TY - JOUR
AU - Sekiguchi, Takashi
AB - Abstract The nonuniformity in hydride vapor phase epitaxy (HVPE)-grown thick GaN was studied using cathodoluminescence (CL) technique. It was found that the nonuniform luminescence feature originated from pit-type defects. Two kinds of pit-type defects were distinguished by their morphology: one was hexagonal V-pit surrounded by {10−11} facets and the other was U-pit with {10−11} facets and the blunt bottom. The {10−11} facets of both pits with N-polarity show the strong CL emission due to the oxygen incorporation, while the matrix with Ga-polarity appeared dark in the CL image. The dim CL contrast was observed in the blunt bottom with the mixed polarities. The blunt bottom suggests the filling of pits during the growth, which may be the key to eliminate pit-type defects in free-standing GaN wafer. cathodoluminescence, GaN, HVPE, pit, impurity luminescence, scanning electron microscopy Introduction Free-standing (FS) GaN wafers are expected as the substrates for high-power laser diodes and high-power white light-emitting diodes. Their strain-free nature and electrical conductivity are outstanding. These substrates can also realize the vertical device structure, which is very attractive for the practical application. Thus, the development of FS-GaN wafers is an important issue in this field. Various growth methods are proposed so far to achieve FS-GaN wafers. They are categorized into three groups, namely melt growth, solution growth and vapor growth. The melt growth is proposed by Porowski and co-workers [1]. Since this technique is conducted at high-temperature and high-pressure condition, it is rather difficult to apply to mass production. The alkali flux [2], solvothermal [3], ammonothermal [4] methods are categorized as the solution growth. Although they can realize GaN wafers over 1 in. at lower temperature and lower pressure condition than melt growth, it is still difficult to achieve the commercially available large-sized GaN wafer due to the low growth rate (50 μm/day for ammonothermal method). For large-sized GaN wafers over 2 in., vapor growth is the most widely used. Hydride vapor phase epitaxy (HVPE) is the most promising method used for the bulk GaN growth due to its high growth rate (typically more than 1 μm/min). Recently, HVPE-grown FS-GaN wafers are commercially available [5,6]. However, the quality of FS-GaN wafers is still far from ideal especially on the uniformity. The nonuniform feature is often reported in the literatures [7,8]. Kim et al. [8] reported that the cross-sectional cathodoluminescence (CL) observation of FS-GaN shows the nonuniform luminescence regions near the growth surface and interface between GaN/sapphire. They speculated that nonuniformity is caused by the surface reconstruction due to the undesired residual gas reaction and column overgrowth. However, this origin has not been clarified yet. In this work, we observed the nonuniform luminescence feature by CL technique and discussed it. Experiment GaN films were grown on the (0001) sapphire substrate using HVPE. First, the substrate was treated with nitridation procedure at 1080°C for 30 min. Then, GaN growth was initiated at 1040°C by introducing the HCl gas. V/III (N/Ga) ratio was fixed at 100 and the growth rate was about 70 μm/h. Finally, three GaN films of 3, 40 and 500 μm thickness were obtained, which were named ‘thin film’, ‘thick film’ and ‘bulk GaN’, respectively. The secondary electron (SE) and CL observations were carried out at room temperature using a Hitachi S4300 field emission scanning electron microscope integrated with CL system (Horiba, MP32S/M). For SE and CL measurement, the electron beam and current were 5 kV and 1 nA, respectively. The in-lens detector was used for the SE observation. CL spectra were taken with parallel detection system of 1 nm wavelength resolution. CL images of band edge emission were recorded at 360 ± 15 nm, which corresponds to the energy