Stage - Thermal transport in van der Waals superlattices by equivariant deep learning models H/F

Vacancy details

General information


The French Alternative Energies and Atomic Energy Commission (CEA) is a key player in research, development and innovation in four main areas :
• defence and security,
• nuclear energy (fission and fusion),
• technological research for industry,
• fundamental research in the physical sciences and life sciences.

Drawing on its widely acknowledged expertise, and thanks to its 16000 technicians, engineers, researchers and staff, the CEA actively participates in collaborative projects with a large number of academic and industrial partners.

The CEA is established in ten centers spread throughout France



Position description


Materials, solid state physics



Job title

Stage - Thermal transport in van der Waals superlattices by equivariant deep learning models H/F


The goal of this internship is to study the thermal transport of chalcogenide superlattices by means of molecular dynamics simulations based on E(3)-equivariant neural network interatomic potentials.

Contract duration (months)


Job description

Among the next generation nonvolatile memories (NVM), Phase Change Memory (PCM) technologies based on chalcogenide materials, such as GeSbTe (GST), are the most mature. Thermally driven phase transitions of the phase change material are achieved using electrical pulses in order to switch between a low resistive crystalline state and a high resistive amorphous state.

Recently, chalcogenide superlattices (SL) formed by the alternate stacking of ultrathin layers separated by van der Waals (vdW)-like interfaces (e.g. GeTe/Sb2Te3 or GeSb2Te4/Sb2Te3) have been reported to improve the energy efficiency of PCMs. SL interfaces are expected to play a key role in the electrical performances of these devices.

Simulations using molecular dynamics (MD) are particularly suitable to study the physical properties of these materials at the atomic scale but require an accurate description of the potential energy surface (PES). In the past couple of years, much progress has been made in the development of machine learning interatomic potentials (ML-IP) to describe complex systems. Recently, ML-IP relying on E(3)-equivariant deep learning models have shown very high accuracy and data efficiency compared to other ML-IP methods.

The goal of this internship is to employ MD simulations to study the heat conduction mediated by lattice vibrations in chalcogenide SL materials using the Green-Kubo formalism and E(3)-equivariant ML-IPs. The candidate will first have to train a potential to study GST materials using a dataset of reference systems obtained from density functional theory (DFT) calculations. Then, the thermal transport in chalcogenide SLs will be studied using MD simulations with the LAMMPS code ( The simulations will be carried out on high-performance computing (HPC) resources with GPU accelerators.

Applicant Profile

  • Background in solid state physics and chemistry or related fields
  • Programming using Python and C++
  • Understanding of Linux and parallel computing is desirable
  • Previous experience with machine learning techniques is a plus

Position location



Job location

France, Auvergne-Rhône-Alpes, Isère (38)



Candidate criteria

Prepared diploma

Bac+5 - Diplôme École d'ingénieurs

Recommended training

M2 in physics, chemistry or related fields

PhD opportunity



Position start date