Allylic oxidations of terminal olefins using a palladium thioether catalyst

Article abstract of DOI:10.1021/ol902905w

"Chemical equation presented" A palladium catalyst that converts terminal olefins to linear allylic acetates at lower catalyst loadings and faster reaction times than current systems is reported. This reaction can be conducted using benzoquinone as the oxidizing agent or catalytic amounts of copper and hydroquinone under one atmosphere of oxygen. Preliminary reactivity studies of π-allylpalladium complexes under our reaction conditions do not provide results similar to those obtained in the catalytic reaction, which may suggest an alternative reaction pathway. The palladium catalyst is ligated by an aryloxyalkyl aryl sulfide, which is identified as a new ligand for homogeneous catalysis.

Full text of DOI:10.1021/ol902905w

ORGANIC
LETTERS
2010
Vol. 12, No. 4
824-827
Allylic Oxidations of Terminal Olefins
Using a Palladium Thioether Catalyst
William H. Henderson, Christopher T. Check, Nicolas Proust, and James
P. Stambuli*
Department of Chemistry, The Ohio State UniVersity, 100 West 18th AVenue,
Columbus, Ohio 43210
stambuli@chemistry.ohio-state.edu
Received December 17, 2009
ABSTRACT
A palladium catalyst that converts terminal olefins to linear allylic acetates at lower catalyst loadings and faster reaction times than current
systems is reported. This reaction can be conducted using benzoquinone as the oxidizing agent or catalytic amounts of copper and hydroquinone
under one atmosphere of oxygen. Preliminary reactivity studies of π-allylpalladium complexes under our reaction conditions do not provide
results similar to those obtained in the catalytic reaction, which may suggest an alternative reaction pathway. The palladium catalyst is ligated
by an aryloxyalkyl aryl sulfide, which is identified as a new ligand for homogeneous catalysis.
The functionalization of allylic C-H bonds via selective
oxidation reactions has the potential to greatly impact the
synthesis of complex molecules.1 Traditionally, allylic oxida-
tions have been mediated by selenium2a and copper,2b,c but
functional group tolerance and low selectivities are prob-
lematic for these metals. Other metals such as mercury2d and
rhodium2e have also been investigated. Over four decades
ago, a stoichiometric Pd(OAc)2-promoted oxidation of the
allylic C-H bond of 1-butene in neat AcOH was reported
to give the corresponding vinyl acetate as the major product
along with 9% of the linear allylic acetate.3 However, the
addition of DMSO provided an 8-fold increase of linear
allylic acetate. Although much work has been done on allylic
oxidations of disubstituted olefins since these findings,4 few
selective methods for allylic oxidation reactions of terminal
olefins to allylic acetates have been reported.5,6 White and
co-workers showed that palladium(II) complexes ligated by
sulfoxides catalyzed the conversion of terminal alkenes to
either linear or branched allylic acetates with very good
selectivities.6 However, some of the challenges that remain
include high catalyst loadings, long reaction times, and the
requirement of excess benzoquinone (BQ). Herein, we
present a unique aryloxyalkyl aryl sulfide palladium catalyst
that converts terminal olefins to linear allylic acetates in good
yields and regioselectivities. This allylic oxidation reaction
(1) Stang, E. M.; White, M. C. Nat. Chem. 2009, 1, 547.
(2) (a) Umbreit, M. A.; Sharpless, K. B. J. Am. Chem. Soc. 1977, 99,
5526. (b) Kharasch, M. S.; Sonsnovsky, G. J. Am. Chem. Soc. 1958, 80,
756. (c) Andrus, M. B.; Lashley, J. C. Tetrahedron 2002, 58, 845. (d)
Rappoport, Z.; Sleezer, P. D.; Winstein, S.; Young, W. G. Tetrahedron
Lett. 1965, 42, 3719. (e) Catino, A. J.; Forslund, R. E.; Doyle, M. P. J. Am.
Chem. Soc. 2004, 126, 13622.
(3) (a) Kitching, W.; Rappoport, Z.; Winstein, S.; Young, W. G. J. Am.
Chem. Soc. 1966, 88, 2054. Trost first isolated allylic C-H activation
products from alkene insertions into Pd: (b) Trost, B. M.; Strege, P. E.
Tetrahedron Lett. 1974, 30, 2603. (c) Trost, B. M.; Metzner, P. J. J. Am.
Chem. Soc. 1980, 102, 3572.
(5) (a) McMurry, J. E.; Kocovsky, P. Tetrahedron Lett. 1984, 25, 4187.
(b) Mitsudome, T.; Umetani, T.; Nosaka, N.; Mori, K.; Mizugaki, T.; Ebitani,
K.; Kaneda, K. Angew. Chem., Int. Ed. 2006, 45, 481. (c) Lin, B.-L.;
(4) (a) Åkermark, B.; Larsson, E. M.; Oslob, J. D. J. Org. Chem. 1994,
59, 5729. (b) Grennberg, H.; Gogoll, A.; Ba¨ckvall, J.-E. J. Org. Chem.
1991, 56, 5808. (c) Hansson, S.; Heumann, A.; Rein, T.; Åkermark, B. J.
Org. Chem. 1990, 55, 975.
Labinger, J. A.; Bercaw, J. E. Can. J. Chem. 2009, 87, 264.
(6) (a) Chen, M. S.; White, M. C. J. Am. Chem. Soc. 2004, 126, 1346.
(b) Chen, M. S.; Prabagaran, N.; Labenz, N. A.; White, M. C. J. Am. Chem.
Soc. 2005, 127, 6970
.
10.1021/ol902905w  2010 American Chemical Society
Published on Web 01/25/2010

