Structure-dependent kinetics for synthesis and decomposition of formate species over Cu(111) and Cu(110) model catalysts

Article abstract of DOI:10.1021/jp002644z

The kinetics and mechanism of formate synthesis by hydrogenation of CO₂ (CO₂ + 1/2H₂ → HCOO_a) and the formate decomposition into CO₂ and H₂ (HCOO_a CO₂ + 1/2H₂

Full text of DOI:10.1021/jp002644z

J. Phys. Chem. B 2001, 105, 1355-1365
1355
Structure-Dependent Kinetics for Synthesis and Decomposition of Formate Species over
Cu(111) and Cu(110) Model Catalysts
Haruhisa Nakano,† Isao Nakamura,† Tadahiro Fujitani,‡ and Junji Nakamura*,†
Institute of Materials Science, UniVersity of Tsukuba, Tsukuba, Ibaraki 305-8573, Japan, and
National Institute for Resources and EnVironment, Tsukuba, Ibaraki 305-8569, Japan
ReceiVed: July 25, 2000; In Final Form: NoVember 1, 2000
The kinetics and mechanism of formate synthesis by hydrogenation of CO2 (CO2 + 1/2H2 f HCOOa) and
the formate decomposition into CO2 and H2 (HCOOa f CO2 + 1/2H2) over Cu(111) and Cu(110) surfaces
were studied by in-situ infrared reflection-absorption spectroscopy (IRAS) using a high-pressure reactor
(∼1 atm). The reaction rates and the apparent activation energy of the formate synthesis were measured for
Cu(111) and Cu(110), indicating that the formate synthesis on Cu was found to be structure-insensitive. The
pressure dependence of CO2 and H2 on the initial formation rate of formate suggested an Eley-Rideal type
mechanism, in which a gaseous CO2 molecule directly reacts with an adsorbed hydrogen atom on Cu. This
is analogous to the well-known mechanism of formate synthesis by organometallic catalysts, in which CO2
is inserted into a Cu-hydride bond. The reaction rates and the activation energy of the decomposition were
measured for Cu(111) and Cu(110). It was found that the formate decomposition on Cu was structure-sensitive
in contrast to the formate synthesis. The promotional effect of coexisting H2 upon the rate of formate
decomposition by 17 times at maximum was incidentally found only on Cu(111). Interestingly, the increase
in the decomposition rate was due to an increase in the preexponential factor of the rate constant for the
formate decomposition with the activation energy being constant. Furthermore, the decomposition kinetics of
the formate prepared by adsorption of formic acid on O/Cu(111) was identical with the H2-promoted
decomposition kinetics of the synthesized formate. The difference in the decomposition kinetics was ascribed
to the ordered structure of formate based on the previous STM results, in which a chainlike structure of
formate was observed for the synthesized formate, whereas no formate chain was observed for the formate
prepared by adsorption of formic acid on O/Cu(111). The unique character of both the decomposition kinetics
and the structure of formate observed only for Cu(111) was discussed from the viewpoint of the mass transport
of copper atoms creating added formate chains.
1. Introduction
techniques, if a catalyst model can once be established on a
single-crystal surface.
A surface science approach is available for studying the
mechanism and kinetics of catalysis on solid surfaces from both
macroscopic and microscopic viewpoints. In conjunction with
industrial catalysts, the catalytic reactions should be examined
under industrial conditions of 1 ∼ 100 atm using a high-pressure
reactor. The Somorjai group originally developed high-pressure
studies in surface science to measure catalytic kinetics on well-
defined surfaces, for example, hydrogenolysis and isomerization
of hydrocarbons on Pt surfaces and ammonia synthesis on iron
surfaces.1 Several research groups have since measured the
kinetics of major industrial catalytic reactions, such as metha-
nation, Fischer-Tropsch synthesis, methanol synthesis, ethylene
epoxidation, CO oxidation, the water-gas shift reaction and so
on.2 Although the surface science approach using a high pressure
reactor has the potential for clarifying the detailed mechanism
of the catalytic reactions at an atomic level, further studies have
not been fully done. That is, one can further look into the
microscopic mechanism, the nature of the active site, and the
kinetics features using a variety of modern surface science
We have studied the kinetics and mechanism of methanol
synthesis from CO2 and H2 on Cu model catalysts by surface
science techniques using high-pressure reactors (1∼18 atm),3-14
while studying methanol synthesis over Cu-ZnO based powder
catalysts.15-18 We found that a Zn-deposited Cu(111) surface
could be regarded as a good model of Cu/ZnO powder catalysts
in terms of the turnover frequency of methanol production and
the apparent activation energy. STM studies showed that the
deposited Zn created a Cu-Zn active site on Cu(111), where
Zn atoms were substituted for Cu atoms.13,19 The assigned role
of the Cu-Zn site in the reaction mechanism was to promote
the hydrogenation of formate (HCOO) to methoxy (H3CO)
species. Surface Cu atoms are also necessary for several
hydrogenation steps such as formate synthesis from CO2 and
H2. That is, the Cu-Zn site and Cu atoms cooperate to catalyze
methanol synthesis. We then tried to describe the mechanism
of methanol synthesis on the Zn/Cu(111) model catalyst by the
kinetics of the constituent elementary steps. The kinetic
measurement of formate synthesis (CO2 + 1/2H2 f HCOOa)
was first carried out on Zn/Cu(111) and clean Cu(111) surfaces
at atmospheric pressure. During XPS and STM studies of
formate, various interesting phenomena were observed regarding
the kinetics of the synthesis and the decomposition of formate
* Corresponding author. Fax: (+ 81) 298 557440, Tel: (+ 81) 298
535279, E-mail: nakamura@ims.tsukuba.ac.jp
† University of Tsukuba.
‡ National Institute for Resources and Environment.
10.1021/jp002644z CCC: $20.00 © 2001 American Chemical Society
Published on Web 01/31/2001

1356 J. Phys. Chem. B, Vol. 105, No. 7, 2001
Nakano et al.
Figure 1. Schematic diagram of an IRAS apparatus equipped with a high-pressure reactor.
as well as the ordered structure.6,10,13,19 Chainlike and non
chainlike structures of formate were seen on Cu(111) depending
on the preparation method.13,19 Also, the decomposition kinetics
of formate depended on the preparation method or the presence
of hydrogen. Because of the very unique character and the
importance of formate on Cu as a reaction intermediate in
methanol synthesis, we concentrated on the systematic study
of the synthesis and decomposition of formate on Cu(111) and
Cu(110) as well as on a Cu/SiO2 catalyst. We report here the
characteristic kinetics for the synthesis and decomposition of
formate on Cu(111) and Cu(110) in relation to adsorption
structure.
