Tuning selectivity in nickel oxide-catalyzed oxidative dehydrogenation of ethane through control over non-stoichiometric oxygen density

Article abstract of DOI:10.1039/d0cy01732a

Despite bulk metal oxide non-stoichiometry often being recognized as a key determinant of catalytic performance, clarifying its catalytic function has remained elusive due in part to the highly complex nature of many catalyst surfaces. In this study, we demonstrate that the non-stoichiometric oxygen (NSO) density for thermally stable nickel oxide cubes can be manipulated without measurable changes in degree of crystallinity, particle size, and morphology. UV-vis spectroscopy, temperature programmed desorption, and H₂-temperature programmed reduction analyses all evidence monotonic decreases in NSO density upon thermal treatment; crucially, these decreases persist upon exposure to ethane and oxygen under oxidative dehydrogenation of ethane (ODHE) reaction conditions. Such control over NSO density under reaction conditions is used to tune ODHE ethene selectivity; independent of reaction conditions, at isoconversion, ethene selectivity tracks qualitatively with NSO density. These trends in selectivity, however, do not extend to conditions resulting in higher fractional hydroxyl coverages, for example, in the presence of water co-feeds, where primary ODHE selectivities vary only at the highest treatment temperature used in our study. Overall, this study points to (i) the need for well-defined oxide materials that enable (ideally) the exclusive variation of a single physicochemical property, (ii) the importance of using multiple characterization techniques for quantifying various properties, and (iii) the sensitivity of selectivity trends to surface coverages of reaction intermediates prevalent during their measurement. For the purpose of elucidating structure-catalytic property relationships, the approach reported herein involving the use of well-defined thermally stable faceted oxide crystals has the potential to be broadly applicable within the field of bulk oxide catalysis.

Full text of DOI:10.1039/d0cy01732a

Catalysis
Science &
Technology
View Article Online
View Journal
PAPER
Tuning selectivity in nickel oxide-catalyzed
oxidative dehydrogenation of ethane through
Cite this: DOI: 10.1039/d0cy01732a
control over non-stoichiometric oxygen density†
Xiaohui Zhao, Mariano D. Susman, Jeffrey D. Rimer * and Praveen Bollini
*
Despite bulk metal oxide non-stoichiometry often being recognized as a key determinant of catalytic
performance, clarifying its catalytic function has remained elusive due in part to the highly complex nature
of many catalyst surfaces. In this study, we demonstrate that the non-stoichiometric oxygen (NSO) density
for thermally stable nickel oxide cubes can be manipulated without measurable changes in degree of
crystallinity, particle size, and morphology. UV-vis spectroscopy, temperature programmed desorption, and
2
H -temperature programmed reduction analyses all evidence monotonic decreases in NSO density upon
thermal treatment; crucially, these decreases persist upon exposure to ethane and oxygen under oxidative
dehydrogenation of ethane (ODHE) reaction conditions. Such control over NSO density under reaction
conditions is used to tune ODHE ethene selectivity; independent of reaction conditions, at isoconversion,
ethene selectivity tracks qualitatively with NSO density. These trends in selectivity, however, do not extend
to conditions resulting in higher fractional hydroxyl coverages, for example, in the presence of water co-
feeds, where primary ODHE selectivities vary only at the highest treatment temperature used in our study.
Overall, this study points to (i) the need for well-defined oxide materials that enable (ideally) the exclusive
variation of a single physicochemical property, (ii) the importance of using multiple characterization
techniques for quantifying various properties, and (iii) the sensitivity of selectivity trends to surface
coverages of reaction intermediates prevalent during their measurement. For the purpose of elucidating
structure–catalytic property relationships, the approach reported herein involving the use of well-defined
thermally stable faceted oxide crystals has the potential to be broadly applicable within the field of bulk
oxide catalysis.
Received 3rd September 2020,
Accepted 10th November 2020
DOI: 10.1039/d0cy01732a
rsc.li/catalysis
the context of this study, nickel oxide non-stoichiometry has
been proposed to influence a variety of properties including
Introduction
10,11
12
Bulk metal oxides deliver promising performances in a range of
p-type semi-conductivity,
electrochromism, and catalytic
Each nickel vacancy—the predominant defect
type in Ni O—is presumed to be compensated by two electron
1
,2
3
13,14
industrial applications including catalysis,
gas sensing,
performance.
4
5–7
energy storage, and photovoltaics,
with variations in oxide
1−x
+
non-stoichiometry commonly used to tailor corresponding
material function. For example, whereas stoichiometric
titanium oxide absorbs light only in the UV range,
holes (h ) that act as charge carriers, endowing the material
with p-type semi-conductivity and a change in color from green
12,15
to black.
understoichiometric titanium oxide (TiO2 ) exhibits extended
While the precise defect structure of nickel oxide remains
a subject of debate, its degree of non-stoichiometry has
repeatedly been suggested as a determinant of catalytic
performance. Li and coworkers were the first to propose a
relationship between non-stoichiometric oxygen (NSO)
density (defined in our study as moles of non-stoichiometric
oxygen per mole total nickel) and ODHE catalytic
−x
absorption into the visible range, which is a feature that enables
8
improvements in photocatalytic efficiency. Cerium oxide is
another case in point, with reversible changes in stoichiometry
constituting the basis for its use as a buffer during lean–rich
9
cycles in automotive exhaust treatment. More specifically in
1
6
performance.
They reported that thermal treatment at
temperatures up to 700 °C led to concurrent decreases in
NSO density—estimated using temperature programmed
desorption (TPD)—and ethane conversion. The lack of ethene
selectivity data at constant reaction temperature and ethane
conversion in this study makes it challenging to interpret the
Department of Chemical and Biomolecular Engineering, University of Houston,
4
726 Calhoun Rd., Houston, TX 77204, USA. E-mail: jrimer@central.uh.edu,
ppbollini@uh.edu
Electronic supplementary information (ESI) available. See DOI: 10.1039/
†
d0cy01732a
Catal. Sci. Technol.
This journal is © The Royal Society of Chemistry 2020

View Article Online
Catalysis Science & Technology
Paper
role of non-stoichiometric oxygen. It was also found that the
non-stoichiometric oxygen content of the sample can be
regenerated upon exposure to gas phase oxygen, meaning
changes in NSO density can occur under ODHE reaction
conditions. Lemonidou and coworkers tried to elucidate,
albeit indirectly, the effect of NSO density on ODHE ethene
selectivity by varying the niobium content in mixed Nb–Ni
surface coordination environment resulting from changes in
distribution of exposed facets has been shown to be a critical
factor affecting bulk metal oxide catalytic performance.
28–30
Thermal treatments used to vary NSO density can result
(through sintering) in changes in particle sizes and exposed
2
7,31,32
crystal facets,
and multicomponent oxides can exhibit
domains that are highly heterogeneous not only from the
standpoint of chemical composition but also of surface
1
7–19
oxides.
Increasing Nb contents were found to result in
3
3,34
decreasing NSO densities, and correlated with increasing
ethene selectivities at constant conversion of ethane. The
authors proposed that non-stoichiometric oxygen, being more
electrophilic than lattice oxygen, plays a role in effecting
structure.
A second major limitation in existing reports is
the fact that although NSO densities have been determined
before exposure to ethane and oxygen, the effect of exposure
to reaction conditions involving gas phase oxygen at high
temperatures (typically greater than 400 °C) on NSO densities
oxidation to CO , causing the reduction in NSO density to
2
1
7,18
result in higher ethene selectivity.
