Nowadays, lithium-ion batteries (LIB) are largely widespread. In most of these power sources, negative electrodes are made of graphite, used as a reversible intercalation material for lithium. Graphite presents good electronic properties, a specific capacity of 372 mAh/g, a low operating potential and a low cost, making it a reliable and cost-effective active material for LIB. Nevertheless graphite-based electrodes undergo a multitude of irreversible physical and chemical changes under cycling, responsible of the major causes of cell capacity fade and performance loss. As examples, electrolyte reacts in contact with intercalated lithium in graphite and precipitates as a Solid Electrolyte Interface (SEI) on active material surface. This latter layer grows continuously and decreases intercalation performance. Under high currents or low temperatures another degradation called lithium-plating occurs. Lithium is locally deposited on the surface of the graphite particles rather than being intercalated, and lithium aggregates risk to be electrically isolated. During cycling, graphite volume changes cause mechanical stresses inside particles, resulting in cracks and fractures, which decrease effective particle size and create additional fresh surface immediately covered with SEI. These phenomena and many others lead to loss of lithium inventory, loss of active material and intercalation performance reduction. At a cell level, ageing mechanisms result basically in a capacity fade and a power performance loss, but even if ageing consequences are simple to quantify, ageing causes are diverse, different and coupled [1]. In consequence, the lack of understanding of the physical origin of capacity and performance fading in LIB is often a bottleneck in the development of simple robust and predictive Battery Management System (BMS). To optimize them, a better understanding and quantification of ageing sources are necessary.

To address this issue, in this work, physics-based models of graphite electrodes have been developed including coupled characteristic ageing models, which are then compared with electrochemical experiments at cell scale. Models provide local information on the intricate physical mechanisms that take place inside electrodes. Thus simulations bring quantitative data to be confronted to operando measurements.

In this study the models have been validated on performances at beginning of life, then during ageing. The studied system consists of common Graphite and NMC (111) electrodes, with different loadings. The coupled ageing mechanisms modeled during cycling are the evolution of SEI, the occurrence of lithium plating and the modification of particle size distribution.

A pseudo two-dimensional (P2D) model, parametrized from literature and experimental data is build. Physical parameters are adjusted with galvanostatic charge and discharge of separate electrodes and a complete pouch cell, including a reference electrode. Geometrical parameters such as active surface area and effective particle dimension are adjusted from analysis of cyclo-voltamperometry curves.

Physical variables along thickness are explored with the validated model at different currents and mass-loadings. Simulations show that lithiation heterogeneity along thickness is caused by the typical electrode equilibrium potential derived from the staging mechanism of graphite. Intercalation current density near separator reaches more often critical values than in deeper area, which could locally influence ageing rate.

Models for the most critical ageing mechanisms are developed and validated. In this work, competition between SEI growing and Lithium-plating is analyze both numerically and experimentally. On a first approach a lumped model of SEI is fitted to cycling data and confirmed the growing mechanism of SEI: lithium consumption depends on the square roots of time but at a SEI thickness threshold, capacity fading becomes time-proportional. Then an original cycling procedure is proposed to favor lithium plating to different aged cells. Analysis of the results shows the reversible loss and the gradual resilience to irreversible loss versus cycling time. Consequently a mechanistic lithium-plating evolution has been added to the SEI lumped model. Model simulates a growing reversible lithium layer between graphite and SEI, which becomes irreversible when it breaks the SEI layer. Finally, validated ageing laws on the lumped model are implemented locally on the electrode model in order to spatially quantify ageing.

[1]: J. Vetter, P. Novak, M. R.Wagner, C. Veit, K.-C.Möller, J. O. Besenhard, M.Winter, M.Wohlfahrt-Mehrens, C. Vogler, and A. Hammouche, J. Power Sources, 147(1–2), 269 (2005).