XAFS study on the specific deoxidation behavior of iron titanate catalyst for the selective catalytic reduction of NOx with NH3

Article abstract of DOI:10.1002/cctc.201300565

The environmentally friendly catalyst iron titanate (FeTiO_x) was reported to be very active for the selective catalytic reduction of NO _x with NH₃ (NH₃-SCR), with high N₂ selectivity and H₂O/SO₂ durability in the medium temperature range, and the specific microstructure of iron titanate crystallites as the active phase was determined. In consideration of the probable existence of a redox cycle between Fe³⁺ and Fe²⁺ species in the NH₃-SCR reaction, the deoxidation behavior of the FeTiO_x catalyst in an H₂ temperature-programmed reduction process was studied extensively by X-ray absorption near-edge structure (XANES) and extended X-ray absorption fine-structure (EXAFS) methods. Owing to the presence of an electronic inductive effect between Fe and Ti species in the unique edge-shared Fe³⁺-(O)₂-Ti⁴⁺ structure, the reducibility of Fe³⁺ species in the FeTiO_x catalyst was greatly enhanced compared with that in pristine Fe₂O₃, leading to the higher oxidation ability of Fe species in FeTiO_x. In the H ₂ temperature-programmed reduction process, the well-dispersed Fe³⁺ species in iron titanate crystallites could be totally converted into Fe²⁺ in the form of ilmenite FeTiO₃ below 500 °C, whereas pristine Fe₂O₃ could only be reduced to Fe₃O₄ up to this temperature point. The typical NH ₃-SCR reaction is usually conducted below 500°C, and the enhanced oxidation ability of Fe³⁺ species in FeTiO_x catalyst is responsible for its excellent catalytic NO_x reduction performance at low temperatures. Based on XANES linear fitting and EXAFS curve-fitting results, the specific deoxidation process of the FeTiO_x catalyst was proposed, which can provide useful information for the characterization of the microstructure and redox ability of active sites simultaneously in mixed oxide catalysts for certain catalytic reactions. Living on the iron edge: The deoxidation behavior of an environmentally friendly iron titanate catalyst for the selective catalytic reduction of NO_x with NH₃ is systematically studied by H₂ temperature-programmed reduction and X-ray absorption fine-structure methods. The significantly enhanced reducibility and redox ability of Fe species in the unique edge-shared Fe ³⁺-(O)₂-Ti⁴⁺ structure improves the NO _x reduction efficiency at low temperatures. Copyright

Full text of DOI:10.1002/cctc.201300565

CHEMCATCHEM
FULL PAPERS
DOI: 10.1002/cctc.201300565
XAFS Study on the Specific Deoxidation Behavior of Iron
Titanate Catalyst for the Selective Catalytic Reduction of
NO with NH
x
3
[a]
Fudong Liu, Hong He,* and Lijuan Xie
The environmentally friendly catalyst iron titanate (FeTiO ) was
pristine Fe O , leading to the higher oxidation ability of Fe spe-
2 3
x
reported to be very active for the selective catalytic reduction
of NO with NH (NH -SCR), with high N selectivity and H O/
cies in FeTiO . In the H temperature-programmed reduction
x 2
3
+
process, the well-dispersed Fe species in iron titanate crystal-
x
3
3
2
2
2
+
SO durability in the medium temperature range, and the spe-
lites could be totally converted into Fe in the form of ilmen-
2
cific microstructure of iron titanate crystallites as the active
phase was determined. In consideration of the probable exis-
ite FeTiO below 5008C, whereas pristine Fe O could only be
3
2
3
reduced to Fe O up to this temperature point. The typical
3
4
3
+
2+
tence of a redox cycle between Fe and Fe species in the
NH -SCR reaction, the deoxidation behavior of the FeTiO cata-
NH -SCR reaction is usually conducted below 5008C, and the
3
3
+
enhanced oxidation ability of Fe species in FeTiO catalyst is
3
x
x
lyst in an H temperature-programmed reduction process was
responsible for its excellent catalytic NOx reduction perfor-
mance at low temperatures. Based on XANES linear fitting and
EXAFS curve-fitting results, the specific deoxidation process of
2
studied extensively by X-ray absorption near-edge structure
(XANES) and extended X-ray absorption fine-structure (EXAFS)
methods. Owing to the presence of an electronic inductive
effect between Fe and Ti species in the unique edge-shared
the FeTiO catalyst was proposed, which can provide useful in-
x
formation for the characterization of the microstructure and
redox ability of active sites simultaneously in mixed oxide cata-
lysts for certain catalytic reactions.
3
+
4+
3+
Fe ꢀ(O) ꢀTi structure, the reducibility of Fe species in the
2
FeTiO catalyst was greatly enhanced compared with that in
x
Introduction
The greatly increased demand for the use of fossil fuels includ-
ing coal and oil nowadays has led to the significant growth in
emission of air pollutants including particulate matter, sulfur
disadvantages still exist such as a narrow operation tempera-
ture window, low N selectivity and poor stability at high tem-
2
peratures, and, more importantly, the biological toxicity of the
active V O phase, restricting its further application especially
dioxide, and nitrogen oxides (NO ), especially in the developing
x
2
5
countries with rapid increase in gross domestic product.
Owing to the widely used dust-removal equipment and desul-
for the NOx reduction process for mobile sources such as
[3,7,8]
diesel engines.
Therefore, more and more researchers now
furizing units, the particulate matter and SO can be effectively
devote themselves to the development of new, highly efficient
2
controlled to a certain extent, whereas the ultrafine particles
and environmentally friendly NH -SCR catalysts with improved
3
and NO become the key targets for air-pollutant control in the
low-temperature deNO efficiency, hydrothermal stability or re-
x
x
next steps. NO , including NO and NO emitted from stationary
sistance to co-existing pollutants, including zeolite catalysts
x
2
and mobile sources, are major air pollutants resulting in acid
promoted by metal cations (e.g., Fe-ZSM-5, Fe-BEA, Cu-ZSM-5,
[
1,2]
[9–20]
rain, photochemical smog, respiratory disease,
and possibly
Cu-beta, Cu-CHA)
and single metal oxide, supported
even severe haze pollution, and need to be controlled effec-
tively in accordance with more and more stringent regulations
type, or mixed oxide catalysts (e.g., g-Fe O nanorods, Fe O /
2 3 2 3
WO /ZrO , Ti Fe O
,
Ce-P-O, CeTiOx, CeWOx, MnOx,
3
2
0.9
[
0.1 2ꢀd
21–29]
worldwide. Selective catalytic reduction of NO with NH (NH -
(Fe Mn ) O4).
3ꢀx x 1ꢀd
x
3
3
SCR) under oxygen-rich conditions is one of the most efficient
In our previous study, we reported a novel, environmentally
technologies for the catalytic removal of NO , and the earliest
benign iron titanate (FeTiO ) catalyst for NH -SCR of NO pre-
x
x
3
x
commercial catalyst system was WO - or MoO -promoted V O /
pared by a facile coprecipitation method, achieving high
deNO efficiency, N selectivity, and H O/SO durability in the
3
3
2
5
TiO , with high NH -SCR performance and SO durability in the
2
3
2
x
2
2
2
[
3–6]
[30–32]
medium or high temperature ranges.
Although this catalyst
medium temperature range.
The NH -SCR mechanisms in
3
system has been utilized for several decades in industry, some
the low- and high-temperature ranges were proposed accord-
ing to in situ diffuse reflectance infrared Fourier transform
spectroscopy and mass spectrometry to detect the surface-ad-
sorbed species and gaseous components, respectively, in the
[a] Dr. F. Liu, Prof. H. He, L. Xie
Research Center for Eco-Environmental Sciences
Chinese Academy of Sciences
[33]
3+
SCR reaction, most likely involving the redox cycle of Fe
1
8 Shuangqing Road, Beijing 100085 (P.R. China)
2
+
$
Fe for the activation of reactants including NH and NO.
