July 1951
INDUSTRIAL A N D E N G I N E E R I N G CHEMISTRY
illdieate either that the asphalt undergoes a slight hydrogenation initially, which is followed by a partial dehydrogenation, particularly a t higher temperatures, or, alternatively, that the variations in hydrogen content reflect the heterogeneity of the asphalt fraction, some-constituents being more easily hydrogenated and split than others. ACKNOWLEDGMENT
The authors are grateful to R. A. Friedel for furnishing the and to Joseph Lederer Of gas ‘pectrometer for the ultimate analyses of products. LITERATURE CITED
(1) Davies, 0. L., “Statistical Methods in Research and Produc-
tion,” pp. 152, 252, London, Oliver and Boyd, 1949.
1525
(2) Fisher, C. H., Sprunk, G . C . , Eisner, A., O’Donnell, H. J., Clarke, L., and Storch, H. H., Bur. Mines Tech. Paper 642
(1942). (3) Kata, M., and Glenn,
R.9., Division of Gas and Fuel Chenllmtry, 115th Meeting, AM. CHEM.SOC., San Francisco, Calif., 1949. (4) Pelipetz, M. G., Kuhn, >I., Friedman, S., and Storch, H. H., IND.ENG.CHEM.,40, 1259 (1948). ( 5 ) Weller, S.,Clark, E. L., and Pelipetz, M. G., I b i d . , 42, 334 (1950). RECEIVEDMarch 1, 1950. Presented before the Division of Gas and Fuel Chemistry, Symposium on Kinetics of Coal Hydrogenation, a t the 118th Meeting of the AMIBRICAA CHEMICAL. SOCIETY,Chicago, Ill. For material supplementary to this article (tables showing elemental balances for all the hydrogenation runs) order Document 3117 from the American Documentation Institute, 1719 N St., N.W., Washington 6, D. C., remitting $1 for microfilm (images 1 inch high on standard 35-mm. motion picture film) or $1 for photocopies (6 X 8 inches) readable without optical aid.
(Kinetics of Coal Hydrogenation)
CONVERSION OF ANTHRAXYLON SOL WELLER, M. G. PELIPETZ, AND SAM FRIEDMAN U . S. Bureau of Mines, Rruceton, Pa.
T h i s study was undertaken in an effort to acquire additional information on the reaction mechanism of coal hydrogenation, and, in particular, to determine whether asphaltene (material soluble in benzene but insoluble in n-hexane) is an intermediate in the conversion of coal to distillable oil. A batch autoclave stud)- was made of the kinetics of hydrogenolysis of bituminous coal anthraxylon. The range of conditions included conversion of coal to a yield of 8Oa/, asphaltene and 8qo oil at one extreme, and a yield of 6qo asphaltene and 76Yo oil at the other. The conversion
of coal to oil can be quantitatively described by the reaction scheme, coal +asphaltene -oil, both reactions being of first order, kinetically. Water and gas are by-products of both reactions. The results furnish eiidence that asphaltene is an intermediate product in coal hydrogenation, and provide quantitative data on the rates of the coal-to-asphaltene and asphaltene-to-oil reactions. This information will be useful in understanding the nature of the coal hydrogenation reactions and, perhaps, in permitting more ratimal design of practical processes for the over-all conversion of coal.
P
gen, 0.8% sulfur, 7.6% oxvgen (by difference), 1.4% ash, and 1.4y0 moisture. The pure asphalt previously studied ( 3 ) was isolated from a pilot plant run in which coal from the same mine was hydrogenated; the results of the experiments with purr asphalt should, therefore, be applicable t o those with the parent coal reported here. The experimental procedures and methods of calculation wcr(identical with those given in detail in previous papers ( 3 , 4). The charge in each experiment was 50 grams of anthraxylon (48.6 grams of moisture- and ash-free coal), 0.5 gram of stannous sulfide, and 0.25 gram of ammonium chloride. No oil vehicle was used. The initial cold hydrogen pressure was always 2500 pounds per square inch gage. In no experiment did the hydro en consumption exceed 177, of the initial hydrogen charge, so t k a t the experiments can be considered to have been made a t essentially constant pressure. The term “liquefaction” employed subsequently in this paper is defined as 100 minus per cent organic benzene-insolubles.
