Machine-learning-enhanced understanding of haze formation: Rapid sulfate formation driven by abnormal aerosol interfacial effects

作者：Yangyang Liu, Qiuyue Ge, Wenbo You, Wei Wang, Lifang Xie, Kejian Li, Le Yang, Runbo Wang, Jilun Wang, Licheng Wang, Minglu Ma, Tingting Huang, Kedong Gong, Tao Wang, Liwu Zhang · 发表于：Cell Reports Sustainability · 年份：2025 · DOI：10.1016/j.crsus.2025.100581 · 被引用次数：2 · 研究领域：Atmospheric chemistry and aerosols、Air Quality and Health Impacts、Atmospheric aerosols and clouds

The rapid formation of sulfate aerosols during haze events represents a pressing challenge in air quality management, yet the underlying mechanisms remain poorly resolved in current models. This study addresses this knowledge gap by combining state-of-the-art spatially resolved spectroscopy and machine-learning strategies to unravel the interfacial process governing rapid sulfate production in nitrate aerosols. We demonstrate that interface-strengthened nitrate photolysis triggers fast sulfate formation, contributing up to 54% among pathways during heavily polluted haze episodes (high ionic strength). These processes are further amplified by uptaking sulfur dioxide, creating a positive feedback loop that further accelerates secondary sulfate and PM 2.5 formation. Our trained model suggests a ∼41% reduction of PM 2.5 by controlling nitrate concentration level during haze-relevant episodes. These findings reshape our understanding of haze occurrence and highlight the need for regulatory strategies targeting nitrate reduction and interfacial processes, offering new directions for refining air quality models and mitigating air pollution.