Bulk GaN crystals grown by HVPE

Bulk GaN crystals grown by HVPE

ARTICLE IN PRESS Journal of Crystal Growth 311 (2009) 3011–3014 Contents lists available at ScienceDirect Journal of Crystal Growth journal homepage...

280KB Sizes 0 Downloads 24 Views

ARTICLE IN PRESS Journal of Crystal Growth 311 (2009) 3011–3014

Contents lists available at ScienceDirect

Journal of Crystal Growth journal homepage: www.elsevier.com/locate/jcrysgro

Bulk GaN crystals grown by HVPE Kenji Fujito a,, Shuichi Kubo a, Hirobumi Nagaoka a, Tae Mochizuki a, Hideo Namita b, Satoru Nagao b a

Optoelectronics Laboratory, Mitsubishi Chemical Corporation, 1000, Higashi-Mamiana, Ushiku, Ibaraki 300-1295, Japan Fundamental Technology Laboratory, Research and Development Division, Mitsubishi Chemical Group Science and Technology Research Center, Inc., 1000, Kamoshida-cho, Aobaku, Yokohama 227-8502, Japan b

a r t i c l e in fo

abstract

Available online 19 January 2009

We succeeded in preparing very thick c-plane bulk gallium nitride (GaN) crystals grown by hydride vapor phase epitaxy. Growth of the bulk GaN crystals was performed on templates with 3 mm GaN layer grown by metal organic chemical vapor deposition on (0 0 0 1) sapphire substrates. Colorless freestanding bulk GaN crystals were obtained through self-separation processes. The crystal’s diameter and thickness were about 52 and 5.8 mm, respectively. No surface pits were observed within an area of 46 mm diameter of the bulk GaN crystal. The dislocation density decreased with growth direction (from N-face side to Ga-face side) and ranged from 5.1 106 cm 2 near the N-face surface to 1.2  106 cm 2 near the Ga-face. A major impurity was Si, and other impurities (O, C, Cl, H, Fe, Ni and Cr) were near or below the detection limits by SIMS measurements. & 2009 Elsevier B.V. All rights reserved.

PACS: 81.05.Ea 81.10.Bk Keywords: A2. Growth from vapor A2. Single-crystal growth A3. Hydride vapor phase epitaxy B1. Nitrides B2. Semiconducting III–V materials

1. Introduction Gallium nitride (GaN) and its related alloys are the enabling materials for green, blue and ultraviolet light-emitting diodes (LEDs) [1,2], blue/violet laser diodes (LDs) [3], and high power, high-frequency electronics devices [4]. Although heteroepitaxially grown devices are widely available in the market, high-performance devices such as LDs with high reliability are mainly achieved by growing on freestanding bulk substrates. Using the freestanding GaN substrates has many benefits, such as low dislocation density, high thermal conductivity and so on. The ultimate way to manufacture GaN substrates at low costs is to slice them out of bulk GaN crystals. Substantial efforts have been directed to the growth of true bulk GaN crystals for many years. However, growing bulk GaN crystals is extremely difficult due to the high melting point and high equilibrium pressure of N2 at high temperatures. Several growth techniques have been demonstrated for GaN crystals by solution growth methods using Ga melt with high N2 pressure [5,6], Na flux [7,8] and supercritical ammonia [9,10]. While impressive results were achieved, low growth rate and high impurity concentration are yet to be resolved. Hydride vapor phase epitaxy (HVPE) is one of the most promising methods for the growth of bulk GaN crystals because of its high growth rate compared to other techniques. Recent development in HVPE growth of thick GaN crystals made it possible for nonpolar and semipolar freestanding GaN substrates

Corresponding author. Tel.: +81 29 841 9089; fax: +81 29 843 3796.

E-mail address: [email protected] (K. Fujito). 0022-0248/$ - see front matter & 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.jcrysgro.2009.01.046

to be manufactured. Bulk GaN crystals grown in the [0 0 0 1] direction (c-direction) are sliced along any direction of interest to achieve the unique polarity of the GaN substrates [11]. However, it is still difficult to obtain uniform crystals with large diameters and thicknesses. Several groups have reported relatively thick GaN crystals grown by HVPE; however, they have numerous V-shaped pits on the surface and/or have a dark brownish color [12,13]. In this work, we have demonstrated large-size bulk GaN crystals grown by HVPE with the thickness exceeding 5 mm and evaluated their quality.

