Characterization of CuO/ZnO under oxidizing conditions for the oxidative methanol reforming reaction

Article abstract of DOI:10.1016/S1381-1169(00)00296-X

Catalytic generation of hydrogen by the reaction of methanol with oxygen in the presence of steam over an industrial copper-zinc oxide catalyst was studied. The Power-Law expression successfully modeled the differential kinetics for the oxidation methanol

Full text of DOI:10.1016/S1381-1169(00)00296-X

Journal of Molecular Catalysis A: Chemical 162 Ž2000. 275–285
www.elsevier.comrlocatermolcata
Characterization of CuOrZnO under oxidizing conditions for
the oxidative methanol reforming reaction
a,)
b
b
b
a
T.L. Reitz , S. Ahmed , M. Krumpelt , R. Kumar , H.H. Kung
a
Department of Chemical Engineering, Northwestern UniÕersity, EÕanston, IL 60208, USA
b
Chemical Technology DiÕision, Argonne National Laboratory, 9700 S. Cass AÕe., Bldg. 205, Argonne, IL 60439, USA
Abstract
Catalytic generation of hydrogen by the reaction of methanol with oxygen in the presence of steam over an industrial
copper–zinc oxide catalyst was studied. Under differential oxygen conversion conditions, the catalyst remained in an
oxidized state, and the main reaction was oxidation of methanol to carbon dioxide and water. The activity was proportional
to the copper oxide surface area. The methanol consumption rate had a small positive order in methanol and oxygen Ž0.18th
order. and was suppressed by water. The catalyst deactivated with time on stream due to agglomeration of copper oxide. As
the reactor temperature increased, the rate of methanol oxidation increased, the oxygen conversion became very high, and
the catalyst away from the reactor entrance became reduced. Then, a significant rate of hydrogen production was observed.
q 2000 Elsevier Science B.V. All rights reserved.
Keywords: CuOrZnO; Oxidizing condition; Oxidative methanol reforming reaction
1
. Introduction
the liquid fuels investigated for potential reforming
feedstock, methanol has shown promise because of
the ease of handling, low cost, and high-energy
content w5,6x.
Recently, there have been substantial efforts to
investigate alternative propulsion systems for auto-
mobiles. The desire to conserve oil reserves, and
concerns over health issues and climate change are
presented as reasons for decreasing emphasis on the
use of the internal combustion engine, w1–4x. Fuel
cells are being reinvestigated as a potential alterna-
tive. Presently, the most viable candidate for auto-
mobile applications is the hydrogen fuel cell incor-
porating a proton exchange membrane. Because of
safety and infrastructure considerations, onboard H2
generation, by reforming a liquid fuel, is being ex-
tensively investigated for commercialization. Among
The steam reforming of methanol to H has been
2
successfully demonstrated w7–11x. Due to the en-
dothermicity of the reaction, significant heat is re-
quired in order to maintain the methanol reforming
reaction ŽEq. Ž1... As a result, heat required for the
steam reforming reaction would have to be supplied
by combustion of methanol, residual hydrogen from
the fuel cell, or other fuel external to the reformer.
Additionally, copper catalysts display significant re-
forming activity only after a complete reductive pre-
treatment. Therefore, sequestration of the catalyst
bed between usage cycles also represents a signifi-
cant drawback to methanol steam reforming for auto-
mobile applications.
)
Corresponding author.
1
381-1169r00r$ - see front matter q 2000 Elsevier Science B.V. All rights reserved.
PII: S1381-1169Ž00.00296-X

2
76
T.L. Reitz et al.rJournal of Molecular Catalysis A: Chemical 162 (2000) 275–285
A second catalytic technique, oxidative methanol
reforming ŽOMR., involves producing H2 from
over 482 kJrmol. CO selectivities, however, were
higher than that reported for steam reforming and the
rate of partial oxidation was found to be a strong
methanol and water while cofeeding with O . The
2
ratio of the three reactants can vary and are often
chosen such that the overall reaction is thermal–neu-
tral or only modestly exothermic. In essence, the
heat necessary to maintain steam reforming of
methanol is supplied by methanol oxidation ŽEq. Ž2..
in the reactor. The stoichiometry of the OMR reac-
tion at an oxygenrmethanol ratio of 0.25 is shown
as Eq. Ž3..
function of copper content. ZnO- and ZrO -sup-
ported Pd catalysts were also found to be highly
selective for methanol partial oxidation to CO, CO2
2
and H w16,17x.
