Effects of oxygen thermal annealing treatment on formation of

ohmic contacts to n-GaN

Wenting Hou (), Theeradetch Detchprohm ( ), and
Christian Wetzela
Department of Physics, Applied Physics, and Astronomy, Rensselaer Polytechnic
Institute, 110 8th Street, Troy, New York 12180, USA

A contact scheme to undoped and n-type GaN was identified that does not require a post
deposition anneal. As-deposited Ti/Al/Ti/Au usually forms a Schottky contact. However,
we find that by means of an oxygen rapid thermal annealing prior to metal deposition, the
contact will develop an ohmic behavior with a specific contact resistance of 3.810-5
cm2. In X-ray photoelectron spectroscopy we find that the Ga 3d electron binding energy
increases with this pre-treatment, indicating a shift of the Fermi level closer to the
conduction band. This sequence reversion of high temperature processing steps allows
important gain in device fabrication flexibility.
Keywords: GaN; light emitting diode; ohmic contact; rapid thermal anneal; oxygen.
Low-resistance ohmic contacts are essential for the fabrication of energy efficient electrooptical conversion devices, such as group-III nitride light emitting diodes (LEDs)1,2 and solar
cells.3 While low contact resistance has readily been achieved to p- and n-type GaN material
separately,4,5,6,7 we find problems when in bipolar devices both types of contacts have to be
integrated in a single wafer fabrication process. The problem originates in the required
temperature differences of thermal annealing steps for p- and n-contacts. The formation of ohmic
contacts to n-type GaN requires rapid thermal annealing (RTA) in a nitrogen ambient at high
temperature (750 900C).7 The formation of ohmic contacts to p-type GaN typically requires
RTA in an oxygen ambient at significantly lower temperature (400 600 C).4,5 In our
experimental findings, the high temperature annealing for n-contacts will degrade the quality of
any p-contact that has been formed prior to the n-contact. In turn, we find that thermal annealing
a

E-mail: wetzel@rpi.edu

in an oxygen ambient for p-contacts degrades the quality of the n-contact if it was formed prior
to the p-contact.
Here, we present a possible solution to the process step sequence dilemma by an oxygen
thermal annealing treatment prior to the n-contact deposition. In this approach, a low-resistance
ohmic contact to n-type GaN can be obtained even without a post-deposition thermal annealing.
The absence of such annealing step solves the problem of thermal deterioration of any previously
formed p-contact. We analyze the effects of thermal treatment time and temperature and employ
X-ray photoelectron spectroscopy (XPS) to elucidate the mechanisms of the proposed thermal
treatment step.
Low contact resistance to n-type GaN layers is readily achieved by various Ti-based
metallization stacks followed by RTA in nitrogen ambient at a high temperature of 600 900
C.7 For example, specific contact resistances as low as 810-6 cm2 in bilayer Ti/Al contacts
have been achieved by a post-deposition high temperature thermal annealing step at 900 C for
0.5 minutes.7 The effect was explained by the formation of a highly n-type doped GaN layer due
to the formation of N-vacancies, and a thin TiN interfacial layer.7,8 Even lower contact resistivity
of 5 10-6 cm2 was reported on Ti/Al contacts when the annealing temperature was lowered to
600 C which was believed to reduce oxidation of the metal itself.9 It has been reported by some
groups10,11 that the contact resistance can further be reduced by a chlorine-based reactive ion
etching (RIE) treatment of the GaN surface prior to contact deposition while others find a contact
deterioration by such treatment,12 attributed to surface lattice disordering. Yan et al. find that an
oxygen plasma treatment prior to the metal deposition can improve the contact resistance to ntype GaN.13 However, when this treatment was performed on p-type GaN prior to Ni/Au contact
deposition, this p-contact performs worse than on untreated p-type GaN.13 Therefore this oxygen
plasma treatment cannot be integrated in any bipolar LED fabrication process. We therefore
propose a pre-treatment of an oxygen RTA step instead.
Epitaxial (0001) GaN thin film layers of 4 m thickness were grown by metalorganic
vapor phase epitaxy (MOVPE) on c-plane sapphire substrate. We distinguish four sample series,
1, 2, 3, and 4, subjected to different treatments (see Table I for the naming scheme). Series 1, 2,
and 3 are n-type GaN layers Si-doped to a free electron concentration of 31018 cm-3. Series 4
are undoped GaN (u-GaN) layers with a residual carrier concentration of ~1016 cm-3. Sample sets
1 and 2 reflect the contact to the n-GaN layer of full LED wafers exposed by mesa etching in
2

