Transfer hydrogenation of levulinic acid under hydrothermal conditions catalyzed by sulfate as a temperature-switchable base

Article abstract of DOI:10.1039/b924648g

It is demonstrated that transfer hydrogenation from formic acid to levulinic acid under hydrothermal conditions can be catalyzed by bases, but also by simple sodium sulfate. The action of salt addition could be clarified and ascribed to the changing dissociation constants at high temperature. This renders sulfate a temperature-switchable base in hydrothermal syntheses. Using such salts can help in preventing waste, as neutralization after reaction is not necessary. By optimizing the reaction conditions, the yield of γ-valerolactone, a sustainable biofuel molecule, could be raised to 12% for a simple passage through a capillary flow reactor with a residence time of less than 20 min.

Full text of DOI:10.1039/b924648g

View Article Online / Journal Homepage / Table of Contents for this issue
PAPER
www.rsc.org/greenchem | Green Chemistry
Transfer hydrogenation of levulinic acid under hydrothermal conditions
catalyzed by sulfate as a temperature-switchable base
Daniel Kopetzki* and Markus Antonietti
Received 24th November 2009, Accepted 18th January 2010
First published as an Advance Article on the web 16th February 2010
DOI: 10.1039/b924648g
It is demonstrated that transfer hydrogenation from formic acid to levulinic acid under
hydrothermal conditions can be catalyzed by bases, but also by simple sodium sulfate. The action
of salt addition could be clarified and ascribed to the changing dissociation constants at high
temperature. This renders sulfate a temperature-switchable base in hydrothermal syntheses. Using
such salts can help in preventing waste, as neutralization after reaction is not necessary. By
optimizing the reaction conditions, the yield of g-valerolactone, a sustainable biofuel molecule,
could be raised to 12% for a simple passage through a capillary flow reactor with a residence time
of less than 20 min.
is nontoxic, has favourable solvent characteristics, has an LD50
Introduction
significantly higher than ethanol, is biodegradable, exhibits a
Currently chemistry is mainly based on fossil resources. Under
high boiling point and very low vapour pressure, and so could
the threat of shortage and ultimately the depletion of this
be considered as green solvent. Furthermore it is a useful energy
feedstock, the green chemistry community is currently trying
storage molecule and can be directly employed as a fuel additive
to set up new platform chemicals based on biomass as a starting
for liquid transportation fuels.7
material for chemical processes. This has led to concepts like
g-Valerolactone can be synthesized by hydrogenation and
biorefinery/biorefining,1 but has also resulted in proposals for
subsequent cyclisation of levulinic acid, either by using
multiple-commodity chemicals based on biomass as a starting
heterogeneous8-11 or homogeneous2,11,12 catalysts. However both
material.2
systems employ noble and heavy metals, which are neither
Being the main component of biomass, carbohydrates will
sustainable nor low-priced. As an equimolar amount of formic
become a primary starting compound for sustainable chemistry.
acid is formed during production of levulinic acid from carbohy-
They can be converted by different means to a variety of
drates, it is convenient to use formic acid as reducing agent in a
compounds, such as fermentation to ethanol, lactic acid or
transfer hydrogenation (Scheme 2). In this way, the side product
citric acid, which is a current biotechnology process. Another
of a first reaction is used as a hydrogen source for the second.
very promising pathway to convert carbohydrates into useful
The usefulness of formic acid concerning hydrogen storage was
chemicals is the chemical dehydration to hydroxymethylfurfural.
recently illustrated by Beller et al.13
This can be obtained by simple heating,3 but also works well
and is more defined under non-aqueous conditions in ionic
liquids catalyzed by metal ions.4 In water, especially under acidic
conditions, this compound further reacts to levulinic and formic
acid, which can be carried out in quite high yield (Scheme 1).5
Scheme 2
The reduction of aldehydes with NaCOOH has been shown
to work without catalyst; however, at very high temperatures
◦
(250–350 C).14 In this paper we will investigate this transfer
hydrogenation of the even more unreactive ketone under green
conditions in water. We will demonstrate that this reaction can
Scheme 1
Levulinic acid itself is an interesting building block and can
be converted to a variety of useful compounds.6 Among those,
g-valerolactone is potentially of highest significance. It possesses
properties that characterize it as an ideal sustainable liquid.7 It
be performe◦d under hydrothermal conditions at temperatures
around 220 C where it is catalyzed even by simple ubiquitous
salts.
