Electronic excitations in elementary reactive processes at metal surfaces

Electronic excitations in elementary reactive processes at metal surfaces

Ricardo Díez MuiñoCentro de Física de MaterialesCentro Mixto CSIC-UPV/EHU

San Sebastián (Spain)

Contributors

Donostia - San SebastiánMaite Alducin (CSIC)Ricardo Díez Muiño (CSIC)Itziar Goikoetxea (PhD, CSIC)Iñaki Juaristi (UPV)Geetha Kanuvakkarai (Post-doc, CSIC)Ludovic Martin (Post-doc, DIPC)

OthersFabio Busnengo (Rosario, Argentina)Antoine Salin (Bordeaux, France)

outline

- brief introduction: adiabatic processes

- electronic excitations at the surface: friction

- electronic excitations at the molecule

gas/solid interfaces

an important goal is to understand how solid surfaces can be used to promote gas-phase chemical reactions

static properties (equilibrium)

- adsorption sites and energies- chemical bonding- induced reconstructions- self-assembling

experimental techniques: - LEED, STM, PE, etc.

dynamical properties

- reaction rates (adsorption, recombination, …)- diffusion- induced desorption- energy and charge exchange

experimental techniques: - molecular beams, TPD, etc.

dissociative adsorption

molecularadsorption

desorption

Fe (111)

Fe (100)

Fe (110)

rates of ammonia synthesis over five iron single-crystal surfaces

surface face and reactivity

N2 + 3H2 2NH3

0.25 0.50 0.75 1.000.00

0.25

0.50

0.75

1.00

W(100)

T=800K

T=300K

T=100K

impact energy Ei (eV)

0.25 0.50 0.75 1.000.00

0.25

0.50

0.75

1.00

W(110)

T=800K

normal incidence

Rettner et al. (1988)Beutl et al. (1997)

Pfnür et al. (1986)Rettner et al. (1990)

stic

king

coe

ffici

ent S

0dissociative adsorption of N2 on W(100) and on W(110)

0.25 0.50 0.75 1.000.00

0.25

0.50

0.75

1.00

W(100)

T=800K

T=300K

T=100K

impact energy Ei (eV)

0.25 0.50 0.75 1.000.00

0.25

0.50

0.75

1.00

W(110)

T=800K

normal incidence

Rettner et al. (1988)Beutl et al. (1997)

Pfnür et al. (1986)Rettner et al. (1990)

stic

king

coe

ffici

ent S

0dissociative adsorption of N2 on W(100) and on W(110)

why the difference in the N2 dissociation rate at low energies between

the (100) and (110) faces of W?

theoretical model: two stepstheoretical model: two steps

dynamics of diatomic molecules on metal surfaces: theory

adiabatic approximation frozen surface approximation 6D PES: V(X, Y, Z, r,, )

1.- calculation of the Potential Energy Surface (PES)

• extended set of DFT energy values, V(X, Y, Z, r,, )

• interpolation of the DFT data: corrugation reduction method

[Busnengo et al., JCP 112, 7641 (2000)]

6D PES construction

Ei

incidence conditions are fixed: (Ei,

sampling on the internal degrees of freedom: (X, Y, , ) and on (azimuthal angle of trajectory)

2.- classical trajectory calculations: Monte Carlo sampling

dissociation of N2 on W(110)

0.0 0.5 1.0 1.5 2.0 2.50.0

0.1

0.2

0.3

0.4

0.5 N2/W(110)

i=0

0.5 1.0 1.5 2.0 2.5

N2/W(110)

i=60

impact energy Ei (eV)

stic

king

coe

ffici

ent S

0

Alducin et al., PRL 97, 056102 (2006); JCP 125, 144705 (2006)

Pfnür et al. (1986)Rettner et al. (1990)

theory

0 25 50 75 1000.00

0.25

0.50

0.75

1.00

final

Z=2.0

Z=3.25

W(110)

final

Z=2.0

Z=3.25

W(100)

impact energy Ei (meV)

prob

abili

ity o

f rea

chin

g Z

Z=2Å

Z=3.25Å

why N2 abundantly dissociate on W(100) and not on W(110)

Alducin et al., PRL 97, 056102 (2006); JCP 125, 144705 (2006)

1 1 .2 1 .4 1 .6 1 .8 2 2 .2

1 .5

2

2 .5

3

3 .5

4

=0o

r(Å)

