Differential Phase Contrast Imaging with Reduced Dynamical Diffraction Effect

Abstract

Phase contrast imaging in STEM has become a very exciting field after the rapid developments in segmented/pixel type detectors, and some powerful imaging techniques have been studied including Differential Phase Contrast (DPC) [1,2] and ptychography [3]. Particularly, DPC is a promising method for characterizing local internal electric and/or magnetic fields of functional materials (ex. magnet, semiconductor device) However, one of the most critical problems on the DPC imaging is artifact from diffraction contrast (ex. bend contours or equal thickness fringe): the acquired images of internal electric and/or magnetic field are severely disturbed by the waving background contrasts due to diffraction contrast. This is particularly critical when we perform DPC STEM for a sample along low index crystal axis at low magnification (that is, large field of view), where many diffracted beams are strongly excited and an intensity distribution of CBED patterns becomes highly complex, due to intensity exchanges between diffracted and direct beams, resulting in confusing image contrasts. To overcome such problems, we adopted the idea of Precession Electron Diffraction (PED) [4] into DPC STEM imaging to avoid these diffraction contrasts. The PED is known as a method for suppressing dynamical scattering effects in electron diffraction. Thus, we tried to apply the PED technique to DPC imaging (precession DPC) to reduce diffraction contrast. A p-n junction in GaAs semiconductor was selected as a model sample. The sample was prepared by FIB, where the thickness of the sample was controlled to be about 200 nm. Detail information about this sample can be found elsewhere [5]. The same sample was characterized by DPC STEM in the previous study [6], where the inner electric field was clearly observed and analysed in a quantitative manner. The STEM observation results reported in this paper was performed using an aberration corrected microscope (JEOL, JEM-ARM200F) equipped with an 8-segmented STEM detector (JEOL, SAAF Octa). The optical condition for DPC STEM was set to be high sensitivity mode (STEM low-mag, incident angle ~ 150 rad, effective camera length ~ 4000 cm). The DPC signals were calculated from the second layer of SAAF Octa detector (signals from channel 5~8). In the experiment, the GaAs [110] zone axis was aligned along the optic axis, and DPC images were acquired with and without precession control. In the former case, the optical system was fixed, and the relative angles between the incident beam and the specimen were manually controlled by X/Y tilt of the specimen holder. The range of precession angle was ± 0.5 degree for X and Y directions around the [110] axis, where the tilting step was set to be about 0.1 degree. A set of raw images was acquired for each step, so that we got 11×11 = 121 data sets from the same sample region. These sequential images were aligned by cross correlation processes, stacked and averaged into single data set, and then used for calculating the precession DPC image. For reference, a DPC image without special care for diffraction contrast was also taken along the [110] crystal axis direction. ± 0.5 degree for X and Y directions around