2. Experiment Fig. 1 shows a schematic drawing of the HVPE system used in this study. GaN crystals were grown in a homebuilt vertical HVPE reactor at atmospheric pressure with metallic gallium (Ga), hydrochloric acid (HCl) and ammonia (NH3) as sources. Templates with 3 mm GaN layer grown by metal organic chemical vapor deposition (MOCVD) on (0 0 0 1) sapphire substrates were employed as starting substrates for the HVPE growth. The starting substrates were placed face up on the silicon carbide-coated graphite susceptor with a rotation of 12 rpm. HCl was reacted with metallic Ga at 800 1C forming GaCl that was carried to the growth zone to react with NH3 on the substrate surface for GaN crystal growth. Partial pressures of GaCl and NH3 were 4.8  102 and 9.7  103 Pa, respectively. A hydrogen and nitrogen gas mixture was used as a carrier. The growth process was performed at 1010 1C for 55 h. The growth rate of the GaN crystal along the c-direction was approximately 105 mm/h under

ARTICLE IN PRESS 3012

K. Fujito et al. / Journal of Crystal Growth 311 (2009) 3011–3014

these growth conditions. Freestanding bulk GaN crystals were obtained through self-separation processes using the two different thermal expansion coefficients of GaN and the sapphire substrate while cooling [12]. The crystal quality was analyzed by X-ray diffraction (XRD) using both symmetrical (0 0 0 2) and asymmetrical (1 0 1¯ 2) reflection. The concentration of impurities was examined by secondary-ion mass spectroscopy (SIMS) with Cs or O sputtering.

The dislocation density, which can be estimated as a dark spot density was obtained by a cathode luminescence (CL) method. To examine the characteristics of depth direction of a bulk crystal, CL and XRD measurements were done on wafers prepared by using a wire saw to slice six pieces out of the bulk crystal in a direction parallel to the c-plane surface.

3. Result and discussion

H 2+N2

NH 3

Fig. 2 shows an as-grown 52 mm diameter, 5.8 mm thick, colorless freestanding bulk GaN crystal. The growth direction was in the c-direction and a relatively smooth c-plane surface was obtained. No surface pits were observed within an area of 46 mm diameter of the crystal. The bulk GaN crystal spontaneously became a dodecagon shape though the growth was started on a round substrate. The inclined facets observed on the sides of the crystal with angles of 621 and 431 on m-plane sides, and 581 on a-plane sides were determined to be {1 0 1¯ 1}, {1 0 1¯ 2} and {11 2¯ 2} planes, respectively (Fig. 2(c) and (d)). The formation of the inclined facets makes it difficult to obtain larger crystals than the seed crystal. Although a detailed faceting mechanism is still under investigation, we believe that the faceting can be controlled by growth temperature, growth pressure and carrier condition as done in MOCVD [14]. In general, a parasitic growth of polycrystalline GaN in a growth reactor is one of the major problems for bulk GaN crystals growth by HVPE [15]. An optimization of gas-nozzles configurations and flow rates can solve this problem, so the parasitic nucleation was not a severe problem in this study. Total growth thickness should only be limited by the volume of the Ga-boat, which creates opportunities to grow thicker crystals by enlarging the Ga-boat volume and growing for longer times. Relatively low growth rate conditions were used in this study. Although higher growth rates can be achieved by increasing the GaCl and/or NH3 supply, it is difficult to obtain smooth surface crystals at higher growth rates under a long growth time with our configuration. We suspect that the cause of rough surface is related to a nucleation of GaN particles in the vapor phase. The growth rate in

HCl

Quarts reactor

Heater

~ 800 °C

Ga

GaCl

NH3

GaN ~ 1010 °C Sub.

Rotation SiC-coated graphite susceptor

Exhaust

Fig. 1. Schematic drawing of the vertical type HVPE system.

