Comparison of enantioselective reductions using bead and monolith 'disk' polymer formulations of CBS catalysts

Article abstract of DOI:10.1016/j.tet.2007.02.048

Two polymer-supported versions of the Corey, Bakshi, and Shibata (CBS) catalyst were prepared and examined. Polymeric beads, with the auxiliary bound in pendant and crosslinked fashion, were prepared utilizing an improved procedure based upon earlier work. Optimization of a procedure for ketone reduction gives results that match those in solution. Attempted reuse gave mixed results. For comparison purposes, CBS functionalized monoliths were formed and tested but performed poorly.

Full text of DOI:10.1016/j.tet.2007.02.048

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
Tetrahedron 63 (2007) 3334–3339
Comparison of enantioselective reductions using bead and
monolith ‘disk’ polymer formulations of CBS catalysts
Michael C. Varela, Seth M. Dixon, Michael D. Price, Jeffrey E. Merit, Patrick E. Berget,
*
Saori Shiraki, Mark J. Kurth and Neil E. Schore
Department of Chemistry, University of California, One Shields Avenue, Davis, CA 95616, United states
Received 15 December 2006; revised 8 February 2007; accepted 9 February 2007
Available online 15 February 2007
Abstract—Two polymer-supported versions of the Corey, Bakshi, and Shibata (CBS) catalyst were prepared and examined. Polymeric beads,
with the auxiliary bound in pendant and crosslinked fashion, were prepared utilizing an improved procedure based upon earlier work. Opti-
mization of a procedure for ketone reduction gives results that match those in solution. Attempted reuse gave mixed results. For comparison
purposes, CBS functionalized monoliths were formed and tested but performed poorly.
Ó 2007 Elsevier Ltd. All rights reserved.
1. Introduction
identical to those found in solution phase. However, a num-
ber of attempts to recycle and/or regenerate the polymer-
bound catalyst have not given satisfactory results. In the
course of this work we determined the enantiopurities of
the chiral monomers used in the polymerization in order to
properly calibrate the polymer-supported reduction results.
The original homogenous CBS catalyst 1 gives excellent
enantiomeric excess and conversion in the course of pro-
moting the borane reduction of prochiral carbonyl com-
pounds.1,2 However, as with many homogenous systems, it
is not ideally suited for separation and recycling of the
catalyst. Other methods of homogenous borane-based asym-
metric reduction have been developed and used with success
similar to that of the CBS model, but also suffer from the
same problem.3 We have developed and optimized a hetero-
genous, polymer-supported method based upon the B-methyl-
ated CBS system. We considered its diphenyl functionality
as a promising site for modification so as to allow well-
controlled polymerization; this was achieved successfully
to form both crosslinked and pendant species 2 and 3
(R¼iPr), respectively, in bead form. As with many hetero-
genous systems based upon homogenous analogs, a drop
was observed in both activity and selectivity.4 In contrast
to the solution-phase versions of these reactions, we found
that the polymer-supported borane reductions suffered to a
varying degree from competition by nonstereoselective solu-
tion-phase reduction. We demonstrated previously that the
polymer-bound crosslinked catalyst 2 could achieve enantio-
meric excess (ee) reasonably close to those of solution-phase
systems.5 The pendant system 3, however, gave poorer
results. Using a method of ‘seeding’ resin 2 with borane
we have been able to achieve enantioselectivies virtually
Ph
Ph
R
O
O
O
N
N
N
B
B
B
1
2
3
2. Results and discussion
2.1. Monomer syntheses
In our initial work the crosslinking monomer was formed
by using commercially available N-Boc-L-proline 4 and
converting it into the corresponding methyl ester 5 using
dicyclohexylcarbodiimide (DCC). Grignard addition of p-
vinylphenylmagnesium bromide provided the polymeric
precursor 6 (Scheme 1). Premature polymerization was oc-
casionally observed by the presence of insoluble solid during
workup of the Grignard reaction. We minimized this compli-
cation by purifying the product immediately after aqueous
workup and storing it under argon as a standardized solution
in chlorobenzene.
Keywords: CBS catalyst; Crosslinking monomer; Enantioselective reduc-
tion; Polymer beads; Polymer disks.
