Mechanistic Study of Carbon Dioxide Hydrogenation over Pd/ZnO-Based Catalysts: The Role of Palladium–Zinc Alloy in Selective Methanol Synthesis

Article abstract of DOI:10.1002/anie.202103087

Pd/ZnO catalysts show good activity and high selectivity to methanol during catalytic CO₂ hydrogenation. The Pd-Zn alloy phase has usually been considered as the active phase, though mechanistic studies under operando conditions have not been c

Full text of DOI:10.1002/anie.202103087

A Journal of the Gesellschaft Deutscher Chemiker
Angewandte
Chemie
International Edition
www.angewandte.org
Accepted Article
Title: Mechanistic study of carbon dioxide hydrogenation over Pd/
ZnO-based catalysts: the role of palladium-zinc alloy in selective
methanol synthesis
Authors: Maxim Zabilskiy, Vitaly L. Sushkevich, Mark A. Newton, Frank
Krumeich, Maarten Nachtegaal, and Jeroen A. van Bokhoven
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: Angew. Chem. Int. Ed. 10.1002/anie.202103087
Link to VoR: https://doi.org/10.1002/anie.202103087

10.1002/anie.202103087
Angewandte Chemie International Edition
RESEARCH ARTICLE
Mechanistic study of carbon dioxide hydrogenation over Pd/ZnO-
based catalysts: the role of palladium-zinc alloy in selective
methanol synthesis
Maxim Zabilskiy,*[a] Vitaly L. Sushkevich,[a] Mark A. Newton,[b] Frank Krumeich,[b] Maarten Nachtegaal,[a]
and Jeroen A. van Bokhoven*[a],[b]
[a]
Dr. M. Zabilskiy, Dr. V. L. Sushkevich, Dr. M. Nachtegaal, Prof. Dr. J. A. van Bokhoven
Laboratory for Catalysis and Sustainable Chemistry,
Paul Scherrer Institute,
CH-5232 Villigen, Switzerland
E-mail: maxim.zabilskiy@psi.ch, jeroen.vanbokhoven@chem.ethz.ch
[b]
Dr. M. A. Newton, Dr. F. Krumeich, Prof. Dr. J. A. van Bokhoven
Institute for Chemistry and Bioengineering
ETH Zurich
Vladimir-Prelog-Weg 1, 8093 Zürich, Switzerland
Supporting information for this article is given via a link at the end of the document.
Abstract: Pd/ZnO catalysts show good activity and high selectivity to
methanol during catalytic CO2 hydrogenation. The palladium-zinc
alloy phase has usually been considered as the active phase, though
mechanistic studies under operando conditions have not been
conducted to verify this. Here, we report a detailed mechanistic study
under realistic conditions of methanol synthesis, using in situ and
operando X-ray absorption spectroscopy, X-ray powder diffraction
and time-resolved isotope labeling experiments coupled with FTIR
spectroscopy and mass spectrometric (MS) analysis. Palladium-zinc
mechanism, and the structure of the active sites, is of a
considerable importance. Under a hydrogen rich atmosphere and
at elevated temperature (≥150 °C), Pd/ZnO is known to form a
palladium-zinc alloy phase, which is considered as providing the
active sites for various catalytic applications including methanol
steam reforming, reverse water gas shift, and hydrogenation
reactions.[16–22] The formation of a palladium-zinc alloy phase is
also often associated with high activity and selectivity towards
methanol in the hydrogenation of carbon dioxide.[5,23–25] Although
the mechanism of methanol synthesis over Pd/ZnO is not fully
understood, it is considered that the palladium-zinc alloy is able
to generate and stabilize formate intermediate species, which
may then be hydrogenated to yield methanol.[24] The association
of the palladium-zinc alloy phase with the active sites has, in most
studies, solely been based on the fact that this phase is formed
during the reductive pre-treatment of the catalyst. However, in situ
and/or operando studies, conducted under reaction relevant
conditions (200-260 °C, P≥10 bar), which could verify this
supposition, have yet to appear. Gentzen et al.,[26] investigated
the behavior of a Pd/ZnO system under operando conditions and
observed formation of palladium-zinc alloy during dimethyl ether
synthesis from carbon monoxide-rich gas mixtures having a
carbon monoxide/hydrogen ratio 1:1, however, methanol
synthesis typically utilizes a significantly more reducing feedstock
(H2/CO2 = 3).
Taking into account recent works,[27–30] which have questioned the
role of copper-zinc alloy in methanol synthesis from carbon
dioxide, an operando investigation of carbon dioxide
hydrogenation over Pd/ZnO system is warranted to understand
the true nature of the active sites and to verify the role that a
palladium-zinc alloy phase may have in facilitating this reaction.
We therefore report a combined high-pressure operando study of
methanol synthesis over Pd/ZnO catalysts using XAS, XRD and
FTIR techniques. We find that, despite the fact that a palladium-
zinc alloy phase is present during reaction, methanol is not formed
when the support is an inert oxide (e.g. Al2O3, SiO2). Instead, we
demonstrate that the co-existence of zinc oxide with palladium or
palladium-zinc phases is required to form methanol with high
activity and selectivity in this industrially relevant reaction.
alloy-based catalysts, prepared through reduction of
a
heterobimetallic PdIIZnII acetate bridge complex, and which do not
contain zinc oxide or any PdZn/ZnO interface, produce mostly carbon
monoxide. The palladium-zinc phase is, therefore, principally
associated with the formation of carbon monoxide, and does not
provide the active sites required to produce methanol from the direct
hydrogenation of carbon dioxide. Instead, we show that the presence
of a zinc oxide phase, in contact with a palladium-zinc phase, is
essential for efficient production of methanol.
