Ind. Eng. Chem. Res. 1988, 27, 1339-1344
1339
KINETICS AND CATALYSIS Carbon Monoxide Hydrogenation Using Cobalt Manganese Oxide Catalysts: Initial Catalyst Optimization Studies Saul Colley, Richard G. Copperthwaite, Graham J. Hutchings,* and Mark van der Riet Catalysis Research Programme, Department of Chemistry, University of the Witwatersrand, P.O.Wits, Johannesburg 2050, South Africa
Cobalt manganese oxide matrix catalysts for CO hydrogenation have been studied, and the initial catalyst optimization studies are reported. A detailed structural characterization of both the bulk and the surface is also reported. Studies have shown that the optimum reduction temperature, based on thermogravimetric and reactivity studies, is ca. 400 "C. In addition, the temperature a t which the catalysts are initially stabilized with CO/H2 was also found to be important, and catalysts stabilized more rapidly at 220 "C and gave final catalysts with low methane yields and high conversion. The optimum Co/(Co Mn) ratio for unpromoted catalysts was determined to be in the range 0.15-0.25, and with these catalysts low methane selectivities (3-4% by mass) could be obtained with high C3 hydrocarbon selectivities at CO conversions of ca. 40% when operated at 220 "C. The notable feature of cobalt manganese matrix catalysts is that under most reaction conditions the catalysts give high propene yields together with low methane; in addition, it is demonstrated that the C3 hydrocarbon selectivity is linearly dependent on CO conversion.
+
The production of synthetic liquid fuels from coal remains an area of economic importance for those countries with limited or no petroleum. Of the possible routes available, i.e., direct conversion (coal liquefaction) or indirect conversion (coal gasification), the production of synthetic fuels and chemicals from CO hydrogenation has remained an intensly studied field, as the process has a long history of commercial operation. The current commercial catalyst formulation is an alkali-promoted iron formulation, and the product distributions and operating criteria have been well documented (Dry, 1981). This catalyst produces a broad range of products both in chemical functionality and carbon number. While the major products are typically 1-alkenes, a number of unwanted byproducts are also unfortunately produced. The major problems associated with the current product distribution are (a) the methane yield is higher than desired, (b) the product distribution is broad, and (c) the production of byproduct organic acids requires costly downstream treatment for their removal. It is clear that suitable alternative catalysts with improved selectivity must be identified to overcome these problems; however, such catalysts must retain the high catalytic activity associated with iron. The identification of more selective CO hydrogenation catalysts has been the subject of intense recent research in a number of centers (Biloen and Sachtler, 1981; Ponec, 1982; Hutchings, 1986). Control of product selectivity to obtain the desired products presents a severe technical problem since it is apparent the majority of published product data agree with a Schulz-Flory polymerization *To whom correspondence should be sent. Present address: Leverhulme Centre for Innovative Catalysis, Department of Inorganic, Physical & Industrial Chemistry, University of Liverpool, P.O. Box 147, Grove Street, Liverpool, Merseyside L69 3BX,
England.
0888-S88S/88/2627-~339$01.50/0
model (Friedel and Anderson, 1950; Henrici-Oliv6 and Oliv6, 1976) which predicts that only C1 products can be identified, and these effects have mainly been associated with strong interaction between the catalyst support and the active metal. Considerable interest has been shown in Fe/MnO matrix catalysts (Maiti et al., 1983, 1985; Lochner et al., 1986; Kugler, 1980; Jensen and Mossoth, 1985), and high selectivities to Cz-C4 alkenes have been observed with this system (Barrault, 1982; Barrault and Renard, 1984, 1985; Barrault et al., 1984; Kolbel and Tillmetz, 1979; Bussemeier et al., 1976). In general, for Fe/MnO catalysts, these high selectivities are only achieved under mild reaction conditions and are not stable at realistic CO conversions or reaction conditions (Cornils et al., 1984). Fe/MnO matrix catalysts have been shown to be comprised of solid solutions of Fe in MnO (Fe,Mn,-,O) and Mn in Fe304(MnyFe3,0,) (Maiti et al., 1983; Barrault and Renard, 1985),and recently it has been found that these solid solutions are unstable with respect to the formation of separate iron carbide and manganese oxide phases under reaction conditions (Hutchings and Boeyens, 1986; Copperthwaite et al., 1985). The formation of iron carbides causes a deterioration in the catalyst selectivity, and due to the propensity of iron to form carbides in the presence of carbon monoxide, we have concluded that iron is an unsuitable metal on which to base an improved catalyst design. Modified cobalt catalysts have also been well studied and indeed a Co/Th02 catalyst formulation was utilized as the first commercial catalyst. More recent studies have shown that modification of cobalt with a shape selective environment (Fraenkel and Gates, 1980) or use of specific reaction conditions (Blanchard et al., 1982) can give catalysts with improved selectivity, particularly to propene, but these systems were not demonstrated to be stable for any appreciable time scale. We have studied the modification of cobalt with manganese oxide, and we have previously shown (van der Riet et al., 0 1988 American Chemical Society
1340 Ind. Eng. Chem. Res., Vol. 27, No. 8, 1988
1986, 1987; Copperthwaite et al., 1987; that a Co/MnO catalyst with Co:Mn mole ratio of unity can give decreased methane yields together with enhanced propene formation. In this paper we describe our initial catalyst optimization studies on a laboratory microreactor scale. In particular, we discuss the effect of Co/Mn ratio on catalyst activity and selectivity.
