1968 DEAKIN AND WII,SMOIiE : SOME REACTIONS OF KErEN.CCIX. --Some Reactions of Kiten.€Iyd?-ocyanic Acid.By STELLA DEAKIN and NORMAN THOMAS MORTIMER WILSMORE.IN a note published by Miss F. Chick and one of us (Proc., 1908,24, 77), it was stated, among other things, that keten combinedwith liquid hydrogen chloride under pressure at the ordinarytemperature to form acetyl chloride :CH,:CO + HCr= CH3*C?OCl,there being no evidence of the formation of the isomeric chloro-acetaldehyde or of its condensation product, dichloroethyl acetate.Under similar conditions keten also combined with hydrogensulphide, forming acetyl sulphide, the so-called thioaceticanhydride: *2CH2:C0 + H2S = (CH,*CO),S.Finally, it was mentioned that keten reacted with hydrocyanicacid, but that the product was not, as might have been anticipated,acetyl cyanide or pyruvonitrile.The reaction between keten and hydrocyanic acid has now beenfurther investigated, and a new compound, liquid at the ordinarytemperature, has been isolated in a pure state.The actual con-stitution of this compound has, however, not yet been elucidated ;but, as our collaboration has come to an end, and as t.he substancein question has somewhat remarkable properties, we venture toplace on record the results so far obtained.Analysis and molecular weight, the latter having been found bothby the vapour density and the cryoscopic methods, showed thesubstance to have the empirical formula C,H,O,N, its formationtaking place according to the equation:2CH2:C0 + HCN = C5H502N.The difficulty in the way of determining its constitution lies inits instability in presence of reagents.For, although it has aboiling point of 1 7 3 O , and its vapour does not dissociate a t thetemperature of boiling aniline, it behaves in all the reactionshitherto studied as if it were merely a mixture of keten and hydro-cyanic acid, were it not that the velocity of reaction is slower.Thus with water, eitber alone or in the presence of acids oralkalis, acetic and hydrocyanic acids are slowly formed :CYomlhatiorL~ withC5H,02N + 2H20 = 2CH,*C02H + HCN.* In the note referred to there is a misprint, the b.p. of the acetyl sulphideobtained being given as 55-58" instead of 155-158"DEAKIN AND WILSMORE : SONE REACTIONS OF RE'I'EN 1960Alcohol in presence of a trace of mineral acid acts in a similarway, giving ethyl acetate and hydrocyanic acid:C,H,02N + 2C2H5*OH = 2CH3*C02*CzH5 + HCN.With aniline a violent reaction takes place, acetanilide andhydrocyanic acid being produced :C5H502N + 2C,H5*NHz= 2CH3*CO*NH*C,H5 + RCN.Saturation of an ethereal solution of the substance with hydrogenchloride at a low temperature, followed by cautious addition ofwater, gave only ammonium chloride together with acetic and formicacids, and an attempt to reduce it by means of hydrogen in presenceof platinum black led to no definite result. Obviously therefore thesubstance is not a derivative of cyclobutan-1 : 3-dione.Although we ha.ve no very definite experimental evidence tooffer in support of it, still we may perhaps be allowed t o hazard aguess as to the constitution and mode of formation of the newcompound.We may suppose that hydrocyanic acid combines inthe first place with a, portion of the keten in a way similar to itsreaction with aldehydes and ketones to form a cyanohydrin, whichin this case would be the nitrile of the unstable a-hydroxyacrylicacid :CH,:CO + HCN = CH2:C(OH)*CN.As this nitrile contains a hydrox'jrl group, the latter would a tonce react with more keten to form the corresponding acetate:CH,:C(OH).CN + CH2:C0 = CR,:C(OAc).CN.It may be objected to this hypothesis that a-hydroxyacrylonitrilewould change, at least in part, into the well-known isomeric acetylcyanide or pyruvonitrile, which should accordingly be found amongthe products of the reaction.It may well be, however, that at thetemperature of the experiment this intramolecular change is com-paratively slow, whereas the reaction between keten and hydroxylis practically instantaneous. A second necessary condition is thatthe combination of hydrocyanic acid with keten shall be slow,otherwise, with excess of hydrocyanic acid present, there wouldbe no keten availab€e for the second stage of the process. Thatthis condition was fulfilled in our experiments was shown by thefact that a large part of the keten had time to polymerise tocyclobutan-l : 3-dione, although hydrocyanic acid was present inconsiderable excess. Following out the hypothesis, the decom-position with water may be supposed to proceed thus:CH,:C(CN)*O*CO*CH, + H20=CH2:C(OH)*0.CO*CH3 + HCN= (CH3*C0),O + HCN.