Hi Gert,

If you know the crystal structure, e.g. space group, and lattice parameters of your material you can create your own material file in the database. If you are using HKL CHANNEL 5 software you can do it in the  "TWIST" tool.

Hope that helps.

This is certainly correct but not the question. If you already have 20 or more different phase descriptions in your databases which is the best one and how you can recognize it? I don't need still another one...:-).

Although EBSD is not the ideal technique for phase identification if you have an unknown material, to answer your question I think you need to check the mean angular deviation (MAD) number. The lower MAD number means a better match between the simulated EBSP and the experimental EBSP. Also you can try increasing the number of Kikuchi bands that is detected and indexed by the software. This increases the chance of your simulated EBSP being unique.

Let me know if that answers your question

MAD is already a good idea although the problem arises how much your pattern center refinement compensates a non-optimal phase selection. Nevertheless, would you really insert e.g. 34 quartz phases in your phase list in order to find out which is the best one? And do you think that the result is independend on the local orientation, i.e. the acquired pattern?

I don't think that the pattern center would have any effect in your case (unless we are talking about different phases with identical space groups and very close lattice parameters). The way that pattern center refinement works is that the software indexes the identified Kikuchi bands and assigns a crystal structure first and then it refines the pattern center position (the default position being saved in the calibration file). The pattern center becomes important only for high resolution EBSD when we need lattice orientation accuracy better than 0.5 degrees.

I'm no mineralogist and I don't know what have you got in your database, but as long as the crystal structures and/or lattice parameters are different I'd trust the results.

And I don't think local orientation changes within one grain (if that's what you mean) influences the results.

Keep me posted

The way a pattern center refinement works is that first ...you are selecting the phase which is perhaps already the first mistake :-). And this is very problematic if you are using only a single pattern since this is perhaps one which is not very sensitive to a c/a or an even more complex ratio, e.g. for triclinic phases. 

The indexing algorithm compares angles between experimentally derived planes (here come the pattern center in) with theoretically calculated, and if there are different phases it looks for the best fit and selects this phase as reliable solution. If you have phases with similar c/a ratios (lets keep at tetragonal as example) the different c/a will be compensated by a variation of the pattern center (here I also consider the detector distance as part of the pattern center). Even for cubic phases where c/a is exactly 1 the pattern center refinement delivers for each single pattern a (hopefully only slightly) different pattern center definition. However, in case you do not exaktly know c/a (since there are many very reliable descriptions in the databases) it becomes therefore difficult since each selected phase will deliver for each single pattern a phase specific MAD, hit rate, orientation description etc. Which is the correct (best) one?

This is the question I was asking for.

Probably the best ones are the ones that fit these criteria:

--Not a phase produced by high pressure, extreme temperature, heavy irradiation, or MBE or some other technique which creates bizarre metastable phases which are of little practical interest but fill the literature since, well since if you can make and solve a strange metastable version of SiO2 then you can publish it.

--Is the reference well cited and in a high impact journal for crystallographic structure reporting?

--When reading the paper does it seem the authors actually know what they are talking about and are talking about a phase that would exist under the conditions that your sample was processed?

After this weeding, I would be surprised if there were many possible candidates left.

Hi Lawrence,

I am still talking about another but from my point of view very irritating question. If you have data bases in your system (e.g. ICSD or AMCSD)...and you know the phase - e.g. quartz - you can search for this phase and you will get...wait a second...127 data base entries for trigonal quartz. They are theoretically all different, either in lattice parameters or in atomic positions. Well, a few of them definitely occur in both data bases. I know that practically I can use each of them, but I also know that the hit rate will slightly change for some of these phases because of tiny differences in c/a ratio...or sometimes also because of the reflector selection (but this should rarely happen). Which one will give the best results? For quartz it is perhaps not a big deal since you can selct always the same and use this in future as well. Moreover, quartz is certainly a "redicules" example since it has a constant chemical composition. For other phases where you have minor constituents or trace elements which slighly shift the lattice parameters the problem becomes more unpredictable.

I think a similar problem also exists in XRD phase ID using implemented search or retrieve algorithms. But perhaps they are perhaps "grouping" phases?

What you mentioned is the fact that you have to deal in general unknown phases which is even more complicated. Indeed, this is a very complex problem as well. The major challange is, that in microstructures small precipitates in a fraction of less than 0.1% may occur so that you have neither an idea about the correct composition (often they are that small that you perhaps get more signal from the matrix than from the precipitate) nor of the non-equilibrium formation of such phases which might be the reson that it hasn't been described in any phase diagram (mostly also because these are multiple element precipitates where no phase diagram exists).

Therefore, sometimes very exotic assumptions of phases are used as analogon to a practically discovered phase which approximately matches the observed diffraction signal. It doesn't automatically means that this applied phase really exists. It only can be misused to determine urientation and local distribution of a "unknown" phase. Of course, this should be pointed out carefully (what is sometimes missing). 

Gert,

Reread my answer. you begin with: "I am still talking about another but from my point of view very irritating question." and then follow by listing problems which I had just addressed. I'm not sure what the other question you were referring to is, since you didn't ask it. Perhaps you asked it in response to my answer? Is it about minor phases of less than 0.1%? If so then that is a very different question, and not the one you asked. So please, what exactly is your question which has not been addressed here by the responses?

Sorry Lawrence. This was certainly my fault. Shortly asked again: If you have a material to investigate by EBSD only. But you know it is a dolomite (and this should be correct). You open your structure database, search for dolomite, and you will get there tens of different dolomite descriptions (different locations, different trace elements, somtimes with some substitutions but still called dolomite) in a list which have slightly different a/c ratios. Which you would select? (You can of course test all of them first but a) usually it works with all of them and b) how reliable is this?)

This practically happens for many phases. For cubic phases it doesn't matter but for all other it may affect your idexing "hit rate" and orientation accuracy. How I can judge them? 

I would go back to my original answer. I often have the same problem since the PDF database has millions of structures. Many of them are outdated and were measured in the 1930's. In this case I will usually go with the most recent if it was published in a reliable journal. With mineral phases there can be many slightly different recent structures which, as you mention, are often due to a solid solution range. This should not change the space group symmetry but just the lattice parameter values. In the EBSD software can you fit the lattice parameter along with the orientations if you know the space group and have a close estimate for the lattice parameters to start with?