Return to Presentations Table of Contents
Recent Research on the Fischer-Tropsch Synthesis
Investigations on the Fischer-Tropsch process on a laboratory and semi-technical scale have been in progress at the Fuel Research Station continuously since 1935. These investigations have been concerned with the factors which affect the activity and life of cobalt catalysts for the synthesis at atmospheric pressure, the synthesis at moderate pressures with cobalt and iron catalysts, the reaction mechanism, and the conversion of the primary products into secondary products, particularly high-grade lubricating oils and fatty acids. Owing to the interruption of the free dissemination of scientific information caused by the war, the results of many of the investigations have not been published, but it is hoped that this will be remedied in due course.
In the present paper, it was considered that is would be of greater interest to report the results of some of the more recent research work. As investigation of the problems selected for discussion is still in progress this paper should be regarded as an “interim report.”
Use of High Gas Throughputs
In the Fischer-Tropsch process as employed in the industrial plants in Germany, France and Japan, a precipitated cobalt catalyst, supported on kieselguhr and containing thoria and magnesia as activators, was employed. This catalyst, in the form of small granules, was arranged in narrow spaces between water-cooled surfaces and was maintained at a temperature in the range 175°-200°C. In order to achieve a 93-94 per cent conversion of the carbon monoxide in the synthesis gas, two or three stages were required and the overall gas space velocity was restricted to about 60 vol. per vol. catalyst space per hour. The space-time yield of the plants was therefore very small and did not, in fact, exceed 0.25 tons primary products per cu. m. of reaction space per day. If attempts were made to increase the space -time yield by increase in gas throughout, the conversion of the synthesis gas decreased, and if the temperature was increased to maintain conversion, the heat liberated by the reaction exceeded the heat-transfer limitations of the water-cooled, fixed-bed system, the catalyst temperature rose rapidly and methane, carbon dioxide and carbon were produced.
The products obtained comprised 40 to 60 per cent of low-grade petrol, 30 percent of high-grade Diesel oil and10 to 30 per cent of waxes.
As a means of producing synthetic petrol, therefore, the German process suffers from two serious disadvantages. The space-time yield is low and the products contain a relatively small proportion of petrol of inferior quality.
By employing an iron catalyst at temperatures in the range 300° to 330°C. and pressures of 250 to 350 lb. per sq. in., and carrying out the synthesis in a “fluid-catalyst” system, American technicians 1,2 claim that both these disadvantages can be overcome. Petrol of 80 octane number can be obtained as the major final product (80 percent of the total) and space-time yields “many times greater” than those obtained in Germany can be achieved.
It was clearly of importance not only to investigate these claims but to explore the implied possibilities of extending very materially the range of reaction conditions which can be employed in the synthesis process.
Iron Catalysts in Fixed Beds
Although in commercial fixed-bed reaction vessels the space velocity cannot be increased much beyond 100 vol. per hour without overheating the catalyst it has been found that in small-scale laboratory reactors this limitation does not apply.
The reaction vessel employed in the experiments was a thick-walled steel tube, 10 mm internal diameter, with a catalyst capacity of 100 ml., embedded in an electrically heated aluminium block, 6 cm. in diameter. The general arrangement of the apparatus is shown in Fig. 1
The catalyst employed was a commercial, fused-iron, synthetic-ammonia catalyst crushed and screened to 7/14 B.S. Test Sieves. Before use for synthesis, it was reduced at 450°C. for 24 hours in pure hydrogen at a space velocity of 2,000 per hour. Synthesis gas (H2:CO=2:1) containing 5 per cent. inert constituents and not more than o.1g. total sulphyr per 100 cu.m., was employed as raw material.
A summary of the conditions used and the results obtained is given in Table 1. The effects of pressure, recirculation of residual gas, and synthesis gas space velocity were studied in experiment 12/26 while adjusting the temperature to maintain a constant carbon monoxide conversion of about 95 per cent.
It will be observed that increase in pressure from 10 to 20 and from 20 to 25 atmos. had a marked beneficial effect, as indicated by the reduction in temperature required to maintain conversion at a fixed space velocity and by the increase in space velocity permissible at fixed temperature without fall in conversion
The CO conversion be maintained at about the 95 per cent level at space velocities up to 1,000 vol. per vol. catalyst per hour. The average velocity over duration of the experiment (128 days of synthesis) was approximately 500 per hour, and the average CO conversion, 95 per cent.
