Chromium - Salen-mediated alkene epoxidation: A theoretical and experimental study indicates the importance of spin-surface crossing and the presence of a discrete intermediate

Article abstract of DOI:10.1002/1521-3765(20020916)8:18<4299::AID-CHEM4299>3.0.CO;2-B

The mechanism of alkene epoxidation by chromium(v) oxo salen complexes has been studied by DFT and experimental methods. The reaction is compared to the closely related Mn-catalyzed process in an attempt to understand the dramatic difference in selectivit

Full text of DOI:10.1002/1521-3765(20020916)8:18<4299::AID-CHEM4299>3.0.CO;2-B

FULL PAPER
Chromium Salen-Mediated Alkene Epoxidation:
A Theoretical and Experimental Study Indicates the Importance of
Spin-Surface Crossing and the Presence of a Discrete Intermediate
Peter Brandt,*[a] Per-Ola Norrby,[b] Adrian M. Daly,[c] and Declan G. Gilheany*[c]
Abstract: The mechanism of alkene
epoxidation by chromiumꢀv) oxo salen
complexes has been studied by DFT and
experimental methods. The reaction is
compared to the closely related Mn-
catalyzed process in an attempt to un-
derstand the dramatic difference in
selectivity between the two systems.
Overall, the studies show that the reac-
tions have many similarities, but also a
few critical differences. In agreement
with experiment, the chromium system
requires a change from low- to high-spin
in the catalytic cycle, whereas the man-
ganese system can proceed either with
spin inversion or entirely on the high-
spin surface. The low-spin addition of
metal oxo species to an alkene leads to
an intermediate which forms epoxide
either with a barrier on the low-spin
surface or without a barrier after spin
inversion. Supporting evidence for this
intermediate was obtained by using
vinylcyclopropane traps. The chro-
miumꢀv) oxo complexes can adopt a
stepped shape or form a more concave
surface, analogous to previous results on
manganese salen complexes.
Keywords: chromium
¥
density
functional calculations ¥ epoxidation
¥ manganese ¥ N,O ligands
-
Introduction
NO3
Chiral, nonracemic manganese[1] and chromium[2] oxo salen
complexes such as 1 and 2 have been used successfully for
asymmetric epoxidation. Complex 1 (with or without chloride
counterion, vide infra) is now the accepted reactive inter-
mediate in epoxidation with Jacobsen×s catalyst[1] and has
been studied by combined mass spectrometric and theoretical
methods,[3] EPR spectroscopy, and other techniques.[4] Com-
plex 2 is the best reagent within the chromium salen series.[2e]
O
O
N
O
N
O
N
O
N
O
Cr
L
Mn
tBu
tBu
Cl
tBu
tBu
CF3
F3C
1
2
Although they share many similarities, there are also some
substantial differences in the characteristics of the two
reagents. The manganese-based system shows a wider spec-
trum of reactivity[5] and therefore has generally been found to
be more useful. On the other hand, the stoichiometric variant
of the chromium system is easily studied since the oxo
complex 2 is isolable,[5a] whereas 1 is only a fleetingly observed
intermediate.[4] However, the major difference between 1 and
2 is in their different substrate selectivity in asymmetric
epoxidation. Manganese salen complexes show high enantio-
selectivities in the epoxidation of Z alkenes, but E alkenes are
poor substrates. Chromium salen complexes show the oppo-
site characteristics (Table 1). Other minor differences in-
clude[1, 2b, 5] the greater effect of O-donor ligand additives in
the chromium series; the type of oxidant, which is more
variable in the manganese series; and the nature of the
substituents on the salen rings: chromium requires electron-
withdrawing substituents for reasonable rates, but these need
not be bulky, in contrast to the manganese case.
[a] Dr. P. Brandt
Department of Chemistry, Organic Chemistry
Royal Institute of Technology
100 44 Stockholm (Sweden)
Fax : (‡46)8-7912333
E-mail: pbrandt@kth.se
[b] Prof. P.-O. Norrby
Department of Chemistry, Organic Chemistry
Danish Technical University, Building 201
Kemitorvet, 2800 Lyngby (Denmark)
[c] Dr. D. G. Gilheany, Dr. A. M. Daly
Chemistry Department and Conway Institute of Biomolecular and
Biomedical Sciences
University College Dublin, Belfield
Dublin 4 (Ireland)
Fax : (‡353)1-7162127
E-mail: declan.gilheany@ucd.ie
Supporting information for this article is available on the WWW under
http://www.chemeurj.org/ or from the author.
Chem. Eur. J. 2002, 8, No. 18
¹ 2002 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
0947-6539/02/0818-4299 $ 20.00+.50/0
4299

P. Brandt, D. Gilheany et al.
FULL PAPER
Table 1. Reported enantioselectivities in epoxidation of E- and Z-b-
methylstyrene with 1[2h, 6] and 2.[2e]
more pronounced in the presence of chloride can be
rationalized by the qualitative PES differences on the quintet
surface, although they also suggested that two different
oxidizing species offer competing pathways.[15b] The relevant
spin surfaces are depicted in Figure 1. By postulating that the
R1
R1
R2
O
1 or 2, [O]
Me
R2
R2
Me
R1
Reagent
ee [%]
H
H
Ph
Ph
Ph
Ph
H
1
92
À 27
22
2[a]
1
2[a]
92
O
H
O
Mn
Mn
[a] L ˆ triphenylphosphine oxide.
