51
Ind. Eng. Chem. Prod. Res. Dev. 1084, 23, 51-56
Comparison of Activity and Selectivity Maintenance for Supported Fe and FeCo Fischer-Tropsch Catalysts John B. Butt' and Lyle H. Schwartzt Ipatleff Laboratory and Depatiment of Chemical Engineering, Northwestern University, Evanston, Illlnois 6020 1
Manfred Baernr and Relner Malessa Lehrstuhl fur Technlsche Chemle, Ruhr-Unlversitiit Bochum, 04630 Bochum, West Germany
The activity and selectivity of Fe/Si02and FeCo/SiO, catalysts, in typical formulation, have been compared for Fischer-Tropsch catalysis at 250 OC and a total pressure of 13.5 atm, with a feed composition of 3:l H,:CO, for moderately extended periods of operation. With a purified feed, avoiding carbonyls, it was found that the iron-cobalt alloy catalyst did not deactivate over a period of 200 h. On the contrary, shift activity and total CO conversion increased moderately over this time. The Fe/Si02exhibited an Initial deactivation that was stabilized at approximately 70 % of initial activity after about 60 h of operation. The olefin yield in this case increased by almost a factor of 2 during the period of deactivation, and then was stable. Oxygenate yields were also different for the two catalysts, and differences in the selectivity patterns are discussed in terms of possible influence of the formation of carbidii
phases.
Introduction We have previously reported data on the activity and selectivity properties of a number of silica-supported iron alloy catalysts, including the effect of pressure (Arcuri et al., 1983). Among these, an FeCo material with a 4:l atomic ratio of the two metals, respectively, was identified as having desirable properties of better olefin selectivity and higher water-gas shift activity than supported Fe. Prior comparisons among these alloy catalysts have been conducted within what might be called the "short-term" steady state, for run periods on the order of 10-12 h, but more extensive studies of longer term behavior have not been carried out. The present paper provides a comparison of Fe/Si02 and FeCo/SiOz in terms of activity and selectivity for operation under synthesis conditions for extended (to 200 h) time-on-stream. Experimental Section The activity and selectivity measurements were conducted in a flow reactor system at 250 OC and a total pressure of 13.5 atm. Isothermal conditions (&l/OC) were maintained by immersing the small reactor (12 mm diameter X 70 mm length) in a fluidized sand bath. A schematic of the experimental system is given in Figure 1,and details of the catalysts and operating conditions are summarized in Table I. Prior to the start of a run the Catalysts were preconditioned in situ by heating in an inert atmosphere (Ar, 450 OC, 1h) followed by reduction (H2,450 "C, 24 h) in accord with the previously described procedure (Arcuri et al., 1983). For purposes of optimizing product composition analysis, the runs were carried out at a nominal conversion (of CO) of ca. 8-10%; the space velocity was adjusted to compensate for catalyst deactivation when necessary to maintain conversion. Product analysis via gas chromatography was carried out through all isomers of C4 paraffii and olefins, C6+and COz,CO, Hz, and Ar (Jacobs, 1981). Prior to gas sampling, oxygenated products and
