Water gas shift reaction for the reformed fuels over Cu/MnO catalysts prepared via spinel-type oxide

Article abstract of DOI:10.1016/S0021-9517(03)00024-1

Cu/MnO catalysts prepared via reduction of Cu-Mn spinel oxide were investigated for development of active Cu catalysts for the water gas shift reaction (WGSR). A Cu-Mn catalyst active for the WGSR was obtained after high temperature calcination at 900°C and subsequent reduction. The optimum Cu/ Mn ratio for catalytic activity of the Cu-Mn oxide system was 1/2. Nonstoichiometric Cu_1.5Mn_1.5O₄ phase existed stably when copper manganese oxide was calcined above 700°C. The optimized Cu-Mn spinel showed excellent WGSR activity when a larger percentage of CO was used, as in hydrocarbon reforming. Cu-Mn spinel oxides calcined above 900°C were easily reduced. This may be responsible for the high activity of the Cu/ MnO catalyst. Carbon dioxide in the reformed gas significantly depressed WGSR activity below 200°C, while CO conversion reached equilibrium at 200°C in the absence of CO₂.

Full text of DOI:10.1016/S0021-9517(03)00024-1

Journal of Catalysis 215 (2003) 271–278
www.elsevier.com/locate/jcat
Water gas shift reaction for the reformed fuels over Cu/MnO catalysts
prepared via spinel-type oxide
Yohei Tanaka,a Toshimasa Utaka,a Ryuji Kikuchi,a Tatsuya Takeguchi,a
Kazunari Sasaki,b and Koichi Eguchi a,∗
a
Department of Energy and Hydrocarbon Chemistry, Graduate School of Engineering, Kyoto University, Yoshida-honmachi,
Sakyo-ku, Kyoto 606-8501, Japan
Department of Molecular and Material Sciences, Interdisciplinary Graduate School of Engineering Sciences, Kyushu University,
b
6-1 Kasugakoen, Kasuga-shi, Fukuoka 816-8580, Japan
Received 24 September 2002; revised 31 December 2002; accepted 13 January 2003
Abstract
Cu/MnO catalysts prepared via reduction of Cu–Mn spinel oxide were investigated for development of active Cu catalysts for the water
◦
gas shift reaction (WGSR). A Cu–Mn catalyst active for the WGSR was obtained after high temperature calcination at 900 C and subsequent
reduction. The optimum Cu/Mn ratio for catalytic activity of the Cu–Mn oxide system was 1/2. Nonstoichiometric Cu Mn
O phase
1.5
1.5
4
◦
existed stably when copper manganese oxide was calcined above 700 C. The optimized Cu–Mn spinel showed excellent WGSR activity
◦
when a larger percentage of CO was used, as in hydrocarbon reforming. Cu–Mn spinel oxides calcined above 900 C were easily reduced.
This may be responsible for the high activity of the Cu/MnO catalyst. Carbon dioxide in the reformed gas significantly depressed WGSR
◦
◦
activity below 200 C, while CO conversion reached equilibrium at 200 C in the absence of CO .
2
2003 Elsevier Science (USA). All rights reserved.
Keywords: Cu/MnO; CuMn O ; Spinel oxide; Water gas shift reaction; CO removal; Reformed gas
2
4
1. Introduction
biles. To solve this problem, on-board reforming of hydro-
carbons, methanol, and dimethyl ether has been proposed
[1,6]. Recently, hydrocarbons were proposed as a hydrogen
source for PEFCs because there exists an infrastructure, such
as petroleum stations and city gas pipelines. However, re-
formed fuels from hydrocarbons contain higher levels of CO
than fuels from methanol, e.g., 1–10% CO, due to the ther-
modynamic equilibrium. Carbon monoxide is irreversibly
adsorbed on the Pt electrode of a PEFC, leading to deteri-
oration in cell performance. Therefore, the CO level must
be reduced to that allowable for a Pt electrode (10–20 ppm)
[7–9].
Hydrogen has attracted much attention as one of the most
important energy media due to its cleanliness and potential
application to various energy conversion processes. Hydro-
gen has been used in refineries and recently was proposed as
a fuel for combustion and fuel cells [1]. In principle, fuel cell
systems have higher efficiency of electricity generation than
internal combustion engine systems since chemical energy
is almost directly converted to electric energy in fuel cells.
Polymer electrolyte fuel cells (PEFCs) constitute a promis-
ing system for power generation on a small scales [2–5].
PEFCs are operated at low temperature, around 80 ◦C, and
have high energy density suitable for small stationary gener-
ators and vehicles.
The water gas shift reaction (WGSR) has been investi-
gated and used in industry [10–13]:
H2O + CO → H2 + CO2.
