Journal of Catalysis 228 (2004) 66–74

[Link]/locate/jcat

Storage and supply of hydrogen by means of the redox

of the iron oxides modified with Mo and Rh species
Sakae Takenaka ∗ , Tomohiro Kaburagi, Chisa Yamada, Kiyoshi Nomura, Kiyoshi Otsuka ∗
Department of Applied Chemistry, Graduate School of Science and Engineering, Tokyo Institute of Technology, O-okayama 2-12-1, Meguro-ku,
Tokyo 152-8552, Japan
Received 20 March 2004; revised 13 August 2004; accepted 17 August 2004
Available online 25 September 2004

Abstract
The redox performance of iron oxides containing Mo and/or Rh species (Fe3 O4 + 4H2
3Fe + 4H2 O) was investigated. The redox cycle
of the reduction of Fe3 O4 with hydrogen and the subsequent oxidation of iron metal with water vapor can be applied for the storage and
supply of pure hydrogen. Addition of Rh to the iron oxide samples improved the reactivity for the redox at a low temperature. The addition of
Rh to iron oxide samples decreased apparent activation energy for the oxidation of iron metal with water vapor. However, Rh species added
to iron oxides promoted sintering of iron species during the redox. The addition of Mo cations to iron oxides containing Rh prevented the
sintering of iron species during the redox. Therefore, iron oxides containing both Rh and Mo species could form hydrogen repeatedly through
the redox. The local structures of Mo and Rh species added to the iron oxide samples were investigated by Mo and Rh K-edge XANES and
EXAFS. Mo cations added to the iron oxide samples were always stabilized in the ferrites Mox Fe3−x O4 during the repeated redox reactions.
Rh species added to the iron oxide samples were present as Rh–Fe alloys after the reduction with hydrogen, while they were present as Rh
metal after the oxidation with H2 O.
 2004 Elsevier Inc. All rights reserved.

Keywords: Iron oxides added with Mo and Rh; Redox of iron oxides; Hydrogen

1. Introduction Previously, we have proposed a new method for the stor-

age and supply of pure hydrogen by applying the redox be-
tween Fe3 O4 and iron metal [7]:
The redox property of a metal oxide is one of the fac-
tors which determine its catalytic activity and selectivity Fe3 O4 + 4H2
3Fe + 4H2 O. (1)
for a catalytic reaction. Frequently, various foreign metals Fe3 O4 is reduced with hydrogen into iron metal by elim-
or metal oxides, which are often termed as promoters, are inating water from the system and subsequently the metal
added into the metal oxide catalysts in order to improve their is oxidized with water vapor into Fe3 O4 to form hydrogen.
redox performances [1–3]. In general, the surface areas of Hydrogen is stored chemically as Fe metal in this method.
metal oxides decrease due to the sintering when they are By means of this method, 1 mol of Fe can store and re-
used continuously in the catalytic reactions. The promoters generate 1.33 mol of H2 , which corresponds to 4.8 wt%
sometimes prevent the sintering of the host metal oxide cat- of Fe metal. However, iron oxide sample without any pro-
alysts [4–6]. Therefore, the promoters are indispensable for moters was deactivated quickly for the hydrogen formation
the design of the catalysts for many catalytic reactions. through the oxidation of iron metal with water vapor due
to the sintering of Fe3 O4 and/or iron metal [8]. In addi-
tion, the redox reaction of iron oxide sample without any
* Corresponding authors. Fax: +81-3-5734-2879. promoters required relatively high temperatures (> 673 K).
E-mail address: stakenak@[Link] (S. Takenaka). It is desirable to perform the redox reactions of iron oxide
0021-9517/$ – see front matter  2004 Elsevier Inc. All rights reserved.
doi:10.1016/[Link].2004.08.027
 S. Takenaka et al. / Journal of Catalysis 228 (2004) 66–74 67

