385
Ind. Eng. Chem. Fundam. 1982, 21, 385-390 Halta, S. Toh&u Imp Unlv. Tech. Repts. 1928, 8 , 1. Olander, D. R. AIChE J. 1960, 6 , 223. Rochelle, 0. T.; King, C. J. Ind. Eng. Chem. Fundarn. 1977, 76, 67. Secor, R. M.; Beutler, J. A. AIChE J. 1987, 73, 365.
Chang. C. S., Ph.D. Dissertation, The University of Texas at Austin, 1979. Chang, C. S.; Rochelle, 0. T. “SO2 Absorption into NaOH and Na,SO, Aqueous Solutlons”; presented at the AIChE 88th Natlonai Meetlng, Philadelphia, June 8-12, 1980. Chang, C. S.; Rochelle, 0. T. AIChE J. 1981, 27, 292. Chang, C. S.; Rochelle,’G. T. AICh€ J. 1982, 28, 261. Danckwerts, P. E. Chem. Eng. Sci. 1968, 23, 1045. Danckwerts, P. V. “Gas-Liquid Reactions”; McGraw-Hill: New York, 1970.
Received for review June 1, 1981 Accepted June 11, 1982
Methanation over Transition-Metal Catalysts. 4. Co/AI,O,. Behavior and Kinetic Modeling
Rate
Pradeep K. Agrawal,” James R. Katzer,’ and Wllllam H. Manogue’ Center for Catalytic Science and Technology. Department of Chemical Engjneering, University of Deb ware, Newark. Delaware 79777
in three pseudosteady-state operating regions-clean Co, carbondeactivated Co, and sulfur-poisoned Co. The intrinsic rate data in all three regions are correlated well by the “carbide”model in which the rate-limiting step is the reaction between a surface carbon atom and a surface hydrogen atom. Kinetic modeling suggests the existence of two types of CO hydrogenation sites-one with the higher heat of CO adsorption as compared to the other. As cobalt undergoes carbon deactivation, the active sites with higher heat of CO adsorption are reduced considerably and are poisoned completely with sulfur. However, no change in either the reaction mechanism or the rate-controlling steps is caused by catalyst deactivation. The kinetic behavior of CO hydrogenation over alumina-supported cobalt has been presented
Introduction Many catalysts undergo changes from their initial activity to a more industrially important pseudo-steady state characterized by a slow continuous activity loss. Kinetic studies have sometimes been carried out on the initial activity behavior of catalysts and on the steady-state activity, but the authors know of no kinetic studies carried out in all activity regions. Furthermore, it is possible for catalysts to exhibit more than one pseudo-steady state activity region; yet examples of such situations have not been clearly documented, and there are no reported examples of kinetic studies. Careful kinetic studies of the initial catalytic activity and of each of the pseudosteady-state activity regions, if more than one exist, can help clarify the nature of the deactivation process. If deactivation is due only to a reduction in the number of active sites, then the kinetic behavior should remain unchanged, and the reduction in the rate constant could be attributed to the fractional reduction in the number of active sites, properly calculated to account for the kinetic order. If the deactivation process is due to other factors, such as electronic changes, then the form of the kinetic rate expression could change, the value of the rate constant could change, and the values of the kinetic rate parameters would be expected to change, leading to new insights into the deactivation mechanism.
Previous papers in this series (Agrawal et al., 1981a,b; Fitzharris et al., 1982) report CO hydrogenation studies over Ni and Co catalysts supported on AlZ0,. Reliable experimental techniques have been developed for measuring intrinsic kinetics of CO hydrogenation at specified CO concentrations,at 1atm total pressure, at temperatures to 673 K, and with controlled amounts of HzS to concentrations as low as 13 ppb. This reactor system allows reaction rates and rates of deactivation to be measured directly rather than being calculated with the aid of an assumed transport, kinetic, and deactivation model. In this research the kinetic behavior of Co/AlZ0, in CO hydrogenation has been determined in three pseudosteady-state regimes having activities that differ by more than four orders of magnitude. The emphasis of this work has been on quantifying the kinetic behavior adequately to determine the best kinetic model which describes CO hydrogenation over Co/A120, and to determine the form of the rate expression in each pseudo-steady-state regime. The information gained provides strong support for the rate-determining step in CO hydrogenation. The kinetic behavior observed in the three pseudo steady-state regions indicates the nature and cause for the deactivation.
