J . Chem. Soc., Faraday Trans. 1, 1985,81, 1345-1355 Tin Oxide Surfaces Part 16.-Infrared Study of the Adsorption of Formic Acid, Acrylic Acid and Acrolein on Tin(rv) Oxide, Tin(1v) Oxide-Silica and Tin@) Oxide-Palladium Oxide BY PHILIP G. HARRISON* AND BARRY MAUNDERS Department of Chemistry University of Nottingham, University Park, Nottingham NG7 2RD Received 1 lth June, 1984 Transmission infrared spectroscopy has been employed to study the surface species formed by adsorption of formic acid, acrolein and acrylic acid on tin(1v) oxide, tin@) oxide-silica and tin(1v) oxide-palladium oxide. The behaviour of the three adsorbents is similar on all three oxides. At low temperatures the two organic acids give surface carboxylate plus coordinated acid, which can be removed by evacuation at higher temperatures.Acrolein is adsorbed as a surface acrylate, although surface-coordinated acrolein is observed at lower temperatures. We have recently shown that ethane, ethene and propene are adsorbed on tin(rv) oxide, tin(1v) oxide-silica and tin(rv) oxide-palladium oxide to give surface acetate species as the ultimate pr0ducts.l In the case of propene on tin(rv) oxide-palladium oxide, a surface acrylate species was also formed. The most probable mechanism for the formation of surface acetate from ethane and ethene is via the oxidation of an initially formed surface-ethoxide species, which appears to be formed by hydrogen abstraction by surface oxide in the case of ethane and by the formal addition of surface hydroxy groups across the carbon+arbon double bond in the case of ethene.The surface ethoxide species was then postulated to undergo oxidation to afford the observed surface carboxylate, by analogy with our previous observation that surface alkoxide groups on tin(1v) oxide undergo facile oxidation to the corresponding carboxylate at temperatures as low as the ambient beam temperature of the spectrometer.2 Furthermore, the probable intermediacy of a surface-coordinated carbonyl species in these reactions gains considerable support from the known chemisorption behaviour of ketones of the type MeCOR (R = Me, Et, Pr", But or Ph) on tin(1v) oxide to give the surface carboxylate, 02CRads.29 Propene adsorption presents a more complicated case, since two initial surface alkoxide species, an isopropoxide (by hydroxide addition to the C=C bond) and an allyloxide (by hydrogen abstraction), may be generated, and also since dissociative chemisorption with C-C bond fission to give both surface acetate and formate, as has been observed on zinc oxide,* may also be possible.In order to resolve some of the possible ambiguities in the adsorption of propene on materials containing tin(1v) oxide (i.e. the possible occurrence of surface formate) and also to corroborate the formation of surface acrylate via the oxidation of coordinated acrolein, we here report the details of an infrared study of the adsorption of formic acid, acrolein and acrylic acid. 13451346 TIN OXIDE SURFACES Table 1. Infrared data for acrylic acid adsorption assignment position of absorption bands/cm-' coordinated SnO, - PdO SnO, SnO, * SiO, acrylate acrylic acid - 1630 1580 15 10-1 520 1410 1325 1270 1160 980 890 820 1425-1430 1355-1365 1245-1250 1050-1065 - 1 630-1 635 1500-1 530 1408-1 4 10 1365 1270 1170 1065 980 890 820 1585-1 590 1 430- 1 43 5 1328-1 330 1 250- 1 25 5 1655 1630 1580-1 590 1500-1 5 10 1430 1410 1325 1270-1273 1365-1368 1245-1250 - C-H head - v(C-C)? P W , ) S(C00) CH, twist CH bend v(C=C) or v(C-0) vc=o - - - CH, scissors SOH v(C-0) or S(C-H) - - v(C-C) - Table 2.Infrared