- Email: [email protected]
Structural properties of GaN grown on LiGaO2 by PLD
ARTICLE IN PRESS
Journal of Crystal Growth 259 (2003) 36–39
Structural properties of GaN grown on LiGaO2 by PLD H. Takahashia, H. Fujiokaa,b,*, J. Ohtaa, M. Oshimaa, M. Kimurac b
a Department of Applied Chemistry, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8656, Japan Kanagawa Academy of Science and Technology, KSP EAST 301, 3-2-1 Sakado, Takatsu-ku, Kawasaki 213-0012, Japan c Advanced Technology Research Laboratories, Nippon Steel Corporation, 20-1 Shintomi, Futtsu 293-8511, Japan
Received 8 May 2003; accepted 14 July 2003 Communicated by S. Hiyamizu
Abstract We have grown GaN films on LiGaO2 (LGO) (0 0 1) substrates by pulsed laser deposition (PLD) for the first time and investigated their structural properties using generalized grazing incidence-angle X-ray diffraction. By the use of PLD, we can grow high quality hexagonal GaN on LGO epitaxially without the use of any buffer layers, which is important to take advantage of the nearly lattice matched substrate. It has been shown that thin GaN films grown on LGO substrates are compressed in the lateral directions and that the lattice constant of GaN in the LGO [1 2 0] direction is slightly larger than that in the LGO [0 1 0] direction. This anisotropy is probably related to the crystal orientation dependence of the thermal contraction rate for LGO during the cooling down from the growth temperature. r 2003 Elsevier B.V. All rights reserved. PACS: 81.15.F; 61.10.N Keywords: A1. Stresses; A1. X-ray diffraction; A3. Laser epitaxy; B1. Nitrides
1. Introduction GaN and related nitrides have been believed to be the most promising materials for short wavelength optical devices such as blue light emitting diodes (LEDs) and laser diodes (LDs). GaN films have been usually grown on Al2O3 substrates in spite of the large lattice mismatch (approximately 16%) [1]. It is well known that the lattice mismatches between the epitaxial films and the substrates cause crystalline defects such as threading dislocations and degrade the device perfor*Corresponding author. Tel.: +81-3-5841-7192; fax: +81-35841-8744. E-mail address: [email protected] (H. Fujioka).
mance. To reduce the crystalline defect density, many researchers have made intensive efforts to utilize nearly lattice matched oxide substrates which include LiGaO2 (LGO). It is true that the crystal structure of LGO is orthorhombic but the symmetry of its (0 0 1) plane is close to that of wurtzite. However, remarkable improvement in the crystal quality of GaN films grown on LGO has not been reported so far. It is quite natural to believe that this problem stems from the nitridation reaction of the substrates just before the epitaxial growth because conventional growth techniques for GaN such as metalorganic chemical vapor deposition (MOCVD) [2,3] and molecular beam epitaxy (MBE) [4,5] use highly reactive nitrogen sources such as NH3 or an N2 plasma.
0022-0248/$ - see front matter r 2003 Elsevier B.V. All rights reserved. doi:10.1016/S0022-0248(03)01579-3
ARTICLE IN PRESS H. Takahashi et al. / Journal of Crystal Growth 259 (2003) 36–39
The unintentional nitridation of the substrate surfaces makes the advantage of the nearly lattice matched oxides futile. We have recently shown that the use of pulsed laser deposition (PLD) for the epitaxial growth of group III nitrides is advantageous over the conventional growth techniques in the reduction of the substrate nitridation reactions. [6–9]. Since the PLD growths proceed in a less reactive N2 ambient, we can grow GaN films on nearly lattice matched substrates which are vulnerable to chemical attacks. In this paper, we report on the epitaxial growth of GaN on LGO with PLD for the first time. We put special emphasis on the lattice distortion of GaN on LGO at the early stage of the film growths because the strains in the thin GaN films are closely related to the formation of defects such as threading dislocations. Although it has been difficult to measure the lattice distortions of thin epitaxial films precisely, we have recently shown that generalized grazing incidence-angle X-ray diffraction (G-GIXD) with the synchrotron radiation (SR) X-ray source is a powerful tool for this purpose. This technique allows us to analyze the lattice constants of materials with small volumes both in the lateral and the normal directions to the surface accurately [10–12]. In this paper, we also discuss the effect of the lattice contraction of the GaN films and the substrates during the cooling down from the growth temperature.
