Quantitative determination of the catalytic activity of bulk metal oxides for formic acid oxidation

Article abstract of DOI:10.1006/jcat.2002.3683

The dissociative chemisorption of formic acid to surface formate species on metal oxide catalysts was studied. Comparison of the number of active surface sites for bulk metal oxides yielded similar adsorption surface site densities for the majority of samples studied, ～ 5-6 μmole HCOO_ads/sq m. Kinetic studies of formic acid oxidation made possible the determination of the apparent activation energies for the various metal oxide catalysts for formic acid oxidation. The turnover frequencies (TOF) values for the various metal oxide catalysts studied varied over 11 orders of magnitude. A strong inverse relationship was observed between the TOF values and the decomposition temperatures of the surface formate intermediates, indicating that the thermal stability of the surface formate intermediate is the only significant kinetic factor. Thus, dissociative adsorption of the formic acid molecule is not kinetically significant for kinetic activity during formic acid oxidation over the metal oxide catalysts. The surfaces of the metal oxide catalysts studied were also oxidized, analogous to in situ Raman observation made with V₂O₅ and MoO₃ during methanol oxidation, since methanol and formic acid both lose two hydrogen atoms during oxidation to H₂CO and CO₂, respectively.

Full text of DOI:10.1006/jcat.2002.3683

Journal of Catalysis 210, 241–254 (2002)
doi:10.1006/jcat.2002.3683
Quantitative Determination of the Catalytic Activity of Bulk Metal
Oxides for Formic Acid Oxidation
David E. Fein and Israel E. Wachs1
Zettelmoyer Center for Surface Studies, Department of Chemical Engineering, Lehigh University, Bethlehem, Pennsylvania 18015
Received June 5, 2001; revised May 15, 2002; accepted May 21, 2002
also been used extensively to determine the structure of
the molecular or dissociatively adsorbed species with re-
spect to coordination with the surface atoms (1–6). The
most reliable method would, however, involve use of the
actual reactant molecule with complementary chemisorp-
tion information obtained in the temperature region of the
reaction (7). Additionally, the chemisorbed surface inter-
mediate should be common on all of the catalytic materials
and care should be taken to ensure that the surfaces of the
catalysts are uniform in pretreatment as well as composi-
tion. Many researchers selected chemical probe molecules
that were utilized to determine reaction rates and give
insights into the quantity and nature of the active sites
(1–14). For example, Ouyang et al. studied the chemisorp-
tion of pyridine, carbon dioxide, and formic acid to deter-
mine the acid–base properties of ZrO2 (10). The decom-
position of chemical probe molecules from the surface has
also received wide spread attention due to its relationship to
the adsorbed species (4, 8). While certain probe molecules
can give insight into the acidic, basic, or redox nature of the
surface active sites, the surface reactions should be identi-
cal on all samples for a truly unambiguous comparison of
catalytic surfaces.
Formic acid (HCOOH) chemisorption satisfies all of the
above criteria for appropriate selection of a probe molecule
for catalytic comparisons. This molecule has been exten-
sively employed in the past due to the commonality of the
chemisorbed formate intermediate on nearly every cata-
lytic surface. Furthermore, the surface formate intermedi-
ate is a very common intermediate in the water–gas shift re-
actions on the surface of metal oxides such as MgO, Al2O3,
and ZnO (6).
The dissociative chemisorption of formic acid, HCOOH, to sur-
face formate species, HCOO-M, on metal oxide catalysts was exam-
ined. The chemisorption studies reveal an average of ∼5–6 µmol/m2
of active surface sites for many of the metal oxide catalysts. From
TPD experiments, the decomposition temperatures of the surface
formate (Tp) were found to vary from 88 to 313◦C. The turnover
frequency (TOF) values for formic acid oxidation were found to
vary over 11 orders of magnitude (10−3–108 −1) at 250◦C. No
s
correlations were found between (i) H2-TPR onset temperatures
and the enthalpies of formation of bulk metal oxides (−ꢀHf),
(ii) TOFs and H2-TPR onset temperatures, or (iii) TOFs and the iso-
topic dioxygen exchange rates with the metal oxide lattice oxygen.
Weak inverse correlations were found for both TOFs and catalytic
activities versus −ꢀHf. A strong inverse relationship was found be-
tween the TOFs and the surface formate decomposition tempera-
tures, supportingtheassertionthattherate-determiningstepduring
formic acid oxidation is the decomposition of the surface formate
c
intermediate. ꢀ 2002 Elsevier Science (USA)
Key Words: formic acid; chemisorption; oxidation; metal oxide
catalysts; TOF; TPD.
1. INTRODUCTION
The ability to quantitatively characterize surface cata-
lyzed reactions is essential for understanding the activity
of a wide range of catalytic materials. The importance of
determining trends in reactivity for surface reaction inter-
mediates on various catalysts is evident from the countless
studies that have appeared on the development of a uni-
versal method for this purpose. From a review of the catal-
ysis literature, it can be concluded that quantification of
the number of active surface sites and turnover frequencies
(TOFs: number of molecules converted per active surface
site per second) and a measurement of surface reaction
intermediate stability are required for the complete char-
acterization of a catalytic material.
As a vapor, formic acid will tend toward the trans-
configuration due to its ability to dimerize; however, the
dissociative adsorption is unimolecular and has a similar
bonding structure on most oxide surfaces. Three surface
bonding structures have been found for the surface formate
intermediate: monodentate (Fig. 1a), bidentate (Fig. 1b),
and surface bridged (Fig. 1c) configurations. It has been
reported that under continuous flow conditions for NiO,
the bidentate or bridged structures are formed to minimize
surface energy (9). Similar findings were reported for the
Several approaches have been previously proposed for
quantifying the number of active sites on surfaces with var-
ious chemical probe molecules. Adsorption studies have
1 To whom correspondence should be addressed. Fax: (610) 758-5057.
E-mail: iew0@lehigh.edu.
241
0021-9517/02 $35.00
c
ꢀ 2002 Elsevier Science (USA)
All rights reserved.

