Reduction of Nitro- and Nitroso-compounds by Tervalent Phosphorus Reagents By J. I. G. Cadogan UNIVERSITY OF ST. ANDRBWS SCOTLAND Tervalent organophosphorus compounds (X,P) such as trialkyl- or triaryl- phosphines and trialkyl phosphites react with a wide variety of oxygen-contain- ing compounds to yield the corresponding quinquevalent derivatives (X,PO),l as in equation (1). The major driving force behind these reactions is the great X,P + zo + x,PO + z equation 1) strength of the P=O bond formed typical values for P=O bond dissociation energies in phosphates and phosphine oxides lying for example in the range of 12&150 kcal./mole,2 which can be compared with values in the range 50-70 kcal./mole for the +N-0- bond in amine oxides. Very many examples of the general reaction outlined in equation (1) have been known for some time but those involving reduction of nitro- and nitroso- compounds have received widespread attention during the last few years only.Such studies have led to the recognition of the reaction of reagents such as triethyl phosphite with aromatic nitro-compounds as one of considerable synthetical value and this Review is directed towards the mechanism and scope of this and related processes. Reduction of Aromatic Nitroso-compounds Most of the reported reactions of nitroso-compounds with tervalent phosphorus reagents have involved aromatic systems. The first record of such a reaction was by Hoffmann and Homer who stated in a reference to unpublished work that substituted nitrosobenzenes (ArNO; Ar = p-C1 p-Me p-NMea but not nitro- sobenzene itself reacted (equation 2) with triphenylphosphine to give the corre- sponding azoxybenzenes (ca.50 %). Bunyan and Cadogan5p6 later showed that ArNO + Ph,P -+ ArN=N(O)Ar + Ph,PO (Equation 2) reduction of nitrosobenzene by triphenylphosphine or triethyl phosphite did give azoxybenzene in low yield together with a polymer if the reaction which was exothermic was carried out in benzene. Reduction of o-ethylnitrosobenzene J. I. G. Cadogan Quart. Rev. 1962 16,208. a S. B. Hartley W. S. Holmes J. K. Jacques M. F. Mole and J. C. McCoubrey Quart Rev. 1963 17,204. * T. L. Cottrell ‘The Strengths of Chemical Bonds’ Butterworths London 1958. H. Hoffmann and L. Homer Angew. Chem. 1956,68,473 P. J. Bunyan and J. I. G. Cadogan Proc. Chem. SOC. 1962 78. P. J. Bunyan and J. I. G. Cadogan J. Chem. Soc.1963 42. 222 Cadogan and p-dimethylaminonitrosobenzene similarly gave the corresponding azoxy- compounds but in the latter case the reaction proceeded more slowly and triethyl N-p-dimethylaminophenylphosphorhidate (1) was isolated either directly or in the case of work-up involving activated alumina as its hydrolysis product diethyl N-p-dimethylaminophenylphosphoramidate (2) (Scheme 1). ArNO+(EtO)3P-+ArN =N (0)Ar + (E tO),PO B>Et-’ + EtO-PTNAr-+(EtO),P(O)NHAr EtO’ \H+ (1) Scheme These observations led to the suggestion ArNO + (EtO),P -t (EtO),PO + ArN ArN + ArNO + ArN=N(O)Ar ArN + (EtO),P 3 (EtO),P=NAr (2) (Ar = P-Me2N.C,H,-) 1 that the reactions could involve the (Equation 3) (Equation 4) (Equation 5 ) intermediacy of nitrenes [reactions (3)-(5)]. Thus in the case of p-dimethyl- aminonitrosobenzene the initial reaction would be expected to occur less readily and the derived nitrene would be more stable as a result of contributions of the form (3) hence increasing the chance of its capture by the strongly nucleophilic triethyl phosphite in a manner analogous to the capture of carbenes by triphenyl- ph~sphine.~ In accord with this equimolar amounts of the reactants gave the azoxy-compound (63.5 ‘4 and the phosphorhidate (1) (13 %) while a tenfold excess of triethyl phosphite over the nitroso-compound led to the predominance of the phosphorimidate (63-5 %) over the azoxy-compound (23 %).On the other hand if the reduction of o-ethylnitrosobenzene proceeds via a nitrene it might be expected that some insertion into the side chain would have been observed (equation 6).Bunyan and Cadogan detected no indoline however but pointed out that the absence of this product could be due to a greater ease of reaction of the nitrene if present with the nitroso-compound to give 2,2’-diethylazoxy- benzene (47 %). They also suggested that the reduction of o-dinitrosobenzene to benzofurazan (4) (Scheme 2) by triphenylphosphine and related reactions D. Seyferth S. 0. Grim and T. 0. Read J. Amer. Chem. SOC. 1960,82 1510. 223 Reduction of Nitro- and Nitroso-compounds by Tervalent Phosphorus Reagents Scheme 2 reported contemporaneouslyY8 might proceed via a nitrene. It is of interest that o-nitronitrosobenzene is smoothly reduced by triethyl phosphite to the furazan oxide ( 5 ; 19%) which in turn gives the furazan (4) on further reduction under more forcing conditions (150°).9 It should be noted that the results of these experiments did not exclude routes (equation 7) to azoxybenzenes or a similar route (equation 8) to benzofurazan.ArNO dimer i- (EtO),P+ (EtO),PO +Ar N = N ( 0 ) Ar 4 (Equation 7) (Equation 8) Regardless of mechanistic detail however this reduction of nitroso-com- pounds under certain conditions provides a good route (equation 9) to azoxy- compounds and has been exploited as such.1° C6F5N0 f (EtO),P -+ C6F5N =N(0)C6F5 (80 %) (Equation 9) A more important synthetical development of this reduction involves the reductive cyclisation of 2-nitrosobiaryl.~~,~ to carbazoles. Thus 2-nitrosobiphenyl reacts (equation 10) within a few minutes with triethyl phosphite in benzene or ether at 0" to give carbazole (76%) and triethyl phosphate (84%).Triphenyl- phosphine is also effective but it is simpler in practice to use triethyl phosphite since both it and triethyl phosphate are easily removed by distillation. Phos- phorus trichloride does not react indicating that a strongly nucleophilic phos- phorus atom is required. With use of triethyl phosphite 3-o-nitrosophenyl- pyridine (6) gave a mixture (64%) of ~(81.5%) and 7-carboline (183%) (equa- tion l l) while 2-o-nitrosophenylpyridine gave an almost quantitative yield of pyrid[l ,Zb]indazole (7) (equation 12). The latter observation was considered6 to be in accord with the concept of an electron-deficient nitrogen intermediate which would react preferentially at the electron-rich ring-nitrogen atom. Further pyrid[lY2-b]indazole (57 %) has been obtained from the decomposition of 2-0-azidophenylpyridine~~~ there being a J.H. Boyer and S . E. Ellzey J. Org. Chem. 1961 26,4684. @ J. I. G. Cadogan M. Cameron-Wood R. K. Mackie and R. J. G. Searle J. Chem. Soc. 1965,4831. lo J. Burdon C. J. Morton and D. F. Thomas J. Chem. SOC. 1965 2621. l1 R. A. Abramovitch and K. A. H. Adams Cunud. J. Chem. 1961 39,2516. 