samples at temperatures as low as possible from an econom- ides could not be observed. After the hydrogen remaining in
ical viewpoint. Thus, the redox performances of the iron the reactor was purged out with Ar, water vapor balanced
oxide samples added with various metal species were in- with Ar (P (H2 O) = 5.0 kPa, total pressure = 101.3 kPa,
vestigated [8,9]. Iron oxides added with Mo cations could total flow rate = 100 ml min−1 ) was contacted with the
form hydrogen repeatedly through the redox, while iron reduced iron oxide samples at 373 K. The temperature at
oxide without any promoters was deactivated quickly for the reactor was increased linearly with time to 873 K at a
the redox. The addition of Rh into iron oxides enhanced rate of 4 K min−1 . The oxidation of the iron oxide sam-
hydrogen formation through the oxidation of iron metal ples with water vapor was continued at 873 K until the
with water vapor at low temperatures. However, the role formation of hydrogen was not observed. During the oxi-
of these promoters on the redox of iron oxides has been dation with water vapor, a part of effluent gases from the
unclear. The role of these promoters on the redox should catalyst bed was sampled out and analyzed by GC. The re-
be investigated on the bases of the local structures and the duction with hydrogen and the subsequent oxidation with
electronic states of metal species added into the iron ox- water vapor for the iron oxide samples were carried out re-
ides. peatedly under the same conditions. XRD studies showed
In the present study, the local structures and the elec- that iron species in all the fresh iron oxide samples used
tronic states of Rh and Mo species added into the iron oxide in the present study were Fe2 O3 mainly, irrespective of
samples were investigated by Rh and Mo K-edge XANES the kind of promoters added into the iron oxide samples.
and EXAFS. On the bases of these results, the roles of Rh The Fe2 O3 in the fresh samples was reduced with hydro-
and Mo species added into the iron oxide samples on the gen into iron metal at the first reduction. Subsequently the
redox reactions between Fe3 O4 and Fe metal will be dis- metal was oxidized into Fe3 O4 at the first oxidation with
cussed. water vapor. The oxidation of iron metal with water va-
por to Fe2 O3 was thermodynamically impossible. After the
first cycles, the redox was performed between Fe3 O4 and Fe
2. Experimental
metal.
X-ray absorption spectra (XANES, X-ray absorption
Iron oxide without any promoters (denoted as FeOx here-
near-edge structure; and EXAFS, extended X-ray absorp-
after) was prepared by precipitation of Fe(OH)3 from an
tion fine structure) were measured on the beam line BL-10B
aqueous solution of Fe(NO3 )3 by using hydrolysis of urea
at the Photon Factory in the Institute of Materials Structure
at 363 K. The iron oxides added with Rh and/or Mo species
Science for High Energy Accelerator Research Organiza-
(denoted as Rh–FeOx , Mo–FeOx and Rh–Mo–FeOx here-
tion at Tsukuba in Japan (Proposal Number 2002G108). Mo
after) were prepared by coprecipitation of the corresponding
and Rh K-edge XANES and EXAFS of iron oxides added
metal hydroxides from mixed aqueous solutions containing
with Rh and Mo were measured with a transmission mode
these metal cations by using hydrolysis of urea at 363 K.
RhCl3 and (NH4 )6 Mo7 O24 were used as metal sources. with a Si(311) channel-cut monochromator at room tem-
The precipitates were dried at 373 K for 12 h and cal- perature. Mo–FeOx and Rh–Mo–FeOx were reduced with
cined at 773 K for 10 h in air. The amount of Rh or Mo hydrogen at 823 K and they were subsequently oxidized
species added into iron oxides (Rh–FeOx or Mo–FeOx ) with water vapor at 873 K prior to the measurement of
was adjusted to be 5 mol% (Rh/(Rh + Fe) = Mo/(Mo + XANES and EXAFS. After the redox reactions, the sample
Fe) = 0.05 as mole ratio). When both Mo and Rh species was packed in a bag made from polyethylene under Ar. The
were added into iron oxides, the content of Rh and Mo analysis of EXAFS data was performed by using an EXAFS
added was adjusted to be 5 mol% each in total metal atoms analysis program, REX (Rigaku Co.). For EXAFS analy-
(Rh/(Rh + Mo + Fe) = Mo/(Rh + Mo + Fe) = 0.05 as mole sis, the oscillation was extracted from the EXAFS data by
ratio). a spline-smoothing method. The oscillation was normalized
The reduction of iron oxides with hydrogen and the sub- by the edge height around 70–100 eV above the threshold.
sequent oxidation of iron metal with water vapor were car- The Fourier transformation of k 3 -weighted EXAFS oscil-
ried out with a conventional gas-flow system with a fixed- lation was performed over a k-range of 3.5 to 14.5 Å−1 .
bed reactor. Iron oxide sample was packed in a tubular reac- Inversely Fourier-transformed data for each Fourier peak
tor made from quartz (an inner diameter and lengths = 1.8 were analyzed by a curve-fitting method, using theoretically
and 60 cm, respectively). The amount of the iron oxide sam- phase-shift and amplitude functions derived by FEFF8 pro-
ples packed in the reactor was 0.20 g as Fe. For the reduc- gram [10].
tion, hydrogen diluted with Ar (P (H2 ) = P (Ar) = 50.7 kPa, X-ray diffraction (XRD) patterns of iron oxide samples
total flow rate = 100 ml min−1 ) was contacted with the were recorded with a Rigaku RINT 2500V diffractometer
iron oxide sample at 573 K. The temperature at the reac- using Cu-Kα radiation at room temperature under air.
tor was increased linearly with time to 823 K at a rate of SEM images of the iron oxide samples were measured
7.5 K min−1 and the temperature was kept at 823 K until the with a Hitachi FE-SEM S-800 (field emission gun scanning
consumption of hydrogen due to the reduction of iron ox- electron microscope).
68 S. Takenaka et al. / Journal of Catalysis 228 (2004) 66–74

