Chemical Imaging of Cathode Materials - The effect of W on the inner workings of a LiNiO2 Cathode - Data analysis procedures using Python for XRD-CT/XRF-CT

Oct 17, 2026

Uploaded by SANS

This project develops Python-based data analysis procedures for X-ray diffraction/fluorescence CT (XRD-CT/XRF-CT) datasets from cathode materials to enable fast, reliable image processing. The workflow selects reconstruction algorithms to build images of battery component distributions from experimental projections, then selects data clustering methods to segment reconstructed images into distinct components. The procedure is applied to LNO and W-doped LNO (LNO-W) charged to 4.7 V to quantify how W doping affects structural stability.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages1

Reading Time3 min

AccessFree to download

Abstract

Motivation

Methods

Simulated sinogram and reconstruction benchmarking

Clustering algorithm selection

Rietveld Refinement benchmarking

NMF for Phase Identification

NMF for Phase Mapping

Rietveld Refinement for c-axis Mapping

Impact / Next steps

References / Acknowledgment

Chemical Imaging of Cathode Materials The effect of W on the inner workings of a LiNiO2 Cathode Kenan Gnonhoue Dokonon Supervisors: Dr. Sam Booth*, Dr. Steve Price** , Dr. Antony Vamvakeros** *The University of Sheffield,**Finden Ltd Abstract This project aims to develop data analysis procedures for X-ray diffraction/fluorescence CT (XRD-CT/XRF-CT) data from cathode materials using Python as to quickly and reliably perform data processing to facilitate analysis. The processing steps include: ▪ Selecting the optimal algorithm to construct an image of the different components of the battery using the projections obtained from the experiment. ▪ Selecting the ideal data clustering technique to segment thereconstructed image into the different components of the battery. The developed data analysis procedure was then applied to datasets obtained from LNO (LiNiO 2 ) and LNO-W (W doped LNO) based cathode materials charged to 4 .7 V to analyse the effect of W doping on the structural stability of such materials. Motivation ▪ Reduce the time spent on data processing. ▪ Quickly generate accurate images of the distribution of the different phases in the cathode. ▪ Apply the developed data processing procedure to map the distribution of W in LNO cathodes to better understand their structural changes. ▪ Deepen our understanding of LNO cathodes as to one-day replace unsustainable LCO (LiCoO 2 ) cathodes. Methods 1. A simulated sinogram of the Shepp-Logan image was generated where the effect of Poisson-Noise and reduced number of projections were tested to benchmark the accuracy of each reconstruction algorithm. 2. Several clustering algorithms were run on a simulated dataset of XRD-CT and that best able to deconvolute the signal into the constituent chemical elements was selected. 3. Rietveld Refinement was then used as a reference to quantify the performance of the selected clustering algorithm on areal dataset. | NMF for Phase Identification

Figure 2. NMF clustering applied to the spectral domains of LNO and LNO-W cathode datasets. For each dataset two XRD patterns were fitted; one for the current collector (identified to be Al) and one for the cathode material (identified to be LNO) [2] . | ▪ The intensity of the H2 peak in LNOW is much greater than that in LNO, signalling W inhibits the formation of the detrimental H3 phase which causes high strain [1],[3] .
▪ The H2 and H3 peaks of the LNO-W spectra superpose, suggesting the
high c-axis strain reduces upon W doping.
▪ Hence, by segregating to the surface, W forms an amorphous 􀢃􀢞 􀡸􀢏 􀢟 􀡻 􀢠 phase, leading to a Ni-rich lattice and an increased c-axis [1] . | | --- | --- | NMF for Phase Mapping 45 μm Figure 3. NMF clustering and Rietveld Refinement applied to the image domain of the LNO cathode dataset. Distribution of a) the H3 phase and c) the H2 phase derived from Rietveld Refinement. Distribution of b) the H3 phase and d) the H2 phase derived from NMF clustering. ▪ ▪ Intens ity (a. .u ) Figure 3a) and b) are similar, suggesting like Rietveld Refinement, NMF can precisely locate the H3 phase as it is the major component of the signal. Figure 3c) and d) show it is more difficult for NMF to detect the H2 phase as it is the minor phase. | Rietveld Refinement for c-axis Mapping | | | --- | --- | |
Figure 4. c-axis distribution derived from Rietveld Refinement for a) the LNO and b) the LNO-W datasets. | ▪ Using the c-axis length as an indicator of the proportion of each phase, figure 4 shows there is more H2 in LNO-W compared to LNO.
▪ Figure 4 indicates the H2 phase segregates towards the electrolyte on the right hand side in LNO, whereas the
H2 and H3 phases distribute more uniformly in LNO-W.
▪ Given the H2 phase is concentrated atthe electrolyte, this suggests the cathode was not fully delithiated. | Impact / Next steps ▪ We were able to demonstrate the power of SIRT and NMFin generating quick image

FAQs

What is the main goal of the project described in the document?

To develop Python data analysis procedures for XRD-CT/XRF-CT cathode datasets that enable quick and reliable image processing and segmentation of battery components.

How does the method identify and map phases in LNO and LNO-W?

It selects an optimal reconstruction algorithm for image construction, then applies an appropriate clustering technique to segment the reconstructed image into constituent components, with phase mapping interpreted against Rietveld Refinement.

What impact does W doping have according to the results and discussion?

The document reports that W inhibits detrimental phase formation (including effects on H3) and leads to structural and c-axis changes, such as increased c-axis parameters and altered phase distribution compared with undoped LNO.

Oct 17, 2026

Uploaded by SANS