Formic acid decomposition over palladium based catalysts doped by potassium carbonate

Article abstract of DOI:10.1016/j.cattod.2015.04.008

The introduction of potassium carbonate into Pd/Al₂O₃, Pd/SiO₂and Pd/C catalysts promoted both the catalytic activities and the hydrogen selectivities for the vapor-phase formic acid decomposition, giving values of the tur

Full text of DOI:10.1016/j.cattod.2015.04.008

G Model
CATTOD-9569; No. of Pages7
ARTICLE IN PRESS
Catalysis Today xxx (2015) xxx–xxx
Contents lists available at ScienceDirect
Catalysis Today
journal homepage: www.elsevier.com/locate/cattod
Formic acid decomposition over palladium based catalysts doped by
potassium carbonate
Lijun Jiaa,b, Dmitri A. Bulusheva,c,d,∗, Julian R.H. Rossa
a
Chemical & Environmental Sciences Department, University of Limerick, Limerick, Ireland
Materials & Surface Science Institute, University of Limerick, Limerick, Ireland
Boreskov Institute of Catalysis SB RAS, Novosibirsk 630090, Russia
Novosibirsk State University, Novosibirsk 630090, Russia
b
c
d
a r t i c l e i n f o
a b s t r a c t
Article history:
The introduction of potassium carbonate into Pd/Al O , Pd/SiO and Pd/C catalysts promoted both the
2
3
2
Received 12 November 2014
Received in revised form 23 March 2015
Accepted 6 April 2015
catalytic activities and the hydrogen selectivities for the vapor-phase formic acid decomposition, giv-
ing values of the turnover frequency (TOF) at 343 K that were 8–33 times higher than those for the
undoped samples. The apparent activation energies over all the K-doped samples increased consider-
ably, this showing that there is a difference in the reaction path between the doped and the undoped
catalysts. Diffuse Reflectance Infrared Fourier Transform Spectroscopy (DRIFTS) has been used to gain
an understanding of the nature of the species formed in the Pd/SiO2 catalysts during the reaction. This
study showed that a considerable fraction of the HCOOH was condensed in the pores of the catalysts and
that the introduction of potassium contributed to the formation of buffer-like solution. The existence of
mobile formate ions present in the buffer solution and stabilized by K ions in a K-doped catalyst is an
essential factor in the promotion of its activity.
Available online xxx
This paper is dedicated to the memory of
Alfred Ross (1901–1948, father of Julian
Ross) Who carried out some of the earliest
work on infrared spectroscopy (A. Ross,
“
Absorption Spectra of Pyrone Derivatives
in the Near Infra-Red”, Proc. Royal Soc. A,
13 (1926) 208–220) and who inspired
1
©
2015 Elsevier B.V. All rights reserved.
Julian’s interest in spectroscopic methods.
Keywords:
Hydrogen production
Decomposition of formic acid
K-doped Pd catalysts
Buffer solution
DRIFTS
1
. Introduction
With the current increase in interest in the production of lev-
ulinic acid and other valuable chemicals from biomass, it becomes
important to develop ways of using the by-product formic acid as
this is otherwise a waste material. For many years, the decomposi-
tion of formic acid according to Eqs. (2) and (3):
Formic acid is a relatively low-volume chemical and it has
limited uses; these include its application as a preservative and
antifungal agent as well as in the production of leather and as a lime
scale remover. It may be produced by chemical methods such as
the hydrolysis of methyl formate but it is also produced in equimo-
lar proportions, together with levulinic acid, by the hydrolysis of
biomass-derived cellulosic raw materials [1]:
HCOOH(g) → CO + H2
(2)
(3)
2
HCOOH(g) → CO + H O
2
was used as a test reaction to distinguish between catalysts with
dehydrogenation (Eq. 2) and dehydration (Eq. 3) properties [2,3].
More recently, however, there has been considerable interest in
the use of the decomposition reaction as a means of producing pure
hydrogen for use, for example, in fuel cell applications; in this work,
the main aims are to produce predominantly hydrogen and CO2,
with minimal CO contents, at as low temperature as possible [4–9].
In our initial work, we showed that a commercial Pd/C catalyst gave
100% conversion of formic acid vapor at a temperature of 433 K with
(
1)
∗ Corresponding author at: Chemical & Environmental Sciences Department, Uni-
versity of Limerick, Limerick, Ireland. fax: +353 61202568.
E-mail addresses: dmitri.bulushev@ul.ie, dmitri.bulushev@gmail.com
D.A. Bulushev).
(
http://dx.doi.org/10.1016/j.cattod.2015.04.008
920-5861/© 2015 Elsevier B.V. All rights reserved.
