TY - JOUR
AU - Shibata,, Naoya
AB - Abstract Differential phase contrast (DPC) imaging in scanning transmission electron microscopy is a technique to visualize electromagnetic field distribution inside specimens at high spatial resolution. However, diffraction contrast strongly hampers electromagnetic contrast in DPC images especially in polycrystalline samples. In this paper, we develop an imaging technique to effectively suppress diffraction contrast in DPC images. It is shown that a magnetic structure in a Nd–Fe–B permanent magnet was clearly visualized by averaging 64 DPC images with various specimen-tilt conditions. This is because the diffraction contrast in DPC images sensitively and randomly varies with crystal orientation and thus almost vanishes by averaging specimen-tilt image series. We further investigated two types of residual diffraction contrast in the tilt-series averaged DPC images: weak contrast inside grains and strong contrast at grain boundaries. We found that the former can be suppressed by averaging more DPC images, whereas the latter can be suppressed by the tilt-series averaging with wider range of specimen tilt. The tilt-series averaging method enables DPC to visualize electromagnetic structures even inside polycrystalline materials. differential phase contrast, diffraction contrast, Nd-Fe-B, permanent magnet, magnetic domain, LACBED Introduction Differential phase contrast (DPC) is a technique to visualize electromagnetic field distribution inside specimens by scanning transmission electron microscopy (STEM) [1–4]. A segmented detector in STEM detects deflection of the transmitted electrons due to electromagnetic fields inside specimens (Fig. 1). Magnetic domains [5–7], p–n junctions [8], quantum wells [9], magnetic skyrmions [10, 11] and atomic electric fields [12–15] have been visualized by DPC STEM at high spatial resolution. Combining with other detectors, DPC images can be obtained simultaneously with other STEM images such as annular dark field (ADF) image and/or energy dispersive X-ray spectroscopy elemental maps. This enables direct correlation between local electromagnetic fields and microstructures at high spatial resolution in the same specimen regions. Therefore, DPC STEM is a promising method for fundamental understanding of the interactions between electromagnetic fields and microstructures inside materials. Fig. 1 Open in new tabDownload slide Schematic illustration of DPC STEM for magnetic field imaging. SAAF detector [23] (16-segmented detector in this figure) detects Lorentz deflection of the transmitted electron. ADF detector provides structural information simultaneously. Fig. 1 Open in new tabDownload slide Schematic illustration of DPC STEM for magnetic field imaging. SAAF detector [23] (16-segmented detector in this figure) detects Lorentz deflection of the transmitted electron. ADF detector provides structural information simultaneously. In permanent magnets, it is well known that the interaction between magnetic domain walls and microstructures is crucially important to control coercivity of the materials. In recent years, high-coercivity permanent magnets such as Nd–Fe–B sintered magnets are highly demanded for the application in electric vehicles. Microstructures such as grain boundaries (GB) and GB phases are considered to be strongly impeding the mobility of magnetic domain walls, and thus improve magnetic coercivity of the polycrystalline magnets [16]. To further improve the coercivity of the Nd–Fe–B sintered magnets, in particular at high operational temperature of the electric motors (~470 K), the design and control of microstructures has been sought for >30 years [17–20]. In the previous studies on the microstructures of Nd–Fe–B sintered magnets, the formation of many kinds of GBs and GB phases with various compositions, structures and magnetic properties have been reported. On the other hand, direct observations of magnetic domain structures have been performed by Lorentz TEM or electron holography. However, in these phase imaging techniques, it is extremely difficult to retrieve detailed microstructural information [21, 22]. Thus, it has been difficult to directly correlate microstructure and magnetic structure from the same specimen regions. Therefore, fundamental interaction mechanisms between GBs and magnetic domain walls in sintered magnets still remain elusive. Direct and simultaneous characterization of microstructures and magnetic structures from the same specimen regions is thus demanded. Although DPC STEM may be usable for such purpose, interpretation of DPC images obtained from sintered materials is currently very complex. In actual experimental DPC images, the image contrast not only contains electromagnetic field induced contrast, but also contains diffraction condition induced contrast (=diffraction contrast). In single crystalline samples, diffraction contrast may be circumvented by controlling specimen tilt condition to avoid strong Bragg excitation. Contrastingly, for polycrystalline samples such as sintered magnets, it is impossible to find suitable specimen tilt condition that sufficiently suppresses diffraction contrasts in all the grains in the field of view. It is worth noting that most of the previous studies by DPC have been limited to single crystals or epitaxially grown hetero structures due to this problem. Therefore, to observe magnetic domain structures in polycrystalline sintered magnets by DPC STEM, it is crucial to effectively remove and/or suppress diffraction contrast from the experimental DPC images. In this paper, we demonstrate an imaging technique to effectively suppress diffraction contrast in DPC images by using specimen-tilt series averaging. We further investigate two types of residual diffraction contrasts in the specimen-tilt averaged DPC images, and develop a method to evaluate and suppress such residual contrasts. Methods Concept of specimen-tilt series averaging method DPC signals are calculated from the intensities detected in each detector segment to measure beam deflection of the bright field (BF) disk due to electromagnetic fields inside specimens (Fig. 1). Diffraction contrast in DPC images is mainly caused by the diffraction patterns inside the bright disk. Typically, black lines, which satisfy the Bragg condition, appear in the bright disk resulting from Bragg reflections, which is called higher order Laue zone lines when the specimen is illuminated along a zone-axis. These patterns cause DPC signals even if there are no electromagnetic fields and thus show up as diffraction contrast in the experimental DPC images. Since diffraction patterns sensitively depend on the crystal orientation, this diffraction contrast is expected to vary with slight crystal orientation differences. In single crystals such as specimens of p–n junctions and magnetic skyrmions, the diffraction contrast may be sufficiently suppressed by selecting the sample tilt condition for avoiding strong Bragg excitations. Contrastingly, in polycrystalline specimens with randomly oriented crystal grains, it is almost impossible to find a single sample tilt condition without any strong Bragg excitations in all the crystal grains in the field of view. To investigate diffraction contrast behaviors in DPC with varying crystal orientations, we simulated large-angle convergent-beam electron diffraction (LACBED) patterns of Nd2Fe14B single crystal from the [100] axis (Fig. 2) by using Many-Beam dynamical-simulations and least-squares FITting software [24]. If the LACBED patterns are trimmed into a circle with the size of the bright-field disk, one can obtain a bright-field disk pattern of a particular crystal orientation as a corresponding position inside the LACBED pattern. This is because the bright-field disks do not overlap with the diffracted disks under a typical illumination condition of DPC imaging, and the LACBED patterns are calculated in the absence of the interference between the bright-field disk and the diffracted disks. The red and yellow disks in Fig. 2a and b are examples of bright-field disks with the interval of 1.0 and 0.2°, respectively. The semiconvergence angle of these bright disks is 852 μrad, which is the same value in the experimental condition as shown later. The [100] LACBED pattern with large area (Fig. 2a) confirms that CBED patterns in the bright-field disks strongly depend on the crystal orientation. Since the bright-field disk patterns are not uniform in the different tilting conditions (Fig. 2b), diffraction contrast should show up in DPC images. These simulated [100] LACBED patterns of four different sample thicknesses (Fig. 2b–e) suggest that thicker samples have stronger diffraction contrast since the intensity of LACBED in thicker samples varies more strongly within the bright-field disk. Fig. 2 Open in new tabDownload slide Simulated [100] LACBED patterns of Nd2Fe14B single crystal at an accelerating voltage of 200 kV. The values of ‘t’ in lower right of the patterns represent sample thicknesses. The yellow and red disks in (a) and (b) are the examples of bright disks. Fig. 2 Open in new tabDownload slide Simulated [100] LACBED patterns of Nd2Fe14B single crystal at an accelerating voltage of 200 kV. The values of ‘t’ in lower right of the patterns represent sample thicknesses. The yellow and red disks in (a) and (b) are the examples of bright disks. Here, we assume that, when tilting the samples, diffraction contrast in DPC images should sensitively and randomly vary, whereas electromagnetic field contrast should be robust. The minimum step of specimen-tilt control in specimen holder is typically around 0.1° (= 1.7 