O
Mn
quintet
The success of these metal salen systems has led to
considerable discussion of the mechanism of oxo trans-
fer[1, 2b, 7 11] and the origin of enantioselectivity.[1, 2b, 12] Recent-
ly, some agreement has been reached on the mechanism of
oxo transfer in the manganese series. The crucial break-
through was recognition of the importance of the spin state of
the manganese oxo moiety in determining the reaction
pathway.[7 10] Both theoretical and experimental studies agree
that the most stable spin state for the MnIII stage of the
catalytic cycle is the quintet (S ˆ 2).[4, 7 10, 12, 13] However, there
with Cl–
triplet
O
Mn
singlet
triplet
without Cl–
quintet
Reaction coordinate
ˆ
Figure 1. Qualitative consensus PES for the reaction of (salen)Mn O
species with alkene, collected from literature sources.[7, 9, 10]
is disagreement about the preferred spin state of the actual
V
epoxidizing species, the MnV O complex. Observable Mn
ˆ
ˆ
X
species (X ˆ O, N; always neutral or anionic complexes) are
generally diamagnetic singlets (S ˆ 0),[4, 13] and some theoret-
ical studies also identify this as the most stable state of the
reaction can proceed in parallel on at least two different spin
surfaces, whereby the preference for and rate of transfer
between the spin surfaces can be affected by the reaction
conditions and mode of production of the MnV O complex,
[8, 10]
ˆ
[(salen)Mn O] species.
However, the energetic differ-
ˆ
ences between available spin states are small and strongly
dependent both on the exact model used (naked cation-
ic,[3b, 7, 8] with neutral ligand,[3d] or with anionic chloride
ligand[8, 9, 10]) and the level of theory employed. For the
related Mn porphyrin series, it was suggested that the singlet
complexes are observable because they are less reactive than
the corresponding catalytically active complexes.[13b] There is
consensus on some important points in the available theoret-
various contrasting experimental observations, such as the
survival of vinylcyclopropane radical traps[16] and the ob-
served isomerization of Z alkenes to trans epoxides,[17] can be
rationalized. Recently, Linde et al. provided further exper-
imental evidence for competing oxidation pathways by study-
ing the electronic effects of the substrate on the diastereose-
lectivity and enantioselectivity in the catalytic epoxidation by
iodosylbenzene of para-substituted cis-stilbenes with Jacob-
sen×s catalyst (1).[11]
With regard to the origin of stereoselectivity in the
manganese system, there is still substantial disagreement.
Most fundamentally, there is no agreement as to the shape of
the active complex. In addition, although nearly all workers
use the concept that the alkene approaches side-on[18] to the
oxygen atom, essentially all possible trajectories of approach
to the metal oxo moiety have been used as a basis for an
explanation. For example, Jacobsen et al. assumed a planar
salen ligand framework and explained enantioselection by
approach of alkene over the cyclohexyl ring of the complex.[19]
They also invoked side-on approach to rationalize the better
selectivity for Z alkenes than for E alkenes. Katsuki et al.
suggested a nonplanar structure and accumulated much
experimental evidence for this.[12b] They suggest that the
alkene approaches along the metal nitrogen bond axis. Houk
et al. independently suggested a similar nonplanar structure
but suggested approach of alkene between the 3,3'-positions
of the salen ligand.[12a] This is similar to our explanation for the
selectivity in the chromium series.[2b, 2f, 20] We also had to
postulate[2b, 20] a nonplanar salen structure for the chromium
complex to retain the side-on approach model for its
ˆ
ical studies. Most importantly, all spin states for Mn O are
energetically accessible, so that if spin change in these systems
is slow,[14] the reaction can proceed completely on the quintet
surface. Indeed, in a recent EPR study, signals which could
arise from a triplet or quintet MnV O species were ob-
ˆ
served.[4a] Furthermore, all comparative studies of transition
states have shown that the addition on the triplet surface
(S ˆ 1) is lower in energy than that on the singlet surface
(S ˆ 0),[7, 10] and experimental evidence also indicates that
addition on the singlet surface is a slow process.[13b] Thus, if
spin change does occur, the addition will proceed on the
triplet surface to give a radical intermediate with subsequent
formation of product either on the triplet or quintet surface.
For this reason, the singlet state, although preferred for less
reactive, isolable complexes, can be neglected for the catalytic
process. Finally, the quintet potential energy surface (PES)
seems very responsive to the charge of the axial ligand. In the
presence of chloride ligand, the quintet PES is very similar to
the triplet PES and proceeds through a radical intermedi-
ate.[9, 10] For the cationic quintet complex, the addition is still
strongly asynchronous, but now formation of epoxide is a
concerted process.[7] Thus, the recent observation by Adam
et al.[15] that isomerization in a radical intermediate is much
4300
¹ 2002 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
0947-6539/02/0818-4300 $ 20.00+.50/0
Chem. Eur. J. 2002, 8, No. 18

[(Salen)Cr]-Mediated Alkene Epoxidation
4299 4307
epoxidation reactions. Recently, Wiest et al. performed cal-
culations on a complete salen ligand and, significantly, they
showed that the catalyst could adopt two different conforma-
tions with a stepped or a concave catalyst/substrate inter-
face.[3c,d] None of this work provides an explanation for the
contrasting selectivity behavior of chromium salen complexes
in asymmetric epoxidation. Therefore, we felt that a detailed
study of the chromium system would shed light on these
differences and in turn give insight into the manganese system.
In recent years, quantum chemical methods have been
successfully employed in the evaluation of mechanistic
alternatives in transition-metal-catalyzed reactions.[21] In par-
ticular, density functional theory has been extensively used,
the most popular methods being B3LYP and BP. While our
study on the chromium system was in progress, several papers,
O
O
N
O
N
O
N
O
N
O
Cr
Cr
Me3PO
3a
3b
Scheme 1. Truncated molecules used in the calculations. Species without
donor ligand are denoted a; with donor b.
•
O
O
N
O
N
N
O
N
O
N
O
N
Cr
Cr
Cr
O
O
O
Me3PO
Me3PO
Me3PO
n=0,1
n=0,1
n=0,1
including theoretical studies into the Mn system, appear-
d, 7 10]
ed.[3b
Our results, in combination with the literature
3
4
5
reports, could therefore be used to compare the two systems
and rationalize some of the differences.
