: : CHEM LIB. WIJ 341 49 V279 B 397740 The Action of Molecular Silver, of Silver Sulfate and Chloride, and of Sulfuric Acid upon Halogen- ated Derivatives of Triphenyl- carbinolchloride. 1 (A CONTRIBUTION TO THE THEORY CONCERNING THE STRUCTURE OF THE COLORED TRIPHENYLMETHANE DERIVATIVES.) DONALD D. VAN SLYKE. 1 } 179 1 ! : The Action of Molecular Silver, of Silver Sulfate and Chloride, and of Sulfuric Acid upon Halogen- ated Derivatives of Tri- phenylcarbinolchloride Chemical Library Q II 341 H9 V279 (A CONTRIBUTION TO THE THEORY CONCERNING THE STRUCTURE OF THE COLORED TRIPHENYLMETHANE DERIVATIVES.) A THESIS SUBMITTED TO THE FACULTY OF THE DEPARTMENT OF LITERATURE, SCIENCE, AND THE ARTS OF THE UNIVERSITY OF MICHIGAN IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY IN JUNE, 1997. BY DONALD D. VAN SLYKE. 14 B12 A Pus The Action of Molecular Silver, of Silver Sulfate and Chloride, and of Sulfuric Acid upon Halogenated Derivatives of Tri- phenylcarbinolchloride CONTENTS: 1. Introduction. 2. Synthesis of Materials. 3. Action of Molecular Silver on Ortho- and Para-halogenated Derivatives of Triphenylcarbinol Chloride. 4. Action of Silver Chloride on Orthobrominated Derivatives of Triphenylcarbinol Chloride in Sulfur Dioxide. 5. Action of Silver Sulfate on Ortho- and Para-halo- genated Derivatives of Triphenylcarbinol Chloride; Inhibitive Effect of Sulfuric Acid upon the Action of Silver Sulfate; Structure of the Sulfates. 6. Summary. 1. Introduction. In previous papers bearing on the constitution of triphenylmethyl and its many analogs the following facts of importance, among others, were brought out: (1) The triarylmethyls are capable of existing in two modifications the colored and the colorless;2 the colored form is not limited to the solid phase alone, but as has been shown by J. Schmidlin,³ the two modifications exist in equilibrium with each other even when in solution. (2) In the colored modification one phenyl group has become apparently changed in its function, for in it the para position is extremely reactive and mobil, in strong contrast to the great stability of the three phenyl nuclei in triphenylmethane itself. (3) A similarly striking change in the functions of one phenyl nucleus, simultaneously with the produc- tion of color and of salt-like properties in the compound itself, has been shown to take place when the colorless triphenylcarbinol halides are dis- solved in sulfur dioxide, or when the triphenylcarbinols-always color- less are changed to the corresponding carbinol sulfates, always colored.5 4 The theoretical explanation which has been advanced in order to account for these newly acquired functions in the phenyl nucleus consists in attrib- uting to that nucleus a tautomeric change from the benzoid to the para- quinoid state. With such a change the para-carbon atom would become in its character more like that in aliphatic than in aromatic compounds. Such an explanation accounts satisfactorily for the tremendous difference as regards the stability of the bromine in the para-brominated triphenyl- carbinol halides according to whether we deal with the colorless solutions of the latter in benzene, chloroform, ether, etc., or with the colored solu- tions of the same compound in liquid sulfur dioxide. The following for- mulas illustrate the view-point. ¹ This paper was published in the Journal of the American Chemical Society, Vol. 33, p. 531, as Paper XX "On Triphenylmethyl," by M. Gomberg and D. D. Van Slyke. 2 Ber., 34, 2729 (1901); 35, 2406 (1902); 37, 2036 (1904); 40, 1881 (1907). 3 Ibid., 41, 2471 (1908). 4 Ibid., 39, 3274 (1906). 5 Ibid., 40, 1847 (1907); 42, 406 (1909). 4 Benzenoid. Quinoid. H Triphenylmethyl...... •[(CgHg)3C]n (C6H5)2=C= CH.Br p-Bromotriphenylcarbinol chloride (CH), c< (CH),=C= C1 схо Br C1 H Triphenylcarbinol sulfate..... (C6H5)2=C= SO4 The hypothesis as regards the existence of the two tautomeric forms of the triphenylmethane compounds has been since extended to the study of the xanthone and thioxanthone¹ derivatives. It was shown that there exists the closest analogy between these two classes of compounds on the one hand and the corresponding triphenylmethane compounds on the other, only that the former are far more prone towards tautomerization than the latter. It was further shown that the tautomeric colored salts obtainable from the xanthones and the thioxanthones must be considered, not as oxonium salts, but as quinocarbonium salts, just like their parent substances, the corresponding triphenylmethane derivatives. The question now arises: Does ortho-quinoidation occur simultane- ously with, or instead of, the para-quinoidation in the above-mentioned instances of tautomerization. If it does, then the halogen placed in the ortho position should appear fully as reactive as it was found to be in the para position in the tautomerized products. On the other hand, if the ortho-halogen be found wholly inert then the conclusion appears. justified as regards the non-occurrence of the orthc-quinoidation. This paper presents the results obtained in the attempt to answer the above question. The relative influence of chlorine and bromine in inducing tau- tomerization has also been studied, and there are also appended some fur- ther data in regard to the formation and behavior of the much-discussed triarylcarbinol sulfates. 