J. CHEM. SOC. FARADAY TRANS., 1994, 90(21), 3347-3356 FTIR Studies on the Selective Oxidation and Combustion of Light Hydrocarbons at Metal Oxide Surfaces Propane and Propene Oxidation on MgCr,O, Elisabetta Finocchio, Guido Busca and Vincenzo Lorenzelli lstituto di Chimica, Facolta di Ingegneria , Universita, P. le Kennedy, 1-16129 Genova, Italy Ronald J. Willey Department of Chemical Engineering, Northeastern University, Boston, MA 021 15,USA The interaction of propene and propane with the surface of the oxidized spinel MgCr204+x has been studied in the temperature range 300-773 K by FTIR spectroscopy. This solid is reduced reversibly by reaction with these organic compounds in the temperature range 300-673 K, giving rise to stoichiometric MgCr,04 and more oxi- dized organic species that finally produce CO,.Comparison with the results of adsorption and oxidation of C, oxygenates (propan-1-01, propan-2-01, allyl alcohol, propionaldehyde, acetone, acrolein, propionic acid and acrylic acid) as well as of C, and C, oxygenates showed that the predominant oxidation pathways for the two molecules are different. Oxidation of propene occurs predominantly through its previous activation at C, to give strongly adsorbed acrolein and acrylate species. These species later burn. Acetone is the primary oxidation product of propane at the surface, at 423 K. Acetone is later oxidized to formate species (which rapidly decompose) and to acetate species that burn at higher temperatures (573-773 K). The different observed paths are rationalized by taking into account the lower C-H dissociation energy at the methylene group in the case of propane and at the allylic methyl group in the case of propene.The data reported here are consistent with the data available on catalytic alkane oxidation over this and similar catalysts. A comparison is made with the behaviour observed with more selective oxidation catalysts like Mg,(VO,), , V,O,-TiO, and MOO,-TiO, . A mechanism for propene and propane catalytic combustion is proposed. The production of oxygenates via catalytic partial oxidation of hydrocarbons can be achieved with high selectivities using light alkenes as the reactants. Accordingly, the selective oxi- dation of alkenes constitutes a prominent area of industrial petrochemistry. '-' However, several potentially useful oxida- tions of alkenes, e.g.acetaldehyde and acetone production from ethene and propene and the one-step propylene oxide synthesis, still cannot be achieved effeciently in the gas phase. Moreover, efforts are focused today on the use of light alkanes arising from natural gas as raw materials in pet- rochemistry. Nevertheless, the partial oxidation of alkanes is far more difficult than that of alkenes and is at present limited, at the industrial scale, to butane oxidation to maleic anh~dride.~Other processes, like the direct ammoxidation of propane to acrylonitrile' and the oxidative dehydrogenation of propane and butane to the corresponding alkenes6 are now under promising development.On the other hand, the hydrocarbon catalytic combustion is also of great interest at present to limit pollutant emissions of waste In particular, the replacement of noble metal-based catalysts with the less expensive transition-metal oxide-based catalysts has been attempted. In order to develop new, more efficient and more convenient processes based on hydrocarbon catalytic oxidation, detailed mechanisms for these reactions are required. However, the factors affecting the selectivities in hydrocarbon oxidations are still far from clear.2 Moreover, according to Spivey," a general theory of catalytic combustion on metal oxides still does not exist. In our laboratory, FTIR spectroscopy has been recently applied to the study of some oxidation and oxidative dehy- drogenation reactions on selective catalysts.'1-14We recently undertook more systematic studies on selective and unselective oxidation of light hydrocarbons at catalyst sur- faces. As a first step, we report here studies on propene and propane oxidation on MgCr,O, . This material was chosen because it is recognized as a good combustion catalyst for alkanes and alkenes,' 5*1 like chromia and other chr~mites."~'~ MgCr,O, was found to be more active than chromia in methanol oxidation," and is more stable than chromia.' Moreover, it displays significant selectivity in some oxidative dehydrogenation reactions.20 It is a typical p-type semiconductor, and this property is associ-ated with its activity both as an oxidation catalyst and as an oxygen21 and ethanol22 sensor.Finally, it can be prepared as an aerogel and it displays excellent optical properties for IR spectroscopic Experimental The preparation of the MgCr,O, aerogel has been described previ~usly.~~A mixture of Mg acetate and Cr acetylacetonate was dissolved in methanol and hydrolysed with the stoichio- metric amount of water. The resulting gel was dried under supercritical conditions in an autoclave (524 K, 127.5 bar), i.e. above the critical point of methanol (512.6 K, 80.9 bar). The powder was then calcined for 6 h at 973 K. The resulting surface area was 53 m2 g-'. FTIR spectra have been record- ed with a Nicolet Magna 750 instrument, using conventional IR cells connected to an evacuation-gas manipulation appar- atus.The catalyst powder was pressed into self-supporting discs, calcined in air at 723 K for 1 h and outgassed at 723 K for 20 min before use in the interaction experiments. Liquid adsorbates were from Carlo Erba (Milano, Italy), while gases were purchased from SIO (Milano, Italy). The chemicals used in the catalyst preparation were from Aldrich. Results Catalyst Characterization The X-ray diffraction (XRD) pattern of the catalyst shows all of the diffraction peaks typical of the spinel MgCr,04 only (JCPDS table 10-351). The FTIR-FT-FIR spectrum (Fig. 1) also shows the typical features of the spinel MgCr,0424 with J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 wavenumber/cm-' Fig.1 FTIR spectra of MgCr,O,: (a) FTIR spectrum (1200-400 cm-') of a KBr pressed disc; (b) FT-FIR spectrum (500-50 cm-') of the powder deposited on a polyethylene disc; (c) FTIR spectrum of a pressed disc of the pure powder after outgassing at 723 K; (d)after (c) and following an interaction with hydrocarbon in the temperature range 573-773 K the four IR-active modes at 630, 480, 415 and 250 cm-'. However, the IR spectrum, which is also sensitive to non- crystalline and surface species, shows an additional weak complex band in the region 1000-800 cm- ', typically associ- ated with Cr-0 stretchings involving high-valency Cr. Although it seems reasonable to identify these species as involving hexavalent Cr6 + species, according to Klissurski et al." it cannot be excluded that Cr5+ is involved. The above spectral features indeed roughly correspond to the most intense bands of Mg chromate(vI), MgCrO, .26 However, the IR spectra of metal chromate(v) species are characterized by Cr-0 stretching bands (900-800 cm-27) at frequencies only