In this study, we investigate the use of artificial neural networks as a potentially efficient method to determine the rate capability of electrodes for lithium-ion batteries with different porosities. The performance of a lithium-ion battery is, to a large extent, determined by the microstructure (i.e., layer thickness and porosity) of its electrodes. Tailoring the microstructure to a specific application is a crucial process in battery development. However, unravelling the complex correlations between microstructure and rate performance using either experiments or simulations is time-consuming and costly. Our approach provides a swift method for predicting the rate capability of battery electrodes by using machine learning on microstructural images of electrode cross-sections. We train multiple models in order to predict the specific capacity based on the batteries’ microstructure and investigate the decisive parts of the microstructure through the use of explainable artificial intelligence (XAI) methods. Our study shows that even comparably small neural network architectures are capable of providing state-of-the-art prediction results. In addition to this, our XAI studies demonstrate that the models are using understandable human features while ignoring present artefacts.

Using three-dimensional (3D) metal foams as current collectors is considered to be a promising approach to improve the areal specific capacity and meet the demand for increased energy density of lithium-ion batteries. Electrodes with an open-porous metal foam as current collector exhibit a 3D connected electronic network within the active mass, shortening the electron transport pathways and lowering the electrodes’ intrinsic electronic resistance. In this study, NMC622 cathodes using an aluminium foam as current collector with a measured areal capacity of up to 7.6 mAh cm⁻² were investigated. To this end, the infiltrated foams were densified to various thicknesses between 200 µm and 400 µm corresponding to an electrode porosity between 65% and 30%. The microstructural analysis reveals (i) the elimination of shrinking cavities and a decrease in the porosity of the infiltrated active mass, (ii) an improved contact of active mass to the current collector structure and (iii) a pronounced clogging of the surface pores. The electrochemical properties such as capacity and rate capability are correlated to the electrode’s microstructure, demonstrating that densification is necessary to improve active material utilization and volumetric capacity. However, strong densification impairs the rate capability caused by increased pore resistance and hindered electrolyte accessibility.

The significant demand for energy storage systems has spurred innovative designs and extensive research on lithium‐ion batteries (LIBs). To that end, an in‐depth examination of utilized materials and relevant methods in conjunction with comparing electrochemical mechanisms is required. Lithium titanate (LTO) anode materials have received substantial interest in high‐performance LIBs for numerous applications. Nevertheless, LTO is limited due to capacity fading at high rates, especially in the extended potential range of 0.01–3.00 V versus Li⁺/Li, while delivering the theoretical capacity of 293 mAh g⁻¹. This study demonstrates how the performance of the LTO anode can be improved by modifying the manufacturing process. Altering the dry and wet mixing duration and speeds throughout the manufacturing process leads to differences in particle sizes and homogeneity of dispersion and structure. The optimized anode at 5 A g⁻¹ (≈17C) and 10 A g⁻¹ (≈34C) yielded 188 and 153 mAh g⁻¹ and retained 73% and 68% of their initial capacity after 1000 cycles, respectively. The following findings offer valuable information regarding the empirical modifications required during electrode fabrication. Additionally, it sheds light on the potential to produce efficient anodes using commercial LTO powder.

Research on battery materials and systems is broader than ever and a large amount of data is being generated. The data collected is often challenging to correlate or compare because they are not compiled in a structured way since no generally uniform data structure exists yet in the battery field. Besides, manufacturing steps and analysis parameters vary enormously from laboratory scale to industrial scale, from research institutions to consortia. Within the "DigiBatMat project" we have created a digital platform for battery material data and processing data that combines collected material data and knowledge within a defined structure based on taxonomies and ontologies. The structure of the digital platform was developed to cover the Battery LabFactory Braunschweig (BLB) pilot-scale production line for lithium-ion cells while being compatible with other battery production processes, for example at the laboratory bench scale. The platform’s underlying data structure is divided into an “electrode production” and “cell production” taxonomy, providing input modules for the most typical battery material and cell characterization methods. This approach offers a commercially available platform with a defined structure, which still allows flexibility for individual electrode and cell productions through option based input. Due to this structure, we aim to go beyond simple data storage by implementing functionalities that allow users to work with and analyze the data within the digital platform. Beside the functionalities we have implemented so far, which are shown here, we also aim to develop further functionalities that enable the identification of new correlations that are otherwise difficult or even impossible to obtain. A consistent data set is necessary to build up and validate such a digital platform. Hence, we specifically generated “reference” data and tested their management and analysis with the platform. We focus on three key use cases of collecting process data and characterizing electrodes and cells properties along the production process. In one of these key use cases we compare state-of-the-art cathodes with Ni-rich active materials, namely NMC811 and NCA. Data are collected from cells with such cathodes with varying loadings and electrode densities. Characterization of the electrodes includes, among other things, microscope data analysis of the microstructure and electrochemical characterization such as electrochemical impedance spectroscopy and rate capability. By showing how raw data can be imported into the database, processed and analyzed, we aim to demonstrate the possibilities and advantages of our digital platform.