Formic Acid Decomposition over ZnPd—Implications for Methanol Steam Reforming

Article abstract of DOI:10.1002/cctc.201800194

Single-phase ZnPd with different elemental composition as well as ZnO-supported ZnPd nanoparticles were prepared and tested concerning their catalytic properties in the formic acid decomposition with the aim to investigate a possible formate pathway in th

Full text of DOI:10.1002/cctc.201800194

Accepted Article
Title: Formic Acid Decomposition over ZnPd - Implications for Methanol
Steam Reforming
Authors: René Kriegel, Dennis Ivarsson, and Marc Armbrüster
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: ChemCatChem 10.1002/cctc.201800194
Link to VoR: http://dx.doi.org/10.1002/cctc.201800194
A Journal of
www.chemcatchem.org

10.1002/cctc.201800194
ChemCatChem
FULL PAPER
Formic Acid Decomposition over ZnPd – Implications for
Methanol Steam Reforming
René Kriegel,[a] Dennis C. A. Ivarsson,[a] and Marc Armbrüster*[a]
Abstract: Single-phase ZnPd with different elemental composition as
well as ZnO-supported ZnPd nanoparticles were prepared and tested
concerning their catalytic properties in the formic acid decomposition
with the aim to investigate a possible formate pathway in the steam
reforming of methanol (MSR). Pd-rich ZnPd showed higher catalytic
activity in comparison to ZnPd with lower Pd-content. All samples
showed high and stable CO2-selectivity. The stability of the materials
was investigated both ex situ by powder X-ray diffraction (XRD),
scanning electron microscopy (SEM) and Raman spectroscopy after
formic acid decomposition as well as by in situ X-ray photoelectron
spectroscopy (in situ XPS) and operando DTA/TG and resulted in
higher stability against oxidation for samples with higher Pd-content,
corroborating earlier observations under MSR conditions. The
supported ZnPd/ZnO material and Zn-rich bulk ZnPd samples showed
a strong modification after reaction, which is attributed to zinc formate
formation during formic acid decomposition. Kinetic data give strong
indications to exclude dehydrogenation of a formate intermediate as
rate limiting step in MSR.
chemical potential (in particular when starting from equimolar
chemical composition).[7] While these studies shed light on the
material changes under reaction conditions, the reaction
mechanism of methanol steam reforming is far from being
completely understood.
To increase the knowledge on the MSR reaction mechanism,
numerous quantum chemical calculations on the catalytic
properties of ZnPd were conducted regarding possible
intermediates.[8] Scheme 1 summarizes the proposed reaction
sequence on different ZnPd surfaces, indicating formaldehyde as
a key intermediate in MSR.
Scheme 1. Proposed reaction pathways for MSR on different ZnPd surfaces
based on quantum chemical calculations. The lowest activation energies in kJ
mol-1 and the involved surfaces for the reaction steps are given in red.[8b, 8d, 8f]
Introduction
Methanol is a promising molecule to serve as hydrogen carrier.[1]
Via methanol steam reforming (MSR, CH3OH + H2O → 3 H2
+
C-H splitting of the formaldehyde intermediate results in
unwanted methanol decomposition and therefore in CO formation.
The dehydrogenation of formaldehyde is kinetically unfavorable
on ZnPd surfaces, unless water activation for MSR is not hindered
sufficient enough.[8b] Experiments as well as quantum chemical
CO2) the chemically stored hydrogen can be easily released on
demand for usage in fuel cell applications.[2] Here, the production
of high purity hydrogen, i.e. avoiding CO contamination is crucial,
since for a long-term stable performance of proton exchange
membrane fuel cells (PEMFCs) CO concentrations should not
exceed 10 ppm.[3]
calculations provide indications for improved water activation by
9]
the presence of ZnO.[6,
This circumstance explains the
The intermetallic compound ZnPd has been shown to be a
promising catalytic material for MSR because of its high CO2-
selectivity of up to 99.4% and thermal stability.[4] Studies on
unsupported ZnPd revealed that the coexistence of ZnPd and
ZnO is necessary for the high CO2-selectivity,[5] whereas single-
phase ZnPd preferentially catalyzes methanol decomposition.[6]
The high CO2-selectivity of unsupported ZnPd results from the
formation of an oxidized Zn-species under MSR conditions, which
leads to the highly selective ZnPd/ZnO interface.[6] The strong
compositional dependence of the catalytic properties results from
the change in chemical potential with the chemical composition of
ZnPd. In this respect, even a small change in the chemical
composition of ZnPd can result in a significant change of the
increasing CO2-selectivity by presence of ZnO, e.g. on partially
oxidized ZnPd surfaces.[5]
To optimize future catalytic materials, the question concerning
the rate limiting step (RLS) in MSR has been addressed by DFT
calculations taking into account the presence of methanol and
water simultaneously on ZnPd(111). The calculated activation
energy of 86 kJ mol-1 for the formaldehyde formation as RLS[8d] is
significantly higher than the experimental apparent activation
energy, reported to be 69 kJ mol-1.[6] This implies that the
calculations on ZnPd(111) do not result in an overall picture of the
underlying mechanism. Considering also other investigated
surfaces, the abstraction of the last hydrogen from a formate
intermediate becomes a likely candidate for the rate limiting step
with an activation energy of 69 kJ mol-1.[10] This assumption gets
encouraged by experimental studies assigning the formate
species as key surface intermediate.[11]
[a]
M. Sc. R. Kriegel, M. Sc. D.C.A. Ivarsson, Prof. Dr. M. Armbrüster
Institute of Chemistry
Technische Universität Chemnitz
Straße der Nationen 62, 09111 Chemnitz (Germany)
E-mail: marc.armbruester@chemie.tu-chemnitz.de
Formic acid decomposition presents a suitable model reaction
to investigate a small part of the likely MSR reaction mechanism.
Formic acid, as precursor for formate, represents a test molecule,
where the MSR reagents methanol and water are already
combined. In addition, formic acid decomposition represents a
Supporting information for this article is given via a link at the end of
the document.
This article is protected by copyright. All rights reserved.

