Presentation Information
[P1-69]Development of data handling tools for high-throughput experiments
*Pierre Le Berre1, William Rigaut1, Wilfried Hortschitz2, Santa Pile2, Harald Oezelt2, Samuel J. R. Holt3,4, Swapneel A. Pathak3,4, Hans Fangohr3,4,5, Thomas Schrefl2, Thibaut Devillers1, Nora M. Dempsey1 (1. Univ. Grenoble Alpes, CNRS, Grenoble INP, Institut Néel, 38000 Grenoble (France), 2. Department for Integrated Sensor Systems, University for Continuing Education Krems, Wr. Neustadt (Austria), 3. Max Planck Institute for the Structure and Dynamics of Matter, Luruper Chaussee 149, 22761 Hamburg (Germany), 4. Center for Free-Electron Laser Science, Luruper Chaussee 149, 22761 Hamburg (Germany), 5. Faculty of Engineering and Physical Sciences, University of Southampton, Southampton SO17 1BJ (UK))
Keywords:
high-throughput,data,thin films,combinatorial
Over the last decades, the combinatorial approach to the exploration of material systems has garnered great interest as a novel way to efficiently optimize functional materials [1]. High-throughput characterization techniques can quickly generate huge amounts of data which needs to be sorted, analyzed and stored, both for human exploitation and for use as input in machine learning models. At Institut Néel, a thin-film based combinatorial approach is being developed to study the effects of elemental substitution and processing conditions on different hard magnetic material systems. Specific software tools have been developed to handle the data sets generated by Energy Dispersive X-Ray (EDX) spectroscopy, mechanical profilometry, X-Ray Diffraction (XRD), Magneto-Optic Kerr effect (MOKE) magnetometry and Scanning Electron Microscopy. To deal with the data, we have created an HDF5 based file system which is particularly suited to handle multi-dimensional datasets, such as those found in combinatorial material science. These files ensure generated data is traceable, convenient to store and easy to share. The file system is integrated with the Magnetic Material Ontology [2] and aggregates within a single file all relevant metadata, raw data and analysis results from a battery of different instruments and measurements. These files also serve as a digital log, recording all the stages of a sample's life with all the necessary information, including user notes. To generate and interface with the HDF5 files, a software suite has been developed, based on Python and Plotly-Dash [3]. The software allows for quick visualization of complex datasets, so that users can evaluate the quality of generated datasets, and it also allows for the customizable export of datasets in a format that can then be used as input in machine learning models. All of this work is open-source and can be expanded on for use with different experimental setups, in an effort to push towards traceable, shareable data.
[1] ML Green et al., J. Appl. Phys. 113 (2013) 231101
[2] https://mammos-project.github.io/MagneticMaterialsOntology/doc/magnetic_material_mammos.html
[3] https://plotly.com/
Acknowledgements: This work is being carried out within the framework of the EU funded MaMMoS project (Grant number 101135546, HORIZON-CL4-2023-DIGITAL-EMERGING-01), the ANR-FWF funded DATAMAG project (ANR-22-CE91-0008 / FWF I 6159-N) and the ANR/DIADEM MIAM project (ANR-23-PEXD-0013).
[1] ML Green et al., J. Appl. Phys. 113 (2013) 231101
[2] https://mammos-project.github.io/MagneticMaterialsOntology/doc/magnetic_material_mammos.html
[3] https://plotly.com/
Acknowledgements: This work is being carried out within the framework of the EU funded MaMMoS project (Grant number 101135546, HORIZON-CL4-2023-DIGITAL-EMERGING-01), the ANR-FWF funded DATAMAG project (ANR-22-CE91-0008 / FWF I 6159-N) and the ANR/DIADEM MIAM project (ANR-23-PEXD-0013).