FEBRUARY, 1938
INDUSTRIAL AND ENGINEEIiING CHEMISTRY
3. The change in phase relations occasioned by variation in the ratio of ester to alcohol in a four-component system involving nitrocellulose, ethyl acetate, ethanol, and toluene is depicted by a comparison of the separate-plane phase diagrams prepared from pure ethyl acetate, the 85 per cent ester grade, and a mixture containing 60 per cent ester and 40 per cent ethanol.
Acknowledgment The author takes pleasure in acknowledging his indebtedness to C. 0. Strother who was instrumental in developing the phase diagram method of solvent evaluation in this laboratory, and likewise the assistance of his associates, R. A. Briggs, G. R. Penn, and R. W. Callard in the preparation of the material presented in this paper.
Literature Cited Atsuki and Ishiwara, Caoutchouc & gutta-percha, 28, 15,462-4 (1930). Baker, S. Chem. Soc., 103, 1653-75 (1913). Bingham, E . C., “Fluidity and Plasticity,” New York, McGrawHill Book Co., 1922. ENG.CHEM.,19, 968 (1927). Brown and Bogin, IND. Brunkow, Ibid., 22, 178 (1930).
203
(6) Calvcrt, Ibid., 21, 213-15 (1929). (7) Davidson, Ibid., 18, 670 (1926). (8) Davidson and Reid, Ihid., 19, 977 (1927). (9) Ibid., 20, 200 (1928). (10) Desparmet, C u i r tech., 16, 217-25 (1927). (11) Doolittle, IND. ENG.CHEM., 27, 1169-79 (1935). 112) Doolittle, Smith, and Penn, P a i n t , Oil Chem. Rev.,99, 26-8, 48-9 (1937). (13) Frazier and Reid, IND. ENG.CHEM.,22, 607 (1930). I 14) Gibson and McCall, S. SOC. Chem. I n d . , 39, 172-6T (1920). (15) Hatschek, Emil, “Viscosity of Liquids,” London, Bell and Sons, 1928. (16) Highfield, Trans. Faraday Soc., 22, 57-81 (1926). ENG.CHEX, 17, 505 (1925). (17) Keyes, IND. (18) Ibid., 17, 558-67 (1925). (19) McBain, S. Phys. Chem., 30, 239-47 (1926). (20) McBain, Grant, and Smith, Ibid., 38, 1217-31 (1934). (21) McBain, Harvey, and Smith, Ibid., 30,312-52 (1926). (22) Mardles, J. SOC.Chem. Ind., 42, 127-361’; 207-llT (1923). (23) Masson and McCall, J . Chem. Soc., 117, 819-23 (1920). (24) Park and Hofmann, IND. ENQ.CHEW,24, 132 (1932). (25) Park and Hopkins, Ibid., 22, 826 (1930). (26) Schwarz, Caoutchouc & gutta-percha, 12, 3859-60 (1914). (27) Sproxton, Brit. Assoc. Advancement Science, Third Colloid Rept., pp. 82-9 (1920). (28) Sproxton, T r a n s . Faradau SOC.,16, Appendix, 78 (1921). (29) Wilson, IND.ENG.CHEM.,21, 592 (1929).
RECEIVED December 13, 1937.
Hydration of Propylene under Pressure Isopropanol of high strength was produced by the direct high-pressure hydration of propylene gas with water in the presence of dilute phosphoric acid catalyst. The investigation was studied over a range of 95 to 503 atmospheres pressure and of 160” t o 290” C. The equilibrium was determined over this range of conditions, and the following free-energy equations were obtained for the formation of alcohol : liquid phase AF” = 23.25 Tv vapor phase A€’ = 34.7 Toc.
- 3005 - 2530
Under favorable conditions of operation, liquid-phase concentrations may reach 200
D
EVELOPMENTS in petroleum technology to produce high antiknock fuels have left refineries overburdened with large quantities of gas of high olefin content. When it is realized that slightly less than two million barrels of oil are cracked daily to produce in excess of one billion cubic feet Of gas containing from to 23 per cent (3, 6 , 11) and 5 to 18 per cent propylene (3, 6, 11), the im-
FRANK M. MAJEWSKI‘ A N D L. F. MAREK2 Massachusetts Institute of Technology, Cambridge, Mass.
grams of isopropanol per liter of alcohol solution, and vapor-phase concentrations as high as 700 grams per liter of condensate. The concentrations of alcohol formed in both liquid and vapor phases increased as the pressure increased and the temperature decreased. The rate of reaction was very rapid a t temperatures above 240” C. Of the two possible alcohols, only the is0 comppund was present. In addition to isopropanol, isopropyl ether and polymer were formed as byproducts a t the higher temperatures and pressures. The boiling range of the polymer changed markedly as the pressure, temperature, and reaction time increased.
portance of refinery gas utilization is more appreciated. To date, considerable research has been done in this direction; much of it has been successfully culminated into industrial processes with which we are all familiar. In the present case attention has been focused on the utilization of propylene. 1 2
Present address, R6hm & Haas Company, Philadelphia, Pa. Present actdress, Arthur D. Little, Inc.. Cambridge, Mass.
INDUSTRIAL AND ENGINEERING CHEMISTRY
204
Its direct hydration to isopropanol, employing dilute acid catalysts, h a s b e e n studied with special attention to the effect of temperature and pressure, the rate of the reaction, and the free energy in the liquid and vapor phases. The reaction was studied over a pressure range of 95 to 503 a t m o s p h e r e s and a temperature range of 160" to 306" C.