The use of residual gas reciculation had a striking effect. By repressing the formation of carbon dioxide by water-gas-shift reaction and increasing the H2:CO utilization ratio it resulted in an increase in the proportion of carbon monoxide converted to hydrocarbons higher than methane from 60 to 80 percent.
The catalyst deteriorated in activity over the course of the experiment as indicated by the increase in temperature necessary to maintain conversion at a given space velocity.
It will be observed that at the high temperatures employed, rather more than half the higher hydrocarbons produced were in the range C2 to C4. This fraction of the products has not been analysed completely, but the average carbon number was 3.3 and the olefin content 75 per cent.
The boiling ranges and olefin contents of the liquid products obtained are shown in Table 2. It will be observed that the products were low-boiling and highly unsaturated and did not change markedly in composition with change in reaction conditions.
Iron Catalyst in a Fluid Bed
It was clearly of considerable interest to determine the behaviour of the catalyst used in the above experiments in a fluidized state but under otherwise similar conditions to those employed in the fixed bed.
The reaction vessel employed consisted of a 3.05 m. length of steel tube, 2.5 cm. internal diameter, fitted with a conical gas inlet at the lower end and with a 30.5 by 7.6 cm. disengaging chamber at the top to allow entrained catalyst to separate from the gas stream. A central tube 0.8 cm. external diameter served as a thermo-couple pocket and as a support for grids cut from steel gauze (4 mesh to the cm.) which were made slightly convex to the gas flow and spaced at 5 cm. intervals up the tube. The use of these grids was found to be essential in order to obtain uniform fluidisation in long, narrow tubes. They did not seriously interfere with catalyst circulation.
The reaction tube and disengaging chamber were heated electrically, no special heat-dissipating means (aluminium block or liquid jacket) being employed. The lower portion of the reaction tube is shown in the drawing, Fig. 2, and the general arrangement of apparatus in the diagram Fig. 3 and the photograph Fig. 4.
The results of an experiment in which 1,000 g. of synthetic ammonia catalyst graded between 72 and 170 B.S. Test Sieves and occupying a compacted volume of 400 ml. was employed for 305°C are recorded in Table 1. The synthesis gas used was of similar composition to that employed in the fixed-bed experiments, but to minimize wax and carbon formation the H2:CO ration was increased to 2.35:1. In order to obtain the linear velocity necessary to maintain the catalyst in a fluid condition (18 cm./sec.) without employing excessively high synthesis-gas rates recycle ratios considerably higher than those used in the fixed bed were employed.
Steady synthesis conditions were maintained continuously for 230 hours. Temperature explorations showed a maximum variation over the column of 4°C., which, in view of the absence of any of the usual cooling devices, is a striking illustration of the temperature-equalising effects of fluidisation. No cyclone separators or filters were fitted to the reaction vessel, but the total amount of catalyst carried over with the gas stream during the experiment amounted to only 5 per cent of the whole. At the conclusion of the experiment, the catalyst was still perfectly fluid and contained only 1 per cent of benzen-soluble organic matter (wax and resins).
Almost complete carbon monoxide conversion was maintained at a constant temperature of 300+5°C. despite increase in space velocity from 750 to 1,050 volumes per hour. This indicates that the full activity of the catalyst was not being utilized, and by comparison with the fixed-bed results, that the catalyst was more active in the fluidized powder form than in the fixed bed, but this may be due to the higher H2:CO ratio employed. The complete elimination of carbon dioxide formation and the H2/CO utilization obtained are attributed to the use of a high recycle ratio. The latter is also believed to be responsible for the greatly increased proportion of C2-C4 hydrocarbons produced in the fluid bed.
At periods in the experiment, negative values for carbon dioxide formation were recorded. It is believed that these many be genuine and result from the hydrogenation of carbon dioxide, a reaction which might be expected to proceed under the conditions prevailing. The composition of the liquid products obtained at similar temperatures in the fixed. (See Table 2)
In Table 3, the results of the experiments in fixed and fluid beds are shown in comparison with those obtained in the German commercial plants. The figures show that, in the present experiments, the efficiency of utilization of the carbon monoxide was not much less than in the cobalt-catalyst process and the space-time yield of higher hydrocarbons was 15 times greater. The high proportion of light hydrocarbons formed with the iron catalyst is not a disadvantage, as the olefins can be polymerized to form high octane petrol, and if this is done, the total yield of petrol will be of the order of 75 per cent of the final products.
Although the results reported at not regarded as the optimum obtainable, they afford confirmation of the claims made for the American process, apart from the question of motor-spirit quality which has not yet been investigated.
