Environmental Science & Technology, Год журнала: 2025, Номер unknown

Опубликована: Май 28, 2025

The development of new adsorbents for water treatment often involves complex adsorption mechanisms, whose individual contributions are unclear, thereby limiting the understanding driving forces, making it difficult to achieve precise design adsorbents. Machine learning (ML) has been used uncover impacts these mechanisms through feature engineering, but progress is limited by data quality training. Herein, we developed a universal ML strategy precisely predicting capacity polymers oxyanions and identifying force based on combination experimental published literature data. mechanism was explored classification RDkit descriptors with different SHAP importance values, electrostatic interaction found be in oxyanion process, which further verified theoretical calculations, experiments, effective targeted adsorbent design. In comparison, analysis relying separate source led decreased model performance, some biased conclusions, invalid Overall, this study proposed set optimization as well dominant identification, could shed light better wastewater.

Язык: Английский