Structure, bonding, and catalytic activity of monodisperse, transition-metal-substituted CeO2 nanoparticles

Article abstract of DOI:10.1021/ja509214d

We present a simple and generalizable synthetic route toward phase-pure, monodisperse transition-metal-substituted ceria nanoparticles (M_0.1Ce_0.9O_2-x, M = Mn, Fe, Co, Ni, Cu). The solution-based pyrolysis of a series of he

Full text of DOI:10.1021/ja509214d

Article
pubs.acs.org/JACS
Structure, Bonding, and Catalytic Activity of Monodisperse,
Transition-Metal-Substituted CeO2 Nanoparticles
Joseph S. Elias,*,† Marcel Risch,‡ Livia Giordano,⊥,∥ Azzam N. Mansour,# and Yang Shao-Horn*,‡,§,∥
‡
§
†Department of Chemistry, Research Laboratory of Electronics, Department of Materials Science and Engineering, and
∥Department of Mechanical Engineering, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, United States
⊥Dipartimento di Scienza dei Materiali, Universita
̀
di Milano-Bicocca, Via R. Cozzi 53, 20125 Milano, Italy
#Naval Surface Warfare Center, Carderock Division, 9500 MacArthur Boulevard, West Bethesda, Maryland 20817-5700, United States
S
* Supporting Information
ABSTRACT: We present a simple and generalizable synthetic route toward phase-
pure, monodisperse transition-metal-substituted ceria nanoparticles (M0.1Ce0.9O2−x, M
= Mn, Fe, Co, Ni, Cu). The solution-based pyrolysis of a series of heterobimetallic
Schiff base complexes ensures a rigorous control of the size, morphology and
composition of 3 nm M0.1Ce0.9O2−x crystallites for CO oxidation catalysis and other
applications. X-ray absorption spectroscopy confirms the dispersion of aliovalent (M3+
and M2+) transition metal ions into the ceria matrix without the formation of any bulk
transition metal oxide phases, while steady-state CO oxidation catalysis reveals an order
of magnitude increase in catalytic activity with copper substitution. Density functional
calculations of model slabs of these compounds confirm the stabilization of M3+ and
M2+ in the lattice of CeO2. These results highlight the role of the host CeO2 lattice in
stabilizing high oxidation states of aliovalent transition metal dopants that ordinarily
would be intractable, such as Cu3+, as well as demonstrating a rational approach to catalyst design. The current work
demonstrates, for the first time, a generalizable approach for the preparation of transition-metal-substituted CeO2 for a broad
range of transition metals with unparalleled synthetic control and illustrates that Cu3+ is implicated in the mechanism for CO
oxidation on CuO-CeO2 catalysts.
substitution in CeO2 on the kinetics of CO oxidation
necessitates the preparation of a library of these catalysts with
superior homogeneity in phase, crystallite size and crystallo-
graphic termination. The presence of inseparable impurities
complicates even the identification of the oxidation state and
local structure of M in MyCe1−yO2−x. Taking a notable example
from heterogeneous catalysis, traditional methods for the
preparation of the solid solution CuyCe1−yO2−x, including
coprecipitation4,12−15 and excess-solution impregnation16−18
techniques, lead to the formation of the solid solution in
addition to dispersed CuO phases in these so-called CuO-CeO2
catalysts, which prevents the rigorous identification of the
catalytic site for CO oxidation.19,20
Here we report a generalizable synthetic strategy for the
preparation of monodisperse, phase-pure nanocrystals of
M0.1Ce0.9O2−x with aliovalent transition metal substitution (M
= Mn, Fe, Co, Ni and Cu). X-ray absorption spectroscopy at
the transition metal L-, K-, cerium L- and oxygen K-edges
confirms that the transition metal species are stabilized in the
3+ oxidation state upon incorporation into the CeO2 lattice, in
contrast to previous studies on CuO-CeO2.4,15 CO oxidation
on Cu0.1Ce0.9O2−x reveals that Cu species in the lattice are
responsible for an increase in catalytic activity by an order of
INTRODUCTION
■
Because of the interesting physical and chemical properties of
cerium(IV) oxide (CeO2, ceria), including reversible surface
oxygen ion exchange1,2 and electronic/ionic conductivity,3
ceria-based materials have been studied for a number of
applications including the oxidation of carbon monoxide (CO)
and hydrocarbon-based fuels,4 the water−gas shift reaction,5
photoinduced water splitting,6 oxygen-conducting membranes
for solid-oxide fuel cells,7 flue-gas desulfurization reactions,8 the
production of hydrogen by the steam reforming of methanol,9
and sensor applications.10 Introduction of redox active cations,
especially of aliovalent transition metals, into the lattice of
CeO2 to form the solid solution MyCe1−yO2−x imparts added
functionality to this class of materials. The dramatic increase in
activity for the water−gas shift reaction with incorporation of
nonmetallic Au and Pt species into the lattice of CeO2 is a
notable example,5 while applications of substitution into CeO2
outside of heterogeneous catalysis include band gap engineer-
ing in photovoltaic devices.11 Unfortunately, traditional
methods for the preparation of aliovalent substituted CeO2,
especially with transition metal substitution, often leads to the
formation of multiple phases, polydisperse crystallite sizes and
disparate crystallite habits. The presence of impurity phases
along with differences in crystallite size and surface termination
all influence the kinetics of CO oxidation on MyCe1−yO2−x
catalysts. To establish the influence of transition metal
Received: September 6, 2014
© XXXX American Chemical Society
A
dx.doi.org/10.1021/ja509214d | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
magnitude in comparison to undoped, nanocrystalline CeO2.
