Miki Fujimoto Article

To reckon construction and deterioration mechanisms of Ta/Ti ohmic contacts that were previously developed for p-GaN, the electrical properties of the Ta/Ti contacts, which were deposited on undoped GaN substrates and subsequently annealed in vacuum (where a slash (/) sign indicates the deposition sequence), were studied. The Ta/Ti contacts displayed good ohmic behavior after annealing at a climate of 800[degrees]C for 10 min in vacuum, although the undoped GaN substrates were used. However, downfall of the present ohmic contacts was observed during room-climate storage. These contact properties were related to those observed in the Ta/Ti contacts able on p-GaN. casino-effect measurements revealed that thin n-type conductive layers were originate to form on surfaces of both the undoped GaN and p-GaN substrates after annealing at 800[degrees]C in vacuum.

Key words: Gallium nitride (GaN), Ta/Ti obmic contact, metal/semiconductor interface

(Received October 14, 2002; accepted May 27, 2003)

presentation

The GaN-based III-nitride semiconductors have been widely used for optoelectronic devices, such as light emitting diodes and laser diodes (LDs) in blue and ultraviolet light regions. These semiconductors are also attractive to fabricate high-temperature devices and high power microwave devices because these semiconductors encompass sweeping bandgaps, high thermal conductivities, and inebriated resolution fields. Although extensive efforts enjoy been made to construct blue LDs, the lifetime of the LDs barely satisfies the engineers' provision. One of the reasons for the short life for the GaN LDs is lack of stable low friction contact to p-GaN. expansion of terribly reliable, ohmic-contact materials for p-GaN with specific contact resistivity ([rho]^sub c) lower than 10^sup -4^ [Omega] cm^sup 2^ is pressing to make the blue However, deficiency of a metal with a low energy barrier to p-GaN and acceptors with low ionization energy without hydrogen passivation hindered evolution of highly faithful, low-friction, ohmic-contact materials for p-GaN.

Annealing in a imperfect O2 ambient after the contact testimony was again found to reform the contact resistance ([rho]^sub c^). Koide et reported that annealing at temperatures of 500-600[degrees]C in a interested O2 ambient reduced the [rho]^sub c^ values of Au-based contacts (Co/Au, Ni/Au, Pt/Au, Pd/Au, and Cu/Au, where a pierce (/) sign indicates the deposition continuity) by a thing of ~3. The reason was explained to be due to the makeup of an intermediary semiconductor sheet with high impasse application by removing hydrogen atoms that bonded with Mg or N atoms in the p-GaN epilayer. Ho et willing a very thin Ni (5 nm)/Au (5 nm) contact by annealing at a temperature of 500[degrees]C in air, and the low contact resistivity of 4 x 10^sup -6^ [Omega] cm^sup 2^ was obtained. They explained the reduction of the contact resistance was due to the formation of a p-type NiO intermediate layer.

upshot of Ta/Ti ohmic contacts for p-GaN was carried by Suzuki et al.,14 which provided pc values as low as 3 x 10^sup -5^ [Omega] cm^sup 2^ after annealing at 800[degrees]C in vacuum. A possible genesis method of this low-resistant Ta/Ti contact was believed to be due to an increase of the hole application doped in the p-GaN epilayer by removing hydrogen from the p-GaN sheet because these contact metals (Ta, Ti) undo the small, negative hydride-formation enthalpy needed to divide the hydrides with Mg atoms. The low-resistance ohmic contact, by enhancing the out-distribution of hydrogen from the p-GaN, was likewise developed by using ZrN/ZrB2 contact

Although the Ta/Ti contacts provided the low contact resistances, the contacts were unstable during room-climate storage after contact The contact resistances of the Ta/Ti contacts increased piecemeal even during storage at room climate. Formation of voids close to the contact edges was again observed after injecting a high electrical In attachment, the contact resistances of the Ta/Ti contacts did not show a conventional U-shape habit on annealing temperatures. The resistances increased by annealing at temperatures lower than 600[degrees]C and reduced after subsequently annealing at 800[degrees]C. These phenomena were difficult to understand based on the preceding hydrogen out-expansion model. Thus, understanding of the contact formation contrivance is important for further improvement of the contact properties of the Ta/Ti ohmic contacts.

