Restless Transition Metal Surfaces in Catalysis

1
MEAM Parameters about their fitting, physics,
and flaws
Paul van Beurden and Gert Jan Kramer MEAM
workshop 24, 25 October 2002

• MEAM the Bond Order picture
• Fit to what?
• Trends in parameter values physical origin
• How to improve the MEAM

2
MEAM Parameters about their fitting, physics,
and flaws
Paul van Beurden and Gert Jan Kramer MEAM
workshop 24, 25 October 2002
/s
http//www.catalysis.nl/paul
3
Subjects addressed....

• Where do we start, what to fit to?
• Physical meaning of (some) parameters
• What goes (systematically) wrong?
• Open ends how en where to improve MEAM/fitting?

• Application to
• Pure fcc metals the Pt-group
• A-B systems CO on Pt ... And how about O-M?

2/16
4
The Modified Embedded-Atom Method
Ei Fi(n) ½?j?ij(rij)
n effective coordination number (electron
density spatial effects)
3/16
5
Fitting parameters for fcc (hcp?) metals

• Baskes
• Separate parameters as much as possible
• Use analytical expressions (for bulk
  properties), if possible

• However
• For most cases in simulations, interplay between
  parameters is crucial!
• These bulk fit-systems are highly symmetric and
  often insensitive to higher order parameters
  w(1-3), b(1-3)

4/16
6
Insensitivity of k 2 parameters to elastic
properties
DE is only 0.02 eV
5/16
7
Fitting parameters for fcc (hcp?) metals

• Better separate only w(k) and b(k)
• ws - via fit-data with bulk distances (rr0)
• all bs drop out because of exp-b(r/r0-1)
  terms
• (energy)
• bs - via relaxations (r?r0), with known ws
• (relaxations)
• Fit data should include surfaces and bulk-defects
  
• (e.g., vacancy energy, elastic properties)

6/16
8
Weighting parameters w importance of sign

• w(3)
• lt 0 for hcp and diamond cubic materials
• gt 0 for fcc and bcc metals

w(2) lt 0 for late 5d metals (Ir, Pt, Au) gt 0
for 4d metals (Rh, Pd, Ag)
Not magnitude but sign of w determines type of
metal!
7/16
9
What does w(2) do?

• w(2) lt 0 appears necessary to predict
• Low vacancy-formation-energy
• Evac/Esub Pt(w(2)lt0) 0.2 , Pd(w(2)gt0) 0.5
• Hex surface reconstruction on (100)
• Exchange diffusion of adatom on (100)

8/16
10
What is the physical origin of w(2) ?

• Fiorentini, Methfessel Scheffler and Yu
  Scheffler
• driving force for (100)-hex and exchange
  diffusion is
• tensile surface stress
• Caused by stronger bonding of low-coordinated
  atoms, a result of relativistic effects in
  electronic structure

w(2) reflects the influence of relativistic
effects in electronic structure
9/16
11
Systematic flaw of the MEAM Alternating
preference for adatom island or chain on Pt(100)
Found with EAM, not with MEAM!
Angular effects may worsen things at low
coordination!
10/16
12
How to improve MEAM?
Better physical justification/understanding of
w-parameters

• EAM tends to over-estimate binding of
  low-coordinated atoms
• Angular effects in MEAM are possibly too dominant
  for
• low-coordinated atoms

• Improve on effective coordination number n!
• n2 f (n(0))2 f -1G
• More specifically, the balance between n(0) and G
  at
• lower coordination (lt7)
• n(0) should decrease less
• G should decrease more

• Improve on effective coordination number n!
• n2 f (n(0))2 f -1G
• More specifically, the balance between n(0) and G
  at
• lower coordination (lt7)
• n(0) should decrease less
• G should decrease more

11/16
13
A-B systems CO on Pt

• CO binds at top or bridge only
• CO prefers (100) 8 over (110) 7 over (111)
  9
• Binds via C atom treat CO as effective atom
• CO-CO is pairwise repulsive E(CO-CO) F(r)

• For alloys F, n, Fii know for each element,
  only Fij is required
• For adsorbate/surface, only F, n, Fii for metal
  are known
• CO-CO differs too much from CO-Pt to use one set
  of F, n, F only
• Further, no reference structure available

12/16
14
A-B systems CO on Pt surfaces
13/16
15
A-B systems
It works! CO induced lifting of Pt(100)-hex
successfully modeled
14/16
16
Flaw of the Nearest-Neighbor approximation CO
binds too strong to (110)
Binding strength at top or bridge is determined
only by Pt atoms CO seesonly 1 or 2 Pt atoms
Recall binding of CO (100) 8 gt (110) 7 gt
(111) 9
CO violates Bond-Order Conservation Principle!
Extension to Next-Nearest-Neighbors necessary for
odd adsorbates to restore BOCP
15/16
17
General Conclusions
MEAM can model a wide range of surface
phenomena, inaccessible for other methods (incl.
EAM)
MEAM provides straightforward links between
properties, by means of sign of weighting
parameters
Of course, MEAM has its limitations, improvement
of n and incorporating NNN would help!
? Thanx to Barend Thijsse et al. for their
contributions ?
Thesis available 12/2002 at http//www.catalysis.n
l/paul
16/16