EXCHANGE 8061 '12 KVP IVd Diphenyl-beta-Naphthylmethyl and the Color of Free Radicals. A Contribution to the Chemistry of Free Radicals A DISSERTATION row T.Q: SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY IN THE UNIVERSITY OF MICHIGAN -By Frederick William Sullivan', Jr. EASTON, PA. ESCHENBACH PRINTING COMPANY 1921 Diphenyl-beta-Naphthylmethyl and the Color of Free Radicals. A Contribution to the Chemistry of Free Radicals A DISSERTATION SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY IN THE UNIVERSITY OF MICHIGAN By Frederick William Sullivan, Jr, EASTON, PA. ESCHENBACH PRINTING COMPANY 1921 KXCHANGfc TABLE OF CONTENTS I Introduction 1 II Experimental 4 Diphenyl-/3-naphthyl Carbinol . . . 4 Diphenyl-/3-naphthylmethane 5 Preparation of Diphenyl-/3-naphthylmethyl 6 Reaction with Oxygen 7 Peroxide 8 Action of Light and of Acids 8 Action of Iodine 8 The Conductivity 8 Molecular Weight 11 Colorimeter for the Study of the Changes in Intensity of Color of Solutions of Free Radicals 16 The Effect of Dilution on the Color of Solutions of Free Radical in Benzene, Nitrobenzene and Cyclohexane at their Respective Freezing Points 18 The Effect of Temperature on the Color of Solutions of the Free Radical in Carbon Tetrachloride and in Toluene 18 III Discussion of Results 19 IV Summary 23 ACKNOWLEDGMENT It is a pleasant duty to express my sincere appreciation of the inspiring supervision and interest of Professor M. Gomberg, at whose suggestion and under whose direction this work was carried out. FREDERICK WIUJAM SULLIVAN, JR. Ann Arbor, Mich. August 23, 1922 DIPHENYL-BETA-NAPHTHYL-METHYL AND THE COLOR OF FREE RADICALS Our purpose in the present investigation has been to prepare the free radical, diphenyl-/3-naphthylmethyl and to study it with regard to the characteristic chemical reactions and physical properties of free radicals ; as a later development, because of the pecular adaptability of this radical to the problem, it was studied es- pecially from the standpoint of the relation between color and chemical constitution. The characteristic phenomena investigated included the absorption of oxygen, the reaction with iodine, the effects produced by light and acids, the conductivity, the dissociation of the hexa-aryl ethane, the formation of additive compounds, and finally, the effect of dilution and temperature changes upon the color of solutions of the free radical. The striking fact that triphenyl methyl, although a hydrocarbon, is colored when in solution has given rise to many explanations in order to account for this phenomenon. The principal theories regarding the re- lation between color and constitution of triphenylmethyl are, briefly stated, the following. 1. Colorless hexaphenylethane dissociates to form the colored free radical, 1 RsCCRj > 2 R 3 C-. 2. Dissociation occurs and is followed by tautomerization of the benzerj phenylmethyl into the quinonoid monomolecular tautomer. The color is attributed to the quinonoid tautomer, C.'- (Cf. p. 1832.) (C 6 H 5 ) 3 C-C(C,H 6 ) 3 ^ (C 6 H 6 ) 3 C + (C 6 H 5 ) 3 C A (C,H,),C -v XTT. C The work on p-halogen substitution products of triphenylmethyl pro- vides definite, strictly chemical, evidence for the existence of tautomerism with the formation of some kind of a quininoid tautomer. 3 The evidence based on purely physical chemical considerations is some- what contradictory. Schmidlin observed that upon shaking a solution of triphenylmethyl with oxygen the color vanishes but reappears on stand- ing, and that this phenomenon takes place several times before the solution is completely and permanently decolorized. By weighing the amount of peroxide formed in a single decolorization, he was able to calculate the amount of the colored form which was present. His results indicated that the ratio of colorless to colored form was about 10 to 1. However, since the molecular weight of triphenylmethyl at room temperature corresponds to the dimolecular formula, Schmidlin concluded that color must be due to tautomerism and not dissociation. Schmidlin also observed the variation in the intensity of color of tri- phenylmethyl solutions at different temperatures and found that at the freezing point of chloroform, 63, the color entirely disappeared. Later 4 he determined the molecular weight of triphenylmethyl in benzene in an apparatus devised to permit the determination by the freezing-point and boiling-point methods on the same sample, and found no changes in dis- sociation at the different temperatures. He then conducted colorimetric determinations over a range of temperatures from the freezing point to the boiling point of benzene and found that the color increased 6-fold. From these results he concluded that the intensification of color of triphenyl- 1 Baeyer, Ber., 35, 1195 (1902). Flurscheim, J. prakt. Chem., 71, 505 (1905). Chichibabin, Ber., 40, 3970 (1907). Wieland, ibid., 42, 3029 (1909). Piccard, Ann., 381, 347 (1911). 2 Gomberg, Ber., 46, 228 (1913). 3 Gomberg and others, ibid., 39, 3294 (1906); ibid., 40, 1886 (1907)- Ann 370, 181 (1909); ibid., 376, 208 (1910); /. Am. Chem. Soc., 33, 538 (1911). 4 Schmidlin, Ber., 45, 3181 (1912). methyl solutions with rise of temperature must be due to something other than dissociation. On the other hand, Wieland, Schlenk and Mair, and Piccard took the very opposite point of view. Wieland 5 concluded from Schmidlin's earlier experiments and from -the molecular weight determinations of Gomberg and Cone that, after allowing for the inherent errors of the cryo- scopic method, the dissociation hypothesis completely explains color forma- tion. Schlenk and Mair 6 gave as evidence that the increase of color with rising temperature is due to dissociation, the results of their molecular weight determinations on triphenylmethyl by the boiling-point method. They found, contrary to Schmidlin, that hexaphenylethane in boiling benzene is dissociated to the extent of 25%. Hence, according to them the increase in color intensity is due to dissociation. What seemed to be the most convincing evidence in favor of the dis- sociation theory is found in the work of Piccard 7 on the increase of color of triphenylmethyl solutions with dilution. This behavior constitutes a deviation from Beer's law, which states that the color absorption of so- lutions should remain constant if there were no chemical changes on di- lution. In the case of hexaphenylethane we are concerned with one of two phenomena, either tautomerism or dissociation. If