O
N
Cr
O
O
N
N
O
N
N
N
V
The spin states of CrIII and Cr O in salen complexes 13
ˆ
Cr
Cr
and 14 were established by Kochi et al.[5] by measurement of
the magnetic moment, and those of CrIII by Bryliakov et al.[22]
by EPR measurements. These provide a useful check on the
calculations reported here, something that is presently
unavailable for the manganese series. We used the B3LYP
method, which has been criticized as unsuitable for spin-state
determinations in the Mn system.[10] However, the results of
the present calculations are in agreement with the experi-
mental determinations, the best measure of their validity.
We now report details of our theoretical and experimental
study, which has led to insight into the mechanism of oxygen
transfer in the chromium system. Results of our studies into
the mode of enantioselection will be the subject of a future
paper. We are aware of one other recent theoretical study of a
(non-salen)chromium(v) oxo system.[23]
O
O
O
O
Me3PO
Me3PO
Me3PO
n=0,1
n=0,1
n=0,1
6
7
8
Scheme 2. Species evaluated in the quantum chemical study (a: n ˆ 0; b:
n ˆ 1).
Table 2. Potential energy surfaces (B3LYP/BSI) of the reactions of
ethylene with complexes 3a and 3b [kJmolÀ1].
a: No donor ligand
b: Axial Me3PO[a]
doublet quartet
doublet
quartet
3 ‡ ethylene
0
64
12
64
14
102
0 (0)
14 (30)
À 23 (À11)
17 (35)
À 69
73 (87)
4
5
6
7
71 (84)[b]
À 19 (À9)[b]
À 50 (À40)[b]
À 162
À 77
8 ‡ epoxide
55
41
À 49
Methods of Calculation
[a] Values in parenthesis from B3LYP/BSII//B3LYP/BSI. [b] Vertical
excitation.
All calculations were performed with the Gaussian98 program package[24]
using the B3LYP hybrid functional.[25] Geometry optimizations were
performed with d95v for C, N, O, and H; d95 for P; and a (14s,11p,6d)
primitive basis from Wachters,[26] augmented by two p and one diffuse d
function contracted to [6s,5p,3d], for Cr (see Supporting Information). This
basis set is referred to as BSIin the following. Final energies were
determined by using 6-311 ‡ G* for Cr, 6-31 ‡ G* for N and O, and 6-31G*
for P, C, and H (BSII).
6
O
84[b]
30
N
O
N
O
100
50
87
0
Cr
L
A reduced model system was used to mimic the full chromium salen
complex and the donor ligand used experimentally (Scheme 1). This
truncated ligand proved to be very similar to the real salen ligand for the
related Mn(salen)(Cl)-catalyzed reaction.[9] Ethene was used as the model
alkene.
35
-9[a, b]
-11
0
-40[b]
O
N
O
N
O
O
N
O
N
O
Cr
L
7
-50
-100
-150
Cr
O
N
Cr
O
L
Results
N
O
O
L
N
O
N
O
4
3
Cr
L
The potential energy surface: The energies of the twelve
species 3a,b 8a,b (Scheme 2) on a nonconcerted PES were
evaluated. Calculated energies are tabulated in Table 2; those
of 3b 8b are given at both BSIand BSIlevels. Figure 2
shows these energies graphically for systems b (with axial
ligand).
5
quartet surface
doublet surface
[a]
This quartet ring-closes without barrier.
[b]
Vertical excitation from the doublet surface.
Figure 2. The PES at the B3LYP/BSII//B3LYP/BSI level for systems b.
0947-6539/02/0818-4301 $ 20.00+.50/0
Chem. Eur. J. 2002, 8, No. 18
¹ 2002 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
4301

P. Brandt, D. Gilheany et al.
FULL PAPER
Table 3. Results of investigations with radical probe 9 and derivatives.
Radical-trap experiments: Radical traps 9 and 15 were
prepared and treated with chromium and manganese salen
complexes 13 and 14 (Scheme 3). The results are listed in
Table 3 and Table 4.
Entry
Substrate
Reactant
Cr complex
[mol%]
11:12
Total yield
[%]
1
2
3
4
5
6
7
8
9
10
11
12
13
14
9
9
14
100
10
100
50
10
100
50
10
-
-
> 20:1
4:1
7[a]
50[a]
90
93
87
83
86
85
96
96
90
86
86
13 ‡ PhIO
10
10
10
10
10
10
10
10
11
11
12
12
13
3:2
2:1
O
13
13
14
14
Ph
O
O
10:1
10:1
10:1
1:1
n.d.[b]
n.d.
n.d.
n.d.
n.d.
n.d.
Ph
Ph
Ph
Ph
Ph
Ph
Ph
14
9
10
11
12
PhIO
none
13
14
13
PF6
PF6
100
100
100
100
PhIO
O
N
O
N
O
N
O
N
Cr
Cr
14
95
O
[a] Reduced yield accounted for by unconsumed alkene. [b] n.d. ˆ no
decomposition.
13
14
O
Table 4. Epoxidation of alkene 15 (mixture of isomers) with various
complexes and oxidants.
Ph
Ph
Ph
Ph
15
Entry Complex mol% Oxidant Yield of 16 [%] Unconsumed 15 [%]
16
cis
trans
Z
E
Scheme 3.
1
2
3
4
13
14
deoxy-1
deoxy-1
10
100
PhIO
none
0
0
0
0
12
2
30
11
11
21
30
0
39
47
Discussion
2.5 NaOCl 33
2.5 PhIO 20
Calculations on intermediates and the PES: Chromium has
one less d electron than manganese, and analogously to the
spin states of manganese salen complexes, the chromium
complexes studied here can be in a doublet or quartet state.
À1
ˆ
(87 kJmol ) causes elongation of the Cr O bond from
1.594 to 1.680 ä and a slight conformational change in the
complex.[28] The high excitation energy makes involvement of
the quartet state in the initial part of the reaction less likely.
Therefore, no attempts were made to locate transition states
on this PES.
