Asymmetric reduction of ketones with catecholborane using 2,6-Bodol complexes of titanium(IV) as catalysts

Article abstract of DOI:10.1002/1521-3765(20010518)7:10<2158::AID-CHEM2158>3.0.CO;2-I

Reductions performed with Ti^IV complexes of ligands based on bicyclo[2.2.2]octane diols 5 and 6 are effective catalysts in the reduction of prochiral ketones to optically active alcohols, with catecholborane as the reducing agent. Methyl ketones are favored and enantiomeric excesses (ee) of ≤98% have been achieved with acetophenone as the substrate. Several other substrates were tested, among them 2-octanone, which gave 2-octanol in 87 % ee. Further details of the method were examined, for example, temperature, solvent composition, amount of molecular sieves (4 A), and catecholborane quality, as well as the sensitivity of the ligands towards acids. NMR spectroscopic methods were used to gain some insight into the complexes formed between the ligands and [Ti(OiPr)₄]. A dimeric structure is proposed for the pre-catalyst.

Full text of DOI:10.1002/1521-3765(20010518)7:10<2158::AID-CHEM2158>3.0.CO;2-I

FULL PAPER
Asymmetric Reduction of Ketones with Catecholborane Using 2,6-BODOL
Complexes of Titanium(iv) as Catalysts
Ian Sarvary,[a] Fredrik Almqvist,[b] and Torbjörn Frejd*[a]
Abstract: Reductions performed with
TiIV complexes of ligands based on
bicyclo[2.2.2]octane diols 5 and 6 are
effective catalysts in the reduction of
prochiral ketones to optically active
alcohols, with catecholborane as the
reducing agent. Methyl ketones are
favored and enantiomeric excesses (ee)
of ꢀ98% have been achieved with
acetophenone as the substrate. Several
other substrates were tested, among
them 2-octanone, which gave 2-octanol
in 87% ee. Further details of the method
were examined, for example, temper-
ature, solvent composition, amount of
molecular sieves (4 Š), and catecholbo-
rane quality, as well as the sensitivity of
the ligands towards acids. NMR spectro-
scopic methods were used to gain some
insight into the complexes formed be-
tween the ligands and [Ti(OiPr)4]. A
dimeric structure is proposed for the
pre-catalyst.
Keywords: asymmetric catalysis
´
diol ligands ´ ketones ´ reduction ´
structure elucidation ´ titanates
Introduction
One of the most studied transformations in asymmetric
catalysis is the asymmetric reduction of prochiral ketones.
Several reasons may be found for this interest; the product
alcohols are often used as starting materials in the synthesis of
drugs, as ligands for catalysis, and in the synthesis of large,
complicated target molecules. Moreover, this reaction also
serves as a test for new catalysts and their ligand systems.
Some of the most efficient catalysts for this transformation
utilize oxazaborolidines as ligands and boranes as reductants
(CBS reduction).[1, 2] Modifications of these ligands have
improved the enantioselectivities so that very high levels can
be achieved, often better than 98% ee.[3]
low ee (24%) in the catalytic asymmetric reduction of
acetophenone. Shortly thereafter, Giffels et al. tested several
ligands together with [Ti(OiPr)4]; this lead to increased
enantioselectivities.[5] The highest values, ꢀ84% ee, were
obtained with bicyclic TADDOL analogues (2) and aromatic
ketones as the substrate. However, much lower eeꢀs were
obtained with nonaromatic ketones.
Systems that employ TiIV-based Lewis acids for reductions
with boranes have not been investigated to the same degree.
In 1994, Lindsley et al. showed that the addition of [Ti(OiPr)4]
accelerated the reduction of ketones with several boranes.[4]
In their investigation, a Ti-TADDOLate (1: TADDOL ˆ
a,a,a',a'-tetraaryl-1,3-dioxolane-4,5-dimethanol) produced a
Other combinations of electron-deficient metals and diol
ligands have been tried recently. Huang et al. employed ZnII
as the electron-deficient metal center and reached eeꢀs of
ꢀ80% for aromatic ketones with a polybinaphthol (3) as the
chiral ligand and catecholborane as the reductant.[6] In
addition to the diol ligands mentioned above, a GaIII complex
that contained two monothiobinaphthol ligands (4) was
reported by Ford and Woodward to give moderate-to-high
eeꢀs (ꢀ93%) in the reduction of ketones with catecholbo-
rane.[7, 8]
In connection with another project, we became interested
to see whether 1,3-diol systems, based on the bicyclo[2.2.2]oc-
tane framework, could be used as ligands for the TiIV-
catalyzed reduction of ketones with catecholborane as the
reductant. The rigidity of the bicycles and the many possible
orientations of metal-coordinating atoms or groups constitute
interesting qualities for ligand construction.
[a] T. Frejd, I. Sarvary
Organic Chemistry
Department of Chemistry, Lund University
P.O. Box 124, 221 00 Lund (Sweden)
Fax : (‡46)46-222-41-19
E-mail: torbjorn.frejd@orgk1.lu.se
[b] F. Almqvist
Department of Organic Chemistry
Ê
Ê
Umea University, 901 87 Umea (Sweden)
E-mail: fredrik.almqvist@chem.umu.se
Supporting information for this article is available on the WWW under
http://www.wiley-vch.de/home/chemistry/or from the author.
2158
ꢁ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 2001
0947-6539/01/0710-2158 $ 17.50+.50/0
Chem. Eur. J. 2001, 7, No. 10

2158±2166
208C. The alcohol formed was then isolated by chromatog-
raphy. A more detailed discussion concerning the reaction
conditions is found below.
The ligand selected for this investigation was based on a
previous observation that anisyl-BODOL (5)[12] produced
higher eeꢀs than phenyl-BODOL (6) (Scheme 1). Picolyl
These bicyclic 1,3-diols may form metal chelates with a six-
membered ring, in contrast to the seven-membered-ring
chelates of the BINOLs[9] and TADDOLs.[10] Thus, different
geometries of the metal complexes would result and, in
consequence, different properties of the active catalysts could
be expected. Although the structure of the catalyst is not
known, it was noted in our earlier report that straight-chain
aliphatic methyl ketones were reduced with higher eeꢀs by the
titanium isopropoxide complexes of the 1,3-diols 5 and 6 as
compared to those of 2.[11] Several other methyl ketones were
successfully reduced with high yields and eeꢀs. At the time it
seemed necessary to apply rigorous reaction conditions in
terms of dryness and quality for all of the reagents (it seemed
crucial to distill the [Ti(OiPr)4], as well as the ketones and
solvents before use). Moreover, the use of large amounts of
molecular sieves seemed necessary to achieve the best results.