Since dense storage or liquefaction of hydrogen is possi-
ble only at high pressure and/or low temperature, hydrogen
storage is an obstacle in use of H2-fueled PEFCs in automo-
The reaction temperature is relatively controllable because
of the moderately exothermic nature (ꢀH298 = −41.1 kJ/
mol) in contrast to the large exothermic heat for CO oxida-
tion to CO2 (ꢀH298 = −283 kJ/mol). Hence, the WGSR
is desirable for CO removal from reformed fuels contain-
ing high concentrations of CO. The difficulty in removing
*
Corresponding author.
E-mail address: eguchi@scl.kyoto-u.ac.jp (K. Eguchi).
0021-9517/03/$ – see front matter
doi:10.1016/S0021-9517(03)00024-1
2003 Elsevier Science (USA). All rights reserved.

272
Y. Tanaka et al. / Journal of Catalysis 215 (2003) 271–278
CO from the reformed gas by the WGSR derives from the
following: Since WGSR is exothermic, the lower reaction
temperature favors the shift in chemical equilibrium in the
forward direction. On the other hand, from a kinetics view-
point, catalysts are not active enough to reach equilibrium at
low temperature. The copper based catalyst for WGSR can
be deactivated by H2O at low temperatures as we have pre-
viously reported [14]. Therefore, the optimum reaction tem-
perature is generally found to be between 200 and 250 ◦C.
We have found out that the Cu–Mn spinel oxide showed
excellent WGSR activity comparable to that of conventional
Cu/ZnO/Al2O3 despite its low surface area. Thus, we in-
vestigated Cu–Mn spinel oxide with low surface area as
a promising catalyst for the WGSR. Cu–Mn spinel oxide
was reduced by H2 in the reformed gas and worked as
Cu/MnO [15]. CuMn2O4, called hopcalite, has been inves-
tigated as a CO oxidation catalyst at ambient temperature in
applications such as gas masks for miners and also oxidation
catalysts for other gas species [16–18].
Pi in P500, P700, P900, and P1100. Temperature-programmed
reduction (TPR) was carried out by feeding 10% H2 in Ar
to 25 mg of catalyst sample in a conventional flow reac-
tor without oxidation treatment prior to measurements. The
flow rate of the reducing gas was set at 40 ml/min. The
temperature of the reactor was raised from room temper-
ature to 800 ◦C at the rate of 10 K/min. The rate of H2
consumption was determined by using a thermal conduc-
tivity detector and recorded on an on-line personal com-
puter. N2O titration (N2O + 2Cu → Cu2O + N2) was im-
plemented at 100 ◦C by injecting N2O pulses into a conven-
tional flow reactor. Samples were reduced with 10% H2/He
at 300 ◦C for 10 min prior to the titration. [Cu]/[Mn] ra-
tios were determined by ICP emission spectrometry (Seiko
Instruments, SPS1700HVR) after copper manganese oxides
were dissolved in a dilute hydrochloric acid solution.
2.2. Evaluation of catalytic activity
In this article, effects of preparation conditions, concen-
trations of gas species, and reduction treatment on the cat-
alytic activity of Cu–Mn spinel oxide for the WGSR in re-
formed fuels are reported. The effect of the method of prepa-
ration of Cu/MnO via spinel-type oxide and the reason for
the high activity are also discussed.
Cu–Mn spinel oxide catalysts were used without prere-
duction treatment and reduced by H2 in the feed gas unless
otherwise noted. Cu/ZnO/Al2O3 was reduced at 220 ◦C for
2 h in a 20% H2/N2 stream prior to evaluation of activity.
Catalytic activity was examined in a conventional flow re-
actor at atmospheric pressure in the temperature range 200
to 350 ◦C and at a constant space velocity of 6400 h−1, us-
ing 1.5 ml of catalyst. A model gas after steam reforming
of methanol containing H2, CO, H2O, CO2, and N2 (bal-
ance) was supplied to the catalyst bed through mass flow
controllers (STEC, SEC-400MK3). The standard composi-
tion of the reaction gas was 37.5% H2, 1.25% CO, 25.0%
H2O, 12.5% CO2, and N2 (balance). The gas composition
before and after the reaction was analyzed by on-line gas
chromatography with a thermal conductivity detector (Shi-
madzu, GC-8A). A molecular sieve 13X column was used
for separation of H2, O2, N2, and CO, and a Porapak Q col-
umn for CO2.
2. Experimental
2.1. Preparation and characterization
Copper manganese oxide catalysts and Cu/ZnO/Al2O3
catalyst were prepared by the coprecipitation method. In the
preparation of Cu–Mn spinel oxides, aqueous ammonia and
a solution of nitrates (Kishida, analytical grade) were added
to a beaker to maintain pH 8. The resultant gel was dried in
air at 100 ◦C for 12 h and precalcined at 400 ◦C for 1 h to
remove nitrate residue. Then the dried powder was calcined
at 900 ◦C for 10 h. The precipitate of Cu/ZnO/Al2O3
was obtained by adding aqueous ammonia to a solution of
nitrates of Cu, Zn, and Al until the precipitate formed at
around pH 7. The resultant slurry was dried at 100 ◦C for
12 h, precalcined at 400 ◦C for 1 h, and then calcined at
500 ◦C for 3 h in air.
