ED Sciences Chimiques
Dynamics of hydrogen interaction on tungsten surfaces: quantum or classical?
by Laura VIAUD (Institut des Sciences Moléculaires)
The defense will take place at 10h00 - salle de conférence 3e EST Institut des Sciences Moléculaires - UMR 5255 Groupe de Chimie Théorique (A12, 3e étage EST) 351 cours de la libération, 33405 Talence cedex
in front of the jury composed of
- Jean-Christophe TREMBLAY - Professeur - Université de Lorraine - Rapporteur
- Jiri VANICEK - Professeur - École Polytechnique Fédérale de Lausanne - Rapporteur
- Corine MATHONIèRE - Professeure des universités - Université de Bordeaux - Examinateur
- Oihana GALPARSORO - Docteure - Universidad del País Vasco - Euskal Herriko Unibertsitatea - Examinateur
Because of its central role in many natural and technological processes, surface reactivity has been the subject of extensive experimental and theoretical studies for several decades. In particular, the interaction dynamics of hydrogen with tungsten surfaces is of strategic interest in the context of plasma-wall interactions in nuclear fusion, where hydrogen and its isotopes are key plasma constituents, and tungsten is one of the preferred materials for the inner wall coatings of reactors. Understanding collisions between plasma species and wall materials is therefore essential to predict and comprehend the behavior of future fusion reactors. In this context, this doctoral thesis proposes a thorough theoretical study of the collision dynamics between dihydrogen (H₂) and tungsten surfaces W(100) and W(110), using molecular dynamics simulations. These simulations rely on multidimensional potential energy surfaces (PES) constructed by interpolating energies obtained via density functional theory (DFT). Two approaches are used: one classical, fast but less precise, and the other quantum, more accurate but computationally costly. One of the main challenges of this work is to reach a precision close to quantum simulations while maintaining the efficiency of classical methods. To this end, semi-classical corrections were developed to introduce purely quantum effects into classical trajectories. The thesis work consisted in implementing, adapting, and refining two types of corrections: a Gaussian weighting, which introduces statistical weights aiming to reproduce Bohr's quantization concept, and an adiabaticity correction, allowing the appropriate treatment of potentially diffracted trajectories. These tools help bridge the gap between classical predictions and quantum results at a much more reasonable computational cost. This framework leads to a more realistic description of interactions between the incident molecule and the metallic surface. The effectiveness of these corrections was evaluated through the calculation of key physical quantities such as sticking and reflection probabilities. To test the robustness and transferability of the method, a third system, H₂/Pd(111), was also studied. The results show a significant improvement in the quantitative agreement between the corrected classical simulations and the quantum references for all considered systems. These works open promising perspectives for developing predictive numerical tools in modeling surface reactivity processes, notably in contexts where quantum effects cannot be neglected.
ED Sciences de la Vie et de la Santé
Role of Arkypallidal neurons in motor symptoms of Parkinson's Disease and L-DOPA-induced-dyskinesia : study in a rodent model
by Léa BONAMY (Institut des Maladies Neurodégénératives)
The defense will take place at 14h30 - Auditorium Centre Broca Nouvelle Aquitaine 146 Rue Léo Saignat, 33000 Bordeaux
in front of the jury composed of
- Jérôme BAUFRETON - Directeur de recherche - Université de Bordeaux - Directeur de these
- Daniela POPA - Directrice de recherche - Institut de Biologie de l'Ecole Normale Supérieure - Rapporteur
- Abdel-Mouttalib OUAGAZZAL - Chargé de recherche - Université Aix-Marseille - Rapporteur
- Patricia BONNAVION - Chargée de recherche - Université Libre de Bruxelles - Examinateur
- Catherine LE MOINE - Directrice de recherche - Université de Bordeaux - Examinateur
The external Globus Pallidus (GPe) is a central hub within the basal ganglia, mainly involved in relaying information through the indirect pathway. It consists primarily of GABAergic neurons, divided into prototypic and arkypallidal (ARKY) subtypes, distinguished by molecular and electrophysiological signatures. Our team recently showed that optogenetic activation of ARKY neurons suppresses ongoing action, suggesting a key role in movement control. Parkinson's disease (PD), the second most common neurodegenerative disorder worldwide, is characterized by severe motor impairments. While dopamine replacement therapy with L-DOPA initially alleviates symptoms, its long-term use often results in involuntary movements known as L-DOPA-induced dyskinesias (LIDs). Despite extensive research on PD and LIDs, the role of ARKY neurons in these conditions remains poorly understood. Therefore, the objective of my PhD has been to investigate functional modifications of ARKY neurons in mice models of PD and LID using multidisciplinary approaches. My work revealed that ARKY neuron's activity is altered in PD and LID behaving mice. I demonstrated that ARKY intrinsic excitability is modified in PD and LID and that ARKY neurons projections to the striatum are enhanced in both pathophysiological conditions. Furthermore, I showed that ARKY neurons are inhibited by L-DOPA in acute brain slices suggesting that L-DOPA exerts its dyskinetic effects partly by acting in the GPe.
