In situ ATR-IR spectroscopic and reaction kinetics studies of water-gas shift and methanol reforming on Pt/Al2O3 catalysts in vapor and liquid phases

Article abstract of DOI:10.1021/jp045470k

Reaction kinetics measurements of the water-gas shift reaction were carried out at 373 K on Pt/Al₂O₃ in vapor phase to investigate the effects of CO, H₂, and H₂O partial pressures. Results of in situ ATR-IR studies conducted in vapor phase under similar conditions suggest that the Pt surface coverage by adsorbed CO is high (～90% of the saturation coverage), leading to a negligible effect of the CO pressures on the rate of reaction. The negative reaction order with respect to the H₂ pressure is caused by the increased coverage of adsorbed H atoms, and the fractional positive order with respect to the water pressure is consistent with non-equilibrated H₂O dissociation on Pt. Results of in situ ATR-IR studies carried out at 373 K show that the presence of liquid water leads to a slight decrease in the Pt surface coverage by adsorbed CO (～80% of the saturation coverage) when the CO partial pressure is the same as in the vapor-phase studies. The rate of the WGS reaction in the presence of liquid water is comparable to the rate under complete vaporization conditions when other factors (such as CO partial pressure) are held constant. Reaction kinetics measurements of methanol reforming were carried out at 423 K over a total pressure range of 1.36-5.84 bar. In situ ATR-IR studies were conducted at 423 K to determine the Pt surface coverage by adsorbed CO in completely vaporized methanol feeds and in aqueous methanol solutions. The decomposition of methanol is found to be slower during the reforming of methanol in liquid phase than in vapor phase, which leads to a lower rate of hydrogen production in liquid phase (0.08 min^-1 at 4.88 bar) than in vapor phase (0.23 min^-1 at 4.46 bar). The lower reaction order with respect to methanol concentration observed for vapor-phase versus liquid-phase methanol reforming (0.2 versus 0.8, respectively) is due to the higher extent of CO poisoning on Pt for reforming in vapor phase than in liquid phase, based on the higher coverage by adsorbed CO observed in completely vaporized methanol feeds (55-60% of the saturation coverage) than in aqueous methanol feed solutions (29-40% of the saturation coverage).

Full text of DOI:10.1021/jp045470k

2810
J. Phys. Chem. B 2005, 109, 2810-2820
In Situ ATR-IR Spectroscopic and Reaction Kinetics Studies of Water-Gas Shift and
Methanol Reforming on Pt/Al2O3 Catalysts in Vapor and Liquid Phases
Rong He, Rupali R. Davda, and James A. Dumesic*
Department of Chemical and Biological Engineering, UniVersity of Wisconsin-Madison,
Madison, Wisconsin 53706
ReceiVed: October 5, 2004; In Final Form: NoVember 22, 2004
2 3
Reaction kinetics measurements of the water-gas shift reaction were carried out at 373 K on Pt/Al O in
vapor phase to investigate the effects of CO, H , and H O partial pressures. Results of in situ ATR-IR studies
2 2
conducted in vapor phase under similar conditions suggest that the Pt surface coverage by adsorbed CO is
high (∼90% of the saturation coverage), leading to a negligible effect of the CO pressures on the rate of
2
reaction. The negative reaction order with respect to the H pressure is caused by the increased coverage of
adsorbed H atoms, and the fractional positive order with respect to the water pressure is consistent with
non-equilibrated H O dissociation on Pt. Results of in situ ATR-IR studies carried out at 373 K show that the
2
presence of liquid water leads to a slight decrease in the Pt surface coverage by adsorbed CO (∼80% of the
saturation coverage) when the CO partial pressure is the same as in the vapor-phase studies. The rate of the
WGS reaction in the presence of liquid water is comparable to the rate under complete vaporization conditions
when other factors (such as CO partial pressure) are held constant. Reaction kinetics measurements of methanol
reforming were carried out at 423 K over a total pressure range of 1.36-5.84 bar. In situ ATR-IR studies
were conducted at 423 K to determine the Pt surface coverage by adsorbed CO in completely vaporized
methanol feeds and in aqueous methanol solutions. The decomposition of methanol is found to be slower
during the reforming of methanol in liquid phase than in vapor phase, which leads to a lower rate of hydrogen
-
1
-1
production in liquid phase (0.08 min at 4.88 bar) than in vapor phase (0.23 min at 4.46 bar). The lower
reaction order with respect to methanol concentration observed for vapor-phase versus liquid-phase methanol
reforming (0.2 versus 0.8, respectively) is due to the higher extent of CO poisoning on Pt for reforming in
vapor phase than in liquid phase, based on the higher coverage by adsorbed CO observed in completely
vaporized methanol feeds (55-60% of the saturation coverage) than in aqueous methanol feed solutions
(29-40% of the saturation coverage).
Introduction
spectroscopic studies of catalytic solid-liquid interfaces have
been conducted using attenuated total reflectance infrared (ATR-
IR) spectroscopy. Previously, in situ ATR-IR was used to study
adsorption of oxygenated compounds (ethyl acetate, carboxylic
acid) on oxide thin films grown on the ATR crystal using sol-
The production of hydrogen for fuel cells and other industrial
applications from renewable biomass-derived resources is a
major challenge as global energy generation moves toward a
“
hydrogen society”. We recently described a catalytic process
5-8
gel methods. Catalytic solid-liquid interfaces involving noble
metals have also been studied using in situ ATR-IR. Zippel et
al. investigated the adsorption of CO on Pt and Pd films (of 5
nm thick) from dry and wet gases and from liquid water with
for production of H2 from aqueous solutions of biomass-derived
oxygenated compounds such as methanol, ethylene glycol,
glycerol, sugars, and sugar alcohols.1 This process takes place
over supported Pt catalysts at relatively low temperatures (e.g.,
-3
9
ATR-IR. Recently, Baiker and co-workers studied the adsorp-
500 K) in the liquid phase, which eliminates the need to vaporize
tion of CO from CH2Cl2 solvent on thin films of Pt (1 nm) and
Pt/Al2O3 (1 nm Pt on top of 100 nm of Al2O3) coated on a Ge
water and the oxygenated hydrocarbon, and produces hydrogen
with low levels of CO (<300 ppm).3,4 Reaction kinetics studies
1
0
crystal. The reaction of CO2 with hydrogen in cyclohexane
to form CO on the same Pt/Al2O3 model catalyst was followed
carried out for the aqueous-phase reforming of methanol and
ethylene glycol suggest that the reforming may take place via
CO as an intermediate product, which is subsequently converted
to CO2 by the water-gas shift (WGS) reaction. In this respect,
monitoring the surface of Pt/Al2O3 catalysts in liquid phase
under conditions similar to reaction kinetics studies should
provide insight into the surface chemistry of this catalytic
process.
1
1
by in situ ATR-IR spectroscopy at 313 K. In another study,
the adsorption of CO, pyridine, quinoline, 2-methylquinoline,
and cinchonidine on Pd/Al2O3 and Pt/Al2O3 model catalysts was
1
2
studied at 283 K in CH2Cl2 solvent. In situ ATR-IR has also
been extended to study supported noble metal catalyst in powder
form. Burgi et al. studied enantioselective hydrogenation of a
pyrone over a 5 wt % Pd/TiO2 catalyst using ATR-IR measure-
In situ spectroscopic studies in the liquid phase are difficult
due to interference from the bulk liquid. In this respect,
13
ments combined with phase-sensitive detection. The adsorption
of CO, formaldehyde, ethanol, and butyronitrile on a 5 wt %
Pt/Al2O3 catalyst was studied in various solvents including
*
To whom correspondence should be addressed. Fax: (608) 262-5434.
1
4
E-mail: dumesic@engr.wisc.edu.
water, ethanol, and hexane. To our knowledge, the above in-
1
0.1021/jp045470k CCC: $30.25 © 2005 American Chemical Society
Published on Web 01/26/2005

Methanol Reforming and Water-Gas Shift Reaction
J. Phys. Chem. B, Vol. 109, No. 7, 2005 2811
situ ATR-IR studies were carried out at near room temperature
prepared via ion-exchange.35,36 The alumina support (150 m2/
g) was placed in contact with an aqueous solution of tetraamine
platinum nitrate (Pt(NH3)4(NO3)2, Strem Chemicals) for 24 h
on a rotary shaker with an oxide surface area per liter of solution
(<313 K) and ambient pressure. In the present study, we have
extended in situ ATR-IR studies of catalytic solid-liquid
interfaces to the elevated temperature of 423 K and a pressure
of 5.84 bar for studying the aqueous-phase reforming reaction
under conditions similar to reaction kinetics studies.
