Akademik Veri Yönetim Sistemi
Materials Today Communications, cilt.52, 2026 (SCI-Expanded, Scopus)
We investigate the temperature-dependent nonlinear optical rectification (NOR) in GaAs/AlGaAs tetrapod core/shell quantum dots containing a central hydrogenic impurity. The electronic structure of these systems is obtained by solving the three-dimensional Schrödinger equation using the finite element method in the effective-mass approximation, with the NOR coefficient calculated using the compact density-matrix formalism. To predict the temperature evolution of the NOR, several machine learning (ML) regression algorithms were evaluated. We find that the temperature induces a redshift of the NOR resonance and an increase in its amplitude, driven by a reduction in the GaAs bandgap and a strengthening of the dipole matrix elements. A comparative analysis of ML models shows that the Random Forest Regression algorithm achieves the highest predictive accuracy across the evaluation metrics, while Gaussian Process Regression provides a robust probabilistic framework for capturing the nonlinear temperature dependence of the NOR coefficient. These results demonstrate that ML models can effectively reproduce the complex behavior of NOR spectra while maintaining strong agreement with theoretical calculations. Our study highlights the effectiveness of combining quantum-mechanical modeling with machine-learning techniques to accurately predict and optimize nonlinear optical processes in semiconductor quantum dots.