Materials Transactions, Vol. 53, No. 5 (2012) pp. 902 to 906
© 2012 The Japan Institute of Metals
Transmission Electron Microscopy of Twins in 10M Martensite
in Ni­Mn­Ga Ferromagnetic Shape Memory Alloy
Mitsuhiro Matsuda1,+1, Yoshihiro Yasumoto1,+2, Kimiaki Hashimoto1,+3,
Toru Hara2 and Minoru Nishida3
1
Department of Materials Science and Engineering, Kumamoto University, Kumamoto 860-8555, Japan
National Institute for Materials Science, Tsukuba 305-0047, Japan
3
Department of Engineering Science for Electronics and Materials, Kyushu University, Kasuga 816-8580, Japan
2
The combination of the 10M martensite plate variants in Ni­Mn­Ga ferromagnetic shape memory alloy was investigated by conventional
transmission electron microscopy (CTEM) and electron diffraction experiments. There were four plate variants commonly designated as A, B, C
Type II, A : C (B : D)
and D, and three fundamental plate combinations can be identified in a given plate group, namely A : B (C : D) of h5 51i
Type I and A : D (B : C) of {105} compound twins in the martensites. The boundary structure of these twins was also observed in the
of f12 5g
edge-on state by high resolution electron microscopy (HREM). The boundary of the Type I and the compound twins was not sharp in
comparison to that in other shape memory alloys. There were neither ledge nor step structures at the irrational boundary in the Type II twin.
[doi:10.2320/matertrans.M2012002]
(Received January 10, 2012; Accepted February 7, 2012; Published March 22, 2012)
Keywords: nickel­manganese­gallium alloy, ferromagnetic shape memory alloy, martensitic transformation, 10M martensite, variants
accommodate twin, conventional transmission electron microscopy (CTEM), high resolution electron microscopy (HREM)
1.
Introduction
Recently, the magnetically controlled shape memory
alloys, i.e., ferromagnetic shape memory alloys, have been
highlighted.1) It has been reported that Ni­Mn­Ga alloys
show a large change in the magnetic moment associated with
a thermoelastic martensitic transformation from the L21
2,
parent structure to 10M with the stacking sequence ð32Þ
2 or 2M with the
14M with the stacking sequence ð52Þ
stacking sequence (1)2 martensitic phases.2­6) The 10M and
the 14M martensites can be characterized by a five (ten)
layered modulated structure and a seven (fourteen) layered
modulated structure, respectively. The 2M martensite is a
non-modulated structure. The deformation of martensite in
ferromagnetic shape memory alloys is induced by the
rearrangement of variants due to the applied magnetic field
and/or stress. It has been reported that the rearrangement of
martensite variants by magnetic field occurs any temperature
for the 10M martensite, while it occurs only in a limited
temperature range for 14M martensite and does not occur
at any temperature range for the 2M martensite.7) Therefore,
the 10M martensite is considered to be mainly responsible for
the magnetic shape memory effect of the Ni­Mn­Ga alloys.
The boundary between martensite plate variants is composed
of the twinning as described below. Thus, twins in the
martensite play an important role of ferromagnetic shape
memory effect as well as those in thermally controlled shape
memory alloys. The characterization of twins and martensite
plate variants is essential to a further understanding of the
shape memory behavior.
The crystallography of twinning is described by either the
K1 (twinning) plane and ©2 direction (intersection of the K2
+1
Corresponding author, E-mail: [email protected]
Graduate Student, Kumamoto University. Present address: Daido Special
Steel Co. Ltd., Tokai 477-0035, Japan
+3
Graduate Student, Kumamoto University. Present address: Riken
Corporation, Kumagaya 360-8522, Japan
+2
plane and the plane of shear) or the K2 plane (another
undistorted plane) and ©1 direction (twinning shear direction).
Type I twinning has a rational K1 plane and ©2 direction.8)
Type II twinning has a rational K2 plane and ©1 direction.
All indices of the four elements are rational in compound
twinning. The two twin crystals in the Type I and the
compound twins are related by the mirror symmetry with
respect to the K1 plane. Those in the Type II twin are related
by a rotation of ³ around the ©1 axis. In order to determine
the twinning mode of Type I and compound twins directly
and conveniently by electron diffraction experiments, the
incident electron beam is required to be parallel to the K1
plane. The obtained pattern consists of two sets of reflections,
which are in mirror symmetry to each other with respect to
the K1 plane. Consequently, the K1 plane can be estimated
unequivocally from the mirror symmetry pattern. On the
other hand, the diffraction pattern obtained along ©1 direction
of Type II twin shows a single pattern.9­18) These incident
beam directions for each twinning mode are so-called the
edge-on state. The edge-on state is also required exactly to
analyze the twin interface structure on atomic scale by
HREM. It is apparent from the above descriptions that there
are plural electron beam directions in the edge-on state for
the Type I and the compound twinnings. On the other hand,
the edge-on state for the Type II twinning is the unique
axis of ©1. In other words, the overlap of two twin crystals
is unavoidable at the Type II twin interface in HREM
observations along the other axes because of an irrational K1
plane, thus clear information is not obtained.
