Metal oxide nanoparticles are promising materials for applications in a wide range of catalytic reactions. Among them, niobium oxide V nanoparticles (Nb 2 O 5 -NPs) present unique acid sites, stability, and electronic properties suitable for application in photocatalysis. However, despite these advantages, the number of studies using Nb 2 O 5 -NPs in photocatalysis remains minor compared to other nanomaterials. In this sense, the application of computational studies using Machine Learning (ML) can help the development of the field, by using algorithms to predict nanoparticle properties and identify cluster regions for further investigations. In this context, the present study aims to evaluate the application of ML study to predict the values of specific surface area and band gap energy of Nb 2 O 5 -NPs. A survey of data available in the literature was carried out for the exploratory analysis of the main synthesis variables and the prediction of specific surface area and bang gap energy values through the evaluation of three tree-based ML algorithms: Random Forest (RF), Extreme Gradient Boosting (XGB), and Gradient Boosting Regressor (GBR). The results of the exploratory analysis showed the presence of cluster regions due to the type of routes and possibilities for further studies to develop new Nb 2 O 5 -NPs for photocatalysis. The ML study showed that the GBR model presents the best performance to predict the values of specific surface area and band gap energy of Nb 2 O 5 -NPs based on the selected synthesis conditions. Therefore, the use of ML is presented as a tool to be used to evaluate Nb 2 O 5 -NPs properties before carrying out experimental steps.
Silva et al. (Sun,) studied this question.