Abstract
The phase transition of cadmium selenide (CdSe) from wurtzite to rocksalt structure has been the subject of extensive research. In this study, we present a novel approach combining machine learning potentials with swarm intelligence-based pathway sampling to elucidate the complex phase transition mechanisms in CdSe. We developed an accurate machine-learning (ML) potential for CdSe, validated against density functional theory calculations, achieving mean absolute errors (MAEs) of 1.8 meV/atom for energies and 33 meV/Å for forces. This potential was integrated with the pathway sampling via swarm intelligence and graph theory (PALLAS) method to explore the potential energy landscape and identify low-energy transition pathways. Our simulations revealed a complex network of transition pathways, and we discovered a multi-step transition mechanism involving an unexpected zinc blende intermediate phase, which appears to play a crucial role in facilitating the transition between wurtzite and rocksalt structures. This finding provides new insights into the structural flexibility of CdSe and offers an explanation for experimentally observed phenomena such as wurtzite/zinc blende coexistence in nanostructures. Our approach not only advances the fundamental understanding of phase transitions in CdSe but also establishes a powerful computational framework for exploring complex materials phenomena, opening new avenues for materials design and discovery in semiconductor systems.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call