data for acrolein adsorption assignment position of absorption bands/cm-l coordinated SnO, PdO SnO, SnO, . SiO, acrylate acrolein 1710-1 720 1655 1630 - 1 5 1 0- 1 520 1425- 1428 1365 1270 - 1150-1 170 1055-1060 980 820 1710 1630 1 5 10-1 520 1675-1680 - 1428- 1430 1405- 141 0 1365 1270 1250 1170 1060 980 - 820-825 1700- 17 10 1655 1630 1570-1590 1510-1515 1428-1432 1405- 14 10 1370 1310 1270-1 273 1250 - v(C=C) vas (C0O)U v, (COO) - S(C-H) d(C-H)" - - v(C-C)? PCH, CH, twist CH bend v(C=O) v(C=O) - CH, scissors - - CH bend v(C-C) Due to coordinated acrylic acid; see text for details.EXPERIMENTAL The preparation of tin(1v) oxide, tin(rv) oxide-silica and tin(1v) oxide-palladium, the manufacture of infrared-transmitted discs therefrom and the general techniques employed have been described previ~usly.~ Infrared spectra were recorded using a Perkin-Elmer 577 spectrometer. Spectral data are summarised in tables 1 and 2.P. G. HARRISON AND B. MAUNDERS 1347 I I I 1 1 I I 1800 1600 1400 1200 wavenumber/cm -' Fig. 1. Infrared spectra of tin(1v) oxide: (1) evacuated 320 K, 16 h, < 1.33 x N m-2, (2) exposed to formic acid vapour, 320 K, 5 min, 13.3 N m-2; subsequently evacuated (3) 320 K, 17 h, < 1.33 x N m-2, (4) 438 K, 3 h, < 1.33 x N rn+ and (5) 476 K, 18 h, < 1.33 x N m-2. RESULTS FORMIC ACID ADSORPTION Exposure of an evacuated tin(1v) oxide disc (320 K, < Torr) to formic acid vapour (1 3.3 N mP2, 320 K) gave very strong infrared absorption bands at 1630, 1600, 1550, 1380,1345 and 1260 cm-l (fig.1). Evacuation (320 K, 17 h) reduced the intensity of the 1630 and 1260 cm-1 bands with respect to the other bands; in addition, it could then be seen that the 1260 cm-1 band was composed of the original band plus a second band at 1285 cm-l and the 1345 cm-l band shifted to 1350 cm-l.Evacuation at 438 K (3 h, < z) reduced still further the 1630 and 1260 cm-l bands, whilst evacuation at 476 K reduced the intensity of all the bands leaving absorptions at 1548, 1382 and 1340 cm-l. Evacuation at 514 K caused the decomposition of the formate species. N mP2, 320 K) to formic acid vapour (13.3 N m-2, 320 K) led to the formation of very strong adsorption bands at 1600,1555,1370,1335 and 1280 cm-l (fig. 2). Evacuation (320 K, 1 h) removed the 1600 cm-l band, leaving a very weak band at 1290 cm-l and three strong bands at 1545, 1380 and 1340 cm-l. Exposure of an evacuated tin(rv) oxide-palladium oxide disc (1.33 x1348 TIN OXIDE SURFACES h E g 39 cd Y .- E 2 56 2 7 4 8 4 . Y - 10 O/O - 1 1 I I I I 1800 1600 1400 1200 wavenumber/cm-' Fig.2. Infrared spectra of tin(1v) oxide-palladium oxide: (1) evacuated 320 K, 96 h, < 1.33 x lop4 N m-2, (2) exposed to formic acid vapour 320 K, 5 min, 13.3 N rnP2 ; subsequently evacuated (3) 320 K, 1 h, < 1.33 x N m-2 and (4) 443 K, 3 h, < 1.33 x N m-2. Exposure of an evacuated tin(1v) oxide-silica disc (1.33 x lo-* N m-2, 320 K) to formic acid vapour (1 3.3 N m-2, 320 K) led to the formation of very strong absorption bands at 1640, 1600-1540, 1375, 1350 and 1260 cm-l (fig. 3). These bands can also be assigned to coordinated formic acid, 1640 and 1260 cm-l, and surface formate groups, 1600-1540, 1375 and 1350 cm-l. Evacuation (320 K, 16 h) left bands at 1580-1530, 1380, 1350 and 1280 cm-l, which agree well with similar bands seen on tin(1v) oxide.Evacuation at 443 K greatly reduced the width of the 1580-1530 cm-l band, exposing a weak broad band centred at 1700 cm-l, which can be assigned to the antisymmetric COO stretch of an unidentate formate, most probably on a silica site. A further weak band was observed at 1815 