Hammett studies of enantiocontrol by PHOX ligands in Pd-catalyzed allylic substitution reactions

Article abstract of DOI:10.1021/ol034610q

(Matrix presented) Electronically modified PHOX ligands 3a-e were synthesized to probe the mechanism of the enantioselective palladium-catalyzed allylic alkylation and amination reactions. Alkylation with dimethyl sodiomalonate produced only a small variation in the ee (89.3% to 93.4%), but amination with benzylamine gave a much wider variation in the ee (16.4% to 66.6%). Hammett analysis suggests that the substituents interact more significantly with phosphorus and supports a combined electronic and steric basis for enantioselection.

Full text of DOI:10.1021/ol034610q

ORGANIC
LETTERS
2003
Vol. 5, No. 13
2279-2282
Hammett Studies of Enantiocontrol by
PHOX Ligands in Pd-Catalyzed Allylic
Substitution Reactions
Ryan N. Constantine, Naomi Kim, and Richard C. Bunt*
Department of Chemistry and Biochemistry, Middlebury College,
Middlebury, Vermont 05753
rbunt@middlebury.edu
Received April 9, 2003
ABSTRACT
Electronically modified PHOX ligands 3a−e were synthesized to probe the mechanism of the enantioselective palladium-catalyzed allylic alkylation
and amination reactions. Alkylation with dimethyl sodiomalonate produced only a small variation in the ee (89.3% to 93.4%), but amination
with benzylamine gave a much wider variation in the ee (16.4% to 66.6%). Hammett analysis suggests that the substituents interact more
significantly with phosphorus and supports a combined electronic and steric basis for enantioselection.
The enantioselective palladium-catalyzed allylic substitution
reaction (eq 1) is a powerful method for the construction of
carbon-carbon and carbon-heteroatom bonds.1 Among the
myriad types and classes of chiral ligands designed for this
and other enantioselective reactions, the phosphinooxazoline
(PHOX) ligands (3) have attracted considerable synthetic and
mechanistic interest.2 The observed stereochemistry of the
products (2) is consistent with either nucleophilic attack trans
to phosphorus in the exo intermediate or trans to nitrogen in
the endo intermediate (Figure 1). The former pathway is
generally favored for the following reasons. Both solution
Figure 1. The two modes of nucleophilic attack on the η3-
1H NMR and X-ray crystallographic studies show that the
allylpalladium intermediates consistent with the observed product
exo diastereomer predominates.3 The π-acceptor nature of
the phosphorus ligand should make the trans allylic carbon
more electron-deficient and therefore more susceptible to
stereochemistry.
nucleophilic addition.4 Most definitively, low-temperature 1H
NMR studies of the initially formed product, the alkene-
palladium complex, show the nucleophile oriented away from
phosphorus, suggesting it arose from a least-motion rotation
following nucleophilic addition trans to phosphorus in the
exo η3-allylpalladium intermediate.5
(1) Trost, B. M.; van Vranken, D. L. Chem. ReV. 1996, 96, 395.
(2) (a) Helmchen, G.; Pfaltz, A. Acc. Chem. Res. 2000, 33, 336. (b)
Helmchen, G. J. Organmet. Chem. 1999, 576, 203. (c) Williams, J. M. J.
Synlett 1996, 705.
(3) (a) Kollmar, M.; Goldfuss, B.; Reggelin, M.; Rominger, F.; Helmchen,
G. Chem. Eur. J. 2001, 7, 4913. (b) Baltzer, N.; Macko, L.; Schaffner, S.;
Zehnder, M. HelV. Chim. Acta 1996, 79, 803. (c) Sprinz, J.; Kiefer, M.;
Helmchen, G.; Reggelin, M.; Huttner, G.; Walter, O.; Zsolnai, L. Tetra-
hedron Lett. 1994, 35, 1523.
(4) Szabo, K. J. Organometallics 1996, 15, 1128.
