Machine learning helps design low-cost biochar to fight phosphorus pollution in lakes

(Press-News.org) Excess phosphorus in lakes and reservoirs fuels harmful algal blooms, threatens drinking water safety, and damages aquatic ecosystems worldwide. Now, researchers have developed a new machine learning–guided strategy to design advanced biochar materials that remove phosphorus efficiently while dramatically lowering treatment costs. The study provides a practical pathway for restoring eutrophic waters at large scale.

Phosphorus concentrations above very small thresholds can trigger ecosystem disruption and toxic algal blooms, making removal of trace amounts of phosphate a major challenge in water treatment. Adsorption using modified biochar is considered one of the most promising approaches for achieving ultra-low phosphorus levels, but high material costs have limited widespread use.

To address this challenge, the research team combined environmental materials science with artificial intelligence. They analyzed data from published studies and trained eight machine learning models to predict how different preparation conditions and metal compositions affect phosphate removal performance. The results showed that tree-based ensemble learning models achieved exceptionally high predictive accuracy, allowing researchers to identify optimal material formulations that balance efficiency and cost.

“Our goal was not only to improve adsorption performance, but to make these materials economically viable for real-world water restoration,” said one of the study’s corresponding authors. “Machine learning allowed us to rapidly explore thousands of possible designs and pinpoint combinations that conventional experiments would struggle to find.”

Using the optimized models, the team designed composite biochars containing lanthanum combined with calcium or iron. Experimental validation confirmed that these materials could reduce phosphate concentrations in water to extremely low levels, closely matching model predictions. Importantly, the optimized materials achieved major cost reductions compared with traditional lanthanum-modified biochar, cutting treatment costs by more than half in some cases while maintaining strong removal performance.

The researchers then simulated how the materials would perform in real lakes with different phosphorus concentrations and water chemistries. The analysis suggested that tailored material choices could significantly reduce remediation costs across diverse environments, from heavily polluted lakes to those with relatively low nutrient levels. By matching specific materials to local conditions, managers could achieve both environmental effectiveness and financial feasibility.

Beyond performance improvements, the study highlights the growing role of data-driven design in environmental engineering. Machine learning made it possible to identify key factors controlling phosphorus adsorption, including solution chemistry and metal loading levels, revealing complex relationships that traditional trial-and-error experimentation might miss.

“This work shows how artificial intelligence can accelerate the development of environmental technologies,” another author noted. “Instead of optimizing materials through long experimental cycles, we can now integrate data, prediction, and validation to design solutions much faster.”

The authors emphasize that practical deployment will require careful consideration of environmental safety, including monitoring potential metal release and developing strategies to recycle phosphorus-loaded biochar as fertilizer or soil amendments. If implemented responsibly, the approach could help close the phosphorus cycle while reducing nutrient pollution.

Overall, the study demonstrates that machine learning–guided material design can overcome one of the main barriers to advanced water treatment technologies: cost. By simultaneously optimizing performance and economics, the method opens new possibilities for restoring nutrient-impaired lakes and protecting freshwater resources worldwide.

 

=== 

Journal Reference: Fu, W., Yao, X., Zhang, X. et al. Machine learning–aided design of La-based composite modified biochar: Efficient materials and cost optimization for low-phosphorus water treatment. Biochar 8, 19 (2026).   

https://doi.org/10.1007/s42773-025-00534-3   

=== 

About Biochar

Biochar (e-ISSN: 2524-7867) is the first journal dedicated exclusively to biochar research, spanning agronomy, environmental science, and materials science. It publishes original studies on biochar production, processing, and applications—such as bioenergy, environmental remediation, soil enhancement, climate mitigation, water treatment, and sustainability analysis. The journal serves as an innovative and professional platform for global researchers to share advances in this rapidly expanding field. 

Follow us on Facebook, X, and Bluesky.  

END