HAL Portal IOGS (nstitut d'Optique Graduate School)
Not a member yet
12589 research outputs found
Sort by
Hybrid Fabry-Perot cavity for multispectral analysis: an experimental demonstration
No full textInternational audienceMultispectral sensors such as Ambient Light sensors (ALS) are becoming increasingly popular due to growing concerns for health, environment, and safety. These sensors provide non-intrusive quantitative and qualitative information on the photonic footprints of interest. To meet the demand for mass production, CMOS image sensors (CIS) are a good basis for these devices. Commercial hyperspectral cameras use a generalization of Bayer-like matrix of Fabry-Perot cavities (FPC) as multispectral filters embedded onto a CIS. However, the delicate fabrication of these filters is tedious and leads to a pronounced surface topology. In this study, we demonstrate experimentally that multispectral sensing can be achieved using a hybrid FPC (h-FPC) which is an improved version of the regular FP cavity, that consists of two silicon mirrors, SiO2 spacer, and a sub-wavelength silicon grating at the center of the cavity. These structures were fabricated using a CMOS compatible process and can be integrated into an imager process flow. The h-FPC optical response can be tuned in the near-infrared region (750-950nm) by changing the filling factor of the grating inside the cavity without varying its height, unlike planar FPC. This feature makes the hybrid FPC a more versatile and efficient option for agile multispectral sensing
Electrically injected metamaterial grating DFB laser exploiting an ultra-high Q electromagnetic Induced Transparency resonance for spectral selection
No full textInternational audienceAbstract The study shows that, in waveguide (WG) configuration, the coupling of a 2D plasmonic metamaterial grating (MMG) having a conventional Bragg period along the guide but a distinct period along the transverse axis can lead to Electromagnetically‐Induced‐Transparency (EIT) behavior. This epitomizes that metamaterials, as functional photonic building blocks, can lead to low losses in many standard devices if properly designed. The study reported the observation in passive WGs of a marked Fano‐type EIT resonance with record high‐quality factor: Q = 5000 and contrast >20 dB. Unlike any standard metal grating, MMG‐assisted waveguides exhibit strong grating coupling strength and low‐loss properties simultaneously. This concept is further applied to demonstrate single‐frequency‐emission electrically‐injected Distributed Feedback (DFB) lasers in the near‐infrared telecom domain. The key point is that laser emission occurs at the peak of EIT, i.e., the maximum in transmission. It therefore addresses one of the main critical issues of DFB lasers related to the single frequency yield. The laser performances are state‐of‐the‐art (I th < 20 mA, P max > 20 mW at I = 200 mA, SMSR > 50 dB, optical feedback tolerance >−21 dB compliant with IEEE 802.3 standard). The presented approach, compatible with existing industrial technologies, is promising for real‐life telecom applications
Propriétés optiques fondamentales de nanocristaux de semi-conducteurs individuels aux températures cryogéniques
No full textSemiconductor nanocrystals exhibit outstanding optical and electronic properties due to the quantum confinement of their charge carriers, making them valuable for various applications in optoelectronics, light-emitting devices, and spin-based technologies. Understanding the physics of the band-edge exciton, whose recombination is at the origin of their photoluminescence, is crucial for developing these applications. This thesis focuses on the experimental study of the optical properties of indium phosphide and lead halide perovskites nanocrystals. Using magneto-photoluminescence spectroscopy onsingle nanocrystals at low temperatures, we reveal spectral fingerprints highly sensitive to nanocrystal morphologies and elucidate the entire band-edge exciton fine structure and charge-complex binding energies. In InP/ZnS/ZnSe nanocrystals, the evolution of photoluminescence spectra and decays under magnetic fields show evidence for a ground dark exciton level lying less than a millielectronvolt below the bright exciton triplet, findings supported by a model accounting for the shape anisotropy of the InPcore. In lead halide perovskites, we demonstrate that the ground exciton state is dark and lies several millielectronvolts below the lowest bright exciton sublevels, settling the debate on the bright-dark exciton level ordering in these materials. Combining our results with spectroscopic measurements on various perovskite nanocrystal compounds, we establish universal scaling laws relating exciton fine structure splitting, trion and biexciton binding energies to the band-edge exciton energy in lead-halide perovskitenanostructures, regardless of their chemical composition. Lastly, preliminary spectroscopy analyses on perovskite nanorods with a high aspect ratio suggest their potential as candidates for quantum light emitters due to their characteristic single emission line.Les nanocristaux de semi-conducteurs présentent des propriétés optiques et électroniques remarquables en raison du confinement quantique de leurs porteurs de charge, ce qui les rend avantageux pour diverses applications en optoélectronique, dans les dispositifs émetteurs de lumière et dans les technologies basées sur le spin. La compréhension de la physique de l’exciton