Machine-learning-assisted modeling of defect-mediated charge transport in anisotropic 2D semiconductors
This study investigates defect-mediated charge transport in anisotropic two-dimensional semiconductors using combined transport modeling and machine-learning analysis. The objective of the work was to determine how structural defects and lattice anisotropy influence electrical conductivity and carrier mobility in low-dimensional quantum materials. Monolayer transition-metal dichalcogenides, black phosphorus, and ReS2-based systems with various defect configurations were analyzed. Transport behavior was modeled within a semiclassical framework incorporating anisotropic carrier mobility and defect-assisted scattering mechanisms. Machine-learning models, including random forest regression, gradient boosting regression, and artificial neural networks, were applied to predict conductivity variations using structural and electronic descriptors. The results demonstrated that conductivity decreases nonlinearly with increasing defect concentration, particularly above 1013 cm−2, where localization effects become dominant. Strongly anisotropic materials exhibited enhanced sensitivity to vacancy-induced disorder due to directional carrier confinement. Among the evaluated algorithms, the artificial neural network achieved the highest predictive accuracy with an R2 value of 0.972. Feature-importance analysis revealed that defect concentration, effective mass, and lattice anisotropy ratio are the primary factors governing transport degradation. The study confirms that accurate prediction of transport properties in anisotropic quantum materials requires simultaneous consideration of defect physics and directional electronic transport. The proposed approach may be useful for the design and optimization of next-generation nanoelectronic and flexible semiconductor devices.
Сведения
- Журнал
- Technobius Physics
- Опубликовано
- 2026 · Т. 4, № 2, С. 0051
- Язык
- Английский
- ISSN
- 3007-0147
- Издатель
- Technobius