Introduction
Despite decades of effort, a commercially viable synthesis of
methanol from carbon dioxide, has yet to be realised, and the
development of an appropriate catalyst formulation is still very
much in progress. The commercial catalyst for methanol
synthesis from syngas (Cu/ZnO/Al2O3), as well as other copper-
based materials, such as Cu/ZrO2, Cu/CeO2, Cu/ZnO/ZrO2 and
Cu/Mo2C, are often the first choice for studying carbon dioxide
hydrogenation to methanol. However, limited activity and poor
selectivity to methanol due to the reverse water gas shift reaction
(RWGS) have forced the investigation of alternative catalyst
compositions, for instance: In2O3 based materials;[1–4] palladium-
containing catalysts;[5–7] and gallium-based intermetallic
compounds.[8–10] Among these Pd/ZnO catalysts, which show
good activity and high selectivity to methanol, are of a great
scientific interest.[11–15]
In order to improve the performance of catalysts for methanol
synthesis from carbon dioxide, an understanding of the reaction
1
This article is protected by copyright. All rights reserved.

10.1002/anie.202103087
Angewandte Chemie International Edition
RESEARCH ARTICLE
Figure 1. HAADF-STEM (a) and BF-STEM (b) of 2Pd-ZnO-i. HAADF-STEM image (c) and corresponding EDX elemental mapping (d) of 2Pd-ZnO-i (zinc–red and
palladium–green). SE-STEM images of 2Pd-ZnO-np (e and f). Results of catalytic carbon dioxide hydrogenation (CO2:H2=1:3) over 2Pd-ZnO-np at 50 bar total
pressure and WHSV=120 L·g-1·h-1 (g).
which show no bands due to possible intermediates such as
carbonate-like species (Figures S1-S2). In contrast to carbon
dioxide, the 13CH3OH response shows a significant delay with
respect to the tracer. This points to the presence of sequential
Results and Discussion
reactions involved in the conversion of carbon dioxide to methanol,
along with possible re-adsorption of methanol on the surface of
the catalyst. Infrared spectra, acquired during the reaction of
12CO2/H2 over a 2Pd-ZnO-np catalyst, show the presence of the
bands centered at 1596 and 1376 cm-1 which may be assigned to
the asymmetric and symmetric vibrations of bidentate formate
species, respectively (Figure S1, Table S2). Our recent studies,[27]
based on FTIR spectroscopy and XAS analysis, revealed that
these formate species correspond to zinc formate species in a
copper-zinc oxide catalyst. No other bands due to methoxy or
carbonate species were found in the spectra. Switching of the
isotopic composition of the reacting feed from 12CO2 to 13CO2
leads to fast exchange of formate species, resulting in a decrease
of the intensity of the bands at 1596 and 1376 cm-1, which is
accompanied by a simultaneous development of bands at 1551
and 1346 cm-1 due to 13C labeled formates (Figure S2). The
difference spectra for both switches show that the exchange is
complete and fully reversible. This indicates the involvement of all
zinc formate species in the reaction of carbon dioxide
hydrogenation to methanol. Moreover, the maximum of the bands
does not shift during the isotope exchange, suggesting the
presence of a single type of zinc formate species with identical
reactivity.
Figure 1 shows electron microscopy results of Pd/ZnO after
calcination in a flow of air at 400 °C. The 2Pd-ZnO-i sample,
prepared by impregnation of palladium nitrate solutions, contains
unevenly distributed palladium nanoparticles with sizes ranging
from 1 to 7 nm (Figure 1 a-d). The sample synthesized using
colloidal palladium shows nicely distributed, and essentially
monodisperse, palladium nanoparticles with a mean size of 2 ±
0.2 nm (Figure 1e-f).
Catalytic carbon dioxide hydrogenation over 2Pd-ZnO-np at total
pressure of 50 bar results in a high selectivity to methanol of
between 65 and 85 %, and with carbon monoxide as the sole by-
product (Figure 1g). Decreasing the reaction temperature from
260 °C to 220 °C yields an even higher methanol selectivity since,
at lower temperature, the methanol synthesis reaction
increasingly dominates the catalytic hydrogenation of carbon
dioxide. The absolute values of methanol productivity are among
the highest for Pd/ZnO-based catalysts thus far reported for this
system (Table S1). In an attempt to understand the fundamental
reasons of such a high activity of palladium-zinc catalyst, we have
focused our attention on 2Pd-ZnO-np sample and used it as the
main object in our operando experiments.
The kinetic evolution of the products and intermediate species
was followed by IR spectroscopy and MS. Figure 2a shows the
normalized MS responses on 2Pd-ZnO-np for 13C-labelled and
unlabeled carbon dioxide and methanol versus the signal of the
tracer (argon), obtained during the isotope switch from 12CO2/H2
to 13CO2/H2. The normalized response of unlabeled carbon
dioxide follows the response of the inert tracer, which points to the
irreversible activation of carbon dioxide to form the reaction
products. This observation is in line with the infrared spectra,
2
This article is protected by copyright. All rights reserved.

