Formal Explanations for Artificial Intelligence
The candidate will contribute to the PyRAT formal analyzer, developped in the lab. This state-of-the-art analyzer is both used as a research sandbox and as an industrial-grade tool. As such, the candidate will work at the boundary of academia and industry.
The candidate missions are the following:
- actively build, update and deliver a state of the art on formal verification, in particular formal verification of machine learning and formal explanations
- contribute to scientific and technical discussions on PyRAT's design and implementations, and pursue said implementations
- investigate and apply the uses of PyRAT for formal explanations
- contribute to funded projects, either national or international, both by institutional and industrial actors, in particular by helping writing deliverables on such projects
- contribute to publications and/or technical reports around PyRAT
- help the dissemination of PyRAT, in particular by contributing to tutorials, courses and presentations and presenting them at scientific and industrial venues
High-throughput PVD deposition of semiconducting materials
Lead halide perovskites are a class of emerging semiconductors that have demonstrated considerable potential for utilization in solar cells.Nevertheless, the release of toxic lead into the environment during the lifespan of the cells is still a concern for their further commercialization.
This 24-months project aims at optimizing the deposition of lead-free double perovskite thin films for photovoltaic applications using PVD (Physical Vapor Deposition). The optimization of the material will be carried out by implementing high-throughput approaches in both the process and characterisation workflows.
Robust path-following solvers for the finite element simulation of cracking in complex heterogeneous media: application to reinforced concrete structures
Path-following (or continuation) procedures are used to describe the unstable responses of structures exhibiting snap-back or snap-through phenomena. These methods consist in adapting the external load during the deformation process in order to satisfy a control constraint, by introducing an additional unknown, the load multiplier. Several variants exist depending on the controlled quantity: degrees of freedom, strain measures, or variables related to energy dissipation.
In addition to enabling the tracing of unstable responses, a major advantage of these approaches lies in improving the convergence of incremental Newton-type solvers by reducing the number of iterations required. This gain often compensates for the additional computational cost associated with the continuation algorithm. Some formulations have proven both efficient and simple to implement.
However, no objective criterion yet allows one to determine which formulation is best suited for the simulation of reinforced concrete structures, where multiple dissipation mechanisms coexist along with a strong spatial variability of the material properties.
The proposed postdoctoral work aims to develop robust path-following algorithms for such structures, building upon previous research carried out at CEA. It will include a critical analysis of existing formulations, an evaluation of their performance (monolithic or partitioned solvers), followed by their implementation. Finally, representative test cases of industrial structures will be simulated to assess the gain in robustness and computational cost compared to standard solvers.
Novel concentrated alloys (HEA/CCA) for nuclear applications: Corrosion and irradiation resistance in molten salts
This postdoctoral position is part of the national PEPR DIADEM program, within the DIAMS project, which aims to accelerate the discovery of new materials by combining computer design and experimental testing. The research focuses on materials for Molten Salt Reactors (MSRs), which require alloys that are resistant to both molten salt corrosion and irradiation. Certain optimized high-entropy alloys and complex concentrated alloys (HEA-CCA), offer superior performance compared to conventional materials such as austenitic stainless steels (ASS).
The postdoctoral research follows on an IA-based alloy design project carried out at IMN (Nantes), which identified promising compositions produced by conventional metallurgy and characterized after corrosion and irradiation on the JANNuS Saclay platform. The tests were carried out sequentially.
Starting in early 2026, the unique JANNuSel device will enable simultaneous corrosion and irradiation testing on both conventionally processed alloys and new compositions produced by AM (additive manufacturing) on the SAMANTA platform.
The samples will be analysed by SEM, TEM/EDX, EBSD, Raman spectroscopy, Atom Probe Tomography (APT), and synchrotron X-ray diffraction (MARS, SOLEIL) to understand the underlying mechanisms and optimize alloy properties.
The position is based at CEA Saclay, in close collaboration with IMN and Mines Saint-Étienne, and benefits from a rich interdisciplinary research environment.
Experimental and Thermodynamic Modeling of Corium Phases Formed During Severe Nuclear Accidents (24 months)
During severe accidents in pressurized water reactors, uranium dioxide (UO2) fuel reacts with zirconium alloy cladding and the steel vessel, forming a mixture of liquid and solid phases known as "in-vessel corium". If the vessel ruptures, this corium interacts with the concrete raft, forming "ex-vessel corium". This phenomenon occurred in the Chernobyl and Fukushima severe accidents. To simulate these stages, multi-physics codes require accurate thermodynamic and thermophysical data for the various phases of corium. This project aims to fill the data gap through experimental measurements and modeling. The work will involve synthesizing samples, measuring liquidus/solidus temperatures and liquid phase densities, and characterizing samples using advanced techniques. Moreover, the laser heating setup combined with aerodynamic levitation (ATTILHA) used to acquire data will be improved. Experimental results will be compared with thermodynamic models (TAF-ID database), and discrepancies will be resolved using the CALPHAD method. Thermophysical data will also be validated using atomistic simulations and other measurement techniques.