The results obtained in the fixed bed, though not markedly inferior to those obtained in the fluid bed, could not be reproduced on a commercial scale owing to heat transfer limitations which do not apply to the fluid catalyst system.
For comparison with the performance of the synthetic ammonia catalyst, some experiments were carried out with the normal cobalt-thoria-magnesia-kieselguhr catalyst at high space velocities in a fixed-bed apparatus similar to that shown in Fig. 1. The normal synthesis gas, H2:CO=2:1, was employed. Some of the results of the experiments are given in Table 1.
It was found possible to operate at a space velocity of 500 per hour with an 80 per cent carbon monoxide conversion by using a temperature of 220°C., but to maintain the conversion, a rapid rise in temperature to 240°C., which brought about a big increase in methane formation, was found to be necessary and no attempt was made to operate at higher velocities. Attempts to push the carbon monoxide conversion to the 95 percent level, maintained in the experiments with the iron catalyst, were unsuccessful and merely resulted in an excessive production of methane and the formation of some carbon dioxide (see last column in Table 1).
In contrast to the experience with the iron catalyst, the use of residual gas recirculation had little beneficial effect. The production of methane was reduced somewhat, but the production of C2 to C4 hydrocarbons was increased at the expense of the liquid products. Under non-recirculation conditions, the cobalt catalyst gave higher yields of liquid hydrocarbons than iron at the same space velocity, due to the low carbon dioxide formation, although methane formation was higher. Using recirculation, the iron catalyst gave an over-all superior performance to the cobalt, a performance, moreover, which could be maintained for prolonged periods without a rapid rise in reaction temperature.
The C2 to C4 fraction had an average carbon number of 3:1 and contained 50 per cent of olefins. The liquid products (see Table 2) were similar in boiling range to those obtained with iron but contained a higher proportion of middle contained a higher proportion of middle oil. They were very similar to the products obtained in the normal atmospheric pressure, cobalt catalyst process. The olefin content of the liquid products was very much lower than that of the iron catalyst products and it is clear that the products as a whole were very much less suitable as a source of high quality motor fuel than those obtained with the iron catalyst.
Thus, although it is possible in the laboratory to employ the normal type of cobalt catalyst at space velocities up to 500 vol. synthesis gas per hour with an 80 percent. carbon monoxide conversions and obtain somewhat similar results to those obtained with a synthetic ammonia iron catalyst, the performance is inferior and cannot be sustained without excessive production of methane.
Rôle of Alcohols
In the synthesis with cobalt catalysts at temperatures in the range 180° to 200°C. and pressures from atmospheric to 10 atmos., as employed in the German commercial plants, the products, apart from water, are almost entirely hydrocarbon in nature. The total oxygen containing substances (alcohols, aldehydes, ketones, acids, etc.) do not amount to more than about 1 per cent of the whole, and it has been suggested that these substances arise as the result of side reaction (e.g., the OXO reaction) and play no part in the main reaction mechanism. When iron catalysts are used in the synthesis at 10 or 20 atmos. pressure, however, appreciable amounts of alcohols are produced. Thus, in the Synol process of the I.G.4,5, in which a synthetic ammonia iron catalyst is used at relatively low temperatures (190° to 220°C.) and with a high gas velocity, straight chain primary alcohols constitute 60 per cent of the liquid products. It seems unlikely that these alcohols can have been produced from olefins by the OXO reaction because, (1) the OXO reaction always gives a mixture of normal and iso-alcohols, (2) the OXO reaction only proceeds at a reasonable rate at pressures above 50 atmos., (3) iron catalyst are of low activity for the OXO reaction.
When a synthetic ammonia iron catalyst is employed at high temperatures (280° to 300°C.) as in the experiments reported in the preceding section of this paper, the alcohol content of the products is low but the olefin content very high.
The possibility, therefore, that alcohols are the true primary products of the Fischer-Tropsch synthesis, as was, in fact, suggested by early workers in the field 6,7, appeared to be worth further investigation.
In the first place, it was of importance to determine whether any evidence could be obtained that alcohols might be intermediate compounds in the synthesis with cobalt catalyst, and it has now been established that if the temperature is reduced materially below the normal synthesis range, appreciable amounts of alcohols appear in the reaction products. An increase in gas rate (decrease in time of contact) also leads to a small, but significant increase in alcohol content. The alcohol contents are appreciably higher at 10 atmos. pressure than at atmospheric pressure.