2
Kumar et al. w5,6x examined a variety of catalyst
systems for OMR and determined unreduced
CuOrZnOrAl O to be highly active and selective
2
3
for H production. Product compositions of greater
2
than 50% ŽH dry., while maintaining less than 1%
2
CH OHqH OsCO q3H
3
2
2
2
CO, were observed for high methanol conversions.
Additionally, the exothermic nature of the reaction
allowed for rapid reformer startup of less than 200 s.
Rapid responses to load changes, necessary for auto-
motive applications, were also observed.
0
2
DH
s130.9 kJrmol
Ž
1
.
988C
3
CH OHq O sCO qH O
3
2
2
2
2
0
Reitz et al. w18x concluded that an operational
DH sy726.6 kJrmol
Ž
Ž
2
3
.
.
c
CuOrZnOrAl O OMR bed consists of regions
2
3
CH OHq0.25O q0.5H OsCO q2.5H
3
2
2
2
2
with activity varying based on the oxidation state of
0
2
the copper catalyst. When conversions of O are
2
DH
sy12.0 kJrmol
988C
low, the copper catalyst remains in the oxidized
form. Under these conditions, the combustion of
methanol was found to be dominant with only minor
selectivity to H . For higher O conversions, the
While steam reforming of methanol for H2 pro-
duction has been studied extensively w7–11x, there
are significantly fewer investigations of the OMR
2
2
reaction. The addition of O to the steam reforming
reaction was initially studied by Huang et al. w12,13x
catalyst bed temperature increased as a result of the
exothermic nature of the combustion reaction. Even-
tually, the temperature increased uncontrollably until
2
who determined the kinetics of the OMR reaction
over a reduced catalyst for high conversions. These
authors suggested that the OMR rate could be mod-
eled by considering a two-step sequence where the
overall reaction rate could be determined by the sum
of partial oxidation and steam reforming rates. For
the purposes of arriving at a rate expression, a
reaction mechanism and a rate-determining step were
assumed based upon the reaction scheme proposed
by Wachs and Madix w14x for formaldehyde oxida-
tion. A Langmuir–Hinshelwood expression was then
developed to model the data under high oxygen
conversion conditions.
Alejo et al. w15x investigated the partial oxidation
of methanol in the absence of steam over copper
complete conversion of O occurred and the catalyst
2
was shown to be in the reduced state. The conse-
quence of this run-away was that the resulting prod-
uct distribution, at complete O conversion, changed
2
dramatically, with H becoming the dominant prod-
2
uct.
Deactivation of the CuOrZnOrAl O system
2
3
during the OMR reaction represents one significant
obstacle to technical implementation. At present, the
exact cause of deactivation is not known because
of lack of information about the surface area of
the active phase, presumably CuO. Currently, the
standard method for determining the activity and
performance of copper containing catalysts is with a
reductive pretreatment. Once the catalyst has been
containing catalysts. O conversion was found to be
2
0
a significant function of temperature with nearly
complete conversion occurring by 488 K with H2
and CO selectivity a strong function of residence
time. Activation energy was observed to vary de-
pending upon copper content from 71 kJrmol to
completely reduced, the surface area of Cu can be
determined by N O chemisorption, w19–23x. Bulk
2
reduction of the sample, however, is likely to result
in significant reconstruction of the CuO particles.
Thus, the technique is not completely appropriate for

T.L. Reitz et al.rJournal of Molecular Catalysis A: Chemical 162 (2000) 275–285
277
the OMR reaction since it has been shown that,
under differential oxidizing conditions, the dominant
phase of copper is in the oxidized state, w18x.
controllers. The feed composition consisted of 10–
40% methanol, 1–12% O , 10–30% H O, 2.5%
2
2
CH Žas a tracer., and the balance N . The tempera-
4
2
The objective of this work is to examine the OMR
ture was varied from 1808C to 2258C and monitored
by use of a thermocouple in a thermocouple well
placed in the catalyst bed. The residence time was
varied by adjusting the catalyst weight from 10 to 50
mg diluted with 450–490 mg of SiC. Gas phase
concentrations were determined by on-line GC anal-
ysis ŽHP 6890. using two TCD detectors incorporat-
reaction over a commercial CuOrZnOrAl O cata-
2
3
lyst. Of particular interest is to examine the reaction
kinetics and the deactivation phenomenon observed
under differential oxidizing conditions, which has
not been studied in detail before. A reaction network
and kinetic information for the oxidizing region of
the OMR catalyst system are also presented.