2. Synthesis of Materials. o-Chlorotriphenylcarbinol Chloride, (CH)2 === C 4 C₂H₁C1(0) C1 o-Chlorotriphenylcarbinol, (CH)₂CH¸CI.C.OH.-This product was synthesized by the action of ortho-chlorobenzoic ester on magnesium bromobenzene. 2 · The ester was prepared from ortho-toluidine. The latter was diazo- tized according to Erdmann, forming ortho-chlorotoluene. From the chlorotoluene, ortho-chlorobenzoic acid was prepared by oxidation with potassium permanganate.³ For esterification, 1 molecule of dry sodium ¹ Gomberg and Cone, Ann., 370, 142 (1909); 376, 183 (1910). 2 Ann., 272, 145 (1893). 3 Ibid., 276, 55 (1893). 5 ortho-chlorobenzoate was heated for two hours at 160° with 1.1 molecules of dimethyl sulfate. The sodium methyl sulfate formed was dissolved in water, and the ester extracted with ether, dried over sodium sulfate, and distilled. The fraction boiling at 228-236° was equal to 80 per cent. of the theoretical amount. For the Grignard reaction 4.1 g. (1 mol.) of magnesium were dissolved in an ethereal solution of 30 g. (1.1 mol.) of bromobenzene, and 12.5 g. (0.4 mol.) of ester added in ethereal solution. After two hours' boiling, the magnesium compound was decomposed with ice and hydrochloric acid. The ethereal extract was freed from bromǝbenzene by steaming, and the residue taken up in hot ligroin, boiled over charcoal and calcium chloride, then concentrated and allowed to crystallize. 11.8 g. of white carbinol were obtained melting at 91°. The pure recrystallized product melts at 95°. 15 Calculated for C₁9H1C1.O: Found: C, 77:40; H, 5.13. C, 77.48; H, 5.60. o-Chlorotriphenylcarbinol Chloride.-11.8 g. of carbinol were dissolved in benzene, and the solution over calcium chloride saturated with hydro- chloric acid. After standing some hours, the solution was concentrated, and on cooling 10 g. of white crystallin carbinol chloride, melting at 134-135°, crystallized. Melting point of pure product, 136°. Calculated for C₁9H₁₁Cl2: Found: Hydrolyzable Cl, 11.33. Hydrolyzable Cl, 11.16. 6 4 5/2 1 o-Chlorotriphenylethoxymethane, CICH(CH)2C.OC₂H.-To a solu- tion of 0.2 gram of sodium in absolute alcohol 2.3 g. carbinol chloride were added, and the solution boiled with reflux condenser for an hour. The sodium chloride was filtered out, and the solution concentrated. The product failed to crystallize in a freezing mixture or on standing for some time. After several weeks, crystallization occurred spontaneously, and 1.4 g. of a white product were obtained, melting at 77°. Easily soluble in ether, ligroin, alcohol, and benzene. Calculated for C₂H₁9C10: Found: 21 C, 78.12; H, 5.94. C, 78.13; H, 6.13. o-Chlorotriphenylanilidomethane, CIC¸H(CHÁ)½C.NHCH5. -3 g. car- binol chloride were dissolved in about 20 cc. benzene and 2.2 g. redistilled aniline added. The anilide is not very soluble in benzene and it separated partially from solution along with the precipitated aniline hydrochloride. The anilide was extracted from the precipitate with hot benzene, and the benzene solution concentrated at room temperature. The product was obtained in 90 per cent. yield in the form of large, white crystals melting at 121°. 6 Calculated for C25H20CIN: Found: N, 3.80. N, 4.11. Unsymmetrical o-Chlorotetraphenylethane, (ClС¸H₁)(C¸H5)2C -CH₂CH5. This compound was prepared by the method of Gomberg and Cone.¹ 3 g. of carbinol chloride in dry benzene solution were added to an ethereal solution of twice the theoretical amount of benzyl-magnesium chloride. The condensation product separated out at once. The mix- ture was heated one-half hour on the water bath, then decomposed with dilute hydrochloric acid and ice. The ethereal extract, dried and con- centrated, gave a product melting at 165°. Recrystallization from ligroin raised the melting point only to 165.5°. The product is most readily soluble in benzene, easily soluble in ether and warm ligroin, somewhat less so in alcohol. Calculated for C26H1C1: Found: C, 84.64; H, 5.74. C, 84.66; H, 6.08. CH,Br(o) C1 o-Bromotriphenylcarbinol Chloride, (CH)2C This was prepared by the action of ortho-bromobenzoic ester on mag- nesium bromobenzene. The ortho-bromobenzoic acid was prepared by diazotizing anthranilic acid in dilute sulfuric, and siphoning the diazonium solution into a mechan- ically agitated cuprous bromide suspension." This gave us better yields (82 per cent. of the calculated) than diazotizing in the presence of cu- prous bromide. A single recrystallization from water gave a pure product, melting at 150°. The methyl ester was obtained by heating the dry sodium salt with methyl sulfate at 160° for two and a half hours, and isolating as in the case of the ortho-chlor ester. The yield was 87 per cent. of ester boiling at 245°-250°. For synthesis of the carbincl chloride the ester was dissolved in abso- lute ether and added to a slight excess of phenyl magnesium bromide. After several hours' boiling, the product, which had separated as a gum, was decomposed with ice and dilute hydrochloric acid, and the resulting oily carbinol steamed to free it from bromobenzene. The carbinol was then dissolved in benzene, the solution boiled over charcoal and calcium chloride, and after filtration, saturated with dry hydrochloric acid in presence of fresh calcium chloride. On driving off the benzene and recrys- tallizing from ligroin, the carbinol chloride was obtained, the yield being 54 per cent. of the calculated. Ber., 39, 1461 (1906). 