Ind. Eng. Chem. Prod. Res. Dev. 1983, 22,31-34
31
Mechanism of Catalytic Hydrodeoxygenation of Tetrahydrofuran Edward Furlmsky Energy Research Laboratories, Canada Centre for Mineral and Energy Technology, Department of Energy, Mines and Resources, Ottawa, Canada, K1A OG1
Product distribution during the hydrodeoxygenation of tetrahydrofuran depends on catalyst pretreatment. Butenes are the major products in the presence of sulfided catalyst while butadiene and propylene are predominant over the reduced catalyst. Temperature effects on product distribution are compared with trends followed from thermodynamic equilibria. Approximate surface reaction routes are proposed to account for the observed differences.
Introduction Since the reserves of conventional crudes are declining, there is a growing need to produce fuels from synthetic crudes. In this respect coal, oil shale, and biomass have been investigated as potential feedstocks (Gorbaty and Harney, 1979). Primary products from processing of these feedstocks always require additional upgrading resulting from high content of aromatics and heteroatom-containing compounds. In liquids derived from coal and biomass, the content of 0 heteroatom is markedly higher than that of N and S. To meet specifications for commercial fuels and to increase their yields, most of the 0 must be removed from the liquids. Hydrodeoxygenation (HDO) is thus one of the major reactions occurring during upgrading. The most important O-containing species present in synthetic liquids are phenols and furanic rings containing heterocyclic compounds. In fact, the former are known to be intermediates in the overall HDO of benzofuran and dibenzofuran (Mallinson et al., 1980). Formation of the intermediates requires the opening of the heterorings, which is based on a preliminary hydrogenation of the ring (Rollmann, 1977). This suggests that saturated furanic rings will be a part of the overall HDO mechanism as well. In the present work the HDO mechanism of tetrahydrofuran (THF) was investigated. Despite the simple structure of this model compound, its gas phase HDO in the presence of a molybdate catalyst has not been r e p o d , although numerous reports on the reactions of similar Sand N-containing heterorings have appeared in the scientific literature. These reports indicate that S and N heteroatoms in feedstocks frequently modify the structure of catalyst surface. It is expected that 0 heteroatoms, if present in feedstocks, will affect the surface structure as well. The high reactivity of THF makes it the ideal model compound to study these effects. Understanding the HDO mechanism of THF appears to be a starting point for such an investigation. Experimental Section Materials. The tetrahydrofuran used was the product of Burdick and Jackson Laboratories (distilled in glass) and used as received. The hydrogen, a Matheson product, was of UHP grade. Catalysts. The catalyst was the extrudate type of Harshaw Co., containing 15 wt 5% Moos and 3 wt % COO on alumina. The extrudates were crushed and sieved through 60 mesh prior to their use in the reactor. The catalyst was subjected to two different pretreatments to obtain the material in its reduced form (catalyst R) and in a sulfided form (catalyst S). Preparation of catalyst R involved heating the material in its oxidic form 0196-4321/83/1222-0031$01.50/0
Published
from room temperature to 400 "C in a stream of hydrogen (80 cm3/min) for 45 min. At this temperature the reduction continued for another 30 min. Catalyst S was prepared by sulfiding the catalyst R at 400 "C for 45 min using a mixture of H2Sand H2containing 10% H2S. These conditions ensured that the major portion of the sulfur needed to sulfide the catalyst was already taken up (Massoth, 1975). Apparatus and Procedure. The basic design of the continuous microcatalytic system is described in detail elsewhere (Owens and Amberg, 1961). A stream of H2 at near atmospheric pressure was passed, via a dispersion tube, through a saturator filled with THF and kept at 0 OC. This corresponds to 6.4 vol % of THF in the mixture of H2. The H2 saturated with THF vapor flowed to the reactor bed. Analysis. Quantitative determination of products was performed chromatographically. The flame ionization detector and an 8 m long column of 20% BMEA on Chromosorb P-AW kept at 30 OC were essential parts of the system. Helium was used as the carrier gas. The injection system consisted of two loops of 5 mL volume. Injection was achieved by switching the valve and sweeping out the sampling loop with the carrier gas. Yields of products in Table I represent the amounts in microliters determined in 5 mL-of injected samples. In case that no HDO reaction of THF took place, the 5 mL of the mixture would contain 320 wL of the reactant. A gas chromatography-mass spectroscopic (GC/MS) technique was used for qualitative identification of the products. The Finnigan quadrupole GC/MS system with INCOS computer attached was used.