The experimental results obtained using a cobalt thoria magnesia kieselguhr catalyst and normal synthesis gas (H2:CO-2:1) are shown in Table 4. The alcohol and olefin contents are expressed both as percentages of the liquid products (b.p. 30° to 300°C.) and of the total non-aqueous products of the reaction, including methane and the wax deposited on the catalyst. The alcohol content of the reaction water amounted to between 1.3 and 1.5 per cent of the total products at 10 atmos. pressure but did not vary markedly with change in reaction conditions.
Distribution of alcohols
The alcohols were distributed throughout the liquid products as shown by the following data for synthesis at 10 atmos. pressure:--
°C 160 167 173
Alcohols, per cent by weight of fraction
30°-200°C 19.0 19.7 14.9
200°-300°C 27.1 19.5 10.5
The results given in Table 4 are consistent with the view that alcohols are the initial synthesis products and that they become dehydrated to olefins which undergo hydrogenation to the paraffins to a degree dependent on the reaction conditions. This view is given support by the behaviour of pure alcohols in the presence of Fischer-Tropsch catalysts at temperatures in the usual synthesis range. The results of some experiments with n-propyl and n-octyl alcohols are summarized in Table 5. the alcohols were passed over the reduced catalyst at a rate similar to that at which products are produced in the Fischer-Tropsch process, i.e., approximately 1 g. per 70 ml. catalyst per hour.
It will be noted that with the cobalt catalyst at the normal synthesis temperatures and with hydrogen or nitrogen as carrier gas, the alcohols were completely decomposed, but in the presence of water gas, some alcohol remained unchanged. At temperatures below the synthesis range, n-propyl alcohol was only converted to the extent of 30 to50 per cent. In the presence of an iron catalyst a temperature of 260°C. was necessary to effect an 80 percent dehydration of n-propyl alcohol in an atmosphere of hydrogen.
The reaction products have not yet been completely examined, but from preliminary examination it seems that at 150°C., the main products are water and the corresponding olefin and paraffin. At higher temperatures, gaseous hydrocarbons were produced from octyl alcohol and the liquid products had an extended boiling range. At 150°C some higher-boiling products were produced from propyl alcohol.
Eidus8 studied the behaviour of methyl alcohol over cobalt catalysts at 180°C. and obtained a range of gaseous and liquid hydrocarbons resembling those obtained in the /Fischer synthesis. The effluent gas in his experiments contained 31 per cent carbon monoxide and 67 per cent hydrogen (on a nitrogen-free basis) and he deduced that the hydrocarbons were formed by synthesis from the gaseous decomposition products of the alcohol. He concluded, therefore, that methyl alcohol cannot be an intermediate product in the Fischer synthesis. In the present experiments with higher alcohols in a nitrogen atmosphere, the maximum carbon monoxide content of the effluent gases (on a nitrogen-free basis) was 2.1 per cent. It is concluded is this case, therefore, that the products are derived not by synthesis from carbon monoxide and hydrogen, but by dehydration of the alcohols followed by reduction and hydrogenation cracking to form lower hydrocarbons, and by condensation reactions to form more complex substances. methyl alcohol is only present in very small amounts when the synthesis is carried out under alcohol producing conditions, and Eidus’ contention that this compound does not function as an intermediate is not disputed.
The fact that the iron catalyst was very much less active than the cobalt for alcohol dehydration is in conformity with the observation that Fischer products obtained with iron catalyst always posses an appreciably higher alcohol content than those obtained with cobalt.
Much more experimental work will be required in order to establish clearly the role of alcohols in the Fischer-Tropsch synthesis, but the preliminary results reported above, strongly suggest that formation of alcohols is a stage in the main reaction mechanism and not the result of an independent side reaction.
It is hoped as a result of further work to build up a picture of the reaction mechanism into which the whole range of syntheses based on carbon monoxide and hydrogen (methane synthesis, methanol synthesis, synthesis of higher alcohols and higher hydrocarbons) can be fitted.
The work described in this paper forms part of the programme of the Fuel Research Board of the Department of Scientific and Industrial Research and this account is published by permission of the Director of Fuel Research. The illustrations are Crown copyright and are reproduced by permission of the Controller of H.M. Stationery Office. In the work on the synthesis process at high space velocities, the author was assisted by Mr. S. L. Smith, Mr. J. Rennie and Dr. D. Gall, and in the study of reaction mechanism, by Mr. E.J. Gibson.