Despite these measures, we experienced difficulties with the
reproducibility. The yields and eeꢀs varied and the catalytic
reaction sometimes failed for no apparent reason.
Scheme 1. The BODOL ligands: anisyl-BODOL (5), phenyl-BODOL (6)
and picolyl-BODOL (7), and the amino alcohol (8) used by Demir et al.[14]
-BODOL (7) has a coordinating pyridine nitrogen at approx-
imately the same distance from the hydroxy groups as the
oxygen of the methoxy group of 5. However, reductions
performed with 7 as the ligand only resulted in traces of the
expected alcohols. When the procedure described above was
followed, a gel-like mass was formed on addition of the
catecholborane. This impaired the stirring process, and the
reaction medium did not seem to allow a proper mixing of the
reactants. Therefore, 5 is used as the preferred ligand
throughout this work, unless otherwise stated.
Good-to-excellent enantioselectivity was achieved for most
of the aromatic methyl ketones used as the substrate when 5
was employed as the ligand in the reductions (entries 1A, 2A,
2B, and 3A ± 3D, Tables 1 ± 3).
When the nonaromatic part of the ketones was changed to
ethyl (entry 2A), the yield decreased; however, it did not
significantly influence the enantioselectivity. 1-Tetralone, in
which the alkyl chain is tied back into a 6-membered ring,
gave the same yield and ee as acetophenone (entry 2B). In
contrast, a somewhat lower ee was found in the case of the
five-membered ring in 1-indanone (entry 2C). The sterically
The reaction system has now been investigated thoroughly
and several details have been changed with respect to our
earlier communication. This provides a reliable and more
robust method for catalytic asymmetric reductions of prochi-
ral ketones. In addition, the synthesis of the ligands has been
improved.
Table 1. Asymmetric reduction of methyl ketones with catecholborane by
Ti-BODOLate complexes of 5, 6, and 7.
Results and Discussion
After some experimentation, it was discovered that the order
of addition of the reagents was important and the following
procedure was found to be reliable. Firstly, [Ti(OiPr)4] was
added to a solution of the ligand in tBuOMe, that contained
the molecular sieves, followed by stirring at 458C for 1.5 h.
During this time we believe that ligand exchange takes place
to generate the catalyst, the structure of which is not known at
present (see below). Subsequently, the ketone was added
followed by a waiting time of 30 min to allow for the
formation of a complex with the chiral Lewis acid. Finally,
the reaction mixture was cooled to 208C before a pre-
cooled solution of catecholborane was added. This generated
a reddish reaction mixture, which was quenched after 24 h at
Entry
Ligand
R
Yield[%][a]
ee[%]
Configuration[d]
1A
1B
1C
1D
1E
1D
5
5
6
6
7
7
Ph
nhexyl
Ph
nhexyl
Ph
nhexyl
quantitative
82
92
80
< 5
< 5
96[b]
87[c]
89[b]
72[c]
±
R
R
R
R
±
[a] Isolated yields. [b] Determined by HPLC on ChiralcelOD-H. [c] De-
termined by GC analysis of the Mosher ester on Supelco alpha-DEX.
[d] Established by optical rotation.
Chem. Eur. J. 2001, 7, No. 10
ꢁ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 2001
0947-6539/01/0710-2159 $ 17.50+.50/0
2159

T. Frejd et al.
FULL PAPER
Table 2. Catalytic asymmetric reduction of aromatic ketones.
eeꢀs and yields. In all cases, the (R) configuration was
obtained.
Entry Ketone
2A
Alcohol
Yield[%][a] ee[%][b] Configuration[c]
Reductions catalyzed by chiral, nonracemic Lewis acids
frequently do not produce high eeꢀs for linear, nonconjugated
ketones. However, some rather high values have been
reported. For the CBS reduction, Demir et al. reported the
highest ee in the reduction of 2-octanone (91% ee and 78%
yield) that used 8 with BH3 as the reducing agent.[14] In our
experiments, the linear methyl ketones (entries 4A and 1B)
produced fairly good eeꢀs. A lower ee was obtained when an
ethyl ketone, 3-octanone (entry 4B), was the substrate. It is
also interesting to note that 1-acetyl-1-cyclohexene (entry 4C)
was reduced with a good yield and ee. The alkene moiety was
intact, showing that there was no over-reduction (Table 4).
50
95
R
2B
2C
quantitative 96
R
R
89
88
±
87
86
±
2D
2E
R
±
Table 4. Catalytic asymmetric reduction of non-benzylic ketones.
[a] Isolated yields. [b] Determined by HPLC, ChiralcelOD-H. [c] Estab-
lished by optical rotation.
Entry Ketone
Alcohol
Yield[%][a] ee[%] Configuration[e]
4A
4B
80
91
85[b]
48[b]
R
R
Table 3. Catalytic asymmetric reduction of aromatic ketones, continued.
Entry Ketone
Alcohol
Yield[%][a] ee[%][b] Configuration[c]
4C
4D
97
96[c]
R
R
3A
Quant.
85
91
98
R
R
quantitative 56[d]
3B
[a] Isolated yields. [b] Determined by GC analysis of the corresponding Mosher
ester on Supelco alpha-DEX column. [c] Determined by GC, Supelco beta-
DEX column. [d] Determined by HPLC on ChiralcelOD-H. [e] Established by
optical rotation.
3C
3D
quantitative 94
R
R
93
96
The reduction of benzylacetone (entry 4D) resulted in an
unexpectedly low ee, although the yield was still high.
Lowering the temperature to 408C did not lead to any
detectable improvement of the ee, while the yield remained
the same.
[a] Isolated yields. [b] Determined by HPLC on ChiralcelOD-H. [c] Estab-
lished by optical rotation.
Reaction conditions: We generally used a catalyst loading of
10% in our reactions; this amount can be decreased to 2.5%
without loss of enantioselectivity or yield, as tested with
acetophenone.
more hindered 1-acetonaphthone also gave a lower yield and
ee than acetophenone (entry 2D). In the case of isobutyro-
phenone (entry 2E), the yield of the alcohol produced was
very low, without any noticeable ee. The amount of product
formed in this case was comparable to that of the background
reduction, that is, the reduction with just the catecholbo-
rane.[13] Apparently, the catalyst did not seem to tolerate steric
hindrance at both a-positions of the carbonyl group. For
isobutyrophenone, there was not even a catalytic effect of
[Ti(OiPr)4] without the ligand, at either 208C or 08C. Again,
low conversions comparable to those of the background
reaction were observed; for example, after 12 h reaction time
at the higher temperature, only 18% of the alcohol was
formed (Table 2).
