phys. stat. sol. (b) 216, 561 (1999)
Subject classification: 61.72.Vv; 73.61.Ey; S7.14
Electrical Properties of the Si Implantation
in Mg Doped p-GaN
Wei-Chih Lai (a), M. Yokoyama (a), Chiung-Chi Tsai (b), Chen-Shiung Chang (b), Jan-Dar Guo (c), Jian-Shihn Tsang (c), and Shih-Hsiung Chan (c)
(a) Department of Electrical Engineering, National Chen Kong University, No. 1, Ta-Hsueh Road, Tainan, Taiwan
(b) Department of Electro-Optical Engineering, National Chiao Tung University, 1001, Ta-Hsueh Road, Hsinchu, Taiwan 30050
(c) National Nano Device Laboratory, 1001-1, Ta-Hsueh Road, Hsinchu, Taiwan 30050 (Received July 4, 1999)
This work performs Si ion implantation to activate and convert the electrical conduction of p-GaN films from p-type to n-type. Multiple implantation method is used to form a uniform Si implanted region in the p-type GaN epitaxial layer. Implantation energies for the multiple implantation are 40, 100, and 200 keV. The implantation dose is 5 1015cm2for each implantation energy. After
implantation, the samples were annealed in an N2 ambient for different annealing temperatures
and annealing times. The activation efficiency reaches as high as 20% when annealing the sample at 1000oC,. The carrier activation energy is about 720 meV. The low activation energy indicates that
the hopping process mechanism is the dominant mechanism for the activation of the Si implant in p-GaN. In addition, the rectifying I±V characteristic of the p±n GaN diode is also examined.
1. Introduction
III±V nitrides are of special interest for their applications as blue light sources and UV detectors [1 to 6]. In addition, this material system is attractive for use in high tempera-ture and high power electronic devices. For integrate electronic circuits, a selective dop-ing technology must be developed to define the device structure. The ion implantation technology with a well controlled doping profile is employed in Si integrated circuits, GaAs metal semiconductor field-effect transistors (MESFET) in digital integrated cir-cuits, and GaAs MESFET in monolithic microwave integrated circuits. Early studies of ion implantation of III-nitride materials focused mainly on investigating the p-type do-pants for the realization of light emitting diodes. Recent studies involving ion-implanted GaN materials have concentrated on studying implanted impurities for n-type donors, p-type acceptors and atoms for electrical isolation. In addition, ion implantation of n-type and p-type dopants in GaN has already been investigated, such as Si and O atoms used for n-type dopants [7 to 9], Mg and Ca atoms used for p-type dopants [7], and H, N, and F atoms used for electrical isolation [7, 10]. In this paper, we elucidate the electrical properties of Si implantation in p-GaN. The Hall measurement of the implanted n+ layer and the I±V characteristic of the n+±p diodes are also
2. Experiment
The p-type GaN layer used in this experiment was grown on a c-plane sapphire sub-strate using a commercial metal-organic chemical vapor deposition (MOCVD) system. During the growth, a 25 nm thick GaN buffer layer was initially grown at 525C and,
then, a 1.88 mm thick Mg-doped p-type GaN was deposited at 1050C. After growth,
the samples were annealed in a nitrogen ambient to activate the p-type dopants. The carrier concentration and Hall mobility of the p-GaN epitaxial layer were 1.5 1017cm±±3 and 11 cm2/Vs, respectively. After activating the p-type dopants, Si
im-plantation was performed in the commercial implanter system. Notably, SiF4 gas was
used as the Si source. To avoid localization of the implanted Si atoms at a fixed depth, multiple implantation with three different implantation energies was used in this study. The implantation energies were 40, 100, and 200 keV. The dose was 5 1015cm±±2 for
each implant. The projective ranges (Rp) simulated from TRIM-95 are 552, 1391, and
2898 A for the Si implant energy of 40, 100 and 200 keV, respectively. The relative projective straggles (DRp) correspond to 319, 707, and 1277 A, respectively. After the
implantation the samples were annealed in an N2 ambient atmosphere. Next, different
annealing temperatures and annealing times were used to realize the effects of thermal annealing treatment for the activation of the Si implanted samples. After annealing, the samples were first characterized by Hall measurements to obtain the room-temperature sheet carrier concentrations and Hall mobilities. The samples were then used to fabri-cate homojunction GaN n+±p diodes. The procedure is described as follows. The
sam-ple was first etched at a depth of 1.2 mm to expose the p-type contact region by using a ICP-RIE system. Ti/Al and Ni/Au were then used as the metals for n-type and p-type ohmic contacts, respectively. After metallization, the I±V characteristics were measured by the HP-4145 semiconductor parameter analyzer.