of 3.3–3.6 eV. The atomic compositions of selected places were determined by energy-dispersive X-ray (EDX) analysis. Before EDX measurement, the plasma cleaning was performed to remove the contamination on the surface. The quantitative values of atoms were evaluated by EDX analysis program with ZAF correction. Result and discussion Pit-type defects in GaN The typical SE and CL images of thin film, thick film and bulk GaN are shown in Fig. 1. In the thin film, SE image (Fig. 1a) shows the flat surface morphology. However, lots of dark spots are observed in the CL image (Fig. 1b) with the density of 3.7 × 109 cm−2. These dark spots correspond to the threading dislocations (TDs), acting as nonradiative carrier recombination centers [9]. In the thick film, SE image (Fig. 1c) shows the hexagonal or round-shaped pits. The size and density of pits are 2–28 μm and 4.7 × 106 cm−2, respectively. These pits are bright with dark spots in their centers in the CL image (Fig. 1d). The large-sized dark spots with a density of 2.1 × 108 cm−2 are also observed in matrix, which corresponds to the bundled TDs. In bulk GaN, the hexagonal pits >70 μm with six triangular facets are observed in the SE image (Fig. 1e) with a density of 1.6 × 104 cm−2. The CL image (Fig. 1f) shows the bright hexagonal pattern with central dark spots, which coincide with that of the hexagonal pits. The inset in Fig. 1f is the brightened CL image. The bundled TDs with a density of 4.7 × 106 cm−2 are also observed. Fig. 1. View largeDownload slide SE and CL images: (a, b) thin film, (c, d) thick film and (e, f) bulk GaN. The inset in Fig. 1f shows the brighten CL image of the matrix region. Fig. 1. View largeDownload slide SE and CL images: (a, b) thin film, (c, d) thick film and (e, f) bulk GaN. The inset in Fig. 1f shows the brighten CL image of the matrix region. These observations indicate that TDs are dominant defects in thin films. When the film grows thicker, TDs are bundled and its density becomes smaller. The consideration of density change suggests that, when the TDs are bundled, a pit is started to grow. In accordance with the growth, the pits become larger and its density becomes smaller. In any case, the pits become the dominant defects in thick or bulk GaN. Two kinds of pits: V-shape and U-shape pits The magnified SE and CL images of typical pits in the thick GaN film are shown in Fig. 2. In this area, two pits of ∼20 μm in lateral size are observed. These two pit-type defects are clearly distinguished based on their morphology in both the SE and CL images. In the SE image (Fig. 2a), the left pits are composed of six bright facets with six valleys. They are distinguished as the {10−11} facets. The right pit is composed of bright hexagon, and the inner dark hypocycloid. Dark contrast of hypocycloid means that the central area of this pit has been filled to a certain amount. The cross-sectional schematic representation of these pits is shown in Fig. 2b. We named these pits as V- and U-pits based on their bottom shape. The bottom of the V-pit is sharp, whereas that of the U-pit is blunt and bowl-shaped. The CL image in Fig. 2c shows that these V- and U-pits are brighter than the matrix area. The {10−11} facets in the V-pit and outer area of the U-pits are the brightest. The blunt bottom in the U-pit has less brighter than the {10−11} facet area. The dark spots are aligned along valleys in the V-pit and distributed over the blunt bottom in the U-pit. This suggests that TDs are rearranged in the V-pit area, whereas they are not well arranged in the U-pit area. Fig. 2. View largeDownload slide Magnified SE and CL images of the V-/U-pits in the thick GaN film: (a) SE image, (b) the schematics of [11−20] cross-section for the V-/U-pits and (c) the CL image of band edge emission. Fig. 2. View largeDownload slide Magnified SE and CL images of the V-/U-pits in the thick GaN film: (a) SE image, (b) the schematics of [11−20] cross-section for the V-/U-pits