Oxygen and silicate glass unknowns

[bgarcia]

Hey All!

I've been analyzing some highly-variable silicate glasses (unknown stoichiometry), using USGS glass standards, and have been getting great oxygen numbers on both standards and unknowns. The person I'm getting the data for has asked to view the numbers in wt % oxide, as that's what she's used to dealing with. Ok fine, no problem.

I recalculate the data into oxides, and lo and behold, I now have anywhere from 7-15 wt% EXCESS oxygen, because the analyzed oxygen is less than the calculated oxygen. What are some explanations for this?

I'm using MPB for the oxygen, and getting good numbers of my glass standards, so I don't have a reason to suspect this is a background setting issue. I've also used the TDI function to minimize charging/damage, but there doesn't seem to be a huge oxygen change, regardless.

I've also been trying to find published data on possible water contents, but the record is sparse. In the few papers I have on this type of glass I'm looking at, no one has looked for water or OH. Also, these glasses were explosively formed so I'm not sure there's any water in them.

How does one deal with excess oxygen?

I've attached an unknown glass analysis, calculated as elemental and as oxide.

From John D.:
 "Also why did you want to analyze for oxygen in the first place? Did you suspect excess oxygen from Fe2O3?"

Nothing so clever, John! I never thought about just doing stoichiometry; my default is usually to try and measure, and then when things go poorly, rely on stoichiometry. Maybe I should't have done it? The expected composition of the glass is broadly silicate, so I didn't see any big hurdles in measuring O.

[Probeman]

Hi Ben,
Some of the largest artifacts in oxygen analyses occur in silicate glasses. E.g., peak shape effects (e.g., when using MgO or Fe2O3 standards for oxygen), MAC issues, beam damage, and other problems, some of which are mentioned in this recent post:

https://probesoftware.com/smf/index.php?topic=1061.msg9282#msg9282

So no, I would not assume that analyzing oxygen is without "hurdles".   

That said, if one works really hard, one can get good oxygen analyses in glasses as seen in this post of some dry to hydrous glass analyses:

https://probesoftware.com/smf/index.php?topic=922.0

I would suggest reading this post and the other posts linked to it very carefully.

You also mentioned you "have been getting great oxygen numbers on both standards and unknowns", but with 106% totals these analyses certainly do not look "great" to me.

Finally, from the negative oxygen numbers visible in the oxide output, you are not getting excess oxygen, you are getting *deficit" oxygen.  Meaning that you are not even measuring enough oxygen to support your cation stoichiometry assumptions.  But since you already have high totals, something is very wrong with these analyses!

All in all, this can be done, but you've got a lot of work ahead of you.  I would start by reviewing the above posts, but there are many ways to proceed with oxygen analysis, not the least is performing "integrated" intensity scans instead of just measuring the (on) peak intensity. This method will take care of the peak/shape and shift problems. The downside to scanning the oxygen peak is that it will dramatically slow down your acquisitions.  But then there's still TDI and MAC issues to deal with.

Oxygen analysis of glasses is not an easy thing to do in my book.  But it can be done, you'll just need to throw every tool "in the book" at the problem!