BCl3- and TiCl4-mediated reductions of β-hydroxy ketones

Article abstract of DOI:10.1021/jo951549x

Syn-selective reduction protocols for β-hydroxy ketones are described exploiting the intermediacy of titanium and boron chelates derived from TiCl₄ and BCl₃, respectively. Reductions are conducted at -78°C in CH₂Cl₂

Full text of DOI:10.1021/jo951549x

868
J . Org. Chem. 1996, 61, 868-873
BCl3- a n d TiCl4-Med ia ted Red u ction s of â-Hyd r oxy Keton es
Christopher R. Sarko, Scott E. Collibee, Allison L. Knorr, and Marcello DiMare*
Department of Chemistry, University of California, Santa Barbara, California 93106
Received August 22, 1995X
Syn-selective reduction protocols for â-hydroxy ketones are described exploiting the intermediacy
of titanium and boron chelates derived from TiCl4 and BCl3, respectively. Reductions are conducted
at -78 °C in CH2Cl2 using a wide range of CH2Cl2-soluble reducing agents. Added acid-scavenging
agents are detrimental to reaction selectivity. An A(1,3)-like interaction involving the stereocenter
responsible for asymmetric induction provides conformational biasing of the intermediate chelates
necessary for high diastereoselectivity.
Ta ble 1. TiCl4-Med ia ted Red u ction s of â-Hyd r oxy
In tr od u ction
Keton es 1a a n d 1b
Recently there has been much activity exploiting the
presence of proximal hydroxyl groups to control the
diastereoselectivity and enantioselectivity of organic
reactions.1 In particular, hydroxyl-directed reductions
have been intensively investigated because of the desire
to control syn and anti 1,3-diol relationships found in
polyacetate- and polypropionate-derived natural prod-
ucts.2 Having recently developed a protocol to allow the
diastereoselective reduction of â-alkoxy ketones,3 we
turned our attention to â-hydroxy ketones. In this paper,
the development of TiCl4- and BCl3-mediated protocols
for the syn-selective reduction of some â-hydroxy ketones
is reported. A(1,3)-like interactions appear to dictate
intermediate chelate conformation and hence are respon-
sible for the diastereoselectivity observed in these reduc-
tions.
1a (R ) Ph)
syn:antia yield (%)b syn:antia yield (%)b
1b (R ) t-Bu)
reductant
pyr‚BH3
Et4NNCBH3
Bu4NBH4
>99:1
98:2
98:2
56:44
87:13
86:14
93:7
81
85
79
87
86
84
80
98:2
84
88
86
84
78
82
76
97:3
97:3
BH3‚THF
58:42
89:11
85:15
91:9
morpholine‚BH3
t-BuNH2‚BH3
DIBALH
a
b
Determined by 1H NMR. Isolated yield of diastereomeric
mixture.
Resu lts a n d Discu ssion
TiCl4-Med ia ted Red u ction s. Given previous experi-
ence in these laboratories,3 initial experiments were
conducted using TiCl4 as the chelating agent and sub-
strates where an adjacent stereocenter would dictate the
course of asymmetric induction and the flanking carbonyl
substituent was large or sp2-hybridized (vide infra). As
envisaged, the â-hydroxy ketone substrate would first be
combined with TiCl4 to form an intermediate chelate A,
which would then be treated with the desired reducing
agent (eq 1). This is akin to what could be called a two-
component approach as outlined by Reetz for chelation
control.4
The presumed generation of HCl was worrisome given
the presence of Brønsted acid sensitive substrates and
nucleophiles. Indeed, initial reductions conducted at
0-25 °C gave large amounts of elimination products.
Reducing reaction temperatures to -78 °C, however,
improved selectivity and yield. Table 1 displays the
results of TiCl4-mediated reductions of ketones 1a (R )
Ph) and 1b (R ) t-Bu) using a standard protocol: a cold
(-78 °C) CH2Cl2 solution of ketone was treated with 1.0
equiv of TiCl4 (10 min), 1.0 equiv of reducing agent (15
min), and then dilute aqueous HCl. The results are
notable in that only BH3‚THF failed to provide useful
levels of diastereocontrol consistent with a chelation-
controlled process.5 Further, only Me3N‚BH3 of the
nucleophiles screened led to no observed addition.6
Although many of the reducing agents used in Table 1
are known to generate hydrogen upon exposure to HCl,
no gas evolution was noted during reductions. This
suggested that B not A is the predominate intermediate
chelate in reductions (eq 2), because chelate B is expected
to be a much weaker acid than the HCl generated upon
forming A. The intermediacy of B is reasonable because
the elimination of HCl from B to afford A comes at a
X Abstract published in Advance ACS Abstracts, J anuary 15, 1996.
(1) For an excellent survey of directed chemistry, see: Hoveyda, A.;
Evans, D. A.; Fu, G. C. Chem. Rev. 1993, 93, 1307-1370.
(2) For the most recent work in this area, see the following and
references cited therein: (a) Bonini, C.; Racioppi, R.; Righi, G.; Rossi,
L. Tetrahedron: Asymmetry 1994, 5, 173-176. (b) Kaneko, Y.; Matsuo,
T.; Kiyooka, S. Tetrahedron Lett. 1994, 35, 4107-4110. (c) Mohr, P.
