December 1948
INDUSTRIAL AND ENGINEERING CHEMISTRY LITERATURE CITED
(1) Adam, N. K., “Physics and Chemistry of Sur‘aces,” 3rd ed., London, Oxford University Press, 1941. (2) Atkins, D. C., Baker, H. R., Murphy, C. M., and Zisman, W A., IND.ENG.CHEM., 39,491 (1947). (3) Baker, H. R., Jones, D. T., and Zisman, W. A,, IND.ENG.CHEM.,
in press. (4) Baker, H. R., and Zisman, W. A,, Am. Assoc. Adv. Science Conference, Gibson Island, Md., Aug. 10, 1945. (5) Baker, H. R., and Zisman, W. A., Naval Research Laboratory Rept. P-2474 (February 1944). (6) Beck, L. W., Cluthe, F. S., and Wolfe, J. K., unpublished inves-
tigation, Naval Research Laboratory Rept. P-2785 (January 1946). (7) Bigelow, W. C., Glass, E., and Zisman, W. A., J . Colloid Sci., 2, 567 (1947). (8) Bigelow, W. C., Pickett, D. L., and Zisman, W. 1., Ibid.. 1, 513 (1946). (9) Bishkin, 8 . L., Natl. Petroleum News,35, R-225 (1943). (10) Bowden, F. P., and Tabor, D., Ann Repts. on Progress Chem. (Chem. SOC.London), 42, 20 (1945). (11) Brockway, L. O., and Karle, J., J . Colloid Sci., 2, 277 (1947). (12) Burdon, R. S., “Surface Tension and the Spreading of Liquids,” London, Cambridge Univ. Press, 1940. (13) Dantsizen, C., Trans. Am. SOC.Mech. Engrs., 63, 491 (1941). (14) Denison, G. H., IND. ENG.CHEM.,36, 477 (1944). (15) Kratzer, J., Green, D., and Williams, D. B., S.A.E. Journal, 54, 228 (1946).
2347
(16) Langmuir, I., and Schaefer, V., J . Franklin Inst., 235, 119 (1943). (17) Maroelin, A,, “Solutions Superficielles,” Paris, Les Presses Universitaires de France, 1931. (18) Mattiello, J. J., “Protective and Decorative Coatings,” Vol. 11, p. 626, New York, John Wiley & Sons, 1944. (19) Pilz, G. P., and Farley, F. F., IND. ENC-.CHEM.,38, 601, 1204 (1946). (20) Rideal, E. K., “Introduction to Surface Chemistry,” London, Cambridge Univ. Press, 1930. (21) Sebba, F., and Briscoe, H., J . Chem. Soc., 1940, 106. (22) Sellei, H., and Leiber, E., Lubrication Eng., 3, 16 (1947). (23) Stirton, A. J., and Peterson, R. F., IND. ENG.CHEM.,31, 856 (1939). (24) Stirton, A. J., Peterson, R. F., and Groggins, P. H., Ibid., 32, 1136 (1940). (25) Texas Co., Lubrication, 25, 97 (1939); 28, 13 (1942); 29, 69 (1943). (26) Thomson, G. P., and Cochrane, W., “Theory and Practice of
Electron Diffraction,” Chap. XII-XV, London, Macmillan Co.. 1939. (27) Trillat’, J: J., “La diffraction des Blectrons dans sea applicaGons,” 269, Paris, Hermann & Cie, 1935. (28) Von Fuchs, G. H., I r o n Age, 46 (Oct. 3, 1946). ENQ. (29) Von Fuchs, G . H., Wilson, N. B., and Edlund, K. R., IND. CHEM.,ANAL.ED.,13,306 (1941). (30) Zisman, W. A., J . Chem. Phys., 9, Pt. I, 534; Pt. 11, 729; Pt. 111, 789 (1941). RECEIVED January 21, 1948. The opinions expressed are those of the authors and not of the Navy Department.
