Machine learning for impurity charge-state transition levels in semiconductors from elemental properties using multi-fidelity datasets

Abstract

Quantifying charge-state transition energy levels of impurities in semiconductors is critical to understanding and engineering their optoelectronic properties for applications ranging from solar photovoltaics to infrared lasers. While these transition levels can be measured and calculated accurately, such efforts are time consuming and more rapid prediction methods would be beneficial. Here, we significantly reduce the time typically required to predict impurity transition levels using multi-fidelity datasets and a machine learning approach employing features based on elemental properties and impurity positions. We use transition levels obtained from low-fidelity density functional theory (DFT) calculations, corrected using a recently-proposed modified band alignment scheme which well-approximates transition levels from high-fidelity DFT. The model fit to the large multi-fidelity database shows improved accuracy compared to models trained on the more limited high-fidelity values. Crucially, in our approach when using the multi-fidelity data, high-fidelity values are not required for model training, significantly reducing the computational cost required for training the model. Our machine learning model of transition levels has a root mean squared (mean absolute) error of 0.36 (0.27) eV versus high-fidelity hybrid functional values, when averaged over 14 semiconductor systems from the II-VI and III-V families. As a guide for use on other systems, we assessed the model on simulated data to show the expected accuracy level as a function of band gap for new materials of interest. Finally, we use the model to predict a complete space of impurity charge-state transition levels in all zincblende III-V and II-VI systems.

Exploration of Deep Learning Models for Accelerated Defect Property Predictions and Device Design of Cubic Semiconductor Crystals

“…To assess the robustness of our models, we determined the standard deviation through a 5-fold cross-validation process. As depicted in Figure , the standard deviation hovers at approximately half the magnitude of the prediction error, a variance deemed acceptable by prevailing reports in the literature. , Although a larger regularization term might bridge the test-train performance gap, such a modification was deemed unnecessary. The regularization parameter has been finely tuned via Bayesian optimization, and further increases might introduce undue bias.…”

Accelerating defect predictions in semiconductors using graph neural networks

“…Inaccuracies will still persist from incorrectly locating VBM and conduction-band minimum (CBM), but appropriate corrections can be applied afterward using different higher-accuracy bulk calculations once PBE-level DFEs are predicted for multiple q and μ conditions. Two such possible correction methods include using the modified band alignment approach based on PBE and HSE06 bandgap values 51 and shifting both band edge positions using GW quasiparticle energies. 52 The focus of the present work is to demonstrate the accelerated prediction of PBE-level defect energetics, and the aforementioned corrections will be considered in the future work.…”

“…2. As an input for higher-fidelity calculations, such as using HSE06 or GW, 43,46 as well as suitable band edge corrections 51,52 that may be applied to shift the predicted DFE (EF = 0 eV) values for more accurate E f vs EF plots. 3.…”

Exploring Multi-Fidelity Data in Materials Science: Challenges, Applications, and Optimized Learning Strategies

“…Different Algorithms functional [14][15][16][17][18] citations (273,60,4,43,25) experiment [19] citations (18) functional and experiment [1,2] citations (1,71) physical models [20] citations (14) functional and physical models [9] citations (55) Different hyperparameters mesh size [21] citations (145) time step [22,23] citations (21,23) different hyper parameters [24] citations (35) Aerospace Science…”