Results and Discussion The distribution and the yield of products was time dependent, as shown in Figure 1. As one would expect, the yields decreased with time. Over catalyst S the decrease of butenes yields was much more pronounced than that of propylene and after 5 h the latter accounted for about 20% of the total amount of products as compared to 14% at the beginning of the experiment. The contribution of butadiene to the total yield of products increased gradually as well. The change in product distribution with time over catalyst R was more uniform. Temperature effects on product distribution over catalyst R and catalyst S are shown in Table I. The results are normalized to 1g although in the experimental runs only about 400 mg of accurately weighed catalyst was used. Results of the runs in which pure A1203support was used as well as the run using catalyst R in the flow of nitrogen are also included. Product distribution in Table I are those 1983 by the American Chemical Society
32 Ind. Eng. Chem. Prod. Res. Dev., Vol. 22, No. 1, 1983
Table I. Effects of Temperature on Yields of Products (pL) ~~~~
temp, "C catalyst R 330
350
catalyst S
400
375
430
N400a 340
375
400
A1203
430
400
N400' ~~
propylene
4.9 0.6 1.5 0.3 0.3 3.0
n-butane 1-butene trans-butene cis- butene butadiene a
16.1 0.8 3.0 0.5 0.5 10.1
11.9 0.9 3.1 0.5 0.5 9.3
38.6 trb 5.2 2.5 2.3 28.6
21.2 0.6 4.1 0.7 0.7 14.7
Experiments performed in the stream of nitrogen. Catalyst S
I
\
40
L
3
-
A 0
L
. .
&
30
.
PROPYLENE I-0UTENE
BUTADIENE T R A N S - BUTENE- 2 CIS-BUTENE-2
\*
.-u E 20
10
0
Catalyst R
- transtcis
Butene - 2
30 -1
2
20
0 .-
E 10
0 0
1
2
3
4
tr 2.0
0.4 0.4 2.4 5.9 4.4 tr
2.8 1.5 7.2 14.5 11.8
tr
12.4 1.0 17.0 11.0 10.4 7.8
17.0 0.2 20.7 10.2 9.6 16.5
3.9 0 0.6 0 0 3.9
4.2 0
0.8 0 0 3.6
tr = traces.
50 A
2.8 tr 0.6 tr
5
T I M E , hrs Figure 1. Yields of products vs. time. of gas mixtures taken after 5 h on stream. Ethylene and propane were identified as the additional volatile products formed in trace quantities in the presence of the catalysts. Propane was not found in the products over catalyst R. Also, several very small peaks of heavier products were observed in chromatograms. The GC/MS analysis confirmed clearly the presence of four products having mass 70 (pentenes or cyclopentane), four products of mass 84 (hexenes), and another three products of mass 98 (heptenes). Although the peaks were more visible over catalyst R than over catalyst S, their contribution to the total yield of products was minor in both cases. Water, as the HDO product, was trapped at the bottom of the tube exiting from the reactor. No attempts were made to measure this product quantitatively. In addition to H20, a tarry material accumulated on the walls of the tube during the runs over catalyst R. The amount of tar increased with temperature. Separate experiments were performed for both catalysts (400 "C and 5 h) to measure the amount of reacted THF
quantitatively. Comparison of reacted amounts with total yields of products in Table I resulted in an estimate of the amount of THF converted to tar and coke. Thus, for catalyst R about 74 pL of T H F reacted compared to 45 p L of detected products. This suggests that about 29 pL of THF was converted to tar and coke. For catalyst S about 65 FL of THF reacted compared to about 60 pL of detected products. No evidence was found of the presence of furan in the product mixture. This suggests that under present experimental conditions, the HDO of THF dominates over its dehydrogenation. Then, for the experiments performed in the presence of the catalyst S, the sum of all products a t a given temperature closely represents the amount of T H F which underwent HDO. The HDO conversions calculated from the sums and the amount of THF in the mixture with Hz entering the reactor varied from 4% at 340 "C to 23% at 430 "C. Similar calculations of HDO conversion for runs performed over the catalyst R give results of a lesser significance because of the tarry product, the quantity of which was not determined. In the experiment performed in the absence of catalyst, no