Effects of nickle oxide on nickel/gold contacts to p-type gallium nitride

MISSING IMAGE

Material Information

Title:
Effects of nickle oxide on nickel/gold contacts to p-type gallium nitride
Physical Description:
xiii, 93 leaves : ill. ; 29 cm.
Language:
English
Creator:
Kim, Junghan
Publication Date:

Subjects

Subjects / Keywords:
Materials Science and Engineering thesis, Ph. D   ( lcsh )
Dissertations, Academic -- Materials Science and Engineering -- UF   ( lcsh )
Genre:
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 2004.
Bibliography:
Includes bibliographical references.
Statement of Responsibility:
by Junghan Kim.
General Note:
Printout
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 003163499
sobekcm - AA00006096_00001
System ID:
AA00006096:00001

Full Text










EFFECTS OF NICKEL OXIDE ON NICKEL/GOLD CONTACTS
TO P-TYPE GALLIUM NITRIDE

















By

JUNGHAN KIM


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA


2004































Copyright 2004

by

JUNGHAN KIM














ACKNOWLEDGMENTS

I would like to express my deepest gratitude to Dr. Michael Kaufman and Dr. Paul

Holloway for their outstanding teaching and warm concern. I appreciate their unselfish

gift of time and help. I admire their passion for research. Financial support from them and

from the department of materials science and engineering is also gratefully

acknowledged.

I would like to express thanks to the members of my supervisory committee Dr.

Stephen Pearton, Dr. Cammy Abernathy, and Dr. Timothy Anderson for their advice and

encouragement. I would also like to express my gratitude to the Major Analytical

Instrumentation Center (MAIC).

Finally, I thank my wife, Yoonhee, and my daughter, Soram. I am very proud of

their patience and support. I also greatly appreciate the endless support and concern of

my parents in Korea. Thank God.














TABLE OF CONTENTS

page

ACKNOWLEDGMENTS .........................................-........................-

LIST O F TA BLES....... ............................. ...-.......... ................................... ... .... vii

LIST O F FIG U RE S ...................................................... ..................... viii

A B ST R A C T ............................................................ .......................................... xii

CHAPTER

S IN TRO DU CTION ............................................................. ............................. 1

2 THEORY AND BACKGROUND ......................... ............................

2.1 Metal-Semiconductor Junctions ..............................................................4
2.1.1 Rectifying C contacts ......................................... ......................................4..
2.1.2 O hm ic C ontacts ........... ........................ .........................................6
2.2 Surface Pinning Effect ................... ................................................................
2.3 N active O xide Effects....................................... .......................................... 9
2.4 Properties of G aN ............................................................................ 11
2.5 Gallium Vacancies in GaN ............................ ...........................11
2.6 N ickel/G old C ontact...................................... ................................................. 12
2.7 Measuring Specific Contact Resistance............................................................16
2.8 Conductivity of NiO ............................................17
2.9 Optical Properties of N iO ......................................... ...................................... 17
2.9.1 O optical G ap of N iO ................................................................. .............17
2.9.2 Nickel Oxide Deposited by Sputtering..................................... ..... 18

3 GOLD/PRE-OXIDIZED NICKEL OXIDE/P-GALLIUM NITRIDE CONTACTS.20

3.1 E xperim ents .................................-............................. ...................................... 20
3.1.1 Sample Preparation....................................... ........................20
3.1.1.1 N i/A u ohm ic contact .......................... ........................................20
3.1.1.2 Au/pre-oxidized NiO/p-GaN contacts...........................................20
3.1.1.3 N iO sheet resistance ................................................................... 23
3.1.2 A analysis .. ................................................................................................23
3.1.2.1 I-V characteristics ............... ............................... ...23
3.1.2.2 A ES analysis ................................. .............................................24








3.2 Ni/Au Ohm ic Contact................................................... ............................24
3.3 Au/pre-oxidized NiO/p-GaN Contact.................................................................24
3.3.1 Current-Voltage (I-V) Characteristics.......................................................24
3.3.2 Depth Profile A analysis ......................................................................... 27
3.3.3 Post Annealing of Au/pre-oxidized NiO/p-GaN Samples........................29
3.3.4 N iO Sheet Resistance ................................................................................ 31
3.3.5 Rectifying Characteristics of Au/pre-oxidized NiO/p-GaN Contact..........36

4 REACTION BETWEEN NICKEL AND P-GALLIUM NITRIDE DURING
ANNEALING IN OXIDIZING AMBIENT ..............................................................41

4 .1 E xperim ents .............................................................................. ........... .......... 43
4.1.1 Sample Preparation.................. .. .....................43
4 .1.2 A analysis ........................................... ...... .................... ... ...................44
4.1.2.1 X RD m easurem ent ......................................................................... 44
4.1.2.2 A ES analysis ................................ ................. ........................... 44
4.1.2.3 TEM analysis............................................ ................................44
4.2 X -ray D iffraction A analysis ................................................................ ...... 44
4.3 Auger Electron Spectroscopy Depth Analysis .........................................45
4.4 Transmission Electron Microscope Analysis ...................................................47

5 INTERPLANAR SPACING CHANGE OF GALLIUM NITRIDE DURING
ANNEALING IN OXIDIZING AMBIENT ................................................ 50

5.1 Selected Area Diffraction Patterns for GaN .................................... ...50
5.2 X-ray Diffraction Analysis of Strain in GaN................................................... 51

6 EFFECT OF GALLIUM NITRIDE STRAIN ON CONTACT RESISTANCE ........55

6.1 E xperim ents ......................................................................................... 55
6.1.1 Sam ple Preparation....................... ........................ ..........56
6.1.2 A analysis ... ........ .... ............................................. 56
6.1.2.1 I-V characteristics ............. ....................... ...56
6.1.2.2 XRD m easurem ent ............................... ......................................56
6.2 Current-Voltage Measurements and XRD Analysis.............................. ....56
6.3 Effect of Annealing in Oxidizing Ambient on p-GaN Strain..............................58
6.4 Strain in G aN for Thin N i...... .................................... .......................................58
6.5 Characterization of Split (0002) GaN Peak............... ...........................64

7 OPTICAL PROPERTIES OF NICKEL OXIDE PRODUCED BY THERMAL
OXIDATION OF NICKEL .............................................67

7.1 Experim ents.......................................................... .... .......... 67
7.1.1 Sample Preparation......... ....................... ................................67
7.1.2 A analysis ...... ........................................................................ ....68
7.1.2.1 X RD m easurem ent ................................. ................................. 68
7.1.2.2 AFM measurement.............................................68








7.1.2.3 Optical transmittance........... ............................................68
7.2 Optical Transmittance of NiO Produced by Thermal Oxidation of Ni ................68
7.3 X -ray D iffraction A analysis ............................................................................... 73
7.3.1 G rain Size A analysis ......................................... ......... .......................... 73
7.3.2 Effects of Preferred Orientation on Optical Transmittance of NiO ..........78
7.3.3 X-ray Peak Position of NiO...................................... .......................81
7.4 Surface Roughness Analysis............................................................................ 81

8 S U M M A R Y ................ .................... ................................................................. 84

LIST OF REFERENCES............................................-. ------.... ................ ........ -.88

BIOG RAPH ICA L SKETCH ......................................................... ..............................93














LIST OF TABLES

Table page

2-1. Selected properties of G aN .................................................................................... 12

3-1. Properties of p-GaN and p-NiO at room temperature ..............................................38

5-1. Interplanar spacing comparison.......................................................................... 51

5-2. Interplanar spacing of GaN (0002) plane ..............................................................52

6-1. XRD condition for the measurement of GaN (0002) peak.......................................58

7-1. Grain size and normalized intensity for NiO produced by annealing Ni in an
oxidizing am bient ............................................................................................... 77














LIST OF FIGURES


Figure page

2-1. Energy band diagram of Schottky contact for n-type semiconductor (a) before
contact (b) after contact.......................................................... ................................. 5

2-2. Energy band diagram of Schottky contact for p-type semiconductor (a) before
contact (b) after contact....................................................................... ............... 7

2-3. Energy band diagram of ohmic contact for n-type semiconductor (a) before contact
(b) after contact .................................................................................................... 8

2-4. Energy band diagram of ohmic contact for p-type semiconductor (a) before contact
(b) aft er contact .................................................................................................... 9

2-5. Final EF position for a number of metals and oxygen on GaSb GaAs and
InP. Note that there is little dependence on the chemical nature of the adatom......10

2-6. W urzite structure of G aN .......................................................................................... 11

2-7. Formation energies of various defects in GaN as a function of the Fermi
level, j, according to theoretical calculations. ................................... ........... 13

2-8. TEM micrograph of Ni/Au on p-GaN after an oxidizing treatment at 5000C for
10min in air; a: Au includes a small amount of Ni, b: face-centered cubic NiO
phase, c: amorphous regions consisting of a relatively larger amount of Ga as well
as N i and O .............................................................................................................14

2-9. TEM micrograph of Ni/Au on p-GaN after an oxidizing treatment at 5000C for
10min in air; the arrow indicates a possible path for high current flow. .................14

2-10. Schematic illustration of the reaction mechanism and diffusion behavior during the
oxidation of Ni/Au on p-GaN. (a) Early stage of oxidation reaction, (b) pronounced
interdiffusion and crystalline NiO grow up to separate the Au-Ni alloy into discrete
islands, (c) Au-rich islands building epitaxially on p-GaN and three possible
reactions producing Ni-Ga-O phases and (d) oxidized contact scheme with
optim um phase distribution. ................................................................................... 15

2-11. Optical absorption coefficient of NiO. 4.0eV is the energy obtained by
extrapolating the flat absorption between 4.3eV and 9eV, and intersecting it








with the tangent at the steepest point of the absorption between 3.1eV and
4 .3eV ............................................................................. ................................... 18

2-12. Transmittance of NiO as functions of oxygen flow ratio at the substrate
tem perature of 400 C ........................................................................................... 19

2-13. Atomic percent of oxygen in the deposited NiO films as a function of
oxygen flow rate ratio. ......................... ................................ ........................19

3-1. Schematic diagram of CTLM patterns .......................................... ................... 21

3-2. Process sequence to make CTLM pattern for pre-oxidized samples..........................22

3-3. Vertical structure of CTLM patterns .......................... ............................................24

3-4. Linear I-V characteristics after annealing at 5000C for 10min in air.
(N i50A /A u50A ) ..................................................................... ...........................25

3-5. I-V curve for pre-oxidized samples annealed at 5000C in air for (a) 0, (b) 10s
(c) 20s, (d) Imin, (e) 2min, and (f) 5min ........................................ ......... 26

3-6. I-V for Au/pre-oxidized NiO/p-GaN samples with different anneal time..................27

3-7. Current measured at (a) 1.0V, (b) 1.5V and (c) 2.0V.................................. ......28

3-8. AES depth profile for pre-oxidized samples with various anneal time. X-axis and y-
axis represent sputter time (sec) and peak intensity (a.u.), respectively ................29

3-9. Effect of post annealing on I-V characteristics for (a) as-deposited sample and
(b) 10s pre-oxidized sample......................... .................. ................................ 30

3-10. Reaction process during post annealing. (a) before post annealing
(b)oxygen diffusion during post annealing. Oxygen flow is designated by arrow(c)
NiO formation near the interface between Ni and p-GaN ....................................31

3-11. I-V curves for NiO formed by annealing Ni at 5000C in air for (a) 10s
(b) 20s (c) 60s (d) 120s and (e) 180s.......................................... ....... ............33

3-12. Linear fitting of In(R/r) vs. Rt curve for (a) 10s (b) 20s (c) Imin (d) 2min
(e) 3min annealed samples.................................. ........................34

3-13. Sheet resistance of NiO formed by annealing Ni at 5000C in air for different
tim es ..................................................................... .................................................35

3-14. High resolution TEM image of (a) as-deposited Ni on p-GaN and (b) after
annealing at 4000C for 10min .................................-.............................. 36

3-15. Energy band diagram of NiO and p-GaN (a) before contact (b) after contact.........37








6-6. GaN (0002) peak comparison for Ni 50A ...........................................................62

6-7. GaN (0002) peak comparison for Ni 50A/Au50A ...........................................63

6-8. GaN (0002) peak comparison for Ni 150A/Au 150A.................................... ..63

6-9. Schematic X-ray penetration depth profile of (a) without tape and (b) with tape
sam ples .................................................................................................................. 64

6-10. GaN (0002) peak with different penetration depth.................................................65

6-11. Dependence of GaN lattice parameters on free electron concentration ...................66

7-1. Schematic diagram of the instrument for optical transmittance measurement...........69

7-2. Optical transmittance of NiO oxidized at 5000C as a function of wavelength...........70

7-3. Normalized intensity of NiO oxidized at 5000C as a function of wavelength............71

7-4. Normalized intensity at 450nm for NiO oxidized at 500C .....................................71

7-5. Optical transmittance of NiO oxidized at 6000C as a function of wavelength...........72

7-6. Normalized intensity of NiO oxidized at 6000C as a function of wavelength...........72

7-7. Normalized intensity at 450nm of NiO oxidized at 6000C.......................................73

7-8. XRD peaks of NiO annealed at 5000C; (a) (111) plane and (b) (200) plane..............74

7-9. XRD peaks of NiO annealed at 6000C; (a) (111) plane and (b) (200) plane.............75

