212 DEWAR AND JONES: THE CHEMICAL REACTIONS OF XXVI1.- The Chemical Reactions of A%cEeZ Caybonyl. Reaction with Aromatic Hydrocarbons Syathesis of Part II. i a prese.iLcc of Alumi.iliicm Chloride. Aldehydes arm! Anthmcene Dwivati/ues. By JAMES DEWAH. and HUMPHREY OWEN JONES. NICKEL carbonyl does not react with aluminium chloride or most of the other metallic chlorides, nor has it by itself any action on aromatic hydrocarbons. If, however, a mixture of nickel csrbonyl and a hydro- carbon, such as benzene or toluene, is treated with aluminium chlorideNICKEL CARBONYL. PAliT 11. 213 in the cold, a rapid evolution of hydrogen chloride a t once occurs, while the mixture darkens in colour and becomes viscid. It is remarkable that during this apparently violent reaction there is no perceptible rise of temperature.The reaction was allowed to proceed in the cold and also a t looo, and the products investigated. In the former case, an aldehyde was produced, the reaction being very similar to that effected by a mixture of carbon monoxide and hydrogen chloride in the presence of aluminium chloride (Gattermann and Koch, Ber., 1897,30, 1622 ; 1898, 31, 1149), and little or no nickel chloride could be detected. I n the latter case, the aldehyde had almost entirely disappeared ; an anthracene derivative was found to have been produced whenever possible, and a consider- able quantity of nickel chloride appeared among the reaction products. The mechanism of this reaction is obscure, and although it is probable that the anthracene is formed by the condensation of two molecules cf the aldehyde with the elimination of the elements of hydrogen per- oxide, a change which may be effected by the metallic nickel present, yet numerous experiments made with the view of throwing further light on the process have given negative results.The aldehyde produced from toluene is p-tolualdehyde, and the dimethylanthracene formed must then be 2 : 6-dirnethylanthracene. m-Xylene gives 2 : 4-dimethylbenzaldehyde, and the tetramethyl- anthracene is therefore the 2 : 4 : 6 : %derivative, since the aldehyde can only condense in one way so as to give an anthracene compound. This tetramethylanthracene is identical with that produced by Anschutz from m-xylene. Consequently the other produced from m-xylem by Friedel and Crafts must be the 1 : 3 : 6 : 8-isomeride, since there are only two such compounds which can be derived from m-xylene.Mesitylene gives the aldehyde and no condensation product, the formation of an anthracene derivative being impossible in this case. Naphthalene behaves in an entirely different manner from the above benzene derivatives; no aldehyde could be detected, and the product contains a hydrocarbon, Cl6HI2, apparently identical with that obtained from ruficoccin by distillatior, with zinc dust and also by the action of methyl chloride and aluminium chloride on naphthalene. I n this case, it would appear that carbon rings have been formed directly from carbon monoxide. Benseihe. A mixture of benzene (4 mols.), aluminium chloride (4 mols.), and nickel carbonyl (1 mol.) was allowed to react in the cold either in a flask provided with a calcium chloride tube with a very small orifice as outlet for the gas, or in a sealed tube, which was opened from time to time to relieve the pressure.After a few days, the dark mass was214 DEWAR AND JONES : THE CEIEMrCAL REACTIONS OF decomposed by means of ice-water (the aqueous layer contained only mere traoes of nickel in solution), and after the addition of hydrochloric acid was submitted to distillation in a current of steam. The distil- late and residue were extracted with benzene, dried over calcium chloride, and distilled. The distillate in steam was found to consist almost entirely of benzaldehyde, which was further identified by oxidation to benzoic acid and by the formation of its phenylhydrazone. The yield obtainable is not large, not exceeding 25 per cent.of the weight of the benzene used. The residue not volatile in steam was found to contain only a small quantity of viscid oil, which was not further examined. On heating a mixture of the reagents, in the same proportions as above, at looo in a sealed