Chromium oxo complexes 3a and 3b: The chromium-
catalyzed epoxidation of alkenes differs from its Mn-catalyzed
counterpart in that the oxidized form of the catalyst has been
isolated and characterized.[5a]
This is possibly due to its lower
reactivity compared to the bet-
ter known Mn analogue. This
trend is not unexpected given
ˆ
the lower M O bond strength
to the right of the periodic
table.[27]
The chromium–salen oxo
complexes studied here (Fig-
ure 3) show a doublet ground
state with the spin density cen-
tered on chromium (s ˆ 1.42 for
3b) and on the oxo oxygen
atom (s ˆ À0.39 for 3b). This
agrees with experimental meas-
urement of meff ˆ 1.9 mB for
[5a]
[(salen)CrV O] complex 14.
ˆ
In the quartet state, an electron
has been promoted from the p
system of one half of salen into
ˆ
an antibonding Cr O orbital
(ligand-to-metal charge trans-
fer, LMCT). This excitation
Figure 3. Structures of the doublet Cr oxo complexes 3a and 3b.
0947-6539/02/0818-4302 $ 20.00+.50/0
4302
¹ 2002 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2002, 8, No. 18

[(Salen)Cr]-Mediated Alkene Epoxidation
4299 4307
Another notable feature in
Figure 3 is that coordination of
a donor-ligand trans to the oxo
group results in a dramatic
conformational
change
in
which salen is pushed towards
the oxo ligand and a stepped
conformation is generated. This
is in agreement with the crystal
structures of [(salen)CrV O]
ˆ
obtained by Kochi et al.[29]
An interesting discovery by
Wiest
et al.[3b,d]
is
that
ˆ
[(salen)Mn O] with an axial
Me3NO ligand can also adopt
different conformations; step-
ped and bowl-shaped confor-
Figure 4. Transition states 4a and 4b in the doublet states.
mations are energetically acces-
sible. For the Cr catalysts examined here two analogous
conformations were located that differ by only 3 kJmolÀ1 in
relative energy. In terms of substrate compatibilty, conforma-
tional differences between the two catalysts might explain
why Mn salen complexes are better suited for the epoxida-
tion of Z alkenes, whereas Cr salen complexes perform
better with E alkenes (Table 1). This rationale is particularly
attractive to explore because of the different requirements of
the 3,3'-substituents for the Mn versus the Cr system. Bulky
substituents at these positions and partially charged groups
will be within van der Waals interaction distance and could
thus affect the conformational behavior of the catalysts.
by Kochi et al.[5a] is confirmed, although our results suggest
that the intermediate may be a radical rather than a cation.
Most significant is that when an axial ligand is present an
energy difference of only 2 kJmolÀ1 separates the doublet and
quartet states of 5b. Excitation from doublet to quartet state
results in direct collapse to the corresponding epoxide. Once
again the effect of the donor ligand on complex conformation
can be seen when the structures of 5a and 5b are compared
(Figure 5).
Transition states 6 for ring closure: Ring closure of the ethyl
radical intermediate on the doublet spin surface (Figure 6) is
calculated to be slightly higher in energy than the reverse
reaction leading back to the alkene (35 versus 30 kJmolÀ1).
The energy difference of only 5 kJmolÀ1 is, however, beyond
the accuracy of the computational method. The experimental
observation of no isomerization of Z-b-methylstyrene implies
that ring closure is faster than the back-reaction. As discussed
previously,[7] vertical excitation from the doublet to the
quartet of 5b requires not more than 2 kJmolÀ1. A small
energy difference offers the possibility of a fast spin inversion,
and ring closure can then take place without barrier on the
quartet PES. Thus, two alternative pathways exist for this step.
Transition states 4a and 4b for the addition of ethylene to oxo
complexes 3a and 3b: Experimentally, the reaction rate and
yield is improved by the addition of various donor ligands to
the reaction.[2a,c,g] Among the ligands experimentally evalu-
ated were various N-oxides, phosphane oxides, DMSO, and
DMF, but triphenylphosphane oxide is usually the most
effective.[2a,e,g] The activation energy for the addition of
ethylene to 3 is 64 kJmolÀ1 without an axial ligand, but this
value drops to 14 kJmolÀ1 when Me3PO is coordinated. This is
in agreement with the experimentally observed ligand accel-
eration. The effect of donor
ligand on the salen conforma-
tion can be seen in Figure 4.
The salen ligand is brought
closer to the active site by axial
coordination. This could be in-
terpreted as a cause of im-
proved enantioselectivity.
Intermediate ethyl radicals 5a
and 5b: Both TSs 4a and 4b
lead to stable intermediates
with a carbon-centered radical
(5a and 5b). The doublet ethyl
radical 5b has a Mulliken spin
density of 2.28 at Cr and À1.03
at the ethyl terminus. Thus, the
stepwise mechanism suggested
Figure 5. Intermediate ethyl radicals 5a and 5b.
Chem. Eur. J. 2002, 8, No. 18
¹ 2002 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
0947-6539/02/0818-4303 $ 20.00+.50/0
4303

P. Brandt, D. Gilheany et al.
FULL PAPER
state, and the singlet and quin-
tet states are 38 and 59 kJmolÀ1
higher in energy (BSI). Thus,
the initial reaction takes place
on the triplet PES and leads to a
radical intermediate, which is
stable both as a triplet and a
quintet. This stability allows
time for rotation about the
À
C C bond to give a mixture of
products, which approaches a
thermodynamic distribution of
cis and trans epoxides.
Summary: The above theoret-
ical investigation suggests that
the epoxidation reaction pro-
ceeds through an intermediate
Figure 6. Transition states 6a and 6b for ring closure of the intermediate ethyl radicals (doublets).