In this paper, we paid special recognition to the effects of putrid-temperature annealing (~800[degrees]C) in vacuum on the electrical properties of the GaN substrates, which is the first step to recognize the ohmic way of the Ta/Ti contacts prepared by annealing at a temperature of 800[degrees]C in vacuum. The resemblance of the electrical properties between the undoped and p-type GaN substrates was made using the indistinguishable Ta/Ti contact metallurgy. moreover, the TiN and TiC contacts (which do not recoil with the GaN at piercing climate) were also prepared on the p-GaN substrates by annealing at 800[degrees]C in vacuum in order to scrutinize the effect of the GaN/metal reaction on the contact properties.

EXPERIMENTAL PROCEDURES

The (0001)-oriented undoped GaN layers and Mg-doped p-type GaN epilayers were grown finally by metal-organic vapor phase epitaxy on the (11-20)-oriented [alpha]-Al^sub 2^O^sub 3^ substrates using thin A1N buffer layers. The thicknesses of the undoped and doped GaN layers were [mu]m and [mu]m, respectively. The hole concentration of the p-GaN layers was 3 x 10^sup 17^ cm^sup -3^, and the resistivity of the p-GaN epilayers (ps) was about 30 [Omega] cm. Prior to loading the substrates in a deposition chamber, the surfaces of the p-GaN epilayers were cleaned by immersing in a buffered HF resolution for 5 min, followed by a deionized water rinse for 1 min. The circular and rectangular electrode patterns were prepared on the GaN epilayers by a photolithographic system, as shown in Fig. 1a and b, respectively. Three types of the contact materials, Ta/Ti, TiN, and TiC, were packed. For the Ta/Ti contacts, Ta and Ti were deposited sequentially in an electron beam evaporator with a base pressure of 5 x 10^sup -7^ torr. The thicknesses of the Ta and Ti layers were 40 nm and 60 nm, respectively. The TiN contacts (50-nm incredible) were able by sputtering a Ti target with an Ar and N^sub 2^ mixture gas using a radio-frequency (RF) reactive magnetron sputtering technique. The TiC contacts (50-nm thick) were deposited on the GaN substrates by a RF magnetron sputtering technique using a TiC target. After lifting off the photoresists, annealing of the contacts was performed at temperatures ranging from 500[degrees]C to 900[degrees]C for a given time ranging from 10 min to 90 min in vacuum lower than x 10^sup -6^ torr. The electrical properties of these contacts were measured using a current-voltage (I-V) method and transfer quantity method (TLM) at room climate before and after annealing. For the I-V measurements, the circular patterns with an central dot of 200-[mu]m orbit and an interspacing of 8 [mu]m were primarily used. For the TLM measurements, the circular patterns with interspacing of 4 [mu]m, 8 [mu]m, 16 [mu]m, and 24 [mu]m, respectively, shown in Fig. 1a, were used to measure the specific contact resistivity ([rho]^sub c^) and the sheet resistance ([rho]^sub s^). To measure the I-V characteristics of the contacts with etched surface, the rectangular patterns shown in Fig. 1b were used. The engraving of the p-GaN channel surface was performed by selectedarea Ga ion bombardment using a focused ion beam (FIB) technique. The foyer-effect measurements were performed at room temperature to determine conduction type and carrier concentrations of the GaN layer before and after annealing.

tentative RESULTS

Contact Properties of Ta/Ti Contacts on Undoped GaN

The electrical properties of the Ta (60 nm)/Ti (40 nm) contacts deposited on the undoped GaN substrates and subsequently annealed in vacuum at temperatures ranging from 500[degrees]C to 800[degrees]C for 10 min were measured. Although the GaN substrates used in this experiment had inherent n-type carriers, the present Hall-effect measurements confirmed that the carrier density of the undoped GaN was extremely small to measure accurately.