242
FEIN AND WACHS
HCOOH(g)
H
HCOO(a) + H(a)
products are formed for the decomposition or oxidation
of formic acid (12). Therefore, TPO and TPD spectra for
CO2/HCOOH were shown to be almost identical in loca-
tion, although much less CO and H2 were formed during the
temperature-programmed oxidation studies (12). Thus, it
was concluded that the oxygen from the oxide surface itself
reacts to form products, creating oxygen vacancies, which
are then replaced by oxygen from the reactant stream via
the Mars–van Krevelen mechanism. Based on these find-
ings, it is possible to utilize the wealth of information de-
scribing surface formate decomposition temperatures from
oxide surfaces to verify oxidation decomposition tempera-
tures during formic acid oxidation experiments.
H
C
O
C
O
H
O O
H
a) Monodentate
b) Bidentate
H
C
Traditionally, the focus has been to quantify the rate-
determining step of surface reactions of chemical probe
molecules. In the case of low conversion (<10%), the
absence of diffusional effects makes it possible to study
only the rate of the surface reaction without heat and mass
transfer limitations (7). For surface formate intermedi-
ates, the rate-determining step for decomposition from
the surface is the breaking of the C–H bond; therefore
the decomposition temperature will be determined and
employed as a measure of the intermediate stability and
bonding strength to the surface. Additionally in the present
study, bulk metal oxide samples were investigated due to
their relatively high surface areas compared to that of pure
metals. Powdered samples of the bulk metal oxides were
employed in the present investigation, which represent
an average of all crystal faces and most surely possess
defects that are representative of samples used in practical
applications as catalysts or supports.
In an attempt to illustrate Sabatier’s principle of an ide-
ally moderate intermediate stability, it has been reported
that so-called “volcano curves” result from diagrams with
a measure of the rate of a catalytic reaction along one
axis and a measure of the stability of the putative inter-
mediate compound on the other axis (13). In previous
studies, the surface reaction rates, which are characteristic
of the surface activity of a particular catalyst for a given
reaction, were correlated with a bulk material property
that was related to the stability of the intermediate com-
pound (14). Thus, the objective of the present investiga-
tion is to consider only surface properties when making
such correlations, since bulk material properties are not
intrinsically related to the stability of the surface reaction
intermediate.
O O H
c) Bridged
FIG. 1. Schematic of dissociative adsorption of formic acid on cata-
lytic surfaces.
adsorption of formic acid onto the surface of ZrO2 pow-
der, where the bidentate surface formate species was iden-
tified (10), and for single-crystal MgO(001) and TiO2(101),
where the bridging and bidentate surface structures, re-
spectively, were theoretically calculated to be more sta-
ble than the monodentate configuration (3, 5–6). However,
analysis of the surface formate configuration from single-
crystal metal oxide studies have revealed all possible con-
figurations. Therefore, metal oxide powders would most
likely exhibit an average of all configurations, depending on
the most favorable disposition on a given crystal face. Al-
though adsorption temperature can vary greatly for many
compounds, it has been reported that formic acid is ad-
sorbed on most metal and metal oxides surfaces dissocia-
tively at room temperature to yield surface formate and
surface hydroxyl species (6, 11). This same result has been
confirmed by inelastic electron tunneling spectroscopic
(IETS) studies for formic acid adsorption on the surface of
Al2O3 (1).
The oxidation products of this simple carboxylic acid are
identical on every surface since complete oxidation prod-
ucts, CO2 and H2O, are obtained. Despite single-crystal
studies on MgO(100) and SnO2(110), which have shown
CO as a major product under ultrahigh-vacuum conditions
(34), it has been found that the TOF values for CO oxida-
tion on many metal oxides is typically ∼10−5 s−1 at higher
pressures (35). Any CO2 observed is therefore a result of
the adsorbed surface reaction, since the conversion of CO
to CO2 is too inefficient to significantly impact the results
2. EXPERIMENTAL
2.1. Catalyst Synthesis
The bulk metal oxide catalysts employed in the present
at higher pressures. In addition, the presence of diatomic investigation were either purchased as high-purity commer-
oxygen in the gaseous reactant stream has been shown cial chemicals or prepared by decomposition of their corre-
to exhibit very little effect on the temperature at which sponding hydroxides or metal salts, as shown in Table 1. The

QUANTITATIVE DETERMINATION OF CATALYTIC ACTIVITY
TABLE 1
243
2.4. Formic Acid Chemisorption
Bulk Metal Oxide Catalyst Preparation
The experimental conditions for quantifying the num-
ber of active surface sites via formic acid chemisorption
were determined over a wide range of temperatures and
formic acid partial pressures in a Cahn TGA microbalance
(Model TG-131) coupled to a PC (Zenith Data Systems
Model CW-282-82) for temperature and weight monitoring.
A detailed flow diagram can be found in a prior publication
(16). The system allowed for a controlled flow of high-purity
gases: air for pretreatment, a mixture of formic acid and he-
lium for chemisorption experiments, and pure helium for
temperature-programmed decomposition experiments.
The following experimental procedure was employed,
which accounts for all measurements associated with the
TGA microbalance. After being weighed and loaded into
the TGA, the oxide catalysts were heated in situ to 350◦C
for 1 h in flowing air (9.5 mL/min; ultrahigh purity; Air
Gas) and helium (80 mL/min; ultrahigh purity; Air Gas)
in order to remove adsorbed moisture and possible car-
bonaceous residues. The pretreated catalysts were then
cooled in flowing helium to 100◦C. Formic acid was quan-
titatively adsorbed onto the catalyst at 100◦C from a flow-
ing HCOOH/He stream (2000 ppm HCOOH; Scott Gas)
for 1 h to obtain the weight gain. Previous surface formate
adsorption studies on metal oxide surfaces (MoO3) have
shown that upon adsorption, H2O formed from the com-
bination of the chemisorbed hydrogen and oxygen atoms
(36). The water formed is expelled from the surface due
to water’s weak interaction with the metal oxide surface.
The net weight gain observed was subsequently corrected
for loss of H2O during HCOOH adsorption since one H2O
molecule is displaced from the surface during the dissocia-
tive adsorption of every two surface formate intermediates
(16). The decomposition temperature of the surface for-
mate intermediate was obtained by ramping the temper-
ature at a constant rate of 10◦C/min and monitoring the
derivative of the weight loss.
Catalyst
Source
MgO
CaO
SrO
MgCO3, calcined at 350◦C for 3 h, Aldrich
CaCO3, calcined at 350◦C for 3 h, Aldrich
Alfa Aesar, 99.5%
BaO
Ba(ClO4)2 · 3H2O, calcined at 400◦C for 3 h, Aldrich
Y2O3
La2O3
TiO2
Alfa Aesar, 99.999%
Alfa Aesar, 99.999%
Degussa P-25
Degussa
Alfa Aesar, 99.9%
ZrO2
HfO2
CeO2
V2O5
Nb2O5
Ta2O5
Cr2O3
MoO3
WO3
Mn2O3
Fe2O3
Co3O4
Rh2O3
NiO
Engelhard
NH4VO3, calcined at 450◦C for 3 h, Alfa Aesar
Niobium Products Company
H. C. Starck Company
Alfa Aesar, 99.997%
(NH4)Mo7O24 · 4H2O, calcined at 300◦C for 5 h, Alfa Aesar
H2WO4, calcined at 400◦C for 48 h, Aldrich
Mn(OOCCH3)2, calcined at 250◦C overnight, Alfa Aesar
Alfa Aesar, 99+%
Aldrich
Rh(NO3)3, calcined at 300◦C for 4 h, Johnson Mathey
Ni(OH)2, calcined at 230◦C overnight, Alfa Aesar
Pd(NO3)2, calcined at 300◦C for 4 h, Johnson Mathey
Pt(NH3)4Cl2, calcined at 400◦C for 4 h, Johnson Mathey
Cu(NO3)2 · 3H2O, calcined at 200◦C for 48 h,
Johnson Mathey
PdO
PtO
CuO
Ag2O
Au2O3
ZnO
Al2O3
Ga2O3
In2O3
SiO2
Alfa Aesar, 99%
Au(OH)2, calcined at 200◦C for 48 h
Aldrich, 99.97%
Engelhard
Alfa Aesar, 99.999%
Alfa Aesar, 99.997%
Carbosil EH-5
SnO2
Bi2O3
Aldrich, 99%
Bi2(C2O4)3, calcined at 400◦C overnight, Alfa Aesar
pretreatment conditions of the precursors were obtained
from the Handbook of Chemistry and Physics (15).
The adsorption temperature for formic acid was 100◦C
since it was above the desorption temperature of physically
adsorbed molecular formic acid, but below the decomposi-
tion temperature of the surface formate intermediates. The
formic acid partial pressure also influenced the amount of
physically adsorbed molecular formic acid that condensed
on the catalyst pores, and 2000 ppm of formic acid in he-
lium was found to essentially eliminate the condensation of
molecular formic acid in the pores of the oxide catalysts at
100◦C. Some of the pure metal oxide catalysts (CuO, PtO,
PdO, Rh2O3, Au2O3, and Ag2O) required lower adsorption
temperatures, 50◦C, because of catalysts reduction during
HCOOH adsorption at 100◦C.
2.2. BET Surface Area
The BET surface area of each sample was determined
by nitrogen adsorption/desorption isotherms on a Quan-
tasorb surface area analyzer (Quantachrome Corporation,
Model OS-9) using a 3 : 7 ratio of a N2/He mixture. Typi-
cally, 0.2–0.3 g of sample was used for the measurement and
the sample was outgassed at 250◦C prior to N2 adsorption
at −195.8◦C.
2.3. XPS Surface Analysis
The surface compositions of the group IB (Cu, Ag, and
Au) oxide samples were examined to determine whether
any surface impurities were present. A Scienta model
ESCA-300 X-ray photoelectron spectrometer was em-
ployed for the surface composition measurements.
2.5. Formic Acid Oxidation Reaction
Steady-state formic acid oxidation was used to examine
the reactivity of the metal oxide catalysts. The reaction was

