Coverage-Dependent Behaviors of Vanadium Oxides for Chemical Looping Oxidative Dehydrogenation

Article abstract of DOI:10.1002/anie.202005968

Chemical looping provides an energy- and cost-effective route for alkane utilization. However, there is considerable CO₂ co-production caused by kinetically mismatched O^2? bulk diffusion and surface reaction in current chemical looping oxidative dehydrogenation systems, rendering a decreased olefin productivity. Sub-monolayer or monolayer vanadia nanostructures are successfully constructed to suppress CO₂ production in oxidative dehydrogenation of propane by evading the interference of O^2? bulk diffusion (monolayer versus multi-layers). The highly dispersed vanadia nanostructures on titanium dioxide support showed over 90 % propylene selectivity at 500 °C, exhibiting turnover frequency of 1.9×10^?2 s^?1, which is over 20 times greater than that of conventional crystalline V₂O₅. Combining in situ spectroscopic characterizations and DFT calculations, we reveal the loading–reaction barrier relationship through the vanadia/titanium interfacial interaction.

Full text of DOI:10.1002/anie.202005968

A Journal of the Gesellschaft Deutscher Chemiker
Angewandte
Chemie
International Edition
www.angewandte.org
Accepted Article
Title: Coverage-dependent Behaviors of Vanadium Oxides for
Chemical Looping Oxidative Dehydrogenation
Authors: Jinlong Gong
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: Angew. Chem. Int. Ed. 10.1002/anie.202005968
Link to VoR: https://doi.org/10.1002/anie.202005968

10.1002/anie.202005968
Angewandte Chemie International Edition
RESEARCH ARTICLE
Coverage-dependent Behaviors of Vanadium Oxides for Chemical
Looping Oxidative Dehydrogenation
Sai Chen1, Chunlei Pei1, Xin Chang, Zhi-Jian Zhao, Rentao Mu, Yiyi Xu, Jinlong Gong* [a]
Abstract: Chemical looping provides an energy and cost effective
route for alkane utilization. However, there is considerable CO2 co-
production caused by kinetically mismatched O2- bulk diffusion and
surface reaction in current chemical looping oxidative
dehydrogenation systems, rendering a decreased olefin productivity.
This paper describes the successful construction of sub-monolayer
or monolayer vanadia nanostructures to suppress CO2 production in
oxidative dehydrogenation of propane by evading the interference of
O2- bulk diffusion (monolayer versus multi-layers). The highly
dispersed vanadia nanostructures on titanium dioxide support
showed over 90% propylene selectivity at 500 °C, exhibiting turnover
frequency of 1.9×10-2 s-1, which is over 20 times greater than that of
conventional crystalline V2O5. Combining in situ spectroscopic
characterizations and density functional theory calculation, we
revealed the loading–reaction barrier relationship through the
vanadia/titanium interfacial interaction. This work demonstrates that
sub-monolayer or monolayer nanostructures have the potential to
serve as a general oxygen carrier platform for redox reactions.
core of CL-ODH using propane as the feedstock involves
complex redox reactions in which propane molecules adsorb
and dissociate on the surface of metal oxides, namely oxygen
[10]
carriers.
It also involves internal lattice oxygen ion (O2-)
diffusion in conventional crystalline oxygen carriers, which could
evolves into electrophilic oxygen species on the surface, leading
[8b,
to CO2 co-production
9a]. In a previous study, we have
demonstrated that the diffusion of O2- significantly influences the
surface reaction kinetics [8a]. Increasing the binding energy of V-
O bonds to inhibit the lattice oxygen diffusion is critical to
enhance the propylene selectivity. However, due to the various
V-O species and uneven oxygen diffusion rates in different
vanadia crystalline phases (e.g. V2O5, VO2, V2O3), it is difficult to
maintain high propylene selectivity during the overall
dehydrogenation reaction.
Considering the unique properties of VOx/MO systems
consisting of isolated and/or polymeric nanostructures that wet
the supporting MO, we can promote the surface reaction by
eliminating the aftermath of O2- bulk diffusion (monolayer versus
multi-layers) for vanadium oxide-based oxygen carriers [11]. It
has been evidenced that the dehydrogenation properties
significantly depend on their surface density and V-O bond types
Introduction
[8a,
12]. Wachs and co-workers mentioned that in gaseous O2
Propylene is one of the most important feedstocks in the
[1]
involved selective oxidation reactions, the V-O-support bonds
(neither V-O-V nor the terminal V=O bonds) appeared to be
chemical industry.
As the demand for propylene is growing,
there exists an imbalance of propylene supply from steam
critical active sites for the reactivity of the supported vanadia
cracking relative to demand, which is made-up by “on-purpose”
propylene production from propane dehydrogenation. Despite
[13]
[2]
species.
However, it still remains unclear of the relationship
between VOx structures and reaction properties over oxygen
carrier systems in chemical looping reactions, due to their
non-oxidative propane dehydrogenation (PDH) being the
commercial technology, the process is either limited to strong
structural complexity and diversity of oxygen species [14]
.
endothermic feature [2b, 3], and using highly expensive platinum
[4]
We report an approach to engineering metal oxide-based
oxygen carriers for chemical looping oxidative dehydrogenation
by designing and synthesizing sub-monolayer or monolayer
vanadia nanostructures on oxide supports. Titanium oxide was
chosen as the support to disperse VOx and easily identify the
different V-O species. A schematic of our materials platform in
the cycled propane dehydrogenation is outlined in Scheme 1B
and Scheme 1C. We experimentally achieve a high propylene
selectivity >90%, which is higher than that over conventional
crystalline V2O5 oxygen carriers. We also find that the turnover
frequency (TOF) decreases with the increasing of V loadings.
