Two recent energy storage modeling and simulation (mod/sim) R&D proposals developed by Scott Roberts (in Sandia’s Thermal/Fluid Component Sciences Dept.) have been funded. Sandia’s participation in these research activities stems from an ongoing Laboratory-Directed Research and Development (LDRD) joint computational/experimental project led by Anne Grillet (also in Sandia’s Thermal/Fluid Component Sciences Dept.) that is exploring degradation mechanisms in lithium cobalt oxide systems at the individual-cathode-particle level.

The LDRD Research Project

As lithium cobalt oxide (LiCoO2) cathodes are charged, LiCoO2 network delithiation increases the crystal lattice’s spacing and the particles swell. When these particles are packed into a percolating network, as is the case in a battery electrode, this swelling generates significant mechanical stress. These stress-induced degradation mechanisms are being investigated in a LDRD project that fuses experimental characterization of fresh and cycled cathodes and binder response to mechanical environments with unique mesoscale modeling capabilities.

In the LDRD study, we performed coupled electrochemical-mechanical LiCoO2 cathode charging simulations to elucidate (1) the stress-generation mechanisms and (2) the effect of charge rate and microstructure on these stresses. The research team used energy-dispersive spectroscopy (EDS) combined with focused ion beam (FIB) cross-sectioning to reconstruct a LiCoO2 cathode in three dimensions and used the conformal decomposition finite-element method (CDFEM) to simulate this reconstructed microstructure. Sandia’s flagship engineering computational mechanics code Sierra was used to simulate the dynamic distribution of lithium in the electrode and a mechanical model was used to elucidate stress and strain generation during this charging process.

Creation of 3D mesh from surface and background meshes using conformal decomposition finite-element method (CDFEM) for a LiCoO2 cathode: (a) reconstructed surface mesh from Avizo for particle phase, (b) background mesh for CDFEM, and (c) resultant 3D mesh for particle and electrolyte phases from CDFEM. Current density in two LiCoO2 cathode models. (left) 3-phase model: particles, binder & electrolyte. (right) 2-phase model: particles & electrolyte. The 3-phase model shows significantly higher current flow, with most current residing in the binder. Stress distribution under uniaxial tension for a cathode with/without a 100 nm binder coating. The presence of the binder significantly reduces the overall stress levels in the network.
Creation of 3D mesh from surface and background meshes using conformal decomposition finite-element method (CDFEM) for a LiCoO2 cathode: (a) reconstructed surface mesh from Avizo for particle phase, (b) background mesh for CDFEM, and (c) resultant 3D mesh for particle and electrolyte phases from CDFEM. Current density in two LiCoO2 cathode models. (left) 3-phase model: particles, binder & electrolyte. (right) 2-phase model: particles & electrolyte. The 3-phase model shows significantly higher current flow, with most current residing in the binder. Stress distribution under uniaxial tension for a cathode with/without a 100 nm binder coating. The presence of the binder significantly reduces the overall stress levels in the network.

Lithium concentration gradients are generated in regions where the open-circuit voltage is constant, but these gradients have very little effect on the network’s mechanical stress state. Instead, the maximum stresses are always found in the fully charged state and are strongly affected by the local details of the microstructure and particle-to-particle contacts.

These stresses shown in our LDRD study are significantly larger than seen in previous single-particle studies—primarily due to the confinement effects of the percolated network. Having a flexible boundary does relax these stresses, but the residual stress is still greater than that of single-particle studies.

Also, our team found that charging rate has very little effect on the particle network’s stress state. While this is inconsistent with experimental observations, investigating the effective Young’s modulus for particle networks coated in thin layers of a polymeric binder shows that the presence of the binder reduces the maximum stresses by over 50%. Our results suggest that using a polymeric binder can greatly mitigate stress generation and is an important area for future research.

These coupled electrochemical-mechanical mesoscale simulations on real microstructures are the first of their kind and represent a leap forward in computational capability for analyzing realistic battery configurations. Developing this new simulation capability is allowing Sandia to extend its impact into the wider community and improve performance and abuse modeling for commercial battery manufacturers as a key player in two new Department of Energy (DOE) Office of Energy Efficiency and Renewable Energy (EERE)-funded proposals.

