Atomic and molecular adsorption on Rh„111…
M. Mavrikakis,
a)
J. Rempel, and J. Greeley
Department of Chemical Engineering, University of Wisconsin-Madison, Madison, Wisconsin 53706
L. B. Hansen and J. K. Nørskov
Center for Atomic-scale Materials Physics, Department of Physics, Technical University of Denmark,
DK-2800 Lyngby, Denmark
共Received 10 January 2002; accepted 24 July 2002兲
A systematic study of the chemisorption of both atomic 共H, O, N, S, C兲, molecular (N
2
, CO, NO),
and radical (CH
3
, OH) species on Rh共111兲 has been performed. Self-consistent, periodic, density
functional theory 共DFT-GGA兲 calculations, using both PW91 and RPBE functionals, have been
employed to determine preferred binding sites, detailed chemisorption structures, binding energies,
and the effects of surface relaxation for each one of the considered species at a surface coverage of
0.25 ML. The thermochemical results indicate the following order in the binding energies from the
least to the most strongly bound: N
2
⬍ CH
3
⬍ CO⬍ NO⬍ H⬍ OH⬍ O⬍ N⬍ S⬍ C. A preference for
threefold sites for the atomic adsorbates is observed. Molecular adsorbates, in contrast, favor top
sites with the exceptions of NO 共hcp兲 and OH 共fcc or bridge tilted兲. Surface relaxation leads to
insignificant changes in binding energies but to considerable changes in the spacing between surface
rhodium atoms, particularly for on-top adsorption where the rhodium atom directly below the
adsorbate is lifted above the plane of the surface. RPBE binding energies are found to be in
remarkable agreement with the available experimental values. All atomic adsorbates, except for H,
have a significant diffusion barrier 关between 0.4 and 0.6 eV 共RPBE兲兴 on Rh共111兲. Atomic H and
molecular/radical adsorbates appear to be much more mobile on Rh共111兲, with an estimated
diffusion barrier between 0.1 and 0.2 eV 共RPBE兲. Finally, the thermochemistry for dissociation of
CO, NO, and N
2
on Rh共111兲 has been examined. In all three cases, decomposition is found to be
thermodynamically preferable to desorption. © 2002 American Institute of Physics.
关DOI: 10.1063/1.1507104兴
I. INTRODUCTION
Rhodium is an extremely valuable and versatile transi-
tion metal for applications in heterogeneous catalysis. For
example, current catalytic converter technology employs the
three-way-catalyst 共TWC兲. The TWC, made of supported
Pt–Rh, can catalyze the simultaneous conversion of nitric
oxide, carbon monoxide, and unburned hydrocarbons.
1
Rhodium can also catalyze the hydroformylation reaction
共aldehyde production from olefins, carbon monoxide, and
hydrogen兲 and the catalytic partial oxidation of methane.
1
Natural reserves of rhodium are currently in rapid de-
cline, and as a result Rh is by far the most expensive pre-
cious metal today. Hence, it is of great importance to find
new catalysts with high activity and selectivity to replace the
current generation of rhodium catalysts. Understanding the
surface chemistry of rhodium might suggest efficient ways of
proceeding with this catalyst design process. In particular,
understanding critical elementary reaction steps such as ad-
sorption and bond breaking/making over rhodium surfaces
can provide substantial insights into why rhodium is so es-
sential in catalysis and as to how it can best be replaced.
The interaction of atomic species with rhodium single
crystal surfaces has been the subject of several research stud-
ies in the past. Adsorbed oxygen atoms on Rh共111兲 have
been investigated with low-energy electron diffraction
共LEED兲共Refs. 2 and 3兲 and with molecular-beam scattering
experiments.
4
The overlayer structure of adsorbed atomic ni-
trogen on Rh共111兲 was studied with scanning tunneling mi-
croscopy 共STM兲.
5
For a detailed review of O and N chemis-
try on Rh共111兲 and other single-crystal rhodium facets, see
Comelli et al.
6
The structures formed by adsorbed sulfur on
Rh共111兲 have been studied by LEED 共Refs. 7 and 8兲 and
STM 共Ref. 8兲 experiments. Finally, high resolution electron
energy loss spectra 共HREELS兲 and laser-induced thermal de-
sorption studies have been carried out on Rh共111兲 to study
hydrogen adsorption
9
and hydrogen diffusion,
10
respectively.
