Morphology-dependent interactions of ZnO with Cu nanoparticles at the materials' interface in selective hydrogenation of CO2 to CH 3OH

Article abstract of DOI:10.1002/anie.201007108

A good face: The (002) polar facet of platelike ZnO nanoparticles gives a much stronger electronic interaction with Cu nanoparticles than other facets (see picture; CB=conductance band; VB=valence band) and shows higher selectivity in the catalytic hydrogenation of CO₂ to methanol. This finding provides the basis for the rational design of new nanocatalysts for CO ₂ hydrogenation.

Full text of DOI:10.1002/anie.201007108

Communications
DOI: 10.1002/anie.201007108
Heterogeneous Catalysis
Morphology-Dependent Interactions of ZnO with Cu Nanoparticles at
the Materialsꢀ Interface in Selective Hydrogenation of CO2 to
CH3OH**
Fenglin Liao, Yaqun Huang, Junwei Ge, Weiran Zheng, Karaked Tedsree, Paul Collier,
Xinlin Hong,* and Shik C. Tsang*
In recent years, carbon dioxide (CO2) has become the focus of
much attention because of the position of CO2 as the primary
greenhouse gas and the implication of its emissions on the
problem of climate change.[1] Various sequestration technol-
ogies for CO2 abatement are being considered. It has been
recently demonstrated that hydrogen gas can be manufac-
tured on large scales from renewable sources, including solar
energy, hydropower, and biomass.[2] Thus, complete or partial
recycling of CO2 through its hydrogenation to high-energy-
density liquid fuels appears to be a very attractive approach.
As a result, catalytic CO2 hydrogenation reactions to meth-
anol, higher alcohols, gasoline, and related higher hydro-
carbons (Fischer–Tropsch-like reactions) have been receiving
much renewed attention. Particularly, the focus is on the
production of methanol, which is a key platform chemical for
present fuel and chemical infrastructures.[3] A recent assess-
ment of economic feasibility for this new green process
supports the possibility.[4]
CO2 and H2 in the carbon dioxide hydrogenation reaction also
promotes the reverse water gas shift reaction to give CO
[Eq. (2)], leading in general to a poorer methanol selectivity
CO2 þ H2 Ð CO þ H2O
ð2Þ
than the syngas route. It would be useful to gain an under-
standing of the structure–activity relationships at the Cu–
ZnO interface for these fundamentally important reactions.
Currently, morphology (shape) control in nanocatalysts
has clearly suggested that catalysis is controlled not only by
the chemical composition and size of the catalyst used but also
by the type of surface sites available at the catalyst surface.[8]
These considerations add exciting variables in tailoring the
properties of nanocatalysts for a wide range of catalytic
reactions. So far, only several examples of single-component
metal nanocrystals and semiconductor oxide nanocrystals
have been demonstrated,[8,9] but the role of shape control at
the materialsꢀ interface, which is the common feature in
practical metal-supported catalysts, is not yet known.
Herein, we present a significant shape effect of ZnO on its
interaction with copper in the synthesis of methanol from
carbon dioxide hydrogenation. The exposed polar (002) face
in platelike ZnO shows a much stronger material synergy with
copper than other crystal facets, which gives higher selectivity
towards methanol from CO2 hydrogenation. Electron para-
magnetic resonance (EPR) clearly indicates a strong elec-
tronic interaction at the interface of the two materials,
rendering it more selective for carbon dioxide activation
and hydrogenation. It is envisaged that this new finding could
lead to the rational design of new nanocatalysts for the
potential use in carbon dioxide hydrogenation.