Experimental Section Cobalt manganese oxide catalysts were prepared by coprecipitation from a solution of mixed nitrates with aqueous ammonia at 70 "C and pH 8.3. The precipitate was dried in vacuo (110 "C, 50 kPa, 16 h) and then calcined in air (500 "C, 24 h). The Co:Mn mole ratio in the prepared catalysts was controlled by selection of the appropriate initial mixed nitrate solution. Calcined catalyst precursors were then pelleted and sieved (0.5-1.0-mm particles) and loaded into a fixed bed stainless steel microreactor previously described (Copperthwaite et al., 1987). Catalysts were reduced in situ in the reactor with hydrogen (typical conditions: 400 "C, 200 h-l, 100 kPa, 16 h) prior to commencement of the CO hydrogenation studies. Catalysts were stabilized with CO/Hz (CO:H2 = 1:1 mole ratio) for ca. 100 h before data collection commenced. Catalysts were stable throughout the time scale of each experiment, and each result quoted represents an average of data collected for at least 120 h. Catalysts were characterized by surface area measurement by using the BET method, X-ray diffractometry using a specially constructed cell enabling in situ studies, and differential scanning calorimetry (DSC, Du Pont 9900). X-ray photoelectron spectroscopic (XPS) measurements were performed in a Vacuum Generators "Solar 300" UHV mbar (1mbar = 0.1 kPa). chamber, base pressure 4 X The chamber was equipped with a VG CLAM XPS/AES module incorporating a common 150" spherical sector electron analyzer. Catalyst samples were treated in a Leybold-Heraeus combined preparation chamber (VLP 10/63) and high-pressure cell (HPZ10) attached to the UHV chamber via a bellows-sealed gate valve. A sample transfer system was designed in order to move the catalyst back and forth between the horizontal sample treatment rod and the vertical UHV manipulator for surface analysis (Copperthwaite et al., 1988). Catalyst discs were prepared using a standard 12.5-mm-diameter press (10 tons). Surface analyses were generally performed within 35 min of mounting the sample. XPS data were taken using unmonochromatized A1 K a or Mg K a radiation at a constant pass energy of 50 eV. Photoelectron peak intensities were corrected for analyzer response to kinetic energy (i.e., KE') and for the variation in photoionization cross sections for the corresponding levels of the various elements (Scofield, 1976). Electron binding energies were calibrated with respect of the metallic cobalt level at 778.2 f 0.3 eV (Moyes and Roberts, 1977), and XPS data were collected in the C ls, 0 Is, Co 2p3j2,and Mn 2p3j2regions. Results and Discussion Bulk and Surface Structural Studies. A detailed study of the bulk structure of the Co:Mn = 1:l catalyst was carried out by using X-ray diffraction, thermogravimetric analysis, and differential scanning calorimetry. By use of X-ray diffractometry, the fresh uncalcined catalyst was found to be comprised of a range of cobalt manganese spinels (CozMn04,CoMnz04),in which the cobalt or manganese ions occupy exclusively either the tetrahedral or octahedral sites, as well as the mixed spinel (Co, Mn)(Co, M&04 in which the cobalt and manganese cations now occupy each site nonexclusively. Calcination of
n
20-
4
- 1.00
;I
:'I - 5
-
L o I
-5
.-200
100
200 I
300
400 O 500
Temperature
600
700 1 .2 0
(OC)
Figure 1. Differential scanning calorimetry and thermogravimetric analysis of a cobalt manganese oxide catalyst (Co:Mn = 1:l by mass) in H,.