7-10. Grain size comparison between 500C and 600C anneal......................................76

7-11. Integrated intensity of NiO annealed at 5000C in air...........................................78

7-12. Integrated intensity ratio of NiO annealed at 5000C in air .....................................79

7-13. Integrated intensity of NiO annealed at 6000C in air...........................................80

7-14. Integrated intensity ratio of NiO annealed at 6000C in air .....................................80

7-15. The surface roughness of NiO annealed at 5000C (a) average, R;, (b) root mean
square, R .............................................................................. ............................82

7-16. AFM surface image of (a) as-depostied Ni (b) NiO produced by annealing Ni 400A
at 5000C in air for 15min. (c) NiO produced by annealing Ni 400A at
500 C in air for h. .......................................... .................................................. 83








6-6. GaN (0002) peak comparison for Ni 50A ...........................................................62

6-7. GaN (0002) peak comparison for Ni 50A/Au50A ....................................................63

6-8. GaN (0002) peak comparison for Ni 150A/Aul50A ...............................................63

6-9. Schematic X-ray penetration depth profile of(a) without tape and (b) with tape
sam p les ................................................ ... ................................................. .64

6-10. GaN (0002) peak with different penetration depth.................................................65

6-11. Dependence of GaN lattice parameters on free electron concentration ...................66

7-1. Schematic diagram of the instrument for optical transmittance measurement..........69

7-2. Optical transmittance of NiO oxidized at 5000C as a function of wavelength..........70

7-3. Normalized intensity of NiO oxidized at 500'C as a function of wavelength............71

7-4. Normalized intensity at 450nm for NiO oxidized at 5000C .....................................71

7-5. Optical transmittance of NiO oxidized at 6000C as a function of wavelength ..........72

7-6. Normalized intensity of NiO oxidized at 6000C as a function of wavelength............72

7-7. Normalized intensity at 450nm of NiO oxidized at 6000C.......................................73

7-8. XRD peaks of NiO annealed at 500C; (a) (11) plane and (b) (200) plane.............74

7-9. XRD peaks of NiO annealed at 6000C; (a) (111) plane and (b) (200) plane............75

7-10. Grain size comparison between 5000C and 600C anneal......................................76

7-11. Integrated intensity of NiO annealed at 5000C in air..............................................78

7-12. Integrated intensity ratio of NiO annealed at 5000C in air .....................................79

7-13. Integrated intensity of NiO annealed at 600C in air................................... ...80

7-14. Integrated intensity ratio of NiO annealed at 600C in air .....................................80

7-15. The surface roughness of NiO annealed at 500"C (a) a\ rage. R,, (b) root mean
square, R .......................................................... .. ..............................................82

7-16. AFM surface image of (a) as-depostied Ni (b) NiO produced by annealing Ni 400A
at 5000C in air for 15min. (c) NiO produced by annealing Ni 400A at
500 C in air for h. ........................................... ........................................... 83
















Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy

EFFECTS OF NICKEL OXIDE ON NICKEL/GOLD CONTACTS
TO P-TYPE GALLIUM NITRIDE

By

Junghan Kim

December 2004

Chair: Michael J. Kaufman
Major Department: Materials Science and Engineering

A determination of the role of NiO on the formation of Ni/Au ohmic contacts to p-

GaN has been carried out by conducting a variety of deposition and annealing

experiments followed by detailed characterization. Current-voltage (I-V) measurements

for Au/pre-oxidized NiO/p-GaN contacts indicate contacts produced this way are not

ohmic in nature but rather exhibit Schottky barrier behavior. The interface between NiO

and p-GaN was analyzed using TEM and no reaction phase forms between Ni and p-GaN

after annealing at 5000C in an oxidizing ambient although this treatment does convert Ni

to NiO. The strain produced in the p-GaN upon Ni deposition is partially released during

annealing in an oxidizing ambient for the samples containing either 600A or 1500A Ni

while a similar annealing of the sample coated with 150A Ni sample did not result in

measurable strain release in GaN. XRD analysis of the GaN sample coated with

Ni50A/Au50A (the industry standard for producing ohmic contacts on p-GaN) indicated








that these contacts produce negligible strain in GaN. Based upon the experimental results,

it is concluded that the NiO formed during annealing in an oxidizing ambient does not

result in the low resistance ohmic contact behavior observed for Ni/Au contacts to p-

GaN. Rather, the current path for the low resistance ohmic contacts of Ni/Au annealed in

an oxidizing ambient appears to result from the direct contact between the Au phase and

the p-GaN. Finally, it is shown that the optical transmittance of NiO produced by thermal

oxidation is dependent on crystallographic orientation (texture).













CHAPTER 1
INTRODUCTION

GaN is an important material because it has a direct band gap which is suitable for

blue and UV light emission. High temperature operation is another promising feature of

GaN due to its wide band gap, high breakdown electric field and high saturation

velocity.' Therefore, many researchers have focused on GaN over the last few decades. In

spite of the desirable features of GaN, there are two major processing challenges. One of

them is related to epitaxial growth of GaN. The single crystal growth of bulk GaN is

difficult because its melting temperature is high (-2500C) and because nitrogen has a

high vapor pressure at these elevated temperatures.2 Consequently, the exact

stoichiometry of GaN is hard to achieve by growing from the melt. Therefore, GaN is

usually grown epitaxially on the basal plane of sapphire (a-Al203) substrates. The

difference in lattice constant between GaN and sapphire tends to result in high defect

density in epitaxially grown GaN films. SiC (silicon carbide) is an alternative to sapphire

but is rarely used because it is considerably more expensive. The other challenge for GaN

is that it is hard to produce low resistance ohmic contacts to p-GaN. There are two major

obstacles to achieving low resistance ohmic contacts. The first is related to the high work

function of p-GaN since, to make ohmic contacts for p-type semiconductors, the work

function of the metal should be higher than that of the p-type semiconductor.

Unfortunately, no metal satisfies this criterion for p-GaN. The other obstacle for low

resistance ohmic contacts is the high activation energy of the acceptor in p-GaN. Since








Mg has the lowest activation energy among possible p-type dopants, Mg is used as an

acceptor for p-GaN. The activation energy of Mg (150meV 200meV) is 3-5 times

higher than that of dopants (- 30-50meV) for Si. Further, the ionization efficiency of Mg

is low (-6% at room temperature). Some researchers have tried to increase the Mg

concentration in order to compensate for its low ionization efficiency.3'4 However,

increasing the Mg concentration above certain levels does not improve conductivity but

results in a decrease in hole concentration. The decrease in hole concentration can be

explained by the solubility limit of Mg in GaN which is due to competing formation of

Mg3N2.5 Finally, Some investigators have attempted to use Be as an acceptor and O as a

reactive codopant to increase its activation efficiency.6

In efforts to produce low resistance ohmic contacts to p-GaN, many different

approaches have been tried and reported.7'8 In light of the Schottky barrier height of the

junction, metals with high work functions should be used for p-GaN contacts because the

Schottky barrier height decreases as the work function of the metal increases.911 Various

metal contacts were applied to p-GaN.12-16 Other investigators have attempted to get rid

of the native oxide between the metal and p-GaN because a native oxide between the

metal contact and p-GaN increases the specific contact resistance.17 The surface treatment

before the deposition of metal also has an effect on specific contact resistance. 1820 Other

investigators have focused on the role of hydrogen in p-GaN.21-24 since hydrogen reacts

with Mg and makes Mg-H complexes which prevent Mg activation and decrease the hole

concentration in p-GaN.25-27 The decrease in hole concentration results in an increase in

specific contact resistance. Therefore, the removal of hydrogen from p-GaN is one

approach for producing low resistance ohmic contacts to p-GaN.28 Finally, other groups








have focused on producing superlattice structures to increase the hole concentration in the

near-surface regions. Thus far, InGaN/GaN and GaN/AlGaN/GaN superlattice structures

have been reported.2932

The low specific contact resistance ofNi/Au contacts is obtained by annealing in an

oxidizing ambient. Although many research papers have been published about Ni/Au

contacts to p-GaN, the mechanism of low resistance ohmic contacts has not yet been

positively identified.

In this research, a determination of the role of NiO on the formation of Ni/Au

ohmic contacts to p-GaN has been carried out by conducting a variety of deposition and

annealing experiments followed by detailed characterization. The optical transmittance of

NiO produced by thermal oxidation is also discussed.













CHAPTER 2
THEORY AND BACKGROUND

2.1 Metal-Semiconductor Junctions

A metal -semiconductor junction may be ohmic or rectifying, depending on the

nature of the semiconductor (p or n type) and the relative work functions of the two

materials.

2.1.1 Rectifying Contacts

Figure 2-1 (a) shows an energy band diagram of the junction between a metal and

an n-type semiconductor. In this figure, DM and Os refer to the work function of the metal

and the semiconductor, respectively. The work function of a material is the energy

difference between the vacuum level and the Fermi energy level. In addition, x is the

electron affinity of the semiconductor and is the energy difference between the top of the

conduction band, Ec, and the vacuum level. The metal-semiconductor (n-type) contact in

which IM > Os is called a Schottky barrier diode as shown in Figure 2-1(a) and this

device shows non-linear I-V (current voltage) characteristics. When the junction is

formed, electrons flow from the semiconductor to the metal because they have higher

energy in the semiconductor. This flow continues until the Fermi levels are aligned as

shown in Figure 2-1(b). The electron flow from the n-type semiconductor to the metal

leaves the surface of the semiconductor depleted of electrons and leaves behind positively

charged donor ions in the semiconductor. In the near-interface region, the Fermi level of

the semiconductor must move towards the valence band and induce upward bending of








the energy band as shown in Figure 2-1(b). The negative charge develops on the metal

side because of the electron transfer from the semiconductor to the metal. Since these two

types of charges exist on either side of the junction, the electric field directed from the

semiconductor to the metal is produced. The equilibrium contact potential of the junction


metal semiconductor


vacuum level



EF --------


vacuum level



EF- ----------


rs Ec
-------- EF


x (DS


Figure 2-1. Energy band diagram of Schottky contact for n-type semiconductor (a) before
contact (b) after contact


represents the difference between the work function potentials OM and Os. The barrier

height OB for the injection of electrons from the metal to the semiconductor is given by

qO, = qO1, qX ---------------------- (1)








This type of metal-semiconductor contact is called a Schottky contact. A Schottky

contact is also formed when a metal with a small work function compared with that of the

semiconductor is deposited on a p-type semiconductor. This condition is shown in Figure

2-2. To align the Fermi levels in this case, electrons must flow from the metal to the

semiconductor, resulting in a positive surface charge in the metal and a negative charge

on the semiconductor. The negative charge exists within a depletion region in which

acceptor ions are left uncompensated by the holes. The potential barrier is Ds OM

designated in Figure 2-2 (b).


2.1.2 Ohmic Contacts

An ohmic contact is a metal-semiconductor contact that has a negligible contact

resistance relative to the bulk or series resistance of the semiconductor. This type of

contact can be formed when metal-semiconductor junctions satisfy the following

requirements: OM < Os for n-type and (M > Os for p-type semiconductors. When (M <

(s and the semiconductor is n-type as shown in Figure 2-3(a), the Fermi levels are

aligned by the transfer of electrons from the metal to the semiconductor. This alignment

raises the electron energies of the semiconductor relative to that in the metal at

equilibrium as shown in Figure 2-3(b). There is no energy barrier to the flow of electrons

from the metal to the semiconductor. When (M > Os and the semiconductor is p-type

(Figure 2-4), electrons flow from the valence band of the semiconductor to the metal.

Since the valence band is not completely filled near the metal junction, electron-hole

pairs in the valence band are formed by exciting valence electrons to higher available

energy levels within the valence band, which results in a junction that does not provide a

barrier to current flow.









metal semiconductor


vacuum level



EF -----------


vacuum level



EF --------


(Ds


x


Figure 2-2. Energy band diagram of Schottky contact for p-type semiconductor (a) before
contact (b) after contact


2.2 Surface Pinning Effect

It is often observed that the Schottky barrier heights for various metals on a

particular semiconductor (especially III-V compound semiconductors) are the same. This

behavior is attributed to the presence of a large number of interface states in the band gap


Ec
Ls


----------------- ------------
Ev






8


of the surface region of the semiconductor; these states arise from the surface dangling

bonds and impurities. As a result, the addition or removal of electrons from the


metal semiconductor


vacuum level

E DM

EF --- ---------


x


Ec

--- EF

Ev


vacuum level


4F Ec
EF ------ ----- --^---- 11 ------------
EF

Ev




(b)



Figure 2-3. Energy band diagram of ohmic contact for n-type semiconductor
(a) before contact (b) after contact


semiconductor does not alter the position of the Fermi level at the surface and the Fermi

level is said to be pinned. Figure 2-5 shows the surface pinning effects for III-V

semiconductors.33 Since GaN does not show a surface pinning effect, the Schottky barrier








height between the metal and GaN must be dependent on the work function of the

metals.34




metal semiconductor


vacuum level


X Ec
(s EF



Ev




(a)


vacuum level


(DM



EF



(


b)



b)


Figure 2-4. Energy band diagram of ohmic contact for p-type semiconductor
(a) before contact (b) after contact


2.3 Native Oxide Effects

According to the metal-semiconductor band theory, the Schottky barrier height can

be influenced by the native oxide thickness between the metal and the semiconductor.

