Highly efficient catalyst for the decarbonylation of lactic acid to acetaldehyde

Article abstract of DOI:10.1039/c0gc00203h

The gas phase decarbonylation of lactic acid was performed over various silica-supported heteropolyacids. The obtained performances were, by far, higher than those previously described in the literature. In particular, the best results were obtained for silicotungstic acid-based catalysts, which showed very high yields of acetaldehyde (81-83%) at high lactic acid conversion (up to 91%). The Royal Society of Chemistry 2010.

Full text of DOI:10.1039/c0gc00203h

View Article Online / Journal Homepage / Table of Contents for this issue
COMMUNICATION
www.rsc.org/greenchem | Green Chemistry
Highly efficient catalyst for the decarbonylation of lactic acid to
acetaldehyde†
Benjamin Katryniok,a,b Se´bastien Paula,b,c and Franck Dumeignil*a,b
Received 11th June 2010, Accepted 27th August 2010
DOI: 10.1039/c0gc00203h
The gas phase decarbonylation of lactic acid was per-
formed over various silica-supported heteropolyacids. The
obtained performances were, by far, higher than those
previously described in the literature. In particular, the best
results were obtained for silicotungstic acid-based catalysts,
which showed very high yields of acetaldehyde (81–83%) at
high lactic acid conversion (up to 91%).
Thus, efficiently realizing the decarbonylation reaction of lactic
acid would open up perspectives for the sustainable production
of acetaldehyde.10 This compound is widely used as a precursor
in the chemical industry with, for example, applications in
condensation, addition and polymerization reactions.†
C–C bond cleavage in lactic acid molecules can follow one
of two different pathways: decarboxylation or decarbonylation
(Scheme 1), both of which require acid catalysts.11 Whereas
the decarbonylation reaction leads to the formation of carbon
monoxide as a co-product together with water, the decarboxyla-
tion reaction releases hydrogen and CO2 as a by-product. In the
latter case, the production of hydrogen can also cause problems
for keeping a high selectivity to acetaldehyde, as it is known that
hydrogen can reduce lactic acid to propanoic acid in the presence
of redox catalysts.8
The use of renewable feedstocks as raw materials for chemical
processes has become more and more important due to the
progressive depletion in fossil resources like oil, natural gas
and coal. For some industrial applications, biomass has already
been proven to be a competitive substitute compared to fossil
reactants.1,2
Unlike petroleum feedstocks, which generally consist of
hydrocarbons, the compounds in biomass-derived resources
usually contain a large number of reactive oxygenated func-
tional groups, leading to difficulties in realizing their selective
transformation. Therefore, one of the challenges in the emerging
renewable chemical industry is the development of processes
that enable a decrease in the oxygen content of these molecules,
thereby directing the conversion selectively to the desired final
product.3
Lactic acid, which contains a hydroxyl and a carboxy group, is
an example of a highly functionalized biomass-derived molecule.
It can be obtained from biomass at low cost by bacterial
fermentation or chemical reactions.4 Due to its potential for pro-
ducing biodegradable polymers (polylactic acid), the production
of lactic acid has recently significantly increased. In addition,
lactic acid has the potential to become a central feedstock for
the chemical industry, e.g. for the synthesis of pyruvic acid.5
Herein, we propose a new strategy for selectively transforming
this bioresource to acetaldehyde by a catalytic process of
decarbonylation. Usually, the decarbonylation of lactic acid
is observed as a side reaction in the dehydration of lactic
acid to acrylic acid. Therefore, the previously published yields
for acetaldehyde using zeolites, niobium oxide or alumina as
catalysts do not exceed 31%.6–9 However, the current production
of acetaldehyde is mainly based on the Wacker process, which
uses fossil feedstock-derived ethylene as a starting material.
Scheme 1 Some valorisation routes for lactic acid.
In regard to these demands for acidic properties, Keggin-type
heteropolyacids (HPA) are excellent candidates for catalysing
this reaction. HPAs are actually known as very strong Brønsted
acids, at least stronger than common inorganic acids such
as HCl, HNO3, H2SO4, etc. and sometimes even classified as
superacids.12,13 The acid strength of a Keggin-type HPA strongly
depends on the nature of its addenda atoms. Hence, HPAs
containing tungsten are more acidic than those containing
molybdenum, while these latter examples exhibit a more pro-
nounced redox character. The strong acidity of HPAs gives
them the ability to efficiently catalyze dehydration, alkylation,
esterification or isomerization reactions. HPAs can also be
strong redox catalysts.14,15 This property can be tuned by the
composition of the Keggin heteropolyanion. Hence, the redox
power of HPAs increases in the series of addenda atoms in
the order W, Mo, V, and in the order Si, P for the central
aUniv. Lille Nord de France, F-59000, Lille, France.
E-mail: franck.dumeignil@univ-lille1.fr; Fax: +33 (0)3.20.43.65.61;
Tel: +33 (0)3.20.43.45.38
bCNRS UMR8181, Unite´ de Catalyse et Chimie du Solide, UCCS,
F-59655, Villeneuve d’Ascq, France
cECLille, F-59655, Villeneuve d’Ascq, France
† Electronic Supplementary Information (ESI) available: Experimental
and analytical details. See DOI: 10.1039/c0gc00203h
1910 | Green Chem., 2010, 12, 1910–1913
This journal is
The Royal Society of Chemistry 2010
©

