Electrical contacts to P-type zinc telluride and gallium nitride


Material Information

Electrical contacts to P-type zinc telluride and gallium nitride
Physical Description:
Mixed Material
Trexler, Jeffrey Todd, 1970-
Publication Date:

Record Information

Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 28638736
oclc - 38827124
System ID:

Table of Contents
    Title Page
        Page i
        Page ii
        Page iii
    Table of Contents
        Page iv
        Page v
        Page vi
        Page vii
        Page viii
    Chapter 1. Introduction
        Page 1
        Page 2
        Page 3
        Page 4
        Page 5
    Chapter 2. Review of the literature
        Page 6
        Page 7
        Page 8
        Page 9
        Page 10
        Page 11
        Page 12
        Page 13
        Page 14
        Page 15
        Page 16
        Page 17
        Page 18
        Page 19
        Page 20
        Page 21
        Page 22
        Page 23
        Page 24
        Page 25
        Page 26
        Page 27
        Page 28
        Page 29
        Page 30
        Page 31
        Page 32
        Page 33
        Page 34
        Page 35
        Page 36
        Page 37
        Page 38
    Chapter 3. Experimental procedure
        Page 39
        Page 40
        Page 41
        Page 42
        Page 43
        Page 44
        Page 45
        Page 46
        Page 47
        Page 48
        Page 49
        Page 50
        Page 51
        Page 52
        Page 53
        Page 54
    Chapter 4. p-ZnTe contacts
        Page 55
        Page 56
        Page 57
        Page 58
        Page 59
        Page 60
        Page 61
        Page 62
        Page 63
        Page 64
        Page 65
        Page 66
        Page 67
        Page 68
        Page 69
        Page 70
        Page 71
        Page 72
        Page 73
        Page 74
        Page 75
        Page 76
        Page 77
        Page 78
        Page 79
    Chapter 5. p-GaN contacts
        Page 80
        Page 81
        Page 82
        Page 83
        Page 84
        Page 85
        Page 86
        Page 87
        Page 88
        Page 89
        Page 90
        Page 91
        Page 92
        Page 93
        Page 94
        Page 95
        Page 96
        Page 97
        Page 98
        Page 99
        Page 100
        Page 101
        Page 102
        Page 103
        Page 104
        Page 105
        Page 106
        Page 107
        Page 108
        Page 109
        Page 110
        Page 111
        Page 112
        Page 113
        Page 114
        Page 115
        Page 116
        Page 117
        Page 118
        Page 119
        Page 120
        Page 121
        Page 122
        Page 123
        Page 124
        Page 125
        Page 126
        Page 127
        Page 128
        Page 129
        Page 130
        Page 131
        Page 132
        Page 133
        Page 134
        Page 135
        Page 136
        Page 137
        Page 138
        Page 139
        Page 140
        Page 141
        Page 142
        Page 143
        Page 144
        Page 145
        Page 146
        Page 147
        Page 148
        Page 149
        Page 150
        Page 151
    Chapter 6. Conclusions
        Page 152
        Page 153
        Page 154
    Chapter 7. Future work
        Page 155
        Page 156
        Page 157
        Page 158
        Page 159
        Page 160
        Page 161
        Page 162
        Page 163
        Page 164
        Page 165
        Page 166
    Biographical sketch
        Page 167
        Page 168
        Page 169
        Page 170
Full Text








First I would like to thank my parents, Henry and

Andrea Trexler, for all their love and support over the

years. I am especially grateful that over the five years I

have been in graduate school, they never once asked the

question "When are you finally going to get out of school?"

I would also like to thank my graduate advisor Dr. Paul

Holloway. I don't think I have ever met a man who knows so

much about so many different things. His knowledge and

support have been influential towards my maturation as a

student and researcher.

I would also like to acknowledge all the people who

aided in the work presented in this dissertation. At the

University of Florida, L. Calhoun for the ZnTe samples, E.

Lambers for help with the AES analysis, M. Puga-Lambers for

the SIMS results, and a special thanks to my TEM pit crew.

I would also like to thank M.A. Khan from APA Optics and B.

Karlicek from Emcore for the GaN samples. Finally, a big

thanks goes out to M. Mier at WPAFB for all his help with

the temperature dependent I-V data.

I appreciate all the help, conversation, and general

good times over the past five years from all the members of

the Holloway group, past and present. It has been enjoyable


working with each and every one of them.

Last but not least, I thank all the friends I have made

during my time in Gainesville, Miller, Rock, Fahey,

Brunstein, Karen, Wish, JV, Brent, and everyone else I may

not have mentioned, for making life in Gainesville bearable.

I have many great memories of my time here that I will carry

with me always.



ACKNOWLEDGMENTS . . . . . . . . . . . ii

ABSTRACT . . . . . . . . . . . . vi


1 INTRODUCTION . . . . . . . . . . . 1

2 REVIEW OF THE LITERATURE . . . . . . . . 6
Schottky and Ohmic Contacts . . . . . . . 6
Carrier Transport . . . . . . . .. 11
Interfacial Reactions . . . . . . .. 15
II-VI Materials . . . . . . . . . .. 17
Growth and Doping . . . . . . . .. 17
ZnSe epitaxial layer growth . . . .. .17
ZnTe epitaxial layer growth . . . .. .18
Schottky and Ohmic Contacts . . . . .. .19
ZnSe . . . . . . . . .. 19
ZnTe . . . . . . . . .. 21
III-V Materials . *.... .. ................... 23
GaN Epitaxial Layer Growth and Doping . . .. .23
Substrates . . . . . . . .. 23
Buffer layers . . . . . . .. 25
Autodoping . . . . . . . .. 26
Contacts . . . . . . . . . .. 28
n-GaN . . . . . . . . .. 28
p-GaN . . . . . . . . .. 32

3 EXPERIMENTAL PROCEDURE . . . . . . . . .. 39
Introduction . . . . . . . . . .. 39
Deposition and Processing . . . . . . .. .39
p-ZnTe Contacts . . . . . . . .. 39
p-GaN Contacts . . . . . . . .. .41
Characterization . . . . . . . . .. 43
Electrical Characterization . . . . .. .43
Surface Composition Analysis . . . . .. .45
Surface Morphology . . . . . . .. .46
Interfacial Reaction Products . . . . .. .46

4 p-ZnTe CONTACTS . . . . . . . . . .. 55
Introduction . . . . . . . . . .. 55

Results . . . . . . . . . . .. 56
I-V Results . . . . . . . . .. 56
AES . . . . . . . . . . .. 57
SIMS . . . . . . . . . . .. 58
Surface Morphology . . . . . . . .. .58
Discussion . . . . . . . . . .. 59
Summary . . . . . . . . . . .. 65

5 p-GaN CONTACTS . . . . . . . . . .. 80
Introduction . . . . . . . . . .. 80
Results . . . . . . . . . . .. 81
Au . . . . . . . . . . .. 81
I-V results . . . . . . .. 81
AES . . . . . . . . .. 81
Pd/Au . . . . . . . . . .. 82
I-V results . . . . . . .. 82
AES . . . . . . . . .. 83
Ni/Au . . . . . . . . . .. 84
I-V results . . . . . . .. 84
AES . . . . . . . . .. 85
Ni/C/Au . . . . . . . . . .. 86
I-V results . . . . . . .. 86
AES and SIMS . . . . . . .. 87
Cr/Au . . . . . . . . . .. 87
I-V results . . . . . . .. 87
AES . ... ... ............. .... **. 88
XTEM and EDS . . . . . . .. 89
Temperature Dependent I-V . . . . . .. .90
Discussion ............. ......................... .91
Metal Contact Schemes . . . . . . .. .91
Au . . . . . . . . .. 92
Pd/Au . . . . . . . . .. 93
Ni/Au . . . . . . . . .. 94
Ni/C/Au . . . . . . . .. 97
Cr/Au . . . . . . . . .. 97
Temperature Dependent I-V . . . . .. 104
Summary . . . . . . . . . . .. 107

