Re: Introduction of non-structural peaks in Fullprof

[Andreas Leineweber]

Dear All,

I also had a similar issue in my PhD thesis when using the that-time 
GSAS: wanting to determing d-spacings of unindexed peaks in 
half-interpreted diffraction pattern (it was magnetic superstructure 
peaks for which I did not have a model).


I may add a small trick to the technique described by Thierry Roisnel, 
which is helpful, if you do not access to the individual peaks of the 
phase (step 4).  You may have a tetragonal cell, e.g. a = b = 0.1 Å and 
c = d, with d to be refined and using an appropriate profile function. 
If you then put an atom with occupancy 1 at 0 0 0 and one of the same 
kind with occupancy -1 at 0 0 0.5, then you get a finite structure 
factor for all 00l with odd l and a zero structure factor for all even 
l. By this method you have fewer higher order reflections of that 
artificial phase, which might interfere with the rest of the pattern. 
The first higher order peak is 003, if I get it correct at the moment. I 
am not sure whether this worked really like this (it was GSAS at the end 
of the 1990s), because some programs do not allow negative occupancy. It 
is also possible that I did the thing by considering antiparallel 
magnetic moments at the mentioned fractional coordinates.


I, however, definitely enjoy using software where you can combine 
Rietveld refinement with individually fitted peaks without relying to 
such tricks.


Best regards

Andreas Leineweber



Am 25.04.2022 um 17:36 schrieb Magnus Sørby:


Dear Thierry and Alan,

@Thierry: Thank you very much! This is a great solution.

@Alan: I agree that we should strive to use as few and as meaningful 
refinable parameters as reasonably possible, not just in Rietveld 
refinements but in all kinds of modelling. The task I’m engaged in now 
is to estimate crystallite sizes in Si particles that are produced 
along with a lot of non-crystalline stuff in a semi-industrial 
process.  Taking more time to produce “clean samples” is therefore not 
an option (although it would surely be the best advice for a pure 
structure determination/refinement study). Since there are a lot of 
sample and time is an issue, I don’t have the luxury to spend a lot of 
efforts on the other features in each dataset; at least not for the 
moment. I’m therefore more than happy to take some shortcuts in the 
description of the non-Bragg parts of the data.


Best regards,

Magnus

*From:*rietveld_l-requ...@ill.fr  *On 
Behalf Of *Alan W Hewat

*Sent:* 25 April 2022 15:50
*To:* rietveld_l@ill.fr
*Subject:* Re: Introduction of non-structural peaks in Fullprof

Dear Thierry and Magnus.

Introducing an extra pseudo-phase, and then editing out all the peaks 
you don't want, is ingenious, but is it really less work than creating 
a background file as Rietveld advocated?


The whole idea of Rietveld was to refine only physically meaningful 
parameters. The more parameters you introduce, the less confidence you 
can have in any of them being physically meaningful, especially if you 
then have to constrain some of them.


Alan


Dr Alan Hewat, NeutronOptics
Grenoble, FRANCE (from phone)
alan.he...@neutronoptics.com
+33.476984168 VAT:FR79499450856
http://NeutronOptics.com/hewat <http://NeutronOptics.com/hewat>
___

On Mon, 25 Apr 2022, 14:53 Thierry Roisnel, 
 wrote:


    Dear Magnus,

    Here is a way to fit independent peaks in FullProf :

1. edit .pcr file and add for every independent peak an artificial
phase to treat in Profile Matching refinement mode (Jbt=2 and
Irf=0). You can use P m m m space group for example and adjust the
a parameter to fit with the diffraction peak 2theta position, b
and c parameters can have arbitrary values, smaller than a. So,
the special reflection will have 1 0 0 indices.

2. add refinement of one special reflection (Nsp_Ref = 1)

3. Run FullProf and update .pcr file, to provide Irf = 2

4. edit the created xxn.hkl file ( n for phase #n) and remove all
reflections excepted the first one (1 0 0)

5. edit .pcr file and refine peak position and profile parameters
for the special reflection you want to fit.

