AROMATIC REARRANGERIENTS By E. D. HUGHES D.Sc. F.R.I.C. F.R.S. and C. K. INGOLD D.Sc. PH.D. F.R.I.C. P.R.S. (UNIVERSITY COLLEGE LONDON) THIS Review is concerned with rearrangements of the following types x e X - O N R H and where X = C1 Br I N,Ar NO Alk OH NO SO,H and NRAr They are formally all very similar a part of a side-chain replaces hydrogen in the aromatic ring and is replaced by hydrogen in the side-chain in the original side-chain there is only one atom between the migrating group and the ring and it possesses unshared electrons the migrating group enters 0- and p-positions exclusively and the reactions are all catalysed by acids. However when we examine the mechanism of these processes we find that they are not all as similar as might be thought. There is a primary distinction of polar type.Of a number of them it is true that the migrat- ing group moves as an electrophilic fragment separating from the side-chain without the electrons by which it was there bound and uniting with aromatic carbon by means of electrons which the latter has to supply. We shall call these reactions aromatic electrophilic rearrangements ; and the property signalised being treated as a primary classificatory character they will be our concern in Section 1. They might alternatively be called arowuitic cationotropic rearrangements the migrating group being regarded as a poten- tial cation whether or not it ever becomes a free cation this would emphasise the analogy with prototropy or more generally cationotropy in the field of unsaturated rearrangements. Among the rearrangements listed above is represented another class in which the group migrates as a nucleophilic fragment taking with it the electrons with which it was bound in the side-chain and using its own electrons for the purposes of establishing a bond with a carbon kernel in the aromatic ring.These reactions will be called aromatic nucleophilic rearrangements and they will be discussed in Section 2. They might alter- natively be called aromatic anionotropic rearrangements the migrating group being regarded as a potential anion in order to signalise the analogy with anionotropy in the domain of unsaturated rearrangements. I n both these classes of rearrangement it is necessary for classificatory purposes to establish that the migrating group a t some stage becomes suf- ficiently free from the parent molecule to enable the question to be con- sidered as to whether it is moving as an electrophilic or as a nucleophilic fragment.Such rearrangements are often (somewhat illogically) called " intermolecular " rearrangements in order to distinguish them from the class next to be mentioned. These are the aromatic intramolecular rearrange- ments which are to be discussed in Section 3. They proceed through a cyclic transition state and in such circumstances it is never possible to 34 HUGHES AND INGOLD AROMATIC REARRANGEMENTS 35 determine which way round the electrons move during reaction or even whether they move heterolytically in pairs or homolytically by uncoupling and recoupling of the pairs. Indeed the uncertainty principle teaches that the denying of this knowledge to us is one of Nature's ways of making the cyclic transition stage as stable as it is and thus of enabling the intra- molecular reaction to go as easily as it does.Accordingly it is unphysical to try to classify intramolecular rearrangements as exclusively electrophilic or nucleophilic or even as heterolytic or homolytic if they are typically intramolecular they will have all these characters though cases may arise in which one character seems to predominate. 1. Aromatic Electrophilic Rearrangements This class of rearrangements will be illustrated with four examples namely (a) the Orton rearrangement of halogeno-amines (b) the rearrange- ment of diazoamino- to aminoazo-compounds (c) the Fischer-Hepp re- arrangement of nitrosamines and (d) the Hofmann-Martius rearrangement of alkylanilines.Reference will be made in passing to reactions which effect the same overall conversions by dther mechanisms usually not exactly known for example to certain photochemical isomerisations. (la) Rearrangements of Halogeno-amines (Orton).-A typical example is the conversion of N-chloroacetanilide into a mixture of o- and p-chloro- acetanilide in the presence of hydrochloric acid and usually in hydroxylic solvents such as acetic acid or water or aqueous acetic acid C,H,*NClAc .-+ (o- and p-)Cl*C,H,*NHAc Up to 1909 the side-chain-to-nucleus migrations of halogen which this example illustrates were regarded as true intramolecular rearrangements. But in that year a different view of the reaction was advanced by Orton and Jones.2 This was that it commenced with a reversible acidolysis of the N-chloro-compound to give acetanilide and elemental chlorine and that then the latter attacked the former in an ordinary process of aromatic C-chlorination HCI (1) (3) (2) C,H,*NClAc + HC1 + C,H,*NHAc + C1 + ( O - p-)CI*C,H,*NHAc + HCl This interpretation though it has often been attacked holds good today; while there is naturally more that can be added.In the early work of Orton and Jones it was pointed out that the overall isomeric change is specifically catalysed by hydrochloric acid there was but little general acid catalysis in the investigated conditions. Acetanilide was isolated from a solution in which N-chloroacetanilide was undergoing rearrangement. Elemental chlorine was aspirated from such a solution. And by going over from N-chloroacetanilide to a derivative with a deacti- vated ring N-chloro-2 4-dichloroacetanilide for example it was found Bender Ber.1886 19 2272 ; Chattaway and Orton J. 1899 75 1046 ; Arm- strong J. 1900 77 1047. 2 Orton and Jones Proc. 1909 25 196 233 305 ; J. 1909 95 1456 ; Brit. Assoc. Reps. 1910 85 ; Orton ibid. 1911 94 ; 1912 116 ; 1913 136 ; 1914 105 ; 1915 82 Orton and King J. 1911 99 1185. 36 QUARTERLY REVIEWS possible to use the chlorine liberated from it to chlorinate some other more reactive aromatic ring for instance that of acetanilide itself or that of anisole. that the proportions in which o- and p-chloroacetanilide are formed are the same whether the starting materials are N-chloroacetanilide and hydrochloric acid or acetanilide and chlorine provided that the solvent is the same. (The proportions are somewhat dependent on the solvent 67% of the p-isomer is formed in pure acetic acid but only 59% of p-compound in 50% aqueous acetic acid.) Let us rewrite the Orton mechanism labelling the component reactions (l) (2) (3) as is done above but adding as reaction (4) the hypothetical intramolecular rearrangement It was subsequently shown Here a mildly subtle point arises.C,H5*NCIAc + HCL. .. *'%. ( 4 ) w ] i w (o- p-)CI-C,H,*NHAc + HCI C,H5*NHAc + C1 Then if the equilibrium (1)-(2) were always established much more rapidly than the isomerisation is observed to proceed we could not distinguish between routes (3) and (4) for the formation of the rearrangement products. For reaction (4) would have the same factors products and rate laws as reaction sequence (1 + 3) and reaction (3) would have the same factors products and rate laws as reaction sequence (2 + 4).However fortun- ately reactions (1) and (2) are not always or even usually very fast in comparison with reaction (3). The equilibrium (1)-(2) depends much on the solvent ; in water the stable system is PhNClAc + HC1 as is doubt- less determined by the strong ionic solvation of HCl; but in acetic acid the stable system is PhNHAc + Cl,. The rate of reactions of form (3) can be varied over a great range by introducing substituents into the aromatic ring. In their original work Orton and Jones showed that in aqueous acetic acid containing less than 65% of the acid the rate of isomerisation of N-chloroacetanilide is less than the rate of C-chlorination of acetanilide by chlorine. This means that reaction (1) is at least partly rate-determining.An incursion of the intramolecular process (4) would make the total rate of isomerisation greater than the rate of the chlorination. Soper subse- quently showed * that in water as solvent reaction (1) becomes wholly rate-determining the rate of isomerisation of N-chloroacetanilide in the presence of hydrochloric acid being just equal to the rate of the production of chlorine. Orton Soper and Williams found several ring-substituted acetanilides namely o- m- and p-chloroacetanilide p-bromoacetanilide and aceto-o- and aceto-p-toluidide for which the rates of reactions (2) and (3) were both measurable in the medium they were to use ; and by employ- ing a strongly aqueous medium 40% acetic acid they secured that the rate of reaction (1) would be negligibly small in comparison then starting with the acetanilide and chlorine they showed in each case that the ratio of the 5 Orton Soper and Williams J.1928 998. Orton and Bradfield J. 1927 986. Soper J . Phys. Chem. 1927 31 1392. HUGHES AND INGOLD AROMATIC REARRANGEMENTS 37 N - to the C-chlorinated product was independent of the time thus satisfying Wegscheider’s test for the simultaneity of two reactions of the same order reactions ( 2 ) and (3) were simultaneous and (3) was not being mistaken for (2) followed by (4). This completes Orton’s case for the intermolecular mechanism (1)-(2)-(3) as against the intramolecular mechanism (4) for the chloro-amine rearrangement in hydroxylic solvents. Confirmation has been furnished by Olson and his collaborators,6 who have examined the isomerisation of N-chloroacetanilide in the presence of hydrochloric acid isotopically labelled with radiochlorine.They found that the final ring-bound chlorine had approximately the radioactivity it would possess if it had once been pooled with the inorganic chlorine in the medium. As remarked already hydrochloric acid is a specific reagent for the isomeric transformation of N-chloroacetanilide. However other halogen acids bring about transformations if not the isomeric one hydrobromic