6 CONCLUSIONS . . . . . . . . . . .. 152
p-ZnTe Contacts . . . . . . . . .. 152
p-GaN Contacts . . . . . . . . .. 153

7 FUTURE WORK . . . . . . . . . . .. 155
p-ZnTe Contacts . . . . . . . . .. 155
p-GaN Contacts . . . . . . . . .. 155

REFERENCES . . . . . . . . . . . .. 157

BIOGRAPHICAL SKETCH . . . . . . . . . .. 167

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




December, 1997

Chairperson: Dr. Paul Holloway
Major Department: Materials Science and Engineering

The electrical and structural properties of sputter

deposited metal contacts to p-ZnTe or p-GaN have been

studied using current-voltage (I-V), Auger electron

spectroscopy (AES) and secondary ion mass spectrometry

(SIMS). Contacts of Au on p-ZnTe were heat treated over a

range of 150-350C for times up to 90 minutes. As-deposited

the contacts had a breakdown voltage of -0.5 V. Heating at

T!200C for 15 minutes resulted in ohmic behavior (linear I-

V curves) due to formation of near-surface acceptors from

doping by Au without interfacial phase formation. The

maximum current density was found to be 2.3 A/cm2 at 5 volts

following a 250C heat treatment for 15 minutes. Diffusion

of Au, which resulted is a highly resistive contact at

350C, was extensive and resulted in compound formation.


For p-GaN, Au contacts were sputter deposited while

Ni/Au, Ni/C/Au, Pd/Au and Cr/Au contacts were deposited by

electron beam evaporation. The contacts were heat treated

from 200-600C for times up to 30 minutes. Contacts of

Pd/Au and Cr/Au were also rapid thermally annealed (RTA) at

900"C for 15 seconds. Analytical data showed that Au did

not react with GaN up to 400"C, therefore a rectifying

contact was observed. A Ni layer between the Au and GaN led

to dissociation of the GaN at 600C. Increased electrical

transport by the Ni/Au contacts was attributed to increased

doping of the GaN near-surface region by interfacial carbon

contamination, where it was calculated that 2x10" cm-3 C

atoms were incorporated during annealing. Pd/Au contacts

were rectifying as-deposited. Increased conduction was

attributed to the formation of a Au:Pd solid solution below

400C and dissociation of the GaN following a 900C, 15

second RTA. Cr/Au contacts resulted in ohmic behavior with

p,:4.3xl0'1 Qcm2 after a 900C, 15 second RTA. The RTA caused

Cr to dissociate GaN, allowing Cr to increase the near-

surface doping density. It was postulated that Cr diffused

into the near-surface region using an interstitialcy

mechanism and formed substitutional acceptors on the Ga sub-

lattice. Temperature dependent I-V measurements showed that

charge transport from Ni/Au, Pd/Au and Cr/Au contacts have

components of both thermionic emission and field emission,


with Cr/Au having the largest contribution from field

emission transport.



Presently there is a strong desire to fabricate

optoelectronic devices which emit light in the blue region

of the visible spectrum. Light emitting diodes (LED's) and

laser diodes (LD's) emitting in this spectral region have

found applications in optical data storage and full color

displays. The present optical storage industry is dominated

by AlGaAs/GaAs based devices which operate in the near

infrared (IR) region. By replacing the infrared devices

with shorter wavelength blue emitting devices, the beam can

be focused to half the minimum spot size of the longer

wavelength devices, yielding the ability to write four times

as much data in the same surface area [And93]. The growing

interest in bright full-color displays has also magnified

the need for blue light emitting devices. Semiconductor

optical devices routinely operate from the IR to green

wavelengths. If this range could be extended into the blue

wavelengths, semiconductor components could then emit and

detect the three primary colors of the visible spectrum

which would have a major impact on imaging and graphics

applications [Str92]. This need for materials which emit in

the blue region with high brightness and efficiency has

stimulated significant research, and while the materials and

devices have been difficult to develop, great advances have

been made in recent years.

There are two groups of blue light emitting materials

that are viable for optoelectronic devices. These are the

materials systems based on ZnSe or GaN. ZnSe is a direct

band gap II-VI compound semiconductor with a zincblende

structure, a 5.688 A lattice parameter [Wea76], a band gap

of 2.67 eV [Shi8O] and an electron affinity of 4.09 eV

[Swa66] at room temperature. ZnSe, doped n-type, has been

grown using Cl doping during molecular beam epitaxial (MBE)

growth [Ohk87], and p-type using a N2 radio frequency (RF)

discharge during MBE growth [Par90]. This ability to

produce both n- and p-type doping has led to the fabrication

of znSe based light emitting diodes (LEDs) [Qui93] and laser

diodes (LDs) [Haa91] which emit in the blue to green

wavelengths of 490-516 nm. Presently, the longest lifetime

devices for laser output consist of a ZnCdSe/ZnSSe/ZnMgSSe

single quantum well separate-confinement heterostructure

laser diode with a wavelength of 514.7 nm, which operated

more than 100 hours at room temperature under continuous

wave (CW) operation [Tan96].

The other material used for blue light emitting

devices, GaN, is a direct band gap III-V compound

semiconductor commonly grown in the wurtzite crystal

structure with lattice parameters of a=3.189 A and c=5.185

A [Maru69], a band gap of 3.4 eV [Str92] and an electron

affinity measured to be between 3.3 eV [Nem96] and 4.1 eV

[Pan74] at room temperature. GaN has been grown by

metalorganic chemical vapor deposition (MOCVD) and doped n-

type with Si donors [Nak92a], and p-type using Mg acceptors

[Nak92b]. The ability to grow both n- and p-GaN led to the

fabrication of p-GaN/n-InGaN/n-GaN double heterostructure

blue LEDs with a peak wavelength of 440 nm and InGaN multi-

quantum well (MQW) laser diodes capable of continuous wave

operation at room temperature with a peak wavelength of

400.23 nm [Nak97].

Currently, GaN based LEDs are the only commercially

available blue light emitting devices. This fact not

withstanding, there are still many areas which require

further advancement before high quality, long life

commercial devices can be manufactured and marketed. One of

the critical problems facing these laser diodes is the short

lifetime under either pulsed or CW operation. Poor contact

performance, particularly in the p-type contact, is a

contributing factor to poor device lifetimes.
An ohmic contact is a non-rectifying

metal/semiconductor contact which exhibits linear current

vs. voltage (I-V) characteristics and has a negligible

resistance in comparison to the bulk resistance of the

semiconductor. Ohmic contacts allow power to be efficiently

applied to the active region of the device as opposed to

being consumed in the contacts.

This work will focus on improving the current contact

schemes to p-ZnSe and p-GaN, and to give an understanding of

the processes responsible for the formation of long-lived

robust ohmic contacts to these materials. Background

information on ohmic contacts and a discussion of the

metallization schemes used and their effects of the

interfacial reactions between metal and semiconductor will

be presented.

A review of the literature is presented in chapter two.

Metal/semiconductor contact theory and processes utilized in

the formation of ohmic contacts are discussed. Then the

literature on growth and doping is presented, focussing

primarily on the formation of p-type material. This is

followed by the various contact schemes that have been

investigated for these materials. This information will

first be presented for the II-VI materials ZnSe and ZnTe and

then for the III-V material GaN.

