Appendix K – The redox state of the Ischia parental mafic magmas

In order to compute the redox state of Ischia parental mafic magmas we used the analyzedOl-hosted Cr-spinels, which most likely reflects the enriched mantle source. To assess the FeO/MgO ratio of the melt, and then retrieve the Fe2O3 content by difference, hence the oxidation state, we used the equation of Maurel and Maurel (1982). It is worthy to point out that the equation cannot be applied to whole-rock compositions, which inevitably yield negative Fe2O3 values in the light of their low FeOtot/MgO ratio, likely due to accumulation of olivine crystals. However, the problem can be easily circumvented by adopting the composition of MIs hosted in Mg-rich olivine crystals. The analyzed Ischia MIs show unusual MgO contents, even lower than those of whole rocks (e.g., analyses in Appendix I; more can be found in Moretti et al. 2013).

Results are reported in the table attached below. We selected two MIs representative of most mafic and intermediate compositions found in this study (I_Vat-B_1 ol6 and I_Vat-A_1 ol3 inc1, respectively, Appendix I) that were coupled to Cr-spinelanalyses reported in Appendix B. Two scenarios have been here considered, in order to embrace the possible range of redox conditions recorded by theserepresentative MIs.

In the first scenario, MIs are considered representative of the magma from which early olivine crystals were crystallizing. In such a case, low-MgO contents are related to a high oxidation state of the parental magma, with Fe3+/Fe spanning from 0.22 to 0.45 (see the tableattached below). The high oxidation state of the most mafic magma returns almost equivalent proportions of divalent and trivalent iron (0.44-0.45), thus suggesting that Fe2+/Fe3+might be the effective couple imposing the redox state to the Ischia source.

In the second scenario, we consider that early olivine crystals could have equilibrated in magmas more mafic with respect to host-rock and trapped MIs. In this case, calculation for post-entrapment crystallization (PEC) of olivine is required. However, a major problem is the lack of any constraints on either initial total iron or oxygen fugacity. Therefore, multiple solutions are possible. We thus decided to use the (FeO/MgO)m ratio obtained via the Maurel and Maurel (1982) equation to explore the maximum FeO and MgO contents of the melt for all available melt-Cr-spinel pairs (table attached below). Based on this ratio, for each Cr-spinel-melt pair we computed the maximum MgO content such that FeOtot can be ascribed fully to FeIIO (i.e., for Fe2O3 = 0). For this maximum MgO content, we thus modelled the effects of PEC, and evaluated its role also on the iron oxide content, searching via the Petrolog3 software package (Danyushevsky and Plechov 2011) for which iron redox conditions the correction for PEC returned the aforecomputed maximum MgO contents (Appendix I). Noteworthy is that, after correction for PEC of olivine in the entrapped melt, the (FeO/MgO)m ratio is no more the one demanded by the Maurel and Maurel (1982) equation, but the one required by the FeO/MgO partitioning between liquid and olivine (Roeder and Emslie 1970). This reflects the implicit decoupling between melt and Cr-spinel throughout the magma evolution at liquidus with olivine.

In order to re-compute oxygen fugacity values, we have to consider the role of water and pressure on iron redox equilibria (Moretti 2005).The pressure estimate and measured water content(Appendix I) of the two MIs here studied come from Moretti et al. (2012), where the Ischia plumbing system has been extensively investigated via determination of major and volatile contents in MIs, and their elaboration to assess pressure of equilibration. Under such hydrous conditions, temperature was set to 1120°C, at melt-olivine equibrium. Finally, application of the iron redox model (Ottonello et al. 2001; Moretti 2005) allows computing the most likely interval for oxygen fugacity in the most mafic Ischia magmas (I_Vat-B-1 ol6; K2O ≈ 2.7  2.9), which is 1.02 NNO  3.20, with the lowest oxygen fugacity value related to compositions corrected for post-entrapment crystallization. On the other side, more evolved melts (I_Vat-A-1 ol3 incl1; K2O ≈ 4.1  4.2) yield a nearly constant oxygen fugacity, independently of the applied correction, comprised in the interval 0.80NNO  0.91. All these conditions individuate a triangular field for Ischia in a K2O vs. NNO diagram (Fig. 10b).

(FeO/MgO)m ratio for Cr-spinels from selected Ischia samples (Appendix B), possible Fe3+/Fe ratios and NNO values resulting for the two MI compositions used. Also reported is the forsterite content of the olivine hosting Cr-spinel, and the MI entrapment pressure and water content (from Moretti et al., 2012).

______

SampleLabel Fo mol% (FeO/MgO)m (*) (Fe3+/Fe)m(§) NNO ($)

______

I-Vat-B_1 ol6C1-sp1 89 0.886 0.4423.15

I-Vat-B_1 ol6C1-sp3 89 0.875 0.4493.20

I-Vat-B_1 ol6Sp1bis 89-85 0.932 0.4142.94

I-Vat-A_1 ol3 incl1ol3OX 83 1.583 0.2300.90

I-Vat-A_1 ol3 incl1C2_sp1 81 1.608 0.2180.80

(*) based on the Maurel and Maurel (1982) function

(§) original EMP analyses from Appendix I

($) based on the model of Moretti (2005)

Reference not listed in the manuscript

Roeder PL, Emslie RF (1970) Olivine-liquid equilibrium. Contrib Mineral Petrol 29:275–289.

1