Communications
either to SA or HPA, depending on which group undergoes mi-
gration. Running the reaction under acidic conditions favors
methyl group migration by a wide margin to give SA in good
drate. Methyl migration leads to the glyoxylic acid ester, which
hydrolyzes to glyoxylate (GA), which itself breaks down on fur-
ther oxidation to formate and CO . However, in no case does
2
[
6]
yield. Remarkably, we found that the selectivity of the migra-
tion could be controlled by switching the pH of the medium
from acidic to basic. Thus, LA was dissolved in 30% aq. H O
this explain the observation of acetone in the reaction mixture.
Here, we propose an alternative pathway involving dehydra-
tion of 3HLA and rehydration to 2-hydroxylevulinic acid
(2HLA). Retro-aldol of 2HLA yields acetone and again GA. To
test these mechanistic postulates, we submitted an independ-
ent sample of MG to the reaction conditions and in fact, only
formate and methanol were observed in the NMR spectrum.
A sample of GA produced only formate under the same condi-
tions. The observation of methanol may also derive from
Baeyer–Villiger oxidation of acetone and subsequent hydrolysis
of the methyl acetate.
2
2
and potassium hydroxide was added. The reaction was heated
to 1158C and further portions of base and H O were added
2
2
over the course of about 10 min. Unlike the same reaction in
acid, rapid evolution of oxygen was observed at each addition.
The mixture was allowed to stir undisturbed for a period of
time before a final portion of H O and KOH was added. The
2
2
entire process was complete within about 90 min. Quantitative
analysis by NMR using an internal standard showed that HPA
was being produced in 47% yield. The mass balance consisted
of acetic acid (89%), formic acid (29%), and methanol (9%).
A volatiles trap further detected traces of acetone (ca. 1%).
Running the same reaction at a lower temperature (608C) re-
duces the yield of HPA (22%) but increases the yield of ace-
tone (10%).
An interesting by-product was observed in the reaction con-
ducted at 608C, which was not detected in the higher-temper-
ature process. It appeared similar to HPA but showed a greater
downfield shift of the methylene group adjacent to oxygen in
1
its H NMR spectrum. We speculated that this was the corre-
sponding hydroperoxide, that is, 3-(hydroperoxy)propanoic
acid (HPPA), and were able to confirm this proposal both by io-
The range of observed products indicated processes other
than those shown in Scheme 1 are operative in this chemistry.
Specifically, the fact that more acetic acid is produced than
HPA demonstrates that more than one route to this product is
available. The most likely scenario is oxidation of the enolate
dometric titration and by derivatization with MeI and Ag O to
2
give methyl 3-(methylperoxy)propanoate. The yield of HPPA at
608C was low, but further reducing the reaction temperature
appeared to favor this product. Ultimately, it was found that
carrying out the reaction between 08C and room temperature
over the course of about 6 h resulted in the production of
[13]
of LA to give 3-hydroxylevulinic acid (3HLA) (Scheme 2).
Retro-aldol cleavage of 3HLA yields acetate and methylglyoxal
MG). Baeyer–Villiger oxidation of MG is favored at the keto
1
(
HPPA in 82% yield by H NMR integration, which was con-
group, since the aldehyde exists mainly in the form of a hy-
firmed by isolation of the product in 80% yield. The expected,
equivalent yield of acetic acid (80%) was also ob-
served, alongside HPA (5%) and formic acid (4%). To
avoid distillation of water in the isolation of HPPA,
the product was extracted with ether. Conversion of
HPPA to HPA by OÀO bond hydrogenolysis over Pd/C
was facile and quantitative, and hence this approach
to HPA is considered to be the method of choice.
The question arises as to how HPPA occurs in the
oxidation of LA with H O . We propose two reasona-
2
2
ble pathways for the generation of HPPA, one or
both of which may be operative. First, and most
straightforward, it is possible that, instead of hydroly-
sis of the acetate, elimination to an acrylate inter-
mediate may occur as shown in Scheme 3. The
higher nucleophilicity of hydroperoxide anion may
explain the selectivity for HPPA over HPA in this
lower temperature reaction. Alternatively, attack on
the acetate carbonyl by hydroperoxide anion would
give a tetrahedral intermediate which could rear-
range as shown to give HPPA and acetate.
In conclusion, we introduce here a new concept
for the derivation of 3-hydroxypropanoic acid from
biomass, as summarized in Scheme 4. The process is
fully chemical-catalytic, comparatively fast, operates
under mild conditions, and gives HPA in high yield
(
>80% overall from LA). The mass balance of the re-
action consists mainly of acetic acid, which is itself
a useful commodity chemical. Since LA can be de-
Scheme 2. Mechanism for the formation of acetone, formate, methanol, and excess ace-
tate in the oxidation of LA to HPA. RA=retro-aldol.
&
ChemSusChem 0000, 00, 0 – 0
2
ꢀ 0000 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!
Wu, Linglin
