CONVERSION or AMINO-ACIDS INTO TERTIARY AMINO-ALCOHOLS. 283 XLVI:.-Conversion of Amino-acids into Tertiary Amino-alcohols. By ALEX. MCKENZIE and GEORGE OGILVIE WILLS. THE first application of Grignard reagents towards esters of amino-acids was made by Baeyer and Villiger (Ber. 1904 37 3191), who converted methyl anthraiiilate into o-aminotriphenylcarbinol by means of magnesium phenyl bromide. Paal and Weidenkaff (Ber. 1905 38 1686) obtained p-amino-ua-diphcnylethyl alcohol from ethyl glycine (compare also Paal and Weidenkaff Ber. 1906, 39 810 4344; Krassusky Cornpt. rend. 1908 146 236). The Grignard reaction has been also applied in this laboratory for preparing amino-alcohols in connexioii with the study of semi-pinacolinic deamination. Since other workers (Thomas and Bett-zieche 2.physiol. Chem. 1924 140 244 261 279; Bettzieche, ibid. 273) have entered the same field quite recently we think it desirable to describe our further results. The elimination of the amino-group in tertiary amino-alcohols has heen dealt with (BlcKenzie and Richardson J . 1923 123 79; McKeiizie and Roger J. 1924 125 844; McKenzie and Dennler, J. 1924 125 2105). The first instance described in the literature as a semipinacolinic deamination arose from the accidental discovery that the transformation OH*CPh,CHPh*NU 3 OH*CPli,*CHPh*O 284 MCKENZIE AND WILLS CONVERSION OF could not be effected by the agency of nitrous acid; the product of the action was phenyldeoxybenzoin and not triphenylethylene glycol. p-Amino-ctap-triphenylethyl alcohol was prepared by McKenzie and Barrow (J.1913 103 1331) by the action of mag-nesium phenyl bromide on r-desylamine hydrochloride and later on r-phenylaminoacetic acid by McKenzie and Richardson. Thomas and Bettzieche have repeated its preparation from phenylamino-acetic acid and confirmed the conversion into phenyldeoxybenzoin. The conversion of r-phenylalanine into r- p-amino-p-benzyl-acc-diphenylethyl alcohol was also carried out in this laboratory and it was shown that the amino-alcohol undergoes semipinacolinic deamination : 7 Fh H ph I (bv HNO,) CH,Ph*y-y=Ph - + CH,Ph*C-yPh + NH OH ":&-OH OH CH2Ph>CH*CO*Ph. 4 bh CH,Ph*QH-(!*Ph * Ph -0 The phenyl group migrates and the product is benzyldeoxybenzoin. Those results are also confirmed by Thomas and Bettzieche.According to Paal and Weidenkaff (Zoc. cit.) the action of nitrous acid on p-amino-am-diphenylethyl alcohol leads to the formation of as-diphenylethylene oxide : I n the light of our work on semipinacolinic deamination it seemed possible that this conclusion was erroneous and that the product was actually deoxybenzoin. By arrangement with us Professor Tiffeneau and M. Or6khov had kindly undertaken the investigation of this point which meanwhile has been solved by Bettzieche. The product is deoxybenzoin and we would interpret its formation on the following lines the phenyl group migrating thus AMINO-ACIDS INTO TERTIARY AMINO-ALCOHOLS. 285 I n a former paper we mentioned that we were continuing our study of the action of nitrous acid on compounds containing the >‘-?< with the object of extending our experience group b H NH, ultimately to optically active amino-alcohols.Accordingly in the present paper we describe the action of magnesium phenyl bromide on r-alanine thus : The amino-alcohol on deaminntion gave met,hyldeoxybenzoin : and thus pfovided another instance of the striking regularity with which the phenyl group migrates in all the cases of semipinacolinic deamination which have been studied. The change may be contrasted with the vinyl dehydration of r-methylhydrobenzoin studied by Tiff eneau and Or6khov (Bull. SOC. chim. 1921 [iv] 29 422) and by McKenzie and Roger inci-dentally to their work on the dehydration of optically active methylhydrobenzoin by concentrated sulphuric acid : The above result has already been arrived at by Bettzieche, who carried out the action with a trace of the amino-alcohol.In the light of the work of McKenzie and Roger as applied to the optically active methyl- and ethyl-hydrobenzoins the extension of this action to the optically active amino-alcohol derived from d-alanine presents obvious points of interest and we are a t present engaged in the investigation of this topic. As already stated the action of nitrous acid on p-amino-asp-triphenylethyl alcohol had been examined with the object of obtaining triphenylethylene glycol. The optically active forms of the latter compound had already been described (McKenzie and Wren J. 1910 97 473; XcKenzie Drew and Martin J., 1915 107 26) and the change in question if carried out with the optically active amino-alcohols would have provided data in connexion with the Walden inversion.This scheme however 286 MCKENZIE AND WILLS CONVERSION OF cannot be effected in practice for the reasons already given. Never-theless we have thought it desirable that the optically active amino-alcohols should a t least be prepared. The description of their preparation is now given in the experimental part. The compounds in question are highly active giving [XI, & 243" in chloroform; they rotate the plane of polarisation in the opposite sense to the phenylaminoacetic acids from which they are prepared. There is no configurative change here however since there is no substitution of a group directly attached to the asymmetric carbon atom so that the compound which is derived from d-phenylamino-acetic acid is designated as d- (3-amino-actp-triphenylethyl alcohol, although it is lzvorotatory d-phenylaminoacetic acid being taken as the reference type.It is rather curious to note that a similar change of sign of rotat'ion also occurs when methyl I-mandelate is converted by magnesium phenyl bromide into triphenylethylene glycol (McKenzie and Wren 7oc. c i t . ) . The action of nitlrous acid on the ci-amino-alcohol gives an optic-ally inactive product owing t o semipinacolii? ic deamination the product being phenyldeoxybenzoin. The isomeric camphorsulphonates of the d- and I-amino-alcohols were prepared but the resolution of the 1.-alcohol by d-camphor-S-sulphonic acid gave a very slow and imperfect separation and the method was obviously unsatisfactory.We are a t present engaged in resolving by means of camphorsulphonic acids the r-amino-alcohols related to desylamine. The changes C<R -+ d- CH,Ph>C- and d- CH,Ph*CH(NH,)*CO,H NH HO are also described but the study of the action of nitrous acid on the amino-alcohols has not yet been completed. Arising from the present research the following points have been investigated. The ethyl ester hydrochloride of Z-phenyl-aminoacetic acid gives on hydrolysis with a slight excess of alcoholic potash a phenylaminoacetic acid with [.ID - 12.4" in hydrochloric acid solution whereas the pure Z-acid has [.ID - 157.8" in the same solvent. This result was anticipated since in phenylaminoacetic acid we have both a phenyl group and a migrational hydrogen atom in direct attachment to the asymmetric carbon atom.McKenzie and Walker (J. 1915 107 1685) have shown that catalytic racemisation occurs when I-phenylbromoacetic acid i AMINO-ACIDS INTO TERTIARY AMINO-ALCOHOLS. 