This finding provides insights into developing novel strategies
to design highly active and uniform catalysts for the oxidation
of CO and hydrocarbon fuels and for other applications.
Single crystals of M-CeIII-3-MeO-salpn (M = CoIII, NiII,
CuII) suitable for X-ray crystal structure analysis were obtained
from the slow evaporation of the complexes in acetonitrile
solutions. The crystal structures of the novel M-CeIII-3-MeO-
salpn complexes were found to share similar structural motifs
(Figures 1 and S1, Supporting Information). Coordination
around the transition metal site is square planar with respect to
the N2O2 pocket of the 3-MeO-salpn ligand. 6-Fold
coordination to the nickel center of NiII-CeIII-3-MeO-salpn
is completed by two axially coordinated acetonitrile molecules
(Figure 1) while the 6-fold coordination of the cobalt center of
CoIII-CeIII-3-MeO-salpn is achieved by two acetate groups that
bridge with the cerium center (Figure S1). The copper center is
5-coordinate with a long (2.4236 Å) axial Cu−O bond to a
nitrate group that bridges the two asymmetric units that make
up the unit cell of CuII-CeIII-3-MeO-salpn (Figure S2).
Trivalent cerium ions were shown to occupy the lacunary O4
site, bridging the transition metal via a pair of μ2-phenoxy
groups, resulting in short M-Ce interatomic distances (3.3613,
3.5617, and 3.5848 Å for the cobalt, nickel and copper crystal
RESULTS AND DISCUSSION
■
Synthesis of M0.1Ce0.9O2−x. Monodisperse, hexanes-
soluble single crystallites of oleylamine-stabilized
M0.1Ce0.9O2−x (M = Mn, Fe, Co, Ni, Cu) were prepared
according to Figure 1. The synthetic scheme involves the
1
structures, respectively). FTIR (Figure S3), H NMR (Figures
S4−S10) and elemental analysis (see Experimental Methods in
the Supporting Information) confirm the formulas we report
here for MnII-CeIII-3-MeO-salpn, FeIII-CeIII-3-MeO-salpn and
CeIII-3-MeO-salpn.
With a library of homologous heterobimetallic precursors in
hand, monodisperse M0.1Ce0.9O2−x and CeO2 nanoparticles
were prepared through the pyrolysis of the precursors in
oleylamine solutions under an argon atmosphere (Figure 1).
Powder X-ray diffraction (PXRD) patterns for as-synthesized
M0.1Ce0.9O2−x (Figure 2) indicate a single crystalline phase,
Figure 1. Synthesis of M0.1Ce0.9O2−x from heterobimetallic 3-MeO-
salpn complexes. Representative crystal structures for NiII-CeIII-3-
MeO-salpn and CuII-CeIII-3-MeO-salpn are included, with thermal
ellipsoids plotted at 50% probability and hydrogen atoms omitted for
clarity.
Figure 2. PXRD patterns for as-synthesized nanocrystalline
M0.1Ce0.9O2−x and CeO2 compounds taken at room temperature
(copper Kα source).
preparation and subsequent decomposition of a library of
heterobimetallic Schiff base complexes21−24 derived from the
ligand N,N′-bis(3-methoxysalicylidene)-propylene-1,3-diamine
(H2-3-MeO-salpn). After the initial coordination of the
transition metal into the inner N2O2 site of H2-3-MeO-salpn,
the outer O4 lacunary site is poised for coordination of hard
Lewis-acidic rare earth ions, such as Ce3+ (Figure 1). The
choice of heterobimetallic Schiff base complexes as inorganic
precursors was made in order to ensure close association
between cerium and transition metal ions during the nucleation
and growth of the M0.1Ce0.9O2−x crystallites in solution, thereby
preventing the formation of separate bulk transition metal oxide
phases.
attributed to the fluorite structure of CeO2 (space group #225,
Fm3m). Scherrer analysis of peak broadening shows average
̅
crystallite sizes of 2−3 nm for M0.1Ce0.9O2−x and CeO2
(Figures S11−12A,C,E and Table S2). Profile fitting of
PXRD data revealed that CeO2 nanoparticles have a lattice
parameter a (5.42 Å) that is slightly larger than that of bulk
CeO2 (5 μm) found in this study and in previous work (5.41
Å),25 which can be attributed to the presence of Ce3+ and
oxygen vacancies on the particle surface.26,27 The lattice
expansion found for 3 nm CeO2 is not as drastic as reported
earlier for similarly sized ceria nanoparticles (5.434328 and
5.5420 Å25). The disparate synthesis and annealing conditions
B
dx.doi.org/10.1021/ja509214d | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
reported elsewhere25,28 may give rise to different amounts of
oxygen vacancies in ceria, and help explain such a discrepancy
in reported lattice parameter values in nanoparticulate CeO2.
Similarly, the lattice parameters found for M0.1Ce0.9O2−x
nanoparticles (Table S2) were also larger than that found for
5 μm CeO2. HRTEM particle-size analysis (Figure 3) confirms
substitution as determined by EDS with a typical experimental
uncertainty of 10% in the measured values is in reasonable
agreement with the values determined by ICP-AES. Since
making CuyCe1−yO2−x with percent substitution y greater than
10% leads to the formation of multiple phases, we set the limit
of substitution for all transition metal samples to y ≈ 10%.
