Item 987654321/17470

(1)

行政院國家科學委員會專題研究計畫成果報告

費米能階釘札效應,準分子雷射加工與乾蝕刻技術應用於 P

型氮化鎵蕭特基二極體製作之研究

計畫類別：個別型計畫

計畫編號： NSC93-2215-E-018-005-

執行期間： 93 年 08 月 01 日至 94 年 07 月 31 日

執行單位：國立彰化師範大學光電工程技術研究所

計畫主持人：林祐仲

計畫參與人員：朱宥霖,林文祥,吳國禎

報告類型：精簡報告

處理方式：本計畫可公開查詢

中華民國 94 年 7 月 4 日

(2)

1

研究計畫中文摘要：

關鍵字﹕ 氮化鎵，蕭特基二極體，表面費米能階釘札，乾蝕刻

目前，氮化鎵族群的藍色發光二極體及雷射的性能提昇上，除必須有良好

的磊晶結構之外，必須有性能良好的 p 型及 n 型歐姆接觸特性。由於目前 p 型氮

化鎵磊晶層之電洞濃度無法有效的提高，因此要達到金屬/p 型氮化鎵的歐姆接觸

較為困難。在 p 型氮化鎵的歐姆接觸研究已有諸多成果被發表，反觀 p 型氮化鎵

的蕭特基接觸的研究並不多，其已發表之文獻報導所觀測的位障高度差異甚大，

並未有明確的機制被建立。此次研究將主持人先前對 p 型氮化鎵的歐姆接觸的研

究成果應用於 p 型氮化鎵蕭特基二極體中歐姆電極製作，因為良好歐姆電極將有

助提昇 p 型氮化鎵蕭特基二極體之電特性，此外，再利用乾蝕刻處理(反應離子

蝕刻或感應耦合電漿蝕刻技術)或準分子雷射加工處理產生 p 型氮化鎵表面施體

型缺陷引發費米能階釘札效應提高 p 型氮化鎵蕭特基二極體之蕭特基電極與 p

型氮化鎵間之障壁，降低漏電流。本次計畫成果將有助於對 p 型氮化鎵之材料特

性有所了解、製作優良之 p 型氮化鎵蕭特基二極體，而且也能改善金屬/p 型氮化

鎵的歐姆接觸進而提高藍光二極體的效能。另一方面，對於發展氮化鎵的光檢測

也有極大的助益。

研究計畫英文摘要：

Keywords: GaN, Schottky diodes, Barrier height, Surface Fermi level pinning,

Dry etching

GaN continues to grow in importance for the high-brightness light emitting

diodes (LEDs) in the visible and ultraviolet wavelength regions. For the case of the

highly efficient LED, several critical problems must be addressed, such as high hole

concentration in p-type GaN (p-GaN) and low-resistance ohmic contact in the

fabrication processes for the devices. High-quality ohmic and Schottky contacts are

required for the improvement of these devices. The ohmic-contact characteristics of

metal contact on p-GaN have been relatively established. Only a few studies of barrier

height on p-GaN have been reported in the literature, likely due to the difficulties in

measuring diodes with large leakage and large series resistance. The contacts tend to

exhibit very leaky Schottky characteristics. Consequently, the mechanism of current

flow through the interface has not been established and the exact value of Schottky

barrier height has not yet been estimated. The p-GaN ohmic-contact technique

reported in our previous published paper shall be applied in this study. Good ohmic

contacts will help to obtain the actual p-GaN Schottky characteristics. In addition, the

applications of the dry etching and an excimer laser treatment in the fabrication of

p-GaN Schottky contacts will induce the formation of the surface defects on the

p-GaN surface, the pinning of the Fermi level on the p-GaN surface and the

enhancement of the Schottky barrier height at the metal/p-GaN interface. So in this

plan, we are devoted to improve the p-GaN Schottky contact.

(3)

2

計畫成果自評：

本次計畫研究內容與原計畫相符程度達 80%，並達成預期目標，而且研究

成果五篇論文均已發表於國際知名期刊，獲得國際學者肯定，極具學術與實際應

用之價值。

結案報告內容：

此次專題計畫研究成果已撰寫成五篇論文並發表於美國國際期刊，依序列於

下：

1. Y. J. Lin, 2005, March

“Appl

i

c

a

t

i

on

of

t

he

t

he

r

mi

oni

c

f

i

e

l

d

e

mi

s

i

on

mode

l

i

n

t

he

study of a Schottky barrier of Ni on p-GaN from current-v

ol

t

age

mea

s

ur

eme

nt

s

”

,

Appl. Phys. Lett. Vol.86, 122109.

2. Y. J. Lin and Y. L. Chu, 2005, May

“

Effect of reactive ion etching-induced

defects on the surface band bending of heavily Mg-doped p-type GaN”

,

J. Appl.

Phys. Vol.97, 104904.

3. Y. J. Lin, Y. L. Chu, Y. S. Huang, and H. C. Chang, 2005, May “

Optical and

electrical properties of heavily Mg-doped GaN upon (NH

4

)

2

S

x

treatment”

, Appl.

Phys. Lett. Vol.86, 202107.

4. Y. J. Lin,

2005,

J

a

n.

/

Fe

b.

“

Electrical properties of Ni/Au and Au contacts on

p-type Ga

N”

,

J. Vac. Sci. Technol. B Vol.23, 48.

5. Y. J. Lin

a

nd C.

W. Hs

u,

2004,

Se

p.

“

Study of Schottky barrier heights of

indium-tin-oxide on p-GaN using x-ray photoelectron spectroscopy and

current-voltage measurements”

,

J

.

El

e

c

t

r

on.

Ma

t

e

r

.

Vol

.33, 1036.

(4)

Application of the thermionic field emission model in the study of a

Schottky barrier of Ni on

p

-GaN from current–voltage measurements

Yow-Jon Lina兲

Institute of Photonics, National Changhua University of Education, Changhua 500, Taiwan, Republic of China

共Received 24 May 2004; accepted 7 February 2005; published online 18 March 2005兲

Barrier height values of Ni contacts to Mg-doped p-type GaN共p-GaN兲 were obtained from current–

voltage measurements in this study. The induced deep level defect band through high Mg doping led to a reduction of the depletion layer width in the p-GaN near the interface and an increase in the probability of thermionic field emission. It also resulted in an increase in current flow under forward bias condition, which was not analyzed using the thermionic emission model. Further, the calculated barrier height value of Ni contacts to p-GaN using the thermionic field emission model is in good agreement with the value of 1.9 eV obtained from x-ray photoelectron spectroscopy

GaN has a 3.4 eV direct gap at room temperature and has attracted a lot of interest because of its application to optical devices in the short-wavelength region. Other devices that have been demonstrated so far include ultraviolet Schottky barrier photodetectors, solar-blind Schottky photo-diodes, metal–semiconductor field effect transistors, and high

electron mobility transistors.1–4 High-quality ohmic and

Schottky contacts are required for the improvement of these devices. Since the sum of the barrier height of the n and

p types equals the band gap 3.4 eV, a high barrier height is

expected after the metals have been deposited on the

p-GaN.5 Recent reviews and studies of Schottky contact

properties on p-GaN have been published.6–11Some reported

values for barrier height for different metals are

0.49– 0.50 eV for Pt,6,80.57 eV for Au,60.65 eV for Ti,6and 1.3 eV for Pd.9For Ni there is a fairly large discrepancy in reported values, ranging from 0.49 to 2.87 eV.6,7,10,11 Conse-quently, the mechanism of current flow through the interface has not been established and the exact value of the barrier height has not yet been estimated using the current–voltage

共I–V兲 measurement. The observed barrier height value

from capacitance–voltage 共C–V兲 measurements has been

reported.7 However, Yu et al.7 did not obtain the barrier

height value from the I – V measurements. In this study, the barrier height value of the Ni/ p-GaN samples was first suc-cessfully obtained from the I – V measurements. Further, the calculated barrier height value of Ni contacts to p-GaN using

the thermionic field emission共TFE兲 model is in good

agree-ment with the value of 1.9 eV obtained from x-ray

photo-electron spectroscopy共XPS兲 measurements.

