AI models forecast 'green' carbon's power to cleanse water of selenium
Machine learning accurately predicts how iron-modified biochar removes the toxic element selenium from water, identifying the key factors for improving environmental remediation efforts
image: Machine learning approach to predict adsorption capacity of Fe-modified biochar for selenium
Credit: Habib Ullah, Sangar Khan & Baoliang Chen
The Selenium Dilemma
Selenium is an element with a dual nature; it is a necessary micronutrient for humans and animals but becomes toxic at high concentrations. Its accumulation in water sources, resulting from both geological processes and human activities, presents a serious environmental and public health issue. Removing excess selenium from water and wastewater is an important goal for sustainable development.
A Sustainable Adsorbent
A promising method for water purification involves biosorption, which uses natural, low-cost materials. One such material is biochar, a stable, carbon-rich substance produced from biomass waste like agricultural byproducts. While standard biochar has many benefits, its performance can be greatly improved. Researchers have found that modifying biochar with iron creates a composite material with a much higher capacity to adsorb, or capture, contaminants like selenium.
Predicting Performance with Machine Learning
To accelerate the development of better adsorbents without extensive trial-and-error experiments, a team of researchers led by Baoliang Chen at Zhejiang University turned to artificial intelligence. They collected data from numerous published studies on the ability of iron-modified biochar to remove selenium. Using this dataset, scientists Habib Ullah and Sangar Khan developed predictive models using three machine learning algorithms: Random Forest RF, Support Vector Regression SVR, and SHAP.
Highly Accurate Forecasting
The machine learning models demonstrated strong performance in forecasting the selenium removal capacity of the biochar. The SVR model was particularly effective, showing a very high accuracy with a low error rate. The results confirm that these computational tools can reliably estimate how well a specific biochar formulation will perform under various conditions, offering a valuable guide for designing new materials.
Identifying the Most Influential Factors
Beyond simple prediction, the models also identified which variables had the most significant impact on selenium adsorption. Across all three models, the most influential factor was the biochar's oxygen content, followed by its carbon content, the water temperature, pH level, and the amount of iron impregnation. This knowledge gives scientists a clear target for optimization.
Guiding Future Water Treatment
The findings provide a comprehensive guideline for developing more effective biochar adsorbents. By focusing on manipulating the most important properties, such as surface oxygen content, researchers can engineer materials specifically tailored for maximum selenium removal. This data-driven approach can improve the treatment of real-world water and wastewater, making the process more efficient and effective. The work by researchers from Zhejiang University, Ningbo University, and other institutions shows the power of combining material science with machine learning to address pressing environmental pollution challenges.
Corresponding Author: Baoliang Chen
Original Source: https://doi.org/10.1007/s44246-023-00061-5
Contributions:
Conceptualization and original draft writing, HU; Modelling, SK; Funding acquisition, BC; Writing—review and editing, LR, AS and NW; All authors have read and agreed to the published version of the manuscript.