Identification of the Local Structure of Transition
Metal Ions in M0.1Ce0.9O2−x. In order to identify the local
atomic structure of these catalysts, especially of the redox-active
transition metal centers, we turned our attention to X-ray
absorption spectroscopy (XAS). The general features of the Ce
LIII-edge spectra (Figures 4A,B and S18−20A,B and Tables S4
and S5) are similar to that of nanocrystalline CeO2, indicating
that the local coordination and oxidation-state of cerium is
maintained in M0.1Ce0.9O2−x. In the XANES region (Figure 4A
and S18−20A) the intensity of the white line at 5729.6 eV
decreases for all M0.1Ce0.9O2−x and 3 nm CeO2 compared to
bulk (5 μm) CeO2, which can be attributed to the decrease in
covalency between Ce(4f) and O(2p) found as the particle size
of CeO2 decreases.31 Importantly, the intensity of the shoulder
at 5724.5 eV, which is attributed to surface Ce3+,31 does not
change with transition metal substitution. Also noteworthy is
that the magnitude of the scattering path (R = 2.3 Å)
corresponding to the first coordination shell in the EXAFS
(Figures 4B, S18−S20B and Tables S4 and S5, representing
eight-coordinate cerium in CeO2) is lower for M0.1Ce0.9O2−x
and CeO2 nanoparticles than for bulk (5 μm) CeO2. Along
with this, we found that the magnitudes of the next-nearest
neighbor scattering paths (Ce−Ce and Ce−O distances,
respectively, R = 3.9 and 4.6 Å) also decrease slightly with
transition metal substitution (Figures 4B, S18−S20B and Table
S4). These results, which we also found for 3 nm CeO2
nanoparticles, support the formation of higher concentrations
of undercoordinated surface CeO2 sites with decreasing particle
size, which is in good agreement with previous work on CeO2
nanoparticles.31 It should noted here, as a caveat, that the
analysis of the EXAFS of Ce LIII spectra of materials based on
CeO2 is rather tenuous.31 The fact that the spectra exhibit two
white lines suggests, as has previously been discussed at
length,32,33 a many-body electronic ground state in CeO2 (i.e.,
one which exhibits substantial configuration interaction
between multiple electronic configurations). Therefore, the
one-electron theory often used in EXAFS simulations34 does
not apply to such oxides. Furthermore, since both the mean
squared displacement (σ2) and scattering path degeneracy (N)
contribute to the magnitude of the EXAFS, a rigorous
determination of coordination number could not be accom-
plished without temperature-dependent studies. The lower
intensity magnitudes in the EXAFS are likely due to a
combination of smaller values for N (as in Table S4) and
larger values for σ2 (as in Table S5), both of which are
consistent with aliovalent substitution into the CeO2 lattice.
This comparison of the Ce LIII-edge XAS spectra confirms
that the local geometric and electronic structure of Ce in
M0.1Ce0.9O2−x remains unaltered compared to CeO2 nano-
particles. This assertion is further supported by spectra at the
oxygen K-edge (Figure 5A). Spectra for Cu0.1Ce0.9O2−x are
identical to unsubstituted 3 nm CeO2, exhibiting peaks at 533.7
eV without features at lower energies, which reflects that the
local density of states around the oxygen ligands is
predominantly unperturbed compared to CeO2. In contrast,
the O K-edge spectra for YBa2Cu3O7−δ (YBCO) and KCuO2
exhibit peaks at 532 eV. The peaks at 532 eV are assigned to
Figure 3. TEM, histogram of particle sizes, HRTEM and FFT images
of as-synthesized M0.1Ce0.9O2−x
.
a narrow size distribution of 3.0 ( 0.4) nm crystallites for
M0.1Ce0.9O2−x. Closer inspection by HRTEM (Figure 3, right
panels) suggests that the individual crystallites exhibit a
truncated octahedral crystal habit, with crystal planes
terminated in eight {111} hexagonal facets and four {100}
square facets (Figure S13B,C,D), a common crystallite shape
reported for ceria nanoparticles.29
Before further characterization, the M0.1Ce0.9O2−x samples
were annealed in air at 400 °C to remove the capping
oleylamine layer, as revealed by thermogravimetric analysis
(TGA, Figure S14). Preliminary ambient pressure XPS
measurements of M0.1Ce0.9O2−x at 400 °C in 75 mTorr O2
suggests that all carbon is removed from the surface of these
catalysts after annealing. PXRD (Figures S11−12B,D,F and S15
and Table S2) and HRTEM (Figure S16) suggest that during
the annealing treatment, the mean size of the crystallites
increases by as much as 1 nm, presumably through an Ostwald
ripening or oriented agglomeration process.30 Energy dispersive
X-ray spectroscopy (EDS) confirms that the average transition
metal substitution is between 8 and 14 mol % (Figure S17 and
Table S2), which is verified by inductively coupled plasma
atomic emission spectroscopy (ICP-AES, Table S2). Percent
C
dx.doi.org/10.1021/ja509214d | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Figure 4. Cerium LIII-edge and transition metal K-edge XAS spectra for annealed CeO2, Cu0.1Ce0.9O2−x and Fe0.1Ce0.9O2−x samples. (A) Cerium LIII-
edge XANES spectra and (B) the magnitude of the k2-weighted Fourier transform of the EXAFS of Cu0.1Ce0.9O2−x (purple), Fe0.1Ce0.9O2−x (green)
and ceria references. (C) Iron K-edge XANES spectra and (D) the magnitude of the k2-weighted Fourier transform of the EXAFS of Fe0.1Ce0.9O2−x
(green) and relevant iron oxides. (E) Copper K-edge XANES spectra and (F) the magnitude of the k2-weighted Fourier transform of the EXAFS of
Cu0.1Ce0.9O2−x (purple) and relevant copper oxides. Ce LIII-edge spectra were collected in transmission mode, while Cu and Fe K-edge spectra were
collected in total fluorescence yield mode. Lighter traces in (B), (D) and (F) are results of EXAFS fitting (see Tables S5 and S6).
transitions into holes in the 2p band of oxygen, an indication of
charge-transfer in formally trivalent copper compounds.35,36
Any contributions from oxygen ligands associated with Cu ions
to the XAS spectrum either share features with the spectra of
CuO or CeO2 or are too low in intensity to be distinguished
from the spectrum of CeO2.
