ARTICLE IN PRESS Journal of Crystal Growth 310 (2008) 3946– 3949
Contents lists available at ScienceDirect
Journal of Crystal Growth journal homepage: www.elsevier.com/locate/jcrysgro
Effect of carbon additive on increases in the growth rate of 2 in GaN single crystals in the Na ﬂux method Fumio Kawamura , Masanori Morishita, Masaki Tanpo, Mamoru Imade, Masashi Yoshimura, Yasuo Kitaoka, Yusuke Mori, Takatomo Sasaki Division of Electrical, Electronic and Information Engineering, Graduate School of Engineering, Osaka University, Yamadaoka 2-1, Suita 565-0871, Japan
a r t i c l e in f o
a b s t r a c t
Available online 7 June 2008
We found that a carbon additive could suppress the unfavorable generation of polycrystals in a crucible without reduction in the yield of GaN in the Na ﬂux method. The suppression of polycrystals due to the effect of carbon signiﬁcantly increased the growth rate of liquid phase epitaxy (LPE), which has been the biggest problem of the Na ﬂux LPE, and enabled an increase in the growth rate above 20 mm/h. A 3 mmthick 2 in GaN crystal was obtained for the ﬁrst time. In addition, the carbon additive was found to have another effect in that the nonpolar face could be widely developed. SIMS measurements revealed that carbon added into a Ga–Na mixed melt was hardly taken into LPE crystals, although carbon did have some favorable effects. & 2008 Elsevier B.V. All rights reserved.
PACS: 61.72.Ef 81.15.Lm Keywords: A2. Single crystal A3. Liquid phase epitaxy (LPE) B1. Carbon B1. Nitrides B2. GaN
1. Introduction Some approaches such as the hydride vapor phase epitaxy (HVPE) [1–3], ammonothermal growth [4–6], high-pressure solution growth [7–9], and the Na ﬂux [10–25] methods have been found to enable the growth of large and high-quality GaN single crystals. The realization of high-quality GaN single crystal substrates with large diameter is expected to bring about the next generation of highly efﬁcient electronic devices. Although 2 in GaN substrates synthesized by the HVPE method have already been commercially produced, the curvature and large dislocation density included in the crystal have been identiﬁed as serious problem for use in future devices. In contrast, the other solution growth methods mentioned above have achieved the growth of GaN crystals with low dislocation density. The high-pressure solution growth method initially demonstrated extremely low dislocation of GaN crystals. In the Na ﬂux method, signiﬁcant reduction of dislocation density was conﬁrmed by applying the liquid phase epitaxy (LPE) technique . Most recently, ammonothermal growth, which has been considered as a disadvantageous method from the perspective of dislocation density, has achieved dislocation density in the order of 103 cm 2 in the growth of 1 in diameter GaN. Based on this success, the potential for the industrial use of ammonothermal growth has been proved because this method is based on the technique of synthesizing quartz. However, the growth rate of these solution growth methods is fairly low and remains as a problem that should be solved immediately.
Corresponding author. Tel.: +81 6 6879 7707; fax: +81 6 6879 7708.
E-mail address: [email protected]
(F. Kawamura). 0022-0248/$ - see front matter & 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.jcrysgro.2008.06.008
Among these techniques, the growth rate of the Na ﬂux method is the fastest (about 10 mm/h). As such, this method is expected to be used for mass production. Furthermore, the lower pressure and temperature of the Na ﬂux method than those of the other solution growth methods are also beneﬁcial. Nevertheless, the growth rate must be improved to several times faster than the current value to be used for mass production techniques. In the Na ﬂux method, GaN crystals are grown in a Ga–Na mixed melt at approximately 800 1C by dissolving a pressurized nitrogen gas. Therefore, increasing the pressure is the simplest method of increasing the growth rate. However, increasing the nitrogen pressure causes superﬂuous supersaturation, particularly near the gas–liquid interface, followed by the generation of polycrystals, as shown in Fig. 1. Once polycrystals are generated, dissolved nitrogen is preferably consumed for the growth of polycrystals rather than for LPE growth, because the place where these polycrystals are generated is around the gas–liquid interface. Eventually, another approach will be necessary to solve the problem of low growth rate since increasing nitrogen pressure cannot be used as an effective solution. In this study, we found that the addition of carbon could markedly suppress the generation of polycrystals, which could improve the growth rate to more than 20 mm/h, and a carbon additive changed the preferential developing face from c to m.