10.1002/anie.202103087
Angewandte Chemie International Edition
RESEARCH ARTICLE
Figure 2. Normalized isotopic transient response curves following the switch from a 12CO2/H2 to 13CO2/H2 mixture during catalytic carbon dioxide hydrogenation
over 2Pd-ZnO-np at 260 °C and 15 bar total pressure measured by mass-spectrometry (a) and FTIR spectroscopy (b). Pd K-edge XANES spectra (c), the
corresponding phase corrected Fourier transform representation of the k3-weighted EXAFS (d) and XRD (e) collected during activation and catalytic carbon dioxide
hydrogenation over 2Pd-ZnO-np catalyst at 260 °C and 15 bar pressure under different sets of conditions (as indicated): under a helium flow (violet); under flowing
hydrogen (blue); at steady state under a catalytic CO2/H2 reaction mixture (aquamarine); and under a flow of carbon dioxide (orange). In each case the solid lines
concern the experimental data, the black dotted lines, fits to that data derived from analysis in EXCURV.[31] Diffraction peaks of the palladium-zinc alloy phase
present in 2Pd-ZnO-np and 2Pd-ZnO-i (red XRD pattern) samples are labeled with diamond symbol (♦)
For kinetic analysis, the area of the band at 1596 cm-1 due to
these formate species and bands due to gas phase carbon
dioxide were integrated and compared (Fig. 2b). The signal due
to the carbon dioxide matches the behavior of the argon tracer in
the MS very closely (Figures 2a-b). The results clearly show that
the kinetic evolution of formate species is significantly delayed
with respect to the evolution of the carbon dioxide signal. This
unambiguously points to the involvement of formate species to
the catalytic cycle of methanol synthesis.
In further experiments, the evolution of both palladium and zinc in
the catalyst was followed. A combined operando XAS-XRD
experiment for 2Pd-ZnO-np was conducted at 260 °C and 15 bar
pressure at the SNBL beamline (BM31, ESRF, France). Figure 2c
shows Pd K-edge XANES spectra collected during catalyst pre-
treatment and during operando carbon dioxide hydrogenation.
After heating in helium to 260 °C, the spectrum of the fresh
catalyst resembles well that of the palladium oxide reference
(Figures S3-S4). Changing the reaction gas to hydrogen leads to
a shift of the white line to lower energy by ~2 eV due to immediate
reduction of palladium oxide and formation of reduced palladium
phase (Figure 2c, Figures S3-S4). According to EXAFS analysis
(Figure 2d, Table S3), XRD (Figure 2e and Figures S5-S6), and
formation of the palladium-zinc alloy phase is facilitated by
hydrogen spillover and accumulation of hydrogen inside metallic
palladium.[22] Furthermore, and irrespective of the difference in
palladium particle size and distribution, the 2Pd-ZnO-i sample
shows very similar behavior during operando XAS study (Figure
S3).
In contrast to our previous study,[27] where we have observed
changes in Cu K-edge XANES that result from an oxidative
desegregation of zinc from a copper-zinc alloy phase upon
switching to a CO2/H2 mixture, the position of the palladium white
line remains constant, not only during carbon dioxide
hydrogenation conditions, but even after switching to pure carbon
dioxide (Figure 2c). Since the white line is sensitive to the
oxidation state of palladium, this suggests that palladium remains
in a metallic state and that the formate species, as observed by
IR (Figures S1-S2 and Figure 2b), are not bound to the palladium.
Moreover, the palladium-zinc alloy phase is stable under reaction
conditions and under high pressures of carbon dioxide. The Zn K-
edge XANES (Figure S9) also stays constant during transient
switches between hydrogen, CO2/H2 mixture and pure carbon
dioxide, for this (2Pd-ZnO-np) sample. Taking into account the
particle size of the zinc oxide phase, which is in the range 10-30
TEM (Figures S7-S8), the resulting phase is a palladium-zinc alloy, nm according to TEM and XRD analysis (Figures 1, 2e and S7b),
in a good agreement with previous studies, where the formation
of this phase was observed even at ambient pressure.[17,22,32] The
any surface changes, even if they take place during catalytic
reaction, will be difficult to observe using
3
This article is protected by copyright. All rights reserved.

10.1002/anie.202103087
Angewandte Chemie International Edition
RESEARCH ARTICLE
Table 1. Structural and statistical parameters derived from the fitting of k3-weighted Pd K-edge EXAFS for the 2Pd-ZnO-np catalyst
under different conditions at 260 °C and 15 bar pressure. Kmin=2.5, Kmax=13, AFAC=0.9
d
EF
Na
r(Å)b
DWc
Re
Chi2 f
Conditions
He
Element
O
2.4
2.3
1.2
2.9
4.4
2.2
0.9
2.02
0.008
0.018
0.016
0.024
0.019
0.029
0.014
Pd
2.74
1.83
28.35
0.55
Pd
3.03
Pd
3.43
Zn
2.59
-6.1
25.3
0.45
0.43
H2
Pd
2.95
(-6.6)
(25.7)
Pd (Zn)
3.51 (3.69)
Zn
Pd
4.4
2.1
2.58
2.92
0.019
0.027
CO2/H2
-5.5
45.8
1.25
a N = Coordination number, b R = bond distance in Angstrom, c DW = Debye-Waller (disorder) factor (22) where 2 = mean squared displacement of the atom pair
with respect to each other, d EF = Fermi energy (eV), e R% = iN 1/i (ie (k) - it (k))2 x 100% Where ie and it are the experimental and theoretical EXAFS respectively,
and k is the photoelectron wave vector (Å-1). i is the uncertainty in the data, with 1/i = kin/ jN kin (ie (j))2, f Chi2 = statistical measure of goodness of fit (x 10-6).