The transition metal K-edge XAFS spectra for M0.1Ce0.9O2−x
are presented in Figures 4C−F and S18−20C,D. Analysis of the
edge-shift in the XANES region (Figure S21) suggests the
presence of formally trivalent transition metal substitution. In
addition, EXAFS fitting (Figures 4D,F and S19−20D and Table
S6) indicates that the transition metal ions incorporate into the
lattice of CeO2 either as isolated ions or as small oxide clusters
rather than form extended MOx phases. Using the binary
transition metal oxides (Fe2O3, Co3O4 and NiO) as starting
models for EXAFS fitting (Table S6), we found that oxygen
coordination around the transition metal was significantly
smaller (∼4) than expected for the binary oxides (6).
Additionally, the number of nearest neighbor transition metal
centers was found to be much smaller (<1) than expected for
extended crystalline MOx phases. It should be noted that the
model did not fit the low-magnitude data well after apparent
distances longer than 3 Å, which suggests a spatial averaging of
several different secondary coordination sphere configurations
in the CeO2 host lattice. These data show that our synthetic
approach is generalizable; a comparison of the intrinsic
properties as a function of transition metal substitution is
meaningful as the M0.1Ce0.9O2−x powders are free from any
“bulk” transition-metal-oxide impurity phases.
The Cu K-edge XAS spectra of Cu0.1Ce0.9O2−x show some
surprising features in Figure 4E,F. The edge position for
Cu0.1Ce0.9O2−x is a full 3.2 eV higher than that of CuO, which
indicates a formal valence higher than 2+ (a mixture of 2+ and
3+, Figure 4E). Indeed, the edge energy (taken as the energy at
half step-height) corresponds to a formal oxidation state of 2.9+
when fit to a calibration curve of the edge energies of copper
oxide references (Figure S21). The weak pre-edge feature at
8977 eV corroborates the presence of Cu3+; this feature
D
dx.doi.org/10.1021/ja509214d | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
data (probing the transition 2p → 3d) can provide information
on the density of states of the copper 3d band. The Cu LII,III
-
edge XAS spectra for Cu0.1Ce0.9O2−x and reference compounds
(CeO2, CuO, YBCO and KCuO2), collected in the partial
fluorescence yield (PFY) mode, are given in Figure 5B. The
prominent local maximum at 931 eV for Cu0.1Ce0.9O2−x has
been assigned to the 2p63d9 → 2p53d10 transition,39−41
indicative of divalent copper centers as can be found in CuO,
YBCO and (as a minor impurity) in KCuO2. The satellite
structures found at 934.5 and 937.5 eV in the XAS spectrum of
Cu0.1Ce0.9O2−x, indicate the presence of Cu3+, as these features
42
are similar to those of the trivalent compound KCu(biuret)2
and other formally trivalent copper compounds.40,41 These
features are significantly lower in energy to be attributed to
transitions to the final state 2p53d94s1 of CuO (∼938.7 eV)35
and are too high in energy to be attributed to Cu+ impurities
(933.7 eV for Cu2O).39
Although further studies are required, including a rigorous
treatment of charge-transfer multiplet theory, suffice to say that
copper sites in Cu0.1Ce0.9O2−x exist as an admixture of Cu2+ and
Cu3+. Thus, we have demonstrated the preparation of a rare
example of an air-stable Cu3+-containing oxide. Typically, the
ternary and quaternary Cu3+ oxides, such as KCuO2 and YBCO,
are unstable in air.43 For example, NaCuO2 and KCuO2 readily
decompose in the presence of moisture to CuO with
concurrent evolution of O2. The CeO2 lattice seems to play
an important role in stabilizing copper in the trivalent state.
Previous DFT studies have also highlighted the role of
electropositive elements (in our case, Ce4+) in stabilizing
Cu3+ in oxides.44 This hypothesis is further supported by our
own DFT + U calculations.
Figure 5. Oxygen K-edge and copper LII,III-edge XAS spectra, collected
in partial fluorescence yield mode, for annealed Cu0.1Ce0.9O2−x. (A)
Oxygen K-edge XAS spectra and (B) copper LII,III-edge XAS spectra of
Cu0.1Ce0.9O2−x (purple) and reference oxides. Dashed gray lines
represent charge-transfer multiplet satellites for Cu3+.
Our DFT + U calculations show that, under our annealing
conditions, Cu exists as a mixture of Cu3+ and Cu2+ (Figure
6A), consistent with our XAS results. At 700 K under 21% O2,
the Gibbs energy of formation (ΔGF, see Experimental
Methods in the Supporting Information) for the
Cu3+2Ce34O71 and Cu2+2Ce34O70 slab models are nearly
identical, suggesting not only that copper exists as a mixture
of Cu3+ and Cu2+, but also that the Cu3+/2+ couple is accessible
in these materials, which may have important implications for
catalysis. Understanding the physical origin for the stabilization
of Cu3+ warrants further study, but presumably it results from a
thermodynamic compromise between the copper species
favoring oxidation states lower than the isovalent case (Cu4+)
and the disruption of the CeO2 lattice by removal of charge-
compensating oxygen atoms for aliovalent (Cu3+ and Cu2+)
substitution. The optimized geometry of the Cu3+2Ce34O71
model (Figure 6C) is also consistent with our EXAFS results.