it were tautomerism the equilibrium would be between two isomers and, Piccard assumes, such an equilibrium should be independent of dilution ; but if it were dissoci- ation then the equilibrium would be between a polymer and the products of its dissociation, and such an equilibrium should be influenced by dilution. He conducted a series of experiments on hexaphenylethane solutions and found a decided increase in color upon dilution. He therefore concluded that the phenomenon must be dissociation. Although Piccard's assumption is in accordance with the general belief 8 that the equilibrium between tautomers is independent of dilution there is much experimental evidence to the contrary. W. Wislicenus 9 found that the equilibrium between the aldo and enol forms of ethyl formyl phenyl acetate was shifted in favor of the enol form by concentration. Kurt H. Meyer and Kappelmeier 10 found that the equilibrium between the enol and keto forms of aceto-acetic ester in hexane varied from a concentration of 9.3% enol in a 90% solution of the ester to 58.6% in a 2% solution. 5 Wieland, Ref. 2. t; Schlenk and Mair, Ann., 394, 178 (1912). 7 Piccard, ibid., 381, 347 (1911); Arch. sci. phys. Nat., Geneva, 1913, 368. Hantzsch Ann., 384, 135 (1911); ibid., 398, 379 (1913). 8 Ostwald, "Lehrbuch der Allgemeine Chemie," Wilhelm Englemann 2nd ed., 1902, p. 604. 9 Wislicenus, Ann., 291, 147 (1896). "' Meyer and Kappelmeier, Ber., 44, 2722 (1911). More recently, O. Miller 11 studied the equilibrium between the two forms of 2-amino-a-naphthoquinone-imide and found that this equilibrium was influenced by dilution. O O ^ [NH 2 IH So it seems, after all, that Piccard's results do not prove that color in triphenylmethyl solutions is necessarily due to dissociation, and the possibility remains that it may be due to tautomerism, or perhaps to dissociation with subsequent tautomerism. The difficulties in the study of this question and the contradictory nature of the conclusions results from the fact that hexaphenylethane is very slightly dissociated, not more than 5% at ordinary temperatures, and therefore conclusions regarding the relation between color and dissoci- ation must be based largely on speculation. Its color is neither sufficiently intense nor sufficiently different from that of its decomposition products for reliable colorimetric study. In diphenyl-/3-naphthylmethyl we have a substance of very considerable and accurately measurable dissociation; its color, bright red in concentrated solutions shading to light yellow in dilute solutions, lends itself readily to colorimetric study; it is quite stable towards light; it is readily soluble in a variety of solvents throughout a wide range of temperature. In addition, it shows striking diminution of color within the temperature range of ordinary cryoscopic determinations. On account of this fortuitous combination of properties, a study of diphenyl- j3-naphthylmethyl was undertaken in an effort to throw some light on the relation between color formation and changes in concentration and in temperature. Experimental Diphenyl-/3-naphthyl Carbinol, (CeHBM/S-CioI^COH. Diphenyl-/3-naphthyl car- binol was prepared by the Grignard reaction from ethyl /S-naphthoate and phenyl- magnesium bromide. Experiments showed this method to be preferable to the prepara- tion from /3-naphthylphenyl ketone and bromobenzene, or from benzophenone and /3-bromonaphthalene. The starting substance in the preparation was the sodium salt of /3-naphthalene sulfonic acid which was converted to the nitrile by dry distillation with potassium ferrocyanide. 12 Two pares of the sulfonate and one part of anhydrous ferro- cyanide were heated in 250g. portions in Pyrex Kjeldahl flasks placed horizontally in an air-bath so that the vapors were removed as fast as formed. The condensate was dried by heating until the water was driven off and was then distilled. The portion boiling between 290 and 310 was collected. The nitrile can be crystallized from petroleum 11 Miller, /. Russ. Phys. Chem. Soc., 47, 1544 (1915). 12 Eckstrand, /. prakt. Chem., 38, 139 (1888). ether and melts at 68. The redistilled substance was hydrolyzed according to the pro- cedure of Baeyer and Besemfelder 13 and the resulting acid purified by dissolving in alkali and reprecipitating with acid. In this manner 650 g. of acid was prepared from 3900 g. of the sulfonate, a yield of 20%. For the esterification of the acid, 20 g. of /3-naphthoic acid, 20 g. of sulfuric acid and 200 cc. of absolute alcohol were boiled under a reflux condenser. The reaction product was poured into an excess of dil. alkali, the ester was extracted from this mixture, dried and distilled, and the fraction between 300 and 304 was collected. The yield of ester was 97.5%. When the proportions were changed to 1 part of 0-naphthoic acid, 1 part of sulfuric acid, and 2 parts, instead of 10, of alcohol, the yield of the ester was reduced, but, since the unchanged acid could be recovered, this latter method was adopted. Ullmann 14 found that the product of the reaction between methyl-/3-naphthoat,e and phenylmagnesium bromide was an oil and he was unable to isolate the carbinol from it. Following the same procedure and using either the methyl or ethyl ester we were able to obtain the crystalline carbinol but the yield was not more than 25%. The yield was improved by replacing the ether after the formation of the Grignard reagent by benzene or, better, by toluene. Xylene was not so satisfactory. The following pro- cedure was finally adopted. A mixture of 10 g. of magnesium turnings in 50 cc. of bromobenzene and 200 cc. of ether was boiled until the metal disappeared; 100 cc. of toluene was then added and the ether was distilled ; 28 g. of ethyl /3-naphthoate was then added gradually; the heat of the reaction kept the toluene boiling, and the temperature of the reaction mixture was about 115. After the ester had been added and when boiling had ceased, the mixture was cooled and decomposed with ice and acetic acid. The product was distilled with steam and the residue was taken up in carbon disulfide. The solution was dried, filtered and concentrated. On the addition of petroleum ether, a light-yellow crystalline product, the carbinol, separated and this was suitable for con- version to the chloride without further purification; yield, 28 g., or 65%. The car- binol is readily soluble in all the usual solvents and can be recrystallized from alcohol or, better, from carbon disulfide, by