The epitaxial layers used in the experiments were grown on c-plane sapphire substrates using a metalorganic chemical vapor deposition system. Trimethylgallium, ammonia, and bis-cyclopentadienylmagnesium were used as the Ga, N, and Mg sources, respectively. An undoped GaN buffer layer with a thickness of 650 nm was grown on the sapphire substrate at 520 ° C, followed by the growth of a Mg-doped p-GaN layer

共762 nm兲 at 1100 °C. Mg concentration 共关Mg兴兲 was ⬃6

⫻1019_cm−3 _{for all samples. The grown samples were}

an-nealed for the purpose of generating holes at 750 ° C for

30 min in N2ambient. According to the Van der Pauw-Hall

measurements, we calculated the hole concentration to be 3.6⫻1017_cm−3_{. The samples were cleaned in chemical clean}

solutions of trichloroethylene, acetone, and methanol. Planar-type Schottky contacts were formed by electron-beam

evaporation. Using the lift-off technique, Ni/ Au

共5 nm/5 nm兲 ohmic contacts were deposited and annealed at

500 ° C in air ambient for 10 min. The fabricated process of ohmic contacts combined with the low specific contact

resis-tance has been previously reported.12 Good ohmic contacts

help to obtain the actual Schottky characteristics from I – V

measurements. Then, Ni共5 nm兲 Schottky contact with

circu-lar patterns 共200␮m in diameter兲 was directly deposited.

The Schottky diodes were measured by the I – V method us-ing a Keithley Model-4200-SCS/F semiconductor character-ization system at 300, 350 and 400 K, respectively.

Figure 1 shows the typical semilog current density–

voltage共J–V兲 characteristics of Ni/p-GaN Schottky diodes

at 300 K. The fitting curve using the TFE model is also shown in Fig. 1. Yu et al.7indicated that the slope of ln I – V curves could not be analyzed using the thermionic emission

共TE兲 model, due to the carrier transport with a tunnel

com-ponent. In addition, these reports13,14 provide theoretical analysis of current tunneling through a Schottky barrier

共q␾B兲. The J–V characteristic in the presence of tunneling

can be described by the relation13,14

a兲

Electronic mail: [email protected]

FIG. 1. J – V curve of Ni/ p-GaN Schottky diodes under forward bias con-dition at 300 K and the fitting curve to the J – V characteristic in the TFE regime.

APPLIED PHYSICS LETTERS 86, 122109共2005兲

(5)

J = J0exp共qV/E0兲, 共1兲 E0= E00coth共E00/kT兲, 共2兲 J0= A*T关␲E00q共␾B− V −␰兲兴0.5 k cosh共E00/kT兲 exp

冋

− q␰ kT − q E₀共␾B−␰兲

册

, 共3兲 E00=共qh/4␲兲共N/m*␧兲0.5, 共4兲

where J is the observed current density of Schottky diodes under forward bias condition, m*_共m*_{= 0.6m}

0, m0is the mass

of hole at rest兲15is the hole effective mass,␧ 共␧=9.5␧₀,␧₀is the permittivity in vacuum兲16 is the dielectric constant of GaN, A*共A*= 103.8 A / cm2K2兲17is the effective Richardson constant of p-GaN, h is Planck’s constant, V is voltage, q is the electron charge, N is the doping concentration, and␰ is equal to共EF− EV兲/q. EV is the valence band maximum and

E_F is the position of the Fermi level. ␰ was assumed to be

equal to 0.1 V in this study. According to the fitting J versus

V curve of samples using Eqs.共1兲–共3兲, the value of E0, E00,

and q␾B can be calculated to be 66 meV, 65 meV, and

1.8 eV, respectively. Then, the calculated value 共⬃7

⫻1019_cm−3_{兲 of N using Eq. 共4兲 is similar to 关Mg兴, which}

indicated that the tunneling current under forward bias

con-ditions took place because of the high Mg doping 关or the

deep level defect 共DLD兲 band induced by high Mg

doping兴.18,19 Kwak et al.18 have suggested that the current transport at the metal/high Mg doped p-GaN interface was dominated by a DLD band which was induced by high Mg doping. Kwak et al.19have also suggested that the DLD band had a large density defect, over 1019cm−3 size, existing in the p-GaN films and that the density of the DLDs increased as关Mg兴 increased. In addition, Shiojima et al.20 have found carrier capture and emission from acceptor-like DLDs for Ni/ p-GaN Schottky diodes. Further, it is worth noting that the calculated value of E00/ kT is slightly larger than 1, which

implies the current transport processes a tunnel component.14 On the other hand, the observed current is too small to obtain the fitting parameters for the J – V characteristic in the field

emission共FE兲 regime for Ni/p-GaN. This suggests that the

FE model cannot be used to study the q␾Bof Ni/ p-GaN in

this case. In addition, the E00of 65 meV is not much greater

than kT 共kT=26 meV, at 300 K兲, so the FE will not take

place.13The effective resistance of the Schottky barrier in the FE regime is quite low, so the FE model is often used for ohmic contact.13

Figure 2 shows the typical semilog J – V characteristics of Ni/ p-GaN Schottky diodes at 350 and 400 K, respec-tively. The fitting curves using the TFE model are also shown in Fig. 2. We find that the barrier heights can readily be obtained by fitting the J – V curves using Eqs.共1兲–共3兲. From the TFE theory, a linear fit to the data for the J – V

measure-ments at 350 K yields E₀= 67 meV and q␾_B= 1.8 eV,

whereas a linear fit to the data for the J – V measurements at 400 K yields E0= 68 meV and q␾B= 1.8 eV. Both values for

the Schottky barrier height are in good agreement with the value calculated for J – V measurements at 300 K. These re-sults indicate that a TFE model can quantitatively explain the observed large forward leakage currents due to the existence of the induced high density of the acceptor-type DLDs by

high Mg doping. We conclude that the creation of a large number of DLDs in p-GaN leads to the reduction of the depletion layer width in the p-GaN near the interface, an increase in the probability of the TFE, and gives rise to large forward leakage. Consequently, the forward current transport with a tunnel component is not analyzed using the TE model. In order to further confirm whether the barrier-height value of Ni/ p-GaN Schottky diodes is 1.8 eV or not, XPS is used to study the surface Fermi level position within the band gap for Ni overlayer on p-GaN. XPS measurements

were performed using a monochromatic Al K_␣ x-ray source.

The barrier height was determined from the XPS data using the following relation.21,22This equation was previously em-ployed by Tracy et al.23 for the calculation of the barrier heights共q␾n兲 of metals on n-GaN:

q␾_n= E_G− Eiv +共Ecore i _{− E} core m _{兲 = E} G−共Ecore m _{− E} VC兲共5兲

EGis the band gap of the semiconductor, Ecore

m

is the binding energy of the semiconductor core-level peak following metal deposition, E_corei is the initial binding energy of the core-level peak, E_Vi is the initial binding energy of the EV of the

semi-conductor, and EVCis equal to共Ecore

i _{− E} V

i_{兲. All binding}

ener-gies are measured relative to the EF. For the calculation of

the barrier heights 共q␾_p兲 of Ni/p-GaN, the equation is ex-pressed as

FIG. 2. Forward J – V characteristic of Ni/ p-GaN Schottky diodes as a function of temperature.

FIG. 3. The left-hand spectrum shows the Ga 3d core-level peak on

p-GaN without a Ni overlayer. The right-hand panel presents the spectrum

of the valence-band region. A linear fit is used to determine the energy of the valence-band edge.

122109-2 Yow-Jon Lin Appl. Phys. Lett. 86, 122109共2005兲

(6)

q␾p= Ecore Ni

− EVC 共6兲

where E_coreNi is the binding energy of the p-GaN core-level peak following Ni deposition.

Figure 3 shows an example of the Ga 3d core level and the valence-band spectrum collected on a p-GaN sample without a Ni overlayer. The EVCis calculated to be 17.8 eV.

This value is in good agreement with the value of 17.8 eV

reported by Hashizume et al.,24 Bermudez,25 Waldrop and

Grant,26and Wu and Kahn.27Figure 4 shows the Ga 3d core

level at the Ni/ p-GaN interface. The spectra determine the Ga 3d binding energy E_coreNi relative to the EF. In Fig. 4, we

can see that the E_coreNi is equal to 19.7 eV. Therefore, the q␾p

was calculated to be 1.9 eV, according to Eq.共6兲. This value is similar to the value of 1.8 eV obtained from J – V measure-ments. In addition, Rickert et al.28suggested that the barrier height value of the Ni/ p-GaN sample is approximately 1.9 eV, which supported our observed results from the XPS and J – V measurements.

In summary, the barrier height value of Ni/ p-GaN Schottky diodes was obtained from J – V measurements in this study. For Ni/ p-GaN Schottky diodes, the calculated barrier height value using the TFE model is in good agree-ment with the value of 1.9 eV obtained from x-ray photo-electron spectroscopy measurements. The induced DLD band through high Mg doping would lead to the reduction of the depletion layer width in the p-GaN near the interface, an increase in the probability of the TFE, and an increase in

current flow under forward bias conditions, which was not analyzed using the TE model.