The syntheses of ZnO nanocrystals of platelike and rod
shapes were based on published methods.[10,11] XRD analysis
confirmed that they all show crystalline wurtzite structure
(see the Supporting Information). The intensity of the (002)
peak of truncated platelike ZnO nanoparticles was higher
than other morphologies with the predominant polar facets
covering the structure, whereas the extended rod structure
showed relatively higher proportions of nonpolar (100) and
(101) planes along the [0001] rod axis, as previously
reported.[9]
The TEM image in Figure 1a reveals platelike ZnO with
the diameter of 50–60 nm, and Figure 1b shows 600–800 nm
rod-shaped ZnO with diameter of 20–30 nm. These images
demonstrate that the uniform ZnO nanoparticles of rod and
plate morphology were prepared by controlled nanochemical
Today, methanol is produced industrially from syngas
containing CO and CO2 (derived from fossil fuels) over Cu/
ZnO/Al2O3 catalysts.[5,6] These Cu/ZnO based systems are
also evaluated to be the most efficient catalysts for the direct
hydrogenation of CO2 [Eq. (1)].[7] Although the two reactions
CO2 þ 3 H2 Ð CH3OH þ H2O
ð1Þ
are likely to share a very similar mechanism (some works
suggested CO2 is the actual species for the production of
methanol from syngas[6]), the relatively high concentration of
[*] F. Liao, Y. Huang, J. Ge, W. Zheng, Dr. X. Hong, Prof. S. C. Tsang
College of Chemistry and Molecular Sciences
Wuhan University, Wuhan 430072 (China)
E-mail: edman.tsang@chem.ox.ac.uk
hongxl@whu.edu.cn
K. Tedsree, Prof. S. C. Tsang
Wolfson Catalysis Centre, Department of Chemistry
University of Oxford, Oxford, OX1 3QR (UK)
P. Collier
Johnson Matthey Technology Centre
Blount’s Court Road, Sonning Common, Reading, RG4 7NH (UK)
[**] This research was financially supported by the NSFC-20903074 and
the Fundamental Research Funds for the Central Universities. The
authors thank Dr. J. Harmer of Oxford for EPR spectra; Y.H. thanks
NSFC of China for a postdoctoral fellowship, and a studentship for
K.T. from the Thai government is acknowledged.
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201007108.
2162
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2011, 50, 2162 –2165

Figure 1. TEM images of a) ZnO plate particles; b) ZnO nanorods.
c) HRTEM image of ZnO rod. d) Crystallographic relationship between
ZnO plate and rod.
Figure 2. TPR curves of a) platelike; b) 400 nm rod ZnO; c) 600–
800 nm rod ZnO, and their corresponding mixtures with copper (d–f).
Testing the physical mixture of Cu nanoparticles with rod
and plate ZnO, respectively, for hydrogenation of CO2 was
therefore conducted. Table 1 clearly shows a consistently
higher selectivity to methanol reaching over 70% [Eq. (1)]
with lesser degree of CO formation from RWGS [Eq. (2)]
synthesis.[9–11] It has been long believed that Cu displays a
strong interaction with ZnO and that the materialsꢀ interface
can generate active sites in traditional methanol synthesis, but
there is a lack of direct experimental evidence. To character-
ize the interface, 35 nm Cu nanoparticles were synthesized
and physically mixed with ZnO nanoparticles to avoid
preferred deposition on any specific surface. It was antici-
pated that physical mixtures do not give as extensive and good
interfaces as industrial catalysts but they serve as model
materials for this study. The interactions between Cu and the
two forms of ZnO, respectively, were probed by temperature-
programmed reduction (TPR), as shown in Figure 2. There
are two types of reducible oxygen atoms in the ZnO structure,
which are reduced at 6008C and 7508C, respectively. Analyz-
ing the relative proportions of these two peaks with refer-
ences to their predominant exposed facets over these samples
suggest that the 7508C peak is associated with lattice oxygen
atoms from the dominant (002) polar facet in ZnO plate,
whereas the 6008C peak arises from oxygen reduction from
nonpolar facets (100, 110) in the rods. The 6008C peak
(nonpolar facets) changed marginally when mixed with Cu,
suggesting that the nonpolar surfaces display poor interaction
with the Cu phase. In a sharp contrast, the 7508C peak almost
disappeared when ZnO plate was mixed with Cu, and a new
broad reduction peak with onset from 2008C and peaked at
5008C, similar to Cu oxide reduction, was observed. Quanti-
tative analysis of the TPR profiles suggests that the addition
of Cu promoted the reduction of ZnO plate (see the
Supporting Information), because there are strong material
interactions between Cu and ZnO platelike forms. Methanol
synthesis over typical Cu–ZnO catalyst takes place at around
2508C and thus such facilitated [O] reduction at the materialsꢀ
interface may be somewhat linked with catalytic perfor-
mance.
Table 1: Catalytic performance of ZnO plate and rod particles mixed with
copper and alumina in the synthesis of methanol from hydrogenation of
CO2.[a]
CO2:H2
T [K] Catalyst
CO2
Methanol
(mole ratio)
conversion selectivity
[%]
[%]
1:2.2
1:2.5
543
553
543
553
Cu/rod ZnO/Al2O3
12.3
10.9
15.3
12.0
15.8
15.5
17.8
14.7
42.3
72.7
39.1
71.6
41.0
64.5
32.0
63.3
Cu/plate ZnO/Al2O3
Cu/rod ZnO/Al2O3
Cu/plate ZnO/Al2O3
Cu/rod ZnO/Al2O3
Cu/plate ZnO/Al2O3
Cu/rod ZnO/Al2O3
Cu/plate ZnO/Al2O3
[a] Total flow rate: 40 stpmLminꢁ1 at 4.5 mPa.