Results and discussion
Max-Planck-Institute of Colloids and Interfaces, Research Campus
Golm, D-14424, Potsdam, Germany.
E-mail: daniel.kopetzki@mpikg.mpg.de; Fax: +49-331-5679502;
Tel: +49-331-567-9538
As levulinic acid and formic acid are produced from car-
bohydrates in aqueous solution, we performed the transfer
hydrogenation in diluted solution with 0.1 M levulinic acid. A
656 | Green Chem., 2010, 12, 656–660
This journal is
The Royal Society of Chemistry 2010
©

View Article Online
slight excess of formic acid (0.15 M) was used. The experiments
were carried out in steel autoclaves by heating for 12 h. For
temperature optimization a tubular flow reactor with less than
20 min residence time was used. As a first test of whether
synthesis of g-valerolactone under hydrothermal conditions was
feasible, the mixture of levulinic acid and formic acid was simply
heated in an autoclave at 220 ◦C for 12 h, resulting in a yield of
1.0% g-valerolactone. By addition of Pd on activated alumina,
the yield increased to 29.0%. However this was accompanied by
a significant occurrence of further reduction towards pentanoic
acid. Furthermore, Pd not only catalyzed the desired hydrogen
transfer, but also the decomposition of formic acid to H2 and
CO2. This decomposition, although at much lower rate, also
occurs in high-temperature water without any additives.15 This
side reaction, together with the demand for a cheap, sustainable
catalytic scheme, makes the use of Pd@Al2O3 obsolete. However,
it was useful in this system as a benchmark value for the possible
conversion range. By using catalysts based on ruthenium, much
higher yields have been reported.11,12 However, Ru has the same
disadvantages as Pd concerning sustainability.
reaction possibilities can be considered. Whereas formic acid
releases CO2 and H2 upon decomposition, its anion formally
only transfers H-. Thus the reducing character of formic acid
is significantly different from that of formates. To perform the
reaction at different pH, variable amounts of either hydrochloric
acid or potassium hydroxide were added to the mixture of
levulinic and formic acid. The concentration of HCl was 0.01 M
or 0.1 M in the final solution, whereas 0.075 M, 0.15 M,
0.2 M or 0.3 M of KOH was added to vary the pH. In the
case of base addition, pH is buffered by formic acid (pKa
=
3.74) and levulinic acid (pKa = 4.59). Since during the reaction
acid is consumed or may decompose, the pH should increase
throughout reaction. Indeed this is validated by the experiment
(Fig. 1). The pH changes are largest in the alkaline region,
presumably due to the missing buffer capacity.
In hydrothermal reactions, even simple salts may activate the
water and can therefore act as catalysts.16 As cations might
activate the carboxy group by coordination, a 0.5 M salt solution
was used as solvent instead of pure water. Different chlorides
(KCl, CeCl3, CoCl2, SrCl2, CuCl2 and ZnCl2) were tested, but
none showed a significant activity for this reaction, even though
transition metals were employed (Table 1).
Taking a closer look at the different halides, one can notice
some effects following the Hofmeister series.17 Potassium fluo-
ride increases the yield compared to the pure aqueous solution,
whereas the other halides do not show any improvement. Also
a remarkable acceleration is found when Na2SO4 is added.
Furthermore, KH2PO4 increases the yield, whereas the more
basic phosphates do not show any catalytic activity, even
inhibiting the reaction. Finally, other basic salts like Na2SO3
and K2CO3 prevent any reaction. This makes pH a potential key
aspect of this reaction.
Fig. 1 The dependence of yield of g-valerolactone upon pH, measured
before (squares) and after (circles) the experiment.
Furthermore, the yield strongly depends on the pH. At very
acidic pH, where both reactants are protonated and uncharged,
the yield is very low. With increasing pH, the yield drastically
increases reaching a maximum at a pH around the pKa of formic
acid. Further increasing the pH leads again to a decreasing yield.
Here also levulinic acid becomes deprotonated, which seems
to inhibit the reaction. Obviously, the reaction is fastest for
levulinic acid being in the neutral and formic acid in the anionic
form. This leads to a rather sharp maximum of optimum pH for
product generation in water with a formal transfer of H- from
formate. So by simple addition of some base to the mixture of
levulinic and formic acid the yield can be drastically optimized.
In media that are too basic, lactonization is not expected to
occur. However, GC–MS did not reveal any other low molecular
weight compounds besides levulinic acid (especially not g-
hydroxyvaleric acid). So no reaction at all seems to occur at
high pH.
