I
GEORGE W . WARREN,' ROY R. WARRENf2 and V. A. YARBOROUGH Union Carbide Chemicals Co., South Charleston, W. Va.
Gas Chromatography Guide to Better €xfracfion Processes The similarity between gas chromatography and extractive distillation is not often readily recognized
GAS
chromatography has been compared to conventional distillation techniques ( 7 , 5, 7 0 ) , and methods have been developed ( 2 ) for calculating the number of theoretical plates in a chromatographic column. T h e efficiency with which two compounds can be separated depends on the difference in their elution time. which a t constant operating conditions is a function of solute vapor pressure and other factors. For a distillation column separation efficiency depends on the difference between vapor pressures of the compounds being separated (assuming a n ideal sysrem). In a chromatographic column, however, some compounds with identical vapor pressures may be separated easily, while others whose vapor pressures differ by 200 mm. of mercury are not. Frequently a compound is eluted before another with a lower boiling point. Thus, chromatographic separations, like extractive distillation separations, depend on the liquid phase used. .-1 study has been made of the relationship between the two processes.
Experimental A Perkin-Elmer Model 154 Vapor Fractometer was employed for all chromatographic determinations. Chromatographic columns were 2 meters X 4.7 mm. (inside diameter). Helium was the carrier gas a t a floic rate of 100 cc. per minute. T h e 0.01-ml. liquid samples were introduced with a hypodermic syringe and needle through a siliconerubber diaphragm into the stream of helium. Detector response was recorded on a strip-chart recorder. Apparent retention volumes were measured by the method of Porter and others ( 9 ) . All distillation experiments were performed \cith a conventional equilibrium still. An equal weight percentage of solute mixture and material used for the chromatographic stationary phase were charged to a N i t e r distillation flask equipped with a stirrer and still head. The charge was refluxed for 2 hours and Present address, University of Alabama, University, Ala. Present address, Union Carbide Olefins Co., Institute, W. Va.
the vapor condensate and kettle mixture were then sampled for gas chromatographic analysis. Boiling point depression data were obtained with a Cottrell ebulliometer. Solubility data were obtained by a modification of the synthetic method-by titrating one component into the other to a "cloud point" indicating equilibrium solution.
Results and Discussion James, Martin, and others (2-J, 6, 8) have recognized the importance of solvent effects and have pointed to the potential application of gas chromatographic measurements in studying nonidealities of solutions. Martin and Svnge (6, 7) provided the basis for a rvlationship between partition coefficient for solute materials and retention volumes for a purely partition process. Porter and others ( 9 ) have applied the following expression for calculating the partition coefficient for the solute: Vi
=
the same group was measured in a n equilibrium still in the piesence of the above liquid phases. T h e most volatile compound was not always enriched in the overhead. However, the compound eluted first from the chromatographic column was always enriched in the still overhead when that particular liquid phase on the chromatographic column was present in the equilibrium still mixture. T o correlate the extent to which compounds of similar vapor prrssures are separated by extractive distillation and gas chromatography, relative volatilities obtained from extractive distillation experiments were compared with partition coefficient ratios (Table I). In general, agreement was good, although a considerable discrepancy was encountered in the n-heptane-1 -propanol system ; n-heptane is only 0.8% soluble in diethylene glycol, and thus the discrepancL between chromatographic and distillation data is probably caused by the twophase system in the equilibrium still.
V, f (AbfJ?T/yoPc)Vs
1% here
= .Zf,/R T / y P o = partition coefficient = retention volume Vs = volume occupied by stationary
H
II
solvent phase in column -1 = activity coefficient of solute in solvent phase, taking pureliquid solute as standard solute and gas phase as ideal y o = activity coefficient at infinite dilution LLf3= moles of stationary liquid phase per unit volume P o = solute vapor pressure R = gas constant T = column temperature K. &,'L = volume occupied by mobile vapor phase in column
Table I. Chromatographic and Equilibrium Still Data Are Well Correlated Parti-
Solvent Diethylene glycol
O
Partition coefficients were calculated using the above expression for two groups of compounds; boiling points ranged from 72" to 85" C. and from 96" to 109' C. Column temperature was 80' C. for measuring retention volumes of the first group of compounds and 100' C. for the second group. Liquid phases for each group were: didecyl phthalate, dibutyl phthalate, dimethyl phthalate, diethylene glycol, and tetraethylene glycol. Relative volatility of compounds in
Didecyl phthalate
Dibutyl phthalate
Solute5
t i o n RelaCoeffi- tive cierit I-olaI l a t i o a a tilityh
n-Heutane 172 1.9c l-propanol 6 . 0 2.0 Ethyl acetate Ethyl alcohol Diethyl ketone 2.9 1.1 I-Propanol Diethyl ketone I-Propanol 2-Propanol Ethylene dichloride
3.0
1.6
3.5
1.3
Diethyl ketone I-Propanol Ethyl acetate Ethyl alcohol
2.0
1.5
1.9
1.1
Dimethyl phthalate
1.9 1.7 Diethyl ketone I-Propanol a From chromatographic data. From equilibrium still data. Two-phase system.
