CAREER: Chemical Network Based Understanding and Prediction of Electrolyte Decomposition in Batteries

为了持续满足不断增长的电力需求以及电动汽车、可再生能源存储和负载均衡等新应用的要求，需要新型电池。开发新型电池电解质的一个核心挑战是对导致电池失效的电解质降解途径的理解不足。这个CAREER项目将开发计算方法来全面阐明这些电解质反应。该项目的成功将改变在昂贵的合成和测试之前使用计算方法设计和实施电解质的方式。由此产生的关于电解质降解化学的新知识和新的模拟方法将使下一代从事电池研究的科学家和工程师受益。该项目的综合教育计划包括与非研究机构的学生和教师一起开发动手研究项目，并为化学工程师开发继续教育内容。本CAREER项目的总体目标是从第一性原理建立电解质降解反应，作为表征、优化和设计新型电池电解质的基础。所提出的方法将利用两项最新突破的新组合，使其成为从计算角度研究电解质降解问题的理想时机。首先，我们将使用现代半经验量子化学提供所需的模拟，以全面描述与电解质降解相关的复杂反应网络。其次，我们将应用迁移学习模型来提高描述凝聚相和电极界面反应的准确性。结合起来，这些策略将解决计算精度和成本之间的权衡，这限制了反应网络表征在液体电解质中的应用。这些模拟方法最初将应用于在当代锂离子和后锂离子电池中广泛使用的液体电解质配方。尽管这些都是已建立的电解质，但驱动固体电解质间相形成和电解质相关失效的降解化学仍未完全解决。因此，通过关注这些电解质的化学性质，我们将能够验证我们的表征方法，同时也提供了这些电解质中发生的降解反应的第一个完整视图。通过这些表征生成的反应数据也将用于创建反应数据库，这将促进机器学习活动，旨在优先考虑甚至绕过更昂贵的基于电解质的物理模拟。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。

New types of batteries are needed to sustainably meet growing power demands and the requirements of novel applications like electric vehicles, renewable energy storage, and load leveling. A central challenge in the development of new electrolytes for batteries is poor understanding of the electrolyte degradation pathways that lead to battery failure. This CAREER project will develop computational methods to comprehensively elucidate these electrolyte reactions. Success in this project will transform how electrolytes are designed and implemented using a computational approach before costly synthesis and testing. The resulting new knowledge about electrolyte degradation chemistry and new simulation methodologies will benefit the next generation of scientists and engineers working on batteries. This project’s integrated education plan includes developing hands-on research projects with students and faculty at non-research institutions and developing continuing education components for chemical engineers.The overarching goal of this CAREER project is to establish electrolyte degradation reactions from first principles as the basis for characterizing, optimizing, and designing novel battery electrolytes. The proposed methods will leverage a novel combination of two recent breakthroughs, making this an ideal time to investigate electrolyte degradation problem from a computational perspective. First, we will use modern semi-empirical quantum chemistry to provide the simulation throughout required to comprehensively describe the complex reaction networks associated with electrolyte degradation. Second, we will apply transfer learning models to improve the accuracy of describing reactions in condensed phases and at electrode interfaces. In combination, these strategies will address the tradeoff between computational accuracy and cost that has limited the application of reaction network characterizations to liquid electrolytes. These simulation methodologies will initially be applied to liquid electrolyte formulations that have widespread usage in contemporary Li-ion and post Li-ion batteries. Despite the fact that these are established electrolytes, the degradation chemistry that drives solid-electrolyte interphase formation and electrolyte-related failure is still incompletely resolved. Thus, by focusing on these electrolyte chemistries, we will be able to validate our characterization methodology while also providing the first complete view of the degradation reactions that occur in these electrolytes. The reaction data generated by these characterizations will also be used to create reaction databases that will facilitate machine learning activities aimed at prioritizing or even circumventing more costly physics-based simulations of electrolytes.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.