J. Chem. Soc., Faraday Trans. 1, 1984,80, 1341-1356 Tin Oxide Surfaces Part 12.-A Comparison of the Nature of Tin(rv) Oxide, Tin(rv) Oxide-Silica and Tin(rv) Oxide-Palladium Oxide : Surface Hydroxyl Groups and Ammonia Adsorption BY PHILIP G. HARRISON* AND BARRY M. MAUNDERS Department of Chemistry, University of Nottingham, University Park, Nottingham NG7 2RD Received 28th April, 1983 The chemical nature of the surfaces of tin(1v) oxide, tin(1v) oxide-silica and tin(1v) oxide-palladium oxide have been compared by surface dehydroxylation, deuterium exchange and ammonia adsorption. Physisorbed and hydrogen-bonded molecular water is lost from all three oxides at temperatures up to ca. 350 K. Further significant loss of water does not occur until ca. 480 K, when condensation of adjacent surface hydroxyl groups occurs.At higher temperatures, a sharp band at 371 5 cm-l is observed for tin(1v) oxide-silica, which is assigned to the v(0H) stretching vibration of isolated surface Si-OH groups. This band is present even at 723 K, when essentially all other surface hydroxyl groups had desorbed. The thermal dehydroxylation behaviour of tin@) oxide and tin(1v) oxide-palladium oxide is similar, although serious transmission problems occur above 523 K for tin(1v) oxide. All three oxides undergo surface deuteration with deuterium oxide at 523 K, although exchange with tin(1v) oxide-silica is not complete until 633 K. Ammonia adsorption demonstrates that tin(1v) oxide and tin(rv) oxide-palladium oxide are predominantly Lewis acidic, although weak Bronsted acidity can be observed in the presence of water vapour.Strong Bronsted-acid sites can be produced by protonation with hydrogen chloride. Surface amide groups are formed on tin@) oxide-palladium oxide after evacuation at 500 K, probably on palladium sites. Tin(1v) oxide-silica exhibits both Lewis and Bronsted acidity, the amount of Bronsted acidity with respect to Lewis acidity decreasing with increasing temperature. Our earlier studies of the chemisorption properties of tin(rv) oxide have included a wide range of adsorbent molecules.1 It is well known that the chemical properties of oxide materials can be modified, sometimes substantially, by the inclusion of other metals or metalloids. We have previously compared the chemisorption behaviour of tin(rv) oxide modified by the inclusion of some first-row transition-metal ions;2 here we report a detailed comparitive study of the nature of the surfaces of tin(rv) oxide, tin(rv) oxide-silica and tin(rv) oxide-palladium oxide.EXPERIMENTAL Tin(1v) oxide, tin(1v) oxide-silica and tin(1v) oxide-palladium oxide were prepared by (C0)-precipitation of the appropriate (mixture of) the metal chloride solution in water by AnalaR aqueous ammonia solution. Assuming that complete conversion to metal oxide occurred, the coprecipitated gels had the stoichiometry SnO, - 0.25Si0, and SnO, * 0.02PdO. Self-supporting discs of the tin(1v) oxide or coprecipitated gels were prepared by finely grinding ca. 60 mg of the oxide with an agate mortar and pestle, drying the powder in air at ca.373 K for 0.5-1 h, then pressing it in a 2.5 cm diameter stainless-steel die with a compacting pressure of 34 MN 13411342 10 - - 8 - h $ 5 - * 6 - E - 3 4 - : - 2 - TIN OXIDE SURFACES TIK Fig, 1. Thermogravimetric results for tin(rv) oxide-silica. Plot of percentage weight loss against temperature of evacuation. Prior to use the discs were pretreated by evacuation in the infrared cell for at least 2 h at room temperature then for at least 3 h in the temperature range from room temperature to ca. 760 K at a vacuum pressure of 1.33 x N m-2. For pretreatment above 523 K, evacuation was usually followed by treatment with oxygen (1.33-3.32 kN m-2) at the evacuation temperature for at least 1 h, then re-evacuation accompanied by cooling of the disc. Adsorbate adsorption was carried out in the range of the ambient temperature of the spectrometer beam (Tab z 320 K) to 760 K.Specific surface-area measurements using the B.E.T. method by adsorption of nitrogen at 77 K gave values of 270 m2 g-' after room-temperature evacuation reducing to 215 m2 g-l after 17 h at 483 K for the tin(rv) oxide-silica sample, and 190 m2 g-' reducing to 99 m2 g-l after 16 h at 423 K for the tin(rv) oxide-palladium oxide sample. Infrared spectra were recorded using a Perkin-Elmer 577 spectrometer. RESULTS AND DISCUSSION SURFACE DEHYDRATION AND DEHYDROXYLATION Results of