Hitachi Review Vol. 61 (2012), No.

6 264

Development of Spin SEM Technology for Observation of

Magnetic Domains

Teruo Kohashi, Dr. Sc. OVERVIEW: Electron microscope technology developed by Hitachi is
Kumi Motai, Dr. Sc. used for testing and analysis of a wide range of semiconductors and other
hardware devices. Spin SEMs are able to observe the magnetic structure
of a device at the nanometer scale and are already used for applications
such as evaluating the shape of storage bits on magnetic media. Hitachi
has succeeded in implementing two new observation techniques in order to
deal with the increasingly diverse range of device functions and evaluation
criteria, namely observation of structural changes in magnetic domains at
high temperatures (up to 500°C) and observation of samples with remanent
magnetization resulting in magnetic stray fields. It is anticipated that these
observation techniques will contribute to improvements in the performance
of magnetic devices, including playing a role in the development of powerful
new magnetic materials.

INTRODUCTION A common cause of magnetism is when the

BECAUSE high-resolution observation of magnetic electron spins inside a sample of magnetic material
domains allows device characteristics to be analyzed become aligned (polarized). It is known that, if an
at the micro level, it is an effective technique for the electron beam is applied to a sample and it causes
evaluation of magnetic devices as they continue to electrons from inside the sample to be emitted,
improve in performance. Electron spin is a physical those electrons will retain this spin polarization.
quantity, and the spin-polarized scanning electron Accordingly, if a tightly focused primary electron
microscope (spin SEM) is an instrument for the beam is applied to the sample and a detector is used to
observation of magnetic domains that works by measure the spin polarization of the emitted secondary
detecting and imaging the spin of electrons inside electrons, the magnetic orientation at the point from
ferromagnetic samples (see Fig. 1). which the secondary electrons were emitted can be
determined. By progressively scanning the primary
electron beam across the surface of the sample, an
Spin detector image of its magnetic domains can be obtained. The
Primary
electron beam key feature of this instrument is the spin detector for
measuring the electron spin polarization.
Spin-polarized It works by Mott scattering whereby spin-polarized
secondary Image of electrons incident on a thin metal foil result in an
electrons magnetic
domains
asymmetric pattern of scattering directions(1). This
Scan principle can be exploited to obtain a quantitative,
three-dimensional map of a material’s magnetic
orientations, with characteristics that include an
excellent ability to handle any sample shapes and
the ability to get an information about magnetization
Fig. 1—Diagram of Magnetic Domain Observation Using Spin separately from physical shape. Hitachi led the
SEM.
world in developing spin SEMs(1), and since then
By exposing a sample to a primary electron beam and detecting
the spin polarization of the emitted secondary electrons,
has made continued improvements to sensitivity and
the magnetic orientation at that point on the sample can be resolution(2),(3), utilizing the instrument to measure the
determined. This primary electron beam can then be scanned characteristics of a wide range of different magnetic
over the sample to build up an image of its magnetic domains. devices.
 Hitachi Review Vol. 61 (2012), No. 6 265

In the past, spin SEM measurement has not been impairment of the electron lensing apparatus located
well suited to observation of permanent magnets and just above the sample.
other materials with strong remanent magnetization, For observations of material with remanent
and examples of such use have been rare. This is magnetization, a magnetic shield is required that
because the magnetic stray field from such samples can block the magnetic stray field and reduce its
affects both the incident primary electron beam and influence on the primary and secondary electron
the emitted secondary electrons that constitute the beams. This mechanism was designed and developed
signal, degrading performance characteristics such as using a three-dimensional simulation of the electron
the resolution and signal-to-noise (S/N) ratio of the path. The magnetic shield was made from 0.5-mm
image. Similarly, experiments that involve raising a permalloy sheet cut to size so that it could be placed
sample to high temperature can have a strong influence like a shroud over the top of the sample holder, and cut
on things like vacuum integrity and electron lensing. with a slit 5 mm long and 1 mm wide. The shield was
Consequently, reports of such use in the literature are all then positioned so that the slit aligned with the region
but non-existent. In recent years, however, performance of sample surface to be observed. The probe electron
improvements and a greater diversity in the nature of beam passes through the slit to the sample surface,
magnetic devices have made these types of observations
more important. They are including observations of
changes in magnetic domains caused by temperature
Under-sample magnet 10 mm
and external magnetic fields, and also observations of
the structure of magnetic domains in materials with
Ceramic
remanent magnetization, such as permanent magnets. heater
Accordingly, we have developed new mechanisms Sample
that allow a spin SEM to be used for observations of Magnetic
shield
samples that have been raised to 500°C, and samples
Thermally
with a magnetic field of 80 kA/m at the material surface. insulating plate
These mechanisms are described here, together Slit in magnetic
shield
with example applications. (a)

