known, the deviation of the results from the correct results, while not calculable, will generally be less than in the extreme example cited. The magnitude of the z/y uncertainty for a given precipitate distribution is also influenced by the nature of the beta rays, weaker beta rays introducing a greater degree of uncertainty. On the other hand, rather pronounced shifts in precipitate distribution from a n assumed distribution cause only minor deviations in results. From this it is evident that, even though a precipitate distribution differs from that represented b y the curves in Figure 1, the results obtained are not expected to differ by much from the results that would have been obtained if the true distribution had been known. A z’y value obtained for a precipitate containing radiocarbon, for example, is expected to differ by less than 10% from the true value. CONCLUSION
The proposed method a l l o w estiniation of the total activity of precipitates on filter paper from the measured ac-
tivities of both sides of the paper. The absorption coefficient, p, for the beta rays and the thickness, g, for the paper with precipitate must be known. Close-geometry conditions of counting must be used, in which the filter paper is close to the counter. A family of curves parametric in pg (Figuie 1) serve to define z/y as a function of z/y, where z is the total activity and z and y are the measured activities from the top and bottom of the filter paper, respectively. Total activity, z, can be converted to absolute total activity (d.p.m.) by multiplying b y the geometrical factor of the counting setup. This is the ratio of 4 8 to the solid angle subtended by the sensitive volume of the counter, based on average precipitate position and corrected for radiations absorbed by the walls of the counter. For an efficient flow counter this factor is expected to be approximately equal to 2. I n the case of small radiocarbon-tagged precipitates, the total absolute activities obtained by means of Figure 1 are expected to differ by less than 10% from the absolute values.
ACKNOWLEDGMENT
The assistance of H. G. Isbell and L. 8. Turcios in the preparation of this manuscript has been invaluable. The authors 11-ish to thank the Bureau of Ships, United States S a v y Department, for permission to publish this material; but all opinions expressed are solely those of the authors. and do not necessarily reflect the official views of the S a v y Department. LITERATURE CITED
(1) Aten, A. W., Jr., Sucleonics 6, S o . 1, 71 (1950). (2) Cook, G. B., Duncan, J. F., “Modern Radiochemical Practice,” Oxford University Press, London, 1952. (3) Schii-eitzer, G. K., Stein, B. R., Sitcleonics 7, S o . 3, 65 (1950). (4) Solomon, A . K., Gould, R G., .infinsen, C. B., Phys. Reo. i 2 , 1097 (1947). (5) Suttle, A . D., Jr., Libby, K. F., r i ~ aCHEU. ~ . 27, 921 (1955). RECEIVEDfor revieiv June 3, 1956. Accepted Sovember 17, 1956.
Application of Thermal Diffusion to Separation of Aliphatic Alcohols and Fatty Acids from Their Mixtures C. W. BLESSIN, C. B. KRETSCHMER, and RICHARD WIEBE Northern Ufilizafion Research Branch, Agriculfural Research Service,
b Although separation of mixtures by thermal diffusion is often very effective -for instance, in mixtures of paraffin hydrocarbons-very little or no separation was found in alcohols and fatty acids. This failure is attributed to hydrogen bonding, which obscures structural differences and prevents their separation.
I
N CONNECTIOS with the possible ap-
plication of thermal diffusion to the analysis of complex mixtures of fatty acids and their derivatives, such as those encountered in vegetable-oil technology, binary mixtures of the lower aliphatic alcohols and fatty acids were studied. APPARATUS AND PROCEDURE
The stainless steel thermal diffusion column used in this n-ork was similar in
408
ANALYTICAL CHEMISTRY
U. S.
Deparfment of Agriculture, Peoria, 111.
design to the one described by Jones and Rlilberger (4). The fractionating section vias 6 feet in length, with an annular space of 0.0115 in. and an annular volume of 22.5 ml. The inner surface was n-atercooled, while the outer one n-as heated electrically. I n older to check the efficiency of the column, the separation of a series of binaiy paraffin hydrocalbon niiutures n a s studied (Table I). -4s \\as to be expected from n-paraffin hydrocarbon mixtures, sepaiation increased viith incieasing differences of the molecular weights between the two components of the mixture. The per cent separation is givcn for a -%hour run in each case and is close to the equilibrium value for this column. RESULTS
I n Table I the density values are quoted from API Research Project 44 ( I ) . The values of final composition listed were determined from experi-
mental plots of refractive index us. volume fraction, which in most cases !yere nearly straight lines. Alcohol Mixtures. No such regulaiity was found with binary mixtures of t h r lower aliphatic alcohols. T h e
0
20
40 60 80 HOURS OF OPERATION
100
Figure 1. Effect of time on separation of 50 volume mixture of propionic acid in butyric acid
yo
Separation of 50 Volume % Binary Mixtures (Duration of runs, 48 hours) Final Composition Density, Hoi \Tall, Cold \Tall, Top Bottom 25" C. c. c. 10% 10%
Table I.
