Production of hydrogen by partial oxidation of methanol over a Cu/ZnO/Al2O3 catalyst: Influence of the initial state of the catalyst on the start-up behaviour of the reformer

Article abstract of DOI:10.1006/jcat.2002.3764

The temperature-programmed start-up of a methanol reformer for the production of hydrogen by partial oxidation of methanol, using air as oxidant, was studied over a Cu/ZnO/Al₂O₃ sample in its oxidised, reduced, and reduced + air-exposed states. The temperature at which the reaction starts was shifted to higher values when the degree of surface oxidation increased: oxidised sample ? reduced sample > reduced + air-exposed sample. Conversely, the same product selectivity is observed irrespective of the initial state of the samples. Methanol partial oxidation over the oxidised sample was accompanied by concomitant surface reduction. This reduction process leads to surface reconstruction with a higher methanol decomposition capacity than that corresponding to prereduced counterparts. These differences appear to be related to changes in the number, but not in the characteristics, of the active sites induced by the different reduction potentials of the reacting gases. For consecutive start-ups, the reaction-reduced catalyst behaves identically to the prereduced counterparts. These observations point to the central role of lattice oxygen during the restructuring of the catalyst surface under reaction conditions.

Full text of DOI:10.1006/jcat.2002.3764

Journal of Catalysis 212, 112–118 (2002)
doi:10.1006/jcat.2002.3764
RESEARCH NOTE
Production of Hydrogen by Partial Oxidation of Methanol over
a Cu/ZnO/Al2O3 Catalyst: Influence of the Initial State of the
Catalyst on the Start-Up Behaviour of the Reformer
R. M. Navarro, M. A. Pen˜a, and J. L. G. Fierro1
Instituto de Cata´lisis y Petroleoqu´ımica, CSIC, Campus UAM, Cantoblanco, 28049 Madrid, Spain
Received April 19, 2002; revised June 17, 2002; accepted July 29, 2002
processes (2): thermal decomposition (TDM) (Eq. [1]), par-
tial oxidation of methanol (POM, Eq. [2]), and steam re-
forming (SMR, Eq. [3]). The overall processes summarised
in Eqs. [1]–[3] are a complex convolution of elementary
steps that involve several organic intermediates.
The temperature-programmed start-up of a methanol reformer
for the production of hydrogen by partial oxidation of methanol,
using air as oxidant, was studied over a Cu/ZnO/Al2O3 sample in
its oxidised, reduced, and reduced + air-exposed states. The tem-
perature at which the reaction starts was shifted to higher val-
ues when the degree of surface oxidation increased: oxidised sam-
ple ꢀ reducedsample > reduced + air-exposedsample. Conversely,
the same product selectivity is observed irrespective of the initial
state of the samples. Methanol partial oxidation over the oxidised
sample was accompanied by concomitant surface reduction. This
reduction process leads to surface reconstruction with a higher
methanol decomposition capacity than that corresponding to
prereduced counterparts. These differences appear to be re-
lated to changes in the number, but not in the characteristics,
TDM: CH3OH → 2H2 + CO (ꢀH0 = +90.5 kJ/mol) [1]
POM: CH3OH + 1/2 O2 → 2H2 + CO2
(ꢀH0 = −191.9 kJ/mol) [2]
SMR: CH3OH + H2O → 3H2 + CO2
(ꢀH0 = +49.8 kJ/mol). [3]
of the active sites induced by the different reduction potentials In the case of hydrogen for nonstationary applications,
of the reacting gases. For consecutive start-ups, the reaction-
reduced catalyst behaves identically to the prereduced counter-
parts. These observations point to the central role of lattice oxygen
during the restructuring of the catalyst surface under reaction
POM and OMR has shown advantages over the endother-
mic SMR in regard to kinetic aspects, reactor start-up dy-
namics, and the fast response necessary to work under vary-
ing loads.
c
conditions. ꢁ 2002 Elsevier Science (USA)
All three reactions [1]–[3] can be carried out over indus-
trial Cu/ZnO/Al2O3 type catalysts (3–7) with high activity
for CH3OH conversion and very high selectivity for hy-
drogen production. For these catalytic systems, under flow
reaction conditions, the local atomic structure of the surface
is itself controlled by the oxidation reaction, since factors
such as temperature and gas composition will determine
the kinetically controlled response of the catalyst to the re-
action conditions. Several authors (8–10) have shown that
in POM, both methanol conversion and the concentration
of products in the exit stream become strongly dependent
on the O2 concentration in the gas phase.
