Combined experimental and theoretical study of the mechanism and enantioselectivity of palladium-catalyzed intermolecular heck coupling

Article abstract of DOI:10.1021/ja802991y

The asymmetric Heck reaction using P,N-ligands has been studied by a combination of theoretical and experimental methods. The reaction follows Halpern-style selectivity; that is, the major isomer is produced from the least favored form of the pre-insertio

Full text of DOI:10.1021/ja802991y

Published on Web 07/12/2008
Combined Experimental and Theoretical Study of the
Mechanism and Enantioselectivity of Palladium-Catalyzed
Intermolecular Heck Coupling
Signe T. Henriksen,† Per-Ola Norrby,*,‡ Pa¨ivi Kaukoranta,§ and
Pher G. Andersson*,§
Department of Chemistry, Technical UniVersity of Denmark, Building 201, KemitorVet, DK-
2800 Kgs. Lyngby, Denmark, Department of Chemistry, UniVersity of Gothenburg, Kemigården
4, SE-412 96 Go¨teborg, Sweden, and Department of Biochemistry and Organic Chemistry,
Uppsala UniVersity, Box 576, 751 24 Uppsala, Sweden
Received April 23, 2008; E-mail: pon@chem.gu.se; Pher.Andersson@biorg.uu.se
Abstract: The asymmetric Heck reaction using P,N-ligands has been studied by a combination of theoretical
and experimental methods. The reaction follows Halpern-style selectivity; that is, the major isomer is
produced from the least favored form of the pre-insertion intermediate. The initially formed Ph-Pd(P,N)
species prefers a geometry with the phenyl trans to N. However, the alternative form, with Ph trans to P,
is much less stable but much more reactive. In the preferred transition state, the phenyl moiety is trans to
P, but significant electron density has been transferred to the alkene carbon trans to N. The steric interactions
in this transition state fully account for the enantioselectivity observed with the ligands studied. The
calculations also predict relative reactivity and nonlinear mixing effects for the investigated ligands; these
predictions are fully validated by experimental testing. Finally, the low conversion observed with some
catalysts was found to be caused by inactivation due to weak binding of the ligand to Pd(0). Adding
monodentate PPh3 alleviated the precipitation problem without deteriorating the enantioselectivity and led
to one of the most effective catalytic systems to date.
Scheme 1. Heck Reaction of 2,3-Dihydrofuran and Phenyl Triflate
Introduction
The Heck reaction was discovered1 more than 25 years ago
and still stands out as one of the most versatile methods for
creating new carbon-carbon bonds. Its applicability to highly
functionalized substrates confers upon the reaction a very wide
substrate scope, and it has been used in a number of syntheses
of complex natural products.2 More recent development of the
Heck reaction has focused on the possibility of controlling its
enantioselectivity. A number of reports3 have addressed the
application of chiral bidentate phosphine ligands such as BINAP
and chiral N,P-donor ligands to the asymmetric Heck coupling
of alkyl and aryl triflates with various olefins in both intra- and
intermolecular fashion. Since the first report by Hayashi et al.,4
the Heck coupling of 2,3-dihydrofuran (S1) and phenyl triflate
(S2) has been frequently used as a standard reaction to study
the asymmetric intermolecular Heck reaction (Scheme 1). In
this system, N,P-donor ligands yield 2-phenyl-2,5-dihydrofuran
(P1) as the major product, whereas bidentate phosphine ligands
mostly give the rearranged species 2-phenyl-2,3-dihydrofuran
(P2).
We recently reported5 the enantioselective Heck reaction
catalyzed by Pd complexes of a new class of chiral thiazole
phosphine ligands that we originally developed for the iridium-
catalyzed hydrogenation of non-functionalized olefins.6 These
ligands induced a high degree of enantioselectivity in the Heck
† Technical University of Denmark.
‡ University of Gothenburg.
§ Uppsala University.
(1) (a) Mizoroki, T.; Mori, K.; Ozaki, A. Bull. Chem. Soc. Jpn. 1971, 44,
581. (b) Heck, R. F.; Nolley, J. P., Jr. J. Org. Chem. 1972, 37, 2320–
2322.
(4) (a) Ozawa, F.; Kubo, A.; Hayashi, T. J. Am. Chem. Soc. 1991, 113,
1417–1419. (b) Loiseleur, O.; Meier, P.; Pfaltz, A. Angew. Chem.,
Int. Ed. Engl. 1996, 35, 200–202. (c) Sprintz, J.; Helmchen, G.
Tetrahedron Lett. 1993, 34, 1769–1772. (d) Vonmatt, P.; Pfaltz, A.
Angew. Chem., Int. Ed. Engl. 1993, 32, 566–568. (e) Dawson, G. J.;
Frost, C. G.; Williams, J. M. J. Tetrahedron Lett. 1993, 34, 3149–
3150.
(2) Bra¨se, S.; de Meijere, A. In Metal-Catalyzed Cross Coupling Reactions;
de Meijere, A., Diederick, F. Eds.; Wiley-VCH: Weinheim, 2004;
Chap. 5.
