Reaction kinetics studies of the conversions of formic acid and butyl formate over carbon-supported palladium in the liquid phase

Article abstract of DOI:10.1016/j.jcat.2012.03.014

Formic acid and butyl formate conversion were studied in the liquid phase over Pd and Ru catalysts. Pd/C was more active, selective, and stable for CO₂/H₂ production in the liquid phase. Kinetic studies were performed over Pd/C at temperatures from 403 to 443 K, at space velocities from 3.8 to 970 h^-1, in the presence of CO and H₂ at partial pressures from 0 to 0.4 and 12.6 atm, respectively, and liquid water. Space velocity studies probed the importance of primary decomposition pathways to CO₂ and CO compared to the secondary water-gas shift reaction. Generally, the rate of the primary pathway was an order of magnitude higher than the rate of the secondary pathway. Over Pd/C, formic acid decomposed primarily via decarboxylation (to CO₂/H₂), whereas butyl formate primarily decomposed via decarbonylation (to CO/butanol). When water was present, the formate ester hydrolyzed, which increased the selectivity toward CO₂ and H₂.

Full text of DOI:10.1016/j.jcat.2012.03.014

Journal of Catalysis 290 (2012) 193–201
Contents lists available at SciVerse ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Reaction kinetics studies of the conversions of formic acid and butyl formate
over carbon-supported palladium in the liquid phase
⇑
Brandon J. O’Neill, Elif I. Gürbüz, James A. Dumesic
Department of Chemical and Biological Engineering, University of Wisconsin, Madison, WI 53705, United States
a r t i c l e i n f o
a b s t r a c t
Article history:
Formic acid and butyl formate conversion were studied in the liquid phase over Pd and Ru catalysts. Pd/C
was more active, selective, and stable for CO2/H2 production in the liquid phase. Kinetic studies were per-
formed over Pd/C at temperatures from 403 to 443 K, at space velocities from 3.8 to 970 hꢀ1, in the pres-
ence of CO and H2 at partial pressures from 0 to 0.4 and 12.6 atm, respectively, and liquid water. Space
velocity studies probed the importance of primary decomposition pathways to CO2 and CO compared to
the secondary water–gas shift reaction. Generally, the rate of the primary pathway was an order of mag-
nitude higher than the rate of the secondary pathway. Over Pd/C, formic acid decomposed primarily via
decarboxylation (to CO2/H2), whereas butyl formate primarily decomposed via decarbonylation (to CO/
butanol). When water was present, the formate ester hydrolyzed, which increased the selectivity toward
CO2 and H2.
Received 9 January 2012
Revised 5 March 2012
Accepted 18 March 2012
Available online 20 April 2012
Keywords:
Formic acid
Formate ester
Decarboxylation
Decarbonylation
Kinetic model
H2 production
Pd/C
Ó 2012 Elsevier Inc. All rights reserved.
1. Introduction
literature exploring the decomposition kinetics of formic acid has
focused on homogenous catalytic systems [38–43], surface science
The need to develop sustainable sources of energy and the po-
tential to use biomass to supplant petroleum are critical to ensure
environmental conservation, stimulate economic development,
and promote national security. A major challenge to achieve the
transition from petroleum-based feedstocks to the use of biorenew-
able resources is the effective deoxygenation of bio-molecules. To
this end, reactions of carboxylic acids and esters have been identi-
fied as important reactions for the conversion of biomass feedstocks
to hydrogen, fuel grade alkanes, and chemicals [1–5]. The deoxy-
genation of carboxylic acids and esters can take place by parallel
pathways involving dehydrogenation/decarboxylation or dehydra-
tion/decarbonylation, as depicted in Scheme 1. Consequently, a fun-
damental understanding of these reactions for carboxylic acids and
esters is necessary to design highly selective processes that effi-
ciently produce fuels and chemicals from biomass.
As the simplest carboxylic acid, formic acid has played a critical
role in the elucidation of carboxylic acid reactivity [6]. Formic acid
has been utilized as a probe molecule for heterogeneous reactions
on metals [7–25], metal-oxides [23,24,26–30], metal carbides [31–
33], and oxide-supported metals [34,35]. Researchers employing
theoretical and computational methods have also investigated for-
mic acid as a molecule to probe reactivity [36,37]. In general, the
studies under ultra-high vacuum (UHV), and gas-phase reactions
over heterogeneous catalysts. However, the literature is limited
with regard to the decomposition of esters and a direct comparison
of how they differ from their associated carboxylic acid [44,45].
Moreover, the parallel pathways in Scheme 1 are connected by
the water–gas shift (WGS) reaction [11–14,31,32,34,46], and this
connection is especially important in biomass conversion pro-
cesses that often are carried out in the liquid phase in the presence
of water [2,47,48].
Another motivation for studying the decomposition of formic
acid is that formic acid is produced as a coproduct during the con-
version of cellulose to levulinic acid. Thus, formic acid can be used
as a source of H2 for the reduction of levulinic acid to gamma-val-
erolactone (GVL), allowing GVL to be produced from biomass with-
out the need for an external source of hydrogen [5]. However, the
use of mineral acids to produce levulinic acid and formic acid from
cellulose deconstruction has posed significant challenges for
downstream upgrading. An interesting strategy to remove the for-
mic and levulinic acids from the acidic aqueous environment is to
employ reactive extraction via esterification by reaction with an
alcohol or an alkene [4]. Accordingly, the present work was under-
taken to explore the reactivity of formic acid and butyl formate as
probe molecules over two metal catalysts used in previous work,
that is Pd/C and Ru/C. Studies at different space velocities were car-
ried out to elucidate the triangular reaction scheme represented in
⇑
Corresponding author.
E-mail address: dumesic@engr.wisc.edu (J.A. Dumesic).
0021-9517/$ - see front matter Ó 2012 Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.jcat.2012.03.014

194
B.J. O’Neill et al. / Journal of Catalysis 290 (2012) 193–201
Liquid samples were taken via a sampling port and quenched in
an ice water bath before analysis.
For all experiments, gas analysis was carried out utilizing a Shi-
madzu 2014 gas chromatograph with a HaySep DB 100/120 column
(Alltech) and a thermal conductivity detector to detect CO and CO2.