Although the exact nature of the active sites involved in
the production of hydrogen from methanol is still unclear,
there is no doubt concerning the major role of surface redox
properties (Cu2+/Cu+/Cu0) of the catalyst in its reaction be-
haviour. In general, the standard method for determining
the activity and performance of copper-containing catalysts
includes a reductive pretreatment in a diluted hydrogen
Key Words: fuel cell; hydrogen production; methanol partial ox-
idation; Cu/ZnO/Al2O3 catalyst; methanol reformer; start-up.
1. INTRODUCTION
Hydrogen gas is an excellent energy vector when used
directly as a fuel in internal combustion engines (1) or, in-
directly to supply electricity using polymer electrolyte fuel
cells. Nevertheless, the use of H2 gas for mobile fuel cell ap-
plications is hindered by problems of storage, safety, refu-
eling, etc. These problems can be overcome by the produc-
tion of hydrogen on board the vehicle from a suitable high
energy/density liquid fuel, such as methanol. Hydrogen
can be obtained directly from methanol by three different
1 To whom correspondence should be addressed. Fax: +34 1 585 4760.
E-mail: jlgfierro@icp.csic.es.
112
0021-9517/02 $35.00
c
ꢁ 2002 Elsevier Science (USA)
All rights reserved.

START-UP BEHAVIOR OF A Cu/ZnO/Al2O3 CATALYST FOR H2 PRODUCTION
113
steam carried out in situ and prior to activity measure- erised Seifert XRD 3000P diffractometer (Cu Kα radia-
ments. The reduction pretreatment controls the structural tion; λ = 0.15418 nm), a PW Bragg-Brentano ꢁ/2ꢁ 2200
and morphological characteristics of the catalyst surface goniometer, equipped with a bent graphite monochromator
(11–14). These initial characteristics play a central role in and an automatic slit. The volume-averaged crystallite sizes
the evolution of the oxidation state and structural morpho- of CuO were determined from the half-width of the CuO
logy during the reaction since, as stated above, the dynamic (111) diffraction line at 2ꢁ = 38.8◦ applying the Debye–
behaviour of the catalyst surface is determined by the gas Scherrer equation. The chemical composition of the cata-
atmosphere conditions under reaction.
lyst was determined by inductively coupled plasma atomic
Because the reduction potential of the reactive envi- emission spectroscopy (ICP-AES) using a Perkin–Elmer
ronment changes the structural and chemical properties Optima 3300DV apparatus. The samples were solubilized
of the surface layer, study of the start-up of the reform- in a solution of HF, HCl, and HNO3 and homogenised in a
ing over Cu/ZnO/Al2O3 catalysts in different initial states microwave oven at a maximum power of 650 W. Aliquots
provides important practical information regarding the of the solution were diluted in order to reach a concen-
activation and handling of catalysts, considering the ser- tration of the elements within the calibration range of the
vice expectations of automobile users. With this aim, here instrument. The nitrogen adsorption–desorption isotherms
we investigated aspects related to the cold start-up be- were obtained at 77 K over the whole range of relative pres-
haviour of the methanol reformer for the production of sures, using a Micromeritics ASAP 2100 automatic device
hydrogen over a Cu/ZnO/Al2O3 catalyst in oxidised, re- on samples previously outgassed at 523 K. BET surface ar-
duced, or reduced + air-exposed forms. In addition, and eas were computed from these isotherms using the BET
bearing in mind that a typical catalytic system for auto- method and taking a value of 0.1620 nm2 for the cross sec-
mobile applications will be subjected to hundreds of start- tion of the physically adsorbed N2 molecule. Table 1 shows
up/shut-down cycles in which the catalysts will be exposed the chemical composition (expressed as weight percent-
to temperature-induced changes in the gas-phase reduc- age), particle size of the CuO crystallites, and the textu-
tion potential, the impact of repeated start-ups and shut- ral parameters of the CuO/ZnO/Al2O3 sample in oxidised
downs on the overall reactor performance is another aspect form.
addressed.