Experimental studies of molecular adsorption on
Rh共111兲 are even more abundant. For example, using tem-
perature programmed desorption 共TPD兲, Hendrickx et al. ex-
amined the adsorbed states of dinitrogen (N
2
) on rhodium
共111兲, 共100兲, 共110兲, and 共210兲 surfaces.
11
LEED, EELS, and
TPD studies have been carried out to study the chemisorp-
tion of nitric oxide 共NO兲 on Rh共111兲 and Rh共100兲.
12–14
LEED has also been used to study coadsorbed overlayers of
NOandOonRh共111兲.
5
The adsorption and energetics of
carbon monoxide 共CO兲 have been the subject of several ex-
perimental studies. LEED experiments have been used to
determine site preferences for CO on Rh共111兲 at both low
and high coverages.
15–17
Numerous overlayer structures for
CO on Rh共111兲 have also been observed with LEED,
12,15
a兲
Author to whom correspondence should be addressed. Electronic mail:
manos@engr.wisc.edu; Phone: 共608兲 262-9053; Fax: 共608兲 262-5434.
JOURNAL OF CHEMICAL PHYSICS VOLUME 117, NUMBER 14 8 OCTOBER 2002
67370021-9606/2002/117(14)/6737/8/$19.00 © 2002 American Institute of Physics
Downloaded 08 Feb 2007 to 128.104.198.19. Redistribution subject to AIP license or copyright, see http://jcp.aip.org/jcp/copyright.jsp
high resolution photoemission 共HRP兲,
17
and STM 共Ref. 18兲
experiments. Coadsorbed systems of CO and O have been
considered in other LEED 共Refs. 2 and 3兲 studies. Finally,
reflection adsorption infrared spectroscopy 共RAIRS兲 and
TPD have been used to study the adsorption of the methyl
radical (CH
3
)onRh共111兲.
19
A substantial amount of experimental research has been
done to gain understanding of the kinetics of adsorption, de-
sorption, and reactions on rhodium surfaces. Borg et al. de-
termined several kinetic parameters involved in nitric oxide
decomposition on Rh共111兲 from TPD experiments.
20
Tem-
perature programmed reaction spectroscopy 共TPRS兲 has
been employed to study carbon monoxide oxidation on
Rh共100兲 and Rh共111兲.
21
The kinetics of the simultaneous re-
action NO⫹ CO have also been widely studied using TPD
共Ref. 22兲 and TPRS 共Ref. 23兲 techniques on Rh共111兲. Finally,
TPD has been used to analyze the recombination of atomic
nitrogen on Rh共111兲 at high surface coverages.
24
Several theoretical studies have been performed to study
chemisorption and reactivity on rhodium surfaces. Periodic
DFT calculations have been used to investigate atomic oxy-
gen binding on Rh共111兲,
25–27
and the adsorption of other
atomic species such as nitrogen,
26
sulfur,
28
and hydrogen
27
has also been considered on this surface. Zhang et al. carried
out periodic DFT studies to determine interactions between
chemisorbed CO and S on Rh共111兲.
28
General trends in the
dissociation of CO have been studied on the 共111兲 surfaces of
Rh, Ru, Pd, Os, Ir, and Pt, yielding the corresponding poten-
tial energy surfaces.
29,30
Loffreda et al. performed periodic
DFT calculations on rhodium 共100兲 and 共111兲 surfaces for
molecular and dissociative chemisorption of nitric oxide.
26
Water formation on Rh共111兲 surfaces has also been explored
using DFT, providing thermodynamic and kinetic
parameters.
27
In the present work we present a systematic survey of
the chemisorption structures and energetics of several atomic
共H, O, N, S, C兲, molecular (N
2
, CO, NO), and radical
(CH
3
, OH) species, all of which play a role in industrial
rhodium-catalyzed reactions, on the Rh共111兲 surface. We use
periodic, self-consistent DFT calculations to determine pre-
ferred binding sites and binding energies for the adsorbed
species. We also estimate diffusion barriers and examine the
thermochemistry of CO, NO, and N
2
dissociation on
Rh共111兲.
II. METHODS
All calculations are carried out using DACAPO.