cm-l, which can be assigned to undissociated formic acid adsorped on a silicon site, since formic acid is known to adsorb undissociated on silica itself.6 This band was removed by evacuation at 470 K. Evacuation at 520 K again decomposed the formate structure. ACRYLIC ACID ADSORPTION Exposure of an evacuated tin(1v) oxide-palladium oxide disc (1.33 x lop4 N m-2, 320 K) to acrylic acid vapour (13.3 N mp2, 320 K) led to an infrared spectrum exhibiting adsorption bands at 1630, 1580, 1520, 1425, 1410, 1325, 1245, 1130, 1050, 980, 890 and 820 cm-l, along with shoulders at ca.1760, 1705 and 1355 cm-l (fig.P. G . HARRISON AND B. MAUNDERS 1349 1900 1700 1500 1300 wavenum berlcm-' Fig. 3. Infrared spectra of tin(rv) oxide-silica: (1) evacuated 320 K, 24 h, < 1.33 x N m-2, (2) exposed to formic acid vapour, 320 K, 5 min, 13.3 N m-2; subsequently evacuated (3) 320 K, 16 h, < 1 . 3 3 ~ 10-4Nm-2, (4) 443 K, 19 h, < 1 . 3 3 ~ 10-4Nm-2 and (5) 470K, 4h, < 1.33 x 1 O - O N m-2. 4). Subsequent evacuation (320 K, 1.5 h) removed the 1760 and 1705 cm-l shoulders, reduced in intensity the 1580, 1410 and 1325 cm-l bands, the 1410 cm-l band now being a shoulder on the 1430 cm-1 band shifted from 1425 cm-l, while the remaining bands became more intense with slight shifts in position (1 520 to 15 10, 1355 shoulder to 1365,1245 to 1250, 1130 to 1160 and 1050 to 1065 cm-l).A shoulder at 1270 cm-l on the 1250 cm-l band was also present. Evacuation up to 400 K had little effect on the spectrum, but evacuation at 463 K removed the 1580, 1410, 1325, 1250 and 1160 cm-l bands, with the appearance of a new band at ca. 1700 cm-l. The removal of the 1250 cm-l band exposed the shoulder at 1270 cm-l as a sharp intense band. At increasing evacuation temperatures up to 597 K, the absorption bands at 1630, 1365, 1270, 1065, 980, 890 and 820 cm-l were steadily reduced in intensity and effec- tively removed, while the two strong absorption bands, at 1510 and 1430 cm-l, were unaltered at 530 K. At 597 K they were slightly reduced in intensity with the 1430 cm-l band shifted to 1395 cm-l.Exposure of evacuated tin(1v) oxide and tin@) oxide-silica discs (1.33 x N m-2, 320 K) to acrylic acid vapour (13.3 N m-2, 320 K) gave infrared spectra which were very similar to those seen for tin@) oxide-palladium oxide (fig. 5 and 6, respectively). Absorption bands were observed on tin@) oxide at 1760 and1350 TIN OXIDE SURFACES 1800 1600 1400 1200 wavenum ber/cm-' Fig. 4. Infrared spectra of tin(rv) oxide-palladium oxide: (1) evacuated 320 K, 48 h, < 1.33 x N m-2, then exposed to acrylic acid vapour, 320 K, 5 min, 13.3 N m-2; subsequent evacuation (2) 320 K, 1.5 h, < 1.33 x N m-2, (4) N m+, (3) 463 K, 17 h, < 1.33 x 597 K, 5.5 h, < 1.33 x lop4 N m-2 and (5) 657 K, 3 h, < 1.33 x N m-2. 1700 cm-l due to weakly held acrylic acid, at 1585-1590, 1408-1410, 1328-1330, 1250-1255 and 1170 cm-l due to acrylic acid coordinatively bonded through the carbonyl oxygen to Lewis-acidic tin sites and at 1630-1635, 1500-1 530, 1430-1435, 1365, 1270, 1065,980,890 and 820 cm-I due to the surface acrylate species. On tin(1v) oxide-silica bands were seen at 1580-1590, 1410, 1325 and 1245-1250 cm-l due to coordinatively bound acrylic acid and at 1630, 1500-1 5 10, 1430, 1365-1 368 and 1270-1273 cm-l due to the surface acrylate species.No bands due to weakly held acid were seen in this case. In addition, a shoulder was observed at 1655 cm-I and it can be assigned to acrylic acid coordinatively bound to a surface site of different strength to the Lewis-acidic tin sites, responsible for the 1580 cm-l band, presumably an unsaturated surface silicon. No absorption bands were observed below 1200 cm-l because of the strong absorptions of the oxide itself.Both oxides showed the appearance of a 1700-1710 cm-l band after evacuation at 456 and 526 K for tin(rv) oxide and tin(1v) oxide-silica, respectively. The absorption bands and assignments for all three oxides are summarised in table 1 .P. G. HARRISON AND B. MAUNDERS 1351 8 6- 9 2. n E W C Y Y .