The structure and the kinetic behavior of formate have been
widely studied by surface science techniques.19-25 On single-
crystal Cu surfaces, formate is known to adsorb in a bridging
bidentate form with two oxygen atoms bound to Cu atoms by
IRAS, HREELS, and EXAFS studies.26-29 The local registry
of formate on Cu(111) has been recently determined by normal
incidence X-ray standing wavefield absorption (NIXSW) in
which the oxygen atoms are located on atop sites.30 The
adsorption site is identical to that reported for formate on
Cu(110) and Cu(100).31,32 STM studies showed that exposure
of O/Cu(110) to formic acid resulted in the formation of c(2 ×
2) and (3 × 1) formate structures depending on the oxygen
coverage,25 where the formate species studied is mostly prepared
by adsorption of formic acid in UHV.
On the other hand, formate synthesis from CO2 and H2 was
carried out on Cu single-crystal surfaces at high pressures (1∼2
atm). Chorkendorff et al. have observed no difference in the
decomposition kinetics on Cu(100) for the formate prepared by
the adsorption of the formic acid in UHV and synthesized by
hydrogenation of CO2.20 However, our previous results indicated
that the structure and the reactivity of formate prepared from
formic acid on O/Cu(111) were quite different from those of
the synthesized formate.13,19 The difference in the kinetic feature
originating from the ordered structures of formate was described
in this paper.
we report the initial suggestion seen in the pressure dependence
of H2 and CO2 upon formate synthesis on Cu(111). Further
studies by a dynamic method and ab initio DFT calculations
support the E-R mechanism, which will be reported soon.
2. Experimental Section
The experiments were conducted using UHV-infrared reflec-
tion absorption spectroscopy (IRAS) as shown in Figure 1. This
apparatus is composed of four chambers: a load-lock chamber
for changing the sample, a preparation chamber equipped with
an ion gun for Ar+ sputtering, an analysis chamber with facilities
for low-energy electron diffraction (LEED), Auger electron
spectroscopy (AES), and quadrupole mass spectroscopy (QMS),
and a reaction chamber in which it is possible to carry out
reactions at atmospheric pressure and take in-situ IRAS
measurements during the reaction. The vibrational spectra in
the range of 700-3000 cm-1 were obtained at a resolution of
4 cm-1 by 100 scans in a total measurement time of 30 s. An
infrared spectrometer with a liquid nitrogen-cooled mercury-
cadmium-telluride (MCT) detector was situated next to the
reaction chamber. Each chamber is shut with a gate valve, and
a sample is transferred using transfer rods. It is possible to
transfer the sample without exposing it to the atmosphere.
The Cu(111) and Cu(110) disks (10 mm diameter, 2 mm
thickness, and 5N purity) used in this study were polished only
on one side. The accuracy of the crystal plane was less than
one degree, and the surface roughness was within 0.03 µm. The
sample was mounted along with two 0.25 mm diameter tungsten
wires for resistive heating. The sample temperature was
measured by a chromel-alumel thermocouple. The sample was
cleaned by repeated Ar+ sputtering at room temperature and
773 K and was then annealed at 773 K for 10-15 min. The
cleanliness of the sample was evaluated by AES. A LEED
pattern showed a clear (1 × 1) structure for Cu(111) and
Cu(110). Because both CO2 and H2 have a low sticking
probability on Cu surfaces at the temperatures used in this
experiment, it is important to remove impurities such as CO
and O2 from the CO2/H2 reaction gas. The CO2 (99.9999% pure)
was thus purified by freeze-pump-thaw cycles using liquid
nitrogen. The H2 (99.9999% pure) was purified by passing it
through a tube cooled by liquid nitrogen.
We also describe the Eley-Rideal (E-R) type mechanism
of formate synthesis strongly suggested by kinetic analysis. As
far as we know, no E-R type mechanism has been reported
for major industrial catalytic reactions over metal catalysts. Here,

Synthesis and Decomposition of Formate Species
J. Phys. Chem. B, Vol. 105, No. 7, 2001 1357
The formate synthesis by the hydrogenation of CO2 on the
clean Cu(111) and Cu(110) surfaces was performed at 323-
453 K, CO2 ) 76-380 Torr and H2 ) 10-380 Torr. After the
sample temperature became constant at the reaction temperature,
CO2 and H2 were subsequently introduced into the reaction
chamber. The sample temperature was controlled within ( 1
K. In the experiment on formate synthesis, the reaction
temperature was set at 323-353 K at which the thermal
decomposition of the formate species was negligibly small.9,10
A typical pressure condition of formate synthesis was CO2/H2
) 380 Torr/380 Torr.
We have also studied the formate prepared by exposure to
formic acid at 333 K in UHV. Formic acid was introduced onto
the clean Cu(111), Cu(110), and oxygen-precovered and/or
oxygen-free surfaces at 1 × 10-8 ∼ 1 × 10-6 Torr using a
stainless steel leak valve. The formic acid (99%) was dried over
copper sulfate anhydride and further purified by freeze-pump-
thaw cycles prior to admission into the vacuum chamber. QMS
was used to evaluate the purity of the formic acid.
It has been reported that the peak intensity of the νs(OCO)
band observed at about 1340-1360 cm-1 is proportional to the
formate coverage (ΘHCOO) on the copper surfaces.26 We thus
estimated formate coverage using the peak intensity of the
νs(OCO). In the XPS measurements of the same formate
synthesis by the hydrogenation of CO2, the saturated formate
coverage was determined to be 0.24 on Cu(111).6 Also, in the
previous STM results, a c(2 × 4) structure as the most densely
packed structure corresponding to ΘHCOO ) 0.25 was observed
for the formate species synthesized from CO2/H2 on Cu(111).13,19
Furthermore, the saturated formate coverage on Cu(110) has
been reported to be 0.25 based on a TPD analysis.26 Thus,
in this study, we adopted the saturation formate coverage
(ΘsHaCt OO) of 0.25 on both Cu(111) and Cu(110).
Figure 2. IRA spectra of formate species on Cu(111) synthesized by
CO2 hydrogenation (CO2/H2 ) 1, 760 Torr) at 353 K.
stretching band (ν(CH)) and the combination band of the
asymmetric OCO stretching and the in-plane CH bending
modes, respectively. No other surface species were detected on
the postreaction surfaces by AES.
To elucidate the kinetics of formate synthesis, the buildup
of formate was repeatedly measured at 333-353 K, at which
the decomposition of formate was neglected on Cu(111). Figure
3 shows the peak intensity at 1353 cm-1 as a function of total
exposure of CO2 and H2. These can be regarded as buildup
curves because the peak intensity of νs(OCO) at 1350 cm-1 is
known to be proportional to the formate coverage.26 The
saturated coverage was identical regardless of the reaction
temperature. In both XPS and STM studies,6,13,19 the saturated
formate coverage was determined to be ΘHCOO ) 0.25 under
the same reaction conditions, where Θ ) 1 corresponds to the
number of Cu atoms on Cu(111), 1.77 × 1015 atoms cm-2. The
initial formation rate of formate was thus obtained by the slope
at zero formate coverage in Figure 3, for example, 9.09 × 10-4
molecules site-1 sec-1 at 353 K.