Chapter three presents the experimental procedure

followed in this study. The deposition and processing

parameters are first presented. This is followed by a

review of the characterization techniques used to study the

electrical properties, surface composition, surface

morphology, structural properties and interfacial reaction


The results and discussion of the contact studies are

presented in chapters four and five. Chapter four presents

the data on Au/p-ZnTe contacts. Heat treatments are

described that lowered the breakdown voltage in these

contacts and led to the formation of ohmic contacts. The

composition and structure of the contacts were analyzed to

identify changes induced by the heat treatments, which led

to the ohmic behavior and the eventual breakdown of the

ohmic contacts. In chapter five, the data for Au, Pd/Au,

Ni/Au, Ni/C/Au and Cr/Au on p-GaN are presented with

emphasis on the interfacial reactions upon heat treatment

and conduction mechanisms in these contacts. Conclusions

from these studies are summarized in chapter six and

suggested future research is discussed in chapter seven.


Schottkv and Ohmic Contacts

When a metal and semiconductor are brought into

electrical contact a potential barrier arises from the

separation of charges at the metal-semiconductor interface

such that a high resistance region devoid of mobile carriers

is created in the semiconductor [Sha84]. Two types of

contacts arise from this orientation: ohmic contacts, which

are defined as a metal-semiconductor contact that has a

negligible contact resistance relative to the bulk or

spreading resistance of the semiconductor [Sze8l], and

Schottky contacts which rectify current flow across the

metal-semiconductor interface. Most metal/semiconductor

contacts are Schottky in nature.

Figure 2.1A shows metal and p-type semiconductor energy

bands before contact. In this figure, 0. and 0. are the

metal and semiconductor work functions, defined as the

amount of energy required to raise an electron from the

Fermi level to the vacuum level; X. is the semiconductor

electron affinity, the energy difference between the vacuum

level and the lower edge of the conduction band; and Eg is


the semiconductor bandgap. In Figure 2.1B the metal and p-

type semiconductor have been brought into electrical contact

in thermal equilibrium. When 0,
metal into the semiconductor until the Fermi level on the

two sides are aligned. After reaching the semiconductor,

the electrons recombine with the majority carrier holes

giving rise to a space charge layer of ionized acceptors.

The concentration of holes in the space charge region is

negligibly small compared to the acceptor concentration. It

follows, that on the semiconductor side of the contact the

space charge region consists of a depletion layer whose

thickness W depends on the concentration of ionized acceptor

atoms. In this configuration the obstacle to current

conduction in contact/p-type semiconductor systems is the

energy barrier at the valence band (OB) which can be defined


OB=X.+Eg-0M. (2.1)

From this equation and Figure 2.1B, it can be seen that for

non-degenerately doped semiconductors, where X,+Egs, a non-

rectifying contact should be observed for a metal work

function greater than the semiconductor work function.

This dependence of 0, on 0m is known as classical

Schottky theory and is only observed in predominantly ionic

semiconductors. For covalently bonded semiconductors

Bardeen [Bar47] pointed out the importance of localized

surface states in determining the barrier height. Dangling

bonds formed due to the covalent nature at the surface give

rise to localized energy states at the semiconductor surface

with energy levels lying in the forbidden gap. These

surface states may be continuously distributed or localized

in the band gap and modify the charge in the depletion

region affecting the barrier height. These states are

characterized by a neutral level 00. The effect of these

surface states on the energy levels in a metal/p-type

semiconductor can be seen in Figure 2.2. When equilibrium

is reached, holes from the semiconductor adjacent to the

surface occupy states below 0o and the Fermi level at the

surface aligns with that in the bulk. The surface then

becomes positively charged and a depletion layer consisting

of ionized acceptors is created in the semiconductor region

near the surface. Because of this dipole formation, a

potential barrier looking from the surface towards the

semiconductor is obtained even in the absence of a metal

contact. When a metal is now brought into contact with the

semiconductor and equilibrium is reached, the Fermi level in

the semiconductor must change by an amount equal to the

contact potential by exchanging charge with the metal. If

the density of surface states at the semiconductor surface

is very large then the charge exchange takes place largely

between the metal and the surface states, and the space

charge in the semiconductor remains almost unaffected. As a

result the barrier height becomes independent of the metal

work function and is given by:

IB=Eg-0o. (2.2)

This barrier height is said to be "pinned" by the surface

states. With Fermi level pinning playing a large role in

the choice of contact materials, the ability to predict and

determine whether this phenomenon will occur is of utmost


In a model applicable to all semiconductors [Bar47],

the Schottky barrier height, 0., can be expressed as:

0>B=S*(s)x'+0o(s), (2.3)

where m and s refer to the metal and semiconductor,

respectively, and 0)(s) represents the contribution of the

surface states. The interface index S*(s) expressed as:

S*(s)=d0o/dx,, (2.4)

gives the dependence of barrier height on the metal

electronegativity, and x, is the metal electron affinity

which is analogous to the metal work function. For

covalently bonded materials, 0o is large and nearly constant

therefore S* is nearly zero and 0, is nearly independent of

Xy. Ionic semiconductors, on the other hand, have a large

S* and 0B is expected to increase linearly with X.. Kurtin

et al. [Kurt69] have suggested that S* is a function of the

electronegativity difference between cation and anion, (AX),

of compound semiconductors which is a measure of the

ionicity of the material, and they have plotted these data

as shown in Figure 2.3. The regime where S=1 is for ionic

bonding, while the tail near S=0 is for covalent bonding.

This plot can be used for a rough estimate as to whether

Fermi level pinning will occur; it is not an exact

determination of bonding character. One example is the case

of Sio2 which is strongly covalently bonded but has a AX


There are also two other types of interfaces prevalent

in metal/semiconductor contacts. These are reactive

interfaces and samples with an insulating, non-reactive

interfacial layer [Sha84]. The majority of work on reactive

interfaces has dealt with silicides e.g. (Ott80, Fre8O,

Bri78], and will not be illustrated here. In most metal-

semiconductor contacts before metal deposition, the

semiconductor surface is prepared by chemical cleaning and a

thin insulating oxide layer is invariably left on the

surface of the semiconductor. When the interfacial layer is

thin enough (i.e.<20A), the potential drop across it is

negligibly small compared to that in the semiconductor

depletion region. Such a thin layer is transparent to the

electrons as the electrons can tunnel through in either

direction. Because of this, the barrier height 0. and the

contact potential difference Vi may be unaffected by the

presence of a thin interfacial layer [Sha84]. Thicker

interfacial layers however, may increase 0.

One final effect on the measured barrier height is a

result of the electric field in the depletion region which

leads to image force barrier lowering. This effect occurs

whether or not an interfacial oxide is present, and the

magnitude of barrier lowering, A
ADB= [ [q3Nd/8n2ed2e,] (Vi-V) ]11/4, (2.5)

where Nd is the carrier concentration, ed is the dielectric

constant in the depletion region, e, is the dielectric

constant in the bulk semiconductor (ed=e.) and Vi is the

built-in potential of the barrier which is related to the

position of the Fermi level. Since image force lowering of

the barrier results from the field produced by an electron,

the measured barrier height is not lowered by those methods

which do not require movement of the electron over the

barrier, (e.g. capacitance-voltage method).

