J. Ballester et al. / Catalysis Communications 47 (2014) 58–62
61
3
. Results and discussion
3
.1. Bases screening
Oxidation of 1-phenylethanol 1a into acetophenone 2a via
H-transfer was first performed with benzophenone 3a as the
H-acceptor in toluene at 90 °C and with alkali alkoxides and hydroxides
as catalysts (10 or 20 mol%). At 10 mol% (Fig. 1, dark), yields of
acetophenone were low (22 to 40%) whatever the base. The effect of
increasing catalyst loading to 20 mol% (Fig. 1, light) was highly
cation-dependent. Indeed, no or little improvement was observed for
lithium and potassium bases. By contrast, oxidation performances
were greatly improved for sodium bases, the yield of 2a increasing
from 30 to 50% for NaOH and from 26 to 70% for NaOtBu. Such alkali cat-
ion influence and superiority of sodium were previously highlighted
in the reverse MPV reduction [25]. Noteworthy, with these sodium
bases, the equilibrium (Oppenauer oxidation, mirror process of MPV re-
duction, is an equilibrium reaction) is almost reached after 9 h at 20 but
also at 10 mol% (with 20 mol% of NaOH, 2a was obtained in 30% yield
after 2 h and 46% after 9 h while with 20 mol% of NaOtBu, 2a was
formed in 50% yield after 2 h and 68% after 9 h, see Supplementary
data for more values). For Li and K bases, the low yields reported after
Fig. 2. Mechanism proposal.
1
8 h (Fig. 1) also correspond to maximum yields that can be obtained
at the equilibrium. Besides, so far, alkali alkoxides were reported to be
efficient only when used in stoichiometric amounts [11]. The ability of
NaOtBu at 20 mol% to efficiently promote Oppenauer oxidation catalyt-
ically is thus unexpected. Additionally, 2a could be selectively obtained
and side aldol condensation usually encountered in Oppenauer oxida-
tions involving stoichiometric amounts of alkali alkoxides occurred to
very few extent (b2%) [11].
3.4. Substrate scope
Oxidation of various 1-arylethanols was tested with NaOtBu
(20 mol%) and H-acceptor 3a in toluene (Table 3, entries 1–5).
Electron-poor substrates were converted into ketones 2b and 2c in
moderate yields (60 and 54% respectively) and H-acceptor 3e did not
allow to significantly improve the performances (64 and 59%, entries
c
c
2
, 3 ). However, the yield of 2c was increased to 93% by increasing
3
.2. Solvents screening
d
the reaction time to 48 h using a cheaper H-acceptor 3a (entry 3 ).
For electron-rich substrates (entries 4 and 5), 2d and 2e were obtained
in 75 and 72% yields respectively and increasing the reaction time
allowed to get 2e with 85% yield (entry 5). Good performances were
also observed for naphthalene-derived alcohols, with 2f and 2g being
formed in 76 and 80% yields respectively (entries 6 and 7). The catalytic
system could also be applied to non-enolizable benzyl alcohols, with 2h
being obtained in 60% yield and 2i quantitatively prepared (entries 8
and 9). The performances of aliphatic alcohols oxidations depended
on the substrate structures. Aliphatic alcohols involving cyclic substitu-
ents were transformed in fair yields (34 and 45% for 2j and 2k, entries 10
and 11) and neither an increase of the temperature or reaction time,
nor the use of H-acceptor 3e allowed to improve yields. By contrast,
linear 6-undecanol, converted in 50% yield at 90 °C, was quantitatively
oxidized at 110 °C (entry 12).
Oxidation of 1-phenylethanol 1a was next performed with NaOtBu
20 mol%) and benzophenone in various solvents (Table 1). Polar
(
solvents like DMSO or acetonitrile allowed to obtain 2a in 37 and 40%
yields respectively but the side aldol condensation and subsequent
crotonization were favored in these media (entries 1 and 2). The selec-
tivities were significantly improved in less polar solvents such as diox-
ane, dichloromethane and THF, with 2a being selectively obtained
however in moderate yields (43, 47, 50% respectively, entries 3–5).
From this study, toluene appeared as the most suitable solvent for
H-transfer oxidation (70%, entry 6).
3
.3. H-acceptor screening
The nature of the H-acceptor was next modified (Table 2) [32].
Acetone 4 did not efficiently promote oxidation of 1a, neither when
used stoichiometrically, nor as solvent (entries 1 and 2). The ability of
pinacolone 5 to act as an H-acceptor was also explored and 2a was
obtained in 35% yield in that case (entry 3). With 3-nitrobenzaldehyde
3.5. Mechanism proposal
To account for the base-catalyzed Oppenauer-oxidation of alcohols,
we propose a reversible catalytic cycle (Fig. 2) consistent with the
mechanism of the reverse reduction of ketones [25,33]. Therefore, de-
protonation of the alcohol (step a, [34]) would give the corresponding
sodium alcoholate. The latter could next activate acetophenone
(step b) via coordination to the cation to give a cyclic intermediate en-
countered in stoichiometric Meerwein-Ponndorf-Verley reductions
promoted by aluminum alkoxides [30]. The H-transfer and release of
expected ketone (step c) could give sodium benzhydrolate which
could next react with another molecule of substrate in step d (alkoxide
exchange). Steps b to d can be described as coordination-activation
and decoordination between the sodium (Lewis acid) and the ketone
or alkoxide (Lewis base). The global order of reactivity (Na N K N Li) ob-
served for tBuOM (M = Li, Na, K) and MOH probably indicates that Na
represents the right balance in terms of Lewis acidity versus K and Li.
6
as the H-acceptor, 2a was obtained in low yields (20%) and side
aldol condensation between 2a and 6 significantly decreased the selec-
tivity (Table 2, entry 4). Benzophenone still thus appeared as the best
candidate and different commercially available 4,4′-disubstituted
benzophenones (entries 5–8) were next tested. The substituents are
expected to modify the H-acceptor capabilities of the benzophenone
by modulating the electron-density onto the carbonyl function.
As expected, electron-rich H-acceptor 3b or 3c revealed to be less
efficient than 3a, with 2a being obtained in 26 and 55% (entries 5
and 6) yields respectively instead of 70%. Despite the presence of
electron-withdrawing fluoro substituents, 3d did not behave as
a better H-acceptor than 3a (entry 7). Interestingly, 3e involving
electron-withdrawing chloride substituents significantly improved
the performances, with 2a being obtained in 85% yield (entry 8).