SAMPLE STAGE INCORPORATING HEATING Magnetic shield

AND MAGNETIC STRAY FIELD SHIELDING
MECHANISMS Thermally
conductive
This section describes the design of a sample copper sheet
Sample
holder that incorporates the mechanisms referred to Ceramic
above. Sample heating is performed using a pyrolytic heater
boron nitride (PBN) ceramic heater. Fig. 2 shows Under-sample magnet
Temperature
how the different parts are positioned in the sample (b) monitoring point
holder. Fig. 2 (a) shows a plan view and (b) shows a
Secondary electrons Primary electron beam
lateral cross section. Although not described here, a
Magnetic shield Magnetic
mechanism for applying a magnetic field to the sample flux lines
was developed in parallel. As this involves locating
a magnet directly under the sample to provide the
Sample
magnetic field, the placement of the heater needed
to leave a gap of approximately 10 mm for this (c)
purpose. The heat from the heater is conveyed to the
sample via a thermally conductive copper sheet. As Fig. 2—Plan View (a), Lateral Cross Section (b), and Detail (c)
overheating the under-sample magnet would degrade of Sample Holder.
Diagrams (a) and (b) show the relative positions of the sample,
its magnetism, a thermally insulating plate (stainless
heater, under-sample magnet, and magnetic shield in the
steel) is positioned between it and the heater. This mechanism for sample heating and observing samples with
design succeeded in minimizing the locations that are remanent magnetization. Diagram (c) is an enlargement of the
raised to a high temperature and limited any negative gray box in diagram (b) and shows how the magnetic shield
effects of sample heating, such as loss of vacuum or blocks the magnetic stray field from the sample.
 Development of Spin SEM Technology for Observation of Magnetic Domains 266

and the secondary electrons are also captured by the It is known that the magnetic easy axis in the
detector through the slit. As the magnetic stray field (0001) orientation that exists in the Co singlecrystal
from the sample is blocked by the magnetic shield, at room temperature changes to a (0001) in-plane
as shown in Fig. 2 (c), its influence on the primary orientation at temperatures above the phase transition
electron beam and secondary electrons is significantly that occurs around 200°C. At temperatures below the
reduced. The mechanism was fabricated and then phase transition, therefore, the magnetization inside
tested on material with a magnetic stray field to the sample is oriented in the (0001) axis, whereas
demonstrate that the shielding functioned adequately. magnetostatic energy causes the structure on the
sample surface to be characterized by small closure
SAMPLE HEATING TESTS magnetic domains. At temperatures above the phase
Using the sample heating mechanism described transition, on the other hand, the (0001) in-plane
above, high-temperature observations were made of orientation (that is, an orientation along the surface)
the magnetic domains on the surface of a cobalt (Co) becomes an easy axis and therefore large stable
singlecrystal (0001). Fig. 3 shows the images captured domains oriented along in-surface axes form both in
at 100°C intervals from 100°C up to 500°C. the interior and on the surface of the sample. In this
While the sample has a magnetic easy axis (axis at experiment, it is assumed that this phase transition was
which the material is easily magnetized) perpendicular the cause of the major change in magnetic domains
to its surface [the (0001) axis], it is also known that that occurred between 200°C and 300°C. Similarly,
its structure is characterized by closure magnetic as Co singlecrystal changes from an hcp structure to
domains on its surface due to magnetostatic energy. an fcc structure at around 450°C, it is assumed that
As the observations show only a small component this was the reason for the second change in domain
perpendicular to the surface, the images in Fig. 3 were structure. Using the spin SEM in this way to observe
produced by color-coding the result of calculating the the magnetic domains of a Co (0001) surface allowed
magnetic orientations that lay on the sample surface. direct observations to be made of the magnetic domain
The 100°C image shows magnetic domains with a changes associated with phase transitions reported
size of 2 to 3 µm, reflecting the symmetry of the crystal in the literature, and therefore it demonstrated that
structure. While any change in domain shape as the the newly developed sample heating mechanism was
temperature increases remained comparatively small working correctly.
up to 200°C, the small domains had all disappeared The intention is to use this technology for the
by 300°C as the structure changed to one with observation of magnetic devices, the high-temperature
large magnetic domains of 10 µm or more. As the magnetic properties of which are a subject of interest.
temperature rose further from 400°C to 500°C, the In the case of the powerful neodymium-iron-boron
formation of small new domains with sizes of 2 to (NdFeB) magnets used in the electric motors that
3 µm was observed to occur inside the large domains. drive hybrid cars, for example, there is a need to assess
changes in magnetic domains at high temperatures in
order to make further improvements in the magnets’
high-temperature coercivity (resistance to becoming
demagnetized).
100°C 200°C 300°C 400°C 500°C
EXAMPLE OBSERVATIONS OF REMANENT
10 µm
MAGNETIZATION
This section describes observations of a permanent
magnet (a material with remanent magnetization)(4).
Fig. 3—Changes in Magnetic Domains with Rising Temperature The sample was a NdFeB magnet with fine-grained
in Co (0001). anisotropic crystals and the experiment, described
The images were color coded to indicate the magnetization below, studied the relationship between changes
vector (as indicated by the color circle on the bottom left).
in the magnetic domains and crystal grain shape.
Major changes in magnetic domain structure were evident
between 200°C and 300°C, and between 400°C and 500°C.
First, spin SEM images of the magnetic domains
These are believed to correspond to changes in the magnetic and surface shape were obtained for the magnet in a
anisotropy and in the crystal structure that occurred between thermally demagnetized state (in which it had zero
these temperatures. magnetization). The sample was then removed and
 Hitachi Review Vol. 61 (2012), No. 6 267