0
Please cite this article in press as: L. Jia, et al., Formic acid decomposition over palladium based catalysts doped by potassium carbonate,
Catal. Today (2015), http://dx.doi.org/10.1016/j.cattod.2015.04.008

G Model
CATTOD-9569; No. of Pages7
ARTICLE IN PRESS
L. Jia et al. / Catalysis Today xxx (2015) xxx–xxx
2
selectivity to hydrogen and CO2 of 95–99% [10]. We also showed
that it was possible to use the same catalyst to hydrogenate an
olefin with formic acid stoichiometrically at the same temperature
in the He flow and single beam scanning was carried out at several
temperatures to provide background spectra for the subsequent in-
situ catalytic measurements at the corresponding temperatures. A
2 vol.% formic acid/He mixture was then introduced into each cat-
alyst sample at 333 K and absorbance spectra were recorded as a
function of time until stable spectra were obtained (30 min). Finally,
the formic acid containing feed gas was replaced by pure He and the
temperature was then increased. The corresponding absorbance
spectrum at each temperature was recorded until it did not change
with time.
[
11].
More recently, we have been probably first to demonstrate that
the rate of formic acid vapor conversion over a Pd/C catalysts could
be increased very significantly by the introduction of alkali metal
ions into the catalyst and that the hydrogen and CO2 selectivities
were also increased very significantly by the alkali promotion
[
6,12]. We presented evidence showing that the promotional effect
in K-promoted Pd/C catalysts is due to the presence in the pores of
the support of a buffer solution consisting largely of potassium for-
mate and formic acid [6]. We suggested that the rate-determining
step in the decomposition reaction is the decomposition of formate
anions at the surface of the Pd crystallites; as soon as a formate
anion decomposes, it is replaced by the dissociation of a formic
acid molecule, the concentration of the latter being replenished
constantly from the gas phase. Evidence for this mechanism was
obtained by careful observation of changes in the gas phase com-
position occurring during the early stages of the reaction prior to
the establishment of steady-state behavior. Despite attempts to
observe the reacting species using infrared techniques, we were
unable to do this because of the lack of transparency to IR radiation
of the carbon support.
The Brunauer–Emmet–Teller (BET) surface areas of all the sam-
ples after pretreatment in a flow of nitrogen at 473 K for 2 h
were measured by nitrogen adsorption using a Micromeritics Gem-
ini system. Transmission electron microscopy (TEM) images were
obtained for the reduced catalysts with a JEOL JEM-2100F (200 kV)
microscope.
3. Results
3.1. Characterization of catalysts
Table 1 shows the BET surface areas and particle sizes of the
catalysts studied. The BET surface area of the Pd/C sample was
2
−1
933 m g , a factor of 2.5 times higher than that of the Pd/SiO2
The aim of the work presented in this paper was initially to
examine whether or not palladium supported on an infrared-
transparent material such as silica or alumina is an effective catalyst
for the decomposition of formic acid and whether or not such mate-
sample and 5 times higher than that of the Pd/Al O3 sample. We
2
showed in our previous work [6,12] using a C support that a very
high potassium content corresponding to a weight ratio of 10:1 of
K:Pd gave the most active catalyst with even distribution of K ions
through the sample. On the basis of the BET surface areas of the
+
rials can then be promoted by the addition of K ions. Then, having
shown that the Pd/SiO catalyst gave significant rates of formic acid
Pd/SiO and Pd/Al O samples, we therefore chose to prepare sam-
2
2 2 3
+
decomposition with good selectivity to hydrogen and that K also
ples with weight ratio values of 4:1 for the K–Pd/SiO2 and 2:1 for
K–Pd/Al O to ensure that the potassium ions were also evenly dis-
gave significant promotion with this material, further experiments
using diffuse reflectance infrared Fourier transform spectroscopy
2
3
persed over the whole surface of these samples; no further attempts
were made to optimize the K/Pd ratios. The average Pd particle
size obtained from TEM images for the silica-supported catalyst
is around 7.4 nm, which is larger than those found for the sam-
ples supported on alumina (4.2 nm) and activated carbon (3.6 nm)
shown in Table 1.
(
DRIFTS) were carried out and the results, which support the mech-
anism previously postulated, are also presented. The value of the
use of in-situ infrared spectroscopy has previously been demon-
strated for formic acid interaction with supported metals [9,13,14].
