A novel approach for investigating enantioselectivity in asymmetric hydrogenation

Article abstract of DOI:10.1016/j.tetasy.2004.05.028

A new and convenient method for investigating the relationship between the position of the enantiodetermining sites in chiral phosphine ligands and enantioselectivity in asymmetric hydrogenation is proposed. We have also synthesized four new diphosphine l

Full text of DOI:10.1016/j.tetasy.2004.05.028

Tetrahedron:
Asymmetry
Tetrahedron: Asymmetry 15 (2004) 2169–2172
A novel approach for investigating enantioselectivity in
asymmetric hydrogenation
Hideo Shimizu, Takero Ishizaki, Takahiro Fujiwara and Takao Saito*
Takasago International Corporation, Central Research Laboratory, Nishi-Yawata, Hiratsuka, Kanagawa 254-0073, Japan
Received 6 May 2004; accepted 17 May 2004
Abstract—A new and convenient method for investigating the relationship between the position of the enantiodetermining sites in
chiral phosphine ligands and enantioselectivity in asymmetric hydrogenation is proposed. We have also synthesized four new
diphosphine ligands, some of which give high enantioselectivity in the asymmetric hydrogenation of an enamide.
Ó 2004Elsevier Ltd. All rights reserved.
Asymmetric hydrogenation is one of the most powerful
tools in organic synthesis.1 As enantioselectivities are
highly affected by chiral ligands in catalysis, various
ligands have been fine-tuned for ultimate selectivity.
Some of these perform with sufficient enantioselectivity
for use in applications to industrial needs.2 Enantio-
selectivity is the result of various factors associated with
a ligand, including steric effects, rigidity of the frame-
work, electronic properties, etc. Of these, we focused on
the location of the carbon atoms, which can impact
enantioselectivity. In many papers, the origin of
enantioselectivity has been explained using a two-
dimensional model, which is divided into four quan-
drants, as illustrated in Figure 1.3 This explanation puts
an emphasis on which quandrants of the four should be
shielded by the ligand. Although this model can account
for the enantiomer observed, it cannot predict the extent
of enantioselectivity. To the best of our knowledge,
there is little guidance in the literature as to where in any
quandrant the enantiodetermining groups should be
during a hydrogenation so as to maximize enantio-
selectivity.4 Moreover there is no one ligand that gives
high enantioselectivity for any substrate, that is, each
substrate has its own ligand of choice. We reasoned that
a substrate must have its best location (or ‘sweet spot’),
at which maximum shielding in the quandrant occurs
leading to high selectivity. A study on the relationship
between the locations of enantiodetermining groups
within several related ligands and enantioselectivity in
asymmetric hydrogenations of various substrates might
ultimately provide a prediction as to the ligand of choice
for future substrates. Described herein is a new ap-
proach for a better understanding of the positioning of
enantiodetermining groups in C2 symmetric diphosphine
ligands. The use of a template serves to position carbon
atoms in ligands at various coordinates during an
asymmetric hydrogenation of an enamide in search of
the reaction’s ‘sweet spot’ (Fig. 2).
R '
M
P
P
P
P
O
M
R
N
H
Figure 2. Where is the ‘sweet spot’?
Figure 1. Quandrants.
For convenience, in order to identify the location
of enantiodetermining groups, we adopted a map-
ping approach based on MM2 calculations.5 The
enantiodetermining groups can be easily found by
* Corresponding author. Tel.: +81-463-25-2189; fax: +81-463-25-2084;
e-mail: takao_saito@takasago.com
0957-4166/$ - see front matter Ó 2004Elsevier Ltd. All rights reserved.
doi:10.1016/j.tetasy.2004.05.028

2170
H. Shimizu et al. / Tetrahedron: Asymmetry 15 (2004) 2169–2172
conformational analysis by MM2 calculations. Taking
DuPHOS6 as an example, a side view of its Rh complex
indicates that two methyl groups (C and C0) out of four
on the phospholane rings are the key enantiodetermin-
ing residues (Fig. 3).
