FINALES Data

The optimisations campaigns utilizing the latest version of FINALES (Fast INteNtion Agnostic LEarning Server) is designed to determine conductivity values and predict end of life for various electrolyte formulations containing the aforementioned chemicals. The first campaign optimising for conductivity is refered to as the singletask phase, due to the single property nature of the optimisation. The second campaign was deployed to optimise for conductivity and EOL (End-of-life) simultaneously and is therefore refered to as the multitask phase.

Tenants

Overall the two campaigns comprises the following tenants (computational/experimental/physical capabilities) grouped by their type of service:

Physical tasks:

  • ASAB: This tenant automatically formulates requested electrolytes starting from stock solutions provided to it. It can also measure the ionic conductivity of the prepared formulation using a symmetric two-electrode cell using two stainless steel electrodes.
  • AutoBASS: This tenant automatically assembles coin cells using electrolyte formulations provided to it.
  • Cycling: This tenant performs the cycling of cells provided to it. It additionally provides a service to reserve cycling channels for cells prior to requesting their assembly to ensure channel availability and proper wetting times.

Computational tasks:

  • 3DS: This tenant performs molecular dynamics simulations and outputs the conductivity. Optional further outputs are the density and the radial distribution function for the requested formulation.
  • DegradationModel: This tenant predicts the end of life for a given capacity trajectory. It uses a threshold of 80 % of the initial capacity after three formation cycles as the definition for the end of life.
  • Overlort: This tenant manages the workflow of formulating electrolytes, assembling cells, cycling cells and predicting the end of life. Based on a request for a prediction of end of life or any other process step in between, it creates the necessary sub-requests to fulfil the task and merges them into one result which is subsequently reported to FINALES.
  • OCond: This tenant follows a Gaussian process for optimizing the electrolyte formulation for maximum conductivity. Based on the currently available data in the FINALES database, it suggests the most promising follow-up formulation targeted towards the maximization of the ionic conductivity.
  • OEOL: This tenant follows a gaussian process for optimizing the electrolyte formulation for maximum end of life of the full cells. Based on the currently available data in the FINALES database, it suggests the most promising follow-up formulation targeted towards the maximization of the end of life. Service tasks:
  • Transportation: This tenant allows for requests for transportation of physical samples. This is used for the transportation of electrolytes and battery cells between the devices connected to the tenants performing the physical tasks.

Formats

Description of the file formats used in the zip files beyond the convetional .png, .csv, .json and .md formats.

Formats in "FINALES (09_2023) Electrolyte Optimization for Maximum Conductivity and for Maximum Cycle Life.zip" and "FINALES (11_2023) Electrolyte Optimization for Maximum Conductivity and for Maximum Cycle Life.zip"

  • db: The binary database file sqlite.db is a dump of the entire database for the FINALES instance pulled directly from an API endpoint. Consult the README in the -zip file for further details.
  • jsonld: Linked Data format using standard json context. Can be uploaded directly into a graph database. Consult the README in the -zip file for further details.

Formats in "FINALES (09_2023 to 12_2023) Electrolyte Optimization for Maximum Conductivity and for Maximum Cycle Life (Additional Data).zip"

  • JDX: The JCAMP-DX file format used in JDX files is a common data format to communicate spectral data. These files contain the (x,y) data of the spectra as well as some comments reporting metadata and parameters related to the measurement. The files can be opened using a text editor and are human readable. A description of the file format can be found in Davies, A. N., & Lampen, P. (1993). JCAMP-DX for NMR. Applied Spectroscopy, 47(8), 1093-1099.

  • PAR: The PAR files provide information about the parameters of a measurement. These files can be opened using a text editor.

  • MNOVA: The MNOVA file format is generated by and can be opened using the MestReNova software (©2020 Mestrelab Research S.L.). These files contain data regarding NMR spectra, their processing and analysis.

Acknowledgements

This project received funding from the European Union’s Horizon 2020 research and innovation program under grant agreement no. 957189 (BIG-MAP). The authors acknowledge BATTERY2030PLUS, funded by the European Union’s Horizon 2020 research and innovation program under grant agreement no. 957213. This work contributes to the research performed at CELEST (Center for Electrochemical Energy Storage Ulm-Karlsruhe) and was co-funded by the German Research Foundation (DFG) under Project ID 390874152 (POLiS Cluster of Excellence). HSS acknowledges funding from DFG EXC 2089/1-390776260 (e-conversion). TV acknowledges funding from the Pioneer Center for Accelerating Materials Discovery (CAPeX), DNRF Grant P3. FFR, FL and GP acknowledge funding by the NCCR MARVEL, a National Centre of Competence in Research, funded by the Swiss National Science Foundation (grant number 205602). GP acknowledges funding by the Open Research Data Program of the ETH Board (project “PREMISE”: Open and Reproducible Materials Science Research).