Figure 4 shows Arrhenius plots of the initial formation rate
derived from the data in Figure 3 as well as our previous data
measured by XPS and IRAS on Cu(111) and Cu(110).6,7 It is
shown that the formation rates on Cu(111) measured in this
study are in agreement with those on Cu(111) measured by XPS
within experimental error. The activation energy and the
preexponential factor of the rate constant on Cu(111) were
determined to be 56.6 ( 4.8 kJ mol-1 and 2.27 × 105 molecules
site-1 sec-1, respectively. Those values measured by IRAS were
comparable with those measured by XPS for the same Cu(111)
sample, that is, 54.6 ( 5.9 kJ mol-1 and 7.92 × 104 molecules
site-1 sec-1. Note that no significant difference in the kinetics
of formate synthesis was observed among Cu(111), Cu(110),
and Cu(100) as has already been reported,6,7,20 indicating a
structure-insensitive reaction. That is, the activation energies
on Cu(110) and Cu(100)20 have been reported to be 59.8 ( 4.1
and 55.6 ( 8.0 kJ mol-1, respectively. The absolute TOF
(turnover frequencies) values were also close to each other.
Figures 5a and 5b show the pressure dependence of H2 and
CO2 upon the initial formation rate of formate on Cu(111) at
The isothermal decomposition of the formate species, pre-
pared by the hydrogenation of CO2 and adsorption of formic
acid, was carried out by maintaining the sample at a constant
temperature (358-403 K) in a vacuum. The decomposition rate
was determined by the IRAS measurement of the formate
coverage. After the decomposition of formate, the carbon residue
was usually below the detection limit of AES.
3. Results
3.1. Formate Synthesis. Typical IRA spectra of formate
synthesized on Cu(111) are shown in Figure 2, where the
synthetic reaction was carried out at 353 K and 380 Torr CO2/
380 Torr H2. The intensity of the peaks increased with increasing
exposure to CO2 and H2. However, no significant peak shifts
were observed upon increasing the exposure. The peak positions
were in good agreement with those reported for bridging
bidentate formate species on Cu surfaces as shown in Table 1.
That is, the peaks at 1354, 2850, and 2930 cm-1 were assigned
to the symmetric OCO stretching band (νs(OCO)), the CH
TABLE 1: Vibrational Frequencies (cm-1) for the Formate Species on Cu Single Crystals
νs (OCO)
ν (CH)
2855
2850
2846-2850
2846-2848
2920, 2960
2848
νcomb
2932
2922
2922-2928
2922-2928
preparation of formate
technique
ref
Cu(111)
Cu(111)
Cu(110)
Cu(110)
Cu(110)
Cu(110)
Cu(110)
Cu(100)
Cu(100)
Cu(100)
1352-1354
1342-1360
1352-1362
1352-1362
1360
CO2 + H2
HCOOH
CO2 + H2
HCOOH
H2CO, CH3OCHO
HCOOH
IRAS
IRAS
IRAS
IRAS
EELS
IRAS
[7]
This work
[7]
This work
[27]
1355
2930
[45]
1348-1358
2891-2900
2946-2955
HCOOH
HCOOH
HCOOH
CO2 + H2
IRAS
[26]
[28]
[44]
[20]
1325
2870
2930
HREELS
HREELS
HREELS
1331
1330
2840, 2910
2879
-
-

1358 J. Phys. Chem. B, Vol. 105, No. 7, 2001
Nakano et al.
the pressure dependencies as described in the following section.
Taylor et al.20 have reported the kinetic analysis of the formate
synthesis on Cu(100) assuming a Langmuir-Hinshelwood
mechanism (L-H mechanism). We first analyze the kinetics
of our data on Cu(111) assuming the L-H mechanism and the
following elementary reaction steps.
1
2
H2,g + / h Ha
(3)
CO2,g + / h CO2,a
(4)
(5)
Ha + CO2,a f HCOOa
Here, / and subscript a stand for a vacant site and an adsorbed
state, respectively, and the decomposition of formate, or the
reverse reaction of eq 5, is neglected at the cited low temper-
atures in our experiments. One can thus obtain the following
equations.
Figure 3. Peak intensity at 1535 cm-1 for formate species on Cu(111)
as a function of exposure to CO2 and H2 (CO2/H2 ) 1 760 Torr).
k1
ΘH )
P1H/2 Θ/
(6)
2
k-1
k2
ΘCO
)
PCO Θ/
(7)
(8)
2
2
k-2
rf ) k3ΘHΘCO
2
Here, k1 and k-1 are the rate constants of step 3 and the reverse
step (3), respectively, k2 and k-2 are those for step 4 and the
reverse step 4, respectively, and k3 is that for step 5. The formate
species is assumed to occupy two copper atoms because it is
known to adsorb in the bridging bidentate state. ΘCO is also
2
neglected in the estimation of Θ* because it is very small as
shown in Figure 6. Θ* is then expressed by the following
equation.
Figure 4. Arrhenius plots for the initial formation rates of formate
species on Cu(111) (b: IRAS, 2: XPS6) and Cu(110) (9: IRAS7).
ΘCO ≈ 0
2
Θ* ) 1 - ΘH - 2ΘHCOO
(9)
TABLE 2: Preexponential Factor and Activation Energy of
Elementary Steps
From eqs 6-9, the following equations are obtained.
preexponential
activation
factor (cm2/s) energy (kJ/mol)
ref
k1
1 - 2ΘHCOO
P1H/2
(10)
H2 + 2* f 2Ha k1
2Ha f H2 + 2* k-1
CO2a f CO2 + * k-2
100.03
60.1
84.0
28.0
Cu(110) [33]
Cu(111) [34]
Cu(100) [20, 35]
ΘH )
2
3.4 × 10-4
k-1
k1
P1H/2
1015
1 +
2
k-1
333, 343, and 353 K. The CO2 (or H2) pressure was fixed at
380 Torr, whereas the H2 (or CO2) pressure was varied from
10 to 380 Torr. As for the H2 pressure dependence in Figure
5a, the increase in rf0 with H2 pressure became small at higher
H2 pressures. On the other hand, the formation rate increased
linearly with CO2 pressure as shown in Figure 5b.
