Catalytic upgrading of lactic acid to fuels and chemicals by dehydration/hydrogenation and C-C coupling reactions

Article abstract of DOI:10.1039/b906869d

Aqueous solutions of lactic acid can be catalytically processed over Pt(0.1%)/Nb₂O₅ into a spontaneously-separating organic layer that can serve as source of valuable chemicals (propanoic acid and C ₄-C₇ ketones

Full text of DOI:10.1039/b906869d

Cutting-edge research for a greener sustainable future
www.rsc.org/greenchem
Volume 11 | Number 8 | August 2009 | Pages 1069–1272
ISSN 1463-9262
Fajula et al.
Etherification of glycerol
Hutchings et al.
Selective formation of lactate
Banerjee et al.
Biodegradable nanoporous materials
Dumesic and Serrano-Ruiz
Catalytic upgrading of lactic acid

COMMUNICATION
www.rsc.org/greenchem | Green Chemistry
Catalytic upgrading of lactic acid to fuels and chemicals by
dehydration/hydrogenation and C–C coupling reactions†
Juan Carlos Serrano-Ruiz and James A. Dumesic*
Received 6th April 2009, Accepted 21st May 2009
First published as an Advance Article on the web 3rd June 2009
DOI: 10.1039/b906869d
Aqueous solutions of lactic acid can be catalytically
processed over Pt(0.1%)/Nb2O5 into a spontaneously-
separating organic layer that can serve as source of valu-
able chemicals (propanoic acid and C4–C7 ketones) and/or
can be used to produce high energy-density liquid fuels
(C4–C7 alcohols). The niobia support plays a key role in
the synthesis by catalyzing dehydration and C–C coupling
reactions.
to valuable chemicals. One advantage of this approach is that we
can accomplish both the removal of oxygen and the upgrading
of intermediates in a single reactor and under high water
environments by using a water-stable bifunctional Pt/Nb2O5
catalyst.
Fig. 1 shows the reaction pathways that we have employed
to upgrade lactic acid into valuable chemicals over Pt/Nb2O5.
The reaction scheme is based on the formation of two less-
oxygenated reaction intermediates: propanoic acid and acetalde-
hyde (each labeled with a square in Fig. 1). The formation
of propanoic acid from lactic acid involves the removal of
the hydroxyl group through C–O cleavage, which can be
achieved by direct C–O hydrogenolysis (on platinum sites) or
can take place over the bifunctional Pt/Nb2O5 catalyst through
dehydration (on the acidic support) followed by hydrogena-
tion (on metal sites). The formation of acetaldehyde involves
C–C bond cleavage through decarbonylation/decarboxylation
reactions, catalyzed by platinum sites and also through acid-
catalyzed decarbonylation.8 After the two intermediates have
been formed, the presence of niobia can selectively guide the
synthesis (by C–C coupling reactions) towards more valuable
products instead of allowing these intermediates to undergo
further reduction reactions (e.g., leading first to propanol and
ethanol and subsequently to propane and ethane). Thus, pro-
pionic acid can be converted into pentanones by ketonization,
and acetaldehyde can be upgraded into C4-carbonyls (mainly
2-butanone) by aldol-condensation reactions. These compounds
can be upgraded to larger ketones (in the C6 and C7 range)
by means of successive aldol condensation reactions with the
previously formed acetaldehyde, which should be present in
the reactor in significant concentrations because the acidity of
niobia favors decarbonylation of lactic acid.8 In this respect,
catalysts consisting of low metal loadings supported on niobia
have been previously reported to be materials for condensation
of ketones, such as acetone, to larger carbonyls such as methyl
isobutyl ketone.9 Additionally, niobia was chosen as a support
because it provides excellent stability as well as sufficiently strong
acidity to carry out dehydration reactions of alcohols under
water environments.10
Utilization of renewable feedstocks for the chemical industry
is becoming ever more important in view of decreasing fos-
sil resources and increased pressure on our environment. In
this respect, biomass is a promising candidate to serve as a
sustainable source of organic carbon for the production of
industrial chemicals.1 However, establishment of a renewable
chemical industry, that is cost-competitive with the current one
based on petroleum, requires development of new processes
to convert biomass-derived feedstocks into valuable products.