Carrier TransDort

Current flow in a metal-semiconductor junction

occurs from charge transport between the metal and

semiconductor. There are four mechanisms by which this can

occur in Schottky contacts: a) thermionic emission over the

barrier; b) tunneling through the barrier; c) carrier

recombination (or generation) in the depletion region; and

d) carrier recombination in the neutral region of the

semiconductor which is equivalent to minority carrier

injection (Figure 2.4 [Sha84]). In general semiconductor

current transport described by classic Schottky theory is

determined by the carrier concentration in the

semiconductor. At moderate temperatures (300K), mechanisms

c and d dominate for very low carrier concentrations (~10"'4

cm-3) and are rarely used to form low resistance ohmic

contacts. For low carrier concentrations (NDI017 cm'3),

conduction across the metal-semiconductor interface is

dominated by thermionic emission (TE) over the potential

barrier, which is the case for Schottky barrier junctions.

For higher carrier concentrations (ND7l109 cm-3), the barrier

width becomes narrow and conduction may take place by

tunneling through the thinned barrier, which is also known

as field emission (FE) transport [Sze85]. Ohmic conduction

is related to a large tunneling component. There also

exists a hybrid case known as thermionic field emission

(TFE), which is actually the mechanism (b) shown in Figure

2.4, in which there is enough kinetic energy for the carrier

to be excited from the ground state to a level at which the

potential barrier is thin enough for tunneling to occur.

The conditions under which each conduction mechanism is

expected to control current flow are fundamental to

understanding formation of ohmic contacts.

Models to determine the dominant conduction mechanism

have been devised for thermionic emission and field

emission. These models allow evaluation of conduction

mechanism from the current-voltage characteristics of the

metal/semiconductor contact. A model for thermionic

emission over a barrier was presented by Wagner [Wag31] and

Schottky and Spenke [Sch39]. In reverse bias the current is

given by:

I=I0[exp(-qV,/kT)-l]1, (2.6)


I=SA*T2exp [(AB-OB)/kT], (2.7)

and A* known as Richardson's constant is given by:

A*=4nm*qk2/h3. (2.8)

In these equations 10 is the saturation current, q is the

electron charge, VR is the reverse bias, k is Boltzmann's

constant, S is the diode area, AB-OB is the effective

barrier height with A0. determined from Equation 2.5 for

image force reduction, m* is the majority charge carrier

effective mass, and h is Planck's constant.

Tunneling can occur in a Schottky barrier junction in

either forward or reverse bias. The probability of

tunneling from the semiconductor to the metal increases as

the doping density of the semiconductor increases or as the

potential barrier decreases, e.g. through proper choice of

the contact metal for an unpinned Fermi level. As mentioned

before, when a metal and semiconductor are brought into

electrical contact, a bending of the semiconductor bands

occurs due to formation of a region depleted of free

carriers whose thickness W (depletion width) is given by


W=[ (2e./qND)tVi-VI]%, (2.9)

where N0 is the doping density, Vi is the built-in barrier

as discussed earlier, and V is the applied bias. This

depletion width is shown in Equation 2.9 to be inversely

proportional to the square root of the carrier density in

the semiconductor. If the tunneling component dominates the

current flow, the tunneling current can be given by the


Jt-exp(-qO,/Eoo00), (2.10)

where E00 is a tunneling parameter proportional to (ND)"

which will be described in detail later in this section.

Equation 2.10 thus indicates that the tunneling current will

increase exponentially with (ND)", and from Equation 2.9 the

tunneling probability is proportional to exp(O./W).

At higher temperature a significant number of carriers

are able to rise to an energy where there is a thinner,

lower barrier and tunneling can occur above the ground state

(i.e. bottom of conduction band for electrons) but below the

top of the barrier. Since the number of electrons decrease

rapidly with energy above the Fermi level due to electron

occupation probability, there exists an energy E. (measured

relative to the bottom of the conduction band) at which the

contribution of TFE becomes maximum. For a semiconductor in

which the Fermi level is not pinned, the magnitude of this

potential barrier can be reduced for p-type materials by

using a metal with a large work function, which will align

the Fermi level as closely as possible to the semiconductor

valence level.

Tunneling through a Schottky barrier has been analyzed

theoretically by Padovani and Stratton [Pad66] and Crowell

and Rideout [Cro69] with the following relationships being


I=I,exp(-qVR/Eo), (2.11)


E0=Eoo00coth(Eoo/kT) (2.12)


Eoo=(qh/4n) (Nd/m*e,), (2.13)

where the symbols have their usual meanings and the pre-

exponential factor I, is only weakly dependent on voltage

and is a complicated function of barrier height, properties

of the semiconductor, and the temperature. The energy E00

is an important parameter in tunneling and kT/Eoo is a

measure of the relative importance of TE versus FE [Pad71].

At low temperatures E00 may become large compared to kT,

then Eo=Eoo and field emission dominates. At high

temperatures where E004ckT, then E0=kT and thermionic emission

dominates. For intermediate values of temperature, Eoo=kT

and thermionic field emission transport dominates current


Interfacial Reactions

Interfacial reactions may also play a large role in the

formation of ohmic contacts. A review by Kim et al.

[KimTJ97] demonstrated the role of interracial reactions for

ohmic contacts to wide bandgap semiconductors. Heat

treatments are often used to form tunneling contacts by

alloying between the contact metal and the semiconductor

[Hal50]. The simplistic idea is to choose a metal or metals

(elements) which create a highly doped semiconductor region

and leads to ohmic behavior through tunneling transports.

In other cases, heat treatments result in interfacial

compounds, such as metal/silicide/silicon contacts. In

these contacts, the metal/semiconductor interface is

eliminated and replaced by two new interfaces, a

metal/compound and a compound/semiconductor interface.

The most widespread use of interfacial reactions to

form ohmic contacts has been in GaAs. The consequences of

these interfacial reactions vary, but dissociation of the

compound semiconductor is a necessary, but not sufficient

condition for formation of good contacts. In those cases

where decomposition results in an ohmic contact, the surface

doping concentration in generally increased [Hol97]. The

proposed mechanism for this incorporation of dopant is an

epitaxial regrowth of the GaAs with the regrown layer having

an increased dopant concentration. Sands et al. [San88]

documented epitaxial regrowth of GaAs at the reaction

interface for the specific case of Si/Ni/GaAs. Through a

series of heat treatments, (250C for 30 minutes followed by

350C for 30 minutes) they detected a regrown GaAs region

using XTEM. Holloway et al. [Hol91] showed that interface

segregation of Si dopants may also occur in the region of

decomposed GaAs. This was the first direct proof that

segregation of dopant from the semiconductor (as opposed to

lattice incorporation from the metallization) could lead to

ohmic contact formation.

II-VI Materials

Growth and DoDing

ZnSe eDitaxial layer growth

II-VI compound semiconductors such as ZnSe (Eg=2.67 eV

at RT) have long been promising materials for the

fabrication of efficient blue light emitting diodes [Qiu91].

One of the objectives of this work is to understand the

factors in the formation of ohmic contacts to p-doped II-VI

wide bandgap semiconductors, so growth and doping of these

p-type materials will be discussed in this section.

Initial efforts to incorporate a substitutional

acceptor impurity into ZnSe during crystal growth focused on

lithium doping during molecular beam epitaxial (MBE) growth

[Dep89]. Lithium was shown to have a maximum net acceptor

density of lxl0" cm-'3, above which strong compensation

occurred resulting in highly resistive ZnSe. In addition,

Li is mobile in ZnSe and doping was non-reproducible.

Oxygen has also been reported as a non-reproducible dopant

in ZnSe layers grown by MBE, [Aki89] with the largest

reported net acceptor density of 1.2x1016 cm-3. In 1990,

Park et al. [Par90] introduced a novel doping technique

using nitrogen atom beam doping from an RF discharge source

during MBE growth. Repeatable net acceptor concentrations

as large as 3xl017 cm-3 were reported using this technique.

