The relative rates of hydrogen atom abstractio~l from ten substituted toluenes by t-butoxy radicals in carbon tetrachloride a t 40 OC have been measured and the data fitted to the Hammett equation. A much better correlation is obtained with the a+ constants than with the u constants of the substituents. In this respect, therefore, the reaction is similar to the majority of hydrogen atom abstraction reactions by radicals of moderate and high electron affinity. That is, the polar properties of the substituents are much more important than their stabilizing effect on the benzyl radical formed in the reaction.The cumyloxy radical has a similar reactivity to the t-butoxy radical. The p-nitrocumyloxy radical appears to be slightly more reactive and also slightly more susceptible to the polar elfects of substituents than are t-butoxy or cu~nyloxy radicals.The relative rates of hydrogen atom abstraction from compounds of the type XC6H4YH by free radicals can frequently be correlated by means of the Hamlnett p u relationship. Data for a \vide variety of such relations were sulnnlarized in a recent paper from this laboratory (1). I t was shown that with few exceptions, notably hydrogen abstraction by t-butoxy radicals fro111 substituted toluenes (2,3), a better correlation was usually obtained with (I+ constants rather than with u constant^.^ Further support for this generalization has been provided by Russell and Willianlson (5) who specifically pointed out that for the t-butoxy radical the proper substituents had not been exanlined to distinguish clearly betxveen a U-and a u+-correlation. With this point in mind we decided to reinvestigate this reaction with particular attention to those substituents having significantly different (I and 6 constants. While this \vork was in progress a similar study by Gilliom and Ward appeared (6). However, we feel that our results and conclusions differ sufficiently from theirs to warrant the present communication. EXPERIMENTAL ReagentsToluene, 112-xylene, p-sylene, ethylbenzene, and cumene were Phillips Research Grade materials. m-Chlorotoluene, p-chlorotoluene, vz-nitrotoluene, p-nitrotoluene, and p-methoxytoluene were obtained from Eastman. These compou~lds were used a s supplied, their purity being checked by gas-liquid chromatography (g.1.c.). Toluene-0-ds was obtained fro111 Merck, Sharp & Dohme; it contained over 95% trideuterotoluene. We are grateful to the Benzol Products Company for a generous sample of p-methoxybenzyl chloride.p-Phenoxytoluene was prepared by the method of Russell and Williamson (5). p-Cyanotoluene was prepared by the dehydration of p-tolualdehyde oxime with acetic anhydride (cf. ref. 7). t-Butyl hypochlorite was prepared as described by Teeter and Bell (8). Cumyl hypochlorite and p-nitrocumyl hypochlorite were prepared from the corresponding alcohols by the method of Zavitsas and Seltzer (9). The p-nitrocu~nyl alcohol was prepared from cumene by nitration (10) followed by oxidation of the purified p-nitrocuinene by the method of Icwart and Francis (...

Oxidation products of these phenols with metal ions vary with the acidity of the solution, the formation of dimers and trimers in alkaline solution giving place to oxidative demethylation as the acidity is increased. Lead tetra-acetate in acetic acid converts the phenols to mixtures of o-and p-quinones, and in benzene to unstable diacetoxycyclohexadienones. The structures of the monoacetates obtained on hydrogenolysis of the diacetates are proved. Oxidation of 4-methoxy-2-t-butylphenol with benzoyl peroxide leads to an unrearranged catechol monobenzoate, while traces of demethylated products are obtained from 4-methoxy-3-t-butylphenol.

The formation of spiroketal trimers (dibenzo[d,f]dioxepins) upon oxidation of alkyl monosubstituted p-methoxyphenols has been shown to be a general reaction. Representative trimers have been hydrolysed, and the methylated reduction product, 2,5,5'-trimethoxy-2'-(2,5-dimethoxy-3-methyiphenoxy)-3,3'-dimethylbiphenyl, of one of them synthesised to prove their structure. The formation of dimers in less alkaline solution is also described, and oxidation products of p-alkoxyphenols are reviewed.THE biological importance of the tocopherols has stimulated interest in the oxidation of highly substituted 9-alkoxyphenols. Oxidation of a-tocopherol in ethanol with alkaline ferricyanide gives an ethyl acetal, and dimer (I) has been isolated by oxidation of 4-methoxy-2,3,5,G-tetramethylphenol in benzene-light petroleum. The oxidation of a-tocopherol with alkaline ferricyanide in hydrocarbon solvents, however, leads to dimeric and trimeric keto-ether~,~ of which compound ( 11) is representative, involving both carbon-oxygen and benzylic coupling. It appears likely that the tocopherol metabolite isolated from mammalian liver is actually the dimer (11) or an isomer of similar structure rather than a 0 Me M e Me Me M e Me Me Me 111 Me R Me (I1 : R = C1,Hss)biphen0-2,2'-quinone.~ However, bipheno-2,2'-quinones are produced by oxidation of the bis-6-hydroxychroman