Unlike petroleum feedstocks which possess low extents of func-
tionality, the compounds in biomass-derived resources contain
an excess of oxygenated groups, leading to high and uncontrolled
reactivity. Consequently, the challenge in the emerging renewable
chemical industry is to develop methods to decrease the oxygen
content of these molecules and, with it, to direct the conversion
to the desired final products.2
Lactic acid (2-hydroxypropanoic acid), containing adjacent
hydroxyl and carboxylic groups, can be considered to be a model
of an over-functionalized biomass-derived molecule. It can be
obtained from biomass with low cost by bacterial fermentation,
and it has a growing market because of its multiple applications.3
In addition, lactic acid has been suggested to be one of the most
promising building blocks for the future growth of a renewable
chemical industry.4 Previous works have reported the direct
conversion of lactic acid to products such as acetaldehyde,5
acrylic acid5,6 and propanoic acid.7 Herein we propose a new
strategy to upgrade this resource into valuable chemicals. The
general approach consists of a partial reduction of the oxygen
content of the molecule leading to less reactive intermediates
that still contain sufficient functionality for further upgrading
Table 1 summarizes experimental results (carbon distribution
and carbon selectivities) for the conversion of a 60 wt%
aqueous lactic acid solution (employing a hydrogen co-feed
at 80 cm3(STP) min-1) at 623 K and 57 bar and at various
space velocities over bifunctional Pt(0.1%)/Nb2O5. These results
were compared with those obtained for a pure monofunc-
tional hydrogenation catalyst, Pt(0.1%)/Vulcan, to study the
important role of niobia in the process. Vulcan was selected
Department of Chemical and Biological Engineering, University of
Wisconsin–Madison, 1415 Engineering Drive, Madison, WI 53706,
USA. E-mail: dumesic@engr.wisc.edu; Fax: +1 608 262 5434;
Tel: +1 608 262 1095
† Electronic supplementary information (ESI) available: Catalyst prepa-
ration and characterization; reaction kinetics studies; acknowledge-
ments. See DOI: 10.1039/b906869d
This journal is
The Royal Society of Chemistry 2009
Green Chem., 2009, 11, 1101–1104 | 1101
©

Table 1 Carbon distribution and carbon selectivities for lactic acid conversion at 623 K and 57 bar over Pt(0.1%)/Vulcan and Pt(0.1%)/Nb2O5.
Values in parentheses refer to methane. COx indicates CO and CO2. The balance in carbon selectivities is closed to 100% by inclusion of other
products such as cycloketones and lactones. The high degree of reproducibility of these data and the high stability of the catalyst with respect to
time-on-stream are documented in the ESI.† Carbon balances range from 88 to 110%
Catalysts
Pt/Vulcan
Pt/Nb2O5
WHSV (h-1)
0.8
75
25
1.7
70
30
—
5.0
64
36
—
0.8
43
10
1.7
42
12
46
5.0
41
14
45
Carbon gas-phase effluent (%)
Carbon aqueous-phase effluent (%)
Carbon organic-phase effluent (%)
Total carbon selectivity (%)
COx
Propanoic acid + esters
Acetaldehyde
—
47
54
9
1
18
47
7
7
12
1
9 (8)
40
5
16
6
0
4 (3)
26
19
0
4
24
22 (2)
4
26
25
2
4
18
17 (2)
4
37
18
9
4
19
4 (1)
2
0
Ethanol
C4–C7 Condensation products
C1-C7 Alkanes (methane)
Propanol/acetone
1
15 (10)
1
0
0
13
0
25
Lactic acid
0
0
Fig. 1 Reaction pathways for lactic acid dehydration/hydrogenation and C–C coupling over Pt/Nb2O5.