VOL. 51, NO. 12
a
DECEMBER 1959
1475
I n order that solute concentration in the still and chromatographic column be similar, the percentage liquid phase was increased in the equilibrium still. Relative volatility for diethyl ketone-lpropanol was increased from 1.1 to 1.6 when diethylene glycol in the equilibrium still was increased from 50 to 7570 (weight). Solute concentration on the chromatographic column in most cases was less than 5%. Thus the partition coefficient ratio does give semiquantitative data to aid in choosing the proper extractant for separating two compounds of similar boiling points. Boiling point depression was used t o s t u d y 2 - b u t a nol-n -he p tane-dimethyl phthalate and n-butyl chlorideethyl alcohol-didecyl phthalate systems (Table 11). Normally a difference of IOo C. in boiling point of each solute wieh a particular solvent indicates that the solvent is satisfactory for extractive distillation. T h e difference in reflux temperatures is a function of solvent concentration, and a IOo C. difference is found only when solute concentration is 30 weight % or less. The partition coefficient ratio is 4.4 for the former system and 3.2 for the latter. The boiling point difference is greater for the former system, which is in agreement with partition coefficient ratios. T h e short retention time of water on the phthalate ester columns and n-butyl chloride on the glycol columns indicates that these compounds may be insoluble in the liquid phase of the chromatographic column. T o relate retention time with the solubility of those compounds in the liquid phase, solubility of various compounds was measured. Solubilities in relation to partition coefficients are given in Table 111. I n most cases, solute retention time increases with increasing solubility in the liquid phase. These data confirm that chromatographic separation is effected by a partition process. Thus, elution order can be predicted from a knowledge of the solubilization tendencies of the solvent. I n most cases the solute is infinitely soluble in the stationary phase; therefore, elution order can be more easily measured than predicted.
Table 111.
Solute Ethyl alcohol n-Butyl chloride Benzene Cyclohexane %-Heptane Water
1476
B.P., C.
78.3 78.8 80.1 80.7 98.4 100.0
Acknowledgment T h e authors thank R. C. Grimm for helpful discussions and H. T. Spengler and E. C. Bell for determining solubility and boiling point depression data.
literature Cited (1) Cvetanovic, R. J.. Kutschke, K. O., in “Vapour Phase Chromatography,” D. H. Desty, ed., p. 87, Academic Press, New York York, 1957 1957. (2) James James, A. T.. Martin, .A. J. P., Bzochem. J . 50,6’ 50,679 (1952). (3) James James, A. T., Martin, A. J. P., Brzt. .Wed. B u l l . 1 ,Wed. 10, 170 (1954). James, A . T., Martin, A. J. P., Howard(4) James. Smith, G., Bzochem. J . 52, 238 (1952).
(5) Keulemans, A . I. M., Kwantes, A., in “Vapour Phase Chromatography,” D. H. Desty, ed., p. 15, Academic Press, New York, 1957. (6) Martin, A. J. P., Biochem. Soc. Syinposia No. 3,4 (1951). (7) Martin, A. J. P., Synge, R. L. M., Biochem. J . 35, 1358 (1941). (8) Zbid., 50,532 (1943). (9) Porter, P. E., Deal, C. H., Stross, F. H., J . Am. Chern. SOG.78, 2999 (1956). (10) Purnell, J. H., in “Vapour Phase Chromatography,” D. H. Desty, ed., p. 5 2 , Academic Press, New York, 1957.
RECEIVED for review February 4, 1959 ACCEPTED July 13, 1959 Division of Industrial and Engineering Chemistry, 135th Meeting, XCS, Boston, Mass., April 1959.
Table II. Boiling Point Depression Data A difference of 10’ C. in solute boiling point shows that the solvent is satisfactory for extractive distillation B.P. of A Mixture, Weight Mole B.P. Hoa Solute O c. Solvent 70 % 282 0 0 119 2-Butanol Dimethyl phthalate 180 133 122 116 108 102 98.3
5 10 20 30 50 75 100
12.1 22.6 38.3 47.1 72.4 88.7 100
Dimethyl phthalate
n-Heptane
27.2
130 115 110 106 101 98 97.6
5 10 20 30 50 75 100
7.3 17.7 31.5 45.4 66.0 85.3 100
Didecyl phthalate
n-Butyl chloride
98.3
400 276 200 105 90 83 80 78
0 5 10 20 30 50 75 100
17.3 34.6 54.7 67.4 82.8 93.5 100
Didecyl phthalate
Ethyl alcohol
260 160 84 81 79 78 77.7
5 10 20 30 50 75 100
33.8 51.9 70.8 80.6 90.7 96.7 100
31.2
50b 18 12 10 7 4
0
16.0c 40 21 9 4 2
Difference in boiling_noints of 2-butanol and a Partition coefficient at infinite dilution. . Difference in boiling points of n-butyl chloride and ethyl n-heptane in dimethyl phthalate. alcohol in didecyl phthalate.
Comparison of Solubility Data with Partition Coefficients
Solute retention time increases with increasing solubility in the liquid p h a s e Tetraethylene Glycol Diethylene Glycol Didecyl Phthalate Dibutyl Phthalate Dimethyl Phthalate Part. Part. Part. Part. Part. 70 Sol. coeff. % Sol. coeff. % Sol. coeff. % Sol. coeff. % Sol. coeff. 53.6 Infinite 52.6 Infinite 31.2 Infinite 259 Infinite 176 Infinite 67.2 Infinite 118 Infinite 98.3 Infinite 13.7 11.3 31.1 24.0 105 Infinite 151 Infinite 120 Infinite 39.4 1.0 68.0 Infinite 30.1 Infinite 72.9 Infinite 85.2 Infinite 9.1 2.3 12.2 3.4 27.2 10.0 62.8 Infinite 55.8 Infinite 0.96 0.8 12.7