REVIOUS data ( 1 ) have indicated that, asphalt may be a n
intermediate product in the over-all hydrogenolysis of coal to distillable oil. (The term “asphalt” is used here to designate material soluble in benzene but insoluble in n-hexane.) To demonstrate the validity of this hypothesis and t o gain additional information about the reaction mechanism, a batch autoclave study has been made of the kinetics of the hydrogenolysis of anthraxylon. -4s a preliminary to this investigation, the kinetics of the hydrogenolysis of pure asphalt were studied (3). It has been found possible to interpret the data on coal hydrogenolysis quantitatively in terms of successive first-order reactions involving asphalt as an intermediate, no additional assumptions being made concerning the rate of asphalt hydrogenolysis. Some earlier Bureau of Mines n-ork on the kinetics of coal hydrogenolysis (8) differed from that reported here in a t least three major respects: In the earlier experiments, Tetralin was employed as a vehicle; the initial hydrogen pressure was 1000 pounds per square inch gage; and extended runs were carried out in stages, between which the autoclaves were cooled to room temperature and recharged with hydrogen, EXPERIMENTAL
I n order t o minimize complications arising from the heterogeneity of coal substance, hand-picked anthraxylon was used in this study. The material was picked from the experimental mine of the Bureau of Mines a t Bruceton, Pa., and it consisted of Pittsburgh-sear,. bituminous coal of the following composition, as used: Ll.Syo carbon, 5.35y0 hydrogen, 1.6% nitro-
RESULTS AND DISCUSSION
The product distributions and hydrogen consumptions observed in these experiments are summarized in Table I. T h e values for “oil” pertain to liquid product which is soluble i n 7 ~ hexane. “Water” includes only water formed from the organic material; it is corrected for the moisture originally present in t h o coal. Within experimental error, the conversion of coal to liquid and gaseous products a t 400’ C. is first-order in coal remaining. (Asphalt is included with the liquid products.) This is shown in Figure 1, which is a semilogarithmic plot of organic benzene.. insolubles, corrected for refractory material (see below), versus
INDUSTRIAL AND ENGINEERING CHEMISTRY
1516
TABLE I. KINETICS OF HYDROGENOLYSIS OF HRUCETON
I
I
I
Vol. 43, No. 1
ANTnRaXmoN (48.6 grams m.a.f. coal, 2500 pounds per square inch Hz, 1% SnS 0.5% NHdCl) Product Distribution, % of M.A.F. Coal Time a t Organic Reaction benaeneHz conTy?., Temp., insoluHydrosumed, Min. bles Asphalt Oil carbons Watero % ' 400 0 57.2 35.8 4.1 2.9 1.2 1.42 400 30 28.4 59.1 6.2 2.5 2.9 1.95 400 60 10.9 76.2 6.6 3.3 3.9 2.73 400 120 4.1 80.5 7.6 4.1 5.3 3.23 400 180 2.1 70.4 20.4 3.9 5 3 3.70
+
\
$ 8 m
3
0 -I
y
-A
\
L
I
0
I
I
50 100 150 NOMINAL TIME AT REACTION TEMPERATURE.MINUTES
1
420 430 420 420 420
0 30 60 120 180
31.1 7.2
430 430 430 430 430
0 30 60 120 180
1.4 2.1
62.6 75.3 70.0 49.0 23.7
2.3 9.9 19.1 39.7 61.3
2.3 4.6 4.9 6.6 9.5
3.1 5 1 6.0 7.0 7.2
2.18 3.42 3.71 4.39 4.65
9.5 2.7 1.4 2.1 1.6
67.6 69.2 43.4 27.4 16.1
13.0 17.9 45.9 57.8 69.0
7.6 8.2 8.6 10 3 11.3
6.2 6.6 6.8 6.8 7.4
3.73 4.20 4.84 5.19 5.23
440 0 9.1 66.5 13.8 5.8 8.2 44n 30 2.1 53.3 30.5 8.0 6.8 440 60 1.6 68.1 18.5 11.3 6.8 440 120 1.9 11.3 70.6 12.1 7.2 440 180 1.2 6.2 76.4 7.2 11.1 a Obtained by oxygen balance, corrected for initiai moisture.
3.75 4.27 5.54 5.17 5.38
3.5
Figure 1. First-Order Plot for Over-All Conversion of Anttiraxylon at 4.00' C.
nominal time a t reaction temperatwe. The line in 3'igure 1 was obtained by the method of least squares from the indicated experimental points. At the higher temperatures, the conversion of coal i8 so rapid that only one, or a t most two, points can be accurately determined for each series, and it is not possible to distinguish between rate laws. [The curvature in corresponding semilogarithmic plots reported by Storch et al. ( 2 ) is believed to be caused primarily by lack of correction for refractory material which will not be hydrogenated even a t very long reaction times, and lack of proper corrections for hea,ting up and cooling periods. ] The remlts shown in Table I may be interpreted on the basis of the following reaction scheme:
p"
"r""
k; ;
where C, is the amount of moisture- and ash-free coal present a t zero time. The following considerations are involved in applying Equations l and 2 to the data on anthraxylon: 1. The hydrogenation data indicate the presence of about 1.6% (based on moisture- and ash-free coal) of refractory organic material which is not liquefied, even at long reaction times. The quantity C, has, therefore, been taken as 98.4%. Similarly, a correction of 1.6% has been subtracted from the observed values of or anic beneene-insolubles in deducing the order of the ~ (Figure 1 ) and in making comparisons primary C O conversion between observed and calculated data (Figures 2 to 5 ) . 2. The fraction of the reacted coal which i s converted to asphalt, a, depends on the quantities of hydrocarbon gas and water which are formed during the primary liquefaction. These quantities were estimated both by considering the observed
k;
Coal (organic benzene-insolubles) I
--+ Asphalt --+ Oil I
4 k;"
t-
H20
z
LT U
(This scheme does not imply that asphalt is the first intermediate product; in these experiments, however, it is the earliest intermediate which is isolated and measured.) The k's are specific rate constants for the reactions indicated.
