EXCHANGE Compound Formation in Phenol- Cresol Mixtures DISSERTATION SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIRE- MENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY IN THE FACULTY OF PURE SCIENCE IN COLUMBIA UNIVERSITY BY JACOB J. BEAVER, B.S., M.A. New York City 1921. The Jackson Press, Kingston 1921 Compound Formation in Phenol- Cresol Mixtures DISSERTATION Submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in the Faculty of Pure Science, Columbia University BY JACOB J. BEAVER, B.S., M.A. i/ New York City 1921 The Jackson Press, Kingston 1921 To W. J. S. ACKNOWLEDGMENT The author wishes to thank Professor James Kendall, at whose suggestion this problem was undertaken, for his generous and helpful advice and to express his appreciation of him both as a friend and as a teacher. The author also desires to thank the other members of the chemistry department of Columbia University for the many cour- tesies extended to him. 51723 COMPOUND FORMATION IN PHENOL-CRESOL MIXTURES 1. What was attempted? (1) The attempt was made to harmonize the results for phenol- cresol mixtures with the general rules previously formulated with respect to addition compound formation in two component systems. The six binary systems formed by phenol and the three cresols were examined by means of conductivity, viscosity and freezing-point depression measurements. 2. In how far were the attempts successful? (2) It has been shown that, in the systems studied, the com- plexes formed are of the nature of substitution compounds (rather than addition compounds) of which the average molecular com- plexity, as demonstrated by the specific conductivity, viscosity and freezing-point depression results, is not greatly different from the pure components. Under this view the observed experimental re- sults fall into line with the general theory of compound formation. 3. What contribution actually new to the science of chemistry has been made? (3a) New standards of purity have been established and more accurate determinations of their chief physical constants made for phenol and the cresols. (3b) New and exact measurements of the specific conductivi- ties and viscosities of the six binary systems formed by phenol and the cresols have been made. The freezing-point depression curves for the pure components and various binary mixtures in benzene solution have also been constructed. (3c) Additional confirmation has been given to a previously formulated and generalized theory of compound formation. COMPOUND FORMATION IN PHENOL-CRESOL MIXTURES. In a series of communications in the past few years 1 addition compound formation in a large number of binary systems has been investigated, particular attention being paid to variations in the extent of compound formation as the constituent radicals of the components were varied. In all more than 100 new compounds have been isolated and simple and definite rules regarding the rela- tive stability of addition compounds in solutions have been derived. It has been shown that the extent of addition compound formation in binary mixtures is controlled primarily by the "chemical con- trast," with the stability of complexes increasing uniformly with the differences in character (i.e., the positive or negative nature of the constituent groups) of the two compounds. In the course of the work, however, it was noted that the re- sults obtained with phenols were abnormal. For example, it was found 2 that phenol and the three cresols gave more stable compounds with the basic substance, dimethylyprone, than did dinitrophenol or even picric acid, both of which substances are more acidic than the phenols. It was further found 3 that weak acids, although of far greater acidic strength than the phenols (as shown by the dissocia- tion constants), give the more stable compounds with trichloracetic acid. To explain this discrepancy a tentative explanation, postulat- ing the existence of phenols in two tautomeric forms of widely different acidic strengths, was proposed. 1 Recently, Dawson and Mountford 2 have made a thorough inves- tigation of compound formation in systems of the type phenol- cresol by the freezing-point method. They found that of the six binary systems possible, five give evidence of compound formation, while the sixth (phenol o-cresol) exhibits formation of solid solu- tions. This result they considered as rather remarkable in view of the close chemical similarity of the components and concluded that iRendall, J.A.C.S., 36, 1222, 1722 (1914) ; S8, 1309 (1906) ; Kendall and Carpenter Ibid., 36, 248 9(1914) ; Kendall and Booge, Ibid., 38, 1712 (1916) ; 89, 2323 (1917) ; Kendall, Booge and Andrews, Ibid., 39, 2303 (1917). 2 Kendall J.A.CS.., 36, 1222 (1910). 3 Kendall, Ibid., 38, 1317 (1916). Kendall, Booge and Andrews, J.A.C.S., 39, 2306 (1917). 