Journal of MoLecular Catalysis, 65 (1991) 337358 337

Review

The bonding, structure and reactions of CO2 adsorbed on clean

and promoted metal surfaces

Frigyes Solymosi
Reaction Kinetics Research Group of the Hungarian Academy of Sciences and Institute
of Solid State and Radiochemistry, University of Szeged P.O. Box 105, H-6701 Szeged
(Hungary)

Abstract

The characteristics of the adsorption and reactions of CO2 on Rh, Pd, Pt, Ni, Fe, Cu,
Re, Al, Mg and Ag metals are discussed with particular emphasis on the activation of
the CO2 molecule. Strong spectroscopic evidence is presented for the formation of
negatively charged COa-, which - depending on the nature of the metal - may dissociate
into CO and 0, or transform into CO3 + CO. The presence of surface adatoms dramatically
influences the adsorption and reactivity of COz. Alkali adatoms increase the binding
energy of adsorbed COa, promote the dissociation and/or the transformation of CO2 into
CO,+O. In the presence of preadsorbed oxygen the formation of carbonate of different
structures predominates.

Introduction
An increasing number of investigations have been concerned with homo-
geneous and heterogeneous catalytic transformations of CO2into more valuable
compounds [l-3] such as carbon monoxide, methanol, formate, oxalate,
organic acids, methane and hydrocarbons. An important observation in this
field is that CO2 is possibly the main source of carbon in the synthesis of
methanol from H,+CO+CO, mixtures [4, 51. Carbon dioxide can also be
converted into useful forms of organic compounds by coupling with an
organic substrate.
Any reaction of CO2 requires its activation by a catalyst. The chemisorption
and reactions of CO2 on oxide surfaces are well-studied processes; the
structure and bonding of adsorbed CO2 and of the various carbonates formed
have been established. Much less information is available on the interaction
of CO2 with metal surfaces. Several modern tools of surface science have
recently been applied to the study of the adsorption of CO2 on metal single
crystals, and attention has also been paid to evaluation of the effects of
surface adatoms or promoters in the activation of COa.
This article surveys the present state of knowledge concerning the bonding,
structure and reactions of CO2 adsorbed on clean and promoted metal
surfaces.

0304-5102/91/$3.50 0 Elsevier Sequoia/Printed in The Netherlands

Clean sm$ace
The interaction of COa and Rh metal has been the subject of a number
of publications, because of the controversy concerning the nature of the
interaction and particularly the dissociation of CO2 on clean Rh surfaces
under UHV conditions.
In the early studies, the adsorption of CO2 was claimed to be weak and
associative on Rh films [6], polyc~s~lline Rh [ 71 and puma-supposed
Rh [S]. Somorjai et al. (g-131, however, found that CO2 does chemisorb
and dissociate on Rh foil and on several faces of Rh single crystals (Rh(l1 l),
(100) and (331)) at 300 K, but this requires a higher CO2 exposure. They
detected the CO formed by means of LEED, EELS (vibration) and thermal
desorption measurements. The calculations of Weinberg [ 141, based on the
available th~~od~amic and kinetic information, showed that the probability
of dissociative chemisorption of CO2 at low pressure ( N 10V8 torr) and at
room temperature is of order of 10-15, i.e. some 14 orders of magnitude
lower than reported previously [g-13]. The rate of dissociation could be
significant only above 300 K and a CO2 pressure of approximately 1 atm.
The probab~~ of the dissociation of COa on Rh( 111) was also calculated
by Goodman et al. [ 151 from an analysis of the kinetic data obtained for
the Hz + COa reaction, and a value of lo- l1 was given for 300 K, which is
consistent with the value of Weinberg. Dubois and Somorjai [ 161 criticized
the conclusions of Weinberg, stating that they were “based on oversimplified
calculation using incorrect kinetic and the~odyn~i~ parameters”. They
argued #at the failure to observe the dissociation of CO2 may simply be
an artifact of the adsorption conditions or of sample preparation.
As the dissociation of CO2 has been experimentally observed [lo-131
several factors causing or preventing the dissociation can be considered,
such as the presence of adsorbed oxygen (which may inhibit the dissociation),
the presence of undetected impurities (which may promote the dissociation)
and the surface irregularities (which could vary from sample to sample and
could function as active sites for the CO2 reaction).
Experiments carried out under carefully controlled conditions on po-
lycrystalline Rh foil and Rh(ll1) single crystal surfaces did not provide
evidence of a strong dissociative adsorption of CO2 117, 181. The adsorption
of COZ was found to be weak and nondissociative on both surfaces; the
sticking coefficient was calculated to be as low as 0.01 at 100-110 K. At
low CO2 exposures, the desorption of COa was characterized by a peak
temperature of Z”,= 244 K, which decreased to 177 K with increasing COa
exposure. An activation energy of 61.2 kJ mol-’ was calculated for the
desorption of CO,. Following adsorption at lower temperature, 95 K, a well-
resolved desorption peak also developed at 115 K; the activation energy for
desorption was 28 kJ mol-’ [ 191. Subsequent TDS and EELS measurements
revealed no indication of the formation of adsorbed CO, which is strongly
bonded to Rh and can easily be detected by both methods. When the
 339

adsorption of COz (180 L) at 3X lo-’ torr was performed at 300 K, no

observable changes were found in the LEED, AES, EELS or TDS.
However, when boron impurity was allowed to segregate onto the Rh
surface, strong evidence was obtained of the dissociation of CO2 and the
formation of adsorbed CO (see next section). As boron is a common impurity
in Rh, and is not easy to detect by AES, it seemed highly probable that the
dissociation of COB observed in previous cases was due to the presence of
boron impurity.
The interaction of CO, with a Rh tip has also been investigated by field
emission microscopy (FEM) [ 201. A value of 60 kJ mol- ’ was obtained for
the activation energy of desorption of COz, which is in excellent agreement
with the value determined on the Rh(ll1) face [ 171.
The FEM pattern indicated that COz adsorption is markedly crystal face-
specific. An analysis of the FEM pattern yielded no evidence of the dissociation
of COz on the (11 l), (100) surfaces. However, signitlcant changes occurred
above 210 K on the (100) surface with (111) steps, such as (533) and (711):
these regions became dark, which was attributed to the dissociation of COa.
The FEM picture resembled that given by a surface 50% covered with CO.
No indication of the presence of adsorbed oxygen was obtained, however,
which we consider to be an important point in further discussion. In the
interpretation of the results, Hendrickx et al. [20] assumed that CO, is
adsorbed in a monodentate structure on top of a Rh atom, on the (111)
terrace of a stepped surface near a (11 l)-like step, which may facilitate the
dissociation of COz.
These results point to the important role of the heterogeneity of the
surface. The number of these sites, however, must be very limited, as Hendrickx
et al. [20] failed to detect a product of the dissociation, e.g. CO, by TDS
measurements. Since FEM is a very sensitive tool for the detection of a
small amount of adsorption, we may speculate whether the changes observed
in the FEM pattern were caused by the adsorption of CO from the background.
This is not absolutely impossible, as the FEM pattern following CO2 adsorption
was the same as observed for adsorbed CO. This idea is apparently supported
by FEM patterns which provided no indication of the presence of adsorbed
oxygen either, which likewise can easily be detected by FEM. This feature
was explained by oxygen diffusion into the bulk of the Rh. However, this
interpretation is in disagreement with the recent findings and considerations
of Fischer et al. [21], who discounted the diffusion of oxygen into Rh even
at higher temperatures, and explained the phenomenon leading to this
assumption in the former paper 1221 in terms of the presence of near-surface
carbon and CO and their reactions with oxygen.
Although it cannot be denied that surface heterogeneity may play an
important role in the dissociation of adsorbed molecules on clean metal
surfaces, the author of the present paper does not agree with the conclusion
of Hendrickx et al. that this finding would explain the controversy concerning
the dissociation of COz. The dissociation of CO2 has been observed by the
Berkeley group on polycrystalline Rh and on various perfect faces of Rh
340