We found that copper is substituted into the CeO2 lattice as
square planar CuO4 moieties that lie flat along the (100) lattice
plane and relax along [100] toward the (100) plane of oxygen
atoms to compensate for a shorter Cu−O bond compared to
that of CeO2 (Figure 6B,C).
An important parameter to determine for oxide materials,
especially in the realm of heterogeneous catalysis, is the oxygen
nonstoichiometry (x in M0.1Ce0.9O2−x), which is related to the
oxygen-ion vacancy content of these materials. The oxygen-ion
vacancy contents for the M0.1Ce0.9O2−x compounds were
determined by TGA analysis and X-ray photoelectron spec-
troscopy (XPS) at the Ce 3d region. The results of these
measurements (Figures S22 and S23 and Table S8 of the
Supporting Information) suggest that the content of oxygen-ion
vacancies in nanoparticulate ceria is relatively unchanged with
corresponds to the quadrupole-allowed 1s → 3d transition,
necessitating an unfilled 3d manifold for Cu.37 It should be
noted here that the calculated valence of +2.9 should be taken
as an upper bound, as transition metal K-edge XAS mainly
probes the dipole-allowed 1s → 4p transition and hence does
not give direct information about the unfilled 3d states of
copper. A more suitable method would be transition metal L-
edge XAS (vide infra), which probes the 2p → 3d transition.
Fitting the EXAFS to a cluster model of CuO (Figure 4F and
Table S6), the dominant Cu−O bond distance was found to be
1.93 Å, which is intermediate to the values for KCuO2 (1.85 Å
fit here and 1.83 Å from neutron diffraction studies38) and CuO
(1.95 Å), which suggests a 20% contribution of Cu3+ to the
EXAFS of Cu0.1Ce0.9O2−x. Additionally, the magnitudes of the
next-nearest neighbor Cu−Cu peaks (indicating edge and
corner sharing square-planar CuO4 motifs) are greatly reduced
in Cu0.1Ce0.9O2−x (0.6) compared to crystalline CuO (8) as was
the case for the other transition metals. This observation
suggests the incorporation of Cu ions into the lattice of CeO2
and not the formation of a separate, amorphous CuO phase as
discussed previously.12 Our DFT + U calculations show that
phase separation of bulk Cu0.1Ce0.9O2−x into stoichiometric
CeO2 and CuO is exothermic by 2.4 electron volts (Table S7),
suggesting that our synthetic route affords the formation of a
metastable solid-solution phase. These computed energies do
not take configurational entropy into account, which
necessitates further computational work on the exergonicity
of phase separation in these materials.
The presence of Cu3+ in Cu0.1Ce0.9O2−x is further supported
by soft X-ray absorption at the Cu L-edge. Cu LII,III-edge XAS
E
dx.doi.org/10.1021/ja509214d | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Figure 6. Phase diagram and local molecular structure computed for
Cu(4−x)+2Ce34O72−x models; (A) computed ΔGF as a function of
oxygen chemical potential for three oxidation states of copper in
Cu(4−x)+2Ce34O72−x; the local coordination of (B) Ce4+ in CeO2 and of
(C) Cu3+ in Cu3+2Ce34O71 as viewed down the [111] zone axis. The
pressure dependence was calculated at 700 K, with 0.21 atm O2
indicated by the dashed gray line and the model Cu4+2Ce34O72 was
taken as the reference for computing ΔGF.
Figure 7. CO oxidation catalysis on annealed 3 nm Cu0.1Ce0.9O2−x, 3
nm CeO2 and commercial 5 μm CeO2 (Sigma-Aldrich). (A) “Light
off” curves and (B) area-normalized Arrhenius plots, measured in 1%
CO, 2.5% O2 balanced in He at a flow rate of 1300 mL min−1 gcat−1 for
20 mg catalyst loadings.
catalysts is inactive for CO oxidation catalysis, with activities
identical to bare, micrometer-sized CeO2.19,20 Our work calls in
to question the origin of the high catalytic activity in mixed
phase CuO-CeO2 catalysts (and mixed-phase ceria-based
catalysts in general), which have traditionally been attributed
to the interface of crystalline CuO particles with
CeO2,4,15,19,20,45,46 and therefore offers a new paradigm for
the design of CeO2-based catalysts.
transition metal substitution, with XPS predicting an upper
bound of as much as 10%. Any disparities in catalytic behavior
are due to the electronic structure of the transition metal
substitute and not to differences in oxygen-ion vacancy content
in these M0.1Ce0.9O2−x compounds.
CO Oxidation Catalysis on CeO2 and Cu0.1Ce0.9O2−x
.
CO oxidation catalysis was performed on annealed 3 nm
Cu0.1Ce0.9O2−x and 3 nm CeO2 nanoparticles as well as on
micrometer-sized CeO2 to examine if Cu in solid solution with
CeO2 can promote CO oxidation. The experimental “light-off”
curves and Arrhenius plots for CO oxidation in oxygen-rich
conditions (1% CO, 2.5% O2) are presented in Figure 7A,B,
respectively. Introduction of aliovalent Cu ions into the lattice
of CeO2 lowers the onset temperature for CO oxidation
(Figure 7A), and increases the catalytic activity by an order of
magnitude compared to CeO2 nanoparticles and 3 orders of
magnitude relative to micrometer-sized CeO2 (Figure 7B).
Preliminary experiments suggest that copper substitution
performs at least an order of magnitude better than the other
transition metals. Notably, Cu0.1Ce0.9O2−x did not exhibit any
irreversible sintering or phase changes during catalysis, even
after exposure to 2.5% CO at 300 °C, which suggests that this
catalyst could meet the rigorous demand of real-world
applications of CO oxidation in catalytic converters or
respirators (Figure S24 and Table S9).
