EXCHANGE A STUDY OF THE SOLUBILITIES OF LIQUIDS IN LIQUIDS. THE PARTITION OF THE LOWER ALCOHOLS BETWEEN WATER AND COTTONSEED OIL. DISSERTATION. Submitted to the Board of University Studies of The Johns Hopkins University in conformity with the requirements for the degree of Doctor of Philosophy by BENJAMIN BLACKISTON WROTH 1916. A STUDY OF THE SOLUBILITIES OF LIQUIDS IN LIQUIDS. THE PARTITION OF THE LOWER ALCOHOLS BETWEEN WATER AND COTTONSEED OIL. DISSERTATION. Submitted to the Board of University Studies of The Johns Hopkins University in conformity with the requirements for the degree of Doetor of Philosophy by BENJAMIN BLACKISTON WROTH 1916. fO Gettysburg Compiler Print. Gettysburg, Pa. TABLE OF CONTENTS. Acknowledgment 4 Introduction 5 Materials 8 Procedure 9 Results 11 Tables 14 Discussion of Eesults 17 Summary 20 Biography 21 371462 ACKNOWLEDGMENT. This work was undertaken at the suggestion of Asso- ciate Professor Reid, and the author desires to take this opportunity of expressing his sincere thanks to him for his untiring and constant interest. The author also desires to express his gratitude and sincere thanks to Professor Morse, Professor Remsen, Associate Professor Reid, Associate Professor Lovelace, Associate Professor Frazier, Professor Ames, former Associate Professor Acree and the late Professor Jones for valuable instruction and assistance received during his university course. A STUDY OF THE SOLUBILITIES OF LIQUIDS IN LIQUIDS. THE PARTITION OF THE LOWER ALCOHOLS BETWEEN WATER AND COTTONSEED OIL. INTRODUCTION. According to the commonly accepted partition law a solute C is partitioned between two immiscible solvents, in contact with each other, _^_. Sa c b s b in which C a and C b are the concentrations of C in the solvents A and B, respectively, r is a constant ratio, and S a and S b are the solubilities of the solute in the two solvents. It has frequently been shown that r is constant only when the solute C exists in the two solvents in the same molecular aggregation. The equality of C a /C b = r = S a /Sb has been proved for iodine partitioned between water and carbon disulfide, bromof orm, and carbon tetra- chloride by Jakowkin. 1 His results are as follows: From solubilities in From r * , partition A. B. S a S b experiments. CS 2 H 2 O 230 -5- 0.3387 = 679 685 GHBrs H 2 O 189.55 -*- 0.3387 = 559 558-5 CCh H 2 O 30.33 -s- 0.3387 = 89.6 89.7 Of course, this law holds strictly only in the ideal case where the solvents A and B are absolutely immiscible and neither modifies the solvent power of the other in the slightest. If any appreciable amount of A dissolves in i Z. physik. Chem. 18. 590 (1895). 5 V 6 B, the solvent power of B for C is thereby altered and vice versa. Steiner 2 and Gordon 3 have shown that the solubilities of gases in water are much altered by the presence of even small quantities of salts in solutions. However, this is not contradicting the partition law, but simply saying that the value of S a and of S b should be the solubilities of C in A saturated with B and in B satu- rated with A, rather than the solubilities in the pure sol- vents. Jakowkin 4 has also shown that r changes progressively with the amount of solute C and that the value of r ap- proaches S a /S b when larger and larger quantities of C are added. When C a and C b approach zero, r approaches another limit which is more properly its ideal value, as only when C a and C b are very small are the solvents A and B unaltered by the presence of C. When larger amounts of C are present A and B become mutually more soluble and the properties of both solvents are altered. In the present work the smallest practicable amounts of C were used. Assuming C a /C b = r = S a /S b , since there are five quan- tities in this double equation, we may determine any three and calculate the other two. Thus if we measure C a and C b and thus find r we need to measure S a in order to calcu- late S b . Such results must, of course, be considered in the light of the above considerations and be accepted with reserve. The present work was undertaken with the aim of ap- plying this to the calculations of solubilities in the case of certain very soluble liquids. The results are considered as suggestive rather than as conclusive. They mean something, but just what they mean may be left for fu- ture consideration. In a series of experiments in which ethyl alcohol was partitioned between cotttonseed oil A, and water B, r was 2. Wied. Ann., 52, 275 (1894) 3 Z. physik. Chem.. 18, 520 (1895). 