Issue XPDF modelling of LaB6

[Nick Gys]

Hi all

From an XPDF measurement campaign at ESRF (ID22), I am performing PDF modelling on a LaB6 standard to determine the instrumental Qdamp and Qbroad parameters. From the experimental XPDF of LaB6 (Qmax=21, see image), the PDF seems to dampen at around 150 angstrom. When performing the modelling in PDFgui in the 1-50 angstrom range, I performed a stepwise optimisation via scale factor, lattice parameter, Qdamp, delta2, etc but I fail to obtain a good fit (see image). 

The following refined parameters are obtained: 

Scale factor=0.279

Lattice parameter=4.169

Qdamp=0.021

ADP La=0.006

ADP B=0.017

Delta2=2.87

Would there somebody be able to provide feedback and advice on this? 

Many thanks in advance!

Greets

Nick

[Simon Billinge]

That's interesting. I am not sure why it is not working but it looks like maybe the adps may be stuck. Did they move from their initial values? Maybe try initializing things by during over a narrower range and changing initial values and make sure it is not stuck.  Feel free to share data and input file and someone else may be able to have a go.

Simon Billinge

Distinguished Professor of Materials Science

Director, CNSI (California NanoSystems Institute)

University of California, Santa Barbara

Tel: 805-893-8716 + Email: sbil...@ucsb.edu

--
You received this message because you are subscribed to the Google Groups "diffpy-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to diffpy-users...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/diffpy-users/07b63f9b-ef3c-4cba-8c1e-7c6c080e8cc2n%40googlegroups.com.

[Nick Gys]

Thanks for the feedback Simon. I started from ADPs of 0.005 for both La and B, and played with various initial values during the refinement but it keeps giving a bad fitting (Rw around 0.4-0.5). I am sharing the experimental PDF and the input cif-file, in case someone wants to give it a try to find the issue. 

Many thanks! 

Op vrijdag 19 juni 2026 om 11:35:01 UTC+2 schreef sbil...@ucsb.edu:

[Till Schertenleib]

Hi Nick,

I gave it a quick go and I'm getting a similarly bad fit like you. I suspect there might be something wrong with the data, rather than your fitting approach. 

I did the fit over 1-10, 1-20, and 1-40. The low-r fit is fine, but for larger r-windows the model struggles and it looks to me as if some peak positions at higher-r are not quite well reproduced by the model. It tries to fix that by inflating the lattice parameters.

Did you do the data integration (from image to 1D diffraction pattern) and calibration of the setup yourself?

It looks like the data was integrated and binned in Q. Are you sure that the calibration was done correctly and that the correct wavelength was used?

Before going insane about your refinement, I would take a step back and make sure that the data is "clean". 

Best,

Till

To view this discussion visit https://groups.google.com/d/msgid/diffpy-users/520092de-4a6c-45cb-8421-37115a2badb7n%40googlegroups.com.

--

Till Schertenleib

Phone: +45 91 66 94 86

[Simon Billinge]

Thanks Till

To view this discussion visit https://groups.google.com/d/msgid/diffpy-users/CA%2B6Bn%2BhXTyTZF%3DQcS4auRuNnvX75ST7ioaKfSxO_R94ZGKViCg%40mail.gmail.com.

--

[Erick]

Hi Nick,

 

I recently had a similar issue, although it involved a different beamline (P02.1 at DESY). I had to perform a parallax correction on the data before generating the PDF. I did this by getting the correction coefficients and applying a 2theta offset during Rietveld. If you can share the scan 2theta v intens file I could try to see if it's the case here.

Best,

Erick

[Simon Billinge]

Just in case it helps, the latest PDFgetx3 versions contain a 2theta 0 offset parameter. 

Also, everyone, you may want to check out diffpy.morph that was recently released and is about to be published in PRM (it is already on arxiv).  It can be helpful for applying ad hoc transformations to data.

S

Simon Billinge

Distinguished Professor of Materials Science

Director, CNSI (California NanoSystems Institute)

University of California, Santa Barbara

Tel: 805-893-8716 + Email: sbil...@ucsb.edu

To view this discussion visit https://groups.google.com/d/msgid/diffpy-users/CALvOQJ_VAupbUM0QWmDs%3DRRb%3DXb1HqtpTXxHP5U%2Bi99AXThUbg%40mail.gmail.com.

[Nick Gys]

Thanks Till and Erick for checking on this, I strongly appreciate it! Somehow, a good sign that a similar bad fitting is obtained and the issue is likely linked to the initial data processing. 

The data processing was done with an in-house python script by the beamline scientists at ID22. 

@erick: interesting to read you faced a similar issue at DESY. Thanks for offering to check my data if a similar correction is needed. In attachment, you can find the twotheta in function of intensity files for both the experimental and simulated LaB6 data. 

Looking forward to your feedback! 

ps: it is a pleasure to see such a supportive community within the PDF world

Greets

Nick

Op zaterdag 20 juni 2026 om 14:42:52 UTC+2 schreef sbil...@ucsb.edu:

[tiankongbuyi]

Dear  Nick

I attempted to plot your LaB6_2theta data and found that the data is acceptable below 10 degrees; however, above 10 degrees, the peaks begin to flatten, which is quite unusual, I suggest you reprocess the TIFF data into 2-theta data. This is just my opinion.

Best Wishes

Xiangyan Hou

---- Replied Message ----

From	Nick Gys<nick....@gmail.com>
Date	6/22/2026 21:13
To	diffpy-users<diffpy...@googlegroups.com>
Subject	Re: [diffpy-users] Issue XPDF modelling of LaB6

To view this discussion visit https://groups.google.com/d/msgid/diffpy-users/3b608c33-6b18-4445-87b7-96e50cdeb0c3n%40googlegroups.com.