Selective etching of GaN polar surface in potassium hydroxide solution studied by x-ray photoelectron spectroscopy

JOURNAL OF APPLIED PHYSICS VOLUME 90, NUMBER 8 15 OCTOBER 2001 Selective etching of GaN polar surface in potassium hydroxide solution studied by x-ray photoelectron spectroscopy Dongsheng Li, a) M. Sumiya, b) and S. Fuke Department of Electrical and Electronic Engineering, Shizuoka University, 3-5-1 Johoku, Hamamatsu 432-8561, Japan Deren Yang and Duanlin Que State Key Laboratory of Silicon Material Science, Zhejiang University, Hangzhou 310027, People s Republic of China Y. Suzuki and Y. Fukuda Research Institute of Electronics, Shizuoka University, 3-5-1 Johoku, Hamamatsu 432-8011, Japan Received 13 February 2001; accepted for publication 20 July 2001 Etching characteristics of nondoped GaN films with the polar surface in KOH solution have been investigated. It is confirmed that the continuous etching in KOH solution takes place only for the GaN films with N-face ( c) polarity independent of the deposition method and growth condition. It is found by x-ray photoelectron spectroscopy XPS analysis for the Ga face ( c) and N-face ( c) GaN films that the atomic composition of the c surface is not changed before and after dipping in KOH solution and that on the other hand, the amount of oxygen oxide on the c surface is significantly decreased by the etching. It is also found that the band bending increases by 0.4 0.2 and 0.6 0.2 ev for the c and c surfaces after etching, respectively. This is discussed in terms of the surface chemistry. Based on the XPS result, the selective etching of the GaN polar surface is pointed out to originate from bonding configuration of nitrogen at the surface. 2001 American Institute of Physics. DOI: 10.1063/1.1402966 I. INTRODUCTION GaN has been intensively used for blue and green optoelectronic devices and high speed and power electrons. 1 3 Since wurtzite GaN has two distinct 0001 planes along the c axis, e.g., 0001 Ga ( c) and 0001 N( c) faces, the surface atomic configuration, composition, and chemical property depend on the polarity faces. It has been well known that the polarity at a GaN surface has a great effect on the growth 4,5 and device performances. 6 8 From these points of view, the determination of the polarity at the GaN surface has been one of the key issues. We investigated the GaN polarity using coaxial impact collision ion scattering spectroscopy CAICISS. 9,10 This technique was found to be reliable for determining the polarity of the material. However, this is a very sophisticated and complicated method as well as the conversion beam electron diffraction technique. 11,12 Therefore, a simple method of determining the polarity has been required. Recently, it was reported that an aqueous solution of KOH etches selectively a certain polar surface of the GaN film grown by molecular beam epitaxy MBE. On the other hand, the surface of the opposite polarity grown by metalorganic chemical vapor deposition MOCVD remains completely unaffected even by a long term dipping in the solution. 13 15 This is in agreement with the result for NaOH solution, and the polarity a Present address: State Key Laboratory of Silicon Material Science, Zhejiang University, Hangzhou 310027, People s Republic of China. b Author to whom correspondence should be addressed; electronic mail: temsumi@ipc.shizuoka.ac.jp determined by the hydroxide solution is consistent with that based on the surface reconstruction. 16 Thus, etching the GaN films in hydroxide solution can be used as a simple method to determine the polarity. However, it is still unclear whether the etching mechanism depends only upon the polarity or not and how the etching in hydroxide solution takes place. In this article, the etching behavior in KOH solution for the GaN films prepared by several kinds of growth techniques has been investigated. X-ray photoelectron spectroscopy XPS analysis was used to study the chemical states of the polar surfaces and to understand the dependence of etching property on the polarity. It is found that the selective etching behavior in the solution is associated with the polarity rather than both the deposition method and the surface morphology of the film. Based on the XPS analysis, an appropriate model is proposed to explain the mechanism of the selective etching of the GaN polar surfaces by KOH solution. II. EXPERIMENT The undoped GaN films were deposited by atmospheric pressure MOCVD, MBE, hydride vapor epitaxy HVPE, and hot wall HW epitaxy. 