Formulation and manufacturing optimization of lithium-ion graphite-based electrodes via machine learning

Abstract

Understanding the formulation and manufacturing parameters that lead to higher energy density and longevity is critical to designing energy-dense graphite electrodes for battery applications. A limited dataset that includes 27 different formulation, manufacturing protocols, and performance properties is reported. Input parameters from formulation and manufacturing are varied: slurry composition, mixing protocol, electrode coating gap size, drying temperature, coating speed, and calendering. Measurable outputs from the rheological characteristics, adhesion, and electrochemical testing are recorded. A database with the inputs and output parameters is populated and used to train an artificial intelligence model. Validation of the model is performed upon test data and an optimized electrode formulation and manufacturing process predicted. The electrode manufactured using the model process shows excellent cycle life and capacity agreement to prediction. The data model can be used to predict and design the formulation and manufacturing process to produce thick, high-coat-weight, graphite-based electrodes.