4 Ibid., 18, 590 (1895). found to be 28.3. The average of a series of experiments gave the solubility of alcohol in the oil, S a as 21.1 g. per 100 cc. Calculation gives S b as the solubility of the alco- hol in 100 cc. water as 600 g. There are many substances of which water dissolves several times its own weight. For instance, 100 cc. of water dissolve 339 g. of cadmium chlorate at and 549 g. at 65. In these cases, however, though the solubility is great, it is still limited. One hun- dred grams of water dissolve 907 g. of calcium iodide at but stops there, and if we shake it with 910 g., 3 g. of the salt are left over, while the 100 g. of water dissolve 600 g. of alcohol, but if more alcohol is added it too disap- pears in the solution. An examination of the phenomena shows that there is an essential difference between a solid solute and a liquid solute in contact with a solvent. The solid solute, except such substances as gelatine, does not absorb or dissolve the solvent, while the liquid solute plays the role of sol- vent and the result is two solutions. Thus, if dry ether be added to water, the ether which is not dissolved does not remain anhydrous, but dissolves a considerable amount of the water. The figure 600 obtained for the solubility of alcohol in water, is for anhydrous alcohol in contact with its solution in water. If the alcohol were separated from the water by a semipermeable diaphragm, through which the alcohol alone could pass, this condition might possibly be realized. Such numbers as these, representing a sort of ideal solubilities, are of interest for the comparison of the properties of the members of a series. They have a prac- tical value in the study of extractions. The lower alcohols have been studied in the hope of be- ing able to arrange them in a series, so as to show the variation of solubility with molecular weight. It is difficult to find a suitable liquid for the solvent A, since most liquids that are insoluble in water, dissolve too much of the alcohols. Cottonseed oil was chosen on ac- count of its insolubility in water, its non-volatility, and its accessibility. It served the purpose only fairly well f Water layer. Oil layer. Cw/Co 3 2.69 0.09 29.8 3 3-90 0.07 557 30 2.64 0.14 18.8 30 3.82 0.16 23.8 8 since it is rather difficult to handle and gives up a small amount of volatile matter when steam distilled. It mixes with the higher alcohols, so only partition ratios could be determined with these. Meyer 5 in a study of the influence of temperature upon the partition coefficient has determined the distribution ratio of ethyl alcohol between olive oil and water. His determinations were made at 3 and 30 with the follow- ing results : G. substance per 100 cc. of Mean. 42.7 21.3 The wide variation in his results is due to the fact that he determined the alcohol in the oil layer by difference, throwing all the error on the smaller quantity. His results at 30 approximate those obtained in the present work for the similar cotttonseed oil. MATERIALS. Cottonseed OIL The ordinary oil was purified by dis- tilling with steam for about two hundred hours. When the density of the distillate approached the density of pure water, the oil was assumed to be ready for use. A small amount of alkali in water was added to neutralize and dissolve any free acid, and the oil filtered through a dry paper in a hot water funnel. Later in the work it was found that Wessen oil required only one day of distilling with steam before being ready for use, and it was then used in the place of the ordinary commercial oil. Ethyl Alcohol. The ethyl alcohol was refluxed with lime for a period of several days and distilled. It had a density of 0.78543 25 / 4 corresponding to 99.92% alcohol. Methyl Alcohol. Difficulty was experienced in dehy- 5 Arch. exp. Path. Pharm., 46, 344 (1901). 