-GaN formed by Si implantation into p-GaN

JOURNAL OF APPLIED PHYSICS

VOLUME 91, NUMBER 4

15 FEBRUARY 2002

n ¿ -GaN formed by Si implantation into p-GaN J. K. Sheua) Optical Science Center, National Central University, Chung-Li 32054, Taiwan

C. J. Tun Institute of Optical Science, National Central University, Chung-Li 32054, Taiwan

M. S. Tsai, C. C. Lee, and G. C. Chi Department of Physics, National Central University, Chung-Li 32054, Taiwan

S. J. Chang and Y. K. Su Department of Electrical Engineering, Institute of Microelectronics, National Cheng-Kung University, Tainan 70101, Taiwan

共Received 29 May 2001; accepted for publication 9 November 2001兲 Si⫹ implantation into Mg-doped GaN, followed by thermal annealing in N2 was performed to achieve n ⫹ -GaN layers. Multiple implantation was used to form a uniform Si implanted region. It was found that the carrier concentration of the films changed from 3⫻1017 cm⫺3 共p-type兲 to 5 ⫻1019 cm⫺3 共n-type兲 when the samples were annealed in N2 ambient at 1000 °C. The activation efficiency of Si in Mg-doped GaN was as high as 27%. In addition, planar GaN n ⫹ – p junctions formed by Si-implanted GaN:Mg were also achieved. © 2002 American Institute of Physics. 关DOI: 10.1063/1.1432118兴 28

I. INTRODUCTION

II. EXPERIMENT

In the mid-1970s, Pankove et al. performed the first ion implantation study using GaN as the host material,1 and reported the photoluminescence properties of the implanted GaN with different implantation elements. In the first half of the 1990s, researches on ion implanted GaN were mostly concentrated on the realization of p-type materials. However, the ionization energy of the implanted group II elements, such as Zn, Mg, Be, and Cd, etc., in GaN are considerably large and thereby results in a low activation efficiency.2 Thus, only few successes in p-type GaN were achieved by ion implantation technique.3,4 Ion implantation is also useful to achieve heavily doped layer with a low specific contact resistance ( ␳ c ) in selected areas of GaN optoelectronic devices, especially for n-type implantation. Previous reports have shown that Si⫹ implants can be efficiently activated in n-type GaN by N2 -ambient annealing at temperatures ⭓1000 °C.5– 8 In this study, n ⫹ -GaN films were formed by multiple 28Si⫹ implantation into p-type GaN. After implantation, samples were thermal annealed in N2 ambient. Hall measurements were performed to characterize the electrical properties of the Si-implanted films. In addition, ␳ c of Ti/Al/Pt/Au Ohmic contact on n ⫹ -GaN, formed by 28Si⫹ implantation into p-type GaN, was evaluated by the transmission line model 共TLM兲. Finally, current–voltage (I – V) characteristic of the GaN n ⫹ – p junction formed by Si-implanted GaN:Mg was also evaluated. Such a GaN n ⫹ – p junction is potentially useful in the lateral GaN-based bipolar junction transistor on an insulating substrate.

In this study, Mg-doped GaN samples were first grown on c-face sapphire substrates by metalorganic vapor phase epitaxy 共MOVPE兲.9 Hall-effect measurement showed that the hole concentration and mobility of the as-grown p-type GaN was 3⫻1017 cm⫺3 and 12 cm2/V s, respectively. 28Si⫹ implantation was then performed to convert the p-type GaN into n-type GaN. Triple implantation was used to ensure a uniform Si-implanted layer. The implantation conditions 共dose/energy兲 were 2⫻1015 cm⫺2 /40 keV, 5 15 ⫺2 ⫻10 cm /100 keV, and 5⫻1015 cm⫺2 /200 keV, respectively. The overall implanted depth was about 0.6 ␮m. After implantation, the samples were capped with another GaN wafer and annealed in an N2 -ambient furnace at different temperatures 共i.e., 750–1000 °C兲 for 30 min. In contrast to previous reports,5–7 the AIN cap layer was not used and the postimplantation annealing temperature was relatively low in this study 共i.e., lower than the GaN growth temperature兲. Samples were then cut into 5⫻5 mm2 , and Ti/Al 共Ref. 10兲 dots were evaporated onto sample surface to form electrical contacts in Van der Pauw geometry for Hall effect measurements. Furthermore, the specific contact resistance ( ␳ c ) of Ti/Al/Pt/Au Ohmic contacts on these 28Si⫹ implanted n ⫹ -GaN were measured by the transmission line model 共TLM兲. The TLM patterns were formed by standard photolithography technique. The thickness of each metal layer is 30 nm, 100 nm, 50 nm, and 150 nm in sequence for Ti, Al, Pt, and Au, respectively. After the formation of TLM patterns, rapid thermal annealing 共RTA兲 in N2 -ambient with temperatures ranging from 500 °C to 650 °C was performed. Room temperature I–V characteristics of the TLM patterns and GaN n ⫹ – p junction were measured by using an HP4145B semiconductor parameter analyzer.

