J . Chem. SOC., Faraday Trans. 1, 1982, 78, 803-815 Analysis of the Factors Affecting Selectivity in the Partial Oxidation of Benzene to Maleic Anhydride Part 3.-Mechanism of Benzene Surface Oxidation on a Vanadium Pentoxide-Molybdenum Trioxide Catalyst BY RAYMOND W. PETTS AND KENNETH C . WAUCH* The Corporate Laboratory, I.C.I., P.O. Box 11, The Heath, Runcorn, Cheshire WA7 4QE Received 6th April, 1981 Rate measurements are described in which benzene, hydroquinone and p-benzoquinone were oxidised over a vanadium pentoxide-molybdenum trioxide catalyst. These experiments showed : (i) that the catalytic oxidation of benzene did not involve any homogeneous component as has been previously reported and (ii) that identical selectivities (ca. 60%) to maleic anhydride could be obtained from benzene or hydroquinone, p-benzoquinone being oxidised mainly to carbon oxides with negligibly small selectivities (< 6%) to maleic anhydride. The selectbe reaction pathway from benzene to maleic anhydride is therefore taken to involve hydroquinone, the oxidation of which can result in maleic anhydride or p-benzoquinone, the latter being the main intermediate in the non-selective pathway.An explanation of these results is found in orbital symmetry conservation arguments which show that the concerted addition of a chemisorbed oxygen molecule (i.e. having an extra electron in an antibonding n orbital) para, across the ring of a chemisorbed benzene molecule (i.e. a benzene molecule deficient of electrons in the highest occupied orbital), is allowed.This adsorbed adduct is presumed to rearrange to form hydroquinone which, by another identical molecular addition of oxygen across the ring, followed by the elimination of a C,H, fragment and of water, forms maleic anhydride. The low surface oxidation activation energy, the necessity for the involvement of molecular chemisorbed oxygen and the identity of the selectivities of benzene and hydroquinone are accounted for in this mechanism. The near inability of p-benzoquinone to form maleic anhydride is also accounted for, since its structure and probable mode of chemisorption (donation of electrons to the catalyst from the carbon+arbon double bond) are unlikely to produce the 1 $-oxygen adduct necessary for maleic anhydride formation. Previous paperslv have examined the detailed kinetics of the vanadium pentoxide- molybdenum trioxide (3 : 1) catalysed oxidation of benzene to maleic anhydride.The combination of the transient techniques of temperature programmed desorption (t.p.d.), temperature programmed reaction spectroscopy (t.p.r.s.) and gas adsorption chromatography have shown the reaction to be rate limited in the desorption of the product, the surface oxidation of an adsorbed benzene molecule to an adsorbed maleic anhydride molecule being particularly facile, having an activation energy of only 36 kJ mol-l. Furthermore no evidence was obtained for the existence of the benzene- oxygen adduct which is proposed by Trimm and coworkers3 to be formed on the surface, to desorb and to oxidise further in the gas phase to maleic anhydride.The experiments described in this paper were undertaken with a two-fold purpose. First, they would provide a rigorous test of the possibility of there being a homogeneous component in the reaction and, secondly, by oxidising two of the more likely intermediates (hydroquinone and p-benzoquinone) in the oxidation of benzene to maleic anhydride they would provide mechanistic information on the selective reaction pathway. 