224 Cadogan (Equation 11) (Equation 10) - + (Equation 12) considerable amount of evidence,12,1s particularly for photo-induced reactions,14 that this type of reaction proceeds via a nitrene. Convincing support for the intermediacy of a nitrene in the reduction of nitroso-compounds by triphenylphosp hine has recently been provided by Odum and Brenner15 following work on the photolysis16 and pyrolysis17 of phenyl azide in amines reactions which proceed via nitrenes which undergo ring expansion in the presence of amines with the ultimate formation of derivatives of 2-amino-3-H-azepines (9).Huisgen and his co-workers17 suggested that the reaction involved the 7-azabicyc10[4,1 ,O]heptatriene (8) e.g. equation (13). The reported isolation of azepines of this type from reactions of nitroso- benzene with triphenylphosphine in the presence of dialkylamines is therefore good evidence in favour of the participation of nitrenes15 in these cases also. Recent work by Sundberg18 also points to the intermediacy of nitrenes in the reactions of o-methyl- o-propyl- and o-butyl-nitrosobenzene carried out in excess of triethyl phosphite in the absence of a solvent. Under these conditions which are very different from those used by earlier worker^,^^^ which involved more nearly equimolar proportions of reactants in a solvent little of the corre- sponding (ca.1 %) azoxy-compounds were isolated and low yields (ca. 5-1 1 %) of triethyl N-arylphosphorimidates (1 ; Ar = O-Me.C,H,; o-Pr-C,H,; o-Bu-C,H,) were indicated by g.1.c. and n.m.r. studies on mixtures of products. l2 P. A. S. Smith and J. H. Boyer J. Amer. Chem. SOC. 1951 73 2435. l3 G. Smolinsky J. Amer. Chem. SOC. 1960 82 4717; 1961 83 2489; J. Org. Chem. 1961 26 4108. l4 G. Smolinsky E. Wasserman and W. A. Yager J Amer. Chent. SOC. 1964 86 3166; G. Bowes A. Reiser and H. Wagner Tetrahedron Letters 1966 2635. l5 R. A. Odum and M. Brenner J. Amer. Chem. SOC. 1966,88,2074. l6 W. von E. Doering and R. A. Odum Tetrahedron 1966 22 81. l7 R.Huisgen D. Vossins and M. Appl Chem. Ber. 1958 91 1; R. Huisgen and M. Appl ibid. p. 12. I6 R. J. Sundbtrg J. Amer. Chem. SOC. 1966 88 3781. 225 Reduction of Nitro- and Nitroso-compounds by Tervalent Phosphorus Reagents In addition small quantities (1-7%) of amines were detected by g.1.c. and in the case of o-nitrosotoluene at O" N-o-tolyl-a-methyl-a-(2-pyridyl) nitrone (10; R=Me; Ar=o-tolyl)) was isolated in varying amounts (ca. 2&30%) according to conditions. At 156" the corresponding anil (11) was also formed by deoxy- genation. The relatively lower yields of azoxy-compounds and higher yields of phosphorimidates (1) obtained in these cases compared with those studied by Bunyan and Cadogan6p6 are as expected as a result of the presence of a large excess of triethyl phosphite.The nature of the amines detected vary with reaction conditions; thus at 0" a-alkylanilines (ca. 2 %) are formed while slightly higher yields (ca. 5-7 %) are produced at 156" together with traces of what are believed to be indolines (12; R=Et or Me) formed by insertion reactions of a nitrene in the side chain. In view of the low yields and lack of evidence based on actual isolation of products little mechanistic significance can be attached to these observations as is readily pointed out by Sundberg. The mode of formation of the nitrone (10) is not yet established but Sundberg has drawn attention to a possible route involving first a nitrene and then a 7-azabicyclo[4,1 ,O]hepta-2,4,6-triene intermediate of the type discussed above which is supposed to react with another molecule of the nitroso-compound (Scheme 3) although at the high temperature and with a relatively low concentration of the nitroso-compound it would be surprising if this reaction between two highly reactive intermediates had a high probability of taking place.Scheme 3 It is clear therefore that several aspects of the mechanism of reduction of nitroso-compounds by tervalent. phosphorus reagents still remain to be clarified (See Appendix). Reduction of Aromatic Nitro-compounds In theory the triethyl phosphite-induced reductive cyclisation of 2-nitrosobiaryls to carbazoles is of some synthetical value. In practice the preparative significance of this reaction is reduced by the difficulties sometimes encountered in the 226 Cadogan preparation of the nitroso-compound from the corresponding nitrobiaryl.This difficulty was overcome and wide synthetical possibilities revealed by the demon~tration~,~~ that reaction of the parent 2-nitrobiphenyl with a slight excess over two equivalents of boiling triethyl phosphite gave carbazole in excellent yield (83%) (equation 14). Extensions of this reaction reported at the included the conversion of both cis- and trans-2-nitrostilbene and a-nitrostilbene into 2-phenylindole 2-nitrobenzylidene anilines into indazoles 2-nitroazoarenes into benzotriazoles and various dinitro-compounds into polycyclic derivatives containing five-membered rings. Since then many additional ramifications of the reductive cyclisation have been reported and for the purpose of this Review it is convenient to classify the results of the various investigations into the scope and mechanism of the reactions in terms of the type of ring system produced.Formation of Carbazoles and Related Systems.-The triethyl phosphite-induced reduction of 2-nitrobiphenyl has been extended to give various bromo- chloro- and methyl-carbazoles in moderate to good yields (35-83 2,2'-Dinitro- biphenyl on the other hand gave only a low (1.5%) yield of benzo[c]cinnoline (13) rather than the possible alternative (14),9 the former compound being readily produced (86 %) by reduction of benzo[c]cinnoline NN'-dioxide under similar conditions. It is of interest that 2,2'-dinitrobiphenyl is converted into benzo[c]cinnoline "'-dioxide by phosphine in allcali,2lU this reagent having previously been used to produce azoxy-compounds from nitroarenes21 Scheme 4 la J.I. G. Cadogan and M. Cameron-Wood Proc. Chem. SOC. 1962 361. *O J. I. G. Cadogan and R. J. G. Searle Chem. and Ind. 1963 1282; 1434. 81 (a) A. G. Bellaart Tetrahedron 1965 21 3285; (b) S. A. Buckler M. Epstein L. Doll and F. K. Lind J. Org. Chem. 1962 27 794. 227 Reduction of Nitro- and Nitroso-compounds by Tervalent Phosphorus Reagents Apart from the above case 1,s-cyclisation is the preferred reaction in phosphite reductions thus 1-o-nitrophenylnaphthalene which presents two possible points of ring closure is reduced to 3,4-benzcarbazole (15) rather than the isomeric mesoben~acridine~ (1 6) (Scheme 4). Reductive cyclisations can be effected by other tervalent phosphorus reagents and a comparative study of their relative efficiencies has been made by Cadogan and Todd22a who followed the rate of removal of 2-nitrobiphenyl on reaction with an excess of the reducing agent (15-20 moles).Although the precision of the g.1.c. method of analysis which they used was not high (15%) the results (Table 1) are useful in that they establish the order of reactivity of the phosphorus reagents thus (EtO),PMe > (Et,N),P N (EtO)P(NEt,) > (EtO),P - (PriO),P > > PCl (inactive) and indicate that the reaction carried out in a polar solvent is not faster. Qualitative experiments established that tributyl- and triphenyl- phosphine were also suitable deoxygenating agents but that neither was as reactive as the phosphorous amides or diethyl methylphosphonite. The results suggest that the rate-determining step involves nucleophilic attack