Vanadium-Catalyzed Epoxidations of Olefinic Alcohols in Liquid Carbon Dioxide

Article abstract of DOI:10.1021/om990467m

The selective epoxidation of olefinic alcohols with t-BuOOH in the presence of vanadium catalysts proceeds in liquid carbon dioxide with high conversions and selectivities. Rates measured in liquid CO₂ for the oxovanadium(V) triisopropoxide cat

Full text of DOI:10.1021/om990467m

4916
Organometallics 1999, 18, 4916-4924
Va n a d iu m -Ca ta lyzed Ep oxid a tion s of Olefin ic Alcoh ols
in Liqu id Ca r bon Dioxid e
David R. Pesiri, David K. Morita, Teri Walker, and William Tumas*
Los Alamos National Laboratory, Chemical Science and Technology Division, MS-J 514,
Los Alamos, New Mexico 87545
Received J une 16, 1999
The selective epoxidation of olefinic alcohols with t-BuOOH in the presence of vanadium
catalysts proceeds in liquid carbon dioxide with high conversions and selectivities. Rates
measured in liquid CO2 for the oxovanadium(V) triisopropoxide catalyzed epoxidation of
allylic and homoallylic alcohols using tert-butyl hydroperoxide are comparable to those
measured in methylene chloride, toluene, and n-hexane. The reactivity of the vanadium(IV)
bis(acetylacetonato) oxide catalyst in liquid CO2 was found to be substantially lower than
in organic solvents, presumably due to its low solubility in CO2. Highly fluorinated acac-
type ligands increased the catalytic reactivity of VO(acac)2-catalyzed epoxidations by
enhancing catalyst precursor solubility. Heterogeneous epoxidation reactions were also
carried out in liquid CO2 using vanadium complexes supported on cation-exchange polymers.
In tr od u ction
dioxide at supercritical conditions in an attempt to
enhance catalysis (e.g., high gas miscibility for hydro-
genations).5 Selective oxidation catalysis has been slower
to emerge as a well-studied class of reactions in dense-
phase CO2 despite several properties which make this
solvent well suited to oxidations: high oxidative stabil-
ity, high O2 solubility, low barriers to mass transfer, and
nonflammability.
Catalytic selective oxidation of organic compounds
continues to be one of the most challenging and impor-
tant areas of synthetic chemistry.1,2 Selective oxidation
catalysis covers a wide range of chemical transforma-
tions ranging from the commodity-scale heterogeneous
oxidation of petroleum feedstocks specialized homoge-
neous transformations with high stereospecificity. Metal-
catalyzed homogeneous epoxidations of alkenes are a
very useful subset of oxidation reactions, with a rich
history that has been described in the literature over
the past 30 years.3 We wish to report the use of dense-
phase carbon dioxide as a medium for oxidation cataly-
sis. Dense-phase fluids4 refer to systems which would
be considered gases at their temperature of use but are
compressed to the point that they have liquidlike
densities (F ) 0.3-1.0 g cm-3). Dense-phase fluids can
be used at temperatures above (supercritical fluids) or
below (liquid dense-phase fluids) their critical point. The
last 5 years have seen a considerable interest in the use
of dense-phase CO2 for catalytic chemistry, motivated
by both environmental and reactivity advantages.5 CO2
has been reported as a solvent for polymerizations,
hydrogenations, hydroformylations, free radical reac-
tions, substitutions, and various other catalytic trans-
formations in both its liquid and supercritical states (Tc
) 30.7 °C, Pc ) 73.8 bar). Many of these reports have
sought to capitalize on the unique properties of carbon
The earliest oxidation catalysis in dense-phase CO2
consisted of mainly nonselective, heterogeneous reac-
tions.6-8 Important contributions in this area include
the catalytic oxidation of organics in supercritical (sc)
CO2 using platinum catalysts supported on metal oxide
supports by Akgerman and co-workers.9,10 Dooley and
Knopf also studied the oxidation of organics in sc CO2
but used supported cobalt catalysts with success in
demonstrating results similar to those reported in
organic solvents.11 Our work set out to explore the
selective oxidation catalysis in dense-phase CO2 by
choosing a well-known reaction, the vanadium-catalyzed
epoxidation of several olefinic alcohols using t-BuOOH
as an oxidant.2 Reactivities and selectivities, as well as
kinetic measurements, were made in organic solvents
in order to benchmark reaction rates for a comparison
with reactions run in dense-phase CO2. A preliminary
communication of this work has already appeared,12
which coincided with a report by Kolis and co-workers
on the epoxidation of olefins with Mo(CO)6 and t-
(6) Srinivas, P.; Mukhopadhyay, M. Ind. Eng. Chem. Res. 1997, 36,
2066.
(7) Occhiogrosso, R. N.; McHugh, M. A. Chem. Eng. Sci. 1987, 42,
2481.
(8) Suppes, G.; Occhiogrosso, R. N.; McHugh, M. A. Ind. Eng. Chem.
Res. 1989, 28, 1152.
(9) Zhou, L.; Akgerman, A. Ind. Eng. Chem. Res. 1995, 34, 1588.
(10) Zhou, L.; Erkey, C.; Akgerman, A. Ind. Eng. Chem. Res. 1995,
34, 2122.
(1) Metal-Catalyzed Oxidations of Organic Compounds; Sheldon, R.
A., Kochi, J . K., Eds.; Academic: New York, 1991.
(2) Sharpless, K. B.; Verhoeven, T. R. Aldrichim. Acta 1979, 12, 63.
Chong, A. O.; Sharpless, K. B. J . Org. Chem. 1977, 42, 1587.
(3) Conte, V.; Di Furia, F.; Licini, G. Appl. Catal., A 1997, 157, 335.
(4) Morgenstern, D. A.; LeLacheur, R. M.; Morita, D. K.; Borkowsky,
S. L.; Feng, S.; Brown, G. H.; Luan, L.; Burk, M. F.; Tumas, W. In
Green Chemistry, Designing Chemistry for the Environment; Anastas,
P. T., Williamson, T. C., Eds.; ACS Symposium Series 626; American
Chemical Society: Washington, DC, 1996; p 132.
(11) Dooley, K. M.; Knopf, F. C. Ind. Eng. Chem. Res. 1987, 26, 1910.
(12) Pesiri, D. R.; Morita, D. K.; Glaze, W. H.; Tumas, W. Chem.