Table I. Catalysts and Operating Conditions catalyst weight loading, g percent exposed calcination reduction synthesis
feed: (vol %) H2/CO feed materials
Fe/SiO FeCo/SiO 3.85(Fe ) ; 1.02(Co ) 4.94 4.7 t 0.2 5.6 t 0.3 450 'C, 1 h, 450 'C, 1 h, 1130 h - ' S V 1130 h-' S V 450 'C, 24 h , 450 'C, 24 h, 1130 h - ' 1130 h-' 250 'C, 95.2 h, 250 OC, 196.8 h, 1650 h - l ; 197.5 h 1140 h - l ; 13.5 990 h-l ; 13.5 atm atm CO = 17.56 t 0.2 CO = 17.46 i 0.25 H, = 63.85 r 0.4 H, = 64.07 i 0.3 Ar = 18.59 t 0.2 Ar = 18.47 0.3 3.67 ?r 0.18 3.63 r 0.16 Ar, 99.997%; H,, 99.999%; CO/H,/Ar, 99.93% (further purified by passage through 3A sieve trap)
higher hydrocarbons were condensed in two traps, kept at 100 and 15 "C, respectively, and then analyzed separately. While a large number of oxygenated products were identified, the major components were C1-CI alcohols. We have not attempted in the present study to interpret oxygenate production in detail, since the condensate material was allowed to accumulate and was analyzed only twice during the course of a run. The resulting compositions, thus, represent integral average values over the sampling period. Product gas sampling and analysis, on the other hand, was conducted at approximately 3-h intervals. In what follows the CO conversion is determined as vol % CO (exit) X C O = 1X f(vo1ume contraction) vol % CO (inlet) where f(vo1ume contraction) =
Department of Materials Science and Engineering, Northwestern University. 0196-4321/64/1223-0051$01.50/0
vol % Ar (inlet) vol % Ar (exit)
(2)
Individual component selectivities are defined as C atom 0
1984 American Chemical Society
Ind. Eng. Chem. Prod. Res. Dev., Vol. 23, No. 1, 1984
52 H,IAr
Fli
v1
v3
'
percent related to the percent CO converted to hydrocarbons
v5
yield C 0 2 scoz =
i
8
I
I EXll
Gas
T
1
Figure 1. Experimental reactor system: V1 to V15, valves; T1 to T3, traps; FIC, flow indicator and controller; TIC,temperature indicator and controller; TI, temperature indicator; PI, pressure indicator; R, reactor; B, fluidized bed; H, heating.
IO
shydrocarbons
--
xco yield hydrocarbon XCO(1 - sco,)
(3) (4)
Results and Discussion Activity and Olefin Selectivity. The basic information concerning activity maintenance is given in Figures 2 and 3 for Fe/Si02 and FeCo/Si02, respectively. For the supported iron catalyst at an initial space velocity of 1650 h-' the CO conversion activity increases over an initial period of 8-10 h and then subsequently decreases monotonically for a period up to about 60 h. The initial increase in activity is probably associated with the formation of a carbide phase from the initially reduced iron (Amelse et
I.
1"
I -
O0
20
40
0
80
111,
m
Is0
la0
1.14
-, Figure 2. Carbon monoxide conversion and ethylene selectivity vs. time-on-stream for Fe/SiO,.
0
PO
$0
60
a0
KK)
I20
140
Experimental conditions as in Table I.
IW
moo 4
196.8
time, L
Figure 3. Carbon monoxide conversion and ethylene selectivity vs. time-on-stream for FeCo/Si02. Experimental conditions as in Table I.
Ind. Eng. Chem. Prod. Res. Dev., Vol. 23, No. 1, 1984 53
*.
*
*
0 .
SV
0.40 0.80
I650
SV (Fa/Si%)
990
t
0
20
40
60
80
timr ,h
m
Ho
120
I60
Figure 4. Comparison of chain growth parameters (upper) and low molecular weight olefin/paraffin ratios: (0) Fe/Si02; ( 0 )FeCo/SiO,.