2
. Experimental
3.2. Catalytic activity data
The unpromoted catalysts used in this work were supplied
Fig. 1(a) shows the results of activity tests for formic acid
by Johnson Matthey (1.0 wt.% Pd/SiO , 1.0 wt.% Pd/Al O ) and
2
2
3
decomposition over the carbon-, silica- and alumina-supported Pd
catalysts as well as the corresponding data for the K-promoted
samples while Fig. 1(b) shows the values of the corresponding
Sigma–Aldrich (1.0 wt.% Pd/C). The incipient wetness impregnation
method was used to deposit potassium carbonate on each of these
Pd-containing samples [6,12]. The measurements of catalytic activ-
ities for formic acid decomposition in the vapor phase were carried
out in a fixed-bed flow reactor. The weight of the catalysts was
chosen to give 0.68 mg of Pd for each set of experiments. All the
selectivities to hydrogen. The catalytic behavior of the Pd/SiO sam-
2
ple was close to that of the Pd/C sample, the conversions for the
former sample being only slightly lower than those for the latter at
all temperatures; the reaction temperatures required to give 50%
conversion of the formic acid over Pd/SiO2 and Pd/C were both
samples were reduced in 1 vol.% H /Ar at 573 K for 1 h and cooled
2
in He to the reaction temperature prior to testing. A mixture of
around 385 K. The Pd/Al O3 sample was significantly less active
2
2
vol.% formic acid in He at a total flow rate of 51 cm3 (STP) min
−1
than the other two samples; the temperature for 50% conversion
was approximately 30 K higher. The differences in selectivities were
even more marked: the hydrogen selectivity remained above 92%
was then introduced into the reactor system using a syringe pump
Sage). The reactants and products were analyzed by a gas chro-
(
matograph (HP-5890) fitted with a Porapak-Q column and a TCD
detector. The details of the activity tests carried out were given in
Refs. [7,10].
Table 1
Characteristics and kinetic data for the catalysts studied.
A SpectraTech-0030 DRIFTS “in situ” cell was used for the
infrared measurements, this being fitted with ZnSe windows. The
cell was mounted in a Nicolet Magna 560 FT-IR spectrometer with
an MCT detector; it was attached to the same gas-flow system
used for the catalytic measurements and the same pretreatments,
flow rates and gas compositions were used. A sample of the cata-
lyst (Pd/SiO , 12.7 mg or K–Pd/SiO , 16.8 mg) to be examined was
Catalysts
BET surface
Mean particle
size (nm)
TOF at
Activation
2
−1
−1
343 K (s )
area (m
g
)
energy
−1
(kJ mol
)
1
1
1
wt.% Pd/C
0:1 K–Pd/C
wt.% Pd/SiO2
933
688
380
–
144
–
3.6 ± 1.7
3.7 ± 1.3
7.4 ± 2.4
–
0.008
0.27
0.0095
0.079
0.010
0.075
65
97
78
95
49
84
4:1 K–Pd/SiO2
2
2
1
2
wt.% Pd/Al2O3
:1 K–Pd/Al2O3
4.2 ± 1.0
introduced into the sample holder of the cell and it was reduced in a
–
flow of 1 vol.% H in Ar at 573 K for 1 h. The sample was then cooled
2
Please cite this article in press as: L. Jia, et al., Formic acid decomposition over palladium based catalysts doped by potassium carbonate,
Catal. Today (2015), http://dx.doi.org/10.1016/j.cattod.2015.04.008

G Model
CATTOD-9569; No. of Pages7
ARTICLE IN PRESS
L. Jia et al. / Catalysis Today xxx (2015) xxx–xxx
3
the range 84–97 kJ mol 1. We have previously reported a similar
increase of the measured activation energies as a result of K-doping
of an unsupported Pd powder [6]; this set of observations points
out that promotion with K ions affects the behavior of the Pd cat-
alysts mainly through Pd sites rather than through support sites.
Not only does the change in the values of Ea indicates that there is
a change in mechanism upon doping the samples with K but that
there is probably a similar promotional reaction mechanism for all
the K-doped catalysts.
−
3.3. DRIFTS measurements
In order to obtain further evidence for the mechanism proposed
previously for the K-promoted Pd/C samples [6], we have car-
ried out infrared (IR) spectroscopy measurements using a DRIFTS
cell under conditions closely resembling the normal reaction con-
ditions used in the flow reactor as described above. There are
restrictions as to the types of samples that can be used in DRIFTS
measurements. In particular, activated carbon supports of the type
used for the samples reported in our previous work [6,10,12] cannot
be used in DRIFTS measurements because of their total absorbance
of IR energy; this means that any reflected IR energy signal after
absorbance by the activated carbon is too weak to be detected.
