Ohmic Contact to High-Aluminum-Content AlGaN ... - Springer Link

1 downloads 0 Views 315KB Size Report
Sep 1, 2009 - We report on the effect of dry etching and the combination of metal stacks used to form ohmic contacts on silicon-doped high-Al-content (>60%) ...
Journal of ELECTRONIC MATERIALS, Vol. 38, No. 11, 2009

Regular Issue Paper

DOI: 10.1007/s11664-009-0924-y Ó 2009 TMS

Ohmic Contact to High-Aluminum-Content AlGaN Epilayers SURENDRA SRIVASTAVA,1 SEONG MO HWANG,1 MD. ISLAM,1 K. BALAKRISHNAN,1 VINOD ADIVARAHAN,2,3 and ASIF KHAN1 1.—Department of Electrical Engineering, University of South Carolina, Columbia, SC 29208, USA. 2.—NITEK, Inc., Irmo, SC 29063, USA. 3.—e-mail: [email protected]

We report on the effect of dry etching and the combination of metal stacks used to form ohmic contacts on silicon-doped high-Al-content (>60%) n-AlGaN layers for deep-ultraviolet light-emitting diodes. The contact characteristics are compared for as-grown and plasma-etched n-AlGaN samples. The Ti/Al/Ti/ Au contacts to as-grown n-AlGaN were linear, with a specific contact resistivity of 5 9 10 5 X-cm2. The same metallic layer combinations yielded nonlinear contacts on the plasma-etched surface of the n-AlGaN layers. However, when Ni was used as the barrier layer instead of titanium, the contacts to plasma-etched AlGaN surfaces became linear, with a specific contact resistivity of 5 9 10 4 X-cm2. Key words: Ohmic contacts, high-aluminum-content n-AlGaN, etching

INTRODUCTION High-aluminum-content AlGaN epilayers are required for a variety of optoelectronic and electronic devices made from III-nitride semiconductors, such as light-emitting diodes, laser diodes, and amplifiers. Deep-ultraviolet light-emitting diodes (DUV LEDs) based on III-nitrides are a potential replacement for current light sources such as mercury vapor lamps, which are environmentally hazardous. Semiconductor-based deep-UV LEDs are also needed for biochemical detection and air, water, and food purification. Recently several groups have reported on deep-UV LEDs.1,2 For deep-UV LEDs with peak emission wavelengths in the range of 240 nm to 300 nm, very high-Al-content AlxGa1 xN (x  0.45 to 0.75) layers are utilized. Silicon is one of the most effective n-type dopants for GaN and low-Al-percentage AlGaN. There are few unintentional dopants such as oxygen and carbon. Oxygen acts as a shallow donor for GaN and low-aluminum-content AlGaN epilayers. As the aluminum percentage is increased to more than 40%, oxygen atoms tend to occupy DX centers and act as a deep-level defect.3,4 Their activation energy increases and hence the carrier concentration in the high-Al-content AlGaN layers decreases. (Received November 24, 2008; accepted August 3, 2009; published online September 1, 2009)

2348

Additionally there are a lot of Al and Ga vacancies in high-aluminum-content n-AlGaN layers. These act as acceptors and compensate the free electron concentration, thereby reducing the electrical conduction.3,4 Due to these mechanisms, the n-type doping efficiency is much lower in high-aluminumcontent AlGaN epilayers as compared with in GaN, and hence it is very difficult to make high-current-carrying linear ohmic contacts to such AlGaN epilayers. Furthermore, to fabricate a lateral-conduction AlGaN-based DUV LED device structure grown on sapphire, it is necessary to dry-etch the epilayer structure to access the n-AlGaN layers to make the ohmic contact. The dry-etching process for high-Alcontent n-AlGaN layers uses a BCl3/Cl2/Ar plasma. Plasma etching makes it even more difficult to fabricate the ohmic contact,5 as new defects are introduced during the etching process. Many studies have been done to find the effect of plasma etching on the formation of contacts to n-GaN.6,7 However, there are very few reports on similar studies for silicon-doped n-AlGaN with an aluminum percentage of greater than 50%.8–10 Miller et al.11 had earlier reported on ohmic contact to 58% AlGaN, by increasing the annealing temperature and also by utilizing a vanadium-based metallization scheme. With increasing aluminum percentage it is widely expected that the difficulty of making ohmic

