Some confusion about the EDS quantification with Hyperspy

[Michael]

Dear all,

While doing the EDS quantification with Hyperspy recently, we have been encountered some confusion as follows:

(1) For proper quantification, should all the X-ray lines belong to the same family be involved in the calculation? For example, in our case of SiGeSn thin film, the Hyperspy uses only the Si_Ka, Ge_Ka and Sn La instead of their whole family lines (e.g. Sn_Lb1, Lb2, and Lg...) to calculate the composition, which is sort of different from some other commercial softwares. Is this correct?

(2) Concerning the Cliff-Lorimer k-factors, normally we choose them either  from the database in the Bruker software Esprit (most of the time) or from other available references (e.g. textbook from Williams and Carter). However, it seems that there are some differences between them, which for sure will influence the final quantification. I wonder which one will be the best way to obtain  reliable results.

By the way, the k-factor is normally for the whole shell. Would it be reasonable to have a k-factor for only a specific line, for example La? 

It will be very appreciated if any EDS expert can give some useful suggestions.

Thanks in advance

[Michael]

 

 Just forgot to mention that the method we used to calculate the composition is Gaussian fitting.

Michael於 2019年3月25日星期一 UTC+1下午5時50分44秒寫道：