2 Graebe, Ann., 276, 57 (1893). 7 19 Calculated for C₁9H₁₁BrCl: Found: Hydrolyzable C1, 9.92. Hydrolyzable C1, 9.82. o-Bromotriphenylcarbinol, (BrCH₁) (CH)₂C.OH.-2 g. of carbinol chloride were boiled in acetic acid solution. The carbinol was then pre- cipitated by the addition of water, and taken up with ether. The ethe- real extract was shaken with fresh portions of water until all the acetic acid was washed away. Concentration of the dried ethereal solution yielded 1.2 g. of pure white crystals. Recrystallized from ligroin the carbinol melts at 104°. It is easily soluble in benzene and ether, some- what less so in alcohol and petroleum ether. Calculated for C₁9H₁5BгO: Found: 19 -15· o-Bromotriphenylanilidomethane, Br, 23.58. Br, 23.18. (BrCH) (CH)2C.NHCH5.-1.2 g. of aniline were added to 2 g. of carbinol chloride in benzene solution. After standing for two days, the solution was filtered from aniline hydro- chloride, and concentrated by evaporation at room temperature. The product separated in pure white crystals, which were washed with ligroin, and dried in vacuo. Yield quantitative, 2.3 g. The anilide melts at 126°, is fairly soluble in hot petroleum ether and alcohol, slightly in cold; easily soluble in benzene. Calculated for C25H20BrN: N, 3.38. Found: N, 3.63. 6 Unsymmetrical o-Bromotetraphenylethane, (BrC¸H₁)(C¸H5)½С.CH₂С¸Н¸·- 2 g. of carbinol chloride dissolved in absolute ether were added to twice the theoretical amount of magnesium benzyl chloride. The mixture was boiled an hour, then decomposed with ice and dilute hydrochloric acid. The ethereal extract was concentrated and steamed, then taken up in petroleum ether, and dried over calcium chloride. From the concen- trated solution the product separated in crystals, melting at 153°. o-Bromotriphenylethoxymethane, (BrCH) (CH5)2C.OC₂H5-2 g. of car- binol chloride were dissolved in absolute alcohol, and added to an alco- holic solution of a slight excess of sodium. The mixture was boiled for an hour, then cooled and diluted with water. The ethoxy compound was extracted with ether, the ethereal solution dried and concentrated. The product refused to crystallize from ether or ligroin, but finally was obtained from alcohol in pure white crystals after several weeks' standing. Melting point 69-70°. o-Bromodi-p-chlorotriphenylcarbinol Chloride, 6 (Þ)-CIC¸H4 (p)-CIC,H₁ -4 CH₁Br(0) Cl -The product was prepared by the action of p-chloromagnesiumiodo- 8 benzene on ortho-bromobenzoic ester. The p-chloroiodobenzene required for the synthesis was prepared by diazotizing p-chloroaniline.¹ 1.1 molecules of magnesium were dissolved in an ethereal solution of one molecule of p-chloroiodobenzene and 0.9 molecule of ortho-bromobenzoic methyl ester added. After five hours' boiling the solution was decom- posed as usual, and the dark, impure carbinol steamed, and then boiled in benzene solution with calcium chloride and charcoal. The filtered benzene solution was saturated with dry hydrochloric acid and, after several hours' standing over calcium chloride, concentrated in vacuo. Hot ligroin was added to the residue, and the entire mass crystallized while still warm. Like other derivatives of o-bromodi-p-chlorotriphenyl- methane, it showed exceptional crystallizability. A 74 per cent. yield was obtained. The recrystallized product melted at 165°. The car- binol chloride is difficultly soluble in ligroin, soluble in chloroform and ether, very soluble in benzene. 12 1 Calculated for C19H₁₂Cl,Br: Found: Hydrolyzable C1, 8.31. Hydrolyzable C1, 8.20. o-Bromodi-p-chlorotriphenylcarbinol, (BrCH₁) (CICH₁)₂C.OH.-3.4 g. of carbinol chloride were dissolved in 20 cc. of glacial acetic acid and boiled. The solution of the carbinol was poured into 200 cc. of water and the pre- cipitated suspension extracted with ether. The extract was washed free of acid, dried over calcium chloride and concentrated. 2.6 g. of crystal- line white carbinol were obtained, melting at 107°. The melting point was not changed by recrystallization from alcohol and ligroin successively. The substance is easily soluble in benzene and ether, fairly soluble in alcohol and ligroin. 13 Calculated for C1,H,Cl₂BrO: Cl2, 17.38; Br, 19.59. Found: Cl2, 17.44; Br, 18.80. 6 2 o-Bromodi-p-chlorotriphenylethoxymethane, (BrC¸H₁)(C¸H¸Cl),C.OC₂H5.- 2 g. carbinol chloride were dissolved in absolute alcohol and added to a solution of 0.14 g. (1.2 molecules) sodium in absolute alcohol. The solu- tion was boiled for an hour, cooled, and poured into water. The sus- pended ethoxy compound was extracted with ether, the extract dried over sodium sulfate, and concentrated. I g. of ethoxy compound crys- tallized on cooling. Recrystallized twice from ligroin it melted at 108°. The product is readily soluble in benzene, ether and ligroin, less so in alcohol. The ethoxy compound, peculiarly, melts at practically the same tem- perature as the carbinol, although both were recrystallized until their melt- ing points became constant. A mixture of the two, however, began to ¹ Gomberg and Cone, Ber., 39, 3281 (1906). 9 soften at 86°, and melted at 93°, showing conclusively their non-identity, which was confirmed by analysis. 21 17 Calculated for C₂H₁,Cl₂BгO: Cl₂, 16.27; Br, 18.33. Cl2, 16.59; Br, 18.02. Found: o-Bromodi-p-chlorotriphenylanilidomethane, (BrC¸H₁)(C¸H₁Cl)2—C— On NHCËÍË.