inductively coupled plasma/reactive ion etching (ICP/RIE) with a power of 500 W and 100 W, a
BCl3/Cl2 flow of 20 sccm /50 sccm, under a pressure of 12 mT. The mesa etched n-GaN samples
provide a more realistic representation of the n-layer condition in actual LED fabrication
including possible surface degradation, etch residues, and changes in the surface morphology.
These factors have previously been found to have an impact on the contact characteristics. Both
sample sets 1 and 2 are nominally identical, but belong to separate preparation batches.
To study the effects of the pre-treatment we subject samples of all series to oxygen RTA
steps prior to the n-metal deposition. We distinguish pretreatments in oxygen and nitrogen gas,
labeling the samples O and N in second position, respectively. The standard annealing
condition O is 550 C for 1 minute in oxygen ambient. To vary the conditions, O2 received a
longer anneal (5 min., 550 C), O3 received a lower (400 C, 1 min.), and O4 a higher
annealing temperature (650 C, 1 min.). N indicates annealing in nitrogen ambient at 550 C
for 1 minute. For comparison, samples of all series were left without pre treatment are designated
as nt.
Subsequently, an n-contact was formed by metal layer deposition in the sequence of
Ti/Al/Ti/Au with thicknesses of 20 nm/100 nm/45 nm/55 nm, respectively, by means of e-beam
evaporation to all samples.
Post contact deposition, two different schemes were applied. Samples were either
annealed in a nitrogen ambient (Sample designator N in the third position) or did not receive
an annealing step (Sample designator na). Two different metal annealing conditions in N
ambient were applied. N indicates 750 C for 1 minute (Samples 1-nt-N, 3-nt-N, 3-O-N) and
N2 indicates 900 C for 0.5 minutes (Samples 3-nt-N2, 3-O-N2). As a further variant, an
oxygen anneal (550 C, 1 min) was applied after the N anneal for sample 1-nt-NO. Samples 1-ntna, 1-O-na, 3-nt-na, 3-O-na and sample sets 2 and 4 were not annealed after the metal deposition.
Sample 1-nt-na, 2-nt-na, and 3-nt-na are the reference samples in each set without any predeposition treatment or post-deposition anneal. For an overview of the treatment conditions see
Table I.
The measurement of the specific contact resistance was performed using the linear
transmission line method (TLM) using four probes with separate pairs for current-driving and
voltage-sensing. For this, contact pads 140 m 100 m in size and 5, 10, 15, 20, 25 m of
respective spacing were used.
3

Table I: Sample naming scheme and treatment conditions overview.

material

pretreatment

nt
1 etched n-GaN O
2 etched n-GaN O2
O3
3 n-GaN
4 u-GaN
O4
N

no treatment
o
O2 550 C 1 min
o
O2 550 C 5 min
o
O2 400 C 1 min
o
O2 650 C 1 min
N2 550oC 1 min

posttreatment

na
na
N
N2
NO

no anneal
N2 750o C 1 min
N2 900o C 0.5 min
N2 750oC 1 min
+O2 550o C 1 min

Standard current-voltage (I-V) measurements at room temperature were performed on

sample set 1 of the contacts on etched n-GaN wafers. Typical data for the differently treated
samples are shown in Figure 1. The as-deposited n-metal on etched n-GaN (sample 1-nt-na, blue
open circles) exhibited a nonlinear I-V behavior, reflecting a rectifying contact with a high
contact resistance. After thermal anneal in N2 at 750 C for 1 minute (sample 1-nt-N, cyan open
triangles), the I-V behavior turns linear. This indicates achievement of an ohmic n-contact and a
specific contact resistance of 6.9410-5 cm2 was derived. However, under subsequential anneal
in oxygen ambient at 550 C for 1 min (sample 1-nt-NO, red star symbols), such as typical for
ohmic p-contact formation, the n-contact resistance was found to increase to 8.5210-5 cm2.
This indicates that the annealed n-contact degrades during the p-metal annealing in oxygen
ambient. On the other hand, the material that has undergone the pre-deposition thermal treatment
consisting of an RTA step at 550 C for 1 minute in an oxygen ambient (sample 1-O-na, black
open squares) shows a linear I-V behavior. Even without any further post-deposition high
temperature step, a specific contact resistance as low as 3.7810-5 cm2 was achieved. This ncontact performance was better than that of the same contact scheme with regular post deposition
nitrogen annealing at high temperature (1-nt-N).
The conditions of the here proposed oxygen thermal pre-treatment step are typically used
as a post-deposition step for ohmic p-contact formation. Therefore, any p-contact metallization
formed before the n-metallization step should be properly annealed during this pre-treatment
step. No further annealing or plasma process is necessary to the samples. Therefore, no
deterioration to the p-contact is expected.
4