slightly lower than, or superimposed upon, those of the chromate(v1) species.The FTIR spectra of pressed discs of the pure catalyst powder show, as usual, a transmission window above 800 cm-', with maximum transmittance at 1300 cm-' and a pro- gressive decrease in transmission towards higher frequencies due to the wavelength dependence of radiation scattering. Also these samples show the above-cited absorption in the region 1050-800 cm-', clearly structured into one sharp band at 1002 cm-', and components at 975,940,920 and 840 cm-'.A broad absorption is also found near 3500 cm-', due to H-bonded surface hydroxy groups (Fig. 2). The IR spectra of the pure catalyst disc after interaction with hydrocarbons or oxygenated organic compounds in the temperature range 573-773 K are different [Fig. l(6) and Fig. 2(b)].The absorption in the region 1050-800 cm-' is either decreased strongly in intensity or disappears completely, leaving weaker harmonic bands of the skeletal vibrations of MgCr,O,. However, if oxygen or air are admitted into the cell in the same temperature range the absorptions in the 1000-800 cm- 'range reappear quickly. This behaviour agrees with the known p-type semicon- ducting behaviour of MgCr,O, .21~22~28,29In fact, MgCr,O, in oxidizing atmospheres is partly oxidized mainly at the surface" giving rise to MgCr,O,+,, and is associated with the appearance of Cr=O stretching bands of high-valency chromium in the region 1050-800 cm-'.On the other hand, interaction with reducing agents, like hydrocarbons, oxygen- ated compounds and hydr~gen,'~ causes the non-stoichiometric oxygen to be destroyed, producing nearly stoichiometric MgCr,O, . Our data show that redox cycles are relatively fast in the temperature range 573-773 K, i.e. in the range where this 1 4000 3800 3600 3400 3200 3000 wavenu mber/cm -' Fig. 2 FTIR spectra (OH-stretching region) of MgCr,O, disc out- gassed at 723 K: (a) oxidized sample; (b) sample previously reduced in hydrocarbon atmosphere (673 K, 150 Torr) compound is active as an oxidation catalyst' 5,16~19-20 and that the analysis of the IR spectrum in the region 1050-800 cm-' allows the oxidation state of the catalyst to be moni- tored.Reduction also causes the growth of a band near 3750-3650 cm-', due to free surface hydroxy groups [Fig. 2(b)], while bands due to adsorbed hydrocarbon residues may also be present in the region 1800-1200 cm-' [Fig. l(d)]. As discussed recently,', the surface acidity may have a rel- evant role in the catalyst oxidation activity and selectivity. Therefore, we tested the surface acidity of oxdized MgCr,O,+, and of nearly stoichiometric MgCr,O, via adsorption of pyridine as probe (Fig.3). The spectrum of pyridine on the 'reduced' nearly stoichiometric catalyst shows sharp bands at 1607, 1576, 1488, 1443, 1219, 1148, 1068 and 1042 cm- I, associated with pyridine coordinated over Lewis acidic Cr3+ sites.30 The spectrum of pyridine adsorbed on the oxidized sample shows the same pattern but with additional bands. Sharp peaks at 1593 and 1033 cm-I persist after outgassing at room temperature (RT) and are associated either with physisorbed pyridine or with species bonded to OH groups. This species is not so evident on the reduced sample, which might suggest the presence of H- bonding with the hydroxy groups that are responsible, in the 1700 1600 1500 1400 1300 1200 1100 1000' wavenumber/cm-' Fig. 3 FTIR spectra of pyridine adsorbed on oxidized MgCr,O, at room temperature (a), and after successive outgassing for 30 min at room temperature (b), 373 K (c) and 423 K (6);(e) as (b) but on a pre-reduced sample J.CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 activated oxidized sample, for the broad band near 3500 cm-' [Fig. 2(a)]. Additionally, two other weak bands at 1640 and 1538 cm-' are also evident and persist after outgassing at 373 K; these bands are associated typically with pyridinium cations, produced by protonation of pyridine by hydroxy groups of sufficient Brarnsted acid strength. These data show that the oxidized catalyst MgCr204+, has medium Lewis acidity, similar to that of the reduced one, but is also a weak Brarnsted acid, having active hydroxy groups able to interact with pyridine by H-bonding and by protonation. Behaviour similar to the above was found in the case of oxidized and reduced Cr,O, (using ammonia as the basic probe molecule3') and is ascribed to the greater cova- lency of the Cr"+-0 bond (n= 6 or 5) compared with the Cr3+-0 bonds, resulting in a stronger acidity of the OH bonded to the high-valency Cr than that bonded to Cr3+.The subtraction spectra relative to pyridine adsorbed on the oxidized sample also show a negative sharp band at 1003 cm-'. This band has been assigned above to Cr=O stretch- ing of a chromate species. Its disappearance upon pyridine adsorption provides evidence for the ability of the corre-sponding species to interact with pyridine.It seems reason- able that the band at 1003 cm-' is due to the Cr=O stretching of a surface chromyl group having one coordi- native unsaturation, which can consequently coordinate one pyridine molecule, resulting in a weakening of the Cr=O bond and a decrease of the stretching frequency. Surface chromyl species perturbed by adsorbed pyridine were also observed on oxidized ~hromia.~, Interaction with Propene Gas The FTIR spectrum of the adsorbed species arising from contact of the activated MgCr,04 with propene gas at RT is reported in Fig. qa). Most bands are assigned to molecularly adsorbed propene as deduced from a comparison with the spectra of propene in the gas-phase and adsorbed on other oxide surface^.^^.^^ The sharp band at 1638 cm-' is due to the C-C stretching (1653 cm- 'in the gas phase), while those at 1454, 1440, 1420, 1377, 1178 and 1048 cm-' are associated 3349 that the spectrum is detectable and that the C=C stretching is definitely shifted to lower frequencies show that the inter- action strength is non-negligible and probably implies elec- tron withdrawal from the C=C double bond.The vibrational perturbation is weaker than that previously observed on reduced E-C~,O~, but is similar to that occurring on the unreduced FeCrO, dehydrogenation catalyst,' whose surface is also 'covered' by chromate species. We can suppose that chromate species, more than reduced Cr3 + centres, are involved in propene molecular adsorption. Molecularly adsorbed propene is desorbed by outgassing at RT.However, other small absorptions that cannot be assigned to adsorbed propene are observed in the spectrum. These bands include a definite peak at 1675 cm-', almost certainly due to the C=O stretching of an adsorbed carbonyl com- pound, a broad band near 1600 cm-' [possibly due to the vas(COO) of a carboxylate species, whose symmetric counter- part is masked by the CH deformations] and sharp peaks at 1350 and 1256 cm-'. Heating the sample in propene at 373 K causes the forma- tion of stronger bands certainly due to transformation pro- ducts [Fig. 4(c) and (41.The carbonyl band at 1680 