View Article Online
heteroatom. The highest redox potential is observed for vanado-
phosphomolybdic acids.12
These results can be discussed in regard to the pathways
presented in Scheme 1. The analyses of the gaseous products
showed that the decarboxylation is more important over the
molybdenum-based HPAs, whereas the tungsten-containing
HPA show nearly no formation of CO2.† Phosphomolybdic acid
and vanadium-substituted phosphomolybdic acid are further-
more known for their increased redox character. This explains
the increasedselectivityfor propanoic acidinthis case, bya redox
process over the catalysts, using the hydrogen co-produced with
CO2.
In this work, we study the selective decarbonylation of lactic
acid to acetaldehyde over various silica-supported HPAs. A
R
commercially available silica (CARiACTꢀ Q-15, Fuji Silysia)
and an SBA-15—prepared by the method of Roggenbuck
et al.16†—were chosen as silica host supports, of which the good
hydrothermal stability was checked and validated prior to use
(Table S1, Fig. S1†). These silica host supports were impregnated
with 20 wt% of commercially available silicotungstic acid,
phosphotungstic acid or phosphomolybdic acid. Additionally,
a catalyst based on H4PMo11VO40 was synthesized by following
the procedure of Rabia et al.17†
The performances of the catalysts were determined at 548 K
using 300 mg of catalyst packed into a fixed-bed reactor fed with
a 20 wt% aqueous solution of lactic acid as a reactant feed. After
1 h of stabilization, the reaction products were condensed over
4 h using ice traps. Analyses of the condensed liquid products
were performed by the HPLC technique, and gas phase analysis
was performed using a gas chromatograph equipped with a
TCD.† The accuracy and reproducibility of the measurements
were carefully checked (Table S2†).
Due to the superior performances of the W-containing HPA-
based catalysts, we further focused our study on this kind of
active phase. In a second series of experiments, we studied the
impact of the HPA surface density on the catalytic performance
at an iso-HPA wt% loading (20 wt%), but on silica supports with
very different specific surface areas. Taking into account a mean
˚
diameter of the Keggin unit of 12 A, we evaluated the theoretical
coverage of the support.12 Accordingly, the surface coverage was
estimated at 36% in the case of the catalyst prepared with 20 wt%
HPA on CARiACT. The surface acidic density was decreased
by using an SBA-15 with a significantly higher specific surface
area (541 vs. 272 m2 g-1), on which the surface coverage was then
14%.
The performances of the catalysts based on HPAs supported
on CARiACT Q-15 are shown in Fig. 1. The activity and
selectivity are strongly dependant on the type of supported
active phase. The catalysts prepared from phosphomolybdic acid
and vanado-phosphomolybdic acids showed comparatively low
conversions of lactic acid (i.e., 70 and 60%, respectively), whereas
the catalysts based on silicotungstic and phosphotungstic acids
were significantly more active, with conversions exceeding
90%. This is not surprising, as molybdenum-containing HPAs
are less acidic than tungsten-based examples. The resulting
acetaldehyde yield was in the region of 80% for the tungsten-
containing compounds, which was twice as much as observed
for the molybdenum-based HPAs, the yields of which were
no more than 34% acetaldehyde. Furthermore, in these latter
cases, the formation of propanoic acid was observed, with
3% yield over the phosphomolybdic acid and even 8% over
the vanado-phosphomolybdic acid, whereas the silicotungstic
and phosphotungstic acids exclusively showed the formation of
acetaldehyde.
The catalytic results are reported in Fig. 2. On the one
hand, a lower density in the active phase, i.e., a lower density
of acid sites, resulted in a decrease in activity (namely 91–
94% over CARiACT vs. 85–86% over SBA-15). On the other
hand, we observed an increase in the yield of acetaldehyde for
silicotungstic acid supported on SBA-15, with a value as high
as 83% (thus corresponding to a selectivity for acetaldehyde
of 98%) compared to a yield of 76% over SBA-15-supported
phosphotungstic acid. For both HPAs, the formation of very
small amounts of propanoic acid was observed over SBA-15
(1%), which was not the case when using CARiACT. Irrespective
of the support, silicotungstic acid showed a slightly lower
activity, but was more selective to acetaldehyde than phospho-
tungstic acid, which is not a significant feature when considering
the well known interdependence of the conversion/selectivity
couple. It is worth mentioning that no decrease in their activity
was observed during the time on stream of these catalysts,
Fig. 1 The catalytic performances of HPAs deposited on CARiACT
Fig. 2 The catalytic performances of W-containing HPAs supported
Q-15.
on CARiACT Q-15 and SBA-15.
This journal is
The Royal Society of Chemistry 2010
Green Chem., 2010, 12, 1910–1913 | 1911
©

