Correlative Multiscale 2D and 3D Imaging Workflow to Reveal Chemo-Mechanical Mechanisms in Batteries

Silicon-based anodes are widely recognized as a key technology for next-generation lithium-ion batteries due to their exceptionally high theoretical specific capacity and low operating potential [1,2]. In combination with state-of-the-art layered oxide cathodes, silicon offers a viable path to higher energy density batteries for electric mobility and stationary storage [2], [3-5]. However, the commercial implementation of silicon-rich anodes remains challenging due to chemical-mechanical degradation associated with lithium insertion and extraction. The large volume changes associated with (de)lithiation generate stresses that alter particle morphology, disrupt electronic percolation and destabilize electrode interfaces, leading to significant irreversible capacity losses [1-8]. Mitigating these degradation processes requires direct insight into the lithiation pathways of silicon, stress development and microstructural evolution across multiple length scales, from electrode architecture down to individual particles. Recent advances in three-dimensional (3D) microscopy, particularly synchrotron-based X-ray computed tomography (SXCT), enable non-destructive, in situ investigation of electrodes under electrochemical operating conditions. When embedded in a correlative multiscale workflow and combined with advanced image analysis and electron microscopy, these techniques enable quantitative, statistically relevant investigations of the chemical and mechanical evolution of complex electrode microstructures. Here, we present a correlative multiscale microscopy workflow that combines in situ 3D synchrotron nano-tomography with convolutional neural network (CNN)-based segmentation and complementary field emission scanning electron microscopy (FESEM), focused ion beam SEM (FIB-SEM) tomography and targeted transmission electron microscopy (TEM) to resolve chemo-mechanical degradation phenomena across multiple length scales and to establish correlations between local microstructural transformations and electrochemical performance. In situ SXCT provides time-resolved three-dimensional gray-scale phase-contrast information of silicon-rich anodes during the electrochemical cycle, enabling direct observation of heterogeneity at the particle and electrode levels. The 3D reconstructions reveal different lithiation pathways across Si particles embedded within the anode. While a classic core-shell transformation characterized by a radially advancing amorphization front is observed in some particles, most exhibit non-core-shell behavior. These particles develop internally connected, spatially complex lithiation patterns that propagate along preferred pathways within the particle volume, pointing towards that silicon lithiation proceeds via heterogeneous reaction fronts rather than a uniform radial mechanism. CNN-based semantic segmentation enables automated extraction and 3D rendering of individual silicon particles embedded in the intact electrode microstructure, allowing particle-resolved analysis within statistically relevant volumes. Correlative FESEM cross-sections and energy-dispersive X-ray spectroscopy (EDS) show pronounced variability in particle deformation, fracture and particle-matrix interaction after cycling, confirming the heterogeneity observed in the X-ray data. FIB-SEM tomography resolves particle-scale microstructural features, including local porosity, composite phase distribution and interface morphology, providing a nanoscale context for the observed 3D lithiation pathways. Targeted TEM analysis extends the workflow to the nanometer and atomic scales, where electron energy loss spectroscopy (EELS) and EDS enable localized chemical and electronic structure analysis of lithiated silicon, interphases and additive-derived reaction products. Together, this correlative workflow links particle-level transformation behavior to local chemical, mechanical and architectural constraints imposed by the surrounding electrode microstructure. These results underscore the strong coupling between electrolyte chemistry, local lithiation pathways and mechanically driven microstructural evolution in silicon-based anodes [11]. Correlative multiscale characterization workflow for silicon-based LIB anodes. (a) Schematic of the in situ lithium-ion battery cell employed SXCT. Below, a rendered SXCT dataset of a selected pristine anode region visualizes the silicon particle morphology. Machine-learning–based prediction models are applied for rapid phase segmentation of grayscale nano-tomography data at different states of charge (SoC). (b) Electrode-scale analysis: Cross-sectional FESEM images of the pristine anode (top) and the delithiated anode after one electrochemical cycle (bottom), revealing pronounced irreversible electrode alterations. Particle-scale analysis: FESEM investigation of individual silicon particles upon cycling. (c) FIB-SEM tomography extends the correlative approach by providing high-resolution three-dimensional reconstructions of the electrode microstructure at different cycling states. (d) TEM in combination with EELS and EDS offers complementary nanoscale and atomic-scale insights into local microstructural and chemical changes within silicon particles and interphases.