cm- ' is now more evident and exhibits a shoulder at 1647 cm-'. The region 1600-1300 cm- ' presents several peaks [1595, 1562, 1452, 1435, 1424, 1382, 1372 (very weak), 1356 and 1330 cm-'1, which are associated with COO stretchings of differ- ent carboxylate species and CH deformations of several adsorbed species.Weak bands are also observed at 1255 and 1241 cm-', while below 1200 cm-' we find peaks at 1180, 1160 (weak), 1115, 1090 and 1030 cm-', (very strong). The features in the 1200-1000 cm-' region are assigned to C-C stretchings and/or C-C/C-0 coupled stretchings of alco- holate species. The bands below 1300 cm -completely disappear upon further heating at 473 K [Fig. 4(f)],when the components near 1680 and 1640 cm-' also decrease in intensity at differ- ent rates, thus demonstrating that they are not due to the same species. Instead, heating causes a further growth of the absorption in the region 1600-1300 cm-', which finally exhibits peaks at 1596, 1560, 1500, 1435, 1385 and 1355 cm-'.with different CH deformations, as dicussed previo~sly.~~,~~ Above 523 K all bands decrease in intensity and CO, is The vibrational perturbation is small and does not give a precise indication of the adsorption mode. However, the fact 0.40---~~ 0.351 0.30 a 0.25 C;0.20 20 m 0.15 0.10 0.05 0.00 2000 19001800170016001500 14001300 12001100 wavenumber/cm-' Fig. 4 FTIR spectra of the adsorbed species arising from propene adsorption over MgCr,O,+, at room temperature (a), after 30 min at room temperature (b), heating in propene at 373 K (c), and after outgassing at 373 K (4,432 K (e),473 K (f)523 K (9) found in the gas phase.Moreover, Cr=O stretching bands are substantially eroded, providing evidence for the reduction of the oxidized catalyst surface according to the oxidation of the adsorbed organic species. This behaviour indicates that propene oxidation at the MgCr,O, surface gives rise above 373 K to alcoholate and carbonyl species that later transform upon heating to carbox- ylate species and finally burn above 523 K. The experiments described below are aimed at the individual identification of these oxygen-containing species. Interaction with Propane Gas The interaction of the activated oxidized catalyst with 50-150 Torr of pure propane gas has also been investigated in the temperature range 300-773 K.Below 373 K, the interaction does not give any appreciable effect. Above 373 K, instead, this interaction gives rise to new absorption bands (Fig. 5) associated with adsorbed species. Simultaneously, the Cr=O stretching bands at 1000-900 cm-', typical of the oxidized catalyst, decrease in intensity, and corresponding negative bands appear in the subtraction spectra. The interaction at 423 K produces two broad and rather strong bands at 1600 and 1440 cm-' and weaker components at 1380 and 1350 cm-'. Together with these absorptions, a pronounced band split at 1702 and 1692 cm- 'and weaker bands at 1245, 1176 3350 0.30 0.25 al 0.20 C 0.15 0 0.10 0.05 0.00 I ,. , . . . , . . . , . . . , .. . , . . . , . . , . . . I . , , . . . , . . . , . .Y 2200 2000 1800 1600 1400 1200 wavenumber/cm-’ Fig. 5 FTIR spectra of the adsorbed species arising from propane adsorption over MgCr,O, and successive evacuation at 423 K (a), 453 K (b),473 K (c), 523 K (6)and 573 K (e) and 1094 cm-’ are found. The features at 1702, 1692, 1245 and 1094 cm-’ disappear progressively upon further heating in vacuum, while the other bands grow and shift slightly to 1595, 1439, 1385 and 1356 cm-’ after heating to 473 K, but later decrease in intensity and disappear above 573 K. Under these conditions, CO, is detected in the gas phase. In the CH-stretching region one weak band is observable at 2880 cm-’ after interaction in the 423-573 K range. A rough assignment of these bands can be proposed at this stage, and will be refined below.The strongest bands at 1595 and 1400 cm-’ are typical of C02 asymmetric and sym- metric stretchings of carboxylate anions, while the split band at 1702, 1692 cm-’ is clear evidence of the formation of a carbonyl compound (C-0 stretching). The bands at lower frequencies are associated with CH deformations and/or C-C stretchings. The spectra show that propane interacts with the surface in the temperature range 373-573 K and undergoes paritial oxidation with a consequent reduction of the catalyst surface. Oxygenate species are produced, whose exact structure will be investigated by comparison with the spectra of any conceivable oxidation product of propane directly adsorbed onto the catalyst surface.These oxygenated species underwent further oxidation and burned in the tem- perature range 573-773 K, with production of gas-phase co, . On the basis of these observations we can conclude that oxidation of propane and propene produces adsorbed species with clearly different IR spectra (cf.Fig. 4 and 5). This sug- gests that two different oxidation paths operate for the two molecules. Identification of the Adsorbed Oxygenate Species and of the Oxidation Pathways The above data showed that partial oxidation products are formed from propene and propane at the surface of our cata- lyst. The gas-phase partial oxidation of propene is carried out industrially to produce a~rolein~~ acrylic acid.36 Other or products can be obtained in non-negligible yields over oxide catalysts, i.e.acet~ne,~*~~-~’ acetal-pr~pionaldehyde,~**~’ deh~de,~’acetic acid and C6 hydrocarbons (hexa-lS-diene and ben~ene).~~.~’ Propane catalytic oxidation also gives rise to a~rolein,~~ acrylic acid41 and propene,42 but acetone and acetic acid were also mentioned as by-products. To determine the nature of the adsorbed oxygenate species observed upon J. CHEM. SOC. FARADAY TRANS., 1994,VOL. 90 propene and propane oxidation over MgCr,04 , we pro-duced them via adsorption of the corresponding carbonyl compounds, carboxylic acids and alcohols. In Fig. 6 the spectra of the species arising from the adsorp- tion at room temperature of acrolein, acetone, propi-onaldehyde and acetaldehyde are reported.The bands assigned to the molecular adsorbed species are summarized in Table 1. In the case of acetaldehyde, strong bands of acetate species (1560, 1435 and 1350 cm-’) are already observed, showing its very fast oxidation. Bands arising from oxidized species are also evident, but weak in the cases of acrolein and propionaldehyde. In contrast the only ketone studied, acetone, is apparently almost stable under these con- ditions. The spectra of carboxylate species produced by adsorption on MgCr204 of formic, acetic and propionic acids are shown in Fig. 7. The spectrum of formate species [Fig. 7(a)] is char- acterized by bands at 2975,2880, 1602, 1390 and 1360 m-’. The four lower-frequency bands are due to CH stretching, asymmetric COO stretching, CH deformation and symmetric COO stretching, re~pectively.