Bubble formation and decrepitation control the CO 2 content of olivine‐hosted melt inclusions

The CO 2 contents of olivine‐hosted melt inclusions have previously been used to constrain the depth of magma chambers in basaltic systems. However, the vast majority of inclusions have CO 2 contents which imply entrapment pressures that are significantly lower than those obtained from independent petrological barometers. Furthermore, a global database of melt inclusion compositions from lowH 2 O settings, indicates that the distribution of saturation pressures varies surprisingly little between mid‐ocean ridges, ocean islands, and continental rift zones. 95% of the inclusions in the database have saturation pressures of 200 MPa or less, indicating that melt inclusion CO 2 does not generally provide an accurate estimate of magma chamber depths. A model of the P ‐ V ‐ T ‐ X evolution of olivine‐hosted melt inclusions was developed so that the properties of the inclusion system could be tracked as the hosts follow a model P ‐ T path. The models indicate that the principal control on the saturation of CO 2 in the inclusion and the formation of vapor bubbles is the effect of postentrapment crystallization on the major element composition of the inclusions and how this translates into variation in CO 2 solubility. The pressure difference between external melt and the inclusion is likely to be sufficiently high to cause decrepitation of inclusions in most settings. Decrepitation can account for the apparent mismatch between CO 2 ‐based barometry and other petrological barometers, and can also account for the observed global distribution of saturation pressures. Only when substantial postentrapment crystallization occurs can reconstructed inclusion compositions provide an accurate estimate of magma chamber depth.