Hydrogen production using Ni-Rh on ZrO2 as potential low-temperature catalysts for membrane reactors

Article abstract of DOI:10.1006/jcat.2002.3687

The dry reforming of CH₄ is an attractive route to produce H₂ from natural gas. The two most wanted features of the catalysts for this process are low-carbon formation and good stability at low reaction temperatures. The CO₂

Full text of DOI:10.1006/jcat.2002.3687

Journal of Catalysis 210, 263–272 (2002)
doi:10.1006/jcat.2002.3687
Hydrogen Production Using Ni–Rh on ZrO2 as Potential
Low-Temperature Catalysts for Membrane Reactors
S. Irusta, L. M. Cornaglia, and E. A. Lombardo1
Instituto de Investigaciones en Cata´lisis y Petroqu´ımica (FIQ, UNL-CONICET), Santiago del Estero, 2829-3000 Santa Fe, Argentina
Received November 7, 2001; revised May 9, 2002; accepted May 30, 2002
use of ZrO2 has a beneficial effect particularly in the case
of metals such as Pt, which is not good for promoting CO2
dissociation (7). The high stability of the Pt/ZrO2 catalysts
has been connected to its ability to activate CO2 (8). This
feature is in turn assigned to the Pt atoms located at the
Pt/ZrO2 interface (9).
In the first part of this project, we investigated the behav-
ior of Ni, Rh, and Ni–Rh formulations supported on La2O3.
In that case, the most active and cleaner catalysts were those
containing only Rh. In this case, in our search to develop
catalysts suitable for membrane reactors, we have prepared
and characterized (XRD, TPR, XPS, hydrogen chemisorp-
tion) several Ni–Rh supported on ZrO2 solids. The nature
and amount of the carbon deposits was further investigated
through Raman spectroscopy, SEM, TGA, DSC, and TPH.
The CO2 reforming of CH4 over Ni–Rh mono- and bimetal-
lic catalysts supported on ZrO2 was investigated. The catalysts
were prepared by wet impregnation with 0.2 wt% Rh and 2 wt%
Ni. The solids were calcined at 823 K in air flow and were re-
duced at 973 K in flowing hydrogen before being tested in a fixed-
bed reactor. The results showed that the bimetallic solid had the
highest activity at 823 K but monometallic Rh was the only sta-
ble catalyst. The catalysts were characterized through XRD, XPS,
and TPR. TGA, DSC, TPH, SEM, and Raman spectroscopy were
used in order to characterize the carbonaceous deposits. Carbon
deposits with different reactivity were detected. A low-carbon de-
position was observed on Rh (0.2%)/ZrO2which exhibited a stable
activity after 100 h on stream. The results suggest a significant
metal–support interaction and a strong metal–metal interaction in
c
ꢀ 2002 Elsevier Science (USA)
the bimetallic solid.
Key Words: hydrogen; CO2 reforming; Ni–Rh catalysts.
EXPERIMENTAL
INTRODUCTION
Catalyst Preparation
The dry reforming of methane is an attractive route to
produce H2 from natural gas. This is even more appealing
if a membrane reactor can be used to produce high-purity
H2 (1).
Catalysts were prepared by conventional wet impreg-
nation of ZrO2 (Degussa) using Ni(NO3)2 · 6H2O and
RhCl3 · 3H2O as precursor compounds. The bimetallic solid
was prepared by simultaneous impregnation. In all cases,
the resulting suspension was then heated at 353 K to evap-
orate the water, and the solid material was dried in an oven
at 383 K overnight. The resulting catalysts were calcined for
6 h at 823 K at a heating rate of 1.8 K/min. Both the Ni and
Rh loadings were the same as those used when La2O3 was
the support (10).
The two most wanted features of the candidate catalysts
for this process are low-carbon formation and good stabil-
ity at low reaction temperatures. Both the metal and the
support used are key players in the development of these
features. Wang and Ruckenstein (2) found that both the
activity and stability of Rh formulations are strongly de-
pendent on the support. On the other hand, Bitter et al. (3)
stated that the activity of Rh-containing catalysts is only a
function of the metal exposed and is not affected by the
support. Ni has been supported over different solids (4).
In all cases, the support plays a key role in coke deactiva-
tion and metal sintering (5). Montoya et al. (6) studied the
catalytic behavior of Ni supported on Ce-modified ZrO2.
It showed good activity but the amount of carbonaceous
deposits was higher in Ni/ZrO2–CeO2than in Ni/ZrO2. The
Catalyst Testing
The catalyst (50 or 300 mg) was loaded into a tubular
quartz reactor (inner diameter, 5 mm) which was placed in
an electric oven. A thermocouple in a quartz sleeve was
placed on top of the catalyst bed. The catalysts were heated
in He at 973 K and then reduced in situ in H2 at the same
temperature for 0.5 h. After reduction, the temperature
was adjusted in flowing He to the reaction temperature
(823–973 K), and the feed gas mixture (33% (v/v) CH4,
33% CO2, 34% He, P = 1 atm) was switched to the reactor.
