Angewandte
Chemie
Attempts to perform the aerobic oxidation of glycerol at
lower catalyst concentrations led to high selectivity for
dihydroxyacetone, but low conversions: using 5 mol% palla-
dium under a balloon of air afforded only 47% conversion
after 24 hours in CD3CN/D2O. Competitive oxidative decom-
position of the catalyst[10a] is a likely cause of the lower
In the presence of three equivalents of benzoquinone in
either [D3]acetonitrile or [D6]dimethyl sulfoxide, the buildup
of trace amounts of lactaldehyde (< 5%) was observed during
the course of the reaction, but disappeared after approx-
imately 80% conversion. The relative concentration of
lactaldehyde appeared to inversely correlate with the
amount of water present in the reaction, as the highest
amount of lactaldehyde (5% of the mass balance) was
observed in dry acetonitrile.
1
conversions and yields: the H NMR spectrum of the final
reaction mixture exhibited resonan-
ces characteristic of the palladium
carboxylate (2) that we had previ-
Whilst NMR studies indicated that lactaldehyde was
formed during the course of the oxidation of 1,2-propanediol,
deuterium-labeling studies suggested that the generation of a
mixture of hydroxyacetone/lactaldehyde and subsequent
isomerization of the aldehyde did not contribute significantly
to the high selectivity for hydroxyacetone. Catalytic oxidation
of [2-D]-1,2-propanediol with 1 in [D6]dimethyl sulfoxide
afforded unlabeled hydroxyacetone (< 1% deuterium-scram-
bling) and oxidation of [1,1-D2]-1,2-propanediol gave
[D2]hydroxyacetone with 96% selectivity and 98% conver-
sion. These experiments suggest that liberation of free
lactaldehyde, followed by palladium- or acid-catalyzed tau-
tomerization was not a major contributor to the high
selectivity for hydroxyacetone. The second-order rate con-
stants for the oxidation of [2-D]-1,2-propanediol and that for
the undeuteriated diol are within experimental error (kH/kD =
1.0(2)), thus implying that b-H elimination was not rate-
limiting.[8a,14] However, an inverse isotope effect of kH/kD =
0.7(2) was evident from the ratio of rate constants for [1,1-
D2]-1,2-propanediol and 1,2-propanediol (Scheme 2).
ously shown to be inactive for alcohol
oxidation.[10a] Thus, high conversions
of glycerol into dihydroxyacetone can
be achieved under aerobic conditions,
but only at relatively high palladium
concentrations.
The oxidation of glycerol and 1,2-
propanediol is faster and more selec-
tive than that of 1,3-diols or primary/secondary alcohols.
Under similar conditions (5 mol% Pd, 3 equiv BQ, DMSO,
238C), oxidation of glycerol was complete within 15 minutes
and oxidation of 1,2-propanediol to hydroxyacetone in
dimethyl sulfoxide was complete within 20 minutes
(Table 1). In contrast, the oxidation of a 1-heptanol/2-
heptanol (1:1) mixture was both slower and nonselective,
requiring 10 hours to reach 78% conversion and affording a
45:55 ratio of the ketone/aldehyde.[8a] Similarly, oxidation of
1,3-butanediol gave a 2:3 mixture of the ketone and aldehyde
products in only 55% conversion after 4 hours.
The high chemoselectivity for the oxidation of the
secondary alcohol of glycerol in the presence of two primary
alcohols is noteworthy. Whilst many stoichiometric oxidants
exhibit a preference for secondary over primary alcohols,[12]
few chemoselective catalytic alcohol oxidations are
known.[8,9b,13]
The lower rates and selectivities observed in the inter- and
intramolecular competition experiments suggest that vicinal
diols have an unusual reactivity with 1. The kinetics of 1,2-
propanediol oxidation with benzoquinone were monitored by
1H NMR spectroscopy in dimethyl sulfoxide. With 1.5–
3.0 equivalents of benzoquinone (relative to the diol), the
disappearance of diol peaks conformed to a mixed-second-
order kinetics analysis [Eq. (1)]:
Scheme 2. Oxidation of deuterium-labeled 1,2-propanediols.
On the basis of previous work,[7c,8a,10,14a] we propose that
isomeric palladium alkoxides are formed by liberation of
acetic acid from the cationic palladium acetate derived from
dimeric 1. b-H elimination from the alkoxides would generate
a palladium hydride that reacts with benzoquinone to
generate a cationic palladium/hydroquinone complex. Reac-
tion of this hydroquinone complex with the diol regenerate
the palladium alkoxide.
To investigate the role of the proton-transfer equilibria on
the rate of reaction, we investigated the kinetics of 1,2-
propanediol oxidation with benzoquinone in the presence of
5, 10, and 20 mol% acetic acid (HOAc, relative to the diol)
and found that the rates were inverse first order in [HOAc]
(k’ = k’’/[HOAc]). This observation is consistent with the
reversible generation of the alkoxide from the reaction of the
cationic palladium acetate (Scheme 3). The first-order
dependence on benzoquinone implies that re-oxidation of
palladium(0) or the Pd-H[15] species is rate-limiting in
ꢀ
ꢁ
ꢀ
ꢁ
½BQꢀ
½PGꢀtt
½BQꢀ
½PGꢀ00
ln
¼ ln
þ kobsð½BQꢀ0 ꢁ ½PGꢀ0Þt
ð1Þ
where [BQ] and [PG] are the concentrations of benzoquinone
and 1,2-propanediol, respectively, and t = time in seconds (see
the Supporting Information). Plots of kobs vs. [Pd] and the
initial rates vs. [BQ] confirmed that the rates were first order
in both [Pd] and [BQ] for [BQ] ꢂ 0.3m, thus yielding a rate
law of [Eq. (2)]:
Rate ¼ k0½Pdꢀ½PGꢀ½BQꢀ
ð2Þ
where kobs = k’[Pd] and k’ = 1.9(3) Mꢁ2 sꢁ1 in [D6]dimethyl
sulfoxide at 238C.
Angew. Chem. Int. Ed. 2010, 49, 9456 –9459
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
9457
Painter, Ron M.
