Efficient conversion of pyruvic aldehyde into lactic acid by lewis acid catalyst in water

Article abstract of DOI:10.1246/cl.130319

Homogeneous and heterogeneous catalysts with watertolerant Lewis acids, such as Sc(OTf)3 and Nb₂O₅·nH₂O effectively promote the hydride transfer of pyruvic aldehyde into lactic acid in water at 383 K, whereas Bronsted acid catalysts cannot function as effective catalysts under mild conditions. NMR measurement in the presence of D₂O suggests that the reaction on a Lewis acid catalyst proceeds via the MeerweinPonndorfVerley (MPV) reduction.

Full text of DOI:10.1246/cl.130319

doi:10.1246/cl.130319
Published on the web May 29, 2013
873
Efficient Conversion of Pyruvic Aldehyde into Lactic Acid by Lewis Acid Catalyst in Water
Yusuke Koito,1 Kiyotaka Nakajima,1,2 Masaaki Kitano,1 and Michikazu Hara*1,3
1Materials and Structures Laboratory, Tokyo Institute of Technology,
4259-R3-33 Nagatsuta, Midori-ku, Yokohama, Kanagawa 226-8503
2PRESTO, Japan Science and Technology Agency (JST), 4-1-8 Honcho, Kawaguchi, Saitama 332-0012
3ALCA, Japan Science and Technology Agency (JST), 4-1-8 Honcho, Kawaguchi, Saitama 332-0012
(Received April 9, 2013; CL-130319; E-mail: mhara@msl.titech.ac.jp)
Homogeneous and heterogeneous catalysts with water-
tolerant Lewis acids, such as Sc(OTf)3 and Nb2O5¢nH2O,
effectively promote the hydride transfer of pyruvic aldehyde into
lactic acid in water at 383 K, whereas Brønsted acid catalysts
cannot function as effective catalysts under mild conditions.
NMR measurement in the presence of D2O suggests that the
reaction on a Lewis acid catalyst proceeds via the Meerwein-
Ponndorf-Verley (MPV) reduction.
While Brønsted acids such as HCl and H2SO4 can, in principle,
enhance these reactions,9 they are not effective catalysts for
lactic acid production. However, some Lewis acids, such as
metal chlorides7 and metallosilicates (Sn-¢ zeolite10,11 and Al-,
Ga-, and Sn-containing MCM-4112), have been reported to be
active catalysts for the production of lactic acid and alkyl
lactates from DHA. Brønsted acids can function as more
efficient catalysts for dehydration than Lewis acids can;
however, previous reports on Lewis acids suggest that the
hydride transfer of pyruvic aldehyde into lactic acid is a key step
for efficient lactic acid production and that Lewis acids that are
workable in water are much more effective catalysts for hydride
transfer than Brønsted acids are. In this study, the hydride
transfer of pyruvic aldehyde into lactic acid in water was
investigated using various Brønsted and Lewis acid catalysts.
H2SO4, HCl, Amberlyst-15 (sulfonated polystyrene), and
Nafion resin (perfluorosulfonated ionomers) were examined as
homogeneous and heterogeneous Brønsted acid catalysts. The
conversion of pyruvic aldehyde into lactic acid in the presence
of rare earth metal triflates (Sc(OTf)3, Y(OTf)3, and Yb(OTf)3)
and Nb2O5¢nH2O, which are Lewis acid catalysts that are active
even in water, was also examined. Rare earth metal triflates are
stable and highly active homogeneous Lewis acid catalysts that
function as versatile catalysts for a variety of acid-catalyzed
reactions in water.13 Recently, we reported that NbO4 tetrahedra
present on the surface of Nb2O5¢nH2O act as Lewis acids, even
in the presence of water.14 Nb2O5¢nH2O is an amorphous metal
oxide composed mainly of distorted NbO6 octahedra and NbO4
tetrahedra. The Nb-O bonds in these distorted polyhedrons
are highly polarized, so that a part of the surface OH groups
function as Brønsted acid sites. In addition, some of the NbO4
tetrahedra are found to work as Lewis acid sites in water.14
All chemicals were used as received. Nb2O5¢nH2O (Com-
panhia Brasileira de Metalurgia e Mineração; CBMM) was
examined as a heterogeneous catalyst. The catalytic reaction
involved the heating of a mixture of catalyst (0.1 or 0.02 g) and
2.0 mL of aqueous pyruvic aldehyde solution (0.1 M) in a sealed
Pyrex tube at 383 or 413 K for 1 h. The solution was analyzed
using high-performance liquid chromatography (HPLC; LC-
2000 plus, Jasco) equipped with a refractive index (RI) detector.