as a support because of its chemical inertness derived from
its low level of functionality. Pt(0.1%)/Vulcan showed lower
activity than Pt(0.1%)/Nb2O5 at the same conditions (lactic
acid conversion of 76% versus 100% at weight hourly space
velocity, WHSV = 5.0 h-1), indicating a promotional effect of
niobia for the overall process. In terms of carbon distribution,
Pt(0.1%)/Vulcan produced a higher amount of product carbon
in the gas phase (75% of reacted carbon at 0.8 h-1 and 100%
conversion)whileinthecaseofthebifunctionalPt(0.1%)/Nb2O5
catalyst, about 60% of the reacted carbon remained in the
liquid phase for all conditions studied. Importantly, unlike
Pt(0.1%)/Vulcan, lactic acid conversion over Pt(0.1%)/Nb2O5
produced the onset of an organic phase that spontaneously
separated from water and accounted for almost the 50% of the
total reacted carbon at 100% conversion. This behaviour is a
clear indication that different reaction pathways are operative
depending on the nature of the catalyst used. Thus, lactic acid
reacted on Pt(0.1%)/Vulcan at high flow rates (WHSV = 5.0 h-1)
to produce mostly CO and CO2 (40% carbon) and acetaldehyde–
ethanol (22%), with lower amounts of methane (3% of the
carbon, included in C1–C7 alkane in Table 1) and propanoic
acid (5%). Remarkably, no C4–C7 condensation products were
1102 | Green Chem., 2009, 11, 1101–1104
This journal is
The Royal Society of Chemistry 2009
©

detected at these conditions, while 25% of the carbon remained
unreacted as lactic acid. When lower space velocities were used
(WHSV = 0.8 h-1), lactic acid conversion was complete, but
CO and CO2 accounted for most of the detected carbon (54%),
while acetaldehyde (1%) was almost completely hydrogenated
to ethanol (18%). Interestingly, the amount of propanoic acid
(and its ethyl-ester) increased at the low space velocity (9%,
with only 1% C4–C7 condensation products) as well as methane,
which accounted for 10% of the detected carbon. In view of
these results, it can be concluded that, at these conditions,
Pt(0.1%)/Vulcan favoured C–C cleavage reactions leading to
C1 species (CO, CO2 and CH4) and C2 species to a lower
extent (acetaldehyde–ethanol), whereas this catalyst was not
effective for C–O cleavage reactions leading to C3 species such
as propanoic acid. In agreement with this behaviour, it has
been reported that metals such as Pt are active and selective
for cleavage of C–C bonds (versus C–O cleavage) in oxygenated
hydrocarbons.11 Importantly, C–C condensation reactions were
not significant on the monofunctional Pt(0.1%)/Vulcan, since
negligible amounts of C4–C7 ketones were observed.
does not change significantly with Pt particle size for catalytic
processing of oxygenated hydrocarbons.13
The results detailed above indicate that a low metal content
bifunctional Pt(0.1%)/Nb2O5 catalyst can efficiently process
aqueous solutions of lactic acid into a spontaneously-separating
organic layer that has several uses. On one hand, this organic
effluent can be employed as source of valuable chemicals.
The composition of the organic phase (specifically the liquid
obtained after lactic acid processing at WHSV = 1.7 h-1, Table 1)
is shown in Fig. 2, A1, and included propanoic acid and esters
(46% of the carbon of this phase) and C4–C7 ketones (35%) in
important amounts, with C4–C7 alkanes, acetone and alcohols
accounting for the rest of the carbon. Taking advantage of the
different boiling points of propanoic acid14 (413 K) and the
C4–C7 ketones15 (C4: 353 K, C5: 373 K, C6: 398 K and C7: 427 K),
the oil could be separated by fractional distillation into its
valuable components: propanoic acid serving as a preservative
for foods and feeds,14 whereas the ketones have widespread use
as solvents and chemical intermediates.15
Lactic acid reacted completely over the bifunctional
Pt(0.1%)/Nb2O5 catalyst at high space velocities (WHSV =
5.0 h-1) to produce mainly CO and CO2 (37% carbon),
acetaldehyde–ethanol (13%), and, importantly, a significant
amount of propanoic acid (18%) and C–C condensation
products (19%). A decrease in the space velocity to 0.8 h-1
caused a slight increase in the amount of C–C condensation
products (24%) at the expense of acetaldehyde that, as pointed
out in Fig. 1, preferentially formed C4-carbonyls versus direct
hydrogenation to ethanol (stable at 4%, Table 1). It is noteworthy
that almost all of the condensation products spontaneously
separated from water to form an organic phase, and only
negligible amounts of these compounds remained in the aqueous
phase. The low space velocities also favored the formation of
C1–C7 alkanes but, unlike Pt(0.1%)/Vulcan, significant amounts
of methane were not detected with the niobia-supported catalyst
(2%). The key role of the niobia support is clearly observed
by analyzing the main differences between both catalysts.