Thermodynamics I n spite of the importance of hydrocarbon compoundsI there is a surprising paucity:of data in the literature that can be reliably employed in thermodynamics without further calculations. The primary sources of trouble in this case were exhibited by wide discrepancies among data on heats of combustion and the complete absence of specific heat data of gaseous hydrocarbons that might be employed in the temperature ranges covered here. I n 1928 Francis (9) summarized the data available for a few paraffins, olefins, and alc5hols, and derived general equations for the free energy, specific heat, and total heat content as functions of temperature. For the reaction involving the conversion of gaseous propylene and gaseous water to gaseous isopropanol, he recommended the following equation for the free-energy change involved : A F o = 17,150
+ 32.6 T'K.
(1)
Although his treatment was excellent, many of the values given owing to the paucity of the data then available are today considerably altered. The absolute entropy values for gaseous propylene and liquid isopropanol have recently been determined by Parks, Huffman, and Barmore (16, 18, 19). For the change in entropy of these compounds a t 298" K. from the elements, Parks and Huffman gave the values -34.5 and -110.3 entropy units, respectively. For the heat of formation of isopropanol the revised results of Richards and Davis (22) on the heat of combustion of n-propanol (483,100 calories at 25" C.) have been accepted. From this, 4000 calories have been subtracted; according to Parks and Huffman (16,page 110), this is the difference between the normal and is0 alcohols, estimating that the error in this assumption may be as high as 2000 calories. The resulting value recommended by them for the heat of formation of liquid isopropanol from the elements is therefore -76,860 calories a t 298" K. On the basis of Thornsen's value of 492,200 calories for the heat of combustion of propylene a t 298" K. (Zd), the value of 4550 calories for the heat of formation of gaseous propylene from the elements of 298" K. is readily obtained. Thus it is possible to write for 298" K. :
VOL. 30, NO. 2
This value, in spite of assumptions in ascertaining the heat of formation of isopropanol from the elements, is in fair agreement with the value obtained as a result of this work. Ext r a p o 1a t i n g t h e experimental data obtained in the course of this investigation from 1 6 0 " C . , t h e v a l u e -1538 calories is obtained for the free-energy change of t h e r e a c t i o n i n t h e gaseous phase a t 298" K. Correcting Equation 2 to the standard free-energy change for the reaction where the isopropanol and dissolved propylene are both a t unit concentration] we obtain a free-energy change of -990 calories for the reaction in the liquid phase. After extensive correlation of the available data, Lewis and McAdams (14) suggested in 1929 the general equation for all hydrocarbon vapor: C,
=
4.4
+ 4.4 n + (0.012 + 0.006 n) T'c.
Parks and Huffman later developed, largely from a consideration of the specific heat curves obtained from petroleum vapors by Balke and Kay (g), the work of Dixon and Greenwood ( 7 ) and that of Thayer and Stegeman (WS),the equation: C p = 4.0
+ 1.3 n + 0.012 n
Although these equations differ by less than 10 per cent, the latter was employed in determining the specific heat of propylene. For gaseous propane the equation becomes : C,(CsHs)
7.9
=
= 14,800 HzO (1) = -56,560 iso-CaH,OH (1) - 76,860 - (298) (- 110.3) = - 44,000 CaHe (9) H20 (1) = iso-CsH,OH (1) AF 0 2 9 8 0 ~ . = -2240 oal. (2)
+
With the vapor pressure of water as 23.7 mm. and that of the isopropanol as 44 mm. a t 25" C., we may obtain the freeenergy change a t 298" K. for the reaction entirely in the vapor phase.
+ 0.036
TOK.
From a study of the rather scant heat-capacity data, Parks and Huffman (16)suggested for gaseous ethylene the equation : C,(CaH4)
=
6.0
+ 0.015 T"K.
From the latter and the equation for ethane suggested by the Parks and Huffman general equation, we can obtain a A value for the loss of hydrogen of 0.6 0.009 T O K .Applying this to the propane equation, it is possible to obtain a good approximation for the specific heat of gaseous propylene:
+
Cp(CsHegas)
=
73
+ 0.027 T'K
(4)
The specific heat equation of isopropanol vapor was developed in a similar manner. Again the data of Landolt and Bornstein suggested for ethanol vapor the equation : C p = 4.5
A H - T A S = AB'" 298" K. CaHe ( g ) - 4550 - (298) (-34.5)
TOK.
+ 0.038 T'K.
Applying the increment for a methylene group given in the general hydrocarbon specific heat equation, an approximation of the specific heat equation for isopropanol vapor is obtained: Cp(iso-CaHTQHgas)
=
5.8
+ 0.050 TOK.
Employing the accepted specific heat equations, together with the previously determined values for the free energy and heats of reaction a t 298' K., we readily obtain for the
ether and polymer were determined by fractionation in the Cooper-Fasce column ( 6 ) .
reaction in the vapor phase, the following equations: C& (g)
+ H20 (g) =iso-CaH,OH +
Liquid Phase (9)
AH = -10.31T 0.01245T' 0.0674Ta - 11,133 ( 5 ) A F O = 23.74T log T - 0.01245T' f 0.0637T3 - 26.552' - 11,133 (6) where T = temp., K .
