May, 1941
INDUSTRIAL A N D ENGINEERING CHEMISTRY
(2) Bonotto, M.,OiE d Soap, 14,310 (1937). (3) Breakey, E.P.,and Olcott, H. S., Science, 87,87 (1938). (4) Brewia, J., J . SOC.Chem. Id., 43, 111 (1924). (5) Campbell, K.N.,Morris, R. C., and Adama, R.. J . Am. Chem. SOC.,59,1723 (1937). (6) Carruth, F. E., Ibid., 40,647(1918). (7) Gallup, W. D.,and Reder, R., J . A v . Reoearch, 52,65 (1936). (8) Garner, J. B., Natural Gas Mag., 16,3(Jan., 1935), (9) Hassel, B., Oil & Fat IndustTka. 4,163 (1927). (10) Hildebrandt, K., Fette u. Seifen, 46, 350 (1939). (11) Jones, D. B., and Waterman, H. C., J . Biol. Chena., 56, 601 (1923). (12) Lewkowitsch, J., "Chemical Technology and Analysis of Oils, Fats, and Waxes", 6th ed., Vol. 2, p. 20, London, Macmillan Co., 1922.
615
(13) MacGee, A.E.,02 & Soap, 14,322,324 (1937). (14) Meyerweisflog, E.W.,Ibid., 14,10 (1937). (16) Olcott, H.S.,and Fontaine, T. D., J. Am. C h a . SOC.,61, 2037 (1939). (16) Olcott, H.S.,and Thornton, C. D. W., Ibid., 61.2417 (1939). (17) Rosenthal, H.,and Trevithick, H. P., Oil & Soap, 11,133(1934). (18) Royce, H.D.,and Lindsey, F. A., Jr., IND. ENG.CHIM.,25,1047 (1933). (19) Bchwartz, A. K..Oil & Fat Industries, 4, 284 (1927). (20) Shrader, J. H., Cotton Oil P T ~ S4,42 S , (1921); 5 , 29 (1922). (21) Siavers, A. F., and McIntyre, T. D., Ibid., 4,44 (1921). (22) Wesson, D.,Oil & Fat Industries, 7 , 217 (1930). (23) Weason, D.,Oil & Soap, 10,151 (1933). COHTBIEUTION from the Multiple Fellowship of the Cotton Research Foundation at Mellon Institute.
Purification of Glvcerol bv J
J
Crystallization Synthetic glycerol, produced by hydrogenolysis processes, is highly impure and the isolation of C. P. or even U. S. P. grades has heretofore proved extremely difficult. A new approach to the problem-namely , crystallization in the presence of suitable solvents-readily yields glycerol of any degree of purity required. As a result of this process it is now possible to produce by a simple one-step treatment with hydrogen any desired quantity of high-quality glycerol from such abundant carbohydrate materials as starch and dextrose. HE production of glycerol from carbohydrate sources by fermentation has been extensively investigated and is reported (6) to have been responsible for the production of 4,000,000 pounds per month of this material in Germany during the war of 191618. More recently the hydroto furnish genolysis of carbohydrates has been found (2,4,6) a ready source of this compound. The conversion of the abundant cornstarch of the Middle West into glycerol via the hydrogenolysis of dextrose is of potential importance from the standpoint of national military preparedness. Glycerol produced by hydrogenolysis upon an experimental scale in the laboratories of the Commercial Solvents Corporation was somewhat impure and failed to meet U. S. P. specification in four respects: It was slightly colored, it had a disagreeable odor not due to acrolein, it tasted bitter, and a black color was produced when it was treated with a n equal volume of sulfuric acid. In order to pass U. S. P. specifications, glycerol must not develop more than a slight color under these conditions. Since conventional purification methods had proved inadequate, an investigation of the applicability of crystallization to the problem was undertaken. Glycerol melts a t 18' C. but is rarely seen in solid form because of the ease with which i t supercools. Apparently the most recently reported attempt to develop a process for the commercial purification of glycerol by crystallization was by Kraut (I) who stated in 1871:
T
1
H. B. HAS$ AND J. A. PATTERSON Purdue University and Purdue Research Foundation, Lafayette, Ind.
For roducing crystals of glycerine in the first instance, distilled Jycerine of 30" Beaume is kept in closed vessels for a week or more at a temperature of from 32" to 43" F., when crystals of pure glycerine will have formed themselves out of the mother liquor. For manufacturing perfectly pure gl cerine from the ordinary glycerine of commerce, such glycerine ofY30' Besum6 is first cooled down under exclusion of moisture from 32" to 43" F., after which a small quantity of the before-mentioned Crystallized gl cerine is introduced and the glycerine is allowed to stand at a low temperature, such as above described for some days, whereupon the crystallization of the glycerine will ensue. When the crystallization is found to proceed no farther the crystals are separated from the remaining mother liquor, first by decantation, and then by a centrifugal drying machine. Should it be desired to carry the process of purification still further the crystals ma be melted down and then recrystallized by the above-descrited process.
So far as we know, the foregoing process invented by Karl &aut and patented by C. D. Abel was never used commercially. One of the obstacles in working with such a process is the difficulty in obtaining seed crystals. Gibson and Giauque reported (8)in 1923 that after "the artifices ordinarily used for starting crystallization in the absence of seed crystals were all tried without success", crystals were produced at will by prolonged cooling of pure glycerol with liquid air and subsequently bringing the sample gradually to 0' C. Oblad and Newton (8) found that cooling by solid carbon dioxide can be substituted for the lower temperatures of liquid air with equally good results. Glycerol seed crystals are most conveniently obtained by keeping pure, anhydrous glycerol overnight in a test tube immersed in a Dewar flask cooled by solid carbon dioxide. Upon removing the test tube, the glassy glycerol appears unchanged because the seed crystals are so small as to be invisible. After standing for a few hours a t 0" C., the crystals have grown enough to resemble sucrose in general appearance. After 2 days a t 10" C. the crystallization is complete. Even assuming that seed crystals of glycerol are obtainable, the process of Kraut is impractical when applied to glycerol containing even relatively small concentrations of im-
INDUSTRIAL AND ENGINEERING CHEMISTRY
616
purities because of the extremely slow rate of crystallization in these Viscous solutions. For example, a 400-ml. sample of U. S. P. glycerol became only about 90 per cent crystalline after several weeks a t 10" C. although a similar saniple of pure glycerol required only 2 days under the same conditions for complete crystallization. According to Oblad (7), 0.01 per cent of water prevents the spontaneous crystallization of glycerol. Ten per cent of water lowers the freezing point- to -1.6'C.
CLUSTER OF GLYCEROL CRYSTALS, NATURAL SIZE Glycerol produced by hydrogenolysis and partially purified by steam-vacuum distillation boiled at 130-140" C. (3.5 mm. pressure) and had a specific gravity of 1.2259. It crystallized incompletely even when maintained for several weeks at 5" C. in the presence of seed crystals. Although crystallization was somewhat accelerated by stirring, the crystals so formed could not easily be separated from the mother liquor because of its high viscosity. An investigation of the effect of thirty solvents upon the crystallization of glycerol is summarized in Table I. ~~
TABLE
FROM VARIOUS I. CRYSTALLIZATION O F GLYCEROL
Solvent Tested Methanol Ethanol 1-Propanol 2-Propanol 1-Butanol 1-Pentanol 2.3-Butylene glycol Diethylene glycol Triethylene glycol 2-Methoxyethanol 2-Ethoxyethanol 2-Butoxyethanol 2-(2-Ethoxy)-ethoxyethanol 2-(2-Butoxy)-ethoxyethanol Furfuryl alcohol 2,2'-Dihydroxydiethyl ether Aminoethanol Triethanolamine Diethylamine Liquid ammonia Trimethylene glycol Propylene glycol Diacetone Furfural 2,2'-Dichlorodiethyl ether Terpineol Ethyl acetoacetate Ethyl acetate Chloroform Carbon disulfide 0
Solubility not tested.