(3) (a) For a recent review of Heck reactions, see: Dounay, A. B.;
Overman, L. E. Chem. ReV. 2003, 103, 2945–2963. (b) For a review
of N,P-ligands, see: Guiry, P. J.; Saunders, C. P. AdV. Synth. Catal.
2004, 346, 497–537.
(5) Kaukoranta, P.; Ka¨llstro¨m, K.; Andersson, P. G. AdV. Synth. Catal.
2007, 349, 2595–2602.
9
10414 J. AM. CHEM. SOC. 2008, 130, 10414–10421
10.1021/ja802991y CCC: $40.75
2008 American Chemical Society

Palladium-Catalyzed Intermolecular Heck Coupling
A R T I C L E S
Scheme 2. Mechanism of Intermolecular Heck Reaction between
2,3-Dihydrofuran and Phenyl Triflate
Figure 1. Thiazole (A) and imidazole (B) ligands used in this study.
aryl moiety is trans to the nitrogen.7b On the basis of this finding,
they proposed a selectivity model for the reaction. In another
paper, Guiry et al. evaluated sterically hindered 2-dialkyl-3-
hydrofurans,9 which led to a better understanding of the steric
repulsions in the reaction. However, neither of these studies
offered a complete understanding of the transition state in the
enantiodetermining step.
We now give a full account of the mechanism and a
rationalization of the enantioselectivity for the asymmetric Heck
coupling using palladium catalysts with chelating N,P-ligands.
We have performed a hybrid density functional study (B3LYP)
that covers a large range of possible intermediates and transition
states in the reaction. We have also carried out an experimental
study to gain support for the conclusions drawn from the
theoretical calculations.
coupling between alkyl and aryl triflates and 2,3-dihydrofuran
under microwave (MW) heating.
Results and Discussion
The asymmetric Heck reaction between 2,3-dihydrofuran (S1)
and phenyl triflate (S2) has become the standard test system
and was therefore chosen as the model reaction for our study
of mechanism and enantioselectivity (Scheme 1). In our earlier
report, a catalyst containing the thiazole-based N,P-donor ligand
A was found to be excellent in the Heck coupling of S1 and
S2, so this ligand was chosen for the study (Figure 1). As a
comparison, we also wanted the study to include a ligand having
a similar structure but different electronic properties. Previously,
we reported the use of chiral imidazole phosphine ligands in
the Ir-catalyzed asymmetric hydrogenation of olefins.10 These
ligands are based on the same design as A and give similar
results in the hydrogenation of unfunctionalized olefins, which
confirms the structural similarity of the two ligand types. We
reasoned that the imidazole phosphine B could also be employed
in the asymmetric Heck reaction of S1 and S2. Initial studies
of the Heck reaction with the imidazole ligand B resulted in
high enantioselectivity (93% enantiomeric excess (ee)), but the
reaction was much slower than with the thiazole ligand A.
Because A and B are pseudoenantiomers,11 Heck reactions using
them as ligands demonstrated enantiodescrimination in opposite
directions.
Although admirable results have been obtained in the well-
studied palladium-catalyzed Heck coupling, most studies of its
mechanism have focused on achiral systems. The generally
accepted mechanism of the Heck reaction with 2,3-dihydrofuran
and phenyl triflate has been given in a number of publications
and is shown in Scheme 2.7 The catalytic cycle starts with the
oxidative addition of phenyl triflate to the Pd(0)-(N,P*)
complex formed in situ to give a Pd(II) complex (step I, Scheme
2). Two types of Pd(II) complexes can be formed because the
phenyl group can be positioned trans to either the phosphorus
or the nitrogen in the N,P-ligand. The coordination of 2,3-
dihydrofuran to the Pd(II) complex may occur from two different
faces, resulting in four possible isomeric Pd(II)-π-alkene
complexes (step II). Subsequent alkene insertion into the Pd-Ph
bond (step III) gives an alkyl-Pd(II) complex that undergoes
ꢀ-hydride elimination to form a hydridopalladium olefin com-
plex (step IV); the alkene dissociates from these species to give
the chiral product. The active Pd(0) complex is finally regener-
ated by reductive deprotonation with the aid of base (step V).8
Considering the large number of reports that have dealt with
the development of the asymmetric Heck reaction, there are
comparatively few reports that discuss the factors governing
the enantiodetermining step of the reaction. So far, all attempts
to rationalize the enantioselectivity have been based on ligand
structure-enantioselectivity relationships. In one of the few
reports that aimed to understand the origin of the reaction’s
enantioselectivity, Uemura et al. isolated the chiral (N,P)-Pd(II)
intermediate formed after oxidative addition of p-carbomethox-
yphenyl triflate and, using NMR spectroscopy, showed that the
Computational Study. Initially, we evaluated the relevant
parts of the catalytic cycle using thiazole ligand A. The free
energy profile is shown in Figure 2. We have concentrated on
steps in which bonds to the hydrofuran moiety are either formed
or broken (colored in Figure 2). These include the migratory
insertion and the ꢀ-hydride elimination. For completeness, we
have also included intermediates (but not transition states) in
(9) Kilroy, T. G.; Hennessy, A. J.; Connolly, D. J.; Malone, Y. M.; Farrell,
A.; Guiry, P. J. J. Mol. Catal. A: Chem. 2003, 196, 65–81.