Z(Å

)

potential well

Difficult access to the precursor well

Once in the precursor well, molecules easily dissociate

in summary, dynamics matters

outline

- brief introduction: adiabatic processes

- electronic excitations at the surface: friction

- electronic excitations at the molecule

theoretical model: two stepstheoretical model: two steps

dynamics of diatomic molecules on metal surfaces

adiabatic approximation frozen surface approximation 6D PES: V(X, Y, Z, r,, )

calculation of the Potential Energy Surface (PES)

• extended set of DFT energy values, V(X, Y, Z, r,, )

• interpolation of the DFT data: corrugation reduction method

[Busnengo et al., JCP 112, 7641 (2000)]

6D PES construction

Ei

incidence conditions are fixed: (Ei,

sampling on the internal degrees of freedom: (X, Y, , ) and on (azimuthal angle of trajectory)

classical trajectory calculations: Monte Carlo sampling

We are assuming the validity of the Born-Oppenheimer approximation

non-adiabatic effects: electron-hole pair excitations

chemicurrentschemicurrents

Exposure (ML)

Gergen et al., Science 294, 2521 (2001).

vibrational promotion of electron transfer

vibrational promotion of electron transfer

Huang et al., Science 290, 111 (2000) White et al., Nature 433, 503 (2005)

NO on Cs/Au(111)electron emission asa function of initial

vibrational state

friction in N2/Ru(0001) friction in N2/Ru(0001)

Luntz et al., JCP 123, 074704 (2005)Díaz et al., PRL 96, 096102 (2006)

adiabatic approximationfor H2 on Pt(111)

adiabatic approximationfor H2 on Pt(111)

Nieto et al., Science 312, 86 (2006).

description of electronic excitations by a friction coefficient

classical equations of motion

mi(d2ri/dt2)=-dV(ri,rj)/d(ri) – (ri)(dri/dt)

for each atom “i” in the molecule

friction coefficient

adiabaticforce:

6D DFT PES

- damping of adsorbate vibrations: Persson and Hellsing, PRL49, 662 (1982)

- dynamics of atomic adsorption Trail, Bird, et al., JCP119, 4539 (2003)

- dissociation dynamics (low dimensions) Luntz et al., JCP 123, 074704 (2005)

previously used for:

description of electronic excitations by a friction coefficient

classical equations of motion

mi(d2ri/dt2)=-dV(ri,rj)/d(ri) – (ri)(dri/dt)

for each atom “i” in the molecule

friction coefficient

adiabaticforce:

6D DFT PES

friction coefficient: effective medium approximation

n(z)

zn0

bulk metal

n0

=n0kFtr(kF)

effective medium: FEG with electronic density n0

- damping of adsorbate vibrations: Persson and Hellsing, PRL49, 662 (1982)

- dynamics of atomic adsorption Trail, Bird, et al., JCP119, 4539 (2003)

- dissociation dynamics (low dimensions) Luntz et al., JCP 123, 074704 (2005)

previously used for:

probability of dissociative adsorption: N2 on W(110)

stic

kin

g co

effi

cie

nt

initial kinetic energy (eV)

polar angle of incidence

i=0

i=45

i=60

0,2

0,4

0,2

0,4

0,5 1,0 1,5 2,0 2,50,0

0,1

0,2

0,3

adiabatic

probability of dissociative adsorption: N2 on W(110)

non-adiabatic

adiabatic

stic

kin

g co

effi

cie

nt

initial kinetic energy (eV)

polar angle of incidence

i=0

i=45

i=60

0.2

0.4

0.2

0.4

0.5 1.0 1.5 2.0 2.50.0

0.1

0.2

0.3

Juaristi et al., PRL 100, 116102 (2008)

but for this system, dissociation is

roughly decided atZ=2.5A

(low energies)

but for this system, dissociation is

roughly decided atZ=2.5A

(low energies)

1 1 .2 1 .4 1 .6 1 .8 2 2 .2

1 .5

2

2 .5

3

3 .5

4

probability of dissociative adsorption: H2 on Cu(110)

non-adiabatic

adiabatic

0.25

0.50

0.75

1.00

0.25

0.50

0.75

stic

kin

g co

effi

cie

nt

initial kinetic energy (eV)

0.5 1.0 1.5 2.00.00

0.02

0.04

0.06

polar angle of incidence

i=0

i=45

i=60

[Salin, JCP 124, 104704 (2006)]

