J. CHEM. SOC. FARADAY TRANS., 1994, 90(21), 3341-3346 3341 Applications of EPR to a Study of the Hydrogenation of Ethene and Benzene over a Supported Pd Catalyst: Detection of Free Radicals on a Catalyst Surface? Albert F. Carley, Hywel A. Edwards, Brynmor Mile, M. Wyn Roberts and Christopher C. Rowlands School of Chemistry and Applied Chemistry, University of Wales College of Cardiff, P.O.Box 912, Cardiff, UK CFI 3TB Fred E. Hancock and S. David Jackson lCl Katalco, R & T Group, Billingham, Cleveland, UK TS23 ILB EPR has not been used extensively in the field of catalysis despite it being the most sensitive technique avail- able for detecting free radicals and paramagnetic metal ions, which are intermediates or participants in many catalytic processes.In this paper, we describe its application to the detection of paramagnetic intermediates on the surface of a palladium catalyst during heterolytic hydrogenation reactions. EPR, in conjunction with spin trapping, is shown to provide a convenient, simple method for detecting hydrogen adatoms generated by the dissociative chemisorption of hydrogen on alumina-supported palladium catalysts (t0.04% Pd) at room tem- perature. By using D, we have also been able to demonstrate directly the occurrence of hydrogen/deuterium spillover onto the alumina surface by H/D atom transfer to surface hydroxy groups. Alkyl and aromatic free radical intermediates formed by reaction of the H’ adatoms with alkenes and benzene have also been observed by EPR at a catalyst surface for the first time.The hydrogen atoms reacting with the spin traps, ethene and benzene, are not strongly chemisorbed hydrogen (Pd-H) but those weakly adsorbed H-adatoms in equilibrium with H,(g). Despite it being a valuable technique for detecting, identify- ing and monitoring free radicals and paramagnetic ions, elec- tron paramagnetic resonance (EPR) spectroscopy has been rather neglected in the field of catalysis. Nevertheless, there have been important contributions by Lunsford,’ Kevan and Naryana,2 Kasai and Bi~hop,~ Rhodes and Webster4 and others on zeolite catalysts, the incisive results of Che and Tench on metal oxides,> the studies of complex aluminas and silicas6 and the clear demonstration of free radical par- ticipation on metal oxide surfaces by Lunsford and co-workers using matrix-isolation methods.’ CH2FH2H There have been relatively few studies by EPR of real cata- lyst systems since they can be complex, often giving poorly &I CH2-CH2 resolved anisotropic spectral transitions which are difficult to assign.Our understanding of catalytic events at the mecha- A/ \&Inistic molecular level is still at a ‘primitive’ state compared with that of gas and liquid phase reactions. This is in spite of the application of powerful new surface science techniques. CH2-CH2-CH2-CH2Supported-palladium catalysts with very low Pd loadings CH3-CH3(g) (<OM% w/w Pd) are used industrially for the removal of hethyne from ethene streams by the selective hydrogenation of the ethyne to less than 5 ppm before passage of the ethene to Scheme 1 polymerisation plants.Gases such as CO are often added to used avoided level crossing muon spin resonance (ALC-pSR) increase selectivity. Besides the hydrogenation reactions, oli- to detect and measure the diffusion characteristics of ethylgomerisation also occurs to yield a complex mixture of and cyclohexadienyl radicals on silica and alumina sur-hydrocarbons ( 2C4), commonly termed ‘green oil ’, which face~,~2-14 the situation in general is clearly unsatisfactory deactivates the catalyst. The mechanism of reactions shown in Scheme 1 is based on that in the recent review by Bos and Westerterp.