2. Experimental procedures The starting materials of metal-Ga (1.0 g), Na (0.88 g), powdery carbon, and a GaN thin ﬁlm, which was grown on a sapphire
ARTICLE IN PRESS F. Kawamura et al. / Journal of Crystal Growth 310 (2008) 3946–3949
N2 Gas (50 atm)
P o ly c ry s ta ls Y ie ld (% )
8 0 0o C , 5 0 a t m
750 o C or 800o C Stainless steel container Alumina crucible
N2 Gas 0% Polycrystal N
N Ga-Na melt
Y ie ld (% )
7 5 0o C , 5 0 a t m
LPE - GaN MOCVD - GaN Fig. 1. Schematic illustration of the growth of LPE and polycrystals in a stainless steel tube.
3. Results and discussions Fig. 1 shows a schematic illustration of the growth of polycrystals. The amount of polycrystals grown near the gas– liquid interface and the growth increment of LPE were measured by weighing the samples after the growth. In the case of the growth of a 2 in GaN substrate, we used the chamber we described previously . Fig. 2 shows the dependence of the yields of polycrystals and the growth increment of LPE on the amount of carbon at 750 and 800 1C. The yields were calculated as the amounts of metal-Ga consumed in the growth of GaN. The total yield was almost 100% in all attempts. It is noteworthy that almost all Ga was consumed for the growth of polycrystals in the case of a non-additive system at both temperatures. Reproducibility was conﬁrmed by multiple experiments for a non-additive system. At 800 1C, when the amount of carbon exceeded 1 at%, the generation of polycrystals was completely eliminated. In the case of growth at 750 1C, 3 at% was needed to completely eliminate polycrystals even though the addition of carbon consistently reduced the generation of polycrystals. The growth thickness reached about 1 mm when polycrystals were not generated. The supersaturation occurring at 750 1C could be estimated to be higher than that at 800 1C if the applied pressure was the same at 50 atm, as we previously reported [16,17]. This seems to be the
Fig. 2. Yield of LPE–GaN and polycrystals in the growth with and without a carbon additive at 750 and 800 1C.
C: 1at.% Pure system
Concentration (cm -3)
substrate by metal organic chemical vapor deposition (MOCVD), were set in an alumina crucible. The crucible containing the starting materials was transferred into a stainless-steel tube resistant to high temperature and pressure. All procedures were conducted in an Ar-ﬁlled grove box until this stage. After the inner part of the tube was pressurized to a set value, the tube was heated in an electric furnace, while adjusting the inner pressure. GaN crystals could be grown in the Ga–Na solution by maintaining the temperature and pressure for 96 h. When the reaction was completed, the tube was taken from the electric furnace and the crucible was then picked up from the tube. In turn, Na was dissolved in cold ethanol and water. In this process, careful handling was required to avoid a strong reaction of Na since the reaction may cause an explosion.
2.0 Depth (um)
Fig. 3. Carbon concentrations incorporated into the LPE crystals grown in a pure system and a carbon-added system measured by SIMS.
reason why more carbon was necessary to completely eliminate the polycrystals at 750 1C. Thereafter, we measured the carbon content incorporated in LPE crystals grown with and without carbon (Fig. 3). The carbon content was almost the same in both samples, regardless of carbon concentrations in the solution, and the concentrations
ARTICLE IN PRESS 3948
F. Kawamura et al. / Journal of Crystal Growth 310 (2008) 3946–3949
P o ly c ry s ta ls Y ie ld (% )
3 mm N o n -a d d itio n
0 .5 a t. %
Fig. 4. Photograph of the 3 mm-thick LPE–GaN crystal grown with 0.5 at% of carbon in the large chamber and the yield of LPE and polycrystals.