3. Results and discussion
3.1. Effect of calcination temperature
We have reported that Cu–Mn spinel oxide calcined at
900 ◦C shows excellent WGSR activity for removal of CO
in reformed fuels despite its low surface area [15]. Under
reducing conditions including a large amount of H2, the Cu–
Mn spinel oxide catalyst is reduced to Cu/MnO. The surface
area of Cu–Mn spinel oxide, so-called hopcalite, is very low
(≈ 1 m2/g). It is expected that a decrease in calcination
temperature would minimize the loss of surface area and
WGSR activity. The calcination temperature of the catalyst
was varied from 500 to 1100 ◦C. Hereafter, Cu–Mn spinel
oxide calcined at X ◦C is denoted as CuMnX (X = 500, 700,
900, and 1100).
Surface area of the catalysts was measured by the
BET method with N2 adsorption using a Yuasa Ionics
NOVA2200. The crystalline phase of the catalysts was
determined by X-ray diffraction (XRD) using a Rigaku
RINT1400. To estimate the crystallinity of Cu–Mn spinel
oxide, 0.150 g of silicon powder (99.9995% purity, Aldrich)
was added to 0.150 g of sample powder as an internal stan-
dard material. Crystallinity of spinel oxides is defined as
Ci = 100Pi/Pmax; Pi = Si/S0, where i represents the calci-
nation temperature of Cu–Mn oxides. Si is the peak area of
(311) diffraction of spinel oxides at 2θ ≈ 36◦, S0 is the peak
area of silicon (111) at 2θ ≈ 28.4◦. Pmax is the maximum

Y. Tanaka et al. / Journal of Catalysis 215 (2003) 271–278
273
Fig. 1. XRD patterns of CuMnX: (") Cu Mn O , (1) Mn O , (Q) CuMn O , (P) Mn O .
1.5
1.5
4
3
4
2
4
2 3
Table 1
Properties of CuMnX catalysts after calcination
a
Sample
Calcination temperature
Crystalline phase
BET surface area
Crystallinity
(%)
Crystallite size
◦
2
( C)
(m /g)
(nm)
CuMn500
CuMn700
CuMn900
CuMn1100
500
700
900
CuMn O
6.6
4.4
0.8
0.6
88
100
94
33
43
32
37
2
4
Cu Mn O , Mn O
1.5
1.5
4
2
3
4
4
Cu Mn O , Mn O
1.5 1.5 4 3
1100
Cu Mn O , Mn O
92
1.5
1.5
4
3
a
Crystallite size was obtained for the 311 peak by the Sherrer equation.
Fig. 1 and Table 1 give the crystalline phase, BET sur-
face area, crystallinity, and crystallite size of CuMnX. BET
surface area of CuMnX decreased with an increase in cal-
cination temperature. This was because particles of CuMnX
shrank as calcination temperature rose. The crystalline phase
of CuMn500 was a stoichiometric CuMn2O4. CuMn700 was
composed of a predominant Cu1.5Mn1.5O4 phase and a mi-
nor Mn2O3 phase. For CuMn900 and CuMn1100 a minor
Mn3O4 phase was detected together with a predominant
Cu1.5Mn1.5O4 phase. This agrees with the results of Hutch-
ings et al., who reported that Cu–Mn solid solution seg-
regated from stoichiometric spinel phase to CuxMn3 − xO4
and Mn2O3 above 600 ◦C [19]. As shown in Table 1, non-
stoichiometric Cu1.5Mn1.5O4 was formed 700 ◦C. It is prob-
able that the phase separation from stoichiometric CuMn2O4
occurred above 700 ◦C, considering the Tamman tempera-
ture of Mn2O3 (403 ◦C), which is nearly half of melting
point in inorganic materials and indicates the temperature
where the solid phase reaction can begin [19,20].
Fig. 2. TPR profiles of CuMnX after calcination. TPR conditions: heating
rate, 10 K/min in 10% H /Ar.
2
spinel started around 600–700◦C. The stable phase at this
composition is the mixture of Cu1.5Mn1.5O4 and manganese
oxide. At 900–1100◦C, reduction of Mn2O3 to Mn3O4 took
place, and the crystallinity and crystallite size of Cu–Mn
spinel oxide decreased compared with those of CuMn700.