ED Sociétés, Politique, Santé Publique
Designing writing supports in french as a second language by considering the concept of attention: the effect of three didactic devices on learners' performance
by Maëlle OCHOA (Laboratoire Cultures, Education, Sociétés)
The defense will take place at 14h00 - Amphi E (à confirmer) 3ter Pl. de la Victoire, Bâtiment E, 33000 Bordeaux
in front of the jury composed of
- Stéphanie ROUSSEL - Professeur des universités - Université de Bordeaux - Directeur de these
- Caroline SCHEEPERS - Professeur - Université Catholique Louvain Saint-Louis Bruxelles - Rapporteur
- Nathalie SPANGHERO-GAILLARD - Professeur des universités - Université Toulouse - Jean Jaurès - Rapporteur
- Maurice NIWESE - Professeur des universités - Sorbonne Université - Examinateur
- André TRICOT - Professeur des universités - Université Paul Valéry Montepellier - Examinateur
- Jérémie SéROR - Full professor - Université d'Ottawa - Examinateur
In both first and second languages, writing is not only ‘a powerful cognitive, identity and epistemic transformer' (Scheepers, 2021, p. 20), but also a key to academic success and social integration in general (Faure, 2011; Scheepers, 2021). These issues, to which political questions may also be added (Petit & Bouchardon, 2017), are encouraging didactics specialists to take an interest in the acquisition of writing skills. This thesis is part of this dynamic. We are looking at the psycho-cognitive aspects of teaching and learning to write, and in particular at the concept of ‘attention', from the point of view of didactics, with the aim of designing writing supports for French as a second language. Writing supports cover, in our view, several situations: the use of a writing strategy by the writer (Graham & Harris, 2019), the implementation of guidance or intervention by means of a didactic arrangement (Alfonso et al., 2016, for example), and the integration of external tools, digital artefacts, into writing environments (Mangenot, 1997). In this work, we test three configurations that correspond to these three types of writing aids. The first involves considering the language switching as a writing strategy; the second is concerned with the study of text models as a guide or didactic intervention; and the third aims to understand better the attentional, writing and didactic issues involved in the use of a digital tool, the neural machine translator, for second-language writing. The experiments we are setting up are part of a complex didactic approach, which involves bringing together concepts from different fields. They are also part of a wider research and development approach, which could aim to bring the world of research closer to that of the field (Loiselle & Harvey, 2007; Guichon, 2007; Adami, 2024; Spanghero-Gaillard & Parpette, 2024). The results of the experiments show that language switching can help expert writers to spread some operations over time, but can also be detrimental to learners with writing difficulties. Furthermore, when the aim is not to perform but to produce, switching languages can encourage creativity. The study of models of texts of the same genre mainly benefits intermediate-level participants. Some participants modified their productions to meet the expectations of the system, which is consistent with several findings in the literature. Finally, the study of models, as well as the memorisation of genre characteristics without contextualising them, is detrimental to advanced-level participants. With regard to the use of neural machine translation (DeepL), we tested two situations: when the tool was used throughout the production process, and when it was introduced at a later stage. The results show that performance increases when learners use DeepL, regardless of when the tool is introduced, but that advanced learners do not benefit from the help. We also show that the use of DeepL can make it possible to delegate or facilitate certain editing and revision operations. The results show that the chosen methodology, and the accompanying experiments, lead to a nuanced view of the process of writing and the appropriation of writing by learners. Individual variations are clearly apparent in our results, which reflect the diversity and complexity of the profiles, and leads us to think of writing supports as complex objects and not as ‘ready-made' solutions (Brassat, 2021, p. 25, our translation ).