Integrated (theoretical, experimental and petrological) study of the stability of zircon and its isotopic systems at dry high-T and hydrothermal medium- to low-T conditions

Abstract

Zircon is one of the most studied minerals by Earth Scientists because of its widespread occurrence in a large diversity of rocks and its key role in the distribution of crucial incompatible trace elements, such as U, Th, Y, Hf and rare earth elements (REE), its generalized application in U-Th-Pb geochronometry and the determination of magma sources throughout the signature of its oxygen and hafnium isotopes. Its remarkable mechanical and chemical resilience makes it possible to carry out all these studies. However, the stability of zircon and its geochemical behavior in environmental conditions very different from its formation are still not well understood. On the other hand, for a better understanding of the geochronological and geochemical information recorded in zircon, it is necessary to know the mechanisms that allow the inheritance of zircon in granites and the formation of zircon in hot mafic magmas that are generally undersaturated in it. This PhD thesis aims to deepen our understanding of zircon stability and its geochemical behavior at extremely high-T conditions, such as those occurring during assimilation processes in basic magmas that can lead to zircon xenocryst entrapment in cumulus mafic minerals and in hydrothermal environments, where it can be a source of trace elements for ore genesis. These purposes have been accomplished by performing annealing experiments using a tubular furnace on the one side, and hydrothermal experiments with Teflon bombs and autoclave reactor, on the other. In addition, the parameters that control zircon inheritance in granitic magmas and the mechanisms of zircon crystallization in low-Zr mafic magmas are modeled numerically. Three samples of zircon separates, chosen based on their morphology and degree of metamictization, were used for the experiments. The samples are The Cambrian-Ordovician orthogneiss SAB50, because it contains numerous pristine zircons that often have inherited cores of Ediacaran or older age. The Variscan tonalite SAB51, because it contains zircons with a very uniform isotopic composition and occasional metamict bands. The Early Paleoproterozoic to Neoarchaean syenite REG20 because it contains highly metamict zircons that are very rich in inclusions and show high 204Pb and radiogenic Pb loss. For the dry high-T experiments, zircon grains embedded in pure cristobalite were placed in open crucibles and heated in a horizontal tubular furnace under an N2 atmosphere at 1300°C for 1, 3 and 6 months. Ultra-pure H2O and 2M NaCl, 2M CaCl2 and 1M NaF solutions were used for the hydrothermal experiments. Zircon + granite powder + solution and granite powder + solution were placed into the teflon bombs and heated at 170 ºC at ca. 10 bar for 43 days. In the autoclave experiments, zircon + solution and zircon + quartz + solution were heated at 550 ºC, at ca. 2 kbar for 3 and 8 days. For determining the most important parameters that control zircon inheritance, we calculated the melt fraction as a function of the temperature in three selected protoliths: a peraluminous greywacke, a metaluminous granodiorite, and a gabbrodiorite. The calculations were done from 650°C to their liquidus temperatures at 3, 6 and 9 kbar and 2, 4, 6, 8 and 10 wt% of H2O using rhyolite-MELTS. Melt production rates were determined using COMSOLTM. For understanding the mechanisms that enable zircon crystallization from low-Zr mafic magmas, diffusion kinetics of Zr rejected by growing minerals in pore-confined MORB melts was determined using moving mesh 2D finite element models computed with COMSOLTM. The annealing experiments show recrystallization of metamict domains, melting of polymineralic inclusions, formation of nanopores and microcracking propagated by thermo-elastic stress accumulated at the interface between domains with different lattice orientations. The zircon-to-baddeleyite transformation occurred by (i) incongruent zircon dissolution in molten mineral inclusions with a high CaO/SiO2 ratio and (ii) recrystallization of metamict domains mediated by silica releasing from the reaction site. Highly metamict zircon was successfully dated after annealing at 1300 ºC because all their common Pb but little radiogenic Pb were lost due to the generation of a melt. Ti concentration increased in the zircon lattice of annealed zircon grains with minute inclusions of rutile or other Ti-bearing minerals. Tungsten could achieve high abundances in zircons from regions with W deposits. An overall increase in W concentration occurred upon annealing because of the dissolution of minute W impurities into the zircon lattice. However, given the limited solubility of W in zircon, a fraction of the released W was consumed in forming W-rich minerals. The oxygen isotope composition of annealed zircon grains embedded in cristobalite drifted quickly to that of this latter. This leads to the conclusion that crustal-derived zircon xenocrysts with high ∂18O from mantle rocks could not have resided in the mantle for a long time. The hydrothermal experiments show that in all low-T, silica-saturated fluids, congruent zircon dissolution took place in the presence of H2O and saline fluids, leading to an increase of zircon permeability along microcracks. In the high-T, silica-saturated experiments, zircon dissolution was congruent for H2O, CaCl2 and NaCl fluids and incongruent for NaF fluids with precipitation of Na-Zr silicate. In all the high-T, silicaundersaturated experiments, zircon was transformed to baddeleyite by coupled dissolution-precipitation mechanisms enhanced by microcracks and pores and permeability creation associated with the lower molar volume of baddeleyite and the silica releasing from the reaction site. The total zircon-to-baddeleyite transformation occurred in the NaCl and NaF fluids, indicating that solutions with Na+ are more reactive than those with Ca2+. Leachates extracted from both high- and low-T hydrothermal experiments presented a remarkable decoupling of Y from HREE, with the former being significantly more partitioned into the fluid. Pb was also partitioned into the leachates with respect to U in all experiments, whereas a significant Th-U fractionation was only present in CaCl2 and NaCl fluids at low-T conditions, with Th being more partitioned into the fluid. The numerical models predicted that the capacity of granitic magmas to inherit zircons from their source was a consequence of the interplay between zircon saturation in the melt and magma mobility, being water content of magmas as the key parameter for controlling the abundance of inherited zircon in granitic rocks. In most cases, the rate of zircon dissolution was significantly faster than this of melt generation. It is estimated that a much more intense heat flux than the one supplied by radioactive decay or asthenospheric upwelling is required for generating melts fast enough to prevent equilibrium with zircon. This requirement is fulfilled only by hot mafic magma underplating, where most of the heat released to the crust comes from its latent heat of crystallization. Zircon inheritance may be substantial in S-type water-rich granite magmas generated at 4.5–6 kbar. In contrast, water-poor granite magmas have moderate or no inheritance because they need higher temperatures to produce a similar melt fraction. The numerical models also predicted that local Zr enrichment at growing interfaces of major minerals could only occur when magma reached a crystallinity high enough to confine residual melts into pores; otherwise, erosion of Zr-rich melt created in the diffusion boundary layers could occur. This process can explain the appearance of Mid-Atlantic syn-magmatic zircons being limited chiefly to cumulate rocks. The calculations showed that local saturation next to growing mineral interfaces seems to be the only possible mechanism for explaining zircon growth before attaining zircon saturation in the bulk mafic magma. On the other hand, the proposed mechanism will be hampered, or even totally inhibited, by the crystallization of major minerals that can partition Zr. Thus, the model predicts that the growth of minerals with Zr mineral/ melt partition coefficient > 0.2, such as clinopyroxene and amphibole, will not produce a zircon-saturated diffusion boundary layer. This implies that the crystallization of these minerals in mafic systems will never reach zircon saturation unless they are much richer in Zr than the MORB considered in the model. These relationships explain the appearance of zircon primocrysts only in the earliest cumulates from the Mid- Atlantic rocks.