Direct observation of enantiomer discrimination of epoxides by chiral salen complexes using ENDOR

Article abstract of DOI:10.1021/ja045491s

Electron nuclear double resonance (ENDOR) spectroscopy was used to investigate the weak enantioselective binding between chiral salen complexes [VO(1)] ((R,R)- and (S,S)-vanadyl N,N-bis(3,5-di-tert-butylsalcylidene)-1,2-cyclohexanediamine) and chiral epoxides (e.g., (R)-/(S)-propylene epoxide, 5) in frozen (10 K) solution. Differences in epoxide binding by enatiomers of [VO(1)] was evidenced by changes to the ¹H epoxide derived peaks in the ENDOR spectra, such that (R,R)-[VO(1)] + (R)-5 and (R,R)-[VO(1)] + (S)-5 yield noticeably different spectra. These changes were assigned to the small structural differences between the diastereomeric metal-epoxide adducts. Simulation of the spectra revealed differences in the VO...¹H_epoxide distances for the diastereomeric pairs, which was confirmed by a complementary set of density functional theory (DFT) calculations. While the epoxide molecule is very weakly coordinated, ENDOR measurements of the racemic complex in racemic epoxide nevertheless indicated the preferential coordination of the (R)-5 to (R,R)-[VO(1)] (likewise (S)-(5) to (S,S)-[VO(1)]), which is favored over the binding of (S)-5 epoxide to (R,R)-[VO(1)] (and likewise (R)-5 epoxide to (S,S)-[VO(1)]). This demonstrates the unique power of the ENDOR technique to resolve weak chiral interactions for which EPR spectroscopy alone lacks sufficient resolution. Copyright

Full text of DOI:10.1021/ja045491s

Published on Web 11/12/2004
Direct Observation of Enantiomer Discrimination of Epoxides by Chiral Salen
Complexes Using ENDOR
Ian A. Fallis,* Damien M. Murphy,* David J. Willock, Richard J. Tucker, Robert D. Farley,
Robert Jenkins, and Robert R. Strevens
School of Chemistry, Cardiff UniVersity, P.O. Box 912 Cardiff, CF10 3TB, UK
Received July 27, 2004; E-mail: fallis@cf.ac.uk; murphydm@cf.ac.uk
Enantiomer discrimination often arises from the small differences
in the energies of diastereomeric states resulting from the interac-
tions of chiral species.1 Many analytical methods and spectroscopic
techniques2 can reveal the presence of diastereomeric interactions
but do not readily yield the spatial arrangement of the individual
chiral moieties that govern a particular diastereomeric response.
As part of a study3 to investigate weak interactions of small mole-
cules with metal complexes, we have adopted a combined ENDOR
spectroscopy and computational approach to elucidate the interac-
tions of chiral species in solution. This is made possible by analysis
of the ENDOR spectra,4 which can often yield nuclear coordinates
of the interacting species in frozen solution. Here we describe how
ENDOR is used to detect weak chiral interactions between chiral
paramagnetic salen complexes [VO(1)] and chiral epoxides (5) in
frozen solution. We suggest that [VO(1)] may be a useful model
for the [Co(1)(X)]5 epoxide hydrolytic kinetic resolution (HKR)
catalyst, developed by Jacobsen, in which a second-order rate depen-
dence of the reaction on catalyst concentration implies activation
of not only the nucleophile (water) but also of the chiral epoxide
substrate.
Figure 1. X-band 1H ENDOR spectra (10 K) of (R,R)-[VO(1)] dissolved
in (a) CH2Cl2, (b) DMF, (c) (R)-5, and (d) (S)-5; (e) simulation of epoxide
peaks in spectrum d.
in [VO(2)], which produced a similar Hlig-VO distance of 3.74 Å,
which we assign to all conformations of the 6-tert-butyl groups.
