Compute xrd for amorphous polymer

LAMMPS LAMMPS General Discussion
1

I am working with Kapton a polymer widely used in space. The first step of my lammps simulation is to find a kapton supercell with a structure similar to the experimental one. For that I want to compare the XRD of my cell with the XRD found experimentally.

Experimentally : 2 sharp peaks linked with crystalline part of Kapton, and a broad peak linked with the presence of defects / non crystalline parts and interchain broadening distance

In lammps I calculate the xrd using the following commands :

compute         1 all xrd 0.711 C H O N 2Theta 0.04 0.65 c 0.04 0.04 0.04 LP 1 manual
fix xrd_output_90 all ave/time 100 5 500 c_1[*] file xrd_output_50K.txt mode vector
run             5000

I build my supercell by starting with a kapton structure perfectly crystalline and in a very low density (chains far away from each other) then I added randomly some dimers and monomers between the chains. After I perform a compression step until the good density is reached and finally I equilibrate before computing the XRD.

I tried initializing my supercell in so many different ways but I always end up with an XRD that only have the sharp peaks linked with the crystalline part. I even tried not adding the crystalline kapton and only random dimers and monomers but the xrd still only show the sharp peaks.

I don’t have acces to the article cited in the lammps compute xrd web. I am wondering if compute XRD enables computing correctly the XRD even for amorphous materials? Does someone know more about how compute xrd works? Or has an idea of improvement?

Thank you in advance for any help,

2

From the LAMMPS manual, compute xrd calculates X-ray diffraction predictions using only three inputs:

  1. The atoms’ coordinates;
  2. A per-element Compton scattering adjustment;
  3. An optional manually-specified reciprocal grid spacing.

You will have to determine whether any of these is responsible for what you are seeing vs what you think you should be seeing. Consider especially that any periodic simulation (with compatible box size) must artificially favor crystalline over amorphous structure – as an extreme example, a simulation box only one unit cell large can never display any non-crystalline (non-periodic) behaviour.