Isotopic and kinetic assessment of the mechanism of reactions of CH 4 with CO2 or H2O to form synthesis gas and carbon on nickel catalysts

Article abstract of DOI:10.1016/j.jcat.2004.02.032

Kinetic and isotopic measurements for catalysts and conditions that rigorously excluded transport and thermodynamic artifacts led to a common sequence of elementary steps for reactions of CH₄ with CO₂ or H₂O and for its stoichiometric decomposition on Ni/MgO catalysts. Turnover rates for forward reactions of CH₄/CO₂ and CH₄/H₂O mixtures were proportional to CH₄ pressure (5-450 kPa) and independent of the partial pressure of the CO ₂ or H₂O coreactants (5-450 kPa). These turnover rates and their first-order rate constants and activation energies are also similar to those measured for CH₄ decomposition, indicating that these reactions are mechanistically equivalent and that C-H bond activation is the sole kinetically relevant step in all three reactions. These conclusions were confirmed by identical CH₄/CD₄ kinetic isotope effects (k_H/k_D=1.62-1.71) for reforming and decomposition reactions and by undetectable H₂O/D₂O isotopic effects. The kinetic relevance of C-H bond activation is consistent with the relative rates of chemical conversion and isotopic mixing in a CH₄/CD ₄/CO₂ mixture and with the isotopic evidence for the quasi-equilibrated nature of coreactant activation and H₂ and H ₂O desorption obtained from reactions of CH₄/CO ₂/D₂ and ¹²CH₄/¹²CO ₂/¹³CO mixtures. These quasi-equilibrated steps lead to equilibrated water-gas-shift reactions during CH₄ reforming, a finding confirmed by measurements of the effluent composition. These elementary steps provide also a predictive model for carbon filament growth and identify a rigorous dependence of the carbon thermodynamic activity on various kinetic and thermodynamic properties of elementary steps and on the prevalent concentrations of reactants and products, specifically given by P_CH4P _CO/P_CO2 (or P_CH4P_H2/P_H2O) ratios. These mechanistic features on Ni surfaces resemble those previously established for supported noble metal catalysts (Rh, Pt, Ir, Ru). These direct measurements of C-H bond activation turnover rates allowed the first direct and rigorous comparison of the reactivity of Ni and noble metal catalysts for CH₄-reforming reactions, under conditions of strict kinetic control and relevant commercial practice and over a wide range of compositions and metal dispersions.

Full text of DOI:10.1016/j.jcat.2004.02.032

Journal of Catalysis 224 (2004) 370–383
www.elsevier.com/locate/jcat
Isotopic and kinetic assessment of the mechanism of reactions of CH4
with CO2 or H2O to form synthesis gas and carbon on nickel catalysts
Junmei Wei and Enrique Iglesia ∗
Department of Chemical Engineering, University of California at Berkeley, Berkeley, CA 94720, USA
Received 24 November 2003; revised 16 February 2004; accepted 24 February 2004
Available online 20 April 2004
Abstract
Kinetic and isotopic measurements for catalysts and conditions that rigorously excluded transport and thermodynamic artifacts led to
a common sequence of elementary steps for reactions of CH with CO or H O and for its stoichiometric decomposition on Ni/MgO
4
2
2
catalysts. Turnover rates for forward reactions of CH /CO and CH /H O mixtures were proportional to CH pressure (5–450 kPa) and
4
2
4
2
4
independent of the partial pressure of the CO or H O coreactants (5–450 kPa). These turnover rates and their first-order rate constants and
2
2
activation energies are also similar to those measured for CH decomposition, indicating that these reactions are mechanistically equivalent
4
and that C–H bond activation is the sole kinetically relevant step in all three reactions. These conclusions were confirmed by identical
CH /CD kinetic isotope effects (k /k = 1.62–1.71) for reforming and decomposition reactions and by undetectable H O/D O isotopic
H
D
4
4
2
2
effects. The kinetic relevance of C–H bond activation is consistent with the relative rates of chemical conversion and isotopic mixing in a
CH /CD /CO mixture and with the isotopic evidence for the quasi-equilibrated nature of coreactant activation and H and H O desorption
4
4
2
2
2
12
12
13
obtained from reactions of CH /CO /D and CH / CO / CO mixtures. These quasi-equilibrated steps lead to equilibrated water–gas-
4
2
2
4
2
shift reactions during CH reforming, a finding confirmed by measurements of the effluent composition. These elementary steps provide also
4
a predictive model for carbon filament growth and identify a rigorous dependence of the carbon thermodynamic activity on various kinetic
and thermodynamic properties of elementary steps and on the prevalent concentrations of reactants and products, specifically given by
P
P
/P
(or P
P
/P
) ratios. These mechanistic features on Ni surfaces resemble those previously established for supported
CH4 CO CO2
CH4 H2 H2O
noble metal catalysts (Rh, Pt, Ir, Ru). These direct measurements of C–H bond activation turnover rates allowed the first direct and rigorous
comparison of the reactivity of Ni and noble metal catalysts for CH -reforming reactions, under conditions of strict kinetic control and
4
relevant commercial practice and over a wide range of compositions and metal dispersions.
2004 Elsevier Inc. All rights reserved.
Keywords: C–H bond activation; Methane reforming; Carbon filaments; Nickel catalysts; Isotopic tracer methods
1. Introduction
for the on-purpose synthesis of carbon nanotubes catalyzed
by metal particles [5–7]. Detailed molecular-level pathways
for conversion of CH4 to H2/CO mixtures and carbon, and
even their kinetic response to reactant concentrations, re-
main controversial and often contradictory, even after many
studies [8–28]. Kinetic measurements are often corrupted
by ubiquitous concentration and temperature gradients at
the high temperatures required for these endothermic reac-
tions, which require the activation of strong C–H bonds, and
by significant but often unrecognized contributions from re-
verse reactions as reactions approach equilibrium.
CH4 reactions with CO2 or H2O on Group VIII metals
form synthesis gas precursors to valuable fuels and chem-
icals, as first shown by Fischer and Tropsch [1]. Industrial
practice relies on Ni catalysts, because of cost and avail-
ability concerns about noble metals. Ni-based catalysts tend
to form unreactive carbon residues that block active sites
and catalyst pores, and which lead ultimately to the for-
mation of carbon filaments and to the elution of Ni crys-
tallites from catalyst pellets and reactors [2–4]. Carbon fil-
ament formation pathways are also of significant interest
The prevailing controversies are illustrated by the broad
range of disparate rate equations for CH4 reforming shown
in Table 1. For instance, a complex rate equation, purported
to resolve previous contradictory results on Ni-based cata-
*
Corresponding author. Fax: +510 642 4778.
E-mail address: iglesia@cchem.berkeley.edu (E. Iglesia).
0021-9517/$ – see front matter
doi:10.1016/j.jcat.2004.02.032
2004 Elsevier Inc. All rights reserved.