Chromium-coordinated epoxide: The structures of the species
produced on ring closure (7a, 7b) in both the doublet excited
states and quartet ground states are shown in Figure 7. The
prediction of a quartet ground state for CrIII salen complexes
agrees with an experimental measurement of meff ˆ 3.5 mB for
complex 13 and with EPR measurements on the (salen)CrIII
spin state.[22]
5 with a doublet ground state of the starting [(salen)CrV O]
(3) and formation of a quartet [(salen)CrIII] product 8. I t
seems most likely that the necessary spin change occurs after
the formation of the intermediate 5 and prior to ring closure
of epoxide. The spin crossing probability pSI will therefore
determine the lifetime of the intermediate.[31] The experimen-
tal observation of small amounts of isomerized epoxide from
ˆ
Recall that Kochi et al. found
strong indications of a cationic
mechanism.[5a]
Byproducts
found in the reaction mixture
were characteristic of a cation
rearrangement. Our failure to
identify carbocationic inter-
mediates in the theoretical in-
vestigations suggests that by-
products in the reaction arise
from Lewis-acid-catalyzed re-
actions mediated by CrIII or
CrV complexes. The observed
rearrangement of epoxide 10 to
aldehyde 11 (see below) is evi-
dence for this.
Diastereoselectivity: investiga-
tion of [(salen)CrIV O]: We re-
ˆ
cently discovered an unusual
experimental effect. Replace-
À
ment of the NO3 counterion
in 2 with BARF[30] led to a clean
conversion of E-olefins to a 1:1
mixture of cis and trans epox-
ides.[2e] We suggested that this
was due to the formation of a
[(salen)CrIV O] species by one-
ˆ
electron reduction of the chro-
mium complex in the presence
of a tetraarylborate. We have
investigated such a species the-
oretically. In short, [(salen)-
CrIV O] has a triplet ground
Figure 7. Product complexes 7a and 7b.
ˆ
4304
¹ 2002 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
0947-6539/02/0818-4304 $ 20.00+.50/0
Chem. Eur. J. 2002, 8, No. 18

[(Salen)Cr]-Mediated Alkene Epoxidation
4299 4307
Z-b-methylstyrene suggests that the spin crossing is relatively
facile. Only in a small number of cases does the intermediate
O
CrV
Ph
Ph
À
live long enough to undergo C C bond rotation prior to ring
9
closure, and therefore only a small amount of trans-b-
methylstyrene oxide (ca. 10%) is produced from Z-b-
methylstyrene.
Ph
Ph
Ph
O
Cr
O
Cr
O
Cr
Ph
Ph
Ph
Terminal oxidants: We have previously noted the failure of
terminal oxidants other than iodosylbenzene.[2e] In light of
these calculations, a possible interpretation of those results
becomes apparent. To oxidize CrIII to Cr O, a spin change
V
ˆ
H
from quartet to doublet must take place. Oxidants containing
a heavy atom could facilitate this spin change, hence the
success of PhIO, but the failure of the other oxidants that we
have tried.[32]
Ph
Ph
Ph
O
O
O
+ CrIII
Ph
+ CrIII
Ph
+ CrIII
Ph
10
11
12
Scheme 4. Possible pathways for the formation of 10 12.
Radical/cation-trap experiments: On the basis of these
theoretical results we sought an experimental method to
confirm the presence of an intermediate in the reaction of a
chromium oxo species with alkenes. We chose vinylcyclopro-
panes, which have been used extensively in recent mechanistic
investigations.[33] A disadvantage of this approach is that
distinction between radical and cationic intermediates would
not be possible. More seriously, the relatively limited sub-
strate scope of the chromium system[2b, 5a] limits the number of
potential test compounds. A radical/cation trap only produces
irrefutable results when the cyclopropyl ring remains intact.
Ring opening strongly suggests an intermediate but does not
prove it conclusively.
intermediate in the reaction of this Cr oxo species with alkene
9. Whether this same intermediate can also lead to epoxide 10
is, of course, unproven.
Hexene-like trap: In an attempt to resolve the issue we
designed an alternative cyclopropylalkene, namely, 1,2-bis-
(trans-2-phenylcylopropyl)ethene (15), which bears a greater
resemblance to substrates that were successfully epoxidized,
38]
albeit poorly, by the chromium system.[36
We first con-
firmed the stability of the bis(cyclopropyl) epoxide 16 to the
reaction conditions. In the presence of 13 or 14 monitoring by
TLC showed that it remained stable for up to 10 h. The
decomposition products which formed after this time, a set of
isomers, had mass spectrometric and 1H NMR data consistent
with the expected[39] rearrangement product, ketone 17.
Alkene 15 was then subjected to both stoichiometric and
catalytic epoxidation with
[(salen)Cr] (Table 4, entries 1
and 2). In neither case was
epoxide 16 or its isomer 17
produced.[40] In both cases just
Styrene-like trap. Substituted styrenes are good substrates for
epoxidation with the chromium system,[2, 5a] because the
incipient cation/radical is stabilized at the benzylic position
that bears the cyclopropyl group. 1-Phenyl-1-(trans-2-phenyl-
cyclopropyl)ethene (9) and derivatives thereof were used to
good effect in the manganese system by Linde et al., who
found that formation of cis epoxide from Z alkene and of trans
epoxide from E alkene did not involve a radical intermedi-
ate.[16b, 34]
Unfortunately 1,1-disubstituted alkenes of this type are not
ideal for the chromium system,[35] and when 9 was treated with
13 as catalyst and PhIO as oxidant, or with a stoichiometric
amount of 14, none of the corresponding epoxide 10 was
generated. Instead, we obtained the dihydropyran 12 and the
aldehyde 11 in a ratio that depended on the amount of
chromium complex used in the reaction (Table 3, entries 1 and
2). This, in itself, suggests the intervention of a reaction
intermediate. However, we also found that, under the reaction
conditions, epoxide 10 reacts with 13 and 14 to give 11 and 12
in various ratios (Table 3, entries 3 8). Linde et al. also found
aldehyde 11 and pyran 12 in the manganese system in cases
when epoxide yield was not quantitative.[16b] The varying
ratios of 11 and 12 in the control reactions strongly suggests
their formation by a combination of the two routes outlined in
Scheme 4. A possible caveat is that spin-change factors might
influence the product distribution in the attempted epoxida-
tion experiment, whereas in the control reactions they would
not come into play. In any event, the fact that ring-opened
products are observed is overwhelming evidence for an
one product was detected by
TLC, but it proved impossible
to isolate.[41]
The failure to observe epoxide 16, known to be stable under
the reaction conditions, as a product of this reaction would
suggest that a radical or cationic intermediate intervened
along the reaction path leading to cyclopropane ring opening.