by phos- phorus but the extremely rapid reaction of diethyl methylphosphonite cannot be explained on this basis alone yet it would be surprising if such an enhance- ment in rate arose from simple steric effects alone.The method of analysis Table 1 Deoxygenation of 2-nitrobbhenyl (Pr 'O),P (EtO),P (EtO),P (EtO),P (EtO)P(NEt,) (Et,N),P x3p (EtO),P + Me2NCO-H (equimolar) (EtO),PMe Temp. 143.5" 155 145 135 144 121 111 61 ti (min.) 63 32 50 83 97 ca. 5 41 154 employed in these kinetic experiments unfortunately could be extended to include only two substituted nitrobiaryls (Table 2) but they show as would be expected for a mechanism involving deoxygenation rather than ring closure as the rate-determining step that the methyl group in 4-methyl-2'-nitrobiphenyl causes no change in rate compared with 2-nitrobiphenyl while the slight increase in rate observed with 4-bromo-2-nitrobiphenyl lends support to the probability of nucleophilic attack by phosphorus at the nitro-group.A key question of mechanism still to be answered in these reactions is whether the nitro-compound is reduced to the carbazole via the corresponding nitroso- compound. In practice this is difficult to establish of course because of the great ease with which 2-nitrosobiphenyl reacts with triethyl phosphite to give carbazole ra (a) J. I. G. Cadogan and M. J. Todd Chem. Cumm. 1967 178; (b) unpublished results; (c) J. I. G. Cadogan S. Kulik and M. J. Todd unpublished results. 228 Cadogan Table 2 Deoxygenation of nitrobiaryls by triethyl phosphite (145.5 ") X H H 4-Br Y th (min.) H 50 4'-Me 50 H 17 (ti = ca. 2 min. at 0").In the absence of evidence to the contrary however it remains a reasonable possibility (See Appendix). The main point of interest concerning the mechanism of the ring closure centres however on the possibility of either the participation of a nitrene formed from the intermediate nitroso-compound or by another more direct route (Scheme 5). The question of the point of initial attack by the phosphorus atom is also open to discussion. In the cases of both the nitro- and nitroso- groups it can be argued that initial attack occurs at the more positively polarised nitrogen atom followed by rearrangement possibly via a three-membered cyclic intermediate to a dipolar structure in which the phosphate leaving group is latent. Alternatively the latter may be formed by direct attack on the oxygen atom.Thus for reduction of the nitro group Scheme 6 might apply a similar dipolar intermediate (RO),P+-O-N-Ar being formulated in the case of the nitroso-compound. Deoxygenation of such acyclic and cyclic intermediates could also lead to the phosphorimidate (1) directly. Scheme 5 + ?- 0- /7 (RO),P-O-NAr-+ (RO),PO - ArNO IR0)jP +O=I$-Ar .T (ROi,P/p ' 0- Scheme 6 'Nil. Evidence for nitrenes is usually taken to be the occurrence of abstraction and insertion reactions with C-H bonds.23 Thus the formation of 8,lO-dimethyl- na R. A. Abramovitch and B. A. Davis Chem. Rev. 1964,64 149. 229 Reduction of Nitro- and Nitroso-compounds by Tervalent Phosphorus Reagents phenanthridine (16; 50%) 2,4,9-trimethylcarbazole (17; 5 %) and 2’-amino- 2,4,6-trimethylbiphenyl (18; 30%) has been attributed to the participation of a nitrene in the decomposition of 2’-azido-2,4,6-trimethylbiphenyl (19 ; X = N3) in hexadecane at 230”.% Following this useful results bearing on the mechanism of the phosphite-induced deoxygenation have been obtained from the reaction of the corresponding 2’-nitro-2,4,6-trimethylbiphenyl (19; X = NO2) with a slight excess of triethyl phosphite i.e.under the normal conditions of the reaction.a2a The formation of 2’-amino-2,4,6-trimethylbiphenyl (1 8; 13 %) in this case suggests abstraction by an electron-deficient nitrogen species such as a nitrene a suggestion supported by the formation of triethyl N-(2’,4’,6-trimethyl- biphenyl-Zyl) phosphorimidate (20; 15 ‘by presumably by reaction of the nitrene with excess of triethyl phosphite (Scheme 7).8,lO-Dimethylphenanthridine (1 6) and 2,4,9-trimethylcarbazole (17) were not formed under these conditions presumably because their formation would have involved the energetically less favourable insertion into strong bonds in competition with the easier pathway involving coupling with excess of phosphite present. In the corresponding reaction of the azide (19; X = N& the latter low- energy pathway is not available hence insertions occur. In accord with this deoxygenation of the nitrobiaryl in an excess (ca. 50 mol.) of (A) isopropyl- benzene and (B) t-butylbenzene under which conditions the coupling reaction to give the phosphorimidate (20) would be expected to be reduced and the inser- IX=N,; 230’ X=NO (EtO),P + PhCHMe (16)+(18)+(20) OMe (22) Me- ArN.+ PhCHMe? + ArNH + PhCMe,.-+ PhCMeiCMezPh (21) ArN. + (EtO),P + (EtO),P=NAr Scheme 7 tion reactions to be enhanced gave in addition to (18) and (20) the insertion product 8,lO-dimethylphenanthridine (16) [(A) 14 %; (B) 12x1. Further in (A) the amount of hydrogen abstraction increased [(18); 32%; cf. 19% in (B)] l4 G. Smolinsky J. Amer. Chem. Soc. 1960 82 4717. 230 Cadogan with the concomitant formation of bi-a-cumyl (21 ; 11 %). Thus when the possibility of coupling with triethyl phosphite is reduced the presumably less energetically favourable abstraction and insertion process occur to an appreci- able extent. Bi-a-cumyl can only arise by dimerisation of a-cumyl radicals produced by abstraction of a hydrogen atom from cumene by a triplet species presumably a nitrene (Scheme 7). The difference between the yields of 2’-amino- 2,4,6-trimethylbiphenyl(l8) formed in experiments (A) and (B) is almost exactly equivalent to the amount of bi-a-cumyl formed in the reaction in cumene t-butylbenzene having no readily abstractable hydrogen atom.In accord with these observations Smolinsky and FeueF have also isolated after hydrolysis of the products of the reaction of triethyl phosphite and 2’-nitro-2,4,6-trimethyl- biphenyl (19; X = NO& a trace of the amine (18) and a compound tentatively identified by n. m . r. as diet h yl N-(2’ 4’,6’-trimet h yl biphenyl-2-yl)p hosp horami- date (22). This appears to be a correct assignment because Cadogan and Todd established22a that the first formed phosphorimidate (20) is easily hydrolysed to the phosphoramidate (22) and in general the formation of phosphoramidate in such reactions usually points to the prior production (equation 15) of the phosphorimidate.6p26 (EtO)sP=NAr --t (EtO)2P(O)Nm (Equation 15) Related to this problem is the suggestion referred to above that phenyl- nitrene and 7-azabicycl0[4,1 ,O]hepta-2,4,6-triene (8) are successive intermediates in the ring expansion which occurs on decomposition of phenyl azide in amines,l6~l7 and as described earlier this ring expansion has been utilised16 to support the suggestion6 of nitrene participation in the deoxygenation of nitrobenzene by triphenylphosphine.Application of this test to the case of nitro- compounds was difficult at first in view of the very much more drastic conditions required but this difficulty disappeared with the demonstration of the high reactivity of diethyl methylphosphonite as a reducing agent.22a Thus reaction of the latter with 2-nitrobiphenyl in an excess of diethylamine gave 2-diethyl- amino-3H-3-phenylazepine (23; 13 in addition to carbazole (67 % compared with 86% in the absence of amine).At fist sight the formation of the azepine derivative suggests the participation of a nitrene in a manner analogous to that generally accepted15,16 following the suggestion by Huisgen and his co-workers17 (Scheme 8). Scheme 8 as G. Smolinsky and B. I. Feuer J. Org. Chem. 1966,31,3882. * J. I. G. Cadogan and H. N. Moulden J. Chem. SOC. 1961,3079. 