6. X. Zhu, Y. Chen, F. W. Zhang, J. R. Niu, M. Xie and J. Ma, RSC Adv., 2014, 4, 2509–2514.
7. F. Adam, J. Andas and I. A. Rahman, Chem. Eng. J., 2010, 165, 658–667.
8. R. E. Owen, J. P. O'Byrne, D. Mattia, P. Plucinski, S. I. Pascu and M. D. Jones, Chem. Commun., 2013, 49, 11683–11685.
9. C. Yacou, M.-L. Fontaine, A. Ayral, P. Lacroix-Desmazes, P.-A. Albouy and A. Julbe, J. Mater. Chem., 2008, 18, 4274–4279.
10. C. Yacou, A. Ayral, A. Giroir-Fendler, M. L. Fontaine and A. Julbe, Microporous Mesoporous Mater., 2009, 126(3), 222–227.
11. K. Moreira Diniz, F. A. Gorla, E. Schwingel Ribeiro, M. B. Olimpio do Nascimento, R. J. Correa, C. R. Teixeira Tarley and M. G. Segatelli, Chem. Eng. J., 2014, 239, 233–241.
12. Z. B. Sun, D. Guo, H. Z. Li, L. Zhang, B. Yang and S. Q. Yan, RSC Adv., 2015, 5, 11000–11008.
13. J. Qu, W. Li, C.-Y. Cao, X. J. Yin, L. Zhao, J. Bai, Z. Qin and W. G. Song, J. Mater. Chem., 2012, 22, 17222–17226.
14. L. Yao, K. Kan, Y. Lin, J. Song, J. Wang, J. Gao, P. Shen, L. Li and K. Shi, RSC Adv., 2015, 5, 15515–15523.
15. P. R. Ohodnicki, J. P. Baltrus and T. D. Brown, Sens. Actuators, B, 2015, 214, 159–168.
16. C. Robertson, R. Beanland, S. A. Boden, A. L. Hector, R. J. Kashtiban, J. Sloan, D. C. Smith and A. Walcarius, Phys. Chem. Chem. Phys., 2015, 17, 4763–4770.
17. L. Liu, D. K. Wang, D. L. Martens, S. Smart, E. Strounina and J. C. Diniz da Costa, RSC Adv., 2014, 4, 18862–18870.
18. R. Igi, T. Yoshioka, Y. H. Ikuhara, Y. Iwamoto and T. Tsuru, J. Am. Ceram. Soc., 2008, 91, 2975–2981.
19. L. Liu, D. K. Wang, D. L. Martens, S. Smart and J. C. Diniz da Costa, J. Membr. Sci., 2015, 475, 425–432.
20. Y. H. Ikuhara, H. Mori, T. Saito and Y. Iwamoto, J. Am. Ceram. Soc., 2007, 90, 546–552.
21. K. Yoshida, Y. H. Ikuhara, S. Takahashi, T. Hirayama, T. Saito, S. Sueda, N. Tanaka and P. L. Gai, Nanotechnology, 2009, 20, 315703.
22. G. Olguin, C. Yacou, S. Smart and J. C. Diniz da Costa, RSC Adv., 2014, 4, 40181–40187.
23. A. Darmawan, S. Smart, A. Julbe and J. C. Diniz da Costa, Materials, 2011, 4, 448–456.
24. B. Ballinger, C. R. Miller, J. Motuzas, S. Smart and J. C. Diniz da Costa, Sci. Rep., 2015, 5, 8210,  DOI:10.1038/srep08210.
25. V. Boffa, D. H. A. Blank and J. E. ten Elshof, J. Membr. Sci., 2008, 319, 256–263.
26. H. Qi, H. Chen, L. Li, G. Zhu and N. Xu, J. Membr. Sci., 2012, 421–422, 190–200.
27. M. Kanezashi, D. Fuchigami, T. Yoshioka and T. Tsuru, J. Membr. Sci., 2013, 439, 78–86.
28. R. Gholami and K. J. Smith, Appl. Catal., B, 2015, 168, 156–163.
29. J.-N. Park, A. J. Forman, W. Tang, J. Cheng, Y.-S. Hu, H. Lin and E. W. McFarland, Small, 2008, 4, 1694–1697.
30. S. Esposito, M. Turco, G. Ramis, G. Bagnasco, P. Pernice, C. Pagliuca, M. Bevilacqua and A. Aronne, J. Solid State Chem., 2007, 180, 3341–3350.
