Japanese Journal of Applied Physics Vol. 47, No. 5, 2008, pp. 3346–3349 #2008 The Japan Society of Applied Physics
Microstructure and Optical Properties of Nonpolar m-Plane GaN Films Grown on m-Plane Sapphire by Hydride Vapor Phase Epitaxy Tongbo W EI, Ruifei D UAN, Junxi W ANG, Jinmin L I, Ziqiang H UO, Jiankun Y ANG, and Yiping ZENG Research and Development Center for Semiconductor Lighting Technology, Institute of Semiconductors, Chinese Academy of Sciences, Beijing 100083, China (Received December 11, 2007; accepted December 24, 2007; published online May 16, 2008)
Thick nonpolar (101 0) GaN layers were grown on m-plane sapphire substrates by hydride vapor phase epitaxy (HVPE) using magnetron sputtered ZnO buffers, while semipolar (101 3 ) GaN layers were obtained by the conventional two-step growth method using the same substrate. The in-plane anisotropic structural characteristics and stress distribution of the epilayers were revealed by high resolution X-ray diffraction and polarized Raman scattering measurements. Atomic force microscopy (AFM) images revealed that the striated surface morphologies correlated with the basal plane stacking faults for both (101 0) and (101 3 ) GaN films. The m-plane GaN surface showed many triangular-shaped pits aligning uniformly with the tips pointing to the c-axis after etching in boiled KOH, whereas the oblique hillocks appeared on the semipolar epilayers. In addition, the dominant emission at 3.42 eV in m-plane GaN films displayed a red shift with respect to that in semipolar epilayers, maybe owing to the different strain states present in the two epitaxial layers. [DOI: 10.1143/JJAP.47.3346] KEYWORDS: HVPE, GaN, sapphire, nonpolar, semipolar
1.
Introduction
GaN-related III–nitride materials have been widely used for visible and ultraviolet light emitting diodes (LEDs) and blue/violet laser diodes (LDs) in the past decade. Up to date, most of these devices are grown on the conventional polar (0001) plane and suffer from undesirable spontaneous and piezoelectric polarization effects. The polarization effects give rise to a significant band bending, and thus reduce the radiative recombination efficiency from quantum wells.1,2) Therefore, the growth of GaN films oriented along nonpolar directions such as {101 0} m-plane and {112 0} a-plane has been receiving considerable attention in order to overcome these problems. Several groups have reported the fabrication of nonpolar a- and m-plane LEDs using different growth methods and substrates.3–6) The m-plane GaN is more stable than the a-plane GaN over a wide range of growth conditions, so more devices are fabricated using nonpolar m-plane GaN epilayers at present.7) However, for m-plane GaN films, they can mostly be obtained on -LiAlO2 (100)8,9) and m-plane 4H- and 6H-SiC substrates,10,11) which have either hydrolyzation and poor thermal stability or higher cost compared with sapphire substrate. Currently, the ZnO substrates with an m-plane are also in use because of its wurtzite structure and a small lattice mismatch with GaN,12) but large-size ZnO substrates are difficult to obtain. In this work, we report that good (101 0) nonpolar GaN can also be grown on m-plane sapphire employing magnetron-sputtered ZnO buffers. The effect of ZnO buffers on the morphology, structural and optical properties of nonpolar (HVPE) films is investigated. 2.