(10) Kaukoranta, P.; Engman, M.; Hedberg, C.; Bergquist, J.; Andersson,
P. G. AdV. Synth. Catal. 2008, 350, 1168–1176.
(6) (a) Ka¨llstro¨m, K.; Hedberg, C.; Bayer, A.; Brandt, P.; Andersson, P. G.
J. Am. Chem. Soc. 2004, 126, 14308–14309. (b) Hedberg, C.;
Ka¨llstro¨m, K.; Brandt, P.; Hansen, L. K.; Andersson, P. G. J. Am.
Chem. Soc. 2006, 128, 2995–3001.
(11) The term “pseudoenantiomers” was coined by Prof. K. B. Sharpless
for ligands that are more dissimilar than true enantiomers, but that
still give a similar magnitude (but opposite sense) of enantioselectivity
in asymmetric reactions: Pearlstein, R. M.; Blackburn, B. K.; Davis,
W. M.; Sharpless, K. B. Angew. Chem., Int. Ed. Engl. 1990, 29, 639–
641.
(7) (a) Loiseleur, O.; Hayashi, M.; Keenan, M.; Schmees, N.; Pfaltz, A.
J. Organomet. Chem. 1999, 576, 16–22. (b) Yonehara, K.; Mori, K.;
Hashizume, T.; Chung, K.-G.; Ohe, K.; Uemura, S. J. Organomet.
Chem. 2000, 603, 40.
(8) Trost, B. M. Chem. Eur. J. 1998, 4, 2405–2412.
9
J. AM. CHEM. SOC. VOL. 130, NO. 31, 2008 10415

A R T I C L E S
Henriksen et al.
Figure 2. Free energy profile for the asymmetric Heck reaction between 2,3-dihydrofuran and phenyl triflate with the thiazole ligand A. Palladium complexes
2-8 are cationic.
the remainder of the catalytic cycle (black in Figure 2). The
normalized Gibbs free energy values for all calculated inter-
mediates and transition states are shown in Table 1. All the
steps are explained and discussed in detail below.
a higher degree of destabilization in this case. This is in
accordance with the Pd-P bond having a larger trans influence
than the Pd-N bond.
Step II. 2,3-Dihydrofuran Coordination. The next step is the
formation of a π complex between Pd and the alkene. Because
2,3-dihydrofuran (DHF) is not a regular alkene, but an enol
ether, it coordinates to Pd almost exclusively through C4, due
to the excess of electron density on this atom. The phosphine
and phenyl moieties still prefer to be cis to each other; however,
the energy difference between cis/trans isomers of complex 3
is smaller than for 2, because the π-coordinated alkene also
exerts a trans influence. Again, this is reflected in the relevant
bond lengths (Table 2). Complexes 3a and 3b, which have Ph
and N positioned trans to each other, show an elongation of
the Pd-N bonds due to the trans influence of Ph, as well as an
elongation of the coordination between Pd and the alkene caused
by the trans influence of P. The complexes 3c and 3d, with Ph
and P trans to each other, have only one elongated bond, the
Pd-P bond. Thus, in the case of complex 3, we have either
two weakly destabilized bonds or one significantly destabilized
bond. The energies (Table 1) tell us that the first situation is
favored by around 40 kJ/mol.
Step I. Oxidative Addition of Phenyl Triflate. In the initial
Pd(0) complex 1, it is beneficial to have an explicit solvent
(THF) molecule coordinated to Pd, even though this means
breaking the Pd-N bond to leave the ligand coordinating in a
monodentate fashion, as 1b is favored over 1a by 17 kJ/mol
(Table 1). We have previously investigated the oxidative
addition in detail,12 and, as it is not expected to influence the
enantioselectivity of the reaction, the transition state (TS) for
this step has not been modeled here. There is a huge energy
difference of 93 kJ/mol (Table 1) between the two possible
oxidative addition products 2a and 2b, which have the phenyl
group cis and trans to phosphorus, respectively. This energy
difference can be ascribed to the large trans influences of both
the phosphine and phenyl ligands, which makes these ligands
prefer to be cis to each other.
The large trans influence of the phenyl group (Ph) is also
obvious from the elongation of the Pd-P and Pd-N bonds when
they are positioned trans to Ph (Table 2). Though the Pd-C
bond is more or less the same length in complexes 2a and 2b,
the Pd-P bond length is increased by 0.17 Å when going from
2a to 2b, whereas the Pd-N bond length is increased by 0.10
Å when going from 2b to 2a. Even though both the Pd-N and
Pd-P bonds lengthen when positioned trans to the phenyl
group, the Pd-P bond is elongated to a larger extent, indicating
Step III. Migratory Insertion. In the next step, the Ph group
is transferred from Pd to the alkene, with concomitant breaking
of the alkene π bond and formation of two new σ bonds. This
means that, from a complex having a strongly destabilizing
C-Pd bond trans to P, 3c and 3d, we now get a complex with
an almost empty site trans to P (no full bond between the phenyl
group and Pd is present), which is very energetically favorable.