In particular, Pt(0.1%)/Nb2O5 achieves C–C coupling reac-
tions, such as ketonization and aldol condensation, leading to
C4–C7 condensation products stored in the organic layer,
whereas Pt(0.1%)/Vulcan catalyzed mainly C–C cleavage re-
actions leading to C1 species (COx and methane accounting for
almost 65% of the reacted carbon at low space velocities) with the
corresponding loss of carbon to the gas phase. Additionally, the
higher amount of propanoic acid produced by Pt(0.1%)/Nb2O5
(19% of the carbon at WHSV = 0.8 h-1) in comparison with
Pt(0.1%)/Vulcan (9%) supports the hypothesis that the higher
reactivity of platinum on an acidic support such as niobia is
caused by a two-step dehydration–hydrogenation process to
achieve C–O cleavage instead of C–C breaking.11 Finally, we
note here that both catalysts were prepared using the same low
loading of Pt (0.1%) to minimize catalyst costs for our process.
While it would be interesting to study the catalytic effect of
varying the dispersion of the Pt nanoparticles, we do not expect
this effect to be large. In particular, hydrogenation processes
(as involved in the bi-functional catalysis described here) are
typically structure insensitive,12 and we have observed elsewhere
that the selectivity for C–C versus C–O bond cleavage reactions
Fig. 2 A) Composition of the organic layer obtained for the conversion
of a 60 wt% solution of lactic acid in water over 1: Pt(0.1%)/Nb2O5 at
623 K, 57 bar and WHSV = 1.7 h-1; 2: Pt(0.1%)/Nb2O5 + Ce0.5Zr0.5O2
double bed at 623 K, 57 bar and WHSV = 1.7 h-1 (based on first bed);
3: organic layer obtained from 2 after hydrogenation over Ru(5%)/C at
373 K, 35 bar and WHSV = 1.3 h-1 (H2 co-feed at 50 cm3(STP)/min).
B) Composition of organic effluent obtained from 3; C2–C7 represent
the number of carbons of the corresponding alcohols.
On the other hand, the spontaneously-separated organic
phase could be upgraded (without the need of separation) to
high energy-density liquid-fuels (composed of C4–C7 alcohols)
in a cascade-mode approach. In this approach, the remaining
propionic acid and esters can first be converted into pentanones
This journal is
The Royal Society of Chemistry 2009
Green Chem., 2009, 11, 1101–1104 | 1103
©

via ketonization by adding a ceria-zirconia bed2 and using the
same conditions of pressure and temperature, as shown in Fig. 2,
A2. The organic effluent obtained with the Pt(0.1%)/Nb2O5 +
Ce0.5Zr0.5O2 double-bed arrangement contained approximately
55% of the feed carbon (versus 46% for the Pt(0.1%)/Nb2O5
single bed configuration, Table 1), and this organic liquid was
enriched in C4–C7 ketones (60% of the carbon, pentanone being
the main compound), with acetone (7%) and C2–C3 alcohols
(16%) also being present in the oil in significant amounts.
Remarkably, the double-bed reactor showed good stability for
the conversion of propanoic acid and its esters to pentanone
for more than 2 days on stream, and the fraction of these
compounds that remained unreacted (17%, Fig. 2, A2) could
be totally converted by using lower space velocities in the
double-bed configuration. Finally, this organic effluent enriched
in C4–C7 ketones can be quantitatively converted into the
corresponding alcohols (only 0.5% the carbon was lost in the
gas phase) with a Ru/C catalyst at low temperatures (Fig. 2,
A3). As a result, the final processed liquid contained a mixture
of alcohols in the range of C4–C7 that could be suitable for
use as high energy-density fuel. We note that the reduction
process with Ru/C at 373 K also caused a partial hydrogenation
of propanoic acid to 1-propanol whereas the esters remained
unreacted. In terms of the alcohol distribution, pentanol was the
most abundant alcohol and ethanol, 1-propanol and 2-propanol
(from the hydrogenation of the acetone) were also present in the
liquid in significant amounts (Fig. 2, B).