This strongly supports the existence of a discrete intermediate
in the epoxidation of alkenes by chromium salen complexes.
We confirmed that 15 was indeed epoxidizable by a metal
oxo species by examining its behavior in the manganese
system, with hypochlorite and PhIO as oxidants (Table 4,
entries 3 and 4). The results were as expected, in that the Z
isomer reacted faster in this case. With hypochlorite (entry 3)
there is a slight discrepancy between the amount of Z alkene
consumed and the yield of cis epoxide, a result similar to those
obtained by Jacobsen et al.[16a] for this type of nonconjugated
alkene substrate. With PhIO (entry 4), there is a somewhat
larger discrepancy, in good agreement with the results of
Linde et al.[16b]
Chem. Eur. J. 2002, 8, No. 18
¹ 2002 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
0947-6539/02/0818-4305 $ 20.00+.50/0
4305

P. Brandt, D. Gilheany et al.
FULL PAPER
2002, 43, 2107 2110; h) A. M. Daly, D. G. Gilheany, Tetrahedron:
Asymmetry, submitted.
Conclusion
[3] a) D. A. Plattner, D. Feichtinger, Angew. Chem. 1997, 109, 1796 1798;
Angew. Chem. Int. Ed. Engl. 1997, 36, 1718 1719; b) D. A. Plattner,
D. Feichtinger, J. El-Bahraoui, O. Wiest, Int. J. Mass. Spectrom. 2000,
195/196, 351 362; c) D. Feichtinger, D. A. Plattner, J. Chem. Soc.
Perkin Trans. 2 2000, 1023 1028; d) J. El-Bahraoui, O. Wiest, D. A.
Plattner, D. Feichtinger, Angew. Chem. 2001, 113, 2131 2134; Angew.
Chem. Int. Ed. 2001, 40, 2073 2076; e) D. Feichtinger, D. A. Plattner,
Chem. Eur. J. 2001, 7, 591 599.
[4] a) EPR: K. A. Campbell, M. R. Lashley, J. K. Wyatt, M. H. Nantz,
R. D. Britt, J. Am. Chem. Soc. 2001, 123, 5710 5719; b) EPR/NMR:
K. P. Bryliakov, D. E. Babushkin, E. P. Talsi, J. Mol. Catal. A: Chem.
2000, 158, 19 35; c) EPR/MS: W. Adam, C. Mock-Knoblauch, C. R.
Saha-Moller, M. Herderich, J. Am. Chem. Soc. 2000, 122, 9685
9691
[5] a) E. G. Samsel, K. Srinivasan, J. K. Kochi, J. Am. Chem. Soc. 1985,
107, 7606 7617; b) K. Srinivasan, P. Michaud, J. K. Kochi, J. Am.
Chem. Soc. 1986, 108, 2309 2320.
[6] E. N. Jacobsen, W. Zhang, A. R. Muci, J. R. Ecker, L. Deng, J. Am.
Chem. Soc. 1991, 113, 7063 7064.
[7] C. Linde, B. äkermark, P.-O. Norrby, M. Svensson, J. Am. Chem. Soc.
1999, 121, 5083 5084.
[8] T. Strassner, K. N. Houk, Org. Lett. 1999, 1, 419 421.
[9] L. Cavallo, H. Jacobsen, Angew. Chem. 2000, 112, 602 604; Angew.
Chem. Int. Ed. 2000, 39, 589 592; H. Jacobsen, L. Cavallo, Chem. Eur.
J. 2001, 7, 800 807.
We have thoroughly investigated the mechanism of
[(salen)Cr]-mediated epoxidation. Both theoretical and ex-
perimental results indicate the existence of an intermediate
along the reaction path. The B3LYP calculations identified
this intermediate as a b-alkoxy radical bound to chromium.
The calculations suggest almost complete degeneracy be-
tween the low- and high-spin state at this point, and this
facilitates spin-surface crossing. Considering the accuracy of
the B3LYP functional, we correctly predicted a low-spin state
V
III
ˆ
for the Cr O reactant and a high-spin state for the Cr
product. The fact that the catalyst must change between high-
and low-spin during the catalytic cycle could affect both the
diastereoselectivity and the rate of reoxidation of the catalyst.
Axial donor ligands significantly lower the barrier of the
reaction and dramatically change the conformation of the
catalyst. This conformational change brings the salen ligand
closer to the substrate in the selectivity-determining transition
state and thereby increases the enantioselectivity. In close
analogy to the [Mn(salen)] catalyst the model [Cr(salen)]
complex could adopt two different conformations with a
stepped or a concave catalyst/substrate interface. We spec-
ulate that a difference in conformational preference is the
cause of differences in substrate selectivity between [Mn(sa-
len)] and [Cr(salen)] catalysts.
[10] Y. G. Abashkin, J. R. Collins, S. K. Burt, Inorg. Chem. 2001, 40, 4040
4048.
[11] C. Linde, N. KoliaÔ, P.-O. Norrby, B. äkermark, Chem. Eur. J. 2002, 8,
2568 2573.
[12] a) K. N. Houk, N. C. DeMello, K. Condroski, J. Fennen, T. Kasuga,
™Origins of Stereoselecivity in Jacobsen Epoxidations∫ in Electronic
Conference on Heterocyclic Chemistry, ECHET96 (Eds.: H. S. Rzepa,
J. P. Snyder, C. Leach), London, 1996, 24/6 22/7 (http://www.ch.ic.a-
c.uk/ectoc/echet96/); b) Y. N. Ito, T. Katsuki, Bull. Chem. Soc. Jpn.
1999, 72, 603 619.