It is important to note
is typically not evaluated. For example, O -TPD profiles in
2
that the incorporation of Nb can lead to changes in a wide
variety of physicochemical characteristics (some of which are
identified below) thereby complicating interpretations in
selectivity trends. A variety of other mixed metal oxide
experiments reaching temperatures as high as 400 °C that
can be reproduced upon cooling under oxygen suggest that
changes in NSO density achieved using thermal treatments
may potentially be partially or entirely erased upon exposure
2
0,21
22,23
24
13
combinations including Ni–Ce,
Ni–W,
Ni–Al, and
to oxygen and/or ethane at reaction temperatures.
2
5,26
Ni–Sn
have also been explored. Effects of aliovalent
Although the effect of NSO density variations on ODHE
performance has received significant attention in prior
literature, less emphasis has been placed on attempting to
exclusively vary NSO density under reaction conditions. In
this study, we synthesize well-defined nickel oxide cubes
exposing thermally stable {100} facets and demonstrate
the ability to selectively tune NSO density under reaction
conditions in order to assess its effect on ODHE ethene
selectivity. In contrast with most prior literature reports
that use control over crystal morphology directly for the
components have typically been interpreted under the
hypothesis that the excess electron density resulting from the
incorporation of high valence dopants contributes towards
neutralizing holes present in the single component oxide,
thereby decreasing the contribution of non-stoichiometric
1
4
oxygen towards catalytic performance. Nickel oxide non-
stoichiometry can also be tuned without the addition of a
second component, merely through thermal treatment at
high temperatures, as shown by Turky, who reported that
electrical conductivities and surface excess oxygen densities
purposes
performance,
of
delivering
improved
catalytic
3
0,35–39
(assessed using the hydrazine method) decrease with
we use well-defined faceted oxide
increasing calcination temperature between 300 and 700
crystals for the selective alteration of specific, catalytically-
relevant material properties that allow for the development
of more rigorous structure–property relationships. To this
end, we discuss, first, the synthesis of nickel oxide
crystals with well-defined morphologies (cubes, octahedra,
and trapezohedra), and their thermal stability. Next, we
track, using three separate characterization techniques,
2
7
°C.
We emphasize that in each of these cases physicochemical
characteristics apart from NSO density are inevitably varied,
thereby rendering inconclusive proposed correlations
between any one of these characteristics and ODHE
performance. For example, Li and coworkers do not report
the effect of thermal treatment on particle size, morphology,
or crystallinity, only the relationship between NSO density
changes in NSO density as
a function of treatment
temperature. Lastly, we report correlations between NSO
density and ODHE catalytic performance. Overall, this
study offers a template for addressing recurrent, persistent
questions as to the nature, density, and catalytic function
of defect sites in bulk oxide catalysis.
1
6
and catalytic performance.
reports significant changes in surface area, crystallite size,
and lattice parameter as function of calcination
Turky, on the other hand,
a
2
7
temperature. In other reports, incorporation of niobium
was found to result not only in an evolution from sponge-like
to plate-like morphology but also the potential formation of
Experimental methods
Materials synthesis
1
7–19
an active nickel niobate phase.
Analogous changes in
surface coordination environment, the formation of new
phases, and the agglomeration of dopants to form
amorphous or crystalline foreign oxide domains are either
conspicuously present, or at the very least challenging to
eliminate in most prior literature attempting to decipher the
role of non-stoichiometric oxygen indirectly through the
Nickel oxide cubes were prepared by a molten salt synthesis
4
0
(MSS) using a reported protocol. All materials were used as
received. In a typical synthesis, 0.1 moles Ni(NO ·6H
3
)
2
2
O
(≥97.0%, Sigma-Aldrich), 0.5 moles KNO (≥99.4%, J. T.
3
Baker), and 0.5 moles NaNO (≥99.6%, J. T. Baker) were
3
1
7–26
incorporation of aliovalent dopants.
A key characteristic
ground with a mortar and pestle for 10 min. The mixture was
transferred to an alumina crucible and calcined in a muffle
in this respect is the overall nickel oxide surface coordination
environment, particularly in view of the fact that variations in
−
1
furnace at 550 °C for 1 h with a 2.5 °C min ramp rate. After
This journal is © The Royal Society of Chemistry 2020
Catal. Sci. Technol.

View Article Online
Paper
Catalysis Science & Technology
3
−1
cooling down to room temperature, alkali salts were removed
by washing with ca. 2 L of deionized (DI) water (Aqua
Solutions RODI water purification system, >18.2 MΩ)
combined with suction-filtration, followed by drying at 60 °C
overnight. The as-synthesized oxide was divided into five
fractions, each of which was calcined in a muffle furnace
before 3 cm min H2 was added to this stream. The
temperature was then increased to 700 °C at a ramp rate of
−
1
10 °C min .
Catalytic testing
Helium, oxygen, ethane, and ethene gas cylinders (Matheson,
UHP grade) were used as purchased. Gas flows were
controlled by mass flow controllers (MKS Instruments,
GE50A) with external calibration, and the flow rates were
measured with a bubble flow meter each time before
reaction. DI water was injected into the reaction lines using a
syringe pump (KD Scientific, model 100) during reaction rate
measurements. All gas lines were heated above 100 °C with
heating tape to avoid condensation.
3
−1
(
Nabertherm, LE 6/11) under 100 cm min of N flow at
2
−
1
different temperatures for 5 h with a 2.5 °C min ramp rate.
Samples were labeled NiO400, NiO500, NiO600, NiO800, and
Ni1000 where the subscript indicates the post-synthesis
thermal treatment temperature in °C. Nickel oxide crystals
with octahedral and trapezohedral habit were synthesized
using a similar procedure as that for cubes, but by replacing
alkali nitrates with 1 mole LiNO3 in the case of octahedra,
and a mixture of KCl and NaCl (0.5 moles each) in the case
of trapezohedra.
The reaction setup consisted of two parallel gas lines—a
pretreatment line and
a reaction line—that could be
interchanged using a four-port valve (VICI-Valco, 400 psi).
The gas composition of each line was analysed using a gas
chromatograph (GC) (Agilent 7890B) equipped with a GS-
CarbonPLOT column (30 m, 0.32 mm, 1.50 μm) whose outlet
was split into a flame ionization detector (FID) and a thermal
conductivity detector (TCD) with a two-way splitter (Agilent,
G3184-60065). Ethane and ethene were quantified using the
FID signal, and O , CO , and H O were quantified using the
Materials characterization
Scanning electron microscopy (SEM) images were acquired
with a LEO 1525 FEG system operated at 3 kV. Powder X-ray
diffraction (PXRD) was performed using a SmartLab Rigaku
diffractometer with a Cu Kα source (λ = 1.5406 Å). The
diffraction patterns were analyzed using the Rigaku PDXL
software equipped with the ICDD database. Nitrogen
physisorption was performed at 77 K in a Micrometrics 3Flex
2
2
2
TCD (CO formation, if any, stayed below the detection limits
of this study).
system. Prior to N
sample were degassed under 140 mTorr vacuum at 250 °C for
2 h. The UV-vis diffuse reflectance spectrum (DRS) of samples
was measured with an Agilent Cary 5000 spectrophotometer
equipped with diffuse reflectance accessory using
Spectralon plate as reference. X-ray photoelectron
spectroscopy (XPS) measurements were conducted on a PHI
800 ESCA system (Physical Electronics) equipped with a
2
physisorption, approximately 0.5 grams of
The sample tube was heated using an insulated single-
zone furnace (Applied Test Systems, Series 3210) with a
thermocouple (Omega Engineering, KMQXL-062U-15) placed
at the top of the catalyst bed. The thermocouple was
connected to an integrated temperature controller (Watlow,
1
a
a
a
EZ-Zone PM) to control bed temperature.