803804 SELECTIVITY I N OXIDATION OF BENZENE EXPERIMENTAL THE CATALYST The method of preparation of the catalyst, a supported 3: 1 molar V,O,: MOO, catalyst, has been described previously.' It had a particle size of 0.5 cm diameter and a surface area of 2.2 m2 g-l. MATERIALS The benzene was AnalaR grade supplied by Hopkin and Williams (Essex, England).Its specification was that not less than 95% of it boiled in the range 352.5-353.5 K and its im- purities were: sulphur-containingcompounds 3 x lo-, %, thiophene 2 x %, water 5 x lo-, %. The hydroquinone and the p-benzoquinone were both supplied by B.D.H. (Poole, England); the former was AnalaR grade having a melting point in the region of 444-448 K, the major impurity being catechol(O.O2%). The latter was technical grade having a melting point in the region of 386-388 K. Its impurities (ca. 1 %) are hydrobenzoquinone and benzoquinol. Both were used without further purification. The gases, oxygen and helium, were supplied by British Oxygen Company. Their specified purities were oxygen: oxygen 99.7%, carbon dioxide < 2 x lo-,%, carbon monoxide < 1 x hydrocarbons as CH, < 2 x lo-,%; helium: helium 99.5%, oxygen (1-2) x lo-,% [the oxygen content was measured at 1.5 x lo-,% on a Hersch oxygen meter (Englehard Industries Limited)], carbon monoxide nil, carbon dioxide nil, nitrogen < %.BENZENE OXIDATION: TUBULAR REACTOR RATE MEASUREMENTS The temperature dependence of benzene conversions and selectivities to maleic anhydride listed in table 1 were obtained in a glass tubular reactor shown in diagram in fig. 1. The reactor, which was mounted vertically, was 91 cm long, 1.9 cm internal diameter and was heated over N FIG. I.-Diagram of the apparatus used for the air oxidation of benzene over the vanadium pentoxide- molybdenum trioxide catalyst. its entire length by a fluidised bed sand bath. To test the possibility of there being a homogeneous-heterogeneous interaction in the benzene oxidation,, i.e.that a component of the overall reaction could include the gas-phase oxidation of a desorbed intermediate, provision was made for the sampling of products (a) immediately after the catalyst bed, (b) after 210 cm3 free space and (c) after 356 x 0.635 cm diameter glass balls packed on top of the catalyst. The benzene + air mixtures were prepared by injecting a continuous stream of benzene from a burette under pressure into an air flow. Minor fluctuations in the reactant stream wereR. W. PETTS AND K. C. WAUGH 805 removed by passing through a large mixing vessel (250 cm3, see fig. 1). The benzene + air mixture was pre-heated by passage down the length of the sand bath and then over a s h r t column (5 cm long) of glass beads.Calibration was made by titrating the amount of benzene injected during the course of the experiment, knowing the total gas flow. The system gave a very stable gas mixture. The reactant and product streams were analysed by gas chromatography using two columns: a poly-2,2-dimethyl propane- 1,3-succinate column [ 10% on Embacel (60-80 mesh)] separated benzene, maleic anhydride, acrylic acid and acetic acid, while a Porapak Q column separated CO and CO,. OX1 D AT10 N OF p-BEN ZOQ U I NON E The line diagram for the apparatus used in the oxidation of p-benzoquinone is shown in fig. 1. However, the apparatus itself differed from that used to oxidise benzene in two major respects: (i) the p-benzoquinone+air feed was achieved simply by passing air over a column ofp-benzoquinone held at 333 K and (ii) the tubular reactor was replaced by a stirred gas solid rea~tor.~ OXIDATION OF HYDROQUINONE Because of the highly reactive nature of hydroquinone neither the tubular reactor described previously nor the stirred gas solid reactor could be used for its oxidation; considerable blank conversions of the hydroquinone to p-benzoquinone were observed in the lengths of heated metal tubing leading to the tubular reactor and also on the walls of the stirred gas solid reactor.A line diagram of the apparatus used in hydroquinone oxidation is shown in fig. 2. M S PR I O2 N2 1 l25OC L - - - J r - - -6cTi , SAV3 , FIG. 2.