X
ing a 8 , 1r80 OD HayeSep Q, ŽCH OH, H O, CO
3
2
2
X
and HCOOH using He as a carrier., and a 10 , 1r80
OD molecular sieve 13= column, ŽO , N , CO, H
2
2
2
2
. Experimental
using Ar as a carrier., for gas separations. Minor
oxygenate concentrations were determined by use of
a HP 5710A GC with a FID detector monitoring the
HayeSep Q separated effluent from the HP 6890
TCD detector. A schematic description of the flow
system is shown as Fig. 1. Mass balance closure
generally exceeded 98% and reactant conversions
2
.1. Kinetic experiments
Kinetic experiments were performed by placing a
powder catalyst of CuOrZnOrAl O diluted in SiC
2
3
ŽElectroAbrasives, 120–170 mesh., into a fused sil-
ica microreactor operating in a steady state flow
system. The catalyst was a commercial low-tempera-
ture shift catalyst, BASF K3-110 Ž120–170 mesh.,
where calculated using Eq. 4 , defined for oxygen
asanjo
Ž .
consisting of 40 wt.% CuO, 40% ZnO, and 20%
Ýn
wP
x
Al O as reported by the vendor. Methanol and
i
i
2
3
i
water were supplied by saturating N2 and O2
ŽMatheson UHP. carriers, respectively, in two sets of
jacketed saturators maintained by external tempera-
ture baths. The flow rates were kept at 100 mlrmin
ŽSTP. total flow by Brooks model 5850E mass flow
O conversions
.
4
Ž .
2
Ýn
wP
xq2wO
x
i
i
2
i
ext
Where n is the number of O atoms in a molecule of
i
product i. An analogous definition can be written for
Fig. 1. System diagram for activity determination.

2
78
T.L. Reitz et al.rJournal of Molecular Catalysis A: Chemical 162 (2000) 275–285
methanol. The products include H , CO , CO, H O,
was used to calibrate the procedure by comparing the
results to the BET surface area. The porous copper
2
2
2
CH O, HCOOH and CH OCH . Product distribu-
2
3
3
tions are reported as a molar percentage of the total
product concentration unless otherwise specified.
oxide was prepared by precipitation of the nitrate salt
q
with NaHCO . After washing to remove the Na ,
3
the filter cake was then dryed at 1008C for 24 h then
calcined in air at 3508C for 6 h.
SEM data was obtained on a Hitachi 4500 FE-
SEM apparatus. Powder XRD patterns were obtained
with a Rigaku Geigerflex XRD diffractometer with
2
.2. Catalyst characterization
Temperature programmed reaction ŽTPR. results
were obtained in a flow apparatus with a fused silica
Ni filtered CuK radiation, Žls0.15418 nm.. Scans
a
microreactor. Gas flows of 60 mlrmin consisting of
were generally performed from 10–808 2u in 0.058
increments for 2 s count times. The Scherrer equa-
tion was used to estimate crystallite sizes using the
FWHM for CuO Ž111. and ZnO Ž100. at 38.88 and
5
% H rAr ŽMatheson., were controlled by a Brooks
2
Model 5850E mass flow controller. The furnace
temperature was adjusted with a temperature con-
troller ŽOmega Engineering. to maintain a constant
3
1.88, respectively.
ramp rate of 5 Krmin. The oxygen impurity in the
H rAr feed was scrubbed by a MnO trap regener-
2
ated periodically by H treatment at 4508C. Water
2
3
. Results and discussion
generated during each TPR run was removed by a
molecular sieve trap. A TCD detector, with output to
a computer was used to monitor H uptake. Pulses
of 1 cm UHP Ar ŽMatheson. were injected periodi-
cally during each run for calibration of the TCD
signal. The system was calibrated by reducing CuO
samples in H rAr with mass balance closure of
greater than 98"2%. Conditions of each run were
controlled in order to optimize the TPR signal by
adjusting the characteristic number K ŽEq. Ž5..,
to below 200 s
w24–27x.
3
.1. Characterization of CuOrZnOrAl O
2
3
2
3
In order to determine the condition to measure
CuO surface area in CuOrZnOrAl O , a calibra-
2
3
tion curve was prepared using pure CuO. Fresh
2
samples of CuO were treated in 5% H rAr at vari-
2
ous temperatures for 60 min. After purging the cata-
lyst in He at 2258C, the uptake of O by the samples
2
y1
were determined. The uptake curves for the pure
according to Monti and Baiker
CuO and the CuOrZnOrAl O are reported in Fig.
2
3
2
. For both samples, a small plateau in the O uptake
2
So
wsx sy1
curve was observed for the samples reduced around
1008C. If this value was assumed to be due to
reoxidation of the reduced copper surface from metal
Ks
,
5
Ž .
VCo
where S is the quantity of reducible species in the
o
sample wsx mmol; V is the volumetric flow rate wsx
mlrs; C is the initial H content in the feed wsx
o
2
mmolrml.
