Homolytic C-S bond cleavage on a heterogeneous Co/Mo/S hydrodesulfurization catalyst

Full text of DOI:10.1021/ja962941a

842
J. Am. Chem. Soc. 1997, 119, 842-843
Homolytic C-S Bond Cleavage on a Heterogeneous
Co/Mo/S Hydrodesulfurization Catalyst
Keenan E. Dungey and M. David Curtis*
Willard H. Dow Laboratory, Department of Chemistry
The UniVersity of Michigan
Ann Arbor, Michigan 48109-1055
ReceiVed August 21, 1996
The hydrodesulfurization (HDS) of petroleum is an industri-
ally important catalytic process whose mechanism still remains
Figure 1. Normalized 1-butene (b), cis-2-butene (9) and trans-2-
butene (2) products from the HDS of CPMT (arrows indicate the
equilibrium concentrations of the respective butenes at 600 K).
1
2
elusive. Organometallic reactions and reactions on single-
3
crystal surfaces have been used to model HDS catalysts and
possible mechanisms. Recent work of Bianchini et al. has
4
demonstrated the first HDS catalyst by a homogeneous system.
catalyst. A Co-promoted MoS2 HDS catalyst supported on
γ-alumina was loaded into a differential flow reactor and
However, few parallels between model systems and the actual
2,5
Co/Mo/S catalysts have been made.
We wish to report the
first direct comparison of an organometallic desulfurization
reaction with the reactivity of a heterogeneous HDS catalyst.
Previously, we reported that 1′, Cp′2Mo2Co2S3(CO)4 (Cp′ )
5
η -C5H4Me), desulfurized organic thiols and sulfides in solution
to form the cluster Cp′2Mo2Co2S4(CO)2 (2′) and the correspond-
6
ing hydrocarbon. A kinetic study revealed that binding of the
thiol to cluster 1′ was the rate-determining step in the desulfu-
rization reaction and that the C-S bond was cleaved homolyti-
activated in the usual manner.10 A solution of CPMT11 was
injected into a stream of hydrogen gas at 1 atm pressure and
passed over the heated catalyst. At 300 °C, almost complete
conversion of the thiol was observed, and the hydrocarbon
products were a mixture of butene isomers in their thermo-
dynamic ratios and a small amount of propene (Figure 1; Table
7
cally. Proof of a radical intermediate was furnished by the
reaction of cyclopropylmethylthiol (CPMT) with the cluster 1′.
The organic product was 1-butene, arising from the rearrange-
ment of the cyclopropylmethyl radical to the butenyl radical
which then abstracted an H atom (eq 1). This rearrangement
1
2
1
, Supporting Information). The conversion was varied by
lowering the temperature. The change in product distribution
to 100% 1-butene at 140 °C (Figure 1, Table 1) indicates that
the initial product of the HDS reaction is 1-butene, which is
then isomerized to the other thermodynamically favored butene
isomers. No methylcyclopropane (MCP) was detected.
In a control experiment, MCP in a hydrogen stream at 1 atm
pressure was passed over the catalyst bed. In addition to
unreacted MCP, the products at 140 °C were 33% 1-butene,
44% cis-2-butene, and 23% trans-2-butene. The 1-butene was
present in large excess of its thermodynamic ratio: 1.0:1.3:0.7
8
occurs at a known rate (k1 ) 1.3 × 10 /s at 25 °C) and is
8
characteristic of a free radical clock reaction. We now report
5
that the reaction of cluster 1 (Cp ) η -C5H5) with a slower
5
radical clock, 5-hexene-1-thiol (k2 ) 1.0 × 10 /s at 25 °C),
(obsd) vs 1.0:5.0:10.0 (equilibrium) for 1-butene/cis-2-butene/
also produced the characteristic free radical rearrangement
product, methylcyclopentane (eq 2).9
trans-2-butene (Table 1). It has been shown that MCP thermally
rearranges by C-C bond homolysis, followed by H-atom shift,1
and the ratio of the cis- to trans-2-butene products was
3
Having demonstrated C-S bond homolysis as the desulfu-
rization mechanism in this model system, we desired to explore
whether or not this mechanism applied to the actual HDS
1
4
approximately 2, as seen here. These results clearly show
that the product slate observed in the HDS of CPMT does not
come from the isomerization of initially formed MCP but is
consistent with the formation and rearrangement of the cyclo-
propylmethyl radical in the HDS process.
