Hello Hai Pham

Your refinement is going pretty well. Have you tried atomic thermal displacement (isotropic or anisotropic), and atoms site occupancy?

Best regards,

Ricardo Tadeu

Hello Ricardo Tadeu Maia

I have tried the site occupancy but the value are still high.

As i tried the anisotropic, the value turned a bit better ( Bragg factor: 13.7 Chi square: 19.7).

However some value for atom anisotropic are negative. Usually the b-factor should not be negative so i dont know if this fitting is okay or not.

Hello Hai Pham. I suggest that before performing a refinement with atomic contributions, perform a less demanding refinement to check if the Chi^2 problem is due to the high number of counts (very high 7.000 in "y" for a small interval and step 10 to 60 Theta/ 0.01). You can check this step by selecting the "Profile Matching with constant scale factor" option in calculation (Fig.1) and keeping the scalar parameter disabled (Fig.2). I usually prefer the number of counts a little above 3.000 counts and an interval between 10 to 110 Theta with a step of 0.02 for structures with relative crystallinity.

If the difficulty persists, you can try to change the mathematical model of the profile using the program's default "Thompson-Cox-Hastings pseudo-Voigt * Axial divergence asymmetry" or refine the asymmetric coefficients.

Hello, few suggestions :

- I would recommend to use a manual background (Winplotr/ open a File / Point selection / Select Background Point, click then save) or at least use a polynomial 12th. Your background seems quite high (for example in the 35-45°) and could mask some datas. You can then refine it.

- With NPR=5, the X values (in UVWX) should be set to 0 (in most case).

- If possible UVW should be refined beforehand from a reference measurement (often LaB6), so you dont have to refine these along your refinement. It is not alway needed, but easier (but your values might be fine).

- I would refine most of the parameters (cell parameters, Zeroshift, UVW, background, ...) in Pattern Matching mode (JBT=2), so you dont need to refine them much when going into Rietveld (JBT =0). At some point if your starting structure is wrong, refining these parameters might mask some datas. If your starting structure is quite close, refining "scale" should manage most of the difference (remaining being atomic positions and Biso). Dont affine scale if you are refining structure at the same time.

.

For Structure : I dont know about Ca2Mn3O8, but I would start back from "perfect" structure values, your actual values seems quite far, these tends to diverge easily, if you dont have a good fit already beforehand.

- You can put R_at / R_an / R_pr (top of your pcr) to lower values (similarly to EPS), to have smaller steps when refining atomic positions, anisotropies and so on. Check the Fullprof manual to know which parameters are affected.

- Occupancies seems quite far from initial one, i would revert them back, and refine them only in case of evidence of it, when most of the refinement is already done. Similarly XYZ atomic positions should not move much from the original one.

- I dont know about Ca2Mn3O8, but some of your Biso are set to 0, which is unlikely. Put them back to 0.5 as starting point.

Also you could check for preferential orientation (depending on how you prepared your sample). Click on the green line to see if you always have missing intensities on the same kind of hkl. If yes you might want to refine Pref1/Pref2

At some point if you want a "perfect" fit you migh need to add Ni in your structure, with linear restraint (so Mn occupancy decrease when Ni one increase), but i am not sure you can refine these from laboratory XRD.

Remember you will not be able to always have a perfect fit depending on measurement conditions and so on, especially from Laboratory XRD. Advanced Rietveld are mostly from Synchrotron, Neutron or really long XRD scan. Maybe you dont need to invest too much time in it, depending on what you are looking for!

Hope it helps,

Bastien Leclercq Hi, I need your help with XRD refinement. my PCR file is not running when I try to fix some parameters, the file gets corrupt and not working.

[email protected] email me, I'm unable to direct message you