TOMOGLASS: Gamma Emission Tomography Applied to the Radiological Characterization of Glass Residues from the Cold Crucible Vitrification Process
The TOMOGLASS project aims to develop an innovative gamma tomography system capable of operating in high-activity environments to characterize in three dimensions the glass residues resulting from the vitrification process of nuclear waste. The objective is to precisely locate platinum-group inclusions, which are poorly soluble in glass, in order to improve the understanding and control of the process. The system is based on a compact gamma imager integrating high-resolution pixelated CZT detectors, pinhole-type collimation, and mounting on a robotic arm. It will enable multi-isotopic reconstruction using advanced tomographic algorithms. This project is part of the modernization of the La Hague facilities and the integration of digital technologies within the framework of the factory of the future.
The first phase of the project will consist in demonstrating the feasibility of implementing a spectro-imager prototype in a constrained environment, building on existing technological components: detection modules and acquisition electronics based on the HiSPECT technology, and image reconstruction algorithms developed at CEA-Leti. The work will focus on conducting a multi-parameter study through numerical simulations (Monte Carlo calculation code) to design an optimized measurement system, and to generate simulated datasets for various representative measurement configurations. Once the concept has been validated, the work will continue in year N+1 with the assembly of the prototype components and their integration on a robotic arm. Experimental tests may then be carried out to demonstrate the system in a representative environment.
Tools and diagnostic methods for the reuse of electronic components
The Autonomy and Sensor Integration Laboratory (LAIC) at CEA-Leti has the primary mission of developing sensor systems for the digitalization of systems. The team's activities are at the interface of hardware (electronics, optronics, semiconductors), software (artificial intelligence, signal processing), and systems (electronic architecture, mechatronics, multiphysics modeling).
In a context of exponential growth in electronics and scarcity of resources, the reuse of electronic components from end-of-life systems represents a promising avenue to limit environmental impact and support the development of a circular economy. The objective of this project is to develop an advanced diagnostic methodology to assess the health status of electronic components, particularly power components, to reintegrate them into a less constraining second-life cycle.
The postdoctoral researcher will be tasked with developing a comprehensive approach to evaluate the reuse potential of electronic components, with the aim of reintroducing them into second-life cycles. This will include:
- Identifying relevant health indicators to monitor the performance evolution of components (e.g., MOSFETs, IGBTs, capacitors, etc.);
- Setting up test benches and sensors adapted to measure electrical, thermal, or mechanical parameters, with the goal of detecting signs of aging;
- Analyzing degradation modes through experimental tests and failure models;
- Developing algorithms for predicting the remaining useful life (RUL) adapted to different usage scenarios;
- Contributing to scientific publications, the valorization of results, and collaboration with project partners.
Analysis of Gas Effluents for More Eco-Responsible Plasma Etching Processes
Traditionally used fluorinated gases, such as CF4 and C4F8, exhibit extremely high Global Warming Potentials (GWPs), significantly contributing to climate change. To address these environmental challenges, the project aims to promote the use of alternative low-GWP gases in combination with efficient exhaust abatement systems at the reactor outlet, while maintaining high-performance plasma etching processes. In this context, the postdoctoral researcher will be responsible for the analysis and characterization of gaseous species in an industrial plasma etching reactor using mass spectrometry. These measurements will be compared with the gas effluent at the outlet of the pumping and abatement systems. The main objectives are (i) to quantify the Destruction Removal Efficiency (DRE) of high and low GWP fluorocarbon gases during plasma processing and within the pumping and abatement stages, and (ii) to identify and propose innovative, environmentally responsible alternatives to minimize the release of high-GWP gaseous effluents.
Understanding and modeling the thermodynamic and kinetic properties of MOX fuel in future reactors
This study is part of the Sodium-Cooled Fast Reactor projects. Uranium and plutonium dioxide (U,Pu)O2, known as MOX, is the reference fuel. During operation, fuel pellets are subjected to a high thermal gradient that induces mass transport, thermodiffusion, and vaporization phenomena, coupled with irradiation effects. Fuel performance codes are developed to simulate the behavior of fuel rods under nominal and incidental conditions, up to and including meltdown.
The objective of this study is to improve the thermokinetic model of MOX used in these codes. This model is based on the description of the U-Pu-O system using the CALPHAD method, coupled with a database of element mobilities developed using DICTRA software. The description of defects will be extended with the introduction of metal vacancies and oxygen clusters. The description of thermodynamic data (oxygen potential and heat capacity) and the phase diagram will also be improved by taking into account the most recent data. Finally, the mobility database, coupled with the Calphad model, will be improved to better describe diffusion in MOX. New experimental data as well as data calculated using atomic-scale calculation methods (molecular dynamics, ab initio) will be used.