2 Jour. Chem. Soc., 113, 923 (1918). "the relations disclosed by the freezing-point diagrams are conse- quently not in accord with what would have been anticipated on the basis of the views advocated by Kendall." The fact that the uniform abnormality of the phenols had already been emphasized was apparently overlooked. A similar study of phenol-cresol systems by Fox and Barker 3 gave widely different results, a stable compound (phenol m-cresol) being isolated in only one of the six binary systems examined. Four of the systems show eutectic points at approximately 50 molecular per cent of the components while the remaining system o-cresol m-cresol gives two eutectic points with no well-define.d maximum between them. These discrepancies are in all probability due to the inadequate purification of the materials employed by Fox and Barker and to the general inferiority of their method. In view of the greater care exercised by Dawson and Mountford in all the details of their work, there is no reason to doubt the accuracy of their experimental data. In order to obtain more experimental data as to the nature of the compounds indicated by Dawson and Mountford, and in the hope of bringing the results for systems containing phenols into line with the rest of the previous work, these six systems have been subjected to additional examination. CONDUCTIVITY MEASUREMENTS Kendall and Gross 1 have shown the validity of the general rule that compound formation and ionization in solutions proceed in parallel. For a series of binary organic mixtures it has been estab- lished experimentally that an increase in compound formation is regularly accompanied by a similar increase in specific conductivity. Where the extent of compound formation was known to be very small (as shown by freezing-point curves of the systems), the con- ductivity was practically zero ; as compound formation increased in amount the conductivity became measurable ; where combination was extensive, the conductivity was very markedly increased, being in some cases over one hundred times that of the higher component. With these facts in mind it was decided that the first line of attack in the present work was to be the determination of the specific 3J. Soc. Chem. Ind., 37, 268 (1918). 1J.A.C.S., 48, (1921). 7 conductivity-composition curves of the phenol-cresol mixtures. As the conductivities to be measured were equal to, or below, the order of pure water (5.5 x 10~ 8 at 25 C.) it was necessary to have a cell with a much larger electrode surface than any available in order to obtain the required degree of accuracy in the experimental results. The problem was further complicated by the necessity of having the cell as small as possible so that only the minimum amount of substance would be used. To fulfil these requirements a cell was constructed using the design described by Beans and Eastlack 1 but having six concentric platinum cylinders for the electrodes instead of two as used by them. The electrodes were 2\ cm. high with the smallest cylinder 1 cm. in diameter and the others increasing in diameter in steps of two millimeters which gives the outer cylinder a diameter of 2 cm. Alternate cylinders were connected to heavy platinum wires sealed into the supporting tubes. The cylinders are separated by small glass beads attached to thin glass rods running through holes at both top and bottom of the cylinders. There are three equally spaced (i.e., at intervals of 120) rods at both ends of the cylinders. The three lower rods are bent downward and continued so that they can be sealed into the bottom of the cell. This method of holding the electrodes in place is different from that used by Beans and East- lack but was found to be necessary in order to hold the cylinders rigidly in place. This rigidity is a necessity if a permanent value for the cell constant is desired. That this end was attained is shown by the following results for the value of the cell constant: at the beginning of the work it was 0.001426; six months later it was 0.001427, a change of only 0.07 per cent (in each case the results are the average of ten determinations). With this cell all the solu- tions, with the exception of o-cresol, gave a resistance of less than 100,000 ohms, and no difficulty was experienced in obtaining dis- tinct minima. Platinized electrodes could not be used, since in the presence of platinum black phenol-cresol mixtures turn yellow, with a measurable increase in conductivity. Taylor and Acree 2 show that if the electrodes are sand blasted to insure a rough surface, the conductivities obtained with unplatinized are not measurably differ- ent from those obtained with platinized electrodes. The electrodes used in the above cell were accordingly sand-blasted, and in that con- iJ.A.C.S., 37, 2674 (1915). 