Changes to Rhyolite-MELTS ± Fluids

[fboschetty]

Dear Paul and Paula,

I'm hoping to test the Rhyolite-MELTS family of compositions against some evolved experimental compositions from LEPR.
I've read Gualda+ 2012 (10.1093/petrology/egr080) and Ghiorso & Gualda 2015 (10.1007/s00410-015-1141-8) to get a handle on how Rhyolite-MELTS was calibrated for evolved compositions.
I have also had a look through the GitHub and GitLab change logs for alphaMELTS for relevant changes since these two papers were published. The only relevant change I can find is that the Tridymite and Cristabolite pure phase models have been adjusted (https://github.com/magmasource/alphaMELTS?tab=readme-ov-file). Could you give me more detail on these changes and any others that I might have missed?

Best wishes,

Felix Boschetty

[Paula]

Hi Felix,

alphaMELTS corrects the melting point of rutile (see Borisov & Aranovich, 2020), which becomes important for mafic pyroxenites. As the error existed when all the liquid calibrations were done, fixing the pure rutile point can mean rutile appears where you are not expecting it (put another way the "phase absent" constraints on rutile at the time of calibration were too loose). If this is a problem just suppress it - it was virtually impossible to crystalize rutile before so you're not going to miss anything!

As you mention, rhyolite-MELTS adjusts quartz properties to better model the granite ternary minimum and in alphaMELTS we also adjust tridymite and cristobalite to model explosive silicic magmatism on Mars. It's described in the supplementary materials of Payré et al., 2022. You may get tridymite or cristobalite crystallizing where you do not expect it but in this case it is likely down to kinetics in the real system. Again if needed just suppress those phases.

Borisov, A., & Aranovich, L., 2020. Rutile solubility and TiO2 activity in silicate melts: An experimental study. Chemical Geology, 556, 119817. https://www.sciencedirect.com/science/article/pii/S0009254120303569

Payré, V., Siebach, K. L., Thorpe, M. T., Antoshechkina, P., & Rampe, E. B., 2022. Tridymite in a lacustrine mudstone in Gale Crater, Mars: Evidence for an explosive silicic eruption during the Hesperian. Earth and Planetary Science Letters, 594, 117694. https://www.sciencedirect.com/science/article/pii/S0012821X22003302

The feldspar in alphaMELTS was split into plagioclase and alkali-feldspar to simplify bookkeeping. That has a small effect on the modeled sanidine content but well within analytical uncertainty for EMP. If you want something even closer to the original feldspar model e.g. to model the granite ternary minimum, then suppress plagioclase.

The standard state volume of apatite was corrected in alphaMELTS recently, but the difference is negligible.

I believe those are the only model changes versus the ENKI/xMELTS codebase. There are many changes replacing Numerical Recipes code with Gnu Science Library etc. but they should not change the results. If I think of anything else I'll update the post.

Best wishes,
Paula