Evaluation of new low-damage GaN etching strategies for the realization of high performance MicroLED
For optical communication and display applications (display, AR/VR), the need for micro-displays with high resolution µLEDs GaN is becoming increasingly important. The formation of the pixel constitutes the critical brick because it generates defects on the sides of the µLED, which limit the emission performance of the device. These degradations have even more impacts on electro-optical performance for aggressive LED pitches.
The subject proposes a comprehensive approach, ranging from the development of new GaN etching strategies limiting damage to the pixel edges, to its characterization through dedicated direct means, and validated on a complete device by an electro-optical test vehicle.
Backpropagation-Free Langevin Sampling for Bayesian Learning on ReRAM Architectures
Bayesian neural networks allow artificial intelligence systems to account for uncertainty in their predictions, but training them requires substantial computational resources. Resistive memories (ReRAM), which can both store data and perform computations directly within electronic circuits, offer a promising solution to reduce this cost.
This PhD project aims to develop a new Bayesian learning method on ReRAM hardware, based on Langevin sampling. This approach efficiently explores neural network parameters by combining random perturbations with gradient estimates. The objective is to exploit the intrinsic physical variability of ReRAM devices to estimate these gradients without relying on conventional backpropagation. The possibility of reusing memory programming attempts will also be investigated to reduce energy consumption and device wear.
The work will combine algorithm development, simulations, and experiments on existing ReRAM circuits at CEA-Leti. Expected outcomes include a demonstration of Bayesian learning on real hardware and an assessment of its performance, accuracy, and energy efficiency compared with existing approaches.
Bayesian Calibration of Multiphysics Models and Inverse Design of Neuromorphic Devices Assisted by Machine Learning
Electrochemical neuromorphic devices, such as ionic synaptic transistors, offer promising opportunities for the development of energy-efficient, brain-inspired computing architectures. However, their design and optimisation remain challenging due to the complex coupling between ion transport, electrochemical processes and electrostatics. Current approaches largely rely on iterative fabrication and experimental characterisation, limiting the exploration of the design space and the identification of optimal device configurations.
This PhD project aims to develop a predictive, physics-informed digital-twin framework to accelerate the modelling and inverse design of electrochemical neuromorphic devices. The proposed approach combines multiphysics modelling, Bayesian inference and machine learning to predict device behaviour, quantify uncertainties and explore device architectures in silico, thereby reducing reliance on trial-and-error experimental approaches.
Starting from an existing Bayesian inference pipeline and neural-network-based surrogate model, the project will investigate advanced learning strategies to efficiently approximate the posterior probability distribution of physical and geometrical model parameters conditioned on experimental measurements and/or target device performances. Particular emphasis will be placed on physics-informed learning approaches combining governing PDE constraints with sparse, punctual experimental observations, inspired by Physics-Informed Neural Operators (PINO). Neural operators, conditional generative models (including diffusion models) and other state-of-the-art probabilistic learning strategies will be investigated to achieve accurate and computationally efficient posterior approximation, uncertainty quantification and inverse design.
Beyond electrochemical neuromorphic devices, the developed framework could be extended to a broader class of nonlinear dielectric systems, including ferroelectric and antiferroelectric devices. These systems involve strongly nonlinear constitutive relationships and complex coupled equations, often making conventional numerical modelling and parameter identification particularly challenging. They therefore represent a promising application for the proposed physics-informed Bayesian and machine-learning methodology.
AI-Driven Multiscale Simulation of Ferroelectric Switching in Hafnium Oxides
What happens at the atomic scale when a bit of information is written in a ferroelectric memory, and how does it determine the behaviour of a real device? This PhD will address these questions by combining machine learning, large-scale atomistic simulations and device-scale modelling.
Ferroelectric Hafnium Zirconium Oxide (HZO) has become a key material for ultra-low-power non-volatile memories such as FeRAM and ferroelectric transistors (FeFETs). It is CMOS-compatible and remains ferroelectric in films only a few nanometres thick. Yet its ferroelectric behaviour is still only partly understood. It depends strongly on composition, defects, microstructure and interfaces, and it is governed by switching mechanisms at the nanoscale. Density functional theory (DFT) is highly accurate but limited to a few hundred atoms, far too few to capture these mechanisms. Machine-learning interatomic potentials, trained on DFT calculations, now overcome this limit. They can simulate the response of materials to an electric field with near-DFT accuracy on systems of up to a million atoms.