acid produces o- and p-bromoacetanilide and hydriodic acid gives o- and p-iodoacetanilide.7 This can be understood on the basis of Orton’s mechan- ism since the intermediate halogens would in these cases be bromine monochloride and iodine monochloride which are respectively brominating and iodinating agents.Of the three reactions (l) (2) and (3) involved in the Orton mechanism reaction (3) needs little comment aromatic C-chlorination having been discussed in other connections.8 We know that molecular chlorine has a combination of thermodynamic stability and electrophilic reactivity that makes i t a very effective chlorinating agent ; so much so that if we want to observe chlorination by a specifically more reactive though less stable reagent such as C1+ or ClOH,+ we have to take steps carefully to remove every trace of molecular chlorine. The effectiveness of chlorine for chlorina- tion is we may believe one of the reasons for the importance of the Orton mechanism in chloramine rearrangements. From the qualitative fact of the specificity of hydrochloric acid it follows that if this substance reacts as its ions as seems probable in view of the highly aqueous condi- tions to which much of the evidence of mechanism applies then the reaction needs both ions.This conclusion is supported by the kinetics the reaction is of third orderYg that is first with respect to the chloramine and second with respect to hydrochloric acid or if we prefer the ionic interpretation first in chloramine first in hydrogen ion and first in chloride ion Rate cc [ chIoramine][HCl] a K [chloramine][ H+][Cl-] Actually Richardson and Soper However reaction (1) deserves further comment. have confirmed the ionic interpretation 6 Olson Porter Long and Halford J . Arner. Chem. SOC. 1936 58 2467 ; Olson Halford and Hornel ibid. 1937 59 1613 ; Olson and Hornel J . Org.Chem. 1938,3 76. 7 Bradfield Orton and Roberts J. 1928 782 ; Richardson and Soper J. 1929 1873. 8Soper and Smith J. 1926 1582; de la Mare Hughes and Vernon Research 1950 3 192 242. Blanksma Rec. Trav. chirn. 1903 22 290 ; Rivett 2. physikal. Chem. 1913 82 201 ; Harned and Seltz J . Amer. Chem. SOC. 1922 44 1475 ; Soper and Pryde J. 1927 2761 ; Dawson and Millet J. 1932 1920. 38 QUARTERLY REVIEWS in the following way. They examined the kinetics of the conversion of N-chloroacetanilide by means of hydrogen bromide into 0- and p-bromo- acetanilide in various essentially aqueous media under conditions in which reaction ( l ) leading to bromine monochloride is rate-determining while reaction (3) in which the aromatic ring is brominated with liberation of hydrochloric acid is instantaneous C,H,*NCIAc +HBr 4 C,H,.NHAc +BrCl -+ Br.C,H,.NHAc +HCl The rate obeyed the expression Rate oc [chloramine][H+][Br-] which cannot be put into an alternative molecular form because two acids are supplying the proton while one has the outstandingly reactive anion namely bromide ion.On this evidence we can plausibly regard reaction (1) as a bimolecular nucleophilic substitution by halide ion a t the chlorine atom of a chlor- ammonium ion a kind of XN2 substitution in an 'onium salt but at chlorine instead of a t carbon slow fast r(+ Hal' + CL-NHAcAr - Hal-CL + NHAcAr This being accepted it follows that reaction (2) is a bimolecular nucleo- philic substitution by the acylanilide molecule a t one halogen atom in the halogen molecule. Reaction (1) is then certainly not a hydrolysis as it is sometimes loosely called.But Soper and Pryde showed9 that it is accompanied in aqueous solutions by 6 comparatively unimportant side-reaction of the nature of hydrolysis this gives hypochlorous acid which in the presence of hydrochloric acid undergoes further conversion into chlorine. Their evidence was that the rate of acidolytic displacement of halogen from N-chloroacetanilide by aqueous solutions of acids other than the halogen acids increased with the acid strength and was identical for the strong acids nitric sulphuric and perchloric acids clearly this was a hydrogen ion reaction the anions of the strong acids taking no part. However if such a reaction were assumed for hydrochloric acid it would account at most for a few units per cent of the observed rate of chlorine production.The mechanism of this relatively slow hydrolytic process is not known. It might as before be a SN2-type substitution in the chlorammonium ion but with water as the substituting agent H,O + CL-NHAcAr - H&CL + NHAcAr P + Or it might be a SN1-type substitution P + + CI-NHAcAr + CL + NHAcAr + H20 + C+L - HZO-Cl Clearly it is still correct to speak of hydrochloric acid as a specific catalyst for the rearrangement not only because the acidolysis with the aid of HUGHES AND INGOLD AROMATIC REARRANGEMENTS 39 chloride ion is so much faster than that involving water but also because in the absence of any chloride ion the product is hypochlorous acid a poor chlorinating agent of itself and one insufficiently converted in equi- librium into active cationic forms to produce a chlorinating agent of efficiency comparable to the chlorine which would be given by chloride ion.In 1912 Orton wrote lo of the chloramine rearrangement " Whether a true intramolecular change is possible under certain conditions has not yet been discovered but it must not be supposed that the possibility is excluded." Orton was one of the earliest workers on reaction mechanism explicitly to repudiate the assumption still apparent in some modern writ- ings that no reaction can have more than one mechanism. And the position with respect to the chloramine rearrangement remains to this day almost as he expressed it. For halogeno-amine rearrangements in aprotic solvents such as chloro- benzene evidence for a one-stage intramolecular process with general acid catalysis has been claimed by Bell l1 in the examples of N-chloroacetanilide N - bromoacetanilide N-bromobenzanilide and N-iodoformanilide.It is contended that the concentration of halogen detected in the system would not account for the observed reaction rates. But according to Soper and his collaborators,12 inadequate account is taken of the formation of acyl- hypohalites HalOAc which as they have shown in this and in other con- nections are undoubtedly produced in such conditions and are good halogenating agents. Dewar l3 supports Bell but with inconclusive argu- ments adding the rider that the rearrangement exemplifies his '' n-bond " theory." But neither party has yet established the unique adequacy of its own view and so the matter remains where Orton left it intramolecular rearrangement is a possibility.On the other hand the proposition that mechanisms other than the Orton mechanism exist even if we do not know exactly what they are can be unequivocally supported. For example the existence of some kind of homolytic mechanism is made clear by the known effect of light in promoting the transformation of N - chloroacetanilide. (16) Rearrangements of Diazoamino-compounds.-The standard illustra- tion is the conversion of diazoaminobenzene into p-aminoazobcnzene HC1 or The reaction can be effected for example by treatment with alcoholic 10 Orton Brit. Assoc. Reps. 1912 116. l1 Bell Proc. Roy. Soc. 1934 A 143 377 ; Bell and Levinge ibid. 1935 A 151 211 ; Bell J. 1936 1154 ; Bell and Brown J. 1936 1520 ; Bell and Danckwerts J. 1939 1774. laIsrael Tuck and Soper J. 1945 547.l3 Dewar " Electronic Theory of Organic Chemistry " Oxford Univ. Press 1949 14Blanskma Rec. Trav. chim. 1902 21 366; Mathews and Williams J. Amer. * This is that the mobile group travels freely round the aromatic ?I shell attached p. 225. Chem. Soc. 1923 45 2574. by a " a-bond " before coming to rest at an o- or p-position. 40 QUARTERLY REVIEWS hydrochloric acid or better by treatment with aniline together with aniline hydrochloride or some other salt of aniline.15 In most examples of the change the azo-group migrates to the p-position. o-Migration is less facile ; but it does occur if the p-position is blocked. Although this reaction has not been investigated as fully and accurately as has the chloramine rearrangement it has had a much less controversial history since 1885 no one seems seriously to have doubted that the diazoamino-aminoazo-conversion is intermolecular.In that year Friswell and Greenl5 advanced the view that the acid-catalysed reaction went in stages an acidolysis which reverses the usual mode of formation of a diazoamino-compound being succeeded by an ordinary process of aromatic diazo-coupling. Their mechanism is formulated below in a way which leaves open the question t o which we shall return of whether (as in the Orton mechanism) the anion of the catalysing acid is directly utilised in the acidolysis or is not so utilised. In other words the question left open is whether the reaction is of second or of first order with respect to the catalysing acid ; or to paraphrase again whether the primary acidolysis product is the covalent diazo-pseudo-salt or the ionised diazonium salt The earliest observation of special significance in relation to the ques- tion of mechanism was that of Nietzski,16 who demonstrated the trans- ference of the diazo-group from a rearranging diazoamino-compound to a foreign aniline molecule by treating p-diazoaminotoluene with the hydro- chloride of aniline or of o-tohidine he obtained the transfer products It was subsequently shown l7 that the foreign molecule receiving the trans- ferred azo-group need not be an aromatic amine but could be a phenol from diazoaminobenzene and phenol p-phenylazophenol and aniline were obtained Similar transfers have been observed l8 when diazoaminobenzene is treated with the hydrochlorides of m-toluidine or of dimethylaniline.It has been noticed l9 that when diazoaminobenzene is rearranged with alcoholic hydro- l5 Griess and Martius 2.Chem. 1866 2 132 ; Kekul6 ibid. p. 688 ; Witt and l6 Nietzski Ber. 1877 10 662. 17 Hermann and Oeconomides Ber. 1887 20 272 ; Fischer and Wimmer ibid. p. 1579; Kidd J . Org. Chem. 1937 2 198. 