TY - JOUR
AU - Motai,, Kumi
AB - Abstract A sample-heating system for spin-polarized scanning electron microscopy (spin SEM) has been developed and used for microscopic magnetization analysis at temperatures up to 500°C. In this system, a compact ceramic heater and a preheating operation keep the ultra-high vacuum conditions while the sample is heated during spin SEM measurement. Moreover, the secondary-electron collector, which is arranged close to the sample, was modified so that it is not damaged at high temperatures. The system was used to heat a Co(1000) single-crystal sample from room temperature up to 500°C, and the magnetic-domain structures were observed. Changes of the domain structures were observed around 220 and 400°C, and these changes are considered to be due to phase transitions of this sample. spin, magnetic domain, high temperature, SEM, secondary electrons, Co Introduction Spin-polarized scanning electron microscopy (spin SEM) [1–3] is one method for observing magnetic domains at the surface of a sample. It is based on the principle that the polarization of secondary electrons from a ferromagnetic sample is anti-parallel to the magnetization vector at the originating point of the secondary electrons [4]. This principle brings advantages such as high resolution, topography independency and the ability to analyze magnetization vectors in three dimensions. Several spin SEMs [5,6] have been equipped with a system for controlling sample temperature over a wide range, and these groups have reported observations of mainly fundamental scientific phenomena at low temperature. On the other hand, high-temperature observations have not been reported very often [5]. These days, however, magnetic characteristics of certain devices at high temperatures have gathered more attention. For instance, permanent magnets based on the neodymium–iron–boron (NdFeB) compound exhibit a high energy-product-to-volume ratio; however, to improve their characteristics, they have still been studied intensively. Recently, the application of this kind of magnet has extended to motors used in hybrid cars or battery vehicles, and it demands further improvement in the magnet's performance, especially at high temperatures. NdFeB magnets, however, are not very stable at high temperatures. In fact, their Currie temperature is not high, namely, below 350°C, while that of a ferrite magnet is around 500°C. Therefore, to improve the characteristics of a NdFeB magnet, it is helpful to microscopically observe how the magnetization changes at high temperatures. In another instance, the temperature dependence of the magnetic characteristics of the hard-disk drive medium has also gathered attention. The recorded bits become smaller and smaller as recording densities continue increasing. As a future high-density magnetic recording system, thermally assisted recording has been intensively studied [7]. In the case of this system, the coercivity of the recording medium is very large at room temperature, so a bit is stable against stray fields even though its size is very small. The magnetic flux from the recording head, therefore, does not change the magnetization of the medium at room temperature. On the other hand, the recording medium is heated to reduce its coercivity at the moment that bits are recorded. The temperature dependence of the magnetic characteristics of the medium is thus very important in regard to coercivity. The curie temperature of the candidate medium has been reported to be nearly 500°C [8]. As just described, spin SEM observation at high temperatures, such as at 500°C, is expected to play an important role in developing and improving the various magnetic devices mentioned above. However, spin SEM measurements at high temperatures face several difficulties. The spin SEM used in this study has a compact secondary-electron collector composed of many tiny electrodes [9,10]. The collector is arranged between the objective lens and the sample so that it collects almost all the secondary electrons and achieves high resolution [thanks to a short working distance (WD)]. If the sample is heated, the collector will be heated too, and some of its parts may be damaged. This damage will result in a reduced number of secondary electrons entering into the spin polarimeter and substantially decreased S/N ratio in the magnetic-domain images. Moreover, the vacuum conditions may be worsened by out-gas from the heated sample and its holder, even though a ultra-high vacuum (UHV) conditions (namely better than 5 × 10−7 Pa) are absolutely necessary for spin SEM. In spin SEM measurement, the sample surface must be clean because the probing depth is around 1 nm. In