Tetrahedron Lett. 1991, 32, 2219-2222. (d) Bonini, C.; Bianco, A.; Di
Fabio, R.; Mecozzi, S.; Proposito, A.; Righi, G. Gazz. Chim. Ital. 1991,
121, 75-80. (e) Evans, D. A.; Hoveyda, A. H. J . Org. Chem. 1990, 55,
5190-5192. (f) Evans, D. A.; Hoveyda, A. H. J . Am. Chem. Soc. 1990,
112, 6447-6449.
(5) Stereochemical assignments were made using 1H and 13C NMR
spectroscopies, see: (a) House, H. O.; Crumrine, D. S.; Teranishi, A.
Y.; Olmstead, H. D. J . Am. Chem. Soc. 1973, 95, 3310-3324. (b)
Heathcock, C. H.; Pirrung, M. C.; Sohn, J . E. J . Org. Chem. 1979, 44,
4294-4299.
(6) The alkylating agents AlMe3 and ZnEt2 were substituted suc-
cessfully for reducing agents with the syn-to-anti ratio exceeding 98:2
and 74-87% yields.
(3) Sarko, C. R.; Guch, I. C.; DiMare, M. J . Org. Chem. 1994, 59,
705-706.
(4) Reetz, M. T. Acc. Chem. Res. 1993, 26, 462-468.
0022-3263/96/1961-0868$12.00/0 © 1996 American Chemical Society

Reduction of â-Hydroxy Ketones
J . Org. Chem., Vol. 61, No. 3, 1996 869
Ta ble 2. Effect of Am in e Ba ses on TiCl4-Med ia ted
Red u ction s of 1a (R ) P h )
conversion
base
(%)a
syn:antia
yield (%)b
none
Et3N
100
81
76
57
43
98:2
94:6
96:4
96:4
93:7
85
89
87
87
85
2,6-di-t-Bu-pyridine
2,6-lutidine
pyridine
its presence. Pyridine, an excellent ligand, would thus
be available to compete with â-hydroxy ketone for binding
sites on titanium, slowing reduction rates. Poorer ligands
and stronger bases such as Et3N would be expected to
deprotonate B to give C. The attenuated Lewis acidity
of the titanium present in C compared to that in B would
then account for the slightly slower rates of reduction in
the presence of bases such as Et3N.
Extension of the TiCl4-mediated reduction procedure
to other â-hydroxy ketones met with mixed results.
Reduction of ketones 1c (R ) C6H11) or 1d (R ) CH3)
using pyr‚BH3 gave 76:24 (89%) and 63:37 (87%) syn-to-
anti ratios, respectively.14 The reduction protocol also
shows pairing of the matched/mismatched type; namely,
ketone 3 is reduced with high syn selectivity (98:2, 86%)
by Et4NNCBH3, while its diastereomer 4 showed little
selectivity (67:33, 86%). Reduction of ketone 5a , which
lacks a substituent at the 2-position, was similarly
unselective. These results parallel those of Oishi and
Nakata using Zn(BH4)2.15
a
1
b
Determined by H NMR. Combined yield of recovered start-
ing material and product.
price. Strong H-Cl (102 kcal/mol7 ) and Ti-O (115 kcal/
mol8 ) bonds are formed in making A, but at the expense
of comparably strong O-H (109 kcal/mol7) and Ti-Cl
(103 kcal/mol8) bonds. In addition, titanium(IV) goes
from a six-coordinate to a five-coordinate state when it
is known to prefer the former.9 The validity of this
analysis is seen in the crystallographically characterized
adducts [TiCl4(HSC5H9)2]10 and [SnCl4(C6H11OH)2]11 and
in the report of Maier and co-workers that combining
TiCl4 and 4-hydroxy-3,3-dimethyl-2-pentanone gives a
solid without the generation of HCl.12
The effect of amines on TiCl4-mediated Et4NNCBH3
reductions of 1a was examined (Table 2). A standard
procedure was followed: a cold (-78 °C) CH2Cl2 solution
of ketone 1a was treated with 1.0 equiv each of TiCl4 (10
min), amine (5 min), and Et4NNCBH3 (15 min) and then
dilute aqueous HCl. No precipitates were observed, and
interestingly, amines slowed the rate of reduction in a
structurally dependent manner. These observations are
difficult to explain if chelate A was the predominate
intermediate. The HCl present due to the formation of
A should generate amine hydrochloride salts, which are
often insoluble in CH2Cl2 at -78 °C (e.g., pyridinium
chloride). The absence of precipitates, though, does not
rule out salt formation. Amine hydrochloride salts and
TiCl4 are known to give organic soluble salts such as
[amine‚H][TiCl5] or [amine‚H]2[TiCl6].13 Thus, the ab-
sence of precipitates can be explained by the formation
of chelates C (eq 3). The dependence of reduction rate
on amine structure, however, cannot be explained by this.
In chelates C, the protonated amine is serving only as a
counterion, and its structure should be irrelevant.