A pressure
transducer (Omega Engineering, PX32B1-050AV) was used to
record the pressure above the catalyst bed.
5
standard achromatic Al Kα X-ray source (1486.6 eV) operating
at 300 W (15 kV and 20 mA) and a concentric hemispherical
analyzer. Data were collected at a 45° takeoff angle, and
measured spectra were analyzed using the MultiPak program,
with the C1s peak calibrated at 284.8 eV.
Temperature programmed desorption (TPD) and H2-
temperature programmed reduction (TPR) were carried out
in a fixed-bed reactor where the outlet gas composition was
monitored using an MKS Cirrus 2.0 quadrupole mass
spectrometer. In a typical TPD measurement, 0.25 grams of
sample were loaded in a quartz tube (1 cm internal diameter
Prior to reaction, catalyst samples were pressed, crushed,
and sieved (40–80 mesh). The pellets were then mixed with
quartz sand (40–80 mesh, acid washed and calcined at 1000
°
C) in a weight ratio ranging between 1 : 1 and 1 : 10. Each
sample was loaded into a 4 mm inner diameter quartz tube
and supported by a quartz frit. The quartz tube was mounted
into the center of the furnace; the sample temperatures
reported are those at the top of the catalyst bed. During the
pretreatment step the reactor was heated to 450 °C at a ramp
−1
rate of 3 °C min under He and then pretreated under a 26
kPa O2 in He flow at 450 °C for 30 min. The reactor was
purged under He for 1 h while the sample temperature was
ramped to the target reaction temperature. Then, the reactant
feed flow was stabilized in the reaction line and its
composition was analyzed by gas chromatography, before
being introduced to the reactor by switching the four-port
valve and reanalyzed. The carbon balance (%C), ethane
conversion (XC2H6), and ethene carbon selectivity (SC2H4) were
calculated using the following equations:
3
−1
3
with a quartz frit) and 50 cm min He (carrier) and 1 cm
−
1
min Ar (internal standard) were fed at room temperature.
Once all mass/charge signals were stable, the temperature
was increased to 1100 °C at a rate of 7 °C min . For testing
−
1
samples pre-exposed to either oxygen or reaction conditions
(ethane + oxygen), pre-TPD treatments were carried out by
exposing the catalyst to either 26 kPa O at 450 °C or 5 kPa
2
C H /5 kPa O at 400 °C for 30 min, then cooling to room
2
6
2
temperature prior to each TPD measurement. In a typical H
2
-
FC H þ FCO =2
TPR measurement, 20 mg of sample was pretreated under 28
2
4
2
%
C ¼ F
(1)
3
−1
3
−1
− F
C2H6;feed
C2H6
cm min He and 1 cm min Ar at 150 °C for 30 min
Catal. Sci. Technol.
This journal is © The Royal Society of Chemistry 2020

View Article Online
Catalysis Science & Technology
Paper
FC H þ FCO =2
2
4
2
XC H ¼
(2)
(3)
2
6
FC H ;feed
2
6
FC H4
2
SC H ¼
2
4
FC H þ FCO =2
2
4
2
where Fi represents the molar flow rate of each species
ethene, CO , and ethane) measured at the reactor outlet,
(
2
and FC2H6,feed represents the molar flow rate of ethane fed to
the reactor. The factor of 2 in eqn (3) reflects the fact that
two CO molecules are formed from one C H molecule.
2 2 6
Conversion and selectivity are reported as steady state values.
Reaction runs in the absence of catalyst conducted at the
corresponding ethane and oxygen residence times at 500 °C
exhibited negligible ethane conversions, eliminating the need
for blank tube subtractions. Negligible changes in ethane
conversion and ethene selectivity over NiO cubes were
observed in extended runs lasting 24 h.
Fig. 1 SEM images of as-synthesized (a) cubic, (d) octahedral, and (g)
trapezohedral NiO crystals, and corresponding images after thermal
treatments at increasing post-synthesis calcination temperatures of (b,
e and h) 700 °C and (c, f and i) 1000 °C for 5 h. Scale bars equal 1 μm.
Results and discussion
Synthesis and characterization of thermally stable NiO
crystals
Nickel oxide crystals with three well-defined morphologies were
40,41
synthesized using MSS techniques reported previously.
Synthesis mixtures prepared with NaNO : KNO (1 : 1 molar
ratio), LiNO , or NaCl : KCl (1 : 1 molar ratio) resulted in cubic,
3
octahedral, or trapezohedral crystals, respectively. The phase
purity of these products was confirmed using PXRD, with only
(Fig. 1d–f) and trapezohedra (Fig. 1g–i) that exhibit a
disappearance of vertices and edges, is consistent with
expectations based on the coordination environment of
pristine, defect-free, unreconstructed surfaces.
3
3
diffraction peaks corresponding to the cubic-structured NiO
phase (space group Fm ¯3 m) being observed (Fig. S1†).
We note that although qualitative trends in crystal stability
reported here can be rationalized on the basis of coordination
environments of defect-free surfaces, we know for a fact that
NSO densities evolve upon thermal treatment, implying that
the relative stabilities of defect-containing surfaces, rather
than defect-free surfaces would likely determine trends noted
in Fig. 1. Complex trends in surface energy as a function of
NSO density and temperature make it challenging to establish
the basis for the greater relative thermal stability of nickel
Cubes, octahedra, and trapezohedra expose {100}, {111},
and {311} facets, respectively (Scheme S1†), and result in
notable differences in surface coordination numbers.
Whereas the exposure of {100} surfaces results in surface
coordination numbers of 5 for both surface nickel and surface
oxygen, both {111} and {311} facets result in significantly
lower coordination numbers—three for either oxygen or
nickel (depending on surface termination) in the case of {111}
surfaces, and 2 for atoms in the topmost layer (nickel or
oxygen) for {311} surfaces. Qualitatively, one may expect {100}
nickel oxide surfaces having a lower degree of coordinative
unsaturation to be more stable compared to {111} and {311}
surfaces—an expectation borne out in calculated surface
energies for defect-free {111} surfaces that are more than two-
fold greater than those of the corresponding {100}
4
2,43
oxide cubes.
We clarify that the basis for the higher
thermal stability of nickel oxide cubes, the elucidation of
which requires a more detailed understanding of surface
energy variations as a function of both treatment temperature
and defect density, is not the focus of our study. The higher
relative stability of cubes—a phenomenological observation in
the context of our investigation—provides a material uniquely
suited to altering, exclusively, NSO density using high-
temperature thermal treatment protocols. We note that unlike
prior literature where methods for varying NSO density also
tend to impact other physicochemical characteristics (vide
supra), the preservation of cubic morphology in our case is
accompanied by imperceptible changes in degree of
crystallinity (Fig. S2†), no apparent changes in particle
morphology (Fig. S3†), and only minor changes in BET surface
area (Fig. S5 and Table S1†). Moreover, particle size
distribution analyses of samples treated at 400 and 1000 °C
suggest a mere 22% increase in particle size upon thermal
4
2,43
surfaces.
Moreover, pristine {111} and {311} surfaces
terminated by single layers of nickel or oxygen are comprised
of alternately charged planes that in theory result in a dipole
moment perpendicular to the surface; divergent surface
energies resulting therefrom are presumed to be
circumvented through either surface reconstruction or
3
6
adsorption of foreign aliovalent impurities. The preservation
of particle morphology in the case of cubes, reflected in the
sharp edges and well-defined terraces observed upon thermal
treatment up to 1000 °C (Fig. 1a–c), in contrast with octahedra
This journal is © The Royal Society of Chemistry 2020
Catal. Sci. Technol.