N BET surface area determination was obtained
2
on a Coulter OmniSorb 360 at 77 K assuming a
molecular cross-section of 0.164 nm. The samples
were first outgassed at 2258C for 2 h under high
vacuum. CuO surface area was determined by partial
reduction in a flowing stream of 5% H rAr for 60
2
min at 1008C, a temperature determined by a calibra-
tion described later. After 60 min, the system was
purged in He and heated to 2258C. Pulses of 1 ml O2
ŽSTP. were then injected over the catalyst and up-
take was determined by a TCD detector. Pure CuO
Fig. 2. CuO surface titration Ža. pure CuO, Žb. CuOrZnOrAl O .
2
3

T.L. Reitz et al.rJournal of Molecular Catalysis A: Chemical 162 (2000) 275–285
279
Table 1
that the bulk CuOrZnO amalgamations are rela-
Material properties of CuOrZnOrAl O , BASF K3-110
2
3
tively separated from the Al O and the carbon filler
2
3
Parameter
Value
Ž
.
Fig. 3a–d . It is well known that Al O and carbon
2
3
2
BET surface area wsx m rg
Pore volume wsx mlrg
Average CuO particle size, A
Average ZnO particle size, A
109"3
0.28"0.01
50"10
40"10
11"3.5
have little activity for methanol oxidation at the
lower temperatures representative of this study. It is
likely then, that either CuO or ZnO or both are the
active phases with the other components being inert.
2
CuO surface area wsx m rg
3
.2. Reaction under oxidizing conditions
It was determined previously that under differen-
to cupric oxide, the surface area of copper oxide was
2
tial conditions, a fresh sample of CuOrZnOrAl O
2
3
determined to be 29.0"3 m rg, assuming a CuO
2
was active primarily for the oxidation of methanol
molecule occupies 11.6P w28,29x. This is in reason-
with a CO rH O product ratio of 1:2 w18x. Small
2
2
able agreement with the BET surface area of 20"2
2
amounts of formic acid was also detected as a pri-
m rg. For the fresh-untreated CuOrZnOrAl O
2
3
mary product, while minor H production was deter-
2
sample, the oxygen uptake at the plateau at 100–
mined to occur as a result of secondary reactions
ŽFig. 4.. Prereduction of the catalyst, under condi-
1
108C corresponded to 3.6"1.4% of the CuO
titrated ŽFig. 2.. The exposed CuO surface area
determined by this method, and other relevant prop-
erties of the BASF sample K3-110 are listed in
Table 1.
tions where bulk reduction of the copper was
assured, revealed that reoxidization occurred when
subjected to differential oxygen conversions and
dominant oxidation selectivity was again observed.
The product distribution under differential conditions
was dramatically different from selectivities ob-
served under complete O2 conversion conditions.
Fig. 3 shows FESEMrEDXS data of a fresh,
untreated CuOrZnOrAl O catalyst. They showed
2
3
that the majority of the CuO and ZnO appear in large
agglomerates in the order of 5 mm in size. Addi-
tional structures existed consisting of Al O and
Once O conversion is complete, it was found that
2
2
3
0
the bulk copper phase is completely reduced to Cu
carbon added as stabilizers and supports. It appeared
and H becomes a dominant reaction product.
2
The reaction kinetics were determined for
methanol and O conversions less than 3% and 20%,
2
respectively, and a temperature range of 180–2258C.
Fig. 3. SEM data of fresh CuOrZnOrAl O , 1000x magnifica-
2
3
tion, clockwise from top: Ža. SEM image, Žb. EDXS Cu, Žc.
EDXS Zn, Žd. EDXS Al.
Fig. 4. Product distribution as a function of O2 conversion for the
OMR reaction: Ža. H O, Žb. H , Žc. CO , Žd. HCOOH.
2
2
2

2
80
T.L. Reitz et al.rJournal of Molecular Catalysis A: Chemical 162 (2000) 275–285
Fig. 5. Correlation plot for reaction rate data for the temperature
ranges of 1808C to 2258C.
Fig. 7. Dependence of methanol consumption rate on Pmethanol, Ža.
T s1808C, Žb. T s1958C, Žc. T s2108C, Žd. T s2258C.
The rate of methanol disappearance can be repre-
sented in the Power-Law form ŽEq. Ž6...
oxidation reactions. If the reaction of methanol re-
0
.18
0.18
q2
q
0
yEa
P
P
0.14
O
duces the Cu to Cu or Cu , which is reoxidized
by oxygen, then a positive order on oxygen is not
unexpected. However, direct oxidation of adsorbed
methanol with adsorbed oxygen would also suggest a
positive order in oxygen.
methanol O2
yRCH 3OH sA exp
.