Therefore, the π complex of highest energy (3d) is converted
into the most stable σ complex (5d). The stabilizing effect of
(12) (a) Ahlquist, M.; Norrby, P.-O. Organometallics 2007, 26, 550–553.
(b) Ahlquist, M.; Fristrup, P.; Tanner, D.; Norrby, P.-O. Organome-
tallics 2006, 25, 2066–2073.
9
10416 J. AM. CHEM. SOC. VOL. 130, NO. 31, 2008

Palladium-Catalyzed Intermolecular Heck Coupling
A R T I C L E S
Table 1. Gibbs Free Energies in Solution for the Compounds
except that here the situation is reversed. The transformation
starts out with a very stable σ complex with an almost empty
site trans to P (only a weak agostic interaction is present).
Transferring the hydride to this empty site gives a product
complex (7d) that is high in energy due to the trans relationship
of the Pd-H and the Pd-P bonds, which both exert strong trans
influences. Again, this uphill transformation is already reflected
in the ꢀ-hydride elimination TS (7), which is of highest energy
for the two reaction paths (7c and 7d) in which H ends up trans
to P.
Enantiodetermining Step: Migratory Insertion (Step III). It
appears from the reaction profile that, in all cases, the insertion
TS (4) is higher in energy than the ꢀ-hydride elimination TS
(7). The insertion step is therefore expected to be irreversible,
which means that only this step, in which the stereochemistry
is set, will influence the enantioselectivity of the reaction.
Because the subsequent steps have lower energy barriers, the
reaction will follow the path that goes through the insertion TS
of lowest energy, and the enantioselectivity of the reaction is
therefore governed by the relative energies of the four possible
insertion TSs. In this case, series d (purple graph in Figure 2)
describes the reaction path with the insertion TS of lowest
energy. This reaction path leads to the formation of the R
product, which is indeed observed experimentally. Furthermore,
the high enantioselectivity of the reaction (88% ee) is reflected
in the large energy difference of 27 kJ/mol (Table 1) between
TS 4d and the lowest energy TS for the S product, 4c.
Shown in Figure 2
a
Gsol (kJ/mol)
b
complex
∆Gsol (kJ/mol)
1a
1b
2a
2b
3a
3b
3c
3d
4a
4b
4c
4d
5a
5b
5c
5d
6a
6b
6c
6d
7a
7b
7c
7d
8a
8b
8c
8d
P1
114
97
0
93
56
51
93
95
135
123
132
105
79
60
40
16
53
46
32
51
55
56
120
96
41
40
17
0
0
93
5
0
42
44
30
18
27
0
63
44
25
0
31
14
0
19
0
1
65
41
1
0
63
45
103
85
18
We have discussed the electronic effects that are the basis
for the energy difference between the cis/trans isomeric
complexes. However, electronic effects cannot account for the
fact that the insertion TS giving the R enantiomer is preferred
over the TS giving the S enantiomer, because the considerations
concerning the trans effect are just as valid for both competing
reaction paths (series c and d, Figure 2). Whereas electronic
effects account for the relative energy of the different cis/trans
isomers, the relative energy of the diastereomers is determined
by steric effects. Figure 3 shows the diastereomeric pair of
insertion TSs that have P trans to the phenyl group (4c and
4d), both as ball-and-stick and as space-filling models. In both
4c and 4d, the phenyl group on the thiazole ring is twisted out
of conjugation due to steric strain in the molecule. However,
there is significantly less steric strain in TS 4d, which leads to
the R product, than in TS 4c, which gives the S product. This
is most evident in the space-filling TS models, which show
several unfavorable van der Waals interactions between the two
interacting phenyl groups in TS 4c, and none in TS 4d. These
steric effects are expected to account for the energy difference
of 27 kJ/mol (Table 1) between the two TSs. The steric crowding
connected with the formation of a S stereocenter and the
unfavorable electronic features of having trans P and C moieties
make TS 4a the most energetically disfavored of all the insertion
TSs (Figure 2, Table 1).
Comparison with Imidazole Ligand B. In studying the effect
of imidazole ligand B on the Heck reaction, we evaluated only
the selectivity-determining insertion TS and the intermediates
preceding it on the reaction path. The free energy profile for
these steps is shown in Figure 4, and the corresponding Gibbs
free energies are presented in Table 3. The overall reaction paths
with the two ligands A and B are very similar. A minor
difference, however, is that the coordination of an explicit
solvent molecule to the Pd(0) complex and consequent mono-
dentate coordination of the ligand are not energetically favorable
for the imidazole ligand (9, Figure 4), in contrast to the thiazole
a Gibbs free energies relative to the complex of lowest energy, 2a.
b Gibbs free energy for each complex relative to the most stable
complex of that type (same numbers, a-d).