The lactic acid yield was estimated from the amount of
introduced pyruvic aldehyde.
Lactic acid has received much attention as a starting
material for the production of industrially important chemicals,
such as biodegradable polymers1 and solvents,2 and for tradi-
tional food, pharmaceutical, and cosmetic chemistry applica-
tions.3 The anaerobic fermentation of dextrose with bacteria, as
shown by lactic acid bacilli, is a conventional industrial method
for the synthesis of lactic acid;1,4 however, this process has some
serious drawbacks, such as a long reaction time of a few days
and careful control of the reaction broth pH by the continuous
addition of Ca(OH)2 is required to maintain bacterial activity.5
An energy-inefficient work-up process for acidification of the
evolved calcium lactate with sulfuric acid is also necessary in
the process, which results in the production of large amounts
of gypsum as a by-product. Therefore, large-scale lactic acid
production requires a more environmentally benign chemical
synthesis route that reduces hazardous work-up processes and
worthless by-products.
Lactic acid can be synthesized from glyceraldehyde (GLA)
or 1,3-dihydroxyacetone (DHA) over an acid catalyst.6 These
molecules are readily produced by the oxidation of glycerol, a
main by-product in biodiesel production. Therefore, an efficient
catalytic system that converts GLA and DHA into lactic acid
without drawbacks in the biochemical process would be a
promising candidate for lactic acid production. The reaction
pathway for lactic acid synthesis from GLA and DHA in the
presence of an acid catalyst is often proposed (Scheme S1).7
Pyruvic aldehyde is first evolved from DHA and GLA by
dehydration. The pyruvic aldehyde is a highly reactive di-
carbonyl compound and is present in water in the original
aldehyde, monohydrate, and dihydrate forms with typical
distributions of trace levels, 57% and 43%, respectively, which
was confirmed by 1H NMR analysis (monohydrate pyruvic
aldehyde: ¤ 2.20 ppm (3H), 5.18 ppm (1H), dihydrate pyruvic
aldehyde: ¤ 1.27 ppm (3H), 4.73 ppm (1H)).8 Monohydrate
pyruvic aldehyde is subsequently converted into lactic acid
through hydride transfer. As the reaction proceeds, most of the
dihydrate pyruvic aldehyde is continuously converted into the
monohydrate form to keep equilibrium distribution, and the
monohydrate pyruvic aldehyde finally turns into lactic acid.
Table 1 shows the catalytic activities of the catalysts for the
reaction. While the reaction proceeds effectively over H2SO4
and HCl at 413 K, these Brønsted acids cannot function as
effective catalysts for the reaction at 383 K. In addition, the
turnover numbers (TONs) of the tested Brønsted acids were
much below 1 even at 413 K, under the present experimental
conditions. In general, most Lewis acids cannot catalyze the
Chem. Lett. 2013, 42, 873-875
© 2013 The Chemical Society of Japan
www.csj.jp/journals/chem-lett/

874
Table 1. Catalytic activity for the conversion of pyruvic aldehyde
into lactic acid in water
Acid site density
Catalyst
weight
/g
Temp.