244
FEIN AND WACHS
TABLE 2
carried out in an isothermal fixed-bed differential reactor,
which was held vertically and made out of Pyrex glass with
a 6.2-mm outer diameter. Approximately 20 mg of catalyst
wastestedforformicacidoxidationatanumberoftempera-
tures for each sample at atmospheric pressure. The reactant
gas mixture of HCOOH/He, molar ratio ∼3.5/15.5/81, was
used with a total flow rate of 100 mL/min. The gas feeds
were controlled by mass flow controllers (Brooks Model
5850). Analysis of the reactor effluent was performed using
an online gas chromatograph (HP 5890 series II) equipped
with a thermal conductivity detector (TCD). A Porapak-
QS packed column was used to separate the components
prior to introduction to the TCD. The catalytic activities
were calculated according to mole products per hour per
square meter of catalyst.
BET Surface Area and Bulk Heats of Formation (−ꢀHf)
Catalyst
Surface area (m2/g)
−ꢀHfa (kcal/mol) (298 K) (33)
MgO
CaO
SrO
23
20
6
2
17
5
50
39
2
33
4
55
4
21
3
15
9
21
3
11
43
17
2
1
1
2
9
143.8
151.7
140.8
133.0
151.8
179.7
107.1
129.3
135.6
N.A.
74.6
N.A.
97.2
269.0
60.1
65.2
76.5
66.2
49.0
22.8
58.4
20.4
N.A.
38.5
BaO
Y2O3
La2O3
TiO2
ZrO2
HfO2
CeO2
V2O5
Nb2O5
Ta2O5
Cr2O3
MoO3
WO3
Mn2O3
Fe2O3
Co3O4
Rh2O3
NiO
PdO
PtO
CuO
Ag2O
Au2O3
ZnO
Al2O3
Ga2O3
In2O3
SiO2
2.6. Temperature Programmed Reduction (TPR) in H2
TPR was carried out with an AMI-100 system (Zeton
Altamira Instruments). A complete description of the ap-
paratus can be found in a previous paper (7).
3. RESULTS
3.1. BET Surface Area Measurements and XPS Analysis
7.0
−11.0
The surface areas of the pure metal oxide catalysts are
presented in Table 2. The results reveal that the highest sur-
face areas correspond to metal oxides that are typically used
as supports (Al2O3, TiO2, ZrO2, SiO2, CeO2, and Nb2O5).
Purchased commercial catalysts were generally of low sur-
face area with the exception of Fe2O3 (21 m2/g), Ga2O3
(18 m2/g), and Y2O3 (27 m2/g), which showed compara-
tivelyhighersurfaceareas. SincetheBETexperimentswere
run with N2 adsorption, significant error might be associ-
ated with lower surface area (<5 m2/g) samples.
83.4
180
18
3
300
8
133.0
86.6
74.0
101.5
69.1
45.7
SnO2
Bi2O3
4
a Normalized per oxygen atom.
Several metal oxides were analyzed by XPS to determine
the composition of their surfaces. Of particular interest
were Au2O3, Ag2O, and CuO since these catalysts exhib-
ited extremely high activities. The XPS results revealed that
the Au2O3 sample had a surface composition of primarily
Au2O3, with trace amounts of C, Cl, Au metal, and Na. The
Ag2O sample had a surface composition of primarily Ag2O,
with trace amounts of C and Na. The CuO sample had a sur-
face composition of primarily CuO, with trace amounts of
C and N. The absence of Pt and Pd impurities from all these
surfaces reveals that the activity for formic acid oxidation
is due entirely to the metal oxide surface.
cation of the number of active surface sites available for
formic acid oxidation. The adsorption temperature is an im-
portant consideration since only active surface sites are de-
sired for activity calculations. It has been found that formic
acid is dissociatively adsorbed on the ZnO surface at room
temperature and any undecomposed formic acid was des-
orbed at 100◦C (17). Associative formic acid adsorption was
reported on NiO at 90 K, which then undergoes heterolytic
dissociation upon heating to >200 K (4). Similar literature
studies of HCOOH adsorption on the metal oxide surfaces
have shown that adsorption temperatures below 100◦C re-
sulted in the coadsorption of surface formate intermedi-
ates and physically adsorbed molecular formic acid, and
adsorption at temperatures significantly higher than 100◦C
3.2. Formic Acid Chemisorption
The oxidation of formic acid involves the dissociative resulted in the decomposition of surface formate interme-
chemisorption of formic acid to form reactive surface diates (5, 18, 19). Therefore molecularly adsorbed species
intermediates, as shown in Fig. 1. A precise determination and decomposition of formate intermediates would alter
of the maximum number of surface formate species formed the true number of surface active sites. Thus, 100◦C was
during formic acid chemisorption is essential for quantifi- chosen as the adsorption temperature for formic acid since

QUANTITATIVE DETERMINATION OF CATALYTIC ACTIVITY
245
it was above the desorption temperature of physically ad- 100 and 150◦C, which indicated that 100◦C is an accept-
sorbed molecular formic acid, at the temperature where able temperature for HCOOH dissociative chemisorption
formic acid readily dissociatively chemisorbed as surface on most metal oxide surfaces.
formate intermediates, but below the decomposition tem-
perature of the surface formate intermediates.
Despite this generalization, chemisorption on certain
metal oxides was conducted at 50◦C. CuO required a
A series of pure metal oxides were examined for lower adsorption temperature because it was reduced at
formic acid chemisorption and the results are presented in the typical adsorption temperature of 100◦C. Formic acid
Table 3. The surface formate concentration is expressed as chemisorption on precious metal group oxides (PtO, PdO,
the number of accessible surface active sites per unit sur- Rh2O3, Ag2O, and Au2O3) was also performed at 50◦C due
face area (Ns). The number of surface formate species was to the extremely reactive nature of these metal oxide sur-
calculated from the weight gain of the catalyst samples af- faces at the typical adsorption temperature. A correction
ter adsorption of formic acid at 100◦C. Several experiment factor of 0.40 is obtained for metal oxides that require a
were also conducted at 150◦C to ensure that the weight gain lower adsorption temperature by comparing the surface ac-
corresponded to dissociative adsorption in the absence of tive site density for TiO2 at 50 and 100◦C, since formic acid
molecular formic acid adsorption. Nearly identical values chemisorption at 50◦C results in both molecular and disso-
were obtained for TiO2, Nb2O5, Fe2O3, and Al2O3, at both ciatively adsorbed formic acid. The extrapolated values of
Ns for CuO, PtO, PdO, Rh2O3, Ag2O, and Au2O3 at 100◦C
are also presented in Table 3.
The adsorption of formic acid on the metal oxide samples
TABLE 3
corresponds to ∼5–6 µmol/m2 HCOOads on average (ex-
Surface Formate Decomposition Temperature (T ) and Number
p
cluding MgO, Ag2O, ZnO, SrO, Bi2O3). The metal oxides
MoO3, V2O5, WO3, and SrO all exhibited much lower val-
ues of Ns compared to the ∼5–6 µmol/m2 average seen on
similar metal oxides. This suggests that either the number of
exposed active surface sites for formic acid chemisorption
is comparatively small or the formic acid molecules do not
utilize the entire surface of these metal oxides. Prior stud-
ies have shown that alcohols only adsorb on the edges of
the MoO3 and V2O5 containing platelet morphologies (7).
The extremely low value of Ns for SiO2 is characteristic of
the well-known unreactive nature of this oxide surface (7).
Significantly higher values of Ns were obtained for MgO,
SrO, Ag2O, ZnO, and Bi2O3. The results for these oxide sur-
faces reveal a much greater number of sites available for
formic acid chemisorption. The number of active surface
sites for SrO, Ag2O, and Bi2O3 might be skewed due to the
low BET surface areas for these samples that contain sig-
nificant error. If the true surface area were higher, the num-
ber of active surface sites would decrease. The Ns values for
MgO and ZnO are thought to be comparatively higher since
these are basic catalysts, which could form bulk carbonates,
thereby distorting the true active surface site value.
of Active Surface Sites (N )
s
Ns (µmol/m2)
HCOOH decomposition
Catalyst 50◦Ca
100◦Ca
150◦Ca
temp (◦C)
MgO
CaO
—
—
—
—
—
—
9.7
—
—
—
—
—
—
—
—
—
—
—
—
8.8
—
12.4
8.3
5.6
54.7
5.1
—
—
—
—
—
—
—
83.0
7.3
20.9
7.5
11.2
10.7
3.9
2.0
3.4
4.1
1.6
3.4
8.2
6.8
1.2
1.9
7.5
11.3
4.1
4.4
6.5
6.2
4.1
2.8
27.2
2.5
22.5
6.3
5.5
5.6
—
—
—
—
—
—
4.0
—
—
—
—
3.4
—
—
—
—
—
9.1
—
—
—
—
—
—
—
—
—
5.2
—
—
—
—
—
306
313
248
309
279
289
228
246
259
280
216
242
255
217
195
238
194
233
190
178
220
96
SrO
BaO
Y2O3
La2O3
TiO2
ZrO2
HfO2
CeO2
V2O5
Nb2O5
Ta2O5
Cr2O3
MoO3
WO3
Mn2O3
Fe2O3
Co3O4
b
Rh2O3
NiO
PdOb
PtOb
88
3.3. Decomposition Temperature of Surface
Formate Species
CuOb
Ag2Ob
Au2O3
ZnO
160
156
136
261
230
181
254
244
229
257
b
In order to obtain a measure of surface stability of the sur-
face formate species, the chemisorption experiments were
extended to perform temperature-programmed decompo-
sition (TPD). TPD studies resulted in the determination
of the decomposition temperature of the surface formate
species, since the adsorption studies were conducted in the
absence of gas phase oxygen.
Al2O3
Ga2O3
In2O3
SiO2
0.1
7.0
123.2
SnO2
Bi2O3
a Adsorption temperature.
b Denotes adsorption data obtained at 50◦C and corrected to 100◦C.
After dissociative adsorption of one monolayer of sur-
face formate species was completed, the temperature was