Theory In spite of an extensive literature devoted to the adsorption of H2 and CO on transition metals, comparatively few studies have been made of the adsorption of these molecules and their mixtures under conditions close to those of catalytic processes. Insufficient characterization of the surface intermediates found under reaction conditions and the great disparity of conditions between surface science studies and reaction studies have kept researchers divided on the identification of intermediate surface complex of methanation. All suggestions in the literature can
*Author to whom all correspondence should be addressed School of Chemical Engineering, Georgia Institute of Technology, Atlanta, GA 30332. ‘Central Research Department, Mobil R & D Co., Princeton, NJ 08540. Experimental Station, E. I. DuPont de Nemours and Company, Inc., Wilmington, DE 19898. 0196-4313/82/1021-0385$01.25/0
0
1982 American Chemical Society
388
Ind. Eng. Chem. Fundam., Vol. 21, No. 4, 1982
Table I. Rate Expressionsa Obtained from “Carbide” Theory and “Enol Complex” Theory rate-determining step
carbide theory
enol complex theory
CO adsorption CO dissociation H, adsorption Rff
1st step in surface reaction
Rff
2nd step in surface reaction a
xCOxH,o’s
(1 + h , X c o t
~ , X H , ~ t’ ~ .
‘
.)’
“CNH,
(1 + h , x c O t
+ . . .)’
Rol
Rff
~,XH,~”
xCOxH,
(1 +
hixCO
kGH,)’
xCOxH,2
(1 + k , X c o t h,XH,
+ . . .)’
hi’s represent a single constant or a combination of terms to simplify the expression.
be summarized into essentially two different mechanisms (Farrauto, 1976; Mills and Steffgen, 1973; Vlasenko and Yuzelfovich, 1969): CO dissociates and forms “carbide”, which is then hydrogenated to form methane; CO is first partially hydrogenated to a CH,O surface species which is subsequently hydrogenated to methane. The first mechanism is commonly referred to as the ”carbide” mechanism, and the reaction sequence can be represented as follows co(g) + * e toad step 1: step 2: step 3: step 4: step 5: step 6: step 7: step 8:
+ * Cad + Oad H2 + 2* + 2Had Cad + Had CH,d + * CH,d + Had CHzad + * CH,,d + Had * CHsad + * CH,,d + Had CHdad + * toad
F+
CH4ad
+
CH,(g)
+*
where * represents a vacant surface site. The second mechanism is referred to as the “enolic complex” mechanism and the reaction sequence can be represented as step 1: co(g) + * toad step 2: step 3: Step 4:
Hz + * + Hzad CO,d + H2,d 6 CHOH,,j CHOH,d
+ HZad
+*
H20ad +CHZ,d
and so on. Kinetic expressions can be developed by assuming one step to be the rate-determining step. Steps occurring after the rate-determining steps are not kinetically important, and steps preceding the rate-determining steps are assumed to be in a state of dynamic equilibrium. Analysis is presented elsewhere (Agrawal, 1979); the results are summarized in Table I. There are large differences in the dependence of methanation rate on Hz and CO concentrations for different rate-controlling steps, but the general form of rate expressions are the same in the “carbide” and the “enol complex” mechanisms for similar rate-controlling steps. Using dissociative chemisorption of Hz in the ”enol complex” mechanism would yield an identical dependence of the methanation rate on Hz concentration in both theories. This comparison shows that in this case a reaction mechanism cannot be established on the basis of kinetic data alone. Experiments which identify with the