231 Reduction of Nitro- and Nitroso-compounds by Tervafent Phosphorus Reagents Scheme 9 Scheme 10 The concomitant formation of carbazole in this reaction calls for a more detailed examination of the possibilities however.From the foregoing the most plausible explanation for the formation of carbazole involves attack by the nitrene at the 2’-position of the biaryl either by radical abstraction of a hydrogen atom followed by radical recombination or by direct insertion into the C-H bond (Scheme 9). The isolation of both carbazole and the azepine derivative suggests the following possibilities (1) That both products arise from the same intermediate which would be either the nitrene or the derived azabicyclo- heptatriene (Scheme S) or (2) that both intermediates are present in equilibrium (cf. Abramovitch and Davis23) carbazole arising from the nitrene and the azepine from the azabicycloheptatriene. If a common intermediate is involved (alternative 1) it appears to be unnecessary to invoke the sequence nitrene- azabicycloheptatriene-carbazole and an alternative explanation does not require the intermediacy of the highly strained azabicycloheptatriene system at all but rather requires a ring expansion concerted with attack by the nucleo- philic amine in competition with reaction of the nitrene with the aromatic ring (Scheme 10).Against this it would be expected that direct reaction of the nitrene with an amine would produce the diethylhydrazine ArNH.NEt,. Regardless of this detail however it is clear that the similarity of mechanism of the decomposi- tion of azides believed to involve nitrenes and of the deoxygenation of nitro- biaryls by triethyl phosphite is established. Formation of Indoles and Related Systems.-By analogy with the cyclisation of 2-nitrobiaryls to carbazole o-nitro-styrenes or -stilbenes would be expected to give rise to indoles and these possibilities have been realised experimentally (Scheme 1 1).Thus cis- and trans-Znitrostilbene and a-nitrostilbene give 2- phenylindole (85 58 and 16 % respecti~ely),~~~~ and a-(o-nitrophenyl)-2- chlorocinnamic acid gives 2-(o-chlorophenyl)-3-ethoxycarbonylindole (22b; 46%) on reaction with triethyl phosphite indicating the occurrence of esteri- 232 Cadogan fication presumably by triethyl phosphate during the reaction. The reductive cyclisation of 2'-nitro-a-stilbazole which presents two possible points of ring closure also gives rise to an indole (24) rather than proceeding via reaction at the electron-rich nitrogen to give a diazepine.22b o-Nitrostyrene and 2,2'- dinitrostilbene gave small yields (ca.1-2 %) only of indole and indolo[2,2-b]- indole (25) respectively while b-aitrostyrene gave no indole and o-nitrocin- namic acid was reduced and esterified to give a low yield of indole-2-carboxylic ester. J Scheme 11 H H Taylor and Garciaz7 have prepared in low yield two of the biologically interesting pyrrolo[3,2-d]pyrimidines (27) from the corresponding 5-nitro-6- styrylpyrimidine derivatives (26) thermally and by the irradiation in the presence of triethyl phosphite. It is not clear in the latter case whether the reaction is photochemically initiated because the experimental details provided indicate that 27 E. E. Taylor and E. C. Garcia J. Org. Chem. 1965 30 655. 233 Reduction of Nitro- and Nitroso-compounds by Tervalent Phosphorus Reagents some warming of the reaction mixture occurred.In this case therefore the products may be arising via a thermal process. It is tempting to extend the analogy between the reductive cyclisation of nitro- biaryls to carbazoles and of o-nitrostyrene derivatives to indoles to include a common mechanism but some interesting results reported by Sundberg in the course of an extension of this indole synthesis indicate that this may not be valid. Thus Sundberg2* confirmed the synthesis of 2-phenylindole from 2-nitro~tilbene~ and extended it to include 2-alkyl- (Me Et ; 50-60 % yields) and 2-acyl- (MeCO PhCO; 16%)-indoles. Of more interest are the by-products of these reactions. Thus deoxygenation of 2-nitro-trans-stilbe gave compounds formulated as 2,2’-diphenyl-3,3’-bi-indolyl(28 ; 7 %) and diethyl2-phenyl-3-indolylphosphonate (29; 1.6%) although the possibility that the latter is the phosphoramidate (30) is not rigorously excluded by the evidence presented.Postulating l-hydroxy-2- phenylindole (31) as an intermediate Sundberg proceeded to isolate this fFom a deoxygenation interrupted well before completion and further demonstrated that on being heated with triethyl phosphite this product gave a product dis- tribution similar to that obtained from the 2-nitro-trans-stilbene (Scheme 12). Bi-indolyls were also obtained from the deoxygenation of p-methyl- and p- propyl-o-nitrostyrene together with small amounts of the corresponding 1- ethoxy-2-alkylindoles but l-ethoxy-2-phenylindole was never detected among the products of the reaction of 2-nitro-trans-stilbene the most thoroughly investigated reaction.Scheme 12 The demonstration that 1-hydroxy-Zphenylindole is an intermediate indicated that in this case Scheme 13 may be followed at least in part thus raising an important point of difference between this and the corresponding reductive cyclisation of 2-nitrobiaryls. The mode of formation of the bi-indolyl in these reactions has not been estab- lished but it is significant that the yield of the diphenylbi-indolyl(28) is increased at the expense of 2-phenylindole if the proportion of triethyl phosphite is reduced by performing the reaction in triethyl phospkate or p-cymene. Thus an inter- s$ R. J. Sundberg J. Org. Chem. 1965 30 3604. 234 Cadogan $9 r:r; \ N-O /N-0 D h ( E t 016 m3P;% ' y-0 OH WPh 0 Scheme 13 mediate in the reaction of triethyl phosphite with 2-nitrostilbene can be diverted to the bi-indolyl (28) when the concentration of the phosphite is reduced but Sundberg notes that the yields of products are not high enough for it to be stated with certainty that this unknown intermediate is converted into 2-phenylindob in excess of the phosphite although this seems likely.If a triplet nitrene is an intermediate in this reaction it would have been expected that some 2-amino- trans-stilbene and ad-bi-p-cymyl would be formed. Sundberg did not comment on this possibility however and it is not clear from the experimental details provided whether these compounds were sought. In the corresponding reaction of #$%disubstituted o-nitrostyrenes,as indoles are again major products. Thus cyclohexylidene(cmitropheny1)methane (32) underwent ring closure with rearrangement to give 5,6,7,8,9,1O-hexahydrocyclo- hept[b]indole (33; 35 %) together with lower yields of the bi-indolinyl(34; 24 YJ and the spiro-indolinone (35; 8%) (Scheme 14) (See Appendix).Similarly @-dimethyl-o-nitrostyrene gave 2,3-dimethylindole (33 %) the indolinone (36; 11 'A and a very low yield of the bi-indolinyl(37) (Scheme 19 Scheme 14 Scheme I5 aB R. J. Sundberg and T. Yamazaki J. Org. Chem. 1967,32,290. 