In our preliminary experiments, the reaction temperature
was set to 508C, as measured in the external cooling bath
(ethanol/solid CO2). However, a cryostat gave better temper-
ature control and allowed the reaction to be performed at
208C without loss of enantioselectivity. At 108C the ee
was lower, and it is therefore recommended to cool the
catecholborane solution to 208C before addition, in order to
achieve the best eeꢀs. No significant change in yield or ee was
noticed when the reaction temperature was 408C, but the
reaction rate was considerably lower.
Catecholborane was used as a 1m solution in THF, which
makes it convenient to use THF as the bulk solvent, although
the reactions may be performed in mixtures of THF and
diethyl ether, toluene, or tBuOMe without significant loss of
eeꢀs and yields. To any of the mentioned solutions, varying
Four aromatic ketones with substituents in the aromatic
ring were tested next; they all gave quite high eeꢀs and yields
(Table 3). The two p-substituted acetophenones (p-methoxy
and p-ethyl, entries 3C and 3D) gave similar results, with high
2160
ꢁ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 2001
0947-6539/01/0710-2160 $ 17.50+.50/0
Chem. Eur. J. 2001, 7, No. 10

Ti-Catalysis of Ketone Reduction
2158±2166
amounts of aliphatic hydrocarbons (pentane, hexane, or
heptane) may be used as co-solvents, as long as the substrate
does not precipitate. It is also important to ensure that the
solvents are well dried.
The ligand sensitivity towards Brönsted acids became
evident when performing NMR spectroscopic analysis in
CDCl3: the ligands deteriorated. NMR investigations were
therefore carried out in solvents in which HCl formation
could not occur (toluene, benzene, etc.). Dissolution of the
ligands in CH2Cl2 for HPLC analysis also caused deteriora-
tion. Thus, the use of halogenated solvents should be avoided.
As expected, picolyl-BODOL (7) was not acid sensitive and
its NMR data were recorded in CDCl3.
To estimate the acid sensitivity of 5 and 6, we studied their
decomposition in the presence of HCl, TFA, and HOAc by
1H NMR spectroscopy (integration of the Hc signal at d ˆ 4).
Not surprisingly, on account of its electron-donating substitu-
ent in the aromatic ring, 5 was more sensitive than 6 towards
acidic conditions. Thus, both of the ligands rapidly deterio-
rated in contact with 0.1m HCl (Table 5, entries 5A and 5B).
gyloxylate-ene reaction that employs [(iPrO)2TiCl2]/BINOL
as a catalyst.[15] However, during our further investigations it
became evident that there was only a minor dependence on
the amount of the molecular sieves. Thus, the reactions may
be performed without the molecular sieves, with only a minor
loss of ee (ꢁ2%, Table 6); however, in this case the dryness of
Table 6. Enantiomeric excess (ee) versus loading of the 4 Š molecular
sieves (ms) in the reduction of acetophenone with 5.
Entry
ms[g][a]
Yield[%][b]
ee%[c]
6A
6B
6C
6D[d]
0.25
0.10
0.05
0
98
95
96
95
96.4
95.7
94.3
94.6
[a] Amount of ms per mmol ketone. [b] Isolated yields. [c] Determined by
HPLC on ChiralcelOD-H. [d] The solvents were dried over sodium/
benzophenone.
the solvent is very important. Such a minor effect of the
molecular sieves has also been observed in Diels ± Alder
reactions with [(iPrO)2TiCl2]/TADDOL as the catalyst.[16] We
believe that the molecular sieves only act as a drying agent
before the addition of [Ti(OiPr)4].
Table 5. Acidic deterioration of 5 and 6 in [D4]methanol measured by
1H NMR spectroscopy.
Entry
Ligand
Acid
Concn[m]
t1/2 [min]
5A
5B
5C
5D
5E
5F
5
6
5
6
5
6
HCl
HCl
TFA
TFA
HOAc
HOAc
0.1
0.1
0.1
0.1
0.15
0.15
< 3
< 3
< 3
The eeꢀs of the alcohols produced in the reductions was
rather insensitive to the ee of the ligand. Anisyl-BODOL (5),
purified only by flash chromatography (95% ee), produced
the same ee in the reduction of acetophenone as the more
rigorously purified sample that had been purified both by
chromatography and recrystallization (>99% ee). When the
reaction was carried out with a ligand of 60% ee, the alcohol
produced had an ee of 90%. This pronounced positive
nonlinear effect[17] is shown in Figure 1.
45
10%[a]
no degradation[a]
[a] Observed deterioration during 5 h.
The use of TFA caused the same rapid deterioration of 5,
while 6 could withstand this acid a little better (entries 5C and
5D). Both ligands could withstand prolonged exposure to
HOAc. Only small amounts of 5 had deteriorated in this
solvent after 5 h, whereas no destruction of 6 was observed,
even after longer periods (entries 5E and 5F).
Because of the sensitivity towards Brönsted acids (see
above), the reaction mixtures were quenched with saturated
NH4Cl. We previously used HCl, but this caused deterioration
of the ligands and resulted in several byproducts that were
difficult to separate from the alcohols produced. Quenching
with NH4Cl left the ligand essentially intact, enabling its
isolation by chromatography and recycling. Thus, normally
>90% of the ligand could be isolated after the reductions, but
small amounts of impurities accumulated on repeated recov-
ery. As a result, the ligand may be reused only five times
before the ee starts to drop. An interesting alternative to
improve the ligand recycling may be to immobilize it on a
solid support; this may also simplify the isolation of the
product alcohols. These measures are now being tested.
In our preliminary experiments, the use of quite a large
amount of molecular sieves appeared to be important to
obtain the best results. Their role could have been to facilitate
the ligand ± alkoxy exchange, as has been proposed in the
Figure 1. The nonlinear effect in the reduction of acetophenone with
catecholborane/[Ti(OiPr)4] in the presence of 5 in various enantiomeric
purities.
We also noticed that the reduction was catalyzed by the
ligand without addition of [Ti(OiPr)4]; however, the alcohols
produced in these reactions showed no or negligible ee. It is
possible that 5 formed borates with catecholborane and that
these may act as Lewis acids, catalyzing the reductions. The
yields from these reactions were, however, much lower (50%)
Chem. Eur. J. 2001, 7, No. 10
ꢁ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 2001
0947-6539/01/0710-2161 $ 17.50+.50/0
2161

T. Frejd et al.
FULL PAPER
compared with those obtained in the presence of both the
ligand and [Ti(OiPr)4].
yields and no ee, despite the presence of [Ti(OiPr)4]. The yield
was essentially the same as that of the background reaction.[13]
In contrast, the use of BH3 ´ THF resulted in a rapid reduction
of the ketones, but produced no or negligible ee.