2. Experimental procedure GaN films were directly grown on LGO (0 0 1) substrate by PLD to take advantage of the nearly lattice matched substrate. After the surface cleaning using ethanol and acetone in an ultrasonic bath, the LGO substrate was introduced into the PLD chamber. The base pressure of the growth chamber was 2.0–4.0 109 Torr. During the growth, N2 gas (99.9999% purity) was introduced into the chamber through a variable leak valve and its pressure was kept at 1.0 101 Torr. The ablated species were ejected with high kinetic energies and deposited onto the substrates that were mounted 5 cm away from the target. The KrF
37
excimer laser (l ¼ 248 nm, t ¼ 20 ns) with an energy density of 3.0 J/cm2 was used at pulse repetition rate of 15 Hz. The substrate temperature was set at 700 C. After the deposition, we observed the surfaces of the films with reflection high-energy electron diffraction (RHEED) with a 25 keV electron gun and atomic force microscope (AFM) operated in the tapping mode. The experimental set-up for the G-GIXD measurements was already reported elsewhere [12]. A SR beam line, BL-3A at the Photon Factory in High Energy Accelerator Research Organization (KEK-PF) was used for the G-GIXD measurements. The wavelength of the ( The accuracy in the X-ray beam was set at 0.9 A. measured lattice constants with this wavelength is ( The incidence-angles of the better than 0.0005 A. X-ray beam were varied from 0.1 to 0.2 under the total reflection condition of GaN. The sample was mounted onto a 4-circle goniometer [10]. An imaging plate (IP) on a spherical-type goniometer [11] was used to detect diffraction peaks. The area detector covers a large area and it allows us to detect many diffraction peaks simultaneously. The reciprocal space mapping was taken by a scintillation counter placed on a spherical-type goniometer [12]. Between the sample and the detector, we placed two slits with an opening of 1 1 mm2.
3. Results and discussion The thickness of the samples used for the measurements was determined to be approximately 200 nm by a grazing incidence-angle X-ray reflectivity (GIXR) measurement. Fig. 1 shows a RHEED pattern from the GaN film grown on LGO (0 0 1) with the electron beam incident direction of GaN ½1 1 2% 0: This figure shows the sharp streaky pattern with spots, which indicates that we can grow high quality epitaxial GaN films with a surface roughness of several nanometers by PLD. The fact that epitaxial GaN films grow on LGO without using any buffer layers is important because it allows us to take advantage of the nearly lattice matched substrate. Assignment of these diffraction spots leaded us to conclude that hexagonal GaN (0 0 0 1) grows on these substrates
ARTICLE IN PRESS 38
H. Takahashi et al. / Journal of Crystal Growth 259 (2003) 36–39
Fig. 1. A RHEED pattern of GaN directly grown on LGO (0 0 1) by PLD. The incident direction of the electron beam is GaN ½1 1 2% 0:
with in-plane alignments of GaN ½1 1 2% 0//LGO ½0 1 0 and GaN ½1 01% 0//LGO [0 0 1]. The AFM image of GaN surfaces on LGO is shown in Fig. 2. Modulation of the surface with a periodicity of approximately 0.1 mm can be seen and the root mean square (RMS) value for the GaN surface morphology on the LGO substrate is 2.0 nm. This result is consistent with the streaky RHEED pattern with spots. G-GIXD measurements with the IP detector were performed to search the diffraction spots from the film. These measurements revealed that GaN films are hexagonal single crystals and they do not contain the other phases such as a cubic phase. Among various diffraction spots detected on the IP, we picked up ð1 0 1% 0Þ and ð1 0 1% 1Þ diffraction spots and carried out the reciprocal space mapping using the scintillation counter. Figs. 3(a) and (b) show the typical reciprocal space mappings for the ð1 0 1% 0Þ and ð1 0 1% 1Þ diffraction spots of GaN on LGO, respectively. The lattice parameters, a and c; were deduced from the experimental data using the Bragg’s law. Since the in-plane lattice constant of ( for LGO [0 1 0] and LGO shows anisotropy (6.372 A ( for LGO [1 0 0]) [13], we performed the in5.402 A plane reciprocal space mappings at every 60 . Table 1 summarizes the calculated lattice constants and shows that GaN layers are compressed in the lateral directions to the surface which can be possibly explained by the smaller in-plane lattice constants of the LGO (0 0 1). This table also shows that the lattice constant of GaN in the LGO [1 2 0] direction is
0.1m Fig. 2. An AFM image of the surface of GaN grown on LGO (0 0 1). Table 1 ( of the GaN films grown on LGO (0 0 1) Lattice constants (A) measured by G-GIXD. The bulk lattice constants are also shown GaN/LGO
Bulk GaN
a-axis
3.173 (LGO [0 1 0]) 3.176 (LGO [1 2 0])
3.185
c-axis
5.066
5.19
slightly larger than that in LGO [0 1 0] direction. This phenomenon, however, cannot be explained solely by the anisotropy in the lattice constants of LGO. The possible explanation for this phenomenon is the large thermal contraction rate coefficient of LGO in the [0 1 0] direction. Since the linear thermal expansion coefficient of LGO for the [0 1 0] direction (11.0 106/K) is much lager than that for the [1 2 0] direction (3.8 106/K) [14], GaN films contracted more in the [0 1 0] direction during the cooling down from the growth temperature. Table 1 also shows that the GaN film on LGO is compressed in the normal direction to the surface in spite of in the lateral compressions. It is obvious that this phenomenon cannot be explained by the simple elastic theory and is possibly related to the existence of the point defects and their preferential
ARTICLE IN PRESS H. Takahashi et al. / Journal of Crystal Growth 259 (2003) 36–39
28x10
39
-3
26
0.64
22
cq ⊥ / 2π
cq ⊥ / 2π
24
20
0.63
18 0.62
16 14
0.61 1.87 (a)
1.88
1.89
1.89
1.90
aq// / 2π
(b)
1.90
1.91
1.92
aq// / 2π
Fig. 3. Typical reciprocal space mappings for (a) ð1 0 1% 0Þ and (b) ð1 0 1% 1Þ GaN diffraction spots on LGO (0 0 1) measured with the scintillation counter.
piling-up [15]. Further study is necessary to clarify this phenomenon.
4. Conclusions PLD makes it possible to grow high quality GaN films on LGO epitaxially without the use of any buffer layers, which is important to take advantage of the nearly lattice matched substrate. G-GIXD measurements with the SR X-ray source have revealed that thin GaN films grown on LGO substrates are compressed in the lateral directions and that the lattice constant of GaN in the LGO [1 2 0] direction is slightly larger than that in LGO [0 1 0] direction. This anisotropy is probably related to the crystal orientation dependence of the thermal contraction rate for LGO during the cooling down from the growth temperature.
References [1] S.D. Lester, F.A. Ponce, M.G. Craford, D.A. Steigerwald, Appl. Phys. Lett. 66 (1995) 1249.