single crystals [ 10-131, whereas both polycrystaIline Rh and Rh(ll1) in a

clean state were found to be inactive for this process in our measurements
[ 17-191. Further investigations on carefully cleaned stepped surfaces of Rh
single crystals would be useful.
The weak interaction between COa and clean Rh surfaces has recently
been confirmed by photoelectron spectroscopy [23]. The adsorption of COa
on clean Rh(ll1) produced three new lines at 6.6, 10.7 and 12.1 eV in the
UP spectrum. The first of these is due to the lrr g orbital, while the emissions
at 10.7 and 12.1 eV correspond to unresolved ln- n/3a u and 4a g orbitals.
The differences in binding energy in the adsorbed layer on Rh(l1 l), 4.1 and
5.5 eV, approximate weIl the values (3.8 and 5.6 eV) measured for gaseous
COa [ 241, which supports the view that COa adsorbs weakly and molecularly
on Rh(ll1) at 90-100 K, without undergoing a strong distortion of its
molecular electronic state. A similar conclusion was drawn from an analysis
of the XPS spectrum of adsorbed COB. The energy difference (AE= 242.7
eV) between the O(ls) (534.7 eV) and C(ls) (292.0 ev> levels approximates
weII the value determined for undisturbed gaseous COz, AE,=243.6 eV.
When the adsorbed layer was heated to 220-250 K, ail the photoemission
peaks disappeared and the spectrum of the clean surface was restored. There
was no sign of the development of new peaks due to adsorbed CO or 0.

Effects of suyfbce adutcnns

The first detailed investigations on the effects of additives and impurities
were performed on B-contaminated Rh foil [ 17, 181. It was observed that
B, a common impurity in Rh, can segregate to the surface, significantly
changing the surface reactivity. Even if the surface was carefully cleaned
and AES initially showed no impurity on the surface, B could accumulate
on the surface during heating of the sample to _ 650-700 K.
Following the adsorption of CO2 on B-contaminated Rh foil, a clearly
detectable amount of CO was produced, which was not the case for the B-
free surface. At the same time, a new loss feature was observed at 9.4 eV
in the ELS. This was a very stable loss and could be eliminated only above
950 K. This loss was never detected following oxygen adsorption on carefully
cleaned B-free Rh surfaces. The 9.4 eV loss was attributed to the oxygen
bonded to B impurity. This assignment was confirmed by a detailed study
of oxygen adsorption on the B-containing surface, where an intense loss at
9.4 eV appeared even at 159 K and was eliminated again only above 950
K (251. The maximum amount of CO formed on the Rh foil exposed to CO2
at 110 K was 8.0~ 1013 molecules cme2.
The dissociation of CO2 was also established on Rh(ll1) contaminated
with B [ 171. The maximum amount of CO formed was approximately 4.6 X 1013
molecules cmv2.
As concerns the eflect of alkali metal additives on the bonding of
COP,the first study was made on the Rh(ll1) [ 191 and Pd(lOO) [ 261 surfaces.
Preadsorbed K exerted a dramatic influence on the adsorption and reactivity
of CO2 on Rh(ll1). It (i) increased the rate of adsorption of C02, (ii) induced
 341

the formation of strongly bonded CO2 species, and (iii) initiated the dissociation
of COz. At lower K coverages, new adsorption states were formed, which
desorbed at temperatures between 343 and 500 K. At around monolayer K
coverage, a very stable form of CO2 was produced, with Tr- ‘724 K. The
amount of CO gradually increased with increasing K coverage and at the
same time the desorption temperature of CO also increased. At monolayer
coverage, CO desorbed with Tp= 704 K.
An interesting feature of the interaction of CO2 with K-dosed Rh(ll1)
was that not only was the binding energy of CO2 increased, but the K was
also stabilized by adsorbed COz. In the presence of adsorbed CO2 the low-
temperature desorption of K ceased and the desorption occurred in a narrow
peak at the same temperature as that at which the highly stabilized CO and
CO2 desorbed.
The adsorption of CO2 on a K-dosed surface resulted in an increase in
the work function. This increase depended on 6,; it was 1.06 eV at S, = 0.07,
when K exhibits mainly an ionic character, and it was 1.8 eV at monolayer
K coverage, when the K is mainly metallic. CO2 adsorption produced new
losses at 15.0; 11.9 and 8.1 eV, which were stable up to 609 K (8.1 and
11.9 eV losses) and 690 K (15.6 eV loss), respectively.
The development of new losses in the EEL spectra and the work function
changes suggested an electronic interaction between CO2 and the K-dosed
surface, a large negative charge on the chemisorbed CO2 and the possible
formation of a partially negatively charged species, COzs-. These features
were attributed to an enhanced back-donation of an electron from the K-
dosed Rh into an empty r orbital of COz. As a result of the enhanced electron
donation, the binding energy of CO2 to the surface is increased, as exhibited
by the higher desorption temperature. The extended electron donation can
cause a basic change in the bonding and structure of the adsorbed COz.
Whereas in the case of weak CO2 adsorption it was assumed that CO2 is
bonded via an oxygen lone pair, with the molecular bond vertical, on K-
promoted Rh the formation of a metal-carbon bond in the form of a
monodentate (A) or bidentate (B) structure has been proposed:

In the latter form (B), an elongation of the ~te~ct~g C-O bond is expected
with respect to the non-coordinated bond, which may lead to an easier
dissociation of COz.
At higher K coverages, when K exhibits a metallic character, a direct
interaction between CO2 and K was assumed:
K+CO,=K*+ -C02*- (1)
which also leads to formation of the COz- anion radical.
342

Furthermore, it was assumed that COZ- is stabilized by forming a dimer

with neutral COZ:
co,+cq- =coz.coa- (21
or dimerizes into oxalate
2coa- = c2042- (3)
Both species can be tr~sfo~ed into c~bonate-lye species and CO,
at higher temperature. The COz evolution above 700 K (T, = 720 K) and the
coincident thermal desorption with K at that temperature were attributed to
the decomposition of carbonate accelerated by contact with Rh. This route
of ~fo~ation of CO*- is based on the findings of Kafafi et al. [27] in
the study of the interaction of COa with alkali metals.
The results of UPS and XPS studies confirmed the strong interaction
between COZ and K-dosed Rh, and the formation of the COa- species at
and above 131 K [ 231. Let us first consider the results obtained at higher
K coverage, where the photo-em~sion peaks are more intense due to the
higher COZ concentration. In this case, the He II UP spectrum exhibits three
emissions, at 5.2, 8.7 and 10.9 eV. These peaks are distinctly different from
those observed following COa adsorption on a clean Rh surface, and agree
well with those derived theoretically I281 and established experimentally for
bent COa- bonded to a clean Ni( 110) surface 1291. As no signals due to
adsorbed CO and 0 were detected in the photoemission spectra at this
temperature, the above features were attributed to formation of the undis-
sociated COZ- radical anion. The observed peaks were assigned to the (laa,
4Bz), (5a,, lbl, 3bz) and (4ar) orbitals of COZ- , respectively. The appearance
of both the C(ls> and the O(ls) signals at lower energies, in the XPS at
290.5 and 532.8 eV respectively, as compared to neutral COa also indicates
that the electron density in the COZ molecule is increased, which further
supports this conclusion.
The orbitals of the bent COZ- were seen in the UPS and XPS up to
238 K, which signifies appreciable stability on this surface. COB- species
may exist even at higher temperatures: the detection is possibly hindered
by the low intensities and by the development of other overlapping pho-
toemission peaks.
As regards the formation of CO, it is very important to note that there
was no sign of a peak at 530.4 eV for O(ls) up to 550-600 K in the XPS
or at -5.5-6 eV (O(2p) in the UPS, which were indicative of adsorbed
oxygen.
Accordingly, it was concluded that the primary source of the fo~ation
of CO was not the dissociation of negatively charged COZ- . As the peaks
of adsorbed CO and a carbonate-like species developed almost simultaneously,
it seemed appropriate that they were formed in the same surface process.
This could be the disproportionation of the COa- radical anion:
z-
2COZW- = CO, + COS(a) (4)
 343

An alternative pathway, via formation and decomposition of oxalate to

carbonate and CO:

Go*(~>z- = WW2 - + C% (5)

still awaits spectroscopic confirmation.
In the case of low K coverage, 0, = 0.1, both UPS and XPS signals
suggested that the activated COO dissociates completely to CO and 0 even
at 131-179 K:

C&C,>- = CO@, + 0, - (6)

as there was no spectral indication of the formation of any other species
during heating of the coadsorbed layer to high temperature.