J. Wei, E. Iglesia / Journal of Catalysis 224 (2004) 370–383
371
Table 1
Proposed rate expressions for CH -reforming reactions on Ni-based catalysts
4
Rate expression
Coreactant
Catalyst
Refs.
r = kP
H O
2
Ni/Kieselguhr
[8]
[9]
CH
4
kP
CH
4
r =
H O
2
Ni foil
1 + a[P
2.5
/P ] + bP
CH H O
4
aP
H
2
H O
2
CO
3
2
2.5
3
2
a
r = a/P
(P
P
− P
P
2 /b)/α + c/P
(P
P
− P
P
2/d)/α
CO
H
H O
2
Ni/MgAl O
2 4
[13,14]
CO
CH H O
CH
2
H
CH
4
2
4
4
P
CH CO
4
2
r =
CO
2
Ni/TiO , Ni/MgO, Ni/SiO
2 2
[16]
(4−x)/2
bP
P
+ (1 + cP
)P
CH CO
4
CO
H
2
2
kP
P
CH CO
4
2
r =
CO
2
Ni/Al O , Ni/CeO –Al O
2 3
[20]
[23]
2
3
2
(1 + K
P
)(1 + K
P
CO CO
2 2
)
CH CH
4
4
aP
P
CH CO
2
r =
CO
2
Ni/La O
2 3
bP
P
+4cP
+ dP
CH CO
4
CH
CO
2
4
a
α = 1 + K
P
+ K
P
+ K
P
+ K
P
H O H O
2 2
/P
H
2
.
H
H
CO CO
CH CH
4
2
2
4
lysts [13,14],
generality of the kinetic conclusions. These steps also pro-
vide a rigorous and accurate kinetic framework for dynamic
treatments of carbon deposition and filament formation dur-
ing high-temperature CH4-reforming and decomposition re-
actions on Ni crystallites [35]. Finally, we report here the
first rigorous comparison of CH4-reforming turnover rates
on Ni and noble metal (Pt, Ir, Rh, Ru) surfaces under demon-
strated kinetic control, corrected for approach to equilib-
rium, and under conditions of relevant industrial practice.
ꢀ
ꢁ
PH3 PCO
ꢂ
a
PH2.5
2
2
rCH
=
PCH PH O
−
[α]
ꢁ
4
4
2
b
2
ꢀ
PH4
P
d
ꢂ
c
PH3.5
CO2
+
PCH PH2 O
−
[α] ,
(1)
2
2
4
2
2
[α] = [1 + KCOPCO + KH PH + KCH PCH
2
2
4
4
+ KH OPH O/PH ],
2
2
2
and widely used [29,30], predicts infinite rates at the reac-
tor inlet and substantial surface coverages by chemisorbed
O atoms and by molecularly adsorbed H2, CH4, and CO at
temperatures well above those required for their desorption
or dissociation. CO2 reforming of CH4 was proposed to oc-
cur via reversible CH4 and CO2 dissociation to form CHxO
[16] as the most abundant reaction intermediate, and CH4
activation and CHxO decomposition as kinetically relevant
steps to give the rate equation:
2. Experimental methods
Ni/MgO-A (7 and 15 wt%) and Ni/MgO-B (7 wt%) were
prepared by incipient wetness impregnation of MgO-A (pre-
pared as described below) or MgO-B (Alfa, CAS 1309-48-4)
with an aqueous Ni(NO3)2 · 6H2O (Alfa, 99.9%) solution
and subsequent drying at 393 K in ambient air and treat-
ment in flowing dry air (Airgas, UHP, 1.2 cm3/(g s)) as the
temperature was increased at 0.167 K/s to 923 K and held
for 5 h. Samples were then treated in H2 (Airgas, UHP, 50
cm3/(g s)) by heating to 1123 K at 0.167 K/s and holding
for 3 h. MgO-A powders were prepared by sol–gel methods
using supercritical drying, a procedure that leads to MgO-A
crystallites of small and uniform diameter and to high ul-
timate dispersion of Ni metal clusters [36]. In this sol–gel
method, Mg(OH)2 was first obtained by hydrolysis of 0.4 M
aqueous MgCl2 (Aldrich, > 98 wt%) at pH 10, maintained
by controlled addition of 14.8 M NH4OH. Precipitates were
filtered and washed with deionized water until Cl ions were
no longer detected in the eluted water using AgNO3 (Cl− <
10 ppm). Powders were then washed three times with an-
hydrous ethanol (10 cm3/g-MgO), with acetone, which is
very soluble in liquid CO2, and then with liquid CO2 within
a supercritical dryer, where they were heated to 311–313 K
at 0.008 K/s while the CO2 pressure was increased to 8.2–
8.9 MPa to reach the critical point (304.5 K, 7.6 MPa). The
pressure was then released to vaporize CO2 directly from its
supercritical state and the sample treated at 923 K for 3 h
in flowing dry air (Airgas, UHP, 1.2 cm3/(g s)) by heating
aPCH PCO
4
2
r =
.
(2)
(4−x)/2
bPCOPH
+ (1 + cPCH )PCO
2
4
2
These conflicting rate equations and reaction pathways
led us to collect and interpret kinetic data that rigorously ex-
clude transport and thermodynamic artifacts and to carry out
isotopic studies to probe the reversibility and kinetic rele-
vance of specific elementary steps. In doing this, we have
developed a unified treatment of CH4 reactions with H2O or
CO2, CH4 decomposition, and water–gas-shift reactions, as
well as mechanism-based rate equations for CH4-reforming
reactions. Here, we postulate and confirm via independent
measurements a catalytic sequence that reveals a rigorous
and intrinsic kinetic and mechanistic equivalence among
these reactions. We also provide rigorous evidence for the
sole kinetic relevance of C–H bond activation steps and for
the equilibrated nature of all elementary steps involved in
water–gas-shift catalytic sequences. These elementary steps
and their kinetic relevance resemble those recently demon-
strated also on noble metal catalysts [31–34], attesting to the

372
J. Wei, E. Iglesia / Journal of Catalysis 224 (2004) 370–383
at 0.167 K/s. Synthesis procedures for the supports used for
Ru [31], Pt [32], Ir [33], and Rh [34] catalysts and their syn-
thesis procedures have been reported elsewhere.
Reduction studies of supported NiO precursors were
carried out in a Quantasorb analyzer (Quantachrome Cor-
poration) using thermal conductivity detection, electronic
mass-flow controllers, a programmable furnace, and on-line
molecular sieve traps to remove the water formed during
reduction. Samples (0.1 g) were reduced in 20% H2/Ar
(0.67 cm3/s, Matheson UHP, certified mixture) as the tem-
perature increased to 1123 K at 0.167 K/s. The thermal
conductivity response was calibrated using CuO reduction
(Aldrich, 99.995%).
Ni metal dispersion was measured on fresh samples from
uptakes of strongly chemisorbed H2 at 313 K (3–50 kPa)
using a Quantasorb chemisorption analyzer (Quantachrome
Corp.), after reducing samples ex situ at 1123 K for 3 h and
then at 873 K for 0.5 h within the adsorption cell. A back-
sorption isotherm was measured by repeating this procedure
after evacuation at 313 K for 0.5 h. The difference between
chemisorption and backsorption isotherms was used to mea-
sure strongly chemisorbed hydrogen uptakes; Ni dispersion
was calculated using 1:1 H:Ni titration stoichiometry [16].
Rates of CH4–CO2 and CH4–H2O reforming and of
CH4 decomposition were measured on catalyst pellets (250–
425 µm) diluted with inert γ -Al2O3 powders (5/25 mg);
these pellets were then mixed with acid-washed quartz pow-
der (500 mg; 250–425 µm) in packed catalyst beds. CO2-
and H2O-reforming rates were measured at 823–1023 K and
100–1500 kPa total pressure using a microreactor system
previously described [31]. Reactant concentrations were in-
dependently varied using 50% CH4/Ar (Matheson UHP) and
50% CO2/Ar (Matheson UHP) certified mixtures (metered
by electronic controllers) and a pure H2O stream (introduced
with a syringe pump; ISCO Model 500D). All transfer lines
after H2O introduction were kept at 500 K to avoid conden-
sation. Reactant and product concentrations were measured
with a Hewlett-Packard 6890 gas chromatograph using a
Carboxen 1000 packed column (3.2 mm × 2 m) and ther-
mal conductivity detection.
CH4 and CD4 decomposition rates on 7 wt% Ni/MgO-A
were measured from dihydrogen and methane isotopomer
concentrations by mass spectrometry (Leybold Inficon,
Transpector Series). Reactant mixtures were prepared from
50% CH4/Ar (Matheson, certified mixture) or CD4 (Isotec,
chemical purity > 99.0%) and Ar (Airgas, UHP). CH4 and
CD4 mass fragments at 15 and 18 amu were used to mea-
sure their respective concentrations, and 2 and 4 amu were
used for H2 and D2, respectively. Decomposition rates were
extrapolated to the time of initial contact with methane to
obtain initial rates before significant carbon formation. Rate
constants were obtained from reaction rates using the ob-
served linear dependence of decomposition rates on CH4
pressure.
Fig. 1. TPR profile of 7 wt% Ni/MgO-A and Ni/MgO-B (100 mg catalyst,
3
20% H /Ar at 0.67 cm /s, temperature increased to 1123 K at 0.167 K/s).
2
delays (< 5 s). Chemical and isotopic compositions were
measured by mass spectrometric analysis of the reactor ef-
fluent (Leybold Inficon, Transpector Series). CD4 (Isotec,
chemical purity > 99.0%), D2O (Isotec, chemical purity
> 99.0%), and 5% D2/Ar and 13CO (Isotec, chemical purity
> 99.0%) were used as reactants. Intensities at 15 and 17–20
amu were used to measure methane isotopomer concentra-
tions. CH4 and CD4 standard fragmentation patterns were
measured, and those for CHD3, CH2D2, and CH3D were
calculated using reported methods [37]. Intensities at 18, 19,
and 20 amu were used to determine water isotopomers and
those at 28, 29, 44, and 45 amu to measure 12CO, 13CO,
12CO2, and 13CO2 concentrations, respectively.
Carbon formation rates were measured during CH4 re-
forming at 873 K using a tapered-element oscillating quartz
element (Ruprecht & Patashnick, 1500 pma) microbalance
while continuously monitoring the effluent by mass spec-
trometry. Samples (30 mg) were reduced at 1123 K for 3 h
and then again within the balance chamber at 873 K for 0.5 h.
3. Results and discussion
3.1. Catalyst characterization
Fig. 1 shows reduction rates for 7 wt% Ni/MgO-A and
7 wt% Ni/MgO-B in H2 during temperature ramping. The
reduction profile of 15 wt% Ni/MgO-A is similar to that
of 7 wt% Ni/MgO-A. Only ∼ 28.2, 46.3, and 43.4% of
the Ni atoms in the samples are reduced to Ni0 after treat-
ment at 1123 K for 7 wt% Ni/MgO-A, 15 wt% Ni/MgO-A,
Isotopic tracer studies on 7 wt% Ni/MgO-A were carried
out using a transient flow apparatus with short hydrodynamic

