A new study is bringing artificial intelligence into the fight against one of the most persistent visual signatures of industrial pollution: the deep orange color of synthetic textile dyes. Researchers Golaki, Azhdarpoor, Samaei and colleagues have investigated a visible-light-driven treatment system designed to break down Acid Orange 25, an azo dye widely used as a model pollutant in water-treatment research. Their approach combines a silver- and nitrogen-modified titanium dioxide photocatalyst with persulfate, while mathematical modeling and AI-based optimization are used to identify the operating conditions that can make the process more effective.
The work, published in Scientific Reports, addresses a problem that is both highly visible and chemically complex. Dyes released from textile, leather, paper and other manufacturing processes can remain in water even after conventional treatment. Their intense colors reduce light penetration, disrupting aquatic photosynthesis, while some dye molecules and their transformation products may display toxicity or resistance to biological degradation. Removing the color alone is not always enough; an effective process must ideally destroy the original molecules and limit the formation of harmful intermediates.
Acid Orange 25 belongs to the azo-dye family, whose characteristic color is produced by one or more nitrogen-nitrogen double bonds linking aromatic chemical groups. These structures are stable because their conjugated electron systems absorb visible light efficiently and resist ordinary chemical breakdown. That stability makes the dye useful in manufacturing, but it also makes contaminated wastewater difficult to treat. The study therefore focuses on advanced oxidation, a family of technologies capable of generating highly reactive chemical species that attack complex organic molecules rather than simply transferring them from water into sludge.
At the center of the proposed system is titanium dioxide, or TiO₂, a semiconductor long studied for photocatalytic water treatment. When TiO₂ absorbs photons with sufficient energy, electrons are promoted from its valence band to its conduction band, leaving behind positively charged holes. These electron–hole pairs can participate in surface reactions, producing hydroxyl radicals and other oxidizing species that break chemical bonds in pollutants. A major limitation, however, is that conventional TiO₂ responds most efficiently to ultraviolet radiation, which represents only a small fraction of sunlight. The researchers address this limitation by modifying the material with nitrogen and silver.
Nitrogen incorporation can alter the electronic structure of TiO₂ and extend its response toward the visible portion of the spectrum. Silver can contribute in several ways: it may help capture photogenerated electrons, reduce the rapid recombination of electrons and holes, and, depending on its chemical state and distribution, enhance light absorption through plasmonic effects. In practical terms, these modifications are intended to help the catalyst use more of the light that reaches the Earth’s surface. By combining the two modifications, the Ag/N/TiO₂ material is designed to create a more active platform for initiating pollutant-degradation reactions under visible illumination.
The second major component is persulfate, an oxidant that can be activated to produce sulfate radicals. These radicals are powerful, short-lived oxidizing agents capable of attacking aromatic rings, azo bonds and other electron-rich sites within dye molecules. Persulfate activation may occur through interactions with catalyst surfaces, photogenerated electrons, or other reactive pathways created during irradiation. Once formed, sulfate radicals can also participate in reaction networks that generate hydroxyl radicals and additional oxidizing species. The resulting chemistry gives the system several routes for dismantling Acid Orange 25 instead of relying on a single degradation mechanism.
What makes the study particularly timely is its use of modeling and artificial intelligence to optimize the treatment rather than testing operating conditions one by one. Advanced oxidation systems are governed by many interacting variables. The acidity of the water can change catalyst surface charge and radical stability; the amount of photocatalyst affects the number of available reactive sites; persulfate concentration can determine whether there is enough oxidant to sustain degradation or whether excess oxidant begins consuming radicals; and the starting dye concentration controls how much pollutant competes for the same reactive species. Light intensity and treatment time add further layers of complexity.
In a conventional experimental program, finding the best combination of these variables could require hundreds of individual tests. A data-driven model can instead learn relationships between experimental inputs and treatment performance, identify influential parameters and predict promising conditions for further verification. The AI component described by the researchers is therefore not a replacement for chemistry but a tool for navigating it. It can reveal nonlinear interactions that are easy to miss in simple experiments, such as situations in which increasing one reagent improves removal only within a narrow range of pH, catalyst loading or irradiation time.
The study’s importance extends beyond the disappearance of an orange dye from laboratory water. A treatment process that works efficiently under visible light could reduce dependence on ultraviolet lamps and potentially make better use of solar radiation. At the same time, the combination of photocatalysis and persulfate raises practical questions that will determine whether the technology can move toward real wastewater applications. Researchers must establish how catalyst particles are recovered, whether silver can leach into treated water, how natural organic matter and dissolved salts affect radical chemistry, and whether the dye is fully mineralized into simpler end products rather than converted into less visible but still problematic compounds.
The Ag/N/TiO₂/persulfate platform also illustrates a broader transformation taking place in environmental engineering. Instead of treating materials development, reaction chemistry and process optimization as separate tasks, researchers are increasingly connecting them through computational tools. AI can help determine which experiments are most informative, reduce unnecessary reagent use and accelerate the search for conditions that balance efficiency, cost and safety. For dye-contaminated water, that integrated strategy could be especially valuable because real effluents contain mixtures of dyes, salts, surfactants and other organic compounds that behave differently from a single laboratory pollutant.
Although Acid Orange 25 serves as a defined target for evaluating the system, the underlying concept may be relevant to a wider group of persistent organic contaminants. The combination of a visible-light-responsive semiconductor, a catalyst modifier that improves charge behavior and an oxidant capable of generating sulfate radicals offers a flexible foundation for advanced water treatment. The next challenge will be demonstrating consistent performance in complex wastewater, confirming the identity and toxicity of intermediate products, and showing that the process remains economically and environmentally responsible at larger scale. By pairing photocatalytic chemistry with AI-guided decision-making, the study points toward a future in which cleaner water may depend as much on intelligent optimization as on the reactive materials themselves.
Subject of Research: Visible-light-driven degradation of Acid Orange 25 in water using an Ag/N/TiO₂/persulfate advanced oxidation system, optimized through modeling and artificial intelligence.
Article Title: Modeling and AI optimization of visible-light-driven acid orange 25 degradation using an Ag/N/TiO₂/persulfate system.
Article References: Golaki, M., Azhdarpoor, A., Samaei, M.R. et al. “Modeling and AI optimization of visible-light-driven acid orange 25 degradation using an Ag/N/TiO₂/persulfate system.” Scientific Reports (2026). https://doi.org/10.1038/s41598-026-65770-4
Image Credits: AI Generated
DOI: 10.1038/s41598-026-65770-4
Keywords: Acid Orange 25, visible-light photocatalysis, artificial intelligence, TiO₂, silver and nitrogen modification, persulfate activation, advanced oxidation, wastewater treatment, azo dyes, environmental engineering