a thermogravimetric analysis (t.g.a.) of the tin(1v) oxide-silica sample are shown in fig. 1. The slight loss up to 323 K can be explained by removal of loosely held molecular water.Removal of hydrogen-bonded water can explain the loss at 348 K, while the large loss in weight beginning with evacuation at ca. 480 K can be attributed to the process of condensation of hydroxyl groups. Even after evacuation at 723 K loss of weight was observed, which suggests that the gel was still partially hydroxylated. The infrared spectra (fig. 2) parallel the t.g.a. observations. Prior to evacuation an intense absorption band was present at 1625 cm-l due to the v, bending mode of molecular water. In water vapour this band has been reported at 1595 crn-l,, while for monomeric, dimeric and polymeric water molecules in a nitrogen matrix it has been reported at 1600, 1620 and 1633 cm-l, re~pectively.~ On evacuation at the intensity of this band was greatly reduced and shifted to 1610 cm-l.This is in agreement with the idea of the molecular water being loosely held to the surface. The remaining band at 16 10 cm-l suggests that some water is more firmly held, presumably by hydrogen bonding to surface hydroxyl groups or oxide ions, or held at strongly Lewis-acidic centres. On evacuation, the broad intense v(0H) absorption between 3760 and 2000 cm-l due to molecular water and surface hydroxyl groups decreasedP. G. HARRISON AND B. M. MAUNDERS 1343 80 - 60 E Q) c * +-I .g 40 E c c 20 0 4000 3500 3000 2500 2 000 Fig. 2. Infrared spectra of tin(1v) oxide-silica at (1) Tab and after evacuation at (2) Tab, (3) 373 K for 4 h, (4) 483 K for 3 h, (5) 544 K for 15.5 h, (6) 625 K for 50 h and (7) 723 K for 5 h. Spectra (6) and (7) are displayed to lower wavenumber by 100 cm-l.wavenumber/cm-' 80 60 % Q) u c * .z 4 0 E 2 U 2 0 0 1 4000 3500 3000 2500 2000 wavenum ber/cm-' Fig. 3. Infrared spectra of tin(rv) oxide-palladium oxide at (1) Tab and after evacuation at (2) 373 K for 15.5 h, (3) 488 K for 23 h, (4) 530 K for 24 h, (5) 573 K for 24 h and (6) 723 K for 23 h. in intensity, and a shoulder appeared at ca. 3700 cm-l which may be assigned to the v(0H) of isolated surface Si-OH groups. Raising the evacuation temperature had little effect on the intensity of the broad band until the disc was calcined at 483 K, when a significant reduction occurred. This indicates the removal of surface hydroxyl groups by elimination of water, and agrees well with the increase in weight loss from1344 TIN OXIDE SURFACES * I 4000 3500 3000 2500 2 000 Fig.4. Infrared spectra of tin(1v) oxide-silica exposed to deuterium oxide: (1) sample evacuated at 708 K for 10 h followed by oxygen treatment; (2) after exposure to D,O vapour (1.33 kN m-,) for 14 h at 568 K and evacuation; (3) after further exposure to D,O vapour (1.33 kN m-2) for 48 h at 633 K and evacuation. wavenum ber/cm-' the thermogravimetric results observed around this temperature. Continued evacuation at increasing temperatures further reduced the intensity of the broad band. At the same time the shoulder at ca. 3700 cm-l developed into a fairly sharp band at 3715 cm-l, but with some broadening on the low-wavenumber side. This band, although reduced in intensity, was still present after evacuation at 723 K.The behaviour of tin(1v) oxide and tin(rv) oxide-palladium oxide is very similar, and both exhibit a broad intense band between 3700 and 2000cm-l [v(OH)], a medium-intensity band at ca. 1630 cm-l (v, bending mode of molecular water) and a broad medium-intensity band between ca. 1250 and 1100 cm-l [a(Sn-OH)]. Room-temperature evacuation removed the v, modes at ca. 1630 cm-l and decreased the intensity of the v(0H) band. Little further change in the intensity of this band takes place until heating in vacuo at ca. 483 K, when a significant decrease in intensity occurs because of the condensation of adjacent surface hydroxyl groups. The tin(1v) oxide-palladium oxide surface was still partially hydroxylated after evacuation at 723 K (fig. 3). Evacuation of tin(1v) oxide at 523 K or above produced discs of very poor optical transmittance which varied in colour from dark brown to black with increasing temperatures.Although not restoring the original white colour of the disc, oxygen treatment did restore it partially, the discs becoming pale yellow in colour. Subsequent prolonged evacuation at Tab again resulted in loss of transmittance accompanied by darkening of the disc. Similar