Table 2. Bond Lengths (Å) between Pd and Its Four Coordinating
Atoms for 2a,b and 3a-d
complex
Pd-P
Pd-N
Pd-C(Ph)
Pd-C(DHF)
2a
2b
3a
3b
3c
3d
2.27
2.44
2.33
2.35
2.49
2.45
2.20
2.10
2.32
2.28
2.22
2.23
2.00
1.98
2.02
2.03
2.03
2.07
2.42
2.41
2.27
2.23
removing the phenyl group from the position trans to P can
already be identified in the insertion TS, which means that the
π complex of highest energy has the lowest TS. This is an
example of a Halpern-type selectivity, in which the least favored
intermediate leads to the most favored product, as first dem-
onstrated for the Rh-catalyzed hydrogenation of enamides.13
Step IV. ꢀ-Hydride Elimination. It is clear from the reaction
paths in Figure 2 that, once the reaction has reached the σ
complex 5, with coordination between the phenyl group and
Pd, it can never revert, as all possible forward processes have
lower activation barriers than the reverse reactions. The forward
reactions will proceed by a 120° rotation about the Pd-C σ
bond, forming the agostic intermediates 6. From here, ꢀ-hydride
elimination via TS 7 leads to hydride complexes 8. We have
not investigated the various possibilities for further reaction of
8, because experimental results show that it is deprotonated by
the external base before any potential isomerizations. This gives
the 2-phenyl-3,4-dihydrofuran product (P1) and re-forms the
Pd(0)-ligand complex needed to close the catalytic cycle.
The same trans influence considerations apply to the ꢀ-hy-
dride elimination step (IV) as for the carbopalladation step (III),
(13) Landis, C. R.; Halpern, J. J. Am. Chem. Soc. 1987, 109, 1746–1754.
9
J. AM. CHEM. SOC. VOL. 130, NO. 31, 2008 10417

A R T I C L E S
Henriksen et al.
Figure 3. Transition states 4c (left) and 4d (right) (migratory insertion).
Figure 4. Free energy profile for the asymmetric Heck reaction between 2,3-dihydrofuran and phenyl triflate, catalyzed by a Pd complex with the imidazole
ligand B. Palladium complexes 10-12 are cationic.
case (1, Figure 2). The oxidative addition product 10 shows
the same large preference (88 kJ/mol, Table 3) for a trans
relationship between the Ph and N as did 2. This preference is
retained, though attenuated, in the π complex 11. As before,
the energetically favored insertion TS is the one having P trans
to the phenyl group that is being transferred from Pd to the
double bond and leading to the S product.
the [L-Pd-Ph]+ cation with the phenyl group trans to N, but
the most preferred reaction path in each case requires the resting
state to first undergo a configurational switch to place the phenyl
group trans to P. The overall barrier from the resting state to
the selectivity-determining carbopalladation TS is also similar
in both cases: 105 kJ/mol using ligand A and 103 kJ/mol using
ligand B. However, we were also interested to see whether direct
competition between the two pseudoisomeric ligands for a
limited amount of Pd would produce a nonlinear effect; that is,
To summarize, the two ligands show very similar behavior.
Of the steps calculated here, the resting state in each case is
9
10418 J. AM. CHEM. SOC. VOL. 130, NO. 31, 2008

Palladium-Catalyzed Intermolecular Heck Coupling
A R T I C L E S
Table 3. Gibbs Free Energies in Solution for the Compounds
Shown in Figure 4
Table 4. Heck Reaction of 2,3-Dihydrofuran and Phenyl Triflate
Utilizing Ligands A and Ba
a
Gsol (kJ/mol)
b
complex
∆Gsol (kJ/mol)
9a
9b
115
124
0
88
55
49
89
84
126
118
130
103
0
9
0
88
6
0
40
35
23
15
27
0
10a
10b
11a
11b
11c
11d
12a
12b
12c
12d
a Gibbs free energies relative to the complex of lowest energy, 10a.
b Gibbs free energy for each complex relative to the most stable
complex of that type (same numbers, a-d).
[Pd]
(mol %)
ligand A
(mol %)
ligand B
(mol %)
convb
(%)
P1 eed
(%)
entry
P1:P2c
1
2
3
4
3
3
3
3
6
>99 (82)
92:8
92:8
99:1
88 (R)
94 (S)
87 (R)
93 (S)
6
36
22
44
whether the concentration ratio of ligands A and B affects the
enantiomeric ratio (er) of the chiral product P1 linearly. To
connect the two studies, we compared the two initial Pd-L,
complexes 1a and 9a with ligands A and B, respectively, and
found that coordination of Pd to ligand B is favored by 9 kJ/
mol. At the Pd(II) stage, the difference is even larger: ligand
B is always preferred over ligand A, with energy differences
ranging from 11 to 23 kJ/mol for corresponding points up to
and including the carbopalladation steps (4 and 12, respec-
tively). Thus, in a mixed system, where A and B compete
for Pd, we predict that the reaction selectivity will be
dominated by ligand B.