-Diagram of the apparatus used for the oxidation of hydroquinone over the vanadium pentoxide- molybdenum trioxide catalyst. C1 = Apiezon column; C2 = poly-2,2-dimethyl propane- 1,3-succinate column; C3 = Porapak columns; C4 = CO to CO, converter; CT = cold trap; FID = flame ionization detector; HC = heating coils; HT = heating tape; HS = hydroquinone saturator; MS = molecular sieve; OM = oxygen meter; PR = pressure regulator; R = rotameter; SAV = sample valve; SW = switching valve; TR = tubular reactor; WGM = wet gas meter.The hydroquinone + air feed was produced by bubbling nitrogen through a saturator containing liquid hydroquinone at 453 K; the saturator was housed in a Pye series 104 gas chromatography oven. Oxygen and nitrogen to make a final inlet composition of 20% oxygen were introduced to the nitrogen + hydroquinone mixture and this feed was diverted by switching valve 1 either to the glass tubular reactor (8 cm long, 1 cm internal diameter) or to the feed analytical column [Apiezon L (2 ft, 10% on Embacel 80-100 mesh)] all of which were housed in the Pye oven.The reactor itself was maintained at the temperature of interest (580-650 K) by heating tape, the majority of the piping being glass. In this way blank conversion of the hydroquinone was minimised to a negligible value. The products were analysed on two columns identical to those used in benzene oxidation.806 SELECTIVITY IN OXIDATION OF BENZENE CALIBRATION Calibration of the gas chromatographic analysis system for the composition of hydroquinone feeds was made by correlating peak areas with weighed amounts of hydroquinone collected in the cold trap after a given length of time at a measured flow rate.Calibration of the flame ionization detector response to p-benzoquinone was made using standard solutions in p-xylene. RESULTS AND DISCUSSION RATE MEASUREMENTS BENZENE OXIDATION It has been reported3 that the oxidation of benzene to maleic anhydride over a vanadium-molybdenum catalyst proceeds mainly by the gas-phase oxidation of a desorbed, partially oxidised, benzene adduct. The tubular reactor used in this work offers a more rigorous test of any homogeneous involvement than was available previ~usly.~ The free space provided after the reactor (2 10 cm3) for the homogeneous reaction to proceed was much greater than before3 where 16.32 cm3 was the maximum volume available. Furthermore it was possible tb sample the gas at the end of the catalyst bed and after the 210 cm3 free space for the same catalyst on the same experiment (in previous experiments it was necessary to change reactors to do this). Inspection of table 1 shows that at 643 K benzene conversions and selectivities to maleic anhydride are identical at the end of catalyst bed and after 210 cm3 free space.As a check that the homogeneous reaction was not being propagated in the 8.8 cm3 volume of the sample tube to the end of the catalyst bed (70 cm long, 0.4 cm internal diameter) the free volume in this tube was reduced by packing it with glass wool (it had also been claimed3 that any inert substance would quench the homogeneous reaction). Results with the packed sample tube were identical to those after 210 cm3 free space.At 618 K the conversions and selectivities show a time dependence (due possibly to some phase change in the catalyst at the outset or to the deposit of a small amount of carbon) but after 180 min the results obtained were again identical at the end of the catalyst bed and after 210 cm3 free space. Equally, in contrast to T ~ i m m , ~ packing the free space above the catalyst with glass balls (356 x 0.635 cm diameter, i.e. 451 cm2 surface area) produced no effect, the results at the end of the catalyst bed and after the glass balls again being identical. We conclude therefore that oxidation of benzene to maleic anhydride occurs totally on the surface of the catalyst and has no homogeneous, i.e. purely gas phase, component. An explanation of the previous author^'^ observations of the effects of free space after the packed bed might be found in the suggestion that their reactor design would produce eddies in the free space after the catalyst and that these would cause a re-mixing of the reactants and catalyst, leading to further reaction.The longer