These findings underscore the strong coverage-dependent
effects through the vanadia/titanium interfacial interaction. This
work will have broader impacts not only on vanadium-based
oxygen carriers for oxidative dehydrogenation, but also on other
oxygen carriers for chemical looping applications such as
carbonaceous fuel conversion and utilization.
or toxic chromium catalysts
. Comparatively, oxidative
dehydrogenation (ODH) by metal oxo species could break the
[3, 5]
thermodynamic equilibrium limitation in PDH reactions
.
However, it remains an unsolved catalysis problem [4b, 6], as it
typically involves consecutive overoxidation or explosion
concerns that are not suitable for industrialization [7]
.
Alternatively, chemical looping oxidative dehydrogenation (CL-
ODH) overcomes the above-mentioned shortcomings [8]. A
typical CL-ODH process involves dehydrogenation and
regeneration reactions taking place in
dehydrogenation reactor) and a regeneration reactor,
a
reducer (or
[9]
shown
in Scheme 1A. In the reducer, the propane feedstock is oxidized
by the lattice oxygen from metal oxides. The (partially) reduced
oxide is subsequently regenerated with air in the regeneration
reactor, regaining its lattice oxygen while producing heat. The
[a]
S. Chen, X. Chang, Prof. Dr. Z. Zhao, Prof. Dr. R. Mu, Y. Xu, Prof.
Dr. J. Gong
Key Laboratory for Green Chemical Technology of Ministry of
Education, School of Chemical Engineering and Technology, Tianjin
University; Collaborative Innovation Center of Chemical Science and
Engineering; Tianjin 300072; Joint School of National University of
Singapore and Tianjin University, International Campus of Tianjin
University, Binhai New City, Fuzhou 350207, China.
E-mail: jlgong@tju.edu.cn
Results and Discussion
VOx supported on TiO2 were prepared by the wetness
impregnation method, and the loading amount of V was
controlled ranging from 0.05 to 4 wt%, named as xV/TiO2,
Supporting information for this article is given via a link at the end of
the document.
This article is protected by copyright. All rights reserved.

10.1002/anie.202005968
Angewandte Chemie International Edition
RESEARCH ARTICLE
crystalline TiO2 (rutile) (Figure 1D-F). EDS mappings show that
VOx are homogenously covering the surface of TiO2 (Figure 1G-
I). When V loading reaches to 4 wt%, small V2O5 nanoparticles
of about 3 nm are deposited on TiO2 (Figure 1C), indicating
formation of crystalline V2O5 with d-spacing of (001) (inserted
image in Figure 1F) [18]. In contrast, when the loading of V is 0.1
wt%, with the absence of continuous amorphous layers and
crystalline particles, some discrete layers appeared, suggesting
the supported VOx were in forms of isolated or deposited into
defects of TiO2 [18]. Therefore, different V coverage induce
various VOx structures, such as isolated or polyvanadate VOx
and crystalline V2O5.
Scheme 1. Chemical looping oxidative dehydrogenation of propane. A)
Schematic of reducer and regeneration reactor in the chemical looping
oxidative dehydrogenation process; B) Structure and products in sub-
monolayer and monolayer VOx vs. conventional crystalline V2O5 oxygen
carrier. C) Schematic cyclogram of dehydrogenation and regeneration steps.
where x is mass weight percent of metal V on TiO2. Considering
propane dehydrogenation at the high temperatures (>500 oC),
rutile is more thermodynamically stable. The V-catalyzed
transformation of anatase to rutile could also occur at high
temperatures. In addition, rutile predominates in single crystal
model systems and theoretical simulation due to the higher
availability of suitable substrates [15]. Such model systems have
been used to study the effects of oxidation and reduction on the
structure of VOx supported on rutile (110). Therefore, rutile was
selected as the supports in this study.
XRD patterns of different VOx/TiO2 catalysts were measured
and showed in Figure S1. One can see only the characteristic
peaks of crystalline TiO2 (rutile) (PDF#21-1276), among which
the diffraction peak of (110) exhibits the strongest intensity. No
characteristic peaks of V2O5 were observed in the samples with
V loadings less than 2 wt%, suggesting that the vanadia species
are highly dispersed on TiO2 supports.[16] Beyond this value, the
characteristic peaks attributed to orthorhombic V2O5 crystallite
(PDH#41-1426) appear (Figure S1B). The BET surface areas of
TiO2 supports are about 21 m2, therefore the V density of
1V/TiO2 was determined to be 0.89 mmol V per 100 m2 of
titanium oxide (namely, 0.081 wt% vanadium oxide per m2 of
titanium oxide), which is close to the previous reported
theoretical monolayer (a value of 0.1 wt% vanadium oxide per
m2 of titanium oxide surface area) [17]. So we suggest that the
VOx with 1 wt% V loadings completely encapsulate the TiO2
substrate and reach to the monolayer dispersion capacity. TEM
and HRTEM images showed that the particle size of TiO2
supports is about 50 nm with a d-spacing of 0.325 nm, whcih is
ascribed to TiO2(110) (Figure S1C and D). For VOx/TiO2
catalysts, on the surface of TiO2, amorphous VOx species
formed (encircled by the white dash line) (Figure 1A-C).
