You are on page 1of 5

Organometallics 2009, 28, 62016205

DOI: 10.1021/om900674y

6201

DFT-Based Explanation of the Effect of Simple Anionic Ligands on the


Regioselectivity of the Heck Arylation of Acrolein Acetals
Signe Teuber Henriksen, David Tanner, Sandro Cacchi, and Per-Ola Norrby*,

Department of Chemistry, Technical University of Denmark, Building 201, Kemitorvet, DK-2800 Kgs.
Lyngby, Denmark, Dipartimento di Studi di Chimica e Tecnologia delle Sostanze Biologicamente Attive,
Universit
a degli Studi La Sapienza, P.le A. Moro 5, 00185, Rome, Italy, and Department of Chemistry,
University of Gothenburg, Kemigarden 4, SE-412 96 G
oteborg, Sweden
Received July 30, 2009

The Heck arylation of acrolein acetal has been studied computationally and compared to the
corresponding reaction with allyl ethers. The reaction can be controlled to give either cinnamaldehydes or arylpropanoic esters by addition of different coordinating anions, acetate, or chloride. The
computational study reveals that coordinating acetate raises the energy of an intermediate sufficiently to block the access to an otherwise favorable -hydride elimination. The reaction path is also
compared to that of allyl ethers, which always give significant amounts of cinnamyl ether products
under all reaction conditions. The difference between the two substrate classes could be rationalized
in terms of relative hydride donating power of the two substrates.

Introduction
Since cinnamaldehyde and its derivatives are industrially
important as flavor additives and as antimicrobial agents,1 it is
of interest to develop efficient and general methods for the
synthesis of such compounds. The Heck arylation2 of acrolein
is an attractive option, but in practice this reaction is hampered
by extensive polymerization of the starting material at the
reaction temperatures usually required for the cross-coupling
process. A better choice is to use acrolein acetals as substrates,
but this introduces a regioselectivity issue, since there are now
two possible -elimination pathways from the organopalladium intermediate (A, Scheme 1), usually leading to a mixture
of coupling products.3 Recently, however, it has been shown
that the regioselectivity of the -elimination step of this
*To whom correspondence should be addressed. Tel: 46 31 7723848.
Fax: 46 31 7723840. E-mail: pon@chem.gu.se.
(1) (a) Friedman, M.; Kozukue, N.; Harden, L. A. J. Agric. Food
Chem. 2000, 48, 57025709. (b) Cocchiara, J.; Letizia, C. S.; Lalko, J.;
Lapczynski, A.; Api, A. M. Food Chem. Toxicol. 2005, 43, 867923. (c)
Burt, S. Int. J. Food Microbiol. 2004, 94, 223253.
(2) (a) Heck, R. F.; Nolley, J. P., Jr. J. Org. Chem. 1972, 37, 2320
2322. (b) For a recent review on the Heck reaction see: Knowles, J. P.;
Whiting, A. Org. Biomol. Chem. 2007, 5, 3144.
(3) Zebovitz, T. C.; Heck, R. F. J. Org. Chem. 1977, 42, 39073909.
(4) Battistuzzi, G.; Cacchi, S.; Fabrizi, G. Org. Lett. 2003, 5, 777780.
(5) Battistuzzi, G.; Cacchi, S.; Fabrizi, G.; Bernini, R. Synlett 2003,
11331136.
(6) For some examples of computational studies of Heck reactions, see:
(a) Siegbahn, P. E. M.; Str
omberg, S.; Zetterberg, K. Organometallics 1996,
15, 55425550. (b) von Schenck, H.; kermark, B.; Svensson, M. J. Am. Chem.
Soc. 2003, 125, 35033508. (c) Deeth, R. J.; Smith, A.; Brown, J. M. J. Am.
Chem. Soc. 2004, 126, 71447151. (d) Balcells, D.; Maseras, F.; Keay, B. A.;
Ziegler, T. Organometallics 2004, 23, 27842796. (e) Datta, G. K.; von
Schenck, H.; Hallberg, A.; Larhed, M. J. Org. Chem. 2006, 71, 38963903.
(f) Kozuch, S.; Shaik, S. J. Am. Chem. Soc. 2006, 128, 33553365. (g) Hansen,
A.-L.; Ebran, J.-P.; Skrydstrup, T.; Ahlquist, M.; Norrby, P.-O. Angew. Chem.,
Int. Ed. 2006, 45, 33493353. (h) Lee, M. T.; Lee, H. M.; Hu, C.-H.
Organometallics 2007, 26, 13171320. (i) Surawatanawong, P.; Hall, M. B.
Organometallics 2008, 27, 62226232.
r 2009 American Chemical Society