Experimental Methods
The ZnO buffers with a 300 nm thickness were initially deposited on m-plane sapphire substrates by RF magnetron sputtering at 400 C. Low-temperature (LT) GaN buffer layers were then grown at 650 C for 5 min to prevent the decomposition of ZnO in the vertical custom-built HVPE reactor. In addition, the LT GaN buffers were also directly
E-mail address:
[email protected]
grown following the in situ annealing of the m-plane sapphire. After the buffer layer growth, the temperature was increased to 1050 C for the subsequent growth of a 50 mm GaN epilayer at a growth rate of approximately 50 mm/h. Prior to the growth of thick GaN, the nitridation with NH3 was maintained to prevent nitrogen evaporation from the LT-GaN layer. In the growth procedures, the NH3 flow rate was held within 500 ml/min, while the HCl flow rate was typically 20 ml/min. The total flow rate of the NH3 /N2 mixture passing the NH3 inlet was fixed at 1.0 l/min. The total flow rate of GaCl/N2 via the showerheads was 720 ml/min. In addition, a 3 l/min main N2 flow rate to drive reactants and purge the complete reactor was used from the reactor bottom. The morphology of thick GaN films was analyzed by JSM-5600LV scanning electron microscopy (SEM) and Nanoscope III atomic force microscopy (AFM) operated in tapping mode. The structural characterization of the epilayers was performed with Bede D1 high-resolution triple axis diffractometer using Cu K1 radiation. The photoluminescence (PL) was excited at low temperature (10 K) using the 325 nm wavelength from a He–Cd laser. In addition, micro-Raman spectra were obtained with HR800 Raman spectrometer attached to a metallographic microscope to evaluate the stress in the films. 3.
Results and Discussion
To determine the crystallographic orientation and crystalline quality, the epitaxial GaN films were characterized by high-resolution X-ray diffraction (XRD). For the conventional two-step growth method without the ZnO buffer, semipolar (101 3 ) GaN films are obtained on m-plane sapphire according to 2–! scans, as shown in Fig. 1(a). Unlike the similar cases of metal organic chemical vapor deposition (MOCVD)13) and HVPE,14) no (112 2)-oriented GaN films are observed, which implies that the initial nucleation layer grown in different ways may lead to the separation of growth orientations of GaN on the same substrate. However, by inserting the ZnO buffer layers, nonpolar (101 0) GaN films are successfully grown while no other phase purity of GaN is examined. Rapid initiation of
3346
Jpn. J. Appl. Phys., Vol. 47, No. 5 (2008)
T. WEI et al.
(a)
[1210]
[3032]
(b)
[1120]
[0001]
Fig. 1. (Color online) (a) 2–! XRD scans for GaN film with 300-nmthick ZnO buffer layer and without buffer layer on m-plane sapphire, (b) XRD (101 0) and (101 1) ! rocking curve scans for GaN epilayer with ZnO buffer at different azimuth angles.
the GaN growth on the well-oriented ZnO-coated sapphire not only improves the quality of GaN films, but also keeps the coherent epitaxial alignment with the m-plane sapphire substrate. The growth toward the semipolar orientation may be effectively eliminated. Nevertheless, the mechanism for the growth of nonpolar and semipolar planes on the msapphire is not fully understood and is still under investigation. The structural quality of the (101 0) GaN was evaluated from the full widths at half maximum (FWHMs) of symmetric and asymmetric X-ray !-scans in Fig. 1(b). The (101 0) GaN films show clean mosaic anisotropy in the onaxis rocking curves, as observed in nonpolar a-plane GaN films. The (101 0) !-scan FWHMs were 0.42 and 1.01 , respectively, corresponding to the projection of the incident X-ray beam parallel to the [0001] and [112 0] directions. The phenomenon may result from the longest migration length of gallium adatoms along the [0001] direction. Furthermore, the FWHM for the 32 off-axis (101 1) reflection reaches 1.16 , indicating a more defined cell structure and higher mosaic content in the HVPE films.15) Figure 2 shows 10 10 mm2 AFM images of the surface of the GaN films. Both (101 3 ) and (101 0) GaN layers show similar slate like surface morphologies that are typical for nonpolar GaN films. These slates are aligned perpendicular to [303 2 ] for the (101 3 ) plane GaN and to [0001] for the (101 0) plane GaN, and their origins are attributable to the intersections of basal plane stacking faults with the growth surface.14) Furthermore, the morphology also reflects the anisotropic defect structure. The root-mean-square roughness (Rrms ) of the (101 3 ) semipolar GaN film is 8.5 nm, while that
Fig. 2. (Color online) AFM images of (a) (101 3 ) GaN and (b) m-plane (101 0) GaN film surfaces.