18 Meyer Ber. 1921 54 2265 ; Rosenhauer and Unger Rer. 1928 61 392. 19Ear1 Ber. 1930 63 1666. Thomas J. 1883 43 112; Friswell and Green J. 1885 47 917. HUGHES AND INGOLD AROMATIC REARRANGEMENTS 41 chloric acid a by-product of the constitution C6H,*N:N*NH.C6H,*N:N.C,H is produced which has obviously arisen from the transfer of a diazo-group from an acidolysing molecule either to an unaltered or to a fully rearranged molecule. A modification of the Friswell-Green mechanism has been developed by Heinrich Goldschmidt,20 particularly as an interpretation of the marked facilitating effect on which all observers agree of aniline and similar bases on the reaction catalysed by such bases in association with acids.In Goldschmidt’s mechanism the aromatic base is assumed to act in just the way in which we allowed that the anion of the catalysing acid might act in the Friswell-Green mechanism ; but when aniline fulfils this function there is no uncertainty about whether a covalent or ionic azo-compound is going to be formed the product is covalent and is the final product so that the second stage of the general Friswell-Green mechanism disappears ON + H,” \ +N*NH2*C6H5 + H,N’ \ N*C,H5 + NH,*C,H + H+ 6) II N*C6H5 The Goldschmidt mechanism can thus be regarded as a particular case of the Friswell-Green mechanism the general acid catalyst HX having been specialised to the anilinium ion Ph*NH + this in essence was Goldschmidt’s final view.The theory implies that the Ii’riswell-Green mechanism of the diazoamino-rearrangement is able to assume a form completely analogous to that of the Orton mechanism of the chloramine rearrangement. Just as in stage (1) of the latter the acidolysis of chlorine was assumed to involve a SN2-like substitution by chloride ion a t the chlorine atom of a chlor- ammonium ion so in stage (1) of the diazoamino-rearrangement the acido- lysis of the azo-group is represented as involving a XN2-like substitution by a nucleophilic conjugate-base a t the azo-group of an azoammonium ion. The evidence on the matter is kinetic and is due entirely to Goldschmidt and his co-workers.21 They used mainly aniline or some other such base as their solvent.They observed the reaction to be of first order with respect t o the diazoamino-compound and to be subject to general acid catalysis. With the strong acids hydrochloric hydrobromic and nitric acid as catalysts the reaction was approximately of first order with respect to the acid; but the three acids had not quite the same absolute kinetic effect and careful examination of the matter showed that a part of the reaction was of second order with respect to acid. Then when these strong acids were replaced by successively weaker acids namely 3 5-dinitro- benzoic o-nitrobenzoic m-nitrobenzoic and o-bromobenzoic acid the order with respect to the catalysing acid rose so that with the weakest acid at not too low concentration the overall order was more nearly two than one.All this is consistent with the view that for the formation of the transition 20 Goldschmidt Ber. 1891 24 2317 ; Goldschmidt and Bardach Ber. 1893 25 1347 ; Goldschmidt and Reinders 2. physikal. Chem. 1896 29 1369 1899 ; Gold- echmidt Johnsen and Overwien ibid. 1924 110 251. Goldschmidt and Salcher 2. physikal. Chem. 1899 29 89 ; Goldschmidt Johnsen and Overwien Eoc. cit. 2 1 Goldschmidt and Reinders Zocc. cit. ; 42 QUARTERLY REVIEWS state of the acidolysis there is needed the diazoamino-compound a proton and a nucleophile ; that when strong acids having weakly nucleophilic anions are used the nucleophilic function is fulfilled mainly by the solvent aniline though halide ions do intervene in place of the aniline to a small extent ; but that when weak acids having strongly nucleophilic anions are employed these anions intervene in place of the aniline to a much larger extent.This interpretation makes the kinetics and mechanism entirely analogous to those of the chloramine rearrangement. (lc) Rearrangements of Nitrosamines (Fischer-Hepp).-It. was discovered by Otto Fischer and Hepp 2 2 that certain aromatic nitrosamines undergo rearrangement on treatment with acids particularly hydrochloric and hydrobromic acids to give ring-nitrosated isomerides as in the following example HC1 C,H,*NMe*NO + p-NO*C,H,*NHMe The main products are usually p-nitroso-compounds in the simpler examples of the benzene series but N-alkyl-N-nitroso-2-naphthylamines give N-alkyl- 1 -nitroso-2-naphthylamines.23 The reaction is usually carried out with ethyl-alcoholic hydrogen chloride or bromide as catalyst ; but ethyl ether acetic acid and water have been used as solvents instead of alcohol. Fischer and Hepp believed their isomerisations to be true intramolecular rearrangements. But in 1912 Fischer showed 24 that the halogenated by- products which are usually formed can be understood as arising from the action of free halogen produced by oxidation of the catalysing halogen acid by nitrous acid liberated during the reaction. In 1913 Houben showed25 that in certain examples in which the yield of rearrangement product was poor a much improved yield could be secured by adding sodium nitrite to the reacting system. As to the mechanism of rearrange- ments the evidential value of these observations is of course very slight ; but on such evidence Houben set up the theory that the reaction is inter- molecular consisting of an acidolytic denitrosation reversing the ordinary method of formation of the nitrosamine followed by direct ring-nitrosation of the formed secondary amine by the simultaneously formed nitrosyl halide or perhaps by some conversion product of the latter such as nitrous acid Such further evidence as has since been secured has tended to confirm this theory.Neber and Rauscher found that hydrogen chloride and bromide are much more effective catalysts than are other strong acids in agreement with general experience to the effect that nitrosyl chloride and bromide have a combination of stability and reactivity which makes them particu- larly useful nitrosating agents.26 It has been found that on acidolysis of 22 Fischer and Hepp Ber.1886 19 2991. 231dem Ber. 1887 20 1247 2471; Morgan and Evens J. 1919 115 1142. 24Fischer Ber. 1912 45 1098. p6 Neber amd Rauscher Annalen 1942 550 182. 25 Houben Ber. 1913 46 3984. HUGHES AND INGOLD AROMATIC REARRANGEMENTS 43 an aromatic nitrosamine in the presence of urea no C-nitroso-isomeride is produced but only the secondary amine.27 Various transfers of the nitroso- group to a foreign aromatic molecule have been reported. When N-methyl- N-nitrosoaniline is treated with ethyl-alcoholic hydrogen chloride in the presence of dimethylaniline thc products are methylaniline and p-nitroso- dimethylaniline. 26 Corresponding products are formed when N-methyl- 2 4-dinitro-N-nitrosoaniline is treated with ethereal hydrogen chloride in the presence of dimethylaniline.28 What is now needed in order to establish the mechanism firmly is a kinetic study of the rearrangement and as far as possible of the separate reactions (l) (2) and (3) just as in the example of the chloramine rearrangement.(Id) Remangements of Alkylanilines (Hohann-Martius).-The re- arrangements which the hydrochlorides and hydrobromides of ,iV-alkyl- anilines and NN-dialkylanilines undergo by thermal decomposition to give salts of ring-alkylated secondary or primary aniline derivatives were discovered by Hofmann 29 HX HX C,H,*NR + (o- or p-)R*C,H,*NHR C,H,*NHR -++ (o- or p-)R.C6H,*NH Our further knowledge of them is due mainly to Hickinbottom. The required temperatures are usually high around 250-300" when the alkyl groups are primary though temperatures below 200' may suffice for the displacement of secondary and tertiary alkyl groups.The alkyl groups enter mainly into p-positions if such are free ; but o-migration will occur if the p-position is occupied ; and a minor proportion of o-migration may accompany p-migration. Thus N-methylaniline when rearranged by heat- ing its hydrobromide gives salts of p-toluidine and a little ~-toluidine.~O Polyalkylation may occur not only in the successive conversions of a ter- tiary amine through secondary amines into primary amines but also in conversions starting from secondary amines. In the latter case there must be alkyl transfer ; for whereas before the change every molecule of base contained one alkyl group after the change some molecules contain none and some two or more.Thus N-n-butylaniline rearranged through its hydrochloride yields p-n- butylaniline as the principal basic product while as by-products aniline and N p-di-n-butylaniline are found together with smaller amounts of more highly butylated anilines. 31 Hickinbottom and his co-workers have shown that alkyl halides and for ethyl and higher alkyl groups olefins are produced in the course of rearrangement-they can be drawn off and identified-naturally with a diminished yield in the actual rearrangement. 32 These investigators have also shown that alkyl groups *'Macmillen and Reade J. 1929 585. s8G1azer Hughes Ingold James Jones and Roberts J. 1950 2657. 29 Hofmann and Martius Ber. 1871 4 742 ; Hofmann Ber. 1872 5 704 720 ; 3O Hickinbottom J. 1934 1700.alReilly and Hickinbottom J. 1920 117 103. 3aHickinbottom and Ryder J . 1931 1281. 1874 7 526. 44 QUARTERLY REVIEWS isomerise during rearrangement in just the way to be expected if they should pass through a carbonium ionic form. Thus when N-isobutylaniline is rearranged as its hydrobromide isobutyl bromide and isobutylene can be drawn off but the rearrangement product is p-tert.-b~tylaniline.~~ And when N-isoamylaniline is similarly rearranged the products are isoamyl bromide trimethylethylene and p-te~t.-amylaniline.~* It is to be noted that the alkyl group in the alkyl bromide is not rearranged but that the olefin and the alkyl group in the C-alkylaniline are rearranged. Hickin- bottom has shown that the easily ionising alkyl halide triphenylmethyl chloride can be used to introduce the triphenylmethyl group into the p-position of dimeth~laniline.