The Analysis of Mixtures of Volatile Fatty Acids

Two main procedures are available for such analyses: the distillation method of Hillig & Knudsen (1942), which is based essentially on the classical Ducleaux (1874, 1900a, b, 1908) procedure, and the method of Osburn, Wood & Werkman (1936) which involves partition of the acids between water and ether phases. In each of these methods formic acid is estimated by a separate procedure.

Preliminary examination of the partition method proved its effectiveness for the assay of mixtures containing acetic, propionic and butyric acids except when one of the components constituted either a very large or a very small proportion of the mixture. As it was expected that such mixtures would be encountered, an improvement in the method was sought.

No difficulties were experienced in the analysis of solutions prepared from mixtures of two acids, when one of them was present in much larger proportion than the other ; but when analyses were made of mixtures containing known concentrations of three acids, the amounts of each, determined by the partition procedure, indicated that serious sources of error are inherent in the method. This is to be expected ; the estimation involves essentially a calculation by difference and so in certain circumstances the implied error is large. Assuming, for example, an accuracy of ± 0·5% in the titration of the acid content of the aqueous phase of each partition, and of ±0·1% in the titration of the original acid mixture, the theoretical maximum error involved in the estimation of propionic acid in a mixture containing acetic acid 1·5 g./l., propionic acid 0·25 g./l. and butyric acid 0·25 g./l. would be 67%; whereas, with the same manipulative accuracy, the maximum error involved in the estimation of this component of a binary mixture containing only acetic and propionic acids in the above concentrations would be reduced to 10%.

If the concentration of butyric acid could be determined by a separate estimation with an accuracy of even 5 %, which may readily be accomplished, then, by applying the partition procedure and substituting the figure for butyric acid in solving the equations, the maximum error involved in the estimation of the propionic acid in the former mixture would be reduced to 19%.

It is clear, therefore, that the accuracy of the determination of the concentrations of each of three acids in a mixture should be very considerably increased if the concentration of one of them is determined by an independent method and the resulting data introduced prior to solving the partition equations. It is also clear that if such a combined procedure is to improve the precision of the method, the accuracy of the independent determination of the third acid must increase with the accuracy of the measurement of the partitions.

The case for improved accuracy arising from the use of combined methods of this sort would apply equally well to both partition and distillation procedures, and so the method of distillation at constant volume (Hillig & Knudsen, 1942), or, since the collection of only one fraction of distillate would be necessary, the simpler semidistillation procedure (Virtanen & Pulkki, 1928) could be adopted. In spite of the fact that the distillation constants of the acids are less widely different than are their partition constants, such a combination of a distillation method for acetic and propionic acids with separate independent estimations of formic acid and of butyric acid, may be preferred: partition calls for considerable quantities of very pure ether and a critical control of the temperature at which equilibrium is established, whereas the distillation procedure involves nothing other than simple, careful manipulation.

A number of 3-acid mixtures containing acetic, propionic and butyric acids, and some 4-acid mixtures containing formic acid in addition, were made up from carefully purified components. Partition constants of the pure acids between water and pure ether at 20° C. were determined; the findings were in accord with the constants reported by Osburn et al. (1933).

In the first series of analyses the results from the partition method when applied to 3-acid mixtures containing acetic, propionic and butyric acids, were compared with those derived from solving partition equations after introducing the data obtained from a separate estimation of the concentration of butyric acid by the method of Kline (1934). The results of analyses of the 4-acid mixtures which contained formic acid are included in this series, the concentration of formic acid in the mixtures being determined by oxidation with mercuric oxide by a procedure slightly modified from that of Osburn et al. (1933), and the data so obtained substituted in the partition equations.

Several variables are encountered, among which the actual error of the titrations themselves, the influence on the partition exerted by the concentrations of the individual acids, and changes in the physical conditions under which the partitions are estimated, are probably the main contributors. The observed partitions differed from the theoretical by amounts varying between o and 0·5 %.

The results from this first series, summarized in Table 1, showed quite definitely that under the conditions of this test, the combined procedure afforded a more accurate analysis when one of the acids was present in relatively high proportion.

In the second series analysed, the results of a distillation procedure were compared with those from the partition method, each being used in conjunction with the separate determination of butyric acid. The semi-distillation method was adopted, but only 50 ml. of the mixtures were distilled instead of the 200 ml. recommended by Virtanen & Pulkki (1928). The distillation constants were determined at known concentrations of approximately 0·2 g./100 ml. for each acid. The results which are set out in Table 2 render it clear that the simpler manipulations involved in this distillation may, without loss of accuracy, be employed in place of the more exacting partition procedure.

1. The analysis of mixtures of acetic, propionic and butyric acids, in which one of the acids is present in either very large or very small proportions has been examined.

2. A combination of a separate method for butyric acid, with the partition method for acetic and propionic acids, provided more accurate estimations than those obtained directly from the application of the partition method to all three acids.

3. Semi-distillation proved a suitable alternative to partition in the combined procedure.

The author is indebted to Mr H. R. Marston, Chief of the Division, for originally suggesting the problem and for his help and criticism.