Accurate reconstruction of the conductive networks in lithium-ion battery electrodes is essential for understanding material-structure-property relationships and optimizing manufacturing processes. The resolution of these networks remains a challenge due to the average size of the microstructure features and lack of contrast to the other materials generally present. Here, a digital materials framework for the three-dimensional reconstruction and quantitative analysis of lithium-ion battery cathode microstructure using focused ion beam–scanning electron microscopy (FIB–SEM) tomography with segmentation by machine learning is introduced. Low acceleration voltage image stacks are recorded and compared to identify the optimal voltage. Manual segmentation aided with standard software provides a ground truth that is used to train a deep neural network model to identify the phases of a cathode material. The model is applied for robust identification and reconstruction of conductive phases. Quantified microstructural descriptors are extracted from the automatic segmentation and systematically linked to structured processing parameters within a digital platform, DataCharge.io, enabling consistent comparison across lab- and pilot-scale electrodes. This integrated data-driven approach facilitates targeted optimization of electrode fabrication, supports transferable process–structure correlations, and advances digitalization strategies for battery materials engineering.

This study aims to employ a machine learning (ML)-based regression model that accurately captures nonlinear relationships between electroplating process parameters and chromium thickness, while enabling the interpretation and visualization of these nonlinear effects. For this purpose, two statistically distinct datasets from laboratory-scale (1L) and pilot-scale (14L) experiments were analyzed. Hyperparameter tuning and fivefold cross-validation are used for training to make the results robust and transparent for different data constellations. Several models were evaluated and achieved coefficients of determination ( ) of up to 75%, often outperforming linear regression (LR) due to nonlinear parameter interactions. Model performance varied depending on the dataset, with no single ML approach proving universally superior. For the entire dataset, the CatBoost model achieved the best result with an average of 70.5%. A deeper analysis of the data was performed using SHAP, permutation importance, and global feature importance. To visualize potential nonlinear relationships, a partial dependence analysis was conducted to assess the influence of individual process parameters. This analysis confirmed the presence of specific nonlinear dependencies for several parameters in relation to chromium thickness. The correlations among the process parameters identified in this study are highly relevant for industrial bath management and process optimization.

Rubber compounds are produced in a discontinuously working internal mixer with the addition of components such as rubber, fillers, plasticizers, and chemicals. In order to ensure a high quality, mixing instructions must be developed for each recipe. Optimal process settings are currently determined iteratively based on expert knowledge. As the scale-up from lab to production machines is not possible today, these process developments take place on production machines, which is limited in terms of time, costs, and material. Therefore, the aim is to predict and subsequently optimize the mixing process depending on the recipe and process parameters. In order to develop initial approaches for describing the entire mixing process, mathematical models are being developed using symbolic regression to describe the ram path and torque as well as the mixing quality based on incorporation, dispersion, and distribution. Overall, sufficient accuracy can be achieved for predicting the ram path and torque. The following optimization using evolutionary algorithms enables a reduction in energy input up to 40%.

Refrigeration based on the magnetocaloric effect (MCE) can contribute to energy-saving, environmentally friendly cooling in private households, or industrial application. The cooling is based on the reversible heat release or uptake during a phase-transformation of the materials that can be controlled by a magnetic field. This process could replace conventional compression-based refrigeration, which often relies on environmentally harmful refrigerants. The MCE is observed in a large number of magnetic alloys, upon them are Heusler alloys. Thus, optimization of MCE materials involves the screening and testing of many different materials systems, generating a large amount of data that, so far, has not been organized in a systematic manner, hindering the progress of the field. Here, were present an approach to digitalize the process chain from synthesis, experiment, and simulation to prototypical applications. Different Heusler alloys have been examined experimentally as model systems for potential applications in magnetic cooling. Templates based on the OTTR technology have been developed and implemented for the acquisition and semantic representation of knowledge in the development of an ontology. The ontology, when combined with unstructured data, can be exploited to train a model that can then be used to predict missing facts, which can help to gain new insights and to generate new hypotheses. Furthermore, tools have been developed that automate and accelerate data acquisition into ontological structures, and workflows have been implemented that provide a fast, easy-to-use theoretical and experimental evaluation of the MCE from first principles and raw data.

Knowledge about the behavior of complex systems and its prediction by modelling is a crucial point in many problems in science and technology. Design of experiments (DoE) is a well-established tool for investigation of the parameter space and development of appropriate models of such systems. Although DoE reduces the number of required experiments enormously, the remaining number of experiments can be still unfeasible in many real-world applications, i.e., when a single experiment is too time-consuming or expensive. This paper proposes a workflow for investigating hydride vapor phase epitaxy (HVPE) for growing of GaN bulk crystals. The suggested workflow is based on DoE and physical simulations of mass transport processes and crystal growth kinetics as an intermediate step between DoE and experiments. Considering the high complexity of real HVPE systems, this study is first focused on the impact of the individual gas flows in the reactor, their compositions, inlet velocities and relations on the growth rate. Possible correlations in the simulated data are investigated using a phenomenological metamodel describing the response of the system on the variation of process parameters and discussed in detail. The presented results can be considered as a step towards a digital twin of the HVPE growth process and can be used for the improvement of crystal quality and the process efficiency in terms of precursor, energy and time consumption.