This project is supported by National Science Council of Taiwan, Republic of China, under Contract No. NSC 93-2215-E-018-005. The XPS was kindly provided from the NTU Instrumentation Center at Taipei National Taiwan University.

1

O. Katz, V. Garber, B. Meyler, G. Bahir, and J. Salzman, Appl. Phys. Lett.

79, 1417共2001兲.

2

M. A. Khan, J. N. Kuznia, D. T. Olson, M. Blasingame, and A. R. Bhat-tarai, Appl. Phys. Lett. 63, 2455共1993兲.

3

M. A. Khan, J. N. Kuznia, A. R. Bhattarai, and D. T. Olson, Appl. Phys. Lett. 62, 1786共1993兲.

4

N. Biyikli, O. Aytur, I. Kimukin, T. Tut, and E. Ozbay, Appl. Phys. Lett.

81, 3272共2002兲.

5

M. W. Wang, J. O. McCaldin, J. F. Swenberg, T. C. McGill, and R. J. Hauenstein, Appl. Phys. Lett. 66, 1974共1995兲.

6

T. Mori, T. Kozawa, T. Ohwaki, Y. Taga, S. Naagai, S. Yamasaki, S. Asami, N. Shibata, and M. Koike, Appl. Phys. Lett. 69, 3537共1996兲. 7

L. S. Yu, D. Qiao, L. Jia, S. S. Lau, Y. Qi, and K. M. Lau, Appl. Phys. Lett. 79, 4536共2001兲.

8

X. A. Cao, S. J. Pearton, G. Dang, A. P. Zhang, F. Ren, and J. M. Van Hove, Appl. Phys. Lett. 75, 4130共1999兲.

9

P. J. Hartlieb, A. Roskowski, R. F. Davis, W. Platow, and R. J. Nemanich, J. Appl. Phys. 91, 732共2002兲.

10

D. L. Hibbard, R. W. Chuang, Y. S. Zhao, C. L. Jensen, H. P. Lee, Z. J. Dong, R. Shih, and M. Bremser, J. Electron. Mater. 29, 291共2000兲. 11

K. Shiojima, T. Sugahara, and S. Sakai, Appl. Phys. Lett. 74, 1936共1999兲. 12

J. K. Ho, C. S. Jong, C. N. Huang, C. C. Chiu, K. K. Shih, L. C. Chen, F. R. Chen, and J. J. Kai, J. Appl. Phys. 86, 4491共1999兲.

13

Michael Shur, Physics of Semiconductor Devices共Prentice-Hall, Engel-wood Cliffs, NJ, 1990兲.

14

H. Morkoç, Nitride Semiconductors and Devices共Springer, Berlin, 1999兲. 15

C. Merz, M. Kunzer, U. Kaufmann, I. Akasaki, and H. Amano, Semicond. Sci. Technol. 11, 712共1996兲.

16

M. Razeghi and A. Rogalski, J. Appl. Phys. 79, 7433共1996兲. 17

J. I. Pankove, S. Bloom, and G. Harbeke, RCA Rev. 36, 163共1975兲. 18

J. S. Kwak, O. H. Nam, and Y. Park, Appl. Phys. Lett. 80, 3554共2002兲. 19

J. S. Kwak, O. H. Nam, and Y. Park, J. Appl. Phys. 95, 5917共2004兲. 20

K. Shiojima, T. Sugahara, and S. Sakai, Appl. Phys. Lett. 77, 4353共2000兲. 21

J. R. Waldrop and R. W. Grant, Appl. Phys. Lett. 52, 1794共1988兲. 22

K. M. Tracy, P. J. Hartlieb, S. Einfeldt, F. Davis, E. H. Hurt, and R. J. Nemanich, J. Appl. Phys. 94, 3939共2003兲.

24

T. Hashizume, S. Ootomo, S. Oyama, M. Konishi, and H. Hasegawa, J. Vac. Sci. Technol. B 19, 1675共2001兲.

25

V. M. Bermudez, J. Appl. Phys. 80, 1190共1996兲. 26

C. I. Wu and A. Kahn, J. Appl. Phys. 86, 3209共1999兲. 28

K. A. Rickert, A. B. Ellis, J. K. Kim, J. L. Lee, F. J. Himpsel, F. Dwikusuma, and T. F. Kuech, J. Appl. Phys. 92, 6671共2002兲.

FIG. 4. Ga 3d core level at the Ni/ p-GaN interface. The binding energy is referenced to the Fermi level.

122109-3 Yow-Jon Lin Appl. Phys. Lett. 86, 122109共2005兲

(7)

Effect of reactive ion etching-induced defects on the surface band bending

of heavily Mg-doped

p

-type GaN

Yow-Jon Lina兲and Yow-Lin Chu

共Received 16 August 2004; accepted 25 February 2005; published online 29 April 2005兲

The effect of reactive ion etching-induced defects on the surface band bending of heavily Mg-doped

p-type GaN共p-GaN兲 was investigated in this study. According to the observed results from x-ray

photoelectron spectroscopy and secondary-ion-mass spectroscopy共SIMS兲 measurements, we found

that the formation of more nitrogen-vacancy-related defects created near the surface by reactive ion etching technique would lead to an increase in the surface band bending, a shift of the surface Fermi level toward the conduction-band edge, the reduction of the current flow at the metal/etched p -GaN interface, and an increase in the barrier height at the metal/etched p-GaN interface. In addition, from the SIMS measurements, it is suggested that the depth of the nitrogen-deficient near-surface region resulting from the dry-etch process is about 60 nm. © 2005 American Institute

of Physics.关DOI: 10.1063/1.1894580兴

I. INTRODUCTION

The recent progress in GaN-based optoelectronic

de-vices, such as blue light-emitting diodes 共LEDs兲 and laser

diodes共LDs兲,1–3points to the need of a deeper understanding of the optical and electrical characteristics of Mg-doped

p-type GaN共p-GaN兲. An important area in dry etching is the

characterization of damage which is produced in the semi-conductor through the bombardment of energetic ions. This induced damage, which may exist as vacancy complexes, dislocations, and recombination centers, can degrade the op-tical and electrical performances of devices. However, very little work has been done in investigating dry etch-induced defects in respect to heavily Mg-doped p-GaN. In this ar-ticle, we report the findings from a study of the effect of reactive ion etching-induced defects on the electrical proper-ties of heavily Mg-doped p-GaN. We find that according to the observed result from x-ray photoelectron spectroscopy

共XPS兲 and secondary-ion-mass spectroscopy 共SIMS兲

mea-surements, the nitrogen-vacancy-related defects induced by the reactive ion etching 共RIE兲 technique will lead to an in-crease in surface band bending and a shift of the surface Fermi level toward the conduction-band edge. In addition, from the SIMS measurements, it is suggested that the depth of the nitrogen-deficient near-surface region resulting from a dry-etch process is about 60 nm. Upon p-type doping, the changed donor defect formation energy will decrease in

tan-dem with the Fermi-level 共EF兲 position.

4,5

The nitrogen-vacancy-related defect is easily formed in the p-GaN films and the native defect of p-GaN.4According to the theoretical predictions,6 the only native defect with a relevant concen-tration in p-GaN is the nitrogen-vacancy-related defect.7As a result, we deduce that further nitrogen-vacancy-related de-fects may exist in the p-GaN near-surface region following etching. Another goal of this paper is to show that the RIE

represents a suitable technique choice for the fabrication of heavily Mg-doped p-GaN Schottky diodes with good quali-ties.

II. EXPERIMENTAL PROCEDURE

The epitaxial layers used in the experiments were grown on c-plane sapphire substrates using a metal-organic chemical-vapor deposition system. Trimethylgallium, ammo-nia, and biscyclopentadienylmagnesium were used as the Ga, N, and Mg sources, respectively. An undoped GaN buffer layer with a thickness of 650 nm was grown on the sapphire substrate at 520 ° C, followed by the growth of a Mg-doped

GaN layer共762 nm,3.6⫻1017_cm−3_{兲 at 1100 °C. The grown}

samples were annealed for the purpose of generating holes,

at 750 ° C for 30 min in N2 ambient. The Mg concentration

共关Mg兴兲 was 6⫻1019_cm−3 _{for all samples. The}_{关Mg兴 in}

Mg-doped GaN was measured using SIMS. The samples were cleaned in chemical clean solutions of trichloroethylene, ac-etone, and methanol共referred to as as-cleaned p-GaN兲. Then, some of the as-cleaned samples were etched using the RIE

technique in BCl3plasma with a rf power of 150 W. During

the etching process, the BCl3flow rate was 5 SCCM

共stan-dard cubic centimeter per minute兲 and the chamber pressure

was 4 Pa. The etched depth is 100 nm. The etched sample was referred to as etched p-GaN. The as-cleaned and etched samples were then inserted into the vacuum chamber of a

ESCA PHI 1600 system. We took a Au 4f7/2peak and a Cu

2p3/2peak for energy reference purposes.