1 To whom correspondence should be addressed. E-mail: nfisico@fiqus.
unl.edu.ar.
263
0021-9517/02 $35.00
c
ꢀ 2002 Elsevier Science (USA)
All rights reserved.

264
IRUSTA, CORNAGLIA, AND LOMBARDO
The stability tests were carried out at W/F = 2.67 × 10−5
g
Thermal analysis experiments were carried out in a
h ml−1, and conversion was measured at W/F = 4.5 × Mettler Toledo DSC 821e in order to find out the exother-
10−6 g h ml−1. The reaction products were analyzed in mic or endothermic nature of the changes that take place
a TCD gas chromatograph (Shimadzu GC-8A) equipped during the calcination of the used solids.
with a Porapak column and a Molecular Sieve column.
Temperature-Programmed Hydrogenation
Surface Area and XRD
TPH experiments were carried out over used catalysts,
For the determination of the surface area of calcined
and used solids, a Quantachrome Sorptometer, Nova
1000model, wasemployed. PriortotheBETmeasurements
the samples were maintained at 473 K under vacuum of
10−3 Torr for 2 h.
and the solids (20 mg) were heated from 300 to 1073 K at
8 K min−1 under H2 with a flow rate of 20 ml min−1. The
temperature was then kept constant for 1 h, and gases were
analyzed by online mass spectrometry.
The X-ray diffraction (XRD) patterns of the calcined
and used solids were obtained with an XD-D1 Shimadzu
instrument, using Cu Kα radiation at 35 kV and 40 mA. The
scan rate was 1◦ min−1 for values between 2θ = 10 and 80◦.
Laser Raman Spectroscopy
The Raman spectra were recorded with a TRS-600-SZ-
P Jasco Laser Raman instrument, equipped with a CCD
(charge coupled device) with the detector cooled to about
153 K using liquid N2. The excitation source was the
514.5-nm line of a Spectra 9000 Photometrics Ar ion laser.
The laser power was set at 30 mW.
X-Ray Photoelectron Spectroscopy
The XPS measurements were carried out using an
ESCA750 Shimadzu electron spectrometer. Nonmono-
chromatic Al Kα X-ray radiation was used. The anode was
operated at 8 kV and 30 mA and the pressure in the analysis
chamber was about 2 × 10−6 Pa.
RESULTS
Catalyst Characterization
The binding energies (B.E.) were referred to the C 1s
signal (284.6 eV). Curve fitting was performed using a
Levenberg–Marquardt NLLSCF routine. The background
contribution was taken into account by assuming an
integral-type background. The surface Rh/Zr and Ni/Zr
atomic ratios were calculated using the areas under the Rh
3d, Ni 2p3/2, and Zr 3d peaks, the Scotfield photoionization
cross sections, the mean free paths of the electrons, and
the instrumental function, which was given by the ESCA
manufacturer.
Table 1 shows that the surface area of the ZrO2 does not
change after metal incorporation and further calcinations.
The Rh dispersion data are also shown in Table 1. Note that
the Ni solid after evacuation at 573 K does not adsorb H2
at all.
Figure 1 shows the TPR traces for the Ni-Rh/ZrO2 cata-
lysts. The monometallic solids exhibit two reduction peaks.
In the case of the Rh (0.2%) sample, the peaks’ maxima are
at 400 and 500 K, corresponding to easily reducible rhodium
oxide. The TPR profile of our Ni/SiO2 presents only one
peak, at 629 K, indicating that there is no significant metal–
support interaction. Xu et al. (11) reported TPR results for
Ni-based catalysts supported on several oxides. The TPR
of those high loading catalysts (∼12 Ni wt%) showed one
reduction peak (660–700 K) coincident with the reduction
of unsupported nickel oxide. The Ni (2%) reduction profile
Metal Dispersion
The Rh dispersion of fresh catalysts, following the H2 re-
duction at 973 K for 0.5 h, was determined by static equilib-
rium H2 adsorption at room temperature in a conventional
vacuum system. As a standard procedure, prior to adsorp-
tion, the catalysts (300 mg) were evacuated (10−5 Torr) at
573 K.
TABLE 1
Temperature-Programmed Reduction
Fresh Catalyst Features
An Ohkura TP-20022S instrument equipped with a ther-
mal conductivity detector (TCD) was used for the TPR
experiments. The 100-mg samples were reduced in a 5%
Sa
TPRc
H2/metal
Catalyst
Rh (0.2%)
Ni (2%)
Ni (2%)–Rh (0.2%)
(m2 g−1
)
H/Rhb
0.45
H2–Ar stream, at a heating rate of 10 K min−1
.
31.9
30.0
29.7
0.98
0.91
0.84
d
—
Thermogravimetric Analysis
0.30
The amount of carbon on the used catalysts was de-
termined by oxidizing the carbon in a Mettler Toledo
TGA/SDTA 851. The used catalysts (usually 10 mg) were
heated at 10 K min−1 to 1173 K in a flow of 90 ml min−1 air.
a
BET surface area of calcined solids. The ZrO2 support has 30 m2 g−1
Measured by H2 chemisorption.
