Machine learning-driven prediction of Cr(VI) adsorption capacity by MOFs and ranking via the innovative QCTE index

Heavy metal pollution, particularly hexavalent chromium, poses critical environmental challenges and demands efficient removal methods. This study introduces a machine learning framework to predict chromium adsorption capacity by metal-organic frameworks (MOFs), accelerating adsorbent discovery. Integrating data from Jiang et al. and ۳۴۷ literature sources, the supervised regression model utilized key features such as specific surface area, initial concentration, adsorbent dosage, contact time, temperature, pH, and numerical atomic properties of the central metal. Among the evaluated algorithms, which include Linear Regression, Ridge, Elastic Net, Random Forest, ExtraTrees, and Gradient Boosting, Extra Trees delivered the best performance with RMSE of ۱۳.۶۰. To rank novel combinations practically, an innovative QCTE index incorporating adsorption capacity, thermal stability, cost and availability, and environmental impact was used, and UiO-۶۶ was identified as the top candidate.

کلیدواژه ها:

Chromium adsorption ، Metal-organic frameworks ، Machine learning ، Supervised regression