too slow, in which case the mobile alkali level on the nickel is too low and the catalyst will not operate a t low steam ratios. Potash loss is thus reduced to the necessary minimum even with a catalyst containing a large quantity of potash by employing what might be termed a “slow release potash reservoir.” Kalsilite can readily be made by firing together an intimate mixture of china clay and potash at temperatures of over 750” C. Further Developments in Naphtha-Reforming Catalysts
The mechanisms by which carbon is formed indicate that the tendency toward carbon formation is likely to be greatest a t that point in the reformer tube where the rate of thermal cracking of naphtha is greatest. Measurements of gas composition profiles show that this position generally occurs about one third of the tube length from the inlet in the temperature range 625“ to 675”C., where both the concentration of naphtha and the cracking velocity constant are high. The first signs of carbon deposition are also known to occur in this region. I t is in this region, therefore, that the mobile alkali concentration needs to be highest and lesser levels of alkali should be sufficient elsewhere to inhibit carbon buildup. When coupled with the desire to reduce the over-all evolution of alkali from the reformer tube, the above considerations suggest that a zoned catalyst system would be beneficial. The upper part of the tube would be filled with catalyst containing a high level of mobile alkali, whereas the hotter lower part would perhaps contain a nonalkalized catalyst. Experiments have shown that this can be done and a modified nickel supported on aluminalime formulation was found to be satisfactory for the bottom catalyst. In addition to reducing the alkali evolution, the elimination of mobile alkali from this catalyst also increases the catalyst activity for the methane-steam reaction very considerably, in agreement with the experiments mentioned above. Typically employing a 3/4 charge of alkalized catalyst (IC1 46-1) in the first part of the
tube and a f/4 charge of the nonalkalized catalyst (IC1 46-2) in the bottom part of the tube allowed tube throughputs to be increased by about one third for any particular set of exit temperatures and gas compositions. No reduction in the ability to withstand carboning up was noticeable. Two-zone catalyst charges of this type have now become common in commercial plants during the past two years. Although most of the researchers in naphtha-reforming catalysts over the past 10 years have been concerned with modification to the support of nickel-based catalysts, other catalytic metals are on occasions suggested. Cobalt is an obvious alternative metal to try, as its catalytic behavior is very similar to that of nickel. Various precious metals have also been suggested as reforming catalysts, and experiments show that all these metals have appreciable reforming activity. In general, however, the cheapness of nickel tends to outweigh any significant advantages from the uses of other metals. literature Cited
Andrew, S. P. S., Eastbourne Conference on Fuel Research and Development, Institute of Fuel, London, 1965. Arnold, M. R., Atwood, K., Baugh, H. M., Smyser, H. D., Id.Eng. Chem. 44,999-1005 (1952). Blayden, H. E., Riley, H. L., Shaw, F., Fuel 22, 64-71 (1943). Borgars, D. J., Ind. Chemist 39, 177-80 (1963). Bridger, G. W., Wyrwas, W., Chem. Proc. Eng. 48, 1015 (1967). Dent, F. J., Moignard, L. A., Eastwood, A. H., Blackburner, W. H., Hebden, D., Trans. Inst. Gas Eng. 95, 604-709 (1945). Harker, H., “Proceedings of Fourth Conference on Carbon,” pp. 125-39, Pergamon Press, Oxford, 1960.
RECEIVED for review February 6, 1969 ACCEPTED July 22, 1969
ROLE OF URANIA AND ALUMINA AS SUPPORTS IN THE STEAM
REFORMING OF n-BUTANE AT PRESSURE OVER
NICKEL-CONTAINING CATALYSTS K .
S .
M. BHATTA
AND
G .