More recently a maximum value of NA-ND=1.8xl0'8 cm-'3 was

reported by Ohtsuka et al. using an electron cyclotron

resonance (ECR) plasma source [Oht93].

ZnTe eDitaxial layer growth

The wide-bandgap semiconductor ZnTe (E,=2.3 eV) is a

II-VI material which has been used to form ohmic contacts to

p-ZnSe [Hie93]. Although bulk crystal ZnTe is naturally

doped p-type by an intrinsic defect, considerable difficulty

has been experienced in achieving reproducible, high quality

p-type ZnTe using MBE. Such difficulty has generally been

associated with the physical incorporation of the dopant

species due to a low sticking coefficient [Han93]. For

example, Kitagawa et al. [Kit81] investigated antimony

doping at flux levels greater than those of the host

elements, with the F,,=2Fzn. Doping levels of 1x10"' cm-'3 were

reported for ZnTe homoepitaxially grown on ZnTe substrates.

Hishida et al. [His89] used elemental phosphorus to obtain

p-doping of ZnTe grown by MBE on GaAs substrates. They

found that the best films were grown with a Te/Zn beam

pressure ratio (BPR) of 4. A maximum doping of 4x10" cm-3

was obtained with phosphorus beam equivalent pressure

comparable to that of Te. However, the crystalline quality

deteriorated with this doping, as shown by a broad x-ray

rocking curve. Han et al. [Han93] succeeded in obtaining

doping levels exceeding 10"' cm-'3 in ZnTe by employing the

same types of nitrogen sources used to dope ZnSe. The

crystalline quality of ZnTe was maintained.

Schottkv and Ohmic Contacts


Many successful ohmic contact schemes to n-ZnSe have

been reported and have been reviewed by Fijol and Holloway.

[Fij96a] An important factor for ohmic contacts is Fermi

level pinning, and experimental data has been collected

indicating that Fermi level pinning does not occur with ZnSe

[Che94, Xu88, Hol97]. Thus ohmic contacts should be

predicted by the metal work function as long as interfacial

contamination and oxide layers do not interfere with charge

transport between the metal and ZnSe. However, with the

electron affinity of ZnSe equal to 4.09 eV, and an Eg=2.67

eV, (Eg+x=6.76 eV), finding a metal with a large enough work

function is impossible. Still, many metallization schemes

have been studied for ohmic contacts to p-ZnSe. The most

basic schemes consisted of non-graded single metal contacts.

Fijol et al. [Fij95] investigated the effects of heat

treatments on Au and Ag contacts to p-ZnSe and concluded

that both metals formed rectifying contacts with minimum

breakdown voltages of 3.0 and 2.3 eV, respectively. The

lower breakdown voltage for Ag was attributed to oxygen

doping of the near-surface region. Pseudo-ohmic behavior

for Au contacts was reported by Akimoto et al. [Aki91] by

using an oxygen plasma to grow an interfacial oxide before

metal deposition. They proposed that 0 doping occurred

leading to the pseudo-ohmic behavior. Chen et al. [Che94]

reported a decrease in the breakdown voltage of about 0.25

eV upon deposition of 2-3 monolayers of Se at the interface

between the p-ZnSe and Au. All of these contacts were

rectifying in nature.

A second contact scheme involved the semimetal HgSe.

Lansari et al. [Lan92] used an in-situ MBE epitaxial layer

of HgSe to reduce the semiconductor interfacial energy

barrier to about 0.6 eV, and Fijol et al. [Fij96b] used an

ex-situ process of reacting Hg with an in-situ Se capping

layer to form HgSe. The potential barrier of this contact

to p-ZnSe was measured to be -0.55 eV, and it remained


But the most successful efforts at ohmic contacts to p-

ZnSe have incorporated ZnTe into the contact scheme. Fan et

al. [Fan93] used a pseudograded Zn(Te,Se) structure

sandwiched between a p-ZnTe top layer and a p-ZnSe bottom

layer. The pseudograded contact region consists of 17

cells, each of 20A thickness. In each cell both the

thickness of the ZnTe and ZnSe layers were varied. The

first cell next to the p-ZnSe layer contained 18A p-ZnSe

and 2A of p-ZnTe, the next cell 17A p-ZnSe and 3A p-ZnTe,

and so on. It was proposed that this pseudograded band gap

region allowed injection of holes from the heavily doped

ZnTe into ZnSe. A specific contact resistance (pc) of 2x10-3

Qcm2 was measured for these contacts with NA=9.5xl017 cm-3.

Hiei et al. [Hie93] employed p*-ZnTe/ZnSe quantum wells

whose sub-bands were aligned in energy to allow resonant

tunneling of holes through the multi-quantum well region.

The ZnSe barriers were 20A thick and the ZnTe wells varied

from 3 to 17A in thickness. This scheme also included a p-

ZnTe capping layer. They recorded a pc as low as 5x10'2 Qcm2

for ZnSe with a hole concentration of 7x1016 cm"3. Fijol and

Holloway studied the thermal stability of this contact

scheme and found them to be thermally unstable. They

deteriorated badly under modest heating, or with high

current densities (:1000 A/cm2) [Fij96].


When graded or MQW contacts of ZnTe/ZnSe are used for

ohmic contacts to p-ZnSe, an ohmic contact is required to p-

ZnTe. For this contact both single and multi-layer metal

contacts have been studied. Ozawa et al. [Oza93]

investigated Au, Au/Pt, Au/Pd, and Au/Pt/Pd contacts to p-

ZnTe and reported the lowest specific contact resistance of

Pc=4.8xl0-'6 Qcm2 for Au/Pd annealed at 200C. They proposed

that a reaction took place between Pd and ZnTe and assumed

that this reaction was more likely to take place at elevated

temperatures. The exact role of the Pd layer in the

formation of the ohmic contacts was not determined. The

contacts were not thermally stable, with an increase in

specific contact resistance occurring after a 250C, 3

minute anneal for all contacts, (e.g. an increase to

pc=2xl0'4 Qcm2 for Au/Pd, after 200C, 3 minutes). Mochizuki

et al. [Moc94] investigated Au/Pt/Ti/Ni and proposed that

the reaction between Ni and ZnTe lowered the specific

contact resistance (pc=7xl0-6 Qcm2 as-deposited). Insertion

of the Ti layer was reported to improve the thermal

stability up to 350C. In later work, this group detected

formation of NiTe2 from the reaction between Ni and ZnTe

upon annealing at 300"C for 3 minutes, which apparently

played an important role in the lowering of the specific

contact resistance of Au/Pt/Ni/Ti [Moc95]. Finally, Ohtsuka

[Oht95] investigated Au/p-ZnTe contacts using an oxygen

plasma cleaning and HCl treatment before Au deposition. A

Te-rich layer was formed on the surface. Specific contact

resistances as low as 5.8xl0"6 Qcm2 were reported on p-ZnTe

(NA=2xl08 cm'3). There was no mention of any reaction of Au

with the underlying ZnTe.

III-V Materials

GaN EDitaxial Layer Growth and DoDinp

As previously mentioned, a second objective of this

research was to understand the formation of ohmic contacts

to p-GaN. The III-V nitrides, in particular GaN, are also

promising materials for semiconductor device applications in

the blue and ultra-violet (UV) wavelength regions [Str93].

InN, GaN, and AlN with direct band gaps of 1.9 eV, 3.4 eV,

and 6.2 eV, respectively [Str93], can be grown in the

wurtzite crystal structure to form a continuous alloy system

with the ability to grade the wavelength of the emitted

light over the entire visible spectrum.

