This article deals with uncertainty in the analysis of strain in silicon nanoscale structures and devices using nanobeam electron diffraction (NBED). Specimen and instrument related errors and instabilities and their effects on NBED analysis are addressed using a nanopatterned ultrathin strained silicon layer directly on oxide as a model system. We demonstrate that zero-loss filtering significantly improves the NBED precision by decreasing the diffuse background in the diffraction patterns. To minimize the systematic deviations the acquired data were verified through a reliability test and then calibrated. Furthermore, the effect of strain relaxation by specimen preparation using a FIB is estimated by comparing profiles, which were acquired by analyzing slices of strained structures in a 220-nm-thick region of the sample (invasive preparation) and the entire strained nanostructures, which are embedded in a thicker region of the same sample (noninvasive preparation). Together with the random deviation, the corresponding systematic shift results in a total deviation of ∼1 × 10(-3) for NBED analyses, which is employed to estimate the measurement uncertainty in the thinner sample region. In contrast, the strain in the thick sample region is not affected by the preparation; the systematic shift reduces to a minimum, which improves the total deviation by ∼50%.
Download full-text PDF |
Source |
---|---|
http://dx.doi.org/10.1017/S1431927611012657 | DOI Listing |
Enter search terms and have AI summaries delivered each week - change queries or unsubscribe any time!