3
Fax: (+86)10-62849123
E-mail: honghe@rcees.ac.cn
Owing to the strong interaction between Fe and Ti species,
ꢀ
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2013, 5, 3760 – 3769 3760

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
the active phase in this FeTiO catalyst was mainly in the form
x
of crystallites, the structure of which was rather difficult to
characterize by conventional XRD, Raman spectra, or TEM,
which usually require long-range order in the target materials.
Therefore, the X-ray absorption fine structure (XAFS) method
was used for the systematic structural characterization of this
[
34]
FeTiO catalyst, which is very sensitive to the electronic state
x
and local structure of specific central atoms. For the vanadium-
free catalyst systems in the reported works mentioned above,
the catalytic deNO performance has usually been investigated
x
in detail, yet the comprehensive investigation of their struc-
ture–activity relationships in the NH -SCR reaction is still largely
3
lacking. In particular, many researchers have only paid atten-
tion to the microstructure of the active phase in NH -SCR cata-
Figure 1. NO
tion over FeTiO
spectively. Reaction conditions: [NO]=[NH
x
conversions as a function of temperature in the NH
catalyst and references hematite Fe and anatase TiO
]=500 ppm, [O ]=5 vol%,
3
-SCR reac-
3
lysts and ignored the redox ability of the active species, which
is actually not beneficial for the understanding of reaction
mechanisms. Herein, the specific deoxidation behavior of Fe
x
2
O
3
2
, re-
3
2
ꢀ
1
0
.6 mL catalyst, and total flow rate of 500 mLmin yielding a gas hourly
ꢀ1
space velocity of 50000 h
.
species in the FeTiO catalyst was sequentially investigated by
x
using the XAFS method in combination with H temperature
2
programmed reduction (H -TPR) experiments, which can pro-
As shown by the high-resolution TEM (HRTEM) images in
2
[32]
2
ꢀ1
vide useful information about the redox ability of Fe species
having a special microstructure, to better understand the in-
trinsic reason for its high catalytic deNO efficiency in the NH -
Figure 2A and B, both the anatase TiO (103.5 m g ) and
2
2
ꢀ1
hematite Fe O (42.5 m g ) were well crystallized with rela-
2
3
tively large particle sizes, lacking rich surface defects for the
x
3
SCR reaction.
adsorption and activation of reactants in the NH -SCR reaction.
3
3
+
Possibly owing to the very similar ion radii of Fe (0.645 ꢁ)
4
+
[35]
and Ti (0.605 ꢁ), in the FeTiO catalyst the Fe species and
x
Results and Discussion
Ti species could interdiffuse in a homogeneous state resulting
in the formation of iron titanate crystallites with small particle
NH -SCR performance of the FeTiO catalyst and its micro-
3
x
2
ꢀ1
size and relatively large surface area (245.3 m g ), which
could supply more surface defects and active sites for the oc-
structure
The NH -SCR performance of the FeTiOx catalyst compared
currence of the NH -SCR reaction. As indicated by the HRTEM
3
3
with that of anatase TiO2 and hematite Fe O3 is shown in
image of the FeTiO catalyst (Figure 2C), no discernible lattice
2
x
Figure 1. As we can clearly see, the FeTiO catalyst exhibited
fringes could be observed, which was in good accordance with
the broad diffraction bumps in XRD patterns and indistinct
scattering bands in Raman spectra as presented in our previ-
x
high NH -SCR performance over a broad temperature range,
3
with over 80% NO conversion obtained from 200 to 4008C.
x
The pristine TiO prepared under
2
our conditions exhibited no
deNOx efficiency below 3508C,
and only 30% NOx conversion
could be achieved at tempera-
tures as high as 4008C. As for
pristine Fe O also prepared by
2
3
ourselves, the maximum NOx
conversion could only reach
60%, and the operation temper-
ature window was rather narrow.
These results indicate that the
coexistence of Fe and Ti species
in the FeTiO catalyst is impor-
x
tant to obtain high deNO effi-
x
ciency, which can be attributed
to the strong interaction be-
tween Fe and Ti species leading
to larger surface areas, smaller
particle sizes, higher NO oxida-
tion ability and more abundant
[32]
Figure 2. HRTEM images of A) hematite Fe
2 3 2 x
O , B) anatase TiO , and C) FeTiO catalyst, and D) proposed structural
[
31,32]
[34]
surface acid sites.
model of FeTiO catalyst derived from EXAFS curve-fitting results
x
.
ꢀ
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2013, 5, 3760 – 3769 3761

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
[
31]
ous study. Therefore, the XAFS method, which is very suita-
ble to be used for the characterization of such amorphous or
microcrystalline materials with no long-range order, was used
to elucidate the microstructure of iron titanate crystallite in
FeTiOx catalyst in detail. Using X-ray absorption near-edge
structure (XANES) and extended X-ray absorption fine structure
(EXAFS) data in combination with DFT calculations to simulate
the electronic density, we concluded that, different from the
3
+
crystal structure of Fe species in hematite Fe O and Fe O /
2
3
3
2
3
4+
+
TiO supported-type catalyst, a homogeneous Fe ꢀ(O) ꢀTi
2
2
structure was clearly formed in the FeTiO catalyst, in which
x
3
+
4+
the Fe species and Ti species were strongly linked in an
edge-shared fashion (as shown by the structural model in Fig-
[31]
Figure 3. H
2 x 2 3 2
-TPR profiles of FeTiO catalyst, hematite Fe O , anatase TiO ,
[
34]
and ilmenite FeTiO .
3
ure 2D). The electronic inductive effect was confirmed to be
3
+
4+
present between Fe species and Ti species by XANES and
[
31,32,34]
X-ray photoelectron spectroscopy (XPS) results,
effective-
extent from the reduction processes observed for g-Fe O
2 3
3
+
ly reducing the electron density around Fe species thus lead-
ing to the enhancement of NO adsorption, oxidation ability
nanorods enclosed by abundant reactive {110} and {100} crys-
[27]
tal facets. However, we could still calculate the H /Fe molar
2
and finally the low temperature NH -SCR performance. Further-
ratio for the reduction peak at low temperature to confirm the
reduction process of Fe species in hematite Fe O below
3
3
+
4+
more, the Fe ꢀ(O) ꢀTi structure also showed high durability
2
2
3
to SO poisoning in the NH -SCR reaction mainly attributable
5008C, which is typically the maximum reaction temperature
2
3
to the close and homogeneous combination of Fe species and
Ti species as well as the easy decomposition of sulfate on Ti
for the NH -SCR reaction. As shown in Figure 3, the H /Fe
3
2
molar ratio for the reduction peak of Fe O at 4058C was calcu-
2
3
[
36]
sites. In previous studies by other researchers, the impor-
tance of the complex generated between the active phase and
catalyst support was also emphasized, such as the AgꢀOꢀAl
lated to be 0.17, which was exactly consistent with the reduc-
tion process from Fe O to Fe O . Further raising the reduction
2
3
3
4
temperature could lead to the formation of FeO or metallic
[
37]
0
species in Ag/Al O catalysts for the CH -SCR of NO
x
and the
Fe , yet different reduction mechanisms have been proposed
2
3
4
[
38]
CeꢀOꢀTi species in CeTiO catalysts for the NH -SCR of NO .
by former researchers including the two-stage reduction of
x
3
x
[
39,40]
Therefore, the study on the microstructure and chemical prop-
erties of such catalytically active species is an important and
universal issue in the field of heterogeneous catalysis.
