Machine Learning-Based Lifetime Prediction of Lithium-Ion Cells.

Precise lifetime predictions for lithium-ion cells are crucial for efficient battery development and thus enable profitable electric vehicles and a sustainable transformation towards zero-emission mobility. However, limitations remain due to the complex degradation of lithium-ion cells, strongly influenced by cell design as well as operating and storage conditions. To overcome them, a machine learning framework is developed based on symbolic regression via genetic programming. This evolutionary algorithm is capable of inferring physically interpretable models from cell aging data without requiring domain knowledge. This novel approach is compared against established approaches in case studies, which represent common tasks of lifetime prediction based on cycle and calendar aging data of 104 automotive lithium-ion pouch-cells. On average, predictive accuracy for extrapolations over storage time and energy throughput is increased by 38% and 13%, respectively. For predictions over other stress factors, error reductions of up to 77% are achieved. Furthermore, the evolutionary generated aging models meet requirements regarding applicability, generalizability, and interpretability. This highlights the potential of evolutionary algorithms to enhance cell aging predictions as well as insights.

Analysis of nuclear actin by overexpression of wild-type and actin mutant proteins.

Compared to the cytoplasmic F-actin abundance in cells, nuclear F-actin levels are generally quite low. However, nuclear actin is present in certain cell types including oocytes and under certain cellular conditions including stress or serum stimulation. Currently, the architecture and polymerization status of nuclear actin networks has not been analyzed in great detail. In this study, we investigated the architecture and functions of such nuclear actin networks. We generated nuclear actin polymers by overexpression of actin proteins fused to a nuclear localization signal (NLS). Raising nuclear abundance of a NLS wild-type actin, we observed phalloidin- and LifeAct-positive actin bundles forming a nuclear cytoskeletal network consisting of curved F-actin. In contrast, a polymer-stabilizing actin mutant (NLS-G15S-actin) deficient in interacting with the actin-binding protein cofilin generated a nuclear actin network reminiscent of straight stress fiber-like microfilaments in the cytoplasm. We provide a first electron microscopic description of such nuclear actin polymers suggesting bundling of actin filaments. Employing different cell types from various species including neurons, we show that the morphology of and potential to generate nuclear actin are conserved. Finally, we demonstrate that nuclear actin affects cell function including morphology, serum response factor-mediated gene expression, and herpes simplex virus infection. Our data suggest that actin is able to form filamentous structures inside the nucleus, which share architectural and functional similarities with the cytoplasmic F-actin.

Preparation and Characterization of Li-Ion Graphite Anodes Using Synchrotron Tomography.

We present an approach for multi-layer preparation to perform microstructure analysis of a Li-ion cell anode active material using synchrotron tomography. All necessary steps, from the disassembly of differently-housed cells (pouch and cylindrical), via selection of interesting layer regions, to the separation of the graphite-compound and current collector, are described in detail. The proposed stacking method improves the efficiency of synchrotron tomography by measuring up to ten layers in parallel, without the loss of image resolution nor quality, resulting in a maximization of acquired data. Additionally, we perform an analysis of the obtained 3D volumes by calculating microstructural characteristics, like porosity, tortuosity and specific surface area. Due to a large amount of measurable layers within one stacked sample, differences between aged and pristine material (e.g., significant differences in tortuosity and specific surface area, while porosity remains constant), as well as the homogeneity of the material within one cell could be recognized.