d/2
C
z
r
θ
M
y
x
z = rsin
θ
x2+ z2= (d/2)2
x2 + y2+ z2 =r2
Figure 5. Equations.
z
O
(R)-1
(S)-2
PR2
PR2
O
O
(S)-3
(S)-4
Figure 3. Enantiodetermining groups.
x
M
O
: enantiodetermining carbon atom
Conformational analysis of metal complexes gives us
information about bond distances, angles, etc. As for C2
symmetric ligands, three kinds of data allow us to
establish the positions of the enantiodetermining groups
in three dimensions: (1) the bond distance between two
enantiodetermining carbon atoms (C and C0) (d), (2) the
bond distance between a central metal atom M and C (r)
and (3) the torsion angle of the line formed by C and M
with the plane containing two phosphine atoms and M
ðhÞ. The complex was oriented within the coordinates as
described in Figure 4. First, the location of M was set to
the origin, the plane containing two phosphine atoms
and M being situated in the x–y plane. Among lines
containing M, one which is parallel to the line formed by
two phosphine atoms was defined as the x-axis, one on
the x–y plane, which is vertical to the x axis to be the y
axis, and one which is vertical to the x–y plane to be the z
axis. Setting the location of the enantiodetermining
carbon atom C to be value (x, y, z) affords three equa-
tions (Fig. 5). By solving these equations, the value of C
can be determined.
PR2
:
P
P
P
P
IPT-SEGPHOS 1
MPL-SEGPHOS 2
IPR-SEGPHOS 3 Cy-SEGPHOS 4
(3.1, 2.0) (3.0, 2.0)
(x, z):
(3.4, 2.3)
(3.3, 1.6)
Figure 6. Designed ligands and their calculated location of their
enantiodetermining carbon atoms.
oxazole as their framework7 and two alkyl groups on
each phosphorus atom.
According to the calculation,8 (4,40-bi-1,3-benzodiox-
ole)-5,50-diylbis(3,3-di-isopropylphosphetane) 1 (IPT-
SEGPHOS)9 should be located at the grid point of (3.4,
0.1, 2.3), whereas (4,40-bi-1,3-benzodioxole)-5,50-diyl-
bis(2,5-dimethylphospholane) 2 (MPL-SEGPHOS)10;11
(3.3, 0.3, 1.6), (4,40-bi-1,3-benzodioxole)-5,50-diylbis(di-
isopropylphosphine) 3 (IPR-SEGPHOS)12 (3.1, 0.4, 2.0)
and (4,40-bi-1,3-benzodioxole)-5,50-diylbis(dicyclohexyl-
phosphine) 4 (Cy-SEGPHOS)13 (3.0, 0.4, 2.0). As dif-
ferences in y values are negligible, only values in the x–z
planes (from viewpoint A) were considered.
z
y
Preparations of the (S)-1 and (aS,S,S)-2 are shown in
Scheme 1. The Grignard reagent derived from 3,4-
methylenedioxybromobenzene was treated with diethyl
chlorophosphate to afford phosphonate 5. Lithiation of
5 followed by coupling with FeCl3 gave diphosphonate
6. Resolution of 6 and reduction with lithium aluminium
hydride gave diphosphine 7. Finally, phosphetane rings
of 1 and phospholane rings of 2 were formed by
exposing 7 with the corresponding cyclic sulfates in the
presence of LDA.
(x, y, z)
P
C
P
M
C'
x
view A
Figure 4. Fixing a metal complex into a coordinate.
Next, we designed and synthesized C2 symmetric
ligands, illustrated in Figure 6, so that their enantio-
determining groups occur at different locations within
the coordination picture above. When discussing a
relationship between the locations and enantioselectivi-
ties, it would be desirable to eliminate electronic effects
to the extent possible as they may also impact enanti-
oselectivity. Therefore, all of the ligands were designed
to have similar electronic conditions, with bisbenzodi-
Synthesis of (+)-3 is described in Scheme 2. Treatment
of the Grignard reagent derived from 1-bromo-3,4-
methylenedioxybenzene with chlorodiisopropylphos-
phine followed by oxidation gave the phosphine oxide.
