Designing the electronic structure and reactivity of metal oxides

through strain, d-band filling and oxidation state

John Kitchin, Paul Salvador Department of Chemical Engineering, Carnegie Mellon

University, Pittsburgh, PA 15213 Department of Materials Science and Engineering, Carnegie Mellon University, Pittsburgh, PA 15213

Funding sources: DOE Early Career Award DOE SECA program DOE NETL-RUA

NSF funding: Proposal is pending

Why do we need to control metal oxide reactivity?

• Applications of oxides – Catalysis – Energy – Electronics – Thermal barrier coatings – Corrosion resistant coatings – Paints – Personal products

• In some applications we want reactivity, in others we do not

• How do we design materials that do what we want?

Understanding the reactivity of d-metals

d-band

Fermi Level

Adsorbate

resonance

Ni, Pt, transition metals

Strong bonding

En

erg

y

B

d-band

Fermi Level

Adsorbate

resonance

Au - Noble metals

Very weak bonding

B

• d-band properties dominate the trends in adsorption behavior • Changing d-band properties changes adsorption behavior

Alloying leads to combinations of strain and ligand effects

Straining metal films leads to changes in d-

band width

“Ligand” effects lead to changes in d-band

width

tensile

compressive

Kitchin, J.R., J.K. Nørskov, M.A. Barteau, and J.G. Chen, Physical Review Letters, 2004. 93(15): 156801. Kitchin, J.R., J.K. Nørskov, M.A. Barteau, and J.G. Chen, The Journal of Chemical Physics, 2004. 120(21): 10240.

Electronic structure to chemistry

Determined by active site composition and geometry

Estimated reactivity Correlation due to constant d-band filling

We are at a stage where we know a lot about metals and their reactivity

İnoğlu, N. and J.R. Kitchin. Physical Review B, 2010. 82(4): 045414. I ̇noğlu, N. and J.R. Kitchin. ACS Catalysis, 2011. 1(4): 399-407.

So far, nothing like this exists for metal oxides

Why metal oxides are harder than metals

• Many more different crystal structures and stoichiometries

• Many different defect and oxidation states that can be determined by the environment

• More complex bonding – Covalent and ionic

• Different adsorption mechanisms

Are there simple concepts in oxide reactivity that are analogous to metals?

– Can strain be used to control reactivity?

– What electronic structure features dominate trends in reactivity?

Part I – Reactivity of supported thin oxide films

• The conductivity of thin oxide films varies in reactive environments (gas sensors)

• Conductivity is related to concentration of oxygen vacancies • Sensitive probe to surface reactivity • Focus here on La0.7Sr0.3MnO3 (LSM) films

– Perovskite used in solid oxide fuel cells, catalysis

Thin films synthesized by pulsed laser deposition

Havelia, S., K.R. Balasubramaniam, S. Spurgeon, F. Cormack, and P.A. Salvador. Journal of Crystal Growth, 2008. 310(7-9): p. 1985-1990.

Heteroepitaxial growth of oxides to introduce strain

• Film – La0.7Sr0.3MnO3 (LSM) a=3.876 Å

• Single crystal substrates – SrTiO3 (100) a=3.905 Å

(heteroepitaxial growth leads to tensile strain)

– NdGaO3 (110) a=3.864 Å (heteroepitaxial growth leads to compressive strain)

• Grow films with varying thickness – Films are smooth and polycrystalline

Characterizing the strain in thin films

Thick, relaxed film

Epitaxial, thin film under tensile strain

Epitaxial, thin film under compressive strain • XRD rocking curves

characterize the lattice parameters of the films

• FWHM of rocking peaks related to defect density

• Thin films (50 nm) are epitaxial and fully strained at the substrate lattice constant

• Thick films relax by defect formation

Strain has a big impact on reactivity

• Tensile strained films have larger kchem (higher O2 exchange rate) than compressively strained films

• Order of magnitude difference!

The question is: how do we understand why? 1. Are vacancies easier to form under strain? 2. Is adsorption more favorable under strain? 3. Is this unique to LSM? How would changing the composition affect reactivity?

Tensile

Compressive

Part II – Computational investigations of oxide reactivity

• Trends in properties of cubic 3d-perovskites – Composition

• B = Sc through Cu • LaBO3 (B+3) • SrBO3 (B+4)

– Strain • At lattice constants of ±5%, ±2.5% and DFT equilibrium volumes

– Electronic structure • Properties of the atom-projected d-band density of states

– Reactivity • Dissociative adsorption energy • Surface oxygen vacancy energy

• DACAPO (DFT/US-PP/planewaves) – No hybrid functionals or DFT+U, no spin-polarization

Oxide reactivity with oxygen

• Adsorption and vacancy formation energies are reactivity prototypes

• We anticipate that trends in these prototype reactions will be representative of other types of reactivity

Clean slab

Reactive surface

Adsorbed oxygen atom

Missing oxygen

2

2

*

2

2

slab + 1/2 O slab+O

slab slab+vac + 1/

( ) 1/ 2

2

1/ 2

O

ads slab O slab O

vac slab vac O slab

E eV E E E

E E E E

12

Perovskite electronic structure evolution with strain

• Strong hybridization between B atom d-orbitals and O s,p orbitals

• Convergence of bands occurs as the B-atom d-band filling increases

• Tensile strain causes band narrowing

13

LaTiO3

LaCuO3

d-band widths of B atom in strained perovskites

– d-s/p orbital

coupling

• The strain dependence of the bulk d-band width depends on the size of the B atom

