Modified nano-crystalline ferrites for high-temperature WGS membrane reactor applications

Article abstract of DOI:10.1016/j.jcat.2007.10.018

In the present study, selected metal ions (M = Cr, Mn, Co, Ni, Cu, Zn, and Ce) were introduced into iron oxide (spinel lattice) and screened for effectiveness for a high-temperature water-gas shift reaction. Simultaneous precipitation of Fe(III) nitrates along with metal nitrate(s) at optimal concentrations resulted in the formation of high-surface area nanosized catalysts. A noticeable interaction between iron and the other substitutent metal was interpreted from the formation of either inverse or mixed spinels of composition A_(1-δ)B_δ[A_δB_(2-δ)]O₄. The incorporation of metal cations into the hematite crystal structure modified the magnetic hyperfine field and also influenced the reducibility of hematite particles, as observed in Moessbauer effect and temperature-programmed reduction studies. These effects strongly depend on the nature of incorporated metal cations. Moessbauer hyperfine fields, isomer shifts and transmission electron microscopy findings support nanoscale nature of the catalysts. Amongst catalysts tested, Fe/Cr and Fe/Ce are found to be the most active, with activity approaching equilibrium conversion at high temperatures.

Full text of DOI:10.1016/j.jcat.2007.10.018

Journal of Catalysis 253 (2008) 91–104
www.elsevier.com/locate/jcat
Modified nano-crystalline ferrites for high-temperature WGS membrane
reactor applications
a
b
b
a,∗
Ataullah Khan , Ping Chen , P. Boolchand , Panagiotis G. Smirniotis
a
Department of Chemical and Materials Engineering, University of Cincinnati, USA
b
Department of Electrical and Computer Engineering and Computer Science, University of Cincinnati, USA
Received 9 July 2007; revised 6 October 2007; accepted 21 October 2007
Available online 26 November 2007
Abstract
In the present study, selected metal ions (M = Cr, Mn, Co, Ni, Cu, Zn, and Ce) were introduced into iron oxide (spinel lattice) and screened
for effectiveness for a high-temperature water–gas shift reaction. Simultaneous precipitation of Fe(III) nitrates along with metal nitrate(s) at
optimal concentrations resulted in the formation of high-surface area nanosized catalysts. A noticeable interaction between iron and the other
substitutent metal was interpreted from the formation of either inverse or mixed spinels of composition A(
B [A B
]O . The incorporation
1−δ)
δ
δ
(2−δ)
4
of metal cations into the hematite crystal structure modified the magnetic hyperfine field and also influenced the reducibility of hematite particles,
as observed in Mössbauer effect and temperature-programmed reduction studies. These effects strongly depend on the nature of incorporated
metal cations. Mössbauer hyperfine fields, isomer shifts and transmission electron microscopy findings support nanoscale nature of the catalysts.
Amongst catalysts tested, Fe/Cr and Fe/Ce are found to be the most active, with activity approaching equilibrium conversion at high temperatures.
©
2007 Elsevier Inc. All rights reserved.
Keywords: HT WGS; Nanosized; Modified ferrites; Mössbauer; TPR
1
. Introduction
[AδB(
]O4 structures, where δ is the degree of inversion
2−δ)
[
11,12]. For a normal spinel, AB2O4, δ = 0, whereas for an
The water–gas shift (WGS) reaction using membrane reac-
inverse spinel, B[AB]O , δ = 1. The physicochemical proper-
4
tors are attractive largely because hydrogen can be selectively
permeated through a membrane, making complete conversion
possible [1,2]. In this manner, the broad consequences of com-
plete conversion can be explored. The underlying issues have
been examined by several authors through feasibility studies
and computational and real-time analyses [1–6]. A membrane
reactor typically operates at high temperatures and high pres-
sures [7]. Among the various membrane tested, silica-based
membranes have been found to be thermally robust, permeable,
and permselective for H over other gases in the syngas reactant
mixture [8–10].
Substitution of ‘Fe’ cation sites by other metal (M) atom
substitutents in the ferrite structure generally promotes forma-
tion of mixed (0 < δ < 1) or inverse (δ = 1) spinel A(1−δ)Bδ-
ties of ferrites are strongly dependent on the site, nature, and
amount of M substitutents at the A and B sites in the struc-
ture. Furthermore, ion sizes also can significantly influence the
+
+
covalency effect (2 to 3 exchange), which is of much in-
terest in the present work. In most instances, these substitu-
tions improve thermal and textural stability while promoting
reducibility (Fe2O3 → Fe3O4) [13,14]. Transition and inner
transition metal substitutents can be expected to strongly mod-
3+
2+
ify the redox properties, that is, the Fe ⇔ Fe charge trans-
2
fer rate.
The present investigation was aimed at developing new high-
temperature catalysts for the WGS reaction that can operate in
extreme conditions as posed in the membrane reactor regime.
In this work, metal (M)-doped ferrite catalysts using either
the first-row transition atoms (M = Cr, Mn, Co, Ni, Cu), the
nontransition atom (M = Zn), or the inner-transition atom
(M = Ce) series were considered. These catalysts were synthe-
*
Corresponding author. Fax: +1 513 556 3473.
E-mail address: panagiotis.smirniotis@uc.edu (P.G. Smirniotis).
0021-9517/$ – see front matter © 2007 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2007.10.018