Clean suyfbce
The first observation concerning the nature of the adsorption of COz
on Pd under UHV conditions came from a study of the oxidation of CO on
several Pd single crystal faces, which indicated that the adsorption of CO2
is weak and non-dissociative [30, 311. This feature was supported by the
more detailed studies of Solymosi and Berko [26, 321. Exposure of a clean
Pd(lOO) surface to COz up to 180 L at a pressure of 3 x 10e7 torr CO2 at
around 300 K produced no observable changes in the AES and EELS of Pd,
and no desorbing products were detected in subsequent thermal desorption
measurements. The situation was different at an adsorption temperature of
100 K, where a significant uptake of COz was observed. However, this CO2
desorbed at rather low temperature, Tp = 135 K, with an activation energy
of 34 kJ mol- ‘.
A similar weak interaction was observed for the Pdflll) surface f33,
341.

Eflect of surface adatom

Potassium adatoms exerted a qualitatively similar influence on the ad-
sorption of COz as in the case of Rh(l1 l), but a number of differences were
exhibited in the details. The presence of K gradually increased the energy
of binding of COz to the Pd surface and induced new adsorption states. A
striking effect was observed in the cy state. The area of the COa desorption
peak in this weakly bonded state markedly increased with increasing pread-
sorbed K. At &= 0.42, it was larger by a factor of 50 as compared to the
clean surface at the same exposure.
The formation of CO was first observed at 6, = 0.21, with a peak
temperature of 573 K. At &=0.36, another CO desorption state emerged
at T,=663 K.
The amount of CO formed increased considerably with the surface
concentration of K, accompa~ed by a shift to higher peak temperatures.
344

Similarly to Rh(lll), K is greatly stabilized in the adsorbed layer. This is

exhibited above &=0.3, when the more stable state of COa developed:
desorption of K observed in the lower temperature range ceased and K
desorbed in a narrow peak with Tp= 695 K. This corresponds to a binding
energy of 175 kJ mol-‘.
The interaction of CO2 with a K-covered Pd surface was also investigated
through work function measurements at two K concentrations. The exposure
of a clean Pd(lOO) surface to K significantly decreased the work function
of Pd, e.g. by 3.38 eV at &=0.32. When CO2 was adsorbed on a K-dosed
surface, it led to a work function increase of 0.79 eV at 0,=0.1, and 2.20
eV at &=0.42. The final values attained were practically the same in the
two cases. These changes indicate a substantial charge transfer from the K-
dosed Pd to an empty n- orbital of COP.
The interaction of CO2 with the K + Pd( 111) surface has been described
in a manner similar to the case of K-promoted Rh(ll1). The key process
is the formation of the COa- radical anion promoted by K.
It is interesting that a marked influence of K on the adsorption of CO2
was not found by Matsushima [33] on the Pd(ll1) surface. Moreover, when
the Pd(ll1) was fully covered by K, this surface interacted with CO, to a
lesser extent than did K-free Pd. However, Matsushima detected considerable
adsorption and stabilization of COB after oxygen treatment of the K-dosed
Pd surface. In this case, a single peak above 800 K (Tp=800-920 K) was
found following CO2 adsorption at 100 K. It was concluded that the high-
temperature form of COa arises from the decomposition of carbonate formed
from COa and K,O. The scrambling of the oxygen isotope in this form
supports this assignment, since the three oxygen atoms in the carbonate are
in equivalent chemical environments.
The stabilizing effect of alkali metal adatoms on the adsorption of COZ
on the Pd( 100) surface was also observed by Egawa et al. [ 351, who deposited
Na on the Pd surface. The Na adatoms significantly increased the binding
energy of COa, which desorbed in three peaks, with T,=380, 550 and 650
K, at monolayer Na. In the case of a clean surface, COz desorbed at 200
K. On an oxygen-covered surface, the peak at 550 K was missing: this site
was probably already blocked by the pre-adsorbed oxygen. The oxygen isotope
distribution in COa desorbed at 650 K after C1602 adsorption on the la0
pre-adsorbed surface showed that random mixing occurred between adsorbed
oxygen and COW, which suggested that the highest peak is derived from
the decomposition of sodium carbonate.
Qualitatively similar features for the effect of Na were observed on the
Pd(ll1) surface [34]. At low Na precoverages (&, = 0.25) CO2 adsorbed as
linear, electronically undistorted or only slightly distorted molecules, with
the molecular axis oriented parallel to the surface plane. The COZ molecules
in this layer dissociated into CO and oxygen when the surface was heated
from 85 K to 120 K. The CO molecules appeared to be bound to the Pd
surface. From the angular dependence of the photoelectron spectra, the
authors found no indication of an intrinsic dissociation precursor such as
 345

COz- on Pd(ll1). The possible reason is the narrow temperature range

of existence of this species, as the dissociation of COz certainly involves
the activation of COP and the transitory formation of the COZ- radical
anion.
Completely different adsorption and reaction behavior was found at high
Na precoverage (&,=2.0). Even at the lowest temperature accessible, i.e.
T= 85 K, CO:! reacts with the surface to form a species identified as carbonate.
Together with the carbonate, electronically undisturbed CO2 physisorbs at
low enough temperature. Upon heating of the surface, the amount of adsorbed
carbonate increased and CO2 was desorbed.
Whether this carbonate is formed via the disproportionation of CO2 or
possibly via the reaction of COZwith oxygen contamination was not established
with certainty, but the fact that more COa forms on increasing the temperature
was considered an indication of the disproportionation reaction. The non-
detection of adsorbed CO formed in the disproportionation of the COZ-
anion was explained by the negligible adsorption of CO on the alkali metal
layer.

Platinum

As concern the adsorption of COP on clean Pt surface, Norton and

Richards [36, 371 were able to measure the X-ray photoelectron spectrum
of adsorbed COz only at 77 K. They concluded that COZ is only physisorbed
on a clean Pt film and the estimated heat of adsorption was 10 kcal mol-‘.
In harmony with this conclusion, exposure of a clean Pt(ll1) surface to
CO2 up to 40 L at a pressure of 1 X lo-’ torr at 110-300 K produced no
observable changes in AES and HREELS of Pt [38, 391.

