TY - JOUR
AU1 - Kohei, Tandokoro,
AU2 - Masayasu, Nagoshi,
AU3 - Takashi, Kawano,
AU4 - Kaoru, Sato,
AU5 - Katsushige, Tsuno,
AB - Abstract Scanning electron microscopy (SEM) is a powerful tool for observing the surface of materials. Modern SEM systems have multiple detectors with different geometries. Consequently, the SEM image contrast depends on the instrument and experimental conditions even for the same sample. Understanding the SEM imaging mechanism is necessary to clarify SEM contrast. In this paper, low-voltage (LV)-SEM image contrast is investigated by comparing LV-SEM images and electron trajectory simulation results. Surface observations of oxides on a steel surface, positive charging contrast and topographic contrast in the image systematically changed with the working distance (WD). The electron trajectory simulation revealed the sharing of emitted electrons by the in-lens and Everhart–Thornley detectors, and systematic changes in the electron sharing caused by changes in WD. The image contrast was reasonably explained by the kinetic energy and take-off angle (acceptance plots) of the detected electrons derived by the electron trajectory simulations. This approach is essential to understanding the SEM image contrast obtained by SEM systems with multiple detectors. Thorough image simulations based on acceptance plots are required in future work. scanning electron microscope, low-voltage SEM, contrast, information selection, trajectory simulation, acceptance plot Introduction Scanning electron microscopy (SEM) is a common technique for investigating material surfaces. Since the advent of the first commercial SEM instrument ~50 years ago [1], SEM technology has progressed remarkably. For example, the field-emission electron source and the in-lens or semi in-lens geometry has improved the spatial resolution drastically. A recent advance is low-voltage (LV)-SEM, with a primary electron energy typically lower than 1 keV. LV-SEM has the advantages of high surface sensitivity and visibility due to a smaller scattering volume or a shorter penetration depth of primary electrons. The development of LV-SEM began around 1985 [2], and the early studies are summarized in textbooks by Reimer [3,4]. The GEMINI system has a unique magnetic–electrostatic hybrid lens, a beam booster system, and an annular in-lens detector for obtaining a fine low-energy primary beam and detecting low-energy electrons efficiently [5]. Jaksch et al. [6,7] demonstrated the system’s performance by conducting high-resolution surface imaging using various materials. Previous studies also demonstrated that topographic and material contrasts were simultaneously extracted by using the two secondary electron detectors in the GEMINI system [8–10], the in-lens detector and the Everhart–Thornley (ET) detector, which provided material contrast and topographic contrast, respectively. This selection of the secondary electron’s information is possible because of electron sharing by the in-lens and ET detectors; the in-lens detector mainly detects electrons with lower kinetic energy. Kumagai et al. [11] demonstrated this behavior experimentally using an ultra-high-vacuum GEMINI system with an electron analyzer. They showed that almost all secondary electrons emitted with low kinetic energy were sucked into the column by the retarding electrostatic field of the GEMINI lens. Nagoshi et al. [12] showed that the contrasts and signal intensity of LV-SEM images taken by the in-lens and ET detectors systematically changed with the working distance (WD). Although the mechanisms of the information separation have gradually been understood, this understand has still been qualitative. To clarify the origin of SEM contrast, understanding the characteristics of the detected electrons is important. Forming a signal acceptance plot by simulating the trajectories of emitted electrons is useful for understanding SEM contrast. However, there are only a few studies [13,14] that describe the simulation of the emitted electron trajectory for specific SEM systems because electron trajectory simulations have been mainly used for optimizing primary electron beams. The lack of information about signal acceptance in the SEM is a big problem for SEM users in interpreting their images properly. In this paper, emitted electron trajectory simulations for a GEMINI column equipped with in-lens and ET detectors are described. Moreover, the LV-SEM image contrast of a steel surface is discussed by comparing LV-SEM images