Procedure and Apparatus The prime variables of the experimental problem are essentially considered as being pressure, temperature, and time. These are important not only to the extent that pressure and temperature influence the thermodynamic equilibria but that they play an important role in guiding the formation of such byproducts as isopropyl ether and propylene polymers. The apparatus must, therefore, allow for the accurate control of these items and permit the withdrawal of liquid samples a t definite intervals. The apparatus is illustrated in Figure I : The reactor was made of copperlined extra-heavy seamless Enduro KA-2 steel pipe, 1.1 inch (2.8 cm.) inside diameter and 36 inches (91.4 cm.) long with a 30-inch (76.2-cm.) heated section. The reactor was heated by three Chrome1 A coils. Between the coil w i n d i n g s and fastened snugly t o the reactor wall, three thermocouples were distributed. The reactor was mounted on a chassis andmade to rock through an arc of 10" while a 7/8-inch (2.2-cm.) ball rolled up and down within the reactor to induce further agitation. The temperature within the reactor was determined by means of a copperconstantan thermocouple which extended 10 inches (25.4 cm.) into the reactor and was calibrated in place against the vapor pressure of water. P r o p y l e n e w a s supplied to the reactor by means of a 3 / ~ 6 - 1 / ~ 6 inch (4.8-1.6 mm.) copper tube, and the same tube was used for removing samples during the course of the run. In taking liquid samples, the reactor was tilted d o w n w a r d . All liquid samples were passed through a cooler and then t o a buret where, in displacing mercury, the dissolved gas in the sample was allowed to separate from the liquid. No gas samples were removed during a run, but at the end of a run all of the liquid was bled from the reactor until the gas or vapor phase appeared. The condensable fraction of the gas was collected and the residual gas metered. The propylene was made by passing the vapors of 95 per cent isopropyl alcohol over a kaolin catalyst maintainedat 400' C. ($). It analyzed 99.0 t o 99.5 per cent p u r i t y a n d w a s liquefied and stored as liquid propylene in cylinders. Isopropanol was determined by the Ponndorf method (do), and the
0
2
4
6
8
I40
120
100
80
60 I eo
100
80
60
40 20 0 MINUTES X 10
205
-*
IO
12
In the liquid phase the principal reaction was that between water and propylene, resulting in the formation of isopropanol; it was surprising that no normal alcohol was formed. I n e v e r y analysis, however, the absence of the latter was established. T h u s t h e 97102" C. cut on the Cooper-Fasce column (6) showed the[ appropriate constants for water, and fractionations of the aqueous solutions in the reactor showed that the amount, of isopropanol o b t a i n e d corresponded to the total amount found by chemical analysis. The results o b t a i n e d i n t h e liquid phase are graphically interpolated in Figures 2 to 8 and summarized in Table I. The lowest teniperature systematically studied w a s 160-165" C. w h i c h w a s selected because a t lower temperatures the rate was extremely slow. The results for runs covering this temperature and pressure range of 95 to 503 atmospheres and using a 12.1 per cent H3P04, catalyst are illustrated in Figure 2. I n every case except a t 95 atmospheres approximately 20 hours were required t o a t t a i n equilibrium. At the higher pressures there was a characteristic d e c r e a s e i n isopropanol c o n c e n t r a t i o n s with reaction periods in excess of 30 hours. This was explained by the formation of large amounts of side-reaction products in the vapor phase, which lowered the concentration of [isopropanol by decreasing the partial pressure of propylene in the vapor phase. In every case where t h e curve dropped off a t 165" C., some polymer was found in the vapor phase. Although the concentration of alcohol was not as high a t 200210' C., the rates were sufficiently high to attain values within 90 per cent of equilibrium in one hour. The results are shown in Figure 3. It was found unnecessary to conduct these runs any longer than 14 hours to be certain of equilibrium. Consequently the effect observed at long reaction periods a t 160165' C. did not appear in these cases. I n Figure 4 the results of a run at 206" C. and 272 atmospheres with 7.7 per cent H3P04 by weight are plotted in comparison with a run employing 12.1 per cent under iden tical conditions of pres-
206
INDUSTRIAL AND ENGINEERING CHEMISTRY