Consortium for Advanced Battery Simulation (CABS)

Internal faults exhibited by jelly-roll layers for pinch with 0.25 in ball (a) and 0.5 in ball (b).

Internal faults exhibited by jelly-roll layers for pinch with 0.25 inch ball (a) and 0.5 innc ball (b).

CABS is an Oak Ridge National Laboratory (ORNL)-led, DOE-EERE Vehicle Technologies Office (VTO)-funded project, with Sandia and Lawrence Berkeley National Laboratory (LBNL) participating. The team will investigate how material properties respond to stresses (e.g., crush tests and penetration tests) at the 100 µm × 100 µm scale. CABS will deliver

  1. new experimental protocols for battery performance measurements,
  2. new experimental data for properties and validation,
  3. new validated models of battery operation, and
  4. efficient and usable software deployed through Virtual Battery Integrated Environment open architecture software (VIBE/OAS) and available to the broad battery research and design community.

In this effort, Sandia will use microscopy data to create mesoscale reconstructions to perform coupled thermal, electrical, electrochemical, and mechanical simulations of both normal and abuse scenarios. These coupled simulations will allow insights into the electrochemical and mechanical performance under a variety of operating environments. They will also capture the effect of battery cycling/aging through multiple reconstructions to understand how it affects the safety of the battery under abuse scenarios.

Sandia will apply its highly scalable multi-physics engineering codes, including the Sierra Mechanics suite, and its large-scale computing platforms (e.g., Red Sky, Chama, and Sky Bridge). The multi-lab team will also use experimental characterization facilities at ORNL and LBNL, along with Sandia capabilities, to obtain images of electrode microstructures.

Implementing Microstructural Models to Extend CAEBAT I Simulation Technology Length-Scale Coverage

DOE-EERE’s Vehicle Technologies program funds the Computer-Aided Engineering for Electric-Drive Vehicle Batteries (CAEBAT) project to develop a suite of simulation and computer-aided engineering (CAE) tools for designing batteries—to speed up the R&D cycle and reduce the number of build-and-break steps, particularly in the automotive industry. These CAE software tools are to be user-friendly, multi-physics, 3D, fully integrated, validated, and address materials, electrodes, cells, and packs for the battery community.

Extending CAEBAT I simulation length-scale coverageThis project, led by CD-adapco, will upscale micro-scale mechanical, electrochemical, and effective-conductivity/thermal-output modeling to the macro scale—at an affordable cost—for the battery industry to use.

  • Develop microstructural models as a tool to design battery electrodes that account for intercalation-induced stress and expansion/contraction. This effort would use supercomputers at LBNL and Sandia and simulations will be carried out using geometrically accurate models of particles (morphology and particle size distribution) as well as binders and conductivity enhancers.
  • Provide greater fidelity of macrohomogenous models with microstructural models at a fraction of the computation cost by enhancing the widely used DUALFOIL code to make greater use of microstructural information such as tortuosity and particle shape.
  • Dramatically improve the computation efficiency of current electrochemical and thermally coupled material, cell, module and battery pack models by converting macrohomogenous models into high-fidelity, high-speed reduced-order 3D electrochemical/thermal models.

Sandia will bring to the table expertise in understanding intercalation-induced mechanical stress and volume expansion/contraction. Sandia has been developing a workflow for characterizing lithium-ion electrode microstructures and using those microstructures to simulate the localized, electrochemically driven mechanical stresses within the electrode particulate network.

Understanding the localization and distribution of lithiation-induced stress within lithium-ion electrodes is important for improving battery design because fracture and isolation of particles from the percolated particulate network is believed to be one of the key mechanisms in lithium-ion battery capacity fade. Besides creating localized stresses, individual particle swelling can lead to macroscopic expansion and contraction of the entire electrode. While volume change of individual particles might be well known, the relative contribution of the particle volume change that goes into reducing the porosity vs macroscopic volume change relies on a detailed understanding of the microstructure.

By leveraging the unique computational mechanics tools in the Sierra code suite, Sandia has quickly become a leader in understanding battery performance of through unique mesoscale modeling tools.