The statistical neural network-based regression approach for prediction of optical band gap of CuO

The design of CuO nanostructured semiconductors and the engineering of its optical band gaps have become important ways to further improve the performance of functional nano-devices such as energy conversion applications and optoelectronics. The optical band gap (Eg) of a semiconductor is a crucial parameter that defines its performance in solar cell applications. Therefore, determining the Eg of the semiconductor and comprehensively examining the relationship between the Eg and the structure of the material will be a guide to improve the performance of optoelectronic device applications. However, the relationship between Eg and structural properties of CuO is complex, and the combinations of variation in the Eg of CuO with various production parameters and doping materials are tremendous. For this reason, employing machine learning techniques can be a cheap, easy and effective approach to predict the Eg value of CuO. In this study, a statistical neural network (SNN)-based regression model is proposed to predict the energy band gap for the CuO semiconductor. A total of 100 CuO materials with optical band gaps between 1.02 and 3.41 eV are examined. With the proposed SNN approach, the optical band gap of CuO is predicted with low mean errors in terms of RMSE and MAE thanks to employing the structural parameters of the semiconductor material. (author)