THE ACTION OF ALDEHYDES ON THE GRTGNARD REAGENT. 527L1V.- The Action oj' Aldehydes on the G r i p ardReagent.By JOSEPH MARSHALL.ALTHOUGH many observers have investigated the reaction occurringbetween magnesium alkyl or aryl haloids and substances containingthe carbonyl group when the magnesium compound was in excess,i t appears that none has hitherto observed that if an excess of analdehyde is added t o an ethereal solution of a magnesium alkylhaloid, on decomposing the product with water there is obtained,along with the secondary alcohol formed in the usual manner, aconsiderable quantity of a substance of higher boiling point. Thepresent paper gives an account of an investigation which owes itsorigin to the fact that, in the residue obtained in the preparationof a large quantity of phenylmethylcarbinol from magnesiummethyl iodide and benzaldehyde, there was present a substancewhich could be distilled in a vacuum a t about 200°, and crystallisedwhen it was allowed to remain a short time.Analyses of this sub-stance gave results from which the empirical formula C,,HI4O wascalculated, and of the many compounds with this empirical formula,the only one which could possibly be formed in this reaction ismethyldeoxybenzoin, C,H,*@H(CH,)*CO*C,H,. This was obtainedby V. Meyer and Oelkers (Ber., 1888, 21, 1297) by the action ofmethyl iodide and sodium on deoxybenzoin, and a comparison ofthe properties of the product of the Grignard reaction with thoseof methyldeoxybenzoin proved that the two substances wereidentical.It was thought that the production of this compound might bedue to the action of a slight excess of benzaldehyde on the first pro.duct of the reaction in a manner indicated by the followingequation :C,H,*CH(CH,)*OMgI + C6H,*CH0 =C6H,*CH(CH3)*CO*C6H, + MgIoOH .. (1)and attempts were made t o increase the yield of this substance bythe addition of a second molecular proportion of benzaldehyde tothe magnesium methyl iodide. It was noticed that the grey,crystalline mass which was produced when one molecular propor-tion of benzaldehyde was allowed to react with one molecular pro-portion of magnesium methyl iodide was converted into a darkbrown, viscous mass when the second molecular proportion ofaldehyde was added, and a t the same time a slight rise in tempera-ture was observed.The mass was not altered in appearance b528 MbRSHALL: THE ACTION OFwarming during several hours on a water-bath, although experi-ments showed that the yield of methyldeoxybenzoin was consider-ably increased by proloriged heating. I f the equation (l), repre-senting the reaction as a double decomposition similar to thattaking place when water is added to the magnesium compound,correctly indicated the manner in which methyldeoxybenzoin isformed, since this compound is readily soluble in ether, it should befound in the supernatant ethereal layer before water is added tothe mixture, whilst the viscous mass referred to above should bechiefly magnesium oxyiodide, mixed with the unchanged Grignardproduct. By treating the two layers separately with dilute acid, itwas found that the ethereal layer contained a small quantity ofbenzaldehyde, but no methyldeoxybenzoin, whilst the viscous massyielded phenylmethylcarbinol and a considerable quantity of methyl-deoxybenzoin.It is suggested that the second molecule of benz-aldehyde combines with the first product of the reaction in thefollowing manner :C6H5*CH(CH3)*OMgI + c6H,*CH0 =C,H,*CH(CH,)*C(C,H,)(OMgI)*OH . . (2)This more complex magnesium compound is then decomposed bythe addition of water, with the formation of methyldeoxybenzoin :C6H5*C'H( CH,)=C( C",H5) (OMgI) *OH + H20 =C6H,*CH((CH3)*CO*C6H, + H,O + MgI-OH . . (3)I n order to determine if the reaction is capable of general appli-cation, experiments were made with other magnesium compounds.By the action of benzaldehyde on magnesium dimethylcarbinyliodide, a quantity of phenyl isopropyl ketone was