287 acted on by ammonia. When the action was conducted in benzene solution that portion of the bromo-acid which had survived the attack of the ammonia had become largely racemised. It was thought possible that soniething of the same kind might occur when Z-phenylaminoacetic acid is acted on by nitrous acid; such, however is not the case. There was one other rather important point in connexion with tlhe study of the Walden inversion in the mandelic acid series which clearly required elucidation. E.Fischer and Weichhold (Ber. 1908 41 1286) acted on ethyl d-phenylaminoacetate with nitrous acid and obtained a change of sign of rotation the resulting ethyl mandelate being highly racemised and lanorotatory . On the other hand Marvel and Noyes ( J . Amer. Chern. Xoc. 1920 42, 2259) by acting on a solution of thc hydrochloride of ethyl 1-plieiiylaminoacctate in dilute sulphuric acid with nitrous acid, obtained an ethyl maidelate of the same sign of rotation as that of the original ethyl I-phenylaminoacetatc. We have repented the work both of Fischer and Weichliolcl and of Marvel and Noyes, and have found that a change of sign of rotation occurs when ethyl Z-phenylaminoacetate is acted on by nitrous acid if the solution, before the ester is extracted is kept sufficiently long after the addition of the nitrite.is therefore correct. 777) have shown that the change occurs the behaviour both of the ester and of the free acid towards nitrous acid is similar. The product obtained by the method of the American workers was contaminated with a nitrogenous com-pound (probably a diazo-compound). It seems that they had proceeded to isolate their ethyl mandelate too quickly after the addition of the sodium nitrite. The work on the action of nitrous acid on compounds containing Fischer and Weichhold’s conclusion, d-CHPh(NH,)*CO,Et + Z-CHPh( OH)*CO,Et, Since fi1cKenzie and Clough (J. 1909 95, d-CHPh( NH,)*CO,H -+ Z-CI-IPh( OH)*CO,H the grouping >?-?< is being continued. OH NH, E x P E R I M E N T A L. Semipinacolinic Dearnination of p-Amino-aa-diphenyl-n-propyl Alcohol .-Ethyl r-alanine hydrochloride (compare Curtius and Koch J .pr. Chem. 1888 [ii] 38 472; Barker and Skinner J . Amer. Chem. Xoc. 1924 46 403) was conveniently prepared as follows A current of dry hydrogen chloride was passed into a suspension of r-alanine (10 g . ) in ethyl alcohol (200 c.c.) for 1 hour, and the mixture after boiling for 1 hour longer was evaporate 288 MCKENZIE AND WILLS CONVERSION OF to dryness a t the ordinary temperature under 15 mm. pressure. The resulting syrup solidified when kept over soda-lime in a vacuum. It crystallised from ethyl acetate in needles (15.5 g. ; 90% of the theoretical). Ethyl r-alanine hydrochloride (20 g.) was added gradually to the Grignard reagent prepared from bromobenzene (245 g.) and the mixture heated for 8 hours.The product was decomposed by ice ammonium chloride and ammonia. The residual solid was extracted with ether and added to the main ethereal solution. After removal of ether and diphenyl as usual, the product (20 g.) was dissolved in hydrochloric acid and pre-cipitated by ammonia This treatment was repeated and then the product was crystallised three times from aqueous alcohol. P- Amino- cccc-diphenyl -n-prop yl alcohol separates from aqueous alcohol in rhombohedra1 plates m. p. 101-5-102~5" is readily soluble in ethyl alcohol methyl alcohol benzene toluene carbon tetrachloride carbon disulphide acetone or ether and very spar-ingly soluble in water (Found C z 79.4; H = 7-7 ; N = 6.2. C1,HI,ON requires C = 79.3 ; H = 7.5 ; N = 6.2%).A trace added to concentrated sulphuric acid assumes a golden-yellow tint and the solution becomes a t first pink and quickly colourless. A solution of 4-5 g. of sodium nitrite in 20 C.C. of water was gradually added (9 hour) to a solution of 5 g. of the amino-alcohol in 150 C.C. of 25% acetic acid cooled at 0". A flocculent pre-cipitate was gradually deposited which was filtered off after 24 hours and crystallised from aqueous alcohol. The silky needles (m. p. 50-51") which separated amounted to 4 g. and were identified as methyldeoxybenzoin (Found C = 85.8 ; H = 7.0. Calc. C = 85.7; H = 6.7%). There was no depression of the melting point when this substance was mixed with a specimen of the methyldeoxybenzoin prepared by McKenzie and Roger (J., 1924 125 844) by the dehydration of methylhydrobenzoin with concentrated sulphuric acid.The coloration observed wit8h con-centrated sulphuric acid was identical with that described by McKenzie and Roger for their product. The identity of the sub-stance was also proved by its behaviour towards magnesium phenyl bromide when accp-triphenylpropyl alcohol m. p. 86-5-87-5" was isolated. The latter compound had already been prepared by Mlle LBvy (Bull. Xoc. chim. 1921 [iv] 29 878) and by McKenzie and Roger. Hydrochlorides of Ethyl d- and 1-Pheny1aminoncetntes.-The optic-ally active phenylaminoacetic acids were prepared by E. Fischer and Weichhold (Ber. 1908 41 1286) who resolved the formyl r-acid by cinchonine and quinine and then hg'drolysed the d-and I-formyl acids with hydrobromic acid.The resolution o AMINO-AC'TDS TSTO TERTTAR'I' AMINO-.iT,C'OHOT>S. 289 ./.-phen~-larninoacetic acid into its opt ic'ally nc t ivc components by means of Reychler's d-camphor- 3-sulphonic acid in aqueous solu-tion has been described by Betti and Xayer (Ber. 1808 41 2071) and by Marvel and Noyes ( J . il?n,eY. Clzem. Soc. 1920 42 2259), the acids being einploj-etl in equiivolecular quantities. I n the inoclification described 1)y Ingersoll a i d Adanis ( J . Arne?.. Phem. Soc. 1922 44 2930) an c s c c ~ of tlic cauil>liorsulphouic acid was used. Since the latter aciti contains Tvater of carystallisation (Pope and Gibson ,J. 19l0 97 2211 ) ? antl sincc this may possibly be ])resent in varying ciiiantitj- T\ c employed this ruethotl.The optically pure 7-acicl was obtainetl by adding the rqiiisite amount of aiiiiiionia to the solntion of the salt o1)tainccl after several c~rystallisations. Thc cl-acid maj- t e preprcd from the acid mixture obtninecl from the first inother-iiqixor of the preceding rcsolutioii either by nieaiis of rl-r.-~rc~~ocaml~lior-,'3-siil~~~i~~iic acid or by formylating and then converting into the quinine salt. For the purification of the d-acid however it is more convenient to com-bine the crude acid with I-camphor-p-sulphonic acid and then to crystsllise until the caiii~~horsulplioiiate is homogeneous. Wc were enabled to conduct the resolution in thc latter iiianiicr owing t o the generous gift of c2 supply of I-csniphor from Sir William Pope.The I-camphorsulplioiiic acid. obtained from E-camphor by snlphon-ation in presence of acetic anhj-clride was crystallised from ethxl acetate and gave in aqueoiis solution 1 = 2 c = 3.8703 c ( ~ - 1-61', I-Camphorsulphonic ac-id combines with I-phenylarninoacetic acid to form a salt which is readily soliible in water from which i t separates in prisms. Its rotation n-as cleterininecl in aqueous solution : The diastereoisonieric salt. prepared froin d-camphorsulphonic acic1 and Lphenylaininoac4c acid has [.ID - 444)'i' (c' = 2.0883, I = 2) in aqueous solution (Hctti and Jfayer loc. cit.). This value enabled us to estimate the progress of the resolution of r-phenyl-amiiioacetic acid by (1- a lit1 Z-cninphorsulplioiiic acids. The polari-metric values for the I - a i d rI-pl~eiiS'1,2minoac.eiic acids obtained agreed with those of Fischer and Weichho1:l.Ethyl 1- P h e ~ ylarniiioacPtntc Ilylroch lor idc .-Fisclier and Weich -hold describe the conversion of I-pheii?.laminoacetic acid into the hydrochloride of its ethyl ester TI liicli they describe as dextro-rotatory [~(]fy f- S8.93" in aqucouc; so1::tjoii (21 = 5.021 d = 1.0087). For example a current of dry hydrogen chloride n-as 1)assecl into a mixture of 3.3 g. of l-phenyl-aminoacetic acid ant1 40 C . C . of etl1j.I alcohol for 1 hours. The solution was then boiled for a few minutes and filtered from a VOL. CXXVIT. L [.ID - 20.8". 