View Article Online
Paper
Catalysis Science & Technology
treatment (Fig. S4†). MSS-derived NiO cubes, unlike more
complex materials that undergo significant changes in surface
area, crystallite size, and lattice parameter(s) as a function of
calcination temperature, provide a platform for tuning NSO
density using thermal treatments, and assessing its role in
effectuating selective as well as unselective alkane oxidation
turnovers. We discuss next the use of characterization
techniques that provide either qualitative or quantitative
insights into NSO density of NiO cubes.
increase in background absorbance in this region has been
associated with an increase in NiO NSO content by several
authors. For example, Newman and Chrenko noted similar
changes in the UV-vis spectrum upon deliberate introduction of
oxygen achieved through exposure of NiO single crystals to air
2
7
12
at 1200 °C. Similar increases in background absorption were
also noted by Haber and Stone upon oxygen adsorption at 20
47
°C, and by Larkins and Fensham upon oxygen adsorption
48
equivalent to merely 1.85% of monolayer coverage. We note
that the origin of this background absorption, although
Tuning NiO non-stoichiometry under reaction conditions
3+
previously discussed in the context of Ni ions, excess oxygen,
12,31,33
UV-visible spectroscopy. Color has commonly been used as
a marker of NiO non-stoichiometry, with reversible transitions
between black and green bleached states being possible not
only in response to changes in NSO density by oxygen exchange
but also through the application of a voltage across the oxide
holes, and impurities,
remains as-yet unresolved.
Also, absorption spectra, although consistent with
expectations based on existing literature, provide only
qualitative insights into NSO density. To obtain quantitative
trends into NSO density, we used TPD and TPR measurements,
as described in the following sections.
44–46
surface.
Adsorption of oxygen onto greenish-yellow
stoichiometric nickel oxide results in darkening of the samples,
Temperature programmed desorption. The evolution in
NSO density as a function of treatment temperature can be
12,47–49
as noted previously by several groups.
Consistent with
these reports, thermal treatment protocols used herein result in
an evolution in coloration from black to green (Fig. 2a),
consistent with reductions in NSO density with increasing
treatment temperature. This evolution in color can be
represented more quantitatively in the form of absorption
spectra in the visible range. Irrespective of thermal treatment
temperature, the samples exhibit two broad absorption bands
centered around 417 and 700 nm (Fig. 2b) that have been
monitored using O
flow rates of O desorbing from the samples are quantified
upon subjection to a pre-set temperature profile under inert
gas flow. Measurements of O -TPD have been reported over a
2
-TPD measurements in which the molar
2
2
wide range of mixed metal oxide catalysts, and reveal
information pertaining not only to the density of oxygen
5
1–53
moieties but also their speciation.
Specifically, in the case
-TPD
of single and multicomponent nickel oxide materials, O
2
12,47,48,50
consistently observed in prior literature,
and
has been applied to quantify non-stoichiometric oxygen largely
predicated on the tacit assumption that the oxide becomes fully
correspond to d–d electron transitions of octahedrally-
2+
47
18,19
coordinated Ni ions in the crystal lattice. Higher treatment
temperatures resulted in similar absorbance corresponding to
these two bands, but lower absorbance within the 500–600 nm
range in the green-yellow region of the visible spectrum. An
stoichiometric upon exposure to high temperatures.
Such loss of non-stoichiometric oxygen can be represented
using the following equation:
Ni1−xO → (1 − xTPD) NiO + (xTPD/2) O2
(4)
where
x
TPD/(1
−
TPD
x )
represents the moles of non-
stoichiometric oxygen (excess O) per mole of nickel (or
equivalently, of the respective stoichiometric Ni), and can be
calculated from O
2
-TPD experiments using the equation:
xTPD
ðMNi þ MOÞ × NO ;desorbed
2
NSO density ðTPDÞ ¼ 1
¼
(5)
− xTPD
2m þ MONO ;desorbed
2
where NO2,desorbed represents the total moles of O
2
desorbed
in a complete TPD experiment performed on a sample of
mass m (derivation in section S3 of ESI†), and MNi and MO
represent the atomic weights of Ni and O, respectively.
Oxygen
monotonically with increasing treatment temperature
Fig. 3a), with onset temperatures being lower than the
desorption
onset
temperatures
increase
(
corresponding treatment temperatures (Table S1†) (except for
the sample treated at 400 °C). Oxygen readsorption onto
thermally treated samples may contribute to these lower
desorption onset temperatures, as indicated by much closer
correspondence between desorption onset and treatment
temperatures for the sample maintained under inert gas flow
after the 600 °C treatment instead of being exposed to air
Fig.
progressive color change after high-temperature thermal treatments
for under (treatment temperatures, in °C, indicated as
subscripts), and (b) corresponding visible absorption spectra calculated
2 (a) Photographs of nickel oxide samples exhibiting a
5
h
N
2
from UV-vis diffuse reflectance spectrum measurements using
a
Kubelka–Munk transformation.
Catal. Sci. Technol.
This journal is © The Royal Society of Chemistry 2020

View Article Online
Catalysis Science & Technology
Paper
could have resulted in the generation of specific oxygen
moieties that desorb at lower temperatures.
Bulk oxygen desorption is a possible contributor to the
broadness of the TPD peaks in Fig. 3, presumably with
surface oxygen desorption preceding bulk oxygen desorption.
The non-stoichiometric oxygen in the NiO400 sample of our
study would theoretically correspond to 4.78 monolayers of
excess oxygen (assuming surface nickel density to be 1.15 ×
1
9
2 31
1
0
surface nickel per m ), indicating that both surface
and bulk oxygen are desorbed upon TPD between 400 and
000 °C. Similar TPD profiles have been reported for other
bulk metal oxides, such as perovskites (e.g. LaMnO and
1
3
SmCoO ), where the breadth of the TPD peaks spans 600 °C,
and have been attributed to bulk oxygen desorption. NSO
3
Fig.
3
(a) O
2
-TPD profiles of as-synthesized Ni1−xO and samples
58
prepared at increasing treatment temperatures (indicated as subscripts
in °C) for 5 h at a 2.5 °C min ramp rate. Data are stacked arbitrarily on
the y-axis for visual clarity. Data for NiO700 are multiplied by 2. Scale
bar: 1 × 10 mol O
by inserting integrated O
densities calculated by integrating NO2,desorbed areas and
applying eqn (5) decrease with increasing treatment
temperatures (Fig. 3b), qualitatively consistent with UV-vis
spectroscopy data (Fig. 2b). Although a concurrent evolution
−
1
−6
2
per g Ni per min. (b) NSO densities are estimated
desorption rates over time into eqn (5).
2
3
+
in Ni density upon thermal treatment remains plausible,
nickel oxidation states remained challenging to access
through characterization methods employed in this study.
(
Fig. S6†); moreover, this sample, which was isolated from
ambient conditions post-thermal treatment and pre-TPD, also
exhibits a lower total moles of O desorbed per gram of
The as-synthesized NiO cubes (without thermal treatment)
2
2
desorbed 1.04 mg O
of 0.9 mg O
2
per m , similar to the TPD-derived value
2
sample compared to the one exposed to ambient conditions,
consistent with the readsorption of O upon exposure to
atmospheric conditions.
prevalence of complex O
that occur during the thermal treatment protocols used in our
study. For the sake of consistency, all samples discussed later
were treated for 5 h under inert gas flow and were exposed to
ambient conditions prior to TPD and ODHE experiments.