17 In MOCVD, the GaN films thickness: 1.2 m were deposited on nitrided and nonnitrided sapphire substrates by changing the V/III ratio in a range from 5000 to 30 000. The films were also grown homoepitaxially on GaN templates prepared on the c plane of sapphire by HVPE and MBE methods HVPE-GaN and MBE-GaN, respectively. Furthermore, the HW method was used to deposit the GaN films on the HVPE- and MBE-GaN templates. The polarity of all the samples was determined by 0021-8979/2001/90(8)/4219/5/$18.00 4219 2001 American Institute of Physics

4220 J. Appl. Phys., Vol. 90, No. 8, 15 October 2001 Li et al. TABLE I. Correlation between polarity and etching characteristics for the GaN films deposited by various techniques. Growth technique Polarity Etching property MOCVD a c inert HVPE c inert MOCVD on HVPE c c inert HW on HVPE c c inert MOCVD b V/III: 5000 30 000 c continuous etching MOCVD on MBE c c continuous etching MBE c continuous etching HW on MBE c c continuous etching a GaN film deposited on a non-nitrided sapphire substrate. b GaN films deposited on nitrided sapphire substrates. The c polarity is independent of the V/III ratio and deposition rate as mentioned in Ref. 4. The details of controlling the polarity of GaN grown by MOCVD were described in Ref. 9. c GaN films were deposited by MOCVD or HW technique on HVPE- or MBE-GaN templates. FIG. 1. Optical microscope images of as-grown: a HVPE-GaN, b MOCVD-GaN with hexagonal facets, and c MBE-GaN films. The films have c, c, and c polarity determined by CAICISS, respectively. The images of each sample after etched for 30 min are also presented in d f, respectively. CAICISS prior to dipping into KOH solution. The analysis condition of CAICISS was described elsewhere. 9 The samples were dipped in KOH (KOH:H 2 O 1:5 in weight solution between 3 and 30 min at room temperature. The surface morphology was observed with an optical microscope and an atomic force microscope AFM at various stages throughout the dipping time. In order to evaluate the change in film thickness, UV-visible spectroscopy measurements were also carried out in a range between 200 and 800 nm. XPS analysis was applied to GaN with the c polar surfaces deposited by MOCVD MOCVD-GaN before and after dipping into KOH solution for 30 min. XPS PHI 5100 ESCA System measurements were carried out using Al K radiation (h 1486.6 ev at a take-off angle 10 to probe a thin layer of the surface. Since XPS analysis was performed for the GaN films exposed to air, oxygen 1s O 1s and carbon 1s C 1s) core level peaks were always detected besides Ga 3d and N 1s. The binding energies of the spectra were referred to that of the C 1s peak 285.0 ev. III. RESULTS AND DISCUSSION A. Etching characteristics of the GaN films Figure 1 shows optical microscope images of the GaN films before and after dipping them into KOH solution. For the HVPE-GaN film with c polarity and rough surface Fig. 1 a, the change in surface morphology is not observed by the optical microscope after etching as shown in Fig. 1 d. For the MOCVD-GaN films with c polarity and smooth surface, the change was hardly detected even after dipping for 70 min not shown here except for the increase in the root mean square rms of the height measured by AFM from 4.3 to 9.3 Å. Since absorption intensity at 3.6 ev above the band gap 3.4 ev of GaN measured by a UV-visible spectrometer for c MOCVD-GaN was constant at various dipping times, the average thickness of the film is not reduced by dipping in the solution. It is confirmed that the other GaN films with c polarity, MOCVD-GaN and HW-GaN on HVPE-GaN templates are not etched by KOH solution as well. Therefore, it is concluded that the film surfaces with c polarity are resistant to the KOH solution, independent of the growth method and surface morphology. In contrast, the surface morphology is dramatically changed for the c GaN films. For c MOCVD-GaN, the hexagonal facets Fig. 1 b, which are a feature of c MOCVD-GaN, disappear after the etching for 30 min as shown in Fig. 1 e. The surface morphology of c MBE- GaN Fig. 1 c becomes rough, similar to that of the c MOCVD-GaN after etching Fig. 1 f. The results obtained for the various films are summarized in Table I. The GaN films with