Enantioselective reduction of prochiral ketones by catecholborane catalysed by chiral group 13 complexes

Article abstract of DOI:10.1002/1521-3765(20001002)6:19<3586::AID-CHEM3586>3.0.CO;2-S

LiGaH₄, in combination with the S,O-chelate 2-hydroxy-2′-mercapto-1,1′-binaphthyl (MTBH₂), forms an active catalyst for the asymmetric reduction of prochiral ketones, with catecholborane as the hydride source. Enantioface differentiation is on the basis of the steric requirements of the ketone substituents. Aryl/ n-alkyl ketones are reduced in 90-93% ee and RC(O)Me (e.g. R = iPr, cycloC₆H₁₁, tBu) in 60-72% ee. Other borane sources and alternative catalyst structures based on indium do not form enantioselective catalysts.

Full text of DOI:10.1002/1521-3765(20001002)6:19<3586::AID-CHEM3586>3.0.CO;2-S

FULL PAPER
Enantioselective Reduction of Prochiral Ketones by Catecholborane
Catalysed by Chiral Group 13 Complexes
Alexander J. Blake,[a] Anthony Cunningham,[a] Alan Ford,[b] Simon J. Teat,[c] and
Simon Woodward*[a]
Abstract: LiGaH4, in combination with the S,O-chelate 2-hydroxy-2'-mercapto-1,1'-
binaphthyl (MTBH2), forms an active catalyst for the asymmetric reduction of
prochiral ketones, with catecholborane as the hydride source. Enantioface differ-
Keywords: asymmetric catalysis
´
entiation is on the basis of the steric requirements of the ketone substituents. Aryl/
gallium ´ indium ´ S ligands
n-alkyl ketonesare reduced in 90 ± 93% ee and RC(O)Me (e.g. R ˆ iPr, cycloC6H11,
tBu) in 60 ± 72%ee. Other borane sources and alternative catalyst structures based on
indium do not form enantioselective catalysts.
are present in numerous targets of synthetic interest. One
effective method for carrying out thisreaction isuse of the
ruthenium transfer hydrogenation catalysts introduced by
Noyori and typified by the structure (R,R)-[RuCl(h6-
Ar)(H2NCH(Ph)CH(Ph)NSO2Ar)] (Arˆ aryl).[1] Super-
lative enantioselectivities, turnover numbers and rates are
attained providing R1 in 1 isan aryl or a cloesly related
unsaturated (Un) group. However, substrates outside this
range, for example, dialkyl or a,b-unsaturated enones are
often much less suitable. Synthetically useful ee valuesfor
such substrates, and many others, can sometimes be at-
tained by using oxazaborolidine catalysed borane reductions
(CBS reduction).[2] In this case, the enantioselectivity results
from the propensity of the CBS catalyst to bind the ketone by
the most electron rich and sterically accessible lone pair (as in
the ªloadedº catalyst structure 3, Scheme 1). The primary
borate product ishydrolysed on workup to afford 2. However,
for CBS reduction relatively high catalyst loadings are
required (5 ± 20 mol%). This, taken together with the rela-
tively narrow substrate range for ruthenium transfer hydro-
genation, indicatesthat there istsill a need to identify new
efficient catalysts in ketone reduction, especially for ªprob-
lematicº substrates.
Introduction
Enantioselective reduction of prochiral ketones R1C(O)R2 (1,
Scheme 1) is an important transformation in asymmetric
synthesis as the resultant chiral sec-alcoholsR 1CH(OH)R2 (2)
s
OH
R2
H
HO
R2
H
O
asymmetric
reduction
R1
R1
R1
R2
1
2
ent-2
π-repulsion if
Ar here
Ph
Ph
R
Me
O
Rbig
O
BH3
B
Ar
O
O
Al
O
H
Li
N
Rbig
Rsmall
O
Rsmall
BH3
O
R
best if Rbig also
electron donating
3
4
Scheme 1. Asymmetric reduction of prochiral ketones and the ªloadedº
states of the CBS and BINAL reagents.
[a] Dr. S. Woodward, Dr. A. J. Blake, A. Cunningham
School of Chemistry, The University of Nottingham
University Park, Nottingham NG7 2RD (UK)
Fax : (‡44)115-951-3564
In 1979, Noyori noted that a 1,1'-(bi-2-naphthol) modified
LiAlH4 reagent (BINAL-H) could reduce aryl/alkyl ketones
in ꢀ100%ee.[3] This remarkable selectivity was proposed to
be attained by the electronically controlled transition state 4.
We determined to attempt the design of a catalytic version of
NoyoriꢀsBINAL-H reagent, aselectronic control israrely
used as an enantioselection mechanism in asymmetric catal-
ysis. This paper describes our full investigation of this area,
E-mail: a.j.blake@nottingham.ac.uk
simon.woodward@nottingham.ac.uk
[b] Dr. A. Ford
Department of Chemistry, University of Hull
Hull, HU6 7RX (UK)
[c] Dr. S. J. Teat
Daresbury Laboratory, Daresbury
Warrington, Cheshire WA4 4AD (UK)
Fax : (‡44)1925-603124
[4]
elementsof which have been communicated.
E-mail: s.j.teat@dl.ac.uk
3586
ꢁ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 2000
0947-6539/00/0619-3586 $ 17.50+.50/0
Chem. Eur. J. 2000, 6, No. 19

3586±3594
chemical yields, with significant enantioselectivity (72%),
were realised for the gallium-MTB complex. Additionally, the
moderate and reversed enantioselectivity shown by the
gallium-DTB species provided the insight that although the
ªsoft ± softº interaction of the catalytic metal/ligand is im-
portant, matching of all the Lewis acid base pairs must be
accommodated. That is, binding of the hard lithium cation in
the bimetallic catalyst must also be achieved. For example, the
poor selectivity of the LiGaH2Àn(OMe)n(DTB) species (n ˆ 1,
2) suggests that the BINAL transition state 4 may involve
additional Li ´´´ BINOL contactsat the binaphthyl alkoxides
as has been suggested before for enantioselective aldehyde
alkylation by MgEt2 in the presence of Li2(BINOL).[3e]
If the catalysis was conducted at a fixed temperature of
À408C, using LiGa(OMe)2(MTB), both the chemical yield
and ee fell to 32 and 64%, respectively. Below this temper-
ature only low yields of essentially racemic 2a were produced.
However, in reactionscatalysed by LiGaH 2(MTB) [prepared
from LiGaH4 and just MTBH2 6], with warming from À788C
to room temperature, a quantitative yield of 2a with 82%ee
was realised. These reactions suggest that the presence of
lithium alkoxide sources can promote a competing achiral
catalytic cycle. Control reactionsof catecholborane and 1a in
the presence of 10 mol% LiOMe led to 45% yield of (Æ)-2a
(À78 to 228C, 16 h), while a similar reaction using LiBHEt3
gave a 60% yield of (Æ)-2a. The simplest explanation of these
findingsisthat alkoxide osurcespromote the formation of
borohydride species [C6H4O2BH(OR)]À and that these re-
duce 1a with negligible enantioselectivity and give a racemic
borate product and regenerate the alkoxide. DiMare[6] and
Arase[7] have noted similar findings. At temperatures below
À208C, the rate of the selective gallium catalysed reaction
becomes slow and the achiral alkoxide manifold significantly
erodesthe overall enantioselectivity of the product 2a.
To maximise the enantioselectivity, the reaction temper-
ature wasfixed at À258C and use of thiolate rather than
methoxide spectator ligands investigated. Addition of
2-HSC10H7 to LiGaH2(MTB) lowered the selectivity to
72%ee, while HSCH2CH2OH waswores (20% ee). Fortu-
nately, addition of a second equivalent of MTBH2 led to a
rather selective catalyst, LiGa(MTB)2, which showed high
activity. Thisformulation provesrather robust to the reaction
conditionsand additional tsirring of the reaction mixture is
not required (runscould be conducted in any cryostat or even
a domestic freezer). The activity and selectivity of this catalyst
system (2 to 2.5 mol%) with a range of substrates were then
investigated (Table 1).