While the mass-normalized rate of CO oxidation found here
for Cu0.1Ce0.9O2−x (0.1 μmol CO s−1 gcat at 35 °C) is
−1
significantly smaller than those found under similar conditions
for state-of-the-art gold-based catalysts (2.5 and 2.9 μmol CO
s−1 gcat−1 for Au/TiO247 and Au/CeO2,48 respectively), they are
comparable to optimized CuO-CeO2 catalysts (0.12 μmol CO
−1
s−1 gcat
at 45 °C).46 Additionally, the as-synthesized
Cu0.1Ce0.9O2−x nanoparticles are extractable in nonpolar
organic solvents and hence are particularly suitable for
postsynthetic modifications, such as exotemplating,49 that
would increase the surface area of these materials (and hence
their mass-normalized activities) for applications in CO
oxidation.
The enhanced activity of CeO2 nanoparticles relative to
micrometer-sized CeO2 is in agreement with previous work.48
The origin of such structure sensitivity has been investigated
previously, where it was found that CeO2 catalysts with exposed
{001} and {110} facets give significantly higher rates for CO
oxidation than catalysts with predominantly {111} termina-
tion.50 In the present case, the activity enhancement is likely
due to a higher concentration of reactive {001} facets in the
The promotion of CO oxidation kinetics with Cu in solid
solution with CeO2 comes as a surprise in light of recent studies
which have proposed that the solid solution in CuO-CeO2
F
dx.doi.org/10.1021/ja509214d | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
nanostructured CeO2 (e.g., Figure S13B,CD) compared to
micrometer-sized CeO2, where the low-energy {111} facets are
more predominant.51 This strategy has been exploited, for
example, for CO oxidation on gold catalysts where researchers
found a two-order of magnitude increase in activity when
employing nanostructured CeO2 over micrometer-sized
CeO2.48
The synthetic strategy adopted here thus decouples geo-
metric effects (dispersion, crystallite size) from intrinsic,
electronic effects in catalysis, thereby paving a way to develop
design descriptors based on the electronic structure of these
materials for heterogeneous catalysis. Apparent rates for CO
oxidation are functions of the size and shape of the catalyst
particles, the dispersion of the active sites on the surface and
the intrinsic activity of each active site. The bottom-up
synthetic approach outlined here enables the study of the
intrinsic activity of these solid solutions as a function of their
electronic structure and transition metal substitution. The
correlation of catalytic activity with the electronic structure of
these catalysts will be explored in future studies to search for
activity descriptors and potentially establish design guidelines
for the development of next-generation CO oxidation catalysts.
first instance that Cu3+ has been implicated in the mechanism
for CO oxidation on CuO-CeO2 catalysts, and, as such,
demonstrate that considerable care must be exercised when
making claims about the active site for mixed-phase
heterogeneous catalysts. The origin of this stabilization of
Cu3+ and its connection to the high catalytic activity of
Cu0.1Ce0.9O2−x warrants further studies, but we hypothesize
that the higher activity is due to the accessible Cu3+/Cu2+ redox
couple in Cu0.1Ce0.9O2−x
.
ASSOCIATED CONTENT
* Supporting Information
■
S
The X-ray crystallographic files (CIF) for CoIII-CeIII-3-MeO-
salpn, NiII-CeIII-3-MeO-salpn and CuII-CeIII-3-MeO-salpn,
experimental and computational methods, NMR, FTIR, PXRD,
HRTEM and XAS characterization. This material is available
free of charge via the Internet at http://pubs.acs.org.
AUTHOR INFORMATION
■
Corresponding Authors
E-mail: jselias@mit.edu.
E-mail: shaohorn@mit.edu.
Notes
CONCLUSIONS
■
The authors declare no competing financial interest.
In summary, we have presented a simple and generalizable
route for the preparation of monodisperse nanoparticles of
transition-metal-substituted ceria (M0.1Ce0.9O2−x, M = Mn, Fe,
Co, Ni and Cu). Unlike preparations of similar catalysts by
conventional methods, the monodispersity in size and shape, as
well as the purity of the resulting single-crystalline
M0.1Ce0.9O2−x, is well controlled by the solution-based pyrolysis
of heterobimetallic Schiff base complexes. To the best of our
knowledge, such synthetic control has not previously been
demonstrated for such a broad range of transition metal
substitutes and has implications not only for the design of novel
catalysts but also for the accurate interpretation of catalytic
trends and active sites in heterogeneous catalysis. As
synthesized, single-crystalline M0.1Ce0.9O2−x materials are
soluble in nonpolar organic solvents and, as such, can be
further processed into hierarchical assemblies,11 thin films and
devices. The bulk and local atomic structure were verified by
PXRD, HRTEM and XAS, which suggest that the aliovalent
transition metal ions substitute completely into the CeO2 lattice
with truncated octahedral crystalline habits and exposed {100}
and {111} facets.