    The .pcr file will look as follows :

Phase #2 : fit  a single peak
!
!Nat Dis Ang Pr1 Pr2 Pr3 *Jbt Irf* Isy Str Furth   ATZ    Nvk
Npr More
   0   0   0 0.0 0.0 1.0 *2   2*   0   0 0  32516.641   0   7   1
!
!Jvi Jdi Hel Sol Mom Ter  Brind   RMua RMub    RMuc   Jtyp
*Nsp_Ref* Ph_Shift N_Domains
   0   0   0   0   0   0  1.  0. 0.  0.    0 
1  0  0
!
!
*P m m m* <--Space group symbol
!---> Profile Parameters for Pattern #   1  > Phase #   2
!  Scale  Shape1  Bov Str1  Str2  Str3  
Strain-Model
 0.1252800E-04   0.0   0.0   0.0 0.0   0.0   0
   0.0 0.000 0.000 0.000 0.000 0.000
!   

Re: Introduction of non-structural peaks in Fullprof

[Alan W Hewat]

OK Magnus, I see your point. But making a background file is not very work
intensive, even for lots of samples. It's just clicking on points on a
computer plot, and if its diffuse/disorder scattering the background will
only vary slowly. If you only have a couple of extra peaks, you can also
exclude their range from your refinement.

Good luck with your analysis. Alan

Dr Alan Hewat, NeutronOptics
Grenoble, FRANCE (from phone)
alan.he...@neutronoptics.com
+33.476984168 VAT:FR79499450856
http://NeutronOptics.com/hewat

++
Please do NOT attach files to the whole list 
Send commands to  eg: HELP as the subject with no body text
The Rietveld_L list archive is on http://www.mail-archive.com/rietveld_l@ill.fr/
++

RE: Introduction of non-structural peaks in Fullprof

[Magnus Sørby]

Dear Thierry and Alan,

@Thierry: Thank you very much! This is a great solution.

@Alan: I agree that we should strive to use as few and as meaningful refinable 
parameters as reasonably possible, not just in Rietveld refinements but in all 
kinds of modelling. The task I’m engaged in now is to estimate crystallite 
sizes in Si particles that are produced along with a lot of non-crystalline 
stuff in a semi-industrial process.  Taking more time to produce “clean 
samples” is therefore not an option (although it would surely be the best 
advice for a pure structure determination/refinement study). Since there are a 
lot of sample and time is an issue, I don’t have the luxury to spend a lot of 
efforts on the other features in each dataset; at least not for the moment. I’m 
therefore more than happy to take some shortcuts in the description of the 
non-Bragg parts of the data.

Best regards,
Magnus



From: rietveld_l-requ...@ill.fr  On Behalf Of Alan W 
Hewat
Sent: 25 April 2022 15:50
To: rietveld_l@ill.fr
Subject: Re: Introduction of non-structural peaks in Fullprof

Dear Thierry and Magnus.

Introducing an extra pseudo-phase, and then editing out all the peaks you don't 
want, is ingenious, but is it really less work than creating a background file 
as Rietveld advocated?

The whole idea of Rietveld was to refine only physically meaningful parameters. 
The more parameters you introduce, the less confidence you can have in any of 
them being physically meaningful, especially if you then have to constrain some 
of them.

Alan

Dr Alan Hewat, NeutronOptics
Grenoble, FRANCE (from phone)
alan.he...@neutronoptics.com<mailto:alan.he...@neutronoptics.com>
+33.476984168 VAT:FR79499450856
http://NeutronOptics.com/hewat
___


On Mon, 25 Apr 2022, 14:53 Thierry Roisnel, 
mailto:thierry.rois...@univ-rennes1.fr>> wrote:

Dear Magnus,

Here is a way to fit independent peaks in FullProf :

1. edit .pcr file and add for every independent peak an artificial phase to 
treat in Profile Matching refinement mode (Jbt=2 and Irf=0). You can use P m m 
m space group for example and adjust the a parameter to fit with the 
diffraction peak 2theta position, b and c parameters can have arbitrary values, 
smaller than a. So, the special reflection will have 1 0 0 indices.

2. add refinement of one special reflection (Nsp_Ref = 1)

3. Run FullProf and update .pcr file, to provide  Irf = 2

4. edit the created xxn.hkl file ( n for phase #n) and remove all reflections 
excepted the first one (1 0 0)

5. edit .pcr file and refine peak position and profile parameters for the 
special reflection you want to fit.

