Scholarlink

<p>Antimony (Sb) contamination in aquatic environments poses serious risks to human health and ecosystem integrity. Conventional remediation techniques for Sb pollution, however, are constrained by inherent limitations, including prolonged treatment durations, intricate optimization of experimental parameters (e.g., pH, temperature, adsorbent dosage), and the potential generation of secondary pollutants, thereby hindering their practical application in real-world scenarios. The emergence of machine learning (ML) technologies has opened innovative avenues to overcome these limitations. In this study, four interpretable ML models were employed to optimize Sb adsorption processes, with metal-organic frameworks (MOFs) selected as representative adsorbents. Model performance was significantly enhanced through Shapley Additive Explanations (SHAP) value analysis and Partial Dependence Plots (PDP), which provided mechanistic insights into feature importance and parameter interactions. Notably, the Support Vector Machine (SVM) algorithm identified ten optimal adsorption conditions, among which MOF-808 consistently demonstrated superior performance for heavy metal adsorption. This robust model further validated MOF-808 as a high-efficacy adsorbent, suggesting its potential for scalable environmental remediation applications. Through a limited number of experimental trials, the synthesis of the MOF-808 material and a series of Sb adsorption experiments were successfully accomplished. The findings demonstrate that MOF-808 exhibits robust adsorption performance across a broad pH range (2–11), achieving a maximum Sb(III) adsorption capacity of 120.60&#xa0;mg&#xa0;g<sup>−1</sup>. Notably, 96% of this maximum capacity is attained within 1&#xa0;h, indicating rapid adsorption kinetics. Mechanistic investigations via fourier-transform infrared spectrometer (FTIR) and X-ray photoelectron spectroscopy (XPS) analyses further confirmed that Sb(III) adsorption is predominantly governed by coordination interactions with unsaturated metal sites and hydrogen-bonding interactions with functional groups. This study not only provides a methodological framework for optimizing existing Sb wastewater treatment systems but also establishes a design principle for developing next-generation adsorbents, thereby advancing the development of efficient and sustainable strategies for heavy metal pollution mitigation.</p> Graphical abstract <p></p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Interpretable machine learning assisted optimization of antimony adsorption by metal–organic frameworks: combined with characterization methods for adsorption mechanism analysis

  • 期刊论文

  • 发表日期：

    2026年3月6日

查看全文
  • Chaojie Wang,
  • Yuxin Zhao,
  • Hanbo Chen,
  • Yurong Gao,
  • Qing Han,
  • Yueyue Leng,
  • Zheng Fang

摘要

Antimony (Sb) contamination in aquatic environments poses serious risks to human health and ecosystem integrity. Conventional remediation techniques for Sb pollution, however, are constrained by inherent limitations, including prolonged treatment durations, intricate optimization of experimental parameters (e.g., pH, temperature, adsorbent dosage), and the potential generation of secondary pollutants, thereby hindering their practical application in real-world scenarios. The emergence of machine learning (ML) technologies has opened innovative avenues to overcome these limitations. In this study, four interpretable ML models were employed to optimize Sb adsorption processes, with metal-organic frameworks (MOFs) selected as representative adsorbents. Model performance was significantly enhanced through Shapley Additive Explanations (SHAP) value analysis and Partial Dependence Plots (PDP), which provided mechanistic insights into feature importance and parameter interactions. Notably, the Support Vector Machine (SVM) algorithm identified ten optimal adsorption conditions, among which MOF-808 consistently demonstrated superior performance for heavy metal adsorption. This robust model further validated MOF-808 as a high-efficacy adsorbent, suggesting its potential for scalable environmental remediation applications. Through a limited number of experimental trials, the synthesis of the MOF-808 material and a series of Sb adsorption experiments were successfully accomplished. The findings demonstrate that MOF-808 exhibits robust adsorption performance across a broad pH range (2–11), achieving a maximum Sb(III) adsorption capacity of 120.60 mg g−1. Notably, 96% of this maximum capacity is attained within 1 h, indicating rapid adsorption kinetics. Mechanistic investigations via fourier-transform infrared spectrometer (FTIR) and X-ray photoelectron spectroscopy (XPS) analyses further confirmed that Sb(III) adsorption is predominantly governed by coordination interactions with unsaturated metal sites and hydrogen-bonding interactions with functional groups. This study not only provides a methodological framework for optimizing existing Sb wastewater treatment systems but also establishes a design principle for developing next-generation adsorbents, thereby advancing the development of efficient and sustainable strategies for heavy metal pollution mitigation.

Graphical abstract