The effect of oxide thickness on Schottky barrier height can be represented by equation












qh = q6( + [4kT/h(2mz) ] -------- (2)

where O<,: Schottky barrier heght without native oxide

m : mean tunneling effective mass of the carriers

S: mean tunneling barrier

3: oxide layer thickness



FERMI LEVEL on
PINING ap


GoA;jilO)


9






0U.- o
0.4- E
SEg
0.2- 0 0
C' Go Au. Oa0 VBt


1n !P(110)

0.- E



C S/ i". ONY V'BM
OVERLAYER PRODUCING PINNING (Sub-Monoloyer)


Figure 2-5. Final EF position for a number of metals and oxygen on GaSb GaAs and
InP. Note that there is little dependence on the chemical nature of the adatom.


According to equation (2), the Schottky barrier height increases with an increase in oxide

thickness and high Schottky barrier heights result in high specific contact resistances.








2.4 Properties of GaN

GaN grown on sapphire substrate has the wurzite structure. The wurzite structure

has a hexagonal unit cell and is similar to the zincblende structure. For wurzite, the

stacking sequence of the (0001) plane is ABABAB, while for the zinc-blende structure,

the stacking sequence of the { 111 } planes is ABCABC. The wurzite structure is shown in

Figure 2-6 and the selected properties of GaN are presented in Table 2-1.36


2.5 Gallium Vacancies in GaN

It has been reported that (1) Ga vacancies are not found in p-type or semi-insulating

Mg-doped layers, (2) Ga vacancies are found in nominally undoped GaN layers, which

show n-type conductivity due to residual oxygen, and (3) Ga vacancy concentrations are

lower in samples, where the n-type doping is done with Si impurities and the amount of

residual oxygen is reduced.37 It is also reported that Ga vacancies in GaN layers depend





t I-
SSlackin order

I;






-I 3/C


Figure 2-6. Wurzite structure of GaN








Table 2-1. Selected properties of GaN


Lattice constant (A) a = 3.189, c = 5.186
Melting point (OC) 2500
Linear thermal expansion ta = 5.59 X 10-6
coefficient (C-1) a =3.17X106

Energy gap (eV) 3.39
Breakdown field (V cm-') 5 X 10


both on the Fermi level and the impurity atoms in the samples. The same general trend is

found in the epitaxial layers as in the bulk crystals: Ga vacancies are formed only in n-

type doping concentrations when oxygen is present. However, if similar doping is done

with Si donors, no Ga vacancies are formed. This behavior can be explained through

association of the observed Ga vacancies with complexes involving oxygen, such as VGa-

ON. Theoretically the formation energies of charged defects in thermal equilibrium

depend on the position of the Fermi level in the energy gap, as shown in the calculated

results of Figure 2-7.37 The negatively charged defects such as Ga vacancies have their

lowest formation energy when the Fermi level is close to the conduction band, i.e. in n-

type material. On the other hand, the formation energy of VGa is high in semi-insulating

and p-type material. These trends correlate with the experimental results showing that Ga

vacancies are observed only in n-type material. It has been reported that N vacancies can

act as donors in GaN with Ga vacancies acting as acceptors.38

2.6 Nickel/Gold Contacts

Ni/Au is the most common metal for p-GaN contacts. The low specific contact

resistance of Ni/Au contacts is obtained by annealing it in an oxidizing ambient.39 Ni has





13


a relatively high work function (5.01eV) among possible metals, and hydrogen can be

removed from the Mg-doped GaN by the deposition of a Ni film on the top surface of p-

GaN and subsequent annealing in N2 ambient.3'40 The specific contact resistance ofNi/Au

contacts is sensitive to anneal condition.39'41'42 Annealing in an oxidizing ambient is more

effective than annealing in a nitrogen ambient for producing low specific contact

resistance. The microstructure of Ni/Au annealed in an oxidizing ambient was reported

by Chen et al.43 Figure 2-8 shows an island-shaped Au phase with a small amount of Ni,

NiO crystalline structure over Au island, and a small region of amorphous phase on the

top of GaN substrate with a relatively large amount of Ga as well as Ni and 0. A possible









-: "V-,,Vv-v
A _-







SON Ga-rich
--T.~N -- -'----------

0.- \ 2.: 3.,
(eV)


Figure 2-7. Formation energies of various defects in GaN as a function of the Fermi
level, p, according to theoretical calculations.

path for high current flow of Ni/Au contact was suggested and designated by arrow in

Figure 2-9.44 Based upon TEM analysis,4 the authors suggested that, during annealing,

Ni diffused out through the Au layer to react with oxygen and formed crystalline NiO as

shown in Figure 2-10. Ho et al. suggested that NiO formation is one of the reasons for





























Figure 2-8. TEM micrograph of Ni/Au on p-GaN after an oxidizing treatment at 500C
for 10min in air; a: Au includes a small amount of Ni, b: face-centered cubic
NiO phase, c: amorphous regions consisting of a relatively larger amount of Ga
as well as Ni and 0.


EPOXY


NiO


NiO:


p-GaN


C Au A
,' .'


10n- m. '-
10nm


Figure 2-9. TEM micrograph of Ni/Au on p-GaN after an oxidizing treatment at 5000C
for 10min in air; the arrow indicates a possible path for high current flow.


...
,e
i~

.*. .~~-
-i










(a) ,



c ).;:.u ,) .
I Aic Au. nh Au-rich




I.0.
(b) proi -nounced interdiffusion and crystalline -NittO grow up to .i separate the.i Au-

(b) o,
o,
:h -Ot--AL) Au 4,,A'."



Ni y int cr i ( A ic is bilin eitia o N































and three possible reactions producing Ni-Ga-O phases and (d) oxidized contact
p-sche


0,


N. ,




-('-.|N





AAu-nc C Au-nclt






Figure 2-10. Schematic illustration of the reaction mechanism and diffusion behavior
during the oxidation ofNi/Au on p-GaN. (a) Early stage of oxidation reaction,
(b) pronounced interdiffusion and crystalline NiO grow up to separate the Au-
Ni alloy into discrete islands, (c) Au-rich islands building epitaxially on p-GaN
and three possible reactions producing Ni-Ga-O phases and (d) oxidized contact
scheme with optimum phase distribution.








low specific contact resistance because NiO was considered as a p-type semiconductor in

direct contact with p-GaN. But this explanation is still controversial. Two different

explanations have been suggested: (1) favorable band line-up when p-type NiO layers

form between Au and p-GaN (Ho et al.) 44, and (2) high hole concentration in p-GaN.45

The effects of NiO on specific contact resistance were reported by several researchers

GaN.46'47 Several investigators examined the role of Au in Ni/Au contacts 43-50 and

reported that Au is necessary because Ni-only contacts showed high specific contact

resistance compared to Ni/Au contacts.


2.7 Measuring Specific Contact Resistance

The specific contact resistance is determined by current-voltage (I-V)

measurements using CTLM (circular transmission line method) patterns.51 For CTLM

measurements, the total resistance, Rt, between two electrodes separated by a circular gap

is represented by equation (3).



R, R 1
R= x In ++ L, 1 ------------(3)
2 I r r R r



where Rsi, is the sheet resistance of p-GaN, R and r represent the radius of the outer and

inner electrode, respectively, and Lt is the transfer length. The total resistance can be

measured and plotted as a function of In(R/r). The fitting of this graph using the least

squares method is conducted to get a linear curve of Rt vs. In(R/r). The slope of this fitted

linear curve is the sheet resistance of the substrate and the intercept at In(R/r) = 0 is Rsh

Lt /(r n). The specific contact resistance can be calculated using the following equation:








P = L,2 x R,, ------------------(4)


2.8 Conductivity of NiO

The conductivity of NiO is due to non-stoichiometry.52,53 The formation of defects

is the following:


1 02 (g) O,,X + x
2

VN, t
V, < VN, + h

where ,,x : an oxygen anion on an anion site
S: a neutral Ni vacancy
'VNi
VNi : singly charged Ni vacancy
VN, : doubly charged Ni vacancy



When oxygen is added to NiO, the nonstoichiometry is enhanced and the hole

concentration increases.

2.9 Optical Properties of NiO

2.9.1 Optical Gap of NiO

It is known that the optical gap of NiO is 4.0eV.54 As shown in Figure 2-11, the

optical absorption of NiO begins at 3.1eV and a first peak is at 4.3eV. A plateau in the

absorption is maintained up to about 9eV and the absorption rises again beyond 9eV. The

optical absorption of NiO, 4.0eV, comes from the point that the line with maximum slope

of absorption intersects the averaged absorption between 4.3eV and 10eV. For

semiconductors, the optical gap is defined as the difference between the smallest

ionization and the smallest electron affinity energy, in other words, the energy one needs





18


to create a free electron-hole pair in the solid. The optical gaps can also be observed in

free molecules like NiC12 and NiBr2. For these cases, the optical gap is the energy it

needs to transfer an electron from a ligand orbital onto a metal orbital and is called a

charge transfer gap.54




/ i



N ,












Figure 2-11. Optical absorption coefficient of NiO. 4.0eV is the energy obtained by
extrapolating the flat absorption between 4.3eV and 9eV, and intersecting it
with the tangent at the steepest point of the absorption between 3.1eV and
4.3eV.


2.9.2 Nickel Oxide Deposited by Sputtering

Some researchers investigated the transmittance of NiO layer produced by RF reactive

magnetron sputtering.55-58 In the research of Lu et al.59, NiO films were deposited under

different O2/Ar ratios. According to their results, the transmittance of NiO films shows a

strong dependence on the 02/Ar ratio. With increasing Oz/Ar ratio, the transmittance

decays as shown in Figure 2-12. It should be noted that all samples have Ni deficiency

and atomic percent of oxygen increases with increasing oxygen flow ratio as shown in






19



Figure 2-13. The resulting high Ni vacancy concentration makes many charge transfer


transitions, which promotes optical absorption in visible light.


1... 4 ... S-. ~ -- --..l.-...
If Iil liii 91P1 ':: (U


Figure 2-12. Transmittance of NiO as functions of oxygen flow ratio at the substrate
temperature of 400C.


58


X
o56






o52




wn


0 20 40 60

Oxyqen flow ratio (%)


80 100


Figure 2-13. Atomic percent of oxygen in the deposited NiO films as a function of
oxygen flow rate ratio.


U








I I I.


~sc~~


50












CHAPTER 3
GOLD/PRE-OXIDIZED NICKEL OXIDE/P-GALLIUM NITRIDE CONTACTS

3.1 Experiments

3.1.1 Sample Preparation

3.1.1.1 Ni/Au ohmic contact

In order to confirm that Ni/Au annealed in oxidizing ambient shows ohmic contact

characteristics under these experimental conditions, and that the quality of the p-GaN is

sufficient to produce ohmic contacts, it is necessary to produce a reference sample. For

this reference sample, Mg- doped p-GaN (hole concentration -2X1017cm'3) was cleaned

in an aqua-regia (HCl:HNO3 =3:1) solution for 5min at room temperature followed by a

DI water rinse and a N2 dry followed. Ni 50A and Au 50A were deposited on pre-cleaned

p-GaN by e-beam evaporation, and annealed at 5000C for 10min in air. E-beam

deposition in this research was conducted under 2X 106 torr base pressure. CTLM

(Circular Transmission Line Method) patterns were made through a lift-off process.


3.1.1.2 Au/pre-oxidized NiO/p-GaN contacts

Mg-doped p-GaN was cleaned in the aqua-regia solution for 5 min at room

temperature to remove the native oxide and/or contamination layers. A DI water rinse and

N2 dry followed. 50A of Ni was then deposited on cleaned p-GaN by electron-beam (e-

beam) evaporation. The deposited Ni was annealed at 500C for different times (0, 10s,

20s, 60s, 120s, 180s and 300s) in air. 1000A of Au was deposited on top of the oxidized

Ni by e-beam evaporation. CTLM patterns (Figure 3-1) were produced through a photo

and dry etch process. AZ-1529 positive photoresist was coated (4000rpm, 40s) and










exposed under I-line (,=405nm) for 105 s. Dry etching was conducted in an ICP dry


etcher using Cl2 and Ar gas to define the CTLM patterns. The experimental sequence of

sample preparation is presented in Figure 3-2.