It has been widely recognized that the lattice invariant
shear of martensite with layered modulated structure is
stacking faults on (001) basal plane.19­25) Therefore, internal
twins are boundaries of martensite plate variants, which are
so-called “variant accommodation twins”. The variant
accommodation twin is introduced as a result of mutual
accommodation of shear strains between variants in the
martensite. For 9R and/or 18R martensites, many researchers
Transmission Electron Microscopy of Twins in 10M Martensite in Ni­Mn­Ga Ferromagnetic Shape Memory Alloy
have reported that there are four plate variants commonly
designated as A, B, C and D around six {110} poles, and
three fundamental plate variant combinations can be
identified in a given plate variant; designated as A : B
(C : D), A : C (B : D) and A : D (B : C) types.20­23) The
intervariant boundaries of these three types are in Type II,
Type I and compound twin relations, respectively.22­25) These
are also classified to three morphological features, i.e., spear,
wedge and fork-like morphologies, respectively.20­23) These
fundamental combinations produce the self-accommodating
morphology with the diamond type in Cu-based alloys19­22)
and the parallelogram type in Ni­Al alloy.23)
The authors have reported on the variant combinations of
the 14M martensite in the Ni­Mn­Ga alloys.26) However,
there is no systematic investigation on those of the 10M
martensite in the Ni­Mn­Ga alloys as far as we know.
The purposes of the present study are to investigate
the twinning modes and variants configuration in the 10M
martensite in the Ni­Mn­Ga ferromagnetic shape memory
alloy and to characterize morphological and crystallographic
aspects of those twins by CTEM and electron diffraction
experiments. In addition, the atomic arrangement of each
twin boundary is also investigated by HREM in the edge-on
state.
2.
Experimental Procedure
The materials used were Ni­28.0 at% Mn­22.0 at% Ga
and Ni­28.5 at% Mn­21.5 at% Ga alloys which were prepared by arc melting in an argon atmosphere using pure
materials of 99.99 mass% Ga, 99.9 mass% Ni and
99.9 mass% Mn. The ingot was annealed at 1123 K for
86.4 ks for homogenization and then plate specimens of
0.2 mm in thickness were sliced by diamond saw. TEM
specimens of 3 mm in diameter were fabricated from the
plates by ultrasonic cutter and then annealed at 1123 K for
3.6 ks in a vacuum-sealed quartz tube. Subsequently, they
were annealed at 923 K for ordering followed by quenching
in iced water. Thin foil specimens are electropolished using
the twin jet method in an electrolyte of 30% nitric acid in
methanol at around 230 K. The transformation temperatures
were measured by differential scanning calorimetory (DSC).
The Curie temperature Tc of Ni­28.0 at% Mn­22.0 at% Ga
and Ni­28.5 at% Mn­21.5 at% Ga alloys is 371 and 369 K,
respectively. The Ms, Mf, As and Af temperatures of Ni­
28.0 at% Mn­22.0 at% Ga alloy were 304, 292, 309 and
321 K, respectively. Those of Ni­28.5 at% Mn­21.5 at% Ga
alloy were 328, 312, 328 and 335 K, respectively. The 10M
and the 14M martensites coexisted in both the alloys. The
10M martensite was dominantly observed in the Ni­
28.0 at% Mn­22.0 at% Ga alloy. On the other hand, the
14M martensite was dominantly observed in the Ni­
28.5 at% Mn­21.5 at% Ga alloy. From the DSC measurements, the martensitic transformation in both the alloys has
been completed at room temperature. However, the 10M
martensite is not stable in TEM specimens of the Ni­
28.0 at% Mn­22.0 at% Ga alloy and transforms to the parent
phase during the TEM observations. Therefore, the Ni­
28.5 at% Mn­21.5 at% Ga alloy was mainly used in the TEM
observations.
903
10µ
µm
Fig. 1 Optical micrograph of Ni­Mn­Ga alloy showing typical martensite
configuration. Spear and wedge-like morphologies are indicated by
arrows.
3.
