[benchmark] low energy reflection and sputtering of tungsten surfaces by MD

[drobnyjt]

Description
A recent paper published sputtering yields and reflection coefficients of several ion species, including tungsten, from tungsten surfaces, calculated with MD. The noble gas ions especially are worth comparing to RustBCA, and the distinction between the sputtered and reflected distributions would be nice to compare as well.

Source
https://www.sciencedirect.com/science/article/pii/S0022311525002508

[drobnyjt]

Some notes

From the paper:

• "No electronic stopping was applied"
• "A total of 5000 impacts were carried out for each parameter set (500 per thermalized surface)."

500 impacts means a Monte Carlo error of ~1/sqrt(500) = 0.044...

Li et al.'s potential, used in the paper, has a Morse pairwise component for W-W with the following form (for RustBCA input):

D0_eV = 2.87454 #eV
r0_nm = 0.238631 #nm
alpha_1_nm = 13.3682 #1/nm
D0 = D0_eV * 1.602e-19
r0 = r0_nm * 1e-9
alpha = alpha_1_nm * 1e9
pot = {"MORSE": {'D': D0, 'r0': r0, 'alpha': alpha}}

I think my methodology will include comparisons to:

RustBCA defaults
Morse potential (ck = 0)
Bjorkas et al. ZBL-cubic spline WW potential (ck = 0)

However, in my initial investigations, I've found that the presented MD results do not necessarily agree with experimental measurements at theta=0, so the value of this as a benchmark of RustBCA may be flawed.

[drobnyjt]

Some preliminary results suggest that, for these cases, sputtering yields from RustBCA do not agree with MD at non-normal angles of incidence:

However, the MD results are ~2x off for the noble gas ions when compared to experiments at normal incidence, which makes this test somewhat inconclusive. This prompted me to look for pre-existing experimental data - unfortunately, none seems to exist for non-normal incidence at energies ~100 eV, but I was able to find normal-incidence yields for W on W and Ar on W:

Interestingly, the best agreement to Ar on W was found with ZBL/Morse and/or ZBL/WW (Bjorkas et al. potential), and is acceptable; the Kr-C yields are a little low. The MD results are close to the experimental data.

The best agreement for W on W was Kr-C or a ZBL/Morse with a higher Es (as suggested by previous work with MD sputtering simulations of tungsten). The MD - which I could only feasibly include one point, the sputtering yield they found for 100 eV was essentially zero - is ~2x off.

The script for this comparison has been uploaded to examples/f_kporha_et_al_benchmark.py on the new_rcpr branch.

It is possible that, for heavy ions, either multi-body effects or crystal structure dominate the angular dependence of sputtering yield, putting limits on the applicability of the BCA for these cases - but in practice, most real targets are polycrystalline and rough, so the latter fact may be of less practical importance. It is also possible that the MD results are not accurate, but that does not necessarily mean the BCA is for these cases. Previous experience with light ions at very low energies was very promising, but heavy ions are a different physics regime. Further investigation is warranted.