Steady-state CO oxidation catalysis on Cu0.1Ce0.9O2−x reveals
that the aliovalent copper ions incorporated into the CeO2
lattice are responsible for an increase in catalytic activity by an
order of magnitude compared to nanocrystalline CeO2. We also
found that nanocrystalline CeO2 is 2 orders of magnitude more
active than bulk CeO2. These results justify the bottom-up
synthetic approach we employ here, where geometrical effects
in catalysis are decoupled from purely electronic effects, by
comparing surface-area-normalized catalytic rates of substituted
CeO2 of nearly identical size and crystalline habit. These results
suggest a new paradigm for the design of CeO2-based catalysts
while simultaneously allowing for systematic correlations
between the electronic structure of M0.1Ce0.9O2−x and catalytic
activity. Notably and in opposition to previous reports
suggesting Cu+ or Cu2+ to be the active sites for
catalysis,4,12,16,17,52 XAS studies on the Cu0.1Ce0.9O2−x catalyst
reveal that copper exists as an admixture of Cu2+ and Cu3+
species. These results report, to the best of our knowledge, the
ACKNOWLEDGMENTS
■
Use of the National Synchrotron Light Source, Brookhaven
National Laboratory, was supported by the U.S. Department of
Energy, Office of Science, Office of Basic Energy Sciences,
under Contract No. DE-AC02-98CH10886. DFT computa-
tions in this work were benefited from the use of the National
Energy Research Scientific Computing Center allocation of the
Center for Nanophase Materials Sciences at Oak Ridge
National Laboratory under Grant Number CNMS2013-292.
Research described in this paper was also performed at the
Canadian Light Source, which is funded by the Canada
Foundation for Innovation, the Natural Sciences and Engineer-
ing Research Council of Canada, the National Research
Council Canada, the Canadian Institutes of Health Research,
the Government of Saskatchewan, Western Economic
Diversification Canada, and the University of Saskatchewan.
This work was partially supported by Philip Morris Interna-
tional. J.S.E. was supported by the National Science Foundation
Graduate Research Fellowship under Grant No. DGE-1122374.
The X-ray diffractometer used for single-crystal studies was
purchased with the help of funding from the National Science
Foundation under Grant No. CHE-0946721. The authors
would like to thank Ms. Kelsey Stoerzinger at MIT and Ms.
Sarah Purdy at University of Saskatchewan for their assistance
with soft XAS, Mr. Thomas Carney at MIT for assistance with
ICP-AES, and thanks are due to the staff members at beamline
X11A at NSLS and beamline SGM at CLS. The authors would
like to thank Professor Yuriy Roman
helpful discussions concerning reactor design and Dr. Peter
́
and his lab at MIT for
Muller at MIT for acquiring the single-crystal diffraction data.
̈
REFERENCES
■
(1) Martin, D.; Duprez, D. J. Phys. Chem. 1996, 100, 9429.
(2) Madier, Y.; Descorme, C.; Le Govic, A. M.; Duprez, D. J. Phys.
Chem. B 1999, 103, 10999.
(3) Mogensen, M.; Sammes, N. M.; Tompsett, G. A. Solid State Ionics
2000, 129, 63.
(4) Liu, W.; Flytzani-Stephanopoulos, M. J. Catal. 1995, 153, 317.
G
dx.doi.org/10.1021/ja509214d | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
(5) Fu, Q.; Saltsburg, H.; Flytzani-Stephanopoulos, M. Science 2003,
301, 935.
(6) Chueh, W. C.; Falter, C.; Abbott, M.; Scipio, D.; Furler, P.; Haile,
S. M.; Steinfeld, A. Science 2010, 330, 1797.
(7) Chueh, W. C.; Hao, Y.; Jung, W.; Haile, S. M. Nat. Mater. 2012,
11, 155.
Karnatak, R. C. Phys. Rev. B: Condens. Matter Mater. Phys. 1989, 39,
1541.
(40) Kaindl, G.; Strebel, O.; Kolodziejczyk, A.; Schafer, W.; Kiemel,
̈
R.; Losch, S.; Kemmler-Sack, S.; Hoppe, R.; Muller, H. P.; Kissel, D.
Phys. B (Amsterdam, Neth.) 1989, 158, 446.
̈
̈
(41) Hu, Z.; Kaindl, G.; Warda, S. A.; Reinen, D.; de Groot, F. M. F.;
Muller, B. G. Chem. Phys. 1998, 232, 63.
̈
(8) Wang, Z.; Flytzani-Stephanopoulos, M. Energy Fuels 2005, 19,
2089.
(9) Liu, Y.; Hayakawa, T.; Suzuki, K.; Hamakawa, S.; Tsunoda, T.;
Ishii, T.; Kumagai, M. Appl. Catal., A 2002, 223, 137.
(10) Jasinski, P.; Suzuki, T.; Anderson, H. U. Sens. Actuators, B 2003,
95, 73.
(42) Bianconi, A.; Castellano, A. C.; De Santis, M.; Rudolf, P.;
Lagarde, P.; Flank, A. M.; Marcelli, A. Solid State Commun. 1987, 63,
1009.
(43) Levason, W.; Spicer, M. D. Coord. Chem. Rev. 1987, 76, 45.
(44) Burdett, J. K.; Sevov, S. J. Am. Chem. Soc. 1995, 117, 12788.
(45) Liu, W.; Sarofim, A. F.; Flytzani-Stephanopoulos, M. Chem. Eng.
Sci. 1994, 49, 4871.
(46) Liu, W.; Flytzani-Stephanopoulos, M. Chem. Eng. J. Biochem.
Eng. J. 1996, 64, 283.
(47) Walther, G.; Mowbray, D. J.; Jiang, T.; Jones, G.; Jensen, S.;
Quaade, U. J.; Horch, S. J. Catal. 2008, 260, 86.
(11) Corma, A.; Atienzar, P.; Garcia, H.; Chane-Ching, J.-Y. Nat.
Mater. 2004, 3, 394.
(12) Wang, X.; Rodriguez, J. A.; Hanson, J. C.; Gamarra, D.;
Martinez-Arias, A.; Fernandez-Garcia, M. J. Phys. Chem. B 2005, 109,
19595.