c polarity are etched in KOH solution, independent of the growth techniques. The details were described in Ref. 15. B. Core level XPS 1. The c GaN film Since it is confirmed that the selective etching depends on the polarity, XPS analysis was performed for the c/ c MOCVD-GaN films deposited under the same condition V/III ratio: 15 000 in order to study the chemical states of the films before and after dipping in KOH solution. Figure 2 shows O 1s a, Ga3d b, andn1s c XPS spectra for as-grown c GaN. The O 1s spectrum is decomposed into three components at 530.0, 531.5, and 533.1 ev, where decomposition parameters are employed as follows: the full width at half maximum FWHM is 1.5 the first peak and 1.8 ev the second and third peaks, and the broadening function, 100% Gaussian. The 531.5 ev peak can be ascribed to chemisorbed oxygen atoms 18 and the 533.1 ev line to hydroxide. 19 However, the chemisorbed oxygen atoms are difficult to distinguish from oxide in XPS. The 530.0 ev

J. Appl. Phys., Vol. 90, No. 8, 15 October 2001 Li et al. 4221 FIG. 2. XPS spectra of: a O1s, b Ga 3d, and c N1s for the as-grown c GaN film. The linear background lines were subtracted from the measured spectra dots and the summed spectra of the decomposed components are superimposed on the measured spectra by solid lines. FIG. 3. XPS spectra of: a O1s, b Ga 3d, and c N1s for the as-grown c GaN film. The background lines were also subtracted as well as in Fig. 2. component would be due to the chemisorbed oxygen atom, which is in different chemical states from those at 531.5 ev. The former state was found when the sample was exposed to a small amount of oxygen. 18 These oxygen species originate from contamination in air. The Ga 3d spectrum is decomposed into the two components at 19.7 and 20.7 ev, which can be ascribed to gallium nitride and oxide and/or hydroxide, 19 respectively. The parameters were employed as follows: the FWHM is 1.4 ev and the broadening function, 78% Gaussian. The N 1s spectrum is also decomposed into the three components at 397.4, 398.9, and 400.4 ev, which can be assigned to nitride, NH 2, and NH 3, 19 respectively. It is assumed that the FWHM is 1.2 the first peak and 1.8 ev the second and third peaks and the broadening function is 90% Gaussian. The latter two species were reported to be formed on the GaN surface by interaction between the surface and NH 3 gas. 20,21 The atomic composition was calculated using the relative sensitivity factor O 1s: 0.733, Ga 3d: 0.43, and N 1s: 0.499 : oxygen, 17.5%, gallium, 36.9%; nitrogen, 45.6%. After dipping in KOH solution, the 530.0 ev peak disappeared, but on the other hand the intensity of the 531.5 ev components was slightly increased not shown here. A slight increase in intensity of the 400.4 ev component was also found. The above result suggests that the chemisorbed O atoms at 530.0 ev are converted to the component at 531.5 ev and the amount of the NH 3 group at the surface increases by etching. Although the slight changes were observed in the spectral features, the surface atomic composition of the c GaN film was hardly changed upon etching in KOH solution. The close surface atomic composition before and after the etching implies that the c GaN surface is stable against the etching by KOH. This is consistent with the result of Fig. 1. It is worth noting that the binding energies for the O 1s, Ga 3d, andn1s spectra are shifted by 0.4 0.2 ev by the etching. This would be due to change in band bending for the intrinsic n-type GaN sample. The bending value is in agreement with that for GaN treated by the NH 4 OH solution. 22 Although the atomic composition is only slightly changed by the etching chemically inert in KOH solution, it should be noticed that the surface chemistry is altered, leading to change in the band bending. 2. The Àc GaN film TheO1s a, Ga3d b, andn1s c XPS spectra for the as-grown GaN sample with c polarity are displayed in