M. D I X O N
Gas Council London Research Station, Michael Road, Fulham, London, S. W.6., England
THEproduction of
town gas (calorific value 500 B.t.u. per cu. foot) based upon carbonization processes ceased to be economic in the United Kingdom in the early 1950’s. At the same time the relative cheapness of light petroleum distillate (LPD) led to the development of gas-making processes based upon the hydrogasification and steam reforming of petroleum feedstocks. The first commercial catalytic steam-reforming process in the United Kingdom, utilizing desulfurized liquid hydro324
I & E C PRODUCT RESEARCH A N D DEVELOPMENT
carbon, was commissioned by IC1 (Andrew, 1965) in 1962 for the production of a hydrogen-rich gas a t high pressure and temperatures in the region of 75WC. It was closely followed by the U. K. Gas Council’s catalytic rich gas (CRG) process (Cockerham et al., 1965; Dent, 1964) for the continuous reforming of desulfurized L P D a t 50PC. over 75 weight % nickel alumina catalyst and a t pressures up to 30 atm. to produce a methane-rich gas; under these conditions the reaction is thermally self-
Catalysts used for the steam reforming of hydrocarbons usually contain nickel as the active component. Promoters-for example, potassium-are also added to neutralize the acidity of the support and so mitigate the effect of coke formation. Coke formation on the active sites of the catalyst is reduced by the introduction of uranium to the catalyst formulation. Kinetic data imply that water is preferentially adsorbed on the active surface of the nickel-urania catalyst, whereas in the case of a nickel-alumina catalyst the active surface is covered with adsorbed hydrocarbons. One of the roles of the support may be to supply oxygenated species, resulting from the adsorption of water, to the surface of the metal.
maintaining, so that no other heat need be supplied. Town gas may then be made by further reforming a fraction of the rich gas at 650°C. and blending the lean gas so formed with the remainder. The main causes for catalyst failure in steam reforming are the sintering of the catalyst, with resultant loss of active area, and the fouling of the catalyst by deposition of coke. A catalyst developed by the North Western Gas Board (Nicklin et al., 1966; Nicklin and Whittaker, 1967), comprising nickel and urania supported on corundum, mitigates the above effects. Urania was chosen for incorporation into a reforming catalyst initially on account of its known resistance to sulfur poisoning, although its ability to promote the dissociation of steam (Nicklin and Whittaker, 1968) and its low surface acidity made it a suitable support material for the steam reforming of hydrocarbons. The product gas composition in steam reforming is governed by equilibrium considerations; the shift reaction, CO + H20 COZ+ 142, and the steam methane reaction, CO + 3H2 CH, -1 H20, attain a close approach to equilibrium. At high temperatures the equilibria are displaced toward the formation of hydrogen; at low temperatures, toward methane. For any one temperature range the steam-hydrocarbon ratio chosen is such that the equilibrium mixture will not contain Boudouard carbon. As the reaction proceeds to equilibrium, however, carbon formation ma:y be favored thermodynamically at one or more points along the way. There is no real agreement about the reactions taking place in the steam-reforming process, but there is general agreement (Akers and Camp, 1955; Bodrov et al., 1964; Eickmeyer and Marshall, 1955; Reisz et al., 1952; Yarze and Lockerbie, 1960) that the initial products are hydrogen and the oxides of calrbon. Formation of methane may result from the hydiogenation of the oxides of carbon and of adsorbed CHZ-type species on the surface of the catalyst. This paper is conclerned with the steam reforming of n-butane a t 30-atm. pressure over a nickel-alumina catalyst and a nickel-and-urania-impregnatedcorundum catalyst. Catalysts siinilar to the ones described in this paper have been used on an industrial scale for the steam reforming of LPD, the exit gas composition being determined by the equilibrium considerations mentioned above.