(CH,*CO),O + H,O = 2CH3*C0,H1970 DEAKIN AND WlLSMORE : SOME REACTIONS OF KETEN.The reaction with aniline may take place as follows :CH, : c (CN) co CH3 + C6H5* NH,= CH,:C(CN)*OH + CH,*CO.NH*C,H,= CH,*C'O.CN + CH3*CO*NH*C,H,.CH3*CTO*CN + C6H5*NH2 = CH3-CO*NH*C6H, + HCN.The reaction with alcohol is less easy to follow, especially as itrequires the presence of mineral acid. It may be mentioned thatthe structure suggested is consistent with the molecular volume,refractivity, and dispersion of the substance.The actions of Grignard's reagent and of nitrosyl chloride onketen have also been studied. I n both cases the keten condensedfor the most part to brown, resinous substances, but from thereaction with Grignard's reagent a small quantity of acetone wasobtained, probably formed thus :CH,:CO -+ CH2:C(OMgT)*CK, -+ CH,:C(OH)*GH, -+CH,* CO*CH,.It was thought that nitrosyl chloride might give nit.rosoacety1chloride or the isomeric oximinoacetyl chloride, but the only sub-stance that could be isolated in a pure state wm chloroacetylchloride.EXPERIMENTAL.Prepa.ration of Keten and of the Hydrocyanic Acid Derivative.The pyrogenic method of preparing keten, previously describedby one of us (Trans., 1907, 91, 1938), has been somewhat improved.Spiral condensers provided with a cylindrical bulb a t the lowerezcl are now used in place of the simpler pattern previouslyillustrated, and it has been found advisable t o insert a secondtrap between the generator and the main condenser.Any ketenwhich condenses in the traps can be readily distilled over into themain condenser a t the end of the operation. The main condenseris kept at - l l O o to -120° by means of a bath of alcohol andether cooled by suitable addition of liquid air, which is not pouredinto the bath itself, but into a kind of flattened, metallic test-tubesuspended from the top of the Dewar vessel. A t a lower tem-perature the spiral of the condenser becomes rapidly blocked withcrystals, while at a higher temperature the condensation of theketen is incomplete, owing to its dilution with other gases. Aplatinum wire of about 0.2 mm. diameter and 5 cm.free lengthhas been found t'o give the best results. With a longer or thickerwire some keten will be carried away by the increased rush of gas,m the wire must be kept nearly at its melting point in any case. Acurrent of about 7 amperes is used to commence with, but this mustbe gradually raised to 12 or 14 amperes as the reaction proceedsDEAKIN AND WILSMORE : SOME REACTIONS OF KETEN. 1971as the wire becomes coated with a layer of conducting carbon.The acetic anhydride is added in charges of about 50 grams, aclean wire being used for each fresh charge. By carefully removingthe layer of carbon by crushing it with pliers, one wire may some-times be made to serve for three charges before breaking. Afterheating for about an hour and a-half, about half of the aceticanhydride will have disappeared, and the remainder will havebecome dark brown.The reaction is then stopped, the spent chargeis removed, a clean wire is inserted, and a fresh charge of aceticanhydride is poured in. Including subsidiary operations, three orfour such charges can be run through in a day, yielding altogether15 to 20 C.C. of crude keten. The acetic anhydride should be aspure as possible, or the yield will be much reduced; but inferiorsamples may be greatly improved in this respect by previouslydistilling from phosphoric oxide. Keten may also be preparedfrom acetone, but is then very impure. Strange to say, glacialacetic acid appears to be quite unacted on by the hot wire.To prepare the hydrocyanic acid derivative, 15 to 20 C.C.of ketenwere distilled into an exhausted bomb-tube cooled in liquid air.About twice this quantity of anhydrous hydrocyanic acid was added,and the tube was then sealed off before the blowpipe. The hydro-cyanic acid was prepared by the action of 50 per cent. sulphuric acidon potassium cyanide, and was dried by passing i t through a longcolumn of calcium chloride. It was stored in a bulb provided withtwo