izarc
I am trying to refine Al occupancies in mordenite structure, but I wish to fix a limit because I have EDX and MP-AES evidence to know that Si/Al ratio is around 6-7:
prm T1 0.1 min 0 max 0.394
prm T2 0.1 min 0 max 0.394
prm T3 0.1 min 0 max 0.5
prm T4 0.1 min 0 max 0.5
prm ttal = (T1*16)+(T2*16)+(T3*8)+(T4*8); : 6 max 8
site SI1 num_posns 16 x 0.19897 y 0.42771 z 0.5418 occ Si =1-T1; beq =BSi;
site Al1 num_posns 16 x 0.19897 y 0.42771 z 0.5418 occ Al =T1; beq =BSi;
site SI2 num_posns 16 x 0.19656 y 0.19092 z 0.5454 occ Si =1-T2; beq =BSi;
site Al2 num_posns 16 x 0.19656 y 0.19092 z 0.5454 occ Al =T2; beq =BSi;
site SI3 num_posns 8 x 0.0871 y 0.384 z 0.25 occ Si =1-T3; beq =BSi;
site AL3 num_posns 8 x 0.0871 y 0.384 z 0.25 occ Al =T3; beq =BSi;
site SI4 num_posns 8 x 0.0866 y 0.228 z 0.25 occ Si =1-T4; beq =BSi;
site AL4 num_posns 8 x 0.0866 y 0.228 z 0.25 occ Al =T4; beq =BSi;
This was my attempt, however when I refine the ttal value goes above 8. Do you know the proper way to do this?
Thank you very much for your help!
johnsoevans
You can't put min/max limits on an equation like this.
You could consider introducing some kind of penalty or restraint to fix the Si/Al ratio. Something like:
penalty = (6.5-ttal)^2;
It is possible to apply a penalty only if the value is <6 or >7, though I don't know if that would really make sense. One way of doing that is given in the Angle_Distance_Restrain macro in topas.inc.
There are some other ways of restraining the chemistry of your samples (including across multiple phases) in the various examples in the "Quantitative" folder. e.g. zro2-restraint-xrf-zr.inp.
izarc
Dear John,
Thank you for your quick response! I am afraid I am still too beginner with all this.
I tried using:
penalty = (6.5-ttal)^2;
I can somewhat force the value of ttal to where I want but it is suspicious that all of a sudden Al1 & Al2 became zero occupied, it feels like it is just mathematically convenient since they are multiplied by 16 in my equation.
I went to open topas.inc file and found:
macro Angle_Distance_Restrain(c, t, t_calc, tol, wscale)
{
prm = c; : t_calc
penalty =
(wscale)
IF c < ((t)-(tol)) THEN
(c - ((t) - (tol)) )^2
ELSE
IF c > ((t)+(tol)) THEN
(c - ((t) + (tol)))^2
ELSE
0
ENDIF
ENDIF
;
}
To be completely honest with you, I am considerably lost in here.
I also found the file named 'zro2-restraint-xrf-zr.inp.':
...
element_weight_percent Zr+4 zr 0 ' refined value is 66.18 without restraint
restraint = zr - 60; : 0
save_best_chi2
'---------------------------
elemental_composition
'---------------------------
...
but I was not able to make the analogy to my case:
element_weight_percent Al alu 0
restraint = alu - 35; : 0
save_best_chi2
'---------------------------
elemental_composition
'---------------------------
or
element_weight_percent Al 0
restraint = Al - 35; : 0
save_best_chi2
'---------------------------
elemental_composition
'---------------------------
or
element_weight_percent Al+3 alu 0
restraint = alu - 35; : 0
save_best_chi2
'---------------------------
elemental_composition
'---------------------------
Gave me the following error.
*** Error loading sstring_in
at LINE 70
See log file C:\TOPAS6\topas.log
Cannot locate a from element in data structures
Please forgive my ignorance, I hope you can shed some light here.
johnsoevans
Are you sure the data you have contains the information you're after?
If you have X-ray data then scattering from Al and Si will be very similar. Extracting 4 site occupancies (especially given their correlation with other parameters) is extremely challenging - there's a lot of specialist literature on this issue in zeolites.
You also have to be concerned if parameters are hitting a hard limit at the end of the refinement.
I suggest you look at the uncertainties on your refined parameters and the correlations between them to get some feel for this.
Another thing that's useful is to just simulate a series of patterns as you change these site occupancies and see how big the changes you'd expect to see are. This will give you a feel for the information content in your data.
izarc
Spot on. After contrasting some opinions I came to the conclusion that expecting to get this type of information would be pushing the analysis too far for a powder diffraction experiment. Thank you very much for your time & useful suggestions!