our system, the sample surface is cleaned by Ar-ion milling in the preparation chamber, and then the sample is transferred to the observation chamber, where the magnetic domain images are taken. If the vacuum conditions of the observation chamber become worse, the magnetic domain images are difficult to be obtained. To substantiate high-temperature measurement by spin SEM, it is thus necessary to overcome these difficulties. In this study, a new approach to achieving high-temperature observation by spin SEM is proposed. A magnetic-field application system will be installed in the sample stage; therefore, a sample-heating system that can cooperate with the system must be designed. To satisfy this design requirement, a sample holder, a stage and secondary-electron optics for high-temperature spin-SEM observations of domain structures, such as those of Co(0001), at up to 500°C were developed. Sample-heating system System structure The system involves a ceramic heater, made of pyrolytic boron nitride (size: 4-mm width; 20-mm length, and 1-mm thickness), at the top of a sample manipulator, where the sample holder is fixed (Fig. 1). Fig. 1 Open in new tabDownload slide (a) Sample stage of a spin SEM and (b) magnified image of the part where the sample holder is set, which is indicated by the square in (a). Fig. 1 Open in new tabDownload slide (a) Sample stage of a spin SEM and (b) magnified image of the part where the sample holder is set, which is indicated by the square in (a). The structure of the sample holder is shown in Fig. 2a. In addition to being used in the sample-heating system, the sample holder is also used with the magnetic-field application system; therefore, a space for the magnet is allocated just below the sample. When the sample is set on the manipulator, the ceramic heater is inserted into the sample holder to form the configuration shown in Fig. 2b. The heater is located at a distance of ∼10 mm from the center position of the magnet just below the sample to make room for the magnet, and the heat is transferred to the sample through a copper plate. In this configuration, the heat from the ceramic heater may influence the magnetism of the magnet; therefore, a heat-masking sheet made of stainless steel is positioned between the ceramic heater and the magnet. This configuration limits the area in which the heat from the ceramic heater is conducted. This limitation is required to minimize out-gas from the components in the spin SEM chamber and keep the UHV conditions. Fig. 2 Open in new tabDownload slide (a) Sample holder for sample-heating system and (b) schematic images of the configuration of the sample and heater in the sample holder. Fig. 2 Open in new tabDownload slide (a) Sample holder for sample-heating system and (b) schematic images of the configuration of the sample and heater in the sample holder. A thermometer was set below the sample holder. If the thermometer was set on the sample surface, it would interfere with the probe's electron beam (irradiating the sample) and with collection of secondary electrons by the collector. The relation between the temperatures on the sample surface and at the reference point below the sample holder was therefore examined in advance for calibration of the sample surface temperature. During the spin SEM observation, the temperature at the reference point is monitored, so the temperature on the sample surface is known. Preheating process We need a method for maintaining the UHV conditions in the observation chamber when the sample is heated up to 500°C. A preheating process was found to minimize out-gas during the observation by heating the sample in the spin SEM chamber in advance. The preheating process is supposed as follows. Before the spin-SEM observation, the sample is inserted in the observation chamber and then heated up to the setup temperature for the observation. At that temperature, the vacuum conditions degrade, but are improved after a certain number of hours (which depends on the materials and the size of the samples and the setup temperature). This preheating process at four elevated temperatures in the case of iron crystals with a size of 10 × 10 mm with a 1-mm thickness was investigated. The times required for the vacuum conditions to recover to 5 × 10−7 Pa for each temperature are summarized in Table 1. After the vacuum has been recovered, the heater is switched off, and the sample is cooled to room temperature. The sample is then transferred to the preparation chamber, and its surface is cleaned by argon-ion sputtering. After that, the sample is transferred to the observation chamber and heated to the setup temperature again. After the preheating process, the vacuum does not become worse than 5 × 10−7 Pa, so spin SEM images can be obtained. Table 1. Preheating time required to recover the vacuum conditions at each temperature Temperature (°C) . Worst vacuum pressure during the preheating process (Pa) . Required time to 5 × 10−7 Pa (h) . 