3.4 The second major steps of the AlGaN/GaN HEMTs fabrication
3.4.4 Electroplating
The purpose of electroplating is used to connect the source pads of the GaN HEMTs devices.
Au electroplating：
Before electroplating, the top Ti layer was first removed in a diluted HF solution, which uncovered the bottom layer Au. The Au metal was used to conduct the planting current through the whole wafer. The wafer was then electroplated with gold to a thickness of 2μm.
The current density of the Au electroplating was 0.3 A/dm2 and the plating time was 9 minutes.
23
Cu electroplating：
The wafer was cleaned before plating to prevent surface contamination. The wafer was dipped in the CuSO4．5H2O solution of Cu plating bath for 5 second. The current density of the Cu electroplating was 2A/dm2 and the plating time was 6 minutes for 2.5 μm thick Cu as shown in Fig.3-12.
3.4.5 Second photoresist removal and thin metal etching
After electroplating, the samples were blank exposed without mask in I-line aligner to decompose the photoresist. And then the samples were immersed into developer (FHD5) to remove the second photoresist.
Next, the thin metal structures were etched by wet chemical etching.
For Au airbridge sample：
The thin metal structure used for Au airbridge was Ti/Au/Ti. The top and bottom Ti layers were etched by diluted HF for 60 seconds. The etching of Ti layer stopped at the underlying Au layer and the color turned from a gray to gold color. The thin layer of Au was etched in KI/I2 solution for 60 seconds.
For Cu airbridge samples：
The thin metal structures used for Cu airbridges were Ti/WNx/Ti/Cu and Ti/Pt/Ti/Cu, which were deposited from bottom to top. The top layer of the thin metal is Cu. It was etched by mixed solution of H2SO4：H2O2：H2O = 5：6：100 for 7 seconds. The etching rate was very high, and the etching of Cu stops at the underlying Ti as the color turned from yellow to grey.
The layers Ti was etched in a dilute HF solution with HF: H2O ratio of 1：100. The etching rate was about 5Å /s. HF is the active ingredient in this etchant, so it also etches oxides.
24
Raising the ratio of HF in the solution increases the etching rate. Ti was readily oxidized, so it was likely to form an oxide layer from the water, which was easily etched by the HF in this solution resulting in the formation of bubbles of oxygen. The diffusion barrier of WNx was etched in H2O2:H2O = 2：1 solution for 5 minutes.
3.4.6 First photoresist removal
The samples were dipped in ACE for 30 minutes and IPA for 5 minutes to remove the first photoresist. The remaining photoresist residuals were stripped by ICP. The airbridge of the GaN HEMT device was success fabricated as shown in Fig.3-13(a) & (b).
25
Figure
Fig. 3- Cross section of the AlGaN/GaN HEMT
26
Fig.3-2 The first part of the AlGaN/GaN HEMTs fabrication
Ohmic contact formation
Active region definition (mesa isolation)
Gate formation
Device
passivation
27
f f s f s s s s s f Fig.3-3 The GaN epitaxial wafer
Fig.3-4 Ohmic contact formation (The source and drain electrodes were formed)
28
Fig.3-5 Active region definition (mesa isolation)
Fig.3-6 Gate formation
29
Fig.3-7 Passivation and nitride via (define the contact via for interconnection)
30
Fig.3-8 The second part of the AlGaN/GaN HEMTs fabrication
The first photolithography for plating vias
Thin metals deposition
The second
photolithography for plating areas
Electroplating
Second Photoresist removal and thin metal etching
First Photoresist removal
31
Fig.3-9 The first photolithography for plating vias
Fig.3-10 Thin metal deposition
32
Fig.3-11 The second photolithography for plating areas
Fig.3-12 Electroplating
33
Fig.3-13(a) The top view of the airbridge of the GaN HEMTs device.
Fig.3-13(b) The side view of the airbridge of the GaN HEMTs device.
34
Chapter 4
Results and discussion
4.1 Ohmic contact resistance measurement
The transmission line method was widely used to determine the specific resistance [21].