3Fe O !2Fe O !6Fe
and the three-stage reduction of
2
3
3
4
[27,41,42]
3Fe O !2Fe O !6FeO!6Fe.
XANES measurements
2
3
3
4
could clearly discriminate the possible mixed Fe species in re-
duced Fe O , which will be discussed in detail below.
2
3
Based on the H -TPR results of TiO and Fe O , the H reduc-
2
2
2
3
2
H -TPR results of FeTiO catalyst and reference samples
2
x
tion peaks observed for the FeTiO catalyst could be only as-
x
Although the microstructure of iron titanate crystallites in the
FeTiOx catalyst has been clearly elucidated in our previous
study, the reducibility of the Fe species has not been well rec-
ognized although it is actually very important for the comple-
cribed to the reduction of Fe species, because the Ti species
were barely reduced under these conditions. In the FeTiO cat-
x
alyst with much higher NH -SCR performance than TiO and
3
2
Fe O , the Fe species started to be reduced at approximately
2
3
tion of the redox cycle in the NH -SCR reaction. Therefore, an
1508C, 508C lower than pristine Fe O . This result clearly
2 3
3
H -TPR experiment for the FeTiOx catalyst was performed,
2
shows that the oxygen species (mainly surface-adsorbed
using hematite Fe O , anatase TiO , and ilmenite FeTiO as ref-
oxygen reduced at low temperatures) in the FeTiO catalyst is
2
3
2
3
x
erences, and the results are shown in Figure 3.
more labile probably owing to the strong interaction between
Fe species and Ti species, leading to rich surface defects for
the activation of oxygen. Interestingly, totally different from
the reduction processes of Fe O , only one composite reduc-
As we can clearly see, no notable reduction peaks were pres-
ent for pristine TiO , which indicates that the anatase support
2
did not participate in the redox cycle for the NH -SCR reaction,
3
2
3
owing to its lack of reducibility, only supplying adsorption sites
for the reactants in the deNO process. As for pristine Fe O ,
tion peak was observed for the FeTiO catalyst with the maxi-
x
mum H consumption rate at 4278C. According to the calculat-
x
2
3
2
3
+
there were two H2 consumption peaks located at 405 and
258C, respectively. According to the baseline of the H signal,
ed H /Fe molar ratio of 0.52, the Fe species in the FeTiO cat-
2 x
2
+
7
alyst could be totally reduced to Fe species below 5008C,
2
the Fe species in hematite Fe O started to be reduced at ap-
probably resulting in the transformation of iron titanate crys-
2
3
3
+
2+
proximately 2008C, similar to the initial reduction process of
tallite Fe TiO to ilmenite Fe TiO . Although the temperature
x 3
[
27]
g-Fe O nanorods prepared by Mou et al.. Possibly owing to
point for the maximum H consumption rate of the FeTiO cat-
2 x
2
3
the relatively large particle sizes and difficult H diffusion, the
alyst (4278C) was higher than that of hematite Fe O (4058C),
2 3
2
first reduction peak at 4058C did not exhibit Gaussian symme-
try, and the second reduction peak at 7258C, overlapping with
a shoulder reduction peak above 8008C, did not return to the
baseline until as high as 9008C, which was different to a certain
the reducibility of Fe species in the former catalyst was actually
much higher than that in the latter sample. Therefore, the syn-
ergistic combination of Fe species and Ti species in the FeTiO
x
3
+
catalyst not only produced a unique edge-shared Fe ꢀ(O) ꢀ
2
ꢀ
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2013, 5, 3760 – 3769 3762

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
4
+
Ti structure, but also resulted
in the easier deoxidation behav-
ior of Fe species at low tempera-
tures, which is beneficial to the
completion of the redox cycle
for the NH -SCR reaction and
3
thus the improvement of low-
temperature deNO efficiency. A
x
similar phenomenon has also
[
38]
been reported by Li et al., in
which the Ce species in the
form of CeꢀOꢀTi linkages in
a CeTiO catalyst with excellent
x
SCR
performance
showed
higher reducibility than that in
pristine CeO . Above 5008C, the
2
H reduction curve of the FeTiOx
2
catalyst was slightly above the
baseline, which could be attrib-
uted to the reduction of surface
2
+
0
layers of Fe TiO to metallic Fe
Figure 4. A) Normalized XANES spectra of FeK edge in Fe-containing samples; B) XANES linear-fitting results of
3
and leaving TiO2 (probably in
the rutile phase after 9008C re-
duction). The in situ formed sur-
FeK edge in hematite Fe
FeK edge in FeTiO catalyst after H
to samples after the H -TPR procedure to 9008C; see Experimental Section.
2
O
3
after H
2
reduction using FeO and Fe as references; C) XANES linear-fitting results of
x
2
reduction using FeTiO and Fe as references. Samples specified by “–H ” refer
3
2
2
0
face metallic Fe might form
2
+
3+
clusters or particles on the Fe TiO phase mainly caused by
of FeK XANES in the FeTiO catalyst indicate that the Fe spe-
3
x
the relatively weak interaction between metal and oxide sup-
cies in this catalyst exhibits the most severe structure distor-
tion and the smallest particle size, which is beneficial to the
generation of rich surface defects and thus high catalytic per-
formance. The similarity of FeK XANES patterns between the
FeTiO catalyst and the Fe TiO reference indicates that the
2
+
port. At the same time, no further reduction of Fe species in
2
+
Fe TiO phase was observed even at temperatures as high as
3
9008C. A similar phenomenon was also found on the ilmenite
FeTiO reference, for which the H -TPR profile revealed no ob-
3
2
x
2
5
2
+
vious reduction peak of Fe species in the whole temperature
iron titanate crystallites may have some FeꢀO or FeꢀOꢀTi coor-
range. This is another interesting characteristic of the H reduc-
dination shells similar to those in Fe TiO , which can be verified
2
2
5
tion behavior of the FeTiO catalyst, and the abovementioned
by EXAFS results. As for the H -reduced Fe O and FeTiO sam-
2 2 3 x
x
points of view are verified by the following XANES linear-fitting
results and EXAFS curve-fitting results.
ples, both XANES patterns of the FeK edge showed marked
changes, with the shift of absorption edges towards the lower
energy range and the substitution of the former pre-edge
0
peaks by new pre-edge peaks characteristic of metallic Fe ,
XANES results of FeK, TiK edges
mixing with other pre-edge peak features. The post-edge re-
gions of FeK XANES in H -reduced Fe O and FeTiO samples
Using Fe foil, FeO, FeTiO , Fe O , and Fe TiO as references, the
3
2
3
2
5
2
2
3
x
XANES spectra of the FeK edge in FeTiO catalysts before and
also exhibited mixed patterns, not similar to those of either
x
after H reduction together with H -reduced Fe O were re-
FeO or FeTiO individually. Therefore, FeK XANES linear fitting
2
2
2
3
3
corded, as shown in Figure 4A. According to the pre-edge
was performed for Fe O and FeTiO samples after H reduction
2
3
x
2
peak position, the Fe species in the FeTiO catalyst was mainly
by using FeO, FeTiO and Fe foil as references to better con-
x
3
3
+
present in the form of Fe , similar to those in Fe O3 and
firm the existing state of Fe species in these H -reduced sam-
2
2
Fe TiO . According to the first-order derivatives of FeK XANES
ples, and the fitting results are shown in Figure 4B and 4C, re-
spectively. By linear combination of FeK XANES data in FeO
and Fe foil, the XANES spectrum of H -reduced Fe O is well si-
2
5
[
34]
as reported in our previous study,
the absorption edge
3
+
energy of Fe species in the FeTiO catalyst (7123.6 eV) was
x
2
2
3
higher than those in Fe O (7123.2 eV) and Fe TiO (7123.4 eV),
mulated with high fitting degree. As the fitting data in Table 1
reveal, the contribution of FeO and metallic Fe to the overall
FeK XANES of H -reduced Fe O is approximately 66.3% and
2
3
2
5
3
+
mainly owing to the presence of the most abundant Fe
ꢀ
4
+
(
O) ꢀTi structure with an electronic inductive effect between
2
2
2
3
3
+
4+
Fe and Ti species, resulting in the higher oxidation ability
33.7%, respectively. This result suggests that after the reduc-
3
+
of Fe species. This result confirmed by XPS data in our previ-
tion of Fe O to Fe O below 5008C, the Fe species could be
2
3
3
4
[
31]
0
ous study is also in good accordance with the H -TPR results,
reduced further to FeO and then to metallic Fe mainly follow-
ing three-stage reduction mechanism as mentioned
above.