If chelate B was the predominate chelate, the effect of
amine structure can be explained. The acidity of the
chelated hydroxyl group of B is enhanced by its associa-
tion with titanium, but even a large increase of 10 pKa
units would still leave pyridine only half-protonated in
The diastereoselectivity afforded by reduction of che-
lates B parallels that observed with the related â-alkoxy
ketone chelates, which adopt half-chair conformations (eq
4).3,16 When R is small (e.g., 1c and 1d ), conformer D is
favored. Since D displays no clear facial biasing of the
ketone, low reduction selectivity results. When R is large
or is an sp2-hybridized center (e.g., 1a and 1b), conformer
E is favored because it avoids the destabilizing A(1,3)-like
interaction17 present in D (e.g., the interaction of an ortho
carbon of a phenyl group with the stereogenic methyl
group). The pseudoaxial methyl group of E affords facial
biasing of the ketone carbonyl leading to the syn reduc-
tion product. With the TiCl4 chelate of ketone 5a , no
diastereoselectivity is observed upon reduction because
there is no A(1,3)-like interaction to favor one of the two
expected half-chair conformers. The substituents within
the organic fragment are too far apart, and the chloride
ligands of titanium are too far away from the organic
fragment.
(7) Carey, F. A.; Sundberg, R. J . Advanced Organic Chemistry, 2nd
ed.; Plenum: New York, 1984; Part A, p 11.
(8) Reetz, M. T. Organotitanium Reagents in Organic Synthesis;
Springer-Verlag: Berlin, 1986; p 37.
BCl3-Med ia ted Red u ction s. The failure of TiCl4-
mediated reductions of â-hydroxy ketones where 1,3-
asymmetric induction was desired pointed toward a
boron-based chelating agent. The absence of conforma-
(9) Shambayati, S.; Crowe, W. E.; Schreiber, S. L. Angew. Chem.,
Int. Ed. Engl. 1990, 29, 256-272.
(10) Winter, C. H.; Lewkebandara, T. S.; Proscia, J . W.; Rheingold,
A. L. Inorg. Chem. 1993, 32, 3807-3808.
(14) Similarly, reduction of ketones 1c and 1d with Et4NNCBH3 led
to 69:31 (84%) and 60:40 (81%) syn-to-anti ratios, respectively.
(15) Oishi, T.; Nakata, T. Acc. Chem. Res. 1984, 17, 338-344.
(16) Sarko, C. R.; Haase, C.; DiMare, M.; Ziller, J . W. Submitted
for publication.
(17) (a) J ohnson, F. Chem. Rev. 1968, 68, 375-413. (b) Hoffmann,
R. W. Chem. Rev. 1989, 89, 1841-1860.
(11) Fournet, F.; Theobald, F. Inorg. Chim. Acta 1981, 52, 15-21.
(12) Maier, G.; Seipp, U.; Kalinowski, H.-O.; Henrich, M. Chem. Ber.
1994, 127, 1427-1436.
(13) McAuliffe, C. A.; Barratt, D. S. In Comprehensive Coordination
Chemistry; Wilkinson, G., Ed.; Pergamon: New York, 1987; Vol. 3,
Chapter 31.

870 J . Org. Chem., Vol. 61, No. 3, 1996
Sarko et al.
Ta ble 3. BCl3-Med ia ted Red u ction s of â-Hyd r oxy
Keton es 5b-f
ratioa
yieldb
ketone
R
R′
reductant
syn:anti
(%)
5b
5b
5b
5b
5c
5c
5c
5c
5d
5d
5d
5d
5e
5e
5e
5e
5f
Ph
Ph
Ph
Ph
Ph
Ph
Ph
Ph
Ph
Ph
Ph
Ph
Ph
Ph
Ph
Ph
Me
Me
Me
Me
C6H13
C6H13
C6H13
C6H13
i-Pr
i-Pr
i-Pr
i-Pr
Ph
Ph
Ph
Ph
Me
Me
Me
Me
Ph
Ph
Ph
Ph
Bu4NBH4
Et4NNCBH3
pyr‚BH3
Me3N‚BH3
Bu4NBH4
Et4NNCBH3
pyr‚BH3
Me3N‚BH3
Bu4NBH4
Et4NNCBH3
pyr‚BH3
Me3N‚BH3
Bu4NBH4
Et4NNCBH3
pyr‚BH3
94:6
95:5
92:8
92:8
93:7
94:6
>95:5
93:7
>95:5
95:5
>95:5
94:6
>95:5
>95:5
>95:5
>95:5
90:10
84:16
81:19
81:19
81
80
90
87
79
84
72
90
68
81
71
96
65
75
74
77
78
91
71
75
F igu r e 1. Ab initio minimized structures for conformers of
the chelate derived from BCl3 and 4-hydroxy-2-pentanone: (top
left) half-chair conformer with stereogenic methyl oriented
pseudoequatorial; (top right) same with stereogenic methyl
oriented pseudoaxial; (bottom) boat conformer. Hydrogen
atoms have been omitted for clarity.
useful reducing agents was limited in BCl3-mediated
reductions. 9-BBN and DIBALH failed to give any 1,3-
diols. THF‚BH3, t-BuNH2‚BH3, and morpholine‚BH3
afforded 1,3-diols, but the preference for the syn 1,3-diol
was relatively low.20 As HCl generation is unavoidable
if intermediate chelates are to form, the range of behavior
displayed by reductants may be due to their modification
by this species. Temperature is an important factor in
these reductions. Selectivity and yields diminish when
reductions are performed at higher temperatures. The
use of the coordinating solvent THF led to the complete
loss of reduction diastereoselectivity.
To support the presence of conformationally biasing
interactions in F (eq 5), the chelate derived from BCl3
and 4-hydroxy-2-pentanone was examined by ab initio
computational methods (see Experimental Section). Three
minima were found (Figure 1). Two of the minima are
half-chair conformers that differ in the orientation of the
stereogenic methyl group. The conformer with a pseu-
doequatorial methyl group is favored by 4.18 kcal/mol.
