Physicochemical Properties of Antimony-containing Photovoltaic (PV) Glass
The proposed PhD thesis is part of the ANR GRISBI project (2026–2030), which aims to optimize the recycling of glass from photovoltaic (PV) panels. These glasses, predominantly manufactured in China, are doped with antimony oxide (Sb2O3) to ensure high transparency while keeping production costs low. However, the presence of antimony currently prevents the recycling of these glasses within the European flat glass industry, which would otherwise greatly benefit from this secondary raw material to reduce its environmental footprint — particularly its greenhouse gas emissions, in line with the carbon neutrality targets set by the Paris Agreement (2015). To make the recycling of PV glass into flat glass production feasible, it is therefore essential to gain a deeper understanding of the physicochemical behavior of antimony in glass, and more generally, within the float process, which involves a hot glass / liquid tin interface.
The core scientific objective of the PhD is to determine the redox equilibria between the multivalent species present in PV glasses, in particular the Sb2O3/Sb and Fe2O3/FeO couples. The work will involve preparing glasses with different Sb2O3 contents, then determining the mechanisms of antimony incorporation into the glass structure, as well as the temperature and oxygen partial pressure (pO2) conditions leading to the reduction of Sb³? to metallic Sb°. Experimental results, based on advanced materials characterizations such as SEM, XRD, EXAFS, and XANES, will be used to enrich thermodynamic databases and to develop a methodology enabling the recycling of Sb-doped PV glasses in flat glass production.
The PhD will be conducted at CEA Marcoule, in collaboration with IMPMC (Sorbonne Université) — two laboratories internationally recognized for their expertise in glass science. All glass samples will be synthesized by the PhD student, and their characterization will primarily rely on facilities available at CEA and IMPMC.
A background in Materials Science is required. This research project will provide the PhD candidate with the opportunity to develop strong expertise in applied glass science and industrial recycling technologies.
Microemulsion model: Towards the prediction of liquid-liquid extraction processes
This multi-scale modeling PhD project aims to develop innovative theoretical approaches and numerical tools to predict the extraction processes of strategic metals, which are essential for the energy transition. Among the existing methods, liquid-liquid extraction is a key process, but its underlying mechanisms remain poorly understood. To address these challenges, the solvent phases will be represented as microemulsions through a synergy of mesoscopic and molecular modeling approaches.
The mesoscopic approach will involve the development of a code based on microemulsion theory using a random wavelet basis. This code will enable the characterization of the structural and thermodynamic properties of the solutions. The molecular approach will rely on classical molecular dynamics simulations to evaluate the curvature properties of the extractants, which are essential for bridging the two scales.
The new high-performance computational code may integrate artificial intelligence techniques to accelerate the minimization of the system’s free energy while accounting for all chemical species present with a minimal number of parameters. This will pave the way for new research directions, such as predicting speciation and calculating thermodynamic instabilities in ternary phase diagrams, thereby identifying unexplored experimental conditions.
This PhD thesis, conducted at the Mesoscopic Modeling and Theoretical Chemistry Laboratory at the Marcoule Institute for Separation Chemistry, will have applications in the recycling domain and extend to the broader field of nanoscience, thereby expanding the impact of this work.
The PhD candidate, with a background in physical chemistry, theoretical chemistry, or physics and a strong interest in programming, will be encouraged to disseminate their scientific results through publications and presentations at national and international conferences. By the end of the thesis, the candidate will have acquired a broad range of skills in theoretical chemistry, modeling, numerical computation, and physical chemistry, providing numerous career opportunities in both academic research and industrial R&D.
Fast charging of lithium-ion batteries : Study of the lithium plating phenomenon using operando NMR
The focus of the thesis is the fast-charging process of lithium-ion batteries and, more specifically, the phenomenon of lithium plating, which will be studied using operando NMR. The target application is electric mobility. The objective of the thesis is to study the dynamics of lithium insertion and lithium metal deposition at the graphite or graphite/silicon-based negative electrode in order to understand the mechanisms leading to plating formation.
Operando NMR is an ideal technique for this study because it offers the unique possibility of simultaneously tracking the signals of the lithiated graphite phases and of deposited lithium during the electrochemical processes. The coupling of electrochemistry and operando NMR will allow us to determine the onset of plating, i.e. the potential of the negative electrode at which deposition begins, and the kinetics of lithium metal deposition and reinsertion at different temperatures and different charging current regimes. We will study Li-ion batteries with a pure graphite negative electrode, but also with graphite-silicon electrodes, in order to investigate the impact of silicon on this phenomenon. The data obtained on the onset mechanisms and the kinetics of lithium metal deposition and reinsertion will be used in a multiphysics model that has already been developed in the laboratory to improve the prediction of plating onset. We will then be able to evaluate the chargeability gains on an NMC 811 // Gr+Si system incorporating optimized electrodes and propose innovative charging protocols.