operates at decreased catalyst loadings, in shorter reaction
times and without the need for excess benzoquinone.
In addition to the synthetic utility of oxidations of allylic
C-H bonds,7 we were attracted to the mechanism proposed
for the palladium-catalyzed allylic oxidation reactions using
sulfoxide ligands.6b Moreover, our group has a specific
interest in the interaction of late transition metal complexes
with sulfur,8 which has been wrongly considered to operate
solely as a poison in the presence of palladium. Using
1-dodecene as the test substrate, we investigated palladium-
catalyzed allylic oxidation reactions using bidentate or
hemilabile bidentate sulfur- and nitrogen-based ligands. We
hypothesized that the tighter binding of the sulfur- and
nitrogen-based ligands (relative to the sulfoxide ligands) may
increase the overall reaction rate while still favoring the
branched regioisomer. We reasoned that since the weaker
binding DMSO ligand favored the linear product3a,6a and
the stronger binding bis-sulfoxide ligand favored the branched
product6b ligands that bound more tightly than bis-sulfoxides
may also favor the branched product. The tighter binding
would increase the overall reaction rate because of the
probability that the ligand would spend more time at the
metal center. Phosphines were not employed because both
McMurry5a and White6b have shown that phosphines increase
the overall reaction rate but provide approximately 1:1
mixtures of branched and linear allylic acetate products.
The use of 10 mol % Pd(OAc)2 in combination with
nitrogen ligands 1-3 or sulfur ligand 4 failed to produce
more than trace amounts of allylic acetate products by GC
(Table 1). However, 1,2-bis(phenylthio)ethane (5) and
optimal solvent as 5 mol % Pd(OAc)2 and 6 gave 85%
conversion of 1-dodecene to products A-D with a 23:1 ratio
of linear to branched products, while suppressing side
products C and D (Supporting Information). The use of
Pd(OAc)2/6 allowed for lower catalyst loadings, which
lessened the amount of side products formed. Our hypothesis
was partially correct: the reaction rate increased, but the
favored product was the linear allylic acetate, not the
branched. To our knowledge, this class of ligand has never
been used in any transition-metal-catalyzed reactions.
With optimized conditions in hand, the scope of this
catalyst system was probed (Table 2). All of the compounds
Table 2. Scope of Allylic Oxidation Reactions Using 5 mol %
of Pd(OAc)2 and 6
Table 1. Ligands Tested in Pd-Catalyzed Allylic Oxidation of
1-Dodecene
a
1
Isolated yields shown are of only the (E) isomer (>95% by H NMR
spectroscopy) and are an average of two (1 mmol) reactions. b Product is
a 2:1 mixture of linear and branched isomers. c Yield is a mixture of 10:1
(E:Z) isomers and contains a trace of an unidentified side product.
Pd(OAc)2 gave 65% conversion of 1-dodecene to allylic
acetates A and B (40:1), along with trace amounts of vinyl
acetate C and methyl ketone D. After this result, we quickly
discovered that a catalyst formed from 2-tolyloxylethyl
phenyl sulfide (6) and Pd(OAc)2 completely converted
1-dodecene to A-D in a 19:2:3:1 ratio, respectively. Further
optimization of this catalyst found acetic acid to be the
1
in the table were isolated as the pure (>95% by H NMR
spectroscopy) linear (E)-allylic acetate unless noted, and
the yields are an average of two reactions. The reaction
times were optimized to favor the greatest conversion with
the least amount of branched allylic acetate, vinyl acetate,
and ketone side products (Table 1, B-D). Nonetheless,
all of the alkenes underwent g85% conversion in all
reactions. Most of the reactions were highly selective for
(7) (a) Fraunhoffer, K. J.; Bachovchin, D. A.; White, M. C. Org. Lett.
2005, 7, 223. (b) Fraunhoffer, K. J.; Prabagaran, N.; Sirois, L. E.; White,
M. C. J. Am. Chem. Soc. 2006, 128, 9032.
(8) (a) Eichman, C. C.; Stambuli, J. P. J. Org. Chem. 2009, 74, 4005.
(b) Lee, K.; Counceller, C. M.; Stambuli, J. P. Org. Lett. 2009, 11, 1457.
Org. Lett., Vol. 12, No. 4, 2010
825