--
Apparatus and Materials
The flow apparatus, shown schematically in Figure 1, was constructed from Aminco Super-pressure equipment. The reactor, R I , 80 cm. long and constructed of Xs-inch internal diameter and Xe-inch external diameter stainless steel was mounted vertically and surrounded by a furnace, 70 cm. in length. The catalyst was supported on a 2-cm layer of 20- to 30-mesh (ASTM) quartz
PRIMARY REACTOR (R1 I
Cu
DEOXO
UNIT
ELlUM CYLINDER
Figure 1. Apparatus
chips held in position by a perforated stainless steel disk; a further 15 cm. of quartz chips above the catalyst bed served as a preheat section. The temperature of the catalyst bed was measured by means of an axial thermocouple of 0.02-inch diameter. The mixer, M , was maintained at 300" C. and all the lines from the on-off valve, 7'2, to the condenser, C1, were heated to approximately 250°C. LS and L7 are pressure-control valves; LI, Lz, LB, and L4 are water-cooled on-off valves; and Fi and F1 are highpressure filters. The secondary reactor, R2, 20 cm. long and similar in construction to the main reactor, was used to convert any butane remaining from the primary reactor which would otherwise freeze out on the diaphragm seating of the dome valve, D. The sample bleed-off line from the air-cooled needle valve, Ls, was heated to 100°C. and a small water condenser, C2, removed most of the water from the effluent gases. Water was pumped into the mixer. Butane was displaced into the mixer from a high-pressure sampling bomb by means of water from a second pump. The flow of diluent helium was controlled by means of a needle valve, N, and monitored by a Honeywell, Type 1028, differential-pressure transducer, D.P.T., connected across a stainless steel capillary. Since the transducer could not withstand a pressure differential of greater than 20 psi., twice its operating differential, a bypass was used to protect it from pressure surges during the initial stages of startup. The total pressure in the system could be closely controlled to the desired value by the dome valve, D,which was externally pressurized. The line pressures in the apparatus and the dome valve were indicated by Bourdon gages and monitored to 0.1 atm. by linear pressure transducers. n-Butane (Phillips 66 pure grade) of purity better than 99.9% was used without further treatment. Cylinder hydrogen (Air Products, Ltd.) was purified by passing it through a Deoxo unit and a molecular sieve dryer (Linde 13X). Cylinder helium (Air Products, Ltd.) was purified by passage through a 6 x % 6 inch i.d. bed of copper filings at 35P C. and dried by means of molecular sieves (Linde 13X). Dissolved air was expelled from distilled water by passing helium through 250-cc. aliquots of the liquid at 100 cc. per minute for 30 minutes. The nickel-alumina catalyst, containing 15 weight 70 nickel, was obtained in the form of %-inch pellets from P. Spence and Sons, Ltd. All experimental runs were carried out on approximately VOL. 8 NO. 3 SEPTEMBER 1969
325
0.35-gram samples of this catalyst of particle size 40- to 60-mesh (ASTM). The total surface area of the reduced catalyst, as determined by the BET method with nitrogen as adsorbate, was 213 sq. meters per gram; the metal area, as estimated by hydrogen chemisorption, was 6.8 sq. meters per gram. X-ray analysis of the reduced catalyst showed that the average size of the nickel crystallites was approximately 40A. and the alumina was present as ?-alumina. The nickel-urania-corundum catalyst was obtained from the North Western Gas Board. It was prepared by impregnating corundum granules with a mixed solution of nickel and uranyl nitrates. Potassium was added by immersing the calcined material in a solution of potassium acetate and calcining the dried catalyst at 550°C. All experimental runs were carried out on approximately 1.3-gram samples of this catalyst particle size, 40- to 60-mesh (ASTM). The total area of the reduced catalyst was 16 sq. meters per gram. Analysis of the catalyst gave nickel, 13.0 weight 7c; uranium, 12.1 weight '%; potassium, 0.3 weight 70; the remainder being the corundum support. X-ray analysis of the calcined catalyst showed the presence of nickel uranate, NiO.3U03, which on reduction at 500°C. gave metallic nickel of crystallite size 250A., and
it has been assumed t h a t the deposition of the carbon is slow. Equations 1 and 2 may then be used to determine the equilibrium concentration of products in the presence of unreacted n-butane and helium. Thus, for a mole ratio of water to n-butane of a and a helium t o n-butane ratio of b the equilibrium constants for Equations 1 and 2 may be expressed as follows:
K1=
Si.X(U - 0.08m + 2~ + Y ) ' ( u + b + 1 + 0.0%~- 2 ~ ) ' (0.04m - x - y ) ( 0 . 1 3 ~~ 4~ - Y ) ~ "
Kz =
S2- y ( a - 0.08m + 2x + y ) ( 0 . 0 4 ~-~x - y ) ( 0 . 1 3 ~~ 4~ - y )
(14
(24
u408.