reactor and the packed length of inerts after the catalyst required for an analysis of the effect of inert packing would reduce this re-mixing effect considerably and so would apparently reverse the reaction. Indeed, the suggestion that a catalyst by the reaction of two adsorbed species could produce a species which could desorb and be decomposed by any inert substance is a definition of a system for the creation of energy.The adsorbed species would contain the heats of adsorption of benzene, oxygen and of the adduct itself, plus its heat of formation. The gaseous benzene and oxygen molecules produced by its desorption and collisional decomposition with an inert substance would contain,TABLE TUBULAR REACTOR RATE MEASUREMENTS FOR THE AIR OXIDATION OF BENZENE OVER THE VANADIUM MOLYBDENUM CATALYST gas-phase composition/mole fraction x 1 O2 ~~ ~ exit (a) exit (b) exit (a) exit (b) time to temp flow rate sample inlet maleic maleic conversion selectivity conversion selectivity /K /cm3 s-' /min benzene benzene anhydride benzene anhydride (%) (%I (%I (%I 618 12.7 12.7 12.7 12.7 12.7 12.7 12.7 12.7 643 643 6.6 6.6 6.6 6.6 30 60 90 120 150 180 210 240 30 60 90 120 1.079 1.079 1.079 1.079 1.079 1.079 1.079 1.079 0.763 0.763 0.763 0.763 0.986 0.0455 - - - - 1.022 0.034 1.025 0.033 - - 1.029 0.033 0.152 0.328 0.155 0.328 - - exit (a) 6.6 6.6 6.6 6.6 6.6 6.6 30 60 90 120 150 180 0.816 0.816 0.816 0.8 16 0.816 0.816 0.149 0.324 - - - - 0.133 0.355 0.126 0.344 - - 0.994 0.048 1.011 0.037 - - 1.029 0.035 1.028 0.035 0.158 0.323 - - 0.154 0.328 exit (c) - - 0.123 0.340 0.127 0.353 0.129 0.340 - - 8.6 48.9 - - - - 5.3 59.7 5.0 61.1 - - - - 4.6 66.0 - - 80.1 53.7 79.7 54.0 exit (a) - - 7.9 48.0 6.3 54.4 - - 4.6 69.6 4.7 68.6 79.3 53.4 - - 79.8 53.9 exit (c) 81.7 48.6 - - - - 83.7 52.0 84.6 49.9 - - - - 84.9 49.1 84.4 51.2 84.2 49.5 - - - - Exits: (a) gas stream sampled at end of catalyst bed, (b) gas stream sampled after 210 cm3 free space, (c) gas stream sampled after 356 x 0.635 cm Weight of catalyst, 15 g ; surface area of catalyst, 2.2 m2 g-l.benzene (inlet) -benzene (exit). Conversion = , selectivity = benzene (inlet) diameter glass balls (ie. 451 em2 glass surface) packed on top of the catalyst bed. maleic anhydride (exit) benzene (inlet) - benzene (exit) *808 SELECTIVITY I N OXIDATION OF BENZENE distributed between them in vibrational or translational degrees of freedom, the heats of adsorption of benzene and oxygen on the catalyst, i.e. these identical reactants would somehow have gained energy. Additionally, were the collisional decomposition energies for the adduct quoted by Brown and Trimm3 correct (a value of 152 kJ mol-l was estimated for the decompo- sition of the adduct to oxygen and benzene by collision with an inert substance), at the temperature studied (ca.670 K) the mean translational energy is only ca. 5 kJ mol-l and so virtually none of the molecules colliding with an inert substance would be expected to decompose. In total then the arguments presented previously3 about the involvement of a homogeneous component in the oxidation of benzene to maleic anhydride over a vanadium pentoxide-molybdenum trioxide catalyst are both infeasible and logically inconsistent. p-BEN Z OQU I N ONE 0 X ID A TI 0 N The results of the rate measurements of the oxidation of p-benzoquinone over the vanadium pentoxide-molybdenum trioxide catalyst in a stirred gas solid reactor4 in the temperature range 543-563 K are listed in table 2. (Blank experiments, i.e.catalyst removed, showed negligible oxidation of the p-benzoquinone and of the maleic anhydride product at these temperatures.) TABLE 2.