2 J.A.C.S., 38, 2396 (1916). dition exerted no action on the solutions but gave perfectly definite and consistent readings. The bridge employed was a 4.7 meter circular slide wire instru- ment, well grounded and carefully calibrated. The resistance coils up to 1000 ohms were bifilar wound ; those over 1000 ohms were of the Curtis type which have practically no inductance or capacity when using a 1000 cycle alternating current. The capacity of the cell was balanced out by means of rotary air condenser connected in parallel with the resistance coils. The source of current was a Vreeland oscillator producing a pure sine wave alternating current of 1000 cycles per second. A high resistance telephone receiver was used to determine the balance point. Most of the measurements were carried out in a large Freas thermostat, regulated to 25 0.005. The absolute temperature was taken from a thermometer graduated in hundredths of a degree and standardized at the Bureau of Standards. Many of the solu- tions which have a higher freezing-point than 25 could be super- cooled sufficiently to enable direct determinations at this temperature to be made. With phenol and mixtures very rich in phenol, how- ever, solidification could not be prevented in the presence of the sand-blasted electrodes, although viscosity measurements at 25 could be made. To complete the specific conductivity curves for phenol-cresol mixtures at 25 it was therefore necessary to perform experiments for these concentrations at higher tempraturs and ex- trapolate the results. 1 For this purpose two smaller thermostats, regulated to 400.05 and 500.10 respectively, were fitted up. Linear extrapolation was assumed to be valid over the small tem- perature range involved. While this assumption is probably not strictly accurate the values so obtained show very good agreement with the direct measurements at 25, as may be seen by referring to curves 4, 5, 6 in figure I. In this diagram, the extrapolated sec- tions of the different curves are indicated by broken lines. Mixtures were made up by direct weighing with the use of a Lunge pipet. The compositions as given in the tables below are accurate to within 0.05 per cent. The specific conductivity values are relatively of the order of 0.1 per cent but no claim is made for this accuracy in the absolute values. This is due to inductance and capacity effects between the apparatus and earth, when large resist- iCompare Kendall and Gross, loc. cit., p. . . 9 ances are measured by the use of a high frequency alternating cur- rent. PURIFICATION OF MATERIALS As in the work of Kendall and Gross it was found that the presence of impurities in quantities insufficient to exert a measur- able effect on the freezing-point or boiling-point caused a very con- siderable change in specific conductivity. A constant value for spe- cific conductivity was therefore made the final criterion of purity. The method of purification was essentially the same in each case. The purest material obtainable was repeatedly fractionated from special stills made of Pyrex glass. The stills were so con- structed that the hot vapors came in contact with nothing but glass, thus eliminating the possibility of contamination from the cork. A middle fraction of constant freezing-point usually gave a product of constant specific conductivity after 3 to 6 additional fractiona- tions. 1 In view of the hygroscopic nature of the materials used and of the marked influence exerted by traces of water upon their spe- cific conductivity, measurements were restricted to cold dry days. The sensitivity of the materials toward moisture constitutes the chief source of error in the whole of the experimental work and all pos- sible precautions were taken to eliminate its effect upon the results. THE SYSTEM : PHENOL O-CRESOL A c.p. sample of phenol, after about twenty fractional distilla- tions using only the middle fraction in each succeeding distillation, gave a final product of specific conductivity 11.98 x 10~ 8 at 40 and 14.07 x 10" 8 at SO . 1 The only previously recorded value is that of Riesenfeld; 43 x lO" 8 at 43. 2 The freezing-point of this sample was 39.700.02 using a standardized thermometer. This is iden- tical with that found by Morgan and Egloff 3 for a specially prepared and purified sample of phenol. Several investigators have reported considerably higher values for phenol, 4 which seems to indicate that l lt was found that the addition of 0.2 grams of anhydrous sodium carbonate to 100 grams of phenol hastened the elimination of impurities. ir rhe fact that twenty fractionations were required to purify the material shows how difficult it is to remove the last traces of impurities. On the average 100 grams of purified material was obtained from 1000 grams of the crude material. 2 Riesenfeld, Z. physik. Chem., 41, 346 (1902). 3 J.A.C.S., 38, 844 (1916). 4 42.4 was found for a Kahlbaum sample (Kendall and Carpenter, J.A.C.S., 86, 2498 (1914). 10 the commercial product is apt to contain some impurity which raises its freezing-point. The original material here employed gave a slightly higher freezing-point than that of the final product. Pure o-cresol was obtained from a c.p. product by similar con- tinued fractionation, the final material possessing a specific con- ductivity of 0.127 x 10- 8 at 25 and a freezing-point of 30.600.02. No previous measurements of the conductivity are recorded; for the freezing-point the highest recorded value is 30.45 . 