Commun. 1998, 9, 1015.
(5) J essop, P. G.; Ikariya, T.; Noyori, R. Chem. Rev. 1999, 99, 475.
10.1021/om990467m CCC: $18.00 © 1999 American Chemical Society
Publication on Web 10/28/1999

Vanadium-Catalyzed Epoxidations of Alcohols
Organometallics, Vol. 18, No. 24, 1999 4917
BuOOH in sc CO2.13 After these two publications
appeared, there were several reports of selective oxida-
tion catalysis in dense-phase CO2 ranging from the
metalloporphyrin-catalyzed homogeneous oxidation of
alkanes and alkenes14 to the diaestereoselective epoxi-
dation of olefins using salen complexes in sc CO2.15
Herein we provide a detailed study comparing oxida-
tion catalysis in liquid carbon dioxide with that in
organic solvents. Reactivities and selectivities as well
as kinetic measurements were examined in organic
solvents in order to benchmark reaction rates for a
comparison with reactions run in dense-phase CO2. The
limited reactivity observed for acetylacetonate (acac)-
based catalyst precursors (i.e., VO(acac)2) led to our use
of solubility-enhancing ligands to promote epoxidation
reactions in CO2. We have found that fluorinated
catalysts are effective in liquid CO2 and exhibit higher
reactivity than unfluorinated systems for certain solu-
bility-limited reactions. Vanadyl cations were also ion-
ically bound to fluorinated and nonfluorinated polymer
backbones to heterogenize the epoxidation catalyst in
what is believed to be the first report of polymer-
supported catalysis in dense-phase carbon dioxide.
mL of CO2 by loading the olefin, oxidant, and an internal
standard into a glass ampule, which was degassed and
flame-sealed. The ampule was placed into the high-
pressure reactor inside a glovebox, and the vanadium
catalyst was subsequently added to the reactor (outside
the ampule). The sealed reactor was then pressurized
with CO2 with the use of a high-pressure manifold, and
the ampule was shattered to begin the reaction. The
solution was magnetically stirred and was temperature-
controlled. Reactions were terminated for analysis by
carefully venting the pressurized contents of the reactor
into an organic solvent such as ice cold acetone and
subsequently rinsing the inside of the reactor with this
solution. These oxidations were homogeneous through-
out the course of the reaction and resulted in high
conversions and selectivities. Reactions were run for 24
h in order to compare conversions and selectivities for
different substrates. All preliminary work was carried
out at 25 °C (i.e., under liquid-CO2 conditions) because
such oxidations are commonly carried out at or below
room temperature in organic solvents and there was no
inherent advantage in running these reactions above
the critical point (vide infra), as was confirmed by a
number of reactions run under supercritical conditions.
The results for a number of vanadium-catalyzed epoxi-
dation reactions in liquid CO2 appear in Table 1.
Conversions and selectivities for a wide range of
substrates are shown in Table 1. In general, they were
in excess of 90% for the majority of the substrates
examined. Yields listed in Table 1 are GC yields.20 The
measured conversions with the VO(OiPr)3/t-BuOOH
system in liquid CO2 are consistent with those reported
for reactions in organic solvents.21,22 The epoxidations
of trans-2-hexen-1-ol (1), geraniol (2), nerol (3), and
1-penten-3-ol (4) each exhibited conversions and selec-
tivities above 95%. Although the selectivities for VO-
(OiPr)3-catalyzed epoxidations of allylic alcohols with
t-BuOOH in CO2 were high in most cases, there were a
few exceptions. For example, the low selectivity for the
aryl-containing cinnamyl alcohol substrate (8) was due
to a cleavage reaction which formed benzaldehyde as
the major product. This result is consistent with litera-
ture reports for aryl-substituted olefins.23 Allyl alcohol
Resu lts a n d Discu ssion
Ep oxid a tion of Allylic Alcoh ols. The epoxidation
of olefinic alcohols with group V catalysts using alkyl
hydroperoxide oxidants was discovered in 1968 by
Sheng and Zajacek16 and concurrently by Kollar.17
Sharpless and co-workers studied similar systems
throughout the 1970s and demonstrated the synthetic
utility of metal-catalyzed epoxidations, in part due to
their high regio- and stereoselectivity.18 This class of
reactions remains an important tool for synthetic chem-
ists and has been well-studied from a kinetic and
mechanistic standpoint. The importance of vanadium-
catalyzed epoxidations with alkyl hydroperoxide oxi-
dants, and the understanding that had resulted from
three decades of research on these reactions in organic
solvents, made this an attractive area of oxidation
catalysis to examine in CO2.
We focused our preliminary experiments on the
vanadium(V) catalyst VO(OiPr)3 (oxovanadium(V) tri-
isopropoxide), which was found to be highly soluble in
CO2. This vanadium complex catalyzed the epoxidation
of a wide range of allylic and homoallylic alcohols in
liquid and supercritical CO2 using tert-butyl hydroper-
oxide as an oxidant. Reactions at elevated pressure were
carried out in custom-built stainless steel high-pressure
reactors with sapphire windows to allow for visual
observation and with provisions for reagent addition and
sampling. These reactors, which have been described
elsewhere, can be safely operated at pressures from
ambient to 345 bar and temperatures from -40 to 150
°C.19 Typical epoxidation reactions were run with 1.27
mmol of olefin and 2.4 equiv of t-BuOOH in the presence
of 3.5 mol % of catalyst and an internal standard in 33
(19) Buelow, S.; Dell’Orco, P.; Morita, D. K.; Pesiri, D. R.; Birnbaum,
E.; Borkowsky, S.; Brown, G.; Feng, S.; Luan, L.; Morgenstern, D. A.;
Tumas, W. Green Chemistry: Frontiers in Benign Chemical Synthesis
and Processing; Anastas, P. T., Williamson, T. C., Eds.; Oxford
University Press: Oxford, U.K., 1998; p 265.
(20) Conversions and selectivities were measured by GC. Conver-
sions were calculated as the percentage of starting material that went
on to react (C/C0), and the selectivity was calculated as the percentage
of epoxide with respect to all products formed in a given reaction (Pi/
∑P). The GC yield can be calculated by multiplying the conversion and
selectivity values. The agreement of the GC yields with isolated yields
for several reaction products supports the accuracy of this quantifica-
tion method. For example, both the olefin and epoxide were isolated
after the epoxidation reaction of trans-2-hexen-1-ol in liquid CO2. The
conversion and selectivity of this reaction in liquid CO2 were both 99%
by GC (a 98% yield). The isolated yield for this reaction, in which the
epoxide was isolated as its acetate derivative, was 91%. Epoxide
products were isolated using a procedure adapted from that described
by Itoh and co-workers.42 The lower value of the isolated yield is likely
to be the result of losses during the isolation but supports the accuracy
of the GC yields to within 10%.
