Development of a molecular kinetic scheme for methane partial oxidation over a Rh/α-Al2O3 catalyst

Article abstract of DOI:10.1016/j.jcat.2006.03.018

A kinetic investigation on catalytic partial oxidation of methane over a Rh/α-Al₂O₃ catalyst was addressed using a structured annular reactor specifically designed for studying ultra-fast reactions at high temperatures and very short contact times, out of thermodynamic control. Information on the kinetics of O₂ consumption, so far not available in the literature, were collected with CH₄/O₂/inert-gas tests at varying temperatures, contact times, and diluting gases. By also running CO/O₂/N₂ and H₂/O₂/N₂ tests, the relative activities of CH₄-, CO-, and H₂-rich combustions were identified. The activation of methane was found to be the rate-determining step of steam reforming based on the results of CH₄/H₂O/N₂ tests, in line with recent literature reports. The bulk of the data confirmed that methane partial oxidation proceeds according to an indirect reaction scheme consisting of CH₄ deep oxidation, secondary reactions of CH₄ reforming responsible for syngas formation, the water-gas shift reaction, and consecutive oxidations of H₂ and CO. Molecular kinetic expressions are proposed to explain the wide-ranging experimental campaign.

Full text of DOI:10.1016/j.jcat.2006.03.018

Journal of Catalysis 241 (2006) 1–13
www.elsevier.com/locate/jcat
Development of a molecular kinetic scheme for methane partial oxidation
over a Rh/α-Al O catalyst
2
3
∗
Ivan Tavazzi, Alessandra Beretta , Gianpiero Groppi, Pio Forzatti
NEMAS – Centro di Eccellenza per l’Ingegneria dei Materiali e delle Superfici Nanostrutturate, Dipartimento di Chimica,
Materiali e Ingegneria Chimica “Giulio Natta ,” Politecnico di Milano, Piazza Leonardo da Vinci 32, 20133 Milano, Italy
Received 4 January 2006; revised 14 March 2006; accepted 22 March 2006
Available online 6 May 2006
Abstract
A kinetic investigation on catalytic partial oxidation of methane over a Rh/α-Al O catalyst was addressed using a structured annular reactor
2
3
specifically designed for studying ultra-fast reactions at high temperatures and very short contact times, out of thermodynamic control. Information
on the kinetics of O consumption, so far not available in the literature, were collected with CH /O /inert-gas tests at varying temperatures, contact
2
4
2
times, and diluting gases. By also running CO/O /N and H /O /N tests, the relative activities of CH -, CO-, and H -rich combustions were
2
2
2
2
2
4
2
identified. The activation of methane was found to be the rate-determining step of steam reforming based on the results of CH /H O/N tests,
4
2
2
in line with recent literature reports. The bulk of the data confirmed that methane partial oxidation proceeds according to an indirect reaction
scheme consisting of CH deep oxidation, secondary reactions of CH reforming responsible for syngas formation, the water–gas shift reaction,
4
4
and consecutive oxidations of H and CO. Molecular kinetic expressions are proposed to explain the wide-ranging experimental campaign.
2
©
2006 Elsevier Inc. All rights reserved.
Keywords: Catalytic partial oxidation; Methane; Rh/α-Al O ; Kinetic investigation; Indirect kinetic scheme; C–H bond activation; Annular reactor
2
3
1
. Introduction
The catalytic partial oxidation of methane is currently con-
units (APUs) on heavy-duty vehicles to supply power to auxil-
iary cab devices (e.g., conditioning systems, GPS) and trailers
(cryogenic circuits). On-board generation of syngas may also
be applied on conventional ICE vehicles to speed up the cold-
start phase of catalytic converters and to serve as a reducing gas
for NO trap regeneration and for the SCR of NO . Recently,
combustion systems using fuel-rich catalytic combustion be-
fore lean-premixed combustion have been tested in advanced-
performance gas turbines (ultra-low-emission GTs) [3].
sidered a promising nonconventional process for syngas gen-
eration. Compared with the existing homogeneous process, the
presence of a catalyst allows to considerably shorten the reac-
tion times, decrease the volumes, and lower the operating tem-
peratures, thus enabling the use of air instead of pure O2 as the
oxidant, and inhibiting carbon black formation [1]. The features
of reactor simplicity and compactness, together with the rapid
response to transient load demands, make this process suitable
for stationary and mobile applications related to the develop-
ment of a medium- to small-scale technology for the production
of syngas and H2 [2]. Such novel applications include small
hydrogen production stations for the refueling of H2-fueled
vehicles, small-scale distribution networks, and fuel cells for
residential power generation. On-board applications deal with
the use of solid oxide fuel cells (SOFCs) for auxiliary power
x
x
Among several effective catalysts for hydrocarbon partial
oxidation [4–7], Rh has been proposed as the best-performing
active element and the least-selective catalyst to carbon for-
mation [8,9]. Schmidt and coworkers studied methane partial
oxidation over Rh, Pt, and Rh/Pt structured catalysts in adia-
−2
−4
batic reactors at contact times as short as 10 –10 s [8,10]
and observed high selectivity to syngas and product distribu-
tion far from the thermodynamic equilibrium, evaluated for the
water–gas shift and steam reforming reactions. A direct route
to H2 and CO was invoked. In previous and recent work over
Rh-coated supports, these authors reported almost complete
conversion of CH4 with selectivity to H2 and CO > 90% [11].
*
Corresponding author. Fax: +39 02 2399 3318.
E-mail address: alessandra.beretta@polimi.it (A. Beretta).
0021-9517/$ – see front matter © 2006 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2006.03.018