taps, from which it could be distilled into the bomb tubes asrequired. As there was an interval of only about loo betweenthe melting point of the hydrocyanic acid and the temperature atwhich the keten began rapidly to polymerise, arrangements had tobe made to keep the keten cold until the hydrocya.nic acid hadmelted, and then to mix the two liquids as rapidly as possible.Accordingly, on removing the bomb tube from the liquid air, thelower end containing the keten was well jacketed in wool, the upperpart containing the hydrocyanic acid being left exposed to the air,and the tube was placed in an inclined position in a shakingmachine and shaken for about two hours, or until it had attainedthe temperature of the room.It was then placed in an ice-safeuntil wanted. Since considerable pressure was necessarily developedin the tubes as the temperature rose, to minimise the effects of apossible explosion, the shaking and removal to the ice-safe wereattended to by one of us a t times when the laboratory was otherwiseunoccupied. The contents of the tubes were only slightly colouredbrown, showing that much less brown resin had been formed thanwhen keten polymerises in the ordinary way.Before opening, thetubes were again cooled to -78O, when, as t,he contents ha1972 DEBKLN AND WlLSMORE : SObIE REACTIONS OF KETEN.solidified, they could be opened with impunity, and they were thenremoved from the bath and allowed to attain the temperature ofthe room. As the contents showed the marked tendency to bumpcharacteristic of hydrocyanic acid, a capillary tube connected witha source of dry hydrogen was passed down to the bottom of thebomb-tube as soon as the contents were sufficiently melted. Withthe help of the current of hydrogen, most of the hydrocyanicacid was then distilled off, the tube being finally warmed to about50° in a water-bath.The contents of three such tubes were frac-tionally distilled under 100 mm. pressure, a Claisen flask beingemployed, to the second neck of which a Young’s ‘‘ pear ” still-headwith four bulbs had been sealed. The liquid separated mainly intotwo fractions, one boiling at about 70--80°, which was chiefly cyclo-butan-1: 3-dione, and the otEer boiling at 100-llOo, which con-tained the bulk of the new substance. Some brown residue wasleft in the flask, and crystals of dehydra.cetic acid, which wereidentified in the usual way, and which were due to the poly-merisation of cyclobutan-1: 3-dione, were deposited in the neck ofthe flask and lower portion of the still-head. After repeatedfractionation of the portion of higher boiling point, 7 grams of acolourless liquid were obtained, which boiled constantly at110*0-110~4°/100 mm.(corr.), this being the yield from about50 C.C. of crude keten. The fractionation was carried out in dryhydrogen. Carbon dioxide mas unsuitable for this purpose, as itwas very soluble in the liquid.Composition and Properties of the Hydrocyanic Acid Derivative.The carbon and hydrogen were determined by the method pre-viously used for acetylketen ” (Trans., 1908, 63, 947). Attemptsto estimate the nitrogen by the Kjeldahl method were unsuccessfulowing to loss of hydrocyanic acid, and a modification of the Dumasmethod was therefore emploxed. The substance was weighed inan exhausted thin-walled glass bulb, the weight of the air removedfrom the bulb (0.66 c.c.=0*8 milligram) being added to theapparent weight of the substance.The bulb was packed in copperoxide in the combustion tube, and, when all the air in the latterhad been swept out with carbon dioxide, i t was broken by meansof a pointed glass rod attached to the inlet tube. The vapourdensity was determined by the Hofmann method, using freshlydistilled aniline in the outer jacket, The volume observed wascorrected for the gas and vapour driven off from the walls of theendiometer, and, in measuring the pressure, allowance was madefor the temperature of the mercury column and for the vapourpressure of mercury at 184O, the boiling point of aniline under756.7 mm. pressure. The volume remained constant for abouDEAKIN AND WILPMORE : SOME REACTIOKS OF KETEN. 