2
of 500 m /L. The pH value of the aqueous solution was adjusted
to 11.5 using NaOH prior to the addition of the alumina support.
The ion-exchanged sample was dried in an oven at 373 K for
24 h and was subsequently heated to 533 K (1.3 K/min) in a
Reaction kinetics studies have been carried out previously
in our group for the aqueous-phase reforming of methanol and
ethylene glycol to investigate the effect of feed concentration
3
mixture of 10% O2/He (300 cm (STP)/min) in a Pyrex cell and
4
and system pressure on the reaction rate. These studies showed
held at 533 K for 2 h. The Pt loading on the catalyst was 0.9
wt % as determined by atomic absorption spectroscopy of the
calcined samples (AAS/ICP).
that the rate of ethylene glycol reforming was similar to the
rate of methanol reforming, suggesting that the C-C cleavage
step involved in reforming of ethylene glycol is not rate-limiting.
Therefore, methanol reforming can serve as a model reaction
for the reforming of oxygenated hydrocarbons with the general
formula CnH2n+2On.
Prior to collection of reaction kinetics data, the calcined
catalyst was reduced on the kinetics apparatus in H2 (100 cm3
(STP)/min). The reactor was heated to 533 K (1.3 K/min) and
held at this temperature for 2 h. Carbon monoxide chemisorption
The decomposition of methanol on Pt has been studied under
2
at 300 K was performed as described elsewhere on the calcined
15-17
UHV conditions by HREELS and TPD.
On a clean Pt-
catalyst after an identical treatment, and the number of catalytic
sites was taken to be the irreversible CO uptake (i.e., 30 µmol/
g).
(
111) surface, methanol decomposes at temperatures above 140
15
K to form adsorbed CO and H. In another study, the
decomposition of methanol on Pt(111) is observed at temper-
ature between 180 and 200 K and is found to be controlled by
the presence of defect sites, because only 1-2% of a monolayer
The apparatus used to conduct reaction kinetics studies of
the WGS reaction (vapor phase) and methanol reforming (vapor
4
and liquid phase) is described in detail elsewhere. Briefly, liquid
16
of CO was formed regardless of the initial methanol coverage.
feed is introduced in an up-flow configuration into the reactor,
and the effluent is cooled with cooling water before entering
the gas-liquid separator. For reaction kinetics studies of the
WGS reaction in vapor phase, CO mixtures balanced with N2
and/or H2, together with deionized water (introduced at 0.01-
Recently, theoretical studies were carried out for methanol
decomposition on Pt(111).18-20 The pathway involving initial
C-H activation is found to be more favorable energetically than
the pathway involving initial O-H activation.
The above studies were conducted in the absence of liquid
water. The electrooxidation of methanol on Pt electrodes in
electrolyte solutions has been extensively studied using in situ
infrared spectroscopy due to the potential of direct methanol
0.06 mL/min with a syringe pump), were preheated before being
introduced into a 25.4-mm o.d. (1 in.) and 23-mm i.d. Pyrex
tubular reactor containing 0.5 g of catalyst. The total system
pressure was regulated from 1.01 to 1.63 bar. The CO
concentration of the effluent gas stream was obtained using a
Siemens Ultramat 5E gas analyzer.
21,22
fuel cells (DMFC).
In situ infrared spectroscopic studies of
methanol electrooxidation have been carried out on polycrys-
2
3-27
28,29
talline Pt,
single-crystal Pt,
and Pt supported on glassy
Reaction kinetics studies of the vapor- and liquid-phase re-
forming of methanol were conducted using solutions of 2 or 5
carbon.3
0-32
In general, these studies identified adsorbed CO
as the predominant species on Pt electrodes. The CO coverage
on a 20% Pt/C electrocatalyst (with an average Pt particle size
of 2.5 nm) was found to be 0.4-0.5 of saturation coverage in
wt % methanol in deionized water. These solutions were fed at
1
0
.05 mL/min with an HPLC pump to a 12.7-mm o.d. ( /2 in.)
and 11.3-mm i.d. stainless steel tubular reactor containing 4.5
g of catalyst. The total pressure of the system was regulated
from 1.36 to 5.84 bar with N2 carrier gas, which bubbled through
the effluent and swept the gaseous products for analysis to gas
chromatographs. The results of replicate measurements agreed
to within (5%.
In Situ ATR-IR Spectroscopy. In situ ATR-IR studies of
the WGS reaction and methanol reforming were carried out
using a 2.1 wt % Pt/Al2O3 catalyst prepared by multiple incipient
wetness impregnation of fine alumina powder (γ-Al2O3, Alfa
Aesar, mean particle size 3 µm) with an aqueous solution of
tetraamine platinum nitrate (Pt(NH3)4(NO3)2, Strem Chemicals).
The impregnated sample was calcined and characterized in ways
identical to that of the sample used for reaction kinetics studies.
The Pt loading on this catalyst was determined with AAS/ICP,
and the amount of irreversible CO uptake at 300 K on this
catalyst is 61 µmol/g.
The calcined catalyst powder was deposited onto the ATR
crystal (ZnSe, 45°, 80 × 10 × 3 mm, Thermo Electron) using
a suspension of the catalyst (20 mg) in 4 mL of deionized water
that had been placed in an ultrasonic bath for 8-12 h (to obtain
a uniform suspension). The suspension was spread onto one
side of the ZnSe crystal and dried under ambient conditions
for at least 12 h before use. The amount of suspension used
was 0.6 mL in the experiments conducted in vapor phase, and
it was 0.2 mL (diluted into 0.4 mL deionized water) for the
0
.1 M methanol at 0 V, based on comparison made between
CO bands formed via methanol dissociation and by direct
adsorption from solutions saturated with CO.32 Jusys et al.
observed that dehydrogenation of methanol was inhibited as the
CO coverage in 0.1 M methanol reaches about half of the
33
saturation coverage on a similar 20% Pt/C electrocatalyst. Park
et al. observed that the extent of CO formation on Pt diminished
with decreasing Pt particle size on various Pt/C samples with
3
4
average Pt particle sizes ranging from 8.8 to 2.0 nm.
In the first part of this paper, we focus on the WGS reaction
on Pt catalysts in vapor phase and in the presence of liquid
water at 373 K. In situ ATR-IR studies are conducted to
investigate the effects of CO and H2 in vapor phase on the
reaction kinetics measured under similar conditions, and these
results are compared with spectra from in situ ATR-IR studies
conducted for the WGS reaction in the presence of liquid water
to probe the effects caused by liquid water. In the second part
of the paper, we study the reforming of methanol at 423 K in
vapor phase and liquid phase using in situ ATR-IR spectroscopy
and reaction kinetics measurements to obtain an understanding
of the differences between methanol reforming in vapor-phase
and in liquid-phase reaction conditions.
Experimental Section
Reaction Kinetics. Reaction kinetics data were collected
using an alumina-supported catalyst (γ-Al2O3 extrudate, Topsøe)

2812 J. Phys. Chem. B, Vol. 109, No. 7, 2005
He et al.
TABLE 1: Operating Conditions and Results of Reaction Kinetics Studies of the Water-Gas Shift Reaction in Vapor Phase
condition no.