discolouration upon thermal treatment was also observed for the tin(1v) oxide-silica and tin(rv) oxide-palladium oxide gels, although for these materials the problem was not serious. SURFACE DEUTERATION Fig. 4 shows the infrared spectra of a tin(rv) oxide-silica disc outgassed at 708 K and then treated with D,O vapour. Hydrogen-deuterium exchange was rapid but notP.G . HARRISON AND B. M. MAUNDERS 1345 - I 1700 1500 1300 11 00 wavenumberlcm-' Fig. 5. Infrared spectra of ammonia adsorbed on tin(rv) oxide: (1) starting surface evacuated (320 K, 24 h); (2) duringexposure to ammonia (1.33 kN m-2, 320 K); (3) subsequent desorption (320 K, 15 min). complete until treatment with the vapour at 633 K. A sharp absorption band at 2740 cm-l can be assigned to the 0-D stretching vibration of isolated SOD groups, while the broad band centred at 2525 cm-l is due to the deuteroxyl groups perturbed by hydrogen bonding. The observed shift for the isolated 0-D stretching vibration (v,/v, = 1.36) is in very good agreement with the calculated shift ( v , / v , = 1.37). Essentially, complete H-D exchange occurs with tin(rv) oxide-palladium oxide at 523 K, to give a broad band centred at ca.2400 cm-l attributable to the hydrogen- bonded deuteroxyl stretching vibrations. A slight shoulder was present at 2640 cm-l which can be ascribed to isolated SnOD groups. Again the observed shift in the isolated 0-D stretching vibration (v,/v, = 1.38) is in very good agreement with the expected shift. The hydroxyl deformation mode would be expected at ca. 870 cm-l; however, this was not observed owing to bulk oxide adsorptions. SURFACE ACIDITY: ADSORPTION OF AMMONIA Fig. 5 shows the infrared spectra of a tin@) oxide disc evacuated (< 1.33 x N m-2) at Tab for 24 h and subsequently exposed to ammonia vapour (1.33 kN m-2) at Tab. Three new absorption bands appeared in the presence of the ammonia vapour at 1620, 1470 and 1240-1250 cm-l.The 1470 cm-l band could be removed by evacuation for a few minutes at qb but was restored on exposure to water vapour ; re-evacuation again removed this band. Evacuation also decreased the intensity of the 1620 and 1240-1250 cm-l bands, with small shifts to 1615 and1346 TIN OXIDE SURFACES 71 - 74 E s 77 5 2 65 e, .- * - 10 - 1700 1500 1300 1100 wavenumberlcm-' Fig. 6. Infrared spectra of ammonia adsorbed on tin@) oxide during exposure to ammonia (1.33 kN mW2, 320 K) on discs heated under vacuum at (1) 320, (2) 369, (3) 413 and (4) 495 K. 1235 cm-l, respectively. The effect of the presence of water vapour was difficult to judge, the water-bending mode being in the same region as the 1615 cm-l band.However, subsequent evacuation left both the 1615 and 1235 cm-l bands slightly reduced in intensity. The broad, intense hydroxyl stretching band prevented the observation of any N-H stretching vibrations. Increasing the thermal pretreatment temperature of the tin(1v) oxide disc had two effects on the spectra. First, the intensity of the 1470 cm-l band decreased with increasing evacuation temperature and was completely removed after evacuation at 495 K (fig. 6). Secondly, the hydroxyl stretching band, which exhibits a maximum at 3200 cm-l after evacuation at 495 K, shifts to lower frequency by ca. 150 cm-l. Subsequent evacuation restored the band to its original position with a slightly increased intensity. Attempts to study this at higher pretreatment temperatures were unsuccessful due to a severe loss in transmittance of the disc upon admitting ammonia vapour.On evacuation of ammonia vapour from the cell the intensities of both the 161 5 and 1235 cm-l bands decreased with increasing evacuation temperature. A tin(1v) oxide disc was heated under vacuum at 367 K, treated with ammonia vapour (4.0 kN m-2, Tab), evacuated and subsequently exposed to D20 vapour (1.33 kN mP2, Tab) (fig. 7). The 1615 and 1245 cm-l bands were greatly reduced in intensity and new bands were observed at 1435, 1205, 1140, 1075,980 and 900 cm-l. Subsequent evacuation had little effect on the spectrum apart from the followingP. G. HARRISON AND B. M. MAUNDERS 1347 - I 1700 1500 1300 1100 900 wavenumber/cm -' Fig. 7. Infrared spectra of a tin(1v) oxide disc heated under vacuum at 369 K and then given the following treatments: (1) exposed to ammonia, followed by evacuation at 320 K; (2) during exposure to D,O vapour (1.33 kN m-,, 320 K); (3) subsequent evacuation (320 K, 10 min).1800 1600 1400 1200 1000 wavenumber/cm -' Fig. 