a兲

Electronic mail: [email protected]

0021-8979/2002/91(4)/1845/4/$19.00

1845

© 2002 American Institute of Physics

Downloaded 03 Mar 2009 to 140.116.208.55. Redistribution subject to AIP license or copyright; see http://jap.aip.org/jap/copyright.jsp

1846

J. Appl. Phys., Vol. 91, No. 4, 15 February 2002

FIG. 1. Resistivity of Si-implanted GaN:Mg as a function of annealing temperature.

III. RESULTS AND DISCUSSION

Figure 1 shows the resistivity of Si-implanted GaN:Mg as a function of annealing temperature. We should note that, prior to Si implantation, the resistivity of p-type Mg-doped GaN annealed at 750 °C in N2 ambient was around 1.5 ⍀ cm. The as-implanted samples show a near insulating property. However, the implanted samples changed from p-type to n-type after postimplantation N2 -ambient thermal annealing. It was also found that the resistivity of the implanted samples decreased as the annealing temperatures increased. The lowest resistivity we obtained was two orders of magnitude lower than those of typical p-GaN samples. Such a result indicates that the implanted Si atoms were efficiently activated after annealing. In this study, we limited our annealing temperatures to 1000 °C so as to prevent severe dissociation occurred at sample surface. Figure 2 shows the electron concentration measured at different temperatures for the 750 °C annealed and 1000 °C annealed samples. It was found that the ionization energy of the 750 °C annealed sample is about 28 meV. This value is close to the previous reported values.11–14 At high temperatures, the 750 °C annealed sample exhibited a typical thermally activated electron concentration profile. That is the electron concentration decreases as the temperature de-

FIG. 2. Temperature-dependent Hall measurement for Si-implanted GaN:Mg annealed at 750 °C and 1000 °C.

Sheu et al.

FIG. 3. Sheet electron concentration 共Ns兲 and Hall mobility of Si-implanted p-type GaN as functions of annealing temperature.

creases. However, when the temperature is lower than 150 K, the electron concentration appears to increase again, indicating the onset of the impurity band conduction. On the other hand, the electron concentration of the 1000 °C annealed sample maintained at approximately 5⫻1019 cm⫺3 even at low temperatures, is also shown in Fig. 2. Therefore, the ionization energy was almost zero for the 1000 °C annealed sample. In other words, electron concentration is independent on temperature. This phenomenon is usually observed in highly doped semiconductors since the impurity band is widened and merged with the conduction band15 in the highly doped semiconductors. Figure 3 shows the sheet electron concentration 共Ns兲 and Hall mobility of the Siimplanted GaN:Mg as a function of annealing temperature. It was found that Ns increased as the annealing temperature increased. In other words, the activation efficiency increased from 0.4% to 27% as the annealing temperature increased from 750 °C to 1000 °C. The highest Ns we obtained in this study is around 3⫻1015/cm2 , which corresponds to a carrier concentration of 5⫻1019/cm3 , when the Si-implanted GaN sample was annealed at 1000 °C for 30 min. With a 1100 °C postimplantation annealing, Zolper et al. reported that the activation efficiency of Si implantation (5 ⫻1015 cm⫺2 /100 keV) in undoped GaN can reach 35%.4 They commented that significant implantation damage remains even after high-temperature annealing, and the activation of Si atoms seems to be insensitive to implantation damages. In other words, the damaged lattice could not be effectively recovered even with an 1100 °C thermal annealing. Figure 4 shows the x-ray spectra of Si-implanted GaN:Mg samples annealed at different temperatures. For the as-implanted sample, we can clearly observed an x-ray shoulder peak originated from the Si-implanted surface layer. It was also found that high temperature thermal annealing can indeed partially recover the damaged layer as evidenced by the decrease of x-ray shoulder peak intensity. Although the sheet electron concentration increased significantly after thermal annealing, as shown in Fig. 3, x-ray data indicated that the implantation induced lattice distortion remains even for the 1000 °C annealed sample. Cao et al. commented that distorted lattice of high-dose Si-implanted GaN could only be recovered with an extremely high 1300 °C thermal