The .pcr file will look as follows :

Phase #2 : fit  a single peak
!
!Nat Dis Ang Pr1 Pr2 Pr3 Jbt Irf Isy Str Furth   ATZNvk Npr More
   0   0   0 0.0 0.0 1.0   2   2   0   0   0  32516.641   0   7   1
!
!Jvi Jdi Hel Sol Mom Ter  Brind   RMuaRMubRMuc   Jtyp  Nsp_Ref Ph_Shift 
N_Domains
   0   0   0   0   0   0  1.  0.  0.  0.0  1  0 
 0
!
!
P m m m  <--Space group symbol
!---> Profile Parameters for Pattern #   1  > Phase #   2
!  Scale  Shape1  Bov  Str1  Str2  Str3   Strain-Model
 0.1252800E-04   0.0   0.0   0.0   0.0   0.0   0
   0.0 0.000 0.000 0.000 0.000 0.000
!   U V  W   X  YGauSiz   LorSiz 
Size-Model
   0.005115  -0.007545   0.020850   0.012780   0.020016   0.00   0.00   
 0
  0.000  0.000  0.000  0.000  0.000  0.000  0.000
! a  b calpha  beta   gamma  #Cell Info
   3.40   0.50   0.50  90.00  90.00  90.00
0.00.00.00.00.00.0
!  Pref1Pref2  Asy1 Asy2 Asy3 Asy4  S_L  D_L
  0.0  0.0  0.0  0.0  0.0  0.0  0.01849  0.02007
 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
! Special reflections:
!  h   k   l  nvk   D-HG^2  Cod_D-HG^2  D-HL  Cod_D-HL  Shift   
 Cod_ShiftDel_U/sig1  Code Del_Y/gam2 Code
   1   0   00   0.   0.000 0.0   0.000 0.0  
 0.000 0.0   0.000 0.0   0.000

I hope you will success in what you want to do.

Regards

Thierry Roisnel


Le 25/04/2022 à 11:18, Magnus Sørby a écrit :
Hi,

I’m playing with Fullprof again for the first time in many years since I’ve 
changed employer and lost my access to Topas.

Is it possible to introduce independent peaks (i.e. peaks whose positions are 
not determined by a unit cell) with refinable positions, intensities and shapes 
in Fullprof? I’ve found this possibility very useful in Topas for dealing with 
features in the datasets that are not of interest and not

Re: Introduction of non-structural peaks in Fullprof

[Thierry Roisnel]

    Dear Magnus,

    Here is a way to fit independent peaks in FullProf :

1. edit .pcr file and add for every independent peak an artificial phase 
to treat in Profile Matching refinement mode (Jbt=2 and Irf=0). You can 
use P m m m space group for example and adjust the a parameter to fit 
with the diffraction peak 2theta position, b and c parameters can have 
arbitrary values, smaller than a. So, the special reflection will have 1 
0 0 indices.


2. add refinement of one special reflection (Nsp_Ref = 1)

3. Run FullProf and update .pcr file, to provide Irf = 2

4. edit the created xxn.hkl file ( n for phase #n) and remove all 
reflections excepted the first one (1 0 0)


5. edit .pcr file and refine peak position and profile parameters for 
the special reflection you want to fit.


    The .pcr file will look as follows :

Phase #2 : fit  a single peak
!
!Nat Dis Ang Pr1 Pr2 Pr3 *Jbt Irf* Isy Str Furth   ATZ    Nvk Npr More
   0   0   0 0.0 0.0 1.0 *2 2*   0   0   0  32516.641   0   7   1
!
!Jvi Jdi Hel Sol Mom Ter  Brind   RMua    RMub    RMuc Jtyp *Nsp_Ref* 
Ph_Shift N_Domains
   0   0   0   0   0   0  1.  0.  0. 0.    0  
1  0  0

!
!
*P m m m* <--Space group symbol
!---> Profile Parameters for Pattern #   1 > Phase #   2
!  Scale  Shape1  Bov  Str1 Str2  Str3   Strain-Model
 0.1252800E-04   0.0   0.0   0.0   0.0 0.0   0
   0.0 0.000 0.000 0.000 0.000 0.000
!   U V  W   X Y    GauSiz   LorSiz 
Size-Model
   0.005115  -0.007545   0.020850   0.012780   0.020016 0.00   
0.00    0

  0.000  0.000  0.000  0.000 0.000  0.000  0.000
! a  b c    alpha  beta gamma  #Cell Info
*3.40   0.50 0.50  90.00  90.00  90.00**
**    0.0    0.0    0.0 0.0    0.0    0.0*
!  Pref1    Pref2  Asy1 Asy2 Asy3 Asy4  S_L  D_L
  0.0  0.0  0.0  0.0  0.0  0.0 0.01849  0.02007
 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
*! Special reflections:**
**!  h   k   l  nvk   D-HG^2 Cod_D-HG^2  D-HL  Cod_D-HL  
Shift Cod_Shift    Del_U/sig1  Code Del_Y/gam2 Code**
**   1   0   0    0   0.   0.000 0.0   0.000 
0.0   0.000 0.0   0.000 0.0   0.000**

*

    I hope you will success in what you want to do.