2
I. Tavazzi et al. / Journal of Catalysis 241 (2006) 1–13
The operating conditions used by Schmidt and coworkers were
extremely severe with respect to both the temperature and re-
actant concentrations, which complicates the discrimination
of the role of thermodynamics, of heterogeneous and homo-
geneous reactions, and of transport phenomena. Chang and
Heinemann [12] threw light on these aspects by demonstrat-
ing that previous results obtained over Co/MgO by Choudary
and coworkers [13–16], who observed a selectivity to syngas
higher than the equilibrium value at short contact times and
suggested the direct formation of H2 and CO from methane,
were biased by incorrect measurements of the actual catalyst
surface temperature. Verykios and coworkers investigated the
reaction mechanism under more favorable conditions, at high
dilution of the gas mixture and of the catalyst [17], over VIII
group noble metals supported on Al2O3, SiO2, and TiO2, and
proposed the indirect production of syngas [18]. Baerns and
coworkers adopted a pulse-feed technique to minimize the hot-
spot temperature on the catalyst [19–21]. They compared the
performance of metallic Rh and Rh/Al2O3 and proposed a re-
dox mechanism in which the initial step of methane dissociation
is catalyzed by partially oxidized Rh sites, stabilized on the
supported system, and disfavored by Rh2O3 species, predom-
inant on the unsupported catalyst. Similar conclusions were
derived by Marin and coworkers [22,23], who reported that the
product distribution is determined by the ratio of oxidized Rh
space velocity (GHSV) that were quantitatively explained by an
indirect kinetic scheme wherein the complexity of the reacting
system [24–29] was reduced to few main molecular stoichiome-
tries. The scheme includes an initial step of complete methane
combustion, followed by secondary reactions of methane steam
and CO reforming, responsible for syngas production, accom-
2
panied by the water–gas shift reaction and consecutive oxida-
tions of CO and H . In this study, effort was focused on defining
2
the main kinetic dependencies that affect the single reactions;
toward this purpose, the main reactions were isolated and stud-
ied separately.
2. Experimental
2.1. Catalyst
The kinetic study was carried out over a 0.5% (wt/wt) Rh/
α-Al O catalyst, prepared by grafting alumina powders with
2
3
Rh (CO) . A detailed description of the preparation procedure
4
12
and the characterization of the catalytic system has been given
previously [33–37]. Measurements of metal dispersion carried
out on fresh samples by H pulse chemisorption showed a Rh
2
dispersion of about 70%.
2.2. Annular reactor
(Rh2O3) to metallic Rh and proposed a mechanism in which
H2 and CO are formed as primary products. Deutschmann and
coworkers [24] proposed a combination of both direct and in-
direct pathways and gave a detailed surface reaction mecha-
nism consisting of 38 reactions among 6 gas-phase species and
Catalytic tests were performed in an annular duct reactor
[38,39] in which a catalytic layer (40–50 µm thick) is deposited
according to an optimized procedure [40] over a mullite tube
that is coaxially inserted into an outer quartz tube, giving rise to
an annular duct through which the gas mixtures flow in laminar
regime (Fig. 1). Catalyst loadings were on the order of 10 mg.
The mullite support also served as a well for a sliding K-type
thermocouple by which the axial profiles of the catalyst temper-
ature were measured. The correspondence of the measurements
to the actual temperatures on the catalyst surface was ensured
by the thermal equilibrium across the section of the ceramic
tube, due to the negligible contribution of the thermal disper-
sion along the mullite tube on the overall heat dissipation [41].
The average catalyst temperature was calculated as an arith-
metic average,
1
1 surface species. Similar models including only elementary
reactions characterized by surface science techniques have been
proposed by Froment et al. [25] and Schmidt et al. [26]. On
the basis of a microkinetic C1 model developed by adopting
a hierarchical multiscale methodology, Mhadeshwar and Vla-
chos [27] suggested that H2 is produced via steam reforming,
whereas CO is directly formed by partial oxidation of CH4.
Iglesia and coworkers corroborated the importance of eliminat-
ing diffusive limitations and accounting for thermodynamics to
properly study the kinetics of methane activation over several
catalysts [28,29].
The development of reaction kinetics is a crucial issue
in reactor design and optimization by mathematical models.
Notwithstanding several previous investigation, there is a gen-
eral lack of kinetic schemes validated in an operating range
relevant to practical applications. In this work, catalytic tests
were run under operating conditions suitable to obtain kineti-
cally informative data, wherein the chemical effects can be de-
coupled from thermodynamic and transfer phenomena effects.
Experiments were performed in a structured annular reactor
specifically designed to investigate the kinetics of very fast re-
actions [30], which allows to control the extent of mass transfer
limitations, to analyze the process far from the thermodynamic
equilibrium and to realize well-controlled temperature condi-
tions with minimum axial gradients.
ꢀ
Tiꢀxi
T =
,
L
where Ti are the temperatures measured along the axial coor-
dinate of the catalytic layer every 2 mm. The reactor was po-
sitioned inside a three-zone tube furnace (Carbolite) equipped
with an inner ceramic wall, heated by thermoelectric resis-
tances. Oven temperatures were recorded with a second K-type
thermocouple sliding inside a mullite tube located along the
heating wall. The effective dissipation of the reaction heat by
radiation helped realize catalytic tests at the expense of accept-
able thermal axial gradients (mostly below 5 K/cm) compared
with those obtained in packed-bed reactors.
In experiments with H2O cofeed, water was produced in situ
by reacting stoichiometric amounts of H2 and O2 (added to the
reactant mixtures) on a Pt layer deposited upstream from the
The data reported herein extend the findings of previous
studies [31,32] on the effects of temperature and gas hourly

I. Tavazzi et al. / Journal of Catalysis 241 (2006) 1–13
3
Rh coating. The Pt layer was placed outside the heated zone
of the furnace and exposed to room temperature, to prevent
methane activation. The efficiency of Pt in H2 burning was
tested by delivering the feed mixtures to the reactor heated at
low temperature (373 K), to ensure negligible activity of Rh
while preventing the condensation of water. At the flow rate
used in the partial oxidation and reforming experiments, H2
conversion was almost complete, and no other reactions were
observed. The residual H2 downstream from the Pt layer was
wall and bulk [Eq. (2)] in which the radial mass flow of each
species is equal to its overall formation rate. Due to high ax-
ial Péclet numbers, axial dispersion was neglected. Resistance
to interphase mass transport was described according to a cor-
relation for the Sherwood number [Eq. (3)], derived from the
literature and analogous to the Graetz–Nusselt problem of heat
exchange in annular ducts with boundary conditions of the
third type [39,42–45]. Molecular diffusivity was approximated
as binary diffusivity of the ith species in N2 and calculated
according to the Füller–Schettler–Giddings correlation [46].
<
0.1% (v/v), and its effect on the catalytic performance of Rh at
higher temperatures could be neglected given its extremely high
combustion rate over hot Rh surfaces by which H2 (together
with the stoichiometric residual O2) was completely depleted
at the very beginning of the catalytic layer.
O intraparticle mass transfer limitations were accounted for by
multiplying the rate of the oxidation reactions by a generalized
internal effectiveness factor [Eq. (4)], calculated on the basis of
2
the overall O consumption rate [47]. The effective diffusivity
2
of O2 was calculated according to the Wakao–Smith model (5)
assuming an average pore diameter of 100 nm (monomodal
2
.3. One-dimensional mathematical model of the annular
reactor
pore distribution) and a macropore void fraction (ε ) of 0.6,
M
determined on the basis of Hg porosimetry measurements.
The annular reactor was modeled according to a one-
dimensional heterogeneous isothermal model whose governing
equations, written in dimensionless form, are given in Table 1.
Gas-phase mass balances along the axial coordinate [Eq. (1)],
are coupled with equations of continuity between the catalyst
2.4. Operating conditions
Experiments were run at atmospheric pressure at a tem-
perature of 623–1193 K and a GHSV of 2.0 · 10 –1.1 ·
6
Fig. 1. Sketch of the annular reactor.
Table 1
Sketch of the annular duct and governing equations in the 1D model of the reactor
∗
∗
dFi
4
ꢁ
b
i
w
ꢂ
F
TOT
◦
Pe
·
= −
· Shloci · x − xi · F
(1)
(2)
m,i
∗
dz
1 + 1/R
TOT
ꢁ
ꢂ
ꢃ
b
w
Sh
Sh
· x − xi
=
ν
· α · r
loci
loci
i
i,j
i
j
j
(
−71.2zSh,i
)
−0.35
= Shinf + 6.874 · e
· (1000 · zSh,i
)
(3)
Generalized internal efficiency
S
C
√
ꢄ ꢅO2
ꢆ1/2
2
1
tanh(φ
)
∞
O2
φO2
η
=
·
D
eff,O2
· rv(c) · dc
φ
=
η
O2
=
(4)
(5)
O
δL · rv(CS
O2
∞
2
)
η
O
O2
2
0
2
D
eff
: random pore model (Wakao–Smith)
ε
μ
· (1 + 3 · εM)
2
M
D
= ε · DM +
· Dμ
eff,O
2
1
− εM