2
3
+
in which the Fe species in the FeTiO catalyst is much more
a
[
x
27,41,42]
easily reduced than that in pristine Fe O . Besides, the highest
As for the H -reduced FeTiO catalyst, no FeO
2 x
2
3
pre-edge peak intensity and the smoothest post-edge region
phase can be abstracted from the FeK XANES spectrum, con-
ꢀ
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2013, 5, 3760 – 3769 3763

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
the conclusions from the FeK XANES experiments. Based on
the first-order derivatives of TiK XANES in our previous
Table 1. XANES linear-fitting results of FeK edge and TiK edge in hema-
tite Fe
2
O
3
and FeTiO
x
catalyst after H
2
reduction.
[34]
4+
study,
the absorption edge energy of Ti species in the
Sample
FeK XANES
TiK XANES
Molar ratio [%]
FeTiO catalyst (4983.4 eV) was lower than those in anatase
x
Ref.
FeO
Fe foil
FeTiO
Fe foil
Molar ratio [%]
Ref.
TiO (4985.5 eV) and Fe TiO (4984.3 eV), which can be simply
2
2
5
Fe
2
O
3
-H
2
66.3
33.7
76.6
23.4
–
–
–
–
explained by the reverse electronic inductive effect between
3+
4+
3+
4+
Fe and Ti species in the most abundant Fe ꢀ(O) ꢀTi
2
FeTiO
x
-H
2
3
FeTiO
rutile TiO
3
72.5
27.5
structures. As for the H -reduced FeTiO catalyst, both the pre-
2
x
2
edge peak and post-edge region revealed a marked change,
4
+
exhibiting a mixed state of Ti species. As presented in Fig-
ure 5B, the XANES spectrum of the TiK edge in the FeTiO cat-
x
firming again that no Fe O phase existed in the as-prepared
alyst after H reduction can be well simulated by using ilmenite
2
3
2
FeTiO catalyst and a new structure of Fe species was indeed
FeTiO and rutile TiO as references, which is in good accord-
x
3
2
formed. Using FeTiO and Fe foil as references, the FeK XANES
ance with the H -TPR and FeK XANES results. The contributions
3
2
of the FeTiO catalyst after H reduction is well simulated with
of ilmenite FeTiO and rutile TiO to the overall XANES result
x
2
3
2
high fitting degree, which is consistent with our viewpoint
from the perspective of TiK edge are approximately 72.5% and
27.5%, respectively (Table 1), confirming again that the forma-
concerning the H -TPR result in Figure 3. As shown in Table 1,
2
0
the contribution of FeTiO and metallic Fe to the overall FeK
tion of metallic Fe resulted in the presence of residual rutile
3
XANES of the FeTiO catalyst after H reduction is approximate-
TiO in an approximately equivalent amount. In brief, not only
x
2
2
ly 76.6% and 23.4%, respectively. Owing to the homogeneous
from the FeK XANES results but also from the TiK XANES re-
3
+
4+
3+
4+
state of the Fe ꢀ(O) ꢀTi structure in the iron titanate crys-
sults, we can conclude that the Fe and Ti species are ho-
mogeneously dispersed at the atomic scale, and the electronic
inductive effect between them results in the enhancement of
oxidation ability of Fe species leading to the more facile deoxi-
2
tallites, the FeTiO catalyst could be uniformly transformed into
x
FeTiO at relatively low temperatures, revealing a deoxidation
3
behavior totally different from that of pristine Fe O . Further in-
2
3
creasing the reduction temperature indeed only led to the for-
dation behavior of the FeTiO catalyst, which is beneficial to
x
0
mation of metallic Fe cluster or particles, leaving a majority of
the completion of the redox cycle in the NH -SCR reaction and
3
Fe species in the form of ilmenite FeTiO . Owing to the stoi-
thus the promotion of deNO efficiency.
3
x
chiometric balance between Fe and Ti species (1:1) in the as-
0
prepared FeTiO catalyst, the formation of metallic Fe in the
x
EXAFS results of FeK and TiK edges
H -reduced sample must result in the formation of excess TiO2
2
phase, which can be verified by the following TiK XANES
results.
To further confirm the existing states of Fe species in the
FeTiO catalyst and Fe O sample after H reduction, the EXAFS
x
2
3
2
The XANES spectra of TiK edge in FeTiO catalysts before
data of the FeK edge were curve-fitted by using the least-
square method. In Figure 6A, the Fourier transforms of filtered
x
and after H reduction were recorded by using Ti foil, anatase
2
TiO2, rutile TiO2, FeTiO3, and
Fe TiO5 as references, and the
2
results are shown in Figure 5A.
As we can clearly see, not only
the pre-edge peak shape but
also the post-edge region of TiK
XANES in the as-prepared FeTiOx
catalyst was totally different
from those in anatase TiO and
2
rutile TiO , indicating the pres-
2
4
+
ence of Ti species in another
coordination state. Notably, the
XANES pattern of the TiK edge
in FeTiO catalyst was similar to
x
that in Fe TiO but with higher
2
5
pre-edge peak intensity and
smoother post-edge region,
suggesting that the iron titanate
crystallites with smaller particle
sizes have TiꢀO or TiꢀOꢀFe co-
ordination shells similar to those
Figure 5. A) Normalized XANES spectra of TiK edge in Ti-containing samples; B) XANES linear fitting results of TiK
in Fe TiO yet with more severe
2
5
edge in FeTiO catalyst after H reduction using FeTiO and Fe as references. Samples specified by “–H ” refer to
x
2
3
2
structure distortion, similar to
2
samples after the H -TPR procedure to 9008C; see Experimental Section.
ꢀ
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2013, 5, 3760 – 3769 3764

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
be uniformly reduced to ilmen-
ite FeTiO phase as a result of
3
3
+
the homogeneous Fe ꢀ(O) ꢀ
2
4
+
Ti structure with unique deox-
idation behavior. These conclu-
sions are well supported by the
curve-fitting results of FeK edge
EXAFS (Table 2) with high
degree of fit between the exper-
imental data and FEFF-calculat-
ed data (Figure 6B). The relative-
ly small coordination numbers
of FeꢀFe shells in the H -re-
2
duced FeTiO catalyst also sug-
x
0
gest that the metallic Fe
formed after high-temperature
reduction exhibited rather small
particle sizes, probably resulting
from the weak interaction be-
0
tween Fe and the oxide sup-
3
Figure 6. A) Fourier transforms of filtered EXAFS oscillations k c(k) into R space and B) corresponding filtered
port and also the hindering of
3
ꢀ1
k c(k) of FeK edge in Fe-containing samples in the k range of 2.5–15 ꢁ . The dotted lines correspond to the
curve-fitting results in k space calculated by using the FEFF8.4 code. * Indicate Fe ; ^ FeO; ! FeTiO
0
Fe aggregation by the in situ
0
3
.
released TiO2.