(21) Bartlett, P. A.; J ernstedt, K. K. J . Am. Chem. Soc. 1977, 99,
4829.
(22) Mihelich, E. D.; Daniels, K.; Eickhoff, D. J . J . Am. Chem. Soc.
1981, 103, 7690.
(23) Vanadyl acetate catalysts have been reported to cleave styrene
analogues to the corresponding aldehyde with yields of 85-98% using
H2O2 as an oxidant: Choudray, B. M.; Reddy, P. N. J . Mol. Catal. 1995,
103, L1.
(13) Haas, G. R.; Kolis, J . W. Organometallics 1998, 17, 4454.
(14) Birnbaum, E. R.; LeLacheur, R. M.; Horton, A. C.; Tumas, W.
J . Mol. Catal. A 1999, 139, 11. Nu, X.-W.; Oshima, Y.; Koda, S. Chem.
Lett. 1997, 1045.
(15) Haas, G. R.; Kolis, J . W. Tetrahedron Lett. 1998, 39, 5923.
(16) Sheng, M. N.; Zajacek, J . G. Canadian Patent 799 502, 1968.
(17) Kollar, J . U.S. Patent 3 350 422, 1968.
(18) Sharpless, K. B.; Michaelson, R. C. J . Am. Chem. Soc. 1973,
95, 6136.

4918 Organometallics, Vol. 18, No. 24, 1999
Pesiri et al.
c
Ta ble 1. Ep oxid a tion of Allylic a n d Hom oa llylic Alcoh ols w ith VO(OiP r )3/t-Bu OOH in Liqu id CO2
a
b
Corresponding triol balance of product. 80% benzaldehyde as major product. c Reactions were run for 24 h at 25 °C in a 33 mL
stainless steel reactor (103 bar). All reactions involved 1.27 mmol of olefin, 3.05 mmol of t-BuOOH (as a 5.6 M solution in decane), 44
µmol of catalyst (3.5 mol %), and 91 µmol of octafluoronaphthalene (internal standard). Reactions were depressurized into acetone.
Conversions and selectivities were determined by gas chromatography (GC) and products identified with authentic standards (when
available) and GC-MS.
(6) was the only allylic alcohol tested which did not
result in a conversion of greater than 80% (presumably
due to limited CO2 solubility of this polar substrate).
With the exception of the sterically hindered isopule-
gol substrate (12), all of the homoallylic alcohol sub-
strates were transformed to their corresponding epox-
ides with conversions of 99%. The selectivity for the
epoxidation of isopugol (12) was the lowest (74%), with
the triol formed as the predominant side product,
apparently from a hydrolysis reaction. cis- and trans-
3-Hexen-1-ol (10 and 11, respectively) were both epoxi-
dized with 99% conversion and 89% selectivity, sug-
gesting no effect of the isomer structure on the
epoxidation of these olefins, a result consistent with
literature reports.
Although there have been no reports of the kinetics
of VO(OiPr)3-catalyzed epoxidation reactions, there have
been a number of reports of other vanadium catalyst
precursors.3 Since all of these vanadium complexes are
likely to be precursors to a common V(V) catalytic
species, these reports are relevant to our work. The
kinetics of olefin oxidation using vanadium catalysts in
organic solvents dates back to the early 1970s, when
Sheldon and van Dorn discussed mechanistic details of
transition-metal-catalyzed reactions using H2O2 and
t-BuOOH.24 Gould and co-workers also reported a
kinetic analysis of the vanadium-catalyzed epoxidation
(24) Sheldon, R. A.; Van Doorn, J . A. J . Catal. 1973, 31, 427.

Vanadium-Catalyzed Epoxidations of Alcohols
Organometallics, Vol. 18, No. 24, 1999 4919
F igu r e 1. Proposed mechanism for the vanadium-catalyzed epoxidation of allylic alcohols.2
of olefins using VO(acac)2 as catalyst precursor.25 The
relative rates of several catalyst systems were reported
as well as several important aspects of the reaction
mechanism, including the role of t-BuOH in inhibiting
the epoxidation. More recently, the work of Modena and
co-workers has detailed the kinetics of VO(acac)2/t-
BuOOH-catalyzed epoxidation reactions of olefins in
organic solvents in which activation parameters were
used to measure the effect of the binding of the -OH
group in the transition state.26 Modena also reported
mechanistic details about the epoxidation of geraniol by
studying the VO(acac)2-catalyzed oxidation of olefin and
thioether substrates.27 The kinetics of the closely related
Ti-tartrate epoxidation catalyst system was reported
by Sharpless in 1991 with a detailed study of the kinetic
and mechanism aspects of the asymmetric epoxidation
of allylic alcohols.28 The proposed mechanism for the
vanadium-catalyzed epoxidation of allylic alcohols is
shown in Figure 1. Given the fact that there were no
results using VO(OiPr)3 as an epoxidation catalyst in
the literature, we set forth to investigate the kinetics
of this system in order to benchmark its reactivity. The
vanadium-catalyzed epoxidation of trans-2-hexen-1-ol
and cis-3-nonen-1-ol was studied in several organic
solvents in order to compare the reactivity in dense-
F igu r e 2. Plot of trans-2-hexen-1-ol and 3-propyloxirane-
methanol concentrations for epoxidation in liquid CO2 at
24 °C. A first-order fit of the data for the epoxidation
appears as an inset.
phase CO2. Kinetic experiments were run in a modified
version of the high-pressure reactor by fitting it with a
high-pressure six-port two-way switching valve, allow-
ing samples to be drawn at high pressures with minimal
disturbances to the reaction pressure and volume.19
When the reaction was carried out with 3.5 mol %
catalyst and a 1:1.6 ratio of olefin to t-BuOOH, overall
first-order kinetics were observed. A linear correlation
between reactant concentration and time was obtained
by a semilogarithmic plot, which was linear over 4 half-
lives (Figure 2).