~~ Finally he has shown that the various olefins encountered in the study of the rearrangement can be condensed with ar@ine in the presence of its hydrobromide under conditions fairly similar to those of the rearrangements to give the actual products of the rearrangements isobutylene for example yielding p-tert.-butylaniline.36 The Hofmann-Martius rearrangement was originally regarded as intra- molecular and this view still has its adherents. Dewar supports it,37 adding that the reactions exemplify his " n-bond " theory of rearrangements.* His main argument is that a formed alkylating agent would lead primarily to polyalkylation because of the activating effect of alkyl groups on ben- zenoid reactivity.However one has to remember that the most reactive position the p-position is occupied first and that the o-positions possibly from steric causes are considerably less reactive. The opposite view that the reaction is intermolecular was first sug- gested by Mi~hael.~8 He thought of the alkyl halides as the active inter- mediates and this idea has received support since.39 Hickinbottom has suggested 35 that the alkyl group is split off from the anilinium ion as a carbonium ion which may then undergo various independent reactions combining with halide ion to give the alkyl halide losing a proton to yield an olefin and attacking the benzene ring to give the rearrangement product. The carbonium ion would react in its internally rearranged form if it is one of those which usually do so.This theory explains all the facts except one namely that when an alkyl group undergoes internal rearrangement during migration it may appear in its unrearranged form in the isolated alkyl halide though it is represented entirely by its rearranged form in the olefin and in the p-alkylaniline. However a mechanism can be suggested which is a combination of those of Michael and of Hickinbottom and which takes account of our general knowledge of nucleophilic substitution and elimination. Recalling that the halide ion is strongly nucleophilic towards carbon but not towards hydrogen it is assumed that the decomposition of 33 Hickinbottom and Preston J. 1930 1566. 34 Hickinbottom J. 1932 2396. 3s.Idem J . 1935 1279; 1937 404. 37 Dewar " Electronic Theory of Organio Chemistry " Oxford Univ.Press 1949 3*Michael Ber. 1881 14 2105; J . Amer. Chem. Xoc. 1920 42 787. 30 Beckrnann and Correns Ber. 1922 55 852. * S e e footnote p. 39. 351&rn J. 1934 1700. p. 227. HUGHES AND INGOLD AROMAT1.C REARRANGEMENTS 45 the anilinium salt in the high concentrations used proceeds by the XN2 mechanism - + Ph*NH,R + Hal -+ Pkt-NH + RHal This makes the first step of the rearrangement analogous to that of the Orton rearrangement and probably also to those of the diazoamino-rearrangement and the Fischer-Hepp rearrangement. It is to be noted that if R is the kind of alkyl group which is ultimately to suffer an internal rearrangement as of isobutyl to tert.-butyl it would not yet be rearranged in the alkyl halide. Next noting that an anilinium salt at a high temperature is to be regarded as a highly polar medium it is suggested that the alkyl halide attacks the aniline by a X,1 process that is by way of an intermediate carbonium ion which is also the source of the olefin formed according to this theory in a reversible side-reaction RHal + R+ + Hal- R+ + Ph-NH + Olefin + Ph*NH,+ R+ + Ph*NH + R-C,H,*NH + H+ H+ + Ph*NH + Ph-NH,+ If the carbonium ion is one which normally isomerises any final product formed through it will have the rearranged alkyl structure.It was discovered by Reilly and Hickinbottom 31 that when N-alkyl- anilines are heated with certain metal halides such as cobaltous and zinc chlorides rearrangement occurs the alkyl group migrating to the ring as in the Hofmann-Martius reaction. However this Reilly-Hickinbottom reaction as we may call it exhibits certain notable differences from the Hofmann-Martius reaction.Alkyl halides are not evolved under Reilly- Hickinbottom conditions ; 4O and neither are 01efins.~~ And alkylanilines such as N-isoamylaniline which by the Hofmann-Martius method would give products with an internally rearranged alkyl group p-tert.-amylaniline in the case cited when treated by the Reilly-Hickinbottom method give products with unrearranged alkyl groups p-isoamylaniline in the present example. 33 2 34 Except that the Reilly-Hickinbottom reaction cannot con- tain a S,1 stage there is little that can be said about its mechanism the function of the metal being at present unknown. 2. Aromatic Nucleophilic Rearrangements The existence of this class of " intermolecular " aromatic rearrangements has only recently been recognised.The leading example is the conversion of arylhydroxylamines under the influence of acids into o- and p-aminophenols. (2a) Rearrangements of Hy&oxylamhes.-When phenylhydroxylamine is treated with dilute aqueous sulphuric acid p-aminophenol is the chief product as was first observed by Bamberger 42 acid C,H,*NH*OH + p-OH*C,H,*NH2 4 O Hickinbottom J. 1927 64. 41 Hickinbottom and Waine J . 1930 1558 ; Hickinbottom J. 1937 1119. 42 Bamberger Ber. 1894 27 1347 1548. 46 QUARTERLY REVIEWS The subsequent investigation of this and of a number of closely related reactions is due chiefly to Bamberger.43 The question of the mechanism of these processes has evoked contrary opinions. Bamberger 44 regarded them as proceeding in an " intermolecu- lar " manner through a univalent nitrogen intermediate APN.His reasons will be mentioned later ; and his conclusion as we shall then see comes fairly close to what we believe today. The question of polar classification did not arise when Bamberger propounded this theory in 1921. Much more recently in 1949 Dewar l3 classified the rearrangements first as intramolecular and secondly as electrophilic with the corollary that they illustrate his " n-bond " theory.* His reasons were that tEe transference of hydroxyl from an arylhydroxylamine to a foreign amine or phenol has not been achieved and that the production of hydrogen peroxide during rearrangement in aqueous solution has not been observed. However the conclusions do not follow from the evidence; and they are at variance with what we believe today.Our present view is indeed the opposite namely that the reactions are " intermolecular " and nucleophilic (and nothing to do with the n-bond theory). With the customary ellipsis of allowing single valency structures to stand for mesomeric molecules this view 46 may be formulated for a pcbra-rearrangement as follows + In the strictly water molecule isomeric change the nucleophilic reagent Y would be a ; but when closely related non-isomeric substitutions with rearrangement are taken into account Y would be any sufficiently acces- sible and reactive nucleophilic molecule or anion. In conformity with the acid catalysis of the reactions they are formulated as starting from the ionic conjugate acid of the arylhydroxylamine which is here written in the form in which it would undergo the indicated heterolysis rather than in its probably more stable form with the extra proton carried by nitrogen.The heterolysis product represented by the second formula is mesomeric having its carbonium ionic charge not only as indicated at the p-position but also in the o-positions ; so that by the use of a different valency struc- ture for the carbonium ion the formation in the general case of o- as well as of p-products can be accommodated. The transition from the third to the fourth formula represents an ordinary prototropic change here written without reference to mechanism. When Y contains a hydroxyl group a proton can of course be lost from the last product formulated. 43 Bamberger Zocc. cit. and many subsequent papers including three summarising articles Annalen 1921 424 233 297 ; 1925 441 207.44 Idem ibid. 1921 424 233. 45 Yukaws J . Chern. Soc. Japan 1950 '71 603 ; Heller Hughes and Ingold Nature 1951 168 909; 1952 169 80. *See footnote p. 39. HUGHES AND INGOLD AROMATIC REARRANGEMENTS 47 The first two steps as written above express a heterolysis to give a mesomeric carbonium ion which subsequently takes up a nucleophilie reagent in a position other than that of the heterolysis this is a familiar form of change a unimolecular nucleophilic. substitution with rearrange- ment aN1’. It is conceivable that in some circumstances the same two steps would become telescoped into a single step so that the change would be a bimolecular nucleophilic substitution with rearrangement SN2’. The mechanism in this form would be expressed thus + Actually a decision between these two forms of the nucleophilic mechanism can at present only tentatively be made in favour of the X,l’ form as the more usual really crucial distinguishing tests have not yet been applied.The present evidence for this mechanism is derived from a study of the products and kinetics of the reaction. The significant work on products was done many years ago chiefly by Bamberger.46 When phenylhydroxyl- amine was rearranged by means of dilute aqueous sulphuric acid p-amino- phenol was the main product. However when ethyl alcohol was used to dilute the acid 0- and p-phenetidines and with methyl alcohol anisidines were formed; and when the acid employed was hydrochloric acid 0- and p-chloroanilines were produced. When phenol was added the product con- tained p-OH*C,H,*C,H,*NH,-p and some C,H,*NH*C,H,*OH-p and when aniline was introduced i t contained some C,H,*NH*C,H,*NH,-p.Formed paminophenol was in some cases accompanied by the ether (p-NH,*C,H,),O. The fact that so many fragments can appear in place-of OH in the aryl- hydroxylamine-aminophenol conversion strongly suggests that the latter is not an intramolecular rearrangement. Furthermore since all the frag- ments come from obvious nucleophiles Y = H,O EtOH MeOH C1- Ph*OH