76 77 D I A M O N D L I G H T S O U R C E A N N U A L R E V I E W 2 0 2 0 / 2 1 D I A M O N D L I G H T S O U R C E A N N U A L R E V I E W 2 0 2 0 / 2 1 Characterising defects in lithium-ion battery particles Related publication: HeenanT. M. M.,Wade A.,Tan C., Parker J. E., Matras D., Leach A. S., Robinson J. B., Llewellyn A., Dimitrijevic A., Jervis R., Quinn P. D., Brett D. J. L. & Shearing P. R. Identifying the Origins of Microstructural Defects Such as Crackingwithin Ni-Rich NMC811 Cathode Particles for Lithium-Ion Batteries. Adv. EnergyMater. 10 , 2002655 (2020). DOI: 10.1002/aenm.202002655 Publication keywords: Batteries; Cathodes; Degradation; Electric vehicles; Microstructure; NMC811; Particle cracking B attery electrodes are composed of many millions of particles. Lithium-ions entering and leaving these particles charge and discharge the battery. Defects in the particles reduce the battery’s capacity. This would impact electric vehicle range in automotive applications, for example, requiringdrivers to stopmore frequently to recharge.Therefore, it is crucial to investigatewhichparticles contain defects, whether different types of defects exist, and howwe might prevent them forming. Electrode particles are spheres with diameters on the order of tens of microns (smaller than a human hair). Due to their small size, investigating their properties requires high-resolution specialist facilities, such as those at the Hard X-Ray Nanoprobe beamline (I14) at Diamond Light Source. The defects have complex origins, requiring the use of multiple investigation techniques at once, for which I14 was specifically designed. Batteries that are nickel-based (around 80%), with manganese and cobalt making up the other 20%, are of particular interest to the automotive industry. By analysing various particles at I14 and comparing the results with data collected in the labs at University College London (UCL), researchers found a correlation between the manganese content and the ordering of the particle’s crystals. However, they attributed the amount of cracking within the particle to the crystal orientations. Preventing the manganese from leaving the particles may maintain the crystal ordering. Aligning the crystals, or making larger single crystals, may avoid - or at least delay - particle cracking. Overcoming these issues would allow electric vehicles to travel for longer before needing to recharge. The automotive sector is being rapidly electrified in an effort to improve air quality and combat global warming. However to achieve this, affordable battery electric vehicles (BEVs) are required that can drive a sufficient distance on a single charge and then recharge quickly, all while maintaining a suitable lifetime. The choice of electrode chemistry and microstructure strongly influence a BEV’s ability to meet these demands and consequently, materials developments have accelerated in recent years. Layered oxide cathodes such as LiNi x Mn y Co z O 2 (NMC) are promising candidates for the next-generation of BEVs. Particularly, Ni-rich variations (e.g. LiNi 0.8 Mn 0.1 Co 0.1 O 2 or NMC811) are of interest because of the toxicity, expense and supply-chain issues associated with Co. Moreover, increasing the Ni content also promotes high electronic conductivities and fast Li + diffusivities, as well as increased volumetric and gravimetric energy densities. However, there is a significant disparity between the practical and theoretical performances of these materials due to complex degradation mechanisms. Furthermore, large voltage windows are employedwith high upper cut-offs (e.g. >4.5V vs. graphite) in order to maximise the accessible capacity, which accelerates degradation. For instance, Ni-rich NMC is known to suffer from: lower thermal runaway temperatures; metal dissolution; surface species; accelerated lattice collapse; oxygen evolution; and surface rock-salt formation. There are many outstanding scientific questions relating to the performance loss of Ni-rich NMC in long term cycling.Therefore this work aimed to correlate several characterisation methods in order to determine cause and effect relationships for such degradation 1 . Commercial polycrystalline NMC811 materials generally consist of broad size distributions (e.g. 3 – 30 μm) of highly spherical particles (e.g. sphericity > 