8. Infrared spectra of tin(1v) oxide: (1) starting surface of disc evacuated (485 K, 17 h); (2) during exposure to hydrogen chloride (1.33 kN m-,, 320 K); (3) subsequent evacuation (320 K, 2 h); (4) during exposure to ammonia (1.33 kN m-,, 320 K); (5) subsequent evacuation.1348 TIN OXIDE SURFACES changes: the 1245 cm-l band became more distinct and increased in intensity, the 1205 cm-l band was reduced in intensity and shifted to 1 180 cm-l, the 1075 cm-l band shifted to 1035 cm-l and the 980 cm-l band was removed. In order to characterise bands due to the adsorbed ammonium cation, a tin(1v) oxide disc was protonated by treatment with hydrogen chloride vapour and then exposed to ammonia vapour (fig.8). In the presence of hydrogen chloride three new bands appeared in the spectrum at 1620, 1410 and 1100 cm-l. Evacuation removed the 1410 cm-l band and reduced the intensity of the 1620 cm-l band, which shifted to 1605 cm-l. Subsequent exposure to ammonia vapour produced strong absorption bands at 1620, 1410 and 1290 cm-l, a broad shallow band at ca. 1750 cm-l and a shoulder on the high-wavenumber side of the 1410 cm-l band. Pumping off the vapour phase reduced the intensity of all the bands, removing the 1750 cm-l band and the shoulder on the 1410 cm-l band completely; the 1620 cm-l band shifted to 1605 cm-l while the 1290 cm-l band shifted to 1270 cm-l.The 1615 and 1235 cm-l bands of adsorbed ammonia can be assigned to the v4(E) or 6,, and the v2(Al) or 6, modes of coordinatively bonded ammonia molecules, re~pectively.~*~ The decrease in intensity and shifts of the two ba1:ds from 1620 and 1240-1250 cm-l in the presence of the vapour to 1615 and 1235 cm-l in its absence suggest that there are two adsorbed ammonia species: one held on the surface by weak hydrogen bonding, and readily removed upon evacuation, the other being more strongly held, either coordinated or hydrogen-bonded to the surface. The position of the v, mode is consistent with the ammonia being coordinated to Lewis-acid tin sites,5 but the decrease in intensity of the bands with increasing evacuation temperature suggests that the ammonia is hydrogen-bonded to surface hydroxyl groups.The 1470 cm-l band is in the correct position for the bending mode of the ammonium ion (NHZ);6 its disappearance on evacuation and reappearance in the presence of water vapour suggests that it results from an interaction between water, or hydroxyl ions, and ammonia. However, on the protonated tin(1v) oxide surface the v4(&) or 6,, mode of the ammonium cation occurs at 1410 cm-l. The difference of 60 cm-l can be attributed to the strength of the Bronsted-acid site and the electrostatic interaction or ammonium ions with the surface. The energy required for the bending vibration of the 1470cm-l band is greater than that for the 1410cm-l band; therefore the interaction of the ammonium ions with the surface is less.This is as expected, since H,O should act as a weak acid, giving NH,+ and -OH ions adsorbed on the surface, while HCl should act as a strong acid, giving the NH: and C1- ions. A stronger electrostatic interaction would therefore be expected between the cation and the chloride ion adsorbed on the surface. Hydrogen bonding of ammonia to surface hydroxyl groups can be seen from the shift in position of the hydroxyl stretching band. The band position is restored on evacuation, which suggests that the remaining ammonia is coordinatively bonded to unsaturated tin ions. No N-H stretches due to coordinated ammonia or amide groups could be discerned against the intense hydroxyl stretching band.Rapid H-D exchange occurs on admitting deuterium oxide to an ammonia-treated tin(rv) oxide disc. The observed infrared bands can all be explained in terms of coordinated or hydrogen-bonded NH,, NH,D, NHD, and ND, species (table 1). In the presence of the D,O vapour the main species was adsorbed ND,. However, on evacuation the intensities of bands related to ammonia species with one or more hydrogens present were increased. The band shifts for deuterated ammonia agreed well with observed shifts on cobalt monoxide and calcium oxide;5 for NH,D v(NH,)/v(NH,D) M 1.08, for NHD, v(NH,)/v(NHD,) M 1.19, for ND, v(NH,)/v(ND,) z 1.34. A band was observed at ca. 2400 cm-l that could be due toP. G. HARRISON AND B. M. MAUNDERS Table 1. Observed vibrational bending modes of NH, D3-,(x = 0,1,2,3) species adsorbed on tin(rv) oxide band position/cm-' in the presence