FISCHER-TROPSCH COBALT CATALYSTS Influence of Type of Kieselguhrs ROBERT B. ANDERSON, ABRAHAM KRIEG, BERNARD SELIGMAN, AND WILLIAM TARN Central Experiment Station, U . S . Bureau of Mines, Pittsburgh, Pa. Testing data are presented for a series of cobalt-thoriamagnesia-kieselguhr catalysts prepared with a number of commercially available American kieselguhrs. Catalysts containing calcined kieselguhrs had lower activity than similar catalysts with natural kieselguhrs. Acid-extracted natural kieselguhrs produced catalysts of the highest activity. The density of the catalyst varied directly with the density of the kieselguhr, and the distribution of products changed with density of the catalysts, the denser catalysts forming a greater percentage of light hydrocarbons and carbon dioxide.
T
HIS study was prompted by a search for a suitable com-
mercially available American kieselguhr for the preparation of cobalt Fischer-Tropsch catalysts. Although the current development of the Fischer-Tropsch synthesis in this country involves the use of iron catalysts, the data are of interest because some use of cobalt catalysts is contemplated in the production of organic chemicals and because the data give some insight into the role of kieselguhr as a carrier in catalysts. At present a number of catalysts of cobalt and nickel supported on kieselguhr are used in catalytic processes. Kieselguhr is an important component of cobalt and nickel catalysts. I n the German work on Fischer‘Tropsch synthesit it was only after development of catalysts supported on kieselguhr that industrial scale use of the synthesis appeared feasible ( l a ) . Little has been published on the suitability of different
types of kieselguhrs. The work of Frana Fischer (8, 9) established suitable ratios of cobalt and nickel to kieselguhr. German documents and interrogations have described some of the kieselguhrs used (IO). The role of kieselguhr in catalysts has been described by Ries (15), de Lange and Visser (6, 7 ) , and Craxford
(6). Catalysts used in this study were of the cobalt-thoria-magnesiakieselguhr (100:6:12:200) type (3, 11). Tests of this type of catalysts and properties of kieselguhrs and unreduced catalysts have been reported previously (8, 3, 4). Studies of the synthesis with iron catalysts now are in progress. EXPERIMENTAL
Methods of preparing and testing these catalysts, as well as reproducibility of preparation and testing procedures, have been described (3). There it was shown that the testing was reproducible (see also tests 21 and 41 of Table I of this paper), but the reproducibility of catalyst preparation was considerably less (see also tests 19 and 27 and tests 14, 23, and 24 of Table I). All of the tests were made with pelleted catalysts a t atmospheric pressure with a 2 to 1ratio of hydrogen t o carbon monoxide synthesis gas at space velocities of 100 (volumes of gas at standard temperature and pressure per volume of catalyst space per hour). The temperature was varied t o maintain a n apparent contraction of 70%. All of the catalysts were reduced a t 400” C. with a space velocity of dry hydrogen of 3000 for 2 hours and inducted by a slow method as described (3). The catalysts were operated
2348
INDUSTRIAL AND ENGINEERING CHEMISTRY
Vol. 40, No. 12
solved arid reprecipita.ted in catalyst preparat,ion,and the resulting solutiun3 !?-ere analyzed for aluminum, iron, (talciurn, phosphate, and sulfate. Thew analyses are given in Table II. \\'ith catalysts containing natural kieselguhr~ sonic correlation \vas found bet\votrn activit'yand the percentage of removahlo iron in the kieselguhr used as a carricr. The more active cat,alysts contaiiietl kieselguhrs wit,h miallcr amounts of wmovable iron. Tho activit,y was not inhibit,ed by sizable amounts of estyactable aluminum. The other impurities were present in slight amounts and ne correlation was found with activity. The catalysts containing acid-extracted natural kieselguhrs had high activities. n'ith calcined kieselguhrs, II\,fio Super-Cel and Johns-Manville 11, onlv a small amount of impurities n-ai ICFigure 1. Fixed Red Cataltst Testing Units i n t h e Rui*eauof Mines Laboratory, Rriiceton, Pa. movable even though their co1npoi.itions were similar to that of F'i1tc.r Cel. The lower activity of cat'alysts with these kieselguhrs R S continuously for &day periods ( AIoiitiay iioon to Saturday carrit,rs cannot be due to the content of removable iron, but may morning) and then were reactivated by a 2-hour trcatmcnt of be due to the lo^ surface areas or the ordered crystal structurc of hydrogen at, 10" C. higher than the previous opcrat,ing tcmt,he silica or both. Hyflo Supei-Cel gave a sharp x-ray diffraction peratiire