N ICKEL-CATALYSTS effective a t temperatures below 1300" F., whereas tJhecommercial reactions are carried out at temperatures above 1300" F., usually 1500 F. I n view of the correlation between catalyst activity and nickel content (Figure 4), without regard to the other catalyst constituents. i t appears unlikely that the commercial catalysts studied in this laboratory are promoted. These catalysts are not selective, but act to bring about thermodynamic equilibrium with respect to all possible reactions of hydrocarbons and steam a t high temperatures (IO). It has been found necessary to maintain close control over the composition of the materials used in preparing the catalysts. Very small amounts of undesirable materials may be released from a catalyst t o be deposited at another point in the plant with serious results. For example, various metal oxides may be reduced at a high temperature and be carried as metal vapor to a lower temperature zone where the reaction may reverse and deposit a very voluminous powder of metal oxide. A few pounds of material deposited in this way are sufficient to plug a packed reaction vessel several feet in diameter. Both the life and activity of catalysts used for the reaction of hydrocarbons with steam in present plant practice are influenced t o a considerable degree by their nickel contents. The'other constituents of a catalyst and the method of preparation are probably chiefly important through effect on the porosity, surface area, and physical strengtp. Nickel shot and screens, for example, make relatively poor catalysts because of low porosity and low7 surface area. O
Catalysts that are inactive either because of low nickel content, . exposure t o a reducing atmosphere at very high temperatures, o r the presence of hydrogen sulfide are relatively selective for t h e decomposition of hydrocarbons to carbon and hydrogen. Highly active catalysts are better able to bring about reaction between hydrocarbons and steam without an accumulation of carbon, providing t h a t sufficient steam is present. LITERATURE CITED
Clark, E . L., Kallenberger, R. H., Browne, R. Y., and Phillips, J. R., Chem. Eng. Progress, 4 5 , 651-4 (1949). Cope, W. C., Chem. Inds., 64, 920-5 (1949). Dieffenbach, O., and Moldenhauser, W., Ger. Patent 2 2 9 , 4 0 8 (1909).
Freyermuth, G. H., Small, J. K., and Hanks, W. V. (to Standard Oil Development Co.), U. S. Patent 1,904,441 (1933). I. G. Farbenindustrie, Brit. Patents 267,535 (1926); 3 2 3 , 8 5 5 (1928).
Mittasch, A., and Schneider, C. (to Badische Anilin and Soda Fabrik), U. S. Patent 1,128,804 (1915). Mond, L., and Langer, C., Brit. Patent 12,608 (1888). Murphree, E. U., Brown, C . I . , rrnd Gohr, E. J., I N D . ENG. CHEM.,32,1203-12 (1940).
Reed, R. M., Trans. Am. Ins!. Chem. Engrs., 4 2 , 3 7 9 - 4 0 1 (1946). Reitmeier, R. E., Atwood, Kenton, Bennett, H. A , , Jr., and Baugh, H. M., I N D .E N G .C H E X . ,40,620-6 (1948). Standard Oil Development Co., Ger. Patent 534,906 (1930). Williams, Roger (to E. I. du Pont de Nemours), U. 8.Patents 1,826,974, 1,830,010, 1,834,115 (1931); 2,119,565, 2,119,566 (1938). RECEIVED for review October 17, 1951.
ACCEPTEDJanuary 14, 1952.
RANEY NICKEL-CATA LYZED HYDROGENATION OF COMMERCIAL ALDOL C. KINNEY HANCOCK' Bureau o f Industrial Chemistry, The Utiiversity of Texas, Austin, l e x . T h e catalytic hydrogenation of aldol i s one step in one of the processes for synthesizing butadiene from natural gas. This reaction was studied in order to more nearly attain optimum conditions. The commercial aldol used appeared to be largely the condensation product between aldol and acetaldehyde. Under optimum hydrogenation conditions, the aldol yielded about 70% butylene glycol, alcohpl being the main by-product. A t higher temperatures, increased dehydration of aldol preceded hydrogenation and led to an increase in yield of butyl alcohol. The addition of water in amounts up to 30% does not greatly affect the yield of butylene glycol or the reaction time. Raney nickel can be re-used if it i s protected properly. Results obtained under various special conditions are reported. The results should be useful to those interested in the reduction of aldol and of related compounds.