The TLM pattern is shown in Fig.4-1, in this study, the distances between TLM electrodes are 3 µm, 5 µm, 10 µm, 20 µm, and 36um, respectively. The resistance between the two adjacent electrodes can be plotted as a function of the spaces between electrodes. The resistance is expressed by the following equation (1)
W
the data to L=0, one can calculate the value for the term Rc. The plot is shown in the Fig.4-2.The specific contact resistivity ρ c (Ω -cm2) is defined by (2)
35
4.2 Thermal stability of the thin metal
Two different thin metal structures (Ti/Pt/Ti/Cu and Ti/WNx/Ti/Cu) were used to fabricate Cu-metalized Airbridges. The top layer of the Ohmic and gate metals was Au. The depth profiles of the Au/Ti/Pt/Ti/Cu and Au/Ti/WNx/Ti/Cu multilayer deposited on GaN blanket wafer were analyzed by Auger electron spectroscopy (AES). The AES depth profile analysis was used to evaluate the thermal stability of the material systems.
4.2.1 Study of material inter-diffusion of Au/Ti/Pt/Ti/Cu multilayer structures by AES
Figure 4-4(a)~(c) show the AES depth profiles of the as-deposited Au/Ti/Pt/Ti/Cu multilayer structure and after 350℃ and 400℃ annealing for 30 minutes. From the results of these profiles, in Fig.4-4(b), there was no inter-atomic diffusion between Cu metal and Au metal after 350℃ annealing for 30 minutes. However, after the multilayer structure was annealed at 400℃ for 30 minutes, as shown in Fig.4-4 (c), Cu and Au began to inter-diffuse into each other through the diffusion barrier layers. These results indicate that Ti/Pt/Ti thin metal structure was thermally stable up to 350℃ annealing for 30 minutes.
36
4.2.2 Study of material inter-diffusion of Au/Ti/WNx/Ti/Cu multilayer structures by AES
Fig.4-5(a) ~ (c) show the AES depth profiles of the Au/Ti/WNx/Ti/Cu multilayer structure as-deposited and after 400℃ and 450℃ annealing for 30 minutes. From the results of the profiles, in Fig.4-5(b) shows there was no inter-atomic diffusion between Cu metal and Au metal after 400℃ annealing for 30 minutes. However, after the multilayer structure was annealed at 450℃ for 30 minutes, Cu and Au started to inter-diffuse into each other through the diffusion barrier layer as shown in Fig.4-5(c). These results indicate that Ti/Wxt/Ti thin metal structure was thermally stable up to 400℃ annealing for 30 minutes.
4.3 DC characteristics of the airbridged device
(a) Device performance of GaN HEMT with Au airbridge：
The DC characteristics of the Au-airbridged GaN HEMTs were shown in Fig.4.6 and Fig.4-7. Fig.4-6 shows the current-voltage (I-V) characteristics of the 50x4 µm gate width HEMTs. The maximum drain current was 735.5 mA/mm and the threshold voltage was Vgs=
-5 V. Fig.4.7 shows the maximum extrinsic transconductance of 186.5 mS/mm was achieved at Vgs= -2.7 V and Vds=10 V.
37 transconductance of 184.4 mS/mm was achieved at Vgs= -2.5 V and Vds=10 V.
(c) Device performance of GaN HEMT with Cu airbridge using Ti/WNx/Ti/Cu barrier layer structure：
The DC characteristics of Ti/WNx/Ti/Cu used to fabricate Cu-metalized Airbridge for GaN HEMTs are shown in Fig.4-10 and Fig.4-11. Fig.4-10 shows the current-voltage (I-V) characteristics of 50x4 µm gate width HEMTs. The maximum drain current was 707 mA/mm and the threshold voltage was Vgs= -5 V. Fig.4-11 shows the maximum extrinsic transconductance of 163 mS/mm was achieved at Vgs= -2.5 V and Vds=10 V.