1 The data for this system are given below in Table 1. The compositions of the solutions are expressed in molecular per cent throughout. The specific conductivities are in reciprocal ohms x 10 s at 25 except as noted above. The viscosity data which are added will be discussed later. TABLE I PHENOL - O-CRESOL Mol. % Phenol Spec. Cond. x 10 8 Viscosity 0,0 0.127 0.07608 13.43 0.375 0.07835 19.73 0.415 0.07930 30.00 0.612 0.08099 38.81 0.693 0.08235 49.10 0.885 0.08404 62.83 1.686 0/08565 69.27 2.583 0.08645 75.79 3.321 0.08731 80.03 4.196 0.08757 87.30 5.422 0.08825 90.10 6.183 0.08851 100.00 8.84 0.08945 THE SYSTEM: PHENOL M-CRESOL Pure m-cresol was prepared by repeated fractionation of a c.p. sample. The effect of the distillation over sodium carbonate in low- ering the specific conductivity was very marked in the case of this substance. The specific conductivity of the final product was 1.397 x 10- 8 at 25; its freezing-point was 11.100.02. The freezing-point as reported 1 is 10.9 ; no conductivity measurements could be found. 1 Dawson and Mountford, loc. cit. iRendall, J.A.C.S., 88, 1315 (1916). Dawson and .Mountford em- ployed a product freezing at 10.0; Fox and Barker used material with a much lower freezing-point, their material apparently containing about 15 per cent, p-cresol. (See Dawson and Mountford ,loc. cit., p. 924). 11 TABLE II PHENOL - M-CRESOL Mol. % Cresol Spec. Cond. x 10 s Viscosity 0.0 8.84 0.08945 6.99 7.431 0.09105 10.79 6.810 0.09206 18.10 5.923 0.09398 26.75 5.094 0.09698 36.48 4.197 0.09961 48.26 3.379 0.1040 54.98 2.988 0.1070 61.52 2.587 0.1095 68.63 2.175 0.1131 75.51 1.887 0.1169 87.41 1.592 0.1250 100.00 1.397 0.1342 THE SYSTEM : PHENOL P-CRESOL The preparation of pure p-cresol from several presumedly c.p. samples was unsuccessfully attempted, the elimination of traces of m-cresol (which possess almost exactly the same boiling-point) 1 not being possible. To obtain pure p-cresol pure p-toluidine(F.P. 43.0) was diazotized in 6 molar hydrochloric acid solution by slow addition of the theoretical amount of sodium nitrite, the temperature being maintained below 10 and thorough mixture secured by vigorous stirring. When the reaction was complete, the temperature was gradually raised to 40 and the solution allowed to stand over night. It was then steam distilled, the distillate extracted with ether, dried and fractionated as described above. The specific conductivity of the material used was 1.378 x 10~ 8 at 25 ; it gave a freezing-point of 34.55 0.02. Practically the same freezing-point was obtained by Kendall and Carpenter 2 and Kendall 3 ; while the material used by Dawson and Mountford possessed a freezing-point of 34.15. No previous determination of specific conductivities are recorded in the literature. It was found that no matter how carefully the para isomer was purified it turned yellow in the course of a day if exposed to sun- light. The other pure substances when purified as outlined above would show no appreciable color at the end of a month. As there 1 The b.p. (under 760 mm. pressure) of m-cresol is 202.1, of p-cresol 202.5. In the cases of phenol and p-cresol or phenol and o-cresol the differences in b.p. are sufficient to permit of their separation by fractiona- tion. (See Dawson and Mountford, loc. cit., p. 937). 2 Kendall and Carpenter, loc. cit. 'Kendall, J.A.C.S., 88, 1315 (1916). 12 was no detectable impurity present it is probable that the para com- pound is considerably more sensitive to light than its homologues. The data presented below in Table III were obtained with the use of freshly-prepared material. TABLE III PHENOL - P-CRESOL Mol. % Cresol Spec. Cond. x 10 8 Viscosity 0.08945 0.09463 0.09835 0.1042 0.1099 0.1175 0.1218 0.1327 0.1474 SYSTEMS CONTAINING TWO CRESOLS In Tables IV, V and VI below, the conductivity and viscosity results for the binary cresol mixtures are given. TABLE IV . % Cresol Spec. Cond. x 10 8 0.0 8.84 12.17 7.151 24.13 5.863 36.15 4.972 47.82 4.201 61.49 3.423 68.09 3.012 84.02 2.210 100.00 1.378 0-CRESOL - M-CRESOL Mol. % meta Spec. Cond, x 10 8 0.0 0.127 11.97 0.178 23.92 0.184 36.29 0.362 49.40 0.633 60.46 0.767 64.64 0.874 69.39 0.977 84.63 1.134 100.00 1.397 Viscosity 0.07608 0.08086 0.08582 0.09208 0.09939 0.1050 0.1075 0.1109 0.1216 0.1342 TABLE V 0-CRESOL - P-CRESOL Mol. % para Spec. Cond. x 10 8 Viscosity 0.0 0.127 0.07608 12.50 0.188 0.08209 24.27 0.190 0.08854 36.96 0.344 0.09612 47.44 0.410 0.1030 57.67 0.507 0.1103 64,75 0.553 0.1163 70.91 0.601 0.1200 84.53 0.726 0.1327 100.00 1.378 0.1474 13 Mol. % para 0.0 8.65 17.33 26.27 29.29 32.67 37.62 44.79 51.77 54.80 69.97 75.39 86.12 100.00 TABLE VI M-CRESOL - P-CRESOL Spec. Cond. x 10 8 397 384 378 ,442 449 ,512 ,551 ,583 ,567 ,560 .728 ,603 ,495 1.378 Viscosity 0.1343 0.1346 0.1352 0.1360 0.1361 0.1369 0.1373 0.1385 0.1393 0.1401 0.1425 0.1432 0.1450 0.1474 The specific conductivity results given in the above tables are represented graphically in Figure I. Curves I, II, III show the data for the binary cresol systems, curves IV, V, VI the data for mix- tures of phenol with cresols. 