INDUSTRIAL AND ENGINEERING CHEMISTRY
334
.kt a metal concentration of O . l % , however, it is again found possible to replace 90% of the tin by zinc without losing catalytic activity (runs 906 and 893); the presence of ammonium chloride is, of course, necessary. At concentrations of O.l%, neither molybdenum trioxide nor ferrous sulfate shows any cataIytic effect. Of the iron catalysts tested a t 1% concentration, ferrous sulfate, pyrite, and dried red mud (Bayermasse) are all about equally effective. Pyrite and red mud should be available in quantity a t low cost and should be considered for use in largescale plants. Other iron ores may prove to be equally suitable. Any comparison of the catalyst concentrations used in these autoclave experiments with those used in a continuous plant should take into consideration the fact that in the c~onventional
Vol. 42, No. 2
LITERATURE CITED
(l) Abe,
R.,Huzikawa, S., Kakutani, T., arid Okaniura, T., J
SOC.Chem. I n d . J a p a n ,
41,
supplenientary binding, 417-18
(1938).
( 2 ) Berthelot, C., A n n . chim. et phys., 20, 526 (1870).
(3) Booth, PI‘. J., J . Sac. Chem. I n d . London, 63,i ( 1 9 4 4 ) . (4) Dept. Sci. Ind. Research (Brit.), Fuel Research, “ILept. for Period Ended March 31, 1932,” p. 54; “Rept. for Year Ihidrd March 31. 1933.” D. 100. ( 5 ) Eccles, A , and McCulloch, A., J . SOC.Chem. Ind., 49, 377T, 383T (1930). (o) Fisher, C. H., Sprunk, G. C., Eisner, A , O’Donnell, EI. ,J,, Clarke, L., and Storch, H. IT., U . 5’. Bur. iMines. T e c h . Paper 642 (1942). (7) Hlavica. B.. Brennstoff-Chem.. 9. 229 (1098), ( 8 ) Kurokawa, M., Hirotla, W., Fuiiwnra: K., and Aqaoka, N , .I. Fuel Sac. J a p e n , 18, 31 (1939). Storch, €1. H., ISD. ENG.CHR:M., 37, 340 (1945). .Varieti, T. E., Bowles, K. W., and Gillnore, R. E,. 1x1).I;vc,. I
_
( Coal Hydrogenation Catalysts)
MECHANISM OF COAL HYDROGENATION SOL WELLER, E. L. CLARK, ANI) AI. G. PELIPETZ Bureau of M i r z e s , P i t t s b u r g h , P a .
Consideration of the influence of catalysts on the hydrogenation of bituminous coal and asphalt has led to the formulation of a descriptive theory of coal hydrogenation. I t is postulated that coal (or asphalt) is thermally split to form reactive fragments, the splitting being catalyzed by halogen acids. The fragments either polymerize to form benzene insoluble products or are stabilized by the addition of hydrogen to form soluble products. The hydrogenation stabilization is catalyzed by tin.
phoric acids were added as G .Y :iqueous ~olutions. h rernovablo Pyrex KO.774 glass liner mas employed in all cases. Three inaterials were studied in t,hcse investigations: whole Bruceton coal (from the Bureau’s oxporimeiital mine a t Bruceton, Pa.), hand-picked Hruoeton anthraxylon (from thc si~nic mine), and crude asphalt isolated froiii the products of a coal hydrogenation pilot plant run 011 Bruceton coal. (Asphalt is dcfined here as material soluble in beiiecne but insoluble in n-hesane.) The ultimate analyses o f these substances nois is ti ire-frcc basis) are prescrited in Table I.