catalytic dehydration of 1,3-butanediol t o yield, finally, 1 , 3 - b u t ~ diene. Some of the results of a study of the fourth reaction of this series are reported in the present paper, There has been considerable interest in the hydrogenation (or reduction) of aldol for a long time. This reaction, depending upon the conditions, may yield either butyl alcohol or 1,3butanediol. Butyl alcohol is the product if dehydration of aldol to yield crotonaldehyde precedes hydrogenation (or reduction); if preliminary dehydration of the aldol does not occur, 1,3-butanediol is the product. This paper is concerned primarily n i t h the latter reaction; however, the former reaction is of some concern since the extent of its occurrence materially affects the yielct of 1,3-butanediol. P R E V I O U S WORK
SERIES of studies related to the synthesis of 1,3-butadiene from natural gas was carried out in 1942 by technologists of the Bureau of Industrial Chemistry of The University of Texas. Research was concentrated on the following series of reactions for accomplishing the over-all synthesis: production of acetylene from natural gas by the Schoch electric discharge process ( I $ ) , catalytic hydration of acetylene, aldol condensation of acetaldehyde, catalytic hydrogenation of aldol, and 1 Present address, Department of Chemistry, The Agricultural and
A
hrechanical College of Texas, College Station, Tex.
May 1952
The literature on the hydrogenation (or reduction) of aIdol is abundant. Soon after t h e introduction of each new carbunyI hydrogenation catalyst or reducing agent, papers have been published on t h e application t o aldol. Among these are reduction by aluminum amalgam (6),electrolytic reduction ( I ) , yeast reduction ( I I ) , hydrogenation with platinum oxide catalyst ( 2 ) , hydrogenation with a supported copper catalyst ( 7 ) ,and hydrogenation with nickel catalyst a t 130Oto 150" C. (8) and a t 50' to 110" C. (9). During the past war, Chemische Werke Huls (IS) produced Buna-S-type synthetic rubbers with a rated capacity of 4000
INDUSTRIAL AND ENGINEERING CHEMISTRY
1003
NICKEL-CATALYSTS .metric tons per month. The 1,3-butadiene required for this process was produced by the classic five-step synthesis, the acetylene being prepared in the first step by the arc-cracking of methane or of methane-ethane mixtures. In the fourth step, the aldol was hydrogenated to 1,3-butanediol under 300 atmospheres and a t 50" to 150' C. over a supported copper-chromium catalyst.
After preparation, the catalyst was stored under ethyl alcohol, arid was used as a slurry in ethyl alcohol (about 50% nickel by weight). All weights and percentages of nickel given in this paper refer t o the actual weight of the metal,
COMMERCIAL ALDOL
The charge was weighed by difference into the autoclave, and the catalyst added by measuring with a calibrated porcelain spoon. The autoclave was assembled and connected through a reservoir t o a commercial hydrogen cylinder carrying a regulator valve adjusted to give the desired pressure. While heating and agitating, the charge was brought to the desired reaction tempmiture (range of 70' t o 120' C.) in 8 to 15 minutes. Generally, hydrogenation became evident a t about 60" C. ilfter reaching the desired reaction temperature, the rate of hydrogenation was noted periodically by closing the bomb head valve and observing the pressure drop over a short time interval. The reaction time reported is the interval between the time at which the charge reached the reaction temperature and the tinie a t which the rate of hydrogen acceptance had fallen to 1 pound per square inch per minute. The bomb was usually allowed to cool while standing overnight. Duplicate bombs were available for autoclave A and the above procedure caused no delay of subsequent experiments. The reaction mixture was separated from the catalyst by decantation, and the 1,3-butanediol content estimated as described belon.. When it was desired to r e u s e the catalyst, it was left covered with a small portion of the reaction mixture.