[2] S. Nakamura, M. Senoh, N. Iwasa, S. Nagahama, Appl. Phys. Lett. 67 (1995) 1868. [3] S. Nakamura, Mater. Sci. Eng. B43 (1997) 258. [4] S. Yoshida, S. Misawa, S. Gonda, Appl. Phys. Lett. 42 (1983) 427. [5] K. Kushi, H. Sasamoto, D. Sugihara, S. Nakamura, A. Kikuchi, K. Kishino, Mater. Sci. Eng. B59 (1999) 65. [6] J. Ohta, H. Fujioka, M. Sumiya, H. Koinuma, M. Oshima, J. Crystal Growth 225 (2001) 73. [7] J. Ohta, H. Fujioka, H. Takahashi, M. Sumiya, M. Oshima, J. Crystal Growth 233 (2001) 779. [8] H. Fujioka, J. Ohta, H. Katada, T. Ikeda, Y. Noguchi, M. Oshima, J. Crystal Growth 229 (2001) 137. [9] J. Ohta, H. Fujioka, H. Takahashi, M. Oshima, Phys. Stat. Sol. A 188 (2001) 497. [10] K. Kawasaki, Y. Takagi, N. Nose, H. Morikawa, S. Yamazaki, T. Kikuchi, S. Sasaki, Rev. Sci. Instr. 63 (1992) 1023. [11] T. Uragami, H. Fujioka, I. Waki, T. Mano, K. Ono, M. Oshima, Y. Takagi, M. Kimura, T. Suzuki, Jpn. J. Appl. Phys. 39 (2000) 4483. [12] T. Uragami, A.S. Acosta, H. Fujioka, T. Mano, J. Ohta, H. Ofuchi, M. Oshima, Y. Takagi, M. Kimura, T. Suzuki, J. Crystal Growth 234 (2002) 197. [13] M. Marezio, Acta Crystallogr. 18 (1965) 481. [14] T. Ishii, Y. Tazoh, S. Miyazawa, J. Crystal Growth 189/ 190 (1998) 208. [15] J. Kruger, . G.S. Sudhir, D. Corlatan, Y. Cho, Y. Kim, R. Klockenbrink, S. Ruvimov, Z. Lilienthal-Weber, C. Kisielowski, M. Rubin, E.R. Weber, B. McDermott, R. Pittman, E.R. Gertner, Mater. Res. Soc. Symp. 482 (1998) 447.
Journal of Crystal Growth 259 (2003) 36–39
Structural properties of GaN grown on LiGaO2 by PLD H. Takahashia, H. Fujiokaa,b,*, J. Ohtaa, M. Oshimaa, M. Kimurac b
a Department of Applied Chemistry, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8656, Japan Kanagawa Academy of Science and Technology, KSP EAST 301, 3-2-1 Sakado, Takatsu-ku, Kawasaki 213-0012, Japan c Advanced Technology Research Laboratories, Nippon Steel Corporation, 20-1 Shintomi, Futtsu 293-8511, Japan
Received 8 May 2003; accepted 14 July 2003 Communicated by S. Hiyamizu
Abstract We have grown GaN films on LiGaO2 (LGO) (0 0 1) substrates by pulsed laser deposition (PLD) for the first time and investigated their structural properties using generalized grazing incidence-angle X-ray diffraction. By the use of PLD, we can grow high quality hexagonal GaN on LGO epitaxially without the use of any buffer layers, which is important to take advantage of the nearly lattice matched substrate. It has been shown that thin GaN films grown on LGO substrates are compressed in the lateral directions and that the lattice constant of GaN in the LGO [1 2 0] direction is slightly larger than that in the LGO [0 1 0] direction. This anisotropy is probably related to the crystal orientation dependence of the thermal contraction rate for LGO during the cooling down from the growth temperature. r 2003 Elsevier B.V. All rights reserved. PACS: 81.15.F; 61.10.N Keywords: A1. Stresses; A1. X-ray diffraction; A3. Laser epitaxy; B1. Nitrides
1. Introduction GaN and related nitrides have been believed to be the most promising materials for short wavelength optical devices such as blue light emitting diodes (LEDs) and laser diodes (LDs). GaN films have been usually grown on Al2O3 substrates in spite of the large lattice mismatch (approximately 16%) [1]. It is well known that the lattice mismatches between the epitaxial films and the substrates cause crystalline defects such as threading dislocations and degrade the device perfor*Corresponding author. Tel.: +81-3-5841-7192; fax: +81-35841-8744. E-mail address: [email protected] (H. Fujioka).