Fig. 2 comprises TPR profiles of CuMnX after calci-
nation. TPR was carried out without preoxidation. Fig. 3
presents the TPR profiles of CuO, Mn2O3, CuMn900, and
CuMn900 after WGSR, suggesting that CuO was more re-
ducible than Mn2O3. From this figure, the lower temperature
region of the reduction peak seen in Fig. 2 was ascribed to
Judging from the results in Table 1, the degree of
crystallization of Cu–Mn spinel oxide increased up to
700 ◦C, then decreased monotonously. The particle size of
Cu–Mn spinel oxide also showed a maximum at 700 ◦C.
This suggests that the stoichiometric CuMn2O4 crystal grew
with an increase in calcination temperature up to 700 ◦C,
and, at the same time, decomposition of stoichiometric

274
Y. Tanaka et al. / Journal of Catalysis 215 (2003) 271–278
Fig. 3. TPR profiles of CuO, Mn O , CuMn900, and CuMn900 after the
2
3
WGSR. TPR conditions: heating rate, 10 K/min in 10% H /Ar. WGSR
2
conditions are the same as in Fig. 4.
reduction of CuO in the spinel lattice, and the higher tem-
perature region, to reduction of Mn2O3 to MnO via Mn3O4,
though it is hard to define a clear boundary between the two
regions. As seen in Fig. 2, there was little difference between
CuMn500 and CuMn700. The amounts of CuO in CuMn900
and CuMn1100 reduced between 300 and 380 ◦C were much
larger than those in CuMn500 and CuMn700. This suggests
that above 900 ◦C the Cu–Mn spinel oxide is calcined at a
higher temperature, the more reducible the spinel is than
CuMn500 and CuMn700. It is considered that decomposi-
tion or reduction of Cu–Mn spinel oxide to Cu1.5Mn1.5O4
and Mn3O4 above 900 ◦C induces microstructural changes.
The porous microstructure may enhance the reducibility of
Cu–Mn spinel oxide.
Fig. 4. Catalytic activity of Cu–Mn spinel oxides for WSGR calcined
at various temperatures: (") CuMn500, (2) CuMn700, (!) CuMn900,
(a) CuMn1100, (- - -) equilibrium conversion. Reaction conditions: H ,
2
.
−1
37.5%; CO, 1.25%; H O, 25.0%; CO , 12.5%; space velocity, 6400 h
2
2
that of CuMn900 above 250 ◦C. It is clear from this result
that the WGSR activity of Cu–Mn spinel oxide depended on
calcination temperature, and a higher temperature was fa-
vorable. After the WGSR, as shown in Fig. 5, the crystalline
phase of CuMnX was composed mainly of Cu and MnO.
Small amounts of Cu2O or CuO were also detected, which
were probably formed by oxidation in air during the XRD
measurements in air. In every sample, no peak ascribed to
spinel oxide was detected after the WGSR.
Table 2 summarizes Cu particle size of CuMnX after the
WGSR estimated by N2O titration. Reduction of CuMnX
calcined at lower temperatures tended to cause aggregation
of Cu particles, leading to low WGSR activity. Although the
Fig. 4 illustrates the WGSR activity of CuMnX as a func-
tion of reaction temperature. CuMn900 exhibited maximum
activity in the temperature range studied. WGSR activity in-
creased with an increase in calcination temperature up to
900 ◦C. CuMn1100 showed catalytic activity comparable to
Fig. 5. XRD patterns of CuMnX after the WGSR: (") Cu, (1) MnO, (Q) Cu O, (!) CuO. Reaction conditions for the WGSR are the same as in Fig. 4.
2

Y. Tanaka et al. / Journal of Catalysis 215 (2003) 271–278
275
Table 2
Properties of CuMnX catalysts after reaction
◦
Sample
Calcination temperature
Crystalline phase
CO conversion at 250 C
(%)
Cu particle size
(nm)
◦
( C)
CuMn500
CuMn700
CuMn900
CuMn1100
500
700
900
Cu, Cu O, MnO
2
CuO, MnO
8.6
13
81
84
44
42
51
Cu, Cu O, MnO
2
1100
Cu, MnO
79
CuMn500 and CuMn700, though CuMn700 showed higher
CO conversion than CuMn500 as displayed in Fig. 4.
It can be said, based on the WGSR data, and TPR and
XRD measurements, that by reduction of easily reducible
CuMn900 and CuMn1100, highly dispersed Cu species are
formed and contribute to high CO conversion at 225–250◦C.
