Liang Chouyuan, Zhang Zhuo, Li Yuanyuan, Wang Yakun, He Mengsi, Xia Fang, Wu He
School of Land Science and Technology, China University of Geosciences (Beijing), Beijing 100083, China.
School of Land Science and Technology, China University of Geosciences (Beijing), Beijing 100083, China; Key Laboratory of Land Consolidation and Rehabilitation, Ministry of Natural Resources, Beijing 100035, China.
Environ Res. 2025 Jan 15;265:120471. doi: 10.1016/j.envres.2024.120471. Epub 2024 Nov 26.
Due to its sea urchin-like structure, Schwertmannite is commonly applied for heavy metals (HMs) pollutant adsorption. The adsorption influence parameters of Schwertmannite are numerous, the traditional experimental enumeration is powerless. In recent years, machine learning (ML) has been gradually employed for adsorbent materials, but there is no comprehensive research on the Schwertmannite adsorbent. In this paper, 27 features and 814 groups of experimental data were used to systematically analyze the adsorption modeling of Schwertmannite first time. The results indicated that the adsorption capacity of Schwertmannite was better predicted by the Random Forest (RF) model (the R was 0.874). Then, the RF model was used to analyze the features importance that affects the adsorption of HMs by Schwertmannite. And the importance of Schwertmannite synthesis conditions, Schwertmannite characteristics, adsorption environment, and HMs properties were 11.88%, 30.01%, 48.26%, and 8.19% respectively. Moreover, the synthesis and adsorption conditions of Schwertmannite were predicted and optimized based on RF model, it was predicted that the better synthesis method of Schwertmannite was biological oxidation > Fe oxidation > Fe hydrolysis. Finally, a predictive Graphical User Interface Web Page for Schwertmannite-HMs was developed. We hope that this paper can promote the integration of machine learning and Schwertmannite.
