Dans le cadre de cette thèse, nous avons cherché à mieux comprendre le comportement des éléments traces – et notamment des éléments légers Li et B – lors des processus de dégazage magmatique par une approche géochimique, basée sur l'analyse de produits volcaniques naturels (laves, gaz, aérosols volcaniques) de deux volcans aux styles éruptifs très contrastés, le Piton de la Fournaise (Réunion) et le Lascar (Chili). Dans un premier temps, cette étude s'est intéressée au rôle des transferts de gaz dans le déclenchement des éruptions majeures et aux échelles de temps impliquées. L'étude des compositions en éléments traces des laves récentes (1998-2008) du Piton de la Fournaise nous a notamment permis d'identifier des anomalies transitoires en éléments volatils (p.e., Li, Cu, B, Tl, Bi, Cd) en début de l'éruption majeure d'avril 2007. La cinétique de fractionnement par diffusion des éléments explique les anomalies observées. Les courts laps de temps nécessaires pour fractionner par diffusion Li par rapport à Cd (minutes à quelques heures) et Bi par rapport à Cd (quelques heures à deux jours) soutiennent l'idée que les magmas ont subi des variations rapides de pression quelques jours avant l'effondrement du cratère du Dolomieu. Dans un second temps, ce travail de thèse a porté sur le dégazage passif du volcan Lascar. Le panache volcanique dilué a été échantillonné au cours de 3 missions d'échantillonnage, menées entre 2009 et 2012, et caractérisé en termes de gaz acides (SO2, HCl, HF) et d’éléments traces. Au cours de ces missions, des mesures de flux de SO2 par DOAS ont également été réalisées. Les données mettent en évidence deux sources principales qui contribuent à l'activité de dégazage observée en surface: un réservoir magmatique profond et un système hydrothermal superficiel. Les contributions des deux sources varient dans le temps en réponse aux changements de l'activité volcanique. Cette évolution temporelle a été démontrée non seulement avec des traceurs répandus comme le SO2 et HCl, mais aussi avec des éléments traces à la fois volatils et très mobiles tel que le B. Pour la détermination des teneurs en B de nos échantillons, nous avons développé une méthode de dosage très précise par dilution isotopique. Appliquées aux laves du Piton de la Fournaise, cette technique nous a permis d'estimer les quantités de B perdues lors des processus de dégazage magmatique (ϵB compris entre 10 et 30%) ainsi que leur dépendance aux conditions de dégazage (continu en système ouvert, processus pré-, syn- et post-éruptifs). Appliquées aux aérosols du Lascar, elle nous a permis de montrer que la volatilité du B est favorisée lors des processus hydrothermaux (interactions gaz-eau, gaz-roche). Enfin, appliquées à des enclaves de péridotites, cette technique nous a permis d'apporter des contraintes nouvelles sur le comportement du B dans le manteau terrestre et d'estimer la teneur en B du manteau primitif (0,26 ± 0,04 ppm). / This study is aimed at better understanding the behavior of trace elements – and notably those of light elements such as Li and B – during magma degassing processes. For this purpose, we used a geochemical approach based on the analysis of fresh lavas and volcanic aerosols from Piton de la Fournaise (Réunion) and Lascar (Chile) volcanoes, respectively. Firstly, this thesis work focused on the role of gas transfers in triggering major eruptions and the time scales involved. Trace element analyses of recent lavas (1998-2008) of Piton de la Fournaise reveal anomalous abundances of volatile elements (e.g., Li, Cu, B, Tl, Bi, Cd) a few days prior to the April 2007 summit collapse. The kinetic (diffusive) fractionation of elements accounts for the observed anomalies. The short time-scales required to fractionate Li from Cd diffusively (minutes to hours) and Cd from Bi (few hours to two days) support the idea that the magmas underwent rapid pressure variations a few days before the summit collapse.Secondly, this study concentrated on the quiescent degassing activity of Lascar volcano. Both major gaseous species and trace element enrichment in gas and aerosols collected in the sustained plume over the period 2009 to 2012 suggest the involvement of two main degassing sources with contrasted geochemical signatures: a deep magmatic reservoir and a shallow hydrothermal system. Contributions from these two dominant sources vary with time in response to changes in volcanic activity. This temporal evolution has been shown not only by well-known tracers such as SO2 and HCl, but also by a trace element both volatile and highly fluid-mobile such as B. To determine the bulk boron concentration of our samples, we have developed a robust low-blank method based on isotope dilution ICP-MS. Applied to lavas of Piton de la Fournaise, this technique allowed us to quantify the amount of B lost during magma degassing (10-30%) and its dependency on degassing conditions. Applied to aerosols of Lascar, it enabled us to show that B volatility is enhanced during hydrothermal processes (gas-water, gas-rock interactions). Finally, applied to fertile peridotite xenoliths, it led us to establish new constraints on the behavior of B during mantle processes and estimate a primitive mantle B content of 0.26 ± 0.04 ppm.