1 = 2 G = 4.630 xD - 6*G4" [.ID - 51.7". Our results were (1 iffcrent 290 MCEENZIE AND WILLS CONVERSION OF small amount of solid.The alcohol was removed from the filtrate by gentle warming under diminished pressure the resulting acid dissolved in 20 C.C. of water and an equal volume of benzene added. The addition of the requisite amount of ammonia caused the separ-ation of the ester which was extracted with benzene. A current of dry hydrogen chloride was passed into the benzene solution for 15 minutes when the crystalline ester h~drochloride (4.2 g.) separ-ated. It had the following rotation i n aqiieous sollition 7 =1, Marvel and Noyes also obtained a laworotatory ester hydro-chloride with [.]* - 84.6" (concentration not quoted) from the I-amino-acid. The dextrorotation recorded by Pischer and Weich-hold is possibly due to a typographical error. Ethyl d-phenylapninoacetate hydrochloride was prepared in a similar manner from d-phenylaminoacetic acid 6.1 g.being obtained from 5.8 g. of d-acid. It was dextrorotatory in aqueous solution: Action of Magnesium Phenyl Bromide on the Hydrochlorides of Ethyl d- and 1-Pheny1aminoacetate.-The d-ester hydrochloride (6 g. ; 1 mol.) was gradually added wit!hin 15 minutes to the Grignard reagent prepared from 52 g. of bromobenzene (12 mols.) 16.5 g. of magnesium and 350 C.C. of ether and the mixture heated for 49 hours. After decomposition of the product with ice and ammonium chloride and remaining over-night the ethereal layer was separated and the aqueous layer extracted with ether. The ether and the diphenyl were removed the latter by steam dis-tillation. The residual yellow solid (8 g.) was crystallised from ethyl alcohol until the product after drying over concentrated sulphuric acid in a vacuum until constant gave a value which remained unchanged on polarimetric examination after repeated cry stallisation.d- @-Amino- axp-triphenylethyE alcohol is somewhat sparingly s o h ble in ethyl alcohol and separates in colourless needles m. p. 129.5-130° whereas the r-isomeride (McKenzie and Barrow J. 1913, 103 1331) melts a t 154-5-155". A trace of it added to con-centrated sulphuric acid gives a pink coloration which quickly becomes orange-brown. It is readily soluble in chloroform ether, benzene or acetone (Found C = 83.1 ; H == 6.8. C,,HISON requires C = 83.0 ; H = 6.6%). The substance has the opposite sign of rotation to thc original ester hydrochloride bcing strongly lzevorotatory in chloroform : I = 2 c = 1.276 a:f - 6-19" [a]:? - 243".In benzene 1 = 2, The I-ester hydrochloride (5 g.) was acted on by magnesium c = 5.070 ED - 4-53' [RID - 89.3". 2 = 1 c = 5.070 ED $- 4-60' [.ID 90.7". c = 2.027 - 9-45" [CC]:" - 233" AMINO-AUIDS INTO TERTIARY AMINO-ALCOHOLS. 291 phenyl bromide under conditions similar to those just described. Yield of crude amino-alcohol = 6 g. It was purified by crystallis-ation from ethyl alcohol, 1-p-Amino-aap-triphenylethyl alcohol has m. p. 129.5-130" (Found : N = 4.9. C,,H1,ON requires N = 4.8%). The compound is strongly dextrorotatory in chloroform 1 = 2 c = 1,304 ug5' + 6*34" + 243". I = 2 c = 1.304 $- 7*63" [~]:4;~ + 293". In bcnzene 1 = 2 c = 2.017 a:' + 9.44" [ a ] r + 234".Action of Nitroes Acid on d- (3 -Amino- a a p -triphen ylethyl Alcohol .-A solution of the d-amino-alcohol (0.68 g.) in 50 C.C. of dilute acetic acid was cooled in a freezing mixture of ice and salt and a solution of 0.5 g. of sodium nitrite in 5 C.C. of water added during 20 minutes. The solid (0.6 g.) which separated was crystallised from ethyl alcohol and 0.42 g. of phenyldeoxybenzoin needles m. p. 134-135" was obtained. It gave the characteristic emerald-green coloration with concentrated sulphuric acid. Its solution in chloro-form was optically inactive. Isomeric Camphorsulphonates of the d- and 1-Amino-Alcohol.s.-I-p-Amino-aap-triphenylefhanol d-camphorsulphonate prepared by combining the 1-amino-alcohol (1 mol.) with d-camphor-p-sulphonio acid (1 mol.) in ethyl-alcoholic solution separates in needles, m.p. 200-201" (decomp.) (Found S = 6.3. C,,H,,O,S requires S = 6.1%). In ethyl alcohol I = 2 c = 0.8568 a g + 2.08", [a]':' + 121"; I = 2 c = 04568 a:::* + 2*41" [a]:!&' + 143". The enantiomorphously related d-p-amino-aap-triphenylethanol 1-camphorsulphonate separates from ethyl aicohol in needles m. p. 200-2001" (decomp.) (Found N = 2.9. C,,H,,?N requires N = 2.7%). I n ethyl alcohol 1 = 2 c = 0.80 a:," - 1.91" [a]:' -119"; 1 = 2 c = 0.80, 1- p-Amino-aap-triphenylethanol I-camphorsulphonate separates from ethyl alcohol in needles m. p. 213.5-214.5" (decomp,). In ethyl alcohol 1 = 2 c = 0,4448 ag5' + 0*68" [a]:G' + 76"; I = 2, c = 0.4448 a::; + 0*76" [a];$ + 85". The enantiomorphously related d-P-amino-aap-triphenylethanol d-camphorsdphonate separates from ethyl alcohol in needles m.p. 213.5-2146" (decomp.) (Found S = 6.2. C1,H,,O,S requires S = 6.1%). In ethyl alcohol 1 = 2 c = 0.40 ag' - 0*61", [a]:"' - 76"; 1 = 2 c = 040 a:& - 0*72" [a]::;1 - 90". It is insoluble in most organic solvents moderately soluble in hot ethyl alcohol or water and sparingly soluble in these solvents a t the ordinary temperature. The concentrations employed in the determination of specific rotatory power were necessarily low so that the values for the specific rotatory power have little significance. Attempts were made to resolve the r-amino-alcohol by d-camphor-- 2-31' [a]iSI - 144". L 292 MCKENZIE AND WILLS CONVERSION OF p-sulphonic acid both in aqueous and ethyl-alcoholic solution.The progress of the resolution was however too slow to enable the method to be used as a preparative one for the d- and 1-amino-alcohols. Action of Magnesium Phenyl Bromide on a-Aminohydratropic Acid.-The amino-acid (6 g.; 1 mol.) prepared according to McKenzie and Clough (J. 1912 101 390) was added in instal-ments (Q hour) to the Grignard reagent (12 11101s.) prepared from bromobenzene (71 g.) and the mixture was heated for lo+ hours. After decomposition with ice and ammonium chloride the remain-ing solid was extracted with ether and the solution added to the main ethereal solution. The ether and diphenyl were removed ; the resulting oil solidified. Yield 5 g. It was tlriturated with light petroleum and the solid then crystnllised from rectified spirit until pure.p- Amino- ma p -triphen yl- p-methyleth yl alcohol separates from re cti-fied spirit in rectangular plates is soluble in benzene and chloro-form and melts a t 113-114" (Pound C = 83.0; H = 7.2; N = 4.8. C2,H210N requires C = 83.1 ; H = 7.0; N = 4.6%). A trace of this compound added to concentrated sulphuric acid produces an orange coloration which changes quickly to a permanent crimson colour. Action of Magnesium Phenyl Bromide on r-Phenyla1anine.-No visible action took place when r-phenylalanine (3 g.; 1 mol.) was gradually added to the Grignard reagent (12 mols.) prepared from 34 g. of bromobenzene. The mixture was heated for 20 hours, and then the additive compound was decomposed with ice and am-monium chloride.The ethereal layer was withdrawn and then the ether and diphenyl were removed from it in the usual manner. The residual solid amounted to 4 g. which were crystallised from ethyl alcohol. The yield of pure amino-alcohol was 2-4 g. cor-responding in crystalline form and melting point with r-p-amino-aa-diphenyl- P-benzylethyl alcohol which was prepared by McKenzie and Richardson (J. 1923 123 79) from the ethyl ester hydro-chloride of phenylalanine by the action of magnesium phenyl bromide. Action of Magnesium Phenyl Bromide on d- Phenyla1anine.