I10pm 20pmn 30pm








60tpm
40pnm 50p.im











90pm
70pm 80m /


300pm


Figure 3-1. Schematic diagram of CTLM patterns









Cleaning: Aqua-regia
(HCl:HN03=3:I) 5min
P-GaN

Ni 50A deposition by e-beam evaporation


P-GaN
Anneal at 5000C in air for various time
S(O-300s)


P-GaN


SAu 000A deposition by e-beam evaporation


-Au


P-GaN



Photolithography to make CTLM pattern


PR



P-GaN


SDry etch by ICP dry etcher

Au

F7 F"Ni and/or NiO


Figure 3-2. Process sequence to make CTLM pattern for pre-oxidized samples








3.1.1.3 NiO sheet resistance

The n-type Si (111) was oxidized at 9500C for 24h in air to form an SiO2 surface

layer in order to electrically isolate the deposited Ni from the Si. A 50A Ni layer was then

deposited on SiO2 by e-beam evaporation using the same chamber as that used to deposit

the Ni on the p-GaN for the Au/pre-oxidized NiO/p-GaN experiments. Therefore, the Ni

deposited on SiO2 has the same thickness and same characteristics as Ni on p-GaN for the

Au/pre-oxidized NiO/p-GaN experiments. After deposition, the samples were annealed at

5000C in air for different times similar to those used for the Au/pre-oxidized NiO/p-GaN

experiments. Since annealing was conducted with Au/pre-oxidized NiO/p-GaN samples,

annealing conditions are the same as those for Au/pre-oxidized NiO/p-GaN samples. It is

evident that characteristics of NiO for different anneal times should be the same as the

NiO which was produced for Au/pre-oxidized NiO/p-GaN experiments. CTLM patterns

were produced on the NiO layer through a lift-off process. Photolithography was

conducted by using NR7-1500P negative photo resist and Au 700A was deposited on

patterned photo resist by e-beam evaporation. A schematic of the samples used for this

experiment is shown in Figure 3-3. To measure the conductivity of NiO, four point probe

and Hall measurement methods were attempted but useful data cannot be achieved

through these measurements due to the high resistivity of NiO.

3.1.2 Analysis

3.1.2.1 I-V characteristics

I-V measurements were conducted using HP4155 Semiconductor Parameter

Analyzer.










Au

FZ NiO

SiO2

Si

Figure 3-3. Vertical structure of CTLM patterns


3.1.2.2 AES analysis

A Perkin-Elmer PHI 660 Scanning Auger Multiprobe was used for depth profile

analysis.


3.2 Ni/Au Ohmic Contact

Measured I-V curves of annealed sample are shown and compared with that of as-

deposited sample in Figure 3-4. It is evident that the current increases and the I-V curve

is almost linear after annealing implying that Ni/Au metal annealed in air shows ohmic

contact characteristic as published by several researchers.


3.3 Au/pre-oxidized NiO/p-GaN Contact

3.3.1 Current-Voltage (I-V) Characteristics

I-V measurements were conducted and are shown in Figure 3-5. The as-deposited sample

shows an almost symmetric but non-linear I-V curve (Figure 3-5 (a)), while the current is

observed to increase for the sample that was annealed in air for 10s. After 20s annealing,

the current decreases and there is no significant change until 300s annealing. I-V curves

for all annealed samples in air shows rectifying I-V characteristics. It is clear that Au/pre-

oxidized NiO/p-GaN contacts are not ohmic in nature but rather exhibit Schottky










0.002

0.0015

0.001

0.0005
--- before anneal
S--*- after anneal
S-1 0 0.5 1 1.5 2
S-0.0005

-0.001

-0.0015

-0.002
Voltage (V)



Figure 3-4. Linear I-V characteristics after annealing at 5000C for 0min in air.
(Ni50A/Au50A)


barrier behavior. In order to compare current among samples with different anneal times,

the fourth largest circular pattern (electrode space is -1201m) was selected from the

CTLM pattern and I-V curves for that size are plotted in Figure 3-6. As can be clearly

seen, the 1Os-annealed sample shows the highest current values within the range of

voltage measured (-2.0V 2.0V). The currents measured at 1.0V, 1.5V and 2.0V are

shown in Figure 3-7 (a), (b) and (c), respectively. The current changes at selected voltage

with anneal times display quite similar trends. The average current value at 1.5V

increases to 260gA for the sample annealed forl0s from 170A for the as-deposited

sample, and then goes down to 50pA, the minimum current in this research, for the 20s-

annealed sample. The sample annealed for 60s shows a slightly higher current (-20p.A)

compared with 20s-annealed sample and the current slightly goes down after 60s

annealing. Since the current variation between 60s and 300s is negligible, it concluded

that there is no significant change after 60s.




































.2 16 14 -2 I 40 406 04


2 1 -1.4 1 -05 -0 01 C4 07 13 1 6 19
V114l

1 t7 14 .1 -04 AS 42 0I 01 04 1 t1 Is to
0~a4.4W


L___ --. _________________
11444.114


(e) (f)



Figure 3-5. I-V curve for pre-oxidized samples annealed at 5000C in air for (a) 0, (b) 10s

(c) 20s, (d) Imin, (e) 2min, and (f) 5min.


come




a oo .. -.. _



OQ6- -



o cx4- __^y)





0 2 .1 .. ......



(b)














omn




u00005

-0 -- ---d


I -I 1It O1 S 0S 02 Or 0 07 1 t3 (8 tf




(d)


12 14 s6 t- 2


0O15


000015


f o3 o


0O000S









0.0006

0,0004
S--- no_anneal
0.0002 --- lOsec
S*r 20sec
0 alllMi 1 min
Sl -- 2min
-0.0002 --- 3min
S....... -+.- 5min
-0.0004

-0.0006
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2
Voltage(V)


Figure 3-6. I-V for Au/pre-oxidized NiO/p-GaN samples with different anneal time


3.3.2 Depth Profile Analysis

The depth profiles were investigated using AES (Auger Electron Spectroscopy) analysis

and are shown in Figure 3-8. The as-deposited sample shows that pure Ni exists on p-

GaN and Au is on Ni. After annealing for 10s, an oxygen peak was detected. The

intensity of this peak is relatively low (-20%) compared with the pure Ni peak. In terms

of peak position showing maximum intensity, the oxygen peak is seen near the interface

between Ni and Au. It is evident through AES depth profiles that NiO layer was formed

through oxidation of Ni at the surface during 10s annealing, but pure Ni still exists near

the interface between Ni and p-GaN after 10s annealing. After 20s annealing, the oxygen

peak is seen over the entire area ofNi and there is no significant change up to 300s

annealing. The oxygen peak position showing maximum intensity is very close to that of

Ni. Based upon these results, Ni was found to partially oxidize during 10sec annealing

and oxidize fully after 20s annealing near the GaN interface.


































no anneal 0lsec 20sec Imin
Anneal time


no_anneal 10sec 20sec 1min
Anneal time


2min 3mm 5mnn


2mm 3min 5mm


0 0007



00000




<.00004
0 0005





00003


00002


00001


0


noanneal 10sec


20sec 1min
Anneal time


2min 3min 5mmn


Figure 3-7. Current measured at (a) 1.OV, (b) 1.5V and (c) 2.0V


0 00012


0001


0 oo00008



000006


000004


000002


0


000035


0 0003


0.00025


< 0.0002


000015


00001


000005


0


(c)


I













----......-- A -- i .... ..---i ---- No an eal
-Au



.. .. a a a....

----- ------------- 1----
T.- J



4 "! ;'" : 20sec

,' -'
I. *, ,
.I


. *Ii. ..... .
.... .. ... ... ..... ......... ...... ....... .


-- -- --- ------ ----
losec



.[-.....j oxygen



.: -^.,- _i "--


H ::i:~:i::
0,


Figure 3-8. AES depth profile for pre-oxidized samples with various anneal time. X-axis
and y-axis represent sputter time (sec) and peak intensity (a.u.), respectively.



3.3.3 Post Annealing of Au/pre-oxidized NiO/p-GaN Samples

Since there is no heat treatment after the Au was deposited on the top of pre-

oxidized NiO, Au/pre-oxidized NiO/p-GaN samples should not display any significant

effect of the Au in terms of reaction during annealing. In order to investigate Au effect on

the reaction and specific contact resistance, as-deposited and 1Os-annealed samples

(Au/pre-oxidized NiO/p-GaN) were annealed at 500C in air for different times (up to

60min). I-V curves for as-deposited and 1 Os-annealed samples after post annealing are

shown in Figure 3-9 (a) and (b) with the I-V curves before post annealing, respectively.

As shown in Figure 3-9 (a), the current increases after post annealing but the I-V curve is

not linear although it was annealed in air with Au on the top of Ni. In this case, it should

be noticed that the Au thickness is too thick (1000A) to make the microstructure (layer

inversion structure) that shows ohmic characteristics. Os-annealed sample showing the

highest current among Au/pre-oxidized NiO/p-GaN samples was also annealed at 500C











0.0015


0.001


0.0005 r- Before anneal

4-- Snin
0--- 10min
I wl25rin

-0.0005 -- -*- 55rin


-0.001


-0.0015
(N (P (NI I) C 0 o CO CD r(
7 9 9 0 -
Voltage (V)

(a)





0.0008


0.0006


0.0004 -.- Before anneal
--- 30sec
Z 0.0002 2min

.. lOmin

--e 30min
-0.0002 ..- -+-- 60min


-0.0004


-0.0006
CM (P (N CO 0 T CO (P4 (0 (N
.- 9 9 0 -
Voltage(V)


(b)




Figure 3-9. Effect of post annealing on I-V characteristics for (a) as-deposited sample and
(b) 10s pre-oxidized sample





31


in air and I-V curve is shown in Figure 3-9 (b). After post annealing (especially after

30min), the current decreased and I-V curves remain rectifying characteristics. It appears

that oxygen in pre-oxidized NiO diffused to remaining pure Ni and produced NiO near

the interface between Ni and p- GaN during post annealing. At the same time, oxygen in

air reacts with Ni at the side-wall to produce NiO. These processes are shown in Figure

3-10 schematically. The fact that the 10s pre-oxidized sample remains rectifying with low

current after post annealing is one of the proofs that pre-oxidized NiO/p-GaN interface

does not make ohmic contact.


3.3.4 NiO Sheet Resistance

In order to investigate the effect of a NiO on determining specific contact

resistance, it is necessary to investigate the conductivity of NiO formed by annealing Ni

at 5000C in air for different times.

The sheet resistance of the substrate layer can be calculated from equation (3). The

linear slope of the graph is Rsh/27 when plotted with Rt vs. In(R/r).



Au Au Au
NiO

~Ni

p-GaN p-GaN p-GaN




Figure 3-10. Reaction process during post annealing. (a) before post annealing
(b)oxygen diffusion during post annealing. Oxygen flow is designated by
arrow(c) NiO formation near the interface between Ni and p-GaN








In Figure 3-11, I-V curves for these samples with different annealing times (10s 180s)

in air are shown. For 1Os-annealed sample, the I-V curve is linear up to -0.4V, but over

0.4V, the current exponentially increases with applied voltage up to 2.0V. This

phenomenon is similar to that of the negative applied voltage indicating that the I-V

curves are symmetric about the origin (0 volts). In contrast, all samples annealed for 20s

or longer show linear I-V characteristics. The non-linearity shown in the 10s-annealed

sample can be explained by the fact that considerable amounts of pure Ni still exist under

a thin NiO layer consistent with AES depth analysis. For low applied voltage less than

0.4V, the current is limited due to high resistance of NiO. As the applied voltage

increases, the electric field applied through the thin NiO also increases. The increase of

applied field can induce tunneling current between the Au and the remaining pure Ni

through the thin NiO layer. The large number of free electrons in Ni and Au act as the

source for the high tunneling current. As the annealing time increases, the thickness of

the NiO layer also increases. After 20s annealing, most of the Ni was oxidized and

current path between two electrodes is Au/ NiO /Au over the entire voltage range. Since

linear I-V characteristics were achieved for 20s, 60s, 120s and 180s-annealed samples,

the sheet resistance of NiO can be calculated by equation (3). For 10s-annealed sample,

short voltage range, 0 to 0.4V, showing linear correlation between current and applied

voltage was selected for sheet resistance calculation. Based upon equation (3), the linear

fitting was conducted to calculate sheet resistance (Figure 3-12). From the slope of the

fitted line, the sheet resistance was calculated and is shown in Figure 3-13. O1s-annealed

sample shows the lowest sheet resistance, 9.92X104 21/D among all samples. The sheet

resistance started to increase to 4.72X 10"' 1/D1 for the sample annealed for 20s and















0.015 5 00E-10


001 300E-10
0005 00E 10




0. -21 00E-10 -
'0 0-2 00E-1000


-4 ODE-1 -
-0 01
-500E-10
0000E-6 010
-0015 0 N Co N0NW 0 -
o0 oo Voltage V)
Voltage (V)


(a) (b)




I 0OOE09 8 00E-10
8 00EO -10 .6 00E-10

4O00E-10
4 00E-10
O 200E-10 -OO2E-10
0000 DE 0 E000E+00


4 E00- 10
-4 00E- 10. .
-60OE-10

e -1 c f O f 600E-g Ni at 10
1 00E09 -8 E-10 s s a e

Vcltage(V) Voltgae (V)




(c) (d)





1 50E-09


1 00E-09


500E010


0 00E*00

.5 00E-10


1 00E-09


1 50E-09

Vontage(V)


(e)






Figure 3-11. I-V curves for NiO formed by annealing Ni at 5000C in air for (a) 10s

(b) 20s (c) 60s (d) 120s and (e) 180s


7


-4 -3













I 4E+11

I 2E-11

1E+11

8E*10




46E10

2E-10


0 0.2 04 06 08
In(R/r)




(a)


30000

30000
25000





. 15000


10000

5000


0


25SE11 r.----


2E+11


15E.11

-


5E.10


0 02 04 06 08 1 0 02 04 06
In(RPr) In(RA)


08 1 12


1 12E+11

1 11El11

1 1E+11

S109E.11

SIOSE+11

. 107E+11

1.06E+11

1.05E+I

1 04E+11


0 02 04 06 08
In(RIr)

(e)