J. Wei, E. Iglesia / Journal of Catalysis 224 (2004) 370–383
373
Table 2
Extent of reduction, dispersion, and average crystal size for Ni/MgO catalysts
a
b
c
Catalyst
Percentage reduction (%)
Dispersion (%)
Average particle size (nm)
7 wt% Ni/MgO-A
15 wt% Ni/MgO-A
7 wt% Ni/MgO-B
28.2
46.3
43.4
14.8
9.6
9.3
6.7
10.4
10.8
a
The fraction of the reduced Ni atoms to the total Ni atoms.
b
c
The fraction of the surface Ni atoms from H chemisorption to the reduced Ni atoms.
2
Estimated from fraction Ni dispersion using D = 1/d [38].
Fig. 3. Net CH turnover rates versus residence time for CH –H O reaction
4
4
2
Fig. 2. Net CH reaction rate for H O–CH reaction as a function of time
4
2
4
on 7 wt% Ni/MgO-A at 873 K ((!) 10 mg catalyst diluted with 100 mg
Al O within pellets, then diluted with 500 mg ground quartz, pellet size
on stream on 7 wt% Ni/MgO-A (973 K, 33 kPa CH , 33 kPa H O, balance
4
2
2
3
Ar, η = 0.2).
250–425 µm; (Q) 10 mg catalyst diluted with 50 mg Al O within pellets,
2
3
then diluted with 500 mg ground quartz, pellet size 250–425 µm; (P) 10 mg
catalyst diluted with 50 mg Al O within pellets, then diluted with 500 mg
ground quartz, pellet size 63–106 µm, extrapolation of the dash line to zero
2
3
and 7 wt% Ni/MgO-B, respectively, while the remaining Ni
atoms remain unreduced within NiO–MgO solid solutions
[38,39]. The number of exposed Ni surface atoms in these
three reduced samples is reported as their fractional disper-
sion in Table 2, defined as the number of exposed Ni atoms,
measured by H2 chemisorption, divided by the number of
reduced Ni atoms.
residence time gives forward CH turnover rate).
4
ground quartz. The absence of intraparticle and bed trans-
port artifacts was confirmed by similar CH4–CO2 reaction
rates measured as the pellet size (250–425 vs 63–106 µm)
or the extent of dilution within pellets (5:1 to 10:1) was
varied (Fig. 3). Fig. 3 shows net CH4 turnover rates (nor-
malized by exposed Ni atoms) for CH4–CO2 reactions at
873 K as a function of residence time. Residence time ef-
fects on CH4 turnover rates reflect reactant depletion and the
approach to thermodynamic equilibrium for reforming reac-
tions. Forward reforming rates can be rigorously obtained
from measured net rates using an approach to equilibrium
parameter (η) evaluated from CH4–CO2 and CH4–H2O ther-
modynamic reaction data [40] and measured reactant and
product partial pressures:
3.2. Kinetic dependence of forward CH4 reaction rates on
reactant and product concentrations
CH4 reaction rates were measured on Ni/MgO catalysts
in the absence of detectable deactivation and of transport or
thermodynamic corruption of rate measurements. For 7 wt%
Ni/MgO-A, all kinetic measurements were carried out at ef-
fluent PCH PCO/PCO ratios below 4 kPa, which as shown
4
2
later, led either to undetectable or to catalytically irrelevant
carbon formation rates.
Fig. 2 shows CH4 reaction rates during H2O–CH4 reac-
tions at 973 K as a function of time on stream on 7 wt%
Ni/MgO-A. CH4 reaction rates remained constant for 100 h.
Transport artifacts were ruled out by dilution of the catalyst
pellet with inert Al2O3 and dilution of the catalyst bed with
2
2
[PCO] [PH
]
1
2
η1 =
η2 =
,
.
(3)
(4)
[P ][P ] KEQ1
CH4
CO2
3
[PCO][PH
]
1
2
[P ][P ] KEQ2
CH4
H2O

374
J. Wei, E. Iglesia / Journal of Catalysis 224 (2004) 370–383
Fig. 4. Effect of CH (a) and H O (b) partial pressure on forward CH reaction rate for CH –H O reaction on 7 wt% Ni/MgO-A (873 K, balance He, average
4
2
4
4
2
pressure is the average of inlet and outlet pressures of the reactor).
Fig. 5. Effects of CH (a) and CO (b) partial pressure on forward CH reaction rate for CO reforming of CH on 7 wt% Ni/MgO-A (balance He, average
4
2
4
2
4
pressure is the average of inlet and outlet pressures of the reactor).
In these expressions, [Pj ] is the average partial pressure of
species j (in atm); it is also used to correct for reactant de-
pletion. KEQ1 and KEQ2 are equilibrium constants for CH4–
CO2 and CH4–H2O reactions, respectively; η values ranged
from 0.05 to 0.20 in this study. Forward turnover rates (rf)
are given by [41]
forward CH4 reaction rates increased linearly with increas-
ing CH4 pressure (5–450 kPa, Figs. 4a and 5a), but they
were not influenced by H2O or CO2 pressures (5–450 kPa,
Fig. 4b). Figs. 4 and 5 also show that forward CH4 turnover
rates using CO2 and H2O coreactants are identical to each
other within experimental accuracy at any given CH4 pres-
sure. These rates also resemble initial CH4 decomposition
rates (Table 2, Fig. 6).
rf = rn/(1 − η),
(5)
H2 (0–20 kPa) and CO (0–10 kPa) products added to
CH4–CO2 or CH4–H2O reactant mixtures influenced net
CH4 conversion rates at 873 K, but forward rates were unaf-
fected (Fig. 7), indicating that H2 and CO influenced only the
extent to which reforming reactions approach equilibrium,
but not the kinetics of forward CH4-reforming reactions.
Thus, previously reported inhibition of CH4 reaction rates
by products may reflect unrecognized contributions from
where rn is the net CH4 conversion turnover rate. The curve
shown in Fig. 3 represents rates predicted from this expres-
sion and the first-order dependence of reaction rates on CH4
concentration reported below.
CH4–H2O reaction rates at 873 K and CH4–CO2 reac-
tion rates at 823 and 873 K and 100–1500 kPa total reactant
pressures are shown as a function of CH4 and coreactant
pressures in Figs. 4 and 5, respectively. For both reactions,