1978half an hour, showing that the vapour of the substance was stableat the temperature of boiling aniline.The benzene used in deter-mining the molecular weight by the cryoscopic method had beer,previously distilled from sodium :0.1705 gave 0.3393 CO, and 0.0672 H20.0.19990.08590.0882, in 16-93 benzene, gave A t = - 0'260O.C=54.2; H=4*4.0.1369 ,, 0.2702 CO, ,, 0.0580 H20. Cz53.8; H=4.7.,, 21.7 C.C. N, (dry) at 14.8O and 755.1 mm.,, 58.6 C.C. at 184O and 375.9 mm.N=12.8.M.W. =111.2.M.W. = 100.2.0.0697, ,, 16.93 ,, ,, A t = - 0.186O. M.W. = 110.8.0.1157, ,, 16.93 ,, ,, A t = - 0.298'. M.W. = 114.8.C,H,O,N requires C = 54.1 ; H = 4.5 ; N = 12.6 per cent.M.W. = 111.04.Under 772 mm. pressure, the liquid boiled at 173O (corr.), butit began t o turn brown a little below that temperature.Oncooling, a white opalescence appeared at about -45O, which dis-appeared again on warming to -42O, but this must have been dueto a trace of moisture or other impurity, as the substance remainedliquid, although it became very viscous, down to -78O. On coolingwith liquid air, it froze to a white solid, which melted at -196'to -195O. Owing t o the presence of the opalescence, the meltingpoint could not be found by visual observation, but it was readilydetermined by means of a thin wire, moisture being excludedby jacketing the wire with calcium chloride in the upper part ofthe tube. On repeated trials, the wire was found to be immovableup to -196O, but it could be moved up and down at - 195O.The density was aetermined at various temperatures.Twosamples of the substance were used, the first of which had beendistilled some weeks, and the second a few days before $he experi-ment. The densities in each series lie very nearly on it straightline, but there is a difference of about five parts in 10,000 betweenthe two series. As a mean of four weighings, the pyknometercontained 0.74325 gram of water weighed in air at 18O. Thetemperatures in the table have been corrected to the hydrogenscale :t.0 -3O10'013.914.216.618.521 '322'924.829.930 '3Weight of substance iii air. T I.0.7S970.79600 '79370-79010-78700.78260.79950.79610.79250.78890.78334&. - I.I I.1'07461 *0743l'C6951.06971'06641.06471'06141.05981.05711'05213'0611974 DEAKIN AND WILSMORE : SOME REACTIONS OF KETEN.Taking the mean of the two series by graphic interpolation, thedensity of the liquid between loo and 30° is given by the equation :46t = 1.0685 - 0.001 12 (t - 15').Consequently, assuming ths molecular weight to be 111.04, themolecular volume at 1 5 O is 103.9, whereas the molecular volumecalculated for the formula CH,:C(CN)*O*CO*CH, from Traube'svalues for the atomic volumes (Traube, Grundriss der pi~ysikalischenChemie, 1904, p. 120 ; Smiles, Chemical Constitution and PhysicalProperties, 1910, p. 125) is 100-2.The refractive indices of the same two samples were measuredby means of the Pulfrich refractometer.The densities wereobtained by interpolation from the values given above, the twoseries being kept separate :Line. Sample. t. 4% AT. M. R.C, I. 16'9" 1.0667 1'42443 26.580C. 11. 22'6 1'0598 1.42150 26-598D. I. 16'0 1.0676 1.42771 26.734D. 11. 23 '1 1 *0592 1.42435 26.771G'. 11. 22 -5 1.0599 1.43756 27-478Taking the mean of the two series, and using the values for theatomic refractivities and dispersions given in Landolt andBornstein's '' Tabellen," we obtain the following values for themolecular refractivity and dispersion :C D at- cFound ... 26-59 26.75 0'89Calculated for CH,:CiCN)*O*CO*CH; 26-34 26-49 0.79At the ordinary temperature the hydrocyanic derivative is acolourless, somewhat oily liquid. It has a rather pleasant, althoughslightly pungent odour, resembling that of the nitriles.It isreadily miscible with all the ordinary organic solvents, but issparingly soluble in cold water. It is more readily soluble in hotwater, separating again as an emulsion on cooling. It is, however,slowly decomposed and dissolved by water on keeping, even in thecold, acetic and hydrocyanic acids being formed, and this whetherthe aqueous solution was originally neutral, acid or alkaline; buton warming it with aqueous alkalis or sodium carbonate, somebrown substance is also formed. This hydrolysis has been studiedquantitatively in both acid and neutral solution. Since methyl-orange cannot be used as an indicator for acetic acid, and phenol-phthalein is useless in presence of soluble cyanides, the followingmethod of titration was adopted. Standard alkali free fromcarbonate was first added in slight excess, that is, in the proportionof rather