235 Reduction of Nitro and Nitroso-compounds by Tervalent Phosphorus Reagents while a-methyl-2’-nitrostilbene gave a high yield of the rearranged indole (38 ; 77%) together with the N-ethylated product (39; 21 %) presumably formed by alkylation of the first formed indole with triethyl phosphate (Scheme 16).Scheme 16 It is noteworthy that in this case in contrast to the reductive cyclisation of simpler o-nitrostyrenes discussed above none of the corresponding 1 -hydroxy- indoles were detected even in the case of interrupted reactions. Thus in these instances it cannot be said whether or not nitrenes or 1-hydroxyindoles are intermediates but it is clear from the rearrangement which occurs particularly in the case of a-methyl-2’-nitrostilbene that an electrophilic nitrogen species is involved and likely possibilities are outlined in Scheme 17. 0’ ‘&-P(OEt) ‘ U I W P’ Scheme 17 The genesis of the by-products is even less obvious; the formation of bi- indolinyls suggests a homolytic process possibly via an intermediate nitroso- compound (Scheme 18) or via a triplet nitrene (Scheme 19).Against Scheme 18 is the requirement of an electron donor for the first step to be possible and the unlikelihood of realising a sufficiently high concentration of the nitroso-com- pound under the experimental conditions. Scheme 19 allows a rationalisation of all the observed products in terms of a common intermediate. Further experi- mentation to obtain a better accountance of starting material is clearly necessary in these cases. Deoxygenation of p-nitrostyrene in the presence of triethyl phosphite has been reported to give a positive test for indole although it is clear that the yield is very Of more interest is the reported isolation of a low yield of phenyla~etonitriIe.~~ That this compound is also produced by photolysis of 30 (a) Ref. 23 refers to a persona1 communication from A.Weinstock; (6) J. H. Boyer W. E. Krueger and G. J. Mikol J. Amer. Chem. SOC. 1967 89 5504. 236 Cadogan l i J’ ti donor Scheme 19 /3-styrylazide has led Boyer and his co-workers tentatively to suggest that /3-styrylnitrene may be an intermediate in both reactions30b [equations (16) and (17).] PhCH-CHNO -+ PhCH=CH.N -+ PhCH,CN PhCH = CHN -t PhCH = CH+N -t PhCH,CN (Equation 16) (Equation 17) Formation of Indazoles Triazoles and Related Systems.-Following the sup- positionlg that an electrophilic nitrogen species such as a nitrene is an inter- mediate in the deoxygenation of 2-nitrobiaryls by triethyl phosphite it was shown early9*l9 that in the analogous reaction of 2-o-nitrophenylpyridine (40) ring closure occurred at the electron-rich ring-nitrogen atom rather than at carbon to give pyrid[l,2-b]indazole (41) in excellent yield.This opened the way to success- ful cyclisations of o-nitrobenzylideneanilines to 2-arylindazoles and of o-nitro- azoarenes to the corresponding benzot~-iazoles,~~~~ except in the case of 2-nitro- 4’-hydroxyazobenzene which undergoes ethylation31 as well as reductive cyclisa- tion to give 2-p-ethoxyphenyl-2H-benzotriazole (Scheme 20). Similarly o-nitro- benzaldehydeazine gives 2,2’-bi-2H-indazolyl (42) bis-[o-nitrobenzylidenel-p- phenylenediamine gives p-di-2H-indazol-2-ylbenzene (43) while the fused five- membered ring system dibenzo[b,fl-l,3a,4,6a,-tetra-azapentalene (44) is readily obtained from 2,2’-dini troazo benzene. 31 Cf. J. I. G. Cadogan J. Chem. SOC. 1957 1079. 237 Reduction of Nitro- and Nitroso-compounds by Tervalent Phosphorus Reagents Ar=Ph,-P-C,H,*Me o-C,H,*Me p-C,H;OMe o-C,H,Br a-CIOH7; 35-83‘%1 Scheme 20 Extensions of these reactions which have been reported since then include (equation 18) the formation of pyrazolo[l,2-a]benzotriazole (46; 18 %) from o-nitrophenylpyrazole (49 32 and (equation 19) of the new benzotriazolo- naphthotriazine (48 ; 46 %) from the naphthotriazine (47).ss (45) (47) 4 N-N equation 19) \ ** Y.Y. Hung and B. M. Lynch J. Heterocyclic Chem. 1965 2,218. 8* H . Sieper Tetrahedron Letters 1967 1987. 238 Cadogan The reaction (equation 19) has been extended to include some C-methyl derivatives and to the formation of the isomeric triazolonaphthotriazine (49) from the corresponding triazine (equation 20),54 and of 13-oxobenzotriazolo [ 2,l-b]benzo[l,2-e]triazine (50) from 3,4-dihydro-4-oxo-l,2,3-benzotriazine (equation 21).= Similarly 2-(o-nitrophenyl)-2benzotriazole (51) 240-nitro- phenyl)-2H-triazole (52) and l-(o-nitropheny1)-lH-benzotriazole (53) are readily converted (61-88 %) into the mono- and dibenzo-tetra-azapentalenes (54) (55) and (56)88 (Scheme 21).Scheme 21 L=/ Scheme 22 Kauer and Carboni in their study of this reactionsg noted that it was strongly affected by the nature of the phosphorus compound; the reactions with rne more nucleophilic tributylphosphine were more rapid but the yields and product purity were higher with triethyl phosphite. The increased nucleophilicity of the former reagent led to a competing reaction at the carbonyl group in the case of the ester (57) to give a product tentatively formulated as (58) (Scheme 22).p4 H. Sieper and P. Tavs Annalen 1967,704 161. *s A. W. Murray and K. Vaughan Chern. Cornrn. 1967 1283. J. C. Rauer and R. A. Carboni J. Amer. Chem. Soc. 1967,89,2633. 239 Reduction of Nitro- and Nitroso-compounds by TervaIent Phosphorus Reagenrs Formation of Anthrani1s.-The reductive cyclisation of nitro-compounds involv- ing an electron-rich centre in the 5-position has also been extended3' to include the formation of anthranils from o-nitrophenyl ketones [equation (22)l. Thus 2-nitrobenzophenone gave 3-phenylanthranil (56 %) and 2-aminobenzophenone (19-5 %) the latter suggesting the intermediacy of a nitrene while 2'-nitro- chalcone and 5-chloro-2-nitroacetophenone similarly gave 3-styrylanthranil K (54 %) and 3-methyl-5-chloroanthranil (37 %).2-Nitroacetophenone on the other hand gave no cyclised product the only product isolated being diethyl N-(2-acetylphenyl)phosphoramidate (1 0 %),22 thus suggesting the intermediacy of the corresponding phosphorimidate and hence a nitrene (Scheme 23). Although Scheme 23 possible extension of this synthesis of anthranils has not received a great deal of attention there is some evidence that it is not of wide appli~ability.~~ Formation of Phenothiazines.