A new open-access resource brings together more than 4.5 terabytes of experimental battery data, making advanced battery imaging datasets easier for researchers to find, analyse, and reuse.
The Battery Imaging Library (BIL) contains experimental results from 48 scientists across 18 institutions worldwide. The collaboration used 13 different imaging modalities and examined batteries at multiple length scales, creating a broad collection of raw data and reconstructed images.
Researchers from Diamond Light Source and the ISIS Neutron and Muon Source contributed to the collection, alongside scientists from synchrotrons and national laboratories internationally.
The resource is intended to support computational modelling, image analysis, research training and the development of new approaches to battery characterisation.
A wider view inside batteries
The collection brings together imaging techniques that reveal different aspects of battery structure and composition.
At Diamond Light Source, scientists used I12-JEEP, the UK’s flagship high-energy X-ray imaging and diffraction beamline, to generate high-resolution static and dynamic X-ray micro-computed tomography data.
Commercial cylindrical lithium-ion, LiFeS2, and alkaline batteries were scanned, producing datasets that reveal their internal morphological structures. The dynamic measurements also capture changes within batteries over time.
The Battery Imaging Library provides access to these datasets, along with results from other techniques, allowing researchers to work with experimental information collected at different scales.
Neutron imaging adds complementary data
Scientists at the ISIS Neutron and Muon Source used the IMAT neutron imaging beamline to collect neutron computed tomography data.
This is significant for battery research because neutron imaging is less commonly available to the battery community and provides complementary information to X-ray measurements.
The technique is particularly useful for examining lighter components within lithium-ion batteries. The datasets can reveal lithium-based electrolytes and the location of lithium species within a cell.
Combining these forms of battery imaging lets researchers examine the same broad problem from multiple perspectives rather than relying on a single imaging method.
Making large datasets reusable
The scale of BIL presented its own challenge. A collection containing several terabytes of experimental information needs to be searchable for researchers to use it in practice.
Ronan Docherty, a PhD student at Imperial College London’s Centre for Doctoral Training for the Advanced Characterisation of Materials, developed a dedicated website to address this issue.
The website lets users browse different imaging modalities and find datasets relevant to their work. Access to both raw data and reconstructed images means researchers can work with the underlying experimental information as well as processed results.
Supporting battery research and education
The Battery Imaging Library was developed with three core aims:
- Expanding access to experimental data
- Supporting the development of imaging and analysis methods
- Providing resources for education and training.
Researchers can use the datasets to test computational models and assess image analysis software using realistic experimental results. Students and professionals can also use the data and associated workflows for practical training.
The collection may also provide useful material for developing AI approaches to scientific research. Its accompanying publication highlights the role that open experimental datasets can play in training AI models for materials research.
By bringing together datasets from multiple facilities, imaging techniques, and length scales, the Battery Imaging Library gives researchers a single resource for accessing experimental battery imaging data and using it to develop, test, and refine new analytical approaches.