Ph*NH, NH,-C,H,-OH-p one is led to assume an active electro- philic intermediate. Bamberger made an extensive comparison between the products obtained from arylhydroxylamines and those produced by nitrogen loss from the corresponding aryl azides in similar conditions. He found the two sets of products to be essentially (and strikingly) identical and was accordingly led to assume a common univalent-nitrogen intermediate ArON.Now if we supply this intermediate with the extra proton shown by the kinetics to be involved i t becomes Ar-NH which is only another valency structure for the mesomeric carbonium ion already assumed as the active electrophilic intermediate. By rearranging p-tolylhydroxylamine in aqueous acid at low tempera- tures Bamberger demonstrated the formation of the compounds + CH3\/_’,\ INH and H O W 46 Bamberger locc. cit. D 48 QUARTERLY REVIEWS along with products derived from them. This is where the already written sequence of reactions representing the nucleophilic mechanism would have to stop when the final prototropic change is blocked by niethyl substitution. It should be mentioned ,that Bamberger usually found aniline and azoxybenzene among the products obtained from phenylhydroxylamine.However these substances probably arose from an irrelevant oxidation- reduction of phenylhydroxylamine. This is one of the points established by a recent kinetic study of the The redox conversion is a chain-reaction easily started by short exposures to atmospheric oxygen but avoidable by arranging that the phenylhydroxylamine has never suffered exposure to oxygen. The acid-catalysed rearrangements of phenylhydroxylamine are completely separate and are not dependent on oxygen or other oxidants. The kinetic study also shows that rearrangement depends on the con- jugate acid of phenylhydroxylamine. At low acidities the rate is propor- tional to the acidity; but it ceases to increase proportionally to the acid when enough acid has been added to ionise nearly the whole of the base.Obviously the matter needs further study yet i t seems a reasonable presumption that the general character of the arylhydroxylamine rearrange- ment has been correctly outlined. 3. Aromatic Intramolecular Rearrangements These reactions will be illustrated by three groups of examples) namely (a) the acid-catalysed rearrangements of arylnitramines ) ( b ) those of aryl- sulphamic acids and ( c ) those of 1 2-diarylhydrazinesY with reference in the last case chiefly to the production of diphenyl derivatives. (3a) Rearrangements of Nitramines.-It was found by Bamberger that phenylnitramine methylphenylnitramine and similar arylnitramines under- go rearrangement on treatment with aqueous strong acids or with hydrogen chloride in organic solvents to yield mainly o-nitroaniline or its derivatives sometimes with a small amount of p-nitroaniline or its derivatives 47 Having found in a parallel series of researches that treatment of primary and secondary amines with neutral or not strongly acid nitrating agents such as dinitrogen pentoxide will often lead to N-nitration he made the suggestion 48 that the aromatic C-nitration of these amines by strongly acidic nitrating agents consists of a N-nitration followed by an acid- catalysed intramolecular rearrangement of the N-nitro-compound to the C-nitro- compound.Now this hypothesis of " indirect nitration " as it has been called was obviously not necessitated by the facts. It would have been equally pos- 47 Bamberger and Landsteiner Ber.1893 26 485 ; Bamberger Ber. 1894 27 48 Idem Ber. 1894 27 584 ; 1895 28 399. 359; 1897 30 1248. HUGHES AND INGOLD AROMATIC REARRANGEMENTS 49 sible on the same facts to set up the a1ternat:ve hypothesis that in analogy with the chloramine rearrangement or the di; ,zoamino-aminoazo-rearrange- ment for example the nitramine rearrangern ent is not intramolecular but is one of the so-called " intermolecular rear] angements " that is a coni- posite process consisting of an acidolysis o-l the N-nitro-group to give a nitrating agent followed by participation 1)f the latter in an ordinary aromatic nitration. In terms of the schenie written below instead of assuming " indirect nitration " that is that reaction (3) is really (2 + a) Bamberger might have assumed " intermolecular rearrangement " that is that (4) is really (1 + 3) C,H,*NH + N0,X' A third alternative hypothesis is equally open namely that both the prc- ceding assumptions are incorrect and that reactions (3) and (4) both exist as independent processes.Holleman Hartogs and van der Linden a t ,empted to test the hypothesis of " indirect nitration " by measuring the pro ?ortions of o- m- and p-nitro- products as obtained by the nitration of aniline and by the action of acids on ~henylnitramine.~g They obtained the results given in the upper part of the Table and concluded that the C-nitrat on of aniline does not always proceed by way of an initial N-nitration. ':'he caution apparent in this statement reflects the circumstances that thc > conditions of nitration and rearrangement were not identical and that the proportions of isomers formed by nitration are sensitive to the condiiions.However Hughes and G. T. Jones have conducted similar experim3nts in which the same con- ditions were used for nitration and for rea.*rangement.5* Their results given in the lower part of the Table leave no doubt that reactions (3) and (4) are essentially independent processes. Proportions of nitro-compounds formed by nitration of aniline Process Conditions 0 - 9 % m- % Nitration Ph*NH,}NO, 95% aq. H,SO, - 20" . 4 39 Y 9 9 80% aq. HNO . . 5 32 9 9 ,Ac,O . . 82 3 Nitration Ph*NH,}NO, 85% aq. H,SO, 1C" . 6 34 Rearrangement Ph*NH*NO, 85% aq. H,SO, loo . . 93 0 by rearrangement of pheny nitramine Holleman Hartogs and Linden (1911) Rearrangement Ph*NH*NO, 74% aq.H,SO, - :!Oo . 95 1.5 Hughes and Jones (1950) and P- % 56 62 15 3.5 59 7 Orton attempted to test the second alternai ive hypothesis namely that the isomerisation of the nitramine might be a] I " intermolecular rearrange- ment ". He had for guidance his own work on the chloramine rearrange- 49Holleman Hartogs and van der Lindeii Ber. 1911 44 704. SoHughes and Jones J. 1950 2678. 50 QUARTERLY REVIEWS ment. However he and his collaborators quickly found that the nitramine rearrangement was very different.lO 51 They noted that it was subject to general acid catalysis. Their main conclusion was that although in special cases nitramines in acid solution could be observed to nitrate a foreign aromatic compound thus proving that a nitrating agent is present in these conditions " no nitrating agent invariably and normally appears in the system in which a nitramine is undergoing isomeric change ',.Hughes and Jones 50 have continued the investigation in two examples which illustrating complementary kinetic situations further elucidate Orton's statement. Their first example was that of X-methyl-p-nitro- phenylnitramine which underwent rearrangement to N-methyl-2 4-dinitro- aniline p-NO,*C ,H,.NMe-NO -+ 24 1 -(NO,)& ,H,-NHMe in the presence of a variety of acids ranging in strength from formic acid to sulphuric acid and in a number of solvents including water ethyl alcohol acetic acid and ethyl ethdr. However they found that under no conditions could any denitration of the nitramine be detected either by the formation of N-methyl-p-nitroaniline in the presence of an easily nitrated foreign substance or by the actual nitration of the added substance.The second example was that of 2 4-dinitrophenylmethylnitramine which suffered rearrangement to N-methyl-2 4 6-trinitroaniline + either in 80% aqueous sulphuric acid or in pure sulphuric acid. They found that the nitramine readily underwent denitration in these conditions as shown both by the isolation of the denitration product N-mcthyl-2 4- dinitroaniline in the presence of added easily nitratable substances such as p-xylene phenol or dimethylaniline and also by the isolation of nitra- tion products of such added materials. However they found that neither the produced nitrating agent nor nitric acid added in equivalent amount was able to nitrate the denitration product N-methyl-2 4-dinitroaniline7 to the rearrangement product N-methyl-2 4 6-trinitroaniline under the conditions in which the rearrangement itself readily took place.Thus in terms of a.scheme of the type of that given on p. 49 H~ighes and Jones's first case established that reaction (4) could not be replaced by (1 -1- 3) because (1) was too slow while their second case proved that it could not when (1) was fast enough because (3) was too slow. It follows that of the three hypotheses originally open we have to choose the third reactions (l) (2) (3) and (4) all exist. That is a nitrating agent can be produced under the conditions of rearrangement but the rearrangement is not dependent on it. It is of interest to consider how it arises that in the acid-catalysed nitramine rearrangement an intramolecular isomerisation plays the impor- tant role in contrast to the chloramine the diazoamino-aminoazo- the Pischer-Hepp the Hofmann-Martius and the arylhydroxylamine rearrange- 51Orton and Pearson J.1908 93 725; Orton Chem. News 1912 106 236; 2 4 l-(NO,),C,H,.NMe*NO 2 4 6 l-(NO,),C,H,*NHMe Bradfield and Orton J. 1929 915. HUGHES AND INGOLD AROMATIC REARRANGEMENTS 51 nients. It would seem that the nitramine retrrangement has an especially facile intramolecular route denied to the other rearrangements. It may be suggested that this is so because the structure of the nitro-group in a nitramine admits of isomerisation to a nitri joamine and thence through a cyclic transition state to an o-nitroaniline. ?'he isomerisations are assumed to take place in the ionic conjugate acids and it will be understood that in the formulz written below t,he curved arrows are given an arbitrary direc- tion they might have been turned the opyosite way or both ways in accordance with the already noted principle tl tat an intramolecular change proceeding through a cyclic transition state is never exclusively electro- philic or nucleophilic or homolytic but deriires some of its facility from a simultaneous possession of the three charazters This theory accounts for the special irnportancc of o-migration in the nitra- mine rearrangement as well as in the rearrang:ement next to be discussed.