of D20 vapour after evacuation assignment 1600 (sh) 1435 (br) 1250 (vw) 1205 (s) 1 140 (vw) 1075 (w) 980 (sh) 900 (vs) 1349 the N-D symmetric stretching mode of ND, hydrogen-bonded to surface oxide ions.However, no other bands due to N-D stretches were discernable against the strong OD absorption band. The 1605 and 1 100 cm-l bands observed on a tin(rv) oxide disc exposed to hydrogen chloride vapour and subsequently evacuated can be assigned to the v, and v2 bending modes, respectively, of the hydroxonium cation. They occur in similar positions to the bending modes of the isoelectronic and isostructural ammonia molecule. Bands for the hydroxonium ion in strong mineral acids have been reported in similar p0sitions.~9 The nature of the 1410 cm-l band, observed prior to evacuation, is more difficult to assign.It occurs in the same position as the v, bending mode of the ammonium cation. However, as it disappears on evacuation it is unlikely to be due to traces of contaminating ammonia. One possibility for its cause is a doubly protonated water molecule involving both lone pairs of the oxygen. However, this is very unlikely, the H402+ ion having only been considered theoretically, whilst in practice it is unlikely to be stable. A more likely explanation would be the formation of an H,OZ, H7O; or H90: species, all of which are known to exist.9 The formation of the 1410 cm-l band on subsequent exposure to ammonia vapour is due to the v4(F2) bending mode of the ammonium cation, NHZ.The Bronsted acid responsible for its formation is the hydroxonium ion, since the 1 100 cm-l band disappears. The shoulder on the 1410cm-l band is probably due to the same species postulated for the 1470 cm-1 band on the untreated disc. Evacuation removed the shoulder and shifted the 1620 and 1290 cm-l bands to 1610 and 1260 cm-l, respectively. These two bands are due to the v4(E) and v2(Al) bending modes, respectively, of the coordinated or hydrogen-bonded ammonia. The 1750 cm-l is also removed by evacuation. A band has been observed in this region for ammonium halideslO-l= (other than phase I NHJ), where it has been attributed to the ~4(F2) + v, (rotary lattice vibration) combination band of the ammonium cation, when the ammonium ion is unable to freely rotate.No one simple adsorbed NH, structure can explain the results. On the untreated disc the ammonia may be hydrogen-bonded to surface oxide or hydroxyl groups or, more probably, coordinated to unsaturated tin sites. Similar bands are observed on the protonated surface, the main difference for adsorbed NH, groups being the position of the S, band, 1235 cm-l on the untreated and 1270 cm-l on the protonated1350 TIN OXIDE SURFACES -1800 1600 1400 1200 wavenum berlcm-’ Fig. 9. Infrared spectra of ammonia adsorbed on tin@) oxide-palladium oxide: (1) starting surface evacuated (320 K, 16 h); (2) during exposure to ammonia (2.6 kN m-2, 320 K); (3) subsequent evacuation (320 K, 1 h); (4) evacuation (365 K, 40 min); (5) evacuation (500 K, 2.5 h).surface. This can be explained as being due to coordinated ammonia molecules also involved in hydrogen bonding, uiz. H 1 H t Sn Sn Sn Sn \ /sn\ /s”\ untreated surface protonated surface \o/ Lo/ \() 0’ 0 0 0 Weak Bronsted-acid sites can be postulated on the untreated oxide surface from the appearance of the 1470 cm-l band in the presence of water vapour. Protonation of the surface causes strong Bronsted-acid sites. The infrared spectra of a tin(rv) oxide-palladium oxide disc evacuated at room temperature then treated with ammonia vapour are shown in fig. 9. Three new bands appeared in the presence of the vapour at 1612, 1470 (shoulder) and 1230 cm-l, the shoulder at 1470 cm-l being removed on evacuating the cell.Exposure to water vapourP. G. HARRISON AND B. M. MAUNDERS 1351 I I I I I I I 1800 1600 1400 1200 wavenumber/cm -' Fig. 10. Infrared spectra of tin(1v) oxide-palladium oxide exposed to hydrogen chloride and then ammonia: (1) starting surface of disc evacuated (320 K, 40 h); (2) during exposure to hydrogen chloride (9.31 kN m-2, 320 K); (3) subsequent evacuation (320 K, 1 h); (4) during exposure to ammonia (9.31 kN m-2, 320 K); (5) subsequent evacuation (320 K, 1 h). caused the 1612 and 1230 cm-l bands to shift to 1620 and 1250 cm-l, respectively, and restored a band at 1450-1460 cm-l. Subsequent evacuation at 500 K reduced in intensity the 1612 and 1230 cm-1 bands, now shifted slightly to 1610 and 1235 cm-l, and