Three-dimensional visualization of dislocations in a ferromagnetic material by magnetic-field-free electron tomography

Abstract

In conventional transmission electron microscopy, specimens to be observed are placed in between the objective lens pole piece and therefore exposed to a strong magnetic field about 2 T. For a ferromagnetic specimen, magnetization of the specimen causes isotropic and anisotropic defocusing, deflection of the electron beam as well as deformation of the specimen, which all become more severe when the specimen tilted. Therefore electron tomography on a ferromagnetic crystalline specimen is highly challenging because tilt-series data sets must be acquired without changing the excitation condition of a specific diffraction spot. In this study, a scanning transmission electron microscopy (STEM) tomography method without magnetizing a ferromagnetic specimen has been developed for three-dimensional (3D) visualization of dislocations in α-Fe, which is a typical ferromagnetic material. Magnetic-field-free environment down to 0.38 ± 0.07 mT at the specimen position is realized by demagnetizing the objective lens pole piece of a commercial STEM instrument. By using a spherical aberration corrector with the magnetic-field-free environment, an "aberration corrected Low-Mag STEM mode" with no objective lens field reaches a convergence semi angle ∼1 mrad and a spatial resolution ∼5 nm, and shows an adequate performance of imaging dislocations under a two-beam excitation condition for a low-index diffracted beam. The illumination condition for the aberration corrected Low-Mag STEM mode gives no overlap between the direct beam disk (spot) and neighboring diffraction disks. An electron channeling contrast imaging technique, in which an annular detector was located at a doughnut area between the direct beam and the neighboring diffracted beams, was effectively employed with the aberration corrected Low-Mag STEM mode to keep image intensity high enough even at large specimen-tilt angles. The resultant tomographic observation visualized 3D dislocation arrangements and active slip planes in a deformed α-Fe specimen.

Five-second STEM dislocation tomography for 300 nm thick specimen assisted by deep-learning-based noise filtering

“…During the 5 s, about 70 frames could be acquired without stopping the goniometer rotation throughout the tilt angle range, from − 70° to + 70°, which is not inferior to the number of tilt-series images acquired by a conventional method. In order to evaluate the quality of rapid STEM tomography comparing to the conventional method, we also acquired tilt-series images from the same field of view in the almost same condition but taking a longer frame time of 1.6 s and intermittent manipulation of the goniometer as performed in the previous studies 28 , 56 . The tilt-series images were acquired every 2° throughout the range from − 70° to + 70°.…”

“…On the other hand, in the scanning TEM (STEM) mode which is basically free from the chromatic aberration of the imaging lens system, the deterioration of image quality due to the specimen thickness is much smaller than that in the CTEM mode. In fact, a 3D arrangement of dislocations in an iron slab, a 400 nm thick specimen, was clearly visualized by STEM tomography 28 . In this particular study using a 300 keV electron beam, dislocation line contrast was visible even when the specimen was tilted by 60°, i.e., the effective specimen thickness reached to 800 nm.…”

Deep learning-based noise filtering toward millisecond order imaging by using scanning transmission electron microscopy

“…Multiple factors, such avoiding chromatic aberration 8 , 9 in the imaging lenses and somewhat improved robustness to dynamical diffraction effects, mean that Scanning TEM (STEM) is the technique of choice for high-resolution analytical studies of crystalline materials. The STEM can visualize even thicker samples, such as three-dimensional structures of dislocations in a steel with a thickness of 300 nm or more 10 , 11 , micro-meter dimensions of polymer samples 12 and biological cells with a thickness of 1 m 9 . The working principle of the STEM that the targeted area is scanned and imaged point-by-point by a converged electron beam leads a lower temporal resolution in principle compared to the CTEM using a continuous illumination.…”

Deep learning-based noise filtering toward millisecond order imaging by using scanning transmission electron microscopy

“…This limitation on the sample thickness is considerably pushed up in the Scanning TEM (STEM) mode which is basically free from imaging lens systems 8,9 . The STEM can visualize even thicker samples, such as three-dimensional structures of dislocations in a steel with a thickness of 300[nm] or more 10,11 , micro-meter dimensions of polymer samples 12 and biological cells with a thickness of 1[m] 9 . The working principle of the STEM that the targeted area is scanned and imaged point-by-point by a converged electron beam, which leads a lower temporal resolution in principle compared to the CTEM using a continuous illumination.…”