k2
1 - 2ΘHCOO
ΘCO
)
PCO
(11)
(12)
2
2
k-2
k1
1 +
P1H/2
2
k-1
2
k1k2k3
(1 - 2ΘHCOO)
rf )
P1H/2PCO
The saturating behavior seen in Figure 5a is probably due to
the coverage of hydrogen atoms, inhibiting surface reactions
on the bare copper surface. We thus estimated the equilibrium
coverage of Ha or CO2,a under the cited reaction conditions using
kinetic data measured by surface science techniques summarized
in Table 2. The coverage of Ha and CO2,a is shown in Figure 6
as a function of H2 and CO2 pressures, respectively. The
equilibrium coverage of Ha at 343 K and 200 Torr H2 was
calculated to be 0.52, whereas the CO2,a coverage was found to
be very small (10-4) at 380 Torr CO2. It is seen in Figure 5a
that rf0 comes to saturation at 200 Torr H2 and 343 K. The
hydrogen coverage is thus responsible for the nonlinear behavior
shown in Figure 5a. We thus tried to analyze the kinetics of
2
2
2
k-1k-2
k1
1 +
P1H/2
(
)
2
k-1
At ΘHCOO ) 0
rf0 )
k1k2k3
1
k1
P1H/2PCO
(13)
2
2
2
k-1k-2
1 +
P1H/2
(
)
2
k-1
The applicability of eq 13 was experimentally evaluated by the
pressure dependence of the initial formation rate. Figures 7a
and 7b show calculated (solid line) and experimental (plots) rf0

Synthesis and Decomposition of Formate Species
J. Phys. Chem. B, Vol. 105, No. 7, 2001 1359
Figure 5. Dependence of the initial formation rate of formate on (a) H2 pressure dependence at PCO ) 380 Torr (b) CO2 pressure dependence at
2
PH ) 380 Torr.
2
rf′ ) k3′P
Θ
CO2 Ha
k1k3′
1 - 2ΘHCOO
)
P1H/2PCO
(16)
(17)
2
2
k-1
k1
1 +
P1H/2
2
k-1
At ΘHCOO ) 0
k1k3′
1
rf0′ )
P1H/2PCO
2
2
k-1
k1
1 +
P1H/2
2
k-1
The major difference between eqs 13 and 17 is the order in (1
+ (k1/k-1)P1/2). In the L-H mechanism, the adsorbed hydro-
H2
gen hinders both H2 dissociative adsorption and CO2 adsorption
so that the order is second in (1 + (k1/k-1)P1/2). Similarly to
H2
the calculation of k3, k3′ was calculated from an experimentally
measured rf0 under the pressure condition of CO2/H2 ) 380
Torr/380 Torr. As shown in Figure 7, the experimental data
were in fairly good agreement with the calculated rf0′ (broken
line) at both different H2 and CO2 pressures. The kinetic analysis
of formate synthesis thus suggested the E-R type mechanism,
which will be further discussed in section 4.1.
3.2. Formate Decomposition. The isothermal decomposition
of formate on Cu(111) was carried out in UHV at 358-403 K.
The formate was synthesized from CO2 and H2 at 343 K, and
the initial coverage was 0.25. The following rate equation can
be obtained for the formate decomposition if the decomposition
is first order in formate coverage.
Figure 6. Equilibrium coverage of (a) Ha and (b) CO2,a at 343 K.
for the dependences of H2 and CO2 pressures. In the calculation
33-35
of rf0, literature values of k1, k-1 and k-2
were used, and
k2 was calculated assuming that the sticking probability of CO2
was unity. The value of k3 in eq 13 was determined from the
measured rf0 at CO2/H2 ) 380 Torr/380 Torr using k1, k-1, k2,
and k-2. Finally, the calculated rf0 (solid line) was thus obtained
at various H2 pressures using eq 13. As shown in Figure 7, the
measured rf0 at different H2 pressures were in disagreement with
the calculated rf0, although good agreement between the
calculation and measurement of rf0 was obtained for CO2
pressure dependence.
dΘ
rd ) -
HCOO ) kdΘHCOO
(18)
(19)
dt
We then tried to calculate rf0 assuming an E-R type
mechanism in which a gaseous CO2 molecule directly reacts
with an adsorbed hydrogen atom on Cu(111). The reaction is
expressed by the following equations.
sat
HCOO
logΘHCOO ) -kdt + log Θ
sat
HCOO
Here, Θ
is the saturation coverage of formate (0.25).
Figure 8 shows log ΘHCOO as a function of reaction time. The
linearity indicates that the decomposition rate is first order in
formate coverage in agreement with the literature data on
Cu(110) measured by TPD.36 The rate constant kd was then
obtained from the slope. At 373 K, for example, the rate constant
1
2
H2,g + / h Ha
(14)
(15)
CO2,g + Ha f HCOOa
was 1.50 × 10-4 sec-1
.
Figure 9 shows Arrhenius plots of the initial decomposition
rate (rd0) derived from the data shown in Figure 8, as well as
our previous data for Cu(110)7 and literature data of Cu(100).20
The formation rate of formate is then expressed by the following
equation.

1360 J. Phys. Chem. B, Vol. 105, No. 7, 2001
Nakano et al.
Figure 7. Dependence of the initial formation rate of formate on (a) H2 pressure (PCO ) 380 Torr constant) (b) CO2 pressure (PH ) 380 Torr
2
2
constant). b, 9, 2: experimental data, solid lines: calculation data assuming an L-H mechanism, dashed lines: calculation data assuming an E-R
mechanism.
Figure 8. Logarithm of the formate coverage as a function of time in
the decomposition of formate species synthesized by CO2 hydrogenation
on Cu(111).
Figure 10. Dependence of the initial decomposition rates of the formate
species on H2 pressure.
comparable with those measured on Cu(111) by XPS.10 The
activation energies on Cu(110) (145.2 ( 7.2 kJ mol-1) and on
Cu(100) (155 kJ mol-1) are greater than that on Cu(111),
showing that the decomposition of formate on Cu surfaces is
structure-sensitive. This is in contrast to the structure-insensitive
character of the formate synthesis, which is discussed in section
4.1.
We accidentally found that the decomposition rate of formate
significantly varied in the presence of hydrogen. Figure 10
shows the dependence of the initial decomposition rate on H2
pressure at 343, 353, and 363 K. Interestingly, the decomposition
rate steeply increased with H2 pressure and then sharply
decreased. At 363 K, the maximum rate at 380 Torr was 17
times greater than that in the absence of H2. We confirmed no
production of formic acid by the hydrogenation of formate. The
effect of hydrogen upon an increase in the decomposition rate
was ascribed to the destruction of the formate chain by adsorbed
hydrogen as discussed in 4.3. At high H2 pressure, coverage of
hydrogen atoms will increase on the formate-covered Cu(111)
surface, which probably hinders the decomposition of formate.
The peaks seen in Figure 10 tend to be shifted to higher H2
pressures at higher temperatures, suggesting that the hydrogen
equilibrium coverage controls the position of the peak maximum
because higher H2 pressure is required at higher temperature to
obtain a certain equilibrium hydrogen coverage. For example,
Figure 9. Arrhenius plots of the initial decomposition rate of formate
species on Cu(111) (b), Cu(110) (9),7 and Cu(100) (dashed line).20
Here, the initial rate was calculated at the saturation coverage
of 0.25 using the rate constant. As shown in Figure 9, it was
found that the activation energy was different depending on the
orientation of the Cu surface. The activation energy and the
preexponential factor of kd were determined to be 107.9 ( 2.8
kJ mol-1 and 1.87 × 1011 sec-1, respectively, which are

Synthesis and Decomposition of Formate Species
J. Phys. Chem. B, Vol. 105, No. 7, 2001 1361
Figure 11. Arrhenius plots for the initial decomposition rate of the
formate species on Cu(111). (b: decomposed in H2 380 Torr, 9:
decomposed in a vacuum).