al., 1981; Raupp and Delgass, 1979), while the decline in activity can be attributed to the formation of an inactive carbonaceous overlayer, possibly graphitic in nature (Dwyer and Somorjai, 1978; Niemantaverdriet et al., 1980, Ponec and van Barneveld, 1979). Note that the CO conversion activity is essentially constant from 60 to 100 h at S V of 1650 h-l. Decreasing the S V to 990 h-' at this point restores the overall conversion to ca. 8%, and there is no apparent further decline in activity up to the end of the run at 197.5 h. (Results are plotted only up to 170 h, as there are no changes beyond this time. Final data points are also indicated on the plots.) The ethylene selectivity for Fe/Si02 is also shown in Figure 2, and it is seen that this quantity, as defined in eq 4, increases during the period of decreasing activity by a factor of about 2 and then becomes stabilized at about the same point that CO conversion activity is stabilized. Similar trends in selectivity were noted for propylene, by about the same factor, while total C4olefins increased only slightly. The net loss in activity for this catalyst up to stabilization was approximately 30%. Stabilization of the activity can be viewed as the establishment of a quasi-steady state with respect to the formation and hydrogenation of coke precursors on the catalyst surface. The behavior of FeCo/Si02, shown in Figure 3, is markedly different. There was, in fact, no discernable loss in CO conversion activity (indeed, there may be a small increase) while the S V was maintained constant at 1140 h-' throughout the run to 196.8 h. Olefin selectivities, C2 through C4, were constant, while the initial olefin selectivities were essentially the same for Fe/Si02 and FeCo/Si02, as has been reported previously by Arcuri et al. (1983) and Amelse et al. (1981) for reaction at 250 OC, 14 atm, 3:l H2:C0 in a similar conversion range. The overall carbon balances (mass basis) averaged 98% for the runs with both catalysts. A comparison of net olefin yield between the two catalysts is interesting, since the stabilized Fe/Si02 has a higher selectivity but lower activity than FeCo/SiOz. This may be determined via rearrangement of eq 4 yc,= = (sc,=)(xco)(~ - sco,)
(5)
Table 11. Comparison of Ethylene Yields for Stabilized Fe/SiO, and FeCo/SiO, cat. Fe/SiO, FeCo/SiO, Fe /SiO, FeCo/SiO, a
time h
--
Xrna
80
8.1
80 160 160
9.0
6.8 9.8
s
=,
%
3.8 2.2 4.0 2.2
Sco , %bz
7.5 8.7 10.0 10.3
Adjusted to S V = 1140 h-' for Fe/SiO,.
YC' x lb3 2.8 1.8 2.4 1.9
Figure 6.
with the results shown in Table 11, assuming direct proportionality between space velocity and conversion, with no changes in selectivity. (The latter assumption we base on the longer term data for both Fe/Si02 and FeCo/Si02 shown in Figures 2 and 3 and the extensive data of Arcuri et al. (1983)J The result here demonstrates that Fe/Si02, equilibrated, is a better low molecular weight olefin producer in terms of overall yield (C3=trends accord) than FeCo/Si02, which differs from prior results based on initial conversion data reported by Arcuri et al. (1983) and Amelse et al. (1981). Olefin/Paraffin Selectivity Ratios. The general picture of olefin yield described above is amplified by further examination of the olefm/paraffin selectivity ratios for the two catalysts, as shown in the lower half of Figure 4. Again, the C2*/C2ratio is constant for the FeCo catalyst, reflective of its stability; thus paraffin and olefin yields are constant over the period of operation. However, C2'/C2 increases by a fador of b o a t 3 in the initial period of operation (SV = 1650 h-l) for Fe/Si02, and this increases even further at the lower S V at longer time-onstream. The change is much greater than the increase in olefin selectivity (Figure 2) and is indicative of a corresponding decrease in paraffin productivity upon catalyst aging. The fact that C2*/C2continues to increase after the point of apparent activity equilibration for Fe/Si02 indicates indirectly that coke formation is independent of the olefin-paraffin functions. the C2-/C2 will certainly become constant eventually; however, the observed increase leads to a conclusion that we would not wish to draw since olefins are commonly regarded as being favored coke
64
I d . Eng. Chem. Prod. Res. Dev., Vol. 23, No. 1, 1984 40 a
a
**.*
0 00
%+
36
a a
a
0 0
. . . . . . . .. . *
.
.
w
-
36-
0
.. . . '
0
1
.
0
I
I
#)
1
I
I
I
1
I
W tW,h
00
40
I
KK)
.
.
-w I
.
**
. . . *
I .
.
.
' I
.
1
I20
I
I
140
I
I60
Figure 6. Cs selectivities vs. time-on-stream: (0) Fe/Si02; ( 0 )FeCo/Si02. Table I11 cat.