followctl by about 10 hours in a slow stream of hydrogen a t about 150" C. hlthoiigh the activity of the catalysts de- . pattern for crist,obalite and Johns-Manville I1 gave a weak cristobalite pattern compared with natural kieselguhrs that were amcreased within each of the operating periods, the average activity orphous or nearly so (4). Aa would be espect.cd, aaid extraction over as inaiiy as fifteen periods didn ot change significantly. The of Hyflo Super-Cel did riot increase the activity of the catalysts catalyst testing units are shown in Figure I . prepared with it. Comparison of activities of catalysts containing The data are not directly suitable for comparing activities; natural kieselguhrs with surface arca of kieselguhr showed no eorhence activities a t 185" C. were computed. I n a previous paper tion. Similarly, no correlation was found between activity and (3)it was shown that, if the contraction was maintained constant, particle size or the external or int,ernal structure of the natural dithe relationship T = . 4 c E I R I ' , where r is t'he rate of synthesis atoms. For example, the activities of catalysts prepared with expressed as cc. (standard temperature and pressure) of 2 to 1 kieselguhrs having the smallest particle size, Snow Floss and hydrogen to carbon monoxide gas consumed per hour per grain of Portuguese, were 86.6 and 1.66.3 cc. per gram per hour, rcspecunreduced catalyst, E has the value of 26 kcal. per mole and A is t i d y ; this is about the maximum difference observed in the testa constant, held over fairly wide ranges of flow and contraction. Since this equation appears satisfact'ory for this and other types of cobalt catalysts, it was used for computation of activities at 185" C. T4BLL 1. z\CTIVITY O F 89 ChTALYbTS WITH L'ARIOl i Several samples of kieselguhr were acid-extracted. Some KIESELGUHRS 4s CARRIERS were extracted with hot' nitric acid as described ( 4 ) . Other Surface Arca Activity" of Kieselguhr, samples were treated by a method recommended by C. C1. Hall Kieselguhrs Catalysts Test Cc./G./H;. Sq. M,,'G. of the British Fuels Research Board; the sample wa3 extracted Flux-calcined b u-ith a solution of hydrochloric and nitric acids (1 volume of conHyflo Super-Cel 89 I, 14 52.4 1 9 Hyflo Super-Cel (acid ex89pi 20 45.2 -1.0 centrated hydrochloric acid, 1volume of concentrated nitric acid, tracted) c HyAo Super-Cel 89 I 23 63.2 1 9 and 2 volumes of water) a t room t'eniperature for 16 hours. The Hvflo Suuer-Cel 89H 24 48.7 I $1 Calc'lnedb sample then was washed with n-ater and dried at 400' C. ACTIVITY A S A FUNCTION O F KIESELGDHR U S E D A S
CARRIER
I n Table I are given the act,ivities of a series of cobalt-thoriamagnesia-kieselguhr (100:6:12:200) catalysts. Those prepared with flus calcined Hyflo Super-Cel were the least active of the series. The preparation containing calcined Johns-lianville I1 was considerably more active than those containing IIpflo Super-Gel but only as active as the least active catalysts containing natural kieselguhrs. The activities of cat,alysts prepared with natural kieselguhrs varied over a wide range. The activities of catalysts prepared with natural kieselguhrs of marine and fresh water origin varied over the same range. Comparison of activities with the analyses of 11-ieselguhrs ( 4 ) indicated that the catalytic activity varied more or 1eF;sinversely with the percentage of iron in the carrier. To study this point further a number of kieselguhrs m r e heated in an 8 AVnit,ric acid solution for 2 hours t o remove any impurities that could be dis-
Johns-Manvillc I1 Natural, inaiino Filter-Cel
Filter-Cel (acid extracted)e Filter-Cel (acid extractedld Snow Floas Katural, fresh water Dicalito, 911 811 (acid exDicalite
tracte'd)a
Dicalit,e. SA6 Dicalite, 65ST PortuRiiese
89BB
53
88.3
5.5
89.1 89J 890 89FF 89DD
21 41 22 42 70
127.4 124.0 166.8 115.5 86.6
22.2 22.2 20.8 24 1 19.1
89V 89GG
38 93
92.7 164.5
29.3 39.2
48 110.7 37.3 4 $1 94 6 23 a 19 166.3 17*5 27 119.8 17.5 German 35 79.6 14.9 Sctivity expressed as cc. (Btandard temperature and yrcssure) of 2 to 1 hydrogen t o carbon monoxide gaa converted pcr gram unreduced catalyst per hour s t 185O C. 6 Methods of preparation of the calcined kieselguhrs were: EIyflo SuperCel heat treated a t abour. 1800' 17. in the presence of a n inorganic fiuxlng agent; Johns-Manville I1 heated a t about 1900' 1,'. w-ithout any fluxing agent. 0 Kieselguhr extracted with hot nitric acid for G hours, washed, dried. a n d h e t t e d at 650" C. for 2 hours; catalysts also contained alkali \vviiiliod magnesia. d Kieselguhr extracted with hydrochloric and nitric acid solution for 16 hours a t room temperature, then washed and dried a t 400' C.
