Machine Learning-Assisted Discovery of Bimetallic Oxides for Highly Efficient Catalytic Ozonation
Changxiao Zhang, Shasha Li, Hanyue Zhang, Jie Miao, Jiatong Zhang, Minghua Zhou
Abstract
Catalytic ozonation stands out as an effective process in the advanced treatment of industrial wastewater, where heterogeneous catalysts play a pivotal role. Here, by screening 1603 bimetallic oxides via machine learning (ML), a pioneering ZnCu 2 O 4 was dug out, validated by density-functional theory and experiments. Compared with the literature, ZnCu 2 O 4 significantly boosted the degradation rate constant for oxalic acid ( k obs = 0.30 min –1 ) by 1.30–61.22 times. Meanwhile, the average ozone treatment efficiency of chemical oxygen demand (COD) and total organic carbon (TOC) for high-salinity coal chemical wastewater (hsCCW), i.e., ΔCOD/ΔO 3 (1.01 kg kg –1 ) and ΔTOC/ΔO 3 (0.30 kg kg –1 ), reached 0.61–4.60-fold and 1.32–4.84-fold of the literature, respectively. Mechanistic studies revealed a unique nonradical pathway dominated by 1 O 2, ensuring resistance to environmental interference. Its particular Cu–O–Zn configuration enhanced stability and active-site exposure, which is critical for scalable applications. Overall, this research and development (R&D) framework encompassing multidimensional “theoretical calculation-machine learning-precision synthesis-mechanism elucidation” establishes a generalizable methodology for intelligent material innovation and environmental application.