this material at 500 "C increases the degree of substitution in the octahedral and tetrahedral sites and consequently increases the relative proportion of (Co, Mn) (Co, M&04 present. It is interesting to note that catalysts obtained from reducing noncalcined materials give higher alkene/ alkane ratios than catalysts prepared by using a calcination step. However, uncalcined catalysts gave very variable catalyst activities, a problem which is totally removed by inclusion of a calcination step. In view of this, it is possible that catalysts derived by reduction of Co,Mn04 or CoMnz04may be more selective for alkene formation and this feature requires further investigation. Reduction of the calcined material in Hz (400 "C, 16 h, GHSV = 200 h-l) results in a material that by X-ray diffraction contains Co metal, MnO, and cobalt/manganese spinels of which the mixed spinel (Co, Mn)(Co, Mn),04 is the dominant phase. Investigation of this reduction process using thermogravimetric analysis and differential scanning calorimetry analysis in a H2 atmosphere indicates that a four-stage process is involved with a total overall mass loss on reduction of ca. 20% (Figure 1). Two major mass losses at ca. 320 and 460 "C are assigned for reduction of cobalt and manganese oxides such that the reduced catalyst contains significant quantities of Co metal and MnO. Analysis of catalysts with different Co/Mn ratios showed very similar trends, and as the cobalt content of the catalyst was increased, the degree of substitution of Co in the octahedral sites also increased. A study of the surface composition of the Co:Mn = 1:l catalyst was also undertaken, using X-ray photoelectron spectroscopy. In this context it was of interest to attempt a correlation of the surface composition data with that pertaining to the bulk after calcination in air and after reaction with hydrogen and carbon monoxide under very similar conditions to those used in the microreactor studies. XPS analysis was performed, before any treatment, on the catalyst calcined in air at 500 "C and after the reactions depicted in Table I and corresponding quantitative data on surface Co:Mn ratios as a function of chemical treatment are also displayed. Clearly, reduction in hydrogen at progressively higher temperatures lowers the Co:Mn ratio until a limiting value of ca. 0.15 is attained under conditions comparable to the microreactor catalyst testing. The subsequent reaction of the reduced catalyst with carbon monoxide leads to only a small increase in the
Ind. Eng. Chem. Res., Vol. 27, No. 8, 1988 1341 Table I. S u r f a c e Treatment Sequence of Catalyst Co:Mn atomic ratio species (k0.03) present Mn3+, Co2+ A. surface analyzed prior to any 0.31 treatment B. reduction in Hz for 12 h at 533 K 0.42 Mn3+, Coo Mn3+, Coo C. reduction in H2 for 6 h a t 633 K 0.19 Mn3+, Coo D. reduction in H2 for 6 h at 673 K 0.15 Mn2+,Co2+ E. pure CO at 493 K for 25 min (flow 0.14 rate 20 cm3/min) F. pure CO at 553 K for 18 min (flow 0.19 Mn2+,Co2+ rate 20 cm3/min) 0.42 Mn2+, Coo G. Ar+ bombardment a t 8 kV for 10 min
Co:Mn ratio (ca. 30%), whereas a final argon ion bombardment (which removes the outer two to three atomic layers) again enhances the relative surface cobalt concentration. Further insight into the composition of the catalyst surface was obtained by an analysis of the measured Co 2p3/, and Mn 2p3/, photoelectron binding energies and the associated presence or absence of high-binding energy satellites. It is well established in the literature that the presence of so-called "shake-up" satellites in the Co 2p3,, region are indicative of Co2+ions (d', high spin configuration), whereas Co3+ions (d6,diamagnetic) do not display shake-up satellites (Frost et al., 1974). Again, Mn2+ions (d5, high spin) are expected to have shake-up satellites, whereas Mn3+ ions (d4) would not. Metallic cobalt and manganese do not possess shake-up satellites and, in any case, are easily distinguished by low-binding energies for their 2~~~~photoelectron signals compared to their ions. These variations in the behavior of MO, M2+,and M3+,with regard to XPS measurements, make it a relatively easy task to identify the chemical natures of the metal species produced at the catalyst surface during this investigation, and they are identified in Table I. From Table I it is clear that, under all reactivity conditions encountered, a depletion of cobalt in the surface layers is apparent, when compared to the bulk composition. Since the Co:Mn = 1:l catalyst displayed a most desirable activity and selectivity, it may be possible to develop formulations which could contain much lower cobalt loadings, possibly dispersed on an inert support. It is also apparent that cobalt enrichment under the influence of CO does not occur with this catalyst, in