I N D U S T R I A L A N D E N G I N E E R I N G CHEMISTRY
December 1948
ing of catalysts containing natural kieselguhrs. It is possible that differences in surface area and particle size and structure are obscurcd by the influence of iron or other impurities on activity and by the rather wide variability oi the activities of different batches o f catalyst prepared j+ith the same kieselguhr (tests 19 and 27 of Table I). There is a rough over-all correlation between catalytic activity and the surface area of kieselguhr in that catalysts with natural kieselguhrs have the highest activity, those with calcined kieselguhrs lower activity, and preparations with the flux-calcined support have the lowest activity. I~ISTRIBUTIONOF PRODUCTS A S A FUNCTION O F KIESELGUHR CARRIER
The distribution of products appears to be fairly independent of the type of kieselguhr used as a carrier, and dependent chiefly on the density of the catalyst. However, the densit'y of thc catalyst depends on the densit,y of the Carrier. Wit>hgranular catalysts the linear relation between mercury densities (determined by the displacement of mercury at at,mospheric pressure) of cat'alyst and kieselguhr is good (W),but the density of pelleted catalysts depends on both the density of kieselguhr and the pelleting procedure. However, with the same setting of the pelleting machine the density of the pellets will vary directly with the density of the carrier. The variation of product distribution with mercury densit,y of the catalyst is shown in Figure 2. With increasing mercury density of catalysts the yields of all of the less desirable products, CH,, C3 3. Cd, Cl - Cd, and COZ, formed per cubic meter a t the same contraction and same space velocity increased at the expense of the yield of liquid products. The points are scattered because of the many variables involved in preparing and tcst,ing catalysts, but the trends are obvious. T o show the general tendency the equation, log Y = UP 6, where Y is the yield in grams per cubic meter, p the mercury density of the catalyst, and a and b constants, was fitted to the data by the method of least squares. The curves in Figure 2 are plots of this equation. The values of a, the relative change in yield with respect to mercury density, are 0.0784,0.0900,0.11054, and 0.4417 for Cs Cp, C1 - C4, CH1, and CO,, respectively; this indicated that the yields of COz and CH1 are most sensitive to changes in mercury density. It is possible that a part of these changes in product distribution is due to changes in flow of synthesis gas per gram of catalyst. However, if this effect occurs, it should be counterbalanced by the fact that the denser catalyst usually operates a t the lower temperature and shodd give lower yields of light hydrocarbons and carbon dioxide.
