Capillary Blank Subtraction & QPA in Synchrotron qXRD (Including Amorphous)

Jandrix1369

Hi all,

I'm working on fully quantitative phase analysis (including amorphous content) using synchrotron XRD data collected at the Australian Synchrotron (PD beamline), and I’m looking for some guidance on best practices for TOPAS (using the JEdit backend). My questions are as follows:

Blank subtraction: I have blank spectra (empty capillary scans) collected separately. What is the best way to subtract this background from the sample scans in TOPAS/JEdit? Should this be handled as a background file (which I'm unsure how to do in JEdit) or mathematically subtracted outside of TOPAS (e.g. excel) and re-imported?
Capillary-specific corrections: Aside from zero error correction (which I have from the LaB6 calibration), and applying empirically refined instrument parameters, are there other capillary-related corrections that should be considered in TOPAS for accurate quantification (especially for amorphous determination)? For example:
Absorption?
Empirical or physical modeling of capillary scattering effects?
Any extra considerations when using internal standards (I've used corundum)?

Any advice or examples would be greatly appreciated — especially from those who’ve worked with synchrotron capillary data in TOPAS. I've done quite a bit of lab qXRD Bragg-Brentano setup, so the synchrotron + capillary are new to me.

Best,
JX

Jandrix1369

I realised that the blank capillary has a higher intensity than the experimental samples... so the result is negative intensities at 5-10 2theta... doesn't seem to make sense...

Would it make sense to scale the capillary value to something that doesn’t give a negative value for the peaks? That might be too much data manipulation though, surely?

rowlesmr

First things first: Pet peave. It's not "spectra" or "spectrum", unless you're doing energy-dispersive diffraction. It's a "pattern".

Right, on to the serious stuff.

Did you use the same capillary ( the same part of the capillary) for both the blank and sample? If not, there will be differences between capillaries. If you want to use the experimental data as a background fit, investigate techref for "Learnt Shapes for Background" and the user_y keyword.

If you want good peakshapes for capillaries, investigate the capillary_diameter_mm and related keywords.

Absorption, see above

capillary scattering effects, see above

Internal stds: be consistent in how you approach modelling the standard. Hopefully your specimens were well-mixed. For quanting the results, have a look at doi:10.1017/S0885715624000460

Jandrix1369

Hi, rowlesmr,

Thanks for the reply. We measured the capillary about a month after we got the experimental measurements. This is because we didn't realise we had to measure the blank for the amorphous QPA; a learning curve on our end. The synchrotron staff mentioned he had the same type/model of our capillary and took the blank of that, so the exact blank measured was not the one we had... but supposedly it has the same dimensions as it's the same model.

Has this kaput'd the experiment?

johnsoevans

Just to add to Matthew's comment. Even with an identical capillary the background correction can sometimes be too high due to absorption by the sample. It will obviously depend on the wavelength and sample you have.
In general, the more you can check your method using known reference materials the better.

rowlesmr

Not kaputed at all. Allow the capillary to be scaled, and ad a 3rd or 4th order chebyshev to account for any other wiggles, and should be good.

As John said, the capillary contribution is going to change due to absorption from the sample, so...

aacorrao

Hi JX,

Do you have an approximation of absorption / transmission % for your samples? Even with relatively low absorption (80-90% transmission) I've found a background can often be scaled well to data at low angles, but as angle increases the background is not representative. As John and Matthew said, this will necessarily depend on your sample and the wavelength.

An approach I've found useful is fitting the background data with a peaks phase (assuming a structural model is not available), then fixing the peak positions. Depending how absorbing your samples are, you may be able to fix the individual intensities and breadths then refine a common scale factor for the intensities. If the intensities are not modeled well at higher angles, you can refine the individual peak intensities while noting the relative intensity changes and fit quality, or even treat the background pattern in multiple sections (e.g., low, mid, high angles) with separate scale factors. Those last 2 comments come with a warning that you can easily start fitting sample contributions as background, and this will be especially problematic if you have broad peaks and / or amorphous content.

I'm unfamiliar with the beamline you collected on, but am assuming transmission geometry and an area detector for the comments below:

If you plan to subtract backgrounds rather than modeling them, I recommend collecting with longer exposure times on your background and averaging many background patterns together. You'll limit the noise propagated to your background-subtracted data.

If the detector was not a photon counting detector, you may see some differences between the perceived background in your sample data and background in your empty capillary data due to differences in the dark current or accumulated charge on the detector. If you did collect data using a photon counting detector and are seeing large differences in the baseline, you may also want to rule out fluorescence depending on your incident beam energy and the elements in your sample.

Hope that's helpful - best of luck with your data!