Eflect of sueace adutoms

The interaction of CO2 with a K-dosed Pt(l11) surface has been studied
by high-resolution electron energy loss (HREELS), and TD spectroscopy [38,
391. Deposition of potassium on a clean Pt(ll1) surface produced only one
loss at 190 cm-’ which was attributed to a K-substrate vibration. Adsorption
of COP on this surface at 112 K gave additional losses at 441, 820, 1340
and 1600 cm-‘.
An increase in the CO2 exposure intensified all the losses, particularly
at 1600 cm- ‘, but their positions remained in the range of accuracy ( + 20
cm-‘) of the determination of peak positions. This coverage-independent
behavior may be an indication of a direct interaction between CO2 and K.
The following losses were tentatively attributed to the asymmetric stretching
(1600&20 cm-‘), symmetric stretching (1340&20 cm-‘) and the bending
mode (820 &-20 cm- ‘) of a bent COZ- species formed by the direct interaction
of metallic potassium and CO2 which may be stabilized in the form of a
dimer or oxalate (eqns. 2, 3).
 At higher COa exposures, loss features at 660 and 2350 cm-’ were also
observed, which were assigned to the bending and asymmetric stretching
modes of weakly adsorbed and only slightly perturbed COa.
Upon heating the saturated layer, the first spectral changes occurred at
200 K. This consisted of (i) complete elimination of peaks assigned to weakly
adsorbed COz (660 and 2340-2350 cm-‘), (ii) significant attenuation of
the loss at 1340 cm-‘, (iii) the formation of a new intense feature at 1440
cm-‘, and (iv) an intensification of the loss at 1600 cm-‘. The weak shoulder
at 1340 cm‘-’ vanished at slightly above 200 K, and the losses at 1440 and
1600 cm-” intensified further. It is important to point out that the intensities
of 1440 and 1660 cm- ’ losses were - 30% higher than those at 1340 and
1600 cm-’ measured at 150 K, which makes it very likely that both losses
belong to new surface species. In addition, several very weak losses were
resolved in the low-frequency region at 780, 870 and 1020 cm-‘. AI1 these
spectral changes was attributed to the occurrence of reaction (4). The loss
at 1440 .cm- ’ is assigned to carbonate and the feature at 1600 cm-’ to
the asymmetric stretch of CO perturbed by potassium.
The above description is supported by the high stability of these spectral
features: between 200 and 650 K, HREEL spectra indicated very little alteration
in the co-adsorbed layer. As one cannot expect such a high thermal stability
from either COz- or CzOd2- anion, an alternative, carbonate, is required.
Attenuation of the intense loss at 1600 cm-” starts above at 650 K, and it
is completely eliminated around 710 and 870 cm-‘. In this temperature
range the majority of potassium is also desorbed, generating free Pt sites
on which CO may adsorb from the backwood. A si~i~~~t attenuation of
the loss at 1440 cm-’ occurred between 680 and 750 K, and it was completely
absent above 800 K. However, even after this temperature treatment, very
weak losses remained at 190, 441 and 1000 cm-‘.
Supplements TPD measurements revealed that CO (Z’, = 640 K) and
COz (T, = 674 and 790 K) desorbed in nearly equal amounts. The characteristic
peak temperature (T,= 640 K) agrees well with the peak temperatures of
CO desorption from 1 monolayer of K on Pt(ll1) [ 40 1. The evolution of
COz points to the decomposition of carbonate-like species.

Nickel

The most detailed picture concerning the nature of the adsorption of

COe and the structure of the adsorbed CO2 is available for Ni surfaces. Angle-
resolved photoemission, high-resolution electron spectroscopy and molecular
beam and near edge X-ray adso~tion fine structure (NEWS) and diffuse
LEED methods were used (41-501.
The first observation regarding the nature of the adsorption on single
crystal surface was made by Madix et al. [41, 42 J, who observed that CO2
 347

adsorbed dissociatively on the Ni( 110) and Ni(lO0) surfaces. A 100 L exposure
of CO2 at 150 K resulted in 4X 1013 molecule COa cmm2 desorbed and
2.1 x lOI molecule CO cmW2 desorbed. No CO2 was observed to desorb
from Ni(lOO) after 100 L exposure at 300 K, but CO was detected at
420 K.
The dynamics of the dissociative adsorption of CO2 on Ni(lOO) was
investigated by means of a supersonic molecular beam technique [43].
Prom a detailed analysis of the angle-resolved photoelectron spectra of
adsorbed CC, on the Ni(lOO) surface, Bartos et al. [44] came to the conclusion
that a bent CO2 species is formed on this Ni surface at 140 K, this being
the precursor of the dissociation. The bent C02- anion is produced by the
transfer of electronic charge from the metal to the CO2 molecule. The
additional electronic charge can be accommodated by the closed-shell CO2
molecule only through the breaking of one of the two C-O double bonds
and the formation of an additional oxygen lone pair. This bond-breaking
process opens space on the carbon atom for the acceptance of the extra
charge. The presence of charge transfer to the adsorbate is corroborated by
the strong work function increase observed at 140 K, where much of the
coadsorbed CO2 was desorbed.
Possible coordination sites with mirror plane symmetry are shown in
Pig. 1. While the sites on the left and right hand sides of Pig. 1 have CzV
symmetry, that shown in the middle has C, symmetry. Analysis of photoelectron
spectra suggested that the adsorbed COz has CsVsymmetry. This interpretation
and assignment were corroborated by subsequent EELS experiments [44].
When CO2 was adsorbed at 140 K, dominant peaks were observed at 7501
745 cm-‘. These values are compatible with the frequencies of the bending
and the symmetric stretching modes of a bent CO2 species. In the specular
direction, no energy loss was observed between 1600 and 1700 cm-‘. In
off-specular scattering geometry, however, a peak appears at 1620 cm-‘,
which was assigned to an asymmetric stretching mode of a bent CO2 species
on the Ni(ll0) surface. The additional losses observed at 670, 1390 and
2350 cm- ’ originate from the vibrational excitation of linear, undisturbed
C02. In agreement with the conclusions drawn from the photoemission results,
this shows the coexistence of linear undisturbed CO2 and a bent CO2 species
at a surface temperature of 140 K, similarly to that found for K-dosed
Rh(ll1) and Pd(lOO) surfaces [19, 23, 321.
The absence of an asymmetric stretching loss in the specular direction,
together with some symmetry considerations, led the authors to conclude

(2” CS C2”

Fig. 1. Possible coordination sites of CO2 together with the corresponding.point group symmetries
1441.
348

that the molecule is coordinated in a CpVcoordination site. For differentiation

between the two possible CzV coordination modes, the assignment of the
410 cm- ’ loss to the Ni-0 stretching vibration helped: this suggested that
a CBycoordination site with direct oxygen coordination is more likely than
a carbon-coordinated site.
Many of these features were confirmed by the EELS study of Lindner
et al. 1471. The bent species was found to be stable up to 220 K: at this
temperature, dissociation into CO and 0 occurred.
Direct evidence as to the bonding geometry of COz adsorbed on the
Ni(ll0) face came from NEXAFS [48] and diffuse LEED studies [48, 491.
An assignment of the spectral features on NEXAFS was made by con-
sideration of the M.O. scheme and by comparison with the EELS for free
COz. From the polarization dependence of the NEXAFS, which is in agreement
with previous ARUPS and EELS measurements [44, 471, it was concluded
that the molecular plane of the bent chemisorbed species is most probably
oriented perpendicular to the surface plane, but there is no preferential
orientation of molecular plane along the (110) or (loo} surface azimuth.
The latter observation is compatible with two different species oriented in
each azimuth. In addition, there is a time dependence of the relative intensity
of the resonance, which was interpreted as a preferential re-alignment along
(1 lo> under the influence of the X-ray beam. A ~-resolution experiment
at the carbon edge revealed that the width of the r resonance due to
physisorbed COz is much smaller than that of the chemisorbed species.
The polarization dependence of the physisorbed species suggests a lying-
down connation. Diffuse LEED structure analysis of adsorbed COz supports
the assumption that a bent COz- species is adsorbed on top of the Ni rows
in the (110) direction, which favors direct oxygen metal bonding over pure
carbon coordination [49]. In a second study, no clear preferential azimuthal
orientation was found (48). R-factor analysis in diffuse LEED favored the
coexistence of molecules aligned along each principal azimuth. Equal numbers
of them on top sites appeared more likely than on hollow, long-bridge or
short-bridge sites.
Although the chemisorbed CO2 is now undoubtedly better characterized,
the O-C-O angle is &ill unknown and it is not clear whether the molecule
is bent upwards or downwards. By analogy with the surface formate species,
the latter geometry is expected [48].