when ZnO plate is in contact with Cu. In contrast, when the
ZnO rod is used, slightly higher CO2 conversion but much
lower methanol selectivity (ca. 40%), is obtained. Thus, the
testing results clearly suggest that the materialsꢀ interface
between ZnO plate and Cu is crucially essential in methanol
production from CO2 hydrogenation. It is important to
elucidate the nature of interactions between ZnO plate and
Cu. ZnO is a well known n-type semiconductor (band gap, Eg
ꢀ 3.3 eV at 300 K); its mobile electrons at conductive band
(CB) may transfer to the Fermi level of metal (i.e. Cu) with
high work function at the interface.[12,13]
Angew. Chem. Int. Ed. 2011, 50, 2162 –2165
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
2163

Communications
Table 2: Quantitative changes of EPR peaks (g=2.00; g=1.96) in the
two forms of ZnO and their mixtures with copper normalized per mg of
ZnO.
Sample
Spin count [mgꢁ1 ZnO],
Spin count [mgꢁ1 ZnO],
g=2.00
g=1.96
Rod ZnO
Cu/rod ZnO
Plate ZnO
2.05ꢀ1013
2.71ꢀ1013
6.98ꢀ1013
6.14ꢀ1013
0
0
1.51ꢀ1014
2.53ꢀ1013
Cu/plate ZnO
plate (see Table 2). It is therefore evident that there is
electron transfer from the CB of ZnO plate to Cu. Thus, this
work offers direct evidence for the existence of strong
electronic interactions between the two materials, and the
information is complimentary to the XPS data.
The mechanism of methanol synthesis over Cu/ZnO by
the syngas route is still controversial despite its industrial
practice for many years. The change in the feedstock to CO2/
H2 will add further complications. It has been proposed that
ZnO hosts highly dispersed Cu nanoparticles, which provide
exclusively the active sites for methanol production from
syngas. However, the nonlinearity frequently observed in
correlations of Cu surface area and catalytic activity suggests
the existence of a strong Cu–ZnO interaction at various
material interfaces.[16] Consequently, Frost stressed the impor-
tance of Schottky barrier establishment at the Cu–ZnO
interface.[12] According to his view, thermal electrons will flow
from the CB of ZnO to the Fermi level of Cu, which facilities
the formation of oxygen vacancies by shifting the equilibrium
[O]Ð[ ] + 1/2O2 + 2eꢁ, which can account for high activity
in methanol synthesis from syngas.[12] Our direct observation
of electron transfer from ZnO to Cu at ambient temperature
by XPS and EPR analysis is consistent with his mechanism.
At elevated temperatures, one would expect more thermal
electrons from ZnO to overcome Schottky barrier (estimated
to be as low as 0.45 eV[12]) to the copper surface and generate
more oxygen vacancies at the materialsꢀ interface. This work
also contributes two further important points. First, the
oxygen atom driven out from the ZnO lattice at the materialsꢀ
interface is retained in the same mixture (higher reduction
peaks in TPR) and seems to have migrated alongside with the
electron flow to the copper as surface CuO (redox reaction
between the two materials). The TPR results clearly show
facilitated oxygen reduction at the interface, possibly the
reduction of ZnO via surface Cu oxide by spilled hydrogen
(activated on zinc oxide) at much lower temperature. This
situation would therefore overcome the above equilibrium
constraints in the extent of electron transfer, and the
formation of vacancies between the two materials by catalytic
reduction. In addition, CO2 is likely to be activated by the
generated vacancies and the Cu phase at the interface assists
molecular rearrangement (formate, dioxomethylene, formal-
dehyde, and methoxy) and hydrogenation to give metha-
nol.[17] Secondly, we have shown that the physiochemical
changes at the interface take place primarily between ZnO
plate with polar (002) facets and the Cu particle. It is not yet
known what unique roles the polar facets play, but these may
have arisen from different surface electronic structures.
Figure 3. XPS data of a) Zn 2p3/2; b) O 1s of plate and rod ZnO
nanoparticles; c) O 1s of plate ZnO and its mixture with copper; and
d) Cu 2p3/2 of metallic copper and its mixture with plate ZnO.
