Spin-Cooperated Catalytic Activities in MnN<sub>4</sub>-Based Single-Atom Nanozyme: Mechanisms and a Brief Charge-Spin Machine Learning Model
Ling Liu, Shaofang Zhang, Xinzhu Chen, Guo Li, Yadong Zhe, Yonghui Li, Xiaodong Zhang
Abstract
The applications of natural enzymes are severely hindered by their limited sources and great tendency to denature. Therefore, developing artificial enzymes with comparable or even superior catalytic performance to that of natural enzymes and understanding multienzyme-mimicking catalytic mechanisms in depth are extremely desirable. Herein, single-atom nanozyme (SAzyme) MnN 4 with multiple enzyme-like activities is reported, and density functional theory (DFT) calculations combined with machine learning (ML) are employed to rationalize the origin of the performance of multienzyme-mimicking activities. Compared to its peroxidase- and catalase-like activities, the MnN 4 SAzyme exhibits outstanding superoxide dismutase (SOD)-like performance in “one-side adsorption” catalytic mechanisms due to a very low barrier of 0.077 eV in the desorption of the O 2 molecule. Such SOD-like activity can be linked to the “spin flip lock”, which appears in the detachment of the O 2 moiety: the O 2 moiety is locked to the MnN 4 SAzyme until the charge is transferred from the superoxide anion to the catalyst. Such a spin-related mechanism suggests a broad correspondence between electronic distribution and spin. With DFT-simulated charge-spin characteristics, a ML model is built based on the natural population analysis data set using the principal component analysis (PCA) and support vector machine algorithms. It displays a w-shaped boundary that separates the high spin from the low spin with respect to PCA-converted charge features. This work not only provides essential guidance for future synthesis of high-performance artificial SAzymes but also proposes rational mechanisms that are responsible for the multienzyme-mimicking catalytic activities of the MnN 4 SAzyme.