J. Wei, E. Iglesia / Journal of Catalysis 224 (2004) 370–383
375
0.01 monolayer at 873 K and even at equilibrium CH4 con-
versions. The lower adsorption enthalpies typically reported
for H2 (92 kJ/mol [43]) would make hydrogen coverages
even lower than for CO. Thus, competitive adsorption is un-
likely to influence the availability of metal sites for CH4
activation and reported inhibition effects predominately re-
flect contributions from reverse reactions.
These kinetic responses to reactant and product concen-
trations are consistent with rate-determining CH4 activation
steps on surfaces essentially free of reactive intermediates
or coadsorbed products. CH4-derived chemisorbed interme-
diates appear to be readily removed via reactions with CO2
or H2O coreactants; as a result, the identity and concentra-
tion of coreactants become kinetically irrelevant and forward
rate data for both CH4–CO2 and CH4–H2O reactions are ac-
curately described by
rf = kPCH
,
(6)
4
Fig. 6. CH decomposition rate as a function of time on stream on 7 wt%
4
Ni/MgO-A (873 K, 20 kPa CH , 100 kPa total pressure, balance Ar).
with similar rate constants for the two coreactants (Table 3)
and even for stoichiometric CH4 decomposition reactions.
These results and conclusions are similar for the three Ni
catalysts in this study and for a wide range of reaction condi-
tions relevant to industrial practice (823–1023 K; 5–450 kPa
CH4, CO2, or H2O pressures). The similar rate constants
4
reverse reaction rates [9,17,22]. CO adsorption enthalpies
are 135 kJ/mol at low coverages (< 0.02) and much lower
at higher coverages on Ni(100) [42]. These adsorption en-
thalpies indicate that CO coverages should be well below
Fig. 7. Effect of H (a) and CO (b) partial pressures on forward CH reaction rate for CH -reforming reactions on 7 wt% Ni/MgO-A (873 K, 20 kPa CH ,
2
4
4
4
20 kPa CO or H O, 100 kPa total pressure, balance He).
2
2
Table 3
Forward CH reaction rates, rate constants, preexponential factors, activation energies, and kinetic isotopic effect (KIE) for CH reactions on 7 wt% Ni/MgO-A
4
4
(873 K, 20 kPa CH , 25 kPa CO or H O, 100 kPa total pressure, balance Ar)
4
2
2
−1
−1
−1
−1
−1
Coreactant
Turnover rate (s
)
Rate constant (s kPa
)
Activation energy (kJ/mol)
Preexponential factor (s kPa )
KIE
b
Measured
Estimated
5
3
CO
2
H O
2
None
4.0
4.1
3.8
0.20
0.20
0.19
105
102
99
3.8 × 10
5.5 × 10
1.62
1.66
1.71
5
3
2.5 × 10
5.5 × 10
a
5
3
1.6 × 10
5.5 × 10
a
Initial CH turnover rate on Ni surface.
4
b
Calculated based on transition-state theory treatments of CH activation steps proceeding via an immobile activated complex [59].
4

376
J. Wei, E. Iglesia / Journal of Catalysis 224 (2004) 370–383
gies of 52.7, 26.8, and 55.6 kJ/mol on Ni(111), Ni(100), and
Ni(110), respectively [57,58].
It is possible that dissolved or chemisorbed carbon
species form during initial contact with reactants in catalytic
reactions and titrate those sites with the highest carbon bind-
ing energy, which would also dissociate C–H bonds most ef-
fectively by stabilizing the required transition state. It is un-
clear, however, why such passivation effects would not also
influence reaction rates on model surfaces. Similar discrep-
ancies between activation energies measured on catalysts
and model surfaces were observed on Rh [31], Pt [32], and Ir
[33], for which CO oxidation rates before and after reform-
ing reactions showed that the number of available surface
metal atoms remained unchanged during CH4-reforming re-
actions. Any unreactive carbon species that formed on Ni
surfaces during initial contact with reactants must be unre-
active, because their concentration and therefore reforming
rates would otherwise depend on the concentration and iden-
tity of coreactants. The reasons for these differences between
steady-state catalytic rates on supported Ni clusters and rates
estimated from C–H bond activation kinetic parameters on
single crystals remain unclear, but may reflect the presence
of specific defect sites on single crystals or the nature of the
molecular beam experiments conducted on lower tempera-
ture surfaces.
Measured preexponential factors for CH4 activation steps
involved in CH4–CO2, CH4–H2O, and CH4 decomposi-
tion reactions are about 100 times greater than predicted
from transition-state theory for immobile activated com-
plexes (Table 3). These differences suggest that activated
complexes can undergo limited two-dimensional translation
in the time scale of C–H bond activation events at typical
reaction temperatures. Significant surface coverages by un-
reactive chemisorbed carbon species would have led instead
to smaller than predicted preexponential factors.
Fig. 8. Arrhenius plots for CO reforming (F), H O reforming (E), and
2
2
CH decomposition (P) rate constants on 7 wt% Ni/MgO-A.
4
on 7 wt% Ni/MgO-A for H2O (kH O) and CO2 (kCO ) re-
2
2
forming and CH4 decomposition (kCH ) in this temperature
4
range (Fig. 8) indicate that activation energies must also
be similar for these three CH4 reactions (Table 3) and that
coreactants are required by stoichiometry, but they do not in-
fluence the dynamics of CH4-reforming reactions. It appears
that previous complex and contradictory rate equations may
reflect varying and unrecognized contributions from concen-
tration and temperature gradients or nonrigorous corrections
for thermodynamic constraints.
The activation energies measured for first-order rate con-
stants for CO2 reforming (105 kJ/mol), H2O reforming
(102 kJ/mol), and CH4 decomposition (99 kJ/mol) resem-
ble those previously reported for CO2 reforming (e.g., 92,
96, and 109 kJ/mol on Ni/MgO, Ni/SiO2, and Ni/TiO2) [44]
and H2O reforming (95 kJ/mol on Ni/Y-ZrO2 [45]) on Ni
catalysts and lie within a broad range of values reported for
Ni supported on Al2O3 (74–118 kJ/mol [46–49]) and mod-
ified Al2O3 (90–107 kJ/mol [17,50]). Density-functional
theory led to activation energy estimates of 85 kJ/mol [51],
93 kJ/mol [52], and 100 kJ/mol [53] for C–H bond activa-
tion of CH4 on Ni(111) surfaces, while embedding methods
led to slightly lower values (72 kJ/mol [54]). These theoret-
ical estimates are in reasonable agreement with experiment
and consistent with the conclusion that the dynamics of CH4
reforming and decomposition reflect only the rate constants
for C–H bond activation elementary steps on Ni surfaces.
These experimental and theoretical activation energies
are, however, much greater than those measured for CH4
activation on Ni single crystals [55–58]. Molecular beam
studies led to activation energies of 52 kJ/mol [55] and
59 kJ/mol [56] on Ni(100). CH4 dissociation rates at
0.13 kPa and 450–600 K, from chemisorbed carbon mea-
sured by Auger electron spectroscopy, gave activation ener-
Catalytic reactions of CH4 with H2O or CO2 to form H2–
CO mixtures and stoichiometric CH4 decomposition to form
C* and H2 on Ni appear to depend only on the rate of initial
activation of C–H bonds catalyzed by surface Ni atoms. Co-
reactant activation is facile and CH4-derived intermediates,
including reactive chemisorbed carbon, are kept well below
monolayer coverages by their rapid reactions with interme-
diates derived from CO2 or H2O coreactants. In the next
section, we describe kinetic isotope effect (KIE) data that
confirm the exclusive kinetic relevance of C–H bond activa-
tion steps in CH4–H2O, CH4–CO2, and CH4 decomposition
reactions on Ni/MgO.
3.3. CH4/CD4 kinetic isotopic effects
Kinetic isotope effects on 7 wt% Ni/MgO-A were mea-
sured for CO2 reforming from forward rates of CH4–CO2
and CD4–CO2 mixtures at 873 K, 25 kPa methane, and
25 kPa CO2. For H2O reforming, isotopic effects were
measured from the corresponding forward rates for CH4–
H2O, CD4–H2O, or CD4–D2O reactants at 873 K, 25 kPa