Figure 13. Arrhenius plots for the initial decomposition rate of the
formate species on Cu(110). The synthesized formate species form CO2
and H2 were decomposed in a vacuum (solid line) and in 380 Torr H2
380 (O). The formate prepared by adsorption of formic on a clean
Cu(110) surface was decomposed in a vacuum (9, broken line).
ponential factors of the decomposition rate constant depending
on the preparation method of the formate and the presence of
H2. The reason for the difference was ascribed to the adsorption
structure of formate on Cu(111) previously observed by STM
as discussed in section 4.2, in which the OCO vibration of
formate toward the surface was regarded to be responsible for
the preexponential factor of the decomposition rate constant or
the frequency factor. Similarly, difference in the decomposition
rate of formate has been reported for Cu(110) with and without
preadsorbed oxygen.23
Figure 13 shows Arrhenius plots of the initial decomposition
rate of formate on Cu(110). The decomposition rate of the
synthesized formate in H2 380 Torr at 383 K (1.01 × 10-4
molecules site-1 sec-1) was comparable with that in a vacuum
at 383 K (4.35 × 10-5 molecules site-1 sec-1). In contrast to
Cu(111), no significant promotion by H2 in the decomposition
kinetics was observed for the synthesized formate. The activation
energy of the decomposition of the formate prepared from
formic acid on a clean Cu(110) surface was determined to be
156.8 kJ mol-1. The activation energy and the absolute
decomposition rate were comparable with those for the syn-
thesized formate. On Cu(110), no significant difference in the
decomposition kinetics of formate was observed depending on
the presence of H2 and the preparation method. Possibly, the
difference in the kinetics is due to the added formate chain
suggested by the previous STM data as discussed in section
4.3.
3.3. Equilibrium Formate Coverage. Figure 14 shows the
equilibrium coverage of formate on Cu(111) under hydrogena-
tion of CO2 at 380 Torr CO2/380 Torr H2. As shown by the
plots, the measured formate coverage was saturated at ΘHCOO
) 0.25 up to 353 K. The coverage steeply decreased above
353 K and was below a detection limit at 453 K. The solid and
broken lines are the calculated formate coverage based on the
measured kinetics for the formate synthesis (3.1.) and the
formate decomposition (3.2.) using kd′ and kd, respectively,
which are the rate constants of the formate decomposition
measured in the presence of 380 Torr H2 and in UHV. The
calculated equilibrium coverage using kd′ is smaller than that
calculated using kd because the decomposition rate was much
greater in the presence of H2 compared to that in UHV as
described in 3.2. The experimental data were in fairly good
agreement with the calculated coverage using kd′. That is, one
Figure 12. Arrhenius plots for the initial decomposition rate of the
formate species on Cu(111). The formate prepared by adsorption of
formic acid on Cu(111) (b) and O/Cu(111) (9) decomposed in a
vacuum and the formate synthesized CO2 and H2 decomposed in a
vacuum (long dashed line) and in H2 380 Torr (dashed line).
equilibrium coverage of hydrogen on Cu(111) in the absence
of formate species can be estimated to be 0.54 at 380 Torr H2
and 363 K.
We then compared the activation energies of the formate
decomposition in UHV and in the presence of 380 Torr H2.
Figure 11 shows the Arrhenius plots of the initial decomposition
rate at ΘHCOO ) 0.25. The activation energy of the decomposi-
tion in 380 Torr H2 (118.4 ( 7.7 kJ mol-1) was found to agree
with that in UHV (107.9 ( 2.8 kJ mol-1), although the
decomposition rate was very different, indicating that only the
preexponential factor of the rate constant was different. The
preexponential factors were determined to be 1.25 × 1014 and
1.87 × 1011 sec-1 for the decompositions in 380 Torr H2 and
in UHV, respectively. Very interestingly, as shown in Figure
12, the decomposition of the formate species prepared by
adsorption of formic acid over an Oa-preadsorbed Cu(111)
(2HCOOH + Oa f 2HCOOa + H2O) in UHV showed the same
kinetics as the decomposition kinetics of the synthesized formate
in 380 Torr H2. On the other hand, the decomposition kinetics
of the formate prepared by adsorption of formic acid on a clean
Cu(111) was same as that in the absence of H2 for the
synthesized formate. In summary, the results described above
indicate that two types of formate exist with different preex-

1362 J. Phys. Chem. B, Vol. 105, No. 7, 2001
Nakano et al.
Figure 16. Model of CO2 insertion into a Cu-H bond of CuH(PH3)2
Figure 14. Equilibrium coverage of formate species on Cu(111) under
hydrogenation of CO2 at 380 Torr CO2/380 Torr H2. b: experimental,
dashed line; calculation using decomposition rate constant in UHV,
solid line: calculation using decomposition rate constant in H2 380 Torr.
or CuH(PH3)3 complex proposed by Sakaki et al.40
al.37 have found that CO2 inserts into a Cu-H bond in
[HCuPPh3]6, leading to the formation of [Ph3P]2CuCOOH by
IR. Sakaki et al.40,41 carried out an ab initio MO study of the
insertion of CO2 into a metal-hydride bond of [RhH(PH3)3] and
[CuH(PH3)3]. They have reported that the CO2 insertion was
significantly exothermic and that its activation barrier was
estimated to be rather small (14.7 kJ mol-1), suggesting that
the CO2 insertion into the Cu(I)-H bond was facile. Figure 16
shows the model of the insertion of CO2 into the Cu-H bond
reported in ref 40, in which Sakaki et al. described that “the
η1-C-coordinated CO2 is significantly distorted, which pushes
down the CO2π* orbital energy.” As a result, “CO2 interacts
with the H ligand through the charge-transfer interaction from
the H ligand to CO2 and the electrostatic interaction between
Cuδ+ and Oδ- atoms. Additionally, the charge-transfer interac-
tion from O to Cu contributes to the Cu-O bond formation in
the last stage of the reaction.” The proposed E-R type
mechanism for the formate synthesis on Cu(111) based on the
kinetic analysis here is analogous to the insertion of CO2 into
the Cu-H bond in the [HCuPPh3]6 complex in terms of the
direct reaction of CO2 and the Cu-H bond without CO2
adsorption on Cu. The E-R type mechanism may further
explain the structure-insensitive character observed for the
formate synthesis on Cu shown in Figure 4. That is, because
the direct reaction of CO2 with the Cu-H bond occurs at a
local site on the Cu surface, the kinetic factors such as activation
energy may be insensitive to the configuration of the surrounding
Cu atoms. Very recently, we have obtained preliminary results
to support the E-R type mechanism, in which the rate of the
formate synthesis by the reaction of heated CO2 molecules with
a hydrogen-preadsorbed Cu(111) surface increased with increas-
ing CO2 gas temperature while maintaining the Cu(111) surface
at room temperature. The results are in preparation for publica-
tion. We are continuing this kind of experiment by varying the
surface temperature and the CO2 pressure in order to obtain
solid evidence for the E-R type mechanism.