sure and temperature to illustrate that the difference in acid concentration does not influence the equilibrium although it does alter the reaction rate; for comparison the curve for water has been inserted in Figure 4. This is more clearly illustrated in Figure 5 which shows the results obtained a t 184 atmospheres and 200-210' C. with phosphoric, hydrochloric, and s u 1 f u r i c acids. The values obtained for equilibrium check within 3 per cent. At h i g h e r temperatures (234' C. and above) the reaction was complicated by the appearance of reactions other than the formation of alcohol and ether, specifically polymerization. Consequently, at 503 atmospheres and 246' C. a decrease was observed in the concentration of isopropanol beyond one hour reaction time. The results are illustrated in Figure 6. Except for the lowest pressure, equilibrium concentrations were obtained within 30 minutes; this temperature range would be most suitable from a consideration of rate, equilibrium concentrations, and polymer formation. Figure 7 depicts the results obtained a t 280-290' C. Here side reactions appearing in the vapor phase have a pronounced effect on the concentration of isopropanol in the liquid phase. This effect was most prominent at the highest temperature and highest pressure (290' C. and 503 atmospheres) that would be conducive to polymer formation. The rate of formation of isopropanol is practically instantaneous a t these temperatures. This effect where the isopropanol concentrations decreased with t i m e w a r r a n t ed further investigation; a series of runs was therefore made at 290' C. and 503 atmospheres. It was shown that, if the liquid could be saturated hurriedly with propylene and removed from the reactor, a high concentration of isopropanol and propylene could be obtained. This is less important a t the lower temperature where these side reactions are less pronounced. This point was clearly illustrated in a study of the reaction a t 280" C. using water alone, without the addition of catalyst; after 8 hours very little polymer was formed. As a consequence, the dissolved propylene and isopropanol were found to be higher than were normally obtained when the reaction was catalyzed with acid. The calculation made on tthe free-energy change by these data checked with those obtained when a catalyst was used. The reverse equilibrium or decomposition runs where known high concentrations of isopropanol were put in the reactor are not shown graphically, but the following table clarifies the results and illustrates the reliability of the method of determining the equilibrium by studying the synthesis : Temp. 196 201 210 200 257 276
Pressure Synthesis Run Decompn. Run Aim. Grams iso-CaHIOH/liter of product 184 130 131 272 150 155 188 503 190 123 184 125 184 71 67 50 184 50
VOL. 30, NO. 2
TABLE I. SUMMARY OF LIQUID-PHASE DATA Propylene Dissolved in Run IsoPres1 co. No. Time CaHrOH sure Temp. Soln., Min. Cc. G./Z. Atm. O C . 40 104.7 49 184 205 3.87 215 6.15 123.8 805 124.3 5.85 68 46 6.6 272 207 4.63 15.2 5.6 440 20.1 5.6 800 135.2 45 45 184 210 5.9 250 6.75 135.0 760 6.4 126.0 43 129.0 70 184 209 6.17 205 129.0 ... 123.0 6.02 810 133.0 51 45 201 7.8 503 225 14.5 194.0 825 10.5 201.5 78.4 38 195 2.8 35 95 98.7 4.5 320 110.0 5.0 780 31 90 108.0 272 206 . 240 142.3 9.0 495 133.0 7.6 126.0 272 206 11.4 23 65 12.9 147 140.0 11.6 139.0 302 495 ... 135.0 0 174.0 200 20 184 8.2 410 146.0 10.1 144.0 620 ... 825 137.0 117.0 184 196 11 75 123.0 168 268 128.0 131.0 375 201 ... 0 230.0 272 56 9.1 174.0 490 8.9 1470 161.0 200 .,. 30 57.0 184 8 101 .o 88 150 112.0 220 117.0 210 , .. 0 274.0 503 55 12.3 189 IO 855 14.4 1500 187.0 205 ... 15 18.1 184 6 40 63.3 86.5 80 120 103.0 200 , ,. 147.0 184 16 0 122.0 230 315 123.0 259 ,. 65.4 184 12 45 63.8 120 195 66.5 69.0 240 257 . 147,O 184 0 16 67.2 180 242 9.2 272 68 93.5 24 9.6 160 94.5 6.6 285 93.5 234 2.68 95 38 37.0 39 3.7 50.0 248 3.7 51.5 640 184 31.4 42 57 24.7 300 32.0 502 45.2 732 261 6.37 40 64.5 184 44 6.36 56.0 185 6.73 58.0 485 240 , 272 30 76.7 47 240 , 272 90.5 60 48 246 6.6 503 120.0 45 50 10 1 113.0 200 9.7 100.0 720 280 184 48.9 18 4 ... 51.5 53 ... 52.3 96 7.6 143 50.2 276 , .. 184 0 174.0 22 7.7 50.0 335 8.2 49.8 575 9.8 53.5 710 272 285 6.3 51.1 30 25 6.4 41 5 135 7.3 40.0 325 272 24 0 40 34 40.2 310 50.7 520
Catalyst
7 . 7 HsPOd
,..
1 2 . 1 Hap01
. . ..
...
1725
124.0
...
..
...
..
6.54
1805
132.0
6.35
6.56
1805
127.0
6.10
7.45
E490
198.0
5.63
1610
110.0
4.75
7.85
1965
138.0
8.3
2260
138.0
12.0
. ..
..
...
...
..
. ..
.
..
..
...
...
.
..
..
...
.
10.8
...
..
6.15
% by wt.
. .
.
l/Kc
IsoCaH?OH PropylEquilib- ene Free rtum Soly. EnSeSeergy lected lected
7 . 7 HaPo4 7 . 7 HaPo4
7 . 7 Hap04 7 . 7 HaPo4
1 2 . 1 HaPo4 1 2 . 1 HaPo&
1 2 . 1 HaPo4
6 0
12.5
Decomposition run
Decompasition run
..
. ..
,
.
...
...
...
..
. ..
...
...
..
..
...
..
...
...
..
...
Decomposition r u n
Decomposition run
...
1 2 . 1 Hap01
Decomposition run 13.3 2650 94.0
9.4
1 2 . 1 Hap04
10.4
2360
50.8
3.7
l.5HtSOa
16.5
2980
59.5
6.49
.. ..
... . ..
... . ..
. ..
1 2 . 1 Hap04 12.1 7.7 HaPo4
12.0
2560
112.0
8.8
1 2 . 1 Hap04
21.9
3400
52
7.6
1 2 . 1 Hap04
25.2 3520 51.1 Decomposition run
8.6
1 2 . 1 H3POa
21.9
6.67
...