92
A. Khan et al. / Journal of Catalysis 253 (2008) 91–104
sized and their performance for the WGS reaction evaluated.
Metal dopant additives (M) influence the physicochemical, the
structural properties and catalytic performance. The experi-
2.2. Catalyst characterization
2.2.1. Surface area and pore size distribution analysis
The BET surface areas were obtained by N adsorption on
a Micromeritics Gemini 2360 instrument. Before analysis, the
◦
ments were performed in a temperature range of 350–550 C.
2
−
1
A relatively high space velocity of 60,000 h was maintained
in the WGS reactor. Different ratios of steam/CO (1, 3.5, and 7)
were used to investigate the performance of the catalysts in the
WGS reaction. The shift activity was found to increase with
increasing temperature and also with increasing steam/CO ra-
tios. Interestingly, at high temperatures, the activity approached
the equilibrium conversion value over Fe/Cr and Fe/Ce cat-
alysts. Among the various catalysts studied, Fe/Ce is found
to be an ideal catalyst for operating in steam-rich conditions.
The information obtained from various techniques, including
X-ray diffraction (XRD), temperature-programmed reduction
◦
samples were oven-dried at 120 C for 12 h and flushed with
argon for 2 h. The pore size distribution analyses were con-
ducted by N physisorption at liquid N temperature using a
2
2
Micromeritics ASAP 2010 apparatus. All samples were de-
◦
gassed at 300 C under vacuum before analysis.
2
.2.2. XRD measurements
Powder XRD patterns were recorded on a Phillips X’pert
diffractometer using nickel-filtered CuKα (0.154056 nm) as
the radiation source. The intensity data were collected over a
(
TPR), pore size distribution (PSD), Mössbauer spectroscopy,
◦ ◦
θ range of 3–80 with a step size of 0.02 using a counting
2
and transmission electron microscopy (TEM), is used to fully
characterize the crystalline structure and morphology of the cat-
alysts and their WGS activity. The findings indicate that WGS
activity is determined by the operating conditions used in a
given case.
time of 1 s per point. Crystalline phases were identified through
comparison with the reference data from ICDD files. Lattice pa-
rameter estimations were done by standard indexing methods,
using the intensity of high 2θ peaks (214) and (300) [17–19].
2
.2.3. TPR measurements
Hydrogen TPR of various catalyst samples was performed
2
. Experimental
using an automated catalyst characterization system (Mi-
cromeritics model AutoChem II 2920) with a thermal con-
ductivity detector (TCD), at a heating rate of 5 C/min. The
2
.1. Catalyst preparation
◦
reactive gas composition was H2 (10 vol%) in argon, and the
flow rate was fixed at 10 mL/min (STP). The total reactive
gas consumption during TPR analysis was measured. The TPR
measurements were done after activation once the sample was
Selected metal-doped iron oxide-based catalysts with nom-
inal composition of Fe1.82M0.18O (where M = Cr, Mn, Co,
3
Ni, Cu, Zn, and Ce) were prepared in 10:1 (Fe:M) atomic ra-
tio by the co-precipitation route. Nitrate precursors were used
for all of the preparations. In a typical preparation, calculated
amounts of iron nitrates and the corresponding dopant metal ni-
trate were dissolved separately in deionized water and mixed
together. Dilute aqueous ammonia was gradually added drop-
wise to the aforementioned mixture solutions under vigorous
stirring until precipitation was complete (pH 8.5). The super-
natant liquid was analyzed for nitrate ions by adding about
◦
cooled to 50 C in a helium flow. The sample was then held at
◦
5
0 C under flowing helium to remove the remaining adsorbed
oxygen until the TCD signal returned to the baseline. Subse-
◦
quently, the TPR experiments were performed up to 800 C.
2.2.4. Mössbauer spectral analysis
57Fe Mössbauer spectra were recorded in transmission
geometry using a constant acceleration spectrometer with a
liquid He metal dewar. Experiments were performed either at
1
mL of concentrated sulfuric acid to 10 mL of the super-
2+
natant; the formation of [Fe(NO)] is indicated by a brown
ring [15]. No brown ring formation was observed in any case.
The precipitate gels thus obtained were further aged overnight
◦
57
room temperature or at −195 C. A 20 mCi sample of Co
(Rh) was used as an emitter, and the spectrometer was cali-
brated using a α-Fe foil and taking the isomer shift of Rh at
−0.116 mm/s with respect to α-Fe. Line widths on the inner
two lines of α-Fe were typically 0.21 mm/s. An 80-mg quan-
tity of the oxide catalyst in fine powder form was spread onto
a thin Teflon sheet using GE varnish as a binder for use as an
absorber. A typical run lasted 24 h, and base line counts were
2 million per channel. Samples that did not reveal a magnetic
◦
and filtered off. The resulting cakes were oven-dried at 80 C
◦
for 12 h and finally calcined at 500 C for 3 h in an inert en-
vironment. The heating and cooling rates were maintained at
◦
5
C/min.
The precipitation process is dictated by solution thermody-
namics. The solution reaches saturation state when the disso-
lution rate equals the precipitation rate, and nucleation starts
when the solution concentration exceeds the saturation concen-
tration. The solution concentration affects the crystallite size of
the precipitated catalyst. In the present investigation, the con-
centration of the nitrate solution was optimized so as yield
nanosized grains [16]. It is notable that all catalysts reported
herein were prepared by analogous procedures, which was nec-
essary to allow direct comparison of their catalytic proper-
ties.
◦
◦
Zeeman splitting at 27 C were examined further at −195 C.
2.2.5. TEM
TEM was performed with a Philips CM 20 electron micro-
scope. The accelerating voltage was 200 keV, with an LaB6
emission current and a point-to-point resolution of 0.27 nm.
Samples were sonically dispersed in ethanol and deposited on a
holey carbon Cu grid before examination. Once the solvent was

A. Khan et al. / Journal of Catalysis 253 (2008) 91–104
93
evaporated, the particles remaining on the borders of the holes
in the carbon film were studied.
2
.3. Catalyst activity
The WGS reaction was carried out in a vertical downflow
fixed-bed differential ceramic microreactor (0.635 cm i.d.) at
atmospheric pressure. In a typical experiment, ca. 0.1 g of
sieved catalyst was placed between two quartz wool plugs. The
reactor was placed vertically inside a programmable tubular
furnace (Lindberg), which was heated electrically. Before the
reaction, the catalyst was pretreated in flowing process gas at
◦
4
00 C for 4 h. The catalyst pretreatment involved the partial
reduction of hematite (Fe2O3) to magnetite (Fe3O4) using a
mixture of H2, CO, CO2 (99.9% pure gases), and water vapor
[
20–22], taking care to avoid overreduction of the magnetite
active phase to lower carbides, oxides, or metallic iron phases.
Metallic iron phases are active catalysts for methanation and
Fig. 1. X-ray powder diffraction patterns of various Fe
Mn, Co, Ni, Cu, Zn, and Ce) catalysts.
M O
1.82 0.18 3
(M = Cr,
(
Fischer–Tropsch processes [20].) The heating and cooling rate
was maintained at 5 C/min. The gas flows were regulated
◦
the addition of different transition/inner-transition metals. The
efficacy of the (dopant ions) foreign cations in modifying the
resultant specific surface area can be directly related to varia-
tions in the rate of crystal growth. The nature and concentration
of the foreign cations present in the system govern this varia-
tion. Interestingly, a narrow pore size distribution (mesoporous
regime) was observed in all of the investigated catalyst systems
in the range of 49–250 Å (Table 1).
through precalibrated mass flow controllers with a digital read-
out unit (MKS Instruments). Water was injected into the system
through a motorized syringe pump (Cole-Parmer type 74900)
◦
to generate steam. The entire system was maintained at 200 C
using heating tapes. Before pretreatment, the reactor setup was
flushed with an inert gas. The pretreatment gas mixture was ini-
◦
tialized only after the catalytic system had reached ꢀ150 C.
◦
The experiments were performed at 350–550 C using a des-
3
.2. XRD studies
ignated amount of steam and CO. The product stream coming
from the reactor was passed through ice cooled trap to condense
water, after which, the product gases were analyzed with an on-
line TCD (Gow Mac series 550) with a Porapak Q column to
separate the gases. The TCD was interfaced to a personal com-
puter using a peak simple chromatography data system. The
results were analyzed using the Peak Simple 2.31 program. The
product gas was injected through a six-port valve, with sam-
pling performed at 20-min intervals. Reported values of con-
versions correspond to steady-state values (0.5 h). A gas hourly
To ascertain the composition and phase purity, the catalysts
were examined by XRD. The powder XRD patterns of vari-
ous Fe1.82M0.18O3 (M = Cr, Mn, Fe, Co, Ni, Cu, Zn, and Ce)
catalysts investigated in the present study are shown in Fig. 1.
Interestingly, all of the modified ferrite samples (except Fe/Mn
and Fe/Ce) exhibit intense and similar XRD patterns, with
well-resolved peaks clearly indicating the polycrystalline and
monophasic nature of the samples. The existence of a Fe2O3-
type phase (PDF-ICDD 33-0664) was identified by comparing
the patterns with standard reference data from the PDF data-
base. The observation of a hematite phase in all of the cata-
lysts is of significance, because during catalyst activation, the
hematite phase is converted to the magnetite phase (Fe3O4), the
active phase for the WGS reaction. In addition, the XRD spec-
tra for the modified ferrite samples show no extra crystalline
phases due to individual oxides of the substitutent cations, such
as MOx (M = Cr, Mn, Co, Ni, Cu, Zn, and Ce). Any WGS cat-
alyst based on iron, has to undergo the required transformation
−
1
space velocity of 60,000 h was maintained throughout the
experiments. To investigate the influence of the steam/dry gas
ratio on the progress of the WGS reaction, different steam/dry
gas ratios (1, 3.5, and 7) were used.
3
. Results and discussion
3
.1. BET surface area and pore size distribution
measurements
(
reduction) from hematite to magnetite, where the active cata-
Table 1 gives various physical properties (BET surface area,
average pore size, crystallite size and lattice variables) of the
metal-doped iron oxide-based catalysts Fe1.82M0.18O3 (M =
Cr, Mn, Co, Ni, Cu, Zn, and Ce) prepared in this work. It is
evident from the table that all of the iron oxide-based catalysts
prepared in the present investigation exhibited remarkably high
surface areas. The surface area measurements clearly reveal that
the surface area of iron-based mixed oxides can be altered by
lyst comprises a crystalline inverse spinel structure in which the
octahedral lattice sites are occupied equally by Fe and Fe
2+
3+
ions. The X-ray diffraction experiments performed on activated
catalysts, confirmed the existence of Fe O phase (PDF-ICDD
3
4
26-1136) (supporting information). Magnetite (Fe O ) crystal-
3
4
lizes in a cubic inverse spinel structure (AB O type) at RT,
2
4
where 1/3 of the Fe ions occupies tetrahedrally coordinated A
sites as Fe3 and remaining 2/3 of the Fe ions with equal num-
+