2-0 t-0 I.METAr-PARA-C R$OL H.ORTHO-M.E.TA -CRJE^OI HLORTHO-PARA-fc RE SOL, I , I , m MOIi* PAA-CR$OLr-*- CURVE EL PHE/ (ObORTHCX:it5OL 100 14 It is quite evident from inspection of this diagram that the spe- cific conductivities of all the solutions examined are very little different from those of the pure components. This type of curve, according to the work of Kendall and Gross, 1 is characteristic of all mixtures in which the two components are essentially similar in character. In such mixtures little or no increase in molecular com- plexity through compound formation is to be expected. Hence the observed conductivity of such solutions should not differ to a marked extent (if the parallelism between molecular complexity and ioniza- tion is accepted) from that obtained by linear extrapolation from the specific conductivities of its constituents. The conductivity results obtained are therefore in apparent con- tradiction with the freezing-point results of Dawson and Mountford, and in agreement with the generalization previously formulated by Kendall and quoted above, (p. 1.) This discrepancy will be discussed more fully below, after the work on two other physical properties of phenol-cresol mixtures has been described. The investigation of the viscosity-composition curves of these systems was chosen as the second line of attack because of the marked effect of changes in molecular complexity upon the viscosity. VISCOSITY MEASUREMENTS In general viscosity curves for binary liquid mixtures fall into three distinct types. 2 The ideal curve, given by solutions in which no interactions at all take place on mixture is not linear but appre- ciably sagged. Where compound formation occurs in the mixtures, the viscosity is abnormally high ; if compound formation is extensive the curve may even exhibit a maximum. These results are in har- mony with the definition of viscosity as "essentially fractional re- sistance encountered by molecules of a solution moving over one another" 3 and with the observed experimental facts as to the in- crease of viscosity with increasing molecular weight in a homologus series. 1 Where dissociation of an associated component takes place on admixture, the viscosity (owing to the production of smaller molecules) is abnormally low; if the dissociation is extensive the curve may even exhibit a minimum. In certain cases both com- iRendall and Gross, loc. cit., p. . . 2 See Kendall and Munroe, J.A.C.S., 43, 115 (1921). 3 Kendall, Medd. K. Veten. Nobelinst., 2, No. 25 (1913). 15- pound formation and dissociation effects may be existent reproduc- ing a curve quite similar to the ideal type. 2 A brief consideration of the chemical character of the components will be sufficient to enable one to distinguish such a system from one which is truly ideal. Usually, however, one effect will predominate sufficiently to give a curve quite distinct from the normal type. It was hoped, therefore, that a study of the viscosity curves of phenol-cresol systems would either make possible a decision between the apparently conflicting conclusions drawn from freezing-point and conductivity measurements, or indicate the true conclusions. The apparatus used for the determination of viscosities was of the Bingham type 3 , the experimental procedure being essentially the same as described by Kendall and Monroe. 4 The tubes used were calibrated by means of pure m-cresol whose absolute viscosity in C.G.S. units was obtained from another tube previously standardized by means of conductivity water. The viscosity value for water was taken as 0.008946. Using this value the constants for the tube are obtained from the complete viscosity formula: t \.iz nPV (<7=aceleration due to gravity ; r=radius of the capillary ; F=trans- piration volume; /=total length of capillary; w=number of capil- laries; P=density of the liquid), which reduces to the form where C and C are the constants of the tube. The relative accuracy of the viscosity data given in Tables I to VI above is 0.2 per cent. With the apparatus employed a some- what higher precision is attainable for most substances, but the tem- perature coefficient of viscosity of the substances employed is ex- tremely high approximately 10 per cent per degree. 1 Accurate density values are not essential for the exact determination of vis- cosity with the Bingham type of instrument since the density enters iDunstan and Thole, "The Viscosity of Liquids," p. 15. 2 Compare Kendall and Brakeley, J.A.C.S., 43, (1921). 3 Bingham J. Ind. Eng. Chem., 6, 233 (1914) ; Bingham, Schlesinger and Coleman, J.A.C.S., 38, 27 (1916). 4 Kendall and Monroe, J.A.C.S., 39, 1787 (1917). ^ramley, loc. cit., p. 10. 16 into the working formula only in the kinetic energy correction- factor which is approximately 1 per cent of the total. Consequently, after it had been found for each system that the density of an approximately equimolecular mixture was, within error limits, iden- tical with the density found by linear interpolation from the specific volume-weight composition curves, no other density values were determined. 