Home

Search

Collections

Journals

About

Contact us

My IOPscience

An improved non-alloyed ohmic contact Cr/Ni/Au to n-type GaN with surface treatment

This article has been downloaded from IOPscience. Please scroll down to see the full text article.
2008 J. Phys. D: Appl. Phys. 41 175107
(http://iopscience.iop.org/0022-3727/41/17/175107)
View the table of contents for this issue, or go to the journal homepage for more

Download details:
IP Address: 128.103.149.52
The article was downloaded on 15/07/2012 at 14:47

Please note that terms and conditions apply.

IOP PUBLISHING

JOURNAL OF PHYSICS D: APPLIED PHYSICS

J. Phys. D: Appl. Phys. 41 (2008) 175107 (4pp)

doi:10.1088/0022-3727/41/17/175107

An improved non-alloyed ohmic contact

Cr/Ni/Au to n-type GaN with surface
treatment
Hyun Kyong Cho1 , Sun-Kyung Kim1 and Jeong Soo Lee2
1
2

LED New Technology Lab, LG Innotek, Seoul 137-724, Korea

LG Electronics Advanced Research Institute, Seoul 137-724, Korea

E-mail: hkchoa@lginnotek.com

Received 22 May 2008, in final form 24 June 2008

Published 21 August 2008
Online at stacks.iop.org/JPhysD/41/175107
Abstract
The Cr/Ni/Au non-alloyed ohmic contact resistance on n-type GaN is obtained by chemical
surface treatment of n-type GaN films following the laser lift-off of the sapphire substrate. The
effects of n-GaN surface treatments on the metal/GaN interface were studied using x-ray
photoelectron spectroscopy. Nitrogen vacancies at the n-type GaN surface are therefore
produced and act as donors for electrons, improving the non-alloyed ohmic contact resistance
induced by the reduction in native oxygen by the surface treatment of chemical solutions. In
addition, the n-GaN surface treatment reduces the forward voltage (Vf ) of the vertical LEDs.
(Some figures in this article are in colour only in the electronic version)

regenerate on the surface prior to metal deposition. In order to

reduce the surface product and improve the contact resistivity,
considerable work has been performed on vertical injection
LEDs [9, 10].
Generally, the thermal annealing process of metal
activation is used for improving the contact resistivity in
conventional lateral GaN-based LEDs, but this inevitably
induces thermal damage such as decomposition and a spiky
interface within the vertical LED structure because the high
temperature process that is used for vertical GaN-based LEDs
is incompatible with other substrates (Si or metal substrates).
Therefore, the pretreatment of the surface before metal
deposition is the key to reducing the contact resistance on GaN.
In our work, we created a non-alloyed ohmic contact
with surface treatment using chemical wet-etching on an airexposed n-GaN surface after LLO and a subsequent undoped
GaN etch.
The GaN-based epilayer structure used in this work
was grown on a sapphire substrate by metal organic
chemical vapour deposition (MOCVD). The LED epilayers
consist of an undoped GaN layer (2 m), an Si-doped
n-GaN layer (3 m), a five-period InGaN/GaN multi
quantum well and a 100 nm thick Mg-doped p-GaN
layer.
Deposition and annealing of a 280 nm thick
transparent indium tin oxide (ITO) film was performed

High power light-emitting diodes (LEDs) are in high demand

for solid-state lighting applications and are expected to replace
conventional lighting applications such as incandescent and
fluorescent lamps [1, 2].
These LEDs require high current injection and they
generate high temperatures during the operation of the LED.
High current injection reduces the internal quantum efficiency
of the active layers in the LED by raising the junction
temperature, and this causes the brightness and operating
voltage of the LED to deteriorate. However, the high current
injection can be avoided by fabricating the LED with a vertical
structure, where a laser lift-off (LLO) process is used to transfer
the GaN film from the sapphire substrate and integrate GaN
with thermally and electrically conductive substrate materials
such as Si, GaAs and Cu to distribute the paths of current
flows and to act as a heat sink [35]. Also, surface patterning
for producing photonic crystal (PhC) structures and surface
texturing have been generated using the dry etching method
to improve the extraction efficiency of vertical GaN-based
LEDs [68].
Unfortunately, these techniques destroy the surface of
the GaN film and thus increase the resistivity of the contact
region. The surfaces of compound semiconductors have very
active chemisorption and production of the native oxide on
the surface. A residual native oxide would still exist or
0022-3727/08/175107+04$30.00

