The influences of supersaturation on LPE growth of GaN single crystals using the Na flux method

The influences of supersaturation on LPE growth of GaN single crystals using the Na flux method

Journal of Crystal Growth 270 (2004) 402–408 www.elsevier.com/locate/jcrysgro The influences of supersaturation on LPE growth of GaN single crystals u...

417KB Sizes 0 Downloads 4 Views

Journal of Crystal Growth 270 (2004) 402–408 www.elsevier.com/locate/jcrysgro

The influences of supersaturation on LPE growth of GaN single crystals using the Na flux method Masanori Morishita, Fumio Kawamura, Minoru Kawahara, Masashi Yoshimura, Yusuke Mori, Takatomo Sasaki Graduate School of Engineering, Electrical Engineering, Osaka University, Sasaki Laboratory, Yamadaoka 2-1, Suita, Osaka, 565-0871, Japan Received 7 June 2004; accepted 9 July 2004 Communicated by K.W. Beng

Abstract The dependency of LPE growth rate and dislocation density on supersaturation in the growth of GaN single crystals in the Na flux was investigated. When the growth rate was low during the growth of GaN at a small value of supersaturation, the dislocation density was much lower compared with that of a substrate grown by the Metal Organic Chemical Vapor Deposition method (MOCVD). In contrast, when the growth rate of GaN was high at a large value of supersaturation, the crystal was hopper including a large number of dislocations. The relationship between the growth conditions and the crystal color in GaN single crystals grown in Na flux was also investigated. When at 800 1C the nitrogen concentration in Na–Ga melt was low, the grown crystals were always tinted black. When the nitrogen concentration at 850 1C was high, transparent crystals could be grown. r 2004 Elsevier B.V. All rights reserved. PACS: 81.05.Ea; 81.10.Dn; 81.15.Lm Keywords: A1. Growth from solutions; A2. Single crystal growth; A3. Liquid phase epitaxy; B1. Gallium compounds; B2. Nitrides; B3. Semiconducting III–V materials

1. Introduction GaN-based semiconductors have already been used practically for short wavelength-emitting Corresponding author. Tel./fax: +81-6-6879-7707

E-mail address: [email protected] (M. Morishita).

devices such as the laser diode (LD) and light emitting diode (LED) [1–4]. In order to realize a highly integrated optical disk and to replace the fluorescent lamps with LED, the emitting efficiency of GaN-based devices should be further increased. GaN-based semiconductors are mainly fabricated on other substrates, such as sapphire, SiC, GaAs, and so on, which generate a great

0022-0248/$ - see front matter r 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.jcrysgro.2004.07.042

M. Morishita et al. / Journal of Crystal Growth 270 (2004) 402–408

number of dislocations in the devices. It is well known that the dislocations in devices lower the emitting efficiency of LD and LED [5,6]. Application of the homoepitaxial growth technique for fabricating the devices is expected to reduce the dislocation density drastically. Therefore, large GaN single crystals with low dislocation density, which can be used as a substrate for fabricating the devices, are now strongly in demand. The Na flux method is one of the growth methods for large GaN single crystals [7–9] with great potential. In the growth of GaN using the Na flux method, applying the nitrogen gaseous pressure to the Ga–Na melt system realizes continuous nitrogen dissolution, and the resultant GaN single crystal grows under conditions of about 800 1C and 50 atm. When GaN crystals are grown in Ga melt by applying nitrogen gas pressure, a severe growth condition (e.g., about 1500 1C and 10,000 atm) is required [10,11]. Crystals with low dislocation density and high crystallinity can be synthesized in the Na–Ga melt system. However, the growth rate is rather low, which hinders the growth of large single crystals. Although growth of a large GaN single crystal with size of 1 cm was reported by Aoki et al. [9] using spontaneous nucleation, a long growth period of 300 h was needed. We previously applied the liquid phase epitaxy (LPE) technique to the Na flux method and reported that a large GaN single crystal with low dislocation density could be grown [12,13]. It is well known that the dislocation density in the crystal grown by the LPE technique varies with supersaturation in other compound semiconductors, such as SiGe, GaAs, InP, and so on [14–18]. In the case of growth of GaN using the Na flux method, an examination of the relationship between supersaturation and dislocation density is essential. However, in the Na flux method, supersaturation is generated by a continuous supply of nitrogen gas to the Ga–Na melt under high pressure, which prevented an estimation of supersaturation. In our previous paper, we showed that the nitrogen concentration in the Ga–Na melt system and supersaturation can be calculated quantitatively [19]. In this study, we investigated the dependency of dislocation density

403

and growth rate of the homoepitaxial LPE film upon supersaturation, and the relationship between the growth conditions and color of the GaN crystals grown.