TPD temperatures have commonly been used to infer
2
per m reported by Lemonidou and coworkers for
NiO prepared by the direct decomposition of nickel nitrate.
17
2
1
6
These results point to the
adsorption–desorption phenomena
NSO densities decreased with increasing treatment
temperature (Fig. 3b), with negligible NSO density measured
for the sample treated at 1000 °C. TPD protocols used here
seem to eliminate residual non-stoichiometric oxygen after
thermal treatment, as indicated by the UV-vis spectrum of the
post-TPD sample treated at 1000 °C that appears similar to
that before TPD (Fig. S7†). Although these characterization
data show that TPD protocols render all samples subjected to
the measurement more stoichiometric in nature, they do not
provide any quantitative indications of the final degree of
(non)stoichiometry after TPD (Fig. S7b†). Existing literature
provides little evidence as to the degree of stoichiometry of
samples post-TPD, leaving open the possibility of incomplete
oxygen desorption, and therefore of underestimations in NSO
density when applying this widely used technique. In fact, the
achievement of stoichiometric nickel oxide samples post-TPD
has not been proven, or even proposed from either a
thermodynamic or kinetic standpoint. Next, we use TPR
experiments performed in the presence of gas phase molecular
hydrogen to address some of these concerns relating to a
potentially incomplete non-stoichiometric oxygen removal.
2
1
6,54,55
oxygen speciation.
For example, Chen et al. proposed
that two forms of non-stoichiometric oxygen exist on bulk
nickel oxide calcined at 500 °C—α oxygen species that desorb
between 300 and 450 °C, and β oxygen species that desorb
between 450 and 620 °C. Mobile, extra-lattice, adsorbed
oxygen was interpreted to desorb at lower temperatures (α
oxygen species), and lattice oxygen was interpreted to desorb
at higher temperatures (β oxygen species). Seiyama and
coworkers tentatively interpreted four distinct oxygen
desorption peaks (α, β, γ and δ) with increasing desorption
temperatures from 30 to 600 °C, with α and β species
1
6
corresponding to molecular oxygen (i.e. physisorbed O
2
and
−
−
O
2
) and γ and δ peaks corresponding to O adsorbed on two
5
4
different sites. A majority of the oxygen desorption in
Seiyama's work occurred below 400 °C, i.e., at temperatures
lower than those reported by other groups,
5
4
2
H -Temperature programmed reduction. TPD protocols
1
4,56,57
including
carried out in inert environments are driven by differences in
Gibbs free energies between the non-stoichiometric and the
stoichiometric surfaces/phases combined with gas phase
oxygen. These thermodynamic driving forces may not
necessarily lead to stoichiometric oxides within temperature
ranges accessible in typical TPD experiments. In contrast, TPR
experiments carried out in the presence of gas phase hydrogen
are driven by Gibbs free energy differences between gas phase
hydrogen and water, in addition to those between non-
ours. Well-resolved desorption peaks were not observed in
our experiments, in contrast with some of these literature
reports that exploit such resolution to elucidate oxygen
speciation. Differences in synthetic procedures and post-
synthesis treatment protocols used may account for some of
these differences. For instance, low temperature oxygen
adsorption followed by evacuation at 0 °C carried out prior to
the TPD experiments, as reported by Seiyama and coworkers,
This journal is © The Royal Society of Chemistry 2020
Catal. Sci. Technol.

View Article Online
Paper
Catalysis Science & Technology
stoichiometric (oxide) and stoichiometric (metal) surfaces,
thereby potentially presenting greater thermodynamic driving
forces for NSO removal at identical temperatures. This greater
propensity towards reduction under hydrogen flow is reflected
in the fact that complete reduction to metallic nickel below 600
1050 °C, pointing to the need for caution in using solely
either UV-vis spectra or temperature programmed desorption
to quantify NSO density. We note that the lack of standard
oxide samples that are (a priori) known to be stoichiometric
makes precise quantitative determinations of NSO density for
bulk metal oxides highly challenging; however, access to a
well-defined material composition post characterization—
metallic nickel in this case—may help circumvent this
limitation. In summary, differences between TPD and TPR-
derived NSO densities (Fig. 3b versus Fig. 4b) emphasize the
importance of not relying solely on TPD measurements, or
exclusively on any one ex situ characterization technique,
when quantifying bulk metal oxide NSO density.
5
9,60
°
C has either been suggested,
literature, for example using in situ time-resolved XRD.
Complete reduction under hydrogen is also reinforced by the
close agreement between H TPR-derived NSO densities and
those obtained by Bunsen–Rupp titrations that assume a 1 : 1
or evidenced directly in the
61
2
3+
62
Ni : NSO ratio. Assuming complete reduction to metallic
nickel below 600 °C, the following equation can be used to
represent the reduction process:
Ni1−xO + H
2
→ (1 − xTPR) Ni + H
2
O
(6)
Evolution in NSO density under reaction conditions
ODHE reactions are commonly interpreted within the context
of Mars–van Krevelen (MvK) redox cycles, with hydrogen
abstraction and water desorption steps constituting the
reduction half of the cycle, and surface oxygen regeneration
The moles of non-stoichiometric oxygen per mole of nickel
can be calculated using the following equation:
xTPR
NSO density ðTPRÞ ¼ 1
(7)
− xTPR
6
3,64
steps constituting the oxidation half of the catalytic cycle.
ðMNi þ MOÞ × NH ;consumed − m
2
The possible depletion and regeneration of NSO within a
single MvK cycle creates the need for assessing its density
upon exposure to ethane and/or oxygen under reaction
conditions. Skoufa et al., for example, reported a 50% drop
in TPD-derived NSO density of bulk nickel oxide upon
¼
m − MONH ;consumed
2
where NH2,consumed represents the total moles of H2
consumed during the H -TPR run (the derivation is provided
2
in section S3†).
1
9
exposure to ethane and oxygen at 400 °C. Changes in NSO
density, albeit only of a qualitative nature, were also reported
by Caps and coworkers, who noted a change in color from
black (non-stoichiometric) to green (more stoichiometric)
NiO samples show smaller differences in reduction onset
temperature compared to O desorption onset temperatures
Fig. 4a and 3a), with the latter varying by more than 100 °C
over the range of treatment temperatures used herein.
Integrated H uptake values incorporated into eqn (7) result
2
(
6
5
upon exposure to ODHE reaction conditions at 400 °C. The
plausibility of NSO loss within MvK cycles, combined with
previous studies reporting NSO desorption under reaction
conditions, necessitates a comparison of NSO densities
before and after exposure to oxygen and oxygen–ethane
mixtures at reaction temperatures. TPD profiles for samples
treated at 600 °C that change only minimally upon exposure
to oxygen (curve ii, Fig. 5a) and oxygen–ethane mixtures
2
in NSO densities that decrease from 0.25 to 0.01 with
increasing treatment temperatures. Importantly, H -TPR
2
derived NSO densities are significantly larger than
corresponding values derived from TPD profiles (Fig. 3b),
consistent with the incomplete desorption of non-
stoichiometric oxygen upon exposure to inert gas flow up to
(
curve iii, Fig. 5a), and NSO densities that are significantly
less sensitive to the presence of oxygen (ii, Fig. 5b) or
oxygen–ethane mixtures under reaction conditions (iii,
Fig. 5b) than to high-temperature thermal treatments suggest
that the protocols used in our study can be used to control
NSO densities not just ex situ, but also in situ under ODHE
conditions. Crucially, this insensitivity of NSO density to the
presence of ethane and oxygen enables an assessment of its
density on catalytic performance in integral packed bed
catalytic experiments resulting in reactant concentrations
that change significantly along the length of the bed. In the
following section, we use this control over NSO density under
reaction conditions, which is atypical for bulk oxide
1
9,65
materials,
to manipulate ODHE catalytic performance.