The disfavored conformer is destabilized by a close
contact between the pseudoaxial chloride and methyl
groups, which are separated by 3.48 Å (sum of van der
Waals radii is 3.75 Å). This 4.18 kcal/mol destabilization
is 1.84 kcal/mol greater21 than that predicted for the 1,3-
diaxial interaction between the chloride and methyl
groups of cis-1-chloro-3-methylcyclohexane. Interest-
ingly, a boat conformer was found to be only 3.56 kcal/
mol less stable than the most stable half-chair con-
former.22 The lack of eclipsing H-H interactions pre-
sumably stabilizes it.
Me3N‚BH3
Bu4NBH4
Et4NNCBH3
pyr‚BH3
5f
5f
5f
Me3N‚BH3
Determined by 1H NMR. Isolated yield of diastereomeric
mixture.
a
b
tionally biasing interactions within the organic fragment
in this situation indicated that the substituents of the
chelating agent must fulfill this role. Boron chelates then
are logical candidates because of the short B-O bond (ca.
1.35 Å) as demonstrated most effectively by the proce-
dures of Narasaka18 and Prasad.19 A shortcoming of
these procedures is their reliance on pyrophoric trialkyl-
boranes to give dialkylalkoxyborane intermediate che-
lates. We rationalized that chlorides could replace the
alkyl groups of these intermediate chelates, affording a
more convenient and practical procedure. The applica-
tion of commercially available BCl3 to the syn-selective
reductions of â-hydroxy ketones follows (eq 5). The origin
of diastereocontrol in this and related systems is also
discussed.
Experimental support for the preceding was obtained
Table 3 shows the result of the application of the
following protocol to ketones 5b-f: a cold (-78 °C) CH2-
Cl2 solution of ketone was treated with 1.0 equiv each of
BCl3 (1.0 M, CH2Cl2, 5 min) and then reductant. After
the reduction was complete (2 h), the solvent was
removed. Silica gel and methanol were then added to
the reaction flask, and the resulting mixture was stirred
for at least 6 h. This workup effectively cleaves inter-
mediate boronate esters under mild conditions. In
contrast to the TiCl4-mediated reductions, the range of
by a VT-NMR experiment.23 1H NMR spectra of a
24
mixture of ketone 5c and BBr3 in CD2Cl2 (ca. 0.1 M) at
-80 °C revealed a new species with the systematic
downfield shifts characteristic of a Lewis acid chelate
(20) BCl3-mediated reduction of 5b afforded the following syn-to-
anti ratios: morpholine‚BH3, 88:12; t-BuNH2‚BH3, 80:20; THF‚BH3,
87:13.
(21) Derived from molecular mechanics (PCMODEL Version 4.51;
Serena Software, Bloomington, IN 47402), strain energies (kcal/mol)
for the following isomers of 1-chloro-3-methylcyclohexane: cis, all
equatorial +7.19, all axial +10.63; trans, chloride axial +7.63, methyl
axial +8.96.
(22) Another boat conformer with the stereogenic methyl occupying
a pseudoaxial site could not be found.
(23) Prasad and co-workers failed to detect the chelate intermediate
predicted for their system using 11B NMR; see ref 19b.
(24) The use of BBr3 (bp 90 °C) was to avoid the expense and
volatility of BCl3 (bp 12.5 °C).
(18) (a) Narasaka, K.; Pai, F.-C. Chem. Lett. 1980, 1415-1418. (b)
Narasaka, K.; Pai, F.-C. Tetrahedron 1984, 40, 2233-2238.
(19) (a) Prasad, K.; Chen, K.; Hardtmann, G. E.; Repic, O.; Shapiro,
M. J . Tetrahedron Lett. 1987, 28, 155-158. (b) Prasad, K.; Chen, K.;
Hardtmann, G. E.; Repic, O.; Shapiro, M. J . Chem. Lett. 1987, 1923-
1926. (c) Kathawala, F. G.; Prager, B.; Prasad, K.; Repic, O.; Shapiro,
M. J .; Stabler, R. S.; Widler, L. Helv. Chim. Acta 1986, 69, 803-805.

Reduction of â-Hydroxy Ketones
J . Org. Chem., Vol. 61, No. 3, 1996 871
having formed.25,26 For example, the methylene hydro-
gens of 5c resonate at 3.01 and 3.16 ppm and at 3.31
and 4.09 ppm after treatment with BBr3.27 Similarly, the
aryl hydrogens ortho to the carbonyl group of 5c resonate
at 7.96 ppm and at 8.36 ppm after treatment with BBr3.
If this intermediate chelate adopted the lowest energy
conformer predicted computationally, the dihedral angles
between the hydrogen of the hydroxy-bearing â carbon
and the methylene hydrogens of the R carbon would be
51 and 170°. The observed coupling constants of 2.0 and
10.5 Hz are consistent with these values.
The half-chair conformer indicated above as an inter-
mediate agrees with the previously proposed mechanisms
for this class of chelate-controlled reductions. The lack
of obvious facial biasing observed in this intermediate,
however, does not explain the asymmetric induction
observed. In analogy to cyclic iminium salt and cyclo-
hexanone enolate chemistry,28 product development con-
trol is suggested (eq 6). Attack of hydride by path a gives
the boronate anion of the major syn 1,3-diol in a chair
conformation, while attack of hydride by path b gives the
boronate anion of the minor anti 1,3-diol in a twist
conformation. Although the presence of so many het-
eroatoms makes the analogy to the twist and chair forms
of cyclohexane tenuous, ab initio calculations show that
the chair boronate anion derived from syn-2,4-pen-
tanediol (eq 6; R, R′ ) Me) is 3.20 kcal/mol29 more stable
that the twist boronate anion derived from anti-2,4-
pentanediol (see Experimental Section).
cols developed will be of use to synthetic chemists, and
the interactions apparently responsible for diastereocon-
trol can be used as design elements in new reactions.