produced a new band at 1510 cm-l. The strong hydroxyl stretching vibrations prevented the observation of any N-H stretching modes.When a tin(1v) oxide-palladium oxide disc was heated in vacuu and treated with oxygen at 563 K, the intensity of the hydroxyl stretching band increased and shifted slightly to lower frequency in the presence of the vapour, and new bands were observed at 1615 and 1225 cm-l. Evacuaticn of the cell did not restore the OH band to its original intensity but reduced in intensity the 1615 and 1225 cm-l bands, shifted to 1605 and 1230 cm-l, and a very weak band was discernible at ca. 535 cm-l. Evacuation at 393 K removed the 1605, 1535 and 1230 cm-l bands. A tin(rv) oxide-palladium oxide disc exposed to hydrogen chloride vapour at the ambient beam temperature exhibited three new bands at 1615, 1405 and 1100 cm-l (fig.10). Pumping off the vapour phase removed the 1405 cm-l band and decreased the intensity of the 1615 and 1100 cm-l bands, moving them to 1580 and 1090 cm-l, respectively. Subsequent exposure to ammonia vapour produced bands at 1730 (broad1352 TIN OXIDE SURFACES I I 1 I I 1800 1600 1400 1200 wavenum ber/cm -' Fig. 11. Infrared spectra of ammonia adsorbed on tin(1v) oxide-silica: (1) starting surface of tin@) oxide-silica evacuated at 320 K; (2) during adsorption of ammonia (1.33 kN mU2, 320 K); (3) subsequent evacuation (320 K, 24 h); (4) disc heated under vacuum at 483 K, during exposure to ammonia (1.33 kN mP2, 320 K); (5) disc heated under vacuum at 693 K, during exposure to ammonia (2.6 kN m-2, 320 K); (6) after evacuation of (5) and during exposure to water vapour (1.33 kN m-2, 320 K).and weak), 1615 (strong), 1410 (very strong) and 1290cm-1 (strong) but removed the 1090 cm-I band. A shoulder on the high-frequency side of the 1410 cm-l band was also observed. Evacuating the cell removed the 1730 cm-l band and the shoulder on the 1410cm-l band, and the 1615, 1410 and 1290cm-l bands were reduced in intensity and shifted to 1605, 1405 and 1270 cm-l, respectively. The 1090 cm-I band was restored but not to its original intensity. The behaviour of tin(1v) oxide-palladium oxide toward ammonia and hydrogen chloride vapour is very similar to the behaviour of tin(1v) oxide. As such, the 1612 and 1230 cm-l bands, produced on adsorption of ammonia, can be assigned to the v4(E) and v2(A,) bending modes, respectively, of coordinatively bonded, or hydrogen- bonded, ammonia molecules, while the shoulder at 1470 cm-' can be ascribed to the bending mode of an ammonium species, formed from weak Bronsted-acid sites.The formation of the band at 1510 cm-l, and decrease in the 1612 and 1230 cm-l bands, on evacuation at 500 K can be ascribed to the formation of surface amido groups from the adsorbed ammonia. Bending vibrations of surface amido groups have been variously reported between 1470 and 1570 cm-l on different aluminas, silicas,P. G. HARRISON AND B. M. MAUNDERS 1353 4 000 3500 3000 2500 wavenumber/cm-' Fig. 12. Effect of ammonia adsorption of the hydroxyl stretching band of tin(rv) oxide-silica heated under vacuum at 693 K: (1) starting surface; (2) during exposure to ammonia vapour (2.6 kN m-2, 320 K); (3) subsequent evacuation (320 K, 4.5 h).aluminium phosphate and germania gels.5 In conjuction with the decrease in the 16 12 and 1230 cm-l band intensities, slight shifts in position to 1610 and 1235 cm-l occurred, thus indicating that there were two forms of adsorbed ammonia present, one of which reacted to form the amide. A possible precursor could be a coordinatively held ammonia molecule involved in hydrogen bonding to surface oxide ions. The intense nature of the hydroxyl stretching band made it impossible to judge whether or not there was a concomitant increase in its intensity with the appearance of the 1510 cm-l band. However, a definite increase in intensity was observed when a disc heated and oxygen treated at 563 K was exposed to ammonia vapour.The 1615 and 1100 cm-l bands, shifting to 1580 and 1090 cm-l on evacuation, observed on exposure of a tin(rv) oxide-palladium oxide disc to hydrogen chloride vapour, can be assigned to the vq and v2 bending modes, respectively, of the hydroxonium cation. The 1 730,16 15,14 10 + shoulder and 1290 cm-l bands observed on subsequent exposures to ammonia vapour and their behaviour on evacuation are analogous to the bands observed on a similarly treated tin@) oxide disc; accordingly the band assignments are the