186 BARNETT AND SMILES : THE INTRAMOLECULARXX.-The Intramolecular Rearrangement of Diphcnyl-amine ortho-Sulphoxides. Part 11.By EDWARD DE BARRY BARNETT and SAMUEL SMILES.IN a previous communication to the Society on this subject it wasshown that the ortho-sulphoxides of diphenylamine (type I) maybe transformed by the action of acid reagents into the sulphoniumhydrates (type 11). A t the same time it was observed that if theNH N NbH i c(1.1 (11.) (111).sulphonium grouping be sufficiently basie, the corresponding salt(type 111) may be formed. The tetranitro- and the two isomericdinitro-sulphoxides yielded the sulphonium hydrates, but the parentcompound (I) furnished phenazothionium chloride. In the fourcases examined, it was observed that the sulphonium derivativescontain an additional molecule of water; but usually this may beremoved by crystallisation from high boiling solvents.The investigation of this reaction has been continued with theobject of determining the mechanism by which the change takesplace, and the results are now collected in the following pages.From the mere inspection of the formulae (I and 11) representingthe initial and final products of the reaction, it might be supposedthat the sulphonium hydrate is formed simply by the migrationof hydrogen from the imino-group to the oxygen of the thionylgroup.In order to test whether this is actually the case or not,we have examined the behaviour of sulphoxides in which thisiminic hydrogen is replaced by alkyl. It has been found that thesederivatives undergo the change as readily as the unsubstitutedcompounds. The two chief examples studied were the dinitro-N-methyldiphenylamine ortho-sulphoxide (IV), which has been preREARRANGEMENT OF DIPHENYLAMINE ORTHO-SULPHOXIDES.187viously obtained by Bernthsen from the nitration of N-methylthio-diphenylamine, and the X-methyldiphenylamine sulphoxide (V),which we have prepared by oxidation of the sulphide.\/\AI l l\/A/\/ so(IV. 1 (V* )Assuming that the course of tThe reaction is the same with theseN-alkyl derivatives as with the imino-compounds, it might thenbe concluded that with the latter the reaction does not involve thedirect movement of the imino-hydrogen atom, for on a priorigrounds it is improbable that the alkyl group should be transferredwith such ease under the conditions favourable to the reaction.This conclusion, however, cannot be accepted until some actualproof is obtained of the immobility of the alkyl group, for manyinstances are known where the transposition of alkyl from one atomto another can be detected, although the treatment then requiredis usually more energetic than that necessary to the presentreaction.The proof that alkyl is not removed from nitrogen duringthe reaction has been obtained in the case of N-methyldiphenyl-amine o-sulphoxide. This sulphoxide, when treated with hydrogenchloride, absorbs one molecular proportion of the halogen acid, beingthen converted into the chloride of a base which is deep red incolour and has properties similar to those of the phenazothioniumderivatives described in a previous paper.It may be remarkedthat the quinonoid structure of this chloride is demonstrated bythe further action of hydrochloric acid, which yields a chloro-derivative of N-methylthiodiphenylamine. If the methyl group istransferred from the nitrogen in this reaction, the only alternativesare that it may either enter the aromatic nucleus (as in VI), orbecome attached to the thionyl group (as in VII). In either casethe substance which is formed would be represented as follows(VI and VII) :f /\/\/\“21 IN021 \/A/\/ soHO CH,\/N N 188 BARNETT AND SMILES : THE INTRAMOLECULARand on reduction it should furnish either a methylthiodiphenyl-amine (from VI) or thiodiphenylamine itself (from VII).We find,however, that reduction of the quinonoid salt does not yield eitherof these compounds, but that instead N-methylthiodiphenylamineis formed. It follows that the constitution of this substance mustbe as represented in the third of the foregoing (VIII) formulae;but it is impossible conclusively to show whether salt formationtakes place at the ammonium or sulphonium groups. Moreover, thesalt is not very stable, since, as previously noticed, it readilychanges to chloro-N-methylthiodiphenylamine, thus affordingfurther proof of the immobility of the methyl group. The experi-ments dealing with these derivatives of N-methylthiodiphenylamineare described in the following paragraphs.EXPERI M E NTALConversion of N-Methyldiphenylamine ortho-Sulphoxide into theSulphoniurn Derivative.Attempts to prepare this sulphoxide were made by treatingN-methylthiodiphenylamine with hydrogen dioxide in acetone solu-tion, but only