3.2
3.2
90:10
a Reaction conditions: 2,3-dihydrofuran (2.0 mmol), PhOTf (0.5
mmol), DIPEA (1.0 mmol), Pd2(dba)3 (1.5 mol %), ligand A or B (mol
%, see table), THF (3 mL), MW 120 °C, 4 h. b Determined by GC with
n-tridecane as internal standard. Isolated yield in parentheses.
c Determined by chiral GC/MS. d Determined by chiral GC/MS. The
conversions and ee values are presented as the mean values of two
reactions.
Scheme 3. Competitive Study with Ligands A and B in the Heck
Reaction
Experimental Study. The theoretical study indicated that Pd
complexes of both ligands should catalyze the Heck reaction
with similar magnitudes of asymmetric induction, and that these
Heck reactions should proceed at similar rates for both ligands.
We also found that imidazole B binds more strongly to
palladium and therefore would be expected to dominate the
reaction in a mixed system if the ligands were competing for
Pd; that is, there should be a strong nonlinear effect upon mixing
the pseudoenantiomers. In order to confirm these predictions,
we performed a simple experimental study utilizing ligands A
and B in the Heck reaction. The Heck coupling of 2,3-
dihydrofuran S1 and phenyl triflate S2 using Pd2(dba)3 as
catalyst precursor was studied using the Hu¨nig’s base and
microwave heating. Opposite enantiomers of thiazole ligand A
and imidazole ligand B were synthesized for the study.
Surprisingly, the two ligands showed quite different behavior
in the experiments. A Heck reaction run in the presence of the
thiazole ligand A (A:Pd ) 2:1) went to completion within 4 h
and was highly enantioselective (Table 4, entry 1). The
corresponding reaction using the imidazole ligand B proceeded
to only 36% conversion, but with even higher enantioselectivity
(Table 4, entry 2). However, when thiazole A was used in equal
amounts with palladium under otherwise identical reaction
conditions, the reaction went to only 22% conversion (Table 4,
entry 3). The Heck reactions with imidazole ligand B gave
nearly identical results, regardless of the ligand-to-palladium
ratio (Table 4, entries 2 and 4). Precipitation of palladium black
was observed in these reactions, and we therefore conclude that
the reactions that give low conversion do so because of catalyst
inactivation early in the reaction. It seems that only ligand A is
able to keep Pd(0) from precipitating, and only when used in 2
equiv relative to Pd.
Scheme 4. Heck Reaction with Imidazole B as Ligand and PPh3
as Additive
mixture (Scheme 3). Because different enantiomers of the
ligands A and B were used and these result in different
enantiomers of the product, it was possible to see which of the
ligands was forming the active chiral catalyst. In a perfectly
linear system, the er of the product should equal the ratio of
the two pseudoenantiomeric ligands, but from the computational
study, we expect ligand B to dominate in the present system.
Indeed, the competitive reaction with ligands A and B gave the
same sense of enantiodiscrimination to give (S) product and an
ee value almost as high as that obtained in reactions run with
ligand B alone (80%, cf. 93% for B alone). This result confirms
that the imidazole ligand B is forming the active chiral catalyst
by binding more strongly to the Pd than does the thiazole ligand
A. To our surprise, the reaction went to completion, which is
not the case when imidazole is the only ligand. One reason for
this could be that the free thiazole is stabilizing the “naked”
In order to confirm the predicted nonlinear effects, the Heck
reaction reaction was run with both A and B in the reaction
9
J. AM. CHEM. SOC. VOL. 130, NO. 31, 2008 10419

A R T I C L E S
Henriksen et al.
Figure 5. Isodesmic comparison of transition states with and without chiral ligand.
Figure 6. Isodesmic comparison of the Pd(0) complexes that can form when both chiral ligands, A and B, and PPh3 are present in the reaction mixture.
Pd(0) complex. To evaluate this possibility, we tested the
reaction with imidazole ligand B and PPh3 as an additive
(Scheme 4). Most gratifyingly, this led to high conversion (93%)
and the same ee (93%) as imidazole B gives in the absence of
PPh3.
Though the addition of a nonchiral ligand could have led to
a deterioration of the enantioselectivity by opening a parallel,
nonselective path, this appears not to be the case. We verified
this computationally by calculating the two pathways for Heck
reaction without any chiral moiety possible in this system: a
pathway with two PPh3 ligands, or a pathway with one PPh3
and one THF solvent molecule coordinated to Pd. The results
are shown as an isodesmic comparison of the relevant transition
states in Figure 5. Clearly, the reaction is completely dominated
even by the relatively weak ligand A. The same comparison
with ligand B would produce energy differences that were larger
by 13 kJ/mol (i.e., 89 and 136 kJ/mol).
Finally, the beneficial effect of PPh3 is shown in Figure 6.