2. Experimental procedure Fig. 1 shows an illustration of the experimental setup. The experimental procedures were as follows: (1) In a nitrogen-charged grove box, Ga: 1.0 g, Na: 0.88 g (Ga:Na=27:73 (molar ratio)), and an 18-mm-thick GaN film grown on a sapphire (0 0 0 1) substrate synthesized by the MOCVD method were first put into an alumina crucible with an inner diameter of 9 mm and a height of

Fig. 1. Illustration of experimental setup.

M. Morishita et al. / Journal of Crystal Growth 270 (2004) 402–408

404

50 mm.The alumina crucible was filled with the starting materials and the GaN film was transferred to a stainless steel tube connected to a nitrogen gas cylinder. Pressure inside the tube can be controlled by adjusting the regulator. The stainless steel tube, in which pressure was adjusted to the fixed pressure, was heated to 850 1C in the electric furnace, and the temperature and pressure were kept constant for 96 h. In this period, GaN crystals were grown on the GaN film by continuous nitrogen dissolution into a solution. After the growth, the stainless steel tube was cooled down to room temperature by furnace cooling. Na flux was decomposed in cold ethanol, and the GaN single crystal grown on the substrate was removed from the alumina crucible. Dislocation density of the GaN obtained on the substrate was estimated by a chemical-etching method. In this process, the crystals were treated in 230 1C etchant (H2SO4:H3PO4=1:1) for 15 min, and the number of etch pits on the surface of the crystal was counted.

3. Results 3.1 The relationship between the growth rate of LPE film and supersaturation The dependence of the growth rate on supersaturation is indicated in Fig. 2. The supersatura-

10

Region A

Growth rate (µm/h)

9

tion in the growing period can be calculated from estimation of the difference between the nitrogen concentration in the Ga–Na melt and that required for growing GaN in the flux [19]. In order to clarify the growth mechanism according to supersaturation, the growing condition was divided into three regions: region A (supersaturation s: 36%–65%), region B (s: 65%–85%), and region C (s: 85%–148%). In region A, the LPE growth starts at supersaturation over 36%, and the growth rate was increased with supersaturation. In region B, a maximum growth rate of 7.9mm/h was achieved at supersaturation of 70%. However, in the region C, the growth rate of LPE film was monotonously decreased with supersaturation. 3.2 Dislocation density and the coloring of crystals We investigated the dislocation density and the coloring of the crystals grown under the growth conditions in regions A–C. Region A: Figs. 3a–c show the differential interference microscope (DIM) photograph of the surface of GaN single crystals grown in regions A, B, and C, respectively. In Fig. 3a, although a great number of giant steps formed by step bunching were confirmed, the whole surface of the crystal was bounded by only the GaN (0001) face.

Region B

Region C

8 7 6 5 4 3 2 1 0 0

20

40

60

80

100

120

Supersaturation (%) Fig. 2. The dependence of the GaN growth rate on supersaturation.

140

M. Morishita et al. / Journal of Crystal Growth 270 (2004) 402–408

405

Despite the fact that the dislocation density of the GaN substrate film grown by the MOCVD method was 8.0  106 (cm 2), the LPE–GaN grown on the film had a dislocation density in the range 4.7  104–1.1  105 (cm 2), which means an unknown mechanism, functions to reduce the dislocation density in the growing stage. This dislocation reduction mechanism is now under investigation. Fig. 4 shows a typical stereoscopic microphotograph of a GaN single crystal grown in region A. The crystal grown at 850 1C in the region A was transparent by visual examination. On the other hand, GaN crystals grown at 800 1C were inevitably tinted black at the beginning of growth despite supersaturation (Fig. 5). However, GaN crystals grown at 800 1C (shown in Fig. 5) also turned transparent in the later term of growth. Region B: Fig. 3b shows a DIM photograph of the crystal surface formed in region B. The morphology of the crystal grown in region B was always hopper because the supersaturation was higher than that in region A. The dislocation density in the crystal was in the range of 2.3  106–2.9  106 (cm 2), which was higher than that of crystals grown in region A. Region C: Fig. 3c shows a DIM photograph of the surface of crystals grown in region C. Although the crystals were grown at extremely high supersaturation, the surface morphology was relatively flat without formation of hopper.