246
FEIN AND WACHS
ramped and the derivative of the weight loss versus temper- showed activity in the temperature range between 200 and
ature was analyzed for the maximum peak in the decompo- 300◦C; therefore, the temperature of 250◦C was chosen as
sition temperatures (Tp). The decomposition temperatures the common temperature to compare the catalytic activities
of the surface formate species for the various metal ox- sincemostmetaloxidesshowedreactioninthisregion. Only
ides catalysts examined are presented in Table 3. All of the six most active catalysts (Rh2O3, PdO, PtO, CuO, Ag2O,
the metal oxides yielded only one decomposition peak, Au2O3), BaO (305◦C), and Bi2O3 (175◦C) were active out-
which occurred over a comparatively narrow range. The side of the 100◦C range targeted. The activities ranged from
decomposition temperatures vary over a range of 225◦C, 1.9 × 10−2 to 1.7 × 109 (µmol/m2 s−1), with SiO2 exhibiting
with PtO exhibiting the lowest surface formate decompo- the lowest activity and PdO the highest activity. The ac-
sition temperature (88◦C) and CaO exhibiting the highest tivities were spread out over the total range, with Bi2O3
surface formate decomposition temperature (313◦C). The exhibiting a relatively high activity 27 (µmol/m2 s−1) com-
oxides of precious metals (PtO, PdO, Ag2O, Rh2O3, and pared to that of the other samples (see Table 4).
Au2O3) as well as CuO exhibited relatively low surface for-
mate decomposition temperatures (88–178◦C). Among the
TABLE 4
transition metal oxides, Ga2O3, CoO3, Mn2O3, and MoO3
Formic Acid Oxidation Activities, Turnover Frequencies (TOF)
exhibited relatively low decomposition temperatures. The
at 250◦C, Measured Ranges, Apparent Activation Energies (Eapp),
majority of the bulk metal oxides exhibited surface formate
and Preexponential Factors
decomposition temperatures in the range 220–300◦C.
HCOOH
activity
Pre-
exponential
factors
Measured
3.4. Temperature-Programmed Reduction
(250◦C)
TOF temperature
Eapp
Catalyst (µmol/m2 s−1
)
(s−1
)
range (◦C) (kcal/mol)a A (s−1
)
a
Results were obtained from previous TPR studies for the
same samples of metal oxides as were used in the present
experiments (7).
MgO
CaO
SrO
1.3
0.081
5.8
0.016
0.011
0.28
175–240
210–310
190–215
305–330
260–290
300–320
175–200
175–205
225–275
200–260
230–260
175–200
205–240
200–235
170–225
170–205
180–200
250–270
260–280
75–115
17
10
15
31
20
26
20
10
25
6
36
18
20
13
11
14
36
26
55
8
44
28
16
31
4
13
107
11
9
1.8E+07
8.7E+02
1.1E+07
3.7E+11
1.2E+07
1.4E+10
1.6E+09
2.5E+03
2.6E+10
5.0E+01
8.1E+15
9.4E+07
6.9E+08
1.7E+07
3.1E+05
2.0E+06
1.5E+16
4.0E+10
1.9E+23
7.5E+05
6.3E+19
1.0E+03
2.1E+14
2.4E+16
3.7E+03
1.7E+08
3.6E+43
5.0E+03
4.4E+03
2.1E+10
2.4E+00
1.3E+07
1.1E+06
BaO
0.047
0.063
0.15
0.0063
0.0057
0.014
1.5
3.5. Formic Acid Oxidation, Turnover Frequency,
and Apparent Activation Energy
Y2O3
La2O3
TiO2
5.8
Surface formate groups are the surface intermediate
speciesduringtheformationoffullyoxidizedreactionprod-
ucts (CO2 and H2O) during formic acid oxidation. Al-
though formic acid can decompose in the vapor phase to
a small extent at room temperature, partial oxidation prod-
ucts were not observed for any of the oxidation experi-
ments. This was expected since the experiments were per-
formedinexcessoxygen. Also, sincetheadsorbedhydrogen
atoms that are generated in the dissociative chemisorp-
tion process combine with adsorbed oxygen to form water,
which subsequently desorbs due to weak interaction with
the surface, it can be concluded that recombination of the
hydrogen or hydroxyl groups with the surface formate does
not occur. Therefore, all of the adsorbed formate species
eventually react to produce CO2 and H2O.
ZrO2
HfO2
CeO2
V2O5
Nb2O5
Ta2O5
Cr2O3
MoO3
WO3
0.27
0.58
0.12
5.8
2.6
4.1
0.84
5.4
0.14
0.17
0.028
3.6
0.78
0.50
0.12
4.3
1.6
3.0
Mn2O3
Fe2O3
Co3O4
Rh2O3
NiO
20
0.62
1.5
2.7
0.055
0.38
68
4.4
3.0E+02
29
200–215
35–50
30–50
PdOb
PtOb
1.7E+09
4.9E+07
3.4E+03
1.2E+02
1.0E+03
0.047
0.12
0.64
1.1
0.019
1.3
27
2.7E+08
1.2E+07
CuO
1.2E+03 160–180
Ag2O
Au2O3
ZnO
Al2O3
Ga2O3
In2O3
SiO2
4.4
3.9E+02
0.0021
0.018
0.12
90–160
45–80
Extensive testing was performed to ensure activity was
only a function of temperature and not mass and heat trans-
fer limitations. Significant apparent activation energies
were found for the most active oxide catalysts (PtO, CuO,
PdO), which mostly exhibited extremely low surface areas.
Therefore, each sample was run at a series of temperatures
while maintaining less than 10% conversion to ensure that
mass and heat transfer limitations did not affect the rate
expression. Thus, the HCOOH conversion as a function of
catalyst temperature for each metal oxide sample was stud-
ied. This information was used to determine the activity of
each metal oxide for formic acid oxidation at a common
250–270
175–240
200–265
250–280
200–250
215–235
145–175
0.20
0.15
0.19
0.22
25
5
17
11
SnO2
Bi2O3
a The values of Eapp and A are only valid over the measured temperature
ranges reported for each surface.
b These metal oxide surfaces most likely encountered heat transfer lim-
itations during the oxidation reaction due to unusually high catalytic ac-
tivity. Extrapolated TOF values are estimated to be higher than the actual
temperature. Of the 33 metal oxide catalysts studied, 25 value by a factor of 101–102.