-r-Phenylalanine was resolved by the alkaloidal method of E. Fischer and Schoeller (Annalen 1907 357 2) and d-phenylalanine was isolated with a specific rotation in aqueous solution of + 35.0" ( I = 2 c = 2.043 aD + 1.43") this value being identical with that quoted by Fischer and Schoeller.Three g. of this amino-acid were treated in the manner recorded in the above experiment, with the exception that the mixture was heated for 24 instead of Yield 2 g AMINO-ACIDS INTO TERTIARY AMINO-ALCOHOLS. 293 20 hours. 4.2 G. of crude omino-alcohol were obtained and were crystallised froin ethyl alcohol several times until a steady value for the specific rotation was obtained in chloroform I = 2 c = 4-06" [CI],;:~;~ + 99.2". In benzene Z = 2 c r= 2.0512 E:!:" + 4-32', This compound has the same sign of rotation as the original amino-acid and since 110 configurative change can take place in the transformation of the amino-acid into the amino-alcohol it is accordifigly designated as d- ~-cr?,iiiio-ax-diphe?~~~- P-be?zxylethyl alcohol.It crystallises from ethyl alcohol in flat glassy needles m. p. 143-1Uo is soluble in acetone h i z z n e chloroform ether or toluene, but is insoluble in T.\.-ater (Found N = 4.6. C',lH,lON requires N = 4-60/,). A trace of the compound added to concentrated sulphuric acid produces an orange-brown coloration which quickly changes to pale pink and then gradually fades. This coloration is also observed with the corresponding r-amino-alcohol. Hydrolysis of Ethfyl 1-Yl~eiz~lnmiiioacetate Hydrochloride.-The Lester hydrochloride (1 g.) in 30 C.C. of ethyl-alcoholic potassium hydroxide (0.4487L\') remained for 24 hours a t the ordinary tem-perature and was then heated on the water-bath for 30 minutes.The alcohol was removed under diminished pressure the residue dissolved in water and the solution extracted with benzene to remove any trace of unchaiiged ester. The aqueous solution was then neutralised with hydrochloric acid and the amino-acid (0.5 g.) was separated. Its specific rotation was determined as follows : 0.3719 g. was dissolved in 3.35 C.C. of X-hydrochloric acid and 1.5 C.C. of water I = 0.5 p = i-193 d = 1.0286 xD - 0*4G" - 12-4". 3'11 c ester Iiydrochloridc was completely hydrolysed . Pa ~t icr l Den in i iiu t ion of I - Phe ii yln nzi 11 oncet ic d c id .-A solution of sodium nitrite (0.41 g.) in water ( 2 c.c.) was added drop by drop (30 minutes) to a solution of the I-amino-acid ( 2 g.) in X-sulphuric acid (30 c.c.) t L t 0". After 5 hours at 0" and 18 hours a t the ordinary tomperaturc the amiiio-acid (1.33 g.) as precipitated by the addition of aniiiioiiia in slight excess.0.7438 G . was dissolved in 6.7 C.C. of n'-hydrochloric acid and 3 c.c. of vcater I = 1 p = 7.193 dJ' = 1.0286 x ~ - L1.36" '[RID - 133-5" whereas thc originzl I-acid had Thc ammoniacal solution from which the amino-acid had been scparated TT~C?S acidified n-ith dilute hydrochloric acid and extracted with cther. The I-esulting mmdelic acid (0.23 6.) was slightly tlestrorotatory in aqueous solution Z = 2 c = 1.055 clD + 0.13". Deuminalioiz. of Ethyl l-Pl~ciiylnmi.Iioacetate.-~~~ethocl of E. F'isclzer und TPeichhoZd. These authors acted on the d-ester with nitrous 2.0472 c(::'* + 3.48" [XI;:'* + 85.0" ; l = 2 c = 2.0472 a3:;' + [a]::' + 105-3"; Z = 2 c = 2.0312 MI,; + 5*03" [cx]~$ + 122.6".- 156' under similar conditions 294 CONVERSION OF AMINO-ACIDS INTO THRTIARY AMINO-ALCOHOLS. acid and obtained a lzevorotatory ethyl mandelate with about [.ID - 10" in acetone solution On following their directions, using ethyl Z-phenylaminoacetate (0.90 g . ) in place of the d-ester, the resulting mandelic ester (0.325 g.) was dextrorotatory. A 10% solution gave ccD + 0.45" in a 0.5 dcm.-tube so that the value for the specific rotation is approximately + 9". The result of Fischer and Weichhold was thus confirmed. It should be stated that after the addition of the nitrite the solution remained over-night a t the ordinary temperature. The large amount of racemisation which accompanied this change is apparent when it is recalled that ethyl E-mandelate has according to Walden (2.physikal. Chem. 1895 17 705) [.ID - 90.6" in acetone solution. The dextrorotatory ester obtained above gave a negative result when tested for nitrogen. It was hydrolysed by 4.5 C.C. of aqueous potassium hydroxide (0.5755N) at the ordinary temperature for 18 hours. The solution was extracted with ether to remove any unsaponified ester and the rnandelic acid isolated from the aqueous solution as usual. The resulting acid (0.14 g.) was dextrorotatory in aqueous solut'ion E = 2 c = 0.622 uD + 0.27". Method of Marvel and Noyes. A solution of sodium nitrite (1.6 g.) in water (2.5 c.c.) was added drop by drop with constant stirring for 30 minutes to a solution of ethyl E-phenylaminoacetate hydrochloride (5.4 g.) in N-sulphuric acid (33 c.c.) the temperature being kept a t 0" throughout.A yellow oil separated. After 1 hour a t 0" and 2 hours a t the ordinary temperature the solution was extracted with ether the ethereal solution dried with anhydrous sodium sulphate and the ether expelled. The resulting oil was distilled and the portion (1.5 g.) b. p. 127-132"/18 mm. collected. I n acetone E = 2 c = 10 uD - 0.35'. The product obtained by Marvel and Noyes was also laevo-rotatory . When this oil was tested for nitrogen it gave a positive result. On hydrolysis under conditions similar to those described in the previous experiment it gave a mandelic acid (0.8 g . ) which was slightly dextrorotatory in aqueous solution I = 2 c = 3.109, This shows that the lzevorotation observed by Marvel and Noyes was not due to ethyl Z-mandelate but rather to the presence of an intermediate lzvorotatory diazo-compound which gradually passes into ethyl d-mandelate with lapse of time.That this view is probable appears from the result of the second experiment quoted below. The unattacked ethyl phenylaminoacetate hydrochloride was recovered from the aqueous solution remaining after the extraotion C(D $- 0.10" NEWBERY THE ACTION OF CAUSTIC ALKALI ETC. 295 of the deamination product with ether. The solution was made alkaline with ammonia extracted with benzene the benzene solution dried and dry hydrogen chloride passed in to precipitate t'he ester hydrochloride Yield 2 g. In aqueous solution 1 = 1 c = 5.07 aD - 469" [ a ] D - 90.5".The regenerated hydrochloride had thus practically the same rotation as the original. In a second experiment where 7 g. of the Lester hydrochloride were employed the conditions of deamination were exactly similar to those just described except that the solution after being kept a t 0" was allowed to remain for 39 instead of 2 hours. The resulting oil (2 g.) which gave a positive test for nitrogen mas in this case dextrorotatory in acetone solution E = 1 c = 10 tcD + 0.30". On hydrolysis with aqueous alkali under similar conditions to those adopted in the other experiments the resulting mandelic acid was dextrorotatory in aqueous solution I = 2 c = 3.32, The amino-ester hydrochloride was recovered as before. In The latter was filtered off.a D + 0.19". aqueous solution 1 = 1 c = 5.07 mD - 4-57" [.ID - 90.1". M7e desire to express our best thanks to the Department of Scientific and Industrial Research and to the Carnegie Trust for their assistance. We are also indebted to Dr. H. J. Plenderleith for his assistance in the deamination of p-amino-act-diphenyl-n-propyl alcohol. UNIVERSITY COLLEGE D UNDEE. UNIVERSITY OF ST. ANDREWS. [Received December 5th 1924. CONVERSION or AMINO-ACIDS INTO TERTIARY AMINO-ALCOHOLS. 