Figure 3-12. Linear fitting of In(R/r) vs. Rt curve for (a) 10s (b) 20s (c) Imin (d) 2min

(e) 3min annealed samples


0 02 04 06 0. 1


7E11

6E+11
5E+11

- SE-11

4E.11









0
3E611

E
3


c~LC









1.00E+14
1.00E+13 .04E+12
1.00E+12 9 E+11
1.E+4.72E10 -~ ~--02.05E+11

1.00E+10
1.00E09 -
1.00E+08
w 1.00E+07
1.00E+06
9 2E+04
1.00E+05
1.00E+04
10sec 20sec rrnin 2min 3min
Anneal time

Figure 3-13. Sheet resistance of NiO formed by annealing Ni at 500C in air for different
times


reached 6.04X1012 nQ/U for the sample annealed for 60s. After 60s, the sheet resistance

decreased slightly reaching 2.05X10I" Q/D at 180s. According to the sheet resistance

data, it is suggested that one of the reasons for the high currents in the Au/ Os pre-

oxidized NiO/p-GaN sample is due to the low sheet resistance of the oxidized Ni layer

(including pure Ni). The relatively low current in the sample annealed for 20s (Figure 3-

7) can be due to the high sheet resistance of NiO. This sheet resistance increase can be

one of the factors for the current decrease observed for Au/pre-oxidized NiO/p-GaN

samples with increasing annealing time. It is noted that 1 Os pre-oxidized sample shows

high current compared with the as-deposited sample shown in Figure 3-7 even though the

10s-annealed NiO should have higher sheet resistance compared with pure Ni. Therefore,

lower current suggests that the reaction between Ni and p-GaN during annealing in

oxidizing ambients is one of the important factors in determining specific contact

resistance. Significantly, Ishikaws et.al60 reported a very thin (-2nm) amorphous layer

between the Ni and p-GaN after deposition of Ni on cleaned p-GaN and suggested
























(a) (b)

Figure 3-14. High resolution TEM image of (a) as-deposited Ni on p-GaN and (b) after
annealing at 400C for 10min.


that this layer could be either a native oxide or a contamination layer as shown in Figure

3-14 (a). They also reported that this amorphous layer disappeared after annealing at

4000C for 10min as shown Figure 3-14(b). If their results are valid, a similar reaction is

expected to have occurred in the 50A Ni on p-GaN after annealing at 5000C for 10s.

Thus, the native oxide or contamination layer removal could explain the observed high

current in spite of the higher sheet resistance of the oxidized Ni layer compared to that of

pure Ni for the as-deposited sample.


3.3.5 Rectifying Characteristics of Au/pre-oxidized NiO/p-GaN Contact

The rectifying characteristic of Au/pre-oxidized NiO/p-GaN samples can be

explained using energy band diagrams. Figure 3-15 shows an energy band diagram for p-

NiO/p-GaN contact. Figure 3-15(a) shows the equilibrium energy band diagram of p-NiO

and p-GaN before contacting each other. The parameters used to construct this energy

band diagram are listed in Table 3-1. According to the results of Nakayama et al., the










p-GaN


E Ni)



E, NO


N iO





Eb N -------------^----------


p-NiO


E GaN





EvGaN
*-v


p-GaN


EcGaN





E GaN
v


Figure 3-15. Energy band diagram of NiO and p-GaN (a) before contact (b) after contact



Fermi level of p-GaN is located at 0.13eV above the top of the valence band at 300K.


The Fermi level of undoped NiO is located at 0.5eV above the top of the valence band.


Based upon these parameters, the offsets of conduction band and valence band can be


calculated to be 2.7 and 2.1eV, respectively. For p-NiO connected to p-GaN, the energy


band diagram was changed and is shown in Figure 3-15 (b). The p-NiO surface bands


p-NiO








bend upwards and p-GaN surface bands slightly bend downwards at the p-NiO/p-GaN

interface. A notch for holes in p-NiO exists near the interface between p-NiO and p-GaN

where holes are trapped because the Fermi level is located below the top of the valence

band. Energy barrier height at the p-NiO/p-GaN interface can be calculated by

Cserveny's concept. Under forward bias conditions, which occurs when a positive

voltage is applied to Au, holes which flow from p-NiO towards p-GaN meet an energy

barrier (0.20-0.31 eV, which depends on the hole concentration of NiO) at the interface

between p-NiO and p-GaN. Under reverse bias, there is relatively low energy barrier

(0.07-0.18 eV) between p-NiO and p-GaN. Since CTLM pattern was used to measure I-V

characteristics for pre-oxidized samples, the current path is Au/p-NiO/p-GaN/p-NiO/Au

as shown in Figure 3-16. Since this current path consists of two p-NiO/p-GaN interfaces,

one of the interfaces has a relatively high energy barrier and the other interface has a

relatively low energy barrier whether positive or negative voltage is applied on Au. The

Schottky barrier (0.27-0.49 eV) between p-NiO and p-GaN is the reason of non-linear I-

V characteristics for pre-oxidized samples.


Table 3-1. Properties of p-GaN and p-NiO at room temperature

Eg (eV) x (eV) Er-Ev (eV)
p-GaN 3.4"' 4.1, 0.13
p-NiO 4.0-' 1.4 0.5


Since the current path has two interfaces (one has relatively high energy barrier and

the other has relatively low energy barrier) whichever voltage applied, it is suggested that

I-V curves for pre-oxidized samples should be symmetric. But I-V curves for pre-

oxidized samples show asymmetric shape. This asymmetric behavior can be explained









n Au

p-NiO


p-GaN


Figure 3-16. Current path of CTLM patterns for pre-oxidized samples


by noting that the sizes of the two electrodes are different as shown in Figure 3-17. It is

evident that the contact area and energy barrier combinations made asymmetric I-V curve

for pre-oxidized samples. If p-NiO/p-GaN with small contact area has high energy

barrier, and p-NiO/p-GaN with large contact area has low energy barrier, the current

measured should be low compared with the opposite case in terms of combination of

contact area and energy barrier height. Figure 3-18 shows a comparison of I-V curves


Figure 3-17. Different electrode size for CTLM patterns


Electrodeod









with different voltage biases on the small electrode curve (a) and large electrode (b). The

I-V curve with different voltage bias is consistent with the explanation that asymmetric

characteristics of I-V for pre-oxidized samples is due to a combination of contact area

and energy barrier height.

0.0005
0.0004
0.0003 A
0.0002

< 0.0001
S 0 .XXXXXXXX.(a)
Co oo qx
5 -0.0001 o o 9o o o o
-0.0002
-0.0003
-0.0004
-0.0005
Voltage (V)


Figure 3-18. I-V curve with different voltage bias (a) voltage on small electrode (b)
voltage on large electrode













CHAPTER 4
REACTION BETWEEN NICKEL AND P-GALLIUM NITRIDE DURING
ANNEALING IN OXIDIZING AMBIENT

As already stated, the theory for the low resistance ohmic contact behavior of

oxidized Ni/Au contacts is still controversial. In fact, the presence of a reaction phase

between the metal and semiconductor may play an important role in determining contact

resistance61, i.e., the reaction between Ni and p-GaN during annealing could be one of the

most critical factors in determining specific contact resistance on p-GaN metal contacts.

Based on this possibility, the reaction between Ni and p-GaN during annealing has been

the focus of numerous studies over the past several years.62-64

The phase diagrams for Ni-Ga and Ni-N are presented in Figure 4-165 while the Ga-

N-Ni ternary phase diagram at different N2 pressure was calculated and reported by

S.E.Money et al.66 Several researchers have reported the reaction between Ni and p-

GaN.62"64 For example, Sheu et al. reported XRD results which indicated that Ni gallide

(Ga3Ni2, Ga4Ni3) and Ni nitride (Ni3N) were formed during annealing at 500C for 10

min in nitrogen ambient.62 Guo et al. analyzed the reaction between Ni and n-GaN, and

reported that Ga4Ni3 and Ni3N formed during annealing at 4000C for 20min. in a nitrogen

ambient.63 Surprisingly, Guo et al. reported a Ni-nitride (Ni3N) peak just after deposition

of Ni without any subsequent heat treatment. However, Venugopalan et al. also reported

through XRD analysis that they could not observe a new reaction phase between Ni and

GaN when it was annealed up to 6000C in a nitrogen ambient.64 They reported that a Ni-

Ga solid solution was formed after a 6000C anneal while Ni gallide (Ni3Ga) was reported













Alunm:, P'e-rrrnt j ii 2> :0 is MI s 73 *a 90 i3
.*ai ---- --- -- ---- ~~~- ~-~ ~~~~-- --- ----- *****

14561C,










i ,i\




I.






l
10 r0n


Atomic Percent Nllr:.;'n
o 0 30 to 0 60 70 a


IIn taph




"(Oi
400
S (I)[

i w

0t




200



too1 2:z



0 II 20 30 10 S0 W0 0
Ni Weight Percent Nitrocen




Figure 4-1. Phase diagrams of (a) Ni-Ga (b) Ni-N systems


(a)









































(b)








after annealing at temperatures above 750C. Thus, their results indicate that no reaction

phase was formed between Ni and GaN at 4000C annealing in direct contrast to the report

by Guo et al.

Besides this controversy over the formation of new reaction phases between Ni and

GaN, it should also be pointed out that most of the annealing treatments in these studies

were conducted in nitrogen or argon ambients. However, the annealing ambient for the

standard Ni/Au contacts that show low specific contact resistance is an oxidizing (air)

ambient. A very limited number of papers focused on the reaction phase between Ni and

p-GaN during annealing in oxidizing ambients have been published.6

In the light of the reaction between Ni and p-GaN, the difference between oxidizing

ambient and nitrogen or argon ambient is whether oxygen (or NiO) exists on the top of Ni

or not during annealing. It is not clear whether oxygen actually plays a role in forming a

reaction phase between Ni and GaN during annealing.


4.1 Experiments

4.1.1 Sample Preparation

Mg-doped p-GaN was cleaned using an aqua-regia solution for 5 min at room

temperature to remove the native oxide or contamination layers. 1500A of Ni was then

deposited on the cleaned p-GaN by e-beam evaporation. This thickness was selected to

ensure that there was enough starting materials for an interfacial reaction to develop.

After deposition, the Ni-coated sample was annealed at 5000C for 4min in air to produce

NiO layer on the top of the Ni but not all of the way through. The annealing time was

carefully controlled because longer times result in spelling of the Ni layer. After

annealing in air, the sample was encapsulated in a quartz ampoule that had been purged








with Ar gas three times prior to sealing. The partial Ar gas pressure inside the quartz

ampoule was selected to maintain approximately 1 atm during annealing at 500C. The

encapsulated sample was then annealed at 5000C for 24 hours in order to produce a

sufficient amount of reaction product which could then be analyzed by TEM. After

annealing, the quartz ampoule was broken and the sample was cooled in air.

4.1.2 Analysis

4.1.2.1 XRD measurement

X-ray diffraction was performed using an analytical high resolution X-ray

diffractometer (Philips X'Pert MRD), operated with Cu K, as the radiation source at

45kV and 40mA.


4.1.2.2 AES analysis

A AES Perkin-Elmer PHI 660 Scanning Auger Multiprobe was used for depth

profile analysis.


4.1.2.3 TEM analysis

High resolution TEM (JEOL 2010F) with Oxford EDX detector was used for

analysis of interfacial reaction between metal and p-GaN. The accelerating voltage was

200kV.


4.2 X-ray Diffraction Analysis

XRD analysis of the sample after annealing revealed the Ni and NiO peaks as well

as GaN peak (Figure 4-2). One additional and relatively strong unknown peak was

observed at a 20 value of 41.7; this peak position is the same as the Ni-nitride peak

suggested by Guo et al.63 In order to investigate this unknown peak, XRD analysis was









Unknown peak
1.0E+12
GaN 0004
1.0E+11 K peak NiO il K peak of -
1.0E10 of GaN / GaN 0004
1.0BE09 0002 / Nio 20
1.OE09 V



1.0EB06 -
--24h anneal
D 1.0E+05 GaN 0002 Ni 11 Ni 200
1.0Ei04

1.OEqC
1.OE40l
1.0E+01
1.0E,00
20 24 28 32 36 40 44 48 52 56 60 64 68 72 76

Angle (two theta)



Figure 4-2. XRD peaks for 24h-anenaled and as-deposited samples (Nil 500A)


conducted for p-GaN without any layer on top of p-GaN and shown in Figure 4-3. The

strong XRD peak at 41.70 can be observed in Figure 4-3. Therefore, this peak didn't

come from Ni-nitride but from the substrate. The position of this peak matches the (0006)

plane of sapphire based upon JCPDS index.


4.3 Auger Electron Spectroscopy Depth Analysis

AES depth profiles for the as-deposited Ni on GaN sample and the sample after

annealing are shown in Figure 4-4. As can be seen, the Ni film exists on GaN before

annealing. After annealing, the oxygen exists over the entire range of Ni. This means

oxygen diffused towards the interface between Ni and p-GaN, and NiO was formed by

oxidation of Ni during the 24h anneal.










7.0E+05

6.0E+05

5.0E+05

S 4.0E+05

3.0E+05

2.0E+05

1.0E+05

n Ec+lnn


St L o m 0 Ange (wo tta) (
Angle (two theta)


Figure 4-3. Strong XRD peak at 41.70 from GaN substrate


kn: l x: 0232

..----,------ ...."-- ...."..-. IN i


-,----- --- ----, ,

.---- ---. ...-- -.. .--- .




