And the overlap 2theta position is ~15, there are TiO2#Rutile, Co-FCC and Co-HCP here.

Dear Lipeng Yao,

In GSAS-II, the peak broadening contributions from size and strain are coupled. A high strain in Co may lead to an apparent increase or decrease in crystallite size. If Co has a strong preferred orientation, peak broadening may not be uniform across different reflections. These factors may be the main causes of the discrepancy in crystallite size.

Other factors, such as the use of different peak shape models (e.g., Gaussian, Lorentzian, or Voigt), can also affect crystallite size determination, so ensure that the same peak shape function is used for all phases. Additionally, check the background subtraction, as improper background fitting can influence peak width and, consequently, crystallite size determination.

I am not able to open GPX files. Do you have a text version that you can share?

Why would it be a surprise that the TiO2 and Co phases be of different sizes? Have you done any electron microscopy on this material to check the sizes? It seems that might be easier than trying to do the entire analysis by XRD. If you see a spread in particle size by EM, I would be more inclined to believe those results than ones derived from lots of mathematical fitting to an XRD pattern.

Fitting is good if you really understand all of the components of peak width and your sample and your instrument. Does your treatment give the right answers for a standard sample? If not, then I would go easy on applying the technique to an unknown sample.

I have seen lots of poorly applied examples of mathematical fitting in the literature of late. There should be some connection to physical reality rather than just turning a fitting routine loose without constraint. (I am not saying you are doing that.) Recent examples of fitting XPS peaks have failed to remember that there are constraints to peak width and position. A recent example of fitting a broad peak in an XRD pattern appeared to allow peaks positions and widths to vary all over the place in order to get the best fit. It should not have. That is why I ask how you are able to handle a known system. Do you find the right answers for it?

In addition to the other comments:

Why do you expect Co size to be same as TiO2 size?

It is common for different phases to have different crystallite sizes.

Also, Co metal may very likely have greater dislocation density (greater microstrain) than the oxide phases, which will also broaden peaks, but with a different angle dependance than size broadening. So, be sure to correctly account for size vs microstrain broadening.

Somesh Chandra

Hi Somesh,

Thank you very much.

Yes, you are right. From the software, I can get the size of different phases, phase fraction, weight fraction and microstrain. Now, I am trying to separate the overlap peak at 2 theta ~ 15, in this position. TiO2#Rutile, Co-HCP and Co-FCC.

Warren Straszheim

Hi Warren,

Thank you very much.

Of course, the TiO2 particle size is different with Co metal. Sorry not to explain the question so clear. I want to analysis the overlap peak at 2 theta ~ 15. TiO2#rutile (2 1 0), Co-HCP (0 0 2), Co-FCC (1 1 1), almost the same 2 theta position. How should I do to separate them?

You can find the attached raw data. I will be appreciated it if you can try to do this. Then we can have a further discuss.

w.s.l. Boyer

Hi w.s.l.,

Yes, it shouldn't the same size. Here, I mentioned the same size: when I use the CIF file, I can get the Anatase size is ~29 nm. when I analysis the data without CIF file. I calculated from gamma, the anatase size is almost the same with using CIF. But the Cobalt metal size is very strange. The problem is how can I get the right Co metal size at the overlap position. Thank you in advance if you have other suggestions.

It took me a while to realize the wavelength was 0.53 angstroms. I had proceeded assuming it was something standard like Cu-Ka at 1.54 Angstroms. You might point that out in your introduction. Was that from a synchrotron or another source? I see it is similar to Cd-Ka with an energy of 23.2 keV.

Now back to your question, I don't think you should try to do too much with the peak at 15 degrees. You would be trying to fit it to peaks of uncertain widths at three different positions. I don't know that the fit would mean much. The uncertainty would be rather large. I would stick to using the clear peaks to estimate the crystallite size if you insist on doing so. Are they giving different size estimates at the different orientations?

You said that the Co size is very strange. You will need to elaborate on that. I have no idea what you mean.

You should be able to image those particles with a good FE-SEM. If the Co particles are much larger, you should be able to clearly see that and their size.

You may not be able to differentiate the particles well by EDS due to their small size unless you prepare them on a TEM grid.

You should be able to differentiate between the bright Co particles and the darker TiO2 particles with a backscattered electron signal.

Warren Straszheim

Hi Warren,

Thank you very much for all of your detailed reply and sorry for the late reply.

Yes, the tests were done at synchrotron beamline, the energy is 23.1 KeV. And the wavelength is 0.536728 A.

Because I guess the stacking faults in this position (Co-FCC and Co-HCP). I want to calculate the percentage. ----Now, I have a solution for this. I considered all the other peaks as background peak and only refined this peak with Rutile, Co-FCC and Co-HCP. It seems that I can get the reasonable results: rutile (2 1 0) - 48 nm, Co-HCP (0 0 2) - 9 nm and Co-FCC (1 1 1) -3 nm.

Perhaps that is reasonable, but I would be cautious about placing much confidence in the results. Are those size result justifiable? How significant are the results to your research hypothesis?

Others have already point out that peak broadening can be due to multiple effects. Can you really claim that there is a three-fold difference in crystallite size? I just recommend you be careful in your claims.