FIG. 4. X-ray diffraction spectra of Si-implanted GaN:Mg measured at 关0004兴 direction.

annealing.7 However, under such a high temperature annealing, dissociation occurred at the GaN surface will be imperative. It is well known that ion implantation would induce anisotropic strain in the lattice. Thus, high-temperature and capped thermal annealing may be needed to remove the implantation-induced damage in GaN and to prevent GaN surface from severe dissociation.16 By diffusing Si into GaN to form a heavily doped n ⫹ -GaN surface layer, Lin et al.17 have observed electron tunneling occurring at the metal/n ⫹ -GaN interface. In this study, Ti/Al/Pt/Au 共30/100/50/150 nm兲 metal contacts were deposited onto Si-implanted GaN:Mg films with an electron concentration of 5⫻1019 cm⫺3 by e-beam evaporation. After metal deposition, samples were alloyed in N2 -ambient with temperatures ranging from 500 °C to 650 °C for 5 min. Noting that all I–V curves exhibit Ohmic behavior even for the nonalloyed sample 共not shown here兲. Such a result implies that Ohmic characteristic without an apparent barrier to current flow can be achieved for the nonalloyed Ti/Al/Pt/Au contacts on the Si-implanted GaN:Mg. In other words, a significant band bending will give rise to a narrow barrier width, allowing easy passage of electrons via tunneling, and thus a good Ohmic characteristic. Specific contact resistance, ␳ c , was calculated by measuring the resistance vs the distance between TLM pattern rings. As shown in Fig. 5, it was found that ␳ c was about 5⫻10⫺5 ⍀ cm2 for the as-deposited sample. It was also found that with a proper 600 °C alloying for 5 min in N2 ambient, we can reduce the value of ␳ c down to 1.5⫻10⫺6 ⍀ cm2 . Such a reduction is attributed to the formation of an interfacial layer at the Ti/GaN interface by solid-state reaction during thermal alloying. In addition, aluminum could diffuse through the Ti layer to also react with GaN and form Al-containing interfacial compounds.18 Another possible reason for the lower ␳ c is the removal of the native oxide layer by interfacial reaction during thermal alloying. On the other hand, as the alloying temperature is higher than 600 °C, the ␳ c value increased again, as also shown in Fig. 5. Such a phenomenon is attributed to the poor adhesion between metallic contact and GaN surface since the low melting point of Al would degrade the metallic contacts on GaN at high temperatures.18

Sheu et al.

1847

FIG. 5. Specific contact resistance of Ti/Al/Pt/Au 共30/100/50/150 nm兲 Ohmic contact to Si-implanted GaN:Mg as a function of alloying temperature.

Although the mobility of Si-implanted GaN:Mg is still smaller than that of the MOVPE-grown Si-doped GaN with the same carrier concentration, the n ⫹ -GaN has been successfully created by Si implantation into GaN:Mg. Figure 6 shows the I–V characteristic of the GaN n ⫹ – p junction created by 28Si⫹ implantation into GaN:Mg, followed by postimplantation thermal annealing at 1000 °C in N2 ambient. The inset in Fig. 6 shows the schematic structure of the GaN n ⫹ – p junction. For this GaN n ⫹ – p junction diode, Ni/Au 共Ref. 19兲 and Ti/Al/Pt/Au were used as the p- and n-type electrode, respectively. It was found that the turn-on voltage and series resistance of this implanted n ⫹ – p junction are much larger than those values reported from GaN n ⫹ – p junctions prepared by normal epitaxial growth. This could be attributed to the degradation of p-type contact during the 1000 °C thermal annealing. Such a high temperature annealing would cause a severe surface dissociation, and therefore results in poor p-type Ohmic contacts. Thus, it might be necessary to implement an Al capping layer so as to improve the electrical characteristics of these GaN n ⫹ – p junctions prepared by Si ion implantation.