1
The X-band H ENDOR spectra of (R,R)-[VO(1)] in CH2Cl2,
DMF, (R)-5 and (S)-5 are shown in Figure 1a-d, respectively. The
ENDOR spectra of the parent [VO(1)] complex will first be briefly
discussed before the analysis of the peaks corresponding to the
coordinated epoxide moiety. A complete angular selective ENDOR
analysis was carried out, but for clarity only the pure perpendicular
component (field perpendicular to the VdO bond at 3290G in the
EPR spectrum; Supporting Information) will be discussed. As
previously reported for metal-Schiff base complexes,6 the proton
ENDOR spectra are complex at this position; therefore, to aid in
the spectral analysis, additional complexes [VO(2)] and [VO(3)]
were also examined (Supporting Information). This analysis
revealed that the largest 1H hyperfine couplings in [VO(1)] can be
systematically assigned to (i) the cyclohexyl methine (Hlig-VO
distances of 3.05 and 3.46 Å), (ii) the imine (Hlig-VO distance of
4.03 Å), and (iii) the 6-tert-butyl protons, respectively. These
distances, extracted from the simulated ENDOR data using the
point-dipole approximation,5a-c were confirmed using DFT cal-
culations (3.04 and 3.48 Å for the cyclohexyl methine and 4.03 Å
for the imine protons) and are in good agreement with those
expected for Salen-type complexes.6 The DFT calculations were
performed on a simplified chiral complex in which the phenyl
groups in [VO(3)] were replaced by carbon-carbon double bonds.
The extended ligand structure of [VO(1)] was also examined by
DFT using semiempirical methods (PM3).7 From this, the Hlig-
VO distances for the 6-tert-butyl groups of [VO(1)] were found to
be in the range 3.38-4.05 Å due to a ∼2 kJ mol-1 difference in
energy between rotamers. By comparison, the ENDOR spectra of
[VO(1)] revealed an additional set of couplings that were absent
ENDOR spectra of [VO(1)] in noncoordinating and weakly
coordinating solvents (CH2Cl2 and DMF, respectively, Figure 1a,b)
are very similar, indicating that these solvents are not strongly
bound. However, the spectrum recorded for (R,R)-[VO(1)] in (R)-5
(Figure 1c) was noticeably different, with the appearance of new
peaks (labeled * and (), which are not visible in the spectra of the
parent complex and must arise from the coordination of the epoxide
ether group to the metal center. ENDOR measurements using 2H-
labeled epoxide (5) revealed that the * peaks originate from the
epoxide, while the ( peak arises from a subtle conformational
change in the ligand protons of [VO(1)]. These new peaks persist
even when the solution is diluted by a factor of 10 with toluene.
Coordination of the epoxide was also verified by measuring
ENDOR spectra of the “diastereomeric” states, namely, (R,R)-[VO-
(1)] dissolved in (S)-5 (Figure 1d). In this case, the pairwise
combinations of (R,R)-[VO(1)] + (R)-5 and (R,R)-[VO(1)] + (S)-5
yield different spectra, as expected for a pair of diastereomers.
Furthermore, the new peaks in Figure 1c,d possess not only different
intensities but also small changes in coupling constants. This
indicates that we can directly observe different structures for the
[(R,R)-VO(1)-(R)-5] and [(R,R)-VO(1)-(S)-5] diastereomeric com-
plexes in frozen solution.
9
15660
J. AM. CHEM. SOC. 2004, 126, 15660-15661
10.1021/ja045491s CCC: $27.50 © 2004 American Chemical Society

C O M M U N I C A T I O N S
the VO‚‚‚Oepoxide distance must be obtained by an alternative meth-
od. Thus, DFT calculations at the B3LYP/6-31G(d) function7 using
the earlier model yielded epoxide adducts in which the VO‚‚‚Oepoxide
distances were calculated as 2.81 and 2.85 Å for (R,R)-[VO(1)] +
(S)-5 and (R,R)-[VO(1)] + (R)-5, respectively. Although these dis-
tances are long, they can be rationalized by considering the known
pyramidalization at V in [VO(1-3)]-type complexes, which renders
the metal center less accessible to donor groups,4b and the poor
donor ability of epoxides, as evidenced by the paucity of structurally
characterized metal-epoxide complexes. Additionally, the corre-
sponding VO‚‚‚1Hepoxide distances for the pair of epoxide vicinal
protons are in good agreement with those obtained experimentally
via the above ENDOR measurements. This comparison reinforces
the assertion that the changes observed in the ENDOR spectra are
due to the inherent chirality in the complex forcing the chiral
epoxide molecules to adopt different binding conformations.