3. Results and discussion tion through the oxidation with water vapor was mitigated
by the addition of Mo cations. The oxidation of Mo–FeOx
3.1. Redox of iron oxides with water vapor produced hydrogen at temperatures higher
than 550 K and the formation rate of hydrogen attained a
The redox performances of FeOx , Rh–FeOx , Mo–FeOx , maximum at around 650 K at the first oxidation with wa-
and Rh–Mo–FeOx were examined. For the reduction of ter vapor. The formation rate of hydrogen in the temperature
these samples, hydrogen was contacted with these samples range of 600 to 700 K for Mo–FeOx was significantly higher
at 573 K and the temperatures at the reactor were increased than that for FeOx . It should be noted that the kinetic curves
to 823 K. After the reduction, the reduced samples were of hydrogen formation for Mo–FeOx did not change appre-
contacted with water vapor at 373 K and the temperature in- ciably from the first oxidation to the fifth oxidation; i.e., the
creased to 873 K. Five redox cycles were repeated for all the formation rates of hydrogen attained a maximum at around
samples except for Rh–FeOx . Fig. 1 shows change of the for- 650 K for every cycle of Mo–FeOx , while the oxidation of
mation rate of hydrogen as a function of temperature at the FeOx with water vapor required higher temperatures with
reactor during the first oxidation, the third oxidation, and the the repeated cycles. On the other hand, the addition of Rh
fifth oxidation with water vapor of each sample. Iron oxide species into FeOx enhanced the formation rate of hydrogen
sample without promoters (FeOx ) produced hydrogen in the at low temperatures. The first oxidation of the Rh–FeOx with
first oxidation with water vapor when temperatures became water vapor formed hydrogen at temperatures higher than
higher than 550 K. The formation rate of hydrogen for FeOx 500 K and the formation rate of hydrogen attained a maxi-
increased with a rise of temperatures and it attained the max- mum at ca. 600 K. The formation rates of hydrogen in the
imum at around 780 K. However, the formation rates of hy- temperature range of 500 to 600 K at the first oxidation of
drogen for FeOx in the temperature range from 550 to 800 K Rh–FeOx with water vapor were considerably higher than
became slower gradually with the repeated cycles. The deac- those at the first oxidation of FeOx and Mo–FeOx . How-
tivation of FeOx for the oxidation with water vapor resulted ever, the Rh–FeOx was deactivated quickly for the hydrogen
from the sintering of iron species, as will be described below formation with the repeated cycles. The formation rate of hy-
in detail. The deactivation of FeOx for the hydrogen forma- drogen at the third oxidation of Rh–FeOx with water vapor
was very slow over the whole temperature range. As will be
described below, the addition of Rh only into the iron oxides
promoted the sintering of Fe3 O4 and/or iron metal during
the redox. Thus, the reactivity of Rh–FeOx for the hydro-
gen formation in the third oxidation with water vapor was
lower than that of FeOx in the third oxidation with water va-
por. In contrast, the addition of both Mo and Rh species into
FeOx improved the durability for the redox as well as the
reactivity of hydrogen formation at low temperatures. Oxi-
dation of Rh–Mo–FeOx with water vapor formed hydrogen
at temperatures higher than 500 K. In addition, the peak of
the formation rate of hydrogen for Rh–Mo–FeOx was posi-
tioned at a lower temperature (ca. 620 K), compared to those
for Mo–FeOx (at 660 K) and for FeOx (at 700 K). The reac-
tivity of Rh–Mo–FeOx for the hydrogen formation through
the oxidation with water vapor was kept high from the first
oxidation to the fifth oxidation. These results implied that
the addition of Mo cations into FeOx prevented the sinter-
ing of Fe3 O4 and/or iron metal during the repeated redox. It
is likely that Rh species added into the iron oxides worked
as the active sites for water molecules during the oxidation
with water vapor, because the addition of Rh species into
FeOx enhanced hydrogen formation at low temperatures.
Apparent activation energies for the oxidation with water
vapor of the iron oxide samples (FeOx , Mo–FeOx , and Rh–
Mo–FeOx ) that had been reduced with hydrogen at 673 K
were estimated. The redox reactions were performed with a
gas-closed and gas-circulation system. The formation rates
Fig. 1. Change of the formation rate of hydrogen as a function of tem- of hydrogen at the early period of the first oxidation with
peratures during the oxidation of reduced iron oxides with water vapor. water vapor were measured at different temperatures in the
", FeOx ; !, Mo–FeOx ; 2, Rh–FeOx ; 1, Rh–Mo–FeOx . range of 453 to 583 K. The apparent activation energies
 S. Takenaka et al. / Journal of Catalysis 228 (2004) 66–74 69