3.2. Effect of Cu/Mn molar ratio
From the above results, the optimal calcination tempera-
ture of Cu–Mn spinel oxide was estimated to be 900 ◦C for
WGSR activity. In this section, calcination temperature is
fixed at 900 ◦C and the molar ratio of Cu to Mn is varied
from 1/8 to 2/1. In Table 3 properties of catalysts are sum-
marized. The catalysts are labeled CuAMnB, where the mo-
lar ratio Cu/Mn = A/B. The BET surface area of CuAMnB
was ca. 1 m2/g and tended to decrease with an increase
in Mn content. Cu1Mn1, composed of only Cu1.5Mn1.5O4
phase, had a relatively high surface area, probably because
the phase was stable at 900 ◦C. However, impurities such as
CuO, Mn2O3, and Mn3O4 were unstable, resulting in low
surface area. Therefore, the BET surface area of CuAMnB
depended on the content of Cu1.5Mn1.5O4 structure. This
is the reason why Cu1Mn1 exhibited the highest BET sur-
face area among CuAMnB catalysts. The Cu/Mn ratios,
which were determined by ICP emission spectroscopy, al-
most agreed with the nominal ones.
Fig. 6. TPR profiles of CuMnX after the WGSR. TPR conditions: heating
rate, 10 K/min in 10% H /Ar. WGSR conditions are the same as in Fig. 4.
2
Cu particle sizes for CuMn700 were comparable to those
for CuMn900, catalytic activities for WGSR were quite
different, indicating that support strongly affected the active
site of Cu particles or WGSR which is discussed later.
To clarify the state of Cu particles, TPR was carried out.
Dow et al. have reported that α, β, and γ peaks existed at
ca. 160, 210, and 250 ◦C, respectively, over Cu/γ -Al2O3 or
Cu/YSZ catalyst [21]. The α and β peaks were ascribed
to reduction of highly dispersed Cu species which were
not detected by XRD analysis. The γ peak was ascribed
to reduction of bulky CuO called isolated Cu species. We
also have reported that two Cu species were observed over
Cu/ZnAl2O4 catalyst and one was a Cu dispersed on the
support, and the other is an isolated Cu species. According
to the report, CuO species dispersed on ZnAl2O4 were easily
reduced [22]. From those results, we can state that highly
dispersed CuOx is reduced at lower temperatures than are
bulky CuO species.
The crystalline phase of CuAMnB was composed mainly
of nonstoichiometric Cu1.5Mn1.5O4 and minor Mn2O3.
When [Cu]/[Mn] = 1/8, Mn3O4 (spinel structure) was the
predominant phase, while Cu1.4Mn1.6O4 was a minor phase.
This result clarified that the nonstoichiometric spinel phase
exists stably in the Cu–Mn system once copper manganese
is calcined above 600 ◦C.
Fig. 6 comprises TPR profiles of CuMnX after the
WGSR, and in Fig. 3 the TPR profiles for CuO, Mn2O3,
and CuMn900 are shown in order to assign the peaks in
Fig. 6. Only a single peak was detected that can be ascribed
to reduction of Cu2O or CuO in CuMn900 and CuMn1100.
In the case of CuMn500 and CuMn700, a shoulder peak was
detected at ca. 270 ◦C, which was ascribable to reduction
of relatively dispersed copper oxide species. The peak
temperature for the reduction of CuMn900 was the lowest
among the catalysts; i.e., the sequence was CuMn900 <
CuMn1100 < CuMn500 < CuMn700. It can be said that
copper species derived from CuMn900 and CuMn1100
were more highly dispersed on MnO than those from
Fig. 7 shows WGSR activity over CuAMnB as a func-
tion of temperature. Above 250 ◦C, equilibrium CO conver-
sion for the WGSR is attained over almost all the CuAMnB
catalysts. CO conversions increased with an increase in re-
action temperature, since the reaction rate limited CO con-
version at low temperatures. At 225 ◦C, the order of WGSR
activity was Cu1Mn2 ≈ Cu1Mn1 > Cu1Mn4 > Cu2Mn1 >
Cu1Mn8. At 200 ◦C, the order was Cu1Mn1 > Cu1Mn2 >
Cu2Mn1 > Cu1Mn4 > Cu1Mn8. The CO conversion at-
tained over the Cu1Mn2 catalyst was ca. 80%, even at
225 ◦C. Although Cu1Mn1 showed the highest CO conver-
sion at 200 ◦C among the catalysts, the value was not enough
for CO removal.

276
Y. Tanaka et al. / Journal of Catalysis 215 (2003) 271–278
Table 3
Properties of CuAMnB catalysts
a
Sample
Cu/Mn molar ratio
Crystalline phase
BET surface area
Cu particle size
(nm)
2
(m /g)
Cu2Mn1
Cu1Mn1
Cu1Mn2 (CuMn900)
Cu1Mn4
Cu1Mn8
2/1.6
1/0.85
0.5/0.43
0.25/0.23
0.125/0.11
Cu Mn O , CuO
0.9
1.4
0.8
0.7
0.6
184
76
42
42
53
1.4
1.6
4
Cu Mn
O
1.5
1.5
4
Cu Mn O , Mn O
1.5 1.5 4 3
4
3
4
Cu Mn O , Mn O
1.5
1.5
4
2
Mn O , Cu Mn
O
3
4
1.4
1.6
a
◦
Set value/value determined by ICP emission spectrometry. All samples were calcined at 900 C for 10 h.