de bord de bande, dont la recombinaison est à l’origine de leur photoluminescence, est cruciale pour le développement de ces applications. Cette thèse porte sur l’étude expérimentale des propriétés optiques des nanocristaux de phosphure d’indium et de pérovskites d’halogénure de plomb. En utilisant une méthode de spectroscopie de magnéto-photoluminescence sur des nanocristaux uniques à basse température, nous révélons des empreintes spectrales très sensibles à la morphologie des nanocristaux et élucidons la structure fine de l’exciton de bord de bande et les énergies de liaison des complexes de charge. Dans les nanocristauxd’InP/ZnS/ZnSe, l’évolution des spectres et des déclins de luminescence sous champ magnétique montrent l’existence d’un niveau d’exciton noir situé à moins d’un millielectronvolt en dessous du triplet brillant de l’exciton, résultats étayés par un modèle tenant compte de l’anisotropie de forme du coeur d’InP. Dans les pérovskites d’halogénure de plomb, nous démontrons que l’état fondamental de l’exciton est noir et se situe plusieurs millielectronvolts en dessous des sous-niveaux d’exciton brillants les plus bas, résolvant ainsi le débat sur l’ordre des niveaux brillants et noirs de l’exciton dans ces matériaux. En combinant nos résultats avec des mesures spectroscopiques sur divers composés de nanocristaux de pérovskite, nous établissons des lois d’échelle universelles qui relient l’éclatement de la structure fine de l’exciton et les énergies de liaison du trion et du biexciton à l’énergie de l’exciton de bord de bande dans les nanostructures de pérovskite d’halogénure de plomb, quelle que soit leur composition chimique. Enfin, des analyses préliminaires de spectroscopie sur des nano-bâtonnets de pérovskite avec un grand rapport d’aspect suggèrent leur potentiel en tant qu’émetteurs de lumière quantique grâce à leur émission composée d’une raie unique
Thermo-temporal physisorption in metal–organic frameworks probed by cyclic thermo-ellipsometry
No full textInternational audienceCyclic thermo-ellipsometry enables exploration of the effects of temperature variation rates on vapor physisorption in MOFs, a key insight for many adsorption-driven applications
Calcul quantique avec des atomes de Rydberg : contrôle et modélisation pour simulation et algorithmes quantiques
No full textRefining our understanding of an unknown system through modelling lays the groundwork for optimally controlling it and opens the door to a myriad of potential applications, exploiting the once enigmatic and unpredictable effects of this now-known system. This thesis applies this paradigm to analog quantum computing with Rydberg atoms, showcasing how careful noise modelling, optimal control and machine learning frameworks can support and enhance the simulation of quantum magnetism and the solving of graph-based optimisation and classification problems. After describing the experimental platform enabling the control of Rydberg atoms, we introduce classical tools such as digital twins of noisy systems, tensor network modelling, robust optimal control, and Bayesian optimisation for variational algorithms. We apply the latter to several applications. We improve the preparation of antiferromagnetic state in the Ising model and benchmark the noisy behaviour of a dipolar XY quantum simulator when probing continuous symmetry breaking and performing quantum state tomography. Using optimisation techniques and machine learning methods, we also tackle industrial use cases such as maximum independent set on graphs representing smart charging tasks, binary classification of toxic or harmless molecular compounds, and prediction of fallen angel companies in financial risk management.Améliorer sa compréhension d'un système en le modélisant permet d'espérer le contrôler de manière plus optimale et ouvre la voie à une myriade d'applications potentielles, exploitant les effets jusqu'alors énigmatiques de ce système désormais familier. Cette thèse applique ce paradigme au calcul quantique analogique avec des atomes de Rydberg, montrant comment à l'aide d'une modélisation minutieuse du bruit, de protocoles de contrôle optimaux et de techniques d'apprentissage automatique, on peut espérer améliorer des expériences de simulation de magnétisme quantique ou la résolution de problèmes d'optimisation et de classification de graphes. Après avoir décrit la plateforme expérimentale permettant de contrôler les atomes de Rydberg, nous introduisons des outils classiques tels que les jumeaux numériques de systèmes enclins à des erreurs, la modélisation d'un grand nombres d'atomes par réseaux de tenseurs, le contrôle optimal robuste et l'optimisation bayésienne pour les algorithmes variationnels. Nous appliquons ces outils à plusieurs applications prometteuses. Nous améliorons la préparation d'états antiferromagnétiques dans le modèle d'Ising et réalisons une évaluation détaillée de l'influence d'erreurs sur l'étude de phases magnétiques du modèle dipolaire XY et lors de la tomographie d'états quantiques. En utilisant des techniques d'optimisation et des méthodes d'apprentissage automatique, nous abordons également des cas d'usage industriels tels que la résolution du problème de stable maximum sur des graphes représentant des tâches de planification de charge de batteries de voitures électriques, la classification de composés moléculaires toxiques ou inoffensifs, et des tâches de prédiction dans la gestion des risques financiers
