J. MATER. CHEM., 1994, 4( 12), 1921-1926 Investigations of Structure and Protonic Conductivity in Composites of Hydrous Antimony(v) Oxide and Mordenite Gary 6.Hix,*" Yves Rouillard," Robert C. T. Slade" and Bernard Ducourantb a Department of Chemistry, University of Exeter, Stocker Road, Exeter, UK EX# #OD Laboratoire des Agregats Moleculaires et Materiaux Inorganiques, URA CNRS 79, Universite Montpellier 2, Place Eugene Bataillon, 34095 Montpellier Cedex 5, France Composites prepared by hydrolysis of antimony pentachloride in the presence of the zeolite mordenite have been characterised by X-ray powder diffractometry (XRD), 29Si and 27AI MAS NMR, "'Sb Mossbauer spectroscopy and investigations of ac conductivity as a function of relative humidity (RH). Both XRD and MAS NMR studies show degradation of the zeolite framework, the extent of the damage depending upon the amount of antimony pentachloride used in the preparation.Mossbauer spectra are consistent with the antimony-containing component being amorphous antimonic acid, as anticipated from the preparative conditions. The composites are protonic (H+) conductors, but do not show the enhanced conductivities characteristic of 'tin mordenites', which show the same disruption of the zeolite framework. Composite proton-conducting solid electrolytes are formed by the bulk mixing of known proton conductors and inert oxide dispersed Enhancements in conductivities over those exhibited by the original conductors are dependent upon the nature of the oxide and its particle size.Zeolites are known to be proton conductor^.^ The protonic conductivity of the zeolite mordenite has been show to be enhanced by the inclusion of a dispersed tin@) oxide pha~e,~-~ giving 'tin mordenites'. The exact nature of these materials is unclear, but the bulk of the evidence indicates that they are composite in nat~re.~ In those materials both the matrix and the dis- persed oxide are proton conductors. The enhancement in conductivity may be due to both an increased number of acid sites in the zeolite, this being due to fragmentation of the framework by dealumination, and the decreased particle size of the microdispersed tin(1v) oxide.7 In terms of the model of a hydrous oxide proposed by England et a[.,' the decreased Sn02 particle size leads to an increased surface area and hence a larger number of charge carriers on the acidic surface of those particles.Antimonic acid (HSbO, .nH,O), sometimes termed 'hydrous antimony(v) oxide' (Sb20,.xH20, x =2n + l), can exist in a variety of structures (ilmenite, cubic, monoclinic, pyrochlore and amorphous forms) within which Sb occurs in SbVOb octahedra. The 'acid' formulation is correct for crystalline materials, protons (H + ) and H,O molecules being present within the crystallites. All forms exhibit proton conduction to a greater or lesser extent. The amorphous form exhibits the highest conductivity,' an observation which may indicate that a component of conduction occurs via hydrated acidic particle surfaces (as in the hydrous oxides').The conductivity of amorphous antimonic acid at ambient temperature is higher than that of hydrous tin(1v) oxide under the same conditions.8 For materials functioning as protonic conductors at near- ambient temperatures, the mechanisms of protonic conduction form a spectrum ranging from mechanical (Grotthus) transfer of H', through vehicular (H,O+) charge transport to liquid- like charge migration." The level of hydration of particle surfaces depends on relative humidity (RH). In systems in which conduction via these surfaces is significant, conductivi- ties increase at high RH as conduction is then more liquid-like. Most materials that conduct protons at ambient tempera- ture dehydrate at ca.60 "C [this is true of both hydrous tin(1v) oxide and antimonic acids8]. Loss of water removes the conduction pathway