the addition of petroleum ether. The pure, white, recrystallized product melts at 117.5. Analyses. Calc. for C 23 H 18 O: C, 89.0; H, 5.84. Found: C, 89.4, 89.4; H, 5.90, 5.92. When boiled with acetic acid in the presence of mineral acids, the carbinol is con- verted to the fluorene. (C 6 H 6 ) 2 (C 10 H 7 )COH > | 6 ' NCH.C 6 H 8 . + H 2 O C 10 H/ This substance, when recrystallized from chloroform by the addition of alcohol, melts at 137 and agrees in all properties with the substance described by Ullmann. Diphenyl-j3-naphthylmethane, (C 6 H 6 )2(/3-C 10 H7)CH. Ten g. of the carbinol is dissolved in 200 cc. of acetic acid and the solution is boiled with 20 g. of zinc dust until a drop no longer gives a red color with sulfuric acid. The acetic acid solution is poured into water and partly neutralized. The ether extract is washed with alkali, dried and concentrated. The hydrocarbon crystallizes on the addition of alcohol; yield, 9 g., or 95%; m. p., 77-78. Analysis. Calc. for C Z3 U, 8 : C, 93.83; H, 6.17. Found: C, 93.44; H, 6.33. Mol. wt. Calc.: 294. Found: 293. 13 Baeyer and Besemfelder, Ann., 266, 188 (1891). 14 Ullmann, Ber., 38, 2218 (1905). 6 The properties of diphenyl-/3-naphthylmethane as here described differ from those which appear in the chemical literature and which are based on the work of Lehne. 15 Diphenyl-/3-naphthylmethyl Chloride, fC 6 H fi ) 2 (i3-C,oH7)CCl. The chloride is prepared by saturating an ether solution of the carbinol with hydrogen chloride in the presence of calcium chloride. After standing, the solution is decanted, boneblack is added and the ether is evaporated. The oily residue is taken up in low-boiling petro- leum ether, filtered and allowed to crystallize. The recrystallized substance melts at 94.5, and the yield is practically quantitative. The use of acetyl chloride instead of hydrogen chloride in this preparation leads to the formation of colored impurities, and is not recommended. Analysis. Calc. for CH 17 C1: Cl, 10.79. Found: 10.71. The chloride gives red additive compounds with stannic and mercuric chlorides. When dissolved in alcohol or, preferably, in an alcoholic solution of sodium ethylate, the chloride is converted to diphenyl--naphthylmethyl ethyl ether, (C6H 5 ) 2 (/3-CioH 7 )- COC 2 Hj, a colorless crystalline substance which melts at 114. The chloride gives an anilide, (CH 5 ) 2 (0-C IO H7).C.NH.C6H 6 , melting at 158.5. Diphenyl-j3-naphthylmethyl Bromide, (C 6 H 5 ) 2 (i3-CioH 7 )CBr.- The bromide can be prepared by the addition of acetyl bromide to a benzene solution of the carbinol. After recrystallizing from petroleum ether it is colorless and melts at 136. On long standing it turns red with the evolution of acid fumes. Analysis. Calc. for C 23 H 17 Br: Br, 21.4. Found: 21.6. Preparation of Diphenyl-0-naphthylmethyl. Five g. of diphenyl- 0-naphthylmethyl chloride and 5 g. of molecular silver are placed in a test- tube and 35 cc. of carbon disulfide is added. The tube is then corked, taking all the necessary precautions, and shaken for several hours. 16 ' 17 When the reaction is complete, the bright red solution is siphoned into the free-radical apparatus. The apparatus is immersed in warm water, suc- tion is applied and the solvent is completely evaporated. Five cc. of acetone is added, the oily residue is dissolved and the apparatus is filled with carbon dioxide. The solution is cooled to a temperature below 0; crystals of the free radical separate after standing for 2 to 3 days. The solvent is drawn off and the crystals washed with acetone, then dried under reduced pressure in a stream of carbon dioxide ; 2 . 5 to 3 g. of pure white or light yellow, finely divided, crystalline powder, the triarylmethyl, is thus obtained. The compound is very soluble in all the usual solvents except petroleum ether. It melts in an atmosphere of carbon dioxade between 135 and 140 to a reddish liquid. The dry substance is quite stable at ordinary temperatures and may be exposed to air for short periods without danger of oxidation. Calc. for C 23 H 17 :C, 94.16; H, 5.84. Found: C, 93.92; H, 6.02. Formation of Additive Compounds with Solvents. Owing to its extreme solubility, it is very difficult to crystallize diphenyl-/3-naphthyl- methyl and the crystallization always requires from 2 to 3 days. The 15 Lehne, Ber., 13, 358 (1880). 16 Gomberg and Schoepfle, J. Am. Chem. Soc., 39, 1659 (1917). 17 Gomberg and Cone, Ber., 37, 2034 (1904). ketones are, as a class, best suited as solvents for recrystallization, although it is possible to recrystallize the free radical from a few solvents of other classes. In order to determine whether additive compounds were formed with different solvents, the procedure already given for the recrystalliza- tion of the substance from acetone was used with the substitution in each case of the solvent whose additive tendency was to be investigated. In case the solvent was one of high boiling point, the free radical after re- crystallization was freed from the non-volatile solvent by washing with a little low-boiling petroleum ether. The solid substance was dried thor- oughly under reduced pressure at room temperature. A weighed sample of the dry free radical was then placed in a porcelain boat and heated at 70 under reduced pressure in an atmosphere of carbon dioxide. The free radical was recrystallized from the following solvents and separated without solvent of crystallization: (I) ketones: acetone, methylethyl ketone, methylhexyl ketone, dipropyl ketone; (II) ethers: diethyl ether, ethyl 250-amyl ether; (III) esters: ethyl acetate. With methylbutyl ketone, diphenyl-/3-naphthylmethyl gives an addi- tive compound in which the proportion of ketone to free radical approaches that required in accordance with the general formula of such compounds with triphenylmethyl, R 3 C CR 3 .C 6 Hi 2 O. Three samples, 1.020, 1.526 and . 481 g. showed . 073, . 175 and . 050 g. loss or 7 . 1, 1 1 . 8 and 10 . 