.
b
c
d
Hydrogen consumption; 5% H2 in Ar, 10 K min−1
This solid does not adsorb H2.
.

LOW-TEMPERATURE Ni–Rh CATALYSTS ON ZrO
265
2
40
50
40
30
20
10
0
Ni(2%)
30
20
Ni(2%)Rh(0.2%)
Rh(0.2%)
10
Ni(2%)
Rh(0.2%)
Ni(2%)Rh(0.2%)
300
400
500
600
700
800
900
1000
0
0
1
2
3
4
0
1
2
3
4
Temperature(K)
W/F (mol h g-1) x 104
W/F (mol h g-1) x 104
FIG. 1. TPR profiles of calcined Ni–Rh catalysts (5% H2 in Ar, 30 ml
min−1, 10 K min−1).
FIG. 2. Catalytic behavior of Ni–Rh solids at 823 K, 33% (v/v) CH4,
33% CO2, 34% He, P = 1 atm.
of our solid reveals two peaks, at 618 and 760 K. The low-
temperature peak may be assigned to well-dispersed nickel catalyst, while the other two formulations do deactivate.
oxide in the support while the highest temperature peak is Note, however, that both TGA and TPH measurements
symptomatic of a strong metal support interaction. One show that no carbon deposition has occurred on the cata-
possibility is the formation of a mixed oxide. However, lysts kept on stream up to 100 h at 823 K.
the Raman spectra and XRD patterns of Ni/ZrO2 catalysts
XPS Data of Fresh and Used Ni–Rh Catalysts
show no presence of Ni phases.
The TPR profile of the bimetallic solid is more complex,
probably due to additional Ni-Rh interaction. The total H2
consumption is reported in Table 1.
The peak positions and FWHM of the Rh 3d5/2, Ni 2p3/2
,
and Zr 3d5/2 are summarized in Table 3. In the monometal-
lic Rh solid, the Rh 3d5/2 binding energy did not change
after 48 h on stream and only a slight peak broadening was
observed.
Stability and Reaction Rates of Ni–Rh/ZrO2 Catalysts
In order to compare the catalytic performance of the
Surface atomic ratios for the fresh and used samples are
solids at low temperature, initial reaction rates of methane given in Table 4. In the case of the fresh Rh (0.2%) solid, the
and carbon dioxide were calculated (Table 2) from CH4 Rh/Zr atomic ratio is 0.012. This ratio remains constant af-
and CO2conversions vs W/F curves (Fig. 2) determined at ter reaction at temperatures between 823 and 973 K. For the
823 K. The bimetallic catalyst is the most active and the Ni bimetallic sample, the high B.E. for Rh 3d5/2 (308.9 eV) and
(2%) had the worst performance. The turnover frequency Ni 2p3/2 (856.3 eV) indicate the presence of oxidized sur-
values (H/Rh from Table 1 were used to calculate TOFs) of face species (12–15). In this sample, the Rh/Zr and Ni/Zr are
the catalysts are also shown in Table 2.
0.022 and 0.08, respectively. However, after 48 h on stream
As clearly shown in Fig. 3, over a testing period of 100 h, under the same reaction conditions, no signal from Rh and
there is practically no loss of activity over the Rh (0.2%) Ni was detected. A large signal from carbon (283.2 eV) was
TABLE 2
Catalytic Behavior of ZrO2-Supported Catalysts
b
b
c
c
XCH
XCO
RCH
RCO
TOFCH
TOFCO
2
4
2
4
2
4
Catalysta
Rh (0.2%)
Ni (2%)
Ni (2%)–Rh (0.2%)
(76.6)
(84.9)
(mol g−1 h−1
)
(mol g−1 h−1
)
(s−1
)
(s−1
)
83.1
65.9
74.7
87.3
67.9
78.9
0.32
0.09
0.47
0.50
0.24
0.67
11.0
—
21.5
16.5
—
31.3
a Solids were reduced in situ at 973 K before reaction. Weight percent of transition metals is in parentheses.
b Initial percentage conversions measured at reaction temperature = 973 K, W/F = 1.05 × 10−5 g h ml−1; total reaction time 48 h.
Equilibrium values, in parentheses, were calculated using a CH4 + CO2 reaction and the reverse WGSR.
c Initial reaction rates measured at 823 K (conversion values, <10%).

266
IRUSTA, CORNAGLIA, AND LOMBARDO
20
whole temperature range investigated when appropriate
30
25
20
15
10
5
space velocities were used over Rh catalysts, as shown in
Table 2. Table 5 shows the surface area of used catalysts
after being on stream for 48 h at temperatures up to 973 K.
The space velocities were low in order to achieve near-
equilibrium conversion all along these runs. Note that nei-
ther monometallic catalyst shows a change in surface area
(see also Table 1) beyond the experimental error. However,
a large increase in surface area is observed in the bimetallic
solid.
