Art. LVI.—The Molecular Complexity of the Fatty Acids and their Derivatives in Phenol Solution.

Transactions and Proceedings of the Royal Society of New Zealand, Volume 35, 1902, Page 452

Art. LVI.—The Molecular Complexity of the Fatty Acids and their Derivatives in Phenol Solution. By P. W. Robertson, Sir George Grey Scholar, Victoria College. Communicated by Professor Easterfield. [Read before the Wellington Philosophical Society, 18th March, 1903.] Plates XLVIII. and XLIX. Phenol, in spite of certain drawbacks, is a popular solvent for the determination of the molecular weight of compounds by the well-known Raoult's method. Not only is it cheap and easy to obtain in a pure state, but it also has a convenient melting-point (41°). Further, it has a large molecular depression, and the magnitude of this quantity increases the accuracy of the determinations. Indeed, Easterfield and Bee* Trans. N.Z. Inst., 1901, 497. have shown that results of the required accuracy can be obtained by the use of a common thermometer and a test-tube. At the proposal of Professor Easterfield I examined the association of water in phenol solution. From this I set about examining other compounds; my experiments with the fatty acids I propose to describe in the following communication. First, however, I shall consider shortly the main points in the history of the cryoscopy of phenol. Raoult in his historical researches gave the value of the molecular depression as 67-5. In 1890 Eykman altered the value to 72. With a view to clearing up this apparent discrepancy, Juillard and Curchod† Bull. Soc. Chim. [3] 6, 237. investigated the question, and came to the conclusion that the compound has two distinct values, 76 and 68-5, for the so-called constant. The reason of this is not far to seek, for the value 68 was obtained for associating compounds—e.g., water, alcohols, and acids—while the higher number resulted when substances known to be normal were employed. The experiments below, however, show that even for one class of organic compounds the results for the value of the molecular in dilute solution (when the freezing-point is depressed 1°) vary from 82 to 64—i.e., nearly 25 per cent.

on the mean. Van't Hoff's law, of course, only applies to infinitely dilute solutions, but even then the results appear to be little or no more in agreement. Thus, for a fall of 0–2° stearic and lauric acids give by a slight interpolation the numbers 81 and 66, and these acids are both normal fatty acids. Calculating. the results from Van't Hoff's equation, D=.02T2/w, where T is the melting-point on the absolute scale and w the latent heat of fusion, we arrive at the result 69. Using, on the other hand, the expression proposed by myself,* Trans. N.Z. Inst., 1901, 501. D = .0087 MT/3√M/d (for mono-derivates of benzene), the value of D becomes 54, which is abnormally low.† Other phenolic compounds give concordant results with those obtained by actual experiment. Thus for the following substances the calculated values are given, the equation being for multi-substituted benzenes, D =.0078 MT/3√M/d Parabromphenol 99 (experimental 98) Orthonitrophenol 74 (" 74) Thymol 70 (" 74) The possibility here presents itself, however, that this irregularity may have some connection with the varying values obtained for the molecular depression. About 1897 a large number of interesting cryoscopical researches were carried on by the Italian school of chemists. Paternò examined, among other solvents, the behaviour of phenol for many compounds, his aim apparently being to examine substances of all types and not to confine himself to any one particular group. From his observations he concludes that phenol is different from most solvents in that there is a tendency for the molecular weights to decrease with the concentration, whilst in general the reverse is found to be the case. The apparatus used in these experiments is essentially that described by Easterfield and Bee,‡ Trans. N.Z. Inst., 1901, 497. only the stirrer is of glass and a side tube is added in order to introduce the substance under investigation more conveniently. Of course, a more accurate thermometer is employed: the instruments used in the investigation could be accurately read to a hundredth of a degree. The mode of experimenting is as follows: 10 to 20 grams of phenol are weighed into the apparatus and two readings of its freezing-point taken. Enough of the foreign substance is now introduced either from a small weighing-tube if a solid or from a Sprengel pipette in the case of it being a liquid to cause a depression of about three-quarters of a degree. After