(36) Rearrangements of Sulphamic Acids.- -The literature of these re- arrangements is so deeply involved with that of ;he sulphonation of aromatic amines that it is convenient to introduce the subject by outlining part of the latter.52 Two practical processes for the sulphonation of aniline are to be dis- tinguished.One involves the treatment of aniline with sulphuric oleum. The second the so-called "baking process " depends on the action of regulated heat on first-formed anilinium hydrogen sulphate. A theoretical point has to be appreciated rlamely that sulphonation being a readily reversible reaction high-tempe qature products tend to be thermodynamically final products their formati m is determined by thermo- dynamic stability independently of mechanism. The experimental evidence indicates that the thermodynamic end-product of the monosulphonation- desulphonation equilibrium involving aniline is siilphanilic acid. It matters neither how the equilibrium is set up nor how riany processes it may con- tain.Sulphanilic acid is rapidly formed from aniline or from orthanilic acid and sulphuric acid a t temperatures near 181)". According to the view stated its production has no bearing on orientation considered as a kinetic phenomenon and none on the mechanism of silphonation. As to products formed a t lower temperatures the main facts of present concern are as follows. From aniline by the 1,aking process orthanilic metanilic and sulphanilic acids are formed thc ugh with relatively little 5 2 Suter " Organic Chemistry of Sulfiir " Wilcy jgew York 1943 p. 245. 52 QUARTERLY REVIEVS metanilic acid. From aniline by the oleum process much more metanilic acid is produced. Dimethylaniline by the oleum method yields a still higher proportion of the m-sulphonic acid.Dimethylaniline has not yet been shown to yield any o-sulphonic acid. Now Bamberger had a theory of what has been called the “indirect sulphonation ” of aniline and similar bases.53 It assumed that anilinium hydrogen sulphate is first dehydrated (reaction 1 below) either by heat or by oleum to phenylsulphamic acid which a t low temperatures experiences a rearrangement (reaction 2 ) to orthanilic acid while this a t high tempera- tures undergoes another rearrangement (reaction 3) to sulphanilic acid Bamberger Hindermann and Kunz reported 53 first that phenylsulphamic acid on treatment with a trace of sulphuric acid in acetic acid just above the freezing point of the solution passed into orthanilic acid (reaction 2 ) ; and secondly that orthanilic acid when heated with excess of sulphuric acid at 180” gave sulphanilic acid (reaction 3).Subsequent discussions of the mechanism of sulphonation of aniline bases have largely taken either of two opposite points of view. One casts doubt on Bamberger’s scheme as a whole for the quite weighty reason that Bam- berger never demonstrated reaction (l) that no one else has succeeded in doing so and that it has been doubted whether phenylsulphamic acid could arise under the conditions of the baking pr0cess.5~ The other line which has been taken is essentially to accept Bamberger’s scheme as a whole associating it chiefly with the baking process; while allowing that direct sulphonation of the anilinium ion is a reaction of main importance in the oleum process and there and elsewhere is responsible for the production of m-sulphonic a~ids.~5 The differences between the baking and the oleum process of sulphonation of aniline and between the behaviour of aniline and emethylaniline in the oleum process can thus largely be understood.It is now easy to extract from this material what is relevant for the sulphamic acid rearrangement. We have not to decide whether Bam- berger’s scheme is right or wrong as a whole. We can admit that reaction (1) has not been demonstrated ; and we can even add that the demonstration of reaction (3) gives no information about mechanism. The fact remains that Bamberger reported the isomerisation (2) under conditions in which aniline would not have been sulphonated. This isomerisation if genuine,* must be an intramolecular rearrangement. 6 3 Bamberger and Hindermann Ber.1897 30 654 ; Bamberger and Kunz ibid. p. 2274. 65For example Alexander J . Amer. Chem. SOC. 1946 68 969; 1947 69 1599. * The caution in the above statement is occasioned by the circumstance that no later report has been found to confirm either directly or indirectly this early observa- tion which should and will be checked. 64For example Huber Helv. Chim. Actu 1932 15 1372. HUGHES AND INGOLD AROMATIC REARRANGEMENTS 53 We ask what property of this particular jystem could make an intra- molecular rearrangement especially facile. 01’ the groups so far considered which migrate from side-chain to nucleus naniely Hal NzR NO Alk OH NOz SO,H the last two are the only ones whi:h have a structure admitting migration by the mechanism illustrated on p. 51. And these groups are the only ones in the list whose migrations appear to depend on intramole- cular processes.The suggested mechanism 2 ,s already illustrated divides the total intramolecular process into two sterel )chemically convenient steps necessarily leading overall to o-migration. Al;ain these two groups are the only ones whose reported migrations are esseiitially pure o-migrations. If we accept these indications as significant the wlphamic acid rearrangement can be written thus It is noteworthy that the assumed first step or the rearrangement the con- version of a sulphamic acid into a sulphitoamine reverses the probable last step of another known reaction namely the formation of phenylsulphamic acid from phenylhydroxylamine and sulphur dioxide. In the representa- tion of the second step of the rearrangement the arrows indicate only one of several ways in which the electrons might move to form the cyclic tran- sition state which as in similar cases mentioned already must be con- sidered to owe some of its stability and ease of formation to the existence of such alternatives.(3c) The Benzidine Rearrangement.-Hyd razo benzene was discovered and its conversion into benzidine under the influence of acids was first observed by Hofmann 56 in 1863. Benzidine was known already Fittig 57 having identified it as a diaminodiphenyl. The positions of its amino- groups were established later by S~hultz.5~ 5lchmidt and Schultz 59 noted the formation along with benzidine of a minor proportion of a second diaminodiphenyl so-called diphenyline ; and with Strasser 6O they deter- mined the positions of its amino-groups.The further investigation of these and related rearrangements is due chiefly to Jacobson,61 who established the formation not indeed from hydrazobencene but from many other benzenoid hydrazo-compounds of two otl ler types of isomerisation 66 Hofmann Proc. Roy. SOC. 1863 12 576. 57 Fittig Annalen 1862 124 282. 59 Schmidt and Schultz Ber. 1878 11 1754. 6o Idem Annalen 1881 207 320 ; idem and Strasser ibid. p. 348. 61 Jacobson many papers from 1892 to 1922 i I cluding the summarising paper 58 Schultz ibid. 1874 174 227. ibid. 1922 428 76. 54 QUARTERLY REVIEWS products. These were both aminodiphenylamines and are usually called o- and p-semidines. Under the name “ benzidine rearrangement ” it is customary to sum- marise the whole family of rearrangements in which aromatic hydrazo- compounds on treatment with acids yield either diaminophenyls or amino- diphenylamines by 0- or p-coupling of the two arylamine residues of which the hydrazo-compound can be considered composed.The possible pro- ducts are 2 2’- 2 4’- and 4 4‘-diaminodiphenyls respectively known as o- benzidines diphenylines and benzidines and 2- and 4-aminodiphenyl- amines known as o- and p-semidines. Not all the products arise in any one example but the types can all be found among known examples and hence their relation may be represented by the omnibus scheme below in which A and B stand for substituents in general NH o - Semidine p-Semidine o-Benzictine Diphenyline Benzidine The acids used to effect these changes are often dilute aqueous or aqueous-alcoholic solutions of strong acids such as hydrochloric or sulphuric acid.Hydrogen chloride in an organic solvent is sometimes employed. With such strong acids the rearrangements are usually rapid. Some of them can be effected by weak acids such as acetic acid. A method which has been considerably employed on account of its convenience for investi- gating the products formed by the rearrangements of hydrazo-compounds side-steps the actual preparation of the hydrazo-compounds and proceeds by reducing the related usually easily prepared azo-compounds with an acid reducing agent the hydrazo-compound is produced and is at once rearranged though a certain amount of reductive fission into primary amines often comes into competition with the processes of rearrangement. When hydrazobenzene itself is rearranged the formed mixture of iso- merides contains about 70% of benzidine and 30% of diphenyline.It has never been established that any o-benzidine or o- or p-semidine is pro- duced. No o-benzidine has ever been detected among the rearrangement products of hydrazo-compounds of the benzene series though o-semidines are commonly and p-semidines are not infrequently encountered. In the naphthalene series on the contrary products of o- benzidine type appear but no product of diphenyline type has ever been detected. Substituents play a large part in determining the course followed by the rearrangement of a hydrazobenzene. They would play a larger part, HUGHES AND INGOLD AROMATIC REARRANGEMENTS 55 but for the