(
1) (a) Topsoe, H.; Clausen, B. S.; Massoth, F. E. In Catalysis: Science
and Technology; Anderson, J. R.; Boudart, M., Eds.; Springer-Verlag: New
York, NY, 1996; Vol. 11. (b) Chianalli, R. R.; Daage, M.; Ledoux, M. J.
AdV. Catal. 1994, 40, 177.
(
2) (a) Angelici, R. J. In Encyclopedia of Inorganic Chemistry; King,
R. B., Ed.; John Wiley & Sons: New York, 1994; Vol. 3, pp 1433-1443.
b) Angelici, R. J. Acc. Chem. Res. 1988, 21, 387. (c) Rauchfuss, T. B.
Prog. Inorg. Chem. 1991, 39, 259. (d) Adams, R. D. Chem. ReV. 1995, 95,
(10) MoO3 (14%, 0.040g) and CoO (3.5%) supported on alumina (Strem)
was diluted to 1% metal loading by mixing with γ-Al2O3 (0.676 g) (Catapal).
This plug of catalyst loosely filled a 0.5 cm i.d. glass tube to a depth of 1
(
3
2
587.
cm, where it was reduced at 400 °C in H2 (20 cm /min) for 4 h, followed
3
(
3) Wiegend, B. C.; Friend, C. M. Chem. ReV. 1992, 92, 491.
by sulfiding for 4 h at 400 °C in 10% H2S/H2 (total flow 20 cm /min).
(
4) Bianchini, C.; Jimenez, M. V.; Meli, A.; Moneti, S.; Vizza, F.;
(11) Prepared by S. H. Druker following the procedure of Cossar (Cossar,
B. C.; Fournier, J. O.; Fields, D. L.; Reynolds, D. D. J. Org. Chem. 1962,
27, 93). A solution in benzene-d6 (0.2 M) was injected into the heated
reagion of a differential flow reactor at a rate of approximately 0.1 mL/h.
The reactor was loaded as described above and the carrier gas was H2 (20
Herrera, V.; Saanchez-Delgado, R. A. Organometallics 1995, 14, 2342.
(
(
5) Jones, W. D.; Chin, R. M. J. Am. Chem. Soc. 1994, 116, 198.
6) Curtis, M. D.; Curnow, O. J.; Riaz, U. J. Am. Chem. Soc. 1994, 116,
4
357.
3
1
(
7) (a) Druker, S. H.; Curtis, M. D. J. Am. Chem. Soc. 1995, 117, 6366.
cm /min). Products were analyzed by GC (PE 8400, 8 ft × /8 in., 0.19%
(
b) Curtis, M. D.; Druker, S. H. J. Am. Chem. Soc. In press.
picric acid on Carbopack-C), quantified by a standardized gas mix (Scott
Specialty Gases), and verified by GC/MS and H NMR.
1
(
8) Griller, D.; Ingold, K. U. Acc. Chem. Res. 1980, 13, 317.
(
9) In an inert atmosphere box, a resealable NMR tube was charged with
(12) Thermodynamic ratios calculated at 600 K from American Petroleum
Institute (API) tables. Rossini, F. D.; Pitzer, K. S.; Arnett, R. L.; Braun, R.
M.; Pimental, G. C. Selected Values of Physical and Thermodynamic
Properties of Hydrocarbons and Related Compounds; Carnegie Press:
Pittsburgh, PA, 1953.
(13) Jakubowski, E.; Sandhu, H. S.; Strausz, O. P. J. Am. Chem. Soc.
1971, 93, 2610.
(14) Chesick, J. P. J. Am. Chem. Soc. 1960, 82, 3277.
1
(23mg, 35 µmol), toluene-d8 (1 mL), and 5-hexene-1-thiol (3 mL, 26
µmol), sealed, and removed. The mixture was heated to reflux for 10 h,
1
while H-NMR spectra were taken periodically to monitor the reaction
progress. Compound 1 was converted to 2 almost quantitatively, while the
thiol was desulfurized, forming methylcyclopentane, as indicated by the
doublet at δ 0.95 (in toluene-d8). GC/MS of the head gases had one peak
with m/e ) 69 and 56.
S0002-7863(96)02941-1 CCC: $14.00 © 1997 American Chemical Society

Communications to the Editor
J. Am. Chem. Soc., Vol. 119, No. 4, 1997 843
Scheme 1
Figure 2. Normalized 1-hexene (b) and internal hexenes (2) yields
from the HDS of 5-hexene-1-thiol (dashed lines indicate the equilibrium
concentrations of the respective hexenes).