-sTIRRED GAS SOLID REACTOR RATE MEASUREMENTS FOR THE AIR OXIDATION OF P-BENZOQUINONE OVER THE VANADIUM PENTOXIDE-MOLYBDENUM TRIOXIDE CATALYST gas-phase composition/mole fraction x 1 O2 exit time to inlet maleic carbon con- selec- temp flow rate sample p-benzo-p-benzo- an- carbon mon- version tivity /K /cm3 s-l /min quinone quinone hydride dioxide oxide (%) (%) 563 9.5 563 9.5 563 9.5 563 9.5 543 9.83 543 9.83 543 9.83 553 9.34 17 42 72 90 30 51 80 93 0.60 0.28 0.60 0.35 0.60 0.37 0.60 0.48 0.62 0.33 0.62 0.42 0.62 0.49 0.60 0.40 0.015 0.013 0.010 0.008 0.0096 0.0058 0.0034 0.062 0.79 0.70 0.51 0.49 0.54 0.37 0.32 0.566 0.25 53.3 4.7 0.24 41.7 5.2 0.14 38.3 4.3 0.12 20.0 6.7 0.10 46.8 3.3 0.039 32.3 2.9 0.026 21.7 2.6 0.102 31.0 3.1 p-benzoquinone (inlet) -p-benzoquinone (exit).p- benzoquinone (inlet) Weight of catalyst, 13.27 g; conversion = 9 maleic anhydride (exit) p-benzoquinone (inlet) -p-benzoquinone (exit) * selectivity = Two points are immediately apparent : (i) significant conversions ofp-benzoquinone are observed at similar contact times (l/flow rate x weight of catalyst) but at lower reaction temperatures than required for comparable benzene conversions (at only 50 % greater contact time the amount of p-benzoquinone converted is 3 times that of benzene, the latter however having been oxidised at a 50 K higher temperature) and (ii) the selectivities to maleic anhydride from p-benzoquinone are much lower than observed in benzene oxidation. Taken in combination these results indicate that theR.W. PETTS AND K. C. WAUGH 809 mechanism and kinetics of p-benzoquinone oxidation over the vanadium pentoxide- molybdenum trioxide catalyst are totally different from those of benzene, the former having an overall activation energy to its main product, carbon dioxide, lower than that of benzene to its main product, maleic anhydride. Germain's mechanism5 which depictsp-benzoquinone as the main intermediate in the oxidation of benzene to maleic anhydride can therefore no longer be taken as valid. [The time dependence of the p-benzoquinone oxidation (table 2) is probably due to carbon deposit.] HYDROQUINONE OXIDATION The results of the rate measurements of the oxidation of hydroquinone in the temperature range 580-660 K are listed in table 3.The difficulty of maintaining a constant hydroquinone feed was overcome by saturating nitrogen with hydroquinone using a low flow rate of the carrier gas (0.83 cm3 s-l). An oxygen+nitrogen mixture was then added to this saturated nitrogen stream to bring the oxygen content to 20% and the total flow rate to 2.5 cm3 s-l. The resulting long contact times (ca. 10 times longer than for benzene) produced complete conversion of the hydroquinone. TABLE 3.-TUBULAR REACTOR RATE MEASUREMENTS FOR THE AIR OXIDATION OF HYDROQUINONE OVER THE VANADIUM PENTOXIDE-MOLYBDENUM TRIOXIDE CATALYST gas-phase composition/mole fraction x 1 O2 exit inlet temp flow rate hydro- hydro- p-benzo- maleic carbon carbon selectivity /K /cm3 s-' quinone quinone quinone anhydride dioxide monoxide (%) 588 2.5 1.45 0 0.41 0.49 2.16 1.08 33.8 597 2.5 1.45 0 0.25 0.62 1.80 1.14 42.8 620 2.5 1.45 0 0.04 0.81 2.52 1.26 55.9 628 2.5 1.45 0 0 0.91 2.10 1.26 62.8 634 2.5 1.45 0 0 0.84 2.40 1.38 57.9 652 2.5 1.45 0 0 0.93 1.80 1.38 64.0 maleic anhydride (exit) hydroquinone (inlet) - hydroquinone (exit) Weight of catalyst, 8.034 g; selectivity = In the temperature range 620-650 K the selectivities to maleic anhydride from benzene and from hydroquinone are virtually identical, suggesting that hydroquinone is an intermediate in the selective reaction pathway from benzene to maleic anhydride.However, at lower temperatures the selectivities to maleic anhydride are significantly lower and considerable quantities of p-benzoquinone are formed.