ChecMatE: A workflow package to automatically generate machine learning potentials and phase diagrams for...

Area of Science:

• Materials Science
• Computational Chemistry
• Solid-State Physics

Background:

• Semiconductor alloy materials offer tunable properties crucial for technological advancements.
• Exploring the vast structural space of these alloys is computationally intensive and limits property control.
• Traditional methods for material discovery are often inefficient and constrained by known chemical and crystallographic principles.

Purpose of the Study:

• To develop an automated framework, ChecMatE (Chemical Material Explorer), for efficient exploration of semiconductor alloy structural spaces.
• To leverage machine learning potentials (MLPs) for accurate prediction of material energy and stability.
• To accelerate the discovery and design of novel semiconductor alloy materials.

Main Methods:

• Development of ChecMatE software package for automated material exploration.
• Generation of machine learning potentials (MLPs) for predicting material properties.
• Integration of global search algorithms for comprehensive screening of alloy structures.
• Application of ab initio accuracy for energy and stability predictions.

Main Results:

• ChecMatE enables efficient and cost-effective exploration of material structural spaces.
• The software accurately predicts the energy and relative stability of semiconductor alloys.
• Demonstrated accelerated structural exploration in a case study of the InxGa1-xN system.
• Reduced computational costs associated with materials discovery.

Conclusions:

• ChecMatE provides a powerful and automated solution for exploring semiconductor alloy structural landscapes.
• The framework significantly enhances the efficiency and reduces the cost of discovering new materials.
• This approach promises to overcome limitations in traditional methods for materials design and property control.