3

2

7

2

ddw

r

d

0

0.5

1

1.5

2

2.5

0.075 0.095 0.115 0.135

d-b

and

wid

th (

Wd

)

d-7/2

LaScO3LaTiO3LaVO3LaCrO3LaMnO3LaFeO3LaCoO3LaNiO3LaCuO3

0

0.5

1

1.5

2

2.5

0.075 0.095 0.115 0.135

d-b

and

wid

th (

Wd

)

d-7/2

SrTiO3SrVO3SrCrO3SrMnO3SrFeO3SrCoO3SrNiO3SrCuO3

LaBO3

SrBO3

14

Pseudo-rectangular band model for perovskite d-band

• The ABO3 series shows a wd/d correlation consistent with a rectangular band model

• Nearly constant d-band filling observed

• This is the same behavior as observed in metals

-3

-2

-1

0

1

2

3

4

5

6

7

8

0.8 1.3 1.8 2.3

Sc

Ti

V

Cr

Mn

Fe

Co

Ni

Cu

d-band width (eV)

d-b

and

cen

ter(

eV

)

-1

0

1

2

3

4

5

0.8 1.3 1.8 2.3

Ti

V

Cr

Mn

Fe

Co

Ni

Cu

d-band width (eV)

d-b

and

cen

ter(

eV)

LaBO3

SrBO3

15 Akhade, S.A. and J.R. Kitchin,. J. Chemical Physics, 2011. 135(10): p. 104702.

Electronic structure-reactivity correlations of adsorption energy

• For most perovskites the adsorption energy is related to the d-band center of the B-atom

• Higher oxidation states tend to have weaker adsorption energies

• LaScO3 and SrTiO3 are different – Probably an error in

the pseudopotentials which do not have semicore states

16

LaCuO3

LaTiO3

LaVO3

LaCrO3

LaMnO3

LaFeO3

LaScO3

LaCoO3

LaNiO3

LaScO3

LaTiO3 LaVO3

LaCrO3

LaMnO3

LaFeO3

LaCuO3

LaCoO3

LaNiO3

LaCuO3

LaTiO3

LaVO3 LaCrO3

LaMnO3

LaFeO3

LaScO3

LaCoO3 LaNiO3

-6.00

-5.00

-4.00

-3.00

-2.00

-1.00

0.00

1.00

2.00

3.00

4.00

-2.00 -1.00 0.00 1.00 2.00 3.00 4.00 5.00

En

ger

y (

eV)

d-band center (ed)

Eads (O2) strain = -5%

Eads (O2) equilibrium

Eads (O2) strain =5%

SrCuO3

SrTiO3

SrVO3

SrCrO3

SrMnO3

SrFeO3

SrNiO3

SrCoO3

SrNiO3

SrTiO3

SrVO3

SrCrO3

SrMnO3

SrFeO3

SrCuO3

SrCoO3

SrCuO3

SrTiO3

SrVO3 SrCrO3

SrMnO3

SrFeO3

SrNiO3

SrCoO3

-6.00

-5.00

-4.00

-3.00

-2.00

-1.00

0.00

1.00

2.00

3.00

4.00

-1 0 1 2 3 4

En

ger

y (

eV)

d-band center (ed)

Eads (O2) strain = -5%

Eads (O2) equilibrium

Eads (O2) strain =5%

Chemical correlations in reactivity

• Adsorption is strongest on materials where oxygen vacancies are hard to make

• Correlated properties make it difficult to identify key factor(s) in reactivity

• We know other correlations also exist

17

0

2

4

6

8

-5 0 5

5%

0%

-5%

Dissociative Adsorption energy (eV)

Vac

ancy

fo

rmat

ion

en

ergy

(eV

)

-2

-1

0

1

2

3

4

5

6

-5 0 5

5%

0%

-5%

Dissociative adsorption energy (eV) Vac

ancy

fo

rmat

ion

en

ergy

(eV

)

LaBO3

SrBO3

Correlations can limit performance Example: oxygen evolution on oxides

• Correlations in reaction energies lead to limits in reactivity

2H2O → HO* + H2O + H+ + e-

HO* + H2O + H+ + e- → O* + H2O + 2H+ + 2e-

O* + H2O + 2H+ + 2e- → HOO* + 3H+ + 3e-

HOO* + 3H+ + 3e- → O2 + 4H+ + 4e-

2 H2O → 4(H+ + e-) + O2

Man, I. C., Su, H.-Y., Calle-Vallejo, F., Hansen, H. A., Martínez, J. I., Inoglu, N. G., Kitchin, J., Jaramillo, T. F., Nørskov, J. K. and Rossmeisl, J. (2011), Universality in Oxygen Evolution Electrocatalysis on Oxide Surfaces. ChemCatChem, 3: 1159.

Thermodynamic limit

Actual material limit

Conclusions • The bulk electronic structure of perovskite

B-atoms is reasonably modeled as a rectangular band

• d-band filling is the dominant factor in determining reactivity, followed by oxidation state

• The oxygen adsorption energy and vacancy formation energies are correlated in this class of perovskites

• Correlations limit degrees of freedom for materials design

• The outlook for systematically understanding oxide reactivity is very bright!

19