Computational methodology to determine rare event chemical reaction dynamics and networks

确定罕见事件化学反应动力学和网络的计算方法

• 批准号: 1764230
• 负责人: Graeme Henkelman
• 金额: $ 46.95万
• 依托单位: University of Texas at Austin
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2018
• 资助国家: 美国
• 起止时间: 2018-08-01 至 2022-01-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1764230&HistoricalAwards=false
• 关键词: Computational; methodology; determine; rare; event

Graeme Henkelman of the University of Texas at Austin is supported by an award from the Chemical Theory, Models and Computational Methods program in the Division of Chemistry to address a major fundamental challenge to the field of computational chemistry and materials science. This challenge is to model the dynamics of active materials over experimental time scales. Conventional computational algorithms allow for simulations of atomic scale systems for very short times: from nanoseconds to microseconds. Such simulations are useful, but they are a million times smaller than the timescale needed to study important systems and processes including catalysis and batteries. Henkelman and coworkers aim to bridge this so-called "timescale gap" through the further development and improvement of a method called adaptive kinetic Monte Carlo (AKMC). In AKMC, an efficient exploration of the potential energy surface is used to determine reaction rates and reaction mechanisms. The goal of this project is to overcome the technical and algorithmic challenges related to applying AKMC to the discovery of new catalysts and functional materials. A significant broader impact of this project is the development, distribution and support of a software tool that is freely available to the entire research community. This software permits scientists to directly model their materials over relevant experimental timescales.The efficiency of AKMC relies on transition state theory, where the simulation timescale is determined by the rate of the slow chemical transitions of interest rather than the vibrational timescale of atomic vibrations, which is the limitation for molecular dynamics. While AKMC has been used routinely for model systems based upon empirical potentials - fit to experiment - it needs to be extended to accurate calculations based upon quantum mechanics which can be used to model systems that are relevant to, for example, energy applications. For that, AKMC need to be coupled to standard chemical and materials modeling software, based upon density functional theory, so that the scientists and engineers working to discover new catalysts and improved materials can use this methodology to simulate the timescales of relevance for their applications. A limitation of AKMC and an ongoing challenge that is addressed in this project is the problem of low barriers. An explicit treatment of the state-to-state kinetics can require an intractable number of KMC transitions over low barriers between the higher barrier events of interest. A strategy for overcoming the low barrier problem is to use an analytic solution of the master equation such as the Monte Carlo with adsorbing Markov chains method. In large and disordered systems, however, the number of states connected by low barriers (superbasin states) grows exponentially with system size. Not only is this a problem for the efficient solving of the rate equations, large global superbasins need to be reconstructed whenever an AKMC step alters the superbasin structure. A localization algorithm is being implemented to avoid the combinatorial increase of superbasin states. Development also focuses on a method to automatically build reaction networks of catalytic systems. Specifically, Cu oxidation and catalytic CO oxidation is being modeled on supported metal alloy nanoparticles. The intrinsic activity of these nanoparticles can be a result of direct and dynamic participation of the nanoparticle and support atoms in the reaction mechanisms. In this project, stable states are defined by a clustering algorithm of reactive events from long time scale trajectories. In this way, long time scale dynamics may be used to determine reactivity descriptors for complex catalytic systems.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

德克萨斯大学奥斯汀分校的Graeme Henkelman获得了化学系化学理论、模型和计算方法项目的奖励，以解决计算化学和材料科学领域的重大基础挑战。这一挑战是在实验时间尺度上对活性物质的动力学进行建模。传统的计算算法允许在很短的时间内模拟原子尺度系统：从纳秒到微秒。这样的模拟是有用的，但它们比研究包括催化和电池在内的重要系统和过程所需的时间尺度小一百万倍。Henkelman和同事的目标是通过进一步开发和改进一种称为自适应动力学蒙特卡罗（AKMC）的方法来弥合这种所谓的“时间尺度差距”。在AKMC中，利用对势能面的有效探索来确定反应速率和反应机理。该项目的目标是克服与应用AKMC发现新催化剂和功能材料相关的技术和算法挑战。这个项目的一个重要的更广泛的影响是开发，分发和支持整个研究社区免费使用的软件工具。该软件允许科学家直接在相关的实验时间尺度上对他们的材料进行建模。AKMC的效率依赖于过渡态理论，其中模拟时间标度由感兴趣的缓慢化学跃迁的速率决定，而不是原子振动的时间标度，这是分子动力学的限制。虽然AKMC已经被常规地用于基于经验电位的系统模型——适合于实验——但它需要扩展到基于量子力学的精确计算，量子力学可以用来模拟与能量应用相关的系统。为此，AKMC需要与基于密度泛函理论的标准化学和材料建模软件相结合，以便致力于发现新催化剂和改进材料的科学家和工程师可以使用这种方法来模拟与其应用相关的时间尺度。AKMC的一个限制和在这个项目中解决的一个持续的挑战是低障碍的问题。状态到状态动力学的明确处理可能需要在感兴趣的高势垒事件之间的低势垒上进行难以处理的KMC转换。克服低势垒问题的一种策略是使用主方程的解析解，如吸附马尔可夫链的蒙特卡罗方法。然而，在大型和无序系统中，由低势垒连接的状态（超盆地状态）的数量随着系统规模呈指数增长。这不仅是有效求解速率方程的问题，而且每当AKMC步骤改变超级盆地结构时，就需要重建大型全球超级盆地。为了避免超盆地状态的组合增加，实现了一种定位算法。同时，研究了一种自动构建催化系统反应网络的方法。具体来说，铜氧化和催化CO氧化是模拟负载金属合金纳米颗粒。这些纳米粒子的内在活性可能是纳米粒子和支持原子直接和动态参与反应机制的结果。在这个项目中，稳定状态是由长时间尺度轨迹的反应事件的聚类算法定义的。这样，长时间尺度动力学可以用来确定复杂催化体系的反应性描述符。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

项目成果

Adaptive kinetic Monte Carlo simulations of surface segregation in PdAu nanoparticles

• DOI: 10.1039/c9nr01858a
• 发表时间: 2019-06-07
• 期刊: NANOSCALE
• 影响因子: 6.7
• 作者: Li, Lei;Li, Xinyu;Henkelman, Graeme
• 通讯作者: Henkelman, Graeme

Selectivity for ethanol partial oxidation: the unique chemistry of single-atom alloy catalysts on Au, Ag, and Cu(111)

• DOI: 10.1039/c9ta04572d
• 发表时间: 2019-11-07
• 期刊: JOURNAL OF MATERIALS CHEMISTRY A
• 影响因子: 11.9
• 作者: Li, Hao;Chai, Wenrui;Henkelman, Graeme
• 通讯作者: Henkelman, Graeme

Embedded atom method potential for hydrogen on palladium surfaces

• DOI: 10.1007/s00894-020-04588-x
• 发表时间: 2020-11-09
• 期刊: JOURNAL OF MOLECULAR MODELING
• 影响因子: 2.2
• 作者: Ciufo, Ryan A.;Henkelman, Graeme
• 通讯作者: Henkelman, Graeme

Pair-distribution-function guided optimization of fingerprints for atom-centered neural network potentials

• DOI: 10.1063/5.0007391
• 发表时间: 2020-06-14
• 期刊: JOURNAL OF CHEMICAL PHYSICS
• 影响因子: 4.4
• 作者: Li, Lei;Li, Hao;Henkelman, Graeme
• 通讯作者: Henkelman, Graeme