* Corresponding author. Tel.: +1 530 752 6263; fax: +1 530 754 7610;
e-mail: schore@chem.ucdavis.edu
The pendant monomer was formed in
a similar
fashion using Weinreb amide 7.6 Treatment of 7 with
0040–4020/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tet.2007.02.048

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
M. C. Varela et al. / Tetrahedron 63 (2007) 3334–3339
3335
p-isopropylphenylmagnesium bromide to form ketone 8,
followed immediately by the addition of p-vinylphenylmag-
nesium bromide gave the pendant polymerization monomer
9 (Scheme 1).
the Teoc group. Both compounds were subjected to Grignard
reactions to form the Teoc-protected monomers 6T and 9T.
Based upon chiral HPLC measurements both of these mono-
mers showed reasonable retention of the integrity of the
original proline stereocenter. Monomer 6T consisted of a
95:5 S/R mixture (90% ee) while 9T exhibited 94:6 S/R
(88% ee). Separation of the diastereomers arising from the
second side-chain stereocenter in 9Twas not observed under
our chiral HPLC conditions.
O
Ar
Ar
OR
OH
N
N
Boc
Boc
4 R = H
Suspension polymerization of monomers 6Tand 9Twas per-
formed successfully in a manner similar to that used to form
the Boc-protected crosslinking and pendant resins to yield
beads of polymers 10T and 11T, respectively. Teoc depro-
tection using tetrabutylammonium fluoride (TBAF) and
KF$2H2O proceeded cleanly to form the CBS precursors
14 and 15 (Scheme 3). In the infrared we observed complete
disappearance of the carbamate stretch at 1690 cmꢁ1 and no
trace of absorption corresponding to polymer-bound cyclic
carbamate.
5 R = Me
6 Ar = p-styryl
O
Ar
R'
R'
Boc
OH
N
N
Boc
7 R' = MeONMe
i
8 R' = p-PrC6H4
9 Ar = p-styryl
Scheme 1. See Ref. 5.
O
Ar1
Ar2
OH
Teoc
a
b
c
R
5 or 7
2.2. Polymer synthesis and deprotection
N
N
Teoc
As we previously reported, suspension co-polymerization of
Boc-protected 6 with styrene to form beads of resin 10 pro-
ceeded in high yield. Divinylbenzene was added as a cross-
linker to monomer 9 form beads of the pendant resin 11.
6T Ar1,2 = p-styryl
9T Ar1 = p-styryl,
Ar2 = p-PrC6H4
5T R = OMe
7T R = MeONMe
i
Subsequent deprotection attempts on these polymers using
methanolic DMSO under basic (KOH) conditions yielded
significant amounts of the cyclic carbamates 12 and 13.
Although we previously found a narrow range of conditions
to permit hydrolysis of these compounds to the desired sup-
ported amino alcohols, this process was difficult to carry out
reproducibly in acceptable yield (Scheme 2). We therefore
considered alternative means of protection to circumvent
this problem.
R
R
d
OH
OH
NH
N
Teoc
10T R = polystyrene
11T R = p-PrC6H4
14 R = polystyrene
15 R = p-PrC6H4
i
i
Scheme 3. (a) Teoc-OSu, Et3N, CH3CN, 0–25 ꢀC, 18 h; (b) For 5T/6T,
p-CH2]CHC6H4MgBr, THF, 0–25 ꢀC, 4 h; for 7T/9T, p-iPrC6H4MgBr,
THF, 45 ꢀC, 24 h, then after workup p-CH2]CHC6H4MgBr, THF, 0–25 ꢀC,
4 h; (c) For 10T, same as for 10; for 11T, same as for 11; (d) TBAF,
KF$2H2O, THF, H2O, 72 h, 65 ꢀC.
R'
R'
We also attempted to make resins with higher catalyst load-
ing by combining monomers 6T and 9T in polymerizations
in which 6T would be the sole crosslinker. However, numer-
ous attempts to make such resins via suspension polymeriza-
tion failed. At best we obtained only minute amounts
(<20 mg) of solids that would not swell in THF, DCM or
benzene.
a
b
OH
6 or 9
O
N
N
Boc
O
10 R' = polystyrene
11 R' = p-PrC6H4
12 R' = polystyrene
13 R' = p-PrC6H4
i
i
Scheme 2. (a) For 6/10, C6H5CH]CH2, (C6H5CO2)2, C6H5Cl, H2O,
85 ꢀC, 24 h; for 9/11, C6H5CH]CH2, p-(CH2]CH)2C6H4,
(C6H5CO2)2, C6H5Cl, H2O, 85 ꢀC, 24 h; (b) KOH, DMSO, MeOH, 65 ꢀC,
4 days.