2 The viscosity results for the six systems are reproduced in graphic form in the accompanying diagram (Figure II). Inspection of these curves shows that they are all very near to the ideal type. When it is remembered, however, that we have independent evidence from other physical properties 3 that phenol and the cresols are highly associated liquids it follows that this approximation to the normal type of curve must be due to the mutual result of the 2 This assumption is in agreement with the actual density determina- tions carried out by Fox and Barker [J. Soc. Chem. Ind., 36, 845 (1917] with less pure materials. 3 See Turner, "Molecular Association"; also data given later. 17 two opposing effects discussed above compound formation and dis- sociation. It appears that neither effect is predominant here, com- bination between the components being practically counterbalanced by mutual depolymerization, leaving the average molecular com- plexity essentially unchanged. To obtain confirmation of this point of view, the molecular weights of the pure substances, and the average molecular weights of their binary mixtures, were determined in an inert solvent. MOLECULAR WEIGHT DETERMINATIONS The freezing-point of a pure liquid, on addition of an ideal solute, will be depressed according to the equation i 1 l n x=(-Q/RT } . (AT/T) (where x is the molecular fraction of solvent in the solution; Q the molecular heat of fusion of the solvent ; T and T the absolute freez- ing-points of the pure solvent and the solution respectively ; AT the freezing-point depression; and R 1.9852). If the solute is not ideal but associated, its average molecular weight in any solution of known composition can be calculated from the above equation by substi- tuting the experimentally determined value for T and solving for x. Benzene was selected as a suitable ideal solvent. The ideal freezing-point depression curve for this liquid has been given in a previous article. 2 The benzene used in the present work was dried over sodium and carefully fractionated from the same substance. The fraction used distilled between 80.2 80.3. (corr.) To determine the freezing-point depression of benzene on addi- tion of phenol and the cresols the standard Beckmann apparatus and method were employed using a thermometer calibrated at the Bureau of Standards. Undercooling was limited to 0.1 to 0.2. For the depressions tabulated in Table VII the accuracy claimed is 0.01. The "compound mixtures" were obtained by weighing somewhere near the approximate amounts of the two components into a small flask and then adding small amount of one of the components until the desired composition was obtained. In all cases the molecular com- position is within 0.1 per cent of the theoretical molecular composi- tion of the compound. iRoozeboom, "Heterogene Gleichgewichte", 2, 273 (1904). 2 Kendall and Monroe, J.A.C.S., 43, 115 (1921). 18 TABLE VII FREEZING-POINT DEPRESSION DATA Phenol Mol. % Solute AT 0.97 0.398 1.81 0.711 3.17 1.190 4.50 1.621 6.38 2.221 7.92 2.635 Mol. Ortho-Cresol % o-cresol 0.88 1.86 2.84 3.81 4.71 5.95 7.53 AT 0.591 1.170 1.752 2.257 2.728 3.341 4.082 Meta-Cresol Para-Cresol Mol. % m-cresol AT 0.67 0.434 1.54 0.925 2.00 1.161 3.10 1.711 4.33 2.161 5.43 2.584 6.01 2.843 7.72 3.424 1 Phenol : 2 m-Cresol 33.33 mol. % Mol.% compound 0.66 1.71 2.78 4.04 5.70 7.25 8.70 10.24 11.74 66.67 mol. AT 0.380 0.908 1.396 1.905 2.490 2.987 3.409 3.890 4.347 Mol. % para AT 0.95 0.590 2.01 1.180 3.18 1.691 4.54 2.225 5.70 2.625 7.10 3.102 8.68 3.569 10.33 4.065 1 Phenol: 1 o-Cresol 49.90 mol. % : 50.10 mol. % Mol.% compound AT 1.15 0.602 2.33 1.180 3.64 1.774 5.07 2.383 6.65 2.998 7.79 3.444 9.46 4.060' 2 Para: 1 Ortho 66.71 mol. % : 33.29 mol. % Mol. % compound AT 0.91 2.02 3.02 4.11 5.34 6.72 8.59 0.590 1.218 1.727 2.239 2.743 3.275 3.953 The freezing-point depression data for three of the systems (those for o-cresol, phenol and for an equimolecular mixture of phenol and o-cresol) are shown graphically in Figure III. The abnormally small depressions obtained for phenol show that this 19 substance is extensively associated. The curve for o-cresol, on the other hand, is very much closer to the ideal curve, indicating that it is much less associated than phenol. The equimolecular mixture gives a curve which falls practically midway between those of its two components. This indicates that the average molecular com- Fic; EL I. IDEAL, H.ORTHOCRESOL J2.PHfiNOL \.\. 