same. The infrared spectra of a tin@) oxide-silica disc evacuated and exposed to ammonia vapour at Tab exhibited three new absorption bands at 1620, 1455 and 1245 cm-l (fig. 11). On evacuation the 1455 cm-l band was reduced in intensity but not removed.Evacuation had little affect on the other bands. Three bands at 1620, 1465 and 1240 cm-l were observed on exposure to ammonia of a disc heated under vacuum at 483 K, although the intensity of the 1465 cm-l band was considerably less 45 FAR 11354 TIN OXIDE SURFACES 1800 1600 1400 1200 wavenum ber/cm Fig. 13. Infrared spectra of hydrogen chloride then ammonia vapour absorption on a tin(1v) oxide-silica disc heated under vacuum at 693 K: (1) starting surface; (2) during adsorption of hydrogen chloride (2.0 kN m-2, 320 K); (3) subsequent evacuation (320 K, 1 h); (4) during adsorption of ammonia (2.66 kN m-2, 320 K); (5) subsequent evacuation (320 K, 15 h). than that of the unheated disc under similar conditions of ammonia vapour.After evacuation at 693 K, the 1465 cm-l band was very weak and shifted to ca. 1490 cm-l. The bands at 1620 and 1245 cm-l were also weaker and shifted to 1615 and 1240 cm-l, respectively. Exposure to water vapour produced a fairly intense band at 1445 cm-l which was removed on subsequent evacuation. The formation of a broad intense band was also observed, with its maximum at 3200 cm-l, on the tin(xv) oxide-silica disc heated under vacuum at 693 K (fig. 12). A tin@) oxide-silica disc heated under vacuum at 693 K and exposed to hydrogen chloride vapour exhibited two weak bands at 1620 and 1410 cm-l (fig. 13). Subsequent exposure to ammonia vapour produced absorption bands at 1750 (broad and shallow), 1610 (medium), 1410 (very strong) and 1280 cm-l (strong). The intensity of the 16 10,141 0 and 1280 cm-l bands were reduced by evacuation and the 1750 cm-l band removed.The 1620 and 1245 cm-l bands can be attributed to the v4(E) and v2(A1) bending modes, respectively, of coordinatively bonded or hydrogen-bonded ammonia mol- ecules, while the 1455 cm-l band can be ascribed to the ~4(F2) bending mode of the ammonium cation. Such a band has been observed in this region on silica-alumina mixed-oxide systems.69 l3 Unlike the other two oxides the 1455 cm-l band is notP. G. HARRISON AND B. M. MAUNDERS 1355 removed by evacuation. The effect of increased pretreatment temperature is to reduce considerably the number of Bronsted-acid sites, as observed from the weaker band intensity. The shift of the v4 band intensity to higher frequency suggests a weaker interaction between the NH,+ ion and the surface at the higher evacuation temperatures.Bronsted-acid sites could be reformed by exposure to water vapour. The 1410 cm-l band observed after ammonia adsorption on the disc treated with hydrogen chloride and heated to 693 K can be assigned to the v,(F,) bending mode of the ammonium cation, involved in a stronger interaction with the surface than the ammonium cation produced on the untreated oxide. The 3200cm-I band can be assigned to the N-H stretching vibrations of the adsorbed ammonia. The broad nature of the band made identification of the nature of the N-H stretches difficult, although the band is at too low a frequency for it to be due to amido N-H stretches. No evidence was seen for amido groups in the N-H bending region.DISCUSSION Infrared spectroscopy of ammonia adsorption on metal oxides has been employed as a method of studying the nature of acidic sites on oxide^.^^ 6~ 14-31 Pure oxides usually exhibit only Lewis l4? 2 o v 2 3 y 25 although this is not always the 2 4 9 29 while mixed oxides exhibit Bronsted acidity.l59 2 4 9 26* 27 The infrared absorption bands of ammonia adsorbed on tin(1v) oxide at 1615 and 1235 cm-l clearly indicate that ammonia is coordinated to Lewis-acidic sites on the tin(1v) oxide surface, i.e. to surface tin sites. Similar bands have been observed for ammonia adsorption on GeO2,ls Ti0,59 2 o q 21* 31 (both rutile and anatase), A1203,5914922-24731 Zn0,29931 Ni0,5731 Mg0,5931 Zr0,,5931 Be0,5931 Ga,O,,,l Ta,05,31 Cr,O, and Fe20,.6* 25 The shift to higher frequency of the 6,(NH,) deformation mode (at 950 cm-I in free ammonia) on coordination has been employed as a guide to the Lewis acidity of metal centres in these oxides.The observed order [6,(NH,) mode in parentheses], Al,O, (1 3 15-1283 cm-l), Al,O, * SiO, (1 3 10-1280 cm-l) > Ga,O, (1275 cm-l) > TiO, (1225 cm-l) > ZnO (1 220 cm-l), Be0 (1220 