IDEA: Inverse Design of Electrochemical Interfaces with Explainable AI

Idea Fellowship是一个为期5年的计划，旨在通过新兴的人工智能、应用于基础电化学工程研究的数字转型，为英国的工业脱碳和数字化铺平道路。电化学工程是21世纪许多关键能源技术的核心，如制氢、二氧化碳减排、储能等。所有这些领域的进一步发展都需要更好地了解电化学系统中的电极-电解液界面，因为几乎所有的关键现象都发生在这种界面上，这最终决定了系统的动力学、热力学和长期性能。设计下一代电化学接口以满足未来的要求是所有类型的电化学应用的共同挑战。传统上，电化学界面的设计依赖于高通量筛选实验或模拟。考虑到设计空间的复杂性，这种蛮力方法高度迭代且成功率低也就不足为奇了，这已成为电化学研究界面临的共同挑战。该奖学金的愿景是通过新颖的可解释人工智能(XAI)和数字解决方案，改变未来电化学界面如何在其整个生命周期中设计、优化和自我发展。它将创建一个反向设计框架，在该框架中，我们使用一组期望的性能指标作为XAI模型的输入，以我们可以解释的物理意义的方式生成满足要求的电化学界面设计。一旦开发出该方法，将解决对净零路线图具有核心重要性的典型挑战，其中包括改进目前的系统，如氢气生产/燃料电池和二氧化碳减少，但也开发新的电化学系统，这些系统目前还不存在于工业规模，如氮气减少和多离子能量储存。

The IDEA Fellowship is a 5-year programme to pave the way for the UK's industrial decarbonisation and digitalisation, via emerging AI, digital transformations applied to fundamental electrochemical engineering research. Electrochemical engineering is at the heart of many key energy technologies for the 21st century such as H2 production, CO2 reduction, energy storage, etc. Further developments in all these areas require a better understanding of the electrode-electrolyte interfaces in the electrochemical systems because almost all critical phenomena occur at such interface, which eventually determine the kinetics, thermodynamics and long-term performance of the systems. Designing the next generation of electrochemical interfaces to fulfil future requirements is a common challenge for all types of electrochemical applications. Designing an electrochemical interface traditionally relies on high throughput screening experiments or simulations. Given the complex nature of the design space, it comes with no surprise that this brute-force approach is highly iterative with low success rates, which has become a common challenge faced by the electrochemical research community. The vision of the fellowship is to make a paradigm-shift in how future electrochemical interfaces can be designed, optimised and self-evolved throughout their entire life cycle via novel Explainable AI (XAI) and digital solutions. It will create an inverse design framework, where we use a set of desired performance indicators as input for the XAI models to generate electrochemical interface designs that satisfy the requirements, in a physically-meaningful way interpretable by us. The methodology, once developed, will tackle exemplar challenges of central importance to the net zero roadmap, which include improving current systems such as H2 production/fuel cell and CO2 reduction, but also developing new electrochemical systems which do not yet exist today at industrial scale such as N2 reduction and multi-ion energy storage.