and acceptance plots from the trajectory simulation results. Materials and methods LV-SEM observations A field-emission SEM with a magnetic–electrostatic hybrid lens and booster system (LEO1530; LEO, now Carl Zeiss) was used in this study. Cold-rolled steel was used for observations using the in-lens and ET detectors with a primary electron energy of 500 eV. The main features of this steel surface include thermal etching steps and spherical Mn oxide particles containing B arising from the manufacturing process. WD was changed from 1 to 5.3 mm (maximum value for these experimental conditions). To avoid the effect of carbon contamination pile-up on the contrast in the SEM images, a fresh area was imaged each time. An ET detector image was recorded after recording the in-lens detector image because the former image is less sensitive to surface contamination. Electron trajectory simulation XENOS (Field Precision LLC) was used for calculating trajectories for emitted electrons from the sample surface. The three-dimensional geometry of the GEMINI column was estimated from the patent. The simulation was carried out for the WDs of 2, 4 and 6 mm. The primary beam energy was set to 500 eV. The electric and magnetic fields were determined as the primary electron focus on the specimen surface for each WD. The trajectory simulation was conducted for electrons emitted with kinetic energies of 0.1, 1, 5, 10, 50, 100 and 500 eV. The take-off angle was measured every 6° from 0° to 90° from the optical axis. Results SEM images of cold-rolled steel Figure 1 shows SEM images obtained by the in-lens and ET detectors with WDs of 1, 2, 3, 4 and 5.3 mm for the cold-rolled steel sheet. Oxide particles with diameters of 50–400 nm were visible on the steel surfaces, with grooving caused by thermal etching. The oxide particles had a strong dark contrast in the images obtained by the in-lens detector with WDs of 3, 4 and 5.3 nm. These images showed weak topographic information. For WDs <3 mm, the dark contrast decreased drastically and the topographic contrast dominated. Fig. 1. View largeDownload slide SEM images obtained by the (a) ET and (b) in-lens detectors with different WDs. WD ≥ 3 mm: topographic contrasts were extracted by the ET detector, whereas the oxide particles showed dark contrast for the in-lens detector. WD ≤ 2 mm: topographic contrast was mainly observed in the in-lens detector image and no signal intensity was detected by the ET detector. Fig. 1. View largeDownload slide SEM images obtained by the (a) ET and (b) in-lens detectors with different WDs. WD ≥ 3 mm: topographic contrasts were extracted by the ET detector, whereas the oxide particles showed dark contrast for the in-lens detector. WD ≤ 2 mm: topographic contrast was mainly observed in the in-lens detector image and no signal intensity was detected by the ET detector. For the ET detector with WDs of 3 to 5.3 mm, the topographic contrast was dominant; no signal intensity was obtained from the ET detector with WDs below 3 mm. The WD dependences on the SEM image contrast obtained by the in-lens and ET detectors were similar to that reported in a previous study [8], although the instrument (Supra 55-VP, Carl Zeiss Microscopy) and primary electron energy (1 keV) were different. The topographic contrast of the oxide particles was different between the in-lens detector images (WD: 1 and 2 mm) and ET detector images (WD: 4 and 5.3 mm). The in-lens detector images had no shadow effect because the in-lens detector looks down on the specimen, whereas the ET detector images showed shadows on the bottom right side of the particles (the opposite direction to the ET detector). This difference is explained by the positions of the detectors. SEM images in Fig. 1 were recorded with various WDs under fixed brightness and contrast values for each detector to investigate the changes in signal intensities as a function of WD. The signal intensities increased with the WD for ET detectors. This result is also similar to those reported previously [8]. To confirm the origin of the dark contrast in the in-lens image, the same area was observed repeatedly. Figure 2 shows the SEM images recorded by the first scan and the 20th scan for the identical surface region using the in-lens detector. As the electron beam scanning was repeated, the dark contrast of the oxide disappeared and the topographic contrast became dominant. This result suggests that contrast change is caused by the improvement of surface conductivity by carbon