Operating Procedure
The catalyst was reduced at 500°C. at atmospheric pressure for 16 hours in a stream of hydrogen flowing at 70 cc. per minute. Helium was then passed through the reactor at the same rate for 1 hour. The reactor temperature was adjusted and the system was pressurized to 30 atm. Water and butane were then pumped into the mixer. Initially the bypass was used for approximately 10 minutes until the three flows had come to equilibrium. To test efficiency of mixing, nitrogen was used as a diluent in the place of helium and a sample of the effluent gas was passed through the gas-chromatography unit with helium as a carrier. Over a wide range of operating conditions the experimentally determined composition of the effluent agreed with that calculated from the input flows. Methods of Analysis. The effluent was analyzed by gas chromatography for H2, CO, COZ, and Ci to C4 hydrocarbons. Helium was used as the carrier gas. The sample (1 cc.) was split between two parallel columns maintained at 40°C. One fraction was passed through an 18-foot, %-inch i.d. silica gel column to separate CO, CO,, Hz, CHa, and C Z H ~and , then analyzed by thermistor detector; the other portion was separated on a 6-foot, %-inch i.d. column of silica gel poisoned by diethylhexyl sebacate and the hydrocarbons were analyzed by a flame ionization detector. Water and helium were estimated from an over-all butane balance, since the mole ratios of helium and water to butane were known. Thermodynamic Analysis
T o elucidate the initial products of reaction and the subsequent reactions leading t o full equilibrium among methane, carbon monoxide, carbon dioxide, hydrogen, and water, the approach t o equilibrium for each component as the parent hydrocarbon is progressively consumed was determined. The reactions which might occur on the catalyst after the initial reaction of the hydrocarbons with steam may be represented by the following equations:
CO + 3H2 2 CH, + Hz0 CO,
+ CH4 2 2CO + 2H2 CH4 2 C + 2H2 2 c o 2 c + coz
(3)
(4) (5)
(6) I n calculating the theoretical equilibrium composition of t h e components a t different conversions of the butane 326
I&EC PRODUCT RESEARCH A N D DEVELOPMENT
where x and y are numbers of moles of methane and carbon monoxide, respectively, at equilibrium, m is the percentage conversion of t h e n-butane, P is the total pressure in atmospheres, and SI and S2 are the fugacity terms for Equations 1 and 2, respectively. Critical data were obtained from the literature (Wagman et al., 1945); from fugacity charts compiled by Newton (1935), SI and S2 were taken as 0.94 and 0.98, respectively. Equations l a and 2a were solved numerically for x and y values of m u p to 100. Besides the possibility t h a t carbon may be deposited by the breakdown of n-butane, C4H10+ 4C + 5H2, it may also be deposited by three reactions involving t h e products of the butane-steam reaction-Reactions 5 , 6, and 7. However, a water t o n-butane ratio may be chosen such that at 100% conversion of the hydrocarbon no carbon is present in an equilibrium mixture. I t does not follow that under these conditions carbon will not be formed as the butane is progressively converted and it is possible t h a t carbon formation will be favored thermodynamically at one or more conversion levels. Results
Preliminary Tests. Prior t o obtaining the experimental data, tests were made t o determine the extent of homogeneous reaction and the catalytic activity of the silica chips and the stainless steel. No decomposition of the butane was detected at 500" C. A series of experiments at 450°C. on the stability of the nickel-alumina catalyst showed that its activity decreased as a n exponential function of time (Bhatta and Dixon, 1967), a region of approximately constant activity being attained after about 3 hours. T h e initial rate of decline in activity was independent of the partial pressure of butane and declined with increasing partial pressure of water. Addition of 1.8 weight % potassium t o the catalyst prior to the pelleting stage had no effect on the initial rate of decline of catalytic activity (Figure 2 ) . With the nickel-urania catalyst the decrease in activity was less marked (Table I ) , and no appreciable effect on the decline in activity was observed for catalyst samples without potassium. Treatment with hydrogen a t 500" C. for 16 hours restored the activity of the nickel-urania and nickel-alumina catalysts t o approximately that of fresh samples. Evidence for the occurrence of high-molecularweight hydrocarbons such as naphthalene, anthracene, and
PARTIAL PRESSURE OF STEAM (ATM)
Figure 2. Effect of partial pressure of steam on initial rate of decline in activity of nickel-alumina catalyst Temp. 450°C. Pressure 30 atm. Flow rate 0.1 male/min. Partial pressure of butane 1.0 atm.
