Analysis of MOFs Using PDFgui

[Kirstin Wilson]

Dear all,

I am a PhD student based at the Institut Laue Langevin working on mixed metal MOF materials. 

I have carried out a neutron experiment on the D4 instrument at the ILL on several different MOF structures and am now trying to analyse the data using pdfgui but am having issues in achieving a successful fit. 

I initially tried refining a large number of parameters using the symmetry constraints (Ce80La20_10K) but after following initial advice from Simon Billinge, I have significantly reduced this and set a single ADP (Ce80La20_10K_2). However, I am still not achieving a good fit and often when I try to refine the ADPs, the calculated fit levels out along 0 rather than following the peaks. I am using Qmax, Qdamp and Qbroad values that have been confirmed by the instrument responsible. 

I have attached  the .ddp3 file I have been using and would be grateful if anyone could have a look and advise me on what the issue may be and what is the best method to fit materials of this nature.

Thank you in advance for all your help,

Best wishes

Kirstin Wilson

[Till Schertenleib]

Hi Kirstin,

It looks to me like there might be something wrong with your G(r). I am not familiar with neutron scattering and data reduction of neutron data, but it's a bit strange that it starts below zero. 

Can you tell us what parameters you used to calculate G(r)? You might have to play around with Qmin to fix this... 

--
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/96f5453c-e577-4f59-848f-893e463d24f3n%40googlegroups.com.

--

Till Schertenleib

EPFL, Switzerland

Phone: +41 78 729 21 31

[Simon Billinge]

Thanks Till,

For the "below zero" issue, actually it is asymptoting back to zero, it just wasn't plotted all the way to zero on the r-axis.   This can be confusing and where possible I prefer to see G(r)'s plotted all the way to r=0, but that is why it is below zero.

With that said, I do agree with Till that it looks a bit odd.  I would normally expect to see a peak around 1.4 coming from the organics in the system, well before the peak that around 2, which presumably is coming from the metal ions in the nodes and whatever is coordinating them (oxygen?).  I would also expect that peak to poke higher onto the positive side of the axis, so the overall shape of the background might be a bit off.  Till's intuition is probably right there.

Here are some thoughts, without knowing too much about what is going on.  First, the G(r) below the peak at 2 is looking too well behaved.  This suggests to me that a low-band-pass filter may have been applied (sometimes done in the neutron world by back-fourier transforming the low-r-region and subtracting it from the data then re-forward-transorming).  I am also not a big fan of this.  It doesn't do any harm per se, but it artificially cleans a part of the G(r) that can give helpful information about data quality and if anything has gone wrong with the measurement....like cleaning  a crime scene before the forensic team has got there.....

One possibility is that there was a significant amount of hydrogen in the sample and the neutron data reduction struggled to remove it and this affected the overall shape?

Regarding the model, it almost looks to me like it has been inverted (picked up a negative scale factor?) because there is a peak at around 1.4 A in that model but it is upside down?  @kirsten, could I ask, is the red curve from a model that has been run through the regression engine, or it is calculated from the model that was uploaded from a CIF file without refining?  That would be my first step, something like:

1) load the cif

2) set all off-diagonal ADPs to zero

3) set all other ADPs to James Bond

4) calculate the PDF without refining anything and plot it on the data

5) manually adjust the scale factor so they are more or less on the same scale and replot.

This is a sanity check that allows you to see where the odel thinks peaks and intensity should be, compared to where your data think they shouljd be.

Feel free to share back the results if you like.

S

On Wed, Feb 5, 2025 at 7:10 AM Till Schertenleib <till.sch...@gmail.com> wrote:

Hi Kirstin, It looks to me like there might be something wrong with your G(r). I am not familiar with neutron scattering and data reduction of neutron data, but it's a bit strange that it starts below zero.   Can you tell us what parameters

ZjQcmQRYFpfptBannerStart

This Message Is From an External Sender

This message came from outside your organization.

 

ZjQcmQRYFpfptBannerEnd

To view this discussion visit https://groups.google.com/d/msgid/diffpy-users/CA%2B6Bn%2Bi804rWcRutmg%3D2%2BnYj8F6%2B%2B5gJgdWMEALa%2BsNmBAYdnQ%40mail.gmail.com.

--

Simon Billinge

Professor, 

Department of Applied Physics and Applied Mathematics

Columbia University

[Kirstin Wilson]

Thank you both.

Yes, the red curve was after refinement of the scale factor which resulted in a negative value. Following the steps you described resulted in the attached image where the peaks in the model and data do not align.

Does this mean I should revisit the G(r) as it has been back fourier transformed and your thought about the hydrogen could also be an issue?

Thanks,

Kirstin

[Simon Billinge]

that's great!  OK, we have the model behaving somehow reasonably now, but there are clearly some issues with the data.  First the low-r part seems to be problematic.  Some of the mess from the crime-scene wasn't cleaned up by the filter and is propagating through is my best bet.  This gets less as r increases, so the fit is becoming more reasonable, for example, in the region around, but noisy beats seem to be reappearing above that, which is likely again a remnant of the low-r issues reappearing.

As I mentioned, a wild guess is that you had hydrogen in your sample and this wasn't properly corrected in the data reduction.  The next steps I am afraid are for you to become expert in neutron data-reduction......

S

On Wed, Feb 5, 2025 at 8:21 AM Kirstin Wilson <kirstin....@gmail.com> wrote:

Thank you both. Yes, the red curve was after refinement of the scale factor which resulted in a negative value. Following the steps you described resulted in the attached image where the peaks in the model and data do not align. Does this mean

To view this discussion visit https://groups.google.com/d/msgid/diffpy-users/d729f8f9-3b9f-448e-aa42-eb8fcbb00f5fn%40googlegroups.com.

[Mikkel Juelsholt]

Hi all 

This is not to say anything bad about any instrument who are kind enough to measure data for us. 

In my experience, the in-house reduction and Fourier transform program at D4 is quite funky. 

I have had exactly the same problem as you, Kirstin. 

If you get a hold of the initial scattering pattern, I(Q), and then just run it through PDFgetN3, then you get a very nice PDF.

Cheers Mikkel