T
HE suggestion has been made ( 3 )that the over-all hydrogen-
ation of coal to distillable oil proceeds via the formation of asphalt as a n intermediate. The coal-to-asphalt conversion appears to be a relatively rapid reaction which is accompanied by the elimination of the bulk of the oxygen in coal, primarily as water. The asphalt-to-oil step seems t o be relatively slow, requiring higher temperatures and longer rraction times than does the primary liquefaction. It is also known, a t least for bituminous and subbituminous coal, that while tin used alone is a fair hydrogenation catalyst and halogen acid alone is a negative catalyst (with the exception of hydriodic acid), the combination tin-halogen acid constitutes perhaps the most effective known coal-hydiogenation catalyst (4). Consideration of these and other results described below has 1 ~ d us t o the formulation of a descriptive theory of coal hydrogenation. hlany of the separate dements of the theory are not unique, nor is it supposed that all the details of coal hydrogenation will be covered by the crude picture presented. It is hoped, however, that the hypothescs will permit a correlation of the major steps in the over-all hydrogenation and of the role played by the separate catalyst constituents, and that they will provide a framework for the design of future experiments. PROCEDURE AND MATERIALS
A detailed description of the hydrogenation equipment and the analytical procedures used has been published (1, 4). In all of the experiments described herein, a n initial hydrogen pressure of 1000 pounds per square inch and a reaction time of 1 hour at the temperature designated were employed. No oil vehicle was added t o the powdered coal or asphalt; hydrochloric and phoa-
Laboratory analysis of the crude asphalt used showed it bo rontain 0.4% water, 1.4% benzene insolubles, 93.8% asphalt, and 4.4% oil (material soluble in n-hexane). CATALYSTS FOR ASPHALT IIYDROGENATIQN
A study was made of the irifluencc of n number of catalysts o n the hydrogenation of crude asphalt. In all of these experiments, ,50 grams of asphalt were hydrogenated for 1. hour a t 450” C. ut an initial hydrogen or helium prcssure of 1000 pounds per square inch gage. Result,s of the tests arc summarized in Tatilc I1 and in Figure 1. It is clear that of the metal catitlysts and combinations teated, tin plus ammonium chloride is as superior for the hydrogenat,ion of asphalt as it is for the hydrogenation of coal (4). It follows that the excellence of tin plus ammonium chloride for coal hydrogenation is not associated with any peculiar structure of coal as a solid, as, for example, the possible existence of graphitic layers between which the catalyst might be intercalated.
INDUSTRIAL AND ENGINEERING CHEMISTRY
February 1950
ON ASPHALT HYDROGENATION TABLE 11. EFFECTOF CATALYSTS
. c
t
(50 g, crude asphalt: 1 hr. a t 450° C.; lOOO-lb./sq. inch initial Hz or H e pressure) Gaseous * Asphalt R?oil droHY- organic Ultimate Analysis of Asphalt Remaining, 7* %amPlus carInsol0 ing, Water, bow, ubles, (differCatalyst G. G. G. G. C H N 6 ence) 40.2 6.2 3.4 1.2 ... 36.5 7.6 3.1 2.8 9o:io i . 2 5 i . 8 3 O.'i5 i.'i7 21.6 5.6 2.1 18.5 89.92 5.77 1.85 0.08 2 . 3 8 0.5% HCl 34.0 6.9 2.8 4.2 90.66 6.31 1.50 0.13 1.40 1 7 HCl 32.6 7.7 4.1 5.1 5% HCl 27.2 7.9 4.2 7.3 si:& 5 . 9 8 i . i o o.'ib i . 2 5 0.5% HCI 17.2 4.3 3.6 23.9 90.19 6.01 1.56 0.24 2.00 38.3 8.2 2.4 1.4 90.13 6.77 1 . 5 1 0.15 1.44 0.5% NHaCl 25.7 21.0 2.9 0.9 91.24 6.43 1.20 0.17 0 . 9 5 Xi-K.G. (1% Ni)q 35.7 10.1 3.2 0.7 90.346.711.680.16~.ll Ni-K.G. (1% NI) 0 5 7 HCl 34.9 8.4 2.9 3.0 80.63 6 . 4 3 1 . 4 3 0 . 1 5 1.36 l % " N ? + 0.5% HCl 33.9 6.9 2.7 4.7 90.356.411.380.171.69 1% Moos 36.9 7.2 3.0 1.9 90.15 6.50 1.82 0 . 1 8 1.35 89.83 6 . 4 4 1.62 0.23, 1.79 36.2 8.9 3.0 1.8 1% Moos 2% Cu-ohromiteh 33.6 9.2 3.8 89.78 6.15 1.77 0 . 3 2 1 . 9 8 2.2 8.4 2.7 11.5 90.53 5.86 1 . 5 8 0 . 1 6 1.87 2% HaPo4 28.2 1% 12 16.0 23.9 4.4 5.3 91.87 5.94 0.92 0.13 1.14
335 h y d r o g e n a t i o n catalyst. It should, and does, result in a relatively high conversion of asphalt and a relatively small production of benzene insolubles (run 1153).