The commercial grade of aldol used in this study was purchased under the specifications of containing not less than 90% aldol, and having a pale yellow color and a specific gravity of 1.098 t o 1.105 a t 20' C. This product was said t o meet t h e U. S. Treasury Department's specifications for use in denaturing ethyl alcohol. These latter specifications require t h a t the material, upon addition of excess sodium bisulfite and subsequent iodometric determination of excess bisulfite, shall titrate not less than 90% nor more than llOyOaldol content. A preliminary study of this material by the author and by Henson (6) disclosed the following characteristics: Ignition yielded 0.2% of ash with sodium carbonate being the main constituent. Atmospheric distillation resulted in decomposition ; crotonaldehyde and water were found among t h e products. Distillation under 3 mm. of pressure yielded the following a p roximate fractions: low-boilin (largely acetaldehyde), 25%; alfol, 65%; and viscous residue ?containing some aldol), 10%. T h e low-boiling material distilled over at first, and not during the distillation of the aldol. T h e first 5 % of aldol that distilled over was accompanied b y sublimation of h e , white crystals. This sublimate was found t o be metaldehyde. T h e first 6 t o lOyoof aldol t h a t distilled over was viscous, subsequent aldol fractions were lim id, and all fractions became viscous upon standing overniggt. Decomposition, yielding crotonaldehyde and water, set in when t h e temperature of the heating- bath was allowed t o rise above about 95"-C. Redistillation, under 10 mm. of pressure, of aldol that had been distilled 5 days earlier under 3 mm. of pressure yielded only about 6% of low-boiline: material. about one half of this volatile fraction'geing crotonalduehyde and water. As described, the commercial aldol used in this study had about the same composition as the product synthesized in this laboratory by other technologists by the procedure of Mueller-Cuni adi and Pieroh (10). Based on the observed properties and the results of hydrogenation t o be presented, it is indicated t h a t the commercial aldol ~ 5 % largely 4-hydroxy-2,6-dimethyl-l,3-dioxane,t h e condensation product between aldol and acetaldehyde, as discussed by Goldstein (4). HYDROGENATION AUTOCLAVES
Autoclave A (supplied by the American Instrument Co.) had inside dimensions of diameter, 1.5 inches; depth, 10.5 inches; and capacity, 300 ml. The bomb charge was agitated by a motor-driven rocking device and heated by a n electrical jacket. Autoclave B (manufactured a t The University of Texas) had inside dimensions of 28.4 cm. over-all depth, was cylindrical to a depth of 24.4 cm. with a diameter of 18.3 cm., had a bowl-shaped bottom and a capacity of 7.0 liters. The autoclave jacket was equipped t o take steam, hot water, or cold water, as needed, t o control the reaction temperature. The autoclave cover was fitted with a thermometer well, a hydrogen inlet tube of sufficient length to extend nearly t o t h e bottom of the autoclave, a U-shaped stirrer (motor-driven) with vertical shaft, a s t u f f i g box for the stirrer shaft, and a pressure gage. A lead gasket was used between t h e cover and the top of the autoclave, and the cover was secured with 12 studs and nuts. PREPARATION OF CATALYST
Raney nickel catalyst was prepared from the commercial Raney nickel-aluminum alloy by the met>hodof Covert and Adkins ( 3 ) .
1004
HYDROGENATION PROCEDURE
ANALYSIS OF HYDROGENATED MIXTURES
Except for very abnormal experiments, it was found that mixtures resulting from the hydrogenation of commercial aldol could be analyzed satisfactorily by a relatively simple physical and chemical investigation of the fractions resulting from distillation. In most cases, boiling points and indexes of refraction of the various distillation fractions were sufficient for estimation of the 1,3-butanediol content with a n accuracy of about &l%. The procedure was to distill over as much of the hydrogenated mixture as feasible under water aspirator pressure without allowing the bath temperature t o exceed 95' C. In this phase, cold water was used in t h e condenser and the receiver was surrounded with ice-salt mixture. The residue from the first phase of distillation was allowed t o cool somewhat; then the distillation was completed under 3 mm. of pressure. The 1,3-butanediol content of the reaction mixture was corrected for the ethyl alcohol introduced with the catalyst in order to calculate the percentage yield of 1,3-butanediol. This study was concerned primarily with yields of 1,3-butanediol. For this reason, no particular attention was paid to yields of other individual products of hydrogenation. However, if the yield of 1,3-butanediol were about 70%, then the yield of ethyl alcohol would be 20 t o 25% and the viscous residue would be 5 to 10%. If the yield of 1,3-butanediol fell appreciably below 70010, then the yield of butyl alcohol increased somewhat proportionately. I n the vacuum distillation analysis of most hydrogenated mixtures (especially of the mixtures resulting from hydrogenation a t temperatures below the depolymerization temperature of metaldehyde), sublimation of metaldehyde occurred near the end of the distillation of t h e ethyl alcohol. From this fact, it was concluded that metaldehyde was not hydrogenated or was only partially hydrogenated below its depolymerization temperature. All percentage yields of 1,3-butanediol given here were calculated on the basis t h a t the commercial aldol was pure. Many of the yields would have been essentially quantitative if they had been calculated on t h e basis of the indicated purity of the commercial aldol.