complete solution a double determination of the temperature of solidification is made, and this process is repeated six or seven times if the solubility permits it. When the dissolved substance is only slightly soluble it is introduced in smaller quantities, and consequently in such cases the accuracy cannot be so great. Besides the error of temperature-determination, which does not exceed 1 per cent., there are several others of more or less moment. Phenol is slightly hygroscopic; but, as in all cases the conditions are practically the same, this source of error, which at the most is only slight, can hardly influence the results. Perhaps the most serious error is that caused by the non-homogeneity of the solution which is being experimented with. Thus, the solute, especially if a liquid, may stay on the thermometer in small quantities above the solution; the solvent also has a tendency to climb up the stirrer and thermometer and sublime to the upper parts of the apparatus. The errors in weighing can be entirely neglected. But in a series of determinations these errors tend to disappear or neutralise one another, and for this reason the rate of association is a more accurate result than the initial molecular depression. Moreover, traces of impurity will, of course, affect the latter, but seem to have little influence on the former. Nevertheless, every effort was made to obtain the material as pure as possible, and different samples were employed where conveniently possible. As an example of the accuracy obtainable in the work I may cite the case of lauric acid. Three determinations gave the following numbers for the rate of association (depression 5°): 58 per cent., 62 per cent., 60 per cent, (different sample of phenol). Most of the fatty acids employed in the investigation were from Merck; acetic acid was frozen several times, and the lower members were fractionated before use. Normal valeric and methylethyl acetic acids were synthesised from ethyl malonate; but, owing to the small quantities at my disposal, their purity cannot be so fully guaranteed. Lauric and stearic acids showed their purity by their melting-points. A number of the other acids were also from Merck. Professor Easterfield kindly placed his own private stock of preparations at my disposal. Among others I prepared the following acids: bromacetic, cinnamic, ethylmalonic, brombutyric, oxy-diphenylacetic, nitrosovaleric, dibrommethylsuccinic, anilido-acetic, toluidoacetic, phenylanilidoacetic, amidoacetic, acet-amidoacetic, mesaconic, adipic, and phenoxyacetic, all of which were used in the investigation. In order to show clearly the method of calculating the results I will give one example in full; in the other cases the final results alone are tabulated.

Table I. Stearic Acid* Owing to its slight solubility a depression of only about 1 ½° could be reached. (molecular weight, 284). 1. Weight of phenol=10-6 grams. Fall in F.P. (d). Weight of Acid. Molecular Depression. ·59° ·2290 76·5° ·395° ·1736 68·5° ·56° ·2906 58° i.e., association increases 1/76.5—1/58 for a fall of 1–07° (.59/2+.395 +-56/2) ∴ increases 147 per cent. for a fall of 5°. 2. Weight of phenol=14.5 grams. Fall in F.P. (d). Weight of Acid. Molecular Depression. ·40° ·2168 76° ·31° ·1888 67° ·22° ·1454 62° ·53° ·3594 60° i.e., association increases 1/76—1/60 for a fall of .99° ∴ increases 135 per cent, for a fall of 5°. Mean increase, 141 per cent. The molecular depression is calculated from the equation D = 284XdXW/100m where W and m are the weights of the phenol and stearic acid respectively. The experimental results are given in detail in Table II. A number of these results are represented graphically, the molecular depression being plotted against the concentration (see Plates XLVIII. and XLIX.). The different acids will be considered under the following heads:— 1. The fatty acids. 2. The mono-, di-, and tri-substituted acetic acids where the substituents are phenyl, alkyls, and halogens. 3. The dicarboxylic acids. 4. Oxy and nitroso acids. 5. The substituted amido acids. 6. Acids whose molecular depression increases with the concentration. 1. (a.) The Normal Fatty Acids (Nos. 1–8, Table II.).—The rate of association alternately increases and diminishes as the series is ascended, each of the even members associating more rapidly than either of the two contiguous odd acids. If alternate members are considered—e.g., the even—it is found that the rate slowly decreases, reaches a minimum probably at C6, and then increases extremely rapidly. From considerations in the next paper, where cases of this nature