[13] a) Z. Gross, G. Golubkov, L. Simkhovich, Angew. Chem. 2000, 112,
4211 4213; Angew. Chem. Int. Ed. 2000, 39, 4045 4047; b) N. Jin,
J. T. Groves, J. Am. Chem. Soc. 1999, 121, 2923; c) J. T. Groves, J. Lee,
S. S. Marla, J. Am. Chem. Soc. 1997, 119, 6269 6273; d) J. Du Bois, J.
Hong, E. M. Carreira, M. W. Day, J. Am. Chem. Soc. 1996, 118, 915
916; e) T. J. Collins, R. D. Powell, C. Slebodnick, E. S. Uffelman, J.
Am. Chem. Soc. 1990, 112, 899 901; f) T. J. Collins, S. W. Gordon-
Wylie, J. Am. Chem. Soc. 1989, 111, 4511 4513.
[14] The rate of spin inversion in a manganocene derivative has been
reported to be 105 sÀ1 at 310 K, a relatively slow process, with an
equilibrium that was strongly affected by environmental effects: N.
Hebendanz, F. H. Kˆhler, G. M¸ller, J. Riede, J. Am. Chem. Soc. 1986,
108, 3281 3289.
[15] a) W. Adam, K. J. Roschmann, C. R. Saha-Mˆller, Eur. J. Org. Chem.
2000, 3519 3521; b) W. Adam, K. J. Roschmann, C. R. Saha-Mˆller,
D. Seebach, J. Am. Chem. Soc. 2002, 124, 5068 5073.
[16] a) H. Fu, G. C. Look, W. Zhang, E. N. Jacobsen, C.-H. Wong, J. Org.
Chem. 1991, 56, 6497 6500; b) C. Linde, M. Arnold, P.-O. Norrby, B.
äkermark, Angew. Chem. 1997, 109, 1802 1803; Angew. Chem. Int.
Ed. Engl. 1997, 36, 1723 1725.
[17] a) W. Zhang, N. H. Lee, E. N. Jacobsen, J. Am. Chem. Soc. 1994, 116,
425 426; b) M. Palucki, P. J. Pospisil, W. Zhang, E. N. Jacobsen, J.
Am. Chem. Soc. 1994, 116, 9333 9334.
[18] The side-on approach model: a) J. P. Collman, X. Zhang, V. J. Lee,
E. S. Uffelman, J. I. Brauman, Science 1993, 263, 1404 1411; b) J. T.
Groves, T. E. Nemo, J. Am Chem. Soc. 1983, 105, 5786 5791.
[19] E. N. Jacobsen, ™Asymmetric Catalytic Epoxidation of Unfunctional-
ised Olefins∫ in Catalytic Asymmetric Synthesis (Ed.: I. Ojima), VCH,
Weinheim, 1993, pp. 159 203.
[20] D. G. Gilheany, Abstracts of Papers, 211th National Meeting of the
American Chemical Society, New Orleans, LA, American Chemical
Society, Washington DC, 1996; ORGN 165.
In addition, our calculations assigned the increased ten-
dency of formation of trans epoxides from cis alkenes with a
BARF counterion to a decreased rate of ring closure of the b-
alkoxy radical formed from the reaction between the alkene
and a CrIV species.
Acknowledgements
We thank Prof. Olaf Wiest, Prof. Douglas Doren, and Prof. Klaus Theopold
for disclosing results prior to publication. Financial support to P.O.N. from
the Danish Technical Research Council (STVF) and
a postdoctoral
fellowship for P.B. from the Swedish Foundation for International
Cooperation in Research and Higher Education are acknowledged. D.G.
and A.D. thank Enterprise Ireland (Grants SC/97/536 and IC/96/055) and
University College Dublin (demonstratorship for A.D.). Supercomputer
time was provided by the Swedish Council for High Performance
Computing (HPDR) and Parallelldatorcentrum (PDC), Royal Institute
of Technology. Mats Svensson is acknowledged for assistance with the
contracted basis set used for Mn and Cr.
[1] a) E. N. Jacobsen, Acc. Chem. Res. 2000, 33, 421 431; b) E. N.
Jacobsen, M. H. Wu in Comprehensive Asymmetric Catalysis (Eds.:
E. N. Jacobsen, A. Pfaltz, H. Yamamoto), Springer, New York, 1999,
chap 18.2, pp. 649 677; c) T. Katsuki in Catalytic Asymmetric Syn-
thesis, 2nd ed. (Ed.: I. Ojima), Wiley, New York, 2000, pp. 287 325;
d) T. Katsuki, J. Mol. Catal. A: Chem. 1996, 113, 87 107.
[2] a) C. Bousquet, D. G. Gilheany, Tetrahedron Lett. 1995, 36, 7739
7742; b) C. T. Dalton, K. M. Ryan, V. M. Wall, C. Bousquet, D. G.
Gilheany, Top. Catal. 1998, 5, 71 90; c) K. M. Ryan, C. Bousquet,
D. G. Gilheany, Tetrahedron Lett. 1999, 40, 3613 3616; d) A. M. Daly,
C. T. Dalton, M. F. Renehan, D. G. Gilheany, Tetrahedron Lett. 1999,
40, 3617 3620; e) A. M. Daly, M. F. Renehan, D. G. Gilheany, Org.
Lett. 2001, 3, 663 666; f) C. P. O'Mahony, E. M. McGarrigle, M. F.
Renehan, K. M. Ryan, N. J. Kerrigan, C. Bousquet, D. G. Gilheany,
Org. Lett. 2001, 3, 3435 3438; g) N. J. Kerrigan, I. J. Langan, C. T.
Dalton, A. M. Daly, C. Bousquet, D. G. Gilheany, Tetrahedron Lett.
¡
[21] M. Torrent, M. Sola, G. Frenking, Chem. Rev. 2000, 100, 439 493.
[22] K. P. Bryliakov, M. V. Lobanova, E. P. Talsi, J. Chem. Soc. Dalton
Trans. 2002, 2262 2265.
4306
¹ 2002 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
0947-6539/02/0818-4306 $ 20.00+.50/0
Chem. Eur. J. 2002, 8, No. 18

[(Salen)Cr]-Mediated Alkene Epoxidation
4299 4307
[23] J. S. Hess, S. Leelasubcharoen, A. L. Rheingold, D. J. Doren, K. H.