small quantities of the substance could be obtainedby this method, almost the whole of the sulphide being recoveredunchanged after remaining with the reagent for about three weeks.Attention is drawn to this result, since thiodiphenylamine is readilyoxidised to the sulphoxide under these conditions.Bernthsen(Annalen, 1885, 230, 92) has shown that N-methylthiodiphenyl-amine may be converted into the sulphone by hot aqueous per-manganate, but we find that this reagent under similar conditionsfurnishes satisfactory yields of the sulphoxide.A solution of six grams of X-methylthiodiphenylamine in acetonewas acidified with dilute sulphuric acid, more acetone being added,if necessary, to retain all in solution.Two grams of finely powderedpotassium permanganate were then gradually added, the temper*ture being kept at about 1 5 O by immersing the flask in cold water.Care was taken to ensure the presence of a slight excess of sulphuricacid during the reaction. The mixture was shaken for a few hours;then the crystalline precipitate was collected, and well washed withwater. The residue was agitated with a cold aqueous solution ofsulphurous acid, and finally recrystallised from alcohol. The leastsoluble portion contained small quantities of the sulphone, and itwas therefore rejected.Nine grams of the crude sulphoxide wereobtain from 18 of sulphide.N-~ethyldil?henyZamine ortho-sdphoxide forms colourless leaflets,which melt a t 1 9 3 O . It is sparingly soluble in cold acetone oREARRANGEMENT OF DIPHENYLAMINE ORTHO-SULPHOXIDES. 189alcohol, and insoluble in water. Concentrated sulphuric or hydro-chloric acid forms bright red solutions, which contain the phenazo-thionium derivative. A sample which had been dried in a vacuumwas analysed:0.1009 gave 0.2520 CO, and 0.0443 H,O.0-2315CIBH,lONS requires C = 68.1 ; H = 4.8 ; N = 6.3 per cent.A ction. of A cia's.-This sulphoxide is readily soluble in concen-trated aqueous hydrochloric acid, being then converted into thedeep crimson hydrochloride. The attempts made to isolate thelatter in the pure condition were unsuccessful, chiefly .on accountof the hygroscopic nature of the substance, and the ease with whichit is transformed into the chloro-derivative of N-methylthiodi-phenylamine. The platinichloride was therefore prepared byaddition of chloroplatinic acid to the aqueous solution of the salt.The precipitate was collected, and after recrystallisation it wasdried in a vacuum before analysis :C = 67.9 ; H = 4-88.,, 12-6 C.C.N, a t 17O and 763 mm. N=6.4.0-2161 gave 0.0474 Pt. Pt =21*93.0.1184 ,, 0-1568 CO, and 0.0238 H,O. C=36.1; H=2*2.(C,3H,10NS,HC1)2PtC14 requires Pt = 22.4 ; C = 35.9.H = 2.7 per lent.The PlatinichZoride of N-rn~et~ylphenasot~ioniunt (VIII) formsminute, reddish-brown needles, which melt with profound decom-position a t about 255O.Reduction.-The hydrochloride was reduced as rapidly as possiblein methyl-alcoholic solution with tin and hydrochloric acid.Whenreduction was complete, the colourless solution was poured intowater, and the precipita,te was collected. The product contained asmall quantity of the chloro-derivative, and, in order to removethis, it was fractionally crystallised from methyl alcohol. Themelting point and solubility of the chloro-compound lie close tothose of N-methyhhiodiphenylamine, and a complete separation isnot easily made. However, after repeated crystallisation, a sample,which melted a t 9B0, was obtained, and this, when mixed withN-methylthiodiphenylamine (m. p. looo), melted at the same tem-perature.It contained traces of chlorine. I n order to obtainfurther proof that thiodiphenylamine is not formed during thereduction of the chloride, both the crude and purified (m. p. 98O)products of reduction were treated with nitric acid (D 1.4) underthe conditions which yield the dinitro-derivatives of the group.The nitro-derivative was insoluble in hot aqueous alkali hydroxide,whereas the nitro-derivatives of thiodiphenylamine are readilyattacked by this reagent, yielding deep red solutions of the alkal190 BARNETT AND SMILES : THE INTRAMOLECULARsalts. Experiments made with artificial mixtures showed that verysmall quantities of thiodiphenylamine may be detected in thepresence of a large excess of the N-methyl derivative by this method.These experiments show that the methyl group of N-methylthiodi-phenylamine remains undisturbed during the conversion to thequinonoid salt.Con version of N-Me t h ytdinit ro p hen ytamine o-SuZphoxide.-Thedinitro-sulphoxide was prepared from N-methylthiodiphenylamineby the action of nitric acid, and it was purified in the mannerrecommended by Bernthsen (Annulen, 1885, 230, 92).To ensurethe purity of the substance employed, a sample was analysed:C = 48.4 ; H = 3.2. 