Figure 2 shows the prevalent I-V characteristics between the Ta/Ti contact pads (measured at room temperature) after annealing at discrete temperatures. The I-V characteristics display rectifying bearing when annealed at temperatures lower than 700[degrees]C. However, the I-V characteristics show ohmic behavior when annealed at a temperature of 800[degrees]C. likewise, the current between the two pads at a given voltage increased with increasing annealing time at 800[degrees]C, resulting in a reduction of the total friction between the pads. The TLM measurements were performed for the contacts annealed at 800[degrees]C for 10 min and 20 min, respectively. The specific contact resistance ([rho]^sub 2^) and the GaN channel resistance ([rho]^sub s^) were measured to be x 10^sup -4^ ohm cm^sup 2^ and x 10^sup 5^ ohm/[white square], respectively, for the contact annealed at 800[degrees]C for 10 min. After annealing at 800[degrees]C for 20 min, the [rho]^sub c^ and [rho]^sub s^ values of the contact were reduced to the values of x 10^sup -5^ ohm cm^sup 2^ and x 10^sup 5^ ohm/[white square], respectively, indicating improvement of the contact properties. These contact resistances are close to those obtained beforehand14 in the Ta/Ti contacts prepared on p-GaN.

The strength of the electrical properties of the Ta/Ti contacts that were prepared by annealing at 800[degrees]C for 10 min was investigated by measuring the change of the resistance, R^sub 0^, between the pads during room-temperature storage. The R^sub 0^ values, which represent the contact resistances, were obtained by measuring the currents between the pads at the applied voltage of + or V. Figure 3 shows the change of the R^sub 0^ values of the Ta/Ti contacts during room-temperature storage. The R^sub 0^ values of the as-prepared sample were measured just after the sample was removed from the chamber, and the R^sub 0^ values of the identical sample were again measured during room-climate storage. The R^sub 0^ values of the Ta/Ti contacts are observed to increase with increasing storage time at room temperature. This weakness during room climate of the electrical properties of the Ta/Ti contacts able on the undoped GaN is similar to that observed already in the Ta/Ti contacts on p-type

The undoped GaN contains extremely negligible amounts of intrinsic carriers before annealing. However, in the present experiment, the density of the carriers in the GaN substrates was observed to increase with increasing annealing climate. To measure the carrier frequency, the gallery-effect measurement was carried out for undoped GaN annealed at 800[degrees]C for 10 min. The conduction type of the GaN substrate was determined to be n-type, but the carrier density could not be measured accurately because of the electrical weakness at room climate. The sheet carrier density and the sheet friction were measured to be ~3 x 10^sup 12^ cm^sup -2^ and ~ x 10^sup 4^ ohm/[silver square], respectively, for the undoped GaN substrate annealed at 900[degrees]C for 10 min. The success of the Hall measurement was due to advancement of the electrical strength of the GaN upon annealing at higher temperature.

It was reported that the Schottky hindrance height of Ti to n-GaN is As the work function of Ta is almost equivalent to that of Ti ( eV), we tacit that the Schottky impediment height of the Ta/Ti contacts to n-GaN was .6 eV. Using Yu's equation,21 the carrier frequency of the n-GaN substrate was calculated to be higher than 10^sup 18^ cm^sup -3^ to give a contact resistance of less than 10^sup -4^ ohm cm^sup 2^. Therefore, a surface layer thinner than 30 nm from the top of the GaN substrate was believed to be eminently conductive after annealing at 800[degrees]C. If the carrier density was uniform throughout the outright undoped GaN substrates (down to 2[mu]m deep), the carrier density should be as low as x 10^sup 16^ cm^sup -3^, which was calculated by the sheet carrier density (=3 x 10^sup 12^ cm^sup -2^) measured in the sample annealed at 900[degrees]C divided by the thickness (=2 [mu]m).