In a similar way, the TiK edge
EXAFS data of FeTiOx catalysts
3
EXAFS oscillations k c(k) of the FeK edge in all Fe-containing
before and after H reduction were curve-fitted by the least-
2
ꢀ
1
samples into R space in the k range of 2.5–15 ꢁ are shown,
square method using different Ti-containing reference models,
3
3
and in Figure 6B, the corresponding filtered k ·c(k) plots are
with the Fourier transformed k c(k) shown in Figure 7A in the
ꢀ
1
3
shown, in which the dotted lines correspond to the curve-fit-
ting results in k space calculated by using the FEFF8.4 code
with different reference models. The corresponding curve-
fitted data of the FeK edge are shown in Table 2. The as-pre-
k range of 2.5–13 ꢁ and the filtered k c(k) shown in Fig-
ure 7B, in which the dotted lines correspond to the FEFF-calcu-
lated scattering pathways. The relevant curve-fitted data of TiK
edge EXAFS results are shown in Table 3. Similar to the case of
FeK edge EXAFS experiments, a well-defined first TiꢀO coordi-
nation shell and a rather small second coordination shell can
pared FeTiO catalyst clearly exhibited a well-defined first FeꢀO
x
coordination shell and a rather small second coordination
shell, in complete contrast to Fe O or Fe TiO , which exhibit
be observed for the FeTiO catalyst, the second shell of which
2
3
2
5
x
[34]
clear bimodal coordination peaks. In our previous study, this
unique second coordination shell was attributed to the scatter-
ing pathway of FeꢀOꢀTi, and the FeꢀOꢀTi bond length was
calculated to be 3.09 ꢁ. This bond length was shorter than
that in well-crystallized Fe TiO (3.18 ꢁ) and belonged to the
is different from those in anatase TiO , rutile TiO , or Fe TiO ,
2
2
2
5
which reveal clear bimodal coordination peaks. According to
the curve-fitting results, the small second coordination shell of
the FeTiO catalyst can be attributed to the scattering pathway
x
of TiꢀOꢀFe with smaller bond length (3.08 ꢁ) than that in
2
5
3
+
4+
edge-shared Fe ꢀ(O) ꢀTi structure, causing the electronic
Fe TiO (3.17 ꢁ). From the perspective of the central Ti species,
2
2
5
3
+
4+
4+
inductive effect between Fe and Ti species to take place
more easily, thus resulting in its specific deoxidation behavior
this bond length can be ascribed to the edge-shared Ti
ꢀ
3
+
(O) ꢀFe
structure, in which the TiꢀOꢀFe coordination
2
and high NH -SCR performance. The small coordination
number is only 2.0 because of the crystallite phase of the
FeTiO catalyst. After H reduction, not only the first but also
3
number of FeꢀOꢀTi (2.6) in the FeTiO catalyst is in good ac-
x
x
2
cordance with the XANES results, confirming again the exis-
tence of abundant structural defects in iron titanate crystallites,
which is advantageous to the adsorption and activation of re-
actants for the deNO process. After H reduction, both the
the second coordination shells of Ti species revealed some no-
table variations, exhibiting overlapping peak shapes. Using il-
menite FeTiO and rutile TiO as references, the TiK EXAFS data
3
2
in the H -reduced FeTiO catalyst could be well simulated with
x
2
2
x
Fe O sample and FeTiO catalyst presented coordination shells
high fitting degree, which is quite consistent with the TiK
edge XANES results.
2
3
x
with a mixed state, for which the common peaks in Figure 6A
0
(
indicated by *) can be attributed to metallic Fe as mentioned
above. For H -reduced Fe O , the FeꢀO and FeꢀOꢀFe shells be-
2
2
3
Proposed deoxidation process of the FeTiO catalyst
x
longing to FeO can be clearly observed (indicated by ^). As for
the H -reduced FeTiO catalyst, the FeꢀO and FeꢀOꢀM (M=Fe
Based on the above-mentioned H -TPR, XANES linear fitting,
2
x
2
or Ti) shells ascribed to FeTiO3 can be observed (indicated
and EXAFS curve-fitting results, the specific deoxidation behav-
by !), confirming again that the iron titanate crystallites can
ior of FeTiO catalyst can be proposed as shown in Scheme 1.
x
ꢀ
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2013, 5, 3760 – 3769 3765

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
Table 2. Calibrated EXAFS curve-fitting results of FeK edge in different samples (M=Fe or Ti).
[
a]
[b]
[c]
[d]
Sample
Ref.
Shell
CN
R
[ꢁ]
DE [eV]
DW [ꢁ]
R factor [%]
0.5
FeꢀFe
FeꢀFe
FeꢀFe
FeꢀFe
8.0ꢁ3.7
6.0ꢁ6.8
2.48ꢁ0.03
2.87ꢁ0.07
4.05ꢁ0.06
4.75ꢁ0.02
ꢀ4.7ꢁ7.4
ꢀ7.1ꢁ12.7
ꢀ11.3ꢁ6.1
ꢀ4.1ꢁ2.0
0.072ꢁ0.033
0.081ꢁ0.080
0.099ꢁ0.063
0.092ꢁ0.024
1
2
3
Fe foil
Fe
12.0ꢁ9.6
24.0ꢁ7.7
FeꢀO
6.0ꢁ2.3
2.10ꢁ0.03
3.06ꢁ0.01
ꢀ11.2ꢁ4.7
ꢀ8.0ꢁ1.8
0.092ꢁ0.051
0.108ꢁ0.018
FeO
FeO
1.8
6.1
FeꢀOꢀFe
12.0ꢁ2.6
FeꢀO
6.0ꢁ1.2
4.0ꢁ1.0
3.0ꢁ1.0
2.10ꢁ0.03
2.98ꢁ0.02
3.38ꢁ0.02
ꢀ7.9ꢁ3.5
2.1ꢁ2.3
0.097ꢁ0.030
0.111ꢁ0.022
0.074ꢁ0.030
FeTiO
3
FeTiO
3
FeꢀOꢀM
FeꢀOꢀM
1
2
17.5ꢁ2.5
FeꢀO
6.0ꢁ1.5
3.0ꢁ0.6
3.0ꢁ1.0
1.95ꢁ0.03
2.97ꢁ0.02
3.36ꢁ0.02
ꢀ15.1ꢁ5.4
ꢀ2.8ꢁ2.3
14.0ꢁ3.2
0.101ꢁ0.037
0.095ꢁ0.017
0.064ꢁ0.028
Fe
2
2
O
3
Fe
2
O
3
FeꢀOꢀFe
FeꢀOꢀFe
1
4.2
2
FeꢀO
3.0ꢁ1.1
3.0ꢁ2.3
3.0ꢁ2.7
1.95ꢁ0.02
3.18ꢁ0.04
3.67ꢁ0.05
ꢀ15.9ꢁ3.3
1.9ꢁ7.1
0.089ꢁ0.018
0.072ꢁ0.043
0.093ꢁ0.051
Fe
TiO
5
Fe
Fe
2
2
TiO
TiO
5
5
FeꢀOꢀM
FeꢀOꢀM
1
9.0
4.5
2
ꢀ16.2ꢁ7.1
FeꢀO
7.7ꢁ1.7
2.6ꢁ2.5
1.96ꢁ0.01
3.09ꢁ0.04
ꢀ8.6ꢁ2.2
4.6ꢁ4.4
0.095ꢁ0.016
0.104ꢁ0.045
FeTiO
x
FeꢀOꢀTi
FeꢀO
4.2ꢁ2.9
8.2ꢁ1.8
0.8ꢁ0.4
0.5ꢁ0.8
1.0ꢁ1.6
2.2ꢁ0.7
2.10ꢁ0.03
3.06ꢁ0.01
2.48ꢁ0.03
2.87ꢁ0.07
4.05ꢁ0.06
4.75ꢁ0.02
ꢀ5.2ꢁ3.8
ꢀ4.0ꢁ1.2
ꢀ2.5ꢁ0.6
ꢀ6.8ꢁ12.2
ꢀ7.9ꢁ0.0
ꢀ4.2ꢁ11.3
0.099ꢁ0.071
0.107ꢁ0.01
0.063ꢁ0.029
0.067ꢁ0.066
0.072ꢁ0.046
0.075ꢁ0.020
FeꢀOꢀFe
*FeꢀFe
FeO
Fe
Fe
2
O
3
-H
2
7.1
*
*FeꢀFe
1
2
3
*
*
FeꢀFe
FeꢀFe
FeꢀO
3.4ꢁ2.5
2.3ꢁ0.6
1.7ꢁ2.9
1.5ꢁ3.5
0.9ꢁ2.2
2.3ꢁ1.9
4.7ꢁ2.9
2.10ꢁ0.03
2.98ꢁ0.02
3.38ꢁ0.02
2.48ꢁ0.03
2.87ꢁ0.07
4.05ꢁ0.06
4.75ꢁ0.02
ꢀ4.4ꢁ4.7
1.8ꢁ2.8
0.080ꢁ0.081
0.105ꢁ0.021
0.070ꢁ0.028