~~ The highest-frequency band is due to the combination v,,(COO) + S(CH).47 The band due to the asymmetric COO stretching actually exhibits shoul- ders on both sides, probably due to the presence of species adsorbed differently or on different sites.0.38 0.36 0.34 0.32 0.30 0.288 0.26 2 0.22 0.24 B 0.20fi rn 0.18 0.16 0.14 0.12 0.10 0.08 0.06 I’~‘I’’~1’”1”’1’”I”’I”’1’”I”’1’’.I’’’ 2200 2000 1800 1600 1400 1200 wavenumber/cm-’ Fig. 6 FTIR spectra of the adsorbed species arising from contact of MgCr,O,+, with acrolein (a), acetone (b), propionaldehyde (c), acetaldehyde (4,all after adsorption and outgassing at RT 2.0 1.8 -1 3200 2800 1700 1500 1300 1100 wavenum ber/cm -’ Fig.7 FTIR spectra of the adsorbed species arising from the adsorption of formic acid (a),acetic acid (b)and propionic acid (c) at RT and after outgassing at RT over MgCr,O, J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 3351 Table 1 Position of the IR absorption bands of adsorbed and pure carbonyl compounds acrolein" acetoneb propionaldehyde' acetaldehyded ads. gas ass. ads. gas ass. ads. gas ass. ads. gas ass. E"} y"} .?il) 1754 1708 1724 v(c-0) 1731 v(C=O) v(c--0) 1743 v(C==O)1695 1710 .7.111 1613 1625 v(C-C) 1454 6,(CH,) 1467 1468 1427 1420 S(CH2Y 1435 1435 4&CH,)1422 1410 6,(CH3) 1413 1423 1366 1360 6[CH(0)lf 1369 1363 6,(CH3) 1395 1395 1347 1340 1281 1275 WH-) 1253 1254 -1176 1158 4C-C) ?"} 1215 v,(CCC) 1178 1138 1170 1232 1094 1090 w(CH,) 1096 1100 " Ref.43. Ref. 44.'Ref. 45. Ref. 46. Splittings due to the presence of different sites. The spectrum of acetate species [Fig. 7(b)]is characterized by strong bands at 1560 and 1435 cm-', the latter being more intense and sharper than the former, assigned to the asymmetric and symmetric COO stretchings. At the lower- frequency side a sharp peak at 1350 cm-'is observed, assign- ed to the symmetric CH, deformation, while a very weak band is found near 1055 cm-', probably due to methyl rocking.48 Acetate species are characterized by extremely weak CH stretchings, with a sharp band at 2935 cm-' and broader ones at 3015 and 2990 cm-'.The spectra of propanoate species are, as expected, more complex. In the V(CH) region three sharp and rather strong bands are observed at 2978, 2944 and 2886 cm-'. In the lower-frequency region, the strong and broad bands at 1560 and 1425 cm-' are associated, as always, with the asym- metric and symmetric COO stretchings. Additional sharp but strong peaks are found at 1471, 1380 (shoulder), 1301 and 1080 cm-',which can be assigned to CH, asymmetric and symmetric bending, CH, deformation (probably a rocking mode) and C-C ~tretching.~' The adsorption and oxidation of C1, C, and C, alcohols has also been investigated; some of these experiments gave valuable information for the identification of propane and propeme oxidation pathways. The spectra of the adsorbed species arising from allyl alcohol adsorption on MgCr,04 are shown in Fig.8. Most features correspond closely with the spectrum of adsorbed acrolein [Fig. qa)], i.e. strong C-0 stretching at 1690 cm-' (moving to lower wavenum- bers with decreasing coverage), weak C-C stretching 1613 cm-',sharp CH deformations at 1427, 1366 and 1281 cm-' (the last weak) and sharp C-C stretching at 1176 cm-'.'2*43 The additional bands at 1640 and 1021 cm-' can be assigned" to C-C and C-0 stretching of allyl alcoholate species. So allyl alcohol is partly oxidatively dehydrogenated to acrolein at RT. Further oxidation of acrolein gives rise to carboxylate species from RT to 423 K, when the aldehyde has almost disappeared.In the complex spectrum of the carbox- ylate species arising from allyl alcohol oxidation, features assignable45 to formate species (1600, 1390 and 1360 cm-') and acetate species (1560, 1435 and 1350 cm-') are certainly present. However, additional components like the sharp shoulder at 1640 cm-' [v(C=C)], the bands near 1500 and 1440 cm-' (COO stretchings) and the bands near 1370 and 1270 cm-'(the last a shoulder, CH deformations) correspond to those reported for acrylate species.' 2*5031 The spectra of the adsorbed species arising from contact of propan-2-01 with the catalyst are shown in Fig. 9. According to previous st~dies,'~.~~ the three bands at 1163, 1128 and ~,(CH,) WH2) -1433 6,(CH,) S[CH(O)] 1395 1395 6[CH(O)]w(CH2) 1346 1352 6,(CH3) t(CH2) 4C-C) 1179 1114 v(C-C), r(CH,) v(C-C) 1102 1107 y(CH3), r(CH,) Coupled modes." ' "''"l~ l~~l~~ ~ ~ ~l 2000190018001700160015001400 1300 1200 1100 wavenumber/cm-' Fig. 8 FTIR spectra of the adsorbed species arising from allyl alcohol adsorption over MgCr,O,_, at RT (a), and after successive outgassing for 30min at RT (b), 373 K (c), 423 K (4,473K (e),523 K (f),573 K (9) 1.1 \,-7-~-.7-#--F.-Y 2000190018001700 16001500 1400 1300 12001100 wavenumber/cm-' Fig. 9 FTIR spectra of the adsorbed species arising from propan-2- 01 adsorption over MgCr,O,+, at RT (a), and after successive out- gassing at 373 K (b), 423 K (c), 473 K (4,523 K (e),573 K (f)and 623 K (9) 3352 1103 cm- ' characterize the 2-propoxy species (coupled C-C and C-0 stretchings) produced by propan-2-01 dissociation, while the broad band at 1284 cm-' characterizes the undis- sociated adsorbed propan-2-01 (COH in-plane deformation).However, the sharp bands at 1710 and 1235 cm-', due to C-0 stretching and C-C-C asymmetric stretching of adsorbed acetone [Fig. 6(b) and Table 11, show that propan- 2-01 is oxidatively dehydrogenated at RT on the catalyst surface. The bands at 1467, 1385 and 1347 cm-' are associ- ated with CH deformations of the three species. After the sample is heated, the 2-propoxy bands disappear near 420 K when the acetone bands rise to their maximum intensity. Later, acetone bands also decrease in intensity and disappear, while the bands at 1580, 1435, 1385 and 1350 cm-' grow. These bands are due to a mixture of formate and acetate species [Fig.7(a) and (b)]. The most intense bands, assigned here to 2-propoxy groups and acetone, are also observed (with small shifts) in the spectrum of the species arising from propene transform- ation (Fig. 4); however, they are in this case very weak. In contrast, the strong bands at 1680, 1640 and 1030 cm-' observed there cannot be assigned to these C, species con- taining oxygen at C,. Finally, the complex spectrum of the carboxylate species arising from propene oxidation is defi- nitely different from that of the species arising from propan- 2-01 oxidation. This allows us to conclude that the pathway involving oxygen insertion at C2, definitely predominant on V20,-Ti0, oxidation catalysts,I2 is of only minor impor- tance for MgCr,O, .However, the most prominent features of the spectra of the adsorbed species arising from propene oxidation (Fig. 4) are definitely similar to those of the species arising from allyl alcohol oxidation (Fig. 8). The spectral region 1200-1000 cm-I is dominated by C-0 or coupled C-C/C-0 stretchings of alcoholate-like species. The spectra obtained after propene interaction at 373 K with the catalyst surface show a prominent sharp band at 1020 cm-', and a complex pattern in the region 1150-1090 cm-'. Comparison with the spectra of the surface species arising from ethanol and propan-1-01 adsorption (Fig.10) allows