20 REILLY THE RESOLUTION OF ASYMMETRIC IV.-The Resolution of Asyniwielric Qriinqueucclent NitqBogen Compounds. Part I. The Salts of d- and 1- Pheriyl benz ylmeth ylallylammonizim 11 ydroxide with d- and l-a-Rrornocamphor-~-sulphonic Acid. By JOSEPH REILLY. THE study of the structure of compounds containing an asymmetric nitrogen (or other) atom has to a large extent developed from the analogy between these compounds and corresponding carbon derivatives. I n general a similarity is observed between the two classes of compounds but in some particulars striking differences are observed. For example the optically active scids show greater * It is quite possible that the substance is formed by the action of duminium chloride on diphenylphthalide a reaction somewhat analogous to that described by Scholl and Seer (Annulen 1912 3W 111; Monulsh.: 1012,33 1) QTJIWQUEVALENT NITROGEN COMPOUNDS.PART I. 21 variation with regard to the facility with which they can be employed as agents for the resolution of compounds containing asymmetric nitrogen atoms than for the resolution of similar carbon compounds. With asymmetric quinquevalent nitrogen com-pounds such as the quaternary ammonium compounds and par-ticularly with the cyclic derivatives this difference is very marked. I n this respect it is also noticed t h a t it is much more difficult to resolve the cyclic ammonium coinpounds than other quinquevalent compounds in which the nitrogen does not form part of the ring. Taking the case of compounds containing a quinquevalent nitrogen atom attached to five different univalent radicles the following brief summary will show the different behaviour of the optically active acids with regard to their power to facilitate resolution (Reilly Proc.Camb. Phil. SOC. 1915 18 177). By fractionally crystallising the compound obtained by displacing the iodine in dl-phenylbenzylmethylallylammonium iodide by the d-or l-P-camphorsulphonate radicle Pope and Peachey (T. 1899 75, 1127) succeeded in resolvirjg the ammonium compound yet Wede-kind (Ber. 1899 32 517) had previously failed to obtain the two modifications by repeated crystallisation of the d-tartrate or d-camphoric acid derivative. Jones later failed t o separate the d- and Z-phenylmethylethylallylammonium d-P-camphorsulphonates even after repeated crystallisation from various solvents whilst he found that the d- or l-a-bromocamphorsulphonate of the inactive base could be readily resolved by fractional crystallisation.It might therefore be considered t h a t the stronger the optically active acid employed to effect the resolution the more easily would such a result be achieved. This however is not dways the case, as the following account shows. The base mentioned above which Pope and Peachey resolved by the aid of camphorsulphonic acid, is much more difficult t o resolve if the agent employed is bromo-camphorsulphonic acid. Recently Komatsu (Mern. Coll. Sci. Kyoto Zmp. Uniu. 1915 1 123) subjected to repeated crystallisa-tion from acetone and ethyl acetate the bromocamphorsulphonates of this base b u t failed t o bring about a separation.Wedekind (Zeitsclb. physikal. Chem. 1903 45 235) aware of the previous work of Pope and his pupils attempted to resolve p-tolylbenzyl-methylallylammonium iodide into its two antipodes by the aid of d-6-camphorsulphonic acid but failed yet Jones (T. 1908 93, 1790; compare also Homer Proc Camb. Phil. Soc. 1907 14 196) succeeded by crystallisation of the hydrogen d-tartrate. Several other instances have been recorded in which resolution is brought about by the aid of the compara.tively weak acids d- or Z-tartaric or d- or Z-camphoric where the much stronger acids such as d- o 22 REILLY THE RESOLUTION OF ASYMMETRIC Z-j3-camphorsulphonic or h or Z-a-bromocamphor-n-sulphonic have been unsuccessful. Resolution may fail in some instances owing to the formation of a partly racemic compound by the crystallisation of which the isomerides cannot be separated.This explanation may also be suggested to account for the difficulty