HAL Portal IOGS (nstitut d'Optique Graduate School)
Not a member yet
12589 research outputs found
Sort by
Propriétés de diffusion des nanotubes de carbone dans des environnements tortueux et biologiques
No full textThe brain’s extracellular space (ECS) is a network of narrow clefts between neurons, glia, and blood vessels, accounting for about one-fifth of brain volume in healthy tissue. This fluid-filled compartment is essential for molecular transport, signalling, and homeostasis. It contains interstitial fluid (ISF) and a structural scaffold provided by the extracellular matrix (ECM). Despite decades of study, a full quantitative description remains elusive. Ensemble measurements of small-ion diffusion have shown that transport in the ECS is slowed two- to threefold compared to free solution. Yet such bulk approaches cannot pinpoint whether this hindrance arises from cellular geometry, ECM organisation, or transient molecular interactions.Single-particle tracking (SPT) offers a way to bridge this gap. By following fluorescent nanoparticles with nanometric precision and video-rate resolution, SPT exposes local heterogeneities invisible to ensemble averaging. A landmark study by Godin et al. introduced single-walled carbon nanotubes (SWCNTs) as near-infrared, photostable probes for SPT in brain tissue, showing that nanoscale trajectories could be recorded deep within scattering tissue. Later studies extended SPT to other tracers and experimental conditions, often contrasting control and pathology. While these provided valuable insights, they remained self-contained, leaving a patchwork of results—sometimes at odds with ensemble measurements and difficult to reconcile across probes.This thesis is motivated by filling this gap. By using novel ultra-short carbon nanotubes (uCCNTs) as tracers better matched to ECS dimensions, and combining them with three-dimensional tracking, it develops an experimental platform tailored to the nanoscale geometry of the ECS. Analytically, it adopts statistical tools from anomalous diffusion studies in membranes and cytoplasm, offering a more rigorous characterisation than conventional MSD fits. These methods allow the contributions of confinement, ECM organisation, and transient interactions to be disentangled.The results show that ECS diffusion is heterogeneous and often anomalous. At short timescales tracer motion is Brownian, but over longer intervals cellular boundaries drive a crossover to non-linear dynamics. Within this landscape, hyaluronan emerges as a key regulator: its depletion tightens local widths and hinders transport, underscoring its role as a buffer that preserves typical ECS dimensions. By integrating these insights, the thesis advances a unified view of extracellular transport that links single-trajectory statistics with structural determinants of the ECS.Beyond local transport, a key question is whether clearance from the ECS relies solely on diffusion or also involves active pathways. This distinction is central to understanding how the brain eliminates metabolites, aggregates, and therapeutic agents, with direct implications for the glymphatic system. To address this, we implemented gradient-index (GRIN) lens–based imaging for in-vivo single-particle tracking, extending nanoscale diffusion measurements from slices into the intact brain. This approach enables visualisation of tracer motion under conditions closer to physiology and provides a first step toward disentangling which mechanisms can drive clearance in the brain ECS.L’espace extracellulaire (ECS) du cerveau est un réseau de fins canaux interstitiels situé entre neurones, cellules gliales et vaisseaux, représentant environ un cinquième du volume cérébral. Rempli de fluide interstitiel et structuré par la matrice extracellulaire (ECM), il assure le transport moléculaire, la signalisation et l’homéostasie. Malgré des décennies d’étude, sa description quantitative reste incomplète. Les mesures d’ensemble de la diffusion d’ions ont montré un ralentissement de l’ordre de deux à trois fois par rapport à la diffusion dans de l’eau. Mais ces approches globales ne permettent pas d’identifier si ce freinage provient de la géométrie cellulaire, de l’organisation de la matrice ou d’interactions transitoires.Le suivi de particules uniques (SPT) apporte une réponse en révélant, avec une précision nanométrique, des hétérogénéités locales invisibles aux méthodes classiques. L’étude fondatrice de Godin et al. a introduit les nanotubes de carbone (SWCNTs) comme sondes photostables dans le proche infrarouge, montrant que des trajectoires nanoscopiques pouvaient être suivies au cœur du tissu cérébral. Depuis, le SPT a été appliqué à divers traceurs et modèles biologiques, souvent en comparant tissus sains et pathologiques. Ces travaux ont fourni des résultats précieux mais fragmentés, parfois contradictoires avec les mesures globales.Cette thèse s’inscrit dans ce contexte en développant une plateforme combinant nanotubes de carbone ultra-courts (uCCNTs), mieux adaptés aux dimensions de l’ECS, suivi tridimensionnel et outils analytiques issus de l’étude de la diffusion anormale. Cette approche permet de dépasser les simples comparaisons