3. Result and Discussion
Fig. 1a illustrates an Arrhenius plot of the room-temperature sheet carrier concentra-tion versus the annealing temperature for the Si implantaconcentra-tion in p-GaN. The annealing time was fixed at 30 min. According to this figure, the conduction type of the samples was inverted from p-type to n-type and the sheet carrier concentration increased with increasing annealing temperature. As the sample was annealed at 750C, the activation
efficiency of the implanted Si atoms was only about 4.5%. When the annealing tem-perature was increased to 1000 C, the amount of activated Si atoms was as high as
20% of the implantation dose. Figure 1a also indicates that the activation energy of the implanted Si atoms can be fitted to be 720 meV. This energy is significantly smaller than that reported by Zopler et al. [9]. The carrier activation energy reported by Zo-pler was 6.7 eV, and the activation mechanism resulted from the substitutional diffusion process. In addition, their results also indicated that the carrier activation energy of 6.7 eV [11] was comparable to the activation energy of the intrinsic diffusion coefficient for GaAs [12]. The difference of the activation energies between 6.7 and 0.72 eV is probably due to the different implantation conditions. The implanted dose in Zopler's studies was lower than that used in this study and was only 1 1015 cm±±2, which was
lower than the critical implantation dose to amorphize the GaN layer. As the crystal structure was amorphized by the high-dose Si implant, much more defects and
disloca-tions might appear in the implanted GaN region. Therefore, when the crystal structure was amorphized, the Si atoms easily occupy a lattice site to be activated as donors dur-ing the thermal annealdur-ing treat-ment.
In addition, a single high-energy implant (100 keV) with a localized Si doping profile would result in a non-uniform implanted region, in-ducing the increase of ment errors in the Hall measure-ment. The multiple implantation used in this study supplied a uni-form Si implantation in the p-GaN layer and, in doing so, the accuracy of the Hall-effect measurement was increased more than before. In addition to the activation en-ergy, the mechanism for Si implantation in this study may also be significantly different from the study of Zopler et al. The high implantation dose in this study will result in a large amount of point defects in the implanted region. Therefore, the activation me-chanism is not only due to the substitutional process. Previous studies found the hop-ping process to be the dominant mechanism for GaAs and InP to activate the im-planted atoms. The activation energy for the hopping process ranges from 0.4 to 1.9 eV [13 to 16]. This process either places an interstitial atom on a vacancy site or breaks up a complex defect of the dopant with a neighboring vacancy. As the multiple energy high-dose implantation created a large amount of defects, the activation mechanism for Si implantation in p-GaN resembles more closely the hopping process than the substitu-tional process as the implantation dose was as high as 5 1015cm±±2.
Figure 1b depicts the relationship between the Hall mobility and the annealing tem-perature. According to this figure, the mobility ranges from 3 to 9 cm2/Vs. Although
markedly differing from the standard n-type epitaxial layer, the mobility is comparable to that of the original p-type GaN (11 cm2/Vs) since most of the atoms were not
acti-vated among the implanted region. This same figure indicates that the room-tempera-ture Hall mobility was increased when increasing the annealing temperaroom-tempera-ture up to 950C. This occurrence was attributed to the fact that the implanted crystal structure
was recrystallized at high annealing temperature. When the temperature exeeded 950C, most of the Si atoms were activated as donors. These activated Si atoms might
Fig. 1. a) Relationship between the sheet carrier concentration and anneal-ing temperature. The annealanneal-ing time was 30 min. The estimated carrier acti-vation is 750 meV. b) Relationship of the Hall mobility was kept versus the annealing temperature. The annealing at 30 min
be attributed to the increase of the impurity scattering and the reduction of the Hall mobility. Regarding the effect of annealing time on the activation efficiency, the annealing time was varied from 15 min to 1 h. Figure 2a re-veals that the carrier concentra-tion did not increase with increas-ing annealincreas-ing temperature.