III. EXPERIMENTAL RESULTS AND DISCUSSION

Figure 1 shows the Ga 3d core-level XPS spectra on the as-cleaned and etched p-GaN surfaces. We found that the Ga 3d core-level spectra of the etched sample shift toward the high binding-energy side, compared with that of the as-cleaned one. This indicated that the Ga 3d core-level spectra of the as-cleaned sample after etching treatment underwent a

a兲_{Author to whom correspondence should be addressed; electronic mail:}

[email protected]

JOURNAL OF APPLIED PHYSICS 97, 104904共2005兲

(8)

shift of 0.7 eV, and that the etching treatment gave rise to a shift of the surface Fermi level toward the conduction-band edge, resulting in an increase in the surface band bending.8,9 In Fig. 1, we also found that the full width at half maximum in the Ga 3d spectra of as-cleaned p-GaN is slightly larger than that of etched p-GaN. This may be due to that the fact that the contaminant and residual oxides on the p-GaN sur-face were removed during the etching process.

Figure 2 shows the XPS spectra of the as-cleaned and etched p-GaN surface near the valence-band edge region.

The valence-band maximum共VBM兲 is defined in a standard

way by the linear extrapolation of the low binding-energy edge of the valence-band spectrum. In Fig. 2, we can see that the VBM of the as-cleaned and etched samples is found at 1.0 and 1.7 eV below the EF, respectively. In addition, for

the as-cleaned or etched samples, the binding energy of the

Ga 3d core level 共18.8 eV for the as-cleaned sample and

19.5 eV for the etched sample, as shown in Fig. 1兲 with

respect to the VBM is 17.8 eV, which is in good agreement with previous results.10,11

To investigate the characteristics of defects induced by RIE technique, the as-cleaned and etched p-GaN surfaces were measured using XPS quantitative analysis. The relative Ga/ N atomic concentration ratio was determined from the integration of Ga 3d and N 1s peaks. In this work, the exact sensitivity factor for p-GaN was not known, so the Ga/ N ratio on the as-cleaned p-GaN surface was taken as 1. Con-sequently, the Ga/ N ratio on the etched p-GaN surface was calculated to be 1.5. XPS surface analysis revealed a lower

nitrogen atomic concentration on the etched sample. This indicates that the RIE technique may create more nitrogen vacancies on the sub-p-GaN surface. The

nitrogen-vacancy-related defect level above the E_F, which is near the

conduction-band edge,12,13 will ionize and the

nitrogen-vacancy-related defects will become the dominant positively charged donors in the etched sample. For the preservation of the charge neutrality on the etched p-GaN surface, an in-crease in the positive charge for the extrinsic nitrogen-vacancy defects will lead to an increase in the negative charge in the surface space-charge region, an increase in the surface band bending, and a shift of the surface Fermi level toward the conduction-band edge. On the other hand, it has been shown that reactive ion etching n-type GaN surface improves the contact resistance,14which can be attributed to the formation of extrinsic donor-type nitrogen-vacancy-related defects. The suggestion is consistent with the above result of XPS analysis. Besides, we found that the reduction of the surface state, related to nitrogen-vacancy defects on the p-GaN surface, led to a reduction in the surface band bending by 0.25 eV in a previous study.15 In contrast, more nitrogen-vacancy-related defects created near the surface by the RIE technique will lead to an increase in the surface band bending of p-GaN.

Figure 3 shows the depth profiles of Ga/ N ratios of as-cleaned p-GaN and etched p-GaN, respectively. The inten-sity of Ga SIMS divided by that of N SIMS equals the Ga/ N ratio. In Fig. 3, we find that the Ga/ N ratio of etched samples is larger than that of as-cleaned samples in the near-surface region. This implies that more nitrogen-vacancy-related de-fects were formed in the etched p-GaN films. We also find that the depth of the nitrogen-deficient near-surface region resulting from a dry-etch process is about 60 nm. The value is in good agreement with the value of 40– 60 nm reported by Cao et al.16 and Kent et al.17

Figure 4 shows the current–voltage共I–V兲 characteristics

of Ni/ Au 共20/20 nm, deposition used an electron-beam

evaporator兲 contacts on as-cleaned and etched p-GaN. The

gap spacing of the Ni/ Au metals pads is 25␮m. These I – V characteristics were observed using the transmission line

method共TLM兲. The fabrication process of the TLM pattern

has been reported previously.18 The I – V characteristics of

FIG. 1. Ga 3d core-level XPS spectra of as-cleaned and etched p-GaN.

FIG. 2. XPS spectra of the as-cleaned and etched p-GaN surface near the valence-band edge region.

FIG. 3. Depth profiles of Ga/ N ratios of the as-cleaned and etched p-GaN samples.

104904-2 Y. Lin and Y. Chu J. Appl. Phys. 97, 104904共2005兲

(9)

the TLM patterns were measured using a Keithley Model-4200-SCS/F semiconductor characterization system. Com-paring the I – V curves shown in Fig. 4, the I – V curve for the as-cleaned sample is better than that of the etched sample. In addition, the I – V curve for the as-cleaned sample, with a smaller turn on voltage and larger slope, is superior to that of the etched sample. The worse performance of the etched samples is attributable to the formation of the higher barrier between the Ni/ Au and p-GaN and an increase in the nitrogen-vacancy-related defects, suppressing an effect of the acceptorlike deep level defects19共DLDs兲 on the carrier trans-port at the Ni/heavily Mg-doped GaN interfaces, as well as the reduction of the hole concentration and the probability of the tunneling transmission through the DLD band for holes at the metal/ p-GaN interface.19,20The explanation for this is given later. Besides, in Figs. 1 and 2, we can see that the downward surface band bending of the etched samples in-deed increases. This indirectly implies that an increase in the surface band bending can lead to the degradation in the I – V characteristics of the etched sample. In this regard, the dry etching technique can be applied in the fabrication process of

p-GaN Schottky diodes, because more

nitrogen-vacancy-related defects formed below the p-GaN surface lead to an increased Schottky barrier for the transport of holes.

For Schottky diodes, secure ohmic contacts help to ob-tain the actual Schottky characteristics. Kwak et al. sug-gested that high Mg doping would lead to the formation of secure ohmic contacts of metals/ p-GaN.19In order to obtain the good qualities of p-GaN Schottky diodes, the heavily Mg-doped p-GaN has to be used to form the low-resistivity ohmic contact. Nevertheless, Kwak et al. found that the cur-rent transport at the metal/high Mg-doped p-GaN interface was dominated by a deep level defect band which was

in-duced by high Mg doping.19Kwak et al. suggested that the

DLD band possesses a large defect density over 1019_cm−3

existing in the p-GaN films, and that the density of the DLDs

was enlarged as 关Mg兴 increased.20 In addition, Shiojima et

al.21 found carrier capture and emission from acceptorlike

DLDs for Ni/ p-GaN Schottky diodes. According to these results reported by Kwak et al. and Shiojima et al.,19–21 we deduced that the heavily Mg-doped p-GaN used to fabricate the Schottky diodes might lead to the formation of the low barrier height and a larger leakage of current at the Schottky

metals/ p-GaN interface. On the basis of this evidence, we concluded that the regions of Schottky contacts of high Mg-doped p-GaN could be etched by the RIE technique after the formation of good ohmic contacts, and more nitrogen-vacancy-related defects were produced in p-GaN near the surface, suppressing an effect of the acceptorlike DLDs on

p-GaN Schottky diodes. Therefore, following the deposition

of metals on the etched Schottky contacts of heavily Mg-doped p-GaN samples, we expect that the higher barrier height and smaller leakage current will be easily obtained. This will be further investigated in the future. Finally, we predict that the application of the RIE technique in the fab-rication of p-GaN Schottky diodes will result in the forma-tion of a high barrier height and a smaller leakage of current at the interfaces.

IV. CONCLUSIONS

In summary, the effect of defects created by the RIE technique on the electrical properties of heavily Mg-doped

p-GaN has been investigated. We found that the formation of

more nitrogen-vacancy-related defects created near the sur-face by the RIE technique would lead to an increase in the downward band bending. This indicates that the etching treatment gives rise to a shift of the surface Fermi level to-ward the conduction-band edge, which is in good agreement with the observed result from the measurements of the valence-band spectrum. Dry etching could be useful for the formation of the high barrier height and the low leakage of current at the metals/heavily Mg-doped p-GaN interface. This fact offers hope for the application of the RIE technique in the fabrication process of heavily Mg-doped p-GaN Schottky diodes with good qualities. We plan to experimen-tally fabricate such Schottky diodes in the near future.