Figure 4 shows the diffractograms of the used catalysts
and the ZrO2 support. In the latter, both the tetragonal
(2θ = 30.1◦) and the monoclinic phases coexist. In the used
catalysts, the main reflection of the tetragonal phase sharply
decreases in intensity, as expected, since this structure is less
stable at higher temperatures. Note that the Ni-containing
solids show a wide reflection at 26 which can be assigned to
graphitic carbon (16). This peak does not appear in the Rh
(0.2%) solid.
15
10
5
0
0
0
20
40
60
80
100
0
20
40
60
80
100
Time (h)
Time (h)
FIG. 3. Stability test carried out at W/F = 2.67 × 10−5 g h ml−1
,
T = 823 K, 33% (v/v) CH4, 33% CO2, 34% He, P = 1 atm; conversion
measurements at W/F = 4.5 × 10−6 g h ml−1
.
TGA was used to quantify the amount of carbon deposits
on the three catalysts. The TGA profiles are shown in Fig. 5,
and the amount of carbon burnt is shown in Table 5. The
Rh (0.2%) catalyst only shows a small mass loss at 587 K.
Note that this peak also appears in the bimetallic solid.
The Ni-containing formulations exhibit similar TGA pro-
files (Fig. 5). The only difference is the lower temperatures
at which the carbon is burnt in the bimetallic catalyst. This
lower temperature may be due to either a catalytic effect
of Rh during carbon burning or different reactivities of the
carbon residues. The DSC data (not shown) confirm that
all the mass changes detected in the TGA experiments are
exothermic processes that are likely to be the burning of
the carbonaceous residues.
Another way to study the reactivity of residues is to use
TPH (Fig. 6). Three peaks of methane (m/e = 16) were de-
tected on the Ni (2%) catalyst, at 573, 800, and 990 K. These
may be assigned to either different types of carbon deposits
or deposits formed on different locations (support, metal,
or metal/support interface). The used bimetallic catalyst
only exhibits two peaks while the Rh (0.2%) only shows a
observed, assigned to residues formed during the reaction.
Similar results were obtained for Ni (2%) supported on
ZrO2. The atomic C/Zr ratio was lower for this solid com-
pared to the bimetallic one. The C 1s spectrum of the used
Rh(0.2%)sampleshowedaveryweakcarbondepositpeak.
Long-term stability tests caused the Rh/Zr and Ni/Zr
atomic ratios to decrease to 0.013 and 0.053 for the Ni
(2%)–Rh (0.2%) sample. No carbon peak at 283.2 eV was
detected. These results are in agreement with the data ob-
tained from TGA and Raman spectroscopy (vide infra).
For this solid, a significant broadening was shown in the Rh
3d5/2 peak. This could be due to a partial reoxidation of the
sample after exposure to ambient conditions. This explana-
tion is supported by the presence of a low binding energy
peak in the Ni 2p3/2 spectrum.
Carbon Deposits after Catalytic Tests
in Equilibrium Conditions
The equilibrium conversion for the CO2 reforming and
the water–gas shift reaction were closely approached in the
TABLE 3
Binding Energies of Ni–Rh Catalystsa
Calcined
Rh 3d5/2
Used (48 h)
Used (100h)
Rh 3d5/2
Catalyst
Rh (0.2%)
Zr 3d5/2
Ni 2p3/2
Zr 3d5/2
Rh 3d5/2
Ni 2p3/2
Zr 3d5/2
Ni 2p3/2
182.2
(1.7)
182.2
(1.7)
182.2
308.2
(2.4)
308.9
(2.8)
—
—
182.2
(1.9)
182.2
308.4
(2.9)
ndb
—
—
—
—
Ni (2%)–Rh (0.2%)
856.3
(3.0)
856.0
(3.2)
nd
nd
182.2
(2.0)
182.2
(2.0)
308.0
(3.6)
—
856.2 (2.7)
853.9 (3.7)
nd
Ni (2%)
182.2
—
a Contamination carbon was taken as reference at 284.6 eV. FWHM values are shown in parentheses.
b nd, not detected.

LOW-TEMPERATURE Ni–Rh CATALYSTS ON ZrO
267
2
TABLE 4
Surface Atomic Ratios of Ni–Rh Catalysts
Calcined
(Ni/Zr)s
Used (48 h)a
Used (100h)b
c
Catalyst
Rh (0.2%)
Ni (2%)–Rh (0.2%)
Ni (2%)
(Rh/Zr)s
(O/Zr)s
(C/Zr)s
(Rh/Zr)s
(Ni/Zr)s
(O/Zr)s
(C/Zr)s
(Rh/Zr)s
(Ni/Zr)s
(O/Zr)s
0.012
0.022
—
—
0.08
0.059
2.6
2.4
2.3
0.3
12.0
9.1
0.017
ndd
—
—
nd
nd
2.2
3.2
4.0
nme
0
—
nm
0.013
—
nm
0.053
—
nm
2.6
—
a Reaction temperature, 823–973 K.
b Reaction temperature, 823 K.