W
e-
< a
m 0 a
-a
Let
w
U c
kl = k ; $. k;' $. k;" kz k; f k:' 4-kh" C = coal (organic benzene-insolubles) remaining a t time t A = asphalt present a t time t a = k;/kl
Earlier work (3) has shown that the reactions involved in asphalt hydrogenolysis are first-order in asphalt remaining. Figure 1 suggests that, in addition, all of the reactions involving the primary conversion of coal are also fht-order. If this is true, it can be shown that C = Coe-kit
a
VI w Ln
m
2z U
z
(L 0
(1) Figure 2.
and
Hydrogenation of Bruceton hnthraxylon a t 400"
c.
July 1951
INDUSTRIAL A N D E N G I N E E R I N G C H E M I S T R Y
1577
z W
0
150 NOMINAL TIME AT REACTION TEMPERATURE MINUTES MINUTES
Figure 3,
Hydrogenation of Bruceton Anthraxylon at
420' C.
amounts of gas and \vatci formcd and by noting the maximum amount of asphalt plus oil formed in each series of runs. The best estimates of the amount of primary gas and water formed are considered t o be 5.8, 8.0, 10.4, and 11.401, at 400", 420°, 43O0, and 440' C., respectively. The corresponding values of LI are 0.939, 0.916, 0.891, and 0.881 * a n estimated error of 0.01. 3. The values of k2,the over-all rate constant for the asphalt hydrogenolysis, were taken directly from the data previously obtained on pure asphalt (a); no adjustment of these figures was made. 4. The value of kl at 400" C. was obtained from Figure 1. A t higher temperatures, because the reaction was so rapid, ICI was fixed by imposing the condition that the calculated value for organic benzene-insolubles should coincide with the observed value for zero nominal reaction time. Because only one ex eri mental point was used in determining values of kl a t each o f t h i higher temperatures, these values are not considered t o be very reliable. I n particular, no accurate value for the activation energy of the primary reaction can be deduced from the temperature variation of kl. At the higher temperatures, errors in estimating kl will have relatively little influence on the calculation of asphalt, as t h e liquefaction is almost complete within 0.5 hour. 5. Because some reaction occurs during the heating up and cooling periods, the "true" value of the reaction time, t , t o be used in Equations 1 and 2 is greater than the nominal time a t reaction temperature (a). I n order to avoid an excessive number
k!
m
2m
70
NOMINAL TIME AT REACTION TEMPERATURE, MINUTES
Figure 4.
Hydrogenation of Bruceton Anthraxylon a t
430" C.
of adjustable parameters, a value of 20 minutes was chosen as the correction to be added to the nominal time at temperature in order t o give the true reaction time at temperature. The value of 20 minutes is required t o fit the data shown in Figure 1 ; i t was assumed that the same value should be applied for all of the other reactions. With this correction, Equations 1 and 2 become modified t o read:
(3)
where
T
is the nominal time a t temperature.
Values of the various constants are summarized in Table 11. With the use of these constants in Equations 3 and 4,value8 of C
$
6:-
-
8
50
OI
X
X 400-C 0 420'C A 430'C 0 440-C
G 40
f
s
__
30
23
,
10
Y 0
Figure 5.
Hydrogenation of Bruceton Anthraxylon at 440" C.
I I
I /
I I I 2 3 4 U p CONSUMED, PERCENT
1 5
Figure 6. Hydrogen Consumption us. Liquefaction for Anthraxylon Hydrogenation
I 6
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
1578
Vol. 43, No. 7
hydrocarbon gas production, t,hese results indicate again the desirability of carrying out, coal hydrogenation in at least tv-o stages. The fist stage should carry the coal only t,hrough t,he primary liquefaction step under conditions of minimal gas formation, nhilc the primary produc hould he processed separately under (presumably different) optimum conditions. The production of carbon dioxide was very small in these experiments (0.0 to 0.4YG),and t'hc oxygen climinated-i.e., the oxygen not present, in residual coal or asphalt-was accounted fot almost exclusively by the JT-ater produced. The close correlation bet,ween oxygen eliminat,ed a s water and coal liquefied is shown in Figure 8. The same relationship has been pointed o u t in earlier m-ork ( 2 ) .