Fig. 8. Temperature dependence of CO conversion for several initial
CO concentrations over CuMn900. Initial CO concentration: (") 1.25%,
Fig. 7. Catalytic activity of CuAMnB with different A/B ratios at
◦
200–250 C. Reaction conditions: H , 37.5%; CO, 1.25%; H O, 25.0%;
2
2
−1
(2) 2.50%, (Q) 5.00%, (!) 10%. Reaction conditions: H , 37.5%; CO,
CO , 12.5%; space velocity, 6400 h
.
2
2
−1
1.25–10.0%; H O, 25.0%; CO , 12.5%; space velocity, 6400 h
.
2
2
Cu particle sizes of the samples after the WGSR are listed
in Table 3. Cu particle size decreased with an increase in Mn
content except for Cu1Mn8. Since the high Cu content led to
a high probability of the existence of neighboring Cu atoms
in the spinel lattice, Cu species tended to form an agglom-
erate on reduction with H2. No obvious correlation could be
found between Cu particle size and WGSR activity. In the
particle size region 30–100 nm, where a morphological ef-
fect would not emerge, TOF tends to be constant and WGSR
depends on Cu loading.
It is inferred from the above results that the increase
in Cu loadings led to the increase in WGSR activity until
Cu/Mn = 1, but excess Cu loading brought about Cu
aggregation during reduction of Cu1.5Mn1.5O4 and Mn3O4
or Mn2O3 to Cu and MnO, which accompanied the decrease
in CO conversion. Therefore, the optimum Cu/Mn ratio is
considered to be ca. 1/2.
hydrocarbons have been proposed as a hydrogen source
for PEFCs because there exists an infrastructure, such as
petroleum stations and city gas pipelines. When hydrogen is
supplied by steam reforming of hydrocarbons, the reformed
gas contains much higher CO concentrations than reformed
methanol gas. The influence of initial CO concentration on
WGSR activity over CuMn900 was investigated.
As shown in Fig. 8, the increase in initial CO concen-
tration up to 5.00% led to the increase in CO conversion
in the measured temperature range. CO conversion reached
a the maximum at 250 ◦C in each case except for the ini-
tial CO concentration of 10%. CO conversion did not reach
equilibrium at 250 ◦C under the 10% CO condition. This
is, as Hutchings and co-workers reported for cobalt oxide-
supported copper catalysts, probably because competitive
adsorption of CO dominated over that of H2O on Cu sites
under the CO-rich condition. As a result, the relative abun-
dance of surface hydroxylgroups, which is believed to be the
key intermediate in oxidation of adsorbed CO, was limited.
And regeneration of the Cu sites can be hindered, which is
considered to lead to lower CO conversion than equilibrium
at 250 ◦C [23]. Fig. 9 illustrates the outlet CO concentra-
tions from CO conversions in Fig. 8. CO concentration was
3.3. Dependence on initial CO concentration
From the results in Sections 3.1 and 3.2, CuMn900
(Cu1Mn2) is the optimal catalyst for WGSR activity. We
have reported that CuMn900 maintains high and stable
WGSR activity in the short-term running test [15]. Recently,

Y. Tanaka et al. / Journal of Catalysis 215 (2003) 271–278
277
Fig. 10. Temperature dependence of WGSR activity over CuMn900
supplied with various concentrations of CO . Reaction conditions: CO ,
2
2
Fig. 9. Outlet CO concentration after the WGSR over CuMn900 supplied
with reformed gases at various CO concentrations. Reaction conditions:
(2) 0%, (Q) 6.25%, (") 12.5%; H , 37.5%; CO, 1.25%; H O, 25.0%;
2
2
−1
space velocity, 6400 h
.
H , 37.5%; CO, 1.25–10.0%; H O, 25.0%; CO , 12.5%; space velocity,
2
2
2
−1
6400 h
.
lowered to ca. 4000 ppm at 250 ◦C over CuMn900 for ini-
tial 5.00% CO gas. From this result, it could be said that
CuMn900 can be a candidate as a WGSR catalyst to remove
relatively high concentrations of CO in reformed hydrocar-
bons. However, preferential oxidation of CO is still needed
after the CO shift reaction to remove residual CO to the level
that PEFCs allow.