Fig. 4 (a) H
estimated from H
oxide samples (subscripts represent treatment temperatures in °C). In
2
-TPR profiles and (b) non-stoichiometric oxygen densities
2
-TPR measurements for thermally treated nickel
Effect of NSO density on ODHE catalytic performance
−4
Ethene selectivity at isoconversion. Here we examine the
(
a), data are stacked on the y-axis and the scale bar equals 5 × 10
mol H per g Ni per min.
2
effect of NSO density on nickel oxide ODHE performance at
Catal. Sci. Technol.
This journal is © The Royal Society of Chemistry 2020

View Article Online
Catalysis Science & Technology
Paper
mixed metal oxides – a discrepancy that may be attributable
2
2,66
to the incorporation of active secondary oxide phases
or
2
5
extra-lattice surface species, a modification of surface acid–
base properties, or a decrease or increase, respectively, in the
1
4
number of nickel–oxygen or Ni–O–M pairs.
Our data,
measured under conditions with negligible contributions
from secondary reactions (vide infra), necessitate the presence
of at least two distinct types of sites that mediate ethene and
2
CO formation, and whose relative surface densities change
with increasing treatment temperature. We note that
differences in catalytic properties presented herein are
unlikely to be a consequence of residual alkali metal content,
with surface potassium content being below XPS detection
limits, and that of sodium being less than 1.4% (Table S2†).
The low percentages of surface alkali metal content,
combined with the absence of a correlation between their
content and ODHE catalytic performance make non-
stoichiometric oxygen, rather than alkali metal content, the
more likely determinant of ODHE performance.
Fig. 5 (a) O
i) before, and after exposure to (ii) 26 kPa O
and (iii) to 5 kPa O and 5 kPa C at 400 °C for 30 min. Data are
stacked on the y-axis. (b) O -TPD derived NSO densities as a function
of treatment temperature. The error bars for NiO400 were obtained
from two individual measurements.
2
-TPD profiles of nickel oxide thermally treated at 600 °C
(
2
at 450 °C for 30 min,
2
2 6
H
2
Possibilities for the identity of these sites include active
oxygen moieties differing in electrophilicity, as has
commonly been proposed for C–H bond activation in bulk
ethane isoconversion. Our findings reveal a direct correlation
between NSO density and ethene selectivity, both of which
1
7,67,68
decrease
monotonically
with
increasing
treatment
oxide catalysis.
active oxygen moieties can be hard to elucidate,
The roles of these different types of
6
9–72
temperature (Fig. 6). Ethene selectivities must be compared
at similar ethane conversions due to possible contributions
from secondary oxidation reactions of ethene. We note that
ethene selectivities that track qualitatively with NSO density
are observed over a wide range of reaction temperatures
and does
not constitute the focus of this study. Other clues as to the
nature of these active sites on nickel oxide, however, do exist
in the literature. For instance, Heracleous and Lemonidou
found that the rates of ODHE and total oxidation were
inhibited to different extents with increasing water co-feed
pressure, possibly due to differences in water reaction orders
(400–500 °C), and these trends persist upon varying ethane
and oxygen partial pressures (Fig. 6). These favorable effects
of NSO density on ethene selectivity contrast with suggestions
made previously in the literature based on the use of complex
1
8
for the selective and unselective reactions. Density function
theory (DFT) calculations have been used to study C–H
abstraction steps over nickel–oxygen pairs that form ethyl
fragments on the nickel atoms, and surface oxygen sites that
form ethoxides, the two of which may have different relative
susceptibilities towards subsequent C–H and C–C bond
7
3,74
cleavage steps.
A combination of nickel–oxygen pair sites
and oxygen sites, the relative abundance of which evolves
with treatment temperature, represents one possible example
of the identities of active sites responsible for the selectivity
trends in Fig. 6.
The need for more accurate active site densities and
turnover frequencies. Elucidating the identity and
mechanistic role of different sites involved in nickel oxide-
2
catalyzed dehydrogenation (to ethene) and oxidation (to CO )
may be aided by the availability of kinetic data at the level of
rate constants. Direct measurement of dehydrogenation and
oxidation rates are complicated, in part, by potential
contributions from secondary oxidation reactions that render
measured ethene formation rates to be lower bounds on
ODHE rates (Scheme 1). Ethene selectivities that decrease
with increasing ethane conversion (Fig. S8†) are consistent
with the reaction network shown in Scheme 1, in which
Fig.
6 Ethene carbon selectivity (colored bars, left y-axis) and
corresponding TPR-derived NSO densities (white diamonds, right
y-axis) as a function of thermal treatment temperature measured at (a)
2.5, (b) 10, and (c) 16% ethane conversion. These conversions were
achieved by adjusting space times for each sample and reaction
condition. The three sets of data correspond to the following ODHE
3
increasing contributions from ethene oxidation (r ) relative to
conditions: (a) 400 °C, P(C
P(C ) = 10 kPa, P(O ) = 5 kPa; and (c) 500 °C, P(C
P(O ) = 10 kPa.
2
H
6
) = 15 kPa, P(O
2
) = 5 kPa; (b) 450 °C,
the oxidative dehydrogenation (r ) can be envisioned with
2
H
6
2
2
H
6
) = 10 kPa,
1
2
increasing conversion. These increasing contributions from
This journal is © The Royal Society of Chemistry 2020
Catal. Sci. Technol.

View Article Online
Paper
Catalysis Science & Technology
in interpreting structure–catalytic property relationships
using phenomenological measures like selectivity, rather
than the rate constants.
Conclusions
Well-defined nickel oxide cubes exposing {100} facets were
used to exclusively alter the non-stoichiometric oxygen (NSO)
density while maintaining the degree of crystallinity, particle
size, and morphology. These cubes, synthesized using molten
salt techniques, were found to be significantly more resistant
to thermal treatments compared to their octahedral and
trapezohedral counterparts, rendering them amenable to
high temperature treatments used to control NSO density.
Scheme
1 Proposed reaction network for ethane oxidative
dehydrogenation and total oxidation involving both primary and
secondary pathways for CO formation in the presence of O
2
2
.
secondary reactions are also reflected in ethene formation
rates that decrease with increasing residence time (Fig. S9†).
CO formation rates, on the other hand, must necessarily
2
UV-vis spectroscopy, TPD, and H -TPR all suggest decreasing
2
NSO densities with increasing treatment temperature.
Importantly, NSO density variations upon exposure to ethane
and oxygen are smaller than those achievable using thermal
treatment protocols employed in our study, meaning that
control over NSO density is enabled not only ex situ, but also
in situ under ODHE reaction conditions. We show that ethene
selectivities at isoconversion track with NSO density over a
wide range of temperatures and reactant pressures, thereby
providing a lever for controlling catalytic performance.
Relative rates of primary reactions on surfaces with greater
fractional hydroxyl coverages, on the other hand, were found
to be rather insensitive to NSO density except at the lowest
value achievable, suggesting that the selectivity control
achieved in our study (and proposed in prior literature) may
only be achievable under specific constraints, for example,
low hydroxyl coverages.
increase with residence time due to a greater relative
preponderance of secondary reactions at higher conversions
achieved at longer residence times. CO formation rates that
2
decrease, rather than increase, with residence time, on the
other hand (Fig. S9†), indicate that the changes in formation
rates, and hence selectivities, as a function of residence time
may be a consequence of product inhibition rather than
being a result of gradients in ethene partial pressure.