94
A. Khan et al. / Journal of Catalysis 253 (2008) 91–104
Table 1
BET surface area, XRD phase, cell parameter, and cell volume measurements
Metal doped
iron oxide based
Physisorption measurements
XRD
phase
XRD analyses
Cell parameter
TEM
grain size
(nm)
a
BET SA
Average pore
diameter (Å)
Axial
ratio c/a
Cell volume
catalysts Fe
MxO
3
2
−x
2
b
(
m /g)
b
b
(Å)
a (Å)
c (Å)
Fe
Fe
Fe O
Cr
O
3
81.7
77
119
250
104
49
120
116
77
Fe O
5.094
nd
13.834
nd
2.716
nd
310.9
nd
∼10
∼8
1
.82 0.18
Mn
2
3
3
3
3
3
3
3
3
c
O
175.9
36.3
41.0
80.1
75.3
46.1
95.9
Fe O
1.82
0.18
3
2
Fe O
2
5.093
5.091
5.097
5.106
5.114
nd
13.912
13.858
14.097
14.109
14.117
nd
2.731
2.722
2.765
2.763
2.760
nd
316.0
311.1
317.2
318.9
319.7
nd
∼15
∼20
5–8
2
3
Fe
Co
O
Fe O
1
1
1
1
1
.82 0.18
3
2
Fe
Fe
Fe
Fe
Ni
O
Fe O
2
.82 0.18
3
Cu
O
Fe O
∼10
5–10
5–8
.82 0.18
3
3
3
2
Zn
O
O
Fe O
2
Fe O
2
.82 0.18
Ce
.82 0.18
a
b
c
Fe O (hematite) PDF-ICDD 33-0664.
Calculated following standard indexation procedures (Refs. [17–19]).
nd: not determined.
2
3
Cr3 and Co (Cr 0.755 Å; Co 0.75 Å) possess larger
+
3+
3+
3+
3
+
ionic radii than Fe thereby lead to lattice (cell volume) ex-
pansion. On the contrary, when Cr and Co replace Fe
3+
3+
3+
ions from their octahedral positions, the distance between is
reduced thereby causing lattice (cell volume) contraction. Mag-
netite is a spinel and possess both tetrahedral and octahedral
iron sites. Hematite, on the other hand possesses only octahe-
dral iron sites. With trivalent iron there are five ‘d’ electrons in
the ‘3d’ orbitals. In the high spin state, each of these would be
unpaired and occupy all of the t2g and e orbitals in an octa-
g
hedrally coordinated environment or all of the t2 and e sites in
a tetrahedrally coordinated environment. Since there are no de-
generacies, the C.F.S.E. for each of these coordinations is zero.
Thus there is no preference for either coordination. The alter-
n+
ation in the lattice parameter upon M (substitutent cation)
n+
substitution suggests the incorporation of M into the spinel
Fig. 2. Relationship between cell volume and ionic radii of substitutent ions in
various Fe
M
1.82 0.18
O (M = Cr, Co, Ni, Cu, and Zn) catalysts.
3
structure.
bers of Fe2 and Fe are located on octahedrally coordinated
+
3+
3
.3. Catalyst activation
B sites. It is well established fact that rapid electron hoping be-
tween Fe2 ⇔ Fe in the Fe3O4 lattice system is needed for
+
3+
The fresh catalysts in the hematite form (α-Fe O ), have to
2
3
be transformed into magnetite form (Fe3O4). This is achieved
by controlled reduction of α-Fe2O3 into Fe3O4 in a flow of
process gas mixture (CO, CO2, H2, and steam-H2O). The rep-
resentative reactions are shown as follows:
carrying out the water gas shift reaction [23]. There was no ev-
idence of the formation of Fe (metallic) or Fe2C (iron carbide)
phases due to possible over-reduction in the present investiga-
tion.
Using the interplanar spacing values, calculation of the cell
parameters was carried out (Table 1), which revealed that all the
substitutent ions enter the hematite lattice under the preparation
conditions employed in this work. The relationship between the
ionic radii of substitutent ion and the resultant cell volume is
3
Fe2O3 + H2 → 2Fe3O4 + H2O, ꢀH = −16.3 kJ/mol,
Fe2O3 + CO → 2Fe3O4 + CO2, ꢀH = +24.8 kJ/mol.
3
The reduction is typically performed during the HTS reactor
startup, and it should be carefully controlled because of a sig-
nificant heat release resulting from the exothermic nature of the
reactions involved, which may damage the catalyst [24]. The
over-reduction should be avoided, as it leads to the formation of
FeO, Fe2C and metallic iron phases, which catalyze undesired
side reactions (i.e., methanation and CO disproportionation).
The ratio of oxidant to reductant (also called “reduction fac-
tor,” R) in the process gas mixture has to be just optimal, in
order to facilitate the reduction of hematitic to magnetitic phase
and should not catalyze further reduction. In the present study
R = 1.4 was utilized; where R = [(CO) + (H2)]/[(CO2) +
3+
presented in Fig. 2. The substitution of Fe (r = 0.785 Å) with
n+
M
(M = Cr, Mn, Co, Ni, Cu, Zn, and Ce) leads to a modifi-
cation of the distances between atomic planes, as well as of the
lattice constant as shown in Fig. 2 and Table 1. Experimental
values of lattice constant were calculated from higher 2θ lines
i.e., from (214) and (300) planes for each sample [19]. By com-
paring the values of the modified ferrite samples with that of
the pure Fe O sample, prepared by an identical method, it is
2
3
observed that the lattice expansion is taking place, except in the
case of Fe/Cr and Fe/Co systems. All the foreign cations [ionic
2+
2+
2+
radii (rion) Ni 0.83 Å; Cu 0.87 Å; Zn 0.88 Å] except
(H O)].
2