ACKNOWLEDGMENT

This project is supported by National Science Council of Taiwan, Republic of China, under Contract No. NSC 93-2215-E-018-005.

1

M. Osiński, Gallium-Nitride-based Technologies 共SPIE, Washington, 2002兲.

2

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

3

J. Neugebauer and C. G. Van de Walle, Phys. Rev. B 50, 8067共1994兲. 5

S. G. Lee and K. J. Chang, Semicond. Sci. Technol. 14, 138共1999兲. 6

J. Neugebauer and C. G. Van de Walle, Mater. Res. Soc. Symp. Proc. 395, 645共1996兲.

7

U. Kaufmann, M. Kunzer, M. Maier, H. Obloh, A. Ramakrishnan, B. Santic, and P. Schlotter, Appl. Phys. Lett. 72, 1326共1998兲.

8

K. A. Rickert, A. B. Ellis, F. J. Himpsel, J. Sun, and T. F. Kuech, Appl. Phys. Lett. 80, 204共2002兲.

9

G. Landgren, R. Ludeke, Y. Jugnet, J. F. Morar, and F. J. Himpsel, J. Vac. Sci. Technol. B 2, 351共1984兲.

10

T. Hashizume, S. Ootomo, S. Oyama, M. Konishi, and H. Hasegawa, J. Vac. Sci. Technol. B 19, 1675共2001兲.

12

C. G. Van de Walle, in Properties of Gallium Nitride and Related

Semi-conductors, edited by J. H. Edger, S. Strite, I. Akasaki, H. Amano, and C.

Wetzel共INSPEC, London, 1999兲, p. 275. 13

T. L. Tansley and R. J. Egan, Phys. Rev. B 45, 10942共1992兲. 14

Z. Fan, S. N. Mohammad, W. Kim, O. Aktas, A. E. Botchkarev, and H. Morkoç, Appl. Phys. Lett. 68, 1672共1996兲.

FIG. 4. I – V characteristics of Ni/ Au contacts on as-cleaned and etched

p-GaN.

(10)

15

Y. J. Lin, Z. L. Wang, and H. C. Chang, Appl. Phys. Lett. 81, 5183 共2002兲.

16

X. A. Cao et al., MRS Internet J. Nitride Semicond. Res. 5S1, W10.8 共2000兲.

17

D. G. Kent et al., Solid-State Electron. 45, 467共2001兲.

18

Y. J. Lin, Z. D. Li, C. W. Hsu, F. T. Chien, C. T. Lee, S. T. Shao, and H. C. Chang, Appl. Phys. Lett. 82, 2817共2003兲.

19

J. S. Kwak, O. H. Nam, and Y. Park, Appl. Phys. Lett. 80, 3554共2002兲. 20

J. S. Kwak, O. H. Nam, and Y. Park, J. Appl. Phys. 95, 5917共2004兲. 21

K. Shiojima, T. Sugahara, and S. Sakai, Appl. Phys. Lett. 77, 4353共2000兲.

(11)

Optical and electrical properties of heavily Mg-doped GaN

upon

„

NH

₄

…

₂

S

_x

treatment

Yow-Jon Lina兲 and Yow-Lin Chu

Institute of Photonics, National Changhua University of Education, Changhua 500, Taiwan, Republic of China

Y. S. Huang

Department of Electronic Engineering, National Taiwan University of Science and Technology, Taipei 106, Taiwan, Republic of China

Hsing-Cheng Chang

Department of Automatic Control Engineering, Feng Chia University, Taichung 407, Taiwan, Republic of China

共Received 12 January 2005; accepted 29 March 2005; published online 10 May 2005兲

We have employed the photoluminescence 共PL兲 and surface photovoltage spectroscopy 共SPS兲

measurements to study the effects of共NH4兲2Sxtreatment on the optical and electrical properties of

p-type GaN共p-GaN兲 in this study. From the PL and SPS measurements, it is suggested that the 共MgGa–VN兲2+ 共MgGa:Ga vacancies occupied by Mg; VN:nitrogen vacancies兲 complexes near the p-GaN surface region were transformed into the 共MgGa-SN兲0 共SN:N vacancies occupied by S兲

complexes after共NH4兲2Sxtreatment, which resulted in the reduction of the ⬃2.8-eV PL intensity

Institute of Physics. 关DOI: 10.1063/1.1926404兴

In recent years, GaN-related wide band gap semiconduc-tors have become the materials of choice for the fabrication of light emitting diodes operating in the green, blue, and violet regions of the visible spectrum.1,2The fabrication of high-quality ohmic contacts on n- and p-type GaN is essen-tial for improving the performance of optoelectronic devices.

However, ohmic contact formation to p-type GaN 共p-GaN兲

alloyed has been difficult to realize due to the absence of metals having a work function larger than that of p-GaN and the absence of p-GaN having a higher hole concentration. Recently, several groups have employed共NH4兲2Sxtreatment

which was designed to improve the electrical properties of

ohmic contacts to p-GaN.3–5 Lee et al. suggested that the

reduction of contact resistivity is due to direct contact of Pt on the clean p-GaN surface, via shift of barrier height for holes at the Pt/p-GaN interface.3 Jang and Seong have indi-cated that surface treatment led to an increase in the hole concentration of p-GaN and the occurrence probability of the field emission for holes at the Pt/p-GaN interface.4However, the effect of 共NH₄兲₂S_x treatment on optical properties of

p-GaN has not been investigated further. To date, the

rela-tionship between the optical and electrical properties of

p-GaN with surface treatment has also not yet been well

understood. For this study, we demonstrate that 共NH4兲2Sx

treatment results in the reduction of⬃2.8-eV

photolumines-cence共PL兲 intensity of the heavily Mg-doped p-GaN sample.

We deduce that the共MgGa–VN兲2+ 共MgGa: Ga vacancies

oc-cupied by Mg, VN: nitrogen vacancies兲 complexes near the

p-GaN surface region were transformed into the 共MgGa

-SN兲0 共SN : N vacancies occupied by S兲 complexes after

共NH4兲2Sxtreatment, which led to the reduction of donor-type

VN-related states near the surface region and an increase in

the hole concentration.4 This can explain why 共NH4兲2Sx

treatment is useful for the improvement of the electrical properties of ohmic contacts to p-GaN.

The epitaxial layers used in the experiments were grown on c-plane sapphire substrates using a metal-organic chemi-cal vapor deposition system. Trimethylgallium, ammonia and bis-cyclopentadienylmagnesium were used as the Ga, N, and Mg sources, respectively. An undoped GaN buffer layer with a thickness of 650 nm was grown on the sapphire substrate at 520 °C, followed by the growth of Mg-doped GaN layer

共762 nm, Mg concentration =6⫻1019_cm−3_{兲 at 1100 °C. The}

grown samples were annealed for the purpose of generating holes, at 750 °C for 30 min in N2ambient. The samples were

cleaned in chemical clean solutions of trichloroethylene,

acetone, and methanol 共referred to as as-cleaned p-GaN兲.

Then, some of the as-cleaned samples were dipped into a

共NH4兲2Sx solution 共S=6%兲 at 60 °C for 30 min 关referred

to as 共NH4兲2Sx-treated p-GaN兴. The as-cleaned and

共NH4兲2Sx-treated samples were then measured via PL and

surface photovoltage spectroscopy共SPS兲. The SPS

measure-ments, which used normalized incident light intensity,6were performed at normal incidence using a fixed grid and probe light chopped at 200 Hz. SPS is a powerful tool for the investigation of surface electronic structure, which has proven to be an extensive source of information about opti-cally active surface states at clean and real surfaces of vari-ous semiconductors, as well as surface with monolayer coverage.7–9

Figure 1 shows the room-temperature PL spectrum of the as-cleaned p-GaN and共NH₄兲₂S_x-treated p-GaN, respectively.

Using a He–Cd laser as an excitation source, the ⬃2.8-eV

PL band was only observed for the all samples due to the

heavily Mg doped.10 According to the reported result by

Kaufmann et al.,11 we find that the generation of the

⬃2.8-eV PL band may be attributed to the MgGa and

Mg_Ga–V_Ncomplexes. In Fig. 1, we find that the PL intensity

a兲_{Author to whom correspondence should be addressed; electronic mail:}

[email protected]

APPLIED PHYSICS LETTERS 86, 202107共2005兲

(12)

of the共NH₄兲₂S_x-treated p-GaN is lower than that of the as-cleaned p-GaN. An explanation for this will be given later.