After coupling of the phosphine oxide, resolution was
performed with (+)-dibenzoyltartaric acid [(+)-DBT].
Reduction of the diphosphine oxide gave (+)-3, where
(+)-4 was prepared following the literature.13

H. Shimizu et al. / Tetrahedron: Asymmetry 15 (2004) 2169–2172
2171
Table 1. Asymmetric hydrogenation of methyl a-acetamidocinnamate
O
P(O)(OEt)2
P(O)(OEt)2
O
O
O
O
H
2 (0.4 MPa)
O
O
a
b
CO2Me
CO2Me
[Rh(cod)(L)]OTf (S/C = 200)
*
P(O)(OEt)2
Br
MeOH, 30oC, 14-16 h
5
NHAc
NHAc
O
8
6
O
Entry
Ligand
Conversion (%)a
>99
Eea (config.)
O
PH2
PH2
c
O
O
d
e
1
(S)-1
63 (S)
75 (S)
97 (R)
95 (R)
98 (R)
98 (R)
P
P
O
O
2
(aS,S,S)-2
(+)-3
4
>99
>99
>99
>99
>99
3
O
4(+)-
5b;c
6b
O
7
(+)-3
(+)-4
1
2
O
a Conversions and ees were determined by HPLC with Daicel Chiralcel
OJ column.
b Hydrogen pressure (0.1 MPa) and THF as solvent was adopted.
c Reaction time was 4h.
P
P
O
O
O
sufficiently interact with substrate 8, the length of which
ꢀ
is estimated to be 8.8 A.
(ii) MPL-SEGPHOS 2, IPR-SEGPHOS 3 and Cy-
SEGPHOS 4 have similar bond distances between their
ꢀ
two enantiodetermining groups (7.3–7.4 A). In this case,
Scheme 1. Synthesis of 1 and 2. Reagents and conditions: (a) (i) Mg,
(ii) ClP(O) (OEt)2; (b) (i) LDA, (ii) FeCl3; (c) (i) (+)-di-p-toluoyltar-
taric acid ((+)-DTT), (ii) NaOH, (iii) LiAlH4, Me3SiCl; (d) (i) cyclic
sulfate of 2,2-diisopropyl-1,3-propane-diol, LDA, (ii) H2O2, (iii)
HSiCl3, PhNMe2; (e) cyclic sulfate of (2R,5R)-2,5-hexanediol, LDA.
IPR-SEGPHOS 3 and Cy-SEGPHOS 4 have more
‘gaps’ in steric repulsion between the two transition
states. Ligands 3 and 4 appear to shield the sweet spot in
these asymmetric hydrogenations, that is, the grid point
around (3.0, 2.0) is the sweet spot for the asymmetric
hydrogenation of 8.
O
P(O)R2
O
O
O
O
O
O
a
b
P(O)R2
P(O)R2
Br
O
Moreover, Me-DuPHOS is known to give 98% ee in
hydrogenation of 8.6 When we convert its enantiodeter-
mining groups to the coordinate scheme in Figure 4, it
locates its methyl groups at the point (3.0, 2.1), very close
to that of 3 and 4, supporting our working hypothesis
that shielding the sweetest spot should give high ees.
O
O
P(O)R2
P(O)R2
PR2
PR2
c
O
O
d
O
O
*
*
*
*
O
O
In conclusion, a new approach for investigating the im-
pact of the location of the enantiodetermining groups on a
ligand on enantioselectivity has been developed. Al-
though more data is necessary to ensure accuracy, map-
ping the location of the key enantiodetermining residue
on a ligand is effective in finding the sweet spot to be
shielded for high enantioselection in asymmetric hydro-
genations. Accumulation of the data with more ligands
should reveal the location of sweet spots. Making alibrary
of sweet spots may make it possible to select a ligand of
choice out of a huge number of diphosphine ligands.
