Machine-learning interaction potentials for non-spherical nanoparticles

The aim is to develop machine-learning models that describe how nanoparticles interact with each other, even when they have complex, non-spherical shapes (such as rods, cubes, triangles, or stars). To do this, we will generate interaction data points using macroscopic physical models (including dispersion forces, magnetic effects, and ligand–solvent interactions), and train modern deep-learning methods to create smooth and reliable energy landscapes. A key goal is predict how nanoparticles prefer to attach to each other.  

The machine-learning models will be validated against detailed atomistic simulations and compared with experimental results on self-assembly. Ultimately, this will allow us to predict how complex nanoparticles organize into superlattices.

Recent publications related to the project: 

Costanzo, S. et al. Nanoscale, 2021, 12, 24020-24029.

Lahouari, A., Piquemal, J.-P. and Richardi, J. J. Phys. Chem. C 2024 128, 1193-1201

The PhD will take place at the Laboratoire de Chimie Théorique at Sorbonne Université (Paris, Latin Quarter), an internationally recognized center in theoretical and computational chemistry.

The student will be supervised by Associate Professor Johannes Richardi and Dr. Riccardo Spezia

Collaboration with an experimental research group is also planned.

The candidate should have a Master (or equivalent) in Theoretical and/or Computational Chemistry or related atomistic/molecular sciences (Computational or Molecular Physics etc.). Speaking French is not required.