2.2. Catalytic Activity in Temperature-Programmed
Start-Ups
2. EXPERIMENTAL
The samples, in oxidised, reduced, and reduced +
2.1. Catalyst Preparation and Characterization
air-exposed forms, were tested in the temperature-
programmed partial oxidation of methanol (O2/CH3OH =
0.3) using a fixed-bed flow reactor at atmospheric pressure.
Temperature-programmed start-up tests were carried out
using 0.1 g of calcined sample diluted with SiC (both in the
0.42–0.5 mm particle size range) at a volume ratio of 2 : 1
in order to avoid adverse thermal effects (12). The catalyst
bed was placed in a 6-mm I.D. quartz tubular reactor with
a coaxially centred thermocouple. Prior to reaction, the
catalysts were flushed in nitrogen at 423 K for 2 h followed
by: (i) nonadditional pretreatment (oxidised sample) or,
The copper–zinc–aluminium catalyst (nominal compo-
sition Cu : Zn : Al = 55 : 40 : 5 wt%) was prepared accord-
ing to the well-documented technique of carbonate co-
precipitation from aqueous nitrate solutions of Cu, Zn,
and Al, using Na2CO3 as a precipitating agent. The so-
lution with the three metal nitrates and the Na2CO3 so-
lution were added, both at 0.83 mL/min, to a vigorously
stirred NaHCO3 solution at 333 K. During the process, the
temperature (333 K) and pH (8.2) of the precipitating so-
lution were strictly controlled. All reagents were used as
received: Cu(NO3)2 · 3H2O from Panreac (ACS reagent),
Zn(NO3)2 · 6H2O (ACS reagent) from Panreac, Al(NO3)3
from Fluka (puriss. p.a.), Na2CO3 from Panreac (puriss
p.a.), and NaHCO3 from Fluka (microselect reagent). The
precipitates were aged with stirring for 90 min at 333 K,
then filtered, and extensively washed with deionized water
at333KuntiltheconcentrationofNa+ ionsinthefiltrateso-
lution was lower than 0.1 ppm, as determined by inductively
coupled plasma emission spectrometry. The precipitate was
dried overnight at 383 K and calcined in air at 623 K for
6 h. Finally, the samples were pelletized and sieved (0.42–
0.50 mm) for the catalytic runs. The fresh oxidised sample
was characterised by X-ray diffactometry according to the
TABLE 1
Chemical Composition, CuO Particle Size, and Textural
Properties of the CuO/ZnO/Al2O3 Oxidised Catalyst
Cu/Zn/Al ratio (wt%)a
BETb
Pore diam.b
(nm)
Dpc CuO
(nm)
Cu
Zn
Al
(m2/g)
57.8
37.9
4.3
80.1
18.0
4.1
a Determined by ICP.
b BET specific area and average pore diameter as measured by N2
adsorption–desorption isotherms at 77 K.
c Determined by XRD using the Scherrer equation for CuO ((111) −
step-scanning procedure (step size 0.02◦) using a comput- 38.8◦).

˜
114
NAVARRO, PENA, AND FIERRO
100
(ii) reduction in situ at 548 K (heating rate 5 K/min) with
55 mL(STP)/min of 10 vol% H2/N2 mixture for 2 h (prere-
duced sample) or, (iii) reduction by method (ii), followed
by catalyst exposure to air at ambient temperature for 24 h
(prereduced + air-exposed sample). The pretreating gases
were flushed from the reactor with N2 before admission
of the methanol–oxygen–nitrogen reaction mixtures.
Methanol was fed into the preheater by means of a liquid
pump (Becton–Dickinson) before being mixed with the
oxygen and nitrogen streams. Gas flows were adjusted
with Brooks model 5850E mass flow controllers. For the
POM reaction, the reactants were introduced into the
reactor at a molar ratio of O2/CH3OH = 0.3, using air as
oxidant agent. In the catalytic tests, total flow rate was
kept at 25 mL(STP)/min (GHSV = 13570 h−1) and the
methanol concentration in the feed gas mixture was fixed
at 45 vol%. For the start-up measurements, the reaction
temperature was programmed according to the following
sequence: starting at 423 K for 1 h, subsequent heating at a
rate of 0.1 K/min up to 573 K, and finally application of this
temperature for 3 h. Taking into account the residence time
(0.1 s) used in the experiments, the slow temperature rise
(0.1 K/min) was selected to ensure isothermal behaviour
of the reactor.