c Carbon at 283.2 eV. Contamination carbon was taken as reference at 284.6 eV. No carbon at 283.2 eV was observed on the calcined samples.
d nd, not detected.
e nm, not measured.
small methane peak. To check the extent of hydrogenation posit completely covers the catalyst particles. On the other
of the residues, a second TGA was performed in air after hand, the SEM micrograph of Rh (0.2%) does not reveal
the TPH experiment. In the TGA profile, two peaks are the presence of carbonaceous residues which were, how-
observable at temperatures similar to the former ones. By ever, detected by the other techniques (TGA, TPH, and
comparing the mg C/g catalyst ratios, it is concluded that LRS), although in tiny amounts (Table 5).
the residues most difficult to oxidize were not significantly
reduced. The second most difficult to oxidize deposits were
DISCUSSION
roughly 50% consumed during the TPH experiments. The
easiest to oxidize residue was completely wiped out by the
reducing stream.
This study originated in the need to find stable catalysts
with the lowest carbon deposition level able to operate
at 823–873 K, for use in membrane reactors. For obvious
All the Raman spectra (Fig. 7) of the used catalysts in-
variably showed two peaks, at 1350 and 1580 cm−1, which
are assigned to graphitic carbon (17). The signal intensity
in the Rh (0.2%) catalyst was very weak compared to the
other two solids. Furthermore, in the Ni-containing solids
no signals from ZrO2 were detected, even though several
pellet areas were scanned. This is consistent with the pres-
ence of significant amounts of carbonaceous deposits which
completely cover the catalyst particles.
The SEM micrographs clearly show the presence or car-
bon filaments on top of both the Ni (2%) and Ni (2%)–Rh
(0.2%) catalysts (Fig. 8). In the latter case, the carbon de-
TABLE 5
Amount of Carbonaceous Residues Formed on Used Catalysts
TGAb
TGAc
Sa
d
d
Catalyst
(m2 g−1) mgcarbon/gcat. Tp (K) mgcarbon/gcat. Tp (K)
Rh (0.2%)
Ni (2%)
29.4
32.4
10
245
95
6
290
135
587
810
940
585
770
900
—
119
82
—
134
112
—
805
963
—
773
929
Ni (2%)–Rh (0.2%) 45.1
a BET surface area of used catalysts.
b Time on stream, 48 h; reaction temperature, between 823 and 973 K;
W/F = 1.07 × 10−5 g h ml−1
.
c Same as above, after TPH experiments.
d Peak temperature.
FIG. 4. DRX patterns of used Ni–Rh catalysts after 48 h on stream
(see Table 5) and support. Scan rate: 1 min−1
.

268
IRUSTA, CORNAGLIA, AND LOMBARDO
Rh(0.2%)
Ni(2%)
Ni(2%)
Ni(2%)Rh(0.2%)
Rh(0.2%)
400
500
600
700
800
900
1000
Temperature (K)
Ni(2%)Rh(0.2%)
FIG. 5. TGA profiles of used solids. Time on stream 48 h; temperature
range, 823–973 K; W/F = 1.07 × 10−5 g h ml−1
.
1200 1300 1400 1500 1600 1700
reasons most of the published studies to date have been
carried out at T > 873 K. In addition, scarce data are avail-
able at any temperature for Rh on ZrO2 systems. The data
presented and the following discussion will be useful for vi-
sualizing the similarities and differences observed between
low- and high-temperature operation.
Wavenumber (cm-1)
FIG. 7. Raman spectra of Ni–Rh catalysts used 48 h in reforming re-
action. Temperature range, 823–943 K; W/F = 1.07 × 10−5 g h ml−1
.
Carbon deposits invariably appear on our catalysts as-
sayed near equilibrium conditions at 823–973 K (Table 5).
Large amounts of carbonaceous material are formed on
both Ni (2%) and Ni (2%)–Rh (0.2%) catalysts. The SEM
micrographs show the abundant formation of filaments
(Fig. 8). Both DRX and LRS indicate the presence of
graphitic carbon (Figs. 4 and 7). The appearance of peaks
in both the TGA (Fig. 5) and TPH (Fig. 6) profiles is symp-
tomatic of either the presence of two carbon species with
different reactivity or carbon deposited on different surface
sites, or both. In Pt/Cex Zr1−x O2, Noronha et al. (7) found
that the different peaks in the TPO profile (similar to ours)
are not due to different forms of carbon but rather to differ-
ent locations on the catalyst surface. The low-temperature
peak was assigned to carbon near the metal particles, and
the high-temperature peak to carbon deposited on the
support.
Carbon Deposits
Carbon deposition is a fundamental problem in the car-
bon dioxide reforming reaction. The origin of the inactive
carbon can be methane decomposition and CO dispropor-
tion (4). The carbon formed is frequently in the form of
filamentous whiskers (18), but the most deactivating one is
the encapsulating type (19). In Rh catalysts supported over
Y-stabilizedZrO2 averylowdeactivationratewasobserved
(20), while in the Ce-doped Ni catalysts over ZrO2 (6) the
coke formation did not affect the catalytic activity.