J. Wei, E. Iglesia / Journal of Catalysis 224 (2004) 370–383
377
Similar H2O–CH4-reforming and CH4 decomposition
rates on Ni foil at 1073 K led previous studies to con-
clude that only C–H bond activation steps limit overall
H2O-reforming rates [12], as also previously proposed for
CO2 reforming on Ni/Al2O3 without direct or specific evi-
dence [15]. These conclusions are similar to those we reach
here on the basis of more extensive experimental evidence.
Several other studies, however, have reported markedly dif-
ferent conclusions. The kinetic misinterpretation of thermo-
dynamic isotope effects led previous studies to conclude that
CO2 activation or reaction of chemisorbed carbon with oxy-
gen limited overall reforming rates [63]. CO2 dissociation
[66] or its reaction with adsorbed CHx species [61] was
proposed to limit CO2-reforming rates. A kinetic analysis
[67] led one study to conclude that CH4 dissociation can
be reversible (a rigorous requirement as overall reactions
approach equilibrium), that more than one elementary step
limited overall rates, and that chemisorbed oxygen was in-
volved in one or more of these limiting steps.
Fig. 9. Forward CH reaction rates with CH /CO (F), CD /CO (Q),
4
4
2
4
2
CH /H O (E), CD /H O (P), and CD /D O (") reactant mixtures on
4
2
4
2
4
2
We note that the kinetic equivalence among CH4 reac-
tions, the specific dependence of rates on CH4 and coreac-
tant pressures, and the kinetic isotope effects reported here
support the exclusive kinetic relevance of C–H bond acti-
vation and the lack of involvement of any species derived
from coreactants in rate-limiting steps during either CH4–
H2O or CH4–CO2 reactions. In the next sections, we provide
additional isotopic evidence for the reversibility and kinetic
relevance of several of the proposed elementary steps in-
volved in these reactions.
7 wt% Ni/MgO-A (5 mg catalyst, 873 K, 25 kPa CO or H O, 25 kPa CH
2
2
4
or CD , 100 kPa total pressure, balance Ar).
4
methane, and 25 kPa water. Both CH4–CO2 and CH4–H2O
reactions showed normal kinetic isotopic effects (Fig. 9)
and these kinetic isotopic effects were identical for the two
coreactants within experimental accuracy (Table 3). For-
ward rates were similar within experimental accuracy for
CD4–H2O and CD4–D2O reactants (Fig. 9), indicating that
neither water activation nor any steps involving adsorbed
species derived from water are kinetically relevant. Kinetic
isotope effects measured from initial CH4 and CD4 decom-
position rates resemble those for catalytic CO2 and H2O-
reforming reactions (Table 2), consistent with the equivalent
rate-controlling steps proposed for the three reactions.
These isotope effects are very similar to those reported
previously for CH4 decomposition at 773 K on Ni/SiO2
(1.60) [60] and for CO2 reforming at 873 K on Ni/Al2O3
(1.45) [61]; they also resemble values we have recently re-
ported on Ru (1.40–1.51) [31], Pt (1.58–1.77) [32], Ir (1.70–
1.81) [33], and Rh (1.54–1.60) [34] at 873 K. On these noble
metals, we also observed similar kinetic isotope effects for
H2O reforming, CO2 reforming, and CH4 decomposition.
Some previous studies have reported similar CO2-reforming
rates with CH4 and CD4 reactants on Ni–La2O3 and Ni/SiO2
and proposed the significant involvement of CO2 activa-
tion and of chemisorbed oxygen in kinetically relevant steps
[62–65]. These previous isotopic measurements were car-
ried out at near-equilibrium conversions; their interpretation
is inconsistent with the measured first-order dependence in
CH4, with the kinetic irrelevance of coreactant activation,
and with the isotope effects reported here on Ni/MgO. They
are likely to reflect isotope effects on equilibrium conver-
sions, which are expected to be much smaller than isotopic
effects on reaction rates.
3.4. Isotopic tracer studies of the reversibility of C–H bond
activation
If C–H bond activation is the only kinetically relevant
step, it must be exactly as reversible as the overall CH4-
reforming reaction, because the reaction affinity must be
identical for the overall reaction and for the rate-determining
step [41]. The reversibility of C–H bond activation can
be measured from the rate of isotopic cross-exchange dur-
ing CH4/CD4/CO2 reactions. Reversible C–H bond activa-
tion steps would rapidly form CHxD4−x (0 < x < 4) iso-
topomers via the microscopic reverse of C–H bond acti-
vation steps; in contrast, methane isotopomers would form
much more slowly than chemical conversion products when
C–H bond activation steps are irreversible.
CH4/CD4/CO2 (1:1:2) mixtures were reacted on 7 wt%
Ni/MgO-A at 873, 923, and 973 K and chemical and isotopic
conversion rates were measured by on-line mass spectrome-
try after removing HxD2−xO at 218 K to avoid interference
among mass fragments for H2O, HDO, D2O species, and
deuterated methane isotopomers. CH4/CD4 cross-exchange
turnover rates, defined as the sum of the rates of formation of
CHD3, CH2D2 (taken twice), and CH3D, are compared with
chemical conversion turnover rates in Table 4. At 923 K,
the methane chemical conversion rate is 8.8 s−1, about 17
times greater than the isotopic cross-exchange rate. The ap-

378
J. Wei, E. Iglesia / Journal of Catalysis 224 (2004) 370–383
Table 4
Methane chemical conversion rate and CH /CD cross-exchange rate during CH /CD /CO reaction on 7 wt% Ni/MgO (12.5 kPa CH and CD , 25 kPa
4
4
4
4
2
4
4
CO , 100 kPa total pressure, balance Ar)
2
−1 a
−1 b
c
d
η
Reaction temperature (K)
873
923
973
Methane chemical conversion turnover rate (s
)
Cross exchange rate (s
)
r
/r
exch reaction
3.5
8.8
19.0
0.24
0.50
0.85
0.07
0.06
0.04
0.06
0.05
0.05
a
b
c
d
Forward methane chemical conversion rate.
Total methane isotopomers formation rate.
Ratio of total methane isotopomers formation rate to the methane chemical conversion rate.
2
2
η = ([P ] [P ] )/([P
][P
]K); approach to equilibrium for CO -reforming reaction.
H
CO
CO
2
CH
2
2
4
proach to equilibrium for this reaction, η, estimated from the
prevalent concentrations of all reactants and products is 0.05,
corresponding to the ratio of the forward reaction rate to the
reverse reaction rate of 20, indicating that CHD3, CH2D2,
and CH3D isotopomers formed merely because the overall
reaction is reversible to some extent under the conditions of
the experiment. These data show that C–H bond activation
steps are exactly as reversible as the overall CH4–CO2 re-
action, as required by their rate-determining nature. These
results and conclusions resemble those obtained from sim-
ilar experiments on Ru [31], Pt [32], Ir [33], and Rh [34].
Other studies [24,65,68] have assumed, without specific ev-
idence, that C–H bond activation is reversible during CO2
reforming. CO2 reforming [68] and partial oxidation [69]
of CH4/CD4 mixtures on Ni/SiO2 showed that C–H bond
activation was reversible and that chemical conversion and
CHxD4−x formation rates were similar, a conclusion that
merely indicates that the overall reaction was reversible at
the high chemical conversions inappropriately used in these
measurements (e.g., 84.7% at 973 K for CO2 reforming
[68]).
The H/D ratio in the dihydrogen formed during CO2
reforming of equimolar CH4–CD4 mixtures on 7 wt%
Ni/MgO-A at 873 K is 1.56. These data confirm the higher
reactivity of CH4 relative to CD4 and the kinetic isotope
effect values measured from independent CH4–CO2 and
CD4–CO2 reactants (1.62). Dihydrogen molecules showed
a binomial isotopomer distribution, indicating that recom-
binative hydrogen desorption steps are quasi-equilibrated,
a conclusion confirmed by the isotopic data presented
next.
Fig. 10. Extent of water–gas-shift equilibration at different reaction tem-
peratures as a function of space velocity on 7 wt% Ni/MgO-A (25
kPa CH , 25 kPa CO , 100 kPa total pressure, balance He, η
=
4
2
WGS
([P ][P
])/([P ][P
]K
WGS
)).
H
2
CO
H O
2
CO
2
and desorption of water occur via quasi-equilibrated elemen-
tary steps.
Dihydrogen and water isotopomers were measured by on-
line mass spectrometry during reactions of CH4/CO2/D2
(1:1:0.2) mixtures on 7 wt% Ni/MgO-A at 973 K. The H/D
fraction expected if all H atoms in chemically converted CH4
molecules and all D atoms in D2 molecules contributed to
the pool of surface intermediates forming water and dihydro-
gen is 3.81. Measured H/D ratios in water and dihydrogen
molecules formed during CH4/CO2/D2 reaction were 3.66
and 3.90, respectively. Thus, complete equilibration occurs
between dihydrogen and water molecules and their respec-
tive chemisorbed precursors. Isotopomer distributions for
both water and dihydrogen were binomial (Fig. 11), consis-
tent with quasi-equilibrated recombination steps involving
H* and OH* to form water and H* and H* to form dihy-
drogen. The quasi-equilibrated nature of the steps forming
water and dihydrogen, the rapid isotopic equilibration of
CO–CO2 mixtures discussed below, and the expected equi-
librium for molecular adsorption of CO under these condi-
tions would require that water–gas-shift reactions also be at
3.5. Isotopic probes of the reversibility of hydrogen and
hydroxyl recombination steps that form H2 and H2O
Some water forms during CO2 reforming of CH4 because
chemisorbed hydrogen (from CH4) reacts with chemisorbed
oxygen (from CO2) in an overall reaction that mimics the
stoichiometry of the reverse water–gas shift. This reac-
tion and therefore all elementary steps required are quasi-
equilibrated during CH4 reforming with H2O or CO2, as
shown from the effluent concentrations measured under all
reaction conditions (Fig. 10). Thus, we expect that formation

