mixed site occupancies

cbirkel

I am trying to do a Rietveld refinement of a compound crystallizing in the half-Heusler crystal structure.

STR(F_-4_3_M, "TiNiSn")
Cubic(@ 5.93412`)

If I put the site occupancies in like this

site Ni x 0 y 0 z 0 occ Ni 1 beq 1
site Sn x =1/2; y =1/2; z =1/2; occ Sn 1 beq 1
site Ti x =1/4; y =1/4; z =1/4; occ Ti 1 beq 1

the intensities do not refine correctly (some intensities are too high, some are too low).

I think there could be mixed occupancies (so some Ni might be on the Sn or Ti site etc). Unfortunately, I have not been able to add constraints which would be that

a) Ni, Sn, Ti should add up to 1 each (e.g. Ni on Ni-site + Ni on Sn-site + Ni on Ti-site = 1), and
b) partial occupancies per site should add up to 1 (e.g. Ni on Ni-site + Sn on Ni-site + Ti on Ni-site = 1).

Can anybody help with that? Thank you!

cc0u50fd

Could you do that by defining an equation for the limit on site occupancy:

site Ni x 0 y 0 z 0 occ Ni 1 min 0 max GNi beq 1
site Sn x =1/2; y =1/2; z =1/2; occ Sn 1 min 0 max GSn beq 1
site Ti x =1/4; y =1/4; z =1/4; occ Ti 1 min 0 max GTi beq 1

prm !gtot = GNi+Gsn+Gti;
prm Gni 1 max 1
prm Gsn 1 max 1
prm GTi 1 max 1

something like this trying to predetermine the max occupancy of each site?

But you would probably have to use penalties to do it. (I also don't know if this method is even possible or practical)

Another way of doing it would be if you know the composition of your material by analytical methods and then restrained (using a really high penalty) the occupancies to fit those values (using equations). You could then allow the refinement to refined the occupancies without directly messing with the amounts in each.

I've used the latter method to keep occupancies sane where phases have overlapped peaks and multiple sites containing the same atom.

mfisch

What type of instrument and radiation are you using?

rowlesmr

The following assumes that your data supports the analysis.

Have you tried just letting the occupancies refine freely to what ever value they want (negative, greater than 1...)? This will give you some indication as to whether or not it could be an occupancy issue. I would fix the beqs of all sites to be the same.

Have you tried letting the beq's refine freely? I would fix the occupancies if you did this.

With regards constraints for refining partial occupancies for three atoms across three sites, it looks a little complex. Using the notation "site:atom", you've got six equations to solve:

1:1 + 1:2 + 1:3 = 1
2:1 + 2:2 + 2:3 = 1
3:1 + 3:2 + 3:3 = 1

1:1 + 2:1 + 3:1 = 1
1:2 + 2:2 + 3:2 = 1
1:3 + 2:3 + 3:3 = 1

You could probably reduce/correlate a few of those, but I don't have time atm to do that.

Matthew

cbirkel

These are all synchrotron XRD data obtained at 11-BM (Argonne).

pamwhitfield

You're in serious danger of over-parameterization this one with a single dataset, even one as good as 11-BM.
It's analogous to simultaneous equations - you need enough information (i.e. known equations) to solve the number of variables. In this case your known equation data-wise is a single 11-BM dataset.

If you assume it's only Ni wandering with a fixed composition it's doable but free refinement of all atoms on all sites is a non-starter. The fixed composition will help with Ti and Ni being such similar scatterers.

te37v

What type of instrument and radiation are you using?

namebraw

Have you tried just letting the occupancies refine freely to what ever value they want

arazbat

Dear All,

I have a similar problem. I am doing MgB2 with Ti and V and want to know how occupancies are changed.
I create another site in site section for Ti and then I saying that Occ. Ti+Mg=1 like that

Ti Occ. X and Mg Occ. =1-X

while I am refining I am getting always zero for Ti occupancy but 1 for Mg! but here for MgB2, lattice parameter a, cell volume and C/a ratio are changing which all are showing there are some doping in MgB2

I wonder am I doing any mistake here?

Thanks for your comments

johnsoevans

It's impossible to comment without seeing details of what you have in the input file. e.g. does Ti refine to exactly zero or a number close to zero? Do you have min/max limits set anywhere?

arazbat

sorry for late answer

@johnsoevans,thanks for you answer, no just zero and I did not sent any min/max or any limit. just I used that formula which I mentioned before.

I already find the way. I am not setting any rule for refinement. Just refine freely Mg site and Ti. Just as they are replacing each other I am writing value near my expectation. for instance, 0.95 for Mg and 0.05 for Ti as I have doped it with 5% Ti.

I hope it is the true way of proceeding

Thanks