Current (mA)

100

50

sample 1-nt-na
sample 1-nt-N
sample 1-nt-NO
sample 1-O-na
5 m between
two TLM pads

-50

-100
-2

-1

Voltage (V)

FIG. 1. Current-voltage characteristics of different contact schemes on etched n-GaN. Contact pad spacing is 5 m. After Ref 14.

We previously reported an integrated LED fabrication process utilizing this predeposition thermal treatment resulting in a smaller series resistance and a lower voltage across
the LED at high current.14
Here, various additional treatment conditions have been examined to explore the effects
of the annealing ambient, temperature, and treatment duration. Figure 2 shows the I-V
characteristics of as-deposited Ti/Al/Ti/Au contacts on the pre-treated (samples 2-N-na, 2-O-na,
2-O2-na, 2-O3-na, 2-O4-na) compared to the reference of untreated (sample 2-nt-na ) n-type
GaN. The specific contact resistance as determined by TLM is listed in Table II.
The contact on untreated n-GaN (sample 2-nt-na) was not ohmic, and shows the highest
specific contact resistance of 1.110-3 cm2. The as-deposited contact on the nitrogen pretreated n-GaN (sample 2-N-na) was not ohmic, and shows a high specific contact resistance of
8.510-4 cm2. A clear advantage is seen in the as-deposited contacts on oxygen pre-treated nGaN (samples 2-O2-na, 2-O-na, 2-O3-na, and 2-O4-na). Their specific contact resistances were
an order of magnitude lower than those of the other two. The best results were achieved with a
pre-treatment in oxygen ambient annealed at 550 C for 1 minute (sample 2-O-na), resulting in a
specific contact resistance of 5.110-5 cm2 without any post-deposition annealing.

100

2-nt-na, reference
o
2-N-na, N2 550 C 1min
o

2-O-na, O2 550 C 1min

Current (mA)

50

2-O2-na, O2 550 C 5min

o

2-O3-na, O2 400 C 1min

o

2-O4-na, O2 650 C 1min

-50

-100
-0.4

-0.2

0.0

Voltage (V)

0.2

0.4

FIG. 2. Current-voltage characteristics of contacts on etched n-GaN with different pre-deposition treatment. Contact pad spacing
is 5 m.
Table II: Specific contact resistances of contacts on n-type GaN with different pre-deposition treatments.

Sample #

c (10-5

Pre-Treatment Condition
Ambient gas

2-nt-na

Max. temperature

Duration

Reference

cm2)

110

2-N-na

N2

550 C

1 min

85

2-O2-na

O2

550 C

5 min

31

2-O-na

O2

550 C

1 min

5.1

2-O3-na

O2

400 C

1 min

7.2

2-O4-na

O2

650 C

1 min

14

Next, we analyzed the relationship between the proposed pre-deposition treatment and
the regular post-deposition annealing. The specific contact resistance for contacts on untreated nGaN wafers (samples 3-nt-na, 3-nt-N, 3-nt-N2) and on oxygen pre-treated n-GaN (samples 3-Ona, 3-O-N, 3-O-N2) are shown in Table III. The as-deposited contacts on untreated n-GaN
(sample 3-nt-na) showed a value of 2.910-4 cm2. The as-deposited contacts on oxygen treated
n-GaN (sample 3-O-na) showed a lower value of 4.710-5 cm2. For both, the specific contact
resistance reduces even further when the contacts received an additional post-deposition
annealing in nitrogen ambient at 750 C for 1 minute (sample 3-nt-N, 3-O-N), or 900 C for 0.5
minutes (sample 3-nt-N2, 3-O-N2). For contacts on untreated n-GaN, the specific contact
resistance reduces to 2.510-4 cm2 (sample 3-na-N) and 2.110-5 cm2 (sample 3-na-N2). For
6