4.4 Comparison of DC characteristics of GaN HEMTs with Au-metalized and Cu-metalized airbridges
The 50x4 µm gate width GaN HEMT with Cu airbridges using Ti/Pt/Ti/Cu and Ti/WNx/Ti/Cu diffusion barriers showed the similar DC characteristics as compared to conventional GaN HEMT with Au airbridges ( Fig.4-12 ~ Fig.4-15). For Au airbridged device：
the drain current was 642 mA/mm and the maximum transconductance was 186.5 mS/mm, for GaN HEMT with Cu airbridges using Ti/Pt/Ti/Cu diffusion barrier, the drain current was
38
630.5 mA/mm and the maximum transconductance was 184.4mS/mm, for GaN HEMT with Cu airbridges using Ti/WNx/Ti/Cu diffusion barrier, the drain current was 627.3 mA/mm and the maximum transconductance was 163mS/mm when biased at Vds= 10 V. The threshold voltage was Vgs = -5V, as shown in Fig.4-12. However, the Vgs bias of the maximum transconductance is different for three devices：For Au airbridged GaN HEMT at Vgs = -2.7V, for Cu airbridged GaN HEMT at Vgs = -2.5V, as shown in Fig.4-13. Fig.4-14 and Fig.4-15 shows the leakage current of GaN HEMT with 50x4 µm gate width using Cu and Au metallizations. The leakage was very small and Cu-metallized showed the similar leakage current characteristics as compared to conventional GaN HEMT with Au-metallized. Finally, Table 4-1 shows the comparison of DC characteristics of GaN HEMT with 50x4 µm gate width using Cu and Au metallizations.
4.5 High temperature stability test
Here, the high temperature stability test is under 300 °C annealing for 30 minutes. From the Fig.4-16 and Fig.4-17, there is no obvious degradation on the DC characteristics performance of using Ti/Pt/Ti/Cu thin metal structures for Cu-metallized GaN HEMT after 300 °C annealing for 30 minutes. And from the Fig.4-18 and Fig.4-19, there is no obvious degradation on the DC characteristics performance of using Ti/WNx/Ti/Cu thin metal structures for Cu-metallized GaN HEMT after 300 °C annealing for 30 minutes, too.
39
4.6 Reliability test
The reliability test is under 350 KV/cm2 high voltage density stress for 24 hours at room temperature. Fig.4-20 shows the current of using Ti/Pt/Ti/Cu thin metal structures for Cu-metallized GaN HEMT after stressed at the high voltage density of 350 KV/cm2 for 24 hours at room temperature. The data shows that the current has no significant change with time. Similarly, Fig.4-21 shows the current of using Ti/WNx/Ti/Cu thin metal structures for Cu-metallized GaN HEMT after stressed at the high voltage density of 350 KV/cm2 for 24 hours at room temperature. The data shows that the current has no significant change with time, too.
40
Table
Au-metallized
Cu-metallized (Ti/Pt/Ti/Cu)
Cu-metallized (Ti/WNx/Ti/Cu) Vds=10V,Vgs=0V
→Ids (mA/mm) 642 630.5 627.3
Vds=4V,Vgs=0V
→Ids,max (mA/mm) 735.5 722.5 707
Gm,max (mS/mm) 186.5 184.4 163.5
Gm,max
→Vgs (V) -2.7 -2.5 -2.5
Threshold voltage (V)
-5 -5 -5
Table 4-1 Comparison of DC characteristics of GaN HEMT with 50x4 µm gate width using Cu and Au metallizations.
41
Figure
Fig.4-1 Transmission line methods (TLM) patterns
Fig.4-2 Utilizing TLM to measure the ohmic contact resistance
42
SPACE(µm) 3 5 10 20 36
TLM(Ω ) 23.4 33 60.4 117.08 203.2
The specific contact resistivity ＝ 1.3404E-06 (Ω-cm2)
Fig.4.3 The specific contact resistivity was measured by TLM of the Ti/Al/Ni/Au Ohmic contacts for GaN HEMT.