tube for periods varying from half an hour t o several hours, the result was invariably the same, a dark mass was formed, considerable pressure developed, and a certain amount of the three reagents was found to be &till present. The product of the reaction was mixed with ice-water and treated as already described ; the aqueous layer now contained considerable quantities of nickel in solution. The distillate in steam yielded a small quantity of benzaldehyde and a trace of a crystalline solid, which was afterwards found to be identical with the substance isolated from the part not volatile in steam.The dark, tarry solid left in the flask after the distillation in steam was taken up in hot benzene and the solution dried over calcium chloride and distilled under reduced pres- sure. A crystalline solid having a pale yellow colour distilled over between 250° and 260° under 10 mm. pressure, leaving behind a small amount of tarry residue, which was not further investigated. In one experiment, 12 grams of the solid were obtained, together with a small amount of benzaldehyde, from 17 grams of nickel carbonyl and 31.2 grams of benzene. The solid, which was repeatedly crystallised from hot alcohol or benzene, was finally obtained in colourless, lustrous plates melting at 211' (uncorr.), and showing a beautiful violet fluorescence; it was identified as anthracene by ultimate analysis and vapour density deter- minations, and was further characterised by the formation of its picrate (m.p. 136-137') and anthraquinone. Xoreover, it was found that both this product and pure anthracene dissolved in pure sulphuric acid to a pale yellow solution, but if a trace of nitric acid were present-as is always the case with the ordinary ' pure ' acid-a brilliant, dark green coloration was produced, and on adding potassium dichromate or a larger amount of nitric acid, the colour changed to a brilliant, reddish-purple, and finally t o brown. This reaction, which we have never seen described, constitutes an ex-NICKEL CARBONYL.PART 11. 215 ceedingIy delicate test for anthracene and its homologuea, and serves to distinguish this series from allied substances. Anthracene is therefore the principal product of the action of nickel carbonyl on benzene in the presence of aluminium chloride a t loo", and it seems that benzaldehyde is certainly an intermediate product. The mechanism of this reaction, if benzaldehyde is assumed to be the intermediate product, consists in the elimination of the elements of hydrogen peroxide. H H This change is of considerable interest, and several experiments have been made with the object of elucidating it ; but, unfortunately, little information has been gained. The metallic nickel, produced by the decomposition of its carbony1 derivative, is very probably the reducing agent which effeots the cbange, since it is only dissolved in any quantity when anthracene is produced.(1) Benzaldehyde, when heated with aluminium chloride alone or with benzene and aluminium chloride, produced no anthracene, but only benzoic acid and a solid melting at 90'. (2) On heating with aluminium chloride and zinc dust, benzaldehyde yielded a solid not volatile in steam; this product dissolved in hot alcohol and crystallised in needles melting at 1343. 0.1223 gave 0.366 CO, and 0.0625 H,O. C = 78.8 ; H = 5.6. (C,H,*CHO), requires C = 79.2 ; H = 5-6 per cent. The substance is identified as being benzoin by the foregoing analysis and by a comparison of its melting point with that of the pure substance.(3) A mixture of benzaldehyde, aluminium chloride, and finely- divided metallic nickel (reduced a t a low temperature), when left in the cold became hot, and R small quantity of gas was produced. The product was heated at 100° for 5 hours and then examined in the manner already described. There was a slight pressure in the tube, and the aqueous solution contained nickel chloride. The steam distil- late contained benzaldehyde and benzoic acid, but no anthracene could be obtained. Hydrogen chloride, in benzene solution, mas found to have only a very slight action on nickel carbonyl. Zinc chloride and anhydrous ferric chloride cause the production of a small amount