INDUSTRIAL AND ENGINEERING CHEMISTRY
Vol. 44, No. 5
NICKEL-CATALYSTS EXPERIMENTAL
Four experiments were carried out in autoclave A at 80" C. using 130-gram charges of commercial aldol. From the results, which are shown in Table I, it was decided t o use 5 grams of nickel per 100 grams of charge in further experiments.
Table 1.
Effect of Amount of Raney Nickel Catalyst
Hydrogen Prwsure, Lb./Sq. Inch 210
Nickel, Grams
Reaction Time, Hours
210
500 500
Yield 1,3Butanediol, ? '& 71 73 69 73
Varying temperature, pressure, or water content, a series of experiments was carried out in autoclave A using 130-gram charges and 5 g r a m of Raney nickel per 100 grams of charge. The results shown in Table I1 do not include experimental reaction times, for these would be of little significance in view of the varying water content and might be misleading; instead, reaction times calculated as hours per 100 grams of commercial aldol are given and these should be of more value for comparison.
Table 11. Temp., O C. 80 90 100 80 80 80 80
Effect of Temperature and Pressure on Reaction Time and Yield of 1,3-Butanediol (Bomb A, 130-gram charge, 6% Ni) Hz Pressure, Reaction Time,' Yield Lb./Sq. In. Hr./100 G . $Idol l,a-Butanediol, % 100% ALDOL 2.7 1.2 0.8 5.7 3.2 2.5 2.2
500 500 500 110 210 625 875 80%
ALDOL(20% WATER) 7.2 3.4
100 80 90 100
3.0
120 70 80 90 100
4.9 5.3 2.4 1.6 1.4
2.9
4.9
110
600
500
73 72 71 71 71 73 73 53 69 67 62 51 41 71 74 69 69
70% ALDOL(30% WATER)
The data of the first three lines of Table I1 show that, with 100% commercial aldol under a hydrogen pressure of 500 pounds per square inch, increasing the temperature from 80" t o 100" C. results in a significant decrease in reaction time accompanied by a slight decrease in yield of 1,3-butanediol. Probably, as indicated by results shown in the second section of Table 11, still higher temperatures would have resulted in increased reaction time and significant decrease in yield. The results shown in Table I1 for lOOyoaldol for experiments run at 80' C. show that over the pressure range of 110 to 875 pounds per square inch there is little variation in the yield of 1,3butanediol, but t h a t reaction time decreases with increasing pressure as would be expected. I n view of these data, there seems little need of exceeding 500 pounds per square inch unless extremely high pressures can be used. The data of Table I1 for experiments with 80% commercial aldol (20% water) and 200 pounds per square inch pressure show a progressive marked decrease in yield of 1,a-butanediol on increasing the temperature from 80' to 120' C. The reaction time May 1952
decreased from 80' t o 100" C., but increased markedly a t 110" and 120' c. The data of the last four lines of the second section of Table I1 show a progressive decrease in reaction time as the temperature was increased from 70" to 100" C. Slightly lower yields were obtained a t 90" and 100' C. With 70% commercial aldol (30% water) at 90" C., increasing the pressure over the range of 55 to 500 pounds per square inch resulted in progressive decrease in reaction time and increase in yield of 1,3-butanediol. Comparing the data of t h e three parts of Table I1 for experiments at 90" C. and 500 pounds per square inch, it can be seen that increasing the water content from 0 to 30% results in slightly lowered yield of 1,3-butanediol and slightly longer reaction time. In order t o determine the effect of special conditions imposed on the commercial aldol prior t o hydrogenation, several experiments were carried out in autoclave A using 115-gram chargeR and 6 grams of Raney nickel under hydrogen pressures of 500 pounds per square inch. The special conditions and experimental results are shown in Table 111.
Table 111.