+
+'
DISCUSSION
In the literature there is some indication that the role of kieselguhr in the cobalt and nickel catalysts is more complex than that of a diluent or bulking agent (6, 7 , 15). I n fact, de Lange and Visser presented x-ray diffraction and electron microscope evidence for the formation of hydrosilicate bonds between the nickel oxide and kieselguhr. Similar studies by the Bureau of Mines (12, I S ) show no evidence of the formation of hydrosilicates, but these data do not preclude its occurrence to a limited extent. I n previous work ( 8 ) it has been found that unreduced catalysts have higher areas than the sum of the areas of cobalt oxide-promoter complex prepared separately and kieselguhr; the areas of catalysts with calcined and flux-calcined kieselguhrs were about 8% greater and those Tsith natural kieselguhrs about 207, greater. These results and data on reduced cobalt catalysts ( I ) , as well as tnose given in the present
paper, indicate that the functions of the carrier are more complex than that of a diluent or bulking agent, possibly owing to formation of cobalt hydrosilicate. The role of the carrier as a bulking agent is nevertheless important, since catalysts containing kieselguhis do not change appreciably in bulk volume on reduction, but the bulk volume of preparations nithout kieselguhrs deereasps severalfold on heating and reduction ( 1 ) . Recently, Craxford ( 5 ) presented data on the effect of kieselguhr in cobalt - thoria-magn e s i a-k i e s e 1g u h r catalysts, 11-hichare in many respects similar to the data in the present paper. He found that catalysts prepared with natural kieselguhr had surface areas and activities greater than those of a catalyst containing calcined kieselguhr. Comparison of data pre-
2349
20
IO
1
0 aw: 40 I-
w I
0 m
3 0
[r
W
30
a v,
5
(r (3
20
IO
00
04
I 1.2
MERCURY DENSITY, GRAMS PER CUBIC CENTIMETER
Figure 2. Variation of Product Distribution with Mercury Density of Cobalt-Thoria-Magnesia-Kieselguhr Catalysts Operated a t Atmospheric Pressure with Contractions Maintained a t 70% Yield data are averages of several operating periods; vertical lines indicate standard deviations of the means. curve 1s plot of log Y = ap + ' b
TABLE11. ANALYSISOF IMPURITIES EXTRACTABL~ FROM KIESELGUHRS WITH ACTIVITY COMPARED
Ai 0.05
Kieselguhr HyfloSuper-Celb
+ +
Hyflo Super-Cei (acid extracted) C Johns-ManvilleIlb Filter Cel
+
0.00
0.17 1.23
Weight % Extractable Fe + C a + 7 Pons SO1= Catalyst 0.03 0.00 0 00 0 . 0 0 89L 891 89H 0.00 0 . 0 0 0.00 0.00 89N +
0.04 0.75
0.00 0.12
0.13 0.09
0.09 0.11
89BB 895
Piltor Pol -1.111
Activitya
Test 14 23 24 20
Cc./G./Hr'.
53 41
21
88.3 127.4 124.0
22 92 70 38 93
166.8 115.5 86.6 92.7 164.5
52.4 63 2 48.7 45.3
-_I
Acid extractede Acid extractedd Snow Floss Dicalite 911 Dicalite 911 (acid extracted)n Dicalite 658T Portuguese
'
0.00 0.00 0.80 1.41 0.00
0.70 1.14 0.00
0.28 0.35 0.00
0.00 0.00
0.00 0.00 0.15 0.00 0.00
0.76 1.61
1.77 0.65
0.14 0.00
0.00 0 . 2 3 0.00 0.11
0.00 0.00
0.00 0.00 0.12 0.26 0.00
890 89FF 89DD 89V 89GG 892 89K
49 94.6 19 166.3 27 119.6 35 79.6 German 0.43 1.49 0 . 0 2 0.00 0.56 Activity expressed as cc. (Etandard temperature a n d pressure) of 2 t o 1 hydrogen t o cabon monoxide gas converted per gram of unreduced catalyst per hour a t 185O C. b Methods of preparation of the calcined kieselguhrs were' Hyflo Super-Cel, heat treated a t about 1800° F. in the presence of on inorganic fluxing i g e n t ; Johns-Manvilie I1 heated a t about 1900' F. without any fluxing agent. C Kieselguhr extracted with hot nitric acid for 6 hours, washed dried, a n d heated a t 650° C. for 2 hours: catalysts also contained alkali washed magnesia'. d Kieselguhr extracted with hydrochloric a n d nitric acid solution for 16 hours a t room temperature, then washed and dried a t 400° C.
"s",8
I N D U S T R I A L AND E N G I N E E R I N G CHEMISTRY
2350
sented here with Craxford’. indicates that the extent of interaction of cobalt and kieselguhr n a s considelably greatei in Ciaxford’s catalysts. This may br due t,o longer tinies of contact of the cobalt nitrate-promoter solution x-it,li kieselguhr or other variations in catalyst preparation. Our data do not, permit any esplttriat,ion as i o how the iron content of the kieselguhr deerrased the activitv of the catalyet. I n preparation of catalysts such as these it appears desirable to use a natural kieselguhr that has ‘been acid-extracted. The choice of this kieselguhr might depend upon ihe density of the catalyst, desired. ACKYOWLEDGMEhT
The authors wish t o give acknonledgnient to man) persons for assistance in various phases of the work, especiallv to W. Dieter and 1s’. Oppenheimer for analytical data, to H. Hemllett .. and K. Stein for catalysl preparation, to It. A. Fricdel and staff for mass spectrometric analyses, and to C . DeI,euac, A. J h d a s h .