A further modification of our earlier experimental details
resulted after we had noticed that the distillation of
[Ti(OiPr)4] was not necessary to achieve good results. There-
fore, we used undistilled [Ti(OiPr)4] in stock solutions
(tBuOMe) in all reductions. These solutions are stable and
can be kept at room temperature for weeks under dry
conditions without any noticeable negative influence on the
reductions.
The ketones used were usually not distilled and we found
that a surprisingly large amount of impurities were tolerated.
Ketones kept in our chemical storage for decades were used
with good results. Most impurities were removed after the
reaction by flash chromatographic purification of the alcohols.
A few percent better yields may be obtained by the use of
freshly distilled ketones.
It has been proposed that the actual reducing agent in the
[Ti(OiPr)4]-catalyzed hydroboration of olefins with catechol-
borane might be BH3, formed by disproportionation between
catecholborane and [Ti(OiPr)4].[18] Naturally, the question
arises as to whether a species formed from BH3 was the real
reducing agent, despite the fact that reductions that used
BH3 ´ THF directly did not produce any noticeable ee. The
deep red color observed in our reductions with catecholbo-
rane indicated the formation of Ti catecholates; this could be
the result of disproportionation (Scheme 2).
After our first report on this reduction, we encountered
severe problems with its reproducibility. After elimination of
several of the evident possible causes, we were left with the
suspicion that the quality of the catecholborane solutions was
crucial. 1H, 13C, and 11B NMR spectroscopy was used to check
the quality and large quantities of several species were
revealed in those batches of catecholborane that had given
inferior results (Figure 2).
Scheme 2. Possible formation of Ti-catecholates and BH3 ´ THF from
catecholborane and [Ti(OiPr)4], as proposed by Burgess and van der
Donk.[18]
The Ti catecholate formed may be essential for the
asymmetric reductions by being part of the catalyst. For
example, Ti catecholates 9 or 10 could exchange two ligand
sites for the diol of the BODOLs. In order to test this idea,
catechol was added to the [Ti(OiPr)4]/BODOL mixture to test
whether the Ti catecholate was formed. Indeed, this was
indicated since the same red color appeared as in the
catecholborane reductions. Next, acetophenone was added
and then the mixture was cooled to 208C followed by the
addition of BH3 ´ THF (pre-cooled to 208C). This resulted in
a rapid consumption of the ketone; however, the alcohol
formed was completely racemic. The addition of catechol
before or after the acetophenone or a reduction of the
temperature further ( 508C) did not produce any ee either.
Thus, the red complex, whatever its structure may be, did not
have any influence on the stereoselectivity when BH3 ´ THF
was used as the reductant. We may conclude that the effective
reducing agent in our reductions was not BH3. With this in
mind, it is very remarkable that (despite the presence of BH3 ´
THF in the good catecholborane solution) the reductions
produce such high eeꢀs.
Figure 2. 11B NMR of 50% catecholborane solution (1m in THF) and 50%
[D8]toluene. A: Good quality of the catecholborane, B: bad quality of the
catecholborane. Both spectra were recorded on a Bruker DRX 500 MHz
spectrometer at 258C with Et2O ´ BF3 in [D8]toluene as the reference.
As seen in the 11B NMR spectrum of the catecholborane
solutions of good quality, the resonance of catecholborane
itself (d ˆ 26.5) dominated relative to the signals from the
impurities. However, in the catecholborane solutions of bad
quality the area of signal was approximately of the same
magnitude as those of the impurities. These signals at d ˆ 15 ±
25 may originate from B(OR)3 compounds formed on
unintended contact of catecholborane with water. It should
also be noted that the quartet at d ˆ 0.5 in the solutions of
both good and bad quality most likely originated from BH3 ´
THF. However, the quartet located at d ˆ 20 in the good
quality catecholborane is not identified, but it is evidently an
indication of the presence of free BH3 or another of its
complexes.[4, 18] Apparently, the storage and handling of
catecholborane solutions must be more carefully controlled
and checked than we anticipated. After we had established
the importance of the catecholborane quality, the reductions
turned out to be very reproducible.
Maturation periods in titanium chemistry are quite com-
mon and this was also observed in our reductions. However,
NMR experiments did not show any evident changes during
several hours after the initial complex formation from a
mixture of 5 and [Ti(OiPr)4] in toluene.
Two other common boranes were tested as reducing agents:
9-BBN and BH3 ´ THF. The more bulky 9-BBN gave low
2162
ꢁ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 2001
0947-6539/01/0710-2162 $ 17.50+.50/0
Chem. Eur. J. 2001, 7, No. 10

Ti-Catalysis of Ketone Reduction
2158±2166
We generally allowed the mixture of the molecular sieves
(4 Š), the ligand, and the solvent to stir for 2 ± 4 hours before
the addition of the [Ti(OiPr)4]; although if the solvents had
been rigorously dried, this drying period was not necessary.
After the addition of [Ti(OiPr)4], a maturation period of 1.5 h
at 458C, or 10 ± 12 hours at room temperature was used
(Table 7, entries 7C and 7F). When this period was less than
was not suitable for an X-ray analysis. Only amorphous
material has hitherto been obtained.
Since the reactions were performed in toluene with good
results, we studied the complexes formed in [D8]toluene
solution by NMR spectroscopy. Thus, addition of one
equivalent [Ti(OiPr)4] to the pure enantiomer 5 resulted in
several downfield shifts of the ligand resonances (Figure 3,
traces A and B), which suggests coordination to TiIV. The most
noticeable changes were the following: a large downfield shift
of the methoxy signal (0.4 ppm), a smaller downfield shift of
the aromatic region and a large downfield shift of Hc
(0.55 ppm). We also noted that the signal of the bridgehead
proton at d ˆ 2.55 was not shifted (not shown). It was also
clear that CH signals of two different iPrO groups appeared at
d ˆ 4.95 and 4.3 (Figure 3, trace B). The large downfield shift
of the methoxy protons indicated that the ether oxygen acted
as an extra coordination site to make 5 a tridentate ligand.
This extra coordination site is not available in 6 and may be
related to the fact that 6 is a less efficient ligand than 5 in most
cases.
Table 7. The effect of varying the catalyst aging conditions for the
reduction of acetophenone with 5.
Entry
Ageing period[h]
T[8C]
ee[%][a]
Yield[%][b]
7A
7B
7C
7D
7E
7F
0.5
4
11
23
23
23
23
45
45
80
89
96.7
75
90
73
92
95
85
90
97
75
48
0.5
1.5
0.5
96.9
77
7G
25 (sonication)
[a] Determined by HPLC on ChiralcelOD-H. [b] Isolated yields.