FIG. 6. I – V characteristic of GaN n ⫹ – p junction created by 28Si⫺ implantation into Mg-doped GaN. The inset shows the schematic structure of the GaN n ⫹ – p junction.


1848


IV. CONCLUSION

In summary, 28Si⫹ implantation has been performed to create a n ⫹ -GaN layer on p-type GaN. The carrier concentration of the films changed from 3⫻1017 cm⫺3 共p-type兲 to 5⫻1019 cm⫺3 共n-type兲 when the Si-implanted p-type GaN was properly annealed in N2 ambience. It was found that the activation efficiency of Si in Mg-doped GaN is as high as 27% when annealed at 1000 °C. Such a GaN n ⫹ – p junction is potentially useful in lateral GaN-based bipolar junction transistor on an insulating substrate. ACKNOWLEDGMENTS

The authors would like to acknowledge the financial support from the National Science Council for their research Grant Nos. of NSC 90-2215-E-008-043 and NSC 90-2112M-008-046. J. I. Pankove and J. A. Hutchby, J. Appl. Phys. 47, 5387 共1976兲. J. H. Edgar, Properties of Group III Nitrides 共INSPEC, London, United Kingdom, 1994兲, p. 273. 3 S. J. Pearton, C. R. Abernathy, C. B. Vartuli, J. C. Zolper, C. Yuan, and R. A. Stall, Appl. Phys. Lett. 67, 1435 共1995兲. 4 J. C. Zolper, H. H. Tan, J. S. Williams, Zou, D. J. H. Cockayne, S. J. 1 2

Sheu et al. Pearton, M. Hagerott Crawford, and R. F. Karlicek, Jr., Appl. Phys. Lett. 70, 2729 共1997兲. 5 H. H. Tan, J. S. Williams, J. Zou, D. J. H. Cockayne, S. J. Pearton, and R. A. Stall, Appl. Phys. Lett. 69, 2364 共1996兲. 6 C. J. Eting, P. A. Grudowski, R. D. Dupuis et al., Appl. Phys. Lett. 73, 3875 共1998兲. 7 X. A. Cao, C. R. Abernathy, R. K. Singh et al., Appl. Phys. Lett. 73, 229 共1998兲. 8 W. C. Lai, M. Yokoyama, C. C. Tsai, C. S. Chang, J. D. Guo, J. S. Chan, and C. Y. Chang, Jpn. J. Appl. Phys., Part 2 38, L802 共1999兲. 9 J. K. Sheu, G. C. Chi, Y. K. Su, C. C. Liu, C. M. Chang, W. C. Hung, and M. J. Jou, Solid-State Electron. 44, 1055 共2000兲. 10 J. K. Sheu, Y. K. Su, G. C. Chi et al., J. Vac. Sci. Technol. B 18, 729 共2000兲, and references therein. 11 I. H. Lee, I. H. Choi, C. R. Lee et al., J. Cryst. Growth 182, 314 共1997兲. 12 L. B. Rowland, K. Doverspike, and D. K. Gaskill, Appl. Phys. Lett. 66, 1495 共1995兲. 13 T. Tanaka, A. Watanabe, H. Amano, H. Kobayashi, I. Akasaki, S. Yamazaki, and M. Koike, Appl. Phys. Lett. 65, 2024 共1994兲. 14 J. Neugebauer and C. G. Van de Walle, Phys. Rev. B 50, 8067 共1994兲. 15 N. F. Mott, Metal–Insulator Transitions 共Taylor & Francis, London 1990兲. 16 E. Solkowski, G. S. Pomrenke, Y. K. Yeo, and R. L. Hengehold, Phys. Scr., T T69, 276 共1997兲. 17 C. F. Lin, H. C. Cheng, G. C. Chi, C. J. Bu, and M. S. Feng, Appl. Phys. Lett. 76, 1878 共2000兲. 18 C. T. Lee and H. W. Kao, Appl. Phys. Lett. 76, 2364 共2000兲. 19 J. K. Sheu, Y. K. Su, G. C. Chi, M. J. Jou, C. M. Chang, C. C. Liu, and W. C. Hung, Appl. Phys. Lett. 74, 2340 共1999兲.