(13) Dow, W. P.; Huang, T. J. J. Catal. 1994, 147, 322.
(14) Wang, X.; Rodriguez, J. A.; Hanson, J. C.; Gamarra, D.;
Martinez-Arias, A.; Fernandez-Garcia, M. J. Phys. Chem. B 2006, 110,
428.
(15) Gamarra, D.; Munuera, G.; Hungria, A. B.; Fernandez-Garcia,
M.; Conesa, J. C.; Midgley, P. A.; Wang, X. Q.; Hanson, J. C.;
Rodriguez, J. A.; Martinez-Arias, A. J. Phys. Chem. C 2007, 111, 11026.
(16) Soria, J.; Conesa, J. C.; Martinez-Arias, A.; Coronado, J. M. Solid
State Ionics 1993, 63 - 65, 755.
(48) Carrettin, S.; Concepcion
Puntes, V. F. Angew. Chem., Int. Ed. 2004, 43, 2538.
́ ́
, P.; Corma, A.; Lopez Nieto, J. M.;
(49) Schuth, F. Angew. Chem., Int. Ed. 2003, 42, 3604.
(50) Zhou, K.; Wang, X.; Sun, X.; Peng, Q.; Li, Y. J. Catal. 2005, 229,
206.
(51) Wang, Z. L.; Feng, X. J. Phys. Chem. B 2003, 107, 13563.
(52) Sedmak, G.; Hocevar, S.; Levec, J. J. Catal. 2004, 222, 87.
(17) Martinez-Arias, A.; Fernandez-Garcia, M.; Soria, J.; Conesa, J. C.
J. Catal. 1999, 182, 367.
(18) Gamarra, D.; Martinez-Arias, A. J. Catal. 2009, 263, 189.
(19) Jia, A.-P.; Jiang, S.-Y.; Lu, J.-Q.; Luo, M.-F. J. Phys. Chem. C
2010, 114, 21605.
(20) Jia, A.-P.; Hu, G.-S.; Meng, L.; Xie, Y.-L.; Lu, J.-Q.; Luo, M.-F. J.
Catal. 2012, 289, 199.
(21) Elmali, A.; Elerman, Y. Z. Naturforsch., B: J. Chem. Sci. 2003, 58,
639.
(22) Vigato, P. A.; Tamburini, S. Coord. Chem. Rev. 2008, 252, 1871.
(23) Akine, S.; Nabeshima, T. Dalton Trans. 2009, 10395.
(24) Andruh, M. Chem. Commun. 2011, 47, 3025.
(25) Deshpande, S.; Patil, S.; Kuchibhatla, S. V. N. T.; Seal, S. Appl.
Phys. Lett. 2005, 87, 133113.
(26) Turner, S.; Lazar, S.; Freitag, B.; Egoavil, R.; Verbeeck, J.; Put,
S.; Strauven, Y.; Van Tendeloo, G. Nanoscale 2011, 3, 3385.
(27) Gilliss, S. R.; Bentley, J.; Carter, C. B. Appl. Surf. Sci. 2005, 241,
61.
(28) Zhang, F.; Chan, S.-W.; Spanier, J. E.; Apak, E.; Jin, Q.;
Robinson, R. D.; Herman, I. P. Appl. Phys. Lett. 2002, 80, 127.
(29) Zhang, F.; Jin, Q.; Chan, S.-W. J. Appl. Phys. 2004, 95, 4319.
(30) Lin, M.; Fu, Z. Y.; Tan, H. R.; Tan, J. P. Y.; Ng, S. C.; Teo, E.
Cryst. Growth Des. 2012, 12, 3296.
(31) Nachimuthu, P.; Shih, W.-C.; Liu, R.-S.; Jang, L.-Y.; Chen, J.-M.
J. Solid State Chem. 2000, 149, 408.
(32) Bianconi, A.; Marcelli, A.; Dexpert, H.; Karnatak, R.; Kotani, A.;
Jo, T.; Petiau, J. Phys. Rev. B: Condens. Matter Mater. Phys. 1987, 35,
806.
(33) Soldatov, A. V.; Ivanchenko, T. S.; Della Longa, S.; Kotani, A.;
Iwamoto, Y.; Bianconi, A. Phys. Rev. B: Condens. Matter Mater. Phys.
1994, 50, 5074.
(34) Rehr, J. J.; Albers, R. C. Rev. Mod. Phys. 2000, 72, 621.
(35) Flipse, C. F. J.; van der Laan, G.; Johnson, A. L.; Kadowaki, K.
Phys. Rev. B: Condens. Matter Mater. Phys. 1990, 42, 1997.
(36) Krol, A.; Lin, C. S.; Ming, Z. H.; Sher, C. J.; Kao, Y. H.; Chen, C.
T.; Sette, F.; Ma, Y.; Smith, G. C.; Zhu, Y. Z.; Shaw, D. T. Phys. Rev. B:
Condens. Matter Mater. Phys. 1990, 42, 2635.
(37) Hahn, J. E.; Scott, R. A.; Hodgson, K. O.; Doniach, S.;
Desjardins, S. R.; Solomon, E. I. Chem. Phys. Lett. 1982, 88, 595.
(38) Brese, N. E.; O’Keeffe, M.; Von Dreele, R. B.; Young, V. G., Jr. J.
Solid State Chem. 1989, 83, 1.
(39) Grioni, M.; Goedkoop, J. B.; Schoorl, R.; de Groot, F. M. F.;
Fuggle, J. C.; Schafers, F.; Koch, E. E.; Rossi, G.; Esteva, J. M.;
̈
H
dx.doi.org/10.1021/ja509214d | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX