Effective Mg activation for p-type GaN in mixed gas ambient of oxygen and nitrogen Wei Lu 1,2, David Aplin 2, A. R. Clawson 2 and Paul K. L. Yu 2 1 Zhejiang University, Zhejiang, PRC 2 Calit2, University of California, San Diego, USA Abstract The effects of mixed gas annealing ambient of O 2 and N 2 on electrical and optical characteristics of p-type Mg-doped GaN grown by MOCVD were investigated at various annealing condition. A mixed gas ambient of 80% O 2 and 20% N 2 at 750 for 2 minutes was found to be an optimum annealing condition with improvements on activation efficiency of Mg acceptor compared with annealing in pure O 2 or N 2 gas ambient. The accompanied results from Hall Effect and photoluminescence measurements of the samples were attributed to the formation of Mg-H and Mg Ga -V N complexes which caused the Mg passivation; we concluded that the existence of N 2 in the annealing ambient can only reduce but not prevent the formation of nitrogen vacancies during high temperature. Keywords: p-gan, Mg dopant GaN, dopant activation, photoluminescence 1. Introduction Since 1985, when H. Amano, et al. achieved high crystalline quality GaN (Gallium Nitride) MOCVD (metal-organic chemical vapor deposition) growth by depositing AlN (Aluminum Nitride) buffer layer technology at low temperature, GaN-based electronic and light emitting devices have made a great development during the past two decades [1]. However, the poor hole mobility and electrical conductivity of p-type GaN remained unsolved and continue to hinder the development of GaN-based electronic devices. Magnesium (Mg) is a common acceptor impurity for p-type GaN, but the as-grown Mg-doped GaN films grown by MOCVD are semi-insulating. This is attributed to the low ionization efficiency of Mg, usually less than a few percent of Mg dopants are ionized at room temperature (the ionization energy of Mg in GaN is ~160meV [2]). Moreover, there are two other mechanisms which can cause the passivation of Mg acceptors: one is self-compensation caused by the formation of a deep donor, Mg Ga -V N, namely a nearest-neighbor association between Mg acceptor and nitrogen vacancy. The nitrogen vacancy, V N, functions as a donor for GaN [3], thus the V N and Mg acceptors, which are oppositely charged, attract each other to form Mg Ga -V N [4]; the other one is the hydrogen passivation effect which results in electrically inactive Mg H complex defects in GaN during deposition [5]. A common method to increase the percentage of ionized Mg acceptors is to perform post-growth thermal annealing treatment in a hydrogen free ambient to release H from Mg-H complex. N 2 is a common gas for Mg-doped GaN thermal annealing activation, and it has been
considered to be capable of compensating nitrogen vacancies during annealing [4, 6, 7]. In this work, a systematic study in a mixed gas ambient of N 2 and O 2 is carried out. Based on our experimental data, a new explanation about the effect of N 2 was put forward. Meanwhile, some recent papers reported that O 2 ambient gives a better activation effect compared with the N 2 ambient. The conjecture is that O 2 ambient helps the hydrogen release from the Mg-H complex by forming H 2 O [8]. We investigated the effects of rapid thermal annealing (RTA) in a mixed gas ambient of O 2 and N 2 and compared with those in either pure O 2 or pure N 2 ambient. In the process, we sought for the optimum RTA condition inc luding optimum annealing time and temperature as well as the gas composition. And also the effects of RTA on photoluminescence (PL) spectrum of Mg-doped GaN were discussed. 2. Experimental Procedure All p-type Mg-doped GaN film samples were grown by Thomas Swan Close Coupled Showerhead MOCVD system [9]. 2-inch c-plane sapphire wafer was used as a substrate. Trimethylgallium (TMGa), ammonia (NH 3 ) and bis-cyclopentadienyl magnesium (Cp 2 Mg) were used as Ga, N and Mg sources, respectively. H 2 was used as the carrier gas. An undoped GaN buffer layer (~1 m) was grown on the sapphire substrate, followed by the growth of a Mg-doped GaN layer (~800nm) at 1020 C. A ~50Å-thick GaN p ++ was grown for good ohmic contact formation. After epitaxy, the samples were diced into 9mm 9mm square pieces. Next, a series of RTA treatments were performed in order to find an optimum thermal annealing condition for activating the Mg acceptors. The heating is done using infrared irradiation from halogen lamps and the temperature ramp can be as fast as 50 /s. Four dot-shape