This content has been downloaded from IOPscience. Please scroll down to see the full text.
Download details:
IP Address: 140.113.38.11
This content was downloaded on 28/04/2014 at 18:31
Please note that terms and conditions apply.
Ti/Pt/Ti/Cu-Metallized Interconnects for GaN High-Electron-Mobility Transistors on Si
Substrate
View the table of contents for this issue, or go to the journal homepage for more 2012 Appl. Phys. Express 5 066503
(http://iopscience.iop.org/1882-0786/5/6/066503)
Ti/Pt/Ti/Cu-Metallized Interconnects for GaN High-Electron-Mobility Transistors
on Si Substrate
Yueh-Chin Lin1, Tza-Yao Kuo1, Yu-Lin Chuang1, Chien-Hua Wu2, Chia-Hua Chang1, Kuan-Ning Huang1, and Edward Yi Chang1;3
1Institute of Materials Science and Engineering, National Chiao-Tung University, Hsinchu 30010, Taiwan, R.O.C. 2Department of Physical Electronics, National Chung-Cheng University, Chiayi 62102, Taiwan, R.O.C.
3Department of Electronics Engineering, National Chiao-Tung University, Hsinchu 30010, Taiwan, R.O.C.
Received March 26, 2012; accepted May 8, 2012; published online May 25, 2012
Cu-metallized interconnects for GaN high-electron-mobility transistors (HEMTs) on Si substrate using a Pt/Cu diffusion barrier layer are investigated. Auger electron spectroscopy (AES) depth profiles indicate that the GaN/Au/Ti/Pt/Ti/Cu thin metal structure is thermally stable up to 350C. The Cu-metallized devices using the proposed metal scheme exhibited DC characteristics comparable to those of conventional Au-metallized GaN devices even after annealing at 350C for 30 min. No degradation in current with time was observed when the device was tested under 28 V high-voltage stress for 24 h at room temperature. These results indicate that the Cu-metallized airbridges with the Ti/Pt/Ti/Cu diffusion barrier layer can be used for GaN HEMT fabrication. # 2012 The Japan Society of Applied Physics
G
aN high-electron-mobility transistors (HEMTs) have been widely studied due to their high break-down voltage (3.3 mV/cm), high electron satura-tion velocity (3 107cm2/s), large band gap, and high operating temperature, which are suitable for power device applications.1–3) In the conventional HEMT process, gold (Au) is usually used as the major metallization metal for interconnects. However, the price of gold is increasing. Copper (Cu) has a lower resistivity (1.67 cm for Cu; 2.2 cm for Au),4,5) a higher thermal conductivity (4.01 W cm1K1 for Cu; 3.18 W cm1K1 for Au), and a lower cost (the price of Cu is 400 times cheaper than Au), and is widely considered as a good candidate to replace gold as the metallization metal for GaN devices.However, Cu can easily diffuse into GaN and cause deterioration in device characteristics when in direct contact with the GaN material without any diffusion barrier.6–8) Therefore, an effective diffusion barrier is necessary for GaN devices to prevent Cu diffusion into the GaN material when Cu is to be used as the interconnect metal.9) Reports have shown that the high-melting-point metal Pt is an effective diffusion barrier to prevent Cu diffusion into III–V devices, and the Ti/Pt/Cu material system is very stable after annealing at 350C for 30 min.10–12)
On the other hand, reports have shown that the peeling off between Au and WNx layers was observed when using WNxas the diffusion barrier for Cu on GaAs devices, which indicates an adhesion problem between these films, a Ti addition layer was applied to the Au/WNx and WNx/Cu interfaces to solve this adhesion problem.13) Also, Ti is widely used as the adhesion layer for Schottky contact metal in the GaAs-based electronic device. In this study, Ti/Pt/Ti thin metal layers are used as the diffusion barrier for Cu airbridges for GaN HEMTs on Si substrate. The DC characteristics of the Au-metallized and Cu-metallized GaN HEMTs are compared, and the thermal stability and reliability of GaN/Au/Ti/Pt/Ti/Cu multilayer structures for Cu-metallized GaN HEMTs are evaluated in this work.
The AlGaN/GaN HEMT device structures were grown by metal organic chemical vapor deposition (MOCVD) on Si substrate. The epi layers of the device, from the Si substrate to the top, were AlN, AlGaN buffer layers, GaN channel
layer, undoped AlN spacer, AlGaN Schottky layer, and GaN cap layer. The AlGaN/GaN HEMT fabrication proc-ess consisted of Ohmic contact formation, device active region definition, gate formation, device passivation, and the airbridge process for AlGaN/GaN HEMT interconnects.
