Tuning the Cation Composition of Ternary Metal Oxides for Light Harvesting Applications \ \ Aron Walsh University College London

Hydrogen: H2 O à H2 + ½O2 ; ΔG° = 237kJ/mol (1.23 eV) Electricity (Photovoltaics) • Production: Photolysis of H2 O (DOE target $2/kg by 2015). • Storage: Physical or chemical (MOFs, metal hydrides). • Energy conversion: Fuel cells (zero emission). “Heterogeneous photocatalyst materials for water splitting” Chem. Soc. Rev. 38, 253 (2009). “Photoelectrochemical hydrogen generation using solar energy” Int. J. Hydrogen Energy. 27, 991 (2002). Solar Energy Utilisation Chemical Energy (Photoelectrochemistry)

Experiment: Dr. Yanfa Yan (National Renewable Energy Laboratory) Prof. Eric Mac Farland (UC Santa Barbara) Prof. Bruce Parkinson (University of Wyoming) Theory: Dr. Su-Huai Wei (National Renewable Energy Laboratory) Prof. Richard Catlow (University College London) Project Collaborators

Photoelectrochemical Cell

Photoelectrode Criteria • Structural stability in solution. • High availability (low cost). • Good catalytic activity. • Band gap: 1.7 – 2.2 eV. • Photocurrent generation. • Band-edge alignment.

Theory and Experiment Simulations: Density Functional Theory. Key Properties: Band structure, optical spectra. Defect and doping effects, structural stability. Surface chemistry. Codes: VASP, WIEN2K, FHI-AIMS. Experiment: Growth and PEC characterization. RF co-sputtering: Co, Al, Ga, In, Cu, Fe targets. Substrates: Quartz / Ag coated stainless steel. Structure: Powder X-ray diffraction. PEC Performance: Three-electrode cell (W lamp with IR/UV filters).

Multiternary Spinel Oxides • Broad theoretical screening of new systems is limited by the lack of a single property defining a good PEC material. • Initial lead from experimental high-throughput screening by the Parkinson group at Colorado State University1. Schematic inkjet salt printing (left) and measured photoresponse (right), relative to the p-type and n-type standards. 1. M. Woodhouse et al., Chem. Mater. 17, 4318 (2005); ibid. 20, 2495 (2008). 2. A. Walsh et al., Phys. Rev. B 76, 165119 (2007); J.P.C. C 112, 12044 (2008). • This lead to the exploration of Co1+x (Al,Ga,In)2-x O4 spinels2.

Tailoring Transition Metal Spinels CoAl2 O4 CoGa2 O4 CoIn2 O4 Spinel: ATETBOCT 2 O4

Electronic Densities of States

Experiment: CoAl2 O4 400 500 600 700 800 900 1000 0.0 0.2 0.4 0.6 0.8 1.0 Wavelength (nm) Absorbance 500°C 800°C 400 500 600 700 800 900 1000 0.0 0.2 0.4 0.6 0.8 1.0 Wavelength (nm) Absorbance Co3 O4 + Al2 O3 CoAl2 O4 1.0 1.5 2.0 2.5 3.0 3.5 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 Al: 400 W Al: 300 W (Anneal @ 900c, 4 hr, air) Photon energy (eV) (αhv)2 (x1010 eV2cm-2) Al: 200 W 1.83 eV Ag epoxy Insulated Cu TCO Glass Cu Film 1. A. Walsh et al., Energy Environ. Sci. 2, 774 (2009).

Defect Chemistry of Co Spinels 1. A. Walsh et al., J. Phys. Chem. C 112, 12044 (2008). 2. C. Greskovich and H. Schmalzried, J. Phys. Chem. Solids 31, 639 (1970). O Co(II) O Vacancy creation1,2: Hole oxidation: Hole mobility: Hole mobility dependence2: (Eh = 0.8 eV in CoAl2 O4 ) (ΔH = 1 eV)

Enhancing Conductivity Spinel polaron conductivity: inversely proportional to d(Co-Co) 1. Increase VCo concentration for defect overlap (transition to band conductivity as in Co3 O4 ). 2. Increase crystal quality/decrease film thickness for improved conductivity (as in Cu2 O). 3. Produce n-type materials, where mobility should be significantly improved (Co(I) is an unstable species). Possible solutions to poor performance: Material a (Å) d (Co-Co) (Å) Δd (%) Co3 O4 8.08 2.86 0 CoAl2 O4 8.10 3.51 23 CoGa2 O4 8.32 3.60 26 CoIn2 O4 8.82 3.81 33

Alternative Ternary Pathways To raise valence band and provide catalytic activity, combine: • nd10 and nd0 cations: (Cu, Ag)x (Ti, W, V, Nb, Ta)y Oz • ns2 and nd0 cations: (Sn, Sb, Pb, Bi)x (Ti, V, Nb, Mo, W, Ta)y Oz Use chemical knowledge to avoid enumeration over all cations.

Delafossites: CuXIIIO2 First came to attention as p-type transparent conducting oxides, e.g. CuAlO2 , CuInO2 (Hosono, Nature 389, 939 1997). Benefits: Stable, conductive. Limitations: Poor optical absorption. CuCrO2 10 20 30 40 50 0 2 4 6 8 10 12 14 16 18 Bulk MesopCuCrO2 MesopMgCuCrO2 H 2 evolution µL/h Illumination hours

Delafossite Alloys: Cu(AIII,BIII)O2 M. N. Huda et al., Appl. Phys. Lett. 94, 251907 (2009).

Lone Pairs: Bismuth Vanadate A. Walsh et al., Chem. Mater. 21, 547 (2009). 0 Clinobisvanite (BiVO4 ): Bi(III) 6s2, V(V) 3d0

Lone Pairs: Sn2 TiO4 • Higher energy 5s states • Avoid (Sn,Ti)O2 • Sn(II), Ti(IV) compound isostructural to Pb3 O4 C. R. A. Catlow et al., Proc. Roy. Soc. A. In Press (2010).

Conclusions and Outlook • Millions of possible metal oxide combinations. • Many potential photocatalysts exist (beyond TiO2 ). Acknowledgements: EU for Marie-Curie Fellowship; All collaborators. • BiVO4 (Bi s / V d) • Sn2 TiO4 (Sn s / Ti d) • Co(Al,Ga,In)2 O4 (Co d / metal s) • CuCrO2 (Cu d / Cr d), Cu(Ga,Y)O2 (Cu d / Ga s) A. Walsh et al., Energy. Environ. Sci. 2, 774 (2009).

Bonus Slides

Hematite (Fe2 O3 ) Benefits: Cheap; good optical absorption, non-toxic. Limitations: Poor conductivity; low performance. A. Kleinman-Shwarzstein et al., Chem. Mater. 22, 510 (2010).

Delafossites: CuCrO2 T. Arnold et al., Phys. Rev. B 79, 075102 (2009). D. O. Scanlon et al., Phys. Rev. B 79, 035101 (2009). Cu d remains dominant at top of the valence band. p-type doping improves.

Lone Pairs: Bismuth Vanadate

Lone Pairs: Bismuth Vanadate A. Walsh et al., Chem. Mater. 21, 547 (2009). • s-p coupling gives valence band dispersion, uncharacteristic of typical metal oxides.