Conductivity Limits of Extrinsically Doped SnO2 Supports

Abstract

Supported catalyst technologies made Pt-based Polymer-Electrolyte Membrane Fuel Cell (PEMFC) electrodes cost-competitive. It has, however, become clear that current standard supports (i.e., high surface area carbons) might not provide the corrosion resistance needed to achieve life-time targets for stationary and/or automotive applications [1]. This has sparked strong interest in alternatives in recent years. Research ranges from exploring different allotropes of carbon (e.g., graphitised carbons, carbon-nano-tubes, and graphene) to new chemistries such as oxides [2]. Especially, oxides in their highest valence state appear attractive for their likely stability under strongly oxidising conditions. Indeed a number of metal oxides promise thermodynamic stability over the full operating range of a PEFC [3]; amongst them TiO2, Bi2O3, and SnO2also appear economically attractive. Being fully oxidised, all of them are intrinsic wide band-gap semi-conductors or even insulators. Hence, achieving and retaining near metallic conductivity over the life-time of a PEFC is a major challenge for oxide supports. We have investigate the possibility of extrinsic n-type doping to tailor the conductivity of SnO2 using Hybrid Density-Functional-Theory (DFT) as implemented in Crystal09. We find that Ta-doping provides limited potential to tailor conductivity of SnO2 for two reasons: (1) the stoichiometric ternary SnTa2O6 competes for thermodynamic stability along the rutile SnO2-TaO2 pseudo-binary potentially limiting the solubility of Ta in SnO2, and (2) collaborative Jahn-Teller distortions [4] tend to localise the Ta donor state leading to a freezing out of the donor state with increasing dopant concentration. Our calculations indicate that the maximum extrinsic carrier concentration should be around 1% Ta-doping. The use of B3LYP as function provided a significantly more accurate description of the band gap of SnO2than semi-local approximations to DFT produce. This enabled accurate description of electron localisation of the donor state without erroneous resonances with the conduction band. Electron localisation is accompanied with a strong Jahn-Teller distortion. Therefore, Ta centres interact via long-ranged elastic effects providing relatively large stabilisation of ordered superlattices. A cluster expansion based search strategy was implemented to bias the search towards these low energy orderings. This can be seen as a collaborative distortion of the crystal structure that leads to a freezing out of the donor state with increasing Ta concentration. Further, the Ta donor state is predicted to be a deep donor state that can only contribute to conductivity at room temperature if secondary defects provide a significant lowering of the conduction band. Using thin-film deposition techniques [5], we have synthesised doped SnO2 thin films. Consistent with our First Principles calculations, we find a declining conductivity in the range 1.0% to 7.1% Ta-doping. Interestingly, our experiments indicate that Nb as an alternative donor impurity shows an increase in conductivity up to at least 2.1% Nb-doping. Given the similarity between the two ions (same ionic radii, same d1 state in four-valent configuration, both are known for Jahn-Teller activity), we currently try to identify the underlying reason for the markedly different behaviour using the methodology described above. This work was supported by CCEM Switzerland and Umicore AG & Co KG within the project DuraCat. The authors acknowledge the use of the ARCHER UK National Supercomputing Service. [1] Tang et al., J. Power Sources 158 (2006) 1306-1312[2] A. Rabis, P. Rodriguez, T.J. Schmidt, ACS Catalysis 2 (2012) 864-890[3] K. Sasaki et al., ECS Transactions 33 (1) (2010) 473-482[4] G. Gehring and K. Gehring, Reports on Progress in Physics 38 (1) (1975) 1-89[5] A. Rabis, D. Kramer, E. Fabbri, M. Worsdale, R. Kötz, and T.J. Schmidt, The Journal of Physical Chemistry C 118 (2014) 11292-11302