and (c) the CL image of band edge emission. CL investigation of pits character The CL spectra obtained from the V-/U-pit regions are shown in Fig. 3. Table 1 shows the variation of peak positions and CL intensities. Those values at (0001) matrix (point A) are 3.435 eV and 2800 counts per second (cps), respectively. The CL spectra at the centers of the V- and U-pits (points B and D) are similar to those at the matrix. On the other hand, at steep {10−11} facets of the V- and U-pits (points C and F), the CL intensities are twice as strong as those of the matrix and the peak positions are shifted about 30 meV lower. The peak position and the intensity of blunt bottom (point E) are between those of the matrix and steep {10−11} facets. Table 1. CL peak variation in thick GaN film CL spectrum  Plane  Peak position (eV)  Intensity (cps)  Point A  (0001) matrix  3.435  2813  Point B  V-pit center  3.430  2278  Point D  U-pit center  3.435  2145  Point E  Blunt bottom of U-pit  3.414  3310  Point C  steep facet of V-pit  3.399  5960  Point F  steep facet of U-pit  3.402  5248  CL spectrum  Plane  Peak position (eV)  Intensity (cps)  Point A  (0001) matrix  3.435  2813  Point B  V-pit center  3.430  2278  Point D  U-pit center  3.435  2145  Point E  Blunt bottom of U-pit  3.414  3310  Point C  steep facet of V-pit  3.399  5960  Point F  steep facet of U-pit  3.402  5248  View Large Fig. 3. View largeDownload slide (a) CL image and (b) CL spectra at various points of the V-/U-pits in the thick GaN film. Fig. 3. View largeDownload slide (a) CL image and (b) CL spectra at various points of the V-/U-pits in the thick GaN film. As for the matrix and pit centers, the emission around 3.430 eV is attributed to the band edge emission. In the steep {10−11} facets, on the other hands, the red-shifted peaks are attributed to the recombination via shallow impurity levels. The emission at the blunt bottom may be regarded as the overlap of band edge emission and this shallow impurity emission. Thus, results from the CL spectra suggest that this shallow impurity may be easily incorporated in the steep {10−11} facets than in the matrix. The influence of impurities in the V-/U-pits EDX analysis is performed at the same positions as CL spectra to confirm the impurity incorporation. EDX spectra were composed of Ga-L, N-K and weak O-K lines. The oxygen concentration was deduced from the quantitative analysis as shown in Table 2. The matrix and the V-/U-pit centers give no oxygen signals. On the other hand, the steep facets of the V-/U-pits contain ∼4% oxygen. The blunt bottom contains 1.3% oxygen. Oxygen concentrations look higher than we expected. At this moment, it is difficult to evaluate the absolution values. However, we can point out that a detectable amount of oxygen is incorporated in the V- and U-pit regions. The facet orientation of the matrix and V-/U-pit centers are (0001), which are terminated with the Ga atoms. The steep facets of the V-/U-pits are {10−11}, which are terminated with the N atoms. The blunt bottom is mixed nature of Ga and N terminated surfaces. Thus, these data suggest that during the growth, oxygen is incorporated in the N-terminated facet but not in the Ga-terminated ones [10]. Such impurity incorporation dependence on the surface atomic arrangement is often pointed out in GaN or ZnO [11–14]. Table 2. Oxygen content variations with different positions CL spectrum  Plane  oxygen (%)  Point A  (0001) matrix  0  Point B  V-pit center  0  Point D  U-pit center  0  Point E  Blunt bottom of the U-pit  1.3  Point C  Steep facet of the V-pit  4.2  Point F  Steep facet of the U-pit  4.1  CL spectrum  Plane  oxygen (%)  Point A  (0001) matrix  0  Point B  V-pit center  0  Point D  U-pit center  0  Point E  Blunt bottom of the U-pit  1.3  Point C  Steep facet of the V-pit  4.2  Point F  Steep facet of the U-pit  4.1  View Large By comparing this tendency with CL spectra, it is clear that CL intensity becomes higher and