mrad), and this is larger than the typical convergence semiangle of DPC imaging. If specimen is tilted, a bright-field disk pattern is largely different and not correlated with the nontilted pattern. Thus, in approximation, the diffraction contrast in DPC images could be randomly changing with respect to the specimen-tilt angle. It is worth noting that the present discussion based on the LACBED patterns is valid only when an electron probe is located inside a single crystal grain, and GBs are out of the scope of this assumption. Under this assumption, it may be recalled that the precession electron diffraction technique has been used to reduce dynamical scattering effects in electron diffraction [25]. A series of raw DPC images acquired at slightly different sample tilt conditions shown by the red and yellow circles in Fig. 2 should contain unchanged electromagnetic field contrast component and randomly varying diffraction contrast component. In the specimen-tilt series of DPC images, diffraction contrast component in DPC signals at a certain probe position are expected to be suppressed by averaging the DPC signals (Fig. 3). Nakamura et al. [26] successfully decreased diffraction contrast component by the tilt-series averaging technique in a single crystal case. They demonstrated the tilt-series averaging method around a zone-axis of GaAs can clearly visualize a p–n junction electric field even under the strong diffraction conditions. In this study, we extend this method to polycrystalline materials and show their validity. Fig. 3 Open in new tabDownload slide Conceptual diagram of tilt-series averaging technique applied to DPC images. In the averaged image of tilt series (tilt-series averaged DPC image), only diffraction contrast could be suppressed. Fig. 3 Open in new tabDownload slide Conceptual diagram of tilt-series averaging technique applied to DPC images. In the averaged image of tilt series (tilt-series averaged DPC image), only diffraction contrast could be suppressed. Experimental details In this study, high-coercivity Nd–Fe–B polycrystalline magnets were selected as model samples. Initial alloy was prepared by melting with high-frequency induction heating followed by the liquid quenching technique. The initial alloy ribbon was crushed into powders and then sintered at 948 K for 180 s. The chemical composition before sintering was Fe67.34Nd30.78B0.95Ga0.40Pr0.38Cu0.12Al0.04 in weight percentage (Fe79.40Nd14.05B5.77Ga0.38Pr0.18Cu0.12Al0.09 in at.%). The c-axis (magnetic easy axis) of each Nd2Fe14B grains were aligned by unidirectional pressing at 1053 K after sintering but other axes were not controlled. Then, the sintered magnet was infiltrated with 2 wt% Nd–Cu of eutectic composition and annealed for 120 min at 893 K. This infiltration process made Nd-rich grain boundary phases surrounding the matrix. The coercivity, |${\mu}_0{H}_c$|⁠, maximum energy product, |${(\mathrm{BH})}_{\mathrm{max}}$| and remanent magnetization, |${\mu}_0{M}_r$|⁠, of this magnet were measured to be 2.10 T, 358 kJ/m3 and 1.35 T, respectively. A specimen for DPC STEM observation was prepared by mechanical grinding, dimpling and ion milling. DPC STEM observation was carried out on an aberration-corrected JEM-2100F (JEOL) at 200 kV. The microscope was operated under objective lens off condition in order not to destroy the magnetic structures. The instrumentation of the 16-segmented SAAF detector used for this study has been reported elsewhere [23]. Semiconvergence angle was set to be about 852 μrad and the edge of the bright disk was positioned in the middle of the third annulus of the segmented annular all-field (SAAF) detector. Figure 1 shows schematic diagram of the segmented/ADF detectors and the bright-field disk. The optical conditions for DPC STEM imaging was fixed throughout the experiment. The focal depth is estimated to be about 6 μm for each DPC imaging. The sample tilt angle was manually controlled by X/Y tilt of the specimen holder. To investigate the influence of the tilt angle range, we obtain two different tilt-series of DPC images (Series A and B). In Series A, the sample tilt angle was varied from 0.0 to 1.4° with 0.2° intervals for X/Y axes of the sample holder; i.e. Series A consists of |$8\times 8=64$| images with slightly different tilt conditions. To avoid image distortion due to sample drift, two experimental images (⁠|$512\times 512\ \mathrm{pixels}$|⁠) with a dwell time of 150 μs/pixel were recorded on each tilt condition and then averaged. In Series B, the sample tilt angle was varied from 0.0 to 7.0° with 1.0° intervals for X/Y axes and thus it also consists of 64 images. The bright-field disks of the tilt conditions of Series A and B correspond to yellow and red disks in Fig. 2a, and b, respectively, but the actual crystal orientations of each grain in the