Figure 2 shows the room-temperature surface photovolt-age 共SPV兲 spectra of the as-cleaned and 共NH4兲2Sx-treated

p-GaN samples, respectively. The slope changes indicated a

one gap state, whose position relative to the conduction band

minimum E_c 共valence band maximum E_v兲 can be deduced

from the position of the downward共upward兲 slope changes

in the spectra.12,13We can see that the upward slope change at⬃3.07 eV indicates population of a donor state, situated at

⬃3.07 eV above the Ev 关Ev+ 3.07 eV, denoted by D1 in

Fig. 2共a兲兴. This state is attributed to the 共Mg_Ga–V_N兲2+

complex.11,14 Kaufmann et al. have suggested that the

MgGa–VN complex may exist in the heavily Mg-doped

p-GaN layer.15The MgGa–VNcomplex which is double

do-nors resulted in the reduction of the hole concentration in the heavily Mg-doped p-GaN sample.15,16However, in Fig. 2共b兲, we did not find population of a donor state, situated at E_v + 3.07 eV. This suggests that共NH4兲2Sxtreatment resulted in

the removal of the donor-type共MgGa–VN兲2+complexes near

the p-GaN surface region. We deduce that that the

共MgGa–VN兲2+complexes near the p-GaN surface region were

transformed into the共MgGa–SN兲0 complexes after共NH4兲2Sx

treatment and the共MgGa–SN兲0 complex near the p-GaN

sur-face region prefers a neutrality charged state, which results in

the reduction of the observed ⬃2.8-eV PL intensity. Kim

et al. have suggested that the reduction of MgGa–VNled to

the reduction of the observed⬃2.8-eV PL intensity.16 In a previous report,17 we have demonstrated that the S atoms might occupy nitrogen-related vacancies near the

共NH4兲2Sx-treated p-GaN surface region. In addition, Oh

et al. have suggested that PL is sensitive to material

surface.18 Besides, we predict that the neutral 共MgGa–SN兲0

complex will not play the role of the compensating donor.

According to the reported results by Lee et al.,3 Jang and

Seong,4and Lee et al.,5we find that the current-voltage char-acteristic of the metals/共NH₄兲₂S_x-treated p-GaN sample is indeed better than that of the metals/as-cleaned p-GaN sample.5This indirectly implies that the formation the neu-tral 共MgGa–SN兲0 complexes near the p-GaN surface region

may lead to an increase in the hole concentration and the improvement in the electrical properties of the metals/

共NH4兲2Sx-treated p-GaN sample. A proof for this will be

given later. On the other hand, we can also see that the downward slope change at 2.9–3.0 eV indicates population of an acceptor state, situated at 0.4–0.5 eV above the E_v,

关denoted by A1 in Fig. 2共a兲兴. The acceptor state is attributed

to the contamination-induced surface state 共i.e., the native oxide or carbon-related contamination兲.19Hartlieb et al.19 in-dicated that the band of surface states may extend as far as 0.6 eV above the E_v, according to the reported results by Wu

and Kahn20and Bermudez.21However, the native oxide and

carbon-related contamination on the p-GaN surface could be effectively removed using共NH4兲2Sxtreatment.

22

As a result, we do not find population of the acceptor state for

共NH4兲2Sx-treated p-GaN关shown in Fig. 2共b兲兴.

Figure 3 shows the room-temperature SPV spectra in the higher photon-energy region. We find that the peak at 3.382 eV represents the band–to–band transition for as-cleaned

p-GaN关shown in Fig. 3共a兲兴. Similarly, we also find that the

peak at 3.396 eV represents the band–to–band transition for

共NH4兲2Sx-treated p-GaN 关shown in Fig. 3共b兲兴. The larger

band gap共Eg兲 of the 共NH4兲2Sx-treated sample than that of the

as-cleaned sample is found 共shown in Fig. 3兲, due to the

reduction of the gap state density during the 共NH4兲2Sx

treatment.23,24This band-gap shift 共⌬Eg兲 is associated with

the Burstein–Moss shift and this shift energy共E兲 can be writ-ten as25

FIG. 1. Room-temperature PL spectra of 共a兲 as-cleaned p-GaN and 共b兲 共NH4兲2Sx-treated p-GaN.

FIG. 2. Room-temperature SPV spectra of 共a兲 as-cleaned and 共b兲共NH4兲2Sx-treated p-GaN.

FIG. 3. Room-temperature SPV spectra of 共a兲 as-cleaned and 共b兲共NH4兲2Sx-treated p-GaN in the higher photon-energy region.

202107-2 Linet al. Appl. Phys. Lett. 86, 202107共2005兲

(13)

E = Eg− Ego= ␲2_h2 2mr *

冉

3N ␲

冊

2/3 , 共1兲

where Eg is the transition energy gap of p-GaN, Ego is the

intrinsic band gap of p-GaN, h is Planck’s constant, N is the carrier concentration of p-GaN, and m_r*is the reduced

effec-tive mass of p-GaN. It can be seen from Eq. 共1兲 that an

increases in E will lead to an increase in N. Kim et al. have indicated that the increase in band gap may be attributed to the increased carrier concentration and the increase in carrier concentration can be explained by the Burstein–Moss shift.26 Lo et al. have also suggested that the band-gap shifts were due to the Burstein–Moss shift, which is related to the carrier concentration.27On the basis of the evidence, we deduce that the共MgGa–VN兲2+ complexes near the p-GaN surface region

were transformed into the 共Mg_Ga–S_N兲0 _{complexes after}

共NH4兲2Sx treatment, which led to an increase in the hole

concentration near the p-GaN surface region and the reduc-tion of the⬃2.8-eV PL intensity, which is in agreement with the reported results by Kim et al.16

In summary, the effects of 共NH4兲2Sx treatment on the

optical and electrical properties of the heavily Mg-doped

p-GaN sample were investigated. After共NH₄兲₂S_x treatment, the共MgGa–VN兲2+ complexes near the p-GaN surface region

were transformed into the neutral 共MgGa–SN兲0 complexes,

which led to the reduction of the donor-type states near the surface region, an increase in the hole concentration near the

surface region, and the reduction of the⬃2.8-eV PL

inten-sity. This could explain why共NH4兲2Sxtreatment is useful for

the improvement of the electrical properties of ohmic con-tacts to p-GaN.

This project is supported by National Science Council of Taiwan, Republic of China.

1

S. Nakamura and S. F. Chichibu, Introduction to Nitride Semiconductor

Blue Lasers and Light Emitting Diodes共Taylor & Francis, London, 2000兲.

3

J. L. Lee, J. K. Kim, J. W. Lee, Y. J. Park, and T. Kim, Phys. Status Solidi A 176, 763共1999兲.

4

J. S. Jang and T. Y. Seong, Appl. Phys. Lett. 76, 2743共2000兲. 5

C. S. Lee, Y. J. Lin, and C. T. Lee, Appl. Phys. Lett. 79, 3815共2001兲. 6

L. Aigouy, F. H. Pollak, J. Petruzzello, and K. Shahzad, Solid State Com-mun. 102, 887共1997兲.

7

M. Leibovitch, L. Kronik, E. Fefer, and Y. Shapira, Phys. Rev. B 50, 1739 共1994兲.

8

Y. Rosenwaks, L. Burstein, Y. Shapira, and D. Huppert, Appl. Phys. Lett.

57, 458共1990兲.

9

J. Szuber, Appl. Surf. Sci. 55, 143共1992兲. 10

S. M. Jeong, H. W. Shim, H. S. Yoon, M. G. Cheong, R. J. Choi, E. K. Suh, and H. J. Lee, J. Appl. Phys. 91, 9711共2002兲.

11

U. Kaufmann, M. Kunzer, M. Maier, H. Obloh, A. Ramakrishnan, B. Santic, and P. Schlotter, Appl. Phys. Lett. 72, 1326共1998兲.