10.1002/cctc.201800194
ChemCatChem
FULL PAPER
common used model reaction to study the catalytic activity of
metals and alloys[12] as well as intermetallic compounds,[13] thus
enabling further understanding of ZnPd by comparison. Lorenz et
al. investigated the decomposition of formic acid on intermetallic
GaPd2/β-Ga2O3 and single-phase β-Ga2O3.[14] Comparing the
results to the data from MSR tests on the same material, they
in hydrogen at 200 °C for 1 h to first reduce palladium oxide to
palladium and subsequently partly reduce the ZnO in the vicinity
of the palladium particles to elemental zinc. This then diffuses into
the palladium and forms the intermetallic compound ZnPd. As
expected, the obtained X-ray diffraction pattern results in the
observation of ZnPd, ZnO and Pd (Figure S2). The latter is only
observed as trace – consistent with earlier publication – and is
expected to be encapsulated in ZnO, according to high resolution
transmission electron microscopy (HRTEM).[5] Thus Pd should
not contribute to the catalytic properties. The lower Pd-loading of
the supported sample, determined by ICP/OES to be 7.8(3) at.-%
compared to the nominal loading of 9.7 at.-%, is likely due to the
unknown amount of water in the used Pd(NO3)2 ∙ xH2O.
assumed that the dehydrogenation of
a formate species
represents the last step in MSR mechanism. Furthermore, they
supposed that the dehydrogenation preferably takes place on the
intermetallic compound. Jeroro and Vohs studied the
decomposition of formic acid on Zn/Pd(111) surface alloys under
ultra-high vacuum (UHV) conditions.[15] Using temperature-
programmed desorption (TPD) and high-resolution electron
energy loss spectroscopy they were able to detect the formation
of a formate intermediate on the surface. This observation affirms
the usability of formic acid for investigations of formate
dehydrogenation.
Catalysis and ex situ Characterization of Bulk ZnPd
The aim of the present work is to enlarge the understanding
of MSR on intermetallic compounds. Therefore, unsupported
single-phase ZnPd as well as ZnO-supported ZnPd particles were
prepared and tested in the formic acid decomposition in a plug-
flow reactor. Owing to the different catalytic behavior of ZnPd with
chemical composition,[6] three samples of bulk Zn100-xPdx (x =
48.1-58.0) were tested. Before the catalytic measurements, bulk
characterization by powder X-ray diffraction (XRD) and ICP-OES
was carried out. SEM and XRD were applied to investigate the
samples additionally after formic acid decomposition. Activity,
selectivity and apparent activation energy were determined for the
three samples.
To achieve direct comparability of the catalytic formic acid
decomposition on ZnPd with different chemical compositions, the
tests were conducted with identical reaction parameters and
identical masses of 50 mg. Figure 1 summarizes the obtained
conversion and CO2-selectivity of the bulk ZnPd samples as a
function of temperature.
Results and Discussion
Synthesis and Characterization
According to the latest phase diagram the intermetallic compound
ZnPd exists between 47 and 62.5 at.-% Pd at 900 °C.[10] Thus the
nominally compositions of the samples were chosen as 48.1 at.-%,
50.1 at.-% and 58.0 at.-% Pd to allow for the formation of single-
phase ZnPd. No phase besides ZnPd could be detected by XRD
(Figure S1). The observed shift of the reflections results from the
different compositions of the samples, which in turn results in a
change of the lattice parameters.[6] ICP-OES analysis yielded Pd-
contents of 46.7(2) at.-%, 49.1(1) at.-% and 57.1(2) at.-% for the
respective ZnPd samples. Since chemical analyses and expected
chemical compositions differed slightly, compositions were also
determined by weighing the samples before and after synthesis.
Because no reaction with the ampoule material was observed, the
difference between the initial and the end mass should clarify if all
provided Zn reacted with Pd during synthesis. By weighing the
nominal compositions of bulk ZnPd samples could be confirmed
(48.1 at.-%, 50.2 at.-% and 58.0 at.-% Pd and a standard
deviation of less than 0.01 at.-%). Therefore, hereafter the
chemical composition determined by weighing will be used.
The pre-calcined PdO/ZnO was used as precursor to prepare
ZnO-supported ZnPd particles by reactive metal-support
interaction (RMSI).[16] The precursor system was directly reduced
Figure 1. Conversion and CO2-Selectivity of bulk Zn100-xPdx (50 mg) in formic
acid decomposition. Lines are a guide to the eyes.
An obvious activity difference between the Pd-rich ZnPd and
the samples with lower Pd-content can be observed. For a
quantitative comparison the activities at 270 °C are considered in
the following. Zn51.9Pd48.1 (A = 21 mmolCOx mmolPd h-1) and
-1
This article is protected by copyright. All rights reserved.