3430
44.2
FEBRUARY, 1938
INDUSTRIAL AND ENGINEERING CHEMISTRY
207
OF LIQUID-PHASE DATA(Continued) TABLEI. SUMMARY
Propylene Dissolved in Run 180Pres1 Go. No. Time CaHiOH sure Temp. Soln. Min. @./I. Atm. CC. C: 40 20 18.0 95 2.41 273 165 19.2 2.95 530 19.9 3.19 41 65 61.0 272 290 10.0 260 77.0 16.2 430 79.4 15.6 52 65 48.4 503 290 6.2 7.3 190 32.8 460 30.6 7.2 50 41.4 95 163 58 2.61 168 80.1 3.3 4.5 588 122.0 1287 133.5 5.6 50 2.2 49.8 184 59 165 168 80.0 3.7 50 52.0 184 60 161 2.6 166 83.0 3.9 1330 152.0 7.0 2220 137.0 50 71.0 272 164 2.6 61 4.5 166 142.0 6.7 1400 191.0 9.5 2245 170.0 ... 57 0 231.0 272 165 825 171.0 6.6 1540 156.0 7.1 50 98.1 503 165 3.7 62 9.2 180 163.0 11.8 1310 213.0 2160 159.0 0 200.0 503 164 ... 54 10.8 340 175.0 14.8 885 189.5 12.5 1620 195.5 13.1 70 72.0 272 164 26 12.2 180 100.0 15.5 360 109.0 13.3 503 290 64 65 93.0 9.1 503 290 65 68.1 65 6.8 190 65.0 5.9 65 22.5 503 290 66 5.7 190 18.5 4.1 440 10.1 60 13.9 184 138 15 190 31.0 305 38.8 465 43.2 5.25 600 123.0 184 203 14 3.90 95 216 600 74.5 3A 217 7.83 128.0 272 4A 300
Is0
Catalyst
1/Ke
C ~ H I O HPropylEquilib- ene Free rium Soly. EnSeSeergy lected lected
% b y wt. .3 07
1 2 . 1 Hap04
23.5
3440
19.6
None
30.2
3810
80.0
15.9
35.3
3980
31.6
7 2
1430
135.0
5.6
7 . 7 Hap04 1 2 . 1 Hap01
5.20
1 2 . 1 Hap04
. . . . . . . .
...
12.1HsPOa
5.59
1485
152.0
7.0
12.1HaPOa
5.00
1403
180.0
8.1
. . . . . . . .
...
...
7 . 7 Hap04 1 2 . 1 Hap04
Decomposition run 6.00
1563 213
. . . . . . . .
7.7
Decomposition run
11.8
...
...
7.7 HZ04 7 . 7 Hap04 7 . 7 Hap04
18.0 17.5
3240 3200
93 66.5
. . . . . . . .
7 . 7 &Po4
...
13.3 7.95
... ...
12.1HsPOa 1 2 . 1 HaPo4 12.1HaPOa
5.45 7.23 7.96
1610 1920 2020
123 74.5 125
5.48 3.90 7 83
!I 4
10 3
9-
With a knowledge of the concentrations of isopropanol and propylene a t each temperature and pressure, it is possible to calculate the free-energy change for the liquid-phase reaction. The data show excellent agreement, and the following equation was obtained: AF = 23.25T"c.
- 3005
where K , was calculated on the basis of concentration in moles per liter. The actual data are plotted in Figure 8. The
20.
IO 0
-
-0
I
2
3 MINUTES X
4
5
6
standard free6energy change obtained by extrapolating to 25 C. is -2424 7calories as compared with 6- 990 obtained by correcting Equation 2 to 5conditions of unit activity. In this and sub4sequent plots (Figures 9 and 10) the tailed 3triangles and DEGREES CENTIGRADE circles refer to data taken from constantpressure experiments to be described in detail later. Liquid-phase samples obtained below 200" C. possessed a faint odor of isopropyl ether but never enough to detect by fractionationlin the Cooper-Fasce column. The absence of
VOL. 30, NO. 2
INDUSTRIAL AND ENGINEERING CHEMISTRY
208
appreciable quantities follows logically from a consideration of its low aqueous solubility and high vapor pressure. Likewise no acetone could be found in the liquid phase despite thermodynamic conditions favorable for its formation. Attempts to elucidate the mechanism, by which the acid catalyzed hydration on the grounds that an intermediate acid ester was formed, were not successful. Rapid titrations of samples immediately after being withdrawn showed that the concentration of acid catalyst remained substantially the same. End points in this titration were well defined and did not fade. The amount of acid was equivalent to that originally in the reaction before any formation of isopropanol. The measurements served to show that, if the reaction proceeded through an intermediate acid ester compound, the concentration of such an intermediate ester remained low during the run.