Zn K-edge XAS. Figure 2d and Figure S10 show the k3-weighted
Pd K-edge EXAFS derived from the 2Pd-ZnO-np sample during
the operando XAS experiment. The presence of Pd-O scattering
at 2 Å and two Pd-Pd scattering shells, at ca. 3.03 and 3.43 Å
respectively, are indicative of a nanoscale, PdO structure (Table
1). Therefore, EXAFS yields a picture of 2Pd-ZnO-np sample
under 15 bar He at 260 C as being predominantly oxidized in
nature, which is consistent with the previously shown XANES
(Figure 2c).
2Pd-ZnO-np sample (Figure S7d), and then calculating the
inverse FFT (IFFT), we have determined the locations of the
corresponding ZnO and PdO phases. Figure S7e shows
superposition of original micrograph of 2Pd-ZnO-np sample with
IFFT of Figure S7d, where circular masks were applied at lattice
spacing of 1/2.60 Å-1, corresponding to the (002) planes of zinc
oxide (red), and at a lattice spacing of 1/2.63 Å-1, which
corresponds to the (101) planes of palladium oxide (blue).
Although these d-values are rather similar, the darker contrast of
the smaller particles indicate the presence of the PdO phase there
(mass contrast).
TEM micrographs of spent 2Pd-ZnO-np catalyst and
corresponding FFT and IFFT (Figure S8) reveal the formation of
PdZn alloy phase after catalytic carbon dioxide hydrogenation. An
FFT of the area represented on Figure S8a also shows two spots
at 1/2.60 Å-1 and 1/1.63 Å-1 which belong to the (002) and (110)
planes of wurtzite zinc oxide phase. Further reflections, at 1/2.05
Å-1 and 1/2.20 Å-1 correspond, to (200) and (111) planes of the
palladium-zinc alloy and match those of the (200) and (111) d-
spacing obtained from XRD and TEM examinations by Malik et
al.[33] The reconstructed IFFT images also show the localization
of the zinc oxide phase (Figure S8c), where circular masks were
applied at 1/2.60 Å-1 and 1/1.63 Å-1, as well as that of PdZn alloy
nanoparticle (Figure S8d), where circular mask was applied at
1/2.05 Å-1 corresponding to (200) planes of palladium-zinc alloy.
To summarize the results of these operando experiments,
catalytic carbon dioxide hydrogenation occurs via a formate
intermediate route. FTIR spectra suggest that the active formate
species are bound to zinc (characteristic bands at 1596 cm-1 and
1376 cm-1). XAS revealed the absence of palladium-zinc de-
alloying or any indication of the surface oxidation of palladium,
which might be expected due to possible palladium-formate
°
A switch to a hydrogen flow results in significant changes in the
FT EXAFS. The peak at 2 Å (due to Pd-O scattering) disappears,
while a new peak at ca. 2.59 Å arises (Figure 2d). This
corresponds to a complete reduction of the palladium present,
along with some of the zinc oxide, to yield a highly dispersed
palladium-zinc alloy phase (RPd−Zn = 2.59 Å, RPd−Pd = 2.95 Å).[22]
Exposure to the reaction mixture, and then carbon dioxide, does
not affect the palladium-zinc alloy (Figure 2d). Overall, the fitting
of k3-weighted Pd K-edge EXAFS of 2Pd-ZnO-np (RPd−Zn = 2.59
Å, RPd−Pd = 2.92 Å) shows that, once reduced in hydrogen, the
nanoparticulate palladium-zinc phase is retained under all of the
conditions applied, irrespective of the gaseous environment
experienced (Tables 1 and S3).
Figure 2e shows the XRD patterns of 2Pd-ZnO-np collected at the
end of every step during the XAS experiment (Figure 2c and d).
The change of gas atmosphere from helium to hydrogen results
in the formation of very broad (FWHM~1) overlapping peaks. For
clarity, the XRD pattern of 2Pd-ZnO-i (containing larger palladium
nanoparticles) collected under hydrogen atmosphere was added
to Figure 2e (other XRD patterns for this sample can be found in
Figure S5). The XRD pattern of the spent 2Pd-ZnO-np sample
shows the (111) diffraction peak of a palladium-zinc alloy phase
after catalytic carbon dioxide hydrogenation. Other diffraction
peaks present in this pattern, correspond to the zinc oxide support, intermediates, during the catalytic cycle. Instead, transient
while we did not observe any peak indicative of a palladium fcc
phase (Figure S6).
experiments show that the palladium-zinc phase is very stable
under different gas atmospheres (hydrogen, CO2/H2 mixture and
carbon dioxide) and it may well be only responsible for hydrogen
activation, while carbon dioxide activation, and its subsequent
hydrogenation, occur on the zinc oxide similar to the Cu/ZnO-
based system.[27,28] However, at this point, it is not clear, whether
Figure S7a-c shows TEM micrographs of fresh 2Pd-ZnO-np
catalyst. A fast Fourier transform (FFT) of TEM micrograph shown
in Figure S7c revealed several spots corresponding to different
crystal lattice vectors of zinc oxide and palladium oxide. By
applying circular masks at a given lattice spot in the FFT of the
4
This article is protected by copyright. All rights reserved.