4
I. Tavazzi et al. / Journal of Catalysis 241 (2006) 1–13
07 NLkg
−1
h
−1
. Calibrated gas mixtures were obtained using
3. Results
1
cat
mass flow controllers and delivered to the reactor through stain-
less steel lines. Reactants were highly diluted (inert gas >90%)
to reduce the axial temperature gradients along the catalytic
layer. The composition of feed and product mixtures was mea-
sured by gas chromatography. N2, O2, CH4, CO, H2O, and CO2
were analyzed in a VARIAN 3400 (Varian Instrument Group)
equipped with a thermal conductivity detector (TCD), a molec-
ular sieve (5 Å) column, and a Porapak QS column with He as
carrier gas. H2 concentration was measured using Ar as carrier
gas in a MEGA 5340C (Carlo Erba Strumentazione) equipped
with a molecular sieve (5 Å) column and a TCD. N2 was used
as an internal standard for quantifying the net molar flows. At
a given feed composition and flow rate, the reaction tempera-
ture was progressively increased stepwise, and product analyses
were recorded under steady-state conditions, 20 min after each
new set point temperature was reached. Blank runs, performed
preliminarily over the whole temperature range with no catalyst
deposited on the mullite tube, indicated the chemical inertness
of the ceramic support and the absence of homogeneous re-
actions below 1173 K. In all experiments, mass balances of
converted C, H, and O were closed between 0.95 and 1.05.
3.1. CH + O tests: experimental effect of temperature and
4
2
contact time
A previous study [32] reported the results of CH /O tests
4
2
in an annular reactor in terms of the effects of temperature
and GHSV. Those data suggested the formulation of an in-
direct kinetic scheme. For the sake of clarity, we summarize
those previous conclusions herein by discussing independent
experiments in which the combined effect of temperature and
contact time on the same reaction system was investigated; re-
sults are reported in Fig. 2. At each GHSV, deep oxidation
of CH occurred in the low-temperature range (roughly be-
4
tween 673 and 843 K), with H O and CO the only species
2
2
produced; reactant conversion increased moderately with tem-
perature. A significant increase in conversion occurred with
further increases in temperature. O consumption was even-
2
tually limited by mass transfer, reaching an asymptotic value
that decreased with increasing GHSV. A significant net produc-
tion of H and CO started above the temperature (depending
2
on the GHSV) at which O2 entered diffusive limitation and in-
Fig. 2. Combined effect of temperature and GHSV on experimental (symbols) and calculated (lines) reactants conversion (a, b) and syngas selectivity (c, d) in
6
−1 −1
6
−1 −1
6
−1 −1
h ;
cat
experiments with the reacting system CH + O : (2) GHSV = 2.0·10 NLkg
h
, (") GHSV = 6.5·10 NLkg
h
, (Q) GHSV = 10.5·10 NLkg
4
2
cat
cat
(
- - -) model without consecutive oxidations of CO and H , (—) model with consecutive oxidations. Feed composition: CH /O /N = 4/2/94% v/v; atmospheric
2
4
2
2
pressure.

I. Tavazzi et al. / Journal of Catalysis 241 (2006) 1–13
5
Fig. 3. Axial net temperature profiles obtained by difference between the temperature inside the mullite tube and the temperature in the oven chamber. (a) Sharpening
of temperature profiles with increasing average catalyst temperature in the low temperature field (only H O and CO produced); (b) progressive smoothening of
2
2
temperature profiles at high temperature (O conversion diffusion limited, increasing production of H and CO). Feed composition: CH /O /N = 4/2/94% v/v;
2
2
4
2
2
6
−1 −1
GHSV = 2.0 · 10 NLkg
h
; atmospheric pressure.
cat
creased with temperature along with CH4 conversion, whereas
CO2 and H2O were progressively consumed (as shown in Fig. 2
by the monotonically increasing selectivity of synthesis gas).
The steep rise in reactant conversion at low temperature was ac-
companied by a significant heat release that produced relatively
pronounced peaks in the axial temperature profiles (Fig. 3).
The increasing formation of partial oxidation species at the ex-
alone. In contrast, the simulations systematically overestimated
H2 selectivity and predicted a stronger dependence of CO se-
lectivity on contact time than was observed in the experiments.
To decouple the net production of the two species, consecu-
tive oxidations of H2 and CO to H2O and CO2 were included
in the scheme [32]. For the same purpose, alternative reaction
paths, as proposed by Mhadeshwar and Vlachos [27], could
have been used that consider the oxidation of CH4 with direct
formation of CO and H2O. Nonetheless, the choice of adopt-
ing the stoichiometries of H2 and CO postcombustion allowed
independent study of the reactions taken into account in the
scheme. Simulations according to such an indirect-consecutive
scheme, with rate equations and kinetic parameters derived in
the present work and discussed in detail below, are reported in
Fig. 2 as solid lines.
pense of CO and H O and the enhanced efficiency of heat
2
2
dissipation by radiation at higher temperatures resulted in in-
creasingly smooth temperature profiles. Notably, a maximum
was always present at the beginning of the catalyst, followed by
a progressive decrease in temperature along the layer, consistent
with the occurrence of an exothermic–endothermic reaction se-
quence. An inspection of the temperature profiles also reveals
that along 90% of the catalytic layer, the temperature increases
were mostly <5 K and never exceeded 10 K, which, consid-
ering the wide range of temperatures investigated, can be con-
sidered isothermal conditions suitable for kinetic analysis. No-
tably, in previous partial oxidation experiments [31] performed
in an optimized packed-bed reactor with a catalyst dilution of
3
.2. CO- and H2-rich combustion tests
Experiments of CO and H2 combustion were performed un-
der rich conditions (CO/O2/N2 and H2/O2/N2 =4/1/95% v/v)
with the intent of reproducing the same conditions that oc-
cur during methane partial oxidation. Tests were carried out
1
:5, axial temperature gradients rose up to 80 K at the same
5
−1 −1
h .
cat
reactant composition and a GHSV of 5 · 10 NLkg
6
6
at 373–973 K at GHSVs of 2.0 · 10 , 4.0 · 10 , and 8.0 ·
Fig. 2 reports simulations according to a purely combustion-
reforming water–gas shift scheme (dashed lines), wherein
6
−1 −1
0 NLkg h .
cat
1
CO-rich combustion started at about 503 K independent
of the contact time, whose reduction instead entailed a de-
crease in reactant conversion (Fig. 4a). Considering the run at
secondary reactions (steam-reforming, CO -reforming, and
2
water–gas shift) were described with rate equations and kinetic
parameters derived from a previous study in a packed-bed re-
actor [31], with the kinetics of methane combustion also de-
duced from a previous study [32]. This model satisfactorily
predicted the trends of reactant conversion but could not cor-
rectly describe the distribution of products. Experimentally, the
6
−1 −1
h , representative of the gen-
cat
a GHSV of 2.0 · 10 NLkg
eral trend of experiments, reactant consumption gradually in-
creased with increasing temperature up to about 603 K. Above
this temperature, the slope of the conversion curve decreased
moderately, possibly due to the onset of internal mass transfer
limitations; yet above 683 K, the dependence of CO consump-
tion on temperature increased again. Inspection of temperature
profiles (Fig. 4b) shows the progressive growth of a hot spot
with increasing average catalyst temperature, roughly up to
603 K; at higher reaction temperatures, axial profiles progres-
sively smoothed, and the hot spot moved upstream from the
selectivity to H decreased markedly at increasing GHSV while
2
the selectivity to CO remained almost unchanged, especially at
high temperature; this resulted in H /CO ratios well below 2
2
at low temperatures and high GHSVs, suggesting that the for-
mation of CO and H2 was partly independent and not fully
constrained by the stoichiometry of the reforming reactions