and leads to drastic reductions in pro- tonic conductivity. The conductivity of 'tin mordeni te' com- posites (0z lop2S cm-I at 100"C and 100% RH4-7)IS among the highest known at temperatures a little above that thresh- old. With that in mind, and noting that antimonic acid is normally considered more conductive than hydrous tin(1v) oxide, we have now undertaken a study of composites formed from mordenite and 'hydrous antimony(v) oxide'. The com- posites have been studied by X-ray powder diffractometry, 29Si and 27Al MAS NMR, "'Sb Mossbauer spectroscopy and electrical (conductivity) measurements. Experimental Zeolites (Na mordenite and H mordenite) were supplied by Laporte Inorganics.Varying quantities (given in Table 1) of antimony(v) chloride (BDH Chemicals) were added clropwise to aqueous suspensions of zeolite (8.0 g in 200 an3), the presence of the zeolite being a modification of the conditions for production of amorphous antimonic a~id.~,~ The resulting suspension was, in each case, freeze-dried and then heated at 100°C under dynamic vacuum for 3 h. The product was re- suspended three times in de-ionised water for 23 h, and samples were stored in desiccators of controlled relative humidity (RH =20, 60, 70 and 80%, above aqueous solutions of sulfuric acid of appropriate concentrations). X-Ray Powder Diffraction ( XRD) XRD profiles were recorded on a Philips diffraytometer (PW1050 goniometer, Cu-Km radiation, i= 1.54178 A).Data were collected over a 10 s period for every 0.1' of 20. Observed profiles were compared with computer-generated simulations for zeolite frameworks. Table 1 Masses of mordenite and SbC15 used in syntheses of composites mass of mordenite/g sample 1 8.0 3.7 sample 2 8.0 8.0 sample 3 8.0" 10.2 " H-Mordenite. MAS NMR 29Si (59.584 MHz) and 27Al (78.152 MHz) MAS NMR spectra at ambient temperature were recorded on a Varian VXR 300 spectrometer by the SERC Solid State NMR Service (Durham). Relaxation delays in recording 29Si spectra were 30s, determined as sufficiently long to avoid all saturation effects and spin rates were ca. 5 kHz. The corresponding delays and spin rates for 27Al spectra were 0.5 s and ca.5 kHz, respectively. "Si spectra were deconvoluted assuming gaussian line- shapes, thereby giving spectral parameters for the different Si sites. "'Sb Mossbauer Spectroscopy Spectra were recorded using a constant-acceleration spec-trometer, with both the sample and the Cal2lrnSnO3 source maintained at T=4.2 K [He (1) cryostat]. Powder samples (containing ca. 45 mg Sb) were dispersed in Apiezon grease. The sample holder presented a geometrical area of 3 cm2 to the oscillator. Data were acquired through a multi-channel scaling analyser as two spectra related through a mirror plane; folding these two spectra improved the counting statistics. Experimental spectra were fitted to calculated model spec- tra, using a program described by Ruenbauer and Birchall.12 Isomer shifts are expressed with respect to the source.Electrochemical Impedance Measurements Powder samples were compressed in a 13 mm die to give pellets of cu. 1 mm thickness. A small quantity of water was added prior to pressing (to aid the binding of the pellets). Pellet faces were coated with conductive silver paint (Acheson Electrodag 915) and copper or nickel discs were then attached using the silver paint as adhesive. Pellets were equilibrated in desiccators over aqueous solutions of sulfuric acid (RH =20, 60, 70 and 80% at ambient temperature). Sample pellets were mounted in a brass cell which held six samples simultaneously. The temperature (233 <T/K <3 13) was controlled by immersion of the cell in controlled-tempera- ture baths, allowing at least 30min for equilibrium to be achieved (this period was found by experience to be in excess of the minimum necessary to give temporal stability of impedance spectra). The