specimen are unknown in the experiment. In a TEM sample with typical thickness of 100 nm, the projection of the vertical plane inside a specimen is estimated to be blurred by 2.4 and 12 nm with the tilt angles of 1.4 and 7.0°, respectively. Thus, blurring effect due to sample tilt in the averaged DPC images should be relatively smaller in Series A than Series B. ADF and BF STEM image (summed image of SAAF segments in the bright-field region) were simultaneously acquired in each sample tilt condition. Series A and B image sets contain 64 experimental DPC, ADF and BF images. DPC images were generated by the center of mass method from all intensities detected in the 16 segments [27]. Figure 4 shows the experimental ADF and DPC STEM images under two different tilt conditions. The inset numbers in the lower right position represent serial numbers of tilt conditions. The tilt conditions of the first and second images are (X, Y) = (0.0, 0.0) and (0.0, 0.2), respectively. It is seen that the contrast of ADF images (Fig. 4a and d) is relatively robust to the sample tilt conditions. The elongated Nd–Fe–B grains are well aligned, and the c-axes of them are in the horizontal direction of the images. The direction of the magnetic polarization should be roughly along the horizontal direction of the images. In DPC color vector image (Fig. 4b and e), the strength and direction of the calculated field are represented by brightness and color, respectively. The color wheels represent the direction of magnetic field. In these DPC color images, the contrast of magnetic domains appears to be severely disturbed by the strong diffraction contrast. Comparing ADF images (Fig. 4a and d) with DPC strength image (Fig. 4c and f), strong diffraction contrast in DPC images may be caused by the local strain or specimen bending. Although the two experimental DPC images (Fig. 4b and e) have similar diffraction contrast as indicated by the arrows, their positions are apparently different. Thus diffraction contrast at each pixel is expected to be randomly varied, and the sample tilt step (0.2°) is adequate for changing the diffraction contrast. Note that the TEM specimen at the top side of the images is thicker than the bottom right, and this thickness slope slightly affect the ADF and DPC strength images. Fig. 4 Open in new tabDownload slide Experimental (a) ADF, (b) DPC color vector map and (c) DPC strength image on tilt condition (X, Y) = (0.0°, 0.0°). Experimental (d) ADF, (e) DPC color vector map and (f) DPC strength image in tilt condition (0.0, 0.2). The color wheel represents direction of magnetic field. Fig. 4 Open in new tabDownload slide Experimental (a) ADF, (b) DPC color vector map and (c) DPC strength image on tilt condition (X, Y) = (0.0°, 0.0°). Experimental (d) ADF, (e) DPC color vector map and (f) DPC strength image in tilt condition (0.0, 0.2). The color wheel represents direction of magnetic field. Image processing To average the tilt-series DPC images, image registration was performed by cross-correlation processes. The simultaneous tilt-series ADF images were used as a reference for the cross-correlation processes because it has more robust contrast to the sample tilt changes than DPC and BF images. Then, DPC and BF images were aligned based on the cross-correlation coefficients of the ADF images. Strongly distorted images due to specimen drift during the scan were eliminated. It is worth noting that the fields of view should include sample edges or characteristic structures for highly precise image registration. After the image registration, averaged images of the aligned tilt-series DPC and BF images were generated. Results Figure 5a shows a color vector map, Fig. 5b the x component of the DPC and Fig. 5c the y component (x, y denoting the lateral and vertical direction of the magnetic fields). We can clearly observe the wedge-shaped magnetic domain consisting of the straight magnetic domain walls (Wall 1, 2 and 3). Because the magnetic easy axis of grains is along the horizontal direction of the image, Wall 1 is almost parallel to the easy axis and regarded as a 180° domain wall. Fig. 5 Open in new tabDownload slide (a) Color vector map, (b) x and (c) y component DPC images averaged over Series A. (d) Histogram of x component of DPC signals measured under 64 tilt conditions in area Q. The standard deviation map (SD map) of (e) x and (f) y components of DPC signals. Fig. 5 Open in new tabDownload slide (a) Color vector map, (b) x and (c) y component DPC images averaged over Series A. (d) Histogram of x component of DPC signals measured under 64 tilt conditions in area Q. The standard deviation map (SD map) of (e) x and (f) y components of DPC signals. Figure 5d is the histogram of x component of DPC signals from Area Q located inside a single grain and a single magnetic domain (40 × 20 pixels × 64 tilt conditions = 51 200 signals). Since the DPC signals approximately follow normal distribution in the single grain region, it is confirmed that the diffraction contrast randomly varies when tilting a sample as expected above. Here, we assume that the confidence intervals of the averaged DPC signal can be evaluated by a similar method of random-noise analysis. To analyze the residual diffraction contrast as discussed later, standard deviations of x and y components of DPC signals were calculated on each pixel and mapped as ‘standard deviation maps’ (SD maps) as presented in Fig. 5e and f. These SD maps should only contain structural contrast but not magnetic contrast. The standard deviations in the bottom right regions that are relatively thin in the field of view and are relatively small, because diffraction contrast is weaker in thinner region as expected in the LACBED simulations. Discussion Comparing the averaged DPC image (Fig. 5a) with DPC images by single image acquisitions (Fig. 4b and e), diffraction contrast can be effectively suppressed by the tilt-series averaging method. We can clearly see a wedge-shaped magnetic domain in the center of the image and some GBs that seem to pin the magnetic domain walls as indicated by the yellow arrows in Fig. 5b, c. The dark contrast at the magnetic domain walls in Fig. 5a suggests that the magnetic polarizations at the domain walls are directed out of plane of the image, indicating that these walls should be Bloch-type walls. In the y component image (Fig. 5c), Wall 1 cannot be seen because the wall is |$180{}^{\circ}$| domain wall and no vertical DPC component is present. However, Walls 2 and 3 can be seen in the y component image. This suggests that the magnetic polarization near Wall 2 and 3 tends to be parallel to the walls and forms an angle against the magnetic easy axis of grains. The wedge angle of the domain walls is thought to be mainly determined by the balance between the pinning force of the GB phase and the anisotropic energy due to the misalignment of magnetization from the easy axis. The DPC signal in a vacuum part, bottom left of Fig. 5a–c, are caused by stray magnetic field, which flows to the magnetic domain outside of the image. Residual diffraction contrast inside grains In the present case, the 64 tilt conditions are enough to observe these magnetic structures. Only to confirm the existence of magnetic domains, fewer tilt-series images may be sufficient. However, Fig. 5a–c still contains some contrast that cannot be regarded as magnetic origin. The nonmagnetic contrast in the tilt-series averaged DPC images is categorized into two different types: weak contrast inside grains and strong contrast near GB. The residual diffraction contrast inside grains is thought to be due to the finite number of images in the tilt-series. To distinguish the magnetic contrast from the residual diffraction contrast, confidence intervals were calculated at each pixel from the SD map on the assumption that the diffraction contrast follows the normal distribution and behaves as random noise. To determine if two specific pixels have a significant difference in their DPC signals, Welch’s t-test [28] was applied considering their confidence intervals. The magnetic domain structures can be visualized without arbitrariness by plotting the results of Welch’s t-test to a reference area. To guarantee the robustness of the t-test plot, we selected multiple pixels as references and counted the number of reference pixels that show a significant difference at each test pixel, which we will call ‘total t-test score’ here. Figure 6b shows a profile of tilt-series averaged DPC signals along the Line |$\alpha$| in Fig. 6a with the error bars of 95% confidence interval. Note that the Line |$\alpha$| runs inside a single grain and across the Wall 3. We selected 10 pixels in the green region as reference pixels. The total t-test scores of each pixel are also shown in Fig. 6b. The pixels with a total t-test score of 10, for example, show significant differences in DPC signals to all of the 10 reference pixels. Although the t-test was performed with a significance level of 5 and 1%, the total t-test score is almost the same. The pixels with the total t-test score 0 and 10 belong to bright and dark magnetic domain in Fig. 6a, respectively. The results of t-test also show that small variation in DPC signals in the dark magnetic domain do not give a significant difference in the total t-test score, so the signal variations can be regarded as residual diffraction contrast. Fig. 6 Open in new tabDownload slide (a) The enlarged DPC image of x′ component averaged over Series A, where x′ axis is defined as indicated by the arrow parallel to Wall 3. The line profile of the DPC signal and the total t-test score (see text for