were fairly low (2 1017 cm 3). This result indicates that a certain amount of carbon existed in the solution as contamination, even in the non-additive system. The distribution coefﬁcient of carbon against GaN crystal is probably quite low in a Ga–Na melt system, which may be the reason why the amount of carbon taken into the crystal was the same in both cases because the carbon concentrations reached saturation in the crystals even in the non-additive system. In order to apply the effect of carbon to the growth of 2 in GaN crystals, a large chamber developed for the growth of large-scale crystals as we previously reported was used. In this experiment, 0.5 at% of carbon was added against a ﬂux (metal-Ga, 45 g; Na, 40 g). LPE growth was performed on a 2 in GaN thin ﬁlm under a nitrogen pressure of 45 atm for 192 h. The temperature distribution was controlled by a three-zone heating system, and an inverse temperature distribution (bottom, 890 1C; upper, 860 1C) was realized, which could generate thermal convection. A photograph of the grown crystal and the yield of poly and LPE crystals are shown in Fig. 4. Dark-colored crystals found on the LPE crystal seemed to be grown in a cooling period because the dark portion was not seen in the LPE portion. The thickness of LPE crystals reached approximately 3 mm, with some portions being signiﬁcantly more than 3 mm. As the growth in the large chamber, it was conﬁrmed that approximately 20 h was needed to start the LPE growth, since a long period was required to increase the nitrogen concentrations to a supersaturated value. Considering the waiting period until saturation, the growth rate was at least 20 mm/h, which was more than twice as large as the maximum value we previously attained with Na ﬂux LPE. Previously, it was known that as the volume of melt increases, the ratio of polycrystals against LPE also increases, which has been the most serious problem in growing large GaN crystals. This problem can probably be attributed to the stagnation that can occur in a largevolume experiment. It was the ﬁrst time that the growth increment of LPE exceeded the amount of polycrystals in the case of the growth of 2 in crystals. Further addition of carbon can probably suppress the generation of polycrystals even more. However, overdoping of carbon made the m-face widely developed, which was another important result in the carbon-added system and will be reported in another paper. The change in the preferential developing face by the addition of carbon limits the amount of carbon in the c-face growth. We are now considering the mechanism of the effect of carbon. Elemental carbon is known to be difﬁcult to dissolve in a Ga–Na melt system, and the solubility of carbon is extremely poor. Therefore, CN ions probably work as an effective additive for the
suppression of polycrystals and for signiﬁcantly changing the preferential developing face. Our approaches for clarifying the behavior of CN ions will be reported elsewhere.