Results and Discussion
3.1 Combination of the 10M martensite plate variants
Figure 1 shows the surface relief of the martensite in Ni­
28.0 at% Mn­22.0 at% Ga alloy at room temperature. There
are two kinds of surface relief. One is self-accommodating
morphology similar to the parallelogram type23) rather than
the diamond type.19­22) The other has rather flat and smooth
morphology, although the martensitic transformation is
completed at room temperature apparently from the DSC
measurements. Substructures of this area will be discussed
later. In the former type, we can clearly see spear and wedgelike morphologies as indicated by arrows. This observation
suggests that the fundamental plate variant combinations
mentioned above are probably completed in the 10M
martensite. Similar surface relief morphologies were observed in the Ni­28.5 at% Mn­21.5 at% Ga alloy.
The plate group configuration consisting of the 10M
martensite and its schematic illustration are shown in
Figs. 2(a) and 2(b), respectively. There are four plate variants
denoted as A, B, C and D in the illustration. The electron
diffraction patterns in Figs. 2(c), 2(d), 2(e) and 2(f ) are taken
from the variants A, B, C and D, respectively. The zone axis
of the plates A and C is the [210]10M and that of the plates B
10M . Although we do not present the
and D is the ½052
electron diffraction patterns taken from the A : C and B : D
boundaries, those patterns are in mirror symmetry with
plane. That is, the A : C and B : D
respect to the ð12 5Þ
10M Type I twin. In the same way, the
boundaries are the ð12 5Þ
A : B (C : D) and A : D (C : B) boundaries are recognized
10M Type II and {105}10M compound twins,
to be ½5 51
respectively. When the specimen is tilted about 35 degrees
axis, the zone axis of all the variant
around the ½12 5
10M
10M which is ©1 direction of
plates changes to the ½5 51
½55110M Type II twin. The single pattern is obtained from
the A : B (C : D) boundary as shown in Fig. 2(g), which is
characteristic of type II twin pattern along the ©1 direction.
Three fundamental combinations of four plate variants
are consistently identified. They are essentially the same
904
M. Matsuda, Y. Yasumoto, K. Hashimoto, T. Hara and M. Nishida
(a)
(b)
C
C
D
B
A
C
200nm
Variant A
(c)
Variant B
(d)
(g)
A : B (C : D)
105A. B
00 10A
125A
Variant C
00 10C
125A. B
125B
200B
EB // [551]A, B
EB // [052]B
EB // [210]A
(e)
000
000
000
Variant D
(f)
200D
000
000
125C
125D
EB // [210]C
EB // [052]D
Fig. 2 (a) Bright field image showing the plate group configuration of 10M martensite in Ni­Mn­Ga alloy. (b) Schematic illustration of
the plate group in (a). (c), (d), (e) and (f ) Electron diffraction patterns taken from the plates A, B, C and D, respectively. (g) Electron
axis.
diffraction pattern showing A : B (C : D) pair after tilting the specimen about 35 degrees around the ½12 5
10M
Table 1
K1
10M martensite
Type I
f12 5g
Type II
h5 51i
ð12 5Þ
ð0:5259 1 2:3710Þ
(105) Compound
(105)
Twinning elements in 10M martensite in Ni­Mn­Ga alloy.
©1
½1:0488 1 0:1902
½5 51
½501
configuration of martensite plate variants with layered
modulated structure in various alloys,22­25) except for the
A : D (B : C) boundary. Some of the A : D (B : C)
boundaries in the 10M martensite are not straight but
randomly curved as seen in (a) and (b). It is quite different
from the kink and fork-like morphologies which are generally
accepted as characteristic features of the A : D boundary as
mentioned above. The observed three twinning modes in the
10M martensite are listed in Table 1. The twinning elements
are calculated by the Bilby­Crocker theory.27) The lattice
parameters used are a = 0.422, b = 0.558, c = 2.103 nm and
¢ = 90.3°, which are determined from the present electron
diffraction experiments.
©2
K2
ð0:5259 1 2:3705Þ
ð125Þ
ð105Þ
S
½55 1
½1:0469 1 0:1910
0.1365
0.1365
[501]
0.0066
The other variant configuration only consisting of A : D
(B : C) type is frequently observed as shown in Fig. 3(a),
where the variant plates are very large over several µm and
the boundary is more largely curved. Such configuration
probably corresponds to the smooth surface region in the
optical micrograph in Fig. 1, since the twinning shear of
the {105}10M compound twin is quite small as presented in
Table 1.