contacts on oxygen pre-treated n-GaN, the specific contact resistance reduces further to 1.910-5
cm2 (sample 3-O-N) and 1.510-5 cm2 (sample 3-O-N2), respectively.
Apparently, contacts on oxygen pre-treated n-GaN can be further improved by the postdeposition nitrogen annealing. They also showed a lower resistance than the contacts on
untreated n-GaN with the same post-deposition annealing. This additional step can be utilized in
unipolar devices, yet it would likely deteriorate any prior formed p-contacts for bipolar devices.
So, when applied to n-type GaN the oxygen pre-treatment before contact deposition improves the
contact quality independent of the application of any post-deposition nitrogen annealing.
Table III: Specific contact resistances of contacts on as-grown and oxygen treated n-type GaN before and after post-deposition
nitrogen anneal.

Sample # / c (10-5 cm2)

Post-Treatment Conditions

As-grown n-GaN

Oxygen pre-treated n-GaN

As deposited

3-nt-na / 29

3-O-na / 4.69

N2 anneal 750 C 1 min

3-nt-N / 25

3-O-N / 1.9

N2 anneal 900 C 0.5 min

3-nt-N2 / 2.1

3-O-N2 / 1.5

The same Ti/Al/Ti/Au contacts were applied on u-GaN (samples 4-nt and 4-O). The asdeposited contacts on both, treated and untreated u-GaN show a rectifying behavior. The current
across the TLM pads at a certain voltage was larger for the contacts on treated u-GaN (sample 4O) than the contacts on untreated u-GaN (sample 4-nt) (data not shown here), indicating that the
thermal treatment reduces the resistance between the TLM pads, and improves the contacts on uGaN as well.
To understand the mechanism of the oxygen thermal pre-treatment, the binding energy of
the material at the surface of n-GaN (sample 3-nt-na and 3-O-na) and u-GaN (sample 4-nt and 4O) with and without oxygen treatment were characterized by XPS using an Al K source. Figure
3(a) shows the XPS spectra of Ga 3d photoelectrons for the u-GaN samples with and without
oxygen treatment. With the oxygen treatment, the peak for Ga shifts by 0.585 eV towards higher
binding energy. XPS spectra of O 1s photoelectrons before and after the thermal annealing
treatment are shown in Figure 3(b). We find that by the oxygen thermal treatment the peak
intensity corresponding to O 1s photoelectrons increases by 50%. When the same analysis is
performed for oxygen treated n-type GaN (data not shown here), the photoelectron peak of Ga is

found shifted by about 0.1 eV towards higher binding energy, and the peak intensity of O 1s
photoelectrons is found to increase 80% by the treatment.
O 1s

Ga 3d
untreated
19.70 eV

Intensity (arb. unit)

Intensity (arb. unit)

treated
20.29 eV

shift
0.585 eV
23

22

21

20

19

18

17

538

no shift in
peak position

increase
150%

536

Binding Energy (eV)

534

532

530

528

526

Binding Energy (eV)

FIG. 3: XPS spectra of (a) Ga 3d and (b) O 1s at the surface of u-GaN with and without oxygen thermal anneal treatment.

The shift of the Ga 3d peak suggests that by the oxygen thermal annealing treatment the Fermi
level EF at the surface of the u-GaN and n-GaN samples moves closer to the conduction-band
edge. This indicates that a highly n-doped layer on the surface was created by the treatment. It is
known that oxygen acts as a donor in GaN and the pre-treatment in oxygen ambient could
possibly induce such a layer.15
The significant improvement in contact resistance by the oxygen thermal treatment can
be explained by considering an energy-band diagram of the interface:16 The contact of an
untreated sample shows a rectifying behavior, which we attribute to a Schottky barrier between
Ti and GaN. When the surface was treated with the oxygen thermal annealing, additional donors
were created, resulting in an n+-layer on the surface. By this the Schottky barrier was reduced,
and the electron transportation by tunneling was increased. This leads to the improvement of
contact resistance on n-type and unintentionally-doped GaN.
In conclusion, we presented an oxygen thermal annealing treatment on n-type GaN and
unintentionally-doped GaN prior to Ti/Al/Ti/Au contact deposition. By exposing the surface of
n-type GaN to an oxygen rapid thermal anneal, the specific contact resistance for as-deposited
Ti/Al/Ti/Au contacts changes from Schottky behavior to a low resistance of 3.810-5 cm2.
Best pre-treatment results were obtained under thermal annealing at 550 C for 1 minute in an
oxygen ambient. Such conditions can simultaneously anneal pre-existing p-contacts without any
degradation. Further improvement can be achieved with a post-deposition nitrogen anneal at 900
8

