High reflectivity and thermal-stability Cr-based Reflectors and

n-type Ohmic Contact for GaN-based flip-chip light-emitting

diodes
Kuang-Po Hsueh,a * Kuo-Chun Chiang,a Charles J. Wang,b and Yue-Ming Hsina
a

Department of Electrical Engineering, National Central University, Jhongli 32054, Taiwan,

R.O.C.
b

LEDARTS OPTO Corporation, Taiwan, R.O.C.

ABSTRACT

We have investigated the thermal stability of three composite metals on their contact resistivities and
luminous intensities for using as the reflector in flip-chip light-emitting diode (FCLED). The composite metals
were simultaneously deposited on n-type GaN without alloy to form n-type Ohmic contact and simplify the
process. The investigated composite metals were Ti/Al/Ti/Au (30/500/30/300 nm), Cr/Al/Cr/Au (30/500/30/300
nm) and Cr/Ti/Au (500/30/300 nm), respectively. The specific contact resistivity of Ti/Al/Ti/Au, Cr/Al/Cr/Au
and Cr/Ti/Au on the n-type GaN Ohmic contact were changed from 5.410-4, 6.610-4 and 7.710-4 -cm2 to
5.310-4, 4.510-4 and 1.310-4 -cm2 respectively after 500 hours thermal stress at 150C in the air. After 96
hours of thermal stress, the luminous intensities at 20 mA of these three structures were decreased 6.2%, 11.1%
and 1.4%, respectively. Therefore, in addition to maintain good n-type ohmic contact and simplify the process,
the Cr/Ti/Au composite metal demonstrates good thermal stability as a reflector in FCLED.

Keywords : flip-chip light-emitting diode, GaN, FCLED, Cr, Al.

*Electronic mail: [email protected] ; Fax : +886-3-4255830

Gallium Nitride Materials and Devices, edited by Cole W. Litton

James G. Grote, Hadis Morkoc, Anupam Madhukar, Proc. of SPIE
Vol. 6121, 61210V, (2006), 0277-786X/06/$15 doi: 10.1117/12.646154
Proc. of SPIE Vol. 6121 61210V-1

1. INTRODUCTION
Group III-Nitride semiconductors are of great technological importance for the fabrication of
optoelectronic devices, such as blue and ultraviolet light-emitting diodes (LEDs), laser diodes, and
photodetectors.1-3 One of the most important applications of GaN-based LEDs is solid-state lighting, which
could replace incandescent bulbs and ultimately fluorescent lamps. Realization of high brightness and high
power GaN-based LEDs are important to solid-state lighting applications, which require high extraction
efficiency in LED structures. The flip-chip LEDs (FCLEDs) have been demonstrated the enhanced light
extraction and reduced thermal dissipation.4-8 However, two issues are important for FCLED to achieve high
performances. One is the highly reflective p-type ohmic contact, which serves as the p-electrode and the mirror
to reflect the light emitting from the active region. The other one is good thermal stability during packaging and
operation. It is well known that the conventional p-GaN contact is a Ni/Au bilayer composed of several hundred
angstroms of each material annealed at around 500C under nitrogen. This contact exhibits good electrical
characteristics but is a poor reflector at visible wavelengths. In this paper, oxidized Ni/Au bilayer contacts were
used to obtain high-quality p-GaN Ohmic contacts, but added the composite metals for using as the reflector and
bonding layer in FCLED. In addition, these composite metals were simultaneously deposited on n-type GaN
without alloy to form Ohmic contact and simplify the process. The investigated composite metals were
Ti/Al/Ti/Au, Cr/Al/Cr/Au and Cr/Ti/Au, respectively. The thermal stability of reflectivity and contact resistivity
at 150C in the air was studied.