3.2 Twin boundary structure
Three twinning modes of the variant plate combinations,
Type II, A : C (B : D) of
namely A : B (C : D) of h5 51i
f125g Type I and A : D (B : C) of {105} compound twins,
Transmission Electron Microscopy of Twins in 10M Martensite in Ni­Mn­Ga Ferromagnetic Shape Memory Alloy
(b)
(a)
105A,D
00 10A
10 5A,D
A
905
Variant A
Compound
000
00 10 D
000
00 10D
EB // [010]A,D
00 10A
K1=(105)A,D
D
200nm
Fig. 3 (a) Lattice image of an A : D boundary, i.e., (105)10M compound
twin. (b) Corresponding electron diffraction pattern taken from the
boundary.
(a)
Variant D
2nm
Variant A
125A,C
0010C
000 0010A
Fig. 5 Two-dimensional lattice image and corresponding diffraction
pattern of the (105) compound twin boundary taken along the [010]
direction.
Type I
K1=(125)A,C
Variant C
(b)
Variant A
5nm
125A,B
00 10A
000
125B
(125)A,B
Variant B
5nm
Fig. 4 Two-dimensional lattice images and corresponding diffraction
Type I (a) and h5 51i
Type II twin boundaries taken
patterns of the f12 5g
B , respectively.
along the [210]A // [210]C and [210]A // ½052
are confirmed as discussed above. In the present section, the
boundary structure of these twins is investigated by HREM.
Figures 4(a) and 4(b) show two-dimensional lattice images
Type I and h5 51i
Type II twin boundaries taken
of the f12 5g
B , respecalong the [210]A // [210]C and [210]A // ½052
tively. It is roughly true that the variants A and C are in
A;C plane in
mirror symmetry with respect to the K 1 ¼ ð12 5Þ
(a). However, the twin boundary is not sharp and it is very
difficult to identify the exact position of the K1 plane in
comparison to that in other shape memory alloys, although
the observation is performed in the edge-on state. In other
words, there is a transition region around the boundary where
the lattice fringes of the variants A and C cross each other.
This will be discussed later together with the (105)
compound twin boundary structure. In the A : B pair, it is
apparent from the corresponding electron diffraction pattern
inserted in the right upper corner in (b) that there are
2 and no
modulated structure with the stacking sequence ð32Þ
modulation in the variants A and B, respectively. Therefore,
Type II twin boundary lies along the
it is likely that the h5 51i
boundary between the modulated and the non-modulated
structures. The trace of the boundary is randomly and largely
curved, and blurred in place. This feature is probably due to
the irrational nature of the Type II twin boundary. The other
reason is that the observation is not completed in the edge-on
condition from the unique ©1 direction. That is, the variants A
and B are overlapped around the boundary.
Figure 5 shows two-dimensional lattice image of the (105)
compound twin boundary taken along the [010] direction.
The twin boundary is not sharp and the lattice fringes of the
variants A and D cross each other as same as those of the
Type I twin, although the
variants A and C in the f12 5g
observation is also performed in the edge-on state. In general,
the boundary of the compound twin was observed not to
be sharp plane in the martensite phase of various shape
memory alloys, since the twinning shear of the compound
twin is smaller than that of the Type I and Type II
twins.9,11,12,21­23,28) In the present alloy, the twinning shear
is very small not only in the {105} compound twin but also
Type I twin as listed in Table 1 in comparison to
in the f12 5g
the other systems such as Ti­Ni and Cu­Al­Ni. Therefore,
the twin boundary is not sharp in both the Type I and the
compound twins.
Finally, in order to understand the irrational nature of the
Type II twin boundary, we observe it in the edge-on
h5 51i
condition from the unique ©1 direction. Figure 6 shows two Type II twin boundary
dimensional lattice image of the h5 51i
taken along ½551 direction. The irrational nature of the
Type II twin boundary is still controversial with respect to the
presence of ledge and step structures.9­18) Neither ledge nor
906
M. Matsuda, Y. Yasumoto, K. Hashimoto, T. Hara and M. Nishida
125A,B
Variant A
000
105A,B
REFERENCES
(105)A,B
Type II
(125) A,B
Variant B
2nm
Fig. 6 Two-dimensional lattice image and corresponding diffraction
pattern of the h5 51i
Type II twin boundary taken along the ½5 51
direction.
step structures are recognized at the irrational boundary of
the 10M martensite in the Ni­28.5 at% Mn­21.5 at% Ga
alloy. This observation suggests that the strain around the
boundary is elastically relaxed by gradual displacement of
the atoms. The same result of the irrational nature of the
Type II twin boundary is obtained for the martensite in
various shape memory alloys by HREM observations from
the ©1 direction.12,14,26)
4.
University, for their kind supply of the Ni­Mn­Ga alloy and
useful discussion. This work was partly supported by the
Grant-in-Aid for Scientific Research (S) (19106013), from
Japan Society for the Promotion of Science, Japan.