100 1 × 10−6 2 200 2 × 10−6 3 300 4 × 10−6 4 500 6 × 10−6 6 Temperature (°C) . Worst vacuum pressure during the preheating process (Pa) . Required time to 5 × 10−7 Pa (h) . 100 1 × 10−6 2 200 2 × 10−6 3 300 4 × 10−6 4 500 6 × 10−6 6 The sample is an iron crystal with a surface area of 10 × 10 mm and a thickness of 1 mm. Open in new tab Table 1. Preheating time required to recover the vacuum conditions at each temperature Temperature (°C) . Worst vacuum pressure during the preheating process (Pa) . Required time to 5 × 10−7 Pa (h) . 100 1 × 10−6 2 200 2 × 10−6 3 300 4 × 10−6 4 500 6 × 10−6 6 Temperature (°C) . Worst vacuum pressure during the preheating process (Pa) . Required time to 5 × 10−7 Pa (h) . 100 1 × 10−6 2 200 2 × 10−6 3 300 4 × 10−6 4 500 6 × 10−6 6 The sample is an iron crystal with a surface area of 10 × 10 mm and a thickness of 1 mm. Open in new tab Heat-proofed secondary-electron optics The secondary-electron collector of the previously reported spin SEM is set close to the sample surface in order to collect as many secondary electrons as possible (Fig. 3). The ground plate, which is located in the lowest part of the secondary-electron collector, is contacted with the sample surface. The other electrodes, such as spherical deflectors and einzel lenses, are set above the earth plate and accelerate the secondary electrons up to several hundred volts. On the other hand, the WD, which is defined as the distance between the objective lens of the electron gun and the sample surface, should be short to achieve high-resolution measurement. The electrodes must, therefore, be tiny and set closely while keeping sufficient distance to avoid electrostatic discharges. In this optics system, each electrode is fixed rigidly on a solid insulator plate made of polyimide materials. When the samples are heated up to 500°C, all the electrodes and insulator plates are heated by thermal conduction through the earth plate and/or heat radiation. The insulator plate is heat-proofed up to 480°C under vacuum. The electrodes and the screws may, however, thermally expand, thereby loosening the rigid arrangement of the electrodes, changing the electric-field distribution, and, sometimes, causing electric discharge. It is, therefore, necessary to stop this heat transfer from the sample to the secondary electrodes. Fig. 3 Open in new tabDownload slide The heat-proofed secondary-electron optics for high-temperature spin SEM measurement. Fig. 3 Open in new tabDownload slide The heat-proofed secondary-electron optics for high-temperature spin SEM measurement. To satisfy that necessity, all the electrodes in the optics system were improved as follows. The ground plate (made of phosphor bronze) was replaced by a gold-coated ceramic one in consideration of the low heat conductivity of the ceramic. The ground plate was welded to the flange, made of cupronickel, which is stable even under high temperature (i.e. 500°C), and set on the base of the insulator. This configuration decreases the heat conductivity of the ground plate, thereby heat-proofing it up to 500°C. The structure and the appearance of the optics are shown schematically in Fig. 3. In addition to that improved ground plate, a heat-reflection plate (made of stainless steel coated with gold) was installed in contact with the ground plate in order to block the heat radiation from the sample to the secondary electrodes. It was confirmed by a simulation of electron trajectories that the heat-reflection plate does not decrease the collection ratio of secondary electrons. The above-mentioned secondary electrodes were designed and constructed, and it was confirmed that they function appropriately under high temperature (500°C). Estimation of drift effect during measurement Drift of the sample stage due to the thermal expansion of various parts of the stage is also a serious problem with measurements performed at high temperature. It takes ∼10–30 min to obtain one magnetic domain image by spin SEM. The drift of the sample stage, therefore, distorts the original shape of the magnetic domain, so it must be minimized, especially in the case of high-magnification images. A waiting time after the sample temperature reaches the setup point is required before the observation starts. This waiting time should be long enough to ensure that the drift of the sample stage becomes negligible. We originally set the acceptable