An Agilent GC6890 equipped with an Rtx column (Agilent) and a
flame ionization detector was used to detect gas-phase alkanes.
Liquid analysis was done by HPLC (Waters 2695 system with either
a Bio-Rad Aminex HPX-87H column and a RI 410 detector or an
Agilent Zorbax SB-C18 column and a UV detector). Liquid and gas
analyses were performed throughout the duration of the experi-
ment. Steady state was usually achieved in less than 8 h over Pd/
C and was not achieved over Ru/C. The system was determined to
be free of mass transfer limitations by the Weisz-Prater criterion.
Scheme 1. Triangle reaction scheme for the decomposition of formic acid and
formate esters. R symbolizes either atomic hydrogen or an alkyl group.
Scheme 1, as well as to probe the effects of water on the hydrolysis
of the ester in solution. The results of these reaction kinetics stud-
ies allow for the evaluation of the individual kinetic contributions
of the primary decomposition pathways, secondary WGS reaction,
and homogeneous chemistry in solution. Finally, a simple kinetic
model is presented to describe the competing reaction pathways
and the observed reaction kinetics.
3. Results and discussion
3.1. Catalyst comparison
Figs. 1 and 2 compare the rates and selectivities of formic acid
and butyl formate decomposition over Pd/C and Ru/C catalysts.
To calculate the turnover frequency (TOF), the rate was normalized
by the number of surface sites as measured by the irreversible up-
2. Experimental
take of CO at 308 K, equal to 126
l
mol gꢀ1 for Ru/C (26% disper-
Reaction kinetics measurements were carried out using a half-
inch stainless steel tube as a flow reactor, followed by a gas liquid
separator at room temperature. A fixed-bed, down-flow configura-
tion was maintained by using a quartz wool plug at the bottom.
The two catalysts studied were purchased from Sigma–Aldrich
and consisted of either 5 wt.% Ru or 10 wt.% Pd supported on car-
bon. The catalyst bed was diluted with crushed, fused SiO2 (4–
16 mesh, Sigma–Aldrich) in a 1:1 volumetric ratio. The reactor
was filled to the top of the heating zone with more crushed, fused
SiO2. The reactor was heated via aluminum blocks in a well-insu-
lated furnace (Applied Test Systems). The temperature was mea-
sured by type-K thermocouples (Omega) and controlled by a PID
controller (Love Controls Series 16A) connected to a variable trans-
former (Staco Energy Products).
Prior to reaction kinetics studies, the catalysts were reduced
in situ at 673 K (0.45 K minꢀ1 ramp) for 2 h. All reaction kinetics
measurements were carried out at a temperature of 423 K, unless
otherwise noted. Helium and H2 (Airgas, industrial grade) were
co-fed for the space velocity studies, and mixtures of 1% CO in he-
lium and 7% H2 in helium (Airgas) were co-fed with additional he-
lium for the reaction order studies. The flow rates of gasses were
controlled using mass flow controllers (Brooks 5850E), and the to-
tal gas flow rate out of the system was measured using a bubble
meter. The liquid reagents (formic acid, Sigma–Aldrich, 98–100%;
1-butanol, Sigma–Aldrich, 99.9%; n-butyl formate, Alfa-Aesar,
97%) were used as purchased, and they were fed to the reactor
from a graduated cylinder via a needle by a high-performance li-
quid chromatography (HPLC) pump (Lab Alliance Series I). The to-
tal pressure in the reactor was controlled by a back-pressure
regulator (GO model BP-60) and maintained at 530 psi.
sion) and 56–110
l
mol gꢀ1 for Pd/C, depending on the
manufacturer batch number (6–12% dispersion). It can be seen in
Figs. 1 and 2 that Pd/C exhibits a higher initial TOF for the decom-
position of formic acid and butyl formate compared to Ru/C
0.08
0.06
0.04
0.02
0
100%
96%
92%
88%
0
1000
2000
3000
4000
5000
Time (min)
Fig. 1. TOF (j) and CO2 selectivity (h) for decomposition of 2 M formic acid in H2O
over Pd/C, compared to TOF (d) and CO2 selectivity (s) over Ru/C at 423 K, with a
He co-feed at 530 psi, and WHSV = 97 hꢀ1
.
0.01
75%
50%
25%
0%
The adsorption uptake of CO was measured on each catalyst at
308 K using an ASAP 2020 (Mircomeritics) after reduction of the
catalyst at 673 K (0.45 K minꢀ1) for 2 h. The number of catalytic
sites was taken to be equal to the irreversible CO uptake at this
temperature.
Experiments were carried out to assess the hydrolysis rate of
butyl formate in a batch reactor at 530 psi in a 50 mL stainless steel
pressure vessel (Parr Instruments). The reactor was heated using
heating tape, and the temperature was maintained at 423 K with
the same equipment, reagents, and gases used in the flow reactor
experiments. Reaction solutions were delivered by an HPLC pump
after heating the pressure vessel to the reaction temperature.
0.005
0
0
1000 2000 3000 4000 5000 6000 7000
Time (min)
Fig. 2. TOF (j) and CO2 selectivity (h) for decomposition of 2 M butyl formate in 1-
butanol over Pd/C, compared to TOF (d) and CO2 selectivity (s) over Ru/C at 443 K,
with a He co-feed at 530 psi, and WHSV = 30 hꢀ1
.

B.J. O’Neill et al. / Journal of Catalysis 290 (2012) 193–201
195
(0.075 sꢀ1 versus 0.021 sꢀ1 for formic acid, and 0.0047 sꢀ1 versus
80
60
40
20
0
0.0045 sꢀ1 for butyl formate), which is in relative agreement for
the two catalysts with recent studies by Solymosi et al. [25] for for-
mic acid in the vapor phase. Additionally, Pd/C has a higher selec-
tivity toward the decarboxylation/dehydrogenation pathway than
Ru/C (97% versus 90% for formic acid and 48% versus 23% for butyl
formate).
The results in Figs. 1 and 2 demonstrate that the rate and H2/CO2
selectivity of formic acid decomposition are higher than for butyl
formate over both Pd/C and Ru/C. In addition, the Pd/C catalyst
exhibits superior stability compared to the Ru/C catalyst. It is prob-
able that the deactivation of the Ru/C catalyst is caused by coking of
the catalyst surface, as suggested by the observation of significant
amounts of CH4 production. Since CH4 production requires the
cleavage of CO bonds and the formation of surface C species, it is
possible that carbonaceous species accumulate on the surface of
the catalyst and block active sites. Support for this hypothesis is
provided by the work and literature review of Mukkavilli et al.
