Scientific direction Development of key enabling technologies
Transfer of knowledge to industry

PhD : selection by topics

Technological challenges >> Numerical simulation
4 proposition(s).

See all positions [+]

3D reconstruction of nanoscale objects from stereoscopic electron microscope images

Département des Plateformes Technologiques (LETI)




Numerical simulation (.pdf)

Keywords: Applied mathematics, Images treatment, Modeling, Inverse problem, Microelectronics. Robust, non-destructive and fast 3D metrology is a world-wide major challenge of microelectronics industry to better improve and control the nanotechnology processes [1]. CEA-LETI has state-of-the-art electron microscopes (SEM) for imaging objects from different points of view (stereoscopy). These equipments could be used in production to reconstruct the 3D topography of objects from reliable SEM imaging models and innovative algorithms. CEA-LETI already has strong expertise in this field [2, 3], and several industrial partners show a strong interest in the development of this technology. The objective of this thesis is to develop a 3D metrology from SEM images the most precise and robust as possible. For this, the PhD student will use the Computational Lithography group's theoretical and simulation resources to improve and develop new SEM imaging models. The scope of these models is broad, from the simulation of micrometric objects to nanoscale structures. The PhD student will train the SEM models on a collection of multi-stereo SEM images of patterns, whose 3D topographies will be measured via 3D reference metrology. He will then investigate different mathematical strategies of 3D reconstruction, allowing rapid convergence and quality. Eventually, 3D reconstruction will be applied to different industrial products of interest. Means: CLG python libraries, Collaborative development SVN, Continuous integration, HPC CPU/GPU, LETI technological platform. [1] B. Bunday, 7/5 nm logic manufacturing capabilities and requirements of metrology, SPIE 9780 (2018) [2] J. Bélissard et al., Limits of model-based CD-SEM metrology, Proc. SPIE 10775, 1077518 (2018) [3] C. Valade, Tilted beam SEM, 3D metrology for industry, Proc. SPIE 10959, 109590Y (2019)

Download the offer (.zip)

Multi-scale modelling approach of a steam storage with Phase Change Materials integrated into a thermal process

Département Thermique Biomasse et Hydrogène (LITEN)

Laboratoire Stockage Thermique



Numerical simulation (.pdf)

Steam storage allows industrial processes and power plants to be more flexible, more reliable, and more stable by time-shifting steam production from steam use. The use of Phase Change Materials (PCM) offers numerous advantages, like steam discharge at constant pressure, a high energy density, and a significant reduction of the pressurized volume. At CEA, a shell-and-tubes PCM storage technology is being studied for several years, through various experimental facilities and modelling activities. A recent work allowed to develop a multi-scale methodology based upon Computational Fluid Dynamics numeric simulations (modelling the PCM melting front and natural convection in a complex geometry) to feed a component model. The aim of the PhD thesis is on one hand to check this predictive approach on new experimental results, and on the other hand to generalize it on other storage geometries, in order to validate its use for the design of industrial-scale storage systems. Finally, another objective is to integrate the component model derived from the multi-scale approach into a system model so as to study the interactions between the storage and its environment and thus to optimize its operation to meet the process' needs.

Download the offer (.zip)

Deep Reinforcement Learning Agents Revealing Uncertainties in Blockchain Systems

Département Ingénierie Logiciels et Systèmes (LIST)

Lab.systèmes d'information de confiance, intelligents et auto-organisants



Numerical simulation (.pdf)

Since its genesis in late 2008, Bitcoin had a rapid growth in terms of participation, number of transactions and market value. This success is mostly due to innovative use of existing technologies for building a trusted ledger called blockchain. A blockchain system allows its participants (agents) to collectively build a distributed economic, social and technical system where anyone can join (or leave) and perform transactions in-between without needing to trust each other, having a trusted third party and having a global view of the system. It does so by maintaining a public, immutable and ordered log of transactions, which provides an auditable trusted ledger accessible by anyone. However, blockchain systems are environments that are too complex for humans to pre-determine the correct actions using hand-designed solutions. Furthermore, the agents performing in these systems have limited observability, and the state and parameter spaces are vast and changing dynamically. Consequently, agents that can learn to tackle such complex real-world uncertain domains are needed. Based on this observation, , the objective of this thesis is to investigate the uncertain constraints of blockchain systems and to propose a deep reinforcement learning decision-making approach based for all agents like that agents will learn to cooperate in a multi-agent setting in blockchain systems and continuously learn the uncertain constraints.

Download the offer (.zip)

Nonsmooth time-stepping methods for 3D frictional contacts with geometrically nonlinear kinematics

Département Intelligence Ambiante et Systèmes Interactifs (LIST)

Laboratoire de Simulation Interactive



Numerical simulation (.pdf)

The simulation of the dynamics of multi-body systems with intermittent contacts has several applications, ranging from the engineering of the design of industrial products (circuit breakers, clockwork mechanisms ...) to the development of real-time simulators of complex systems (teleoperated robots operating in a hostile environment, offshore lifts, prototyping of assembly processes in the manufacturing industry, etc.) through the study of granular media. Even if numerical methods provided by nonsmooth mechanics nowadays lead globally to robust and efficient simulations of such systems, a certain number of application cases reach the limits of the state-of-the-art schemes and associated solvers. In particular, it is often necessary to invoke models of dry friction in 3D contact models (e.g. the Signorini-Coulomb contact law), in the presence of nonlinear contact kinematics. Indeed, the nonlinearity of those kinematics can come not only from the curvature of the surfaces in contact, but also from the kinematics of relative motion of the solids, which are often intrinsically nonlinear due to the presence of large rotations. This thesis aims to overcome the current limitations of the numerical methods in this type of situation, by proposing new numerical schemes as well as solvers adapted to application constraints. In this sense, particular attention will be paid to the robustness of the proposed methods (energy behavior, solvability of the constructed algebraic systems, etc.) and to the overall efficiency of the methods (achievable performance levels, possibilities of parallelization, applicability to real-time simulation contexts).

Download the offer (.zip)

See all positions