The GaN HEMT fabrication process started with Ohmic contact formation. An Ohmic metal of Ti/Al/Ni/Au was deposited on the wafer using an e-gun evaporator,14–16) the wafer was then annealed by rapid thermal annealing (RTA) at 800C for 60 s in N2ambient. The Ohmic contact resistance of3:2 106 ( cm2) was obtained by the trans-mission line method (TLM). Then, mesa regions were formed by etching in the inductive couple plasma (ICP) reactor using Cl2 in Ar ambient. After device isolation, the wafer was coated with photoresist AZ 2020 and exposed using an I-line aligner to define the gate region. Then, the wafer was dipped in 10% HCl solution to remove the native oxide on the GaN surface, and then gate metal Ni/Au was deposited by e-gun evaporation on GaN with the gate length of 1.5m. After gate formation, a 100-nm-thick Si3N4 film was deposited on GaN by plasma-enhanced chemical vapor deposition (PECVD) at 300C for device passivation.17) The contact vias of the SiN film were etched by reactive ion etching (RIE) with CF4/O2 plasmas. Finally, Pt and Ti were used as the diffusion barrier layer and adhesion layer, respectively, for the device interconnects. The thin metal structure of Ti/Pt/Ti/Cu was used as a seed layer for Cu airbridges and was deposited from the bottom to the top by sputtering.18,19) Then, the CuSO4
5H2O electrolyte was used for Cu electroplating at room temperature. The Cu airbridges fabricated on GaN HEMT devices are shown in Fig. 1.The Au/Ti/Pt/Ti/Cu multilayer deposited on the GaN blanket wafer was analyzed by Auger electron spectroscopy (AES) to evaluate the thermal stability of the material system. Au was used because it was the top layer of the Ohmic metal. The AES depth profiles are shown in Fig. 2. Figure 2(a) is the data for the as-deposited sample. Figures 2(b) and 2(c) are the data of the samples after annealing at 350 and 400C for 30 min, respectively. The data from Fig. 2 indicate that there was no obvious interatomic diffusion between Cu and Au after 350C annealing for 30 min. However, Cu and Au began to diffuse into each other through the Pt diffusion barrier layer after the
E-mail address: edc@mail.nctu.edu.tw
Applied Physics Express 5 (2012) 066503
066503-1 # 2012 The Japan Society of Applied Physics DOI:10.1143/APEX.5.066503
sample was annealed at 400C for 30 min. A Cu4Ti phase formed after the annealing process and resulted in the device deterioration. The phenomenon is the same as that reported in ref.20.
To evaluate the device performances of GaN HEMT with Cu airbridge interconnects,50 4-m-gate-width GaN HEMTs with Cu airbridges using a Ti/Pt/Ti/Cu diffusion barrier structure were fabricated. The performances of the HEMT devices with Cu airbridges were compared with those GaN HEMTs with Au airbridges. The DC characteristics of GaN HEMT with Au airbridges and GaN HEMT with Cu airbridges before and after 350C annealing are shown in Fig. 3. The performances of the Cu-metallized devices are listed below: the drain–source current (IDS) was 631 mA/mm (Au: 642 mA/mm), the
maximum drain–source current (IDSmax) was 1020 mA/mm
(Au: 1010 mA/mm), the threshold voltage (Vt) was 4:1 V
(Au:4:2 V), and the maximum extrinsic transconductance (Gm,max) was 184 mS/mm (Au: 187 mS/mm) atVDS¼ 10 V.
The Cu airbridge devices were also tested after annealing at 350C for 30 min to determine their stability at high temperature. IDS of 638 mA/mm, IDSmax of 1012 mA/mm,
Vt of4:3 V, and Gm,maxof 181 mS/mm were obtained for
the Cu-metallized HEMTs after annealing at 350C for 30 min, as shown in Fig. 3. Comparison of the Cu airbridge devices with and without annealing shows no obvious device degradation and no obvious decay of the IDS and Gm
Fig. 1. SEM image of the GaN HEMT with Cu airbridges by using Ti/Pt/ Ti/Cu multilayer diffusion barrier.
(a)
(b)
(c)
Fig. 2. AES depth profiles of the Au/Ti/Pt/Ti/Cu multilayer structure: (a) as-deposited, (b) after 350 and (c) 400C annealing for 30 min.
(a)
(b)
Fig. 3. DC characteristics of the50 4-m-gate-width Cu-metallized GaN HEMT before and after annealing at 350C for 30 min: (a)IDSversus
VDScurves, and (b)GmandIDSversusVGScurves.
Y.-C. Lin et al. Appl. Phys. Express 5 (2012) 066503
for Cu-metallized GaN HEMT was observed even after annealing. Figure 4 shows the Cu-metallized GaN HEMT with a Ti/Pt/Ti/Cu seed layer after being stressed at the high voltage of 28 V for 24 h at room temperature. Little change was observed after the stress, indicating that the material system was quite stable.