HTAG-eNN: Hardening Technique with AND Gates for Embedded Neural Networks
No full textInternational audienceEmbedded Neural Networks (NNs) face significant challenges due to Single-Event Upsets (SEUs), compromising their reliabil- ity. To address this challenge, previous works study SEU layers sensitivity of AI models. Contrary to these techniques, remaining at high level, we propose a more accurate analysis, highlighting that, except for the last layer, faults transitioning from 0 to 1 sig- nificantly impact classification outcomes. Based on this specific behavior, we propose a simple hardware block able to detect and mitigate the SEU impact. Obtained results show that HTAG protec- tion efficiency is near 96.85% for the LeNet-5 CNN inference model, suitable for an embedded system. This result can be improved with other protection methods for the classification layer. Additionally, it significantly reduces area overhead and critical path compared to existing approaches
Étude expérimentale et théorique des effets photo-thermiques ultra-rapides dans des réseaux de nanoparticules - application au contrôle local de la chimie de surface
No full textThe excitation of metal nanoparticles through short pulses of lightinduces localized photo-thermal effects capable of altering their surface chemistry. Thisresearch aims to investigate and harness these effects for the precise manipulation ofmolecule distribution on nanoparticles at a local level. Initially, employing both pumpprobespectroscopy measurements and a thermo-optical numerical model with minimalfree parameters, we outlined the heterogeneous nature of photo-thermal effects withinasymmetric cross-shaped nanostructures. Then, a methodology was developed tospecifically label the surface chemistry using silica nanoparticles. This labelingrevealed the localized degradation of molecules on the nanostructure's surface exposedto very short pulses. Under low-power illumination, only molecules within highelectricfield zones undergo degradation, enabling the experimental delineation andvisualization of electric field intensity distribution on the structure surfaces with a fewtens of nanometers resolution. These findings pave the way for the development ofplasmonic sensors optimized for the detection of molecules at very low concentrations.L'excitation de nanoparticules métalliques par des impulsions delumière ultra-brèves génère des effets photo-thermiques localisés capables d'altérer leurchimie de surface. Ce travail de recherche a pour objectif d'étudier et d'utiliser ceseffets dans le but de contrôler localement la distribution de molécules d'intérêt sur lesnanoparticules. Dans un premier temps, l'utilisation conjointe de mesures despectroscopie pompe-sonde et d'un modèle numérique thermo-optique ayant un nombreminimum de paramètres libres ont permis de mettre en évidence l'hétérogénéité deseffets photo-thermiques se produisant dans des nanostructures en forme de croixasymétriques.Par la suite, un protocole permettant de marquer spécifiquement la chimiede surface avec des nanoparticules de silice a été développé. Celui-ci a permis demettre en lumière la dégradation locale de molécules à la surface de nanostructuresilluminées par des impulsions très brèves. Pour une illumination à faible puissance,seules les molécules dans les zones de forts champs électriques sont dégradées. Ainsi,cela permet de marquer et de visualiser expérimentalement la distribution de ce dernierà la surface des structures avec une résolution de quelques dizaines de nanomètres. Cesrésultats ouvrent la voie au développement de capteurs plasmoniques optimisés pour ladétection de molécules en très faible concentration
Heterogeneous SplitFed: Federated Learning with Trainable and Untrainable Clients
No full textInternational audienceWith the advent of edge computing and distributed learning paradigms, the integration of low-resource devices and embedded systems within these frameworks has become a focal point of numerous research initiatives. These resource constraints, particularly in memory and computational capacity, are exacerbated by the demands of increasingly complex neural network models that are deployed on such devices. SplitFed Learning (SFL) has been proposed as an innovative approach that combines two prominent distributed machine learning strategies, namely federated learning (FL) and split learning (SL), which facilitates model usage among clients with resource limitations while preserving their privacy. However, there are specific devices where training is difficult or impossible, which SFL, FL, or SL do not consider. For instance, devices based on Field Programmable Gate Arrays (FPGAs) may face such challenges. Despite this, these devices could still benefit from the federation and contribute to it with their own data. Therefore, in this paper, we introduce a new federated learning approach for deep neural networks, called Heterogeneous SplitFed Learning (HSFL), designed to support low-resource clients that are only capable of performing model inference and that can cope with heterogeneous data, thus enabling their active participation. This enhances privacy while improving model performance in collaboration with clients owning more substantial computational resources. We demonstrate empirically on image classification benchmarks and common deep learning models that HSFL can match the performance of other FL approaches that accommodate heterogeneous data, maintaining efficacy and including clients with limited resources.</div