Another potential reason for the low yield under acidic
conditions is the fact that addition of HCl promotes the
autodecomposition of formic acid at those temperatures (Fig. 2),
which is – beside reaction with levulinic acid – a second reaction
channel.
Knowing the sensitivity of the transfer hydrogenation of
levulinic acid with formic acid on pH, the initial experiments
on the influence of pure salts onto the reaction can be
Therefore, we had to investigate the pH dependence of the
transfer hydrogenation without additional salt to distinguish
salt and pH effects. As both educts are acids, a variety of
Table 1 Effect of salt addition (0.5
g-valerolactone
M
salt) on the yield of
Yield (%)
Additive
No additive
Pd@Al2O3
CeCl3
CoCl2
SrCl2
CuCl2
ZnCl2
KF
KCl
1.0
29.0
1.4
1.0
1.1
0.0
0.1
11.3
1.4
1.2
0.7
11.0
0.0
0.0
3.4
0.7
0.0
KBr
KI
Na2SO4
Na2SO3
K2CO3
KH2PO4
K2HPO4
K3PO4
This journal is
The Royal Society of Chemistry 2010
Green Chem., 2010, 12, 656–660 | 657
©

View Article Online
-
Fig. 3 Variation of pKa of HSO4 (A) and formic acid (B) with
Fig. 2 Amount of levulinic acid (circles), formic acid (squares) and the
product g-valerolactone (triangles) versus pH prior to reaction.
increasing temperature, calculated from thermodynamic data.
re-analyzed. The pH probably accounts for the acceleration
pattern of the different potassium phosphates. It is now also
evident why very basic salts like Na2SO3 or Na2CO3 inhibit any
reaction. In contrast, slightly basic salts could shift the pH in
the narrow optimum range. This might be the case for KF and
Na2SO4.
-
the dissociation of HSO4 . At the high reaction temperatures in
our experiments sulfate can in fact act like a weak base with
a strength comparable to carboxylic acid salts and becomes
protonated to a significant extent while deprotonating formic
acid. We speculate that this also causes the high yield of
g-valerolactone in the presence of this salt.
The catalytic activity of sulfate, which is a quite neutral salt
For the other tested salts the pKa values at high temperature
-
(pKa = 1.99 for the deprotonation of HSO4 under ambient
◦
were also calculated for a temperature of 220 C.18 In the case
conditions) is still striking at first sight, whereas fluoride is more
basic (pKa = 3.17 under ambient conditions). When judging
about basicity of the tested salts, it is a striking result that the
pKa values at elevated temperatures can be quite different from
those under ambient conditions. So to really relate the effect
to pH, we have to compare the dissociation constants at the
reaction temperature of 220 ◦C. The temperature dependence
of the dissociation constant can be expressed by the van’t Hoff
equation
◦
of halides, data for 200 C were used (Fig. 4).22 In fact, the
-
pKa of hydrogen fluoride is quite near to the value of HSO4 ,
which results in a similar yield of product by adding KF.
The difference in dissociation constant between ambient and
-
reaction condition is, however, less pronounced than for HSO4 .
In a green, sustainable approach, KF cannot be recommended
due to its toxicity. On the other hand there is no concern in
using simple sulfates. In any case, the KF example allows us
to conclude that the observed effects of salt addition are not
attributed to Hofmeister effects, but rather to their influence on
pH at elevated temperatures.
(1)
with DH◦ being the standard molar enthalpy change for the
dissociation of the acid. Integration, neglecting the increased
pressure at high temperatures and the temperature dependence
of the heat capacity change Dc0p leads to
(2)
with T0 being the reference temperature (T0=298.15 K).18 With
data for formic acid (DG0 = 21.45 kJ mol-1, DH0 = 1.03 kJ mol-1
and Dc0p = -175 J K-1 mol-1)19 and HSO4- (DG0 = 11.342 kJ mol-1,
DH0 = -22.4 kJ mol-1 and Dcp0 = -258 J K-1 mol-1)18 the
-
temperature dependence of the pKa of formic acid and HSO4
was calculated (Fig. 3).
As for most simple acids, the dissociation constant of formic
acid decreases only slightly with increasing temperature. Even
though the pressure dependence was not taken into account, the
calculated pKa variation is in good agreement with experimental
values for both formic acid19,20 and hydrogensulfate.21 On the
other hand, the dissociation constant of HSO4 decreases very
strongly and even crosses that of formic acid at elevated
temperatures. The reason is the very high enthalpy change for
Fig. 4 Yield of product versus the pKa of the corresponding acids of
the tested salts at 220 ◦C and 200 ◦C for the halides respectively.