Effect of Special Conditions on Reaction Time and Yield of 1,f-Butanediol
Yield 1,3Preliminary Treatment Tzmj., Hz0 Reaction Time Butanediol, of Commercial Aldol % ' Hr./100 G . Aldbl % 0029' in NaOH 90 20 1.3 69 90 20 1.5 0:20% in NaOH 67 0.9'7 in NanSO4, 0.1% in CfIpCOONa 80 20 72 2.8 259r In CHaCHZOH 80 0 75 3.0 2 0 g stripped at 70° C. and 25 mm. 80 2.7 0 88 Fresh1 distilled commercial aldor 80 20 3.1 87
The results indicate that a very small concentration of sodium hydroxide may be permissible, but t h a t larger concentrations may adversely affect the reaction time and yield of 1,3-butanediol. Small amounts of sodium sulfate and sodium acetate have little or no effect. The inconclusive results of a single experiment indicate t h a t the original presence of ethyl alcohol may increase the yield of 1,3butanediol. The yield of 1,a-butanediol from freshly distilled aldol was slightly lower than t h a t from stripped aldol because decomposition was much more extensive during distillation than during stripping, thereby resulting in a considerably larger yield of butyl alcohol in the former case. USE OF HYDROGEN-METHANE MIXTURE. A 132-gram charge of 70% commercial aldol (30% water) with 6 grams of Raney nickel was hydrogenated in autoclave A at 90°C. with a hydrogen-, methane mixture under a total pressure of 1025 pounds per square inch (partial pressure of hydrogen of 450 pounds per square inch). The reaction time was 1.7 hours per 100 grams of commercial aldol and the yield of 1,a-butanediol was 68%. Comparing these data with those of the last two lines of Table 11, it is indicated that the use of a hydrogen-methane mixture will yield about the same results as the use of pure hydrogen under a total pressure comparable t o the partial pressure of hydrogen in the mixture; however, the reaction time will be slightly longer as would be expected. Re-use of Raney Nickel. A 115-gram charge of 80% rommercial aldol (20% water) with 6 grams of Raney nickel was hydrogenated at 100' C. in autoclave A under a hydrogen pressure of 500 pounds per square inch. The reaction mixture was filtered by suction, and the catalyst in the form of a thick slurry was returned t o the autoclave. A fresh 115-gram charge of 80% commercial aldol (20% water) was added and the mixture hydrogenated as before. A similar procedure was used for the third experiment. The results are shown in Table IV. Only the per-
INDUSTRIAL AND ENGINEERING CHEMISTRY
1005
NICKEL-CATALYSTS centages of 1,3-butanediol found in the reaction mixtures are reported, since these are sufficient to evaluate retention of catalytic activity. The percentage of 1,3-butanediol found in the first product was corrected for the ethyl alcohol (6 grams) added with the catalyst to obtain the value of 55y0 given in the table. Probably the loss in catalytic activity resulted from overexposure of the catalyst to air during recovery between experiments; however, the use of aqueous aldol may have contributed to the loss in activity.
Table
IV.
Expt. 1 2
3
Table
Reaction Time and Yield of I,3-Butanediol
Expt.
Effect of Re-use of Raney N i c k e l o n Reaction Time and Yield of 1,3-Butanediol Reaction Time, Hr./100 G. Aldol 1.4 1.9 3.5
1,a-Butanediol in Reaction Mixture, % 55 51
V. Effect of Re-use of Protected Raney N i c k e l on Temp., O C. 80 80-83 80-83 80-83 81 for 1.9 hr. 91 for 1.4 hr. 81 for 1.4 hr. 91 for 1.6 hr. 81 for 1.4 hr. 91 for 1.9 hr. 81 for 1.3 hr. 91 for 2.0 hr. 90-93
Reaction Time, Hours
1,a-Butanediol in Reaction Mixtuoe,
n
2: 1 2.2 2.5 3.3
64
67 68 68 67
3.0
67
3.3
67
3.3
64
2.0
65
Indeterminate reaction time caused by several mechanical failures d the stirring apparatus. a
47
