PhD Studentship: Decoding Charge Dynamics in Triboelectric Nanogenerators with Physics-Informed Neural Networks
University of Exeter
Exeter
CDD
Temps plein
Physical & Environmental Sciences
Publié le 19 novembre 2026
Description du poste
Every mechanical contact between two dissimilar materials initiates a complex, microscopic migration of charge such as a transient of electrons and ions governed by local surface states, overlapping wavefunctions, and local potential barriers. Despite the rapid development of triboelectric nanogenerators (TENGs), i.e. devices able to harvest charge transfer occurring at the contact/separation of materials, mapping this interface dynamics in real-time remains a formidable boundary in applied physics. The complex convolution of environmental parameters, material degradation, and multi-axial mechanical strain yields highly non-linear displacement currents that elude standard analytical formulas. This proposal introduces a novel methodology to map these hidden interface physics by utilizing deep neural networks as universal function approximators. By treating the neural network not just as a statistical black box, but as a framework bound by the fundamental conservation laws of electrostatics and quantum charge trap states with characteristic energies and relaxation, this research will decode the micro-scale charge transfer and relaxation mechanisms that ultimately dictate macro-scale energy harvesting performance. The doctoral researcher will fabricate TENGs using films derived from Ghanaian agricultural residues and investigate the effects of surface functionalisation. They will vary contact force and relative humidity independently while controlling approach and separation speeds. Independent measurements of charge transfer and decay will provide reference data for testing a compact neural network that incorporates the physics of charge evolution and electrical readout. The researcher will compare the network with a conventional physical model on devices held out from training to determine when both quantities can be recovered reliably and when a further measurement is needed. Decoding these charge dynamics would provide the physical basis for multimodal sensing with a single triboelectric element. The project will test whether contact force and local humidity affect charge transfer and decay in sufficiently different ways for both quantities to be estimated from the same waveform when they vary together. This would open a route to wearables that measure mechanical and environmental conditions through a single electrical readout. The same approach could later be tested for contact speed or changes in the local chemical environment. The project builds on Exeter’s EPSRC-funded research in quantum materials, its work on triboelectric nanogenerators and its recent use of AI to reconstruct optical spectra from a single two-dimensional perovskite photodetector by the team of supervisors. Prof Saverio Russo (Physics) will guide device measurements and AI-based inference, Dr Steven Hepplestone (Physics) will guide physical modelling and interpretation of charge dynamics at material interfaces and Prof Monica F. Craciun (Engineering) will guide film development and controlled testing, drawing on her work on cellulose–graphene nanogenerators and textile sensor stability. The findings will inform planned ERC and EPSRC proposals on what AI can infer from experimental signals and how to design measurements that resolve ambiguities.