of hydrogen218 DEWAR AND JONES: THE CHEMICAL REACTIONS OF chloride, but the action is very slight and the liquid remains clear and colourless even after prolonged heating.Toluene. On mixing nickel carbonyl (1 mol.) with toluene (4 mols.) and aluminium chloride (4 mols.), torrents of hydrochloric acid were im- mediately evolved, and the evolution of gas from the mixture continued slowly for several days. The reaction product was treated successively with &-water and hydrochloric acid, being then submitted to steam distillation and worked up as already described. The aqueous solution contained only a trace of nickel chloride. Toluene distilled over first, and then a small quantity of oil boiling a t 204' was obtained, the yield being about 16 per cent. of the toluene used. This product was identified as p-tolualdehyde by the preparation of methyl tere- phthalate from the product of i t s oxidation.The phenylhydrazone of this aldehyde was prepared; it separated from alcoholic solution in almost colourless plates which melt at 114-115' and readily turn pink on exposure to sunlight. There was a small amount of non-crys- tallisable oil with a high boiling point, which was not further examined. The product of the reaction at loo*, worked up in a similar way, gave about 9 per cent. of the aldehyde, and then the residue gave a solid, which, on crystallising from hot alcohol, separated in plates melting a t 215-216'. 0.1566 gave 0.5330 CO, and 0.099 H,O. C = 92.9 ; H= 7-02. C,,H,, requires C = 93.2 ; H = 6.8 per cent. The substance is therefore a di~net?iylanthl.ucene. This hydrocarbon gives the same colour reactions as anthracene with sulphuric and nitric acids, the colour being a little more intense, On oxidation with an acetic acid solution of chromic acid, it gave a quinone which melted at 159-160'.Elbs and Wittich (Beg.., 1885, 18, 348), by the action of chloroform and aluminium chloride on toluene in carbon disulphide solution, ob- tained a dimethylanthracene melting at 2 15-2 16O, and producing a quinone melting at 16 1-1 62O. This hydrocarbon is probably iden- tical with that described above. A similar hydrocarbon melting at 331--232O, usnally assumed to be identical with t h a t of Elbs and Wittich, has been prepared by Friedel and Crafts (Arzm. Chim. Phys., 1884, [vi], 1, 482) by tho action of benzyl chloride and alumin- ium chloride on toluene, and also by the action of methylene chloride and aluminium chloride on toluene (Zoc.cit., 11, 266). The constitution of the dimethylanthracene prepared by these methods has not been determined.NICKEL CARRONYL. PART 11. 217 If, as is probable, p-tolualdehyde is an intermediate product in the formation of the anthracene derivative, then the constitution of the dimethylanthracene thus produced, and consequently that of Elbs and Wit tich, is at once fixed, since y-tolualdehyde can only condense in one way giving 2 : 6-dimethylanthracene, K H H H H &\,,/ \/yq \A/\/H I 1 : H;//\CH 0 H /\C1T3 -+ 1 I c H,' H H H CH3i)JS H U H C \P H This dimethylanthracene is therefore probably not the same as that obtained by Louise ( A m , . Chim. Phys,, 1885, [vi], 6, IS) ; this sub- stance, which melted at 218-219O and yielded a quinone melting at 170°, was prepared by passing benzplmesitylene through a red hot tube.These two hydrocarbons could not, however, be identical unless some profouud molecular rearrangement had taken place. rn-Xylene. A mixture of m-xylene, nickel carbonyl, and aluminium chloride in the proportions employed in the preceding examples at once darkened and a rapid evolution of hydrogen chloride occurred. The reaction product was allowed to remain for several days and then worked up in the manner already described. The distillate in steam gave an approxi- mately 20 per cent. yield of an aldehyde boiling at 215--220", and a very small quantity of a n oil with a high boiling point; there was a small amount of tarry residue left in toe distillation flask.The aldehyde, on oxidation with chromic or nitric acids or when left exposed to the air, gave an acid which, after recrystallisation from hot water, was obtained as long, lustrous prisms melting a t 1269 0.1355 gave 0.3560 CO, and 