Results and Discussion
HSABmatching studies : When one designs a catalytic version
of the BINAL-H reagent, two significant problems have
to be overcome. Firstly, the reactivity of the catalyst system
must be improved relative to BINAL-H [typically 300 mol%
BINAL-H isrequired to reduce PhC(O)Me ( 1a) at À1008C].[3]
Secondly, only release of the product sec-alkoxide (produced
from 4) should occur and not the bound chiral auxiliary [the
dialkoxide of 1,1'-(bi-2-naphthol)]. If thisisnot realised, the
enantioselectivity of the catalyst system will soon be compro-
mised by chiral ligand loss. To attempt to solve these problems
we applied Pearsonꢀs ªHard/Soft Acid/Baseº (HSAB) prin-
ciple.[5] Asthe product sec-alkoxide containsa ªhardº oxygen
donor we reasoned that use of ªhardº catecholborane would
rapidly remove the product alkoxide from the catalytic centre,
whilst a suitable metal hydride is regenerated. To avoid chiral
ligand dissociation at the metal centre, we proposed to use
analoguesof 1,1 '-(bi-2-naphthol) that contain ªsoftº thiolate
donorsin conjunction with a ªsofterº gallium Lewisacid. That
is, the reagent and catalyst requirements are matched into two
HSAB pairs (ªhard/hardº for the substrate/reagent and ªsoft/
softº for the catalytic metal/chiral ligand). To test this
proposal a series of trial reactions was carried out using
LiMH4 (M ˆ Al, Ga) and the ligandsBINOLH 5, MTBH2 6
2
and DTBH2 7 (Scheme 2). Mixturesof LiMH 4 and 5 ± 7 were
i) 5 mol% catalyst
o
-78 C to RT
O
O
OH
O
B
H
+
Ph
Me
ii) workup
Ph
Me
2a
1a
with (R)-ligand (5 mol%)
Al Ga
Ligand
LiMH4
MeOH
6% R[a] 3% R[a]
BINOL
MTB
BINOLH2 5 A = B = OH
A
B
1% R[b]
72% R[b]
MTBH2 6 A = SH B = OH
DTBH2 7 A = B = SH
26% S[a]
30% S[b]
4% R[b]
DTB
(Ra)-5-7
[b] with LiM(OMe)2(Ligand)
[a] with LiMH(OMe)(Ligand)
Scheme 2. HSAB matching experimentsin the reduction of acetophe-
none 1a.
stirred at room temperature and in the presence of 1 ± 2
equivalentsof methanol and then cooled to À788C. Once
twenty equivalentsof acetophenone ( 1a, R1 ˆ Ph, R2 ˆ Me)
and catecholborane had been added (at À788C), the mixture
wasallowed to come to room temperature overnight.
Typically 70 ± 80% yieldsof alcohol 2a were isolated in all
cases. The table in Scheme 2 shows the enantiomeric purity of
the 2a produced asa function of catalyst type. It ishelpful to
know at this point that the Noyori transition state 4 predicts
that the (Ra) binaphthyl ligand leadsto the ( R) sec-alcohol
product and that in the original formulationsmethanol was
added in the BINAL-H preparation to improve the enantio-
selectivity.[3] The initial screen matrix soon revealed the
apparent usefulness of the HSAB matching approach as high
In general, substrates with a phenyl/n-alkyl motif are
reduced in good chemical yield and ee (ketones a ± c, f ± g).
The reaction is slightly affected by the presence of (Æ)-I
substituents on the 4-position. However, the presence of more
sterically demanding alkyl substituents is detrimental both to
the reaction rate and the enantioselectivity (ketones d± e, h ± i).
Similarly, the enantioselectivities realised are dependent on
the size of the aryl substituent in the ketone; both smaller
(ketone j) and larger (ketones k ± l) aryl substituents lead to
lower enantioselectivities. As the enantioselectivity criterion
for these substrates appeared to be steric rather than
electronic, as in the BINAL transition state 4, further ketones
Chem. Eur. J. 2000, 6, No. 19
ꢁ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 2000
0947-6539/00/0619-3587 $ 17.50+.50/0
3587

S. Woodward et al.
FULL PAPER
Table 1. Enantioselective reduction of ketones 1 by (Ra,Ra)-LiGa(MTB)2
(2 or 2.5 mol% unless otherwise stated) and catecholborane (1.1 equiv-
alents).
ligand). Under thispropoasl, an initial gallium hydride ( 8)
reactswith the ketone 1 to provide a gallium alkoxide (9).
Interaction of
9 with catecholborane [represented in
Ketone R1
R2
Temp [8C] Time [h] Yield [%][a] ee [%][b]
Scheme 3 asH-B(OR) ] leadsto eslective decomplexation
2
of the alkoxide, while the MTB ligand remainsanchored to
the gallium centre by virtue of a strong thiolate bond. To
investigate this hypothesis, stoichiometric reaction studies of
LiGaH4, MTBH2 (6) and catecholborane were carried out.
When treated with two equivalentsof MTBH 2 (6) at room
temperature, 20mm solutions of LiGaH4 in Et2O/THF
mixtures spontaneously lose four equivalents of hydrogen.
The nature of the clear solution formed can be investigated by
1H NMR spectroscopy. The undeuterated solvents cause
intense peaks in the spectrum at approximately d ˆ 1.1
(Et2O), 1.7 (THF), 3.3 (Et2O) and 3.6 (THF). However, the
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
Ph
Ph
Ph
Ph
Me
Et
nBu
iPr
iBu
Me
Me
À 25
18
18
18
60
60
18
18
18
19
18
20
20
60
18
18
18
18
18
18
18
18
90
96
80
65
65
80
95
60
88
76
82
83
50
70
60
88
50
81
93
72
76
90
93
92
24
92
87
87
70
68
81
59
73
0
75
63
22[d]
12[d]
69
46
72
79
À 25
À 25
À 15[c]
À 15[c]
À 25
Ph
4-BrPh
4-MePh
Ph
À 25
CH2Br À 25
CH2Ph À 20
nC6H13 À 25
Ph
2-furyl
1-C10H7
2-C10H7
Ph
Me
Me
À 20
À 20
C CH À 15[c]
ꢁ
ˆ
PhCH CH Me
À 25
À 25
À
MTB aryl C H region isclear and reproducibly indicatesthe
ꢁ
C CEt
Me
presence of two different types of MTB ligand in a 3:1 ratio
consistent with the formulation [Ga2(MTB)3][Li2(MTB)]. An
identical spectrum is obtained from mixtures of GaCl3
etherate, MTBH2 (6) and BuLi after filtration of the
precipitated LiCl. In the signal derived from the three
equivalent MTB ligands, one of the H3(3') binaphthyl protons
suffers a large downfield shift relative to the free ligand
MTBH2 (6). These electronic effects are also apparent in the
13C NMR spectrum of [Ga2(MTB)3][Li2(MTB)], in which one
of the signals appears at appreciably higher frequency
(13C NMR: d ˆ 158.4) than the rest (13C NMR: d ˆ 120.0 ±
139.3). Full assignment of the carbon spectrum was not
possible because of problems in obtaining adequate signal/
noise in the 1H:13C COSY spectrum. However, complete
connectivity for all the MTB 1H environmentscould be
established by 1H:1H COSY studies. Mass spectrometry is an
ideal technique to probe the constitution of such species: two
natural gallium isotopes (69/71Ga) exist approximately in a
60:40 ratio. When these solutions, assayed by 1H NMR
spectroscopy as containing only [Ga2(MTB)3][Li2(MTB)], are
subjected to negative ion electrospray mass spectrometry, a
major peak at m/z 669 with a Ga1 isotope pattern can be
identified. Thispeak correspondsto the formulation [LiGa-
(MTB)2]. A small (ꢀ1% of the base peak) signal is observed
ꢁ
C CH
nC5H11 À 25
ˆ
CH CH2
nC5H11 À 25
iPr
iBu
cC6H11
tBu
Me
Me
Me
Me
À 20
À 20
À 20
À 20
s
t
u
[a] Determined by 1H NMR, GC or by isolation. [b] Determined by GC,
HPLC or MTPA ester formation. [c] 4 mol% 10. [d] Absolute stereo-
chemistry not determined.
were selected to shed light on the mechanism. Consistent with
electronic control PhC(O)CCH (m) gave a low yield of
racemic product, while significant selectivities were realised
with benzylidene acetone (n) and 3-hexyn-2-one (o). How-
‡
ever, the former result may be indicative of poor Li binding
in the transition state due to the poor electron density in the
ketonic lone pairsof m. The negligible enantioselectivities
shown in the reduction of ynone (p) and enone (q) are not
indicative of an electronically controlled transition state. Few
asymmetric studies of the reduction of q have been reported
but stoichiometric BINAL reduces ynone p in 84%ee;[3] this
suggests that the gallium and aluminium systems may operate
by different mechanisms. Studies with dialkyl ketones also
support steric discrimination in the transition state of the
gallium catalysed process. The ee valuesfor the reduction of
ketones r± u correlate well with the steric demands of the
larger group.
at m/z 1347 and itsiostope pattern indicatesit is[Ga
-
2
(MTB)3][Li(MTB)].
