Acid and Polarity Effects in Benzene Photoreactions with Alkenes and Alkynes

D. Bryce-Smith*, A. Gilbert, N. Al-Jalal, R. R. Deshpande, J. Grzonka, M. A. Hems, and P. Yianni Chemistry Department, University of Reading, Whiteknights, Reading, Berkshire, RG6 2 AD Z. Naturforsch. 38b, 1101-1112 (1983); received May 2, 1983

Photoreactions, Benzene, Alkenes, Alkynes, Acid Effects

The effects of polar solvents and proton donors on various photoreactions of the benzene ring with electron-acceptor alkenes and alkynes are described and discussed in relation to the various electronic excitation mechanisms involved. Proton donors prove valuable as mechanistic probes for polar intermediates in such processes, and can in some systems both initiate and divert reaction pathways.

In 1970 we observed that the course of the photo- anhydride and certain maleimides, a second mole- chemical cycloadditions of certain ethylenic and cule may add thermally to give the remarkably acetylenic compounds to benzene, earlier reported stable 2:1 adducts (2) and (3) [1, 12]. The ortho- by us [1, 2], was profoundly altered in the presence adducts of acetylenes (4) undergo valence isomeri- of proton donors, leading to a type of photochemical zation to give cyclo-octatetraenes; e.g. dimethyl Friedel-Crafts reaction [3]. Since then, we have acetylenedicarboxylate gives exclusively the isomer described, mostly in preliminary form, numerous (5) having a single bond between the substituents, further examples of acid catalysis in the photo- apparently for steric reasons [2], Addition mode (a) reactions of benzene, viz. with amines [4], ethers shows the greatest sensitivity to proton donors and [5], a ketone [6], a carboxylic acid [6], and hexa- polar solvents, and discussion of these effects forms fluorobenzene [7]. Our objective in this present the main part of the present paper. paper is to bring together the previous preliminary Addition mode (b) is observed with alkenes having reports on acid catalysis in the photoreactions of an ionization potential similar to that of benzene; ethylenic and acetylenic compounds with benzene e.g. cis-cyclooctene gives an approximately 4:1 in conjunction with unpublished recent findings, mixture of the meta adducts (6) and (7) with the and to provide a mechanistic rationale together preparatively useful combined quantum yield of with relevant experimental details. 0.38 [13]. No example of analogous meta photo- But before we proceed to consider the acid-cata- addition of an acetylene has yet been convincingly lysed processes, it is desirable to summarise the main established. types of reaction that have been shown to occur in Addition mode (c) is observed less frequently, the absence of acids. Ethylenes and/or acetylenes though para photoadducts (8) have occasionally photoadd to benzene in the following four main been reported as minor products, e.g. from cis-but- ways: (a) ortho cycloaddition; (b) meto-cycloaddi- 2-ene [14] and cyclobutene [15] (but not cyclopro- tion; (c) £>ara-cycloaddition; and (d) para-ene addi- pene, which adds meta [13]). Only allenes have so tion [8, 9]. The mode of addition varies with the far been found to undergo para cycloaddition as the addend, and in some cases more than one pathway predominant mode; e.g. allene and benzene mainly may be followed. In the case of (a), the adducts (1) give the adduct (9), with a minor proportion of the of ethylenes may be obtained as such when the meta adduct (10) [16]. ethylenic addend is not a strong thermal dienophile, Ene-addition, mode (d), is only observed with e.g. acrylonitrile [10], 2,3-dihydropyran [11]. When relatively electron-donor alkenes having the group it is a strong dienophile, as in the case of maleic Me2C=, and is the predominant mode in the photoaddition of tetramethylethylene to benzene, giving the adduct (11) [13]. * Reprint requests to Prof. D. Bryce-Smith. The orbital symmetry aspects of these processes 0340-5087/83/0900-1101/$ 01.00/0 have been analysed by one of us [17], and the con- 1102 D. Bryce-Smith et al. • Acid and Polarity Effects in Benzene Photoreactions elusions reached have been found to have good pre- terion (13) or a related dipolar exciplex species [13], dictive value [8]. It is particularly relevant to the With CH3OD as solvent, we have shown that D is mechanism of acid effects that the analysis shows incorporated at the 4-position of the ring in (11). the three new bonds formed in meta cycloadditions We come now to the ortho cycloadditions, mode to arise in a partially or fully concerted manner (a), which are by far the most interesting in their from Si benzene plus So alkene, whereas concerted susceptibility to polar factors. ortho cycloaddition can only arise from So benzene As mentioned above, orbital symmetry con- plus Si alkene, or via charge transfer either to or siderations suggest that homopolar ortho cyclo- from the arene. additions of So ethylene or acetylene to Si benzene Not surprisingly, we and others have observed are forbidden as concerted processes. The only that the largely homopolar concerted meta addition, allowed ortho cycloadditions involve Si ethylene or mode (b), usually shows little or no sensitivity acetylene plus So benzene. Charge-transfer excita- either to the presence of proton donors or to solvent tion is also fully "allowed', though polar effects pre- polarity [13, 18]. The only exception to this so far dominate in practice. These predictions have proved noted is in the photoaddition of ethyl vinyl ether to valid to a remarkable degree, and no clear excep- anisole, giving adducts (12), which is markedly pro- tions have yet been discovered. Thus ortho cyclo- moted by solvent acetonitrile in comparison with additions, when they occur normally in the pre- cyclohexane [19]. There is however no diversion of sence of a excess of benzene, have involved (i) the reaction pathway. The only mechanistic con- ethylenic or acetylenic addends having an Si state clusion we have yet reached on this apparent below Si benzene (in practice having significant anomaly is that the transition state(s) for formation absorption near to or beyond the benzene solvent of adducts (12) is or are markedly more polar than front beyond ca. 280 nm but not showing charge- the system at earlier stages along the reaction co- transfer absorption in benzene at wavelengths ordinate, and more polar than the corresponding >280nm, (ii) charge-transfer excitation at wave- transition states for formation of analogous adducts lengths > 280 nm, (iii) systems where the reactants where one or both -OEt groups are absent. The comprise a marked donor acceptor pair (in practice most odd feature of this anomalous case is that the having an ionisation potential difference > ca. 1 eV) sensitivity to solvent polarity is manifest in a but show no charge-transfer or other absorption reaction which is expected a priori to be essentially beyond the benzene solvent front. homopolar in view of the closely similar ionization N-(w-Butyl)maleimide and dimethyl acetylene- potentials of anisole and ethyl vinyl ether (8.54 and dicarboxylate could in principle fall either in 8.6 eV respectively). The observed meta rather than category (i) or category (iii): the ortho adduct ortho cycloaddition is however in accord with the (4:R=R'=C02Me) from the latter can be trapped orbital symmetry analysis [17]. with tetracyanoethylene, but otherwise undergoes As for the rather rare para cycloaddition, mode rapid valence isomerization to the corresponding (c), no studies of solvent polarity or proton donor cyclo-octatetrane (5) [20] (which does not appear effects have yet been reported, and we ourselves to undergo the expected bond exchange to the con- have not yet investigated these matters. former/tautomer (14) to any significant extent [21]). Maleimide and its N-alkyl derivatives give the 2:1 On the other hand, the para ene-addition, mode adducts (3) [22], via the 1:1 adducts (15): the latter (d), is markedly promoted by proton donor solvents can be trapped by tetracyanoethylene (which does such as methanol, but not by aprotic polar solvents not itself photoadd to benzene) and maleic anhy- such as acetonitrile. As we have discussed our studies on this process in some detail elsewhere [13], dride as the 1:1:1 adducts (16) and (17) respectiv- it will be sufficient for the present purposes to point ely [23]. out that para-ene photoaddition is non-stereospe- The maleimides and the acetylene photoadd to cific and occurs by a non-concerted mechanism, in benzene on irradiation at wavelengths >290nm, contrast with the concerted mechanism normally so direct photoexcitation of benzene can be ruled attributed to conventional thermal ene-additions. out. Furthermore, although both the maleimides We have suggested the intermediacy of the zwit- and dimethyl acetylenedicarboxvlate have electron 1103 D. Bryce-Smith et al. • Acid and Polarity Effects in Benzene Photoreactions acceptor properties relative to benzene, the ultra- to acid implies the substantial absence of a polar violet spectra of these in benzene show no signs of intermediate capable of being intercepted by tri- significant charge-transfer absorption at wave- fluoroacetic acid, but the mechanistic significance lengths beyond the benzene solvent front at ca. of this observation is better appreciated by com- 280 nm: the absorption due to the maleimides and parison with the effects, described below, of acids the acetylene above 280 nm is virtually identical in on the seemingly analogous photoaddition of maleic benzene, cyclohexane, and chloroform, and extends anhydride to benzene. almost to the visible region, though the acetylene In contrast, irradiation of dimethyl acetylene- absorbs much more weakly than the maleimides. dicarboxylate in benzene in the presence of tri- Direct photoexcitation of the addends therefore fluoroacetic acid gave the results shown in Table I. clearly occurs. The most obvious mechanism would Separate experiments showed that dimethyl phtha- involve a symmetry allowed concerted ortho cyclo- late was the only product formed by irradiation of addition of the Si addends to So benzene, via a rela- dimethyl cyclo-octatetraene 1,2-dicarboxylate in tively non-polar cyclic transition state. This ap- benzene, and that this photoreaction is.markedly pears to operate in the case of maleimides, but not acid catalysed. with dimethyl acetylenedicarboxylate, as revealed These results have several very interesting by the differing effects of acids on these processes. features. Firstly, use of the strong acid appears Most of our mechanistic studies of the maleimide essential: methanol was ineffective. Secondty, tri- additions have employed the N-(w-butyl) derivative, fluoroacetic acid only affected the relative yields of which was selected for reasons of experimental con- products, not the overall combined yields which venience. Maleimide and all its n-alkyl and N-aral- were closely similar in all three experiments. kyl derivatives so far examined appear to photoadd Thirdly, since the reduced formation of the cyclo- similarly to benzene. octatetrane with increased proportions of acid was Thus the photoadditions of maleimide and N-bu- only partly accounted for by the increased pro- tylmaleimide to benzene at wavelengths > 290 nm duction of dimethyl phthalate, presumably by the are relatively insensitive to the presence of trifluoro- process shown in Scheme 1,* we conclude that the acetic acid. In the presence of 0.825 molar CF3CO2H, the yield of the 2:1 adduct (3:R=Bun) was repro- ^^ C02 Me C02Me H co Me ducibly reduced by about 20%, but no other products were detected. A particular search was made C02Me CO,Me H C0,Me for N-butylphenylsuccinimide, but no trace was 18 22 Scheme 1 found. The photoaddition of maleimide itself was similarly insensitive to acid, though the low solub- dimethyl phenylfumarate and -maleate arose from ility of maleimide in benzene renders this system a precursor of the cyclo-octatetraene, and at its less convenient for study. The relative insensitivity expense. For as noted above, these products are not formed from the cyclo-octatetraene under the Table I. Effect of trifluoroacetic concentration on the experimental conditions employed. In principle, relative percentage yields of products formed during processes of the type depicted in Scheme 2 may be 5-hour irradiation of a 2% v/v solution of dimethyl acetylenedicarboxylate in benzene. considered. It seems reasonably certain that the bicyclic Product percentage yield ortho adduct (18) depicted as formed by path B is a Trifluoro- Dimethyl Dimethyl Dimethyl Dimethyl acetic acid phthalate phenyl- cyclo- phenyl- true intermediate since this can be trapped by concentra- fumarate octa- maleate tetracyanoethylene, giving (19), under conditions -1 tion mol l tetraene- where the cyclo-octatetrane (5) does not react 1,2-di- carbox- thermally with tetracyanoethylene. As previously ylate noted, its formation in a concerted process of type