10.1002/anie.202103087
Angewandte Chemie International Edition
RESEARCH ARTICLE
the palladium-zinc alloy alone governs the activity and selectivity
of the catalyst, or the interface between the alloy phase and bulk
zinc oxide represents the active site for the selective
transformation of carbon dioxide to methanol.
intensity of the low binding energy pre-edge feature observed in
XANES (Figure 3d), the EXAFS, in this case, is dominated by a
broad low Z (O) scattering shell (Figure 3f and Figure S14). We
also find the shell at ca 2.6 Å, indicative of the presence of the
palladium-zinc alloy, but at a much reduced level in terms of its
contribution to the overall EXAFS envelope. Together with the
XANES (Figure 3d) and TEM (Figures S11-S12), this sample is
therefore comprised of both palladium-zinc alloy and zinc oxide
phases, with a much heavier weighting to the latter than the
former. Overall, we found that palladium-zinc alloy phase in these
samples is stable during carbon dioxide hydrogenation, and
similarly to 2Pd-ZnO-np sample, does not undergo oxidation or
de-alloying during methanol synthesis.
To investigate the role of palladium-zinc alloy in this system, we
have designed the following experiment to distinguish the activity
of palladium-zinc alloy from that of the PdZn/ZnO interface in this
reaction. From operando XAS and XRD experiments, we have
found that, under carbon dioxide hydrogenation conditions,
palladium is present in the form of a palladium-zinc alloy.
Therefore, we supported a heterobimetallic PdIIZnII acetate bridge
complex on silica and alumina and reduced it to obtain palladium-
zinc alloy supported catalysts,[32,34,35] PdZn/SiO2 and PdZn/Al2O3,
which do not contain any zinc oxide, and therefore cannot present
a PdZn/ZnO interface. Alternatively, a palladium-zinc catalyst with
excess of ZnO (i.e. PdZn/ZnO/SiO2) was synthesized by a double
impregnation of silica, firstly with the PdIIZnII acetate complex, and
secondly with zinc nitrate. This sample was then calcined in air
and then reduced in hydrogen, such that the role of zinc oxide and
of the PdZn/ZnO interface on the catalysis could be determined.
Both samples (i.e. PdZn/ZnO/SiO2 and PdZn/ZnO/SiO2) possess
similar particle size distributions, comparable to that of 2Pd-ZnO-
i (Figures S11-S12).
Figures 3a-b show Pd K-edge XANES, and the first derivatives
thereof, of PdZn/SiO2 and PdZn/Al2O3 after calcination and
reduction, as well as the spectrum of the 2Pd-ZnO-np catalyst
discussed earlier. Pd K-edge XANES spectra of samples
prepared via the PdIIZnII acetate route resemble those of both
2Pd-ZnO-np and a palladium foil standard. According to EXAFS
analysis (Figure 3c and Table S4), both of these systems are
characterized by a Pd-Zn scattering shell at ca. 2.57 – 2.6 Å. As
such, the Pd K-edge XANES and EXAFS in both of these cases
confirms the formation of a palladium-zinc alloy (see Figure S13
and corresponding discussion there).
From Zn K-edge XANES (Figure 3d), the most significant variation
observed between these three systems (PdZn/SiO2, PdZn/Al2O3
and PdZn/ZnO/SiO2) lies in the magnitude of the low binding
energy feature at ~9658 eV. This feature we may associate with
the formation of either, reduced zinc (Zn0), and therefore the
presence of palladium-zinc alloys, and/or the presence of zinc
oxygen vacancies.[36] The double peak structure of the derivative
spectra (Figure 3e) could be indicative of the presence of both Zn0
and ZnII containing phases in each of the catalyst systems. We
might further intuit, from the relative magnitudes of the pre-edge
feature (Figure 3d), that the proportions of the two zinc states are
markedly different across the samples, with significantly higher
levels of reduced zinc being indicated for PdZn/SiO2 and
PdZn/Al2O3 compared to PdZn/ZnO/SiO2.
For PdZn/SiO2, PdZn/Al2O3 the phase-corrected FTs (Figure 3f)
are dominated by the scattering shell at ca 2.6 Å that we may
again associate (see also Table S5) with the significant presence
of the palladium-zinc alloy phase. In these two cases, this is
accompanied by a very broad feature in the FT, indicative of low
Z (modelled as oxygen) coordination to some fraction of the zinc
atoms. Given that the XANES from these two cases (Figure 3d)
points to a predominance of zinc being present in a reduced form,
this rather significant level of coordination might suggest a
segregation of the zinc toward the surface of the alloy
nanoparticles particles where they may be in contact with either
reactant species or the underlying support. Different from both of
these cases is the PdZn/ZnO/SiO2. Consistent with the reduced
5
This article is protected by copyright. All rights reserved.

10.1002/anie.202103087
Angewandte Chemie International Edition
RESEARCH ARTICLE
Figure 3. Pd K-edge XANES (a), the first derivatives of corresponding XANES spectra (b) as well as the corresponding Fourier transforms (c) of the k3-weighted
data of PdZn/SiO2, PdZn/Al2O3 and 2Pd-ZnO-np catalysts collected in situ after reduction in hydrogen at 260 °C and 15 bar pressure. Zn K-edge XANES (d), the
first derivatives of corresponding XANES spectra (e) as well as the corresponding Fourier transforms (f) of the k3-weighted data of PdZn/SiO2, PdZn/Al2O3, and
PdZn/ZnO/SiO2. The latter all measured in situ post reaction after catalytic carbon dioxide hydrogenation (260 °C, 30 bar).
