Periodic Reporting for period 3 - SINTBAT (Silicon based materials and new processing technologies for improved lithium-ion batteries)
Berichtszeitraum: 2019-03-01 bis 2020-02-29
A characterization toolbox was developed to analyse the lithiation and ageing mechanisms of Si-based anode material. Microscopy results showed that the ageing creates a core-shell structure on the long-term, the core remaining formed by nanoscale domains embedded in an amorphous Si-phase, and the star-shaped shell being surrounded by SEI. Synchrotron tomography was used to study the evolution of the Si-compound particles and their impact on the proximate pore network on m-scales, and structural analysis of aged Si-based electrodes using a combination of x-rays/neutron scattering techniques was developed. Combined operando SAXS/WAXS synchrotron experiments allowed to obtain the real-time evolution of graphite and silicon phases during cycling. The sequential contribution of both phases to the total capacity during lithiation and delithiation was quantified, and the effects of cycling rates, ageing and pre-lithiation were monitored. The combination of scattering and 2D/3D imaging techniques with modelling allowed to identify heterogeneities in lithiation at the level of the active Si-particles, and to establish the correlation between pore network ageing and Li-Ion diffusion rate restrictions through the whole electrode after long-term cycling. Moreover, insights into the role of the nanoscale crystalline “buffer” particles were established by NMR, showing the importance of the interfacial regions with amorphous silicon network, which help in protecting the core silicon and homogenizing Li-diffusion. A key finding is that the nanoscale design of the active silicon phase inhibits pulverization and limits capacity fading.
Energy storage performance of full cells containing Si-Graphite anode combined with LiNi1/3Mn1/3Co1/3O2 (NMC111) cathode has been studied. Electrolytes with different degrees of fluorination and the impact of lithium salt and additives have been investigated. The highly fluorinated electrolyte formed a fluorine-rich SEI layer and featured high capacity retention at high current densities compared with other electrolytes. In contrast, the fluorine-free electrolyte formed a stable oxygen-rich SEI layer on the anode’s surface and showed good electrochemical performance at low current, but the cycling stability was limited at higher currents. Although a fluorinated electrolyte was ultimately preferred within the project, the possibility to obtain stable SEI layers on silicon-based anodes with a fluorine-free electrolyte was shown. This indicates that the latter are potential candidates for more sustainable and less toxic high energy batteries.
Various electrochemical methods were adopted to evaluate Li-ion diffusion, the resistivity of the SEI layer and parameters related to electron transfer kinetics. Contrary to the literature, the independence of charge transfer resistance value from a current direction at low C-rate tests, and analysis of potential changes during cycling, confirmed single-phase mechanism on both: lithiation and delithiation of silicon alloy. New, modified EIS technique combined with DC current flow was used to estimate changes in the electrode real electroactive surface. Almost 60% increase in surface area due to silicon alloy swelling was detected. Parameters such as exchange current density, charge transfer coefficient and lithiation/delithiation reaction symmetry factor were determined and used in electro-mechanical modelling and capacity fade of the cell behaviour.
A probabilistic regression approach was employed to predict capacity fade and resistance increase of a Sintbat battery consisting of 75% silicon composite and 25% graphite. The capacity loss exhibited a linear fade with cycle number, compared to a non-linear resistance increase. The linear capacity fade favoured the covariance functions utilised in the probabilistic regression which lead to the improved forecasting. The modelling approach, therefore, highlighted that the prediction performance is highly determined by the adopted covariance function. To predict non-linear resistance trends, new methods are required whereby the covariance function is motivated by the ageing physics of a battery and can, thus, lead to physically informed data-driven models in the future.
A multiscale model based on mean-field homogenisation enabled us to investigate effects of stress on the kinetics of two-phase lithiation process within Si-particles at the microscale and link it to the macroscale via relevant parameters defining anode material compositions and material properties of anode material components. This included a theory of mechanochemistry based on the chemical affinity tensor, chemo-mechanical coupling, non-linear constitutive modelling of anode material behaviour, model parametrization with available experiments, and numerical implementation within a non-linear finite-element framework. The entire homogenisation framework can be used to provide useful insight into structure-property relations in the anode during battery performance and guide material engineers to design optimum material compositions for the anode.
In the end, a module with Sintbat cells and the developed BMS was assembled and successfully tested in a standard VARTA energy storage system.