(The selectivities to p-benzoquinone are 28.3% at 588 K, 17.2% at 597 K and 2.8% at 620 K.) From the inverse relationship between the maleic anhydride selectivities, which increase over the temperature range 588-628 K, and those to p-benzoquinone, which decrease over the same temperature range, it is inferred that the activation energy for the adsorption of hydroquinone into that state which producesp-benzoquinone has a small advantage over the energy barrier for adsorption into the species which eventually forms maleic anhydride. However, no hydroquinone was ever detected in the products of benzene oxidation over the vanadium pentoxide-molybdenum trioxidecatalyst, while only trace quantities 21 FAR 1810 SELECTIVITY I N OXIDATION OF BENZENE ofp-benzoquinone have been noted, which, when considered with our previous finding that the maleic anhydride is formed at the site at which the benzene was first adsorbed,2 suggests that the difficulty in adsorbing hydroquinone propitiously does not exist in benzene oxidation itself.The reaction is confined entirely to the surface of the catalyst. MECHANISM A rationale for the observation that benzene and hydroquinone have identical selectivities, while that of p-benzoquinone is extremely low, and for our previous finding that the selective oxidant is molecularly held oxygen2 is to be found by consideration of the electronic perturbations of the reactants on chemisorption and the reactions allowed, on the basis of orbital symmetry arguments,g to these electronically perturbed adsorbed species.[While acknowledging that the Woodward- Hoffmann rules are strictly qualitative, dealing mainly with the angular contribution to the total wavefunction, in effect neglecting the radial contribution, they are introduced as a form of justification of the feasibility of reactions which are postulated to occur between chemisorbed oxygen and benzene and chemisorbed oxygen and hydroquinone (in the absence of this justification these reactions might otherwise be viewed as speculative). The arguments are considered solely in respect of the adsorptions and reaction on the V5+, V4+ ions at the surface but apply, with minor modifications for ionic size, to Mo6+, Mo5+ ions, should they also occur at the surface.] The 30% molar solution of MOO, in V205 coincides with its maximum solubility in V205.7 The role of the former, by the substitution of the larger Mas+ ion for the V5+ ion, has been judged to produce inclusion strain in the vanadium pentoxide lattice7 and therefore to maximise the number of V4+ ions.Shvets et aZ.,8r9 using e.s.r. spectroscopy, have shown that the mechanism of oxygen adsorption on V,05 is O,(gas) -+ 02(ads) + O;(ads) -+ 20-(ads) + 202-(1at) (1) [where 02- (lat) is oxygen of the lattice] and that 0; (ads) and 0- (ads) co-exist on the surface. The 0; species is formed by interaction of the precursor 0, (ads) with a V4+ ion on the surface, donation of the electron being made from either the d,, or d,, orbitals of the vanadium ion to one of the antibonding II orbitals of the oxygen Q D Q D FIG.3.-Symmetry allowed combination of the oxygen K* orbitals and of the dyL and d,, orbitals of the vanadium ion V4+.R. W. PETTS AND K. C. WAUGH 81 1 atom. These are symmetry allowed reactions (see fig. 3). The electronic configuration of the 0, (ads) will therefore be [+)I2 [o*(s)l2 [a(p)I2 [zI4 [z*I3, the extra electron being located in either the x or y, n* orbitals. [The oxygen molecule in fig. 3 is depicted as adsorbing ' end-on ', although a ' sideways-on ' configuration is also symmetry allowed. The spatial arrangement of the n* orbitals of oxygen in 0; and of the dvz or dzz orbitals of the V4+ ion suggest that the end-on configuration will lead to the maximum orbital overlap and hence the stronger bond.This conclusion derives from charge density considerations for 0; lo which show that the 0.002 contour of the total charge density (95% of the total charge) in 0; has the dimensions of 4.45 A along the intermolecular axis and 3.39A at right angles to it, while the maximum charge density of the d,, FIG. 4.