GeoDict Battery Applications

GeoDict provides powerful capabilities for the digital analysis and microstructure simulation of lithium-ion battery materials. From importing and segmenting 3D scan data, to computing material properties, to coupled electrochemical and mechanical simulations, GeoDict helps battery researchers efficiently identify material weaknesses in a virtual environment, significantly shortening R&D cycles and reducing experimental costs.

The GeoApp Battery Capacity Test offers a convenient method to compute the accessible cell capacity for different charge rates of a battery or an electrode.Fast charging reduces the accessible cell capacity, and GeoDict simulates this capacity change with varying charge rate.The GeoApp automatically creates a four‑step charging protocol: charging with constant current, charging with constant voltage, relaxation, and discharging with constant current.Simulations are run for different charge rates, and well‑arranged results allow easy comparison of accessible capacities.Users can also compare charging simulations with experimental data by selecting the voltage range, charge rates, and switching conditions.The GeoApp visualizes charge‑rate‑dependent transferred capacities, cell profile over time, and the Ragone plot containing power density and energy density for each tested charge rate.

Application Examples

1. Anisotropic Diffusion in Charging Simulations

Graphite and other common battery active materials exhibit significant direction‑dependent lithium‑ion diffusion due to their atomic structure.In graphite, diffusion along the graphene plane is about 4.4×10⁻¹⁰ m²/s, while perpendicular to it is about 8.7×10⁻¹⁶ m²/s—a difference of roughly six orders of magnitude.GeoDict's BatteryDict can account for anisotropic diffusion properties and their interaction with electrode microstructure during charge/discharge simulations.This capability can be applied to both artificially generated graphite electrode models and microstructure models reconstructed from µCT or FIB‑SEM scans, using GrainFind to identify individual graphite particles and apply orientation‑dependent diffusion characteristics.

2. Segmentation Methods for FIB‑SEM Images

FIB‑SEM provides nanoscale images of lithium‑ion battery cathodes, where active material, binder, and pores can be distinguished.However, these images present processing challenges—if the sample is not impregnated before imaging, structures behind pores may appear, and ion‑beam cutting of different materials can create curtain artifacts.Three machine learning segmentation methods are compared for accurately distinguishing material phases: Boosted Tree segmentation (fast training, less manual annotation, no GPU required, but feature selection significantly affects results), 2D U‑Net (deep learning‑based, requires GPU, better particle segmentation but may show discontinuities between slices), and 3D U‑Net (utilizes 3D contextual information, performs best for binder phase segmentation with no discontinuities between slices).

3. Experimentally Validated Aging Simulation

In a joint study by Math2Market and ETH Zürich, experimental research used 3D in‑situ X‑ray tomography and electrochemical characterization to analyze microstructural changes in NMC cathodes and graphite anodes during battery cycling, while digital research used GeoDict to compute microstructural stress and swelling caused by local lithium‑ion concentration.GeoDict couples 3D electrochemical charging simulations with mechanical deformation simulations, linking local lithium concentration, volume changes, material deformation, and damage.The simulation results were validated against experimental data—for a graphite anode half‑cell, the measured particle swelling rate was 10%, while the simulation result was 9.8%.This experimentally validated digital workflow can be used to study the relationship between microstructural changes and performance degradation during battery cycling.

4. Balancing Fast Charging and High Energy Density

Achieving both fast charging and high energy density requires not only new electrode material chemistries but also optimization of the microstructure of existing materials.GeoDict can both import and segment CT scan data of electrode materials and generate realistic 3D microstructures for studying geometric parameters, transport properties, and charging behavior.The workflow includes: importing 3D images (using ImportGeo‑Vol and GrainFind to identify active material, binder, conductive additives, and electrolyte), analyzing material properties (using DiffuDict and PoroDict to compute effective diffusion coefficients, tortuosity, and other structural parameters), running fast‑charging simulations (using BatteryDict to calculate voltage, local lithium concentration, and lithium plating risk), generating and modifying digital microstructures (using GrainGeo to vary solid volume fraction, porosity, particle size, shape, and orientation), and comparing different designs.Volkswagen's innovation department has used this workflow to study design criteria for fast‑charging anodes.

These application examples demonstrate GeoDict's core value as a "Digital Material Laboratory" throughout the entire R&D process for lithium‑ion battery materials—from data import and analysis to property computation and optimization—helping accelerate the development of next‑generation high‑performance, high‑safety battery materials.