Cu-metallized GaN HEMT using a Ti/Pt/Ti/Cu metal scheme was successfully fabricated and showed electrical characteristics comparable to those of the Au-airbridged GaN HEMT. The AES depth profiles of the Au/Ti/Pt/Ti/ Cu multilayer after thermal annealing at 350C for 30 min indicate that the material system is quite stable. Also, the Cu-metallized and Au-metallized devices exhibited compa-rable DC characteristics even after 350C annealing for 30 min. In conclusion, we have demonstrated that Pt is an effective diffusion barrier for Cu-metallized interconnects on GaN HEMT. The Cu-metallized GaN HEMT device shows good thermal stability and the performances are comparable to those of Au-metallized GaN HEMT devices.
Acknowledgment The authors would like to thank the Ministry of Education and the National Science Council of the Republic of China for supporting this research under contract nos. 98-2923-E-009-002-MY3 and 100-2120-M-009-010.
1) S. N. Mohammad, A. Salvador, and H. Morkoc:Proc. IEEE83 (1995) 1306.
2) H. Morkoc, S. Strite, G. B. Gao, M. E. Lin, B. Sverdlov, and M. Burns:
J. Appl. Phys.76 (1994) 1363.
3) U. K. Mishra, P. Parikh, and Y. F. Wu:Proc. IEEE90 (2002) 1022.
4) J. D. Plummer, M. D. Deal, and P. B. Griffin: Silicon VLSI Technology (Prentice Hall, Upper Saddle River, NJ, 2000) 695.
5) J. R. Lloyd, J. Clemens, and R. Snede:Microelectron. Reliab.39 (1999) 1595.
6) V. Rajagopal Reddy and P. Koteswara Rao:Microelectron. Eng.85 (2008) 470.
7) E. R. Weber:Appl. Phys. A30 (1983) 1.
8) A. Cros, M. O. Aboelfofotoh, and K. N. Tu:J. Appl. Phys.67 (1990) 3328.
9) S. W. Chang, E. Y. Chang, C. S. Lee, K. S. Chen, C. W. Tseng, and T. L. Hsieh:IEEE Trans. Electron Devices51 (2004) 1053.
10) S. W. Chang, E. Y. Chang, D. Biswas, C. S. Lee, K. S. Chen, C. W. Tseng, T. L. Hsieh, and W. C. Wu:Jpn. J. Appl. Phys.44 (2005) 8.
11) Y. C. Wu, E. Y. Chang, Y. C. Lin, S. H. Chen, W. C. Wu, L. H. Chu, and C. Y. Chang:IEEE Microwave Wireless Components Lett.17 (2007) 133.
12) S. W. Chang, E. Y. Chang, C. S. Lee, K. S. Chen, C. W. Tseng, Y. Y. Tu, and C. T. Lee:Jpn. J. Appl. Phys.44 (2005) 899.
13) C. S. Lee, Y. C. Lien, E. Y. Chang, H. C. Chang, S. H. Chen, C. T. Lee, L. H. Chu, S. W. Chang, and Y. C. Hsieh:IEEE Trans. Electron Devices53 (2006) 1753.
14) R. Gong, J. Wang, S. Liu, Z. Dong, M. Yu, C. P. Wen, Y. Cai, and B. Zhang:Appl. Phys. Lett.97 (2010) 062115.
15) M. Piazza, C. Dua, M. Oualli, E. Morvan, D. Carisetti, and F. Wyczisk:
Microelectron. Reliab.49 (2009) 1222.
16) B. Jacobs, M. C. J. C. M. Kramer, E. J. Geluk, and F. Karouta:J. Cryst. Growth241 (2002) 15.
17) B. M. Green, K. K. Chu, E. M. Chumbes, J. A. Smart, J. R. Shealy, and L. F. Eastman:IEEE Electron Device Lett.21 (2006) 268.
18) B. Johnson, R. Amster, and L. Vanasupa:J. Electron. Mater.27 (1998) 923.
19) S. Shibuki, H. Kanao, and T. Akahori:J. Vac. Sci. Technol. B15 (1997) 60.
20) Y. C. Lien, E. Y. Chang, S. H. Chen, L. H. Chu, P. C. Chen, and Y. C. Hsieh:Appl. Phys. Lett.89 (2006) 083517.
Fig. 4. Drain–source current of the Cu-metallized GaN HEMT after being stressed at 28 V for 24 h at room temperature.
Y.-C. Lin et al. Appl. Phys. Express 5 (2012) 066503