0 3% 1% 96% trace 0.27 4% 12% 82% 1% * One and possibly both of these photoreactions are 1.90 18% 39% 39% 4% acid-catalysed. The formation of acetylene is presumed. 1104 D. Bryce-Smith et al. • Acid and Polarity Effects in Benzene Photoreactions

CO,Me CO,Me pare the case of w-butylmaleimide already mentioned, and contrast the case of maleic anhydride and benzene discussed below. Fourthly, only in the effects of acid does addition of the acetylene to benzene differ from that of maleimides. Yet since the acid-insensitivity of maleimide additions seems to rule out electron-transfer analogous to path (A), the acetylene would be even less likely to follow S, Dimethyl this route, for it appears to be a weaker electron- Acetylene CO,Me dicarboxylate JQ acceptor than ft-butylmaleimide to judge from ultra- + SnBenzene violet/visible absorption spectra of mixtures with anisole and dimethoxybenzenes as donors. More- over, the maleimide intermediate (15) analogous to (18) would not be expected to follow an acid-cata- CO,Me lysed pathway analogous to D since the imide nitro- Scheme 1 gen rather than carboxyl oxygen would preferen-

Dimethyl phthalate tially undergo protonation.

Scheme 2 We therefore conclude that the products shown in Table I arise by paths B, C, D, and E. Although path A cannot be totally ruled out as a minor con- B is symmetry allowed; and as we have shown that tributor, none of the findings require it, and the photoaddition of dimethyl acetylenedicarboxylate above considerations argue against it. In other to benzene occurs at wavelengths > 290 nm where words, the mechanism appears to be of type (i) only the acetylene absorbs, process B privides a rather than type (iii). more economical route to the intermediate (18) than the alternative electron-transfer route A fol- Returning to Scheme 1, it may be noted that the lowed by ring-closure. In the absence of acid, all the cyclo octatetraene (5) appears not to undergo photo- products shown in Table I may reasonably be chemical reversion to (18) since irradiation of (5) in accounted for by an initial symmetry-allowed the presence of tetracyanoethylene does not give cyclisation step B, followed mainly by E to give the the adduct (19). Evidently (22) is the preferred pro- cyclo-octatetraene (5) and thence dimethyl phtha- duct. This observation is consistent with the ob- late as in Scheme 1. The traces of maleate (20) and servation that irradiation of the cyclo-octatetraene fumarate (21) could arise by the photochemical (5) in benzene in the presence or absence of tri- 1,3-hydrogen shifts shown in step C: the relative fluoroacetic acid gives only dimethyl phthalate and inefficiency could simply reflect the difficulty in no trace of the maleate (20) or fumarate (21). The exciting (18) at low concentrations in an excess of evident preferred tendency of the cyclo-octatetraene the reactants. (5) to undergo valence-bond isomerization to (22) rather than (18) is in accord with our earlier ob- In the presence of acid, the increased formation servation that (5) exists exclusively in the indicated of maleate (20) and fumarate (21) could in principle form, not (14) [21]. arise either via path A or path B followed by D. Of these, we prefer B followed by I) for the following We come now to cycloadditions of type (ii), reasons. Firstly, it utilises the allowed concerted namely those that occur via initial charge-transfer pathway B. Secondly, it explains the apparent need photoexcitation between a donor-acceptor pair. to excite the acetylene rather than the benzene. These show some particularly interesting acid Thirdly, if the indicated zwitterionic intermediate effects that shed considerable light on the mechan- were really formed through electron-transfer as in isms involved. The photoaddition of maleic an- path A, one might have expected some sign of hydride to benzene, giving the remarkably stable charge-transfer absorption in mixtures of the ace- 2:1 adduct (2), is by far the most extensively tylene and benzene, yet the absorptions were strictly studied reaction falling into this category, and was additive over the observable range > 260 nm: com- in fact the first known example of photoaddition to 1105 D. Bryce-Smith et al. • Acid and Polarity Effects in Benzene Photoreactions benzene [1] : Professor Schenck discovered that this mal) yield of the 2:1 adduct (2). The phenylsuc- reaction could also be photosensitized by benzo- cinic anhydride may reasonably be considered phenone [24], though as we shall show, the me- to arise following protonation of the zwitter-ion chanism is different in this case even though the (23) as in Scheme 3. adduct formed is the same as in the unsensitized H 0 H ? Ph , process. benzene hv H* Firstly, we consider the chemistry of the un- maleic e-transfer" anhydride OH sensitised addition before coming to the sensitised (weakly complexed) 0 process and the photoexcitation mechanisms. Although we had originally [1] suggested the inter - ® maleic anhydride mediacy of an or/Äo-adduct of maleic anhydride and 0 benzene, the failure of tetracyanoethylene to inter- Scheme 3 cept this intermediate led us subsequently [12] to reject it in favour of the zwitter-ion (23). Later, This observation is of interest on several counts. Hartmann et al. synthesised the or^o-adduct (24) Firstly, it shows that proton donors can provide an by an independent route and rather surprisingly effective mechanistic probe to test for polar inter- showed that, under reaction conditions analogous mediates formed by charge-transfer excitation. Se- to those used for the photo-process, it is far less condly, it demonstrates a clear difference between thermally reactive to addition of tetracyanoethylene the mechanisms of photoaddition of maleic an- than to maleic anhydride [25]. We were then able to hydride and N-(n-butyl) maleimide to benzene, and confirm the intermediacy of (24) in the photoaddi- one that is consistent with the observed differences tion of maleic anhydride to benzene by trapping it between the initial photoexcitation steps - the very efficiently with the stronger dienophile N- former charge-transfer, and the latter direct excita- phenylmaleimide as the 1:1:1 adduct (25) [26]: tion of the maleimide. Thirdly, it provides a rare none of the 2:1 adduct (2 a) was obtained. N-Phenyl- example of photoelectrophilic substitution in the maleimide itself is almost totally unreactive photo- benzene ring, and was the first case of photochemical chemically towards benzene. It is therefore clear Friedel-Crafts alkylation. Incidentally, the cor- that the 2:1 adduct (2) arises via thermal Diels- responding thermal reaction between benzene and Alder addition of a second molecule of maleic an- maleic anhydride catalysed by aluminium chloride hydride to the intermediate photo-adduct (24). follows an acylation pathway, giving benzoylacrylic Concerning the excitation process, it was shown acid; so the thermal and photochemical processes [27] that addition occurred on irradiation