/K
Conv. Yield
/% /%
¹1
/mmol g
Catalyst
TON
BASa
LASb
Blank
383
®
®
®
15.5 0.0
®
383
413
0.1
0.1
15.7 14.8
63.3 51.4
0.02
0.05
concd H2SO4
19.6
9.6
®
®
383
413
0.1
0.1
15.1 11.5
89.0 73.7
0.02
0.2
Scheme 1. Proposed reaction pathway for Lewis acid-catalyzed
hydride-transfer reaction of pyruvic aldehyde in water.
concd HCl
Amberlyst-15 383
Nafion SAC13 383
0.1
0.1
4.8
0.13
®
®
9.4 5.9
11.1 0.0
0.02
®
The Lewis acid catalysts such as Sc(OTf)3 and Nb2O5¢nH2O
are clearly superior to the Brønsted acid catalysts for the
hydride-transfer reaction of pyruvic aldehyde into lactic acid in
water. Two reaction pathways are possible (Scheme 1) for lactic
acid formation from pyruvic aldehyde by Lewis acids: the
classical ene-diol-type rearrangement and the Meerwein-
Ponndorf-Verley (MPV) reduction. It has been reported that
Lewis acids can effectively catalyze the hydride transfer of
glucose into fructose in water through MPV reduction.16 In the
case of hydride transfer via MPV reduction, Lewis acid catalysts
would be coordinated with the monohydrate form of pyruvic
aldehyde in water to form a five-membered intermediate that is
subsequently converted into lactic acid via a 1,2-hydride-transfer
reaction (Scheme 1). In this study, NMR spectroscopy (Avance
III 400, Bruker, 1H: 400.1 MHz, 13C: 100.6 MHz) was employed
to clarify the reaction mechanism over a Lewis acid catalyst. The
reaction was performed in pure H2O and D2O (>99 vol %) in the
presence of Sc(OTf)3 at 383 K. Figure 1 shows 1H and 13C NMR
383
0.1
0.02
0.02
>99 94.4
86.8 77.2
>99 90.4
0.9
3.8
4.4
Sc(OTf)3
383
413
2.0
383
413
0.1
0.02
72.3 40.2
>99 50.7
0.4
2.7
Y(OTf)3
®
1.9
383
413
0.1
0.02
74.3 47.7
>99 60.3
0.6
3.7
Yb(OTf)3
®
1.6
383
413
0.1
0.02
>99 78.3 52
>99 85.4 285
Nb2O5¢nH2O
0.14c
0.03c
aBrønsted acid site. bLewis acid site. cWater-tolerant acid
site.14
reaction because they do not function well or because they
are decomposed in water. However, Sc(OTf)3, Y(OTf)3, and
Yb(OTf)3, which are water-tolerant homogeneous Lewis acids,13
exhibit high catalytic performance for the reaction; the lactic
acid yield with Sc(OTf)3 reached 94.4% at 383 K. When the
reaction was performed with 0.02 g of catalyst at 413 K, TONs
of these triflates exceed 2, indicating that the tested Lewis acids
can work catalytically. In particular, Sc(OTf)3 and Nb2O5¢nH2O
have a significant advantage over Brønsted acids for the hydride
transfer of pyruvic aldehyde in water. The efficient hydride
transfer of pyruvic aldehyde was also observed on Nb2O5¢nH2O,
which is a heterogeneous acid catalyst with both Brønsted and
Lewis acid sites.14,15 This suggests that the Lewis acid sites on
Nb2O5¢nH2O are quite effective for hydride transfer in water,
despite the low density of Lewis acid sites. In addition, no
significant decrease in activity was observed for three reuse of
Nb2O5¢nH2O (Figure S2), indicating that Nb2O5¢nH2O works
as a reusable catalyst for the reaction. However, mass balance is
not satisfied for Lewis acid catalysts, including Nb2O5¢nH2O.
