| EPSRC Reference: |
EP/Y014405/1 |
| Title: |
Designing and optimizing polar photovoltaics with physics informed machine learning |
| Principal Investigator: |
Butler, Dr K T |
| Other Investigators: |
|
| Researcher Co-Investigators: |
|
| Project Partners: |
|
| Department: |
Chemistry |
| Organisation: |
UCL |
| Scheme: |
New Investigator Award |
| Starts: |
01 July 2024 |
Ends: |
30 June 2027 |
Value (£): |
422,894
|
| EPSRC Research Topic Classifications: |
| Materials testing & eng. |
Numerical Analysis |
| Solar Technology |
|
|
| EPSRC Industrial Sector Classifications: |
|
| Related Grants: |
|
| Panel History: |
|
|
Summary on Grant Application Form |
Solar photovoltaics (PV) are predicted to be a key enabling technology on the road to net zero - the international energy agency predicts that solar electricity production should increase from 430 Tw-h in 2017 to 6,410 Tw-h in 2040. [1] To achieve this, new materials, that are cheaper to process and manufacture than the current market leader (crystalline Si) are urgently needed. Inorganic materials with a spontaneous polarization have attracted significant interest for photovoltaics, due to their ability to overcome several limitations of traditional PV materials, but there are a limited number of such materials with the required optical absorption for effective PV. In this project we will use our expertise in data-driven materials science to design new polar PV (PPV) absorbers with the potential to offer excellent efficiencies from cheap, earth-abundant materials.
Machine learning (ML) has had a transformative effect on many important industries. However, in materials science, the lack of large labelled datasets of important properties proves a barrier to adopting some of the most commonly applied methods from fields such as image analysis or language processing, where data is not an issue. Recent work has shown that incorporating physical knowledge into ML greatly enhances learning ability, and allows powerful predictive models on limited datasets. In the project we will develop ML methods for PPV materials, that include prior physical knowledge, thus allowing us to learn accurate models for the important properties of PPV materials. These ML models, once trained are highly scalable and will allow us to virtually screen through spaces of hundreds of billions of candidate materials - a scale unimaginable with traditional simulations, let alone with physical experiments.
The award will fund the development of a team with a unique combination of expertise in machine learning, photovoltaics and materials design. The proposal includes Queen Mary University of London (QMUL), which has recently invested more than £16 million in state of the art facilities for data science and the Physical Sciences Data Infrastructure, who are developing the technology for facilitating the future of data-driven physical science in the UK. The project will also work closely with experimental groups at QMUL who are world-leading in the development of polar photovoltaic materials and devices.
[1] International Energy Agency (IEA) World Energy Outlook 'Sustainable Scenario' https://www.iea.org/weo2018/
|
|
Key Findings |
|
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk
|
|
Potential use in non-academic contexts |
|
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk
|
|
Impacts |
|
| Description |
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk |
| Summary |
|
| Date Materialised |
|
|
|
|
Sectors submitted by the Researcher |
|
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk
|
| Project URL: |
|
| Further Information: |
|
| Organisation Website: |
|