within are complementary. the benzene-maleic anhydride charge-transfer band We come now to processes of the foregoing type [28] at wavelengths > 295 nm. This observation led that appear to follow a triplet pathway. As pre- us to postulate the initial formation of the zwitter- viously noted, Schenck and Steinmetz found that ion (23) which, as explained above, we at one time the photoaddition of maleic anhydride to benzene were led to suggest as the species that underwent could be effectively sensitized by benzophenone [24]. addition of the second molecule of maleic anhydride. Hardham and Hammond subsequently reported The effect of acids on the photoprocess was then that triplet benzophenone is effectively quenched studied to test the hypothesis that zwitter-ion (23) by maleic anhydride plus benzene, but not by was a precursor of the ortho-adduct (24). maleic anhydride or benzene alone [29]. These ob- Irradiation of solutions of maleic anhydride in servations clearly show that photosensitization of benzene in the presence of trifluoroacetic acid as the addition to benzene involves the following ini- proton donor gave none of the normal 2:1 adduct tial steps: Benzophenone (Ti) + Benzene-Maleic (2): contrast N-(n-butyl) maleimide. Instead, small Anhydride Complex (So) -> Benzophenone (So + amounts of phenylsuccinic anhydride were pro- Benzene-Maleic Anhydride Complex (Ti). duced. With acetic acid as a weaker proton donor, We now consider the nature of this triplet com- phenylsuccinic anhydride was also obtained to- plex. While some degree of charge-transfer within gether with a greatly reduced (to ca. 30% of nor- it is to be expected, the effectively complete charge- 1106 D. Bryce-Smith et al. • Acid and Polarity Effects in Benzene Photoreactions transfer involved in formation of the zwitter-ion (23) and Hammond [29] that the normal unsensitized would obviously not be possible in the absence of and ketone-sensitized photoadditions of maleic an- spin-pairing. For this reason, we propose the hydride to benzene all involve the same triplet com- polarised diradical structure (26) for the triplet plex species. Our conclusion is that the unsensitized complex between benzene and maleic anhydride. photoaddition in the absence of a heavy atom sol- The ortho adduct (24) is presumed to be formed by vent involves only excited singlet intermediates intramolecular cyclisation of this complex, and this whereas triplet intermediates are involved in the then undergoes Diels-Alder addition of maleic an- presence of a heavy atom solvent, or with benzo- hydride, or of N-phenylmaleimide when this is used phenone photosensitization. as a trap, as in the unsensitized process. This We now come to the case of 2?-benzoquinone as a mechanistic scheme therefore implies that the key photoaddend to benzene. Despite the similarity be- difference between the sensitized and unsensitized tween p-benzophenone and maleic anhydride as processes lies in the different degrees of polarity be- thermal dienophiles, irradiation of solutions of tween the corresponding intermediate complexes(26) p-benzophenone in benzene leads to no observable and (23). reaction. Although these solutions do show some Confirmation of this has been provided by the broad charge-transfer absorption tailing out well relative unsusceptibility of the benzophenone-sen- beyond the benzene solvent front at ca. 275 nm [28], sitized photoaddition to trifluoroacetic acid, in con- 2max. for this lies within the benzene absorption trast with the marked susceptibility of the un- range and cannot readily be observed. In benzene, sensitized process that we have already described. the n-n* absorption over the range 400-490 nm Specifically, 0.825 M trifluoroacetic acid decreased shows considerable broadening of the peaks, but the rate of formation of photoadduct (2) by about the positions ofAmaxShow only hypsochromic shifts 50%, but gave no trace of phenylsuccinic anhydride. of 1-2 nm. In the presence of trifluoroacetic acid The reason for the decreased rate is not well under- (0.415 M) the n-n* absorption in cyclohexane, and stood, but might perhaps be connected with pro- the charge-transfer absorption in benzene, show tonation of ground-state and/or excited benzo- pronounced bathochromic shifts, especially in the phenone. Be that as it may, there can be no doubt latter case, and there are alterations in intensity that intermediate (26) would be much less suscept- but not Amax, for the n-n* peaks. ible to protonation than (23), and therefore much The chemical effect of trifluoroacetic acid was less likely to give rise to phenylsuccinic anhydride. examined in the hope of trapping a zwitter-ionic Further confirmation of the distinction between intermediate, as in the case of maleic anhydride as the triplet and excited singlet complexes (26) and addend. Control experiments showed that no 'dark' (23) respectively is provided by the demonstration reaction occurred in the presence of trifluoroacetic that diverting the unsensitised process into a triplet acid. Irradiation of a solution of ^-benzoquinone in pathway by the use of dibromomethane as a 'heavy benzene containing 0.415 M trifluoroacetic acid gave atom' solvent renders it relatively insensitive to the 1:1 adduct 4-phenoxyphenol (27) in isolated trifluoroacetic acid. No trace of phenylsuccinic an- yields of up to ca. 50%. The reaction occurred either hydride could be detected under these conditions, on irradiation at wavelengths > 290 nm, or with and the rate of formation of the 2:1 adduct (2) was visible light alone. The latter observation shows reduced by only ca. 35-40% in contrast with the that the reaction involves initial n^n* rather than total inhibition observed in the absence of the heavy the direct charge-transfer excitation involved in the atom solvent. We conclude that the unsensitized corresponding photoaddition of maleic anhydride. reaction in the heavy atom solvent proceeds by p-Benzoquinone is however highly complexed in initial direct photoexcitation of the ground-state benzene solution [28], and this phenomenon would benzene-maleic anhydride complex followed by tend strongly to facilitate any charge-transfer from intersystem crossing to give the same polarized benzene to the excited ^-benzoquinone moiety of a diradical species (26) as is formed by benzophenone complex. Scheme 4 is in accordance with these ob- sensitization in the absence of a heavy atom solvent. servations and considerations, though it should be These findings exclude the proposal by Hardham noted that the weak ground-state complex irself 1107 D. Bryce-Smith et al. • Acid and Polarity Effects in Benzene Photoreactions undergoes some protonation in the presence of tri- envisaged as capable of competing effectively with fluoroacetic acid. electron-transfer. Furthermore, in the related case of maleic anhydride, benzene, and trifluoroacetic acid, a mechanism analogous to that in Scheme 5 °Oo • Q—°0-° • O could certainly account for the formation of phenyl-