Juaristi et al., PRL 100, 116102 (2008)

classical equations of motion

mi(d2ri/dt2)=-dV(ri,rj)/d(ri) – (ri)(dri/dt)

for each atom “i” in the molecule

friction coefficient velocity

n(z)

zn0

bulk metal

why the excitation of electron-hole pairs is not relevant

energy loss of reflected molecules: N2 on W(110)

i=0i=45i=60

energy losses in the reflected molecules due to electronic excitations are < 100 meV

Ei=1.5 eV

Juaristi et al., PRL 100, 116102 (2008)

Energy loss of reflected molecules: N2 on W(110)

• Ts=1200K

• Trot <5K (J=0)

• Normal incidence

and detection

Experimental conditions

Experimental data by Hanisco et al.

J.Vac.Sci.Technol. A 11 ,1907 (1993)

Energy loss of reflected molecules: N2 on W(110)

• Ts=1200K

• Trot <5K (J=0)

• Normal incidence

and detection

Experimental conditions

Experimental data by Hanisco et al.

J.Vac.Sci.Technol. A 11 ,1907 (1993)

adiabatic

electronic friction

Energy loss of reflected molecules: N2 on W(110)

• Ts=1200K

• Trot <5K (J=0)

• Normal incidence

and detection

Experimental conditions

Experimental data by Hanisco et al.

J.Vac.Sci.Technol. A 11 ,1907 (1993)

GLO: phonons

Phonon excitationsare responsible for the

energy transfer observed experimentally

adiabatic

conclusions

• A local description of the friction coefficient shows that electronic excitations play a minor role in the dissociation of N2/W and H2/Cu: The Born-Oppenheimer approximation remains valid in these systems.

• Open questions still remain about the role of electron-hole pair excitations in other situations, in which non-adiabatic effects are due to the crossing of two or more potential energy curves, with possible transfer of charge included.

outline

- brief introduction: adiabatic processes

- electronic excitations at the surface: friction

- electronic excitations at the molecule

O2 on metal surfaces

Yourdshahyan et al., PRB 65, 075416 (2002)

Behler et al., PRL 94, 036104 (2005)Carbogno et al. PRL 101, 096104

(2008)

non-adiabatic effectsin the incoming O2 molecule

electronic excited statesmay play a role

excited state

triplet O2

singlet O2

3g

1 eVtriplet to singlet excitation energy(gas phase O2)

1g

we prepare the incoming O2 moleculein an excited state

: can we enhance dissociation?

adsorption of O2 on flat Ag surfaces Ei

Ts < 150K: O2 adsorbs only molecularly (Ei < 1eV)

• Ag (100)/Ag(110)

L. Vattuone et al., Surf. Sci. 408, L698

(1998).

A. Raukema et al., Surf. Sci. 347, 151 (1996).

70%

Low probability

• Ag (111)

molecular beam experiments

O2/Ag(100) - theoretical calculations

Ei

incidence conditions are fixed: (Ei, )

Monte-Carlo sampling on the internal degrees of freedom: (X, Y, , ) and on (parallel velocity)

Born-Oppenheimer approximation

frozen surface approximation 6D PES: V(X, Y, Z, r,, )

calculation of the Potential Energy Surface (PES)

classical trajectory calculations

XY

Z

x

y

z

surface unit cell

r

building the 6D PES

• about 2300 spin-polarized DFT values

• interpolation of the DFT data: Corrugation reducing procedure[Busnengo et al., JCP 112, 7641 (2000)]

numerical procedure

• O2 in vacuum spin-triplet ground state: • DFT - GGA (PW91) calculation with VASP

• plane-wave basis set and US pseudopotentials

• periodic supercell: (2 x 2) and 5-layer slab

DFT energy data

z

top

hollow

bridge

x

y

top view front view

dissociation probability of spin-triplet O2/Ag(100)

Alducin, Busnengo, RDM, JCP 129, 224702 (2008)

dissociation probability of spin-triplet O2/Ag(100)

Alducin, Busnengo, RDM, JCP 129, 224702 (2008)

2D Potential energy surface

Energy barrier: E~ 1.1 eV

Position: Z≈1.5 Å

General features:

• Activation energy: ~1.1eV

• Low dissociation probability

Reason:

Only configurations around

bridge lead to dissociation

Alducin, Busnengo, RDM, JCP 129, 224702 (2008)

dissociation probability of spin-triplet O2/Ag(100)

the question

Gas phase O2

3g

1 eVtriplet to singlet excitation energy

1g

can we enhance O2 dissociation on clean Ag(100) ?