Fig. 2. XRD patterns of Rh–Mo–FeOx before and after the oxidation with
water vapor and the reference samples (Fe2 O3 and Fe3 O4 ). (a) Fe2 O3 ref-
erence; (b) fresh Rh–Mo–FeOx ; (c) Rh–Mo–FeOx after the first oxidation;
(d) Rh–Mo–FeOx after the third oxidation. (e) Fe3 O4 reference.

for the first oxidation with water vapor of the FeOx , Mo–
FeOx , and Rh–Mo–FeOx were estimated to be 62, 59, and
37 kJ mol−1 , respectively. The addition of Rh species into
the Mo–FeOx decreased the apparent activation energy for
the oxidation of iron metals with water vapor, while the acti-
vation energy did not change appreciably by the addition of Fig. 3. Mo K-edge XANES spectra of Mo–FeOx after the redox and of the
reference samples (MoO3 and Mo foil). (a)–(d) Mo–FeOx samples: (a) after
Mo cations into FeOx . These results implied that Rh species
the first reduction; (b) after the first oxidation; (c) after the third reduction;
added into the iron oxides activated water molecules during (d) after the third oxidation. (e) MoO3 . (f) Mo foil.
the oxidation, while Mo cations in the samples could not ac-
tivate them.
ples were reduced with hydrogen at 823 K and they sub-
3.2. Characterization of Rh–Mo–FeOx sequently were oxidized with water vapor at 873 K prior
to the XRD measurements. Thus, iron metals which had
Fig. 2 shows XRD patterns of Rh–Mo–FeOx before been formed by the reduction of Rh–Mo–FeOx with hydro-
and after the redox reactions and of the reference samples gen were oxidized with water vapor completely into Fe3 O4 .
(Fe2 O3 and Fe3 O4 ). The XRD patterns of the used Rh–Mo– A very weak peak due to crystallized Rh metal was observed
FeOx were measured after the first oxidation and after the at 2θ = 41.5◦ for the XRD pattern of Rh–Mo–FeOx after the
third oxidation with water vapor. The diffraction peaks due third oxidation. In contrast, no diffraction signals due to the
to Fe2 O3 and Fe3 O4 were observed in the XRD pattern of compounds containing Mo cations appeared in the XRD pat-
the fresh Rh–Mo–FeOx sample (spectrum b), although the terns of the Rh–Mo–FeOx samples after the redox.
peaks due to Fe3 O4 were significantly smaller than those due As described earlier, the addition of Mo cations into
to Fe2 O3 . Any diffraction peaks assignable to compounds FeOx prevented the deactivation for the hydrogen formation
containing Mo or Rh species were not observed for the fresh through the oxidation of iron metals with water vapor. The
Rh–Mo–FeOx . These results implied that Rh and Mo species structure of Mo cations in Mo–FeOx could not be clarified
in the fresh Rh–Mo–FeOx were highly dispersed. XRD pat- by XRD studies. Thus, Mo K-edge XANES and EXAFS
terns of Rh–Mo–FeOx after the first oxidation (spectrum c) for Mo–FeOx were measured in order to examine the local
and after the third oxidation (spectrum d) were consistent structures around Mo atoms added in FeOx . Fig. 3 shows Mo
with that of Fe3 O4 (spectrum e). These Rh–Mo–FeOx sam- K-edge XANES spectra of Mo–FeOx after the redox and of
70 S. Takenaka et al. / Journal of Catalysis 228 (2004) 66–74