9 drating methyl alcohol. It was first refluxed with lime for several days, then with anhydrous copper sulf ate, then finally with metallic calcium. Density 0.79580 25 / 4 cor- responding to 99.95% alcohol. Propyl, Isobutyl and Isoamyl Alcohols. All three alco- hols were refluxed with metallic calcium for several days and assumed to be practically anhydrous. Their densi- ties were 0.80715 25 / 4 , 0.79949 25 / 4 and 0.81225 25 / 4 , respec- tively, not corrected for air displacement. PROCEDURE. Weighed amounts of oil and water were placed in a 200 cc. glass stoppered bottle. With the lower alcohols sev- eral volumes of oil were used to one of water, so as to in- crease the alcohol in the oil layer. To this mixture a quantity of alcohol, weighed from a pycnometer, was added. A piece of sheet rubber was stretched over the stopper and securely tied around the neck of the bottle. The bottles were then placed in the shaking machine and shaken for one hour in a constant temperature bath at 25. The bottles were completely submerged in the bath during the shaking. It was found that it required three weeks' standing at 25 for the layers to separate clear. To obviate this delay, the bottles containing the mixture were centrifuged at 1350 revolutions per minute until the layers became clear. The temperature of the centrifuge was not regulated but was usually not far from 25. It was found to require about six hours' centrifuging for the ethyl and methyl alcohol mixtures while propyl, isobutyl and isoamyl required about three hours. The bottles were then placed in the 25 bath until ready for analysis. Estimation of Alcohol in Oil Layer. The oil layer con- taining the alcohol was drawn off by means of a special pipet, shaped something like an Ostwald pycnometer. A suitable amount of the oil was weighed out of the pipet into a Kjeldahl flask, out of which it was distilled with steam. The flask was provided with an efficient trap to prevent splashing over. The receiver was provided with 10 a cork having a small opening, and was placed in an ice bath to diminish evaporation. The alcohol in the weighed distillate was estimated either by the usual density method or by the interfero- meter. For the interferometer work a table was pre- pared for each alcohol. Small weighed bulbs were filled with the anhydrous alcohol, sealed and weighed. These bulbs were broken in glass stoppered bottles containing weighed amounts of water. The interferometer readings of a number of these mixtures were plotted against the percentages of the alcohols. From this a table was cal- culated, which was used in determining the composition of the distillate. The interferometer readings with weighed mixtures of alcohols and water are given in the following table : Methyl Alcohol. CH 3 OH% 0.821% 1,50 3.99 Reading 0.62 1.08 2.87 Reading for i% 0.75 0.72 0.72 Propyl Alcohol. CH 7 OH% 1.84% 2.67 3-39 4.24 Reading 5.75 8.57 11.07 14.48 Reading for i%...3.i2 3.20 3.26 3.41 Isobutyl Alcohol. C4H 9 OH% 1.89% 2.95 3.18 4.21 Reading 7.31 11.30 12.17 16.40 Reading for i%..3.86 3.82 3.82 3.89 Isoamyl Alcohol. C 5 H W OH% 1.42% 2.31 Reading 5.06 9.72 Reading for i% 4.19 4.20 When cottonseed oil is distilled with steam, no matter how long the process is continued, the distillate always contains a small amount of something which changes its density slightly and also gives a reading in the interfero- meter. The composition of this distillate remains nearly constant for a long time. The densities of the alcoholic distillates were corrected for this by taking densities of 11 distillates obtained with the same lot of oil containing no alcohol. In the interferometer work the steam distilla- tion of the sample of the oil layer was continued for a con- siderable time after all of the alcohol had passed over. A specimen of the distillate collected at the end of this distillation was used, instead of pure water, in the other cell of the interferometer. Since the interferometer is a differential instrument, the reading thus obtained rep- resented the amount of alcohol present. A weighed portion of the filtered water layer was dis- tilled in the Kjeldahl flask and the alcohol determined as above. For the experiments with ethyl alcohol the pycnometer was used and for the other alcohols the interferometer for estimation of the alcohol. The absolute solubilities of ethyl and methyl alcohol in the