0.058ꢁ0.027
0.062ꢁ0.061
0.077ꢁ0.049
0.080ꢁ0.021
FeꢀOꢀM
FeꢀOꢀM
*FeꢀFe
1
2
16.0ꢁ0.0
ꢀ4.2ꢁ0.1
ꢀ12.6ꢁ0.7
ꢀ13.2ꢁ1.1
ꢀ3.7ꢁ0.1
FeTiO
Fe
3
FeTiO
x
-H
2
1.2
*
*FeꢀFe
*FeꢀFe
*FeꢀFe
1
2
3
[a] For information on the calibration of (reference) coordination numbers and the treatment of error margins, see the Experimental Section. [b] Coordina-
tion number. [c] Bond length. [d] Debye–Waller factor.
3
+
4+
Owing to the homogeneous state of Fe species and Ti
species with electronic inductive effect in the edge-shared Fe
tallites in the FeTiO catalyst results in the specific deoxidation/
x
3
+
redox behavior of the Fe species, thus leading to high NH -SCR
3
4
+
3+
ꢀ
(O) ꢀTi structure of iron titanate crystallites, the Fe spe-
performance at low temperatures.
2
cies with higher oxidation ability can be totally converted into
2
+
Fe species below 5008C, which is a much easier process
Conclusions
3
+
than the corresponding reduction of Fe species in pristine
Fe O . It is noteworthy that the NH -SCR reaction mainly
The advantageous combination of Fe species and Ti species in
the FeTiO catalyst resulted in its high NO reduction efficiency
2
3
3
occurs over the FeTiO catalyst below 5008C with high deNO
x
x
x
x
efficiency, which is closely related to the enhanced redox abili-
in the NH selective catalytic reduction (NH -SCR) of NO with
3 3 x
3
+
3+
4+
ty of Fe species in the specific Fe ꢀ(O) ꢀTi structure. The
a broad operation temperature window and good SO durabili-
2
2
2
+
3+
4+
as-formed Fe species in the H -reduced sample combines
ty. The unique edge-shared Fe ꢀ(O) ꢀTi structure in iron ti-
2
2
4
+
with Ti species to exist as ilmenite FeTiO , but not as FeO as
tanate crystallites was confirmed to be the true catalytically
3
3
+
in the Fe O case, indicating again that the Fe species and Ti
active site, with an electronic inductive effect between Fe
2
3
4
+
species are present in a high dispersion state at the atomic
species and Ti species leading to the enhanced oxidation
3
+
3+
scale. Above 5008C, the surface layers of as-formed FeTiO can
ability of Fe species. In the H -TPR process, the Fe species
2
3
2
+
be further reduced, which results in the formation of partial
in the FeTiO catalyst can be totally reduced to Fe species
x
0
3+
metallic Fe clusters or particles and residual rutile TiO . How-
below 5008C in the form of ilmenite FeTiO , whereas the Fe
2
3
2
+
ever, the reduction of Fe
species in bulk FeTiO3 phase
species in Fe O can only be reduced to Fe O . The typical
2
3
3
4
cannot be achieved even at temperatures as high as 9008C. In
short summary, the unique microstructure of iron titanate crys-
NH -SCR reaction mainly occurs below 5008C, and the en-
3
3
+
hanced redox ability of Fe species in FeTiO catalyst is bene-
x
ꢀ
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2013, 5, 3760 – 3769 3766

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
were controlled as follows: [NO]=
[
NH ]=500 ppm,
[O ]=5 vol%,
2
3
and N2 balance; 0.6 mL catalyst,
2
5
0–40 mesh; total flow rate of
ꢀ1
00 mLmin , and gas hourly
velocity (GHSV) of
0000 h . The effluent gas was
space
5
ꢀ1
continuously analyzed by using an
FTIR spectrometer (Nicolet Nexus
6
70) equipped with a heated, low-
volume multiple-path gas cell
2 m). The HRTEM images of the
FeTiOx catalyst, hematite Fe O ,
(
2
3
and anatase TiO with high mag-
2
nification were obtained on
a JEOL 2011 TEM (JEOL).
H -TPR experiments
2
Prior to the H -TPR procedure, the
2
samples (100 mg) were pretreated
in
a
flow of 20 vol% O /He
2
3
ꢀ1
Figure 7. A) Fourier transforms of filtered EXAFS oscillations k c(k) into R space and B) corresponding filtered
(30 mLmin ) at 3008C for 0.5 h
and cooled down to RT. Then the
temperature was raised linearly to
3
ꢀ1
k c(k) of TiK edge in Ti-containing samples in the k range of 2.5–13 ꢁ . The dotted lines correspond to the
curve-fitting results in k space calculated by using the FEFF8.4 code. * Indicate rutile TiO
; ! FeTiO
2
3
.
ꢀ
1
9
a
008C at a rate of 108Cmin in
flow of 5 vol% H /Ar
2
ꢀ
1
(
30 mLmin ). The H signal (m/z=2) was monitored
2
online by using
a quadrupole mass spectrometer
(
HPR20, Hiden Analytical Ltd.) to obtain the H -TPR pro-
2
files. After the reduction temperature reaching 9008C,
the reduced Fe O3 and FeTiO samples were naturally
2
x
cooled to RT under the H protection, and then put into
2
sealed bags for the following XAFS measurements.
XAFS measurements
2 x
Scheme 1. Proposed H reduction process of the FeTiO catalyst.
The XANES and EXAFS spectra of FeK and TiK edges
were measured in transmission mode at RT on the BL-7C
beam line, Photon Factory, KEK, Japan and BL14W1 beam line,
Shanghai Synchrotron Radiation Facility (SSRF). The storage ring
was operated at 2.5 GeV with 300 mA as an average storage cur-
rent for the BL-7C beam line and 3.5 GeV with 200 mA as an aver-
age storage current for the BL14W1 beam line. The synchrotron ra-
diation beam line was monochromatized with a Si (111) double
crystal monochromator, and mirrors were used to eliminate higher
harmonics. The incident and transmitted beam intensities were
ficial to enhance its catalytic performance. Based on XANES
linear-fitting and EXAFS curve-fitting results, the unique deoxi-
dation behavior of the FeTiO catalyst in the H -TPR process
x
2
could be clearly elucidated. The results in this study can supply
some ideas for the investigation of the local structure and
redox ability of active sites in mixed oxide catalysts simultane-
ously for certain catalytic reactions.
monitored using ionization chambers filled with pure N . XAFS
2
data were analyzed by using the REX2000 program (Rigaku Co.).