(25) Gould, E. S.; Hiatt, R. R.; Irwin, K. C. J . Am. Chem. Soc. 1968,
90, 4573.
(26) Bortolini, O.; Di Furia, F.; Modena, G. J . Mol. Catal. 1982, 16,
61. Bonchio, M.; Conte, V.; Di Furia, F.; Modena, G. J . Org. Chem.
1989, 54, 4368.
(27) Cenci, S.; Di Furia, F.; Modena, G. J . Chem. Soc., Perkin Trans.
2 1978, 979.
(28) Woodard, S. S.; Finn, M. G.; Sharpless, K. B. J . Am. Chem.
Soc. 1991, 113, 106.
A rate law was established with respect to olefin,
vanadium, and alkyl hydroperoxide by conventional
kinetic analysis using a large molar excess of reagents

4920 Organometallics, Vol. 18, No. 24, 1999
Pesiri et al.
Ta ble 2. Ra te of Hom oa llylic Alcoh ol Ep oxid a tion
to achieve “constant” concentrations for the remaining
reagents. When the concentration of olefin was doubled
with a large molar excess of both oxidant and catalyst,
the rate constant doubled, confirming a first-order
dependence on olefin. Similar results were obtained for
experiments establishing the order of the reaction with
respect to t-BuOOH and catalyst, confirming that the
reaction is nearly first order for all three.29 A fourth
term is present in the rate law and corresponds to an
inverse dependence of t-BuOH, the reduced product of
tert-butyl hydroperoxide. This nonintegral denominator
likely arises from a term such as ka + kb[t-BuOH], which
describes the inhibitory role of tert-butyl alcohol in the
mechanism’s ligand dissociation step. An expression for
the measured rate law appears in eq 1.
in Va r iou s Solven ts a t 25 °Ca
R1 ) C5H12
R2 ) C2H4OHb
,
R1 ) C3H7,
R2 ) CH2OHc
k′
krel
(CO2)
k′
krel
(CO2)
solvent
(s-1 M-1
)
(s-1 M-1
)
dichloromethane
acetonitrile
toluene
carbon dioxide
carbon tetrachloride
n-hexane
30
18
17
9
5
3
3.3
2.0
1.9
1.0
0.6
0.3
252
176
158
146
1.7
1.2
1.1
1.0
tetrahydrofuran
34
0.2
-d[olefin] [epoxide]
a
Initial observed first-order rates were converted to second-
order rate constants. All reactions were carried out at 24 ( 0.1
°C. Rates measured in CO2 were carried out at 29 bar in high-
pressure reactors. Reactions involved [VO(OiPr)3] ) 1.47 µM (3.5
mol %), [t-BuOOH] ) 100 mM, and [olefin] ) 42 mM. Rates
reported are accurate to 10%. Activation parameters for the
epoxidation of trans-2-hexen-1-ol were 3.6 kcal mol-1 and -35 eu
for ∆H and ∆S, respectively. ∆H and ∆S values for cis-3-nonen-
1-ol were 2.5 kcal mol-1 and -38 eu, respectively. The enthalpies
of activation for both reactions were lower than those in literature
reports for cyclohexene epoxidation using VO(acac)2. b cis-3-Nonen-
1-ol. c trans-2-Hexen-1-ol.
rate )
)
)
dt
dt
k[olefin]1.0[t-BuOOH]0.95[VO(OiPr)3]0.98
[t-BuOH]0.4
(1)
At the initial reaction times used for our kinetic
studies, the dominance of the ka term in the denomina-
tor (i.e. low [t-BuOH]) simplifies the rate law to the form
expressed in eq 1. This rate law is consistent with that
proposed for the epoxidation of allylic alcohols by group
V metal centers in which olefin, oxidant, and catalyst
are each involved in a prebinding step that forms the
active catalytic species.2 An intramolecular oxygen atom
transfer from the bound t-BuOOH to the olefin results
in a bound epoxide and the reduced t-BuOH group.
Subsequent displacement of the product and alcohol by
olefin and 1 equiv of t-BuOOH completes the catalytic
cycle.
Rate constants were calculated for the epoxidation of
allylic and homoallylic alcohols in the following manner.
Apparent first-order rate constants, k (min-1), were
obtained from plots of ln(C/C0) vs time for the disap-
pearance of olefin. These values were then converted
into second-order rate constants (s-1 M-1) to give k′. At
the low levels of t-BuOH during initial conditions the
effect on the rate can be excluded from the calculation
of k′. Rates of epoxidation for two olefins in liquid CO2
and several organic solvents are listed in Table 2. The
rate constant, k′, for cis-3-nonen-1-ol in CO2 was deter-
mined to be 9 s-1 M-1, and the rate for the epoxidation
of trans-2-hexen-1-ol was 146 s-1 M-1at 25 °C.
of both allylic and homoallylic alcohols. The superiority
of chlorinated alkanes as solvents for metal-catalyzed
epoxidations has been reported by Sheldon.30 In general,
however, there was not a large rate dependence on the
solvent, a result not unexpected for such a nonpolar,
nonionic reaction. The rate of allylic alcohol epoxidation
using VO(OiPr)3 was higher that that for homoallylics,
a result consistent with literature reports that claim
allylic alcohols react roughly 5 times faster than homo-
allylic alcohols during epoxidation with vanadium and
titanium catalysts.31 The data in Table 2 suggest that
carbon dioxide does not inhibit the rate of the vanadium-
catalyzed epoxidation reaction; rather, the rates are
roughly in line with the polarity of the solvent for both
allylic and homoallylic alcohols. The solubility and
reactivity properties of dense-phase CO2 are commonly
compared to those of alkanes such as pentane and
hexane, but these results suggest that toluene or carbon
tetrachloride may be a better model for reactivity in
CO2, particularly for nonpolar reactions.