-(a) Symmetries of the constituent atomic orbitals of the allowed combinations which comprise the six molecular orbitals of benzene. (b) Bonding orbitals involved in the chemisorption of benzene on the V6+ ion of vanadium pentoxide; the yl(E,a) orbital of benzene overlaps with the d,, or the d,, of the vanadium ion. 21-2812 SELECTIVITY I N OXIDATION OF BENZENE or dz5 orbitals is 0.8 A apart, 95% of the total charge lying within a 3 A distance.llU The end-on adsorption therefore more closely coincides with the spatial distribution of charge in the d,, or dzz orbitals.] The weakly bound oxygen which has been shown to be the selective oxidant in benzene oxidation2 is likely to be 0;.The signs of the atomic wavefunctions of the allowed combinations which constitute the six molecular orbitals of the electron system of benzene are shown in fig. 4(a).12 The energies of these orbitals are y/(A) < y/(Ela) = y/(E,b) < y/(E2a) = y/(E,b) < y/(B) so that the ground-state electronic distribution in benzene is [y/(A)I2, Chemisorption of benzene will probably form a compound shown in fig. 4(b), which is strictly analogous to the chromium-benzene complexes well known in transition- metal chemistry.llb The bond will be formed, on distance and overlap considerations, by electron donation from the y/(Ela) orbital to either the duz or the dz5 orbitals of a surface V5+ ion [fig.4(b)]. The electronic configuration of chemisorbed benzene will The correlation diagram for the concerted reaction of chemisorbed oxygen with [y/(El 412, [y/(El b)I2, ME2 41°, [y/(E2 b)I09 "3)1°. therefore be [W)l2, [y/(El 4 3 ' 9 [y/Y(El 412, [y/(E2 41°, [ W 2 b)I0, [y/(B)I0. chemisorbed benzene [reaction (2)] is shown in fig. 5. CJc FIG. 5.--Correlation diagram for the 1,4- addition of a chemisorbed 0; species across a chemisorbed benzene molecule.R. W. PETTS AND K. C . WAUGH 81 3 When orbital electronic occupancy is considered the combination of an electron deficient y/(E,a) orbital and a singly occupied 7t* oxygen orbital is both symmetry allowed and makes a net bonding contribution of two electrons in the o o orbital. Surprisingly, reaction (2) is also perfectly feasible on bond-distance considerations, since in spite of the oxygen molecule bond length being only 1.2 A as described earlier, the charge density of the 0; extends to 4.45 A, while the 1,4- distance across the benzene ring is only 2.79 A.(Indeed, 1,3- addition of 0; to chemisorbed benzene, which is also bonding and symmetry allowed, is less favoured on bond-distance considerations, it being only 2.42 A; the resultant overlap with the oxygen antibonding orbitals will be smaller and therefore weaker.) The mode of transport of the end-on chemisorbed oxygen molecule anion to the chemisorbed benzene molecule (a previous paper2 has shown that it is the oxygen which migrates to the immobile, adsorbed benzene) is envisaged as involving a pendular vibration of the oxygen molecule as a whole with respect to the vanadium cation to which it is attached.Adsorbed vicinal to the chemisorbed benzene molecule, the rocking vibrational mode of the oxygen molecule [indicated by the bent arrow, n, in reaction (2)] could easily result in 1,4- addition to the benzene molecule, which itself is held planar to the catalyst at the $6- positions. The product of reaction (2), the benzene-oxygen adduct (identical to that proposed by Dmuchovsky et all3), will because of the heat released in forming two carbon-oxygen bonds and because of the extra electron in the antibonding oxygen orbital, probably rearrange to form hydroquinone.Although the addition of gas-phase oxygen to the electronically deficient, chemisorbed benzene molecule is symmetry allowed, the net bonding contribution is only one electron in the 0-0 orbital; this would result in a lower heat release on formation of the benzene-oxygen adduct, which, together with the absence of an extra electron in the oxygen antibonding orbital, would result in the adduct having a higher activation energy (lower probability) of rearranging to hydroquinone. As always, the apparently lower energy reaction pathway (in this case the one involving the chemisorbed 0; species) is favoured, but this nevertheless accords with our previous observation2 that the overall activation energy for the oxidation of an adsorbed benzene molecule to an