标题： 基于第一性原理的表面反应反应物通量理论：多通道与穿越动力学 显示英文标题

标题： First Principles Reactive Flux Theory for Surface Reactions: Multiple Channels and Recrossing Dynamics

摘要： 异质反应通常由多个基元步骤组成，其速率系数在阐明这些过程的机理和微观动力学方面具有根本重要性。 计算表面反应速率系数的过渡态理论（TST）通常仅依赖于吸附物振动的简谐近似，并忽略了回溯动力学。 在这里，我们首次将一种高效的元动力学增强采样方法与更一般的反应通量方法相结合，以计算任何阶数的表面反应和/或具有多个反应坐标的速率系数，克服了TST的这些局限性。 我们将这种方法应用于一个教科书中的表面反应，即Pt(111)上的CO氧化，对于该反应，速率常数已经精确测量，使用从第一性原理数据构建的全维神经网络势能面。 通过引入三个协同变量，获得了准确的多维自由能面，得到了CO氧化和竞争性CO脱附的速率系数，与实验数据相符。 有趣的是，我们的结果揭示了两个通道中显著的动力学回溯，然而这源于不同的物理机制。 这种方法代表了一种从第一性原理计算基元表面过程速率系数的准确且通用的框架，这对于开发异质催化预测动力学模型至关重要。 显示英文摘要

摘要： Heterogenous reactions typically consist of multiple elementary steps and their rate coefficients are of fundamental importance in elucidating the mechanisms and micro-kinetics of these processes. Transition-state theory (TST) for calculating surface reaction rate coefficients often relies solely on the harmonic approximation of adsorbent vibrations and neglects recrossing dynamics. Here, we combine, for the first time, an efficient metadynamics enhanced sampling method with a more general reactive flux approach to calculate rate coefficients of surface reactions of any order and/or with multiple reaction coordinates, overcoming these limitations of TST. We apply this approach to a textbook surface reaction, CO oxidation on Pt(111), for which rate constants have been precisely measured, using a full-dimensional neural network potential energy surface constructed from first-principles data. An accurate multi-dimensional free-energy surface is obtained by incorporating three collective variables, yielding rate coefficients for both CO oxidation and the competing CO desorption that are in good agreement with experimental data. Interestingly, our results reveal significant dynamic recrossing in both channels, which however arises from distinct physical mechanisms. This approach represents an accurate and general framework for calculating rate coefficients of elementary surface processes from first-principles, which is vital for developing predictive kinetic models for heterogenous catalysis.