Le réacteur à sels fondus rapides (MSFR) est un des six concepts de réacteur nucléaire retenu lors du Forum Génération IV en 2001. La particularité de ce concept est d'utiliser un combustible liquide constitué d'un sel fondu, LiF-ThF₄-UF₄/UF3₃ (77-19-4 mol%) et d'intégrer un procédé de traitement du sel usé. Ce traitement est constitué d'étapes successives de séparation chimiques basées sur les propriétés redox et acido-basiques des éléments produits dans le réacteur par des réactions nucléaires : produits de fission solubles et gazeux, éléments métalliques et actinides mineurs solubles. L'une des étapes majeures du procédé de traitement est une extraction réductrice qui consiste à mettre en contact le sel fondu et un métal liquide, le bismuth, contenant un élément réducteur, le lithium. Cette étape permet notamment de séparer les actinides mineurs des lanthanides. Les actinides mineurs sont réintroduits dans le réacteur nucléaire afin d'y être brûler alors que les lanthanides seront confinés en stockage profond.Le travail réalisé au cours de cette thèse avait deux objectifs : (i) vérifier la faisabilité de l'extraction réductrice des actinides et des lanthanides, étape qui avait été validée au préalable uniquement sur la base de calculs thermodynamiques et (ii) étudier la chimie des sels fluorures fondus (et notamment le sel combustible LiF-ThF₄-UF₄) en développant une méthodologie pour la détermination de données fondamentales telles que les coefficients d'activité dans les milieux fluorures, coefficients qui quantifient les propriétés de solvatation.La première étape pour réaliser expérimentalement une extraction réductrice consiste à préparer une nappe métallique de Bi-Li liquide de composition pré-définie. Une technique d'électrolyse en milieu LiCl-LiF fondu à 550°C a été retenue pour réaliser ces solutions métalliques. Nous avons montré que seul ce milieu fondu pouvait être utilisé pour la fabrication de ces alliages métalliques. Des tests d'extraction ont ensuite été réalisés par contact entre LiF-ThF₄ (dans lequel sont introduits UF₄ et NdF ₃ pour simuler respectivement les actinides et les lanthanides) et Bi-Li à 650°C. Les principaux résultats montrent que l'extraction du néodyme et de l'uranium a été obtenue avec des rendements respectivement de l'ordre de 3% et 15% dans les meilleures conditions. Ces valeurs sont faibles comparées aux calculs thermodynamiques prévisionnels. On explique la faible efficacité de l'extraction par une extraction simultanée du thorium dans la nappe métallique liquide qui forme des composés intermétalliques à l'interface métal/sel et bloque le transfert interphasique. Des méthodes ont été développées pour atteindre des données fondamentales qui font défaut en milieu fluorures fondus, en particulier les propriétés de solvatation. La spéciation de plusieurs cations métalliques par les ions fluorures à haute température a notamment été étudiée et les constantes de complexation calculées par simulation des résultats expérimentaux. Réalisée pour deux lanthanides, le néodyme et le lanthane, deux actinides, le thorium et l'uranium et également pour un métal de transition, le nickel, cette étude permet d'atteindre les coefficients d'activité de ces éléments dans tous les sels fluorures fondus. En particulier, l'étude de la spéciation du thorium a été une étape importante dans la connaissance de la chimie du sel combustible LiF-ThF₄ puisque nous avons pu en déduire le coefficient d'activité de l'ion fluorure dans ce milieu à 650°C.Enfin, l'ensemble de ce travail a conduit à donner une première estimation de la réactivité de chaque élément de la classification périodique (présent dans le réacteur nucléaire après opération) à chaque étape du traitement du sel combustible usé. / The molten salt fast reactor (MSFR) is one of the six nuclear reactor concepts retained during the Forum GEN IV in 2001. The particularity of this concept is to use a liquid fuel consisting of a molten salt, LiF-ThF₄-UF₄ /UF ₃ (77-19-4 mol%) and to have an integrated spent fuel treatment process. This treatment consists of successive chemical separation steps based on redox and acid-base properties of the elements produced in the reactor by nuclear reactions: soluble and gaseous fission products, metals elements and soluble minor actinides. One of the major steps of the treatment method is a reducing extraction which consists to contact the molten salt and a liquid metal, bismuth, containing the reducing element, lithium. This step allows separating the minor actinides and lanthanides. Minor actinides are reintroduced in the nuclear reactor to be burned while the lanthanides are confined in deep storage.The work in this thesis had two objectives: (i) assess the feasibility of reducing extraction of actinides and lanthanides, a step that had previously only been validated on the basis of thermodynamic calculations and (ii) study the chemistry of molten fluoride salts (and especially the fuel salt) by developing a methodology for the determination of fundamental data such as the activity coefficients in fluorides media, coefficients activities which quantify the solvation properties.To experimentally realize a reducing extraction, the first step is to prepare a metal layer of liquid Bi-Li with predefined composition. An electrolysis technique in molten salt LiCl-LiF at 550°C was chosen to achieve these metal solutions. We have shown that only this molten medium could be used for the manufacture of such metal alloys. Extraction tests were then carried out by contact between LiF-ThF₄ (with UF₄ and NdF ₃ are introduced to simulate respectively the actinides and lanthanides) and Bi-Li at 650°C. The main results show that the extraction of neodymium and uranium was obtained with yields of around 3% and 15% respectively in the best conditions. These values are low compared to previous thermodynamic calculations. Low efficiency of the extraction is due to a simultaneous extraction of thorium in the liquid metal phase which forms intermetallic compounds at the metal/salt interphase and blocks the transfer.Methods have been developed to achieve fundamental data that are lacking in molten fluoride medium, in particularly the solvation properties. Speciation of some metallic cations by fluoride ions with high temperature was particularly studied and calculation of complexation constants by simulated experimental results was done. Carried out for two lanthanides, neodymium and lanthanum, two actinides, thorium and uranium, and also for a transition metal, nickel, this study achieves to calculate the activity coefficients of these elements in different fluoride molten salt. The study of the speciation of thorium was an important step to understand the chemistry of the fuel salt LiF-ThF₄. We were able to calculate the activity coefficient of the fluoride ion in this environment at 650°C.Finally, all of this work allows giving a first estimate of the reactivity of each element of the periodic table (present in the nuclear reactor after operation) at each stage of the treatment of the spent fuel salt.

