T H E ADSORPTIOS OF WATER FROM ETHYL ALCOHOL-KATER MIXTURES BY SILICA GEL* BY HESRY hlAUZEE DAYIS A S D LLOYD E. SWEARINGEN
Introduction The problem of producing ethyl alcohol of more than 94 to 95%; purity from mixtures of the alcohol and water has been the object of many investigations. The problem reduces almost entirely to one of devising a method for breaking up or changing the composition of the azeotrope formed by these tvio substances. Both chemical and physical methods have been used to accomplish this result. Of the chemical procedures for removing water from alcohol-water mixtures, distillation from quick lime or sodium are the most common. The nearness of the constant boiling temperature of the azeotrope to the boiling point of pure alcohol makes ordinary fractional distillation impractical. Young and Fortey,' by adding benzene to the alcohol-water azeotrope converted it into a lower boiling point ternary system. Alcohol of high purity is recoverable from this ternary system on distillation. Wade and Merriman,' making use of the fact that the composition of the azeotrope is influenced by pressure, have succeeded in producing alcohol of high purity by fractional distillation under reduced pressure. Thus, starting with an alcohol-water mixture containing j.1 4 7 water, ~ after ten fractionations under pressures of 57-59 mm. mercury, only 0.95% water remained. Grimm and W ~ l f f and ,~ Grimm, Raudenbusch and Wolff' have successfully broken up azeotropic mixtures of ethyl alcohol and carbon tetrachloride by adsorption processes, using silica gel as the adsorbent. Since it has been demonstrated that the composition of an azeotropic mixture can be changed by the preferential adsorption of one component of the mixture, this procedure should yield results with the alcohol-wat#ersystem. Since silica gel is know to have a high adsorptive capacit'y for water, it should be a good medium for the removal of water from the alcohol-water mixtures. The adsorption of fluids by silica gel is generally conceded to be a process of capillary condensation.5 Within the minute capillaries of this adsorbent, the surface of any condensed liquid must have a radius of curvature about equal to that of the capillary in which it stands. The vapor pressure of a liquid in a capillary tube is lower than the vapor pressure of the liquid over * Contribution from the Chemical Laboratory, University of Oklahoma.
* J. Chem. 9
Soc., 81, 717, 739
(1902).
J. Chem. SOC., 99, 997 (1911). 2. angew. Chem., 41, 98 (1928). 2. angew. Chem., 41, to4 (1928). Patrick and Eberman: J. Phys. Chem., 29, 2 2 0
(192j).
.iDSORPTION O F W'ATER BY SILICA GEL
P the plane surface, The Kelvin equation, In - = P O
-
I309
2 uv affords a means RTr ~
of predicting the magnitude of the vapor pressure lowering in small capillaries. Po and P are the vapor pressures over the plane and concave surfaces respectively, at temperature T, R is the gas constant, u the surface tension, r the radius of the capillary and v the molar volume of the liquid. Shereshefsky' has shown that the vapor pressure lowering for water in small capillaries is much greater than can be accounted for by the Kelvin equation. His results indicate that this lowering of the vapor pressure is due to the increase in surface tension of the water. The Kelvin equation predicts that the greater the surface tension of the liquid, the greater will be the reduction in vapor pressure. That is, P:Po is proportional to e-u, The surface tension of water will vary from three to four times greater than that of ethyl alcohol between o°C. and the boiling point of alcohol. The molar volumes of water and alcohol, near the boiling point of alcohol will be roughly in the ratio of 18 to 63. The radius of the capillaries being the same for both water and alcohol, then the lowering of the vapor pressures for each of these liquids should be proportional to their respective (o-v 'T) values. It is seen that the (av/T) value for water is slightly greater than the (o-v/T) for alcohol. Consequently the vapor pressure lowering for water in the capillaries of silica gel should be greater than the vapor pressure lowering for alcohol. It follows that the component suffering the greatest vapor pressure lowering will be more readily condensed. The preferential adsorption of water from an alcohol-water mixture, which this treatment predicts, is somewhat meager. It is further complicated by the fact that the heats of wetting of silica gel by alcohol and water are slightly different. Grimm, Raudenbusch and Vrolff2 report the heats of wetting of silica gel by alcohol and water to be 23.7 and 20.95 calories per gram, respectively. However, it was believed that if the proper conditions could be discovered, the passage of vapor from the alcohol-water mixture over silica gel should result in the adsorption and removal of a t least a part of the water vapor from the mixture. Subsequent condensation of the vapor phase remaining should yield an alcohol-water mixture enriched in alcohol.