amount of drift to 0.2 μm min–1. However, if the waiting time is too long, the sample surface will be contaminated even if it is in the UHV chamber. We, therefore, determined the appropriate waiting time as follows. First, we heated up the sample to the setup temperature (200°C), at which point the amount of drift was determined to be 20 μm min−1. It took 30 min to reduce the drift to 0.2 μm min−1. In the case of observation at 500°C, it took 120 min. SEM images of the surface [i.e. Co (0001)] of the sample at 500°C after a waiting time of 120 min are shown in Fig. 4. Image (a) was taken exposing the sample for 15 min (in the same manner as usual spin SEM). This image appears to include the effect of drift due to thermal expansion. Image (b) was taken at a 15-s exposure just after (a) was taken so that it would not be affected much by the drift. The amount of drift can be estimated by comparing both images. The dashed line is drawn between two points in (a): a small particle magnified in (c) and the point indicated by +. The same dashed line is drawn in (b) from the same particle as in (a), but the other edge of the segment does not fit the point shown indicated by + and has moved by ∼0.6 μm. This shows that the relative position between the small particle and the point indicated by + was changed because of the drift. In images (c) and (d), which are high-magnification images of the small particle shown in images (a) and (b), there is also a small difference – ∼0.1 μm – between the lengths of the particle. This amount of drift has not been significant up until today. Contamination on the sample surface during the waiting time was confirmed to be not serious (as described in the next section). Fig. 4 Open in new tabDownload slide Confirmation of the amount of the drift at 500°C after a waiting time of 120 min. (a) Sample surface topography of Co(0001) taken for 15 min and (b) the same area taken for 15 s just after taking (a). (c) and (d) are the magnified image of the part in (a) and (b), respectively. Fig. 4 Open in new tabDownload slide Confirmation of the amount of the drift at 500°C after a waiting time of 120 min. (a) Sample surface topography of Co(0001) taken for 15 min and (b) the same area taken for 15 s just after taking (a). (c) and (d) are the magnified image of the part in (a) and (b), respectively. These results indicate that the effect of the drift can be substantially decreased by setting the waiting time after the sample temperature reaches the setup (high) temperature. The drift can be further reduced by using an image-processing method such as phase correction [11] for higher magnification images. Results and discussion To confirm the performance of the sample-heating system, we observed the magnetic domain structure of Co (0001). The images were taken from room temperature up to 500°C in increments of 100°C (Fig. 5). As shown in the figures, two in-plane components of the sample magnetizations were detected simultaneously within 15 min. Fig. 5 Open in new tabDownload slide Spin SEM images of Co(0001) as functions of temperature from room temperatures to 500°C. Three images containing in-plane-magnetization components (X,Y) and vector-mapping color images are shown for each temperature. At room temperature, well-known closure-domain structures due to a hexagonal crystal structure are observed. Between 200 and 300°C, however, domain structures with sizes of 1–2 μm change into the structures with sizes larger than 10 μm. Between 400 and 500°C, small domains with sizes of 1–2 μm appear inside the large domain. Fig. 5 Open in new tabDownload slide Spin SEM images of Co(0001) as functions of temperature from room temperatures to 500°C. Three images containing in-plane-magnetization components (X,Y) and vector-mapping color images are shown for each temperature. At room temperature, well-known closure-domain structures due to a hexagonal crystal structure are observed. Between 200 and 300°C, however, domain structures with sizes of 1–2 μm change into the structures with sizes larger than 10 μm. Between 400 and 500°C, small domains with sizes of 1–2 μm appear inside the large domain. This sample had an easy direction of magnetization along the crystal axis of (0001) at room temperature, namely, the direction perpendicular to the observed sample plane. The magnetization, however, inclined in the in-plane direction due to its magneto-static energy, and closure domains were formed at the sample surface, as previously reported by the National Institute of Standards and Technology group [12]. In this study, we confirmed that the perpendicular component is not so large. The perpendicular components in Fig. 5 can