Targeting key toxic nanoscale particulate matter for precision control of coal power emissions

Coal-fired power plants represent a major anthropogenic source of nanoscale particulate matter, yet conventional mass-based regulations overlook the distinct and potent health risks posed by specific components. Here we combine single-particle elemental profiles (169 plants across China) with cellular toxicity (human lung cells). Using interpretable machine learning, we reveal iron-rich nanoparticles as key toxic driver, explaining 27.4% of the observed oxidative stress and 16.9% of cytotoxicity. We then develop a high-resolution national inventory of iron-rich nanoparticles, estimating total emissions of 236 tons in 2020, with Eastern China as a hotspot contributing 38.2%. Tailored regional strategies could achieve a 77.5% reduction in national emission, with electrostatic precipitator upgrades identified as the most cost-effective measure. Our findings provide an actionable framework to advance air pollution policy beyond total emissions control toward component-specific reduction of the most toxic nanoparticles, ultimately mitigating their associated public health impacts. Iron rich nanoscale particles from coal plants are identified as the main toxic component, revealed through single particle chemical profiling and cell-based toxicity analysis combined with interpretable machine learning.