Exp er im en ta l Section
1H NMR spectra were acquired at 199.98 MHz in CDCl3 and
referenced to residual CHCl3 unless otherwise noted. 13C NMR
spectra were acquired at 50.29 MHz in CDCl3 and referenced
to CDCl3. Coupling constants are reported in Hz. All reactions
were conducted in oven-dried (130 °C) glassware under a dry
nitrogen atmosphere. Stereochemical assignments were ac-
complished by direct comparison to the literature or by use of
spectroscopic methods.5 Compound 1d , 1.0 M BCl3 in CH2-
Cl2, and TiCl4 were purchased from Aldrich Chemical Com-
pany. Chromatography refers to “flash” chromatography on
SiO2.
Gen er a l TiCl4 Red u ction P r oced u r e. To a cold (-78 °C)
solution of ketone (1.0 mmol) in 10 mL of CH2Cl2 was added
TiCl4 (111 µL, 1.0 mmol) to give immediately a bright yellow
solution, which was stirred 10 min at this temperature. The
nucleophile (1.0 mmol) in 5 mL of CH2Cl2 was then added.
After 15 min, 25 mL of 1 N HCl was added, and the reaction
was warmed to rt. The organic layer was separated, the
aqueous layer was washed with CH2Cl2, and the combined
organics were concentrated in vacuo. The resulting residue
was partitioned between Et2O and H2O. The ethereal layer
was washed with water and brine, dried over anhyd Na2SO4,
and concentrated in vacuo. Chromatography gave diastereo-
meric mixtures of diols in 84-96% yield.
Gen er a l BCl3 Red u ction P r oced u r e. To a cold (-78 °C)
solution of ketone (0.5 mmol) in 5 mL of CH2Cl2 was added
prechilled BCl3 (0.5 mmol, 1 M CH2Cl2). After 5 min, reduc-
tant was added, and the reaction was stirred for 2 h. While
still cold, CH3OH was added (gas evolution evident), and the
reaction vessel was then allowed to warm to rt. After
concentration in vacuo, silica gel (ca. 0.75 g) and CH3OH (ca.
10 mL) were added, and the resulting slurry was stirred
overnight.30 Chromatography gave diastereomeric mixtures
of diols in 70-90% yield.
1-Hyd r oxy-2,4,4-tr im eth yl-3-p en ta n on e (1b). A cold
(-78 °C) solution of LDA generated by the addition of n-BuLi
(12.0 mL, 1.6 M hexanes) to (i-Pr)2NH (2.05 g, 20 mmol) in 20
mL of THF was transferred to ethyl propionate (5.0 g, 18.0
mmol) in 10 mL of cold (-78 °C) THF. After 1 h, pivaldehyde
(1.72 g, 20.0 mmol) was added in one portion. The resulting
solution was stirred for 30 min at -78 °C, warmed slowly to
rt where it remained 90 min, and then treated with 30 mL of
saturated aqueous NH4Cl. The products were extracted into
CH2Cl2, dried over anhyd Na2SO4, concentrated in vacuo, and
chromatographed (15% EtOAc:hexanes) to give 2.51 g (74%)
of ethyl 3-hydroxy-2,4,4-trimethylpentanoate31 as a colorless
oil.
This ester was then dissolved in 100 mL of Et2O and chilled
(0 °C). Solid LAH (0.51 g, 13.5 mmol) was added portionwise
over 15 min, and the reaction was warmed to rt. After 2 h,
0.5 mL of water, 0.5 mL of 10% NaOH, and then 1.5 mL of
water were added sequentially. Water was added to the
reaction, which was then extracted with Et2O. The combined
organic extracts were washed with brine, dried over anhyd
Na2SO4, concentrated in vacuo, and chromatographed (20%
EtOAc:hexanes) to give 1.67 g (84%) of 2,4,4-trimethyl-1,3-
pentanediol (see below).
Con clu sion s
Two diastereoselective reduction procedures have been
developed for â-hydroxy ketones. When TiCl4 is used as
the chelating agent, the â-hydroxy ketone must be R
substituted and the flanking carbonyl substituent large
or sp2-hybridized if highly diastereoselective reduction
is to occur. The resulting 1,3-diols have the newly formed
hydroxyl syn to the R substituent. When BCl3 is used
as the chelating agent, â-hydroxy ketones are reduced
with high diastereoselectivity to give syn 1,3-diols. The
avoidance of an A(1,3)-like interaction appears responsible
for the high diastereoselectivity in both protocols. With
TiCl4 as a chelating agent, a half-chair conformer is
favored that strongly biases approach of the reducing
agent. With BCl3 as a chelating agent, a half-chair
conformer is strongly favored that is subject to product
development controlled reduction. Hopefully, the proto-
(25) See, for example: (a) Keck, G. E.; Castellino, S. J . Am. Chem.
Soc. 1986, 108, 3847-3849. (b) Keck, G. E.; Castellino, S.; Wiley, M.