Fig. 3. A DIM photograph of the surface of the crystal obtained in region A (a), in region B (b), and in region C (c). Fig. 4. A stereoscopic microphotograph of the crystal grown at 850 1C.

406

M. Morishita et al. / Journal of Crystal Growth 270 (2004) 402–408

Fig. 5. A stereoscopic microphotograph of the crystal grown at 800 1C.

The degree of dislocation density included in the crystal is as same as that in the crystals grown in the region A; 4.1  104–1.5  105 (cm 2).

GaN polycrystals

substrate

4. Discussion 4.1 The relationship between the growth rate of LPE film and supersaturation As shown in Fig. 2, the threshold supersaturation required for LPE growth with the Na flux method is about 30%, which is supersaturation far higher than that in LPE growth with SiGe, GaAs, InP, and so on [14–18]. The supersaturation in the Na flux system is considered as the gap between the nitrogen concentration in the Na–Ga melt and the minimum nitrogen concentration required for the growth of GaN [19]. Therefore, the value shown here is only valid in the case of the gas–liquid equilibrium condition. The real supersaturation in the Na flux system should be lower than the calculated value because the nitrogen in the melt is consumed as the crystal growth progresses. In Fig. 2, the LPE growth rate of GaN increases with the supersaturation in Na flux system up to supersaturation of about 70%, and then starts to decrease monotonously. When the supersaturation reaches 70%, heterogeneous nucleation on the alumina crucible occurs near the gas–liquid interface, as shown in Fig. 6, and results in decrease of the growth rate. When that happens, nitrogen supply to the substrate should be reduced because

Fig. 6. Schematic illustration of GaN nucleated on the crucible near the gas–liquid interface.

of the consumption of nitrogen by the growth of GaN nucleated near the gas–liquid interface. Another possible reason for the reduction of growth rate at high supersaturation is the decrease in the area of gas–liquid interface that results when the polycrystals grow along the gas–liquid interface. 4.2 Dislocation density and the coloring of crystals Region A: GaN single crystals grown at 850 1C in region A were transparent, as shown in Fig. 4. The crystals grown at 800 1C were always tinted black at any supersaturation, as described in Fig. 5, though the surface, which had grown in the late term of the growth period, was transparent. We reported that the coloring of GaN single crystals grown in the Na flux system was probably due to the nitrogen vacancies in the crystal [13,20,21]. The nitrogen concentration in the Na–Ga melt at 800 1C at the beginning of LPE growth is estimated at about 0.11 at% (the N/Ga ratio in the solution is 4.6  10 3), which is the

M. Morishita et al. / Journal of Crystal Growth 270 (2004) 402–408

minimum value to grow GaN at 800 1C [19]. In the late term of LPE growth, the N/Ga ratio in the Na–Ga melt rises because Ga is consumed in the LPE growth. Consequently, transparent crystals with low nitrogen vacancies are easily grown. For the transparent crystal shown in Fig. 4, nitrogen concentration in the early term of LPE growth is about 0.23 at% (the N/Ga ratio in Na–Ga melt is 7.4  10 3) using the same calculation shown in the prior paragraph. The higher N/Ga ratio at 850 1C (relative to that at 800 1C) probably leads to the growth of transparent crystals. Region B: Hopper crystal with higher dislocation density than that in region A was grown in region B. Fig. 7 shows a scanning electron microscope (SEM) photograph of the surface of hopper crystal treated with 230 1C H3PO4–H2SO4 etchant in 15 min. It was confirmed, as shown in Fig. 7, that etch pits on the hopper crystal occurred in a row, probably along the boundaries between facets. In the growth from the hopper crystals, frequent contact between facets generates a large number of new boundaries, which leads to high dislocation density. Furthermore, hopper crystals tend to have a lot of inclusions. This is another possible reason for the high dislocation density in the crystals grown in region B. Region C: The dislocation density in the crystals grown in region C was in the range of 4.1  104–1.5  105 (cm 2), which is almost the

407

same as that in region A despite the fact that region C has higher supersaturation than region B. The reduction of the supersaturation by consumption of nitrogen in the growth of polycrystals nucleated on the crucible [6] is surely the cause of low dislocation density in the crystals. 5. Conclusions We revealed the dependency of dislocation density and the growth rate of GaN single crystals on supersaturation in the Na flux system, using the LPE technique. The LPE growth rate was increased with supersaturation up to 70%, over which it was decreased due to the nucleation of GaN on the crucible near the gas–liquid interface. The dislocation density of the homoepitaxial LPE film was far lower than that of the GaN thin film grown by the MOCVD method. However, the hopper crystal, which was grown at high supersaturation, had a large number of dislocations due to the inclusions taken into the crystals and because the boundaries generated by the contacts between facets cause an outbreak of dislocations. The N/Ga ratio in the melt relates to the density of nitrogen vacancies in GaN crystals, and the increase in nitrogen concentration leads to improvement of the color and transparency of the grown crystals. References

Fig. 7. A SEM photograph of the hopper crystal treated with 230 1C H3PO4–H2SO4 etchant in 15 min.