Gentle treatment with acid converts the ferricyanide oxidation product, believed to be the dienone (111). of 4methoxy-2-t-butylphenol into the hemi-acetal (IV). The latter's structure is proved by reduction to the trihydroxybiphenyl (V), and by formation of the azodibenzofuran (VII) on reaction with 2.4-dinitrophenylhydrazine.In acid, oxidation of either the hemi-acetal or of the trihydroxybiphenyl (V) gives the dibenzofuran-I ,2-quinone (XIII), but in neutral or alkaline solution the dibenzofuran-1.4-quinone (XV). Electron spin resonance specrrometry has been used to examine some of these oxidations.SOLUTIOKS of the blue, ferricyanide oxidation product of 4-methoxy-2-t-butylpheno1, first described by Baltes and Volbert as the 2,2'-biphenoquinoiie (I), deposit colourless crystals on cooling, to which these authors ascribed structure (11). It has been suggested that the colourless material is the cyclohexadienone (111) rather than the cyclic peroxide (11). Our spectroscopic data are in agreement with structure (111). The nuclearWte 1111) magnetic resonance (n.1n.r.) spectrum of a carbon tetrachloride solution of the freshly dissolved colourless material measured a t 60 Mc./sec. shows two equivalent aromatic protons (z 3.38), two coupled vinylic protons (7 4.59, 4-77, two doublets, J = 1.0 c./sec.), two non-equivalent methoxyl groups ( T 6-23, 6-35) and two non-equivalent t-butyl groups (z 8.55, 8.75). After three days a t room temperature the spectrum still showed resonances a t only these positions.The absence of resonances from the quinonoid form (I) is not surprising, as Baltes and Volbert have estimated that at equilibrium at 20°, reached after 12 hours, only 2% of the material exists in this form. A band at 1672 cm.-l in the infrared spectrum in cldoroform is also consistent with structure (111). As mentioned in Part I1 of this Series, the blue oxidation product is converted on silica gel into the hemi-acetal (IV). Evidence for structure (IV) is now presented.We have obtained compound (IV) by two methods; the first using silica gel, the second by treating the blue, light-petroleum solution of oxidised 4-methoxy-2-t-butylphenol with a small amount of toluene-9-sulphonic acid. The yellow hemi-acetal so obtained was reduced by sodium borohydride, or catalytically over palladiumcharcoal, requiring 1 mol. of hydrogen, to the trihydroxybiphenyl (V). Treating the latter with pyridinium chloride then acetic anhydride gave the diace-

Further evidence for the structure of the trimer (IIa) is provided by hydrolysis with toluene-p-sulphonic acid, nuclear magnetic resonance spectrometry, and hydrogenolysis.MULLER, Kaufmann, and Rieker 2 have recently described the oxidation of 4-methoxy-2,5-di-t-butylphenol (Ia) to a trimer to which they have assigned structure (IIa). We had also observed the formation of this trimer, and had carried out degradations of the molecule by methods differing from those used by Muller et al. As our results provide further evidence for structure (IIa) we report them here.It was found by Miiller et al. that ferricyanide oxidation of either phenol (Ia) alone in methanol, or (most significantly) of a mixture of phenol (Ia) and the phenoxy-phenol (IVa) in benzene led to the trimer (IIa). However, only 2,5-di-t-butyl-1 ,kbenzoquinone (111) resulted from ferricyanide oxidation of phenol (Ia) in benzene. In contrast, we have found that ferricyanide oxidation of phenol (Ia) in light petroleum gave the trimer quantitatively. Hydrogenolysis of the trimer in cyclohexane using palladium-charcoal catalyst gave the phenoxy-phenol (IVa) and phenol (Ia) in the molar ratio 1 : 1.It has been shown by Matsuura and Cahnmann3 that quinol ethers of type (V) are split by toluene-9-sulphonic acid to diphenyl ethers with loss of isobutene. When trimer (11) in ethyl acetate was treated with toluene-p-sulphonic acid the phenoxy-quinone (VIa) , phenol (Ia), the phenoxy-phenol (IVa), and quinone (111) were obtained in the molar ratio 12 : 11 : 1 : 0-6. Bearing in mind the volatility of the quinone (111) it will be seen that the products fall into two pairs, each pair containing a quinone and a phenol. The fact that no debutylation occurred eliminates the possibility of the trimer possessing structure (IIc). On the basis of the products isolated, consideration of protonation and cleavage of structures (IIa) and (IIb) provides additional evidence in favour of the former.Protonation at oxygen (i) in structure (IIa), as in (VII), followed by cleavage of bond " a " as in classical acetal hydrolysis should lead to phenol (Ia) and the carbonium ion (VIII). Carbonium ions of this type have been suggested as intermediates in the oxidative dealkylation of o-and @-alk~xy-phenols,~ and intermediate (VIII) would therefore lead to the phenoxy-quinone (VIa). The less likely fission of bond '' b '' should give the same final products. The minor pair of products can be accounted for by protonation of the carbonyl oxygen as in structure (IX), giving the phenoxy-phenol (IVa), and the quinone (111) via the corresponding carbonium ion. The duality of mechanism is analogous to