2008 IOP Publishing Ltd

Printed in the UK

J. Phys. D: Appl. Phys. 41 (2008) 175107

H K Cho et al

to serve as a p-contact to the p-GaN. Subsequently,

an Ni(10 )/Ag(2000 )/Ni(1000 )/Cu(70 m) layer was
deposited by an e-beam evaporator on the p-GaN surface. The
multi-metal layer acted not only as a reflective mirror layer but
also as a conductive seed layer on which a 70 m thick Cu was
electroplated. Using an ArF excimer laser with a wavelength
of 193 nm, an LLO process was performed to separate the
sapphire substrate from the GaN-based LED structure. The
GaN attached with Cu was rinsed in an HCl : DI (1 : 1) solution
in order to remove the uppermost low quality Ga nucleation
layer. The air-exposed undoped GaN surface was dry etched
using inductively-coupled-plasma reactive ion etching (ICPRIE) with a Cl2 /BCl3 etchant by 2 m in order to expose the
n-GaN layer.
The specific contact resistivities were measured using the
linear transmission line method (TLM). For the measurement
of contact resistivity on n-GaN of the vertical LED, an active
rectangular mesa region was defined using ICP-RIE. The
TLM test structure was patterned with a photoresist. Contact
dimensions were 10050 m2 , and the TLM spacing between
the contact pads was 5, 10, 15, 20 and 25 m.
Prior to metal evaporation of Cr/Ni/Au (200 /250 /
5000 ) using the electron beam evaporator, the surfaces of
the samples were treated with various chemical etchants such
as buffered oxide etch (BOE). The currentvoltage (I V )
characteristics of the contacts were measured using a fourpoint-probe arrangement at room temperature. Also, x-ray
photoelectron spectroscopy (XPS) (PHI5400 system) with
Al K was used to analyse the surface state in n-GaN with
different surface treatments.
Figure 1 shows the change in surface morphology in
n-GaN as measured by atomic force microscopy (AFM) of
the samples after LLO and LLO/ICP. For the samples that
were processed with LLO operation, the rms roughness was
increased to 46 , while that of as-grown samples was 2 . No
distinct change in roughness was found after the dry etching
of GaN for exposing n-GaN in the LLO/ICP sample (39 ).
Figure 2(a) shows the I V curves of Cr/Ni/Au nonalloyed ohmic contacts to n-GaN with and without BOE
treatment. The I V curves were measured between the TLM
pads with a gap spacing of 5 m. The ohmic contact exhibits
linear I V behaviour after BOE surface treatment, while the
I V curve for the as-is sample (without surface treatment)
was non-linear. The specific contact resistivity as a function
of the BOE cleaning duration is shown in figure 2(b). The
specific contact resistivity was 1.2 104  cm2 after BOE
cleaning for 10 min. As the cleaning time was increased to
10 min, the contact resistivity was decreased. However, the
contact resistivity of the Cr/Ni/Au ohmic metal on the n-GaN
that was grown on the sapphire substrate was lower than that
of the result by two orders of magnitude.
Generally, the ohmic contact resistivity for n-GaN along
with the Ga-face surface grown on the sapphire substrate was
lower than 3 106  cm2 using Cr/Ni/Au. In our case, the
remaining n-GaN epitaxial films of the vertical structure LEDs
after removing the sapphire substrate displayed an N-face
(0 0 0 1) surface. The (0 0 0 1) surface (Ga-face) is composed
of three nitrogen dangling bonds that point upwards to the

(a)

RMS:46
(b)

RMS:39
Figure 1. The surface morphology in n-GaN as measured by AFM
of the samples after LLO (a) and LLO/ICP (b).

c-plane surface, while the (0 0 0 1) surface (N-face) has

a single nitrogen dangling bond that points upwards. The
difference in the structures of the surfaces affects the device
characteristics, especially the ohmic contact properties. The
Ga-face films have larger surface band bending than the N-face
film by about 3.5 eV and 1.4 eV in the theoretical calculation
and experimental results, respectively [1113]. It was found
that the Ga-face GaN has a larger surface band bending than the
N-face GaN, resulting in a higher Schottky barrier height [12].
After the surface treatment of n-GaN using BOE, the peak
of the GaN bond in the Ga 3d core level shifts towards higher
binding energies (not shown). The shift of the Ga 3d core
level peak is closely associated with the shift of the surface
Fermi level, increasing the conductivity of electrons at the
n-GaN surface. Figure 3 shows XPS spectra of O 1s core levels
with LLO, LLO/ICP and subsequent BOE cleaning. All O ls
spectra were deconvoluted into two components. OGa and
OC bonds can be seen. The peak intensity corresponding to
O ls on n-GaN increased after the LLO and subsequent u-GaN
etching by ICP. This shows that a large number of Ga oxides
were formed. When the etched sample was treated with BOE,
the peak of the Ga oxides was reduced. This suggests that the
Ga oxides that were unintentionally formed under the LLO and
Cl2 plasma could be removed by the BOE treatment.
Also, the relative Ga/N ratio from the n-GaN sample
of LLO/ICP is normalized to one. The Ga/N ratio of the
BOE treated sample exceeds that of the untreated sample.
2