therefore be omitted. The magnetization directions are indicated by colors corresponding to the color wheel at the bottom corner of the figure. At room temperature, small domains of 2–3 μm and various magnetization directions were produced from the hcp structure of a six-folded-symmetry crystal. As the temperature increased, the features of the domains did not change up to 200°C. The small domains disappeared and domains larger than 10 μm appeared at 300°C, and small (2–3 μm) domains appeared inside the large domains at 500°C. In other words, the domain structure changed conspicuously at two temperatures. We observed the two changes of magnetic-domain structure in more detail as the temperature was increased in 5°C increments. The observed spin SEM images for the transitions between 200 and 300°C are shown in Fig. 6. Below 210°C, the magnetic domains stayed small (2–3 μm) and the magnetization was orientated in various directions. The anisotropy of this Co (0001) sample at room temperature was directed along the c-axis, which was perpendicular to the observed sample plane, but the magnetizations formed closure domains at the sample surface due to the magneto-static energy. In addition, the magnetizations were orientated in various directions as a result of the six-folded symmetry of the crystal structure. Significant transitions of the magnetic-domain structure occurred from 210 to 240°C, where the small magnetic domains joined together to form larger domains. In this transition process, the contrast in the Y-component became weaker, which means that the magnetization was orientated in the X-direction. It seems that this transition was caused by the changes of the magnetic anisotropy of the sample, which are summarized in Fig. 7. Below 210°C, closure domains of 2–3 μm were formed on the sample surface. Several studies have demonstrated that the easy-magnetization axis changes from in-the-c-axis to in-the-c-plane direction at high temperatures [13–15]. Although not all of the reported transition temperatures are the same, they range from around 180 to 290°C, which is close to our results. The closure domains disappear above this temperature because the magnetization lies in the sample surface plane, so the domain sizes increase and the magnetizations align in the same direction. The magnetization directions are not parallel to the domain wall in the case at 300°C of Fig. 5, which is not preferable in terms of the magneto-static energy. The reason behind this domain formation is not yet clear. One possibility is that the magnetization at some points on the surface cannot rotate freely because of the surface strain, thus resulting in a frustrated domain structure at the surface. Fig. 6 Open in new tabDownload slide Spin SEM images of Co(0001) as functions of temperature from 210 to 240°C every 5°C. Small closure domains gather in a large one, and the magnetization aligns in the X-direction as temperature increases. Fig. 6 Open in new tabDownload slide Spin SEM images of Co(0001) as functions of temperature from 210 to 240°C every 5°C. Small closure domains gather in a large one, and the magnetization aligns in the X-direction as temperature increases. Fig. 7 Open in new tabDownload slide Presumed magnetic-domain structure of Co(0001) at (a) room temperature and (b) above 230°C. Fig. 7 Open in new tabDownload slide Presumed magnetic-domain structure of Co(0001) at (a) room temperature and (b) above 230°C. The magnetic domains from 390 to 400°C are shown in Fig. 8. Small domains of ∼1 μm were gradually formed inside the large domains. The reason for this transition is not yet clear. However, the crystal structure of the sample changed from hcp to fcc at 450°C [16], and we conclude that the magnetic domain changes observed in this experiment are related to that change. Fig. 8 Open in new tabDownload slide Spin SEM images of Co(0001) as functions of temperature from 390 to 400°C every 5°C. Small domains with sizes of several microns are created inside the large domains as temperature increases. Fig. 8 Open in new tabDownload slide Spin SEM images of Co(0001) as functions of temperature from 390 to 400°C every 5°C. Small domains with sizes of several microns are created inside the large domains as temperature increases. The results presented in this section show that the developed sample-heating system works as designed. It can image transformations of magnetic domains in Co(0001) at high temperature that are thought to be caused by the phase transitions previously reported [13–16]. Concluding remarks A sample-heating system was developed and installed in a spin SEM. Problems with high-temperature observations were investigated and