Components
% separations
Hydrocarbons n-Heptane n-Octane n-Hept ane n-Decane n-Heptane n-Hexadecane n-Octane n-Decane n-Octane n-Hexadecane n-Decane n-Hexadecane
0 6795 0,6985 0.6795 0.7262 0,6795 0.7700 0.6985 0,7262 0.6986 0,7700 0 , 7262 0.7700
90
24
90
25
55 0 45 0
38.5
37.0 63 0 21 5 78 5 7 0 930 25.5 74.5
26
93.0 7.0
6.5 93.5
86.3
30
83 5 16 5
I1 0 89 0
72 7
68 5
31 5 90
25
115
26
115 115
93 5 6 5 61.5
18 2 46 9 86 7 35.6
.Zlcoholsb Methanol Ethyl alcohol Methanol n-Propyl alcohol RIethanol Isopropyl alcohol Methanol tert-Butyl alcohol Ethyl alcohol n-Propyl alcohol Ethvl alcohol Isopropyl alcohol Ethyl alcohol n-Butyl alcohol Ethyl alcohol tert-Butyl alcohol n-Propyl alcohol Isopropyl alcohol n-Propyl alcohol tert-Butyl alcohol n-Butyl alcohol tert-Butyl alcohol
0 0 0 0 0 0 0
7865 7851 i865 7991 7865 7808 7865 0 . 7807 0,7851 0 . 7994 0,7851 0.7808 0 , '7861 0,8057 0 7851 0 7807 0 7994 0 7808 0 7994 0 7807 0 8057
0 7807
50 50 51 48 50 50 56
0 0 5 5 0 0 0 41 0 51.0 49.0 46.0 54.0 51.0 49.0 52 0 48 0 50 0 50 0 50 0 50 0 51 5 48 5
49 5 50 5 49 0 51 0 47 5 52 5 40 0 60 0 49.0
51.0 50.0 50.0 49.0 51.0 45 0 5.5 0 50 0 50 0 50 0 50 0 45 5 54.5
0 6 2 3 2 7 15 4
1.7 2.6 1.6
7 0 0 0 0.0 4 9
Fatty Acids
b
Propionic 0,9880 51.5 46.5 Butyric 48.5 53.5 0.9532 Propionic 0 9880 51.5 47.5 Valeric 0 9345 48.5 52.5 0 9880 Propionic 53.5 44.5 0 9230 46.5 55.5 Caproic Propionic 0.9880 51.0 45.5 Enanthic 0 9137 49.0 54.5 Propionic 0 9880 52.5 47.0 Caprylic 0.9066 48.0 54.0 Propionic 0 9880 52.0 46.0 Pelargonic 0 9017 48.0 54.0 A ~ (betn-een L ~ top and bottom fractions) X 100 - % separation. AnD (between pure compounds) Hot \\-all temp. 50' C., except for n-hutyl-tert-butyl, which is 75' C.
0 9 0 f-. La c/.
c
5.3
3.7 9.2 5.5 5.6
LITERATURE CITED
6.1 (1) Am. Petroleum Institute, Research Project 44, "Selected Values of
Cold wall temp.
Hot wall temp. 100" C. Cold wall temp. 27-30" C.
largest per cent separation was only 1570, observed with a 50:5@mixture of methanol and tert-butyl alcohol. Hydrogen bonding in alcohols is very strong ( 3 ) . As alcohols can form tn-o hydrogen bonds per molecule, polymers
alcohol. showing the widest variation in structure. The importance of structural difference in alcohols in facilitating separation is also shown in the system benzyl alcohol-ethylene glycol, where Jones and Milberger found a 29% separation ( L ) . N o seaaration occurred in a 50350 mixture of n-propyl with isopropyl alcohol and with fert-butyl alcohol, respectively, although i t might have been expected to take place. A ternary mixture consisting of equal percentages b y volume of methyl, ethyl, and lert-butyl alcohols was investigated in order to find out whether ethyl alcohol might improve separation of the other two components. S o such effect 11-as observed, and separation occurred a s if ethyl alcohol was not present. Fatty Acid Mixtures. T h e esperimental results giving t h e separation of butyric, valeric, caproic, enanthic, caprylic, and pelargonic acids with respect to propionic acid are also shown in Table I. All acids had been purified in a highly efficient fractionating column. Hydrogen bonding is again evident, as only small separation occurred. The irregularity up to caproic might be attributed to the differences between odd- and even-numbered acids, which is also shown by differences in melting points. Beyond caproic acid, differences betm-een even- and odd-numbered acids appear to become negligible a s far as thermal diffusion behavior is concerned. I n all pairs of fatty acids investigated, the acid having a lower density niigrated to the bottom because of thermal diffusion and thus opposed the purely thermal density gradient. This situation may result in the &-called forgotten effect (a), a n example of which is given by Jones and RIilberger (4), where it is shown that this results in a reversal of the direction of concentration. The propionic-butyric acid mixture was selected to test whether the forgotten effect m-as influencing the separation of the acids. As shown in Figure 1, equilibrium n-as approached normally.
will consist of molecular chains. As far as thermal diffusion is concerned, all n-alcohols therefore appeared to have a more or less identical structure and significant separation was obtained only between methanol and tert-butyl
Properties of Hydrocarbons and Related Compounds," Carnegie Institute of Technology, Pittsburgh, Pa., 1953. (2) deGroot, D. T., Physica 9, 923 (1942). (3) Hildebrand, J. H , Scott, R. L., "Solubility of Nonelectrolytes," 3rd ed., p. 172, Reinhold, \-e\v Tork, 1950. (4) Jones, A. L., Nilberger, E. C., Ind. Eng. Chem. 45, 2689 (1953).
RECEIVED for review September 7, 1956. Accepted h-ovember 16, 1956. VOL. 29, NO. 3, MARCH 1957
409