molecular processes at the surface (surface intermediates) are required. Experimental Section An all-quartz internal-recycle reactor, developed in our laboratory (Fitzharris and Katzer, 19781, was used for kinetic measurements. In addition to quartz being unreactive toward CO hydrogenation, it is also inert toward sulfur adsorption as compared to glass or stainless steel construction. The flow behavior of gases through the reactor was observed to be that of an ideal continuous flow stirred-tank reactor, based on tracer studies. The absence of external transport effects was demonstrated by varying the stirrer speed and observing no change in the reaction rates. High purity (99.9%) fused -pAl,O, plates were used as the catalyst support because they had negligible internal surface area; therefore, all the surface metal atoms were exposed to a uniform gas environment. This support configuration ensured the absence of intraparticle (pore diffusion) effects which is particularly important for studying the metal poisoning by sulfur. The inertness of quartz reactor and alumina support toward CO hydrogenation was evident from a complete absence of any traces of CHI in the reactor effluent stream at typical reaction conditions. The Co/Al2O3catalysts used in our studies were prepared by single-impregnation,multi-impregnation, and evaporation techniques (Agrawal et al., 1981a). The methanation turnover numbers for these catalysts prepared by different methods were reproducible within &20%; also, the kinetic behavior (e.g., activation energy, CO partial pressure dependence, etc.) was independent of the method of catalyst preparation. The above reactor configuration allowed reaction rates and rates of deactivation to be measured directly in the absence of any transport effects. All gases were analytical grade and were purified further. The H2S handling part of the reactor flow system was especially designed using all Teflon tubing to prevent HzS uptake and back contamination (Fitzharris, 1978; Fitzharris et al., 1982). During the measurements of methanation rate dependence on reactant (CO or H,) concentration, the concentration of only one of the reactants was varied, while that of the other was held constant. Furthermore, rate dependence measurements were made with both decreasing and increasing reactant concentrations. Typical kinetic measurements were made over a period of 5-6 h; during this period the activity changed at most by 6-8%. In all kinetic measurements a return to the earlier operating conditions resulted in the same activity (within a 6-8% error), confirming that the kinetic behavior observed represents intrinsic behavior of the
Ind. Eng. Chem. Fundam., Vol. 21, No. 4, 1982 387 10’
II
F
Table 11. Kinetic Behavior of Co/A1,0, in Different Regions of Methanation Activity ~~
NCH, at 673 K,
fresh carbon sulfur catalysta deactivateda poisonedb 10.0 0.10 0.001
S-
E,, kcal/mol effect of Pco, NCH4
effect of NCH4 a 0
15
30
45
EO
(pCO)” PH
28k 2
-0.24
0.5
’
16+2 0.3 to 1.0
0.5
16t 2 0.3 t o 1.0
_-_
@
Agrawal e t al. (1981a).
Agrawal e t al. (1981b).
75
TIME, HR
Figure 1. Effect of CO concentration on deactivation of Co/A1203. Reaction conditions: (1) 0-4% CO in H2 at 673 K; (2) (0) 0.5% CO in H2at 673 K, 0-4% CO in H2at 673 K (Agrawal et al., 1981a).
catalyst and not any artifacts resulting from catalyst deactivation. The surface and subsurface composition of the reaction-aged catalysts was examined using Auger electron spectroscopy (AES). All experimental details of reactor setup and surface analysis equipment are described in earlier papers (Agrawal, 1979; Agrawal et al., 1981a,b; Fitzharris et al., 1982; Fitzharris, 1978).