In the HKR,5 and related [CrCl(1)] catalysts,8 the active species
bear an anionic π-basic donor group trans to the bound epoxide,
similar to the current [VO(1)] model. A recent mechanistic study5b
indicates that the key step in enantiodiscrimination in the HKR of
epoxides by [Co(1)(X)] species (X ) anion) is the interaction of
the activated nucleophile [Co(1)(H2O)(OH)] and the activated
epoxide complex [Co(1)(OH)(epoxide)] and not the enantioselective
binding of epoxide substrates by [Co(1)(X)] species. The current
ENDOR study supports this by showing that chiral Lewis acids
such as [VO(1)] bind the “mismatched” epoxide more strongly. If
this mismatched species were involved in the hydrolysis step, the
opposite enantiomers would be observed as products. The implica-
tion of these results is that although, for example, the (R,R)-[Co-
(1)(OH)] + (R)-5 complex is likely to have a higher formation
constant than (R,R)-[Co(1)(OH)] + (S)-5, it is the more rapid
reaction of the latter complex with [Co(1)(H2O)(OH)] that deter-
mines the stereochemical outcome.
Figure 2. X-band 1H ENDOR spectra (10 K) of the diastereomeric states
formed between enantiomers of [VO(1)] dissolved in (R)- or (S)-5. (a) (R,R)-
[VO(1)] in (S)-5, (b) (S,S)-[VO(1)] in (R)-5, (c) (R,R)-[VO(1)] in (R)-5,
(d) (S,S)-[VO(1)] in (S)-5, and (e) racemic (R,R/S,S)-[VO(1)] in racemic
(R/S)-5.
The diastereomeric nature of these interactions was further
explored by comparing all possible combinations of (R,R)-/(S,S)-
[VO(1)] and (R)-/(S)-5. Expansions of the resulting spectra are
shown in Figure 2a-d. A high degree of correlation between the
enantiomeric (R,R)-[VO(1)] + (S)-5/(S,S)-[VO(1)] + (R)-5 (Figure
2a,b) and (R,R)-[VO(1)] + (R)-5/(S,S)-[VO(1)] + (S)-5 (Figure
2c,d) spectra is obvious, while differences in the diastereomeric
states (R,R)-[VO(1)] + (S)-5/(R,R)-[VO(1)] + (R)-5 (Figure 2a,c)
and (S,S)-[VO(1)] + (R)-5/(S,S)-[VO(1)] + (S)-5 (Figure 2b,d) are
notable. These results are strong evidence for the formation of
diastereomeric complexes between enantiomers of [VO(1)] and
enantiomers of the weakly interacting epoxide (5). Furthermore,
these ENDOR observations appear to be general, yielding similar
enantiomeric discrimination for other aliphatic epoxides also
examined (e.g., 1,2-epoxybutane, epichlorohydrin).
1H ENDOR spectra of the racemic [VO(1)] complex in racemic
epoxide (5) were also recorded (Figure 2e). Significantly, this
spectrum is identical to the spectra of the enantiomeric pairs (R,R)-
[VO(1)] + (R)-5 and (S,S)-[VO(1)] + (S)-5. This result represents
clear, unambiguous proof for the preferential binding of (R)-5 by
(R,R)-[VO(1)] and (S)-5 by (S,S)-[VO(1)]. Therefore, ENDOR
spectroscopy can clearly reveal that in frozen solution, one
diastereomeric complex is strongly preferred.
Acknowledgment. We acknowledge EPSRC funding for the
ENDOR Centre (GR/R17980/01) and postgraduate support for R.R.
Strevens (GR/L80447/01) and R.J. Tucker. We also acknowledge
the EPSRC national mass spectrometry service (Swansea).
Supporting Information Available: Figures 1 and 2 and corre-
sponding EPR/ENDOR spectra (PDF). This material is available free
of charge via the Internet at http://pubs.acs.org.