[49]. Furthermore, CH4 production is not observed at these condi-
tions over the Pd/C catalyst, consistent with the observation that
this catalyst is stable at the reaction conditions of the present study.
Accordingly, the Pd/C catalyst is the focus of the further kinetic
studies due to its higher activity, selectivity, and stability.
0
0.005
0.01
0.015
0.02
1/WHSV (h)
Fig. 4. CO2:CO ratio with respect to inverse WHSV for decomposition of 2 M formic
acid in liquid water at 423 K and 530 psi with 35% H2 in He co-feed.
The effect of hydrogen on the decomposition pathway was
investigated because decomposition of formic acid is a potential
route for production of hydrogen and because many biomass pro-
cessing strategies involve reactions under hydrogen atmospheres.
Fig. 4 illustrates the effect of changing the gaseous co-feed from
helium to a mixture of 35% H2 in helium. The increase in the
CO2:CO ratio at lower space velocity that is seen in Fig. 3 is largely
absent after the addition of H2 because the predicted WGS equilib-
rium ratio of CO2:CO has shifted to 54:1 in the presence of H2.
Extrapolating to infinite space velocity using only the data in the
region before WGS equilibrium is reached (1/WHSV = 0.0055 h)
gives a ratio of CO2:CO of 42:1 (97.7% selectivity), similar to the va-
lue of 35:1 (97.2% selectivity) obtained in the absence of H2.
3.2. Primary pathway for formic acid decomposition
The results of this study show that the decarboxylation/dehy-
drogenation pathway appears to be favored for the decomposition
of formic acid in the liquid phase over Pd/C. However, it was not
clear whether this high selectivity to CO2 was due to the primary
decarboxylation/dehydrogenation pathway or to the secondary
WGS reaction pathway. Accordingly, studies were performed to
measure the CO2:CO ratio in the product stream at varying weight
hourly space velocities (WHSV), starting with a feed of 2 M formic
acid in water. At high space velocity, it can be assumed that the
contributions of secondary reaction pathways were minimized.
The trend illustrated in Fig. 3 can be extrapolated to infinite space
velocity to predict a CO2:CO ratio of 35:1, which is equal to the rate
of dehydrogenation/decarboxylation divided by the rate of dehy-
dration/decarbonylation. This ratio of 35:1 corresponds to a CO2
selectivity greater than 97%, showing that the high CO2 selectivities
for Pd/C are predominantly due to the dehydrogenation/decarbox-
ylation primary decomposition pathway. Fig. 3 shows that at lower
space velocities, the CO2:CO ratio can be increased by more than
one order of magnitude to greater than 430:1 by allowing the
WGS reaction to occur. This order of magnitude increase in the
CO2:CO ratio corresponds to an improvement in selectivity up to
99.8%. Operating at even lower space velocities would allow WGS
to reach equilibrium and could further increase the CO2:CO ratio
to 1800:1 (99.9% CO2 selectivity).
3.3. Primary pathway for butyl formate decomposition
To gain insight into the selectivity differences between the
decompositions of formic acid and butyl formate demonstrated
in Figs. 1 and 2, studies were carried out at different space veloci-
ties for the decomposition of butyl formate, analogous to those
studies carried out for formic acid. Comparison of the results in
Fig. 5 for butyl formate with the behavior in Fig. 3 for formic acid
demonstrates that changing the functional group from a carboxylic
acid to an ester decreases the CO2:CO ratio at infinite space velocity
due to a change in the primary decomposition pathway. The sharp
increase in CO2:CO ratio at high space velocity is attributed to the
decomposition of a formic acid impurity (up to 3 wt.% of pure butyl
formate) in the feed that reacts at a much higher rate than butyl
formate, as shown in Figs. 1 and 2. Thus, the CO2:CO ratio at lower
space velocities (1:10, 8.5% selectivity to CO2) is a better indication
of the ratio of CO2:CO due to the primary pathways of formate es-
ter decomposition. At these higher conversions, the effect of the
small impurity of formic acid is mitigated by the higher conversion
450
375
300
225
150
75
0.6
0.4
0.2
0
0
0
0.005
0.01
0.015
0.02
0.025
0
0.1
0.2
0.3
1/WHSV (h)
1/WHSV (h)
Fig. 3. CO2:CO ratio with respect to inverse WHSV for decomposition of 2 M formic
Fig. 5. CO2:CO ratio with respect to inverse WHSV for decomposition of 2 M butyl
acid in liquid water at 423 K and 530 psi with He co-feed.
formate in 1-butanol at 423 K and 530 psi with He co-feed.

196
B.J. O’Neill et al. / Journal of Catalysis 290 (2012) 193–201
3
2
1
0
0.0075
100%
75%
50%
25%
0%
0.005
0.0025
0
0.1
0.2
0.3
0
0
1/WHSV (h)
2
4
6
Fig. 6. CO2:CO ratio with respect to inverse WHSV for decomposition of 2 M butyl
formate in 1-butanol at 423 K and 530 psi with 35% H2 in He co-feed.
Water Concentration (M)
Fig. 7. TOF (j) and CO2 selectivity (s) for the decomposition of 2 M butyl formate
in 1-butanol as a function of water concentration with a co-feed of 35% H2 in He at
423 K, 530 psi, and a WHSV = 15 hꢀ1. The predicted CO2 selectivity (e) is calculated
assuming that the increase in rate is entirely due to the decomposition of formic
acid formed from the hydrolysis of butyl formate.
of butyl formate, and the reaction conditions (absence of water or
H2) effectively eliminate the secondary WGS reactions (in both the
forward and reverse directions).