R. J . Org. Chem. 1986, 51, 5480-5482. (c) Keck, G. E.; Castellino, S.
Tetrahedron Lett. 1987, 28, 281-284.
Oxidation of the preceding diol to 1b was accomplished
using the procedure of Stevens and co-workers.32 To a solution
of the above diol in glacial AcOH (6.6 mL) was added aqueous
20% NaOCl (6.6 mL) in a dropwise manner over a 1-h period.
An ice bath was used to maintain the reaction temperature
between 15 and 25 °C. The mixture was stirred for 1 h after
(26) See Experimental Section for more details.
(27) The 1H NMR spectra of uncomplexed ketone 5c were acquired
in CDCl3 at rt. Our experience is that chemical shifts of such
compounds are not greatly dependent on temperature or the chlori-
nated solvent used.
(28) Deslongchamps, P. Stereoelectronic Effects in Organic Chem-
istry; Pergamon: Oxford, 1983; pp 209-221, 274-284.
(29) This is a large part of the 4.6-6.2 kcal/mol destabilization of
the twist form of cyclohexane relative to the chair form, see: Eliel, E.
L.; Wilen, S. H. Stereochemistry of Organic Compounds; Wiley: New
York, 1994; pp 689-690.
(30) Six hours of stirring is sufficient at this point. If an ester is
present in the substrate, transesterification has been observed with
the longer reaction time.
(31) Meyers, A. I.; Yamamoto, Y. Tetrahedron 1984, 40, 2309-2315.
(32) Stevens, R. V.; Chapman, K. T.; Weller, H. N. J . Org. Chem.
1980, 45, 2030-2032.

872 J . Org. Chem., Vol. 61, No. 3, 1996
Sarko et al.
the addition was complete. Saturated aqueous NaHSO3 was
added until the color of the reaction mixture changed from
pale yellow to white and the mixture gave a negative KI-
starch test. The resulting mixture was poured into an ice-
brine mixture and extracted with Et2O. The combined organic
layers were washed with 10% aqueous NaOH. The combined
aqueous washes were back extracted with Et2O. All the
organic extracts were combined, dried over anhyd Na2SO4,
concentrated in vacuo, and chromatographed (10% EtOAc:
hexanes) to give 1.46 g (90%) of 1b: 1H NMR δ 3.57 (dd, J )
7.54, 10.6, 1H), 3.35 (dd, J ) 4.9, 10.6, 1H), 3.19-3.03 (m, 1H),
2.56 (br s, 1H), 1.09 (s, 9H), 0.98 (d, J ) 7.1, 3H); 13C NMR δ
220.10, 65.16, 44.65, 41.68, 25.78, 14.84; IR (film) 3504 (br),
7.27-7.35 (m, 5H), 4.96 (d, J ) 3.3, 1H), 3.51 (d, J ) 9.0, 1H),
1.95 (m, 1H), 1.74 (m, 1H), 0.99 (d, J ) 6.9, 3H), 0.81 (d, J )
6.9, 3H), 0.76 (d, J ) 6.9, 3H); 13C NMR δ 133.38, 128.72,
128.29, 125.87, 79.19, 76.22, 43.37, 31.21, 19.96, 16.96, 15.97.
Red u ction s of â-Hyd r oxy Keton es 5. â-Hydroxy ketones
5a ,2f 5c,39 5d ,39 5e,40 and 5f41 were prepared by aldol reactions
similar to that reported for 5b (see below) and matched
physical characteristics previously reported for them. The
corresponding BCl3-mediated reduction products syn-6a ,42 syn-
6c,43 syn-6d ,44 and syn-6e44 similarly matched physical char-
acteristics previously reported for them.
3-Hyd r oxy-1-p h en yl-1-n on a n on e (5b). To a cold (-78
°C) solution of i-Pr2NH (4.80 mL, 36.6 mmol) in 150 mL of
THF was added a solution of n-BuLi (19.5 mL, 1.6 M hexanes).
After 30 min, a solution of acetophenone (3.60 mL, 51.5 mmol)
in 15 mL of THF was added followed 60 min later by a solution
of heptanal (4.8 mL, 34.4 mmol) in 15 mL of THF. This was
stirred for 5 min and then treated with 50 mL of saturated
aqueous NH4Cl. After warming to rt, the resultant mixture
was extracted with ether (4 × 100 mL). The combined organic
extracts were washed with 50 mL of brine, dried over anhyd
MgSO4, and chromatographed (20% EtOAc:hexanes) to give
3.20 g (44%) of 5b as a white solid: mp 28-29 °C; 1H NMR δ
0.84 (t, J ) 6.5, 3H), 1.09-1.67 (m, 10H), 2.90-3.16 (m, 2H),
3.32 (s, 1H), 4.17 (m, 1H), 7.36-7.57 (m, 3H), 7.88-7.93 (m,
2H); 13C NMR δ 14.0, 22.3, 25.6, 29.0, 31.8, 36.4, 45.0, 67.6,
127.8, 128.2, 133.2, 136.3, 200.5; IR (neat) 3420, 2925, 2859,
1684, 1605, 1582, 1450, 1212, 1020, 757, 691 cm-1; HRMS-CI
m/ z calcd for C15H23O2 (MH) 235.169805, found 235.170563.