A large quahtity of benzeneinsoluble matter was produced in the two experiments in which ... a n inert (helium) atmosphere was used (runs 1185 and 1148). I n the absence of high-pressure hydrogen, no hydrogenation stabilization of reactive fragments + can occur. Some splitting to + form fragments, however. does occur in the absence of hydrogen, even if no splitting catalyst is present. The fragments (free radicals?) formed under these conditions can act as polymerization initiators and acceleraNi on kieselguhr; 46% Ni. tors, inducing polymerization h Adkins' catalyst, obtained by the courtesy of R. A. Glenn a t Carnegie Institute of Technology. in the residual asphalt with the resultant formation of large amounts of insolubles. The relative proportion of insolubles, asphalt, and oil in the The data of Table I1 and Figure 1 show that, with the excepproduct depends on the relative rates of splitting, hydrogenation of tin plus ammonium chloride, all the catalysts which intion stabilization of fragments, and polymerization by fragments. crease the hydrogenation of asphalt t o oil also increase the proThe product distribution depends, therefore, on the presence of duction of organic benzene-insoluble material. The production splitting catalysts, hydrogenation catalysts, and high-pressure of hydrocarbon gas seems to be random, however, and independhydrogen. The strajght line drawn in Figure 1 may be interent of the nature of the catalyst present. Large oil productions preted t o represent the locus of materials which, roughly speakwere observed only with tin plus ammonium chloride and iodine. ing, have either no catalytic effect or which catalyze the splitting The following scheme is suggested as a working hypothesis to explain these results: of asphalt but not the hydrogenation of fragments. Points lying to the right of the line are for those materials or conditions which Asphalt favor the polymerization by fragments over their hydrogenation ZHCl / (phosphoric acid, inert atmosphere). Points lying t o the left of Reactive fragments Sn Gas the line represent materials which catalyze both the splitting and --n rc( Benzene insolubles Oil the hydrogenation stabilization of fragments. Run No. Gas 844 Hz 1033 Hz 1185 He 1225 Hz 847 Hz 1073 H2 1148 He 1136 Hz 877 Hz 897 Hz 1166 112 1220 IIZ 876 HZ 917 Hz 915 IIz 1156 Hz 1153 HZ
I
.
.
!%gi
It is possible t o devise other schemes which correlate the data and which cannot be ruled out at present. One of these is the possibility that asphalt is susceptible t o a n acid-catalyzed polymerization and t o a stannous chloride-catalyzed depolymerization-hydrogenation. This has several undesirable features, however, among them the fact t h a t both thermodynamic calculations and experimental results indicate the conversion of stannous chloride t o stannous sulfide under reaction conditions. While the suggested scheme is not unique, therefore, it seems t o be at least as simple and self-consistent as any other.
HYDROGENATION OF WHOLE BRUCETON COAL AND BRUCETON ANTHRAXYLON
The similarity between coal and asphalt in their response t o hydrogenation catalysts suggests t h a t the scheme formulated above for asphalt hydrogenation may be applicable t o coal as well. The authors assume, as in the case of asphalt,, that the first reaction which coal undergoes is a thermal splitting into reactive fragments, the splitting being catalyzed by halogen acids.