Several experiments under liquid and supercritical
CO2 conditions revealed that pressure had no measur-
able effect on the conversions or the selectivities of these
reactions, a result not surprising in such a high-density
(liquid CO2) regime. For example, at 103 bar and 25 °C
the rate from Table 2 for the epoxidation of trans-2-
hexen-1-ol is 146 s-1 M-1. At the same temperature and
a pressure of 345 bar the rate of the reaction is 151 s-1
M-1. Under liquid CO2 conditions (T ) 25 °C) there
appears to be no difference between the rates within
experimental error. When the same reaction was run
at 35 °C and 103 bar (above the critical point), the
reaction rate was 382 s-1 M-1. A pressure increase to
The reaction rate in CO2 is 3 times that in hexane
and roughly equal to that in carbon tetrachloride for
the homoallylic alcohol substrate. Reaction rates for this
substrate were measurably greater in dichloromethane,
acetonitrile, and toluene. The trend in reaction rates
was similar for the trans-2-hexen-1-ol substrate: dichloro-
methane, acetonitrile, and toluene reactions were faster
than those in CO2. The epoxidation of the allylic alcohol
in CO2 was faster than in tetrahydrofuran. Methylene
chloride gave the highest rates for epoxidation reactions
(29) A practical limitation was encountered with the use of t-BuOOH
as an anhydrous 5-6 M solution in decane. Reactions that required a
large excess of t-BuOOH were run with reduced quantities of olefin
and catalyst to avoid unacceptably high decane levels in the CO2. It is
important to point out that cosolvents in dense-phase fluids have been
shown to have a potentially large effect on the dense-phase media’s
properties and the extent of this effect, even at levels below 10%, may
be worthy of investigating.
(30) Sheldon, R. A.; Van Doorn, J . A.; Schram, W. A.; De J ong, A. J .
J . Catal. 1973, 31, 438.
(31) Dehnel, R. B.; Whitman, G. H. J . Chem. Soc., Perkin Trans. 1
1979, 4, 953.

Vanadium-Catalyzed Epoxidations of Alcohols
Organometallics, Vol. 18, No. 24, 1999 4921
Ta ble 3. Ep oxid a tion of tr a n s-2-Hexen -1-ol Usin g Su bstitu ted VO(a ca c)2 Liga n d sa
a
Reactions were run for 6 h at 24 °C in 33 mL high-pressure reactors. Each experiment used 3 mmol of olefin, 4 mmol of t-BuOOH,
and 75 µmol of catalyst.
345 bar at 35 °C resulted in a rate of 385 s-1 M-1. The
effect of pressure on the epoxidation reaction above the
critical point was negligible. Clearly these reactions are
not in the mass-transfer-limited regime. The lack of any
measurable pressure effect or any benefits in carrying
out these reactions above room temperature led to the
majority of these reactions being carried out under
liquid-CO2 conditions.
VO(a ca c)2. VO(acac)2 is known to be one of the most
effective epoxidation catalysts in organic solvents. We
did not observe high conversions in preliminary experi-
ments using VO(acac)2 in dense-phase CO2, however.
Reactions with VO(acac)2 were turbid and contained a
precipitate, suggesting poor solubility. In an attempt to
enhance the activity of the vanadium acetylacetonate
complex we examined fluorinated, less polar acetyl-
acetonate analogues which would impart higher solubil-
ity in dense-phase CO2.
The use of highly fluorinated structures to improve
chemistry in supercritical CO2 has been reported by
several researchers for reactions ranging from hydro-
formylations32 to metalloporphyrin-catalyzed autoxida-
tions.14 Leitner and co-workers increased the CO2
solubility of catalysts to improve chemical transforma-
tions with modified catalysts which incorporate perfluo-
roalkyl substituents to enhance solubility (and reactiv-
ity) in sc CO2.32 This example of a widely used ligand
system which was modified to improve its reactivity in
dense-phase CO2 may expand the field of homogeneous
catalysis in sc CO2 by overcoming solubility limitations.
Holmes and Carroll33 as well as Tumas and co-workers34
have also demonstrated the effectiveness of fluorinated
substituents on catalyst complexes for palladium-
catalyzed carbon-carbon bond-forming reactions in
supercritical CO2.
included reaction rates for vanadium-catalyzed epoxi-
dations of olefins using fluorinated acetylacetonates
with increased reactivity. Our experiments involved the
use of commercially available fluorinated acac com-
pounds, including 1,1,1-trifluoro-2,4-pentanedione (tfac),
6,6,7,7,8,8,8-heptafluoro-3,5-octanedione (hfac), and 5,5-
dimethylhexane-2,4-dione (dmac), to prepare catalyst
precursors by binding them to vanadyl salts in aqueous
solutions to form VO(acac)2 complexes according to
literature methods.35
The results from Table 3 demonstrate that increasing
the catalyst solubility increases the reactivity. The
highest conversion achieved for the VO(acac)2-catalyzed
epoxidation of trans-2-hexen-1-ol in liquid CO2 was 64%
at 24 °C with a selectivity of 89% in 6 h. The presence
of two methyl groups on the acac ligand increases the
conversion of this reaction to 91%. A further increase
in the conversion resulted when a fluoromethyl or
fluoroalkyl plus methyl was substituted on the acac. The
low conversion in the reaction with the unsubstituted
acetylacetonate ligand is likely attributable to observed
precipitates in the liquid CO2, an indication of limited
solubility. The relative rates for the epoxidation of trans-
2-hexen-1-ol using the difluoromethyl-substituted acac
ligand was 7.6 × 10-3 s-1, compared to 5.3 × 10-3 s-1
for the unsubstituted acac in CO2. Both the fluoro-
methyl- and fluoroalkyl-substituted acetylacetonate
ligands gave rates of 1.7 × 10-2 s-1 in liquid CO2. These
reactions were again roughly 3 times slower than
reactions run in dichloromethane under identical condi-
tions. The similarity in the initial rates of reactions
which exhibit different conversion levels is analogous
to a published report of palladium-catalyzed cross-
coupling reactions using several phosphine ligands.33,34
The higher conversions in both cases appear to arise
from catalyst solubility and suggest the substituted
acac’s role in enhancing the conversion is due to an
improvement in the CO2 solubility of the vanadium
complex.
Oxidation catalysis using fluorinated ligands for
increased reactivity in organic solvents actually dates
back three decades. In 1968 Gould and co-workers
studied the vanadium-catalyzed epoxidation of cyclo-
hexene using VO(acac)2 and t-BuOOH.35 This report also
P olym er -Su p p or ted Va n a d iu m Ca ta lysis in Liq-
u id CO2. The concept of attaching homogeneous cata-
lysts onto fixed organic or inorganic supports to offer
the separation advantages of a heterogeneous system
has received a great deal of attention during the past
two decades. For example, a number of researchers have
explored the potential for polymers to act as heteroge-
(32) Kainz, S.; Koch, D.; Baumann, W.; Leitner, W. Angew. Chem.,
Int. Ed. Engl. 1997, 36, 1628.