A. Khan et al. / Journal of Catalysis 253 (2008) 91–104
95
◦
Fig. 3. The WGS activity results of various Fe
M O
.73 0.273 4
(M = Cr, Mn, Co, Ni, Cu, Zn, and Ce) activated catalysts in the temperature range 350–550 C;
2
−
1
steam/CO = 1, 3.5, and 7.0; gas hourly space velocity (GHSV) = 60,000 h ; (A) shows the variation of the thermodynamic equilibrium conversion of CO (%) as
a function of steam-to-CO (S/CO) ratio in the feed at different temperatures; (B) represents WGS activity over pure Fe O (undoped) sample.
3
4
3
.4. WGS activity
increase in steam-to-CO (H2O/CO) ratio. No pressure drop
across the catalyst bed was observed during the experiments
nor was CH4 formed in the effluent stream (except in the case
of Fe/Mn). The results of the experiments carried out at dif-
ferent temperatures and different H2O/CO ratios, over various
The WGS reaction is a moderately exothermic reaction.
As shown in Fig. 3A, the equilibrium conversion of CO de-
creases as the reaction temperature increases and increases with

96
A. Khan et al. / Journal of Catalysis 253 (2008) 91–104
modified ferrite-based shift are shown in Figs. 3C–3I. For the
sake of better comparison WGS activity results over undoped
Fe3O4 activated catalyst are presented in Fig. 3B. The activ-
ity increased with the temperature, due to kinetic factors [25]
and this also shows that the catalysts are stable in these condi-
tions. It is well known that under low steam to CO ratio or high
temperatures metallic iron or a carbide phase may be formed.
These phases can act as Fischer–Tropsch (FT) catalysts. How-
ever, except for the presence of Fe O phase, no other XRD
the reduction profile compared with that of the pristine Fe2O3
sample [31].
The TPR profiles of the various metal-doped iron oxide-
based catalysts are presented in Figs. 4a and 4b, and the cor-
responding Tmax values for each reduction step of iron oxide
are given in Table 2. The TPR experiments were repeated three
times on of all these samples to check for reproducibility; simi-
lar line shapes were observed for all of the catalyst samples. For
reference purposes, the TPR profile of pristine Fe2O3 (free from
any substitutent ion) prepared by a similar procedure is shown
in Fig. 4A. In the TPR of pristine hematite (Fe2O3), the first re-
3
4
phases could be detected in the present study (except in the case
◦
of Fe/Mn). The lower WGS activity observed at 350 C, can be
◦
related to the difficulty in sustaining the active magnetite phase,
as observed from TPR experiments (presented later). It is also
suggested that at lower temperatures the activity of magnetite
is limited by the dissociation of steam [26]. The WGS reaction
was performed at different steam-to-CO ratios (S/CO = 1, 3.5,
and 7.0) to mimic steam deficient to steam rich conditions re-
spectively, and to study the influence of steam on the eventual
activity performance of our catalysts. A relatively high space
duction peak appeared at 348 C, corresponding to the reductive
◦
transition of Fe2O3 to Fe3O4. The peak at 621 C corresponds
to the transformation of Fe3O4 to FeO. Within the experimen-
tal conditions of the present study, formation of metallic iron
species was either delayed (prolonged) or not observed, which
can be attributed to the preparation method used in this investi-
gation.
From Figs. 4B–4H and Table 2, it is obvious that each
promoter (M = Cr, Mn, Co, Ni, Cu, Zn, and Ce) influenced
the reduction profile of iron oxide in a unique manner. In the
−1
velocity of 60,000 h was maintained in all of the experiments.
The WGS reaction was performed in the temperature range of
◦
case of Fe/Cr catalyst, the first reduction peak at 225 C corre-
◦
3
50–550 C. In general, WGS activity increased with increas-
sponded to the reduction of Cr6 → Cr (Fig. 4B); interest-
+
3+
ing temperature. In addition, the WGS activity increased with
increasing steam-to-CO ratio. The activity approached the equi-
librium conversion at higher temperatures over Fe/Cr and Fe/Ce
catalysts. On the whole, it is observed that, among the various
modified ferrites investigated in this study, Fe/Ce is a promising
system for performing the WGS reaction under steam-rich and
high-temperature conditions. In a membrane reactor operation,
the catalyst is exposed to a steam-rich environment and high
temperatures (severe operation conditions). All of the present
results were successfully reproduced twice within the limits of
permissible error.
3+
2+
ingly, further partial reduction of Cr → Cr , which would
◦
be expected at 490 C [32], was not observed. Reduction of
◦
Fe2O3 → Fe3O4 was observed at Tmax 350 C, whereas fur-
ther reduction to FeO occurred at higher temperatures. Adding
chromium to Fe2O3 did not improve the reducibility of hematite
to magnetite. In the Fe/Mn catalyst, the first reduction peak
◦
4+
3+
at 200 C corresponded to the reduction of Mn → Mn
,
3+
2+
◦
whereas the reduction of Mn → Mn occurred at 528 C
(
Fig. 4C). In particular, the Fe/Mn system presents a typical ex-
ample of the cooperative synergistic influence of Mn and Fe on
each other’s reduction profile. An important feature observed in
the TPR profile of Fe/Mn is a shift of the iron oxide TPR peaks
3
.5. TPR
Iron can exist in the form of three oxides: hematite (Fe O ),
◦
to lower temperatures (by ∼50 C) as manganese is introduced
into the hematite host lattice. Thus, this shift indicates that the
3+
2
3
presence of Mn facilitates the reduction of Fe species (Fe O
2
3
magnetite (Fe3O4), and wustite (FeO). The latter is unstable
◦
to Fe3O4). The 613 C Tmax peak corresponds to the reduction
of magnetite to wustite, beyond which total reduction to metal-
lic iron would be expected. The reduction profile for pristine
◦
below 570 C, where it decomposes to α-Fe and Fe3O4. At
◦
temperatures below 570 C, the reduction of Fe2O3 to Fe metal
◦
proceeds in two steps via an Fe3O4 intermediate [27]. Fe2O3 →
Fe3O4 reduction is exothermic, whereas reduction to the metal
is endothermic [28]. In reduction of either hematite or mag-
Co3O4 comprises a low-temperature peak below ∼300 C [33]
◦
or ∼330 C [34] and a high-temperature peak between 350 and
◦
6
00 C [34] or, as reported by Brown et al. [33], between 300
◦
◦
netite at temperatures above 600 C, the formation of a stable
and 700 C, corresponding to the following reductive processes:
wustite phase also must be considered. In TPR of pure hematite
Co O → CoO → Co. The addition of cobalt to iron oxide low-
3
4
◦
(
Fe2O3), the first reduction peak appeared at 302 C, corre-
ered the temperature of the maximum rate of hydrogen uptake
◦
◦
sponding to the transition of Fe2O3 to Fe3O4. The second peak
[for Fe O → Fe O ] by 20 C (from 350 to 330 C), demon-
2
3
3
4
◦
at 354 C was attributed to the transformation of Fe3O4 to FeO.
strating that the presence of cobalt makes the hematite easier to
reduce (Fig. 4D). The observed TPR pattern matches well with
the literature reports on iron-cobalt catalysts [33]. In the TPR
profile of Fe/Ni (Fig. 4E), the reduction of hematite to mag-
◦
The third and final peak at 475 C corresponds to the transi-
tion of FeO to Fe [29]. In another report, the TPR curves of
◦
hematite exhibited a peak at 510 C, attributed to magnetite for-
◦
◦ 2+
netite occurred at 327 C, followed by the reduction of Ni to
mation, and another peak at around 770 C, attributed to the
0
◦
formation of metallic iron [30]. The position of the temperature
maxima may vary from sample to sample depending on the par-
ticle size and other parameters, such as temperature ramp rate.
The addition of a substitutent/dopant ion significantly modifies
Ni at ∼510 C. Finally, the reductive transition of magnetite
◦
to wustite occurred at 590 C. The formation of metallic iron
◦
could not be observed even up to 800 C. In the Fe/Cu catalyst
(Fig. 4F), the first peak at 143 C corresponds to the reduction
◦