2.4. Oxazaborolidine formation and enantioselective
reductions
2.3. Teoc-protected monomers: enantiopurity and
polymerization
Treatment of the supported amino alcohols 14 and 15 with
trimethylboroxine yielded oxazaborolidines 2 and 3, respec-
tively.5 The newly formed catalysts were readied for use by
means of multiple azeotropic distillations with toluene to
remove excess boroxine and any moisture that could inter-
fere with borane reduction. The resin was subsequently dried
under reduced pressure. Acetophenone was chosen as the
prochiral ketone due to the ease of NMR and chiral GC anal-
ysis of the reaction mixture and product. Reductions were
Compounds 5 and 7 were deprotected with TFA and imme-
diately reprotected using 1-[{2-(trimethylsilyl)ethoxy}-
carbonyloxy]pyrrolidin-2,5-dione (Teoc-OSu) to form the
corresponding silyl carbamates. In the NMR both the
Teoc-protected ester 5T and amide 7T showed distinct roto-
mers at 25 ꢀC, presumably due to restricted rotation about
the bond from the ring nitrogen to the carbonyl carbon of

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
3336
M. C. Varela et al. / Tetrahedron 63 (2007) 3334–3339
Table 1. Reductions of acetophenone to 1-phenylethanol using beads of
polymeric CBS catalysts 2 and 3
Table 3. Regeneration of beads of previously recycled catalyst 2
Reuse #
Yield
ee (corr)
Catalyst
Seeding time
Yield
ee (corra)
1
2
3
100
100
100
46 (51)
31 (34)
52 (58)
2
2
2
2
3
w
55
96
96
97
70 (78)
86 (96)
81 (90)
89 (99)
55 (63)
15 min
30 min
60 min
60 min
100
unpredictable way the ability of the reagents and substrates
to access catalytic sites in the interior. Note, however, that
examination of the swelling characteristics of the beads
before and after use showed no significant changes.
a
Values in parentheses are corrected for the enantiopurity of the polymer
precursor.
initially performed by swelling the resin in a minimum
amount of THF, followed by cooling and addition of
BH3$THF, and followed immediately by the ketone. We
found that this method often failed to give complete conver-
sion of the ketone to the alcohol over the course of 24 h. To
address the possibilities that either some moisture may have
still remained on the resin or that complexation of borane to
the polymer-bound catalyst might be slow, we explored
a method in which we ‘seeded’ the polymer with borane
prior to addition of the ketone. Borane was added to the
swelled resin and cooled to 0 ꢀC for a period of time, after
which a second aliquot of borane was added together with
the ketone. The reduction was allowed to proceed for 24 h
and the reaction worked up. Seeding for short periods gave
somewhat erratic results, but the results using a one-hour
pre-incubation time were consistently quite good. Not only
did this seeding method provide nearly complete conversion
of the ketone to the alcohol, but it also gave superior enan-
tioselectivity results (Table 1).
In any event we chose to attempt to completely regenerate
the catalyst using the same methods used in its initial forma-
tion. The resin recovered from the experiments described in
Table 2 was dried, swollen in THF, and trimethylboroxine
was added. The regeneration attempt was allowed to proceed
for 3 days followed by azeotropic distillation with toluene.
The resulting resin was seeded for 60 min and used in three
successive reductions, giving the slightly less erratic but still
unsatisfactory results shown in Table 3.
The same recycling and regeneration experiments were per-
formed on beads of the pendant resin 3 with little success.
All recyclings gave complete conversion, but all products
were racemic. In these cases it is clear that non-selective
solution phase reduction completely dominates with respect
to catalysis by polymer-supported material. A partial expla-
nation may be found in an examination of the CBS-derived
material that remains subsequent to solution phase use. We
have carried out reductions using the original diphenylproli-
nol-derived methylboroxazoline-based CBS catalyst in suf-
ficient quantity to allow the fate of the catalyst to be at
least partially ascertained by NMR. We find that little or
no intact CBS material survives the procedure. Instead,
a complex mixture is obtained exhibiting NMR signals
consistent with dialkoxy and aminoalkoxymethylboranes,
as well as derivatives of methylboronic acids.7 Thus it
appears that the catalyst breaks down to fragments suitable
neither for efficient reuse nor regeneration under the condi-
tions that we have employed.
Note that the ee values obtained in several of these reduc-
tions using crosslinking monomer 14 (i.e., polymer 2)
approach those of the monomers prior to suspension poly-
merization. Thus the actual enantioselectivity of reduction
corrected for the enantiopurity of the catalysts in several of
these cases is quite high.