The hydrodesulfurization of 5-hexenethiol by the heteroge-
15
neous catalyst was also performed, and in this case the product
was not that expected from free radical rearrangement (Table
2
, Supporting Information). At 200 °C, the conversion was low,
and the products were internal hexenes (62 mol %) and 1-hexene
different rates of hydrogen abstraction. For example, if metal
hydrides are the hydrogen donors on the surface of the
heterogeneous catalyst (as presumed in Scheme 1), we would
expect the reaction rate with formed radical to be greater than
the rate of hydrogen abstraction from thiols in solution, based
on organometallic models.
The formation of free radicals in the vicinity of surfaces has
(
(
27 mol %), the latter in excess of its thermodynamic ratio
equilibrium: 94% internal/6% terminal). Some hydrogenation
product (hexane) was observed but no methylcyclopentane was
detected. As the reaction temperature was raised, the product
slate approached the equilibrium ratios (Figure 2).16 In control
experiments, 1-hexene was quickly isomerized to internal
hexenes, while methylcyclopentane showed no rearrangement.
The product distribution is consistent with the initial formation
of 1-hexene, followed by isomerization to the thermodynami-
cally favored internal hexenes.
A mechanism for the heterogeneous HDS of the free radical
clocks is presented in Scheme 1. This mechanism parallels that
proposed for the homogeneous desulfurization of thiols by
cluster 1′. The steps include adsorption of the thiol on the
surface, homolytic C-S bond cleavage, and rearrangement of
the resulting radical. For cyclopropylmethyl radical, this occurs
faster than hydrogenation of the radical, presumably by a
hydrogen atom on the surface of the catalyst. The results of
the HDS of 5-hexenethiol can also be explained by Scheme 1
if the rate of hydrogenation is greater than the rate of cyclization
of the 5-hexenyl radical (k2 ) 1.5 × 10 /s at 300 °C). We
recognize that other mechanisms are possible (e.g., initial
hydrogenation of the olefin, E2 elimination of the thiol to
produce 1-hexene, followed by isomerization to internal hexenes)
and experiments are underway in our lab to address these issues.
The difference in reactivity between the cluster model and
the heterogeneous catalyst may stem from the drastic differences
in conditions. The concentration and reaction rate of hydrogen
atom sources (metal hydrides or thiols) should be very different
from solution state to heterogeneous systems, resulting in
18
been explored under UHV (ultrahigh vacuum) conditions with
1
9
alkyl halides and metal crystals. Recently, a study of the
desulfurization of cyclopropylmethylthiol on a Mo(110) crystal
2
0
was performed under UHV conditions. Homolytic cleavage
was proposed as the mechanism for the C-S bond cleavage on
the surface, but no rearrangement products typical of the
cyclopropylmethyl radical were observed. The lack of rear-
rangement products was rationalized by assuming that C-S
bond breaking and C-H bond forming were nearly simultaneous
(at least the hydrogenation occurred at a much faster rate than
the radical rearrangement). We have demonstrated here that
this is not the case for the HDS of CPMT over a conventional
catalyst operating at the usual temperatures and atmospheric
pressure. It appears that in this particular case, at least, an
organometallic cluster in solution models the heterogeneous
catalyst better than a clean metal surface under UHV conditions.
7
17
Acknowledgment. This work was supported by the National
Science Foundation under grant CHE-9523056. K.E.D. is grateful to
the Department of Chemistry and the College of Literature, Sciences
and Arts for a Regent’s Fellowship.
Supporting Information Available: Tables of product distribution
(1 page). See any current masthead page for ordering and Internet
access instructions.
(
15) A 0.28 mM solution of 5-hexenethiol (for preparation, see: Walling,
JA962941A
C.; Pearson, M. S. J. Am. Chem. Soc. 1964, 86, 2262) in toluene was injected
via syringe into the heated zone of the differential flow reactor at approx.
(18) Bullock, R. M.; Samsel, E. G. J. Am. Chem. Soc. 1990, 112, 6886.
(19) Lin, J.-L.; Bent, B. E. J. Am. Chem. Soc. 1993, 115, 2849. Lin,
J.-L.; Bent, B. E. J. Am. Chem. Soc. 1993, 115, 6943.
(20) Wiegand, B. C.; Napier, M. E.; Friend, C. M.; Uvdal, P. J. Am.
Chem. Soc. 1996, 118, 2962.
0
.1 mL/h. The reactor was loaded with 0.36 g of catalyst and the carrier
3
1
1
gas was H2 (10 cm /min). Products were analyzed as described above.
1
2
(
16) Equlibrium calculated from API tables.
17) Calculated from tables in ref 8.
(