3.4. Effect of CO2 on WGSR activity
Since the water gas shift reaction is reversible, CO con-
versions are influenced by CO2 and H2. Although the WGSR
has been well investigated, there have been few surveys of
WGSRs in concentrated CO2 and H2. We have reported
that CO2 suppresses the WGSR over Cu/ZnO/Al2O3 cat-
alyst [24]. Fig. 10 shows the effect of coexistence of CO2 in
the feed gas. From the thermodynamic equilibrium, the CO
conversion decreases in particular at higher temperatures as
the initial CO2 increases. The experimental data agree well
with equilibrium conversions above 250 ◦C. At 200 ◦C, the
WGSR almost reached equilibrium when no CO2 was in-
cluded in the reformed gas and 99% CO conversion was ob-
tained, whereas the WGSR was suppressed when more than
6.25% CO2 was present.
Fig. 11. WGSR activity over CuMn900 after several reduction treatments.
Reduction was carried out by feeding 10% H /N prior to the WGSR.
2
2
Reaction conditions: H , 37.5%; CO, 1.25%; H O, 25.0%; CO , 12.5%;
2
2
2
−1
space velocity, 6400 h
.
3.5. Effect of reduction treatment against Cu–Mn spinel
oxide on WGSR activity
As discussed in Section 3.1, the Cu–Mn spinel oxide
catalyst is reduced in a hydrogen atmosphere or under the
reaction conditions. From Fig. 3, it is expected that the
reduction of Cu–Mn spinel oxide is facilitated at 350–
480 ◦C. We have reported that reduction at 400 ◦C degraded
the catalytic activity of Cu–Mn spinel oxide [15]. As
depicted in Fig. 11, reduction of Cu–Mn spinel oxide at
250 ◦C prior to the WGSR greatly enhanced CO conversions
below 225 ◦C, and the WGSR reached thermodynamic
equilibrium. It was found that CO conversion of Cu–
These results suggest that CO2 coexisting in the reformed
gas suppresses CO conversion at low temperatures, even
though higher CO conversion is expected from the thermo-
dynamics. It is inferred that the low desorption rate of CO2
or its high surface coverage at low temperatures brings about
low activity at 200 ◦C in the presence of CO2. It is clari-
fied that promotion of CO2 desorption at lower temperatures
is one of the keys in developing low temperature WGSR
catalysts. The material and surface modification of supports
would also affect desorption of CO2.

278
Y. Tanaka et al. / Journal of Catalysis 215 (2003) 271–278
Mn spinel oxide with reduction treatment at 250 ◦C is
comparable to that of conventional Cu/ZnO/Al2O3.
formed gas depressed CO conversion below 200 ◦C, while
CO conversion easily attained equilibrium at 200 ◦C in the
absence of CO2. The rate of CO2 desorption is considered
to be related to catalytic activity at low temperatures where
high CO conversion is attained thermodynamically. Mild
reduction treatment against Cu–Mn spinel oxide enhanced
WGSR activity.
Considering the effect of reduction temperature, it is
assumed that the mild reduction of Cu–Mn spinel oxide
results in high dispersion of Cu on the MnO support. This
is probably because Mn dispersed atomically near the Cu
sites in the spinel lattice can hinder Cu agglomeration under
mild reduction conditions. However, a clear correlation
between Cu particle size and WGSR activity was not
observed by N2O titration. This may be because MnO
may play a role in CO adsorption on the Cu/MnO surface
and promote the WGSR. Xu et al. have reported that
addition of MnO into Fe/silicalite catalyst enhances CO
adsorption capacity for CO and CO2 hydrogenation [25,
26]. For CO2 hydrogenation for hydrocarbon formation, a
two-step reaction mechanism involving a reversible WGSR
and a Fischer–Tropsch reaction has been suggested [27,28].
Trevin´o et al. have also proposed that formate may be formed
by adsorption of nondissociative CO on the MnO surface
of Mn-promoted zeolite Y-supported Rh catalyst for CO
hydrogenation [29]. Therefore, in our case, the MnO support
itself may contribute to the WGSR as an adsorption site of
CO and may be one of the reasons for high WGSR activity
despite low surface area.
Acknowledgments
This research was partially supported by Grants-in-Aid
for Scientific Research from the Ministry of Education,
and also by the New Energy and Industrial Technology
Development Organization (NEDO), Japan.
References
[1] J.N. Armor, Appl. Catal. 176 (1999) 159–162.
[2] P.G. Patil, J. Power Sources 37 (1992) 171.
[3] J.C. Amphlett, R.M. Baumert, R.F. Mann, B.A. Peppley, P.R. Roberge,
J. Electrochem. Soc. 142 (1995) 1.
[4] L. Clarkin, D. Epp, in: Proceedings of 2nd International Fuel Cell
Conference, 1996, p. 370.
[5] G.G. Schere, Solid State Ionics 94 (1997) 249.
[6] V.V. Galvita, G.L. Semin, V.D. Belyaev, T.M. Yurieva, V.A. Sobyanin,
Appl. Catal. 216 (2001) 85–90.
[7] S.H. Oh, R.M. Sinkevitch, J. Catal. 142 (1993) 254.