For this reason, and because silica is often considered to have
similar properties to those of activated carbon, the Pd/SiO2 cata-
lysts discussed above were examined in this work. Figs. 2–4 show
the DRIFT spectra obtained for the experiments carried out with
the Pd/SiO2 and K–Pd/SiO2 samples. In the experiments shown
in Figs. 2 and 3, the formic acid was admitted at 333 K and the
sample was subsequently heated in He in the absence of formic
acid while in that shown in Fig. 4, the sample was heated in the
presence of formic acid. Each figure includes only those parts of
the spectra that show significant features, these appearing in the
Fig. 1. Temperature dependence of conversion (a) and hydrogen selectivity (b) in
the formic acid decomposition over undoped catalysts, Pd/Al2O3, Pd/SiO2, Pd/C, and
doped catalysts, 2:1 K–Pd/Al2O3, 4:1 K–Pd/SiO2, 10:1 K–Pd/C.
at all temperatures examined over both the Pd/SiO2 and Pd/C cat-
alysts while those for the Pd/Al O3 sample fell in the range of only
2
7
9–83%.
Similar time-dependent effects during the start-up phase as
−
1
regions 4000–2800 cm
(a region containing bands associated
those found with the K/Pd/C catalysts [6] were observed with the
K/Pd samples supported on silica and alumina examined here;
these effects have not been quantified. When the catalyst sam-
ples were doped with K, there was marked improvement in the
steady-state behavior over each of the materials. As reported pre-
viously, the temperature necessary to give 50% conversion for the
with hydroxyl groups and C–H bond vibrations; note that some
of these bands are negative and others are positive, as discussed
−
1
further below) and 2000–1400 cm (a region associated with C–O
vibrations in formic acid and formate species); in each set of spec-
−
1
tra, there are also small bands in the region 3000–2600 cm and
expanded versions of the spectra in this region are also given for
clarity. Table 2 summarizes the vibration frequencies and the attri-
bution of the bands observed; details of these attributions will be
discussed further below as appropriate.
1
0K–Pd/C sample is approximately 50 K lower and 100% conver-
sion was obtained at a temperature of about 350 K. With both the
silica and alumina supports, 50% conversion is now obtained at
about 360 K and 100% conversion also occurs at significantly lower
temperatures compared with the unpromoted materials; although
the alumina-supported material is not as effective as the silica-
supported one at higher temperatures, the improvement compared
with the unpromoted sample is much more significant. Similar
considerable improvements are found in the selectivities towards
hydrogen with all three samples: all give hydrogen selectivities
well above 96% for the whole range of temperature examined. The
improvement in the hydrogen selectivity is particularly marked for
the alumina-based sample. Generally, both the activity and hydro-
gen selectivity were improved by doping K ions into all the Pd
catalysts with different supports.
Table 1 includes the catalytic results obtained for the three dif-
ferent materials without and with K as promoter, showing the
values of the turnover frequency (TOF) calculated for each sample
at a temperature of 343 K (calculated on the basis of the metal parti-
cle size determined by TEM) as well as the activation energies (Ea)
calculated from the Arrhenius plots. As discussed above, the TOF
values show quite clearly the significant promotion by potassium
species for all three supports. Particularly noteworthy are the val-
ues of the activation energies; while for the unpromoted samples,
there are significant differences between the different samples, the
values for the K-promoted materials are all very similar, lying in
3.3.1. Interaction of HCOOH with the Pd/SiO2 catalyst
Fig. 2 shows the main IR bands observed when HCOOH was
introduced to the unpromoted Pd/SiO2 sample at 333 K as well
as the behavior of these bands after the sample had been heated
to 453 K after substituting the flow of HCOOH/He by pure He. In
−
1
−1
the spectral region from 4000 cm to 2800 cm , a negative band
occurs at ca 3740 cm
−
1
following admission of HCOOH at 333 K
and this is accompanied by a broad positive band centered at about
−
1
−1
3200 cm and a sharp band at 2938 cm . On removing the HCOOH
at 333 K, each of these bands decreases and almost disappears
on heating to temperatures of 393 K and above. As shown more
−
1
clearly in the expanded spectrum in the region 3000–2600 cm
,
−
1
the sharp band at 2938 cm formed in the presence of HCOOH
is reduced on heating till 453 K without formic acid but it does
not disappear completely. The same applies to the band appear-
ing at 1739 cm . A band is also found at 1578 cm , but it is very
weak.