produced.Asomewhat different result was obtained by heating magnesiumphenyl bromide with two molecular proportions of benzaldehyde, ofwhich reaction benzophenone and benzyl alcohol were the productsinstead of phenyldeoxybenzoin. This result, however, may easilybe explained by assuming the reaction to proceed in a manneranalogous to that expressed by equations (2) and (3), and that thenthe phenyldeoxybenzoin is hydrolysed to benzophenone and benzylalcohol. It is well known that phenyldeoxybenzoin does not existin the ketonic form, but only as the unsaturated triphenylvinylalcohol, C(C~H,)Z:C(OH)*C~H, (H.Biltz, Ber., 1899, 32, SSO), andthe addition of a molecule of water to this substance in the mannerindicated by equation (4) would decompose it, with the formationof the above-mentioned substances :C(C6H5),:C(OH)eC6H5 + H20 =(C,H,),CO + C,H,*CH,*OH . (4)A similar theory would account for the production of a certaiALDEHYDES ON THE GRIGNARD REAGENT. 529amount of benzyl alcohol in the above-mentioned reaction betweenbenzaldehyde and magnesium dimethylcarbinyl iodide. In thiscase, phenyl isopropyl ketone is apparently partly hydrolysed intoacetone and benzyl alcohol:cH( cH,),*C"o* C,H,+ c (CH,), : c( OH) C,H,-+(c'H,),co + C6H,*CH2*OH.An analogous result was obtained when acetaldehyde was allowedto react with magnesium phenylmethylcarbinyl bromide, in whichreaction, however, no trace of a-phenylethyl methyl ketone couldbe found.Instead of this, the products of the reaction were aceto-phenone and phenylethyl bromide, C6H,*CHBr*CH,, the formationof the former being easily explained by the above theory. No ex-planation is offered a t present for the production of the phenyl-ethyl bromide, but it is in agreement with results obtained inexperiments made on the reaction between Grignard reagents andformaldehyde. It was thought that the reagent might react withformaldehyde to produce a substituted acetaldehyde in a mannerexpressed by the equation :R*CH(OMgX)*R' + CH20 =R*CH[CH(OH)*OMgX].R' + R*CH(CHO)*R' . , (5)and that formic ester might react in a similar manner to form sub-stituted acetic esters.The products of these reactions were, how-ever, substituted methyl haloids, along with a certain quantity ofdecomposition products. These reactions appear to consist in theremoval of magnesium oxide from the Grignard product, thus :C6H5*~H(CH,).0MgI = c6H5*CHI*CH3 + MgO,but the mechanism of the reaction is not yet clear. The trioxy-methylene or formic ester disappears during the reaction, with theproduction of a viscous mass similar in appearance to that observedin the earlier experiments, so that it would appear that some com-bination takes place between the Grignard product and the trioxy-methylene or formic ester.Further experiments are in progress, but it was thought that theresults obtained u p to the present were of sufficient interest towarrant their publication.EXPERIMENTAL.Action of Benzaldehyde on Magnesium Methyl Iodide.*Twelve grams of magnesium turnings were allowed to react with20 C.C.of a mixture of 72 grams of methyl iodide and 100 C.C. ofanhydrous ether. When the first violent reaction was over, 100 C.C.* The author is indebted to Mr. J. A. Hartley for assistance in the execution ofthis experiment530 MARSHALL: THE ACTION OFof ether were added, and the remainder of the methyl iodide-ethermixture was allowed to run slowly on to the magnesium, the wholebeing then gently warmed on the water-bath t o effect completesolution of the magnesium. This solution was cooled in ice, and amixture of 52 grams of benzaldehyde and 50 C.C.of ether wasslowly added to it, care being taken that no rise in temperaturetook place. (In an experiment in which phenylmethylcarbinol wasrequired, the usual precaution was not observed, and it was foundthat, instead of almost a theoretical yield of the carbinol, thequantity obtained was only about 20 per cent. of that theoreticallypossible. The principal