10.1021/ol034610q CCC: $25.00 © 2003 American Chemical Society
Published on Web 05/29/2003

Although the PHOX ligands seem like a clear case of
electronic control of enantioselectivity (i.e., nucleophilic
attack trans to the better π-acceptor ligand phosphorus rather
than the donor ligand nitrogen), steric interactions between
the chiral ligand and product alkene complex can also explain
the enantioselectivity (Figure 2).6 With a series of N,S-chiral
no variation in ee with changing substituents,7 the analogous
phosphine ligand system (6) showed a dramatic change in
the ee as the substituents were varied.12 These results have
been used to argue for either steric or electronic origins for
enantiocontrol, respectively.
The objective of this study is to investigate whether the
trans to phosphorus transition state for the PHOX ligands is
primarily electronic or steric in origin. Toward that goal, we
undertook the synthesis and Hammett study of ligands
3a,b,d,e, which are electronically different but sterically
identical (around the sites of ligation) to the parent PHOX
ligand (3c). The modified ligands were prepared according
to literature protocols starting from the corresponding 4-sub-
stituted acids or acid chlorides and (S)-valinol.13 Directed
lithiation of the 4′-substituted 2-aryloxazolines and coupling
with chlorodiphenylphosphine provided the desired chiral
ligands (eq 2).2a,14
Figure 2. “Sector model”6a for the four possible alkene-palladium
complexes arising from the nucleophilic addition pathway indicated.
Hydrogen is the small group (S), the i-Pr and axial-like Ph are
considered medium groups (M) and the equatorial-like Ph is
considered large (L). Pathways a and d give the major enantiomer
and pathways b and c give the minor enantiomer.
ligands nucleophilic attack trans to the arguably harder ni-
trogen was observed and the chiral recognition was specif-
ically attributed to steric interactions rather than electronic
differences in the ligand.7 Additionally, theoretical calcula-
tions of several types of P,N-chiral ligands attributed the
enantioselectivity to steric interactions with nucleophilic
attack occurring trans to nitrogen in some cases.8
Previous studies of electronic differences between other
types of chiral ligands have produced disparate results in
the alkylation of 1a.9,10 A large difference in reactivity was
observed for different substituents (X) with 4′-substituted
pyridinyloxazolines (4) but the enantioselectivity was “scarcely
affected”.11 Whereas imine-sulfide ligands 5 showed almost
The electronic differences between the ligands became
apparent during the synthesis. Both electron-donating (a,b)
and electron-withdrawing (d,e) groups gave higher chemical
yields in the directed-lithiation/coupling step than 3c. The
ligands with electron-withdrawing groups (3d,e) also proved
more sensitive to decomposition and were repurified by flash
chromatography directly before use in the palladium-
catalyzed reactions. Attempts to study the X ) CF3 PHOX
ligand have thus far been unsuccessful for this reason.
The electronically modified PHOX ligands were first tested
in what has become the standard test reaction: the alkylation
of 1a with dimethyl sodiomalonate to give 2a (eq 1a).15 In
contrast to the dramatic results observed with Morimoto’s
phosphine-imine ligand 6,12 much less variation in the ee of
2a was observed over a similar range of substituents (Table
1). The trend toward higher ee’s with more electron-
withdrawing substituents is likely real (i.e., ee differences
are greater than error range for 3a to 3e) but not definitive.
(5) (a) Junker, J.; Reif, B.; Junker, B.; Felli, I. C.; Reggelin, M.;
Griesinger, C. Chem. Eur. J. 2000, 6, 3281. (b) Steinhargen, H.; Reggelin,
M.; Helmchen, G. Angew. Chem., Int. Ed. Engl. 1997, 36, 2108.
(6) (a) Kollmar, M.; Steinhagen, H.; Janssen, J. P.; Goldfuss, B.;
Malinovskaya, S. A.; Vazquez, J.; Rominger, F.; Helmchen, G. Chem. Eur.