31
A three-
layer slab of rhodium, periodically repeated in a super cell
geometry with five equivalent layers of vacuum between any
two successive metal slabs, is used. A 2⫻ 2 unit cell, corre-
sponding to a surface coverage of 0.25 ML, is used. Adsorp-
tion is allowed on only one of the two exposed surfaces, and
the electrostatic potential is adjusted accordingly.
32
Initial
computations are performed with metal atoms fixed in their
bulk-truncated positions. The calculations are then repeated
for all geometries allowing the top metal layer to relax. Ionic
cores are described by ultrasoft pseudopotentials,
33
and the
Kohn–Sham one-electron valence states are expanded in a
basis of plane waves with kinetic energy below 25 Ry. The
surface Brillouin zone is sampled at 18 special k points. In
all cases, convergence with respect to the k point set and
with respect to the number of metal layers included is con-
firmed. The exchange-correlation energy and potential are
described by two generalized gradient approximations, self-
consistently with GGA-PW91 共Refs. 34 and 35兲 and non-
self-consistently with RPBE.
31
The RPBE functional has
been shown to give better chemisorption energies compared
to PW91.
31
The self-consistent PW91 density is determined
by iterative diagonalization of the Kohn–Sham Hamiltonian,
Fermi population of the Kohn–Sham states (k
B
T⫽ 0.1 eV),
and Pulay mixing of the resulting electronic density.
36
All
total energies have been extrapolated to k
B
T⫽ 0 eV. The cal-
culated lattice constant for bulk rhodium, 3.83 Å, is within
1% of the experimental value of 3.797 Å.
37
III. RESULTS AND DISCUSSION
This section contains a description of the chemisorption
properties for all the adsorbates studied, including binding
energies, diffusion barrier estimates, site preferences, and the
effect of surface relaxation. Binding energies are presented
for both PW91 and RPBE functionals, with the latter given
FIG. 1. Binding energies of atomic, molecular, and radical adsorbates on
Rh共111兲, calculated with the 共a兲 PW91 and 共b兲 RPBE functionals 共solid line,
fixed surface; dashed line, relaxed surface兲. Reference zero corresponds to
gas phase species at infinite separation from the surface. Lines are only
guides to the eye.
6738 J. Chem. Phys., Vol. 117, No. 14, 8 October 2002 Mavrikakis
et al.
Downloaded 08 Feb 2007 to 128.104.198.19. Redistribution subject to AIP license or copyright, see http://jcp.aip.org/jcp/copyright.jsp
in square brackets. The effect of surface relaxation on the
binding energies and that of coordination numbers on the
binding site preferences are discussed. A summary of the
calculated binding energies for fixed and relaxed surfaces for
atomic, molecular, and radical adsorbates is presented in Fig.
1. The thermochemistry of adsorption and decomposition of
carbon monoxide, nitric oxide, and dinitrogen is then pre-
sented. All results are compared to available experimental
and theoretical data from the literature.
A. Adsorption of atoms
Atomic hydrogen is the least strongly bound atom
among those we studied on Rh共111兲, with a binding energy
of ⫺2.79 关⫺2.62兴 eV 共Table I兲. An experimental value for
the binding energy can be estimated by combining the energy
of dissociative adsorption of H
2
on Rh共111兲, ⫺0.807 eV 共ob-
tained from TPD experiments
38
兲, with the H
2
gas-phase bond
enthalpy, ⫺4.52 eV.
37
These values yield an estimated
atomic hydrogen binding energy of ⫺2.66 eV, in excellent
agreement with the RPBE value. In addition, a previous cal-
culation indicates a binding energy of ⫺2.82 eV,
27
consistent
with our findings. The most energetically favorable configu-
ration in our calculations is the fcc site with the hydrogen
atom very close to the surface. The geometric parameters for
this configuration are presented in Table II, while Fig. 2
shows a schematic representation of the corresponding struc-
tural parameters. This site preference is consistent with
HREELS experiments
9
and with the theoretical study men-
tioned above.
27
The hydrogen potential energy surface is
found to be relatively flat in our calculations; an estimate for
the H-diffusion barrier is 0.10 关0.10兴 eV 共Table III兲. These
findings are in close agreement with laser-induced thermal
desorption experiments that indicate a diffusion barrier of
0.14 eV for a 0.3 ML coverage.
10
Finally, the effect of sur-
face relaxation is found to be very small in terms of energet-
ics 共0.02 eV destabilization兲.