S0 so si so succinic anhydride, but would not explain the v ^ ' s ^ ' strong inhibition of cycloaddition to give the adduct weak ground-state . complex je-transfer (2) via (24). On balance therefore, we are inclined to favour the mechanism depicted in Scheme 4, though we would not at this stage completely rule out at least a contribution from that in Scheme 5.

27 (and other canonical for.Tis) In conclusion, wre consider acid effects in donor- 28 acceptor systems of type (iii) where there is no

Scheme U charge-transfer or other absorption beyond the arene solvent front, i.e. where only the arene is We presume that the apparent failure to react in the excited, and undergoes only the ortho mode of cyclo- absence of acid probably results from ready dissoci- addition. Acrylonitrile and its derivatives as ad- ation of the zwitter-ion (28). An alternative expla- dends fall most clearly into this category. nation is that the initial photoaddition involves an It is interesting to compare our present findings excited protonated p-benzoquinone (29). This could on the acrvlonitrile-benzene system with those for wrell contain a strongly electrophilic oxygen centre acrylonitrile and the stronger donors naphthalene capable of reacting with benzene by electrophilic [30] and methoxybenzenes [31]. Irradiation (254nm) substitution to give (27), as in Scheme 5. We do not of solutions of acrylonitrile in benzene or benzene exclude this possibility, although at the concentra- and excess of cyclohexane gives a mixture of the tions employed, trifluoroacetic acid does not pro- ortho adducts (30a and b) in which the endo/exo duce any change in the absorption spectrum of p- ratio is ca. 5. The formation of these adducts is benzoquinone in benzene at wavelengths > 290 nm. strongly promoted in the more polar solvents This tends to argue against any substantial protona- acetonitrile and methanol, but the endo/exo ratio is tion in the ground-state, though protonation could unchanged, and no other 1:1 adducts could be of course occur after photoexcitation. Another ob- detected. In particular, a careful search revealed no servation that tends to weigh more heavily in trace of 1-phenyl-1-cyanoethane (31), or meta- favour of Scheme 4 than Scheme 5 is that the use cycloadducts, and under the conditions employed, the 2:1 photoadduct first described by Job and Littlehailes [10] was not formed to any significant extent.