J. Wei, E. Iglesia / Journal of Catalysis 224 (2004) 370–383
379
Fig. 11. Distribution of dihydrogen (a) and water (b) isotopomers during reactions of CH /CO /D mixtures on 7 wt% Ni/MgO-A (973 K, 16.7 kPa CH ,
4
2
2
4
16.7 kPa CO , 3.3 kPa D , 100 kPa total pressure, balance Ar).
2
2
equilibrium during CO2 and H2O reforming on Ni catalysts,
as indeed found under all reaction conditions in this study
(Fig. 10).
3.6. Isotopic probes of reversible coreactant activation
steps
The reversibility of CO2 activation steps during CO2–
CH4 reactions was probed using 12CH4/12CO2/13CO (1:1:
0.4) reactant mixtures on 7 wt% Ni/MgO-A at 973 K. Mea-
sured 13C fractions were very similar in CO (0.24) and CO2
(0.23) molecules leaving the catalyst bed, even when the
overall reforming reaction was far from equilibrium (η =
0.06). These 13C contents reflect complete chemical and
isotopic equilibration between CO and CO2 molecules in
the contacting gas phase; they confirm that CO2 activation
steps are much faster than kinetically relevant C–H bond
activation steps, and that CO2 dissociation to CO and oxy-
gen occurs in both directions many times during each CH4
chemical conversion turnover. CO2 activation steps are re-
versible and quasi-equilibrated during CH4–CO2 reactions
and we conclude from kinetic analogies with CH4–H2O re-
actions that H2O activation steps are also reversible and
quasi-equilibrated during the latter reaction. This conclusion
is consistent with the equilibrated nature of water–gas-shift
reactions during steam reforming; it will be confirmed in fu-
ture studies via 18O isotopic tracing using CH4/H218O/C16O
reactants.
Scheme 1. Sequence of elementary steps for CH -reforming and wa-
4
ter–gas-shift reactions on Ni-based catalysts (→ irreversible step,
quasi-equilibrated step, ꢀ reversible step, k is the rate coefficient and K
i
i
is the equilibrium constant for a given step i).
12CO2/13CO reactions. H2 and H2O form in quasi-equilibra-
ted steps.
This evidence led us to consider the sequence of el-
ementary steps shown in Scheme 1, which accounts for
H2O and CO2 reforming, CH4 decomposition, and water–
gas-shift reactions and also provides steps required to form
chemisorbed carbon precursors to carbon filaments. CH4
decomposes to chemisorbed carbon (C*) via sequential
elementary H-abstraction steps, which become faster as
H atoms are sequentially abstracted from CH4 reactants.
Density-functional theory led to an activation energy of
142 kJ/mol for the first H-abstraction step in CH4 on Ni
clusters, which decreased to 25–40 kJ/mol for CH2 for-
mation from CH3 [70]. This cascade process leads to low
3.7. Elementary steps involved in CH4-reforming reactions
These isotopic tracing and exchange measurements con-
firmed the mechanism first proposed based on measured
kinetic effects of reactant and product concentrations of
forward reaction rates. The sole kinetic relevance of CH4
activation is consistent with the results of CH4/CO2/D2,
CH4/CD4/CO2, and 12CH4/12CO2/13CO reactions. The qua-
si-equilibrated nature of CO2 activation was confirmed by
the similar 13C fraction in CO and CO2 during 12CH4/

380
J. Wei, E. Iglesia / Journal of Catalysis 224 (2004) 370–383
Fig. 12. Forward CH turnover rates for CO (a) and H O (b) reforming of CH on different metal clusters as a function of metal dispersion on various
4
2
2
4
supports (873 K, 20 kPa CH , (Q) ZrO , (") γ -Al O , (E) ZrO –CeO as support, (!) MgO-A, (F) MgO-B).
4
2
2
3
2
2
CHx coverages and to C* as the most abundant carbon-
containing reactive intermediate. Chemisorbed carbon is
then removed using CO2 or H2O as coreactants. These el-
ementary steps are consistent also with kinetic and isotopic
measurements on Ru [31], Pt [26], Ir [32], and Rh [33] cata-
lysts. When exposed metal atoms (*) are the most abundant
surface species, only the rate constant for the activation of
the first C–H bond in CH4 appears in the rate expression and
reaction rates become first order in CH4 and independent
of the presence or concentration of CO2 or H2O coreac-
tants. We note that these elementary steps provide pathways
for reactions of CH4 with either CO2 or H2O and also
reactions. Next, we compare measured C–H bond activation
turnover rates on noble metal (Rh, Ru, Pt, Ir) and Ni cata-
lysts.
∗
3.8. Effect of the identity of metal clusters on C–H bond
activation
Metal dispersion and support effects on forward CH4
turnover rates for CH4–H2O, CH4–CO2, and CH4 decom-
position reactions were reported earlier on individual no-
ble metal catalysts (Rh, Ru, Pt, Ir) [31–34]. For each no-
ble metal, CH4-reforming turnover rates (normalized by
exposed metal atoms and corrected by their approach to
equilibrium) increased with increasing metal dispersion, but
they were not influenced by the support (Al2O3, ZrO2, and
ZrO2–CeO2) [31–34]. The demonstrated kinetic irrelevance
of coreactant activation implies that even when supports cat-
alyze coreactant activation, they cannot influence overall re-
action rates.
Fig. 12 shows forward CH4-reforming turnover rates with
CO2 (Fig. 12a) and H2O (Fig. 12b) on supported Pt, Ir, Rh,
and Ru catalysts with a relatively wide dispersion range and
on the three Ni catalysts of this study. The exclusion of
transport and thermodynamic artifacts, the rigorous correc-
tion to the forward rates, and the measurement of rates nor-
malized by the relevant exposed surface metal atoms make
these rate comparisons a rigorous reflection of the kinetic
reactivity of such metal surfaces. One previous comparison
of reactivity was based on CH4 conversion during CO2 re-
forming for various metals on Al2O3 supports, but neither
metal dispersions nor turnover rates were reported [74]. An-
other study compared CH4 turnover rates on noble metals
supported on Al2O3–MgO and observed a reactivity order
(Ru > Rh > Ir > Ni > Pt) [75], very different from the or-
der reported here for both CO2–CH4 and H2O–CH4 reac-
for water–gas-shift reactions (H2O + CO KWGS CO2 + H2),
which have been typically, but inappropriately and nonrigor-
ously, treated as independent kinetic processes during CH4
reforming.
The general features of this sequence are in essential
agreement with previous mechanistic proposals on sup-
ported Ni catalysts [21,59,71] and supported noble catalysts
[72,73], which lacked steps required for complete water–
gas-shift sequences and which did not provide direct evi-
dence for the reversibility of the elementary steps [21,59].
It appears that some accumulation of unreactive carbon
in some of these previous studies, possibly as a result of
transport restrictions, led to the incorrect conclusion that
reactions between chemisorbed carbon and oxygen limit
CO2-reforming rates on Ni surfaces [21].
Detailed kinetic and isotopic tracer studies of CH4–H2O,
CH4–CO2, and CH4 decomposition and water–gas-shift re-
actions on noble metals (Rh, Ru, Pt, Ir) have been reported
elsewhere [31–34]. These studies also provided kinetic and
isotopic evidence for the identity and reversibility of elemen-
tary steps involved in CH4-reforming reactions and for the
sole kinetic relevance of C–H bond activation and mecha-
nism equivalence among CH4-reforming and decomposition