12

N. Kinrot and Y. Shapira, Phys. Rev. B 65, 245303共2002兲. 13

S. Kuźmiński, Vacuum 54, 189共1999兲. 14

M. A. Reshchikov, G.-C. Yi, and B. W. Wessels, Phys. Rev. B 59, 13176 共1999兲.

15

U. Kaufmann, P. Schlotter, H. Obloh, A. Ramakrishnan, K. Köhler, and M. Maier, Phys. Rev. B 62, 10867共2000兲.

16

S. W. Kim, J. M. Lee, C. Huh, N. M. Park, H. S. Kim, I. H. Lee, and S. J. Park, Appl. Phys. Lett. 76, 3079共2000兲.

17

Y. J. Lin, C. D. Tsai, Y. T. Lyu, and C. T. Lee, Appl. Phys. Lett. 77, 687 共2000兲.

18

E. Oh, B. Kim, H. Park, and Y. Park, Appl. Phys. Lett. 73, 1883共1998兲. 19

P. J. Hartlieb, A. Roskowski, R. F. Davis, W. Platow, and R. J. Nemanich, J. Appl. Phys. 91, 732共2002兲.

20

V. M. Bermudez, J. Appl. Phys. 80, 1190共1996兲. 22

J. K. Kim, J. L. Lee, J. W. Lee, Y. J. Park, and T. Kim, J. Vac. Sci. Technol. B 17, 497共1999兲.

23

W. Mönch, J. Vac. Sci. Technol. B 4, 1085共1986兲. 24

C. Tejedor, F. Flores, and E. Louis, J. Phys. C 10, 2163共1977兲. 25

S. Ray, R. Banerjee, N. Basu, A. K. Batabyal, and A. K. Barua, J. Appl. Phys. 54, 3497共1983兲.

26

H. Kim, A. Piqué, J. S. Horwitz, H. Mattoussi, H. Murata, Z. H. Kafafi, and D. B. Chrisey, Appl. Phys. Lett. 74, 3444共1999兲.

27

C. Y. Lo, C. L. Hsu, Q. X. Yu, H. Y. Lee, and C. T. Lee, J. Appl. Phys. 92, 274共2002兲.

202107-3 Linet al. Appl. Phys. Lett. 86, 202107共2005兲

(14)

Electrical properties of Ni/ Au and Au contacts on

p

-type GaN

Yow-Jon Lina)

(Received 3 August 2004; accepted 25 October 2004; published 28 December 2004)

The electrical properties of Ni/ Au and Au contacts on p-type GaN共p-GaN兲 were investigated in

this study. From the experimental result, it is suggested that the current–voltage characteristic of Au/ Ni/ p-GaN is better than that of Au/ p-GaN. The secondary-ion mass spectroscopy measurements revealed that hydrogen is effectively removed from the p-GaN layer by the existence of the Ni film. These results suggest that a Ni film of Au/ Ni/ p-GaN significantly enhances hydrogen desorption from the p-GaN film, which leads to an increase in the hole concentration, the occurrence of the tunneling transmission for holes at the interface, and the improvement of electrical

I. INTRODUCTION

GaN has drawn much interest as a promising material for the fabrication of optoelectronic devices because of the suc-cess in commercialization of blue light-emitting diodes.1–3 To date, the contact resistance is not low enough to be ap-plied for blue laser diode because there is no metal with a high work function comparable to that of p-GaN. In order to manufacture low-resistivity ohmic contacts, various contact

schemes have been investigated.4–16 In recent years, the

Au/ Ni/ p-GaN ohmic contacts annealed in air have been studied extensively because they provide low specific contact resistances and high transparency.7–16 Kim et al. concluded that the improvement of oxidized Ni/ Au contact properties is due mainly to the formation of an intermediate NiO layer, rather than to an enhancement in p-type activation.9Qiao et

al. indicated that the presence of oxygen during annealing

appears to increase the conductivity of the p-GaN.10Koide et

al. indicated that the reason for reduction of the contact

re-sistance by the oxygen gas addition was believed to be due to formation of the p-GaN epilayer with high hole concentra-tion, caused by removal of hydrogen atoms that bonded with Mg atoms.11Mistele et al. suggested that the layer structure of the Ni/ Au layers do not seem to be that important than the selection of the annealing gas.16 The previous studies7–16 were focused on the effect of annealing in a partial oxygen ambient on electrical properties of ohmic contacts to

p-GaN. To our knowledge, the formation mechanism of

ohmic contacts of Au/ Ni/ p-GaN annealed under nitrogen ambient and the effects of Ni film on electrical properties of Ni/ Au ohmic contacts to p-GaN have not yet been well un-derstood. In this study, we find that a Ni film plays an im-portant role in the formation of Ni/ Au ohmic contacts to

p-GaN. According to the experimental result, we deduce that

a Ni layer on p-GaN enhances the activation of the acceptor and improves the ohmic performance of Au/ Ni/ p-GaN.

II. EXPERIMENTAL PROCEDURES

The epitaxial layers used in the experiments were grown on c-plane sapphire substrates using a metalorganic chemical

vapor deposition system. Trimethylgallium, ammonia共NH3兲,

and bis-cyclopentadienylmagnesium were used as the Ga, N, and Mg sources, respectively. An undoped GaN buffer layer with a thickness of 650 nm was grown on the sapphire sub-strate at 520° C, followed by the growth of an p-GaN layer

共762 nm兲 at 1100°C. Mg concentration 共关Mg兴兲 was 9

⫻1019_cm−3 _{for all samples. The}_{关Mg兴 in p-GaN was}

mea-sured using secondary-ion mass spectroscopy (SIMS). The

grown samples were annealed at 750° C for 30 min in nitro-gen ambient using a thermal annealing furnace for the pur-pose of generating holes.12 The specific contact resistance

was measured using the transmission line method(TLM) for

Ni/ Au or Au contacts to p-GaN. A Ni/ Au 共50/600 nm兲

metal mask was employed to form rectangular mesa regions

with dimensions of 100⫻100␮m2. A reactive ion etching

system was used with BCl3 gas to etch the p-GaN wafers.

The dimensions of the metal contact pads in the TLM pattern

was 100⫻100␮m2. The gap spacings between the contact

pads were designed to be 5, 10, 15, 25, 35, 50, and 60␮m.

After the removal of the metallic mask, the samples were cleaned in chemical cleaning solutions of trichloroethylene,

acetone, and methanol(as-cleaned p-GaN). The surfaces of

the as-cleaned p-GaN samples were treated with aqua regia for 10 min prior to respective deposition of the Ni/ Au

共20/20 nm兲 and Au 共40 nm兲 metals using an electron-beam

evaporator. Next, the contacts were treated in a rapid thermal

annealing furnace at 300° C for 1 min (or 2 min) under

ni-trogen ambient. The current–voltage共I–V兲 characteristics of the TLM patterns, before and after annealing, were measured using a Keithley Model-4200-SCS/F semiconductor param-eter analyzer.

III. EXPERIMENTAL RESULTS AND DISCUSSION

Figure 1 shows the respective I – V characteristics of Au/ Ni/ p-GaN and Au/ p-GaN, measured between metal

pads with a gap spacing of 25␮m. The Au/ Ni/ p-GaN

sample does not show a linear I – V behavior. The

a)_{Electronic mail: [email protected]}

(15)

Au/ p-GaN sample does not also show a linear I – V curve. However, in Fig. 1, we find that the I – V characteristic of the Au/ Ni/ p-GaN sample is better than that of the Au/ p-GaN sample. On the other hand, Qiao et al. suggested that the Au

layer has a larger work function than the Ni layer.10 This

implies that the contact conductivity of the Au/ p-GaN sample may be higher than that of the Ni/p-GaN sample. This cannot explain why the I – V characteristic of the Au/ Ni/ p-GaN sample is better than that of the Au/ p-GaN sample. An explanation for this will be given later.

Figure 2 shows the respective I – V characteristics of the

annealed Au/ Ni/ p-GaN sample and the annealed

Au/ p-GaN sample, measured between metal pads with a gap

spacing of 25␮m. The Au/ p-GaN samples were annealed at

300° C for 1 min in nitrogen ambient. Some of the Au/ Ni/ p-GaN samples were annealed at 300° C for 1 min in

nitrogen ambient (1-min-annealed Au/Ni/ p-GaN sample).

Some of the Au/ Ni/ p-GaN samples were annealed at 300° C

for 2 min in nitrogen ambient (2-min-annealed Au/Ni/ p

-GaN sample). We find that the 1-min-annealed Au/Ni/ p

-GaN samples show an approximate linear I – V behavior with a specific contact resistance of 7.5⫻10−3⍀ cm2. Simi-larly, the 2-min-annealed Au/ Ni/ p-GaN samples also show an approximate linear I – V behavior with a specific contact resistance of 4.2⫻10−3 _{⍀ cm}2_{. On the other hand, we find}

that the I – V curve of the annealed Au/ p-GaN sample is similar to that of the unannealed Au/ p-GaN sample. Accord-ing to the experimental result, it is suggested that a Ni film played an important role in the ohmic contact formation of Au/ Ni/ p-GaN.

