Ch~'mlc ~t! f(n~ineer~¢tg 5tience, Vol. 5t, No. I I, pp. 2885-2890. 1996
Pergamon
Copyright© 1996ElsevierScienceLtd Printed in GreatBritain. All rightsreserved 0009-2509/96 $15.00 + 000 S0009-2509(96)00169-8
C O M P O U N D C A T A L Y S T FOR H I G H YIELDS OF OLEFINS FROM SYNTHESIS GAS Catalytic Reaction Steps H. de LASA, L. HAGEY, S. RONG and A. PEKEDIZ Chemical Reactor Engineering Centre. University of Western Ontario, l.ondon, Ontario, Canada Abstract - Compound catalysts involving methanol-synthesis/methanol- conversion functions offer promise for selective conversion of synthesis gas into C2-C3 hydrocarbons and particularly of C2-C 3 olefins. This approach was followed, by modifying a ZSM5 zeolite with the addition of P compounds. The catalyst was characterized using SEM-EDX, BET, XRD and TPD. To elucidate the reaction mechanism, experiments were conducted in a Betty well mixed reactor to establish the following: a) methanol is a key intermediate present in minute amounts, b) water gas shift reaction takes places under conditions of equilibrium, c) hydrogenation and condensation of light olefins into higher hydrocarbons is mild and allows for significant fractions of C2-C3 olefins to be produced.
INTRODUCTION Compound catalysts with the double function of methanol synthesis (MSC) and methanol conversion (MCC) can perform efficient transformation of synthesis gas into gasoline range hydrocarbons (Simard et al, 1991a). This process, developed in conjunction with advanced schemes for methane reforming (Jarosch, 1995), can help considerably in the economic conversion of natural gas. Performance of a compound catalyst such as the one described was already demonstrated in a pilot plant unit at CREC. This unit was based on the Pseudoadiabatic Operation reactor concept (de Lasa 1989, Simard et al., 1991b). It produced gasoline containing high fractions of aromatics and consequently high RON. However, given new environmental regulations, there exists an incentive to produce light olefins from natural gas. It was expected that high selectivity towards light olefins of a combined methanol synthesis (MSC) and methanol conversion (MCC) catalyst could be obtained by proper MCC modification with phosphorus compounds (Rong, 1994). CATALYST PREPARATION AND CHARACTERIZATION The MCC prepared was a ZSM-5 zeolite. It was synthesized according ot the method proposed by Gabelica et al. (19831) and Simard (1991). The protonated form HZSM-5 was then modified using PC13 in accordance with Butter et al. (1976). Different batches of "PZSM-5", each having a different phosphorus level, were prepared. The zeolites had a phosphorus loading in the 2 - 6 wt% range. The MSC used in this study was ZnO-Cr203. It was prepared by a co-precipitation method as detailed in Rong (1994). The goal was to achieve low Zn/Cr weight ratios (between 0/1 and 1/7) which would be indicative of zinc being highly dispersed in a chromium matrix. This MSC offers reasonable rates of methanol synthesis, and shows minimum olefin hydrogenation (Simard et al, 1995). SEM-EDX, BET, ESCA, TPD, DCP and TPR were adopted to characterize the catalyst prepared. Zn/Cr levels obtained were close to the ones expected: 0/1 to 1/7. Moreover, SEM-EDX was further employed to corroborate the above mentioned P levels. XRD was utilized to confirm both the ZnO-Cr203 structure as well as the structure of ZSM5 before and after P modification, by fnding the unit cell size through d-spacing of the diffracting planes. Regarding the effect of P on the zeolite structure it was observed that increasing P, the d-spacing and hence unit cell size remained unchanged. This suggests that P is not present as a constituent of the crystalline framework. There is no silicon or aluminum atoms substitution, since the unit cell dimensions of the zeolite remain unchanged when the P atoms are incorporated. It was further observed with XRD that while increasing the phosphorus content (0 to 3.3 wt%) there was a decrease in the relative intensities of the diffracted X-rays at d = 11.11 A, and d = 10.05 ,~ of the P modified zeolites (Table 1). This observation was consistent with results reported by Butter (1976) and was indicative that even if P was 2885