Negligible contributions from secondary oxidation reactions
are also reinforced by the similar oxidation rates measured
for ethane and ethene at equivalent reactant pressures (Fig.
S10†). These similar oxidation rates, combined with the low
ethene to ethane ratios prevalent at the low conversions used
in the rate measurements (i.e. less than 1% ethane
conversion) suggest the products formation are almost
exclusively from primary reactions. Consistent with these
findings, co-feeding 1–10 kPa water results in ethene and
Using the role of non-stoichiometric oxygen in ODHE as
an example, this study emphasizes three specific, disparate
aspects of bulk oxide catalysis: (1) the importance of
synthesizing well-defined oxide materials that allow for the
exclusive variation of a small number of physicochemical
characteristics, (2) the utility of exploiting multiple
2
CO formation rates that are lower than those measured in
the absence of water co-feeds. These formation rates are also
rather insensitive to residence time (Fig. S11†) resulting from
a flattening of water partial pressure gradients through the
bed (Scheme S2, ESI†). Inhibition of oxidation rates by water
1
7,75
characterization
techniques
in
quantifying
these
have been noted previously over nickel oxide,
for light alkane ODH over other catalysts, such as supported
vanadium and molybdenum oxides.
and also
characteristics, and (3) the need for rigorously measured
reaction rates as a prerequisite for clearly elucidating active
site requirements for both selective and unselective steps in
partial oxidation reaction networks. We envision that the
approach used here for elucidating structure–catalytic
property relationships via the combination of advanced
synthesis, characterization, and kinetic analysis may be
applied more broadly within the domain of bulk oxide
catalysis.
7
6–78
Water co-feeds can
therefore be used to artificially create differential beds that
reveal rates of primary ODHE and oxidation, which can then
be compared as a function of NSO density. The ratios of
1 2
primary ODHE to oxidation rates (r /r ) in the presence of
water co-feeds that are rather invariant in NSO density except
for the sample with the lowest NSO density (Fig. S12†)
suggest that monotonic trends in ethene selectivity noted in
the absence of water co-feeds (Fig. 6) may not necessarily
translate to reaction conditions resulting in surfaces with
large fractional hydroxyl coverages. The non-monotonic
trends in the ratio of primary reaction rates also point to the
need for developing characterization techniques that allow
for a more accurate characterization of surface active site
density. Moreover, the differences in trends in ethene
selectivity as a function of NSO density in the presence and
absence of water co-feeds point to the need for more caution
Conflicts of interest
There are no conflicts to declare.
Acknowledgements
PB acknowledges UH startup funds that were used to set up
the reactor used in this study. JDR acknowledges support
from The Welch Foundation (Award E-1794).
Catal. Sci. Technol.
This journal is © The Royal Society of Chemistry 2020

View Article Online
Catalysis Science & Technology
Paper
References
27 A. M. Turky, Appl. Catal., A, 2003, 247, 83–93.
2
8 M. Capdevila-Cortada, M. García-Melchor and N. López,
1
2
3
J. C. Védrine, ChemSusChem, 2019, 12, 577–588.
W. Huang, Acc. Chem. Res., 2016, 49, 520–527.
N. Barsan, D. Koziej and U. Weimar, Sens. Actuators, B,
J. Catal., 2015, 327, 58–64.
29 Z. A. Qiao, Z. Wu and S. Dai, ChemSusChem, 2013, 6,
1821–1833.
2
007, 121, 18–35.
S. Fang, D. Bresser and S. Passerini, Adv. Energy Mater.,
020, 10, 1902485.
30 Q. Hua, T. Cao, X.-K. Gu, J. Lu, Z. Jiang, X. Pan, L. Luo, W.-X.
Li and W. Huang, Angew. Chem., 2014, 126, 4956–4961.
31 E. E. Platero, G. S. Coluccia and A. Zecchina, Langmuir,
1987, 3, 291–297.
32 G. A. El-Shobaky, I. F. Hewaidy and T. H. El-Nabarawi, Surf.
Technol., 1981, 12, 309–315.
33 X. Xu, L. Li, J. Huang, H. Jin, X. Fang, W. Liu, N. Zhang, H.
Wang and X. Wang, ACS Catal., 2018, 8, 8033–8045.
34 H. Zhu, H. Dong, P. Laveille, Y. Saih, V. Caps and J. M.
Basset, Catal. Today, 2014, 228, 58–64.
4
2
5
6
C. T. Sah, IEEE Trans. Electron Devices, 1964, 11, 324–345.
M.-J. Lee, S. I. Kim, C. B. Lee, H. Yin, S.-E. Ahn, B. S. Kang,
K. H. Kim, J. C. Park, C. J. Kim, I. Song, S. W. Kim, G.
Stefanovich, J. H. Lee, S. J. Chung, Y. H. Kim and Y. Park,
Adv. Funct. Mater., 2009, 19, 1587–1593.
7
A. Pérez-Tomás, A. Mingorance, D. Tanenbaum and M. Lira-
Cantú, Futur. Semicond. Oxides Next-Generation Sol. Cells,
2
018, pp. 267–356.
35 Q. Ren, Z. Feng, S. Mo, C. Huang, S. Li, W. Zhang, L. Chen,
M. Fu, J. Wu and D. Ye, Catal. Today, 2018, 332, 160–167.
36 H. X. Mai, L. D. Sun, Y. W. Zhang, R. Si, W. Feng, H. P.
Zhang, H. C. Liu and C. H. Yan, J. Phys. Chem. B, 2005, 109,
24380–24385.
37 E. Aneggi, D. Wiater, C. De Leitenburg, J. Llorca and A.
Trovarelli, ACS Catal., 2014, 4, 172–181.
38 Y. Kwon, A. Soon, H. Han and H. Lee, J. Mater. Chem. A,
2015, 3, 156–162.
8
9
N. Serpone, J. Phys. Chem. B, 2006, 110, 24287–24293.
J. Kašpar, P. Fornasiero and M. Graziani, Catal. Today,
1
999, 50, 285–298.
0 S. Mrowec and Z. Grzesik, J. Phys. Chem. Solids, 2004, 65,
651–1657.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
1
1 A. Bielański, J. Dereń, J. Haber and J. Słoczyński, Trans.
Faraday Soc., 1962, 58, 166–175.
2 R. Newman and R. M. Chrenko, Phys. Rev., 1959, 114,
1
507–1513.
3 T. Chen, W. Li, C. Yu, R. Jin and H. Xu, Stud. Surf. Sci. Catal.,
000, 130 B, 1847–1852.
4 E. Heracleous and A. A. Lemonidou, J. Catal., 2010, 270,
7–75.
5 R. Farhi and G. Petot-Ervas, J. Phys. Chem. Solids, 1978, 39,
169–1173.
6 T. Chen, W. Li, C. Yu, R. Jin and H. Xu, Stud. Surf. Sci. Catal.,
000, 130, 1847–1852.
7 E. Heracleous and A. A. Lemonidou, J. Catal., 2006, 237,
62–174.
8 E. Heracleous and A. A. Lemonidou, J. Catal., 2006, 237,
75–189.
9 Z. Skoufa, E. Heracleous and A. A. Lemonidou, J. Catal.,
015, 322, 118–129.