Figure 4 shows the schematically perpendicular conformation of the undoped GaN annealed at 800[degrees]C deduced from the present results. It was found that the undoped GaN substrate annealed at 800[degrees]C formed an n-type conductive sheet near the surface. This conductive sheet produced the Ta/Ti ohmic contacts prepared on the undoped GaN.

Contact Properties of Ta/Ti Contacts on p-GaN

We able the Ta/Ti contacts on the p-type GaN, which had a impasse density of 3 x 10^sup 17^ cm^sup -3^, to investigate whether the current transport mechanism of the Ta/Ti contact on p-GaN is the same proposed for the contact prepared on the undoped GaN.

The Hall-effect determination was carried out for the p-GaN substrate, which was annealed at 800[degrees]C. To measure the electrical properties near the GaN surface (

The I-V characteristics were measured using the rectangular pattern shown in Fig. 1b. For the Ta/Ti contacts annealed at 800[degrees]C for 20 min, the I-V measurements between the two pads showed linear behavior, as shown in curve (a) of Fig. 5. Then, the GaN surface of the channel domain between the contacts was etched to a intensity of [mu]m from the surface by the FIB technique to measure the thickness of the n-type conductive layer in the p-GaN substrate. The scanning ion micrograph of the cross section of the etched sample is shown in Fig. 6, where the etched neighborhood is shown by a dashed queue. After etching, the sample was dipped into a HCl:H2O = 1:1 clarification for l min to remove the Ga droplets, which jurisdiction be formed on the p-GaN surface during FIB etching. The membrane friction of the Ta/Ti metal did not change within experimental errors after immersing in the weak HCl solution. The I-V characteristics of the etched sample measured at room climate is shown in curve (b) of Fig. 5. The I-V curve shows rectifying guise, and the current at an applied voltage of 5 V was only 65 [mu]A. In this rectangular pattern, the domain bombarded by Ga ions during FIB treatment was limited to the channel territory exposed to the GaN surface to minimize the damage below the Ta/Ti contact area. Although we could not avoid the effect of GaN surface damage by the FIB treatment, the change of the I-V deportment by etching the GaN channels was believed to be due to removal of the conductive layer, which was estimated to be ~30 nm. The present provisional result indicated that the n-type conductive sheet formed near the GaN surface influenced significantly the I-V presence of the Ta/Ti contact willing on the p-GaN as well as the undoped GaN substrate.

To avoid the effect of reception between the Ta/Ti contact and the GaN on the I-V behaviors, the TiN and TiC contacts, which do not boomerang with GaN at 800[degrees]C, were deposited on the p-GaN and annealed at 800[degrees]C for 40 min in vacuum. Figure 7 shows the I-V characteristics of these three samples measured immediately after annealing. The I-V characteristics of both the TiN and TiC contacts display good ohmic morals. atrophy was also observed for these contacts. The reduction of the contact resistances was obtained by increasing the annealing time at 800[degrees]C. This experiment showed that the acknowledgment of the contact and the GaN surface had little effect on the I-V performance.

DISCUSSION

Formation of the n-Type Conductive sheet on the Undoped GaN Substrate

The present I-V and Hall-effect measurements indicated that the n-type conductive layer was formed on the GaN surface after annealing at 800[degrees]C in vacuum. The formation apparatus of the n-type conductive layer on the GaN surface will be discussed here. To prepare the Ta/Ti contact on both the undoped and p-type GaN substrates, which display excellent ohmic deportment, the combination of alpine-climate ((greater than over equal to)800[degrees]C) and high vacuum annealing, was essential. It was reported that the GaN luminous is unstable because of easy dissociation of nitrogen atoms from the GaN at high Madar et construct that the dissociation pressure of the GaN increased promptly with increasing annealing temperature and reached 1 atm at a temperature of around 840[degrees]C. Thus, the GaN crystal would dissociate upon annealing at 800[degrees]C in the vacuum whose condition was used in the present experiment to prepare the Ta/Ti ohmic contacts. Extensive experiments on injury of N from GaN, AlN, InN, InGaN, and InAlN were carried out by Vartuli et and ingrained a loss of N from these surfaces by Auger electron spectroscopy.