0 50 100 150 200 250
0 10 20 30 40
Resistance(Ω)
Width(µm)
43
Figure 4-4(a) AES depth profiles of the as-deposited Au/Ti/Pt/Ti/Cu multilayer structure.
Figure 4-4(b) AES depth profiles of the Au/Ti/Pt/Ti/Cu multilayer structure after 350°C annealing for 30min.
44
Figure 4-4(c) AES depth profiles of the Au/Ti/Pt/Ti/Cu multilayer structure after 400°C annealing for 30min.
Figure 4-4(a)~(c) show the AES depth profiles of the Au/Ti/Pt/Ti/Cu multilayer structure as-deposited and after 350℃ and 400℃ annealing for 30 minutes.
45
Figure 4-5(a) AES depth profiles of the as-deposited Au/Ti/WNx/Ti/Cu multilayer structure.
Figure 4-5(b) AES depth profiles of the Au/Ti/WNx/Ti/Cu multilayer structure after 400°C annealing for 30min.
46
Figure 4-5(c) AES depth profiles of the Au/Ti/WNx/Ti/Cu multilayer structure after 450°C annealing for 30min.
Figure 4-5(a)~(c) show the AES depth profiles of the Au/Ti/WNx/Ti/Cu multilayer structure as-deposited and after 400℃ and 450℃ annealing for 30 minutes.
47
Fig.4-6 Ids versus Vds curves of the GaN HEMT with 50x4 µm gate width and Au metallization.
Fig.4-7 Extrinsic transconductance and Ids versus Vgs bias characteristics of the GaN HEMT curves for 50x4 µm gate width with Au metallization.
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5
48
Fig.4-8 Ids versus Vds curves of the GaN HEMT of 50x4 µm gate width with Ti/Pt/Ti/Cu thin metal structure as diffusion barrier for Cu- metalized airbridges
Fig.4-9 Extrinsic transconductance and Ids versus Vgs bias characteristics of the GaN HEMT of 50x4 µm gate width with Ti/Pt/Ti/Cu thin metal structure as diffusion barrier for
49
Fig.4-10 Ids versus Vds curves of the GaN HEMT with 50x4 µm gate width with Ti/WNx/Ti/Cu thin metal structure as diffusion barrier for Cu-metalized airbridges
Fig.4-11 Extrinsic transconductance and Ids versus Vgs bias characteristics of the GaN HEMT with 50x4 µm gate width and with Ti/WNx/Ti/Cu thin metal structure as diffusion
50
Fig.4-12 Comparison of I-V characteristics of GaN HEMT with 50x4 µm gate width using Cu and Au metallizations.
Fig.4-13 Extrinsic transconductance and drain to source current versus Vgs bias characteristics of the GaN with 50x4 µm gate width using Cu and Au metallizations.
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5
51
Fig.4-14 Comparison of gate leakage current of GaN HEMT with 50x4 µm gate width using Cu and Au metallizations.
Fig.4-15 Comparison of drain and gate leakage current of GaN HEMT with 50x4 µm gate width using Cu and Au metallizations in the off-state.
0 2 4 6 8 10
52
Fig.4-16 Ids versus Vds curves of the GaN HEMT of 50x4 µm gate width with Ti/Pt/Ti/Cu thin metal structure as diffusion barrier for Cu- metalized airbridges before and after annealing at 300℃ for 30 minutes.
Fig.4-17 Extrinsic transconductance and Ids versus Vgs bias characteristics of the GaN HEMT of 50x4 µm gate width with Ti/Pt/Ti/Cu thin metal structure as diffusion barrier for Cu-metalized airbridge before and after annealing at 300℃ for 30 minutes.
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5
53
Fig.4-18 Ids versus Vds curves of the GaN HEMT of 50x4 µm gate width with Ti/WNx/Ti/Cu thin metal structure as diffusion barrier for Cu- metalized airbridges before and after annealing at 300℃ for 30 minutes.