product of the reaction in this case was aliquid, boiling a t 148-150°/15 mm., which did not react withsodium. On warming with phosphorus pentachloride, this liquidyielded phenylethyl chloride, and it was probably diphenylmethyl-carbinyl ether.) The flask containing the mixture was allowed toremain overnight, and a further quantity of 52 grams of benz-aldehyde mixed with 50 C.C.of ether was slowly added. A veryslight rise in temperature was observed, while the supernatant etherbecame yellow, and the grey, crystalline mass, which constitutedthe first product of the reaction, became viscous. After all thebenzaldehyde had been added, no trace of crystalline structure couldbe observed. The flask was now heated gently on the water-bathduring twelve hours, the only noticeable changes being that theether became less yellow in colour, whilst the viscous mass graduallybecame dark brown. The mixture, after cooling, was poured intodilute sulphuric acid, ice being added to prevent any rise intemperature. The ethereal layer was now separated, washedsuccessively with dilute acid, dilute sodium hydrogen sulphite solu-tion, dilute sodium carbonate solution, and finally with water, andafter dehydration it was submitted to fractional distillation in avacuum.The product was thus divided into two fractions, thcfirst (about 40 grams) boiling below 150°/15 mm., whilst the secondfraction distilled between ZOOo and 220° under the same pressure,and it solidified on cooling. The first fraction contained about20 grams of benzaldehyde, which was separated as the sodiumhydrogen sulphite compound, and the remainder of this fractionconsisted of phenylmethylcarbinol. The second fraction wasdrained, and after recrystallisation several times from alcohol itmelted sharply a t 53O. (Meyer and Oelkers give the melting pointof methyldeoxybenzoin as 53O, whilst Beilstein's " Handbuch "13rd ed., Vol.II., p. 2301 gives 58O.) From 20 t o 30 grams ofpure product were thus obtained in fine, faintly yellow needles.(Found, C = 85-56 ; H = 6.42. C',,H,,O requires C = 85-71 ; H = 6-66per cent.ALDEHYDES ON THE GRIGNARD REAGENT. 531The oxime and semicarbazone of this ketone are difficult to prepare, but a simple means of identifying the substance consists inthe formation of a bromine derivative. I f a solution of brominein carbon disulphide is added to a solution of the ketone in thesame solvent, the colour of the bromine disappears, and heat isdeveloped. No evolution of hydrogen bromide is observed, but oncooling the solution to the ordinary temperature a solid separates,which may be recrystallised from alcohol, in which solvent the sub-stance is not very soluble.The pure product melts a t 159-160°:0.2260 gave 0.2320 AgBr. Br =43*55 per cent.(r,,H,,0Br2 requires Br = 43.25 per cent.The substance is therefore the dibromide of the ketone.An attempt was made to determine whether derivatives of thetautomeric form of the ketone could be obtained. A quantity ofthe ketone was boiled for some time with an excess of acetyl chloride,and the solution was then allowed to cool. Crystals separated,which were collected and recrystallised from alcohol. The sub-stance was not readily soluble in this solvent, and separated fromi t in colourless leaflets, which melted a t 111-112O. It containedchlorine, and an estimation of this element gave a result (C1= 15-52)which agreed with the formula C6H,*C(CH3):CC1.C6H,.