peak is red-shifted when oxygen concentration is higher. It is reasonable to think that a contained fraction of oxygen existing in the N-site as substitutional, ON, act as a shallow donor with an activation energy of 33 meV [15,16]. This energy agrees with the peak energy shift at the steep facets of the V-/U-pits. The small shift at the blunt bottom of the U-pit may be explained as the overlap of band edge emission and oxygen-related emission. The above discussion indicates that the existence of N-polar facets at the V-/U-pits is the cause of nonuniform luminescence features, namely patch pattern, in the bulk GaN. If we can eliminate such V-/U-pits in the GaN films, the quality of GaN will be significantly improved. The dark spot at the V-pit center in the CL image suggests that a bundle of dislocation exists at the center. The dislocation bundle may suppress the planar growth of GaN to form the V-pits. After generating the V-pits, they are filled with nonfacetted growth of GaN to form the U-pits, as indicated by the blunt bottom of the U-pit. This nonfacetted growth may dissipate the dislocations at the center to the blunt bottom, and dark spots are spread over the blunt bottom of the U-pits. Thus, it is possible to eliminate the V-pits by enhancing the filling of the V-pits. The increase in material supply during HVPE growth may be one possible way to realize this ‘filling-in’ process. Conclusion The nonuniformity of HVPE-grown thick GaN was studied using the CL technique. It was found that the nonuniform luminescence feature of bulk GaN originated from the pit-type defects. Two kinds of pits were distinguished by their morphology. Both V- and U-pits have the N-terminated {10−11} side facets, whereas the matrix region is the Ga-terminated (0001) facet. The bottom of the V-pit is sharp, whereas the bottom of the U-pit is blunt. The {10−11} side facets show the red-shifted luminescence due to the oxygen incorporation, whereas the pit center and matrix area show the weak band edge emissions without the impurity incorporation. The blunt bottom of the U-pit shows the mixed CL characters due to the weak oxygen incorporation by mixed polarities. This blunt bottom of the U-pit is regarded as the intermediate state of V-pit filling. Thus, V-pit can be eliminated by the enhancement of U-pit growth to suppress the nonuniformity of bulk GaN. Funding J.H.C. was supported by the MKE (the Ministry of Knowledge Economy), Korea, under the ITRC (Information Technology Research Center) support program supervised by the NIPA (National IT Industry Promotion Agency) (NIPA-2010-C1090-1021-0015). Acknowledgement The authors are grateful to Dr Ogiwara and Dr Hasegawa (NIMS). References 1 Boćkowski M,  Grzegory I,  Krukowski S,  Łucznik B,  Wróblewski M,  Kamler G,  Borysiuk J,  Kwiatkowski P,  Jasik K,  Porowski S.  Deposition of bulk GaN from solution in gallium under high N2 pressure on silicon carbide and sapphire substrates,  J. Cryst. Growth ,  2004, vol.  270 (pg.  409- 419) https://doi.org/10.1016/j.jcrysgro.2004.06.060 Google Scholar CrossRef Search ADS   2 Sekiguchi T,  Yamane H,  Aoki M,  Araki T,  Shimada M.  Cathodoluminescence study of h- and c-GaN single crystals grown by a Na or K flux,  Sci. Tech. Adv. Mater. ,  2002, vol.  3 (pg.  91- 94) https://doi.org/10.1016/S1468-6996(02)00003-7 Google Scholar CrossRef Search ADS   3 Ehrentraut D,  Sato H,  Kagamitani Y,  Sato H,  Yoshikawa A,  Fukuda T.  Solvothermal growth of ZnO,  Prog. Cryst. Growth Charact. Mater. ,  2006, vol.  52 (pg.  280- 335) https://doi.org/10.1016/j.pcrysgrow.2006.09.002 Google Scholar CrossRef Search ADS   4 Dwiliński R,  Doradziński R,  Garczyński J,  Sierzputowski L P,  Puchalski A,  Kanbara Y,  Yagi K,  Minakuchi H,  Hayashgi H G.  Bulk ammonothermal GaN,  J. Cryst. Growth ,  2009, vol.  311 pg.  3015  https://doi.org/10.1016/j.jcrysgro.2009.01.052 Google Scholar CrossRef Search ADS   5 Oshima Y,  Yoshida T,  Eri T,  Shibata M,  Mishima T.  