-All the cyclisation reactions so far described have resulted in five-membered rings even in those cases where the possibility of six- membered ring formation also exist. In accord with this phenoxazine and di- hydrophenazine were not isolated from the reactions of 2-nitrodiphenyl ether and 2-nitrodiphenylamine with triethyl phosphite respecti~ely,~ although in these cases the tarry products should be further investigated.The corresponding reaction (23) of 2-nitrodiphenyl sulphide on the other hand gave a moderately good yield of phenothiazine (54%) together with a small amount of N-ethyl- phenothiazine formed by alkylation of the former and it appeared that satis- factory route to phenothiazines had been found.37 A more detailed examination of the reaction has recently revealed the opera- tion of a new molecular rearrangement in this reaction however. Thus while 4-methyl-2-nitrodiphenyl sulphide gives the expected 2-methylphenothiazine (36 %) and the corresponding N-ethyl derivative (25 %) the isomeric 4'-methyl-2- 37 J. I. G. Cadogan R. K. Mackie and M. J. Todd Chem. Comm. 1966,491. 38 Altaf-Ur-Rahman and A. J.Boulton Tetrahedron 1966 Suppl. 7 49. 240 Cadogan nit rodip hen yl sulp hide gives 3 -met h ylp heno t hiazine and its N-e t h y 1 derivative instead of the expected 2-methyl-derivatives (Scheme 24). Similarly 4’-t-butyl-2- Scheme 24 nitrodiphenyl sulphide gives 3- rather than 2-t-butylphenothiazine [equation (24)] while 4’-chloro-2-nitrodiphenyl sulphide gives 3- rather than 2-chloro- phenothiazine. These observations suggest that the six-membered ring is being formed after rearrangement of a five-membered intermediate formed by electro- philic attack at the electron-rich 1’-position (Scheme 25). It is probable although not yet established that the ring-closure involving cyclisation on to an unsubstituted ring also proceeds by such a route which is similar to that involved in the Hayashi rearrangement of carboxybenzophenones in strong acid39 (Scheme 26).It will be of interest to determine whether the reported thermally 0 ‘ 0 Scheme 26 39 Cf. R. B. Sandin R. Melby R. Crawford and D. McGreer J . Amcr. Chem. Soc. 1956 78 3817. 24 1 Reduction of Nitro- and Nitroso-compowtds by Tervalent Phosphorus Reagents induced cyclisation of 2-azidodiphenyl sulphide to phen~thiazine~~ also pro- ceeds via a similar rearrangement. Thus the formation of phenothiazines in the phosphite-induced deoxygenation conforms with other successful cyclisations in that five-membered ring forma- tion is preferred in the first instance. It is significant that the success of this cyclisation is reagent-dependent; thus reduction by the highly nucleophilic tributylphosphine and hexaethylphos- p hor ous t r iamide or with tripheny lp hosp hine or diet h ylmet h ylp hosphoni te gave tars only.In the last case dilution with hexadecane gave phenothiazine in 9% yield. Dilution thus appears to have a beneficial effect on the yield of cyclised products and in the cases of 3-t-butyl- 2-chloro- and 3-chloro-phenothiazines dilution increased the yields in each case (55 to 74; trace to 55; and 63 % respec- tively). The absence of bi-a-cumyl in these cases suggests the absence of triplet nitrene by comparison with the reaction of 2’-nitro-2,4,6-trimethylbiphenyl discussed above. It is possible that in the case of the phenothiazine a nitrene generated in a singlet state (see below) undergoes the cyclisation before d e activation to the triplet state occurs. Whereas in the case of 2’-nitro-2,4,6- trimethylbiphenyl where cyclisation is not the preferred reaction deactivation to the triplet state is competitive giving rise to some hydrogen abstraction from the solvent.Deoxygenatfon of 0-Nitroalkylbenzenes and of Nitrobenzene.-Sundberg has reported that treatment of 0-methyl- o-propyl- o-butyl- o-cyclohexyl-,18 and o-ethyl-nitrobenzene28 with excess of boiling triethyl phosphite [equation (25)] gives the corresponding triethyl N-alkylphosphorimidate (59) as the major identified product (ca. 35-50%). In addition minor amounts of products o-R.C,HdNO + (EtO),P -+ (EtO)sP-N.C,H,(o-R) (Equation 25) ascribed to abstraction and insertion reactions of intermediate nitrenes were detected. Thus o-propylnitrobenzene gave 2-methylindoline (7 %) o-propyl- aniline (6 %) and o-allylaniline (6 %); o-butylnitrobenzene gave 2-ethylindoline (9 “A 1,2,3,4-tetrahydr0-2-methylquinoline (2 %) and compounds believed to be isomeric butenylanilines (ca.5 %) ; while o-cyclohexylnitrobenzene gave a mixture of amines (16 %) containing cis- and trans-1,2,3,4,4a,9a-hexahydro- carbazole (Scheme 27). These results are considered to indicate the intermediacy of a nitrene in these deoxygenations a conclusion c o h e d by the recent work by Smolinsky and on the triethyl phosphite-induced reduction of optically active 2-nitro-(2’-methylbutyl)benzene (60; X = NO& in an extension of earlier work41 involving the pyrolysis of the corresponding 2-azido-derivative (60; X = N$. The latter reaction carried out in the vapour phase and in solu- B. B. Brown R. K. Putney R.F. Reinisch and P. A. S . Smith J. Amer. Chem. SOC. 1953 B. I. Feuer and G. Smolinsky J. Amer. Chem. Soc. 1964 86 3085. 75 6335. 242 Cadogan W o R > a K + mMe + rR (R=Me,Et) [ R = E t ) + unsaturated alkylanilines Scheme 21 tion gave active 2-ethyl-2-methylindoline (61) in each case thus suggesting direct insertion of a singlet nitrene into the G H bond at the 2-position of the side chain via a transition state such as (62). The alternative reaction involving triplet nitrene via radical abstraction to give the intermediate (63) followed by recombination would have led to extensive racemisation (Scheme 28). Scheme 28 It is noteworthy that a significantly greater degree of retention of configuration was observed in the vapour-phase reaction suggesting as would be expected that collisional deactivation of the singlet nitrene to the triplet occurred in the liquid phase followed by reaction via intermediate (63).However it appears that even if pyrolysis of an azide does give a singlet nitrene this will only react as such in the presence of a suitably constituted side chain since thermal de- composition of p-methoxyphenylazide in ~ u m e n e ~ ~ produces substantial yields of bi-a-cumyl a product which can only arise via radical intermediates. These observations are particularly relevant to the interpretation of some of the phosphite-induced deoxygenations of nitrobiaryls and nitrodiphenyl sulphides referred to above. In the case of the deoxygenation by triethyl phosphite of active 2-nitro-(2’- methylbuty1)benzene (60; X =NO& to which we return the product was partially active (ca.50 %) 2-ethyl-2-methylindoline (61) thus paralelling the results and hence the conclusions obtained from the pyrolysis of the corre- sponding azide in solution. Still further support for the concept of the inter- mediacy of a nitrene in these reactions comes from a recent reinvestigationa2b P. Walker and W. A. Waters J. Chem. Soc. 1962 1632. 