QUANTITATIVE DETERMINATION OF CATALYTIC ACTIVITY
247
The commonality of the surface formate intermedi- A (preexponential factor), and activity for formic acid oxi-
ate during dissociative chemisorption of formic acid and dation for each oxide catalyst. The extensive set of formic
steady-state formic acid oxidation on metal oxide sur- acid activity and number of active surface site data allowed
faces allows quantitative determination of the turnover the determination of the intrinsic TOFs of the different
frequency (TOF) values by normalizing the activities at catalysts, which can be correlated with other catalytic pa-
250◦C to the number of active surface sites determined rameters to obtain additional fundamental insights about
for each catalyst. For each metal oxide studied, all of the these bulk metal oxide catalysts for formic acid oxidation.
adsorbed surface formate intermediates reacted to form
3.6.1. Turnover Frequency Correlations
products in only one narrow temperature region. In the
absence of multiple decomposition peaks, it can be con-
cluded that active surface sites for the adsorption of the
formate intermediate are identical to active surface sites
for the oxidation reaction. The TOF values are shown
in Table 4. The results reveal that formic acid oxida-
tion TOF values vary over 11 orders of magnitude, from
10−3 to 108 s−1. PdO exhibits the highest turnover fre-
quency (2.7 × 108 s−1) and ZnO the lowest (2.1 × 10−3 s−1).
Formic acid oxidation TOFs are plotted against the en-
thalpies of formation of selected bulk metal oxides (Mx Oy)
normalized per mole of oxygen atom (see Fig. 2). The TOFs
do not appear to correlate with the bulk metal–oxygen bond
strength, however, a very weak correlation is seen that sug-
gests that TOF decreases with increasing ꢀHf for the bulk
metal oxides. The temperature required to reach a constant
activity of 31.6 (µmol/m2 s−1) for HCOOH oxidation for
each oxide catalyst is plotted against the enthalpies of for-
mation in Fig. 3. Note that the temperatures are plotted
inversely to emphasize that more active catalysts require
lower reaction temperatures to achieve this activity. This
plot again resulted in a weak correlation, suggesting that
activity decreases with increasing ꢀHf for the bulk metal
oxides. The onset temperatures for H2 reduction were plot-
ted against the enthalpies of formation for the available
metal oxide samples in Fig. 4, and no correlation was ob-
tained. The TOF values were also plotted against the onset
temperature for reduction, obtained from H2-TPR of the
bulk metal oxide catalysts, which is shown in Fig. 5. How-
ever, no correlation was found between TOF and H2-TPR
onset temperature for reduction. The isotopic dioxygen ex-
change rate constants for the available metal oxides were
obtained from Boreskov (20), at a common temperature of
300◦C, and normalized to the surface area of each sample.
The TOFs were plotted against the isotopic dioxygen ex-
change rate constant K (molecules/cm2 s−1), in Fig. 6, for
selected metal oxide catalysts with no apparent correlation.
A strong inverse relationship was found to exist between
the formic acid oxidation TOFs and the surface formate in-
termediate decomposition temperatures, as shown in Fig. 7.
A plot of the preexponential factor against Eapp shows a
strong relationship, as shown in Fig. 8.
Most of the precious metal oxides PtO (1.2 × 107
s
−1),
PdO (2.7 × 108
s
−1), Rh2O3 (6.8 × 101
s
−1), and Au2O3
(3.9 × 102 s−1), as well as CuO (1.2 × 103 s−1), possess the
highestturnoverfrequencies. Severaltransitionmetaloxide
catalysts, TiO2 (1.5 × 100 s−1), V2O5 (3.6 × 100 s−1), MoO3
(4.3 × 100 s−1), WO3 (1.6 × 100 s−1), Mn2O3 (2.7 × 100 s−1),
and NiO (4.4 × 100 s−1), as well as Ag2O (4.4 × 100 s−1),
also exhibited comparatively high turnover frequencies.
In addition to ZnO, comparatively low turnover frequen-
cies were demonstrated by BaO (6.3 × 10−3 s−1) and Y2O3
(5.7 × 10−3
s
−1).
In addition to the activity data at the common temper-
ature of 250◦C, the conversion at each temperature was
used to construct an Arrhenius plot of HCOOH oxida-
tion for each catalyst. The Eapp values ranged from 4 to
107 kcal/mol, with Ag2O exhibiting the lowest value and
ZnO exhibiting the highest (Table 4). The majority of the
metal oxides studied exhibited apparent activation ener-
gies in the range 10–36 kcal/mol. Comparatively low appar-
ent activation energies were shown by CeO2 (6 kcal/mol),
Rh2O3 (8 kcal/mol), Ag2O3 (4 kcal/mol), SiO2 (5 kcal/mol),
and Ga2O3 (9 kcal/mol). The highest values were shown
by Co3O4 (55 kcal/mol), NiO (44 kcal/mol), and ZnO
(107 kcal/mol). The preexponential factor for each surface
was also calculated and values for A ranged from 2.4 × 100
to 3.6 × 1043, with SiO2 exhibiting the lowest value and
ZnO the highest (Table 4). The majority of the metal ox-
idesexhibitedpreexponentialfactorsbetween1.0 × 103 and
2.4 × 1016. Comparatively low values are noted for SiO2
(2.4 × 100), CeO2 (5.0 × 101), and CaO (8.7 × 102), and high
values for NiO (6.3 × 1019), Co3O4 (1.9 × 1023), and ZnO
(3.6 × 1043).
4. DISCUSSION
The results presented above reveal both qualitative and
quantitative information about the intrinsic activity of bulk
metal oxide catalysts for formic acid chemisorption and ox-
idation and their relationship to other thermodynamic and
kinetic parameters. It has previously been shown by in situ
Raman spectroscopy that during the oxidation of methanol,
3.6. Correlations
The quantitative techniques employed in the analysis of the surfaces of Rh, Pd, and Pt metals remain oxidized
these samples resulted in values for Ns (number of active as Rh2O3, PdO, and PtO respectively (21–23). A similar
surface sites), Tp (decomposition temperature), Eapp values, finding was reported for in situ Raman spectroscopy studies

248
FEIN AND WACHS
1E+9
1E+8
1E+7
1E+6
1E+5
1E+4
1E+3
1E+2
1E+1
1E+0
1E-1
1E-2
1E-3
1E-4
PdO*
CuO
Au2O3
Rh2O3
MoO3
V2O5
Mn2O3
Ta2O5
NiO
Ag2O
WO3
TiO2
Co3O4
HfO2
ZrO2
SrO
SiO2
In2O3
Ga2O3
Bi2O3
SnO2
Fe2O3
Cr2O3
Al2O3
BaO
La2O3
CaO
Y2O3
MgO
ZnO
-10
40
90
−∆Hf (kcal/mol) of Bulk Metal Oxides (298K)
140
190
* Extrapolated TOF value estimated higher than the actual value by a factor of 101-102 due to heat transfer limitations
FIG. 2. Semilog plot of TOF vs enthalpy of formation of bulk metal oxide catalysts.
200
400
600
800
PdO
Au2O3
Ag2O
Rh2O3
CuO
NiO
Mn2O3
In2O3
V2O5
Fe2O3
ZnO
SrO
Ta2O5
TiO2
Bi2O3
HfO2
BaO
Co3O4 MoO3
MgO
SnO2
La2O3
Cr2O3
WO3
Y2O3
Ga2O3
1000
ZrO2
1200
1400
CaO
1600
2700
Al2O3
Al2O3
3200
190
-10
40
90
140
−∆Hf (kcal/mol) of Bulk Metal Oxides (298K)
FIG. 3. Plot of temperature to achieve activity of 31.6 µmol/m2 s−1 (reverse order) vs enthalpy of formation of bulk metal oxide catalysts.