4%, respectively, whereas the calculated loss is 14.6%. Among other solvents which were tried but from which the free radical failed to crystallize altogether were benzene, benzonitrile, caprylene and cyclohexane. Reaction with Oxygen. Diphenyl-/3-naphthylmethyl rapidly com- bines with the amount of oxygen necessary for the formation of the peroxide in accordance with the equation, 2(C 23 Hi 7 ) + O 2 >(C 2 3Hi7)O-O(C 2 3Hi7). The apparatus for the measurement of the oxygen absorption 18 consists of an absorption bottle which is connected by a heavy rubber tube to a gas buret. To determine the oxygen absorption, a weighed quantity of di- phenyl-/3-naphthylmethyl chloride and an equal weight of silver were placed in a tube, bromobenzene was added and the tube was sealed and shaken for several hours. The tube containing the reaction mixture was placed in bromobenzene in the absorption bottle, the apparatus was sealed, exhausted and filled with oxygen. When the system had come to equi- librium, the tube was broken by shaking the bottle. Shaking was con- tinued until the color of the free radical had permanently disappeared and the reading was taken half an hour later. The oxygen absorbed by the previously isolated, crystalline, free radical was also determined and, of course, silver was left out. The following table gives the results of typical experiments. 18 Ref. 17, p. 1661. 8 TABUJ! REACTION WITH OXYGEN Weight of sample Oxygen absorption C 23 H 17 C1 C2sH 17 Cc. at Stand. Cond. % of Calc. G. G. 1 1.200 ... 40.8 99.6 2 1.200 ... 41.6 102 3 1.200 ... 41.0 100 4 ... 1.029 39.3 100 5 ... 1.000 39.0 102 Peroxide. The yield of peroxide varies greatly under different con- ditions. When an ether solution of the free radical is oxidized with air the yield is 72%, while the use of benzene as a solvent reduces the yield to 46%. The peroxide when recrystallized from benzene is pure white and melts at 166. Analyses. Calc. for C4 8 H 34 O 2 : C, 89.3; H, 5.54. Found: C, 89.3, 89.6; H, 5.54, 5.50. Action of Light and of Acids. The free radical is not very sensitive to light. A 5% solution in bromobenzene which was exposed to full day- light during three days in December still absorbed 75% of the calculated amount of oxygen. When a 10% solution of the free radical in bromo- benzene was treated with an equal volume of bromobenzene saturated with hydrogen chloride, the color of the free radical rapidly disappeared and the solution, after 3 hours, absorbed no oxygen. 19 Action of Iodine. Diphenyl-jS-naphthylmethyl reacts with iodine in accordance with the following equation, 2(C6H 5 ) 2 (/3 Ci H 7 )C + 1% > 2(C 6 H 5 )2(/3-CioH5)CI but an equilibrium is reached before the reaction has gone to completion. A weighed sample of the free radical in benzene was titrated with a 0. 1 Absolution of iodine in benzene. The color of the iodine disappeared rapidly until 70% of the calculated amount was added, then the absorption of iodine ceased. The change from the color of the free radical to the color produced by excess of iodine can be readily de- tected. Three samples of 1.000 g. each required 23.0, 23.7 and 25.0 cc. of . 1 N iodine solution, or 67 . 5, 70 . and 73 . 5% of the calculated amount, respectively. The presence of the iodide in solution was demonstrated by its conver- sion to the anilide, which substance proved to be identical with diphenyl- /3-naphthylmethyl anilide prepared from the corresponding chloride. The Conductivity. For the determination of the conductivity a cell resembling the one used in the earlier work 20 was used. It differed from the other in that the electrodes were placed horizontally instead of verti- cally. The cell constant was determined using a 0.02 N potassium 19 Ref. 17, p. 1663. 20 Gomberg and Cone, Ber., 38, 1342 (1905). 9 chloride solution at 25. The sulfur dioxide was passed through a tube of phosphorus pentoxide 50 cm. long, and was then condensed in the cell, which was surrounded by a freezing mixture. The cell was then stoppered and placed in a Dewar flask which contained liquid sulfur dioxide. By such a procedure it was possible to maintain in the surrounding bath a constant temperature of 8, the boiling point of liquid sulfur dioxide. It was found that the cell could be removed from the cooling solution long enough to weigh it without generating too much pressure. After weighing, the cell was connected into the usual Wheatstone bridge appa- ratus, and the conductivity of the pure solvent was determined. In all experiments, the conductivity of the solvent was between 10 ~ 5 and 10~ 8 mhos. The material whose conductivity was to be determined was intro- duced in 3 portions. The cell was shaken after each addition until all the substance was dissolved, as indicated by the fact that the reading had become constant. Determinations were made of the conductivity of triphenylmethyl and of its chloride and bromide at 8, with the cell immersed in liquid sulfur dioxide, and at in a mixture of ice and water. The conductivities of diphenyl-a-naphthylmethyl chloride and bromide, and those of diphenyl-0-naphthylmethyl chloride and bromide were de- termined at 8 only. Contrasting with its behavior in other solvents, diphenyl-/3-naphthyl- methyl was found to be quite insoluble in sulfur dioxide. For this reason the substance was dissolved in toluene and measured quantities of this solution were added to sulfur dioxide. The same procedure was carried out with triphenylmethyl and it was found that in both cases the re- II CONDUCTIVITIES IN SULFUR DIOXIDE (C 6 H 6 ) 3 CC1 (C 6 H*) 3 CBr (C 6 H 5 ) 2 (a C 10 H 7 )- (C 6 H 5 ) 2 (-C 10 H 7 )- CC1 CBr M- 8 Mo M-g Mo M-g V M-, 1 154 24. 2 20. 2 237 113 118 2600 128 1180 138 2 43 .7 14. 11. 3 112 108 112 179 70 223 118 3 18 .2 9. 1 7. 8 39.6 97 .6 99.5 59.4 50 55.7 102 1 205 27. 9 23. 1 189 117 122 1271 117 683 124 2 61 .0 16. 6 13. 4 60.7 105 107 334 84.6 184 108 3 26 .3 11. 5 9. 2 28.2 96 ,4 97.9 46.9 44.8 46.7 95.6 (C,H 5 ) 3 C- in SO 2 + C 7 H 8 V M-g % CyHs by Vol (C.H,),(0-CioH 7 )C- (CH.)i(0-Cio- (C 6 H 6 in SO 2 + C 7 H 8 H 7 )CC1 H 7 . V /*_ g % C 7 H 8 by Vol. V p_ 8 V )CBr M-g 1 211 31.4 2.62 227 34 2.57 1072 73 .8 692 131 2 108 26.3 5.11 116 27. 1 5.02 212 41 .4 148 110 3 57 20.2 9.72 60.2 22. 9.57 30.5 20 .1 25.9 96 .5 1 195 28.6 2.81 261 39 ,4 2.24 . 892 129 2 100 26.6 5.47 133 31 4 4.39 283 42 .2 325 123 3 52.9 19.5 10.4 68.4 22. 