Ni(2%)Rh(0.2%)
In agreement with what has been reported by Nagaoka
et al. (8) the TGA experiments, performed after the TPH,
show that the carbon which oxidizes at higher temperature
cannot be removed by reaction with hydrogen up to 1173 K
(Table 5). The carbonaceous deposits formed in our Ni-
Rh catalysts could be deposited on the metals and on the
support. The deposits on the support would require tem-
peratures over 1173 K to react with H2.
The monometallic Rh catalyst exhibits a very-low-carbon
deposition. For this solid, the XPS surface C/Rh ratio was
equal to 15 (Table 4). No changes were observed in the
surface Rh/Zr ratio after 48 h on stream, suggesting that
Ni(2%)
Rh(0.2%)(x 10)
400
600
800
1000
T = 1173 K
Temperature (K)
FIG. 6. TPH profiles of used solids. Time on stream 48 h; 823–973 K;
W/F = 1.07 × 10−5 g h ml−1
.

LOW-TEMPERATURE Ni–Rh CATALYSTS ON ZrO
269
2
FIG. 8. SEM micrographs of Ni (2%) and Ni (2%)–Rh (0.2%) after 48 h on stream at W/F = 1.07 × 10−5 g h ml−1; temperature range, 823–973 K.

270
IRUSTA, CORNAGLIA, AND LOMBARDO
no sintering occurred due to the high reaction tempe- we will make frequent reference to the related system of
rature.
Pt/ZrO2. For the latter system, a number of studies (3, 4,
7) have been published. They are mainly aimed at under-
standing the nature of the activity and stability behavior of
these solids. Ross and co-workers (21), using a TAP reactor
(temporal analysis of products), observed that on Pt/ZrO2
Stability and Reaction Rates of Ni–Rh/ZrO2 Catalysts
at Low Temperature
Bitter et al. (9) have suggested that the oxide supports both methane and carbon dioxide dissociate independently
whicharegoodoxygendonors, suchasZrO2, yieldexcellent of one another. The dissociation of carbon dioxide acts as
catalysts for the dry reforming of methane. When Rh is the an oxygen supplier, while the decomposition products of
active metal used in this reaction, most of the published methane scavenge the oxygen from the catalyst. According
work concerns the effect of several supports upon the cata- to this mechanism, CH4 decomposition would take place
lytic performance. The most common supports are ZrO2, on the metal (22), resulting in the production of H2 and the
TiO2, Al2O3, SiO2, and MgO (20).
formation of carbonaceous deposits. The role of the sup-
Zhang et al. (20), comparing the initial activity of Rh cata- port would be to adsorb CO2 and allow the dissociation at
lysts and the acidity of the carriers, stated that a correlation the metal–support interface.
between TOF and surface acidity seems to exist. It appears
In the Pt/ZrO2 system, Stagg et al. (23) found that when
that the acidic nature of the carrier may promote methane the rate of CH4 decomposition is initially greater than the
dissociation on the sites at the periphery of metal crystal- rate of CO2 dissociation, carbon deposition occurs and is re-
lites (metal–support interface). They found that the initial sponsibleforthedeactivationofthecatalyst. Thestabilityof
intrinsic activity and rate of deactivation of Rh decrease the catalysts depends on the balance between the two paths.
with increasing metal particle size. They maintained that
They proposed a cleansing mechanism where the carbon
the higher the dispersion, the higher the metal–support in- isremovedfromthemetalparticles. ThedissociationofCO2
terfacial area, which results in more intense metal–support leads to the formation of CO and adsorbed O, which can
interactions.
then combine with carbon on the metal to form additional
Table 6 shows the TOF of our Rh/ZrO2 and Rh/La2O3 CO. Another possibility is that the oxygen that reacts with
(10) catalysts together with data reported in the literature the carbon comes from the support.
for the same systems. The data show that both the specific
Using pulse experiments with 13C-labeled CH4, they
activity and the activity per mole of Rh of our Rh/ZrO2 found that Pt is needed to catalyze the dissociation of
catalyst is the highest of them all. Note that this catalyst CO2. Then, the dissociation can take place near the metal–
(Fig. 3) and the Rh/La2O3 (10) are stable after 100 h on support interface or can occur on oxygen vacancies gen-
stream at 823 K.
erated during the previous reduction of the support. Our
In the case of the Ni monometallic solid, low activity and monometallic Rh catalyst supported on ZrO2 is very stable
a high deactivation rate were observed (Table 2 and Fig. 3). and does not present carbon deposition. Considering that
Adding Rh to the Ni solid produced an active catalyst but CO2 dissociates on Rh surface (24), a similar cleansing reac-
the deactivation rate was increased. However, no carbon tion mechanism could be proposed for our stable Rh/ZrO2
formation was observed in these catalysts.
solid. Besides, the XPS Rh/Zr ratio did not change after 48 h
Since there are scarce data concerning the nature of the on stream (Table 4), suggesting that Rh dispersion remains
Rh/ZrO2 interaction for dry reforming, in what follows unchanged.