placed in a separate apparatus where it was exposed impurities visible in the shape image (b) were used as
to a magnetic field of 80 kA/m. After the magnetic landmarks to align the viewpoints for each observation.
field was turned off, leaving the material in a state of The position of a particular crystal grain at the bottom
remanent magnetization, the sample was then returned left of the shape image (b) is enclosed by a white border
to the spin SEM and further observations were made. and indicated by an arrow. The same crystal grain is
This procedure was then repeated with the applied also shown in the magnetic domain images to indicate
magnetic field increased by 80 kA/m each time until the relationship between the viewpoints in each image.
the magnetization became identical in the field of
view. The changes in the remanent magnetization of
the magnetic domains at each stage were observed.
Images (a) to (k) in Fig. 4 show some example
results. The images are oriented so that the magnetic
easy axes of the crystal grains point upwards. In
the thermally demagnetized state shown in image
(a) (b) (c)
(a), the bands of light and dark represent the
magnetic domains. The light regions indicate an
upward orientation, the dark regions a downward
orientation, and non-magnetic regions (corresponding
to impurities, etc.) appear gray. The areas of the light
and dark regions in this view are roughly equal. (d) (e) (f)
Image (b) shows the shape for the same area as
image (a). A crack can be seen running up the left side
of the image and an impurity with a size slightly less
than 1 µm is visible on the right. The granular structures
with sizes in the 0.3 to 0.5-µm range observable
elsewhere in the image as thin lines contrasted against (g) (h) (i)

the background correspond to the individual crystals

that make up the magnet. As the contrast is inadequate
in some places, not all the crystals can be resolved.
To investigate the relationship between the shape 1 µm
image and magnetic domain walls, image (c) was (j) (k)
generated by overlaying the magnetic domain walls
obtained from image (a) on image (b). This shows Fig. 4—Spin SEM Images of Sample at Varying Degrees of
some places where the magnetic domain walls Magnetization.
coincide with the crystal grain boundaries as well The images show the magnetic domains and shape respectively
as some places where they do not. As the magnetic at varying degrees of magnetization, from a thermally
domain walls tend to run in straight lines in order to demagnetized state up to remanent magnetization after exposure
minimize their area, they sometimes transect crystal of a field of 640 kA/m. Images (a) and (b) show the magnetic
domains and shape images respectively when the sample was
grains. Grains where this occurs have multiple
in a thermally demagnetized state, and image (c) shows the
magnetic domains, which means they have two magnetic domain walls overlaid on image (b). Images (d) to
different directions of magnetization within the same (k) show the magnetic domains for the sample in a state of
grain. As the spin SEM can be used in this way to remanent magnetization after exposure of increasingly strong
study shape and magnetization in parallel, it provides a magnetic fields, the respective field strengths being: 80 kA/m
way to analyze the relationship between crystal shapes (d), 160 kA/m (e), 240 kA/m (f), 320 kA/m (g), 400 kA/m (h),
and magnetic domain walls. 480 kA/m (i), 560 kA/m (j), and 640 kA/m (k). In the magnetic
Images (d) to (k) in Fig. 4 show the magnetic domain images, the light regions indicate an upward magnetic
orientation and the dark regions a downward magnetic
domains for the sample in a state of remanent
orientation. Although the viewpoint shifts slightly in the images
magnetization resulting from the successive exposure that show the sample in a state of remanent magnetization, the
of magnetic fields oriented in the upward direction. As position of a particular crystal grain is enclosed by a white
exposure of the magnetic fields required the sample border and highlighted by an arrow. Reprinted with permission
to be removed from the microscope, the crack and from the Journal of The Magnetics Society of Japan(4).
 Development of Spin SEM Technology for Observation of Magnetic Domains 268