0 Nickel-alumina A Nickel-alumina
+ 1.6 wt. 'Yo K 2 0
pyrene was found by gas-chromatographic analysis of material extracted from the used catalysts. No elemental carbon was detected Ion the surface of the used catalysts. Kinetic Data. The kinetic data were obtained after the activity of the catalysts had declined to an approximately constant value. I n these experiments the conversion of butane was in the range of 1.0 to 4.0%, so that the rates are the initial r,stes of reaction. T h e reaction rates were determined from the relatiop, rate = ( F / W ) x ,where F is the feed rate of n-butane to the reactor in moles per minute, W is the weight of catalyst in grams, and x is the fraction of butane converted. _
_
_
_
_
~
~
~
~
~
Table 1. Decline in Catalytic Activity with Time
Total pressure Butane pressure Water pressure Flow rate Temperature
Time, Min. 10 30 60 90 120 150 180 210 240
30 atm. 1.5 atm. 8.0 atm. 0.1 mole/min. 450°C.
Moles Butane x 10' Conuerted Per Minute Der Gram o f Catalyst Nickel-alumina Nickel-urania 1.38 0.94 0.54 0.24 0.11
0.025 0.022 0.021
0.077
0.021
0.074 0.069 0.061
0.024
0.022 0.023 0.021 0.020
NICKEL-ALUMINA. Bhatta and Dixon (1967) have shown that the initial rate of reaction, ro, between butane and steam may be expressed as follows: -AEJRT
r, = h(C4H10)o(H20)1e where AE has the approximate value of 13 kcal. per mole over the temperature range 420" to 480" C., with a partial pressure of butane of 1.63 atm. and a partial pressure of steam of 9.1 atm. NICKEL-URANIA. I n the temperature range of 404" to 491" C. and a total pressure of 30 atm. the partial pressures of butane and water were varied between 0.6 and 2.2 atm. and 4.3 and 19.7 atm. respectively. I n the range of input flows used, 5.0 to 10.0 moles per hour, the resistance to mass transfer of the components between the gas stream and the surface of the catalyst was shown to be negligible by varying the mass velocity through the reactor while holding all other conditions constant. No significant change in catalytic activity was also found on using particles of 30- to 40-mesh size. The initial rate of reaction a t 450" C. and over a range of partial pressures 0.6 to 2.2 atm. of butane and 4.3 to 19.7 atm. of steam, may be expressed as follows: -AE/RT
e
where AE has a value of 24 kcal. per mole. T h e apparent energy (Figure 3) was obtained over the temperature range 404" to 491"C., with a partial pressure of butane of 1.65 atm. and a partial pressure of steam of 11.0 atm. VOL. 8 N O . 3 S E P T E M B E R 1 9 6 9
327
3
1.45
1hCJ x103
I
1'45
I
150
Figure 3. Decline in activity of nickel-alumina catalyst with time Temp. 450" C. Pressure 30 atm. Flow rate 0.1 mole/min. Mole ratio woter-n-butane 25.0
100
to approximately its original value by hydrogen treatment a t 500" C. NICKEL-URANIA. Typical data obtained on reforming n-butane to partial conversion in the temperature range 404" to 491°C. a t various steam to hydrocarbon ratios are presented in Table 11. At all conversion levels the reaction products consisted almost entirely of conventional reformed products and unreacted feedstock. At low conversions traces of ethane and propane were present in the effluent. The ratios of the equilibrium constants calculated from the gas analysis to those calculated from thermodynamic data for the shift reaction
90 80
-E
70
60
z
0
JI
6 50 >
0
LO
30 20
and the methanation reaction 10 -
0
1
I
I
I
I
I
I
I
I
I
_
CO + 3Hz
..
I
Figure 4. Arrhenius plot for nickel-urania catalyst Pressure 30 atm. Partial pressure of butane 1.65 atm. Partial pressure of steam 11 .O atm. Flow rate 0.1 mole/min.