Eflects of surface adato~

The effects of pre-adsorbed oxygen on the adsorption and reactivity of
COz on Ni(ll0) have been investigated by photoelectron spectroscopy [44].
It was concluded that COz at 85 K interacts at least weakly with adsorbed
oxygen. Angle-resolved photoemission spectra suggested amolecular geometry
similar to the geometry on the oxygen-free surface, and the COz molecule
is oriented parallel or almost (*ZOO) parallel to the Ni(ll1) surface.
EELS studies showed that adsorbed COz is much more stable on oxygen-
dosed surface than on the clean surface. It was an important observation
 349

that a bent COz- molecule was not detected on the oxygen-dosed surface
and no indication of the presence of adsorbed CO was found either.
When the sample was heated to 170 K, a new species was formed on
the surface, which was assumed to be bidentate carbonate.
The reaction between adsorbed 0 and COz was investigated on Ni(lO0)
by TDS and XPS, and the formation of surface carbonate was detected [50].
The C(ls) and binding energies for CO,/Ni(lOO) species were 283.0 and
531.2 eV, respectively, i.e. very close to those for bulk NiCOa. Furthermore,
it was concluded that reaction occurs only between adsorbed COz and an
activated form of oxygen on the surface. It does not take place with the
stable ordered Ocaj or O,, and nor by the direct impact of C02(g). Tridentate
bonding was proposed, with each 0 on the Ni surface in approximately
equivalent positions.

Iron

A signiticant contribution to the understanding of the nature of the

interaction of COz with Fe surfaces was provided by Wedler et al. [ 5 l-541.
An interesting face specificity was observed by means of angle-integrated
UP spectroscopy. At 77 K photoemission peaks characteristic of weakly held
linear CO2 molecules dominated the spectra. However, new photoemission
signals attributed to formation of the COz- anion were observed on Fe film
even at 77 K. These signals were detected for the Fe(ll1) surface at - 100
K.
When identical experiments were carried out with Fe(1 lo), no adsorption
or decomposition of CO2 occurred at ah. There was only condensation of
CO2 at 77 K when the COz pressure exceeded the saturation value. No
interaction between CO2 and Fe(ll0) was observed at 300 K either. (This
was in contrast with an early observation that COz dissociates on the Fe(ll0)
surface at 300 K; a LEED pattern similar to that observed with CO adsorption
was found [ 551. As soon as regular steps were produced on the Fe(ll0)
surface by means of ion bombardment, exposure to COP led to results very
similar to those described for Fe(ll1). The only difference was that the
decomposition of COP started at slightly elevated temperatures
It was concluded that the reactivity of Fe surfaces towards the adsorption
of COz is governed by their work function. Samples (Fe(ll0) and UnaMeaIed
Fe film) with high work function (-4.4 eV) do not adsorb COz, whereas
the formation of both COB(phys) and COz- occurred on Fe surfaces (Fe(ll1)
and annealed Fe film) of lower work function ( - 4.1 ev).
The results obtained for Fe(ll1) by low-resolution UP spectroscopy
were verified in an angle-resolved photoemission study [54]. Linear COz
molecules desorbing at low temperature (below 140 K) and anionic COz-
species stable at 160-180 K were detected.
From a comparison of the angular dependence of the CO&-rduced peaks
with a similar data set for the COJNi(ll0) system, it was concluded that
the CO2 molecule is not oriented perpendicular to the surface, but the angle
of inclinatian of the COa molecular axis is likely to be larger than f20”
with respect to the surface plane.
The bent COa species dissociates completely to adsorbed CO i-0 at
300 K.

Copper

Renewed interest in the interaction of COz with Cu was initiated by the

recent discovery that COa is the main source of ‘carbon’ in the synthesis
of methanol from Ha+ CO i- COa mixtures [56-611. The key step in the
reaction scheme is the dissociation of COa, i.e. the formation of Oca) and
CO,,.
The early classical work of Collins and Trapnell [6] found no adsorption
of COa on evaporated Cu films at 195 and 273 K, which was attributed to
the full d-bands precluding covalence. This feature was confirmed by a number
of studies using tools of modern surface science [ 63-701.
At exposures up to 350 L, Rodriguez et al. [69] were unable to react
any measurable amount of CO2 with clean Cu(ll0) surface at 110 or 250
K, as evidenced by the complete absence of an oxygen signal in the XPS
and ,AES after exposure. The same conclusion was reached by Bugyi et al.
[70] for the Cu(ll1) surface from AES, TPD, UPS and XPS studies.
In stark contrast with these results, Wachs and Madix [ 71) reported
that 99% of adsorbed CO2 dissociated to form CO and a surface oxygen
species. Although this observation was cited frequently by the ICI group
[60], certain caution is required, as this statement was merely a side remark
in a paper dealing with the adsorption and oxidation of several organic
compounds on the Cu(100) surface. In addition, neither coverage nor tem-
perature were given, which does not make this observation very useful.
The results obtained on polyc~st~l~e Cu suggest an exceptionally high
reactivity toward CO2 [72]. When the Cu surface was exposed to 115 L of
COa at 80 K, the XPS spectrum in the C(ls) region exhibited two main
features, one at 289 eV and the other at 291 eV. These peaks were assigned
to the COz- anion and physisorbed COz, respectively. A weak signal at 293.4
eV was attributed to a COa multilayer. At 93 K, the 289 eV peak dominated
the spectrum, with a new peak at 285 eV ascribed to Ccal*.This assignment
would mean that the complete dissociation of a COz molecule could occur
on the Cu sample at a temperature as low. as 93 K (!), which clearly needs
confirmation. Unfortunately, the oxygen region of the XPS spectrum was not
shown and the behavior of these peaks above 93 K was not reported. The
authors gave no concentra~on values, either for adsorbed COa or for surface
carbon.
The dissociation of COa above 300 K was reported by the ICI group
on Cu powder and on polycrystalline Cu foil [60, 611. In this case, however,
the pressure range was higher and the state of cleanness of the surface was
not checked by electron spectroscopy.
 351

Eflect of surface adatoms

Wachs and Madix [ 711 reported that the presence of oxygen adatoms
on the Cu(ll0) surface diminished the sticking probability of COs, and the
total amount of CO2 adsorbed decreased following initial exposure to oxygen.
A similar observation was made by Bugyi et al. [ 701 for the Cu(ll1) surface.
A significant adsorption of CO2 was observed by Copperthwaite et al.
[ 721 on the pre-oxidized Cu(2 11) surface at 80 K. As regards the promoting
or inhibiting effect of pre-adsorbed oxygen, no conclusion can be drawn, as
no comparison with the clean Cu(2 11) surface was made. On heating of the
adsorbed layer to 130 K, the CO2 multilayers were desorbed and peaks
ascribed to C02(aj- and physisorbed COB appeared in the XPS spectrum.
A much more drastic effect was observed in the presence of preadsorbed
alkali metals [69, 701 which exerted an influence on the adsorption of CO2
on Cu surface similar to the case of Pt metals.
It was found that the amount of adsorbed CO2 and the probability of
adsorption of CO2 are enhanced as Cs is added to the Cu(ll0) surface.
Several new adsorption states were produced, which desorbed in the intervals
110-170 K and 500-700 K. The low-temperature CO2 was attributed to
neutral CO2 stabilized by Cs, and the high-temperature desorption to the
thermal cracking of Cs . &JO,,,, sp ecies. The stabilization of Cs in the adsorbed
layer was also observed: Cs desorbed in a coincident peak with COZ. The
formation of adsorbed CO likewise occurred but, in contrast with the behavior
observed for Pt metals, the CO formed was not stabilized by Cs; it was
released below 250 K.
An interesting feature of this system is that carbonate formation, identified
by XPS measurements, occurred at the adsorption temperature of 110 K.
This is probably the reason why the authors found no spectral evidence for
the existence of COZ- anion species. In the further description of the reactions
of COB on the K-dosed Cu surface, the same processes were taken into
account as for Rh and Pd surfaces [ 19, 321.