Experimental The general plan for attacking the problem may be stated in the following manner. The vapors from the boiling alcohol-water mixture were passed through a column of silica gel, maintained a t some suitable temperature above the boiling point of alcohol and below the boiling point of water. The surviving vapors were then condensed and collected in I O C.C.portions. The composition of each of these portions was determined by specific gravity methods. Fifty cubic centimeters of the alcohol-water mixture were used in each experiment. Fifty-six grams of silica gel, prepared by Patrick's method, 2
J. Am. Chem. SOC.,50, 2966 (1928). LOC.cit.
1310
HENRY MAUZEE DAVIS AND LLOYD E. SWEARINGEN
were used in each experiment, except as otherwise noted. The silica gel was packed loosely in a straight piece of glass tubing, 45 cm. long and 2 cm. in diameter. The gel column was kept at the desired temperature by means of a water bath. The gel was activated by heating to 130'C. under a pressure of from 25-28 mm. of mercury.
The Effect of Temperature on the Adsorption Process. The effect of temperature on the adsorption process is complicated somewhat by the fact that considerable quantities of heat are evolved during the adsorption process. The data in Table I show that the temperature of the adsorbent may vary from 2' to IS'C. above the temperature of the constant temperature bath. These experiments were carried out with a thermometer imbedded in the gel column, about midway of the column. The temperature of the gel column and that of the constant temperature bath were compared several times during the distillation of an alcohol-water sample. These experiments were made on an alcohol-water mixture containing 7.4'34 water. A typical set of data are given in Table. I
TABLE I The Variation in the Temperature of the Adsorbent due to Heat of Wetting. Time of Reading
3:55 P.M.
3 :58 4:or 4 :06 4 :09
4:18
Temperature of Bath
Temperature of Adsorbent
Remarks
82 C. 81.8" 82.0' 81.7'
82°C. 96' 97'
Vapors enter gel column. Vapors emerge from gel column.
82 .oo
85'
82.0'
84.5'
82.0'
84.0'
90'
Temperature of adsorbent never below 84'C.
Table I1 and Fig. I give the results of experiments carried out for the purpose of determining the optimum temperature for water removal from a given alcohol-water mixture. The samples used in these experiments contained 7.470 water, weight percent. All conditions except temperature were maintained uniform throughout these experiments. From Table I it is seen that the heat of wetting is sufficient to keep the gel a t least 2'C. above the temperature of the bath as long as vapors are passing through. I n experiments number 8 to 12, Table 11, the bath temperature was kept below the boiling temperature of the mixture, with the expectation that the heat of wetting would keep both components largely in the vapor phase. However, due to the fact that the walls of the tube containing the gel remained a t the temperature of the bath, a considerable amount of purely thermal condensation occurred. This condensate was trapped in
1311
ADSORPTION O F WATER BY SILICA GEL
TABLEI1 The Effect of Temperature of the Adsorbent on Water Removal Experiment Kumber
Bath Temperature
Beat Fraction
2
0.76 1.12 I . 14
2
1'
I
I
1.76 2.39 2.71
I
2.20
I
11.
76.5 75.4 74.5
12.
72 0
I
2.17 2.31 2.62 2.62
I. 2 .
3. 4. 5. 6. 7.
8. 9. IO.