operating time, h
Fe/SiO, Fe/SiO FeCo/SiO FeCo/SiO,
3.5-95.2 96.3-197.5 2.7-87.5 99.6-196.8
Fe/SiO, Fe/SiO, FeCo/SiO, FeCo/SiO,
3.5-95.2 96.3-197.5 2.7 -8 7.5 99.6-196.8
input c, g
CH,OH
C,H,OH
C,H,OH
C,H,OH
SV,h-l
a. Alcohol Products as g of Carbon Converted 75.0 49.7 48.6 55.7
0.254 0.207 0.194 0.359
0.210 0.181 0.207 0.379
b. Alcohol Yields a t Common 0.175 51.8 0.238 57.2 48.6 0.194 0.359 55.7
precursors compared to paraffins; further study is clearly necessary. However, the rate of increase of C2=/C2after the point of equilibration is less than that before, which may provide some comforting evidence for an increased rate of coke precursor hydrogenation. Growth Parameter and Higher Molecular Weight Products. The values of a reported here are based on the C3and C4hydrocarbon fractions. The absolute value may be slightly changed if the higher molecular weight fractions are included. On the top of Figure 4 are provided the experimental resulta on the chain growth parameter, a. While these are scattered, particularly for Fe/Si02, they can be considered, within the precision of the experimental data (and the sensitivity of a to such), to be constant over the period of experimentation. Further investigation of this leads to an examination of C5+selectivity and yields; these are shown in Figure 5. Again, there is scatter in the data for Fe/Si02 (although the scales are substantially expanded). An overall conclusion must be, however, that these long-term yields are not substantially affected by the deactivation of Fe/Si02, and for both catalysts differences in Cz-C4 selectivities seem to compensate for overall hydrocarbon selectivity as far as a is concerned as a parameter of measurement. The Cs+ selectivities do
0.067 0.053 0.064 0.102
0.032 0.026 0.026 0.042
1650 990 1140 1140
0.022 0.030 0.026 0.042
1140 1140 1140 1140
Space Velocity 0.145 0.208 0.207 0.379
0.046 0.060 0.064 0.102
appear to be lower over the longer term operation for FeCo/Si02; this is in accord with previous results for fresh catalysts reported by Jacobs (1981) and Amelse et al. (1981). Shift Activity. Shift activity, as represented by Sco2, increases with time-on-stream for both catalysts, as shown in Figure 6. For the FeCo/Si02 this invites speculation that shift-active sites differ from those for synthesis and the small overall increase in XCOnoted in Figure 3 is the result of increased shift activity. The similarity in the trends for the two catalysts suggests that this change in activity is associated with the iron function in the alloy catalyst, poasibly due to partial oxidation after long periods of exposure to product HzO. Unfortunately, in the present experimental configuration this could not be investigated. To our knowledge the stability of the mixed e' and x carbide phases (Amelse et al., 1981) formed on materials such as Fe/Si02 under synthesis conditions to water vapor for long periods of time has not been reported. This could be a fruitful area for study, since the most striking difference between Fe and FeCo is that the latter does not form carbidic phases. Alcohol Yields. As mentioned above, oxygenates were trapped out of the product stream and analyzed separately.
Ind. Eng. Chem. Prod. Res. Dev., Vol. 23, No. 1, 1984 55
3
19
IO
-
-
1 0
-
stop
8 -
-... ..
0
.
. s
0.6
04
0.4 0
0
il'
-A
1s.. 0
0'3 0.2
0. I
0
.:
Chongr SV
lh
b-.
I
I
c
I
1
...* . 0
.
.
I
-
/ L - = - -
,
0
.
0
0
-
I
-
.*
I
E
1
Figure 7. Alcohol selectivities in terms of weight C incorporation per total C (inlet): (A) Fe/Si02; 3.5-95.2h; (B)FeCo/Si02; 2.7-87.5 h; (C) Fe/SiOz; 96.3-197.5 h; (D)FeCo/SiOz; 99.6-196.8 h.