80
60
40
20
0
350
400
450
500
550
600
Temperature (K)
FIG. 1. Temperature dependence of methanol conversion during
temperature-programmed start-up over the Cu/ZnO/Al2O3 catalyst in
different initial states: (ꢀ) oxidised, (ꢁ) reduced, and (ꢂ) reduced + air-
exposed. Start-up conditions: atmospheric pressure, feed ratio O2/
CH3OH = 0.3, ramp rate = 0.1 K/min.
The reaction products were analysed online by GC
with TCD (Varian chromatograph, Model Star 3400 CX)
equipped with Porapack Q (CO2, water, methanol,
formaldehyde, methyl formate, formic acid, and dimethyl
the sample was quite different from that observed for the
prereduced counterparts. The oxidised sample behaved
similarly, but with lower conversion values than the
prereduced samples at temperatures up to about 513 K. At
higher temperatures, the increase in methanol conversion
induced by the increase in temperature occurred more
quickly over the oxidised sample than over the prereduced
counterparts. Thus, at temperatures over 521 K, the
methanol conversion achieved on the oxidised catalyst
surpassed the values corresponding to the prereduced
samples.
˚
ether), and molecular sieve 5-A (H2, O2, N2, CO) packed
columns connected in series, using He as carrier gas. The
composition of the product gas was determined at time in-
tervals of 30 min (3 K in the temperature programme of the
reaction). The methanol conversion and product selectivi-
ties are based on mole percentages.
3. RESULTS
3.1. Temperature-Programmed Start-Up over Oxidized,
Prereduced, and Prereduced + Air-Exposed Samples
The conversion patterns for oxygen, Fig. 2, followed
the same trend as those observed for the conversion of
The influence of temperature on the activity of the methanol. Complete conversion of oxygen was reached at
Cu/ZnO/Al2O3 catalyst in its oxidised, reduced, and 485 K for the prereduced samples and at 512 K for the ox-
reduced + air-exposed forms during the temperature- idised sample. The evolution of product selectivity shown
programmed POM start-up is shown in Fig. 1. For all the in Fig. 2 indicates the existence of several pathways that
pretreatments applied, the methanol conversion profile contribute to the overall reaction. Three ranges of temper-
follows a typical sigmoid-like shape, with a drastic increase ature (I, II, III), with characteristic product distributions,
in methanol conversion within a narrow temperature range. can be identified in Fig. 2. In range I (423–460 K), the reac-
For the prereduced sample, methanol conversion started at tion begins with the formation of formic acid, methyl for-
416 K, while for the prereduced, air-exposed, and oxidised mate, H2O, and CO2. In this temperature window, the rise
samples this point shifted to higher temperatures: 422 and in temperature was accompanied by an increase in the se-
434 K, respectively. Both reduced samples, with/without air lectivity of water and CO2 at the expense of oxygenated
exposure, displayed the same activity profiles in the whole compounds. A second temperature-dependent kinetic re-
temperature range, with a temperature shift to slightly gion was defined in range II (460–500 K). In this region,
higher values in the case of the air-exposed sample. For the reaction developed with a sharp decrease in the selec-
the unreduced catalyst sample, the activity behaviour of tivity of water, with a simultaneous increase in hydrogen

START-UP BEHAVIOR OF A Cu/ZnO/Al2O3 CATALYST FOR H2 PRODUCTION
115
the temperatures shifted to 472–491 K and 493–512 K, re-
spectively.
3.2. Temperature-Programmed Start-Up/Shut-Down/
Restart Cycle over an Oxidised Sample
To know how the working conditions might affect the
initial performance of the oxidised catalyst, a second start-
up was performed. For this, after the first start-up mea-
surement, the reactor temperature was lowered under a
static reactive atmosphere from 573 K to room temper-
ature. Once cooling was complete, the second start-up
was performed, following the temperature protocol de-
fined in the experimental section for the start-up of the
reformer. Figure 3 compares the methanol conversion pro-
file of the oxidised sample subjected to the start-up/shut-
down/restart-up cycle. The product distribution obtained
over the oxidised sample during this cycle is shown in Fig. 4.