us to exclude the formation of ethoxy and 1-propoxy species from propene in significant amounts. In fact, both these species are characterized by a main sharp band near 1040 cm-' and other components in the range 1120-1040 cm-'. In contrast, the bands of alkoxy groups arising from propene are those typical of allyl alcoholate species [lo20 cm-', Fig. 7(a)] and 2-propoxy species [1150-1090 cm-', Fig. 6(a)]. Analogously, no evidence is found for adsorbed acetaldehyde and propionaldehyde in the spectra of the species arising from propene adsorption. In conclusion, com- parison of the spectrum of the adsorbed species arising from propene oxidation with those arising from allyl alcohol oxi- dation shows that the same species are formed and dominate the spectra in both cases, although 2-propoxy species and adsorbed acetone can also be found, in small amounts.In the spectra of the species arising from propane oxida- tion (Fig. 5), bands due to acetone (near 1700 and 1240 cm-') can be found. For this reason acetone adsorption and oxida- tion has been studied in detail (Fig. 11). At room tem-perature, adsorbed acetone gives rise to sharp and intense bands, as summarized in Table 1. In the temperature range 423-573 K acetone is converted to carboxylate species. Com- parison of the spectra in Fig. 1 l(d) and (e)with those reported in Fig. 7 shows that the former are due to a mixture of ace- tates and formates. At this temperature ketone species are no longer observed: the typical bands at 1710 and 1233 cm-' have disappeared.Obviously, the spectra of the species arising from acetone and propan-2-01 transformation are very similar (cf Fig. 8 and 11). J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 1, l.,..lv1100 wavenumber/cm-' Fig. 10 FTIR spectra of the adsorbed species arising from contact of MgCr,O,+, with ethanol (a) and propan-1-01 (b)at room tem- perature Comparison of the spectra of the adsorbed species arising from oxidation of propane with those due to species arising from adsorption and oxidation of acetone strongly supports the assignment of the carbonyl species formed by propane oxidation as acetone. This is due to the close correspondence of most band positions, as well as of the further oxidation products of both molecules. The absence of the strong bands at 1470, 1300 and 1080 cm-' in the spectra of the adsorbed species arising from propane oxidation on the MgCr,O, surface clearly argues against the presence of propanoate species.On the other 0.20.41 0, c(D nin 2000 1800 1600 1400 1200 wavenumber/cm-' Fig. 11 FTIR spectra of the adsorbed species arising from acetone adsorption at room temperature over MgCr,O, and following evac- uation at RT (a),423 K (b),473 K (c), 523 K (4,573 K (e),623 K (f), 673 K (9) J. CHEM.SOC. FARADAY TRANS., 1994, VOL. 90 3.41 3.03*21 2.8 1 y 2.6/2.42*2L2.0 11.84 8 ' ' ' I . ' ' 2000 1800 1600 1400 1200 ' 1000 waven u m ber/cm -Fig.12 FTIR spectra of adsorbed species arising from formic acid adsorption over MgCr,O, at 373 K (a)and following decomposition at 473 K (b)and 573 (c) hand, the absence of a sharp v(C=C) band in the region near 1635 cm-' and of a COO asymmetric stretching component near 1500 cm -characterizing adsorbed acrylate species (see above), precludes the presence of acrylate and, consequently, C, carboxylate species are apparently not formed at all. Simi- larly, no trace is found of the C, aldehydes acrolein and pro- pionaldehyde (cf: Fig. 5 and 6). In contrast, the spectra of carboxylate species arising from both propane and acetone oxidation can easily be obtained from a sum of the spectra of acetate and formate species. From this analysis we conclude that carboxylate species are obtained by the oxidative cleav- age of the C(l)-C(2) bond of acetone and propane. Thermal Evolution of Carboxylate Species The study of the species arising from adsorption of C,-C, acids as well as of the corresponding aldehydes and alcohols allowed us to determine the stability range of these adsorbed species on MgCr,O,.For completeness, we also studied the stabilities of the carboxylate species produced by propane oxidation, i.e. formates and acetates. As mentioned above, the IR spectrum of the formate species shows, after adsorption at room temperature, three main bands (1602, 1390 and 1360 cm-'). Above 473 K these bands quickly decrease in intensity (Fig. 12) giving rise to carbon oxides in the gas phase and Q1 C e n8 bands at 1500 and 1350 cm-' assignable to surface carbonate species,26 while there is a very slight decrease in the intensity of the Cr=O stretching bands.This supports the idea that formate decomposes following the reaction : HCOO--+ CO + -OH. This reaction is a simple decompo- sition, and does not imply catalyst reduction. However, some of the CO can be oxidized to CO,, with consequent catalyst reduction. The evolution of acetate species with temperature is shown in Fig. 13. Acetates, characterized by bands at 1560, 1435, 1350 and 1055 cm-' (see above), are substantially stable up to 573 K, while they progressively disappear in the range 573-773 K, giving rise to bands due to carbonate species and to gas-phase carbon oxides.However, their disappearance is accompanied by a strong decrease of the Cr=O stretching bands in the region 1000-800 cm-'. This provides evidence that the disappearance of these species is due to their overox- idation to carbon oxides, with a parallel reduction of the catalyst surface. Our data show that formate species undergo decomposi- tion below 573 K, so they are very unstable at the tem- perature at which propane begins to be oxidized. Acetate species, instead, burn at higher temperatures with a conse- quent reduction of the catalyst; they are almost stable up to above 573 K. Discussion Mechanism of Propene Oxidation at the MgCr,O, Surface The present results show that propene reacts with the surface of oxidized MgCr,O, at room temperature, but this reaction becomes more evident near 373 K when adsorbed acrolein and allyl alcoholate species are observed, although acrylate species (and perhaps other carboxylate species) have already appeared.However, acrolein and allyl alcohol in the tem- perature range 373-473 K are rapidly oxidized, giving rise to acrylate species as the predominant products, which become unstable above 523 K, being burnt. Thus, we have provided evidence for the main oxidation pathway of propene on MgCr,O, (Scheme 1). This pathway implies the breaking of the allylic C-H bond with a subsequent oxygen insertion to give allyl alco- holate species, which are, in fact, observed oia the detection of + C H2= CH-C HZO H It'\ 2000 1800 1600 1400 1200 1000 wavenumber/cm-' Fig.13 FTIR spectra of adsorbed species arising from acetic acid co -co2 adsorption over MgCr,O, at 473 K (a),573 K (b),673 K (c), 773 K Scheme 1 Propane and propene oxidation pathways and their con- (4 nection at the surface of the oxidized MgCr,O, catalyst the typical sharp C-0 stretching band at 1020 cm-'. Further evolution strictly parallels allyl