2-0 M.OL% 40.0 4-O plexity of the mixture is of the same order as that of the pure sub- stances. Exactly the same result is obtained from the other data given in Table VII above. Each of the two mixtures tested (phenol and m-cresol in molecular proportion I to 2; o-cresol and p-cresol in the molecular proportion 1 to 2) corresponds to a definite compound isolated by Dawson and Mountford. In each case the freezing-point depression curve for the compound mixture falls intermediate be- tween the curves for the respective components, and nearer to that component in which the mixture is richer. If these phenol-cresol compounds were true addition compounds (their formation involving an increase in molecular complexity) en- 20 tirely different results would have been obtained. Although the dissociation of such compounds into their simpler components would be favored by the dilution of the mixture with a large excess of solvent benzene we should still find appreciable combination indi- cated by abnormally small freezing-point depressions. Thus, to give an example, it has been found 1 that the freezing-point depression of a solution of ethyl acetate in benzene is practically unchanged on the addition of an equimolecular amount of trichloracetic acid, 2 showing that the acid-ester addition compound formed is only slightly broken up into its components in benzene solution. Ethyl acetate and acetic acid give a similar, but much less decided increase in molecular complexity when mixed in benzene solution. In the system containing phenol and the cresols, however, the freezing- point depression curves resemble the viscosity curves in showing practically no change in molecular complexity ; any compound forma- tion between the components must therefore be compensated for by simultaneous depolymerization. In order to supply a definite idea as to the relative molecular complexities in the different solutions examined, the following table is given. As an arbitrary concentration for comparing the different systems, 5 molecular per cent has been chosen. The added specific conductivity and viscosity data will be discussed later. TABLE VIII ASSOCIATION OF PHENOL, ETC., IN BENZENE SOLUTION Solute concentration 5 molecular per cent. AT Mol.Wt. AvgMol. Spec.Cond. Viscosity Complex- x 10 8 at 25 at 25 ity phenol 1.780 178.4 1.897 8.84 0.08945 o-cresol 2.861 125.1 1.157 0.127 0.07608 m-cresol 2.450 146.9 1.358 1.397 0.1342 p-cresol 2.400 150.2 1.390 1.378 0.1474 1 phenol + 1 o-cresol ..2.342 144.5 1.427 0.979 0.08415 1 phenol + 2 m-cresol .. 2.258 153.2 1.482 2.230 0.1121 1 o-cresol +2 p-cresol. . 2.613 137.6 1.274 0.570 0.1177 The results given above are in good agreement with those of previous investigators 3 in indicating that phenol is much more highly Unpublished data obtained by Dr. J. E. Booge in this laboratory. 'Kendall and Booge, J.A.C.S., 88, 1712 (1916) show that the two substances form a stable equimolecular addition compound freezing at 27.5. 3 Beckmann, Z. physik. Chem., 2, 715 (1888) ; Aurvers, Ibid., 12, 689 (1893); Mascarelli and Benati, Gazz. chim. ital., 37, 527 (1907); 39 B, 642 (1909) ; Hewitt and Winmill, J. Chem. Soc., 91, 441 (1907). 21 associated than the cresols. The specific conductivity of phenol, it will be noted, is also considerably greater than the values obtained for the cresols a fact in exact accordance with the work of Kendall and Gross 2 in which a parallelism between specific conductivity and molecular complexity in a series of liquids of similar type was pre- dicted. O-cresol is much less associated than its two isomers, for which the association factors, like their other physical properties, are quite similar. As was to be expected from their greater association the specific conductivities of m-cresol and p-cresol are much higher than that for o-cresol. The influence of this association factor is also evident in the relative viscosity values. GENERAL CONCLUSIONS The following experimental facts have been established with regard to phenol-cresol mixtures : (a) Dawson and Mountford have isolated compounds in five of the six possible binary systems by the freezing-point method, (b) the specific conductivity-composition curves show no significant divergence from the normal curves, (c) the viscosity-composition curves are all "pseudo-ideal" in type, (d) the freezing-depression curves (which are also relative molecular weight curves) given by binary mixtures in benzene solution fall intermediate between the curves found for the pure components. As noted above the conclusions drawn from these facts are not in harmony with one another. It now remains to be seen if a rational explanation can be given which will harmonize these results with the general hypothesis correlating chemical contrast, compound formation and conductivity in solution developed by Kendall and his co-workers, or whether the whole behavior of systems containing phenols must continue to be classed as "abnormal." A careful consideration of the problem shows that an extension to these solutions of the views presented by Kendall and Gross (loc. cit., p. . .) regarding conductivity and molecular complexity in pure associated liquids, and binary mixtures of the same, supplies a complete explanation of the points in dispute. Phenol and the cresols, as already shown, are highly associated in the liquid state. The association of phenol, in particular, has been critically investigated by several investigators, 1 the conclusions 2 Loc. cit. 1 Yamnaoto, Sakuri Memorial Papers, Article 12, p. 33 (1908) ; Mascarelli and Benati, loc. cit. ; Beckmann and .Maxim, Z. physik. Chem., 89, 411 (1914). 