(Invited) Towards Model-Based Predictive Battery Electrode Design

Abstract

Lithium ion batteries contain porous electrodes, whose design determines battery performance, cost and life-time. Yet, we are far from knowing what is the best structure, and it is prohibitively time- and resource-consuming to do experimental trial and error production, as there are too many parameters in the production process and in the electrode itself. This motivates to use physico-chemical models that yield insight into performance and limitations of given electrode structures as well as allow optimal design. The need for models that allow structure optimisation, i.e. rapid evaluation of many design variants, basically excludes large 3-dimensional models with complex inhomogeneous structures. On the other hand, homogeneous pseudo-two-dimensional models face limitations regarding the prediction of new electrode designs as the crucial influence of conducting additives and binder are not considered sufficiently. This talk will elucidate a pathway that allows model-based predictive battery design. It shows the application of microstructure models to identify the effective electric and ionic conductivity and electroactive area of electrode structures containing additives, and the transfer of this knowledge via surrogate models into electrochemical P2D models to evaluate the respective electrochemical performance. The effect of additive and active material content and distribution is demonstrated at the example of Li ion battery cathode and all solid state Li battery electrodes. The models are applicable to predict the performance of electrodes in cells. We further elucidate how uncertainties in the production process of electrodes lead to distributed electrode design parameters and performance. Finally, we show and compare two pathways of identifying optimal electrode designs at the example of dual layer electrodes: a coarse-grained scanning of thousands of design variants using 8 parameters reveals strongly different designs for optimal charge and discharge, whereas a high resolution optimisation of two design parameters shows flat performance plateaus. Model-based design optimization without extensive experimental studies will reduce costs and experimental efforts in the development of next-generation batteries.