The behaviour of metals during differentiation of subduction-related magmas: A case study of active submarine volcanoes on the Hunter-Ridge, SW Pacific

In subduction zone environments, the behaviour of metals and other associated elements during magmatic differentiation has a profound impact on the potential to form magmatic-hydrothermal ore deposits. Understanding the partitioning of ore metals between the melt, crystal and volatile phases is of key importance in their removal from, or concentration in, evolving magmas, and hence, the potential to form economically important mineral deposits. This study focuses on the partitioning of Cu, Zn, Pb, V, Co and Sc between melts and crystallising mineral phases in subduction-related magmas from the Hunter Ridge. This locality offers a unique opportunity to compare calc-alkaline and adakite rock suites. In addition, effects of contamination by continental material can be excluded due to the intra-oceanic setting of this region. The main aim of this work is to determine how magma ascent, crystallisation and degassing can lead either to the enrichment or depletion of ore-forming metals. Furthermore, the common association between porphyry copper deposits and high Sr/Y (i.e. adakitic signature‚ÄövÑvp) magmatic rocks is examined, which may be useful for mineral exploration. Rocks from the Hunter Ridge calc-alkaline series record a protracted history of magma evolution from the very earliest stages of fractionation (i.e. before volatile loss) to their subsequent eruption. Groundmass and glass compositions (rather than whole-rock geochemistry data) together with a detailed trace-element study, have been used to reproduce the liquid line of descent (LLD) for the evolution of the calc-alkaline magmas. Subsequently, it has been demonstrated that in the case of the Hunter Ridge calc-alkaline series, the minerals and melts are genetically related and that fractional crystallisation of olivine + clinopyroxene + plagioclase + magnetite ¬¨¬± amphibole is the primary process recorded in these rocks. Following the same approach for the adakite magma series, it was established that the minerals and melts are genetically related and that fractionation of olivine + clinopyroxene occurred in the more primitive rocks. However, no examples with an MgO wt% lower than 5.2 are represented in our collection. Metal partitioning behaviour in the principal magma components (of both adakitic and calcalkaline rocks) was examined (i.e. for olivine, clinopyroxene, plagioclase, magnetite, and melt). Mineral/melt partition coefficients (Ds) for Cu, Zn, V, Pb, Co and Sc used in the modelling were calculated directly from LA-ICP-MS analyses for phenocryst-groundmass/glass equilibrium pairs. For comparison purposes, modelling was also conducted using Ds values obtained from the literature (where available). Although basaltic rocks from the Hunter Ridge calc-alkaline samples have initial metal abundances similar to those in mid-ocean ridge basalts (MORB), the contents of chalcophile elements such as Cu increase early during magma evolution beyond the range found in typical MORB (from 8.3 to 6.5 MgO wt% while Cu contents in the melt increase from ~89 ppm to ~128 ppm). This behaviour can be linked to the presence of higher water contents and higher oxidation states of arc magmas compared to MORB, which suppresses the formation of sulphide liquids that might otherwise scavenge Cu from a silicate magma. Plagioclase saturation occurs at a melt MgO wt% of 6.5, which is accompanied by a rapid decrease in Cu contents in the magma. This most likely marks the onset of degassing. Zinc and V increase during magma evolution (from ~50 ppm and 230 ppm respectively at 8.3 MgO wt% to a maximum of ~66 ppm and 275 ppm respectively at 5 wt% MgO). Similar behaviour is observed in FeO and TiO2 during magmatic fractionation. Once magnetite enters the liquidus at ~5 wt% MgO, Zn and V both decrease within decreasing MgO (falling to ~50 ppm and 20 ppm respectively at MgO <0.5 wt%). Given that, both DZn mt/mel and DV mt/melt are high at 5.7 and 9.8 respectively (at ~ 5 wt% MgO), this behaviour can be explained by the partitioning of Zn and V into magnetite. Lead contents increase from ~1 ppm at 8.3 wt% MgO to 4.5 ppm at 0.3 wt% MgO. Modelling shows that partitioning of Pb into olivine, clinopyroxene and plagioclase is negligible. In addition, it would appear that Pb does not enter a volatile phase (as with Cu). In contrast to the above elements, Co and Sc concentrations decrease with decreasing MgO wt% of the melt (from ~37 ppm at 8.3 wt% MgO to ~2 ppm and ~8 ppm respectively at 0.3 wt% MgO). This can be explained by the partitioning of Co into olivine and to lesser extent clinopyroxene, and Sc into clinopyroxene (only), during the early stages of magmatic fractionation (as indicated by the high DCo ol/melt of 5.6 - 22.2, DCo cpx/melt of 1.1, DSc cpx/melt of 2 ‚Äö- 14, measured DSc ol/melt of 0.1-0.4). Co and Sc also partition into magnetite once it enters the liquidus, as indicated by DCo mt/melt from 6 to 18 and DSc mt/melt from 0.9 to 2.2 (over a MgO range of 5.3 to 2.3 wt%). These results provide evidence that: i) Cu partitions into a fluid phase at crustal levels before sulphide saturation occurs; ii) Zn and V are greatly affected by magnetite saturation; iii) Pb behaves as an incompatible element during magmatic fractionation and does not enter a fluid phase; iv) Co behaves as compatible element and partitions into olivine and clinopyroxene (and to lesser extent magnetite); v) Sc also behaves as compatible element and partitions into clinopyroxene (and to lesser extent magnetite). There are notable differences in the crystallisation sequence between the calc-alkaline and adakitic magmas from the Hunter Ridge. Basalts from both magma series contain olivine and clinopyroxene as early liquidus phases. In the calc-alkaline series, plagioclase enters the liquidus at ~6.5 MgO wt%, whereas in the adakite series plagioclase saturation is suppressed below 5.2 MgO wt%. The adakite magmas are, therefore, consistent with differentiation under higher water and/or volatile contents and at deeper crustal levels than the calc-alkaline magmas. This has an important effect on the partitioning of metals that have affinity for a volatile phase. The results show that, at least for the Hunter Ridge adakites, there is no a direct contribution of slab-derived melts to the metal budget of these magmas. This is evidenced from the Cu, Zn, Pb, Co and V contents, which in the most primitive adakite samples are comparable (if not lower) than those from the most primitive calc-alkaline samples (i.e. demonstrating that slab melts were not enriched in these elements). Even so, adakitic magmas have more potential to generate economically important ore deposits because they are more likely to exsolve a single-phase fluid at deep crustal levels, which is considered favourable to the formation of large porphyry deposits.