Conclusions
The combination of the 10M martensite plate variants in
Ni­Mn­Ga ferromagnetic shape memory alloy was investigated by CTEM and electron diffraction experiments. There
were four plate variants commonly designated as A, B, C
and D, and three fundamental plate combinations can be
identified in a given plate group, namely A : B (C : D) of
Type II, A : C (B : D) of f12 5g
Type I and A : D
h5 51i
(B : C) of {105} compound twins in the martensites. The
boundary structure of these twins was also observed in the
edge-on state by HREM. The boundary of the Type I and the
compound twins was not sharp in comparison to that in other
shape memory alloys. There were neither ledge nor step
structures at the irrational boundary in the Type II twin.
Acknowledgements
The authors would like to express their sincere appreciation to Prof. T. Kakeshita and Prof. T. Fukuda of Osaka
1) K. Ullakko, J. K. Huang, C. Kantner, R. C. O’Handley and V. V.
Kokorin: Appl. Phys. Lett. 69 (1996) 1966­1968.
2) V. A. Chernenko, E. Cesari, V. V. Kokorin and I. N. Vitenko: Scr.
Mater. 33 (1995) 1239­1244.
3) F. Zuo, X. Su and K. H. Wu: Phys. Rev. B 58 (1998) 11127­11130.
4) B. Wedel, M. Suzuki, Y. Murakami, C. Wedel, T. Suzuki, D. Shindo
and K. Itagak: J. Alloy. Compd. 290 (1999) 137­143.
5) J. Pons, V. A. Chernenko, R. Santamarta and E. Cesari: Acta Mater. 48
(2000) 3027­3038.
6) J. Kim, F. Inaba, T. Fukuda and T. Kakeshita: Acta Mater. 54 (2006)
493­499.
7) N. Okamoto, T. Fukuda and T. Kakeshita: Mater. Sci. Eng. A 481­482
(2008) 306­309.
8) J. W. Christian: The Theory of Phase Transformations in Metals and
Alloys, (Second Edition, Oxford, 1975) p. 45.
9) K. M. Knowles and D. A. Smith: Acta Metall. 29 (1981) 101­110.
10) K. M. Knowles: Philos. Mag. A 45 (1982) 357­370.
11) T. Onda, Y. Bando, T. Ohba and K. Otsuka: Mater. Trans., JIM 33
(1992) 354­359.
12) T. Hara, T. Ohba, S. Miyazaki and K. Otsuka: Mater. Trans., JIM 33
(1992) 1105­1113.
13) M. Nishida, H. Ohgi, I. Itai, A. Chiba and K. Yamauchi: Acta Metall.
43 (1995) 1219­1227.
14) M. Nishida, K. Yamauchi, I. Itai, H. Ohgi and A. Chiba: Acta Metall.
43 (1995) 1229­1234.
15) M. Nishida, T. Hara, Y. Morizono, A. Ikeya, H. Kijima and A. Chiba:
Acta Mater. 45 (1997) 4847­4853.
16) M. Nishida and S. Ii: Mater. Sci. Forum 327­378 (2000) 103­110.
17) Y. Liu and Z. L. Xie: Acta Mater. 51 (2003) 5529­5543.
18) Z. L. Xie and Y. Liu: Philos. Mag. 84 (2004) 3497­3507.
19) H. Tas, L. Deleay and A. Deruyttere: Metall. Trans. A 4 (1973) 2833­
2840.
20) T. A. Schroeder and C. M. Wayman: Acta Metall. 25 (1977) 1375­
1391.
21) T. Saburi and C. M. Wayman: Acta Metall. 27 (1979) 979­995.
22) K. Adachi, J. Perkins and C. M. Wayman: Acta Metall. 34 (1986)
2471­2485.
23) Y. Murakami, K. Otsuka, S. Hanada and S. Watanabe: Mater. Sci. Eng.
A 189 (1994) 191­199.
24) S. Ii, T. Hara and M. Nishida: Scr. Mater. 46 (2002) 549­554.
25) S. Ii, M. Nishida, T. Hara and K. Enami: Mater. Trans. 43 (2002) 902­
907.
26) M. Nishida, T. Hara, M. Matsuda and S. Ii: Mater. Sci. Eng. A 481­482
(2008) 18­27.
27) B. A. Bilby and A. G. Crocker: Proc. R. Soc. Lond. A 288 (1965) 240­
255.
28) T. Hara, T. Ohba, K. Otsuka and M. Nishida: Mater. Trans., JIM 38
(1997) 277­284.