the reference samples (MoO3 and Mo foil). The spectra of

Mo–FeOx were measured after the reduction with hydrogen
at 823 K or after the subsequent oxidation with water vapor
at 873 K. Fig. 3B shows the threshold of XANES spectra
of these samples in an energy range of 19990 to 20020 eV.
XANES spectra for all the Mo–FeOx samples (spectra a–d)
were similar to one another regardless of different treatments
for the samples (after the reduction with hydrogen or after
the oxidation with water vapor). These spectra of Mo–FeOx
were not consistent with that of MoO3 (spectrum e) or that
of Mo foil (spectrum f). It is well accepted that the posi-
tion of threshold for XANES spectra of any metal species is
sensitive to the oxidation states of the corresponding metal
species; i.e., the threshold of XANES spectra of the metal
species is shifted to higher energy with the higher oxidation
state of the metal species [11–13]. The threshold of XANES
spectra shown in Fig. 3B was positioned to higher energy in
the following order of Mo foil < Mo–FeOx < MoO3 , sug-
gesting that the oxidation number of Mo cations in the Mo–
FeOx after the redox was lower than 6+ (VI) and higher than
zero [14,15]. It was reported that Mo cations were stabilized
on the B sites in the ferrites of the formula Mox Fe3−x O4
when Mo cations were added into Fe3 O4 [16,17]. The oxida-
tion number of Mo cations in the ferrites was evaluated to be Fig. 4. Mo K-edge k 3 -weighted EXAFS of Mo–FeOx after the redox.
3+ (III) or 4+ (IV). The present results of the XANES spec- (a) After the first reduction; (b) after the first oxidation; (c) after the third
reduction; (d) after the third oxidation.
tra of Mo–FeOx were consistent with these previous reports.
It should be noted that the threshold of XANES spectra for
the Mo–FeOx after the reduction with hydrogen was po- Fourier transforms for Mo K-edge k 3 -weighted EXAFS
sitioned at slightly lower energy compared to that for the of Mo–FeOx after the redox were performed in order to ob-
tain radial structural functions (RSFs) of the spectra. The
Mo–FeOx after the oxidation with water vapor. Thus, the
RSFs for Mo–FeOx are shown in Fig. 5. The RSFs for Mo–
fraction of Mo3+ in all the Mo cations (Mo3+ and Mo4+ )
FeOx after the reduction with hydrogen (spectra a and c)
in the Mo–FeOx after the reduction with hydrogen would
were different from those of Mo–FeOx after the oxidation
be higher than that in the Mo–FeOx after the oxidation with
with water vapor (spectra b and d). Bouet et al. reported
water vapor.
Fourier transforms of Mo K-edge EXAFS for the ferrites
Fig. 4 shows Mo K-edge k 3 -weighted EXAFS of Mo–
of the formula Mox Fe3−x O4 [17]. The RSFs of the ferrites
FeOx after the redox. The EXAFS spectrum for Mo–FeOx
Mox Fe3−x O4 were quite similar to ones shown in Fig. 5.
after the first reduction (spectrum a) was similar to that of Thus, Mo cations in Mo–FeOx were always stabilized in the
Mo–FeOx after the third reduction (spectrum c). In addition, ferrites Mox Fe3−x O4 during the repeated redox. It is likely
the EXAFS spectrum for Mo–FeOx after the first oxidation that changes of the RSFs for Mo–FeOx during the redox re-
(spectrum b) was consistent with that of Mo–FeOx after the sulted from the change of the fraction of Mo3+ and Mo4+ in
third oxidation (spectrum d). The spectra of Mo–FeOx after the ferrites Mox Fe3−x O4 .
the reduction were different from those of Mo–FeOx after Fig. 6 shows Rh K-edge XANES spectra of Rh–Mo–
the oxidation. Thus, the local structure around Mo atoms in FeOx after the redox, and the spectrum of Rh foil. The
Mo–FeOx after the reduction with hydrogen was changed spectra of the Rh–Mo–FeOx were measured after the re-
by the oxidation with water vapor; however, it was recov- duction with hydrogen at 823 K or after the oxidation with
ered again by the subsequent reduction of the Mo–FeOx with water vapor at 873 K. The XANES spectrum of Rh–Mo–
hydrogen, i.e., the reversible change of the local structure FeOx after the first oxidation with water vapor (spectrum b)
around Mo cations during the repeated redox of Mo–FeOx . was consistent with that of Rh–Mo–FeOx after the third ox-
XANES spectra of Mo–FeOx showed that the fraction of idation (spectrum d). These spectra of Rh–Mo–FeOx after
Mo3+ in all the Mo cations (Mo3+ and Mo4+ ) depended the oxidation with water vapor were similar to that of Rh
on the state of the Mo–FeOx (after the oxidation with wa- foil (spectrum e), suggesting that Rh species in the Rh–Mo–
ter vapor or after the reduction with hydrogen). The change FeOx after the oxidation with water vapor were present as
of the local structures around Mo cations in Mo–FeOx dur- Rh metal. On the other hand, XANES spectra of Rh–Mo–
ing the repeated redox must result from the redox between FeOx after the reduction with hydrogen (spectra a and c)
Mo3+ and Mo4+ in the ferrites Mox Fe3−x O4 . were different from that of Rh foil. Thus, Rh species in the
 S. Takenaka et al. / Journal of Catalysis 228 (2004) 66–74 71

Fig. 7. Rh K-edge k 3 -weighted EXAFS of Rh–Mo–FeOx after the redox

Fig. 5. Fourier transforms of Mo K-edge k 3 -weighted EXAFS of Mo–FeOx and of Rh foil. (a–d) Rh–Mo–FeOx : (a) after the first reduction; (b) after
after the redox. (a) After the first reduction; (b) after the first oxidation; the first oxidation; (c) after the third reduction; (d) after the third oxidation.
(c) after the third reduction; (d) after the third oxidation. (e) Rh foil.