alcohol oxidation. The key step in this mechanism is the breaking of the allylic C-H bond. The active site in our case is identified as a Cr"+=O site (n= 5 or 6), which should be reduced to CF2)+.A likely intermediate in this process is the forma- tion of an allyl species, either anionic or radical, as proposed for the selective oxidation of propene to acrolein.2 However, such an intermediate can be observed only if its formation is faster than its further reaction with surface oxide species to give allyl alcoholate.Anionic allyl species have been unequivocally identified on Zn0,53 and 21-0,, 34 which are not active allylic oxidation catalysts (so the subsequent oxygen insertion should be slow or may not occur at all) and have a strong basic character. On true allylic oxidation cata- lysts such as uranyl antimonate,', acrolein was found upon propene adsorption, but neither the allyl intermedite nor allyl alcoholate were detected spectroscopically. This is prob- ably because anionic or radical-like allyl species are oxidized so rapidly to allyl alcoholate species that they cannot be observed.This implies that the rate-determining step is the first hydrogen abstraction from the methyl group. Also allyl alcoholates can be rapidly dehydrogenated to acrolein at the temperature at which they form, so their detection can be difficult too. This situation differs from that found in the case of toluene oxidation to benzaldehyde on vanadia-titania.' In this case, the rate-determining step is probably the reac- tion of the intermediate benzyl species with surface oxygen to give benzyl alcoholates, which are later rapidly dehydroge- nated. For this reason benzyl species are observed while benzyl alcoholates are not." The reverse is observed here for allyl and allyl alcoholate species. Besides the allylic oxidation pathway at C(l), another propene oxidation pathway is evident on MgCr,O, (although it is definitely of minor importance) with the pro- duction of species oxidized at C(2) (Scheme 1).According to our previous this pathway is associated with the electrophilic attack of weakly Bransted acidic OH groups of the alkene double bond, according to the Markovnikov rule, giving rise to secondary propoxy species that are later oxi- dized to acetone, and finally undergo oxidative breaking of the C(2)-C/3) bond leading to acetate and formate species. This pathway (called oxidative hydration2) was shown to be predominant on vanadia-titania," which is less active as an oxidation catalyst and gives mixtures of acetone and acrolein upon propene ~xidation.'~ Mechanism of Propane Oxidation at the MgCr,O, Surface The present data show that propane gas interacts with the surface of oxidized MgCr,O, starting from above 373 K giving rise to adsorbed oxygen-containing species that can further be oxidized to carbon oxides.Complete burning of hydrocarbon and oxygenated organic compounds by the oxi- dized MgCr,O, catalyst is observed at 773 K. The data pre- sented here also give a relatively simple picture of the propane combustion pathway on this surface. In fact, the first detectable adsorbed oxidation product is acetone, which is later converted into a mixture of formate and acetate species, by oxidative cleavage of the C(l)-C(2) bond. Formate species rapidly decompose while acetate species burn only near 773 K (Scheme 1, left-hand side).Surprisingly, com- pounds oxidized at C( 1) like acrolein and acrylate species are very evident upon popene oxidation under the same condi- tions, but are not found at all with propane. So, the pathways of propane and propene oxidation (Scheme 1, right-hand side) are considerably different. This supports the data reported several years ago by YaoS7 which showed that the J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 oxidation pathways of alkanes and alkenes are different on a-Cr203. The formation of acetone from propane implies propane activation at C(2). This agrees with the lower dissociation energy of C-H at secondary carbons (-CH,- methylene groups, 94 kcal mol-') with respect to C-H at primary carbon atoms (methyls, 99 kcal mol- ,).Accordingly, also the C-H bond dissociation energy at the primary carbon in the case of propene is in agreement with the lower energy for C-H bond cleavage at an allylic methyl group (77 kcal mol-') with respect to that of a vinylic group (105 kcal mol-'). The lower temperature at which propene oxidation to acrolein is observed, with respect to propane oxidation to acetone, is justified by the lower C-H bond dissociation energies of allylic methyls with respect to alkane methylenes. It seems reasonable, although it is not strictly proved here, that cleavage of the first C-H bond of propane is followed by insertion of an oxygen before the second C-H bond at C(2) is broken.This hypothesis is confirmed by the detection of allyl alcoholates from propene, as well as by studies which are now in progress5' concerning the oxidation of other hydrocarbons. Therefore, 2-propoxy species should be the first intermediates in propane oxidation, although they cannot be observed easily because at the temperature at which their formation is sufficiently fast, their further trans- formation to acetone is even faster, making their concentra- tion zero. The active site for C-H bond scission is certainly associ- ated with the high-valency Cr ions responsible for the Cr=O stretching bands in the region 1000-800 cm-'.It is not excluded that the sites active for propane activation do not entirely coincide with those active for propene oxidation.In fact, it is possible that only the most active sites can attack both molecules, the least active ones being able to attack propene only. On the other hand, sites able to break the C-H bonds of the propene methyl groups can cooperate with sites able to interact with the alkene double bond. Acetone is a rather reactive molecule, and this causes its strong adsorption and further oxidation at the C(2)-C(3) bond to formate and acetate species. The last stages in propane combustion correspond to the evolution of acetate and formate species. However, the behaviour of these two species is different. Formates decompose rapidly, giving CO, while acetates need further oxidation to give finally CO, . Comparison with Other Catalytic Systems: Activity-Selectivity in Heterogeneously Catalysed Oxidation of Propene and Propane The data reported above concern surface reactions observed to take place from an oxidized MgCr,O, catalyst surface with propane and propene.However, they do allow a devel-opment of the theory of catalytic hydrocarbon combustion mechanisms on metal oxides. The conditions under which these interactions are observed (373-773 K) agree roughly with those at which this compound was found to act efi- ciently as a combustion catalyst for propene and light alkanes.15*58Our data show that propene combustion can be obtained in the absence of gaseous oxygen, totally at the expense of the Cr=O oxidized sites of the catalyst, which in the 573-773 K range are quickly reoxidized by gas-phase oxygen.This strongly supports the idea that catalytic com- bustion of propene and propane over this catalyst can occur completely with a Mars-Van Krevelen type me~hanism.'