... ... ....... ......... ..... ...I .. .
r-----r---~----r--- -






0 20 40 60 80 100 120 1280 180 2
Cycles

(a)


Ni
x6871 ................







...........O ..... .
Oxygen i








0 80 10 24 320 40
Cycles

(b)


Figure 4-4. AES Depth profiles of (a) as-deposited sample (b) the sample annealed for
24h at 500C. X-axis and y-axis represent sputter time (sec) and peak intensity
(a.u.), respectively.


-I c


u.uI r,








4.4 Transmission Electron Microscope Analysis

Since there was no evidence for a new reaction phase after 24h annealing based on

XRD analysis, it can be suggested that the 4min annealing in air plus 24h annealing in Ar

at 500C either produces no new reaction phase between Ni and p-GaN, or that the

reaction phase is very thin near the initial Ni/GaN interface. In order to investigate this

possibility, TEM analysis was conducted to examine the interfacial region. The low

magnification TEM image of the interface between Ni and p-GaN is shown in Figure 4-5

and indicates that fine-grain-NiO formed during annealing. Figure 4-6 shows an SAD

(Selected Area Diffraction) pattern from NiO that is consistent with its fcc structure. The

interface between NiO and p-GaN was observed at higher magnification and a thin,

relatively bright layer was observed at the interface (see arrow in Figure 4-7). However,

this layer appears to simply result from a slight overlapping of the two different crystal

structures that meet at the interface. The other possible explanation for this layer's

formation is from damage or locally thin area, induced by ion bombardment during

FIBing.6870 In an effort to determine if there was anything at the interface, an EDX line

scan was conducted across the interface (Figure 4-8). As can be seen, a strong Ni peak

was detected on the NiO side of the interface with Ga and N peaks on the GaN side. A

gradual transition from one side to the other was observed consistent with no additional

phases at the interface. Thus, the HRTEM and EDX data indicate this bright layer is not a

reaction phase indicating that no reaction phase forms between Ni and p-GaN after

annealing at 5000C in an oxidizing ambient.











/ I. .**

fc. ^'S ? !*.


Figure 4-5. TEM image of NiO and p-GaN after 24h annealing










NiO(ll11


NiO(200)



a.^^


Figure 4-6. SAD pattern of NiO produced by annealing Ni 1500A






















Figure 4-7. High magnification image of the interface between NiO and p-GaN


D


NiO


L "


GaN .' j7
Figure 4-8. EDX analysis across the interface between NiO and p-GaN
Figure 4-8. EDX analysis across the interface between NiO and p-GaN












CHAPTER 5
INTERPLANAR SPACING CHANGE OF GALLIUM NITRIDE DURING
ANNEALING IN OXIDIZING AMBIENT

5.1 Selected Area Diffraction Patterns for GaN

SAD (selected area diffraction) patterns of GaN, previously discussed in chapter 4,

were obtained in the as-deposited condition and after annealing at 500"C for 24h (Figure

5-1). Pure Au was used as a standard to calibrate the camera length. The d-spacings of the

major GaN planes are provided in Table 5-1, where it is apparent that the interplanar

spacing of the (0002) plane is lower after the 24h anneal while the (2110) plane has an

increased d-spacing after annealing. Each plane is presented in Figure 5-2. It is evident

that the interplanar spacing for the plane parallel to the surface, i.e., the (0002) plane,

contracts during annealing, and that for the plane parallel to the c-axis, i.e., (2110),




0002 110 0002








2110



(a) (b)

Figure 5-1. SAD patterns of GaN (a) as-deposited (b) after annealing at 5000C for 24h








Table 5-1. Interplanar spacing comparison

Plane Before anneal After anneal

0002 2.452 2.440

2110 1.609 1.634


(b)


Figure 5-2. Crystallographic view of (a) 0002 (b) 2110 planes of GaN


expands during annealing.

5.2 X-ray Diffraction Analysis of Strain in GaN

The change in the interplanar spacings of GaN, produced after annealing at 500"C for

4min in air and the 24h in Ar, was analyzed by XRD. The as-deposited and annealed

samples were compared with p-GaN with no metallic surface layer (Figure 5-3). The

results show that the (0002) peak of GaN shifts 0.230 towards lower 20 angles (higher d








4.5E+07

4.0E+07

3.5E+07

3.0E+07

2.5E+07

2.0E+07

1.5E+07

1.0E+07

5.0E+06

O.OE+00
O


I
*


- -no depo
- no_anneal
- 24h anneal


Angle (two theta)


Figure 5-3. GaN (0002) peak shift after Ni 1500A deposition and annealing in oxidizing
ambient



spacing) after Ni deposition and reverses 0.07 upon annealing, consistent with the

smaller d-spacing expected from the TEM analysis. The change in interplanar spacing

was calculated and is shown in Table 5-2. It is noticed that the d-spacing of (0002) plane

measured by TEM (Table 5-1) is smaller than that by XRD (Table 5-2). This

phenomenon can be induced by several factors. One possibility is related to the lattice

damage caused by Ga ions during FIB.


Table 5-2. Interplanar spacing of GaN (0002) plane


Before anneal After anneal

20 34.50 34.57

d-spacing 2.5975 2.5932








The other is that the oxidation of GaN can decrease the d-spacing of the (0002) plane. It

was published that Gallium oxynitride had smaller (0002) plane d-spacing compared with

that of GaN.71 It is also possible that this phenomenon is a measurement error from

analyzing SAD patterns. Based upon both TEM and XRD analysis, deposition of 1500A

Ni on cleaned p-GaN results in an elastic dilation in the [0002] direction of GaN. This

tensile strain is partially released by annealing Ni in an oxidizing ambient as shown in

Figure 5-3. The data show that Ni deposited on p-GaN has compressive stress along with

p-GaN surface (parallel to surface) and this compressive stress makes compressive strain

to the parallel plane to c-axis of p-GaN such as (2110) plane. Since the parallel plane to

c-axis of GaN has compressive strain, GaN (0002) plane can have tensile strain after

deposition of Ni. During annealing in an oxidizing ambient, the tensile strain in the

[0002] direction of GaN is partially released due to either NiO formation or relaxation by

annealing. The schematic diagram of this process is shown in Figure 5-4.





c-axis


Ni


NiO

/1 M


Figure 5-4. Schematic diagram of partial strain release of GaN after annealing in
oxidizing ambient (a) before deposition ofNi (b) as-deposited (c) after
annealing in oxidizing ambient


~I












CHAPTER 6
EFFECT OF GALLIUM NITRIDE STRAIN ON CONTACT RESISTANCE

Lattice strain in GaN may affect contact resistance because strain in GaN,

especially the near-surface regions of GaN, may shift the energy bands and/or produce

defects at energy levels inside the band gap. The discrete deep levels within the GaN

band gap can be formed due to (1) native defects, (b) metal-induced bonding and (c)

reaction products.72 Deep level defects at the metal-GaN interface were reported by

Brillson et al.; these metals on GaN induced new deep level emission across the band

gap72. Krispin et al. reported that the internal strain near the as-grown GaAs surface leads

to the formation of an intrinsic defect within the band gap.73 Singh et al. also reported that

the increasing strain with higher carbon fraction in SiC layer plays an important role in

producing deep level defects.74 The defects may result in the production of charge trap

centers or recombination centers near the surface.75 Auret et al. reported that sputter

deposition of metal produces disorder in the semiconductor surface, which results in a

new energy level in the semiconductor band gap; this new energy level may produce

trapping or recombination centers.76 Other authors reported that the barrier heights of

sputter-deposited Schottky contacts on p-type Si 77 and GaAs 78 are higher than those of

similar metal contacts deposited by other less damaging processes. It was shown that this

barrier height alteration was accompanied by the introduction of donor-like defects at and

below the semiconductor surface.


6.1 Experiments








6.1.1 Sample Preparation

Pre-cleaned (aqua-regia, 5min at room temperature) p-GaN samples were coated

with different thicknesses of Ni (1 50A and 600A) by e-beam evaporation to measure

stress versus Ni film thickness.

6.1.2 Analysis

6.1.2.1 I-V characteristics

I-V measurements were conducted using HP4155 Semiconductor Parameter

Analyzer.


6.1.2.2 XRD measurement

For the analysis of strain in GaN, X-ray diffraction was performed using an

analytical high resolution X-ray diffractometer (Philips X'Pert MRD), operated with Cu

K, as the radiation source at 45kV and 40mA.


6.2 Current-Voltage Measurements and XRD Analysis

I-V curves for as-deposited samples with different Ni thickness all display non-linear

behavior as expected (Figure 6-1). Further, it is evident that currents for the samples with

600A of Ni are higher than those from samples with 150A of Ni. XRD analysis (See

Table 6-1 for details) was conducted on the samples coated with 0, 150 or 600A of Ni.

The 0002 GaN peaks (Fig. 6-2) for the different samples indicate that the Ni does cause

this peak to shift to smaller angles (larger d-spacings). Furthermore, the shift of the peak

is highest for the sample coated with 150A of Ni indicating a larger strain in this sample.

Based upon XRD analysis and I-V measurements, it is concluded that the increase of the

strain in GaN induces the decrease of the current value.












8.0E-04

6.0E-04

4.0E-04

2.0E-04

O.OE+00

J -2.0E-04

0 -4.0E-04

-6.0E-04

-8.0E-04

-1.0E-03

-1.2E-03
C (D CN O 0
T- 0 o

Voltage (V)




Figure 6-1. I-V curves for different Ni thickness





6.0E+07



5.0E+07



4.OE+07



3.OE+07



2.0E+07



1.OE+07



0.OE+00


; --150A




600A
----
--*-


00 CN C N
a 0 -- ,-


- no Ni
- ni150
I- ni600


Figure 6-2. XRD peaks of GaN (0002) plane


- c 0 -Angle (two tha) 0 -
Angle (two theta)








Table 6-1. XRD condition for the measurement of GaN (0002) peak

Wavelength (A) 1.54056

Step width ( o) 0.002

Time per step (sec) 2.0


6.3 Effect of Annealing in Oxidizing Ambient on p-GaN Strain

The (0002) GaN peak shift after annealing in oxidizing ambient is due to the

reduced stress caused either by annealing and/or by the formation of NiO. In order to

investigate the effect of annealing on the strain produced by deposition of Ni, the samples

with different Ni thicknesses were annealed at 5000C for 5min in nitrogen. Subsequent

XRD analysis (Figure 6-3) indicates that the (0002) GaN peak for the 600A Ni sample

shifted to higher 20 (0.040). However, the (0002) GaN peak from the 150A Ni sample

showed a negligible difference after annealing.

These samples were annealed at 5000C for 1 Oh in air to investigate whether or not

the formation of NiO can release the strain. Grazing angle XRD patterns from the

oxidized samples revealed only NiO peaks (Figure 6-4) indicating that the Ni was fully

oxidized by this anneal. XRD analysis of the (0002) GaN peak (Figure 6-5) after

annealing in oxidizing ambient indicates that the difference of peak position between the

sample annealed in nitrogen and the sample annealed in air is negligible. Therefore, it is

concluded that the strain release in GaN after annealing in oxidizing ambient is not due to

the formation of NiO but due to just annealing


6.4 Strain in GaN for Thin Ni

In order to investigate the GaN strain effect for thin Ni (less than 150A), 50A Ni












4.5E+07

4.0E+07

3.5E+07

3.0E+07

2.5E+07

2.0E+07

1.5E+07

1.0E+07

5.0E+06

O.OE+00


M a* CD UO 'J* C*' 'J* CD a n U) (' D vt
mc, c c Ao Cl Co th e0 a
Angle (two theta)


-





i as-deposited I
after anneal



1 "~""a


c (i' D co ^ (4 CAg (two t c Uto) c
C M C M C') C' C' (') M CIii U U U I)
Angle (two theta)


Figure 6-3. GaN (0002) peak comparison for (a) Ni 600A (b) Ni 150A before and after
annealing at 500oC for 5min in nitrogen


- as-deposited
-- after anneal


4.0E+07

3.5E+07

3.0E+07

2.5E+07

2.0E+07

1.5E+07

1.0E+07

5.0E+06

O.OE+00














50 N 111 NO 200
45 -

40 --- -- -

35


S25

. 20

15

10

5

0
0 0 0 0 0 0 0 0 m 0 0 0 0 0-

Angle (two theta)


180 ..

160 NiO 111 NiO 20(

140

S 1 2 0 ..