10.1002/cctc.201800194
ChemCatChem
FULL PAPER
Zn49.8Pd50.2 (A = 25 mmolCOx mmolPd h-1) differ only by 19% in
activity (Zn49.8Pd50.2 Zn51.9Pd48.1), whereas Zn42.0Pd58.0
value roughly 4-times higher (A = 85 mmolCOx
acid decomposition. This alteration must involve consumption of
hydrogen.
-1
>
exhibited
a
X-ray diffraction patterns of the catalyst samples after formic
acid decomposition (Figure 3) were recorded to investigate the
changes appearing under reaction conditions. For this purpose,
the catalyst samples were cooled down to room temperature in
inert atmosphere (N2, 60 mL min-1) after finishing the (second)
heating step and were directly analyzed afterwards. The patterns
were compared to the measured diffraction pattern of the filling
material graphite, as well as to calculated patterns of ZnPd and
ZnO. At a first view only the reflections of graphite and ZnPd can
be observed. By magnification of the patterns in the range
between 30 and 40 ° also the main peaks of ZnO become visible
in the patterns of Zn49.8Pd50.2 and Zn51.9Pd48.1, while for the
sample Zn42.0Pd58.0 no signal is observed in this region.
mmolPd h-1). This difference in the catalytic properties of the
chemically differently composed ZnPd samples is also displayed
in the selectivity towards CO2. While Zn51.9Pd48.1 and Zn42.0Pd58.0
show no significant temperature dependence of the
CO2-selectivity (98.5-99.9% and 99.5-99.9%, respectively),
Zn49.8Pd50.2 reveals a reversible decrease of the CO2-selectivity
(down to 96%) with increasing reaction temperature.
-1
Figure 3. X-ray diffraction patterns of bulk Zn100-xPdx after formic acid
decomposition.
Friedrich et al. have shown that ZnO is formed on the surface
of Zn51.9Pd48.1 under MSR conditions.[6] The authors attribute the
high activity of ZnPd in MSR to this phenomenon. In contrast, for
Pd-rich ZnPd stronger resistance in surface oxidation was
observed, which resulted in lower MSR activity. Here, it is
assumed that Zn51.9Pd48.1 and Zn49.8Pd50.2 samples pass an
oxidation/reduction during the heating/cooling in formic acid,
which results in a change of the H2/CO2 ratio as well as of the
catalytic properties. At low temperatures (≤ 240 °C) the reaction
atmosphere possesses a more oxidative character due to the
lower conversion and thus a lower hydrogen content. This leads
to the oxidation of the Zn-rich samples and the formation of ZnO
as observed by ex situ XRD. The partial coverage of the metal
surface with oxidized species seems to have a negative influence
of the catalytic activity as was already seen for GaPd2/β-
Ga2O3.[14] The generated ZnO further reacts with formic acid to
zinc formate as shown by Noto et al. using in situ infrared
spectroscopy during formic acid decomposition on pure ZnO.[17]
Higher temperatures (> 250 °C) cause the decomposition of zinc
formate under formation of ZnO, hydrogen, carbon monoxide and
carbon dioxide as main products whereas hydrogen and carbon
Figure 2. H2/CO2-ratio of bulk ZnPd in formic acid decomposition. Lines are a
guide to the eyes.
In Figure 2 the ratio between H2 and CO2 in the product gas
as function of reaction temperature is depicted. If only
dehydrogenation and/or dehydration of formic acid proceed, this
ratio should be one. Therefore, this value provides information
about additionally ongoing other processes.
For all samples the H2/CO2-ratio was initially 1.0 within the
standard deviation. With increasing temperature the ratio started
to decrease. Above 240 °C Zn51.9Pd48.1 and Zn49.8Pd50.2 showed
a drastic drop with a minimum value of around 0.8 and 0.9,
respectively. The drop of the H2/CO2-ratio was reversible during
cooling and the second heating. However, the palladium-rich
sample Zn42.0Pd58.0 did not share such a drop of the H2/CO2-ratio
and reached a constant value of about 1.0. Thus, the significant
change of the H2/CO2-ratio for Zn51.9Pd48.1 and Zn49.8Pd50.2 is
connected to a reversible change of the samples during formic
This article is protected by copyright. All rights reserved.

10.1002/cctc.201800194
ChemCatChem
FULL PAPER
The obtained activity (1268 mmolCOx mmolPd h-1 at 270 °C)
-1
monoxide are able to form formaldehyde in a second reaction
step.[18] If only decomposition of zinc formate above 240 °C would
influence the H2/CO2 ratio one would expect a short decrease of
the values followed by an equalization with time. Because of the
fact that the H2/CO2 ratio is rather temperature- than time-
dependent (decreasing H2/CO2 ratio with increasing
temperature), there has to be a second reason for the unusual
behavior. An emerging side reaction could be an explanation.
Conceivable side reactions as water gas shift reaction and
formaldehyde formation by hydrogenation of CO can be excluded
because the first one would not alter the H2/CO2 ratio and
formation of formaldehyde out of syngas is reported as not being
feasible.[19] Methanol synthesis from CO2, reported by Iwasa et al.
at ambient pressure over ZnPd supported on ZnO,[20] could
explain the observed decrease of the H2/CO2ratio. The
necessary simultaneous presence of ZnPd and ZnO is in our case
only achieved for the samples Zn49.8Pd50.2 and Zn51.9Pd48.1 above
240 °C. In contrast, Zn42.0Pd58.0 is more resistant against
oxidation and therefore, the methanol synthesis is not feasible.
Thus, methanol synthesis explains the observed catalytic
properties as well as the behavior of the H2/CO2-ratio.
was much higher than for the tested bulk ZnPd. The observed
CO2-selectivity was stable and as high as for the bulk samples
(99.2-99.8%). A view of the H2/CO2-ratio during formic acid
decomposition (Figure 5) shows a characteristic, which is similar
to Zn49.8Pd50.2 and Zn51.9Pd48.1 regarding the temperature where
the ratio changes strongly (240 °C). The second heating of the
catalyst showed that this phenomenon is reproducible and
reversible.
Catalytic Properties of ZnO-Supported ZnPd
For the formic acid decomposition on the supported catalyst
ZnPd/ZnO identical reaction conditions as for bulk ZnPd were
applied. Because of the higher activity of the supported material,
a lower sample mass of 10 mg was used. Figure 4 shows the
conversion and CO2-selectivity of the supported sample. A strong
difference in activity between the first and second heating is
observed (not shown). During the first heating up to 240 °C
ZnPd/ZnO shows lower conversion in comparison with the
second heating session. Between 240 and 270 °C – the assigned
region for zinc formate decomposition – conversion and selectivity
of the first and second heating suddenly assimilate. In difference
to the unsupported samples, ZnPd/ZnO was not crushed before
the catalytic tests. Therefore, the difference between the first and
second heating cannot be explained by healing of surface
irregularities caused by mechanical stress.
Figure 5. H2/CO2-ratio of ZnPd/ZnO in formic acid decomposition. Lines are a
guide to the eyes.
To check for material changes during catalysis, X-ray
diffraction patterns after formic acid decomposition were recorded
(Figure 6). For this purpose, the catalyst sample was cooled down
to room temperature in inert atmosphere (N2, 60 mL min-1) after
finishing the (second) heating step and was directly analyzed
afterwards. The pattern showed the expected peaks of ZnPd and
the support, while the prior observed Pd-reflections are no longer
detectable. Furthermore, the relative intensity of the ZnO
reflections is much lower in comparison to the pattern of the
material after reduction.
Figure 6. X-ray powder diffraction patterns of ZnPd/ZnO after pretreatment and
Figure 4. Conversion and CO2-selecivity of ZnPd/ZnO (10 mg) in formic acid
after formic acid decomposition. The inner window shows a magnification of the
decomposition. Lines are a guide to the eyes.
This article is protected by copyright. All rights reserved.