Asphalt is assumed to give either gas, by a noncatalytic reaction, or reactive fragments of some kind (possibly free radicals). The cleavage to give reactive fragments is catalyzed by halogen acid-producing substances (hydrochloric acid, ammonium chloride, etc.). These reactive fragments may recombine to produce asphalt again; they may polymerize still further to produce benzene-insoluble material; or they may be stabilized with addition of hydrogen t o produce a n oil. This hydrogenation stabilization is assumed to be catalyzed especially well by tin. This descriptive theory leads to a number of predictions, some of which are: Tin used in the absence of hydrochloric acid should behave as no catalyst at all, since the splitting t o form fragments, which is assumed t o precede the hydrogenation s t a b i l i z a t i ~ ~proceeds ,, 1s 1s found only slowly in the absence of a splitting catalyst. t o be the case (run 1135). Phosphoric acid, which is used as a polymerization catalyst @), might be expected t o produce especially large quantities of benzene insolcbles. It is observed.to do so (run 1155). Hydriodic acid, t o which iodine is rapidly converted under reaction conditions, is almost unique in being both a splitting and a
Figure 1. Effect of Catalysts on Asphalt Hydrogenation
336
INDUSTRIAL AND ENGINEERING CHEMISTRY
Vol. 42, No. 2
run (1035), for which the ratio of polymerized fragments to TABLE111. EFFECTO F C.4T.4LYSTS O X HYDROGENATION OF unreacted coal should be higher, have a lower oxygen content WHOLEBRUCETON COALAND BRCCETON ANTHRAXYLON than those from the catalyzed run (1069). '(50 g coal; 1 hr. a t 450" C ; lOOO-lb./sq. inch initial Hz pressure) Anthraxylon, which is considered to be the petrographic conRun stituent of coal most amenable t o hydrogenation, surprisingly so. Coale Cat a1ys t 7' Liquefaction shows a very low liquefaction in the absence of any catalyst (run 311 B. , . . 55.0 616 B. 1% Sn 66.8 482, Table 111), and a n even lower liquefaction when ammonium 518 B. 0.5% SH4C1 40.8 chloride alone is added (run 783). This is attributed to a greater 481 B. 1% Sn + 0.5% XH4C1 86.1 482 B.A. ... 29.5 rate of primary splitting, relative to the rate of fragment stabili483 B.A. 1% Sn 85.3 783 B.A. 0.0% KH4C1 22.6 zation, in the case of anthraxylon than in the case of other petro484 B.A. 1% Sn + 0.5% KHaC1 92,9 graphic constituents. The splitting rate is enhanced by the a B. = whole Bruceton coal; B.A. = Bruceton anthrasylon. addition of ammonium chloride, in the manner indicated above. When the hydrogenation catalyst, tin, is used, most of the fragments formed have a chance t o be stabilized before they polymerize. As a result, tin used by itself causes a striking increase in A competition may then occur between the stabilization of fragliquefaction with anthraxylon (run 483). The combination cataments by hydrogenation (catalyzed by tin) and the polymerizalyst, tin plus ammonium chloride, results in still further increase tion of fragments t o form benzene insolubles. I n this case, the In liquefaction, as in the case of whole coal. fragments are assumed to be of a molecular size corresponding t o that of asphalt, so t h a t the stabilization of fragments results in asphalt production. I n Table I11 are summarized the results of several catalyst tests on whole Bruceton coal and Bruceton anthraxylon ( 4 ) . These results are consistent with the reaction scheme postulated, as may be seen from the following considerations. I n the absence of added catalyst, whole Bruceton coal shows a liquefaction (under the given conditions) of 55%.
- benzene insolubles yoliquefaction = original coaloriginal coal
x
100
TABLE IV. HYDROGENATION OF WHOLEBRUCETON COAL (100 g. coal; 1 hr. a t 450' C ; 1000-lb / s q . inch initial Hn pressure)
Ultimate Analyis of Benzene Insolubles (Yo of LI 4 l?. Material)a
Run Untreated coal 1035
Liquefaction
Catalyst
... 0 Sone 49.0 1% Sn 0.5% SHiCl 76.9 moisture- and ash-free.