1
2
3
4
5
6
7
8
system pressure (bar)
1.65
277
1.68
574
1.67
990
1.64
1000
68.1
0.02
1.22
0.41
0.54
0.68
6.0
1.64
1005
69.6
0.02
1.21
1.03
1005
70.6
0.01
0.88
1.25
1005
71.0
0.03
0.82
1.65
1005
72.6
0.06
0.82
dry gas CO concentration (ppm)
3
dry gas flow rate (cm /min)
69.6
0.02
0.34
0.44
1.21
0.00
5.7
71.5
0.02
0.97
0.44
1.24
0.00
9.3
68.7
0.02
1.17
0.42
1.25
0.00
3.6
water feed (mL/min)
CO partial pressure (mbar)
H
2
H
2
N
2
O partial pressure (bar)
partial pressure (bar)
partial pressure (bar)
0.43
0.15
0.43
0.83
-
4
-4
-4
-4
1.9 × 10
0.8 × 10
1.0 × 10
1.8 × 10
1.21
0.88
0.82
0.81
conversion (%)
16.0
8.7
12.3
22.3
TOF (min-1)
0.0086
0.0076
0.0081
0.0128
0.0343
0.0189
0.0270
0.0496
experiments conducted in liquid phase to ensure a good seal
made by the O-ring. The film thickness is estimated to be ∼5
µm based on the density of alumina when 0.6 mL of suspension
was used, and the alumina support used for this catalyst has an
average particle size of 3 µm. The evanescent wave generated
at the interface (with a penetration depth of ∼1 µm) thus probes
part of the catalyst film, along with the interstitial solution.
spectrum was collected of the calcined catalyst layer in liquid
water at 303 K with flowing He. The catalyst layer was then
heated to 373 K (1.75 K/min) in liquid water with flowing He,
and another reference spectrum was collected. The catalyst layer
in liquid water was exposed to 333 and 1000 ppm CO/H2
sequentially for 6 and 1.5 h, respectively, and a spectrum was
taken at the end of each period. The catalyst layer was then
cooled to 303 K in 1000 ppm CO/H2, and a final spectrum was
taken in the presence of liquid water. In another experiment,
the catalyst layer in liquid water was first exposed to 333 ppm
CO/H2 for 6 h, and then exposed to 333 and 1000 ppm CO/He
sequentially (both for 1.5 h), with a spectrum taken at the end
of each period. The catalyst layer was subsequently cooled to
303 K in 1000 ppm CO/He, and a final spectrum was taken in
the presence of liquid water.
ATR-IR spectra were recorded using a horizontal flow cell
(Thermo Electron), in which the coated ATR crystal was sealed
with a viton O-ring to a Teflon-coated aluminum frame. The
length and width of the oval-shaped exposed area are 70 × 8
mm, respectively. The total volume of the flow cell is ∼65 µL.
The flow cell was heated to 423 K with heating cartridges
integrated in the aluminum frame and pressurized to 5.84 bar.
The cell was mounted onto an ATR accessory (Thermo
Electron) within the FTIR spectrometer (Matteson Galaxy 3020).
The cell was connected to an apparatus similar to the reaction
kinetics apparatus. Liquid feed was introduced into the cell with
an HPLC pump. Prior to entering the flow cell, the reactant
stream was preheated to the same temperature as the flow cell
In situ ATR-IR studies of the vapor-phase reforming of
methanol were carried out at 423 K and at the pressure of 1.36
or 3.08 bar. A reference spectrum was first collected of the dry
3
catalyst layer at 303 K in He (50 cm (STP)/min). The catalyst
layer was then heated to 423 K (2 K/min) in He. The He gas
flow into the ATR-IR flow cell was stopped, deionized water
was introduced into the ATR-IR flow cell at 0.05 mL/min for
1 h, and a reference spectrum was collected. The catalyst layer
was then dried in flowing He for 0.5 h and reduced in flowing
1
by flowing through a heated titanium tube (3.2-mm o.d. ( /8
in.), 0.61-m long).
In situ ATR-IR studies of the WGS reaction in vapor phase
were carried out at 373 K and 1.63 bar. A reference spectrum
was first collected on the dry catalyst layer at 303 K in He (70
3
H2 for 0.5 h, both gases flowing at 50 cm (STP)/min. The H2
3
cm (STP)/min). The catalyst layer was then heated to 373 K
flow was stopped, aqueous methanol feed was introduced into
the flow cell (where the feed was completely vaporized) at 0.05
mL/min for 1 h, and a spectrum was collected. The catalyst
layer was then dried in flowing H2 for 0.5 h at 423 K, and cooled
to 303 K (2 K/min). A final spectrum was collected at 303 K
after the H2 was replaced with 1000 ppm CO/H2 for 0.5 h.
(1.75 K/min) in He with deionized water flowing at 0.02 mL/
min (leading to 0.4 bar H2O in vapor phase), and another
reference spectrum was collected. The calcined catalyst layer
was subsequently exposed to (and reduced by) 333 ppm CO/
3
H2 (dry basis, 70 cm (STP)/min) with 0.4 bar H2O in vapor
phase for 1.5 h, and a spectrum was taken. The CO concentration
in H2 was increased to 1000 ppm, and another spectrum was
collected. The flow of deionized water was then stopped, and
another spectrum was taken over the drying period of 0.5 h.
The catalyst layer was subsequently cooled to 303 K in 1000
ppm CO/H2, and a final spectrum was collected. In another
experiment, the catalyst layer was first exposed to 333 ppm
CO/H2, and then exposed to 333 and 1000 ppm CO/He
sequentially, all in the presence of 0.4 bar H2O vapor. Each
gas mixture was given 1.5 h to reach steady state, and a spectrum
was taken at the end of each period. The catalyst layer was
then dried and cooled to 303 K in 1000 ppm CO/He, and two
more spectra were taken at 373 and 303 K, respectively.
In situ ATR-IR studies of the liquid-phase reforming of
methanol were carried out at 423 K and 5.84 bar. A reference
spectrum was first collected of the calcined catalyst layer at
3
303 K with He flowing at 50 cm (STP)/min and deionized
water flowing at 0.1 mL/min. The catalyst layer was then heated
3
to 423 K (2 K/min) with He flowing at 3 cm (STP)/min and
deionized water flowing at 0.2 mL/min, and another reference
spectrum was collected. The catalyst layer was subsequently
dried in flowing He for 0.5 h and reduced in flowing H2 for 0.5
3
h, both gases flowing at 50 cm (STP)/min. The H2 flow was
3
then reduced to 3 cm (STP)/min, 2 or 5 wt % methanol in
deionized water was introduced into the flow cell at 0.2 mL/
min, and three spectra were collected sequentially over a period
of 1 h. The catalyst layer was then dried briefly (<1 min) in
In situ ATR-IR studies of the WGS reaction in the presence
of liquid water were carried out at 373 K and 2.22 bar (to keep
the partial pressures of dry gases the same as in the vapor-
phase studies conducted at 1.63 bar). The flow rate of deionized
water was maintained at 0.10 mL/min throughout the experi-
3
H2 flowing at 50 cm (STP)/min, followed by cooling to 303 K
(2 K/min). The dry catalyst layer was exposed to 1000 ppm
3
CO/H2 flowing at 50 cm (STP)/min for 1 h, deionized water
was then introduced into the flow cell at 0.1 mL/min, and a
final spectrum was taken.
3
ments, and the gas flow rate was 70 cm (STP)/min. A reference

Methanol Reforming and Water-Gas Shift Reaction
J. Phys. Chem. B, Vol. 109, No. 7, 2005 2813
Figure 2. (A) ATR-IR spectra of CO adsorbed on Pt/Al
exposed to (a) 333 ppm CO/H (dry basis) with 0.4 bar H
b) 1000 ppm CO/H (dry basis) with 0.4 bar H O at 373 K, (c) 1000
ppm CO/H (dry basis) after the water feed was stopped at 373 K, and
d) 1000 ppm CO/H (dry basis) after the sample was subsequently
cooled to 303 K; (B) ATR-IR spectra of CO adsorbed on Pt/Al
collected under conditions similar to the spectra in (A), except that the
gas mixtures are He-balanced instead of H -balanced.
O
2 3
catalyst
2
2
O at 373 K,
(
2
2
2
(
2
2 3
O
2
in the presence of 1.2 bar of H2 and 0.4 bar of H2O
corresponding to columns 1-3 in Table 1). The reaction order
with respect to CO partial pressure is 0.02 between 0.34 and
.17 mbar of CO. Figure 1B shows the rate of the WGS reaction
(
1
versus the H2 partial pressure in 1.2 mbar of CO and 0.4 bar of
H2O (corresponding to column 3-5 in Table 1). The reaction
order with respect to H2 partial pressure is -0.22 between 1.9
-
4
×
10 and 1.25 bar of H2. The effect of H2O partial pressure
in the presence of 0.82-0.88 mbar of CO is shown in Figure
C (corresponding to columns 6-8 in Table 1). The reaction
1
order with respect to H2O partial pressure is 0.55 between 0.15
and 0.83 bar of H2O. Previous studies by Genoble et al. reported
that the order with respect to CO is -0.21 and the order with
Figure 1. (A) The effect of CO partial pressure on the rate of WGS
reaction over the range of 0.34-1.17 mbar at 373 K and a total pressure
2
of 1.6 bar; (B) the effect of H partial pressure on the rate of WGS
-
4
37
reaction over the range of 1.9 × 10 to 1.25 bar at 373 K and a total
respect to H2O is 0.74 on a Pt/Al2O3 catalyst at 543 K. Lam
pressure of 1.6 bar; and (C) the effect of H O partial pressure on the
2
et al. found a reaction order of 0.45 with respect to CO, 0.37
rate of WGS reaction over the range of 0.15-0.83 bar at 373 K and a
total pressure range of 1.0-1.6 bar.
with respect to H2O, and -0.23 with respect to H2 on Pt/Al2O3
3
8
at a much higher temperature of 817 K.