Specificially, the mean thickness of amorphous VOx layer was
identified as 0.8 nm over 1V/TiO2 (Figure 1B). By HRTEM
analysis, the facets of (110), (001) and (101) were identified to
Figure 1. TEM and HRTEM images of A) and D) 0.1V/TiO2, B) and E) 1V/TiO2,
C) and (F) 4V/TiO2. EDS mappings of 1V/TiO2 G: Ti; H: V; I: O. The scale bar
is 5 nm.
To discuss the interaction between the dispersed VOx and
TiO2 support, the surface structure of TiO2 supports as one of
the key factors should be considered. As suggested by the
incorporation model, the dispersed metal oxide cations might
incorporate into the defective sites available on the surface of
the support with the accompanied anion as capping anions for
[19]
charge compensation
. Considering the impurities and
contaminations, the VOx/TiO2 catalysts are calcined in air at 500
oC before the XPS characterization. The calculated ratios of
OII/OI (OI: lattice oxygen, OII: adsorbed -OH in water at oxygen
defects) ascribed from XPS of O 1s gradually decreased with
the increasing of V loadings (Figure S3A and B), suggesting the
VOx species might be incorporated into the defective sites of
TiO2 support and induced the decreasing of defective sites.
However, further increasing V loadings leads to the increasing
ratios of OII/OI due to the formation of small and defective V2O5
nanoparticles (<3 nm). The turning point with the V loadings at
about 1 wt% is consistent with the monolayer VOx (dispersion
capacity) mentioned above. Therefore, the schematic diagram
This article is protected by copyright. All rights reserved.

10.1002/anie.202005968
Angewandte Chemie International Edition
RESEARCH ARTICLE
for the dispersion of vanadia species on the TiO2 surface was
showed in Figure S3D. When the loading of V is very low, VOx
Figure 2. A) Visible (532 nm) and B) UV (325 nm) Raman spectra of VOx/TiO2 with different V loadings. C) The calculated IV-O-V/IV-O-Ti as a function of V loadings.
D) Schematic representation of various VOx species formed on the surface of TiO2.
species were mainly located on the defective sites of TiO2
surface. With the increasing of V loadings, the isolated and
polymerized VOx species were gradually formed. However, more
V loadings would contribute to the formation of disordered VOx
layers and crystalline V2O5 on TiO2 support.
the ratios of V-O-V to V-O-Ti bonds as well as the relative
amount of the polymerized to isolated vanadia species
increased accordingly. Therefore, with the increasing of V
loadings, the surface around isolated VOx species are occupied
gradually, resulting in the interaction between isolated species
and their nearest neighbors (either isolated or polymerized
vanadia species) through bridging V-O-V at the expenses of V-
O-Ti bonds. The schematic representation of various VOx
species formed on the surface of TiO2 was showed in Figure 2D.
The oxidative dehydrogenation of propane was further tested.
To eliminate the disturbance of direct propane dehydrogenation
under high temperature, the reaction performance was tested
under 500 °C. At 5th minute during the initial reaction period, as
shown in Figure 3A, the instantaneous propane conversion
increases with the increasing of V loadings. While the selectivity
of propylene is relatively low of about 80% over crystalline V2O5
(4V/TiO2). Comparatively, highly dispersed VOx/TiO2 with the
loadings from 0.05 to 1 wt% show much better catalytic
performance, where 0.1V/TiO2 possesses the highest selectivity
of propylene (~95%) (Figure 3B). In addition to C3H6 and COx,
CH4, C2H6 and C2H4 are the main other products (Figure S4).
Compared with conventional ODH[25] catalysts (Figure S5), the
VOx/TiO2 catalysts showed very high selectivity of propylene in
the chemical looping oxidative dehydrogenation at a comparable
conversion of propane. Moreover, to demonstrate the
importance of the kinds of VOx species on propylene selectivity,
we detected the reaction performance at the close degree of
propane conversion of about 10% over different VOx/TiO2
catalysts. As showed in Figure S6, with the increasing of V
loadings, the instantaneous selectivity of propylene is gradually
decreased, further indicating that the highly dispersed vanadia
supported on TiO2 are the main active species, which are
responsible for selective oxidative dehydrogenation of propane.
During the overall dehydrogenation step, about 89 % integral
selectivity of C3H6 was achieved over 1V/TiO2 (Figure S7B) at 5th
To reveal the configuration of various VOx species on TiO2,
Raman spectroscopies were further employed. Figure. 2 shows
the visible (532 nm) (Figure 2A) and UV (325 nm) (Figure 2B)
Raman spectra of the mixed VOx/TiO2 catalysts with different V
loadings. The Raman peaks at 143, 238, 445, and 609 cm-1 are
assigned to B1g (two-phonon scattering process), Eg (planar O-O
[16,
interaction), and A1g (Ti-O stretch) of rutile TiO2, respectively
20]. At the low loadings of V less than 0.25 wt%, the main feature
of the isolated VOx is the existence of V=O stretching (1024 cm-
1) and V-O-Ti bridging bonds (800 cm-1) [14e], indicating the
formation of isolated VOx [21]. With the increasing of V loadings to
1 wt%, the bonds of V-O-V (920 cm-1) appeared and gradually
dominated compared with V-O-Ti bonds. The fact of more
dominated V-O-V bonds indicated that the polymeric (bonding
also to neighboring V) species were present [22]. When the
loadings of V were higher than 1 wt%, a new band at 994 cm-1
corresponding to V=O stretching in V2O5 crystallites appeared
[18,
on 2 wt% VOx/TiO2 and 4 wt% VOx/TiO2 samples
23]. In
addition, it should be noted that as the loading of VOx was
further increased, the Eg peak of TiO2 shifted to lower
wavenumbers maybe due to the lattice distortion of TiO2 at the
high loading of VOx [24]. The relative amount of these two
bridging bonds in the catalysts is closely related to the V
loadings, where higher V loadings contribute to more dominated
V-O-V bonds than V-O-Ti bonds. Consequently, the intensity
ratio of V-O-V to V-O-Ti bonds (IV-O-V/IV-O-Ti) increased linearly
with the increasing of V loadings but at different rates before and
after its dispersion capacity (Figure 2C). The two straight lines
come across at the V loading of 0.98 wt%, which is close to the
dispersion capacity measured above (1 wt%). In other words,
This article is protected by copyright. All rights reserved.