243 Reduction of Nitro- and Nitroso-compounds by Tervalent Phosphorus Reagents of the deoxygenation by triethyl phosphite of o-methyl- and o-ethyl-nitrobenzene. It will be recalled that Sundberg1*s2a reported good yields of the triethyl N-alkyl- phosphorimidate in these reactions. Reinvestigation22 has confirmed that although these compounds are indeed formed but in a lower yield than pre- viously reported an additional product is present in each case.Thus 2-ethyl- nitrobenzene gives diethyl 2-ethyl-3-H-azepin-7-ylphosphonate (64; 21 %) and ethylene (ca. 50% based on the azepine). These most interesting products pre- sumably arise by a mechanism similar to that suggested above for the formation of diethylaminoazepines with the difference that triethyl phosphite acts as the nucleophile in this case to give a phosphonium intermediate (65) which then eliminates ethylene rather than undergo Arbusov-type isomerisation to give a C- or N-ethylated product (Scheme 29). The other products of the reaction were Scheme 29 triethyl N-(2-ethylphenyl)phosphorimidate (66) its hydrolysis product diethyl N-2-ethylpheny1)phosphoramidate (67) and diethyl N-ethyl-N-(2-ethylphenyl)- phosphoramidate (68).(EtO),P = NAr -+ (Et O),P(O)NH Ar (EtO),P(O)NEtAr (66) (67) (68) (Ar = o.EtC,H,) The reaction of o-nitrotoluene with triethyl phosphite followed a similar course as did that of nitrobenzene. In the latter case for example diethyl N-phenylphosphoramidate diethyl N-ethyl-N-phenylphosphoramidate diethyl 3H-azepin-7-ylphosphonate and ethylene were formed control experiments having established that triethyl N-phenylphosphorimidate is partially converted into the diethyl N-ethyl-N-phenylphosphoramidate but not into the corre- sponding azepin-7-ylphosphonate. This is in contrast to earlier failure to detect identifiable products in this r e a c t i ~ n . ~ ~ ~ ~ These observations coupled with the isolation of bi-a-cumyl (8%) from the reaction of o-nitroethylbenzene with triethyl phosphite in cumene,22b therefore provide a considerable body of circumstantial evidence in favour of the participa- tion of nitrenes in the reactions described in this section.Reactions of Bifunctional Aromatic Nitro-compounds.-Reactions of trialkyl phosphites with o-dinitrobenzene proceed mainly without deoxygenation to give 244 Cadogan dialkyl o-nitrophenylphosphonates (69 ; 75-85 %) and the alkyl nitrite?Sa Although a small quantity of triethyl phosphate is formed indicative of some deoxygenation as a competing side reaction the corresponding reduction products of the nitro-compound have not been isolated. The simplest route to the products of this reaction involves direct aromatic nucleophilic substitution by the phos- phite (Scheme 30) but a mechanism involving attack on one of the nitro-groups QTEt '6% Scheme 30 followed by rearrangement cannot be excluded at this stage.The reaction has been extended to include other tervalent phosphorus reagents such as diethyl methylphosphonite and ethyl diphenylphosphonite which give the correspond- ing o-nitrophenyl-phosphinate and -phosphine oxide.43 The reaction is of interest not only because it affords convenient routes to compounds containing o-nitrophenyl-phosphorus bonds but also because established examples of heterolytic aromatic substitution by phosphorus compounds are rare. It is also of interest in that it possibly sheds light on observations previously recorded but unexplained. Thus Horner and KliipfelM showed that o-dinitrobenzene and triethylphosphine gave a 1 :1-adduct of unknown structure while the reaction of triphenylphosphine with 4-nitropyridine 1 -oxide at 200" has been reported to lead to the evolution of nitrous fumes the fate of the remainder of the molecule being unkn0wn.4~ In view of the foregoing it is obviously possible these reactions proceed via displacement of a nitro-group from the aromatic ring.Triphenyl- phosphine and o-dinitrobenzene on the other hand are reported to give triphenylphosphine oxide after reaction in boiling benzene.8 The nitro-group displacement is most successful with o-dinitrobenzene. rn- and p-Dinitrobenzenes and o-chloronitrobenzene react with a variety of tervalent phosphorus compounds to give a complex mixture of products which although they have not been resolved yet do not include compounds such as diethyl 0- or p-nitrophenylphosphonate.It is also significant that Griffin and Obrycki have ob~erved"~ photolytic oxidation of triethyl phosphite by o-halogenonitrobenzenes but identification of the transformation products of the latter was not possible. In these cases the many possible competing side reactions e.g. formation of azepine amines. phosphorimidates and phosphoramidates are probably favoured. 4-Methyl-2'-nitrophenyl phenyl sulphoxine on the other hand gives a low yield (ca. 5 %) of diethyl o-nitrophenylphosphonate in addition to the pheno- 43 (a) J. I. G. Cadogan D. J. Sears and D. M. Smith Chem. Comm. 1966,491 ; (b) Unpub- lished results. 44 L. Horner and K. Kliipfel Annalen 1955 591 69. 45 E. Howard and W. F. Olszewski J. Amer. Chem. Soc. 1959 81 1483. 46 C. E. Griffin and K. Obrycki J. Org.Chem. 1968,33 632. 245 Reduction of Nitro- and Nitroso-compounds by Tervalent Phosphorus Reagents thiazine (5 %) on reaction with triethyl phosphite (Scheme 31).43b 1,2,4-Trinitro- benzene as might be expected gives a good yield of diethyl 2,4-dinitrophenyl- phosphonate on reaction with triethyl phosphite and the same compound is produced in lower yield from the corresponding reaction of 2,4dinitrochloro- benzene.& b. 0 Scheme 31 An interesting displacement closely related to those described above has been reported recently by Sieper.*' Treatment of 2H-2-(2-nitrophenyl)-naphtho [ 1,8-de]-1,2,3-triazine (70) with triethyl phosphite gave diethyl o-nitrophenyl- phosphonate and the ethylated triazines (71) and (72) in addition to the expected te t r a-azapent alene (7 3) (Scheme 3 2).Diet h y 1 o-ni tr op hen y Ip hos p honate was also produced as a side product in the reaction of triethyl phosphite with 3,4-dihydro-4-oxo- 1,2,3-ben~otriazine.~~ + + Scheme 32 Reduction of Aliphatic Nitro- and Nitroso-compounds Fewer reactions of aliphatic than of aromatic nitro- and nitroso-compounds with tervalent organophosphorus reagents have been reported. Simple nitro- alkanes** do not react at low temperatures and reactions at higher temperatures have not been investigated. Reactions involving the more reactive halogeno- nitroalkanes have been reported however although in some cases the low accountances of identified products lead to uncertainty over the participation of O7 H. Sieper Tetrahedron Letters 1967 1987. 48 S. Trippett B. J. Walker and H. Hoffmann J. Chem.SOC. 1965 7140. 246 Cadogan the nitro-gr0up.4~ Allen50 has shown that 2-chloro-2-nitropropane reacts with triethyl phosphite to give diethyl isopropylideneaminophosphate (74) triethyl phosphate and ethyl chloride. It is not known whether reduction to the nitroso- compound fist occurs followed by reaction of this with another molecule of phosphite it having been shown that 2-chloronitrosopropane and triethyl phosphite also give the same product (74) but much more easily or whether deoxygenation of an intermediate is involved (Scheme 33). Alternative routes involving nucleophilic attack on halogen to give an intermediate (RO),PCl+ Me,CNO,- cannot be discounted. A 0 79 ?