2. DEVICE FABRICATION
The InGaN/GaN MQW LED wafers were grown by metal-organic chemical vapor deposition (MOCVD)
on c-plane sapphire substrates using a GaN buffer layer followed by a thickness of 3 m, 51018 cm-3 Si-doped,
n-type GaN current spreading layer. The MQW active region consists of 5 periods of InGaN wells and Si doped
GaN barriers. The structure was completed with a thickness of 50 nm and 21017 cm-3 Mg-doped AlGaN layer,
followed by a 50 nm 51017 cm-3 Mg-doped GaN contact layer. The fabricated LEDs were designed to operate
at 460 to 470 nm. The schematic device structure with flip-chip submount is shown in Fig. 1.
Devices were fabricated using standard lithographic techniques to define features. The Mg acceptors in the
p-GaN and p-AlGaN were activated by the furnace at 750C for 30 minutes in nitrogen ambient. High density
plasma etching was used to expose the n-type GaN current spreading layer. Ohmic p-type contacts were first
formed by evaporating a Ni/Au metal stack which was annealed by the furnace at 500 C for 5 minutes under
oxygen ambient. The size of the p-type contact is 200 285 m2. Following the p-type contacts, three different
types of composite metals (including diffusion barrier, metal reflector and bonding pad layers) were formed on

Proc. of SPIE Vol. 6121 61210V-2

top of the Ni/Au p-electrode. Three types of composite metals in this study, as shown in Fig. 1, were Ti/Al/Ti/Au
(30/500/30/300 nm), Cr/Al/Cr/Au (30/500/30/300 nm) and Cr/Ti/Au (500/30/300 nm), respectively. In addition,
these composite metals were simultaneously deposited on n-type GaN without alloy to form Ohmic contact and
simplify the process. The deposition of composite metals was carried out by electron beam evaporation at a
chamber pressure of 310-6 torrs. After metallization, devices were characterized for thermal stability under
thermal stress at 150C in the air. Current-voltage (I-V) measurements performed by the parameter analyzer
(HP4156) were used to evaluate the contact resistances. The luminous intensity was measured by Instrument
System Spectrometer (CAS140B) at wavelength of 460 nm to compare the relative reflectivity after thermal
stress. The surface morphology and roughness of the contact metal were studied using an atomic force
microscope (AFM).

Type I

Type II

Type III

M1

Ti

Cr

Cr

M2

Al

Al

M3

Ti

Cr

Ti

M4

Au

Au

Au

Sapphire
GaN buffer
n-GaN
MQW
p-AlGaN
solder

Ni/Au p-electrode

p-GaN

M1 : Diffusion Barrier Layer

M2 : Metal Reflector Layer
M3 : Bonding Pad
M4 : Bonding Pad

submount

Fig. 1 Schematic cross section of the FCLED with the different composite metals in this study.

3. RESULTS
Alumina is usually considered to achieve a good metal reflector in FCLED. For the type I of composite
Ti/Al/Ti/Au, Al is good reflective material and could be formed Ti-Al alloy with the adjacent metal of Ti. And
the Ti-Al alloy can prevent the Al diffusion.9 The outside metal of Au could avoid the oxidization of Al. Due to

Proc. of SPIE Vol. 6121 61210V-3

the similar characteristic on the n-type Ohmic contact, Cr was used to replace Ti to form type II of composite
Cr/Al/Cr/Au for this study.10 Another reason is the better thermal stability in Cr than Al at higher temperature
operation.11 Type III of composite Cr/Ti/Au was included to study the composite metal without Al.
Figure 2 shows the specific contact resistances (c) of Ti/Al/Ti/Au, Cr/Al/Cr/Au and Cr/Ti/Au on n-type
GaN contacts versus stress time at 150C in the air obtained from TLM measurement. All the Ti/Al/Ti/Au,
Cr/Al/Cr/Au and Cr/Ti/Au contacts show linear I-V curves with and without thermal stress. The measured data
shows that c decrease from 5.410-4, 6.610-4 and 7.710-4 -cm2 to 5.310-4, 4.510-4 and 1.310-4 -cm2,
respectively, after 500 hours thermal stress. The specific contact resistivity of Cr/Ti/Au demonstrates the most
reduction of 83.1% after 500 hours thermal stress while other two composite metals do not decrease
significantly. This is due to the Al metal tends to oxidize during the thermal stress.