of the (101 0) GaN layers is only 5.2 nm, indicating a relatively planar surface. Because of the underlying ZnO buffer layer, the m-plane GaN film exhibits elongated slate like features and shallower depressions compared with the semipolar film, which indicates that the island nucleation growth mode is suppressed at the initial stage. In addition to the striated morphology, another characteristic feature is the occurrence of triangular-shaped pits on the surface as can be observed from Fig. 3(a). Unlike the triangular-shaped pits with a blunt corner previously reported for the MOCVD grown a-plane GaN,16) the pits with a sharp corner appear on the film surface, suggesting a relatively higher growth rate on the (11 00) facet than on the (112 0) facet. This is related to the higher stability of the mplane GaN than the a-plane GaN. These pits also have a common orientation with the base perpendicular to the caxis. The vertical facets opposite the tips are nitrogen-face (0001) c-plane, while the two inclined facets belong to stable {11 01} family of planes. The formation of the pits arises from the in-plane asymmetric growth rates of the N and Ga faces of the nonpolar GaN film. For the c-plane film, the hexagonal pits are open, inverted pyramids, which are associated with nanopipes and are likely the surface depressions formed by dislocations. Here, triangular pits likely decorate threading dislocation terminations at the free surface, which are indeed a part of the hexagonal pyramids revealed by the TEM images and originate from the island growth.17) A panchromatic cathodoluminescence (CL) map of the triangular-shaped pit is given in Fig. 3(b), which shows a clear spatial nonuniformity of luminescence intensity. The emission characteristics of the large-scale triangular-shaped pits reveal that they are formed during growth accompanied by stronger impurity incorporation, which may be oxygen donors.
3347
Jpn. J. Appl. Phys., Vol. 47, No. 5 (2008)
(a)
(b)
[0001]
[1120]
20 µm
20 µm
(c)
T. WEI et al.
(d) Fig. 4. LT PL spectra of GaN epitaxial layers grown without a buffer and with sputtered ZnO buffer layer.
5 µm
5 µm
Fig. 3. (a) SEM and (b) panchromatic CL image of triangular-shaped pits on m-plane GaN film, surface morphologies of m-plane GaN film (c) and (101 3 )-oriented film (d) after etching in boiled KOH for 2 min.
Defect-selective etching is a fast and simple way of determining the density and distribution of defects in semiconductor single crystals, epitaxial layers and devices. Although some studies showed that the etch pit density (EPD) of GaN is usually one or two orders lower than the actual dislocation density as determined by plan-view transmission electron microscopy (TEM), an agreement between the two techniques could be achieved by optimizing the etching process.18) Here, we use boiled KOH as the etchant to understand the etched feature and evaluate the dislocation density of the nonpolar GaN film. Figure 5(c) shows the surface morphology of the m-plane GaN after etching for 2 min. After etching, the stripes become ambiguous and many triangular-shaped pits with diameters of about 1– 5 mm appear on the film surface. The triangularshaped pits tend to align uniformly with the tips pointing to the c-axis. A rough estimation by AFM yields an EPD of 9 107 cm2 , which is on the similar order of the typical dislocation density of GaN film on sapphire. However, basal-plane-faulted dislocations are not revealed by the orthodox etching. For the (101 3 ) semipolar GaN film, the surface also becomes rough after etching, while the resulting feature consists of oblique hillocks as in Fig. 3(d). Anisotropic etching results in the local surface decomposition, and many inclined hillocks with stable (101 1 ) facets are eventually formed. According to Baker et al.,19) the