circumstance that these rearran,;ements are very " strong " processes often able to cause the ejection of a substituent which stands in their way.The groups SO,H and C0,H we thus ejected more easily than most groups C1 and OAc rather less easily OR less easily still and NRAc NR, and Alk not a t all. For example hydrazobenzene-4-carboxylic acid gives benzidine in high yield with loss c f the carboxyl group while 4-acetoxyhydrazobenzene gives benzidine only in low yield the rearrange- ment being largely diverted by the substitiient in the direction of a diphenyline which retains the substituent. The introduction of one or two substituenls into o-positions or of one or two into m-positions in a hydrazobenzene has the effect of reducing minor products of rearrangement and increasir g the yield of the benzidine derivative. It is p-substituents which produc:e the large changes in the direction of reaction.A single p-substituent if not ejected blocks the formation of a benzidine ; and according t c its nature it may render either diphenylines or o-seniidines or p-seniidines the chief products. Although as Jacobson emphasises allowances must be made for difficul- ties of separation his results indicate that the ,;ingle p-substituents NMe, Hal OAc favour diphenylines while OR Me favour o-semidines and NHAc NH, favour p-semidines. Assuming that the group NMe acted as NHMe,+ in the relevant experiment a polar series may possibly be recognised here the regularity being that the more electronegative and less electropositive groups favour substitution c n m-carbon rather than on p-nitrogen while the most electropositive groups have the reverse effect.Thus we pass along the series from favoured di1)henylines and o-semidines in the formation of either of which the ahead,{ substituted ring receives additionally a C-substituent to p-semidines ir i the production of which the nitrogen bound to the substituted ring receives the substituent. Two firm-standing p-substituents in a hydrazobenzer e block all the commonly observed rearrangements except that leading to an o-semidine. These points are illustrated in the Table on next page which reproduces some of Jacobson's qualitative indications of the relati. re amounts in which the different rearrangement products arise. Three types of theory have been advanced with reference to the mechan- ism of the benzidine change. The first assumes Tihat we should now under- stand as a preliminary homolysis of the N-N bold of the hydrazine.The original author of this theory is Tichwinsky,62 .,vho wrote i t as follows C,H,.hTH C,H,*NH,,HCl C6H5'NH C,H,.NH,,HC1 The precise mode of incursion of the acid is an adjuritable detail in this theory and it is not necessary to consider detailed vari;,tions. What is essential to the theory is the assumed dissociation of the hydrazo-compound into free radicals whose radical centres become transferred to p - or o- positions. Jacobson advanced 61 two strong arguments agtinst theories of this type. I + 2HC1 -+ 2(*C6H,*NH,,HC1) + [ 6 2 Tichwinsky J . Russ. Phys. Chem. SOC. 1903 35 667. 56 QUARTERLY REVIEWS Rearrangement products from substituted hydraxobenxenes (Jacobson) Substituents Types of Rearrangement Products r-- 7 A A 2 -Me 2-OEt 2 -Me 2-OMe 2-c1 2-CO,H 3-Me 3-OMe 3-NH 3-Me - 3-OH 3-OH 3-C1 3-cozH 3-SO3H 4-me2 4 -NH 2 4-NHAc 4-Me 4-OMe 4-OAc 4-C1 4-COZH 4-SO3H 4-Me 4-C1 B - - - 2-Me 2-OMe 2-c1 2-C02H - - - 3-NH2 3-Me 3-OH 3-C1 3-C02H 3-SO,H - - - - - - - - - 4-Me 4-C1 Benzidirie 1 111 111 111 111 111 111 111 111 111 111 111 111 111 111 111 111 - - _ ._ - -_ (11) (11) (111) (111) - - o-Semidine p-Semidine - - - ? ? - - - - - - __ 1 - 111 - 111 l l l A - l l l A 11 11A 1 - - H ydrazonaphthalenes.o-Benzidine type. 1 1' 111 111 2 2' - 111 1 The parentheses indicate expulsion of the substituent. 2The A indicates the ring which receives the new C-substituent in semidine - - - - formation. First he cited Wieland's work63 on the tetra-arylhydrazines which are dissociated in solution into free radicals with special facility but do not undergo the benzidine change under conditions in which such homolytic dissociation is known to be considerable Secondly Jacobson pointed out that he had studied the rearrangement of many unsymmetrically substituted hydrazo-compounds AB including a number which having two free p-positions gave benzidines ; and always he had isolated the unsymmetrical benzidine AB rather than either sym- metrical benzidine AA or BB which should have been formed along with AB from dissociated radicals.This argument is strongly reinforced by recently recorded experiments 64 on unsymmetrical hydrazobenzenes so substituted that the absence from the rearrangement product of syrn- metrical benzidines could be proved. Wheland and Schwartz rearranged 2- methyl-2'-ethoxyhydrazobenzene having radiocarbon in the methyl group ; 63 Wieland Annalen 1912 392 127 ; Ber.1915 48 1095. 64 Wheland and Schwartz J. Chem. Phys. 1949 17 425 ; Bloink and Pausecker The C-C coupling is in the 1 1'-positions. (' gH5 )2N*N(C6H5 18 * 2(c sH5hN J. 1950 950. HUGHES AND INBOLD AROMATIC :%EARRANCEMENTS 57 and then they added and subsequently separa;ed ordinary 3 3’-dimethyl- benzidine observing that the recovered material was radiochemically inactive. Bloink and Pausacker rearranged hydrazobei izene-3-carboxylic acid and were able to show that no non-acidic benzidine i v a s contained in the product. The second type of theory assumes a form of N-N splitting which we should now interpret as a heterolysis. Stiegliiz first advanced this idea,65 in the following form C:,H,*NH*NH-C,H + HX + C,H,*&;’< + XH,N*C,H / H,N*C,H + >c,H,:NH + H,N*C,H,*C,H,*NH The essential meaning of this expression beconies clear when it is realised that the conjugate acids of the represented unkralent nitrogen and bivalent carbon intermediates are only different valency structures of the carbonium ionic heterolysis fragment (C,H5:NH)+ the reactive positions of which are assumed to combine with those of the other heterolysis product the aniline molecule.Thus the central idea of the Stieglitz mechanism could today be expressed in a simpler form H+ C6H,-NH-NH0C,H + (C,H,:NH)+ + NH,*(>,H --+ Benzidine efc. Jacobson’s first argument against the homo1 ytic-dissociation theory has no relevancy for this theory. And his second argument is also not par- ticularly damaging here because fragments forr zed by heterolytic dissocia- tion are functionally differentiated like lock and key and so when an unsymmetrical hydrazobenzene AB heterolyse 3 in its preferred direction the functionally complementary parts which w2 may symbolise A B will necessarily unite in a “ lock-and-key ’’ fashion to form the unsymmetrical benzidine AB.However the heterolytic-dissoc iation theory has been dis- proved by rearranging two similar but symmel ricd hydrazobenzenes AA and BB in the same solution two pairs of fimctionally complementary fragments A A B B should certainly unite t l > give not only the sym- metrical benzidines AA and BB but also the ‘crossed” benzidine AB. Experiments on this principle were made pith 2 2‘-dimethoxy- and 2 2’-diethoxy-hydrazobenzene and it was shown by study of the freezing- point diagram that the product was a mixture of dimethoxy- and diethoxy- benzidine containing no third substance of any kind and therefore no ethoxymethoxybenzidine + - + - + - NH NH H+ Meoo OoMe __+ NH-NH XH NH E t O O O O E t \ L ___/ 6 6 Stieglitz Amer.Chem. J. 1903 29 62-63 footnote. 661ngold and Kidd J. 1933 984. 58 QUARTERLY REVIEWS The total effect of these arguments and demonstrations is to show that the benzidine transformation does not involve any kind of preliminary splitting homolytic or heterolytic and is therefore a true intramolecular rearrangement proceeding through a cyclic transition state. The acid catalysis of the benzidine change shows that one or more protons as well as the hydrazobenzene molecule are required to build the transition state.The precise nature of this requirement follows from the kinetics of the process. They were first studied by van Loon,67 whose somewhat rough results suggested what was much more recently established conclusively by Hammond and Shine,68 namely that the reaction has specific hydrogen-ion catalysis with second-order dependence on hydrogen ions Rate cc [PhNH.NHPh][H+l2 Carlin Nelb and Odioso 69 have added the point that since the ratio of formed benzidine to diphenyline (70 30) is independent of t'he acidity the same kinetic equation must hold for t'he formation of each of these pro- ducts. The conclusion is that cach transitlion state is formed from one hydrazobenzene molecule and two protons.70 Our further discussion of mechanism will be restricted to the formation of benzidines and diphenylines.For as yet nothing is known about the kinetics of scmidine formation. They may well be the same; but this cannot be taken for granted. So many rearrangements occur through the same form of prior splitting that when we encounter a true intramolecular process having a cyclic transition state i t seems natural to try to suggest some special reason for the stability of the latter apart from the general reason of ambiguity in the directions of electron displacements which applies to all such cyclic processes. I n the examples discussed in the two preceding Sections the nitramine and the sulphamic acid rearrangements the suggested special reason was " exceptional stereochemical facility " gained through the two-stage mechanism.This cannot be the answer in the example of the benzidine rearrangement. The special reason which has been offered 71 emphasises the quite exceptional amount of extra resonance energy in the transition stage of the benzidine change an amount which i t is assumed is energetically adequate to set up an atomic configuration widely different from that of any normal molecule. I n order the more easily to follow this theory let us first recall in a siniple example the valency-bond description of transition states noting how it accommodates a conjugative orienting effect. Consider a p-substi- tution by a diazonium ion oriented by a dialkylamino-group. The rate of reaction will depend on the stability of the transition state. For fixed nuclear positions in this state the electron distribution can be described 6 7 van Loon Rec.Trav. chin&. 