relative humidity in the cell was J.MATER. CHEM, 1994, VOL. 4 controlled by placing the appropriate acid solution (as described previously) in the bottom of the cell. ac Impedance spectra (admittance or impedance plane) were collected in the range 5 Hz-1 MHz (with an oscillation voltage of 100mV) using a Hewlett-Packard 4192A LF impedance analyser controlled by an TBM-compatible com- puter using a program embedding the EQUIVCRT program (fitting impedance spectra to a user-defined equivalent circuit13).Results and Discussion Chemical Analyses The parent zeolites and the composites were analysed by X-ray fluorescence (XRF). The Si:Al ratio can be calculated simply from the XRF results. 29Si NMR spectra allow the Si:A1 ratio for a zeolite framework to be calculated; appli- cation of Loewenstein's rule, which disallows Al-0-A1 linkages in zeolite frameworks, gives m =0.4 m=0.4 where 14,,, is the relative intensity of a line assigned to a Q4(mAl) silicon site (rn is the number of -OAl linkages around the Si atom). The results of the various analyses are given in Table 2, and 29Si NMR line intensities are given in Table 3. The Si:Al ratios for the composites (from both XRF and NMR) are greater than those of the parent materials.This shows that more of the Si nuclei in the remaining zeolite framework in composites have fewer Als as nearest neighbour tetrahedrally coordinated atoms and indicates a degradation of the zeolitic framework by acid-leaching of framework aluminium. The acid responsible, HC1, is generated by the reaction of the antimony(v) chloride with water. X-Ray Powder Diffraction Studies The XRD profiles obtained for composites (Fig. 1) show the zeolite component in composites to be distinctly less crystalline than the pristine zeolites. The decrease in crystallinity is seen to become more pronounced as the amount of antimony(v) chloride used in the preparation increases. In the case of sample 1, the material with the lowest antimony content, the Table 2 Chemical analyses of composites and parent zeolites XRF Si :Al mass of SiO, (%) A1203 (Yo) Sb,O, (Yo) XRF NMR product/g ~~ ~ sample 1 60.38 7.13 3 1.08 7.18 7.04 9.35 sample 2 47.95 5.57 45.72 7.26 6.85 11.95 sample 3 Na-mordenite 42.67 81.80 4.91 10.90 52.15 - 9.37 6.15 9.44 6.11 13.56 ~ H-mordenite 90.90 7.31 - 9.15 8.91 ~ Table 3 29Si MAS NMR spectral parameters for composites and parent zeolites Si (OAl) Si (1Al) Si (2x1) sample 1 sample 2 sample 3 Na-mordenite H-mordenite 112.6 288 43 105.7 434 52 98.3 346 5 112.9 298 48 105.8 418 45 99.1 465 7 112.6 292 49 105.8 420 46 100.1 27 1 6 112.6 293 40 105.9 355 55 98.9 245 5 112.7 313 58 105.3 359 39 98.1 314 3 J.MATER. CHEM., 1994, VOL. 4 Fig.1 X-Ray powder diffraction profiles of Na mordenite and composites: (a) Na-mordenite, (b) sample 1, (c) sample 2 and (d) sample 3 principal zeolite peaks are seen, though much reduced in intensity. The sample with the highest antimony content (sample 3, made from H-mordenite) can be seen to be almost completely amorphous, with no discernible peaks. Increasing the amount of antimony(v) chloride used in the preparation increases damage to the zeolitic framework. There are no peaks due to the presence of antimonic acid or Sb205 in any of the profiles, this indicating that the non- zeolitic component has a very small particle size. MAS NMR Studies The results of NMR studies are fully consistent with those from chemical analysis and XRD studies.The 29Si MAS NMR spectra of the composites (Fig. 2) show Si peaks at chemical shifts similar to those observed for the parent mordenite phases.14 The line assigned to the Si( 1Al) site remains largely unchanged with respect to the parent zeolite, but its relative intensity changes (Table 4).For samples 1 and 2 the intensity of the line due to the Si(lA1) site decreases on composite formation, this being consistent with