contrast to the Fe/MnO system (Copperthwaite et al., 1985), and this may, in turn, be related to the ease of formation of bulk carbide phases in the case of iron. The effect of argon ion bombardment (condition G in Table I) is to remove the outer layer of cobalt(2+) oxide and to reveal a near-surface region relatively enriched with metallic cobalt. Although we were unable to perform surface studies on the catalyst under synthesis conditions (i.e., CO and HJ, we are confident that the working catalyst is composed of substantial amounts of cobalt(2+) and manganese(2+) oxides, apart from metallic cobalt. Other interesting features, which may have an important part to play in the activity of this catalyst, include the resistance to reduction of Mn3+specifically. It seems likely that, under synthesis conditions, Mn3+remains the predominant manganese species in the catalyst and its stability could be due to the formation of mixed spinel phases, as determined by XRD (see previous section). Again, the reaction of carbon monoxide is observed to have opposing effects in this mixed catalyst; whereas Mn3+is reduced to Mn2+,the metallic cobalt (from hydrogen reduction) is oxidized at the surface (Betts et al., 1988). Such information suggests that electronic effects
Table 11. Effect of Different Reduction Temperatures on the Performance of a Co:Mn 1:l Catalyst d u r i n g CO Hydrogenationa reduction temp, OC 200 400 430 1 31 14 conversion, % time on line, h 79 47 88 290 310 269 GHSV,b h-' hydrocarbon selectivity, % by mass 100 6.1 6.6 CH, Tr 1.7 1.0 C2H4 0 1.0 0.5 CZH6 Tr 6.7 2.1 C3H6 0 1.1 0.6 C3H8 c4 o 8.8 2.7 c5 0 2.6 c6 0 10.0 CI 0 74.6' 4.2 C,+ 0 45.3 oxygenate selectivity, % by mass C,OH 0 0 0 CzOH 0 0 19.2 C30H 0 0 3.9 0 0 0 C,OH+ CH3CHO 0 0 0.9 COP selectivity, % v/v 0 0 0 "CO/H2 = 1:1,190 "C, 600 Wa. bMeasured at 101.3 kPa, 25 "C. Total C5+ hydrocarbons.
may have some influence on the mechanism of hydrocarbon formation on these catalysts. Effect of Reduction Temperature. To investigate the process occurring during catalyst reduction, thermogravimetric analysis of a cobalt manganese oxide catalyst (Co:Mn = 1:l mole ratio; calcined 500 "C, 24 h) was carried out in a hydrogen atmosphere. Mass losses were observed at ca. 230, 320, and 460 "C, indicating that thermal treatment during reduction occurred in a three-stage process. Three samples of a cobalt manganese oxide catalyst (Co:Mn = 1:l mole ratio; calcined 500 "C, 24 h) were then reduced in situ in the reactor at 200,400, and 430 "C (H2GHSV 200 h-l, 16 h). The catalysts were then studied for CO hydrogenation under identical conditions, and the results are shown in Table 11. It is clear that the catalyst reduced at 200 OC (corresponding to below the first TGA mass loss) was inactive compared to the catalyst reduced at 400 "C. The catalyst reduced at 430 "C (corresponding to a temperature on the onset of the third TGA mass loss) showed a steady decline in activity, and selectivity to oxygenates was markedly increased. In view of these results, all subsequent catalyst reductions were carried out at 400 "C. Effect of Catalyst Stabilization Temperature. Prior to data collection for all catalyst studies, catalyst were stabilized in CO/Hz for 100-200 h during which time the CO conversion and product selectivity attained values which were reasonably stable for the next 500-600 h catalyst testing. The temperature at which the catalysts are initially stabilized with CO/H, was found to be important for catalysts with low Co loadings (Co/(Co + Mn) < 0.5), which when stabilized at 220 "C, rather than 190 "C, more rapidly, giving final catalysts with low methane yields and high conversion (Table 111). Catalysts initially stabilized at the lower temperature of 190 "C required very long stabilization periods (>500 h). Consequently all subsequent catalyst stabilizations were carried out at an initial temperature of 220 "C. Effect of Co:Mn Ratio. Cobalt manganese oxide catalysts were prepared with a range of Co/Mn ratios, and CO hydrogenation was carried out for a number of conditions and full product distributions are given in Tables IV-VI. A t comparable reaction conditions, increasing the Co concentration increases the methane yield and de-
1342 Ind. Eng. Chem. Res., Vol. 27, No. 8, 1988 Table 111. Pretreatment Effect on CO Hydrogenation over a 25% Co/MnO Catalysta catalyst Ab catalyst Be reaction time,d h 253 375 GHSV,' h-I 240 292 CO conversion, % 39 25 hydrocarbon selectivity, o/s by mass CHI 3.5 7.4 C*H4 2.6 4.2 C2H6 1.6 2.5 C3H6 14.1 13.0 C3H8 2.9 2.3 c4 10.8 12.5 c5 11.1 9.6 c6 8.6 9.2 c, 7.6 8.8 32.9 28.8 C8+ alcohol selectivity, % by mass CH30H 1.5 0.03 CPH5OH 1.6 0.1 1-C30H 0 0.02 1-C40H 0 0.1 1-C50H 0.1 0.3 ~-C~OH 0.7 0.5 1-C70H 0.1 0.7 0.3 3.4 C8+ COz selectivity, 70v/v 0.34 0
.