J. Phys. D: Appl. Phys. 41 (2008) 175107

H K Cho et al

Normalized Intensity (arb.units)

O1s

LLO/ICP/BOE

LLO/ICP
O-Ga

O-C
LLO

532

534

536

538

540

542

544

546

Binding Energy (eV)

Figure 3. XPS spectra of O 1s core levels with LLO, LLO/ICP and

subsequent BOE cleaning.
100

VLED (w/o surface treatment)

VLED (with surface treatment)

Current (mA)

80

60

40

20

0
-2

-1

1
2
Voltage (V)

Figure 4. The I V characteristics of the vertical GaN LED with

LLO, u-GaN etching and subsequent surface treatment with BOE,
compared with the vertical LED without the surface treatment.

Figure 2. (a) The I V curves of Cr/Ni/Au non-alloyed ohmic

contacts to n-GaN with and without BOE treatment. (b) The specific
contact resistivity as a function of the BOE cleaning duration.

The increase in the Ga/N ratio after LLO, ICP etching and
subsequent BOE cleaning is indicative of the formation of
a Ga-rich layer near the surface due to the generation of N
vacancies.
Figure 4 shows the I V characteristics of the vertical
GaN LED with LLO, u-GaN etching and subsequent surface
treatment with BOE, compared with the vertical LED without
the surface treatment. The chip size is 500 500 m2 . It is
seen that the forward voltage (Vf ) drop in the vertical LED
at 60 mA is 3.3 V, which is about 39% lower than that of the
LED that had no surface treatment. Compared with the LED
without the surface treatment, the reduction in Vf of the LED
with the surface treatment is mainly attributed to low contact
resistivity because the surface treatment using BOE enables
relatively much less surface plasma damage during ICP.
The Cr/Ni/Au non-alloyed ohmic contact resistance of the
n-GaN in the vertical GaN LED was reduced by the BOE

surface treatment of n-GaN following LLO and undoped GaN

etching. XPS shows the reduction in the native oxide on the
n-GaN surface following the BOE treatment. The specific
contact resistivity of 1.2 104  cm2 was achieved in nonalloyed Cr/Ni/Au contacts on n-GaN of the vertical GaN LED
with BOE treatment. This suggests that the Ga oxides that
were unintentionally formed under the LLO and Cl2 plasma
could be removed by BOE treatment. Also, the increase in
the Ga/N ratio after LLO, ICP etching and subsequent BOE
cleaning is indicative of the formation of a Ga-rich layer near
the surface due to the generation of N vacancies.

References
[1] Wierer J J et al 2001 Appl. Phys. Lett. 78 3379
[2] Fujii F, Gao Y, Sharma R, Hu E L, DenBaars S P and
Nakamura S 2004 Appl. Phys. Lett. 84 855

J. Phys. D: Appl. Phys. 41 (2008) 175107

H K Cho et al

[8] David A, Fujii T, Sharma R, Moran B, Nakamura S,

DenBaars S P, Weisbush C and Benisty H 2006 Appl. Phys.
Lett. 88 133514
[9] Jang H W, Kim J K, Lee J-L, Schroeder J and Sands T 2003
Appl. Phys. Lett. 82 580
[10] Su S-H, Hou C-C, Lin M-H, Yokoyama M, Chen S-M and
Kuan H 2007 Japan. J. Appl. Phys. 46 965
[11] Karrer U, Ambacher O and Stutzmann M 2000 Appl. Phys.
Lett. 77 2012
[12] Jang H W, Lee J-H and Lee J-L 2002 Appl. Phys. Lett. 80 3955
[13] Osvald J, Kuzmik J, Konstantinidis G, Lobotka P and
Georgakilas A 2005 Microelectron. Eng. 81 181

[3] Wong W S, Sands T, Cheung N W, Kneissl M, Bour D P,

Mei P, Romando L T and Johnson N M 1999 Appl. Phys.
Lett. 75 1360
[4] Wong W S, Sands T, Cheung N W, Kneissl M, Bour D P,
Mei P, Romando L T and Johnson N M 2000 Appl. Phys.
Lett. 77 2822
[5] Chen S-L, Wang S-J, Uang K-M, Chen T-M, Lee W-C and
Liou B-W 2007 IEEE Photon. Technol. Lett. 19 351
[6] Chen Y C et al 2003 Appl. Phys. Lett. 82 2221
[7] Wierer J J, Krames M R, Epler J E, Gardner N F, Craford M G,
Wendt J R, Simmons J A and Sigalas M M 2004 Appl. Phys.
Lett. 84 3885