particular Heck reaction can be controlled by the use of


appropriate bases and additives (Scheme 1), allowing selective
formation of either cinnamaldehydes4 or -arylpropanoic
esters5 after hydrolytic workup (products B and C, Scheme 1).
Several aspects of the Heck reaction have been studied
theoretically in recent years.6,7 We have previously made a
combined experimental and computational study of the
closely related Heck cross-coupling of allyl ethers and aryl
halides,8 which, in contrast to the Heck arylation of acrolein
acetals, always favors the cinnamyl ether product over the
regioisomeric enol ether product (Scheme 2). With an acetate
base the reaction yields the regioisomeric products in the
ratio 8:1, while the ratio is approximately 2:1 when an
alternative base (such as triethyl amine) is used.
The reasons for this regioselectivity were not obvious, since
the product distribution was shown experimentally to be kinetically controlled, whereas DFT calculations indicated that the
-elimination transition state leading to the major product is
actually of higher energy than the alternative transition state
leading to the minor regioisomer. Intriguingly, the computational study showed that, instead of arising from a direct competition between the two possible -elimination transition states,
the regioselectivity could be ascribed to a high-energy barrier
for the interconversion of the agostic intermediates, resulting in
preferred elimination to the side where the aryl group was
inserted in the carbopalladation step (cf. the agostic intermediate 5, Figures 1 and 2).8 Other possible explanations for the
observed regioselectivity9 could be discounted after a thorough
computational investigation of the reaction pathway.
(7) Henriksen, S. T.; Norrby, P.-O.; Kaukoranta, P.; Andersson, P.
G. J. Am. Chem. Soc. 2008, 130, 1041410421.
(8) Ambrogio, I.; Fabrizi, G.; Cacchi, S.; Henriksen, S. T.; Fristrup,
P.; Tanner, D.; Norrby, P.-O. Organometallics 2008, 27, 31873195.
(9) (a) Kang, S.-K.; Lee, H.-W.; Jang, S.-B.; Kim, T.-H.; Pyun, S.-J.
J. Org. Chem. 1996, 61, 26042605. (b) Ono, K.; Fugami, K.; Tanaka, S.;
Tamaru, Y. Tetrahedron Lett. 1994, 35, 41334136.
Published on Web 10/13/2009

pubs.acs.org/Organometallics

6202

Organometallics, Vol. 28, No. 21, 2009

Henriksen et al.

Scheme 1. Heck Arylation of Acrolein Acetala

a
In the presence of Bu4NOAc, K2CO3, and KCl the cinnamaldehyde product B is formed as the main product (81%) and C is formed as the minor
product (5%). With Bu4NCl and Bu3N the aryl propanoate ester C is formed exclusively (91%).

Figure 1. Free energy profile of the Heck phenylation of acrolein methyl acetal with acetate as the ligand on palladium.
Scheme 2. Heck Arylation of Allyl Ethera

a
With and without an acetate base the product ratio is 8:1 and 2:1,
respectively.8

In the present paper we present the results of a computational study of the Heck arylation of acrolein acetals, with a
detailed comparison to the previously studied Heck arylation of allyl ethers (Scheme 2).8 The experimental behavior of
the two reactions is similar in the presence of excess acetate,
yielding the cinnamaldehyde and cinnamyl ether product,
respectively, and we can expect the two reactions to have
relatively similar energy profiles under these conditions.
However, in the absence of excess acetate, the product
profiles of the two reactions differ (vide supra). This difference will be one of the main focuses of the current study.

Computational Details
The calculations have been performed using Schr
odingers
software Jaguar, version 6.5 release 106.10 We have used the
(10) Jaguar, version 6.5; Schrodinger, LLC: New York, 2005.