5 8.41 27.1 23 .2 27.4 96 .6 10 suiting solutions of the mixed solvents were good conductors, much better than a solution of triphenylmethyl in sulfur dioxide alone. The readings were taken 1 hour after the additions were made. The accompanying tables and curves show the results of these experiments. L/TERS Conductivity in Hydrocyanic Acid. The same apparatus was used as in the determination of the conductivities in sulfur dioxide. The hydro- cyanic acid was generated by the addition of a concentrated solution of sodium cyanide to a 50% solution of sulfuric acid and the gas evolved was dried by calcium chloride and phosphorus pentoxide. It was then condensed in the conductivity cell. The specific conductivity of the hydro- cyanic acid was never greater than 10~ 5 mhos. It was found that the conductivity of the triarylmethyl halides in hydro- cyanic acid did not remain constant, but changed gradually. For this reason it was not possible to determine the conductivity by dissolving the solid substance in this solvent. Experiments showed that the con- ductivity of triphenylchloromethane in hydrocyanic acid to which 15% of benzene was added, was not greatly affected by the further addition of benzene up to 20%. It was found that benzene and hydrocyanic acid mixed without a change in volume. It was, therefore, decided to add the halides dissolved in benzene to hydrocyanic acid; 22 cc. of the solvent was condensed in the cell and 3 cc. of benzene was added. To this mixture was added, first, . 07 cc. of benzene containing the substance whose con- ductivity was to be determined, second, . 23 cc. more and, finally, . 7 11 cc., a total of 1 cc. of solution. The readings were taken 5 minutes after each addition and the series of three readings was always completed in 15 minutes. The conductivities were all determined at using an ice-bath. The density of hydrocyanic acid at this temperature is 0.7112. III CONDUCTIVITY IN HYDROCYANIC ACID (C 6 H 5 ) 3 CC1 (-C 10 H 7 )(C 6 H 6 ) 2 CCl V . 1659 391 121 1668 393 121 111 77 44 115 75 45 211 206 166 212 202 165 1794 422 130 1704 424 131 161 117 77 170 118 76 (/3-C 10 H 7 )(C 6 H 6 ) 2 CBr V (C 6 H 5 ) 3 CBr V i 1790 420 130 1726 406 125 (/3-C 1 oH 7 )(C 6 H 6 ) 2 CCl V Mo 166 125 86 166 122 85 The conductivities of the chlorides and bromides of triphenylmethyl, diphenyl-a-naphthylmethyl and diphenyl-/3-naphthylmethyl were de- termined in hydrocyanic acid. Unfortunately the conductivities of the free radicals themselves could riot be determined because when a benzene solution of a free radical is added to hydrocyanic acid an amorphous precipitate is formed and the solvent remains colorless and nonconducting. (a-C 10 H 7 )(C 6 H 6 ) 2 CBr V Mo 1830 200 416 197 129 174 1740 192 409 198 126 174 1690 404 125 1711 404 125 1767 382 127 1704 401 124 216 203 191 193 206 191 2. Cououcr/v/r r /v Z / TERS COA/7%/MWG/ Mot ECUL A/? Wr V Molecular Weight. The molecular weight of dipheny-/3-naphthyl- methyl was determined in 8 solvents, covering a temperature range of 12 from 22 to 80. The standard Beckmann apparatus with electromag- netic stirrer was used. Details for the manipulation for solvents freezing above have been previously published. 21 The determination of the molecular weight in solvents freezing at low temperatures is described below. A large number of solvents were tried out, including benzonitrile, dimethyl resorcin, ethylene chlorohydrin and *< ^ ^^ / /OL cCtsif Mfc* y/r MT --^ ^Qo/fQ """"""" fc . "-x CeHsA/Oz - 3 . 1 c: * , ^ .. mm CeH^c/2\ 7 HrB/- Q) 51 ^^MHM "S^: ^S 'S.HG h ^ ^ 3Z !H^=a "- -. * ~ k ^_ ^"^ ~ _ X: *" C*. C/4 $ s X $"* s> / 2 3 4 p" The Quantitative Determination of the Number of Decolorizations Produced by the Action of Oxygen on Solutions of Diphenyl-j8-Naph- thylmethyl. The fact that solutions of free radicals are decolorized by 15 shaking with oxygen and that the color reappears upon standing has been referred to in the introductory part of this paper. The approximate number of such decolorizations has been determined by several investi- gators. 25 We found that such results varied greatly with the conditions of the experiment. Therefore, a definite procedure was adopted and an apparatus was devised for the investigation of this phenomenon. The apparatus was quite similar to the usual form of absorption apparatus which has already been described, except that the absorption vessel was replaced by a lOOcc. Nessler tube. This tube was tightly stoppered and provided with a fairly close fitting cork piston which was moved by an iron rod passing through the stopper. In order to make a gas-tight joint, yet one through which the rod would slide easily, an iron tube slightly larger than the rod was inserted in the stopper and a rubber tube was fitted tightly over both the iron sleeve and the rod. By the use of a little glycerine this joint was made perfectly gas tight and yet it did not hinder the free motion of the piston. A tube connecting with the gas buret was also inserted through the rubber stopper. For the determination, a Ig. sample of the free radical was placed in a test-tube of 15cc. capacity, sufficient bromobenzene was added to fill the tube, and the tube was sealed. The substance was quickly dissolved by shaking. The tube was then placed in the Nessler tube just described, enough bromobenzene was added to make a total volume of 50 cc. and the tube was connected with the buret of the absorption apparatus. The apparatus was exhausted and filled with oxygen. The test-tube was broken by pressing down the p'ston, which was adjusted so that its under surface was in contact with the surface of the liquid, thus sealing the solution al- most completely from the oxygen above. The liquid in the gas buret was leveled and the reading was taken. The piston was then withdrawn as far as possible in order to admit a fresh supply of oxygen and the solution was shaken for a few seconds until the color of the free radical disappeared. The piston was then quickly pushed in and the absorption was determined. The same procedure was repeated at 2-minute intervals. After the last easily perceptible color change had taken place, the tube was shaken vigorously until no more absorption of oxygen occurred. The last reading represents not merely the oxygen required to produce perceptible decolori- zation but the amount necessary to complete entirely the oxidation of the free radical. The values for the percentage of oxygen absorbed at each interval are based, not on the amount which would be required by theory for the entire oxidation, but on the actual amount of oxygen which was absorbed in the experiment under consideration. This hap- 25 Schmidlin, Ber., 41, 2476 (1908). Schlenk and Weickel, Ann., 372, 7 (1909). Schlenk and Renning, ibid., 394, 194 (1912). Schmidlin and Garcia-Banus, Ber., 45, 3186 (1912). 