The next least strongly bound atom is oxygen. The pre-
ferred binding site is fcc with a binding energy of ⫺4.88
关⫺4.31兴 eV 共Table I兲, and the estimated diffusion barrier is
0.45 关0.41兴 eV 共Table III兲. The geometric parameters for the
fcc site are presented in Table II. The preference for the fcc
site is in agreement with LEED experimental findings
2
and
with previous DFT calculations.
25
However, we find that the
fcc site is preferred by only 0.03 eV over the hcp site. This
very small difference 共discussed further below兲 raises the
question of whether or not the site preferences could be sig-
nificantly altered by zero-point energy effects; to assess the
magnitude of such effects, we calculated the three normal
vibrational modes and frequencies for oxygen in both the fcc
and hcp sites. The total zero-point energy 共including all three
normal modes兲 of oxygen in the hcp site is roughly 100 cm
⫺1
共⬃0.01 eV兲 greater than the corresponding energy in the fcc
site, suggesting that zero-point effects are unlikely to alter
site preferences for this system. In any case, at high surface
FIG. 2. Schematic representation of top and side views of the atomic and
molecular adsorbates on threefold sites of Rh共111兲. 共a兲 Atomic adsorbates.
共b兲 Molecular and radical adsorbates. Geometrical parameters shown are
used in Tables II and V.
TABLE I. Binding energies of atomic adsorbates on a relaxed Rh共111兲 surface. Preferred sites are shown by shaded entries. The reference zero corresponds
to gas-phase atoms at infinite separation from the rhodium slab. 共Experimental estimates are obtained from the gas-phase diatomic bond energy.
b,c
兲
Best site
Binding energy 共eV兲
PW91 共RPBE兲
Experimental
estimates
共eV兲
Calc. Expt. Top Bridge fcc hcp
H fcc fcc
a
⫺2.57 关⫺2.44兴 ⫺2.68 关⫺2.52兴 ⫺2.79 关⫺2.62兴 ⫺2.75 关⫺2.59兴 ⫺2.66
b,c
O fcc fcc
d
⫺3.58 关⫺3.14兴 ⫺4.43 关⫺3.90兴 ⫺4.88 关⫺4.31兴 ⫺4.85 关⫺4.26兴 ¯
Nhcp ¯ ⫺3.24 关⫺2.97兴 ⫺4.62 关⫺4.24兴 ⫺5.06 关⫺4.65兴 ⫺5.31 关⫺4.87兴 ¯
S fcc fcc
e,f
⫺3.66 关⫺3.27兴 ⫺4.96 关⫺4.49兴 ⫺5.42 关⫺4.95兴 ⫺5.39 关⫺4.92兴 ¯
Chcp¯ ⫺5.10 关⫺4.79兴 ⫺6.51 关⫺6.09兴 ⫺6.78 关⫺6.35兴 ⫺7.11 关⫺6.65兴 ¯
a
Reference 9.
d
Reference 2.
b
Reference 37.
e
Reference 7.
c
Reference 38.
f
Reference 8.
TABLE II. Vertical distance between the adsorbate and the plane of three
metal atoms defining the corresponding site on relaxes surfaces (d
A–Rh
),
displacement of the plane of three metal atoms defining the corresponding
sites with respect to the clean, relaxed surface (⌬z
Rh
), and metal–metal
bond length (d
Rh–Rh
) on the relaxed surface. A schematic representation is
given in Fig. 2共a兲. d
Rh–Rh
on a clean surface⫽ 2.71 Å.
Adsorbate
d
A–Rh
共Å兲
⌬z
Rh
共Å兲
d
Rh–Rh
共Å兲
H 共fcc兲 0.98 0.02 2.74
O 共fcc兲 1.23 0.08 2.75
N 共hcp兲 1.03 0.09 2.64
S 共fcc兲 1.66 0.06 2.74
C 共hcp兲 1.00 0.09 2.62
6739J. Chem. Phys., Vol. 117, No. 14, 8 October 2002 Atomic and molecular adsorption on Rh(111)
Downloaded 08 Feb 2007 to 128.104.198.19. Redistribution subject to AIP license or copyright, see http://jcp.aip.org/jcp/copyright.jsp
coverages, population of either one of these sites can be ob-
served. An experimental estimate of the binding energy can
be obtained by combining the dissociative adsorption energy
of molecular oxygen on Rh共100兲, ⫺4.00 eV, obtained from
femtomole adsorption calorimetry experiments,
39
and the
gas-phase O
2
bond enthalpy, ⫺5.17 eV.