The promoting effect of methanol in this reaction V l-H* Exciplex appears to be more a result of increased dielectric 29 27 constant than of acidity since irradiation of an Scheme 5 equimolar mixture of benzene and arylonitrile gave exactly the same result in the presence and absence of toluene in place of benzene gives only quin- of 10% of trifluoroacetic acid. The effects of solvent hydrone and bibenzyl as products of hydrogen atom polarity clearly indicate the involvement of an abstraction, and no trace of methyl-substituted exciplex or formally bonded intermediate having derivatives of (27): ^-xylene behaved similarly. Yet markedly polar character, whereas the intensitivity toluene and xylene would be expected to be more to proton donors implies that this intermediate is reactive than benzene towards electrophilic attack (a) insufficiently dipolar in character and/or (b) of by the excited species (29), whereas hydrogen atom too short a lifetime relative to ring closure for it to transfer from toluene within a weak excited com- be intercepted by protonation. If charge-transfer plex of the type depicted in Scheme 4 is readily from Si benzene to So acetonitrile were complete, 1108 D. Bryce-Smith et al. • Acid and Polarity Effects in Benzene Photoreactions the polar intermediate could be depicted as (32); substitution products analogous to (31). Results but this probably exaggerates the degree of charge from the use of naphthalene and methoxy benzenes separation actually involved since it is hard to as donors confirm this expectation. believe that a fully developed carbanionic centre Thus irradiation of naphthalene and acrylonitrile could excape protonation in the presence of tri- in a range of proton-donor solvents is reported to fluoroacetic acid. We therefore consider it likely give mixtures of the 1,2-cycloadducts (33) and (34) that the degree of dipolar character which develops with the cyanoethyl substitution products (35) and in the course of the ortho cycloaddition is sufficient (36) [30]. In general, and with only one minor to escape the restrictions imposed by orbital sym- exception (acetic acid), formation of the substitu- metry considerations on a fully concerted homo- tion products relative to the cycloadducts is polar ortho cycloaddition of Si benzene to an S0 strongly promoted by increased solvent acidity. In alkene [17], but insufficient to permit interception the non-donor solvents acetonitrile and dioxane, by a proton donor. little reaction occurred, and it would appear that It follows from these considerations that the use only traces of the cycloadducts were formed. In of a stronger electron donor than benzene might CH3OD or CH3CO2D, the labelled substitution pro- possibly give a dipolar intermediate having suffi- ducts (37) and (38) were produced, thereby con- ciently well-developed carbanionic character [cf. firming the proton-donor role of the solvent. The (32)] for it to be intercepted by proton donors, with authors attributed these findings to protonation of resultant diversion of the reaction pathway to give

15: x = n - r \ A. 2:x = o 24:x = 0 Cn 3:x=n-r (R= H , alkyl, benzyl, 2,6-dimethylphenyl)

ncn CO,Me cn; NC - NC- CO,Me

5; R = C02Me, r' =H 19

14: R= H, r'=C02Me 16 : X = N -nBu, R = r' =R2=R3=CN 17: X = N -nBu, R = R' = H, R2-R3 = - CO-0-C0- 25: X =0, R=r' = H, R2-R3 = -C0-N(Ph)-C0-

2 6:R = R=R =H, R'-R = -(CH2)6- 1 2 3 4 7 : R = R = R =H, R -R = -(CH2)6- 1 2 3 4 12 : R = OMe, R , R , R or R =0Et 0 6" I cr 26

Me Me

10 0" 30:a.R = CN, R1 = H 23 forms and other canonical b, R = H, R' =CN 1109 D. Bryce-Smith et al. • Acid and Polarity Effects in Benzene Photoreactions the exciplex (39), a precursor of the cycloadducts specifically bonded intermediate is perhaps more (33) and (34). The zwitter-ion (40) analogous to (32) consistent with production of a proportion of the may be considered either as an extreme form of the kinetically less favoured /^-substitution product (36). exciplex (39) or as derived from collapse of a On the whole, we favour the zwitter-ion (40) or a radical-ion pair (41). It should be noted however highly polarized exciplex that approaches it in that protonation of (41) would be expected to lead structure. to the substitution products (42) and (43), and in- The behaviour of anisole and dimethoxybenzenes corporation of deuterium as in (44) and (45), towards acrylonitrile is broadly similar to that of contrary to what is observed. Concerning exciplex naphthalene, except that cycloaddition occurs (39) in relation to the a-zwitter-ion (40), the latter readily in aprotic solvents. or^o-Cycloadducts and better accounts both for the position and specificity cyanoethyl substitution products are formed, of the deuterium labelling, whereas the former less dependent on the solvent. As with benzene and naphthalene, excitation of the arene, not charge- transfer excitation, is involved. We have observed CH,-CH-CN CH, 3 I that in cyclohexane as solvent, only ortho cyclo- Ph addition occurs, giving (46) from anisole and (47) • CN 31 from p-dimethoxybenzene, together with cyclo- 32 butane dimers of acrylonitrile. In methanol however, the reaction pathway is diverted away from cycloaddition and dimerization (in comparison with the partial diversion observed with naphthalene), and leads to the corresponding cyanoethyl substitu- 33: R =CN, R' = H 35: R = -CH(CH3)CN, R' = H tion products (48) and (49) [31], together with 34: R = H, R' = CN 36: R = H, R' = CH(CH3)CN substantial amounts of a homopolymer of acrylo- 37: R = -CHICH2D) CN, R' = H nitrile (contrast ref. [31]). As previously noted, 38: R = H. R'=-CH(CH2D)CN methanol promoted the ortho cycloaddition of 42: R = -CH2-CH2CN, R'=H acrylonitrile to the weaker donor benzene, but did 43: R = H, R' = -CH2-CH2CN not divert the reaction pathway. 44: R =-CH2-CHID)CN, R' = H