The experimental data point to considerable differences in
catalytic behaviour between the first and second start-ups.
Duringthesecondcycle, ameasurablemethanolconversion
was observed at a significantly lower temperature (465 K).
Above this temperature, the methanol conversion achieved
during the second start-up was lower than that attained
during the first start-up.
3.3. Consecutive Start-Up/Shut-Down Cycles
To estimate the impact of repeated start-ups and shut-
downs on the overall catalyst performance, the effect of
100
80
60
40
20
0
FIG. 2. Temperature evolution of the products: (ꢃ) H2O, (ꢁ) H2,
(ꢄ) CO2, (ꢂ) CO, (ꢅ) HCOOCH3, (ꢆ) HCOOH, and (ꢀ) O2 con-
version during temperature-programmed start-up over the Cu/ZnO/
Al2O3 catalyst in different initial states: (a) oxidised, (b) reduced, and
(c) reduced + air-exposed. Three temperature ranges: I, II, and III, with
characteristic product distributions, were included (see text). Start-up con-
ditions: atmospheric pressure, feed ratio O2/CH3OH = 0.3, ramp rate =
0.1 K/min.
and CO2 selectivities. Finally, in range III (T > 500 K), the
reaction evolved with unchanged H2 selectivity, accompa-
nied by a steady increase in the CO/CO2 selectivity ratio
with temperature. The same selectivity patterns were ob-
served for all the catalytic samples, irrespective of the ini-
tial state. Only the transition temperature that produced
the change in the reaction regime depended on the pre-
treatment applied to the sample. For the prereduced sam-
ple, the transition temperatures were set at 463 and 487 K,
350
400
450
500
550
600
Temperature (K)
FIG. 3. Temperature dependence of methanol conversion over the
sample in the oxidised initial state during temperature-programmed
start-ups following the cycle: (ꢀ) first start-up/shut-down/ (ꢁ) consecutive
while for the reduced + air-exposed and oxidised samples start-up.

˜
116
NAVARRO, PENA, AND FIERRO
100
temperature cycles was investigated. For this purpose, af-
ter a temperature-programmed reaction over the catalyst
in the oxidised state, the catalytic bed was subjected to shut-
down procedures, varying the length of the stand-by period.
During the different shut-downs, the sample was cooled to
room temperature under a static reactive atmosphere, re-
maining under this atmosphere for 4, 72, and 96 h. After
these stand-by periods, the activity was evaluated again in
accordance with the reactor temperature programme de-
scribed previously. Figure 5 shows the changes in CH3OH
conversion during the cycled temperature-programmed re-
action after catalyst stand-by operations for periods of 4,
72, and 96 h. A negligible effect, neither methanol con-
version nor product distribution were seen for stand-by
periods under the reactive atmosphere of less than 72 h.
Only for the catalyst subjected to a stand-by period of 96 h
was the onset of the reaction shifted to a slightly higher
temperature.
80
60
40
20
0
350
400
450
500
550
600
Temperature (K)
FIG. 5. Temperature dependence of methanol conversion for POM
during temperature-programmed cyclic start-ups after stand-by periods
under a static reactive atmosphere of different durations: (ꢃ) 0 h, (ꢀ) 4 h,
(ꢂ) 72 h, and (ꢆ) 96 h.
4. DISCUSSION
4.1. Influence of the Initial State of the Catalyst
on the Start-Up Behaviour of the Reformer
The product selectivity patterns (Fig. 2) clearly indicate
that the same potential reaction networks (9) operate for
the catalyst in reduced, air-passivated, or oxidised forms.
Only the profile of methanol conversion and the onset tem-
peratures for each chemical region that contributes to the
overall conversion vary with the pretreatment applied.
In the case of prereduced samples, it is evident that the
reduced/oxidised catalyst state is affected by the kinetics of
the catalytic reaction since the catalytic reactions involved
in POM consume oxygen in competition with copper oxide
formation:
4Cu + O2 → 2Cu2O.