XANES spectra were normalized with edge height. EXAFS oscilla-
tion c(k) was extracted by using spline smoothing with a Cook–
Experimental Section
[43]
3
Sayers criterion and weighted by k in order to compensate for
the diminishing amplitude in the high k range because of the
Catalyst preparation, NH -SCR performance test, and HRTEM
3
characterization
3
decay of the photoelectron wave. Thereafter, the filtered k -weight-
ꢀ
1
FeTiO catalyst with Fe/Ti molar ratio of 1:1 was prepared by a co-
ed c(k) was Fourier transformed into R space (k range: 2.5–15 ꢁ
x
ꢀ1
precipitation method using Fe(NO ) ·9H O and Ti(SO ) as precur-
for FeK EXAFS and 2.5–13 ꢁ for TiK EXAFS) with a Hanning func-
tion window. In the curve-fitting step, the possible backscattering
amplitude and phase shift were calculated by using the FEFF8.4
3
3
2
4 2
sors and NH ·H O (25 wt%) as precipitator. The resulting precipi-
3
2
tate was filtrated, washed with distilled water, dried at 1008C for
2 h, and finally calcined in air at 4008C for 6 h. As reference sam-
ples, pristine hematite Fe O and anatase TiO were also prepared
[44]
1
code. For standard reference samples, the coordination numbers
(CN), the bond lengths (R), the edge corrections (DE) and the
Debye–Waller factors were all set to be adjustable. And for the cat-
alysts containing confirmed components by XANES linear fitting re-
sults, the bond lengths of specific shells were fixed exactly to the
2
3
2
by the same method. The steady-state NH -SCR performance of
prepared samples was tested in a fixed-bed quartz tube reactor at
atmospheric pressure, and the reaction conditions of standard SCR
3
ꢀ
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2013, 5, 3760 – 3769 3767

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
Table 3. Calibrated EXAFS curve-fitting results of TiK edge in different samples (M=Fe or Ti)
[
a]
[b]
[c]
[d]
Sample
Ti foil
Ref.
Ti
Shell
Ti-Ti
CN
R
[ꢁ]
DE [eV]
4.4ꢁ1.3
DW [ꢁ]
R factor [%]
1.2
6.0ꢁ0.9
2.90ꢁ0.01
0.095ꢁ0.015
TiꢀO
6.0ꢁ1.0
3.0ꢁ0.9
3.0ꢁ1.6
1.98ꢁ0.01
3.02ꢁ0.02
4.03ꢁ0.03
12.5ꢁ2.0
4.4ꢁ2.9
0.101ꢁ0.019
0.077ꢁ0.027
0.088ꢁ0.043
anatase TiO
2
anatase TiO
2
TiꢀOꢀTi
TiꢀOꢀTi
1
2
1.7
2.9
3.6
ꢀ7.6ꢁ4.1
TiꢀO
6.0ꢁ1.3
2.0ꢁ1.2
4.0ꢁ1.4
1.98ꢁ0.02
2.96ꢁ0.04
3.57ꢁ0.02
ꢀ6.4ꢁ2.9
ꢀ17.1ꢁ6.5
ꢀ13.0ꢁ3.2
0.096ꢁ0.024
0.071ꢁ0.059
0.066ꢁ0.035
rutile TiO
2
rutile TiO
2
TiꢀOꢀTi
TiꢀOꢀTi
1
2
TiꢀO
6.0ꢁ1.6
4.0ꢁ0.8
3.0ꢁ1.3
1.86ꢁ0.02
2.96ꢁ0.01
3.38ꢁ0.03
ꢀ17.2ꢁ4.3
ꢀ8.4ꢁ2.0
9.9ꢁ3.5
0.082ꢁ0.039
0.060ꢁ0.024
0.081ꢁ0.037
FeTiO
3
FeTiO
3
TiꢀOꢀM
TiꢀOꢀM
1
2
TiꢀO
6.0ꢁ1.1
3.0ꢁ1.4
3.0ꢁ2.8
1.96ꢁ0.02
3.17ꢁ0.03
3.72ꢁ0.05
ꢀ8.4ꢁ2.7
ꢀ8.1ꢁ4.5
11.0ꢁ7.2
0.105ꢁ0.023
0.082ꢁ0.036
0.071ꢁ0.094
Fe
2
TiO
5
Fe
Fe
2
2
TiO
5
5
TiꢀOꢀM
TiꢀOꢀM
1
3.3
6.9
2
TiꢀO
6.9ꢁ1.7
2.0ꢁ1.7
1.97ꢁ0.02
3.08ꢁ0.04
ꢀ5.8ꢁ2.5
ꢀ2.6ꢁ2.3
0.125ꢁ0.021
0.119ꢁ0.052
FeTiO
FeTiO
x
TiO
TiꢀOꢀFe
TiꢀO
2.8ꢁ1.7
1.9ꢁ0.4
1.5ꢁ1.5
3.2ꢁ0.7
1.1ꢁ0.7
2.1ꢁ0.7
1.86ꢁ0.02
2.96ꢁ0.01
3.38ꢁ0.03
1.98ꢁ0.02
2.96ꢁ0.04
3.57ꢁ0.02
ꢀ13.0ꢁ7.5
ꢀ7.6ꢁ1.6
0.0ꢁ0.0
0.062ꢁ0.065
0.043ꢁ0.022
0.072ꢁ0.049
0.111ꢁ0.036
0.095ꢁ0.023
0.102ꢁ0.020
TiꢀOꢀM
TiꢀOꢀM
*TiꢀO
1
FeTiO
rutile TiO
3
2
x
-H
2
5.8
*
2
ꢀ7.5ꢁ11.2
ꢀ25.6ꢁ15.2
19.7ꢁ1.0
*
*
TiꢀOꢀTi
1
2
TiꢀOꢀTi
[a] For information on the calibration of (reference) coordination numbers and the treatment of error margins, see the Experimental Section. [b] Coordina-
tion number. [c] Bond length. [d] Debye–Waller factor.
values derived from reference samples, and the rest parameters
were set to be adjustable. Thus, we adopted the same error inter-
vals of the fixed bond lengths for the studied catalysts as those in
reference samples. During this curve-fitting step, the numbers of
free parameters for all samples comply with the Nyquist criterion.
After the EXAFS curve fitting of standard reference samples (includ-
Keywords: iron titanate · nitrogen oxides · selective catalytic
reduction · deoxidation behavior · redox ability · X-ray
absorption fine-structure
ing Fe foil, FeO, FeTiO , Fe O , Fe TiO , Ti foil, anatase TiO , and
3
2
3
2
5
2
[1] S. Beirle, K. F. Boersma, U. Platt, M. G. Lawrence, T. Wagner, Science
2011, 333, 1737–1739.
rutile TiO ), their coordination numbers and bond lengths were all
2
calibrated to the corresponding values in theoretical crystal struc-
tures. Thus, more reasonable fitting data could be obtained that
were closer to the real situations. Based on the calibrated fitting
data of standard reference samples, both the coordination num-
bers and bond lengths as well as the corresponding error intervals
in the studied catalysts were also calibrated.