16 pened to be somewhat less than that calculated for the amount of free radical taken. TABLE V DECOLORIZATION BY OXYGEN Sample I Weight of sample, 1.000 g. Solvent, 50 cc. of C 6 H 6 Br Temp., 21. Bar., 744 mm. % of calc. absorption, 87.3 Oxygen Total abs. for % oxygen ab- single absorp- sorption decolorization tion Interval sorpti< 1 5.1 2 9.2 3 12.9 4 16.9 5 19.6 6 22.3 7 23.8 8 25.6 9 26.7 10 28.0 Total 34.5 Sample II Weight of sample, 1.000 g. Solvent, 50 cc. of C 6 H 5 Br Temp., 22. Bar., 730 mm. % of calc. absorption, 96.7 Oxygen Total abs. for % of total oxygen single absorp- absorption decolorization tion 5.1 14.9 4.1 26.7 3.7 37.5 4.0 49.0 2.7 56.9 2.7 64.7 1.5 68.9 1.8 74.2 1.1 77.3 1.3 81.2 6.5 100 5.2 8.9 13.2 17.3 19.9 22.6 25.3 27.6 29.2 30.8 37 5.2 3.7 4.3 3.9 2.6 2.7 2.7 2.3 1.6 1.6 6.2 13.9 24.0 35.6 64.9 53.9 61.1 68.3 74.5 78.9 83.2 100 Corrected to standard conditions. Similar experiments were conducted with triphenylmethyl and di- phenyl-a-naphthylmethyl but the turbidity produced by the immediate precipitation of the peroxide obscured the end-point. Twelve clearly dis- tinguishable changes were observed with the former and only three with the latter. Such a method as here employed is suited for use only with compounds which show a decided color change on oxidation and which give a clear solution by virtue of the solubility of the corresponding peroxide, as happened to be the case with diphenyl-/3-naphthylmethyl. Colorimeter for the Study of the Changes in Intensity of Color of Solutions of Free Radicals. For the quantitative determination of the color of a free radical solution of a Campbell-Hurley colorimeter 26 was used, with such changes as were necessary in order to work in the absence of oxygen. The tube, B, which was to contain the solution under investigation was fitted with a glass plate sealed to the top and with 2 side-arms placed near the top. The lower side- arm was fitted with a ground-glass joint to which was attached a 3-way stopcock, a. One opening of the stopcock connected with a funnel for the introduction of solvent while the other led to the carbon dioxide generator. The upper side-arm was connected through a stopcock, b, to the suction pump. The other colorimeter tube, A, which was to contain the standard solution, was also sealed with a plate at the top and fitted with 2 side-arms, one as near the bottom as possible and the other near the top. The 26 Campbell and Hurley, J. Am. Chem. Soc., 33, 1112 (1911). 17 lower arm was connected by a ground-glass joint to a tube leading to the Drechsel bottle, C, which served as a reservoir for the standard solution. To the other arm of the Drechsel bottle was fused a 3-way stopcock, c, one arm of which connected with the suc- tion pump and the other with the carbon dioxide generator. This arm of the Drechsel bottle was attached also to a second 3-way stopcock, d, one outlet of which connected with the funnel for the introduction of solvent and the other led back to the upper side- arm of the colorimeter. This upper side-arm was also fitted with a 2-way stopcock, e, and trap for the outlet of carbon dioxide. The carbon dioxide for these experiments was freed from all traces of oxygen by passing over reduced copper at a dull red heat. To prepare the standard solution, . 5 g. of the free radical was placed in the Drechsel bottle, C. The apparatus was evacuated, filled with car- bon dioxide and 100 cc. of benzene was added. The free radical dissolved Fig. 5. rapidly. The apparatus was kept under a positive pressure of carbon dioxide all the time. To raise the column in the colorimeter tube, the stopcock d was closed and the stopcock e was opened. The pressure forced the solution into the colorimeter tube. To empty the colorimeter tube the stopcock d was opened and e was closed. By repeating this procedure a few times, the solute was completely dissolved and the solution was well mixed. The reliability of the apparatus was thoroughly tested. A solution in B when made up to the same strength as the standard in A 18 and compared with it was invariably balanced exactly by a column of the standard solution of the same height. The Effect of Dilution on the Color of Solutions of Free Radical in Benzene, Nitrobenzene and Cyclohexane at their Respective Freezing Points. One-quarter g. of the free radical was dissolved in 5 cc. of solvent in the colorimeter tube B. The solution was frozen by immersing in a cooling mixture and then allowed to melt until a few crystals remained. It was then compared with the standard in A and 5 readings were taken. Readings were easily duplicated within 1 mm. The solution was then di- luted by the addition of 5 cc. of solvent, frozen and compared with the standard again. Finally it was diluted to 25 cc. This procedure was carried out with each of the 3 solvents. The experiments were carried out in duplicate on solutions made by dissolving different samples of free radicals, but since the two experiments checked exactly only one is reported. TABLE VI CHANGE OF COLOR WITH DILUTION Read- ings Volume soln. Nitrobenzene % Cone. Height by wt. standard Benzene % Cone. Height by wt. standard. Cyclohexane % Cone. Height by wt. standard. Cc. Mm. Mm. Mm. 1 5 4.15 13 5.68 11.5 6.85 10 2 10 2.08 17 2.84 16 3.43 14 3 25 .83 25 1.14 23 1.37 19 The Effect of Temperature on the Color of Solutions of the Free Radical in Carbon Tetrachloride and in Toluene. A solution containing 2 g. of the free radical per 100 cc. of carbon tetrachloride was frozen in a mixture of carbon dioxide and toluene. The color at the freezing point was compared with that of the standard solution at room temperature. A number of comparisons were made throughout a temperature range from the freezing point of carbon tetrachloride, 22, to +30. Similar com- TABLE VII CHANGE OF COLOR WITH TEMPERATURE Height of solution, 17 mm. Temp. CC1 4 ' Height stand. Temp. C 7 H { C. Mm. c. -39 -26 -22 -15 10 20 30 6 15 23 33 50 -18 -10 -5 5 12 20 25 30 Height stand. Mm. 3 4 6. 