When the reaction mixture from the addition of two
equivalentsof MTBH 2 (6) to LiGaH4 islayered with pentane,
colourless crystals form. This procedure reproducibly gives
not the solution species [Ga2(MTB)3][Li2(MTB)] but [Li-
(THF)3Ga(MTB)2] (10), whose x-ray structure is shown in
Figure 1. This compound is isostructural with [Li(THF)3Al-
(BINOL)2], which has been characterised by Shibasaki from
Mechanistic studies: One working model for the gallium/
MTB catalysed ketone reductions is shown in Scheme 3
(where X is the spectator ligand, normally a second MTB
O
reactionsof AlCl and BINOLH2 (5) in the presence of
3
BuLi.[8] If one changesfrom an Al-BINOL to a Ga-MTB
R1
R2
R2
1
À
complex, the average M Y (Yˆ O, S) distances increase from
O
H
X
H
R1
S
S
Ga
Li
1.75 Š in [Li(THF)3Al(BINOL)2] to 1.89 (Yˆ O) and 2.24
(Yˆ S) Š in [Li(THF)3Ga(MTB)2] (10).
Ga
O
O
X
On redissolution in THF, the solid-state structure of
[Li(THF)3Ga(MTB)2] (10) isnot retained. Proton spectra of
these solutions indicate regeneration of [Ga2(MTB)3]-
[Li2(MTB)]. However, thisistuation iscomplicated by the
apparent decomposition of some 10 by minor protonolysis
Li
(THF)n
H
B(OR)2
(RO)2B
R1
(THF)n
O
(R)-8
(R)-9
R2
Scheme 3. Working model for the gallium catalysed reduction (X ˆ MTB
ligand).
3588
ꢁ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 2000
0947-6539/00/0619-3588 $ 17.50+.50/0
Chem. Eur. J. 2000, 6, No. 19

Enantioselective Reduction of Prochiral Ketones
3586±3594
isquite different and not all of the chloridesare liberated. The
solution structure of 11 could not be unequivocally deter-
mined because of its relatively low solubility, fluxionality and
its pronounced tendency to undergo MTB ligand solvolysis.
For example, positive ion FAB mass spectra of 11 show
‡
[Li2InCl(MTB)2] as the highest mass ion. Additional peaks
assigned to In4Cl4O4 and itsdaughter ionsare obesrved in
some samples of aged 11. The proton NMR spectra of 11 at
ambient temperature indicate that the MTB ligandsare
equivalent in solution. Use of freshly prepared mixtures of
Li2(MTB) and InCl3(THF)3 leadsto the clean reduction of
cyclohexylmethylketone 1t in 70% yield but negligible enan-
tioselectivity at a 2 mol% catalyst loading.
Addition of a slight excess of catecholborane to a 20mm
solution of [Ga2(MTB)3][Li2(MTB)] in THF (prepared from
LiGaH4 and MTBH2 6) leadsto the clean formation of a new
species. Only very broad and uninformative NMR resonances
could be observed at room temperature. On cooling the
sample to À508C, these sharpen significantly. In addition to
unreacted catecholborane, the spectra suggest that two non-
equivalent MTB ligandsare preesnt together with a bound
catechol fragment. However, no gallium or boron hydride
signal could be detected. Layering of these 1H NMR samples
with pentane led only to the formation of oils. A second
catalytic state can be identified spectroscopically by sequen-
tially treating 20mm solutions of [Ga2(MTB)3][Li2(MTB)]
with catecholborane followed by pinacolone 1u (asa repre-
Figure 1. Molecular structure of [Li(THF)3Ga(MTB)2] (10) shown with
30% probability ellipsoids. Hydrogens have been omitted for clarity and
only one component of the disorder is shown. Selected bond lengths and
À
À
À
angles[Š, 8]: Ga(1) O(1) 1.880(5), Ga(1) O(2) 1.898(7), Ga(1) S(1)
À
À
À
À
2.238(2), Ga(1) S(2) 2.243(2), Li(1) O(1) 1.900(14), O(1) Ga(1) S(1)
À
À
À
À
104.75(15), O(2) Ga(1) S(2) 108.52(19), O(1) Ga(1) O(2) 102.7(2),
S(1) Ga(1) S(2) 106.08(9).
À
À
reactions. It is clear that in solution there exists only a minor
energy difference between [Li(THF)3Ga(MTB)2] (10) and a
dimer of constitution [Ga2(MTB)3][Li2(MTB)] but at present
unknown structure.
To try and access the structural type of the dimer,
[InCl3(THF)3] wastreated with two equivalentsof MTB
dianions(generated from MTBH 2 6 and BuLi) in THF. It was
1
sentative ketone). However, the H NMR spectra in the aryl
region are complex and fluxional and complete assignment of
the spectrum is not possible. The speciation of the catalyst at
the enantio-discriminating step was also investigated by
searching for a Nonlinear Effect (NLE) in the catalytic
asymmetric reaction.[9] No significant NLE was detected. The
data (six points for the reduction of 1a) fit a straight line:
eeobs ˆ 0.93eelig with R ˆ 0.999. The absence of any NLE may
be taken as indicative of three possibilities: either the
asymmetric transition state contains only one MTB ligand
(despite the 2:1 MTB:Ga mixing formulation); or alterna-
tively diastereomeric mixtures of catalyst species that are
formed all react at the same rate with 1a; or the active catalyst
isnot able to form heterochiral psecie,s asisobesrved in
À
hoped that because of longer In Y (Yˆ O, S) bonds, any
potential steric clashes in the dimer would be minimised and
thistsructure favoured. However, after removal of the
precipitated LiCl and layering with pentane, colourless blocks
of [Li2(THF)5InCl(MTB)2] (11) are obtained in moderate
yield. The molecular structure of [Li2(THF)5InCl(MTB)2]
(11), together with selected bond lengths and angles, is shown
in Figure 2. Clearly the chemistry for the indium analogue
KaufmannꢀsB (BINOL)3 species.[10] The present data do not
2
allow differentiation between these possibilities. The working
model of the gallium catalysed reduction is suggested to
involve attack of a gallium hydride 8 on the ketone 1 to yield
an alkoxide 9 (Scheme 3). If thismechanims doesindeed
operate then the enantioselectivities realised should, for a
given ketone (1), be independent of the borane used in the
reaction. This is not the case. Five representative boranes
were compared in their efficiency for the reduction of
acetophenone 1a under standard conditions (Table 2).[6, 11±14]
The absence of any enantioselection in reactions using
BH3 ´ THF and (iPrO)2BH (runs1 ± 2) can be attributed to the
formation of reactive borohydridesLi[BHY 2{(R)-OCH-
(Me)(Ph)}] (2 mol%, Yˆ H, OiPr). Such species could be
formed by Li[(R)-OCH(Me)(Ph)] abstraction from 9 by
BHY2 and would be efficient catalytic activatorsof the
remaining BHY2, but would be expected to show negligible
enantioselectivity. As judged by literature 11B NMR shifts, the
Figure 2. Molecular structure of [Li2(THF)5InCl(MTB)2] (11) shown with
30% probability ellipsoids. Hydrogens have been omitted for clarity.
À
À
Selected bond lengthsand angles[Š, 8]: In Cl 2.452(3), In O1 2.206(7),
À
À
À
À
À
In O2 2.196(7), In S1 2.480(2), In S2 2.471(3), Li O1 1.92(2), Li2 O2
À
À
À
À
À À
1.91(2), O1 In S1 91.11(17), O2 In S2 91.09(17), O1 In O2 173.9(3),
À
À
À
À
À
À
À À
S1 In S2 124.37(10), Cl In O1 91.64 (18), Cl In O2 82.29(19), Cl In S1
À
À
115.63(11), Cl In S2 119.85(10).
Chem. Eur. J. 2000, 6, No. 19 ꢁ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 2000
0947-6539/00/0619-3589 $ 17.50+.50/0
3589

S. Woodward et al.
FULL PAPER
Table 2. Enantioselective reduction of acetophenone 1 by (R,R)-LiGa-
(MTB)2 (2 mol%) and variousboranes.
Mass spectra were obtained on FinniganMAT 1020 or Autospec VG
(electron impact ionisation, EI), FinniganQMS (electrospray ionisation,
ESI), VGZAB or AutospecVG (fast atom bombardment, FAB) machines.
Run Borane
Equiv. used d 11
B
Yield [%][a] ee [%][b]
Elemental analyses were performed by using
a Fisons Instruments
À 0.6[11] 95
< 2
< 2
EA1108 CHN elemental analyser. Optical rotations were measured on
an Optical Activity AA10 instrument in units of 10À1 8cm2 gÀ1 (c in
g/100 cm3). Light petroleum refersto that fraction with b.p. 40 ± 60 8C.
1
2
BH3 ´ THF
(iPrO)2BH
4.5
4.5
27.0[6] 81
The ligands 5 ± 7[15±17] and ketones 2j,[18] 2m,[19] and 2p ± q[20, 21] were
O
3
1.5
28.7[12] 41
< 2
B
H
obtained by literature procedures. All boranes, except BH ´ THF and
3
O
catecholborane, were obtained by known techniques(ese Table 3). All
other compoundswere commercial products(Aldrich).