Table 2 shows results of catalytic carbon dioxide hydrogenation
for the PdZn/SiO2 and PdZn/Al2O3 samples. Despite significant
carbon dioxide conversion (2.6-4.4 %), both materials produce
carbon monoxide as the main product, and methanol selectivity
for PdZn/SiO2 and PdZn/Al2O3 catalysts is correspondingly low,
at 11 and 6 % respectively. Thus, methanol productivity for both
materials is below 30 gMeOH·kgcat-1·hour-1, which is significantly
lower than attained in the 2Pd-ZnO-np case (vide supra). These
results permit a fundamental conclusion to be drawn: that the
presence of the palladium-zinc alloy phase of itself does not
guarantee that the selective hydrogenation of carbon dioxide to
methanol will result. PdZn/SiO2 and PdZn/Al2O3 exhibit significant
activity in the RWGS reaction, which produces carbon monoxide
and water.
zinc oxide support (the 2Pd-ZnO-np sample). Thus, considering
the results of the operando study of the 2Pd-ZnO-np catalyst, we
can conclude that palladium-zinc alloy itself does not provide the
active sites required to produce methanol from the direct
hydrogenation of carbon dioxide. Instead, we have shown that
zinc oxide phase is responsible for carbon dioxide activation,
while the palladium-zinc alloy in this bifunctional catalyst activates
hydrogen and enables the further hydrogenation of the formate
on zinc to methanol.[27,30] Our data therefore points to selective
catalytic activity taking place predominantly on the zinc oxide and
likely at the interface with the palladium-zinc alloy.
The addition of zinc oxide leads to a much improved selectivity to
methanol (Table 2) and PdZn/ZnO/SiO2 produces six times more
methanol per catalyst mass than the PdZn/Al2O3 or PdZn/SiO2.
Though similar, in terms of carbon dioxide conversion (3.6 % at
30 bar total pressure), the PdZn/ZnO/SiO2 sample, produces 184
gMeOH·kgcat-1·hour-1 as a result of the greatly enhanced selectivity
to methanol (49.8 %). Furthermore, at 50 bar pressure, the
performance level achieved in the PdZn/ZnO/SiO2 case, in terms
of activity and selectivity toward the production of methanol, is
very similar, and even slightly better, to that achieved using a bulk
6
This article is protected by copyright. All rights reserved.

10.1002/anie.202103087
Angewandte Chemie International Edition
RESEARCH ARTICLE
beamtimes. M.A.N. acknowledges Shell Global Solutions for the
partial funding of his position. Electron microscopy work was
performed at the Scientific Centre for Optical and Electron
Microscopy (ScopeM) ETH Zurich.
Table 2. Result of catalytic carbon dioxide hydrogenation over different
catalysts. Unless otherwise indicated, catalytic experiments were performed at
260 °C and 30 bar, catalyst mass: 100 mg, CO2:H2 ratio equals 1:3 and flowrate:
50 ml/min.
Catalyst
Carbon dioxide
conversion, %
Methanol
productivity,
gMeOH·kgcat
1·hour-1
Methanol
selectivity, %
Keywords: CO2 hydrogenation • methanol • PdZn alloy,
operando XAS • isotope-labelling experiment.
-
PdZn/SiO2
2.6
4.4
3.6
3.3
2.8
30
11.2
6.0
[1]
O. Martin, A. J. Martín, C. Mondelli, S. Mitchell, T. F. Segawa, R.
Hauert, C. Drouilly, D. Curulla-Ferré, J. Pérez-Ramírez, Angew.
Chemie Int. Ed. 2016, 55, 6261–6265.
PdZn/Al2O3
27
PdZn/ZnO/SiO2
[a]PdZn/ZnO/SiO2
[a]2Pd-ZnO-np
184
443
384
49.8
65.3
66.5
[2]
[3]
A. García-Trenco, A. Regoutz, E. R. White, D. J. Payne, M. S. P.
Shaffer, C. K. Williams, Appl. Catal. B Environ. 2018, 220, 9–18.
J. L. Snider, V. Streibel, M. A. Hubert, T. S. Choksi, E. Valle, D. C.
Upham, J. Schumann, M. S. Duyar, A. Gallo, F. Abild-Pedersen, T.
F. Jaramillo, ACS Catal. 2019, 9, 3399–3412.
[a] 2Pd-ZnO-np and PdZn/ZnO/SiO2 catalysts were also compared at 260 °C
and 50 bar total pressure, catalyst mass: 50 mg, CO2:H2 ratio equals 1:3 and
flowrate: 50 ml/min.
[4]
[5]
M. S. Frei, C. Mondelli, R. García-Muelas, K. S. Kley, B. Puértolas,
N. López, O. V. Safonova, J. A. Stewart, D. Curulla Ferré, J. Pérez-
Ramírez, Nat. Commun. 2019, 10, 3377.
H. Bahruji, M. Bowker, G. Hutchings, N. Dimitratos, P. Wells, E.
Gibson, W. Jones, C. Brookes, D. Morgan, G. Lalev, J. Catal. 2016,
343, 133–146.
Conclusion
[6]
[7]
[8]
[9]
[10]
F. Lin, X. Jiang, N. Boreriboon, Z. Wang, C. Song, K. Cen, Appl.
Catal. A Gen. 2019, 585, 117210.
In contrast to copper-zinc alloys, we have shown through the
application of operando XAS, that nanoparticulate palladium-zinc
is stable and does not undergo oxidative disruption under the
conditions required for the direct hydrogenation of carbon dioxide
to methanol. More significantly, however, through comparing the
behavior of catalyst systems specifically designed to yield
nanoparticulate palladium-zinc in the absence or presence of a
zinc oxide phase, we have shown that the palladium-zinc alloy
phase by itself is not particularly good at selectively hydrogenating
carbon dioxide to methanol. Instead, in the absence of a co-
existing zinc oxide phase, the palladium-zinc alloy primarily yields
carbon monoxide. When zinc oxide and the palladium-zinc alloy
nanoparticles are both present, the system becomes a far more
active and selective catalyst. At the highest pressures used in this
work, a catalyst based on zinc oxide and palladium-zinc alloy
supported on silica matches the methanol synthesis performance
of typical Pd/ZnO catalysts.
