Demonstrating faster multi-label grey-level analysis for crack detection in ex-situ & operando micro-CT images of NMC electrode

During charge/discharge cracking occurs in Nickel Manganese Cobalt (NMC) secondary particles. Secondary particles with cracks will have observably lower grey-levels in micro-CT datasets compared to an otherwise identical pristine particle. This is due to the ‘partial volume effect’ where voxels representative of space containing both void and solid phases result in grey-levels that are intermediate between voxels of void or solid only. In this work, we present a method for automatically tracking changes in grey-level due to this effect in large statistically relevant micro CT datasets of 10,000+ discrete particle instances. This work extends previous work from our group where the GREAT algorithm was used to track the grey-level change in tomography images of NMC particles. This study processed datasets of hundreds of particles with different electrochemical histories and was capable of processing ca.1400 particles per day. In this work, we develop the GREAT2 algorithm which is capable of processing 10,000 similar particles in under a minute. This was achieved with an automated particle tracking method allowing the same particle to be tracked through different states of charge in an operando experiment. Additionally, we demonstrate methods for processing the data in-order to extract useful insights. This method can expedite tomographic analysis of electrode particle cracking from upto 100 particles per day with nano-CT to 10,000+ particles per day with micro-CT (or much more at synchrotron micro-CT beamlines). This may benefit both academic and commercial research. These methods and the GREAT2 algorithm have been packaged into the GRAPES python toolkit and GUI for open source distribution.