* Surface hydrogen atoms, ethylidene, vinyl, ethyl and higher alkenyl and alkyl radicals are implicated as inter- mediates.Elegant labelling and spectroscopic techniquesg support this postulated mechanism from which rate expres- 1 sions have been derived and kinetic parameters extracted. lo spin trapHowever, no surface participants other than the ethylidene Uintermediate, which is a spectator species,” have been char- acterised spectroscopically. Although there is the recent 1 impressive work of Roduner and co-workers, who have rlspin addud ~~~~ ~ t This paper was presented at the 27th International ESR con- ference at the University of Wales, Cardiff, 21st-25th March, 1994. Scheme 2 H2or H2+ C2H4 --T----sinter degassedspin trapsolution in flask F \ catalyst Fig. 1 Schematic of apparatus for exposure of the supported Pd catalyst to gas mixtures, followed by treatment with spin trap solu- tions and transfer to EPR tubes and hence this appeared to be an area where EPR could be usefully deployed.The present work illustrates how free atoms and free radicals present on a palladium catalyst (Scheme 2) can be detected by the method of spin trapping.” Our experiments are based on the pioneering, but neglected, work of Howe and co-workers16 in their studies of hydrogen atoms on the surface of zinc oxide using spin traps. Experimental The apparatus used is shown in Fig. 1. The crushed ‘egg shell’ commercial catalyst (ca. 0.04% w/w Pd on either an alumina or calcium aluminate support with total surface areas of 55 m2 g- ’ and 10 m2 g- ’,respectively, and Pd par- ticles of 1.5-3.0 nm dispersed near the primary support parti- cle surface) were pretreated in situ at 400°C for 2 h under vacuum.By opening tap A they were exposed to atmospheres of pure hydrogen or hydrogen and ethene mixtures at 25°C for up to 10 min. The degassed spin trap solutions [O.Ol mol dm-N-tert-butyl-a-phenylnitrone (PBN) or 2,4,6-tri-tert-butylnitrosobenzene (BNB) in degassed CH,Cl,], which were isolated in flask F from the catalyst during these oper- ations, were than poured onto, and stirred with, the catalysts for prescribed periods of time before being tipped into the quartz EPR sample tube through the coarse glass sinter. After closing tap B, the sample was inserted into the EPR cavity of a JEOL JES-RE2X spectrometer operating at X band with 100 kHz modulation and detection at this fre- quency.The spectra were collected and averaged over 10 multiple scans. In some experiments, the catalyst was evac- uated to mbar for ca. 30 min to remove all but strongly chemisorbed hydrogen adatoms before exposure to the spin trap solution under helium gas. The catalyst was prepared by established methods at ICI Katalco.” H2(> 99.99%), D2(> 99.8%) and ethene (~99.98%) were supplied by BOC Special Gases; benzene, CH2Cl,( >99Y0) and the spin traps were supplied by Aldrich. Results and Discussion Hydrogen Dissociation and Spillover Fig. 2(a) shows the EPR spectrum of a PBN solution after admixing with the hydrogenated catalyst for 10 min.The iso- J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 1 rnT I-+ I Fig. 2 (a) EPR spectrum at 25°C of a PBN solution in CH2C12 (0.01 mol dm-3 PBN) after mixing with a Pd catalyst treated with H, (1 x lo5 Pa) for 10 min at 25 “C; (b) simulation of spectrum using parameters given in the text tropic spectrum is clearly a triplet of triplets and is accurately simulated [Fig. 2(b)] by the hyperfine parameters a&) = 1.53 mT with one nitrogen nucleus and 42) = 0.82 mT, with two equivalent hydrogen nuclei. The spectrum can be assigned unequivocally to the hydrogen atom spin adduct, A (Scheme 3)’ by comparison with well established spectra and hyperfine parameters for the monohydrogen atom adduct of PBN.16*18 The present spectra are superior to those previously reported in that M,(N) = -1, M,(2H) = +1 and M,(N) = 0, MA2H) = -1 together with the corresponding high-field transitions are clearly resolved here compared with the unre- ’ilsolved doublets previously observed.’ No EPR tran-sitions were detected when the undoped support was similarly treated.0 *. 