References
(1) For an excellent historical overview, see: Mason, S. F. Molecular Optical
ActiVity and the Chiral Discriminations; Cambridge University Press:
Cambridge, 1980.
To determine the structures of each diastereomeric complex in
solution, iterative cycles of ENDOR simulation and DFT calcula-
tions were carried out. Previous ENDOR studies of vanadyl
complexes have shown that coordinating solvents predominantly
coordinate trans to the oxo group and in the case of [VO(1)] it is
a reasonable assumption that epoxides coordinate in a similar
manner. Simulation of the peaks labeled * in Figure 1c,d and
analysis of the resulting hyperfine tensors yielded two V‚‚‚1Hepoxide
distances of 3.72 and 3.56 Å for (R,R)-[VO(1)] in (R)-5 and 3.76
and 3.66 Å for (R,R)-[VO(1)] in (S)-5 for the pair of vicinal protons
on the epoxide. Essentially identical data are obtained for a similar
treatment of (S,S)-[VO(1)] + (S)-5 and (S,S)-[VO(1)] + (R)-5,
respectively. It is important to note that the EPR spectra of
diastereomeric combinations of [VO(1)] and 5 are identical,
regardless of the enantiomeric composition of the reaction mixture
(Supporting Information). Thus, we need to use ENDOR’s ability
to resolve the weak electron-nuclear interactions in order to reveal
the subtle structural differences between diastereomeric species.
While ENDOR is useful for determining VO‚‚‚1Hepoxide distances,
(2) For a recent relevant example, see: Bobb, R.; Alhakimi, G.; Studnicki,
L.; Lough, A.; Chin, J. J. J. Am. Chem. Soc. 2002, 124, 4544.
(3) (a) Murphy, D. M.; Fallis, I. A.; Farley, R. D.; Tucker, R. J.; Avery, K.
L.; Willock, D. J. Phys. Chem. Chem. Phys. 2002, 4, 4937. (b) Tucker,
R. J.; Fallis, I. A.; Farley, R. D.; Murphy, D. M.; Willock, D. J. Chem.
Phys. Lett. 2003, 380, 757.
(4) (a) Rirst, G. H.; Hyde, J. S. J. Phys. Chem. 1970, 52, 4633. (b) Hoffman,
B. M.; Martinsen, J.; Venters, R. A. J. Magn. Reson. 1984, 59, 110. (c)
Hurst, G. C.; Henderson, T. A.; Kreilick, R. W. J. Am. Chem. Soc. 1985,
107, 7294. (d) Hoffman, B. M. Proc. Natl. Acad. Sci. U.S.A. 2003, 100,
3575. (e) Smoukov, S. K.; Kopp, D. A.; Valentine, A. M.; Davydov, R.;
Lippard, S. J.; Hoffman, B. M. J. Am. Chem Soc. 2002, 124, 2657. (f)
Bolm, C.; Martin, M.; Gescheidt. G.; Palivan, C.; Neshchadin, D.;
Bertagnolli, H.; Feth, M.; Schweiger, A.; Mitrikas, G.; Harmer, J. J. Am.
Chem. Soc. 2003, 125, 6222. (g) Schweiger A.; Jeschke, G. Principles of
Pulse Electron Paramagnetic Resonance; Oxford University Press: New
York, 2001.
(5) (a) Tokunaga, M.; Larrow J. F.; Kakiuchi, F.; Jacobsen, E. N. Science
1997, 277, 936. (b) Nielsen, L. P. C.; Stevenson, C. P.; Blackmond, D.
G.; Jacobsen, E. N. J. Am. Chem. Soc. 2004, 126, 1360.
(6) (a) Attanasio, D. J. Phys. Chem. 1986, 90, 4952. (b) Schweiger, A.;
Structure and Bonding; Springer: Berlin, 1982; Vol. 51, p 1.
(7) See Supporting Information for details of the computational work.
(8) Hansen, K. B.; Leighton, J. L.; Jacobsen, E. N. J. Am. Chem. Soc. 1996,
118, 10924.
JA045491S
9
J. AM. CHEM. SOC. VOL. 126, NO. 48, 2004 15661