45: R = H, R' = -CH2-CH(D)CN Experimental 'CH- CN All irradiation experiments were performed under rvr CH a nitrogen atmosphere at 20-25 °C unless otherwise CHCN stated. The experiments with the maleimides and maleic anhydride as addends involved medium singlet exciplex pressure mercury arc lamps and pyrex apparatus 39 whereas the dimethyl acetylenedicarboxylate ben- 40 zene and acrylonitrile-arene systems were studied out in fused silica tubes and employed low pressure OMe mercury arc lamps. Irradiation of the ^-benzo- - CN quinone-arene systems employed medium pressure CH,-CHCN mercury arc lamps and pyrex apparatus: for the irradiation with visible light, an array of quartz- halogen lamps with a nitrite-hydrogen phthalate 41 46 : R = H filter solution was used [32], 47 : R = OMe Irradiation of dimethyl acetylenedicarboxylate and benzene Irradiation of the acetylene (5 ml) in benzene (100 ml) produced four volatile products in a time invariant ratio of 3:1:96:0.1 (respective relative retention times on Apiezon L g. 1. c. columns of 0.47, 48: R = R2=H, R1 = 0Me 0.70, 1.00, 1.26). After 72 h irradiation, removal 49: R=R2=0Me, R' = H of the reactants under vacuum produced 4 g of an 1110 D. Bryce-Smith et al. • Acid and Polarity Effects in Benzene Photoreactions orange oil which partially crystallised. Extraction Irradiation of dimethyl acetylenedicarboxylate of this residue with cold cyclohexane (3 X 50 ml) and benzene in the presence of trifluoroacetic acid gave a yellow solution which on concentration The following three solutions were irradiated and yielded yellow crystals (2.8 g) of 5 (m.p. 209.5 to the progress of the reactions folloAved by g.l.c. 210.5° following recrystallisation from methanol). (Apiezon L column): the results are given in Table I C12H12O4 in the text. The products Avere identified from comparison of their g.l.c. properties and mass Found C 65.44 H 5.50, spectra with those of authentic samples. Calcd C 65.45 H 5.50. (i) Dimethyl acetylenedicarboxylate (1.0 ml), ben- M+ = 220 m.u. m NMR spectra (CC14) (100 MHz) zene (50 ml). showed resonances at 7.06, 2 H, (d J = 3.0Hz); (ii) Dimethyl acetylenedicarboxylate (1.0 ml), tri- 6.2-5.7, 4 H, (m), and 3.67 ppm, 6 H, (s). Other fluoroacetic acid (1.0 ml, 0.014 mol), benzene spectroscopic features of (5) were identical Avith (50 ml). those reported elsewhere [2]. The three minor (iii) Dimethyl acetylenedicarboxylate (1.0 ml), tri- reaction products (relative abundance 3:1:0:1) fluoroacetic acid (8.0 ml, 0.112 mol), benzene AATere identified as dimethyl phthalate, dimethyl (50 ml). phenylfumarate and dimethyl phenvlmaleate respectiATely by comparison of their g.l.c. retention Irradiation of 5 in the presence of times on two columns and their mass spectra Avith trifluoroacetic acid those of authentic materials. A solution of (5) (0.5 g) and trifluoroacetic acid Irradiation of a 5% v/v of the acetylene in (4.0 ml) in benzene (500 ml) Avas irradiated and the benzene in pyrex apparatus AAdth a medium pressure progress of the reaction followed by g. 1. c. After 48 h mercury arc lamp produced the same products as exposure approximately 80% of (5) had been those described aboATe (comparison of g.l.c. prop- converted to a single product Avith a retention time erties and mass spectra). Yields were very IOAV and mass spectrum identical to those of dimethyl- and reflect the absorption properties of the phthalate: traces of benzene-trifluoroacetic acid acetylene under the two sets of irradiation condi- irradiation products were also evident [6]. The tions. cyclo-octatetraene (5) was recovered quantitatively from a "dark" control experiment. Irradiation of dimethyl acetylenedicarboxylate, Irradiation of maleimide and N-n-butylmaleimide tetracyanoethylene and benzene in benzene A solution of the acetylene (1 ml), tetracyano- A solution of maleimide (2 g, 0.02 mol) in benzene ethylene (1 g) and benzene (100 ml) was irradiated (150 ml) Avas irradiated for 6h and the arene for 12 d during which the formation of (5) was removed from the turbid solution by rotary evapo- examined periodically and found to be forming at ration. Diethyl ether (100 ml) A\as added to the ca. 20% the rate in the absence of the ethylene. A semi-solid residue and the insoluble (3) (R=H) dark broAAn solid (1.3 g) was filtered off and from its (100 mg) filtered off. The 2:1 adduct (M+ = 272m. u.) structureless IR spectrum, insolubility in common did not melt below 400 °C and had iwc (Nujol mull) solvents and non-sublimation at 250°/0.1 mm Hg at 3225, 3075, 1765, and 1700 cm-1 and <3 values it was assumed to be polymeric. The deep orange (CF3CO2H solution) at 9.87, 1 H, (brs); 9.69, 1 H, filtrate produced a brown solid (0.7 g) on concentra- (br s); 6.60, 2 H, (t, J = 3.5 Hz), 360, 2 H (brs), tion A\7hich on recrvstallisation from methanol and overlapping multiplets centered at 3.16, 3.09, produced (19), m.p. 209-211 °C, M+ = 348 m.u., and 2.91 ppm, 6 H. Vmax (Nujol mull) 2250, 1720, and 1642 cm-1. d values (100 MHz, acetone -D6) 6.53, 2 H, (q J = C14H12N2O4 4.0, J1 = 3.1 Hz collapsed to singlet by irradiation Found C 61.55 H 4.59 N 10.39, at 4.28 ppm), 4.28, 2 H. (m); 3.75, 6H (s); and Calcd C 61.76 H 4.44 N 10.29. 3.60 ppm, 2 H, (q / = 2.5, J' = 1.5 Hz collapsed to singlet by irradiation at 4.28 ppm). Sunlight irradiation of a similar solution of maleimide in benzene in a pyrex flask for six days C1SH12O4N4 in April again produced (2a) (R=H) (0.66 g). Found C 62.05 H 3.48 N 16.02, Irradiation of N-butylmaleimide (0.05 ml) in ben- Calcd C 62.07 H 3.47 N 16.09. zene (150 ml) for 7 h produced the diethyl ether insoluble 2:1 adduct (3) (R = w-butyl) (0.8 g) M+ = Irradiation of tetracyanoethylene (1 g) and 5 384 m.u. The adduct Avas recrystallised from ethyl (1 g) in benzene (100 ml) for 12 d produced a deep alcohol (m.p. 175.5-176.5 °C) and had vm3lX (Nujol orange solution and polymeric solid (0.8 g). Ex- mull) at 1755 and 1695 cm-1 and d values (CF3CO2H amination of the solution yielded tetracyano- solution) centered at 6.50, 2 H, (t, J = 3.5 Hz); 3.7. ethylene (0.5 g) and (5) (0.6 g): no evidence Avas 6 H, (m); overlapping broad singlets at 3.07 and obtained for the formation of (19). 2.86, 6 H; 1.5, 8 H, (m); and 1.0 ppm, 6 H, (m). 1111 D. Bryce-Smith et al. • Acid and Polarity Effects in Benzene Photoreactions

C22H28N0O4 C1sH1505N Found C 68.63 H 7.27 N 7.44, Found C 68.15 H 4.28 N4.17, Calcd C 68.73 H 7.34 N 7.29. Calcd C 68.77 H 4.30 N4.01.