Figure 15. Equilibrium coverage of formate species on Cu(110) under
hydrogenation of CO2 at 380 Torr CO2/380 Torr H2. b: experimental;
solid line: calculation using decomposition rate constant in UHV.
should be careful of the effect of coexisting gases when
discussing the microscopic kinetics of surface elementary steps.
At a typical reaction temperature of 523 K for methanol
synthesis, the formate coverage is as low as 1.3 × 10-3 in Figure
14. However, the coverage will be increased at higher pressures.
In comparison with Cu(111), the equilibrium formate cover-
age on Cu(110) is shown in Figure 15. The measured coverage
was in good agreement with that calculated based on the kinetics
of formate synthesis and formate decomposition shown in
Figures 4 and 13. It was found that the equilibrium coverage
on Cu(110) was much greater than that on Cu(111) at 400∼500
K, which was mainly due to the absence of the promotion effect
on the decomposition rate by H2. Thus, it is important to clarify
that the promotion effect of H2 on the decomposition should
appear only for Cu(111). That is the same question raised
regarding the results of Figures 11 and 13, which is discussed
in section 4.3.
In contrast to the formate synthesis, the formate decomposi-
tion on the Cu surfaces showed a structure-sensitive character
as described in section 3.2. We propose a model to explain this
unique character with respect to structure sensitivity as shown
in Figure 17. The vibration of the OCO molecular plane in a
bidentate formate toward the Cu surface is required in order to
overcome the transition state (TS) for the formate decomposi-
tion. The surrounding Cu atom thus takes part in the decom-
position process so that the position of the neighboring Cu atoms
affects the kinetics of the decomposition, leading to the structure-
4. Discussion
4.1. Structure Sensitivity. In organometallic chemistry, it
is well-known that formate is synthesized by the reaction of
CO2 with transition metal complexes such as [HCuPPh3]6,37
[Ph2P(CH2)2PPh2]2RhH,38 or (PMePh2)3Cu(BH4).39 Beguin et

Synthesis and Decomposition of Formate Species
J. Phys. Chem. B, Vol. 105, No. 7, 2001 1363
preparation method was suggested by the IRA spectra of
formate. The peak position of νs(OCO) vibration for the formate
on Cu(111) synthesized by CO2 hydrogenation slightly shifted
to high frequencies (2 cm-1, 1352 ∼ 1354 cm-1) with increasing
formate coverage. However, significant peak shifts of 18 cm-1
(1342 ∼ 1360 cm-1) were observed depending on formate
coverage for the formate species prepared from formic acid and
O/Cu(111). The difference in IRAS can be also explained by
the adsorption structure of formate in the following. For the
chain structure, the local environment of the formate does not
change with formate coverage because the chain structure starts
to grow at very low formate coverage. On the other hand, for
the formate from formic acid on O/Cu(111), the distance to the
nearest neighbor formate was continuously decreased with
increasing formate coverage, leading to a continuous change in
the local environment of formate. The repulsive force between
formate species thus continuously varies as a function of the
formate-formate distance or formate coverage, leading to the
change in the IRA spectra of formate. The more interesting
difference seen in IRAS was the sensitivity of IR peak intensity.
In the XPS measurements of formate buildup, the saturated
formate coverage was 0.25 and 0.33 for the formate synthesized
from CO2 and H2 on Cu(111) and the formate prepared from
formic acid on O/Cu(111), respectively. That is, both saturated
coverages were not very different. In similar buildup measure-
ments of formate by IRAS, however, the intensity of the IR
peak for the synthesized formate at saturation was one-third of
that for the saturated formate prepared from formic acid on
O/Cu(111). The difference in the sensitivity was reproducibly
observed. The decrease in intensity for the synthesized formate
forming the formate chain was explained by the screening effect
known in IRAS measurement in which closely located formate
in a chain screens the neighboring dipole moment. In Figure
14, the equilibrium formate coverage below 360 K is consistent
with the synthesis/decomposition kinetics using kd (for the
decomposition of the synthesized formate), whereas above 360
K the equilibrium coverage is in accordance with the kinetics
using kd′ (for the decomposition of the formate prepared from
formic acid as well as the decomposition of the synthesized
formate in H2). Apparently, the equilibrium property seems to
change at 360 K, which is probably due to a change in the
structure of formate, i.e., chain and nonchain structures.
Figure 17. Model of formate chemistry on Cu.
sensitive character. We thus infer that the path in the reaction
coordinate to overcome the barrier to the formate synthesis is
structure-insensitive, whereas the path going through TS toward
the decomposition is structure-sensitive. The difference in the
activation energy for the decomposition may be related to the
adsorption energy of the formate species, although the detailed
microscopic mechanism is ambiguous.
4.2. Kinetics Depending on the Ordered Structure of
Formate. As shown in Figure 12, the decomposition rate of
formate prepared from formic acid on O/Cu(111) was greater
than that for the formate synthesized from CO2 and H2.
Interestingly, the apparent activation energy for the decomposi-
tion in both cases was almost identical, whereas the preexpo-
nential factor of the rate constant was very different between
those two formate species. These phenomena were also con-
firmed by our XPS studies.10 We then studied the adsorption
structure of formate by STM in order to examine the difference
in the decomposition kinetics of formate.13,19 It was found that
the synthesized formate adsorbed in a chain structure with the
distance to the nearest neighbor formate being twice that of the
nearest neighbor Cu. Even at low formate coverage, a single
formate chain was observed by STM, indicating that an
anisotropic attractive interaction operated between the formate
species. This further means that the attractive interaction should
be quite strong and the barrier of diffusion is remarkably high.
On the other hand, no chain structure was observed for the
formate prepared from formic acid on O/Cu(111). However, (4
× 4) and (3 × 7/2) structures appeared, where the formate
species adsorbed more dispersedly. The spacing between formate
species was varied from four to three times that of the nearest
neighbor Cu atoms in contrast to the synthesized formate with
the spacing twice that of the nearest neighbor Cu atoms
independent of formate coverage. It is possible that the
decomposition of formate proceeds via the OCO plane vibration
as shown in Figure 17, which has been proposed by Bowker et
al. for acetate decomposition on Rh(111).46 In this case, the
frequency of the plane vibration should significantly influence
the preexponential factor of the rate constant of the formate
decomposition. The observed significant decrease in the pre-
exponential factor with no variation in the activation energy is
thus possibly due to the plane vibration of formate. That is, the
OCO plane vibration may be hindered by the attractive
interaction between the nearest neighbored formate species in
a chain. Such an attractive interaction is absent between formate
species prepared from formic acid. The difference in the
adsorption structure thus reflects the reaction kinetics of the
formate decomposition.
