Electron differential microscopy
of magnetic thin films
T. Tanji, S. Manabe and K. Yamamoto
CIRSE, Nagoya University, Chikusa, Nagoya 464-8603, Japan
Electron holography has been successfully used in the quantitative observation of phase
objects such as magnetic fields and electrostatic potentials. Electron holograms are constructed by the
interference of a modulated object wave and a well-defined reference wave, and in most cases the
reference wave is required to be a plane wave. Frequently, however, we want to observe the area far
from a specimen edge, that is far from a vacuum area or a plain carbon film area. In such cases we
have to use the wave transmitting some area of the specimen as a reference wave. The reconstructed
wave horn the hologram made with a distorted reference wave does not accurately express the object
wave but the difference from the reference wave [l].
Differential microscopy is a useful technique when a well-defined reference wave is not
available. Some techniques using two-beam illumination for electron differential microscopy have
been proposed [2-41. We also have proposed a technique for differential microscopy by conventional
electron off-axis holography [5]. Shearing of the object wave is essential for differential microscopy.
In our experiment, the shearing was achieved by changing the potential applied to the prism. For
convenience, an electron trapezoidal prism [6] has been devised and applied to the differential
microscopy. The electron trapezoidal prism, which has two equi-potential filament-electrodes between
two grounded platelet-electrodes and has a trapezoidal electric potential distribution, is illustrated in
Fig. 1. An electron wave passing through between the two filament-electrodes, area II, travels
straight. Only an electron wave passing between each grounded electrode and the adjacent filament
electrode, area I or III, is tilted and interferes with one coming straight. Here, we use the former
wave traveling area II as the reference wave and the latter wave tilted in the area I or IIl as the
reference wave. If the two filaments were positioned closely enough, three waves might interfere, but
we set them about 0.3 mm and used only one side of the prism for two-wave interference. An
interference area was about 3 ,um. Only the object wave is shifted by changing the potential of the
trapezoidal prism, while the reference wave maintains its position. Making two holograms where
object waves are sheared each other by the trapezoidal prism and reconstructing the complex wave
function from each hologram, we calculated the difference between the two waves’ phases by
dividing one wave by the other. The effect of the distorted reference wave is included in the
differentiated phase no longer, because exactly the same components of reference waves in the two
reconstructed waves were compensated.
The magnetic fine structure of a permalloy thin film was observed with a Hitachi HF-2000
holographic TEM. Two holograms were recorded with a slow-scan CCD camera and reconstructed on
a personal computer. The difference between two images including interference fringes and
surrounding area is shown in Fig. 2. Applied potentials were 7.25 V and 7.50 V. It is seen clearly
that ‘theleft side of the interference region has subtracted and shows uniform contrast. This shows that
only the wave coming from the right side of the prism-filament (object wave) was sheared. Shifting
the prism potential by 0.25 V corresponds to the shear of 90 mrr in the specimen plane. Onedimensional differentiation gives us one component of the magnetic flux density projected along the
observation direction [7], so we need to rotate the specimen or the prism to obtain two-dimensional
density maps.
The results are shown in Fig. 3. The vectors in the two dimensional map of magnetic flux
density (a) are in the average direction of the top left. The plot of the vector angles from the horizontal
line (counter clockwise) along the line A-B (b) and its differential (c) show clearly the fluctuation in
the direction of magnetic vectors. This fluctuation, i.e. magnetic ripple, causes a weak and streaky
contrast in its Lore& micrograph (d). Maxima and minima in the intensity profile (e) along C-D well
correspond to those in the differential curve in (c).
In conclusion, the performance of the electron trapezoidal prism was confirmed in the
shearing of the object wave. This new prism was applied to electron differential microscopy, where
the effect of the distorted reference wave is removed. Electron differential microscopy was applied to
the observation of magnetic flux density in a thin permalloy film, and it was shown that the fluctuation
of magnetic vectors corresponds to the ripple contrast in the L0rent.z microscopy.
References
1. G. Maneucciet al., J. Appl. Phys. 69 (1991) 1835.
2. T. Leothner, H. Lichte and K. -H. Herrmann, Phy. .Stafas Solidi A 116 (1989) 113.
3. M. Mankos, M. R. Scheinfein and J. M. Cowley, J. A&. Whys. 75 (1994) 7418.
4. P. Kruit et al,, F’roc.of Microscopyrmd Microanalysis (1995) 606.
5. T. Tanji, Q. Ru and A. Tonomura, Appl. Phys. Lett. 69 (19%) 2623.
6. T. Tanji and S. Manabe, Microscopy md Micromdysis 3, Supl.2 (1997) 515.
7. G. Lai, T. Hiiayama, A. Fukuhara, K. Ishizuka, T. Tanji and A. Tonomura, J. Appl. Phys. 75 (1994) 4593.
q&&fjj
rv&Q
X
FIG. 1 Schematic diagmm of an electron
trapezoidal prism and its potential distribution.
A
FIG. 3
40
80
FIG. 2 Subtraction between two images including interference
fringes and surrounding area shows that the wave coming form the
left side of the prism was not sheared. Applied prism potentials
were 7.25 V and 7.50 V.
120 B
C
40
80
120 D
a: Two dimensional distribution of the magnetic flux density of a thin pamalloy film; b: plot of the vector angles
and c: its differential along A-B; d: corresponding Lorenti micrograph and d: intensity profile along C-D.