Multiphysic modeling of sintering of nuclear fuel pellet: effect of atmosphere on shrinkage kinetics
Uranium dioxide (UO2) fuels used in nuclear power plants are ceramics, for which solid-phase sintering is a key manufacturing step. The sintering stage involves heat treatment under controlled partial O2 pressure that induces coarsening of UO2 grain and then consolidation and densification of the material. Grain growth induce material densification and macroscopic shrinkage of the pellet. If the green pellet (powder obtained by pressing, manufacturing step before sintering) admit a highly heterogeneous density, this gradient leading to differential shrinkage and the appearance of defects. Furthermore, the sintering atmosphere, i.e., the gas composition in the furnace, impacts grain growth kinetics and thus the shrinkage of the pellet. Advanced simulation is the key to improving understanding of the mechanisms observed as well as optimizing manufacturing cycles.
The PhD thesis aims at developing a Thermo-chemo-mechanical modeling of sintering to simulate the impact of the gas composition and properties on the pellet densification. This scale will enable us to take into account not only the density gradients resulting from pressing, but also the oxygen diffusion kinetics that have a local impact on the densification rate, which in turn impacts the transport process. Therefore, a multiphysics coupling phenomenon has to be modelled and simulated.
This thesis will be conducted within the MISTRAL joint laboratory (Aix-Marseille Université/CNRS/Centrale Marseille CEA-Cadarache IRESNE institute). The PhD student will leverage his results through publications and participation in conferences and will have gained strong skills and expertise in a wide range of academic and industrial sectors.
Magnetar formation: from amplification to relaxation of the most extreme magnetic fields
Magnetars are neutron stars with the strongest magnetic fields known in the Universe, observed as high-energy galactic sources. The formation of these objects is one of the most studied scenarios to explain some of the most violent explosions: superluminous supernovae, hypernovae, and gamma-ray bursts. In recent years, our team has succeeded in numerically reproducing magnetic fields of magnetar-like intensities by simulating dynamo amplification mechanisms that develop in the proto-neutron star during the first seconds after the collapse of the progenitor core. However, most observational manifestations of magnetars require the magnetic field to survive over much longer timescales (from a few weeks for super-luminous supernovae to thousands of years for Galactic magnetars). This thesis will consist of developing 3D numerical simulations of magnetic field relaxation initialized from different dynamo states previously calculated by the team, extending them to later stages after the birth of the neutron star when the dynamo is no longer active. The student will thus determine how the turbulent magnetic field generated in the first few seconds will evolve to eventually reach a stable equilibrium state, whose topology will be characterized and compared with observations.
Exploration of VACNTs in Anode-less Batteries: Mechanism and Cell Optimization
Anode-less or anode-free batteries are getting increasing attention owing to their excellent energy density, cost efficiency, and ease of process upscaling. Exploring anode-less battery will offer a breakthrough in energy storage devices by using the lithium reserve already present in the NMC cathode to reversibly cycle after an initial formation process, which will reduce the overall thickness, processing steps, and cost of materials, and provide excellent energy density. Vertically aligned CNTs (VACNTs) on metal substrates can be an interesting choice for this application due to their low thickness, reproducible synthesis process, and uniform surface properties, which have already proven their applicability in supercapacitors. In this PhD project, we will investigate their newer avenue of applications- anode-less batteries, where VACNTs act as the lithium or sodium plating substrate. We will study the electrochemistry of VACNT in lithium anode-less batteries (in liquid and solid electrolytes) and in sodium anode-less batteries in a liquid electrolyte. The PhD student will work on the synthesis optimizations of VACNT to tune the thickness and density to match their electrochemistry. Post-cycling studies (Raman and SEM) will be carried out to study the effect of cycling and the electrolytes on the VACNT layers. The project aims to explore the possibility of the application of VACNTs in various energy storage systems, which can open up new application possibilities and valorization
Magneto-convection of solar-type stars: flux emergence and origin of starspots
The Sun and solar-type stars possess rich and variable magnetism. In our recent work on turbulent convective dynamos in this type of star, we have been able to highlight a magneto-rotational history of their secular evolution. Stars are born active with short magnetic cycles, then slow down due to braking by their magnetized particle wind, their magnetic cycle lengthens to become commensurate with that of the Sun (lasting 11 years) and finally, for stars that live long enough, they end up with a loss of cycle and a so-called anti-solar rotation (slow equator/fast poles). The agreement with observations is excellent, but we are missing an essential element to conclude: What role do sunspots/starspots play in the organization of the magnetism of these stars, and are they necessary for the appearance of a stellar magnetic cycle, e.g. the so-called “paradox of spotty dynamos”? Indeed, our HPC simulations of solar