2886
H. DE LASAet aL
not incorporated in the crystalline framework, it was somewhat included in the zeolite structure. Table 1 Comparison of relative peak intensities for MCC with various phosphorus levels. Peak Position 20 (degrees) 7.98 8.8 23.11
d-spacing (A)
Relative Peak Intensity (I/Io) P=0.0 wt %
P=2.2 wt %
11.1 155 132 10.05 91 73 100 100 3.85 23.86 3.72 65 54 24.4 3.65 38 38 eference peak for the relati, e intensity a nalysls
P=3.3 wt % 90 61 100 57 37
BET was utilized for assessing specific surface areas of both the MSC as well as the PZSM5. It was observed that the Zn/Cr ratio and the calcination temperature are major factors influencing the MSC surface area. The reduced form of the ZnO-Cr203 has a substantially higher surface area than the non reduced form. All other factors being equal, the surface area of ZnO-Cr203 increased while the zinc oxide content decreased. For instance, a sample (Zn/Cr = 1/7) calcined at 773 K had a surface area of 11.5 m2/g. Its surface area increased to 12.7 m2/g for a Zn/Cr of 1/30 and to 18.2 m2/g when the Zn was reduced to zero. On this basis it was hypothesized that the main role of chromium oxide is to maintain a high dispersion of zinc oxide. When the calcination temperature increased from 623 K to 773 K the surface areas of the samples decreased. For instance, MSC with Zn/Cr = 1/20 had a surface area of 33.7 m2/g, 23.7 m2/g and 12.7 m2/g when calcined at 623 K, 673 K and 773 K respectively. Therefore, high temperature thermal treatment or uncontrolled temperatures during calcination can be detrimental for the MSC catalyst. Furthermore, while the MSC with a 1/7 Zn/Cr ratio had a 20.6 m2/g surface area before reduction, its surface area increased significantly to 118.0 m2/g following reduction. According to this, MSC catalyst reduction condition is extremely important for yielding a catalyst with a large surface area. This surface area is comparable with the 103 m2/g obtained with the Zn/Cr weight ratio = 1/6.5 under similar reduction conditions (Simard et al., 1995). Also, this result is consistent with the values obtained by Carruthers and Sing (1967) and Tronconi et a!.(1987). Consequently mild conditions, 623 K and 573-583 K in diluted hydrogen, were adopted for both calcination and reduction respectively of the MSC of the nine pellets of compound catalyst prepared. BET analysis for HZSM5 before P modification gave 366 m2/g while after P modification (P=3.77%) was 351 m2/g. Thus, results indicated that the modification by P has no significant influence on the surface areas of the zeolite catalyst. Ammonia TPD (Micrometrics TPD/TPR 2900 Analyzer) was used to evaluate acidity changes on the zeolite surface as a result of the P addition. TPD was controlled linearly from 373 K to 873 K with a ramping temperature of 30°C/min. Comparing TPDs, before and after modification, it was observed that the stronger acid sites, recorded at 731 K, were significantly reduced after modification with P and this was observed at two P levels considered : P = 3.77% and P = 5.25%. For P = 3.77% the highest TPD peak temperature appeared at 532 K while for P = 5.25% the highest TPD peak temperature developed at 492 K. This shows that the density of the stronger acid sites was decreased significantly by proper zeolite modification with P. Moreover, it was observed that with higher P levels there was a significant reduction on the total acidity strength and this was due to a lower density of the stronger acid sites. It is hypothesized that reduction of strong acid sites is of major importance for a catalyst yielding high fractions of light olefins.