- 5 5 5 65 - 10% - I I I I I I 1800 1600 1400 1200 wavenumber/cm -' Fig. 5. Infrared spectra of tin@) oxide: (1) evacuated 320 K, 3 h, < 1.33 x N m-2, then exposed to acrylic acid vapour, 320 K, 5 min, 13.3 N m-2; subsequent evacuation (2) 320 K, 18 h, < 1.33 x N mP2 and (3) 456 K, 3.5 h, < 1.33 x loP4 N m-2.ACROLEIN ADSORPTION The adsorption of acrolein on tin(1v) oxide-palladium oxide, tin(1v) oxide and tin(rv) oxide-silica gave infrared spectra attributable to surface acrylate species. In addition, weak absorption bands were observed at ca. 1720-1700, 1655-1675, 1405-1410, 1250 and 1150-1 170 cm-l. Detailed discussion and diagrams of spectra are not shown since the main features closely resemble those of acrylic acid absorption. The absorption bands are summarised in table 2. DISCUSSION FORMIC ACID ADSORPTION The adsorption bands observed upon exposure of tin(1v) oxide to formic acid vapour are readily assignable to surface formate and coordinated formic acid. The 1550, 1380 aiid 1345-1350 cm-l bands can be assigned to the v,,(COO), djeP.(CH) and y,(COO) modes, respectively, of a surface formate group, the bands being in good agreement with absorption bands reported for other formate species (see table 3). The 1630 and 1260 cm-1 bands can be assigned to the v(C=O) and v(C-0) bands of adsorbed formic acid, respectively. The large shift of the carboxy stretching band for the ad- sorbed formic acid is evidence for the acid being coordinatively bonded through the1352 TIN OXIDE SURFACES 1800 1600 1400 1200 wavenumber/cm - l Fig. 6. Infrared spectra of tin(1v) oxidesilica: (1) evacuated 320 K, 16 h, .c 1.33 x N m-,, then exposed to acrylic acid vapour, 320 K, 5 min, 13.3 N m-2; subsequent evacuation (2) 320 K, 0.5 h, < 1.33 x lop4 N m-,, (3) 448 K, 3.25 h, < 1.33 x N rn-, and (4) 526 K, 16 h, < 1.33 x lop4 N rn+.Table 3. Absorption band positions for formate species compound band position of formate ion/cm-l ref. ~ ~~~ Sn(OOCH), 1563 1385 1339 7 Sn(OOCH), 1608 1404 1368 7 (CW, Sn(OOCH), 1588 1373 1390 8 CH(CH,), Sn(O0CH) 1595 1325 1373 8 HCOONa 1567 1377 1366 9 assignment va(C00) &i.p.(CH) vs(C00) - - 1370 131 1 carboxy oxygen to Lewis-acid sites, as in structure (I). Hydrogen bonding between the acid and surface hydroxy groups may also occur. The nature of the bonding of the surface formate species may be distinguished by the separation of the symmetric and antisymmetric by surface oxide formate. The formation of both bridging [structure (11)] and chelating [structure (III)] surface formateP. G. HARRISON AND B. MAUNDERS 1353 groups may be readily rationalised by mechansims involving an initial coordination of the carbonyl oxygen atom to a Lewis-acidic tin site, the electron donation having the effect of increasing the negative charge density at a neighbouring surface oxide or hydroxy group, whilst also increasing the group or water molecule can then form (scheme 1).acidity of the acid proton. A hydroxy leaving the surface formate group .c\ 6+ H 0-H H O@ / / H I H c or 2"\ /sz Sn Sn i n i n / \ / \ / \ / \ 0 0 0 0 0 0 0 0 0 (11) H I Sn Sn Sn Sn ' ~ n Sn / \ / + \ / \ / \ / \ / \ 0 0 0 0 0 0 0 0 0 The bands