10.1002/cctc.201800194
ChemCatChem
FULL PAPER
X-ray powder diffraction patterns (the signals at 42.5 ° and 47.5 ° belong to
grbaphite and ZnO, respectively).
of the ZnO during reaction. As shown above, ZnO can form zinc
formate under formic acid decomposition conditions. The
transformation of ZnO into zinc formate and backwards
apparently results in lower crystallinity. This dynamic behavior
might also well be the reason for the disappearance of the Pd
reflections. Here, the originally encapsulated elemental Pd was
transformed to ZnPd.
The discrepancies can either be explained by a significant
leaching of Zn during reaction or by a strong Zn-enrichment of the
intermetallic ZnPd. Analysis of the Zn/Pd mass ratio from the
catalyst sample before – Pd: 8.8(4) wt.-% / Zn: 63.9(8) wt.-% –
and after – Pd: 0.23(1) wt.-% / Zn: 1.26(2) wt.-% - formic acid
decomposition determined by ICP-OES was conducted to
reassess possible leaching. The smaller mass percent of Pd and
Zn after reaction are explained by the dilution of the catalyst
sample through the inert filling material. After normalizing to 100,
the Zn/Pd ratios before – Pd: 12.1(5) wt.-% / Zn: 88(1) wt.-% –
and after – Pd: 15.4(7) wt.-% / Zn: 85(1) wt.-% - formic acid
decomposition showed a slight Zn decrease after reaction, not
explaining the observations by XRD after reaction. The lattice
parameters of the supported ZnPd before (a = 2.8962(1) Å, b =
3.3402(3) Å) and after (a = 2.8979(3) Å, b = 3.3405(6) Å) reaction
do not show a large deviation, thus excluding a vast zinc-
enrichment of the ZnPd. Since no additional Zn-species can be
detected by XRD but the total zinc content stays constant, some
of the zinc must be present in amorphous form.
Reactivity
XPS measurements were conducted to affirm the different
behavior with changing chemical composition and to explore the
predominant states of the near-surface region before, during and
after catalysis. Since in the supported material the presence of
large amounts of ZnO can hinder the detection of changes, the
bulk Zn100-xPdx (x = 48.1-58.0) samples were investigated after
pre-reduction, operando during formic acid decomposition as well
as after subsequent reduction. After pre-reduction signals with
low intensity of Au4d5/2 at 335.4 eV and Au4p3/2 at 546.2 eV
binding energy (BE) are detectable in the regions scans of Pd3d
and Pd3p, respectively. The Au signals were too small to detect
them beforehand in the survey measurement and are most likely
a contamination from the XPS chamber.
Figure 7. SEM images (BSE micrographs) of ZnPd/ZnO before (a, b) and after
(c, d) formic acid decomposition.
SEM investigations show strong differences between
ZnPd/ZnO directly after reduction and the catalyst sample after
reaction (Figure 7). The morphology of the support has drastically
changed due to the formic acid decomposition. After reaction, the
ZnO support shows fine pores and possesses a less compact
morphology.
Most likely, the reason for the different relative reflection
intensities (shown in Figure 6) lies in the change of the crystallinity
This article is protected by copyright. All rights reserved.

10.1002/cctc.201800194
ChemCatChem
FULL PAPER
Figure 8. XP spectra of the Zn3d region measured at Zn52.1Pd47.9 and
Zn49.8Pd50.2 in the pre-reduced state. The spectra correspond to a photon
energy of 237 eV.
subsequent reduction points out that the developing oxidized Zn
species under formic acid conditions is only partially reversible –
in line with the observations during the catalytic measurements in
the reactor. Pd-rich and equimolar ZnPd nearly reach the same
Zn(Ox)/Zn(IMC) ratios as before formic acid decomposition
whereas especially at the most surface-sensitive measurements
a slight increase of oxidized Zn species is present. However, in
the case of Zn-rich the Zn(Ox)/Zn(IMC) ratio changes completely
from 21 at.-% before to 32 at.-% after formic acid decomposition.
These observations of changing oxidative behavior with changing
chemical compositions and partially irreversible reaction state of
Zn-rich ZnPd samples fits well with the inferences drawn out of
the other characterization methods.
By evaluating the Zn3d region of the Zn52.1Pd47.9 and
Zn49.8Pd50.2 sample two different Zn species with characteristic
spin orbit coupling could be identified (Figure 8). The BE of the
corresponding Zn3d5/2 signals were determined to be at 10.5 eV
and 9.65 or 9.49 eV, respectively. While the quantity of the Zn
species at higher BE was decreasing with increasing information
depth, an inverse behavior was observed for the Zn species at
lower BE (Figure S3). In the case of Pd-rich Zn42.0Pd58.0 only one
signal at 9.30 eV appeared.
In contrast to the Zn signal, the Pd3d signal (recorded at BE
The signal at 10.5 eV in the samples can be assigned to a
layer of an oxidized Zn-species, which is in line with the observed
change in depth. According to that, the shifting signal with
chemical composition can be described as Zn in the intermetallic
compound ZnPd. The determined peak positions assort with
former XPS investigations on ZnPd by Friedrich et al. where the
oxidized Zn was also assumed to originate from the presence of
zinc oxide.[6] The shift of the intermetallic Zn3d signal with different
chemical compositions is an indication for changes in the
electronic state of the observed species within the samples and is
also seen for the intermetallic Pd3d signals (Figure S4). While
Friedrich et al. observed this compositional shift in their depth
profiles due to zinc enrichment on the surface caused by
mechanical stress, the newly developed synthesis protocol can
circumvent this effect. The homogeneity of the bulk ZnPd samples
could also be confirmed by calculating the Zn:Pd atomic ratios
based on the fitted peak areas of the Zn3d and Pd3d signals. The
ratios show no dependence with depth, ruling out segregation.
Like in the study by Friedrich et al., the ZnPd samples show
different oxidative behavior in such a way that Zn52.1Pd47.9 and
Zn49.8Pd50.2 are not fully reducible at 200 °C in contrast to Pd-rich
between 335.7 eV for Zn42.0Pd58.0 and 336.2 eV for Zn52.1Pd47.9
)
was only slightly changing with different reaction conditions
(Figure S5). The detected position range as well as the symmetric
peak shape are characteristic for Pd in intermetallic ZnPd.
Investigation of the C1s region revealed the appearance of a
signal at the BE of 289.2 eV under reaction conditions which can
be assigned to formate carbon[22] (Figure S6). Both before and
after formic acid decomposition only the characteristic carbon
depositions from the in situ chamber could be detected. Therefore
emerging of the formate signal together with an increase of the
Zn42.0Pd58.0
.
After UHV XPS investigations, 0.2 mbar formic acid were
applied in the chamber while heating the samples to 200 °C. Once
pressure and temperature were stabilized, complete depth
profiles (Pd3d, Pd3p, Zn3d, O1s, C1s) under operando conditions
were recorded. After formic acid decomposition, the ZnPd
samples were reduced in 0.5 mbar hydrogen at 200 °C and full
depth profiles were recorded in UHV (≈ 10-8 mbar) at room
temperature. Figure 9 displays the atomic Zn(Ox)/Zn(IMC) ratios
of all three samples under formic acid decomposition conditions
and after subsequent reduction. One can directly assess that
Zn52.1Pd47.9 and Zn49.8Pd50.2 are significantly stronger oxidized
than the Zn42.0Pd58.0 sample. At the most surface-sensitive
photon energy (237 eV), more than 50 at.-% of the total Zn of Zn-
rich ZnPd consist of oxidized species under reaction conditions,
whereas at Pd-rich ZnPd less than 10 at.-% of oxidized Zn can be
determined at the same photon energy. The difference between
Zn-rich and equimolar ZnPd is not that pronounced but the
general tendency reveals that Zn-rich ZnPd gets slightly stronger
oxidized. Associated with the grade of oxidation the area of the
O1s signal, which is localized at 530.4 eV,[21] is changing,
approving the alteration of the Zn3d region as oxidation
phenomenon. The quantification of the Zn3d region after
oxidized Zn3d signal provides a strong indication for the presence
of a zinc formate species during formic acid decomposition at
200 °C.
Figure 9. Depth-depending atomic Zn(Ox)/Zn(IMC) ratios of Zn52.1Pd47.9
,
Zn49.8Pd50.2 and Zn42.0Pd58.0 during formic acid decomposition (200 °C, 0.2
mbar formic acid) and after subsequent reduction in hydrogen (200 °C, 0.5
mbar) derived from XPS measurements.
This article is protected by copyright. All rights reserved.