9
Ti/Al/Ti/Au
Cr/Al/Cr/Au
Cr/Ti/Au

-4

c (x10 -cm )

8
6
5
4
3
2
1
0
-100

100

200

300

400

Stress Time (hr)

500

600

Fig. 2 The specific contact resistivities versus thermal stress time for three different composite metals on n-type GaN. (Metal
contact area is 50m 100m.)

Proc. of SPIE Vol. 6121 61210V-4

Table I The changes of the specific contact resistivities after thermal stress time for three different composite metals on
n-type GaN.

Specific Contact Resistivity (W-cm )

As-deposited
150C / 500hours

Metal

-4

Ti/Al/Ti/Au
Cr/Al/Cr/Au
Cr/Ti/Au

50

30

40

Current (mA)

40

Current (mA)

5.3210
-4
4.4910

5.4010
-4
6.5810
7.7010

-4

1.310

-4

as-dep

as-dep
500hrs
as-dep

500hrs
as-dep

500hrs
30

%
-1.5
-31.8
-83.1

-4

Ti/Al/Ti/Au
Cr/ Al/Cr/Au
Cr/ Ti/Au

as-dep

500hrs

500hrs

20

as-dep
20
3.1

3. 2

3.3

Voltage (V)

3.4

3.5

500hrs

10
Ti/Al/Ti/Au
Cr/Al/Cr/Au
Cr/Ti/Au

0
-10

2.6

2.8

3.0

3.2

3.4

Voltage (V)

Fig. 3 Typical forward I-V characteristics of LEDs fabricated with three composite metals after 500 hours at 150C in air. For
comparison, the I-V behaviors without thermal stress (as-dep) are also shown.

Figure 3 shows the corresponding forward I-V characteristics of LEDs measured up to 3.5V. It shows the

Proc. of SPIE Vol. 6121 61210V-5

increased series resistance of 2.8 and 0.8 in both Ti/Al/Ti/Au and Cr/Al/Cr/Au metals after 500 hours thermal
stress. Since the contact resistance on the n-type GaN is decreased, the degradation would result from the p-type
Ohmic contact with Al oxidation. In addition, the root-mean-square surface roughness of Ti/Al/Ti/Au,
Cr/Al/Cr/Au and Cr/Ti/Au on p-electrode shows the increase from 3.65, 7.17 and 3.67 nm to 4.23, 17.24 and
3.76 nm respectively after 150C 500hrs thermal stress. The least change in morphology in Cr/Ti/Au composite
metal is due to the elimination of Al. The thermal stability of relative luminous intensity of FCLEDs using
Ti/Al/Ti/Au, Cr/Al/Cr/Au and Cr/Ti/Au composite metal is showed in Fig. 4. The measured luminous intensities
at 20 mA in these three structures (Ti/Al/Ti/Au, Cr/Al/Cr/Au and Cr/Ti/Au) without thermal stress are 27.6, 27.9
and 35.4 mcd respectively. The improvement of using metal reflector is significant while compared with the
normal LED without flip-chip process (16.6 mcd). In addition, FCLED with Cr/Ti/Au demonstrates the highest
luminous intensity than those with Al composite reflector; it is possible due to the wafer uniformity (7 mcd
variation) and the existence of diffusion barrier layer. After 96 hours of thermal stress, the luminous intensities
were decreased 6.2%, 11.1% and 1.4% for diodes with Ti/Al/Ti/Au, Cr/Al/Cr/Au and Cr/Ti/Au, respectively.
The Cr/Ti/Au composite metal demonstrated the least reduction in luminous intensity and thus excellent thermal
stability.

Normalized Luminous lntensity

@ 20 mA (%)

110

Ti/Al/Ti/Au
Cr/Al/Cr/Au
Cr/Ti/Au

100

90

80

24

48

72

96

hours

Fig. 4 Normalized luminous intensity versus thermal stress time for three FCLEDs operated at current of 20 mA.

Proc. of SPIE Vol. 6121 61210V-6

Table II The changes of Normalized luminous intensity after thermal stress time for three different composite metals.

