AI accelerates material discovery for advanced perovskiet Sunnight Technology
A collaboration team at Beijing University and the Shenzhen Graduate School has developed Machine Learning models that can quickly and accurately predict critical properties of Halide Perovskites – important materials in solar cells of the next generation. Their work is intended to streamline the search for optimum connections by concentrating on essential parameters such as guideline minimum (CBM), Valence Band Maximum (VBM) and BandGap energy.
Halide -perovskites, with their ABX3 crystal structure, are promising materials because of their impressive photovoltaic performance, manufacturing bag and low costs. These materials are very adjustable, allowing researchers to optimize electronic properties to improve the efficiency of the power conversion (PCE), which has now surpassed 27% in a single junction and more than 30% in tandem solar cells. However, persistent challenges – such as lead oxity and stability problems – require the discovery of improved compositions with ideal band structures.
Accurate knowledge of the CBM, VBM and Bandgap of a Perovskiet is of fundamental importance for optimizing device efficiency, because these properties determine light absorption and loading transport options. Traditional methods for analyzing these factors, such as screening with high transit and density Functional theory (DFT) simulations, are reliable but resource-heavy.
To tackle this, the researchers used Extreme Gradient Boosting (XGB) to build predictive models that are able to estimate band structure functions, both inorganic and hybrid halide perovskites. Their XGB model resulted in high accuracy and reached the test set R-values of 0.8298 for CBM, 0.8481 for VBM and 0.8008 for BandGap predictions using the Heyd-Scuseria-Orzerhof (HSE) functional. With the help of the Perdew-burke-Ternzerhof (PBE) functional for a wider data set, the model further improved with an R of 0.9316 and an average absolute error (Mae) of only 0.102 EV.
In addition, Shap analysis (Shapley Additive Designations) revealed which chemical and structural characteristics most electronic energy levels influence the most, which offer a route map for designing better performing perovskites. This approach not only speeds up the pace of the discovery, but also offers environmentally friendly and cost -effective alternatives to traditional methods.
Looking ahead, the researchers want to integrate the interpretability of shallow machine learning models with the depth of neural networks to further refine the discovery of materials. Their approach is promising for developing the next generation of solar technologies with improved efficiency, stability and environmental safety.
Research report:Machine Learning for prediction of energy band from Halogenide Perovskites