    Regards

    Thierry Roisnel


Le 25/04/2022 à 11:18, Magnus Sørby a écrit :


Hi,

I’m playing with Fullprof again for the first time in many years since 
I’ve changed employer and lost my access to Topas.


Is it possible to introduce independent peaks (i.e. peaks whose 
positions are not determined by a unit cell) with refinable positions, 
intensities and shapes in Fullprof? I’ve found this possibility very 
useful in Topas for dealing with features in the datasets that are not 
of interest and not possible to possible to describe with a 
crystalline phase e.g. diffuse scattering from an amorphous component.


I’m aware of the option to include a self-defined background (.bac 
file) with refinable scale, but this is time-consuming workaround.


Best regards,

Magnus



Magnus H. Sørby, PhD

Senior Scientist  - Materials Science and Analytical Chemistry

Cenate AS

Norway


++
Please do NOT attach files to the whole list
Send commands to  eg: HELP as the subject with no body text
The Rietveld_L list archive is onhttp://www.mail-archive.com/rietveld_l@ill.fr/
++


--

* Thierry Roisnel*
Centre de Diffractométrie X (CDIFX)
Institut des Sciences Chimiques de Rennes
UMR6226 CNRS - Université de Rennes 1
Bât. 10B, p. 153
Campus de Beaulieu
Avenue du Général Leclerc
35042 Rennes cedex, France

Tél: 02 23 23 59 02
Email : thierry.rois...@univ-rennes1.fr 
 

Centre de Diffractométrie X, ISCR UMR6226 




ISCR UMR6226 





Membre de ЯÉCIPROCS 	Réseau ЯÉCIPROCS 

++
Please do NOT attach files to the whole list 
Send commands to  eg: HELP as the subject with no body text
The Rietveld_L list archive is on http://www.mail-archive.com/rietveld_l@ill.fr/
++

Re: Introduction of non-structural peaks in Fullprof

[Alan W Hewat]

IMHO, it's not a good idea to introduce independent peaks with refinable
positions, intensities, shapes and other parameters, simply to reduce your
R-factor or save time. You risk reducing the confidence you can have in
your structural parameters.

That is why Rietveld defined the background rather than refined it. Take
time to produce clean samples, more time to understand unexpected features
that are "not of interest", and even more time to measure and define the
background with a minimum of extra parameters. Computer programs should not
be sausage machines.

Alan

Dr Alan Hewat, NeutronOptics
Grenoble, FRANCE (from phone)
alan.he...@neutronoptics.com
+33.476984168 VAT:FR79499450856
http://NeutronOptics.com/hewat
___


On Mon, 25 Apr 2022, 11:20 Magnus Sørby,  wrote:

> Hi,
>
>
>
> I’m playing with Fullprof again for the first time in many years since
> I’ve changed employer and lost my access to Topas.
>
>
>
> Is it possible to introduce independent peaks (i.e. peaks whose positions
> are not determined by a unit cell) with refinable positions, intensities
> and shapes in Fullprof? I’ve found this possibility very useful in Topas
> for dealing with features in the datasets that are not of interest and not
> possible to possible to describe with a crystalline phase e.g. diffuse
> scattering from an amorphous component.
>
>
>
> I’m aware of the option to include a self-defined background (.bac file)
> with refinable scale, but this is time-consuming workaround.
>
>
>
> Best regards,
>
> Magnus
>
>
>
> 
>
> Magnus H. Sørby, PhD
>
> Senior Scientist  - Materials Science and Analytical Chemistry
>
> Cenate AS
>
> Norway
>
>
> ++
> Please do NOT attach files to the whole list  >
> Send commands to  eg: HELP as the subject with no body
> text
> The Rietveld_L list archive is on
> http://www.mail-archive.com/rietveld_l@ill.fr/
> ++
>
>
++
Please do NOT attach files to the whole list 
Send commands to  eg: HELP as the subject with no body text
The Rietveld_L list archive is on http://www.mail-archive.com/rietveld_l@ill.fr/
++