10.1002/cctc.201800194
ChemCatChem
FULL PAPER
To corroborate the XPS measurements, operando DTA/TG
measurements on Zn-rich ZnPd under formic acid decomposition
conditions at ambient pressures were conducted to investigate
the supposed transition between zinc formate and zinc oxide
(Figure 10). To achieve direct comparability, the ground
Zn51.9Pd48.1 was pre-reduced in hydrogen inside the DTA/TG
before formic acid decomposition. Formic acid decomposition
was conducted by three consecutive heating/cooling cycles each
with isothermal steps at the maximum (360 °C) and minimum
(150 °C) temperature. After three cycles, the experiment was
finished at 150 °C and the sample was cooled to room
temperature while flowing He. During the first heating under
formic acid decomposition conditions no sharp mass change can
be observed. Only a minor mass increase, assumedly provoked
by the surface formation of ZnO, is visible. During the first cooling
the mass increases by 0.5 wt.-% with an onset temperature of
250 °C. The reversible character of the mass increase is
becoming clear in the subsequent heating step where a mass loss,
also occurring at an onset of 250 °C, is detected. The mass
change during the heating and cooling steps is reproducible in the
following cycles. In addition, the mass loss is not fully reversible.
The observation perfectly fits to the picture of the previous
investigations. If only formation of ZnO would take place, an
irreversible mass gain would result under formic acid
decomposition conditions. Thus, the increase in mass during the
cooling steps beginning at 250 °C can be explained by the
formation of a zinc formate species, which gets decomposed to
zinc oxide above 250 °C. The irreversible mass gain is assigned
to the formation of zinc oxide.
Characteristic modes for zinc oxide – appearing between 100 and
800 cm-1 – were not detectable.[24]
Figure 11. Raman spectrum of Zn51.9Pd48.1 after operando DTA/TG.
Correlations Between Formic Acid Decomposition and
Methanol Steam Reforming
In general bulk ZnPd seems to exhibit similar changes during
formic acid decomposition and methanol steam reforming
concerning its oxidation behavior with composition. To retrieve
more information, Table 1 lists the apparent activation energies of
bulk ZnPd, ZnPd/ZnO and pure ZnO in formic acid decomposition
and MSR collected from literature. The Arrhenius graphs of the
apparent activation energies, determined in this study, are
depicted in Figure S7-S10.
In formic acid decomposition two apparent activation energies
were calculated for Zn49.8Pd50.2, Zn51.9Pd48.1 and ZnPd/ZnO
because of the different behavior in the low (< 240 °C) and high
temperature (> 240 °C) regime. By comparison of formic acid
decomposition and MSR similar trends are realized with
composition. In both reactions, the apparent activation energy
decreases if intermetallic compound and oxide are
simultaneously present. Correlation between both reaction
mechanisms is thus conceivable. In MSR, the presence of ZnO is
necessary to activate the water and thus achieve high CO2-
selectivity. In contrast, formic acid decomposition also works on
Pd-rich ZnPd compounds where formation of ZnO is suppressed.
At Zn-rich ZnPd samples, the oxidation of Zn and
adsorption/formation of formate seems to hinder the overall
reaction rate. This is a hint that the dehydrogenation of the
postulated formate intermediate takes place preferentially on the
surface of the intermetallic compound and correlates with the
observations of Lorenz et al. who investigated formic acid
decomposition on β-Ga2O3 and intermetallic GaPd2.[14] It is
conspicuous that the apparent activation energies in Table 1 of
Figure 10. Operando DTA/TG measurement of Zn51.9Pd48.1 under formic acid
decomposition conditions. The sample was pre-reduced in 10 vol.-% H2/He
(total gas flow 45 mL min-1) at 200 °C for 1 h. After flushing for 30 min with pure
He (45 mL min-1) and simultaneously cooling down to 150 °C formic acid was
initiated (0.021 mL min-1).
The DTA/TG was finished under formic acid decomposition
conditions below 250 °C and the ZnPd sample was subsequently
investigated by Raman spectroscopy (Figure 11). The most
intense vibrational modes of formate (νs(COO-): 1380, 1363 cm-1;
2νa(COO-): 2795 cm-1; ν(CH): 2910, 2895; νa(COO-)+ν(COO-):
2985 cm-1) with corresponding relative intensities could be
identified, thus strengthening the arguments above.[23]
This article is protected by copyright. All rights reserved.