dynamos do not have yet the angular resolution to resolve the spots, and yet we do observe cycles in our simulations of stellar dynamos for Rossby numbers < 1. So, are the spots simply a surface manifestation of an internal self-organization of the cyclic magnetism of these stars, or do they play a decisive role? Furthermore, how do the latitudinal flux emergence and the size and intensity of the spots forming on the surface evolve during the magneto-rotational evolution of these stars? To answer these key questions in stellar and solar magnetism in support of the ESA space missions Solar Orbiter and PLATO, in which we are involved, new HPC simulations of stellar dynamos must be developed, allowing us to get closer to the surface and thus better describe the process of magnetic flux emergence and the possible formation of sun/starspots. Recent tests showing that magnetic concentrations inhibiting local surface convection form in simulations with a higher magnetic Reynolds number and smaller-scale surface convection strongly encourage us to continue this project beyond the ERC Whole Sun project (ending in April 2026). Thanks to the Dyablo-Whole Sun code that we are co-developing with IRFU/Dedip, we wish to study in detail the convective dynamo, the emergence of magnetic flux, and the self-consistent formation of resolved spots, using its adaptive mesh refinement capability while varying global stellar parameters such as rotation rate, convective zone thickness, and surface convection intensity to assess how their number, morphology and latitude of emergence change and if they contribute or not to the closing of the cyclic dynamo loop.
Surface technologies for enhanced superconducting Qubits lifetimes
Materials imperfections in superconducting quantum circuits—in particular, two-level-system (TLS) defects—are a major source of decoherence, ultimately limiting the performance of qubits. Thus, identifying the microscopic origin of possible TLS defects in these devices and developing strategies to eliminate them is key to superconducting qubit performance improvement. This project proposes an original approach that combines the passivation of the superconductor’s surface with films deposited by Atomic Layer Deposition (ALD), which inherently have lower densities of TLS defects, and thermal treatments designed to dissolve the initially present native oxides. These passivating layers will be tested on 3D Nb resonators than implemented in 2D resonators and Qubits and tested to measure their coherence time. The project will also perform systematic material studies with complementary characterization techniques in order to correlate improvements in qubit performances with the chemical and crystalline alteration of the surface.
Bayesian Neural Inference Using Ferroelectric Memory Transistors
An increasing number of safety-critical systems now rely on artificial intelligence functions that must operate under strict energy constraints and in environments characterized by data scarcity and high uncertainty. However, conventional deterministic AI approaches provide only point estimates and lack principled uncertainty quantification, which can lead to unreliable or unsafe decisions in real-world deployment.
This PhD is positioned within the emerging field of Bayesian electronics, which aims to implement probabilistic inference directly in hardware by leveraging the intrinsic stochasticity of nanoscale devices to represent and manipulate probability distributions. While memristive devices have previously been explored for Bayesian inference, their limited endurance and high programming energy remain critical bottlenecks for on-chip learning.
The objective of this thesis is to investigate ferroelectric field-effect memory transistors (FeMFETs) as building blocks for hardware Bayesian neural networks. The work will involve characterizing and modeling the exploitable ferroelectric randomness for sampling and probabilistic weight updates, designing Bayesian neuron and synapse architectures based on FeMFETs, and evaluating their robustness, energy efficiency, and system-level performance for safety-critical inference under uncertainty.
Adaptive and explainable Video Anomaly Detection
Video Anomaly Detection (VAD) aims to automatically identify unusual events in video that deviate from normal patterns. Existing methods often rely on One-Class or Weakly Supervised learning: the former uses only normal data for training, while the latter leverages video-level labels. Recent advances in Vision-Language Models (VLMs) and Large Language Models (LLMs) have improved both the performance and explainability of VAD systems. Despite progress on public benchmarks, challenges remain. Most methods are limited to a single domain, leading to performance drops when applied to new datasets with different anomaly definitions. Additionally, they assume all training data is available upfront, which is unrealistic for real-world deployment where models must adapt to new data over time. Few approaches explore multimodal adaptation using natural language rules to define normal and abnormal events, offering a more intuitive and flexible way to update VAD systems without needing new video samples.
This PhD research aims to develop adaptable Video Anomaly Detection methods capable of handling new domains or anomaly types using few video examples and/or textual rules.
The main lines of research will be the following:
• Cross-Domain Adaptation in VAD: improving robustness against domain gaps through Few-Shot adaptation;
• Continual Learning in VAD: continually enriching the model to deal with new types of anomalies;
• Multimodal Few-Shot Learning: facilitating the model adaptation process through rules in natural language.