10.1002/anie.202005968
Angewandte Chemie International Edition
RESEARCH ARTICLE
minute according to the Eq. (2d) in Supporting Information,
whcih is slightly lower than the instantaneous value (92%) due to
Figure 3. A) The instantaneous conversion and B) selectivity as a function of V loadings over VOx/TiO2 in the oxidative dehydrogenation of propane. C) The TOF
of VOx/TiO2 as a function of V loadings. Catalytic test conditions: 500 °C, 1.4 atmospheric pressure, GHSV = 2500 h-1, 250 mg of sample, C3H8 with balance N2
for total flow rate of 21 mL min-1, reaction data at 5th minute in the initial oxidative dehydrogenation period.
the formation of COx in the initial dehydrogenation period (Figure
S7A). In addition, considering the formation of coking, the
propylene selectivity (including coking formation) was evaluated
from the atomic C balance according to the Eq. (2b) (Figure S8).
Because of lattice oxygen involvement and relatively low
samples (Figure S15 and Table S2). All the results imply the
superior regenerability of this VOx/TiO2 catalyst.
In order to get deeper insights into the influence of the
dispersion state of vanadia species on the intrinsic reaction
reactivity, the turnover frequency (TOF) of VOx/TiO2 catalysts
are calculated. It is important to stress that the TOF and propane
conversion rates were tested under a special condition, which
was determined to eliminate the mass transfer and the
conversion below 15% to ensure differential reaction. TOF was
calculated based on the total number of V atom from Eq. (3) in
Support Information. As VOx could be well dispersed on TiO2 as
a sub-monolayer and monolayer, TOF based on the number of
V atom will be same as TOF based on active sites. Nevertheless,
for the catalysts with crystalline V2O5, the calculated TOF could
be slightly smaller than the actual TOF based on the active sites.
This calculation method has been widely applied in many other
works [26]. As shown in Figure 3C, the TOF values gradually
decrease with the increasing of V loadings. The highest TOF
value of 1.9×10-2 s-1 is observed at 0.1V/TiO2, namely the
o
reaction temperature (500 C), at which the coking formation is
limited and the calculated carbon balance is determined to be
about 98% (Figure S4), indicating the coke formation over
VOx/TiO2 catalysys is very weak. Moreover, no matter of O2-
TPO profiles (Figure S9) or Raman spectra (Figure S10) of the
spent VOx/TiO2 also showed little amount of coke formation
occured in the dehydrogenation step.
A redox test was then conducted over 1V/TiO2 to recover the
crystalline structures and combust the coking. Few changes of
product compositions at 5th minute in the dehydrogenation step
were observed during the 10 redox cycles (Figure S11),
indicating that a stable CL-ODH performance was achieved. In
addition, we find that after 1 and 10 redox cycles, no matter of
the morphology (Figure S12 and S13) or the crystalline structure
(Figure S14) can maintained unchanged. Moreover, the pore
structures and BET surface areas of 1V/TiO2 after 10 redox
cycles showed no significant changes compared with the fresh
isolated VOx supported on TiO2, which are at a similar level with
[26d]
respect to the VOx-based catalysts
.
Comparatively,
crystalline V2O5 (4V/TiO2) showed much lower activity for
propylene production. Therefore, it seems reasonable to
This article is protected by copyright. All rights reserved.

10.1002/anie.202005968
Angewandte Chemie International Edition
RESEARCH ARTICLE
consider that the sub-monolayer and monolayer VOx exhibit for propylene production.
higher activity than conventional crystalline V2O5 oxygen carriers
To reveal the relationships between
V
loadings and
O2-TPD was also performed to characterize relative amounts
dehydrogenation reactivity, temperature programmed reduction of surface chemisorbed and bulk lattice oxygen species. For
(TPR) of a series of VOx/TiO2 samples with different V loadings three samples (Figure 4C), there is a significant peak at about
were further performed. As shown by H2-TPR profiles (Figure 200 °C, which can be assigned to chemisorbed α-oxygen peak
4A), tiny H2 consumption has been observed for pure TiO2 under of electrophilic oxygen on the surface and nonselective for
o
700 C. Compared with TiO2 supports, the main feature of the propylene production. The peak shifting to lower temperatures
TPR profiles over VOx/TiO2 samples is the obvious shift of their with increase of V loadings explains its decreasing C3H6
reduction peaks to a lower temperature region, revealing the fact selectivity toward complete oxidation, which corresponds to XPS
that, comparatively, the supported vanadia species are easier to data that amount of electrophilic surface oxygen species
be reduced. Noteworthy, when V2O5 loading exceeds its increase when the V loading exceeds the monolayer capacity.