EXPERIMENTS AND DISCUSSION OF RESULTS Nine pellets with different zinc/chromium ratios and phosphorus contents were considered as part of the experimental program (Table 2). These various catalysts were pelletized as follows: 63% of the MSC, 27% of MCC and about 10% of alumina binder. One hundred experimental runs were developed in a Betty reactor. Additional details about the absence of external mass transfer limitations for the pellets used in the experimental unit and the flow diagram
Compound catalyst for high yields of olefins
2887
of the experimental set up can be found in Simard et al.(1991a). Various operating conditions such as reactor temperature, total pressure, feed flow, cumulative time-on-stream were systematically changed during the program (Rong, 1994). These operating parameters were changed as follows: a) 7 reaction pressures between 3200 to 4560 kPa, b) 7 reaction temperature between 623 to 773 K, c) 7 feed flow between 425 to 783 cm3/min. Products were analyzed with FID and TCD. Estimations of overall mass balances were within 4% range. Table 2 Composition of the Compound Catalysts Catalyst Code A
B
C
D
E
F
G
H
1
Zn/Cr (wt) P (wt%)
1/20 2.31
1/7 2.20
0/1 4.57
1/20 3.77
1/7 2.65
0/1 5.95
1/20 5.25
1/7 4.3
0/1 3.24
It was observed that for the runs performed CO conversions were in the 10-70% range and that the addition of P was very effective in modifying reaction selectivities towards light hydrocarbons. Under optimum conditions C 2 selectivities were in the 45-60% range with selectivities for all hydrocarbons including C 1, C2 and C~ in the 64% range. As well ethylene and ethane were at similar yield levels to each other. One peculiarity of this compound catalyst is its catalytic activity change with time-on-stream. In this respect, Simard (1991a) recommended tests after 1500 min of time-on-stream, once the catalyst activity for methanol synthesis was stabilized. On the basis of the expected behaviour of the individual components of the compound catalysts it can be postulated that the conversion of synthesis gas proceeds via with the following reaction steps: CO + 2 H 2 2 CH3OH CH3-O-CH 3 CO + H2O CHz=CH 2 + H 2 CO + 3 H 2
---~ ---, "-~ ~ --"
CH3OH CH3-O-CH 3 + H20 CH2=CH 2 + H20 CO 2 + H z CH3-CH 3 CH 4 + H20
(1) (2) (3) (4) (5) (6)
While reviewing more carefully the reaction network it was observed that for all the conditions studied no methanol or di-methyl-ether were detected at the reactor outlet. This was consistent with observations of Simard et al.(1991a) and allowed to speculate that while methanol was formed, via Reaction 1 controlled by reaction equilibrium, methanol was readily converted to hydrocarbons via Reactions 2 and 3. Thus, in practice the reaction network was influenced by a mechanism drifting chemical equilibrium for methanol synthesis towards hydrocarbon formation and the reaction network could be simplified as follows: 2 CO + 4 H a CO + H20 CH2=CH 2 + H 2
--, -* --,
CH2=CH 2 + 2 H20 CO 2 + H 2 CH3-CH 3
(7) (8) (9)
Moreover, one significant aspect of the research was to establish that the water gas shift reaction (Reaction 8) took place, in the compound catalyst and for almost all conditions studied, very close to reaction equilibrium. With this end theoretical and experimental values of equilibrium constants were calculated. Detailed procedures for equilibrium constant calculations, which includes evaluation of fugacity coefficients and compressibility factors, are detailed in Rong (1994). To illustrate the trends typical results for two catalysts (Catalysts B and C) are reported in Figure 1. It was observed that the theoretical and experimental values were very close which allowed to hypothesize that reaction equilibrium was achieved. These trends were repeated for all temperatures (673-758 K) and for the nine catalysts studied. As well, from the same data it was possible to establish that water gas shift reaction (Reaction 8), given the high values of the equilibrium constants, was shifted towards the right and in practice the combined Reactions (7) and (8) were close to the following:
2888
H. DE LASAet 4C0 + 2H 2
al.
(lO)
CH2=CH 2 + 2 CO 2
12 --Theoretical 11
0006
0
Measured
(Pellet
B)
[]
Measured
(Pellet
C}
10
.=
~9
o<\D \o
3 E
0.004
~7
3. o
II/ //
I
0002-~ '/a
5 0000 4
660
I
i
I
i
680
700
720
740
-
0 000
760
780
-
, 0.002
0 004
0 006
CH4 Concentration(mo[/cm1)
Temperature (K)
Figure l Comparisons of theoretical and experimental equilibrium constants for Water-Gas Shift reaction (Pellets B and C)
Figure 2 Comparison of the amount of methane formed with that of ethane and ethylene. Temp. range: 674-743 K, CO conversion: 4-70 %
Experimental runs were also analyzed in terms of trends in productivity of various fractions with encouraging yields towards the desired C2-C3 light hydrocarbons. An interesting consideration was to assess the potential influence of ethylene hydrogenation reaction (Reaction 9) and to what extent ethylene hydrogenation approached reaction equilibrium. Theoretical chemical reaction equilibrium constants were evaluated for all pellets at various conditions studied. While theoretical values were in the 1000 - 14300 range, constants assessed with experimental data were in a 0.024-0.7 range. Thus, ethylene hydrogenation took place far from chemical reaction equilibrium and consequently, even if the compound catalysts cannot completely suppress ethylene hydrogenation, these catalysts do not strongly favour this reaction either. The fact that this reaction is very far from chemical equilibrium is important for a successful catalyst yielding high levels of light olefins from synthesis gas. Methanol formation described as Reaction 1 of the network can be viewed as a non-dissociative hydrogenation of carbon