0 spin addud A Scheme 3 There can be little doubt that the PBN has been able to scavenge surface hydrogen atoms to give the adduct A (Scheme 3). The rate constants for radical reactions in solu- tion with PBN are of the order of 1 x 10’ dm3 mol-” s-’ and clearly large enough in the case of adsorbed hydrogen atoms to ensure that they can be removed efficiently from the metal surface.We sought to confirm our conclusions by substituting pure deuterium for hydrogen in an identical experiment, expecting the triplet of triplets from the hydrogen atom adduct to be replaced by a triplet of doublets of triplets with a,,(l) = 0.12 mT from the monodeuteriated, monohydrogenated spin adduct. Fig. 3(a)shows that these were indeed the major tran- J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 1 mT Pd-Pd -0-Al-O-Al-O- Al-f O-AI-O-Al-O-Pd-Pd+-d dl 0 Al 0 Al 0 A1 0 Pd-Pd -0-AI-0-Al-0-AllO-AI-0-Al-0-Pd-Pd11 b 0 11< Al 0 Al 0 Al 0 Scheme 4 1 mT+ 1 mT I--+ Fig. 3 (a) EPR spectrum of a PBN solution after mixing with the Pd catalyst exposed to D, for 10 min at 25°C; H denotes lines from the dihydro PBN spin adduct, all other transitions arise from the monohydro-monodeuterio spin adduct; (b) simulation of a 15 :85 composite spectrum from the dihydro and monohydro-mono PBN spin adduct mixture.(c) spectrum of a PBN solution after mixing with a Pd catalyst treated with D,O before conditioning at 400°C and then exposure to pure D, for 10 min at 25 "C. sitions observed (ca. 85%) but, surprisingly, despite the high isotopic purity of the D, used (>99.8%), there were still sig- nificant transitions from the dihydrogen adduct [labelled in Fig. 3(a)by HI, indicating directly that hydrogen atoms were generated by adsorbing D, on a supported palladium surface. The simulation in Fig. 3(b)is that for a 15 :85 HD composite adduct spectrum. The most likely source of hydrogen is the surface hydroxy groups of the alumina support (usually about five OH groups per 100 A,).This was proven by repla- cing the surface hydroxy groups with deuterioxy groups during multiple treatment of the catalyst with D20 at 80°C (1 :100 catalyst to D,O ratio). This D,O-exchanged catalyst was pretreated at 400°C and then exposed to D, gas and PBN solution as already described. The EPR spectrum, Fig. 3(c), shows that, as expected, the dihydrogen signals have been considerably reduced. We believe these results to be direct evidence for hydrogen spillover, where hydrogen atoms formed on metal particles spill over onto the surface of the support, which does not, by itself dissociatively chemisorb molecular hydrogen. The We now consider the nature of the hydrogen/deuterium atoms that react with the spin trap to give the hydrogen/ deuterium spin adduct.Two possibilities exist: (a) the hydro- gen atoms are those strongly chemisorbed to the palladium; (b) they are mobile hydrogen atoms only weakly adsorbed at the palladium surface. The following experiment was per- formed to distingush between these two alternatives. The H, treated catalyst was evacuated for 30 min at 21 "C and mbar and then exposed to the PBN spin trap solu- tion under helium. Thus, only strongly chemisorbed hydro- gen adatoms were left on the surface to react with the PBN. This procedure resulted in the complete loss of the dihydro- gen spin adduct showing conclusively that (b) is correct, i.e.the active hydrogen is not that strongly chemisorbed at the palladium surface. It is interesting to speculate that these hydrogen adatoms may be those that are active in alkene hydrogenation reactions. Experiments are in progress to check whether the rate of hydrogenation can be correlated with the concentrations of these active, mobile hydrogen ada t oms. The ratio for the relative H and D atom spin adduct con- centrations (15/85 = 0.18) is a measure of the equilibrium constant for the following equilibrium at the palladium and alumina surfaces, assuming the H moieties are weakly bound. ---Pd---D+ H-O-Al---=---Pd---H + D-0-Al---We make the reasonable assumption that the surface H and D atoms are transferred to PBN with equal facility. The dif- ference in zero-point energies calculated by using vpdWH= ~1850 cm-' and v ~ =-3650 cm-' and, assuming2, that vD = vJJ2 is 5.2 kJ mol-' which is higher than the value of 4.3 kJ mol-' estimated from the ratio of spin adduct popu- lations.The difference is close to the experimental error but is in the direction expected for a more loosely bound H adatom reacting with PBN. Formation and Trapping of Alkyl and Cycloalkadienyl Radicals at a Pd Surface Ethene As already discussed and illustrated in Scheme 1, ethyl rad- icals have been suggested as important intermediates in ethene hydrogenation. The involvement of ethyl species