Equilibrium Thermodynamic Modeling of Arclogites: Are We There Yet? An Example From The Andean Volcanic Zone (NVZ)

Abstract

The Northern Volcanic Zone (NVZ) in the Andes is a continental magmatic arc emplaced in >50 km-thick continental crust. In S Colombia, a Pleistocene eruption known as the Granatifera Tuff exhumed a variety of (ultra)mafic xenoliths derived from lower-crust and mantle, providing a unique opportunity to study petrologic processes operating in the roots of the Andean arc. Here, we focus on garnet-clinopyroxenite (aka ‘arclogite’) xenoliths, which are lithologies expected to drive density instabilities in arcs. These rocks consist predominantly of garnet (30-57%) and clinopyroxene (20-66%), with amphibole as an additional primary phase (10-47%) in some cases.Phase equilibria modeling of pyroxenites using free-energy minimization tools such as MELTS has been shown to be troublesome, but more recent optimizations of thermodynamic databases are yet to be tested. We used Perple_X coupled with recent thermodynamic databases and solution models to calculate pyroxenite phase equilibria to: i) test the extent to which these results agree with phase relations and partial melting conditions of pyroxenites for which experimental data exist; ii) compare the results of the pyroxenite models from the NVZ with traditional thermobarometry, to confirm previous suggestions that the Mercaderes xenoliths represent a ‘hot’ end-member of arclogite localities worldwide; and ii) better understand the petrology and density structure of the lower NVZ arc. For amphibole-free xenoliths, we constructed P-T diagrams with garnet compositional isopleths to obtain P-T conditions. For amphibole-bearing xenoliths, we created isobaric T-X(H2O) diagrams using previous P estimates, to evaluate the predicted stability of amphibole as a function of water content. For all samples, diagrams were first created assuming all iron as ferrous (FeOT) from bulk-rock XRF data. Ferrous iron was then gravimetrically determined in whole-rock aliquots using potentiometric titration, and ferric iron contents were calculated using mass balance. With these results, new phase diagrams were recalculated using appropriate Fe2O3/FeO for each, to evaluate the effects of Fe speciation in the models. Our calculation results from MIX-1G, a pyroxenite for which extensive experimental data exist in the literature, yield better agreement between recent thermodynamic databases and experimental constraints than in the past, indicating that recent optimizations of thermodynamic data are more suitable for the study of nominally anhydrous pyroxenites. For amphibole-free (anhydrous) NVZ arclogites, we found the P-T conditions of low Fe3+/Fe2+ samples to agree well with previous P-T estimates when Fe2O3 is taken into account. However, the quality of compositional isopleth intersections degrades for samples with high Fe3+/Fe2+, indicating Fe speciation is an important parameter to consider in more oxidized arclogites but that existing solution models may not yet be completely reliable handling Fe3+ in these models. Calculation results for OCA-2, an amphibole-bearing pyroxenite for which experimental data exists, shows that phase assemblages and melting could not be replicated by the model. Furthermore, results from amphibole-bearing Mercaderes pyroxenites do not predict amphibole to be stable at the P-T conditions established from equilibrium thermobarometry, therefore precluding independent verification of the accuracy of the latter. From a more general perspective, our results indicate that modeling of amphibole-bearing arclogitic cumulates remains a challenge, and thus that P-T-X modeling of amphibole bearing arclogites using existing databases should be approached with caution or altogether avoided until experimental results can be shown to be accurately replicated.