Polycyclic aromatic compounds (PACs), which comprise a complex class of condensed multi-ring benzenoid compounds, are important environmental pollutants originating from a wide variety of natural and anthropogenic sources. PACs are generally formed during incomplete combustion of pyrolisis of organic matter containing carbon and hydrogen. Because combustion of organic materials is involved in countless natural processes or human activities, PACs are omnipresent and abundant pollutants in air, soil, and water. Chemical analysis of PACs is of great environmental and toxicological importance. Many of them are highly suspect as etiological agents in human cancer. Because PACs carcinogenic properties strongly depend on molecular structure and differ significantly from isomer to isomer, it is of paramount importance to determine the most toxic isomers even if they are present at much lower concentrations than their less toxic isomers. Gas chromatography (GC), high-resolution GC, and high-performance liquid chromatography (HPLC) are the basis for standard PACs identification and determination. Many cases exist where GC, HPLC, and even HR-GC have not been capable to provide unambiguous isomer identification. The lack of reliable analytical data has lead to serious errors in environmental and toxicological studies. This dissertation deals with the development of novel instrumentation and analytical methods for the analysis of PACs in environmental samples. The developed methodology is based on two well-known high-resolution luminescence techniques, namely Shpol'skii Spectroscopy (SS) and Fluorescence Line Narrowing Spectroscopy (FLNS). Although these two techniques have long been recognized for their capability in providing direct determination of target PACs in complex environmental samples, several reasons have hampered their widespread use for the problem at hand. These include inconvenient sample freezing procedures; questions about signal reproducibility; lengthy spectral acquisition, which might cause severe sample degradation due to prolonged excitation; broadband fluorescence background that degrades quality of spectra, precision of measurements and detection limits; solvent constrains imposed by the need of optically transparent media; and, most importantly, the lack of selectivity and sensitivity for unambiguous determination of closely related PACs metabolites. This dissertation presents significant advances on all fronts. The analytical methodology is then extended to the analysis of fluoroquinolones (FQs) in aqueous samples. FQs are one of the most powerful classes of antibiotics currently used for the treatment of urinary tract infections. Their widespread use in both human and animal medicine has prompted their appearance in aquatic systems. The search for a universal method capable to face this new environmental challenge has been centered on HPLC. Depending on the FQ and its concentration level, successful determination has been accomplished with mass spectrometry, room-temperature fluorescence (RTF) or UV absorption spectrometry. Unfortunately, no single detection mode has shown the ability to detect all FQ at the concentration ratios found in environmental waters. We provide a feasible alternative based on FLNS. On the instrumentation side, we present a single instrument with the capability to collect multidimensional data formats in both the fluorescence and the phosphorescence time domains. We demonstrate the ability to perform luminescence measurements in highly scattering media by comparing the precision of measurements in optically transparent solvents (Shpol'skii solvents) to those obtained in "snow-like" matrixes and solid samples. For decades, conventional low-temperature methodology has been restricted to optically transparent media. This restriction has limited its application to organic solvents that freeze into a glass. In this dissertation, we remove this limitation with the use of cryogenic fiber-optic probes. Our final efforts deal with low-temperature absorption measurements. Recording absorption spectra via transmittance through frozen matrixes is a challenging task. The main reason is the difficulty to overcome the strong scattering light reaching the detector. This is particularly true when thick samples are necessary for recording absorption spectra of weak oscillators. In the case of strongly fluorescent compounds, additional errors in absorbance measurements arise from the emission reaching the detector, which might have comparable intensity to that of the transmitted light. We present a fundamentally different approach to low-temperature absorption measurements as the sought-for-information is the intensity of laser excitation returning from the frozen sample to the intensified-charge coupled device (ICCD). Laser excitation is collected with the aid of a cryogenic fiber optic probe. The feasibility of our approach is demonstrated with single-site and multiple-site Shpol'skii systems. 4.2K absorption spectra show excellent agreement to their literature counterparts recorded via transmittance with closed cycle cryogenators. Fluorescence quantum yields measured at room-temperature compare well to experimental data acquired in our lab via classical methodology. Similar agreement is observed between 77K fluorescence quantum yields and previously reported data acquired with classical methodology. We then extend our approach to generate original data on fluorescence quantum yields at 4.2K.