However, two intrinsic material problems have inhibited

the growth of high quality p-GaN. First, no lattice matched

substrate for GaN has been found so all films have had a

large defect density (_11012 cm'2) Second, GaN

intrinsically dopes n-type. The methods and treatments

discovered to overcome these difficulties and resulting

production of high quality p-GaN will be discussed below.


Despite the large lattice mismatch (-14%) and the

difference in thermal expansion coefficients, the majority

of GaN growth studies have used the basal plane (0001) of

sapphire (A1203) [Str92]. Other substrates investigated

with limited success have included SiC [Kim94 and Bot95],

ZnO [Sit90], and MgO [Pow90], all of which are zinc-blende

crystal structure. More recently Kuramata et al. [Kura95]

and George et al. [Geo96] have grown GaN on the (111) and

(100) planes of magnesium aluminate (MgAl204) which has a

spinel crystal structure and a lattice mismatch of -10% with

GaN. One of the advantages of this substrate and SiC is the

ability to cleave (not possible with sapphire) which

provides the possibility of facets to be used as a cavity

mirror in a laser diode.

Lithium aluminate (LiA0l2) and lithium gallate (LiGaO2)

single crystals are also attracting attention as substrates

for epitaxial growth of GaN [Nic96]. The small mismatches

with GaN (-1.5% for LiAlO2 and 0.2% for LiGaO2) should lead

to lower defect concentrations [Wil96]. However, the

crystal quality of MBE grown GaN films on LiGa02, as

determined by photoluminescence (PL) measurements, was

similar to films grown on sapphire. This was attributed to

the presence of strain due to the difference in thermal

expansions coefficients between layer and substrate [And97].

More recent work has occurred on GaN layers grown

homoepitaxially on bulk single crystal GaN using MBE [Gas96]

and metalorganic chemical vapor deposition (MOCVD) [Pon96].

While homoepitaxial approaches are promising, the quality of

epilayer GaN has not improved dramatically (p,~107-108 cm-2).

Buffer layers

Sapphire is still the primary substrate for GaN growth,

therefore all research groups have used low temperature

buffer layers between the substrate and GaN film. The

purpose of these buffer layers is to reduce the detrimental

effects of the lattice mismatch by trapping defects in the

layer and not allowing them to propagate to the deposited

film. Amano et al. [Ama90] used a 500A thick AlN buffer

layer grown at 600C and Khan et al. [Kha92] used a 250A

thick AIN layer grown at 550C. Nakamura et al. [Nak94]

have incorporated a 250A GaN buffer layer grown at 500C.

In all these cases, the buffer layers were grown by MOCVD.

More recently, Li et al. [Li95] have reported that a double

buffer layer with GaN and AlN as the first and second layers

leads to mirror-like films across the entire substrate.

Kuznia et al. [Kuz93] investigated the crystalline quality

of both GaN and AlN buffer layers grown at 500C. They

determined, using low-energy electron diffraction (LEED) and

cross-sectional transmission electron microscopy (XTEM),

that the AlN buffer layer was amorphous upon deposition and

become single crystal following deposition of the GaN

epitaxial film at 1000C. Under similar conditions, the GaN

buffer layer was also amorphous before GaN epilayer growth,

and then appeared to convert mostly to single crystal form

with some buried polycrystalline or amorphous regions.


As pointed out earlier, unintentionally doped GaN has

in all cases been observed to be n-type [Str92]. There is

no consensus on the cause of this autodoping, but many

believe nitrogen vacancies are the intrinsic defect

responsible for n-type conductivity [Pan73a]. By proper

growth, the electron concentrations of GaN doped with Si

were in the range of 1017-2x10" cm"3 [Nak92a].

Many potential p-dopants have been investigated,

including Zn and Mg [Maru69] and Be and Li [Pan73b], with

all showing highly compensated material in the absence of

post-deposition treatments. In addition to compensation of

donors resulting in semi-insulating layers, low activation

of the acceptor species was also observed. One of the

causes of this low activation is the relatively deep level

of acceptors in GaN. For example, the ionization level of

Mg in GaN is -150-165 meV [Aka9l]. This limits the

electrically active acceptors at room temperature to less

than 1% of substitutional Mg. This leads to the discrepancy

between dopant concentration and free hole concentration.

Currently using Mg, in order to obtain 1x1017 cm3 free holes,

there would need to be IxlO10 cm"3 Mg atoms in the GaN. At

this time, there is no shallow acceptor known for GaN to

overcome this problem.

Along with the relatively deep acceptor levels in p-

GaN, low activation was also attributed to introduction of

hydrogen during the growth of GaN, leading to an in-situ

hydrogenation process which caused the formation of

acceptor-H neutral complexes and compensation [Nak92b].

This conclusion was supported by Brandt et al. [Bran94] who

investigated intentional hydrogenation of acceptors and

found a reduction in hole concentration due to the formation

of H-acceptor complexes. Pearton et al. [Pea96a] studied

the role of atomic hydrogen introduced during processing in

passivating the electrical activity of Mg and C acceptors

using secondary ion mass spectrometry (SIMS). They

determined that H had a diffusivity of >10-11 cm2s-1 at 170C

and passivated the Mg and C acceptors. They found that

annealing at 450-500C restored the electrical activity, but

the hydrogen did not physically leave the films until much

higher (~800C) temperatures. In another study with H ion

implanted samples, they determined that the dissociation of

Mg-H complexes and the loss of hydrogen from GaN were

sequential processes. Reactivation occurred at s700C

annealing under N2, while significant concentrations of

hydrogen remained in the crystal even at 900C [Pea96b]. In

another approach to re-activate the dopant, Amano et al.

used a low-energy electron beam irradiation (LEEBI) process

with the accelerating voltage far below the threshold energy

for atom displacement. They achieved a hole concentration

of 2xl016 cm3 [Ama89]. The LEEBI treatment was limited to

the penetration depth of the electron beam, therefore only a

thin surface layer was converted to p-type material. On the

other hand, N2 annealing re-activated the dopant throughout

the entire epilayer.

Attempts to limit the presence of H in the growth

environment have also been reported. Abernathy et al.

obtained a hole concentration of 3x10'7 cm-3, doping with C

introduced as CCI4 using a He carrier gas [Abe95]. Zolper

et al. obtained p-GaN using ion-implantation of Ca acceptors

requiring an RTA treatment at 1100C [Zol96a], while Brandt

et al. have reported hole concentrations of up to 1019 cm-'3,

with doping efficiencies of up to 10% at room temperature,

grown by MBE with no post-deposition treatment [Bran94].


With n-type material being much easier to grow than p-

GaN, the large majority of ohmic contact studies are focused

on n-GaN. In this section, advances made in contacting n-

GaN are discussed and related to contact schemes for p-GaN.

The contacts and heat treatments used to obtain low contact

resistances are described as well as work investigating the

interfacial reactions which lead to ohmic contacts. Finally,

previously reported contact schemes for p-GaN will be



A major topic of discussion for GaN is whether the

Fermi level is pinned as for GaAs or is unpinned as for

ZnSe. A pinned E, would lead to barrier heights not

dependent on the metal work function. If EF is not pinned,

small work function metals should give lower potential

barriers and form ohmic contacts to n-type materials, since

the Schottky contact theory predicts that the barrier height

is equal to:

OB =M-XS. (2.13)

Recent studies have reported that the barrier height

was not solely dependent on the metal work function. Guo et

al. used capacitance-voltage (C-V) measurements to

determine that Schottky barrier heights of Pt and Pd on n-

GaN were 1.04 and 0.94 eV [Guo95]. They proposed that since

the difference in work function between Pt and Pd was

greater than the difference in measured Schottky barrier

height, that 0. did not uniquely determine 0.. Wang et al.

measured Pt and Pd Schottky diodes on GaN and reported no

dependence of the barrier height on the difference between

the metal work functions [Wan96].