91. j"C. 86. j 84.6 8.3 . 5 82 . o 80.0 78.0
2
77.2
Maximum Percent Water lost
I
I
I I
44
Percent of Total Sample retained by Gel
16 16.2 22.8 24.8 .... 2j.8 24.0
34.4 41 .o 48.8 54.0 66.0
the gel column. Data given in Table I1 show how important purely thermal condensation becomes a t temperatures below 78OC. Table I1 makes the meaning of Fig. I rather apparent. At temperatures well below the boiling point of the mixture, (Ex. No. 11 and 1 2 ) the percent of ater lost (in the liquid phase) is approximately constant. As we approach the boiling point of the mixture, (Ex. No. IO, 9 and 8) while we have increasingly larger quantities of vapor phase present, the liquid present apparently effectively prevents it from being adsorbed to any appreciable extent. Once the boiling point of the mixture is passed, the amount of water adsorbed increases rapidly to a maximum, just above the boiling point of pure alcohol. When considerable quantities of both liquid and vapor phases are in contact with the gel, the amount of water removed is smaller than for either of the phases present in large excess. The amount of water removed again falls off as the boiling point of water is approached. The results indicate that the removal of water from the alcohol mixtures is more pronounced in the vapor than in the liquid phase. The optimum bath temperature for water removal is 78°C.
The Effcct of Water Content of W t x t u r e s on the A m o u n t of Tl'ater adsorbed f r o m Alcohol- Water Mixtures. Data for other alcohol-water mixtures, including the true azeotrope are given in Table I11 and Fig. 2 . These data were obtained in experiments similar to those used for the data in Table 11, except that all conditions except composition were maintained constant. The composition of the true azeotrope for the particular pressure desired (737.4 mm.) was obtained by interpolation of the data of Wade and Merriman. It contained 9j.62TG alcohol and boiled with utter constancy a t 77.26"C. The gel column was maintained a t 78" to 78.2"C.
1312
HESRT lhIhUZEE D A W S A S D LLOYD E. SWE.4RISGE9
FIG.2
ADSORPTIOS O F WATER BY SILICA GEL
13'3
TABLE 111 The Relation of Water Content of Mixture to Water adsorbed. %'eight Percwit .4lrohd in Mixture
True Azeotrope.
Perctwt
Water Lost
92.6 94.8 gj.6~~ 99.0
3
1.95 1.72
0.33
Change in Composition of Condensate during Prqreas of Distzilation. The results given in Table I V show how the compositiori of the condensate changed during the progress of a single distillation. Fifty cubic centimeters of the true azeotrope were distilled over 1 5 5 grams of gel, the gel temperature being maintained a t 7 8 O C . The successive I O c.c. portions collected showed the following compositions. TABLE IP The Relation of Alcohol Content to Fraction col!ected. Frartiort Sunlber I. 2.
3. 4.
Alccihol Content
Water Rmovrd
9y.11C; 97.82 96,8Y
3.435;
96.2j
o.b3C;;
2 , ?G
1.27
E$ect of T'arzntl'on
i:i Quaiitl'tU 0.f Gcl. Experiments similar t o those recorded in Table 11 were carried out with I j j grams of gel, all other factors being maintianed the same. The maximuni amount of water lost with this larger quantity of gel was 3,445, as compared with 1.72Vc. The azeotropic mixture, with g5.6zc,*C alcohol was used in these experiments. With 2 . 7 6 times more gel present, the water removed is exactly twice as great. Summary I, Water may be preferentially adsorbed from alcohol-water mixtures by silica gel. 2. Adsorption of wit,er from the mixtures is greater when adsorption takes place from the vapor than from the liquid phase. 3. The optimum temperature for the preferential adsorption of water vapor from the mixtures has been found to be 78°C. 4. The fraction of the total water present which is removed by adsorption decreases as the total water content of the mixture decreases. 5 . The true azeotrope was studied and one fraction of the condensate (about zoyG of the quantity started with) recovered showed an alcohol content better than 99%. 6. The adsorption of water increases with an increase in quantity of adsorbent, but not as a direct proportion.