As judged by its NMR spectrum, this initially formed
complex seemed to be quite stable for several hours and was
not effected by the addition of three equivalents of iPrOH.
However, when this complex was kept in toluene solution for
48 h at room temperature, other complexes were formed.
These aged complexes could account for the decreased ee and
yield mentioned above (Table 7, entry 7D).
Further structural information on the pre-catalyst was
obtained by the use of a partially enantiomerically enriched
sample of the ligand. If the complex contained two ligands,
two diastereomeric complexes would be formed that could be
detected by their NMR spectra. We therefore recorded the 1H
MNR spectrum of a sample prepared by the addition of one
equivalent of [Ti(OiPr)4] to a 40% enantiomerically enriched
sample of 5, which had been obtained by mixing the two
enantiomers of 5 in the proper portions (70:30). The resulting
1H NMR spectrum clearly showed that two complexes were
formed. They had similar, but different spectral features
(Figure 3, trace C). The appearance of the signals at d ˆ 3.6,
1 hour at room temperature, there was a drop in ee of 17%
(entry 7A). A mixture kept at 458C for 30 min or at room
temperature (ꢁ218C) for 4 h gave approximately the same
results (entries 7E and 7B). The maturation period must not
be too long, however, because when the mixture was kept at
room temperature for more than 24 hours, the reaction
became sluggish and gave a considerably lower ee (entry 7D).
It should be mentioned that sonication of the mixture for
30 min (entry 7G) did not improve the ee.
To test the reliability of the method, a reaction was
performed on a somewhat larger scale (3 g) for 4-meth-
oxy-acetophenone with
a 5% catalyst loading. There
was no significant change in either ee or chemical yield,
relative to the 150 mg experiments. Therefore, we do not
expect any special difficulties for the application of the
method on even larger scales.
Structure of the pre-catalyst: In
the absence of more precise
data concerning the structure
of the active catalyst, it could
still be of interest to acquire
structural information of the
pre-catalyst, that is the com-
plex(es) formed on mixing the
metal component with the li-
gand. As already discussed, our
catalyst showed
a nonlinear
behavior; the ee of the product
was considerably higher than
that of the ligand. This indicat-
ed that the actual catalyst may
contain more than one ligand.
Unfortunately, the solid mate-
rial obtained from equimolar
portions of 5 and [Ti(OiPr)4]
Figure 3. 1H NMR spectra in [D8]toluene of A: free ligand, B: mixture of enantiopure 5 and one equivalent of
[Ti(OiPr)4], C: mixture of 5 (40% ee) and one equivalent of [Ti(OiPr)4].
Chem. Eur. J. 2001, 7, No. 10
ꢁ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 2001
0947-6539/01/0710-2163 $ 17.50+.50/0
2163

T. Frejd et al.
FULL PAPER
4.5, and 7.25, which have the expected areas of the minor
diastereomeric component, indicated a structure which con-
tains two ligands.
that LiBr is formed in a Wurtz coupling between ArLi and
BuBr and then may form an ªateº complex with CeCl3 to give
Li [CeCl3Br] , which may not give a useful Ar± Ce reagent.
‡
Complexes involved in the Sharpless Ti ± tartrate-catalyzed
asymmetric epoxidation of allylic alcohols have been studied
with NMR spectroscopic methods by Potvin et al.[19] These
structures have the possibility of extra coordination to Ti from
the carbonyl of the ester or amide groups. Despite the fact
that there is no X-ray data available for the Ti ± diethyl-
tartrate esters, it is assumed that they coordinate in a
tridentate fashion. It is also assumed that these complexes
are dimers,[20±23] [(TiL)2], although it has been disputed that
these are the active catalysts.[24] The situation has been
summarized recently.[25] Naturally, similar questions arise in
our case. Thus, when all the
Our previously described method of inversion of ( )-12
gave access to (‡)-12.[27] During these transformations, we
observed that the ArLi ± CeCl3 reaction with 15, which has a
bulky TBDPS (tert-butyldiisopropylsilyl) protective group,
did not work under the same reaction conditions that were
used for the synthesis of 14. The starting material was
recovered to a large extent. Therefore, it was necessary to
change the protective group from TBDPS to TBDMS (tert-
butyldimethylsilyl) for the conversion of 15 to ( )-5.
Conclusion
data is taken into considera-
tion, it seems likely that the
dominating species is a rather
stable C2-symmetric complex,
such as 11, which is composed
of two ligands and two titanium
atoms. However, further spec-
troscopic evidence must be ob-
tained before a more rigorous
The presented enantioselective reduction conditions are mild
and provides a useful method for the catalytic asymmetric
reduction of ketones. It is also worth noting that, although
sensitive towards branching at the carbonyl a-position, the
Ti ± anisyl-BODOL complex (Ti-5) is quite substrate tolerant
and produced good enantioselectivities and chemical yields
for a wide range of substrates. Based on preliminary structural
data, a tentative structure of the pre-catalyst is suggested to be
the dimeric complex 11.
structure determination of the
pre-catalyst system can be
made.
Experimental Section
Synthesis of the ligands: BODOLs 5 ± 7 were synthesized
according to a published procedure with some adjustments
(Scheme 3).[11] Although this scheme seems straightforward,
General remarks: GC analyses was performed with either an alpha- or
beta-DEX column (Supelco, 30 m  0.25 mm i.d.  0.25 mm film thickness).
HPLC analyses were performed on a ChiralcelOD-H column (0.46 cm Â
25 cm, 5 mm particle size), with iPrOH and hexane as the eluents. NMR
spectra were recorded at 400 MHz (BrukerDRX spectrometer), if not
stated otherwise. Optical rotations were measured with a Perkin ± Elmer
241LC polarimeter at 238C. Preparative chromatographic separations
were performed on normal-phase silica gel 60 (0.035 ± 0.070 mm, Matrex
Amicon). Thin-layer chromatography was performed on TLC plates
precoated with silica gel60F-254, 0.25 mm (Merck). After elution, the
TLC plates were visualized with UV light followed by spraying with a
solution of p-methoxybenzaldehyde (26 mL), glacial acetic acid (11 mL),
concentrated sulfuric acid (35 mL), and 95% ethanol (960 mL), followed
by heating. All solvents were dried over 4 Š molecular sieves (5% m/w) for
24 h prior to use, unless stated otherwise. The molecular sieves were
activated at 4008C for 6 h and then allowed to cool under argon. [Ti(OiPr)4]
(97%, Aldrich), phenyllithium (2.0m in benzene/ether, Aldrich), and
catecholborane (1m in THF, Aldrich) were used as delivered. In our
Scheme 3. Synthetic route to the BODOL ligands. a) TBDMSCl, DMF,
imidazole; b) ArLi, CeCl3, THF, 788C ! RT; c) Bu4NF, THF. Synthetic
route to
( )-5: d) i) DHP (dihydro-2H-pyran), PPTS (pyridinium-p-
toluenesulfonate), CH2Cl2; ii) NaBH4, MeOH/THF; iii) TBDPSCl, DMF,
imidazole; iv) PTS, iPrOH; v) Swern oxidation; e) i) Bu4NF, THF;
laboratory, the catecholborane solutions were stored at
reductions have not been optimized for any particular reaction; however,
the temperature selected was that which was best suited to acetophenone.
