Operando characterisation techniques for graphene-based lithium-sulfur batteries

Lithium-sulfur (Li-S) batteries have the potential to enable high-gravimetric capacity, low-cost energy storage, but their widespread implementation is presently limited by their rapid capacity degradation due to irreversible loss of electrical connectivity with their active material. Electrically conductive cathodic host matrices, with a porous structure tuned to retain the soluble sulfur and polysulfides (PS) formed during cycling, are an active research area, with graphene-related materials (GRMs) of particular interest due to their compatibility with high-throughput manufacturing processes. This thesis presents a meta-analysis of recent literature on Li-S host matrices, for which a graphical user interface was developed to collate reported parameters and directly link them to attained cell capacities. This meta-analysis highlights the wide array of approaches and parameter combinations reported in the literature, and the corresponding lack of convergence in cell performance to date. A novel method of synthesising sulfur/ carbon composites is presented, entailing co-high pressure homogenization (co-HPH) of graphite and sulfur, and is evaluated against a conventional thermal-infiltration synthesis method. A new technique for analysing galvanostatic cycling data is employed to compare their electrochemical performance, suggesting the more homogenous novel composites undergo gradual dissolution of sulfur over multiple cycles compared to the dissolution of sulfur within ca. 20 cycles in thermally-infiltrated composites. This thesis also reports an attempt to characterise the morphology of these composites using X-ray tomography, with the key outcome being a novel method for labelling and segmenting heterogeneous tomography data using Isomap-based dimensionality reduction. Spatially-resolved operando Raman spectroscopy was applied to acquire 2d maps of the sulfur and PS in electrodes undergoing cycling, and monitor the (re-)distribution of active material in electrodes prepared via different synthesis pathways and cycled using different electrochemical parameters. This was facilitated by the development of a data-processing pipeline, incorporating dimensional reduction via principal component analysis and de-noising using K-nearest neighbours classification, and is demonstrated in a series of case studies to elucidate reasons for inter-sample variability. The data-processing pipelines developed here can be readily adapted to other heterogenous, spatially and/ or temporally resolved datasets with low signal-to-noise ratio to enable real-time data visualisation for experimental diagnostics and analysis.

Name dc:type.qualificationname

Doctor of Philosophy (PhD)

Level dc:type.qualificationlevel

Doctoral

Grantor dc:publisher.institution

University of Cambridge

Year dc:date.issued

2024

Author dc:creator

• Bird, Lauren

Advisors dc:contributor.advisor

• Ducati, Caterina
• Ferrari, Andrea
• Kumar, Ramachandran

• Battery
• Graphene
• Lithium-sulfur
• Meta-analysis
• Principal component analysis
• Raman
• Synchrotron X-ray CT

Licence

• https://apollo8-f-pro.lib.cam.ac.uk/bitstreams/75f8e39f-45d1-4025-af74-07907ed450e5/download
• https://creativecommons.org/licenses/by/4.0/

Language dc:language

eng

Author Identifier

0000-0002-5310-946X

OAI identifier oai:identifier

oai:www.repository.cam.ac.uk:1810/373000