Results Three activity regions for methanation over Co/A1203 were observed and studied (Agrawal et al., 1981a,b). Under sulfur-free reaction conditions, two steady-states were observed: (i) an upper pseudo steady state and (ii) a lower pseudo steady state. A third steady state was observed for sulfur-poisoned Co catalyst. General deactivation behavior is first reviewed; then the specific kinetic behavior in each region is discussed. Under sulfur-free reaction conditions, Co catalysts are deactivated about 102-fold by carbon even in a large excess of H2 After initial reactor start-up transients have passed, there is a period in which the catalytic activity decreases only slowly (-30% per 24-h period) over a period whose duration is shorter at higher temperatures or higher CO concentrations; this period is referred to as the upper pseudo steady state. Figure 1 shows the effect of CO concentration on the duration of upper pseudo steady state at 673 K; with 0.5% CO in H2 the steady state lasts for more than 24 h, whereas with 4 % CO in H2 the steady state lasts for only 5 h. At the end of the upper pseudosteady-state period, the rate of catalyst deactivation accelerates and the activity drops almost 50-fold before approaching the lower pseudo steady state. The rate of deactivation in the lower pseudo steady state is about the same as that in the upper pseudo steady state. The deactivation and the transition from upper pseudo steady state to the lower pseudo steady state are quite complex processes and have been discussed in detail elsewhere (Agrawal et al., 1981a). In the presence of 13 ppb of H2S at 663 K, methanation activity with 1% CO in H2 is reduced by almost four orders of magnitude; the poisoning results from the formation of a two-dimensional, saturated surface sulfide layer corresponding to one sulfur atom per two surface Co atoms. Analysis of transient rate data shows that the methanation rate varies linearly with the square of the fraction of sulfur-free surface. The residual steady-state methanation activity with 13 pph of H2S is relatively insensitive to subsequent increase in the H2S concentrations (Agrawal et al., 1981b). The surface compositions of reaction-aged catalysts were examined using AES,and there were observed significant
differences in the surface compositions of catalysts depending upon whether they were in the upper pseudo steady state or lower pseudo steady state or sulfur-poisoned state. Although a detailed analysis of these results has appeared in earlier papers (Agrawal et al., 1981a,b),a brief discussion is presented here. Examination by AES of catalyst removed from the reactor while it was still in the upper pseudo steady state showed mainly Co on the surface with very small amounts of carbon (less than 10-20% of a monolayer). For Co catalysts deactivated to the lower pseudo steady state, AES studies show an almost complete absence of Co on the surface with as little as 1% present in some; there is extensive coverage with graphitic carbon and bulk carburization to a depth of several hundred angstroms. We believe that slow catalyst deactivation in the upper pseudo steady state was due to the buildup of carbon-containing species on the cobalt surface. However, these species were hydrogenated off the surface during the reactor shutdown. Hence, we term them as “reactive” carbon-containing species. The term “reactive” is appropriate here because a fully deactivated cobalt catalyst (lower pseudo steady state) showed large amounts of carbon indicating that this carbon was not “reactive” enough to be hydrogenated off the catalyst surface during an identical reactor shutdown. For sulfur-poisoned catalysts, AES studies showed that the loss in methanation activity resulted from two-dimensional surface sulfide formation; no sulfur was present in the subsurface regions. While there was a 102-foldreduction in activity going from upper pseudo steady state to lower pseudo steady state, the selectivity to higher hydrocarbons was unchanged, suggesting the same precursor for CHI formation and for higher hydrocarbon formation. The methanation activation energy in the upper pseudo steady state is 28 kcal/mol; for both sulfur-poisoned and carbon-deactivated Co/A120, it is the same, 16 kcal/mol. This suggests that there is a common rate-determining step in methanation over sulfur-poisoned and carbon-deactivated C0/A1203and that it may be different from the one for methanation over a clean Co surface. Kinetic Modeling Table I1 summarizes the kinetic behavior of Co/A1203 in different regions of methanation activity in terms of a power series model to facilitate comparison with those of Vannice (1975). The data for fresh catalyst (upper pseudo steady state) are in good agreement; the kinetic behavior of carbon-deactivated and sulfur-poisoned Co/A1203has not been reported previously. Kinetics in the Upper Pseudo Steady State. The effect of CO concentration on the rate of methanation over Co/A1203was investigated at only 573 K because the period for which the upper pseudo steady state could be maintained at 673 K was too short (-5 h) to make good kinetic measurements. The concentration of H2 was kept
388
Ind. Eng. Chem. Fundam., Vol. 21, No. 4, 1982 0 I O 1
l l r 9 0 7 0 Xc0l2
Lnq
( i + 5 9 4 x c 0 12
( 1 + 02 X c o P
i
008
0 IO
+z
5
004
z
1
1
L
0 02
0 06
004
xco
0
002
001
Figure 5. Methanation rate as a function of CO mole fraction in the lower pseudo steady state at 673 K. The solid line represents the single-term rate equation, and the dashed line represents the twoterms rate equation.