22 reached being that the main equilibrium is represented by the equa- tion: 3(C 6 H 5 OH)^(C 6 H 5 OH) 3 . Some investigators 2 postulate the existence of still more complex molecular types of phenol. Now it has been noted by Kendall and Gross that in a solution containing two highly associated components of similar character, compounds of the general type (AB) X . (CD) y are undoubtedly formed in quantity. The average molecular complexity of such a mixture, however, (and hence its specific conductivity also) will not differ greatly from that of its pure components since disassociation of these is also involved. The complexes here' existent are to be regarded not as addition but rather as substitution compounds, for disintegration and recombination of the various molecular and ionic types present will finally result in an almost "haphazard" replace- ment of the different radicals by one another at all points of the original molecules (AB) and (CD). If the attractive forces be- tween the constituent groups are such that at a certain pafticular substitution complex is predominantly stable, then it is only logical to expect that such a complex may be definitely isolated under suit- able conditions (e.g., by freezing the solution). The phenol-cresol compounds described by Dawson and Mountford are consequently not true addition compounds as the physical properties given above sufficiently demonstrate, but substitution compounds formed by the replacement of part of an associated molecule by part of a different associated molecule. It is very significant that all five of the com- pounds isolated by the freezing-point method are trimolecular, e.g., 2 m-cresol+1 phenol, 2 p-cresol-fl phenol. Such complexes could be readily formed from the predominating phenol complex (C 6 H 5 OH) 3 by replacement of two phenol groups. No increase in the average molecular complexity of the mixture is involved in such replacements, and the apparent contradiction between Dawson and Mountford's results and those of the present work is therefore re- moved. It must also be noted that mixtures of this type, if no particular substitution complex predominates in stability, will necessarily tend to give "mixed crystals" on solidification, as in the case of the phenol - o-cresol system. Analogous behavior is shown by binary 2 Mascarelli and Benati, loc. cit. 23 mixtures of fused salts of similar character, e.g., neutral salts of the alkali metals. 1 Thus potassium sulfate gives a continuous series of solid solutions with sodium sulfate and also with potassium chro- mate. Lithium sulfate and potassium fluoride, on the other hand, form stable equimolecular compounds with potassium sulfate. These latter systems are evidentally identical in character with the five compound-producing phenol-cresol mixtures. SUMMARY By repeated f ractionation, very pure samples of phenol and the three cresols have been prepared, and their chief physical constants determined. The specific conductivity and viscosity curves for the different systems have been carefully determined. Freezing-point depression curves for phenol, the cresols, and various binary mix- tures in benzene solution have also been constructed. Without exception, the results indicate that no increase in molecular complex- ity occurs in the different systems. This is in agreement with the views correlating addition compound formation with diversity in character of the components previously formulated by Kendall and his co-workers, but apparently in disagreement with the fact that in five of the six systems definite compounds were isolated by Dawson and Mountford. A brief consideration of the equilibria existent in binary mix- tures of associated liquids has shown that the compounds present in these systems are to be regarded as substitution rather than as addi- tion compounds. Under this view the results of Dawson and Mountford and those of the present work fall directly into line with the general theory. iLandolt-Bornstein, Tabellen, 1912, p. 611-37. 24 VITA Jacob J. Beaver was born in Schenectady, N.Y., on February 23, 1893, attending the grade and high schools of that city. He received the degree of B.S. from Union University in June, 1915. From September, 1915, until June, 1917, he attended the graduate school of Columbia University, receiving the degree of M.A. in June, 1916. From June, 1917, until he resumed his studies at Columbia he was in the government service. In 1919 he was laboratory assistant in physical chemistry, in 1920 Goldschmidt Fellow in Chemistry and in 1921 lecturer in chemistry in Columbia. UNIVEESITY OF CALIFORNIA LIBRARY, BERKELEY THIS BOOK IS DUE ON THE LAST DATE STAMPED BELOW Books not returned on time are subject to a fine of 50c per volume after the third day overdue, increasing to f 1.00 per volume after the sixth day. Books not in demand may be renewed if application is made before expiration of loan period. 1930 50m-7,'29 Gaylord Bros. Makers Syracuse, N.Y. PAT. JAN. 21 ,1308 51.72(1 UNIVERSITY OF CALIFORNIA LIBRARY