O
O
Procedure for the preparation of LiGaH4: Thiscompound wasprepared by
a modification of Shriver and Shirkꢀsprocedure. [22] A solution of anhydrous
GaCl3 (0.57m in Et2O) wasprepared by slow addition of Et 2O to a chilled
(08C) suspension of GaCl3 so that generation of HCl was minimised. A
sample (10 mL, 5.7 mmol) of this solution was added dropwise to an ice-
cooled suspension of LiH (finely powdered under argon, 0.74 g, 92 mmol)
in Et2O (10 mL) and the resulting mixture was stirred overnight and then
filtered through an oven-dried medium porosity Schlenk frit. The resulting
solution was analysed for active hydrogen by quenching into a 1:1 mix of
THF/1m HCl and measuring the gas evolved. The titre was typically 0.23 ±
0.28m and the solution could be kept in a suitable storage flask at À208C
under argon for at least four months without appreciable decomposition.
4
5
1.5
1.5
29.9[13] 90
28.1[14] 49
90 (R)
B
H
H
H
Me
N
5 (R)
B
O
Me
N
6
1.5
22.3[14]
9
< 2
B
S
[a] Determined by GC. [b] Determined by GC, absolute stereochemistry in
parentheses.
Representative reduction of ketone 1a (cryostat method): A solution of
(R)-(À)-MTBH2 (6, 15 mg, 0.05 mmol) in THF (5 mL) wastreated with
LiGaH4 (0.25m in Et2O, 0.025 mmol) and the resulting colourless solution,
nominally containing [Li(THF)3Ga(MTB)2] (10), wastsirred at room
temperature for 25 minutesand then placed in a pre-equilibrated cryostat
at À258C. Acetophenone 1a (145 mL, 1.25 mmol) wasadded and after five
minutesa solution of catecholborane (1.0 m in THF, 1.5 mmol) wasadded
dropwise over five mins. The mixture was swirled in the cryostatic bath and
allowed to stand overnight at À258C. After quenching with dilute HCl, the
reaction wasdiluted with Et 2O. The aqueousphaes wasextracted with
Et2O. The combined organic phases were washed successively with water,
1m NaOH (Â2), water and then brine, dried over MgSO4 and concentrated
to afford, after filtration through SiO2 (9:1 hexanes/Et2O), 137 mg (89%)
of (R)-(‡)-1-phenylethan-1-ol (90%ee by GC). 1H NMR (270 MHz,
CDCl3): d ˆ 7.37 ± 7.21 (m, 5H; ArH), 4.84 (q, J ˆ 6 Hz, 1H; CH), 1.19 (brs,
1H; OH), 1.46 (d, J ˆ 6 Hz, 3H; Me). Stirring the cooled mixture had no
affect on the chemical yield or enantioselectivity realised. Equivalent
reactionscould be carried out in a domestic freezer. Other ketoneswere
reduced by similar procedures. Chemical yields were determined by
Lewisaciditiesof pinacolborane (run 3), catecholborane
(run 4), and the very closely related oxazaborole (run 5) and
thiazaborole (run 6) are very similar, as are their structures.
These values are not significantly changed for samples run in
THF with the exception of 1,2-C6H4S(NMe)BH, which
resonates at higher frequency (11B NMR: d ˆ 41.0, JBH
ˆ
158 Hz), and catecholborane itself, which shows the presence
of the free borane and itsTHF adduct. Only reactionsusing
catecholborane reduced acetophenone 1a in high yield and
ee. Thispsecific dependence on catecholborane isnot in
accord with the proposed working model for the catalytic
cycle (Scheme 3). One explanation of these observations is
that the role of [Li(THF)3Ga(MTB)2] (10) isto bind the
catecholborane and activate it by Lewisacid/base interactions
akin to those in the CBS ªloadedº catalyst state 3. In thiscase
asthe hydride isdelivered from a coordinated borane, the
enantioselectivity for the reduction of 1 isexpected to be
highly dependent on the nature of the R2BH species used. In
further support of this idea the enantioselective transition
state is strongly affected by even minor solvent changes. For
example, use of (Æ)-2-methyl-THF in runsotherwise identical
to Table 1 (substrate 1 a, THF solvent, 90% ee, R-product)
leadsto the alcohol being isolated in only 43% ee (R). Further
experimentsare required to prove thishypotheissand to
identify the exact reasons for the specific need for catechol-
borane in thischemistry.
1
isolation, GC or H NMR conversion.
Representative reduction of ketone 1a (warm-up method): A stirred
solution of 10 (prepared from LiGaH4 and (R)-MTBH2 6; 0.025m, ca.
0.025 mmol) wasdiluted with THF (1 mL) and the solution wascooled to
À788C. Acetophenone 1a (58 mL, 0.5 mmol) wasadded and after five
minutesa solution of catecholborane (1.0 m in THF, 0.55 mmol) wasadded
dropwise over 5 mins. The mixture was stirred and allowed to warm slowly
overnight, then quenched with dilute HCl and worked up asdecsribed
above to yield (R)-(‡)-1-phenylethan-1-ol (58 mg, 95%, 90%ee by GC).
Other ketoneswere reduced by similar procedures.
(R)-(‡)-1-Phenyl-1-propanol (2b): Reduction of propiophenone 1b
(2.5 mol% 10, À258C, 18 h); yield 96% (93%ee). 1H NMR (270 MHz,
CDCl3): d ˆ 7.37 ± 7.23 (m, 5H; ArH), 4.59 (t, J ˆ 6.5 Hz, 1H; CH), 1.89
(brs, 1H; OH), 1.82 ± 1.72 (m, 2H; CH2), 0.92 (t, J ˆ 7 Hz, 3H; Me).
(R)-(‡)-1-Phenyl-1-pentanol (2c): Prepared by the reduction of valero-
phenone 1c (2.5 mol% 10, À258C, 18 h); yield 80% (92%ee). 1H NMR
(400 MHz, CDCl3): d ˆ 7.45 ± 7.24 (m, 5H; ArH), 4.68 ± 4.64 (m, 1H; CH),
1.85 (d, J ˆ 3.2 Hz, 1H; OH), 1.83 ± 1.70 (m, 2H; CH2), 1.42 ± 1.26 (m, 4H;
CH2CH2), 0.88 (t, J ˆ 7 Hz, 3H; Me).
(R)-(‡)-1-Phenyl-2-methyl-1-propanol (2d): Prepared by the reduction
isobutyrophenone 1d (4 mol% 10, À158C, 60 h); yield 65% (24%ee).
1H NMR (270 MHz, CDCl3): d ˆ 7.37 ± 7.26 (m, 5H; Ar), 4.37 (dd, J ˆ 6.5,
3 Hz, 1H; CHO), 1.96 (octet, J ˆ 6.5 Hz, 1H; CHMe2), 1.85 (d, J ˆ 3 Hz,
1H; OH), 1.00 (d, J ˆ 6.5 Hz, 3H; CHMe2), 0.80 (d, J ˆ 6.5 Hz, 3H;
CHMe2).
Experimental Section
General: Infrared spectra were recorded by using a Perkin-Elmer 983G
infrared spectrophotometer and a Perkin-Elmer 882 infrared spectropho-
tometer. Proton and 13C NMR spectra were recorded on either Jeol
(JNMGX270, JNMLA400) or Bruker (DRX500) spectrometers (tetra-
methylsilane as a standard, J valuesgiven in Hz). The 11B NMR spectra
were acquired on the JNMLA400 or Bruker DRX500 machines. All
spectra were recorded at ambient temperature unless otherwise noted.
(R)-(‡)-1-Phenyl-3-methyl-1-butanol (2e): Prepared by the reduction of
isovalerophenone 1e (4 mol% 5, À158C, 60 h); yield 65% (92%ee).
1H NMR (400 MHz, CDCl3): d ˆ 7.38 ± 2.28 (m, 5H; Ar), 4.79 ± 4.71 (m,
3590
ꢁ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 2000
0947-6539/00/0619-3590 $ 17.50+.50/0
Chem. Eur. J. 2000, 6, No. 19

Enantioselective Reduction of Prochiral Ketones
3586±3594
Table 3. Methodsused for the determination of the enantiopurity of the alcohols 2.
