Compositional gap when spinel crystallizes

[AdrienBoucher]

Hello, I am trying to compute the differenciation of a basaltic-andesitic melt and get some results that I am not sure to understand.

First of all, the composition I am using is that of a late archean feldspar macrocrysts bearing flow from which I removed the macrocrysts prior to analysis (when runing calculations with a flow still containing those macrocrysts I get feldspar crystallizing first, leading to an initial increase of MgO in the residual melt, the results are much more consistent with the macrocrysts removed). The sample is relatively fresh (according to mass balance calculations using Trépanier et al. 2016 routine).

As most of the other samples I collected were altered I am specifically looking at elements such as Cr, Ni or at ratios such as Zr/TiO2 in order to infer the differenciation level of each flow in the sequence I am studying.

The initial parameters I am using are those :

ALPHAMELTS_DELTAT -1
ALPHAMELTS_MAXT 2400
ALPHAMELTS_MINP 1
ALPHAMELTS_FRACTIONATE_SOLIDS true
ALPHAMELTS_MINT 500
ALPHAMELTS_MODE isobaric
ALPHAMELTS_DELTAP 0
ALPHAMELTS_MAXP 30000
ALPHAMELTS_CELSIUS_OUTPUT true
ALPHAMELTS_VERSION MELTS
ALPHAMELTS_DO_TRACE true


The input file is :

Title: UCB0005CG02
Initial Composition: SiO2 48.9
Initial Composition: TiO2 0.91
Initial Composition: Al2O3 15.65
Initial Composition: Fe2O3 12.05
Initial Composition: FeO 0.00
Initial Composition: K2O 0.02
Initial Composition: MnO 0.18
Initial Composition: MgO 8.26
Initial Composition: CaO 10.05
Initial Composition: Na2O 1.72
Initial Composition: P2O5 0.06
Initial Composition: H2O 0.2
Initial Trace: Cr 300.0
Initial Trace: Ni 162.0
Initial Trace: Th 0.19
Initial Trace: Nb 2.2
Initial Trace: Ta 0.4
Initial Trace: Zr 50.0
Initial Trace: Hf 1.4
Initial Trace: La 2.7
Initial Trace: Ce 7.4
Initial Trace: Pr 1.21
Initial Trace: Sr 103.5
Initial Trace: Nd 6.4
Initial Trace: Sm 2.15
Initial Trace: Eu 0.78
Initial Trace: Ga 17.6
Initial Trace: Gd 2.92
Initial Trace: Tb 0.5
Initial Trace: Dy 3.47
Initial Trace: Y 20.5
Initial Trace: Er 2.38
Initial Trace: Tm 0.34
Initial Trace: Yb 2.53
Initial Trace: Lu 0.34
Initial Temperature: 1500.00
Final Temperature: 1000.00
Increment Temperature: 3.00
Initial Pressure: 500.00
Final Pressure: 500.00
Increment Pressure: 0.00
dp/dt: 0.00
log fo2 Path: FMQ
Log fO2 Delta: 0.00
Mode: Fractionate Solids

When looking at the results of the calculations everything seems fine (except for a slight increase of Cr during the first steps of calculation as olivine, which is first to crystallize, takes much less Cr than clinopyroxene), but when spinel starts to crystallize, a compositional gap in several elements such as SiO2, TiO2, Zr, Ni or Cr, leading to discontinuity in the model.

I runned the calculations again with spinel supressed and got smoother results, but Cr remains pretty high for intermediate/felsic compositions and I get a rhm-oxide crystallizing instead of spinel. I am not sure to grasp all the implications of this.

Could anyone enlighten me?

[matrix]

Hi, I found another similar case in which certain mineral disappeared (mineral gap) under some conditions where the mineral was supposed to be there as some researchers suggested. That mineral is just suddenly gone completely. I can not find the causes. But there is indeed some cases where alphaMELTS would not converge even with smaller P/T step. The gaps are filled by certain algorithm (e.g., nearest neighboring algorithm). See the paper "Evolution of young oceanic lithosphere and themeaning of seafloor subsidence rate", Korenaga, 2016. You can contact Paula directly for suggestions, she's very kind.




Best,
matrix

[Paula]

Sorry I didn't see this post/reply before. Do you mean discontinuities in liquid composition (or discontinuities in mineral composition)?

When spinel comes in, does a lot come in, all in one go? It sounds like something the improved algorithms for detection of phase saturation in Rhyolite-MELTS might solve. You could run the Rhyolite-MELTS GUI or alphaMELTS 2 to test (neither has trace elements yet though). There are workarounds in alpahMELTS 1.X also, but they are cumbersome, so it would worth checking first.

Paula