Rationally designed improvement of the bis(phospholano)ethane ligand for asymmetric hydrogenation leads to a reappraisal of the factors governing the enantioselectivity of Duphos catalysts

Article abstract of DOI:10.1039/b002994g

Enhancement of enantioselectivity in hydrogenations catalysed by δ vs. λ, rhodium chelate complexes of trans-1,2-bis(phospholano)cyclopentanes cannot be rationalised using the current quadrant model for Duphos ligands and therefore a new consistent model

Full text of DOI:10.1039/b002994g

Rationally designed improvement of the bis(phospholano)ethane ligand for
asymmetric hydrogenation leads to a reappraisal of the factors governing the
enantioselectivity of Duphos catalysts
Elena Fernandez,a Amy Gillon,b Katie Heslop,b Emily Horwood,b David J. Hyett,b A. Guy Orpen*b and
Paul G. Pringle*b
a Universitat Rovira i Virgili, Departament de Química Física Inorgànica, Plaça Imperial Tàrraco, 1, Tarragona,
Spain
b School of Chemistry, University of Bristol, Cantocks Close, Bristol, UK. E-mail: paul.pringle@bristol.ac.uk
Received (in Cambridge, UK) 13th April 2000, Accepted 18th July 2000
Enhancement of enantioselectivity in hydrogenations cata-
lysed by d vs. l rhodium chelate complexes of trans-
1,2-bis(phospholano)cyclopentanes cannot be rationalised
using the current quadrant model for Duphos ligands and
(3)
the l conformer of 1 would give the higher enantioselectivity
(see below)3 and we reasoned that this hypothesis could be
tested by comparing the optical yields obtained with 2 and 3, the
two diastereoisomers of trans-1,2-bis(R,R-phospholano)cyclo-
pentane since 2 would give exclusively a l-conformer chelate
complex and 3 would give a d-conformer.
therefore a new consistent model is suggested.
Diphosphines containing the 2,5-dialkylphospholane moiety
(e.g. R,R-Duphos, R,R-bpe) are currently the most efficient
The new ligands 2 and 3 were prepared from resolved trans-
1,2-diphosphinocyclopentane8 and the 1,4-diol cyclic sulfate1
as shown in eqn. (4) for 2. The rhodium(
I
) chelate complexes l-
(4)
ancillary ligands for the asymmetric hydrogenation of many
alkenes;1–4 several of these catalytic processes are of commer-
cial interest.4 The enantioselectivity is a sensitive function of
the ligand backbone.1–7 For example, for the hydrogenations
shown in eqns. (1) and (2), the enantioselectivity increases with
4 and d-5 have been synthesised and fully characterised but we
have been unable to obtain crystals suitable for X-ray
crystallography. However the crystal structures of the diiodo-
platinum(II) complexes of 2 and 3, l-6 and d-7 have been
determined (Figs. 1 and 2).† These confirm the assignment of
the chelate conformations. The bond lengths and angles around
platinum in l-6 and d-7 are not substantially different (lengths
differ by 0.01–0.02 Å and angles by 1–2°).11 The principal
effect of the change in backbone stereochemistry seems to be in
the orientation of the phospholane rings. Changing the MP2C2
chelate conformation from l to d leads to a rotation about the
M–P bond of ca. 30°, as measured by I–Pt–P–C torsions (see
Figs. 1 and 2). As a consequence, the methyl groups are closer
to the other ligands in the metal coordination plane (here iodine)
in l-6 (I…CH3 3.70 Å) than in d-7 (I…CH3 4.11, 4.20 Å).
Conversely the axial hydrogen atoms adjacent to the phospho-
rus are closer to the iodo ligands in d-7 (I…H 3.07, 3.00 Å) than
(1)
(2)
increasing rigidity of the ligand backbone. Here, we show an
example of how the enantioselectivity for the bis(phospholane)-
ethane complex 1 can be improved by rational design of the
ligand backbone; our results challenge the currently accepted
explanations for the selectivity of Duphos ligands.
Fig. 1 Molecular structure and numbering scheme for l-6. All but
phospholane tertiary hydrogen atoms have been omitted for clarity.
Important molecular dimensions: bond lengths (Å) Pt(1)–P(1) 2.237(2),
Pt(1)–I(1) 2.6453(8); bond angle (°) P(1)–Pt(1)–P(1A) 88.25(10); torsion
angles (°) I(1)–Pt(1)–P(1)–C(2) 46.5(2), I(1)–Pt(1)–P(1)–C(5) 269.2(2).
The chelate ring of the R,R-bpe complex 1 is flexible and an
interconverting mixture of diastereomeric l and d chelate
conformers would be anticipated [eqn. (3)]. It was predicted that
DOI: 10.1039/b002994g
Chem. Commun., 2000, 1663–1664
This journal is © The Royal Society of Chemistry 2000
1663