are discussed, it appears probable that the acids with an odd number of carbon atoms will behave in the same way. This peculiar behaviour is shown by several of the properties of the fatty acids, including their melting-points and the differences of their boiling-points. Biltz found nothing of this nature when he determined the rates of association of the alcohols in benzene solution. It is noteworthy, however, that neither the melting-points nor the differences of the boiling-points of these compounds show a behaviour like that of the aliphatic acids. Table II. Acid Class considered under Percentage Rate of Association for Depression of 5°. Molecular Depression for 1°. 1. Acetic (C2) 1 33 71 2. Propionic (C3) 1 18 70 3. Butyric (C4) 1 20 69 4. Valeric (C5) 1 16 67-5 5. Hexoic (C6) 1 18 67 6. Heptoic (C7) 1 13 66 7. Lauric (C12) 1 60 64 8. Stearic (C18) 1 141 70-5 9. Isovaleric 1 25 72-5 10. Methylethylacetic 1 and 2 12 72 11. Chloracetic 2 20 71 12. Bromacetic 2 17 72 13. Brombutyric 2 11 77 14. Phenylacetic 2 20 69 15. a Chlorphenylacetic 2 4 71-5 16. Trichloracetic 2 2-5 68 17. Ethylmalonic 3 44 72 18. Methylsuccinic 3 67 66 19. Sebacic 3 23 70 20. Dibrommethylsuccinic 3 16 21. Lactic 4 23 22. Mandelic 4 32 70 23. Benzilic 4 7 69-5 24. Oxydiphenylacetic 4 5-5 73 25. Nitrosovaleric 4 50 63-5 26. Anilidoacetic 5 24 69-5 27. Hippuric 5 and 6 —20 72-5 28. Levulinic 6 —2 82

The results obtained for the initial molecular depression, on the other hand, do not show this wavy character, but diminish regularly for some distance, apparently reaching a minimum value in the neighbourhood of lauric acid, and then increasing at about the same rate, or perhaps slightly faster. The rate of association was shown to increase much more quickly than it diminished for the lower members. 1. (b.) The Isomeric Fatty Acids.—When the chain branches beyond the a carbon the rate of association increases: this is shown in the case of isovaleric acid thus— Isovaleric. CH3 CH3 CH · CH2 · CO2H Butyric. CO2H · CH2 · CH2 · CH3 25 20 Normal Valeric. CH3 · CH2 · CH2 · CH2 · CO2H 16 The numbers placed beneath the acids are their rates of association. The influence of such a grouping probably becomes less as it is further removed from the carboxyl. In the acids with one of the a hydrogens replaced by an alkyl group, such as methylethylacetic acid, the rate of association is found to be less than that of the corresponding normal acid. That this behaviour is general is confirmed by the results obtained for other disubstituted acetic acids, which are discussed under 2. Scanty as this evidence is, there is presented a simple method for discriminating the isomers of higher members. Compared with chemical methods it possesses two great advantages—namely, quickness of execution, and the small quantity of material required for the experiment. 2. (a.) Monosubstituted Acids.—In these acids the halogens, methyl, ethyl, and phenyl have practically the same influence, the numbers varying from 17 in the case of bromacetic to 20 for phenylacetic acid. Thus the negative groups behave like the positive; if the substituents are of the same nature the heavier group appears to have the greater influence—e.g., in the case of chlor- and brom-acetic acids, the values of which are 20 and 17. On the other hand, methyl-acetic acid associates more slowly than ethylacetic. 2. (b.) Disubstituted Acids. — Dichloracetic acid unfortunately decomposes in the presence of phenol, giving a solution with a beautiful fluorescence, transmitting light of greenish hue and reflecting red. The only dialkyl acid examined was methylethylacetic, which associates at the rate of 12 per cent.—i.e., slower than either of the acids from which it is derived.

