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) MnN4 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 MnN4 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 MnN4 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 MnN4 SAzyme.

Machine Learning Assisted Graphdiyne-Based Nanozyme Discovery

Design and performance analysis of multi-enzyme activity-doped nanozymes assisted by machine learning

Accelerated Design of Catalytic Water-Cleaning Nanomotors Via Machine Learning.

Machine-Learning-Assisted Nanozyme Design: Lessons from Materials and Engineered Enzymes

Machine‐Learning‐Assisted Nanozyme Design: Lessons from Materials and Engineered Enzymes

Machine learning facilitating the rational design of nanozymes.

Machine Learning in Nanozymes: From Design to Application

Prediction and Design of Nanozymes using Explainable Machine Learning

ChatGPT Combining Machine Learning for the Prediction of Nanozyme Catalytic Types and Activities

Functional Graphdiyne for Emerging Applications: Recent Advances and Future Challenges

Machine Learning-Guided Yield Optimization for Palladaelectro-Catalyzed Annulation Reaction

Spin-Cooperated Catalytic Activities in MnN4-Based Single-Atom Nanozyme: Mechanisms and a Brief Charge-Spin Machine Learning Model

Machine-learning-accelerated Discovery of Single-Atom Catalysts Based on Bidirectional Activation Mechanism

AI-Powered Knowledge Base Enables Transparent Prediction of Nanozyme Multiple Catalytic Activity

Graphdiyne Oxide: a New Carbon Nanozyme

Molecular insights of nanozymes from design to catalytic mechanism

Accelerated discovery of superoxide-dismutase nanozymes via high-throughput computational screening

Boron doped graphdiyne: A metal-free peroxidase mimetic nanozyme for antibacterial application

Filling in the Gaps Between Nanozymes and Enzymes: Challenges and Opportunities

Accelerating Atomic Catalyst Discovery by Theoretical Calculations‐Machine Learning Strategy

Nanozyme Catalytic Performance Enhancement through Defect and Electronic Structure Regulation of Metal-Doped NiCo2O4@Pd