+
1069
The per cent liquefaction is computed on a moisture- and ash-free basis. Addition of ammonium chloride alone increases the rate of splitting but does not accelerate the rate of fragment hydrogenation. As a result, in the presence of ammonium chloride an increased polymerization to insolubles occurs before the hydrogenation stabilization can take place. This accounts for the negative catalysis by ammonium chloride or hydrochloric acid (run 518). Sddition of tin, on the other hand, catalyzes the hydrogenation of fragments but not the primary splitting. I n the absence of catalyst some of the fragments formed polymerize before they can be stabilized; use of tin, therefore, preferentially promotes the stabilization over the polymerization and results in some increase in liquefaction (run 516). Use of both ammonium chloride and tin increases the rates both of fragment formation and of fragment stabilization, with a resultant liquefaction greater than that achieved with either component separately. One interesting implication of the suggested mechanism is the following: Under the reaction conditions used, a significant portion of the benzene-insoluble material should be made up of reactive fragments which polymerized before they could be stabilized. These benzene insolubles should, therefore, be qualitatively different from unreacted coal, and this is indeed found to be the case, as consideration of their ultimate analyses shows. In Table IV are summarized data for untreated coal and for whole Bruceton coal hydrogenated at 450 ' C. in the presence and in the absence of catalyst. (The liquefactions are lower than the corresponding values in Table I11 because of the lower average hydrogen pressure in these experiments; with the large coal charge, 100 grams, a significant portion of the hydrogen originally present is consumed by the reaction.) It is reasonable to assume that the primary splitting of coal to form reactive fragments is accompanied by a n appreciable elimination of oxygen. Polymerization of the fragments should produce benzene insolubles which are much poorer in oxygen than the original coal. It may be seen from Table IV t h a t the benzene insolubles from runs 1035 and 1069 do contain much less oxygen than the unreacted coal. Although the result is within experimental error, it also appears that in this case the benzene insolubles from the uncatalyzed
n "
%
a M.A.F. =
(difference)
H
C
N
8
5.6 4.3
83.5 89.4
1.6 3.0
1.4 1.4
7.0 2.8
4.3
87.0
2.1
3.4
3.2
Thus it appears possible to interpret the gross hydrogenation behavior of bituminous coal and of asphalt on the basis of the descriptive theory presented above. It is hoped that formulation of the theory, crude as i t is in its present form, will suggest further experiments which can lead to a more detailed understanding of the coal hydrogenation process. SUMMARY
Consideration of the influence of catalysts on the hydrogenation of bituminous coal and asphalt has led to the formulation of a descriptive theory of coal hydrogenation. It is postulated that coal (or asphalt) is thermally split to form reactive fragments, the splitting being catalyzed by halogen acids. The fragments either polymerize to form benzene insoluble products or are stabilized by the addition of hydrogen to form soluble products. T h e hydrogenation stabilization is catalyzed by tin. ACKNOWLEDGMENT
The authors would like to express their gratitude to S.Friedman and J. Lederer for assistance in obtaining the data presented in this paper, and to M. Orchin and H. H. Storch for many helpful discussions. LITERATURE CITED
(1)
(2) (3) (4)
Fischer, C. H., Sprunk, G. C., Eisner, A., O'Donnell, H. J., Clarke, L., a n d Storch, H H., U . S . Bur. Mines, Tach. Paper 642 (1942). Ipatieff, V. N., aiid Pines, H., IND. ENG.CHEM.,27,1384 (3936). Pelipets, M., Kuhn, E. bf., F r i e d m a n , S., a n d Storch, €1. H., Ibid., 40, 1259 (1948). Weller, S.,Pelipets, M. G., Friedman, S , a n d Storch, H. H.,
Ibid., 42, 330 (1950).
RECEIVEDJuly 27, 1949. Presented before the Division of Gas and Fuel Chemistry a t the 116th hfeeting of the -4MERICAN CHEMICAL SOCIETY, Atlantic City, N . J.