ohmic contacts were fabricated on four corners of each square sample by photolithography and depositing a 250Å-thick Ni and a 250Å-thick Au sequentially (Au on the top), followed by annealing in air ambient at 500 for 5 minutes [10]. A small ratio of contact diameter d to sample side length D (d/d=0.6mm/9mm 0.067) was used to ensure the accuracy of the Hall Effect measurement [11]. The quality of the GaN crystal was evaluated by rocking curve X-ray Diffractometry (XRD) and the rectifying property of the Au/Ni/p-GaN contacts was evaluated via I-V (current-voltage) measurements. Next, hole concentration, mobility and material resistivity were found from a series of Hall Effect measurements. For PL measurements, the samples were mounted on the cold finger of a closed cycle helium cryostat in which temperature can be controlled anywhere from 12 K to RT. A 325nm line of a 35mW He-Cd laser was used as the optical pumping source. 3. Results and Discussion 3.1 XRD and I-V measurement
The Full-Width-at-Half-Maximum of the XRD rocking curves at symmetric (0002) plane and asymmetric (101 2) plane of the p-gan are 291 arcsec and 487 arcsec, respectively, consistent with those reported in the literature [12, 13], linear I-V curves are confirmed for each contact on each sample before performing Hall Effect measurement to make sure the contacts are ohmic. 3.2 Hall Effect measurement Each series of experiments was done twice and each sample used in the same series of experiments originated from the same wafer. The hole mobility of p-type GaN used in this experiment is found to be in the range of 10~15cm 2 /V sec. A series of RTA experiments, as summarized in Fig, 1, were performed with different concentration of oxygen in nitrogen ranging from 0% to 100% at 750 for 5 minutes. The total gas flow rate is kept at 2l/min. The highest carrier concentration of 8 10 17 cm -3 and the lowest layer resistivity of 0.75Ω-cm were obtained at 80% O 2. Compared with that in pure N 2 and pure O 2, the improvement in carrier concentration is about 149% and 48% respectively, while the corresponding improvement in resistivity is 105% and 39% respectively. This result showed that the coexistence of O 2 and N 2 can be better than either in pure N 2 or pure O 2 for Mg acceptors thermal activation. We regard the formation of Mg-H complex as the dominant reason for Mg acceptor passivation, and that O 2 is more instrumental for releasing hydrogen by forming H 2 O, therefore a higher percentage of O 2 is preferred during the anneal. This is consistent with the results in the literature [8] and our experimental data: annealing in pure O 2 results in an improvement of 68% and 47% for carrier concentration and resistivity respectively, compared with in pure N 2. Furthermore, the coexistence of N 2 is also helpful in the activation of Mg acceptors as an overpressure of N 2 reduces the decomposition of GaN to form V N sites which then form Mg Ga -V N during high temperature anneals, as inferred from our following experimental results. Fig. 1. Carrier concentration and resistivity after RTA under various concentration of oxygen in nitrogen.
Using the optimum gas ambient of mixed gas of O 2 and N 2 found above, we explored the optimum annealing time and temperature. We kept the same annealing ambient of O 2 80% in N 2. Fig. 2 shows the measurement results for different annealing time from 30 seconds up to 10 minutes at 750. The maximum carrier concentration of 8 10 17 cm -3 in this series of experiments corresponds to an annealing time of 2 minutes, with increased annealing time the carrier concentration decreases before leveling off. The lowest resistivity of 0.73Ω-cm corresponds to an annealing time of 6 minutes, and then increases rapidly. During the annealing, the H released from the Mg-H complex out-diffuses to the surface of sample, and the surface desorption of H adatoms is considered to be rate controlling in the H release [14]. O 2 can promote the surface desorption of H adatoms by forming volatile H 2 O to leave the surface of sample so that more surface sites are available for the out-diffusion of H to the surface of sample. The required annealing time for optimal release of the H from Mg-H complex is about 2 minutes from our data which is roughly the same order of magnitude as the diffusion time of H in GaN (<10mins) based on a simple L 2 /D calculation (L is the thickness of the p-gan layer (~800nm), and D is the diffusion coefficient of H in GaN (>10-11 cm 2 /s [15])). And if the annealing time is longer than the hydrogen diffusion time, due to the volatility of nitrogen at high temperature, more and more nitrogen atoms will out-diffuse from GaN leaving behind nitrogen vacancies for the formation of Mg Ga -V N, which can cause the re-passivation of Mg acceptors and reduces the carrier concentration. That is to say, the existence of the N 2 in the annealing ambient possibly can only reduce but not prevent the formation of nitrogen vacancies during high temperature anneal. This inference can be supported by our following PL measurements. Fig. 2. Carrier concentration and resistivity after RTA at various annealing times. In the same token, we set the annealing time at 2 minutes and the annealing gas ambient at 80% O 2 in N 2, and find an optimum annealing temperature. Fig. 3 shows the measurement results for annealing temperature from 400 to 850 (ohmic contact can not be made for temperature higher than 850 due to the decomposition of GaN [16]). The highest carrier concentration of 8 10 17 cm -3 and the lowest resistivity of 0.63Ω-cm in this series of experiments are obtained at an annealing
temperature of 750, which is the same annealing temperature as we used in our previous series of experiments. For annealing temperature lower than 750, there is not enough thermal activation energy to break the bond of Mg-H complex efficiently, but for the temperature higher than 750, the nitrogen atoms become more volatile so that more nitrogen vacancies form at that high temperature, which re-passivate the Mg acceptors by forming more Mg Ga -V N. Fig. 3. Carrier concentration and resistivity after RTA at various annealing temperatures. 3.4 Photoluminescence measurement It has been reported that, in p-gan photoluminescence measurement, four bands of Mg-related emissions are observed.. The two bands at 3.455 and 3.27 ev are associated with the decay of excitons bound to neutral acceptor sites and the donor acceptor pair (DAP) emission due to the optical transition from a shallow donor to a shallow Mg acceptor [17, 18]. The other PL bands appear near 3.1 3.2 ev and 2.5 2.95 ev. The former is related to the optical transitions from the conduction band to the Mg acceptor[2, 19, 20]. The later, which is called the 2.8 ev band, is attributed to the optical transitions from the conduction band to a deep acceptor level or an Mg complex [17, 19-21]. This 2.8 ev band dominates at room temperature (RT), and has a characteristically broad peak in heavily Mg-doped GaN [13]. Fig. 4 shows the RT PL spectrum for an unannealed and an annealed sample at 750, for 2 minutes in 80% O 2. The main peak of the spectrum appears around 2.8eV, which is attributed to the optical transitions from the conduction band to a deep acceptor level or an Mg complex. From the figure we can see that the annealed sample shows a larger PL intensity, which probably means more Mg complex formed after annealing, and according to the RT PL peak (~2.8eV) and the bandgap of GaN (~3.4eV) we believe the Mg complex is Mg Ga -V N (deep donor level: 430meV [20]). This result is consistent with our two previous explanations for annealing gas ambient and time: the existence of the N 2 during annealing can only reduce but not prevent the formation of nitrogen vacancies at high temperature and the Mg-H complex is the dominant reason for Mg acceptor passivation (if it is not the
case, the increasing Mg Ga -V N concentration during annealing will deteriorate the carrier concentration and resistivity). Fig. 4. Room temperature PL spectrum for an unannealed and an annealed at 750, for 2 minutes, in 80% O 2 sample. Fig. 5 show the low temperature (LT: 12K) PL spectrum for an unannealed and an annealed sample at 750, for 2 minutes in 80% O 2. The PL intensity at low temperature is much larger (~100 times) than at RT. The main peak at ~3.22 ev is attributed to the zero-phonon DAP transition and the first satellite at ~3.16 ev is assigned as the first phonon replica, and the line at ~3.09 ev as the second phonon replica of the DAP transition, which is related to the optical transition from a shallow donor to a shallow Mg acceptor [17]. And the intensity of LT PL spectrum for the annealed sample is much larger than the unannealed one. This is good evidence that the passivated Mg acceptors has been activated successfully for the sample annealed in the optimum annealing condition. Fig. 5. PL spectrum at 12K for an unannealed and an annealed at 750, for 2 minutes, in 80% O 2 sample. DAP is the zero phonon donor-acceptor pair transition. DAP-1LO and DAP-2LO are the first and second phonon replicas. 4. Conclusions (will work on this later) The annealing condition using a mixed gas ambient of O2 and N2 was investigation for optimum activation of p-type Mg-doped GaN acceptors. It was found that a mixture of 80% O 2 and
