If I have a mineral in which I did a SEM analysis on and got the data (elements and their wt% with Std. Dev., MDL, atomic % oxide % and k ratio) how would I go about quantifying the sample's composition? I know I would need standards to do so but I not sure how to go about this. Can someone please assist? Thank you!


2 Answers 2


You would be lucky if you could identify it with only elemental composition. XRD would be the best method to identify material ( the elemental composition is also needed ). Our lab normally used XRD and XRF , and only went to the SEM when special information was needed.

  • $\begingroup$ I agree on X-ray diffraction to check what you have in your sample before SEM. You get structural data and probably the name of mineral, it's really useful in case you have polymorphs. On the other hand, some SEMs can do electron backscatter diffraction and that helps with structure as well. $\endgroup$
    – Gabija
    Dec 3, 2017 at 11:31
  • $\begingroup$ I expect the difference in answers may come from the type of samples to be evaluated. We evaluated mostly corrosion products and related alloys and minerals. basically the SEM /EDS could not do it as well as XRF +XRD , and still the techs would spend time looking in the Fink Index. But I loved the SEM , it was great looking at various layers in a corrosion deposit . $\endgroup$ Dec 4, 2017 at 20:55

I have to disagree with with @blacksmith37and @Gabija. Using an SEM is one of the best methods there are to identify minerals.

XRD is only useful if you have a nice single crystal, or if you have the material powdered. If it's a mineral in a rock, then the powder will be a mixture of everything, unless you can separate the mineral out. SEM-EDS analysis (which is what you probably did) gives you the chemical composition of the mineral, which is usually sufficient for identification.

Some cases where it doesn't work:

  1. Polymorphs. Quartz or coesite, calcite or aragonite. They will have the same composition. But ideally you have some prior information on the rock so you know what to expect.
  2. X-ray energy overlap. Some elements overlap in the energies, for example Mo, Pb and S. You will only get one peak for them, making it hard to distinguish. But, usually galena (PbS) and molybdenite (MoS2) look so different, so it shouldn't be too hard. Another example is Ti and Ba. If you analyse a feldspar and you get some Ti, it's most likely to be Ba because that's what goes into feldspar, not Ti.
  3. Light elements. It's not possible to analyse Li and Be so no spodumene or beryl.

Even though there are problems, it's still an excellent way to do it.

I know I would need standards to do so but I not sure how to go about this

Too late for that. Measuring standards is part of the instrument's calibration, and you do that before you analyse your minerals. It might have already been done by the SEM operator - just ask them. Unfortunately, many SEMs have standard-less EDS systems, where it's calibrated once in the factory and that's it. Fortunately, the results you get from such systems are surprisingly good, particularly for simple minerals. The problem is that they normalise the results, so it's hard to know if you have a hydrous mineral or just a bad analysis.

how would I go about quantifying the sample's composition?

You already have everything. Mineral compositions are usually reported in oxide wt%, so you get this information straight off the SEM. Mineral formulas are a bit more tricky. You first need to know which minerals it is, and then you need to know how much oxygens it has. For example, olivine has four oxygens: Mg2SiO4. Then you need to normalise your (atomic%) data to that number of oxygens. Here's a website that explains this very well:


And in general, a web search for "calculating mineral formula" results in many sites that have worked out examples on how to do it.

If you have the composition, but you have no idea what mineral it is, there are two websites that can assist you:

  1. http://webmineral.com/ This site lists the composition of each minerals in oxide% and wt%, so you can compare it with your analyses.
  2. https://www.mindat.org/ This site has a search by chemistry. You just type in the elements you have from your analysis and it gives you a list of minerals that have them. Then try to look at the relative proportions of you atomic% to try and figure out which mineral it can be.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.