Détermination expérimentale du coefficient de partage F entre une phase fluide et des minéraux hydratés en zone de subduction

Résumé en français indisponible.Mechanisms of volatile transfer from subducting slab to the melting region beneath arc volcanoes are probably the least understood process of arc magma genesis. Fluorine, which suffers minimum degassing in arc primitive melt inclusions, retains the information about the role of volatiles during magma genesis at depth. Experimentally determined solubility of F in aqueous fluid, and partition coefficients of F between fluid and minerals provide first order geochemical constraints about the volatile-transporting agent.My thesis experimentally determined F solubility in fluid and its partition coefficients among several phases. The systems are in equilibrium with hornblende and a humite group mineral (some contain melt or pyroxene) at 1 – 2 GPa, from 770 to 1047 °C, or equilibrium with hydrogrossular, pyroxene and norbergite or chondrodite at 2.5 – 3 GPa and 877 °C. The experiments were conducted with piston cylinder and cold sealing technique. The oxygen fugacity conditions were controlled by NNO buffer, while some were unbuffered. The fluids were extracted into volumetric flasks, and their compositions were determined by mass balance calculations. Moreover, the consistency was verified by HPLC for fluorine ion, and ICP-MS or ICP-AES for major cations of the quenched fluids.In 1 GPa experiments, the quench phases are so rare that the majority of the fluid compositions from direct analyses are consistent with mass balance results in their uncertainties. Moreover, my mass balance procedure takes into account all the measurements errors, which leads to large uncertainties on fluid compositions. The consistency demonstrates that most of fluorine aflter annealing in the capsule is present as fluorine ion. Futhermore, increases of the masses of starting materials, fluid proportions and analytical precisions will improve the uncertainties performances. ������������ can be represented by a single value 0.135 ± 0.036, which is independent of temperature, bulk composition and buffer conditions at 1 GPa. Df between fluid and humite group minerals is much less. Xf of hornblende and norbergite decrease from 1 to 2 GPa, while F partitionig between them doesn't change much. It indicates that F partitioning between fluid and minerals increases. Moreover, F concentrations in norbergite between NNO buffered and unbuffered experiments are significantly different. Meanwhile, Fe concentration variations of norbergite indicate that unbeffered experiments have higher oxygen fugacity than the NNO buffered ones. According to high temperature improves the free radical exchange reactions, H2O + 0.5O2 ⇄ 2OH. It indicates that both water fugacity and oxygen fugacity contribute to OH fugacity in fluid. I developed a simple model in which XF in humite group minerals are correlated to the ratio between F and OH. It is sucessfully applied to estimate the F concentration in the fluid, which co-exists with clinohumite, using Xf value.With the knowledge of my study, a new constraint can be framed on slab flux. The average F concentration in the fluid is 2700 ppm for F-rich experiments and it constrains the maximum amount of F carried by fluid in the presence of amphibole. Using partition coefficient of F to estimate F abundance in subducting slab, one can conclude that the increase of F concentration in the subarc mantle by fluid, in equilibrium with hornblende, to be less tan 5 ppm. Significant F enrichments found in arc lavas cannot be derived from aqueous fluid of subductiong slab in the presence of amphibole. Therefore, this result highlights the role either 1) slab melt, 2) fluid in equilibrium with eclogite, or perhaps 3) supercritical fluid for the element transfer from slab to mantle wedge