QUANTITATIVE DETERMINATION OF CATALYTIC ACTIVITY
249
700
600
500
400
300
200
100
0
MoO3
V2O5
WO3
SnO2
La2O3
Bi2O3
In2O3
Ta2O5
Co3O4
Y2O3
Ga2O3
CuO
NiO
Cr2O3
Ag2O
Fe2O3
Mn2O3
Rh2O3
PdO
0
20
40
60
80
100
120
140
160
−∆Hf (kcal/mol) of Bulk Metal Oxides (kcal/mol)
FIG. 4. Plot of onset temperature for reduction of metal oxides vs enthalpy of formation of bulk metal oxide catalysts.
1E+9
PdO*
1E+8
1E+7
1E+6
1E+5
1E+4
1E+3
1E+2
1E+1
1E+0
1E-1
1E-2
1E-3
1E-4
PtO*
CuO
NiO
Rh2O3
V2O5
Ag2O
MoO3
Mn2O3
WO3
Ta2O5
Co3O4
Cr2O3
SrO
Ga2O3
CaO
Bi2O3
SnO2
In2O3
Fe2O3
CeO2
La2O3
BaO
Y2O3
0
100
200
300
400
500
600
700
Onset Temperature for Reduction of Metal Oxides Tonset (oC)
* Extrapolated TOF values estimated higher than the actual value by a factor of 101-102 due to heat transfer limitations
FIG. 5. Semilog plot of TOF vs onset temperature for reduction of metal oxides.

250
FEIN AND WACHS
10000
1000
100
10
CuO
MoO3
NiO
V2O5
Mn2O3
WO3
TiO2
1
Nb2O5
Ta2O5
SiO2
Co3O4
Bi2O3
SrO
SnO2
ZrO2
Ga2O3
Cr2O3
Fe2O3
HfO2
0.1
MgO
CaO
0.01
Y2O3
ZnO
0.001
1E+0
1E+2
1E+4
1E+6
1E+8
1E+10
1E+12
1E+14
1E+16
K (molecules/cm2 sec)
FIG. 6. Log–log plot of TOF vs isotopic dioxygen exchange rate constant.
1E+9
PdO*
PtO*
1E+8
1E+7
1E+6
1E+5
1E+4
1E+3
1E+2
1E+1
1E+0
1E-1
1E-2
1E-3
CuO
Au2O3
Rh2O
NiO
Nb2O5
MoO3
V2O5
Ag2O
WO3
SrO
TiO2
Mn2O3
Ta2O5
HfO2
Bi2O
In2O3
ZrO2
Co3O4
Ga2O3
SnO2
Cr2O3
SiO2
3 CeO2
La2O3
Fe2O3
MgO
Al2O3
CaO
Y2O3
BaO
ZnO
50
100
150
200
250
300
350
Surface Formate Decomposition Temperature Tp (oC)
* Extrapolated TOF values estimated higher than the actual value by a factor of 101-102 due to heat transfer limitations
FIG. 7. Semilog plot of TOF vs surface formate decomposition temperature.

QUANTITATIVE DETERMINATION OF CATALYTIC ACTIVITY
251
1.E+45
1.E+42
1.E+39
1.E+36
1.E+33
1.E+30
1.E+27
1.E+24
1.E+21
1.E+18
1.E+15
1.E+12
1.E+09
1.E+06
1.E+03
1.E+00
ZnO
R2 = 0.9114
Co3O4
NiO
Mn2O3
CuO
PtO
TiO2
V2O5
Fe2O3
HfO2
Au2O3
Cr2O3
Ta2O5
BaO
La2O3
In2O3
Y2O3
Nb2O5
MgO
SrO
WO3
Bi2O3
Rh2O3
Ga2O3
SnO2
Ag2O
PdO
ZrO2
CaO
CeO2
SiO2
Al2O3 MoO3
0
20
40
60
80
100
120
Ea (kcal/mol)
FIG. 8. Semilog plot of preexponential factor vs apparent activation energy.
performed during the oxidation of methanol over polycrys- surfaces agree with the limited amount of previous data
talline Ag (24) and for X-ray absorption spectroscopy of reported for similar studies (see Table 5). Based on these
methanol oxidation over Cu (25, 26). Analogous in situ findings, it can be concluded that the values agree very well
Raman observations were also made with V2O5 and MoO3 within experimental error for metal oxides presented for
duringmethanoloxidation(27). Itcanbeconcludedthatthe the different methods employed to determine the number
surfaces of all of the metal oxide catalysts examined in the of active surface sites.
present study are also oxidized since methanol and formic
The most important benefit of calculating active surface
acid both lose two hydrogen atoms during their oxidation site densities is that different bulk metal oxide catalysts with
to H2CO and CO2, respectively. This conclusion is further broadly varying surface areas can by normalized and com-
supported by the fact that the present investigation employs pared explicitly through their turnover frequencies. Com-
excess oxygen (O2-to-formic-acid ratio of 13 : 3). Thus, the parison of the number of active surface sites for bulk metal
excess gas-phase oxygen serves to ensure complete surface oxides yields similar adsorption surface site densities for the
oxidation under the current reaction conditions.
majority of samples studied, ∼5–6 µmol HCOOads/m2. This
Formic acid chemisorption provides the number of sur- indicates the similar tendencies for the metal oxide catalytic
face active sites for bulk metal oxide catalysts. Since most surfaces to dissociatively chemisorb formic acid despite the
catalytic solids are polycrystalline, each particle may pos- various surface arrangements. In addition, catalyst prepa-
sess vacancies and defects with a different number of active ration conditions do not appear to play a critical role in the
surface sites. The chemisorption method employed in this
experiment enables the determination of an average active
surface site density for the bulk metal oxides. It has been
TABLE 5
shown that all adsorption sites are occupied by formic acid
chemisorption on TiO2 at monolayer coverages (28). Thus,
the number of active surface site values represents the max-
imum number of sites that can chemisorb a surface formate
species and hydroxyl species for a given surface, taking into
account steric interactions as well as available surface area.
The quantitative values for the number of active sur-
face sites for formic acid chemisorption onto metal oxide
Active Surface Site Comparison to Reported Literature Values
Reported number
of active sites
Present study
calculation
Catalyst
TiO2 (powder)
2.3 µmol/nm2 (2)
3.9 µmol/nm2
2.0 µmol/m2
ZrO2 (powder) 1.0 µmol/m2 (10)
ZnO
8.7 µmol/m2, crystal faces: 22.5 µmol/nm2 (powder)
(100), (010), (001) (11)