004
003
xco
Figure 2. Methanation rate as a function of CO mole fraction in the upper pseudo steady state at 573 K. The solid line represents the power law rate equation, and the dashed line represents the twoterms rate equation.
0 573'K A673-K
-
Y
' 0
I
020
,
040
1
1
060
000
,
100
0
025
050
xHP
IO
xHP
Figure 3. Methanation rate as a function of Hz mole fraction in the upper pseudo steady state at 573 K, 1%CO, and balance He. The solid line represents the square-root dependence of the methanation rate on Hz mole fraction.
_--,------
5x
075
2~~
+-
Figure 6. Methanation rate as a function of Hz mole fraction in the lower pseudo steady state at 573 K and at 673 K. The solid line represents the square-root dependence of the methanation rate on Hz mole fraction; 1%CO, Hz, balance He. 20,
-------i---
0
I 366Xc0
I
,
I
,
0138Xc0
(1+277 Xco12
(1+%14Xc012
00351 Xco
Y
oi
0025
0 4 t{C -' H4'
"0
0050
0075
01100
Olh5
01150
Oli5
xco
Figure 4. Methanation rate as a function of CO mole fraction in the lower pseudo steady state at 573 K. The solid line represents the single-term rate equation, and the dashed line represents the twoterms rate equation.
essentially constant (96% to loo%), while that of CO was varied between 170 ppm and 4%. Increasing CO concentration inhibits methanation in this regime (Figure 2). In another series of runs with 1% CO, the hydrogen concentration was varied between 25 and 99% by dilution with high-purity He. The results (Figure 3) show that methanation rate is proportional to the square root of the H2 mole fraction. If we make the reasonable assumption that the concentration dependence at 673 K is the same as that at 573 K, the upper pseudo-steady-state data can be represented by the expression NCHloc e[(-2S*z)x1031/RTxCo-00.24f0.2xH20.5 (1) Lower Pseudo Steady State. The effect of CO concentration on the rate of methanation over Co/A1203 in
( lC393XCO1*
002
004
006
008
010
xco
Figure 7. Rate of methanation over sulfur-poisoned Co/AlZO3as a function of CO mole fraction (Xco).Reaction conditions 663 K, 15 ppb H2S in CO and H2. The measurements were made after steady-state methanation activity had been achieved for Co/A1203 at 663 K with 1% CO and 15 ppb of HzS in H2 The solid line represents the fit of single-term rate equation.
the lower pseudo steady state at 573 and 673 K is shown in Figures 4 and 5. Catalyst performance is qualitatively similar at both temperatures with the rate increasing with CO concentration, in marked contrast with the inhibition by CO observed in the upper pseudo steady state. A square root dependence was again observed for the effect of H2concentration on methanation rate in 1% CO at both 573 and 673 K (Figure 6). Sulfur-Poisoned Catalyst. The kinetic behavior of catalyst poisoned with 15 f 1ppb of H2S resembles that of catalyst severely deactivated by carbon. The methanation rate increased as the mole fraction of CO in the feed increased to 10% (Figure 7). Variations in the concentration of H2 which constituted the balance of the feed
Ind. Eng. Chem. Fundam., Vol. 21, No. 4, 1982
were small enough to have little effect on the methanation rate.