6
I. Tavazzi et al. / Journal of Catalysis 241 (2006) 1–13
6
−1 −1
h ;
cat
Fig. 4. (a) CO conversion in CO combustion tests under rich conditions at different space velocities. Experimental data: (2) GHSV = 2.0 · 10 NLkg
6
−1 −1
6
−1 −1
h . Model predictions reported as continuous lines. (b) Axial net temperature profiles at
cat
(
") GHSV = 4.0 · 10 NLkg
h
; (Q) GHSV = 8.0 · 10 NLkg
cat
6
−1 −1
h
cat
GHSV = 2.0 · 10 NLkg
on increasing catalyst temperature. Feed composition: CO/O /N = 4/1/95% (v/v), atmospheric pressure.
2
2
6
−1 −1
h ;
cat
Fig. 5. (a) H conversion in H combustion tests under rich conditions at different space velocities. Experimental data: (2) GHSV = 2.0 · 10 NLkg
2
2
6
−1 −1
6
−1 −1
h . Model predictions reported as continuous lines. (b) Axial net temperature profiles at
cat
(
") GHSV = 4.0 · 10 NLkg
h
; (Q) GHSV = 8.0 · 10 NLkg
cat
6
−1 −1
h
cat
GHSV = 2.0 · 10 NLkg
on increasing catalyst temperature. Feed composition: H /O /N = 4/1/95% (v/v), atmospheric pressure.
2
2
2
catalytic layer, suggesting the onset of gas-phase reactions pos-
sibly initiated by the catalyst.
and thus improving the chemical consistency of the kinetic
scheme. Catalytic runs were performed on a second Rh-coated
mullite tube with a catalyst layer similar in weight, length, and
thickness to that used in the experiments described above.
4
H oxidation under rich conditions started at about 373 K
2
(Fig. 5a), at temperatures more than 100 K lower than those
observed in the CO-rich combustion experiments. A stepwise
trend was observed; the reaction was in fact so rapid that a dif-
fusive regime was established at low temperatures. At about
Preliminarily, experiments of CH partial oxidation were re-
6
−1 −1
h to compare the
cat
peated at a GHSV of 2.6 · 10 NLkg
performance of the two reactors. Data reported in Fig. 6 show
that at low temperatures, in a chemical regime, reactant con-
version was congruent with measurements in the first reactor at
5
73 K, reactant conversion exhibited a moderate step increase,
similar to that observed in the CO oxidation tests. Temperature
6
−1 −1
h , indicating the equivalence
cat
profiles progressively smoothed on increasing H conversion
a GHSV of 2.0 · 10 NLkg
2
(
Fig. 5b) while the hot spot shifted backward and became po-
sitioned upstream from the catalyst inlet at temperatures above
03 K, suggesting the presence under these conditions of ho-
mogeneous contributions to reactant conversion.
of the reactors from a kinetic standpoint; in contrast, a more
significant impact of diffusive limitations was observed at high
temperatures, as evidenced by a significantly lower asymptotic
6
conversion of O . The increased mass transfer resistance was
2
interpreted to result from some degree of eccentricity of the an-
3
.3. CH + O + H O and CH + O + CO tests
nular duct in the second reactor, due to the flexibility of the
long, thin mullite tube. In annular ducts, the external diffusion
is strongly affected by the eccentricity; with an aspect ratio
of 0.53, the asymptotic Sherwood number of 5.724 with no ec-
4
2
2
4
2
2
The experimental campaign was extended with the aim of
better defining the kinetic dependencies of the main reactions

I. Tavazzi et al. / Journal of Catalysis 241 (2006) 1–13
7
centricity is reduced to 2.27 with an eccentricity of 0.5 [45]. By
assuming an asymptotic Sherwood number (Sh∞ = 3.9) lower
than the value assumed for the first reactor, corresponding to
ately affected by the improved mass transfer in He, suggest-
ing a mixed chemical-diffusive regime induced by the high
rate of secondary reactions. Simulations matched the exper-
imental trends satisfactorily, considering the degree of sim-
plification of the mono-dimensional assumption in which the
complex fluodynamic effect of the eccentricity was incorpo-
rated in a single mass-transfer parameter. The adequacy of the
model in accounting for such phenomena allowed a kinetic
analysis of data properly decoupled from diffusive limitation
effects.
∗
a perfect concentric configuration (Sh∞ = 5.724, R = 0.53),
this effect could be parametrically accounted for (Fig. 6, solid
lines). Direct evidence of the impact of external diffusion on
reactor performance was obtained by changing the dilution gas
from N2 to He (with a feed composition of CH4/O2/N2/He =
4
/2/8/86% v/v), which roughly tripled the molecular diffusivity
of each gas species. At low temperatures (up to about 773 K),
the reduced resistance to the interphase mass transfer did not
cause any change in reactant conversion due to the prevail-
ing chemical regime established at these temperatures (Fig. 6).
Conversely, at high temperatures, whereas the amount of O2
available at the catalytic surface tended to zero, the asymp-
totic consumption of O2 was considerably higher in He than
in N2. Even CH4 conversion at high temperature was moder-
The effect of H O and CO cofeed on the CH +O reaction
2
2
4
2
system was investigated by enriching the gas mixture with 2
and 4% of H2O and with 2 and 4% of CO2, respectively. En-
richment with H2O (Fig. 7) depressed the oxidation of CH4 at
temperatures below 773 K, as evidenced by the significant loss
of oxygen conversion at low temperature. The inhibiting effect
was remarkable at 2% added water and increased only slightly
from doubling the amount of H O. Above 823 K, O conversion
2
2
was independent of the water content in the feed stream, with
oxygen consumption under diffusive control. At low tempera-
tures, along with O conversion, CH conversion also decreased
2
4
on the addition of H2O to the reactant mixture. With increasing
temperature, methane consumption (being under partial kinetic
control over the entire temperature range, as discussed above)
did not vary, suggesting the independence of secondary reac-
tions on the partial pressure of H O.
2
Fig. 8 reports the trend of oxygen and methane conversion on
enriching the feed stream with CO2. In the case of carbon diox-
ide addition, data were collected only at 733 and 1033 K (i.e.,
at temperatures representative of the low- and high-temperature
fields), to avoid prolonged exposure of the catalyst to CH - and
4
CO2-rich mixtures, which in previous experiments of methane
dry reforming caused deactivation, possibly due to the forma-
tion of surface carbonaceous deposits. At inhibiting effect was
not observed at either low or high temperatures, indicating the
independence of both CH4 combustion and reforming reactions
Fig. 6. Effect of external diffusive limitation on reactants conversion in par-
tial oxidation tests. (") and (2) experimental O and CH conversion in
2
4
N , respectively; (F) and (Q) O and CH conversion in He. Model predic-
2
2
4
tions: (solid lines) simulation in N ; (dashed lines) simulation in He. Feed
2
composition: CH /O /N = 4/2/94% and CH /O /N /He = 4/2/7/87% v/v.
4
2
2
4
2
2
6
−1 −1
GHSV = 2.6 · 10 NLkg
h
; atmospheric pressure.
cat
2
on CO concentration.
Fig. 7. Effect of H O enrichment to the CH + O2 reacting system on O (a) and CH (b) conversion. Experimental results: (2) feed: CH /O /N
=
2
4
2
4
4
2
2
4
/2/94% (v/v); (") feed: CH /O /H O/N = 4/2/2/92% (v/v); (Q) feed: CH /O /H O/N = 4/2/4/90% (v/v). Model predictions reported as continuous lines.
4
2
2
2
4
2
2
2
6
−1 −1
GHSV = 2.6 · 10 NLkg
h
, atmospheric pressure.
cat