In order to clarify the effects of the Ni film on electrical properties of the Au/ Ni/ p-GaN sample, we investigated the hydrogen content in the p-GaN film by SIMS measurements

(Cameca IMS-4f system). First, Au and Ni/Au were

depos-ited on all p-GaN surfaces. Next, these samples were

an-nealed at 300° C for 1 min (or 2 min) in nitrogen ambient.

To remove the Au 共Au/Ni兲 film, the annealed Au/p-GaN

共Au/Ni/p-GaN兲 samples were treated in aqua regia solution

for 10 min, rinsed in deionized water for 1 min, and dried by a nitrogen blower. Next, these samples were measured using

SIMS. Figure 3(a) shows the SIMS profile of hydrogen

con-tent for the as-cleaned p-GaN samples. Figure 3(b) shows

the SIMS profiles of hydrogen content of Au/ p-GaN

an-nealed at 300° C for 1 min. Figure 3(c) shows the SIMS

profiles of hydrogen content of Au/ Ni/ p-GaN annealed at

300° C for 1 min. Figure 3(d) shows the SIMS profiles of

hydrogen content of Au/ Ni/ p-GaN annealed at 300° C for 2 min. It is noteworthy that the hydrogen content of the

1-min-annealed(or 2-min-annealed) Au/Ni/ p-GaN samples

is lower than that of the annealed Au/ p-GaN sample or that of the as-cleaned p-GaN samples. This is in good agreement with the previous report17 that the Ni film significantly en-hances hydrogen desorption from the p-GaN film, which leads to an increase in the hole concentration and the im-provement of electrical properties of Au/ Ni/ p-GaN. As a result, the Ni film enhances activation of Mg acceptors as catalysts at low temperature, causing the improvement of the electrical properties of Au/ Ni/ p-GaN, as shown in Figs. 1 and 2. We also find that the I – V characteristic of the 2-min -annealed Au/ Ni/ p-GaN sample is better than that of the 1-min-annealed Au/ Ni/ p-GaN sample, because the hydro-gen content of the 2-min-annealed Au/ Ni/ p-GaN sample is lower than that of the 1-min-annealed Au/ Ni/ p-GaN sample, as shown in Figs. 2 and 3. On the other hand, Rick-ert et al. have indicated that the barrier-height value of

FIG. 1. Typical I – V curves of(a) Au/Ni/ p-GaN and (b) Au/ p-GaN before

annealing. FIG. 2. Typical I – V curves of_{1 min,} _{(b) Au/ p-GaN annealed at 300°C for 1 min, and (c) Au/Ni/ p}(a) Au/Ni/ p-GaN annealed at 300°C for -GaN annealed at 300° C for 2 min.

49 Yow-Jon Lin: Electrical properties of Ni/ Au and Au contacts 49

(16)

Au/ p-GaN or Ni/p-GaN is approximately equal to 1.9 eV.18 Yu et al. have indicated that the value of tunneling parameter

共E00兲 of unannealed Ni/p-GaN was calculated to be

0.056 eV.19The deduced value of E00 is likely to contain a

tunneling component.19 This indirectly suggests that the Ni film of Au/ Ni/ p-GaN absorbs hydrogen and enhances acti-vation of Mg acceptors, causing the occurrence of the tun-neling transmission for holes at the Ni/p-GaN interface. Con-sequently, we deduce that the selection of the Ni/ Au layers seems to be an important factor for improving the ohmic contacts of p-GaN.

In Fig. 3, we can also see that the hydrogen content of the annealed Au/ p-GaN sample is close to that of the as-cleaned

p-GaN sample. Nakamura et al. have indicated that the

low-resistivity p-GaN films were obtained by N2-ambient thermal

annealing above 700° C.20 Nakamura et al. also suggested

that almost no change in resistivity was observed when the

p-GaN sample was annealed between room temperature and

400° C.20 This implies that low-temperature activation of

p-GaN and low-temperature hydrogen desorption from the p-GaN film were difficult to achieve by the absence of Ni

film.

IV. SUMMARY

In this study, the I – V characteristics of Au/ Ni/ p-GaN and Au/ p-GaN, and the effects of Ni layer on contact resis-tivity of Au/ Ni/ p-GaN have been investigated. SIMS mea-surements have revealed that the Ni layer on p-GaN en-hances the activation of the acceptor, as well as the occurrence of the tunneling transmission for holes at the Ni/p-GaN interface, and improves the ohmic performance of Au/ Ni/ p-GaN. This is probably due to the catalytic effect of Ni for the hydrogen desorption.

ACKNOWLEDGMENTS

The author would like to thank the National Science Council of the Republic of China for financially supporting this research under Contract No. NSC 93-2215-E-018-005. The SIMS was kindly provided from the NTHU Instrumen-tation Center at Hsinchu National Tsing Hua University.

1

M. Osiński, Gallium-Nitride-Based Technologies (SPIE, Washington, 2002).

2

H. Morkoç, Nitride Semiconductors and Devices(Springer, Berlin, 1999). 3

E. F. Schubert, Light-Emitting Diodes(Cambridge University Press, Cam-bridge, 2003).

4

J. S. Kwak, O. H. Nam, and Y. Park, Appl. Phys. Lett. 80, 3554(2002). 5

H. W. Jong and J. L. Lee, J. Appl. Phys. 93, 5416(2003). 6

J. O. Song, D. S. Leem, J. S. Kwak, O. H. Nam, Y. Park, and T. Y. Seong, Appl. Phys. Lett. 83, 2372(2003).

7

Y. J. Lin, Z. D. Li, C. W. Hsu, F. T. Chien, C. T. Lee, S. T. Shao, and H. C. Chang, Appl. Phys. Lett. 82, 2817(2003).

8

J. S. Kwak, J. Cho, S. Chae, O. H. Nam, C. Sone, and Y. Park, Jpn. J. Appl. Phys., Part 1 40, 6221(2001).

9

H. Kim, D. J. Kim, S. J. Park, and H. Hwang, J. Appl. Phys. 89, 1506 (2001).

10

D. Qiao, L. S. Yu, S. S. Lau, J. Y. Lin, H. X. Jiang, and T. E. Haynes, J. Appl. Phys. 88, 4196(2000).

11

Y. Koide, T. Maeda, T. Kawakami, S. Fujita, T. Uemura, N. Shibata, and M. Murakami, J. Electron. Mater. 28, 341(1999).

12

J. K. Sheu, Y. K. Su, G. C. Chi, P. L. Koh, M. J. Jou, C. M. Chang, C. C. Liu, and W. C. Hung, Appl. Phys. Lett. 74, 2340(1999).

13

J. K. Ho, C. S. Jong, C. N. Huang, C. Y. Chen, C. C. Chiu, and K. K. Shih, Appl. Phys. Lett. 74, 1275(1999).

14

J. K. Ho, C. S. Jong, C. N. Huang, C. C. Chiu, K. K. Shih, L. C. Chen, F. R. Chen, and J. J. Kai, J. Appl. Phys. 86, 4491(1999).

15

S. H. Wang, S. E. Mohney, and R. Birkhahn, J. Appl. Phys. 91, 3711 (2002).

16

D. Mistele, F. Fedler, H. Klausing, T. Rotter, J. Stemmer, O. K. Semchi-nova, and J. Aderhold, J. Cryst. Growth 230, 564(2001).

17

I. Waki, H. Fujioka, M. Oshima, H. Miki, and A. Fukizawa, Appl. Phys. Lett. 78, 2899(2001).

18

K. A. Rickert, A. B. Ellis, J. K. Kim, J. L. Lee, F. J. Himpsel, F. Dwikusuma, and T. F. Kuech, J. Appl. Phys. 92, 6671(2002). 19

L. S. Yu, D. Qiao, L. Jia, S. S. Lau, Y. Qi, and K. M. Lau, Appl. Phys. Lett. 79, 4536(2001).

20

S. Nakamura, T. Mukai, M. Senoh, and N. Iwasa, Jpn. J. Appl. Phys., Part 2 31, L139(1992).

FIG. 3. SIMS profile of hydrogen content of (a) as-cleaned p-GaN, (b) Au/ p-GaN annealed at 300° C for 1 min,(c) Au/Ni/ p-GaN annealed at 300° C for 1 min, and(d) Au/Ni/ p-GaN annealed at 300°C for 2 min.

50 Yow-Jon Lin: Electrical properties of Ni/ Au and Au contacts 50

(17)

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

Study of Schottky Barrier Heights of Indium-Tin-Oxide on p-GaN Using X-ray Ph...

Yow-Jon Lin; Chou-Wei Hsu

Journal of Electronic Materials; Sep 2004; 33, 9; ProQuest Science Journals pg. 1036

(18)

(19)

(20)

(21)