more than three equivalents of alkali t o one moleculeof the substance, and the cyanide wits titrated with standard silveDEAKIN AND WILSMORE : SOME REACTIONS OF BETEN.1975nitrate, the end-point being shown, as usual, by the formation of apermanent opalescence. A second equal volume of silver solutionwas then added, in order to replace all the CN' in solution byNO,', after which the excess of alkali could be found by meansof standard acid and phenolphthalein. Proceeding in this way,the alkali used corresponded with the hydrocyanic as well as withthe acetic acid, so that, t o find the amount of t.he latter, the hydro-cyanic acid as found from the silver titration had to be subtractedfrom the total acid.To carry out the hydrolysis in acid solution,a weighed amount of the substance was placed in a stoppered flask,together with 50 C.C. of water and 1 C.C. of 0-1092N-sulphuric acid.The mixture was kept overnight, warmed for a few minutes on thewater-bath, cooled, and titrated, the sulphuric acid being allowedfor.0.3041 required 13.33 C.C. silver and 76-91 C.C. alkali; or 1 mol.gave 0.973 mol. HCN and 1.988 mol. C,H,O,0.2943 required 12-90 C.C. silver and 72-69 C.C. alkali; or 1 mol.gave 0.975 mol. IECN and 1-919 mol. C2H402.For the hydrolysis in alkaline solution, the substance was addedat once to excess of the standard alkali.After about two hours atthe ordinary temperature, the reaction appeared to be complete,and the solution was titrated :0.2820 required 12.16 C.C. silver and 70.85 C.C. alkali; or 1 mol.gave 0.967 mol. HCN and 1.975 mol. C,H402.0.3142 required 13.86 C.C. silver and 80.80 C.C. alkali; or 1 mol.gave 0.979 mol. HCN and 2.032 mol. C2H402.The reaction is thus not quite quantitative, but i t is sufficientlyso to prove the validity of the equation given on p. 1968.An attempt was next made to hydrolyse the CN group withoutbreaking up the molecule. 2.33 Grams of the compound weredissolved in about 20 C.C. of dry ether; the solution was cooled to-78O, and dry hydrogen chloride was passed in until two layersbegan to separate, the lower one being a solution of ether in liquidhydrogen chloride.Two molecular proportions of water, dissolvedin 20 C.C. of ether and cooled to -78O, were then added, and themixture was kept overnight, the temperature gradually rising toabout Oo. As no reaction appeared to have taken place, even onwarming to the boiling point of ether, a third molecular proportionof water was added, and ether was evaporated until water beganto separate. On again keeping overnight at the ordinary tem-perature, ammonium chloride crystallised out, and the solution wasthen fractionally distilled. After the ether had been expelled, afraction passed over at 85--103O, which gave all the reactions offormic acid. A second fraction, which passed over at 103-135O,The silver nitrate was 0*1N, and the alkali 0.1055N 1976 DEAKIN AND WILYMORE : SOME REACTIONS OF KETEN.was also acid.It was redistilled from concentrated sulphuric acidto destroy any formic acid, and it then answered to the tests foracetic acid. A small quantity of liquid which still remained inthe flask was heated under a pressure of 35 mm. to 230°, but i tmerely charred without distilling. The hydrolysis of the hydro-cyanic acid derivative had therefore followed practically the samecourse as in aqueous solution.The substance did not appear to react with absolute alcohol, buton adding dry hydrogen chloride the odour of ethyl acetate wasnoticed. On treating another portion of the alcoholic solutionwith lime, a vigorous reaction took place, hydrocya.nic acid andethyl acetate being given off.To a larger portion of the solutiona small quantity of ethylsulphuric acid was added, and the mixturewas left for two days, when hydrocyanic acid and ethyl acetatewere again noticeable. On passing a current of dry air throughthe solution and then into water, hydrocyanic acid was readilyidentified in the latter. The alcoholic solution was then boiled forsome time over phosphoric oxide in a flask fitted with a refluxcondenser, and finally distilled, when, after all the hydrocyanic acidhad been driven off, a sample of ethyl acetate was obtained, whichboiled steadily at 7 7 O (corr.).The compound reacted vigorously with aniline, hydrocyanic acidbeing given off, and a solid substance being