6D PES calculation: Non spin polarized DFT

differences between SP and NSP PESs

Non spin polarized

Spin polarized

1 eVtriplet to singlet excitation energy

Gas phase O2

3g

1g

XY

Z

x

y

z

surface unit cell

r

dissociation is enhanced for singlet O2

spin-triplet O2 spin-triplet O2 spin-triplet O2

spin-singlet O2 spin-singlet O2 spin-singlet O2

dissociation occurs for Ei < 1 eVdissociation can increase in one order of magnitude

Alducin, Busnengo, RDM, JCP 129, 224702 (2008)

dissociation occurs for Ei < 1 eVdissociation can increase in one order of magnitude

spin-triplet O2 spin-triplet O2 spin-triplet O2

spin-singlet O2 spin-singlet O2 spin-singlet O2

Gas phase O2

3g

1 eVtriplet to singlet excitation energy

1g

But there is a trick here!The total energy is 1 eV larger

for the singlet O2

dissociation is enhanced for singlet O2

Alducin, Busnengo, RDM, JCP 129, 224702 (2008)

spin-triplet O2 spin-triplet O2 spin-triplet O2

spin-singlet O2 spin-singlet O2 spin-singlet O2

1 eV 1 eV 1 eV

dissociation occurs for Ei < 1 eVdissociation can increase in one order of magnitude

for ≠ 0o, singlet-O2 is more efficient than triplet-O2 with the same total energy

dissociation is enhanced for singlet O2

why is that?

1 eV

1 eV

spin-triplet O2

spin-singlet O2

available paths to dissociation are different(and more!)

spin-triplet O2

spin-singlet O2

why is that?

it is not the same road

triplet O2 singlet O2

+ 1 eV

conclusions

• Dissociation increases in about one order of magnitude, if singlet–O2 molecular beams are used.

• The enhancement of the dissociation rate is not only due to the extra energy that we are adding to the system. A different spin state in the incoming O2 molecule opens new paths to dissociation.

thank you for your attentionthank you for your attention

DIPC(2000)

CIC Nanogune (June 2007)CFM

(Febr. 2008)

chemistry faculty UPV/EHU (80’s)

Korta UPV/EHU (2007)

differences between SP and NSP PESs

Non spin polarized

Spin polarized

1 eVtriplet to singlet excitation energy

Gas phase O2

for Z < 2A, the SP and NSP PESs merge

3g

1g

XY

Z

x

y

z

surface unit cell

r

0 25 50 75 1000.00

0.25

0.50

0.75

1.00

final

Z=2.0

Z=3.25

W(110)

final

Z=2.0

Z=3.25

W(100)

impact energy Ei (meV)

prob

abili

ity o

f rea

chin

g Z

why N2 abundantly dissociate on W(100) and not on W(110)

Alducin et al., PRL 97, 056102 (2006); JCP 125, 144705 (2006)

it is not the same road

triplet O2 singlet O2

+ 1 eV

tripletO2

singletO2

Time-dependent density functional theory: Energy loss and friction

R

z axis

• TDDFT

-calculation from the force:-calculation from the energy:

time-dependent evolution of the electronic density

-25 -20 -15 -10 -5 0 5 10 15 20 25

20

15

10

5

0

-5

-10

-15

-20

z (a.u.)

x (a.u.)

-0.015 -0.005 0.005

n-n0 (a.u.)

cluster: N=254 rs=2 Rcl=12,67 a.u.

antiproton velocity: v=1,5 a.u.

TDDFT calculation of the induced electronic density

Borisov et al., Chem. Phys. Lett. 387, 95 (2004)Quijada et al., Phys. Rev. A 75, 042902 (2007)

Time-dependent density functional theory: Energy loss and friction

antiproton with v=2 au interacting with a metal cluster N=18induced dissipative electronic density ndiss=nTDDFT-nadiab

Time-dependent density functional theory: Energy loss and friction

exchange correlation functionals in DFT: testing the full potential energy surface

Exp: Pfnür et al., JCP 85 7452 (1986)

PW91

the RPBE XC functional:

- fits better the experimental

chemisorption energies

- describes better the interaction

very near the metallic surface

- but worst in the long distances!

stic

king

coe

ffic

ient

initial kinetic energy (eV)

