fractional crystallization-oddity with cpx

[garnet2014]

I've been trying to calculate an ideal liquid line of descent for a set of basaltic glass samples using the amoeba program. However, the LLD calculated using the primary melt composition I get from the amoeba routine has a skip at approximately T=1150, where olivine stops crystallizing and cpx comes in and then about 6 T steps later olivine comes back in to crystallize with cpx. This doesn't really make sense to me but I'm not sure how to fix the problem.

Here's the MELTS file I've been using from the amoeba result:

Title: Calculated liquid composition
Initial Composition: SiO2 47.817632
Initial Composition: TiO2 2.246321
Initial Composition: Al2O3 14.807020
Initial Composition: Fe2O3 0.822715
Initial Composition: FeO 9.324307
Initial Composition: MnO 0.168945
Initial Composition: MgO 9.880000
Initial Composition: CaO 11.471355
Initial Composition: Na2O 2.315975
Initial Composition: K2O 0.324780
Initial Composition: P2O5 0.366020
Initial Composition: H2O 0.454931
Initial Mass: 100.000000
Initial Temperature: 1241.635462
Initial Pressure: 1000.00
Log fO2 Path: None
Suppress: orthoamphibole
Suppress: chlorite
Suppress: knorringite

The environmental conditions I've been using are from the fractional crystallization example I've gotten from the website. Here are the set variables:
ALPHAMELTS_VERSION         MELTS
ALPHAMELTS_CELSIUS_OUTPUT   true
ALPHAMELTS_MODE          isobaric
ALPHAMELTS_DELTAP        0
ALPHAMELTS_DELTAT        -1
ALPHAMELTS_MAXP          30000
ALPHAMELTS_MINP          1
ALPHAMELTS_MAXT          2400
ALPHAMELTS_MINT          1000
ALPHAMELTS_FRACTIONATE_SOLIDS   true

Any advice would be greatly appreciated.

[Paula]

Hi,

I tried running your files in alphaMELTS. Olivine went out and cpx came in at ~1160 ^oC. A lot of cpx was added all at once, which suggested that the routines that detect phase saturation were not working for cpx. I checked and it's not an outright failure of those routines, but the estimated affinity and composition of cpx must be off. alphaMELTS uses the older algorithm (Ghiorso 1994), and the simplifying assumption of ideal mixing between end members is not so reliable for phases that have lots of dependent end members and / or ordering.

Rhyolite-MELTS has faster and more robust routines for detecting general and specific phases, including pyroxenes (see Ghiorso 2013). Running your melts_file in Rhyolite-MELTS, cpx joins the assemblage at ~1170 ^oC and olivine continues to precipitate. At higher temperatures, the alphaMELTS and rhyolite-MELTS results agree properly. Despite the wrong assemblage in the alphaMELTS version, the final liquid compositions at 1000 ^oC are similar. Looks like the biggest differences are in TiO2 and Fe2O3.

I am incorporating the Rhyolite-MELTS routines into the next version of alphaMELTS at the moment (see the Library tab on the alphaMELTS website). I will try to get something out before going on leave but there won't be any support for it for a while, and it is unlikely to include amoeba until after I get back. In the meantime, I suggest using rhyolite-MELTS for your 'final' results (move the 'Initial Mass:' line after the fO2 one; that way it will generate a warning in the GUI but not actually affect the calculation). Alternatively, your could follow the workaround described here. I know it's not ideal that the phase saturation routines are misbehaving during the amoeba calculation too but hopefully that is not affecting the calculated primary melt composition too adversely.

Sorry I don't have a proper solution right now.

Happy Holidays!

Paula