J. 2002, 8, 3103. (b) von Matt, P.; Lloyd-Jones, G. C.; Minidis, A. B. E.;
Pfaltz, A.; Macko, L.; Neuburger, M.; Zehnder, M.; Ruegger, H.; Pregosin,
P. S. HelV. Chim. Acta 1995, 78, 265. (c) Brown, J. M.; Hulmes, D. I.;
Guiry, P. J. Tetrahedron 1994, 50, 4493.
(7) Adams, H.; Anderson, J. C.; Cubbon, R.; James, D. S.; Mathias, J.
P. J. Org. Chem. 1999, 64, 8256.
(8) (a) Windhalm, M.; Nettekoven, U.; Kalchhauser, H.; Mereiter, K.;
Calhorda, M. J.; Felix, V. Organometallics 2002, 21, 315. (b) Blochl, P.
E.; Tongi, A. Organometallics 1996, 15, 4125.
(9) For an example where the major enantiomer reverses with changing
electronics, see: Clyne, D. S.; Mermet-Bouvier, Y. C.; Nomura, N.;
RajanBabu, T. V. J. Org. Chem. 1999, 64, 7601.
(12) Saitoh, A.; Achiwa, K.; Tanaka, K.; Morimoto, T. J. Org. Chem.
2000, 65, 4227.
(13) (a) Gant, T. G.; Meyers, A. I. Tetrahedron 1994, 50, 2297. (b) Peer,
M.; de Jong, J. C.; Kiefer, M.; Langer, T.; Rieck, H.; Schell, H.; Sennhenn,
P.; Sprintz, J.; Steinhagen, H.; Wiese, B.; Helmchen, G. Tetrahedron 1996,
52, 7547.
(14) Koch, G.; Lloyd-Jones, G. C.; Loiseleur, O.; Pfaltz, A.; Pretot, R.;
Schaffner, S.; Schnider, P.; von Matt, P. Recl. TraV. Chim. Pays-Bas 1995,
114, 206.
(15) See Supporting Information for full experimental details of pal-
ladium-catalyzed allylic-alkylations and aminations of 1a,b.
(10) For examples of electronic ligand tuning in other asymmetric
reactions, see: (a) RajanBabu, T. V.; Casalnuovo, A. L.; Ayers, T. A.;
Nomura, N.; Jin, J.; Park, H.; Nandi, M. Curr. Org. Chem. 2003, 7, 301.
(b) Casey, M.; Smyth, M. P. Synlett 2003, 102. (c) Morimoto, T.; Nakajima,
N.; Achiwa, K.; Tetrahedron: Asymmetry 1995, 6, 23. (d) Jacobsen, E.
N.; Zhang, W.; Guler, M. L. J. Am. Chem. Soc. 1991, 113, 6703.
(11) Chelucci, G.; Deriu, S. P.; Saba, A.; Valenti, R. Tetrahedron:
Asymmetry 1999, 10, 1457.
2280
Org. Lett., Vol. 5, No. 13, 2003

Table 1. Enantioselective Alkylation/Amination Results
2a
2b
ligand
X
% eea
errorb
% eec
errorb
3a
3b
3c
3d
3e
NMe2
OMe
H
F
Cl
89.3 (S)d
89.7 (S)
89.9 (S)
93.0 (S)
93.4 (S)
(2.21
(1.11
(1.86
(1.15
(0.55
16.4 (R)e,f
22.7 (R)
28.4 (R)
44.9 (R)
66.6 (R)
(4.14
(2.81
(2.61
(3.96
(4.03
a HPLC Chiralpak AD 90:10 hexane/2-propanol, 1.0 mL/min. b One
standard deviation based on 5-7 trials. c HPLC Chiralcel OJ 85:15 hexane/
2-propanol, 0.5 mL/min. d Determined from (-) sign of optical rotation,
ref 16a. e Determined from (-) sign of optical rotation, ref 16b. f (R)-2b
has the same “sense” of chirality as (S)-2a, but the priority rules reverse.