100

c 80

60 -----

40

20

0
0 0 0 0 0 0 0 0 0 0 0 0 0 0
v t Co r- co 0- 'Ag (w ) v O-a
c Moo ) M To '
Angle (two theta)


(b)





Figure 6-4. Grazing angle XRD analysis of NiO produced by annealing (a) Ni 150A (b)
Ni 600A after 500C for 10h in air











4.5E+07

4.0E+07

3.5E+07

3.0E+07

2.5E+07- -

2.0E+07 -

1.5E+07

1.0E+07

5.0E+06

O.OE+00

Angle (two theta) )
Angle (two theta)


4.0E+07

3.5E+07

3.0E+07

2.5E+07

2.0E+07

1.5E+07

1.0E+07

5.0E+06


-- anneal in nitrogen
-- anneal in air


- anneal in nitrogen
-- anneal in air


0.OE+00 L ----
SAngle ( t o theta)
Angle (two theta)


Figure 6-5. GaN (0002) peak comparison for (a) Ni 600A (b) Ni 150A before and after
annealing at 5000C for 10h in air









was deposited on pre-cleaned p-GaN. XRD analysis (Figure 6-6) revealed no peak shift

suggesting that 50A of Ni is insufficient to produce any strain in the p-GaN. In terms of

strain effect, the selection of thin Ni is needed to avoid energy band shift and/or the

formation of energy level defects. Since Ni50A/Au50A coatings annealed in oxidizing

ambient are the typical metal contacts used on p-GaN3, it is necessary to investigate their

influence on the lattice strain of GaN. For these experiments, Ni50A and Au 50A were


3.5E+07 -

3.0E+07

2.5E+07

2.0E+07 -

c 1.5E+07 no Ni
C: -- ni50
1.0E+07

5.0E+06 -

O.OE+00
O( NT O) ') LA (0 r,.- 0 o' L (O CO

Angle (two theta)


Figure 6-6. GaN (0002) peak comparison for Ni 50A


deposited sequentially on pre-cleaned p- GaN by e-beam evaporation. After deposition,

the sample was annealed at 500C in air for 10min. XRD analysis was performed on (1)

p-GaN before deposition, (2) as-deposited Ni50A/Au50A on p-GaN, and (3) after

annealing (Figure 6-7). Since the (0002) GaN peaks do not shift between these three

conditions, it can be concluded that Ni50A/Au50A contacts produce negligible strain in

GaN.

A Au film 150A thick was deposited on Ni 150A in vacuum by e-beam evaporation

to investigate the effect of Au in terms of strain release in GaN. The (0002) GaN XRD






63



peak (Figure 6-8) does not shift for the GaN coated with Nil50A/Aul50A compared with


that for the sample coated with 150A Ni only.


4.5E+07

4.0E+07

3.5E+07

3.0E+07

2.5E+07 .

c 2.0E+07
C
1.5E+07

1.0E+07

5.0E+06

O.OE-OO
O W C D CR O r P CR O U) (P U)
o M Mc co co o o ) Co Cn c c o )
Angle (two theta)



Figure 6-7. GaN (0002) peak comparison for Ni 50A/Au50A



4.0E+07

3.5E+07

3.0E+07

S2.5E+07

2.0E+07

1.5E+07

1.0E+07

5.0E+06

O.OE+00
c) C C') 'c C') cO Co C ( C O C')
Angle (two theta)


Figure 6-8. GaN (0002) peak comparison for Ni 150A/Aul50A


- nodepo
-noanneal
- anneal


-nil50
- N50/Aul50








Based upon this XRD analysis, Au does not have a measurable effect on the strain of

GaN for the sample ofNil50A/Au 150.


6.5 Characterization of Split (0002) GaN Peak

As noted in Figure 6-8, the (0002) peak of GaN usually appears to be split into two

peaks. This implies that the GaN has regions with different lattice parameters. GaN is

grown on a buffer layer (AIN) on sapphire and the GaN is also divided into two regions.

One is an undoped region located on the buffer layer and the other region is a Mg doped

region (-2.5pm) at the top of the GaN deposit. In order to understand which peak comes

from where in the GaN, the sample was covered with a thin piece of cellulose tape to

decrease the penetration depth of the X-rays (see schematic in Figure 6-9). The XRD

patterns from the untaped and taped samples (Figure 6-10) indicate that the intensity ratio

of the split peaks is different between the two samples. Specifically, the first and second

peaks were assigned and designated in Figure 6-10 for convenience. The relative


X-ray X-ray
X-ray




Mg-doped GaN Mg-doped GaN

Undoped GaN Undoped GaN


sapphire sapphire


(a) (b)

Figure 6-9. Schematic X-ray penetration depth profile of (a) without tape and (b) with
tape samples











First peak Second peak


4.OE+07

3.5E+07

3.OE+07

n 2.5E+07
S--w/tape
S2.0E+07
rc w/o tape
E 1.5E+07

1.0E+07

5.0E+06

O.OE+00
C' Ci t) 1O I? OP O9 0 P'N
V V V V V to U ) Ui
Angle (two theta)


Figure 6-10. GaN (0002) peak with different penetration depth



intensity ratio is defined as the intensity of the first peak divided by the intensity of the

second peak. As shown in Figure 6-10, the relative intensity ratio is low for the taped

sample compared with the untaped sample. This suggests that the higher angle peak

comes from the Mg doped GaN whereas the first peak comes from the undoped GaN

located below the Mg-doped region. Indeed, the relative intensity ratio 11/12 decreases as

the coating thickness is increased consistent with the conclusion that the second peak

comes from the upper Mg-doped region. Kirchner et.al reported that the GaN lattice

constants are slightly different when it has different carrier concentrations79 and the free

electrons expand the lattice constant of GaN by the following equation:80



A/V -nD
/V~ = B


V: volume


where









n: free electron concentration

D: deformation potential of conduction band minimum

B: bulk modulus


Figure 6-11 shows the dependence of relaxed lattice parameters a and c on the free

electron concentration.



C 1 -





F 1e*s











Figure 6-11. Dependence of GaN lattice parameters on free electron concentration
Figure 6-11i. Dependence of GaN lattice parameters on free electron concentration













CHAPTER 7
OPTICAL PROPERTIES OF NICKEL OXIDE PRODUCED BY THERMAL
OXIDATION OF NICKEL

Since the band gap of GaN is suitable for the emission of blue light, one of the most

important applications for GaN is blue light LD (Laser Diode) and LED (Light Emitting

Diode). One of the reasons for adopting Ni/Au metal contacts annealed in oxidizing

ambients is that the NiO produced during annealing has high transparency for blue light.

NiO is a wide band gap semiconductor, with the absorption edge in the UV region and no

absorption in the visible region. Although there should be no absorption in the visible

region for NiO based upon the band gap value, some visible light is absorbed when it

goes through NiO. It is known that Ni3+ induces absorption in the visible region.81

Therefore, a high concentration of Ni3+ ions in the NiO lattice decreases optical

transmittance. Since high transparency is essential for better performance of blue light

LEDs or LDs, the blue light transmittance of NiO annealed in different oxidizing

ambients should be an interesting factor in terms of GaN applications.

Since NiO was produced through thermal oxidation for LD or LED applications, it

is necessary to investigate the optical properties of NiO produced by thermal oxidation

under various conditions.


7.1 Experiments

7.1.1 Sample Preparation

Ni 400A was deposited on a polished quartz plate by e-beam evaporation. Polished

quartz plates were used as substrates because quartz plates are transparent to visible light.








Thus, 1/16" thick quartz plates were bought from Quartz Scientific Inc. Before the

deposition of Ni, the quartz plates were cleaned by acetone, methanol and DI water, and

dried with N2 gas. After deposition, samples were annealed at 5000C for different times

(from 15min to 24h), and at 600oC for different times (from 3min to 3h) in air,

respectively.


7.1.2 Analysis


7.1.2.1 XRD measurement

X-ray diffraction was performed using XRD Philips APD 3720, operated with Cu

Ka as the radiation source at 40kV and 20mA.


7.1.2.2 AFM measurement

The surface roughness was measured using Digital Instruments Dimension 3100

AFM and contact mode was applied.


7.1.2.3 Optical transmittance

Light of wavelength of 400-500nm was used to measure optical transmittance since

the wavelength of blue light is -450nm.The schematic diagram of the instrument used for

optical transmittance measurement is shown in Figure 7-1. The optical fiber was used

between the optical source and the sample plate to make a point source of white light.


7.2 Optical Transmittance of NiO Produced by Thermal Oxidation of Ni

The graphs in Figure 7-2 show transmitted intensity as a function of wavelength for

the samples annealed at 5000C for different times. The curve showing the highest

transmitted intensity is for the polished quartz plate without Ni. The curve showing the













Monochromator




Mirror Mirror

I I I
I 'I
t \ is t tss
I \ / \
\ \ t \ / / s
I I
i \ I
/

,,/ v ,, *
I\\ i
I /



*i *" *"-
/


i / 1 1
I VI


I I
I
I~
I,
III
III

I'


I
I Grating




Focusing lens






I :







Sample Ii
stage Xk*


Optical fiber


Photo
Mutliplier
Tube









Controls







Computer


Optical
source


Figure 7-1. Schematic diagram of the instrument for optical transmittance measurement





70


350000

300000 Polished quartz plate source
S-- Source
250000 15rrin
30nin
I 200000

150000 -- 2h
5h
Soo00 As deposited --
50000 -- 24h
0 -- As deposited

-50000
400 410 420 430 440 450 460 470 480 490 500
Wavelength(nm)


Figure 7-2. Optical transmittance of NiO oxidized at 5000C as a function of wavelength


lowest transmitted intensity (designated by arrow) is for the as- deposited sample without

annealing. It is evident that the as-deposited sample is almost opaque over the wavelength

range measured. The transmitted intensity of the polished quartz plate was selected as the

reference and the transmitted intensities for other samples were normalized by dividing

by the transmitted intensity through this plate (Figure 7-3). As shown in Figure 7-3, the

sample annealed for 15min shows the highest normalized intensity and the sample

annealed for 5h shows the lowest normalized intensity although all are in a narrow band.

The normalized intensity for blue light at 450nm (Figure 7-4) has a U-shape with a

minimum point at 62.2% for the sample annealed for 5h, compared with the high point at

69.5% for the sample annealed for 15min. For times longer than 5h, the normalized

intensity increases to 66.2% for the sample annealed for 24h. For the samples annealed at

600C the transmitted intensity varies with wavelength in a manner similar to the samples

annealed at 500C (Figure 7-5). Again, the normalized intensity and the intensity at

450nm are shown in Figure 7-6 and Figure 7-7, respectively. The significant trend for the










15 min



Z, A


(U


'F 0.65
C
a)
C
-o
a"
N 0.6

E
0


- 15nin

- 30mn

Sh

2h

- 5h

-210h

-24h


0 0 0 0 0 0
0 W O ( 0
o (v 0 (o to

V\Bvelength (nr


Figure 7-3. Normalized intensity of NiO oxidized at 500"C as a function of wavelength


* 15min






30min


* 1h


* 2h


*-24hL


* 10h


* 5h


Figure 7-4. Normalized intensity at 450nm for NiO oxidized at 500C


0 1 2 3 4 5 6 7 8






72


300000


250000 Polished quartz plate
S source
200000 As deposited
3nin
S150000 5rin
C -_ 20nin

100000 1h
As deposited 3h
50000 -


0 -
400 410 420 430 440 450 460 470 480 490 500
V\velength (nrr


Figure 7-5. Optical transmittance of NiO oxidized at 600C as a function of wavelength




0.8 -- 3min

0.75





.1 0.6 20mn

N 0.5lh
E 3 h 3h
o 0.5
z
0.45

0.4
0 C0 C) CO CO 0

V\veleth (nrr)


Figure 7-6. Normalized intensity of NiO oxidized at 6000C as a function of wavelength






73


0.73
0.72
3min
0.71
0.7
0.69
S* 5min lh
| 0.68
S0.67 20min
06
S0.66
S0.65
0.64

0.63 3h
0.62
0 1 2 3 4 5 6


Figure 7-7. Normalized intensity at 450nm of NiO oxidized at 6000C


samples annealed at 6000C is that the normalized intensity decreases as the anneal time

increases. Only the h sample deviated from this trend in that it had higher transmittance

compared with the sample annealed for 20min.

Many features such as film density, grain boundaries, surface roughness and crystal

defects can act as scattering centers, and should be considered in determining light

transmittance for thin films.



7.3 X-ray Diffraction Analysis

7.3.1 Grain Size Analysis

XRD analysis was conducted to investigate the grain size and crystal structures of NiO.

XRD peaks for NiO (111) and (200) annealed at 5000C and at 6000C are shown in Figure

7-8 and 7-9, respectively. The grain size was calculated by equation (5) based upon

measured XRD peaks.












2500

24h

2000



1500oo







500

15min

co CN (0 co I) N- C' IT (D cO cO CN 'T- (D OD
C C CO NM C CO nY C) m
Angle(tw o theta)


- 15nin
-- 30in
--1h
--2h
I--5h
--10h
-- 24h


S -- 15rin
-- 30min
---lh

-2h
S5h
1-10h
^4 1Nh1 j- 24h


o c9 'J CD U) 0 CV: 1P CD U 0 CJ (P (0 U
0 v ri c mc a co 0 -vo M IV va
Angle(tw o theta)


Figure 7-8. XRD peaks of NiO annealed at 500C; (a) (111) plane and (b) (200) plane


2500



2000



1500

U)
S1000



500












2500

3h
2000

3rrin
" 1500

i"- 20in
C:
S1000 h
3h


500

3min
0
(a CN 'T (D 00 1- (N (D 00 cO CN IT CPD (
(0 CD (0 CD 6- C0- 6- Cca- co co co
co o o o Mo M o Mo M M o o Moo
Angle(tw o theta)



(a)




2500


2000 A



S3mmminn
1500



A 1nl000ew o


500b

3min

(C Go M (0 (0 C) N N Cr (0 0 (N ( CD C0

Angle(tw o theta)

(b)


Figure 7-9. XRD peaks of NiO annealed at 600C; (a) (111) plane and (b) (200) plane









0.92 (5)
t=-
B cos O


where B is the line broadening measured at full width half maximum and has units of

radians, X is X-ray wavelength, 0 is the Bragg angle and t is the average grain diameter.