B3LYP hybrid functional11 and the basis set LACVP*.12 The


complexes have first been optimized in the gas phase and
subsequently in solvent (DMF: epsout=38, radprb=2.47982)
modeled by a Poisson-Boltzmann self-consistent reaction field
(PB-SCRF).13 Our previous study showed that the calculation
of solution phase free energy is essential for the correct description of the reaction.8 Since the solution and gas phase geometries
are very similar, we have approximated the free energy in
solution by performing a vibrational analysis in the gas phase,
using the analytic Hessian and adding the thermodynamic
contributions at 363.15K to the solution phase energy. From
earlier experience, we assume that in ideal cases the current
methodology has an accuracy around 1-2 kJ/mol when comparing very similar (e.g., diastereomeric) barriers. In such cases,
known inadequacies of the current DFT implementation like the
self-interaction error and neglect of dispersion are expected to
give similar errors for all compared structures and, thus, to cancel in the final energy difference. Accounting for these known
(11) (a) Lee, C.; Yang, W.; Parr, R. G. Phys. Rev. B 1998, 37, 785789.
(b) Becke, A. D. J. Chem. Phys. 1993, 98, 56485652. (c) Stephens, P. J.; Devlin,
F. J.; Chabalowski, C . F.; Frisch, M. J. J. Phys. Chem. 1994, 98, 1162311627.
(12) Hay, P. J.; Wadt, W. R. J. Chem. Phys. 1985, 82, 299310.
(13) (a) Tannor, D. J.; Marten, B.; Murphy, R.; Friesner, R. A.;
Sitkoff, D.; Nicholls, A.; Ringnalda, M.; Goddard, W. A.III; Honig, B.
J. Am. Chem. Soc. 1994, 116, 1187511882. (b) Marten, B.; Kim, K.; Cortis,
C.; Friesner, R. A.; Murphy, R. B.; Ringnalda, M. N.; Sitkoff, D.; Honig, B.
J. Phys. Chem. 1996, 100, 1177511788.

Article

Organometallics, Vol. 28, No. 21, 2009

6203

Figure 2. Free energy profile of the Heck phenylation of acrolein methyl acetal with chloride as the ligand on palladium.
errors, it is reasonable to assume that errors could increase
by about an order of magnitude when comparing significantly different barriers. An error estimate of ca. 12-20 kJ/mol
(3-5 kcal/mol) is frequently quoted even in cases where small
computational models are employed in lieu of the experimental
systems,14 and we can expect the current treatment of a realistic
experimental system to be more accurate.
As a representative example of the general reaction between
acrolein acetals and aryl halides, we have chosen to examine the
cross-coupling of acrolein dimethyl acetal and phenyl chloride.
We have used formamide to model explicit solvent molecules in
situations where direct coordination makes the implicit solvent
model of DMF inadequate. We focus only on the formation of
(E)-cinnamaldehyde, since formation of the Z isomer has not
been experimentally observed for the Heck process.

Scheme 3. Catalytic Cycle for Heck Arylation of Allyl Ether and


Acrolein Acetal

Results and Discussion


As shown in Scheme 1 the reaction conditions for the
synthesis of cinnamaldehydes (a) or arylpropanoate esters
(b) differ only in base and additives. In our previous study8
we found that the most likely ligand on palladium under the
first-mentioned conditions is acetate coordinated in bidentate mode, while in the latter case it is a chloride; we assume
that the same situation holds for the present system. We have
therefore investigated the reaction pathway for the Heck
coupling under the two different reaction conditions, omitting the oxidative addition of aryl halide to palladium(0)
since the mechanism of this step has been described in detail
elsewhere15 and should have no influence on the regioselectivity. However, we note that under the reaction conditions
(no phosphine, excess of coordinating anions), the Pd(0)
(14) See, for example: Himo, F.; Siegbahn, P. E. M. Chem. Rev. 2003,
103, 24212456.
(15) (a) Ahlquist, M.; Fristrup, P.; Tanner, D.; Norrby, P.-O. Organometallics 2006, 25, 20662073. (b) Ahlquist, M.; Norrby, P.-O. Organometallics 2007, 26, 550553.
(16) For leading references, see: (a) Amatore, C.; Jutand, A. Acc.
Chem. Res. 2000, 33, 314321. (b) de Vries, J. G. Dalton Trans. 2006, 421
429, and references therein.

species reacting in the oxidative addition is most likely


already coordinated to an anion.15,16 Thus, the anion coordinated to Pd(II) in the catalytic cycle (X in Scheme 3)
need not be the leaving group from the aryl halide, but rather
the most prevalent coordinating anion in the reaction mixture, acetate or chloride, respectively. The first step of our
study is the coupling of the alkene and the phenyl moiety to
form intermediate 3 (Figures 1 and 2). During the transfer of
the phenyl group from palladium to the alkene the chloride
switches its coordination site on palladium by donating a
lone pair into the neighboring free orbital on the metal, in
order to avoid the trans influence from the Pd-C -bond.
The bidentate acetate stays in place, but the bond lengths are
modified so that the longer Pd-O bond is always trans to the
atom with the stronger trans influence (-bonded carbon
or hydride). The need for the chloride to switch position
during the phenyl transfer explains why the activation energy