adsorbed maleic anhydride molecule is low, 31.4 kJ mol-l.Adsorption of oxygen, now to a position vicinal to a chemisorbed hydroquinone molecule, followed by 1,4- addition of the adsorbed oxygen molecule anion to the chemisorbed (and therefore electronically perturbed) hydroquinone molecule produces a hydroquinone-oxygen adduct, strictly analogous to that formed between chemisorbed benzene and chemisorbed oxygen [reaction (3)]. The correlation diagram for this reaction is identical to that employed for the formation of the benzene oxygen adduct (fig. 5) and the reaction is therefore symmetry allowed and bonding. Like the benzene-oxygen adduct, the hydroquinone-oxygen adduct will rearrange but in this case to adsorbed maleic anhydride, an adsorbed acetylide residue and water.The complete reaction sequence is shown in reaction (3), all the reactions of which are confined to the surface of the catalyst [the reaction of gas-phase molecular oxygen with gas-phase hydroquinone, or with the benzene-oxygen adduct as required by Trimm and c o ~ o r k e r s , ~ is symmetry forbidden, while the formation of the comparable anthracene-oxygen adducts, common in organic chemistry, requires the involvement of the excited singlet state of oxygen (a state not far removed from the 0; species) for the reaction to occur].14 The near identity of the selectivities of benzene and hydroquinone is explained by this mechanism as is the necessary involvement of a weakly bound, molecularly held oxygen.The stability of the strongly chemisorbed2 maleic anhydride molecule to814 SELECTIVITY I N OXIDATION OF BENZENE \ I 3 \ I + v5+ 5 +(+el 1 \ / 5 + t + e ) + V 4 + O-(ads) OzPi[at I I o=y+ + C,H,( ads) - \ I + H20 \ / ,, 5 t (+el COz+ CO+ HZO t v4+ 1 further oxidation can also be understood since 1,2- addition of 0; to the carbon-carbon double bond of the chemisorbed maleic anhydride molecule is forbidden on bond- distance considerations; the reaction therefore must involve the 0- (ads) or 02- (lat) species, the latter of which, according to Bond,15 is difficult to remove. The non-involvement ofp-benzoquinone in forming maleic anhydride (the low selectivities to maleic anhydride obtained from it could almost be due to hydroquinone impurities) follows from its structure. It will not be involved in 1,4- molecular oxygen addition, 2,3- reaction with the carbon-carbon double bond being forbidden on distance considerations, and so, like maleic anhydride, it can only be involved with the non- selective reactions of 0- (ads) or 02- (lat). J. Lucas, D. Vandervell and K. C. Waugh, J. Chem. SOC., Faraday Trans. I , 1981, 77, 15. J. Lucas, D. Vandervell and K. C. Waugh, J. Chem. SOC., Faraday Trans. I , 1981, 77, 31. D. M. Brown and D. L. Trimm, Proc. R. SOC. London, Ser. A, 1972,326, 215. M. L. Brisk, R. L. Day, M. Jones and J. B. Warren, Trans. Inst. Chem. Eng., 1968, 46, No. I , T3. J. E. Germain, Catalytic Conversion of Hydrocarbons (Academic Press, New York, 1969), chap. 5, p. 259. F. D. Mango, Aduances in Catalysis (Academic Press, New York, 1969), vol. 20, p. 291. I. I. Ioffe, Z. E. Ezhkova and A. G. Lyubarskii, Kinet. Katal., 1962, 3, 194.R. W. PETTS AND K. C. WAUGH 815 V. A. Shvets, V. M. Vorotyntser and V. B. Kazanskii, J. Catal., 1968, 10, 287. * V. A. Shvets, M. E. Sarichev and V. B. Kazanskii, J. Catal., 1968, 11, 378. lo P. E. Cade, R. F. W. Bader and J. Pelletier, J . Chem. Phys., 1971, 54, 3517. L. E. Orgel, An Introduction to Transition Metal Chemistry: Ligand Field Theory (Methuen, London, 1960), (a) p. 21, (b) p. 160. l2 F. A. Cotton, Chemical Applications of Group Theory (Wiley-Interscience, New York, 1971), p. 136. l3 B. Dmuchovsky, M. C. Freerks, E. D. Pierron, R. H. Munch and F. B. Zienty, J. Catal., 1965,4,291. l4 H. Heaney, Comprehensive Organic Chemistry, ed. J. F. Stoddart (Pergamon Press, Oxford, 1979), I6 G. C . Bond, Catalysis by Metals (Academic Press, New York 1962), chap. 1, p. 2. vol. I, p. 330. (PAPER 1 /545)