prevented. The magnetic domains of Co(0001) were observed up to 500°C. This observation shows that this system is useful for studying the characteristics of magnetic materials at high temperature. In the future, this system will be combined with the magnetic-field application system in a spin SEM and used to study the change of magnetic domains and evaluate the coercivity or anisotropy of materials. References 1 Koike K , Hayakawa K . Scanning electron microscope observation of magnetic domains using spin-polarized secondary electrons , Jpn. J. Appl. Phys. , 1984 , vol. 23 (pg. L187 - L188 ) 10.1143/JJAP.23.L187 Google Scholar Crossref Search ADS WorldCat Crossref 2 Unguris J , Hembree G G , Celotta R J , Pierce D T . High resolution magnetic microstructure imaging using secondary electron spin polarization analysis in a scanning electron microscope , J. Microsc. , 1985 , vol. 139 (pg. RP1 - RP2 ) 10.1111/j.1365-2818.1985.tb02628.x Google Scholar Crossref Search ADS WorldCat Crossref 3 Oepen H P , Kirschner J . Imaging of magnetic microstructures at surfaces. The scanning electron microscope with spin polarization analysis , Scanning Microsc. , 1991 , vol. 5 (pg. 1 - 16 ) OpenURL Placeholder Text WorldCat 4 Chrobok G , Hofmann M . Electron spin polarization of secondary electrons ejected from magnetized europium oxide , Phys. Lett. , 1976 , vol. 57a (pg. 257 - 258 ) Google Scholar Crossref Search ADS WorldCat 5 Allenspach R . Ultrathin films: magnetism on the microscopic scale , J. Magn. Magn. Mater. , 1994 , vol. 129 (pg. 160 - 185 ) 10.1016/0304-8853(94)90108-2 Google Scholar Crossref Search ADS WorldCat Crossref 6 Konoto M , Kohashi T , Koike K , Arima T , Kaneko Y , Kimura T , Tokura Y . Direct imaging of temperature-dependent layered antiferromagnetism of a magnetic oxide , Phys. Rev. Lett. , 2004 , vol. 93 pg. 107201 10.1103/PhysRevLett.93.107201 Google Scholar Crossref Search ADS PubMed WorldCat Crossref 7 Saga H , Nemoto H , Sukeda H , Takahashi M . New recording method combining thermo-magnetic writing and flux detection , Jpn. J. Appl. Phys. , 1999 , vol. 38 (pg. 1839 - 1840 ) 10.1143/JJAP.38.1839 Google Scholar Crossref Search ADS WorldCat Crossref 8 Thiele J-U , Coffey K R , Toney M F , Hedstrom J A , Kellock A J . Temperature dependent magnetic properties of highly chemically ordered Fe55−xNixPt45L10 films , J. Appl. Phys. , 2002 , vol. 91 (pg. 6595 - 6600 ) 10.1063/1.1470254 Google Scholar Crossref Search ADS WorldCat Crossref 9 Matsuyama H , Koike K . Twenty-nm resolution spin-polarized scanning electron microscope , J. Electron Microsc. , 1994 , vol. 43 (pg. 157 - 163 ) OpenURL Placeholder Text WorldCat 10 Kohashi T , Motai K . Spin-polarized scanning electron microscopy for analysis of magnetic devices , AMTC Letters , 2008 , vol. 1 pg. 70 OpenURL Placeholder Text WorldCat 11 Tsuneta R , Koguchi M , Nakamura K , Nishida A . A specimen-drift-free EDX mapping system in a STEM for observing two-dimensional profiles of low dose elements in fine semiconductor devices , J. Electron Microsc. , 2002 , vol. 51 (pg. 167 - 171 ) 10.1093/jmicro/51.3.167 Google Scholar Crossref Search ADS WorldCat Crossref 12 Unguris J , Scheinfein M R , Celotta R J , Pierce D T . Magnetic microstructure of the (0001) surface of hcp cobalt , Appl. Phys. Lett. , 1989 , vol. 55 (pg. 2553 - 2555 ) 10.1063/1.101979 Google Scholar Crossref Search ADS WorldCat Crossref 13 Zener C . Classical theory of the temperature dependence of magnetic anisotropy energy , Phys. Rev. , 1954 , vol. 96 (pg. 1335 - 1337 ) 10.1103/PhysRev.96.1335 Google Scholar Crossref Search ADS WorldCat Crossref 14 Sucksmith W , Thompson J E . The magnetic anisotropy of cobalt , Proc. Roy. Soc Lond. A Mat. , 1954 , vol. 225 (pg. 362 - 375 ) 10.1098/rspa.1954.0209 Google Scholar Crossref Search ADS WorldCat Crossref 15 Pauthenet R , Barnier Y , Rimet G . Anisotropy and relaxation in metals , J. Phys. Soc. Jpn. , 1972 , vol. 17 Suppl. B1 (pg. 309 - 313 ) OpenURL Placeholder Text WorldCat 16 Lee B W , Alsenz R , Ignatiev A . Surface structures of the two allotropic phases of cobalt , Phys. Rev. B. , 1978 , vol. 17 (pg. 1510 - 1520 ) 10.1103/PhysRevB.17.1510 Google Scholar Crossref Search ADS WorldCat Crossref © The Author 2013. Published by Oxford University Press [on behalf of Japanese Society of Microscopy]. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com
TI - Sample heating system for spin-polarized scanning electron microscopy
JF - Microscopy
DO - 10.1093/jmicro/dfs132
DA - 2013-08-01
UR - https://www.deepdyve.com/lp/oxford-university-press/sample-heating-system-for-spin-polarized-scanning-electron-microscopy-rWjS1xSlmp
SP - 429
EP - 436
VL - 62
IS - 4
DP - DeepDyve
ER -