Ketone
Product
Conditionsfor ee assay
Elution order and retention time [min]
1a
1-phenylethanol 2a
GC. Lipodex A:
initial temp 758C
ramp rate 18CminÀ1
GC. Cyclodex B:
1208C isothermal
GC. Cyclodex B:
1208C isothermal
GC. Lipodex A:
initial temp 1008C
ramp rate 18CminÀ1
HPLC. Chiralcel
OD:
0.5 mLminÀ1 20:1 hexane:isopropanol
GC. Lipodex A:
initial temp 1008C
ramp rate 0.38CminÀ1
GC. Cyclodex B:
1408C isothermal
Cyclodex B
S 16:47
R 17:12
R 14:52
S 15:20
1b
1c
1-phenyl-1-propanol 2b
1-phenyl-1-pentanol 2c
R 22:38
S 23:35
S 24:41
R 25:07
1d
1e
1-phenyl-2-methyl-1-butanol 2d
1-phenyl-3-methyl-1-butanol 2e
S 22:44
R 26:00
S 27:35
R 28:37
1 f
1g
1h
1-(4-bromophenyl)-ethanol 2 f
R 37:50
S 38:58
R 20:30
S 21:50
S 13:52
R 14:09
1-(4-methylphenyl)-ethanol 2g
1208C isothermal
GC on epoxide
Lipodex A:
1-phenyl-2-bromo-ethanol 2h (asepoxide)
initial temp 1008C
ramp rate 0.38CminÀ1
HPLC. Chiralcel
OD:
1i
1,2-diphenyl-1-ethanol 2i
1-(2-furyl)hexanol 2j
R 23:18
S 29:54
1.0 mLminÀ1 39:1 hexane:isopropanol
Lipodex A:
1j
S 25:50
R 26:19
initial temp 1008C
ramp rate 0.38CminÀ1
HPLC. Chiralcel
OD:
0.5 mLminÀ1 9:1 hexane:isopropanol
HPLC. Chiralcel
OD:
0.5 mLminÀ1 9:1 hexane:isopropanol
Lipodex A:
1k
1l
1-(1-naphthyl)-ethanol 2k
1-(2-naphthyl)-ethanol 2l
1-phenyl-propyn-1-ol 2m
R 24:36
S 39:12
R 25:06
S 26:36
1m
21:16[a]
21:54[a]
initial temp 1008C
ramp rate 0.38CminÀ1
MTPA ester
1n
1o
4-phenyl-3-buten-2-ol 2n
3-hexyn-2-ol 2o
see text
R 13:18
S 13:28
Lipodex A:
initial temp 508C
ramp rate 0.38CminÀ1
Lipodex A:
initial temp 758C
ramp rate 0.38CminÀ1
Lipodex A:
initial temp 758C
ramp rate 0.38CminÀ1
GC on acetate
g-Cyclodex[b]
508C isothermal
GC on acetate
g-Cyclodex[b]
508C isothermal
GC on acetate
g-Cyclodex[b]
1p
1q
1r
1s
1t
1-heptyn-3-ol 2p
15:14[a]
15:30[a]
1-hepten-3-ol 2q
9:51[a]
10:08[a]
3-methyl-butan-2-ol 2r
4-methyl-pentan-2-ol 2s
1-cyclohexylethanol 2t
R 6:05
S 7:12
R 8:31
S 10:20
R 14:20
S 15:26
initial temp 608C
ramp rate 5.08CminÀ1 to 908C
GC on acetate
g-Cyclodex[b]
1u
3,3-dimethyl-butan-2-ol 2u
R 8:21
S 9:12
508C isothermal
[a] Absolute configuration order not determined. [b] Oktakis-(6-O-methyl-2,3-di-O-pentyl)-g-cyclodextrin (6-me-2,3-pe-g-CD).
Chem. Eur. J. 2000, 6, No. 19
ꢁ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 2000
0947-6539/00/0619-3591 $ 17.50+.50/0
3591

S. Woodward et al.
FULL PAPER
1H; CHOH), 1.81 ± 1.63 (m, 2H; CH2), 1.58 ± 1.43 (m, 1H; CHMe2), 0.91
(2  d, J ˆ 6.5 Hz, 6H; CHMe2).
(R)-(À)-4-Methyl-2-pentan-2-ol (2s): Prepared by the reduction of meth-
ylisobutylketone 1s (2 mol% 10, À208C, 18 h); yield 93% (46%ee).
Determination of ee by GC analysis on acetate. 1H NMR (400 MHz,
CDCl3): d ˆ 3.88 (m, 1H; CHO), 1.75 (m, 1H; CHMe2), 1.55 (brs, 1H;
OH), 1.41 (ddd, J ˆ 13.5, 8, 6 Hz, 1H; CH2), 1.23 (ddd, J ˆ 13.5, 8, 6 Hz, 1H;
CH2), 1.19 (d, J ˆ 6 Hz, 3H; Me), 0.92 (d, J ˆ 7 Hz, 3H; CHMe2), 0.91 (d,
J ˆ 7 Hz, 3H; CHMe2).
(R)-(À)-1-Cyclohexylethanol (2t): Prepared by the reduction of methyl-
cyclohexylketone 1t (2 mol% 10, À208C, 18 h); yield 72% (72%ee).
Determination of ee by GC analysis on acetate. 1H NMR (400 MHz,
CDCl3): d ˆ 3.54 (quintet, J ˆ 6 Hz, 1H; CHO), 1.86 ± 1.65 (m, 5H; cC6H11),
1.39 (brs, 1H; OH), 1.15 (d, J ˆ 6 Hz, 3H; Me), 1.29 ± 0.86 (m, 6H; cC6H11).
(R)-(‡)-1-(4-Bromophenyl)ethanol (2 f): Prepared by the reduction of
4-bromoacetophenone 1 f (2.5 mol% 10, À258C, 18 h); yield 80%
1
(87%ee). H NMR (270 MHz, CDCl3): d ˆ 7.49 ± 7.44 (m, 2H; Ar), 7.27 ±
7.21 (m, 2H; Ar), 4.89 ± 4.82 (m, 1H; CH), 1.90 (brd, J ꢀ 3 Hz, 1H; OH),
1.47 (d, J ˆ 6.2 Hz, 3H; Me).
(R)-(‡)-1-(4-Methylphenyl)ethanol (1g): Prepared by the reduction of
4-methylacetophenone 1g (2.5 mol% 10, À258C, 18 h); yield 95%
(87%ee). 1H NMR (270 MHz, CDCl3): d ˆ 7.29 ± 7.23 (m, 2H; Ar), 7.18 ±
7.82 (m, 2H; Ar), 4.85 (q, J ˆ 6.5 Hz, 1H; CH), 2.34 (s, 3H; ArMe), 1.47 (d,
J ˆ 6.5 Hz, 3H; Me).
(S)-(À)-Styrene epoxide (from 2h): Prepared by reduction of 2-bromo-
acetophenone 1h (2.5 mol% 10, À258C, 16 h). The (S)-2h produced was
cyclised directly during the workup to yield (S)-styrene epoxide (60%,
70%ee). 1H NMR (400 MHz, CDCl3): d ˆ 7.36 ± 7.21 (m, 5H; Ar), 3.83 (dd,
J ˆ 4, 2 Hz, 1H; CH), 3.10 (dd, J ˆ 5.5, 4 Hz, 3H; CH2), 2.76 (dd, J ˆ 5.5,
2.5 Hz, 1H; CH2).
(R)-(À)-3,3-Dimethyl-butan-2-ol (2u): Prepared by the reduction of
pinacolone 1u (2 mol% 10, À208C, 18 h); yield 76% (79%ee). Determi-
nation of ee by GC analysis on acetate. 1H NMR (400 MHz, CDCl3): d ˆ
3.48 (q, J ˆ 6 Hz, 1H; CHO), 1.26 (brs, 1H; OH), 1.13 (d, J ˆ 6 Hz, 3H;
Me), 0.90 (s, 9H; tBu).
Chromatographic conditions for separation of the enantiomers: All
compounds, with the exception of 2i ± h, 2k ± l, 2n and 2r± 2u, were
baseline separated by using GC with Lipodex-A (25 m, Machery-Nagel) or
Cyclodex-B (J&W Scientific). The analyses were run on a Varian 3380
machine with a head pressure of approximately 12 psi (1508C injector port
temp. 2508C detector temp). Detailsof the temperature programming and
enantiomer elution order are given in Table 3. The configuration of the
product alcohols was confirmed by comparison with authentic samples
where possible. For 2d, 2i and 2k ± l, the ee measurement was carried out
(R)-(À)-2-Phenyl-1-phenylethanol (2i): Prepared by the reduction of
deoxybenzoin 1i (2 mol% 10, À208C, 19 h); yield 88% (68%ee).
1H NMR (400 MHz, CDCl3): d ˆ 7.18 ± 7.35 (m, 10H; Ar), 4.90 (m, 1H;
CH), 3.04 (dd, J ˆ 14, 5 Hz, 1H; CH2), 2.99 (dd, J ˆ 14, 8 Hz, 1H; CH2), 1.93
(d, J ˆ 3 Hz, 1H; OH).
(R)-(‡)-1-(2-Furyl)hexanol (2j): Prepared by the reduction of 2-hexanoyl-
furan 1j (2.5 mol% 10, À258C, 18 h); yield 76% (81%ee). 1H NMR
(400 MHz, CDCl3): d ˆ 7.38 ± 7.36 (m, 1H; H-5), 6.43 ± 6.31 (m, 1H; H-4),
6.24 ± 6.22 (m, 1H; H-3), 4.68 (t, J ˆ 8 Hz, 1H; CH), 1.9 ± 1.8 (m, 2H; CH2)
1.45 ± 1.3 (m, 6H; (CH2)3), 0.83 (t, J ˆ 12 Hz, 3H; Me).
