Development of Semiconductor Electrical Conductivity Prediction System through Integration of Machine Learning and Solid State Physics Theory for Electronic Material Design

Machine learning has rapidly emerged as an effective system in various fields, including materials science, due to its ability to efficiently uncover complex patterns from big data. This research aims to develop a machine learning-based semiconductor electrical conductivity prediction system integrated with solid state physics theory. In the context of electronic material technology development, the ability to predict the electrical properties of materials before the synthesis process is crucial for improving efficiency and accuracy in material design. This study employs a qualitative descriptive method to create a machine learning-based system and qualitatively describe the research results. The dataset is obtained from various literature and previous experiments, with physical parameters such as crystal structure, band gap, and density. The machine learning model used is the random forest algorithm. Evaluation results show that the system can classify materials based on their electrical conductivity properties with high accuracy. Materials such as GaAs, Ge, and Si are identified as potential candidates for electronic device design. The integration of solid-state physics theory into this system enhances the physical and statistical validity of predictions. This research demonstrates that a machine learning approach based on physical principles can be an effective solution for future semiconductor material engineering.