Understanding Capacity Fade in Organic Flow Batteries by Combining Experiments with Modeling and Uncertainty Quantification

氧化还原液流电池（rfb）是未来最有前途的技术之一，它可以廉价地储存间歇性可再生能源（如太阳能和风能）产生的电网规模的电力。因此，它们对于减少对化石燃料能源的依赖非常重要。这些电池需要含有电活性物质的液体电解质。有机分子作为电活性物质是有吸引力的候选者，因为它们可能使rfb的总成本低于传统（例如锂离子）电池。然而，大多数实验室规模的有机液流电池在循环过程中表现出很高的容量损失率，这使得它们不适合商业用途。该项目将通过将实验测量与建模和统计方法相结合，推进对这些能力损失根源的认识。此外，与密歇根大学自然历史博物馆合作，研究人员将建立一个研究站，突出电池储能在电网脱碳中的作用。了解有机液流电池的容量损失是具有挑战性的，因为候选电解质反应物包含了很大范围的分子类别，因此容易受到各种各样的转换和损失机制的影响，这些机制可能相互作用和/或在重叠的时间尺度上起作用。该项目侧重于通过建模和贝叶斯统计学习来解决这个组合问题，贝叶斯统计学习能够识别多个动态交互输入如何组成给定的实验输出。贝叶斯统计框架为噪声和不确定性提供了严格的数学表征，因此提供了一种途径，可以根据循环数据衡量各种假设的容量损失机制的概率，并指导未来的实验设计，以产生最有用的数据。通过在此基础上分析有机液流电池的性能，并结合基于物理的电池电化学模型，将根据与电解质酸度、电池的电荷状态和反应物浓度等参数相关的化学和电化学机制来描述容量损失。实验、建模和基于数据的统计学习的协同使用补充了更传统的技术，并将为减少容量损失的合成化学和电化学战略提供信息。研究工作也将导致对实际操作条件下RFB性能权衡的更好理解。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。

Redox-flow batteries (RFBs) are among the most promising future technologies for cheaply storing grid-scale electricity from intermittently available renewable (e.g. solar and wind) power. Hence, they are important for reducing dependence on energy from fossil fuels. These batteries require liquid electrolytes containing electro-active species. Organic molecules are attractive candidates as electro-active species because they may allow the overall cost of RFBs to be less than conventional (e.g. Li-ion) batteries. However, most laboratory-scale organic flow cells exhibit high rates of capacity loss over time during cycling, which render them unsuitable for commercial use. The project will advance knowledge of the origins of these capacity losses by combining experimental measurements with modeling and statistical methods. Also, in collaboration with the University of Michigan’s Museum of Natural History, the investigators will construct a research station highlighting the role of battery-based energy storage in decarbonizing the electric grid. Understanding capacity losses in organic flow cells is challenging, because candidate electrolyte reactants encompass a large range of molecular classes and are therefore susceptible to a wide variety of conversion and loss mechanisms that may mutually interact and/or operate on overlapping timescales. This project focuses on addressing this combinatorial problem via modeling and Bayesian statistical learning, which is capable of discerning how multiple, dynamically interacting inputs compose a given experimental output. The Bayesian statistical framework offers a rigorous mathematical characterization of noise and uncertainty, and therefore provides a pathway to weighing the probabilities of various hypothesized capacity loss mechanisms in light of cycling data and guiding future experimental designs that produce the most useful data. By analyzing organic flow cell performance on this basis and together with a physics-based electrochemical model of the cell, capacity loss will be described in terms of chemical and electrochemical mechanisms related to parameters such as the electrolyte acidity, the state of charge of the cell, and reactant concentration. The synergistic use of experiment, modeling, and data-informed statistical learning complements more traditional techniques and will inform synthetic chemical and electrochemical strategies for capacity loss mitigation. The research efforts will also lead to an improved understanding of RFB performance tradeoffs under realistic operating conditions.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.