Characterization of ZnO plate, rod, and their mixture
with copper was therefore conducted by X-ray photoelectron
spectroscopy (XPS). Figure 3a reveals that the binding
energy (BE) of Zn 2p3/2 is about 1021 eV for all the samples,
which agrees with reported Zn2+.[14] However, Figure 3b
shows that the BE value of O 1s for the plate sample
(530.1 eV) is slightly lower than that of the rod (530.4 eV),
implying electrons are easier to be excited from the plate
sample. The BE value of O1s of the plate form after addition
of Cu reverted to that of rod form (Figure 3c). There is also a
corresponding down-shift of Cu 2p3/2 with reference to
932.8 eV Cu0 (Figure 3d). In contrast, the BE value of O 1s
of the rod sample remains constant regardless of whether it is
in contact with Cu (see Figure S2 in the Supporting Informa-
tion). The [O] terminated polar facet is known to be electron
richer than those of nonpolar facets (i.e. 100) that contain
equal number of cations and anions.[13] There is indeed
electron transfer from ZnO plate to Cu as a result of the
formation of a Schottky–Mott junction[12] at the interface, thus
affecting the BE values.
Further characterization of ZnO samples with and with-
out Cu using electron paramagnetic resonance (EPR) was
carried out. Figure S3 (see the Supporting Information)
reveals two intense signals for plate ZnO at g = 2.00 and g =
1.96, whereas only the former signal at lower intensity is
detected for rod sample. The signal at g = 2.00 has commonly
been assigned to the unpaired electrons deeply trapped in
oxygen vacancies, probably via adsorbed oxygen species from
ꢁ
air (O2 ), and the signal at g = 1.96 represents unpaired
electrons trapped from the CB by shallow donors or
impurities.[15] Thus, some strongly and weakly trapped elec-
trons (excited electrons in the CB) were detected on the ZnO
polar facets. The plate form appears to give more oxygen
defects than the rod form under identical treatment. After the
addition of copper, the electron spin count of the signal
normalized per milligram of ZnO at g = 2.00 showed a
marginal change for both types of the ZnO samples, and the
signal at g = 1.96 decreased substantially in the case of ZnO
2164
www.angewandte.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2011, 50, 2162 –2165

in vacuum at 1008C for 2 h. Catalyst tests in the hydrogenation of CO2
were carried out at a total pressure of 4.5 mPa using a tubular fixed
bed reactor (12.7 mm outside diameter) with a catalyst weight of
0.3 g. CO2/H2 reaction mixtures with molar ratios of 1:2.2 and 1:2.5
were fed at a rate of 40 stpmLminꢁ1 (stp = standard temperature and
pressure; P = 101.3 kPa, T= 298 K) through the catalyst bed. Before
each test, the catalyst was pre-reduced in situ at 573 K for 2 h under
the H2 flow (20 stpmLminꢁ1). The products were analyzed by a gas
chromatograph equipped with a thermal conductivity detector.
Clearly, the ZnO plate gives a higher degree of electron
trapping as shown in our experiments, which must be
intimately related to its defect formation, surface morphol-
ogy, and electronic properties.[13] Regarding design of future
catalysts for the CO2 hydrogenation reaction, it appears to be
crucial to maximize their interfacial area. Without the
appropriate interface (for example, nonpolar facets or polar
facets in remote distance from Cu), we show that CO2 can still
be activated and hydrogenated on oxygen vacancies of ZnO,
but the main products would be CO and H2O [Eq. (2)], which
can be prevalent at high CO2 coverage under the reaction
conditions. Traditionally, it is difficult to gain an understand-
ing of the interface of materials in industrial catalysts.[18] With
the introduction of nanosciences to material synthesis, it is
shown to be possible to separate complex factors in catalyst
formulation (size, shape, support effects) and study them
individually in a systematic manner. Herein, we have clearly
demonstrated the importance of morphology-controlled
SMSI (strong metal–support interaction), which plays a
great influence in selectivity for this important reaction.
Received: November 12, 2010
Published online: January 20, 2011
Keywords: copper · heterogeneous catalysis · hydrogenation ·
.
interfaces · zinc
[1] K. Man, K. Yu, I. Curcic, J. Gabriel, S. C. E. Tsang, ChemSus-
Chem 2008, 1, 893 – 899.
[2] J. Turner, G. Sverdrup, M. K. Mann, P. C. Maness, B. Kroposki,
M. Ghirardi, R. J. Evans, D. Blake, Int. J. Energy Res. 2008, 32,
379 – 407.
[3] L. K. Rihko-Struckmann, A. Peschel, R. Hanke-Rauschenbach,
K. Sundmacher, Ind. Eng. Chem. Res. 2010, 49, 11073 – 11078.
[4] G. A. Olah, Angew. Chem. 2005, 117, 2692 – 2696; Angew. Chem.
Int. Ed. 2005, 44, 2636 – 2639.