Metal
Ti/Al/Ti/Au
Cr/Al/Cr/Au
Cr/Ti/Au

Normalized luminous intensity (%)

As-deposited
150C / 96hours
100
93.8
100
88.9
100
98.6

%
-6.20
-11.10
-1.40

4. CONCLUSION
To summarize, three composite metals were investigated to form n-type Ohmic contacts and reflectors for
FCLEDs. The specific contact resistivity of Ti/Al/Ti/Au, Cr/Al/Cr/Au and Cr/Ti/Au on the n-type GaN Ohmic
contact were decreased from 5.410-4, 6.610-4 and 7.710-4 -cm2 to 5.310-4, 4.510-4 and 1.310-4 -cm2
respectively after 500 hours thermal stress at 150C in the air. The luminous intensity at 20 mA of the FCLEDs
in these three structures (Ti/Al/Ti/Au, Cr/Al/Cr/Au and Cr/Ti/Au) were significantly improved compared with
the normal LED without flip-chip process. After 96 hours of thermal stress, the luminous intensities at 20 mA
of these three structures were decreased 6.2%, 11.1% and 1.4%, respectively. Therefore, FCLEDs fabricated
with the Cr/Ti/Au reflector demonstrated good n-type Ohmic contact and thermal stability in reflectivity. By
using this process, the n-type Ohmic contact, the reflector and the bonding pads can be formed at the same
time. It is not only to simplify the process and thus manufacture cost, but also maintain good thermal stability
on Ohmic contact and luminous intensity.

Acknowledgments
The authors would like to thank the National Science Council of the Republic of China, Taiwan for
financially supporting this research under Contract No. NSC 94-2215-E-008-005.

Proc. of SPIE Vol. 6121 61210V-7

REFERENCES

1. S. J. Pearton, J. C. Zolper, R. J. Shul, and F. Ren, J. Appl. Phys., 86, 1 (1999).

2. Shuji Nakamura, Masayuki Senoh, Shin-ichi Nagahama, Naruhito Iwasa, Takao Yamada, Toshio Matsushita,
Hiroyuki Kiyoku, Yasunobu Sugimoto, Tokuya Kozaki, Hitoshi Umemoto, Masahiko Sano, and Kazuyuki
Chocho, Appl. Phys. Lett., 72, 2014 (1998).

3. M. Razeghi and A. Rogalski, J. Appl. Phys., 79, 7433 (1996).

4. J. J. Wierer, D. A. Steigerwald, M. R. Krames, J. J. OShea, M. J. Ludowise, G. Christenson, Y. C. Shen, C.
Lowery, P. S. Martin, S. Subramanya, W. Gtz, N. F. Gardner, R. S. Kern, and S. A. Stockman, Appl. Phys.
Lett., 78, 3379 (2001).

5. J. O. Song, W. K. Hong, Y. Park, J. S. Kwak, and T. Y. Seong, Appl. Phys. Lett., 86, 133503 (2005).
6. J. O. Song, D. S. Leem, J. S. Kwak, O. H. Nam, Y. Park, and T. Y. Seong, Appl. Phys. Lett. 83, 4990 (2003).
7. J. O. Song, J. S. Kwak, and T. Y. Seong, Appl. Phys. Lett. 86, 062103 (2005).
8. A. Chitnis, J. Sun, V. Mandavilli, R. Pachipulusu, S. Wu, M. Gaevski, V. Adivarahan, J. P. Zhang, M. Asif
Khan, A. Sarua, and M. Kuball, Appl. Phys. Lett. 81, 3491 (2002).

9. D. F. Wang, F. Shiwei, C. Lu, A. Motayed, M. Jah, S. N. Mohammad, K. A. Jones, and L. Salamanca-Riba, J.

Appl. Phys., 89, 6214 (2001).

10. A. C. Schmitz, A. T. Ping, M. A. Khan, Q. Chen, J. W. Yang, and I. Adesida, Semicond. Sci. Technol., 11,
1464 (1996).

11. David R. Lide, Handbook of Chemistry and Physics, editor in chief 82nd Edition page 12-133.

Proc. of SPIE Vol. 6121 61210V-8