contamination. Thus, the dark contrasts in the in-lens images in Fig. 1 are caused by the positive charging effect due to the low electron conductivity of the oxides [10]. The positive charging usually causes surface potential of only a few electronvolts [4]. Emitted electrons with a kinetic energy lower than the surface potential are retracted by the potential and cannot escape from the oxide surface. Therefore, the dark contrast of the oxide particle remained as long as the poor electron conductivity was retained. Fig. 2. View largeDownload slide SEM contrast transformation obtained by multiple scanning. (a) First and (b) 20th scan (EHT = 500 V, WD = 4 mm). With repeated electron beam scanning, the dark contrast of the oxide disappeared. Fig. 2. View largeDownload slide SEM contrast transformation obtained by multiple scanning. (a) First and (b) 20th scan (EHT = 500 V, WD = 4 mm). With repeated electron beam scanning, the dark contrast of the oxide disappeared. Electron trajectory simulation for emitted electrons Figure 3 shows a cross-sectional schematic of the simulation model of the LEO1530 SEM system. The grid bias of the ET detector was set to +300 V. Figure 4 shows the calculated trajectories of electrons emitted with kinetic energies of 5 and 50 eV from specimens at WDs of 2, 4 and 6 mm. The side view of each trajectory was expressed as lines from the specimen surface to the termination points, such as the detectors, pole piece and chamber wall. Three-dimensionally spread trajectories were projected to a cross-sectional plane including the optical axis and the center of the ET detector. Fig. 3. View largeDownload slide Model constructed for the electron trajectory simulation. Fig. 3. View largeDownload slide Model constructed for the electron trajectory simulation. Fig. 4. View largeDownload slide Simulated electron trajectories with emitted electron energies of (a) 5 and (b) 50 eV. As the WD and energy increase, the emitted electrons tend to deviate from the optical axis and are detected by the ET detector. Fig. 4. View largeDownload slide Simulated electron trajectories with emitted electron energies of (a) 5 and (b) 50 eV. As the WD and energy increase, the emitted electrons tend to deviate from the optical axis and are detected by the ET detector. The emitted electrons exhibited the following features, as shown in the trajectory simulation in Fig. 4. All electrons with a kinetic energy of 5 eV were drawn into the column when the WD was set to 2 or 4 mm. Emitted electrons were detected by both the in-lens and the ET detectors for a WD of 6 mm. The electrons with a kinetic energy of 5 eV were drawn into the column and collimated to the optical axis more easily than those with a kinetic energy of 50 eV. This tendency was more pronounced for shorter WDs. The electrons with a kinetic energy of 50 eV deviated more from the optical axis and hit the wall of the SEM column or were detected by the ET detector. The deviation from the optical axis was large for longer WDs. These results can be explained by the effect of the electrostatic field of the objective lens between the pole piece and specimen. Low-energy electrons are easily attracted into the column by the electrostatic field of the objective lens. At shorter WD, higher-energy electrons are also drawn into the column. Figure 5 shows the acceptance plots for WDs of 2, 4 and 6 mm as obtained from the trajectory simulation results. The plots show emitted electrons on the plane that includes the optical axis and the center of the ET detector. The arrival points of the emitted electrons were categorized in the matrix of their original kinetic energy and take-off angle measured from the optical axis. Emitted electrons reaching the in-lens and ET detectors are denoted by open circles (○) and double circles (◎), respectively. Closed circles (●) denote emitted electrons going through the hole in the in-lens detector; crosses (×) denote those hitting the pole piece wall of the SEM chamber. An annular in-lens detector with a center hole of 5 mm in diameter was put on the optical axis at a height of 110 mm from the pole piece. Fig. 5. View largeDownload slide Acceptance plot generated by each trajectory simulation. Double circles (◎): reaching the ET detector; open circles (○): reaching the in-lens detector; closed circles (●): going through a hole in the in-lens detector; crosses (×): hitting the pole piece or wall of the SEM chamber. Detected electrons shift to a higher angle and lower energy as WD