~ Our data also show that adsorbed partial oxidation pro- ducts are formed upon propene and propane oxidation at the surface of MgCr,O,, well known to be essentially a non- J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 selective combustion catalyst. These adsorbed partial oxida- tion products are intermediates in the formation of carboxylate species, already recognized as intermediates of propene combustion over chromia-based catalysts.' 'v60 The data concerning propene oxidation on this com-bustion catalyst can be discussed in relation to previous data obtained on selective oxidation catalysts.' '-14 Similar surface reaction pathways are observed for selective and non-selective oxidation catalysts.However, important differences can be found in the temperature at which the same steps occur on different catalysts. Propene activation is observed to commence on MgCr,O, at 373 K, while on Mg3(V0,), it starts at near 523 K.'4*61The much lower reactivity of Mg vanadate with respect to Mg chromite with propene explains why the former acts as a selective catalyst for propane oxida- tive dehydrogenation to propene,42 while the latter is anon-selective combustion catalyst for both propane and propene.' Acrolein is also definitely oxidized on MgCr,O, at 373 K, while it is much more stable on the surfaces of V,0S-Ti02 l2 and MOO,-TiO,.62 This could explain why these catalysts are rather selective in the oxidation of propene to a~rolein~~ while MgCr,O, is not. The reaction pathway proposed in Scheme 1 implying propane activation at C(2) on MgCr,O, also agrees with the behaviour reported for another chromite, ZnCr,O,, as a catalyst for n-butane oxidation. In fact, over this catalyst complete butane combustion has been observed at 623 K, while at 493 K butane oxidation gives acetic acid and methyl vinyl ketone [i.e. compounds functionalized at C(2)] with rather high ~electivity.~~ On the other hand, acetic acid is frequently reported as a product of propane oxidation over oxide catalyst^:^' this could be the result of activation at C(2) and agrees with the stability of acetate species to overoxida- tion, evidenced above, allowing desorption of acetic acid.According to Scheme 1, propene is not an intermediate in the main propane oxidation path, so that propene and propane oxidation follow predominantly two different path- ways on the MgCr,O, surface. On the other hand, our data suggest that decomposition of 2-propoxide groups (the first intermediate starting from propane) to propene and an OH group is competitive with its oxidative dehydrogenation giving rise to acetone (Scheme 1). This competition could provide a 'cross-roads' between alkane oxidative dehydroge- nation and oxidation at C(2). On MgCr,O,, the dehydroge- nation of 2-propoxy groups to acetone is much faster than their decomposition to propene, which is consequently not an intermediate for further oxidation.This makes MgCr,O, an oxidation (combustion) catalyst rather than an oxidative dehydrogenation catalyst for propane. This picture finds support in the comparison between the behaviour of MgCr,O, and Mg vanadate catalysts, which are completely opposite. On Mg vanadates, in fact, alcoholate species decompose quickly at very low temperature, giving only traces of oxidized species. Therefore, the equilibrium propene s2-propoxide is displaced left (towards the alkene) on Mg vanadates, in contrast to Mg chromite where it is shifted to the right (towards the alkoxide). This agrees with the behaviour of Mg vanadates as oxidative dehydrogenation catalysts for production of propene from propane.,, Comparison of the behaviour of the different metal oxide catslysts allows us to apply the generalization of the mecha- nism of Scheme 1 to all of them.The behaviour of an oxida- tion catalyst is strongly influenced by its acid-base properties. In fact, the decomposition of the alcoholate species giving rise to the alkene is a purely acid-base reac-tion. According to our data, the equilibrium propene e2-propoxide is displaced left (towards the alkene) on both very basic and very acidic catalysts (in fact, alcohol dehydration is either acid- or base-catalysed66), while it is shifted right (towards the alkoxide) for catalysts with medium Brsnsted acidity.In fact, Mg chromite shows Brransted acidity while Mg vanadate does not.61 The propane oxidative dehydrogenation pathway (through propene) can be followed by allylic oxidation, giving rise to acrolein and/or acrylic acid. Therefore, catalysts with no Brsnsted acidity but strong Lewis acidity (allowing decompo- sition of 2-propoxy groups to propene) and very active allylic activation centres are the best candidates for propane oxida- tion to acrolein. These data support the idea that the main difference between a selective and a non-selective oxidation catalyst is in its ability to overoxidize selected products. This leads us to propose that a combustion catalyst is one that is able to perform selective oxidation, but from which the partial oxidation product is unable to be desorbed without further oxidation.This is supported by two observations : (i) Efficient partial oxidation processes from hydrocarbons are limited to the production of compounds such as acrolein, acrylonitrile, maleic and phthalic anhydrides, which have intrinsic chemical stability owing to their ability to delocalize ionic charges. Compounds whose ionic charges are more localized (like acetone or the non-conjugated aldehydes, with respect to acrolein) are more reactive towards the oxide cata- lyst surface (which always contains electrophilic and nucleo- philic sites) and are desorbed less easily. (ii) Basic dopants frequently increase the selectivity but decrease the activity of partial oxidation catalysts, becase they lower the adsorption strength of the oxygenate compounds (which are always Lewis bases) on the surface (which always contains Lewis acid sites), but this tends to limit the activity for C-H bond scission (which needs Lewis acidity',).Therefore, the main requirement of a partial oxidation catalyst is inactivity towards overoxidation of the desired product, in spite of being a poorly active catalyst. This is the case for V2O5- and Moo3-based catalysts, which are gener- ally much less active than typical combustion catalysts, like, among others, those based on Cr203. This view is only partially in disagreement with that of Haber2*67 which associates selective oxidation with nucleo- philic oxygen species (0'-lattic oxide ions) and combustion with electrophilic oxygen (O,, 0,-and O,,-).Our study confirms that lattice oxygen performs selective oxidation, but suggests that it may also be involved in non-selective com- bustion pathways. This does not preclude the involvement of electrophilic oxygen (together with nucleophilic