1202.5005 (Alexander Stukowski)

Structure identification methods for atomistic simulations of
crystalline materials    [PDF]

Alexander Stukowski

We describe a new computational analysis technique to identify lattices and
structural defects in large-scale atomistic computer simulations of crystalline
materials. Our approach is based on a user-supplied catalog of template
structures (motives). We use a graph-based pattern matching algorithm to find
occurrences of periodic and non-periodic atomic arrangements in atomistic
snapshots, and to generate a high-level description of a simulated
microstructure. The method covers defects such as stacking faults, grain
boundaries, crystal interfaces, point defects, and defect clusters in a wide
range of crystal lattices. In contrast to existing methods, the proposed
pattern matching algorithm is able to identify crystal structures with a
polyatomic basis. Furthermore, we discuss how the local lattice orientation can
be determined to measure crystal rotations, and how a defective crystal can be
mapped to an ideal reference state. Finally, we derive a computational method
for detecting and characterizing disclination defects in the analysis data.

View original: http://arxiv.org/abs/1202.5005