[4]
At low reaction temperatures, the prereduced sample is
expected to be oxidised at surface level by gas oxygen ac-
cording to reaction [4]. Reitz et al. (8) and Gu¨nter et al.
(10) observed by time-resolved XANES Cu+ as transient
species at low temperatures under reaction flow over re-
duced Cu/ZnO samples. However, some authors (16)—
taking into account the low O2/CH3OH molar ratio used
and the mismatch between the lattice parameters of cubic
Cu and Cu2O—have indicated that complete oxidation of
the surface is not to be expected, as indicated by Eq. [4],
its being more plausible for there to be a two-dimensional
FIG. 4. Evolution with temperature of the reaction products: (ꢃ)
H2O, (ꢁ)H2, (ꢄ)CO2, (ꢂ)CO, (ꢅ)HCOOCH3, (ꢆ)HCOOH, and(ꢀ)O2
conversionoverthesampleintheoxidisedinitialstateduringtemperature-
programmed start-ups following the cycle: (a) first start-up/shut-down/
(b) consecutive start-up.

START-UP BEHAVIOR OF A Cu/ZnO/Al2O3 CATALYST FOR H2 PRODUCTION
117
copper suboxide surface species. The exposure of the pre- methanol conversion for O2/CH3OH = 0.3), the excess of
reduced sample to air for 24 h leads to an increase in the methanol is converted through the endothermic decompo-
abundance of these suboxidized structures, with the devel- sition pathway on metallic Cu atoms (23, 24).
opment of Cu2O structures, but maintaining the Cuo core
The higher activity shown by the oxidised sample once
particle (17, 18). In the case of the oxidised sample only the reaction has started points to the different evolution
CuO surface species are present. of the catalyst surface under the reaction conditions of this
Besides chemisorbed oxygen, the reactions involved in sample with respect to the prereduced counterparts. Vari-
the beginning of partial oxidation of methanol require the ations in the surface structure of Cu in Cu/ZnO catalysts
existence of metal sites. It is evident that higher degrees due to changing either the reducing agent or the reduction
of surface oxidation in the samples induce a shift to higher conditions have been previously described in the literature
values in the reaction temperature at which the reaction (11, 12, 14). Changes in the relative abundance of differ-
starts (see Fig. 1). To explain this behaviour, the ability of ent crystallographic Cu metal planes (12), the formation
the surface to promote the adsorption of CH3OH must be of Cu–Zn alloys (13), or the coalescence of copper crystal-
taken into account. Several studies (19, 20) have pointed lites (25) induced by the exothermicity of reduction may
to the importance of the amount of oxygen at the surface be involved in the differences in catalytic activity found as
in the probability of methanol sticking to copper atoms. In a result of changes in the reduction/reaction conditions. In
these studies, it was demonstrated that the methanol stick- spite of the differences in reactivity found for the samples
ing probability was enhanced at low oxygen coverage, while tested, analysis of the products versus methanol conver-
at high oxygen coverage the sticking probability decreased. sion reveals that the product distribution at an equal value
Therefore, lower methanol conversion is expected when the of methanol conversion does not depend significantly on
extent of surface oxidation increases. In the case of oxide the pretreatment applied to the sample. Only for the oxi-
surfaces, the metal sites are difficult to access due to the in- dised sample is the participation of the reverse water gas-
herentlylargesizeofoxygenionsrelativetocations. Defects shift reaction observed when practically all the methanol
inclosed-packedsurfacesandincompletelycoordinatedox- has been converted. The absence of marked differences
ide polyhedra are, however, accessible for the chemisorp- in product distribution at methanol isoconversion strongly
tion of adsorbates being able to deprotonate methanol to suggests that the differences in activity between oxidised
methoxide (21). In this case, the exposed Cu cations and and prereduced samples can be ascribed to a change in the
oxide anions behave as Lewis acid and Brønsted conjugate number, but not in the characteristics, of the active sites in-
base sites. Once the reaction begins, the oxygen is continu- duced by the different reduction potentials of the reacting
ously consumed by reaction with CH3OH, the surface be- gases. Therefore, it is feasible to propose that Cu is more
ingprogressivelyactivatedbyincreasingnumbersofdefects dispersed or that its dispersion is more stabilised when the
generated by the reaction itself (22). Therefore, at high oxy- starting up of the reformer is performed over the catalysts
gen coverage the reaction starts with a low reaction prob- in oxidised form rather than in prereduced form. Further
ability but increases as more vacancies are created in the work concerning this relationship will be carried out in the
oxygen layer.
future.