-
To gain further insight into the effect of added Na2SO4, stan-
dard mixtures of levulinic and formic acid were reacted in the
presence of different concentrations of Na2SO4. Increasing the
658 | Green Chem., 2010, 12, 656–660
This journal is
The Royal Society of Chemistry 2010
©

View Article Online
salt concentration increases the yield, until a plateau is reached
around 0.1 M of salt (Fig. 5). Interestingly, the yield increases
roughly with the logarithm of added salt concentration.
decreased yield at the standard reaction temperature of about
220 ◦C. This is due to the fact that in the tubular flow reactor the
residence time is less than 20 min, as compared to the batch mode
where it is about 12 h. The short reaction time therefore results
in incomplete conversion. Higher yields are expected when
pumping the reaction mixture into the reactor for a second time.
The highest potential for an increased yield is given when the
g-valerolactone is extracted prior to re-reacting the unconverted
starting educts, but such optimization by chemical engineering
is out of the scope of the present paper. Selective extraction can
be carried out for example with supercritical CO2.23
Conclusions
We have shown that the transfer hydrogenation from formic
acid to levulinic acid can be performed under hydrothermal
conditions. Optimization of pH allows one to increase the
yield of g-valerolactone, a solvent or fuel molecule with great
potential for sustainable chemistry based on biomass. Even
simple sulfates can promote the reaction. We have found that
this “salt catalysis” is probably not attributable to Hofmeister
effects, but purely to the pH influence at elevated temperatures.
Interestingly, sulfates in high temperature water are “switchable”
bases, and become more and more basic with increasing
temperature. This temperature dependence is more pronounced
for sulfate compared to other anions. It can be exploited for easy
preparation of reaction solutions that are neutral and thus easy
to handle under ambient conditions but drastically change their
properties at high temperature.
Such a temperature switch may be used to substitute con-
ventional bases that are traditionally employed as catalysts,
which require neutralization after reaction and thus create waste
salt. Furthermore, a temperature-switchable base will become
neutral when cooling to ambient conditions after reaction and
can therefore be reused. This adds another sustainable feature to
the high atom economy of the presented hydrothermal synthesis.
Fig. 5 Catalytic effect of Na2SO4 on the transfer hydrogenation.
For further optimization of the reaction conditions we
screened the temperature dependence of the reaction within an
8 mL tubular flow reactor for a fixed flow of 0.4 mL min-1 and
a pressure of 200 bar, corresponding to a reaction time between
18.1 min (at 175 ◦C) and 14.7 min (at 300 ◦C) (Fig. 6). With
its continuous product stream, such a setup is certainly more
adapted to industrial processes. In these experiments we used
0.125 M Na2SO4, as the reaction performed in autoclaves gave
satisfactory yields at that concentration. This experiment shows
that a sufficiently high temperature to activate the formic acid
is necessary to start the reaction. Product yield under those
conditions passed through a maximum, as a temperature above
275 ◦C again lowers the yield. We assume that the spontaneous
decomposition of formic acid is too fast in this range. Hydrogen,
formed by this process, cannot reduce levulinic acid. It is the
direct hydrogen transfer which creates the g-valerolactone.
Experimental
Levulinic acid (98% purity) and g-valerolactone (99% purity)
were purchased from Sigma-Aldrich, and formic acid (98%
purity) from Acros Organics. All chemicals were used without
further purification.
Experiments were carried out in 45 mL acid digestion vessels
from Paar, equipped with a Teflon inlet and a glass vial. In case
of samples with high pH or when KF was added, the reaction
was performed directly in the Teflon inlet, as glass was etched to
a great extent under these conditions at high temperature. The
autoclaves were put in an oven and heated up to 220 ◦C for 12 h.
About 10 mL of solution were used for each experiment.
For measuring the temperature dependence of the reaction, we
used the X-Cube Flash flow reactor from Thales Nano. Solution
was pumped through an 8 mL coil at a flow rate of 0.4 mL min-1.
This corresponds to residence times in the range of 18.1 min (at
175 ◦C) to 14.7 min (at 300 ◦C), calculated from the density ratio
of water under reaction and ambient conditions. Because of the
relatively low concentration of solute, the density of pure water
was used.24
Fig. 6 Variation of yield with reaction temperature measured with a
Thales Nano X-Cube Flash flow reactor and a residence time in the
range of 18.1 min (at 175 ◦C) and 14.7 min (at 300 ◦C) with 0.125 M
Na2SO4 added.