0.7) that are labelled ‘secondary’ and are composed of many ‘primary’ particles that are around an order of magnitude smaller in size (e.g. 0.1 – 1 μm). The primary particles are mostly single crystals and are often assembled randomly into the secondary particle agglomerate, resulting in the polycrystalline definition. The secondary particles are then made into a slurry with a carbon and binder domain (CBD), to be printed and calendared onto an aluminiumcurrent collector. See Fig. 1 for a visual representation of this material of interest. For this work, commercial NMC811 electrode material was initially examined in the pristine state using lab-based nano X-ray computed tomography (CT) at UCL. This revealed that even before operation around one-third of the secondary particlesexhibitedsomeformofdefect,especiallytowardstheelectrode-separator interface where calendaring affects duringmanufacturing are thought to produce localised pressure maxima which are dissipated by the cracking of particles. Cracking, that is particularly evident in the larger secondary particles, reduces the effective particle size distribution, increasing the surface area and the population of possible sites for undesirable side-reactions. To observe such substantial particle cracking before operation is significant if it triggers severe degradation and capacity loss. It is therefore advised that novel and improved manufacturing methods are explored in order tomitigate this issue. To examine how these defects progress with operation, the NMC811 material was assembled into coin cells against graphite anodes and then cycled to various uppercut-offvoltages(4.2–4.5V).Usinglab-basedmicroX-rayCTmicrostructural cracking defects were observed to deteriorate throughout the entire electrode volume after high voltage operation, thought to be triggered by the rapid contraction of the c lattice parameter above ca. 60 % State of Charge (SoC) - the level of charge of an electric battery relative to its capacity.Through multi-length- scale correlation a novel defect algorithm was developed that is capable of detecting defects within low resolution tomograms (three dimensional datasets), consequently improving materials statistics substantially, and allowing an order of magnitude more secondary particles to be examined at once. Developments such as this lend themselves well to advanced data processing such as machine learning and other artificial intelligence and as such, endeavours should be made to release such data openly for development and analysis by the wider computing andmodelling community 2 . To expand the investigation into crystallographic and chemical analysis required synchrotron radiation and the world-class facilities of beamline I14 at Diamond. Focusing on the high voltage (4.5 V) material, individual secondary particles ranging from2–8 μm in diameter were extracted fromthe bulkmaterial and imaged in UCL’s lab-based nano-CT before being transferred to the beamline. At I14 the Ni:Mn:Co content ratio could be studied via X-ray fluorescence (XRF) analysis (Fig. 2). It became evident that the amount of Mn within the secondary particles varied considerably, with some regions as low as 4%. Moreover, Mn-rich regions (detected using XRF) had formed not as coatings on the particles but as distinct surface clusters and, when correlated to the lab-based nano X-ray CT, were found to be of significantly lower density than the NMC. The importance of Mn- mobility during early-stage cycling has two key implications. Firstly, the correlative data confirms that the surface cluster is not the NMC that is desired during manufacturing and therefore will not contribute to the cell capacity as intended, occupyingpreciousvolumethatcouldotherwisebefilledbyusefulNMC.Secondly, if the Mn cluster is not inert/inactive in its current state, it may travel to the anode and deposit on the surface, impeding Li+ transport pathways. For future work it is therefore of interest tomonitor theMnmobility during early cycling inmore detail and particularly possibility of its presence on the anode 3 . Due to the specialist set-up at I14, X-ray diffraction (XRD) data could be collected simultaneously to the XRF data, providing information of the particle crystal structure. Correlating the XRF and XRD data revealed a greater disordering in the particles