In Situ ATR-IR Studies of WGS Reaction. Results of in-
situ ATR-IR studies of the WGS reaction in vapor phase at
73 K and 1.63 bar are shown in Figure 2. Spectra a and b in
All infrared spectra were collected in the absorbance mode
-
1
using 1024 scans with a resolution of 4 cm . Spectra reported
in the present study are spectra of the catalyst under controlled
conditions minus the IR spectrum of the calcined sample
hydrated similarly at the same temperature and pressure.
3
Figure 2A were collected while the Pt/Al2O3 catalyst was
exposed to 333 and 1000 ppm CO/H2 sequentially with 0.4 bar
-
1
H2O vapor at 373 K. A band at 2011 cm and another broad
-
1
band at ∼1810 cm were observed in spectrum a, which can
Results
be assigned to linear-bonded (L) CO and bridge-bonded (B)
-
1
Reaction Kinetics Studies of WGS Reaction. Table 1
summarizes the operating conditions and results of reaction
kinetics measurements of the WGS reaction in the vapor phase.
The partial pressures of gases are calculated on the basis of the
water feed rate, gas composition, and gas flow rate. Figure 1A
shows the rate of WGS reaction versus the CO partial pressure
CO, respectively. The frequency of 2011 cm for the CO(L)
-
1
band is lower than the normal range of 2040-2070 cm
observed for CO adsorption on Pt/Al2O3. We have observed
-
1
CO(L) bands at 2044 and 2057 cm on this catalyst in 10%
CO in H2 and He, respectively, at 373 K without H2O vapor,
which can be assigned to CO adsorbed on Pt sites with low

2814 J. Phys. Chem. B, Vol. 109, No. 7, 2005
He et al.
coordination numbers.39-41 The Pt dispersion of this catalyst
reduced at 533 K is approximately 80% based on CO chemi-
sorption results, which indicates that the average size of the Pt
particle is ∼1.5 nm, assuming the surface of the Pt particles
consists of equal percentages of (111), (110), and (100) surfaces.
-
1
Therefore, the stretching frequency of 2057 cm observed in
0% CO/He is consistent with the small Pt particle size. The
1
-
1
slightly lower frequency of 2044 cm observed in 10% CO/
H2 can be attributed to coadsorption of H2, because previous
studies have shown that coadsorption of H2 with CO on Pt leads
to a small decrease of CO stretching frequency by 5-10
-
1 42,43
cm .
Importantly, several studies have shown that coad-
sorption of H2O with CO leads to a decrease in frequency of
linear-bonded CO. In particular, Bourane et al. observed a shift
-
1
of the CO stretching frequency from 2075 cm in 1% CO/He
-
1
44
to 2045 cm in 1% CO/3.5% H2O/He on Pt/Al2O3 at 300 K.
-
1
A shift of the CO(L) band from 2065 to 2050 cm was
observed on Pt/Al2O3 with the introduction of an excess of H2O
vapor. On a thin Pt layer, the presence of H2O vapor led to a
shift of the CO(L) band from 2060 to 2030 cm . Therefore,
on the basis of these previous studies, we attribute the low
frequency of 2011 cm observed to the adsorption of CO in
the presence of H2O vapor and H2.
4
5
-
1 9
-
1
The CO(L) band in Figure 2A shifted to 2014 cm-1 and
increased slightly in intensity as the CO concentration in H2
was increased to 1000 ppm. The CO(L) band then shifted to
-
1
Figure 3. (A) ATR-IR spectra of CO adsorbed on Pt/Al
catalyst in liquid water was exposed to (a) 333 ppm CO/H (dry basis)
at 373 K, (b) 1000 ppm CO/H (dry basis) at 373 K, and (c) 1000 ppm
CO/H (dry basis) after the sample was cooled to 303 K; (B) ATR-IR
spectra of CO adsorbed on Pt/Al collected under conditions similar
to the spectra in (A), except that the gas mixtures are He-balanced
instead of H -balanced.
2 3
O as the
2023 cm during the drying period of 0.5 h in 1000 ppm CO/
2
H2, and the integrated intensity of this band in spectrum c is
essentially the same as in spectrum b, although the band has
lower intensity and larger half-width. Bourane et al. reported
similar changes in band shape, and they found that the extinction
coefficient of linear-bonded CO species is not modified strongly
in the presence of H2O.44 The intensity of the CO(L) band then
increased slightly after the sample was cooled to 303 K in 1000
ppm CO/H2 (spectrum d). The ratios of integrated intensities
of the CO(L) bands in spectra a and b to that in spectrum d in
Figure 2A are 0.94 and 0.97, respectively.
2
2
2 3
O
2
-
1
observed a CO(L) band at 2058 cm with a shoulder at 2000
-
1
cm in ATR-IR spectra of CO collected on a model Pt/Al2O3
1
0
catalyst in liquid water at room temperature. The adsorption
of CO on a powdered Pt/Al2O3 catalyst in liquid water at room
-
1
temperature led to a CO(L) band at 2052 cm in ATR-IR
Spectra a and b in Figure 2B correspond to CO adsorbed on
Pt/Al2O3 under conditions similar to spectra a and b in Figure
14
spectra. We attribute the lower frequency observed here to
the presence of low-coordination Pt sites on our catalyst due to
the low calcination and reduction temperature. The ratios of
integrated intensities of the CO(L) bands in spectra a and b to
that in spectrum c in Figure 3A are 0.80 and 0.84, respectively.
The spectra in Figure 3B were collected under conditions similar
to the spectra in Figure 3A, except that the 333 and 1000 ppm
gas mixtures were balanced using He, instead of H2. The ratios
of integrated intensities of the CO(L) bands in spectra a and b
to that in spectrum c in Figure 3B are 0.78 and 0.81,
respectively.
2
1
A (with 0.4 bar H2O vapor at 373 K), except that the 333 and
000 ppm CO gas mixtures are balanced using He, instead of
H2. The CO(L) bands in spectra a and b were observed at 2021
-1
and 2022 cm , respectively, with similar intensity, as compared
-
1
to the frequency of 2011 and 2014 cm observed for the CO-
L) bands in spectra a and b in Figure 2A (in H2-balanced gas
(
mixtures). Drying the Pt/Al2O3 catalyst in 1000 ppm CO/He
for 0.5 h at 373 K leads to an increase in CO(L) band frequency
-
1
to 2035 cm , and a slight increase in integrated intensity of
the CO(L) band. The intensity of the CO(L) band then increased
slightly after the sample was cooled to 303 K in 1000 ppm CO/
He (spectrum d). The ratios of integrated intensities of the CO-
We observed a considerable decrease in the intensity of the
CO(L) band when the catalyst layer in 333 ppm CO/H2 was
subsequently exposed to 333 ppm CO/He in the presence of
(L) bands in spectra a and b to that in spectrum d in Figure 2B
0
.4 bar H2O, as shown in spectra a and b in Figure 4A,
are 0.88 and 0.91, respectively.
respectively. A similar decrease in the intensity of CO(L) band
Results of in situ ATR-IR studies of the WGS reaction at
was observed when the catalyst layer in 333 ppm CO/H was
2
373 K and 2.22 bar in the presence of liquid water are shown
subsequently exposed to 333 ppm CO/He in the presence of
in Figure 3. A linear-bonded CO band was observed at 2007
liquid water, as shown in spectra a and b in Figure 4B,
respectively. To interpret these results, another experiment was
conducted using an in situ transmission IR flow cell with short
path length (∼5 mm) under conditions identical to those in
Figure 4A, and the CO(L) band intensity remained constant as
the gas flow was switched from 333 ppm CO/H2 to 333 ppm
CO/He in the presence of 0.4 bar H2O vapor. Therefore, the
changes in intensity of CO(L) bands observed in Figure 4A and
B are not due to changes in surface CO coverage. These artifacts
-
1
cm when the Pt/Al2O3 catalyst was exposed to 333 ppm CO/
H2 at 373 K in the presence of liquid water, as shown in
spectrum a in Figure 3A. The intensity of the CO(L) band
increased slightly as the CO concentration was increased from
-
1
3
33 to 1000 ppm, and the band shifted to 2011 cm (spectrum
-
1
b). The CO(L) band shifted to 2020 cm and gained intensity
as the Pt/Al2O3 catalyst was then cooled to 303 K in 1000 ppm
CO/H2 in liquid water (spectrum c). Previously, Ferri et al.