0.1037 gave 0.1841 CO, and 0.0298 H,O.CI3Hg'OjN3S requires C = 48-9 ; H= 2.8 per cent.The conversion of the sulphoxide into the sulphonium derivativewas effected by means of sulphuric or hydrochloric acid.(a) Concentrated Sulphuric Acid.-A solution of dinitro-N-methyldiphenylamine o-sulphoxide in this acid was set aside forhalf an hour at the atmospheric temperature, and then poured onto powdered ice.The solid precipitate was collected and washedwith water until free from splphuric acid; it was then trituratedwith dilute aqueous sodium hydroxide to remove a small quantityof soluble material. The latter is probably formed by replacementof the nitro- by the hydroxyl group (see Barnett and Smiles, Trans.,1909, 95, 1262), but it was not further investigated. After re-newed filtration and washing, the reddish-brown product wascrystdlised from hot glacial acetic acid. Samples of the once (1)and four times recrystallised (2) product were analysed after beingdried in the steam-oven.(6) Hydrogen C1doride.-Methyl alcohol was saturated withgaseous hydrogen chloride a t Oo, and excess of the solution wasmixed with the dinitro-sulphoxide.The mixture was shaken forsix hours, and then set aside at the atmospheric temperature.After the lapse of forty-eight hours, the dark brown liquid con-tained in suspension an orange-red, crystalline substance, and, sincethis evidently was different from the more soluble portion, it wascollected and separately treated. The solution, which was shown bysubsequent experiment to contain the bulk of the desired sulphoniumcompound, was mixed with a large quantity of water, and the nowinsoluble product was collected and purified by recrystallisationfrom glacial acetic acid. Analysis (3) was then made with asample from the fourth recrystallisation. The crude product con-tained traces of non-ionisable halogen REARRANGEMENT Ol? DIP€IENYLAMINE ORTHO-SULPHOXIDES.'1 91(1) 0*1010 gave 0.1774 CO, and 0,0296 H20. C=4?.9; H=3.26.(2) 0.1005 ,, 0,1752 CO, ,, 0.0260 H20. C=47.5; H=2*87.(3) 0.1006 ,, 0.1752 CO, ,, 0.0287 H20. C=47*5,; H=3*16.CI3H9O5N3S,&H20 requires C = 47.5 6 ; H = 3.0 per cent.The analyses show that, as with the nitro-derivatives of thio-diphenylamine which have been previously studied (Trans., 1909,95, 1257), the conversion of the sulphoxide to the sulphoniumderivative is accompanied by absorption of water; but here themolecular proportion is half that previously observed, and it ismore firmly retained, since it cannot be removed by heating or byrecrystallisation from high boiling solvents.Din~tro-N-~ethylplielzazotitiolziuml hydrate forms reddish-brownleaflets, which do not melt below 250O.It is sparingly soluble incold benzene or glacial acetic acid. By reduction and subsequentoxidation, the substance is converted into the greenish-blue dyewhich is obtained by the same process from the dinitro-sulphoxide;but in contrast with the latter it does not exhibit the reactioncharacteristic of the thionyl group (Trans., 1906, 89, 696), sinceit does not yield the S-phenetyl derivative when treated withphenet.de and sulphuric acid. The only formula which can beassigned to the anhydrous compound is the following:c=3 k/\A\AINOZI 0 ",I I \/\I/\/S(IX. )but from reasons which will presently be stated, it is probable thatthe additional water in these substances is of constitution.The insoluble portion of the product which was obtained by theaction of hydrochloric acid on the dinitro-sulphoxide was separatelycollected and recrystallised from hot glacial acetic acid.It wasthus obtained in bright orange-red leaflets, which remained un-decomposed at 250O. Samples from two different preparations wereanalysed :C=48*3; H=3*51. (1) 0.1004 gave 0.1777 CO, and 0.0303 H,O.(2) 0'2028 ,, 0'3616 CO, ,, 0'0516 H20. C=48*6; H=2*8.CI3H9O,N3S requires C = 48.90 ; H = 2-82 per cent.When crystallised from hot nitrobenzene, the substance is0'2000 gave 0.3718 CO, and 0.0656 H,O.obtained with solvent of crystallisation :C=50-7; H=3.6.2C&W&3S,C&j02N requires C = 50.6 ; H = 3.0 per cent192 BARNETT AND SMILES : THE INTRAMOLECULARThe presence of nitrobenzene in this substance was proved bymixing the sulphuric acid solution with water and passing a currentof sterrm through the liquid. The substance melts and decomposesa t about 280O.Further experiment showed that this compounddoes not contain the sulphonium grouping,- and since it yields the8-phenetylsulphonium salt when treated with phenetole in sulphuricacid solution, it may for the present be regarded as an isomericform of the original sulphoxide. Probably the isomerism involvesthe nitro-grouping.Assuming that the conversion of sulphoxide (I) to sulphoniurnbase (11) takes place in the same manner with the imino-compoundsas with these N-methyl derivatives, it is clear that the hypothesisof migration of hydrogen from the imino- to the thionyl groupcannot be accepted as an adequate explanation of the change.Theortho-sulphoxides of diphenylamine contain the imino- and thionylgroups, and it is well known that each of these in simple derivativespossesses basic properties. So far as the thionyl group is concerned,it has been shown that the aliphatic (Saytzeff, Annulen, 1867, 144,148) and aroma;tic (Smiles and Le Rossignol, Trans., 1906, 89, 697)sulphoxides form nitrates and hydrochlorides. From these con-siderations and from the fact that the reaction can be effectedonly by acid reagents, it seems probable that the salts of thesulphoxides are formed in an initial stage of the reaction, andthat these are subsequently changed into the sulphonium derivatives.That a causal relation exists between the formation of these saltsand the production of the sulphonium compounds may be shownby the following experiments.(1) The Action of the same Acid Reagent o n Sulpkoxides ofVarying Basic Power.-Experiments were made with the tetra-nitro-, di-p-nitro-, isodinitro-, and the unsubstituted diphenylaminesulphoxides, the reagent being alcohol saturated with hydrogenchloride at the atmospheric temperature.In each case the con-ditions adopted were the same: the sulphoxide was shaken with alarge excess of the acid reagent a t about 15--17O, and theapproximate time required for the complete conversion to thesulphoxide was observed.The results are collected in the followingtable :Substance. Approximate time required.Diphenylamine o-sulphoxide.. , . . . . . . , . . . .. .. .N-Methyldiphenylamine o-sulphoxide ... . ..Di-p-nitrodiphenylamine o-sulphoxide.. , . . .isoDinitrodiphenylamine o-sulplioxicle.. . ...Tetranitrocliyhenylamine o-sulphoxida . , . about three weeks.apparently instantaneous.apparently instantaneous.about five hours.about five hours.The data cannot be regarded as anything more than very roughapproximations; but they serve to show that by decreasing thREARRANGEMENT OF DIPHENYLAMINE ORTHO-SULPHOXIDES. 193basic function of the sulphoxides, the tendency to form the sul-phonium base is correspondingly diminished.(2) The Action of Acids of Varying Strength o n t h e sameSuZphoxide.-In these experiments the diphenylamine o-sulphoxidewas employed as the standard.The colourless solutions of thissubstance become deeply coloured on the addition of acids, thesalt of phenazothionium being then formed. Equal volumes(50 c.c.) of 2N-aqueous solutions of acetic, monochloroacetic, tri-chloroacetic, and hydrochloric acids were separately mixed at thesame moment with 50 C.C. of an alcoholic solution (0.1 per cent.) ofthe sulphoxide. After a suitable lapse of time (thirty minutes),the solutions were examined in layers of equal depbh. A differencebetween the intensity of colour in these solutions could be readilydetected, but quantitative measurements of the intensity were notmade, since the solutions containing the stronger acids were ofslightly different colour, probably owing to the formation of di-acidsalts. When arranged in order of increasing intensity of colour,the solutions fall into the following order:F i f t y C.C.of 0.1 per ce,nt. Solution of Diphenylamine Sdphoxide inA Zcohol.Relative coefficientsWith 50 C.C. of of saponification2N-solution of velocity. Remarks.Acetic acid ........................ 0-003 colour appeared after 1 hour.Monochloroacetic acid ......... 0 -043 faint colour immediately produced.Trichloroacetic acid ............ 0.682 immediate coloration.Hydrochloric acid ............... 1 '000 immediate coloration.It may be remarked that after the further lapse of time theintensity of colour in these solutions increases, and the differencebetween each solution becomes even more pronounced.The centrecolumn of the table shows the relative affinities of the acids; thedata were obtained by Ostwald ( J . pr. Chem., 1883, [ii], 28, 449)from measurements of the velocity of hydrolysis of methyl acetate.Assuming that the intensity of colour of these solutions isroughly proportionate to the quantity of azothionium salt present,it follows that the amount of this salt formed at the end of a, givenperiod is controlled by the strength of the acid employed. It isclear that these two sets of experiments justify the hypothesis thatthe conversion of the sulphoxides to the sulphonium compoundsdepends on the preliminary formation of salts of the former sub-stances with the acid reagent.It is, however, difficult conclusively to determine the process bywhich these salts are converted to the quinonoid compounds, sincethey cannot be isolated in the pure condition for separate trea8bVOL.XCVII. 