increases in each detector. Fig. 5. View largeDownload slide Acceptance plot generated by each trajectory simulation. Double circles (◎): reaching the ET detector; open circles (○): reaching the in-lens detector; closed circles (●): going through a hole in the in-lens detector; crosses (×): hitting the pole piece or wall of the SEM chamber. Detected electrons shift to a higher angle and lower energy as WD increases in each detector. The matrix clearly showed that emitted electrons with lower kinetic energy and lower take-off angle went through the hole in the in-lens detector and those with higher kinetic energy and high take-off angle did not go into the column. Consequently, electrons in a band-like region ranging from low energy and high angle to high energy and low angle were detected by the in-lens detector (○). However, the ET detector collects higher-energy and higher take-off angle electrons. These results clearly revealed that the two detectors detected electrons with different kinetic energy and take-off angle characteristics. Moreover, each detected region shifted in the lower kinetic energy and lower take-off angle direction when WD was increased. Thus, the detectable region of the in-lens detector also shifted to the lower-energy and lower-angle region, similar to the ET detector. Discussion In this section, the contrast formation mechanisms are discussed by comparing the SEM images (Fig. 1) and acceptance plots (Fig. 5) for WD ≥ 3 mm and WD ≤ 2 mm. The SEM images shown in Fig. 1 have two distinct characteristics depending on the WD. For WD ≥ 3 mm, the material contrast was extracted by the in-lens detectors, and the topographic contrast was observed by the ET detectors. For WD ≤ 2 mm, the topographic contrast was mainly observed by the in-lens detector image and no signal intensity was detected by the ET detector. Contrasts in SEM images with WD ≥ 3 mm In this WD region, Figs. 5(b) and (c) show that the low kinetic energy (less than several electronvolts) emitted electrons were collected by the in-lens detector. This result explains the dark material contrast that appears in the SEM images observed by the in-lens detector with longer WD. For the ET detector, the electrons with kinetic energy higher than a few tenths of an electronvolt and a high take-off angle were detected (Figs. 5(b) and (c)). High kinetic energy electrons were hardly affected by the positive charging with the potential of a few electronvolts. Moreover, the ET detector geometry produces the shadow effect in the SEM contrast, and the ET image formed by these electrons shows topographic contrast, as shown in Fig. 1. These simulations focus on the emitted electrons from the sample surface. However, SE3 and SE4 electrons are also important for understanding the ET image contrast. As shown in Fig. 4(b), SE3 electrons seem to be generated, although they are not considered in this paper for simplicity. In addition, the number of emitted electrons for a given pair of energy and angle is crucial for SEM imaging simulations. The contributions of SE3 and SE4 electrons and the number of emitted electrons should be examined more quantitatively in the future. Contrasts in SEM images with WD ≤ 2 mm Compared with WD ≥ 3 mm, the detectable electrons shifted to a high kinetic energy as WD decreased (Fig. 5). The result for WD = 2 mm indicates that electrons emitted with a kinetic energy below several electronvolts were not detected anymore by the in-lens detector and passed through the central hole of the detector. This can explain the suppression of the material contrast (due to positive charging) in the SEM images obtained by the in-lens detector with WDs of 1 and 2 mm. The acceptance plot for WD = 2 mm (Fig. 5(a)) shows that almost no electrons were detected by the ET detector. This is in good agreement with the SEM image (Fig. 1) showing no signal intensity for the ET detector at WDs of 1 and 2 mm. A short WD has the advantage of high spatial resolution due to a finely focused electron beam. However, to separate the material and topographic contrasts using the in-lens and ET detectors, WD longer than 3 mm was needed, as demonstrated in Figs 1 and 5. Therefore, the SEM observations and trajectory simulation indicated that a long WD is better for signal sharing. Generally, a high voltage and short WD are preferred for high-resolution observations. However, from these results, signal acceptance is as important as fine probe formation in the case of SEM. Consequently, there are some