Materials characterisation by angle-resolved scanning transmission electron microscopy

Abstract

Solid-state properties such as strain or chemical composition often leave characteristic fingerprints in the angular dependence of electron scattering. Scanning transmission electron microscopy (STEM) is dedicated to probe scattered intensity with atomic resolution, but it drastically lacks angular resolution. Here we report both a setup to exploit the explicit angular dependence of scattered intensity and applications of angle-resolved STEM to semiconductor nanostructures. Our method is applied to measure nitrogen content and specimen thickness in a GaNxAs1−x layer independently at atomic resolution by evaluating two dedicated angular intervals. We demonstrate contrast formation due to strain and composition in a Si- based metal-oxide semiconductor field effect transistor (MOSFET) with GexSi1−x stressors as a function of the angles used for imaging. To shed light on the validity of current theoretical approaches this data is compared with theory, namely the Rutherford approach and contemporary multislice simulations. Inconsistency is found for the Rutherford model in the whole angular range of 16–255 mrad. Contrary, the multislice simulations are applicable for angles larger than 35 mrad whereas a significant mismatch is observed at lower angles. This limitation of established simulations is discussed particularly on the basis of inelastic scattering.

Angular dependence of fast-electron scattering from materials

“…The simulation clearly diverges from the experimental data around 35 mrad, as do other simulations shown in Ref. [11]. Here, the experimental data were normalized to the right-most data point.…”

“…More recent examples of the lack of agreement between the measured angular dependence of scattering and theory using standard simulations incorporating only elastic scattering and phonon excitations were provided by Müller-Caspary et al [11] who used what they dubbed angle-resolved STEM to explore several semiconductor specimens. Effectively using a series of detectors, each incrementally covering a small angular range, they found an underestimate of measured intensities compared to simulations at angles roughly three to four times α (9 mrad).…”

An optical configuration for fastidious STEM detector calibration and the effect of the objective‐lens pre‐field

“…Necessarily the spacing of these lines depends on the magnification and pixel‐sampling of the scan‐output of the STEM imaging settings. However, importantly for the practicality of our approach, as this scan makes use of the microscope scan‐generator it takes only a few seconds to acquire an entire detector scan rather than several hours (Müller‐Caspary et al ., ). Further, as the same aperture is used for the calibration and STEM imaging, the scan approach is compatible with either the so‐called ‘dropped gain’ or ‘dropped‐current’ approaches described by Jones (), where as always care should be taken to avoid saturation and ensure linearity (LeBeau & Stemmer, ; Ishikawa et al ., ; Yamashita et al ., ).…”

The atomic lensing model: New opportunities for atom-by-atom metrology of heterogeneous nanomaterials

“…These counting results enable one to retrieve the 3D atomic structure [8,9,10,11,12]. For heterogeneous nanocrystals, the dependency of the image intensities on the atomic number Z may be used to detect dopant atoms [13,14] or determine the chemical concentration per atomic column [15,16,17,18].…”