reaction of THF occured. Results in Table I indicate the effect of catalyst pretreatment on product distribution. The main difference resulting from different pretreatment is the appearance of propylene and butadiene as major products over catalyst R as compared to more even distribution of products over catalyst S. The n-butane was the other fully saturated product detected. At 430 "C this product was formed only in trace quantities. Contribution of n-butane to the total yield of hydrocarbons decreased as temperature increased, and at 330 OC, it accounted for about 6% as compared to 2% at 400 "C. Temperature effects on the yields of products formed over catalyst R differ from those over catalyst S. For example, temperature increase from 330 OC to 430 "C resulted in about a threefold increase in the yield of 1butene over catalyst R. A similar change of temperature in the presence of catalyst S resulted in about a tenfold increade in the yield of this product. Also, the change in propylene yield was more pronounced over catalyst S than over catalyst R. Yields of trans- and cis-butenes in the presence of the catalyst S reached a maximum at about 375 "C. Further temperature increase resulted in a slight decrease in the yields of these products. Because of the small yields of trans- and cis-butenes over the catalyst R, their temperature trends are inconclusive. The appearance of products over A1,03 and catalyst R in the presence of nitrogen, although in very small quantities, is rather surprising. This suggests that the HDO of THF can occur, to some extent, in the absence of external hydrogen. Over A1,0,, after 5 h on stream, only propylene, butadiene, and 1-butene were detected while over the catalyst R, in the absence of hydrogen, traces of
Ind. Eng. Chem. Prod. Res. Dev., Vol. 22, No. 1, 1983 33 H
' H
HC -CH
II H&
II CH,
H,
Figure 3. Approximate reaction route for hydrodeoxygenation of tetrahydrofuran on sulfide catalyst.
777jT77 Figure 2. Approximate surface reaction routes for hydrodeoxygenation of tetrahydrofuran on reduced catalyst.
trans- and cis-butenes were also detected. Occurrence of the reaction in the absence of hydrogen suggests that intramolecular migration of H atoms in THF while adsorbed on the surface play an important role. To show the participation of internal hydrogen in the HDO reactions, the speculative mechanism shown in Figure 2 is proposed. Route 1 depicts the formation of H 2 0 and butadiene as the result of H atom migration towards the heteroatom. In step la, a complex is formed after reactant molecules contact the surface. In the complex, the dotted lines represent bonds being cleaved while the broken lines are of those being broken. The difference in bond strength of bonds formed and that of bonds broken is the driving force of reactions. The strength of the C-H bond in butadiene (110 kcal/mol) and the 0-H bond in H,O (120 kcal/mol) as compared to the C-H and C-O bonds in THF (92 and 95 kcal/mol, respectively) suggests that the driving force for such reactions is large. Route 2 shows the cleavage of only one C-0 bond in the first step with the assistance of internal H atoms. This leads to a transition state in which the hydrocarbon part is attached to the surface via 0 atom (steps 2a and 2b). Decomposition of the transition state leads either to 1butene (C-0 bond cleavage) or to propylene (C-C bond cleavage). Step 2 d yields a formaldehyde-like species. The failure to detect such a compound in the products suggests that it remains attached to the surface. In this form the species might participate in the formation of polymeric products or carbonaceous deposits which were formed on catalyst surfaces in all experiments (Furimsky, 1983). The higher yields of butenes than of butadiene over the catalyst S are attributed to more efficient surface hydrogen transfer to reactant molecules. The much more pronounced increase in 1-butene yield with temperature over catalyst S than catalyst R is an indication of more than one reaction route for 1-butene formation. One of the routes might be based on a participation of