Other work has also shown that the Fermi level is not

pinned. Foresi and Moustakas deposited Al and Au contacts,

and found Al to be ohmic as deposited while Au required a

575C anneal to obtain ohmic characteristics. They

speculated that Au ohmic contacts were due to Au diffusion

into the GaN, but showed no evidence to support this

speculation. This work supported the idea that the Fermi

level was not pinned since OA

Ag, Au, and Ti contacts and showed that as deposited Ag

((D,=4.3 eV) contacts were weakly rectifying while Au

provided strongly rectifying contacts. Ti (OTi=4.4 eV) was

rectifying as-deposited but became ohmic upon annealing due

to the formation of a TiN layer [Mil96]. Finally, Fan et

al. employed a Ti/Al/Ni/Au contact scheme with an in-situ

reactive ion etching (RIE) before deposition to remove the

surface oxide. They obtained a low specific contact

resistance (p,=8.9xl0-' Qcm2) and speculated that the surface

defect density following the RIE was not large enough to pin

the Fermi level [Fan96].

Thus the data show that either the Fermi level is

completely or only partially pinned. It is generally

speculated that partial Fermi level pinning is due to

interfacial contamination at the GaN surface rather than an

intrinsic property of the semiconductor itself.

The majority of these studies concentrated on

determining the mechanism of ohmic contact formation. Lin

et al. grew an InN/GaN short period superlattice (SPS)

sandwiched between a GaN channel and an InN capping layer.

This SPS allowed greater tunneling to occur through sub-band

states which led to specific contact resistances as low as

6x10"5 Qcm2, with GaN doped -5x10" cm-' without any post-

annealing [Lin94a]. Another mechanism for ohmic contact

formation has been to increase the probability of tunneling

by increasing the doping concentration of the near surface

region, as discussed above. Since N vacancies act as donors

in GaN [Jen89], increasing their concentrations either by

high temperature anneals or interfacial reactions have been

used. Lester et al. deposited non-alloyed Ti/Al contacts on

Si implanted GaN where the ohmic character (pc=lxl0O5 Qcm2)

was believed to be due to the 1120C implant activation

anneal which generated nitrogen vacancies [Les96], Zolper

et al. have collected Auger electron spectroscopy (AES) data

to support the hypothesis that an AlN encapsulant will

reduce N loss from the GaN substrate during an anneal at

1100C. Without a capping layer, N loss may create an n*

region at the surface [Zol96b]. Cole et al. investigated W

contacts on n-GaN and concluded that upon annealing at up to

1000C, formation of P-W2N and W-N interfacial phases were

responsible for the ohmic behavior (pc=8xl0-5 Qcm2 for

ND=1.5xl0'9 cm-'3). They proposed that the N to form these

phases out-diffused from the GaN without changing the

original structure, leading to an accumulation of N

vacancies near the GaN surface [Col96]. A similar result

was reported for Ti/Al contacts, in which formation of a

TiNX, interfacial phase was postulated to create N vacancies

[Bin94]. Lin et al. have speculated that only 1-2

monolayers of TiN are needed to form N vacancies which would

generate a 100A layer of GaN with an electron density of

1020 cm-3 [Lin94b]. Ruvimov et al. detected a TiN layer by

XTEM at the Ti/GaN interface after annealing at 900C

[Ruv96]. Finally, Smith et al. investigated Al contacts and

found the formation of a thin AiN layer on the surface led

to increased carrier conduction, presumably due to increased

N vacancies [Smi96].

Various other studies of multilayer contact schemes

leading to either extrinsic doping of the near surface

region or phase formation led to decreased contact

resistances. Miller and Holloway postulated that in

Au/Si/Ti contacts, Si diffused to the GaN surface and served

as an n-type dopant [Mil96]. For PtIn2 annealed at 800C

for 1 minute in a high purity Ar atmosphere, In was

incorporated into the GaN lattice producing (InGa1.x)N and

Pt(In,Ga)2. Formation of (InxGa1.,)N resulted in ohmic

behavior with pclxl0'3 Qcm2 [Ing97]. Ping et al. showed

that annealed Pd/Al/n-GaN contacts formed a Pd2Al3 phase

with pc=l.2xl05 Qcm2 upon annealing at 650C for 30 seconds

[Pin96]. Luther et al. reported that Ti in Ti/Al contacts

reduced the native interfacial oxides, Al diffused through

the Ti to the interface, and an Al-Ti intermetallic formed

which contacted the GaN. These contacts became ohmic upon

annealing at 400C for 3 minutes in an Ar/4%H2 atmosphere

(pc=7xl0-6 Qcm2 for ND=5xl017 cm-3) [Lut97].


In contrast to n-GaN contacts, which have been widely

studied, there is very little information on p-GaN contacts.

A limited number of studies have reported Schottky barrier

heights and that interfacial contamination has limited

contact formation. However, most reports simply mention the

p-contact used in a particular device. There has been very

limited information on the formation of interfacial phases

or conduction mechanisms in p-GaN. Due to the large Eg (3.4

eV) and X. (4.1 eV) of GaN, a metal would need 0m7.5 eV to

achieve an ohmic contact with the Fermi level unpinned.

Unfortunately there are no metals with work functions larger

than 5.6 eV, [Sze81c] thus interfacial reactions, doping of

the near surface region, or graded band offsets must be used

for ohmic contact formation.

Two groups have reported the effect of metal work

function on contact properties to p-GaN. Mori et al.

measured the Schottky barrier heights and a contact

resistances of Pt, Ni, Au and Ti to p-GaN. Their data

showed both 0, and pc decreased with increased metal work

function, as expected for an unpinned Fermi level [Mor96].

Ishikawa et al. investigated the effects of surface cleaning

on the electrical properties of metal/GaN interfaces, and

measured pc for Pt, Ni, Pd, Au, Cr, Ti, Al and Ta on p-GaN.

They showed that removal of the interfacial contamination

layer between the metal and the GaN did not significantly

reduce the contact resistance. However, the resistance

decreased exponentially with increased metal work function.

Also no reaction between any of these metals and GaN could

be detected after annealing at temperatures below 500C. It

was proposed that for ohmic contacts, a metal is needed

which would react with GaN during annealing at high

temperatures [Ish97]. The remainder of information on p-GaN

contacts consists of the metals used for devices. Early

work on p-n junctions and p-GaN/n-InGaN/n-GaN double

heterostructure blue Light emitting diodes (LEDs) used Au as

the p-contact [Nak91 and Nak93]. Subsequently, a Ni

interfacial layer was added between the Au and p-GaN

[Nak95]. Other groups also used this Ni/Au scheme [Kha95]

with Molnar et al. reporting thicknesses of 200A Ni/2000A

Au [Mol95]. All subsequent reports by Nakamura et al.

regarding their LEDs and laser diodes (LDs) used this Ni/Au

scheme (e.g. [Nak96]). Other groups have used Ni [Sak95],

Au + Zn [Kug95] and Ti/Mo/Au (non-ohmic) [Gol93]. None of

these studies reported the electrical properties of the

contacts or metallurgical reactions occurring between the

metals and semiconductor. In summary, while there has been

a great deal of work on n-GaN contacts, the understanding of

contacts to p-GaN is very limited and needs to be elevated

for GaN based blue light emitting devices to reach their



0s Xs



.. . . ......... E F

Figure 2.1A. Metal and p-type semiconductor before contact.





- -


Figure 2.1B. Metal and p-type semiconductor in thermal
equilibrium after the contact has been made.



