Experimental and numerical simulations of Li isotope fractionation during H2O degassing of rhyolitic magma

Ascending hydrous magma in nature often undergoes segregation of vapor phase. Given that this is a fundamental driving force of volcanic eruption, there have been many efforts to understand degassing process of magma. In this study, we have focused on the aspect of Li segregation during degassing simulated in laboratory experiments. Natural rhyolitic obsidian was pre-saturated in H₂O at 210 MPa, 800 ˚C giving approximately 6 wt%. Subsequently the pre-saturated samples were isothermally decompressed at the rate of 500 kPa/s. The final pressure before quench was approximately 70 MPa. The pressure of bubble nucleation in these experimental conditions is equal to 90 MPa, so the duration of bubble growth was 40 sec. Some experiments were kept at constant pressure conditions immediately after decompression for up to 3 days.

In the sample quenched directly at the end of the decompression step, the Li abundance in degassed melt varies between above and below the initial concentration (71 ppm): from 69 to 74 ppm. The isotopic composition of Li is heavier by 3 to 5 permil compared to the initial. The sample annealed for 3 days shows similar heavy isotope values, but Li abundance is only higher than the initial, no concentration depletion is detected. Numerical simulation accounting for bubble growth and Li diffusion shows profiles in melt that are variable in concentration, with isotopically heavy value due to diffusion fractionation. Specifically, significant variation of Li concentration and shift to the heavy value are favored for the condition at which Li is compatible to gas phase over melt. When Li is incompatible, insignificant quantify of Li leaves the melt thus resulting in negligible isotopic fractionation. The profile caused by degassing relaxes during steady state anneal. This relaxation yields patterns consistent with the composition observed in the degassing-anneal run. Our experimental results also indicate that gas-melt partitioning of Li must be between 1 and 10. Further tuning of the model is required to determine the range of Li partition coefficients and equilibrium isotope fractionation factors that are consistent with observations.