Building on our in-house machine-learning framework, you will first run large-scale molecular dynamics simulations of ferroelectric switching in HZO films and in complete TiN/HZO/TiN capacitors under applied voltage. You will study domain nucleation, domain-wall motion and the role of oxygen vacancies and electrode interfaces. You will then use these results to parameterize a device-scale model of the full memory cell, in order to interpret electrical measurements on FeRAM devices fabricated at CEA-Leti. You will work closely with the CEA-Leti teams in charge of FeRAM integration and advanced characterization.
Multiscale 3D Characterization of Nanodevices Using EDX Tomography and Electron Ptychography
The increasing complexity of nanodevices requires multiscale 3D characterization techniques spanning from the nanometer to the atomic scale. STEM-EDX tomography provides three-dimensional elemental mapping at the nanometer scale. Recent deep learning-based approaches have achieved near-isotropic spatial resolution of approximately 5 nm in specimens 60–100 nm thick. Electron ptychography offers complementary capabilities for atomic-resolution structural characterization. This PhD aims to develop a correlative methodology combining both techniques on the same specimen. STEM-EDX tomography will first be performed on relatively thick TEM lamellae. The specimens will subsequently be thinned to a target thickness below 30 nm for 4D-STEM acquisition. Multislice electron ptychography will provide depth-resolved structural information. Simulations will be used to optimize experimental conditions and acquisition parameters. EDX spectral processing and limited-angle tomographic reconstruction methods will be optimized. Ptychographic reconstruction algorithms will be evaluated in terms of resolution, robustness, and sensitivity to light elements. A major challenge will be the spatial registration of chemical and structural volumes before and after specimen thinning. Reconstruction accuracy and limitations will be assessed using simulations and reference specimens. The methodology will be applied to phase-change memory (PCM) devices and InGaN-based nano-emitters. The ultimate goal is to establish a correlative 3D characterization approach linking chemical composition, atomic structure, and nanodevice properties.
Cryogenic exploration of 1/f and Random Telegraph noise effects in advanced transistors used in Quantum System control/readout Electronics
With the CMOS transistor downscaling, low-frequency noise (LFN) or “1/f noise” due to charge trapping/ detrapping in gate dielectric interface defects has become increasingly important, as its amplitude is inversely proportional to the gate area, thus limiting the performance of mainly analog and mixed signal circuits. For nano-scale transistors, another type of fluctuation appears, called "Random Telegraph Noise" (RTN), due to the activity of only a handful of traps. RTN consists of sudden, high-amplitude transitions of the current, which can jeopardize the stability of memory cells and the precision of imager circuits. Counterintuitively, not only both LFN and RTN present at deep cryogenic conditions, but in certain cases even more prevalent than at room temperature[4]. This is an enormous obstacle for quantum computers, because cryo-CMOS circuits need to be physically very close to the qubit devices in order to perform the tasks of control, readout, and signal amplification. While technological advancements could potentially improve function below 10 K and accurate IC design could mitigate the read/write or readout amplification issues, a significant effort on the physical understanding of the trapping/detrapping cryogenic phenomena and a precise mathematical description is required, not only for noise but also for aging effects such as Bias Temperature Instabilities (BTI).
This thesis aims to provide a deep understanding of trap-related noise effects at cryogenic temperatures down to 4 K through thorough experimental studies, while also developing physics-based compact noise models for cryo-circuit design. Recent interpretation approaches will be evaluated, and new methods of noise parameter extraction will be elaborated. In terms of technologies, studied devices will be FD-SOI MOSFETs 28nm node from ST and used for Q-LOOP project (IRT - UE project.
Development and Application of Time-Resolved Cathodoluminescence for Nanoscale Characterization of MicroLEDs
Context: MicroLEDs based on III-nitride semiconductors are promising for display and optoelectronic technologies. Their performance depends on carrier recombination and transport at the nanoscale, particularly near sidewalls, interfaces and defects. Time-Resolved Cathodoluminescence (TRCL) combines electron microscopy spatial resolution with time- and spectrally-resolved luminescence, enabling investigation of local carrier dynamics with nanometric spatial and sub-nanosecond temporal resolution. The thesis aims to develop TRCL methods for III-nitride microLEDs.