(C,,Hl3CIrequires C1= 15.54 per cent.)Attempts were made to hydrolyse methyldeoxybenzoin intoacetophenone and benzyl alcohol, but they were not successful, thesubstance being almost unchanged by boiling with sulphuric acid(lH,SO, : lH20) or by heating with concentrated hydrochloric acidin a sealed tube a t 150O. The only decomposition product thatcould be isolated was a very small quantity of benzaldehyde.Action of Benzaldehyde on Magnesium Dimethylcarbinyl Iodide.This experiment was carried out as in the preceding case, theformation of a brown, viscous mass being again observed. Afterdecomposition of the product with dilute acid, the ethereal extractwas dehydrated and distilled with a fractionating column to pre-vent dimethylcarbinol passing over with the ether. When thethermometer had reached 50° the receiver was changed, and a frac-tion (A) was collected up to 130O.The residue was distilled underdiminished pressure, when two fractions, B (100-150°/ 15 mm.)and C (150-210°/15 mm.), were obtained. The distillate (A) hada strong odour of acetone, and it gave iodoform when treated withiodine and sodium carbonate solution. It was probably a mixtureof acetone and isopropyl alcohol, but the quantity of substance wastoo small to admit of the separation of the two. Fraction (B) wasVOL. cv. N 532 MARSHALL: THE ACTION OFredistilled under the ordinary pressure a t which it boiled, between195O and 220°, leaving a small quantity of residue, which w aadded to (C).The distillate contained a small quantity of phenylisopropyl ketone, for it gave a phenylhydrazone which melted a t126O (Claus, J . pr. Chem., 1892, [ii], 46, 480), but the greater partof the fraction consisted of benzyl alcohol, which was identified byconversion into benzyl chloride and pnitrobenzyl chloride(m. p. 72-73O). The distillate (C) was redistilled under 15 mm.pressure, and a fraction, which boiled steadily a t 165O, was thusobtained. It gave a phenylhydrazone which separated from alcoholin faintly yellow leaflets, melting a t 145O.0.2276 gave 23 C.C. N, a t 19O and 755 mm.This would agree with a formula, CISHl8N2 (N=11.76 per cent.),isomeric with that of the phenylhydrazone of phenyl isopropylketone.An oxime (m.p. l l O o ) was also obtained, but the constitution ofthe fraction (C), which is apparently a ketone, has not yet beendefinitely settled.N=ll.82.Action of Benaaldehyde on Magnesium Phenyl Bromide.A solution of 12 grams of magnesium in a mixture of 75 gramsof bromobenzene and 150 C.C. of ether was dropped very slowly intoa cooled solution of 106 grams (2 mols.) of benzaldehyde in 200 C.C.of ether. A t first a bright yellow solution was obtained, but as theaddition proceeded a viscous oil separated from the ether. Themixture was allowed t o remain overnight, and was then warmedduring twelve hours on the water-bath, after which time the pro-duct was decomposed in the usual manner. Two chief fractionswere obtained on distillation in a vacuum, the former (A) boilingbelow 150°, and the second (B) distilling between 150° and 200°,whilst a small quantity of tetraphenylethane (m.p. 209O) was alsoobtained. The fraction (A) was shaken with sodium hydrogensulphite solution to remove excess of benzaldehyde, and the remain-ing liquid distilled a t 204-208O under the atmospherlc pressure.Boiling with concentrated hydrochloric acid converted this liquidinto benzyl chloride (b. p. 176O), which on nitration gave pnitro-benzyl chloride.Fraction (B) was redistilled, and the greater part or it distilledconstantly a t 180°/15 mm. The distillate did not solidify for someconsiderable time, but the addition of a trace of benzophenone pro-duced instant crystal'lisation of the whole mass.The substance wasrecrystallised from alcohol, when it melted a t 48O, and it did notaffect the melting point of a specimen of benzophenone obtained inThe fraction (A) was therefore benzyl alcoholALDEHYDES ON THE GRIGNARD REAGENT. 