Ga-face {1012} {1120} 5.8mm {1011}

[0001] ~3mm <1010> N-face

Ga-face

{1122}

{1010}

5.8mm

[0001]

<{1120}> N-face Fig. 2. (a) Freestanding bulk GaN crystal grown by HVPE at 1010 1C for 55 h with a thickness of 5.8 mm and diameter of 52 mm. (b) Bird-eyes-view photograph of the bulk GaN crystal. Cross-sectional schematic images of the bulk GaN crystal for (c) {11 2¯ 0} and (d) {1 0 1¯ 0} planes.

ARTICLE IN PRESS K. Fujito et al. / Journal of Crystal Growth 311 (2009) 3011–3014

the c-direction that produces the smoothest surface is below 110 mm/h so far. Further improvements will be required to achieve faster growth rate for increasing productivity. The crystalline quality was evaluated by XRD rocking curve measurements (o-scans) for each sample sliced from the bulk GaN crystal. The width of the (0 0 0 2) peak is used for evaluating dislocations with a screw component and the width of the (1 0 1¯ 2) peak is used for evaluating dislocations with an edge component. Fig. 3 shows the position in which the wafers were sliced out of the crystal. The values of the full-widths at half-maximum (FWHM) of o-scans measured for the (0 0 0 2) and (1 0 1¯ 2) reflections as a function of slicing position are depicted in Fig. 4. The FWHM of o-scans for both (0 0 0 2) and (1 0 1¯ 2) reflections decreases rapidly at distances below 2.5 mm from the N-face, and turns into a gradual slope after distances of 2.5 mm or more. The origin of an inflection point at around 2.5 mm from the N-face is

3013

still under investigation. These results showed that the dislocations of both screw and edge components decreased similarly up to at least 5.8 mm in growth thickness. The dislocation density of each sliced wafers was also determined by CL. Fig. 5 shows the dislocation density as a function of slicing position. The dislocation density decreased with growth direction from N-face side to Ga-face side, and ranged from 5.1 106 cm 2 near the N-face surface to 1.2  106 cm 2 near the Ga-face surface. The dislocation densities observed along the c-direction were in good agreement with the results of the o-scan’s FWHM. SIMS measurements were carried out for the bulk GaN crystal (Table 1). This crystal was unintentionally doped sample and background Si concentration level which is caused by the quarts reactor tube was 1.5  1017 cm 3. The concentrations of other impurities (O, C, Cl, H, Fe, Ni and Cr) were near or below the

6.0E+06 Dislocation density [cm-2]

Ga-face (as-grown surface) 5800 5220 No.6 (565µm No.6 (565µm thickness) thickness) 4330

4.0E+06 3.0E+06 2.0E+06 1.0E+06

Growth direction

0.0E+00 0.0

1.0

3470

2.0 3.0 4.0 5.0 Slicing position [mm]

6.0

No.4 (565µm No.4 (565µm thickness) thickness)

Fig. 5. The dislocation density of each sliced wafers, which can be estimated as a dark spot density by the CL measurements, as a function of slicing position.

No.3 (560µm No.3 (560µm thickness) thickness)

Table 1 Si, O, C, Cl, H, Fe, Ni and Cr concentration for unintentionally doped bulk GaN sample by SIMS measurements.

2570

1730

[0001] No.2 (570µm No.2 (570µm thickness) thickness) Si O C H Cl Fe Ni Cr

820 No.1 (570µm No.1 (570µm thickness) thickness) 0 N-face Fig. 3. The position in which the wafers were sliced out from the crystal.

80 60 50 40 30 20

Growth direction

10

Detection limits cm

1.5  1017 o7  1015 o2  1015 o2  1016 o5  1014 o1  1015 o3  1015 2.2  1014

8  1015 7  1015 2  1015 2  1016 5  1014 1  1015 3  1015 1  1014

80

(0002)

70

Undoped GaN

Detection limits of each element are also indicated in the table.