The change in reaction product with temperature (Fig. 2)
indicates the sensitivity of the mechanism to the local
atomic structure, which is itself controlled by the reaction.
In keeping with this, the reaction evolution is accompanied
by surface/bulk reduction/reconstruction. The high concen-
4.2. Consecutive Start-Up/Shut-Down Cycles
over an Oxidised Sample
As shown in Fig. 3, a different methanol conversion pro-
tration of oxygen adatoms at the first steps of the reaction file was obtained when the sample in oxidised state was sub-
allows the evolution of the superficial methoxide toward jected to a consecutive start-up. During the second start-up
products of total oxidation. In all cases, methanol conver- the reaction began at a lower temperature, displaying a con-
sion by total oxidation does not reach its maximum stoichio- version profile and product distribution identical to those of
metric value (20% for O2/CH3OH ratio of 0.3), a maximum the prereduced sample, thus confirming the reduction of the
value—irrespective of catalyst pretreatment—of 9% being catalysts under reaction. As has been pointed out above, it
achieved. The high stoichiometric amount of O2 required appears reasonable to assume that the surfaces developed
for the total oxidation of methanol entails a fast decrease under reaction on oxidised samples are different from those
of the O2 content in the gas. This increase in the reduc- developed on prereduced counterparts. In spite of these
ing potential of the gas phase, together with the excess of structural differences at the beginning of the reaction, both
energy, implies a progressive removal of surface oxygen the prereduced sample and the reaction-reduced counter-
species, which favours the dehydrogenation pathway un- part showed the same activity behaviour (Figs. 1 and 3). The
der substoichiometric amounts of surface oxygen. When all surface changes at high temperature under a reducing at-
oxygen available for oxidation has been consumed (60% of mosphere have been associated with either the sintering of

˜
NAVARRO, PENA, AND FIERRO
118
copper (25) or the migration of ZnOx onto the Cu surface have important implications in reformers for automobile
(13, 26). These phenomena could be the origin of the differ- use since no additional equipment included to activate the
ences found between the first and the consecutive start-ups catalyst would be necessary.
over the oxidised sample. This interpretation is, however,
inconsistent with the fact that the catalyst maintained at
573 K for 3 h did not show evidence of deactivation. Addi-
tionally, the absence of thermal sintering effects associated
with the experimental protocol is supported by the consec-
utive start-ups carried out over the reaction-reduced sam-
ple, in which activity remained virtually unchanged (Fig. 5).
Therefore, the differences found between both prereduced
and reaction-reduced catalysts and the oxidised counter-
part must be related to the lattice oxygen present in the
latter. The question as to how the lattice oxygen present in
the oxidised sample influences the structural changes oc-
curring under reaction and the reason these changes lead
to higher catalytic activities in methanol decomposition re-
main obscure and require further investigation.
ACKNOWLEDGMENTS
Financial support of this work from the European Regional Develop-
ment Fund under Contract 2FD97-1405-C02-01 is acknowledged. RMN
gratefully acknowledges financial support (I3P-PC2001-2 programme)
from the European Social Fund.
REFERENCES
1. Peschka, W., Int. J. Hydrogen Energy 23, 27 (1998).
¨
2. Trimm, D. L., and Onsan, Z. I., Catal. Rev. 43(1–2), 31 (2001).
3. Takahashi, K., Takezawa, N., and Kobayashi, H., Appl. Catal. 2, 363
(1982).
4. Amphlett, J. C., Evans, M. J., Mann, R. F., and Weir, R. D., Can. J.
Chem. Eng. 63, 605 (1985).
5. Alejo, L., Lago, R., Pen˜a, M. A., and Fierro, J. L. G., Appl. Catal., A
162, 281 (1997).
6. Agrell, J., Hasselbo, K., Jansson, K., Ja¨ra¨s, S. G., and Boutonnet, M.,
Appl. Catal., A 211, 239 (2001).
5. CONCLUSIONS
7. Mizsey, P., Newson, E., Truong, T., and Hottinger, P., Appl. Catal., A
213, 233 (2001).
