226280 research outputs found
Sort by
Thermomigration de microgouttes d'Au-Ge sur Ge(111) : mécanismes atomiques et effets sur la morphologie de surface
No full textInternational audienceThe motion of microdroplets under thermal gradients, also known as thermomigration (TM), plays a critical role in a wide range of surface phenomena and technological applications. In this work, we study the TM behavior of Au-Ge micro-droplets on Ge(111) surfaces and show that the droplet motion leads to the formation of atomically flat terraces, which are essential for thin film epitaxy. Experimental observations are combined with a microscopic model to investigate droplet motion and the formation of flat atomic terraces. Our results show that, at low temperatures, droplet velocity increases with size, but this size dependence diminishes as the temperature rises. The microscopic model we develop is based on thermally activated detachment, diffusion, and attachment of solute (Ge) atoms across the droplet, driven by a temperature-dependent concentration difference between the leading edge and the trailing edge. The model reproduces the experimental trends. Our results provide insights into the fundamental mechanisms of droplet TM and suggest ways to control surface morphology
The Rhône Sediment Observatory (OSR) monitoring network for suspended sediment and contaminants long-term assessment
No full textInternational audienceIntroduction: The Rhône Sediment Observatory (OSR), created in 2009, aims to provide scientific knowledge for promoting a sustainable management of channel forms and sediment processes of this river [1], the largest by mean discharge in France and one of the largest tributaries of the Mediterranean Sea. We propose here a summary of the major developments and findings obtained on the monitoring of suspended sediment and associated priority substances in the Rhône River basin. We also focus on the evolution of the techniques and tools developed and applied for research and diffusion to stakeholders. Methods: Monitoring of concentrations and fluxes of suspended particulate matter (SPM) and contaminantsis based on continuous measurements of water discharge and turbidity/SPM, coupled with integrative sediment traps collected monthly for contaminants analyses. The OSR network includes 12 permanent stations in the Rhône River and major tributaries (Fig. 1, [2]). Different geochemical methods and models have also been developed and tested at various temporal and spatial scales to estimate the sources of SPM and associated contaminants. Results: This comprehensive network has permitted to establish event-related, annual and interannual SPM and contaminants mass budgets across the entire Rhône River basin [3]. Sedimentary export to the Mediterranean was on average 5.21 Mt per year for the period 2009-2023, with very high annual variability(between 1.8 and 12.8 Mt). Nonetheless, on average,the Rhône system has a balanced sediment budget (inputs from tributaries versus outflow to the sea). In contrast, for contaminants (e.g., polychlorobyphenyls - PCBs, aromatic polycyclic hydrocarbons – PAHs, trace metal elements – TMEs, radionuclides), the annual outputs are often lower than the tributary inputs, suggesting that other sources must be takeninto account. The various tributaries show very contrasting levels of contamination according to the contaminants studied, which can be explained by past or present activities on their watersheds. Overall, in the Rhône River, the concentrations of particulate contaminants have generally decreased or remained stable over the past 10 years and over the longer term. Nonetheless, contaminants concentrations remain higher in the Rhône River at the outlet than upstream
Étude comparative de réponses humaines et de grands modèles de langue à des QCM en pharmacie
No full textNational audienceCet article propose d'étudier les réponses générées par plusieurs Grands Modèles de Langue à un ensemble de Questions à Choix Multiple en pharmacie. Ces réponses sont comparées aux réponses données par des étudiants, afin de comprendre quelles sont les questions difficiles pour les modèles par rapport aux humains et pour quelles raisons. Nous utilisons les logits internes des modèles pour construire des distributions de probabilité et analyser les caractéristiques principales qui déterminent la difficulté des questions via une approche statistique. Nous apportons aussi une extension du jeu de données FRENCH MEDMCQA avec des paires question-réponses en pharmacie, enrichies avec les réponses des étudiants, la ponctuation assignée aux réponses, les thématiques cliniques correspondantes et des annotations manuelles sur la structure et certains traits sémantiques des questions
Raffinage des représentations des tokens dans les modèles de langue pré-entraînés avec l’apprentissage contrastif : une étude entre modèles et entre langues
No full textNational audienceLes modèles de langue pré-entraînés ont apporté des avancées significatives dans les représentations contextuelles des phrases et des mots. Cependant, les tâches lexicales restent un défi pour ces représentations en raison des problèmes tels que la faible similarité des representations d'un même mot dans des contextes similaires. Mosolova et al. (2024) ont montré que l'apprentissage contrastif supervisé au niveau des tokens permettait d'améliorer les performances sur les tâches lexicales. Dans cet article, nous étudions la généralisabilité de leurs résultats obtenus en anglais au français, à d'autres modèles de langue et à plusieurs parties du discours. Nous démontrons que cette méthode d'apprentissage contrastif améliore systématiquement la performance sur les tâches de Word-in-Context et surpasse celle des modèles de langage pré-entraînés standards. L'analyse de l'espace des plongements lexicaux montre que l'affinage des modèles rapproche les exemples ayant le même sens et éloigne ceux avec des sens différents, ce qui indique une meilleure discrimination des sens dans l'espace vectoriel final
Connaissances factuelles dans les modèles de langue : robustesse et anomalies face à des variations simples du contexte temporel
No full textNational audienceCe papier explore la robustesse des modèles de langue (ML) face aux variations du contexte temporel dans les connaissances factuelles. Il examine si les ML peuvent associer correctement un contexte temporel à un fait passé valide sur une période de temps délimitée, en leur demandant de différencier les contextes corrects des contextes incorrects. La capacité de distinction des ML est analysée sur deux dimensions : la distance du contexte incorrect par rapport à la période de validité et la granularité du contexte. Pour cela, un jeu de données, TimeStress, est introduit, permettant de tester 18 ML variés. Les résultats révèlent que le meilleur ML n'atteint une distinction parfaite que pour 11% des faits étudiés, avec des erreurs critiques qu'un humain ne ferait pas. Ces travaux soulignent les limites des ML actuels en matière de représentation temporelle