0.0838 H,O. C6H,(CH,),*C0,H requires C = 72.0 ; H = 6.7 per cent. The acid is therefore 2 : 4.dimethylbenzoic acid (xylic acid), identical with that produced by the action of carbonyl chloride and aluminium chloride on m-xylene (Ador and Meier, Ber., 1879, 12, 1968). Accordingly the aldehyde is 2 : 4-dimethylbenzaldehyde, identical with that produced by the action of carbon monoxide, hydro- gen chloride, and aluminium chloride on m-xylene by Gitttermann and Koch (ZOO. cit.). The phenylhydrazone of this aldehyde crystallises from alcohol in pale yellow, rhombic plates melting at 82-84O.It has a great tendency t o separato as an oil, and decomposes rapidly in sunlight, becoming coloured and gummy. C = 71.66 ; H = 6.87.218 DEWAR AND JONES: THE CHEMICAL REACTIONS OF The mixture was also heated at 100' for several hours in a sealed tube and the products worked up as beFore. The part which volatilised in steam gave an approximately 20 per cent. yield of the aldehyde, just a8 if the reaction had proceeded in the cold. The residue in the distilling flak yielded a quantity of a solid boiling above 280' under 20 mm. pressure and a non-crystallisable oil with a higher boiling point. The solid mas crystallised successively from hot benzene and from hot acetic acid, and was obtained in beautiful, lustrous plates melting at 280' and having a slight yellow tinge with a brilliant green fluorescence.This hydrocarbon is sparingly soluble in cold ether, alcohol, benzene and acetic acid, but more soluble in the hot solvents; it is fairly soluble in cold chloroform. 0,2698 gave 0.9097 CO, and 0.1875 H,O. C,,H,, requires C = 92.3 ; H = 7.7 per cent. The compound gives a yellowish-brown coloration when dissolved in sulphuric acid with a trace of nitric acid; the solution, on further addition of the latter acid, assumes a claret, or an intense reddish- purple colour. On oxidation with chromic acid dissolved in acetic acid, a quinone was obtained, whicb, on recrystallisation from hot alcohol or acetic acid, formed pale yellow prisms melting a t 228-230'.The hydrocarbon is therefore a tetmmethylanthracene, and appears to be identical with that prepared by Anschiitz (Annulen, 1886, 235, 174) by the action of acetylene tetrabromide and aluminium chloride on m-xylene. Since the aldehyde C = 91.9 ; H = 7.72. can only condense in one way, the compound produced must be 2 : 4 : 6 : 8-tetramethylanthracene having this constitution : CH,H H unless, during the reaction, a considerable transformation has taken place resulting in the change of position of the methyl groups. The quinone described above is therefore 2 : 4 : 6 : S-tetramethyl- ant hraquinone. Friedel and Crafts (Ann. Chim. Phys., 1887, [vi], 11, 268) obtained a tetramethylanthracene by the action of methylene chloride andNICKEL CARBONYL.PART 11. 219 aluminium chloride on nz-xylene, which is certainly diiferent from that obtained by the authors and by Anschutz, since i t melts at 162-163' and gives a quinone melting at 235". This hydrocarbon must be 1 : 3 : 6 : €l-tetrarnethylanthracene, since this is the only other possible derivative obtainable from nz-xylene. MesityZene. Mixtures of mesitylene (4 mols.), aluminium chloride (4 mols.), and nickel carbonyl (1 mol.) were either allowed to remain in the cold or heated in sealed tubes for several hours at 100'. The result in both cases was the same, On opening the tubes, there was but little pres- sure due t o hydrogen chloride as compared t o that observed with benzene and toluene, and on treating wit8h water only a comparatively small quantity of nickel went into solution.After acidification, the liquid was submitted to distillation in steam, when the whole of the oil passed over leaving only a very small quantity of tarry matter in the distilling flask. The distillate was extrwted with ether, dried over calcium chloride, and submitted to fractional distillation. The main portion OF the liquid was found to be unchanged mesitglene, the remainder was an oil boiling at 234-240' and having the character- istics of an aldehyde. This oil, when oxidised with the calculated quantity of potassium permanganate in an a1 kaline solution, yielded an acid which was sparingly soluble in cold water and, on crystal- lisation from hot water, melted at 152'. 