threading dislocation line directions were predominantly in the h101 0i directions on the (0001) plane for the semipolar GaN film, so we are not able to trace the defect-related character by etching similarly to the c-plane GaN. In Fig. 4, typical LT PL spectra of two samples without and with ZnO buffer with similar thickness are shown. Both the spectra are dominated by an emission band at about 3.42 eV, which has been found to spatially correlate with the basal plane stacking faults (BSFs)20) in nonpolar a-plane GaN owing to the planar anisotropic nature of the growth mode. The main peak position of the sample with the ZnO
buffer displays a slight red shift, probably caused by the different strain states present in the two epitaxial layers. At higher energies, the common GaN near the band-edge emission (NBE) is only observed in the spectra with ZnO buffer at 3.46 eV, implying a relatively better crystal quality in contrast with the sample without buffer. In addition, two broad peaks located at 3.28 and 3.19 eV are also revealed for the m-plane GaN film. The 3.28 eV band behaves like a donor–acceptor pair emission band often observed in c-plane GaN, and its origin may be related to the partial dislocations terminating the BSFs. However, the strong emission at 3.19 eV is peculiar and not the longitudinal optic (LO)phonon replica of the 3.28 eV band because of its higher intensity. It is not clear what kind of defect causes the luminescence. More detailed studies on the emission are underway. On the other hand, a well-pronounced emission at 3.33 eV is observed in the semipolar GaN film without the ZnO buffer. The emission can also be seen in MOCVD a-plane layers and is attributable to prismatic plane stacking faults (PSFs).20) Having in mind that the 3.33 eV band exists in the sample without buffer, one can conclude that the PSFs are somehow favorable in semipolar growth mode. Polarized Raman scattering was performed to study the layer orientation and stress states of nonpolar GaN films with ZnO buffer. As shown in Fig. 5, three Raman phonon peaks can be observed, including the A1 (TO) at 532:7 cm1 , the E1 (TO) at 560:0 cm1 , and the E2 (high) at 568:9 cm1 , and the FWHM of the E2 (high) peak is only 2.5 cm1 . However, the selection rules partially break down in the yðx; xÞy configuration owing to the appearance of E1 (TO), which is forbidden according to the selection rules.21) This is may be caused by the high density of defects in nonpolar epitaxial layers grown on sapphire. Another reason may be that the propagation and polarization directions of the incident and scattered light are not exactly along the coordinate axes. Raman scattering spectra are particularly sensitive to the stresses in the epilayer. The shifts of phonon frequencies for each mode = A1 (LO and TO), E1 (LO and TO) or E2 (high) can be expressed as22)
3348
! ¼ 2a xx þ b zz ;
Jpn. J. Appl. Phys., Vol. 47, No. 5 (2008)
T. WEI et al.
Program of China under Grant Nos. 2006AA03A111 and 2006AA03A143. We would also like to thank Professors Hong Chen and Yulong Liu of the Institute of Physics, Chinese Academy of Sciences, for their assistance in the DCXRD test and Raman analysis, respectively.
Fig. 5. Unpolarized and polarized Raman scattering spectra of m-plane GaN epilayers in various backscattering configurations along the ydirection. The Cartesian axes x, y, and z correspond to the [112 0], [11 00], and [0001] directions, respectively.
where x k GaN[112 0] and z k GaN[0001]. As the reference phonon frequencies of unstrained GaN films, !A1(TO) ¼ 531:8, !E1(TO) ¼ 559, and !E2(High) ¼ 568 cm1 are taken.23) Then we obtain the in-plane anisotropic stresses xx ¼ 0:48 GPa and zz ¼ 0:23 GPa for nonpolar film, consistent with the anisotropy of thermal expansion and lattice mismatches between GaN and sapphire. 4.