The major products were n-C1 to n-C4 alcohols and the basic yield data are given in Table IIIa. Alcohol selectivities have been shown by Arcuri et al. (1983) to be independent of conversion under the conditions of these experiments for the two catalystq hence we can adjust the yield data to a common space velocity for purposes of comparison, as shown in Table IIIb. Comparison of these values, expressed as (carbon appearing in alcohol/total carbon input), is given in Figure 7 for the two catalysts. In both cases the alcohol selectivity increases with timeon-stream; FeCo/Si02 is seen to have the higher overall selectivity at both short and longer run times. Somewhat unexpected is the higher selectivity for ethanol than for
.*
0
I
I
I
I
I
I
1
I
I
methanol of this catalyst. Conclusions The complex and interrelated sequence of events depicted in Figures 2-7 produces several surprises and our conclusions are to a certain extent questioning ones. First among these is the nondeactivation of FeCo/Si02 accompanied by increased shift activity and CO conversion with time-on-stream; this is to be compared to the extensive deactivation and subsequent stabilization of Fe/Si02, but also accompanied by increasing shift activity. Second is the increase in olefin selectivity for Fe/Si02 that ultimately turns it into a better olefin producer than FeCo/Si02 even at the lower CO conversions on the stabilized catalyst. In general it would appear that the FeCo/Si02 is a very stable catalyst that, in some sense (shift activity and alcohol yield) actually improves with age. To some extent this applies also to Fe/Si02, since the only deactivation, in the classical sense, noted in all of these observations was the decline in XCOover the first 60-80 h of operation. It is tempting to relate the increases in shift activity and alcohol selectivity to partial oxidation of the Fe via generated H20 in both catalysts, as has been reported by Niemantsverdriet et al. (1982) for Fe, but this would depend on the as yet unknown stability of carbidic phases to the reaction atmosphere over long periods of time. The specific nature of the surface of both Fe/Si02 and FeCo/Si02 after reaction is now under investigation via XPS, and hopefully we shall have in the future a more quantitative understanding of this. Overall, one can state that both catalysts exhibited much better activity and selectivity stability than would have been expected on the basis of the reports of Dwyer and Somorjai (1978), Niemantsverdriet et al. (1980),and Ponec and van Barneveld (1979). Acknowledgment This research was supported by the NATO Collaborative Research Grants program under Grant RG.117.81 and by the Exxon Education Foundation.
Ind. Eng. Chem. Prod. Res. Dev. 1984, 23, 56-62
58
Registry No. Fe, 7439-89-6; FeCo, 50954-72-8; co, 630-08-0; CH,OH, 67-56-1; C2H50H,64-17-5; C3H,0H, 71-23-8; C4H,0H, 71-36-3; ethylene, 74-85-1; propylene, 115-07-1.
Literature Cited Amelse, J. A.; Grynkewich. G.;Butt, J. B.; Schwartz, L. H.J . Phys. Chem. 1981, 85, 2484. Amelse, J. A.; Schwartz. L. H.; Butt. J. B. J . Catel. 1981, 72, 95. Arcuri, K. B. Ph.D. Dissertation, Department of Chemical Engineering, Northwestern University.
Arcuri, K. B.; Schwartz, L. H.;Piotrowski. R. D.; Butt, J. B. J . Catal., in press. Dwyer, D. J.; Somorjai, G. A. J . Catal. 1978, 52, 291. Jacobs, J. Ph.D. Dissertation, Ruhr-Universltiit Bochum, 1981. Niemantsverdriet, J. W.; van der Kraan. A. M.; van Dijk, W. L.; van der Baan. H. S. J . Phys. Chem. 1980, 84, 3363. Ponec, V.; van Barneveld, W. A. Ind. Eng. Chem. Prod. Res. Dev. 1979, 18, 268. Raupp, G, B,; Delgass, w, N, J , 1979, 58, 337,
Received for review May 6, 1983 Accepted September 12, 1983
Structure of the Cerium-Molybdenum-Tellurium Oxide Acrylonitrile Catalyst Jan C. J. Bart’ and Nicoia Glordanot Montedison Research Laboratories, Petrochemical Dlvlslon, 2002 1 Bollate, Mllano, Ita&
The structure of a microcystailine multicomponent molybdate catalyst for ammoxidation of propylene, consisting of 25 wt % (Ce, Mo, Te)O active phase supported on 75 wt % SO2, was studied by X-ray diffraction, TGA and DTA, electron microscopy, and chemical analysis, and conclusions are drawn (in part) also from studies of unsupported catalysts. The catalyst, prepared from Ce(N03),.6H20, (NH4)6M07024-4H20, HeTeOe, and microspheroklal silica, and subjected to calcination at 500 O C for 8 h, is a complex muttiphase system consisting essentially of a ternary (Ce,Mo,Te) oxide, a-Ce2M040,, and/or P-Ce2Mo,0,,, and does not contain (Te,Mo)O and (Te,Ce)O systems compounds or substantial amounts of free metal oxides. Evidence is available for surface enrichment in tellurium.