Ohmic Contact to High-Aluminum-Content AlGaN Epilayers

contacts will increase owing to the increased ionization energy and decreased mobility of carriers.3 In this paper, we report on the fabrication of ohmic contacts to n-AlGaN with an aluminum percentage as high as 68%, which to the best of our knowledge, is the highest percentage of aluminum-based AlGaN used for an ohmic contact study. This paper also discusses the effect of dry etching and the metallization schemes to form an ohmic contact after inductively coupled plasma (ICP) etching on relatively thick (greater than 2 lm) n-AlGaN layers with an Al content more than 65%. Generally a Ti/Al/Ti/Au metal scheme is used to form ohmic contacts to n-GaN12–14 and n-AlGaN.15,16 Some researchers have also investigated Pt,17 Mo,18 and Ni19–21 as the third layer. In this investigation, an effort has been made to study the effect of dry etching and the employed contact metal scheme for n-type contacts to n-AlxGa1 xN layers with an aluminum content as high as 70%. EXPERIMENTAL PROCEDURES For this experimental study, Si-doped n-AlxGa1 xN epilayers were grown by metalorganic chemical vapor deposition (MOCVD) on a 2-inch ˚ -thick AlN c-plane sapphire substrate. First, a 300-A layer was grown at low temperature (500°C). This buffer layer is followed by the growth of an AlN/AlGaN superlattice to relieve stress and to act as a dislocation filter. Next, a 2.0-lm-thick Si-doped n-AlGaN layer was grown. The carrier concentration and Hall mobility were found to be 5 9 1018 cm 3 and 60 cm2/V-s, respectively, as measured by Hall measurement.22,23 Three pieces of the same sample were taken. Sample details are given in Table I. First, all the samples were cleaned and boiled in acetone and isopropanol for 5 min. They were then dipped in 1:3 hydrofluoric acid to remove the native oxide (if any) from the surface. The acid cleaning was followed by rinsing the samples in deionized (DI) water. Samples B and C were dry etched in a BCl3/Cl2/Ar plasma environment. The flow rates were maintained at 25 sccm, 10 sccm, and 45 sccm for BCl3, Cl2, and Ar, respectively. The chamber pressure was maintained at 10 mTorr during the dry etching process and the inductively coupled plasma (ICP) radio-frequency (RF) power was set at 1000 W. After etching, rectangular transmissionline method (TLM) patterns were defined on the