in separating into d- and Z-modifications certain cyclic ammonium compounds such as the a- and &substituted pyridinium compounds or the tetrahydro-quinolinium derivatives of the type CSH,,:NR/R/’X. These com-pounds should be capable of existing as optical antipodes since their molecules have no plane of symmetry (Jones T. 1903 83, 1400). Hydrolytic dissociation of the salts it is thought in other cases will account for the results obtained. Identical solubilities of the two derivatives formed from the d- and Z-base and the optic-ally active acid could be offered as another explanation but this suggestion is not sufficient.This selective action of optically active acids in their power to bring about the resolution of extern-ally compensated compounds noticeable much more frequently in the case of nitrogen than of carbon derivatives would suggest that the non-resolution in particular cases is due to some other causes than to those indicated The present work was undertaken in order to investigate more fully this selective action of the optically active acids. The original quaternary iodide resolved by Pope and Peachey is readily obtained in the d- and Z-forms by fractional crystallisation of the d- and Z-B-camphorsulphonates of the inactive ammonium base and formation anew of the iodides by the action of aqueous potassium iodide.Substituting a-bromocamphor-n-sulphonic acid for P-camphorsulphonic acid in the above experiments on fractional crystallisation of the a-bromocamphor-n-sulphonates a separation of the two isomerides is not so readily obtained. I n order t o ascer-tain if racemisation sufficiently accounts for the different behaviour of the P-camphorsulphonates and the halogenated compounds an attempt was made to obtain the pure d- and Z-compounds of the base with d- and Z-a-bromocamphor-n-sulphonic acid by an indirect method and to investigate the compounds formed. By the reaction of anhydrous silver a-bromocamphor-n-sulphonate with d-phenylbenzylmethylallylammonium iodide in dry acetone under certain conditions the compound d-phenylbenzylmethylallyl-ammonium d-a-bromocamphor-?r-sulphonate was obtained pure.By combination of the d- and Z-iodides with silver d- and Z-a-bromo-camphor-n-sulphonates under conditions stated later the following compounds were obtained dBdA dBZA ZBZA and IBdA. These compounds are of the same order of stability as the correspondin QUINQUEVALENT NITROGEN COMPOUNDS. PART I. 23 P-camphorsulphonates and show very little tendency t o racemise. I n alcohol and chloroform solution small racemisation effects have been observed. The further study of these racemisation effects is much facilitated both by the very high specific rotations of the bromocamphorsulphonates and by their solubility in both water and non-aqueous solvents. EXPERIMENTAL. R esolut i o n of dl-Ph e l ~ y l b emylm et hylallylammonium Iodide.The details of the method employed in the resolution of the iodide differ somewhat from those previously employed. The quaternary iodide in dry acetone was heated with a molecular pro-portion of silver d-P-camphorsulphonate for several hours. The solution was filtered and the insoluble portion extracted three or four times with boiling acetone. The acetone from the extract was distilled off and on cooling the syrupy residue quickly crystal-lised forming a hard mass which consisted of a mixture of the d-P-camphorsulphonates of the d- and Z-bases. The acetone filtrate from the sulphonates on evaporation on the water-bath gave a partly viscous residue which was extracted with dry ethyl acetate and most of tohe solvent evaporated off. It was t'hen placed in a vacuum desiccator when the main crop of crystals was obtained.The two fractions together gave a nearly theoretical yield. The separation of the isomerides and the preparation of the pure d-iodide were carried out according to Harvey's modification of the method devised by Pope and his co-workers (T. 1905 87 1482). The following determinations of the rotatory power were made with a chloroform solution of the iodide containing 0.3065 gram made up to 30 C.C. a t 1 7 O in a 4-dcm. tube: Hggree,,. Hgyellow. Na elluw-a ............ +2-83" $- 2.44" + 2.30" [MI ......... 253 218 205 [u) ......... 69.2 59.7 56.3 The rotation dispersion ratio for Hggree,l/NaJ-allo,v = 1.230 and for Hgyellow/ The value [a]= 56.7O obtained by Harvey (loc. cit.) is in agree-Nayellow= 1.061.ment with these results. Preparation of d-PhenylbenzyZmethylallylammonium d-a-Bromo-cam phor-?r-sulph onat e. Silver a-bromocamphor-n-sulphonate (70 grams dried at looo) was digested with dry acetone (300 c.c.) under reflux and a mole-cular proportion of the pure d-quaternary iodide added in smal 24 REILLY THE RESOLUTION OF ASYMMETRIC quantities a t a time. The mixture was boiled for about two hours and then filtered. To the filtrate were added washings with acetone of the insoluble silver iodide. The greater portion of the acetone was distilled off and the residue set aside overnight when it set to a crystalline mass which was extracted in a Soxhlet apparatus xrith dry ethyl acetate. On cooling the ethyl acetate solution deposited the sulphonate in colourless crystals which were collected.The filtrate on the addition of light petroleum (b. p. 40-60°) gave a further deposit of crystals. The substance was subsequently crystallised twice more from dry ethyl acetate. It melted and decomposed a t 155-159O. It was very readily soluble in alcohol or chloroform readily so in water or acetone sparinglyso in ethyl acetate and insoluble in light petroleum. It was dried a t looo in a vacuum until constant in weight before being analysed : 0.1174 gave 0.2550 CO and 0.0692 H,O. 0.1425 , 0.0487 AgBr and 0*0585 BaS04. Br=14.54; C=59*23; H=6.55. S = 5.63. C,,H,,O,NBrS requires C = 59-13 ; H= 6-21 ; Br = 14.59 ; S = 5-85 per cent. The compound was again crystallised from the same solvent, An aqueous solution containing 0.1655 gram in 30 C.C.a t 14O when the final determinations of the rotatory power were made. in a 4-dcm. tube gave: Hayreen. Hg e ~ l c w . Nayc~low. a ............ +2*23" + 1-91" + 1.80" [a] ......... 101.1 86.5 81.5 [MI ...... 554 474 44 7 The rotation dispersion ratio for Hgg,ee,,/Nayellow= 1.239 and for Hgyellow/ Na ellow = 1.06 1. The corresponding values for the ammonium salt of the acid are 1.243 and 1.059 (Pope' and Read T. 1910 97 2201). Taking the molecular rotation of the d-bromocamphorsulphonic acid ion as [MIL 279O (Pope and Read Zoc. cit.) this gives the molecular rotatory power of the d-phenylbenzylmethylallylammon-ium ion [Nln + 168O agreeing well with t'he value [MI +166*4O obtained by Pope and Harvey (T.1901 79 828) from the study of the d-P-camphorsulphonate derivatives. The quaternary iodide was recovered by the addition of an aqueous solution of pot'assium iodide and it gave in chloroform solution [a] + 56*1°. The bromocamphorsulphonate in aqueous solution showed only a very slight tendency towards racemisation and after two months in a thermostat a t 20° the substance gave [ a ] 79.9O QUINQUEVALENT NITROGEN COMPOUNDS. PART 1. 25 The compound separated from water in large well-defined, anhydrous colourless needle-shaped crystals in radiant bunches. The specific rotatory power however had fallen to [aID 79'1O. On warming an aqueous solution of the compound it became cloudy, and after some time oily globules consisting of the tertiary base, appeared.With prolonged boiling the solution gradually lost its activity owing not to optical inversion but probably t o decomposi-tion. The inactive ammonium iodide was also caused to react with silver d-a-bromocamphor-.rr-sulphonate in dry acetone solution under similar conditions to those employed. in the above experiments. Working with the inactive iodide there was greater difficulty in obtaining a crystalline derivative. On filtering from the silver iodide and concentrating the solution the product tended t o separate in a partly crystalline mass. I n this form it was macer-ated with dry acetone and the mixture boiled and filtered the filtrate afterwards being evaporated to small bulk. The solid matter which separated was spread on a porous tile and dried in a vacuum desiccator over sulphuric acid until quite hard and recrys-tallised several times from dry acetone.It was then obtained in a crystalline form melting indefinitely from 145O to 1 5 0 O . After re-peated crystallisation from acetone the specific rotation determined in aqueous solution altered only from [a]Eo 51.3O t o [a] 53*0° a t which figure it remained constant. The latter figure gives [MI 291° a value sufficiently near that of ammonium bromo-camphorsulphonate ([MI 279O) t o indicate that very little resolu-tion has been accomplished under these conditions. A series of fractional crystallisations from ethyl acetate showed indications of resolution but the pure active compound was not easily obtained. The acid used in the above experiments was prepared from d-a-bromocamphor.This compound was converted into the sulphonic acid derivative by fuming sulphuric acid and then purified by means of its ammonium salt. The latter salt was decomposed by baryta solution and the silver salt obtained by the action of silver hydroxide as described by Pope and Read (T., 1910 97 2200). d-Ph enyl b en z y lm e t h ylally lamm onizi m 1-a-Bro m oca m p h OFT-sulphonn t e. The d-iodide and silver 1-a-bromocamphor-.rr-sulphonate do not react readily in very dry organic solvents whilst if a little water is added the sulphonate obtained is partly racemised. The follow-ing procedure gave satisfactory results. One molecular proportion of anhydrous silver a-bromocamphor-.rr-sulphonate was added t 26 REILLY THE RESOLUTION OF ASYMMETRIC dry ethyl acetate boiling under reflux ; one molecular proportion of the d-base was added in small quantities a t a time and the mixtlure boiled for about two hours.After filtration the soluble portion together with two or three washings with ethyl acetate of the silver iodide was evaporated to dryness and the viscous residue left over sulphuric acid in a desiccator until the whole mass crystal-lised. When very hard this mass was pulverised and extracted in a Soxhlet apparatus with ethyl acetate. On evaporation t o a small bulk and adding a little light petroleum the required com-pound was precipitated and it became crystalline very quickly. It melted and decomposed a t 146-148O. The compound had [a] -23O (in aqueous solution) but on three crystallisations from ethyl acetate this number changed to [aID -20-56O.This value on subsequent crystallisation from non-aqueous solvents did n o t alter. The substance crystallised from water in large needles and clusters of radiant crystals. Although no exact measurements have yet been made of the solubility of this compound it appears however from general examination to be more readily soluble than d-phenylbenzylmethyl-allylammonium d-a-bromocamphor-?r-sulp'honate in ethyl acetate. It would be expected therefore that ethyl acetate might be a suit-able solvent to bring about the separation of the isomerides from an inactive mixture of the two by fractional crystallisation but, as already shown this is not the case. By the interaction of silver d-a-bromocamphor-71.-sulphonate with partly resolved &iodide the pure d-a-bromocamphor-r-sulphonate of the d-base was obtained after several crystallisations from dry ethyl acetate.The following determinations of the rotatory power were made with an aqueous solution containing 0.1240 ++ gram in 30 C.C. a t 20° in a 4-dcm. tube: Hggreeii. Hgyellow. Nayellow. a ............ - 0.45" - 0.37" - 0.34" [a] ......... 27-22 22.4 20.5 [MI ......... 