that had lost moreMn. It is thought that the presence of Mnwithin theNMCstructurealleviatesJahn-TellerdistortionbythedonationofaMnelectron at low voltages; if Mn is leaving the structure during early cycling, distortions may be reintroduced.The secondary particles were then examined using the lab-based nano-CT data to explore if there was a correlation between the crystal disordering (and Mn loss) with the degree of secondary particle cracking. Interestingly, there was no clear correlation at this stage suggesting that the c lattice contraction and the random orientation of the crystals likely dominates the secondary particle cracking at low cycle numbers but undesirable crystal strain accumulates in the primary particles due to Mn loss that is possibly not yet visible at the secondary particle scale. High resolution lab-based nano CT of single secondary particles at UCL strengthened this conclusion by the observation of cracking produces similar structures within NMC811 as within lower Ni content NMC (e.g. NMC111) 4 . Consequently, it was concluded that single crystal alternatives may provide some alleviation for cracking induced by lattice contraction but accumulative crystallographic strain, particularly in the presence of Mn loss, could remain problematic 5 . In this work we explore the various mechanisms that may lead to particle defects that are responsible for capacity loss and ultimately, the deterioration of BEV performance. Many of these findings were only possible through correlative characterisation methods requiring the specialist facilities such as those at I14. Multi-modal imaging such as this is highly powerful and will likely continue to advancemanyaspectsofmaterialsdevelopmentsforyearstocome,notonly inthe field of Li-ion batteries but throughout various fields of science and engineering. References: 1. LiW. et al. High-nickel layered oxide cathodes for lithium-based automotive batteries. Nat. Energy 5 , 26–34 (2020). DOI: 10.1038/s41560-019-0513-0 2. HeenanT. M. M. et al. Data for an AdvancedMicrostructural and Electrochemical Datasheet on 18650 Li-ion Batteries with Nickel-Rich NMC811 Cathodes and Graphite-Silicon Anodes. Data Br. 32 , 106033 (2020). DOI: 10.1016/j.dib.2020.106033 3. DoseW. M. et al. Effect of Anode Slippage on Cathode Cutoff Potential and DegradationMechanisms in Ni-Rich Li-Ion Batteries. Cell Reports Phys. Sci. 1 , 100253 (2020). DOI: 10.1016/j.xcrp.2020.100253 4. Tsai P. C. et al. Single-particlemeasurements of electrochemical kinetics in NMC and NCA cathodes for Li-ion batteries. Energy Environ. Sci. 11 , 860–871 (2018). DOI: 10.1039/c8ee00001h 5. LiuY. et al. Microstructural Observations of“Single Crystal”Positive Electrode Materials Before and After LongTermCycling by Cross-section Scanning ElectronMicroscopy. J. Electrochem. Soc. 167 , 020512 (2020). DOI: 10.1149/1945-7111/ab6288 Funding acknowledgement: This work was carried out with funding from the Faraday Institution (faraday. ac.uk ; EP/S003053/1), grant number FIRG001 and FIRG003; and the EPSRC grant EP/M014045/1.The authors acknowledge Diamond Light Source for time on beamline I14 under proposals sp20841 andmg23858.The authors would like to acknowledge the Royal Academy of Engineering (CiET1718\59) for financial support. Corresponding authors: DrThomas Heenan, University College London/Faraday Institution,
[email protected] ; Prof. Paul Shearing, University College London/Faraday Institution,
[email protected] Imaging andMicroscopy Group Beamline I14 Figure 1: Microstructure of NMC811. (a) Diagrams of the NMC structure; (b) The number of defected secondary NMC particles with respect to electrode depth; (c) Visual aid for the various types of secondary particle defects; and (d) The average defect composition for a commercial NMC material. Figure 2: Multi-modal imaging for the correlation of X-ray absorption, fluorescence and diffraction data. (a) Aligned Mn fluorescence and (all material) attenuation maps of a single NMC particle, with a magnified section of Mn-rich surface cluster; (b) Correlated diffraction and fluorescence data for six individual NMC particles, displaying the r factor and minimum Mn content; and c) An absorption nano-CT cross-sectional ortho-slice of an NMC particle. All particles reported within this figure were cycled to 4.5 V for 5 cycles, before imaging.