View Article Online
Table 1 The textural properties of supports and catalysts based on
the performances of which were thus very stable (Fig. S2†).
Furthermore, one can state that the observed performance over
HPAs in the catalytic decarbonylation of lactic acid largely
outperforms the results previously published for other acid
catalysts such as zeolites, niobium oxide and alumina, which
do not exhibit a yield in acetaldehyde exceeding 31%.6–9
The aforementioned selectivity for propanoic acid that is
only observed over SBA-15 and not over CARiACT-supported
silicotungstic acid leads to a question about the structural
integrity of the active phase on this former support. Indeed,
the decomposition of silicotungstic acid yielding tungsten oxide
could be an explanation for the presence of the newly formed
redox properties. Nevertheless, the IR spectrum of fresh and
spent catalyst containing 20 wt% silicotungstic acid over SBA-
15 exhibited the characteristic absorption bands of the Keggin
unit at 981 and 928 cm-1 (Fig. 3).† This suggests the structural
integrity of the HPA after impregnation and also, after test, we
cannot even exclude the notion that a small undetectable part
might be decomposed.
silicotungstic acida
Silica support:
SBA-15
0
CARiACT
H4SiW12O40 amount (wt%):
20
0
20
S
BET/m2 g-1
541
1.20
9
431
1.05
9
272
1.25
15
246
1.06
15
Vp/cm3 g-1
Dp/nm
a SBET = specific surface area (BET), Vp = pore volume, Dp = pore
diameter.
Linkage of the HPA to the silica surface was experimentally
probed using nuclear magnetic resonance spectroscopy with
magic angle spinning (MAS-NMR) on the silicon nucleus.† The
29Si MAS-NMR spectra of the initial SBA-15 support and of
the catalyst containing 20 wt% silicotungstic acid are shown
in Fig. 4. Deconvolution of the spectra (Fig. S3; Table S3)†
clearly shows that the relative intensities of the Q2 [(SiO)2–Si–
(OH)2; -94 ppm] and Q3 [(SiO)3–Si–OH; -104 ppm] signals
decreased after impregnation of the SBA-15 host support
with the HPA, whereas the Q4 [Si–(SiO)4; -113 ppm] signal
relative proportion remained constant. This means that some
of the silanol groups (Si–OH) on the surface were covered
with silicotungstic acid.† Therefore, new silica species, namely
Q3-HPA [(SiO)3–Si–HPA] and Q2-HPA [(SiO)2–Si–HPA], were
formed with an identical chemical shift of -108 ppm, which
made discrimination between them impossible by this technique.
However, the deconvolution clearly shows that the loss in
the Q2 and Q3 signal relative contribution is proportionally
compensated-for by the appearance of the contribution of the
combined Q3-HPA and Q2-HPA signals. This contribution is
quantified at 14% (Fig. S3; Table S3†). Accordingly, the coverage
of the silica species on the surface was subsequently calculated
as the ratio of the number of covered surface silica species
(combined Q2-HPA and Q3-HPA peak area) to the total number
of surface silica species in the initial SBA-15 support (Q2 and
Q3 peak areas). This gives an experimental coverage of 17%,
which is consistent with the theoretically obtained value of
14%. This confirms the good distribution of the HPA on the
SBA-15 surface, as already suggested by the results of the N2
physisorption experiments.
Fig. 3 FT-IR spectra of (a) bare SBA-15 support, catalyst containing
20 wt% silicotungstic acid (b) before and (c) after test, and (d) pure
silicotungstic acid.
Furthermore, the dispersion of the active phase in the
silica host support was evaluated using nitrogen physisorption
experiments.† The results of these textural analyses, which were
carried out on catalysts based on 20 wt% silicotungstic acid
on SBA-15 and CARiACT, are given in Table 1. A ca. 20%
decrease in specific surface was observed after impregnation.
This decrease is mechanically linked to the weight of the
introduced active phase. Due to the non-porous character of
HPAs, the HPA layer does not generate a new surface, meaning
that the specific surface decreases proportionally to the weight
of the introduced active phase. Therefore, one can state that
the impregnation leads to the formation of a well-distributed
layer over the silica surface. Overloading of the active phase
would indeed result in the plugging of pores, which would
induce a significant decrease in the specific surface area. As
this phenomenon was not observed in the present case, one
can consider that pore blocking is locally limited, and that
the majority of the porous network of the catalyst should be
accessible to the reactants.
Fig. 4 29Si MAS-NMR spectra for (a) SBA-15 and (b) the catalyst
containing 20 wt% silicotungstic acid on SBA-15.
1912 | Green Chem., 2010, 12, 1910–1913
This journal is
The Royal Society of Chemistry 2010
©