Rh–Mo–FeOx after the reduction with hydrogen were not

crystallized Rh metal. It should be noted that the threshold
for XANES spectra of the Rh–Mo–FeOx after the reduction
with hydrogen was consistent with that of Rh foil. This result
implied that Rh species in the Rh–Mo–FeOx after the reduc-
tion with hydrogen were of zero valent, although they were
not Rh metal crystallite itself. As will be described below
in detail, Rh species in the Rh–Mo–FeOx after the reduc-
tion with hydrogen were present as Rh–Fe alloys. Therefore,
Rh K-edge XANES spectra of the Rh–Mo–FeOx after the
reduction with hydrogen were assignable to that of Rh–Fe
alloys.
Fig. 7 shows Rh K-edge k 3 -weighted EXAFS of Rh–
Mo–FeOx after the reduction with hydrogen or after the suc-
cessive oxidation with water vapor, and EXAFS of Rh foil.
Fourier transforms of these EXAFS spectra (RSFs) were
shown in Fig. 8. Fourier transforms were performed over
a k range of 3.5 to 14.5 Å−1 . EXAFS oscillations for the
Rh–Mo–FeOx after the first oxidation (spectrum b in Fig. 7)
and after the third oxidation (spectrum d) were compatible
with that for Rh foil (spectrum e), although the intensity of
EXAFS oscillation for the Rh–Mo–FeOx was smaller than
that for Rh foil. In addition, the RSFs of the Rh–Mo–FeOx
after the oxidation with water vapor (spectra b and d in
Fig. 8) were similar to that for Rh foil (spectrum e). Thus,
Fig. 6. Rh K-edge XANES spectra of Rh–Mo–FeOx after the redox and of
Rh foil. (a–d) Rh–Mo–FeOx : (a) after the first reduction; (b) after the first Rh species in the Rh–Mo–FeOx after the oxidation with wa-
oxidation; (c) after the third reduction; (d) after the third oxidation. (e) Rh ter vapor were present as Rh metal crystallites. On the other
foil. hand, EXAFS spectra of Rh–Mo–FeOx after the reduction
72 S. Takenaka et al. / Journal of Catalysis 228 (2004) 66–74

Table 1
Curve-fitting results for Rh K-edge EXAFS of Rh–Mo–FeOx
Condition Shell C.N.a R b (Å) σ c (Å)
After the first reduction Rh–Fe 6.6 ± 1.0 2.55 0.057
After the first oxidation Rh–Rh 5.9 ± 0.4 2.66 0.058
After the third reduction Rh–Fe 7.4 ± 0.6 2.55 0.056
After the third oxidation Rh–Rh 6.2 ± 0.5 2.67 0.056
a Coordination number.
b Interatomic distance.
c Debye–Waller factor.

Table 2
Results of adsorption of H2 and N2 on the iron oxide samples
Samples Surface areaa (m2 g−1 ) H2 ads.b
Fresh Reduced (µmol g−1 )
FeOx 36 11 54
Mo–FeOx 77 54 46
Rh–Mo–FeOx 63 40 40
a Specific surface area estimated by N adsorption at 77 K with BET
2
methods.
b Hydrogen was adsorbed on the samples reduced with hydrogen. Ad-
sorbed amounts of H2 were estimated from the isotherm assuming that
hydrogen was adsorbed atomically on a metal atom with Langmuir-typed
adsorption.
Fig. 8. Fourier transforms of Rh K-edge k 3 -weighted EXAFS of Rh–
Mo–FeOx after the redox and of Rh foil. (a–d) Rh–Mo–FeOx : (a) after Rh–Rh bond in Rh metal crystals. The coordination number
the first reduction; (b) after the first oxidation; (c) after the third reduction; of Rh–Rh bond for the Rh–Mo–FeOx after the first oxida-
(d) after the third oxidation. (e) Rh foil.
tion was estimated to be 5.9. The coordination number of
Rh–Rh bond did not change significantly after the third oxi-
with hydrogen (spectra a and c in Fig. 7) were different from dation with water vapor. Thus, Rh metal crystallites present
that of Rh foil (spectrum e). The envelope function of the on the Rh–Mo–FeOx after the oxidation with water vapor
EXAFS for the Rh–Mo–FeOx after the reduction had the were not aggregated seriously during the repeated redox cy-
maximum at a k value of 7–9 Å−1 , while the maximum cles.
of the envelope function for Rh foil was positioned at a In order to estimate the fraction of the metallic surface
k value of 9–12 Å−1 . This result implied that Rh atoms in of FeOx , Mo–FeOx , and Rh–Mo–FeOx after the first reduc-
the Rh–Mo–FeOx after the reduction with hydrogen were tion with hydrogen, the adsorption of nitrogen and hydro-
surrounded by lighter atoms than Rh, i.e., Fe atoms. RSFs gen on these samples was performed. These samples were
of the Rh–Mo–FeOx in Fig. 8 also supported this idea. The reduced with hydrogen at 673 K prior to the adsorption ex-
intense peak was observed at 2.1 Å in the RSFs of the Rh– periments. The specific surface areas of the samples were
Mo–FeOx after the reduction with hydrogen (spectra a and c evaluated by BET methods of nitrogen adsorption at 77 K.
in Fig. 8), while a strong peak due to a specific Rh–Rh bond Hydrogen adsorption on these samples was performed at
appeared at 2.3 Å in the RSF of Rh foil. Thus, Rh atoms 273 K. On the basis of these results, the fraction of the
in the Rh–Mo–FeOx after the reduction with hydrogen were metallic surface area to the specific surface area of these
surrounded by Fe atoms, i.e., the formation of Rh–Fe alloys. samples after the reduction with hydrogen was evaluated.
The local structures around Rh atoms in the Rh–Mo– The results are shown in Table 2. Specific surface areas
FeOx during the redox were examined furthermore by of the fresh FeOx , Mo–FeOx and Rh–Mo–FeOx were esti-
curve fitting for Rh K-edge EXAFS. Inversely Fourier- mated to be 36, 77, and 63 m2 g−1 , respectively. The spe-
transformed data for the Fourier peak at a R range of 1.7– cific surface areas for all the samples decreased after the
2.9 Å on the RSFs of Rh–Mo–FeOx in Fig. 8 were analyzed first reduction with hydrogen. Nevertheless, the decrease in
by a curve-fitting method. The structural parameters evalu- the surface areas of Mo–FeOx and Rh–Mo–FeOx after the
ated by the curve fitting are listed in Table 1. The strong peak first reduction with hydrogen was not so serious as that
in the RSFs for the Rh–Mo–FeOx after the reduction with of FeOx ; that is, the specific surface areas of Mo–FeOx
hydrogen could be assigned to a specific Rh–Fe bond. This (54 m2 g−1 ) and Rh–Mo–FeOx (40 m2 g−1 ) were signif-
result indicates the formation of Rh–Fe alloys in the Rh– icantly larger than that of FeOx (11 m2 g−1 ). This result
Mo–FeOx after the reduction with hydrogen. On the other suggests that Mo cations added into FeOx prevented the
hand, the peak in the RSF of the Rh–Mo–FeOx after the sintering of iron metal particles during the reduction with
oxidation with water vapor could be assigned to a specific hydrogen.
 S. Takenaka et al. / Journal of Catalysis 228 (2004) 66–74 73

