Daniel Lambert1 Alison Lennon1 Patrick Burr1

1, University of New South Wales, Kensington, New South Wales, Australia

In the search for low-cost, high efficiency solar cells, researchers have been investigating new materials to use as carrier-selective contact layers. Sub-stoichiometric molybdenum trioxide (MoO3-x ) is a promising material for interface contacts due to its hole selectivity and low parasitic absorption. It has been investigated as a contact for a number of different photovoltaic devices using different absorber materials, including ZnO/PbS quantum dot, CdS-CdTe nanotubes, CuInSe, and kesterite, as well as organic materials, perovskites, and crystalline silicon.

The performance of these contacts depends on the electrical and chemical properties of MoO3-x. Of particular interest are the bandgap, electron affinity, and the presence of a defect band, the latter imparting n-type semiconducting properties to the metal oxide. These material properties have been found to be sensitive to the presence of intrinsic defects, especially oxygen vacancies. An understanding of how the defect chemistry of MoO3-x varies with preparation conditions could allow the material properties of MoO3-x to be optimised for interfaces with different absorber materials.

This paper reports on the use of density functional theory (DFT) simulations to predict defect concentrations as a function of temperature and oxygen partial pressure for crystalline MoO3-x, by constructing Brouwer diagrams. Additionally it is shown that samples prepared in contact with silicon may be prone to contamination under common preparation conditions, and this contamination can affect the electronic structure of the material. It is therefore reasonable to assume that contamination of MoO3-x may also occur for other absorber materials and the implications of this contamination in terms of device performance and durability needs to be considered. We then extend this analysis into the properties of amorphous MoO3-x by means of reverse Monte Carlo modelling based on experimental thin film diffraction data. This extension is of particular value because most commonly-used preparation methods typically result in an amorphous metal oxide. This study provides a critical theoretical contribution to our understanding of the role of defects in transition metal oxide functionality as a carrier-selective contact for photovoltaic devices.