A. Khan et al. / Journal of Catalysis 253 (2008) 91–104
97
Fig. 4. (a) TPR profiles of titled Fe
M
O (M = Cr, Mn, and Co) catalysts.
1
.82 0.18 3
of Cu2 → Cu and the subsequent peak at 162 C corresponds
+
+
◦
[34]. Promoting iron oxide with cerium causes a shift in the
reduction peaks of both hematite-to-magnetite and magnetite-
to-wustite to lower temperatures (Table 2). In the Fe/Ce system,
+
0
to the reduction of Cu → Cu [35,36]. An interesting observa-
tion from the present study is that the addition of Cu to Fe2O3,
decreased the reduction temperature of hematite to magnetite
considerably, to 190 C, compared with 348 C for the pristine
hematite sample. Interestingly, the incorporation of Cu did not
affect the stability of the Fe O phase, which was found to be
stable up to 600 C, after which transformation of the magnetite
to wustite was observed. The TPR profile of the Fe/Zn catalyst
Fig. 4G) was very similar to that of the pristine-Fe O sample.
The peak at 333 C can be attributed to the formation of mag-
netite phase, whereas reduction of Zn to Zn was observed
at about 588 C, beyond which formation of wustite phase was
◦
the ceria surface shell reduction occurs at 380 C, instead of
◦
◦
◦
485 C as in the pristine ceria sample; however, the ceria bulk
reduction was not affected by the presence of iron [34]. Ac-
cording to Giordano et al. [37], the TPR profile for ceria is
not controlled by the rate of diffusion of the oxygen vacancies;
instead, surface reduction process and the difference of both
thermodynamic and kinetic properties in the ceria microcrys-
tals (as a function of their size) are critical factors. The WGS
3
4
◦
(
2 3
◦
3+
2+
0
reaction is a redox process wherein facile reducibility of Fe
species enhances the WGS activity.
◦
noted. The enhanced reducibility observed in the Fe/Ce catalyst
(Fig. 4H) may be related to the close interaction between Fe
3.6. Mössbauer spectroscopy studies
and Ce cations occurring during the co-precipitation process.
57Fe Mössbauer spectroscopy provides an elegant local
probe of Fe cations in metal-doped ferrites. These results are
Pristine CeO exhibited two characteristic reduction regimes:
2
◦
◦
surface shell reduction (485 C) and bulk reduction (850 C)

98
A. Khan et al. / Journal of Catalysis 253 (2008) 91–104
Fig. 4. (b) TPR profiles of titled Fe
M
1.82 0.18
O (M = Ni, Cu, Zn, and Ce) catalysts.
3
Table 2
the crystal field stabilization energy (CFSE) of metallic ions
with different electronic configurations, polarization, and re-
pulsion effects, apparently affect the total energy and preferred
symmetry of the resulting compounds. Thus, doping effects
can lead to the crystallization of an ideal inverse spinel or
a normal or mixed spinel structure. Substitution of iron sites
with other substitutent cations apparently modifies the redox
ability of ferrites, and also influences their chemical stabil-
ity. In magnetite-type structures, octahedral sites are occupied
◦
Tmax ( C) values pertaining to each reduction step of iron oxide
◦
Metal-promoted
Tmax ( C)
Fe
MxO catalysts
3
2−x
Fe O → Fe O
Fe O → FeO
FeO → Fe
2
3
3
4
3
4
Fe
Fe
Cr
O
3
O
0.18 3
350
300
348
330
327
190
333
319
302
510
∼800
613
621
623
590
650
∼700
578
354
–
>800
>800
>800
>800
>800
>800
>800
>800
475
1
.82 0.18
Mn
1.82
Fe O
2
3
Fe
Fe
Fe
Fe
Fe
Fe O
Fe O
2
Co
O
O
3
1
1
1
1
1
.82 0.18 3
Ni
.82 0.18
Cu
O
O
O
.82 0.18
3
+
+
Zn
by 2 and 3 ions, whereas tetrahedral sites are occupied by
.82 0.18
3
3
Ce
+
.82 0.18
3
ions only. Reithwisch and Dumesic [38] studied a number
a
2
3
3
b
of spinel-based materials and concluded that only inverse and
mixed spinel structures readily undergo rapid electron exchange
770
a
Ref. [29].
Ref. [30].
+ +
between the 2 and 3 states, thereby catalyzing the WGS re-
b
+
+
action [38]. The significance of the 2 ⇔ 3 redox couple also
was demonstrated by Boreskov [39]. Encouraging tetrahedral
sites to expand and octahedral sites to contract results in im-
proved covalency of the system and promotes electron-hopping
important, because small variations in the substitution of Fe(II)
or Fe(III) leads to major changes in the chemical properties
and functionality of these materials. Some parameters, such as

A. Khan et al. / Journal of Catalysis 253 (2008) 91–104
99
57
Fig. 5. (a) Fe Mössbauer spectra of titled Fe
M
1.82 0.18
O (M = Cr, Mn, and Co) catalysts.
3
between these sites (Fe2 ⇔ Fe ). Mössbauer spectroscopy is
an ideal probe for gaining insight into the structure/coordination
and valence state [40]. We performed Mössbauer spectroscopy
measurements to elucidate the nature of the Fe sites formed and
their dynamics.
+
3+
previous work [42,43] are also given in Table 3. Bulk crystalline
hematite is characterized by an internal magnetic hyperfine
field of 51.8 T at room temperature. Nanocrystalline hematite,
on the other hand, has a smaller quadrupole splitting Δ and
a room temperature magnetic hyperfine field of about 51 T.
Deviations from ideal chemistry and crystallinity usually re-
late to particle size effects and/or substitution effects [42]. In
the present investigation, Mössbauer spectra were recorded at
room temperature for most of the samples. Enhanced contri-
bution due to super-paramagnetic nanoparticles was observed
in room temperature Mössbauer spectra of Fe/Ni, Fe/Zn, and
Fe/Ce samples. Therefore, Mössbauer spectra of these sam-
Hematite crystallizes in the Al O (corundum) structure
2
3
3+
with a closely packed oxygen lattice, with Fe cations occu-
pying octahedral sites. It exhibits a complex magnetic behav-
ior, being antiferromagnetic at low temperatures (T < 80 K)
and undergoing a Morin transition at TM > 260 K involving
a spin flip, leading to a weak ferrimagnetic state, before be-
coming paramagnetic (Curie temperature) at high temperatures
∼
(
T = 955 K). Its structure can be visualized as being composed
ples also were recorded at liquid N temperature (78 K). At
C
2
of Fe–O3–Fe units (triplets) of closely packed oxygen atoms
with Fe(III) on either side. The Fe(III) atoms in each of these
Fe–O3–Fe units have opposite spins, being antiferromagneti-
cally coupled as a result of superexchange interaction through
the triad of oxygen atoms [41]. Mössbauer spectra of selected
samples investigated in the present study are shown in Fig. 5,
and the corresponding hyperfine structure parameters obtained
by fitting the line shape are summarized in Table 3. For com-
parison, relevant hyperfine structure parameters obtained from
78 K, the recoil-free fraction increased, and the size of the
effect increased. More significantly, as the sample is cooled,
spin-lattice relaxation slowed, and a magnetic hyperfine struc-
ture developed, facilitating identification of various iron phases
in the samples [41]. This finding is in agreement with the report
of Janot et al. [44], who confirmed the variations in magnetic
properties for natural hematites of small particle size. In gen-
eral, samples containing particles smaller than 8 nm showed a
super-paramagnetic doublet at room temperature. Interestingly,