252
FEIN AND WACHS
activesurfacesitedensitysincemostsampleswereprepared orders of magnitude, from 10−3 to 108 s−1. These values
under various conditions and exhibit similar values. represent a true catalytic parameter for a given metal ox-
In comparison to analogous studies performed using ide catalytic surface toward formic acid oxidation. Thus,
other chemical probe molecules for metal oxide chemisorp- the TOF values can be compared to other catalytic param-
tion, there also appears to be similar values for the active eters explicitly since they embody an accurate normalized
surface site density. For methanol (CH3OH) chemisorp- surface property of the metal oxide surfaces.
tion, most metal oxide catalysts were shown to exhibit ∼3–
Kinetic studies of formic acid oxidation also made pos-
4 µmol CH3Oads/m2 on average (7). Thus, it appears that sible the determination of the apparent activation energies
the majority of metal oxides demonstrate comparable num- and preexponential factors for the various metal oxide cata-
bers of active surface sites for the dissociative adsorption lysts for formic acid oxidation. These apparent activation
of small chemical probe molecules.
energies are not corrected for the enthalpy of adsorption
The surface formate decomposition temperature, as ob- (30); however, they do give insight into the strength of
tained from the TPD studies for each metal oxide sam- chemisorption of the surface intermediate formate species.
ple, was used as the measure of surface formate stability, Although previous studies have reported apparent activa-
which is strictly a surface property. According to Poulston tion energies for formic acid decomposition on metal oxide
et al., a main surface formate decomposition peak to CO2 catalysts, no information could be found in the literature
from CuO is observed at 272◦C, with a minor peak at 157◦C for steady-state oxidation experiments. Therefore, the cal-
(29). The present analysis revealed a primary decomposi- culated values are presented (Table 4) without comparison
tion peak at 241◦C and a minor peak at 160◦C. The latter for- to previous literature values.
mate decomposition temperature, which would be achieved
To determine the effect of heat transfer on the oxida-
first during steady-state kinetic studies in which the temper- tion of formic acid for the metal oxide surfaces, a plot of
ature is ramped until oxidation products are observed, is the preexponential factor versus the apparent activation
nearly identical to the value reported by Poulston et al. TPD energy was constructed (Fig. 8). This plot clearly shows a
experiments for formate decomposition on ZnO were also direct correlation between these parameters, as would be
compared with literature values and found to be similar. expected as a compensation effect for the activity of each
It has been reported that the decomposition of the surface of the surfaces. The two catalysts exhibiting the highest ac-
formate intermediate occurs at about 200◦C for the surface tivity, PdO and PtO, deviate most significantly from the lin-
of ZnO(0001) under ultrahigh-vacuum (UVH) conditions ear correlation. It can be assumed that these surfaces most
(17), compared to 261◦C observed for powdered ZnO in likely encountered heat transfer limitations during the oxi-
the present study. Low formic acid exposures resulted in dation reaction due to their unusually high catalytic activity.
decomposition of the surface formate intermediate to CO2 Based on a comparison of the experimental results for PdO
and H2O at 202◦C for TiO2(110) (8). The decomposition and PtO to the majority of the metal oxides studied, the
temperature obtained in the present study for a powder calculated TOF values for these two surfaces are estimated
TiO2 sample was 237◦C, which is similar to that obtained in to be higher than the actual value by a factor of 101–102.
the single-crystal study. The present analysis exhibits accu- Note that due to the unusually high activity demonstrated
rate decomposition temperatures for powder metal oxide for formic acid oxidation by these surfaces, any overstate-
catalysts, which represent the average of all crystal faces ment in the reported values not corrected for heat transfer
and are used to quantify the surface formate intermediate limitations will not affect any of the correlations, trends, or
stability.
conclusions discussed in this study. Also, the reported val-
Formic acid oxidation over the bulk metal oxide catalysts ues for both Ea and A are only accurate over the measured
yielded exclusively CO2 and H2O. Therefore, selectivity of temperature range for each surface for formic acid oxida-
a particular surface toward the production of partial oxida- tion. Any extrapolation outside of these narrow ranges will
tion products (either H2 and CO2 or H2O and CO), as seen introduce error and will not be representative of the
in formic acid decomposition studies, did not play a factor actual values at the desired temperature. However, for
in the present investigation. Oxidation was more suitable the purpose of this study it is important to recognize
for comparison purposes since certain metal oxide surfaces that the measured temperature range for each surface is
exhibit all of these products for formic acid decomposition very close to the desired normalization point of 250◦C;
(8, 9, 17, 19). Thus, all the active surface sites determined therefore the reported values for both Ea and A apply
from the chemisorption studies can be assumed to con- directly.
tribute to the activity of the metal oxide catalyst for formic
Correlation of the TOF values with other catalytic pa-
acid oxidation to CO2 and H2O. The average activity of rameters may provide fundamental insights into these bulk
a surface site for a particular metal oxide is reported metal oxide catalysts for the formic acid oxidation reaction.
as a TOF.
Figure 2 demonstrates that a very weak inverse correlation
Comparison of the formic acid TOF values at 250◦C over results from the plot of TOF versus the bulk heat of forma-
the metal oxide catalysts reveals that they vary over 11 tion of the metal–oxygen bond strength of the metal oxide.

QUANTITATIVE DETERMINATION OF CATALYTIC ACTIVITY
253
Thus, TOF qualitatively decreases with increasing −ꢀHf tion. Therefore, no correlation should exist between a plot
for the bulk metal oxides. No correlation is seen for metal of these two parameters, as shown.
oxides with moderate-to-high bulk enthalpies of formation.
Figure 6 shows no correlation when the TOFs were plot-
It was previously stated that there appears to be a corre- ted against the isotopic dioxygen exchange rate constant
lation between the bulk heats of formation of the metal K (molecules/cm2 s−1). The isotopic dioxygen exchange
oxide catalysts divided by the number of oxygen atoms and rate constant K (molecules/cm2 s−1) is characteristic of
the catalytic activity toward the total oxidation of some the rate of exchange of bulk lattice oxygen molecules per
chemical compounds (31). However, the surface reaction unit surface per unit time for a metal oxide. The dioxy-
rates, which are characteristic of the surface activity of a gen exchange rate constant represents the exchange be-
particular catalyst for a given reaction, should not be in- tween gas phase oxygen and the surface and bulk lattice
herently related to a bulk material property of the metal oxygen atoms of the metal oxide catalysts (20). This ex-
oxide catalytic material. Thus, when the rate of a catalytic change rate, although it does play a role in the Mars–van
reaction is plotted versus the stability of the metal–oxygen Krevlen mechanism for oxidation from these surfaces, is not
bond strength, no correlation is expected, although a weak the rate-determining step for formic acid oxidation. TOF
qualitative relationship is seen.
values are dependent on the decomposition of the surface
Figure 3 shows a weak inverse correlation between the formate intermediate, which is characteristic of the break-
inverse temperature to achieve a constant catalytic ac- ing of the C–H bond representing the rate-determining step
tivity and the bulk metal–oxygen bond strength, suggest- for the reaction. Therefore, since these measures of surface
ing that activity decreases with increasing −ꢀHf for the properties do not correspond to the same step in the reac-
bulk metal oxides. The existence of the correlation is ex- tion mechanism, one should not expect that a correlation
plained with the same reasoning that can be applied for exists between TOF and the isotopic dioxygen exchange
the TOF versus −ꢀHf correlation seen in Fig. 2. One pos- rate.
sible explanation for the low activity of Al2O3 is that it is
The TOF values were also plotted against the surface for-
known to form surface carbonates as high as 300◦C, which mate decomposition temperatures obtained from the TPD
would inhibit chemisorption HCOOH (32). It was previ- studies (see Fig. 7). Both the oxidation TOF and the TPD
ously shown that a plot of the temperature required for a decomposition temperature proceed by means of the same
surface to achieve a particular rate of activity in reverse rate-determining step, the decomposition of the surface for-
order versus a bulk property will exhibit “volcano curve” mate intermediate. Although the plot is semilog in order to
behavior for a range of materials (14). However, when the show the 11 orders of magnitude change in the TOF val-
surface catalytic activity is plotted against the bulk metal– ues, a qualitative inverse relationship is obtained between
oxygen bond strength, one would not expect a correlation the TOF values and the decomposition temperatures of the
since this would represent a correlation of surface and bulk surface formate intermediates. This result indicates that
material properties. An important feature of Fig. 3 is that the thermal stability of the surface formate intermediate
although the temperature required to reach a constant ac- is the only significant kinetic parameter. Thus, dissociative
tivity increases with the bulk metal–oxygen bond strength, adsorption of formic acid is not kinetically significant dur-
no “volcano curve” is obtained.
ing HCOOH oxidation over metal oxide catalysts. Simi-
Temperature-programmed reduction (H2-TPR) studies lar findings were reported when methanol was used as the
are commonly used to reflect the ease of oxygen removal chemical probe molecule for the same set of metal oxide
from metal oxide catalysts. The onset temperature for re- catalysts (7).
duction corresponds to either the dissociative adsorption of
The data in Fig. 7 demonstrate that only the right-hand
molecular H2 or the recombination of surface hydroxyls to side of the classic “volcano plot” (14) is obtained when
form H2O, which may vary among the different metal ox- the TOF is plotted against the surface formate decomposi-
ide catalysts. When the onset temperature for reduction of tion temperature. It has been proposed that these classic
the metal oxide catalysts is plotted against the bulk metal– “volcano plots” suggest that there is a shift in the rate-
oxygen bond strength, as shown in Fig. 4, no correlation determining step of the reaction as one crosses from the
exists.
“left-hand” side to the “right-hand” side of the plot. The
The TOF values are plotted against the H2-TPR-Tonset of catalysts on the left side of the curve exhibit adsorption
the metal oxide catalysts in Fig. 5. Each of these parameters limiting behavior. Consequently, the corresponding reac-
usesadifferentchemicalprobemoleculetoquantifytheH2- tion rate is small since the reactant molecules are not read-
TPR-Tonset. The onset temperature for reduction uses H2 to ily dissociatively adsorbed onto the active surface sites. On
establish the rate-determining steps of either H2 dissocia- the right side of the curve, decomposition from the surface
tive adsorption or hydroxyl recombination on the surface, has been proposed to be the rate-limiting process. These
while TOF values were calculated using formic acid as the surfaces dissociatively adsorb the reactant easily; however,
probe molecule and depend on the breaking of the C–H the bond to the surface is very strong, thereby limiting
bond as the rate-determining step during HCOOH oxida- the reaction rate since surface reaction and desorption are