Vapor Phase There was a little debate and question a t first as to the existence of two distinct phases in the reactor. It was deemed possible that what was normally regarded as the vapor phase might be a separate layer composed of isopropyl ether, isopropanol, polymer, and water, which existed above the aqueous layer of isopropanol and between the propylene gas. The layer of organic liquid was assumed to be so highly saturated with propylene gas that it must be vapor. Several methods of attack were employed to disprove this concept and to show that only one liquid phase (aqueous isopropanol) and one gaseous phase (propylene, isopropyl ether, and polymer) existed in the reactor. The most direct way was to prove that, a t the temperature in question, what was thought to be a possible intermediate layer was above its critical temperature. This was demonstrated experimentally with various mixtures of isopropanol, isopropyl ether, polymer, and water. Thus the vapor condensate from runs a t 240" C. had am experimentally determined critical temperature of 220" C. ; that obtained a t 280" C. was 252" C. These temperatures would have been considerably lower had they been saturated with propylene gas a t the pressures equivalent to those within the reactor. The second conclusive evidence against it was the fact that, when the amount of metered gas leaving the reactor in the vapor a t the end of the run was plotted against the pressure, almost straight lines with negative slopes were obtained. If an intermediate layer had existed in the reactor, a sharp break in pressure would have been expected a t first until all of the intermediate layer had been released; the curve would have followed a linear function of negative slope. Such, of course, was not the case.s The equilibrium in the vapor phase was determined from a consideration of the composition of the gas bled from the reactor a t the termination of the run after all of the aqueous isopropanol had been removed. At the termination of the run all of the aqueous solution of isopropanol and catalyst were bled out until the appearance of the gas or vapor phase. This transition was easily recognized by the sudden burst of gas issuing from the sample line. This apparently unjustifiable procedure was demonstrated to give satisfactory results by means of check runs made with samples withdrawn a t con-
- --
a If i = initial conditions in the reactor before bleeding, v = conditions in the reactor during bleeding, and m metered gas, then: moles of metered gas, Mm Mi Mr
or Since Vr and V , are equal and oonstrtnt, and Pi is a constant for every experiment, we find on differentiating Pr with respect to Vm:
stant pressure. The condensable fraction in the vapor was removed by cooling the gas to 28" F. (-2.2" C.), and the remainder was metered at room temperature. Analyses had shown that in runs made a t 290" C. this noncondensable gas was pure propylene; this condition was assumed to hold at all lower temperatures as well. The condensable fraction consisted of two layers, the aqueous and the organic. Both layers were analyzed for isopropanol by the Ponndorf method (bo), and the upper or organic layer was fractionated in a Cooper-Fasce column (6),using a salt-ice mixture in the cooler. The range of the column permitted fractionations up to 140 O C., but this was not a serious limitation, inasmuch as samples formed below 250" C. were usually completely distilled before this temperature was reached. In the calculation of the free-energy change of the hydration reaction from these results, attempts were made wherever possible to identify the products or assign to them a reasonable molecular weight and density commensurate with the determined boiling point and refractive index. At runs below 210" C. isopropyl ether could be identified with certainty, but above this reaction temperature the ether fraction of the Copper-Fame column included a polymerization product boiling in the same range. I n these cases a molecular weight, of 80 and a density of 0.7 were assigned to the fraction boiling below the alcohol (up to 75" C . ) . The fractions above the alcohol, except where water was definitely present, gave no true boiling point curve with fractionation. For these compounds up to 140" C. the molecular weight of 110 and density of 0.75 were assumed. This assumption is in agreement with those applied to propylene polymers by Gayer (IO). Although this method of arriving a t the total number of moles of polymer can hardly be called precise, the error introduced by these assumptions was small for several reasons, since by far the greatest part of the total number of moles present was represented by water and propylene. Table 11presents a summary of these data.
TABLB 11. SUMMARY OF VAPOR-PHASB DATA Uncor- CorPresrected rected sure Temp. AF AF Atm. O C . SA 197 276 7210 7140 7A 272 215 4725 4470 6A 184 202 4170 4130 5A 95 217 4790 4870 184 211 4660 4620 45 42 184 263 6470 6420 272 294 7710 7520 30 41 272 251 7510 7340 23 272 207 4600 4330 272 285 6970 6740 25 52 503 290 8100 7510 22 184 276 6200 6150 24 272 242 5700 5480 50 503 246 6640 6000 503 290 7710 7180 65 503 290 7770 7200 64 272 360 9240 9120 32 503 166 4080 3235 62 272 205 4920 4665 35 272 206 4880 4610 31 184 210 4560 4510 43 503 201 4770 4025 51 272 201 4760 4500 56b 503 210 4960 4250 55b 38 95 195 4420 4640 Average a As determined from Equation 6. Run
No.
I
Correoted 1" CsHrOH
Moles 25.08 0.0571 26.08 0.2540 25.63 0.2475 25.42 0.1245 25.47 0.1888 25.72 0.0695 25.70 0.0350 25.65 0.0640 25.75 0.254 26.62 0.0537 25.47 0.0130 27.00 0.0841 26.08 0.156 26.27 0.106 26.08 0.019 26.00 0.085 26.90 0.007 24.84 0.358 24.80 0.180 25.10 0.214 25.63 0.193 25.88 0.291 24.86 0.221 26.17 0.288 24.05 0.120 25.60 b Reverse.
Gas
Fraction
phase CsH7
0.498 0,334 0.285 0,3570 0.3458 0.597 0.551 0.617 0.328 0.445 0.533 0.525 0.303 0.348 0.515 0,462
0.427 0.354 0.389 0.496 0.368 0.270 0.212 0.304 0.346 0.223 0.081 0.276 0.475 0.380 0.185 0.380 0.055 0.208 0.268 0.354 0.302 0.206 0.402 0.247 0.463
HzO
0.800
0.365 0.442 0.374 0.403 0.434 0.315 0.402 0.353
J
Making these assumptions and calculating the equilibrium constants, Figure 9 was obtained, which shows the effect of temperature on the free-energy change of the hydration reaction. The points were slightly scattered because they represent data taken over a wide variation in pressure, and proper fugacity corrections have not been applied; but a t
FEBRUARY, 1938
INDUSTRIAL AND ENGINEERING CHEMISTRY
most the maximum deviation falls within 15" C . of the best curve. The equation for this curve is: AF = 34.7T"c. - 2530
Figure 10 was obtained after fugacity corrections were applied. Here the agreement has been greatly improved. The fugacity data for propylene were taken from charts from an article of Lewis and Luke (IS). These fugacity charts were also employed for isopropanol where direct data were unavailable. The fugacity data for water were calculated from Keenan's Steam Tables (12). The equation for the freeenergy change of the reaction corrected for fugacity became = 33.3T.c.
AFO
- 2370
and by extrapolation we obtain - 1538 calories for the standard free-energy change, compared with -2610 obtained by Equation 4 according to the third law of thermodynamics.
Temp.
1A
203 216 217
C.