—To 2 g. carbinol chloride in 20 cc. benzene solution was added 20 per cent. excess of aniline (1.1 g.) and the solution allowed to stand for a day. The anilide is unusually insoluble, and a considerable portion precipitated with the aniline hydrochloride. The latter was extracted with hot benzene and the extract united with the main solution. concentrating at room temperature the anilide was cbtained in almost quantitative yield, 2.0 g. It is difficultly soluble in all the usual solvents except hot benzene. When heated it turns green at 211° and melts to a dark liquid at 212°. 18 Calculated for C25H₁gCl₂BrN: N, 2.90. Found: N, 2.94. Unsymmetrical o-bromodi-p-chlorotetraphenylethane, (BrCH₁)(CH¸Cl)2 C-CH,CH.-This was obtained by the same method as described for the other unsymmetrical tetraphenylethanes. Recrystallized from alcohol it melts at 162°. The substance is readily soluble in benzene, less so in ether and ligroin, and is fairly insoluble in alcohol, from which it can be recrystallized almost quantitatively. Calculated for C28H19Cl₂Br: Found: Cl2, 14.71; Br, 16.59. Cl2, 14.78; Br, 16.19. CH₁Br(þ) p-Bromodi-p-chlorotriphenylcarbinol Chloride, (Þ) CICH₁ (p) CICH₁ C1 -This product was made from the corresponding carbinol, which was prepared by the Grignard reaction. 2.29 g. magnesium were dissolved in an ethereal solution of 24.4 g. (1.1 molecules) of p-chloroiodobenzene. To the solution was added 10.1 g. (1 molecule) of p-bromobenzoic methyl ester, and the mixture boiled six hours. It was decomposed as usual, and the product, after steaming off the excess of p-chloroiodobenzene, taken up in ether and boiled under reflux condenser with calcium chloride and charcoal. The colorless carbinol crystallized from the concentrated solution, and was washed with ligroin. The product recrystallized twice from petroleum ether melts at 106°. Yield, 12 g. which is 65 per cent. of the calculated amount. Calculated for C,,H,,Cl₂BrO: Found: 13 Cl2, 17.38; Br, 19.59. Cl2, 17.68; Br, 19.05. The carbinol chloride obtained by the action of hydrochloric acid on | ΙΟ the above carbinol and melting at 122° has already been described by Gomberg.¹ p-Bromodi-p-chlorotriphenylethoxymethane, (BrCH₁)(C¸H¸CI)₂C.OC₂H5. -The ethoxy compound was prepared similarly to the corresponding o- bromo derivative as described above. Recrystallized from ether the sub- stance melts at 188°. The substance is soluble in benzene, moderately soluble in ether, comparatively insoluble in alcohol. Calculated for C₂H,,Cl₂BrO: Cl2, 16.27; Br, 18.33. Found: Cl2, 16.65; Br, 18.07. p - Bromodi - p - chlorotriphenylanilidomethane, (BrСH¸)(C¸H¸CI)¿C. NHCH.-This anilide was obtained similarly to the above-described anilides. Recrystallized from a mixture of benzene and petroleum ether, the product on heating turns green at 178° and melts at 182°. It is soluble in benzene, fairly scluble in ether, insoluble in alcohol and ligroin. Calculated for C25H18C12BrN: N, 2.90. N, 2.92. Found: The synthesis of the di- and of the tri-ortho-chlorotriphenylcarbinol chlorides was attempted by the Grignard reaction but the reaction, when it took place at all, furnished such unsatisfactory yields of the desired products that the preparation of these compounds was given up for the present. 3. Action of Molecular Silver on ortho- and para-Halogenated Deriva- tives of Triphenylcarbinol Chloride. When para-halogenated derivatives of triphenylcarbinol chloride in ben- zene solution are shaken with an excess of molecular silver, not only the carbinol chlorine is removed, but part of the para-nuclear halogen as well, indicating the quinoid nature of the triphenylmethyl analogs in the bril- liantly colored solutions. Approximately one-half of an atɔm of para- halogen was removed when one such atom was present, three-fourths when there were two para-halogens, seven-eighths when there were three. Gomberg² explained this by the hypothesis that the reaction products successively couple together as they are formed, with elimination of an atom of para-halogen from each complex at each coupling, on the assump- tion that in solution triphenylmethyl exists partially in the quinol state as suggested by Jacobson, (CH),C == CHA H 4 C(C₂H₂)3 The theoretical amount of para-halogen removable in this manner should be one-half, three-fourths, or seven-eighths of an atom for each original ¹ Ber., 40, 1863 (1907). 2 Ibid., 40, 1886 (1906). II molecule of carbinol chloride, according as the latter possesses one, two, or three atoms respectively of para-halogen. The requirements of the theory were thus found to harmonize in a remarkably close way with the results obtained. The present experiments were undertaken in order to obtain experi- mental data from derivatives containing both para-bromine and para- chlorine in the same molecule, and to ascertain from ortho-halogenated derivatives whether evidence could be obtained of ortho-quincid forma- tion. 0.3 to 0.5 g. portions of the carbinol chlorides tabulated below were sealed with benzene and 1.5 to 2.5 g. portions of molecular silver; the tubes were then shaken at room temperature for several months and their contents analyzed. TABLE I.