2.5. Recycling and regeneration experiments
Having found that the 60 min seeding time gave the best re-
sults, we attempted reuse and regeneration of this particular
resin. The used resin was recovered, dried, seeded with bo-
rane, and treated with acetophenone. The reductions were
allowed to proceed for 24 h; their results are summarized
in Table 2.
2.6. Synthesis and use of functionalized monolith-
derived polymer disks
In addition to resins in bead form, we prepared monoliths
from the Teoc-protected monomers 6T and 9T, using
a method similar to that previously described.8 To our
knowledge, these are the first monoliths of their kind. In or-
der to maintain good swelling properties we chose a cross-
linking level of 2%, about half that of the previously
prepared beads, resulting in lower loading of monomer 6T.
After polymerization, the monoliths were cut to form disks
roughly 2 mm in height and 5 mm in diameter. These disks
were then washed in a Soxhlet apparatus to remove any un-
reacted monomer and styrene. Upon drying under reduced
pressure, the disks were then deprotected using TBAF and
treated with trimethylboroxine to presumably form CBS-
functionalized disks. In addition to crosslinker and pendant
disks, 2D and 3D, respectively, a set of ‘mixed’ disks
(2/3D) were also formed using a mixture of the crosslinker
and pendant monomers in an attempt to increase net loading
of catalytic sites.
Based upon these incomplete conversions and erratic enan-
tioselectivities we considered the possibility that decompo-
sition of the oxazaborolidine may have occurred during
the reduction, perhaps including reversion back to the amino
alcohol. In addition, although the beads exhibited no visible
effects of their prior use, it seemed possible that the internal
pore structure may have been compromised, affecting in an
Table 2. Recycling of beads of catalyst 2 (all seeded for 60 min)
Reuse #
Yield
ee (corr)
1
2
3
29
98
55
19 (21)
61 (68)
40 (44)

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
M. C. Varela et al. / Tetrahedron 63 (2007) 3334–3339
3337
Each of these three types of disks was placed in a batch-flow
type reactor, the entire apparatus was dried under vacuum
and flushed with argon to ensure a moisture free environ-
ment. The disks were swollen in THF, cooled to 0 ꢀC fol-
lowed by seeding for 60 min with borane. After this
period, the ketone was slowly added via syringe and the re-
actor agitated for 24 h. The solution was collected, the disks
were washed three times with THF, and the filtrates com-
bined. Although this method gave nearly complete conver-
sion in all cases, only disk 2D gave significant selectivity.
The enantioselectivities obtained from the others were
very low (Table 4).
monitored from 200–400 nm with a 0.2 mL/min flow rate
installed in a system consisting of a Waters 2695 HPLC,
a Waters PDA 996, and a Waters Micromass ZQ detector
for obtaining low-resolution MS (ESI) data at concentrations
of w1 mg/mL. Reduction products were analyzed using an
Astec G-TA b-cyclodextrin trifluoroacetylated 25 m GC chi-
ral column. Elemental analyses were provided by Midwest
and Galbraith Laboratories.
4.1.1. (2S)-1-([2-Trimethylsilyl]ethoxycarbonyl)-2-(me-
thoxycarbonyl)pyrrolidine (5T). A solution of L-proline
methyl ester (2.82 g, 21.8 mmol) in 17 mL of CH3CN was
treated with 8.7 mL of Et3N and cooled to 0 ꢀC. Commer-
cially available 1-[2-(trimethylsilyl)ethoxycarbonyloxy]-
pyrrolidin-2,5-dione (Teoc-OSu) (10.1 g, 40.0 mmol) was
added at 0 ꢀC and the mixture was allowed to warm to rt,
and stirred for 18 h. The solution was then diluted with
Et2O and extracted with water (3ꢂ), 3% HCl, water, 10%
NaHCO3, and satd NaCl, and dried (Na2SO4). The solvent
was removed under reduced pressure to give 5.99 g (99%
crude yield) of 5T as an orange oil. Upon chromatography
this compound co-eluted with unreacted Teoc-Su (1H NMR
br s at d 2.88 ppm) that could not be completely removed
without significant loss of material; thus 5Twas used in sub-
sequent transformations without further purification. NMR
spectroscopy indicated the presence of two rotomers in ap-
Table 4. Reductions using CBS-functionalized disks
Disk
Yield
ee (corr)
2D
3D
2/3D
100
93
95
40 (44)
3 (3)
5 (6)
The low selectivities are clearly attributable to the low load-
ing, perhaps combined with poor diffusion of the reagents
into the disks, thus favoring the uncatalyzed solution-phase
reduction. Numerous attempts were made to circumvent this
problem, e.g., by increasing seeding time; however, all re-
ductions gave similarly poor results. Thus, in contrast to
our successful application of the disk methodology to transi-
tion-metal catalyzed processes,9 it is clear that disks are
a poor option in circumstances where competition from a
solution-phase process interferes.
