The journal Biochar recently published a study by Weilin Fu, Xia Yao, Xueyan Zhang, Shiyu Lv, Tian Yuan, Yi An, and Feng Wang exploring how artificial intelligence can improve the way we clean our water. The researchers focused on the challenge of phosphorus pollution, a major cause of harmful algae blooms in lakes and rivers that can kill fish and ruin drinking water. While a rare earth metal called lanthanum is incredibly effective at trapping phosphorus, its high price makes it difficult for many communities to use on a large scale. By using machine learning, the team sought to find a way to mix lanthanum with cheaper metals to maintain high performance at a much lower price point.

The findings reveal that advanced computer models, specifically tree-based ensemble learning, can predict how well a filter will work with nearly 100 percent precision. This allowed the researchers to skip months of trial-and-error laboratory work. Instead, the computer identified the ideal ratio of lanthanum to calcium, a much more common and affordable element. This optimized “smart” material achieved a 99 percent removal rate of phosphorus, leaving behind a residual concentration of only 0.02 milligrams per liter. This level of purity is significant because it meets the strictest global environmental standards for sensitive water bodies.

Beyond the technical success of cleaning the water, the study provides a detailed economic breakthrough. The researchers found that by replacing a portion of the expensive lanthanum with calcium, they could reduce the cost of the raw materials by 31.5 percent. This cost reduction does not come at the expense of efficiency; the new bimetallic biochar actually performed better than many traditional materials. It works effectively across a wide range of water conditions, including different acidity levels and the presence of other salts, which often interfere with standard water treatment processes. The material also showed it could be reused multiple times, further increasing its value for long-term municipal use.

The study concludes that integrating machine learning into environmental engineering is a game-changer for sustainability. By moving away from expensive, single-metal treatments toward optimized, multi-metal composites, cities can afford to implement higher standards of water protection. The ability of the computer to analyze complex factors like metal loading and reaction times means that future water treatment plants can be custom-designed for the specific pollutants in their local area. This approach ensures that we can protect our natural water resources without placing an impossible financial burden on taxpayers or industry.

Source: Fu, W., Yao, X., Zhang, X., Lv, S., Yuan, T., An, Y., & Wang, F. (2026). Machine learning-aided design of La-based composite modified biochar: Efficient materials and cost optimization for low-phosphorus water treatment. Biochar, 8(19).