Propionic. Methylethylacetic. CH3 · CH2 · CO2H CH3 · CH · CO2H 18 Butyric. CH3 · CH2 CH3 · CH2 · CH2 · CO2H 12 20 On replacing the methyl in this compound by bromine the rate of association remains practically the same: this was also found in the case of bromacetic acid. Propionic. Bromacetic. CH3 · CH2 · CO2H Br · CH2 · CO2H 18 17 Methylethylacetic. Bromethylacetic. CH3 · CH2 · CH(CH3) · CO2H CH3 · CH2 · CHBr · CO2H 12 11 Whereas the phenyl group has an effect approximately equal to an alkyl or a halogen in the monosubstituted acids, its influence is strongly normalising in the diacetic acids. It is interesting to compare the rates of association of the following acids:— Phenylacetic. Butyric. C6H5 · CH2 · CO2H C2H5 · CH2 · CO2H 20 20 Chlorphenylacetic. Chlorethylacetic. C6H5 · CHCl · CO2H C2H5 · CHCl · CO2H 4 (12) The number in brackets is calculated by analogy from the similarly constituted brombutyric acid. Diphenylacetic acid also probably associates very slowly, as its oxy-derivative shows the low value of 5–5. Thus we see that when one of the groups in a disubstituted acetic acid is a phenyl the acids associate much less rapidly than the compounds with alkyl or halogen groups. The explanation of this perhaps lies in the stereo-chemistry of the phenyl in the molecule. At any rate, it renders the a hydrogen which causes the association in these acids practically inactive. It is interesting to note that all the acids of this type have a high value for the initial molecular depression. 2. (c.) Trisubstituted Acids.—As would be expected, tri-chloracetic acid possesses a rate of association of only 2–5 per cent. The only other acid of this class in the list is benzilic acid, (C6H5)2 · COH · CO2H, whose association (7 per cent.) is no doubt increased owing to the action of the hydroxyl group (see under 4). 3. The Dicarboxylic Acids.—These acids, as well as the tricarboxylic acids, with their halogen and oxy derivatives, are

characterized by their sparing solubility. On introducing an alkyl group, however, the compounds are rendered much more soluble, and can therefore in most cases be examined (see under “Solubilities,” p. 463). Of the groups examined, carboxyl exerts, in general, the most influence, always increasing the rate of association of the monocarboxylic acid from which it is derived. The magnitude of the change is clearly seen in the case of the isomeric acids—ethylmalonic and methylsuccinic. In ethyl malonic the two carboxyls are attached to an isocarbon atom, but in methylsuccinic acid one is united to an iso and the other to a carbon united to two atoms of hydrogen. As would be expected, the association is more rapid in the latter case. In sebacic acid the carboxyls are much further separated, and the mutual influence of the two groups is thus less noticeable. Nonylic. Sebacic. H(CH2)8 · CO2H CO2H · (CH2)8 · CO2H (11) 23 If the a hydrogens are replaced, the mutual effect of the carboxyls is much diminished. This is seen in the case of dibrommethylsuccinic acid, which associates only at the rate of 16 per cent. Methylsuccinic. Dibrommethylsuccinic. CH2 · CO2H CBr2 · CO2H CH3 · CH · CO2H CH3 · CH · CO2H 67 16 Oxy and Nitroso Acids.—In phenol solution the molecular depressions of hydroxyl compounds tend to increase with rising concentration; but a hydroxyl group increases the association of an acid. From this it seems probable that there is some mutual influence between the two groups, as a diminution of the rate of association is to be expected. This being so, there is reason to expect that the hydroxyl would have less and less effect the further it is removed from the carboxyl group.

Of the acids examined it is interesting to compare lactic and mandelic with the compounds from which they are derived. Propionic. Lactic. CH3 · CH2 · CO2H CH3 · CH(OH) · CO2H 18 23 Phenylacetic. Mandelic. C6H5 · CH2 · CO2H C6H5 · CH(OH) · CO2H 20 32 The greater magnitude of the increase in the case of mandelic acid is especially noteworthy when we remember that all the disubstituted acetic acids containing a phenyl radicle are characterized by their extremely low rate of association. That the high value in the case of mandelic acid is due to the remaining a hydrogen is evident from the fact that benzilic acid, in which this hydrogen is replaced by phenyl, has the low value of 7 for its rate of association. Thus it seems that the oxy group protects the hydrogen from the normalising influence of the phenyl; other groups, however, have quite a different effect. Thus— Chlorpropionic. Chlorphenylacetic. CH3 · CHCl · CO2H C6H5 · CHCl · CO2H (13) 4 Lactic. Mandelic. CH3 · CH(OH) · CO2H C6H5 · CH(OH) · CO2H (23) 32 When the hydroxyl occurs in the benzene nucleus of a substituted fatty acid its influence is only slight. This is shown for othooxydiphenylacetic, which associates at about the rate that we should expect in the case of diphenylacetic acid. Nitrosovaleric acid possesses the constitution CH3 · C · CH2 · CH2 · CO2H NOH It associates very rapidly, but it differs from most of the acids with high association numbers in the fact that it possesses a very low value for the initial molecular depression. Again, it is the only acid of which the rate of association clearly diminishes with the concentration. (See Plate XLIX.) Hence it seems probable that the associating groups disappear in the process of association. This favours the possibility of the formation of complexes of the type