Experimental Section
The synthesis of platelike ZnO particles was based on the method by
Zhu and co-workers.[10] Zinc acetate dihydrate (Zn(CO2CH3)2·2H2O,
9.0 g) was dissolved in deionized water (72 mL) and then hexame-
thylenetetramine (C6H12N4, 5.76 g) was added. The solution was
stirred for 10 min, transferred into a 200 mL teflon-lined autoclave
and maintained at 978C for 24 h, and then cooled to room temper-
ature slowly. The white precipitate was collected by centrifugation at
5000 rpm for 10 min, after which the supernatant was decanted and
discarded. The solid was washed repeatedly with ethanol and water to
remove excess precursor. Rodlike ZnO was synthesized according to
the method of Liu and Zeng.[11] Zinc nitrate (Zn(NO3)2·6H2O,
1.487 g) and NaOH (6 g) were dissolved in deionized water (10 mL;
molar ratio of Zn2+ to OHꢁ 1:30). Pure C2H5OH (100 mL) was added,
followed by ethylenediamine (C2H4(NH2)2, 5 mL). The mixed solu-
tion was then transferred to a covered plastic container with a volume
capacity of 250 mL. The container was kept at room temperature
under constant stirring. The length of the rods was controlled by the
reaction time: 1 day for the short rods and 3 days for the long ones.
After the synthesis, a white crystalline product was centrifuged and
washed with deionized water and pure ethanol. The precipitate was
dried at 608C for 12 h. All the samples were calcined at 723 K for 2 h.
The catalysts were prepared by mixing equal quantities of ZnO, Cu,
and Al2O3 mechanically using pestle and mortar. Cu was synthesized
as follows: 0.8 g copper acetate dihydrate (Cu(CO2CH3)2·2H2O) was
dissolved in ethanol (60 mL) and NaBH4 solution (0.4 g NaBH4 in
10 mL ethanol) was added to reduce Cu2+. The black precipitate was
collected by centrifugation at 5000 rpm for 10 min. The solid was
washed repeatedly with ethanol and water. The precipitate was dried
[5] J. G. Wu, M. Saito, M. Takeuchi, T. Watanabe, Appl. Catal. A
2001, 218, 235 – 240.
[6] G. C. Chinchen, P. J. Denny, J. R. Jennings, M. S. Spencer, K. C.
Waugh, Appl. Catal. 1988, 36, 1 – 65.
[7] M. Saito, Catal. Surv. Jpn. 1998, 2, 175 – 184.
[8] R. Narayanan, M. A. El-Sayed, J. Phys. Chem. B 2005, 109,
12663 – 12676.
[9] A. Mclaren, T. Valdes-Solis, G. Q. Li, S. C. Tsang, J. Am. Chem.
Soc. 2009, 131, 12540 – 12541.
[10] G. R. Li, T. Hu, G. L. Pan, T. Y. Yan, X. P. Gao, H. Y. Zhu, J.
Phys. Chem. C 2008, 112, 11859 – 11864.
[11] B. Liu, H. C. Zeng, Langmuir 2004, 20, 4196 – 4204.
[12] J. Frost, Nature 1988, 334, 577 – 580.
[13] H. Morkoꢁ, ꢂ. ꢃzgꢄr, Zinc Oxide: Fundamentals, Materials and
Device Technology, 1st ed, Wiley-VCH, Berlin, 2009.
[14] Y. G. Chang, J. Xu, Y. Y. Zhang, S. Y. Ma, L. H. Xin, L. N. Zhu,
C. T. Xu, J. Phys. Chem. C 2009, 113, 18761 – 18767.
[15] V. Ischenko, S. Polarz, D. Grote, V. Stavarache, K. Fink, M.
Driess, Adv. Funct. Mater. 2005, 15, 1945 – 1954.
[16] K. C. Waugh, Catal. Lett. 1999, 58, 163 – 165.
[17] S. A. French, A. A. Sokol, S. T. Bromley, C. R. A. Catlow, P.
Sherwood, Top. Catal. 2003, 24, 161 – 172.
[18] T. Ressler, B. L. Kniep, I. Kasatkin, R. Schlogl, Angew. Chem.
2005, 117, 4782 – 4785; Angew. Chem. Int. Ed. 2005, 44, 4704 –
4707; and also Y. Lin, Y. Hsu, S. Chen, Y. Lin, L. Chen, K. Chen,
Angew. Chem. 2009, 121, 7722 – 7726; Angew. Chem. Int. Ed.
2009, 48, 7586 – 7590.
Angew. Chem. Int. Ed. 2011, 50, 2162 –2165
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
2165