definition) along (b) line |$\alpha$| and (c) line |$\beta$| in (a). Fig. 6 Open in new tabDownload slide (a) The enlarged DPC image of x′ component averaged over Series A, where x′ axis is defined as indicated by the arrow parallel to Wall 3. The line profile of the DPC signal and the total t-test score (see text for definition) along (b) line |$\alpha$| and (c) line |$\beta$| in (a). To compare the precision of the present method with those of other phase imaging techniques, we evaluate the phase resolution. The confidence intervals are averagely 15.6 μrad, which corresponds to the phase gradient of |$2\pi /161$| rad/nm [29]. If there is a field with a magnitude of the precision in an area larger than the probe size, DPC can detect the field. Thus, to detect the field, there should be the phase shift of |$2\pi /161\ \mathrm{rad}/\mathrm{nm}\times 1.8\ \mathrm{nm}=2\pi /90\ \mathrm{rad}$|⁠, where the probe size is evaluated by |$0.61\lambda /\alpha$|⁠. This value can be compared with the phase resolution in electron holography. The present value is limited by the number of DPC images in the specimen-tilt series, and can be improved by averaging more DPC images. Fundamental limitations may be electron dose and a detection system, and there may be a room to improve the phase resolution up to |$2\pi /1000$| [30]. Residual contrast at GB Figure 6c shows the tilt-series averaged DPC signals and the total t-test scores along the Line |$\beta$| in Fig. 6a, which run across the Wall 3 and a GB. The results of t-test near Wall 3 are similar to Fig. 6b, but DPC signals near GB gives a significant difference. However, it should be noted that GBs are exception of the assumption that the diffraction contrast randomly varies with the specimen-tilt angle. Moreover, signals at the GB are too strong to be considered as magnetic signals. Similar strong contrasts also remain at many GBs within the field of view (Fig. 5a–c). Therefore, the contrast near the GBs in the tilt-series averaged DPC image should be not due to magnetic origin. To investigate the origin of the residual contrast at GBs, we compare the tilt-series averaged DPC image with the tilt-series averaged BF image. Figure 7a shows the averaged BF image of Series A. Because of the uniformity of the grain compositions in this sample, the differences of BF intensity between different grains should be originated from the difference in crystal orientation. The contrast of the tilt-series averaged BF image thus represents the difference of crystal orientation, which is not averaged out by the sample tilting in Series A. It is worth noting that the present DPC signal is obtained by approximating the center-of-mass of the intensity in the BF disk, and the total intensity in the BF disk essentially does not affect the DPC signal. Comparing the tilt-series averaged BF image (Fig. 7a) and the DPC image (Fig. 7b and c), DPC signals at GBs appear to be correlated with the differences of the BF intensities of the neighboring grains. For example, the grain indicated by the yellow arrow has dark contrast in the BF image (Fig. 7a) and is surrounded by the strong signals along the GBs (Fig. 7b and c). Fig. 7 Open in new tabDownload slide (a) BF, (b) DPC color vector map and (c) DPC signal strength image averaged over Series A. The strong DPC signals near grain boundaries correspond to the intensity gradient between the bright to dark grains in BF image. (d–f) show BF, DPC color vector map, DPC signal strength image averaged over Series B, respectively. The grain indicated by the yellow arrow has strong GB signal in Series A but does not in Series B. Fig. 7 Open in new tabDownload slide (a) BF, (b) DPC color vector map and (c) DPC signal strength image averaged over Series A. The strong DPC signals near grain boundaries correspond to the intensity gradient between the bright to dark grains in BF image. (d–f) show BF, DPC color vector map, DPC signal strength image averaged over Series B, respectively. The grain indicated by the yellow arrow has strong GB signal in Series A but does not in Series B. Here, it is expected that the residual contrast at GBs can be suppressed if the averaged BF intensities in each grain get closer to each other. In the dark grains in the averaged BF image, multiple Bragg reflections may be excited and the total intensities in the bright-field disk decrease. This tendency can be discussed in the simulated LACBED pattern shown in Fig. 2a. If the tilt conditions are set as the yellow circles around the [100] axis, the tilt-series averaged BF intensity is expected to be weaker than off-axis conditions, in which Bragg reflections are not strongly excited. To make the tilt-series averaged BF intensity normal, the tilt-range should be set with a wider tilt-range as indicated by the red circles (Series B). Figure 7d–f show