fore must be re-face bound). Indeed they were able to confirm
this assignment of adduct stereochemistry by NMR experiments
and by analogy with iridium chemistry. By implication, for R,R-
phospholanes, as here, the minor adduct is si-face bound (and
the R-configuration product results from its Halpern-type
hydrogenation). This directly contradicts the standard quadrant
model prediction noted above.
In l-6 the phospholane methyl groups are closer to the metal
coordination plane than in d-7 and therefore would be expected
to present more steric interaction at the sites where the alkene
substrates would bind. However in d-7 (whose analogue d-5 is
the more stereoselective catalyst) the axial hydrogen atoms are
closer to the substrate binding site. This leads us to suggest an
alternative stereochemical model for these phospholane li-
gands: it is the axial hydrogens rather than the equatorial
methyls that offer the critical, diastereofacially-discriminating
steric interactions with a prochiral substrate. The corollary of
this suggestion is that the diagonal of blocked quadrants [Fig.
3(b)] is orthogonal to that currently accepted. The new quadrant
diagram predicts (in accord with ref. 10) that the less stable
(minor) diastereoisomer is formed when the substrate is si-face
bound, Fig. 3c. A Halpern-type hydrogenation mechanism
would then lead to the observed R-configuration of the product,
again in accord with the results reported in ref. 10.
Fig. 2 Molecular structure and numbering scheme for d-7. All but
phospholane tertiary hydrogen atoms have been omitted for clarity.
Important molecular dimensions include: bond lengths (Å) Pt(1)–P(1)
2.248(2), Pt(1)–P(1A) 2.258(2), Pt(1)–I(1) 2.6533(9), Pt(1)–I(1A)
2.6690(8); bond angle (°) P(1)–Pt(1)–P(1A) 86.44(9); torsion angles (°)
I(1)–Pt(1)–P(1)–C(2) 78.4(2), I(1)–Pt(1)–P(1)–C(5) 238.3(2), I(1A)–
Pt(1)–P(1A)–C(2A) 70.6(2), I(1A)–Pt(1)–P(1A)–C(5A) 247.5(2).
in l-6 (I…H 3.37 Å). The PC4 rings in R-phospholanes have d-
conformations and so in d-7 the conformations are ddd for the
PC4, PtP2C2 and PC4 rings, respectively, while in l-6 these rings
show dld conformations.
Hydrogenation of very bulky enamides (e.g. H2CNCButN-
HAc) catalysed by R,R-Duphos complexes leads to a product of
S-configuration. While this inversion of stereochemistry poses
problems for any quadrant model, we note that it is possible that
extreme bulk of the a-substituent may render the concentration
of the minor adduct diastereoisomer effectively zero, thereby
leading to no product from that pathway regardless of the rate of
hydrogenation of the major (only) species.
Note added in proof: The quadrant models discussed in this
paper implicitly assume that crowding in the plane of the metal–
diphosphine moiety is important in enantioselection. A recent
report (I. D. Gridnev, N. Higashi, K. Asakura and T. Imamoto,
J. Am. Chem. Soc., 2000, 122, 7183) suggests that octahedral
cis-dihydride species are critical.
The results of the hydrogenations shown in eqns. (1) and (2)
catalysed by the rhodium catalysts l-4, d-5 and 1 are shown in
Table 1. It is clear that the optical yields obtained with l-4 are
inferior to 1 and those for d-5 are superior to 1. The unequivocal
conclusion is that the d-chelate gives higher enantioselectivities
than the l-chelate, i.e. d-5 is the matched diastereomeric
catalyst.6 Since this is the opposite of what was predicted,3 we
decided to re-examine the basis of the current heuristic model
for Duphos catalysts.12
Table 1 Optical yields for the hydrogenations of methyl-(Z)-2-acet-