15
-
5-
C
53c ?E
3,: ne ih)
on
Figure 2. Catalytic performance of a Co:Mn = 1:l catalyst: conversion as a function of time on line, 220 "C, GHSV = 303 h-l; 600 kPa, CO:Hp = 1:l by vol. 3:
"220 "C, 600 kPa. Brought on line at 220 "C. Brought on line a t 190 "C for 190 h and then increased to 220 "C. dCumulative run time. eMeasured a t 101.3 kPa, 25 "C.
creases the alkenelalkane ratio. In addition, it was observed that catalysts with high Co loadings (e.g., Co/(Co + Mn) = 0.75, 1.0) exhibited decreased stability and catalyst lifetimes. These effects were more pronounced when the catalysts were operated at high CO conversion (>50%) or at high reaction temperatures (e.g., 250 "C). For catalysts with lower Co loadings (Co/(Co + Mn) C 0.5), long lifetimes were observed for significant CO conversion levels, and in general the useful catalyst lifetime of these materials exceeded 700 h under these test conditions. The changes in catalyst activity/selectivity with time on stream are shown in Figures 2 and 3 for a typical experiment. In general catalysts, required 100-200 h to stabilize, during which time the conversion usually increased with increasing reaction time. Subsequently, catalyst activity slowly declines, indicating that structural changes are
35
-
Y
U
53'
1
500 Time on line
-~ li3i
(17)
Figure 3. Catalytic performance of a Co:Mn = 1:l catalyst; conditions as for Figure 1; selectivity as a function of time on line: (A) CH4, (n)Cz hydrocarbons, (0) C3 hydrocarbons.
possibly still occurring even at these extended reaction times. In conjunction with this slow activity decline, the methane and C2hydrocarbon selectivities increased at the expense of the C3 hydrocarbon selectivity. Low cobalt concentrations (Le., Co/(Co + Mn) = 0.15-0.25) give catalyst with low methane yields, high propene yields, and long lifetimes. However, these catalysts are not particularly active at 190 "C and must therefore be operated at a higher temperature (220 "C). Operation at 250 "C shifts the product distribution to
Table IV. CO Hydrogenation Using Co/MnO Catalysts at 190 "C" catalyst Co/(Co + Mn) 0.15 0.25 reaction time,bh 249 415 GHSV,' h-l 239 257 11 17 CO conversion, % hydrocarbon selectivity, % by mass CH4 7.8 2.2 C2H4 4.2 2.1 C2H6 1.3 0.4 C3H6 6.4 6.2 1.7 1.7 C3H8 9.1 5.1 c 4 7.4 5.2 c5 c6 14.7 8.4 c7 5.6 19.0 41.8 46.7 C8+ alcohol selectivity, % by mass CH30H 0 0.1 CZH5OH 0 0.4 1-C30H 0 Tr 1-CdOH 0 0 1-CbOH 0 0 1-CBOH 0 0.02 1-C70H 0 0.2 ct3+ 0 2.4 CO, selectivity, % v/v 0.72 0
0.35 249 220 9
0.50 421 272 24
0.75 415 250 44
1.00 410 277 29
5.7 2.2 0.8 4.2 1.2 5.7 4.8 9.6 7.6 58.2
5.4 2.2 2.2 8.3 2.9 7.9 6.9 10.1 13.2 36.3
9.0 0.9 3.9 9.6 6.4 11.2
15.5 0.7 2.4 3.7 2.3 5.6 5.6 7.9 7.5 46.9
0 0 0 0 0 0 0 0 0
0.5 1.1 0
"600-kPa reactor pressure. bCumulative run time. 'Measured at 101.3 kPa, 25 "C.