monoxide. Formation of methanol leads to the formation of di-methyl-ether and the formation of ethylene through dehydration. Competing with the non-dissociative hydrogenation of CO there is a dissociative hydrogenation of CO yielding methane. Regarding the dissociative CO hydrogenation (Reaction 6), this competitive step should be limited and even suppressed if possible. In the process of converting natural gas to hydrocarbons, methane synthesis is not a desired reaction. Methane is reformed first into synthesis gas, the reverse of Reaction 6, and synthesis gas is then converted into hydrocarbons. Large yields of methane makes this process non-viable. Therefore, the competition between dissociative and non-dissociative hydrogenation of CO and the ways of controlling the importance of the dissociative CO hydrogenation is a key issue while developing a compound catalyst. To clarify the relative importance of the non-dissociative hydrogenation of CO, yielding methanol first and then to C2, and the dissociative formation of CO yielding CH4, the molar concentrations of C2 and molar concentrations of CH 4 were compared. Graphs reporting these molar concentrations provided invariably very close to straight lines. Figure 2 illustrates this behaviour with an essentially constant [C2]/[Ct] relationship (Rong, 1994). Given that the ratios between molar concentrations in the Berty reactor remained unchanged and given as well that this was found consistently for all pellets, it can be argued that the kinetic models for both the dissociative and non-dissociative hydrogenation of CO are comparable and thus [C2]/[C l] becomes kl/k 2. [C2]/[C l] was also constant at different temperatures, and it is possible on this basis to speculate that these two competing CO hydrogenation reactions should have very similar energies of activation (i.e., E1-E2 is close to zero). Furthermore, it was noticed that the kJk 2
Compound catalyst for high yields of olefins
2889
relationship was strongly affected by P levels in the PZSM5 as well as somewhat by the MSC composition. Figure 3 shows that at higher P levels there is an increased catalyst ability to favour formation of C2 hydrocarbons. In summary, it was observed that P content raises C2 selectivity. P levels greater than 4 wt% were the ones yielding maximum C2 values. For instance, comparing three batches of compound catalyst with the same Zn/Cr = 1/7 and the three levels of phosphorous (2.2 wt%, 2.65 wt%, 4.3 wt%) the pellet with the higher level of P yielded the higher C2 selectivity: 47.2% for 4.3 wt% of P, 37.1 wt% for 2.65 wt% of P and 31.2% for 2.2 wt% of P. 80
~
-.
.
.
.
.
.
.
j I - 0 - C Selectivity, 15 0
.
.
.
.
.
.
.
.
.
.
70 I
G @
f
i ~ •
2 C
.
.
.
.
Pei'.et [,
Setectivit;', Ccmver.<~[on,
iell,,t }: [~dler :
[ ii
•
I 25 ~'
~> 504
100 ,
II 075
DO
,
0(30
II
40
"
- I
•
i
050 !
025
o
EO
2O
/-
J
•
C •
~----
I
--T
I
1
2
3
4
- 3 - -
5
6
-1 7
Phosphorus ( ~ )
660
670
680
690
700
710
720
730
740
750
Temperature (K)
Figure 3 Comparisons of kl/k 2 for different pellets, labelled according to Table 2
Figure 4 Changes in C2 selectivity and CO conversion with temperature. (WHSV of 87 and 151 mmol/gc,rhr, Pellets D and H respectively).
Regarding the C5 + fraction and runs developed with the 9 different pellets, no liquid hydrocarbon (C5 +) products were observed in the product stream. This was true for all operating conditions studied. Moreover, in all experiments the product mixture, as already stated, contained methane, ethylene and ethane, and in only a few cases C 3. The Cz selectivity for the majority of the runs was approximately 4055 % with an ethylene selectivity of 32.3 % (for Pellet C, run C6). These results compare very favourably with those obtained previously by Simard (1990) using a compound catalyst (no P added) in which a significant C5 + liquid fraction was obtained. In the case of Simard (1990) more than twenty hydrocarbon components, high in aromatics, were obtained with a maximum C2 selectivity in the hydrocarbon fraction of only 9.1%. Results in the present study demonstrate that these compound catalysts are effective in eliminating the C5 + fraction, and in increasing as well both C2 yields and selectivities. Thus, the procedures adopted for the preparation of the MSC catalyst, for the preparation of the MCC catalyst and its modification were successful. Concerning the influence of various parameters, it was observed that the temperature was a major factor on C2 selectivity and CO conversion. This trend is illustrated in Figure 4 (Pellets D and H). It was observed that when temperature increased the CO conversion augmented while the C2 selectivity decreased. Thus, it seems that there is an optimum operating temperature range for maximum C2 yields (Yc2). This optimum temperature was identified as being in between 733 K and 743 K. Moreover, it was further observed that total pressure (3200 kPa to 4560 kPa) did not have a strong effect towards the C 2 selectivities and CO conversions. Comparing results at different temperatures with various pellets (Figure 5) one can postulate that higher temperatures severely depress ethylene hydrogenation leading to maximum {C2=/(C2 = +nee) } values.This parameter was as high as 48 - 50% at 743 K and only 20-25% at 673 K. Under the conditions considered, catalysts with the higher phosphorous levels, adequate Zn/Cr ratios and operated at the higher temperatures (733 K to 743 K) yielded optimized Xco, So_ and {C2=/(C2 = +nC2) }. Optimum values for C2=-Selectivities were close to 13% with 70% CO conversions:
IO Sc2= = S~ ([C2=]/([C2---]+ [nC2])) 100 Sc~= = 0.25 x 0.50 x 100 = 12.5 % Thus, 12.5% ethylene, with comparable ethane selectivities, represent encouraging prospects for this compound catalyst.