To explicitly account for the role of H2 in the decomposition of
butyl formate, the gaseous co-feed was switched from He to 35%
H2 in He. As demonstrated in Fig. 6, the presence of H2 shifted
the predicted equilibrium ratio of CO2:CO toward CO due to the re-
verse WGS reaction, while simultaneously shifting the CO2:CO ra-
tio at infinite space velocity toward CO2 (from 1:2 to 4:1). The
latter effect is caused by a decrease in the rate of butyl formate
decomposition upon addition of H2, which thus increases the rela-
tive contribution to the rate by decomposition of the formic acid
impurity at a given space velocity. It should be noted that the
behavior at low space velocities in Fig. 6 is different from that seen
in Fig. 4, with the ratio of CO2:CO decreasing at low space velocity
rather than increasing. This behavior is due in part to a lack of
water, which moves the reaction into the reverse WGS regime,
and in part to the depletion of the formic acid impurity, which
leads to a decrease in the relative contribution of formic acid com-
pared to butyl formate. The CO2:CO ratio was equal to 1:4 at low
space velocities, with a predicted equilibrium CO2:CO ratio of
1:15 due to reverse water–gas shift.
Table 1
Thermal hydrolysis of 2 M butyl formate in 1-butanol at 423 K under a pressure of
530 psi as a function of water concentration. The maximum rate and equilibrium
conversion are calculated as averages of the rate or extent of generation of formic acid
and the rate or extent of consumption of butyl formate.
Water
concentration
(M)
Maximum rate
mol minꢀ1 mLꢀ1
Equilibrium
conversion (%) equilibrium
conversion (%)
Predicted
(
l
)
2
4
6
4.3
7.6
15.7
4
10
12
4
9
15
decomposition, indicating that all of the formic acid produced dur-
ing hydrolysis should be consumed as the solution passed over the
catalyst bed. This is in agreement with the observation that formic
acid was never detected as a product. While the amount of formic
acid that was produced in the feed before the catalyst bed is not
sufficient to explain the entire increase in rate produced by the
addition of water, the hydrolysis reaction likely continued to occur
over the catalyst bed. In addition, the hydrolysis of alkyl esters is
expected to take place on the catalyst surface, as reported in the
heterogeneous catalysis literature [50].
Studies of the decomposition of butyl formate were carried out
at different space velocities to probe the effect that the addition of
water has on the primary decomposition pathway. The CO2:CO ra-
tio increased at higher concentrations of water across the range of
space velocities, as shown in Fig. 8. In addition, Fig. 8 shows that
the CO2:CO ratio extrapolated to infinite space velocity for butyl
formate decomposition increased significantly for all water con-
centrations compared to Figs. 5 and 6 in the absence of water.
Importantly, the value of the CO2:CO ratio for butyl formate in
the presence of water approached that of formic acid decomposi-
tion. At the highest space velocity, the selectivity to CO2 for decom-
position of butyl formate with 6 M water nearly equaled that of
pure formic acid (95.5% versus 97.7%) and was much higher than
the selectivity for butyl formate decomposition without water at
high space velocity (95.5% versus 79%). This comparison indicates
that hydrolysis of the ester and the subsequent decomposition of
the formic acid is the primary cause of the increase in selectivity
when co-feeding water.
3.4. Effect of water on butyl formate decomposition
In a previous study by Gürbüz et al. [4], it was observed that the
addition of water during the decomposition of butyl formate over
Pd/C increased both the rate of decomposition and the selectivity
toward the decarboxylation pathway. The effect of water concen-
tration on the rate and selectivity of butyl formate decomposition
was thus further investigated, and the results of these experiments
are shown in Fig. 7, where the concentration of water was varied
from 0 to its solubility limit of 6 M in the butyl formate/1-butanol
reaction mixture. The addition of 2 M water improved the rate by
more than a factor of 2. Further increasing the water concentration
to 6 M gave a smaller additional increase in the rate of the reaction.
Fig. 7 also shows that the selectivity improvement from the addi-
tion of water matches the improvement predicted by assuming
all of the increased activity can be attributed to the decomposition
of formic acid formed from hydrolysis of butyl formate.
The possible role of hydrolysis contributing to the reactivity of
butyl formate was investigated by measuring the rate of hydrolysis
in a batch reactor in the absence of the Pd/C catalyst. The hydroly-
sis reaction was observed to be autocatalytic, because of acid-cat-
alyzed processes enhanced by the formation of formic acid. This
behavior suggests that the hydrolysis reaction can occur in the
heated zone of the flow reactor even before reaching the catalyst
bed. Table 1 shows the calculated equilibrium conversion and the
maximum observed rate for autocatalytic hydrolysis of butyl for-
mate at the three different water concentrations. The fastest rate
of hydrolysis was lower than the observed rate of formic acid
The addition of water increased the value of the CO2:CO ratio
that is predicted by the WGS equilibrium (up to 200:1 for the case
of 6 M water). Despite this prediction, the observed ratio of CO2:CO
decreased at low space velocity to a value less than 4:1 for all
water concentrations. This behavior indicates that WGS is not the
dominant factor in determining the change in selectivity with

B.J. O’Neill et al. / Journal of Catalysis 290 (2012) 193–201
197
30
20
10
0
-2
-1
0
1
0
-4
-8
(a)
X = -0.02
X = -0.01
0
0.1
0.2
0.3
2) (atm)
ln(PH
1/WHSV (h)
-5
-4
-3
-2
0
(b)
Fig. 8. CO2:CO ratio with respect to inverse WHSV for decomposition of 2 M butyl
formate with 2 M ( ), 4 M ( ), and 6 M ( ) H2O in 1-butanol with 35% H2 in He co-
feed at 423 K, and 530 psig.
-2
X = -0.2
-4
decreasing space velocity for the decomposition of butyl formate in
the presence of water, but rather the trend in the CO2:CO ratio is
caused by the interplay between ester hydrolysis and ester decarb-
onylation. More specifically, at high space velocity, the selectivity
is dominated by the decomposition of formic acid formed from bu-
tyl formate hydrolysis in the presence of water. At low space veloc-
ities, the formic acid has been completely consumed, and the trend
in overall selectivity is dominated by the decomposition of butyl
formate that has a significantly lower primary pathway selectivity
toward CO2 (observed to be as low as 8.5% experimentally).
We note that the CO2 selectivity in this paper differs from the
value of 100% reported previously because the decomposition of
butyl ester in the present study was performed in excess 1-butanol
rather than with excess water and formic acid in the co-feed [4].
Thus, the equilibrium conversion and the rate for ester hydrolysis
were lower in the present work compared to the previous study.