In summary, we report a catalytic route to selectively guide
the conversion of lactic acid (an important low-cost biomass-
derived commodity chemical with an expected growing market)
toward valuable products (propanoic acid and C4–C7 ketones)
using a low metal-content Pt(0.1%)/Nb2O5 catalyst. A novel as-
pect of the approach outlined here is that by using a bifunctional
catalyst containing metal and acid sites, the series of required
reactions leading to the desired products (e.g., dehydration–
hydrogenation and C–C coupling) can be carried out in a single
flow reactor, thus reducing capital and operating costs for the
process. The niobia support plays a crucial role in directing the
synthesis to valuable compounds by catalyzing C–C coupling
reactions such as aldol-condensation and ketonization, instead
of C–C cleavage reactions that take place over monofunctional
Pt/Vulcan and lead to loss of carbon in the gas phase.
This conversion of lactic acid over Pt(0.1%)/Nb2O5 maintains
approximately 50% of the feed carbon in an organic effluent
phase (rich in valuable products) that spontaneously separates
from the aqueous layer, thus eliminating the need to remove these
products from water. This oil could be used as a direct source of
valuable chemicals and/or can be potentially upgraded to liquid
fuels. Finally, the approach outlined here of oxygen removal and
subsequent upgrading of intermediates in a single reactor is
flexible in that it could be applied not only to lactic acid but also
to other over-functionalized biomass-derived molecules. One
especially interesting example would be the catalytic processing
of levulinic acid, which can be derived from waste biomass
sources (e.g., paper mill sludge, urban waste paper, agricultural
residues) by the Biofine process.16
Notes and references
1 C. H. Christensen, J. Rass-Hansen, C. C. Marsden, E. Taarning and
K. Egeblad, ChemSusChem, 2008, 1, 283.
2 E. L. Kunkes, D. A. Simonetti, R. M. West, J. C. Serrano-Ruiz, C. A.
Ga¨rtner and J. A. Dumesic, Science, 2008, 322, 417.
3 Y. Y. Wee, J. N. Kim and H. W. Ryu, Food Technol. Biotechnol., 2006,
44, 163.
4 T. Werpy and G. Petersen, Top Value Added Chemicals
from Biomass, US Department of Energy, Office of Scientific
and Technical Information, No.: DOE/GO-102004-1992, 2004,
http://www.nrel.gov/docs/fy04osti/35523.pdf.
5 L. Li, J. R. Portela, D. Vallejo and E. F. Gloyna, Ind. Eng. Chem.
Res., 1999, 38, 2599.
6 C. T. Lira and P. J. McCrackin, Ind. Eng. Chem. Res., 1993, 32, 2608.
7 B. Odell, G. Earlam and D. J. Cole-Hamilton, J. Organomet. Chem.,
1985, 290, 241.
8 W. S. Mok, M. J. Antal and M. Jones, J. Org. Chem., 1989, 54, 4596;
G. C. Gunter, R. H. Langford, J. E. Jackson and D. J. Miller, Ind.
Eng. Chem. Res., 1995, 34, 974.
9 T. Higashio and T. Nakayama, Catal. Today, 1996, 28, 127; M. Paulis,
M. Mart´ın, D. B. Soria, A. D´ıaz, J. A. Odriozola and M. Montes,
Appl. Catal., A, 1999, 180, 411.
10 R. M. West, D. J. Braden and J. Dumesic, J. Catal., 2009, 262, 134.
11 J. N. Chheda, G. W. Huber and J. A. Dumesic, Angew. Chem., Int.
Ed., 2007, 46, 7164; G. W. Huber, R. D. Cortright and J. A. Dumesic,
Angew. Chem., Int. Ed., 2004, 43, 1549.
12 P. Biloen, F. M. Dautzenberg and W. H. M. Sachtler, J. Catal., 1977,
50, 77.
13 J. W. Shabaker, R. R. Davda, G. W. Huber, R. D. Cortright and J. A.
Dumesic, J. Catal., 2003, 215, 344.
14 U. R. Samuel, W. Kohler, A. O. Gamer and U. Keuser, Propionic acid
and derivatives in Ullmann’s Encyclopedia of Industrial Chemistry,
VCH Publishers Inc, 1993.
15 Y. L. Hwang and T. C. Bedard, Ketones in Kirk-Othmer Encyclopedia
of Chemical Technology, 2001, John Wiley and Sons Inc.
16 S. W. Fitzpatrick, Production of levulinic acid by the hydrolysis
of carbohydrate-containing materials, World Pat. 9640609, 1997 to
Biofine Incorporated.
1104 | Green Chem., 2009, 11, 1101–1104
This journal is
The Royal Society of Chemistry 2009
©