linear allylic oxidation products. Allylic (entries 1-3) and
homoallylic (entry 4) arenes provided the corresponding
linear products in good yields regardless of the substitution
present on the arene. The reaction conditions also tolerated
amides, carbamates, esters, ethers, and acetals (entries
6-12). The only substrate that did not provide good L:B
selectivity contained an amide, which likely coordinates
to the metal center and disrupts the Pd-S dative bond
present in the high selectivity catalyst in these reactions
(entry 6). The low yield obtained for the PMB-ester (entry
9) may be a result of some type of oligomerization or
other side reaction as g90% conversion of the alkene was
observed. The GC spectrum contained only minor amounts
of the corresponding vinyl acetate and methyl ketone
products, and attempts to isolate any other major products
were fruitless. In a separate experiment, 1-dodecene was
submitted to the catalytic conditions in the absence of
benzoquinone or air, and <5% of the terminal alkene was
isomerized to 2-dodecene. Therefore, it is not likely that
the unknown side products are emanating from oxidation
products formed from the corresponding internal alkene.
The oxidation of sulfides to sulfoxides is facile, and
thus it is reasonable to believe that the reaction conditions
employed may oxidize sulfide ligand 6 to the correspond-
ing sulfoxide ligand 9 and that 9 binds to palladium to
form the active catalyst. Sulfide ligand 6, not sulfoxide
ligand 9, is observed in the GC spectrum upon analysis
of the oxidation reaction under our standard conditions
shown in Table 2. Nonetheless, we prepared ligand 9 and
tested its reactivity with Pd(OAc)2 under optimal reaction
conditions. The major product of this reaction was the
corresponding vinyl acetate, which was isolated in 46%
yield. The reaction did not go to completion as ap-
proximately 30% of the starting alkene was observed in
the GC spectrum (Scheme 1).
Table 3. Pd-Catalyzed Allylic Oxidations Without Excess BQ
a
1
Isolated yields shown are of only the (E) isomer (>95% by H NMR
spectroscopy) and are an average of two (1 mmol) reactions.
to or slighly decreased in comparison to the yields
obtained using 2 equiv of BQ shown in Table 2.
The mechanism of this process is unclear at this stage of
our research. Although Pd complexes coordinated to mixed
(S,O) ligands have been previously isolated,10 aryloxyalkyl
aryl sulfide palladium complexes have not. Observation of
the interaction of 6 with Pd(OAc)2 showed coordination of
6 to Pd by 1H NMR spectroscopy (Supporting Information);
however, any attempts to isolate the complex that gave rise
to these resonances only produced intractable oils. The
expected coordination of ligand 6 to palladium is mainly
through sulfur with the aryl ether providing only moderate
support as a potential hemilabile ligand. Surprisingly, we
were able to isolate 10 Pd[PhS(CH2)2OAr]2Cl2 (Ar )
p-CH3C6H4) (Figure 1). The X-ray structure of this complex
Scheme 1. Allylic Oxidation with Sulfoxide Version of 6
The requirement of excess BQ is one drawback to this
catalytic process. For example, allylic oxidation of 10 g
of 4-phenyl-1-butene requires approximately 16 g of BQ.
We recently discovered that our reaction conditions can
be modified to replace the 2 equiv of BQ with a catalytic
amount of copper and dihydroquinone (DHQ) under 1 atm
of O2.9 The catalyst loading had to be increased to 10
mol % in Pd(OAc)2 and ligand (6). However, these
conditions allowed the production of the corresponding
linear allylic acetates from a range of terminal alkenes
(Table 3). The products shown in the table are of the (E)
isomer. The yields of linear allylic acetates are equivalent
Figure 1
probability.
. ORTEP structure of 10 shown at 50% ellipsoid
showed the ligands bound solely through sulfur with no close
contact between the metal center and oxygen atoms. Submis-
sion of this complex to the standard reaction conditions did
not provide any allylic acetate product until 10 was pretreated
with 2 equiv of AgOAc (Scheme 2). The yield of this
reaction is lower than observed under the standard reaction
(9) Ba¨ckvall, J.-E.; Hopkins, B. R.; Grennberg, H.; Mader, M. M.;
Awasthi, A. K. J. Am. Chem. Soc. 1990, 112, 5160.
(10) Esterhuysen, M. W.; Brull, R.; Raubenheimer, H. G.; Esterhuysen,
C.; Kruger, G. J. J. Organomet. Chem. 2001, 619, 164.
826
Org. Lett., Vol. 12, No. 4, 2010