8
I. Tavazzi et al. / Journal of Catalysis 241 (2006) 1–13
Fig. 8. Effect of CO enrichment to the CH + O2 reacting system on O (a) and CH (b) conversion. Experimental results: (2) feed: CH /O /N
=
2
4
2
4
4
2
2
4
/2/94% (v/v); (") feed: CH /O /CO /N = 4/2/2/92% (v/v); (Q) feed: CH /O /CO /N = 4/2/4/90% (v/v). Model predictions reported as continuous lines.
4
2
2
2
4
2
2
2
6
−1 −1
GHSV = 2.6 · 10 NLkg
h
, atmospheric pressure.
cat
Fig. 9. Effect of CH partial pressure in steam reforming tests (reacting system CH + H O) on (a) CH conversion and (b, c, d,) products molar fraction.
4
4
2
4
Experimental data: (2) feed: CH /H O/N = 0.5/4/95.5% (v/v); (") feed: CH /H O/N = 1/4/95% (v/v); (Q) feed: CH /H O/N = 2/4/94% (v/v); (ꢀ) feed:
4
2
2
4
2
2
4
2
2
6
−1 −1
h , atmospheric pressure.
cat
CH /H O/N = 3/4/93% (v/v). Model predictions reported as continuous lines. GHSV = 2.6 · 10 NLkg
4
2
2
3
.4. CH4 + H2O tests
Mixtures with a H2O/CH4 ratio >1 (the stoichiometric ratio
for steam reforming) were fed to depress coke formation. For
the same reason, the experiment with the lowest H O/CH ra-
tio (4/3) was carried out at temperatures above 973 K, because
carbon formation is favored at lower temperatures.
Fig. 9 shows the results of steam-reforming tests at vari-
ous partial pressures of methane. The reaction started at about
693 K, and reactant consumption increased progressively with
The reactivity of the CH4 + H2O system was studied by fix-
2
4
ing the concentration of one species and changing the amount
of the other to investigate the kinetic dependence on the reac-
tant partial pressure. The effect of water was investigated at 2%
of CH4 and at 4, 6, and 8% (v/v) of H2O; the effect of CH4
was analyzed at 4% of H2O and 0.5, 1, 2, and 3% of methane.

I. Tavazzi et al. / Journal of Catalysis 241 (2006) 1–13
9
Fig. 10. Effect of H O partial pressure in steam reforming tests (reacting system CH + H O) on (a) CH conversion and (b, c, d,) products molar fraction.
2
4
2
4
Experimental data: (2) feed: CH /H O/N = 2/2/96% (v/v); (") feed: CH /H O/N = 2/4/94% (v/v); (Q) feed: CH /H O/N = 2/6/92% (v/v). Model predictions
4
2
2
4
2
2
4
2
2
6
−1 −1
h , atmospheric pressure.
cat
reported as continuous lines. GHSV = 2.6 · 10 NLkg
temperature, although remaining significantly lower than the
predicted values at thermodynamic equilibrium (dotted lines).
H2 and CO production also increased monotonically with tem-
perature, whereas CO2 formation passed through a maximum
due to the thermodynamics of the water–gas shift reaction,
which favors the forward reaction at low temperatures and the
reverse reaction at high temperatures. The conversion of CH4
was weakly dependent on its feed concentration, suggesting
that the forward rate of steam reforming was first-order with re-
spect to the partial pressure of methane. The residual moderate
dependence, with an increasing trend of conversion at decreas-
ing partial pressure of CH4, was interpreted as the result of
an inhibiting effect due to the competitive adsorption of CO,
the concentration of which increased with increasing CH4 feed
content. It is noteworthy that at low temperature, the extent of
methane conversion was comparable to that measured in partial
oxidation tests, indicating that the rate of intrinsic reforming ki-
netics in the absence of oxygen was similar to the combustion
rate under fuel-rich conditions.
distribution, the increasing amount of water in the feed mix-
ture promoted the water–gas shift reaction, which progressively
shifted the distribution of products toward hydrogen and CO2.
4. Kinetic analysis
In this work, extending the experimental investigation al-
lowed us to better define the kinetic dependencies of several
molecular reactions in the indirect scheme; these are discussed
below. The corresponding kinetic parameters, derived by non-
linear regression on experimental data, are reported in Table 2.
4.1. CH4 combustion
Low-temperature data on the effect of temperature, GHSV,
nature of the carrier, and H O and CO addition are best de-
2
2
scribed by assuming that the oxidation of methane is propor-
tional to its partial pressure, independent of O2 concentration
and inhibited by H2O adsorption with kinetics
On increasing the inlet partial pressure of H2O (Fig. 10),
methane conversion was totally independent of the excess of
water, evidencing a zero-order dependence of the forward re-
action rate of steam reforming on H2O concentration and in-
dicating that thermodynamic constraints played no significant
role under the investigated conditions. Concerning the product
kTO · PCH
H2O
4
rTO =
.
(6)
ads
2
(1 + K
· PH O)
2
The lack of an explicit dependence on O is congruent with the
picture of high surface coverage from oxygen with few sites
available for CH4 adsorption, in line with findings on Pt and
2

