Hi Vivian, I'm not certain if I understand your question. By "assign," do you mean that you positively and unambiguously identify a material using only compositional analysis techniques, such as SEM-EDS?

If that is the case, then your question really depends on what you mean by "unknown." For instance, let's say you have four pieces of glass from a manufacturer. These amorphous glasses differ only in their compositions, and those nominal compositions are given by the manufacturer. Perhaps all four pieces are from the same glass, or perhaps each piece comes from a different glass. In this instance, SEM-EDS is an incredibly useful tool for identifying these materials. Here, what is unknown is the identity of the glass, and the assignment is made based on compositional analysis and comparison to the given composition.

Let's say that you have a set of olivine minerals with a range of compositions. Olivine forms a solid solution with O, Mg, Si and Fe, with the two end members of the solution being Fayalite (Fe2SiO4) and Forsterite (Mg2SiO4). Again, SEM-EDS compositional analysis will allow you to determine where each olivine mineral occurs in this solid solution. Here, the unknown is the position of the mineral along the solid solution curve.

Now let's say that you have three polymorphs of titanium oxide, rutile, anatase and brookite. All three of these samples have the identical composition, so their SEM-EDS spectra will be very similar. What is unknown is the crystal structure, and pure SEM-EDS will not allow you to determine this. In this case, SEM-EDS cannot be used to identify the material. Similarly, if your compositional analysis shows that you have titanium oxide, that is not sufficient to positively identify an unknown mineral as rutile.

Finally, let's say you have a true unknown. It is a lump of something, and you have no idea where it came from or what it is. In this case SEM-EDS can provide a lot of information, but it can't be the only technique used to identify the material. SEM-EDS is not sensitive to crystal structure, bonding state, isotopic composition or the presence of hydrogen. It is a surface technique, telling you only about the composition of the first few micrometers of a sample. It provides complementary and supplemental information, but only as a part of a much larger, multi-technique study. Laboratories that specialize in this type of analysis typically use mass spectrometry, X-ray diffraction, neutron activation, NMR, FTIR, light and electron microscopy, among other techniques. It is only after careful review of the data from multiple sources that any material identification can be made.

To summarize, there are a limited number of cases where SEM-EDS can be used to positively and unambiguously identify a material. Generally, this is when you have a limited set of unknowns that vary in their compositions. There are materials, such as glasses, where composition is a major clue as to the identity of the material. In all other cases, the SEM-EDS data will provide clues to the composition of the surface of the material, but this information must be combined with information from other techniques to positively identify a material.

I mean that is it possible to assign multiple compositions to each unknown when using SEM-EDS? The question might focus on how to find the multiple materials in the unknown if just only using these techniques? Can it be posible or not to do so?

I'm afraid I still do not understand your question. What do you mean by "assign?" Do you mean that you can you can group similar compositions together and state that the unknown material has a range of compositions, as identified by the compositional variance of the spectral group?

For instance, let's say we have two "unknown" materials in the specimen, A and B. We have 10 spectra we have measured with SEM-EDS, numbered 1 to 10. Looking purely at the spectra themselves, we arrive at two groups:

Group A: 1,3,5,6,7

Group B: 2,4,8,9,10

We notice from the analysis that the spectra in Group A were taken from a region of the sample containing specimen A, likewise for B.

Are you asking if it is possible to make the following statement:

The composition of specimen A is in a range of X-Y, given the compositions of Group A.

Vivian, it sounds to me like you are asking about phase variation within a single sample. This is possible using a combination of mapping, point-EDS, and a backscatter detector. More detail on what you are looking at would make it easier to provide an answer.

Hello Jeffrey,

I mean that can I assign multiple compositions, for example, if an unknown is dopted with something else or a sample possibly composite of multiple materials by using SEM-EDS?

The most general answer is "no". EDS provides you with some information on elemental composition in some volume below the surface of your sample. This is all. You don't know much about this volume, you don't know much about the crystal structure, etc. In fact, you can only compare one part of your sample with another. If you are lucky (if some conditions are fulfilled), you can compare one sample to another, "reference" sample.

What are your possibilities? First, you may assume the crystal structure of your sample by analogy with some reference study (published earlier). Or you can check the phase composition of your sample with some other technique, say, XRD. Second, use your EDS results not directly, but as a difference between different areas of your sample. Find a point on the surface of your sample that seems most representative (most ordinary, most standard) to you. Compare compositions of other surface features (EDS data) with this "standard" point, find defferences. Compare the differences with your data on phases present in your sample (XRD data), or with your assumptions (based on literature).

If you need more reliable results, use EBSD (diffraction of back-scattered electrons) instead of EDS. Or, better, a combination of these techniques.

Hi Peter, how about combining SEM-EDS and Raman Spectroscopy to identify them?

Raman, as much as I remember, provides overall composition of your sample. It gives you less than necessary: you can only guess which phases (chemical and crystal structure) are giving the numbers you see. Maybe if your structure is very simple, or really well studied, then taking into account this preliminary knowledge you may guess composition of the parts of your sample based on your EDS data.

Hi Peter,

My sample is coated with Al support stubs using epoxy. So in the Raman, peaks signal of elements is the epoxy might be overlapped with some elements in my unknown. I have Raman spectra for both epoxy and my unknown. I am wondering that how to interpret the peak signals to identify the chemical structure of my unknown based on Raman. Besides, I already knew the chemical compositions of my unknown by using SEM-EDS. May you take a look at my spectra and share your thought? Thank you.