4A
Constant-Pressure Experiments I n the course of the work it was felt that the accuracy of the data was handicapped by the method of removing the samples; for, in permitting the pressure to decreaseslightly during the withdrawal of samples, the equilibrium existing between the dissolved propylene and the propylene in the gaseous phase was disturbed. The net effect was to distort the results somewhat so as to yield lower propylene solubilities in the liquid phase and a vapor phase slightly richer in propylene. Although conditions of supersaturation are known to be possible during such short periods as are required to withdraw samples, the danger still existed that the solution might lose propylene; therefore a series of runs was devised in which the apparatus was altered in order that samples might be removed under conditions of constant pressure. For obtaining a liquid sample under pressure, gaseous propylene was heated to the temperature a t which the reaction was being conducted and
TABLE111. COMPARISON OF DATAFOR CONSTANT-PRESSURE AND NORMAL RUNS
Run No.
SA
209
Liquid Phase -Constant-pressure runs-Alcohol per C I H ~per Pressure liter cc. liquid AFO Atm. Grams cc.
184 95 272
123 75 125
5.3 3.9 7.8
1610 1920 2020
--
Normal runs Alcohol CaHp per per liter 00. liquid Grams cc.
7 -
123 89 128
The data of Dohse and Kalberer (8) and of Allardyce ( 1 ) are included on Figure 10. The work of Dohse and KPlberer covers the pyrolysis of isopropanol vapor a t 190-205' C. and 6.7 mm., and was carried out with the necessary precautions essential to this work, such as accurate temperature and pressure measurements. In the work of Allardyce, which covered the pyrolysis of isopropanol a t 360" C. and 1 atmosphere, the temperature reported was that of the electric furnace. The exact gas temperature was unknown; but if we assume a 40' C . temperature drop from the furnace to the alcohol vapor, we find that his data agree quite well with the results of this investigation, in spite of the great difference in experimental approach. The data of Dohse and KPlberer substantiate the work at low temperatures. Although of the same slope, the data are consistently 57.00 calories higher than those obtained by the Francis equation. It is surprising that no acetone was formed in the course of the reaction as predicted by thermodynamic calculations. According to the work of Parks and Kelley (17)and Rideal (21) who passed hydrogen gas saturated with either acetone or isopropanol over a specially prepared copper catalyst, the following free-energy equation was suggested for the dehydrogenation of isopropanol to acetone and hydrogen : AF = 13,500
- 4.OT In T - 3.OT
According to this equation, the dehydrogenation becomes more favorable as the temperature increases, and a t 290 " C. the free-energy change is -2700 calories. The dehydrogenation reaction appears to be largely one involving a proper choice of catalyst. In the work of Allardyce ( I ) , for instance, chemical tests likewise showed that acetone, hydrogen, methane, and aldehyde were absent. However, according to Adkins and his co-workers a t 345" C. over a specially active zinc oxide catalyst, the extent of dehydrogenation was 11 per cent, whereas a t 394" C. it was 20 per cent. It is reasonable to conclude that a negligible amount of dehydrogenation may be expected to take place below 290" C.
5.8 4.3 8.2
Vapor Phase
c-AF'-
AF'
Run No.
1680 2020 2040
5A
Temp. C. 217 202 215 276 a
6A 7A 8A
Constant- Normal Pressure pressure runs runs Atm. 95 4870 4830 184 4130 4350 272 4470 4780 197 7120 6820
was introduced into the reactor, holding the pressure constant, while the liquid sample was removed. In obtaining vapor samples a t constant pressure, dilute isopropanol of equilibrium composition was heated to the reaction temperature under pressure and introduced slowly into the reactor while the pressure was held constant and the vapor sample removed. In either case the total amount of material exchanged was never very large, since the liquid samples were only 20 cc. and the pressure drops to be corrected were rarely over 10 per cent. The results obtained in the liquid- and vapor-phase runs and made with due precaution to hold the pressure constant during the withdrawal of the sample are summarized in Table 111; for comparison the results of the previous runs are also included. Experimental data indicate that there is fair agreement between both experimental methods, and neither possesses any marked trend in being consistently higher or lower than the other. The data are included on Figures 8, 9, and 10, and are designated by the tailed circles and triangles.
Other Products in Vapor Phase Accompanying the reaction that yielded isopropanol was that where the ether was formed; although it was not the purpose of this investigation to examine this reaction completely, a few points are of definite value. At temperatures below 210' C. the identity of ether formed in the vapor phase could be established quantitatively, since the fractionation involved only alcohol, ether, and water. At this and lower temperatures polymers were not formed which prevented accurate fractionation. The isopropyl ether obtained ( p 20/4 = 0.738, p z O o c. = 1.3740) was found to check the constants obtained for the ether sold by the Eastman Kodak Company fairly closely ( p 20/4 = 0.736, ZOO C . = 1.3735). In the fractionation in the Cooper-Fasce column, the ether came over as constant-boiling mixture containing 10 per cent alcohol by weight.