Fig.4-19 Extrinsic transconductance and Ids versus Vgs bias characteristics of the GaN HEMT of 50x4 µm gate width with Ti/WNx/Ti/Cu thin metal structure as diffusion barrier for Cu-metalized airbridge before and after annealing at 300℃ for 30 minutes.
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5
54
Fig.4-20 The current of using Ti/Pt/Ti/Cu thin metal structures for Cu-metallized GaN HEMT after stressed at the high voltage density of 350 KV/cm2 for 24 hours at room temperature.
Fig.4-21 The current of using Ti/WNx/Ti/Cu thin metal structures for Cu-metallized GaN
55
Chapter 5
Conclusions
In this thesis, Cu-airbridged GaN HEMT using Ti/Pt/Ti/Cu and Ti/WNx/Ti/Cu metal schemes were successfully fabricated, and the thermal stability of the metal schemes were verified by the AES depth profiles analysis.
AES was used to study the inter-diffusion of Cu with Au based ohmic metal. The AES depth profiles of the Au/Ti/Pt/Ti/Cu multilayer structure shows that there was no inter-atomic diffusion between Cu metal and Au based ohmic metal after 350℃ annealing for 30 minutes.
However, after 400℃ annealing for 30 minutes, Cu and Au began to inter-diffuse into each other through the Pt diffusion barrier layer. The results indicate that Ti/Pt/Ti/Cu thin metal structure can be used as diffusion barrier for Cu-metallized GaN HEMT with thermal stability up to 350℃ annealing for 30 minutes.
Similarly, no inter-diffusion was observed for Au/Ti/WNx/Ti/Cu multilayer structure between Cu metal and Au metal after 400℃ annealing for 30 minutes by the AES depth profiles. However, Cu and Au began to inter-diffuse to each other through the WNx diffusion barrier layer. It means using Ti/WNx/Ti/Cu thin metal structure can be used as diffusion barrier for Cu-metallized GaN HEMT with thermal stability up to 400℃ annealing for 30 minutes.
In this study, Cu-airbridged GaN HEMTs using Ti/Pt/Ti/Cu and Ti/WNx/Ti/Cu as the diffusion barriers showed the comparable electrical characteristics with the Au-airbridged GaN HEMT. For Au airbridged device, the maximum drain current was 735.5 mA/mm and the maximum transconductance was 186.5 mS/mm, for Cu airbridged device using
56
Ti/Pt/Ti/Cu metal scheme, the maximum drain current was 722.5 mA/mm and the maximum transconductance was 184.4mS/mm; for Cu airbridged device using Ti/WNx/Ti/Cu metal scheme, the maximum drain current was 707 mA/mm and the maximum transconductance was 163mS/mm when biased at Vds= 10 V and Vgs= 0 V. The Cu-metallized interconnect on GaN HEMTs have shown comparable electrical characteristics with Au-metallized interconnect on GaN HEMTs.
In conclusion, these experimental results demonstrate that using Pt and WNx as diffusion barriers for Cu-metallized interconnects on GaN HEMT can effectively prevent Cu diffusion.
57
References
[1] K. Holloway and P. M. Fryer, “Tantalum as a diffusion barrier between copper and silicon,”
Appl. Phys. Lett. 57, 1736 (1990).
[2] K. Holloway, P. M. Fryer, C. Cabral, Jr., J. M. E. Harper, P. J. Bailey, and K. H. Kelleher,
“Tantalum as a diffusion barrier between copper and silicon: failure mechanism and effect of nitrogen additions,” J. Appl. Phys. 71, 5433 (1992).
[3] D. S. Yoon, H. K. Baik, and S. M. Lee, “Effect on thermal stability of a Cu/Ta/Si heterostructure of the incorporation of cerium oxide into the Ta barrier,” J. Appl. Phys. 83, 8074 (1998).
[4] E. R. Weber, “Transition metals in silicon,” Appl. Phys. A, Solids Surf. 1, 1 (1983).
[5] A. Cros, M. O. Aboelfofotoh, and K. N. Tu, “Formation, oxidation, electronic, and electrical properties of copper silicides,” J. Appl. Phys. 67, 3328, (1990).