Rhenium

The adsorption of CO2 was studied on the Re(OOO1) surface [ 731.

Adsorption kinetics consistent with an extrinsic precursor state were observed
at a crystal temperature of 80 K. A dissociation probability of 0.85 was
found at coverages of less than 8= 0.1 and was coverage-dependent, decreasing
to 0.35 at saturation CO2 coverage. The dissociation probability was crystal
temperature-dependent, decreasing by a factor of 10 when the adsorption
temperature was increased from 80 K to 310 K. This decrease was attributed
to a reduction in the sticking probability of CO,&, at the surface, although
a mechanism in which CO,, desorption competes with the dissociation
reaction was not ruled out. The CO,, and O,, fragments formed an ordered
(2 x 2) structure.
More details on the CO,-Re interaction were revealed by HREEL spec-
troscopy [74]. Following the adsorption of CO2 (0.8 L) on Re(OO01) at 83
352

K, the most intense loss feature was seen at 650 cm-” with relatively weak
modes at 1600, 1935 and 2350 cm- ‘. In the 10” off-specular direction, not
onIy was the intensity at 1600 cm-i enhanced reiative to aII other transitions,
but a new vibrational mode at 1285 cm- ’ was also distinctly observed. After
a short annealing of the sample at 120 K, most of the 650 cm-’ peak and
that at 2350 cm- ’ disappeared. The vibrational mode at 1600 cm- ’ persisted
and was somewhat enhanced. Further heating of the sample to 135 K or
above produced a new loss feature at 1970 cm-‘, indicative of a low coverage
of CO on the Re(OOO1) surface. In the presence of adsorbed oxygen, the
peak previously observed at 1600 cm-’ was shifted to 1650 cm-’ and a
peak was clearly seen at 1290 cm-‘.
The HREELS peaks at 650, 1290 and 2350 cm-’ were assigned to the
vibrations of a linear molecular COa lying parahel to the surface, while the
observed extra modes at 1600 cm-’ (on a clean surface) and 1230 and
1650 cm-’ (on an oxygen-precovered surface) were considered to be in-
dications of the presence of new surface species. These were attributed to
a bent CO*-, which dissociates to CO, and O,, above 130 K.

The interaction of COawith cleanAl foil was followed by XPS measurements

at 80-295 K f75]. From an analysis of the O(ls) and C(k) spectra, it was
concluded that me&stable surface carbonate is formed at 80-120 K. The
formation of carbonate involves the dissociation of adsorbed COz and the
subsequent recombination of adsorbed 0 and COz. The carbonate is thermally
unstable and dissociates readily above 120 K to give carbon in two forms,
carbide, C, ‘--, and graphitic C,*.
At 295 K and low COz pressures (lo-* torr), there was no evidence of
adsorption or reaction even at high exposures (3 X lo3 L). At high pressure
(1 torr), the formation of both carbidic and graphitic-type species was
es~b~hed, suggesting that the metastable carbonate is a precursor state
for dissociation.
InterestingIy, no COz adsorption was observed on the AI(100) surface
at around 100 K. Surface atoms with low coordination numbers and significantly
higher pressures were considered to be the main reasons for the high reactivity
of a polyc~~~e Al sample.

Effect of suyfbce adatom

A significant promotion of COa adsorption was observed in the presence
of Na adatoms [ 761. The nature of the interaction and its expkmation were
practically the same as those proposed first for K-promoted Rh and Pd
surfaces.
(i) The probability of CO2adsorption increases with alkali metal coverage.
(ii) AIkali metal promotion Ieads to CO2 reduction, giving an anion
radical,which is transformed into oxalate and carbonate at higher temperatures.
 353

The HREEL spectra revealed two different states of CO2 adsorption at

100 K and monolayer Na coverage: weakly bound molecular CO2 and oxalate
ions with low, possibly rmidentate, coordination. The latter species was
characterized by vibrational bands at 460, 1310 and 1480 cm-‘. Heating
of the co-adsorbed layer to 330 K produced strong bands at 1050, 1350
and 1570 cm-‘, which were attributed to surface carbonate bound to the
substrate via two oxygen atoms. The oxalate-carbonate transformation (as
discussed previously) was accompanied by the release of CO at N 285 K.

Magnesium

Atomically clean Mg(OOO1) surface exhibited a high reactivity towards

COz [ 771. Both XPS and EELS indicated the dissociation of CO2 even around
100 K.
When the magnesium surface was exposed to carbon dioxide at room
temperature, the resulting O(ls) spectra showed two peaks. The oxide peak
at 530.7kO.2 eV is close to its position at low temperature, but the O(ls)
peak associated with the C(ls) feature at 292 eV now appears at a slightly
higher binding energy of 534.0f0.2 eV. The ratio of the intensities of the
peaks at 292 eV and 534 evindicates a stoichiometric ratio of C:O = 1:3.2 k 0.5.
This, together with the observation of similar peaks in the spectrum of bulk
MgCOB, constitutes strong evidence for a carbonate species. The data showed
that the carbonate is confhred to the topmost layer, while the oxide forms
a thicker, uniform layer beneath.
The EELS spectra of the sample heated to 131 K exhibited losses at
850, 1090, 1390 and 1630 cm-‘, which were explained by the formation
of bidentate carbonate, although the high intensity of 1390 cm-’ (attributed
to the asymmetric 0 -C-O stretch) is not in accord with this assignment.

Silver

Clean swface
The study of the interaction of COPwith silver metal has been the subject
of several thorough works [78-941. This interest was primarily motivated
by the fact that silver is a catalyst industrially applied in the epoxidation
of ethylene. The results unequivocally showed that ‘the COZ adsorbs only
very weakly on clean Ag surfaces.
The properties of COP physisorbed on a Ag(ll1) surface at 40 K were
studied by HREELS and LEED [ 78 1. The observed loss energies at 83, 167
and 290 meV coincide with the vibrational energies of bending, symmetric
stretching and asymmetric stretching modes of CO2 in the gas phase within
the accuracy of the measurements ( f 3 meV). No diffraction spots of ordered
structures of CO2 layers were observed. Neither annealing to 70 K nor
condensation at elevated substrate temperature (85 K) produced any change
354

in the LEED pattern. This agrees with the He diffraction experiment, in

which no diffraction spots were registered [ 791. It was concluded that layer-
by-layer growth is not probable and that island formation occurs preferentially.
It was assumed that at 40 K the deposited layer consists of three-dimensional
microcrystals even at submonolayer coverages.