These data show that carbon dioxide hydrogenation to methanol
requires a multifunctional catalyst, with zinc oxide activating
carbon dioxide, and the palladium-zinc alloy splitting hydrogen. In
this system, as with the archetypal Cu/ZnO/Al2O3 case, selective
methanol synthesis is therefore the result of a synergy that exists
between an oxidized zinc phase, upon which active formates may
be hosted, and a metallic phase whose primary function is to
dissociate hydrogen. This hydrogen may then spillover to effect
the selective hydrogenation of the formates and result in the
desired synthesis of methanol, rather than the undesired
formation carbon monoxide.
J. Xu, X. Su, X. Liu, X. Pan, G. Pei, Y. Huang, X. Wang, T. Zhang,
H. Geng, Appl. Catal. A Gen. 2016, 514, 51–59.
H. Choi, S. Oh, S. B. Trung Tran, J. Y. Park, J. Catal. 2019, 376,
68–76.
Q. Tang, W. Ji, C. K. Russell, Z. Cheng, Y. Zhang, M. Fan, Z. Shen,
Appl. Energy 2019, 253, 113623.
E. M. Fiordaliso, I. Sharafutdinov, H. W. P. Carvalho, J. Kehres, J.
D. Grunwaldt, I. Chorkendorff, C. D. Damsgaard, Sci. Technol. Adv.
Mater. 2019, 20, 521–531.
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
[20]
[21]
[22]
D. Wu, K. Deng, B. Hu, Q. Lu, G. Liu, X. Hong, ChemCatChem
2019, 11, 1598–1601.
N. Iwasa, H. Suzuki, MasaoTerashita, M. Arai, N. Takezawa, Catal.
Letters 2004, 96, 75–78.
X.-L. Liang, X. Dong, G.-D. Lin, H.-B. Zhang, Appl. Catal. B Environ.
2009, 88, 315–322.
X.-L. Liang, J.-R. Xie, Z.-M. Liu, Catal. Letters 2015, 145, 1138–
1147.
Y. Yin, B. Hu, X. Li, X. Zhou, X. Hong, G. Liu, Appl. Catal. B
Environ. 2018, 234, 143–152.
M. Armbrüster, M. Behrens, K. Föttinger, M. Friedrich, É. Gaudry, S.
K. Matam, H. R. Sharma, Catal. Rev. 2013, 55, 289–367.
K. Föttinger, J. A. Van Bokhoven, M. Nachtegaal, G. Rupprechter, J.
Phys. Chem. Lett. 2011, 2, 428–433.
V. Lebarbier, R. Dagle, A. Datye, Y. Wang, Appl. Catal. A Gen.
2010, 379, 3–6.
H. Zhou, X. Yang, L. Li, X. Liu, Y. Huang, X. Pan, A. Wang, J. Li, T.
Zhang, ACS Catal. 2016, 6, 1054–1061.
Acknowledgements
F. Lin, X. Jiang, N. Boreriboon, Z. Wang, C. Song, K. Cen, Appl.
Catal. A Gen. 2019, 585, 117210.
We acknowledge the Swiss Light Source (SuperXAS beamline)
and ESRF (Swiss-Norwegian beamline) for providing access to
these facilities for XAS measurements. Dr. O. Safonova and Dr.
W. van Beek are acknowledged for local contacting during
L. Yang, Y. Guo, J. Long, L. Xia, D. Li, J. Xiao, H. Liu, Chem.
Commun. 2019, 55, 14693–14696.
M. W. Tew, H. Emerich, J. A. Van Bokhoven, J. Phys. Chem. C
2011, 115, 8457–8465.
7
This article is protected by copyright. All rights reserved.

10.1002/anie.202103087
Angewandte Chemie International Edition
RESEARCH ARTICLE
[23]
[24]
[25]
[26]
J. Díez-Ramírez, J. L. Valverde, P. Sánchez, F. Dorado, Catal.
Letters 2016, 146, 373–382.
J.-S. Chang, G. Férey, M. Daturi, Chem. - A Eur. J. 2012, 18,
11959–11967.
H. Bahruji, M. Bowker, W. Jones, J. Hayward, J. Ruiz Esquius, D. J.
Morgan, G. J. Hutchings, Faraday Discuss. 2017, 197, 309–324.
Y. Yin, B. Hu, X. Li, X. Zhou, X. Hong, G. Liu, Appl. Catal. B
Environ. 2018, 234, 143–152.
[48]
[49]
C.-H. Kim, J. S. Lee, D. L. Trimm, Top. Catal. 2003, 22, 319–324.
D. Lee, J. Y. Lee, J. S. Lee, Stud. Surf. Sci. Catal. 2004, 153, 169–
172.
[50]
[51]
J. Díez-Ramírez, P. Sánchez, A. Rodríguez-Gómez, J. L. Valverde,
F. Dorado, Ind. Eng. Chem. Res. 2016, 55, 3556–3567.
F. Liao, X.-P. Wu, J. Zheng, M. Meng, J. Li, A. Kroner, Z. Zeng, X.
Hong, Y. Yuan, X.-Q. Gong, S. Chi, E. Tsang, Green Chem. 2016,
19, 270.