On Cu(110), no significant difference depending on the
presence of H2 or the preparation method of formate was
observed in the decomposition kinetics as shown in Figures 13
and 15. The peak intensity of formate on Cu(110) was
comparable with that for the formate prepared from formic acid
on O/Cu(111), suggesting the absence of the screening effect.
This is also supported by a peak shift of the νs(OCO) vibration
(1352 ∼ 1362 cm-1) at formate coverages between 0 and 0.25
for both formate species prepared from formic acid and
synthesized by CO2 hydrogenation on Cu(110). The peak shift
of 10 cm-1 has also been reported in the literature for formate
species prepared from formic acid on Cu(110).26 The absence
of screening and the peak shift of the νs(OCO) vibration for
formate on Cu(110) indicate the absence of the strong attractive
interaction seen for the formate chain on Cu(111). However,
Leibsle and Haq have reported a chainlike structure of formate
on Cu(110).42,47,48 We think that the interaction between formate
in a chain is very different on Cu(111) and Cu(110).
4.3. Possibility of Added Structure. As described in sections
3.2 and 4.2, several unique features of the decomposition
kinetics were observed only for Cu(111). Here, we discuss the
mechanism. First, the questions are summarized as follows: (i)
The difference in the adsorption structure depending on the

1364 J. Phys. Chem. B, Vol. 105, No. 7, 2001
Nakano et al.
why is the decomposition rate exceptionally high for the formate
prepared from formic acid on O/Cu(111) as well as the
decomposition of the synthesized formate in H2? and (ii) why
was the unique different kinetic behavior observed only for
Cu(111)? A possible explanation for the two questions is that
the added-row-type reconstruction may occur upon chain
formation for all formate species except for the formate prepared
from formic acid and O/Cu(111). As mentioned above, on
Cu(110), the chain structure of formate has been reported for
the formate prepared by adsorption of formic acid on an oxygen-
covered Cu(110) surface.42 The formate chains may be an added-
row-type reconstruction such as the well-known (-O-Cu-O-)
added rows formed on Cu(110) upon O2 adsorption. It has been
reported that Cu atoms at the step edges are removed and
transported onto the terrace to create the -O-Cu-O- added
chain on Cu(110).43 This kind of mass transport may occur in
the formation of the formate chain on Cu(111) because the
corrugation of the STM image for the synthesized formate
(10-20 Å) was much greater than that for the formate prepared
from formic acid on O/Cu(111) (1-4 Å). As for the formate
prepared from formic acid on clean Cu(111), the corrugation
was also high (10-20 Å). This is inconsistent with the same
character of the decomposition kinetics, the IRAS intensity and
the peak shift between the synthesized formate and the formate
prepared from formic acid on the clean Cu(111) surface. The
substrate copper atom on Cu(111) might easily migrate by
reacting with hydrogen atoms in the presence of high-pressure
H2, leading to a stable added-row-type formate chain. That is,
the driving force to create the formate chain different from the
structure of formate prepared by formic acid on O/Cu(111) is
related to the presence of adsorbed hydrogen in equilibrium with
high-pressure H2 gas (380 Torr). The (111) planes of Cu are
the most stable compared to the other planes; therefore, the
surface copper atoms required for the added-row formation
would hardly migrate in the absence of adsorbed hydrogen
atoms. In the presence of high-pressure H2 gas, adsorbed
hydrogen probably breaks the added-row chain by reacting with
Cu atoms. It is thus possible that the attractive interaction
between nearest neighbor formate species is vanished and the
decomposition rate becomes the same value as the formate
prepared from formic acid on O/Cu(111).
4.4. Comparison with Cu/SiO2 Powder Catalyst. We have
reported the synthesis and decomposition of formate on a Cu/
SiO2 powder catalyst.18 The apparent activation energy of the
formate synthesis on Cu/SiO2 was 58.8 kJ mol-1, which was
comparable with those obtained for Cu(111), Cu(110), and
Cu(100) as described in section 3.1. As for the decomposition,
the activation energy and preexponential factor for Cu/SiO2 were
115.7 kJ mol-1 and 5.38 × 10-11 s-1, respectively. These values
were in good agreement with those obtained for Cu(111) rather
than Cu(110) and Cu(100). Thus, the surface of Cu particles in
the Cu/SiO2 powder catalyst is expected to comprise the most
densely packed Cu(111) plane. Furthermore, the promotion
effect of hydrogen in the formate decomposition was also
observed on the Cu/SiO2 powder catalyst.18 The promotion curve
measured for Cu/SiO2 was very similar to that for Cu(111)
shown in Figure 10. It is thus concluded that the Cu(111) surface
can be regarded as a model of the Cu/SiO2 powder catalyst.
an Eley-Rideal type mechanism in which an adsorbed hydrogen
reacts with a gaseous CO2 molecule. (3) The decomposition of
formate on copper in UHV is structure-sensitive in contrast to
the formate synthesis. (4) Measured coverage of formate on
Cu(111) and Cu(110) in equilibrium with 1 atm of CO2/H2 is
in good agreement with that calculated by the kinetics of the
formate synthesis and the formate decomposition. (5) The
decomposition kinetics of formate on Cu(111) uniquely depend
on the preparation method or the coexistence of H2, where the
preexponential factor of the rate constant is increased only while
the activation energy was identical. The difference in the
decomposition kinetics is probably due to the surface structure
of formate, diffusion of Cu atoms, or H-induced reconstructing
of Cu(111). (6) The selectivity to form a chain or nonchain
structure is probably controlled by the mass transport of Cu
atoms at the step edges, which is further thermodynamically
controlled by temperature and coexisting adsorbates. (7) The
Cu(111) surface can be regarded as a model of a Cu/SiO2
powder catalyst in terms of the kinetics of formate synthesis,
formate decomposition, and the unique character of the promo-
tion on the formate decomposition.
Acknowledgment. This research was supported by the
Grant-in Aid for Scientific Research on Priority Areas “Mo-
lecular Physical Chemistry” from the Ministry of Education,
Science, Sports, and Culture.
References and Notes
(1) Rodriguez, J. A.; Goodman, D. V. Surf. Sci. Rep. 1991, 14, 1.
(2) Campbell, C. T. AdV. Catal. 1989, 36, 1.
(3) Nakamura, J.; Nakamura, I.; Uchijima, T.; Kanai, Y.; Watanabe,
T.; Saito, M.; Fujitani, T. Catal. Lett. 1995, 31, 325.
