AdsorbateAdsorbate Interactions and Chemisorption at Different Coverages Studied by Accurate ab init

1
Adsorbate-Adsorbate Interactions and
Chemisorption at Different Coverages Studied by
Accurate ab initio Calculations CO on
Transition Metal SurfacesSara E. Mason, Ilya
Grinberg, and Andrew M. Rappe
Introduction and Motivation
Discussion of Results
Systems Modeled and Results

• In general we are interested in molecules on
  metal surfaces because the surface helps make and
  break bond to take reactant to product -
  heterogeneous catalysis
• Chemisorption is the process by which a molecule
  adsorbs to a surface through the formation of
  chemical bond(s)
• Prototypical CO/transition metal chemisorption
  systems continue to be of interest
• Accurate Modeling (The CO/metal Puzzle)
• Trends in bonding properties on one metal often
  to not carry over to other metals
• We focus here on Evolution of Chemisorption with
  Coverage (?)

Hammer-Morikawa Nørskov (HMN) model for the
d-band contribution to top site chemisorption
where f is the idealized filling of the metal
d-bands, ? are orbital energies, ?d denotes the
average energy of the d-band, referred to as the
d-band center, and V and S are coupling and
overlap matrix elements, respectively.
Is the difference between the low-coverage and
higher coverage
Hammer-Nørskov Model, Hammer et al. PRL (1996)
Nearest-Neighbor Interactions and Perturbations
In overlayer patterns with exclusive top site
occupation, through-space repulsion is the only
interaction. We compare Eint (the difference
between the low-coverage and higher coverage)
with Eint (the interaction energy of CO,
plotted). At nearest-neighbor separations for Pt
(2.76 Å), Pd (2.73 Å), and Rh (2.66 Å), Eint is
0.19, 0.21, and 0.27 eV, respectively. The
corresponding Eint values for tt (111) from the
chemsiorption data are 0.30, 0.32, and 0.32 eV,
respectively.
DFT-GGA with norm-conserving pseudopotentials
(opium.sourceforge.net) Ecut 50
Ry First-Principles Extrapolation Procedure for
Accurate Chemisorption Energies
gas
Methodology
gas
How CO Bonds to Metal Surfaces
The CO/Metal Puzzle
Blyholder Bonding Mechanism, Blyholder, J. Phys.
Chem (1964)

• DFT fails to predict the preferred adsorption
  site for CO on several FCC metal surfaces.
• Erroneously favors hollow site adsorption.

2p orbital
5s orbital
Overlayer patterns marked by indicate that
lateral forces on CO were not minimized.
Overlayer patterns for which Echem is reported in
italics indicates that pattern is experimentally
observed at that coverage (?).
O
O
back-bonding accepts e-
The first comprehensive evaluation of this
puzzle points to the conclusion that DFTcannot
be used to determine the correct adsorption site,
even though this is a basic property of CO/metal
surface interactions Feibelman et. al. J. Phys.
Chem. B 2001
C
C
Metal Atom Sharing Bonding Competition and
Electron Delocalization
donates e-
Bonding competitionWhen the surface metal atoms
involved in chemisorption participate in more
than one carbon-metal bond. We differentiate the
HMN model with respect to the external parameter
?,representing the number of chemisorption bonds
per metal atom
Metal Surface
(a)
(b)
(a)
(b)
First-Principles Extrapolation Procedure
Step 1 Generate pseudopotentials for C and O
that when used to model CO give different values
of ?ES-T Step 2 Determine DFT values of Echem
using different sets of C and O pseudopotentials
to get the slope of Echem vs. ?ES-T Step 3
Extrapolate fit of Echem vs. ?ES-T out to the CI
value of ?ES-T Corresponding y-value gives the
corrected chemisorption energy Example to left
is for adsorption at the HCP hollow on Pt(111)
The opposite signs of the two terms in the
derivative mean that any change in ?d has two
competing effects on Echem. This competition
vanishes as f --gt 1, so Echem on Pt and Pd
(f0.9) should be more sensitive to bonding
competition than Rh (f0.8).
DFT Data
(c)
(d)
(c)
(d)
Linear Fit
The change of Echem with bonding competition (?)
also depends on ?, the shift in ?d with respect
to ?. The effect is explored in the above table,
which shows that ? is much smaller for Rh than
for Pt or Pd.
Extrapolation
(a) Schematic of (111) surfaces with CO at
?0.25. Overlayer t is indicated by squares, b by
circles, and h by X''. The cell is indicated
by the shaded region. (b) Schematic of (111)
surface with CO at ? 0.5 in overlayer tt
(squares), in the bb1 overlayer pattern (circles)
and hh pattern (X''). (c) Schematic of the
(111) bb2 structure. (d)Schematic of (111) bb3
structure.
(a) Schematic of (100) surface with CO at 0.25.
Top site pattern t is indicated by squares,
overlayer b by circles, and h by X''. (b)
Schematic of (100) tt1, bb1, and hh1 overlayers
(? 0.5). (c) Schematic of the (100) tt2, bb2,
and hh2 overlayers. (d) Schematic of (100) bb3
structure with CO at ? 0.5.
CI value for DES-T
Extrapolated Echemcorr
S.E. Mason, I. Grinberg, and A.M. Rappe, Phys.
Rev. B Rapid Comm. 69 161401R (2004)
In addition to the destabilizing effect of
bonding competition, metal atom sharing can also
enable electron delocalization
bb1 No metal atom sharing, No delocalization
bb2 Metal atom sharing, Electron delocalization
(a) Schematic of (111) tb structure at ??0.5.
Occupied top sites are shown with squares, and
occupied bridge sites are shown with circles. (b)
Schematic of (111) hf and thf structures at
?0.5 and ?0.75. In the former, only hcp and
fcc sites (X'') are occupied. In the latter,
the top sites (squares) are also occupied.
Here for the first time we apply our
Extrapolation Procedure to CO adsorption at
multiple ? and in multiple patterns
Summary and Conclusions

• Our corrected values for Echem at higher ? with
  CO occupying different sites and patterns show
  the experimentally observed structures to be the
  most favorable
• Key interactions are identified as
• Adsorbate-adsorbare through-space repulsion
• Bonding Competition
• Substrate-mediated electron delocalization
• This work is in press with Journal of Physical
  Chemistry B

(a)
(b)
Acknowledgements AFOSR NSF-CRIF
HPCMO