Theopold, J. Am Chem. Soc. 2002, 124, 2454 2455.
[34] Linde et al. quoted a rate constant for ring opening of the cyclopropyl-
containing radical derived from 9 of 3.6 Â 108 sÀ1 [16b]
.
The cyclopropyl
[24] Gaussian98, Revision A.3, M. J. Frisch, G. W. Trucks, H. B. Schlegel,
G. E. Scuseria, M. A. Robb, J. R. Cheeseman, V. G. Zakrzewski, J. A.
Montgomery, R. E. Stratmann, Jr., J. C. Burant, S. Dapprich, J. M.
Millam, A. D. Daniels, K. N. Kudin, M. C. Strain, O. Farkas, J. Tomasi,
V. Barone, M. Cossi, R. Cammi, B. Mennucci, C. Pomelli, C. Adamo,
S. Clifford, J. Ochterski, G. A. Petersson, P. Y. Ayala, Q. Cui, K.
Morokuma, D. K. Malick, A. D. Rabuck, K. Raghavachari, J. B.
Foresman, J. Cioslowski, J. V. Ortiz, B. B. Stefanov, G. Liu, A.
Liashenko, P. Piskorz, I. Komaromi, R. Gomperts, R. L. Martin,
D. J. Fox, T. Keith, M. A. Al-Laham, C. Y. Peng, A. Nanayakkara, C.
Gonzalez, M. Challacombe, P. M. W. Gill, B. Johnson, W. Chen, M. W.
Wong, J. L. Andres, C. Gonzalez, M. Head-Gordon, E. S. Replogle,
J. A. Pople, Gaussian, Inc., Pittsburgh, PA, 1998.
ring opening in such a stabilized system should be slower than bond
rotation in the intermediate radical, which has an estimated rate of
>1010 sÀ1 [4c]
but the complete absence of isomerized product with
,
intact cyclopropyl traps[16b] shows that it is substantially faster than
ring closure to give the epoxide. Note that both these processes are
substantially faster than the expected rate of spin inversion (ca.
105 sÀ1).[14]
[35] For example, we previously found that a-methylstyrene gave poor
epoxide yields and low ee with complex 14: C. Bousquet, D. G.
Gilheany, unpublished results.
[36] It may be compared to E-3-hexene, which gave 10% yield and 33% ee
in the reaction with 2 (L ˆ Ph3PO),[2e] and E-2-hexene, which gave
20% yield in achiral epoxidation.[5a]
[25] a) A. D. Becke, J. Chem. Phys. 1993, 98, 5648 5652. b) C. Lee, W.
Yang, R. G. Parr, Phys. Rev. B 1988, 37, 785 789.
[26] A. J. H. Wachters, J. Chem. Phys. 1970, 52, 1033 1036.
[27] R. H. Crabtree, The Organometallic Chemistry of the Transition
Metals, 2nd Ed. Wiley, New York, 1994, pp. 302 306.
[37] Although 15 was unknown, the analogous tetraphenyl-substituted
alkene has previously been used in the investigation of epoxidation by
Cytochrome P-450 models: A. J. Castellino, T. C. Bruice, J. Am. Chem.
Soc. 1988, 110, 7512 7519; M.-H. Le Tadic-Biadatti, M. Newcomb, J.
Chem. Soc. Perkin Trans. 2 1996, 1467 1473.
ˆ
[28] Starting electronic configurations with an excited Cr O p bond could
[38] A complication with alkene 15 is the presence of four chiral centers.
The two E isomers were distinguishable from the two Z isomers by
NMR spectroscopy, and all four were distinguishable by GC. Due to a
lack of a large-scale method of separation, 15 was used as the 1:1
mixture of E and Z isomers obtained from the Wittig reaction used for
its synthesis.
ˆ
be generated by slightly elongating the Cr O bond, but during
optimization, the LMCT state reformed. The triplet to quintet
excitation in the Mn system corresponds to breaking the double bond
between the metal and the oxo ligand to generate an oxygen-centered
radical. The Mn- and Cr-based systems thus differ significantly in this
respect.
[39] J. March, Advanced Organic Chemistry, 4th Ed., Wiley, New York,
1992, pp. 1073 1074.
[29] a) T. S. Siddall, N. Miyaura, J. C. Huffman, J. K. Kochi, J. Chem. Soc.
Chem. Commun. 1983, 1185 1186; b) K. Srinivasan, J. K. Kochi,
Inorg. Chem. 1985, 24, 4671 4679.
[40] The catalytic reaction was slow, and substantial residual starting
material remained. In the stoichiometric case, 1H NMR spectroscopy
of the crude reaction mixtures showed that the E alkene reacted
somewhat faster, since when the E alkene had been consumed, 11%
of the Z isomer remained.
[41] In the 1H NMR spectrum of the crude product the only new signals
identifiable were in the methylene (broad multiplet: d ˆ 1.6) and
aldehyde (singlet: d ˆ 10.0) regions.
[30] BARF: [B(3,5-(CF3)2C6H4)4]À.
[31] For a discussion on spin-crossing effects in organometallic chemistry,
see: D. Schrˆder, S. Shaik, H. Schwarz, Acc. Chem. Res. 2000, 33, 139
145.
[32] For recent examples of this type of external heavy-atom effect, see:
Z. S. Romanova, K. Deshayes, P. Piotrowiak, J. Am. Chem. Soc. 2001,
123, 2444 2445; M. A. Anderson, Y. Xu, B. Grissom, J. Am. Chem.
Soc. 2001, 123, 6720 6721.
Received: January 30, 2002
Revised: June 12, 2002 [F3836]
[33] For a leading reference, see: M. Newcomb, Acc. Chem. Res. 2000, 33,
449 455.
Chem. Eur. J. 2002, 8, No. 18
¹ 2002 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
0947-6539/02/0818-4307 $ 20.00+.50/0
4307