100
A. Khan et al. / Journal of Catalysis 253 (2008) 91–104
57
Fig. 5. (b) Fe Mössbauer spectra of titled Fe
M
1.82 0.18
O (M = Ni, Cu, Zn, and Ce) catalysts.
3
samples of Fe/Ni, Fe/Zn, and Fe/Ce ferrites were composed of
to 8 nm particles, as confirmed by TEM (presented below).
Table 3
5
a
Magnetic field (H), quadrupole splitting (Δ) and isomer shift (δ ) measure-
ments
57
Fe Mössbauer spectra of pristine Fe2O3 sample (Fig. 5C)
exhibits a six-line Zeeman spectrum at room temperature. The
value of the isomer shift of Fe2O3 is typical of a trivalent iron
a
Metal doped
iron oxides
Fe2−xMxO3
Cryst.
size
(nm)
Tempe-
rature
H
Δ
δ
(T)
± 0.2
(mm/s)
± 0.01
−0.109
−0.112
−0.107
0.056
(mm/s)
± 0.01
b
(
(
Table 3). Deviations in isomer shift (δ), quadrupole splitting
Δ), and magnetic hyperfine field (Hint) parameters between
Fe
Cr
1.82 0.18
O
12.5
nd
300 K
300 K
300 K
78 K
50.0
49.7
50.4
53.1
49.5
52.7
52.3
50.1
51.6
51.75
–
0.365
0.365
0.345
0.456
0.374
0.363
0.377
0.354
0.387
0.37
3
O
the present nanocrystalline α-Fe2O3 and reported bulk α-Fe2O3
could be simply a grain size effect. To further study the as-
sociation between grain size and the hyperfine magnetic para-
Fe
Mn
1.82
0.18 3
Fe1.82Co0.18O3
16.3
11.8
10.7
10.1
7.1
Fe
Fe
Ni
O
1.82 0.18 3
Cu O
1.82 0.18 3
300 K
78 K
−0.095
−0.080
−0.095
−0.087
−0.077
−0.197
−0.22
meters, a sample of pristine α-Fe2O3 was further calcined at
◦
Fe1.82Zn0.18O3
7
00 C for 12 h in flowing N2 gas and analyzed by Mössbauer
Fe
Ce
O
78 K
1
.82 0.18
3
spectroscopy. On calcination of nanocrystalline α-Fe2O3, the
magnetic field strength (H) was found to increase from 50.1 to
c
Fe O
2
Fe O
2
Fe O
2
Fe O
2
12.8
28
300 K
300 K
295 K
80 K
3
3
3
3
c,d
e
5
1.6 T (Table 3); this increase appears to be related to particle
Bulk
Bulk
size. Changes in isomer shift and quadrupole splitting parame-
ters with increasing particle size were also seen (Table 3). These
findings support the idea that particle size through local fields
influences the hyperfine structure. Doping metal cations into
hematite (α-Fe2O3) modifies the magnetic hyperfine field and
also influences other properties, such as reducibility and cova-
lency of hematite particles. These effects appear to depend on
f
–
a
Relative to α-Fe.
b
c
d
e
f
Calculated from XRD using Scherrer equation.
Present study.
Calcined at 700 C for 12 h in N flow.
Ref. [42].
Ref. [43].
◦
2

A. Khan et al. / Journal of Catalysis 253 (2008) 91–104
101
the nature of the dopant metal cation. In general, the radii of
the Mössbauer spectroscopy through hyperfine magnetic inter-
action parameters is a rather sensitive probe of the size, crystal-
lite distortion, and substructure of metal-doped α-Fe2O3 par-
ticles.
n+
3+
substitutent ions (M ) differ from those of Fe . Such differ-
ences produce distortions in the crystal structure. A size mis-
match between the Fe3 and substitutent (M ) cations leads
to an electric field gradient (EFG) that measures the aspheric-
ity or distortion of the FeO6 octahedra. Mössbauer spectra of
our metal-doped iron oxide samples were analyzed as a super-
position of a Zeeman sextet and a doublet. The former comes
from the large grain-sized ferromagnetic particles, whereas the
later feature is due to super-paramagnetic small particles. The
physics underlying these observations can be summarized as
follows.
+
n+
3.7. TEM
To study surface morphology and explore the structural fea-
tures at the grain level, TEM studies were performed on various
metal-doped iron oxide-based Fe1.82M0.18O3 (M = Cr, Mn, Co,
Ni, Cu, Zn, and Ce) catalysts. The corresponding TEM im-
ages thus obtained are presented in Figs. 6A–6H. In general,
the existence of grains with an associated size of 5–20 nm was
observed (Table 1). To gain insight into their crystal structure,
digital diffraction patterns (DDPs) acquired from the nanopar-
ticles assemblies of each sample are presented as insets in the
respective TEM images. The different rings and spots account
for the existence of periodic contrasts in the original experimen-
tal micrographs, which correspond to different sets of atomic
planes of the crystalline structure. The geometric arrangements
of these reflections are directly related to the structural aspects
of the analyzed crystals. Because of this, the DDPs are very
useful in phase recognition [47]. Table 4 displays the measured
lattice spacing based on the rings in the diffraction pattern and
compares them with the known lattice spacing for bulk Fe2O3
along with their respective hkl indexes from the PDF database.
The position and relative intensity of all of the diffraction rings
closely match the standard Fe2O3 powder diffraction data. All
of the catalysts prepared in this investigation exhibited nanosize
grains. The high surface area-to-volume ratio in nanoparticles
leads to modification in the surface structure and hence to en-
hanced surface features. Nanocrystalline particles are known to
be more reactive than larger particles due to the higher relative
concentration of crystallographic areas where atoms exist with
low coordination numbers. On the whole, all of the samples in-
vestigated exhibited uniform particle size distribution; however,
a significant number of crystals displayed a pattern of paral-
lel high-contrast light and dark bands extending across their
surfaces. As reported previously, it may be a diffraction phe-
nomenon caused by the presence of a surface layer of different
phase or crystallographic orientation from that of the crystal un-
derneath [47]. Substitution of iron by larger substitutent metal
ions in the hematitic crystal lattice generates considerable stress
in the system. To overcome this stress, the equilibrium parti-
cle size is shifted toward smaller particles, because the ratio of
strain to surface effects becomes greater for larger particles.
Difference in line shape between samples with large and
small particles come largely because the net magnetic field is
finite in the former but vanishes in the later. In microcrystalline
samples, the magneto-anisotropic energy may be of the same
order of magnitude as the thermal energy (kT ), where k rep-
resents the Boltzmann constant and T represents the absolute
temperature. The magneto-anisotropic energy represents the en-
ergy required to rotate the magnetization direction and is found
to be proportional to the particle volume (V ) and magneto-
crystalline anisotropy constant (K). In a simplistic form, the
spin-lattice relaxation time, τ, can be written [45] as
τ = τ0 exp(KV/kT ).
The spin-lattice relaxation time τ is short for small-sized (V <)
particles, whereas the reverse is the case for large-sized (V >)
particles. For an 57Fe nucleus to see a magnetic field Hint, the
nuclear magnetic moment must precess about the local mag-
netic field for at least one cycle, also called the Larmor preces-
sion time, τl. In general, a super-paramagnetic doublet will be
observed when τ < τl, a circumstance in which the magnetic
field averages out to zero, because atomic spins (producing
the field) fluctuate at a rate exceeding the time (τl) needed to
sense that field. τl may be visualized as the interaction time,
that is, the time needed for a nucleus to sense the local field.
At the other extreme, a well-resolved six-line magnetic hy-
perfine structure will be observed when τ > τl, that is, when
fluctuations in magnetization are much slower than the Lar-
mor precession time. In these Mossbauer effect experiments,
−
8
−9
the characteristic time, τl, is on the order of 10 –10 s, which
fixes a threshold particle size of about 10 nm to observe a mag-
netic hyperfine structure at room temperature. Cooling a small
grain-sized sample from room temperature to 78 K slows the
spin-lattice relaxation time, leading to a six-line Zeeman spec-
trum [46].
Spatial variations in interaction fields can lead to broaden-
ing of lines. Generally, the incorporation of metal cations with
charges other than 3 in α-Fe2O3 will change the electrical
3.8. Structure–WGS activity correlated
+
Our findings demonstrate that simultaneous precipitation of
n+
3+
neutrality of the crystal lattice, and countercharges will be re-
substitutent metal cation (M ) with Fe leads to the forma-
quired to restore electrical neutrality. In the case of formation of
tion of hematite-like Fe1.82M0.18O phase, which on activa-
tion transforms into either an inverse or mixed Fe2.73M0.27O4
3
n+
solid solutions of M (n
ꢀ
balance is most likely achieved by the creation/formation of
oxygen vacancies. The variation in the quadrupole-splitting val-
ues observed in this study (Table 3) provides a direct measure
of the distortion caused by oxygen vacancies. In conclusion,
spinel. The overall WGS activity depends on various parame-
ters, including operation conditions, the nature of the substi-
III
II
tutent ion incorporated, and the covalency of the Fe ↔ Fe
redox couple in the resultant catalyst. Topsoe and Boudart [48]