TABLE111. ULTIMATE ANALYSESOF ASPHALTFROX ANTHRAXYLOR HYDROGES.ATION Time a t Reaction Temp.,
Temp.,
Ultimate Analysis of .Isylialt, "eight C N S
C.
Min.
I
sts tried, aluminum and titanium oxides and phosphoric arid were the most active. While alumina w a s the most efficient for nitrogen alkylation, all these catalysts eEected more nuclear alkylation than had
been anticipated. This led to a study of the rearrangement of alkyl groups from nitrogen to carbon, which was found to be accompanied by extensive disproportionation. Although conversions of primary to tertiary amines exceeding 90% were obtained, more selective catalysts than any here used must be found to permit the vapor phase preparation of N,N-dialkylanilines in a reasonable degree of purity. Means were found for the syntheses from aniline of a wide variety of aromatic amines, including toluidines, xylidines, mesidine, and diphenylamine.
T
washed and ignited. The coconut' charcoal mas a n activated form. Particles were roughly cubical and about 2 mm. on a side. Pumice was screened and part'icles of a size less than 12 mesh were reject8ed. Particles used averaged a,bout 0.55 cc. each. AIXJMINVM OXIDE. Aluminum oxide A was a preparation rcsulting from the action of amalgamated aluminum on water. Coiled strips of condenser aluminum, 99.95 yo pure, were dipped successively in 10% hydrochloric acid and 10% potassium hydroxide until a uniform react'ion over the surface resulted. With a minimum exposure to the air, the metal was washed with distilled water and immersed for 1 minute in a 0.5% solution of mercuric chloride. The coils were then rinsed thoroughly in distilled water, after which they were suspended in distilled water from the lip of a large beaker. Aluminum hydroxide formed on the metal and dropped'to tjhe bottom of the beaker. After filtering, the cake was put int,o pans, cut into l/d-inch cubes, and dried a t room temperature in the air. Aluminum oxide B was prepared in t h e same manner except t h a t it was dried a t an elevated temperature. The temperature was raised from 50" t o 110' C. over a period of 6 hours. Under these conditions, there was little shrinkage in comparison to the considerable shrinkage exhibited in the formation of oxide A4. Oxide A had a particle density of 0.9 and a bulk density of 0.4. Oxide B had a particle density of 0.4 and a bulk density of 0.25. Aluminum oxide C was prepared by precipitating aluminum Iiydroxide on pumice. The pumice was first impregnated ITith aluminum nitrate and then dropped into concentrated ammonium hydroxide containing ammonium nitrate. After standing 24 hours, the pumice was filtered off and thoroughly xvashed. This catalyst was dried a t 105O C. I t contained 5.3 grams of aluminum oxide in a bulk volume of 150 cc. PROMOTED ALUMIKUNOXIDE CATALYSTS.Aluminum oxide promoted with blue oxide of tungsten was prepared by impreggrams of asbestos with a suspension of 15 grams of 1 j.l'roxi+p 13 and 15.5 grams of tungstic acid in 100 ml.
HE production 01 alkylated aromatic amines, such as dimethylaniline, bv the high pressure liquid phase reaction of
alcohols and amines has been well established for many years The corresponding catalytic vapor phase reaction, xhich offers the advantages of continuous operation a t low pressures, has been frequently mentioned but not exhaustively investigated. A wide variety of catalysts has been advocated for this reaction by 8 number of workers (1-4, 9-13), Unfortunately, because of differences in equipment, operating conditions, and analytical techniques, i t is not possible to compare adequately these various catalysts from published results. The purpose of the present investigation was to determine the relative efficiencies of a large number of possible catalysts for the vapor phase formation of dimethylaniline from aniline and methanol. Using the most active catalyst found, a study Mas then made of certain process variables, including an extension of the results to higher alcohols. PREPARATION OF C4TALYSTS
I n general, the catalysts were prepared from reagent grade chemicals. Precipitated catalysts were washed thoroughly t o free them from salts. After careful drying, the usual procedure was to activate the c,atalyst for 1 hour in the reaction tube in a stream of nitrogen a t 360" C. SUPPORTING MATERIALS.In certain instances, asbestos, coconut charcoal, pumice, or silica gel was used as a supporting material. T h e asbestos \r.as in coarse threads, which had been acid 1 Present address, -4inerican Cyanamid Co., Calco Chemical Dirision, Bound Brook, N. ,J. 2 Present address, E. I. du Pont de Nemours & Co.. l n c . , TVilmineton, Del.
