MultiScale and MultiTime Image-Based Control and Characterization of Lithium-Ion Batteries and Materials

Lithium Ion batteries are the main energy storage devices today, from portable devices to electric vehicles. From improving the understanding of their properties, their ageing, or their modes of failures, to investigating new materials and designs, imagebased analyses and measurements can play a key role to support the industry and research in this sector. From X-ray Computed Tomography (CT) to FIB-SEM (Focused-Ion Beam – Scanning Electron Microscopy), including recent Plasma FIB, images can be acquired to display the whole assembly, down to the microstructure of the materials involved. Thermo ScientificTM AvizoTM Software (Thermo Fisher Scientific, Waltham, MA, US) allows for visualizing, processing, and quantifying such data in a controlled, robust and reproducible manner, and can therefore help at all stages. In this paper, we present images and analyses results from different experiments, at the microand nano-metric scales, covering problematics of battery ageing, defect identification and detailed analysis using correlative microscopy from microCT to FIBSEM.


Introduction
The analysis of batteries and materials involved in their fabrication raises many challenges for image acquisition and analysis, from temporal evolution through their life cycle to the multiscale study of their electrical, thermal or mechanical properties.Starting from a microCT of an entire Lithium-Ion cell, we present methods for visual inspection of the data, and then for its segmentation and the extraction of the precise geometry of the rolled electrodes.This automated segmentation allows us to extract quantitative information about the surface of the electrodes, detailing their respective heights and 'unrolled length', as well as the local thickness.As we performed two µCT scans from the same battery, first in a fresh condition and then after cycles of charge/discharge until failure [1], we present both visual and quantitative results demonstrating the evolution of the assembly.Finally, we present a correlative experiment, where a spot of interest is detected on the micro-CT, and then imaged and analyzed with FIB-SEM.High-resolution imaging, using SEM or nanotomography opens the path to very detailed microstructure characterization, as described e.g. in [2,3], allowing precise defect investigations or estimation of physical properties.

Micro-CT imaging
MicroCT enables non-destructive image acquisition of complete cells.This allows for characterizing the sample for controlling the production process, or to monitor the evolution of the structure after specific usage conditions.We present analyses carried out on a complete INR18650 energy cell, scanned in its entirety at 7µm resolution using Thermo Scientific HeliScan™ microCT, in a fresh state and after failure due to extended cycles of charge and discharge.

Fresh Cell
We propose visual inspection techniques accounting for the specific cylindrical and spiral geometry, and methods and results for segmenting the main components, and quantifying the thickness, surface, height and length of rolled electrodes.Avizo Software propose off-the-shelf tools for interactive visualization and data exploration: volume rendering, arbitrary slicing or unrolling cylindral slices without any prior segmentation.The high quality images obtained using the Heliscan microCT allows for simple automated segmentation of several structures inside the cell, including the detection of cathode and the anodealthough in the latter case, it is the copper collector of the anode that is most easily segmented.The central area, including the central pin, is also easily segmented to measure its area throughout the height of the battery.
As the cathode is composed of 2 separate parts, while the anode is made of single sheet, we used the latest to extract an explicit geomtetrical model of the spiral representing the roll.This was done by using a centreline tracing algorithm on a set of orthogonal slices, except for the very top and bottom borders where the tracing was done in 3D.Given such a spiral mode, we investigated a prototype (not included in the product) tool for proposing a visual inspection of the virtually unrolled eletrodes.Finally, we cut the traced centreline obtained at different height levels, which corresponding to the anode collector, to start and terminate at the same level than the cathode roll.All the spirals were then sampled uniformly along their length, so as to obtain a precise, regular and reproducible geometrical representation of the electrodes.This representation allows for a straightforward estimation of the height and its unrolled length.The thickness of the electrodes is measured at each point of the spiral as the distance to the closest spiral point on the 'next layer' (by discarding nearest points which have a curvilinear distance that is too low).The average thickness between 2 layers of the anode collector is estimated to 375µm, with a standard deviation of 22µm.