J. Wei, E. Iglesia / Journal of Catalysis 224 (2004) 370–383
381
tions. These differences arise, at least in part, from disper-
sion differences among the noble metal catalysts compared
in Ref. [75], which lead to marked differences in reactivity
(Fig. 12), and to varying approaches to equilibrium at the
different CH4 conversions used in this previous comparison.
The data shown in Fig. 12 appear to be the first rigorous
comparison of the C–H bond activation rates on supported
noble metal and Ni clusters over a common dispersion range.
Among these different metal clusters, Pt clusters show the
most reactive surfaces for C–H bond activation.
3.9. Mechanistic implications for the dynamics of carbon
filament formation during CH4 reforming and
decomposition
Here, we discuss briefly the implications of the proposed
CH4 reaction mechanism for kinetic treatments of filamen-
tous carbon formation. All rate constants and steps refer to
those in the Scheme 1. The concentration (C*) or thermo-
∗
dynamic activity (aC ) of chemisorbed carbon depends on
prevalent pressures and on the rate and equilibrium con-
stants for individual elementary steps. A pseudo-steady-state
analysis based on C* and * as the most abundant reactive in-
termediates gives
Fig. 13. Effect of P
P
/P
or P
P
/P on carbon for-
H O
2
H
CO CH
CO
CH
2
4
2
mation rate on Ni/MgO-A and Ni/MgO-B at 8734K (P
P
/P
or
CO
2
CO CH
4
P
P
/P
ratios were changed by varying CH and CO or H O
H
CH
H O 4 2 2
2
4
partial pressur2es or space velocity during CH /CO and CH /H O reac-
4
2
4
2
tions. (!,") CH /CO reaction, (P,Q) CH /H O reaction, and (E) addi-
4
2
4
2
ꢃ
ꢄ
tion of H in CH /CO reaction).
2
4
2
k1P
k
−3PCO
CH4
∗
aC = kc
+
PCO
.
(7)
K2k3K4P
K2k3K4P
CO2
CO2
k3
pressure, CO2 or H2O pressure, H2 addition). Electron mi-
croscopy detected the formation of hollow multiwalled car-
bon nanotubes with diameters similar to those of the Ni crys-
tallites attached to their advancing tips. Comparing the car-
bon formation rate on Ni/MgO-A and Ni/MgO-B, we found
When carbon removal by surface oxygen step (C∗ + O∗ ꢁ
CO∗ + ∗) is quasi-equilibrated,
k
−3
k
−3(CO) ꢀ k1(CH4),
(8)
(9)
that for a given value of PCH PCO/PCO or PCH PH /PH O
,
4
2
4
2
2
Eq. (7) becomes
carbon formation rates (per surface Ni atom) increase with
increasing Ni crystallite diameter, apparently as a result of
the more facile nucleation and lower thermodynamic car-
bon activity for the larger carbon filaments prevalent in such
larger crystallites.
kck1P PCO kP PCO
CH4
CH4
∗
aC
or
=
=
=
K2k3K4P
P
CO2
CO2
kPCH PH
kꢁPCH PH
4 2
4
2
∗
aC
=
,
(10)
The observed linear dependence of carbon formation
rates on these two parameters is consistent with diffusion-
limited growth of filaments driven by a gradient in ac-
tivity between the exposed Ni surface and the carbon
filament [76,77], which increases as PCH PCO/PCO or
KWGS
P
H2O
P
H2O
in which PCH PCO/PCO and PCH PH /PH O are related
4
2
4
2
2
to each other by the water–gas-shift equilibrium constant;
thus, these two ratios are proportional to each other at
each given temperature. The thermodynamic carbon activ-
ity (aC ) that provides the driving force for carbon filament
formation is proportional to PCH PCO/PCO (or equivalently
4
2
PCH PH /PH O increases. The rate of carbon formation for a
4
2
2
∗
given value of PCH PCO/PCO or PCH PH /PH O increased
4
2
4
2
2
with increasing Ni particle size, even though exposed Ni sur-
face areas are lower and carbon must diffuse through larger
Ni crystallites. These effects of crystallite size appear to
reflect the lower carbon activity required to nucleate large
filaments (the x intercept in Fig. 13) and the temperature
dependence of the groupings of kinetic and thermodynamic
constants that relate PCH PCO/PCO and PCH PH /PH O to
4
2
PCH PH /PH O).
4
2
2
Fig. 13 shows carbon formation rates as a function of
these two ratios, which were varied by changing CH4, CO2,
or H2O partial pressures or the space velocity, as well
as by adding H2 during CH4–CO2 reactions at 873 K on
Ni/MgO-A and Ni/MgO-B catalysts. For PCH PCO/PCO
4
2
4
2
4
2
2
ratios in the range of 1.1–4.5 kPa (PCH PH /PH O ∼ 2.7–
4
2
2
the carbon thermodynamic activity. A more detailed treat-
ment of filament formation rate and of parallel mechanisms
for catalyst deactivation via encapsulation of Ni particles
will be reported elsewhere [35].
11.0 kPa), carbon formation rates increased linearly with
increasing carbon activity on 7 wt% Ni/MgO-A, irrespective
of how these ratios were varied (e.g., residence time, CH4