dispersion capacity, the H2 consumption peak shifts to Additionally, it should be noted that there is a significant peak of
apparently higher temperatures with the increase of loading. In bulk lattice oxygen at about 650 °C over 4V/TiO2. During the
other words, the vanadia species associated with the TiO2 propane dehydrogenation reactions under high temperature,
supports are more active. These results suggest that the lattice outward diffusion of bulk O2− could evolve into surface
oxygen in VOx/TiO2 catalyst might be the main active species in electrophilic oxygen species, leading to the overoxidation of pre-
ODH reactions, of which the reactivity decrease with the produced propenyl species and decreasing propylene
increasing of V loadings. In addition, the H2 consumption amount production [9a]
of each vanadium ion was shown in Figure 4B, which decreased To further evaluate intrinsic activity of C-H activation, propane
.
rapidly with the increasing of V loadings and then reached to a temperature-programmed surface reaction (C3H8-TPSR)
rather flat line. Interestingly, the inflexion point curve was also experiments were operated (Figure S16). The temperature
positioned around the dispersion capacity of VOx (1 wt%) on where C3H6 signal begins to emerge is determined as the C3H6
TiO2. The result was consistent with the argument that with the production temperature. As shown in Figure 4D, C3H6 starts to
increasing of the V loadings, the isolated VOx species in the appear over 0.1V/TiO2 at temperature down to 207 °C, lower
samples tend to interact with their nearest neighbors through than that over 4V/TiO2 at 253 °C, demonstrating higher oxygen
bridging V-O-V at the expenses of V-O-Ti bonds resulting in the reactivity for propylene production of isolated VOx compared with
decrease of the reactivity of the associated oxygen anions. With crystalline V2O5. Furthermore, the co-production of CO2 (Figure
further increase of V loadings, V2O5 crystallites formed, in which 4E) over 4V/TiO2 is significantly higher than that over 0.1V/TiO2,
the oxygen anions associated with V-O-V bonds are less which states the suppressed propylene selectivity and significant
reactive and only partially exposed on the surface.
CO2 co-production over crystalline V2O5.
This article is protected by copyright. All rights reserved.

10.1002/anie.202005968
Angewandte Chemie International Edition
RESEARCH ARTICLE
Figure 4. A) H2-TPR profiles of VOx/TiO2 catalysts with different V loadings. B) The calculated H2 consumption of H2/V from H2-TPR profiles of VOx/TiO2 catalysts
with different V loadings. C) O2-TPD profiles of VOx/TiO2 catalysts with different V loadings. D) C3H6 signals and (E) CO2 signals of C3H8-TPSR mass spectra over
0.1V/TiO2 catalyst and 4V/TiO2 catalyst. F) The TOF values and IV-O-V/IV-O-Ti obtained from Raman spectra as a function of H2 consumption obtained from H2-TPR
profiles.
It has been elucidated that sub-monolayer and monolayer
VOx/TiO2 showed higher oxygen reactivity for propylene
production than crystalline V2O5 oxygen carriers. However, the
detailed analysis over V-O bond evolution and surface reaction
V-O-V bonds than V-O-Ti bonds. Taking the fact into
consideration that increasing V loading leads to the decreasing
of TOF values and reducibility of the supported vanadia species
(as shown by TPR results above), it is suggested that for the
oxidative dehydrogenation of propane, the oxygen anions
associated with V-O-Ti bonds are more reactive compared with
those associated with V-O-V bonds. Correlating the ratios of
H2/V with TOF and IV-O-V/IV-O-Ti obtained from Raman spectra, we
found the linear relationships between them (Figure 4F), which
revealed that the vanadia coverage determined the reaction
properties over VOx/TiO2 catalysts for ODH reactions.
Specifically, with the increasing of V loadings, the decreasing
ratios of V-O-Ti to V-O-V bonds result in the decreasing
reactivity of associated oxygen anions and hence the decreasing
TOF values. When the loading amount of V is higher than its
dispersion capacity, the propylene selectivity decreases more
rapidly with the increase of V loading due to the formation of
V2O5 crystallites in which the bulk O2- could diffuse and evolve
into surface electrophilic oxygen species, leading to the
overoxidation of pre-produced propenyl species and co-
production of CO2.
To reveal the loading–reaction relationship through the
vanadia/titanium interfaces, DFT calculations were further
employed. The slab model of TiO2(110) were selected and
simulated by a three-layer (4 × 2) surface unit cel considering
the (110) plane exhibits the greatest thermal stability. For rutile
TiO2(110) and VxOy supported on TiO2(110), a 2 × 2 × 1 k-points
grid was chosen. As showed in Figure S18, the structure has
become more stable with the increase of the V coverage,
corresponding to the increase of V loadings. As has been
studied, the local bonding ability of the lattice oxygen atoms can
be measured using an H atom as the probe [12]. Therefore, the
adsorption energy of the H atom was detected to estimate the
oxygen reactivity and ability to activate the C-H bond of the
established models with various V loadings over VOx/TiO2
catalysts. At the low loading (VO3/TiO2) of isolated VOx/TiO2, the
mechanism has not been well achieved. In situ Raman and
[27]
DRIFTS, which are more sensitive to surface VOx structure
,
were further performed to monitor the evolution of V-O bonds
and reaction intermediates. As shown in Figure 5A, in situ
Raman spectra over 1V/TiO2 indicated that the V=O and V-O-Ti
species are very reactive and it could be consumed at about 200
oC. By comparison, the mode of V-O-V stretching was relatively
stabilized and gradually reduced above 200 °C. This
demonstrated that the oxygen anions associated with the
bridging V-O-Ti bonds are more active than those with V-O-V
bonds as mentioned before. In terms of surface reaction, in situ
DRIFTS showed that C3H8 adsorption occurs in forms of C–H
distortion. Then propyl complex is oxidatively dehydrogenated to
C3H6 (ν(C=C), 1645 cm-1) by subtracting H to oxygen sites,
presenting a positive hydroxyl band (V-OH, 3650 cm−1) and a
[28]
negative V=O band (2050 cm-1) over 1V/TiO2 (Figure 5B)
.