9 11 13 17.5 20.5 28 36 41 19 parisons were made upon a 2% toluene solution at various temperatures, from the freezing point of mercury, 39, to -+-30. The standard so- lution for these comparisons was, as before, a 0.5% solution of the free radical in benzene. The fact that the color of the free radical solution upon cooling does not decrease beyond a certain limit, may be due to the presence of colored im- purities. This suspicion is strengthened by the fact that the solid free radical itself is colored slightly. If this surmise be correct, then the read- ings given are all somewhat too high and should be corrected by subtracting the value of the least reading, 3 mm. We were unable to obtain a solution of the free radical in carbon tetrachloride or toluene whose color disap- peared entirely even when cooled to the lowest temperature obtainable with a mixture of carbon dioxide and toluene. Discussion of Results Diphenyl-/3-naphthylmethyl is a pure white substance which rapidly becomes yellow on standing, even in an atmosphere of carbon dioxide. In contrast to its a-naphthyl isomer, it is very soluble in all the common solvents and crystallizes only on long standing. It absorbs the theoretical amount of oxygen to form the peroxide but does not give a high yield of the crystalline substance. The free radical is quickly decomposed by hydrogen chloride and slowly acted on by light. The equilibrium between the free radical and iodine is reached when 70% of the theoretical amount of iodine has been added. The concentrations of the products present when equilibrium has been reached are inter- mediate between those in the case of triphenylmethyl and in the case of diphenyl-a-naphthylmethyl. The results of the conductivity determinations show that the property of electrolytic dissociation is a general one for this class of compounds. The bromides give strongly conducting solutions with but little difference between the conductivities of the individuals. The chlorides are not such good conductors and there is a variation among them. The con- ductivities of the chlorides are related in the same way as the dissociation of the corresponding free radicals; the chloride of diphenyl-a-naphthyl- methyl, highest; of diphenyl-/3-naphthylmethyl, next; and of triphenyl- methyl, least. Triphenylmethyl chloride has a large negative temperature coefficient of conductivity. The same relation holds for the free radicals themselves; diphenyl--naphthylmethyl is a better conductor than tri- phenylmethyl. The conductivity of the triarylmethyl halides in hydrocyanic acid is uniformly higher than in sulfur dioxide. The bromides show molecular conductivities comparable to that of an aqueous solution of potassium hydroxide or about twice the conductivity of the alkali halides in water. 20 This is a striking fact, in view of the high molecular weight, 373, the great complexity of the molecules, and in view of the absence of any basic groups like the amino group. The chlorides conduct less than the corre- sponding bromides and the individual variations are the same as those ob- served for the same substances in sulfur dioxide. The colors of the solutions of the free radicals in sulfur dioxide are: triphenylmethyl, orange yellow; diphenyl-a-naphthylmethyl, green; diphenyl-/3-naphthylmethyl, red. The halides of these free radicals give the same colors when dissolved in sulfur dioxide, hydrogen cyanide or sulfuric acid. The color is very intense and a minute quantity of sub- stance suffices to impart a deep color to a large volume of solvent. Merely from a visual comparison of the solutions of diphenyl-/3-naphthylmethyl or of its halogen derivatives in sulfur dioxide with a benzene solution of the free radical itself, it is evident that the phenomenon of electrolytic dissoci- ation with the accompanying color is entirely different from the phenomena of dissociation and color formation of the free radical which takes place in non-dissociating solvents. Also, the same difference is apparent when the bright green sulfur dioxide solutions of the a-naphthyl derivatives are compared with the dark brown solution of this free radical in benzene. Kurt H. Meyer and H. Wieland 27 had previously come to a similar con- clusion in regard to triphenylmethyl and its derivatives. As a result of spectroscopic investigations, they found that the colors of triphenylmethyl solutions in sulfur dioxide and in benzene, while apparently identical, in reality were distinctly different. As a result of previous molecular weight determinations upon a number of free radicals certain generalizations have been drawn concerning their dissociation. Diphenyl-/3-naphthylmethyl conforms to the generaliza- tion that dissociation increases with dilution except in a few cases where deviations can be attributed to experimental errors. The results of the determinations of the molecular weight of triphenyl- methyl indicate that, within certain limits, the dissociation of hexaphenyl ethane is proportional to the temperature of the solution and independent of the nature of the solvent. This hypothesis was borne out by subse- quent work on diphenyl-a-naphthylmethyl which showed a very regular increase in dissociation with increase in temperature, quite independent of the nature of the solvent. In a set of experiments upon the j8-naph- thyl free radical, performed to test this point, using nitrobenzene, benzene and cyclohexane, it was found that there are exceptions to this rule. The solvents all possess approximately the same freezing points, 5 to 6. In nitrobenzene and benzene, the molecular weights are practically the same. In cyclohexane, however, the molecular weight of the free radical is very high and increases rapidly with concentration until, in a 5% solu- 27 Meyer and Wieland, Ber., 44, 2557 (1911). 21 tion, it is almost completely associated. This peculiar behavior is in har- mony with the observations of Mascarelli and Benati 28 who found that, as compared with the ordinary cryoscopic solvents, cyclohexane was very strongly associating. It might be expected that a solvent possessing this tendency would certainly exhibit it toward a free radical. The accom- panying graph shows that, in general, dissociation increases with temper- ature. The results of the experiments on the relation between dissociation and the oxygen necessary to produce decolorization proves conclusively that the color is not proportional to the dissociation. A 2% solution of the hexa-aryl ethane in ^-bromotoluene at its freezing point is about 30% dissociated. A bromobenzene solution of the same concentration at the same temperature is completely decolorized by only 14% of the oxygen necessary for complete oxidation. How are we to explain the fact that the decolorization is produced by only x /2 of the oxygen which would be necessary to oxidize the whole amount of the monomolecular form present in solution? Evidently only on the assumption that the monomolecular form exists as two tautomers, one of which is colored and which reacts more rapidly with oxygen than the colorless form. Of course, the results obtained by decolorization cannot be assumed to represent the true amount of the colored tautomer, first, because the colorless form may also be ab- sorbing some oxygen and, second, because the colorless form undoubtedly tautomerizes partly into the colored form during the process of decolori- zation. The values do certainly represent the maximum possible amount of the colored tautomer in the solution. These results differ from Schmid- lin's on triphenylmethyl in that the decolorization experiments here described are paralleled by a series of accurate molecular-weight deter- minations on a substance of appreciable dissociation, and we are therefore 28 Mascarelli and Benati, Gazz. chim. ital, 39, [2] 642 (1909). 