The liability of formation of the n-type conductive sheet by dissociation of nitrogen will be discussed here. We deem that the nitrogen dissociation created nitrogen vacancies in the GaN. It was reported that the N vacancies in GaN acted as the We consider that the nitrogen vacancies formed by the dissociation transformed the undoped or p-type GaN surface layer into the heavily doped n-type sheet, resulting in enhancement of electron tunneling through the metal/n-GaN interface. Therefore, flux of the Schottky to ohmic behavior of the Ta/Ti contacts was observed upon heating the undoped GaN. Based on this model, the deterioration mechanism of the Ta/Ti contact is interpreted to be caused by a reduction of density of the N vacancies in the GaN. The vacancies formed during 800[degrees]C annealing were supersaturated at room climate, and thus, the vacancy frequency would be reduced close to that estimated by the thermodynamic equilibrium. Therefore, the reduction of the carriers occurred during room-temperature storage.

If the nitrogen atoms were dissociated to destroy the stoichiometry of the GaN crystalline, noncontinuous Ga islands might be formed on the GaN surface. It is familiar that the electrical properties of island-type metal films are different from those of continuous Because the electrical conductivity of the island-type films is strongly dependent on the interdistances between the neighbor islands, the Ta/Ti contacts annealed at 800[degrees]C were dipped into a HCl solution (which was reported to impress Ga) for 1 min to exclude a Ga metal sheet that might be formed on the GaN surface. The difference of the I-V characteristics of the contacts was not observed before and after HCl engraving. accordingly, we believe that the current flew into the n-type GaN surfaces (which were formed by the dissociation of nitrogen atoms) rather than in the noncontinuous Ga film.

Formation of the n-Type Conductive Layer on the p-GaN Substrate

The n-type conductive layer was also found to be formed near the p-GaN surface after annealing in vacuum. The alteration of the conduction type close to the GaN surface upon heating to 800[degrees]C is schematically shown in Fig. 8. In the p-type GaN, p and n carriers coexist with densities of 3 x 10^sup 17^cm^sup -3^ and 1 x 10^sup 16^ cm^sup -3^ (estimated roughly) before annealing, respectively. After annealing the p-GaN at temperatures lower than 600[degrees]C, the carrier densities do not change significantly. However, the n-type carrier density rapidly increases with increasing annealing temperature in vacuum, and the n-type carrier frequency becomes close to the p-type carrier density at the temperature T^sub c^ marked in Fig. 8. Upon subsequent annealing at higher temperatures, the p-type carrier density may increase because of the activation of the carriers. However, the amounts of the increase in the n-type carriers are believed to be much larger than that of the p-type carriers because of the dissociation of the GaN surface. accordingly, the effective carrier concentration, Nefr, which is defined by N^sub eff^ = |N^sub h^ - N^sub e^|, near the p-GaN surface is close to N^sub e^, and the p-type conduction changes to an n-type one. This model agrees well with the result where annealing at 900[degrees]C improved the ohmic-contact properties because the annealing at higher temperature increased the N^sub e^ by further dissociation. It also explains the erstwhile result15 that the contact resistances of the Ta/Ti contacts increased by annealing at a climate of around 600[degrees]C and again decreased upon subsequent annealing at higher temperatures, as shown schematically by a dotted curve in Fig. 8. The dependence of the contact resistances of the Ta/Ti contacts on the annealing temperature indicates that the resistances strongly hang on the practical carrier frequency in the GaN substrates. The contact resistance has the highest stature at the annealing climate, T^sub c^, because the N^sub eff^ value is minimum at T^sub c^.