283 XLVI:.-Conversion of Amino-acids into Tertiary Amino-alcohols. By ALEX. MCKENZIE and GEORGE OGILVIE WILLS. THE first application of Grignard reagents towards esters of amino-acids was made by Baeyer and Villiger (Ber. 1904 37 3191), who converted methyl anthraiiilate into o-aminotriphenylcarbinol by means of magnesium phenyl bromide.Paal and Weidenkaff (Ber. 1905 38 1686) obtained p-amino-ua-diphcnylethyl alcohol from ethyl glycine (compare also Paal and Weidenkaff Ber. 1906, 39 810 4344; Krassusky Cornpt. rend. 1908 146 236). The Grignard reaction has been also applied in this laboratory for preparing amino-alcohols in connexioii with the study of semi-pinacolinic deamination. Since other workers (Thomas and Bett-zieche 2. physiol. Chem. 1924 140 244 261 279; Bettzieche, ibid. 273) have entered the same field quite recently we think it desirable to describe our further results. The elimination of the amino-group in tertiary amino-alcohols has heen dealt with (BlcKenzie and Richardson J . 1923 123 79; McKeiizie and Roger J.1924 125 844; McKenzie and Dennler, J. 1924 125 2105). The first instance described in the literature as a semipinacolinic deamination arose from the accidental discovery that the transformation OH*CPh,CHPh*NU 3 OH*CPli,*CHPh*O 284 MCKENZIE AND WILLS CONVERSION OF could not be effected by the agency of nitrous acid; the product of the action was phenyldeoxybenzoin and not triphenylethylene glycol. p-Amino-ctap-triphenylethyl alcohol was prepared by McKenzie and Barrow (J. 1913 103 1331) by the action of mag-nesium phenyl bromide on r-desylamine hydrochloride and later on r-phenylaminoacetic acid by McKenzie and Richardson. Thomas and Bettzieche have repeated its preparation from phenylamino-acetic acid and confirmed the conversion into phenyldeoxybenzoin.The conversion of r-phenylalanine into r- p-amino-p-benzyl-acc-diphenylethyl alcohol was also carried out in this laboratory and it was shown that the amino-alcohol undergoes semipinacolinic deamination : 7 Fh H ph I (bv HNO,) CH,Ph*y-y=Ph - + CH,Ph*C-yPh + NH OH ":&-OH OH CH2Ph>CH*CO*Ph. 4 bh CH,Ph*QH-(!*Ph * Ph -0 The phenyl group migrates and the product is benzyldeoxybenzoin. Those results are also confirmed by Thomas and Bettzieche. According to Paal and Weidenkaff (Zoc. cit.) the action of nitrous acid on p-amino-am-diphenylethyl alcohol leads to the formation of as-diphenylethylene oxide : I n the light of our work on semipinacolinic deamination it seemed possible that this conclusion was erroneous and that the product was actually deoxybenzoin.By arrangement with us Professor Tiffeneau and M. Or6khov had kindly undertaken the investigation of this point which meanwhile has been solved by Bettzieche. The product is deoxybenzoin and we would interpret its formation on the following lines the phenyl group migrating thus AMINO-ACIDS INTO TERTIARY AMINO-ALCOHOLS. 285 I n a former paper we mentioned that we were continuing our study of the action of nitrous acid on compounds containing the >‘-?< with the object of extending our experience group b H NH, ultimately to optically active amino-alcohols. Accordingly in the present paper we describe the action of magnesium phenyl bromide on r-alanine thus : The amino-alcohol on deaminntion gave met,hyldeoxybenzoin : and thus pfovided another instance of the striking regularity with which the phenyl group migrates in all the cases of semipinacolinic deamination which have been studied.The change may be contrasted with the vinyl dehydration of r-methylhydrobenzoin studied by Tiff eneau and Or6khov (Bull. SOC. chim. 1921 [iv] 29 422) and by McKenzie and Roger inci-dentally to their work on the dehydration of optically active methylhydrobenzoin by concentrated sulphuric acid : The above result has already been arrived at by Bettzieche, who carried out the action with a trace of the amino-alcohol. In the light of the work of McKenzie and Roger as applied to the optically active methyl- and ethyl-hydrobenzoins the extension of this action to the optically active amino-alcohol derived from d-alanine presents obvious points of interest and we are a t present engaged in the investigation of this topic.As already stated the action of nitrous acid on p-amino-asp-triphenylethyl alcohol had been examined with the object of obtaining triphenylethylene glycol. The optically active forms of the latter compound had already been described (McKenzie and Wren J. 1910 97 473; XcKenzie Drew and Martin J., 1915 107 26) and the change in question if carried out with the optically active amino-alcohols would have provided data in connexion with the Walden inversion. This scheme however 286 MCKENZIE AND WILLS CONVERSION OF cannot be effected in practice for the reasons already given. Never-theless we have thought it desirable that the optically active amino-alcohols should a t least be prepared.The description of their preparation is now given in the experimental part. The compounds in question are highly active giving [XI, & 243" in chloroform; they rotate the plane of polarisation in the opposite sense to the phenylaminoacetic acids from which they are prepared. There is no configurative change here however since there is no substitution of a group directly attached to the asymmetric carbon atom so that the compound which is derived from d-phenylamino-acetic acid is designated as d- (3-amino-actp-triphenylethyl alcohol, although it is lzvorotatory d-phenylaminoacetic acid being taken as the reference type. It is rather curious to note that a similar change of sign of rotat'ion also occurs when methyl I-mandelate is converted by magnesium phenyl bromide into triphenylethylene glycol (McKenzie and Wren 7oc.c i t . ) . The action of nitlrous acid on the ci-amino-alcohol gives an optic-ally inactive product owing t o semipinacolii? ic deamination the product being phenyldeoxybenzoin. The isomeric camphorsulphonates of the d- and I-amino-alcohols were prepared but the resolution of the 1.-alcohol by d-camphor-S-sulphonic acid gave a very slow and imperfect separation and the method was obviously unsatisfactory. We are a t present engaged in resolving by means of camphorsulphonic acids the r-amino-alcohols related to desylamine. The changes C<R -+ d- CH,Ph>C- and d- CH,Ph*CH(NH,)*CO,H NH HO are also described but the study of the action of nitrous acid on the amino-alcohols has not yet been completed.Arising from the present research the following points have been investigated. The ethyl ester hydrochloride of Z-phenyl-aminoacetic acid gives on hydrolysis with a slight excess of alcoholic potash a phenylaminoacetic acid with [.ID - 12.4" in hydrochloric acid solution whereas the pure Z-acid has [.ID - 157.8" in the same solvent. This result was anticipated since in phenylaminoacetic acid we have both a phenyl group and a migrational hydrogen atom in direct attachment to the asymmetric carbon atom. McKenzie and Walker (J. 1915 107 1685) have shown that catalytic racemisation occurs when I-phenylbromoacetic acid i AMINO-ACIDS INTO TERTIARY AMINO-ALCOHOLS. 