(R)-(‡)-1-Naphthylethanol (2k): Prepared by reduction of 1-acetonaph-
thone 1k (2 mol% 10, À208C, 20 h); yield 82% (59%ee). 1H NMR
(400 MHz, CDCl3): d ˆ 8.08 ± 8.05 (m, 1H; Ar), 7.86 ± 7.83 (m, 1H; Ar), 7.74
(d, J ˆ 8 Hz, 1H; Ar), 7.63 (d, J ˆ 7 Hz, 1H; Ar), 7.51 ± 7.42 (m, 3H; Ar),
5.60 (q, J ˆ 6 Hz, 1H; CH), 2.18 (brs, 1H; OH), 1.62 (d, J ˆ 6 Hz, 3H; Me).
(R)-(‡)-2-Naphthylethanol (2l): Prepared by reduction of 2-acetonaph-
thone 1l (2 mol% 10, À208C, 20 h); yield 83% (73%ee). 1H NMR
(400 MHz, CDCl3): d ˆ 7.84 ± 7.80 (m, 4H; Ar), 7.51 ± 7.43 (m, 3H; Ar), 5.06
(q, J ˆ 6.5 Hz, 1H; CH), 1.94 (brs, 1H; OH), 1.58 (d, J ˆ 6.5 Hz, 3H; Me).
by HPLC by using
a Hewlett Packard Series 1100 machine under
conditions described in Table 3. The enantioselectivity of 2h wasdeter-
mined by derivatisation to styrene epoxide and subsequent analysis on
Lipodex-A. For 2n, the assay was carried out by formation of the
(trifluoromethyl)-phenylacetic acid (MTPA) esters. For 2r± 2u, the ee
values were determined by GC analysis of the derived acetates (see below)
on a 25 m oktakis-(6-O-methyl-2,3-di-O-pentyl)-g-cyclodextrin (6-me-2,3-
pe-g-CD; g-Cyclodex) column by using the general procedure given above
and the conditionsin Table 3.
Preparation of MTPA ester of 4-phenyl-3-buten-2-ol (2n): A mixture of
4-phenyl-3-buten-2-ol (74 mg, 0.5 mmol), dicyclohexylcarbodiimide
(113 mg, 0.55 mmol), (S)-(À)-a-methoxy-a-trifluoromethylphenylacetic
acid (129 mg, 0.55 mmol) and dimethylaminopyridine (7 mg, 0.05 mmol)
in CH2Cl2 (about 5 mL) wasstirred at room temperature for 24 h and then
filtered. The solution was concentrated, taken up in 1:1 light petroleum/
Et2O and then filtered through SiO2. After removal of the solvents, the
1-Phenyl-prop-2-yn-1-ol (2m): Prepared by the reduction of 1-phenyl-
propynone 1m (4 mol% 10, À158C, 60 h); yield 50% (<2%ee). 1H NMR
(270 MHz, CDCl3): d ˆ 7.57 ± 7.35 (m 5H; Ar), 5.47 (s, 1H; CHO), 2.68 (s,
1H; alkyne CH), 2.29 (brs, 1H; OH).
1
residue was examined directly by H NMR spectroscopy.
(R)-(‡)-4-Phenyl-3-buten-2-ol (2n): Prepared by the reduction of benzyl-
idenacetone 1n (2.5 mol% 5, À258C, 18 h); yield 82% (75%ee). 1H NMR
(270 MHz, CDCl3): d ˆ 7.39 ± 7.36 (m, 2H; Ar), 7.33 ± 7.29 (m, 2H; Ar),
Preparation of acetates of aliphatic alcohols (2r± 2u): Pyridine (2.5 mL)
and acetic anhydride (0.15 mL, 1.6 mmol) were added to the crude
alcohol 2 and the mixture wasstirred (16 h). The reaction wasquenched
with 2m hydrochloric acid and extracted with diethyl ether. The organic
phase was washed successively with 2m HCl, 2m NaOH, water and brine.
The solution was dried (MgSO4) and analysed directly by GC (Table 3).
ˆ
7.25 ± 7.21 (m, 1H; Ar), 6.56 (d, J ˆ 16 Hz, 1H; PhCH ), 6.26 (dd, J ˆ 16,
ˆ
6.4 Hz, 1H; CH), 4.52 ± 4.45 (m, 1H; CH), 1.64 (brs, OH), 1.37 (d, J ˆ
6.4 Hz, 1H; Me).
(R)-(‡)-3-Hexyn-2-ol (2o): Prepared by the reduction of 3-hexyn-2-one 1o
(2.5 mol% 5, À258C, 18 h); yield 52% (63%ee). 1H NMR (270 MHz,
CDCl3): d ˆ 4.47 ± 4.55 (m, 1H; CH), 2.27 ± 2.18 (m, 2H; CH3CH2), 1.42 (d,
J ˆ 7 Hz, 3H; Me), 1.13 (t, J ˆ 8 Hz, 3H; MeCH2).
Oct-1-yn-3-ol (2p): Prepared by the reduction of oct-1-yne-3-one 1p
(2.5 mol% 10, À258C, 16 h); yield 88% (22% ee). 1H NMR (270 MHz,
CDCl3): d ˆ 4.29 (dt, J ˆ 7, 2 Hz, 1H; CHO), 2.39 (d, J ˆ 2 Hz, 1H; alkyne
CH), 1.89 (brs, 1H; OH), 1.69 ± 1.18 (m, 8H; (CH2)4), 0.83 (t, J ˆ 7 Hz, 3H;
Me). Absolute stereochemistry not determined.
Investigation of potential nonlinear effects: Portionsof scalemic MTB were
made up asfollows: 20% ee from 3 mg (À) and 12 mg (Æ), 33%ee from
5 mg (À) and 10 mg (Æ), 50%ee from 7.5 mg (À) and 7.5 mg (Æ) and
66%ee from 10 mg (À) and 5 mg (Æ). These were each dissolved in THF
(5 mL) and treated with LiGaH4 (0.25m in Et2O, 0.025 mmol) and the
solution was stirred at room temperature for 25 minutes and then placed in
a pre-equilibrated cryostat at À258C. Acetophenone 1a (116 mL, 1 mmol)
wasadded and after five minutesa oslution of catecholborane (1.0 m in
THF, 1.1 mmol) wasadded dropwise over five minute.s The mixture was
allowed to stand for 18 h at À258C, then quenched with dilute HCl and
worked up asdescribed above to afford the product, which wasexamined
by GC. The results of this study showed (eeSM, eePROD.): (20%, 20%), (33%,
33%), (50%, 47%), (66%, 63%) and (100%, 92%).
Oct-1-en-3-ol (2q): Prepared by the reduction of oct-1-en-3-one 1q
(2.5 mol% 10, À258C, 16 h); yield 50% (12%ee). 1H NMR (270 MHz,
ˆ
ˆ
CDCl3): d ˆ 5.80 (m, 1H; CH ), 5.21 (dt, J ˆ 17, 1.5 Hz, 1H; CH2), 5.10
ˆ
(dt, J ˆ 10, 1.5 Hz, 1H; CH2), 4.10 (m, 1H; CHO), 1.66 ± 1.20 (m, 9H;
(CH2)4 and OH), 0.89 (t, J ˆ 7 Hz, 3H; Me). Absolute stereochemistry not
Catalysis of the reduction of 1a with catecholborane by LiOMe: A solution
of MeOH (8 mL, 0.2 mmol) in THF (2.5 mL) wastreated with n-BuLi (2.5m
in hexanes, 0.08 mL, 0.2 mmol) and the resulting solution cooled to À788C.
Acetophenone (240 mg, 232 mL, 2.0 mmol) wasadded followed by
catecholborane (1.0m in THF, 2.2 mmol). The mixture wasallowed to
warm to room temperature overnight, quenched with 2m HCl and
determined.
(R)-(À)-3-Methyl-2-butanol (2r): Prepared by the reduction of methyl-
isopropylketone 1r (2 mol% 10, À208C, 18 h); yield 81% (69%ee).
Determination of ee by GC analysis on acetate. 1H NMR (400 MHz,
CDCl3): d ˆ 3.56 (quintet, J ˆ 6 Hz, 1H; CHO), 1.68 (brs, 1H; OH), 1.61
(m, 1H; CHMe2) 1.15 (d, J ˆ 6 Hz, 3H; Me), 0.92 (d, J ˆ 7 Hz, 3H;
CHMe2), 0.91 (d, J ˆ 7 Hz, 3H; CHMe2).
pentadecane (100 mL) wasadded. The organic phase waswashed with 1
m
NaOH, after which GC analysis revealed 42% yield of alcohol.
