Abstract - MXene, a family of 2D materials, was recently shown to offer electrochemical qualities that have excellent applicability in catalysis and electronics. MXene offers high tunability by varying different properties such as constituent elements and lattice structure. Understanding those properties is a crucial step in tuning a more efficient and versatile catalyst. MXenes’ electric and physical properties, such as workfunction, orbital radii, and electronegativity, are collected from Computational 2D Materials Database (C2DB). Machine learning is used as an alternative data analysis method to extract unseen insights and construct an accurate result model. Fundamental statistical analyses are conducted, followed by diverse machine learning tools such as sure independence screening and sparsifying operator (SISSO), AUTOFEAT feature engineering & selection, and different ensemble methods like stacking and boosting. The high accuracy of models indicates the applicability of these models in future predictions. These models are significantly less time-consuming and expensive to produce than widely used analysis methods, such as quantum mechanical modelling methods like density functional theory (DFT) or direct experimentation. Further research, such as using data of MXene across different environments, should be explored for further industrial application.
Sadly when I was working on this project I wasn't the greatest at file organization ;-; can't find all the vital and final calculations and files, but I hope whatever I share could give you a good overview. Any feedbacks welcomed, and questions welcomed too.