This can be attributed to the formation of polymerized species
by intermolecular reactions among molecules in the reaction
system. Matrix-assisted laser desorption-ionization-time-of-
flight (MALDI-TOF) mass spectrometric analysis revealed that
the reaction solution for Nb2O5¢nH2O contains numerous
polymerized species with m/e = 300-900. Pyruvic aldehyde
has reactive carbonyl (C=O) groups, and complex intermolec-
ular reactions, such as aldol condensation, by Lewis acid
catalysts would result in the formation of complex polymerized
species. It should be noted that Y(OTf)3 and Yb(OTf)3 are not
superior to Sc(OTf)3 and Nb2O5¢nH2O as shown in Table 1,
meaning that the former are distinct from the latter in reactivity
such as acid strength. The details for Lewis acidity are currently
under investigation.
1
spectra of lactic acid synthesized with Sc(OTf)3. The H NMR
spectra have two signals at around 4.0 and 1.2 ppm that are
assignable to CH and CH3 atoms at the C2 carbon, respectively,
and there is no significant difference between the 1H NMR
spectra of lactic acid synthesized in H2O and D2O
(Figures 1A(a) and 1A(b)). 13C NMR measurement also indi-
cated no significant difference (Figure 1B). If the reaction
proceeded via an ene-diol-type rearrangement in D2O, then a
deuterium atom would be incorporated at the C2 position of the
resultant lactic acid. Therefore, the NMR spectra suggest that the
reaction catalyzed by the Lewis acid does not proceed through
an ene-diol rearrangement but via MPV reduction. Such a
mechanism was also proposed for the formation of methyl
lactate from pyruvic aldehyde over a solid Lewis acid catalyst.17
In this study, the reaction was also examined in the presence
of a Brønsted acid in a similar manner. Although homogeneous
and heterogeneous Brønsted acids cannot function as efficient
catalysts for the reaction at 383 K, as shown in Table 1, an
increase in the reaction temperature improved the lactic acid
yield. For example, the pyruvic aldehyde conversion and lactic
acid yield with conc. HCl for 1 h at 383 K are only 15 and 12%,
respectively (Table 1). However, the reaction with conc. HCl at
413 K for 2 h gave a pyruvic aldehyde conversion of >99% and
a lactic acid yield of 82%, which indicates that Brønsted acids
can also function as effective catalysts for lactic acid production
from pyruvic aldehyde at higher temperatures. Figure 2 shows
1H and 13C NMR spectra of lactic acid synthesized with HCl in
pure H2O and D2O (413 K, 2 h). The signals at 4.0 and 1.2 ppm,
attributable to CH and CH3 atoms at the C2 carbon, respectively,
1
are observed for the H NMR of lactic acid synthesized in H2O
Chem. Lett. 2013, 42, 873-875
© 2013 The Chemical Society of Japan
www.csj.jp/journals/chem-lett/

875
(A)
(A)
(B)
(B)
Figure 2. (A) 1H NMR and (B) 13C NMR spectra of lactic acid
synthesized with conc. HCl in (a) H2O and (b) D2O dissolved in
(A) DMSO-d6 and (B) D2O.
Figure 1. (A) 1H NMR and (B) 13C NMR spectra of lactic acid
synthesized with Sc(OTf)3 in (a) H2O and (b) D2O. For NMR
measurement, the products in H2O and D2O are dissolved in DMSO-d6
and D2O, respectively. The resonances at around 120 ppm are
attributed to the CF3 groups of Sc(OTf)3.
mechanism for the hydride transfer of pyruvic aldehyde over
these Lewis acids is distinct from that over Brønsted acids.
This work was supported by the Core Research for Evolu-
tional Science and Technology (CREST, No. JY230195) pro-
gram of the Japan Science and Technology Agency (JST).