254
FEIN AND WACHS
necessary for products to be formed. Compounds at the
peak of the volcano curve, commonly precious metal cat-
alysts, exhibit an optimal balance of intermediate stability
as put forth by Sabatier and, therefore, should exhibit the
highest rates of reaction (13). These classical interpreta-
tions are likely correct for surface reactions where the ad-
sorption of chemical probe molecules on the catalyst sur-
face is an activated process, but these interpretations do
not appear to hold for surface reactions where adsorption
is relatively easy and the decomposition of the surface in-
termediate is the rate-determining step (see Fig. 7). Similar
findings were also reported for analogous methanol oxida-
tion studies over bulk metal oxide catalysts (7).
REFERENCES
1. Lewis, B. F., Mosesman, M., and Weinberg, W. H., Surf. Sci. 41, 142
(1974).
2. Root, T. W., and Duncan, T. M., J. Catal. 101, 527 (1986).
3. Vittadini, A., Selloni, A., Rotzinger, F. P., and Gratzel, M., J. Phys.
Chem. B 104, 1300 (2000).
4. Truong, C. M., Wu, M. C., and Goodman, D. W., J. Phys. Chem. 97,
9447 (1992).
5. Nakatsuji, H., Yoshimoto, M., Umemura, Y., Takagi, S., and Hada, M.,
J. Phys. Chem. 100, 694 (1996).
6. Nakatsuji, H., Yoshimoto, M., Hada, M., Domen, K., and Hirose, C.,
Surf. Sci. 336, 232 (1995).
7. Badlani, M., and Wachs, I. E., Catal. Lett. 75, 137 (2001).
8. Henderson, M., J. Phys. Chem. B 101, 221 (1997).
9. Bandara, A., Kubota, J., Wada, A., Domen, K., and Hirose, C., J. Phys.
Chem. B 101, 361 (1997).
10. Ouyang, F., Nakayama, A., Tabada, K., and Suzuki, E., J. Phys. Chem.
B 104, 2012 (2000).
5. CONCLUSIONS
11. Noto, Y., Fukuda, K., Onishi, T., and Tamaru, K., Trans. Faraday Soc.
63, 3081 (1967).
12. Cordi, E. M., O’Neill, P. J., and Falconer, J. L., Appl. Catal. B 14, 23
(1997).
13. Ertl, G., Kno¨zinger, H., and Weitkamp, J., Eds., “Handbook of Het-
erogeneous Catalysis,” Vol. 1. VCH, Weinheim, 1997.
The catalytic oxidation of formic acid on bulk metal
oxides was studied using (i) dissociative chemisorption
to form surface formate intermediates for determination
of the number of active surface sites, (ii) temperature-
programmeddesorptiontoobtainthesurfaceformateinter- 14. Sachtler, W. M. H., and Fahrenfort, J., in “Proceedings, 2nd Interna-
tional Congress on Catalysis, Paris, 1960,” p. 831. Technip, Paris, 1961.
15. Weast, R. C., Ed.,“Handbook of Chemistry and Physics,” CRC Press,
Boca Raton, FL., 1986–1987.
16. Briand, L. E., Farneth, W. E., and Wachs, I. E., Catal. Today 62, 2–3
mediate decomposition temperatures, and (iii) steady-state
oxidation kinetic studies to calculate TOFs. Comparison of
the number of active surface sites for bulk metal oxides
yields similar adsorption surface site densities for the ma-
(2000).
jority of samples studied, ∼5–6 µmol HCOOads/m2. Kinetic 17. Akhter, S., Lui, K., and Kung, H. H., J. Phys. Chem. 89, 1958 (1985).
18. Fukuda, K., Noto, Y., Onishi, T., and Tamaru, K., Trans. Faraday Soc.
studies of formic acid oxidation made possible the determi-
nation of the apparent activation energies for the various
metal oxide catalysts for formic acid oxidation. The TOF
63, 3072 (1967).
19. Eischens, R. P., Pliskin, W. A., in “Proceedings, 2nd International
Congress on Catalysis, Paris, 1960,” p. 789. Technip, Paris, 1961.
values for the various metal oxide catalysts studied were
found to vary over 11 orders of magnitude (10−3–108 s−1).
The variation in TOF values was not related to the isotopic
dioxygen exchange rate constant or the onset temperature
for reduction of the metal oxide surfaces. Weak inverse cor-
relations were seen between TOF and activity and the bulk
metal–oxygen bond strength, however, a “volcano curve”
over the range of bulk metal oxide catalysts was not ob-
served. A strong inverse relationship was found between
the TOF values and the decomposition temperatures of the
surface formate intermediates. This result indicates that
20. Boreskov, G., in “Catalysis Science and Technology” (J. R. Anderson
and M. Boudart, Eds.), Vol. 3, p. 62. Springer-Verlag, New York, 1982.
21. Williams, C. T., Takoudis, C. G., and Weaver, M. J., J. Phys. Chem 102,
406 (1998).
22. Williams, C. T., Chan, H. Y. H., Tolia, A. A., Weaver, M. J., and
Takoudis, C. G., Ind. Eng. Chem. Res. 37, 2307 (1998).
23. Chan, H. Y. H., Williams, C. T., Weaver, M. J., and Takoudis, C. G.,
J. Catal. 174, 191 (1998).
24. Wang, C. B., Deo, G., and Wachs, I. E., J. Phys. Chem. 103, 5645 (1999).
25. Knop-Gericke, A., Havecker, M., Schedel-Niedrig, T., and Schlogl, R.,
Catal. Lett. 66, 215 (2000).
26. Werner, H., Herein, D., Schulz, G., Wild, U., and Schlogl, R., Catal.
Lett. 49, 109 (1997).
the thermal stability of the surface formate intermedi- 27. Deo, G., Hu, H., and Wachs, I. E., unpublished results.
28. Enriquez, M. A., and Fraissard, J. P., J. Catal. 74, 89 (1982).
29. Poulston, S., Rowbotham, E., Stone, P., Parlett, P., and Bowker, M.,
Catal. Lett. 52, 63 (1998).
30. Burcham, L. J., and Wachs, I. E., Catal. Today 49, 467 (1999).
ate is the only significant kinetic factor. Thus, dissocia-
tive adsorption of the formic acid molecule is not kineti-
cally significant for kinetic activity during formic acid ox-
idation over the metal oxide catalysts. Consequently, it
is concluded that one must consider only surface proper-
ties when attempting to establish correlations for catalytic
parameters.
31. Hucknall, D. J., “Selective Oxidation of Hydrocarbons.” Academic
Press, London, 1974.
32. Turek, A. M., Wachs, I. E., and DeCanio, E., J. Phys. Chem. 96, 5000
(1992).
33. Perry, R. H., and Green, D. W., Eds., “Perry’s Chemical Engineering
Handbook,” 7th ed. McGraw–Hill, New York, 1997.
34. Stubenrauch, J., Brosha, E., and Vohs, J. M., Catal. Today 28, 431
(1996).
ACKNOWLEDGMENT
35. Wachs, I. E., manuscript in preparation.
36. Farneth, W. E., Staley, R. H., and Sleight, A. W., J. Am. Chem. Soc.
108, 2327 (1986).
The support of the Lehigh University Presidential Scholar’s Program is
gratefully acknowledged.