r el-(1S,3S)-1-P h en yl-1,3-n on a n ed iol (syn -6b). Ketone
5b (92 mg, 0.392 mmol) was reduced using the standard BCl3
procedure and Et4NNCBH3. Chromatography (20% EtOAc:
hexanes) afforded syn-6b (74 mg) in 81% yield: mp 67-69 °C;
1H NMR δ 0.83 (t, J ) 6.5, 3H), 1.12-1.81 (m, 10H), 3.28 (s,
2H), 3.37-3.89 (m, 2H), 4.79-4.85 (m, 2H), 7.12-7.37 (m, 5H);
13C NMR δ 14.1, 22.5, 25.3, 29.2, 31.8, 38.0, 45.1, 72.9, 75.3,
125.6, 127.4, 128.3, 144.4; IR (neat) 3290, 2915, 2852, 1464,
1366, 1075, 1020, 1002, 756, 700 cm-1; HRMS (EI) m/ z calcd
for C15H24O2 236.177630 (M), found 236.178375.
2969, 2877, 1700 (vs), 1479, 1457, 1465, 1049, 1070, 1029 cm-1
;
MS-CI m/ z (relative intensity) 145 (MH, 9), 127 (9), 103 (23),
87 (27) 57 (100); HRMS-CI m/z calcd for C8H17O2 (MH)
145.1228, found 145.1229.
3-Hydr oxy-1-cycloh exyl-2-m eth yl-1-pr opan on e (1c). The
same series of steps used for the preparation of 1b was used
to prepare 1c. Substitution of cyclohexanecarboxaldehyde
(2.24 g, 20.0 mmol) for pivaldehyde followed by chromatogra-
phy (15% EtOAc:hexanes) gave 2.62 g (68%) of ethyl 3-hydroxy-
3-cyclohexyl-2-methylpropanoate as a colorless oil.31 Reduction
with LAH (0.63 g, 16.5 mmol) and chromatography (20%
EtOAc:hexanes) gave 1.68 g (80%) of 1-cyclohexyl-2-methyl-
1,3-propanediol (see below). Oxidation and chromatography
(10% EtOAc:hexanes) gave 1.51 g (93%) of 1c: 1H NMR δ 3.70
(dd, J ) 6.8, 10.1, 1H), 3.53 (dd, J ) 4.4, 10.1, 1H), 2.98-2.80
(m, 1H), 2.46 (m, 2H), 1.90-1.55 (m, 4H), 1.40-1.10 (m, 6H),
1.05 (d, J ) 6.5, 3H); 13C NMR δ 218.14, 64.35, 49.75, 46.08,
28.50, 27.90, 25.65, 13.29; IR (film) 3830, 2942, 2864, 1711,
1456, 1380, 1152, 1046, 1003, 896 cm-1; MS-CI m/ z (relative
intensity) 171 (MH, 20), 129 (88), 111 (22), 83 (132); HRMS-
CI m/ z calcd for C10H19O2 (MH) 171.1385, found 171.1382.
r el-(1S,2R)-2-Met h yl-1-p h en yl-1,3-p r op a n ed iol (syn -
2a ).33 General reduction was performed on 1a :33,34 1H NMR δ
7.41-7.23 (m, 5H), 4.92 (d, J ) 5.2, 1H), 3.65-3.21 (m, 3H),
2.63 (br s, 2H), 2.08 (m, 1H), 0.83 (d, J ) 7.7, 3H); 13C NMR δ
142.63, 128.15, 127.83, 127.20, 126.75, 126.15, 76.45, 66.28,
41.30, 10.73.
Ch ela te fr om BBr 3 a n d Keton e 5c. Ketone 5c (20 mg,
0.104 mmol) was placed in an NMR tube equipped with a J .
Young valve, frozen in liquid nitrogen, and then evacuated to
150 mTorr. A solution of BBr3 (33.8 mg, 0.135 mmol) in CD2-
Cl2 (1.0 mL) was then vacuum transferred to the NMR tube,
which was kept in liquid nitrogen until placed in the chilled
NMR probe: 1H NMR (500 MHz, CD2Cl2, -80 °C) 1.02 (d, J )
6.8, 3H), 1.04 (d, J ) 6.8, 3H), 1.92 (m, 1H), 3.31 (dd, J ) 10.5,
20.0, 1H), 4.09 (dd, J ) 2.0, 20.0, 1H), 4.41 (ddd, J ) 2.0, 7.2,
10.5, 1H), 7.72 (app. t, J ) 8.0, 2H), 8.00 (t, J ) 7.5, 1H), 8.36
(d, J ) 8.0, 2H). As the sample was warmed, two sets of new
signals appeared at the expense of the those assigned to the
chelate formed from BBr3 and ketone 5c. The two new
compounds were very similar, present in equal amounts, and
only differentiated because one displayed line broadening at
15 °C.
r el-(2R,3S)-2,4,4-Tr im eth yl-1,3-p en ta n ed iol (syn -2b).35
General reduction was performed on 1b: 1H NMR δ 3.90-
3.71 (m, 2H), 3.36 (m, 1H), 2.24 (br s, 2H), 1.90 (m, 1H), 1.02
(d, J ) 7.3, 3H), 0.92 (s, 9H); 13C NMR δ 84.93, 69.96, 35.83,
35.35, 26.24, 13.02.