3592
ꢁ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 2000
0947-6539/00/0619-3592 $ 17.50+.50/0
Chem. Eur. J. 2000, 6, No. 19

Enantioselective Reduction of Prochiral Ketones
3586±3594
Spectral studies of LiGaH4/MTBH2 (6)/catecholborane/(1u) mixtures:
Solution 1H NMR spectra were recorded by using a sample prepared from
(S)-(‡)MTB (30 mg, 0.1 mmol) and LiGaH4 (0.25m in Et2O, 0.05 mmol) in
THF (2.5 mL). Proton NMR spectra were recorded at 400 MHz relative to
days. Representative crystals of (Ra,Ra)-10 and (Ra,Ra)-11 were mounted
[24]
using oil. Compound 10 wascollected on Station 9.8
at the Daresbury
Synchrotron Radiation Source by using a BrukerSMARTCCD area
detector diffractometer and silicon monochromated radiation (l ˆ
1
2
external [D6]acetone. The D and D spectra indicate the presence of two
0.6890 Š). Compound 11 wascollected on a SMART area detector
different MTB ligandspreesnt in the ratio 3:1.
1H NMR (400 MHz,
diffractometer by using graphite monochromated MoK radiation (l ˆ
0.71073 Š). The SAINT[25] software was used to integrate the data sets
and apply the Lorentz and polarisation corrections. Crystal data and details
of the data collection and refinement are given in Table 4. An absorption
and incident beam decay correction were performed by using SADABS for
10.[26] No absorption or decay correction was made for 11.
The structures were solved by direct methods by using SHELXS-97 for
10[27] and SIR92 for 11.[28] The structures were refined on F2 by using full-
matrix least squares (SHELXL-97).[29] For 10, the disorder in two of the
THF ligands was modelled with the aid of 1,2 and 1,3 distance restraints.
For 11, all five THF ligandswere restrained to have local C2 symmetry and
to have similar geometry. For both compounds, all ordered non-hydrogen
atoms were refined with anisotropic displacement parameters. Also, H
atomswere placed geometrically and refined by riding modelswith
Uiso(H) ˆ 1.2Ueq(C).
[D6]acetone): d ˆ 7.44 (d, J ˆ 8.7 Hz, 1H; H-3 major), 7.33 (d, J ˆ 8.7 Hz,
1H; H-4' major), 7.32 ± 7.19 (m, 4 1/3H; H-4, 5, 5' major, H-4, 4', 5, 5'
minor), 6.95 (d, J ˆ 9.3 Hz, 1/3H; ArH minor), 6.81 (d, J ˆ 8.9 Hz, 1H; H-3'
major), 6.73 ± 6.71 (m, 1H; ArH major), 6.71 ± 6.65 (m, 1/3H; ArH minor),
6.63 ± 6.57 (m, 1 1/3H; ArH major, ArH minor), 6.53 ± 6.43 (m, 4H; 3 ArH
major, 3 ArH minor), 6.38 (d, J ˆ 9.3 Hz, 1/3H; ArH minor), 6.25 (d, J ˆ
8.7 Hz, 1/3H; ArH minor), 6.22 (d, J ˆ 9.3 Hz, 1H; ArH major); 13C NMR
(125.7 MHz, THF): d ˆ 158.4, 139.3, 139.0, 135.4, 134.3, 134.0, 131.5, 127.9,
127.5, 127.1 (2C), 126.9 (2C), 126.1, 125.6, 125.2, 124.5, 124.0 (2C), 123.6,
122.6, 120.0. Signalsfrom the quaternary carbonsof the minor MTB ligand
could not be detected because of insufficient intensity. Electrospray MS
data were collected on
a FinniganLCQ instrument in the negative
ionisation mode which gave: m/z 1347 [Ga2(MTB)3][Li(MTB)2] (dimer
M À Li, 1%), 669 [LiGa(MTB)2] (100).
The absolute configurations of 10 and 11 were confirmed from the value of
the Flack parameter (x)[30] at the end of the refinement, for 10 x ˆ 0.06(2)
and for 11 x ˆ 0.00(4). Neutral atom scattering factors and anomalous
dispersion corrections were taken from the International Tables for
Crystallography.[31]
When [Ga2(MTB)3][Li2(MTB)] wastreated with catecholborane, only
broad resonances were observed in the 1H NMR spectrum of the reaction
mixture at room temperature. On cooling, the signals sharpened: 1H NMR
(500 MHz, THF, À508C, aryl region only): d ˆ 8.55 (d, J ˆ 8 Hz, H-4 or 5),
8.17 ± 8.12 (m, 1H; Ar), 8.09 (d, J ˆ 8 Hz, 1H; H-4 or 5), 8.05 ± 7.98 (m, 2H;
Ar), 7.91 (d, J ˆ 8 Hz, 1H; H-4 or 5), 7.56 ± 7.53 (m, 2H; Ar), 7.48 ± 7.42 (m,
2H; Ar), 7.39 ± 7.27 (m, 6H; Ar), 7.24 ± 7.19 (m, 2H; Ar), 7.15 ± 7.12 (m, 1H;
Ar), 7.09 ± 7.02 (m, 2H; Ar), 7.02 ± 6.88 (m, 7H; Ar). On addition of
pinacolone at room temperature followed by cooling to À508C, a new
species appeared: 1H NMR (500 MHz, THF, À508C, aryl region only): d ˆ
8.24 ± 8.15 (m, 1H; Ar), 8.14 ± 8.06 (m,
Crystallographic data (excluding structure factors) for the structures
reported in this paper have been deposited with the Cambridge Crystallo-
graphic Data Centre asuspplementary publicationsno.s CCDC-141195
[compound (Ra,Ra)-10] and CCDC-141196 [compound (Ra,Ra)-11]. Copies
of the data can be obtained free of charge on application to CCDC, 12
1H; Ar), 8.05 ± 8.01 (m, 2H; Ar),
8.01 ± 7.94 (m, 3H; Ar), 7.92 ± 7.86 (m,
1H; Ar), 7.52 ± 7.48 (m, 2H; Ar), 7.45 ±
7.40 (m, 3H; Ar), 7.38 ± 7.26 (m, 8H;
Ar), 7.22 ± 7.14 (m, 4H; Ar), 7.05 ± 7.00
(m, 1H; Ar), 6.98 ± 6.89 (m, 2H; Ar).
Table 4. Crystallographic data for (Ra,Ra)-10 and (Ra,Ra)-11.
Crystal data
(Ra,Ra)-10
(Ra,Ra)-11
formula
Mw
C52H48GaLiO5S2
893.68
C60H64ClInLi2O7S2
1125.4
crystal system
space group
a [Š]
b [Š]
c [Š]
hexagonal
P65
10.1729(12)
10.1729(12)
73.623(8)
90
90
120
6598.3(13)
6
1.349
monoclinic
P21
11.437(4)
17.656(7)
14.280(5)
90
106.518(7)
90
2765(2)
2
1.352
0.603
1168
0.25 Â 0.12 Â 0.05
Preparation of (Ra,Ra)-[Li2(THF)5-
InCl(MTB)2] (11):
A solution of
MTBH2 (302 mg, 1.00 mmol) in
7
THF (1.0 mL) wasdeprotonated with
BuLi (0.8 mL of 2.5m solution in
a [8]
b [8]
hexane, 2.00 mmol).
A solution of
[InCl3(THF)3][23] (4.4 mL of 0.11m sol-
ution in THF, 0.50 mmol) wasadded at
08C. The mixture wasallowed to come
to room temperature while stirring
(16 h) and the suspension formed
filtered under argon to remove LiCl.
The solution was layered with pentane
to afford colourless blocks. 1H NMR
(500 MHz, THF, aryl region only): d ˆ
7.76 (d, J ˆ 8.5 Hz, 1H; Ar), 7.71 (d,
J ˆ 8.5 Hz, Ar), 7.56 (apparentt, J ˆ
8 Hz, 2H; Ar), 7.47 (d, J ˆ 8.5 Hz,
1H; Ar), 7.26 (apparents, 1H; Ar),
7.13 (apparentt, J ˆ 8 Hz, 1H; Ar),
7.03 (d, J ˆ 8 Hz, 1H; H-8), 6.94 (ap-
parentt, 2H; Ar), 6.86 (apparentt,
1H; Ar), 6.79 (d, J ˆ 8 Hz, 1H; H-8);
MS (FAB): m/z: 764 [Li2InCl-
g [8]
V [Š3]
Z
1calcd [gcmÀ3
]
absorption coefficient [mmÀ1
F(000)
]
0.769
2796
0.20 Â 0.18 Â 0.04
crystal size [mm]
data collection
T [K]
qmin ± qmax
scan type
h, k, l ranges
total reflectionscollected
independent reflections7227 (
reflectionswith I > 2s(I)
absorption
correction
refinement
data/restraints/parameters
final R indices
[I > 2s(I)]
final R indices
(all data)
goodness-of-fit on F 2
absolute structure parameter
final (D/s)max
150(2)
2.24 ± 25.00
150(2)
1.86 ± 26.00
w
w
À 12 to 9, À12 to 12, À90 to 69
À 15 to 13, À24 to 18, À18 to 18
24614
17820
Rint ˆ 0.060)
8211 (Rint ˆ 0.107)
6579
3959
Tmin ˆ 0.86
0.96
Tmax ˆ none
‡
(MTB)2] .