6
Ž .
o
ž /
RT
PH
2
The correlation plot of the observed versus calcu-
lated rates, as well as the dependence of the rate on
P , P
, and PH are shown in Figs. 5–8. The
The activation energy as a function of component
partial pressure are shown in Figs. 9–11. The activa-
tion energy for the reaction was found to be 115"6
kJrmol and the pre-exponential factor to be 6.0"
O
CH 3OH
O
2
2
inhibitory influence of steam was attributed to com-
petitive adsorption of water with O2 andror
methanol. It is also possible that the high concentra-
tion of water in the gas phase simply inhibits desorp-
tion of surface bound water produced during
methanol oxidation. The positive order of the O2
partial pressure with reaction rate is common for
8
0.22 y1
0
.2=10 mol ŽminPgcatPkPa
.
. No statistical
trend was observed in the activation energy as a
function of O or methanol partial pressure suggest-
2
ing a power-law kinetic expression fits the data
Fig. 6. Dependence of methanol consumption rate on Poxygen, Ža.
T s1808C, Žb. T s1958C, Žc. T s2108C, Žd. T s2258C.
Fig. 8. Dependence of methanol consumption rate on Psteam , Ža.
T s1808C, Žb. T s1958C, Žc. T s2108C, Žd. T s2258C.

T.L. Reitz et al.rJournal of Molecular Catalysis A: Chemical 162 (2000) 275–285
281
Fig. 11. Dependence of activation energy on steam partial pres-
sure.
Fig. 9. Dependence of activation energy on oxygen partial pres-
sure.
ŽFig. 12.. The initial activity for fresh samples pre-
treated in various atmospheres are shown in Table 2.
appropriately. However, the activation energy varied
more significantly with water partial pressure, rang-
ing from 134 kJrmol at 8.2 kPa to nearly 106
kJrmol at 27 kPa. A possible explanation is that as
the surface coverage of water increases, there is a
monotonic decrease in the heat of adsorption. Since
adsorbed water suppresses the reaction, a lower heat
of adsorption lowers the apparent activation energy
of the reaction. This is consistent with the findings
that, for methanol synthesis, the heat of adsorption of
water over a reduced CurZnOrAl O is a linear
The results show that treatment with N or air and
2
steam at 2258C did not cause deactivation of the
CuOrZnOrAl O sample. However, treatment at
2
3
3
508C caused significant deactivation independent of
the atmosphere. Samples treated in steam at 3508C
showed the most deactivation, with only ;33% of
the activity remaining. Preliminary evidence suggests
that the rate of deactivation decreased with increas-
ing P .
O
2
2
3
In order to test whether deactivation is associated
with reduction of the CuO in the sample, TPR was
performed on a sample deactivated to 33% of origi-
nal activity. As shown in Fig. 13, negligible differ-
ence in the amount of reducible CuO was observed
compared with a fresh, undeactivated sample. H2
consumption was determined to be 0.9 " 0.05
decreasing function with surface coverage w30x.
3
.3. DeactiÕation of CuOrZnOrAl O under oxi-
2
3
dizing conditions
As shown previously w18x, CuOrZnOrAl O was
2
3
found to deactivate with time-on-stream at 2258C
Fig. 12. Deactivation profile for CuOrZnOrAl O under oxidiz-
2
3
Fig. 10. Dependence of activation energy on methanol partial
pressure.
ing conditions, Ža. mole % H in product, Žb. O conversion, Žc.
2 2
CH OH conversion.
3

2
82
T.L. Reitz et al.rJournal of Molecular Catalysis A: Chemical 162 (2000) 275–285
Table 2
Reaction rates with deactivation treatments for various CuOrZnOrAl O samples
2
3
Deactivation treatment
Methanol conversion at 2258C
ŽWrFmethanol s11.8"0.7.
Initial rate
Mmolrg catalyst min
MoleculesrCuO site min
Fresh, no treatment
0.9"0.03%
1.6"0.11%
1.1"0.07%
1.5"0.13%
0.6"0.03%
0.4"0.02%
0.3"0.02%
0.78"0.05
1.20"0.10
0.93"0.08
1.34"0.14
0.50"0.04
0.34"0.03
0.26"0.02
5.5"0.6
4.7"0.5
5.5"0.6
6.3"0.7
4.4"0.5
2.6"0.3
4.6"0.8
N2 and steam at 2258C, for 3 h
N2 and steam at 2258C, for 24 h
Air and steam at 2258C, for 24 h
N2 only at 3508C, for 24 h
N2 and steam at 3508C, for 24 h
Air and steam at 3508C, for 24 h
molrmol CuO for the deactivated sample compared
with 0.8"0.05 molrmol CuO for the fresh sample.