Vol. 40, No. 12
It Kelly, 15. O’Neill, L. Shaw, and \I’Thomas of the crew of operators. LITERATURE CITED
Anderson, Hall, and Hafer, J . Am. Chem. Soc., 70, 1727 (1948). Anderson, HalI, Hewlett, and Seligman. Ibid., 69, 3114 (1947). Anderson, Krieg, Seligman, and O’Neill, IND.ENG.CHEM.,39, 1548 (1947). Anderson, McCartney, Hall, and Hofer, I b i d . , 39, 618 (1947). Craxford, Fuel, 26, 119 (1947). De Eange, private communication. De Lange and T’isser, Ingenieur (Utrecht) 58, 24 (1946). Fisoher and Koch, Rrennstof-Chem., 13, 61 (1932)” Fischer and Meyer, Ibid., 12, 225 (1931). Hall, Craxfovd, and Gall. “Interrogation of 0. Roelen,” British Intelligence Objectives Sub-committee, 1945. Hall and Smith, J . SOC.Chem. I n d . (London), 65, 128 (1946). Hofer, unpublkhed work. McCartney, unpublished n-ork. Pichler, U . S.B U Y Mines, . Tech. P a p e r , in press. Rim, J . Am. Cham. SOC.,67, 1242 (1945); J . Chem. Ph,ys., 14, 465 (1946); Ind E n g . Cham., 37, 310 (194-5). RECEIVED October 24, 1947. Published by perniission of the director, B~~~~~ of hfines, u, s. Department of the Interior,
J R I C H i R D N. Wl[LRELXI AND DONALD W. COLLIER’ Princeton University, Princeton, .V.J . Vapor-liquid eyuilibriuni measurements of the system butadiene-styrene were performed at 0 C. in butadienerich and styrene-rich solutions. The Hildebrand-Scatchard equation correcting for deviations in the liquid phase from Raoult’s law as well as from the gas laws was found to fit the data satisfactorily. This equation was used to compute equilibria from 15’ to 80 O C. for the full range of compositions. Partial pressures, partial heats of mixing, and z - y relations are given.
-
APOR-liquid equilibria of the butadiene-styrene system are important as thermodynamic background in the manufacture of the GR-S polymer. These data cannot be estimated simply by the use of Raoult’s law because there is a significant heat of mixing of the liquids as well as a deviation of vapors from the perfect gas law. Direct measurements of the equilibria are subject t o error because of a tendency to polymerize, particularly above 2 5 ” C. This paper on butadiene-styrene equilibria prespnts one of several studies on the subject of monomer recovery from GR-S latices that were undertaken by Princeton University a t the request of the Office of the Rubber Director, acting through the government-owned Rubber Reserve Company. The method of procedure was t o select first the most probable thermodynamic equation relating vapor and liquid compositions; secondly, to verify this equation by equilibrium measurements conducted a t 0 ” C.; and, finally, using the selected function, to compute the desired equilibria over a wide range of conditions in which direct experiment was not feasible. The work is reported in these three steps. 1 Present
address, The Sharples Corporation Research Laboratories
Philadelphia, Pa.
THEORY
The fundamental equilibrium equation for a liquid mixture is given in terms of the partial molal free energy of mixing:
Each of the terms of this equation may be defined for the component butadiene as follows:
For a regular solution in which the orientation of molecules may be assumed to be completely random, the partial entropy nf mixing is:
The further assumption regarding the butadiene-styrene system is made that it is nonpolar, nonassociating, and that any deviations from Raoult’s law are due to a heat of mixing resulting from differences in molecular volumes of the two components. The Hildebrand-Scatchard relation ( 1 ) for the partial heat of mixing based on such a model then may be written for butadiene in styrene:
This equation is derived using the assumptions that: Separation of like molecules in the mixture is the same as in the pure liquid; attractive forces between like molecules in the mix-