mance. To reduce the crystalline defect density, many researchers have made intensive efforts to utilize nearly lattice matched oxide substrates which include LiGaO2 (LGO). It is true that the crystal structure of LGO is orthorhombic but the symmetry of its (0 0 1) plane is close to that of wurtzite. However, remarkable improvement in the crystal quality of GaN films grown on LGO has not been reported so far. It is quite natural to believe that this problem stems from the nitridation reaction of the substrates just before the epitaxial growth because conventional growth techniques for GaN such as metalorganic chemical vapor deposition (MOCVD) [2,3] and molecular beam epitaxy (MBE) [4,5] use highly reactive nitrogen sources such as NH3 or an N2 plasma.
0022-0248/$ - see front matter r 2003 Elsevier B.V. All rights reserved. doi:10.1016/S0022-0248(03)01579-3
ARTICLE IN PRESS H. Takahashi et al. / Journal of Crystal Growth 259 (2003) 36–39
The unintentional nitridation of the substrate surfaces makes the advantage of the nearly lattice matched oxides futile. We have recently shown that the use of pulsed laser deposition (PLD) for the epitaxial growth of group III nitrides is advantageous over the conventional growth techniques in the reduction of the substrate nitridation reactions. [6–9]. Since the PLD growths proceed in a less reactive N2 ambient, we can grow GaN films on nearly lattice matched substrates which are vulnerable to chemical attacks. In this paper, we report on the epitaxial growth of GaN on LGO with PLD for the first time. We put special emphasis on the lattice distortion of GaN on LGO at the early stage of the film growths because the strains in the thin GaN films are closely related to the formation of defects such as threading dislocations. Although it has been difficult to measure the lattice distortions of thin epitaxial films precisely, we have recently shown that generalized grazing incidence-angle X-ray diffraction (G-GIXD) with the synchrotron radiation (SR) X-ray source is a powerful tool for this purpose. This technique allows us to analyze the lattice constants of materials with small volumes both in the lateral and the normal directions to the surface accurately [10–12]. In this paper, we also discuss the effect of the lattice contraction of the GaN films and the substrates during the cooling down from the growth temperature.
2. Experimental procedure GaN films were directly grown on LGO (0 0 1) substrate by PLD to take advantage of the nearly lattice matched substrate. After the surface cleaning using ethanol and acetone in an ultrasonic bath, the LGO substrate was introduced into the PLD chamber. The base pressure of the growth chamber was 2.0–4.0 109 Torr. During the growth, N2 gas (99.9999% purity) was introduced into the chamber through a variable leak valve and its pressure was kept at 1.0 101 Torr. The ablated species were ejected with high kinetic energies and deposited onto the substrates that were mounted 5 cm away from the target. The KrF
37
excimer laser (l ¼ 248 nm, t ¼ 20 ns) with an energy density of 3.0 J/cm2 was used at pulse repetition rate of 15 Hz. The substrate temperature was set at 700 C. After the deposition, we observed the surfaces of the films with reflection high-energy electron diffraction (RHEED) with a 25 keV electron gun and atomic force microscope (AFM) operated in the tapping mode. The experimental set-up for the G-GIXD measurements was already reported elsewhere [12]. A SR beam line, BL-3A at the Photon Factory in High Energy Accelerator Research Organization (KEK-PF) was used for the G-GIXD measurements. The wavelength of the ( The accuracy in the X-ray beam was set at 0.9 A. measured lattice constants with this wavelength is ( The incidence-angles of the better than 0.0005 A. X-ray beam were varied from 0.1 to 0.2 under the total reflection condition of GaN. The sample was mounted onto a 4-circle goniometer [10]. An imaging plate (IP) on a spherical-type goniometer [11] was used to detect diffraction peaks. The area detector covers a large area and it allows us to detect many diffraction peaks simultaneously. The reciprocal space mapping was taken by a scintillation counter placed on a spherical-type goniometer [12]. Between the sample and the detector, we placed two slits with an opening of 1 1 mm2.