31. L. Liu, D. K. Wang, P. Kappen, D. L. Martens, S. Smart and J. C. Diniz da Costa, Phys. Chem. Chem. Phys., 2015, 17, 19500–19506.
32. A. Mastalir, B. Rac, Z. Kiraly and A. Molnar, Catal. Commun., 2008, 9, 762–768.
33. R. W. J. Scott, O. M. Wilson and R. M. Crooks, Chem. Mater., 2004, 16, 5682–5688.
34. C. Cantalini, M. Post, D. Buso, M. Guglielmi and A. Martucci, Sens. Actuators, B, 2005, 108, 184–192.
35. C. R. Miller, D. K. Wang, S. Smart and J. C. Diniz da Costa, Sci. Rep., 2013, 3, 1648,  DOI:10.1038/srep01648.
36. B. Ballinger, J. Motuzas, S. Smart and J. C. Diniz da Costa, J. Membr. Sci., 2014, 451, 185–191.
37. B. Ballinger, J. Motuzas, S. Smart and J. C. Diniz da Costa, J. Membr. Sci., 2015, 489, 220–226.
38. L. Radev, M. Khristova, D. Mehandjiev and B. Samuneva, Catal. Lett., 2006, 112, 181–185.
39. D. L. Mowery and R. L. McCormick, Appl. Catal., B, 2001, 34, 287–297.
40. K. Yan, C. Jarvis, T. Lafleur, Y. Qiao and X. Xie, RSC Adv., 2013, 3, 25865–25871.
41. G. Uhde and U. Hoffmann, Chem. Eng. Sci., 1997, 52, 1045–1054.
42. J. Y. Park and O. Levenspiel, Chem. Eng. Sci., 1975, 30, 120–1214.
43. S. Penner, D. Wang, B. Jenewein, H. Gabasch, B. Klötzer, A. Knop-Gericke, R. Schlögl and K. Hayek, J. Chem. Phys., 2006, 125, 094703,  DOI:10.1063/1.2338028.
44. R. Kumar, D. Varandani, B. Mehta, V. Singh, Z. Wen, X. Feng and K. Müllen, Nanotechnology, 2011, 22, 275719.
45. N. K. Raman, M. T. Anderson and C. J. Brinker, Chem. Mater., 1996, 8, 1682–1701.
46. C. J. Brinker and G. W. Scherer, Sol Gel Science: The Physics and Chemistry of the Sol Gel Processing, Academic Press, San Diego, USA, 1990.
47. C. J. Brinker and G. W. Scherer, J. Non-Cryst. Solids, 1985, 70, 301–322.
48. G. Orcel, L. L. Hench, I. Artaki, J. J. Jonas and T. W. Zerda, J. Non-Cryst. Solids, 1988, 105, 223–231.
49. A. J. Vega and G. W. Scherer, J. Non-Cryst. Solids, 1989, 111, 153–166.
50. A. Bertolluzza, C. Gagnano, M. A. Morelli, V. Gottardi and M. Guglielmi, J. Non-Cryst. Solids, 1982, 48, 117–128.
51. A. Duran, C. Serna, V. Fornes and J. M. Fernadez Navarro, J. Non-Cryst. Solids, 1986, 82, 69–77.
52. W. G. Fahrenholtz, D. M. Smith and D. W. Hua, J. Non-Cryst. Solids, 1992, 144, 45–52.
53. C. J. Brinker, G. W. Scherer and E. Roth, J. Non-Cryst. Solids, 1985, 72, 345–368.
54. F. Orgaz and H. Rawson, J. Non-Cryst. Solids, 1986, 82, 57–68.

---