208C. The
ii) TBDMSCl, DMF, imidazole; iii) ArLi, CeCl3, THF,
iv) Bu4NF, THF.
788C ! RT;
(1R,2R,4S,6S)-2-(2-Anisyl)-bicyclo[2.2.2]octane-2,6-diol (5): CeCl3 (3.6 g,
9.7 mmol) was placed in a round-bottomed flask and dried at 1108C under
vacuum (<1 mm Hg) overnight (14 h). The flask was cooled on ice under
Ar, then ice-cold, dry THF (25 mL) was added. The resulting mixture was
stirred for 5 min at 08C and then for at least 4 h at RT before use. This
should result in a ªmilky suspensionº without any large fragments.[28] The
suspension was then cooled to 788C, followed by addition of anisyllithi-
um [prepared by addition of nBuLi (1.6 mL, 12 mmol), to anisole (1.6 mL,
15 mmol) in THF(20 mL)] and the resulting yellow mixture was stirred for
1 h. A solution of (1R,4S,6S)-6-(t-butyldimethylsilyloxy)bicyclo[2.2.2]oc-
tane-2-one (13, 1.5 g, 5.9 mmol) in THF (5 mL) was then added, and the
mixture was allowed to reach RT overnight. A saturated aqueous solution
of NH4Cl (25 mL) was added, the phases were separated, and the water
phase was extracted with ether (2 Â 20 mL). The combined organic phases
were dried (Na2SO4) and filtered through a pad of SiO2, in order to remove
some comments should be made. The treatment of the
protected keto-alcohol 13 with aromatic Grignard and lithium
reagents did not result in a 1,2-addition to give 14 unless CeCl3
was added to the reagent.[26] The most likely explanation for
this is the formation of an enolate in the absence of CeCl3
since 13 could be recovered after aqueous work-up
(Scheme 3).
It was also necessary to introduce the lithium by direct
metalation, since ArLi syntheses based on a halogen ± metal
exchange with BuLi were nonproductive, even in the presence
of CeCl3. The reason for this is unclear; however, it is possible
2164
ꢁ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 2001
0947-6539/01/0710-2164 $ 17.50+.50/0
Chem. Eur. J. 2001, 7, No. 10

Ti-Catalysis of Ketone Reduction
2158±2166
inorganic cerium salts. The solvent was removed under vacuum, and the
residue was diluted with cold, dry THF (30 mL). Bu4NF (3 g, 9.5 mmol) was
added to this solution and the deprotection was monitored by TLC. The
reaction was complete after 4 hours. The reaction mixture was concen-
trated under reduced pressure to yield an orange oil, which was diluted with
EtOAc (50 mL). The organic solution was washed with brine (3 Â 20 mL)
and dried (Na2SO4). Filtration and removal of the solvent under reduced
pressure gave an oil that was purified by flash chromatography (SiO2,
heptane/EtOAc 50:50, Rf ˆ 0.26). The resulting clear syrup crystallized in
the refrigerator to give 1.18 g (81%) of white crystals, which were
recrystallized (tBuOMe/hexane, 20:80) to afford 5, m.p. 82 ± 838C.
1H NMR data in CDCl3 were identical with those previously reported.[11]
However, since we found that the compound was acid sensitive, we give
here the NMR data in benzene: 1H NMR ([D6]benzene, 258C): d ˆ 7.08
(dd, J(H,H) ˆ 1.65, 7.74 Hz, 1H), 7.01 (ddd, J(H,H) ˆ 1.65, 7.43, 8.2 Hz,
1H), 6.83 (td, J(H,H) ˆ 7.47, 1.2, Hz, 1H), 6.45 (dd, J(H,H) ˆ 1.2, 8.2 Hz,
1H), 4.55 (d, J(H,H) ˆ 11.2 Hz, 1H; OH), 4.45 (s, 1H; OH), 4.07 (m, 1H;
CHOH), 3.05 (s, 3H; OMe), 2.56 (m, 1H), 2.2 (m, 2H), 2.02 (dt, J(H,H) ˆ
14.7, 2.5 Hz, 1H), 1.93 (m, 1H), 1.25 (m, 2H), 1.08 (m 2H); 13C NMR
([D6]benzene, 258C): d ˆ 157.9, 135.8, 128.2, 126.8, 121.0, 112.0, 78.3, 71.0,
54.8, 42.8, 39.8, 39.5, 26.5, 23.5, 21.8.
cept that racemic endo-6-hydroxy-bicyclo[2.2.2]octan-2-one was employed
instead of the optically active material. The product was obtained as white
crystals, m.p. 101 ± 1028C. 1H and 13C NMR spectral data were identical to
those of 5.