1
proximately of 52:48 ratio. H NMR d 4.34 and 4.28 (both
dd, J¼3.4, 8.6 Hz, total 1H), 4.20–4.08 (m, 2H), 3.71 and
3.70 (two s, 3H), 3.60–3.38 (m, 2H), 2.26–2.15 (m, 1H),
2.02–1.84 (m, 3H), 1.00 and 0.92 (dd, J¼8.2, 8.6 Hz and
dd, J¼7.0, 10.2 Hz, total 2H), 0.01 and 0.00 (two s, total
9H) ppm. 13C NMR d 175.0 and 174.9, 156.8 and 156.3,
65.1 and 64.9, 60.5 and 60.3, 53.7 and 53.6, 48.2 and 47.8,
32.5 and 31.4, 25.9 and 25.0, 19.34 and 19.31, 0.1 and
0.0 ppm. LRMS calcd for C12H24NO4Si (M+1) 274.14;
found 274.06. Anal. Calcd for C12H23NO4Si: C, 52.72; H,
8.48; N, 5.12. Found: C, 54.54, 54.79; H, 8.73, 8.69; N,
4.75, 5.09.
3. Conclusions
We have completed optimization studies of several versions
of polymer-supported CBS catalysts. The resin derived from
crosslinking monomer 6T is clearly and consistently supe-
rior to the pendant resin derived from 9T, and gives results
comparable to solution-phase model reductions. The physi-
cal and chemical nature of the polymer significantly affects
the outcome of reduction. Attempts to both recycle and re-
generate failed, most likely due to a combination of disas-
sembly of the oxazaborolidine and structural breakdown of
the polymer itself. Polymer disks derived from monoliths
could not be prepared with high loading and showed consis-
tently poor results.
4.1.2. (2S)-2-(N-Methoxy-N-methylcarbamoyl)-1-([2-tri-
methylsilyl]ethoxycarbonyl)pyrrolidine (7T). To a solu-
tion of 5.00 g (19.0 mmol) of 7 in 2.5 mL of CH2Cl2 was
slowly added 47.5 mL of trifluoroacetic acid (TFA). After
2 h TLC showed the complete absence of starting material.
After removal of CH2Cl2 and most of the TFA under reduced
pressure, the oily residue was taken up in 1:1 ether/hexane,
filtered, and solvent was again removed under reduced pres-
sure. The deprotected ester was not characterized, but used
immediately for Teoc protection. The crude oil was dis-
solved in 17 mL of CH3CN, 8.7 mL of Et3N was added,
and the solution was cooled to 0 ꢀC. Teoc-OSu (10.1 g,
40.0 mmol) was added at 0 ꢀC and allowed to warm to rt
and stirred for 18 h. After workup as for 5T, above, 5.82 g
of 7T was obtained (99% crude yield) as an orange oil.
NMR indicated the presence of two rotomers in a ca.
70:30 ratio. As in the case of 5T this material was used after
chromatographic separation from most but not all unreacted
Teoc-OSu. 1H NMR d 4.73 (major) and 4.67 (minor) (both br
dd, J¼3.2, 8.4 Hz, total 1H), 4.20–4.07 (m, 2H), 3.78 (ma-
jor) and 3.73 (minor) (both s, total 3H), 3.59 (m, 1H), 3.44
(m, 1H), 3.19 (s, 3H), 2.17 (m, 1H), 2.04 (m, 1H), 1.87
(m, 2H), 1.02 (major) and 0.92 (minor) (both m, total
2H), 0.02 (major) and 0.01 (minor) (both s, total 9H) ppm.
4. Experimental
4.1. General
All reactions were performed in oven-dried glassware unless
otherwise stated under an atmosphere of dry argon. 1H
and 13C spectra were recorded at 400 MHz in CDCl3. IR
spectra of liquids and solids were recorded neat using a
FTIR. Detailed preparations of 5–11 are described else-
where.5 Optical purities of chiral auxiliaries and their pre-
cursors were determined using a Daicel Chiralcel OD-RH
0.46 cm by 15 cm (I.D. by length) 5.0 mm column employ-
ing a gradient elution of 0–5 min: 100% A, 5–25 min:
0–100% B, 25–30 min: 100–0% B, 30–35 min: 100% A
(solvent A: H2O/0.1% TFA; B: Acetonitrile/0.1% TFA)