Since the ratio of water to butyl formate was much higher in the
previous study, it allowed for more extensive hydrolysis of the es-
ter, thus allowing the overall reaction to proceed almost entirely
through formic acid. Additionally, since very low space velocities
were used to ensure full conversion in the previous study, there
was ample time for WGS to further increase selectivity toward
CO2, similar to the results observed at low space velocity for formic
acid decomposition in the present work.
-6
X = -0.4
-8
-10
ln(PCO) (atm)
-4
-3
-2
-1
0
(c)
-2
X = 0.3
X = 0.7
-4
-6
-8
-10
ln(PFA) (atm)
Fig. 9. TOF for the decomposition of FA versus partial pressures of (a) H2, (b) CO,
and (c) FA. Experimental data are given by solid symbols: TOF for decarboxylation/
dehydration (j) and TOF for decarbonylation/dehydrogenation (d). Solid lines
show predictions of the kinetic model. Values of X are reaction orders calculated
from experimental data.
had detrimental effects on the rate, and thus these species were in-
cluded in a site blocking term. The low order of H2 suggested that
the surface coverage by H-atoms was only significant at high par-
tial pressures. The rate was inhibited strongly by CO even at low
partial pressures, indicating a strong interaction of CO with the cat-
alyst surface. Because the binding of CO with a metal surface is
known to be highly dependent on the surface coverage, Eq. (6)
was introduced to modify the free energy of CO adsorption as a
function of CO partial pressure, similar to the strategy employed
by Grabow et al. [51]. The data in Fig. 9c show that the reaction or-
ders for the dehydration/decarbonylation and dehydrogenation/
decarboxylation reactions were fractional with respect to formic
acid, thus necessitating its inclusion in the sight blocking term as
well.
3.5. Reaction order studies and kinetic modeling
To help describe and quantify the reaction kinetics observed
experimentally in this study, a simple kinetic model was developed
and implemented in MATLAB by solving the material balance
equations for a plug flow reactor (PFR). The complex nature of
these reactions with multiple pathways taking place in the liquid
phase makes it difficult to extrapolate model parameters from
the literature and necessitates the use of a simplified model with
a limited number of parameters. This type of simplified model is
based on mechanistic reasoning, but does not attempt to explain
all the intricacies of the underlying mechanisms of formic acid
(FA) or butyl formate (BF) decomposition. Rather the model is used
to describe the competing reaction pathways and observed kinetics
under the current set of reaction conditions.
Rate expressions based upon the partial pressures of reactants
and products were developed by assuming that the compositions
of the liquid and gas phases were quasi-equilibrated. The rates of
adsorption and desorption from the catalyst surface of all species
were assumed to be quasi-equilibrated, the decomposition of for-
mic acid was assumed to be irreversible, and the water–gas shift
reaction was assumed to be reversible. Studies of the formic acid
decomposition rate at various partial pressures of CO and H2
(Fig. 9a and b) demonstrated that CO, and to a lesser extent H2,
The experimental data of this study for the liquid-phase decom-
position of formic acid can be described by the rate expressions in
Eqs. (1)–(3),
rco ¼ kco
P
FAH2ꢁ
ð1Þ
ð2Þ
rCO ¼ kCO2 PFAH3ꢁ
2
!
PCO PH
rWGS ¼ kWGSP0:125PH
1 ꢀ
ð3Þ
2
2
2O
CO
Keqm;WGSP0:125PH
CO
2

198
B.J. O’Neill et al. / Journal of Catalysis 290 (2012) 193–201
1
0.00232 0.00236 0.0024
0.00244 0.00248 0.00252
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
H
¼
ð4Þ
(a)
ꢁ
-1
1 þ KCOPCO
þ
KH PH þ KFAPFA
2
2
ꢀ
ꢁ
ꢀ
D
RT
Gi
Ki ¼ exp
ð5Þ
ð6Þ
Ea = 67 kJ/mol
Ea = 88 kJ/mol
-5
P
D
GCO
¼
D
GoCO þ 0:75 CO RT
Pref
where rCO is the rate of the dehydration/decarbonylation reaction,
rCO2 is the rate of the dehydrogenation/decarboxylation reaction,
rWGS is the rate of the water–gas shift reaction, Pi is the partial pres-
sures of species i, Ki is the adsorption equilibrium coefficient for
-9
1/T (K-1)
0.00232
species i,
DGi is the free energy of adsorption for species i, Keqm,WGS
0.00224
0.00228
0.00236
0.0024
is the equilibrium constant of the water–gas shift reaction, h repre-
(b)
⁄
-5.5
sents the vacant sites on the catalyst, Pref is the reference pressure at
standard conditions, R is the universal gas constant, T is the reaction
temperature, and 0.75 is a parameter fit for this model. These rate
expressions imply the following rate controlling steps for formic
acid decomposition
Ea = 86 kJ/mol
-6.5
FA þ 2h ! CO þ H2O
ð7Þ
ꢁ
Ea = 87 kJ/mol
FA þ 3h ! CO2 þ H2
ð8Þ
ꢁ
where formic acid requires two and three vacant sites in the dehy-
dration/decarbonylation and dehydrogenation/decarboxylation
reactions, respectively. The different orders with respect to the va-
cant sites were determined empirically to describe the dependence
of the rates on the partial pressure of formic acid (Fig. 9c). A justifi-
cation for the different dependence of rCO and rCO2 on the fraction of
vacant sites is that decarbonylation may take place by the reaction of
a monodentate carboxyl species with an adjacent vacant site, and
decarboxylation may take place by reaction of a bidentate formate
species with a vacant site. Additionally, the low order (approxi-
mately 0.1) with respect to CO partial pressure for the rate of WGS
was determined empirically, because the apparent rate of WGS var-
ied little when the initial partial pressure of CO was varied. This va-
lue was the highest fractional order that could be assigned without
the model violating the experimentally observed result of net CO
production. A low order with respect to CO was also observed exper-
imentally in results communicated to us by Mehta and Ribeiro for
their work currently in progress on liquid-phase water–gas shift
over a PtMo/C catalyst [52]. Also, we note that the solvent, in this
case water, does not appear in the site blocking term, because its
partial pressure, and hence its coverage, was constant throughout
the experiments. Accordingly, it was lumped with the reaction rate
constants to decrease the number of model parameters. Similarly, a
reaction order of 1 with respect to water in the WGS rate expression
was used because of the nearly constant partial pressure of water.