3. Results and discussion The thickness of the samples used for the measurements was determined to be approximately 200 nm by a grazing incidence-angle X-ray reflectivity (GIXR) measurement. Fig. 1 shows a RHEED pattern from the GaN film grown on LGO (0 0 1) with the electron beam incident direction of GaN ½1 1 2% 0: This figure shows the sharp streaky pattern with spots, which indicates that we can grow high quality epitaxial GaN films with a surface roughness of several nanometers by PLD. The fact that epitaxial GaN films grow on LGO without using any buffer layers is important because it allows us to take advantage of the nearly lattice matched substrate. Assignment of these diffraction spots leaded us to conclude that hexagonal GaN (0 0 0 1) grows on these substrates
ARTICLE IN PRESS 38
H. Takahashi et al. / Journal of Crystal Growth 259 (2003) 36–39
Fig. 1. A RHEED pattern of GaN directly grown on LGO (0 0 1) by PLD. The incident direction of the electron beam is GaN ½1 1 2% 0:
with in-plane alignments of GaN ½1 1 2% 0//LGO ½0 1 0 and GaN ½1 01% 0//LGO [0 0 1]. The AFM image of GaN surfaces on LGO is shown in Fig. 2. Modulation of the surface with a periodicity of approximately 0.1 mm can be seen and the root mean square (RMS) value for the GaN surface morphology on the LGO substrate is 2.0 nm. This result is consistent with the streaky RHEED pattern with spots. G-GIXD measurements with the IP detector were performed to search the diffraction spots from the film. These measurements revealed that GaN films are hexagonal single crystals and they do not contain the other phases such as a cubic phase. Among various diffraction spots detected on the IP, we picked up ð1 0 1% 0Þ and ð1 0 1% 1Þ diffraction spots and carried out the reciprocal space mapping using the scintillation counter. Figs. 3(a) and (b) show the typical reciprocal space mappings for the ð1 0 1% 0Þ and ð1 0 1% 1Þ diffraction spots of GaN on LGO, respectively. The lattice parameters, a and c; were deduced from the experimental data using the Bragg’s law. Since the in-plane lattice constant of ( for LGO [0 1 0] and LGO shows anisotropy (6.372 A ( for LGO [1 0 0]) [13], we performed the in5.402 A plane reciprocal space mappings at every 60 . Table 1 summarizes the calculated lattice constants and shows that GaN layers are compressed in the lateral directions to the surface which can be possibly explained by the smaller in-plane lattice constants of the LGO (0 0 1). This table also shows that the lattice constant of GaN in the LGO [1 2 0] direction is
0.1m Fig. 2. An AFM image of the surface of GaN grown on LGO (0 0 1). Table 1 ( of the GaN films grown on LGO (0 0 1) Lattice constants (A) measured by G-GIXD. The bulk lattice constants are also shown GaN/LGO
Bulk GaN
a-axis
3.173 (LGO [0 1 0]) 3.176 (LGO [1 2 0])
3.185
c-axis
5.066
5.19
slightly larger than that in LGO [0 1 0] direction. This phenomenon, however, cannot be explained solely by the anisotropy in the lattice constants of LGO. The possible explanation for this phenomenon is the large thermal contraction rate coefficient of LGO in the [0 1 0] direction. Since the linear thermal expansion coefficient of LGO for the [0 1 0] direction (11.0 106/K) is much lager than that for the [1 2 0] direction (3.8 106/K) [14], GaN films contracted more in the [0 1 0] direction during the cooling down from the growth temperature. Table 1 also shows that the GaN film on LGO is compressed in the normal direction to the surface in spite of in the lateral compressions. It is obvious that this phenomenon cannot be explained by the simple elastic theory and is possibly related to the existence of the point defects and their preferential
ARTICLE IN PRESS H. Takahashi et al. / Journal of Crystal Growth 259 (2003) 36–39
28x10
39
-3
26
0.64
22
cq ⊥ / 2π
cq ⊥ / 2π
24
20
0.63
18 0.62
16 14
0.61 1.87 (a)
1.88
1.89
1.89
1.90
aq// / 2π
(b)
1.90
1.91
1.92
aq// / 2π
Fig. 3. Typical reciprocal space mappings for (a) ð1 0 1% 0Þ and (b) ð1 0 1% 1Þ GaN diffraction spots on LGO (0 0 1) measured with the scintillation counter.