CH, + HzO
a t different percentage conversions of the butane are reported in Table 11. The gas compositions calculated from Equations l a and 2a, assuming that the above two reactions are in equilibrium at all conversions of the parent hydrocarbon, are shown in parentheses in Table 11. Discussion
Product Distribution. NICKEL-ALUMINA. For the nickelalumina catalyst the reaction products were a t equilibrium (Bhatta and Dixon, 1967) a t all conversions of n-butane, irrespective of the input conditions used. The decline in catalytic activity a t 450°C. for a water to n-butane mole ratio of 25 is shown in Figure 4. Under these conditions any carbon would be removed from the catalyst at all conversions of the parent hydrocarbon by the reverse of Reactions 5, 6, and 7. Carbonaceous material was again present on the catalyst and the activity was restored 328
I&EC PRODUCT RESEARCH A N D DEVELOPMENT
Comparison of the initial rate data shows a marked difference in the affinity of the same reactant for the surface of the two catalysts. For the nickel-alumina catalyst the surface is fully covered with adsorbed hydrocarbon which reacts with water from the gaseous phase or with water adsorbed onto sites other than those occupied by the hydrocarbons. I n the case of the nickel-urania catalyst water is preferentially adsorbed on the surface and reacts with hydrocarbon fragments occupying the remaining sites.
Table II. Effluent Composition over Nickel-Urania Catalyst
Mole Ratio Conuer-~ Composition, Mole % a sion, Temp.. H20/ He/ c. C4HlO GHio CH, co GO2 H% 7c CJIio 0.233 0.059 1.571 4.361 9.98 475 4.41 15.10 4.203 (2.741) (0.860) (0.025) (1.024) (4.305) 3.794 4.325 0.136 0.046 1.316 7.97 475 4.41 15.10 (2.584) (0.591) (0.021) (0.916) (4.412) 1.028 2.99 5.01 0.096 0.046 5.47 475 2.07 15.10 (0.489) (0.022) (0.678) (1.886) (5.14) 0.188 0.164 1.327 5.689 7.42 10.053 4.01 451 0.58 (1.323) (10.525) (1.093) (0.023) (0.643) 3.944 0.345 0.122 2.064 8.012 13.83 455 5.29 13.78 (4.191) (1.494) (0.020) (1.177) (2.95) 2.69 1.073 0.239 2.569 10.937 462 5.28 17.86 26.5 (2.95) (2.733) (0.036) (1.487) (2.864) 491 6.66 10.43 0.162 4.81 0.036 1.689 4.352 8.9 (4.91) (0.726) (0.020) (1.18) (3.337) 491 6.66 24.89 2.58 0.161 0.052 1.283 3.814 12.7 (2.63) (0.53) (0.029) (0.966) (2.808) Figures in brackets are theoretical equilibrium amounts.
Although nickel itself is active for the steam reforming of hydrocarbons, it is difficult to explain the radical change in reaction orders for the two catalysts without assuming the support influences the over-all reaction as well as serving to prevent sintering of the nickel crystallites. Certainly, in the case of urania, it is possible that a complex with nickel is formed a t the nickel-urania interface which is active for both the breakdown of hydrocarbon and the dissociative adsorption of water. On the other hand, the support may act mainly as a reservoir for the supply of oxygenated species from the breakdown of water, the rate a t which these are transferred to the surface of the nickel depending upon their concentration and rate of migration across the surface of the support. I t has been reported (Dowden et al., 1968) that urania is more active in promoting steam cracking than is 7-alumina on account of being an oxygen-deficient compound with compositions ranging from UO, to u308. If, therefore, adsorption equilibrium between water and the support is rapidly attained, the supply of oxygenated species to the nickel would be controlled by the concentration and rate of migration of these species on the surface of the support. A suggested reaction scheme is: C4H10
where hl is the rate of adsorption of water onto the support, hl is the rate of desorption of water, K3 is the rate of migration of water across the surface of the support, and k4 is the rate of adsorption of butane onto the nickel surface. The chemical step for the over-all reaction is assumed to involve adsorbed hydrocarbon species and adsorbed oxygenated species. If for urania
HzO 19.25 (18.95) 19.59 (19.22) 10.22 (9.85) 4.907 (5.028) 22.43 (23.29) 17.1 (18.1) 33.81 (33.52) 18.59 (18.04)
He 70.32 (72.04) 70.78 (72.21) 80.56
Shift Methanation Constant, Constant, Experimentall Experimentall Theoretical Theoretical 1.04