I n another study of re-use of Raney nickel, a 4-kg. charge of commercial aldol with 150 grams of catalyst was hydrogenated in autoclave B under a hydrogen pressure of 600 pounds per square inch with stirring a t 900 r.p.m. After settling, the majority of the product was decanted, and the residual catalyst, protected in a small portion of the product, plus 120 grams of fresh catalyst were used with the next 4-kg. rharge of commercial aldol. Thereafter, only the residual catalyst from the previous experiment, protected in a small portion of the previous product, was used in succeeding experiments with 4-kg. charges of commercial aldol. The results are shown in Table V. The percentages of 1,3butanediol found are sufficient to compare the relative activities of the re-used catalyst. The percentages found in the first two products were corrected for the ethyl alcohol added with the catalyst to obtain the first t n o values given in the table. It can be seen from the results shown in Table V t h a t Raney nickel retains its catalytic activity quite well when protected as described. ACKNOWLEDGMENT
The author thanks E. P. Schoch, director of the over-all project, for his endorsement and release of this paper for publication. He also thanks H. R. Heme for many helpful suggestions, and
acknowledges the help and part,icipation of Melvin 1.Nobles and C. Weldon Chaffin in the procedures for hydrogenation and analysis. LITERATURE CITED
(1) Bayer and Co., Brit. Patent 940 (Jan. 13, 1913). (2) Carothers, \IT. H., and Adams, R., J . A m . Chern. Soe., 46, 1675 (1924). (3) Covert, L. W., and Adkins, H., Ibid., 54, 4116 (1932). (4) Goldstein, R. F., “The Petroleum Chemicals Industry,” p. 292, New York, John Wiley &- Sons, 1950. (5) Halpern, J. H., Monatsh., 22, 63 (1901). (6) Henson, D. D., Progress Report to Bureau of Industrial Cheniistry, The University of Texas, Austin, Tex., March 28, 1941. (7) I. G. Farbenindustrie -4kt.-Ges., Brit. Patent 328,083 (hlarch 9, 1929). (8) I. G. Farbenindustrie Akt.-Ges., French Patent 668,103 (April 27, 1928). (9) Rlueller-Cunradi, M . (to I. G . Farbenindustrie Akt.-Gee.), U. S. Patent 1,907,855 (May 9, 1933). (10) Mueller-Cunradi, M., and Pieroh, K., Ibid., 1,881,863 (Oct. 11, 1932). (11) Neuberg, C., and Kerb, E., Biochem. Z., 9 2 , 96 (1918). (12) The University of Texas, Austin, Tex., Pub.‘Aro. 5011 (June 1, 1950). (13) U. S. Dept. Commerce, Office Pub. Board, R e p t . S o , 189 (1945). ACCEPTEDMarch 8, 1952. Presented a t the Seventh Southwest Regional Meeting of !he A a i r : R I c A s CHEXICAL SOCIETY, Austin, Tex., December 1951. RECEIVED for review January 28, 1932.
C A T A L Y T I C A C T I V I T Y OF N I C K E L BORIDES RAYMOND PAUL, PAUL BUISSON, AND NICOLE JOSEPH Soci&t& des Usines Chimiques RhOne-Poulenc, Paris B y reacting sodium or potassium borohydride with various nickel salts under various conditions of temperature, pH, and solvents, voluminous black precipitates were produced which invariably comprised one atom of boron to two atoms of nickel. These products are neither magnetic nor pyrophoric and do not dissolve as quickly as Raney nickel in hydrochloric acid or potassium triiodide. Their catalytic activity has been studied in comparison to that of Raney nickel in the hydrogenation of safrole, furfural, and benzonitrile. When prepared from nickel acetate or nickel chloride, these catalysts have an activity equal to or dightly inferior t o that of Raney nickel, Their resistance to fatigue appeared particularly high. O n the other hand, by reacting alkaline borohydride with a nickel salt solution containing small quantities (about 2%) of metals of known promoting activity such as chromium, molybdenum, tungsten, or vanadium, a whole series of complex catalysts was prepared
1006
which in a number of cases clearly proved more active than Raney nickel. The simplicity and rapidity of this technique render it highly suitable to the study of promoters (activators).
N T H E course of his work on borohydrides, Schlesinger (4) observed that although the alkali borohydrides reduced a certain number of salts t o the metallic state (silver, mercury bismuth salts, etc.) they gave only the corresponding bolides with nickel salts or with cobalt salts. Since the products thus obtained have a finely divided form and since on the other hand the atomic volume of boron is small compared t o that of nickel, the authors believed they might possess a certain activity in the reactions for which metallic nickel is usually employed as a catalyst. This hypothesis was found to
I
INDUSTRIAL AND ENGINEERING CHEMISTRY
Vol. 44, No. 5