Application of phase-retrieval x-ray diffractometry to carbon doped SiGe(C)∕Si(C) superlattice structures. II. High resolution reconstruction using neural network root finder technique

A neural network root finder approach for finding complex roots of high-degree complex polynomials was applied as part of the phase-retrieval x-ray diffractometry technique to reconstruct strain profiles in SiGe(C)∕Si(C) superlattice structures. The high spatial resolution, 5Å, as a result of significantly higher degree of the complex polynomial, 1400 roots, allowed us to obtain more accurate results for the strain profile distribution in SiGe(C)∕Si(C) superlattice structures. Previously, such high quality analysis was fundamentally impossible due to the limitations imposed by conventional numerical methods of finding complex polynomial roots.