The reaction rate constants for formic acid and butyl formate
were described using the Arrhenius form shown in Eq. (9), and
the water–gas shift rate constant was described using Eq. (10):
-7.5
1/T (K-1)
Fig. 10. TOF versus temperature for decomposition of (a) FA and (b) BF plus the FA
impurity. Solid points show experimental data for decarboxylation/dehydration (j)
and decarbonylation/dehydrogenation (Ç). Solid lines show predictions of the
kinetic model.
stant because we are using an empirical rate expression for this
reaction, in contrast to using a mechanistically based rate expres-
sion. Fig. 10a shows the experimentally determined activation ener-
gies, along with the experimental and model-predicted values of
TOF for decomposition of formic acid with respect to temperature
and Fig. 10b shows data for the decomposition of BF. (A kinetic
model for the decomposition of BF will be addressed later in this
paper.)
The relative values of the activation energies for the decompo-
sition of formic acid versus the decomposition of butyl formate
indicate that the high CO2 selectivity for formic acid decomposition
is dominated by the change in activation energy of the decarboxyl-
ation pathway (from 88 kJ/mol to 67 kJ/mol) upon changing the
functional group from the ester to carboxylic acid, while the activa-
tion energy of the decarbonylation pathway remains relatively un-
changed (from 85 kJ/mol to 88 kJ/mol). This change in activation
energy may indicate that the formate species thought to be respon-
sible for production of CO2 is more readily produced from the acid
than the ester, whereas the presence of an ester functional group
has a potentially small stabilizing effect on the formation of a po-
tential carboxyl intermediate. The values of Ea and ko are listed in
Table 2. Experimental values of activation energies and approxi-
mate pre-exponential factors were extracted from plots of experi-
mental data and used as initial guesses for the model. The
uncertainties of these values are reported as the standard error
of the slope and the standard error in the y-intercept of Fig. 10,
respectively, as calculated by the LINEST function of Excel. The
approximate pre-exponential factor was calculated by dividing
the y-intercept of Fig. 10 by the respective partial pressure and sur-
face vacancy terms from Eqs. (1), (2), (11), and (12). A value of 0.66
was used for the vacancy term as it was a value typical of the initial
vacancy according to the model calculations. The model activation
energies, pre-exponential factors, and free energies of adsorption
were treated as model parameters and optimized using the ‘nlinfit’
ꢀ
ꢀ
ꢁ
ꢀ
ꢀEa
ki ¼ ko exp
ð9Þ
RT
ꢁꢁ
ꢀEa
1
T
1
Tavg
ki ¼ ko exp
ꢀ
ð10Þ
R
where ki is the reaction rate constant, ko is the pre-exponential fac-
tor, Ea is the reaction activation energy, R is the gas constant, T is the
temperature of the reaction, and Tavg is the average temperature of
the data set. The alternative expression for the WGS rate constant
was used because it is a more efficient parameterization for numer-
ical optimization. This mathematical form decouples the effect of
the activation energy and pre-exponential factor on the overall rate
constant. We have employed this approach for the WGS rate con-

B.J. O’Neill et al. / Journal of Catalysis 290 (2012) 193–201
199
Table 2
Pre-exponential factors and activation energies for model reactions.
(a)
400
200
0
a
c
Parameter Model ko
Model Ea
Approximate ko
Experimental
(kJ/mol)
Ea (kJ/mol)
kBF,CO
1.6 0.1 ꢂ 108
5.2 0.1 ꢂ 107
3.1 0.1 ꢂ 10ꢀ4
4.3 0.9 ꢂ 109
5.7 0.1 ꢂ 108
85.3 0.4
1.5 0.1 ꢂ 108
6.6 0.5 ꢂ 107
86
87
N/A
3
5
kBF,CO2
88
1
b
kWGS
90 30
N/A
kFA,CO
kFA,CO2
88
67
1
1
2.49 0.08 ꢂ 109 88
2
3
5.2 0.2 ꢂ 108
67
a
b
c
The units for ko are atm sꢀ1 except for kWGS, which has units of atm1.125 sꢀ1
kWGS was not measured experimentally.
The approximate value of ko was calculated for comparison with the model
.
0
0.5
1
using h of 0.66, typical of initial vacancies calculated by the model.
⁄
Fractional Distance Along PFR
(b)
100
(a)
1.E+06
1.E+03
1.E+00
75
50
25
0
0
0.5
1
0
25
50
75
100
Fractional Distance Along PFR
Experimental Conversion (%)
Fig. 12. (a) Ratio of the decarboxylation rate of FA to the WGS rate calculated by the
kinetic model, plotted versus the fractional distance along the PFR. (b) Ratio of the
decarbonylation rate of BF to the WGS rate calculated by the kinetic model, plotted
versus the fractional distance along the PFR.
100
98
(b)
Analogous to the reaction order studies in Fig. 9 for the decom-
position of formic acid, we carried out reaction order studies for
the decomposition of butyl formate. Fig. 13 shows the turnover fre-
quency for the decomposition of butyl formate versus the partial
pressures of H2, CO, and BF along with the results from the kinetic
model that is described below. Note that the trends in the TOF for
CO2 production in these plots are dominated by the amount of CO2
produced from the formic acid impurity, which reacts to complete
conversion and leads to nearly zero order with respect to CO partial
pressure and the strong positive reaction order with respect to bu-
tyl formate (and consequently formic acid) partial pressure.
The experimental data of Figs. 10b and 13 were used to formu-
late a kinetic model for the decomposition of BF by adding the fol-
lowing rate expressions to the rate expressions described
previously for the decomposition of formic acid (the latter of which
were needed to account for the formic acid impurity in the feed),
96
96
98
100
Experimental Selectivity (%)
Fig. 11. Rectification plots of the (a) conversion and (b) selectivity for the
decomposition of FA at various space velocities in He ( ) and 35% H2 in He ( ),
at different temperatures ( ), and for different partial pressures of FA ( ), CO ( ),
and H2
and 3.
( ). The values of the parameters used in the model are given in Tables 2
function of MATLAB, and the uncertainties of the optimized param-
eters were calculated with the ‘nlparci’ function.