287 acted on by ammonia. When the action was conducted in benzene solution that portion of the bromo-acid which had survived the attack of the ammonia had become largely racemised.It was thought possible that soniething of the same kind might occur when Z-phenylaminoacetic acid is acted on by nitrous acid; such, however is not the case. There was one other rather important point in connexion with tlhe study of the Walden inversion in the mandelic acid series which clearly required elucidation. E. Fischer and Weichhold (Ber. 1908 41 1286) acted on ethyl d-phenylaminoacetate with nitrous acid and obtained a change of sign of rotation the resulting ethyl mandelate being highly racemised and lanorotatory . On the other hand Marvel and Noyes ( J . Amer. Chern. Xoc. 1920 42, 2259) by acting on a solution of thc hydrochloride of ethyl 1-plieiiylaminoacctate in dilute sulphuric acid with nitrous acid, obtained an ethyl maidelate of the same sign of rotation as that of the original ethyl I-phenylaminoacetatc.We have repented the work both of Fischer and Weichliolcl and of Marvel and Noyes, and have found that a change of sign of rotation occurs when ethyl Z-phenylaminoacetate is acted on by nitrous acid if the solution, before the ester is extracted is kept sufficiently long after the addition of the nitrite. is therefore correct. 777) have shown that the change occurs the behaviour both of the ester and of the free acid towards nitrous acid is similar. The product obtained by the method of the American workers was contaminated with a nitrogenous com-pound (probably a diazo-compound).It seems that they had proceeded to isolate their ethyl mandelate too quickly after the addition of the sodium nitrite. The work on the action of nitrous acid on compounds containing Fischer and Weichhold’s conclusion, d-CHPh(NH,)*CO,Et + Z-CHPh( OH)*CO,Et, Since fi1cKenzie and Clough (J. 1909 95, d-CHPh( NH,)*CO,H -+ Z-CI-IPh( OH)*CO,H the grouping >?-?< is being continued. OH NH, E x P E R I M E N T A L. Semipinacolinic Dearnination of p-Amino-aa-diphenyl-n-propyl Alcohol .-Ethyl r-alanine hydrochloride (compare Curtius and Koch J . pr. Chem. 1888 [ii] 38 472; Barker and Skinner J . Amer. Chem. Xoc. 1924 46 403) was conveniently prepared as follows A current of dry hydrogen chloride was passed into a suspension of r-alanine (10 g .) in ethyl alcohol (200 c.c.) for 1 hour, and the mixture after boiling for 1 hour longer was evaporate 288 MCKENZIE AND WILLS CONVERSION OF to dryness a t the ordinary temperature under 15 mm. pressure. The resulting syrup solidified when kept over soda-lime in a vacuum. It crystallised from ethyl acetate in needles (15.5 g. ; 90% of the theoretical). Ethyl r-alanine hydrochloride (20 g.) was added gradually to the Grignard reagent prepared from bromobenzene (245 g.) and the mixture heated for 8 hours. The product was decomposed by ice ammonium chloride and ammonia. The residual solid was extracted with ether and added to the main ethereal solution. After removal of ether and diphenyl as usual, the product (20 g.) was dissolved in hydrochloric acid and pre-cipitated by ammonia This treatment was repeated and then the product was crystallised three times from aqueous alcohol.P- Amino- cccc-diphenyl -n-prop yl alcohol separates from aqueous alcohol in rhombohedra1 plates m. p. 101-5-102~5" is readily soluble in ethyl alcohol methyl alcohol benzene toluene carbon tetrachloride carbon disulphide acetone or ether and very spar-ingly soluble in water (Found C z 79.4; H = 7-7 ; N = 6.2. C1,HI,ON requires C = 79.3 ; H = 7.5 ; N = 6.2%). A trace added to concentrated sulphuric acid assumes a golden-yellow tint and the solution becomes a t first pink and quickly colourless. A solution of 4-5 g. of sodium nitrite in 20 C.C. of water was gradually added (9 hour) to a solution of 5 g. of the amino-alcohol in 150 C.C.of 25% acetic acid cooled at 0". A flocculent pre-cipitate was gradually deposited which was filtered off after 24 hours and crystallised from aqueous alcohol. The silky needles (m. p. 50-51") which separated amounted to 4 g. and were identified as methyldeoxybenzoin (Found C = 85.8 ; H = 7.0. Calc. C = 85.7; H = 6.7%). There was no depression of the melting point when this substance was mixed with a specimen of the methyldeoxybenzoin prepared by McKenzie and Roger (J., 1924 125 844) by the dehydration of methylhydrobenzoin with concentrated sulphuric acid. The coloration observed wit8h con-centrated sulphuric acid was identical with that described by McKenzie and Roger for their product. The identity of the sub-stance was also proved by its behaviour towards magnesium phenyl bromide when accp-triphenylpropyl alcohol m.p. 86-5-87-5" was isolated. The latter compound had already been prepared by Mlle LBvy (Bull. Xoc. chim. 1921 [iv] 29 878) and by McKenzie and Roger. Hydrochlorides of Ethyl d- and 1-Pheny1aminoncetntes.-The optic-ally active phenylaminoacetic acids were prepared by E. Fischer and Weichhold (Ber. 1908 41 1286) who resolved the formyl r-acid by cinchonine and quinine and then hg'drolysed the d-and I-formyl acids with hydrobromic acid. The resolution o AMINO-AC'TDS TSTO TERTTAR'I' AMINO-.iT,C'OHOT>S. 289 ./.-phen~-larninoacetic acid into its opt ic'ally nc t ivc components by means of Reychler's d-camphor- 3-sulphonic acid in aqueous solu-tion has been described by Betti and Xayer (Ber.1808 41 2071) and by Marvel and Noyes ( J . il?n,eY. Clzem. Soc. 1920 42 2259), the acids being einploj-etl in equiivolecular quantities. I n the inoclification described 1)y Ingersoll a i d Adanis ( J . Arne?.. Phem. Soc. 1922 44 2930) an c s c c ~ of tlic cauil>liorsulphouic acid was used. Since the latter aciti contains Tvater of carystallisation (Pope and Gibson ,J. 19l0 97 2211 ) ? antl sincc this may possibly be ])resent in varying ciiiantitj- T\ c employed this ruethotl. The optically pure 7-acicl was obtainetl by adding the rqiiisite amount of aiiiiiionia to the solntion of the salt o1)tainccl after several c~rystallisations. Thc cl-acid maj- t e preprcd from the acid mixture obtninecl from the first inother-iiqixor of the preceding rcsolutioii either by nieaiis of rl-r.-~rc~~ocaml~lior-,'3-siil~~~i~~iic acid or by formylating and then converting into the quinine salt.For the purification of the d-acid however it is more convenient to com-bine the crude acid with I-camphor-p-sulphonic acid and then to crystsllise until the caiii~~horsulplioiiate is homogeneous. Wc were enabled to conduct the resolution in thc latter iiianiicr owing t o the generous gift of c2 supply of I-csniphor from Sir William Pope. The I-camphorsulplioiiic acid. obtained from E-camphor by snlphon-ation in presence of acetic anhj-clride was crystallised from ethxl acetate and gave in aqueoiis solution 1 = 2 c = 3.8703 c ( ~ - 1-61', I-Camphorsulphonic ac-id combines with I-phenylarninoacetic acid to form a salt which is readily soliible in water from which i t separates in prisms.Its rotation n-as cleterininecl in aqueous solution : The diastereoisonieric salt. prepared froin d-camphorsulphonic acic1 and Lphenylaininoac4c acid has [.ID - 444)'i' (c' = 2.0883, I = 2) in aqueous solution (Hctti and Jfayer loc. cit.). This value enabled us to estimate the progress of the resolution of r-phenyl-amiiioacetic acid by (1- a lit1 Z-cninphorsulplioiiic acids. The polari-metric values for the I - a i d rI-pl~eiiS'1,2minoac.eiic acids obtained agreed with those of Fischer and Weichho1:l. Ethyl 1- P h e ~ ylarniiioacPtntc Ilylroch lor idc .-Fisclier and Weich -hold describe the conversion of I-pheii?.laminoacetic acid into the hydrochloride of its ethyl ester TI liicli they describe as dextro-rotatory [~(]fy f- S8.93" in aqucouc; so1::tjoii (21 = 5.021 d = 1.0087).For example a current of dry hydrogen chloride n-as 1)assecl into a mixture of 3.3 g. of l-phenyl-aminoacetic acid ant1 40 C . C . of etl1j.I alcohol for 1 hours. The solution was then boiled for a few minutes and filtered from a VOL. CXXVIT. L [.ID - 20.8". 