TABLE 6
Reaction Rates, TOF Values, and Deactivation of Rh Catalysts
a
a
RCH
RCO
TOF (s−1
)
4
2
b
D
(%)
Temp.
(K)
θreact
(h)
Deacc
(%)
Catalyst
mol h−1 g−1 mol h−1 g−1 Rh mol h−1 g−1 mol h−1 g−1 Rh CH4
CO2
Ref.
Rh (0.2%)/ZrO2
Rh (0.2%)/La2O3
Rh (0.6%)/La2O3
Rh (0.5%)/ZrO2
Rh (1.0%)/ZrO2
45d
64d
14d
90d
16e
823
823
823
875
873
0.32
0.20
0.26
—
160
100
43
—
5
0.50
0.42
0.51
0.34
—
250
210
85
68
—
11.0
4.4
13.8
—
16.5
9.1
16.7
2.2
100
100
100
70
0
0
0
0
0
This work
10
10
6
0.05
0.9
—
48
11
a Initial reaction rates, calculated from data given by the authors.
b Time on stream.
c Deactivation.
d Dispersion measured by H2 chemisorption.
e Dispersion measured by CO chemisorption.

LOW-TEMPERATURE Ni–Rh CATALYSTS ON ZrO
271
2
Van Keule et al. (21) maintained that the oxygen pool be the result of an interplay between CH4 dissociation and
used for the reactions described is present in the vicinity of CO2 decomposition. The former is favored by higher tem-
the Pt crystallites. They emphasized that ZrO2 can be par- peratures and perhaps the CO2 decomposition does not
tially reduced, leading to the formation of Pt–Zr surface proceed fast enough to burn the carbonaceous residues left
alloys. This surface alloy helps to maintain a high Pt disper- on the surface. An alternative view is that the ZrO2 support
sion. Since Rh–Zr alloys do exist (25), the same beneficial could provide lattice oxygen to clean the surface at a slow
effect could be present in the Rh/ZrO2 catalyst.
rate at high temperature. In any case, the experimental data
ZrO2 is well-known for its ability to interact strongly with indicate that CO disproportionation is not responsible for
the metal component and, therefore, to stabilize the metal- carbon formation on the Ni–Rh catalysts.
lic Rh cluster (15). In the Ni–Rh bimetallic solid the TPR
Previously, we used Ni–Rh/La2O3 catalysts for dry re-
profile (Fig. 1) suggests the presence of a strong interaction forming of methane (10). We found that the formation of
between the metals, and the low-binding-energy Ni 2p3/2 carbonaceous residues does not affect the catalytic activity
peak (Table 3) indicates the formation of stable reduced of these formulations. We have also detected strong metal–
species in the catalysts after 100 h on stream. These species support interaction associated with the high specific activ-
did not reoxidize after being exposed to air. The results ity and stability of the Ni–Rh/La2O3 formulations. Table 6,
of both techniques are consistent with the formation of a which includes the activity data for the La2O3 catalysts,
Ni–Rh surface alloy (12). As no carbon deposition has oc- shows that Rh/ZrO2 is the most active catalyst, as stable as
curred on this solid when kept on stream for up to 100 h at those supported on La2O3 (100 h on stream).
823 K, one explanation for the deactivation observed in the
Ni–Rh solid could be that the presence of Ni decreases the
interaction between the Rh and the support at the particle–
support interface. A similar effect was previously reported
by Stagg et al. (23) in Pt–Sn/ZrO2 catalysts.
Bradford and Vannice (26) have reported that the pres-
ence of a TiOx overlayer on the Pt surface reduces the
number of large ensembles of Pt atoms and geometrically
inhibits CO dissociation and CH4 decomposition. This is
consistent with Zhang et al. (20). They attributed the high
activity and stability of Ni/La2O3 catalysts to LaOx species
on the Ni surface.
Considering available literature data, Bradford and
Vannice (26) assigned the higher stability and coking re-
sistance of Pt/ZrO2 to strong Pt–Zrn+ interactions which
block active sites for carbon deposition and results in the
formation of ZrOx species on the Pt surface. Roberts and
Gorte (27) indicated that ZrO2 influences the Pt surface via
a strong Zrn+–Pt interaction, altering particle morphology
and lowering the CO desorption barrier.
Comparing the XANES results with TPR, Bitter et al.
(3) assigned the high-reduction-temperature peak at 632 K
(from TPR) to the partial reduction of ZrO2. They specu-
lated that hydrogen is activated on Pt and partially reduces
ZrO2 in its direct vicinity. In our TPR profiles for the Ni–
Rh/ZrO2 catalyst, the reduction temperature peak at 632 K
was not observed; therefore, no partial reduction of the sup-
port seems to occur.