Table I. Details of Various Samples Used for Experiment Sample A B C

Surface

Metal Combination

As grown Plasma etched Plasma etched

Ti/Al/Ti/Au Ti/Al/Ti/Au Ti/Al/Ni/Au

2349

dry-etched and the as-grown samples using standard photolithography. Rectangular TLM pads of dimension 50 lm 9 150 lm with spacings of 2 lm, 4 lm, 6 lm, 8 lm, 10 lm, and 12 lm were defined to extract the sheet resistance and specific contact resistivity. The contact metals were then deposited using an electron-beam evaporator. During the metal deposition, the chamber pressure was maintained at 8 9 10 7 mTorr. A Ti (40 nm)/Al(120 nm)/ Ti(40 nm)/Au(80 nm) metal stack was used for both the sample types A and B. In addition, a Ti(40 nm)/ Al(120 nm)/Ni(40 nm)/Au(80 nm) metal stack was also deposited on another plasma-treated sample (C, etched under the same conditions as before) from the same wafer. Metal deposition was followed by lift-off and cleaning procedures. All the samples were then annealed in N2 ambient at 875°C for 15 s in a rapid thermal annealing (RTA) furnace under forming gas ambient. Current–voltage (I–V) measurements were then performed for all the TLM patterns using an Agilent parameter analyzer. To analyze the effect of plasma etching, cathodoluminescence measurement was also carried out on the as-grown as well as the plasma-treated n-AlGaN epilayers. In addition, x-ray photoelectron spectroscopy was used to analyze the surface chemistry after annealing. X-ray diffraction was also used to find the reaction products formed by the metal annealing. RESULTS AND DISCUSSION Figure 1a depicts the current–voltage (I–V) characteristics of contacts fabricated on the as-grown n-AlGaN (sample A) and the n-AlGaN epilayer etched with BCl3/Cl2/Ar plasma in an ICP (sample B). Contacts for both the etched and unetched samples consisted of the Ti/Al/Ti/Au metal scheme. The metal contacts formed on the as-grown n-AlGaN sample shows good ohmic characteristics after annealing at 875°C with a specific contact resistivity of 5 9 10 5 X-cm2. However, the contacts fabricated on the plasma-etched n-AlGaN sample show nonlinear I–V characteristics after annealing at 875°C. Figure 1b shows I–V characteristics of contacts fabricated on plasma-etched n-AlGaN (sample C) in which titanium (Ti) was replaced with nickel (Ni) in the third metal layer of our n-contact stack. As seen in Fig. 1b, the Ti/Al/Ni/Au metal scheme leads to an ohmic behavior in sample C. The specific contact resistivity for these linear Ti/Al/Ni/Au ohmic contacts on the plasma-etched n-AlGaN sample C was found to be 5 9 10 4 X-cm2. It should be noted that this is only one order of magnitude lower than the specific contact resistivity for the contact fabricated on as-grown n-AlGaN. Room-temperature cathodoluminescence (CL) measurements were performed on the plasmaetched and as-grown samples to study the effect of the etching on the epilayer properties. The beam

2350

Srivastava, Hwang, Islam, Balakrishnan, Adivarahan, and Khan

Fig. 1. (a) I–V characteristics for Ti/Al/Ti/Au metal scheme on as-grown and plasma-etched sample. (b) I–V characteristics for Ti/Al/Ti/Au metal scheme and Ti/Al/Ni/Au on plasma-etched sample.

energies were varied from 10 kV to 30 kV to study the surface or the bulk regions. Figure 2 shows CL spectra of the as-grown and plasma-etched n-AlGaN epilayer. When a high beam energy (30 kV) was used in the CL study, the CL spectra were the same for both samples A and B. As seen from Fig. 2a, for the high energy, the CL spectrum consists of a nearband-edge peak at 260 nm and a deep defect level peak at around 490 nm for the as-grown and plasma-etched epilayer. When the beam energy was reduced to 10 kV, the intensity of the 490-nm parasitic peak increased in the plasma-etched sample B as compared with the as-grown sample A, as shown in Fig. 2b. These data suggest that the origin of this peak is related to defects created during the etching process. Furthermore, the defect concentration is higher near the sample surface. It should be emphasized that the plasma etching had little effect on the bulk of the n-AlGaN epilayer. We believe that the generated surface traps act as nonradiative

Fig. 2. (a) CL spectrum for as-grown and plasma-etched n-AlGaN sample at high bias voltage. (b) CL spectrum for as-grown and plasma-etched n-AlGaN sample at low bias voltage.

recombination centers for the generated electronhole pairs. Therefore, after the plasma etching, the carrier concentration available for electrical conduction decreases and the I–V curve becomes nonlinear for the contacts. Yan et al.14 have also reported nonlinearity in the the I–V characteristics because of oxide formation due to the plasma treatment. Kim et al.3 and Wu et al.20 also mentioned the presence of oxygen near the surface which affects the etch rate and the surface properties of the etched layers. To gain a better understanding of the chemistry near the surface, x-ray photoelectron spectroscopy (XPS) and x-ray diffraction (XRD) studies were carried out. From Table II, which shows the atomic concentration of different elements near the surface, it is clear that the oxygen concentration under the contact is higher in the plasma-etched samples as compared with the as-grown sample. We believe that the reason for the linear characteristics of the Ti/Al/Ni/Au contacts on the plasmaetched sample C is the formation of nitrogen