Eflects of sm$ace adatowts

The adsorption of COz on an oxygen-dosed powdered silver specimen
was studied first by Czanderna and Bregen [80-821 using a vacuum micro-
balance thermal desorption technique. They found that pre-adsorbed oxygen
promoted the adsorption of COZ, which desorbed in a first order process
with an activation energy of 84 to 102 kJ mol- ’ and a pre-exponential factor
in the range 10”-1013 s-‘.
Similar observations were made by Madix et al. [83-871 on Ag single
crystal surfaces under UHV conditions. They have shown that an adsorbed
carbonate species (CO,(,,) can be produced via the adsorption of CO2 onto
an Ag(ll0) surface predosed with oxygen atoms [ 831. The reaction of C1602
with preadsorbed 180 atoms showed that all oxygens in the CO,,, species
are chemically equivalent. It was postulated that this equivalence was achieved
by a facile interconversion of monodentate (MOC02) and bidentate (MO&O)
species.
The thermal desorption spectrum following an exposure of 1.5 x 1015
molecules cmm2 of CO2 at 100 K to the oxygen-covered Ag(ll0) surface
showed two peaks at 130 and 480 K [85]. The high-temperature state is
due to the decomposition of surface carbonate, CO,,,, to yield CO,,, and
Ocaj. The low-temperature state is due to the desorption of molecular C02.
This form of CO2 was thought to be induced by adsorbed carbonate, but
its presence - which means a stabilization of physisorbed CO2 - could be
also due to chemisorbed oxygen.
EEL spectra following a CO2 exposure of 1.5 x lOI molecules cmp2 to
the oxygen covered surface at 100 K conflrmed the presence of molecularly
adsorbed CO2 and CO3 species [85]. The losses observed were assigned as
follows: 260 (Ago); 830 71(C03); 1050 -y(CO); and 1360 cm-’ r,(OCO). The
@OS) notation refers to the out-of-plane carbonate deformation, while
y,(OCO) represents the symmetric OCO stretch. No loss was observed for
the asymmetric stretch (expected to be around 1480 cm-‘). This indicates
that the axis through the ‘up’ oxygens of the carbonate is parallel to the
surface according to the dipole selection rule. The absence of a peak near
1600 cm- ’ clearly rules out a structure of the form
0
II

oic\o
Photoelectron spectroscopy of the CO3 formed demonstrated little per-
turbation of the molecular orbitals or of electron distribution on the oxygen
atoms due to bonding with the Ag(ll0) surface [86].
 355

Suggestions for future work

As the surface chemistry of adsorbed COa on metal single crystal surfaces

has been rather neglected in the past, this subject is open for several studies
using the tools of modern surface science. Without going into detail, we list
some selected topics which need further consideration for the better un-
derstanding of the features summarized in the present review:
(1) Study of the effects of defects of transition metals on the adsorption
and dissociation of COz.
(2) Photochemical activation of COz on metal surfaces with particular
emphasis on the formation of COs- radical anion.
(3) Determination of the structure and coordination mode of CO2 in the
presence of alkali promoters by NEXAFS, diffuse LEED and HREELS methods.
(4) Detection of the formation of alkali-COa surface complex on transition
metal surfaces by SIMS and/or by field desorption mass spectrometry.
(5) Further study of the nature of the interaction of CO2 with different
faces of a carefully cleaned Cu single crystal, as regards the formation of
COz- anion and the extent of dissociation.
(6) Study of the reactions of COa, activated by alkali metals, with other
gas molecules on different metal surfaces.
(7) Extension of the investigation of the adsorption of CO:! ,to other
metal surfaces.

Conclusions

It can be generally stated that COa adsorbs weakly and molecularly on

carefully cleaned Pt metals at 100-300 K under UHV conditions. It appears
likely that the dissociation is also very limited on Cu surfaces, if it occurs
at all. The adsorption of COz is dissociative on Fe, Ni, Re, Al and Mg surfaces.
The presence of oxygen adatoms promotes the bonding of CO2 on metals
in form of different carbonates and leads to its stabilisation.
The bonding, structure and reactivity of adsorbed CO2 can be drastically
influenced by alkali adatoms through the formation of surface COP- radical
anion. Depending on the state of alkali metals, COz- may dissociate to CO
and 0, or transform into COBand CO species. It is likely that a COz- species
plays a dominant role in the thermal and photocatalytic reactions of the COz
molecule.

References

1 R. P. A. Sneeden, in G. Wilkinson (ed.), Comprehensive Organometallic Chemistry,

Pergamon, New York, 1982, Vol, 8.
2 D. J. Darensbourg, C. G. Bauch and C. Ovalles, Rev. Inorg. Chem., 7 (1985) 315; D. J.
Darensbourg and R. A. Kudaroski, Adv. Organometall. Chem., 22 (1983) 129.
3 W. M. Ayers (ed.), Catalytic Activation of COz, ACS Symp. Ser., American Chemical
Society, Washington D.C., 1988.
356

4 A. Ya. Rozovskii, G. Lint, L. B. Liberov, E. V. Silvinski, 8 M. Loktev, Yu. B. Kagan and

A. N. Bashkirov, Kinet. Catai. (Engl. Transl.), 18 (1977) 691.
5 G. C. Chinchen, P. J. Denney, D. G. Parker, M. C. Spencer and D. Whan, Appl, Catal.,
30 (1987) 333.
6 A. C. Coilins and B. M. W. Trapneli, Trans Fa?o.&z~ Sot., 53 (1957) 1436.
7 C. T. Campbell and J. M. White, J. Cut&!., 54 (1978) 289.
8 A. C. Yang and C. W. Garland, J. Phys. C&em., 61 (1957) 1504.
9 B. A. Sexton and G. A. Somorjai, J. Catal., 46 (1977) 167.
10 D. G. Castner, B. A. Sexton and G. A. Somorjai, Sur$ Sci., 71 (1978) 519.
11 D. G. Castner and G. A. Somorjai, Sur$ Sci., 83 (1979) 60.
12 L. H. Dubois and G. A. Somorjai, Su@ Sci., 88 (1979) L213.
13 L. H. Dubois and G. A. Somorjai, Su@ Sci., 91 (1980) 514.
14 W. H. Weinberg, Sur$ Sci., 128 (1983) L224.
15 D. W. Goodman, D. E. Peables and J. M. White, Surf. Sci., 140 (1984) L239.
16 L. H. Dubois and G. A. Somorjai, Su@. Sci., 128 (1983) L231.
17 F. Soiymosi and J. Kiss, Su?cf: Sci., I7 (1985) 149.
18 F. Solymosi, J. Kiss, Cti. Phys. L&t., 110 (1984) 639.
19 F. Solymosi and L. Bugyi, J. Chem. Sot., Faraday Tram. I, 83 (1987) 2015.
20 H. A. C. M. Hendrickx, A. P. J. M. Jongenelis and B. E. Nieuwenhuys, Sur$ Sci., 154
(1985) 503.
21 G. B. Fischer, B. A. Sexton and J. L. Gland, J. Vuc. Sci. Techrwl., 17 (1980) 144.
22 P. A. Thiel, J. T. Yates and W. H. Weinberg, Sur$ Sci., 82 (1979) 22.
23 J. Kiss, K. R6vesz and F. Solymosi, Sur$ Sci., 207 (1988) 36.
24 D. W. Turner, H. D. Baker, C. BakerandC. R. Brundle,MokculurPhotoekctronSpectroscopy,
W‘lley, New York, 1970.
25 J. Kiss, K. Rev&s and F. Solymosi, Appl. Surf: Sci., 37 (1989) 95.
26 A. Berk6 and F. Solymosi, Surf: Sci., 201 (1986), L489.
27 Z. H. Kafaii, R. H. Hauge, W. E. Billups and J. L. Margraven, J. Am. Chem. Sot., 105
(1983) 3886.
28 H. J. Freund and R. P. Messmer, Sue Sci., I (1986) 172.
29 B. Bartos, H. J. Freund, H. Kuhlenbeck, M. Neumann, H. Linder and K. Miiller, Sur$ Sci.,
59 (1987) 179.
30 T. Engel and G. Erti, A&. CutaL, 1 (1979) 28.
31 T. Engel and G. Erti, J. C&m. Phys., 69 (1978) 1267.
32 F. Soiymosi and A. Berk6, J. CataL, 101 (1986) 458.
33 T. Matsushima, J. Phys. Chem., 91 (1987) 6192.
34 J. Wambach, G. Odiirfer, H. J. Freund, H. Kuhlenbeck and M. Neumann, Surx Sci., 209
(1989) 159; S. Wohirab, D. Ehriich, J. Wambach, H. Kuhlenbeck and H. J. Freund, Sur$
Sci., (1990) in press.
35 C. Egawa, I. Doi, S. Naito and K. Tamaru, Sud Sci., 176 (1986) 491.
36 P. R. Norton, Surf: Sci., 44 (1974) 624.
37 P. R. Norton and P. J. Richards, Sur$ Sci., 49 (1975) 567.
38 Z. M. Lin, Y. Zhou, F. Solymosi and J. M. White, J. Phys. Chem., 93 (1989) 4383.
39 Z. M. Lin, F. Solymosi and J. M. White, SurJ Sci., in press.
40 C. M. Greenlief, P. L. Radloff, S. Akhter and J. M. White, Sur$ Sci., 186 (1987) 563; J.
E. Croweil, E. L. Gerfunkel and G. A. Somorjai, Surf. Sci., 121 (1982) 303.
41 I. McCarty, J. Faiconer and R. J. Madii, J. CataL, 30 (1973) 235.
42 J. B. Benzinger and R. J. Madii, Sue Sci., 79(1979) 394.
43 M. P. D’Evelyn, A. V. Hanua, G. E. Gidowski and R. J. Madix, Sufl Sci., 167 (1986) 451.
44 B. Bartos, H. J. Freund, H. Kublenbeck, M. Neumann, H. Lmdner and K. Mtiiler, Sur$
Sci., 59 (1987) 179.
45 H. J. Freund, H. Kossmann and Y. Schmidt, Cht?m. Phys. l&t., 123 (1986) 463.
46 H. J. Freund and R. P. Messmer, Surf. Sci., 1 (1986) 172.
47 H. Lindner, D. Rupprecht, L. Hammer and K. Miiller, J. EkctronSpectrosc. Relat. Phmwm.,
44 (1987) 141.
 357