CONCLUSIONS
Rh (0.2%)/ZrO2 is the most active, stable catalyst for
CO2 reforming at 823 K. Only small amounts of carbona-
ceous residues (TGA, XPS) were formed after 48 h on
stream at temperatures up to 973 K. No residues were de-
tectedat823Kafter100honstream. Therefore, thiscatalyst
is suitable for use in membrane reactors. Besides, compared
with published data of Rh supported on different ceramic
oxides, the Rh (0.2%)/ZrO2 solid results in the most ac-
tive formulation on a per-gram-of-rhodium basis. Our data
suggest that the origin of this remarkable performance is a
strong interaction between rhodium and ZrO2.
Carbon deposits formed on Ni-containing catalysts used
at high temperature are mostly graphitic filaments, prob-
ably formed on different surface sites, exhibiting different
reactivities.
In the bimetallic formulation, the formation of Ni–Rh
alloys is likely to be responsible for the deactivation of this
solid at 823 K when carbonaceous deposits are negligible.
The formation of Ni–Rh alloys would impair the strong
Rh/ZrO2 interaction.
ACKNOWLEDGMENTS
The authors acknowledge the financial support received from UNL,
CONICET, and ANPCyT. Thanks are given to the Japan International
Cooperation Agency (JICA) for the donation of the major instruments,
to Elsa Grimaldi for editing the English, to Fabricio Charles and John
Munera for their technical assistance, and to Prof. Marı´a Alicia Ulla for
her helpful discussions.
Three factors (20) may influence the deactivation of these
catalysts, namely the deposition of carbonaceous residues,
metal particle sintering, and blockage of Rh surface sites
by species originated in the support. The absence of car-
bon deposition at 823 K eliminates the first factor. Metal
sintering seems to be minimized due to the metal–support
interaction evidenced by TPR and XPS.
REFERENCES
1. Kikuchi, E., Wemoto, Y., Kejiwara, M., Vemiya, S., and Kojima, T.,
Catal. Today 56, 75 (2000).
2. Wang, H., and Ruckenstein, E., Appl. Catal. A 204, 143 (2000).
3. Bitter, J., Seshan, K., and Lercher, J., J. Catal. 176, 93 (1998).
When the reaction is conducted at high temperature, car-
bon deposition occurs on the bimetallic solids. This could

272
IRUSTA, CORNAGLIA, AND LOMBARDO
4. Bradford, M., and Vannice, M., Catal. Rev.–Sci. Eng. 41(1), 1
(1999).
15. Pechi, G., Reyes, P., Go´mez, R., Lo´pez, T., and Fierro, J., Appl. Catal.
B 17, L7 (1998).
5. Wang, S., and Lu, G., Appl. Catal. B 16, 269 (1998).
6. Montoya, A., Romero, E., Guimon, C., and Monzo´n, A., Catal. Today
63, 71 (2000).
7. Noronha, F., Fendley, E., Soares, R., Alvarez, W., and Resasco, D.,
Chem. Eng. J. 82, 21 (2001).
16. Wang, S., Energy Fuels 12, 248 (1998).
17. Tuinstra, F., and Koenig, J., J. Chem. Phys. 53, 1126 (1970).
18. Takenaka, S., Ogihara, H., Yamanaka, I., and Otsuka, K., Appl. Catal.
A 217, 101 (2001).
19. Kroll, V., Swaan, H., and Mirodotos, C., J. Catal. 161, 409 (1996).
20. Zhang, Z., Tsipouriari, V., Etstethiou, A., and Verykios, X., J. Catal.
158, 51 (1996).
8. Nagaoka, K., Seshan, K., Lercher, J., and Aika, K., Catal. Lett. 70, 109
(2000).
9. Bitter, J. H., Seshan, K., and Lercher, J. A., Top. Catal. 10, 295
(2000).
21. Van Keule, A., Seshan, K., Hoebink, J., and Ross, J., J. Catal. 166, 306
(1997).
10. Irusta, S., Cornaglia, L. M., and Lombardo, E. A., J. Catal., in press.
22. Bitter, J., Seshan, K., and Lercher, J., J. Catal. 171, 279 (1997).
11. Xu, Z., Li, Y., Zhang, J., Chang, L., Zhou, R., and Duan, Z., Appl. 23. Stagg, W., Romeo, E., Padro, C., and Resasco, A., J. Catal. 178, 137
Catal. A 210, 45 (2001). (1998).
12. Lei, H., Song, Z., Tan, D., Bao, X., Mu, X., Zong, B., and Min, E., Appl. 24. Van Tol, M., and Gielbert, A., Appl. Surf. Sci. 67, 166 (1993).
Catal. A 214, 69 (2001).
25. Moffett, W., “The Handbook of Binary Phase Diagrams,” Vol. 4. Gen-
eral Electric Co., 1978.
26. Bradford, M., and Vannice, M. A., J. Catal. 173, 157 (1998).
27. Roberts, S., and Gorte, R., J. Phys. Chem. 95, 5600 (1991).
13. McIntyre, N., and Cook, H., Anal. Chem. 47(13), 2208 (1975).
14. Sigl, M., Bradford, M., Kno¨zinger, H., and Vannice, A., Top. Catal. 8,
211 (1999).