dox cycles of FeOx , particles of Fe3 O4 and/or iron metal in

FeOx were aggregated significantly to form particles larger
than 200 nm. The aggregation of iron species (Fe3 O4 and/or
iron metal) after the third redox cycles was more serious in
the case of Rh–FeOx . The particle size of Rh–FeOx after
the third redox cycles was significantly larger than that of
FeOx after the third redox cycles. These results implied that
the addition of Rh species into FeOx promoted the sintering
of Fe3 O4 and/or iron metal during the redox. As described
in Fig. 1, Rh–FeOx was deactivated more quickly for the
hydrogen formation through the oxidation with water vapor
than FeOx ; i.e., the formation rate of hydrogen in the third
oxidation of Rh–FeOx with water vapor was lower than that
for FeOx in the third oxidation. This rapid deactivation of
Rh–FeOx would result from the serious aggregation of iron
species. In contrast, average particle sizes of the Mo–FeOx
and Rh–Mo–FeOx were relatively smaller even after the fifth
redox cycles. Thus, we concluded that Mo cations added into
FeOx prevented the sintering of iron species during the re-
peated redox reactions.
The results of the hydrogen adsorption at 273 K on
FeOx , Mo–FeOx , and Rh–Mo–FeOx after the reduction with
hydrogen (Table 2) would provide a relative fraction of
metal surface of these samples, while whole surface areas
of these samples were estimated by nitrogen adsorption at
77 K. The amounts of hydrogen adsorbed on the Mo–FeOx
(46 µmol g−1 ) and Rh–Mo–FeOx (40 µmol g−1 ) after the re-
duction with hydrogen were smaller than that on the FeOx
(54 µmol g−1 ) after the reduction with hydrogen, although
the specific surface areas for the Mo–FeOx (54 m2 g−1 ) and
Rh–Mo–FeOx (40 m2 g−1 ) were significantly larger than
that for the FeOx (11 m2 g−1 ). As described earlier, XANES
and EXAFS spectra indicated that Mo species were present
as the ferrites Mox Fe3−x O4 and Rh species were stabilized
as Rh–Fe alloys in the samples after the reduction with hy-
drogen. As for the hydrogen adsorption at 273 K on the iron
oxide samples after the reduction with hydrogen, hydrogen
atoms would be adsorbed on metallic sites of zero valent,
i.e., Fe-metal sites on the FeOx and Mo–FeOx , and Fe-metal
sites and Rh-metal sites on the Rh–Mo–FeOx . The results in
Table 2 implied that the fraction of metal surface area in the
whole surface area of the Mo–FeOx and Rh–Mo–FeOx af-
ter the reduction with hydrogen was very low compared with
the fraction of metal surface for the FeOx after the reduction
with hydrogen. Thus, considerable fractions of the surface of
the iron metal particles in the Mo–FeOx and Rh–Mo–FeOx
after the reduction with hydrogen would be covered with the
Fig. 9. SEM images of FeOx , Rh–FeOx , Mo–FeOx , and Rh–Mo–FeOx
ferrites Mox Fe3−x O4 . The ferrites Mox Fe3−x O4 that are lo-
samples before and after the redox. calized on the surface of iron metal particles in the Mo–FeOx
and Rh–Mo–FeOx after the reduction with hydrogen should
inhibit the contact between iron metal particles during the
Fig. 9 shows SEM images of FeOx , Rh–FeOx , Mo–FeOx , repeated redox. We believed that this is the main reason why
and Rh–Mo–FeOx before and after the redox. All the fresh the addition of Mo cations into the iron oxides prevented the
iron oxide samples shown in Fig. 9 were composed of Fe2 O3 sintering of iron species (Fe3 O4 and/or iron metal) during
particles with diameters of 20 to 30 nm. After the third re- the redox.
74 S. Takenaka et al. / Journal of Catalysis 228 (2004) 66–74