The calculated grain size from XRD peaks is shown in Figure 7-10. It is observed that the

grain size of the samples annealed at 6000C is larger than that of the samples annealed at

5000C. The grain size and normalized intensity at 450nm are summarized in Table 7-1.

The normalized intensities of the sample annealed for h at 5000C and at 6000C are

66.0% and 68.5%, respectively. The measured grain sizes for the corresponding samples

are 248A and 299A, respectively. According to grain size analysis, it is concluded that


330 -

310 6000C uh 3h/
\s 1h
290 5000C

270
S3nii n v 20nin
a 250 -1h
.N ----- 2h ^y
m + 15rin
.E 230
5h
210
S10h
190

170

150
0 2 4 6 8 10 12


Figure 7-10. Grain size comparison between 5000C and 6000C anneal










Table 7-1. Grain size and normalized intensity for NiO produced by annealing Ni in
an oxidizing ambient.


Anneal Annealing
Normalized
temperature time Grain size (A)
intensity
(C)

15min 236 69.5

30min 244 66.7

lh 248 66.0

500 2h 245 63.0

5h 222 62.2

10h 200 62.8

24h 241 66.2

3min 258 71.6

5min 254 68.5

600 20min 256 67.2

lh 299 68.5

3h 317 63.1



the transmittance of NiO with large grains is higher than that with small grains. Since

large grain size means relatively small grain boundary area and grain boundaries act as

scattering centers for light transmittance, the samples annealed at 6000C show higher

transmittance compared with those annealed at 5000C.










7.3.2 Effects of Preferred Orientation on Optical Transmittance of NiO

In an effort to determine if there was a preferred orientation/texture to the NiO film,

the integrated intensities of the 111 and the 200 peaks for NiO annealed at 5000C were

measured (Figure 7-11). The integrated intensity of 200 peak is always higher than that of

111 peak except for the sample annealed for 2h. The integrated intensity ratio, 1111/1200,

(Figure 7-12) for the sample annealed for 15min is 0.854. It increases with annealing time

and reaches 1.040 for the sample annealed for 2h. After 2h, the integrated intensity ratio


1900 -

1800

1700
S1600

S1500 .-- --111 peak

1400 /7 --- 200 peak

1300 -
1200

1100

1000
15min 30rnin 1h 2h 5h 10h 24h
Anneal time

Figure 7-11. Integrated intensity of NiO annealed at 5000C in air


decreases to 0.852 for the sample annealed for 24h. The integrated intensity ratio curve

displays a convex shape having a maximum value for the sample annealed for 2h. The

optical transmittance of NiO annealed at 5000C decreases as the annealing time increases

to 5h, and then increases as anneal time increases as shown in Figure 7-4. The optical

transmittance graph is concave with a minimum for the sample annealed for 5h. The

correlation between optical transmittance and integrated intensity ratio indicates that








1.1


1.05
1.04

1

i A 0.97
0.95
-o / 0.94
| ^0.93 0.95
0.9


0.85 -0.85 -
0.85
0.8
15min 30min 1h 2h 5h 10h 24h
Anneal time


Figure 7-12. Integrated intensity ratio of NiO annealed at 5000C in air


optical transmittance decreases as the integrated intensity ratio increases. It is suggested

that optical transmittance decreases as the number of grains with (111) orientation

increases compared with that of (200) orientation. This similar correlation can be seen for

the samples annealed at 6000C. The integrated intensity and the integrated intensity ratio

for NiO annealed at 6000C are shown in Figure 7-13 and Figure 7-14, respectively. In

Figure 7-14, the integrated intensity ratio increases with annealing time up to 1.174 for

the sample annealed for 20min and drops to 1.119 for the sample annealed for 1h. The

integrated intensity ratio for the sample annealed for 2h is 1.333, which is the highest

value among samples. The optical transmittance of NiO annealed at 6000C shows that it

decreases with annealing time as shown in Figure 7-7. Only one deviation from this

correlation is the sample annealed for 1 h. For the samples annealed at 6000C, the optical

transmittance decreases as the amount of grains with (111) orientation increases

compared with that of (200) orientation. It is known that the change of atomic density










2200



2000



S1800
Cu


1600



j 1400
C


1200



1000
3min 5min 20min 1h 3h
Anneal time



Figure 7-13. Integrated intensity of NiO annealed at 6000C in air


1.2







1



0.9
0.8
C 1.1







0.9



0.8


---111 peak
1- 200 peak i


1.33


5min


20min
Anneal time


Figure 7-14. Integrated intensity ratio of NiO annealed at 6000C in air








may cause the alteration of optical properties such as refractive index.8283 Ferreira et al.

reported that the sample with relatively large amounts of (11) preferred orientation in

NiO deposited by sputtering showed low optical transmittance.84 Thus, it can be

concluded that having relatively large amount of grains with (111) orientation can

decrease optical transmittance of NiO, which indicates that the optical transmittance of

NiO is dependent on crystallographic orientation.


7.3.3 X-ray Peak Position of NiO

The other interesting point in the XRD analysis is the peak position of NiO. The

(111) peak of NiO annealed at 5000C with different anneal time was shown in Figure 7-8

(a) and it is evident that the peak moved towards higher 20 values as anneal time

increases until anneal time reaches 1Oh. The peak position is 43.36 and 43.41 for the

sample annealed for 15min and the sample annealed for 1Oh, respectively. The peak

position of the sample annealed for 24h is 43.35o and is almost the same as that for the

sample annealed for 15min. The (200) peak of NiO annealed at 5000C is shown in Figure

7-8(b). The peak shift with annealing time for NiO (200) is similar to that for the (111)

peak. For the samples annealed at 6000C, the peak shifts of NiO (111) and (200) are

similar to the samples annealed at 500C as shown in Figure 7-9. It is evident that the

stress level of the NiO film changes with annealing time. However, in terms of

relationship with optical transmittance, the XRD peak shift with annealing time does not

show any systematic correlations.


7.4 Surface Roughness Analysis

Generally, one of the factors in determining optical transmittance is surface









roughness. In order to check if surface roughness plays a role in determining optical


transmittance of NiO, AFM (Atomic Force Microscope) was used to get surface

roughness of NiO annealed at 5000C. The contact mode was applied, and the average


roughness data (Ra) and root mean square roughness (Rq) were determined (Figure 7-15).

1.3
1.25
1.2

1.1
1.00


S0.9 0.95
0.94 0.96 0.92 0.93
0.8

0.7

0.6 0.59

0.5
no_anneal 15min 30min 1h 2h 5h 10h 24h
anneal time

(a)


1.7

1.64
1.6

1.5


1.4

1.33
1.3 1.23 1.24 1.22 1.28

1.2 1 1.23

1.1
noanneal 15rmin 30min 1h 2h 5h 10h 24h
anneal time


(b)

Figure 7-15. The surface roughness of NiO annealed at 500C (a) average, Ra (b) root
mean square, Rq.








Surface images for no anneal, and after 15min and lh anneals at 5000C are shown in

Figure 7-16 (a), (b) and (c), respectively. Ra is the most common index of surface

roughness while Rq is used in optical applications since it is more directly related to the

optical quality of a surface. It is evident that the surface of the sample annealed for 15min

is rougher than those of other samples. There is no significant difference in surface

roughness for samples other than the sample annealed for 15min. Therefore, it can be said

that surface roughness is not a critical factor in determining optical transmittance of NiO

films produced by annealing Ni 400A at 5000C in air at 450nm wavelength.












(a) (b) (c)

Figure 7-16. AFM surface image of (a) as-depostied Ni (b) NiO produced by annealing
Ni 400A at 500C in air for 15min. (c) NiO produced by annealing Ni 400A at
5000C in air for lh.













CHAPTER 8
SUMMARY

In order to investigate the effect of NiO on specific contact resistance, 50A of Ni

was deposited on cleaned p-GaN by e-beam evaporation. The deposited Ni was annealed

at 500oC for different times (0, 10s, 20s, 60s, 120s, 180s and 300s) in air. 1000A of Au

was deposited on top of the oxidized Ni by e-beam evaporation. CTLM patterns were

produced through a photolithography and dry etch process. I-V curves for all annealed

samples in air show rectifying I-V characteristics. It is clear that Au/pre-oxidized NiO/p-

GaN contacts are not ohmic in nature but rather exhibit Schottky barrier behavior. Based

upon AES depth analysis, it can be said that 50A Ni was partially oxidized during 10s

annealing and fully oxidized after 20s annealing. According to the sheet resistance of the

oxidized Ni layer, it is suggested that one of the reasons for the high currents in the

Au/10s pre-oxidized NiO/p-GaN sample is due to the low sheet resistance of the oxidized

Ni layer. The relatively low current in the sample annealed for 20s may be due to the high

sheet resistance of NiO. For Au/pre-oxidized NiO/p-GaN contacts under forward bias

conditions, which occurs when a positive voltage is applied to Au, the holes that flow

from p-NiO towards p-GaN meet an energy barrier at the interface between p-NiO and p-

GaN. Under reverse bias, there is a relatively low energy barrier between p-NiO and p-

GaN. This Schottky barrier between p-NiO and p-GaN is the reason for the non-linear I-

V characteristics for pre-oxidized samples.

There was no evidence for a new reaction phase between Ni and p-GaN after 24h

annealing at 500C based on XRD analysis. The interface between NiO and p-GaN was








observed using TEM and a thin, relatively bright layer was observed at the interface

between NiO and p-GaN. The high resolution TEM image and EDX data indicate that

this bright layer is not a reaction phase indicating that no reaction phase forms between

Ni and p-GaN after annealing at 5000C in an oxidizing ambient.

The pre-cleaned p-GaN samples were coated with different thicknesses of Ni

(150A and 600A) by e-beam evaporation in order to produce different stress levels. These

samples were then subjected to both I-V measurements and strain analysis by XRD in an

effort to determine the role of strain in the p-GaN on contact resistance. Based upon XRD

analysis and I-V measurements, it is concluded that the increase in the strain in GaN

induces a decrease in the current value. The effect of annealing in an oxidizing ambient

on p-GaN strain was investigated and, according to the XRD results for the 150A, 600A,

and 1500A Ni deposits, it appears that the strain in GaN for 600A and 1500A Ni samples

was partially released during annealing in an oxidizing ambient but the annealing for the

150A Ni sample did not result in measurable strain release in GaN. A similar shift of the

(0002) GaN peak was observed after annealing the 600A Ni sample in a nitrogen ambient

indicating that the strain release in GaN after annealing in an oxidizing ambient is not due

to the formation of NiO but due to the relaxation associated with annealing. According to

XRD analysis of the GaN sample coated with Ni50A/Au50A, it can be concluded that

Ni50A/Au50A contacts produce negligible strain in GaN.

Based upon the experimental results, (1) Au/pre-oxidized NiO/p-GaN contacts are

not ohmic in nature but rather exhibit Schottky barrier behavior, (2) no reaction phase

forms between Ni and p-GaN after annealing at 5000C in an oxidizing ambient and (3)

Ni50A/Au50A contacts do not cause significant strain in p-GaN and strain release in








GaN after annealing in an oxidizing ambient is not due to the formation of NiO but due to

the relaxation associated with annealing, Thus, it is concluded that the NiO formed

during annealing in an oxidizing ambient does not result in the low resistance ohmic

contact behavior observed for Ni/Au contacts to p-GaN. It can be concluded that the

current path for low resistance ohmic contacts of Ni/Au annealed in oxidizing ambients is

due to the direct contact between Au and the p-GaN. Ni appears to serve as a necessary

wetting layer that removes either the native oxide and/or the contamination layer on the

p-GaN surface during annealing. After removal of these layers, the direct contact of Au

on p-GaN is produced through a "layer inversion process". It is known that the solubility

of nitrogen in Au is negligible. It is also well-known that nitrogen vacancies in p-GaN act

as compensation centers for holes. Therefore, the formation of nitrogen vacancies in p-

GaN during annealing results in a decrease of hole concentration in p-GaN. Since Au/p-

GaN direct contacts do not produce significant nitrogen vacancy levels during annealing

due to negligible nitrogen solubility, the direct contact of Au/p-GaN can produce lower

contact resistance compared with the direct contact of Ni/p-GaN.

In order to investigate the transmittance of thermally-oxidized NiO, Ni 400A was

deposited on a polished quartz plate by e-beam evaporation. After deposition, samples

were annealed at 5000C for different times (from 15min to 24hours) and at 6000C for

different times (from 3min to hours) in air, respectively. According to grain size

analysis, it is concluded that the transmittance of NiO with large grains is higher than that

with small grains. This result can be explained by the fact that the large grain size means

relatively small grain boundary area and less scattering by the grain boundaries. Since,

the optical transmittance decreases as the amount of grains with (111) orientation





87


increases compared with that of (200) orientation, it can be concluded that the relatively

large amount of grains with (111) orientation can decrease optical transmittance of NiO

indicating that the optical transmittance of NiO is dependent on crystallographic

orientation.