Conclusions
The effect of ZnO buffer layers on the morphology and structural and optical characteristics of HVPE-grown GaN films on m-plane sapphire was investigated. The ZnO buffer layers were found to change the growth orientation and improve significantly the crystalline quality in comparison with the conventional two-step growth method, although the anisotropic growth remained. After etching in boiled KOH, triangular-shaped EPD of about 9 107 cm2 appeared on the m-plane GaN surface, whereas many oblique hillocks were formed on the semipolar epilayers without ZnO buffers. Except for the 3.42 eV emission, three emission bands including NBE, 3.28 and 3.19 eV were also observed in the LT spectra of m-plane GaN films. On the other hand, the emission at 3.33 eV related to the prismatic plane stacking faults was only observed in semipolar films. In addition, the anisotropy of stress states in the nonpolar GaN epilayers was also revealed by Raman spectra. Acknowledgements This work is supported by the National High Technology
1) F. Bernardini, V. Fiorentini, and D. Vanderbilt: Phys. Rev. B 56 (1997) 10024. 2) P. Lefebvre, A. Morel, M. Gallart, T. Taliercio, J. Allegre, B. Gil, H. Mathieu, B. Damilano, N. Grandjean, and J. Massies: Appl. Phys. Lett. 78 (2001) 1252. 3) N. F. Gardner, J. C. Kim, J. J. Wierer, Y. C. Shen, and M. R. Krames: Appl. Phys. Lett. 86 (2005) 111101. 4) A. Chakraborty, B. A. Haskell, S. Keller, J. S. Speck, S. P. Denbaars, S. Nakamura, and U. K. Mishra: Appl. Phys. Lett. 85 (2004) 5143. 5) A. Chakraborty, B. A. Haskell, H. Masui, S. Keller, J. S. Speck, S. P. Denbaars, S. Nakamura, and U. K. Mishra: Jpn. J. Appl. Phys. 45 (2006) 739. 6) M. C. Schmidt, K. C. Kim, H. Sato, N. Fellows, H. Masui, S. Nakamura, S. P. Denbaars, and J. S. Speck: Jpn. J. Appl. Phys. 46 (2007) L126. 7) B. Imer, F. Wu, M. D. Craven, J. S. Speck, and S. P. Denbaars: Jpn. J. Appl. Phys. 45 (2006) 8644. 8) R. R. Vanfleet, J. A. Simmons, H. P. Maruska, D. W. Hill, M. M. C. Chou, and B. H. Chai: Appl. Phys. Lett. 83 (2003) 1139. 9) J. W. Gerlach, A. Hofmann, T. Hoche, F. Frost, B. Rauschenbach, and G. Benndorf: Appl. Phys. Lett. 88 (2006) 011902. 10) R. Armitage, M. Horita, J. Suda, and T. Kimoto: J. Appl. Phys. 101 (2007) 033534. 11) T. Kawashima, T. Nagai, D. Iida, A. Miura, Y. Okadome, Y. Tsuchiya, M. Iwaya, S. Kamiyama, H. Amano, and I. Akasaki: J. Cryst. Growth 298 (2007) 261. 12) A. Kobayashi, S. Kawano, Y. Kawaguchi, J. Ohta, and H. Fujioka: Appl. Phys. Lett. 90 (2007) 041908. 13) X. Ni, U. Ozgur, A. A. Baski, H. Morkoc, L. Zhou, D. J. Smith, and C. A. Tran: Appl. Phys. Lett. 90 (2007) 182109. 14) T. J. Baker, B. A. Haskell, F. Wu, J. S. Speck, and S. Nakamura: Jpn. J. Appl. Phys. 45 (2006) L154. 15) B. Heying, X. H. Wu, S. Keller, Y. Li, D. Kapolnek, B. P. Keller, S. P. Denbaars, and J. S. Speck: Appl. Phys. Lett. 68 (1996) 643. 16) P. P. Paskov, R. Schifano, B. Monemar, T. Paskova, S. Figge, and D. Hommel: J. Appl. Phys. 98 (2005) 093519. 17) F. Wu, M. D. Graven, S. H. Lim, and J. S. Speck: J. Appl. Phys. 94 (2003) 942. 18) G. Kamler, J. L. Weyherm, I. Grzegory, E. Jezierska, and T. Wosinski: J. Cryst. Growth 246 (2002) 21. 19) T. J. Baker, B. A. Haskell, F. Wu, P. T. Fini, J. S. Speck, and S. Nakamura: Jpn. J. Appl. Phys. 44 (2005) L920. 20) R. Liu, A. Bell, F. A. Ponce, C. Q. Chen, J. W. Yang, and M. A. Khan: Appl. Phys. Lett. 86 (2005) 021908. 21) T. Azuhata, T. Sola, K. Suzuki, and S. Nakamura: J. Phys.: Condens. Matter 7 (1995) L129. 22) J. M. Wagner and F. Bechstedt: Appl. Phys. Lett. 77 (2000) 346. 23) V. Y. Davydov, N. S. Averkiev, I. N. Goncharuk, D. K. Nelson, I. P. Nikitna, A. S. Polkovnikov, A. N. Smimov, M. A. Jacobson, and Q. W. Semchinova: J. Appl. Phys. 82 (1997) 5097.
3349