0
0 0.5 0.7 1.8 0
10.5
10.3 9.1 22.3 0.2 1.1 0.2 0.1 0.2 0.3 0.5 4.2 0.15
0.03 0.01 0 0
0.01 0.01
0.04 1.8 0
Ind. Eng. Chem. Res., Vol. 27, No. 8, 1988 1343 Table V. CO Hydrogenation Using Co/MnO Catalysts at 220 OC" catalyst Co/(Co + Mn) 0.15 0.25 reaction time,b h 418 253 GHSV,' h-' 217 240 CO conversion, % 36 39 hydrocarbon selectivity, % by mass CH4 5.2 3.5 3.2 2.6 C2H4 2.0 1.6 C2H6 13.0 14.1 C3H6 2.7 2.9 C3H8 11.3 10.8 c4 c5 9.7 11.1 c6 10.6 8.6 c7 9.1 7.6 C8+ 26.2 32.9 alcohol selectivity, 70 by mass CH30H 0.4 1.5 C2H50H 1.1 1.6 1-C30H 0.4 0 1-C40H 0.5 0 1-C50H 0.6 0.1 1-c60~ 0.6 0.7 1-C70H 0.6 0.1 C8+ 2.8 0.3 COP selectivity, % v/v 0 0.34
0.35 418 236 38
0.50 275 250 43
0.75 188 247 59
1.00 122 278 44
6.0 2.7 2.4 13.2 3.1 9.7 8.3 9.7 8.7 29.6
8.5 3.3 3.6 16.1 3.7 10.0
19.7 0.3 8.3 5.2 17.9 13.1 8.6 6.8 4.2 12.2
19.6 0.2 3.4 2.1 3.9 5.7 6.3 6.7 6.4 44.1
0.3 1.4 0.6 0.5 0.3 0.2 0.1 0.3 3.38
0.2 0.1 0.02 0 0.01 0.01 0.04 1.2 0
46.7d
0.4 1.2 0.5 0.7 0.7 0.6 0.5 2.0 0
0.2 1.7 0.7 0.5 5.0d 0.42
"600-kPa reactor pressure. *Cumulative run time. CMeasured at 101.3 kPa, 25 "C. dTotal C5+ products. Table VI. C O Hydrogenation Using Co/MnO 250 "C" catalyst Co/(Co Mn) 0.15 reaction time,* h 585 GHSV,' h-l 262 CO conversion. 9i 53 hydrocarbon selectivity, % by mass 7.9 CH4 0.8 C2H4 5.3 C2H6 15.1 C3H6 6.4 C3H8 12.3 c 4 9.3 CL 6.7 c6 3.9 c7 24.2 C8+ alcohol selectivity, % by mass CHzOH 0.4 1.6 CZHSOH 1.1 l-C*OH 1.2 1-C40H 0.9 1-C50H 0.7 1-CBOH 0.4 1-C70H 1.8 C8+ 8.7 COPselectivity, % v/v
+
-
Catalyst at
25
0.35 585 269 56
20 -
0.50 370 258 87
N
15
-
-
r?
u
16.6 0.3 7.9 9.0 12.1 10.6 8.0 6.1 4.0 18.2
13.2 0.1 6.5 4.4 15.5 9.8
0.4 1.3 1.0 1.5 1.2 0.6 0.5 0.7 9.9
0.04 0.3 0.5 1.8
" 600-kPa reactor pressure. Cumulative run time. at 101.3 kPa, 25 "C. dTotal C,+ product.
45.7d
Y
E 10 -
5 -
od
'
10
'
20"
30"
40"
50
60
"
Conversion X
2.2d 21.1 Measured
alkanes, particularly methane, indicating that at this elevated temperature the catalysts have a high hydrogenation activity and exhibit catalytic performance similar to a typical cobalt thorium catalyst (Pichler et al., 1967). Operation at 250 "C also effects a structural change in the catalyst since when the operating temperature is returned to 220 "C or even 190 "C the product distribution remains predominantly alkanes. Hence, the optimum operating temperature is considered to be ca. 220 "C. A t 220 "C, all the catalysts with a low Co loading (Co/(Co + Mn) < 0.5) gave product distributions best fitted to a Schulz-Flory plot with a = 0.77, whereas catalysts with Co/(Co + Mn) ratios of 0.75 and 1.0 fitted Schulz-Flory plots with a values of 0.57 and 0.85 respectively. This implies that the former group of catalysts (Co/(Co + Mn) = 0.15-0.5) operates with a similar chain propagation mechanism. The Schulz-Flory equation de-
Figure 4. Selectivity to C3hydrocarbons as a function of conversion; 0.35, (X) 0.5, (+) key to Co/(Co + Mn) ratios: (A)0.15, (m) 0.25, (0) 0.75; (0) unsupported Co catalysts.