10
I. Tavazzi et al. / Journal of Catalysis 241 (2006) 1–13
Table 2
Kinetic equations and parameters obtained in this work
8
73 (K)
−1 −1
−1
atm )
Reaction
Kinetic expression
k
(mol g
s
Eatt/R (K)
j
k
· PCH4
CTO
−3
Methane combustion
R
=
· σO2
2
)
H O
2
8.061 · 10
6795
CTO
(
1 + K
· P
CH
4
ads,H O
CH + 2O → 2H O + CO
2
4
2
2
2
k
· P
· (1 − ηSR
)
SR
−3
Steam reforming
CH + H O → 3H + CO
R
SR
=
· σ
)
6.2 · 10
7069
7069
3910
7000
H O
2
(
1 + Kads,O2 · PO2 + K
· PCO 2
ads,CO
4
2
2
kDR · P
· (1 − ηDR)
CH
4
−3
CO -reforming
RDR =
· σCO2
)
6.2 · 10
2
(
1 + Kads,O2 · PO2 + K
· PCO 2
ads,CO
CH + CO → 2H + 2CO
4
2
2
k
· PCO2 · PH2 · (1 − ηRWGS
)
RWGS
−1
−1 −1
−2
atm )
Reverse water–gas shift
R
RWGS
=
6.07 · 10 (mol g
s
2
)
(
1 + Kads,O2 · PO2 + K
· PCO
ads,CO
CO + H → CO + H O
2
2
2
2.54 · 10 1
−
CO oxidation
R
Ox,CO
= kOx,CO · PCO · σO2
CO + (1/2)O → CO2
2
H oxidation
2
H + (1/2)O → H O
2
R
Ox,H
= kOx,H2 · PH2 · σO2
4.196
5000
2
2
2
8
73 (K)
−1
(atm )
Surface adsorption
K
ꢀH /R (K)
ads,j
ads
O
2
42.53
2.216 · 10
15
−7000
−19900
−3120
−
1
H O
2
CO
Pd [48] and recent results on Rh [49]. Note that on the basis of
preliminary data, a negative kinetic dependence on O2 had been
initially assumed [32]; however, subsequent experiments [37]
clearly indicated that the kinetics of CH4 activation by O2 is in-
dependent of O2 partial pressure and has an overall order lower
than 1.
increasing trend of CH4 conversion at decreasing inlet concen-
tration of methane (Fig. 9) was accounted for by invoking an
inhibiting effect of CO [Eqs. (7) and (8)].
The extent of CH4 conversion at low temperature in partial
oxidation tests indicated that in the presence of O2 at tempera-
tures below 773 K, the contribution of reforming reactions was
negligible, contrary to what was observed at the same tem-
peratures in CH4/H2O experiments. The effect of O2 on the
kinetics of steam reforming at low temperatures (a complex ef-
fect that may be related to modifications of Rh coordination)
could be accounted for (Fig. 2, solid lines) by introducing an
inhibition term. Accordingly, the following expression was pro-
posed:
On the addition of water to the inlet mixture, the deep ox-
idation of CH4 was markedly depressed, indicating competi-
tive adsorption of H2O on the catalyst surface and decreased
availability of O2-free sites for CH4 activation. This trend can
be accounted for (Fig. 7, solid lines) by assuming in Eq. (6)
a negative dependence with respect to H2O; the exponent of
the adsorption term is consistent with a dissociative adsorption
of CH4 involving two catalytic sites. The apparent saturation
of the inhibiting effect of water in the series at 0–2–4% is
likely correlated with the progressive increase of the tempera-
ture range in which the kinetic data were collected at increasing
partial pressure of H2O, that is, to the attenuation of the inhibi-
tion at increasing temperature. A rather high value of apparent
heat of H2O adsorption equal to −165 kJ/mol (Table 2) was
functional for capturing the observed trends according to the
assumed Langmuir–Hinshelwood rate expression.
kSR · PCH · (1 − ηSR)
4
rSR =
.
(7)
(
1 + K · PO + K · PCO)2
ads
ads
O2
2
CO
The kinetic dependencies found with respect to reactants par-
tial pressure were in agreement with recent studies of Wei and
Iglesia [29], who reported the independence of methane ac-
tivation rate on the nature and amount of coreactants. They
investigated CH4 reforming over Rh clusters by means of
kinetic and isotopic tracer methods and provided a simple
and unifying mechanistic proposal for it. They found that
activation of the C–H bond was the sole kinetically rele-
vant step, and that turnover rates were dependent only on
Rh crystallite size and independent of the nature of the sup-
port. Although Wei and Iglesia provided data concerning the
negligible role of competitive adsorption on methane activa-
tion, in this work surface saturation was taken into account
as an important kinetic effect. It should be noted that the
present experimental work spanned a very wide temperature
window.
4
.2. Steam and CO2 reforming of CH4
For secondary reactions, the approach to equilibrium was ac-
counted for by introducing the (1 − ηj ) factor, where ηj is the
ratio between the experimental reaction quotient (KP,j ) and the
thermodynamic equilibrium constant (Keq,j ), so as to identify
the actual reaction order and temperature dependence of for-
ward reaction rates. Experimental evidence in reforming tests
led to the formulation of rate equations for steam and CO2
reforming proportional to the partial pressure of CH4 and in-
dependent of the partial pressure of H2O and CO2. The slightly
Concerning CO reforming, catalyst deactivation phenom-
2
ena on exposure to CH4/CO2-rich mixtures did not allow us to

I. Tavazzi et al. / Journal of Catalysis 241 (2006) 1–13
11
directly investigate the reaction in the annular reactor. Analysis
of data and previous investigations support the kinetic relevance
of this consecutive step. In CH4/O2/N2 experiments, we sys-
tematically observed that at increasing temperature, the H2/CO
ratio tended to increase to an asymptotic value (∼1.7) below
the equilibrium value of 2. Such a dependence on tempera-
ture could not be accounted for by invoking only the steam-
reforming and water–gas shift reactions; an important contri-
bution from CO2 reforming had to be included to decrease
the calculated value of the H2/CO ratio at high temperature
and match the experimental data. Besides, previous studies per-
formed in an optimized packed-bed reactor under kinetically
informative conditions [31] indicated the activity of Rh toward
CO2 reforming, with a resulting reaction rate comparable in or-
der of magnitude to the rate of steam reforming (with a ratio
Finally, note that to account for the annihilation of reac-
tion rates in the case that the partial pressure of the coreac-
tant approached zero, which does not appear explicitly in the
proposed kinetic equations, the rate expressions of CH4 com-
bustion, steam reforming and CO2 reforming were multiplied
by σO , σH O, and σCO , respectively, where σi is a “limiting”
2
2
2
factor defined as
pi
σi =
.
(9)
−
6
10 + p
i
−
6
The threshold value (10
atm) at the denominator, under
which the kinetics tend to be first-order with respect to the
partial pressure of the coreactant as well, was chosen to be
the smallest number that guaranteed numerical convergence in
simulations. In this respect, here we did not imply a specific
physical meaning. An analysis of sensitivity revealed negligi-
kCO -R/kSR equal to about 0.25 at all temperatures). On the ba-
2
−4
ble variations in model predictions up to values of 10 atm.
sis of these considerations and the mechanistic findings of Wei
and Iglesia mentioned above, we assumed for the reaction of
CO2 reforming the same rate equation proposed for the reac-
Of course, kinetic equations that are independent of the partial
pressure of the limiting reactant cannot be used in reactor sim-
ulations, because they will lead to overconsumption of coreac-
tant, exceeding the limit allowed by molecular stoichiometries
at the catalyst wall.
Concerning the water–gas shift reaction, an accurate descrip-
tion of the experimental results (Figs. 9 and 10) was achieved by
adopting the following rate equation (written in reverse form):
tion of steam reforming, with kCO -R = kSR:
2
kCO -R · PCH · (1 − ηCO -R)
2
4
2
rCO -R =
.
(8)
2
ads
ads
CO
· PCO)2
(
1 + K · PO + K
O2
2
Notably, under conditions of high dilution and low conversion
i.e., at low reaction temperature) the kinetics of CH4 steam
(
kRWGS · PCO · PH · (1 − ηRWGS)
2
2
reforming and CH4-rich combustion tend to reduce to the sim-
r
RWGS
=
,
(10)
(
1 + K · PO + K · PCO)2
ads
ads
ple form rj = kj · PCH , in agreement with previously reported
O
2
CO
4
2
results [29]. According to our results, kTO ∼ kSR, in line with
the hypothesis that both oxidation and reforming are kinetically
controlled by C–H bond activation. According to our interpreta-
tion, the rate-determining step involved in either the combustion
or the reforming of CH4 is independent on the state of the cata-
lyst surface (O2-covered or clean). In contrast, the prevailing
stoichiometry is determined by the surface environment sur-
with negative dependence on O2 and CO partial pressure con-
served to ensure congruence with equations of reforming reac-
tions. The sensitivity of the calculated CO2 molar fraction to
the kinetics of the water–gas shift reaction allowed tuning of
the corresponding parameters even though the reaction was not
independently investigated. Note, however, that above 873 K,
the observed kinetics of the water–gas shift were significantly
affected by the approach to equilibrium conditions (with ηWGS
between 0.2 and 0.5), which could have biased estimation of
the kinetic parameters.
*
*
rounding the adsorbed C and H intermediates; in this respect,
the conversion of CH4 by steam reforming was depressed when
high O2 surface coverage was present. Referring the values re-
ported in Table 2 to the number of exposed surface Rh atoms
measured by H2 chemisorption (at Rh dispersion of 70%), kTO
4.3. CO and H combustion
2
−
1
−1
and kSR at 873 K are equal to 0.24 and 0.18 s kPa , re-
spectively. These estimates are very close to the values reported
by Wei and Iglesia relative to methane conversion in reform-
ing reactions [29]. Nonetheless, in contrast to the findings of
Wei and Iglesia [49], our study evidenced a very similar depen-
dence of the reaction rates of methane combustion and steam
reforming on temperature, with activation energies equal to 56
Kinetic parameters for H and CO oxidation were derived by
2
assuming for these reactions the simplest possible expressions
(i.e., first-order with respect to the fuel species), given the lack
of direct information on the reaction orders [Eqs. (11) and (12)].
Experiments with H2/CO/O2 mixtures under rich conditions
gave evidence of the inhibiting effect by CO on the oxidation
of H2; nonetheless, no inhibition contributions were included
in the kinetics given the difficulty in extrapolating the adsorp-
tion constants to the temperature field of interest for CH4 partial
oxidation:
−1
and 59 kJ mol , respectively. These values resemble those re-
ported by Liu and Hu for dissociative adsorption of CH4 on flat
Rh(111) surfaces calculated by density functional theory [50];
however, they are significantly lower than those reported by
several authors for CO2 reforming over Rh-supported catalysts,
rH -comb = kH -comb · PH
(11)
−
1
2
2
2
which range from 90 to 100 kJ mol [29]. Actually, for CH4-
rich combustion, Wei and Iglesia found an activation energy of
and
−1
5
5 kJ mol [49], very close to our estimate reported here, but
rCO-comb = kCO-comb · PCO.
(12)
for methane activation in absence of O2, they derived an activa-
−
1
Both expressions in the model mass balances were multiplied
tion energy of about 110 kJ mol , in agreement with the values
reported in the literature [29].
by σO , the coreactant-limiting factor. Fitting to the experimen-
2