in the hydrogenation of ethene was first established by Kemba1123 using deuterium, where evidence for the monodeuteriated species C,H,D(g) was obtained mass-spectrometrically.These ethyl species were assumed to be chemisorbed. More recently, Thomson and Webb24 concluded that only a weakly held C2H4 species was hydrogenated, the more strongly che- occurrence of such spillover is still a topic of controver~y'~*~~ and is difficult to trace experimentally, as illustrated by NMR studies of the Euro Pt/SiO, catalyst. These were first inter- preted to show the occurrence of spillover, and then reinter- preted to demonstrate its absence.21 We envisage the hydrogen spillover of hydrogen atoms to occur by a 'Grotthus type' chain or 'knock-ony effect as illustrated in Scheme 4. misorbed ethene being inactive. Since this paper was sub- mitted, Bowker et a1.,25on the basis of a mass-spectrometric temperature-programmed desorption approach have further discussed the role of weakly adsorbed ethene in catalytic hydrogenation at Rh(ll1) surfaces.Here we report a demons- tration of the participation of ethyl radicals as intermediates using the spin-trapping technique. Fig. 4 EPR spectrum of a PBN solution after mixture with a Pd catalyst which had been exposed at 25°C to a 50: 50 mixture of H,-C2H, for 10 min The pretreated catalyst was exposed at 298 K to a 50 :50 mixture of H, and C2H4 for 10 min before the PBN spin trap was added and the resulting solutions were then manipulated in the way already described. Fig. 4 shows the spectrum of the resulting solution which exhibits no hydrogen atom PBN adduct transitions, indicating that the ethene has reduced the surface hydrogen atom concentration to undetectable levels.The triplet of triplets spectrum of Fig. 2(a) is now replaced by a triplet of doublets with uN = 1.47 mT and u,(l) = 0.23 mT which arise from a radical spin adduct, B (Scheme 5), rather than a hydrogen atom adduct. The nitrogen and hydrogen hyperfine parameters are consistent with X' being an alkyl radical, although an alkoxy radical or alkyl peroxyl spin adduct cannot be excluded because of the similarities of the hyperfine parameters for these three adducts. spin addud B Scheme 5 In order to distinguish between these two possibilities, we used the nitroso spin trap BNB which is much more defini- tive in this regard since a-proton hyperfine parameters from the parent radical can usually be observed.On repeating the experiment with 50 :50 H2-C,H, mixtures, but using BNB solutions, the spectrum, shown in Fig. 5, was observed. The main transitions show unambiguously that X' has a CH; ter- minal group. They can be accurately simulated by the follow- ing parameters; g = 2.007 & 0.0005,a, = 1.31 mT, a,(2H) = 1.81 mT, ad2H) = 0.08 mT which are close to those for the ethyl BNB adduct.26 We have, therefore, conclusive proof of alkyl (probably ethyl) radical formation on the palladium surface, and have observed the occurrence of surface hydro- gen atom addition to ethene to give ethyl or higher C,H;,+ 3 2) radicals.The additional triplet of quartet fea- tures, nsrked R in Fig. 5, arise from a radical addition to the oxygeri md (1: the BNB as indicated by an upward shift of *r.L;-cemz of riie spectrum and, hence, a downward shift in ~-k:a-to g = 2,0037. Such oxygen addition only occurs for ._ ~~ic-~k~~.4. -28 indicating that oligomeric and/or poly- i-w-3: -xii-:dz are being formed and trapped in addition to J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 R 9 Fig. 5 EPR spectrum at 25°C of a BNB solution after mixing with a Pd catalyst exposed to a mixture of 50 :50 H,-C2H, at 1 bar for 10 min at 25 "C. Lines A are from the ethyl spin adduct with addition at the nitrogen end of the nitroso groups and lines R are from a secondary alkyl radical RCHR attaching at the oxygen of the nitroso group. small primary radicals such as ethyl.The quartet of lines indicate three equivalent hydrogens; two are probably the rn-protons of the benzene ring which usually have hyperfine parameters of 0.186 mT(a, = 1.0 mT), while the third is a single proton on the attacking radical, i.e. a secondary radical, R-CH-R. This constitutes