-ACTION OF MOLECULAR SILVER on ORTHO- AND PARA-HALOGEnated De- RIVATIVES OF TRIPHENYLCARBINOL CHloride. Carbinol chlorine. Days shaken with silver. Found. cent. Per Calculated. Per cent. Per cent Atoms. Found. Ring chlorine. Per cent cal- culated atom. Per cent. Ring bromine. Found. Atoms. Per cent. cal- culated. atom. Total ogen, I Atoms. ring hal- No. Carbinol chloride. I o-chloro 2 0-bromo 106 11.15 11.34 0.00 0.00 222 8.98 9.92 3 0-bromodi-p-chloro 4 0-bromodi-p-chloro 206 8.31 5 p-bromodi-p-chloro 198 8.31 6 p-chlorodi-p-bromo 157 7.53 7 p-chlorodi-p-bromo 157 7.53 7.53 2.17 0.288 206 8.31 8.31 5.90 0.711 8.31 6.05 0.728 8.31 3.31 0.398 7.53 2.17 0.288 II.34 0.00 0.00 0.00 22.37 0.00 8.31 0.00 0.00 18.75 0.711 8.31 0.00 0.00 18.75 0.728 8.31 5.74 0.307 18.75 0.706 7.53 9.22 0.542 16.98 0.830 7.53 9.76 0.575 16.98 0.863 The conclusions afforded by the above results may be summarized as follows: 1. The results of numbers 1, 2, 3, and 4 show that no trace of ortho- halogen was removed in any case. There is no indication of the forma- tion of ortho-quinoid nuclei. 2. The total amounts of para-halogen removed, as related to the num- ber of para-halogen atoms present, agree (except No. 5, which is too low, probably from experimental error) with those of Gomberg and Cone mentioned above, and substantiate strongly the explanation given for the reaction and based upon successive coupling of the products. 3. Bromine in the para-position is somewhat more reactive towards molecular silver than is chlorine. A larger portion of a bromine atom is removed in Nos. 6 and 7 than of chlorine in No. 5; also of bromine in No. 5 than of chlorine in Nos. 6 and 7. Comparative Influence of Halogen in Ortho and Para-positions upon Color caused by the Action of Molecular Silver upon Halogenated Triphenylcar- binol Chlorides. The solutions contained 0.3 to 0.5 g. carbinol chloride 12 dissolved in 12-15 cc. benzene and sealed in tubes with the molecular silver, free from all traces of air. The derivatives are arranged in pairs differing, except in the last, only in the position of the halogen atoms, so that the comparative effect of halogen in the ortho and para positions is clearly brought out. TABLE II. Color caused by action of silver. Substance. Triphenylcarbinol chloride.. • Mono-p-chlorotriphenylcarbinol chloride.. Mono-o-chlorotriphenylcarbinol chloride. Mono-p-bromotriphenylcarbinol chloride. Mono-o-bromotriphenylcarbinol chloride. • • Mono-p-bromodi-p-chlorotriphenylcarbinol chloride Mono-o-bromodi-p-chlorotriphenylcarbinol chloride. ❤ · Di-p-bromotriphenylcarbinol chloride. Di-o-chlorotriphenylcarbinol chloride.. Mono-p-chlorodi-p-bromotriphenylcarbinol chloride..... Deep yellow Light red Deep red Brown-red Deep red. Tinge purple. Orange Deep purple Orange Deep violet Orange The greater influence of the crtho-halogen substituents in deepening the color is apparent, especially in the di- and tri-halogenated compounds. The para derivatives give colors varying from orange to red, the ortho from red to purple and violet. The similarity in the greater influence of ortho substituents in the analogs of triphenylmethyl and in the mala- chite green derivatives as observed by Noelting¹ indicates a similar struc- tural cause of color. 4. Action of Silver Chloride on ortho-Brominated Derivatives of Tri- phenylcarbinol Chloride. When tri-p-bromotriphenylcarbinol chloride is dissolved in sulfur dioxide and shaken with silver chloride the three bromine atoms are suc- cessively replaced by chlorine. The carbinol chloride in solution is at least partially in the quinoid form, an equilibrium existing, when part of the bromine has been replaced, between molecules containing brom- inated nuclei in the quinoid state and those in which a chlorinated nucleus is the quinoid one. The silver chloride reacts with the former, the dis- turbance of the equilibrium resulting in the shifting of the quinoidation to fresh bromine-containing nuclei as those formerly brominated become chlorinated. The replacement of bromine by chlorine consequently continues until it is complete, whether one, two, or three para-bromine atoms are present at the start. In the following diagram is shown the mechanism covering the replacement of the first two bromine atoms only, assuming, as we may, an equilibrium between products (3) and (4), and also between products (6) and (7). 1 Ber., 39, 2041 (1906). (1) C 6 SO₂ CH₁Br CH_Br SO C&H Br Cl 4 (2) C (4) C CH4Br CHBr SO, CH CI (5) C-CH, Br AgCl (6) C6H4C1 13 -C.H,Br -CH Br AgC1 Br (3) C1 Cl CH4Br CH4Br CA CHICC < C1 4 C1 CH,Br C1 CH 4 C1 CH,C1 C=c (7) C Cl -CH,Br -C,H,C1 -CH C1 --Cl As will be shown by experiments with the sulfates described below, (p. 14) para-bromine exerts a stronger influence than chlorine in directing the quinoidation towards the nucleus in which it is situated. Even when there are two para-chlorine atoms to one bromine the quinoidation takes place almost entirely in the brominated nucleus. We have now tested the ability of sulfur dioxide to induce ortho as well as para quinoidation. 0.3 g. portions of ortho-bromo- and ortho- bromodi-p-chlorotriphenylcarbinol chloride were sealed in bomb tubes with sulfur dioxide and 0.6 g. portions of silver chloride. The tubes were shaken one month at 20° and one month at 50°. No silver bromide was formed in either case. The tautomerization resulting from the influ- ence of the solvent on a carbinol chloride cannot apparently be influenced by ortho-halogen, even bromine, to form ortho-quinoid nuclei. 