Machine-learning enabled design of multiferroic optoelectronic materials

This project will develop a machine-learning-enabled multiscale modelling framework to accelerate the discovery and design of multiferroic materials for next-generation optoelectronic devices. Multiferroics uniquely couple ferroelectric, magnetic, and ferroelastic orders, enabling ultra-low-power computing, multifunctional sensors, and energy-efficient memory technologies. Yet their rational design is hindered by the absence of scalable simulation tools that can capture coupled structural, electronic, and polarization responses across atomic to device length scales.The fellowship will establish physics-informed machine learning approaches that integrate latent Ewald summation and Hamiltonian-based models with first-principles calculations. This will enable accurate treatment of long-range electrostatics, electronic excitations, and ferroic switching dynamics. The methodology progresses from (i) developing advanced ML–DFT workflows, (ii) investigating domain wall dynamics, defect interactions, and their impact on local electronic structures, to (iii) scaling up to device-level simulations of charge transport and switching under realistic operating conditions. Iterative validation with experimental collaborators in spectroscopy and microscopy will ensure predictive accuracy and technological relevance.The expected outcomes include a validated, transferable open-source platform for modelling multiferroics, new insights into structure–property–function relationships at ferroic interfaces, and design principles for energy-efficient devices. By bridging artificial intelligence, quantum simulations, and device modelling, the project contributes to the Horizon Europe objectives of advancing digital and sustainable technologies, while enhancing the researcher’s independence and career prospects at the intersection of physics, materials science, and AI.