observed for the adsorption of formic acid on tin(1v) oxide-silica and tin(1v) oxide-palladium oxide can be assigned in a similar fashion. All three oxides exhibited a decrease in the intensity of the hydroxy stretching bands upon evacuation of the formic acid vapour at 320K, suggesting that the former mechanism operates.The separation between the p(C=O) and p(C-0) stretching frequencies is 665 cm-l in the formic acid monomer and ca. 550cm-l in unidentate methyl and ethyl formates,1° while a separation of ca. 225 cm-l has been reported for a bidentate formate species adsorbed on ultrastable zeolite.ll In the present case, the difference in wavenumber between the antisymmetric and symmetric COO stretches of 205-210 cm-l suggests that the formate is adsorbed in a bidentate manner. However, a bidentate formate species may be either chelating or bridging, although no distinction between these two possibilities can be made on the infrared data alone.The formation of the formate species can be rationalised by a mechanism which involves the initial coordination of the carbonyl oxygen of the acid to a Lewis-acidic tin site. The electron donation has the effect of increasing the negative charge density 45 F A R 11354 TIN OXIDE SURFACES at a neighbouring oxide or hydroxy group whilst also increasing the acidity of the acid proton, which is abstracted by neighbouring surface oxide or hydroxy groups to give either bridging [structure (111)] (abstraction by surface hydroxyl) or chelating [structure (11)] (abstraction by surface oxide). However, a tin (IV) oxide-silica disc that has been evacuated at 740 K, exposed to formic acid vapour and then subsequently evacuated exhibited an increase in the hydroxy stretching band along with the appearance of the formate structure, which suggests that the second mechanism can occur also.Decomposition of the formate structure occurs above 500 K, which is close to the decomposition temperature of tin(I1) formate (47 1-473 K).12 ACRYLIC ACID ADSORPTION By comparison with the gas-phase spectrum of acrylic acid,13 the two shoulders at 1760 and 1705 cm-l can be assigned to the v(C=O) stretch of weakly held monomeric and dimeric acrylic acid; their easy removal at 320 K suggests only weak hydrogen- bonding interactions with the surface. The bands at 1580, 1410, 1325, 1245-1250 and 1160 cm-l can also be assigned to coordinated acrylic acid since their intensities alter in a similar manner and are removed on evacuation at 463 K.The large shift in the carbonyl stretching frequency, from 1725 to 1580 cm-l, suggests that the acid is coordinatively bonded through the oxygen to Lewis acid sites. The 1410 cm-l band can be ascribed to the CH, scissors mode of the methylene group, the 1325 and 1245-50 cm-’ bands are tentatively assigned to an OH deformation and either the v(C-0) or S(C-H) modes, respectively, while the 1160 cm-l band is best assigned to the y(C-C) stretching mode. The expected y(C=C) stretching mode will occur in approximately the same position as in the acrylate and is not, therefore, seen as a separate band. M M M / \ / \ / \ 0 0 0 0 The bands at 1630, 1510-1520, 1425-1430, 1355-1365, 1275, 1050-1065, 980, 890 and 820 cm-l are in similar positions to bands observed for sodium acrylate13 and can be attributed to the v(C=C), v,,(COO), vs(COO), S(C-H), (C-H) bending, p(CH,), CH, twisting, S(CO0) and (C-H) bending modes, respectively, of a surface-adsorbed acrylate.The 1 160-1 170 cm-l band assigned to the v(C-C) stretch- ing mode of coordinated acrylic acid may also be partly due to the same mode of the adsorbed