8. Reitz, T. L., Lee, P. L., Czaplewski, K. F., Lang, J. C., Popp, K. E., and
Kung, H. H., J. Catal. 199, 193 (2001).
9. Murcia-Mascaro´s, S., Navarro, R. M., Go´mez-Sainero, L., Costantino,
U., Nocchetti, M., and Fierro, J. L. G., J. Catal. 198, 338 (2001).
10. Gu¨nter, M. M., Ressler, T., Jentoft, R. E., and Bems, B., J. Catal. 203,
133 (2001).
11. Grunwaldt, J. D., Molenbroek, A. M., Topsøe, N. Y., Topsøe, H., and
Clausen, B. S., J. Catal. 194, 452 (2000).
The temperature-programmed start-up of the reformer
for hydrogen production by POM using a feed molar ra-
tio of O2/CH3OH = 0.3 was studied under Cu/ZnO/Al2O3
catalysts in oxidised, reduced, and reduced + air-exposed
forms. The importance of the initial state of the catalyst in
the start-up behaviour of the reformer was demonstrated.
The profile of methanol conversion and the onset temper-
atures for each chemical region contributing to the over-
all conversion differ, depending on the initial state of the 12. Sakakini, B. H., Tabata Baei, J., Watson, M. J., and Waugh, K. C.,
J. Mol. Catal. A: Chem. 162, 297 (2000).
13. Topsøe, N. Y., and Topsøe, H., J. Mol. Catal. A: Chem. 141, 95 (1999).
14. Fujita, S.-I., Moribe, S., Kanamori, Y., Kakudate, M., and Takezawa,
N., Appl. Catal., A 207, 121 (2001).
15. Mears, D. E., J. Catal. 20, 127 (1971).
catalyst. Oxidative dehydrogenation and combustion are
faster over the prereduced samples. Passivation of samples
by contact with air at room temperature only slightly re-
tards the start of the reaction. The reduction of oxidised
sample takes place under reaction and leads to a surface re- 16. Knop-Gericke, A., Ha¨vecker, M., Schedel-Niedrig, T., and Schlo¨g, R.,
Top. Catal. 15(1), 27 (2001).
17. Urban, J., Sack-Kongehl, H., and Weiss, K., Catal. Lett. 49, 101
construction with a higher methanol decomposition capac-
ity than those derived from reduction pretreatments. The
higher decomposition capacity found for the oxidised sam-
(1997).
18. Bart, J. C. J., and Sneeden, R. P. A., Catal. Today 2, 1 (1987).
ple is most likely related to differences in the number, but
not in the characteristics, of the active sites induced by the
different reduction potentials of the reacting gases. The dif-
ferences in surface features shown by the oxidised sample
under the first reaction are not maintained during the sec-
ond start-up. In this case, the reaction-reduced catalyst be-
19. Clarke, D. B., Lee, D. K., Sandoval, M. J., and Bell, A. T., J. Catal. 150,
81 (1994).
20. Bowker, M., Top. Catal. 3, 461 (1996).
21. Jones, P. M., May, J. A., Reitz, J. B., and Solomon, E. I., J. Am. Chem.
Soc. 120, 1506 (1998).
22. Dadykov, V. A., Tikhov, S. F., and Popovskii, V. V., Kinet. Katal. 27,
133 (1986).
haves identically to the prereduced counterpart. A negligi- 23. Sato, S., Iijima, M., Nakayama, T., Sodesawa, T., and Nozaki, F.,
J. Catal. 169, 447 (1997).
24. Vanderborg, N. E., Goodby, B. E., and Sringer, T. E., in “Proceedings
of the 32nd International Power Sources Symposium,” p. 623, 1986.
25. Gines, M. J. L., and Apesteguia, C. R., Latin Amer. Appl. Res. 25, 215
ble effect on methanol conversion during repeated start-up
and shut-down cycles over the oxidised sample is observed.
Consequently, the prereduction of the Cu/ZnO/Al2O3 cata-
lyst is not necessary for high conversion and selectivity for
(1995).
H2 production in POM reactions to be achieved. This may 26. Fujitani, T., and Nakamura, J., Appl. Catal., A 191, 111 (2000).