Based on the I-V measurements of the contacts able on p-GaN, the possible current transport paths are schematically shown in Fig. 9. During annealing at 800[degrees]C, nitrogen atoms located close to the GaN surface dissociate from the uncovered channel terrain, as shown in Fig. 9a, and the GaN, which immediately contacts the Ta or Ti contact, will react with the nitrogen atoms in the GaN to form TaN or TiN, which was previously confirmed by x-ray diffraction Therefore, the GaN surface sheet would lose crystalline stoichiometry, and the enrichment of the Ga atoms would result in formation of the n-type conductive layer. again, two current transfer paths would be formed, as shown in Fig. 9b: the n-sheet and p-layer paths. These resistances are expressed as R^sub total(n)^ = R^sub s^(n) + 2R^sub c^(n) and R^sub total(p)^ = R^sub s^(p) + 2R^sub c^(p) for the n- and p-layers, respectively. The total resistance (R^sub total^) measured between the contact pads after annealing at 800[degrees]C is disposed by

R^sub total^ = (R^sub total(n)^ [middot] R^sub total(p)^)/(R^sub total(n)^ + R^sub total(p)^)

To find out the extensive current aisle, the GaN surface was etched away, as shown in Fig. 6. The I-V measurement of the contacts with the etched surfaces showed rectifying behavior. The result indicated that R^sub c^(p) is much higher than the R^sub c^(n) esteem. accordingly, the resistances measured for the Ta/Ti contacts, which were packed by annealing at 800[degrees]C in vacuum, render the contact resistance to n-type GaN. This upshot is consistent with the result of the TiN contact, which provided the low contact friction for

CONCLUSIONS

To understand the constitution machinery of the Ta/Ti ohmic contacts to p-GaN, the effects of high-climate annealing of the GaN substrates in vacuum were premeditated by measuring the electrical properties of the Ta/Ti contacts prepared on the undoped and p-type GaN substrates. The Ta/Ti contacts on the undoped GaN substrates, packed by annealing at 800[degrees]C in vacuum, displayed good ohmic performance. Also, an increase of the contact friction was observed during room-temperature storage, as was already observed in the Ta/Ti contacts on p-GaN. The passage-effect size revealed that the n-type conductive sheet was formed near the p-GaN surfaces by annealing at 800[degrees]C in vacuum. By removing the n-type conductive layer, the I-V characteristic of the Ta/Ti contact showed rectifying guise. It gimmick that the contact resistance of the Ta/Ti contacts on p-GaN was very high, although the resistance on n-GaN was very inconsequential. We considered that the n-type conductive layer was formed because of the dissociation of the GaN substrate by tipsy-temperature annealing in vacuum.

ACKNOWLEDGEMENTS

This duty was partially supported by a Grant-in-Aid for Scientific Research from the Ministry of civilization (Grant No. 13450287).

REFERENCES

1. M. Murakami and Y. Koide, Crit. Rev. real State Mater. Sci. 23, 1 (1998).

2. H. Ishikawa, S. Kobayashi, Y. Koide, S. Yamasaki, S. Nagai, J. Umezaki, M. Koike, and M. Murakami, J. Appl. Phys. 81, 1315 (1997).

3. . Lee, . Kim, . Lee, . Park, and T. Kim, Solid-State Electron. 43, 435 (1999).

4. . Lee and . Kim, J. Electrochem. Soc. 147, 2297 (2000).

5. . Kim, . Bae, . Kim, . Baik, . Myoung, and . Lee, J. Electron. Mater. 30, 183 (2001).

6. . Kim, . Lee, . Lee, . Park, and T. Kim, J. Vac. Sci. Technol. B 17, 497 (1999).

7. . Kim, . Kim, . Cho, . Je, . Kwak, . Park, and . Lee, J. Electron. Mater. 30, 170 (2001).

8. . Jang, . Park, and . Seong, J. Vac. Sci. Technol. B 17, 2667 (1999).

9. C. Huh, . Kim, . Kim, . Kim, and . Park, Appl. Phys. Lett. 78, 1942 (2001).

10. Y. Koide, T. Maeda, T. Kawakami, S. Fujita, T. Uemura, N. Shibata, and M. Murakami, J. Electron. Mater. 28, 341 (1999).

