Prediction of the absorption coefficient in multi-quantum dot structures using deep neural networks
Materials Today Communications, cilt.57, 2026 (SCI-Expanded, Scopus)
- Yayın Türü: Makale / Tam Makale
- Cilt numarası: 57
- Basım Tarihi: 2026
- Doi Numarası: 10.1016/j.mtcomm.2026.116313
- Dergi Adı: Materials Today Communications
- Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Chemical Abstracts Core, Compendex, INSPEC
- Anahtar Kelimeler: Deep neural network, Finite element method, GaAs/AlGaAs, Machine learning, Nanostructures, Total absorption coefficient
- Sivas Cumhuriyet Üniversitesi Adresli: Evet
Özet
In this work, we propose an efficient physics-generated data-driven deep learning framework to accurately predict the total absorption coefficient (TAC) in GaAs/Al0.3Ga0.7As multi-quantum dot (MQD) structures under external electric fields. The quantum-mechanical behaviour of confined electrons was first modelled by solving the effective-mass Schrödinger equation using the finite element method (FEM), generating a comprehensive dataset that captures the dependence of energy levels and dipole transition matrix elements on the dot width, barrier thickness, and applied electric field. These data were subsequently employed to train a deep residual neural network incorporating two residual blocks and multiple dense layers with batch normalisation and dropout regularization. The proposed model exhibits outstanding predictive performance, achieving a coefficient of determination R2≈0.999 and a mean absolute error below 0.02 across all test configurations. The results reveal a pronounced quantum-confined Stark effect, characterized by a redshift and amplitude suppression of the absorption peak with increasing field strength, while variations in dot and barrier dimensions enable precise modulation of optical response characteristics. This hybrid FEM-DNN approach offers a fast and accurate surrogate model for predicting the optical response of nanoscale optoelectronic and photonic devices with respect to individual structural and field parameters based on semiconductor quantum dots.