Ohmic Contact to High-Aluminum-Content AlGaN Epilayers

2351

Table II. Atomic Concentration of Elements (%) Sample A B C

Surface

Metallization Scheme

N (%)

O (%)

Ga (%)

Al (%)

Ti (%)

As grown Plasma etched Plasma etched

Ti/Al/Ti/Au Ti/Al/Ti/Au Ti/Al/Ni/Au

31.3 34.8 29.8

3.5 5.4 4.2

10.6 2.5 10.1

40.1 37.3 39.7

0.6 5.5 0.0

Fig. 3. XRD scan for contact with Ti/Al/Ni/Au metal scheme on plasma-etched n-AlGaN epilayer.

vacancies, which are well known to act as donors in the AlGaN material system.24 According to Table II, the nitrogen concentration is higher at the surface for the Ti/Al/Ti/Au contacts. This confirms our assertion that a nitrogen-deficient surface under the Ti/Al/Ni/Au contact helps in making the ohmic contacts linear by enhancing tunneling. Contacts based on Ni also show good stability at high temperature.19 This can be attributed to the formation of NiAl at the metal semiconductor interface.19 An XRD scan of the Ti/Al/Ni/Au contains a peak that corresponds to the formation of a NiAl compound at the interface which has a very high melting temperature of 1639°C.25 Ingerly et al.19 have also reported a NiAl contact to n-AlGaN, with linear electrical characteristics, good stability at high temperatures, and an oxidationresistant behavior. We then fabricated deep-UV LEDs emitting at 280 nm where the n-AlGaN layers of this study were used as the bottom n-contact layer. Two sets of devices were fabricated, only differing in the n-contact metals. For the first Ti/Al/Ti/Au metals were used, while for the other Ti/Al/Ni/Au contact metals were used. Note that, for both device types, ICP was used to access the bottom n-AlGaN layer (Fig. 3). Figure 4 presents the I–V characteristics of

Fig. 4. I–V characteristics of UV LED fabricated with Ti/Al/Ti/Au and Ti/Al/Ni/Au contacts.

these two LED types. The LED fabricated with the Ti/Al/Ti/Au-based contact shows a higher turn-on voltage of 6.2 V as compared with 5 V for the LED fabricated with the Ti/Al/Ni/Au metal stack.

2352

Srivastava, Hwang, Islam, Balakrishnan, Adivarahan, and Khan

We attribute the higher turn-on voltage of the Ti-contact-based LED to the high contact resistance. CONCLUSIONS We studied ohmic contacts to n-AlxGa1 xN (x  0.7) layers with and without plasma etching of the surfaces. Two metal schemes, Ti/Al/Ti/Au and Ti/Al/Ni/Au, were employed. The Ti/Al/Ti/Au-based contacts show linear I–V characteristics on the as-grown epilayer and nonlinear behavior on plasmaetched surfaces. The nonlinear I–V characteristics of the Ti/Al/Ti/Au contact are attributed to the damage created in the region of the epilayer under the contact. The Ti/Al/Ni/Au contact shows linear I–V characteristics on plasma-etched samples. Although the exact mechanism is not clear, our preliminary conclusion is that a nitrogen-deficient region underneath the Ni-based contact is created that aids in the formation of contacts with ohmic behavior. ACKNOWLEDGEMENTS The authors would like to thank Dr. Q. Fareed and B. Zhang for useful discussions. REFERENCES 1. V. Adivarahan, S. Wu, A. Chitnis, R. Pachipulusu, V. Mandavilli, M. Shatalov, J.P. Zhang, M.A. Khan, G. Tamulaitis, A. Sereika, I. Ilmaz, M.S. Shur, and R. Gaska, Appl. Phys. Lett. 81, 3666 (2002). 2. A. Yasan, R. McClintock, K. Mayes, S.R. Darvish, P. Kung, and M. Razeghi, Appl. Phys. Lett. 81, 801 (2002). 3. C. Stampfl and C.G. Van De Walle, Appl. Phys. Lett. 74, 459 (1998). 4. V. Adivarahan, G. Simin, G. Tamulaitis, R. Srinivasan, J. Yang, M. Asif Khan, M.S. Shur, and R. Gaska, Appl. Phys. Lett. 79, 1903 (2001). 5. H.S. Kim, D.S. Lee, J.W. Lee, T.I. Kim, and G.Y. Yeon, Vacuum 56, 45 (2000).