48 G. Iliing, D. Heskett, E. W. Plummer, H.J. Freund, J. Somers, Th. Lmdner, A. M. Bradshaw,

U. Buskotte, M. Neumann, U. Starke, K. Heinz, P. L. de Andres, D. Saidin and J. B. Pendry,
Surf. Sci., 206 (1988) 1.
49 K. Heinz, D. K. Saldm, J. B. Pendry, Phys. Rev. L.&t., 55 (1985) 231.
50 R. J. Behm and C. R. Brundle, J. Vat. Sci. Technol., AI (1983) 1223.
51 R. Dziembaj and G. Wedler, Su$ Sci., I34 (1983) 283.
52 H. Behner, W. Spess, G. Wedler and D. Borgmann, Sz& Sci., 175 (1986) 276.
53 R. Bauer, H. Behner, D. Borgmann, M. Pimer, W. Spess and G. Wedler, J. Vuc. Sci.
TechnoZ., (1987) 1110.
54 H. J. Freund, H. Behner, B. Bartos, G. Wedler, H. Kuhlenbeck and M. Neumann, Surj
Sci., 180 (1987) 550.
55 R. Yoshida, G. Somorjai, SurJ Sci., 75 (1978) 46.
56 A. Ya. Rozovskii, Kin&. Katal., 21 (1980) 97.
57 Yu. B. Kagan, G. I. Lin, A. Ya. Rozovskii, S. M. Loktev, E. V. Slivinskii, A. N. Bashkirov,
I. P. Naumov, I. K. Khludenev, S. A. Kudinov and Yu. I. Golovkin, Kinet. Katal., I7 (1976)
440.
58 Yu. B. Kagan, A. Ya. Rozovskii, G. I. Lin, E. V. Slivinskii, S. M. Loktev, L. G. Liberov and
A. N. Bashkirov, Kinet. Katal., 16 (1975) 809.
59 A. Ya. Rozovski, Yu. B. Kagan, G. I. Lm, E. V. Siivinskii, S. M. Loktev, L. G. Liberov, A.
N. Bashkirov, Kinet. Katat., 16 (1975) 810.
60 G. C. Chinchen, M. S. Spencer, K. C. Waugh, D. A. Whan, J. Chem. Sot. Faraday Trans.
1, 83 (1987) 2193.
61 G. C. Chinchen, P. J. Denny, D. G. Parker, M. S. Spencer and D. A. Whan, Appl. Catal.,
30 (1987) 333.
62 F. H. P. M. Habraken, E. Ph. Kieffer and G. A. Bootsma, SurJ Sci., 83 (1979) 45.
63 P. R. Norton and R. L. Trapping, Ch.em. Phys. L&t., 38 (1976) 207.
64 A. B. Anderson, Surf: Sci., 62 (1977) 119.
65 N. K. Ray and A. B. Anderson, Sue Sci., 119 (1982) 35.
66 F. Solymosi and J. Kiss, Surf: Sci., IO8 (1981) 368.
67 F. Solymosi and J. Kiss, Surf: Sci., IO4 (1981) 181.
68 F. Solymosi and A. Berko, Sur_- Sci., 122 (1982) 275.
69 J. A. Rodriquez, W. D. Clendering and C. T. Campbell, J. Phys. Chem., 93 (1989) 5238.
70 L. Bugyi, J. Kiss and F. Solymosi, J. Catal., in press.
71 J. E. Wachs and R. J. Madii, J. Cat&., 53 (1978) 208.
72 R. G. Copperthwaite, P. R. Davies, M. A. Morris, M. W. Roberts and R. A. Ryder, Catal.
L&t., I (1988) 11.
73 H. Peled and M. Asscher, SuG Sci., 183 (1987) 201.
74 H. Asscher, C. T. Rao and G. A. Somorjai, J. Phys. Chem., 92 (1988) 2711.
75 A. F. Carley, B. E. Gallagher and M. W. Roberts, SurJ Sci., I83 (1987) L263; idem,
Spectrochim. Acta, 43A (1987) 1447.
76 J. Paul and F. M. Hoffmann, Catal. L&t., 1 (1988) 445.
77 S. Campbell, P. Hoilins, E. McCash and M. W. Roberts, J. Electron Spectrosc. Relat.
Phewm., 39 (1986) 145.
78 M. Sukurai, T. Okano and Y. Tuzi, J. Vat. Sci. TechnoZ., (1987) 431.
79 T. H. Eilis, G. Scales, U. Vaibusa, H. Johnson and J. H. Weare, Surj Sci., I55 (1985)
492.
80 A. W. Czandema, J. CoL!oid Irztq$ Sci., 22 (1966) 482.
81 A. W. Czandema, J. Phys. Chem., 70 (1966) 2120.
82 A. W. Czandema and J. B. Bregen, J. Vat. Sci. Technol., 8 (1971) 594.
83 M. Bowker, M. A. Barteau and R. J. Madix, Surj Sci., 92 (1980) 528.
84 M. A. Barteau and R. J. Madix, Sur_f Sci., 97 (1980) 101.
85 E. M. Stuve, R. J. Madix and B. A. Sexton, Chem. Phys. L&t., 89 (1982) 48.
86 M. A. Barteau and R. J. Madix, J. Ckem. Phys., 74 (1981) 4144.
87 M. A. Barteau and R. J. Madix, J. Electron Spectrosc. Relat. Pherwm., 31 (1983) 101.
88 C. Backx, C. P. M. De Groot and P. Biloen, Appl. Sue Sci., 6 (1980) 256.
358

89 C. Backx, C. P. M. De Groot, P. Biloen and W. M. H. Sachtler, Sur$ Sci., 128 (1983) 81.
90 K. C. Prince and A. M. Bradshaw, Sm.-.- Sci., 126 (1983) 49.
91 C. T. Campbell and M. T. Paffett, Su@ Sci., 143 (1984) 517.
92 J. Segner, C. T. Campbell, G. Doyen and G. Ertl, SurJ Sci., 138 (1984) 505.
93 C. T. Campbell, G. Ertl, H. Kuipers and J. Segner, Chem. Phys. L&t., 73 (1980) 5862.
94 C. T. Campbell, G. Ertl, H. Kuipers and J. Segner, SurJ Sci., IO7 (1981) 207, 220.