expected allylic acetate in a 20:1 (L:B) ratio as observed in
the catalytic reaction; rather, a 1:1 mixture of the linear and
branched products was observed. The addition of 1 equiv of
ligand 6 to π-allylpalladium complex 11 provided <4% of
allylic acetate products, while the addition of 40 equiv of
BQ favored the branched product in a 3:1 (B:L) ratio. The
preliminary results obtained using this π-allylpalladium dimer
along with the ability to conduct this reaction without excess
benzoquinone suggests that the allylic oxidation of 1-dodecene
under our reaction conditions does not proceed via this
specific π-allylpalladium complex.
Under the optimized catalytic conditions shown in Table
2, isolated linear allylic acetate (48:1) L:B deteriorates to a
30:1 (L:B) ratio of allylic acetate over 12 h.13 Under these
identical conditions, isomerically pure branched allylic
acetate is transformed to a mixture of branched and linear
allylic acetates in a 2:1 (B:L) ratio over 12 h. However,
monitoring the reaction progress of the allylic oxidation of
1-dodecene (Table 2, entry 5) by GC shows that the reaction
does not begin to produce branched product until about 3 h
when a 58:1 (L:B) ratio is observed in the GC spectrum.
This high L:B ratio decreases to 43:1 after 5 h, 34:1 after
8 h, and ultimately around 20:1 after 12 h. Therefore, we do
not believe that the catalyst is first producing branched allylic
acetate product that is quickly isomerized to the linear allylic
acetate.
Scheme 2
.
Allylic Oxidations Using an Isolated Bisthioether
Palladium Complex
conditions that consist of a 1:1 ratio of metal to ligand.
Complex 10 contains two thioether ligands on palladium,
and the use of a 1:2 ratio of metal to ligand under the typical
conditions shown in Table 2 provides the (E)-linear allylic
acetate product in 50% yield. The slightly lower yield
observed in Scheme 2 may be caused by insufficient
formation of the L2Pd(OAc)2 (L ) 6) complex. The origin
of this ligand effect is not known. However, when catalytic
transformations proceed more slowly in the presence of
excess ligand rather than an equimolar amount of ligand, it
is sometimes because the presence of excess ligand retards
the rate of the ligand dissociation step that may be a
prerequisite for catalyst activation.11 Therefore, the active
catalyst in this system likely requires one ligand per
palladium.
Palladium-catalyzed allylic oxidation reactions are ex-
pected to proceed via π-allylpalladium intermediates or
through a mechanism that begins with an initial acetoxyp-
alladation step.12 To begin investigations into the reaction
pathway, we prepared a π-allylpalladium acetate complex
of dodecene (11) (Scheme 3). This complex was reacted with
In summary, we have demonstrated a palladium-catalyzed
allylic oxidation of terminal olefins using a unique thioether
catalyst that occurs at lower catalyst loadings, faster reaction
times than previous systems, and without the need for
superstoichiometric benzoquinone. Further studies involving
the use of this catalyst in other processes as well as
developing a clear mechanistic picture of this allylic oxida-
tion reaction continue.
Scheme 3. Intermediacy of a π-Allylpalladium Acetate Dimer
Acknowledgment. This work was generously supported
by The Ohio State University. We thank Dr. Judith Gallucci
(OSU) for the X-ray structure of complex 10.
Supporting Information Available: Experimental pro-
cedures and characterization of all compounds. This material
is available free of charge via the Internet at http://pubs.acs.org.
40 equiv of BQ and 1 equiv of 6 in AcOH at 40 °C for 4 h
to mimic the actual reaction conditions.6b Observation of the
crude reaction mixture by GC analysis did not reveal the
OL902905W
(11) Stambuli, J. P.; Buhl, M.; Hartwig, J. F. J. Am. Chem. Soc. 2001,
123, 2677.
(13) For a potential mechanism of this process, see: Marion, N.;
Gealageas, R.; Nolan, S. P. Org. Lett. 2007, 9, 2653, and references
contained therein.
(12) Grennberg, H.; Ba¨ckvall, J.-E. Chem.sEur. J. 1998, 4, 1083, and
references contained therein.
Org. Lett., Vol. 12, No. 4, 2010
827