[6] C. A. Chang, “Formation of copper silicides from Cu(100)/Si(100) and Cu(111)/Si(111) structures,” J. Appl. Phys. 67, 556 (1990).
[7] Y.C. Wu, “Low cost, low power consumption SPDT GaAs switches with copper metallization and dielectric, ”(2009).
[8] H. C. Chang, E. Y. Chang, Y. C. Lien, L. H. Chu, S. W. Chang, R. C. Huang and H. M. Lee,
“Use of WN
X as diffusion barrier for copper airbridged low noise GAAs PHEMT,” Electron.
Lett. 39, 1763 (2003).
[9]Cheng-Shih Lee, Yi-Chung Lien, Edward Yi Chang, Huang-Choung Chang, Szu-Houng Chen, Ching-Ting Lee, Li-Hsin Chu, Shang-Wen Chang, and Yen-Chang Hsieh
“Copper-airbridge low-noise GaAs PHEMT with Ti/WNx/Ti diffusion barrier for high-frequency applications,” IEEE Trans. Electron Devices 53, 8 (2006).
58
[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, “Gold-Free Fully Cu-Metallized InGaP/GaAs Heterojunction Bipolar Transistor,” Jpn. J. Appl. Phys. 44, 8 (2005).
[11] Y. C. Wu, E. Y. Chang, Y. C. Lin, H. T. Hsu, S. H. Chen, W. C. Wu, L. H. Chu, and C. Y.
Chang, “SPDT GaAs switches with copper metalized interconnects, ”IEEE Micro. Wireless Compon. Lett., 17, 133(2007).
[12] S. W. Chang, E. Y. Chang, C. S. Lee, K. S. Chen, C. W. Tseng, Y. Y. Tu, and C.T. Lee, “A gold-free fully copper-metallized InP heterojunction bipolar transistor using non-alloyed ohmic contanct and platinum diffusion barrier, ” Jpn. J. Appl.phys., 44, 899( 2005).
[13] U.K. Mishra, P. Parikh, Y.F. Wu, ”AlGaN/GaN HEMTs: An overview of device operation and applications,” Processdings of The IEEE, Vol. 90,No.6, June(2002).
[14] Chien-chi Lee,”GaN-Based Heterostructure Field Effect Transistors,“ June (2006).
[15]Vorgelegt von,M.Sc. Eng.,Ibrahim Khalil,and Barisal, Bangladesch,” Intermodulation Distortion in GaN HEMT,”17.07.2009
[16] V. Rajagopal Reddy *, P. Koteswara Rao,” Annealing temperature effect on electrical and structural properties of Cu/Au Schottky contacts to n-type GaN,” Microelectronic Engineering 85 (2008) 470-476
[17] Yi-Chung Lien, Edward Yi Chang, Szu-Hung Chen, Li-Hsin Chu, Po-Chou Chen, and Yen-Chang Hsieh,” Thermal stability of Ti/Pt/Cu Schottky contact on InAlAs layer,” Applied Physics Letters 89, 083517(2006).
[18] H. C. Chang “Copper-metallized interconnects on GaAs low noise pseudomorphic high electron mobility transistors, ”(2004).
[19] Peter Madakson, and Joyce C. Liu, “Interdiffusion and resistivity of Cu/Au, Cu/Co, Co/Au and Cu/Co/Au thin films at 25-550 ℃,” J. Appl. Phys. 68, 2121 (1990).
59
[20]Bruce M. Green, Student Member, IEEE, Kenneth K. Chu, E. Martin Chumbes, Student Member, IEEE, Joseph A. Smart, James R. Shealy, Member, IEEE, and Lester F. Eastman, Life Fellow, IEEE,” The Effect of Surface Passivation on the Microwave Characteristics of Undoped AlGaN/GaN HEMT’s,” IEEE Electron Device Letters, Vol. 21, No. 6, June (2000).
[21]Dieter K. Schroder, “Semiconductor material and device characterization,” Wiley Interscience.