If, on the other hand, two molecules united through the nitroso groups a dibasic acid would be formed, and then the association should be expected to increase with the depression of freezing-point. 5. Substituted Amido Acids.—The a amido acids are characterized by extreme insolubility in phenol. First it was thought that this was due to the internal combination of the acid and basic parts of the molecule, the constitution of these acids then becoming X · CH · C · O · NH3 ‖ O This, however, is rendered improbable for the following reasons:— (a.) In the case of amidosuccinic acid one carboxyl group should become neutralised, forming a monocarboxylic acid, which would then dissolve. This substance, however, is extremely insoluble. (b.) Anilidoacetic acid has a considerable rate of association, whereas if the carboxyl group were neutralised this would hardly be expected. Hippuric (i.e., benzoylamidoacetic) acid is characterized by its great rate of negative association. (It must be remembered, however, that the total depression is only a little above 1°.) The slight solubility cannot be the cause (for anilidoacetic-acid associates). It is well worth remarking, however, that the saturated solutions of these two acids have the same molecular depression, 71, which is about the mean value for the so-called constant. 6. Acids whose Molecular Depression increases with the Concentration.—Levulinic acid gives the abnormally high value 82 for the molecular depression. This, however, is probably correct. The acid obtained by fractionation under reduced pressure gave the value as 75. On solidifying this in a freezing mixture and separating the solid from the liquid the former gave the mean value 82. In 1887 Bredt* “Annalen,” 236, 225. gave the constitution of levulinic acid as

This alcoholic constitution is, I believe, at present not generally accepted, but it is probable that in phenol solution the molecules join up with themselves to give an alcohol, for (a) the alcohols have a molecular depression increasing with the concentration; (b) the alcohols have a high initial molecular depression. Hippuric acid has already been discussed: it possesses a resemblance to levulinic acid only in that there exists a carboxyl group in the molecule. The Connection Between the Rate of Association and Other Properties of the Acids. Sudborough and Lloyd* Trans. Chem. Soc., 1899, 467. have determined the rates of esterification of a number of substituted acetic acids in the presence of hydrochloric acid. The qualitative agreement between their results and those in Table II. is remarkable. The comparison is shown in the following table:— Table III. Acid. Esterification Constant. Rate of Association. Acetic 3-661 32 Propionic 3-049 18 Chloracetic 2-432 20 Phenylacetic 2-068 20 Bromacetic 1-994 17 Diphenylacetic 0-0559 (4) Trichloracetic 0-0372 2-5 For the esterification constant the numbers are given in decreasing order. With the single exception of propionic acid the values for the rate of association are in exactly the same order. The reason of this is not far to seek. The most probable theory of esterification is due to Henry: his explanation is based on an additive hypothesis. This is what occurs with the acids in phenol solution; but in this case like molecules are added together, while in the process of esterification an unstable addition product is formed with the alcohol, and this immediately decomposes into the ester. Looking somewhat ahead, it is to be expected that the ortho acids will associate less rapidly than their isomers, as these compounds are the most difficult to esterify. Further, the diortho substituted acids should scarcely associate at all, as, according to the well-known researches of V. Meyer and his pupils, these acids either do not form esters by this method or do so only with the greatest difficulty. Whether this supposition is true or false I hope to ascertain in the near future.