Discussion As Co/A1203 undergoes carbon deactivation, the methanation rate dependence on XHzand the selectivity to higher hydrocarbons do not change significantly (Agrawal et al., 1981a). This suggests that the reaction mechanism has not changed with catalyst deactivation since a change in mechanism would be expected to alter the product selectivity significantly. On the other hand, a change in the activation energy from 28 to 16 kcal for both the sulfided and the carbon-deactivated catalysts does indicate a change in the electronic structure of the active sites. A survey of the rate expressions derived by considering the “carbide” theory and the “Enol Complex” theory (Table I) indicates that only those rate expressions in which the surface reaction is rate controlling are able to predict both a positive and a negative dependence on Xco, along with a positive dependence on XH2.Both mechanisms can predict a square root dependence of the methanation rate on XH . However, the observed severe carbon deactivation o! Co/Alz03which is confirmed by AES suggests that CO hydrogenation proceeds at least partially via a surface carbon intermediate, in agreement with earlier reports (Araki and Ponec, 1976; Joyner and Roberts, 1974; Wentrcek et al., 1976). The effect of hydrogen (Figures 3 and 6) and CO concentration on methanation rate is correlated by a simplified form of the rate expression based on “carbide” theory with the surface reaction between Cad and Had as the rate-determining step
Solid lines in Figures 4 and 5 show the nonlinear leastsquare fit of this model to the data obtained in the lower pseudo steady state. While eq 2 qualitatively predicts the methanation rate dependence on CO concentration, a quantitative comparison between the model and the data shows a less than satisfactory agreement. In the upper pseudo steady state (Figure 2) the agreement between the model (eq 2) and the data was not at all satisfactory and is not shown. On the other hand, it was possible in all cases (Figures 2, 4, and 5) to fit the kinetic data using eq 2 in either one of the two extreme ranges of CO concentration. Agreement between the data and the model is improved if it is assumed that the catalyst surface consists of two types of sites active for methanation, with each type having different heats of adosorption for CO and Hz.This gives the rate expression
Figure 2 shows the fit of this model to the methanation data in the upper pseudo steady state. The constants (eq 4) for data in the upper pseudo steady state were evaluated by nonlinear estimation
To visualize physically the concept of two types of active sites for methanation, one should note that constants in the denominator of the rate expression (eq 4) differ about two orders of magnitude. Each constant is a combination of several parameters including a CO adsorption equilib-
389
rium constant. A 5 kcal/mol change in the heat of CO adsorption can change the CO adsorption equilibrium constant (Kco)by a factor of 100 with a comparable change in the value of the constant in the denominator of the rate expression. Thus,the first type of surface sites have a high heat of CO adsorption and yield a maximum rate of methanation at very low CO concentrations; the second type of sites have a lower heat of CO adsorption which shifts the CO concentration required to achieve the maximum rate of methanation to a much higher value. Instead of using the concept of a heterogeneous surface with two types of sites, each having a different heat of CO adsorption, one may consider a continuous decrease in the heat of CO adsorption with increasing coverage of the catalyst surface by CO. Adsorption studies on single crystal surfaces of metals have suggested such a possibility. A polycrystalline surface, such as the one used in the present study, is likely to have a distribution of several single crystal faces, each of which may undergo changes in the surface structure induced by carbon deactivation. While we accept this argument in principle, we do not believe that it can result in any practical improvement over the correlation presented here. Similar arguments can be made for the methanation behavior of C0/Alz03in the lower pseudo steady state. Use of a two-term rate expression, shown as a dashed line in Figures 4 and 5, does improve the fit to the data at low values of Xco. Actually the agreement between the data and