−
1
−1
−
1
We attribute the negative band at 3740 cm
to the dis-
appearance of vibrations associated with at least some of the
isolated surface OH groups of the silica, these becoming associated
by hydrogen bonding to physically adsorbed/condensed HCOOH
Please cite this article in press as: L. Jia, et al., Formic acid decomposition over palladium based catalysts doped by potassium carbonate,
Catal. Today (2015), http://dx.doi.org/10.1016/j.cattod.2015.04.008

G Model
CATTOD-9569; No. of Pages7
ARTICLE IN PRESS
4
L. Jia et al. / Catalysis Today xxx (2015) xxx–xxx
in He
453 K
in He
93 K
3
in He
33 K
3
in HCOOH
33 K
3
4
000 3600 3200 2800 3000
2800
26002000 1800 1600 1400
-1
Wavenumber, cm
Fig. 2. DRIFT spectra of the Pd/SiO2 catalyst in HCOOH at 333 K and in He at different temperatures.
in He 493 K
in He 453 K
in He 393 K
in He 333 K
in HCOOH
333 K
4
000 3600 3200 28003000
2800
26002000 1800 1600 1400
-1
Wavenumber, cm
Fig. 3. DRIFT spectra of the 4:1 K–Pd/SiO2 catalyst in HCOOH at 333 K and in He at different temperatures.
species after the introduction of the HCOOH. The broad positive
The second set of spectra of Fig. 2 show that the bands at
−
1
−1 −1
1739 cm and 2943 cm as well as the broad band centered at
band centered at 3200 cm
is characteristic of the vibration of
−
1
hydrogen bonded OH groups in the support and condensed HCOOH.
The previous IR studies [9,13,14] have reported that physically
adsorbed/condensed hydrogen-bonded formic acid is generated
after the introduction of formic acid to catalysts and this suggestion
is consistent with the explanation given here.
3200 cm all decreased after switching off the HCOOH flow and
replacing it by a flow of pure He at 333 K. This result indicates that
gas phase HCOOH is in equilibrium with condensed HCOOH over
the Pd/SiO catalyst. Further desorption of condensed HCOOH took
2
place with an increase in temperature (top two spectra); however,
−
1
−1
The weak band at 2938 cm occurs in the C–H stretching range
the band at 1739 cm still remained even at a temperature as high
−1
and can be ascribed to the ꢀCH vibration of condensed HCOOH
as 453 K. Both the negative band at 3740 cm and the broad band
−
1
−1
molecules. The band with high intensity visible at 1739 cm
is
centered at 3200 cm disappeared with the increase in temper-
probably due to the C O stretching mode of HCOOH condensed
ature, this indicating that the desorption of HCOOH freed up the
isolated hydroxyl groups on the silica support.
−
1
on SiO2 while the weak and broad band at 1578 cm is probably
due to the asymmetric stretching vibration of a formate species
present in a very small concentrations. The bands of gas phase
3.3.2. Interaction of HCOOH with the 4K–Pd/SiO2 catalyst
The interaction of HCOOH with a sample of the K-doped Pd/SiO2
catalyst was carried out following the same procedure as that used
−1
formic acid at about 1790 cm contribute negligibly to the band of
−
1
condensed formic acid at 1739 cm as a shoulder on the left hand
side.
for the undoped Pd/SiO sample and the results are shown in Fig. 3.
2
Please cite this article in press as: L. Jia, et al., Formic acid decomposition over palladium based catalysts doped by potassium carbonate,
Catal. Today (2015), http://dx.doi.org/10.1016/j.cattod.2015.04.008

G Model
CATTOD-9569; No. of Pages7
ARTICLE IN PRESS
L. Jia et al. / Catalysis Today xxx (2015) xxx–xxx
5
in HCOOH
393 K
373 K
353 K
333 K
3000
2800
Wavenumber, cm
26002000 1800 1600 1400
4000 3600 3200 2800
-1
Fig. 4. DRIFT spectra of the 4:1 K–Pd/SiO2 catalyst in HCOOH at different temperatures.
−1
Table 2
−1
−1
Three bands, at 2916 cm , 2822 cm and 2726 cm , all cor-
responding to C–H stretching frequencies, appeared at 333 K on
Summary of the major vibration frequencies (in cm−1) observed following interac-
tion of the catalysts with HCOOH.
−
1
switching to He in place of the single band at 2937 cm . (The addi-
tional two bands may have been present with formic acid but, if so,
Assignment
Pd/SiO2
4:1 K–Pd/SiO2
−
1
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
2
(C O) formic acid (liquid)
(CH) formic acid (liquid)
as(OCO) formate ions
as(OCO) bulk formate
(CH) mainly bulk formate
ı(CH) mainly bulk formate
1739
2938
1578 (very weak)
Very weak
Not found
Not found
1729–1742
2935
1580–1590
1606–1611
2822, 2809
2716, 2700
were very indistinct.) The band at 2937 cm found in the pres-
ence of formic acid is probably a combination of bands due to C–H
stretching in the condensed HCOOH species as well as in formate
species; this band attenuated and shifted to 2916 after removing
the HCOOH from the feed gas. As noted above for the band occurring
−
1
−1
at about 1606 cm , these additional bands shifted to 2809 cm
−
1
and 2700 cm when the temperature was increased to 393 K.
