Premium
Reliability of high‐resolution electron backscatter diffraction determination of strain and rotation variations using phase‐only and cross correlation
Author(s) -
Riedl T.,
Wendrock H.
Publication year - 2014
Publication title -
crystal research and technology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.377
H-Index - 64
eISSN - 1521-4079
pISSN - 0232-1300
DOI - 10.1002/crat.201300217
Subject(s) - standard deviation , rotation (mathematics) , optics , cross correlation , noise (video) , physics , phase (matter) , correlation function (quantum field theory) , diffraction , resolution (logic) , displacement (psychology) , correlation coefficient , computational physics , mathematics , spectral density , mathematical analysis , statistics , geometry , psychology , quantum mechanics , artificial intelligence , computer science , image (mathematics) , psychotherapist
This contribution analyzes the reliability of strain and rotation variation determination using cross‐ as well as phase‐only correlation of experimental wide‐angle electron backscatter diffraction (EBSD) patterns. For both rotation and three‐point bending the resulting displacement gradient tensor components are examined in terms of magnitude and statistical scatter as a function of various correlation procedure parameters and signal‐to‐noise ratio. It is shown that on the one hand the Fourier filter width has a major impact on the strain results. At higher noise level a smaller filter width has to be applied for obtaining maximum precision. On the other hand, the influence of the degree of overlap of the regions of interest positioned in the patterns is less important. For both rotation and bending experiments the cross‐correlation variant yields a smaller standard deviation with respect to phase‐only correlation, in particular for elevated noise level. The difference is attributed to the stronger propagation of noise effects in the course of phase‐only correlation function calculation and fitting. In the rotation experiment a standard deviation of ∼1.0 × 10 −4 , averaged over the displacement gradient tensor components, and a rotational precision of ∼1.5 × 10 −4 rad have been achieved by using optimized evaluation settings.