descriptives pour relier la structure de la matrice, le confinement géométrique et les interactions transitoires aux propriétés de diffusion mesurées.Les résultats montrent que la diffusion dans l’ECS est hétérogène et souvent anormale. Le mouvement est brownien à court terme mais devient non linéaire à plus longue échelle, sous l’effet des obstacles cellulaires. L’hyaluronane émerge comme un régulateur clé : sa déplétion resserre les canaux et freine le transport, soulignant son rôle de « tampon » qui maintient les dimensions typiques de l’ECS. Ces observations conduisent à une vision unifiée de l’ECS, non pas comme un milieu homogène, mais comme un réseau structuré où géométrie et matrice organisent ensemble le transport.Enfin, la question d’éventuelles voies actives dans le drainage cérébral reste ouverte. Pour l’aborder, nous avons développé une approche d’imagerie in vivo basée sur des lentilles à gradient d’indice (GRIN), permettant de suivre des particules uniques dans le cerveau intact, dans des conditions plus proches de la physiologie. Cette méthode constitue une première étape vers l’identification des mécanismes qui gouvernent le drainage des métabolites, agrégats pathologiques et agents thérapeutiques, au cœur du débat sur l’existence d’un système glymphatique dans le cerveau
Effective properties of resonant nanoparticle suspensions: Impact of the elementary volume shape and eigenmodes analysis
No full textInternational audienceThe study of the effective properties of random particulate materials is a storied subject, which gave rise to several effective-medium theories establishing a link between the microstructure and the macroscopic properties of inhomogeneous materials. This link is typically provided by the effective refractive index. However, effective-medium theories are defective when resonant interactions between the particles take place. Homogenization must then be performed via Maxwell's equations from the rigorous calculation of the electromagnetic fields averaged over a statistical ensemble of realizations of the random medium in a finite domain, i.e., different configurations of the geometrical particle distribution. In even more restricted regimes originating from interparticle coupling, one observes markedly high field intensities at the surface of the particles or in localized regions of the agglomerate, casting doubts on the representativeness of the calculated mean field to extract an effective refractive index. Here we show that the value of the extracted effective refractive index may strongly depend on the shape of the ensemble's configurations, irrespective of the statistical fluctuations of the states. This result is obtained by performing homogenization on systems of particles arranged in the shape of elliptic domains, with different eccentricities. The impact of the boundary effects on the effective refractive index is analyzed through the calculation of the quasinormal modes of the particulate systems, evidencing either strong fluctuations of the eigenfrequencies in the complex plane or Mie-type eigenmodes
A current and LFN method for individually assessing layer quality in III-V multi-junctions
No full textInternational audienc
Continuous scanning full-field OCT for fast volumetric imaging of multi-cellular aggregates
No full textFull-field optical coherence tomography (FF-OCT) offers label-free, high-resolution imaging of biological samples but remains limited by slow acquisition due to piezoelectric mirror modulation. We present a continuous-scanning FF-OCT method that eliminates piezoelectric displacement and synchronization by continuously translating the sample with a motorized stage while recording images on the fly. Depth-resolved information is retrieved via Fourier analysis of the temporal signal at each pixel. This approach enables volumetric imaging over several hundred micrometers within tens of seconds and provides a higher contrast-to-noise ratio than traditional four-phase FF-OCT. Continuous-scanning FF-OCT thus represents a simpler and faster alternative for 3D bio-imaging of living tissues and organoids
Deep sub-wavelength scale focusing of heat flux radiated by magneto-optical nanoemitters in the presence of an external magnetic-field
No full textInternational audienc
Emergence of second-order coherence in the superradiant emission from a free-space atomic ensemble
No full textInternational audienceWe investigate the evolution of the second-order temporal coherence during the emission of a superradiant burst by an elongated cloud of cold Rb atoms in free space. To do so, we measure the two-times intensity correlation function g(2) N (t 1 ,t 2 ) following the pulsed excitation of the cloud. By monitoring g(2) N (t,t) during the burst, we observe the establishment of second-order coherence, and contrast it with the situation where the cloud is initially prepared in a steady state. We compare our findings to the predictions of the Dicke model, using an effective atom number to account for finite size effects, finding that the model reproduces the observed trend at early time. For longer times, we observe a subradiant decay, a feature that goes beyond Dicke's model. Finally, we measure the g(2) N (t 1 ,t 2 ) at different times and observe the appearance of anti-correlations during the burst, that are not present when starting from a steady state.</p