Methanol Reforming and Water-Gas Shift Reaction
J. Phys. Chem. B, Vol. 109, No. 7, 2005 2815
Figure 4. (A) ATR-IR spectra of CO adsorbed on Pt/Al
exposed to (a) 333 ppm CO/H (dry basis) and (b) 333 ppm CO/He
dry basis) with 0.4 bar H O in vapor phase at 373 K; (B) ATR-IR
spectra of CO adsorbed on Pt/Al catalyst exposed to (a) 333 ppm
CO/H (dry basis) and (b) 333 ppm CO/He (dry basis) in the presence
of liquid water at 373 K.
2 3
O catalyst
2
(
2
2
O
3
2
of ATR-IR were generally induced by a change in gas
atmosphere at elevated temperatures and, importantly, in the
presence of water, especially liquid water. To minimize these
artifacts, the catalyst layers were saturated with CO by cooling
to 303 K in the same 1000 ppm gas mixtures as at 373 K, instead
of by increasing the CO concentration in the gas mixtures at
373 K.
Reaction Kinetics Studies of Methanol Reforming. Figure
5
A shows the rates of hydrogen production from methanol re-
forming at 423 K versus the system pressure. The reaction order
with respect to system pressure is 0.3 in vapor phase (up to
4
.46 bar) using 2 or 5 wt % methanol feed. In contrast, the
reaction order with respect to system pressure in liquid phase
above 4.88 bar) is highly negative (ca. -6 order) for 2 and 5
(
wt % methanol feeds. A similar inhibiting effect of system
pressure on rate of hydrogen production has been observed for
the liquid-phase reforming of methanol at 498 K and was mainly
attributed to an increase in hydrogen partial pressure as the
Figure 5. (A) The effect of total system pressure on the rate of
hydrogen production from the vapor- and liquid-phase reforming of 2
and 5 wt % methanol at 423 K over a total pressure range of 1.36-
5
.84 bar; (B) the effect of hydrogen partial pressure on the rate of
4
system pressure increases. In short, the gaseous bubbles formed
hydrogen production from the liquid-phase reforming of 2 and 5 wt %
methanol feed at 423 K over a total pressure range of 4.88-5.84 bar;
and (C) the effect of methanol concentration on the rate of hydrogen
production from the vapor and liquid-phase reforming of methanol at
during the liquid-phase reforming of methanol are composed
of the reaction products plus the water and methanol at their
respective vapor pressures. The hydrogen partial pressure in the
catalyst bed is then calculated assuming the total pressure of
the bubbles is equal to the total pressure of the system. Figure
4
23 K over the range of 2-5 wt % at various system pressures.
Figure 5C shows the effect of feed concentration on the rate
5
B shows a plot of the rate of hydrogen production from liquid-
of hydrogen production at various system pressures. For vapor-
phase reforming, the reaction order with respect to methanol
feed concentration is 0.2 at 1.36, 1.77, and 3.22 bar. In contrast,
for liquid-phase reforming, the reaction order with respect to
feed concentration is 0.7 and 0.8 at 5.15 and 5.84 bar,
respectively.
phase methanol reforming versus the calculated hydrogen partial
pressure. The reaction order with respect to the calculated
hydrogen partial pressure is -0.7 for 2 and 5 wt % methanol
feed. We also note that the rate of hydrogen production at 4.46
bar (vapor-phase reforming) is 0.23 min , which is much higher
than the rate of 0.08 min observed at 4.88 bar (liquid-phase
reforming).
-1
-
1
The results of reaction kinetics measurements of methanol
reforming are summarized in Table 2. For the liquid-phase

2816 J. Phys. Chem. B, Vol. 109, No. 7, 2005
He et al.
TABLE 2: Operating Conditions and Results of Reaction Kinetics Studies of the Vapor- and Liquid-Phase Reforming of
Methanol
vapor-phase partial pressures
feed (wt %)
system P (bar)
H
2
TOF (min-1)
conversion (%)
H
2
CO
2
4
CH O
H
2
O
% vaporization of water
2
2
2
2
2
2
2
2
1.36
1.77
3.08
3.91
4.46
4.88
5.15
5.84
0.17
0.18
0.22
0.23
0.23
0.08
0.06
0.03
25.3
28.9
31.2
33.4
31.8
13.4
9.2
0.011
0.016
0.033
0.044
0.050
0.033
0.238
0.749
0.004
0.006
0.011
0.014
0.016
0.012
0.083
0.261
0.012
0.014
0.024
0.029
0.034
0.140
0.140
0.140
1.33
1.74
3.01
3.82
4.37
4.69
4.69
4.69
100
100
100
100
100
59
6
1
4.9
5
5
5
5
5
1.36
1.77
3.22
5.29
5.84
0.20
0.21
0.26
0.11
0.06
11.6
13.2
15.8
6.1
0.013
0.019
0.042
0.242
0.651
0.005
0.007
0.014
0.076
0.218
0.034
0.044
0.077
0.355
0.355
1.30
1.70
3.09
4.61
4.61
100
100
100
11
3.4
2
1
reforming of 2 wt % methanol feed at system pressures of 4.88,
.15, and 5.84 bar, we calculate that 59%, 6%, and 1% of water
in the feed was vaporized, respectively. For the liquid-phase
reforming of 5 wt % methanol at system pressures of 5.29 and
frequency of 2020 cm- observed for the CO(L) band in
spectrum b is most likely due to the higher hydrogen pressure
at the high system pressure of 3.08 bar, as compared to the
5
-
1
frequency of 2032 cm observed in 1000 ppm CO/H2 at 303
K and 1.36 bar. The ratio of integrated intensity of the CO(L)
band in spectrum a to that in spectrum b is 0.40, as compared
to the corresponding ratio of 0.55 observed at the lower system
pressure of 1.36 bar. This result suggests that an increase in
water partial pressure in vapor phase at 423 K at constant
methanol partial pressure leads to a decrease in surface CO
coverage.
Figure 7A and B shows the results of in situ ATR-IR studies
of liquid-phase methanol reforming at 5.84 bar using 2 and 5
wt % methanol feed, respectively. Spectra a, b, and c in Figure
7A were collected over the time periods of 0-20, 20-40, and
40-60 min after the Pt/Al2O3 catalyst layer was exposed to
the 2 wt % methanol feed in liquid phase at 423 K. In these
three spectra, the intensity of the CO(L) band decreased with
respect to the time elapsed as the catalyst layer was exposed to
the liquid feed. Nevertheless, the CO(L) bands were at the same
frequency of 1986 cm . Therefore, the decrease in intensity
of the CO(L) band is unlikely due to a decrease in CO coverage
and is mostly likely due to a change in the catalyst film
morphology (we note that the majority of the catalyst film did
not detach from the ATR crystal), which decreases the amount
of catalyst probed by the evanescent wave generated at the
interface between the ZnSe crystal and the wet catalyst layer.
To estimate the surface coverage of CO in 2 wt % aqueous
methanol solution at 423 K, the catalyst layer needs to be
exposed to a gas mixture containing CO to saturate the surface.
As we have shown in the ATR-IR studies of WGS reaction, a
change in gas atmosphere in the presence of liquid water at
373 K could lead to a change in the CO(L) band intensity. To
minimize these possible artifacts, the catalyst layer was dried
rapidly (<1 min) in flowing H2 and cooled to 303 K, and it
was then saturated with CO by exposing to 1000 ppm CO/H2.
Spectrum d in Figure 7A was collected after the CO-saturated
catalyst layer was exposed to liquid water again. The ratio of
integrated intensity of the CO(L) band in spectrum c to that in
spectrum d is 0.29.
5.84 bar, we calculate that 11% and 2% of water in the feed
was vaporized, respectively.