scribed by Henrici-Oliv6 and OlivB (1976) was used for all calculations. The effect of reaction pressure and variation in reactant space time on product selectivity has been discussed previously for the cobalt manganese oxide catalyst with Co:Mn = 1.1mole ratio (Copperthwaite et al., 1987). In general, the findings for the previous study are also obtained for catalysts with different Co:Mn ratios, and increasing space time or increasing the reaction pressure increases the alkane selectivity of the catalysts. A notable feature of cobalt manganese oxide catalysts is that these formulations give high propene yields together with low methane and C2 yields under most reaction conditions. In general, C3 yields increase with increasing conversion whether this is achieved by increasing reaction pressure or temperature (Figure 4). While we have previously shown this for the Co/(Co + Mn) = 0.5 catalyst (van der Riet et al., 1987),this study shows this is a general feature for these catalysts. Co in the absence of MnO does not exhibit this trend, and this indicates that MnO modifies the Co catalyst so that a reaction pathway is promoted, whereby high yields of C3's are produced. In previous studies (van der Riet et al., 1987), we have shown using 13C isotopic studies that the high C3yield is the result of a reaction between a C1 and C2 surface species, and consequently C1 and C2yields are correspondingly low. The
1344
Ind. Eng. Chem. Res. 1988,27, 1344-1348
results of the present study demonstrate that this mechanism must be common to cobalt manganese oxide catalysts with Co/(Co + Mn) ratios < 0.5 since these formulations demonstrate the same 01 value for the Schulz-Flory distribution in addition to showing a linear dependence of C3production on CO conversion. The nature of the C1 and C2 reactive species is as yet not defined, but the operation of such a mechanism with cobalt manganese oxide catalyst explains why methane yields lower than those expected from a Schulz-Flory distribution are obtained. However, the production of high yields of C3 and C4 hydrocarbons compared to the current commercial catalysts is of significant technological and economic importance. For optimal unpromoted formulations, the C3and C4hydrocarbons comprise ca. 28 mass % with a methane selectivity of only 3.5%. The C3and C4 hydrocarbons contain >80% alkenes and hence would be a suitable feedstock either for oligomerization to produce diesel-range hydrocarbons, e.g., via the Mobil MOGD process, or for alkylation under acid conditions to produce gasoline-range hydrocarbons. Hence, use of a cobalt manganese oxide CO hydrogenation catalyst could provide a synfuel process with flexibility in the diesel/gasoline ratio produced.
Conclusions The results of this study indicate that cobalt manganese catalysts can give high activity, long-lived catalysts for CO hydrogenation with high propene and low C1 and C2yields. The optimized unpromoted catalyst formulation has a Co/(Co + Mn) ratio in the range 0.15-0.25 and must be operated at ca. 220 "C. While upper limits for the optimum conditions for reaction pressure and reactant space velocity have yet to be determined, the catalysts should be operated at >600 kPa and GHSV >300 h-l. Further studies are now required to determine if this catalyst performance can be enhanced by addition of suitable promoters, and these studies are now in progress in our laboratories. Acknowledgment We thank Mark Dry, Egbert Gritz, and Dirk Vermaire for useful discussions. We thank Sasol Technology, the University of the Witwatersrand, and the Foundation for Research Development for financial assistance. We also thank Peter Loggenberg and Mark Betts for obtaining the X-ray photoelectron spectra.
115-07-1; C3H3,74-98-6; CH,OH, 67-56-1; CZHSOH, 64-17-5; l-C,OH, 71-23-8; 1-C40H, 71-36-3; l-C,OH, 71-41-0; l-C60H, 111-27-3; l-CVOH, 111-70-6.