formed, which proved tobe acetanilide, melting at 1 1 3 O (corr.).This was confirmed bytaking a mixed melting point with a sample of pure acetanilide fromother sources. The reaction appeared to be quantitative.Action of Grignctrd’s Reagent on Keten.The reagent (magnesium methyl iodide) in dilute ethereal solutionwils cooled to -50°, and gaseous keten was slowly passed in. Avery vigorous reaction took place, and much heat was evolved, aswas shown by the amount of solid carbon dioxide which wits requiredto maintain the low temperature. As only resinous substancesappeared to be formed, the experiment was repeated, but this timethe keten was largely diluted with dry hydrogen, the dilutionbeing effected by passing the hydrogen through liquid keten at-78O to -7OO.The ethereal solution became dark reddish-brownand very viscous, and a yellow, resinous solid separated. Themixture wits decomposed with powdered ice, and the magnesiumhydroxide was dissolved by addition of dilute hydrochloric acid,leaving a considerable amount of brown resin, which was not solublein either the water or the ether. This was removed by filtration,and the filtrate wa.s fractionally distilled. After the ether hadbeen removed, st small quantity of liquid passed over at abouDEAKIN AKD WILSMORE: SOME REACTIONS OF KETEN. 1977looo, which had a strong odour of acetone The presence of acetonewas confirmed by the iodoform reaction, by the alkaline mercuricchloride and nitroprusside tests, and by the formation with benz-aldehyde of distyryl ketone (m.p. 111-112°, corr.). No othercompound could be isolated.Action of Nitrosyl Chloride on Keten.The nitrosyl chloride was prepared by passing nitric oxide intoliquid chlorine at about -60° until saturated. The chlorine wasobtained from hydrochloric acid and potassium permanganate. Itwas washed with water and dried with sulphuric acid. The nitricoxide was produced by the action of nitric acid on copper, and waspurified by passing first through 50 per cent. potassium hydroxidesolution and then through concentrated sulphuric acid. Thenitrosyl chloride was fractionally distilled before use to remove anyexcess of chlorine.A t first, attempts were made to carry out the reaction by con-densing the keten with excess of nitrosyl chloride in bomb-tubes bymeans of liquid air, and allowing the temperature gradually to rise,but violent explosions took place as soon as the nitrosyl chloridemelted.Accordingly, another method was tried, gaseous keten beingpassed slowly into excess of nitrosyl chloride at a temperature justabove the melting point of the latter (-61O). Vigorous reactiontook place at once, with evolution of much heat, hydrogen chlorideand an odour of carbonyl chloride being given off, and a whitesolid separating out, which melted at about -5OO. I f the tem-perature was allowed to rise during the reaction, much gas wasevolved, which, on analysis, proved to be a mixture of hydrogenchloride, carbonyl chloride, carbon dioxide, acetylene, and nitrogen.The acetylene and some or all of the carbon dioxide were impuritiesin the original keten. Afterall the keten had been added, the temperature was allowed to rise,and a current of hydrogen was passed through the mixture in orderto remove the excess of nitrosyl chloride, when a dark brown, veryviscous liquid was left. On distilling this under a pressure of 95to 100 mm., a colourless liquid passed over at 58-68O. This liquiddid not contain nitrogen. Under atmospheric pressure it boiled at106O. It reacted with water to form chloroacetic acid, and withaniline it gave chloroacetanilide, melting at 1 3 3 O , which was con-firmed by taking a mixed melting point with a sample of chloro-acetanilide from another source. The colourless liquid was thereforechloroacetyl chloride, in the formation of which the nitrosyl chloridehad apparently acted merely as a source of chlorine. A largequantity of a brown, resinous substance was left in the distillingThe gas did not contain nitric oxide.vor,. XCVII. G 1978 CHICK AND WILSMORE : THE POLYMERISATION OF KETEN.flask. This was found to contain nitrogen, but a definite com-pound could not be obtained from it. Attempts to obtain il cleanreaction by previously diluting the nitrosyl chloride with dry ethermet with no better success.U N I V E R S I T T COLLEGE,UNIVERSITY OF LONDON