Figure 4. Hammett plot of log(er) data vs σM for ligands 3a-e
giving 2a (() and 2b (2) and ligand 6 (O) (also giving 2a).
the free energy difference between the diastereomeric transi-
tion states.20
Using benzylamine as the nucleophile (eq 1b), however,
produced a much larger variation in the ee of 2b with the
same trend as the malonate alkylations (both (S)-2a and
(R)-2b have the same “sense” of chirality16). In neither
reaction was a dramatic effect on reactivity observed with
varying substituents, although the amination reactions were
slower than the malonate alkylations on the whole and
required heating to 40 °C.
The substituent (X) in the modified PHOX ligands (3) is
both meta to phosphorus and para to the nitrogen-containing
oxazoline ring. To ascertain which interaction (with P or N),
if either, is more significant, Hammett analysis of the
enantioselectivity with respect to both σP and σM was carried
out (Figures 3 and 4).17 For comparative and interpretive
purposes the literature data for ligand 6 was also included
in this analysis.12,18 The log of the enantiomeric ratio (er)19
was used for these plots because log(er) relates directly to
The Hammett plots for alkylations with ligand 6 are the
most straightforward to interpret. Structurally, all substituents
in ligand 6 are para (i.e., imines of 4-substituted benzalde-
hydes) and only conjugated with nitrogen. So it is not
surprising that the log(er) data fit σP (R2 ) 0.910) much better
than σM (R2 ) 0.604). The negative slope of the plot suggests
that electron-donating groups increase the er by making
nitrogen a stronger donor ligand (accentuating the difference
between nucleophilic attack trans to P or N), whereas
electron-withdrawing groups lower the er by making nitrogen
a weaker donor ligand (minimizing the electronic difference
between nucleophilic attack trans to P or N).21
For the data with ligands 3a-e giving 2a and 2b, the log-
(er) data fit σM (Figure 4) better than σP (Figure 3) with a
positive slope in both cases. The overall quality of the fits
is lower than with 6.22 The substituents in 3 likely affect
both phosphorus (meta) and nitrogen (para) to some extent.
Nonetheless, it appears that the substituents in ligand 3 affect
(16) For absolute configuration determinations and optical rotation
correlations, see: (a) Hayashi, T.; Yamamoto, A.; Hagihara, T.; Ito, Y.
Tetrahedron Lett. 1986, 27, 191. (b) Hayashi, T.; Yamamoto, A.; Ito, Y.;
Nishioka, E.; Miura, H.; Yanagi, K. J. Am. Chem. Soc. 1989, 111, 6301.
(17) All substituent constants (σP and σM) taken from: Hansch, C.; Leo,
A.; Taft, R. W. Chem. ReV. 1991, 91, 165.
(18) The raw ee data for the dimethyl malonate alkylation of 1a to give
2a was taken from ref 12. Those authors did not undertake the formal
Hammett analysis shown here, nor are these interpretations and conclusions
about functioning of ligand 6 necessarily supported by those authors.
(19) Since 2a and 2b have opposite (R/S) configurations, the enantiomeric
ratio (er) is defined here as the relative amount of the major enantiomer
divided by the relative amount of the minor enantiomer (e.g., 80% ee gives
an er of 9.0).
(20) The er reflects the net ratio of rates of formation of the (R) and (S)
products through the four pathways shown in Figure 2.
(21) For an example of a similar electronic trend with several P,N-ligands
see: Porte, M. A.; Reibenspies, J.; Burgess, K. J. Am. Chem. Soc. 1998,
120, 9180.
(22) Regression analysis with other types of substituent constants (e.g.,
σ+, σo, etc.) or dual-parameter fits (e.g., F/R, σI/σR, etc.) did not provide
Figure 3. Hammett plot of log(er) data vs σP for ligands 3a-e
giving 2a (() and 2b (2) and ligand 6 (O) (also giving 2a).
better data fits. Considering the “simple” system (6) only gave an R2
)
0.91 it is not clear that better fits should be expected.
