Zero Hunger (SDG 2), Good Health & Well-being (SDG 3)
Machine learning-driven prediction of photocatalytic degradation of Tetracycline using graphitic carbon nitride-based heterojunctions
Olalekan C. Olatunde * , Damian C. Onwudiwe Department of Chemistry, North-West University, Mafikeng Campus E-mail: 32587139@mynwu.ac.za
The persistence of antibiotics like tetracycline in water sources poses a significant environmental threat due to potential toxicity. Photocatalytic degradation, particularly using semiconductor heterostructures, has emerged as a promising method for treating such contaminants. In this study, we explore the machine learning-based prediction of photocatalytic degradation efficiencies for tetracycline using graphitic carbon nitride (g-C 3 N 4 ) heterojunctions. Ten different g-C 3 N 4 -based heterojunctions were evaluated, with key photocatalytic parameters such as band gap energy, surface area, and pollutant concentration serving as input variables. Various ensemble machine learning models, including Random Forest, Catboost, Xgboost, and Adaboost, were trained and assessed based on their performance metrics, including root mean squared error (RMSE) and coefficient of determination (R²). The Xgboost model demonstrated the highest accuracy in predicting the degradation efficiency, particularly with NiAl 2 O 4 /g-C 3 N 4 as a photocatalyst. Our findings highlight the significant role of catalyst weight, reaction time, and pollutant concentration in influencing photocatalytic degradation, offering a predictive
framework for optimizing future wastewater treatment processes. Key words: Antibiotics, Tetracycline, g-C3N4, Heterojunction, Machine learning
P28
© The Author(s), 2025
Made with FlippingBook Learn more on our blog