General method for the reduction of the ketones: A solution of the ligand
(25 mg, 0.1 mmol) in tBuOMe was dried by stirring with activated 4 Š
molecular sieves (0.4 g) for 2 h at room temperature. Then [Ti(OiPr)4]
(0.5 mL, 0.19m in tBuOMe, 95 mmol) was added. The mixture was kept for
1.5 h at 458C before addition of the ketone (1 mmol, neat or dissolved in a
minimum amount of THF). The wall of the reaction vessel was rinsed with
hexane (0.75 mL) to ensure that all of the ketone had been added to the
reaction mixture. The mixture was stirred at room temperature for another
30 min and was then cooled to 208C. Cold ( 208C, from the freezer)
catecholborane (1.5 mL, 1m in THF, 1.5 mmol) was added, and the reaction
was monitored by TLC. After 24 h, only traces of the starting material
could be detected. Aqueous saturated NH4Cl (5 mL) was added at 208C
to the reaction mixture, which was then stirred for 10 min. Diethyl ether
(25 mL) and another portion of aqueous saturated NH4Cl (15 mL) were
added at RT, followed by stirring for 1 h. The phases were separated, and
the aqueous phase was extracted with diethyl ether (2 Â 25 mL). The
combined ether phases were washed with NaOH (1m, 2 Â 25 mL), dried
(Na2SO4), and the solvent was removed under reduced pressure. The
residue was purified by flash chromatography (SiO2, diethyl ether/pentane,
and then diethyl ether) to elute the unreacted ketone, then the alcohol, and
then the ligand. This procedure was used for the preparation of the
following alcohols: (1R)-1-phenylethanol (entry 1A), (2R)-octan-2-ol (en-
try 1B), (1R)-1-phenyl-propan-1-ol (entry 2A), (1R)-1,2,3,4-tetrahydro-
naphth-1-ol (entry 2B), (1R)-indan-1-ol (entry 2C), (1R)-1-(1-naphthyl)-
ethanol (entry 2D), (1S)-1-(2-methoxyphenyl)-ethanol (entry 3A), (1R)-1-
(3-methoxyphenyl)-ethanol (entry 3B), (1R)-1-(4-methoxyphenyl)-ethanol
(entry 3C), (1R)-1-(4-ethylphenyl)-ethanol (entry 3D), (2R)-hexan-2-ol
(entry 4A), (3R)-octan-3-ol (entry 4B), (1R)-1-(cyclohex-1-en-1-yl)-ethan-
1-ol (entry 4C), (2R)-4-phenylbutan-2-ol (entry 4D). For further details see
Tables 1 ± 7 and the Supporting Information.
(1R,2R,4S,6S)-2-Phenyl-bicyclo[2.2.2]octane-2,6-diol (6): This compound
was prepared by the procedure described for 5 except that phenyllithium in
benzene/ether was used instead of anisyllithium. Flash chromatography
(SiO2, heptane/EtOAc 2:1, Rf ˆ 0.21) gave a clear syrup which crystallized
in the refrigerator (78%). Recrystallization from tBuOMe/hexane (20:80)
gave 6, m.p. 85 ± 878C. 1H NMR data in CDCl3 were identical to those
previously reported.[11] However, since we found that the compound was
acid sensitive we give here the NMR data in benzene solution: 1H NMR
([D6]benzene, 258C): d ˆ 7.35 (brd, J(H,H) ˆ 6.6 Hz, 2H), 7.2 (brt,
J(H,H) ˆ 7.7 Hz, 2H), 7.1 (tt, J(H,H) ˆ 7.3, 1.9 Hz, 1H), 4.1 ± 3.7 (brs;
1H; OH), 3.8 (m; 1H; CH), 3.8 ± 3.5 (brd, 1H; OH), 2.19 (m, 1H), 1.99 (m,
1H), 1.9 (m, 1H), 1.76 (m, 1H), 1.66 (m, 2H), 1.2 ± 0.84 (m, 4H); 13C NMR
([D6]benzene, 258C): d ˆ 148.0, 128.4, 127.3, 126.8, 76.9, 71.0, 42.7, 42.5,
38.3, 26.4, 23.3, 20.7.
Reductions for the investigation of the nonlinear effect: Samples that
contained the ligand in different enantiomeric compositions were prepared
as stock solutions (0.10m) by mixing the appropriate amounts of the
racemate and the pure enantiomer in tBuOMe. The enantiomeric
compositions of the product and the purity of the ligand samples were
determined by HPLC analysis (ChiralcelOD-H, hexane/iPrOH 80:20)
before the reductions, and for the ligand also after the reductions. This
showed that the ligand retained its enantiomeric composition.
(1R,2R,4S,6S)-2-(2-Picolyl)-bicyclo[2.2.2]octane-2,6-diol (7): This com-
pound was prepared by the procedure described for
5 except that
2-picolyllithium was used instead of anisyllithium. Flash chromatography
(SiO2, heptane/EtOAc 20:80, Rf ˆ 0.29) resulted in a light yellow syrup
(77%). [a]2D1 ˆ 23.8 (c ˆ 6.5, CHCl3); 1H NMR (CDCl3, 258C): d ˆ 8.37
(brd, J(H,H) ˆ 4.7 Hz, 1H), 7.66 (td, J(H,H) ˆ 7.7, 1.8 Hz, 1H), 7.2 (dd,
J(H,H) ˆ 7.7, 5.6 Hz, 1H), 7.16 (d, J(H,H) ˆ 7.8 Hz, 1H), 4.93 (brs, 1H;
OH), 3.8 (brs, 1H; CH), 2.97 (ABq, J(H,H) ˆ 14.9 Hz), 2.13 (m, 1H), 1.81
(m, 2H), 1.65 ± 1.5 (m, 4H), 1.4 ± 1.20 (m, 4H); 13C NMR (CDCl3, 258C):
d ˆ 159.5, 148.6, 137.6, 124.8, 122.2, 76.5, 71.0, 47.1, 44.0, 39.3, 38.9, 26.1, 23.6,
21.3; HRMS (FAB ‡ , direct inlet) calcd: 234.1494; found: 234.1493.
Acknowledgements
(Æ)-endo-6-Hydroxy-bicyclo[2.2.2]octane-2-one: The synthetic sequence
was adopted from the literature.[29] The first step, the synthesis of 3-allyl
cyclohexanone, was performed according to the literature[30] from 2-cyclo-
hexenone (3.0 g, 31 mmol), TiCl4 (3.7 mL, 33 mmol), and allyltrimethylsi-
lane (5.0 mL, 32 mmol), with the exception that after the addition of water
(200 mL) the mixture was filtered through Celite to remove the thick
precipitate of titanium oxide. In the second step, the crude product from
above (4 g) was diluted with MeOH (100 mL) and cooled to 768C, and O3
was bubbled through the stirred solution until it turned blue-violet. The
solution was purged with argon gas until colorless, followed by the addition
of methyl sulfide (5.0 mL, 68 mmol). The temperature was then allowed to
rise slowly to reach RT (over 14 h). Concentration under reduced pressure
gave an yellow oil (4.3 g) that was diluted with acetone (40 mL) and HCl
(40 mL, 1m). This solution was heated under reflux (1 h) and then allowed
to cool to room temperature. The mixture was diluted with brine (75 mL)
and then extracted with EtOAc (4 Â 25 mL). The combined organic phases
were washed with saturated NaHCO3 (25 mL) and brine (25 mL), and then
dried (Na2SO4), filtered, and concentrated under reduced pressure. The
oily residue was purifed by chromatography (SiO2, heptane/EtOAc 1:3) to
yield the compound as a white solid (0.93 g, 6.5 mmol, 21%). M.p 200 ±
2038C (sublimation and decomp); lit.: m.p. 201[31] and 165.3 ± 167.2[32]).