surface hydrogen. Increasing yields of 1-butene with temperature are in agreement with a more efficient absorption of hydrogen on the catalyst surface with increasing temperature, as reported in the literature (Lipsch and Schuit, 1969). The involvement of surface hydrogen as shown in Figure 3 results in opening of the THF ring. This leaves a fully saturated butyl radical attached to the surface via an 0 atom. The subsequent 1-butene formation can occur with participation of internal hydrogen. The formation of 1butene from the transition state is believed to be more favorable than that of 2-butenes because the former re-
' I T x 1000,
'/OK
Figure 4. Thermodynamic equilibria for the reactions listed in Table 11. Table 11. Thermodynamic Equilibria Equations T H F 2 furan THF 2 CH,CH=CH, + CH,O THF Z CH,=CHCH=CH, + H,O n-butyl alcohol 2 CH,CH,CH=CH, + H,O THF t H, 2 CH,CH,CH=CH, + H,O THF t 2H, 2 CH,CH,CH,CH, + H,O THF t M@+ Z CH,CH,CH=CH, + MeO, CH=CHCH=CH, + H, 2 CH,CH,CH=C&
action requires less intramolecular rearrangments.
Thermodynamic Equilibria Considerations It is of interest to see how the experimental observations agree with predictions which follow from thermodynamic data. The equilibrium equations via which the identified products might be formed are summarized in Table 11. Their log K vs. 1/T correlations are shown in Figure 4. The calculations were based on thermodynamic data compiled by Stull et al. (1967). Equations 1to 3 represent formations of product from THF without participation of an external hydrogen. The correlations show that the probability for THF dehydrogenation is rather low as compared to the reactions in which the reactant is converted to propylene and butadiene. Temperature effects on log K values for eq 2 and 3 agree qualitatively with the effects on yields of propylene and butadiene observed over catalyst S as well as catalyst R. Equilibria 5 and 6 involve an external hydrogen. Trends which follow from the equilibria contradict temperature effects on the yields of n-butene and 1-butene. Thus, n-butane if detected was always a minor product, while yields of 1-butene always increased with temperature. Low n-butane yields can be attributed to either kinetic effects or a limited availability of surface hydrogen. It has been reported that a near-atmospheric pressure of hydrogen, as applied in the present work, is insufficient to produce fully hydrogenated products (Hargreaves and Ross, 1979). Increased yields of 1-butene with temperature increase agree better with temperature effects on log K of eq 4. In
Ind. Eng. Chem. Prod. Res. Dev.
34
this case, the n-butyl alcohol reflects, as closely as possible, a complex such as that shown in Figure 2 and 3. Here, the 0 remains attached to the surface; therefore, a reaction resembling eq 7 may describe better the real situation. In eq 7 the Mez+represents a cation of Al, Mo, or Co. Unfortunately, a calculation of log K values for equilibrium 7 is not possible because the form of MeZ+on the catalyst surface is not clearly understood (Furimsky, 1980). The log K vs. 1/T correlations for the formation of trans- and cis-butenes from THF with participation of an external hydrogen follow closely that of eq 5. Temperature effects on the yields of trans- and cis-butenes observed experimentally over 375 O C agree qualitatively with trends which follow from the correlations. The appearance of trans- and cis-butenes as major products over the catalyst S, which has a much better hydrogen transfer efficiency (Furimsky, 1980), is supportive evidence for the involvement of an external hydrogen in the reactions. Another source of butenes might be the hydrogenation of butadiene as shown in equilibrium reaction 8. It was, indeed, reported by Kolboe and Amber (1966) that under conditions similar to those applied in the present work, butadiene is hydrogenated to trans-, cis- and 1-butene. The formation of n-butane as a minor product was also confirmed by these authors. The presence of butadiene in trace quantities over the catalyst S at low temperatures and the increased contribution of n-butane to the total yield of hydrocarbons a t low temperatures (Table I) is in support of the reaction route. Conclusions