0.1072 gave 0.2865 CO, and 0.0725 H,O. C,H2(CH,),*C02H requires C = 73.17 ; H = 7.32 per cent.The substance is therefore mesitylenecarboxylic acid (m. p. 152O), the aldehyde being 2 : 4 : 6-trimethylbenzaldehyde, the boiling point of which is given as 235-240' by Feith (Ber., 1891, 24, 3544) ; this compound cannot condense to form an anthracene derivative. I n the present instance, the reaction is very incomplete, thus, after heating at 100' for 6 hours, only 2 grams of the aldehyde were ob- tained from 24 grams of mesitglene. I n a similar experiment, in which the heating lasted 16 hours, only a small quantity of aldehyde was obtained, but a little mesityleaecarboxylic acid was also produced (compare the formation of benzoic acid, page 215). C=72*89 ; Hx7.51. iVGp?&cclene, Mixtures of naphthalene, nickel carbouyl, and aluminium chloride in the proportions indicated in the preceding experiments, when allowed t o remain in the cold, gradually became very dark and slowly220 DEWAR AND JONES: THE CHEMICAL REACTIONS OF evolved hydrogen chloride.After about:a week, the mass was worked up in the manner already described. The distillate in steam which solidified was found to be practically pure naphthalene, and no product of an aldehydic nature was found. The black solid left in the distill- ing flask was dissolved in benzene, dried over calcium chloride, and, after evaporating off the benzene, was distilled under diminished pressure. At first a smdl amount of naphthalene distilled over, at about 280' under 17 mm. pressure an oil distilled which afterwards solidified, then above 300° the oily fraction, which had an orange colour, only partially solidified, and a small amount of black residue was left in the distilling flask. When the same mixture had been heated at looo f o r n few hours in a sealed tube and worked up as before, the same products were ob- tained, but in different proportions.There was now only a small quantity of naphthalene, together with a certain amount of the solid distillate and more of the oily fraction, but the main product remained in the distilling flask as a hard, black, resinous residue decomposing at the temperature at which the glass began to soften. The two last- mentioned substances still await further investigation. The principal solid product obtained when the reaction was carried out in the cold or when the heating only continued for a very short time, was purified by repeated crystallisation from hot benzene, and was then isolated in lustrous plates having a yellow tinge and melting at 180-1 81'.This substance is very sparingly soluble in cold alcohol, benzene and acetic acid, much more so in the hot solvents, and especi- ally in benzene or ethylene dibromide ; its solutions are fluorescent. The colour of the hydrocarbon was diminished by crystallisation from hot alcohol with the addition of animal charcoal, but the melting point remained unchanged, From its solution in ethylene dibromide, the compound was obtained in colourless plates melting a t 180-181', the fluorescence of which still persists. 0.1203 gave 0.4100 CO, and 0*0610 H,O. C = 93.7 ; H = 5.63.0.1371 ,, 0.4722 CO, ,? 0.0711 H,O. C= 93.9 ; H 4 - 7 6 . (C,H,), requires C = 93.75 ; H = 6.25. (C,H,), ,, C=94*11 ; Hn5.89. (C,H,), ,, C-92.7 j H=7*3 per cent. A determination of the vapour density was made by V. Meyer'g 0.0712 gave 8.4 C.C. at 12' and '774 mm. After cooling the apparatus, it was found that the hydrocarbon bad Several determinations of the molecular weight were made by the method, using a lead bath heated to a high temperature. M. W. = 196.4. aublimed in lustrous plates without charring.NICKEL CARBONYL. PART 11. 221 cryoscopic method with carefully purified ethylene dibromide as solvent. Test experiments with naphthalene and ant hracene served to show that the solvent was pure and gave very Satisfactory numbers. The experimental errors in the determination of the depression are necessarily large, since with solutions nearly saturated, as, for example, with 0.11 gram of the substance in 41 grams of solvent, the depression wa8 only 0*16', so that accurate results could not be expected. The following values mere obtained : M.W. = 193. 