5. Action of Silver Sulfate on ortho- and para-Halogenated Derivatives of Triphenylcarbinol Chloride. The first action of silver sulfate on para-brominated triphenylcarbinol chloride is to remove the aliphatic chlorine attached to the central carbon atom, with formation of intensely colored neutral sulfates of the triphenyl- methyl radicle. The second step in the reaction is the removal of one— and no more-para-bromine atom from each triphenylmethyl radicle. This removal can only be explained on the assumption that the sulfates undergo tautomerization at the moment of their formation and conse- quently have a quinoid nucleus as represented below, because the halo- gen in benzoid rings is absolutely inert towards silver salts. (C,H,Br),C.SO,→→ (CH,Br),C Br SO The experiments tabulated below were undertaken in order to ascer- tain the comparative influence of chlorine and bromine in directing the quinoidation to or from the rings in which the halogens are substituted, and further to test the possibility of ortho-quinoid formation. The silver sulfate (1.5 to 2.5 g.), carbinol chloride (0.3 to 0.5 g.), and methyl · sulfate as a solvent were sealed in a tube and shaken mechanically in a 14 thermostat at 50°. The different carbinol chlorides are designated by the positions of their ring halogens. TABLE III.-ACTION OF SILVER SULFATE ON CARBINOL CHLORIDES IN METHYL SULFATE. Carbinol chlorine Ring chlorine removed. Ring bromine removed. removed. Carbinol No. chloride. Days shaken. Calculated. Found. Calculated. Found. Calculated. Found. I Mono-o-chloro- 7 II.32 10.84 11.32 0.00 2 Mono-0-bromo- 3 9.92 9.82 22.37 0.00 66 (6 3 I I 9.92 9.65 22.37 0.00 4 0-Bromodi-p-chloro- 2/3 8.31 8.31 8.31 7.72 18.75 0.00 (( 5 6 31/2 8.31 8.31 8.31 8.17 18.75 0.00 5 8.31 8.31 8.31 8.28 18.75 0.00 IO 8.31 8.31 8.31 8.24 18.75 0.00 78 ུ 13 8.31 8.31 8.31 8.35 18.75 0.00 9 p-Bromodi-p-chloro- 9 8.31 8.31 8.31 0.25 18.75 16.03 ΙΙ 8.31 8.31 8.31 0.80 18.75 15.92 ΙΟ 19 8.31 8.31 8.31 0.00 18.75 16.46 II 26 8.31 8.31 8.31 0.77 18.75 15.34 12 13 p-Chlorodi-p-bromo- 6 7.53 7.53 7.53 0.04 16.98 16.83 14 7.53 7.53 7.53 Ο.ΙΟ 16.98 16.89 14 The solutions all became deep red at once upon shaking with silver sulfate, the substance containing ortho-halogen giving a deeper color, almost black, than those with only para-halogen. The "ring chlorine" was estimated by subtracting the theoretical carbinol chlorine from the total. The bromine was estimated, as usual, by igniting the mixed halides in chlorine gas. From the data presented we may draw the following conclusions: 1. When bromine and chlorine occupy para-positions in different nuclei of the same substance, the bromine is removed in preference to the chlorine. When one bromine and two chlorine atoms occupy the para- positions (Nos. 9-12) not more than a tenth of an atom of ring-chlorine is removed, while nearly nine-tenths of the single atom of bromine is taken out. When there are two para-bromines to one para-chlorine (Nos. 13-14) an atom of bromine is taken out quantitatively, the amount of chlorine in excess of the carbinol atom being scarcely above the limit of error. In each case, however, the total amount of ring-halogen taken out by the silver sulfate never exceeds one atom. The para-bromine atom consequently influences the quinoid transformation, occurring when a sulfate is formed, strongly towards the nucleus which contains the bromine. 2. No ortho-quinoid rings are formed in the colored sulfates. Even when an ortho-bromine atom is the only halogen in the three nuclei 15 (Nos. 2 and 3), in which case it would presumably exert a strong influ- ence towards the formation of an ortho-quinoid ring, if such were possible, not the slightest trace of bromine is removed. When in addition to the ortho-bromine in one nucleus there are para-chlorines in the other two (Nos. 4-8), one atom of para-chlorine is removed quantitatively, showing that in the ortho position bromine completely loses its influence to direct the quinoidation towards its own nucleus. A series of experiments was also carried out with sulfur dioxide instead of methyl sulfate as a solvent. The results obtained agree with the pre- ceding, in showing that one para-halogen atcm is removed from the colored neutral sulfate but that ortho-halogen is untouched. Inhibitive effect of sulfuric acid upon removal of para-halogen from col- ored sulfates by silver sulfate. It was found, in the case of the p-bromodi-p-chloro- and o-bromodi- p-chloro-sulfates, as by Gomberg in the cases of the p-bromo- and tri-p- bromo-sulfates, that the presence of slight amounts of free sulfuric acid in the methyl sulfate reaction mixture protects the para-halogen from removal by silver sulfate. In the experiments tabulated below 0.3 g. portions of carbinol chloride with 15 g. portions of silver sulfate were weighed into test tubes, 10 CC. of methyl sulfate and the indicated amount of sulfuric acid then added, and the tube at once sealed and placed on the shaker at 50°. The results of Table IV were obtained after nineteen days' shaking. TABLE IV.—INHIBITIVE EFFECT OF SULFURIC ACID ON REMOVAL OF PARA-BROMINE BY SILVER SULFATE FROM р-BROMODI-p-CHLOROTRIPHENYLCARBINOL SULFATE. Molecules sulfuric acid added per i mol. carbinol chloride. O No. I 2 3 I 4 2 1/2 Per cent. bromine removed. 16.46 10.77 1.32 0.13 Calculated 1 atom of bromine. 