(33) Carroll, M. A.; Holmes, A. B. Chem. Commun. 1998, 13, 1395.
(34) Morita, D. K.; Pesiri, D. R.; Glaze, W. H. Chem. Commun. 1998,
13, 1397.
(35) Su, J . C.; Reed, J . W.; Gould, E. S. Inorg. Chem. 1973, 12, 337
and references therein.

4922 Organometallics, Vol. 18, No. 24, 1999
Pesiri et al.
neous supports for catalytic transformations,36 including
a number of vanadium-catalyzed epoxidations.37 We
were intrigued by the idea of using anionic polymers as
supports for cationic vanadyl cations for epoxidations
in dense-phase CO2. Two important aspects of polymer-
supported catalysis in dense-phase CO2 which compelled
our investigation were the extreme nonplanarity (and
correspondingly low solvation power) of CO2, making the
leaching of cationic catalysts less favorable, as well as
the potential for CO2 to swell certain polymer struc-
tures. We studied cationic vanadyl species ionically
exchanged onto anionic sites of two different polymer
structures, the commercially available Nafion (a per-
fluoronated alkane backbone with sulfonate functional-
ity), and Amberlite IRC50 (a polystyrene backbone with
carboxylate functionality).
catalytic oxidation transformations. In the case of the
VO(acac)2-catalyzed epoxidation of allylic alcohols, lim-
ited solubility of the catalyst or catalyst precursor
limited the reactivity. The use of highly fluorinated
analogues of the acetylacetonato ligands did increase
the reactivity of this system in what is believed to be
the first demonstration of enhanced oxidation catalysis
in a dense-phase fluid through improvements in catalyst
solubility. Ion exchange polymer supports did not offer
any significant control over vanadium catalyst leaching
for epoxidation reactions in liquid CO2, however. While
the reaction rates are similar in liquid CO2 and organic
solvents, a result that is not surprising for a nonpolar
reaction, we believe that other selective oxidations
deserve further investigation. The opportunity for en-
hanced catalysis in dense-phase carbon dioxide war-
rants an exploration of reaction systems that capitalize
on the unique properties favoring oxidation chemistry.
The two polymer-supported catalysts were examined
for the epoxidation of trans-2-hexen-1-ol in liquid CO2.
A Nafion-supported catalyst had a reaction rate (kobs
)
0.015 s-1) that was lower than that of the corresponding
homogeneous reaction using VO(OiPr)3 in liquid CO2
(0.126 s-1). The lower rates for polymer-supported
catalysts are likely due, in part, to a lower catalyst
loading. Reaction rates for the polymer-supported and
homogeneous reactions were equal when corrected for
catalyst concentration. The Nafion-supported catalyst
gave a conversion of 80% with a very low selectivity
(21%) during a 6 h reaction. Extraction experiments
identified the majority of products to be oligomeric
species intimately associated with the polymer back-
bone. The oligomer was the result of a ring-opening
reaction at the Lewis acidic sulfonate sites on the
polymer. To avoid the reactivity of the Lewis acid
polymer structure, an Amberlite-supported catalyst was
used for the epoxidation of trans-2-hexen-1-ol. Although
it exhibited a slightly lower reaction rate than the
Nafion system (kobs ) 0.0085 s-1), the epoxide product
was found to be stable. Conversions of 90% and selec-
tivities of epoxide of over 95% were achieved with this
system. Although epoxidation reactions were shown to
proceed using polymer-supported vanadium catalysts in
liquid CO2, the degree of catalyst leaching was signifi-
cant (greater than 3% of the catalyst loading leached
over a 6 h reaction).
Exp er im en ta l Section
High -P r essu r e Ba tch Rea ction s. All high-pressure reac-
tions were carried out in specially designed, custom-built 316
stainless steel reactors with sapphire windows to allow visual
inspection at high pressure. These reactors have been de-
scribed in detail elsewhere.4,19 In addition to the window
flange, these reactors were fitted with a pressure gauge, inlet
and outlet valves, a rupture disk, and a thermocouple. Reac-
tions were mixed with Teflon stirbars in situ using magnetic
stirplates. Glass ampules were used in all high-pressure
reactions as a means of introducing reagents and catalyst.
Constant temperatures of up to 90 °C were achieved with
heating tape wrapped around the outside of the reactor with
control through a temperature controller (Omega). A typical
epoxidation reaction with VO(OiPr)3 and t-BuOOH involved
loading the olefin (1.27 mmol), 2.6 equiv (3.3 mmol) of a 5-6
M solution of anhydrous alkyl hydroperoxide in decane (Ald-
rich), and octafluoronaphthalene (92 µmol, used as an internal
standard, Aldrich) into a 2 mL glass ampule which was
degassed (freeze-thaw cycling), flame-sealed under vacuum,
and placed in the reactor. tert-Butyl hydroperoxide was
obtained as a 5-6 M solution in decane (Aldrich). All reagents
were used as purchased, without further purification. The
peroxide solution was assayed for t-BuOOH using a titration
method published by Sharpless.38 Alkyl hydroperoxides were
found to be soluble in both liquid and supercritical CO2 (>23%
v/v) and were used in all of the epoxidation experiments.
Vanadyl triisopropoxide catalyst (44 µmol, 0.035 mol %, Strem)
was then added to the reactor in a glovebox. A sealed ampule
containing the balance of reactants and a dry stirbar were also
added to the reactor in the glovebox. The reactor was then
hand-tightened at the window flange using Teflon O-rings on
each side of the sapphire window. The reactor was removed
from the glovebox, tightened by wrench, and pressurized using
a high-pressure manifold system inside a fume hood. Ultra-
high-purity carbon dioxide (Matheson) was passed through
purification columns (an oxygen trap (Alltech), charcoal trap
(Alltech), a moisture trap (Alltech), and 8 Å molecular sieves)
to ensure purity. Upon pressurization with CO2, the ampule
imploded, resulting in the mixing of the catalyst with reagents,
marking the beginning of the reaction. Hydrostatic pressures
of 103 bar were chosen for liquid-CO2 reactions because this
is above the vapor pressure of CO2 (60.6 bar) at 25 °C and
results in homogeneous solutions in a single liquid-CO2 phase
in the pressure cell. At 25 °C, pressures between 61 and 76
bar result in a two-phase system with condensed (liquid) CO2
Con clu sion s
Our results clearly indicate that selective oxidation
reactions can be carried out in liquid CO2. The high
selectivities and conversions achieved in several vana-
dium-catalyzed epoxidation reactions in liquid carbon
dioxide are important evidence that this medium does
not limit the catalytic activity of electrophilic, peroxidic
transition-metal catalysts. In virtually every case the
epoxide yield was comparable to reactions run in organic
solvents such as n-hexane, methylene chloride, and
toluene. Rates measured for VO(OiPr)3-catalyzed ep-
oxidation of olefinic alcohols fall squarely in the range
of the organic solvent reactions examined. This result
suggests the validity of considering dense-phase CO2 as
a potential organic solvent alternative for homogeneous
(36) Bergbreiter, D. F. Catal. Today 1998, 42, 389.