37
These values give
an estimated oxygen binding energy of ⫺4.58 eV on
Rh共100兲, which is reasonably close to our findings on
Rh共111兲. Interestingly, there is quite a bit of controversy
about the magnitude of the O binding energy in previous
DFT calculations; values of ⫺5.03 eV,
26
⫺5.22 eV,
25
and
⫺5.51 eV 共Ref. 27兲 have been reported. As with atomic hy-
drogen, the effect of surface relaxation on the oxygen bind-
ing energy is small. The vertical displacement of the coordi-
nated rhodium atom is found to be 0.08 Å 共Table II兲,
consistent with a 0.06⫾ 0.05 Å experimental value.
2
The
oxygen to rhodium vertical distance, 1.23 Å 共Table II兲,is
also in agreement with LEED experimental findings of 1.24
⫾ 0.06 Å.
2
The favored binding site for atomic nitrogen is hcp with
a binding energy of ⫺5.31 关⫺4.87兴 eV 共Table I兲. This site
preference does not agree with previous DFT calculations
(
⫽ 0.25 ML) that predict that the most stable site is fcc
with a binding energy of ⫺4.55 eV.
26
This discrepancy might
be explained by differences between that calculation and
ours; the functionals 共Becke 88/Perdew 86 versus PW91兲,
the treatment of core electrons 共nonrelativistic frozen-cores
versus relativistically corrected pseudopotentials兲, and the
approximations for surface dynamics 共no surface relaxation
versus one surface layer relaxed兲 are different between the
two calculations. In addition, it is worth noting that DFT
calculations can sometimes lead to incorrect site preferences
for small adsorbates. This problem has been analyzed in de-
tail for the case of CO;
40
a similarly extensive study would
probably be needed to sort out the site preference discrepan-
cies in the present case, and such a study is beyond the scope
of this work. In any case, our computations yield a 0.25 eV
preference of the hcp site over the fcc site. The calculated
estimate for the diffusion barrier of atomic nitrogen is 0.68
关0.64兴 eV 共Table III兲. Surface relaxation yields an insignifi-
cant change in binding energy and a small vertical displace-
ment of the rhodium atoms 共Table II兲.
Even stronger bonding is exhibited by sulfur in fcc sites
with a binding energy of ⫺5.42 关⫺4.95兴 eV 共Table I兲.Asin
the case of atomic oxygen, sulfur has only a very weak pref-
erence 共⬇0.03 eV兲 for fcc over hcp sites. The estimated dif-
fusion barrier is 0.46 关0.46兴 eV 共Table III兲. LEED experi-
ments confirm that for coverages below 0.33 ML, sulfur
occupies fcc sites.
7,8
Previously reported DFT calculations
yield a binding energy of ⫺5.25 eV 共fcc adsorption兲,
28
in
reasonable agreement with the present findings. For the ad-
sorption in the fcc site, the change in binding energy due to
surface relaxation is small. The geometric parameters of the
preferred site are given in Table II.
The most strongly bound atom studied is carbon. The
preferred binding site is hcp with a binding energy of ⫺7.11
关⫺6.65兴 eV 共Table I兲. The binding energy of the next most
stable site, hcp, is 0.33 eV higher in energy. The barrier for
the diffusion of C is estimated to be 0.60 关0.56兴 eV 共Table
III兲. The change in binding energy due to surface relaxation
is negligible. The geometric parameters of the preferred site
are given in Table II.
For three of the atomic adsorbates discussed above 共H,
O, and S兲, the magnitudes of the calculated binding energies
共B.E.’s兲 for the hcp sites 共the second most favorable sites in
each case兲 are only 30–40 meV less than the corresponding
B.E.’s at the most favorable fcc sites. While we cannot say
with certainty that these differences are numerically signifi-
cant, the result that zero-point energy differences do not af-
fect the site preferences 共in the case of oxygen, at least兲, and
the fact that our findings agree well with experimental re-
sults, allow us to be reasonably confident about our calcu-
lated site preferences. Furthermore, we note that our data
support the explanation proposed by Ganduglia-Pirovano
and Scheffler
25
for the preference of oxygen for fcc sites.