Product distribution in HDO of T H F depends on the structure of the catalyst surface. It appears that an interaction of reactant molecules with some surface species enhances selectivity for certain reactions. This is supported by a marked increase of 1-butene yield with temperature increase over the catalyst S. It is proposed that
1983,22, 34-38
1-butene may be formed in at least three different reaction routes. Appearance of butenes as major products over the catalyst S as compared to propylene and butadiene over catalyst R c o n f i i the expected beneficial effects resulting from catalyst sflidation. More efficient transfer of surface hydrogen to the adsorbed THF molecules in the presence of catalyst S results in a suppression of propylene-forming reactions (C-C bond cleavage). The formation of heavier products and tarry material is suppressed as well. The latter is attributed to enhanced hydrogenation of butadiene to butenes. In other words, faster removal of readily polymerizable butadiene from the products is achieved. To a certain extent the HDO of THF occurs even in the absence of surface hydrogen. To account for this observation, the mechanism based on an intramolecular migration of THF hydrogen atoms is proposed. It is further suggested that the presence of surface hydrogen improves the chances for THF ring opening, as the first step in the overall HDO mechanism. Registry NO.THF,109-99-9; M003,1313-27-5; COO,1307-96-6. L i t e r a t u r e Cited Furimsky, E. Ind. Eng. Chem. Prod. Res. Dev. 1983. following paper in this issue. Furimsky, E. Catal. Rev. Sci. Eng. 1960, 22, 371. Gorbaty, M. L.; Harney, E. M. A&. Chem. Ser. 1979, 779. Hargreaves, A. E.; Ross, J. R. H. J. Catal. 1979, 56, 363. Kolboe, S.;Amberg, C. H. Can. J. Chem. 1966, 4 4 , 2623. Lipsch, J. M. T. G.; Schult, G. C. A. J. Catal. 1969, 15, 179. Mallinson, R. G.; Chao, K. C.; Greenkorn, R. A. f r e p r . . Div. fetr. Chem., Am. Chem. SOC. 1980, 25, 120. Massoth, F. E. J. Catal. 1975, 3 6 , 164. Owens, P. J.; Amberg, C. H. A&. Chem. Ser. 1961, 3 3 , 182. Rollmann, L. D. J. Catal. 1977, 4 6 , 243. Stull, D. R.; Westrum, E. F.; Sinke, G. C. "The Chemical Thermodynamics of Organic Compounds"; Why: New York, 1967; p 420.
Received for review November 30, 1981 Accepted August 30, 1982
Deactivation of Molybdate Catalyst during Hydrodeoxygenation of Tetrahydrofuran Edward Furlmsky Energy Research Laboratories, Canada Centre for Mineral and Energy Technology, Depaflment of Energy, Mines and Resources, Ottawa, Canada, KIA OG1
The formation of carbonaceous deposits was observed during catalytic hydrodeoxygenation of tetrahydrofuran in the presence of cobalt molybdate catalyst. Rates of deposit-forming reactions depend on catalyst pretreatment and are significantly lower for the sulfided catalyst as compared to the reduced catalyst. The high rates coincide with high yields of propylene and butadiene and with the accumulation of tarry liquid under catalygt bed. On the basis of high aromaticity of the liquid, the mechanism for the deposit formation is proposed. This includes a combination of two surface radicals followed by an intramolecular rearrangement, giving an aromatic structure.
Introduction
Formation of carbonaceous deposits on a catalyst surface is the main cause for the loss of catalytic activity. The mechanism of deposit-forming reactions is only vaguely understood, mainly because of the many factors involved. The complexity of the deactivation process was confirmed by examining components of deposits extracted from used
catalyst pellets (Furimsky, 1982). An approach based on a well-understood model reaction can provide valuable information. One such reaction appears to be hydrodeoxygenation (HDO) of tetrahydrofuran (THF). Under conditions applied in the present work, a sufficient amount of deposits was formed to permit an accurate determination. It was further observed that the
0196-432118311222-0034$01..50/0 Published 1983 by the American Chemical Society