C3&,
Literature Cited Barrault, J. Metal Support and Metal Additive Effects in Catalysis; Imelik, B., Ed.; Elsevier: Amsterdam, 1982; p 225. Barrault, J.; Renard, C. French Patent 2 571 274, 1984. Barrault, J.; Renard, C. Appl. Catal. 1985, 14, 133. Barrault, J.; Renard, C.; Yu, L. T.; Gal, J. Proc. 8th Znt. Congr. Catal., Berlin 1984, 2, 101. Betts, M. J.; Copperthwaite, R. G.; Hutchings, G. J.; Loggenberg, P. M., submitted for publication in Surf. Sci. 1988. Biloen, P.; Sachter, W. M. H. Adu. Catal. 1981, 30, 165. Blanchard, M.; Vanhove, D.; Laine, R. M.; Petit, F.; Mortreux, A. J. Chem. SOC., Chem. Commun. 1982, 570. Bussemeier, B.; Frohning, C. D.; Horn, G.; Kluy, W. German Patent 2 518 964, 1976. Copperthwaite, R. G.; Hack, H.; Hutchings, G. J.; Sellschop, J. P. F. Surf. Sci. 1985, 164, L827. Copperthwaite, R. G.; Hutchings, G. J.; van der Riet, M. Znd. Eng. Chem. Res. 1987,26, 869. Copperthwite, R. G.; Derry, T. E.; Loggenberg, P. M.; Sellschop, J. P. F. S. Vacuum 1988, in press. Cornils, B.; Frohning, C. D.; Morow, K. Proc. 8th Int. Congr. Catal., Berlin 1984, 2, 23. Dry, M. E. Catalysis-Science and Technology; Springer Verlag: Berlin, 1981; Vol. 1, p 159. Fraenkel, D.; Gates, B. C. J . Am. Chem. SOC. 1980, 102, 2478. Friedel, R. A.; Anderson, R. B. J . Am. Chem. SOC.1950, 72, 1212, 2307. Frost, D. C.; McDowell, C. A.; Woolsey, I. S. Molec. Phys. 1974,27, 1473. Henrici-Oliv6, G.; Oliv6, S.; Angew. Chern., Znt. Ed. Engl. 1976, 15, 136. Hutchings, G. J. S. Afr. J. Chem. 1986, 39, 65. Hutchings, G. J.; Boeyens, J. C. A. J. Catal. 1986, 100, 507. Jensen, K. 5.;Mossoth, F. E. J . Catal. 1985, 92, 98, 109. Kolbel, H.; Tillmetz, T. US Patent 4 177 203, 1979. Kugler, E. L. Prep.-Am. Chern. SOC.,Diu. Pet. Chem. 1980, 564. Lochner, U.; Papp, H.; Baerns, M. Appl. Catal. 1986,23, 339. Maiti, G. C.; Melessa, R.; Baerns, M. Appl. Catal. 1983, 5, 151. Maiti, G. C.; Melessa, R.; Lochner, U.; Papp, H.; Baearns, M. Appl. Catal. 1985, 16, 215. Pichler, H.; Schulz, H.; Elstner, M. Brennst. Chem. 1967, 48, 78. Ponec, V. Catalysis-Specialist Periodical Report; Royal Society of Chemistry: London, 1982; Vol. 5, p 48. Scofield, J. H. J. Electron Spectrosc. 1976, 8, 129. Van der Riet, M.; Hutchings, G. J.; Copperthwaite, R. G. J . Chem. SOC.,Chem. Commun. 1986, 798. Van der Riet, M.; Hutchings, G. J.; Copperthwaite, R. G. J . Chem. SOC.,Faraday Trans. 1 1987, 83, 2963.
Registry No. Co/MnO, 51845-85-3; CO, 630-08-0; Co, 744048-4; MnO, 1344-43-0; CHI, 74-82-8; CzH4,74-85-1; C5Hs, 74-84-0;
Received for reuiew December 23, 1987 Accepted March 30, 1988
Oxidation of HS03- by O2 David Littlejohn, Ke-Yuan Hu,+and Shih-Ger Chang* Applied Science Division, Lawrence Berkeley Laboratory, University of California, Berkeley, California 94720
The oxidation of bisulfite ion by dissolved oxygen has been studied using a high-pressure flow system and Raman spectroscopy. We have detected an intermediate in the reaction which appears to be , formed from 2HS03- and lo2. We propose that the intermediate is disulfate ion, S 2 0 T 2 - which s-l hydrolyzes into SO2-, H f , and/or HSOd2-with a hydrolysis rate constant of 1.33 f 0.26 X a t 25 "C. An improved understanding of the kinetics of the oxidation of dissolved SO2 by oxygen in aqueous solutions would benefit flue gas desulfurization programs in which On leave from t h e Research Center for Eco-Environmental Sciences, Academia Sinica, Beijing, People's Republic of China.
0888-5885/8812627- 1344$01.50/0
control of the oxidation of SO2is desirable (Hudson, 1980). This oxidation reaction has been studied for many years, yet the oxidation mechanism is unclear (Alyea and Backstrom, 1929; Backstrom, 1934; Braga and Connick, 1982). Previous studies of this reaction have reported conflicting results because the reacton is very sensitive to 0 1988 American Chemical Society