oil were found by adding a weighed amount of alcohol to a weighed amount of oil in a small separating funnel. The mixture was then shaken in a constant temperature bath at 25 for two hours. The funnels were allowed to remain in the bath until the layers became clear. A sample of the oil layer was then drawn off into a Kjeldahl flask and steam distilled, the amount of alcohol being determined as in the above experiments. The funnel was weighed before and after to get weight of sample. Since alcohol evaporates from this saturated oil solution very readily, much care was taken to effect this transfer with the least possible ex- posure to the air, but even then it was difficult to obtain concordant results. Weighed portions of the alcohol layers were evaporated in wide-mouth weighing bottles on a steam bath. The film of oil gained weight on long heating. This materi- ally interfered with the accuracy of the results. RESULTS. The results are given in the table below. The experiments are given in the order in which they were made and none are omitted except preliminary ex- periments at beginning of first two series. 12 In Column 1 is given volume of oil, which was obtained by dividing the weight of oil put in by 0.922, the density of the oil used. In Column 4 is given weight of alcohol for the total oil layer, calculated from the analysis of a weighed portion of the oil layer, while in Column 5 is the amount of alcohol for the total water layer found in the same way. The sums of these amounts of alcohol are given in Column 6. These sums should equal the amounts put in, which are given in Column 3. The amounts found are usually less, indicating loss of alcohol some- where in the operations. It is likely that most of this alcohol is lost in handling the water layer as this had to be filtered. Excess of alcohol found over that put in may be explained by error in pycnometer estimation or by presence of volatile matter other than alcohol in the dis- tillate. The distillates from the oil layer were usually about 60 cc. so that the total alcohol present amounted to only about 2% in the distillate. Columns 7 and 8 contain the amounts of alcohol calcu- lated from 100 cc. of oil and water, and the ratio of these is given in Column 9, as the partition ratio for alcohol be- tween oil and water. As the greatest error seemed to be in the handling of the water layer, the ratios in Column 10 were also calcu- lated, assuming that the values in Column 4 for alcohol in oil layers are correct and estimating the alcohol in water layer by difference, i. e., the difference between the values in Columns 6 and 4. The values calculated in this way are usually higher and are somewhat more regular. The values of the partition ratios have not been calcu- lated by difference in the case of isobutyl and isoamyl alcohols since the amounts of these alcohols in the two layers are more nearly equal, and the probable errors of estimating the alcohols in the two layers were nearly the same. Some results which appear erratic are in parentheses and have not been used in making the averages. 13 I i o +s + O 2 O O X 1 - 1 HH d o* d d o d o 6 .S -u ~u q q q q q S o o d d d d Cott HH*O O C Rod vd NO -; I -Sri d d d d d o" d !^o *-; O ^00 ""> *> i > K 2 s ? 2 . 10 f5 ON O txOO *o J> oo ob 5 ^ 00*00 o? 14 - -.a > S *> ^ ^QQ <^ q ffi ci ro oi oi ci ro co R< ON CO Q* -* txOO OO OO 15 * +1 !* U o w > .S ' .ts vd "> vd vd vd vd o M 6 *EON~ 01 tr io\o co G 01 ooo CO i w o >, -Bo O ^1Hllfl| 5 ffi Q 3 ! Qu OO 0\00 *S.a .K 3 . 01 01 10 10 O M n: f? o Tt- q M 01 O OJ ro ^O ro i-O ro GO 00 OOOO OOOO 16 SO 3| ffi SQ8 E 8 U c^S O 04' CO 04 04 w o HH I *m 73 : O ^jffi J ^83 ! E8 g u~ o ffi , U ^g-HS- CO V 1O ID H-i O ON IH u H * o 6 M o o K as tX O\ t^ IO & 1 ! K c o J- to H o' ON C 8 i "ffi U CO CO OJ 04* 04* CO I K Tj to txOO ON 04 *O *"^ \O COOO *^ co O - ' TJ- OJ HH Tf- IO tX O tX tX tX tX tX tX ..ioqoq >a?oooo oo o?oo K > 00 00 00 00 00 00 DISCUSSION OF RESULTS. The average results are brought together in table, j Partition Ratio Between Water and Cottonseed Oil. ] Alcohol. Methyl. Ratio 103.6 Square root of ratio .. 