Hi Sarah,

I think this is a typical matrix effect issue that is related to the ZAF correction, in short, the material used in the calibration standard and your unknown samples should have similar (ideally identical) ZAF factors so they would be cancelled out in the concentration calculations. These ZAF factors are related to material properties and compositions. Gabi wrote a very detailed page about this on the Rice website (https://eeps.rice.edu/about-epma-method), highly recommend to have a check!

Yes, but then why do I get basically the same results when I use Diopside rather than San Carlos olivine as primary standard.

My feeling is if it was a matrix effect, I should get good result on the natural olivine (and san carlos as unknown) when I use san carlos as a primary standard and bad results with diopside.

But I get good results on San calors olivine and bad results on natural olivine independently of the primary standard I use...

That being said, I'm going to read this post in details

Then I guess both Di and Ol are in a crystalline silicate structure so the ZAF is similar but not exactly identical, and a silicate glass may have a different silicate structure comparing to the crystalline phase. I guess you could test by using a pure Si metal as primary standard, you should also get different results given the material is in different structure