10.1002/cctc.201800194
ChemCatChem
FULL PAPER
bulk and supported ZnPd determined in MSR are double as high
as for formic acid decomposition on bulk ZnPd. This excludes the
dehydrogenation of a formate species as rate determining step for
the overall MSR reaction mechanism on ZnPd and ZnPd/ZnO,
respectively, in difference to what was proposed earlier by
quantum chemical calculations.[8d]
represent the rate limiting step in MSR on the intermetallic
compound ZnPd.
Experimental Section
For the preparation of Zn100-xPdx (x = 48.1-58.0) samples, Pd-foil
(ChemPur, 99.9%, thickness: 0.25 mm, used in one piece) and zinc
granules (ChemPur, 99.999%, diameter: 1-5 mm) were used. To obtain
dense samples, the recently developed solid-vapor synthesis method was
applied.[26] Direct contact of Zn and Pd was avoided by means of a neck.
The tube was placed in a furnace and the temperature was increased with
60 °C h-1 to 500 °C. After 1 day of annealing, the temperature was raised
to 900 °C with the same heating rate and held for 10 days. Finally, the
ampoule was quenched in water. No interaction of the reactants with the
quartz glass ampoule was observed. ZnO-supported ZnPd particles were
prepared using 4.99911 g ZnO (AlfaAesar, 99.99% metals basis) and
1.52558 g Pd(NO3)2 ∙ xH2O (AlfaAesar, 99.9% metals basis) following a
standard incipient wetness impregnation based on the synthesis route of
Iwasa et al..[4] The single-phase nature of the intermetallic bulk samples
was verified by powder X-ray diffraction (Huber image plate G670, Cu Kα1,
λ = 1.54056 Å, quartz monochromator, 3° < 2θ < 100°). Unit cell
parameters were determined using the program WinCSD[27] with LaB6 as
internal standard (a = 4.15692 Å). Elemental analysis was conducted by
ICP-OES (Vista RL, Varian) by dissolving the samples in aqua regia. The
elemental composition was analyzed in triplicate after matrix matched
calibration.
Table 1. Apparent activation energies of bulk ZnPd, ZnPd/ZnO and ZnO in
formic acid decomposition and MSR. Values without given reference were
measured in the current study.
Sample
Apparent activation energy
(kJ mol-1)
Formic acid
MSR
decomposition
Zn42.0Pd58.0 (150-240 °C)
53(2)
-
Zn49.8Pd50.2 (150-210 °C / 240-330 °C)
Zn49.8Pd50.2 (300-500 °C)
32(3) / 47(3)
-
-
120[6]
Zn51.9Pd48.1 (150-210 °C / 240-330 °C)
ZnPd/ZnO (150-210 °C / 240-330 °C)
ZnPd/ZnO (180-350 °C)
34(3) / 45(2)
-
32(3) / 28(3)
-
-
75[5]
-
ZnO (170-210 °C)
105(8)[25]
-
Catalytic measurements were carried out in a plug-flow reactor system
(Microactivity Reference, PID Eng&Tech). For the catalytic tests the Zn100-
xPdx samples were ground in air, sieved to a particle size between 20 and
32 µm and mixed with 200 mg of graphite (ChemPur, >99.9%) to dilute the
catalyst bed. For better comparison the herein described catalytic tests of
bulk ZnPd were all conducted with a sample mass of 50 mg. The used
graphite was tested before to be inactive in formic acid decomposition as
well as the water gas shift reaction. ZnPd/ZnO was also mixed with 200
mg graphite before catalytic tests. Catalysts were placed inside a silica-
coated stainless steel tube (inner diameter 7.9 mm) which was mounted
inside a hotbox to prevent condensation of liquids. All samples were pre-
treated directly before formic acid decomposition to reach a defined
starting point. The ZnPd bulk samples were reduced in 100 vol.-% H2
(Praxair, 99.999%, 3 ml min-1) at 200 °C for 1 h identical to the chosen pre-
treatment conditions of Friedrich et al..[6] The supported material was
calcined at 500 °C in synthetic air (Praxair, 99.999%, 40 ml min-1) for 3 h
and subsequently reduced in 5 vol.-% H2 (total gas flow 40 ml min-1) at
500 °C for 2 h. Catalytic tests consisted of two heating sessions and an
intermediate cooling session. The samples were heated/cooled in 30 °C
steps with 1 h of holding time at every step to allow for equilibration before
taking data from which the apparent activation energy was determined.
The reactive feed consisted of 0.021 ml min-1 formic acid (Sigma Aldrich,
≥ 98%) which was evaporated before being mixed with 41 ml min-1 N2
(Praxair, 99.999%) and 4 ml min-1 He (Praxair, 99.999%). Helium was
applied as internal standard to calibrate the gas volumes while N2 was
used as carrier gas. For determination of the gas composition in the
product stream a gas chromatograph (Varian Micro GC CP4900) was used,
allowing quantitative determination of CO (column: Al2O3/KCl, carrier gas:
He, sensitivity down to 20 ppm), CO2 (column: Pora Plot U, carrier gas:
He) and H2 (column: MS 5A, carrier gas: Ar from Messer with 99.999%
purity) every four minutes. To determine conversion and selectivity all GCs
per isothermal step were averaged. Unconverted formic acid, as well as
produced water were not determined by GC because they were separated
by a cooling trap and a subsequent Nafion® membrane. Remaining traces
of liquid in the product gas stream were finally removed by cooling the
ZnO (240-340 °C)
144[9]
Conclusions
Single-phase samples of the intermetallic compound ZnPd with
different chemical compositions (Zn51.9Pd48.1 Zn49.8Pd50.2
,
,
Zn42.0Pd58.0) and ZnO-supported ZnPd/ZnO were prepared. The
samples were tested in formic acid decomposition and
characterized by XRD, ICP-OES and SEM before and after
reaction. All tested samples catalyze the decomposition of formic
acid with high CO2-selectivity up to 99.9%. Among the bulk ZnPd
samples Zn42.0Pd58.0 showed the highest activity. Unlike the Zn-
richer samples, Zn42.0Pd58.0 seemed to be more stable against
surface oxidation under reaction conditions as has been proven
by operando XPS measurements. XRD investigations after formic
acid decomposition and operando DTA/TG indicate that Zn-rich
ZnPd underwent a transformation caused by oxidative/reductive
processes during formic acid, which effected the catalytic
properties. ZnO-supported ZnPd showed the highest activity, as
expected. Investigations of ZnPd/ZnO before and after reaction
affirm a modification of the support caused by formation and
decomposition of zinc formate. The apparent activation energies
of bulk ZnPd and ZnPd/ZnO in MSR sample are two times higher
than for formic acid decomposition. The experimental results
imply that dehydrogenation of a formate intermediate does not
This article is protected by copyright. All rights reserved.