[1] S. Nakamura, M. Senoh, N. Iwasa, S. Nagahama, Jpn. J. Appl. Phys. 34 (1995) L797. [2] S. Nagahama, N. Iwasa, M. Senoh, T. Matsushita, Y. Sugimoto, H. Kiyoku, T. Kozaki, M. Sano, H. Matsumura, H. Umemoto, K. Chocho, T. Yamamoto, T. Mukai, Phys. Status Solidi A 188 (2001) 1. [3] S. Uchida, M. Takeya, S. Ikeda, T. Mizuno, T. Fujimoto, O. Matsumoto, S. Goto, T. Tojyo, M. Ikeda, IEEE J. Sel. Top. Quant. Electron. 9 (2003) 1252. [4] S. Goto, M. Ohta, Y. Yabuki, Y. Hoshina, K. Nagamura, K. Tamamura, T. Hashizu, M. Ikeda, Phys. Status Solidi A 200 (2003) 122. [5] T. Miyajima, T. Hino, S. Tomiya, K. Yanashima, H. Nakajima, T. Araki, Y. Nanishi, A. Satake, Y. Masumoto, K. Akimoto, T. Kobayashi, M. Ikeda, Phys. Status Solidi B 228 (2001) 395. [6] H. Amano, I. Akasaki, J. Phys.: Condens. Mater 13 (2001) 6935.

408

M. Morishita et al. / Journal of Crystal Growth 270 (2004) 402–408

[7] H. Yamane, M. Shimada, S. Sarayama, F.J. DiSalvo, Chem. Mater. 9 (1997) 413. [8] M. Aoki, H. Yamane, M. Shimada, T. Sekiguchi, T. Hanada, T. Yao, S. Sarayama, F.J. DiSalvo, J. Crystal Growth 218 (2000) 7. [9] M. Aoki, H. Yamane, M. Shimada, S. Sarayama, F.J. DiSalvo, J. Crystal Growth 242 (2002) 70. [10] J. Karpinski, J. Jun, S. Porowski, J. Crystal Growth 66 (1984) 1. [11] I. Grzegory, M. Bockowski, B. Lucznik, S. Krukowski, Z. Romanowski, M. Wrobiewski, S. Porowski, J. Crystal Growth 246 (2002) 177. [12] F. Kawamura, T. Iwahashi, K. Omae, M. Morishita, M. Yoshimura, Y. Mori, T. Sasaki, Jpn. J. Appl. Phys. 42 (2003) L4. [13] F. Kawamura, T. Iwahashi, M. Morishita, K. Omae, M. Yoshimura, Y. Mori, T. Sasaki, Jpn. J. Appl. Phys. 42 (2003) L729.

[14] A.M. Sembian, M. Konuma, I. Silier, A. Gutjahr, N. Rollbu¨hler, F. Banhart, S. Moorthy Babu, P. Ramasamy, Thin Solid Films 366 (1998) 116. [15] A.M. Sembian, F. Banhart, M. Konuma, J. Weber, S.M. Babu, P. Ramasamy, Thin Solid Films 372 (2000) 1. [16] A. Tanaka, N. Izumi, M. Kimura, T. Sukegawa, J. Crystal Growth 208 (2000) 33. [17] Z. Yan, S. Naritsuka, T. Nishinaga, J. Crystal Growth 198/199 (1999) 1077. [18] Z. Yan, S. Naritsuka, T. Nishinaga, J. Crystal Growth 203 (1999) 25. [19] F. Kawamura, M. Morishita, M. Yoshimura, Y. Mori and T. Sasaki, J. Mat. Sci., in press. [20] F. Kawamura, M. Morishita, T. Iwahashi, M. Yoshimura, Y. Mori, T. Sasaki, Jpn. J. Appl. Phys. 41 (2002) L1440. [21] M. Morishita, F. Kawamura, T. Iwahashi, M. Yoshimura, Y. Mori, T. Sasaki, Jpn. J. Appl. Phys. 42 (2003) L565.