Type-I intermittency route to the chaos in passively Q-switched Tm:YLF laser emitting at 2.3 µm
No full textInternational audienceWe report on an original chaotic dynamic behaviour of a Tm-doped Q-switched laser operating on the 3 H4 → 3 H5 transition emitting at 2.3 µm using a Tm:YLF crystal, with optional co-lasing at 1.9 μm via the ³F₄→³H₆ transition.This study specifically investigates the chaotic and intermittent dynamics originally observed in this laser. Experimental observations reveal an atypical type-I intermittency route to chaos, linked to cascade transitions and characterized using Poincaré maps, entropy calculations, and phase-space reconstructions. The analysis highlights synchronization losses between the Q-switched pulse train and relaxation oscillations in the 1.9 μm cascade laser as a driver of chaotic instabilities. These findings deepen the understanding of chaotic regimes in Tm-doped lasers and provide insights for optimizing the stability of MIR laser systems for advanced photonic applications. Further modeling is needed to fully elucidate the interplay of dual-wavelength dynamics and the observed chaotic behavior.</div
Clustering de résumés LLM guidés par l'utilisateur : vers une approche constructiviste et réaliste unifiée
No full textNational audienceNous introduisons un cadre hybride combinant grands modèles de langage et techniques de regroupement pour extraire, résumer, évaluer et structurer automatiquement les connaissances de larges collections textuelles. Après avoir sélectionné, via une métrique d'entropie sémantique, la stratégie de prompt la plus stable, un LLM génère des résumés modulables qui font l'objet d'une évaluation factuelle assurant leur fiabilité. Ces résumés validés sont ensuite vectorisés, projetés en basse dimension et regroupés en thématiques. Optionnellement, un second LLM affine ensuite leurs libellés pour renforcer l'interprétabilité. Expérimentée sur un corpus majeur d'incidents aériens, cette approche augmente la cohérence et la granularité des clusters thématiques par rapport à une analyse directe des textes, ouvrant de nouvelles perspectives pour la recherche d'information et l'exploration de bases documentaires
Unrolled-SINDy: A Stable Explicit Method for Non linear PDE Discovery from Sparsely Sampled Data
No full textIdentifying from observation data the governing differential equations of a physical dynamics is a key challenge in machine learning. Although approaches based on SINDy have shown great promise in this area, they still fail to address a whole class of real world problems where the data is sparsely sampled in time. In this article, we introduce Unrolled-SINDy, a simple methodology that leverages an unrolling scheme to improve the stability of explicit methods for PDE discovery. By decorrelating the numerical time step size from the sampling rate of the available data, our approach enables the recovery of equation parameters that would not be the minimizers of the original SINDy optimization problem due to large local truncation errors. Our method can be exploited either through an iterative closed-form approach or by a gradient descent scheme. Experiments show the versatility of our method. On both traditional SINDy and state-of-the-art noise-robust iNeuralSINDy, with different numerical schemes (Euler, RK4), our proposed unrolling scheme allows to tackle problems not accessible to non-unrolled methods
Physics-Informed Machine Learning for Modeling CO2 Capture from Scarce Data
No full textInternational audienceAccurate modeling of complex industrial processes often relies on costly mechanistic simulations grounded in physical principles. In this paper, we investigate the subject of CO2 capture, a major environmental challenge, through the absorption column of an amine-based post-combustion process. The modeling of such unit at industrial scale faces two difficulties: (i) theoretical models, efficient at laboratory scale, might fail to fully reflect the complexity of the numerous intertwined phenomenaoccurring in the absorber, (ii) the cost and uncertainty of iindustrial observation data make purely data-driven approaches unfeasible. To tackle both this low data regime and inaccurate physical models, we envision this CO2 capture problem through the lens of Physics-informed Machine Learning (PiML). We present a hybrid (data+knowledge) model where the scarce observation data complement the physical model, while the latter ensures that the predictions remain physically consistent. Beyond the standard use of simulation data for learning and the embedding of physical laws as regularization, the originality of our PiML algorithm compared to other methods in the literature lies in a physicalprior assumption about the network architecture and its countercurrent flow learning process inspired by the column’s operation. Our experimental results showcase a significant improvement in accuracy and highlight the potential of our augmented model for generalizing across domains, especially when data is scarce