tilt-series averaged BF and DPC images of Series B. These images appear to be more blurred than the tilt-series averaged images of Series A because of the larger tilt angle range (0.0–7.0°). As expected, the averaged BF image of Series B (Fig. 7d) has much less contrast than that of Series A, and the tilt-series averaged DPC images of Series B (Fig. 7b and c) show much less GB signals than those of Series A. The grain indicated by the yellow arrow does not have dark contrast in the BF image and the GB signal is not observed in DPC images. These results suggest that the GB signals appear to be related to the gradient of the tilt-series averaged BF image intensity. Concluding remarks It is shown that the tilt-series averaging method can dramatically suppress the strong diffraction contrast in DPC STEM images. The tilt-series averaged DPC images of polycrystalline Nd–Fe–B clearly visualize the presence of magnetic domain structures. These results can be qualitatively explained by the simulated LACBED patterns of Nd2Fe14B single crystal. Welch’s t-test is useful to distinguish the magnetic contrast from the residual diffraction contrast in the tilt-series averaged DPC images. Although strong contrast near the GB due to the difference in the crystal orientation of the adjacent grains was observed, such strong GB contrast can be suppressed by increasing the tilt-angle ranges. Thus, the tilt-series averaging method is useful for applying DPC STEM to polycrystalline samples. Finally, an equivalent effect should be achievable by tilting the electron probe instead of specimens. The probe-tilt system may enable much easier operation and faster acquisition than the present sample tilt-series averaging method. Acknowledgments We thank Prof. K. Tsuda for providing the LACBED simulation software. Funding This work was supported by the SENTAN (JPMJSN14A1) from Japan Science and Technology Agency (JST) and the KAKENHI Grant Numbers JP19H05788 and JP17H06094 from Japan Society for the Promotion of Science (JSPS). A part of this study was supported by “Nanotechnology Platform” (JPMXP09A19UT0375) from the Ministry of Education, Culture, Sports, Science, and Technology in Japan (MEXT). References 1. Rose H ( 1974 ) Phase contrast in scanning transmission electron microscopy . Optik 39 : 416 – 436 . Google Scholar OpenURL Placeholder Text WorldCat 2. Dekkers N H , and de Lang H ( 1974 ) Differential phase contrast in a STEM . Optik 41 : 452 – 456 . Google Scholar OpenURL Placeholder Text WorldCat 3. Rose H ( 1977 ) Nonstandard imaging methods in electron microscopy . Ultramicroscopy 2 : 251 – 267 . Google Scholar Crossref Search ADS PubMed WorldCat 4. Shibata N , Findlay S D, Matsumoto T, Kohno Y, Seki T, Sánchez-Santolino G, and Ikuhara Y ( 2017 ) Direct visualization of local electromagnetic field structures by scanning transmission electron microscopy . Acc. Chem. Res. 50 : 1502 – 1512 . Google Scholar Crossref Search ADS PubMed WorldCat 5. Chapman J N , Batson P E, Waddell E M, and Ferrier R P ( 1978 ) The direct determination of magnetic domain wall profiles by differential phase contrast electron microscopy . Ultramicroscopy 3 : 203 – 214 . Google Scholar Crossref Search ADS PubMed WorldCat 6. Chapman J N , McFadyen I R, and McVitie S ( 1990 ) Modified differential phase contrast Lorentz microscopy for improved imaging of magnetic structures . IEEE Trans. Magn. 26 : 1506 – 1511 . Google Scholar Crossref Search ADS WorldCat 7. Chapman J N , Ploessl R, and Donnet D ( 1992 ) Differential phase contrast microscopy of magnetic materials . Ultramicroscopy 47 : 331 – 338 . Google Scholar Crossref Search ADS WorldCat 8. Shibata N , Findlay S D, Sasaki H, Matsumoto T, Sawada H, Kohno Y, Otomo S, Minato R, and Ikuhara Y ( 2015 ) Imaging of built-in electric field at a p-n junction by scanning transmission electron microscopy . Sci. Rep. 5 : 2 – 4 . Google Scholar OpenURL Placeholder Text WorldCat 9. Lohr M , Schregle R, Jetter M, Wächter C, Wunderer T, Scholz F, and Zweck J ( 2012 ) Differential phase contrast 2.0-opening new “fields” for an established technique . Ultramicroscopy 117 : 7 – 14 . Google Scholar Crossref Search ADS PubMed WorldCat 10. Matsumoto T , So Y-G, Kohno Y, Sawada H, Ishikawa R, Ikuhara Y, and Shibata N ( 2016 ) Jointed magnetic skyrmion lattices at a small-angle grain boundary directly visualized by advanced electron microscopy . Sci. Rep. 6 : 35880 . Google Scholar Crossref Search ADS PubMed WorldCat 11. Matsumoto T , So Y G, Kohno Y, Sawada H, Ikuhara Y, and Shibata N ( 2016 ) Quantum physics: direct observation of ∑7 domain boundary core structure in magnetic skyrmion lattice . Sci. Adv. 2 : e1501280 . Google Scholar Crossref Search ADS PubMed WorldCat 12. Shibata N , Findlay S D, Kohno Y, Sawada H, Kondo Y, and Ikuhara Y ( 2012 ) Differential phase-contrast microscopy at atomic resolution . Nat. Phys. 8 : 611 – 615 . Google Scholar Crossref Search ADS WorldCat 13. Shibata N , Seki T, Sánchez-Santolino G, Findlay S D, Kohno Y, Matsumoto T, Ishikawa R, and Ikuhara Y ( 2017 ) Electric field imaging of single atoms . Nat. Commun. 