preferential removal of Als linked to Sis of type Si( 1Al) in Na mordenite. The case of sample 3 (prepared from H-mordenite) is different, the near-total disruption of the zeolite being evident in the XRD profile [Fig.l(d)]. The lines due to Si sites with A1 tetrahedral neighbours are broader than those of the host zeolites, the largest difference being in excess of 200 Hz. This line broadening is a result of the fragmentation of the zeolite microcrystals into smaller particles by acid leaching, resulting in a range of environments for the Si nuclei and a distribution of chemical shifts. 27Al spectra (Fig. 3) are dominated by the signal of tetra- hedrally coordinated A1 (in the zeolite framework), but also indicate the presence of some octahedrally coordinated alu- minium, i.e, a peak at 6~0typical of Al"' octahedrally coordinated by 0 atoms [c.f A1(H20),3+(aq), which is the reference]. Peak positions are given in Table 5.The octahedral A1 is that acid leached from the framework. It can again be seen that the extent of leaching is more pronounced with higher antimony content, this being evident in the increase in the intensity of the line corresponding to six-coordinate aluminium (Table 4). I2'Sb Mossbauer Studies In 12'Sb Mossbauer spectroscopy, eight quadrupole trans- itions are expected for the 7/2* 5/2 decay. Quadrupole split- 50 Fig. 2 Proton-decoupled high resolution 29Si MAS NMR spectra of (a) sample 1, (b)sample 2 and (c) sample 3. Spectra were recorded at ambient temperature, with spin rate ca. 5 kHz and 42 rf pulses. The recorded spectrum is given as a dotted line and the fitted gaussians are shown by the solid lines. The difference between the observed and calculated spectrum is given as a solid line below each spectrum.Table4 "A1 MAS NMR spectral parameters for composites and parent zeolites sample 1 53.2 82.2 -0.9 17.8 sample 2 53.4 70.1 -0.9 29.9 sample 3 53.4 65.6 -0.6 34.4 ~Na-mordenite 54.1 100 -H-mordenite 54.7 100 -- 00 Fig. 3 Proton-decoupled high resolution 27Al MAS NMR spectra of composites: (a) sample 1, (h)sample 2 and (c) sample 3. Spectra were recorded at ambient temperature, with spin rates of ca. 5 kHz and employing ~16rf pulses. *Indicates spinning side bands. Table 5 '"Sb Mossbauer parameters for composites (numbers in brackets are standard deviations arising in the fitting of individual spectra) 4," I mm s-' AEq/ mms-' FWHM/ mm s-' intensity x2 sample 1 sample 2 0.61(1) 0.74(3) -0.1 l(2) 5.20(2) 9.84(4) 4.88(2) 1.06(1) 0.76(4) 0.76(2) 1.00 0.75 0.25 1.36 1.52 sample 3 0.73(2) 6.92(3) 1.20(1) 1.00 1.79 tings are detected in spectra of antimony compounds, but line splittings are less than experimental 1ine~idths.l~ The '"Sb spectra for the composites in this work (Fig.4) all showed a single line at close to zero velocity, consistent with Sb present in SbV06 octahedra (as in the source; diso z0mm s-' is typical of that environment16). The anti- mony-containing component of the composites is amorphous and SbV06octahedra are therefore likely to be present with a distribution of local environments and distortions. It is therefore impossible to be completely non-arbitrary in fitting of the individual spectra.The approach adopted was to assume the minimum number of 'Sb sites' (each giving a contribution to the experimental line) necessary to give an adequate fit (judged visually) to each spectrum. The spectra of samples 1 and 3 were satisfactorily fitted with the inclusion of only one SbV site type, giving Mossbauer parameters (Table 5) similar to those for SbVO, in Sb205.16-18 The likely Sb environment in the composites is, however, in amorphous antimonic acid; Sb20, is fabricated from antimonic acids only on heating to higher temperatures (275 "C) than used in this work. The spectrum of sample 2 was, however, satisfactorily fitted only as the sum of a minimum of two contributions (Table 5).That is likely to reflect differences in the distri- butions of local geometries for SbVO, octahedra. J. M'4TER. CHEhl., 1994, VOL. 4 150 000 140000 130 000 120 000 130 000 c 120000v) 3 8 110000 120 110 100 90 ,II I I,, I II, I I1 <*--20 0 velocity/mm s-' Fig. 4 '"Sb Mossbauer spectra of composites: (a)sample 1, (6)sample 2 and (c) sample 3 The '"Sb Mossbauer spectra are consistent with description of the antimony component in the composites as amorphous antimonic acid. ac Conductivity Studies Impedance spectra consisted of a high-frequency arc and a sloping rise of reactance with respect to resistance at lower frequency, this being typical of ionic conduction and blocking electrodes. The conductance of the sample was extracted as the intercept of the low-frequency line with the resistance (real) axis.Over the experimental temperature range (233 < T/K <3 13), and at all relative humidities, all three samples showed Arrhenius-like behaviour in the temperature dependence of their conductivities (Fig. 5). Empirical acti- vation energies E, for protonic conduction are given in Table 6. The conductivity for a given composite and tempera- ture increases with RH, as discussed in the Introduction. The degree of hydration of the conducting surfaces will vary with temperature and it follows that E, values contain a contri- bution from this variation and are best treated as empirical parameters only. Sample 1 has conductivities of the same order of magnitude as those for the parent zeolite (Na mordenite, 0% 10-8-10-5 S cm-' at 50% RH3)under similar conditions.Table 6 Activation energies for protonic conduction in composites at different relative humidities EJkJ mol-' RH (Yo) sample 1 sample 2 sample 3 20 38k 1 45+1 23k2 60 42k 1 45k1 38f I 70 44+2 42f1 38k1 80 29k2 45f2 39f2 J. MATER. CHEM., 1994, VOL. 4 -2 r -3 --4--5 --6-\. --2 --3 2 7 ‘E -4-0 9 I-b -5-v +\ +\+ -2 r --5 I 1 I I I Fig. 5 Temperature-dependent protonic conductivities for (a)sample 1. (b)sample 2 and (c) sample 3, at RH=20 (+), 60 (x), 70 (0)and 80% ( ). Lines shown are the linear regressions and correspond to the activation energies given in Table 6.All experimental values are less than those observed for antimonic acid (including the less-conductive crystalline forms), investigations of ~hi~h~,~~-~~have found 0% 10-3-10-4 S cm-l at ambient temperature. Sample 2 is similar in nature to sample 1 (though less crystalline zeolite is present) and has conductivities similar to those of sample 1. Sample 3 (in which disruption of the zeolite framework is most extensive) exhibits conductivities of the same order of magnitude as the parent H-mordenite, except at RH =80% where the conductivity of sample 3 is higher. The composites in this work do not show the enhancement of protonic conductivities (relative to those of parent zeolites) exhibited by ‘tin m~rdenites’,~-’ but do show disruption of the zeolite framework, as in the case of ‘tin mordenite’ and other ‘tin zeolite^'.^ Antimonic acid [‘hydrous antimony(v) oxide’] itself differs from hydrous tin(1v) oxide in having an intraparticle protonic conduction pathway in addition to surface conduction [in hydrous tin(1v) oxide the intraparticle structure is that of SnO,]. The formation of composites in this work was aimed at giving a high surface area for the antimonic acid component.The results in this study may indicate that the intraparticle conduction pathway is of much greater significance in amorphous antimonic acid than in crystalline antimonic acids (reflecting a disordered/glassy intraparticle structure for the amorphous form) and that surface conduction is much less significant than has been supposed previously (see Introduction).Alternatively. the anti- monic acid may have been deposited in particles too large for a continuous conduction pathway via them alone. Conclusions In ‘hydrous antimony(v) oxide/mordenite’ compo4tes, the zeolite component exhibits extensive damage to the zeolite framework. The extent of that damage increases with the amount of antimony pentachloride used in the preparation. This is evident both from the XRD profiles, in which there is a broadening or disappearance of peaks due to the zeolite and from the MAS NMR studies, in which there is a reduction in the intensity of the line corresponding to the Si(1Al) site at the same time as octahedrally coordinated A1 sites are detected.”