4222 J. Appl. Phys., Vol. 90, No. 8, 15 October 2001 Li et al. Fig. 3, where the fitting parameters are almost the same as described above. The surface atomic composition is calculated to be 20.5%, 43.9%, and 35.6% for oxygen, gallium, and nitrogen, respectively. The three components for the O 1s spectrum are found at 530.7, 532.2, and 533.6 ev. These can be ascribed to the same components as in Fig. 2 a, though the binding energies of the O 1s lines are higher by 0.6 0.2 ev than those for Fig. 2 a. This shift might be due to a partial charge-up effect in XPS measurements. However, since the binding energies of the Ga 3d and N 1s lines to be described later are close to those in Fig. 2, the other uncertain effects would contribute to the shift. The Ga 3d spectrum is decomposed into the three components at 18.4, 19.8, and 20.7 ev. The latter two can be ascribed to nitride and oxide hydroxide, respectively, as shown in Fig. 3 b. The 18.4 ev peak originates from metallic gallium 19 because the Ga-rich surface was reported to be formed for the c GaN film. 23 The three components at 397.6, 398.9, and 400.6 ev in Fig. 3 c can also be assigned to GaN, NH 2, and NH 3, respectively. These binding energies are in agreement with those in Fig. 2 c within 0.2 ev. TheO1s, Ga3d, andn1s XPS spectra for the etched c GaN film are shown in Fig. 4. The surface atomic composition was calculated to be 10.9%, 38.6%, and 50.5%, respectively. It is noticed that oxygen is significantly removed while the OH intensity at 533.5 ev is relatively increased by the etching Fig. 4 a. The metallic Ga component at 18.4 ev disappears. The 20.7 ev oxide/hydroxide and 19.8 ev nitride components are decreased and increased in intensity, respectively Fig. 4 b. The result on O 1s and Ga 3d suggests that the Ga metal and oxide are dissolved in KOH solution. As for N 1s spectra, the 397.6 and 400.6 ev peaks are increased and the 398.9 ev is decreased in intensity Fig. 4 c. The increase of the 19.8 and 397.6 ev peak intensity is due to exposing GaN to the surface by removing Ga metal and oxide. The decrease and increase in intensity for the NH 2 and NH 3 components, respectively, would be ascribed to conversion of NH 2 to NH 3 due to reaction between NH 2 and OH. The peak shift by etching except for O 1s is found to be 0.6 0.2 ev, which is also due to the increase in the band bending by the etching. This value is larger than that for c GaN, which would be due to a large decrease of oxygen on the surface by the etching. FIG. 4. XPS spectra of: a O1s, b Ga 3d, and c N1s for the etched c GaN film in KOH solution. The background lines were also subtracted as well as in Fig. 2. C. Mechanism of the selective etching We discuss the different etching behaviors for both surfaces using simplified ideal atomic configuration. It is well known that gallium oxide can be dissolved in alkali solution. The increase in the rms for c GaN after etching suggests that the oxide on the surface is dissolved in the solution. However, oxygen still remains on the surface after etching. This might be due to adsorption of oxygen and H 2 Oinair and solution at step edges where the Ga atoms appear on the surface after etching. Once the Ga layer is removed, the surface is converted into the nitrogen termination. Hydroxide ions in KOH solution cannot attack the N-terminated surface because of large repulsion between OH and three occupied dangling bonds of nitrogen. This is a possible reason why the c polar surface is resistant to the etching in KOH solution. In contrast, there is the single dangling bond of nitrogen atom upward on the c polar surface as shown in Fig. 5. The OH ions can attack the back bond of the Ga atoms coordinated tetrahedrally and be adsorbed on the c polar surface as shown in Fig. 5 b. This is consistent with the XPS result that the OH component is relatively increased in intensity by etching. The OH ion reacts with GaN, forming gallium oxide and NH 3 Fig. 5 c. Assuming the ideal stoichiometry, the following reaction would take place on the c GaN surface: KOH 2GaN 3H 2 O Ga2 O 3 2NH 3. 1 Here, KOH is working as a catalyst. The key point of etching for c GaN is the existence of OH ions that attack the Ga atoms in Fig. 5 b. Therefore, alkali solution KOH and NaOH can etch the c GaN surface. Once the gallium oxide is formed, it can be dissolved in KOH solution, which is consistent with the XPS result that the Ga O peak decreases in intensity in Fig. 4. The first layer of the Ga atom would be removed between stages of Figs. 5 a and 5 c, and then the surface structure would be converted into Fig. 5 d. This surface structure is the same dangling bond configuration of nitrogen as that in Fig. 5 a. The etching of the c GaN film

