Exploring Large-Lattice-Mismatched Interfaces with Neural Network Potentials: the Case of the CdS/CdTe Heterostructure

Mao Su,Ji-Hui Yang,Zhi-Pan Liu,Xin-Gao Gong
DOI: https://doi.org/10.1021/acs.jpcc.2c02424
2022-01-01
The Journal of Physical Chemistry C
Abstract:The heterostructure CdS/CdTe is one of the most important interfaces in solar energy cells, however, its exact structure is not well-known due to the large lattice mismatch, thus hindering the studies of interfacial properties. In this paper, we present a scheme to explore the interface structures by combining the differential evolution (DE) algorithm and the neural network potential. A high-accuracy neural network potential is trained to describe the interatomic interactions, and then the interfacial structures are searched using our DE program that started from a CdS/CdTe interface model that consists of about 1740 atoms. We identify a series of interfacial structures with defects and assess their stabilities by considering the interfacial energies. We find that the CdS-like interface I is generally energetically more favorable and prefers to be defect-free under the Cd-poor condition. Therefore, the Cd-poor condition might be beneficial for improving the solar cell efficiencies by suppressing possible defect recombination at the interface. Our work provides a general scheme to study structures of complex interfaces with a large number of atoms.
What problem does this paper attempt to address?