‘Sb Mossbauer spectra are consistent with the aiitimony- containing component of the composites being amorphous antimonic acid, as would be anticipated from the prcparative conditions. ‘Hydrous Sb20,-mordenite’ composites do not show the enhanced protonic conductivities characteristic of ‘tin morden- ite~’,~-~even though they do show the framework disruption found to be characteristic of those system^.^ The chemistry of the ‘oxide’ component in composites formed from zeolites and ‘hydrous oxides’ is therefore of considerable Significance in determining the protonic (H’ ) conductivity of those composites. We thank Dr D. J. Jones of 1’Universite Montpellxer 2 for useful discussions on the fitting of Mossbauer spectra.We thank Laporte Inorganics for provision of zeolite samples. We thank the SERC National Solid State NMR service (Ihrham) for recording of the NMR spectra and for subsequent decon- volutions. We thank SERC for a studentship for G.B.H. We thank NATO for a travel grant enabling joint studies at Exeter and Montpellier. This work was part-funded under the Commission of the European Communities JOULE programme. References 1 R. C. T. Slade, H. Jinku and J. A. Knowles. Solid Sttrte lonics, 1992,50,287. 2 R. C. T. Slade and J. A. Knowles, Solid State Ionics, 199i. 46,45. 3 E. Krogh-Andersen, 1. G. Krogh-Andersen and E. Skou, in Proton Conductors: Solids, Membranes and Gek-Materials untl Devices, ed.Ph. Colomban, Cambridge University Press, Climbridge, 1992. ch. 14 and references therein. 4 N. Knudsen, E. Krogh-Andersen, I. G. Krogh-Andersen and E. Skou, Solid State Zonics, 1988,28-30, 621. 5 N. Knudsen, E. Krogh-Andersen, I. G. Krogh-Andersen and E. Skou, Solid State Zonics, 1989,35. 51. 6 N. Knudsen, E. Krogh-Andersen, I. G. Krogh-Andersen and E. Skou, Solid State Ionics, 1991,46, 89. 7 G.B. Hix, R. C. T. Slade, K. Molloy and B. Ducourant, J. Mater. Chem., 1994,4, 1913. 8 W. A. England, M. G. Cross, A. Hamnett, P. J. Wiseman and J. B. Goodenough, Solid State lonics, 1980, 1, 23 1. 9 M. Abe and T. Ito, Bull. Chem.Soc. Jpn, 1968.41,333. 10 Ph. Colomban, in Proton Conductors: Solidy, Membranes and Gels-Materials and Devices, ed.Ph. Colomban, Cambridge University Press, Cambridge, 1992, ch. 3. 1926 J. MATER. CHEM., 1994, VOL. 4 11 R. von Ballmoos, Collection of Simulated XRD Powder Patterns 18 G. G. Long, J. G. Stevens and L. H. Bowen, Inorg. Nucl. Chem. for Zeolites, Butterworth, Guildford, 1984. Lett., 1969, 5, 799. 12 13 K. Ruenbauer and T. Birchall, Hyperjne Interactions, 1979,7,125. B. A. Boukamp, Solid State lunics, 1986, 18& 19 136; 1986,20, 31. 19 X. Turrillas, G. Delabouglise, J. G. Joubert, 1 Fournier and J. Muller, Solid State lonics, 1987, 17, 169. 14 15 16 G. R. Hays, W. V. Van Erp, N. C. M. Alma, P. A. Couperus, R. Huis and A. E. Wilson, Zeolites, 1984,4,377. S. L. Ruby, G. M. Kalvius, R. E. Snyder and G. B. Beard, Phys. Rev., 1966, 148, 176. N. N. Greenwood and T. C. Gibb, in Mossbauer Spectroscopy, Chapman and Hall, London, 1972, p. 444. 20 21 D. J. Dzimtrowicz, J. B. Goodenough and P. J. Wiseman, Muter. Res. Bull., 1982, 17, 971. S. Yde-Andersen, J. S. Lundsgaard, J. Malling an4 J. Jensen, Solid Slate Iunics, 1984, 13, 81. 17 V. S. Shpinel, V. A. Bryukhanov, V. Kothekar, B. Z. Iofa and S. I. Semenov, Symp. Faraday Soc., 1968, I, 69. Paper 41036625; Received 16th June, 1994