1H NMR and 13C NMR spectral data were in full agreement with those
reported.[32]
We thank the Swedish Natural Science Research Council, the Crafoord
Foundation, the Knut and Alice Wallenberg Foundation, and the Royal
Physiographic Society in Lund (Sweden) for financial support.
[1] S. Wallbaum, J. Martens, Tetrahedron: Asymmetry 1992, 3, 1475 ± 1504.
[2] L. Deloux, S. Morris, Chem. Rev. 1993, 93, 763 ± 784.
[3] E. J. Corey, C. J. Helal, Angew. Chem. 1998, 110, 2092 ± 2118; Angew.
Chem. Int. Ed. 1998, 37, 1986 ± 2012.
[4] C. W. Lindsley, M. DiMare, Tetrahedron Lett. 1994, 35, 5141 ± 5144.
[5] G. Giffels, C. Dreisbach, U. Kragl, M. Weigerding, H. Waldmann, C.
Wandrey, Angew. Chem. 1995, 107, 2165 ± 2166; Angew. Chem. Int. Ed.
Engl. 1995, 34, 2005 ± 2006.
[6] W.-S. Huang, Q.-S. Hu, L. Pu, J. Org. Chem. 1999, 64, 7940 ± 7956.
[7] A. Ford, S. Woodward, Angew. Chem. 1999, 111, 347 ± 349; Angew.
Chem. Int. Ed. 1999, 38, 335 ± 336.
[8] A. J. Blake, A. Cunningham, A. Ford, S. J. Teat, S. Woodward, Chem.
Eur. J. 2000, 6, 3586 ± 3594.
[9] K. Mikami, Y. Motoyama, in Encyclopedia of Reagents for Organic
Synthesis (Ed.: L. A. Paquette), Wiley, Chichester, 1995, pp. 397 ± 403.
[10] R. Dahinden, A. K. Beck, D. Seebach, in Encyclopedia of Reagents for
Organic Synthesis (Ed.: L. A. Paquette), Wiley, Chichester, 1995,
pp. 2167 ± 2170.
(Æ)-o-Anisyl-2,6-BODOL: The same procedure was used as for the
synthesis of (1R,2R,4S,6S)-2-(2-anisyl)-bicyclo[2.2.2]octane-2,6-diol, ex-
Chem. Eur. J. 2001, 7, No. 10
ꢁ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 2001
0947-6539/01/0710-2165 $ 17.50+.50/0
2165

T. Frejd et al.
FULL PAPER
[11] F. Almqvist, L. Torstensson, A. Gudmundsson, T. Frejd, Angew.
Chem. 1997, 109, 388 ± 389; Angew. Chem. Int. Ed. Engl. 1997, 36, 376 ±
377.
[12] We suggest that the diols based upon the bicyclo[2.2.2]octane frame-
work be named BODOLs (bicyclo[2.2.2]octanediols).
[13] The background reaction is defined as the reduction of the ketone
with the borane, in the absence of [Ti(OiPr)4] and the ligand.
[14] A. S. Demir, I. Mecitoglu, C. Tanyeli, V. Gülbeyaz, Tetrahedron:
Asymmetry 1996, 7, 3359 ± 3364.
[24] E. J. Corey, J. Org. Chem. 1990, 55, 1693 ± 1694.
[25] A. C. Oehlenschlager, in Houben-Weyl Methods of Organic Chem-
istry: Stereoselective Synthesis, Vol. 8 (Eds.: G. Helmchen, R. W.
Hoffmann, J. Mulzer, E. Shaumann), Thieme, Stuttgart, 1996,
pp. 4683 ± 4695.
[26] T. Imamoto, Y. Sugiura, N. Takiyama, Tetrahedron Lett. 1984, 25,
4233 ± 4236.
[27] F. Almqvist, T. Frejd, Tetrahedron: Asymmetry 1995, 6, 957 ± 960.
[28] If there are large fragments or if the CeCl3 forms a solid on the bottom
of the flask, a dry glass rod may be used to manually crush the
fragments. This generally results in a milky solution.
[15] K. Mikami, M. Terada, T. Nakai, J. Am. Chem. Soc. 1990, 112, 3949 ±
3954.
[16] D. Seebach, R. Dahinden, R. E. Marti, A. K. Beck, D. A. Plattner,
F. N. M. Kühlne, J. Org. Chem. 1995, 60, 1788 ± 1799.
[17] D. G. Blackmond, in Catalysis of Organic Reactions (Ed.: F. E.
Herkes), Marcel Dekker, New York, 1998, pp. 455 ± 465.
[18] K. Burgess, W. A. van der Donk, Organometallics 1994, 13, 3616 ±
3620.
[29] K. Mori, E. Nagano, Biocatalysis 1990, 3, 23 ± 36.
[30] A. Hosomi, H. Sakurai, J. Am. Chem. Soc. 1977, 99, 1673 ± 1675.
[31] K. Mori, Y. Nakahara, M. Matsui, Tetrahedron 1972, 28, 3217 ± 3226.
[32] B. De Santis, A. L. Iamiceli, R. M. Bettolo, L. M. Migneco, R.
Scarpelli, G. Gerichelli, G. Fabrizi, D. Lamba, Helv. Chim. Acta
1998, 81, 2375 ± 2387.
[19] P. G. Potvin, B. G. Fieldhouse, Can. J. Chem. 1995, 73, 401 ± 413.
[20] I. D. Williams, S. F. Pedersen, K. B. Sharpless, S. J. Lippard, J. Am.
Chem. Soc. 1984, 106, 6430 ± 6431.
[21] S. F. Pedersen, J. C. Dewan, R. R. Eckman, K. B. Sharpless, J. Am.
Chem. Soc. 1987, 109, 1279 ± 1282.
Note added in proof: For a recent method for the enantioselective
reduction of linear ketones see: B. A. Persson, A. L. E. Larsson, M. Le Ray,
J.-E. Bäckvall, J. Am. Chem. Soc. 1999, 121, 1645 ± 1650.
[22] K. A. Jörgensen, R. A. Wheeler, R. Hoffmann, J. Am. Chem. Soc.
1987, 109, 3240 ± 3246.
[23] C. J. Burns, C. A. Martin, K. B. Sharpless, J. Org. Chem. 1989, 54,
2826 ± 2854.
Received: August 4, 2000
Revised version: December 21, 2000 [F2647]
2166
ꢁ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 2001
0947-6539/01/0710-2166 $ 17.50+.50/0
Chem. Eur. J. 2001, 7, No. 10