oxygen) in a different combustion pathway. Conclusions The FTIR study of propene interaction with the surface of oxidized MgCr,O, shows that the alkene is totally oxidized in the temperature range 573-773 K at the expense of surface Cr"+=O (n = 5 or 6) species. However, this occurs through two different 'selective oxidation ' pathways at the surface, one of which involves activation at C(1) via ally1 alcoholate species, and the other activation at C(l) via 2-propoxy species.Combustion seems to involve essentially successive overoxidation of the partial oxidation products. Propane is oxidized at the same surface to acetone, which further transforms into acetates and formates and, finally, into carbon oxides; this occurs in the 573-773 K range. This oxidation pathway is completely different from the main one undergone by propene. The surface reactions observed allow us to propose a detailed reaction pathway for both selective and unselective oxidation of propane and propene. On the basis of the comparison of these data with data available for 3356 J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 other catalytic systems we propose this pathway as a general 34 G.Busca, V. Lorenzelli, G. Ramis and V. Sanchez Escribano, mechanism for alkane and alkene catalytic oxidation. Muter. Chem. Phys., 1991,29, 175. 35 R. K. Grasselli and J. D. Burrington, Adu. Catal. Relat. Subj., 1981,30,133.This work has been supported in part by MURST (Rome, 36 K. Waissermel and H. J. Arpe, Industrial Organic Chemistry, Italy) and by NATO (CRG n. 900463.). Verlag Chemie, Weinheim, 1978, p. 256. 37 Y. Moro-Oka, S. Tan and A. Ozaki, J. Catal., 1968,12,291. 38 N. Giordano, N. Meazzo, A. Castellan, J. C. J. Bart and V.References Ragaini, J. Catal., 1977,50, 342. 1 Chemical Processing Handbook, ed. J. J. McKetta, Marcel 39 R. F. Howe and H. Minming, in Proc. 9th Int. Congr. Catal., Dekker, New York, 1993. Calgary, The Chemical Institute of Canada, Ottawa, 1988, p.2 A. Bielanski and J. Haber, Oxygen in Catalysis, Marcel Dekker, 1585. New York, 1991. 40 Y. C. Kim, W. Ueda and Y. Moro-oka, in New Developments in 3 New Developments in Selective Oxidation 11, ed. V. Cortes Corb- Partial Oxidation, ed. G. Centi and F. Trifirh, Elsevier, Amsterd- eran and S. Vic Bellon, Elsevier, Amsterdam, 1994. am, 1990, p. 491. 4 G. Centi, F. Trifiro, J. Ebner and V. Franchetti, Chem. Rev., 41 M. Ai, J. Catal., 1986, 101,389. 1988,88,55. 42 M. A. Chaar, D. Patel, M. C. Kung and H. H. Kung, J. Catal., 5 BP America, US Pat. 5008427, 1991; SOHIO, Jpn. Pat., H3-1987,105,483. 157356,1991. 43 R. K. Harris, Spectrochim. Acta, 1964,20, 1129. 6 D. Bhattacharayya, S. K. Bey and M. S. Rao, Appl. Catal.A, 44 G. Dellepiane and J. Overend, Spectrochim. Acta, 1966,22,593. Gen., 1992,87, 29. 45 E. F. Worden, Spectrochim. Acta, 1962,18, 1121. 7 R. Prasad, L. A. Kennedy and E. Ruckenstein, Catal. Rev. Sci. 46 K. B. Wiberg, V. Walters and S. D. Colson, J. Phys. Chem., 1984, Eng., 1984,26, 1. 88,4723. 8 J. J. Spivey, Ind. Eng. Chem. Res., 1987,26, 2165. 47 G. Busca, J. Lamotte, J. C. Lavalley and V. Lorenzelli, J. Am. 9 M. F. M. Zwinkels, S. G. Jaras and P. G. Menon, Catal. Reu. Chem. SOC., 1987,19,5197. Sci. Eng., 1993,35, 319. 48 F. Vratny, C. N. R. Rao and N. Dilling, Anal. Chem., 1961, 33, 10 J. J. Spivey, in Catalysis, The Royal Society of Chemistry, 1455. London, 1989, vol. 8, p. 157. 49 D. Lin-Vien, N. B. Colthup, W. G. Fateley and J. G. Grasselli, 11 G. Centi and G.Busca, J. Am. Chem. SOC., 1989,111,46. Handbook of Infrared and Raman Characteristic Frequencies of 12 V. Sanchez Escrbano, G. Busca and V. Lorenzelli, J. Phys. Organic Molecules, Academic Press, San Diego, 1991. Chem., 1990,94,8939,8945; 1991,95,5541. 50 W. R. Feairheller and J. E. Katon, Spectrochim. Acta, Part A, 13 G. Busca and V. Lorenzelli, J. Chem. SOC., Faraday Trans. 1, 1967,23,2225. 1992,88,2783. 51 A. A. Davydov, Infrared Spectroscopy of Adsorbed Species on the 14 G. Busca, V. Lorenzelli, G. Oliveri and G. Ramis, in ref. 3, p. Surface of Transition Metal Oxides, Wiley, New York, 1990, p. 253. 195. 15 L. Ya. Margolis, Adv. Catal. Relat. Subj., 1963, 14,429. 52 P. F. Rossi, G. Busca, V. Lorenzelli, 0.Saur and J. C. Lavalley, 16 F. G.Dwyer, Catal. Rev. Sci. Eng., 1972,6,261. Langmuir, 1987,3, 52. 17 Y. Yu Yao, J. Catal., 1974, 28, 139. 53 A. L. Dent and R. J. Kokes, J. Am. Chem. SOC., 1970,92,6709. 18 A. Terlecki-Baricevic, B. Grbic, D. Jovanovic, S. Angelov, D. 54 G. W. Keulks and Z. Yu, in Proc. 8th Int. Congr. Catal., Mehandziev, C. Morinova and P. Kirilov-Stefanov, Appl. Catal., Dechema, Berlin, 1984, vol. 111,p. 289. 1989,47, 145. 55 G. Busca, J. Chem. SOC., Faruday Trans., 1993, 89, 753; G. 19 A. A. Arve, G. Miliades and J. C. Vickerman, J. Catal., 1980, 62, Gusca, in Catalytic Selective Oxidation, ed. S. T. Oyama and J. 202. W. Hightower, ACS, Washington, 1993, p. 168. 20 R. J. Rennard and W. L. Kehl, J. Catal., 1971,21,282. 56 A. Doulov, M. Forissier, M.Noguerol Perez and P. Vergnon, 21 Y. Shimizu, S. Kusano, H. Kuwayama, K. Tanaka and M. Ega- Bull. SOC. Chim. Fr., 1979,I-129. shira, J. Am. Ceram. SOC., 1990,73,818. 57 Y. Yu Yao, J. Catal., 1974,28, 139. 22 Z. Chen and K. Colbow, Sensors and Actuators, B, 1992,9,49. 58 E. Finocchio, G. Busca, V. Lorenzelli and R. J. Willey, work in 23 R. Willey, P. Noirclerk and G. Busca, Chem. Eng. Commun., progress. 1993, 123, 1. 59 P. Mars and D. W. Van Krevelen, Chem. Eng. Sci., 1954,3,41. 24 N. W. Grimes and A. J. Collett, Phys. Stat. Sol., 1971,43, 591. 60 V. A. Kuznetsov, S. V. Gerei and Ya. B. Gorokhovatskii, Kinet. 25 D. Klissurski, K. Hadjiivanov and A. A. Davydov, J. Catal., Katal., 1977, 18,710. 1988,111,421. 61 G. Ramis, G. Busca and V. Lorenzelli, J. Chem. SOC., Faruday 26 0.Muller, W. B. White and R. Roy, Spectrochim. Acta, Part A, Trans., 1994,90, 1293. 1969,25,1491. 62 G. Ramis, G. Busca and V. Lorenzelli, Appl. Catal., 1987, 32, 27 R. G. Darrie and W. P. Doyle, Reactivity of Solids, ed. J. W. 305. Mitchell, R. C. DeVries, R. W. Roberts and P. Cannon, Wiley, 63 T. Ono, Y. NaKagawa, H. Miyata and Y. Kubokawa, Bull. New York, 1969, p. 281. Chem. SOC. Jpn., 1984,57,1205. 28 K. Balasubramanian and V. Krishnasami, J. Chem. SOC., 64 G. Centi, F. Trifiro, A. Vaccari, G. M. Pajonk and S. J. Teichner, Faraday Trans 1, 1986,82,2665. Bull. SOC. Chim. Fr., 1981,I-290. 29 B. Gillot and F. Jemmaly, Muter. Chem. Phys., 1987, 18, 139. 65 J. Barrault, L. Magaud, M. Ganne and M. Tournoux, in ref. 3, p. 30 G. Busca, V. Lorenzelli, G. Ramis and R. J. Willey, Langmuir, 305. 1993,9, 1492. 66 K. Tanabe, New Solid Acids and Bases, Elsevier, Amsterdam, 31 K. Hadjiivanov and G. Busca, Langmuir, in the press. 1989. 32 A. Zecchina, E. Guglielminotti, L. Cerruti and S. Coluccia, J. 67 J. Haber, in Catalysis of Organic Reactions, ed. J. R. Kosak and Phys. Chem., 1972,76,571. T. A. Johnson, Marcel Dekker, New York, 1994, p. 151. 33 G. Busca, G. Ramis, V. Lorenzelli, A. Janin and J. C. Lavalley, Spectrochim. Acta, Part A, 1987,43,489. Paper 4102726D; Received 9th May, 1994