Massive sulfide deposits in the Noranda mining camp, northwestern Québec, are mainly associated with extensive footwall alteration defined by intense chloritization and sericitization. However, Fe-carbonate alteration also occurs in proximity to some deposits. To test the exploration significance of carbonate alteration in the camp, two areas of intense carbonate alteration were examined, around the small Delbridge deposit and near the new Pinkos occurrence in the Cyprus Rhyolite. Between 1969 and 1971, the Delbridge deposit produced 370,000 t of ore grading 9.6% Zn, 0.61% Cu, 110 g/t Ag, and 2.1 g/t Au. Recent drilling at the new Pinkos occurrence intersected 2.64 m of massive to semi-massive sulfides grading 8.1% Zn and 18.2 g/t Ag. Alteration mapping has shown that the distribution of Fe-carbonates can be used to identify vertically extensive zones of hydrothermal upflow at both properties. At Delbridge, intense Fe-carbonate alteration in brecciated rhyolite defines a pipe-like upflow zone that extends vertically for up to 300 m within the stratigraphic footwall of the massive sulfides and 100 m into the hanging wall. The location of known massive sulfide mineralization coincides with the intersection of the alteration pipe and a favorable horizon marked by the occurrence of fine-grained volcaniclastic rocks. At Pinkos, a similar zone of Fe-carbonate alteration occurs in outcrops of coherent rhyolite. Fe-carbonate alteration is most intensely developed along polygonal cooling fractures in massive rhyolite and decreases in intensity towards the centers of the columns. Fe-carbonate stringers and locally abundant matrix carbonate occur in fragmental rocks at the stratigraphic top of the coherent rhyolite flows and are most intense at the location of sulfide-bearing outcrops that mark the known mineralized horizon. Whereas Fe-carbonate alteration defines the central part of the hydrothermal upflow zones at both properties, disseminated pyrite occurs at the margins and is widespread outside the main upflow zones. This may indicate that Fe-carbonate in the main upflow zones formed at the expense of earlier disseminated sulfides. Replacement of pyrite by synvolcanic Fe-carbonate alteration at Delbridge and Pinkos can probably be attributed to a relatively high concentration of dissolved CO2, possibly of magmatic origin, in the main-stage ore-forming fluids.

Massive sulfide deposits in the Noranda mining camp, northwestern Québec, are mainly associated with extensive footwall alteration defined by intense chloritization and sericitization. However, Fe-carbonate alteration also occurs in proximity to some deposits. To test the exploration significance of carbonate alteration in the camp, two areas of intense carbonate alteration were examined, around the small Delbridge deposit and near the new Pinkos occurrence in the Cyprus Rhyolite. Between 1969 and 1971, the Delbridge deposit produced 370,000 t of ore grading 9.6% Zn, 0.61% Cu, 110 g/t Ag, and 2.1 g/t Au. Recent drilling at the new Pinkos occurrence intersected 2.64 m of massive to semi-massive sulfides grading 8.1% Zn and 18.2 g/t Ag. Alteration mapping has shown that the distribution of Fe-carbonates can be used to identify vertically extensive zones of hydrothermal upflow at both properties. At Delbridge, intense Fe-carbonate alteration in brecciated rhyolite defines a pipe-like upflow zone that extends vertically for up to 300 m within the stratigraphic footwall of the massive sulfides and 100 m into the hanging wall. The location of known massive sulfide mineralization coincides with the intersection of the alteration pipe and a favorable horizon marked by the occurrence of fine-grained volcaniclastic rocks. At Pinkos, a similar zone of Fe-carbonate alteration occurs in outcrops of coherent rhyolite. Fe-carbonate alteration is most intensely developed along polygonal cooling fractures in massive rhyolite and decreases in intensity towards the centers of the columns. Fe-carbonate stringers and locally abundant matrix carbonate occur in fragmental rocks at the stratigraphic top of the coherent rhyolite flows and are most intense at the location of sulfide-bearing outcrops that mark the known mineralized horizon. Whereas Fe-carbonate alteration defines the central part of the hydrothermal upflow zones at both properties, disseminated pyrite occurs at the margins and is widespread outside the main upflow zones. This may indicate that Fe-carbonate in the main upflow zones formed at the expense of earlier disseminated sulfides. Replacement of pyrite by synvolcanic Fe-carbonate alteration at Delbridge and Pinkos can probably be attributed to a relatively high concentration of dissolved CO2, possibly of magmatic origin, in the main-stage ore-forming fluids.