cm-l) > Cr,O, (1 220 cm-l) > ZrO, (1 200- 1 160 cm-l) > NiO (1 1 80-1 1 30 cm-l) > MgO (1 170 cm-l) > Ni,O, (1 145 cm-l) > COO (1 130 cm-l), therefore ranks the Lewis acidity of the bare tin sites of tin(rv) oxide between those of Ga,O, and TiO,. Analogous bands observed for ammonia on tin(1v) oxide-palladium oxide (1 605-1 6 10 and 1230-1235 cm-l) and tin(1v) oxide-silica (1620 and 1240-1245 cm-l) indicate that the strengths of the Lewis-acid sites are slightly less and slightly stronger, respectively, than those on the pure oxide.Weak Bronsted-acid sites are observed for all three oxides. An important difference, however, is the appearance of the band at 1510-1 535 cm-l, together with an increase in intensity of the hydroxyl stretching band, after evacuation at 500 K, which is attributed to dissociative chemisorption giving rise to surface amido groups on palladium sites similar to those found on palladium metal.30 Several other metals also chemisorb ammonia to give surface amido 22* 23* 32 including silica, which does not exhibit any Lewis acidity towards amrnonia.l7-l9 CONCLUSIONS Tin(rv) oxide is predominately Lewis acidic, with weak Bronsted acidity evolving in the presence of water vapour.Strong Bronsted-acid sites can be produced by protonation with hydrogen chloride. 45-21356 TIN OXIDE SURFACES The nature of tin@) oxide-palladium oxide is quite similar to tin(1v) oxide, exhibiting predominately Lewis acidity, although weak Bronsted acidity can be observed in the presence of water-vapour. Strong Bronsted-acid sites can be produced by treating the surface with hydrogen chloride vapour. The main difference is that surface amido groups are formed on tin(1v) oxide-palladium oxide after evacuation at 500 K, whereas no bands ascribable to surface amido groups were observed on tin(rv) oxide under any conditions. It seems reasonable, therefore, to assume the amide is forming on palladium sites. Unlike tin@) oxide or tin(1v) oxide-palladium oxide, tin@) oxide-silica exhibits both Lewis and Bronsted acidity, the amount of Bronsted acidity with respect to Lewis acidity decreasing with increasing evacuation temperature.P. G. Harrison and E. W. Thornton, J. Chem. Soc., Faraday Trans. 1, 1976, 72, 1310; 1317; 2484; and references cited therein. P. G. Harrison and E. W. Thornton, J. Chem. SOC., Faraday Trans. I , 1978, 74, 2703. M. Van Thiel, E. D. Becker and G. C. Pimentel, J. Chem. Phys., 1957, 27, 486. K. Nakamoto, Infrared Spectra of Indrganic and Coordination Compounds (Wiley, London, 1970). A. A. Tsyganenko, D. V. Pozdnyakov and V. N. Filimonov, J. Mol. Struct., 1975, 29, 299. M. L. Hair, Infrared Spectroscopy in Surface Chemistry (Arnold, London, 1967). ' D. E. Bethel1 and N. Sheppard, J. Chem. Phys., 1953, 21, 1421. C. C. Fenso and D. F. Hornig, J. Chem. Phys., 1955, 23, 1464. I. Olovsson, J. Chem. Phys., 1968, 49, 1063. lo E. L. Wagner and D. F. Hornig, J. Chem. Phys., 1950, 18, 296; 305. l1 R. C. Plumb and D. F. Hornig, J. Chem. Phys., 1953, 21, 366. l2 R. C. Plumb and D. F. Hornig, J. Chem. Phys., 1955, 23, 947. l3 L. H. Little, Infrared Spectra of Adsorbed Molecules (Butterworths, London, 1966). l4 R. P. Eischens and W. A. Pliskin, Adv. Catal., 1958, 10, 1. l5 J. B. Pen, Discuss. Faraday Soc., 1971, 52, 55. l6 M. J. D. Low and K. Matsushita, J. Phys. Chem., 1969, 73, 908. l8 N. W. Cant and L. H. Little, Can. J. Chem., 1965, 73, 1252. IQ B. A. Morrow and I. A. Cody, J. Phys. Chem., 1976,80, 1098. 2o G. D. Parfitt, J. Ramsbottom and C. H. Rochester, Trans. Faraday Soc., 1971,67, 841. 21 M. Herrmann and H. P. Boehm, 2. Anorg. Allg. Chem., 1969, 73, 1368. 22 A. A. Tsyganenko, D. V. Pozdnyakov and V. N. Filimonov, Usp. Fotoniki, 1975,5, 150. 23 J. B. Pen, J. Phys. Chem., 1965, 69, 231. 24 A. A. Davydov and Yu. M. Shchekochikhin, Kinet. Catal., 1969, 10, 523. 25 G. Blyholder and E. A. Richardson, J. Phys. Chem., 1962, 66, 2599. 26 J. E. Mapes and R. P. Eischens, J. Phys. Chem., 1954,58, 1059. 27 N. W. Cant and L. H. Little, Nature (London), 1966, 69, 21 1. 28 M. M. Mortland, J. J. Fripiat, J. Chaessidon and J. Uytterhoeven, J. Phys. Chem., 1963, 67, 248. T. Morimoto, H. Yani and M. Nagao, J. Phys. Chem., 1976,80,471. 30 D. V. Pozdnyakov and V. N. Filimonov, Kinet. Catal., 1972,13, 522. 31 N. E. Tret'yakov and V. N. Filimonov, Kinet. Katal., 1973, 14, 803. 32 G. M. Zhabrova and E. V. Egorov, Russ. Chem. Rev., 1961,30, 338. J. B. Peri, J. Phys. Chem., 1966, 70, 2937. (PAPER 3/667)