The XRD patterns of some deactivated samples are
shown in Fig. 14. The XRD pattern of the fresh,
oxidized copper catalyst Žplot a. shows broad peaks
decreased with increasing CuO crystallite size, the
dependence on ZnO was much less obvious. Interest-
ingly, the fresh catalyst is much less active than
predicted by the CuO crystallite size. The major
difference between the fresh sample and a mildly
at 13.18, 24.28, and 28.18 2u indicative of the miner-
als hydrozincite and malaquite. These peaks disap-
treated sample ŽN rwater at 2258C for 24 h. are the
2
dramatic increase in the ZnO particle size followed
by the disappearance of the mineral phases. This
suggests that perhaps the CuOrZnO mineral phases
discussed previously potentially inhibit the exposure
of the active CuO site. TPR evidence also supported
this finding. While there was no difference in the
quantity of the CuO reduced ŽFig. 13b and c. the
peared after treating the samples in N and steam at
2
2
258C without affecting catalytic activity. Treatment
of the catalyst at 2258C caused a slight sharpening of
the CuO peak at 38.88 2u and the ZnO peak at 31.88
2
u. The sharpening was most noticeable after treat-
ment at 3508C, where substantial deactivation was
observed.
sample of mild treatment showed a lower Tm over
Fig. 15 shows the variation of catalytic activity,
measured at 2258C under similar space times, with
average CuO and ZnO crystallite size, as estimated
by the Scherrer equation. While catalytic activity
the fresh, untreated CuOrZnOrAl O . It is, there-
2
3
fore, postulated that a slight thermal treatment is
necessary to cause the decomposition of the mineral
phases into more active, separate constituents.
In order to confirm that deactivation was a result
of thermal sintering of CuO, the surface area of CuO
Fig. 13. TPR plots for: Ža. fresh CuOrZnOrAl O , Žb. CuOr
Fig. 14. XRD diffraction patterns for differing extents of deactiva-
2
3
ZnOrAl O treated in N and steam at 2258C for 3 hrs, Žc.
tion. Ža. Fresh catalyst, Žb. 3 hr N rSteam @2258C, Žc. 24 hr
2
3
2
2
sample with 67% deactivation.
N rsteam @2258C, Žd. 24 hr N rsteam @3508C.
2
2

T.L. Reitz et al.rJournal of Molecular Catalysis A: Chemical 162 (2000) 275–285
283
can be oxidized to formic acid, which has been
detected by GC. Under low O2 conversion condi-
tions, sufficient O2 is still available, favoring the
complete oxidation of formic acid to CO and H O
2
2
ŽEq. Ž8... However, as O conversion increases, the
2
decomposition pathway ŽEq. Ž9.. becomes more sig-
nificant relative to Eq. Ž8., consistent with experi-
mental data showing H as a secondary product ŽFig.
2
3
.. Because they are so energetically favorable, Eqs.
Ž7. and Ž8. are likely to be very labile relative to Eqs.
Ž9. and Ž10.. As a result, at lower O conversions,
2
higher formic acid production would be expected,
increasing the likelihood for desorption of formic
acid ŽEq. Ž10.. w31–34x.
Fig. 15. Dependence of average CuO crystallite size on methanol
consumption rate, measured at 2258C and WrFmethanol s11.8q
ry0.7 ŽWwsx weight of catalyst in grams, Fwsx flow of
methanol in molermin., Ža. ZnO, Žb. CuO.
CH OHqO sHCOOH
qH O
Ž
Ž
Ž
7
8
9
.
.
.
.
3
2
Žads.
2
Žads.
HCOOHŽads. q1r2O2Žads. sCO
qH O
2
2
HCOOHŽads. sCO qH
2
2
was measured using the technique discussed earlier.
The dependence of the CuO surface area on the
methanol conversion activity is shown in Fig. 16. As
with Fig. 14, a direct correlation was observed,
suggesting that the majority or all of the oxidizing
activity is due to exposed CuO. The turnover fre-
quency was calculated to be 4.7"1 moleculesPCuO
HCOOHŽads. sHCOOHŽg.
