Mechanical Failure Mechanism of Silicon-Based Composite Anodes under Overdischarging Conditions Based on Finite Element Analysis

Abstract

Overdischarge is a severe safety issue that can induce severe mechanical failure of electrode materials in lithium-ion batteries. A considerable volume change of silicon-based composite anodes undoubtedly further aggravates the mechanical failure. However, the mechanical failure mechanism of silicon-based composite anodes under overdischarging conditions still lacks in-depth understanding despite many efforts paid under normal charging conditions. Herein, we have modeled and tracked the mechanical failure evolution of silicon/carbon nanofibers, a typical silicon-based anode, under overdischarging conditions based on the finite element simulation, with derived optimization strategies of optimal Young’s modulus and stable microstructure. The severe contact damage between silicon nanoparticles and carbon nanofibers, which causes larger shedding and breakage risks, has been found to contribute to mechanical failure. To improve the electrode stability, an optimal Young’s modulus interval ranging from ∼75 to ∼150 GPa is found. Furthermore, increasing the embedding depth of silicon nanoparticles in carbon nanofibers has proven to be an effective strategy for improving electrochemical stability due to the faster lithium salt diffusion and more uniform current density distribution, which was further verified by the experimental capacity retention ratio of carbon-coated silicon and silicon/carbon nanofibers (84 vs 75% after 100 cycles). Our results provide meaningful insights into the mechanical failure of silicon-based composite anodes during overdischarging, giving reasonable guidance for electrode safety designs and performance optimization.

Breathing of a Silicon-Based Anode: Mechanical Discrete Approach Using DEM

“…The most promising part occurs at low Young'modulus (E < 5 GPa), where some part of the swelling seems absorbed by the softness of particles, presenting an eventual way to limit the breathing by reducing the particles stiffness. Nevertheless, in practice, the material used for anode is located at much higher range of Young's modulus (E > 5 GPa, 73,74 colored rectangle in Fig. 6b), where the system rigidity becomes such that little breathing changes occurs.…”

Epitaxial growth of hexahedral Fe2O3@SnO2 nano-heterostructure for improved lithium-ion batteries

“…The internal thermal strain ( ε t ) derived from thermal expansion can be given by. 37 ε t = α ( T − T ref ) = α Δ T where α is the thermal expansion coefficient, T ref is the reference temperature, which represents a state of zero stress caused by lithiation. Δ T was employed to represent the state of expansion caused by lithiation during the discharging process in this study.…”

“…The stress and strain induced in the electrochemical process can be modeled by the stress and strain in the thermodynamic process, based on the previous study. 37 The thermal stress is defined by ε t = α ( T − T ref ) where the α , T and T ref are the thermal expansion coefficient, temperature, and reference temperature, respectively. The stress is σ t = E t α Δ T where E t is Young's modulus of active substance and the value of Δ T is equal to the value of T − T ref .…”

A Review of Performance Prediction Based on Machine Learning in Materials Science

“…Yang et al [ 50 ] used the Gaussian random field (GRF) method to create a synthetic microstructure image dataset of materials with different compositions and dispersion patterns. In addition, there are several common means of generating data today, such as the use of the finite element method (FEM) to generate high-contrast composites [ 51 ], two-dimensional (2D) mosaic composites [ 52 , 53 , 54 ], as well as microstructure [ 55 , 56 , 57 , 58 ] and datasets [ 59 , 60 , 61 , 62 , 63 ] of three-dimensional (3D) materials. The existing studies suggest that combining diverse knowledge from materials science, solid mechanics, and other related fields can generate datasets that are more representative of the design space and thus give better results with the applied models.…”