piling-up [15]. Further study is necessary to clarify this phenomenon.
4. Conclusions PLD makes it possible to grow high quality GaN films on LGO epitaxially without the use of any buffer layers, which is important to take advantage of the nearly lattice matched substrate. G-GIXD measurements with the SR X-ray source have revealed that thin GaN films grown on LGO substrates are compressed in the lateral directions and that the lattice constant of GaN in the LGO [1 2 0] direction is slightly larger than that in LGO [0 1 0] direction. This anisotropy is probably related to the crystal orientation dependence of the thermal contraction rate for LGO during the cooling down from the growth temperature.
References [1] S.D. Lester, F.A. Ponce, M.G. Craford, D.A. Steigerwald, Appl. Phys. Lett. 66 (1995) 1249.
[2] S. Nakamura, M. Senoh, N. Iwasa, S. Nagahama, Appl. Phys. Lett. 67 (1995) 1868. [3] S. Nakamura, Mater. Sci. Eng. B43 (1997) 258. [4] S. Yoshida, S. Misawa, S. Gonda, Appl. Phys. Lett. 42 (1983) 427. [5] K. Kushi, H. Sasamoto, D. Sugihara, S. Nakamura, A. Kikuchi, K. Kishino, Mater. Sci. Eng. B59 (1999) 65. [6] J. Ohta, H. Fujioka, M. Sumiya, H. Koinuma, M. Oshima, J. Crystal Growth 225 (2001) 73. [7] J. Ohta, H. Fujioka, H. Takahashi, M. Sumiya, M. Oshima, J. Crystal Growth 233 (2001) 779. [8] H. Fujioka, J. Ohta, H. Katada, T. Ikeda, Y. Noguchi, M. Oshima, J. Crystal Growth 229 (2001) 137. [9] J. Ohta, H. Fujioka, H. Takahashi, M. Oshima, Phys. Stat. Sol. A 188 (2001) 497. [10] K. Kawasaki, Y. Takagi, N. Nose, H. Morikawa, S. Yamazaki, T. Kikuchi, S. Sasaki, Rev. Sci. Instr. 63 (1992) 1023. [11] T. Uragami, H. Fujioka, I. Waki, T. Mano, K. Ono, M. Oshima, Y. Takagi, M. Kimura, T. Suzuki, Jpn. J. Appl. Phys. 39 (2000) 4483. [12] T. Uragami, A.S. Acosta, H. Fujioka, T. Mano, J. Ohta, H. Ofuchi, M. Oshima, Y. Takagi, M. Kimura, T. Suzuki, J. Crystal Growth 234 (2002) 197. [13] M. Marezio, Acta Crystallogr. 18 (1965) 481. [14] T. Ishii, Y. Tazoh, S. Miyazawa, J. Crystal Growth 189/ 190 (1998) 208. [15] J. Kruger, . G.S. Sudhir, D. Corlatan, Y. Cho, Y. Kim, R. Klockenbrink, S. Ruvimov, Z. Lilienthal-Weber, C. Kisielowski, M. Rubin, E.R. Weber, B. McDermott, R. Pittman, E.R. Gertner, Mater. Res. Soc. Symp. 482 (1998) 447.