Ž
10
Another possibility is that H2 is formed by the
water gas shift reaction ŽWGS Eq. Ž11... However,
for WGS to have a significant role in H production,
2
CO would need to be at least as high as the equilib-
rium concentration under these conditions, which is
in the order of 1 ppm. To detectable limits of
approximately 50 ppm, no CO was observed over the
range of this study. H2 formation from methanol
decomposition ŽEq. Ž12.. is also unlikely, because no
y1
y1
site Pmin
for all of these catalysts. These data
strongly suggest that the decrease in methanol oxida-
tion activity over an oxidizing catalyst is primarily
due to a decrease in the dispersion of the active CuO
phase. It becomes obvious then that, in order to
maintain a high methanol oxidizing activity, it is
essential to retain high CuO dispersion. ZnO, how-
ever, is relatively resistant to sintering under the
conditions experienced during methanol oxidation.
This supports the conclusion that ZnO is not signifi-
cantly involved during the oxidation of methanol.
The production of H2 during oxidization of
methanol allows some conclusions to be drawn about
potential reaction networks for the oxidative region
during methanol reforming. As mentioned previ-
ously, CuOrZnOrAl O has been shown to deacti-
2
3
vate under oxidizing conditions, but with no change
in product selectivity w18x. As a result, H2 produc-
tion can be assumed to be produced on the same site
as methanol oxidation. One possibility is that it is
formed by the decomposition of formic acid into H2
Fig. 16. Dependence of methanol consumption rate, measured at
2
258C and WrF
s11.8qry0.7ŽWwsx weight of catalyst
methanol
in grams, Fwsx flow of methanol in molermin., on exposed CuO
and CO ŽEq. Ž8... In the presence of O , methanol
surface area.
2
2

2
84
T.L. Reitz et al.rJournal of Molecular Catalysis A: Chemical 162 (2000) 275–285
CO was detected. However, rapid CO removal by
occurs at high reaction rates. Consequently, acceler-
ated sintering of CuO occurs at the hot spots. There-
fore, under typical operating conditions, the catalyst
near the reactor entrance, which is responsible for
methanol oxidation, will deactivate with time-on-
stream. Thus, the oxidation section lengthens down
the bed with time. The point where the catalyst is
reduced also propagates down the bed resulting in an
overall decrease in the bed efficiency to produce H2.
Practically all of the H2 produced occurs over cop-
per metal in the latter section of the reactor where
steam reforming of methanol is the dominant reac-
tion.
Supported copper oxide catalysts have been shown
to be highly active for the OMR reaction. The above
results strongly suggest that the catalyst performs
multiple functions, the activity of which is primarily
determined by the oxidation state of the copper.
Because of the bifunctional aspect of this system, it
is possible that other catalysts may be more effective
for the overall generation of hydrogen. This could be
a subject of future investigations.
oxidation to CO cannot be ruled out in either case.
2
H OqCOsH qCO
Ž
Ž
11
12
.
.
2
2
2
CH OHsCOq2H
3
2
4
. Conclusions
Differential kinetics for the oxidation of methanol
over CuOrZnOrAl O has been shown to be suc-
2
3
cessfully modeled by a Power-Law expression.
Under these conditions, products of complete com-
bustion are dominant. By increasing the reaction
temperature slightly, a non-linear increase in the O2
consumption is observed followed by a complete
reduction of the copper oxide to metallic copper.
Under these conditions, the dominate reaction is
steam reforming. As a result, an operational CuOr
ZnOrAl O catalyst bed can be partitioned into two
2
3
regions with a transition region in between. Under
differential conditions, the entire catalyst bed re-
mains oxidized with activity primarily for the oxida-
tion of methanol to water and carbon dioxide. Under
typical operation conditions of high oxygen and
methanol conversions, this region of oxidized cata-
lyst is short. Methanol oxidation proceeds in this
region until the oxygen is practically all consumed.
The heat of reaction causes the catalyst temperature
to rise rapidly. After this region, the atmosphere is
substantially reducing resulting in the autothermal
reduction of the catalyst with a prominent shift to
methanol reforming as the dominant reaction.
Deactivation under oxidizing conditions has been
shown to be due to a decrease in exposed CuO. As
the sample deactivates with time-on-stream, the CuO
surface area was found to vary linearly with the
methanol oxidation rate. Consequently, the activity
was observed to decrease with increasing CuO crys-
tallite size. Similar trends were not observed with
ZnO. This data is fairly conclusive that CuO is the
principal active phase responsible for methanol com-
Acknowledgements
Support of this work by the Argonne National
Laboratory of the Department of Energy is gratefully
acknowledged.
References
w1x US Congress, Office of Technology Assessment, OTA-ETI-
6
38, Washington, DC, US Gov. Printing Office, Sept., 1995.
w2x US Congress, Office of Technology Assessment, OTA-ETI-
89, Washington, DC, US Gov. Printing Office, July, 1994.
w3x US Congress, Office of Technology Assessment, OTA-O-
12, Washington, DC, US Gov. Printing Office, July, 1989.