12
I. Tavazzi et al. / Journal of Catalysis 241 (2006) 1–13
tal data was carried out at 353–523 K for H2 combustion and
at 473–673 K for CO combustion, that is, in temperature win-
dows in which no evidence of homogeneous contributions to
the reaction was observed. At increasing temperature, on en-
tering a diffusive regime, radicalic reactions (possibly induced
by the presence of the catalyst) occurred because reactant con-
sumptions were higher than the asymptotic values predicted by
the model (Figs. 4a and 5a) representing the upper conversion
limit for the catalytic reaction. The comparison among the ki-
netic constants reported in Table 2 at 873 K indicates a scale
of rich combustion reactivity, H2:CO:CH4, in order of mag-
nitude equal to 700:30:1, in agreement with estimates derived
preliminarily by fitting the experimental data for methane par-
tial oxidation [32]. The estimated CO/CH4 relative oxidation
rate closely resembles the reactivity ratio of 40 observed under
different conditions in previous work [49].
This implies that H2O enrichment doest not exert a kinetic pro-
motional effect on H yield in catalytic partial oxidation over
2
Rh catalyst, but a thermodynamically controlled promotion can
be realized at sufficiently low GHSV. Our findings agree with
and extend the results reported by Wei and Iglesia concerning
the independence of CH activation on the nature and amount
4
of coreactant for single reactions; however, adsorption and in-
hibition phenomena play significant roles at high conversion
values, and the kinetics of steam reforming are markedly inhib-
ited in the presence of O . By means of simple kinetics, a fine
2
description of the CPO reacting system was achieved (Fig. 2,
solid lines) within a very wide range of operating temperatures
6
7
−1 −1
h ). Molec-
cat
(623–1173 K) and GHSVs (10 –2.5·10 NLkg
ular equations developed herein are proposed as a kinetic output
of practical interest for reactor design and optimization.
According to the scenario depicted in the present study, the
overall process of methane partial oxidation proceeds through
different phases determined by the nature of the species ad-
sorbed on the catalyst surface. In the presence of high O2
surface coverage, which is thermodynamically favored at low
temperatures, only CO2 and H2O are produced by deep ox-
idation of methane; methane reforming is strongly inhibited.
The rate of O2 consumption increases with temperature, caus-
ing a reduction in the O2 surface coverage, which promotes the
reactions of reforming. H2 and CO are formed and converted
to H2O and CO2 along the entire catalyst length, in turn caus-
ing depletion of the surface O2, which eventually drives O2 into
mass transfer limitations. The cleaning of O2 from the surface
considerably enhances the reforming reaction rates and thus
CH4 consumption. Such a self-promoting mechanism induces
a “chemical lightoff” of the process associated with a stepwise
cleanup of the catalyst surface from O2, after which a net pro-
duction of synthesis gas can be established, in line with findings
of Lyubovsky et al. [3]. At the high temperatures representative
of practical applications, reforming reactions are active from
the very start of the catalyst; O2 is rapidly depleted by H2 and
CO postcombustion, which is predicted to occur mostly near
the entrance of the catalytic layer.
Acknowledgments
This work was supported by MIUR-Rome, the EU Cathlean
Project, and the Center for Nano-Engineered Materials and Sur-
faces.
Appendix A. Nomenclature
w(b)
xi
Di
z
wall (bulk) molar fraction
molecular diffusivity (m s
axial coordinate (m)
2
−1
)
∗
Fi
z
dimensionless axial coordinate (= z/d ) (–)
h
−1
molar flow of species ith (moli s
)
◦
−1
F
F
R
total inlet molar flow (mol s
)
TOT
∗
◦
Fi/F
dimensionless flow of species ith (–)
i
∗
TOT
Rint/Rout (aspect ratio) (–)
hydraulic diameter (m)
dh
Pem,i
v
Péclet number, Re · Sci (–)
−1
average gas velocity (m s
)
Re
Reynolds number = vdh/ν (–)
Sci
Kc,i
Shi
ZSh,i
rj
Schmidt number = ν/Di (–)
−1
mass transfer coefficient (m s
)
local Sherwood number = Kc,idh/Di (–)
∗
z /Pem,i (–)
5
. Conclusion
−1
−1
reaction rate (mol s kg
O2 concentration at the catalyst wall (mol m
)
S
O2
−3
)
C
The potentialities of the annular reactor were exploited to
2
−1
DM
macroporous diffusivity (m s
)
obtain highly informative kinetic data on extremely fast reac-
tions involved in the partial oxidation of CH over Rh/α-Al O .
2
−1
)
Dμ
microporous diffusivity (m s
4
2
3
The comparison between experimental results and predictions
of a mono-dimensional model of the annular reactor corrob-
orated the evolution of the process according to an indirect-
consecutive scheme (previously proposed by us) consisting of
a primary reaction of CH4 deep oxidation to CO2 and H2O,
followed by secondary reactions of steam and CO2 reforming
of CH4, responsible for syngas production, the water–gas shift
reaction, and the consecutive combustion of H2 and CO. The in-
variance of reactants conversion at high temperatures in partial
oxidation experiments after enriching the feed stream with CO2
and H2O was confirmed by the independence of the forward re-
action rate of CH4 reforming on the coreactant partial pressure.
Greek symbols
cinematic viscosity (m2 s
−1
)
ν
νi,j
stoichiometric coefficient (–)
−
1
−1
−1
i
−1
αi
Wcat · dh · S · C · D (mol s kg)
tot
∞
η
effectiveness factor at φ → ∞ (–)
η
effectiveness factor (–)
δ
thickness of the catalytic layer (m)
generalized Thiele modulus (–)
macropore void fraction (–)
micropore void fraction (–)
L
φ
ε
M
εμ