In situ ATR-IR Studies of Methanol Reforming. Figure
6
A shows ATR-IR spectra collected for the vapor-phase
methanol reforming at 423 K and 1.36 bar. A band at 1986
-
1
cm was observed when the catalyst layer was exposed to
completely vaporized 2 wt % methanol feed (without co-feeding
H2) at 423 K (spectrum a) and is attributed to linear-bonded
CO species formed on Pt via methanol decomposition. The CO
-
1
band then shifted to ∼2010 cm and broadened (leading to
higher integrated intensity) when the catalyst layer was dried
and cooled to 303 K in flowing H2 (spectrum not shown). The
-
1
CO band subsequently shifted to 2032 cm and gained more
intensity after the catalyst layer was exposed to 1000 ppm CO/
H2 at 303 K subsequently (spectrum b). The ratio of integrated
intensity of the CO(L) band in spectrum a to that in spectrum
b is 0.55, which suggests that the surface CO coverage in
completely vaporized 2 wt % methanol feed at 1.36 bar and
-
1
4
23 K is ∼55% of the saturation coverage. The low frequency
-1
observed for the CO band in spectrum a (1986 cm ) is
consistent with the estimated low coverage of CO and with the
presence of 1.3 bar H2O vapor.
Analogous experiments were conducted using 5 wt %
methanol feed to examine the effect of methanol partial pressure,
and the results are shown in Figure 6B. Exposing the catalyst
layer to completely vaporized 5 wt % methanol feed (without
co-feeding H2) at 423 K leads to a linear-bonded CO band at
-
1
1
988 cm (spectrum a). The ratio of integrated intensity of
the CO(L) band in spectrum a to that in spectrum b (collected
for a CO-saturated catalyst at 303 K) is 0.60, as compared to
the corresponding ratio of 0.55 obtained using 2 wt % methanol
feed.
Figure 6C shows ATR-IR spectra collected for the vapor-
phase methanol reforming at 3.08 bar using 0.9 wt % methanol
feed. The methanol partial pressure in vapor phase is the same
as in the experiment conducted at 1.36 bar using 2 wt %
methanol feed (Figure 6A), whereas the water partial pressure
is considerably higher. The linear-bonded CO band was
Spectra a-d in Figure 7B were collected under conditions
similar to the spectra in Figure 7A, except that the methanol
feed concentration was increased to 5 wt %. The linear-bonded
CO band is observed at the frequency of 1990 cm , as
compared to the frequency of 1986 cm observed in 2 wt %
methanol feed, and the intensity of this band also decreased
with respect to the time elapsed as the catalyst layer was exposed
to the liquid feed. The ratio of integrated intensity of the CO-
-
1
observed at 1980 cm in completely vaporized 0.9 wt %
methanol feed (spectrum a), as compared to the frequency of
-
1
-
1
-1
1
986 cm observed at the lower system pressure of 1.36 bar.
-
1
The CO band then shifted to 2020 cm and increased in
intensity when the catalyst layer was subsequently exposed to
1000 ppm CO/H2 at 303 K and 3.08 bar (spectrum b). The lower

Methanol Reforming and Water-Gas Shift Reaction
J. Phys. Chem. B, Vol. 109, No. 7, 2005 2817
2 3
Figure 7. (A) ATR-IR spectra of CO formed on Pt/Al O as the catalyst
was exposed to 2 wt % aqueous methanol feed solution at 423 K over
the periods of (a) 0-20, (b) 20-40, and (c) 40-60 min, and of CO
(
d) adsorbed on Pt/Al
ppm CO/H
spectra of CO formed on Pt/Al
aqueous methanol feed solution at 423 K over the periods of (a)
-20, (b) 20-40, and (c) 40-60 min, and of CO (d) adsorbed on
Pt/Al as the catalyst was then exposed to 1000 ppm CO/H and
liquid water at 303 K at 5.84 bar.
2 3
O as the catalyst was then exposed to 1000
2
and liquid water at 303 K at 5.84 bar; and (B) ATR-IR
as the catalyst was exposed to 5 wt
2 3
O
%
0
O
2 3
2
to the saturation coverage (∼90%). Importantly, the CO(L) band
-
1
observed in 1000 ppm CO/H2 is at a frequency of 2014 cm ,
-
1
which is 8 cm lower than the CO(L) band observed in the
absence of co-feeding H2 (in 1000 ppm CO/He). Previous
studies have shown that coadsorbing H2 with CO on Pt can
lead to a small decrease of CO stretching frequency by 5-10
-
1 42,43
cm .
Therefore, the lower frequency of CO(L) observed
Figure 6. (A) ATR-IR spectra of CO (a) formed on Pt/Al
catalyst was exposed to completely vaporized 2 wt % methanol feed
at 423 K and (b) adsorbed on Pt/Al as the catalyst was then exposed
to 1000 ppm CO/H at 303 K at 1.36 bar; (B) ATR-IR spectra of CO
a) formed on Pt/Al as the catalyst was exposed to completely
vaporized 5 wt % methanol feed at 423 K and (b) adsorbed on Pt/
Al as the catalyst was then exposed to 1000 ppm CO/H at 303 K
at 1.36 bar; and (C) ATR-IR spectra of CO (a) formed on Pt/Al as
the catalyst was then exposed to completely vaporized 0.9 wt %
methanol feed at 423 K and (b) adsorbed on Pt/Al as the catalyst
was exposed to 1000 ppm CO/H at 303 K at 3.08 bar.
2
O
3
as the
in the presence of H2 gas suggests that the Pt surface has a
higher coverage by adsorbed H atoms. A kinetically significant
step of the WGS reaction on Pt is the dissociation of H2O to
give adsorbed OH and H species, and the reaction proceeds
subsequently by reaction of OH species with CO, based on
predictions from DFT calculations. An increase in the surface
coverage by H atoms is unfavorable for H2O dissociation on
Pt, because H atoms can block surface sites for water dissocia-
tion and H atoms can combine with OH species to convert these
species back to water. Both of these effects would lead to a
decrease in the rate of the WGS reaction.
Results of reaction kinetics studies of the vapor-phase WGS
reaction at 373 K show that the CO partial pressure has a
negligible effect on reaction rate in the presence of 1.2 bar H2.
In situ ATR-IR studies carried out using 333 and 1000 ppm
CO/H2 (with 0.4 bar H2O at 373 K) show that the intensity of
the CO(L) band increases slightly with an increase in CO
concentration, and the surface CO coverage is close to saturation
in both CO gas mixtures. Increasing CO partial pressure on a
Pt surface with high CO coverage leads to a more poisoned
surface, and hence to a lower reaction rate in the presence of
1.2 bar H2. A negative order with respect to CO equal to -0.21
2 3
O
2
(
2 3
O
20
O
2 3
2
2 3
O
2
O
3
2
(L) band in spectrum c to that in spectrum d (collected for a
CO-saturated catalyst in liquid water) is 0.40.
Discussion
Water-Gas Shift Reaction. Results of reaction kinetics
studies of the vapor-phase WGS reaction at 373 K show that
H2 has an inhibiting effect on the reaction rate; that is, the
reaction order with respect to hydrogen is -0.22. In situ ATR-
IR studies carried out under conditions similar to reaction
kinetics studies (with 0.4 bar H2O at 373 K) suggest the surface
CO coverages in 1000 ppm CO/H2 and CO/He are both close

2
818 J. Phys. Chem. B, Vol. 109, No. 7, 2005
He et al.
has been reported on Pt/Al2O3 at 543 K.37 On the other hand,
as the surface coverage by CO becomes high (at the low
temperatures of the present study), then increasing the CO
pressure has a small effect on the CO coverage because the
differential heat of CO adsorption becomes a strong function
of coverage. Therefore, the negligible effect of CO on the
reaction rate observed in 1.2 bar H2 can be attributed to the
fact that the Pt surface has become essentially saturated with
adsorbed CO at a gas-phase concentration of 333 ppm.
Methanol Reforming. The low order (0.2) with respect to
methanol partial pressure observed for vapor-phase methanol
reforming at 423 K cannot be attributed to a high methanol
coverage on the Pt surface under reaction conditions, because
18
methanol adsorbs weakly on Pt. In situ ATR-IR studies carried
out under similar conditions at 1.36 bar and 423 K show that
the surface coverage by adsorbed CO increases from ∼55% to
∼60% of the saturation coverage as the methanol feed concen-
tration is increased from 2 to 5 wt %. This observation suggests
the Pt surface becomes more poisoned with CO as the methanol
partial pressure increases in vapor phase. Therefore, the positive
effect of methanol partial pressure on reaction rate is reduced
by a decrease in the number of vacant sites available for
methanol to adsorb and decompose, which leads to the apparent
order of 0.2 with respect to methanol partial pressure in vapor
phase.