Comparatively,
A significant additional band of acetone
(νs(C=O), 1660 cm-1), an important intermediate of overoxidation
to COx, emerges and band of ν(C=C) gradually submerges over
4V/TiO2 over 350 C (Figure 5C). Considering the O2- diffusion
o
from bulk at high temperatures over crystalline V2O5, the O2-
species could evolve into surface electrophilic oxygen species,
leading to the overoxidation of pre-produced propenyl species
(emerged νs(C=O) vs. submerged ν(C=C)) and thus CO2 co-
production.
Apparently, the structure and bonding environment of the
supported vanadia species have strong impacts on the reactivity
of the associated oxygen anions and hence the catalytic
reactivity for propylene production, i.e., the oxygen anions
associated with the bridging V-O-Ti bonds are more active than
those with V-O-V bonds. The relative amount of these two
bridging bonds in the catalysts is closely related to the V
loadings, where higher V loadings contribute to more dominated
interfacial
V-O-Ti
site
shows
the
strongest.
Figure 5. A) In situ Raman spectra over 1V/TiO2 catalysts under propane dehydrogenation step ranging from 50 oC to 550 oC.. B) In situ DRIFTs over 1V/TiO2
catalysts under propane dehydrogenation step from 50 oC to 500 oC. C) In situ DRIFTs over 4V/TiO2 catalysts under propane dehydrogenation step from 50 oC to
500 oC..
This article is protected by copyright. All rights reserved.

10.1002/anie.202005968
Angewandte Chemie International Edition
RESEARCH ARTICLE
(V=O and V-O-Ti) over monomeric vanadium oxide (VO3/TiO2)
are higher than other structures indicating the good electrons
affinity which brings about the extraordinary C-H bond rupture
ability (Table S4).
Conclusion
In the present study, we investigated the structure and
catalytic performance of sub-monolayer and monolayer VOx
deposited on TiO2 support and conventional crystalline V2O5
oxygen carriers. By conducting catalytic studies, spectroscopic
characterizations and theoretical simulations, we found that the
oxygen reactivity was dependent on the dispersion states of VOx
on TiO2. With the increasing of V loadings, the isolated vanadia
species formed and then interacted with their nearest neighbors
(either isolated or polymerized vanadia) through bridging V-O-V
at the expenses of V-O-Ti bonds, resulting in decreasing
reactivity of the associated surface oxygen anions.
Consequently, although the activity increases with the increasing
of V loadings, the propylene TOF of the VOx/TiO2 catalyst
decreases linearly. Especially, the isolated VOx exhibited highest
turnover frequency of 1.9×10-2 s-1, which is over 20 times greater
than that of conventional crystalline V2O5 oxygen carriers, where
O2− diffusion from bulk and its subsequent evolution into
electrophilic oxygen species on the surface suppress the
propylene production. The established loading–reaction barrier
relationship clearly illustrates the loading effect on the catalytic
performance of oxygen carrier systems. Importantly, considering
the limited surface area of TiO2 in our model catalysts, the large
surface area supports are supposed to be preferentially needed
to obtain the sub-monolayer and monolayer nanostructures for
industrial utilization.
Figure 6. A) H adsorption energy on different V-O sites of VO3/TiO2, V2O5/TiO2,
V3O6, V4O8, 1 monolayer VOx/TiO2 and crystalline V2O5(001). B) Calculated
potential energy diagrams of VO3/TiO2, V2O5/TiO2 and crystalline V2O5(001).
adsorption of H atoms, and increasing V loading weakens its
adsorption due to the more increasing sites of V-O-V bonds than
V-O-Ti bonds (Figure 6A and Figure S19). Noticeably, the
calculated H adsorption energy with various V coverage showed
a very similar trend of experimental TOF values as a function of
VOx density on TiO2 supports in Figure 4C. Additionally, the
trend of oxygen vacancy (OV) formation energy is consistent with
the trend of H adsorption energy and VO3/TiO2 shows the most Acknowledgements
negative H adsorption energy and lowest OV formation energy,
which further indicates the isolated VOx may feature the highest
lattice oxygen activity (Table S3). It's worth mentioning that the
reaction network used in Figure 6 is the same for PDH or ODH
process when evaluating the thermodynamics and kinetics of
propane dehydrogenation to propylene. The calculated potential
energy diagrams confirm that a facile dehydrogenation process
can occur over VO3/TiO2 whose barrier of the first C-H bond
activation is much lower (~0.2 eV) and the second C-H bond
cleavage barrier is merely a little bit higher (increased by ~0.3
eV) due to the strong adsorption of C3H7* on the surface oxygen
(Figure 6B). With increase of V loadings, the activation barrier
rises due to the formation of V-O-V sites, whose
dehydrogenation barrier is comparable to crystalline V2O5(001).