As with the spectra obtained for Pd/SiO2 (Fig. 2), introduction of
HCOOH to the 4:1 K–Pd/SiO catalyst at 333 K gave rise to a sharp
These changes brought about by switching from a flow contain-
ing HCOOH to one of pure He and heating is possibly due to the
appearance of another form of formate species in addition to for-
2
1
−
negative band at 3740 cm due to the transformation of isolated
hydroxyl groups to hydrogen bonded species as well as to a broad
positive band at 3300 cm due to the OH vibration in the hydrogen
bonded hydroxyl groups of condensed HCOOH and of silica, both
bands increasing slightly with time (not shown in the figure).
The intensities of both the bands ascribed to the condensed
and 1740 cm ) were lower than those
observed with the undoped Pd/SiO . The most significant differ-
ence between the unpromoted and K-promoted samples is seen
in the region 2000–1400 cm with the appearance, in addition to
the band at 1740 cm , of a strong absorption at 1586 cm char-
acteristic of an asymmetric OCO vibration of formate species. This
band appears to be due to the presence of formate ions in a solution
of potassium formate in formic acid or water [8]. This observation
confirms the formation of formate anions on the K-promoted cata-
lyst, showing that quite a considerable proportion of the adsorbed
HCOOH dissociates into formate anions over the K–Pd/SiO sample
at 333 K in contrast to the undoped sample (Fig. 2).
Fig. 3 also shows spectra taken after switching from the flow
containing HCOOH to pure He at 333 K and increasing the tem-
perature to 493 K. The C O vibration band at 1740 cm
−
1
−
1
mate anion (1585 cm ) existing in a solution of KHCOO [8]. The
−
1
band at 1606 cm could be assigned to “bulk” formate (crystalline
or molten KHCOO). The bulk formate provides more intense C–H
stretching bands compared to those given by formate anions in
−
1
−1
−1
solution. The bulk formate band at 1606 cm is still observed in
the spectrum at 493 K indicating high stability and low reactivity of
this formate. The disappearance of the bands associated with con-
densed HCOOH at lower temperatures in the case of the K-doped
sample than in the case of the undoped material would appear to
be due to its involvement in the decomposition reaction and by the
much higher reactivity of the doped catalyst (Fig. 1).
HCOOH (2937 cm
2
−
1
−1
−1
3
4
.3.3. HCOOH decomposition at different temperatures over the
K–Pd/SiO2 catalyst
Fig. 4 shows equivalent DRIFTS results to those shown in Fig. 3
2
for the 4K–Pd/SiO2 catalyst but in which formic acid was not
removed from the gas flow when the temperature was raised. The
−
1
−1
bands at 2934 cm and 1740 cm , both of which are characteristic
of condensed HCOOH, decreased with increasing temperature, this
confirming the existence of an equilibrium with gas phase HCOOH.
These bands are still observed at 393 K in contrast to the data for
heating in He (Fig. 3). The bands assigned to formate anions and
bulk formate species increase with temperature.
−
1
due to
condensed HCOOH disappeared totally at 393 K and this con-
trasting with the result obtained with the undoped Pd/SiO catalyst
with which a small band was still found after heating to 453 K
Fig. 2). Both the negative band at 3740 cm and the broad band
2
−1
(
−1
centered at 3300 cm disappeared with increasing temperature.
−1
However, the formate band at 1586 cm discussed above shifted
4. Discussion
−1
−1
from 1586 cm to 1606 cm , growing slightly in magnitude, after
switching from the HCOOH to pure He on heating to 393 K; this
band decreased slightly in magnitude once more on heating the
sample further to 493 K.