The rate of the WGS reaction in vapor phase is fractional
order (0.55) with respect to H2O partial pressure. The positive
effect of H2O can be attributed to increased OH coverage
through dissociation of H2O. A fractional order is typically
observed for a reactant when the surface coverage of the
adsorbed reactant becomes high, which is not consistent with
46
the fact that adsorption of H2O on Pt is weak. Results from
DFT calculations predict that the dissociation of H2O on Pt-
The above interpretation is also supported by reaction kinetics
measurements and microkinetic modeling carried out for the
(111) has a high activation energy barrier of 142 kJ/mol, and
48
the subsequent step involving reaction between CO and OH
decomposition of methanol on Pt/SiO2 at 473 K. The reaction
order with respect to methanol partial pressure was determined
to be 0.16 by reaction kinetics measurements. In agreement with
this result, microkinetic modeling predicts that the order is equal
to 0.28 with respect to methanol partial pressure as the methanol
concentration is increased from 1 to 10 wt %, which is mainly
due to a decrease in the number of vacant sites as the result of
an increase in the CO surface coverage.
4
7
species has a lower activation energy barrier of 86 kJ/mol.
Accordingly, the activation energy barrier for reaction of OH
species with adsorbed CO is similar to the barrier for reaction
of OH with adsorbed H atoms (89 kJ/mol);47 therefore, the H2O
dissociation step may not be quasi-equilibrated. Consequently,
the surface OH coverage may not be directly proportional to
the surface coverage of H2O, which is directly proportional to
the partial pressure due to the weak adsorption of H2O on Pt.
Thus, the nonlinear relation between H2O partial pressure and
surface OH coverage may lead to the fractional reaction order
with respect to H2O.
The low order (0.3) with respect to system pressure observed
for vapor-phase methanol reforming can be attributed to the
effect of increasing methanol, water, and hydrogen partial
pressures inside the reactor for vapor-phase methanol reforming,
as shown in Table 2. We have shown that methanol partial
pressure has a positive effect on the rate of hydrogen production.
Reaction kinetics measurements of the WGS reaction in vapor
phase at 373 K show that water partial pressure has a positive
effect on the WGS reaction rate. In this respect, results of in-
situ ATR-IR studies at 1.36 and 3.08 bar in completely
vaporized methanol feeds show that the surface coverage by
adsorbed CO decreases from 55% to 40% of the saturation
coverage as the H2O partial pressure increases (at constant
methanol partial pressure). This observation suggests that an
increase in H2O partial pressure leads to higher rate of CO
removal from Pt surface via the WGS reaction. Adsorbed H
atoms can have a negative effect on the rate of methanol
decomposition by blocking surface sites that are not poisoned
with CO. Reaction kinetics measurements of the WGS reaction
in vapor phase at 373 K show that hydrogen has an inhibiting
effect on the WGS reaction rate. Therefore, the positive effects
of methanol and water partial pressures on the rate of hydrogen
production from vapor-phase methanol reforming are reduced
by the negative effect of hydrogen partial pressure, which leads
to the apparent order of 0.3 with respect to system pressure.
In the presence of liquid water, in situ ATR-IR studies of
the WGS reaction on the Pt/Al2O3 catalyst at 373 K show that
the Pt surface is still dominated by adsorbed CO in 333 and
1
000 ppm CO in either H2 or He, based on the ratios of
integrated intensities of CO bands observed at 373 K to those
of CO bands observed in 1000 ppm CO at 303 K, that is, about
0.8. We note that the corresponding ratios in vapor-phase ATR-
IR studies are about 0.9, which suggests that the presence of
liquid water may decrease slightly the surface CO coverage
under the current WGS reaction conditions at 373 K. The lower
CO coverage for WGS reaction in the presence of liquid water
could be due to a slightly weaker binding of CO to the Pt surface
caused by displacement from the surface of weakly adsorbed
water molecules. In addition, the lower CO coverage for WGS
reaction in the presence of liquid water could be attributed to
the presence of other adsorbed species, which are able to
compete with CO for adsorption sites at high CO coverage.
Results from DFT calculations suggest that the presence of liquid
water may decrease the activation energy barrier of H2O
dissociation on Pt(111) from 142 to 75 kJ/mol.47 Furthermore,
the corresponding barrier could be even lower on low coordina-
tion Pt sites, which are predominant on our catalyst based on
the low frequency of CO adsorbed under dry conditions.
Therefore, the lower coverage of CO observed in the presence
of liquid water could be caused by a higher coverage of OH
species. Importantly, based on the above arguments, it appears
that the rate of the WGS reaction in the presence of liquid water
is comparable to the rate under complete vaporization conditions
when other factors (such as partial pressure of CO) are kept
constant. The rate of WGS reaction at 423 K with 2.5 mbar of
In contrast to the low order with respect to system pressure
observed in vapor-phase methanol reforming, reaction kinetics
measurements of liquid-phase methanol reforming show that
system pressure has a strong inhibiting effect on the rate of
hydrogen production. An increase in system pressure leads to
an increase in hydrogen partial pressure inside the gas bubbles,
which most likely inhibits methanol reforming in liquid phase
by blocking surface sites with adsorbed hydrogen atoms.
The rate of hydrogen production from vapor-phase methanol
reforming at 4.46 bar is much higher than that from liquid-
-1
CO in vapor phase was found to be 0.38 min when only 37%
of the water feed was vaporized, which is comparable to the
-1
phase methanol reforming at 4.88 bar (0.23 versus 0.08 min ).
The inhibiting effect of hydrogen observed for liquid-phase
methanol reforming cannot account for this observation, because
-
1
value of 0.36 min measured when the water feed was
completely vaporized.

Methanol Reforming and Water-Gas Shift Reaction
J. Phys. Chem. B, Vol. 109, No. 7, 2005 2819
the hydrogen partial pressures are similar between the two
conditions (0.05 versus 0.03 bar, respectively). The rate of the
WGS reaction at 423 K in the presence of liquid water is
comparable to when the water is just fully vaporized, as shown
in the WGS reaction part of this discussion. Importantly, results
from in situ ATR-IR studies of methanol reforming show that
the surface coverage by adsorbed CO in 2 wt % aqueous
methanol solution is only ∼29% of the saturation coverage, as
compared to the value of 55% observed in completely vaporized
positive effect of H2O on the WGS reaction rate is most likely
due to an increase in the surface coverage by adsorbed OH
species. The results of DFT calculations carried out for the WGS
on Pt(111), and the fractional order with respect to H2O partial
pressure, suggest that the dissociation of H2O on Pt is not quasi-
equilibrated for the WGS reaction in vapor phase.
In situ ATR-IR studies of the WGS reaction at 373 K in the
presence of liquid water show that the Pt surface is still
dominated by adsorbed CO in gas mixtures containing 333 and
2
wt % methanol feed, which suggests that the methanol
1
000 ppm CO balanced with H2 or He; however, the surface
decomposition step is more kinetically limiting in liquid phase
than in vapor phase. The difference of ∼3 times between the
reaction rate in vapor phase and in liquid phase at 423 K
corresponds to an increase of only ∼4 kJ/mol in the activation
energy barrier that is involved in the rate-limiting step of
methanol decomposition. The decomposition of methanol on
Pt(111) is predicted to occur mainly through a pathway
involving initial C-H bond activation by DFT calculations, and
the highest activation energy barrier for this pathway is the initial
coverage by adsorbed CO is slightly lower (∼80% of the
saturation coverage) than that under complete vaporization
conditions. The lower CO coverage for WGS reaction in the
presence of liquid water could be due to a slightly weaker
binding of CO to the Pt surface caused by displacement from
the surface of weakly adsorbed water molecules, or it could be
caused by a higher coverage of OH species on low coordination
Pt sites as a result of faster dissociation of H2O on the surface
facilitated by liquid water. The rate of WGS reaction in the
presence of liquid water is comparable to the rate under complete
vaporization conditions when other factors (such as the CO
partial pressure) are held constant.
18-20
C-H bond activation (65 kJ/mol) leading to CH2OH species.
Under liquid-phase reaction conditions for methanol decomposi-
tion, the methanol molecules are present in hydrogen-bonded
clusters with water molecules,49,50 and the surface Pt sites that
would be vacant in vapor phase are in contact with physisorbed
water molecules present in the liquid phase. These differences
from vapor-phase conditions could increase the activation energy
barrier for the initial C-H bond activation, leading to a lower
rate of methanol decomposition in the liquid phase.