Les défis éthiques de l’intelligence artificielle dans les services de santé : revue systématique de littérature et agenda de recherche
No full textInternational audienceThe deployment of AI technologies in the healthcare is shaking up individuals, organizations and healthcare systems, in particular through major ethical challenges. Based on a systematic review of the literature, this research analyses five specific ethical challenges posed by AI in healthcare and proposes a critical reflection on their tensions leading to a research agenda.Le déploiement des technologies d’IA dans les services de santé bouleverse les individus, les organisations et les systèmes de santé, notamment à travers des défis éthiques majeurs. A partir d’une revue systématique de la littérature, cette recherche analyse cinq défis éthiques propres à l’IA en santé et propose une réflexion critique de leurs tensions débouchant sur des perspectives de recherche
Non-reconnaissance (provisoire ?) par la France de l'Etat palestinien
No full textInternational audienc
PVD coatings on open-cell 3D foams for electrochemical applications: A review
No full textInternational audienceThe deposition of functional coatings by Physical Vapor Deposition (PVD) on open-cell 3D foams represents a burgeoning area within material science, especially for electrochemical applications. Due to the novelty of this field and the unique geometry of the foams, the use of PVD on these substrates is a breakthrough innovation for functional material development. However, several challenges remain, e.g. understanding film growth mechanisms on foams, their impact on electrochemical processes, and optimizing the performance of coated foams across various applications through an understanding of the electrochemical phenomena occurring inside and on the surface of the coated foams. This review provides the first thorough overview of the current state-of-the-art in this area and suggests innovative solutions to the challenges encountered. It reports the various properties of films on foams reported in literature, compares the electrochemical performance of PVD-coated foams for Oxygen Evolution Reaction (OER)/Hydrogen Evolution Reaction (HER) catalysis, and energy storage applications, and discusses the mechanisms that explain their performance. Additionally, the review offers an analysis of existing research and introduces a novel numerical methodology, integrating Direct Simulation Monte Carlo (DSMC), Particle-in-Cell Monte Carlo (PICMC), and kinetic Monte Carlo (kMC) techniques to facilitate the characterization of coatings within the foams
Genome-scale metabolic modeling of Ruminiclostridium cellulolyticum : a microbial cell factory for valorization of lignocellulosic biomass
No full textInternational audienceThe development of sustainable biotechnological processes requires a transition from the traditional fermentation of refined substrates toward the valorization of waste materials such as lignocellulosic biomass. Although these so-called recalcitrant substrates cannot be degraded by model industrial organisms, they can be degraded by microbial consortia through a process of anaerobic digestion, where different community members are able to break down polysaccharides of varied complexity. Among these microbes, Ruminiclostridium cellulolyticum stands out as a promising candidate for fermentation of lignocellulose due to its ability to degrade both cellulose and hemicellulose. In this work, we present an updated genome-scale metabolic model for R. cellulolyticum strain H10. The model was manually curated with experimental data, and the pathways for degradation of cellulose and hemicellulose (arabinoxylan and xyloglucan) were reconstructed and annotated with full detail. The model enables the simulation of the fermentation profile of lignocellulosic materials of various compositions, facilitating the use of this organism as a potential workhorse for sustainable biotechnology, and it provides a valuable template for the reconstruction and optimization of lignocellulose degradation pathways in related organisms. IMPORTANCE In this work, we present a manually curated genome-scale metabolic model for Ruminiclostridium cellulolyticum , one of the few species known to fully degrade cellulose and hemicellulose. The model was extensively curated with experimental data obtained from the literature, covering approximately 25 years of research on this organism. We use this model to simulate the fermentation of mixed lignocellulosic polysaccharides and observe a good agreement with experimental data. This organism is therefore a promising microbial cell factory for sustainable transformation of lignocellulosic residues into valuable industrial products
Synergies between cellulosomal and non-cellulosomal bacterial cellulases complexed in designer cellulosomes
No full textInternational audienceThe development of cellulases cocktails to efficiently solubilize cellulose for biotechnological applications is a constant demand. Cellulases exhibit diverse mode of action, operating in the free state or in multi-enzyme complexes (cellulosomes), and potentially acting synergistically for a higher efficiency. The efficacy of three non-cellulosomal bacterial cellulases originating from three different species that naturally act in the free state: Cel9A from Lachnoclostridium phytofermentans, Cel5H from Saccharophagus degradans and two variants of Cel5I from Ruminiclostridium cellulolyticum, were evaluated in bifunctional or trifunctional designer cellulosomes, and in combination with the most efficient pair of cellulases reported in R. cellulolyticum, Cel48F and Cel9G.The activity of the newly tested cellulases in designer cellulosomes, reached a stimulation factor of 1.25. Cel5H and Cel5I can act in a cooperative manner with other cellulases. Low synergy was observed when Cel9A was included in the complexes with the pair Cel48F and Cel9G. However, all the complexes containing Cel9A exhibited the highest activities. Finally, the most efficient enzymatic combination is a bifunctional designer cellulosome comprising Cel48F and Cel9G, in association with free Cel9A. Interestingly this mixture releases high levels of glucose and cellobiose, making this association an especially attractive option for biotechnological conversion of cellulose to chemicals