Using Machine Learning to enhance Perovskite Solar Cell performance

This PhD project develops physics-based and machine-learning-enhanced models to optimise perovskite and perovskite–silicon tandem solar cells. It addresses coupled optical, charge, and ion transport processes to improve device efficiency across different operating conditions, supporting the advancement of high-performance photovoltaic technologies for net-zero energy systems.

Solar photovoltaics (PV) accounts for almost 80% of global renewable power capacity and is being rapidly expanded to support the target of Net Zero Emissions by 2050. Among emerging PV technologies, perovskite solar cells are particularly promising, with power conversion efficiencies increasing from 4% in 2009 to 27% in 2025. Their compatibility with silicon photovoltaics also makes them well suited for high-efficiency tandem solar cells.

This PhD project focuses on the development and application of physics-based models to support the design and optimisation of both single-junction perovskite solar cells and perovskite–silicon tandem devices. The research will combine advanced mathematical modelling with machine learning techniques to optimise device architectures for efficient light harvesting and electrical power generation across different geographical latitudes.

A central challenge is the complex coupling between optical behaviour and charge and ion transport within the device, including the motion of electrons, holes, and mobile ions in the perovskite layer. Addressing these multi-physics interactions requires sophisticated modelling approaches. Established physical models will therefore be integrated with machine learning methods to efficiently explore the high-dimensional device parameter space. Techniques such as Bayesian parameter estimation will also be used to infer material and device parameters from experimental data where these are not directly known.

The supervisory team brings complementary expertise in photovoltaic device physics, computational modelling, and machine learning. The interdisciplinary training provided will equip the student with highly transferable skills for careers in both industry and academia.

Entry requirements:

A 2:1 in Physics, Mathematics, Engineering or Chemistry

How to apply:

Please direct enquiries to Giles Richardson G.Richardson@soton.ac.uk and attach a cv and degree transcripts.

Funding Notes

Funding will be sought from the Centre for Doctoral Training in AI for Sustainability (SustAI). This will fully fund UK and EU students.