In order to gain an insightful comprehension of these variable
performances, electron transfer of the model catalysts was
analyzed. At the highly dispersed VOx/TiO2, VOx transfers
electrons to TiO2 via interfacial V-O-Ti bonds with increasing of
VOx loadings (Figure S17 and S20), resulting in the promoted
oxygen reactivity. It further revealed the existence of loading–
reaction barrier relationship through the vanadium-titanium
interface. Additionally, the p-band center of surface oxygen
This work was financially supported by the National Key
Research and Development Program of China
(2016YFB0600901), the National Science Foundation of China
(Nos. 21525626, 21676181) and the Program of Introducing
Talents of Discipline to Universities (No. B06006).
Keywords: vanadium oxides • monolayer • loading-reactivity
relationship • chemical looping oxidative dehydrogenation
[1]
[2]
E. McFarland, Science 2012, 338, 340-342.
a) Y. Bian, M. Kim, T. Li, A. Asthagiri, J. F. Weaver, J Am
Chem Soc 2018, 140, 2665-2672; b) M. D. Marcinkowski,
M. T. Darby, J. Liu, J. M. Wimble, F. R. Lucci, S. Lee, A.
Michaelides, M. Flytzani-Stephanopoulos, M. Stamatakis,
E. C. H. Sykes, Nat. Chem. 2018, 10, 325-332; c) A. A.
Latimer, A. R. Kulkarni, H. Aljama, J. H. Montoya, J. S. Yoo,
C. Tsai, F. Abild-Pedersen, F. Studt, J. K. Norskov, Nat
Mater 2017, 16, 225-229.
[3]
[4]
J. J. Sattler, J. Ruiz-Martinez, E. Santillan-Jimenez, B. M.
Weckhuysen, Chem Rev 2014, 114, 10613-10653.
a) J. T. Grant, C. A. Carrero, F. Goeltl, J. Venegas, P.
Mueller, S. P. Burt, S. E. Specht, W. P. McDermott, A.
Chieregato, I. Hermans, Science 2016, 354, 1570-1573; b)
C. A. Gartner, A. C. van Veen, J. A. Lercher, J. Am. Chem.
This article is protected by copyright. All rights reserved.

10.1002/anie.202005968
Angewandte Chemie International Edition
RESEARCH ARTICLE
Soc. 2014, 136, 12691-12701.
Wang, Y. Cui, L. Zeng, R. Mu, J. Gong, ACS Catal. 2018,
8, 5526-5532.
G. Liu, Z.-J. Zhao, T. Wu, L. Zeng, J. Gong, ACS Catal.
2016, 6, 5207-5214.
Y. Chen, L. Zhang, Catal. Lett. 1992, 12, 51-62.
a) J. C. Parker, R. W. Siegel, J. Mate. Res. 2011, 5, 1246-
1252; b) T.-D. Nguyen-Phan, S. Luo, D. Vovchok, J. Llorca,
J. Graciani, J. F. Sanz, S. Sallis, W. Xu, J. Bai, L. F. J.
Piper, D. E. Polyansky, E. Fujita, S. D. Senanayake, D. J.
Stacchiola, J. A. Rodriguez, ACS Catal. 2015, 6, 407-417.
X. Bolian, F. Yining, L. Liu, L. Ming, C. Yi, Sci.in China, 45,
407-415.
T. H. Vuong, J. Radnik, J. Rabeah, U. Bentrup, M.
Schneider, H. Atia, U. Armbruster, W. Grünert, A. Brückner,
ACS Catal. 2017, 7, 1693-1705.
A. Khodakov, B. Olthof, A. T. Bell, E. Iglesia, J. Catal. 1999,
181, 205-216.
a) X. Zhang, X. Li, J. Wu, R. Yang, Z. Zhang, Catal. Lett.
2009, 132, 253-258; b) Z. Zhang, V. E. Henrich, Sur. Sci.
1992, 277, 263-272; c) D. E. Keller, S. M. K. Airaksinen, A.
O. Krause, B. M. Weckhuysen, D. C. Koningsberger, J.
Am. Chem. Soc. 2007, 129, 3189-3197; d) A. M. Rao, P. C.
Eklund, S. Bandow, A. Thess, R. E. Smalley, Nature 1997,
388, 257-259; e) B. Santara, P. K. Giri, K. Imakita, M. Fujii,
J. Phy. D 2014, 47, 215302.
[5]
[6]
a) Z. Liang, T. Li, M. Kim, A. Asthagiri, J. F. Weaver,
Science 2017, 356, 299–303; b) L. C. Buelens, V. V.
Galvita, H. Poelman, C. Detavernier, G. B. Marin, Science
2016, 354, 449-452.
a) S. H. Joo, J. Y. Park, C. K. Tsung, Y. Yamada, P. Yang,
G. A. Somorjai, Nat Mater 2009, 8, 126-131; b) W. Qi, P.
Yan, D. S. Su, Acc Chem Res 2018, 51, 640-648.
L. Zeng, Z. Cheng, J. A. Fan, L.-S. Fan, J. Gong, Nat. Rev.
Chem. 2018, 2, 349–364.
a) S. Chen, L. Zeng, R. Mu, C. Xiong, Z. J. Zhao, C. Zhao,
C. Pei, L. Peng, J. Luo, L. S. Fan, J. Gong, J Am Chem
Soc 2019, 141, 18653-18657; b) D. Li, R. Xu, X. Li, Z. Li,
X. Zhu, K. Li, Energy Fuels 2020, 34, 5381-5413.
a) Y. Gao, L. M. Neal, F. Li, ACS Catal. 2016, 6, 7293-
7302; b) Y. Liu, L. Qin, Z. Cheng, J. W. Goetze, F. Kong, J.