r el-(1R,2R)- a n d r el-(1R,2S)-1-Cycloh exyl-2-m eth yl-1,3-
p r op a n ed iol (syn - a n d a n ti-2c).35 General reduction was
performed on 1c: 1H NMR δ 3.87-3.41 (m, 3H), 2.62 (m, 1H),
1.86-1.12 (m, 11H), 0.83 (d, J ) 6.9, 3H), 0.72 (d, J ) 6.3,
3H); 13C NMR δ 81.42, 78.21, 67.66, 65.75, 40.68, 40.56, 36.19,
35.55, 30.07, 29.67, 28.90, 26.45, 25.99, 25.78, 13.92, 8.86.
r el-(2R,3R)-2-Meth yl-1,3-bu ta n ed iol (syn -2d ).36 General
reduction was performed on 1d : 1H NMR δ 3.52-3.74 (m, 3H),
2.95 (br s, 1H), 2.55 (br s, 1H), 1.62 (m, 1H), 1.29 (d, J ) 6.3,
3H), 0.92 (d, J ) 7.5, 3H); 13C NMR δ 76.45, 69.06, 41.76, 22.01,
15.65.
r el-(1S ,2R ,3S )-2,4-Dim e t h yl-1-p h e n yl-1,3-p e n t a n e -
d iol.37 General reduction was performed on 3:38 1H NMR δ
7.27-7.35 (m, 5H), 5.01 (d, J ) 2.7, 1H), 3.51 (dd, J ) 1.9,
9.4, 1H), 1.95 (m, 1H), 1.74 (m, 1H), 1.03 (d, J ) 6.9, 3H), 0.83
(d, J ) 7.0, 3H), 0.80 (d, J ) 7.0, 3H); 13C NMR δ 143.92,
128.63, 127.52, 126.24, 83.17, 79.35, 41.58, 32.21, 20.12, 19.41,
4.65.
Com p u ta tion a l Meth od . All structures were optimized
at the RHF/6-31G* level of theory45 using the Gaussian 92
suite of programs.46 Frequency calculations were performed
(39) Hasegawa, E.; Ishiyama, K.; Horaguchi, T.; Shimizu, T. J . Org.
Chem. 1991, 56, 1631-1635.
(40) Oare, D. A.; Henderson, M. A.; Sanner, M. A., Heathcock, C.
H. J . Org. Chem. 1990, 55, 132-157.
(41) Surov, I.; J und, H. Acta Chem. Scand. Ser. B 1986, 40, 831-
r el-(1S ,2R ,3R )-2,4-Dim e t h yl-1-p h e n yl-1,3-p e n t a n e -
d iol.37 General reduction was performed on 4:38 1H NMR δ
838.
(42) Anwar, S.; Davis, A. P. Tetrahedron 1988, 44, 3761-3770.
(43) Kiyooka, S.; Kurada, H; Yajoi, S. Tetrahedron Lett. 1986, 27,
3009-3012.
(44) Todesco, R.; Bockstaele, D. V.; Gelen, J .; Martens, H.; De
Schryer, F. C. J . Org. Chem. 1983, 48, 4963-4968.
(45) Hariharan, P. C.; Pople, J . A. Chem. Phys. Lett. 1972, 16, 217-
219.
(46) Frisch, M. J .; Trucks, G. W.; Head-Gordon, M.; Gill, P. M. W.;
Wong, M. W.; Foresman, J . B.; J ohnson, B. G.; Schlegel, H. B.; Robb,
M. A.; Replogle, E. S.; Gomperts, R.; Andres, J . L.; Raghavachari, K.;
Binkley, J . S.; Gonzalez, C.; Martin, R. L.; Fox, D. J .; Defrees, D. J .;
Baker, J .; Stewart, J . J . P.; Pople, J . A. Gaussian 92, Revision B;
Gaussian Inc.: Pittsburgh, PA 1992.
(33) Nakata, T.; Tani, Y.; Hatozaki, M.; Oishi, T. Chem. Pharm. Bull.
1984, 32, 1411-1415.
(34) Zimmerman, H. E.; English, J ., J r. J . Am. Chem. Soc. 1954,
76, 2294-2300.
(35) Masamune, S.; Sato, T.; Kim, B. M.; Wollmann, T. A. J . Am.
Chem. Soc. 1986, 108, 8279-8281.
(36) White, J . D.; Theramongkol, P.; Kuroda, C.; Engebrecht, J . R.
J . Org. Chem. 1988, 53, 5909-5921.
(37) Bloch, R.; Gilbert, L.; Girard, C. Tetrahedron Lett. 1988, 29,
1021-1024.
(38) Stille, J . K.; Labadie, S. S. Tetrahedron 1984, 40, 2329-2336.

Reduction of â-Hydroxy Ketones
J . Org. Chem., Vol. 61, No. 3, 1996 873
at the same level. All optimized geometries proved to be
minima, and energies were corrected for zero-point energies
(unscaled). No restrictions were placed on the molecules
during geometry optimizations.
thanks the American Cancer Society for a junior faculty
research award.
Su p p or tin g In for m a tion Ava ila ble: 1H NMR spectra are
available for 1b, 1c, 5b, and syn-6b (4 pages). This material
is contained in libraries on microfiche, immediately follows this
article in the microfilm version of the journal, and can be
ordered from the ACS; see any current masthead page for
ordering information.
Ack n ow led gm en t is made to the University of Cali-
fornia and to the donors of the Petroleum Research
Fund, administered by the American Chemical Society,
for support of this research. C.R.S. acknowledges the
support of a J ohn H. Tokuyama Fellowship, and M.D.
J O951549X