10.18 ^ 9 5 6 I 7 $ 6 Ethyl. Propyl. Isobutyl. IsoamyL 28.3 6.41 1.70 047 5.32 2.53 1.30 0.68 t 5' 012345 Number of Carbon Atoms in AJcoho/s. 17 18 The ratios decrease rapidly as the number of carbon atoms in the alcohol increase, each ratio being approxi- mately one-fourth of the preceding ratio. Taking these ratios as fractions, the numerators which represent the solubilities of the alcohols in water, in- crease as number of carbon atoms decrease, while the de- nominators, or the solubilities of the alcohols in the oil increase in the opposite directions. For this reason the square roots of these ratios have been plotted against number of carbon atoms in the curve. (See page 17). The solubility of methyl alcohol in cottonsed oil appears to be 4.84 g. per 100 cc. while for ethyl alcohol it is 21.2 or about four times as great. We have 4.84 x 103.6 = 505 and 21.2 X 28.3 = 600. From this 505 g. methyl alcohol and 600 g. ethyl alcohol should dissolve in 100 cc. water. These represent, in the same approximation, what may be termed ideal solubilities. At any rate the numbers 4.84 and 103.6 represent something real as de- termined by experiment and their product must have some meaning also. Further work will have to be done before these results can be satisfactorily interpreted. It is surprising that methyl alcohol gives the smaller num- ber 505, while ethyl alcohol gives the larger number 600. It was expected that methyl alcohol would give the larger number as the fact that methyl alcohol is more difficult to salt out of a solution than ethyl, seems to indicate a greater affinity for water and hence a greater solubility in water. The difficulty may be in the numbers 4.84 and 21.2, the directly determined solubilities of the two alcohols in the oil. These do not represent the amounts of alcohol taken up by the oil when in contact with the pure alcohols, but the amounts in the oil when the oil is in contact with solu- tions of the oil in the two alcohols. If we consider mole- cular solubilities, the order is what we should expect, since 505 -f- 32 = 15.8 and 600 -4- 46 = 13.0. Taking the higher and more concordant values we find that 100 g. of methyl alcohol dissolve 8.45 g. of cottonseed oil and 100 g. ethyl alcohol, 11.75 g. Experiments showed that propyl, isobutyl, and isoamyl 19 alcohols mix with cottonseed oil in all proportions. This prevented the determinations of their direct solubilities. The solubilities of isobutyl and isoamyl alcohols in 100 cc. water are given in the tables as 9.55 180 and 2.67 220 . Ac- cording to the above reasoning, the solubilities of these two alcohols in the oil should be 9.55 -H 1.70 = 5.6 and 2.67 -H 0.47 = 5.6. This shows that we must be guarded in drawing conclusions and that there are factors which are not taken into consideration in the reasoning as stated above. These alcohols, when they contain a few per cent, of water, do not mix with the oil. The phenomena of mu- tual solubility are quite complex and require much fur- ther study. This work is being continued in this labora- tory. SUMMARY. 1. The partition ratios of methyl, ethyl, propyl, iso- butyl and isoamyl alcohols between water and cottonseed oil at 25 are found to be 103.6, 28.3, 6.41, 1.70 and 0.47, respectively. 2. These are found to change regularly with increased number of carbon atoms. 3. The solubilities of methyl and ethyl alcohols in cottonseed oil are 4.84 and 21.2 g. per 100 cc. oil. 4. The attempt has been made by using these num- bers and the partition law to calculate the ideal solubili- ties of the lower alcohols in water. 20 BIOGRAPHY. Benjamin Blackiston Wroth was born November 10, 1887, in Chestertown, Maryland. He received his early education in the public schools of Chestertown. In the autumn of 1904 he entered Washington College, Mary- land, from which he graduated in 1908 receiving the de- gree of Bachelor of Arts. After graduation he taught for four years in the grammar and high schools of this State. In 1912 he entered Johns Hopkins University as a graduate student in Chemistry. His subordinate sub- jects were Physical Chemistry and Physics. 21 UNIVERSITY 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 $1.00 per volume after the sixth day. Books not in demand may be renewed if application is made before expiration of loan period. PAT. JAN. 21 ,1908 371462 > UNIVERSITY OF CALIFORNIA LIBRARY