20% N 2, and annealing time of 2 minutes and temperature of 750 yields the highest carrier concentration. The improvement of Mg activation efficiency is explained by the formation of Mg-H complex as the dominant reason for Mg passivation. We noted that the existence of the N 2 in the annealing ambient can only reduce but not prevent the formation of nitrogen vacancies during high temperature anneals. The demonstrated scheme of thermally activating Mg acceptors in GaN represents a detailed balance of releasing the hydrogen from Mg-H complex in GaN during annealing while minimizing the formation of nitrogen vacancies. Acknowledgements Wei Lu like to acknowledge the sponsorship of Chinese Scholarship Council for visiting UC San Diego, the work is partially supported by National Science Foundation under the program NIRT0506902. References 1. Amano, H., et al., Metalorganic Vapor-Phase Epitaxial-Growth of a High-Quality Gan Film Using an Ain Buffer Layer. Applied Physics Letters, 1986. 48(5): p. 353-355. 2. Gotz, W., et al., Activation of acceptors in Mg-doped GaN grown by metalorganic chemical vapor deposition. Applied Physics Letters, 1996. 68(5): p. 667-669. 3. Pankove, J.I. in Mater. Res. Soc. Symp. Proc.. 1990. 4. Kim, S.W., et al., Reactivation of Mg acceptor in Mg-doped GaN by nitrogen plasma treatment. Applied Physics Letters, 2000. 76(21): p. 3079-3081. 5. Nakamura, S., et al., Hole Compensation Mechanism of P -Type Gan Films. Japanese Journal of Applied Physics Part 1-Regular Papers Short Notes & Review Papers, 1992. 31(5A): p. 1258-1266. 6. Hwang, J.D., et al., Enhancing P-type conductivity in Mg-doped GaN using oxygen and nitrogen plasma activation. Japanese Journal of Applied Physics Part 1-Regular Papers Brief Communications & Review Papers, 2005. 44(4A): p. 1726-1729. 7. Hwang, J.D. and G.H. Yang, Activation of Mg-doped P-GaN by using two-step annealing. Applied Surface Science, 2007. 253(10): p. 4694-4697. 8. Wen, T.C., et al., Activation of p-type GaN in a pure oxygen ambient. Japanese Journal of Applied Physics Part 2-Letters, 2001. 40(5B): p. L495-L497. 9. E.J. Thrush, A.R.B., in: Z.C. Feng (Ed.), III-Nitride Semiconductor Materials. 2006, London: Imperial College Press. 73 116. 10. Qiao, D., et al., A study of the Au/Ni ohmic contact on p-gan. Journal of Applied Physics, 2000. 88(7): p. 4196-4200. 11. Van der Pauw, L.J., A method of measuring specific resistivity and Hall effect of discs of arbitrary shape. Philips Research Reports 1958. 13(1). 12. Youn, D.H., et al., Investigation on the P-type activation mechanism in Mg-doped GaN films grown by metalorganic chemical vapor deposition. Japanese Journal of Applied Physics Part 1-Regular Papers Short Notes & Review Papers, 1999. 38(2A): p. 631-634. 13. Jeong, T.S., et al., Room-temperature luminescence study on the effect of Mg activation annealing on p-gan layers grown by MOCVD. Journal of Crystal Growth, 2005. 280(3-4): p. 401-407. 14. Myers, S.M., et al., Influence of ambient on hydrogen release from p-type gallium nitride. Journal of Applied Physics, 2004. 95(1): p. 76-83. 15. Pearton, S.J., et al., The incorporation of hydrogen into III-V nitrides during processing. Journal of Electronic Materials, 1996. 25(5): p. 845-849.
16. Ambacher, O., et al., Thermal stability and desorption of Group III nitrides prepared by metal organic chemical vapor deposition. Journal of Vacuum Science & Technology B, 1996. 14(6): p. 3532-3542. 17. Viswanath, A.K., et al., Magnesium acceptor levels in GaN studied by photoluminescence. Journal of Applied Physics, 1998. 83(4): p. 2272-2275. 18. Ilegems, M. and R. Dingle, Luminescence of Be-Doped and Mg-Doped Gan. Journal of Applied Physics, 1973. 44(9): p. 4234-4235. 19. Smith, M., et al., Mechanisms of band-edge emission in Mg-doped p-type GaN. Applied Physics Letters, 1996. 68(14): p. 1883-1885. 20. Kaufmann, U., et al., Nature of the 2.8 ev photoluminescence band in Mg doped GaN. Applied Physics Letters, 1998. 72(11): p. 1326-1328. 21. Kaufmann, U., et al., Origin of defect-related photoluminescence bands in doped and nominally undoped GaN. Physical Review B, 1999. 59(8): p. 5561-5567.