ω-scan’s FWHM [arcsec]

ω-scan’s FWHM [arcsec]

Growth thickness [µm]

No.5 (560µm No.5 (560µm thickness) thickness)

5.0E+06

(1012)

70 60 50 40 30 20

Growth direction

10 0

0 0.0

1.0

2.0

3.0

4.0

Slicing position [mm]

5.0

6.0

0.0

1.0

2.0

3.0

4.0

5.0

6.0

Slicing position [mm]

Fig. 4. The FWHM of o-scans of each sliced wafers measured for the (a) (0 0 0 2) and (b) (1 0 1¯ 2) reflections as a function of slicing position.

3

ARTICLE IN PRESS 3014

K. Fujito et al. / Journal of Crystal Growth 311 (2009) 3011–3014

detection limits. These results indicate high purity of the bulk GaN crystal grown by HVPE. The SIMS analysis of an intentionally Sidoped sample which can be obtained to flow SiH2Cl2 diluted with hydrogen is still in progress.

4. Conclusions In summary, we have succeeded in preparing as-grown 52 mm diameter, 5.8 mm thick, colorless freestanding c-plane bulk GaN crystals. No surface pits were observed within an area of 46 mm diameter of the bulk GaN crystals. The dislocation density decreased with growth direction and achieved 1.2  106 cm 2 near the as-grown surface. The decreasing tendency of the dislocation density is in good agreement with the FWHM measured from the o-scans for both (0 0 0 2) and (10 1¯ 2) reflections. These results show that the dislocations of both screw and edge components decreased similarly for at least 5.8 mm in growth thickness. A major impurity is Si and the other impurities (O, C, Cl, H, Fe, Ni and Cr) were near or below the detection limits by SIMS measurements. This study shows that HVPE is an effective approach to produce high quality, thick bulk GaN crystals.

Acknowledgements The authors would like to thank Yoichiro Yoshida for his contribution of SIMS measurements. The authors are also grateful

to Takanori Suzuki, Masanobu Ishida, Yoko Mashige, Isao Fujimura and Motoko Furukawa for fruitful discussions and comments to this work.

References [1] S. Nakamura, T. Mukai, M. Senoh, Jpn. J. Appl. Phys. 32 (1993) L16. [2] T. Nishida, N. Kobayashi, Phys. Status Soldi (a) 188 (2001) 113. [3] S. Nakamura, G. Fasol, S.J. Pearton, The Blue Laser Diode, Springer, New York, 2000. [4] L.F. Eastman, U.K. Mishra, IEEE Spectrum 39 (2002) 28. [5] S. Porowski, MRS Internet J. Nitride Semicond. Res. 4S1 (1999) G1.3. [6] T. Inoue, Y. Seki, O. Oda, S. Kurai, Y. Yamada, T. Taguchi, Phys. Status Solidi (b) 223 (2001) 15. [7] H. Yamane, M. Shimada, T. Sekiguchi, F.J. DiSalvo, J. Cryst. Growth 186 (1998) 8. [8] F. Kawamura, M. Morishita, K. Omae, M. Yoshimura, Y. Mori, T. Sasaki, Jpn. J. Appl. Phys. 42 (2003) L879. [9] R. Dwilin´ski, R. Doradzin´ski, J. Garczyn´ski, L. Sierzputowski, J.M. Baranowski, M. Kamin´ska, Diamond Relat. Mater. 7 (1998) 1348. [10] T. Hashimoto, F. Wu, J.S. Speck, Shuji Nakamura, Jpn. J. Appl. Phys. 46 (2007) L889. [11] K. Fujito, K. Kiyomi, T. Mochizuki, H. Oota, H. Namita, S. Nagao, I. Fujimura, Phys. Status Solidi (a) 205 (2008) 1056. [12] A.D. Williams, T.D. Moustakas, J. Cryst. Growth 300 (2007) 37. [13] M. Weyers, E. Richter, C. Hennig, S. Hagedorn, T. Wernicke, G. Tra¨nkle, Proc. SPIE 6910 (2008) 69100I. [14] K. Hiramatsu, K. Nishiyama, A. Motogaito, H. Miyake, Y. Iyechika, T. Maeda, Phys. Status Solidi (a) 176 (1999) 535. [15] B. Łucznik, B. Pastuszka, I. Grzegory, M. Boc´kowski, G. Kamler, E. LitwinStaszewska, S. Porowski, J. Cryst. Growth 281 (2005) 38.