1904 23 62. 68Hammond and Shine J . Amer. Chem. SOC. 1950 72 220. 69 Carlin Nelb and Odioso ibid. 1951 73 1002. 70 Other kinetic investigations have contributed new analytical methods Biilmaiin and Blom J. 1924 125 1719; Dewar J . 1946 777. 71 Hughes and Ingold J . 1941 608. HUGHES AND INGOLD AROMATIC ZEARRANGEMENTS 59 by superposition of the wave-functions of certs in valency structures whose differences express the uncertainty of electron position and therefore deter- mine the electronic energy of the transition sta ,e. As an initial approxima- tion one usually considers only structures such LS (I) and (11) corresponding apart from the nuclear deformation to the factor and product of reaction. (In the present case each structure stands fcr a set of structures since factor and product themselves are mesomeric.; What the orienting effect does in the same approximation is to add strtcture (111) thereby increas- ing the general uncertainty of electron position and in particular making it quite indeterminate as indicated by expressions (IV) and (V) where the electrons come from which bind the diazoniurr.ion a t successive moments during the determinative period of the substii,ution this it is which in accordance with the uncertainty principle sta bilises the transition state ; so much so that the coupling reaction occurs er,sily with the dialkylaniline though it cannot be realised a t all with benzsne (Iv) IV) Proceeding to a description of the transitioii state of benzidine forma- tion we make two preliminary assumptions.'I'he first is that on account of forces still to be described the two benzene sings lie in roughly parallel planes in the transition state. This is indicated by the overall geometrical result of the rearrangement. The second assumption is that the transition state although partly covalent is also largely onic. One reason for this is that even if the two protons which Hammoiid and Shine showed to be included in the transition state are both cova1l:ntly bound a t the outset two protons are set free from covalent attachmelit by the electronic change [as indicated in formulz ( A ) and ( E ) below] therefore from two to four protons must be partly thus set free in t'he tramsition state of rearrange- ment. To this i t can be added that any polarjty involving the aromatic carbon or the nitrogen atoms will automatical1:i be distributed by meso- meric processes so that if ionic character is set up anywhere as it must be according to the argument just given it will become generalised through the system.A valency-bond description of the transition state of rearrangement therefore involves an enumeration of covalent and ionic structures. In some of them the rearranging residues are helcl together a t one end or the other by a covalency they will be termed 1 he " unsplit " structures some of them are covalent and some ionic. In the majority however the residues are held by an electrostatic bond they will be called the " split " 60 QUARTERLY REVIEWS structures and they are all ionic since a structure is thus classified if it has at least one electrovalent bond.With the simplification of allowing as before one structure to do duty for the set describing a benzene ring and in the initial approximation already illustrated we should if it were not for the orienting effect of the amino-groups have to enumerate only two component structures (A) and (E) below. The orienting effect however adds three more structures (B) (C) and (D). Neglecting spatial perspective they may be depicted as follows * We may refer first to the unsplit structures. + + H H H+ H+ - I - + H,N NH -~ 1 H H H+ H + H,N NH 00 H H+ There are octet-preserving routes by which any one of these structures can be converted into any other some involve circulation of the electrons one way round some the other way and some either or both ways.The essence of the theory is that the existence of all this free intercommunica- tion and more still to be described between the different conventional electron distributions and of course between the infinitude of intermediate unconventional ones determines on account of the uncertainty principle a very strong transition state one able and even prone to form itself despite its great difference of shape from that of any normal molecule. The split structures doubtless play a large part in establishing this situation. According to our simplified system of representation we should in the absence of an orienting effect have to enumerate only four split structures namely those obtained from (A) and ( E ) by heterolysis of the central bond in either direction. However the orienting effect introduces many additional structures.All are mutually interconvertible and are interconvertible with the unsplit structures the whole assembly forming a highly elaborate mesomeric system. *For lack of the knowledge which in 1950 Hammond and Shine supplied these structures were originally (1941) written in a slightly generalised form as was explained in the following words " As we do not yet know whether one or two adding protons are included in this [the transition] state we shall omit them and quite formally write negative charges on any atoms which ultimately receive protons " (Hughes and Ingold ref. 71 p. 611). Now that we know definitely that the number of adding protons is two we include them in the structures as rewritten in the present text. HUGHES AND MGOLD AROMATIC RE.LRRANGEMENTS 61 Some of the ionic structures both unsplit t,nd split have properties which seem especially significant for the formatj on of the transition state of rearrangement.Hammick and Mason have j>ointed out 7 2 that if the unsplit ionic structure ( F ) were the only contributing structure one nitrogen atom (that written on the left) would be out of the plane of the adjacent ring with the result that the geometrical paramchers of a normal molecule being assumed the p-carbon atom of that ring mould (but for steric repul- sion) be only 1-50 A from the p - and only 1.51 I I from the o-carbon atom of the other ring. It is true that other structur3s contribute to the tran- sition state which have the property that if any of them was alone relevant the similarly computed distances would be much greater for structure ( A ) for instance 4.3 1(1 from p - to p-carbon and 4-1 from p - to o-carbon.However any contributions by structures such as (P) which tend to a short distance where they have no bond but where a bond is to be established must facilitate the development of the bond. H2N () +I H Y H It has also been emphasised 73 that in all split i;tructures the rearranging fragments are held by an electrostatic bond and t iat in a number of them such as (a) this is located where the new covalmcy is to be established. The significance of the point is that an electrostatic bond provides attrac- tive forces a t greater interatomic distances than does a pure covalency already at 4 A an electrostatic bond can be fairly strong. Thus the form- ing bond could begin its development as an ionic bond but go over into a covalency as it shortens.As we have seen hydrazobenzenes in general we converted more easily into benzidines than into diphenylines and mole easily into these than into o-benzidines which have not in fact been isolated. It is most un- likely that the reason for this is stereochemical. The suggested reason is greater resonance energy in transition states assoziated with p - than with o-rearrangement~.~~ 7 2 More electrons are of n3cessity disturbed by a transition from a benzenoid to a p-quinonoid than by one to an o-quinonoid structure and thus p-quinonoid structures such a;; (B) (C) and (D) intro- duce more resonance energy into the transition litate in which they par- ticipate than do corresponding o- quinonoid structures.In the naphthalene series this difference of resonance energy is largely annulled by the better conjugation of an o- 6r p-quinonoid than of a p - or a-quinonoid ring with the adjoining benzenoid ring. This may explain w1.y products of o-benzidine type are approximately as important as those cf benzidine type in the 72Hammick and Mason J. 1946 638. '*Hughes and Ingold J. 1950 .638. 62 QUARTERLY REVIEWS naphthalene series ; though i t is still not clear why products of diphenyline type have not there been found. Other theories of the benzidine rearrangement have been advanced by Robinson 74 and by D e ~ a r ’ ~ which also accept its intramolecular character. But these theories involve the assumption that the rearranging entity is the univalent cation of hydrazobenzene the dissymmetry of charge in which leads to a differentiation of electronic function between the benzene rings.Hainmond and Shine’s subsequent proof that the rearranging entity in its covalent form is the bivalent cation having similar rings does much to convince us of the essential truth of the “exceptional resonance ” explanation of these intramolecular changes. 74R. Robinson J. 1941 220. This discussion involves the author’s theory of electronic oscillations some dif’ficulties of which have been indicated by the present writers (ibid. p. 608). 75 Dewar J . 1946 406 ; Discuss. Faruday SOC. 1947 2 5.0 “ Elect,ronic Theory of Organic Reactions ” Oxford Univ. Press 1949 p. 235. This is an application of the author’s “ n-bond ” theory the electron-surfeited n-shell of one aromatic ring is assumed to form with the electron-depleted n-shell of the other a r-bond around which after severance of the N-N bond the rings undergo relative rotation until “ anchored ” by the establishment of a C-C bond.