good evidence for the occurrence of either an intermolecular hydrogen transfer between an oligomeric radical and an adjacent preformed macromolecule or an intramolecular hydrogen transfer or back biting by a large oligomeric or polymeric primary alkyl radical, Scheme 6. Attack at the oxygen of BNB, indicates a Scheme 6 large secondary or tertiary radical and is consistent with the presence of a large and bulky oligomeric or polymeric radical on the palladium surface.Fig. 6 shows the spectrum when a 80 : 20 H,-C,H4 mixture was used. The decrease in the R components is evidence for this assignment to an oligomeric or polymeric component which should decrease with the Fig. 6 EPR spectrum of a BNB solution after mixing with a Pd catalyst exposed to a 80 :20 H,-C2H, mixture J. CHEM.SOC. FARADAY TRANS., 1994, VOL. 90 decrease in ethene levels. Further experiments are now in progress to study the reactions of C2D4, CH2CD, and dial- kenes on a range of catalysts. Benzene Because the addition of hydrogen atoms to benzene is less efficient than to alkene~,~’,~~ we increased the benzene con- centration substantially by using it as the solvent for BNB.The procedure was as described but a 0.01 mol dm-3 BNB solution in benzene was mixed with the palladium catalyst after it had been exposed to pure H, for 10 min. The resulting spin adduct spectrum, shown in Fig. 7(a), again shows the complete loss of the hydrogen atom adduct tran- sitions and their replacement by a well resolved major sextet of pentets arising from the following interactions: a&) = 1.22 mT, 43) = 1.22 mT and 444) = 0.07 mT. Fig. 7(b) shows a simulation of the spectrum based on these param- eters. We assign the experimental spectrum to the cyclo- hexadienyl spin adduct (C, Scheme 7) resulting from the spin adduct C Scheme 7 two-step sequence of a surface hydrogen atom reaction with benzene to give the 2,3-cyclohexadien- 1-yl radical which is then scavenged by the BNB to give adduct C.The large sextet proton hyperfine parameters arise from the three protons at the 1 and 2 positions of the cyclohexadienyl moiety and the small proton quintet hyperfine parameters from the two protons at the 4 and 5 positions of the cyclo- hexadienyl and the two at the 3 and 5 positions of the 1,3,5- tri-tert-butylbenzene. This is the first report of the EPR spec-trum of the cyclohexadienyl radical adduct of BNB but the 1 mT w Fig. 7 (a) EPR spectrum of a BNB solution in benzene after addi- tion to a Pd catalyst exposed to H,at 1 mbar for 10 min at room temperature; (b)simulation of spectrum using the parameters given in the text parameters are those expected by comparison with the allyl spin adduct reported by Terabe and Konaka,26 uH (from one CH,) = 1.81 mT; aN(1) = 1.34 mT and a,(2 m-H) = 0.08 mT.No coupling is observed from the central proton of the allyl entity or those of the other CH2 group which were equivalent in the parent allyl radical, i.e. we do not have allyl x-type bonding across the two terminal carbons as observed in many transition-metal allyl complexes. These assignments will be confirmed by using suitably deuteriated benzenes and a combination of H2 and D, experiments. Again, the spin trapping technique has revealed, for the first time, the presence of precursors en route to the complete hydrogenation of benzene to cyclohexane. We are presently conducting a comprehensive study of a range of substrates and catalysts.Note that the samples were ‘as used’ in industrial plants and examined without prior treatment other than the normal activation procedures. Conclusions In the present study it is clear that at palladium surfaces only weakly chemisorbed hydrogen interacts with the spin trap and is therefore observed by EPR. This raises the question as to whether the spin trap intercepts the hydrogen adatom, H(s), just subsequent to the cleavage of the dihydrogen bond but prior to it being chemisorbed or whether the spin trap interacts with the weakly chemisorbed hydrogen adatom, H(a): 9H2(g)*H(s) H(a)-+ With increasing coverage the heat of H(a), a chemisorption will decrease and only under these conditions would we anticipate the reversible step to be