102
A. Khan et al. / Journal of Catalysis 253 (2008) 91–104
Fig. 6. TEM images of various Fe
M O
.82 0.18 3
(M = Cr, Mn, Co, Ni, Cu, Zn, and Ce) catalysts. The insets in the respective images are digital diffraction patterns
1
(DDPs) obtained from the corresponding nanoparticles assemblies.

A. Khan et al. / Journal of Catalysis 253 (2008) 91–104
103
Table 4
Measured lattice spacing, d (Å), based on the rings in the insets of Figs. 6A–6H and standard atomic spacing for Fe O along with their respective hkl indexes from
2
3
the PDF database
Ring/d (Å)
1
2
3
4
5
6
7
8
Sample
Fe
Fe
Fe O
Fe
Fe
Fe
Fe
Fe
Cr O
.82 0.18 3
3.641
–
3.932
3.389
3.932
–
2.731
–
2.457
–
2.34
2.521
2.397
–
2.184
–
–
2.184
2.04
2.286
2.048
–
1.787
–
1.638
–
–
1.666
1.585
1.638
1.611
–
–
–
–
–
–
–
–
–
1.404
–
–
–
1
1
Mn O
0.18 3
.82
2.73
2.808
2.89
2.808
2.731
2.731
1.966
1.966
1.854
1.787
1.855
1.724
2
3
Co O
1
1
1
1
1
.82 0.18 3
Ni
.82 0.18
O
–
3
Cu
O
O
O
1.404
–
1.445
.82 0.18
3
3
3
Zn
3.933
–
2.397
–
.82 0.18
Ce
.82 0.18
a
Fe O
2
hkl
3.684
012
30
2.7
104
100
2.519
110
70
2.2070
113
20
1.8406
024
40
1.6941
116
45
1.4859
214
30
1.4538
300
30
3
a
a
Int-f
a
PDF-ICDD: 033-0664.
found that expansion of tetrahedral sites and contraction of oc-
tahedral sites leads to increased covalency in the system and
improved electron-hopping capabilities between FeIII ↔ Fe
octahedral sites. Moreover, improving the covalency of the cat-
alysts improves the WGS activity [39]. Boreskov has demon-
shifts, δ. The synergistic effect can be attributed primarily to
the formation of either inverse or mixed spinels of composition
II
3+
A(
B [A B
]O , which have a highly facile Fe
⇔
1−δ)
δ
δ
(2−δ)
4
2
+
Fe redox couple. The resultant WGS activity can be related
3+
2+
to the degree of covalency (Fe ⇔ Fe redox couple), which
in turn depends on the operation conditions and catalyst formu-
lation.
strated that the octahedral Fe2 and Fe ions located in the
magnetite-based structure function as a redox couple, and that
magnetite-based catalysts can be highly effective for the com-
plete dissociation of water into hydrogen and adsorbed oxygen
under reaction conditions. Water dissociation causes the oxida-
+
3+
4
. Conclusion
tion of Fe2 centers to Fe and liberates hydrogen. The oxi-
+
3+
Simultaneous precipitation of Fe(III) nitrates along with sub-
dized iron centers may subsequently be reduced by CO, thereby
stitutent metal nitrates (of Cr, Mn, Co, Ni, Cu, Zn, or Ce)
results in the formation of high-surface area nanocrystalline
catalysts. XRD data analyses of activated catalysts have estab-
producing CO to complete the reaction loop [39]. Cr O is
2
2
3
a well-known refractory oxide. The presence of chromium re-
stricts the rapid thermal sintering of Fe O , thereby enhanc-
3
4
lished the existence of Fe2.73M0.27O (M = Cr, Mn, Co, Ni,
4
ing the WGS activity of the Fe/Cr catalyst. It is proposed that
Cu, Zn, and Ce)-type spinels. The synergistic effect between
iron and the other metal substitutent leads to the formation
3
+
iron-chromia forms an inverse spinel structure and that Cr
2+
3+
replaces equal amounts of Fe and Fe from the octahedral
of either inverse or mixed spinel of composition A
Bδ-
sites, with the displaced Fe2+ consequently located in tetrahe-
(1−δ)
[AδB(
]O4. Incorporation of metal cations into the hematite
2−δ)
dral sites [49]. In the case of our Fe/Mn catalysts, formation of
FT product (methane) was observed at the expense of the H2
produced by the WGS reaction. The addition of Cu to iron ox-
(
α-Fe2O3) crystal structure modifies the magnetic field and
also influences the reducibility of hematite particles, as was
observed in Mössbauer spectroscopy and TPR studies, re-
spectively. Mössbauer spectroscopy through hyperfine mag-
netic interaction parameters is a rather sensitive probe of the
size, crystallite distortion, and substructure of metal-doped
α-Fe2O3 particles. Based on the present findings, it can be
concluded that Fe/Ce-modified ferrite is a promising system
for the WGS reaction during steam-rich operations. The con-
ditions of operation play a vital role on the eventual WGS ac-
tivity exhibited by each individual catalyst investigated in this
study.
ide resulted in improved reducibility of Fe3 into Fe species,
as evidenced by the TPR results, thereby promoting WGS ac-
tivity. On doping with Cu, the mobility of lattice oxygen and
hydroxyl groups increased, due to the greater electronegativ-
ity of Cu (1.9) relative to Fe (1.8), thereby improving catalytic
activity [50]. Iron–ceria-based WGS catalysts are promising,
because the oxygen storage capacity (OSC) of ceria and the
cooperative effect of Ce–Fe leads to active sites. Interestingly,
both iron and ceria undergo facile charge-transfer reactions
+
2+
III
II
IV
III
between Fe ↔ Fe and Ce ↔ Ce redox couples, re-
spectively; the synergism between the two couples could be
responsible for the improved WGS activity. In addition, the in-
creased WGS activity with increase in temperature could be
due to the improved OSC of ceria [51]. A noticeable interac-
tion between iron and other substitutent metal was interpreted
in terms of the changes in the lattice parameters and isomer
Acknowledgments
Financial support was provided by the U.S. Department of
Energy (grant DE-PS36-03GO93007) and the National Science
Foundation (grant DMR 04-56472).

