Artificial intelligence boosts eco‑friendly dye cleanup

As industries continue to produce colorful textiles and plastics, the dyes that make them vibrant pose a serious threat to rivers and lakes. A new study published in Carbon Research shows how machine learning can help design better materials to capture and remove toxic dyes from water, improving both environmental safety and industrial sustainability.

The research team led by Chong Liu of Dongguan University of Technology developed artificial‑intelligence models that predict how effectively biochar, a carbon‑rich material made from waste biomass, can adsorb different dyes. By training and testing nine different machine‑learning algorithms with hundreds of laboratory datasets, the scientists identified the CatBoost model as the most accurate prediction tool, achieving a performance score of R2=0.988, meaning its predictions almost perfectly matched experimental data.

Biochar is a promising, low‑cost adsorbent for treating dye‑contaminated wastewater from textile, paper, and chemical plants. Yet, the performance of each biochar product varies depending on its source material and production method. Traditional experiments to test every combination are time‑consuming and expensive. The new machine‑learning approach eliminates much of this guesswork by revealing which factors matter most.

The analysis found that operating conditions, such as solution concentration and temperature, determined over half of the variation in adsorption performance, followed by biochar structure and chemical composition. According to the model, key properties that enhance dye removal include high carbon content and large surface area. The team also used an explainable‑AI technique called SHAP to visualize how each factor contributed to performance, helping researchers interpret the complex model results in simple physical terms.

To confirm its predictions, the group performed new laboratory experiments using cotton‑straw biochar to remove common dyes such as methylene blue and congo red. The tests matched the algorithm’s forecasts with impressive precision. Building on these results, Liu’s team designed an easy‑to‑use computer interface that allows scientists and engineers to input key parameters and instantly estimate dye adsorption efficiency. The software, created in Python and freely available on GitHub, is intended to make machine‑learning tools accessible to environmental engineers worldwide.

“By uniting advanced computing with green chemistry, we can accelerate the design of next‑generation water‑purification materials,” Liu said. “Machine learning provides a smarter, faster way to protect our aquatic ecosystems.” 

=== 

Journal Reference: Liu, C., Balasubramanian, P., Nguyen, X.C. et al. Enhanced machine learning prediction of biochar adsorption for dyes: Parameter optimization and experimental validation. Carbon Res. 4, 46 (2025). https://link.springer.com/article/10.1007/s44246-025-00213-9  

=== 

About Carbon Research

The journal Carbon Research is an international multidisciplinary platform for communicating advances in fundamental and applied research on natural and engineered carbonaceous materials that are associated with ecological and environmental functions, energy generation, and global change. It is a fully Open Access (OA) journal and the Article Publishing Charges (APC) are waived until Dec 31, 2025. It is dedicated to serving as an innovative, efficient and professional platform for researchers in the field of carbon functions around the world to deliver findings from this rapidly expanding field of science. The journal is currently indexed by Scopus and Ei Compendex, and as of June 2025, the dynamic CiteScore value is 15.4.

Follow us on Facebook, X, and Bluesky.