the two-term model is much better for the lower pseudo steady state than for the upper pseudo steady state. In going from the upper pseudo steady state to the lower pseudo steady state, the number of sites highly active for methanation appears to be substantially reduced. The kinetic behavior is then dominated by the second type of active sites. While there was need for assuming two types of active sites for methanation over Co/Al,03 under sulfur-free conditions, a single rate expression is quite adequate for sulfur-poisoned catalyst (Figure 7). Even at very low CO concentrations the agreement between the kinetic model and the data is excellent. This suggests that the active sites with a higher heat of CO adsorption are completely poisoned by sulfur. This is not surprising since it has been reported (Demuth et al., 1974; Legg et al., 1977) that sulfur preferentially adsorbs on high coordination sites on the catalyst surface; the same high coordination sites are also considered to be the first type of active sites for methanation, on which CO adsorbs with bridge-bonded structure with a high heat of CO adsorption (Rewick and Wise, 1978). Additional support for using eq 2 to correlate the data comes from poisoning studies (Agrawal et al., 1981b) where the rate data obtained during in situ poisoning showed that the rate of methanation is proportional to the square of the fraction of unpoisoned sites. This indicates that the rate-controlling step in methanation requires two adjacent surface sites. In spite of the limitations of eq 2, it provides a useful comparison of the different catalyst operating regimes. Table I11 compares the numerical values of the constants for the upper pseudo steady state, the lower pseudo steady state, and the sulfur-poisoned catalyst. The constant in the denominator of the rate expression (eq 2) decreases as the catalyst deactivates; it is further reduced by sulfur poisoning. This results from a decrease in the value of the CO adsorption equilibrium constant or the heat of CO adsorption as has been reported (Bonze1 and Ku, 1973; Garland, 1959; Kishi and Roberts, 1975; Sacco, 1977;
390
Ind. Eng. Chem. Fundam., Vol. 21, No. 4, 1982
Table 111. Rate-Expressions for Co/A1,0, Upper Pseudo-Steady State
Lower Pseudo-Steady State
Weller, 1947). The observed temperature dependence of the constant in the denominator is unexpectedly low. The value of Kco would be expected to decrease more than 10-fold with an increase in the temperature from 573 to 673 K. Since the constant in the denominator of eq 3 includes not only the CO adsorption equilibrium constant, but terms which are dependent on CO and Hz surface coverage and as well as other rate constants (Agrawal, 1979), it seems probable that the effect of temperature is diminished by the lumping process. Our data do not permit more than speculation on this point. There are no literature data available on the kinetics of methanation over Co/A1203 for comparison with the present study. Some observations made for methanation over Ni/A1203,however, are in agreement with those made for Co/A1203. The inhibiting effect of CO concentration on the rate of methanation over Ni/A1203 has been reported (Fitzharris et al., 1982; Van Herwijnan et al., 1970; Fontaine, 1973). Schoubye (1969,1970) observed a square root dependence on H, concentration for the rate of methanation over Ni supporting dissociative chemisorption of H2 on Ni. Until recently, few data have been obtained on the intrinsic kinetics of methanation because of the presence of heat- and mass-transfer limitations in the experiments. The rate equations obtained from data-fitting differ widely in their basic nature as well as their implications. In most cases, the range of experimental variables is too small to lead to a statistically sound rate equation. The objective of fitting the data by a power-law model is merely to represent them in the conventional form which have appeared in the literature and to show that our data are typical of those reported in the literature. However, such power-law models shed no light on the reaction mechanism itself. A better understanding of the reaction mechanism and the catalyst behavior (such as two types of reaction sites) can be inferred if one attempts to fit the data using a model based on a postulated reaction mechanism which is also supported by other physicochemical experiments.