533the usual manner.its properties with those of benzophenoneoxime (m. p. 140O).The oxime was prepared, and this agreed inAction of Acetaldehyde on Magnesium Phenyl Bromide.Twelve grams of magnesium were dissolved in a mixture of75 grams of bromobenzene and 50 C.C. of ether, and to this solutionwas slowly added, with cooling, a solution of 45 grams ofacetaldehyde in 90 C.C. of ether. The grey, crystalline mass whichfirst separated became viscous only when the mixture had remainedovernight, and it was then heated during eight hours on the water-bath, decomposed with acid, and, after dehydration, the etherealextract was distilled.The distillate passing over below120°/20 mm. weighed 40 grams, and above that temperature afurther quantity of 7 grams was obtained, whilst no appreciableamount of residue was left. After further fractionation of the firstdistillate, a quantity of substance was obtained which boiled between90° and 100°/15 mm., whilst a small amount of diphenyl (m. p. 70°)was isolated. The distillate, which was evidently a mixture, con-tained a bromine compound, which was identified by allowingsodium t o react with an ethereal solution of the substance.Diphenyldimethylethane (m. p. 1 2 4 O ) was isolated in this reaction,from which it is inferred that phenylethyl bromide was present inthe original distillate.The remaining component of the distillatewas identified as acetophenone. By roughly estimating the amountof the phenylhydrazone formed from a weighed quantity of thedistillate, it was conchded that acetophenone formed about 60 percent. of the product.A ction of Ethyl Formate on Magnesium Diphenylcarbinyl Bromide.A solution of 12 grams of magnesium in a mixture of 50 C.C. ofether and 75 grams of bromobenzene was obtained; t o this wasadded a mixture of 50 C.C. of benzaldehyde in 50 C.C. of ether, andthe whole was allowed to remain for some hours. A solution of40 grams of ethyl formate in 50 C.C. of ether was slowly added,and the mixture was heated for eight hours.The reaction mixturewas now decomposed with dilute acid; the ethereal layer wasseparahed, dehydrated, and distilled under 15 mm. pressure. Threefractions were obtained: (A) 12 grams boiling a t about looo,(B) 26 grams boiling a t 1 7 8 O , (C) 4 grams boiling a t about 250°,as well as a large amount of residue, which was not examined.The fractions were examined in order. (A) consisted of benzylbromide. The liquid, boiling a t 178O, was also a bromo-derivative,and by allowing sodium to react with an ethereal solution of thisN N 534 THE ACTION OF ALDEHYDES ON THE GRIGNARD REAGENT.substance, tetraphenylethane was obtained, so that the originalsubstance must have been diphenylmethyl bromide. This was con-firmed, first, by the preparatio'n of benzhydryl ethyl ether, and,secondly, by the production of benzhydrylamine from it.Thesubstance was not obtained in the crystalline condition; but thiswas probabb due to the presence of some benzhydrol, from whichi t could not be separated by distillation. The fraction (C) solidifiedin the receiver, and it was found to be tetraphenylethane. Itcrystallises from acetic acid in needles, melting a t 2 0 9 O .Action of Ethyl Formate on Magnesium Ph.estylmethylcarbiny1Zodide.The reaction was carried out in a manner similar to that immedi-ately preceding, and from the products, which soon became dis-coloured owing to the separation of iodine, only phenylethyl iodide,diphenyldimethylethane (produced by decomposition of phenyl-ethyl iodide), and a trace of methyldeoxybenzoin were isolated. Asethyl a-phenylpropionate was the product expected in this reaction,the whole of the fraction boiling below 110°/15 mm. (which frac-tion contained the phenylethyl iodide, and should have containedand ethyl a-phenylpropionate produced in the reaction) was boiledwith alcoholic potassium hydroxide solution, but no trace ofa-phenylpropionic acid could be found.A ction of Triolcyme t hylene on Magnesium Phenglmet hylcarbinylIodide.One molecular proportion of dry, finely-powdered trioxy-methylene was added to the Grignard mixture obtained frommagnesium methyl iodide and benzaldehyde, and the whole wasthen heated during several hours on the water-bath. The productsof this reaction were the same as those mentioned in the precedingcase. No trace of a-phenylpropaldehyde could be found,although, as this aldehyde very easily combines with sodiumhydrogen sulphite, it would have been an easy matter to detect itspresence.THE UNIVERSITY, LEEDS