0.55
0.50 t
I
--~-
Pellet
--~
Pellet
D F
--~
Pellet
H
0.45
0.40 CONCLUSIONS ,'/ I 1. Methanol synthesis on the MSC component "~ 0.35 of the compound catalyst was the controlling u0 / kinetic step while the water-gas shift reaction o" 0.30 took place at conditions very close to J chemical equilibrium. This allowed to 0.25 propose an overall stoichiometry for as 0.20 follows: I A/ 4CO + 2 H 2 ~ CH2=CH 2 + 2CO2 O. 1 5 -~ ~ -- - ~ 2. Experimental results demonstrated that when 660 670 580 690 700 710 720 730 740 750 the ZSM5 zeolite was modified with P, high Temperature(K) selectivity for light olefins with optimum C_,z= Figure 5 Selectivity changes of ethylene over the total selectivities of 13% and C2 selectivities of C 2 fraction with temperature (Pellets D, F, H) 26% were achieved. 3. The relative importance of formation of C2 and Cz hydrocarbons was favourably affected by high P levels and was effectively represented by a kJk 2 ratio. 4. The compound catalysts developed in this study eliminated the undesired C5+ fraction. /.
,
+
NOTATION combined pentane and higher hydrocarbons lump C5 + [CI] molar concentration of methane, mol/cm3 [C2] molar concentration of ethane and ethylene, mol/cm3 [C2 = ] molar concentration of ethylene, mole/cm3 [nC2] molar concentration of ethane, mole/cm3 d-spacing,/~, d activation energy of dissociative hydrogenation reaction E1 activation energy of non-dissociative hydrogenation reaction E: kinetic constant for the rate of C2 formation kl kinetic constant for the rate of methane formation ks selectivity to the combined ethane plus ethylene lump (moles C2 formed / moles of CO converted) Sc2 Sc2= selectivity to the ethylene lump molar conversion of CO XCO ACKNOWLEDGMENTS The financial support of Imperial Oil Grant Program and the Natural Sciences and Engineering Research Council is gratefully acknowledged. REFERENCES Butter S.A., Kaeding W.W.,1976, Phosphorus containing zeolite catalyst, U.S. Patent, 3,972,832. Carruthers J. and Sing K., 1976, Chem. and Ind., 11, 1919. de Lasa H., 1989, Pseudoadiabatic reactor for exothermal catalytic conversions, U.S. Patent, 4,929,798. Gabelica Z., Blom N., Derouane E.G., 1983, Appl. Catal. 5, 227. Jarosch K., 1995, MESc Thesis, University of Western Ontario, London, Ontario, Canada. Rong S.,1994, MESc Thesis, University Western Ontario London, Ontario, Canada. Simard, F., Mahay, A., Jean G., de Lasa,H., 1991a, Can.J.Chem.Engng., 69, 898. Simard, F., Mahay, A., Ravella, A., Jean G., de Lasa, H., 1991b, Ind. Chem. Eng. Research, 30, 1448 Simard, F., 1990, PhD Dissertation, The University of Western Ontario. Simard F. Sedran U.,Figoli N., de Lasa H., 1995, Appl. Catal., 125, 81. Tronconi E., Cristiani C., Ferlazzo N., Forzatti P., Villa P., Pasquon I., 1987, Appl. Catal., 32, 285.