The Influence of Undercooling and Sector Zoning on Clinopyroxene–Melt Equilibrium and Thermobarometry

Abstract

Abstract Thermobarometry provides a critical means of assessing locations of magma storage and dynamics in the lead up to volcanic eruptions and crustal growth. A common approach is to utilise minerals which have compositions sensitive to changes in pressure and/or temperature, such as clinopyroxene which is ubiquitous in mafic to intermediate magmas. However, clinopyroxene thermobarometry may carry significant uncertainty and require an appropriate equilibrium melt composition. In addition, the degree of magma undercooling (ΔT) affects clinopyroxene composition and zoning, with common sector zoning potentially obfuscating thermobarometry results. Here, we use a set of crystallisation experiments on a primitive trachybasalt from Mt. Etna (Italy) at ΔT = 25 – 233 °C, P = 400 – 800 MPa, H2O = 0 – 4 wt.% and fO2 = NNO+2, with clinopyroxene crystals defined by Al-rich zones (prisms and skeletons) and Al-poor zones (hourglass and overgrowths) to assess common equilibrium models and thermobarometric approaches. Under the studied conditions, our data suggest that the commonly applied Fe-Mg exchange (cpx-meltKdFe–Mg) is insensitive to increasing ΔT and may not be a reliable indicator of equilibrium. The combined use of DiHd (i.e., CaMgSi2O6 + CaFeSi2O6) and EnFs (Mg2Si2O6 + Fe2Si2O6) models indicate the attainment of equilibrium in both Al-rich and Al-poor zones for almost all investigated ΔT. In contrast, CaTs and CaTi models reveal substantial deviations from equilibrium with increasing ΔT, particularly in Al-rich zones. We postulate this reflects slower diffusion of Al and Ti in the melt compared to Ca and Mg and recommend the concurrent application of these four models to evaluate equilibrium between clinopyroxene and melt, particularly for sector-zoned crystals. Thermobarometers calibrated with only isothermal–isobaric experiments closely reproduce experimental P–T at low ΔT, equivalent to natural phenocrysts cores and sector-zoned mantles. Models which also consider decompression experiments are most accurate at high ΔT and are therefore suitable for phenocryst rims and groundmass microlites. Recent machine learning approaches reproduce P–T conditions across all ΔT conditions but carry larger uncertainties. Applying our experimental constraints to sector-zoned microphenocrysts and groundmass microlites erupted during the 1974 eccentric eruption at Mt. Etna, we highlight that both hourglass and prism sectors are suitable for thermobarometry, given that equilibrium is sufficiently tested for. The combination of DiHd, EnFs, CaTs and CaTi models identifies compositions closest to equilibrium with the bulk melt composition, and results in smaller differences in P-T calculated for hourglass and prism sectors compared to applying only DiHd and EnFs equilibrium models. This provides a framework to tackle crystallisation conditions in sector-zoned clinopyroxene crystals in mafic alkaline settings.