Yakabe et al.
1917
with a FID through a 2 m × 5 mm diameter glass column
packed with 5% PEG-20M on Chromosorb WAW-DMCS
and interfaced with a Shimadzu Chromatopac C-R6A inte-
grator, with temperature programming. Mass spectra were
determined on a JEOL SX-102A mass spectrometer which
was coupled with a Hewlett Packard GC5890 Series II GC
apparatus via a heated capillary column. IR spectra were re-
corded for thin films (neat) or KBr disks on a JASCO A-100
spectrophotometer. Melting points were determined on a
Yanagimoto MP-S3 melting point apparatus and are uncor-
rected.
pure (GC, NMR, and TLC) benzhydrol 2e in 98% yield
(0.180 g), based on the starting benzophenone (entry 26 in
Table 1): mp 63.2–63.8°C, reported mp 65–67°C (13).
The recovered alumina was washed with portions of deio-
nized water (in total 50 mL) to remove inorganic residues
and then oven-dried (500°C, 3 h). After treating with deio-
nized water as above, it was able to be used for repeat runs
(Table 2).
The reductions of the other ketones and aldehydes with
the NaBH – moist alumina system were performed under
4
conditions determined in terms of their reactivities and yields
of the corresponding alcohols 2, followed by the usual work-
up as described above, affording 2. Alcohols thus obtained
were fully characterized by spectroscopic comparisons (IR,
NMR, and MS) with the commercial authentic samples and
were usually of satisfactory purity for a further use, for ex-
ample, as substrates for the selective oxidation with chemi-
cal manganese dioxide in hexane (14).
It is noteworthy that although certain carbonyl compounds
and alcohols have only limited solubilities in hexane under
the conditions employed, the reductions and the work-ups of
the products were achieved without any difficulty. In addi-
tion, we also encountered no difficulty upon a large-scale re-
duction of 1e (40 mmol), yielding 2e quantitatively (entry 27
in Table 1). The recovery of acetophenone 1a and the yield
of 1-phenylethanol 2a in Tables 2 and 3 were determined by
GC using biphenyl as an internal standard.
Reagents
Hexane was dried (CaCl ), distilled, and stored over mo-
2
lecular sieves. Commercial sodium borohydride (Yoneyama
Chemical, Japan) was ground to a fine powder in a dry box
and was stored in a desiccator. Ketones and aldehydes to be
reduced were available from commercial sources and were
used without further purification, the purities of which were
checked by GC prior to use. The corresponding alcohols as
reference compounds were commercially available. Addition
of deionized water (0.02, 0.04, 0.11, 0.2, and 0.3 g) in por-
tions to chromatographic neutral alumina (ICN Biomedical,
alumina N, Super I; 1.0 g), which had been predried in an
oven at 500°C for 6 h (dry alumina), followed by vigourous
shaking of mixtures after every addition for a few minutes
until free-flowing powder was obtained, afforded moist
aluminas with 2, 4, 10, 17, and 23 wt.% loading of water, re-
spectively, which were immediately employed for the reduc-
tion. Acidic and basic aluminas (ICN Biomedical, alumina A
and B, respectively, Super I), silica gel (Fuji–Silysia BW-
Sodium borohydride has enjoyed a widespread use as a
selective and mild hydride donor for a broad range of func-
3
00), florisil (Kishida Chemical), Japanese acid clay
(
Wako), bentonite (Wako), Montmorillonite K10 (Aldrich),
tional groups (10). Established procedures of the NaBH re-
4
kaolin (Wako), and zeolites A-3 and F-9 (Wako), and cal-
cium oxide (Kokusan Chemical Works) were predried and
were then treated with deionized water as described above.
Commercial Hammett indicator reagents (3a) were used as
received.
duction have been carried out in protic solvents so as to
accomplish reasonably rapid reactions, and therefore an
aqueous treatment during work-ups of products has been
usually required. However, in order to simplify the method,
to increase reagent applicability, and to eliminate aqueous
wastes, our synthetic approach involves only introduction of
an inorganic support material into a slurry of finely pulver-
Reduction procedure
The reduction procedure for benzophenone 1e was repre-
sentative. A 30 mL, round-bottom, two-necked flask, equipped
with a 1.5 cm long Teflon-coated stirrer bar, a 25 cm long
condenser, and a glass gas-inlet tubing connected to an
argon-filled balloon was arranged in order to perform the re-
action under a dry atmosphere by linking the top of the con-
denser to a liquid paraffin trap via a flexible silicone rubber
tubing. The flask was charged with 1e (1 mmol, 0.182 g),
ized NaBH and a carbonyl compound in an aprotic solvent,
4
followed by the efficient stirring of the resultant heteroge-
neous mixture. The reduction milieu was of close resemble
to supported reagent systems (2, 3) and therefore retained
the practical advantage of the latter as regards facile product
isolation requiring only filtration followed by evaporation of
the solvent.
It is frequently observed that efficiency of supported re-
agents depends strongly on the choice of support materials
and the presence of water (3a). In fact, we have experienced
in the previous studies that these are decisive factors to de-
termine the facility of the reactions, to control selectivity
and yield of products (5–9). Thus, effects of supports on the
reactivity of a carbonyl compound and the selectivity and
the yield of the alcohol were investigated by using
acetophenone 1a as the test substrate (Table 3). The results
showed an important role of a support whose absence gave
no indication of the reaction in hexane (entry 1). In addition,
hexane (10 mL), finely ground NaBH (0.4 mmol, 0.151 g),
4
and freshly prepared moist alumina (0.5 g; H O content, 10
2
wt.%) and was then deaerated by passing a gentle stream of
dry argon throughout the system. The flask was immersed in
an oil bath and was kept at 60°C while efficient stirring of
the resultant heterogeneous mixture was continued in order
to ensure smooth reduction and to attain reproducible re-
sults. After 3 h, the reaction mixture was cooled rapidly to
room temperature by external cooling and was then filtered
through a sintered glass funnel. The filter cake was washed
thoroughly with portions of dry ether (ca. 50 mL). The com-
bined clear solvent was evaporated on a rotary evaporator
under a reduced pressure to dryness to give satisfactorily
no appreciable reaction occurred when 1a and NaBH were
4
stirred with commercial or dry alumina (entries 2 and 3,
respectively). There have been numbers of reports that a
©
1998 NRC Canada