39 Y. Xu, H. Wang, Y. Yu, L. Tian, W. Zhao and B. Zhang,
J. Phys. Chem. C, 2011, 115, 15288–15296.
40 M. D. Susman, H. N. Pham, X. Zhao, D. H. West, S. Chinta,
P. Bollini, A. K. Datye and J. D. Rimer, Angew. Chem., Int. Ed.,
2020, 59, 15119.
41 M. D. Susman, H. N. Pham, A. K. Datye, S. Chinta and J. D.
Rimer, Chem. Mater., 2018, 30, 2641–2650.
42 P. M. Oliver, G. W. Watson and S. C. Parker, Phys. Rev. B:
Condens. Matter Mater. Phys., 1995, 52, 5323–5329.
43 P. M. Oliver, S. C. Parker and W. C. Mackrodt, Modell. Simul.
Mater. Sci. Eng., 1993, 1, 755–760.
44 F. Lin, D. Nordlund, T. C. Weng, D. Sokaras, K. M. Jones,
R. B. Reed, D. T. Gillaspie, D. G. J. Weir, R. G. Moore, A. C.
Dillon, R. M. Richards and C. Engtrakul, ACS Appl. Mater.
Interfaces, 2013, 5, 3643–3649.
45 C. M. Lampert, T. R. Omstead and P. C. Yu, Sol. Energy
Mater., 1986, 14, 161–174.
46 G. A. Niklasson and C. G. Granqvist, J. Mater. Chem.,
2007, 17, 127–156.
47 J. Haber and F. S. Stone, Trans. Faraday Soc., 1963, 59, 192.
48 F. P. Larkins and P. J. Fensham, Trans. Faraday Soc.,
1970, 66, 1755–1772.
2
6
1
2
1
1
2
0 J. L. Park, S. K. Balijepalli, M. D. Argyle and K. J. Stowers,
Ind. Eng. Chem. Res., 2018, 57, 5234–5240.
1 B. Solsona, P. Concepción, S. Hernández, B. Demicol and
J. M. L. Nieto, Catal. Today, 2012, 180, 51–58.
2 B. Solsona, J. M. López Nieto, P. Concepción, A. Dejoz, F.
Ivars and M. I. Vázquez, J. Catal., 2011, 280, 28–39.
3 H. Zhu, D. C. Rosenfeld, M. Harb, D. H. Anjum, M. N.
Hedhili, S. Ould-Chikh and J. M. Basset, ACS Catal., 2016, 6,
49 S. J. Teichner and J. A. Morrison, Trans. Faraday Soc.,
1955, 51, 961–966.
2
852–2866.
2
2
4 Z. Skoufa, G. Xantri, E. Heracleous and A. A. Lemonidou,
Appl. Catal., A, 2014, 471, 107–117.
50 K. Klier, Catal. Rev., 1968, 1, 207–232.
51 C. Dong, Z. Qu, Y. Qin, Q. Fu, H. Sun and X. Duan, ACS
Catal., 2019, 9, 6698–6710.
52 S. Anke, G. Bendt, I. Sinev, H. Hajiyani, H. Antoni, I.
Zegkinoglou, H. Jeon, R. Pentcheva, B. R. Cuenya, S. Schulz
and M. Muhler, ACS Catal., 2019, 9, 5974–5985.
53 Y. Zheng, S. Thampy, N. Ashburn, S. Dillon, L. Wang, Y.
Jangjou, K. Tan, F. Kong, Y. Nie, M. J. Kim, W. S. Epling, Y. J.
5 D. Delgado, B. Solsona, A. Ykrelef, A. Rodríguez-Gómez, A.
Caballero, E. Rodríguez-Aguado, E. Rodríguez-Castellón and
J. M. López Nieto, J. Phys. Chem. C, 2017, 121, 25132–25142.
6 B. Solsona, P. Concepción, B. Demicol, S. Hernández, J. J.
Delgado, J. J. Calvino and J. M. López Nieto, J. Catal.,
2
2
012, 295, 104–114.
This journal is © The Royal Society of Chemistry 2020
Catal. Sci. Technol.

View Article Online
Paper
Catalysis Science & Technology
Chabal, J. W. P. Hsu and K. Cho, J. Am. Chem. Soc.,
019, 141, 10722–10728.
66 B. Savova, S. Loridant, D. Filkova and J. M. M. Millet, Appl.
Catal., A, 2010, 390, 148–157.
2
5
5
5
5
5
4 M. Iwamoto, Y. Yoda, M. Egashira and T. Seiyama, J. Phys.
Chem., 1976, 80, 1989–1994.
5 M. Iwamoto, Y. Yoda, N. Yamazoe and T. Seiyama, J. Phys.
Chem., 1978, 82, 2564–2570.
67 V. Fleischer, R. Steuer, S. Parishan and R. Schomäcker,
J. Catal., 2016, 341, 91–103.
68 G. I. Panov, K. A. Dubkov and E. V. Starokon, Catal. Today,
2006, 117, 148–155.
69 K. Aika and J. H. Lunsford, J. Phys. Chem., 1978, 82,
1794–1800.
70 E. Morales and J. H. Lunsford, J. Catal., 1989, 118, 255–265.
71 K. Aika and J. H. Lunsford, J. Phys. Chem., 1977, 81,
1393–1398.
6 Y. Wu, Y. He, T. Wu, W. Weng and H. Wan, Mater. Lett.,
2
007, 61, 2679–2682.
7 Z. Skoufa, E. Heracleous and A. A. Lemonidou, Catal. Today,
012, 192, 169–176.
2
8 N. Ashburn, Y. Zheng, S. Thampy, S. Dillon, Y. J. Chabal,
J. W. P. Hsu and K. Cho, ACS Appl. Mater. Interfaces,
72 M. B. Ward, M. J. Lin and J. H. Lunsford, J. Catal., 1977, 50,
306–318.
73 X. Lin, Y. Xi and J. Sun, J. Phys. Chem. C, 2012, 116,
3503–3516.
2
019, 11, 30460–30469.
9 J. L. Brito, J. Laine and K. C. Pratt, J. Mater. Sci., 1989, 24,
25–431.
5
4
6
6
0 V. C. F. Holm and A. Clark, J. Catal., 1968, 11, 305–316.
1 J. A. Rodriguez, J. C. Hanson, A. I. Frenkel, J. Y. Kim and M.
Pérez, J. Am. Chem. Soc., 2002, 124, 346–354.
74 J. J. Varghese and S. H. Mushrif, J. Phys. Chem. C, 2017, 121,
17969–17981.
75 Y.-F. Y. Yao and J. T. Kummer, J. Catal., 1973, 28, 124–138.
76 S. T. Oyama, A. M. Middlebrook and G. A. Somorjai, J. Phys.
Chem., 1990, 94, 5029–5033.
77 M. D. Argyle, K. Chen, A. T. Bell and E. Iglesia, J. Phys. Chem.
B, 2002, 106, 5421–5427.
78 K. Chen, A. T. Bell and E. Iglesia, J. Phys. Chem. B, 2000, 104,
1292–1299.
6
6
2 N. K. Kotsev and L. I. Ilieva, Catal. Lett., 1993, 18, 173–176.
3 K. Chen, A. Khodakov, J. Yang, A. T. Bell and E. Iglesia,
J. Catal., 1999, 186333, 325–333.
6
6
4 X. Li and E. Iglesia, J. Phys. Chem. C, 2008, 112, 15001–15008.
5 H. Zhu, S. Ould-Chikh, D. H. Anjum, M. Sun, G. Biausque,
J. M. Basset and V. Caps, J. Catal., 2012, 285, 292–303.
Catal. Sci. Technol.
This journal is © The Royal Society of Chemistry 2020