18.75 18.75 18.75 18.75 From the o-bromodi-p-chlorocarbinol chloride an atom of para-chlorine is quickly and quantitatively removed in the absence of sulfuric acid. But here again in the presence of acid, the action of silver sulfate is greatly hindered. The effect of time of reaction as well as that of the amount of acid was studied in this case. Table V.-EFFECT OF AMOUNT OF SULFURIC ACID AND DURATION OF REACTION ON AMOUNT OF PARA-CHLORINE REMOVED BY SILVER SULFATE FROM 0-BROMODI- Molecules sulfuric acid added per i mol. carbinol chloride. O p-CHLOROTRIPHENYLCARBINOL SULFate. Percentage of para-chlorine removed. Calculated, 16 hrs. 3½ days. Io days. I atom. 7.72 7.36 8.17 7.21 8.24 8.31 7.72 8.31 I 5.24 6.40 7.08 8.31 2 1.13 2.72 3.30 8.31 1/2 16 It is apparent that there is no simple ratio between the amount of sulfuric acid added and the amount of para-halogen protected. But it is equally apparent that the inhibitive effect of sulfuric acid is much smaller in case of chlorine than in case of bromine in the para position. Tri-p-halogen Acid Sulfates. Baeyer¹ obtained by the action of concentrated sulfuric acid upon an excess of carbinol in solution brown crystallin sulfates of tri-p-chloro- and tri-p-iodo-triphenylcarbinol. The sulfates showed a green metallic luster, and were of the constitution (R,C)SO,H.H₂SO4. Gomberg2 pre- pared similar sulfates of tri-p-chloro- and tri-p-bromocarbinol, using methyl sulfate instead of chloral as a solvent, the methyl sulfate having the marked advantage of being a solvent for sulfuric acid. Prepared with methyl sulfate, the products were found to contain not one mcle- cule of excess sulfuric acid, as Baeyer's results indicated, but from 11/2 to 2 molecules. The ortho-bromodi-p-chlorosulfate could not be brought to crystal- lize, but the p-bromodi-p-chlorocarbinol gave by Gomberg's method a beautifully crystallin sulfate with green metallic luster, the yield being about 80 per cent. of the calculated amount. If the sulfuric acid solu- tion in methyl sulfate was added to a cold benzene methyl sulfate solu- tion of the carbinol, the carbinol sulfate usually separated quickly in fine crystals, while if the solution was warm, crystallization occurred slowly, with formation of large, compact crystals, in form suggestive of cubical salt crystals. The analyses gave compositions similar to those of Gom- berg's polysulfates, the percentages of (SO,) varying from that correspond- ing to (R,C)SO,H.1'/₂H₂SO, to that corresponding to (R,C)SOH.2H₂SO. Either the substance is a mixture of the sulfates containing, respectively, one and a half and two extra molecules of sulfuric acid, or, as seems more probable, it contains two extra molecules of sulfuric acid, the analysis falling low because of the readiness with which the sulfate hydrolyzes, even by short contact with air. 4 The polysulfate containing three molecules of sulfuric acid, from p- bromodi-p-chlorotriphenylcarbinol, was dissolved in both methyl sul- fate and sulfur dioxide, and both solutions shaken two months at 50° with silver sulfate. The residue from the methyl sulfate tube gave too slight an amount of silver halide for estimation, although it could be qualitatively determined. That from the sulfur dioxide gave an amount of silver halide corresponding only to 0.37 per cent. of bromine. It is clear from the above results that these acid sulfates, unlike the normal sulfates, do not give up their para-halogen on treatment with ¹ Ber., 38, 1162 (1905). 2 Ibid., 40, 1853 (1907). 17 silver sulfate, although every precaution has been taken to insure against the possible reverse tautomerization to the benzenoid form during the treatment with the silver salt. Similar negative results with the acid sulfates, obtained by Baeyer previously in point of time to the positive results on the normal sulfates as described by one of us, led Baeyer to the conclusion that these, and also all the other similarly colored salts are not quinoid in constitution. Baeyer's conclusion based, as it is, on evidence of negative nature only, does not seem to us warranted. Cer- tainly it is not warranted as regards the colored carbinol chlorides or the normal sulfates, the quinoid nature of which has been inferred from results of strictly positive character. We have shown that the normal sulfates, like the colored carbinol chlorides, contain a labil para-halogen and consequently, like the latter, must be quinoid. It has been further shown that in these very same sulfates the para-halogen becomes partially protected against the action of silver salt by the presence of free sulfuric acid, and the protection becomes almost complete when the relative concentration of the free acid is equal to two molecules. The action of the free acid upon the nor- mal sulfates consists primarily, no doubt, in the formation of acid sul- fates. Does the acid at the same time induce the tautomerization of the active quinoid to the non-reactive benzenoid form, and this without change of color? Such an inference as regards the action of sulfuric acid is entirely contrary to what we know of the action of hydrochloric acid. Gomberg and Cone¹ have demonstrated that certain normal, colorless carbinol chlorides of this series are converted by excess of hydrochloric acid into acid chlorides which are colored and are quinoid. It does not seem reasonable to assume that sulfuric acid would act in the very oppo- sit direction from that in which hydrochloric acid acts. Consequently the non-reactivity of the para-halogen in the acid sulfates must find its explanation in some hypothesis which would still admit of a struc- tural analogy between the normal and the acid sulfates. Some years ago, soon after the publication of Baeyer's negative results with the acid sulfate, Stieglitz and Barnard² have made the suggestion that the non-activity of these acid sulfates can still be reconciled with their having a quinoid constitution, if we assign to them the constitution I as chloronium" or "bromonium" salts. (" (BrC₂H₁)₂C:CH₁:Br.OSO₂H I. (CHÁ)₂C:C¸H₁:H.OSO₂H. 