(Figure 2A(a)). For lactic acid synthesized in D2O, the quartet
peak at 4.0 ppm, attributable to CH at the C2 carbon, becomes
a singlet, and the CH3 signal at 1.2 ppm disappears completely
(Figure 2A(b)). The 13C NMR spectrum of lactic acid synthe-
sized in D2O is also different from that synthesized in H2O. The
13C NMR spectra have some signals at around 20, 70, and
180 ppm that are assignable to CH3, CH, and COOH, respec-
tively. The weak signals marked with asterisks in Figure 2B(a)
are attributed to lactic acid anhydride formed by intramolecular
dehydration of the evolved lactic acid. The CH3 signal at 20 ppm
in Figure 2B(b) is much smaller than that in Figure 2B(a). These
results are clearly due to the formation of a CD3 moiety in lactic
acid (Figure 2A, inset), which results in the disappearance of the
1H-1H spin-spin coupling of CH with CH3 and a reduction of
References and Notes
1
R. E. Drumright, P. R. Gruber, D. E. Henton, Adv. Mater. 2000, 12,
1841.
2
3
4
S. Paul, A. R. Das, Tetrahedron Lett. 2012, 53, 2206.
R. Datta, M. Henry, J. Chem. Technol. Biotechnol. 2006, 81, 1119.
E. Haslam, Shikimic Acid: Metabolism and Metabolites, John Wiley &
Sons, New York, 1993.
5
6
A. Corma, S. Iborra, A. Velty, Chem. Rev. 2007, 107, 2411.
R. M. West, M. S. Holm, S. Saravanamurugan, J. Xiong, Z. Beversdorf,
E. Taarning, C. H. Christensen, J. Catal. 2010, 269, 122.
Y. Hayashi, Y. Sasaki, Chem. Commun. 2005, 2716.
I. Nemet, D. Vikić-Topić, L. Varga-Defterdarović, Bioorg. Chem. 2004,
32, 560.
7
8
9
M. J. Antal, Jr., W. S. L. Mok, Carbohydr. Res. 1990, 199, 111.
1
the H-13C spin-spin coupling of CH3. H-D substitution of the
10 M. S. Holm, S. Saravanamurugan, E. Taarning, Science 2010, 328, 602.
11 E. Taarning, S. Saravanamurugan, M. S. Holm, J. Xiong, R. M. West,
C. H. Christensen, ChemSusChem 2009, 2, 625.
12 L. Li, C. Stroobants, K. Lin, P. A. Jacobs, B. F. Sels, P. P. Pescarmona,
Green Chem. 2011, 13, 1175.
13 S. Kobayashi, C. Ogawa, Chem.®Eur. J. 2006, 12, 5954.
14 K. Nakajima, Y. Baba, R. Noma, M. Kitano, J. N. Kondo, S. Hayashi,
M. Hara, J. Am. Chem. Soc. 2011, 133, 4224.
CH3 group would occur rapidly on a Brønsted acid through
keto-enol tautomerization. Deuterium-substituted pyruvic alde-
hyde may be subsequently converted into lactic acid through
methyl anion transfer, although it is generally more effectively
catalyzed by a base catalyst through benzilic acid rearrangement.
The details are currently under investigation.
In conclusion, water-tolerant Lewis acids such as Sc(OTf)3
and Nb2O5¢nH2O can effectively catalyze the hydride transfer
of pyruvic aldehyde in water by MPV reduction under mild
reaction conditions. NMR experiments revealed that the reaction
15 K. Nakajima, T. Fukui, H. Kato, M. Kitano, J. N. Kondo, S. Hayashi, M.
Hara, Chem. Mater. 2010, 22, 3332.
16 Y. Román-Leshkov, M. Moliner, J. A. Labinger, M. E. Davis, Angew.
Chem., Int. Ed. 2010, 49, 8954.
17 Supporting Information is available electronically on the CSJ-Journal
Web site, http://www.csj.jp/journals/chem-lett/index.html.
Chem. Lett. 2013, 42, 873-875
© 2013 The Chemical Society of Japan
www.csj.jp/journals/chem-lett/