We have designed, purchased, and commissioned a unique glovebox-integrated suite of thin-film growth setups that combine high experimental throughput and chemical breadth, including access to sulfur, phosphorus, and most metals in the periodic table. This setup has enabled us to grow some of the first phosphosulfide thin films ever reported. We have conducted high-throughput combinatorial materials discovery campaigns in three ternary phosphosulfide systems: Cu-P-S, Sb-P-S, and Ba-P-S. On the modeling/computation side, we have conducted high-throughput screening of about 1000 unique ternary phosphosulfides by density functional theory.
We have been able to synthesize a number of single-phase thin-film phosphosulfides, among which Cu3PS4. This compound is a 2.5 eV semiconductor. We measured rather high absorption coefficient, carrier mobilities, and carrier lifetimes in our synthesized samples. Thus, Cu3PS4 could find applications in LEDs and photoelectrochemical cells. Simultaneously, we have studied the properties of Cu3PS4 with high-level first-principles computational methods to understand the origin of the favorable properties found by experiment.
To make our high-throughput experimental and computational data findable, accessible, interoperable, and reusable, we have developed a cloud-based FAIR data infrastructure that will be the foundation of the inverse design process proposed in this project. The infrastructure is based on a local customized version of the public NOMAD database. The data is stored in the local database until it is validated and published in scientific journals, at which stage it can easily be pushed to the public database. The hybrid experimental/computational database will be an essential tool to apply artificial intelligence techniques to establish relationships within the data collected in this project and aid the inverse design process.