5/2 II. Baeyer criticized this interpretation because on the basis of this view, in the salts obtainable from carbinols which contain in place of the para- ¹ Ann., 370, 142 (1909). 2 J. Am. Chem. Soc., 27, 1016 (1905). : 18 halogen some other group, that group, or even the para-hydrogen itself must likewise become basic and trivalent (Formula II).¹ We believe, however, that the kernel of Stieglitz and Barnard's view is none the less correct. So modified as to meet Baeyer's criticism, that interpretation furnishes a most satisfactory and plausible explanation of the entire behav- ior of these salts. Structure of the Sulfates. The normal sulfates must be considered as quinoid and on that basis can have only cne possible constitution as given on page 13. The pai- tial protection of the para-halogen in the mono-acid and the complete protection in the poly-acid sulfates is most rationally explained by assuming a structural connection between the para quinoid halogen and the combined acid radicles. The structure of the mono-acid sulfates may be represented as I or II, and that of the poly-acid sulfates as III, where X = halogen. X R₂C=C¿H I. 4 SO,H H X-H X< SO H R₂C=CH, R₂C=C¸H¸ SO SO,H II. III. The halogen thus assumes in II the trivalent condition, as in chlorites, bromites, and in iodonium compounds, and in III perhaps even a higher valence, as in chlorates and bromates. The halogen, having become tri- valent, has changed its properties, and is no more capable of precipita- tion by silver than the chlorine in the chlorate or in some of the complex inorganic double salts of Werner that contain non-ionizable halogen. Such halogen we will term, after the iodonium compounds, bromonium or chloronium. For the polysulfates only the bromonium or the chloronium formula III applies. It expresses the chromatic nature of these substances as well as the complete protection of the para-halogen from the action of the silver salts. For the mono-acid sulfates, present results indicate that both I and II apply, I being in considerable evidence when the para- halogen is chlorine, II almost the exclusive form when the para-halogen is bromine; i. e., bromine shows much greater tendency to assume the bromonium form than does chlorine to assume the chloronium. This relative tendency of the two halogens is in harmony with our knowledge as regards the basicity of iodine in the iodonium compounds. The prob- ability that para-chlorinated mono-acid sulfates are largely of form I follows from the fact that the greater part of their halogen appears unpro- tected. Table V (p. 15) shows that the addition of one molecule of sul- furic acid to the normal sulfates—a little more than sufficient to form the mono-acid sulfate-does not protect the greater part of the para-chlo- 1 Ber., 40, 3083 (1907). 19 rine. This seems still more evident when one compares Table V with Table IV. The latter shows that a molecule of acid under the same conditions almost completely protects the para-halogen when the latter is bromine. Gomberg's results with mono-p-bromo- and tri-p-bromocarbinol chlo- rides are similar. The following experiments show more definitly the inertness of para- bromine in a para-brominated mono-acid sulfate. One molecule of sul- furic acid, just enough to form the mono-acid sulfate, was dissolved in methyl sulfate as a solvent with one molecule of mono-p-bromotriphenyl- carbinol chloride, and the liberated hydrochloric acid driven off with dry carbon dioxide, until the carbinol chlorine was almost quantitatively removed. The carbinol chloride was thus changed to acid sulfate, RCSOH. To the solution of the organic sulfate was now added an excess of silver sulfate and the mixture sealed off and shaken for thirty days at 50°. 0.036 atom of chlorine was obtained, corresponding nearly to the slight amount of carbinol chlorine not driven off by the previous treatment with sulfuric acid, but no bromine whatever could be detected. In a similar experiment with the p-bromodi-p-chlorotriphenylcarbinol chloride less than 0.1 atom of bromine was freed. It is, therefore, apparent that all the para-bromine in mono-acid sulfates is protected, a fact which is explainable on the basis of structure, the para-brominated mono-acid sulfates having almost entirely the bromonium structure II. The Action of Sulfuric Acid upon Triphenylcarbinol Chlorides. In order to gain some idea as to the avidity with which the sulfuric acid is held in the bromonium salts the following set of experiments was carried out. Exactly one molecule of sulfuric acid was allowed to act upon two molecules of various triarylcarbinol halides. In case of the simple triphenylcarbinol chloride the reaction may be expected to pro- ceed in these two steps: (a) (b) H R¸C.C1 + H₂SO¸ = R₂C : CH₁