acrylate. The same situation may exist for the assignment of the 14 10 cm-l band. Decomposition of the acrylate at 657 K caused absorption bands at 1590, 1510, 1440, 1390 and 11 80 cm-l. A possible explanation for these bands could be the formation of a polymeric acrylic acid ~ a 1 t . l ~P. G. HARRISON AND B. MAUNDERS 1355 The weak infrared bands observed at ca. 1700 cm-l on evacuation at 463 K occur with the disappearance of the coordinated acrylic acid and may be due to a carbonyl stretch of a decomposition product from acrylic acid at this temperature.The mechanism for acrylate formation is the same as for formate formation from formic acid. However, the position of the v(C=C) stretching band in all three cases is 5-10 cm-1 lower than in either the free-acid vapour or sodium acrylate,13 indicating that the carbon-carbon double bond is involved in interaction with surface Lewis-acidic sites [structures (IV) and (V)] or surface Bronsted-acid sites [structure (VI)]. ACROLEIN ADSORPTION Infrared spectra obtained from the adsorption of acrolein were very similar to those obtained from acrylic acid adsorption, indicating the very facile oxidation of acrolein by these tin(1v) oxide materials.The 1720-1700, 1405-1410, 1250 and 1150-1 170 cm-l bands can be attributed to the v(C=O) stretching, CH, scissors, CH bending, and p(C-C) stretching modes, respectively, of physisorbed acrolein, while the 1655-1675 cm-l band may be assigned to acrolein coordinated through the carbonyl oxygen to Lewis-acidic sites and be due to the v(C=O) stretching mode.15 The acrolein responsible for the 1720-1700 cm-l band may be physisorbed through the carbon- carbon double bond to Lewis-acidic sites or hydrogen-bonded to surface hydroxy groups. Note that two weak bands at 1570-90 and 1310 cm-l are observed after acrolein adsorption on tin(1v) oxide-silica; these agree well with the positions of the v(C=O) stretching and d(0H) deformation modes, respectively, of coordinated acrylic acid. This observation is not surprising since the tin(rv) oxide-silica is Bronsted acidic and may well protonate the acrylate (scheme 2). H, ,c= H Sn Sn Si / \ / \ / \ 0 0 0 0 Scheme 2. H H >c=c' 0-H H 'C' II 1 I Sn Sn Si / \ / \ / \ 0 0 0 0 0 oo We thank the S.E.R.C. and the International Tin Research Institute for support in the form of a CASE award (to B.M.). P. G. Harrison and B. Maunders, J. Chem. SOC., Faraday Trans. I , 1985,81, 1309; 1327. E. W. Thornton and P. G. Harrison, J. Chem. SOC., Faraday Trans. I , 1975, 71, 2468. P. G. Harrison and B. Maunders, J . Chem. SOC., Faraday Trans. I , 1984,80, 1329. Y. Kubokawa, H. Miyata, T. Ono and S. Kawasaki, J. Chem. SOC., Chem. Commun., 1974, 655. P. G. Harrison and B. Maunders, J. Chem. SOC., Faraday Trans. I , 1984,80, 1341. K. Hirota, K. Fueki, K. Shindo and Y. Nakai, Bull. Chem. SOC. Jpn, 1959, 32, 1261. J. D. Donaldson, J. F. Knifton and S. D. Ross, Spectrochim. Acta, 1964, 20, 847. R. Okawara, D. E. Webster and E. G. Rochow, J. Am. Chem. SOC., 1960,82, 3287. K. Nakamoto, Infrared Spectra of Inorganic and Coordination Compounds (Wiley, London, 1970). lo A. R. Katitzky, J. M., Lagowski and J. A. T. Beard, Spectrochim. Acta, 1964, 20, 847. l 1 T. M. Duncan and R. W. Vaughan, J. Catal., 1981, 67,469. IL2 D. Hadzi and M. Pintar, Spectrochim. Acta, 1958, 12, 162. l3 W. R. Fairheller and J. E. Katan, Spectrochim. Acta, 1967, 23, 2225. l 5 For the infrared spectrum of acrolein see R. K. Harris, Spectrochim. Acta, 1964, 20, 1129. J. C. Leyle, L. H. Zuiderweg and H. J. Viedder, Spectrochim. Acta, 1967, 23, 1397. (PAPER 4/975) 45-2