A. Fan, L. S. Fan, Nat Commun 2019, 10, 5503.
a) X. Zhu, Q. Imtiaz, F. Donat, C. R. Müller, F. Li, Energy
Environ. Sci. 2020, 13, 772-804; b) Y. Gao, X. Wang, J.
Liu, C. Huang, K. Zhao, Z. Zhao, X. Wang, F. Li, Sci. Adv.
2020, 6, eaaz9339.
[18]
[19]
[20]
[7]
[8]
[21]
[22]
[9]
[23]
[24]
[10]
[11]
[12]
[13]
[14]
C. A. Carrero, R. Schloegl, I. E. Wachs, R. Schomaecker,
ACS Catal. 2014, 4, 3357-3380.
C. Xiong, S. Chen, P. Yang, S. Zha, Z.-J. Zhao, J. Gong,
ACS Catal. 2019, 9, 5816-5827.
I. E. Wachs, B. M. Weckhuysen, Appl. Catal. A 1997, 157,
67-90.
[25]
[26]
a) J. T. Grant, C. A. Carrero, F. Goeltl, J. Venegas, P.
Mueller, S. P. Burt, S. E. Specht, W. P. McDermott, A.
Chieregato, I. Hermans, Science 2016, 354, 1570-1573; b)
A. Christodoulakis, J. Catal. 2004, 222, 293-306; c) O.
Ovsitser, R. Schomaecker, E. V. Kondratenko, T. Wolfram,
A. Trunschke, Catal. Today 2012, 192, 16-19.
a) G. Liu, Z.-J. Zhao, T. Wu, L. Zeng, J. Gong, ACS Catal.
2016, 6, 5207-5214; b) U. Rodemerck, M. Stoyanova, E. V.
Kondratenko, D. Linke, J. Catal. 2017, 352, 256-263; c) U.
Rodemerck, S. Sokolov, M. Stoyanova, U. Bentrup, D.
Linke, E. V. Kondratenko, J. Catal. 2016, 338, 174-183; d)
C. A. Carrero, R. Schloegl, I. E. Wachs, R. Schomaecker,
ACS Catal. 2014, 4, 3357-3380; e) J. T. Grant, C. A.
Carrero, A. M. Love, R. Verel, I. Hermans, ACS Catal.
2015, 5, 5787-5793; f) P. Hu, W.-Z. Lang, X. Yan, L.-F.
Chu, Y.-J. Guo, J. Catal. 2018, 358, 108-117.
a) S. Kattel, P. J. Ramírez, J. G. Chen, J. A. Rodriguez, P.
Liu, Science 2017, 355, 1296-1299; b) Q. Fu, H. Saltsburg,
M. Flytzani-Stephanopoulos, Science 2003, 301, 935-938;
c) M. Cargnello, V. V. Doan-Nguyen, T. R. Gordon, R. E.
Diaz, E. A. Stach, R. J. Gorte, P. Fornasiero, C. B. Murray,
Science 2013, 341, 771-773; d) A. Chen, X. Yu, Y. Zhou, S.
Miao, Y. Li, S. Kuld, J. Sehested, J. Liu, T. Aoki, S. Hong,
M. F. Camellone, S. Fabris, J. Ning, C. Jin, C. Yang, A.
Nefedov, C. Wöll, Y. Wang, W. Shen, Nat. Catal. 2019, 2,
334-341; e) M. Ek, Q. M. Ramasse, L. Arnarson, P. Georg
Moses, S. Helveg, Nat. Commun. 2017, 8, 305; f) I. Ro, J.
Resasco, P. Christopher, ACS Catal. 2018, 8, 7368-7387.
a) L. Artiglia, S. Agnoli, A. Vittadini, A. Verdini, A. Cossaro,
L. Floreano, G. Granozzi, J Am Chem Soc 2013, 135,
17331-17338; b) D. J. Stacchiola, S. D. Senanayake, P.
Liu, J. A. Rodriguez, Chem Rev 2013, 113, 4373-4390.
K. Zhuang, J. Qiu, F. Tang, B. Xu, Y. Fan, Phys Chem
Chem Phys 2011, 13, 4463-4469.
[15]
[27]
[28]
Z. Wu, H. S. Kim, P. C. Stair, S. Rugmini, S. D. Jackson, J
Phys Chem B 2005, 109, 2793-2800.
Y. He, H. Ji, J. Xu, L. Wang, J. Nat. Gas Chem. 2009, 18,
[16]
[17]
359-364.
a) G. Centi, Appl. Catal. A 1996, 147, 267-298; b) H. Li, S.
Zha, Z.-J. Zhao, H. Tian, S. Chen, Z. Gong, W. Cai, Y.
This article is protected by copyright. All rights reserved.

10.1002/anie.202005968
Angewandte Chemie International Edition
RESEARCH ARTICLE
This Article describes loading–
reaction barrier relationship over
VOx/TiO2 catalysts for chemical
looping oxidative dehydrogenation
of propane, where sub-monolayer
Sai Chen, Chunlei Pei, Xin Chang,
Zhi-Jian Zhao, Rentao Mu, Yiyi Xu
and Jinlong Gong*
Page No. – Page No.
or
monolayer
vanadia
Coverage-dependent Behaviors of
Vanadium Oxides for Chemical
nanostructures, compared with
conventional
oxygen carriers,
suppress the CO2 co-production by
evading the O2- bulk diffusion and
its subsequent evolution into
crystalline
V2O5
dramatically
Looping
Oxidative
Dehydrogenation
surface
species
electrophilic
oxygen
This article is protected by copyright. All rights reserved.