Work description: The candidate will develop TRCL measurements: optimize electron-beam energy, current and temperature; investigate carrier dynamics and the influence of sidewalls, interfaces and defects; process multidimensional TRCL datasets. The work will be performed within the PFNC platform at CEA-Leti, in interaction with researchers working on microLEDs.
Required skills: Strong background in solid-state physics and motivation to learn characterization and data analysis. Knowledge of luminescence is appreciated. The candidate will develop skills through academic collaborations and interaction with LETI partners.
Development and characterization of bio-based photoresists for lithography applications
Context :
CEA-Leti designs the microelectronic components of tomorrow. This involves developing new materials for devices, as well as essential materials for integrated circuit manufacturing, such as photoresists. These innovations also aim to reduce environmental impact by creating products without PFAS or toxic substances. In this context, the Advanced Lithography Processes Laboratory (LALP) collaborates with the Materials for Sensors and Delivery Laboratory (LMCD) at CEA to co-develop new bio-based lithography resists containing methacrylate functions. The goal of the PhD is to study the performances of these new photoresists in lithography and to evaluated their environmental and economical impact. To do so, an emphasis will be made on the influence of the chemical synthesis on the optical properties of the resists, their polymerization mechanisms, and the mechanical characteristics of the resists after exposure. Finally, tools will be given to conduct life cycle analysis on the different products.
Required skills :
The study will examine the entire photolithography process: from the synthesis of the base product and its formulation, to the final structuring. An initial study will analyze the rheological properties of the material after synthesis. Next, its behavior will be evaluated during the spin-coating step (ellipsometry and TGA/DSC). An analysis will focus on the product's sensitivity to a wavelength of 365 nm. Characterization of the product will be carried out by examining the shape of the obtained patterns (SEM) and the physical properties of the resist (ellipso-porosimetry and AFM). Etching tests will also be conducted.
Study of cryogenic etch of metals for he realization of high aspect ratio devices
An increasing number of devices require high-aspect-ratio etching to increase the effective surface area while maintaining a small 2D footprint—3D memory architectures are a prime example. In the case of 3D FeRAM or FeFET devices, it is often necessary to etch these deep structures directly into metals such as W or TiN; however, such etch processes are typically constrained by poor mask selectivity, low etch rates at small critical dimensions (CD), and distorted profiles. Cryogenic etch offers the potential for innovative improvements in these areas—as it was demonstrated for silicon and dielectrics like SiO2/SiN—and investigating new, low-GWP (global warming potential) chemistries (e.g., Cl2, BCl3, F2) that have rarely been explored for cryoetch presents a compelling scientific opportunity
Chemical Engineering of Active Surfaces in II–V Material Technologies for Cooled Infrared Detection
Voici la traduction anglaise complète, avec un vocabulaire scientifique adapté à un résumé de thèse.
Écriture
This PhD thesis builds on previous research aimed at improving the passivation of HgCdTe photodiodes, particularly those based on P-i-N architectures. It seeks to understand the physicochemical mechanisms responsible for the degradation of interfaces between the semiconductor material and the passivation layer. Previous studies have highlighted the influence of residual contaminants, surface oxidation, and excess tellurium on the electrical properties of devices.
The main objective is to develop surface engineering processes that enable better control of the chemical composition and defects at these interfaces. Various chemical and electrochemical treatments will be investigated to limit oxidation, reduce contamination, and improve surface stability. Characterization will rely on advanced techniques such as X-ray Photoelectron Spectroscopy (XPS), Hard X-ray Photoelectron Spectroscopy (HAXPES), and Time-of-Flight Secondary Ion Mass Spectrometry (ToF-SIMS), complemented by electrical and electrochemical measurements. Particular attention will be paid to establishing quantitative correlations between the chemical composition of interfaces, their electrochemical properties, and their electrical behavior. These results will help identify the key parameters governing photodiode performance and optimize passivation treatments. The developed processes will then be evaluated on structures representative of P-i-N photodiodes, particularly with a view to operation at elevated temperatures. Finally, this thesis incorporates an eco-innovation component aimed at making manufacturing processes more environmentally friendly. The development of in-situ electrochemical treatments and new, so-called “green” chemistries could make it possible to replace certain conventional bromine-based solutions. Ultimately, this research will contribute to the development of higher-performance, more stable infrared detectors that require less demanding cooling systems, while potentially reducing the environmental impact of manufacturing processes.