On the detection and classification of material defects in crystalline solids after energetic particle impact simulations

Abstract

We present a fingerprint-like method to analyze material defects after energetic particle irradiation by computing a rotation invariant descriptor vector for each atom of a given sample. For ordered solids this new method is easy to use, does not require extreme computational resources, and is largely independent of the sample material and sample temperature. As illustration we applied the method to molecular dynamics simulations of deuterated and pristine tungsten lattices at 300 K using a primary knock-on atom (PKA) of 1 keV with different velocity directions to emulate a neutron bombardment process. The number of W atoms, that are affected after the collision cascade, have been quantified with the presented approach. At first atoms at regular lattice positions as well as common defect types like interstitials and vacancies have been identified using precomputed descriptor vectors. A principal component analysis (PCA) is used to identify previously overlooked defect types and to derive the corresponding local atomic structure. A comparison of the irradiation effects for deuterated and pristine tungsten samples revealed that deuterated samples exhibit consistently more defects than pristine ones.

On the classification and quantification of crystal defects after energetic bombardment by machine learned molecular dynamics simulations

“…Although, this analysis is limited by the definition of spatial region around the W atoms, we have a good agreement by finding the two Frenkel pairs (single vacancy and a single self-interstitial atom) formed at the location in the damaged W lattice. These W atoms, called SIA, have the lowest probability to be at a lattice position [20]. However, our method is capable to identify the W atoms that are in the vicinity of the SIAs, which is observed as a distorted region.…”

“…In the same figure, we add a fitting curve to the number of SIA and atoms in the distorted Counts ML potential a) Figure 1: Histogram of the distance difference between the interstitial DV and the W atoms in the damaged sample after relaxation process. MD simulation were performed by the ML potential in a), J-T-ZBL in our previous work [20], and the AT-EAM-FS potential in c). vacancy, v V (T = 0), in a bcc unit cell, at 0 K [20].…”

“…MD simulation were performed by the ML potential in a), J-T-ZBL in our previous work [20], and the AT-EAM-FS potential in c). vacancy, v V (T = 0), in a bcc unit cell, at 0 K [20]. In Fig.…”

“…In Fig. 3, we report the average of the number of Self-Interstitial-Atom (SIA) and atoms in its distorted region in a), W atoms next to a single vacancy in b), Frenkel pair formation in c), and type-A (W atom in the vicinity of a split-vacancy or di-vacancy [20]) in d) defects as a function of the impact energy. Since, a fitting curve to the formation of material defects as atoms in SIAs in different metal materials has been proposed by Bacon et al and Stoller et al [39,40], where Counts = αE β PKA ; whith E PKA as the PKA energy, α and β are fitting parameters.…”

Computational study of crystal defect formation in Mo by a machine learning molecular dynamics potential

“…which is then approximated in terms of spherical harmonic functions, Y lm (r), and a set of basis functions in radial directions g n (r) as c (i) nlm = g n Y lm |ρ i [18,19]. The sum over the order m of the squared modulus of the coefficients c nlm is invariant under rotations around the central atom [28].…”