10.1002/cctc.201800194
ChemCatChem
FULL PAPER
product stream to -60 °C. The conversion of formic acid is calculated as C
= (COx_out/COx_max) x 100, wherein COx_out is the summed concentration
of CO and CO2 in the product gas and COx_max is the concentration of
carbon oxides that can be generated if all formic acid is converted
according to the reaction equations HCOOH → H2O + CO and HCOOH
→ H2 + CO2, respectively. The CO2-selectivity S(CO2) is calculated as
CO2_out/COx_out x 100, wherein CO2_out is the concentration of CO2 in the
product gas. The activity is defined as A = COx_out/(nPd x t), wherein nPd is
the molar amount of palladium used and t is time.
Fitted areas were corrected considering ring current, photon flux and the
respective cross sections.[30] The calculation of information depths relies
on the inelastic mean free path (IMFP) through the NIST Standard
Reference Database 71[31] based on the work of Tanuma et al..[32] As an
approximation the IMFP of elemental Pd was used. The information depth
is defined as three times the IMFP, i.e. 95% of the excited electrons
originate from the respective depth.[33]
Acknowledgements
Scanning electron microscopy (SEM, FEI Quanta 200F, cold FEG) was
conducted to investigate the morphology of the supported sample before
and after catalytic measurement. Prior to SEM imaging, graphite – which
was used as filling material during formic acid decomposition – was mostly
separated by sieving. Samples were mounted on Al-holders using
conducting carbon tape prior to analysis.
The authors thank J. Grin for his unfathomable support especially
during the move from the MPI CPfS in Dresden to the Technical
University Chemnitz. We acknowledge the Helmholtz Zentrum
Berlin für Materialien und Energie – Electron storage ring BESSY
II – for providing beamtime at the ISISS beamline (Project No.
14201526ST) and J. Velasco-Vélez, T. Keilhauer, M. Meier and
R. Blume for supporting the XPS measurements. The authors
also thank W. Reschetilowski for valuable discussion.
Operando DTA/TG measurements (Netzsch, STA 449 F3 Jupiter) of
Zn51.9Pd48.1 under formic acid decomposition conditions were conducted
to investigate changes of the catalyst sample during heating and cooling
cycles. Similar to catalytic investigations, bulk ZnPd was ground in air to
powder. Prior to formic acid decomposition the sample was reduced in 10
vol.-% H2/He (total gas flow 45 mL min-1) at 200 °C for 1 h. After flushing
for 30 min with pure He (45 mL min-1) and simultaneously cooling down to
150 °C formic acid was introduced (0.021 mL min-1). Three heating/cooling
cycles, each with a rate of 5 K min-1 and one hour isothermal steps at the
minimum (150 °C) and maximum (360 °C) temperature were conducted
before cooling down to room temperature. The measurement was
background corrected to eliminate any non-sample effects.
Keywords: Formic acid decomposition • Intermetallic compound
• Methanol steam reforming • ZnPd
[1]
[2]
Y. Brunet, Energy Storage, Wiley-VCH, Weinheim, 2013.
P. Gruss, F. Schüth, Die Zukunft der Energie, C. H. Beck, München,
2008.
[3]
[4]
[5]
[6]
G. Bender, M. Angelo, K. Bethune, R. Rocheleau, J. Power Sources
2013, 228, 159-169.
Raman spectra of Zn52.1Pd48.1 after operando DTA/TG were measured
with a Horiba Jobin-Yvon Labram 010 spectrometer at ambient conditions
in backscattering geometry. For excitation the green Ar ion laser line (488
nm) with a power of 30 mW at the sample was used. Acquisition time per
spectrum ranged from 100 to 2000 s. The sample was measured at
different positions to ensure characteristic spectra.
N. Iwasa, S. Masuda, N. Ogawa, N. Takezawa, Appl. Catal., A 1995, 125,
145-157.
M. Friedrich, S. Penner, M. Heggen, M. Armbrüster, Angew. Chem. Int.
Ed. 2013, 52, 4389-4392; Angew. Chem. 2013, 125, 4485.
M. Friedrich, D. Teschner, A. Knop-Gericke, M. Armbrüster, J. Catal.
2012, 285, 41-47.
[7]
[8]
T. H. Chiang, H. Ipser, Y. A. Chang, Z. Metallkd., 1977, 68, 141-147.
a) Z. X. Chen, K. M. Neyman, K. H. Lim, N. Rösch, Langmuir 2004, 20,
8068-8077; b) K. H. Lim, Z.-X. Chen, K. M. Neyman, N. Rösch, J. Phys.
Chem. B 2006, 110, 14890-14897; c) Z. X. Chen, K. H. Lim, K. M.
Neyman, N. Rösch, J. Phys. Chem. B 2005, 109, 4568-4574; d) S. Lin,
D. Xie, H. Guo, J. Phys. Chem. C 2011, 115, 20583-20589; e) S. Lin, D.
Xie, H. Guo, J. Mol. Catal. A: Chem. 2012, 356, 165-170; f) G. K. Smith,
S. Lin, W. Lai, A. Datye, D. Xie, H. Guo, Surf. Sci. 2011, 605, 750-759.
H. Lorenz, M. Friedrich, M. Armbrüster, B. Klötzer, S. Penner, J. Catal.
2013, 297, 151-154.
In situ X-ray photoelectron spectroscopy (in situ XPS) was conducted at
the ISISS-PGM beamline, Helmholtz Zentrum Berlin für Materialien und
Energie – Electron storage ring BESSY II. The setup is described in detail
elsewhere[28] Bulk ZnPd samples with different chemical composition were
prepared by the foil method as already described. To ease the mounting
of the samples on the sapphire holder the Pd-foils were punched into round
pieces with a diameter of 8 mm prior to synthesis. Pre-reduction in
hydrogen was performed in situ at 0.5 mbar hydrogen pressure and 200 °C
while measuring the Zn3d signal to monitor the surface reduction. After
reaching a stable state, the samples were cooled to ambient temperature
and the core levels of Pd3d, Pd3p, Zn3d, O1s and C1s were recorded at
pressures of 10-8 mbar. To achieve depth profiling of the samples, core
levels were measured at five different kinetic energies (232, 382, 532, 682
and 832 eV). Experiments under operando formic acid decomposition
conditions were conducted at 0.2 mbar formic acid and 200 °C. Formation
of reaction products as an indication of catalytic activity during the
measurements was detected by a mass spectrometer (Pfeiffer Prisma).
After formic acid decomposition, the samples were reduced and measured
once again to check the reversibility in the changes to the near-surface
region. The signals were energy corrected by the fermi edge before
qualitative and quantitative evaluation using the software Casa XPS.[29]
Prior to the fitting of the signal areas the background was subtracted by
applying the Shirley-type of background. During the fitting of the signals
the ratio between the doublet signals (Pd3d5/2:Pd3d3/2 and
Zn3d5/2:Zn3d3/2 = 3:2, Pd3p3/2:Pd3p1/2 = 2:1) as well as the distance
[9]
[10] M. Armbrüster, M. Behrens, K. Föttinger, M. Friedrich, E. Gaudry, S. K.
Matam, H. R. Sharma, Cat. Rev. - Sci. Eng. 2013, 55, 289-367.
[11] a) E. S. Ranganathan, S. K. Bej, L. T. Thompson, Appl. Catal., A 2005,
289, 153-162; b) A. Haghofer, D. Ferri, K. Föttinger, G. Rupprechter,
ACS Catal. 2012, 2, 2305-2315.
[12] G. C. Bond, Catalysis by Metals, Academic Press, 1962.
[13] a) G. Rienäcker, Z. Elektrochem. 1934, 40, 487-488; b) G. Rienäcker, Z.
Anorg. Allg. Chem. 1936, 227, 353-375.
[14] H. Lorenz, R. Thalinger, E.-M. Köck, M. Kogler, L. Mayr, D. Schmidmair,
T. Bielz, K. Pfaller, B. Klötzer, S. Penner, Appl. Catal., A 2013, 453, 34-
44.
[15] E. Jeroro, J. M. Vohs, Catal. Lett. 2009, 130, 271-277.
[16] S. Penner, M. Armbrüster, ChemCatChem 2015, 7, 374-392.
[17] Y. Noto, K. Fukuda, T. Onishi, K. Tamaru, Trans. Faraday Soc. 1967, 63,
3081-3087.
[18] a) K. A. Hofmann, K. Schumpelt, Ber. Dtsch. Chem. Ges. 1916, 49, 303-
317; b) Dollimor.D, K. H. Tonge, J. Inorg. Nucl. Chem. 1967, 29, 621-
627.
between the doublet signals (∆Pd3d5/2/Pd3d3/2
=
5.26 eV,
∆Zn3d5/2/Zn3d3/2 = 0.4 eV, ∆Pd3p3/2/Pd3p1/2 = 22.6 eV) was constrained.
This article is protected by copyright. All rights reserved.