(37) Sherrington, D. C. Pure Appl. Chem. 1988, 60, 401. Yokoyama,
T.; Nishizawa, M.; Kimura, T.; Suzuki, T. M. Bull. Chem. Soc. J pn.
1985, 58, 3271.
(38) Gao, Y.; Hanson, R. M.; Klunder, J . M.; Ko, S. Y.; Masamune,
H.; Sharpless, K. B. J . Am. Chem. Soc. 1987, 109, 5765.

Vanadium-Catalyzed Epoxidations of Alcohols
Organometallics, Vol. 18, No. 24, 1999 4923
in contact with a CO2 vapor above it. Two-phase reaction
conditions are sometimes desirable due to the low pressures
associated with these liquid-CO2 reactions, but the presence
of the liquid- and vapor-CO2 phases can introduce complica-
tions in homogeneous catalysis reactions due to uncertainty
in reaction volumes. In summary, liquid-CO2 pressures of 103
bar were chosen as the lowest possible pressure that was still
safely in the single-CO2-phase region.
H2SO4 was used.40 This solution was heated to 75 °C for 20
min, at which time the precipitate was filtered. To the filtrate
was added 1 g of 2,2-dimethyl-6,6,7,7,8,8,8-heptafluoro-3,5-
octanedione (Aldrich) dropwise. The mixture was then neu-
tralized with 5% Na2CO3 to precipitate the complexes, which
were recrystallized from benzene and dried in vacuo over P2O5.
Acetyla tion of Allylic a n d Hom oa llylic Alcoh ols a n d
E p oxy Alcoh ols. The epoxides were difficult to isolate
directly; therefore, their acetate derivatives were prepared for
isolation in the following manner. The reactor contents were
depressurized and collected in 20 mL of acetone, and this
solution was then used to rinse the reactor as described earlier.
The acetone solution was then concentrated by rotovaporation
at room temperature to roughly 5 mL. The acetylation was
carried out by adding 1.5 mmol of triethylamine, 1.5 mmol of
acetic anhydride, and 0.08 mmol of 4-(dimethylamino)pyridine
for every 1 mmol of analyte to be derivatized. The deep yellow
solution was stirred in a 100 mL round-bottom flask, during
which time the acetylation solution turned a deep gray-green
color and evolved a significant amount of heat. The reaction
was complete within 6 h. The entire solution was diluted with
10 mL of diethyl ether and washed in a 250 mL separatory
funnel with 1 M HCl (1 × 20 mL). The acetlyation products
were then extracted with an additional volume of 20 mL of
diethyl ether. The organic layer was subsequently washed in
succession with saturated aqueous solutions of the following:
NaHSO3 (1 × 15 mL), NaHCO3 (1 × 20 mL), and NaCl (2 ×
20 mL). The acetylated product was dried over MgSO4 and
filtered using filter paper and a glass wide-mouth funnel. The
organic filtrate was concentrated in vacuo using a rotavap,
yielding an oil which was transferred to a sealed Schlenk
storage flask. Thin-layer chromatography with 1:1 hexane-
ethyl acetate gave good separation of the products with an
acceptable Rf value. The acetylated product was separated by
flash chromatography. A 62 cm glass column, 2.5 cm width
(Aldrich), was filled with silica gel (Merck, grade 9385, 230-
400, mesh 60 Å). Roughly 5 mL of the acetylated oil was added
to a sea sand plug at the top of the column, and the hexane-
ethyl acetate mobile phase was forced through with a flow rate
of approximately 5 mL/min. Ten milliliter fractions were
collected and stored. Each fraction was analyzed by GC-MS,
and those containing pure acetylated product were combined
and concentrated in vacuo by rotavaporation. The pure product
(as an oil) was weighed and a portion solubilized in acetone
for GC-MS analysis; another portion was solubilized in benzene-
Batch reaction contents were collected by venting the CO2
pressure into a glass vial filled with an organic solvent (20
mL of acetone, hexane, or ethyl acetate). Routine analyses were
accomplished by gas chromatography (GC) using a Hewlett-
Packard 5890 Series II instrument with FID detector and 30
m capillary (1.5 µm i.d.) DB 624 column (Alltech). Calibration
curves with commercially available standards were used when
available to quantify conversions and selectivities. Reaction
products were identified using a Hewlett-Packard 5971 A
instrument with mass-selective detector (GC MS) and 30 m
capillary DB 624 column (1.5 µm i.d.) as well as 1H and 13C
NMR for isolated products with and without derivatization.
Kin etics. Kinetic experiments in CO2 were carried out in
identical 33 mL stainless steel reactors which had been
modified for sampling. A Hastelloy high-pressure (344 bar) six-
port two-way switching valve (Valco Inc.) was used to take
samples from the reactor at high pressure. A 100 µL sample
loop was used in all experiments to provide a uniform sample
size. The contents of the loop were vented into a dry 1.8 mL
glass GC vial, the loop was rinsed with solvent via syringe,
and the rinsing solution was collected in the same GC sample
vial. The vial was then sealed and analyzed directly by GC.
The use of an internal standard enabled quantitative analysis
of the olefin and epoxide with good reproducibility and
accuracy. Reactions in organic solvents were carried out in 100
mL glass three-necked round-bottom flasks fitted with a
condenser. Molecular sieves (4 Å, 8-12 mesh, ca. 100) were
added, and VO(OiPr)3 catalyst was added via syringe to the
flask. The round-bottom flask was connected to nitrogen via
a Schlenk line. Samples were taken via a syringe needle and
were quenched with acetone and analyzed by GC. The epoxides
were difficult to isolate directly; therefore, their acetate
derivatives were prepared for isolation by the method de-
scribed by Itoh and co-workers.39 Second-order rate constants
were calculated by dividing the apparent first-order rate
constant by 60 (seconds per minute) and by the concentrations
of catalyst (3.9 × 10-3 M), t-BuOOH (136 mM), and olefin (48
mM) to give k′.