Irradiation of the above N-w-butylmaleimide Irradiation of the maleic anhydride solution in solution in benzene in the presence of trifluoroacetic benzene in the presence of benzophenone (5 g) acid (10 ml) and under similar conditions produced produced 1.2 g of (2) in 8 h. Incorporation of tri- 0.6 g of (3) (R = w-butyl). Thin layer chromato- fluoroacetic acid (0.825 M) into this solution reduced graphic analysis of the irradiated solution alongside the yield of (2) to 0.6 g. that of authentic phenyl-N-(w-butyl) succinimide showed that this compound had not been formed in Solvent effect on the formation of (2) the photoreaction. Maleic anhydride (4.25 g) in solution in benzene (65 ml) and the diluent (95 ml) was irradiated for 5 h and the yield of the 2:1 adduct (2), isolated by its Irradiation of N-n-butylmaleimide, maleic anhydride insolubility in diethyl ether, was estimated by and benzene weight. The averaged weight of (2) from three A solution of N-w-butylmaleimide (1 g) and maleic experiments with each diluent were 110 mg (cyclo- anhydride (5 g) in benzene (200 ml) was irradiated hexane), 120 mg (acetonitrile), and 190 mg (di- for 7 h produced, on removal of the arene, a viscous bromomethane). Irradiation of the solution of pale yellow oil. Ethyl alcohol (10 ml) was added to maleic anhydride in benzene dibromomethane in the the oil and the solid (17) (R = w-butyl) (1.0 g) presence of trifluoroacetic acid resulted in the filtered off. The crude 1:1:1 adduct (M+ = 329) was formation of 115 and 125 mg of (2) in duplicate recrvstallised from ethvl alcohol (m.p. 223-225 °C) irradiations. and'had vmax (Nu.jol mull) 1853, 1778, and 1695 cm-1 and 6 values (CF3CO2H solution) centred at 6.64, Irradiation of p-benzoquinone in arenes 2 H, (t, J = 3.5 Hz); 3.70, 4 H, (m); 3.31, 2 H, A solution of freshly sublimed ^>-benzoquinone (br s); 3.05, 2 H, (br. s), 2.85, 2 H, (br. s); 1.50. 4 H, (1.0 g) in benzene (75 ml) containing trifluoroacetic (m), and 0.99 ppm, 3 H, (m). acid (0.425 M) was irradiated for 6 h either with radiation from medium pressure mercury lamps or C18H19NO5 light from an array of quartz-halogen lamps. The Found C 65.70 H 5.94 N4.41, benzene and acid were removed from the brown Calcd C 65.64 H 5.81 N 4.25. solution and the viscous dark brown residue was extracted four times with 50 ml of boiling cyclohexane. On concentration and cooling the cyclo- Photoreactions of maleic anhydride and benzene hexane solution gave (27) (0.95 g) (m.p. 83.5 °C no Irradiation of maleic anhydride (7 g) in benzene depression on mixed m.p. with authentic material). (150 ml) for 8h produced (2) (0.8 g) which was The XH NMR and infrared spectra were identical readily isolated by its total insolubility in diethyl with authentic material. ether. Incorporation of trifluoroacetic acid (10 ml) into the above irradiation solution totally sup- C12H10O2 pressed the formation of (2). The residue from this Found C 77.61 H 5.4, reaction was refluxed for 5 h with methanol (50 ml) Calcd C 77.42 H 5.38. and conc. sulphuric acid (0.1 ml) and from analysis Similar irradiation of the solutions of the quinone by g.l.c. (Apiezon L), combined m.s./g.l.c. and in acidified toluene and ^-xylene (orange and dark comparison with authentic phenyl dimethyl suc- red respectively) produced varying amounts of cinate, it was calculated that 100 ± 10 mg (repeat quinhydrone, and bibenzyl and its 4,4'-dimethyl runs) of phenyl succinic anhydride was formed in derivative respectively. the photoprocess. The combined esterified product from several reactions was combined and vacuum Irradiation of acrylonitrile-arene systems distilled. The fraction distilling between 80-90° at Formation, isolation, and identification of the 2.5 nm Hg was crystallised and had a m.p. of endo and exo ortho cycloadducts (30 a) and (30b) of 52-53 °C which was not depressed by mixing with acrylonitrile and benzene has been previously de- authentic phenyl dimethyl maleate. The infrared scribed [33]. Irradiation of acrylonitrile (1 ml), and *H NMR spectra of the reaction product were benzene (1 ml) in diluent (8 ml) (cyclohexane, also identical with those of the authentic compound. acetonitrile, and methanol) produced the same two Irradiation of maleic anhydride (7 g) in benzene adducts in the same ratio as deduced from h.p.l.c. solution (150 ml) in the presence of N-phenyl- analysis of the irradiated solutions on Partisil maleimide (10 g) produced a turbid tolution. Re- PX10/20 using 5% v/v ethanol in cyclohexane as moval of the arene by rotary evaporation and eluent: the rate of formation of (30a) and (30b) in addition of diethvl ether (70 ml) gave (25) (200 mg), the latter two solvents was approximately 1.5 times M+ = 349 m.u., *Wx 1715 cm-1. that in the hydrocarbon. The yields of the two 1:1 1112 D. Bryce-Smith et al. • Acid and Polarity Effects in Benzene Photoreactions