Pd(0) strongly prefers to be ligated by two phosphines rather
than by the P and N termini of A and B. Coordination by two
phosphines leads to complexes that resist precipitation due to
their high stability. Importantly, the stabilization occurs only
for the high-energy Pd(0) state. The Pd(II) complexes are too
crowded to allow facile coordination of the relatively bulky PPh3
ligand; thus, we see no detrimental effect of PPh3 on reaction
rates.
effects, which were predicted on the basis of the relative binding
strengths of two ligands, were verified experimentally. Though
some experiments were inefficient due to catalyst precipitationsan
effect that cannot easily be addressed computationallyscatalyst
deactivation could be avoided without deterioration of the high
stereoselectivity by addition of an achiral ligand. The effect of
the achiral ligand was rationalized by additional computational
studies. The role of the achiral ligand as a stabilizing but
nonparticipating agent could be extremely useful in other metal-
catalyzed processes, the turnover of which is limited by catalyst
inactivation due to precipitation.
Experimental Section
Computational Details. All calculations were performed using
Jaguar, version 6.0, release 12.14 We employed the B3LYP hybrid
functional15–17 with the LACVP**basis set,18 which uses an
effective core potential for Pd and the 6-31G** basis set for the
other atoms.
All complexes were optimized in the gas phase, and single-point
solvation energies were then determined using parameters suitable
for THF (dielectric constant, epsout ) 7.43; probe radius, radprb
) 2.5241416). Vibrational analysis using the analytic Hessian was
performed for the gas-phase geometries, and the final free energies
(Gsol) were obtained by adding the thermodynamic contributions
(at 393.15 K) to the solution-phase potential energies. The Cartesian
coordinates for all optimized geometries are available in the
Supporting Information.
An achiral ligand can successfully stabilize a highly selective
and active catalyst that suffers from metal precipitation, as long
as this achiral ligand does not produce an active catalyst. This
concept could have wide-ranging implications for homogeneous
catalysis. Many reactions catalyzed by noble metals are plagued
by problems of precipitation. Though adding ligand that binds
well to the low-valent state might inhibit the reaction slightly,
it can also confer a definite advantage by inhibiting metal
precipitation, thus yielding a much more effective catalytic
system overall. Naturally, this concept requires the stabilizing
ligands to be incompetent in the selectivity-determining step
of the catalytic cycle. The use of PPh3 in the Heck reaction
fulfills this requirement beautifully, and we envision that the
same principle could be applied to many other metal-catalyzed
processes.
Experimental Details. THF was freshly distilled from sodium
benzophenone ketyl under N2 prior to use. 2,3-Dihydrofuran, phenyl
triflate, Pd2(dba)3, and tridecane were purchased from Sigma-
Aldrich and used as received. Diisopropylethylamine (DIPEA) was
distilled from ninhydrin and then from potassium hydroxide.
Ligands A6b and ligand B10 were synthesized according to reported
procedures. Flash chromatography was performed using silica gel
60 Å (37-70 µm). Analytical thin-layer chromatography was
carried out utilizing 0.25 mm precoated plates, silica gel 60 UV254
,
and spots were visualized under UV light. NMR samples were
dissolved in CDCl3 and analyzed at room temperature; 1H (500
MHz) and 13C (126 MHz) NMR spectra were recorded on a 500
MHz spectrometer. Chemical shifts for protons were referenced to
the shift of the residual CHCl3 (δ 7.26). Carbon spectra were
referenced to the shift of the 13C signal of CDCl3 (δ 77.0).
Conversions and ee values were determined by chiral GC-MS (B-
DM) using the following method: 80 °C, 0.3 °C/min 90 °C, 5 °C/
min 125 °C, 14.5 psi, 1.5 mL/min. The retention times (tR, min)
Conclusions
Using DFT calculations, we have elucidated the source of
the high enantioselectivity conferred in the Heck reaction by
the current class of P,N-ligands. The reactions follow a Halpern-
style selectivity. It is important to note that, in cases like this,
structural outcomes cannot be rationalized on the basis of
observable complexes, because the major enantiomeric product
is formed from the least populated intermediate. The nonlinear
(14) Jaguar 6.0; Schrodinger, LLC: Portland, OR, 2005; for the most recent
version, see www.schrodinger.com.
(15) Lee, C.; Yang, W.; Parr, R. G. Phys. ReV. B 1988, 37, 785–789.
(16) (a) Becke, A. D. J. Chem. Phys. 1993, 98, 1372–1377. (b) Becke,
A. D. J. Chem. Phys. 1993, 98, 5648–5652.
(17) Stephens, P. J.; Devlin, F. J.; Chabalowski, C. F.; Frisch, M. J. J.
Phys. Chem. 1994, 98, 11623–11627.
(18) Hay, P. J.; Wadt, W. R. J. Chem. Phys. 1985, 82, 299–310.
9
10420 J. AM. CHEM. SOC. VOL. 130, NO. 31, 2008

Palladium-Catalyzed Intermolecular Heck Coupling
A R T I C L E S
are as follow: phenyltriflate, 6.6; tridecane, 32.0; (S)-P1, 32.6; (R)-
P1, 35.3; (S)-P2, 27.3; (R)-P2, 29.5. Tridecane was used as an
internal standard for determining conversion of the reaction.