Massive sulfide deposits in the Noranda mining camp, northwestern Québec, are mainly associated with extensive footwall alteration defined by intense chloritization and sericitization. However, Fe-carbonate alteration also occurs in proximity to some deposits. To test the exploration significance of carbonate alteration in the camp, two areas of intense carbonate alteration were examined, around the small Delbridge deposit and near the new Pinkos occurrence in the Cyprus Rhyolite. Between 1969 and 1971, the Delbridge deposit produced 370,000 t of ore grading 9.6% Zn, 0.61% Cu, 110 g/t Ag, and 2.1 g/t Au. Recent drilling at the new Pinkos occurrence intersected 2.64 m of massive to semi-massive sulfides grading 8.1% Zn and 18.2 g/t Ag. Alteration mapping has shown that the distribution of Fe-carbonates can be used to identify vertically extensive zones of hydrothermal upflow at both properties. At Delbridge, intense Fe-carbonate alteration in brecciated rhyolite defines a pipe-like upflow zone that extends vertically for up to 300 m within the stratigraphic footwall of the massive sulfides and 100 m into the hanging wall. The location of known massive sulfide mineralization coincides with the intersection of the alteration pipe and a favorable horizon marked by the occurrence of fine-grained volcaniclastic rocks. At Pinkos, a similar zone of Fe-carbonate alteration occurs in outcrops of coherent rhyolite. Fe-carbonate alteration is most intensely developed along polygonal cooling fractures in massive rhyolite and decreases in intensity towards the centers of the columns. Fe-carbonate stringers and locally abundant matrix carbonate occur in fragmental rocks at the stratigraphic top of the coherent rhyolite flows and are most intense at the location of sulfide-bearing outcrops that mark the known mineralized horizon. Whereas Fe-carbonate alteration defines the central part of the hydrothermal upflow zones at both properties, disseminated pyrite occurs at the margins and is widespread outside the main upflow zones. This may indicate that Fe-carbonate in the main upflow zones formed at the expense of earlier disseminated sulfides. Replacement of pyrite by synvolcanic Fe-carbonate alteration at Delbridge and Pinkos can probably be attributed to a relatively high concentration of dissolved CO2, possibly of magmatic origin, in the main-stage ore-forming fluids.