1
d6 for H NMR analysis. The analysis of the acetylated product
confirmed the presence of high-purity material.
P r ep a r a t ion of t h e F lu or in a t ed Va n a d iu m -Acet yl-
a ceton a te Com p lexes. Literature methods were used for the
synthesis of the vanadium complexes of the fluorinated acetyl-
acetonate ligands.35 Vanadyl trifluoroacetylacetonate (VO-
(tfac)) was prepared by dissolving 1.0 g of VOSO4‚4H2O (Strem)
in 5 mL of deionized water, adjusted to pH 3 with the addition
of aqueous Na2CO3, followed by the addition of 1.6 mL of 1,1,1-
trifluorohexane-2,4-dione. During the addition of the ligand
the pH was kept at 3.0 with the addition of a 5% aqueous
Na2CO3 solution. The mixture of green and white precipitates
was filtered and washed with deionized water to remove white
precipitate, and the crude green product was recrystallized
from benzene and dried under vacuum. The product was stored
cold, under N2. Vanadyl dimethylacetylacetonate was also
prepared by this aqueous method, in which 1 g of VOSO4.4H2O
and 250 µL of 5,5-dimethylhexane-2,4-dione were combined
at pH 3.0. The complex was recrystalized from benzene and
dried as above. Vanadyl heptafluorodimethylacetylacetonate
could not be obtained in high yield with the previous method;
therefore, a procedure in which 1 g of V2O5 was added to 4
mL of ethanol and 5 mL of water with 0.5 mL of concentrated
P olym er -Su p p or ted Ca ta lysis. Nafion-Na+ (purchased as
Nafion-H+ as Nafion NR 50 beads, Aldrich) was washed in a
solution of pH 5 aqueous NaOH (200 mL, 0.05 mol NaOH) and
stirred until the pH dropped to a constant level. The beads
were washed in deionized water, and the NaCl washing
process was repeated. The beads were then washed a third
time with a stronger NaCl solution (200 mL, 0.7 mol NaOH)
and washed once the pH leveled off. Nafion-K+ was prepared
using an identical method involving KOH. Vanadium was
exchanged onto the polymer in several different solvents,
including water, methanol, and 2-propanol. These methods
were evaluated, and the most effective was determined to use
vanadyl sulfate (VOSO4‚3H2O, Fisher) because of its simplicity
(other methods required inert atmospheres) and stable van-
adyl-polymer product. Quantification of the vanadium on the
polymer support was achieved by analyzing the preparation
solution after the exchange and subtracting the difference from
the quantity added at the beginning of the reaction. This
method of subtraction required an accurate quantification of
the hydrated vanadium salt added to the exchange solution.
The hydration number of the VOSO4 was measured by
dissolving 0.1 g of VOSO4 (Fisher Lot No. 964622) in 30 mL of
(39) Itoh, T.; J itsukawa, K.; Kaneda, K.; Teranishi, S. J . Am. Chem.
Soc. 1979, 101, 159.
(40) Rowe, R.; J ones, M. M. Inorg. Synth. 1957, 5, 114.

4924 Organometallics, Vol. 18, No. 24, 1999
Pesiri et al.
deionized water. ICP analysis of vanadium yielded a concen-
tration of 742 µg of V/g of solution, or 0.003 473 mol of
vanadium/g of VOSO4‚nH2O, where n ) 4.93. This hydration
number was constant over time and was used in all vanadium
sulfonate measurements. The vanadyl cation was exchanged
with the Nafion-Na by combining Nafion-Na (12 g) with a
solution of vanadyl sulfate in deionized water (830 mg of
VOSO4, 100 mL of H2O). This aqueous mixture was stirred
while the concentration of vanadium was followed by UV-vis
spectroscopy. When the concentration in solution indicated the
desired vanadium loading on the polymer, the beads were
filtered and rinsed with deionized water. The dried beads were
sealed under an atmosphere of argon to prevent deactivation
(exposure of the Nafion beads to air turned them a black-brown
color and inactivated them). Amberlite exchange with VO2+
was accomplished in an identical manner. Leaching studies
were carried out in hexane as an analogue to CO2. The loading
of vanadium on a typical polymer was 0.1656 g of V/g of
Amberlite IRC 50. Polymer-supported catalytic reactions were
run by adding octafluoronaphthalene (0.09 g, internal stand-
ard) and the catalyst beads (equivalent of ∼0.009 mmol of
vanadium) to the high-pressure batch reactor. tert-Butyl
hydroperoxide (7:1 molar ratio of tBHP to trans-2-hexen-1-ol)
and trans-2-hexen-1-ol (2.1 mmol) were added to a glass
ampule that was evacuated, sealed, and added to the reactor.
Kinetic analyses were performed in a manner identical with
that described for VO(OiPr)3-catalyzed reactions.
Sa fety Note. There are inherent risks in conducting reac-
tions at elevated pressures, although the danger can be
managed with the appropriate design and procedural practices.
No research of this type should be conducted in the absence
of the necessary safety precautions. All reactions were carried
out in a fume hood which was modified with 1/2 in. polycar-
bonate blast shielding. Reactors at high pressures were fitted
with pressure relief valves or rupture disks and were fixed
into place with a steel framework to reduce the risk of personal
injury.4141
Ack n ow led gm en t. This work was supported as part
of the Los Alamos Catalysis Initiative by the Depart-
ment of Energy through a Laboratory Directed Research
and Development (LDRD) grant and a grant from the
U.S. EPA, Office of Pollution Prevention and Toxics
(Grant Number 1877-97-1). We acknowledge many
helpful discussions with R. Tom Baker, William Glaze,
and Steve Buelow. We also thank Phil Dell’Orco for
assistance on preliminary polymer support work.
OM990467M
(41) Clavier, J . Y.; Perrut, M. In High-Pressure Chemical Engineer-
ing, Process Technology Proceedings; von Rohr, P. R., Trepp, C., Eds.;
Proceedings from the 3rd International Symposium on High-Pressure
Chemical Engineering; Elsevier Science: Amsterdam, 1996; Vol. 12,
p 627.