Modeling effect of xH2O on feldspar equilibria

[jtepper]

I am using Melts 1.0.2 to constrain the water content of hypersolvus "A-type" granites.  As one would expect, with low (<2 wt.%) H2O contents plagioclase appears late or not at all.  Early crystallization of quartz and late crystallization of biotite are correctly modeled, but none of the runs produce Na-amphibole or aegerine, which are present some the rocks in minor (1-2 modal %) amounts (opx or fayalite appear instead).  My question is whether it is reasonable to trust that the felspar results are broadly correct even if some of the mafic phase results are not.  Any advice on this would be much appreciated.  Thank you.

[Paula]

The problems with amphibole in MELTS are well-known and should not affect other phases e.g:

https://magmasource.caltech.edu/forum/index.php/topic,833.msg1420.html#msg1420

None of the constraints on the original MELTS model involved amphibole and so far as I know none of the rhyolite-MELTS ones did either. The rhyolite-MELTS calibration concentrated on the feldspar and quartz phase relations so I think you can consider those results to be reasonable reliable.

The amphibole project mentioned in the link above did not work out at the time due to Covid. But I am working on it a bit at the moment from the calibration data side - trying to get the experimental results that we have compiled for amphibole into the LEPR database (lepr.earthchem.org). And the ENKI group have been extending MELTS to include F, Cl and S. They will be demonstrating these and other exciting developments at Goldschmidt this summer:

https://conf.goldschmidt.info/goldschmidt/2025/meetingapp.cgi/Session/7308