Primitive ankaramitic melts in island arcs : evidence from melt inclusions

Primitive ankaramite rocks differ from picrite by their richness in clinopyroxene (cpx>ol), their high CaO/Al2O3 values (>1, wt%) and their high normative di/ol values (>0.7, mol%). Ankaramite rocks are generally interpreted as a variety of basalt enriched in clinopyroxene crystals yet some studies suggest that ankararnite rocks crystallized from primitive ankaramitic melts. The origin of ankaramitic melts is of interest because their compositions differ from the picritic compositions of experimentally produced partial melts in equilibrium with mantle peridotite, yet many of their of their characteristics are similar to those of primitive mantle melts, i.e. high Mg# values, high Ni and Cr, and magnesian olivine phenocrysts. The four primitive ankaramite suites studied in this thesis are from the Ulakan Formation in Bali, and the Rinjani volcano in Lombok (Sunda arc), and from Merelava and Epi (Vanuatu arc). Primary melt inclusions in magnesian olivine phenocrysts (Fo>90) from each suite are studied to investigate the early stages of ankaramite magma evolution. The composition of these melt inclusions after homogenization are quite unlike picrite and have CaO/Al203 values which range up to ‚ÄövÑvÆ1.7, and are therefore ankaramitic. Additionally, the CaO/Al203 values of melt inclusions within the same grain are similar, but vary between phenocrysts. Critical assessment of the data suggests that the composition of melt inclusions trapped in olivine can be modified by Fe-Mg re-equilibration with the host, before the magma is erupted. This process might lower the original FeO* content of the trapped melt by several wt% (\Fe-loss\"). The extent of \"Fe-loss\" depends on (1) the time spent by the host olivine phenocryst in the magma before eruption (residence time) and (2) the rate at which the re-equilibration occurred. Most of the compositions of homogenized melt inclusions in this study are affected by \"Fe-loss\" and are therefore not directly representative of the original trapped melt. A recalculation procedure is developed to reconstruct the original composition of the melt at the moment of trapping. Both before and after recalculation the compositions of melt inclusions display di/ol values >1 thus retaining their ankaramitic affinities. The CaO/Al203 values of melt inclusions are unaffected by this procedure. The recalculated melt inclusion compositions are also more silica-undersaturated (ne- lc- and cs-normative) than the host ankaramite rocks which range from ne- to hy-normative. If the melt inclusions are aliquots of the parent melts these results suggest that the ankaramite magma formed by the aggregation of strongly-silica-undersaturated primitive ankaramite melts with high CaO/Al203 and high di/o/ values. A link may therefore exist between these trapped melts the hypothetical primary melts and the formation of ankaramitic magmas. This link is explored experimentally at high pressures using the composition of a representative melt inclusion in Fosq from the Lombok ankaramite suite as a starting mix composition. Liquidus and near-liquidus phase relations of this melt under dry hydrous CO2-H20- and CO2-bearing conditions and between 1 to 3 GPa pressures lack orthopyroxene near the liquidus. Therefore a direct origin is excluded for this composition by partial melting of mantle peridotite (i.e. not a primary melt). Instead the trapped melt inclusions may have been derived from even more mafic primary melts that were generated at higher temperatures and pressures than those prevailing when the ankaramitic magma aggregated. Two mechanisms are considered for the formation of these primary melts that have high CaO/Al203 values (>1): (a) partial melting of lherzolite at high pressures (>5 GPa) and (b) partial melting of lherzolite at lower pressures (<5 GPa) in the presence of CO2-rich mantle fluids. The natural phenocryst assemblage and composition of the host ankaramite rock are only duplicated experimentally under hydrous conditions. With increasing CO2 content in the fluid the compositions of synthesized clinopyroxene crystals become less calcic and are unlike the natural phenocrysts. Thus the crystallization and aggregation of the ankaramite magma may have taken place under hydrous conditions at depths where olivine and clinopyroxene are co-crystallizing as phenocryst phases. In the case of the Lombok ankaramite suite these conditions correspond to pressures equivalent to depths of around ‚ÄövÑvÆ35 km below the base of the arc crust. The aggregated ankararnite magma may mix and continue to crystallize isobarically and in-situ thus producing the wide compositional range observed in olivine (‚ÄövÑvÆFo75.91) and clinopyroxene (‚ÄövÑvÆWo45-47En50-41Fs5-12) phenocrysts as well as the variable zoning and resorption textures. Reaction and re-equilibration of this aggregated magma with the sub-arc mantle before eruption may cause the silicaenrichment that distinguishes the wholerock ankaramite compositions from the melt inclusions in their olivine phenocrysts. Tiny pleonastic daughter spinel crystals with up to 65 wt% Al2O3 and virtually no chromium occur within melt inclusions in olivine phenocrysts. These aluminous spinels could form in melt inclusions provided the trapped melt becomes sufficiently Al-enriched and Cr-depleted following the fractional crystallization of olivine on the walls. Daughter aluminous spinels were found in the melt inclusions of olivine phenocrysts from all four ankaramite suites studied and may therefore also occur in melt inclusions of other basaltic rocks. The occurrence of aluminous spinels in melt inclusions is therefore not evidence for the trapping of contaminant aluminous melts. This thesis attempts to 1) demonstrate that ankaramitic melts with CaO/Al203 values (>1) and normative di/ol values >0.7 exist and can be parental to ankaramite rocks; 2) describe the Fe-Mg reequilibration (\"Fe-loss\") process that affects melt inclusions in olivine phenocrysts before magma eruption; and 3) re-interpret the formation of aluminous spinel crystals in melt inclusions that have previously been interpreted as evidence for contaminant aluminous melts in basaltic magma chambers."