Comparing the overall yield from this experiment with the
reaction performed in autoclaves (see Table 1) one notices the
This journal is
The Royal Society of Chemistry 2010
Green Chem., 2010, 12, 656–660 | 659
©

View Article Online
Qualitative analysis was carried out with GC–MS. The sample
was silylated with a mixture of bis-N,O-trimethylsilyl trifluo-
roacetamide with 1% trimethylchlorosilane prior to analysis.
Quantitative analysis was performed by HPLC equipped
with a refractive index detector. A C18 column was used for
separation with an eluent consisting of 1% acetonitrile in 10 mM
HCl. Citric acid was used as internal standard.
10 Z. P. Yan, L. Lin and S. J. Liu, Energy Fuels, 2009, 23, 3853–
3858.
11 H. Heeres, R. Handana, D. Chunai, C. B. Rasrendra, B. Girisuta and
H. J. Heeres, Green Chem., 2009, 11, 1247–1255.
12 L. Deng, J. Li, D. M. Lai, Y. Fu and Q. X. Guo, Angew. Chem., Int.
Ed., 2009, 48, 6529–6532.
13 A. Boddien, B. Loges, H. Junge and M. Beller, ChemSusChem, 2008,
1, 751–758.
14 T. A. Bryson, J. M. Jennings and J. M. Gibson, Tetrahedron Lett.,
2000, 41, 3523–3526.
15 P. E. Savage, Chem. Rev., 1999, 99, 603–621.
16 M. Siskin and A. R. Katritzky, Chem. Rev., 2001, 101, 825–835.
17 M. G. Cacace, E. M. Landau and J. J. Ramsden, Q. Rev. Biophys.,
1997, 30, 241–277.
18 R. N. Goldberg, N. Kishore and R. M. Lennen, J. Phys. Chem. Ref.
Data, 2002, 31, 231–370.
19 M. H. Kim, C. S. Kim, H. W. Lee and K. Kim, J. Chem. Soc., Faraday
Trans., 1996, 92, 4951–4956.
20 J. L. S. Bell, D. J. Wesolowski and D. A. Palmer, J. Solution Chem.,
1993, 22, 125–136.
21 A. G. Dickson, D. J. Wesolowski, D. A. Palmer and R. E. Mesmer,
J. Phys. Chem., 1990, 94, 7978–7985.
22 B. N. Ryzhenko and O. V. Bryzgalin, Geokhimiya, 1987, 137–142.
23 R. A. Bourne, J. G. Stevens, J. Ke and M. Poliakoff, Chem. Commun.,
2007, 4632–4634.
References
1 S. Fernando, S. Adhikari, C. Chandrapal and N. Murali, Energy
Fuels, 2006, 20, 1727–1737.
2 H. Mehdi, V. Fabos, R. Tuba, A. Bodor, L. T. Mika and I. T. Horvath,
Top. Catal., 2008, 48, 49–54.
3 Y. Roman-Leshkov, J. N. Chheda and J. A. Dumesic, Science, 2006,
312, 1933–1937.
4 H. B. Zhao, J. E. Holladay, H. Brown and Z. C. Zhang, Science,
2007, 316, 1597–1600.
5 Y. Takeuchi, F. M. Jin, K. Tohji and H. Enomoto, J. Mater. Sci.,
2008, 43, 2472–2475.
6 B. Girisuta, L. P. B. M. Janssen and H. J. Heeres, Chem. Eng. Res.
Des., 2006, 84, 339–349.
7 I. T. Horvath, H. Mehdi, V. Fabos, L. Boda and L. T. Mika, Green
Chem., 2008, 10, 238–242.
8 H. Broadbent and T. Selin, J. Org. Chem., 1963, 28, 2343–2345.
9 P. Maki-Arvela, J. Hajek, T. Salmi and D. Y. Murzin, Appl. Catal.,
A, 2005, 292, 1–49.
24 E. W. Lemmon, M. O. McLinden and D. G. Friend, ‘Thermophysical
Properties of Fluid Systems’ in NIST Chemistry WebBook, NIST
Standard Reference Database Number 69, National Institute of Stan-
dards and Technology, Gaithersburg MD, 20899, http://webbook.
nist.gov, retrieved November 3, 2009.
660 | Green Chem., 2010, 12, 656–660
This journal is
The Royal Society of Chemistry 2010
©