A different way to estimate H2O mass fraction

[Brian Joy]

Let me refine the “machinery” of the calculations first; I’ll then focus on the user input.  The latter is more likely to produce intractable problems.  If that ends up being the case, then I’ll at least have produced code that I can use for my own purposes .  Perhaps I’ll ask “other Brian” for some advice.

I took a look at the amphibole code written by Ague -- he is using essentially the same approach that I outlined briefly above (the list of possibilities is not very long).  I haven’t looked at his code closely enough, but it looks like he may be using an average of stoichiometrically allowable Fe2O3/FeO calculations in a manner similar to Holland and Blundy* (1994, CMP 116:433-447, Appendix B -- attached); if so, this is bad practice.  I haven’t looked at the biotite code yet.

I’m hoping to make some headway on this in the next week or two.  My workload has been brutal for the past several months, but I think I can see light at the end of the tunnel (or is it just an illusion?).

*I don't like their use of "average Fe2O3" (p. 436) because, even if a recalculation is stoichiometrically allowable, this doesn't necessarily mean that it is geologically reasonable.  Still, Holland and Blundy give a nice, compact summary of the various amphibole recalculations in Appendix B.

[DavidAdams]

I have to say that this topic is really exciting and extremely relevant! As John pointed out Heather Lowers and I have used the amphibole and biotite calculations in PfEPMA quite extensively. It's been noted many times on this forum that the addition of unanalysed elements (H, B, C, etc) can have a profound effect on the matrix corrections affecting the calculated concentrations of other elements in the analysis up to multiple percentage points. For people who are working on very sensitive projects like that of the Libby, Montana amphibole legal case that Heather and I participated in about a decade ago, getting results for things like amphibole not only have implications for scientific/geological accuracy but they also can have legal and regulatory impact. The change of a few percentages of elements in an amphibole can change where that amphibole lies on the classification diagram. This in turn can result in an asbestiform amphibole being regulated or not. Obviously how an amphibole is classified has implications towards industry and human health regulation. There are many, many other examples of why this is important but I won't dive into those here.

Thank you Brian and John for continuing to advance EPMA science!

[John Donovan]

Let me refine the “machinery” of the calculations first; I’ll then focus on the user input.  The latter is more likely to produce intractable problems.  If that ends up being the case, then I’ll at least have produced code that I can use for my own purposes .  Perhaps I’ll ask “other Brian” for some advice.

Hi Brian (I wondered what your profile pic was),
What do you think about the idea of calculating excess oxygen using Droop (which is already implemented) and calculating H2O from a charge balanced formula (my proposal)?  All inside the matrix iteration loop for correction of matrix effects...

[Brian Joy]

Let me refine the “machinery” of the calculations first; I’ll then focus on the user input.  The latter is more likely to produce intractable problems.  If that ends up being the case, then I’ll at least have produced code that I can use for my own purposes .  Perhaps I’ll ask “other Brian” for some advice.

Hi Brian (I wondered what your profile pic was),
What do you think about the idea of calculating excess oxygen using Droop (which is already implemented) and calculating H2O from a charge balanced formula (my proposal)?  All inside the matrix iteration loop for correction of matrix effects...

Profile pic?  It's a guy named Brian who was born in 1969.

Yes, I've written the code in the manner you suggest.  The user has the choice of implementing the Fe2O3/FeO calculation (or not) and then also has the choice of calculating H2O (minus halogens where appropriate) by applying information from the formula unit (i.e., by specifying the number of monovalent anion sites).  It can also handle molecular H2O as a separate entity.  And yes, all of it within the iterations required to determine the matrix correction factors.  Basically I'm exploiting the formula unit for all it's worth.  Note that, in the amphibole example that I present at the beginning of the topic, I've only applied general constraints with no information on site occupancies (and so have ignored Fe2O3).  My goal is to incorporate as much flexibility as possible.

But my code is a mess, and I need to re-think how it should be organized.