Summary There is sufficient recent evidence in the literature to suggest that the adsorption-desorption characteristics of the methanation reactants are significantly altered by carbon deactivation or sulfur poisoning. Such changes are expected to change the observed kinetic behavior as a result of a change in reaction mechanism, rate-determining step, or electronic nature of the sites active in catalyzing the reaction. However, essentially no reaction studies have been made to investigate such possibilities. This study has supplied such data. This work presents both reaction-kinetic and surfaceanalysis evidence that methanation on Co proceeds via a surface carbon or carbide intermediate. Some researchers (Farrauto, 1976; Vlasenko and Yuzelfovich, 1969) find carbide to be unacceptable as an intermediate in methanation on the grounds that the rate of methanation exceeds that of carbide formation. However, this viewpoint overlooks two factors: (i) formation of surface carbon does not necessarily follow the same kinetics as those followed by metal carbiding, and (ii) H2in the methanation process removes surface carbon as methane (Araki and Ponec, 1976; Wentrcek et al., 1976), thus making more metal surface area available for CO chemisorption. Our data further suggest that formation of hydrocarbons from enolic species (Eidus, 1943;Joyner, 1975; Hall et al., 1957) may be a secondary process. Literature Cited Agrawai, P. K. Ph.D. Thesis, University of Delaware, Newark, DE, 1979. Agrawal, P. K.; Katzer, J. R.; Manogue, W. H. J . Catal. 1961, 69, 312. Agrawai, P. K.; Katzer, J. R.; Manogue, W. H. J . Catal. 1981, 6 9 , 327. Araki, M.; Ponec, V. J . Catal. 1976, 4 4 , 439. Bonzei, H. P.; Ku, R. J . Chem. Phys. 1973, 58(10), 4617. Demuth, J. E.; Jepsen, D. W.; Marcus, P. M. Phys. Rev. Left. 1974, 32(21), 1182. Eidus, Y. T. Izv. Akad. Nauk SSSR-ad. Khim Nauk 1943, 65. Farrauto, R. J. J . Cafal. 1976, 4 7 , 482. Fitzharris, W. D.; Katzer, J. R. Ind. Eng. Chem. Fundam. 1976, 77, 130. Fitzharris, W. D. Ph.D. Thesis, University of Delaware, Newark, DE 1978. Fitzharrls, W. D.; Katzer, J. R.; Manogue, W. H. J . Cafal. 1982. Fontaine, R. Ph.D. Thesis, Corneii University, Ithaca. NY, 1973. Garland, C. W. J . Phys. Chem. 1959, 63, 1423. Hail, W. K.; Kokes, R . J.; Emmett, P. H. J . Am. Chem. SOC. 1957, 79, 2983. Joyner, R. W.; Roberts, M. W. Chem. Phys. Left. 1974, 29, 447. Joyner, R . W. Faraday Discuss. Chem. SOC. 1975, 60, 172. Kishi, K.; Roberts, M. W. J . Chem. SOC. Faraday Trans. 7 1975, 7 7 , 1715. Legg, K. O., Jona, F.; Jepsen, D. W.; Marcus, P. M. Surf. Sci. 1977, 66, 25. Miiis, G. A.; Steffgen, F. W. Cafal. Rev. 1973. 8 , 159. Rewick, R. T., Wise, H. J . Phys. Chem. 1978, 82, 751. Sacco. A,, Jr. Ph.D. Thesis, Massachusetts Institute of Technoioqy, -. Cambridge, MA, 1977. Schoubye, P. J . Cafal. 1969, 7 4 , 238; 1970, 18, 118. Van Herwijnan, T., Van Doesburg, H.; DeJong, W. A. J . Catal. 1970, 2 8 , 39 1
V a i i c e , M. A. J . Cafal. 1975, 3 7 , 449. Viasenko, V. M.; Yuzeifovich, G. E. Russ. Chem. Rev. 1969, 38(9), 728. Weiler, S. J . Am. Chem. SOC. 1947, 69. 2432. Wentrcek. P. R.; Wood, B. J.; Wise, H. J . Cafal. 1978, 4 3 , 363
Received for review June 19, 1981 Revised manuscript received July 7, 1982 Accepted August 12, 1982 Support by the Division of Basic Energy Sciences, U S . Department of Energy, under Contract No. E (11-1)-2579is gratefully acknowledged.