5
4
w4x Congressional Research Service, Environmental and Natural
Resource Policy Division, June 3, 1992.
w5x R. Kumar, S. Ahmed, M. Krumpelt, K.M. Myles, Argonne
National Laboratory Report, ANL-92r31, Argonne, IL, USA,
1992.
w6x R. Kumar, S. Ahmed, M. Krumpelt, 1996 Fuel Cell Seminar
bustion in the presence of gas phase O . Tempera-
2
Program and Abstracts, 1996, p. 750.
w7x C.J. Jiang, D.L. Trimm, M.S. Wainwright, N.W. Cant, Appl.
ture appears to be the most significant cause for CuO
sintering. Despite attempts to minimize hot spots in
the catalyst by inclusion of diluents, the temperature
control was still not sufficient such that local heating
Catal. 93 Ž1993. 245.
C.J. Jiang, D.L. Trimm, M.S. Wainwright, N.W. Cant, Appl.
Catal. 97 Ž1993. 145.
8
w x

T.L. Reitz et al.rJournal of Molecular Catalysis A: Chemical 162 (2000) 275–285
285
w9x N. Takezawa, N. Iwasa, Catal. Today 36 Ž1997. 45.
w10x B.A. Peppley, J.C. Amphlett, L.M. Kearns, R.F. Mann, Appl.
Catal. 179 Ž1999. 21.
w24x D.A.M. Monti, A. Baiker, J. Catal. 83 Ž353. Ž1983..
w25x P. Malet, A. Caballero, J. Chem. Soc., Faraday Trans. 84
Ž2369. Ž1988..
w11x B.A. Peppley, J.C. Amphlett, L.M. Kearns, R.F. Mann, Appl.
w26x G. Fierro, M. Lo Jacono, M. Inversi, P. Porta, R. Lavecchia,
Catal. 179 Ž1999. 31.
F. Cioci, J. Catal. 148 Ž1994. 709.
w12x T.J. Huang, S.W. Wang, Appl. Catal. 24 Ž1986. 287.
w13x T.J. Huang, S.L. Chren, Appl. Catal. 40 Ž1988. 43.
w14x I.E. Wachs, R.J. Madix, J. Catal. 53 Ž1978. 208.
w15x L. Alejo, R. Lago, M.A. Pena, J.L.G. Fierro, Appl. Catal.
w27x G. Fierro, M. Lo Jacono, M. Inversi, P. Porta, F. Cioci, R.
Lavecchia, Appl. Catal. 137 Ž1996. 327.
w28x A.F. Wells, Structural Organic Chemistry, 3rd edn., Oxford
Univ. Press, 1962, 868 pp.
1
62 Ž1997. 281.
w29x J.J.F. Scholten, J.A. Konvalinka, Trans. Faraday Soc. 65
w16x M.L. Cubeiro, J.L.G. Fierro, J. Catal. 179 Ž1998. 150.
w17x M.L. Cubeiro, J.L.G. Fierro, Appl. Catal. 168 Ž1998. 307.
w18x T.L. Reitz, S. Ahmed, M. Krumpelt, R. Kumar, H.H. Kung,
Ž1969. 2465.
w30x J. Skrzypek, J. Sloczynski, S. Ledakowiez, Methanol Synthe-
sis, Polish Scientific Publishers, Warsaw, 1994, ISBN 83-01-
Proc. 12th Int. Cong. Catal., accepted for publication.
w19x R.M. Dell, F.S. Stone, P.F. Tiley, Trans. Faraday Soc. 49
Ž1953. 195.
11490-8.
w31x N.W. Cant, S.P. Tonner, D.L. Trimm, M.S. Wainwright, J.
Catal. 91 Ž1985. 197.
w20x R.M. Dell, F.S. Stone, P.F. Tiley, Trans. Faraday Soc. 49
w32x R. Ramos, G. Ruiz, M.L. Rojas, J.L.G. Fierro, Appl. Catal.
Ž1953. 201.
68 Ž1991. 217.
w21x J. Osinga, B.G. Linsen, W.P. Van Beek, J. Catal. 7 Ž1967.
w33x L. Domokos, T. Katona, A. Molnar, A. Lovas, Appl. Catal.
142 Ž1996. 151.
2
77.
w22x K.J. Soerensen, N.W. Cant, Catal. Lett. 33 Ž1995. 117.
w23x A. Dandekar, M.A. Vannice, J. Catal. 178 Ž1998. 621.
w34x S. Sato, M. Iijima, T. Nakayama, T. Sodesawa, F. Nozaki, J.
Catal. 169 Ž1997. 447.