Thermal and electrical properties of high quality freestanding GaN wafers with high carrier concentration,  Jpn J. Appl. Phys. ,  2006, vol.  45 (pg.  7685- 7687) https://doi.org/10.1143/JJAP.45.7685 Google Scholar CrossRef Search ADS   6 Motoki K,  Okahisa T,  Nakahata S,  Matsumoto N,  Kimura H,  Kasai H,  Takemoto K,  Uematsu K,  Ueno M,  Kumagai Y,  Koukitu A,  Seki H.  Growth and characterization of free standing GaN substrates,  J. Cryst. Growth ,  2002, vol.  237 pg.  912  https://doi.org/10.1016/S0022-0248(01)02078-4 Google Scholar CrossRef Search ADS   7 Wang F,  Lu H,  Xiu X,  Chen D,  Zhang R,  Zheng Y.  Dislocation clustering and luminescence nonuniformity in bulk GaN and its homoepitaxial film,  J. Electron. Mater. ,  2010, vol.  39 (pg.  2243- 2247) https://doi.org/10.1007/s11664-009-1040-8 Google Scholar CrossRef Search ADS   8 Kim H M,  Oh J E,  Kang T W.  Nonuniformities in free-standing GaN substrates,  Mater. Sci. Eng. B ,  2001, vol.  79 (pg.  16- 19) https://doi.org/10.1016/S0921-5107(00)00543-2 Google Scholar CrossRef Search ADS   9 Ino N,  Yamamoto N.  Low temperature diffusion of excitons in gallium nitride measured by cathodoluminescence technique,  Appl. Phys. Lett. ,  2008, vol.  93 pg.  232103  https://doi.org/10.1063/1.3040310 Google Scholar CrossRef Search ADS   10 Tuomisto F,  Saarinen K,  Lucznik B,  Grzegory I,  Teisseyre H,  Suski T,  Porowski S,  Hageman P R,  Likonen J.  Effect of growth polarity on vacancy defect and impurity incorporation in dislocation free GaN,  Appl. Phys. Lett. ,  2005, vol.  86 pg.  031915  https://doi.org/10.1063/1.1854745 Google Scholar CrossRef Search ADS   11 Peng C X,  Weng H M,  Zhu C F,  Ye B J,  Zhou X Y,  Han R D,  Fong W K,  Surya C.  Influence of GaN polarity and intermediate-temperature buffer layers on strain relaxation and defects,  Physica B ,  2007, vol.  391 (pg.  6- 11) https://doi.org/10.1016/j.physb.2006.05.431 Google Scholar CrossRef Search ADS   12 Shiojiri M,  Chuo C C,  Hsu J T,  Yang J R,  Saijo H.  Structure and formation mechanism of V defects in multiple InGaN/GaN quantum well layers,  J. Appl. Phys. ,  2006, vol.  99 pg.  073505  https://doi.org/10.1063/1.2180532 Google Scholar CrossRef Search ADS   13 Yeh Y H,  Chen K M,  Wu Y H,  Hsu Y C,  Lee W I.  Hydrogen etching on the surface of GaN for producing patterned structures,  J. Cryst. Growth ,  2011, vol.  314 (pg.  9- 12) https://doi.org/10.1016/j.jcrysgro.2010.10.063 Google Scholar CrossRef Search ADS   14 Urbieta A,  Fernandez P,  Piqueras J,  Hardalov Ch,  Sekiguchi T.  Cathodoluminescence microscopy of hydrothermal and flux grown ZnO single crystals,  J. Phys. D: Appl. Phys. ,  2001, vol.  34 (pg.  2945- 2949) https://doi.org/10.1088/0022-3727/34/19/303 Google Scholar CrossRef Search ADS   15 Frayssinet E,  Kanp W,  Krukowaki S,  Perlin P,  Wisniewski P,  Suski T,  Grzegory I,  Porowski S.  Evidence of free carrier concentration gradient along the c-axis for undoped GaN single crystals,  J. Cryst. Growth ,  2001, vol.  230 (pg.  442- 447) https://doi.org/10.1016/S0022-0248(01)01294-5 Google Scholar CrossRef Search ADS   16 Cruz S C,  Keller S,  Mates T E,  Mishra U K,  Denbaars S P.  Crystallographic orientation dependence of dopant and impurity incorporation in GaN films grown by metalorganic chemical vapor deposition,  J. Cryst. Growth ,  2009, vol.  311 (pg.  3817- 3823) https://doi.org/10.1016/j.jcrysgro.2009.02.051 Google Scholar CrossRef Search ADS   © The Author 2011. Published by Oxford University Press [on behalf of Japanese Society of Microscopy]. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com
TI - Cathodoluminescence study of nonuniformity in hydride vapor phase epitaxy-grown thick GaN films
JF - Journal of Electron Microscopy
DO - 10.1093/jmicro/dfr093
DA - 2011-12-08
UR - https://www.deepdyve.com/lp/oxford-university-press/cathodoluminescence-study-of-nonuniformity-in-hydride-vapor-phase-jkIbr80uCp
SP - 25
EP - 30
VL - 61
IS - 1
DP - DeepDyve
ER -