Moreover, it will also enable the design of new ligands for
a particular substrate based on the position of its sweet
spot. That is, ligand design will be performed so that
the ligand should effectively shield a targeted sweet
spot.
R = i-Pr: 3
R = Cy: 4
Scheme 2. Synthesis of 3 and 4. Reagents and conditions: (a) (i) Mg,
(ii) ClP(i-Pr)2, (iii) H2O2 or (i) Mg, (ii) ClP(O)Cy2; (b) (i) LDA, (ii)
FeCl3; (c) (i) DBT, (ii) NaOH; (d) HSiCl3, (i-Pr)2NEt (for 3) or
PhNMe2 (for 4).
Cationic Rh catalysts bearing these ligands were used in
the asymmetric hydrogenation of the commonly studied
substrate, methyl a-acetamidocinnamate, 8. Results are
shown in Table 1. A complex bearing 1 gave a com-
paratively moderate ee, whereas that bearing 2 gave a
somewhat higher enantioselectivity. Curiously, (S)-1
gave the same configuration of the product derived from
8 with (aS,S,S)-2, although (S)-1 should shield com-
plementary quandrants relative to that of (aS,S,S)-2. On
the other hand, Rh complexes ligated by (+)-3 or (+)-4
gave excellent ees. Further optimization revealed that
reduced hydrogen pressure (0.1 MPa) and using THF as
solvent improved the ee up to 98%.
It should be noted that in this study, only (x, z) values
are discussed. The effects of protrusion of enantio-
determining groups over a metal atom (y values) and
electronic properties are currently under investigation,
which may explain the inversion of configuration ob-
served with 1 and 2. The absolute configurations of 3
and 4 are also yet to be determined. Efforts to synthesize
more ligands, which shield other locations within the
coordinate are also in progress.
Our interpretation of these results is as follows:
(i) enantiodetermining carbon atoms of IPT-SEGPHOS
ꢀ
1 are too far from each other (ca 8.4 A) and thus cannot

2172
H. Shimizu et al. / Tetrahedron: Asymmetry 15 (2004) 2169–2172
2H), 1.28 (dd, J ¼ 12.6, 23.1 Hz, 2H), 0.89 (d, J ¼ 6.6 Hz,
References and notes
6H), 0.88 (d, J ¼ 6.6 Hz, 6H), 0.71 (d, J ¼ 6.6 Hz, 6H),
0.68 (d, J ¼ 6.6 Hz, 6H); 31P NMR (CD2Cl2) d )27.0; EI-
1. For a recent review: Tang, W.; Zhang, X. Chem. Rev.
2003, 103, 3029.
MS m=z 554(M) þ; ½aꢀ ¼ )226.5 (c 1.0, CHCl3 98.8% ee).
D
10. Diphospholane ligands with biaryl backbone were synthe-
sized by Schmid and Achiwa (a) Schmid, R.; Broger, E.
A.; Cereghetti, M.; Crameri, Y.; Foricher, J.; Lalonde, M.;
2. Blaser, H.-U.; Malan, C.; Pugin, B.; Spindler, F.; Steiner,
H.; Studer, M. Adv. Synth. Catal. 2003, 345, 103.
3. Koening, K. E.; Sabacky, M. J.; Bachman, G. L.;
Christopfel, W. C.; Barnstorff, H. D.; Friedman, R. B.;
Knowles, W. S.; Stultz, B. R.; Vineyard, B. D.; Weinkauff,
D. J. Ann. N.Y. Acad. Sci. 1980, 16, 333.
€
Muller, R. K.; Scalone, M.; Schoettel, G.; Zutter, U. Pure
Appl. Chem. 1996, 68, 131; (b) Fujie, N.; Matsui, M.;
Achiwa, K. Chem. Pharm. Bull. 1999, 47, 436.