Reaction kinetics studies were carried out for the reforming
of methanol in vapor and liquid phase at 423 K over a total
pressure range of 1.36-5.84 bar. The reaction order with respect
to methanol partial pressure is 0.2 for vapor-phase methanol
reforming, and it is near first order (0.7, 0.8) for liquid-phase
methanol reforming. System pressure has a positive effect on
the rate of hydrogen production from vapor-phase methanol re-
forming (order of 0.3), and it has a strong inhibiting effect on
the rate of hydrogen production from liquid-phase methanol re-
forming. The rate of hydrogen production from vapor-phase
methanol reforming at 4.46 bar is much faster than that from
liquid-phase methanol reforming at 4.88 bar (0.23 versus 0.08
Results from reaction kinetics measurements of liquid-phase
methanol reforming at 423 K show that the rate of hydrogen
production is near first order with respect to methanol concen-
tration as the feed concentration increases from 2 to 5 wt %, as
compared to the corresponding low order of 0.2 observed in
vapor-phase methanol reforming. In situ ATR-IR studies of
liquid-phase methanol reforming show that the surface CO
coverage is only ∼29% of the saturation coverage in 2 wt %
aqueous methanol feed, and the coverage increases to ∼40%
of the saturation coverage as the methanol feed concentration
is increased to 5 wt %. The corresponding CO coverages in
completely vaporized 2 and 5 wt % methanol feed are 55%
and 60% of saturation coverage (at 1.36 bar), respectively. These
observations suggest that the Pt surface is less significantly
covered by adsorbed CO for liquid-phase methanol reforming,
leading to a higher reaction order with respect to methanol
concentration for the reforming of methanol in liquid phase than
in vapor phase.
-1
min ).
In situ ATR-IR studies were conducted at 423 K to estimate
the surface coverage by adsorbed CO on the Pt/Al2O3 catalyst
in completely vaporized methanol feeds and in aqueous
methanol feed solutions. The surface coverages by adsorbed
CO are estimated to be 55% and 60% of the saturation coverage
in completely vaporized 2 and 5 wt % methanol feed at 1.36
bar, respectively. The surface coverage by CO decreased from
5
5% to 40% of the saturation coverage as the H2O partial
pressure increases at constant methanol partial pressure (in
completely vaporized 0.9 wt % methanol feed at 3.08 bar). In
contrast, the surface coverages by adsorbed CO are estimated
to be 29% and 40% of the saturation coverage in 2 and 5 wt %
aqueous methanol solution at 5.84 bar, respectively. To our
knowledge, the present study is the first application of in situ
ATR-IR spectroscopy to study a catalytic solid-liquid interface
at the elevated temperature of 423 K and the pressure of 5.84
bar.
Conclusions
Reaction kinetics studies of the WGS reaction were carried
out at 373 K in vapor phase, giving reaction orders with respect
to CO, H2, and H2O partial pressures equal to 0.02, -0.22, and
0
.55, respectively. In situ ATR-IR studies carried out in vapor
phase under similar conditions (in 333 and 1000 ppm CO/H2
with 0.4 bar H2O) suggest that the Pt surface coverage by
adsorbed CO is high (∼90% of the saturation coverage) under
reaction conditions. ATR-IR spectra collected in gas mixtures
containing 1000 ppm CO in H2 or He, with 0.4 bar H2O in
vapor phase, indicate that the surface coverage by adsorbed CO
is similar in both gases (near the saturation coverage), and the
surface coverage by adsorbed H atoms is higher in the
H2-balanced CO gas mixture. The negative reaction order with
respect to H2 partial pressure is attributed to the inhibiting effect
of adsorbed H atoms on the dissociation of H2O, which is a
kinetically significant step in the WGS reaction on Pt. The
The decomposition of methanol is found to be slower during
the reforming of methanol in liquid phase than in vapor phase
(and hence the lower rate of hydrogen production), which could
be due to a higher activation energy barrier for the initial C-H
bond activation in the presence of aqueous methanol solution.
The low reaction order with respect to methanol partial pressure
observed for vapor-phase methanol reforming is caused by a
decrease in the number of vacant sites on the Pt surface as a
result of the increased extent of CO poisoning when the
methanol partial pressure is increased. The surface coverage by
CO during the reforming of methanol in liquid phase is lower
than that in vapor phase; therefore, the number of vacant sites

2820 J. Phys. Chem. B, Vol. 109, No. 7, 2005
He et al.
on the Pt surface, available for interaction with methanol, is
not as strongly affected by CO poisoning as the methanol
concentration is increased in liquid phase, which leads to the
near first-order reaction with respect to methanol concentration
observed for liquid-phase methanol reforming. The positive
effect of system pressure on the rate of hydrogen production
observed for vapor-phase methanol reforming is due to the
positive effects of increasing the methanol and water partial
pressures, combined with the negative effect of increasing H2
partial pressure. The inhibiting effect of system pressure on the
rate of hydrogen production from liquid-phase methanol re-
forming is mainly caused by the negative effect of adsorbed
hydrogen, which blocks the surface sites that are not poisoned
by CO.
3
Acknowledgment. We wish to acknowledge funding from
National Science Foundation (NSF), Department of Energy
(DOE), and ConocoPhillips.
2
References and Notes
1) Cortright, R. D.; Davda, R. R.; Dumesic, J. A. Nature 2002, 418,
(34) Park, S.; Xie, Y.; Weaver, M. J. Langmuir 2002, 18, 5792.
(35) Spieker, W. A.; Regalbuto, J. R. Chem. Eng. Sci. 2001, 56, 3491.
(36) Regalbuto, J. R.; Navada, A.; Shadid, S.; Bricker, M. L.; Chen, Q.
9
J. Catal. 1999, 184, 335.
3) Davda, R. R.; Dumesic, J. A. Chem. Commun. 2004, 36.
(37) Grenoble, D. C.; Estadt, M. M.; Ollis, D. F. J. Catal. 1981, 67, 90.
(38) Lam, C. W.; Stacey, M. S.; Trimm, D. L. Chem. Eng. Sci. 1981,
36, 224.
(39) deMenorval, L. C.; Chaqroune, A.; Coq, B.; Figueras, F. J. Chem.
Soc., Faraday Trans. 1997, 93, 3715.
(40) Kappers, M. J.; Vandermaas, J. H. Catal. Lett. 1991, 10, 365.
(41) Greenler, R. G.; Burch, K. D.; Kretzschmar, K.; Klauser, R.;
Bradshaw, A. M.; Hayden, B. E. Surf. Sci. 1985, 152, 338.
(42) Eischens, R. P. Acc. Chem. Res. 1972, 5, 74.
(43) Vannice, M. A.; Twu, C. C.; Moon, S. H. J. Catal. 1983, 79, 70.
(44) Bourane, A.; Dulaurent, O.; Bianchi, D. Langmuir 2001, 17, 5496.
(45) Primet, M.; Basset, J. M.; Mathieu, M. V.; Prettre, M. J. Catal.
1973, 29, 213.
1
5
5
(
(
10) Ferri, D.; Burgi, T.; Baiker, A. J. Phys. Chem. B 2001, 105, 3187.
11) Ferri, D.; Burgi, T.; Baiker, A. Phys. Chem. Chem. Phys. 2002, 4,
(46) Wagner, F. T.; Moylan, T. E.; Schmieg, S. J. Surf. Sci. 1988, 195,
403.
(47) Desai, S.; Neurock, M. Electrochim. Acta 2003, 48, 3759.
(48) Gokhale, A. A.; Kandoi, S.; Greeley, J. P.; Mavrikakis, M.;
Dumesic, J. A. Chem. Eng. Sci. 2004, 59, 4679.
(49) Wei, S.; Shi, Z.; Castleman, A. W. J. Chem. Phys. 1991, 94, 3268.
(50) Wakisaka, A.; Abdoul-Carime, H.; Yamamoto, Y.; Kiyozumi, Y.
J. Chem. Soc., Faraday Trans. 1998, 94, 369.
2
667.
(
(
(
(
(
12) Ferri, D.; Burgi, T.; Baiker, A. J. Catal. 2002, 210, 160.
13) Burgi, T.; Baiker, A. J. Phys. Chem. B 2002, 106, 10649.
14) Ortiz-Hernandez, I.; Williams, C. T. Langmuir 2003, 19, 2956.
15) Sexton, B. A. Surf. Sci. 1981, 102, 271.
16) Gibson, K. D.; Dubois, L. H. Surf. Sci. 1990, 233, 59.