(4) Fujitani, T.; Nakamura, I.; Watanabe, T.; Uchijima, T.; Nakamura,
J. Catal. Lett. 1995, 35, 297.
(5) Nakamura, J.; Nakamura, I.; Uchijima, T.; Kanai, Y.; Watanabe,
T.; Saito, M.; Fujitani, T. J. Catal. 1996, 160, 65.
(6) Nakamura, J.; Kushida, Y.; Choi, Y.; Uchijima, T.; Fujitani, T. J.
Vac. Sci. Technol. 1997, A15, 1568.
(7) Nakamura, I.; Nakano, H.; Fujitani, T.; Uchijima, T.; Nakamura,
J. J. Vac. Sci. Technol. 1999, A17, 1592.
(8) Nakamura, J.; Nakamura, I.; Uchijima, T.; Watanabe, T.; Fujitani,
T. Studies Surf. Sci. Catal. 1996, 101, 1389.
(9) Nakamura, I.; Nakano, H.; Fujitani, T.; Uchijima, T.; Nakamura,
J. Surf. Sci. 1998, 402-404, 92.
(10) Nishimura, H.; Yatsu, T.; Fujitani, T.; Uchijima, T.; Nakamura, J.
J. Mol. Catal. 2000, A155, 3.
(11) Fujitani, T.; Nakamura, I.; Uchijima, T.; Nakamura, J. Surf. Sci.
1997, 383, 285.
(12) Nakamura, I.; Fujitani, T.; Uchijima, T.; Nakamura, J. Surf. Sci.
1998, 400, 387.
(13) Fujitani, T.; Choi, Y.; Sano, M.; Kushida, Y.; Nakamura, J. J. Phys.
Chem. B 2000, 104, 1235.
(14) Fujitani, T.; Nakamura, I.; Ueno, S.; Uchijima,; Nakamura, J. Appl.
Surf. Sci. 1997, 121/122, 583.
(15) Nakamura, J.; Uchijima, T.; Kanai, Y.; Fujitani, T. Catal. Today
1996, 28, 223.
(16) Nakamura, I.; Fujitani, T.; Uchijima, T.; Nakamura, J. J. Vac. Sci.
Technol. 1996, A14, 1464.
(17) Nakamura, J.; Nakamura, I.; Uchijima, T.; Watanabe, T.; Fujitani,
T. 11th Int. Congr. Catal. Stud. Surf. Sci. Catal. 1996, 101, 1389.
(18) Yatsu, T.; Nishimura, H.; Fujitani, T.; Nakamura, J. J. Catal. 2000,
191, 423.
(19) Fujitani, T.; Nakamura, J. Appl. Catal. 2000, A 191, 111.
(20) Taylor, P. A.; Rasmussen, P. B.; Ovesen, C. V.; Stoltze, P.;
Chorkendorff, I. Surf. Sci. 1992, 261, 191.
(21) Chorkendorff, I.; A. Taylor, P.; Rasmussen, P. B. J. Vac. Sci.
Technol. 1992, A10, 2277.
(22) Taylor, P. A.; Rasmussen, P. B.; Chorkendorff, I. J. Chem. Soc.
Faraday Trans. 1995, 91, 1267.
5. Conclusions
(23) Bowker, M.; Rowbotham, E.; Leibsle, F. M.; Haq, S. Surf. Sci.
1996, 349, 97.
(24) Leibsle, F. M.; Haq, S.; Frederick, B. G.; Bowker, M.; Richardson,
N. V. Surf. Sci. 1995, 345, L1175.
(25) Leibsle, F. M. Surf. Sci. 1998, 401, 153.
From this work we draw the following conclusions. (1) The
formate synthesis on copper at 1 atm is structure-insensitive in
terms of the absolute formation rate as well as the activation
energy. (2) The kinetics of formate synthesis is consistent with

Synthesis and Decomposition of Formate Species
J. Phys. Chem. B, Vol. 105, No. 7, 2001 1365
(26) Hayden, B. E.; Prince, K.; Woodruff, D. P.; Bradshaw, M. A. Surf.
Sci. 1983, 133, 589.
(27) Sexton, B. A.; Hughes, A. E.; Avery, N. R. Surf. Sci. 1985, 155,
366.
(37) Beguin, B.; Denise, B.; Sneeden, R. P. A. J. Organo. Chem. 1981,
208, C18.
(38) Burgemeister, T.; Kastner, F.; Leitner, W. Angew. Chem. Int. Ed.
Engl. 1993, 32, 739.
(28) Dubois, L. H.; Ellis, T. H.; Zegarski, B. R.; Kevan, S. D. Surf. Sci.
1986, 172, 385.
(29) Crapper, M. D.; Riley, C. E.; Woodruff, D. P.; Puschmann, A.;
Haase, J. Surf. Sci. 1986, 171, 1.
(30) Sotiropoulos, A.; Milligan, P. K.; Cowie, B. C. C.; Kadodwala,
M. Surf. Sci. 2000, 444, 52.
(31) Crapper, M. D.; Riley, C. E.; Woodruff, D. P. Surf. Sci. 1897, 184,
121.
(39) Bianchini, C.; Ghilardi, C. A.; Meli, A.; Midollini, S.; Orlandini,
A. Inorg. Chem. 1985, 24, 924.
(40) Sakaki, S.; Ohkubo, K. Inorg. Chem. 1989, 28, 2583.
(41) Sakaki, S.; Musashi, Y.; J. Chem. Dalton Trans. 1994, 3047.
(42) Haq, S.; Leibsle, F. M. Surf. Sci. 1997, 375, 81.
(43) Jensen, F.; Besenbacher, F.; Legsgard, E. L.; Stensgaard, I. Phys.
ReV. 1990, B 41, 10233.
(44) Sexton, B. A. Surf. Sci. 1979, 88, 319.
(32) Woodruff, D. P.; McConville, C. F.; Kileoyne, A. L. D.; Lindner,
T.; Somers, J.; Surman, M. Surf. Sci. 1988, 201, 228.
(33) Campbell, J. M.; Campbell, C. T. Surf. Sci. 1991, 259, 1.
(34) Anger, G.; Winkler, A.; Rendulic, K. D. Surf. Sci. 1989, 220, 1.
(35) Rasmussen, P. B.; Taylor, P. A.; Chorkendorrf, I. Surf. Sci. 1992,
269/270, 352.
(45) Bowker, M.; Haq, S.; Holroyd, R.; Parlett, P. M.; Poulston, S.;
Richardson, N. J. Chem. Soc., Faraday Trans. 1996, 92, 4683.
(46) Li, Y.; Bowker, M. Surf. Sci. 1993, 285, 219.
(47) Jones, A. H.; Poulston, S.; Bennett, R. A.; Bowker, M. Surf. Sci.
1997, 380, 31.
(48) Stone, P. S.; Poulston, S.; Bennett, R. A.; Price, N. J.; Bowker, M.
Surf. Sci. 1998, 418, 71.
(36) Ying, D. H. S.; Madix, R. J. J. Catal. 1980, 61, 48.