significant (see above).It is therefore either the H(s) species or the weakly chemisorbed hydrogen, H(a), that we observe by EPR and not the strongly chemisorbed hydrogen, H(a). An analogous argument may also apply to the hydro- genation of ethene: C2H4(g) C2H4(a); strong chemisorbed and inactive in hydrogenation C2H4(g)-’ C2H%s); weakly adsorbed at high coverage and active C,HX(s) + H(s) --* C,HT(s); two dimensional gas reaction : first step of hydrogenation C2HZ(s)+ H(s)+C,H,(g); desorption of ethane This mechanism is essentially that of Kemball,23 and Thomson and Webb.24 The C2H, spin trap species has been observed unequivocally in the present EPR study.If we accept that the spin trap is specific to radical-type species only, then the present observations support a mechanism involving the participation of ethyl radicals present in a two- dimensional gas-like overlayer composed of weakly adsorbed ethene C,HX(s) and weakly adsorbed H adatoms H(s), in equilibrium with H2(g). Both H(s) and C,H;(s) are observed by EPR. Analogous free radical chemistry is observed in the catalytic hydrogenation of benzene. We are grateful to ICI Katalco for their generous support of this work. H.A.E. thanks the SERC for the award of a CASE studentship and ICI Katalco for their sponsorship. We also thank Professor M. S. Spencer for illuminating discussions and comments.3 346 J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 References 17 ICI, UK Pat., 7940086, 1979. 1 J. H. Lunsford, Adv. Catal., 1972,22,265. 2 L. Kevan and M. Naryana, Intrazeolite Chemistry, ed. G. D. Stucky and F. G. Dwyer, ACS Symp. Ser. 218, 1984, p. 283. 3 P. H. Kasai and R. J. Bishop Jr., Zeolite Chemistry and Cataly- sis, ACS Monograph 171, ed. J. A. Rabo, 1977, p. 350. 4 C. J. Rhodes and B. C. Webster, J. Chem. SOC., Faruday Trans., 1993,89,1283. 5 M. Che and A. J. Tench, Adv. Catal., 1983,32, 1. 6 M. Che and E. Giamello, Catalysis Characterisation: Physical Techniques for Solid Materials, ed. B. Imelik and J. C. Vedrine, Plenum Press, New York, 1994, p. 131. 7 D. J. Driscoll, W. Martin, J. X. Wang and J. H. Lunsford, Adsorption and Catalysis on Oxide Surjiaces, ed.M. Che and G. C. Bond, Elsevier, Amsterdam, 1965, p. 403. 8 A. N. R. Bos and K. R. Westerterp, Chem. Eng. Process., 1993, 32, 1. 9 S. Le Viness, V. Nasr, A. H. Weiss, Z. Schay and L. Guczi, J. Mol. Catal., 1984, 25, 131. 10 L. Z. Gra and K. G. Kho, Kinet. Catal., 1988,29,381. 11 F. Zaera and G. A. Somorjai, J. Am. Chem. SOC.,1984,106,2288. 12 I. V. Reid, T. Azuma and E. Roduner, Nature (London), 1990, 345, 328. 13 E. Roduner, Chem. SOC. Rev., 1993,22,337. 14 E. Roduner, Phys. Rev. B, in the press. 15 E. G. Janzen in Free Radicals in Biology, ed. W. A. Pryor, Aca- demic Press. New York. 1980. vol. 4. D. 115. 16 T. Matsuzaki, T. Uda,’A. Kkusak;, G. W. Keulks and R. F. Howe, J. Am. Chem. SOC., 1980,102,7511.18 0. Augusto, P. R. Ortiz de Montellano and A. Quintanilha, Biochem. Biophys. Res. Commun., 1981,101,1324. 19 M. S. Spencer, Appl. Catal., 1989,55, N23. 20 New Aspects of Spillover Eflect in Catalysis, Stud. Surf. Sci. Catal., ed.T. Inui, Elsevier, Amsterdam, 1993, vol. 77. 21 M. A. Chesters, A. Dolan, D. Lennon, D. J. Williamson and K. J. Packer, J. Chem. SOC., Faruday Trans., 1990, 86, 3491; M. A. Chesters, D. Lennon, H. Viner and K. J. Packer, J. Chem. SOC.,Faruday Trans., submitted. 22 L. Lelander and W. H. Saunders Jr., Reaction Rates of Isotopic Molecules, Krieger Publishing Co., Malabar, 1991, 109. 23 C. Kemball, Proc. Chem. SOC., 1954,30,1164. 24 S. J. Thomson and G. Webb, J. Chem. SOC., Chem. Commun., 1976,526. 25 M. Bowker, J. L. Gland, R. W. Joyner, Y. Li, M. M. Slin’ko and R.Whyman, Catal. Lett., 1994,25,293. 26 S. Terabe and R. Konaka, J. Chem. SOC., Perkin Trans. 2, 1972, 369. 27 B. Mile, C. C. Rowlands, P. D. Sillman and M. Fildes, J. Chem. SOC.,Perkin Trans. 2, 1992, 1431. 28 B. Mile, C. C. Rowlands, P. D. Sillman and A. Holmes, J. Chem. SOC.,Perkin Trans. 2, 1993,2141. 29 J. E. Bennett and B. Mile, J. Chem. SOC., Faruday Trans. 1, 1973, 69, 1398. 30 C. H. Bamford and C. F. H. Tipper, Comprehensive Chemical Kinetics, 1969,3,97. Paper 4102235A; Received 14th April, 1994