The DRIFT spectra obtained during the HCOOH interaction with
the K-promoted Pd/SiO2 catalyst studied here show typical strong
bands due to the C O stretching frequencies of condensed formic
Please cite this article in press as: L. Jia, et al., Formic acid decomposition over palladium based catalysts doped by potassium carbonate,
Catal. Today (2015), http://dx.doi.org/10.1016/j.cattod.2015.04.008

G Model
CATTOD-9569; No. of Pages7
ARTICLE IN PRESS
6
L. Jia et al. / Catalysis Today xxx (2015) xxx–xxx
Fig. 5. Schematic representation of the K doped catalyst (a) in HCOOH and (b) in He (Pd particles are not shown).
acid together with broad bands attributable to OH species in
condensed HCOOH and support OH species hydrogen bonded to
HCOOH. The results above show that a significant proportion of
the HCOOH introduced is condensed on the catalysts, this being
consistent with the results published by the other research groups
condensed in the pores, bulk potassium formate (crystalline or
molten) exists in equilibrium with formate ions and K . In our pre-
+
vious publication [6], we speculated on the importance of the liquid
phase for the catalytic promotional effect as depicted in Fig. 6; this
is clearly confirmed by the DRIFTS results here. The concentration of
formate ions present in aqueous formic acid is negligible because
of its low dissociation constant. However, the presence of alkali
metal ions completely changes this situation, giving a significant
concentration of formate ions in a buffer-like solution consisting
of formate anions, K ions, protons and HCOOH explained by our
previous work [6].
[
9,13,14] and with the non-steady state results presented in our
previous paper [6]. We concluded from the latter results that a con-
siderable fraction of the HCOOH introduced to a fresh K-promoted
catalyst is condensed in the pores of the catalyst (Fig. 5a) and that
formate anions stabilized by K ions in solution are also formed.
It is notable that the peaks attributed to formate species became
even stronger after switching from HCOOH to He and that some
of the bands due to formate species were still discernible at tem-
peratures as high as 493 K. It can be suggested that two different
states of the formate species exist: mobile formate anions present
in liquid HCOOH; and bulk potassium formate species formed after
most of the HCOOH is desorbed from the catalyst pores (Fig. 5b) or
present even in the condensed liquid if potassium content is taken
in excess. The bands of typical OCO stretching vibration of formate
The dissociation of formate ions to give adsorbed hydrogen on
Pd and forming gaseous CO2 appears to be the rate determining
step in the all-over reaction as the reaction rate is dependent on
+
the concentration of K ions, this determining the concentration of
formate ions [6]. In the experiments reported here, after switching
from the HCOOH-containing flow to pure He, bulk formate crys-
tallized in the pores from the decreasing volume of liquid HCOOH
(Fig. 5b) and was stable at the high temperature of 493 K. The liq-
uid formic acid when present provides easy access of the formate
anions to all the surface Pd sites.
The presence of water is inevitable in the reaction system as
it could either be formed as a side product of formic acid dehy-
dration (with CO and water as products) or be introduced to the
system during the transformation of the potassium carbonate used
for impregnation into potassium formate via Reaction (4):
−1
−1
shifted from 1586 cm to 1606 cm after switching from HCOOH
in He to pure He and heating showing that there was a change in
the form of the formate species during this process. The existence
of a stable bulk formate at 493 K implied that it dissociates less eas-
ily and has decreased mobility over the surface in the absence of
condensed HCOOH.
The introduction of K into Pd based catalysts on different sup-
ports significantly accelerates the decomposition of formic acid.
The promoted reaction steps were represented in Fig. 6. In the
first step, a phase consisting of liquid formic acid is formed within
the pores of the catalyst and this phase then provides a reservoir
for the formation of formate anions with participation of K+ ions.
The formate anions then decompose to form CO2 and give gaseous
hydrogen via hydrogen species on the Pd surface. In the liquid
K CO
2
+ 2HCOOH(l) → 2KHCOO(aq) + H O + CO2(g)
(4)
3(aq)
2
(l)
The change of the standard pressure Gibbs free energy for this
−
1
reaction is −98 kJ mol at 343 K. Thermal conversion of formates
to carbonates in inert atmosphere normally takes place at temper-
atures higher than those used for the performed DRIFTS study [15].
This explains why carbonates are not observed in our conditions.
The introduction of extra water (2.5 vol.%) into the reactant flow
did not have a significant influence on the DRIFT spectra obtained
for the K-doped catalyst (see Fig. S1, Supplementary information
file); nor were substantial changes observed during the non-steady
state period (see Fig. S2). Balarew et al. [16] showed that potassium
formate dissolves easily in both formic acid or water. The final step
in the reaction scheme is the recombination of two atoms of hydro-
gen adsorbed on the Pd active sites to produce gaseous hydrogen.
This step is reversible.
Fig. 6. Schematic representation of the mechanism of the formic acid decomposition
on Pd catalysts doped with potassium ions. The characteristic IR bands positions are
indicated.