C for 0.5 minutes. In this case, the specific contact resistance decreases further to 1.510-5
cm2, yet at the risk of destroying p-contacts in bipolar devices. An XPS analysis shows that with
the oxygen thermal annealing pre-treatment the Fermi level at the semiconductor surface shifts
closer to the conduction band edge, which leads to a reduction of the Schottky barrier height, and
the contact resistance.
This work was supported by a DOE/NETL Solid-State Lighting Contract of Directed
Research under DE-EE0000627. This work was in part also supported by the National Science
Foundation (NSF) Smart Lighting Engineering Research Center (# EEC-0812056).
References
1

I. Akasaki and H. Amano, Jpn. J. Appl. Phys. 36, 5393 (1997).

C. H. Chiu, H. H. Yen, C. L. Chao, Z. Y. Li, P. Yu, H. C. Kuo, T. C. Lu, S. C. Wang, K. M. Lau, and S. J. Cheng,
Appl. Phys. Lett. 93, 081108 (2008)
3
Y. Kuwahara, T. Fujii, T. Sugiyama, D. Iida, Y. Isobe, Y. Fujiyama, Y. Morita, M. Iwaya, T. Takeuchi, S.
Kamiyama, I. Akasaki, and H. Amano, Appl. Phys. Express 4, 021001 (2011)
4
J. K. Ho, C. S. Jong, C. C. Chiu, C. N. Huang, K. K. Shih, L. C. Chen, F. R. Chen, and J. J. Kai, J. Appl. Phys. 86,
4491 (1999).
5
D. H. Youn, M. S. Hao, H. Sato, T. Sugahara, Y. Naoi, and S. Sakai, Jpn. J. Appl. Phys. Part 1 - Regul. Pap. Short
Notes Rev. Pap. 37, 1768 (1998).
6
Wenting Hou, Christoph J.M. Stark, Shi You, Liang Zhao, Theeradetch Detchprohm, Christian Wetzel, Appl. Opt,
51, 5596 (2012).
7
M. E. Lin, Z. Ma, F. Y. Huang, Z. F. Fan, L. H. Allen, and H. Morkoc, Appl. Phys. Lett. 64, 1003 (1994).
8
S. Ruvimov, Z. LilientalWeber, J. Washburn, K. J. Duxstad, E. E. Haller, Z.F. Fan, S. N. Mohammad, W. Kim,
A. E. Botchkarev, and H. Morko, Appl. Phys. Lett. 69, 1556 (1996).
9
B. P. Luther, S. E. Mohney, T. N. Jackson, M. Asif Khan, Q. Chen, and J. W. Yang, Appl. Phys. Lett. 70, 57
(1997).
10
H. W. Jang, C. M. Jeon, J. K. Kim, J-L. Lee, Appl. Phys. Lett. 78, 2015 (2009).
11
Z. F. Fan, S .N. Mohammad, W. Kim, O. Aktas, A. E. Botchkarev, H. Morko, Appl. Phys. Lett. 68, 1672 (1996).
12
J. Y. Chen, C. J. Pan, G. C. Chi, Solid-State Electron. 43, 649 (1999).
13
J. Yan, M.J. Kappers, Z.H. Barber, C.J. Humphreys, Applied Surface Science 234, 328 (2004).
14
W. Hou, T. Detchprohm, C. Wetzel, Int. J. High Speed Electronics and Systems 20, 521 (2011).
15
C. Wetzel, T. Suski, J. W. Ager, E. R. Weber, E. E. Haller, S. Fischer, B. K. Meyer, R. J. Molnar, and P. Perlin,
Phys. Rev. Lett. 78, 3923 (1997).
16
S.M. Sze, and Kwok K. Ng, Physics of Semiconductor Devices 3rd Edition (Wiley-Interscience, New Jersey,
2006) p. 151.
2