6204

Organometallics, Vol. 28, No. 21, 2009

is 11 kJ/mol higher when chloride is the ligand rather than


acetate, which does a much smaller geometry change. In the
initial carbopalladation product (3) the phenyl group coordinates to palladium, and this interaction has to be
broken before the reaction can proceed. This is done by
rotation around the C-C single bond that originates from
the substrate double bond, through the transition state 4,
forming the agostic intermediate 5. At this point there are
two routes that the reaction can follow: either -elimination
(transition state 6) can occur directly from the agostic
intermediate 5, resulting in the cinnamaldehyde acetal product complex 7, or the other -hydrogen can be eliminated,
forming the ketene acetal complex 11. The latter elimination
requires that the agostic interaction present in intermediate 5
is broken and a new one, involving the relevant -hydrogen
(9), is established. In order to get to the agostic intermediate
9, the reaction must proceed via the intermediate 8. This
intermediate is the key to the regioselectivity of the reaction,
since its energy is highly dependent upon which ligand is
present on palladium. With acetate as the ligand the energy
of the intermediate 8 is higher than the energy of the
-elimination transition states, and thus the low-energy path
will be the direct -elimination (transition state 6a) from the
agostic intermediate 5a. When chloride is the ligand on
palladium, the position trans to the alkyl group is vacant,
and the intermediate 8b is of lower energy than both
-elimination transition states. This leads to a CurtinHammett situation where the agostic intermediates can
interconvert rapidly, and the reaction will go through the
-elimination transition state of lowest energy (10b), resulting in the ketene acetal product complex 11.
The key intermediate 8a, with the acetate carbonyl oxygen
occupying the site trans to the strongly trans-influencing
alkyl moiety, is so unstable that the agostic intermediate 9a
and the -elimination transition state 10a are not even a
minimum and maximum, respectively, on the free energy
surface, but must simply be considered as points along the
C-H bond stretching reaction coordinate for the conversion
of 8a to the product complex 11a. On the potential energy
surface these two particular intermediates correspond to
well-defined stationary points, but when entropy is taken
into account, this is no longer the case. Without the entropy
and solvent contributions, the intermediate 8a is lower in
energy than the -elimination transition states 6a and 10a,
leading to the prediction that the ketene acetal should be the
major product, in clear contrast to the experimental results
for these reaction conditions. This emphasizes the importance of using the free energy, as opposed to merely using
potential energy, when investigating reaction paths, and is in
accordance with our previous findings.8
In the agostic intermediate 5 there are two interactions of
importance for the position of the ligand. The Pd-C bond
exerts a strong trans influence on the ligand, while the agostic
interaction has a minor trans influence. Hence, it is favorable
for the chloride ligand in 5b to be positioned cis to the Pd-C
bond, despite the steric crowding in this position, which
results in a Cl-Pd-C bond angle in 5b of 99, instead of the
ideal angle of 90. In acetate complex 5a, the Pd-O bond
lengths show the main interaction between acetate and
palladium to be through the oxygen in the cis position to
the Pd-C bond (Figure 3). For the agostic intermediate 5 to
proceed through -elimination transition state 6, the ligand
needs to adjust to the change in trans influence, where the
major contribution now comes from the hydride that is being

Henriksen et al.

Figure 3. Bond lengths and angles in the DFT structures of key


intermediates and transition states.

transferred to palladium, while the forming double bond


gives rise only to a smaller repulsion.7 The bidentate acetate
ligand can easily accommodate this change by adjusting
Pd-O bond lengths to have the main coordination taking
place through the oxygen in the trans position to the breaking Pd-C bond. Chloride, on the other hand, has to completely change coordination in order to avoid the increasing
repulsion from the hydride (Figure 3). The energy cost of
doing so makes the reaction less favorable (cf. the carbopalladation step) and accounts for the 9 kJ/mol higher
activation energy between 5b and 6b than between 5a and
6a (Figures 2 and 1, respectively). The other option from
intermediate 5 involves breaking the agostic interaction
leading to intermediate 8, where a solvent molecule is
coordinating to the new free coordination site on palladium.
Computationally, for this intermediate it is necessary to use
an explicit solvent molecule to describe the coordination, due
to the lack of orbital interactions in the implicit electrostatic
solvent model.13 In intermediate 8 only the Pd-C bond
exerts a trans influence on the ligand. This means that the
ligand will prefer to be positioned cis to this bond. The
chloride ligand in 5b is already very close to the optimal
position, and the Cl-Pd-C bond angle changes by only 9
upon going from 5b to 8b (Figure 3). In contrast, the acetate
ligand in 8a has almost completely lost a Pd-O coordination. Since the Pd-O coordination is present in 5a, the
conversion of 5a to 8a is connected with an unfavorable
enthalpic contribution that is not present in the conversion
of 5b to 8b. This explains why the conversion of 5a to 8a is
8 kJ/mol more endergonic than the conversion of 5b to 8b
(Figures 1 and 2). These observations account for the experimentally found selectivities, in that the reaction with acetate
as the base will proceed via the -elimination transition
state 6a, which is of lower energy than the intermediate 8a on
the alternative reaction path, giving the cinnamaldehyde product complex 7a (Figure 1). In contrast, when chloride is the
ligand, the intermediate 8b is so stable that it has no influence
on the reaction outcome, which is instead determined by the
energy difference between the two competing -elimination
transition states. Since transition state 10b is 10 kJ/mol more
favorable than 6b, the reaction will go through 10b, leading to the ketene acetal product complex 11b (Figure 2).
The energy difference of 10 kJ/mol corresponds to a product distribution of 27:1 at 90 C, which explains why the