M. Gentzen, D. E. Doronkin, T. L. Sheppard, A. Zimina, H. Li, J.
Jelic, F. Studt, J. Grunwaldt, J. Sauer, S. Behrens, Angew. Chemie
Int. Ed. 2019, 58, 15655–15659.
[27]
[28]
M. Zabilskiy, V. L. Sushkevich, D. Palagin, M. A. Newton, F.
Krumeich, J. A. van Bokhoven, Nat. Commun. 2020, 11, 2409.
E. Frei, A. Gaur, H. Lichtenberg, L. Zwiener, M. Scherzer, F.
Girgsdies, T. Lunkenbein, R. Schlögl, ChemCatChem 2020, 12, 1–
6.
[52]
[53]
[54]
S. E. Collins, J. J. Delgado, C. Mira, J. J. Calvino, S. Bernal, D. L.
Chiavassa, M. A. Baltanás, A. L. Bonivardi, J. Catal. 2012, 292, 90–
98.
H. Bahruji, M. Bowker, G. Hutchings, N. Dimitratos, P. Wells, E.
Gibson, W. Jones, C. Brookes, D. Morgan, G. Lalev, J. Catal. 2016,
343, 133–146.
[29]
[30]
[31]
[32]
E. Lam, G. Noh, K. Larmier, O. V. Safonova, C. Copéret, J. Catal.
2020, DOI 10.1016/j.jcat.2020.04.028.
M. Zabilskiy, V. L. Sushkevich, M. A. Newton, J. A. van Bokhoven,
ACS Catal. 2020, 14240–14244.
S. E. Collins, M. A. Baltanás, A. L. Bonivardi, J. Catal. 2004, 226,
410–421.
N. Binsted, EXCURV98, CCLRC Daresbury Laboratory computer
program, 1998.
O. P. Tkachenko, A. Y. Stakheev, L. M. Kustov, I. V Mashkovsky, M.
Van Den Berg, W. Grünert, N. Y. Kozitsyna, Z. V Dobrokhotova, V.
I. Zhilov, S. E. Nefedov, M. N. Vargaftik, I. I. Moiseev, Catal. Letters
2006, 112, 155–161.
[33]
[34]
A. S. Malik, S. F. Zaman, A. A. Al-Zahrani, M. A. Daous, H. Driss, L.
A. Petrov, Appl. Catal. A Gen. 2018, 560, 42–53.
A. A. Veligzhanin, Y. V. Zubavichus, N. Y. Kozitsyna, V. Y. Murzin,
E. V. Khramov, A. A. Chernyshov, J. Surf. Investig. 2013, 7, 422–
433.
[35]
[36]
[37]
N. Y. Kozitsyna, S. E. Nefedov, F. M. Dolgushin, N. V. Cherkashina,
M. N. Vargaftik, I. I. Moiseev, Inorganica Chim. Acta 2006, 359,
2072–2086.
E. Frei, A. Gaur, H. Lichtenberg, C. Heine, M. Friedrich, M. Greiner,
T. Lunkenbein, J. D. Grunwaldt, R. Schlögl, ChemCatChem 2019,
11, 1587–1592.
A. L. Bugaev, M. Zabilskiy, A. A. Skorynina, O. A. Usoltsev, A. V.
Soldatov, J. A. Van Bokhoven, Chem. Commun. 2020, 56, 13097–
13100.
[38]
[39]
[40]
[41]
[42]
[43]
[44]
[45]
[46]
[47]
W. Van Beek, O. V. Safonova, G. Wiker, H. Emerich, Phase
Transitions 2011, 84, 726–732.
P. J. Chupas, K. W. Chapman, C. Kurtz, J. C. Hanson, P. L. Lee, C.
P. Grey, J. Appl. Crystallogr. 2008, 41, 822–824.
P. M. Abdala, H. Mauroy, W. Van Beek, J. Appl. Crystallogr. 2014,
47, 449–457.
R. Frahm, M. Nachtegaal, J. Stötzel, M. Harfouche, J. A. Van
Bokhoven, J. D. Grunwaldt, AIP Conf. Proc. 2010, 1234, 251–255.
A. H. Clark, J. Imbao, R. Frahm, M. Nachtegaal, J. Synchrotron Rad
2020, 27, 551–557.
N. Binsted, PAXAS: program for the analysis of X-ray absorption
spectra, 1988.
B. Ravel, M. Newville, IUCr, J. Synchrotron Radiat. 2005, 12, 537–
541.
S. J. A. Figueroa, C. Prestipino, J. Phys. Conf. Ser. 2016, 712,
012012.
A. Rossberg, K.-U. Ulrich, S. Weiss, S. Tsushima, T. Hiemstra, A.
C. Scheinost, Environ. Sci. Technol. 2009, 43, 1400–1406.
S. Wuttke, P. Bazin, A. Vimont, C. Serre, Y.-K. Seo, Y. K. Hwang,
8
This article is protected by copyright. All rights reserved.

10.1002/anie.202103087
Angewandte Chemie International Edition
Entry for the Table of Contents
In this operando study of methanol synthesis over Pd/ZnO catalysts, we show that the palladium-zinc alloy phase by itself is
not particularly good at selectively hydrogenating carbon dioxide. Instead, and in the absence of a co-existing zinc oxide phase,
the palladium-zinc alloy primarily yields carbon monoxide. When zinc oxide and palladium-zinc alloy nanoparticles are both
present, the system becomes a far more active and selective catalyst.
9
This article is protected by copyright. All rights reserved.