194 BARNETT AND SMILES : THE INTRAMOLECULARment. There are, however, only two alternatives: either (a) thatthey lose the elements of water, or ( b ) that migration of hydroxyltakes place. These will be separately dealt with in the ordergiven.To assume that elimination of water takes place in the mono-acid sulphoxide salt ( X a ) is evidently incorrect, for on this basisthe conversion of the N-methyl derivative would be impossible,and no account would be forthcoming of the additional moleculeof water in the products.In all the examples studied, a largeexcess of acid is necessary completely to effect the reaction, andthis fact lends colour to the assumption that the salts which undergothe change are of the di-acid type (X). It is apparent that theseNH/\/\/\I l l 1\/\/\/ s/\OH C1HCl*k*H/\/\/\I I I I\/\/\/Ssalts would be converted into the azothionium salts by the loss ofthe elements of water, the complete process being represented asfollows :HNH C1.N -H/\OH c1HO H HO H\' \/N N61(XII.)6 H(XIII.)Moreover, it is apparent that whether the resulting quinonoidsalt (XI) can be isolated or not must depend on the basic characterof the sulphonium and ammonium groups, and this is controlledby the nature and number of the groups present in the aromatiREARRANGEMENT OF DIPHENY LAMINE ORTHO-SULPHOXIDES.195nuclei (Kehrmann, Ber., 1906,39, 914; Smiles and Hilditch, Trans.,1908, 93, 1691). For example, the parent phenazothionium com-pound is obtained as the monohalide (XII) (Trans., 1909,95,1259),whilst the di- and tetra-nitro-derivatives, being completely hydro-lysed during the process of isolation, appear as sulphoniumhydrates (XIII).It is necessary to observe that this explanation of the processsatisfactorily accounts for the invariable occurrence in the azo-thionium compounds of an additional molecule of water, which isso difficult to remove as t o give rise t o the suspicion that it is notmerely water of crystallisation.Moreover, a satisfactory account is given of the conversion ofthe N-methyl sulphoxide to the sulphonium compound.The follow-ing formulae represent the process :C1 CH, C1 CH3\/ \/NH NHO CH,\/NIn this case it must remain doubtful whether hydrolysis takesplace at the ammonium or sulphonium group, but the formulagiven for the product is the more probable of the two alternatives,for with the parent compound there can be no doubt that theproduct is the sulphonium halide, since it has been isolated in theanhydrous condition (Kehrmann, Ber., 1901, 34, 4170).The alternative hypothesis of migration of hydroxyl fromquadrivalent sulphur to tervalent nitrogen in the sulphoxide saltaffords an equally satisfactory explanation of the change.Fromthis point of view the reaction must be regarded as a true intra-molecular rearrangement, being represented as follows :HO H (or R)\/ NH (or R) N/\/\/\ /\//\/\I l l 1 -+ t i l l\A/\/ \/A/\/S SBut, as previously remarked, we are a t present unable to offerany decisive evidence for the preference of either alternativehypothesis. On general grounds, the former of the alternativesgiven seems the more probable.0 196 BEKRY: THE ADSORPTIOX OFIn conclusion, it must be stated that the hypothesis described(1) The immobility of the alkyl group during the conversion of(2) The more basic the character of the sulphoxide, the more(3) The conversion of a given sulphoxide proceeds more easily(4) The absorption of the elements of water during the reaction.The results of the study of this reaction may be summed up asfollows :(1) The conversion of diphenylamine sulphoxides t o phenazo-thionium derivatives does not take place by direct migration ofhydrogen from the imino-group t o the oxygen of the thionyl group.(2) The salts of the imino-sulphoxides are first formed with theacid reagent, and these are then converted into the phenazothioniumsalt either (u) by loss of the elements of water, or ( b ) by migrationof hydroxyl from the quadrivalent sulphur to tervalent nitrogen.Preliminary experiments have shown that the conversion of thethionyl group to the quinonoid thionium arrangement may beeffected not only in other cyclic systems, but also in hydroxy- andamino-sulphoxides.is advanced to account for the following facts:N-methyldiphenylamine o-sulphoxide.readily does the change take place with a given acid reagent.the stronger the acid employed.These substances are now being investigated.In conclusion, we desire to thank the Research Fund Committeeof the Chemical Society for a grant which has defrayed the expenseof this research.THE O~~GANIC CHEMISTRY LABORATORY,UNIVERSITY COLLEGE, LONDON