4. Conclusions It was found that the addition of carbon can markedly suppress the unfavorable generation of polycrystals. The effect of a carbon additive was more effective in the growth at higher temperatures, resulting in the complete suppression of polycrystals at 1 at% of carbon at 800 1C. The carbon additive was not easily taken into the grown crystals, and the concentrations incorporated into the crystal were 2 1017 cm 3, which was almost the same as the contamination in the growth of the pure system. In the growth of large-scale GaN crystals, the addition of carbon enabled the easy growth of thick 2 in GaN because the Ga source is considerably consumed for the growth of the seed substrate and not for that of polycrystals. As a result, we succeeded in growing a 3 mm-thick 2 in GaN substrate. References  A. Usui, T. Ichihashi, K. Kobayashi, H. Sunakawa, Y. Oshima, T. Eri, M. Shibata, Phys. Status Solidi (a) 194 (2002) 572.  S. Bohyama, K. Yoshikawa, H. Naoi, H. Miyake, K. Hiramatsu, Y. Iyechika, T. Maeda, Phys. Status Solidi (a) 194 (2002) 528.  K. Motoki, T. Okahisa, S. Nakahata, N. Matsumoto, H. Kimura, H. Kasai, K. Takemoto, K. Uematsu, M. Ueno, Y. Kumagai, A. Koukitu, H. Seki, J. Crystal Growth 237–239 (2002) 912.  A. Yoshikawa, E. Ohshima, T. Fukuda, H. Tsuji, K. Oshima, J. Crystal Growth 260 (2004) 67.  T. Hashimoto, F. Wu, J.S. Speck, S. Nakamura, Nat. Mater. 6 (2007) 568.  B. Wang, M.J. Callahan, Crystal Growth Design 6 (2006) 1227.  J. Karpinski, J. Jun, S. Porowski, J. Crystal Growth 66 (1984) 1.  S. Porowski, I. Grzegory, J. Crystal Growth 178 (1997) 174.  M. Bockowski, P. Strak, P. Kempisty, I. Grzegory, S. Krukowski, B. Lucznik, S. Porowski, J. Crystal Growth 307 (2007) 259.  H. Yamane, M. Shimada, T. Sekiguchi, F.J. Disalvo, Chem. Mater. 9 (1997) 413.  M. Aoki, H. Yamane, M. Shimada, S. Sarayama, H. Iwata, F.J. Disalvo, Jpn. J. Appl. Phys. 42 (2003) 5445.  F. Kawamura, M. Morishita, T. Iwahashi, M. Yoshimura, Y. Mori, T. Sasaki, Jpn. J. Appl. Phys. 41 (2002) L1440.  M. Morishita, F. Kawamura, T. Iwahashi, M. Yoshimura, Y. Mori, T. Sasaki, Jpn. J. Appl. Phys. 42 (2003) L565.  F. Kawamura, T. Iwahashi, K. Omae, M. Morishita, M. Yoshimura, Y. Mori, T. Sasaki, Jpn. J. Appl. Phys. 42 (2003) L4.  F. Kawamura, T. Iwahashi, M. Morishita, K. Omae, M. Yoshimura, Y. Mori, T. Sasaki, Jpn. J. Appl. Phys. 42 (2003) L729.  F. Kawamura, M. Morishita, K. Omae, M. Yoshimura, Y. Mori, T. Sasaki, Jpn. J. Appl. Phys. 42 (2003) L879.  M. Morishita, F. Kawamura, M. Kawahara, M. Yoshimura, Y. Mori, T. Sasaki, J. Crystal Growth 270 (2004) 402.
ARTICLE IN PRESS F. Kawamura et al. / Journal of Crystal Growth 310 (2008) 3946–3949
 F. Kawamura, M. Morishita, K. Omae, M. Yoshimura, Y. Mori, T. Sasaki, J. Mater. Sci. Mater. Electron. 16 (2005) 29.  M. Morishita, F. Kawamura, M. Kawahara, M. Yoshimura, Y. Mori, T. Sasaki, J. Crystal Growth 284 (2005) 91.  K. Omae, T. Iwahashi, F. Kawamura, M. Yoshimura, Y. Mori, T. Sasaki, Jpn. J. Appl. Phys. 43 (2004) L173.  J.L. Weyher, P.D. Brown, J.L. Rouviere, T. Wosinski, A.R.A. Zauner, I. Grzegory, J. Crystal Growth 210 (2000) 151.
 F. Kawamura, H. Umeda, M. Kawahara, M. Yoshimura, Y. Mori, T. Sasaki, H. Okado, K. Arakawa, H. Mori, Jpn. J. Appl. Phys. 45 (2006) 2528.  F. Kawamura, H. Umeda, M. Morishita, M. Kawahara, M. Yoshimura, Y. Mori, T. Sasaki, Y. Kitaoka, Jpn. J. Appl. Phys. 45 (2006) L1136.  T. Iwahashi, Y. Kitaoka, F. Kawamura, M. Yoshimura, Y. Mori, T. Sasaki, R. Armitage, H. Hirayama, Jpn. J. Appl. Phys. 46 (2007) L227.  T. Iwahashi, Y. Kitaoka, M. Kawahara, F. Kawamura, M. Yoshimura, Y. Mori, T. Sasaki, R. Armitage, H. Hirayama, Jpn. J. Appl. Phys. 46 (2007) L103.