198. 214. 227. Attempts were made to use acetic acid or benzene for the cryoscopic method, but the hydrocarbon was too sparingly soluble in these solvents. With phenol, the valw 215 was obtained, but the depres- sion observed with a nearly saturated solution was only O.OSo. By the ebullioscopic method with alcohol as solvent, the elevation was too small to give reliable results, but in benzene the values 263 and 334 were obtained. Mr. G. Barger, B.A., of King's College, kindly determined the molecular weight; of the hydrocarbon by means of his microscopic method (this vol, p.286). The sparing solubility again makes the determinations difficult, and largely increases the experimental error. A solution in benzene containing 9.07 grams to the litre mas found t o be isotonic with a benzil solution containing 0.0345 gram-molecule per litre, hence the molecular weight is 263. A solution in ethylene dibromide containing 11 -5 grams per litre mas isotonic with a solution containing between 0.057 and 0,064 gram- molecule of triphenylmethane per litre. Hence molecular weight is 180-202, giviog 191 as a mean value. A solution containing 12.7 grams per litre was isotonic with a scjlution containing between 0.055 and 0.066 gram-molecule of benzil per litre, hence the molecular weight is 192-231, the mean being 211.The analyses and molecular weight determinations correspond best with the formula C16H,2, and the properties of the hydrocarbon corre- spond fairly well with those of a hydrocarbon having this formula, which had already been prepared by several chemists. Liebermann and Dorp (AnnaZsn, 1873, 163, 112), by the distillation of ruficoccin with zinc dust, obtained a hydrocarbon, C16H12, melting at 183-18S0, and giving a quinone melting at 250". The constitution Cl,H,<8z is suggested by these authors for this hydrocarbon. By distilling either coccinin or carmine with zinc dust, Fiirth (Bey., 1883, 16, 2169) obtained a hydrocarban having this formula and melting at Bischoff (Ber., 1890, 23, 1905 and 3200), by the action of methyl chloride on naphthalene in the presence of aluminium chloride, obtained, VOL LXXXV, Q 186-1 87'.228 THE CHEMICAL REACTIONS OF NICKEL CARBONYL. among other products, a hydrocarbon, boiling at about 36O0, from which a crystalline substance was isolated having the empirical formula C,H, and melting at 179-181O.The melting point and solubility of Bischoff’s compound correspond exactly with those of the compound prepared by the authors. It may or may not be identical with the compound from ruficoccin. BischoE put forward the following formula : for his hydrocarbon, suggesting that it might be derived from 1 : 4 : 5 : 8-tetramethylnaphthalene. Whatever the formula of the hjdrocarbon may be, it is clear t h a t unless the highly improbable assumption is made that some of the naphthalene molecules are partially broken down, and the hydrocarbon then built up from intact naphthalene molecules and the degradation products, it must be coicluded that a carbon ring bas been synthekised from carbon monoxide. The only other case in which this has been shown to occur is in tho production of potassium hexaoxybenzene by the action of carbon monoxide on potassium. Reaetion of Nickel CarboPqZ with BeTazlene in the Pressnee of Aluminium Bromide. In this case, a reaction readily takes place, and the product was treated in the manner already described for aluminium chloride. No benzaldehyde and no anthracene could be detected, but a crystalline product distilling at 300’ under 30 mm. pressure, and an uncrystaliis- able oil with a higher boiling point were obtained. The crystalline product, on repeated crystallisation from hot alcohol, was obtained in lustrous plates melting a t 181-181.5O, and appeared to be identical in every respect with that obtained from naphthalene, a mixture of the two substances melting a t 181-181.5°. The authors desire to express their thanks t o Dr. Mond for his aid in supplying the means for the conduct of these investigations. U N I v E R s I T Y C H E MI c A I, LA BO I: AT o RY , CAMBRIDGE.