[2] R. A. Perry, D. L. Siebers, Nature 1986, 324, 657–658.
[3] G. Busca, L. Lietti, G. Ramis, F. Berti, Appl. Catal. B 1998, 18, 1–36.
[
[
[
[
[
4] H. Bosch, F. Janssen, Catal. Today 1988, 2, 369–379.
5] N.-Y. Topsøe, Science 1994, 265, 1217–1219.
6] P. Granger, V. I. Parvulescu, Chem. Rev. 2011, 111, 3155–3207.
7] D. M. Chapman, Appl. Catal. A 2011, 392, 143–150.
8] M. Casanova, K. Schermanz, J. Llorca, A. Trovarelli, Catal. Today 2012,
1
84, 227–236.
9] D. W. Fickel, E. D’Addio, J. A. Lauterbach, R. F. Lobo, Appl. Catal. B 2011,
02, 441–448.
[
Acknowledgements
1
[
[
10] S. Roy, M. S. Hegde, G. Madras, Appl. Energy 2009, 86, 2283–2297.
11] M. Iwasaki, K. Yamazaki, K. Banno, H. Shinjoh, J. Catal. 2008, 260, 205–
This work was supported by the National Natural Science Foun-
dation of China (51108446), the Ministry of Science and Technolo-
gy, China (2012AA062506, 2013AA065301), and the Photon Fac-
tory, KEK, Japan (Project No. 2012G537). We sincerely thank Prof.
Kiyotaka Asakura from Catalysis Research Center, Hokkaido Uni-
versity, Dr. Yasuhiro Niwa from KEK and Prof. Yuying Huang,
Zheng Jiang, Shuo Zhang and other staffs from BL14W1 beam
line, SSRF for their generous help in XAFS experiments.
216.
[
12] U. Deka, A. Juhin, E. A. Eilertsen, H. Emerich, M. A. Green, S. T. Korhonen,
B. M. Weckhuysen, A. M. Beale, J. Phys. Chem. C 2012, 116, 4809–4818.
[13] L. Ren, L. Zhu, C. Yang, Y. Chen, Q. Sun, H. Zhang, C. Li, F. Nawaz, X.
Meng, F.-S. Xiao, Chem. Commun. 2011, 47, 9789–9791.
[
[
14] J. H. Kwak, R. G. Tonkyn, D. H. Kim, J. Szanyi, C. H. F. Peden, J. Catal.
010, 275, 187–190.
15] C. H. F. Peden, J. H. Kwak, S. D. Burton, R. G. Tonkyn, D. H. Kim, J.-H. Lee,
2
H.-W. Jen, G. Cavataio, Y. Cheng, C. K. Lambert, Catal. Today 2012, 184,
2
45–251.
16] M. Høj, M. J. Beier, J.-D. Grunwaldt, S. Dahl, Appl. Catal. B 2009, 93,
66–176.
[
[
1
17] M. Schwidder, S. Heikens, A. De Toni, S. Geisler, M. Berndt, A. Brꢂckner,
W. Grꢂnert, J. Catal. 2008, 259, 96–103.
ꢀ
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2013, 5, 3760 – 3769 3768

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
[
[
[
[
[
[
[
[
[
[
[
[
18] H. Sjçvall, L. Olsson, E. Fridell, R. J. Blint, Appl. Catal. B 2006, 64, 180–
88.
19] S. Brandenberger, O. Krçcher, A. Tissler, R. Althoff, Catal. Rev. 2008, 50,
92–531.
20] J. Hun Kwak, H. Zhu, J. H. Lee, C. H. F. Peden, J. Szanyi, Chem. Commun.
012, 48, 4758–4760.
21] F. Li, Y. Zhang, D. Xiao, D. Wang, X. Pan, X. Yang, ChemCatChem 2010, 2,
416–1419.
22] S. Yang, C. Wang, J. Li, N. Yan, L. Ma, H. Chang, Appl. Catal. B 2011, 110,
1–80.
23] S. Roy, B. Viswanath, M. S. Hegde, G. Madras, J. Phys. Chem. C 2008, 112,
002–6012.
24] J. Li, H. Chang, L. Ma, J. Hao, R. T. Yang, Catal. Today 2011, 175, 147–
56.
25] W. Shan, F. Liu, H. He, X. Shi, C. Zhang, Chem. Commun. 2011, 47, 8046–
048.
26] W. Shan, F. Liu, H. He, X. Shi, C. Zhang, ChemCatChem 2011, 3, 1286–
289.
27] X. Mou, B. Zhang, Y. Li, L. Yao, X. Wei, D. S. Su, W. Shen, Angew. Chem.
012, 124, 3044–3048; Angew. Chem. Int. Ed. 2012, 51, 2989–2993.
[31] F. Liu, H. He, C. Zhang, Z. Feng, L. Zheng, Y. Xie, T. Hu, Appl. Catal. B
2010, 96, 408–420.
[32] F. Liu, H. He, J. Phys. Chem. C 2010, 114, 16929–16936.
[33] F. Liu, H. He, C. Zhang, W. Shan, X. Shi, Catal. Today 2011, 175, 18–25.
[34] F. Liu, K. Asakura, P. Xie, J. Wang, H. He, Catal. Today 2013, 201, 131–
138.
[35] M. A. Khan, S. I. Woo, O. B. Yang, Int. J. Hydrogen Energy 2008, 33, 5345–
5351.
[36] F. Liu, K. Asakura, H. He, W. Shan, X. Shi, C. Zhang, Appl. Catal. B 2011,
103, 369–377.
[37] X. She, M. Flytzani-Stephanopoulos, J. Catal. 2006, 237, 79–93.
[38] P. Li, Y. Xin, Q. Li, Z. Wang, Z. Zhang, L. Zheng, Environ. Sci. Technol.
2012, 46, 9600–9605.
[39] H.-Y. Lin, Y.-W. Chen, C. Li, Thermochim. Acta 2003, 400, 61–67.
[40] H.-J. Wan, B.-S. Wu, C.-H. Zhang, H.-W. Xiang, Y.-W. Li, B.-F. Xu, F. Yi,
Catal. Commun. 2007, 8, 1538–1545.
[41] Y. Wu, H. Yu, F. Peng, H. Wang, Mater. Lett. 2012, 67, 245–247.
[42] W. K. Jozwiak, E. Kaczmarek, T. P. Maniecki, W. Ignaczak, W. Maniukie-
wicz, Appl. Catal. A 2007, 326, 17–27.
[43] J. W. Cook, D. E. Sayers, J. Appl. Phys. 1981, 52, 5024–5031.
[44] A. L. Ankudinov, B. Ravel, J. J. Rehr, S. D. Conradson, Phys. Rev. B 1998,
58, 7565–7576.
1
4
2
1
7
6
1
8
1
2
28] N. Apostolescu, B. Geiger, K. Hizbullah, M. T. Jan, S. Kureti, D. Reichert, F.
Schott, W. Weisweiler, Appl. Catal. B 2006, 62, 104–114.
29] Z. Huang, X. Gu, W. Wen, P. Hu, M. Makkee, H. Lin, F. Kapteijn, X. Tang,
Angew. Chem. 2013, 125, 688–692; Angew. Chem. Int. Ed. 2013, 52,
660–664.
Received: July 13, 2013
[30] F. Liu, H. He, C. Zhang, Chem. Commun. 2008, 2043–2045.
Published online on October 2, 2013
ꢀ
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2013, 5, 3760 – 3769 3769

Copyright of ChemCatChem is the property of Wiley-Blackwell and its content may not be
copied or emailed to multiple sites or posted to a listserv without the copyright holder's
express written permission. However, users may print, download, or email articles for
individual use.