Effect of the grain arrangements on the thermal stability of polycrystalline nickel-rich lithium-based battery cathodes

One of the most challenging aspects of developing high-energy lithium-based batteries is the structural and (electro)chemical stability of Ni-rich active cathode materials at thermally-abused and prolonged cell cycling conditions. Here, we report in situ physicochemical characterizations to improve the fundamental understanding of the degradation mechanism of charged polycrystalline Ni-rich cathodes at elevated temperatures (e.g., ≥ 40 °C). Using multiple microscopy, scattering, thermal, and electrochemical probes, we decouple the major contributors for the thermal instability from intertwined factors. Our research work demonstrates that the grain microstructures play an essential role in the thermal stability of polycrystalline lithium-based positive battery electrodes. We also show that the oxygen release, a crucial process during battery thermal runaway, can be regulated by engineering grain arrangements. Furthermore, the grain arrangements can also modulate the macroscopic crystallographic transformation pattern and oxygen diffusion length in layered oxide cathode materials.


Supplementary Note 1: XANES-3DTXM measurements
X-ray tomography measurement can be achieved by taking TXM projections of a rotating sample at each fraction of a degree. The 2D images collected at all angles are then reconstructed into a 3D volume. The hard X-rays can penetrate through the sample, enabling non-destructive 3D analysis of structure and morphology within the sample. Compared with 2DTXM which is a projection through the entire sample, the 3D structure from X-ray tomography can detect the cracks, porosity, and tortuosity within the sample with a spatial resolution of tens of nanometers.
A more powerful technique actively developed in recent years is coupling TXM with Xray absorption near edge structure (XANES) spectroscopic capability. This is achieved by collecting and reconstructing TXM projections at a series of X-ray energies across the absorption edge of an element. A XANES spectrum can be generated at each voxel of the reconstructed 3D volume by imposing and aligning the absorption objects at different energies, which involves sophisticated image alignment in data processing. The shape of the XANES spectra represents the local chemical fingerprints of the element, therefore the 3D information on oxidation states can be achieved by fitting the spectrum on every voxel of the object. A great advantage of XANES-3DTXM over the most widely used 2D approaches is that the information is depth-averaged in 2D measurements, chemical depth profile along the beam direction is not available, while XANES-3DTXM offers the great capability for finer chemical and morphological features inside the sample. 1 With the high flux of synchrotron X-rays and the development of optics, detectors, and algorithms for fast data acquisition, one tomography measurement could be carried out in minutes, which makes in situ/operando experiments possible. 2 This in situ XANES-3DTXM approach allows us to observe the real-time morphological and chemical evolution at elevated temperatures.
The large field of view (20-40 µm) and reasonably high spatial resolution (~30 nm) offer the capability to analyze NMC cathodes at different length scales, ranging from voids of tens of nanometers, to cracks of hundreds of nanometers, to secondary particles of tens of micrometers.
This so-called 5D mapping approach can greatly help to understand the reaction mechanisms of cathode active materials, and the quantitative analysis methodology is broadly applicable for other energy materials.
The powders for in situ TXM measurements for this study were prepared by dissembling the 4.5 V charged cell, then collecting the powders on the cathode disk. The powders were rinsed immediately with dimethyl carbonate, dried, and then sealed in an Ar-filled quartz capillary. The capillary was mounted on a holder base and then inserted into the chamber. The 3DTXM was conducted by acquiring tomographic images of the particles at 519 angle positions in 180° range.
The XANES-3DTXM was done by repeating the tomographic scans of the same particles in an Xray energy range across the white-line of Ni K-edge in 1 eV step. The heating rate was 5°C/min when changing the temperature. After reaching the target temperature, a 20 mins waiting time was given to stabilize the temperature. The total XANES-3DTXM collection time was ~30 mins for one particle at each temperature.
In order to do quantification and further analysis of the morphological characteristics, the greyscale absorption mapping is normalized and then binarized using Otsu's thresholding method, which is effective in minimizing the intraclass variance of the black and white pixels after binarization. 3

Supplementary Note 2: Neutron Diffraction
Chemically delithiated NMC powders were used for in situ ND study for several reasons: • Firstly, for a better signal, the required amount of materials is large (300~400 mg). Considering obtaining such NMC powders with high purity from charged cells is very challenging, the chemical delithiation was chosen as an alternative. • Secondly, XANES-3DTXM results show the NMC thermal instability originates from two parts: the intrinsic Ni reduction and oxygen release, as well as the morphological factors (e.g., microcracks, voids, etc.), and these two parts cannot be decoupled by solely TXM approach.
Chemical delithiation can open up the secondary particles, resulting in similar cracks ratios, porosity, and tortuosity for gravel-and rod-NMC. Therefore, the intrinsic factor for the stability in NMC can be focused on, while the effect of microcracks can be ruled out. • Thirdly, cell variation and parasitic reactions can lead to some over-or under-estimation of the absolute amount of lithium intercalated into and deintercalated from the host material for electrochemical processes, 4,5 making a direct comparison between two delithiated NMCs difficult. • Finally, the presence of carbon black, polymer binders, and proton-containing electrolyte residual from electrochemically cycled cathodes will impact the scattering signals, making Li site occupancy calculation less convincing.
Neutron diffraction data were collected continuously from 25°C to 250°C, with a collection time of 5 mins per dataset. In addition, a longer exposure time of 30 mins was used at 25°C, 100°C, 150°C, 200°C, and 250°C for better quality. For each in situ dataset, six detector banks with nominal diffraction angles of 7°, 15°, 31°, 65°, 120°, and 150° were simultaneously measured.
For each sample, data from different detector banks with different Q-spacing coverages were analyzed simultaneously for structure refinement at each temperature.

Supplementary Note 3: Refinements on Diffraction Patterns
The average structure before heating was first investigated by performing Pawley fits on the as collected ND data. For Pawley fit the background terms, unit cell parameters, and peak profiles were refined. The results of the Pawley fit were used as starting models for the analysis using the Rietveld method where the scale factor, zero offset, phase fractions, atomic positions, occupancies, and atomic displacement parameters were also refined. The instrument parameters were calibrated by refining a standard Si powder (SRM 640e from NIST) at corresponding temperatures.
In the following analysis of in situ ND patterns, the refinement of phases (phase fraction and lattice parameters) and atoms (position, occupancy, and thermal displacements) is performed in the sequential order, meaning that the values obtained from temperature ! will be the initial values for refining at the next temperature !"# . We applied constraints to atoms on the same site to have the same thermal displacement, for example, the Li and Ni (intermixed on Li layer) on 3a sites have the same displacement parameter; while Li (intermixed on TM layer), Ni, Mn, and Co on 3b sites will share the same value. For Li/Ni intermixing ratio, we refine Ni occupancy on Li layer (3a site), and the Li occupancy on the TM layer (3b site) will change accordingly. The thermal displacement parameters for Li will be fixed once they exceeded the theoretical upper limit. Otherwise, we found that the refinement process can hardly converge. Ni/Mn/Co were fixed at the nominal composition obtained from 25˚C during in situ refinement, since the TM ratio didn't change according to our ICP-MS measurements on 25˚C and 250˚C heated samples. Supplementary Table: Supplementary Table 1 Crystallographic information of pristine gravel-and rod-NMCs after simultaneous refinement using SXRD and ND patterns. *Note site fraction is not restricted to stoichiometry.