11. . Ho, . Jong, . Chiu, . Huang, . Chen, and . Shih, Appl. Phys. Lett. 74, 1275 (1999).

12. . Ho, . Jong, . Chiu, . Huang, . Shih, . Chen, . Chen, and . Kai, J. Appl. Phys. 86, 4491 (1999).

13. . Jang, . Kim, . Kim, . Hwang, . Yang, . Lee, . Son, and . Lee, Appl. Phys. Lett. 79, 1822 (2001).

14. M. Suzuki, T. Kawakami, T. Arai, S. Kobayashi, Y. Koide, T. Uemura, N. Shibata, and M. Murakami, Appl. Phys. Lett. 74, 275 (1999).

15. M. Suzuki, T. Arai, T. Kawakami, S. Kobayashi, S. Fujita, Y. Koide, Y. Taga, and M. Murakami, J. Appl. Phys. 86, 5079 (1999).

16. L. Zhou, W. Lanford, . Ping, I. Adesida, . Yang, and A. Khan. Appl. Phys. Lett. 76, 3451 (2000).

17. . Jang, . Park, and . Seong, Appl. Phys. Lett. 76, 2898 (2000).

18. E. Kaminska, A. Piotrowska, A. Barcz, D. Bour, M. Zielinski, and J. Jasinski, Mater. Sci. Eng. B 82, 265 (2001).

19. T. Arai, H. Sueyoshi, Y. Koide, M. Moriyama, and M. Murakami, J. Appl. Phys. 89, 2826 (2001).

20. . Binari, . Dietrich, G. Kelner, . Rowland, K. Doverspike, and . Gaskill, Electron. Lett. 30, 909 (1994).

21. .C. Yu, Solid-State Electron. 13, 239 (1970).

22. . Thurmond and . Logan, J. Electrochem. Soc. 119, 622 (1972).

23. R. Madar, G. Jacob, J. Hallais, and R. Fruchart, J. Cryst. Growth 31, 197 (1975).

24. . Lorentz and . Binkowski, J. Electrochem. Soc. 109, 24 (1962).

25. J. Karpinski, J. Jun, and S. Porowski, J. Cryst. Growth 66, 1 (1984).

26. R. Groh, G. Gerey, L. Bartha, and . Pankove, Phys. Status Solidi (a) 26, 353 (1974).

27. . Vartuli, . Pearton, . Abernathy, . MacKenzie, . Lambers, and . Zolper, J. Vac. Sci. Technol. B 14, 3523 (1996).

28. . Maruska and . Tietjen, Appl. Phys. Lett. 15, 327 (1969).

29. M. Ilegems and . Montgomery, J. Phys. Chem. Solids 34, 885 (1973).

30. . Jenkins and . Dow, Phys. Rev. B 39, 3317 (1989).

31. J. Neugebauer and . Van de Walle, Phys. Rev. B 50, 8067 (1994).

32. K. van Steensel, Philips Res. Rep. 22, 246 (1967).

33. S. Miller and . Holloway, J. Electron. Mater. 25, 1709 (1996).

34. . Dimitriadis, T. Karakostas, S. Logothetidis, G. Kamarinos, J. Brini, and G. Nouet, valid-State Electron. 43, 1969 (1999).

35. Y-F. Wu, . Jiang, . Keller, S. Keller, D. Kapolnek, . Denbaars, . Mishra, and B. Wilson, Solid-State Electron. 41, 165 (1996).

IPPEI FUJIMOTO,1 HIROKUNI ASAMIZU,1 MASAHIRO SHIMADA,1 MIKI MORIYAMA,1 NAOKI SHIBATA,2 and MASANORI MURAKAMI1,3

of Materials information and Engineering, Kyoto University, Kyoto 606-8501, Japan. Department, Optoelectronics, Toyoda Gosei Co., Ltd., Aichi 452, Japan. : ..jp