Article

Organometallics, Vol. 28, No. 21, 2009

6205

compared to the competing paths, neatly explaining the


experimentally observed selectivity difference between the
two closely related systems. Under the alternative reaction
conditions where an acetate base is present, the reaction
outcome is independent of the -elimination transition state
leading to the ketene acetal or enol ether, respectively, and
depends only on the -elimination transition state leading to
cinnamaldehyde/cinnamyl ether and the high-energy intermediate 8a. Since these are fundamentally conserved between the two systems, so is the regioselectivity under these
reaction conditions.
Figure 4. -Hydride elimination transition states.

arylpropanoate ester (C, Scheme 1) is the only isolated


product under these conditions. For the acetate pathway
the energy difference gives only the lower bound to the
selectivity, since we are comparing a transition state (6a)
with an intermediate (8a), but the 3 kJ/mol energy difference
between 6a and 8a indicates that it should be at least 3:1 at
90 C in favor of the cinnamaldehyde product.
Under the conditions where no acetate base is added, there
is a significant selectivity difference between the arylation of
acrolein acetal and allyl ether. To analyze this difference in
detail, we will take a closer look at the -hydride elimination
transition states: 6b and 10b for acrolein acetal, 6c and 10c
for allyl ether (Figure 4). The acetal hydrogen is the most
hydride-like; the overlap between the oxygen lone pairs with
the C-H *-bond provides a strong driving force for the
reaction. Thus, 10b is much earlier than any of the other
transition states. This is most notable in the preferred position of the chloride; in 10b, the position trans to the migrating alkyl group is preferred, whereas in the other three
transition states the hydride has developed enough of a
negative charge so that the chloride prefers the position cis
to the developing hydride ligand (Figure 4). In accordance
with the Bell-Evans-Polanyi relationship,17 this corresponds to a much lower transition state energy for 10b
(17) (a) Bell, R. P.; Proc, R. Soc. London, Ser. A 1936, 154, 414429.
(b) Evans, M. G.; Polanyi, M. J. Chem. Soc., Faraday Trans. 1936, 32,
13331360. (c) For a discussion see: Jensen, F. Introduction to Computational Chemistry; Wiley: Chichester, 1999.

Conclusions
Our study of the Heck reaction between acrolein acetals
and aryl halides verifies that the regioselectivity of the final
elimination step does not necessarily arise from competition
between the two different -hydride elimination transition states, since under some circumstances there is a high
energy barrier for interconversion that negates the CurtinHammett conditions. This means that there will be a predisposition for elimination of the -hydrogen that requires only
a slight reorganization of the carbopalladation product to be
in the correct position as opposed to elimination of the
hydrogen that requires a more comprehensive rearrangement of the complex. Furthermore, our study explains why
the regioselectivity is conserved between the present system
and the related Heck arylation of allyl ethers when an acetate
base is used, but not when chloride is the ligand on palladium. A small structural change has no influence on the highenergy intermediate and the -elimination transition state of
the benzylic hydrogen, which govern the reaction outcome in
the presence of acetate. The change in structure does, however, influence the elimination of the alternative -hydrogen,
since there is a significant difference in hydride donating
ability between the two substrates. Since it is the relative
-elimination transition state energy that controls the reaction outcome when acetate is absent, the regioselectivity is in
this case dependent on the substrates.
Supporting Information Available: This material is available
free of charge via the Internet at http://pubs.acs.org.

You might also like