Used Cell
After the first scan, the battery was charged and discharged until failure.A new scan was performed with similar conditions as previously, and the images were rigidly aligned to facilitate comparison.Direct observation shows significant evolutions happening within the cell.Most striking is the motion of the central pin toward the positive tab, and deformations in the electrodes.The same image analysis protocol could be reapplied without difficulty.Thanks to the robust geometrical reprentation chosen previously, the measures generated are directly comparable.They reveal a global expansion of the electrode, with an average increase of the electrode height from 60.06 to 60.35 mm (+0.48%), and of its unrolled length from 591.1 to 598.8 mm (+1.28%).
The measured average thickness also increased slightly from 0.36mm to 0.37mm.The precise observation of the deformation reveals a general swelling and slight unrolling of the electrode, together with substantial localized deformations and folding.
Figure 7: Left: spirals traced and overlaid on a section of the aged battery scan.Right: analysis of the shape of electrodes before and after ageing.

Correlative microCT / FIB-SEM analysis
Especially in batteries and cells, macroscopic properties -electrical, but also thermal and mechanicalfind their origins at a micro-or nano-scale.We will present our approach for correlating microCT and SEM, relying on a specific sample holder, alignment of surfaces extracted from microCT images on one side, and multi-view SEM images.Such multimodal correlative workflows allow investigating the multiscale nature of these properties, but also to finely characterize defects that can be detected at the microCT level.

From microCT to FIB-SEM
After identifying a region of interest from the first microCT experiment, the battery was cut near the sport of interest.A small sample of the electrode was extracted and mounted on a sample holder.New microCT scans were made at 0.7µm/pixel, before performing multivew SEM acquisition of the sample surface.Avizo software was again used to rigidly align both scans, and selecting new regions of interest.Finally, a FIB/SEM structural analysis with Thermo Scientific Helios TM G3 Plasma FIB at 18nm in plane resolution for a total volume of 150µmx150µmx50µm.

FIB-SEM data analysis
Using Electron Microscopy (FIB-SEM), high resolution images can be acquired which can reveal the structure of the electrodes or the separator.Besides the volume fraction or surface areas of the different phases, the connectivity of pores and/or particles, their surface of contacts, path tortuosity or constrictivity, permeability or molecular diffusivity, become accessible.

Conclusion
We propose two experiments about the image acquisition and analysis of Li Ion batteries for the purpose of monitoring their ageing, and investigate sites of interest in a correlative study setup.The first experiment investigates the evolution of the electrode before and after cycles of charges and discharges until failure, from micro-CT acquisitions of the entire battery acquired at 7µm resolution using Thermo Scientific HeliScan.An image analysis protocol implemented using the Avizo enables reproducible quantitative analysis of the electrode size, thickness and shape, and revealing general expansion of the electrodes, slight unrolling, as well as localized foldings of the structure.Regions of interest are selected and a correlative experiment is then proposed to further investigate the microstructure using Electron Microscopy.

Figure 2 :
Figure 2: Left: Subset of the extracted anode collector spiral surface, Right: virtually unrolled volume

Figure 3 :
Figure 3: Left: Traced spiral, with starting and ending points corresponding to the extent of the cathode.Upper Right: the Central Area is defined as the area enclosed by the inner portion of the anode collector.Lower Right: plot displaying the shape of the traced anode spiral on the central cross section.

Figure 4 :
Figure 4: Quantitative analysis of the fresh battery.The central image shows the estimated electrode thickness, after a virtual unrolling of the electrode.The left side of the image corresponds to the inner part of the battery.

Figure 5 :
Figure 5: Fresh and Aged battery, slices taken at the same positions after rigid alignement

Figure 6 :
Figure 6: Quantitative comparison between the Fresh and Aged scans of the battery.

Figure 9 :
Figure 9: Left: Connectivity of Ni particles in Li-Ion battery cathode.Right: Permeability simulation and estimated pressure field