- (EtO)3P*b0 =&-CMez -+ (EtO),P(O)Or;l= CMe + EtCl (EtOl,P 7) 1 (EtOi,i’\ O=&-CMc -+ (EtO),P(O)ON =CMez (74) Scheme 33 In an extension of these reactions it has been shown that gem-halogeno- nitr~so-~l and -nitr~-~~cycloalkanes give rise to intermediates which undergo the Beckrnann rearrangement.Thus at room temperature the nitroso-compound reacts to give a good yield of the corresponding lactam presumably as outlined in Scheme 34. The related nitro-compounds react similarly but at higher tem- peratures and again there is no evidence for or against the participation of the nitroso-compound (Scheme 35). / I = 1,2,3,4 8 + Ph,PO Scheme 34 Scheme 35 Certain 1 -bromo-1 -nitro-alkanes including 1 -bromo-1 -nitro-octane l-bromo- 1-nitropropane bromonitrophenylmethane and ethyl bromonitroacetate react $@ A E. Arbusov €3. A. Arbusov and B. P. Lugovkin Bull. Acad. Sci. U.R.S.S. CIasse xi.chim. 1947 538; G. Kamai Doklady Akad. Nauk S.S.S.R. 1951 79 795. so J. F. Allen J. Amer. Chem. SOC. 1957 79 3071. 61 M. Ohno and I. Sakai Tetrahedron Letters 1965,4541. 6a M. Ohno and N. Kawabe Tetrahedron Letters 1966 3935. 247 Reduction of Nitro- and Nitroso-compounds by Tervalent Phosphorus Reagents with triphenylphosphine to give the phosphine oxide and the corresponding nitrile possibly via attack on the aci-form of the nitro-compound (Scheme 36),53 but it is more likely that the reaction proceeds via initial nucleophilic attack on bromine as suggested by Speziale and Smith54 (Scheme 37). No evidence for the Scheme 36 P- P hj P-L Br -Gk- NO __I) P h,PB r+ R C H = ?- O- 4 0- + - + RCN RC 2 N- 0 t- Ph3P70-N= C - R ?-I H Scheme 37 intermediacy of the postulated nitrile oxide was found which is not surprising in view of the now known reactivity of triphenylphosphine towards nitrile oxides.55 Bromonitromethane and 1-bromo-1-nitroethane on the other hand give the corresponding or-hydroxyiminoalkylphosphonium bromides [equation (26)] and the change in the course of the reaction following increase in size of + 2Ph3P + RCHBraNO -+ Ph3P0 + Ph,PCR=NOH Br- (Equation 26) the alkyl group is clearly a point of interest.The possibility that both processes proceed via reaction (equation 26) as a first step followed in the case of R = Me and Et by addition (equation 27) of triphenylphosphine to the nitrile oxide - ..a + - + PhSP RCE N- 0 + Ph,PCR=N - 0 (Equation 27) probably can be discounted because of the relative ease with which triphenyl- phosphine reduces nitrile oxides although quantitative data would be necessary completely to rule out this alternati~e.~~ Smolinsky and F e ~ e r ~ have isolated phenylacetonitrile from the phosphite reduction of 1 -nitro-2-phenylethane thus recalling the corresponding reduction of P-nitrostyrene referred to above.That phenylacetaldoxime also gives the same product suggests in this case also that prior reduction of the mi-form of the nitro-compound occurs [reaction (28)]. PhCH,CH,NO + PhCH,*CH=N(O)OH -+ PhCH,*CH=NOH -+ PhCH,C_N (Equation 28) 53 S. Trippett and D. M. Walker J . Chem. SOC. 1960 2976. 54 A. J. Speziale and L. R. Smith J. Amer. Chem. Soc. 1962 84 1868. 55P. Griinanger Atti Accad. naz. Lincei 1964 36 387 (Chem. Abs. 1965 62 3973); C. Grundman and H. D. Frommeld J. Org. Chem. 1965 30 2077.248 Cadogan Appendix added in Proof It is known5s that reaction of 5-dimethylaminobenzofuroxan with nitrous acid gives 4-dimethylamino-7-nitrobenzofurazan (76). From this fact the formation of an intermediate nitroso-compound (75) was inferred this compound then being assumed to undergo rapid rearrangement to the furazan (76) (Scheme 38). I 0- Extension of this idea has led to evidence which tends to support the transient intermediacy of a nitroso-compound in the phosphite reduction of 3-methyl-7- nitroanthranil (77). 57 In this case the major product is 4-acetylbenzofuroxan (78) (Scheme 39). The analogy with Scheme 38 is obvious. In the absenceof Scheme 39 instances of actual isolation of an intermediate nitroso-compound in phosphite reductions of nitro-compounds this together with the observed formation of phenylacetonitrile in the reduction of 1 -nitr0-2-phenylethane~~ mentioned above is the best evidence so far produced for the participation of nitroso- intermediates in the phosphite reduction.Photochemically induced phosphite reduction of aromatic nitro-compounds (79) at room temperature has now been reported.58 In general yields of products are low and it is also stated that cyclisation of 2-nitrobiphenyl and 2-nitro- stilbene under these conditions occurs in lower but unstated yield compared with the thermal cyclisation. In all the cases (79) studied triethyl N-arylphos- phorimidates (80) are formed in moderate yields. o-Methyl nitro-compounds also give the corresponding N-aryl-2-acetimidylpyridines (8 1). The results of the reactions are summarised in Scheme 40 and Table 3.It will be recalled that products of this type (81) had been isolated previously from the reaction 68 A. J. Boulton P. B. Ghosh and A. R. Katritzky J. Chem. SOC. (B) 1966 1004. 68 R. J. Sundberg W. G . Adams R. H. Smith and D. E. Blackburn Tetrahedron Letters 1968 777. A. J. Boulton I. J. Fletcher and A. R. Katritzky Chem. Comm. 1968 62. 249 Reduction of Nitro- and Nitroso-compounds by Tervalent Phosphorus Reagents of nitroso-compounds with triethyl phosphit&* and that a not very satisfactory rationalisation of the mode of formation had been advanced (Scheme 3). The new and very significant experimental data presented by Sundberg and his co-workers68 now clearly indicates that the skelatal rearrangement to give the pyridine derivative is more complex than previoiisly suspected in that it involves the formation of a C-1 to C-3 link (Scheme 40).On the basis of these results Scheme 3 has been withdrawn and a satisfactory rationalisation is now awaited. Table 3 Products of photochemical phosphite reduction of (79) (Scheme 40) (79) Yield (%) H H H H 1 0 H H Me H <3 0 M e H H H 13 37 Me Me H H 12 18 Me H Me H 51 10 Me H H Me 14 4 H H Me0 H 34 0 R1 R2 R3 R4 (80) (81) Scheme 40 9 (EtO),P 3. ArNO,-+ArNO-+Ar~~O. (OEt) or ArN-P (OEt) azepines 5 -membered ' 0 heterocycles ArN P (OEt) ArN(O):NAr ArN -7 (OE t)3 or ArN .O - h (OEt) e tc. @ + (EtO),PO Scheme 41 250 Cadogan Such a rationalisation should embrace all of the various types of products which have been obtained from phosphite deoxygenations of nitro-compounds. A tentative interim suggestion for consideration while we await more experimental results is outlined in Scheme 41 which includes the new intermediate (83) which could in theory be stabilised when the nitro compound contains an o-methyl group and hence play a more significant part in the reaction. Scheme 42 An extension of the synthesis of indoles by deoxygenation of o-nitrostyrenes has been reported59 and is summarised in Scheme 42. These are the first instances of migration of a /3-substituent in a monosubstituted styrene earlier work having shown that one of the substituents in /Ip-disubstituted o-nitrostyrenes was prone to migration during deoxygenation. ~4 R. J. Sundberg J. Org. Chem. 1968,33 487. 251