Microwave heating was carried out using an automatic single-mode
synthesizer from Biotage, which produces a radiation frequency
of 2.45 GHz.
conversion determination before the mixture was MW-heated at
120 °C for 4 h. After cooling, the mixture was diluted with Et2O
and filtered through silica. The filtrate was analyzed by chiral GC-
MS (B-DM) (>99% conversion, 80% ee (S), P1:P2 (98:2)). The
filtrate was concentrated in Vacuo to give the crude product, which
was purified by flash column chromatography on silica gel using
CH2Cl2:pentane (1:1) as the eluent to afford a clear oil as the pure
Representative Procedure for MW-Assisted Asymmetric
Intermolecular Heck Reaction of 2,3-Dihydrofuran and
Phenyl Triflate (Table 4, Entry 1). Ligand A (12.4 mg, 0.03
mmol) and Pd2(dba)3 (6.8 mg, 0.0075 mmol) were weighed into a
MW vial, and dry THF (3 mL) was added. The vial was sealed
and then evacuated and backfilled with N2 (three times). The
mixture was gently heated to reflux and then allowed to cool to
room temperature. Phenyl triflate (81 µL, 0.5 mmol, 1 equiv), 2,3-
dihydrofuran (151 µL, 2.0 mmol, 4 equiv), DIPEA (174 µL, 1
mmol, 2 equiv), and tridecane (30 µL, 0.125 mmol, 0.25 equiv)
were added. A small sample was removed for conversion deter-
mination before the mixture was MW-heated at 120 °C for 4 h.
After cooling, the mixture was diluted with Et2O and filtered
through silica. The filtrate was analyzed by chiral GC-MS (B-DM).
The conversion of the reaction was >99% with 87% ee (S). The
filtrate was concentrated in Vacuo to give the crude product, which
was purified by flash column chromatography on silica gel using
CH2Cl2:pentane (1:1) as the eluent to afford a clear oil as the pure
1
product (0.065 g, 89% yield). The H NMR spectrum of the pure
compound corresponded to the reported data.19
Asymmetric Heck Reaction with Ligand B and Additive
PPh3 (Scheme 4). Ligand B (11.9 mg, 0.03 mmol) and Pd2(dba)3
(6.8 mg, 0.0075 mmol) were weighed into a MW vial, and dry
THF (2 mL) was added. The vial was sealed and then evacuated
and backfilled with N2 (three times). The mixture was gently heated
and then allowed to cool to room temperature. A solution of
triphenylphosphine (7.9 mg, 0.03 mmol) in dry THF (1 mL) was
added, and the mixture was stirred for 10 min at room temperature.
Phenyl triflate (81 µL, 0.5 mmol, 1 equiv), 2,3-dihydrofuran (151
µL, 2.0 mmol, 4 equiv), DIPEA (174 µL, 1 mmol, 2 equiv), and
tridecane (30 µL, 0.125 mmol, 0.25 equiv) were added. A small
sample was taken for conversion determination before the mixture
was MW-heated at 120 °C for 4 h. After cooling, the mixture was
diluted with Et2O and filtered through silica. The filtrate was
analyzed by chiral GC-MS (B-DM) (93% conversion, 93% ee (S),
P1:P2 (99:1)). The filtrate was concentrated in Vacuo to give the
crude product, which was purified by flash column chromatography
on silica gel using CH2Cl2:pentane (1:1) as the eluent to afford a
1
product (0.060 g, 82% yield). The H NMR spectrum of the pure
product corresponded to the reported data.19
Competitive Asymmetric Heck Reaction Utilizing Ligands
A and B (Scheme 3). Ligands A (12.4 mg, 0.03 mmol) and B
(11.9 mg, 0.03 mmol) were weighed into separate vials, and dry
THF (1.5 mL) was added to each vial. The two solutions were
mixed together and then added to a MW vial containing Pd2(dba)3
(6.8 mg, 0.0075 mmol). The vial was sealed and then evacuated
and backfilled with N2 (three times) before the mixture was heated
gently. After the mixture cooled, phenyl triflate (81 µL, 0.5 mmol,
1 equiv), 2,3-dihydrofuran (151 µL, 2.0 mmol, 4 equiv), DIPEA
(174 µL, 1 mmol, 2 equiv), and tridecane (30 µL, 0.125 mmol,
0.25 equiv) were added. A small sample was removed for
1
clear oil as the pure product (0.059 g, 81% yield). The H NMR
spectrum of the pure compound corresponded to the reported data.19
Acknowledgment. We thank The Swedish Research Council
(VR). We also thank Dr. T. L. Church for careful reading of the
manuscript and for providing the starting material for the synthesis
of the thiazole ligand A.
Supporting Information Available: Cartesian coordinates for
all optimized geometries. This material is available free of
charge via the Internet at http://pubs.acs.org.
(19) Loiseleur, O.; Hayashi, M.; Schmees, N.; Pfaltz, A. Synthesis 1997,
11, 1338–1345.
JA802991Y
9
J. AM. CHEM. SOC. VOL. 130, NO. 31, 2008 10421