INDUSTRIAL AND ENGINEERING CHEMISTRY
2 10
Several interesting properties were exhibited in the reaction resulting in polymer formation. These appeared above 210 " C. and became more predominant as the temperature was raised. I n Figure 11 the effect of reaction time on the constitution of the polymer is illustrated. These curves represent the results of separate runs made a t 290" C. and 503 atmospheres. Fractionation revealed that a t the end of an hour 60 per cent of the polymer boiled below 70" C. The greatest quantity that could be successfully distilled was 14.5 cc. from 150,
I30 a
110
Y u)
y 90 E
d n 70
less low-boiling material. The fractionation curve obtained from a run made at 360" C. and 272 atmospheres is also given on Figure 12, showing that temperature may serve the same purpose as pressure in effectively increasing the amount of high-boiling constituents in the polymer. At 285" C., 7.7 per cent sulfuric acid was more effective in producing high-boiling constituents in the polymer than a similar concentration or 12.1 per cent phosphoric acid, although they produced approximately the same amount of polymer. Also, 1.5 per cent hydrochloric acid produced more high-boiling constituents in the polymer than 1.5 per cent sulfuric acid, although neither produced as much as 7.7 per cent phosphoric acid. Both catalysts, however, produced approximately the same amount of polymer in the same time period. It is interesting to compare the results obtained on the composition of the polymer in this investigation (at 285" C.) with those of Gayer (IO). He polymerized propylene a t 340" C. and atmospheric pressure with an alumina-on-silica catalyst. The following table illustrates this comparison : .--Authors'
50 0
2
4
6
a
IO
12
id
16
la
Fraction
c.
Density, 20/4
50-58.5
58.5-83.5 64-76 76-100 100-133 133-150 150
0
I
I
I
2
4
6
I
I
8 IO CUBIC CENTIMETERS
I
1
12
1 4 ' 1 6
1
1 I8
VOL. 30, NO. 2
0.705 0.695 0.696 0.711 0.740 0.747 0.806
ResultsRef r aotive Index, 200 c.
-Gayer's
1.3885 1.393 1.399 1.410 1.425 1.430 1.455
Density, 20/4
0.667 0.672 0.693 0.710 0.733 0.751 0,805
Results-Ref raotive Index 200 c:
1.383 1.385 1.402 1.407 1.418 1.425 1.458
The values obtained for the densities and refractive indices indicate that the compounds present in the polymer were not aromatic or complex alcohols, since both possess high densities and indices. Rather, they exhibit the properties of long-chain unsaturates. Although no work was done to determine the extent of unsaturation of each fraction, it might be interesting to do so. Gayer had found that' in most of the cuts the extent of unsaturation exceeded 90 per cent.
Literature Cited an original 20 cc., for above this temperature the polymer decomposed. At the end of 3 hours the polymerization had proceeded farther a t the expense of the lighter polymer, a higher boiling mixture being formed. At the end of 8 hours, more high-boiling polymers formed until very little of the lighter polymer remained. The effect of using 12.1 per cent phosphoric acid in place of the 7.7 per cent acid is also illustrated on Figure 11 ; all other conditions remained the same, and the lower boiling constituents practically disappeared, forming instead a mixture of high-boiling constituents. Clearly, then, the mechanism of polymer formation under these circumstances is, first, a light relatively unstable polymer which decomposes when heated higher than 110" C. and then the conversion of this light polymer into a heavier, more stable mixture boiling a t much higher temperatures. Pressure introduced some interesting results. Figure 12 demonstrates the effect of increasing the pressure from 184 to 272 to 503 atmospheres for runs of the same duration (8 hours) and approximately the same temperature range (276280" C.). At 95 atmospheres insufficient polymer (8 cc.) was formed under these circumstances to distill, but evidence showed that it was very low boiling in character. At 184 atmospheres 70 cc. of polymer were formed, with a large fraction distilling below 70" C. At 272 atmospheres 114 CC. of polymer were formed, but the amount of low-boiling constituents in this fraction was considerably reduced. At 503 atmospheres 155 cc. of polymer were formed, possessing still
Allardyce, Trans. Roy. SOC.Can., [3]21, Sect. 3. 315 (1927). Balke and Kay, IND.ENQ.CHEM.,21, 942 (1929). Bowen and Nash, Refiner Natural Gasoline Mfr., 12, 361 (1933). Brezinski, thesis, Mass. Inst. Tech., 1929. Brooks, Refiner Natural Gasoline Mfr., 12, 359 (1933). Cooper and Fasce, Mass. Inst. Tech., Chem. Eng. Dept., Pub. 222.
Dixon and Greenwood, Proc. Roy. SOC. (London), A105, 199 (1924).
Dohse and Kdberer, 2. physik. Chem., B5, 131 (1929). Francis, IND.ENQ.CREM.,20, 283 (1928). Gayer, Ibid., 25, 1122 (1933). Gerr, Pipik, and Mezhebovskaya, Refiner Natural Gasoline iMjr., 12, 2 (1933).
Keenan, Steam Tables, New York, John Wiley & Sons, 1932. Lewis and Luke, IND.ENQ.CHEM.,25, 725 (1933). Lewis and McAdams, Chenz. & Met. Eng., 36, 336 (1929). Parks and Huffman, "Free Energies of Some Organic Comoounds." A. C. 8. Monograph 60, New York, Chemical Catalog Co., 1932. Parks. Huffman, and Barmore, J . Am. C?iem. S o c , 48, 2788 (1926).
Parka and Kelley, J . Phys. Chem., 32, 740 (1928). Ibid., 52, 4387 (1930). Zbid., 53, 3884 (1931). Ponndorf, 2.anal. Chem., 80, 401 (1930). Rideal, Proc. Roy. SOC.(London), A99, 153 (1921). Swietoslawski and Bobinska, J. Am. Chem. SOC.,49.2478 (1927). Thayer and Stegeman, J . Phys. Chem., 35, 1505 (1931). Thomsen, 2. physik. Chem., 52, 346 (1905). RECEIVZDAugust 6,1936. Presented by F. M. Majewski as a thesis in partial fulfillment of the requirements for the degree of doctor of scienoe at Massachusetts Institute of Technology.