amidocinnamate [eqn. (1)] and methyl-2-acetamidoacrylate [eqn. (2)]a
Methyl-(Z)-2-
Methyl-2-
Catalyst
acetamidocinnamate
acetamidoacrylate
We would like to thank Dr Guy Lloyd-Jones for useful
discussions, EPSRC and Albright and Wilson for a CASE
studentship (to E. L. H.), Acciones Integradas for a travel grant
and Johnson-Matthey for a loan of precious metals.
l-4
1b
d-5
77 (R)
85 (R)
98 (R)
73 (R)
91 (R)
95 (R)
a Experimental conditions: solvent MeOH, 2 atm H2, 20–25 °C, 0.05–0.1%
Rh catalyst, reaction time, 1–16 h. Conversions and enantiomeric excesses
were determined by GC using a Hewlett-Packard 5800 A with a L-Chirasil-
Val column. b Results from ref. 3.
Notes and references
† Crystal structure analyses: l-6: C17H32P2I2Pt, M = 747.26, trigonal,
space group P3221 (no. 154), a = 12.699(3), c = 11.995(2) Å, U =
1675.1(5) Å3, Z = 3, m = 9.186 mm21, T = 173 K, 2589 unique data, R1
In the stereochemical model of refs. 3, 6 and 12, it is assumed
that the alkyl substituents of the phospholanes block the
diagonal of quadrants shown in Fig. 3(a) for bis(R-phospholane)
chelates. In turn this implies that the favoured, major, adduct
diastereoisomer is that in which the enamide substrate is bound
through its si face (see ref. 12 for explicit confirmation of this
view).
=
0.0343. Molecules of l-6 lie at sites of exact crystallographic C2
symmetry and show signs of some disorder in the cyclopentane ring
[leading to artificial flattening of the ring at C(9)]. d-7: C17H32P2I2Pt, M =
747.26, monoclinic, space group P21 (no. 4), a
= 8.3470(19), b =
13.844(4), c = 10.3968(19) Å, b = 111.574(11)°, U = 1117.2(4) Å3, Z =
2, m = 9.183 mm21, T = 173K, 5096 unique data, R1 = 0.0361.
CCDC 182/1724. See http://www.rsc.org/suppdata/cc/b0/b002994g/ for
crystallographic files in .cif format.
Seminal mechanistic work9 revealed that the major enantio-
mer formed in asymmetric hydrogenation by rhodium com-
plexes of ‘traditional’ chiral diphosphines such as chiraphos
was derived from the minor diastereoisomer of the prochiral
alkene complex. Burk and coworkers showed10 that, when the
hydrogenation shown in eqn. (1) is catalysed by [Rh(S,S-
Duphos)(cod)]+, the major product enantiomer (having S-
configuration) is derived, in Halpern-like manner, from the
minor diastereoisomer of the substrate complex (which there-
1 M. J. Burk, J. Am. Chem. Soc., 1991, 113, 8518.
2 M. J. Burk, J. E. Feaster and R. L. Harlow, Organometallics, 1990, 9,
2653.
3 M. J. Burk, J. E. Feaster, W. A. Nugent and R. L. Harlow, J. Am. Chem
Soc., 1993, 115, 125.
4 M. J. Burk, M. F. Gross, T. G. P. Harper, C. S. Kalberg, J. R. Lee and
J. P. Martinez, Pure Appl. Chem., 1996, 68, 37.
5 M. J. Burk and M. F. Gross, Tetrahedron Lett., 1994, 35, 9363.
6 M. J. Burk, A. Pizzano, J. A. Martín, L. M. Liable-Sands and A. L.
Rheingold, Organometallics, 2000, 19, 250.
7 M. J. Burk, J. E. Feaster and R. L. Harlow, Tetrahedron: Asymmetry,
1991, 2, 569.
8 C. Eckert, L. Dahlenburg and A. Wolski, Z. Naturforsch., Teil B., 1995,
50, 1004.
9 W. S. Knowles, Acc. Chem. Res., 1983, 16, 106; C. Landis and J.
Halpern, J. Am. Chem. Soc., 1987, 109, 1746 and references therein.
10 S. K. Armstrong, J. M. Brown and M. J. Burk, Tetrahedron Lett., 1993,
34, 879.
11 A. Martín and A. G. Orpen, J. Am. Chem. Soc., 1996, 118, 1464.
12 M. J. Burk, Acc. Chem. Res., 2000, 33, 363.
Fig. 3 (a) Quadrants that are reportedly3,6,12 blocked by the methyl
substituents in Rh(R,R-bpe) chelate; (b) quadrants that are proposed to be
blocked by the axial hydrogens in a Rh(R,R-bpe) chelate; (c) si-face binding
of substrate alkenes.
1664
Chem. Commun., 2000, 1663–1664