La croissance de la phase vésiculée, moteur de l'éruption, est contrôlée par les processus de diffusion qui permettent la migration des gaz (et notamment des gaz rares) dans les bulles. On utilise la haute volatilité des gaz rares comme traceur géochimique de l'évolution d'une phase gazeuse sans interaction chimique. Ainsi, documenter précisément les mécanismes de diffusion des différents gaz rares (He, Ne, Ar) lors de l'éruption (c'est-à-dire en fonction de la chute de température et de pression du système), permet de quantifier les phénomènes de fractionnement de la phase gazeuse. La compréhension des processus de fractionnements cinétiques, permet dès lors de prédire le temps nécessaire pour atteindre une certaine quantité de gaz rares dans une bulle (située au sein d'un système magmatique), lors de l'éjection des laves. Pour cela, la compréhension de l'influence de la température et de la structure du réseau silicaté sur les coefficients de diffusion est nécessaire. Cependant, la compréhension physique des processus de diffusion ainsi que l'évolution des coefficients de diffusion en fonction de la température, n'est pas suffisante pour dériver des temps caractéristiques d'une éruption volcanique de type Plinian. La complexité symptomatique de tels systèmes, nécessite une résolution numérique des équations de diffusion prenant en compte la dépendance des coefficients de diffusion à la température. Plusieurs verres synthétiques et naturels de composition basaltique ont été fabriqués dans le but de déterminer la vitesse de diffusion des gaz rares. Les données de diffusivités expérimentales mesurées sur ces systèmes, depuis l'état vitreux de basse température (T = 423 K) jusqu'à des températures sur-liquidus (T = 1823 K), documentent nos connaissances des processus physiques de diffusion dans ces milieux. Un modèle numérique intègre ces données et permet de suivre en continue la variation des coefficients de diffusion lors de la trempe d'une lave. On a pu ainsi montré : - La relation particulière entre la structure du milieu diffusif et les espèces diffusantes. La quantité de formateurs de réseaux (SiO2) et de modificateurs (CaO - MgO - etc.), joue sur la connectivité des chemins de diffusions de chaque gaz rare, avec un effet antagoniste entre l'ouverture globale du réseau et la connexion des tétraèdres de la structure. - La présence de comportements non-arrheniens des gaz rares proches de la Tg, due à la relaxation du réseau silicate. - L'importance des données expérimentales dans l'étude des mécanismes de dégazage des magmas basaltiques. En effet, les études précédentes utilisent des extrapolations des coefficients de diffusions, mesurés dans le verre pour extrapoler les diffusivités dans le liquide silicaté. Nos données montrent que le fractionnement cinétique des gaz rares pendant le dégazage de lave basaltique, est surestimé par ces extrapolations basées sur les vitesses de diffusions aux basses températures (T << Tg) / Noble gas geochemistry is an important tool for constraining the history of the volatile phase during magmatic eruptions. Degassing processes control the gas flux from liquid to bubble, leading to solubility- or kinetic-control of the fractionation mechanisms. Noble gases have no chemical interactions at magmatic conditions and are therefore well adapted to tracing gas fractionation mechanisms during the evolution of the gas phase. Well constrained diffusion coefficients, and their dependence on temperature, of several noble gases are critical for estimating the timescale of a plinian eruption, for example. During the quench phase of the lava ejected in the plume, atmospheric noble gases will diffuse through the liquid/glass shell surrounding gas bubbles. Diffusion of these atmospheric gases determine the gas content measured in the eruption products, which are therefore a function of the timescale of the eruption, the initial and final temperatures, the glass/liquid shell thickness and the cooling rate of the magma. Therefore, it should be possible to calculate plinian eruption timescales from noble gas fractionation patterns trapped in pumice. However, in order to perform the diffusion calculations, it is first necessary to model the diffusive system: a numerical resolution of the diffusion equations for hollow sphere geometry is required as there are no analytical solutions (for complex thermal histories such as for a plinian ash column). In order to constrain the diffusion mechanisms (He, Ne and Ar) in silicate glasses and liquids, several synthetic basaltic glasses were produced. Diffusion coefficients were measured from low temperatures (423 K) to the Tg (glass transition temperature) of the system (1005 K). These experiments allowed us to investigate the physical processes that limit diffusion in glassy media: He, Ne and Ar diffusion in silicate glasses show non-Arrhenian behavior as the Tg is approached thought to be due to structural relaxation of the silicate network itself. Complementary diffusion experiments (on He and Ar) at super-liquidus conditions (1673 K and 1823 K) provide important information on the temperature dependency of He/Ar fractionation in silicate liquids. These diffusion measurements required that a new experimental protocol was developed in order to investigate noble gas diffusivities in silicate melts. The results show that relative He and Ar diffusion (i.e. DHe/DAr) decreases with temperature, from 165 at temperatures close to the Tg to 3.2 at high (>1823K) temperature. The measured coefficient diffusions are incorporated to a numerical model of the diffusion equations for a hollow sphere geometry that were developed as a MatLab code as part of this thesis work. This enabled us to determine the likely timescales of plinian eruptions from existing noble gas measurements. These results also have important implications for mechanisms of degassing in basaltic magmas: previous work used diffusivities measured on glasses in order to extrapolate to noble gas diffusivities at magmatic temperatures. Our measurements show that kinetic fractionation of noble gases during degassing of basaltic magmas has likely been overstated because noble gas diffusion in the glass cannot be extrapolated to the liquid state