10.1002/cctc.201800194
ChemCatChem
FULL PAPER
[19] A. M. Bahmanpour, A. Hoadley, A. Tanksale, Green Chem. 2015, 17,
[26] D. C. A. Ivarsson, U. Burkhardt, K. Richter, R. Kriegel, L. Rößner, M.
Neumann, M. Armbrüster, J. Alloys Compd. 2018, accepted manuscript.
[27] L. G. Akselrud, Y. P. Zavalii, Y. N. Grin, V. K. Pecharski, B. Baumgartner,
E. Wölfel, Mater. Sci. Forum 1993, 133-136, 335-342.
3500-3507.
[20] N. Iwasa, H. Suzuki, M. Terashita, M. Arai, N. Takezawa, Catal. Lett.
2004, 96, 75-78.
[21] E. Agostinelli, C. Battistoni, D. Fiorani, G. Mattogno, J. Phys. Chem.
[28] M. Salmeron, R. Schlögl, Surf. Sci. Rep. 2008, 63, 169-199.
[29] Casa XPS (Version 2.3.16), 2010.
Solids 1989, 50, 269-272.
[22] S. Yumitori, J. Mater. Sci. 2000, 35, 139-146.
[30] J. J. Yeh, I. Lindau, At. Data Nucl. Data Tables 1985, 32, 1-155.
[31] NIST Electron Inelastic-Mean-Free-Path Database, 2010.
[32] S. Tanuma, C. J. Powell, D. R. Penn, Surf. Interface Anal. 1991, 17, 911-
926.
[23] J. Hedberg, S. Baldelli, C. Leygraf, J. Electrochem. Soc. 2010, 157,
C363-C373.
[24] N. Panwar, J. Liriano, R. S. Katiyar, J. Alloys Compd. 2011, 509, 1222-
1225.
[33] J. C. Vickerman, I. S. Gilmore, Surface Analysis
Techniques, Vol. 2, Wiley, Chichester, 2009.
– The Principle
[25] T. Kondo, T. Ogawa, K. Kishi, K. Hirota, Z. Phys. Chem. 1969, 67, 284-
290.
This article is protected by copyright. All rights reserved.

10.1002/cctc.201800194
ChemCatChem
FULL PAPER
FULL PAPER
The reaction mechanism of methanol
steam reforming (MSR) on ZnPd was
investigated by use of catalytic formic
acid decomposition as model reaction.
On the basis of determined apparent
activation energies we conclude that
dehydrogenation of a formate
René Kriegel, Dennis C. A. Ivarsson,
and Marc Armbrüster*
Page No. – Page No.
Formic Acid Decomposition over
ZnPd – Implications for Methanol
Steam Reforming
intermediate does not represent the
rate limiting step.
This article is protected by copyright. All rights reserved.