1 = 2 G = 4.630 xD - 6*G4" [.ID - 51.7". Our results were (1 iffcrent 290 MCEENZIE AND WILLS CONVERSION OF small amount of solid. The alcohol was removed from the filtrate by gentle warming under diminished pressure the resulting acid dissolved in 20 C.C. of water and an equal volume of benzene added. The addition of the requisite amount of ammonia caused the separ-ation of the ester which was extracted with benzene.A current of dry hydrogen chloride was passed into the benzene solution for 15 minutes when the crystalline ester h~drochloride (4.2 g.) separ-ated. It had the following rotation i n aqiieous sollition 7 =1, Marvel and Noyes also obtained a laworotatory ester hydro-chloride with [.]* - 84.6" (concentration not quoted) from the I-amino-acid. The dextrorotation recorded by Pischer and Weich-hold is possibly due to a typographical error. Ethyl d-phenylapninoacetate hydrochloride was prepared in a similar manner from d-phenylaminoacetic acid 6.1 g. being obtained from 5.8 g. of d-acid. It was dextrorotatory in aqueous solution: Action of Magnesium Phenyl Bromide on the Hydrochlorides of Ethyl d- and 1-Pheny1aminoacetate.-The d-ester hydrochloride (6 g.; 1 mol.) was gradually added wit!hin 15 minutes to the Grignard reagent prepared from 52 g. of bromobenzene (12 mols.) 16.5 g. of magnesium and 350 C.C. of ether and the mixture heated for 49 hours. After decomposition of the product with ice and ammonium chloride and remaining over-night the ethereal layer was separated and the aqueous layer extracted with ether. The ether and the diphenyl were removed the latter by steam dis-tillation. The residual yellow solid (8 g.) was crystallised from ethyl alcohol until the product after drying over concentrated sulphuric acid in a vacuum until constant gave a value which remained unchanged on polarimetric examination after repeated cry stallisation. d- @-Amino- axp-triphenylethyE alcohol is somewhat sparingly s o h ble in ethyl alcohol and separates in colourless needles m.p. 129.5-130° whereas the r-isomeride (McKenzie and Barrow J. 1913, 103 1331) melts a t 154-5-155". A trace of it added to con-centrated sulphuric acid gives a pink coloration which quickly becomes orange-brown. It is readily soluble in chloroform ether, benzene or acetone (Found C = 83.1 ; H == 6.8. C,,HISON requires C = 83.0 ; H = 6.6%). The substance has the opposite sign of rotation to thc original ester hydrochloride bcing strongly lzevorotatory in chloroform : I = 2 c = 1.276 a:f - 6-19" [a]:? - 243". In benzene 1 = 2, The I-ester hydrochloride (5 g.) was acted on by magnesium c = 5.070 ED - 4-53' [RID - 89.3". 2 = 1 c = 5.070 ED $- 4-60' [.ID 90.7". c = 2.027 - 9-45" [CC]:" - 233" AMINO-AUIDS INTO TERTIARY AMINO-ALCOHOLS.291 phenyl bromide under conditions similar to those just described. Yield of crude amino-alcohol = 6 g. It was purified by crystallis-ation from ethyl alcohol, 1-p-Amino-aap-triphenylethyl alcohol has m. p. 129.5-130" (Found : N = 4.9. C,,H1,ON requires N = 4.8%). The compound is strongly dextrorotatory in chloroform 1 = 2 c = 1,304 ug5' + 6*34" + 243". I = 2 c = 1.304 $- 7*63" [~]:4;~ + 293". In bcnzene 1 = 2 c = 2.017 a:' + 9.44" [ a ] r + 234". Action of Nitroes Acid on d- (3 -Amino- a a p -triphen ylethyl Alcohol .-A solution of the d-amino-alcohol (0.68 g.) in 50 C.C. of dilute acetic acid was cooled in a freezing mixture of ice and salt and a solution of 0.5 g. of sodium nitrite in 5 C.C. of water added during 20 minutes.The solid (0.6 g.) which separated was crystallised from ethyl alcohol and 0.42 g. of phenyldeoxybenzoin needles m. p. 134-135" was obtained. It gave the characteristic emerald-green coloration with concentrated sulphuric acid. Its solution in chloro-form was optically inactive. Isomeric Camphorsulphonates of the d- and 1-Amino-Alcohol.s.-I-p-Amino-aap-triphenylefhanol d-camphorsulphonate prepared by combining the 1-amino-alcohol (1 mol.) with d-camphor-p-sulphonio acid (1 mol.) in ethyl-alcoholic solution separates in needles, m. p. 200-201" (decomp.) (Found S = 6.3. C,,H,,O,S requires S = 6.1%). In ethyl alcohol I = 2 c = 0.8568 a g + 2.08", [a]':' + 121"; I = 2 c = 04568 a:::* + 2*41" [a]:!&' + 143". The enantiomorphously related d-p-amino-aap-triphenylethanol 1-camphorsulphonate separates from ethyl aicohol in needles m.p. 200-2001" (decomp.) (Found N = 2.9. C,,H,,?N requires N = 2.7%). I n ethyl alcohol 1 = 2 c = 0.80 a:," - 1.91" [a]:' -119"; 1 = 2 c = 0.80, 1- p-Amino-aap-triphenylethanol I-camphorsulphonate separates from ethyl alcohol in needles m. p. 213.5-214.5" (decomp,). In ethyl alcohol 1 = 2 c = 0,4448 ag5' + 0*68" [a]:G' + 76"; I = 2, c = 0.4448 a::; + 0*76" [a];$ + 85". The enantiomorphously related d-P-amino-aap-triphenylethanol d-camphorsdphonate separates from ethyl alcohol in needles m. p. 213.5-2146" (decomp.) (Found S = 6.2. C1,H,,O,S requires S = 6.1%). In ethyl alcohol 1 = 2 c = 0.40 ag' - 0*61", [a]:"' - 76"; 1 = 2 c = 040 a:& - 0*72" [a]::;1 - 90". It is insoluble in most organic solvents moderately soluble in hot ethyl alcohol or water and sparingly soluble in these solvents a t the ordinary temperature.The concentrations employed in the determination of specific rotatory power were necessarily low so that the values for the specific rotatory power have little significance. Attempts were made to resolve the r-amino-alcohol by d-camphor-- 2-31' [a]iSI - 144". L 292 MCKENZIE AND WILLS CONVERSION OF p-sulphonic acid both in aqueous and ethyl-alcoholic solution. The progress of the resolution was however too slow to enable the method to be used as a preparative one for the d- and 1-amino-alcohols. Action of Magnesium Phenyl Bromide on a-Aminohydratropic Acid.-The amino-acid (6 g.; 1 mol.) prepared according to McKenzie and Clough (J.1912 101 390) was added in instal-ments (Q hour) to the Grignard reagent (12 11101s.) prepared from bromobenzene (71 g.) and the mixture was heated for lo+ hours. After decomposition with ice and ammonium chloride the remain-ing solid was extracted with ether and the solution added to the main ethereal solution. The ether and diphenyl were removed ; the resulting oil solidified. Yield 5 g. It was tlriturated with light petroleum and the solid then crystnllised from rectified spirit until pure. p- Amino- ma p -triphen yl- p-methyleth yl alcohol separates from re cti-fied spirit in rectangular plates is soluble in benzene and chloro-form and melts a t 113-114" (Pound C = 83.0; H = 7.2; N = 4.8. C2,H210N requires C = 83.1 ; H = 7.0; N = 4.6%).A trace of this compound added to concentrated sulphuric acid produces an orange coloration which changes quickly to a permanent crimson colour. Action of Magnesium Phenyl Bromide on r-Phenyla1anine.-No visible action took place when r-phenylalanine (3 g.; 1 mol.) was gradually added to the Grignard reagent (12 mols.) prepared from 34 g. of bromobenzene. The mixture was heated for 20 hours, and then the additive compound was decomposed with ice and am-monium chloride. The ethereal layer was withdrawn and then the ether and diphenyl were removed from it in the usual manner. The residual solid amounted to 4 g. which were crystallised from ethyl alcohol. The yield of pure amino-alcohol was 2-4 g. cor-responding in crystalline form and melting point with r-p-amino-aa-diphenyl- P-benzylethyl alcohol which was prepared by McKenzie and Richardson (J.1923 123 79) from the ethyl ester hydro-chloride of phenylalanine by the action of magnesium phenyl bromide. Action of Magnesium Phenyl Bromide on d- Phenyla1anine.