The solubility of CO₂ in hydrous basaltic andesite was examined in f_O2-controlled experiments at a temperature of 1125 °C and pressures between 310–1200 MPa. Concentrations of dissolved H₂O and CO₂ in experimental glasses were determined by ion microprobe calibrated on a subset of run glasses analyzed by high-temperature vacuum manometry. Assuming that the solubility of H₂O in mafic melt is relatively well known, estimates of 𝑋H2Ofluid and 𝑃H2Ofluid in the saturating fluid were modeled, and by difference, values for 𝑋CO2fluid and 𝑃CO2fluid were obtained (X_CO2 ~0.5–0.9); f_CO2 could be then calculated from the fluid composition, temperature, and pressure.

Dissolved H₂O over a range of 2.3–5.5 wt% had no unequivocal influence on the dissolution of CO₂ at the pressures and fluid compositions examined. For these H₂O concentrations, dissolved CO₂ increases with f_CO2 following an empirical power-law relation: dissolved CO₂ (ppmw) = 14.9−3.5+4.5[f_CO2 (MPa)]^0.7±0.03. The highest-pressure results plot farthest from this equation but are within its 1 standard-error uncertainty envelope.

We compare our experimental data with three recent CO₂-H₂O solubility models: Papale et al. (2006); Iacono-Marziano et al. (2012); and Ghiorso and Gualda (2015). The Papale et al. (2006) and Iacono-Marizano et al. (2012) models give similar results, both over-predicting the solubility of CO₂ in a melt of the Pavlof basaltic andesite composition across the f_CO2 range, whereas the Ghiorso and Gualda (2015) model under-predicts CO₂ solubility. All three solubility models would indicate a strong enhancement of CO₂ solubility with increasing dissolved H₂O not apparent in our results. We also examine our results in the context of previous high-pressure CO₂ solubility experiments on basaltic melts. Dissolved CO₂ correlates positively with mole fraction (Na+K+Ca)/Al across a compositional spectrum of trachybasalt-alkali basalt-tholeiite-icelandite-basaltic andesite. Shortcomings of current solubility models for a widespread arc magma type indicate that our understanding of degassing in the deep crust and uppermost mantle remains semi-quantitative. Experimental studies systematically varying concentrations of melt components (Mg, Ca, Na, K, Al, Si) may be necessary to identify solubility reactions, quantify their equilibrium constants, and thereby build an accurate and generally applicable solubility model.