hydrogen influence on neutron diffraction PDF

[chenguang yang]

Dear all, 

I  just got some neutron diffraction data from NOMAD in ORNL. The sample is Si particle electrode in half-cell Li battery. The instrument scientist told me there is appreciable amount of hydrogen in the sample. Could you please tell me how to identify the hydrogen effect on the Pair distribution function? Is it possible to still extract useful information from the PDF data  when hydrogen is present? I attached a plot of the data fitted in PDFgui. 

Thank you!

Chenguang Yang 

Materials Science and Engineering Department

University of Utah

[Mikkel Juelsholt]

Hi Chenguang

As such the hydrogen does not have any direct effect on the PDF. The main issue with hydrogen is the large incoherent scattering cross section of hydrogen which lowers your signal-to-noise and gives you a massive background from incoherent diffuse scattering which does not carry any information.You have to correct for the background, because if you do not then your PDF will just be a a series of rapid oscillations and you cannot extract any information from it. Correction can either be done using NOMADs own data reduction program or by using PDFgetN3. You cannot do anything about the extra noise, except you might have to play around with your Qmax in the Fourier transform. 

If you can make the necessary correction and get a dataset with a good enough signal-to-noise ratio, then the PDF will show interatomic distances containing hydrogen. Hydrogen is a bit special that it has a negative coherent scattering length which gives rise to negative peaks in the PDF. Since the bond lengths between hydrogen and other atoms are very short, the presence of hydrogen is easily seen from a negative peak around 1 Å. 

So in short if you can correct for background, then there is absolutely no problem having hydrogen in you sample and you can still extract structural information.

If you are worried about the fit that you have attached, then it looks like you have an extra phase in your sample. Remember with neutrons then light atoms, such as C, is very visible. It could be an electrolyte or binder for example. Your PDF also seem somewhat noisy, maybe you need to adjust your Qmax.

Hope it helps. Mikkel

[Peter Metz]

As a follow-up to Mikkel's remarks, here's a paper on the subject. Figures 3-7 show the data effects in question. The nuclear incoherent hydrogen scattering forms an approximately gaussian contribution centered at Q=0 to the scattered intensity, and thus perturbs the low-r region of the reduced PDF.

I think the first correlation in your figure is the 2.35 AA Si-Si correlation? (It's a little hard to tell with the scale) If you're refining delta1 or delta2 (correlated motion parameters) PDFgui will return a null peak width if the peak width model becomes negative. You may just have to fix the correlated motion parameter to something sensible until the refinement is better converged.

This PDF looks pretty clean to my eye. It's possible the empirical correction Mikkel referenced has already been applied-- the beamline scientist could confirm this for you.

Happy modeling! 

[Peter Metz]

To write the intuitive statement "...thus perturbs the low-r region of the reduced PDF..." a bit more concisely, the effect in real space would be something mathematically like the convolution of a gaussian and a sinc function, the latter arising because of the truncation at finite Q of the diffraction signal.

Here's a rough demonstration with and without the influence of termination ripples.

Cheers!

[Peter Metz]

[chenguang yang]

Thanks Mikkel! That helps a lot!

Chenguang Yang 

[chenguang yang]

Thank you Peter for your help! I appreciate it!

Chenguang Yang 

在2020年11月10日星期二 UTC-7 上午11:42:29<Peter Metz> 写道：