I. Tavazzi et al. / Journal of Catalysis 241 (2006) 1–13
13
References
[26] D.A. Hickman, L.D. Schmidt, AIChE J. 39 (1993) 1164–1177.
[27] A.B. Mhadeshwar, D.G. Vlachos, J. Phys. Chem. B 109 (2005) 16819–
1
6835.
[
[
[
1] S.S. Bharadwaj, L.D. Schmidt, Fuel Process. Technol. 42 (1995) 109–127.
2] D. Neumann, M. Kirchhoff, G. Veser, Catal. Today 98 (2004) 565–574.
3] M. Lyubovsky, L.L. Smith, M. Castaldi, H. Karim, B. Nentwick, S. Ete-
mad, R. LaPierre, W.C. Pfefferle, Catal. Today 83 (2003) 71–84.
4] A.T. Ashcroft, A.K. Cheetham, J.S. Foord, M.L.H. Green, C.P. Grey, A.J.
Murrel, P.D.F. Vernon, Nature 344 (1990) 319–321.
5] P.D.F. Vernon, M.L.H. Green, A.K. Cheetham, A.T. Ashcroft, Catal.
Lett. 6 (1990) 181–186.
6] P.D.F. Vernon, M.L.H. Green, A.K. Cheetham, A.T. Ashcroft, Catal. To-
day 13 (1992) 417–426.
[
[
[
28] J. Wei, E. Iglesia, J. Catal. 224 (2004) 370–383.
29] J. Wei, E. Iglesia, J. Catal. 225 (2004) 116–127.
30] G. Groppi, W. Ibashi, E. Tronconi, P. Forzatti, Chem. Eng. J. 82 (2001)
57–71.
[
[
[
[
[
31] L. Majocchi, Ph.D. Thesis (1999), Politecnico di Milano.
32] I. Tavazzi, A. Beretta, G. Groppi, P. Forzatti, in: X. Bao, Y. Xu (Eds.),
Studies Surface Science Catalysis—Natural Gas Conversion VII, vol. 147,
Elsevier, Amsterdam, 2004, pp. 163–168.
[
33] L. Basini, M. Marchionna, A. Aragno, J. Phys. Chem. 23 (1992) 9431–
9
441.
34] L. Basini, A. Aragno, J. Chem. Soc., Faraday Trans. 90 (5) (1994) 787–
95.
[7] S.C. Tsang, J.B. Claridge, M.L.H. Green, Catal. Today 23 (1995) 3–15.
[8] D.A. Hickman, L.D. Schmidt, J. Catal. 138 (1992) 267–282.
[9] J.B. Claridge, M.L.H. Green, S.C. Tsang, A.P.E. York, A.T. Ashcroft, P.D.
Battle, Catal. Lett. 22 (1993) 299–305.
[
7
[
[
[
35] L. Basini, D. Sanfilippo, J. Catal. 157 (1995) 162–178.
36] L. Basini, A. Aragno, G. Vlaic, Catal. Lett. 39 (1996) 49–55.
37] T. Bruno, A. Beretta, G. Groppi, M. Roderi, P. Forzatti, Catal. Today 99
[
10] D.A. Hickman, E.A. Haupfear, L.D. Schmidt, Catal. Lett. 17 (1993) 223–
237.
(2005) 89–98.
[
[
[
[
[
[
11] K.L. Hohn, L.D. Schmidt, Appl. Catal. A: Gen. 211 (2001) 53–68.
12] Y.F. Chang, Heinemann, Catal. Lett. 21 (1993) 215–224.
13] V.R. Choudary, A.M. Rajput, B. Prabhakar, J. Catal. 139 (1993) 326–328.
14] V.R. Choudary, V.H. Rane, A.M. Rajput, Catal. Lett. 22 (1993) 289–297.
15] V.R. Choudary, A.M. Rajput, V.H. Rane, Catal. Lett. 16 (1992) 269–272.
16] V.R. Choudary, S.D. Sansare, A.S. Mamman, Appl. Catal. A: Gen. 90
[
[
38] J.G. McCarty, Catal. Today 26 (1995) 283–293.
39] A. Beretta, P. Baiardi, D. Prina, P. Forzatti, Chem. Eng. Sci. 54 (1999)
765–773.
[
40] M. Valentini, G. Groppi, C. Cristiani, M. Levi, E. Tronconi, P. Forzatti,
Catal. Today 69 (2001) 307–314.
[
[
41] W. Ibashi, G. Groppi, P. Forzatti, Catal. Today 83 (2003) 115–129.
42] R.D. Hawthorne, After catalysts effects of heat and mass transfer between
gas and catalyst surface, AIChE Symp. Ser. 70 (1974) 428–438.
43] E. Tronconi, P. Forzatti, AIChE J. 38 (1992) 201–210.
(1992) L1–L5.
[
[
[
[
17] Y. Boucouvalas, Z. Zhang, X.E. Verykios, Catal. Lett. 27 (1994) 131–142.
18] T. Ioannides, X.E. Verykios, Catal. Lett. 47 (1997) 183–188.
19] O.V. Buyevskaya, D. Wolf, M. Baerns, Catal. Lett. 29 (1994) 249–260.
20] K. Walter, O.V. Buyevskaya, D. Wolf, M. Baerns, Catal. Lett. 29 (1994)
[
[
44] A. Beretta, G. Groppi, L. Majocchi, P. Forzatti, Appl. Catal. 187 (1999)
49–60.
2
61–270.
21] O.V. Buyevskaya, K. Walter, D. Wolf, M. Baerns, Catal. Lett. 38 (1996)
1–88.
22] E.P.J. Mallens, J.H.B.J. Hoebink, G.B. Marin, J. Catal. 167 (1997) 43–
6.
23] K.H. Hofstad, J.H.B.J. Hoebink, A. Holmen, G.B. Marin, Catal. Today 40
1998) 157–170.
24] R. Schwiedernoch, S. Tischer, C. Correa, O. Deutschmann, Chem. Eng.
Sci. 58 (2003) 633–642.
25] D. Wang, O. Dewaele, A.M. De Groote, G.F. Froment, J. Catal. 159 (1996)
[
[
[
45] R.K. Shah, A.L. London, Laminar Flow Forced Convection in Ducts, Aca-
[
[
[
[
[
demic Press, New York, 1978.
46] R.C. Reid, J.M. Prausnitz, B.E. Pauling, The Properties of Gases and
Liquids, McGraw-Hill, New York, 1987, pp. 587–589.
47] G.B. Froment, K.B. Bischoff, Chemical Reactor Analysis and Design, Wi-
ley, New York, 1990.
48] R. Burch, D.J. Crittle, M.J. Hayes, Catal. Today 47 (1999) 229–234.
49] M. Li, J. Wei, E. Iglesia, Proceedings of the 5th World Congress on Oxi-
dation Catalysis, Sapporo, Japan (2005) pp. 74–75.
8
5
(
[
[
[
50] Z. Liu, P. Hu, J. Am. Chem. Soc. 125 (2003) 1958–1967.
418–426.