Connection Between the Molecular Depression and the Rate of Association. Among the normal fatty acids both constants tend to increase at the same time, though this is not absolute. The disubstituted acids tend to have a slightly higher value for the depression than the mono acids. Other regularities, if they do exist, remain hidden. Even if the acids associate at the same rate the molecular depressions may vary widely. Thus nitrosovaleric acid (rate of association 50 per cent.) has a molecular depression of 64; ethylmalonic acid. on the other hand (association 44 per cent.), gives the value of the constant as 72. Again, chlorphenylacetic and trichloracetic acids associate at the rates of 4 and 2–5, but their depression constants are 71-5 and 68. Of the seven acids whose association numbers vary between 18 and 24, all but one give molecular depressions between 69 and 71. In spite of these minor relationships, it is clear that the conditions that influence the molecular depression are indistinct; in all probability it depends to a large extent on the spatial arrangement of the molecule in solution. The Solubility of the Fatty Acids And Their Derivatives In Phenol. The following generalisations may be drawn up:— (1.) The fatty acids are readily soluble, but the solubility becomes slight when stearic acid is reached. (2.) All the oxy, halogen, and phenyl derivatives are readily soluble. (3.) The polycarboxylic acids are characterized by their extreme insolubility: this applies also to their unsaturated compounds, the halogen and oxy derivatives. Their solubility, however, is much increased by the introduction of an alkyl group, the effect being smaller the greater the number of carboxyls. (4.) The oxy derivatives of the dibasic acids are less soluble than the original acids. (5.) Of the unsaturated dibasic acids the cis compounds are much more soluble than their isomers possessing the trans configuration. (6.) The a amido acids are extremely insoluble; their substituted derivatives possess a slight solubility. In the following table the differences between the freezing-points of phenol and its saturated solution are given for a number of acids. This depression is roughly proportional to the number of dissolved molecules in a saturated solution of a constant weight:—

Table IV. Acid. F.P. Depression of Saturated Solution. 1. Malonic ·15° 2. Succinic ·22 3. Adipic 1·3 4. Sebacic 2·1 5. Methylsuccinic > 3 6. Ethylmalonic > 3 7. Mesaconic > ·27 8. Citraconic > 5 9. Fumaric ·09 10. Maleic 2·3 11. Bromsuccinic ·45 12. Dioxysuccinic (tartaric) ·10 13. Tetraoxyadipic (mucic) ·25 14. Amidoacetic ·08 15. Phenylamidoacetic 1·4 16. Benzoylamidoacetic 2·05 17. Amidophenylacetic ·15 18. Phenylamidophenylacetic 1·7 19. Amidosuccinic ·29 20. Tricarballylic ·20 21. Isopropyltricarballylic ·38 Acids 1–4 are homologous dibasic acids: it is seen that the solubility rises with the molecular weight. Mesaconic and fumaric acids are trans acids; they are much less soluble than their isomers, citraconic and maleic acids. It is interesting to note that with these acids the compounds that most readily form anhydrides are the most soluble. This was also observed in the case of succinic and methylsuccinic acids. The addition of a methyl group increases both the solubility and the ease of anhydride formation. Tartaric and mucic acids (12 and 13) are considerably less soluble than the acids from which they are derived (2 and 3). In the dibasic acids the introduction of an alkyl group causes an enormous increase in the solubility. In the case of the tribasic isopropyltricarballylic acid the increase is only comparatively slight (20 and 21). Among other insoluble di- or tri-basic acids may be mentioned oxalic, meconic, camphoric, and aconitic acids. Summary And Conclusion. (1.) In freezing phenol the fatty acids and their derivatives associate more or less rapidly with rising concentration. (2.) The rates of association of the normal fatty acids alternately increase and decrease for each member. If the

even compounds alone are considered the association decreases, reaches a minimum, and then rapidly rises again. The initial molecular depression, however, steadily falls to a minimum, and then rises again after about the twelfth member. (3.) Disubstituted acetic acids associate less rapidly than mono-derivatives. The trisubstituted acids have the smallest rate of association. (4.) Dicarboxylic acids associate more strongly than the monocarboxylic acids. (5.) Hydroxyl and nitroso groups tend to increase the rate of association. (6.) The rate of association of the substituted acetic acids shows a qualitative relationship with their velocity of esterification. (7.) The dicarboxylic and a amido acids are characterized by their sparing solubility. In conclusion, I wish to express my thanks to Professor Easterfield for the encouragement and advice which he has given me during the course of my work.