In this project we want to explore the suitability of Physics Informed Neural Networks (PINNs) to support advanced metrology techniques in semiconductor manufacturing. PINNs are a form of neural networks that incorporate physics equations and corresponding boundary conditions to improve robustness and efficiency. We will focus on a use case involving InfraRed Atomic Force Microscopy (IR-AFM), a technique allowing IR-spectroscopy with nm-scale lateral resolution. This technique is relevant for characterizing the increasingly complex structures on computer chips, along with their production processes. In this project, we will use PINNs to predict the Electro-Magnetic field strength as function of a measured AFM tip shape, based on training on the EM field obtained from FEM simulations with idealized tip shapes. We will compare PINN simulation results to measurements. This project is to be the first step in a larger effort towards reverse modelling of sample characteristics from measurements using PINNs.You will work in a team with a strong background in developing AFM technologies consisting primarily of physicists. You will however also interact with a team based in The Hague with a strong background and infrastructure related to Artificial Intelligence.