[8] V.M. Schmidt, P. Bröcherhoff, B. Höhlein, R. Menzer, U. Stimming,
J. Power Sources 49 (1994) 299.
4. Conclusions
[9] N. Hashimoto, H. Kudo, J. Adachi, M. Shinagawa, N. Yamada, A.
Igarashi, in: Proceedings of 32nd IECEC, 1997.
[10] C.T. Campbell, K.A. Daube, J. Catal. 104 (1987) 109.
[11] G.C. Chinchen, M.S. Spencer, Catal. Today 10 (1991) 293.
[12] A.J. Elliot, R.A. Hadden, J. Tabatabali, K.C. Waugh, F.W. Zemicael,
J. Catal. 157 (1995) 153.
[13] H.E. Curry-Hyde, M.S. Wainwright, D.J. Young, Appl. Catal. 77
(1991) 89.
[14] Y. Tanaka, T. Utaka, R. Kikuchi, K. Sasaki, K. Eguchi, Appl. Ca-
tal. 238 (2003) 11–18.
[15] Y. Tanaka, T. Utaka, R. Kikuchi, K. Sasaki, K. Eguchi, Appl.
Catal. 242 (2003) 285–293.
[16] S. Veprek, D.L. Cocke, S. Kehl, H.R. Oswald, J. Catal. 100 (1986)
250.
[17] P. Porta, G. Moretti, M. Musicanti, A. Nardella, Solid State Ionics 63–
65 (1993) 257–267.
[18] G. Fierro, S. Morpurgo, M.L. Jacono, M. Inversi, I. Pettiti, Appl.
Catal. 166 (1998) 407–417.
[19] G.J. Hutchings, A.A. Mirzaei, R.W. Joyner, M.R.H. Siddiqui,
S.H. Taylor, Appl. Catal. 166 (1998) 149–150.
[20] CRC Handbook of Chemistry and Physics, 74th ed., CRC Press, Boca
Raton, FL, 1993–1994.
[21] W.-P. Dow, Y.-P. Wang, T.-J. Huang, J. Catal. 160 (1996) 157–163.
[22] T. Takeguchi, Y. Kani, M. Inoue, K. Eguchi, Catal. Lett. 83 (2002) 50.
[23] G.J. Hutchings, R.G. Copperthwaite, F.M. Gottschalk, R. Hunter, J.
Mellor, S.W. Orchard, T. Sangiorgio, J. Catal. 137 (1992) 421.
[24] K. Sekizawa, S. Yano, K. Eguchi, H. Arai, Appl. Catal. 169 (1998)
291–297.
The preparation of Cu/MnO from Cu–Mn spinel oxides
and the concentrations of gas species significantly affected
the catalytic activity for the water gas shift reaction. Calci-
nation of Cu–Mn spinel oxide at high temperature was fa-
vored for promoting the WGSR despite the low BET sur-
face area. The crystallinity of Cu–Mn spinel oxides was at a
maximum when the calcination temperature was at 700 ◦C.
Decomposition or reduction of spinel to Cu1.5Mn1.5O4 and
Mn3O4 is considered to have occurred at 900 and 1100 ◦C.
Cu–Mn spinel oxides calcined above 900 ◦C showed higher
reducibility than those calcined at 500 and 700 ◦C in TPR
measurements. Those results suggest that easily reducible
Cu species were highly dispersed and contributed to high
CO conversion at low temperatures. This characteristic mi-
crostructure was verified by XRD and TPR measurements.
The Cu/Mn ratio of Cu–Mn spinel oxide affected the
Cu particle size of Cu/MnO, but the particle size was not
directly related to WGSR activity. The optimum Cu/Mn
ratio was 1/2. This is probably because in the particle
size region 30–100 nm, TOF tends to be constant and
WGSR depends on Cu loading. It was confirmed that
nonstoichiometric Cu1.5Mn1.5O4 existed stably.
The Cu–Mn spinel oxide in optimized composition
showed excellent CO conversion even when a large amount
of CO was present, which strongly suggests the potential
possibility of this catalyst system for CO removal in re-
formed hydrocarbon gas. It is expected that MnO species
promote the WGSR by adsorbing CO. CO2 coexisting in re-
[25] L.Y. Xu, Q.X. Wang, Y.D. Xu, J.S. Huang, Catal. Lett. 25 (1994) 177.
[26] L. Xu, Q. Wang, D. Liang, L. Lin, W. Cui, Y. Xu, Appl. Catal. 173
(1998) 22–25.
[27] M.D. Lee, J.F. Lee, C.S. Chang, T.Y. Dong, Appl. Catal. 72 (1991)
267.
[28] K. Fujimoto, T. Shikada, Appl. Catal. 31 (1987) 13.
[29] H. Trevin´o, G.D. Lei, W.M.H. Sachtler, J. Catal. 143 (1995) 251.