adducts from irradiation of equivolume mixtures and 2.48 ppm, 4-H, (m) and rmax 2245 cm-1. The (10 ml) of the addends in the absence and presence clear irradiated solution contained the two acrylo- of trifluoroacetic acid (1 ml) were, within experi- nitrile dimers and a 1:1 adduct (M+ = 191 m.u. mental error, the same. eluting as a single component with 3.3 relative We have previously described the photoaddition retention time) as the major product. Flash of acrylonitrile and anisole and the effect of cyclo- chromatography gave the 1.2-cycloadduct fraction hexane and acetonitrile solvents on the reaction [34]. (2.0 g) in high'purity (>99%), d values (CDCls) Ohashi and co-workers have reported that in centered at 6.34, 1-H, (d of dj; 5.83, 1-H (d); 4.81, methanol as solvent irradiation of acrylonitrile and 1-H. (d); 3.8-3.5, 1-H, (m) with 3.55, 3-H, (s) super- anisole yields only the substitution product (48) [31]. imposed; 3.32 1-H, (t); 3.15, 3-H, (s); 2.47, 1-H, (q), In contrast we find that irradiation of acrylonitrile and 1.69 ppm, 1-H, (q). The iwx (liquid smear) at (1.0 M) and anisole (0.15 M) in methanol yielded 2210 and 2240 cm-1 indicated the presence of two large amounts of insoluble white polymeric material stereoisomers. From this system Ohashi reports and the 1,2-cycloadduct and the substitution prod- formation of the substitution product (49) from uct (48) in a ratio of 2.3:1 respectively. methanol solution [31] but again in our hands the Irradiation of acrylonitrile (1.0 M) and p-di- results from the irradiation differ significantly. Thus methoxybenzene (0.1 M) in cyclohexane (400 ml) from acrylonitrile (1.0 M) and ^-dimethoxybenzene produced an imiscible oil and a clear water-white (0.1 M) in both acetonitrile and methanol large solution. By combined g.l.c./m.s. analysis the oil amounts of insoluble polymeric material were formed comprised two dimers (M+ = 106 m.u.) of acrylo- and only trace amounts of volatile material were nitrile (relative retention times on Carbowax 20 M obtained. of 1:4): these were separated by flash chromatography and are assigned cyclobutane structures on We thank the Science and Engineering Research the basis of spectroscopic analysis. Short retention Council for Studentships (to J. G. and M. A. H.) and time product (1.5 g), 6 values (CDCI3) centered at an Assistantship (to P. Y.), and the University of 3.62, 2-H, (m) and 2.55 ppm, 4 H, (m) and vm&x Kuwait for a Studentship (to N. A.-J.). The (liquid smear) 2245 cm-1; longer retention time P. C. M. U. (Harwell) are thanked for the 220 MHz product (1.0 g) 6 values centered at 3.55, 2-H, (m), *H NMR spectra and accurate mass measurements.

[1] H. J. F. Angus and D. Bryce-Smith, Proc. Chem. [18] R. Srinivasan and J. A. Ors, Chem. Phys. Lett. Soc. 1959, 326. 42, 506 (1976). [2] D. Bryce-Smith and J. E. Lodge, Proc. Chem. [19] A. Gilbert, G. N. Taylor, and A. Collins, J. Chem. Soc. 1961, 333; J. Chem. Soc. 1963, 695. Soc. Perkin I, 1980, 1218. [3] D. Bryce-Smith, R. R. Deshpande, A. Gilbert, [20] D. Bryce-Smith, A. Gilbert, and J. Grzonka, and J. Grzonka, J. Chem. Soc. Chem. Commun. J. Chem. Soc. Chem. Commun. 1970, 498. 1970, 561. [21] D. Bryce-Smith, A. Gilbert, and J. Grzonka, [4] M. Bellas, D. Bryce-Smith, M. T. Clarke, A. Angew. Chem. Int. Ed. Engl. 10, 746 (1971). Gilbert, G. Klunklin, S. Krestonosich, C. Manning, [22] D. Bryce-Smith and M. A^ Hems, Tetrahedron and S. Wilson, J. Chem. Soc. Perkin I 1977, 2571. Lett. 1966, 1895; J. S. Bradshaw, ibid. 2039. [5] D. Bryce-Smith and G. B. Cox, J. Chem. Soc. [23] M. A. Hems, Ph. D. Thesis, Reading University, Chem. Commun. 1971, 915. Reading, UK (1967). [6] D. Bryce-Smith, G. B. Cox, and A. Gilbert, [24] G. O. Schenck and R. Steinmetz, Tetrahedron J. Chem. Soc. Chem. Commun. 1971, 914. Lett. 1960, 1. [7] D. Bryce-Smith, A. Gilbert, and P. J. Twitchett, [25] W. Hartmann, H. G. Heine, and L. Schräder, J. Chem. Soc. Perkin I 1979, 558. Tetrahedron Lett. 1974, 883; ibid. 3101. [8] D. Bryce-Smith and A. Gilbert, Tetrahedron 33, [26] D. Bryce-Smith, R. R. Deshpande, and A. Gil- 2459 (1977). bert, Tetrahedron Lett. 1975, 1627. [9] A. Gilbert, Pure Appl. Chem. 52, 2669 (1980). [27] D. Bryce-Smith and J. E. Lodge, J. Chem. Soc. [10] B. E. Job and H. D. Littlehailes, J. Chem. 1962, 2675. Soc. (C) 1968, 886. [28] L. J. Andrews and R. M. Keefer, J. Am. Chem. [11] A. Gilbert, G. N. Taylor, and M. W. bin Samsudin, Soc. 75, 3776 (1953). J. Chem. Soc. Perkin I 1980, 869. [29] W. M. Hardham and G. S. Hammond, J. Am. [12] D. Bryce-Smith, Pure Appl. Chem. 16, 47 (1968). Chem. Soc. 89, 3200 (1967). [13] D. Bryce-Smith, B. E. Foulger, J. Forrester, [30] R. M. Bowmann, T. R. Chamberlain, C. W. Huang, A. Gilbert, B. H. Orger, and H. M. Tyrrell, J. and J. J. McCullough, J. Am. Chem. Soc. 96, 692 Chem. Soc. Perkin I 1980, 55. (1974). [14] K. E. Wilzbach and L. Kaplan, J. Am. Chem. [31] M. Ohashi, Y. Tanaka, and S. Yamada, Tetra- Soc. 93, 2073 (1971). hedron Lett. 1977, 3629. [15] J. Cornelisse and R. Srinivasan, Chem. Phys. [32] D. Bryce-Smith, A. Gilbert, and M. G. Johnson, Lett. 20, 278 (1973). J. Chem. Soc. 1967, 383. [16] D. Bryce-Smith, B. E. Foulger, and A. Gilbert, [33] A. Gilbert and P. Yianni, Tetrahedron 37. 3275 J. Chem. Soc. Chem. Commun. 1972, 664. (1981). [17] D. Bryce-Smith, J. Chem. Soc. Chem. Commun. [34] A. Gilbert and P. Yianni, Tetrahedron Lett. 1982, 1969, 806. 255.