SOLVENTS Result a (1
0 G
(I
Sol. Sol. Sol. a
Sol. a
Sol. a
Sol. Sol. Sol. Sol. Sol. .e
Partially sol. Insol. Insol. Insol. Insol. Insol. Insol. Insol.
Rate of Crystallization Slow Slow Slow but complete Slow Fairly rapid and complete Fairly rapid Slow Slow Slow Very slow Very slow Slow Very slow None Very slow Very slow Slow but complete
....
Fairly rapid Very rapid and complete Slow Slow None
.... .... .... .... ....
Vol. 33, No. 5
Careful vacuum rectification of the hydrogenolysis product showed that it contains impurities with boiling points very close to that of glycerol. These substances have not been completely identified but may tentatively be regarded as triols derived from hydrocarbons having more than three carbon atoms. Glycerol which passed U. S. P. specifications was easily obtained in good yield from this mixture by the following procedure: The material distilling close to the boiling point of glycerol was dissolved in an equal volume of 1-butanol, laced in a water-tight bottle, ooolcd, seeded, and slowly revolve! in an ice water bath until crystallization had taken place. The impurities and most of the solvent were centrifuged away from the glycerol crystals which were washed with cold acetone or isopropyl ether. Traces of solvcnt were removed from the glycerol by evaporation. Over 60 per cent of the material boiling at 290' C. was thus isolated as crystalline glycerol. Of the thirty solvents tested, 1-butanol, I-pro anol, l-pentanol, and liquid ammonia were found most suitatle. When 25 ml. of glycerol were dissolved in 25 ml. of liquid ammonia and seeded, crystallization at a temperature just below the boiling point (-33" C.) of liquid ammonia was complete in 30 minutes. At about 5' C. glycerol separated slowly from 1-propanol in large, perfectly formed crystals. 1-Butanol was the best solvent tried for this purpose since it could be handled more conveniently than liquid ammonia, and glycerol crystallized from it more readily than from the other alcohols.
It would seem that the function of the solvent here is to diminish the viscosity of the glycerol solution without too greatly lowering the fugacity of the solute. A fairly low viscosity is necessary for rapid crystallization, but substances such as water, methanol, and ethanol are such good solvents for glycerol that they greatly lower the fugacity of the glycerol in the solution. This requires the use of lower temperatures which, in turn, increase the Viscosity of the solution. Butanol, with only one hydroxyl group and four carbon atoms, is a relatively poor solvent for glycerol; in the presence of small quantities of water these liquids readily separate into two layers a t 0" C. Because of this wide departure from ideality, butanol yields relatively dilute, mobile glycerol solutions from which nearly all of the glycerol can be crystallized. The next homolog, pentanol, is a still poorer solvent for glycerol; concentrations of glycerol in this alcohol are so low that butanol is to be preferred.
Acknowledgment It is a pleasure to express thanks to the Commercial Solvents Corporation for furnishing the funds defraying the cost of this investigation and for supplying the samples of hydrogenolysis product which were used. Literature Cited (1) Abel, C. D., Brit. Patent 873 (1871). (2) Assoc. Am. Soap and Glycerine Producers, Inc., French Patent 816,952 (1937); Brit. Patent 490,211 (1938). (3) Gibson, G . E., and Giauque, W. F., J . Am. Chem. SOC.,45, 102-3 (1923) Farbenindustrie A-G., Brit. Patent 299,373 (1927); French (4) . , I. G. Patent 662,874 (1928). ( 5 ) Larchar, A. W.. U. S. Patents 1,963,997and 1,963,999 (1934). (6) Lawrie, "Glycerol and the Glycols", New York, Chemical Catalog Co.. 1928. (7) Oblad, A. G., private communication. (8) Oblad, A. G., and Newton, R. F., J. Am. Chem. SOC.,59, 2497 (1937). I
ABSTRACTED from a portion of the thesis submitted by J. A. Patterson t o the faculty of Purdue University in partial fulfillment of the reauirements of the degree of doctor of philosophy, June, 1940. The prooesses described are covered by U. 8. Patent 2,233,606issued t o H. B. Hass and assigned t o the Purdue Researoh Foundation.