J. Appl. Phys., Vol. 90, No. 8, 15 October 2001 Li et al. 4223 The etching characteristics for the polar GaN surfaces in KOH solution have been investigated by an optical microscope and XPS. It is found that the c polar surface is chemically inert to the KOH solution while the c polar surface continues to be etched, independent of the deposition method and growth condition. The XPS result shows that the surface atomic composition of the c surface is not changed before and after dipping and that the amount of oxygen oxide on the c surface is significantly decreased by the etching. The hydroxide ions in KOH solution attack the Ga atoms, and gallium oxide thus formed on the c surface is dissolved in an alkali solution. Consequently, the continuous etching of the c GaN surface can take place. It is also found that the band bending increases by 0.4 0.2 and 0.6 0.2 ev for the c and c surfaces after etching, respectively. Based on the XPS data, the appropriate mechanism of the selective etching is proposed, taking the bonding configuration at the polar surface into account. The dangling bond of nitrogen is suggested to play an important role in the selective etching of the polar GaN surface. ACKNOWLEDGMENTS The author D.L. was supported by an exchange program between Zhejiang and Shizuoka University. The authors wish to thank Professor Fujinami for fruitful discussions and Dr. Tanoue for supplying them with HW-GaN samples. FIG. 5. Schematic diagrams of the cross sectional GaN film viewed along the 1 1 20 direction for c GaN to explain the mechanism of the selective etching. is continued through repeating stages a d in Fig. 5. In the case of the Ga-terminated surface, the etching would start from Fig. 5 c and be continued according to the same scenario. It should be emphasized that the dangling bond configuration of nitrogen on the surface plays an important role in the selective etching, regardless of the surface terminated by Ga or N. IV. CONCLUSION 1 S. Nakamura, M. Senoh, N. Iwasa, S. Nagahama, T. Yamada, and T. Mukai, Jpn. J. Appl. Phys., Part 2 34, L1332 1995. 2 M. W. Shin and R. J. Trew, Electron. Lett. 31, 498 1995. 3 M. Asif Kahn, A. Bhattarai, J. N. Kuznia, and D. T. Olsen, Appl. Phys. Lett. 63, 1214 1993. 4 M. Sumiya et al., J. Appl. Phys. 88, 1158 2000. 5 L. K. Li, M. J. Jurkovic, W. I. Wang, J. M. Van Hove, and P. P. Chow, Appl. Phys. Lett. 76, 1740 2000. 6 G. Nowak, S. Krukowski, I. Grzegory, S. Porowski, J. M. Baranowski, K. Pakula, and J. Zak, MRS Internet J. Nitride Semicond. Res. 1, 5 1996. 7 S. Keller, B. P. Keller, Y.-F. Wu, B. Heying, D. Kapolnek, J. S. Speck, U. K. Mishra, and S. P. DenBaars, Appl. Phys. Lett. 68, 1525 1996. 8 E. S. Hellman, MRS Internet J. Nitride Semicond. Res. 3, 11 1998. 9 M. Sumiya, M. Tanaka, K. Ohtsuka, S. Fuke, T. Ohnishi, I. Ohkubo, M. Yoshimoto, H. Koinuma, and M. Kawasaki, Appl. Phys. Lett. 75, 674 1999. 10 S. Shimizu, Y. Suzuki, T. Nishihara, S. Hayashi, and M. Shinohara, Jpn. J. Appl. Phys., Part 2 37, L703 1998. 11 F. A. Ponce, D. P. Bour, W. T. Young, M. Sauders, and J. W. Steeds, Appl. Phys. Lett. 69, 337 1996. 12 L. T. Romano, J. E. Northrup, and M. A. O Leefe, Appl. Phys. Lett. 69, 2394 1996. 13 M. Seelmann-Eggebert, J. L. Weyher, H. Obloh, H. Zimmermann, A. Rar, and S. Porowski, Appl. Phys. Lett. 71, 2635 1997. 14 J. L. Rouviere, J. L. Weyher, M. Seelmann-Eggebert, and S. Porowski, Appl. Phys. Lett. 73, 668 1998. 15 D.-S. Li, M. Sumiya, K. Yoshimura, Y. Suzuki, Y. Fukuda, and S. Fuke, Phys. Status Solidi A 180, 357 2000. 16 A. R. Smith, R. M. Feenstra, D. W. Greve, M.-S. Shin, M. Skowronski, J. Neugebauer, and J. E. Northrup, Appl. Phys. Lett. 72, 2114 1998. 17 F. Tanoue, S. Sakakibara, M. Ohbora, K. Ishino, A. Ishida, and H. Fujiyasu, J. Cryst. Growth 189Õ190, 47 1998. 18 R. A. Beach, E. C. Piqutte, T. C. McGill, and T. J. Watson, MRS Internet J. Nitride Semicond. Res. 4S1, G6.26 1999. 19 C. D. Wagner, W. M. Briggs, L. E. Davis, J. F. Moulder, and G. E. Muilenberg, Handbook of X-ray Photoelectron Spectroscopy Perkin-Elmer, Eden Prairie, MN, 1979. 20 S. Sloboshanin, F. S. Tautz, V. M. Polyakov, U. Starke, A. S. Usikov, B. Ja. Ber, and J. A. Schaefer, Surf. Sci. 427Õ248, 250 1999. 21 S. W. King, E. P. Carlson, R. J. Therrien, J. A. Christman, R. J. Nemanich, and R. F. Davis, J. Appl. Phys. 86, 5584 1999. 22 V. M. Bermudez, J. Appl. Phys. 80, 1190 1996. 23 A. R. Smith, R. M. Feenstra, D. W. Greve, J. Neugebauer, and J. E. Northrup, Phys. Rev. Lett. 79, 3934 1997.