The rectification plot of Fig. 11 demonstrates the ability of the
simplified model to predict the experimental data for the decom-
position of formic acid, assuming the reactor operates as a PFR.
In particular, it demonstrates the ability of this model to correctly
predict the trends in selectivity observed during the space velocity
studies and reaffirms the conclusion that these trends are evidence
of the secondary role of the water–gas shift reaction in determin-
ing the selectivity of CO2 versus CO compared to the primary role
of the direct decomposition pathways. Further evidence for the
secondary role of water–gas shift is provided by Fig. 12a, which
shows that the ratio of the primary decomposition pathway to
the water–gas shift rate is greater than 100 even at conversions
of formic acid higher than 90% where WGS would have its largest
effect. Finally, the model allows the calculation of the rates at the
inlet of the reactor, corresponding to the infinite space velocity
condition. The calculated CO2 selectivity (97.5%) of 2 M formic acid
in water at 423 K agrees very well with the value predicted exper-
imentally (97.2%). The values of the adjustable parameters used in
the model are provided in Tables 2 and 3.
rco ¼ kco
P
BFH2ꢁ
ð11Þ
rCO ¼ kCO2 PBFH3ꢁ
ð12Þ
2
and the site vacancy term was modified to account for the presence
of butyl formate on the surface.
1
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
H
ꢁ
¼
ð13Þ
1 þ KCOPCO
þ
KH PH þ KFAPFA þ KBFPBF
2
2
The temperature dependence of each rate constant was de-
scribed with an Arrhenius expression (Eq. (9)). The values of the
additional parameters are listed in Tables 2 and 3. The model was
solved assuming the parameters from the formic acid model were
constant, and only the new parameters for butyl formate were al-
lowed to vary. Fig. 10b shows that the kinetic model captured the
trends of the experimental data versus reaction temperature. The
rectification plot of Fig. 14 shows that the model describes the

200
B.J. O’Neill et al. / Journal of Catalysis 290 (2012) 193–201
Table 3
Values of standard
calculated by the model.
50
(a)
DG of adsorption for CO, H2, BF, and FA
Goads ðkJ=molÞ
Species
D
COa
H2
ꢀ0.65 0.01
25
0
18.0 0.5
BF
FA
ꢀ0.55 0.07
ꢀ4.6 0.9
Goads;CO is a function of the
a
The model value for
partial pressure of CO according to Eq. (6). The value in
the table is for PCO equal to 0.
D
0
25
50
Experimental Conversion (%)
-6
-4
-2
0
2
80
60
40
20
0
-6.5
(b)
(a)
X = -0.02
X = 0.01
-7.5
0
20
40
60
80
ln(PH2) (atm)
Experimental Selectivity (%)
-4
-2
0
-6.8
-7.2
-7.6
-8
Fig. 14. Rectification plots of the (a) conversion and (b) selectivity for the
decomposition of BF at various space velocities in He ( ) and 35% H2 in He ( ),
at different temperatures ( ), and for different partial pressures of BF ( ), CO ( ),
(b)
(c)
X = -0.2
and H2
and 3.
( ). The values of the parameters used in the model are given in Tables 2
X = -0.01
formate, which gives a CO2:CO ratio equal to 1:10. This value agrees
favorably with the experimental value of 8.5% selectivity observed
at low space velocity. Additionally, Fig. 12b shows that the ratio of
the primary decarbonylation reaction was more than 10,000 times
greater than the water–gas shift rate throughout the reactor.
ln(PCO) (atm)
-2
-1
0
-6.4
-6.8
-7.2
-7.6
-8
4. Conclusions
X = 0.7
X = 0.2
Palladium supported on carbon was found to be an active, selec-
tive, and stable catalyst for the liquid-phase conversion of formic
acid and butyl formate. Studies at different space velocities showed
that formic acid decomposes primarily through a decarboxylation
pathway (leading to CO2 and H2), whereas butyl formate primarily
decomposes through a decarbonylation pathway (leading to CO
and butanol). In the presence of water, the formate ester undergoes
hydrolysis to form formic acid, which increases the selectivity to-
ward the production of CO2 and H2. When excess water is present,
CO2 and H2 can be formed with almost quantitative selectivities
from butyl formate due to hydrolysis leading to the decomposition
of formic acid rather than butyl formate and due to increased
water–gas shift activity.
ln(PBF) (atm)
Fig. 13. TOF for the decomposition of BF versus partial pressures of (a) H2, (b) CO,
and (c) BF. Experimental data are given by solid symbols: TOF for decarboxylation
(j) and TOF for decarbonylation (Ç). Solid lines show predictions of the kinetic
model. Values of X are reaction orders calculated from experimental data.
Acknowledgments
experimentally observed trends for the conversion and selectivity
for the decomposition of BF versus space velocity, temperature,
and the partial pressures of H2, CO, and BF. The model demon-
strated that the true selectivity of the primary pathway for butyl
formate can be found at low space velocity, where the effect of
the formic acid impurity is diminished and where forward and re-
verse WGS are negligible. It was possible with the model to separate
the effect that the formic acid impurity had on the CO2:CO ratio at
infinite space velocity. From the model, it is predicted that the
selectivity for the production of CO2 from 2 M butyl formate in
1-butanol at 423 K is equal to 9.0% for the decomposition of butyl
This material is based upon work supported as part of the Insti-
tute for Atom-efficient Chemical Transformations (IACT), an Energy
Frontier Research Center funded by the US Department of Energy,
Office of Science, Office of Basic Energy Sciences and part of DOE
Great Lakes Bioenergy Research Center (www.greatlakesbioener-
gy.org), which is supported by the US Department of Energy, Office
of Science, Office of Biological and Environmental Research,
through Cooperative Agreement DE-FC02-07ER64494 between
The Board of Regents of the University of Wisconsin System and
the US Department of Energy.

B.J. O’Neill et al. / Journal of Catalysis 290 (2012) 193–201
[29] S.D. Senanayake, D.R. Mullins, J. Phys. Chem. C 112 (2008) 9744.
201
References
[30] X.D. Peng, M.A. Barteau, Langmuir 7 (1991) 1426.
[31] A. Koos, F. Solymosi, Catal. Lett. 138 (2010) 23.