cases where WD control is more important for optimizing signal detection. The intensities of electrons caught by the in-lens and ET detectors can be summarized as follows: ET detector Short WD (2 mm): No electrons reach the ET detector, so the image is completely dark. Long WD (6 mm): Higher-energy electrons (>10 eV) reach the ET detector. The oxide particles and the matrix have the same brightness in the image contrast. In-lens detector Short WD (2 mm): Only higher-energy electrons (>10 eV) reach the in-lens detector. The oxide particles and matrix have a similar brightness in the image contrast. Long WD (6 mm): Only lower-energy electrons (<10 eV) reach the in-lens detector. The oxide particles are dark and the matrix is brighter in the image contrast. Concluding remarks LV-SEM image contrast was examined by comparing the SEM images and the trajectory simulations for emitted electrons. The SEM image contrast difference between the in-lens and ET detector was satisfactorily explained by the acceptance plot derived from the simulations. Moreover, the WD dependence of the SEM contrast was also clarified by the acceptance plot. The detection systems of the latest SEM instruments are complex. For example, recent high-end instruments have multiple in-lens detectors, energy filtering systems and stage bias. The approach demonstrated in this study of investigating the SEM contrast by comparing SEM images with acceptance plots for various experimental conditions will be required to understand the contrast in the SEM images obtained by these complex SEM systems. Our results are not limited to steel surfaces and can be applied to SEM images from a wide range of materials. This approach will also contribute to optimizing the detection geometry of SEM. References 1 McMullan D ( 1995 ) Scanning electron microscopy 1928–1965 . Scanning 17 : 175 – 185 . Google Scholar Crossref Search ADS 2 Pawley J ( 1984 ) Low voltage scanning electron microscopy . J. Microsc. 136 : 45 . Google Scholar Crossref Search ADS PubMed 3 Reimer L ( 1993 ) Image Formation in Low-Voltage Scanning Electron Microscopy , ( SPIE , Washington ). 4 Reimer L ( 1998 ) Scan Electron Microsc , 2nd ed , ( Springer-Verlag , Berlin, Heidelberg ). 5 Martin J P , Drexel V , Jaksch H , and Weimer E ( 1995 ) New scanning electron microscope opens up new horizons in materials sciences and biology . Zeiss Inf Jena Rev 4 ( 5 ): 14 . 6 Jaksch H , and Martin J P ( 1995 ) High-resolution, low-voltage SEM for true surface imaging and analysis . Fresenius J. Anal. Chem. 353 : 378 . Google Scholar Crossref Search ADS 7 Pohl D , and Jacksh H ( 1996 ) Advances in scanning electron microscopy . Prakt. METallogr. 33 : 235 . 8 Nagoshi M , Kawano T , and Sato K ( 2003 ) Practical material surface imaging by super-low-voltage scanning electron microscopy . J. Surf. Finishing Soc. Jpn. 54 : 31 . (in Japanese). Google Scholar Crossref Search ADS 9 Sato K , Nagoshi M , Kawano T , and Homma Y ( 2004 ) Ultra low voltage scanning electron microscopy . OYO BUTSURI (Applied Physics) 73 : 1325 . (in Japanese). 10 Nagoshi M , Kawano T , and Sato K ( 2006 ) Proc. Asia Steel International Conference 2006, 836 . 11 Kumagai K , and Sekiguchi T ( 2009 ) Sharing of secondary electrons by in-lens and out-lens detector in low-voltage scanning electron microscope equipped with immersion lens . Ultramicroscopy 109 : 368 . Google Scholar Crossref Search ADS PubMed 12 Nagoshi M , Kawano T , and Sato K ( 2016 ) Simple separations of topographic and material contrasts using one annular type in-lens detector of low-voltage SEM . Surf. Interf. Anal. 48 : 470 . Google Scholar Crossref Search ADS 13 Kazemian P , Mentink S A , Rodenburg C , and Humphreys C J ( 2007 ) Quantitative secondary electron energy filtering in a scanning electron microscope and its applications . Ultramicroscopy 107 : 140 . Google Scholar Crossref Search ADS PubMed 14 Müllerová I , and Konvalina I ( 2009 ) Collection of secondary electrons in scanning electron microscopes . J. Microsc. 236 : 203 . Google Scholar Crossref Search ADS PubMed © The Author(s) 2018. 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 - Low-voltage SEM contrasts of steel surface studied by observations and electron trajectory simulations for GEMINI lens system
JF - Microscopy
DO - 10.1093/jmicro/dfy030
DA - 2018-10-01
UR - https://www.deepdyve.com/lp/oxford-university-press/low-voltage-sem-contrasts-of-steel-surface-studied-by-observations-and-iL4Z6bA240
SP - 274
VL - 67
IS - 5
DP - DeepDyve
ER -