6. H.W. Jang, C.M. Jeon, J.K. Kim, and J.L. Lee, Appl. Phys. Lett. 78, 2015 (2001). 7. X.A. Cao, F. Ren, and S.J. Pearton, Crit. Rev. Solid State Mater. Sci. 25, 279 (2000). 8. X.A. Cao, H. Piao, S.F. LeBoeuf, J. Li, J.Y. Lin, and H.X. Jiang, Appl. Phys. Lett. 89, 082109 (2006). 9. X.A. Cao, H. Piao, J. Li, J.Y. Lin, and H.X. Jiang, Phys. Stat. Sol. 204, 3410 (2007). 10. M.A. Miller, B.H. Koo, K.H.A. Bogart, and S.E. Mohney, J. Electron. Mater. 37, 567 (2008). 11. M.A. Miller, S.E. Mohney, A. Nikiforov, G.S. Cargill, and K.H.A. Bogart, Appl. Phys. Lett. 89, 132114 (2006). 12. B.P. Luther, S.E. Mohney, T.N. Jacson, M. Asif Khan, Q. Chen, and J.W. Yang, Appl. Phys. Lett 70, 1 (1996). 13. B. Boudart, S. Trassaert, X. Wallart, J.C. Pesant, O. Yaradou, D. Theron, Y. Crosnier, H. Larhreche, and F. Omnes, J. Electron. Mater. 29, 5 (2000). 14. J. Yan, M.J. Kappers, A. Crossley, C. McAleese, W.A. Phillips, and C.J. Humphreys, Phys. Stat. Sol. (b) 241, 2820 (2004). 15. S. Ruvimov, Z. Liliental-weber, J. Washburn, D. Qiao, S.S. Lau, and P.K. Chau, Appl. Phys. Lett. 77, 2582 (1998). 16. M.W. Fay, G. Moldovan, P.D. Brown, I. Harrison, J.C. Birbeck, B.T. Hughes, M.J. Uren, and T. Martin, J. Appl. Phys. 92, 1 (2002). 17. B. Shen, H.M. Zhou, J. Liu, Y.G. Zhou, R. Zhang, Y. Shi, Y.D. Zheng, T. Someya, and Y. Arakawa, Opt. Mater. 23, 197 (2003). 18. L. Wang, F.M, Mohammed, and I. Adesida, J. Appl. Phys. 101, 013702 (2007). 19. D.B. Ingerly, Y. Chen, R.S. Williams, T. Takeuchi, and Y.A. Chang, Appl. Phys. Lett. 77, 382 (2000). 20. A.N. Bright, N. Sharma, and C.J. Humphreys, J. Electron. Microsc. 50, 489 (2001). 21. F. Roccaforte, F. Lucolano, A. Alberti, F. Giannazzo, V. Puglisi, C. Bongiorno, S. Di Franco, and V. Raineri, Superlattice. Microst. 40, 373 (2006). 22. J.P. Zhang, H.M. Wang, M.E. Gaevski, C.Q. Chen, Q. Fareed, J.W. Yang, G. Simin, and M.A. Khan, Appl. Phys. Lett. 80, 3542 (2002). 23. W.H. Sun, J.W. Yang, J.P. Zhang, M.E. Gaevski, C.Q. Chen, J.W. Li, Z. Gong, M. Su, and M.A. Khan, Phys. Stat. Sol. 7, 2083 (2005). 24. Z. Yang, L.K. Ll, J. Alperin, and W.I. Wang, J. Electrochem. Soc. 144, 3474 (1997). 25. W. Huang and Y.A. Chang, J. Phase Equilib. 19, 361 (1998).