382
J. Wei, E. Iglesia / Journal of Catalysis 224 (2004) 370–383
4. Conclusions
[15] Z.L. Zhang, X.E. Verykios, Catal. Today 21 (1994) 589.
[16] M.C.J. Bradford, M.A. Vannice, Appl. Catal. A 142 (1996) 97.
[17] U. Olsbye, T. Wurzel, L. Mleczko, Ind. Eng. Chem. Res. 36 (1997)
5180.
[18] T. Osaki, T. Horuchi, K. Suzuki, T. Mori, Appl. Catal. A 155 (1997)
229.
[19] V.C.H. Kroll, G.J. Tjatjopoulos, C. Mirodatos, Stud. Surf. Sci. Ca-
tal. 119 (1998) 753.
[20] S. Wang, G.Q. Lu, React. Eng. Pollut. Prev. (2000) 75.
[21] S. Bebelis, A. Zeritis, C. Tiropani, S.G. Neophytides, Ind. Eng. Chem.
Res. 39 (2000) 4920.
[22] K. Hou, R. Hughes, Chem. Eng. J. 82 (2001) 311.
[23] V.A. Tsipouriari, X.E. Verykios, Catal. Today 64 (2001) 83.
[24] A. Becerra, M.E. Iriarte, A.E. Castro-Luna, React. Kinet. Catal. Lett.
79 (2003) 119.
[25] R.W. Stevens, S.S.C. Chuang, J. Phys. Chem. B 108 (2004) 696.
[26] K. Takehira, T. Shishido, P. Wang, J. Catal. 221 (2004) 43.
[27] X.E. Verykios, Appl. Catal. A 255 (2003) 101.
[28] A. Shamsi, C.D. Johnson, Catal. Today 84 (2003) 17.
[29] Z. Chen, Y. Yan, S.S.E.H. Elnashaie, AIChE J. 49 (2003) 1250.
[30] M. Zanfir, A. Gavriilidis, Chem. Eng. Sci. 58 (2003) 3947.
[31] J. Wei, E. Iglesia, J. Phys. Chem. B, in press.
Isotopic studies and forward rate measurements led to
a simple and comprehensive mechanistic picture for CH4–
CO2 and CH4–H2O reforming, CH4 decomposition, and
water–gas-shift reactions on Ni-based catalysts. The mech-
anistic equivalence among CH4-reforming and decomposi-
tion reactions was confirmed by identical forward CH4 re-
action rates, rate constants, kinetic isotopic effect, and acti-
vation energies measurements for the three CH4 reactions.
CH4 reaction rates were limited solely by C–H bond activa-
tion steps and unaffected by the identity or concentration of
coreactants. Isotopic tracer and exchange studies confirmed
that C–H bond activation is the only kinetically relevant step
and that it is exactly as reversible as the overall reforming re-
action. Isotopic evidence also showed that activation of H2,
CO2, and H2O was reversible and quasi-equilibrated dur-
ing CO2 and H2O-reforming reactions of CH4. The rigorous
exclusion of transport artifacts and the accurate correction
of net rates for thermodynamic constraints led to the first
comparison of true turnover rates among Ni and noble metal
catalysts. The elementary steps proposed and confirmed by
kinetic and isotopic evidence provide a rigorous basis for
treating the dynamics of carbon formation during CH4 re-
forming. The thermodynamic activity of chemisorbed car-
bon species, which provides the kinetic driving force for fila-
ment formation, and the rate of carbon formation depend lin-
[32] J. Wei, E. Eglesia, J. Phys. Chem. B 108 (2004) 4094.
[33] J. Wei, E. Iglesia, Angew. Chem. Int. Edit., in press.
[34] J. Wei, E. Iglesia, J. Catal. (2004), in press.
[35] J. Wei, E. Iglesia, manuscript in preparation.
[36] K.S. Suslick, T. Hyeon, M. Tang, Chem. Mater. 8 (1996) 2172.
[37] G.L. Price, E. Iglesia, Ind. Eng. Chem. 28 (1989) 839.
[38] E. Ruckenstein, Y.H. Hu, Appl. Catal. A 154 (1997) 185.
[39] K. Tosishige, O. Yomazaki, Y. Chen, K. Yokoyama, X. Li, K. Fuji-
moto, Catal. Today 45 (1998) 35.
[40] D.R. Stull, F. Edgar, J. Westrum, G.C. Sinke, In the Thermodynamics
of Organic Compounds, Krieger, Malabar, FL, 1987.
[41] M. Boudart, G. Djega-Mariadassou, The Kinetics of Heterogeneous
Catalytic Reactions, Princeton Univ. Press, Princeton, NJ, 1984.
[42] A.J. Muscat, R.J. Madix, J. Phys. Chem. 100 (1996) 9807.
[43] K. Christmann, O. Schober, G. Ertl, M. Neumann, J. Chem. Phys. 60
(1974) 4528.
[44] M.C.J. Bradford, M.A. Vannice, Appl. Catal. A 142 (1996) 73.
[45] K. Ahmed, K. Foger, Catal. Today 63 (2000) 479.
[46] S. Wang, G. Lu, Energy Fuels 12 (1998) 248.
[47] L.M. Aparico, A.G. Ruiz, I.R. Ramoa, Appl. Catal. A 170 (1998) 177.
[48] O. Tokunaga, S. Ogasawara, React. Kinet. Catal. Lett. 39 (1989) 69.
[49] T. Horiuchi, K. Sakuma, T. Fukui, Y. Kubo, T. Osaki, T. Mori, Appl.
Catal. A 144 (1996) 111.
early on the prevalent PCH PCO/PCO (or PCH PH /PH O
ratios in the contacting gas phase.
)
4
2
4
2
2
Acknowledgments
This study was supported by BP as part of the Methane
Conversion Cooperative Research Program at the University
of California. Helpful technical discussions with Drs. John
Collins and Theo Fleisch (BP) throughout these studies are
gratefully acknowledged.
[50] A.A. Lemonidou, L.A. Vasalos, Appl. Catal. A 228 (2002) 227.
[51] I.M. Ciobica, F. Frechard, R.A. van Santen, A.W. Kleyn, J. Hafner,
J. Phys. Chem. B 104 (2000) 3364.
References
[52] A.P.J. Jansen, H. Burghgraef, Surf. Sci. 344 (1995) 149.
[53] P. Kratzer, B. Hammer, J.K. Norskov, J. Chem. Phys. 105 (1996) 5595.
[54] H. Yang, J.L. Whitten, J. Chem. Phys. 96 (1992) 5529.
[55] I. Chorkendorff, I. Alstruo, S. Ullmann, Surf. Sci. 227 (1990) 291.
[56] B. Olgaard Nielsen, A.C. Luntz, P.M. Holmblad, I. Chorkendorff,
Catal. Lett. 32 (1995) 15.
[57] T.P. Beebe, D.W. Goodman, B.D. Kay, J.T. Yates, J. Chem. Phys. 87
(1987) 2305.
[58] X. Jiang, D.W. Goodman, Catal. Lett. 4 (1990) 173.
[59] J.A. Dumesic, D.F. Rudd, J.E. Rekoske, A.A. Trevino, The Microki-
netics of Heterogeneous Catalysis, Am. Chem. Society, Washington,
DC, 1993.
[1] V.F. Fischer, H. Tropsch, Brennstoff-Chemie 3 (1928) 39–46.
[2] J.H. Edwards, A.M. Maitra, Fuel Process. Technol. 42 (1995) 269.
[3] S.B. Wang, G.Q. Lu, Energy Fuels 10 (1996) 896.
[4] M.C.J. Bradford, M.A. Vannice, Catal. Rev.-Sci. Eng. 41 (1999) 1.
[5] J.W. Snoeck, G.F. Froment, Ind. Eng. Chem. Res. 41 (2002) 3548.
[6] H.S. Bengarrd, J.K. Norskov, J. Sehested, B.S. Clause, L.P. Nielsen,
A.M. Molenbroek, J.R. Rostrup-Nielsen, J. Catal. 209 (2002) 365.
[7] D.E. Resasco, W.E. Alvarez, F. Pompeo, L. Balzano, J.E. Herrera,
B. Kitiyanan, A. Borgna, J. Nano. Res. 4 (2002) 131.
[8] W.W. Akres, D.P. Camp, AIChE J. 1 (1955) 471.
[60] K. Otsuka, S. Kobayashi, S. Takenaka, J. Catal. 200 (2001) 4.
[61] T. Osaki, T. Horiuchi, K. Suzuki, T. Mori, Catal. Lett. 44 (1997) 19.
[62] Y. Schuurman, V.C.H. Kroll, P. Ferreira-Aparicio, C. Mirodatos, Catal.
Today 38 (1997) 129.
[63] V.C.H. Kroll, H.M. Swaan, S. Lacombe, C. Mirodatos, J. Catal. 164
(1997) 387.
[9] N.M. Bodrov, L.O. Apel’baum, Kinet. Katal. 5 (1964) 614.
[10] N.M. Bodrov, L.O. Apel’baum, M.I. Temkin, Kinet. Katal. 8 (1967)
821.
[11] T.Q.P. Quach, D.J. Rouleau, Appl. Chem. Biotechnol. 25 (1975) 445.
[12] P. Munster, H.J. Grabke, J. Catal. 72 (1981) 279.
[13] J. Xu, G.F. Froment, AIChE J. 35 (1989) 88.
[14] J. Xu, G.F. Froment, AIChE J. 35 (1989) 97.
[64] J. Chang, S. Park, J.W. Yoo, J. Park, J. Catal. 195 (2000) 1.

J. Wei, E. Iglesia / Journal of Catalysis 224 (2004) 370–383
383
[65] S. Tang, L. Ji, J. Lin, H.C. Zeng, K.L. Tan, K. Li, J. Catal. 194 (2000)
[71] J.A. Lercher, J.H. Bitter, W. Hally, W. Niessen, K. Seshan, Stud. Surf.
Sci. Catal. 101 (1996) 463.
424.
[66] J.J. Gamman, G.J. Millar, G. Rose, J. Drennan, J. Chem. Soc., Faraday
Trans. 94 (1998) 701.
[67] L.M. Aparicio, J. Catal. 165 (1997) 262.
[68] V.C.H. Kroll, H.M. Swaan, C. Mirodatos, J. Catal. 161 (1996) 409.
[69] V.C.H. Kroll, P.S. Delichure, C. Mirodatos, Kinetic. Catal. 37 (1996)
698.
[72] S.M. Stagg, Stud. Surf. Sci. Catal. 111 (1997) 543.
[73] A.M. O’Connor, F.C. Meunier, J.R.H. Ross, Stud. Surf. Sci. Catal. 119
(1998) 819.
[74] F. Solymosi, G. Kutsan, A. Erdohelyi, Catal. Lett. 11 (1991) 149.
[75] J.R. Rostrup-Nielsen, J.H. Bak Hansen, J. Catal. 144 (1993) 38.
[76] I. Alstrup, T. Tavares, C.A. Bernardo, O. Sorensen, J.R. Rostrup-
Nielsen, Mater. Corrosion 49 (1998) 367.
[70] H. Burghgraef, A.P.J. Jansen, R.A. van Santen, Surf. Sci. 324 (1995)
345.
[77] S.A. Safvi, E.C. Bianchini, C.R.F. Lund, Carbon 29 (1991) 1245.