Figure 2b presents the I±V curve of the Si implanted GaN homojunction diode. The turn on voltage of the p±n junction diode was 12 V. The forward current was about 30 mA at a bias voltage of 30 V. Although the I±V curve is not good enough, a rectifica-tion characteristic is obviously observed. The high turn-on vol-tage may be due to the damages induced by the high-energy implantation of 200 keV [17]. To improve the I±V charac-teristics, a low implantation energy and optimized implantation dose are required to reduce the turn-on voltage.
4. Conclusions
Using conventional thermal treatment, this study performed Si implantation in p-GaN to activate and convert the electrical conduction of GaN from p-type to n-type. A car-rier activation energy of 720 meV was estimated. In addition, the hopping process was considered and appeared to be the dominant mechanism of the activation process for the Si implantation in p-GaN. The activation of Si implantation in p-GaN heavily de-pended on the annealing temperature, and was insensitive to the annealing time. For the first time, this study has successfully fabricated GaN p±n junction diodes by using the multiple high-dose Si implantation in p-GaN. This technique is highly promising for fabricating GaN-base optical and electrical devices.
Acknowledgement The authors would like to thank the National Science Council of the Republic China (Contract No. NSC-88-2215-E009-015) and Epistar Corporation for par-tially supporting this research.
Fig. 2. a) Sheet carrier concentration, which almost did not change with the annealing time. b) I±V characteristics of the Si implanted p±n junction diode
Reference
[1] H. Morkoc, S. Strite, G. B. Gao, and M. E. Lin, J. Appl. Phys. 76,1363 (1994). [2] S. Nakamura, M. Senoh, and T. Mukai, Appl. Phys. Lett. 62, 2390 (1993).
[3] M. A. Khan, J. N. Kuznia, A. R. Bhattarai, and D. T. Olson, Appl. Phys. Lett. 62, 1768 (1993).
[4] S. C. Binari, L. B. Rowland, W. Kruppa, G. Kenlner, K. Doverspike, and D. K. Gaskill, Elec-tron. Lett. 30, 1248 (1994).
[5] M. A. Khan, M. S. Shur, J. N. Kuznia, Q. Chen, J. Burm, and W. Schaff, Appl. Phys. Lett. 66, 1083 (1995).
[6] T. P. Chow and R. Tygai, IEEE Trans. Electron Devices 41, 1481 (1994).
[7] S. J. Pearton, C. R. Abernathy, C. B. Vartuli, J. C. Zopler, C. Yuan, and R. A. Stall, Appl. Phys. Lett. 67, 1435 (1995).
[8] J. C. Zopler, R. G. Wilson, S. J. Pearton, and R. A. Stall, Appl. Phys. Lett. 68,1945 (1996). [9] J. C. Zopler, H. H. Tan, J. S. Williams, J. Zou, D. J. H. Cockayne, S. J. Pearton, M. Hagerott
Crawford, and R. F. Karlicek, Appl. Phys. Lett. 70, 2729 (1997).
[10] S. C. Binari, H. B. Dietrich, G. Kelner, L. B. Rowland, K. Doverspike, and D. K. Wickenden, J. Appl. Phys. 78, 3008 (1995).
[11] T. Y. Tan and U. Gosele, Appl. Phys. Lett. 48, 1467 (1988).
[12] S. J. Pearton (Ed.), GaN and Related Materials, Gordon & Breach, New York/London 1997. [13] N. J. Barrett, J. D. Grange, B. J. Sealy, and K. G. Stephens, J. Appl. Phys. 56, 3503 (1984). [14] K. D. Cummings, S. J. Pearton, and G. P. Vella-Coleiro, J. Appl. Phys. 60, 163 (1986). [15] S. J. Pearton, J. S. Williams, K. T. Short, S. T. Johnson, D. C. Jacobsen, J. M. Poate, J. M.
Gibson, and D. O. Boerma, J. Appl. Phys. 65, 1089 (1989).
[16] S. S. Kular, B. J. Sealy, Y. Ono, and K. G. Stephens, Solid State Electronics 27, 83 (1983). [17] H. H. Tan, J. S. Williams, J. Zou, D. J. H. Cockayne, S. J. Pearton, and R. A. Stall, Appl.