4. Conclusion References

[1] S. Poulston, N.J. Price, C. Weeks, M.D. Allen, P. Parlett, M. Steinberg,

M. Bowker, J. Catal. 178 (1998) 658.
We concluded as follows based on the results described [2] G. Costentin, J.C. Lavalley, F. Studer, J. Catal. 200 (2001) 360.
above: [3] S. Golunski, R. Rajaram, N. Hodge, G.J. Hutchings, C.J. Kiely, Catal.
Today 72 (2002) 107.
[4] D.G. Barton, S.L. Soled, G.D. Meitzner, G.A. Fuentes, E. Iglesia,
1. The addition of Rh species to iron oxides enhanced the
J. Catal. 181 (1999) 57.
formation of hydrogen at low temperatures through the [5] M.V. Twigg, M.S. Spencer, Top. Catal. 22 (2003) 191.
oxidation of iron metal with water vapor. The addition [6] J. Zao, Z. Feng, F.E. Huggins, N. Shar, G.P. Huffman, I. Wender,
of Rh to the iron oxide samples decreased apparent acti- J. Catal. 148 (1994) 194.
[7] K. Otsuka, A. Mito, S. Takenaka, I. Yamanaka, Int. J. Hydrogen En-
vation energy for the hydrogen formation. However, Rh ergy 26 (2001) 191.
species in iron oxides promoted sintering of iron species [8] K. Otsuka, C. Yamada, T. Kaburagi, S. Takenaka, Int. J. Hydrogen
during the redox. The addition of Mo cations to Rh– Energy 28 (2002) 335.
FeOx prevented the sintering of iron species during the [9] K. Otsuka, T. Kaburagi, C. Yamada, S. Takenaka, J. Power Sour-
ces 122 (2003) 111.
redox. The Rh–Mo–FeOx could produce hydrogen re- [10] A. Ankudinov, B. Ravel, J.J. Rehr, S.D. Conradson, Phys. Rev. B 58
peatedly through the redox. (1998) 7565.
2. Rh species in Rh–Mo–FeOx were present as Rh metal [11] T. Tanaka, H. Yamashita, R. Tsuchitani, T. Funabiki, S. Yoshida,
J. Chem. Soc., Faraday Trans. 1 84 (1988) 2987.
after the oxidation with water vapor, whereas they were
[12] T. Tanaka, T. Hanada, S. Yoshida, T. Baba, Y. Ono, Jpn. J. Appl.
present as Rh–Fe alloys after the reduction with hydro- Phys. 32 (1993) 481.
gen. [13] Y. Wang, Y. Ohnishi, T. Shishido, Q. Zhang, W. Yang, Q. Guo, H. Wan,
3. Mo cations in Rh–Mo–FeOx and Mo–FeOx were al- K. Takehira, J. Catal. 220 (2003) 347.
[14] W. Ding, S. Li, G.D. Meitzner, E. Iglesia, J. Phys. Chem. B 105 (2001)
ways stabilized in the ferrites Mox Fe3−x O4 during the 506.
redox. The surface of iron metal particles in Rh–Mo– [15] K. Asakura, Y. Noguchi, Y. Iwasawa, J. Phys. Chem. B 103 (1999)
FeOx and Mo–FeOx after the reduction with hydrogen 1051.
was covered with the ferrites Mox Fe3−x O4 . The ferrites [16] B. Domenichini, D. Aymes, P. Perriat, B. Gillot, Mater. Chem.
Phys. 39 (1994) 80.
Mox Fe3−x O4 inhibited the contact between iron metal [17] L. Bouet, Ph. Tailhades, A. Rousset, K.R. Kannan, M. Verelst, G.U.
particles during the redox. Kulkarni, C.N.R. Rao, J. Solid State Chem. 102 (1993) 414.