-r-Phenylalanine was resolved by the alkaloidal method of E. Fischer and Schoeller (Annalen 1907 357 2) and d-phenylalanine was isolated with a specific rotation in aqueous solution of + 35.0" ( I = 2 c = 2.043 aD + 1.43") this value being identical with that quoted by Fischer and Schoeller. Three g. of this amino-acid were treated in the manner recorded in the above experiment, with the exception that the mixture was heated for 24 instead of Yield 2 g AMINO-ACIDS INTO TERTIARY AMINO-ALCOHOLS.293 20 hours. 4.2 G. of crude omino-alcohol were obtained and were crystallised froin ethyl alcohol several times until a steady value for the specific rotation was obtained in chloroform I = 2 c = 4-06" [CI],;:~;~ + 99.2". In benzene Z = 2 c r= 2.0512 E:!:" + 4-32', This compound has the same sign of rotation as the original amino-acid and since 110 configurative change can take place in the transformation of the amino-acid into the amino-alcohol it is accordifigly designated as d- ~-cr?,iiiio-ax-diphe?~~~- P-be?zxylethyl alcohol. It crystallises from ethyl alcohol in flat glassy needles m. p. 143-1Uo is soluble in acetone h i z z n e chloroform ether or toluene, but is insoluble in T.\.-ater (Found N = 4.6. C',lH,lON requires N = 4-60/,). A trace of the compound added to concentrated sulphuric acid produces an orange-brown coloration which quickly changes to pale pink and then gradually fades.This coloration is also observed with the corresponding r-amino-alcohol. Hydrolysis of Ethfyl 1-Yl~eiz~lnmiiioacetate Hydrochloride.-The Lester hydrochloride (1 g.) in 30 C.C. of ethyl-alcoholic potassium hydroxide (0.4487L\') remained for 24 hours a t the ordinary tem-perature and was then heated on the water-bath for 30 minutes. The alcohol was removed under diminished pressure the residue dissolved in water and the solution extracted with benzene to remove any trace of unchaiiged ester. The aqueous solution was then neutralised with hydrochloric acid and the amino-acid (0.5 g.) was separated. Its specific rotation was determined as follows : 0.3719 g.was dissolved in 3.35 C.C. of X-hydrochloric acid and 1.5 C.C. of water I = 0.5 p = i-193 d = 1.0286 xD - 0*4G" - 12-4". 3'11 c ester Iiydrochloridc was completely hydrolysed . Pa ~t icr l Den in i iiu t ion of I - Phe ii yln nzi 11 oncet ic d c id .-A solution of sodium nitrite (0.41 g.) in water ( 2 c.c.) was added drop by drop (30 minutes) to a solution of the I-amino-acid ( 2 g.) in X-sulphuric acid (30 c.c.) t L t 0". After 5 hours at 0" and 18 hours a t the ordinary tomperaturc the amiiio-acid (1.33 g.) as precipitated by the addition of aniiiioiiia in slight excess. 0.7438 G . was dissolved in 6.7 C.C. of n'-hydrochloric acid and 3 c.c. of vcater I = 1 p = 7.193 dJ' = 1.0286 x ~ - L1.36" '[RID - 133-5" whereas thc originzl I-acid had Thc ammoniacal solution from which the amino-acid had been scparated TT~C?S acidified n-ith dilute hydrochloric acid and extracted with cther.The I-esulting mmdelic acid (0.23 6.) was slightly tlestrorotatory in aqueous solution Z = 2 c = 1.055 clD + 0.13". Deuminalioiz. of Ethyl l-Pl~ciiylnmi.Iioacetate.-~~~ethocl of E. F'isclzer und TPeichhoZd. These authors acted on the d-ester with nitrous 2.0472 c(::'* + 3.48" [XI;:'* + 85.0" ; l = 2 c = 2.0472 a3:;' + [a]::' + 105-3"; Z = 2 c = 2.0312 MI,; + 5*03" [cx]~$ + 122.6". - 156' under similar conditions 294 CONVERSION OF AMINO-ACIDS INTO THRTIARY AMINO-ALCOHOLS. acid and obtained a lzevorotatory ethyl mandelate with about [.ID - 10" in acetone solution On following their directions, using ethyl Z-phenylaminoacetate (0.90 g .) in place of the d-ester, the resulting mandelic ester (0.325 g.) was dextrorotatory. A 10% solution gave ccD + 0.45" in a 0.5 dcm.-tube so that the value for the specific rotation is approximately + 9". The result of Fischer and Weichhold was thus confirmed. It should be stated that after the addition of the nitrite the solution remained over-night a t the ordinary temperature. The large amount of racemisation which accompanied this change is apparent when it is recalled that ethyl E-mandelate has according to Walden (2. physikal. Chem. 1895 17 705) [.ID - 90.6" in acetone solution. The dextrorotatory ester obtained above gave a negative result when tested for nitrogen. It was hydrolysed by 4.5 C.C. of aqueous potassium hydroxide (0.5755N) at the ordinary temperature for 18 hours.The solution was extracted with ether to remove any unsaponified ester and the rnandelic acid isolated from the aqueous solution as usual. The resulting acid (0.14 g.) was dextrorotatory in aqueous solut'ion E = 2 c = 0.622 uD + 0.27". Method of Marvel and Noyes. A solution of sodium nitrite (1.6 g.) in water (2.5 c.c.) was added drop by drop with constant stirring for 30 minutes to a solution of ethyl E-phenylaminoacetate hydrochloride (5.4 g.) in N-sulphuric acid (33 c.c.) the temperature being kept a t 0" throughout. A yellow oil separated. After 1 hour a t 0" and 2 hours a t the ordinary temperature the solution was extracted with ether the ethereal solution dried with anhydrous sodium sulphate and the ether expelled.The resulting oil was distilled and the portion (1.5 g.) b. p. 127-132"/18 mm. collected. I n acetone E = 2 c = 10 uD - 0.35'. The product obtained by Marvel and Noyes was also laevo-rotatory . When this oil was tested for nitrogen it gave a positive result. On hydrolysis under conditions similar to those described in the previous experiment it gave a mandelic acid (0.8 g . ) which was slightly dextrorotatory in aqueous solution I = 2 c = 3.109, This shows that the lzevorotation observed by Marvel and Noyes was not due to ethyl Z-mandelate but rather to the presence of an intermediate lzvorotatory diazo-compound which gradually passes into ethyl d-mandelate with lapse of time. That this view is probable appears from the result of the second experiment quoted below.The unattacked ethyl phenylaminoacetate hydrochloride was recovered from the aqueous solution remaining after the extraotion C(D $- 0.10" NEWBERY THE ACTION OF CAUSTIC ALKALI ETC. 295 of the deamination product with ether. The solution was made alkaline with ammonia extracted with benzene the benzene solution dried and dry hydrogen chloride passed in to precipitate t'he ester hydrochloride Yield 2 g. In aqueous solution 1 = 1 c = 5.07 aD - 469" [ a ] D - 90.5". The regenerated hydrochloride had thus practically the same rotation as the original. In a second experiment where 7 g. of the Lester hydrochloride were employed the conditions of deamination were exactly similar to those just described except that the solution after being kept a t 0" was allowed to remain for 39 instead of 2 hours. The resulting oil (2 g.) which gave a positive test for nitrogen mas in this case dextrorotatory in acetone solution E = 1 c = 10 tcD + 0.30". On hydrolysis with aqueous alkali under similar conditions to those adopted in the other experiments the resulting mandelic acid was dextrorotatory in aqueous solution I = 2 c = 3.32, The amino-ester hydrochloride was recovered as before. In The latter was filtered off. a D + 0.19". aqueous solution 1 = 1 c = 5.07 mD - 4-57" [.ID - 90.1". M7e desire to express our best thanks to the Department of Scientific and Industrial Research and to the Carnegie Trust for their assistance. We are also indebted to Dr. H. J. Plenderleith for his assistance in the deamination of p-amino-act-diphenyl-n-propyl alcohol. UNIVERSITY COLLEGE D UNDEE. UNIVERSITY OF ST. ANDREWS. [Received December 5th 1924.