22 in a position actually to correlate intensification of color with increase of dissociation. It was found that solutions of the free radicals in cyclohexane, benzene and nitrobenzene are of about the same intensity of color and that the intensity of color changes equally in all cases on dilution, in spite of the differences in dissociation in these solvents. Assuming the color of a 5% solution to be unity, we can calculate the relative intensity of color at dilutions down to 1% in ref- erence to this unit. Plotting these color values against the dilution, that is, the number of grams of solvent containing 1 g. of solute, we obtain curves a, 6, c, (Fig. 7), showing the relative change of color inten- sity with dilution. Similarly, we assume the degree of disso- ciation at 5% as our unit and calculate the relative degree of dissociation at other dilutions with reference to this unit. These values are also plotted against the dilution, and the resulting curves a', b f , c', show the relative change in dissocia- tion on dilution. It is appar- ent, in nitrobenzene and ben- D/L unfit *cc. COLOR //v 0/ss'w //v zene the intensity of color al- most doubles with the decrease of concentration from 5% to 1% and yet the change in dis- sociation throughout the same range is very slight. On the other hand in cyclohexane, the dissociation increases three times as much as the color. This again shows that color is independent of dissociation, that is, an increase in dissocia- tion may occur without an equal increase in color intensity. It was found that a solution of the free radical in carbon tetrachloride at its freezing point, or in toluene at 40, is only slightly colored although it is dissociated to the extent of from 15 to 20%. On gradually warming these very light colored solutions, an increase in color begins to take place between 30 and 20 and this color increases 10 times with a rise of temperature of 50 while the dissociation increases only l /2 with the same 23 change of temperature. If we assume as the unit of color intensity the color of a solution at 20 and calculate the relative color intensity at other temperatures with reference to this unit, then plot these values against the temperature, we obtain curves a, b, (Fig. 8), showing the relative increase of color with temperature. If we now take as our unit of dis- sociation the amount of dissociation at 20 we can calculate the relative dissociation at other temperatures with reference to this unit. Plotting these values against the temperature we obtain a curve, c, showing the increase of dissociation with rise of temperature. The accompanying curves show the striking differences between the change in the degree of dissociation and the change in color intensity which occurs when the tem- perature of the solution of the free radical is varied. D/SSN* C, It is not our purpose to explain in what manner the effect of dilution, or that of rising temperature, operates to disturb the equilibrium between the colored and colorless modifications of the free radical. However, our results do definitely prove that changes in color intensity take place independently of changes in dissociation. The evidence presented is convincing. This being M>, then the observed facts can best be explained by the assumption that we are dealing here with an equilibrium between benzenoid and quinonoid tautomers of the triaryl methyl, 29 and that an increase in the proportion of the quinonoid tautomer is the cause of the in- crease of color. Summary ' 1. Diphenyl-0-naphthyl carbinol has been synthesized from ethyl /3-naphthoate and phenylmagnesium bromide. A number of its de- *Cf. P. 1811. 24 rivatives were prepared. Diphenyl-/3-naphthylmethyl chloride was obtained by the action of hydrogen chloride on the carbinol. When the chloride was treated with molecular silver, the free radical, diphenyl- /3-naphthylmethyl, resulted. 2. The radical was isolated in the crystalline state. A careful study was made of its reactions with oxygen, iodine and acids, and of the effect of light. The conductivity of the free radical, diphenyl-/3-naphthyl- methyl, was determined in sulfur dioxide. The conductivities of the chlorides and bromides of triphenylmethyl, diphenyl-a-naphthylmethyl and diphenyl-/3-naphthylmethyl were determined in both sulfur dioxide and hydrogen cyanide. These compounds, especially the bromides, were found to be good conductors. The molecular weight of diphenyl-/3- naphthylmethyl was determined in 8 solvents whose freezing points covered the range from 22 to + 80 . It was found that the hexa-arylethane was dissociated from 15% to 50%. 3. Experiments were carried out to find what effect changes in con- centration of the free radical would have upon the dissociation and upon the color of the solutions. It has been found that the resulting changes in color intensity are independent of the changes in dissociation. It has also been found that the changes in color intensity which result from vari- ations in temperature are, again, not parallel to the changes in dissociation which are thus produced. These facts point to the conclusion that color formation in solutions of free radicals is not due wholly to dissociation of the hexa-aryl ethane into the tri-aryl methyl. The most satisfactory explanation of the facts is the hypothesis that in addition to dissociation we have tautomerization of the benzenoid tri-arylmethyl into the quin- onoid form. The equilibrium between the dimolecular and monomolec- cular forms on the one hand and the equilibrium between the two mono- molecular tautomers on the other hand, are not equally influenced by changes either in concentration or in temperature. This investigation was made with the assistance of The National Ani- line and Chemical Company Fellowship, and we wish to express our ob- ligations for the generous aid we have thus received. BOOK a MOV 6 1932 MOV 1932, ^ LD 2i_ 32 507573 UNIVERSITY OF CALIFORNIA LIBRARY