8 : 15631 . Google Scholar Crossref Search ADS PubMed WorldCat 14. Ishikawa R , Findlay S D, Seki T, Sánchez-Santolino G, Kohno Y, Ikuhara Y, and Shibata N ( 2018 ) Direct electric field imaging of graphene defects . Nat. Commun. 9 : 8 – 13 . Google Scholar Crossref Search ADS PubMed WorldCat 15. Sánchez-Santolino G , Lugg N R, Seki T, Ishikawa R, Findlay S D, Kohno Y, Kanitani Y, Tanaka S, Tomiya S, Ikuhara Y, and Shibata N ( 2018 ) Probing the internal atomic charge density distributions in real space . ACS Nano 12 : 8875 – 8881 . Google Scholar Crossref Search ADS PubMed WorldCat 16. Sagawa M , Fujimura S, Togawa N, Yamamoto H, and Matsuura Y ( 1984 ) New material for permanent magnets on a base of Nd and Fe . J. Appl. Phys. 55 : 2083 – 2087 . Google Scholar Crossref Search ADS WorldCat 17. Li D , Suzuki S, Boundary G, Tsubokawa Y, Hirosawa S, Shimizu R, Takahashi M, and Miyazaki T ( 1987 ) Related content Nd – Fe – B permanent magnet materials . Jpn. J. Appl. Phys. 26 : 785 – 800 . Google Scholar OpenURL Placeholder Text WorldCat 18. Fidler J , and Knoch K G ( 1989 ) Electron microscopy of Nd-Fe-B based magnets . J. Magn. Magn. Mater. 80 : 48 – 56 . Google Scholar Crossref Search ADS WorldCat 19. Vial F , Joly F, Nevalainen E, Sagawa M, Hiraga K, and Park K T ( 2002 ) Improvement of coercivity of sintered NdFeB permanent magnets by heat treatment . J. Magn. Magn. Mater. 242–245 : 1329 – 1334 . Google Scholar Crossref Search ADS WorldCat 20. Li W F , Ohkubo T, Hono K, and Sagawa M ( 2009 ) The origin of coercivity decrease in fine grained Nd-Fe-B sintered magnets . J. Magn. Magn. Mater. 321 : 1100 – 1105 . Google Scholar Crossref Search ADS WorldCat 21. Janovec V , and Park Y ( 1984 ) Related content Lorentz electron microscope observation of domain walls in Fe – Nd – B alloy permanent magnets . Jpn. J. Appl. Phys. 23 : L412 – L423 . Google Scholar Crossref Search ADS WorldCat 22. Sepehri-Amin H , Ohkubo T, Shima T, and Hono K ( 2012 ) Grain boundary and interface chemistry of an Nd-Fe-B-based sintered magnet . Acta Mater. 60 : 819 – 830 . Google Scholar Crossref Search ADS WorldCat 23. Shibata N , Kohno Y, Findlay S D, Sawada H, Kondo Y, and Ikuhara Y ( 2010 ) New area detector for atomic-resolution scanning transmission electron microscopy . J. Electron Microsc. 59 : 473 – 479 . Google Scholar Crossref Search ADS WorldCat 24. Tsuda K , and Tanaka M ( 1999 ) Refinement of crystal structural parameters using two-dimensional energy-filtered CBED patterns . Acta Crystallogr. Sect. A Found. Crystallogr. 55 : 939 – 954 . Google Scholar Crossref Search ADS WorldCat 25. Vincent R , and Midgley P A ( 1994 ) Double conical beam-rocking system for measurement of integrated electron diffraction intensities . Ultramicroscopy 53 : 271 – 282 . Google Scholar Crossref Search ADS WorldCat 26. Nakamura A , Kohno Y, Sasaki H, and Shibata N ( 2017 ) Differential phase contrast imaging with reduced dynamical diffraction effect . Microsc. Microanal. 23 : 1412 – 1413 . Google Scholar Crossref Search ADS WorldCat 27. Close R , Chen Z, Shibata N, and Findlay S D ( 2015 ) Towards quantitative, atomic-resolution reconstruction of the electrostatic potential via differential phase contrast using electrons . Ultramicroscopy 159 : 124 – 137 . Google Scholar Crossref Search ADS PubMed WorldCat 28. Welch B L ( 1947 ) The generalization of `Student’s problem when several different population variances are involved . Biometrika 34 : 28 – 35 . Google Scholar PubMed OpenURL Placeholder Text WorldCat 29. Seki T , Sánchez-Santolino G, Ishikawa R, Findlay S D, Ikuhara Y, and Shibata N ( 2017 ) Quantitative electric field mapping in thin specimens using a segmented detector: Revisiting the transfer function for differential phase contrast . Ultramicroscopy 182 : 258 – 263 . Google Scholar Crossref Search ADS PubMed WorldCat 30. Seki T , Ikuhara Y, and Shibata N ( 2018 ) Theoretical framework of statistical noise in scanning transmission electron microscopy . Ultramicroscopy 193 : 118 – 125 . Google Scholar Crossref Search ADS PubMed WorldCat © The Author(s) 2020. Published by Oxford University Press on behalf of The Japanese Society of Microscopy. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)
TI - Magnetic-structure imaging in polycrystalline materials by specimen-tilt series averaged DPC STEM
JF - Microscopy
DO - 10.1093/jmicro/dfaa029
DA - 2020-10-30
UR - https://www.deepdyve.com/lp/oxford-university-press/magnetic-structure-imaging-in-polycrystalline-materials-by-specimen-hSy0VFi5Ow
SP - 312
EP - 320
VL - 69
IS - 5
DP - DeepDyve
ER -