0,0 0,5 1,0 1,5 2,0 2,50,0

0,1

0,2

0,3

0,4

0,5

polar angle of incidence

i=0

0,0 0,5 1,0 1,5 2,0 2,50,0

0,1

0,2

0,3

0,4

0,5

RPBEBocan et al., JCP 128, 154704 (2008)

N2/W(110)

W(100) W(110)

adsorption energyDFT = 7.4 eV

Exp. = 6.6-7 eVadsorption distance

DFT = 0.63 Å

adsorption energyDFT = 6.8 eVExp. = 6.6 eV

adsorption distanceDFT = 1.15 Å

final state features: N adsorption on W

approach to surface: vertical over a surface atom

2.62Z(Å): 2.5

1091 meV

223 meV

868 meV

2.651.9Z(Å):

705 meV

395 meV

310 meV

W(100) W(110)

dynamic trapping in a well in both faces

Friction coefficient for a slow atomic particletraveling through a free electron gas ()

- The potential created by the projectile in the electron gas of density n0 is calculated using density functional theory for a static atomic particle of atomic number Z1 embedded in a free electron gas (non-linear theory of screeening)

Z1

v

The stopping power S (the energy loss per unit path length) is calculated in terms of the momentum transferred per unit time to a uniform current (n0v) of electrons scattered by the screened static impurity:

j = n0v

Z1

S = n0v vF tr(vF)

vF: Fermi velocity

tr(vF): Transport cross section at the Fermi level

l(vF): scattering phase-shifts at the Fermi level = S / v

Description of electronic excitations by a friction coefficient Friction coefficient: effective medium approximation

n(x,y,z)

z

bulk metal

Dissociative adsorption: Role of electron-hole pair excitations

Non-Adiabatic Electronic Effects in dissociative adsorptionGRC-Dynamics at surfaces

2009

n0

n0

= n0 kFtr(kF)

Effective medium approximation: FEG with electronic density n0

Friction coefficient of an atom in a FEG

Transport cross section of the

electrons scattered by the atom FEG Fermi

momentum

FEG electron density

unidad de física de materialesUFM

energy & force along the particle trajectory

energy transfered to the cluster force on the antiproton

-0,4

-0,3

-0,2

-0,1

0

0,1

0,2

0,3

-150 -100 -50 0 50 100 150

For

ce (

atom

ic u

nits

)

antiproton position (atomic units)

-1

0

1

2

3

4

5

6

7

-150 -100 -50 0 50 100 150

Ene

rgy

(ato

mic

uni

ts)

antiproton position (atomic units)

E

antiproton with velocity v=1,5 a.u. and cluster with rs=2 and 254 electrons (Rcl=12,67 a.u.)

unidad de física de materialesUFM

size dependence of the Energy loss

0,001

0,01

0,1

0,1 1 10

rs= 2

18 electrons 58 electrons138 electrons254 electronsFEG (linear theory)FEG (non-linear theory)

effe

ctiv

e st

oppi

ng p

ower

(a.

u.)

velocity (a.u.)

0,001

0,01

0,1

0,1 1 10

18 electrons 58 electrons132 electrons254 electronsFEG (linear theory)FEG (non-linear theory)

effe

ctiv

e st

opp

ing

pow

er (

a.u.

)

velocity (a.u.)

rs= 4

unidad de física de materialesUFM

size dependence of the Energy loss

0,001

0,01

0,1

0,1 1 10

rs= 2

18 electrons 58 electrons138 electrons254 electronsFEG (linear theory)FEG (non-linear theory)

effe

ctiv

e st

oppi

ng p

ower

(a.

u.)

velocity (a.u.)

•“Universal” behaviour

•Size dependence for small clusters and low velocities

Explanation:

-0,4

-0,3

-0,2

-0,1

0

0,1

0,2

0,3

-40 -20 0 20 40

For

ce (

atom

ic u

nits

)

antiproton position (atomic units)

-local process inside the cluster

-size effects due to interference with reflections of the travelling perturbation

unidad de física de materialesUFM

summary & conclusions

Size effects appear only when the reflection of the electronic excitation at the cluster surface interfers with the screening hole that accompanies the antiproton along its trajectory

Universal behaviour of the energy loss as a function of projectile velocity, independent of size and in agreement with the bulk curves, for nanoparticle sizes as small as ~1 nm

Most of the contribution to the stopping of the antiproton takes place inside the cluster. The process is very local, which explains the universal behaviour.

Size effects found only for velocities v~0.1 a.u. and small clusters