104
A. Khan et al. / Journal of Catalysis 253 (2008) 91–104
Supplementary material
[22] E. Xue, M. O’Keeffe, J.R.H. Ross, Catal. Today 30 (1996) 107.
[23] J.A. Dumesic, H. Tøpsoe, Adv. Catal. 26 (1977) 121.
[
24] M.V. Twigg, Catalyst Handbook, second ed., Wolfe Publishing, London,
The online version of this article contains additional supple-
mentary material.
1989, pp. 268–288.
[25] D.S. Newsome, Catal. Rev.-Sci. Eng. 21 (1980) 275.
[26] E.F. Armstrong, T.P. Hilditch, Proc. R. Soc. A 97 (1920) 265.
[27] J.O. Edstrom, J. Iron Steel Inst. 175 (1953) 289.
Please visit DOI:10.1016/j.jcat.2007.10.018.
References
[28] M.V.C. Sastri, R.P. Vishwanath, B. Vishwanath, Int. J. Hydrogen Energy 7
1982) 951.
(
[
[
29] P.A. Webb, MIC Technical Publications, January 2003, pp. 1–12.
30] J.C. Gonzalez, M.G. Gonzalez, M.A. Laborde, N. Moreno, Appl. Catal. 20
[
[
[
[
[
1] E. Kikuchi, S. Uemiya, N. Sato, H. Inoue, T. Matsuda, Chem. Lett. 489
1989) 1.
2] S. Uemiya, N. Sato, H. Ando, E. Kikuchi, Ind. Eng. Chem. Res. 30 (1991)
85.
3] A. Basile, E. Drioli, F. Santella, V. Violante, G. Capannelli, G. Vitulli, Gas
Sep. Purif. 10 (1996) 53.
(
(1986) 3.
[31] O.J. Wimmers, P. Arnoldy, J.A. Moulijn, J. Phys. Chem. 90 (1986) 1331.
[32] E.P. Reddy, B. Sun, P.G. Panagiotis, J. Phys. Chem. B 108 (2004) 17198.
[33] R. Brown, M.E. Cooper, D.A. Whan, Appl. Catal. 3 (1982) 177.
5
4] A. Basile, A. Crscuoli, F. Santella, E. Drioli, Gas Sep. Purif. 10 (1996)
[34] A. Khan, P.G. Smirniotis, J. Mol. Catal. A Chem. (2007), in press.
[35] J.Y. Kim, J.A. Rodriguez, J.C. Hanson, A.I. Frenkel, P.L. Lee, J. Am.
Chem. Soc. 123 (2003) 10684.
[36] G.C. de Araujo, M. do C. Rangel, Catal. Today 62 (2000) 201.
[37] F. Giordano, A. Trovarelli, C. de Leitenburg, M. Giona, J. Catal. 193
(2000) 273.
[38] D.G. Reithwisch, J.A. Dumesic, Appl. Catal. 97 (1986) 109.
[39] G.K. Boreskov, Kinet. Katal. 11 (1970) 374.
[40] S. Morup, J.A. Dumesic, H. Topsoe, in: R.L. Cohen (Ed.), Applications
of Mössbauer Spectroscopy, vol. II, Academic Press, New York, 1980,
p. 1.
2
43.
5] A. Criscuoli, A. Basile, E. Drioli, O. Loiacono, J. Membr. Sci. 181 (2001)
1.
2
[
[
[
[
6] A. Criscuoli, A. Basile, E. Drioli, Catal. Today 56 (2000) 53.
7] Y.S. Lin, Sep. Purif. Technol. 25 (2001) 39.
8] S.G. Deng, Y.S. Lin, Ind. Eng. Chem. Res. 34 (1995) 4063.
9] Y.S. Lin, I. Kumakiri, B.N. Nair, H. Alsyouri, Sep. Purif. Methods 32
(2002) 229.
[
[
10] J. Dong, W. Liu, Y.S. Lin, AIChE J. 46 (2000) 1957.
11] R.C.C. Costa, F. Lelis, L.C.A. Oliveira, J.D. Fabris, J.D. Ardisson, R.R.A.
Rios, C.N. Silva, R.M. Lago, Catal. Commun. 4 (2003) 525.
12] M.S. Selim, G. Turky, M.A. Shouman, G.A. El-Shobaky, Solid State Ion-
ics Diffusion React. 120 (1999) 173.
[41] N.N. Greenwood, T.C. Gibb, Mössbauer Spectroscopy, Chapman & Hall,
London, 1971.
[42] E. Murad, J. Phys. E 17 (1984) 736.
[
[
[
13] C.G. Ramankutty, S. Sugunan, B. Thomas, J. Mol. Catal. 187 (2002) 105.
14] L.C.A. Oliveira, R.V.A. Rios, W.N. Mussel, J.D. Fabris, R.M. Lago, Stud.
Surf. Sci. Catal. 130 (2000) 2165.
[43] J. Gastebois, J. Quidort, Compt. Rend. 253 (1961) 1257.
[44] C. Janot, H. Gibert, C. Tobias, Bull. Soc. Fr. Miner. Cristallogr. 96 (1973)
281.
[
[
[
15] G.C. Araujo, M.C. Rangel, Catal. Today 62 (2000) 201.
16] B.M. Reddy, A. Khan, Catal. Surv. Asia 9 (2005) 155.
17] H.P. Klug, L.E. Alexander, X-Ray Diffraction Procedures for Polycrys-
talline and Amorphous Materials, second ed., Wiley, New York, 1974.
18] R.D. Shannon, Acta Crystallogr. A 32 (1976) 751;
[45] L. Neel, Ann. Geophys. 5 (1949) 99.
[46] S. Morup, in: G.J. Long (Ed.), Applications of Mössbauer Spectroscopy
Applied to Inorganic Chemistry, vol. 2, Plenum, New York, 1988.
[47] C. Lopez-Cartes, J.A. Perez-Omil, J.M. Pintado, J.J. Calvino, Z.C. Kang,
L. Eyring, Ultramicroscopy 80 (1999) 19.
[48] H. Topsoe, M. Boudart, J. Catal. 31 (1973) 346.
[49] H. Topsoe, J.A. Dumesic, M. Boudart, J. Catal. 28 (1973) 477.
[50] W.F. Podolski, Y.G. Kim, Ind. Eng. Chem. Process. Des. Dev. 13 (1974)
415.
[
[
A. Navrotsky, O.J. Kleppa, J. Inorg. Nucl. Chem. 29 (1967) 2701.
19] A.D. Krawitz, Introduction to Diffraction in Materials Science and Engi-
neering, Wiley–Interscience, New York, 2001.
[
[
20] C. Rhodes, G.J. Hutchings, A.M. Ward, Catal. Today 23 (1995) 43.
21] C. Rhodes, B.P. Williams, F. King, G.J. Hutchings, Catal. Commun. 3
[51] F.J. P-Alonso, I. M-Cabrera, M.L. Granados, F. Kapteijn, J.L.G. Fierro,
J. Catal. 239 (2006) 340.
(2002) 381.