7224/269/543
R1 ˆ 0.0799, wR2 ˆ
0.175
8211/453/658
R1 ˆ 0.0646, wR2 ˆ
0.123
X-ray structure determination of (Ra,
Ra)-[Li(THF)3Ga(MTB)2] (10) and
(Ra,Ra)-[Li2(THF)5InCl(MTB)2] (11):
Crystals of 10 ± 11 were grown by
layering solutions in THF (ca. 1 mL
containing 0.01 mmol of complex)
with dry pentane (ca. 3 mL) under
argon. Small, moisture-sensitive, col-
ourless crystals formed over several
R1 ˆ 0.0912, wR2 ˆ
R1 ˆ 0.143, wR2 ˆ
0.179
0.144
1.166
0.06(2)
0.001
0.88/ À 0.76 eŠ
0.837
0.00(4)
0.001
0.98/ À 1.07 eŠ
À3
À3
largest diff. peak and hole
Chem. Eur. J. 2000, 6, No. 19
ꢁ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 2000
0947-6539/00/0619-3593 $ 17.50+.50/0
3593

S. Woodward et al.
FULL PAPER
Union Road, Cambridge CB21EZ (UK) (Fax: (‡44)1223-336-033; e-mail:
[11] H. Nöth, B. Wrackmeyer, Nuclear Magnetic Resonance Spectroscopy
of Boron Compounds, Vol. 14 NMR Basic Principles and Progress
(Eds.: P. Diehl, E. Fluck, R. Krosfeld), Springer, New York, 1978.
[12] C. E. Tucker, J. Davidson, P. Knochel, J. Org. Chem. 1992, 57, 3482 ±
3485.
[13] R. Goetze, H. Nöth, H. Pommerening, D. Sadlak, B. Wrackmeyer,
Chem. Ber. 1981, 114, 1884 ± 1893.
[14] K. K. Knapp, P. C. Keller, J. V. Rund, J. Chem. Soc. Chem. Commun.
1978, 971.
[15] D. W. Cai, D. L. Hughes, T. R. Verhoeven, P. J. Reider, Tetrahedron
Lett. 1995, 36, 7991 ± 7994.
[16] S. M. Azad, S. M. W. Bennett, S. M. Brown, J. Green, E. Sinn, C. M.
Topping, S. Woodward, J. Chem. Soc. Perkin Trans. 1 1997, 687 ± 694.
[17] D. Fabbri, G. Delogu, O. De Lucchi, J. Org. Chem. 1993, 58, 1748 ±
1750.
deposit@ccdc.cam.ac.uk).
Acknowledgements
We thank EPSRC for their support under grants GR/K91982 and GR/
M84909, for access to their Mass Spectrometry Service (Swansea) and for a
studentship (A.C.). We are indebted to the COST-D12 programme for their
support (project number D12/0009/98). We thank Prof. S. Gladiali (Sassari)
for a gift of 1,1'-bi(2-thionaphthol). We thank Drs. David Ewing (Hull) and
Martin Wills (Warwick) for help with NMR spectroscopy and HPLC
advice, respectively. We thank a referee for pointing out the work of
Kaufmann to us.
[18] T. Miyakoshi, H. Togashi, Synthesis 1990, 407 ± 410.
[19] N. G. Kundu, M. Pal, C. Chowdhury, J. Chinmay, J. Chem. Res.
Miniprint 1995, 101 ± 123.
[20] L. E. Overman, K. L. Bell, J. Am. Chem. Soc. 1981, 103, 1851 ± 1853.
[21] M. L. F. Cadman, L. Crombie, S. Freeman, J. Mistry, J. Chem. Soc.
Perkin Trans. 1 1995, 1397 ± 1408.
[22] A. E. Shirk, D. F. Shriver, Inorg. Synth. 1977, 45, 45 ± 47.
[23] R. L. Wells, S. S. Kher, R. A. Baldwin, P. S. White, Polyhedron 1994,
13, 2731 ± 2735.
[24] a) R. J. Cernik, W. Clegg, C. R. A. Catlow, G. Bushnell-Wye, J. V.
Flaherty, G. N. Greaves, I. Burrows, D. J. Taylor, S. J. Teat, M.
Hamichi, J. Synchrotron. Radiat. 1997, 4, 279 ± 286; b) W. Clegg,
M. R. J. Elsegood, S. J. Teat, C. Redshaw, V. C. Gibson, J. Chem. Soc.
Dalton Trans. 1998, 3037 ± 3039.
[1] Review: R. Noyori, S. Hashiguchi, Acc. Chem. Res. 1997, 30, 97 ± 102.
[2] Review: E. J. Corey, C. J. Helal, Angew. Chem. 1998, 110, 2092 ± 2118;
Angew. Chem. Int. Ed. 1998, 37, 1986 ± 2012.
[3] a) Review: M. Nishizawa, R. Noyori in Comprehensive Organic
Synthesis, Vol. 1 (Eds.: B. M. Trost, I. Flemming), Pergamon, Oxford,
1991, pp. 159 ± 182; b) R. Noyori, I. Tomino, Y. Tanimoto, J. Am.
Chem. Soc. 1979, 101, 3129 ± 3131; c) R. Noyori, I. Tomino, Y.
Tanimoto, M. Nishizawa, J. Am. Chem. Soc. 1984, 106, 6709 ± 6716;
d) R. Noyori, I. Tomino, M. Yamada, M. Nishizawa, J. Am. Chem. Soc.
1984, 106, 6717 ± 6725; e) R. Noyori, Asymmetric Catalysis in Organic
Synthesis, Wiley-Interscience, New York, 1994, especially Chapters 5
(pp. 255 ± 297) and 6 (pp. 298 ± 322).
[4] A. Ford, S. Woodward, Angew. Chem. 1999, 111, 347 ± 349; Angew.
Chem. Int. Ed. 1999, 38, 335 ± 336.
[25] SMART and SAINT Area-Detector Control and Integration Software,
BrukerAXS, Madison, 1997.
[5] R. G. Pearson, Coord. Chem. Rev. 1990, 100, 403 ± 425.
[6] C. W. Lindsley, M. DiMare, Tetrahedron Lett. 1994, 35, 5141 ± 5144.
[7] A. Arase, M. Hoshi, T. Yamaki, H. Nishiwasza, J. Chem. Soc. Chem.
Commun. 1994, 855 ± 856.
[26] G. M. Sheldrick, SADABS: Program for Scaling and Correction of
Area Detector Data, University of Göttingen, 1996.
[27] G. M. Sheldrick, SHELXS-97: Program for the Solution of Crystal
Structures, University of Göttingen, 1997.
[8] M. Shibasaki, H. Sasai, T. Arai, Angew. Chem. 1997, 109, 1290 ± 1310;
Angew. Chem. Int. Ed. Engl. 1997, 36, 1237 ± 1256.
[28] A. Altomare, G. Cascarano, C. Giacovazzo, A. Guagliardi, J. Appl.
Crystallogr. 1993, 26, 343 ± 350.
[9] a) Review: H. B. Kagan, C. Girard, D. Guillaneux, D. Rainford, O.
Samuel, S. Y. Zhang, S. H. Zhao, Acta Chem. Scand. 1996, 50, 345 ±
352; b) D. G. Blackmond, J. Am. Chem. Soc. 1997, 119, 12934 ± 12937;
c) D. Guillaneux, S. H. Zhao, O. Samuel, D. Rainford, H. B. Kagan, J.
Am. Chem. Soc. 1994, 116, 9430 ± 9439.
[29] G. M. Sheldrick, SHELXL-97: Program for the Refinement of Crystal
Structures, University of Göttingen, 1997.
[30] H. D. Flack, Acta Crystallogr. Sect. A 1983, 39, 876 ± 881.
[31] International Tables for Crystallography, Vol. C (Ed.: A. J. C. Wilson),
Kluwer Academic, Dordrecht, 1992.
[10] D. Kaufmann, R. Boese, Angew. Chem. 1990, 102, 568 ± 569; Angew.
Chem. Int. Ed. Engl. 1990, 29, 545 ± 546.
Received: March 3, 2000 [F2339]
3594
ꢁ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 2000
0947-6539/00/0619-3594 $ 17.50+.50/0
Chem. Eur. J. 2000, 6, No. 19