View Article Online
Conclusion
Notes and references
1 C. H. Christensen, J. Rass-Hansen, C. C. Marsden, E. Raarningen
and K. Egeblad, ChemSusChem, 2008, 1, 283.
2 B. Katryniok, S. Paul, M. Capron and F. Dumeignil, ChemSusChem,
2009, 2, 719.
3 E. L. Kunkes, D. A. Simonetti, R. M. West, J. C. Serrano-
Ruiz, C. A. Ga¨rtner and J. A. Sumesic, Science, 2008, 322,
417.
4 M. Ai and K. Ohda, Appl. Catal., A, 1997, 150, 13.
5 R. Datta and M. Henry, J. Chem. Technol. Biotechnol., 2006, 81,
1119.
6 P. Sun, D. Yu, K. Fu, M. Gu, Y. Wang, H. Huang and H. Ying,
Catal. Commun., 2009, 10, 1345–1349.
7 H. Wang, D. Yu, P. Sun, J. Yan, Y. Wang and H. Huang, Catal.
Commun., 2008, 9, 1799–1803.
8 J. C. Serrano-Ruiz and J. A. Dumesic, ChemSusChem, 2009, 2,
581.
9 C. Paparizos, W. G. Shaw, S. R. Dolhyj (Standard Oil Company),
Eur. Pat. 0181718, 1985.
10 R. Jira, Angew. Chem., Int. Ed., 2009, 48, 9034.
11 W. S.-L. Mok and J. Antal Jr., J. Org. Chem., 1989, 54, 4596.
12 M. Misono, T. Okuhara and N. Mizuno, Catalysis by Heteropoly
Compounds: Successful Design of Catalysis, Amsterdam, Elsevier
Science Publishers, 1988, 267.
In this work, a highly selective decarbonylation of lactic acid to
acetaldehyde was achieved. When using less acidic heteropoly-
compounds like phosphomolybdic acid or H4PMo11VO40, the
decarboxylation pathway became more important. Due to
the increased redox character of these latter compounds, the
hydrogen liberated during decarboxylation supposedly easily
reduces lactic acid to propanoic acid.
The best results were obtained when using 20 wt% silico-
tungstic acid supported on SBA-15 or CARiACT. Whereas
the CARiACT-supported silicotungstic acid catalyst showed a
higher conversion of lactic acid (91% vs. 85%), the SBA-15-based
catalyst was more selective for acetaldehyde, whereby a slightly
higher yield of 83% (vs. 81% over CARiACT) was achieved.
Both results represent remarkable performances, especially when
compared to conventional acid catalysts such as zeolites, alu-
mina and niobium oxide, which are selective to acrylic acid but
give no more than a 31% yield of acetaldehyde.6–9 However, when
considering the aim of elaborating an economical industrial
process, the use of CARiACT seems to be a more reasonable
option regarding the catalysts’ preparation costs.
13 K. Nowinska, J. Chem. Soc., Chem. Commun., 1990, 44.
14 N. Mizuno, T. Watanabe and M. Misono, J. Phys. Chem., 1985, 89,
80.
15 N. Mizuno, T. Watanabe and M. Misono, J. Phys. Chem., 1990, 94,
890.
Acknowledgements
16 J. Roggenbuck, G. Koch and M. Tiemann, Chem. Mater., 2006, 18,
4151.
The authors thank the Japanese company Fuji Silysia for kindly
providing the CARiACT silica.
17 C. Rabia, M. M. Bettahar, S. Launay, G. Herve´ and M. Fournier, J.
Chim. Phys., 1995, 92, 1442.
This journal is
The Royal Society of Chemistry 2010
Green Chem., 2010, 12, 1910–1913 | 1913
©