Org. Lett., Vol. 5, No. 13, 2003
2281

phosphorus more than nitrogen. This agrees with the predic-
tion that electron-withdrawing groups increase the er by
making phosphorus a better acceptor ligand (accentuating
the electronic difference between nucleophilic attack trans
to P or N) and electron-donating groups decrease the er by
making phosphorus a weaker acceptor ligand (minimizing
the difference between nucleophilic attack trans to P or N).
That is, the electronic effects of ligand 3 are opposite in
nature to those of ligand 6.
The changes in enantioselectivity could also be explained
by a change in the exo to endo ratio of intermediates resulting
in an increased (or decreased) rate of formation of the minor
enantiomer via nucleophilic addition trans to phosphorus in
the endo intermediate (Figure 2, path b).23 However, a
change in the exo to endo ratio based on electronics seems
unlikely as this ratio is reported to be steric in origin and
varies with the size and structure of the allyl moiety and the
size of the groups on phosphorus.2,24 Thus, it seems reason-
able to conclude that the ligand substituents (X) do affect
the ratio of nucleophilic attack trans to phosphorus or
nitrogen. This effect would apply to both the exo and endo
complexes, but because the major enantiomer arises from
the more abundant exo complex,5 changing the reactivity for
attack trans to phosphorus or nitrogen would be amplified
by the exo to endo ratio and thus alter the net enantiose-
lectivity.
tronically sensitive amination reaction (eq 1b) seems more
influenced by the electronic structure (i.e., phosphorus vs
nitrogen) of the η3-allylpalladium intermediate (Figure 1) and
less influenced by sterics. Similarly, the differing positions
of the transition states may influence the transmission of
electronic effects. A later transition state is expected for the
less reactive amine nucleophile. Although a late transition
state has generally been used to argue for the increased
importance of steric interactions in the product alkene-
palladium complex,6c the greater degree of bond formation
may accentuate the electronic differences between the allylic
termini as well.
In conclusion, these Hammett studies of modified ligands
3a-e support a mechanism whereby the PHOX ligands direct
nucleophilic attack trans to phosphorus in the exo intermedi-
ate for electronic reasons but suggest that steric interactions
may also play a role in determining the enantioselectivity
that can vary with the nature of the nucleophile.
Acknowledgment. This research was supported by an
award from Research Corporation and acknowledgment is
also made to the donors of the Petroleum Research Fund,
administered by the American Chemical Society, for support
of this research.
Supporting Information Available: Experimental condi-
tions for 2a,b, experimental procedures and full characteriza-
tion data for 3a,b,d,e and 7a,b,d,e, and tables of data used
for Hammett analyses. This material is available free of
charge via the Internet at http://pubs.acs.org.
Finally, the differing susceptibilities of the alkylation and
amination reactions to electronic influences on the enantio-
selectivity (i.e., ee data or slopes of the Hammett plots)
suggest some differences in the origin of the enantioselec-
tivities.25 These reactions differ in both the size and charge
of the nucleophile. The less electronically sensitive nature
of the alkylation reaction (eq 1a) suggests that it is more
influenced by sterics either in the transition state due to the
bulkier malonate anion nucleophile or in the product alkene-
palladium complex (Figure 2). Conversely, the more elec-
OL034610Q
(23) This explanation could avoid any involvement of nucleophilic
addition trans to nitrogen.
(24) Kudis, S.; Helmchen, G. Angew. Chem., Int. Ed. 1998, 37, 3047.
(25) The difference between 89.3% ee and 93.4% ee corresponds to a
∆G of 0.30 kcal/mol, while the difference between 16.4% ee and 66.6% ee
corresponds to a ∆G of 0.75 kcal/mol.
2282
Org. Lett., Vol. 5, No. 13, 2003