[32] D.W. Flaherty, S.P. Berglund, C.B. Mullins, J. Catal. 269 (2010) 33.
[33] J. McCarty, R.J. Madix, J. Catal. 38 (1975) 402.
[34] G. Jacobs, P.M. Patterson, U.M. Graham, A.C. Crawford, B.H. Davis, Int. J.
Hydrogen Energy 30 (2005) 1265.
[1] A. Corma, S. Iborra, A. Velty, Chem. Rev. 107 (2007) 2411.
[2] G.W. Huber, S. Iborra, A. Corma, Chem. Rev. 106 (2006) 4044.
[3] J.C. Serrano Ruiz, D. Braden, R.M. West, J.A. Dumesic, Appl. Catal., B 100 (2010)
184.
[4] E.I. Gürbüz, D.M. Alonso, J.Q. Bond, J.A. Dumesic, ChemSusChem 4 (2011) 357.
[5] D.J. Braden, C.A. Henao, J. Heltzel, C.C. Maravelias, J.A. Dumesic, Green Chem.
13 (2011) 1755.
[6] P. Sabatier, Ber. Deutsch. Chem. Gesellsahft 44 (1911).
[7] E. Iglesia, M. Boudart, J. Catal. 81 (1983) 214.
[8] E. Iglesia, M. Boudart, J. Catal 81 (1983) 224.
[9] E. Iglesia, M. Boudart, J. Phys. Chem. 95 (1991) 7011.
[10] E. Iglesia, M. Boudart, J. Phys. Chem. 90 (1986) 5272.
[11] S.D. Senanayake, D. Stacchiola, P. Liu, C.B. Mullins, J. Hrbek, J.A. Rodriguez, J.
Phys. Chem. C 113 (2009) 19536.
[12] D.A. Bulushev, S. Beloshapkin, J.R.H. Ross, Catal. Today 154 (2010) 7.
[13] D.A. Bulushev, J.R.H. Ross, Catal. Today 163 (2010) 42.
[14] M. Ojeda, E. Iglesia, Angew. Chem. Int. Ed. 48 (2009) 4800.
[15] Y.-K. Sun, W.H. Weinberg, J. Chem. Phys. 94 (1991) 4587.
[16] M.R. Columbia, A.M. Crabtree, P.A. Thiel, J. Am. Chem. Soc. 114 (1992) 1231.
[17] M. Bowker, R.J. Madix, Surf. Sci. 102 (1981) 542.
[18] M.A. Barteau, M. Bowker, R.J. Madix, Surf. Sci. 94 (1980) 303.
[19] J.B. Benziger, G.R. Schoofs, J. Phys. Chem. 88 (1984) 4439.
[20] J.L. Falconer, J.G. McCarty, R.J. Madix, Surf. Sci. 42 (1974) 329.
[21] S.W. Jorgensen, R.J. Madix, J. Am. Chem. Soc. 110 (1988) 397.
[22] D. Sander, W. Erley, J. Vac. Sci. Technol. A 8 (1990) 3357.
[23] R.J. Madix, J.L. Gland, G.E. Mitchel, B.A. Sexton, Surf. Sci. (1983) 481.
[24] J. Kubota, A. Bandara, A. Wada, K. Domen, C. Hirose, Surf. Sci. 368 (1996) 361.
[25] F. Solymosi, Á. Koós, N. Liliom, I. Ugrai, J. Catal. 279 (2011) 213.
[26] P.A. Dilara, J.M. Vohs, J. Phys. Chem. 97 (1993) 12919.
[27] R. Larrson, M.H. Jamroz, M.A. Borowiak, J. Mol. Catal. A: Chem. 129 (1998) 41.
[28] M. Ai, J. Catal. 50 (1977) 291.
[35] F. Solymosi, A. Erdohelyi, J. Catal 91 (1985) 327.
[36] N. Akiya, P.E. Savage, AIChE J. 44 (1998) 405.
[37] H.F. Wang, Z.-P. Liu, J. Phys. Chem. C 113 (2009) 17502.
[38] A. Boddien, B. Loges, H. Junge, F. Gartner, J.R. Noyes, M. Beller, Adv. Synth.
Catal. 351 (2009) 2517.
[39] A. Majewski, D.J. Morris, K. Kendall, M. Wills, ChemSusChem 3 (2010) 431.
[40] C. Fellay, N. Yan, P.J. Dyson, G. Laurenczy, Chem. Eur. J. 15 (2009) 3752.
[41] C. Fellay, P.J. Dyson, G. Laurenczy, Angew. Chem. Int. Ed. 47 (2008) 3966.
[42] W. Gan, P.J. Dyson, G. Laurenczy, React. Kinet. Catal. Lett. 98 (2009) 205.
[43] Y. Gao, J.K. Kuncheria, H.A. Jenkins, R.J. Puddephatt, G.P.A. Yap, J. Chem. Soc.,
Dalton Trans. (2000) 3212.
[44] T. Moriyoshi, S. Keisuke, Y. Uosaki, High Press. Res. 20 (2001) 491.
[45] W.K. Metcalfe, J.M. Simmie, H.J. Curran, J. Phys. Chem. A 114 (2010) 5478.
[46] Y. Yasaka, K. Yoshida, C. Wakai, N. Matubayasi, M. Nakahara, J. Phys. Chem. A
110 (2006) 11082.
[47] G.W. Huber, J.N. Chheda, C.J. Barret, J.A. Dumesic, Science 308 (2005) 1446.
[48] R.D. Cortright, R.R. Davda, J.A. Dumesic, Nature 418 (2002) 964.
[49] S. Mukkavilli, C.V. Wittmann, L.L. Tavlarides, Ind. Eng. Chem. Process Des. Dev.
25 (1986) 487.
[50] O. Jogunola, T. Salmi, K. Eranen, J.-P. Mikkola, Appl. Catal. A (2008).
[51] L.C. Grabow, A.A. Gokhale, S.T. Evans, J.A. Dumesic, M. Mavrikakis, J. Phys.
Chem. C 112 (2008) 4608.
[52] D.D. Mehta, M.C. Akatay, E.A. Stach, W.N. Delgass, F.H. Ribeiro, Kinetics of the
water–gas shift reaction in the liquid phase, in: AIChE Annual Meeting,
Minneapolis, MN, October 19, 2011.