Those authors note that the workfunction of the rhodium
surface decreases when adsorbed oxygen moves from the fcc
to the hcp site 共we calculate ⌬
⫽⫺0.14 eV). At the same
time, the oxygen s-orbital energy decreases 共we calculate
⌬
⑀
s
⫽⫺0.17 eV). Both of these observations suggest that
more charge is transferred from the surface to the oxygen
atom in the fcc site, thereby leading to a more ionic bond
with the surface and to a higher binding energy.
For the investigated atomic adsorbates a consistent pref-
erence for threefold sites is observed. This result indicates
that gas-phase bonding trends are not always followed by
atomic adsorbates on Rh共111兲. For example, oxygen and sul-
fur bind to at most two atoms in the gas phase while both
adsorb in fcc sites on Rh共111兲. Similarly, hydrogen only
binds to one atom in the gas phase, yet it prefers fcc sites on
the rhodium surface. Nitrogen typically forms three bonds in
the gas phase 共although four bonds can be observed in spe-
cies like ammonium兲, but it prefers hcp sites with four neigh-
boring atoms on Rh共111兲. In the case of carbon, however, the
gas phase tetravalent coordination is consistent with its pref-
erence for the hcp sites.
A general trend is observed in the geometry changes of
top-layer rhodium atoms during surface relaxation for the fcc
and hcp sites. For species adsorbed at fcc sites 共H, O, S兲,
Table II shows that there is an increase in d
Rh–Rh
from the
equilibrium value. In other words, movement of rhodium
atoms directed radially outward is observed for the fcc site.
This trend has been observed in LEED experiments of ad-
sorbed O on Rh共111兲.
2
On the other hand, for adsorbates at
hcp sites 共N, C兲, the corresponding d
Rh–Rh
decreases from
the equilibrium value. In this case there is movement of
TABLE III. Estimates for the diffusion barriers of atomic species on
Rh共111兲.
Adsorbate
Diffusion barrier 共PW91兲
共eV兲
Diffusion barrier 共RPBE兲
共eV兲
H 0.10 0.10
C 0.60 0.56
N 0.68 0.64
O 0.45 0.41
S 0.46 0.46
6740 J. Chem. Phys., Vol. 117, No. 14, 8 October 2002 Mavrikakis
et al.
Downloaded 08 Feb 2007 to 128.104.198.19. Redistribution subject to AIP license or copyright, see http://jcp.aip.org/jcp/copyright.jsp
rhodium atoms directed radially inward. We could not find
experimental data for comparison with these results.
For three of the above atomic adsorbates 共H, O, and S兲,
the energetic preference for the most favored site 共fcc, in
these cases兲 is very weak 共⬍0.05 eV, see above discussion兲.
Thus, it is possible that these adsorbates will occupy hcp
sites 共in addition to the weakly favored fcc sites兲 under a
variety of experimental conditions. In our calculations, we
observed that these adsorbates induced a radially inward
movement of rhodium atoms when placed at hcp sites. This
result is consistent with the trend observed for adsorbates
whose best site is clearly the hcp site 共N, C兲. The result
suggests that the movement of surface atoms does not de-
pend exclusively on the adsorbate in question and that a
more general statement about the relationship between adsor-
bates, site preferences, and surface atom movement can be
made. Namely, for atomic adsorbates where adsorption at fcc
or hcp sites is energetically competitive, fcc adsorption leads
to outward radial movement of the rhodium atoms, and hcp
adsorption leads to inward radial movement, irrespective of
whether or not the site in question is the most favorable site
for the adsorbate.
It is important to note that, while the above trends seem
to depend strongly on the adsorption site 共fcc or hcp兲, there
does exist a dependence on the identity of the adsorbate.
Some molecules do not follow the observed trend; ethyli-
dyne (CCH
3
), for example, binds in hcp sites on Rh共111兲 but
has been observed experimentally to induce outward radial
movement of rhodium surface atoms.