Introduction supported on a microspheroidal silica carrier. In particular, attention has been paid to a number of subsystems, both Multicomponent multiphase catalysts for (a”)oxidawith regard to the solid-state properties and catalytic action of olefins were introduced in about 1970 and mostly tivity in (amm)oxidation of propylene: (Te,Mo)O (Bart, consist of combinations of oxides of Ni, Co, Mg, Fe, Bi, Petrini, and Giordano, 1975a,b; Bart, Marzi, Pignataro, Te, Ce, Sb, P, and Mo supported on SiOz (Hucknall, 1974). Castellan, and Giordano, 1975; Bart, Petrini, Perissinoto, General structural aspecta concerning some of these comand Giordano, 1974; Bart and Giordano 1979,1980; Bart, mercial catalysts have been reported previously, for exCariati, and Sgamellotti, 1979; Bart, Van Truong and ample with regard to the systems (U,Sb)O (Aykan and Giordano, 1980); (Ce,Mo)O (Bart and Giordano, 1975, Sleight, 1970; Grasselli and Suresh, 19721, (Bi,P,Mo)O 1976, 1981/82; Castellan, Bart, Bossi, Perissinoto, and (Batist et al., 1974), (Bi,Sb,Mo)O (Ohdan et al., 1969), Giordano, 1976), (Te,Ce)O (Bart, Giordano, and Gianoglio, (Bi,Fe,Mo)O (Batist et al., 1971; Daniel and Keulks, 1973; 1981; Bart and Giordano, 1982), (Te,Si)O (Castellan et al., Notermann et al., 1975), (M,Mo,Te)O for M = Cp, Ni 1975),and (Mo,Si)O (Castellan, Bart, Vaghi, and Giordano, (Kozlowski and Sloczyfiski, 1976; Sloczyfiski and Sliwa, 1978),N ~ , C O $ ~ ~ B ~ P & . ~ M O (Praaada ~ ~ O ~ ~Rao , ~and / S ~ O ~1976; Giordano et al., 1977,1977; Ragaini et al., 1978; Vaghi et al., 1976). Menon, 1978) and (Mn,Mm,Bi,Mo)Ofor Mn = Ni, Co, Mg, From these preliminary studies it became suspect that Mn and Miu = Fe, Cr, Al, Ce (Wolfs, 1974; Wolfs and Te2MoO7 might be a principal catalytic phase in the Batist, 1974; Matsuura and Wolfs, 1975; Wolfs and Van multicomponent formulation, thus requiring an explanaHooff, 1975). tion for the synergistic action of cerium. On the other Development, properties, and performance of the inhand, cerium molybdate is another strong candidate for dustrial (Ce,Mo,Te)O ammoxidation catalyst have been catalytic action, in this case by enhanced action of telludescribed in recent work from this laboratory (Giordano rium. In this respect, we recall the ammoxidation activity and Caporali, 1969; Caporali et al., 1972; Caporali, 1972, of scheelite phases (Aykan et al., 1973,1974). In this paper 1977; Giordano et al., 1984). In an accompanying series we report various conditioning procedures together with of papers efforts have been made to defiie the nature and results derived from X-ray diffraction, thermal and properties of the phases in binary systems relevant to the chemical analysis, X-ray microanalysis, microchemical complex (Ce,Mo,Te)O formulation, which in its original attack, and XPS (Barr and Fries, 1982), which allow the form consists of a 20-35 wt % (Ce,Mo,Te)O active mass formulation of a structural model for the fresh ACN catalyst, sustained by experimental facts and in accordance with all available background information. The oxidative * Montedison “G. Donegani” Research Laboratories, Via G. and chemical states of the constituents of the fresh Fauser 4, 28100 Novara, Italy. (Ce,Mo,Te)O system are identified. Related studies of Cattedra di Chimica Industriale, Universitl di Messina, Via Salita S. Lucia Sopra Con39,98013 Pistunina-Messina, Italy. continuous use provide further evidence as to the chemical .. .
0 1984 American Chemical Society