PHotonique Et hydrodynamique par simulatioN Intelligente pour la structuration eXtrême)

PhD student: Mael JOUSSET

ABSTRACT
The structuring of materials using ultrafast femtosecond laser pulses is an advanced technology that opens up vast prospects for research and various applications, particularly in optics, energy, and biomedicine. In this context, the functionalization of future surfaces requires extreme mastery of nanostructuring, enabling the creation of complex surfaces at the nanometric scale. These surfaces result from self-organized patterns such as nanocavities, peaks, streaks, or hexagonal periodic structures, and their design necessitates innovative approaches that combine structured optical beams and hydromechanical phenomena.
This PhD work aims to explore and model the complex interactions between hydrodynamic and photonic phenomena generated by ultrashort lasers in the context of laser nanostructuring. The objective is to push the boundaries of current technologies by proposing innovative numerical approaches to understand and control these interactions. The project is structured around several key areas:
• Advanced hydrodynamic modeling: Development of a numerical method to solve the three-dimensional stochastic Navier-Stokes equations, incorporating terms representing laser energy deposition. The implementation leverages high-performance computing (GPU) capabilities using the Julia programming language.
• Machine learning for multi-scale simulation: Design of a neural network combining supervised and self-supervised approaches, integrating real data, artificial data, and physical knowledge. This network must be capable of modeling fluid dynamics at the nanometric scale using limited data, with strong generalization to multi-scale phenomena.
• Study of self-organization dynamics: Analysis of self-organization mechanisms with a focus on convective instabilities underlying these phenomena.
• Prediction and control of chiral structures: Integration of optical torque effects (spin and orbital angular momentum) into numerical and machine learning models to predict and generate chiral or asymmetric nanometric structures.



 
 


No publication available yet.