11. (aS,S,S)-(4,40-Bi-1,3-benzodioxole)-5,50-diylbis(2,5-di-meth-
ylphospholane) (aS,S,S)-2: Mp 163–165 °C; 1H NMR
(CDCl3): d 7.19 (dd, J ¼ 3.3, 8.2 Hz, 2H), 6.95 (d,
J ¼ 8.2 Hz, 2H), 6.03 (m, 2H), 5.98 (d, J ¼ 1.1 Hz, 2H),
2.65–2.50 (m, 2H), 2.42–2.26 (m, 2H), 2.23–2.05 (m, 4H),
1.64(dddd, J ¼ 2.2, 4.9, 13.2, 24.2 Hz, 2H), 1.50–1.37 (m,
2H), 1.33 (dd, J ¼ 7.1, 18.1 Hz, 6H), 0.90 (dd, J ¼ 7.1,
10.4, 6H); 31P NMR (CDCl3): d )2.3; EI-MS m=z 469
4. Lipkowitz, K. B.; Kozlowski, M. C. Synlett 2003,
1547.
5. MM2 calculation was carried by CAChe.
6. (a) Burk, M. J. Am. Chem. Soc. 1991, 113, 8518; (b) Burk,
M. J.; Feaster, J. E.; Nugent, W. A.; Harlow, R. L. J. Am.
Chem. Soc. 1993, 115, 10125.
7. Saito, T.; Yokozawa, T.; Ishizaki, T.; Moroi, T.; Sayo, N.;
Miura, T.; Kumobayashi, H. Adv. Synth. Catal. 2001, 343,
264.
(M)H)þ; ½aꢀ ¼ +255.4( c 1.0, CHCl3).
D
12. (+)-(4,40-Bi-1,3-benzodioxole)-5,50-diylbis(diisopropyl-phos-
ꢀ
8. Our calculation afforded values below; 1: d ¼ 8.36 A,
1
ꢀ
ꢀ
ꢀ
phine) (+)-3: Mp 84–86 °C; H NMR (CDCl3): d 7.03 (d,
r ¼ 4.18 A, h ¼ 33.4°; 2: d ¼ 7.44 A, r ¼ 3.73 A,
ꢀ
ꢀ
J ¼ 7.9 Hz, 2H), 6.88 (d, J ¼ 7.9 Hz, 2H), 5.95 (2H,
J ¼ 1.5 Hz, 2H), 5.80 (d, J ¼ 1.5 Hz, 2H), 2.19–2.06 (m,
2H), 1.84–1.72 (m, 2H), 1.10 (dd, J ¼ 7.1, 14.2 Hz, 6H),
1.07 (dd, J ¼ 7.1, 14.2 Hz, 6H), 0.93 (dd, J ¼ 7.1, 13.1 Hz,
6H), 0.90 (dd, J ¼ 7.1, 11.0 Hz, 6H); 31P NMR (CDCl3):
h ¼ 26.0°; 3: d ¼ 7.32 A, r ¼ 3.68 A, h ¼ 33.0°; 4:
ꢀ
ꢀ
d ¼ 7.30 A, r ¼ 3.68 A, h ¼ 33.2°.
9. (S)-(4,40-Bi-1,3-benzodioxole)-5,50-diylbis(3,3-di-isopropyl-
phosphetane) (S)-1: Mp 68–70 °C; 1H NMR (CD2Cl2): d
6.98–6.94(m, 2H), 6.87 (d, J ¼ 7.7 Hz, 2H), 5.97 (d,
J ¼ 1.1 Hz, 2H), 5.84(d, J ¼ 1.1 Hz, 2H), 2.13–2.07 (m,
2H), 1.92–1.87 (m, 2H), 1.70 (dd, J ¼ 12.6, 23.1 Hz, 2H),
1.66 (dd, J ¼ 6.6, 6.6 Hz, 2H), 1.47 (dd, J ¼ 6.6, 6.6 Hz,
d ꢁ 3:0; CI-MS: m=z 475 (M)þ; ½aꢀ ¼ +6.5 (c 1.0, CHCl3).
D
13. Saito, T.; Yokozawa, T.; Zhang, X.; Sayo, N. U.S.
5,872,273, 1999.