149 123 113 The rotation dispersion ratio for Hggrecn/NZtye\low = 1.323 and for Hgyellow/ Nayellow= 1.088. The number for the molecular rotation compares well with the molecular rotatory power obtained by calculation from the mole-cular rotatory power of d-phenylbenzylmethylallylammonium d-a-bromocamphor-.Ir-sulphonate. The molecular rotation of the basic ion is 166.4O.Pope and Peachey have shown (T. 1899 75 1084) that the molecular rotatory powers of the basic and acidic ions may be * Dried at 100" QUINQUEVALENT NITROGEN COMPOUNDS. PART I. 27 calculated from the molecular rotatory power of the salts dBdA, dBlA as follows provided the determinations are made in dilute solution : [MI of dBdA + [MI of dBlA= twice [MI of dB ion. [MI of dBdA- [MI of dBZA= twice [MI of dA ion. Applying these formulae toi the values found above we obtain the following molecular rotations for sodium light : dB ion 167O. d A ion 279’6O. The Z-a-bromocamphor-7r-sulphonate was prepared in a similar manner to the d-compound starting from 1-camphor as described by Pope and Kipping (T. 1893 63 576).l-Phenylbenzylmethylallylammonium l-a-Bromocamphor-7;-sulphona.te. The solvent from the mother liquor obtained from the fractional crystallisation of the d-P-camphorsulphonates was distilled off and the pure Z-p’heenylbenzylmethylallylammonium Z-P-camphorsulphon-ate prepared in the manner described by Harvey (T. 1905 87, 1482) using acetone as solvent. The Z-iodide was obtained by adding a concentrated solution of potassium iodide to an aqueous solution of the camphorsulphonate and subsequent treatment as in the case of the d-compound. A determination of the rotatory constants with a chloroform solution containing 0.1215 gram in 30 C.C. of solution at 15O in a 4-dcm. tube gave: H g g r e m . Hgyellow. Nayellom-a. ........... - 1.12” - 0.97” -0.91” [u] .........69.1 59.9 56.2 [MI ......... 252 219 205 The rotation dispersion ratio for Hggraen/N&yellow = 1.230 and for Hgyellouv/ Nayellow= 1-066. The figure for the rotatory constant with sodium light is in close agreement with the value [a] -56-4O obtained by Harvey (Zoc. The method of preparation and purification of Z-phenylbenzyl-methylallylammonium Z-a-bromocamphor-.lr-sulphonate is similar t o that used in the case of its enantiomorphously related isorneride. After one crystallisation the substance had [a]= - 7 7 * 3 O and sub-sequent crystallisation altered this to the constant value [alD - 81.2O. The following determinations of the rotation con-stank of a purified sample were made: cit .) 2 8 ASYMMETRIC QUINQUEVALENT NITROGEN COMPOUNDS. 0.1154 gram in 30 C.C.of aqueous solution a t 20° in a 4-dcm. tube gave: Hggreen- Hgyellow- Nayellow. a ............ - 1.54" - 1.32" - 1-25' [u] ......... 100.1 85.8 81.2 [MI ......... 548 470 445 The rotation dispersion ratio for Hgg,ee,,/Na.,eil.w = 1.232 and for HgyellOw/ The molecular rotatory power of the l-phenylbenzylmethylallyl-ammonium ion is [MI -166O a value agreeing well within experi-mental error with that obtained for the d-base. NaJ7ello,,. = 1 * 0 5 6. l-Ph enyl b e n z ylm e t h ylallylmn monium d-a-BTomocaniph or-7r-sulphonat e. This compound was also prepared in a similar manner to its isomeride and had similar properties. It melted a t 147-149O. The rotatory constants were determined in aqueous solution ; 0.2360 gram made up to 30 C.C. a t 1 7 O in a 4-dcm. tube gave: Hggreet Hgyellow. N%allow. a. ........... + 0-85" + 0.70" + 0.64" [Q] ......... 27.0 22.2 20.3 [MI ...... 148 122 111 The rotation dispersion ratio for Hgg,een/Nayellow = 1.328 and for Hg,ellc,,r/ Nayellow = 1.094. The molecular rotatory power of this salt ([MID +11l0) gives a value which agrees within experimental errors with t h a t assigned to its enantiomorphously related isomeride and also with the value obtained by calculation from d-phenylbenzylmethylallylammonium d-a-bromocamphor-.rr-sulphonate. The recovered quaternary iodide had [aID -56*5O. There is no immediate prospect of continuing this work or the comparative study of the corresponding derivatives of the optically active acids with the cyclic ammonium compounds owing to the undertaking of other duties. For this reason the work is pub-lished in its present form. I desire to acknowledge my indebtedness t o Prof. Pope for his suggestion of this work and for his kind interest during the course of the experiments. THE CHEMICAL LABORATORY, THE UNIVERSITY, CAM RRIDGE. [ I?ccci~ecl A'occiriher 161h 1916