(Table 1) Composition of rocks from ODP Hole 140-504B before and after leaching experiments

Hydrothermally altered rocks from the base metal depleted reaction zone in the lower sheeted dyke complex (ODP Hole 504B) show significant losses of rare earth elements (REE) in extensively altered domains (alteration patches). While very similar in major element composition and alteration mineralogy, the alteration patches have up to 50% lower REE concentrations than the surrounding weakly altered diabases. In order to test if there is a dependence of REE mobility on the alteration history, we carried out leaching experiments on a weakly altered diabase and an adjacent alteration patch. The leaching experiments are used as a probe to identify the hosting phases of major and trace elements before and after alteration; they are not designed to simulate natural alteration processes. Major element compositions of the leaching solutions show that small amounts (9 to 14%) of the rocks and similar proportions of plagioclase, amphibole, and chlorite were dissolved from both rock types. In contrast to the major elements, up to 100% of the REE could be leached from the strongly recrystallised alteration patch after 20 h. All leaching solutions were light REE-enriched and had negative Eu anomalies compared to the host rock. Geochemical modelling suggests that mainly interstitial material (formerly late-stage magmatic liquids) and probably some apatite are the main REE sources in the rocks affected during experimental leaching. Dissolution of plagioclase, amphibole, clinopyroxene, and chlorite cannot explain the REE compositions of the leachates. The higher REE leachability of the heavily altered rock suggests that the extensive hydrothermal alteration resulted in a complete redistribution of the REE from lattices of the magmatic minerals to the grain boundaries of secondary minerals. This can explain the marked increase in REE accessibility during ongoing alteration, which finally led to REE depletions in the highly altered rock. We propose that late-stage interstitial material provides the most important REE source of rocks in the oceanic crust. This interpretation is in contrast to a model by Klinkhammer et al. [Klinkhammer, G.P., Elderfield, H., Edmond, J.M., Mitra, A., 1994. Geochemical implications of rare-earth element patterns in hydrothermal fluids from mid-ocean ridges. Geochim. Cosmochim. Acta, Vol. 58, pp. 5105-5113.] who suggested that the similarity in the REE patterns of plagioclase and that of hydrothermal vent fluids is a result of chemical exchange between plagioclase and hot fluids in the ocean crust. Our leaching results indicate that, although dramatic dissolution of plagioclase occurred, it did not control the REE patterns of the leachates. Moreover, REE patterns inferred for fluids that mobilised REE from the highly altered rock are different from those of vent fluids. We, therefore, propose that the characteristic REE pattern of vent fluids may result from differences in aqueous chloride complex stabilities, sorption-controlled REE fractionation during fluid migration and mineral precipitation in the discharge zone.