Thermodynamic modeling tools for the interpretation of melt inclusions and volcanic gases

Kayla Iacovino , Ery Hughes, Penny Wieser, Alain Burgisser, Philippa Liggins, Shuo Ding, Chenguang Sun, Geoff Kilgour

H2O, CO2, and S are the most abundant volatiles in basaltic magmatic systems and are critical to understanding magma storage and the size and style of volcanic eruptions. Models for calculating melt–vapor (±mineral) equilibria are abundant in the literature but are not interoperable. Consequently, few comparisons of model outputs have been performed. To address this, we comprehensively compare five H-O-C-S melt–vapor equilibrium tools (D-Compress, VolFe, EVo, Sulfur_X, MAGEC) to evaluate how and why they diverge. We present a series of degassing scenarios for MORB, Kīlauea, Fuego, and Fogo basalts revealing that often understated model assumptions such as fO2 buffer equations and fO2–Fe3+/ΣFe relationships have outsized effects on results. All models consider both a reduced and oxidized sulfur melt species but with different approaches to incorporating their solubility. Differences in model implementation lead to clear output divergence of modeled gas compositions, melt S and Fe speciation, and fO2. We also compare tool outputs with those of S-free models MagmaSat, VolatileCalc, and Iacono-Marziano using VESIcal and find that the addition of S has little to no effect on the calculation of the vapor saturation pressure (Pv_sat). Conversely, the calibration dataset underlying an H2O–CO2 model implemented into these tools exerts the strongest control on Pv_sat particularly for compositions less represented in the experimental literature (here, Fuego and Fogo). We recommend that new high temperature and pressure solubility experiments target the conditions where models most severely disagree rather than aiming to "fill in the gaps" in parameter space. The success of future assessments of model suitability rely on a continued push for regular evaluation of code usability, transparency, interoperability, and benchmarking. These factors must be the codified pillars of the peer-review process, providing the confidence for wider community uptake and implementation.