41
B. Adsorption of molecules and radicals
The least strongly bound of the molecular adsorbates is
dinitrogen with a maximum binding energy of ⫺0.68
关⫺0.34兴 eV at the top site, with the molecule perpendicular
to the surface 关Table IV and Fig. 3共a兲兴. We note here that
RPBE gives a stable N
2
molecular state only for the top site,
and thus estimation of a diffusion barrier for N
2
on Rh共111兲
is meaningless. The RPBE result is in excellent agreement
with an estimate of the binding energy based on thermal
desorption spectroscopy experiments, ⫺0.36 eV.
11
The effect
of surface relaxation on the binding energy is small, only
⫺0.06 eV. A significant upward shift of 0.20 Å is calculated
for the location of the rhodium atom coordinated to the N
2
molecule 关Table V, Fig. 3共a兲兴.
The methyl radical also favors the top site, binding
through the carbon atom with hydrogen atoms pointed to-
ward the neighboring hcp sites 关Fig. 3共b兲兴. A binding energy
of ⫺1.84 关⫺1.49兴 eV is calculated 共Table IV兲, and a small
diffusion barrier, 0.10 关0.20兴 eV 共Table VI兲, is estimated. The
effect of surface relaxation on the binding energy is very
small. The CH
3
-coordinated rhodium atom is shifted upward
by 0.12 Å 共Table IV兲.
Another diatomic molecule that prefers the top site is
carbon monoxide, bound through the carbon atom and ori-
ented perpendicular to the surface 关Fig. 3共c兲兴. A binding en-
ergy of ⫺2.04 关⫺1.68兴 eV is calculated 共Table IV兲, and a
diffusion barrier of 0.06 关0.14兴 eV is estimated. Surface re-
laxation leads to an upward shift of the CO-coordinated
rhodium atom by 0.20 Å 共Table V兲. Vibrational spectra of
adsorbed carbon monoxide and a core-level photoemission
study also indicate a preference for the top site at low
coverages.
15,16
A high resolution photoemission study indi-
cates that at coverages above 0.5 ML the threefold sites be-
come populated, consistent with our finding that the binding
FIG. 3. Preferred binding modes for molecular and radical adsorbates on
Rh共111兲; 共a兲 N
2
top, 共b兲 CH
3
top, 共c兲 CO top, 共d兲 NO hcp, 共e兲 OH fcc, 共f兲
OH bridge tilted.
TABLE IV. Binding energies of molecular and radical adsorbates on a relaxed Rh共111兲 surface. All configurations are perpendicular to the surface, unless
indicated otherwise. Best sites are shown by shaded entries. Reference zero corresponds to gas phase species at infinite separation from the rhodium slab.
NS⫽ not stable.
Best site
Binding energy 共eV兲
PW91 共RPBE兲
Experimental
findings
共eV兲
Calc. Expt. Top Bridge fcc hcp
N
2
top ¯ ⫺0.68 关⫺0.34兴 ⫺0.10 关0.30兴 NS NS ⫺0.36
a
CH
3
top ¯ ⫺1.84 关⫺1.49兴 ⫺1.74 关⫺1.29兴 ⫺1.40 关⫺0.97兴 ⫺1.32 关⫺0.91兴 ¯
CO top top
b,c
⫺2.04 关⫺1.68兴 ⫺1.91 关⫺1.49兴 ⫺1.89 关⫺1.45兴 ⫺1.99 关⫺1.54兴 ⫺1.65
d
NO hcp three fold
e,f
⫺1.98 关⫺1.53兴 ⫺2.25 关⫺1.73兴 ⫺2.26 关⫺1.72兴 ⫺2.39 关⫺1.83兴 ⫺1.48,
g
⫺1.17
h
OH fcc ¯ ⫺2.09 关⫺1.61兴 ⫺2.68 关⫺2.12兴 ⫺2.89 关⫺2.30兴 ⫺2.74 关⫺2.17兴 ¯
OH bridge ¯ ⫺2.65 关⫺2.17兴 ⫺2.89 关⫺2.33兴 ¯¯ ¯
共tilted兲
a
Reference 11.
e
Reference 5.
b
Reference 13.
f
Reference 18.
c
Reference 14.
g
Reference 22.
d
Reference 21.
h
Reference 20.
6741J. Chem. Phys., Vol. 117, No. 14, 8 October 2002 Atomic and molecular adsorption on Rh(111)
Downloaded 08 Feb 2007 to 128.104.198.19. Redistribution subject to AIP license or copyright, see http://jcp.aip.org/jcp/copyright.jsp