Please cite this article in press as: L. Jia, et al., Formic acid decomposition over palladium based catalysts doped by potassium carbonate,
Catal. Today (2015), http://dx.doi.org/10.1016/j.cattod.2015.04.008

G Model
CATTOD-9569; No. of Pages7
ARTICLE IN PRESS
L. Jia et al. / Catalysis Today xxx (2015) xxx–xxx
7
5
. Conclusions
Appendix A. Supplementary data
It can be concluded that the addition of potassium to a Pd/C
Supplementary data associated with this article can be found,
in the online version, at http://dx.doi.org/10.1016/j.cattod.2015.04.
008
catalyst has a remarkable promotional effect on the catalytic activ-
ity and hydrogen selectivity for formic acid decomposition, the
reaction rate being increased by 1–2 orders of magnitude. Having
shown that similar promotional effects were exhibited by K-doped
Pd/SiO2 and Pd/Al O materials, DRIFT spectra of the interaction
References
2
3
[
1] D.J. Hayes, S. Fitzpatrick, M.H.B. Hayes, J.R.H. Ross, in: B. Kamm, P.R. Gruber,
M. Kamm (Eds.), Biorefineries-Industrial Processes and Products, Vol. 1, Wiley-
VCH, Weinheim, 2006, p. 139.
of HCOOH with Pd catalysts supported on silica were obtained to
provide a greater understanding of the intermediate species par-
ticipating in the reaction. The DRIFTS results demonstrate that a
considerable amount of HCOOH was condensed on the catalyst
and the presence of potassium over doped catalyst contributed
to the formation of a buffer-like solution, this giving rise to the
high concentration of formate ions in the reaction system. The
DRIFTS results confirmed that the liquid phase of HCOOH and
mobile formate ions are two essential features with respect to the
promotion of the catalysts by providing formate ions to the Pd
sites.
[2] J. Fahrenfort, L.L. van Reyen, W.M.H. Sachtler, in: J.H. de Boer (Ed.), Proceedings
of the Symposium on the Mechanism of Heterogeneous Catalysis, Elsevier,
Amsterdam/London/New York/Princeton, 1960, p. 23.
[
3] P. Sabatier, A. Mailhe, Compt. Rend. 152 (1911) 1212.
[4] S. Enthaler, J. von Langermann, T. Schmidt, Energy Environ. Sci. 3 (2010) 1207.
[5] M. Grasemann, G. Laurenczy, Energy Environ. Sci. 5 (2012) 8171.
[
6] L. Jia, D.A. Bulushev, S. Beloshapkin, J.R.H. Ross, Appl. Catal. B 160–161 (2014)
5.
3
[
7] L.J. Jia, D.A. Bulushev, O.Y. Podyacheva, A.I. Boronin, L.S. Kibis, E.Y.
Gerasimov, S. Beloshapkin, I.A. Seryak, Z.R. Ismagilov, J.R.H. Ross, J. Catal. 307
(
2013) 94.
[
[
8] K. Jiang, K. Xu, S.Z. Zou, W.B. Cai, J. Am. Chem. Soc. 136 (2014) 4861.
9] F. Solymosi, Á. Koós, N. Liliom, I. Ugrai, J. Catal. 279 (2011) 213.
Acknowledgements
[10] D.A. Bulushev, S. Beloshapkin, J.R.H. Ross, Catal. Today 154 (2010) 7.
[
[
[
11] D.A. Bulushev, J.R.H. Ross, Catal. Today 163 (2011) 42.
12] D.A. Bulushev, L. Jia, S. Beloshapkin, J.R.H. Ross, Chem. Commun. 48 (2012) 4184.
13] G.C. Cabilla, A.L. Bonivardi, M.A. Baltanás, Appl. Catal. A 255 (2003) 181.
This publication has emanated from research conducted with
the financial support of Science Foundation Ireland under Grant
Number 06/CP/E007. We highly appreciate the assistance of Dr.
Yina Guo (MSSI, University of Limerick) with TEM measure-
ments.
[14] G.J. Millar, C.H. Rochester, K.C. Waugh, J. Catal. 155 (1995) 52.
[
[
15] T. Meisel, Z. Halmos, K. Seybold, E. Pungor, J. Therm. Anal. 7 (1975) 73.
16] C. Balarew, T.P. Dirkse, O.A. Golubchikov, M. Salomon, S. Trendafilova,
S. Tepavitcharova, T. Ageyeva, P. Baldini, G. D’Andrea, J. Phys. Chem. Ref. Data
30 (2001) 1.
Please cite this article in press as: L. Jia, et al., Formic acid decomposition over palladium based catalysts doped by potassium carbonate,
Catal. Today (2015), http://dx.doi.org/10.1016/j.cattod.2015.04.008