Looking at the shapes of the magnetic domains, no CONCLUSIONS

major changes are evident between the thermally Nearly 30 years after the spin SEM was first
demagnetized sample in image (a) and the 240-kA/m developed, advances in this field continue to be
image (f), except for a gradual thinning of the dark made. This article has described an observation
regions and an expansion of the light regions. In the technique for investigating changes in magnetic
320-kA/m remanent magnetization image (g), the large domains. Meanwhile, further planned developments
dark magnetic domain visible in the earlier images has include work on achieving higher resolutions and on
shrunk and split in two, and this shrinkage becomes satisfying user demands such as the ability to make
rapidly more pronounced in images (h) through (j) (400 observations of the grain boundary regions in NdFeB
to 560 kA/m) until it is seen to disappear. magnets. Hitachi aims to continue contributing to
Through this process, the shapes of the magnetic improvements in magnetic device performance
domain walls can be observed changing from smooth through the use of microscopic analysis, while also
and linear in the thermally demagnetized image (a) to making enhancements to detector sensitivity and other
more complex and undulating lines in the images for aspects of microscope technology.
320 kA/m and higher. At around 0.3 to 0.5 µm, the
size of these undulations is similar to that of the crystal
grains, suggesting that the magnetic orientation is being REFERENCES
determined at the level of single grains. For example, (1) K. Koike et al., “Scanning Electron Microscope Observation
the crystal grain highlighted at the bottom-left of the of Magnetic Domains Using Spin-Polarized Secondary
magnetic domain images remains part of the large dark Electrons,” Japanese Journal of Applied Physics, Pt.2, Letters
domain from the thermally demagnetized image (a) up 23, L187–188 (Mar. 1984).
(2) H. Matsuyama et al., “Twenty-nm Resolution Spin-polarized
to the 240-kA/m remanent magnetization image (f). In
Scanning Electron Microscope,” Journal of Electron
the 320-kA/m image (g), this crystal grain still retains Microscopy 43, pp. 157–163 (Jun. 1994).
its magnetic orientation but the magnetization of part (3) T. Kohashi et al., “High-resolution Spin-polarized Scanning
of the surrounding area has flipped polarity, leaving this Electron Microscopy,” Journal of Electron Microscopy 59,
crystal grain protruding from the area of dark magnetic pp. 43–52 (Feb. 2010).
domain and causing the magnetic domain wall to take (4) T. Kohashi et al., “Analysis of Magnetization Mechanism
on an undulating shape. In the 400-kA/m image (h), for NdFeB Magnet Using Spin-Polarized Scanning Electron
the flipping of polarity in the surrounding area is more Microscopy (Spin SEM),” Journal of The Magnetics Society
of Japan 33, No. 4, pp. 374–378 (Sep. 2009).
advanced, and by the 480-kA/m image (i) the crystal
grain has become its own isolated magnetic domain.
Finally, in the 560-kA/m image (j), the magnetic
polarity of this crystal grain also flips. ABOUT THE AUTHORS
These results show how the retention and flipping
of magnetic polarity occur at the level of crystal Teruo Kohashi, Dr. Sc.
Joined Hitachi, Ltd. in 1991, and now works at the
grains and are different to the multi-domain structure
Advanced Measurement and Analysis Center, Central
suggested by the sample when it was in a thermally Research Laboratory. He is currently engaged in work
demagnetized state. Also flipping of the magnetization on spin-polarized electron optics and its application
polarity occurs at the level of individual crystal grains to the analysis of materials and devices. Dr. Kohashi
at different magnetic fields. The use of data such is a member of The Magnetics Society of Japan, The
Japan Institute of Metals, and The Japanese Society of
as these to provide pointers on how to improve the Microscopy.
coercivity of the magnet will be a topic for future work.
During this experiment, the remanent magnetization
of the sample was progressively increased until it Kumi Motai, Dr. Sc.
Joined Hitachi, Ltd. in 1994, and now works at the
finally reached more than 0.2 T. While there were Advanced Measurement and Analysis Center, Central
concerns about the magnetic stray field at the surface Research Laboratory. She is currently engaged in
of the sample, the measures described earlier meant work on surface analysis of materials and devices
that magnetic domain images could be obtained using auger electron spectroscopy and spin-polarized
scanning electron microscopy. Dr. Motai is a member
without magnetic stray fields influencing the results. of The Surface Science Society of Japan, The Physical
The research described in this article was conducted Society of Japan, and The Japan Society of Applied
jointly with the NEOMAX Company, Hitachi Metals, Ltd. Physics.