# X-Ray Diffraction: Phase Identification Strategy

I am having some trouble with qualitative phase identification using x-ray diffraction (XRD). Here follows some basic information about the experiment:

Instrument: PANalytical X'Pert$^3$ Powder

Software: PANalytical Highscore Plus

Database: PAN ICSD

Sample: respirable fraction aerosol samples from ferro- and silica manganese plants.

The goal: which manganese phases are present, and which dominate?

Grasping the software is not what I am concerned about. I lack an analytical strategy. I know how to determine the background, but I am not sure when the background is appropriately determined. Too high? Too low? Does this strongly affect the phase identification?

Then I do a peak search, and 50 peaks are detected. Should I focus just on the major peaks? Are the minor ones important too? When I do a search-match analysis, I get 400 suggested phases, where 20 phases share the highest score. Then another 20 phases share the highest score minus 1. I lack criteria needed to accept a phase as present, and discard other phases as not present.

I am not able to find good literature on this subject. I did find an article called "Hints on Phase Identification Using Powder X-ray Diffraction", which certainly raised a lot of potential problems and procedures, but which did not necessarily solve these problems or explain why the procedures would help. Does anyone know of any books or other literature that could help me?

My problem boils down to lack of "strategy" or "analytical procedure", and maybe someone here could offer some advice. I hope my question is "answerable", and not too broad; I know there is a "specific-and-precise-question" policy in here.

I don't have much information on your particular system: are they nanocrystals? Are you expecting just one or two phases? Dozens of phases? Do you have a lot of amorphous material present?

I have one publication you can look at which addresses a rather complicated sample with amorphous, nanocrystalline, and crystalline objects. It is (hopefully) overkill for your sample, but it will give you ideas on how to partition the problem (and some of the references may help too):

Gainsforth, Z., Brenker, F. E., Simionovici, A. S., Schmitz, S., Burghammer, M., Butterworth, A. L., et al. (2013). Stardust Interstellar Preliminary Examination VIII: Identification of crystalline material in two interstellar candidates. Meteoritics and Planetary Science, 49(9), 1645–1665. doi:10.1111/maps.12148

You may also find it helpful to play with simple fitting softwares such as CrystalMaker/CrystalDiffract where you can try out different mixtures and fits. Sometimes the peak-finding softwares are a little simplistic with a large number of false positives. If you try manually combining the possible phases, you can probably rule many of them out and simplify your list.

A few tips:

1) Look for sets of peaks with FWHM that are similar. This works with nanocrystalline materials. If you have a nanocrystal with a larger crystal, then the larger crystal will have sharp peaks and the nanocrystal will have broad peaks. Then you can select which peaks apply to Phase A and Phase B and simplify your search space drastically.

2) If Peak A is present in your experimental data, and you get a "hit" for phase X, but Phase X also predicts Peak B but you don't see it in your experimental data, then you can rule Phase X out. You may need to understand the characteristics of the phase in a little detail because you can have situations where one atom substitutes for another and you get a change in peak intensities or positions.

3) Do you know the chemical composition of your sample? If one of your "found" phases is a Yttrium based superconductor, and you're pretty sure you don't have Yttrium, then you can drop that phase.

4) Try looking at it with an electron microscope (SEM or TEM). This will provide basic constraints as to whether you have many phases, or just a couple, and how big the crystals are. This is a lot of work, but maybe less work than spending the next month chasing false positives. TEM can also give you diffraction on any phase, no matter how small, and will be pretty unambiguous.

5) You can also provide constraints using techniques such as XANES, NMR, FTIR, etc., depending on what kind of sample you have. Which of these are useful is also dependent on the resources you have present at your university/company/facility.

Hope that helps!

• Thank you for your help. I have just done an element analysis using ICP-OES, which certainly will help. SEM analysis is also planned, along with ion chromatography. I will keep in mind your points; they are very helpful! I have realized that experience is alpha and omega in phase ID! – Yoda Nov 8 '14 at 10:36
• Glad to help! I wish you success with the analysis. – ZSG Nov 9 '14 at 23:09