Filled States



Figure 2.2. Metal/p-type semiconductor energy bands, showing
the effects of surface states, after contact has been made.


1. AON A 101 SeT.O)
1.0 .*ZnS GiN ,K.o.
*- S:02 Z;O I__ KTZ3

p[ GoS GO?,

-9 C4
P4 .6 4e

R ~*Znse

/G ,eCdSe

E- 2 cd*GOP

0 0 I I I -- ,I II
0 .4 .8 1.2 1.6 2.0 2.4


Figure 2.3. Index of interface behavior, S as a function of
lattice electronegativity difference AX between the cation and
anion for various semiconductors [Kur69].


" Ec


0 Hole

Figure 2.4. Energy band diagram of a forward biased Schottky
barrier junction on an n-type semiconductor showing different
transport processes; (a) thermionic emission, (b) thermionic
tunneling through the barrier, (c) carrier recombination in
the depletion region, and (d) hole injection from the metal
into the semiconductor [Sha84d].



The following chapter describes the experimental

process that was followed to prepare electrical contact

samples to p-ZnTe and p-GaN. The procedure consisted of

initial cleaning of the samples followed by metal contact

deposition. These contacts were then heat treated and

characterized in terms of their electrical properties,

surface composition, surface morphology and interfacial

reaction products.

Deposition and Processina

D-ZnTe Contacts

Investigation of Au contacts was performed on ZnTe/ZnSe

samples grown on semi-insulating GaAs substrates using a

custom made Molecular Beam Epitaxy (MBE) system and the

sample configuration shown in Figure 3.1 [Jeo95]. An

undoped 0.5 =un ZnSe buffer layer was epitaxially deposited

on the GaAs, followed by a 2.3 um p-ZnSe epilayer. The

ZnSe/ZnTe multi-quantum well (MQW) structure described by


Hiei et al. [Hie93] was epitaxially deposited onto the p-

ZnSe layer. The MQW structure was capped by a 1100A

heteroepitaxial p-ZnTe layer. An Oxford Applied Research

Systems nitrogen free radical source was used for the p-type

dopant layers. The free hole density was measured by a Hall

experiment to be -3x10" cm'3.

No further surface preparation was done on the as-grown

ZnTe samples before metal deposition. The Au contacts were

deposited by sputtering [Ohr92] in a custom built sputter

deposition system [Tru92]. The system contained two radio

frequency (RF) and two direct current (DC) 2" diameter

planar magnetron sputter guns manufactured by US Inc. The

vacuum chamber was a quartz bell jar with a Viton seal,

seated on a stainless steel base plate. The system

consisted of a liquid nitrogen cooled (trapped) diffusion

pump (Varian VH5-6) which was backed by a Sargent-Welch

mechanical pump, model #1397. The ultimate pressure reached

by the system was -2x10-7 Torr.

For deposition, the samples were loaded in the sputter

chamber (Figure 3.2) which was pumped to a base pressure of

at least 2x10"6 Torr. The 1500A thick Au contacts were

deposited from a 2 inch Pure Tech Inc. sputter target,

(99.99% purity) using a DC sputter gun with a potential of

375 V, and a power of -100 W. The gas pressure was 18 mTorr

argon and a deposition rate of -188 A/minute was achieved.

To improve the uniformity of the film thickness, the sample

was rotated about the system axis at 25 rpm. The Au

contacts were patterned as dot contacts (0.8 mm diameter)

during deposition using a stainless steel shadow mask. This

method resulted in electrically isolated discrete Au dot

contacts separated by -0.4 mm (Figure 3.3).

Following deposition, the samples were heat treated in

a quartz tube furnace in flowing forming gas (10% H2, 90%

N.). Separate Au/ZnTe samples were individually heat

treated at 150, 200, 250, or 350C using 15 minute

increments for total times up to 90 minutes.

D-GaN Contacts

Metal contacts were deposited on a variety of GaN

epitaxial films grown by metalorganic chemical vapor

deposition (MOCVD) on a polycrystalline GaN buffer layer on

a (0001) sapphire substrate [Kha91]. Mg acceptors were used

to obtain p-type conduction and carrier concentrations at

296K were measured by a Hall technique to be between 5x1016-

4.5x10"' cm-'3. The contact schemes investigated consisted

of Au, Ni/Au, Ni/C/Au, Pd/Au, and Cr/Au with the first metal

listed being deposited first and adjacent to the GaN. All

samples were degreased prior to deposition using acetone,

methanol, and de-ionized water (DI H20) for 5 minutes

followed by a N2 blow dry. Any native oxide was removed

using a 10:1 DI:HC1 etch for 5 minutes followed by a 5

minute DI rinse and N2 blow dry. The samples were then

immediately introduced to the deposition chamber.

The 2000A thick Au contacts were DC magnetron sputter

deposited using the previously described sputter system and

the same target as was used for the Au/p-ZnTe contacts. For

GaN contacts, the sputter atmosphere was 25 mTorr Ar with a

power of 40 W. For all GaN contacts, a smaller stainless

steel shadow mask was used which defined contacts with a

diameter of -0.5 mm and a spacing of -0.2 mm.

All contacts other than Au were deposited in a Davis

Wilder vacuum system with an Airco Temescal four pocket

electron beam evaporator [Ohr92], powered by a model ES-6

power supply (Figure 3.4). The system was pumped with a

liquid nitrogen trapped Varian VH5-6 oil diffusion pump

backed by an Edwards 40 two stage mechanical pump, providing

a base pressure of ~6x10-7 Torr. For all electron beam

evaporations, the deposition pressure was sl-3x10"6 Torr and

the charge consisted of metal pellets with the following

purity's: Ni, (Cerac Inc., 99.95%); Au, (Materials Research

Corporation, 99.95%); Pd, (AESAR Johnson Matthey Inc.,

99.8%); Cr, (Union Carbide, purity unknown), and C,

(graphite slug, purity unknown). The metal layer thickness

was monitored using a quartz crystal oscillator controlled

by a Kurt J. Lesker Company QXM-500 controller. The GaN

contact thickness varied with different experiments and will

be noted in the results section of Chapter 5.

Following deposition, all metallization schemes were

heat treated in the quartz tube furnace previously mentioned

using either flowing forming gas or N2 as the ambient.

Separate samples were heated at 200, 400 or 600C for 5, 15,

and 30 minutes. The Ni/Au, Pd/Au and Cr/Au samples were

also heated to 900C for 15 seconds in a 50 cm quartz tube

custom rapid thermal annealing (RTA) furnace with a 25 cm

hot zone and flowing N2 as the ambient.


Both the p-ZnTe and p-GaN samples were characterized

as-deposited and following each of the previously described

heat treatments in terms of their electrical properties, and

also for their surface composition, surface morphology, and

interfacial reaction products as required.

Electrical Characterization

The electrical properties of all contacts were

investigated using room temperature current-voltage (I-V)

measurements. The I-V data were obtained by measuring the

current flow between two adjacent dot contacts under an

applied bias. An automated system consisting of an IBM PC

with IEEE-488 communications, a Hewlett-Packard 6112 A DC

power supply and either a Keithely 488 picoammeter or a

Hewlett-Packard 3478A multimeter, depending on the current

range measured, was used to obtain the I-V data. A

schematic of this setup is shown in Figure 3.5. The

resulting data, collected over a range of -5 to +5 V, was

modeled as two back-to-back diode barriers, one forward

biased and the other reverse biased, and the ohmic or

rectifying nature of the contacts could be determined by the

linearity of the I-V curve. The reverse-bias breakdown

voltage could also be determined from the I-V

characteristics. Barrier heights were also determined from

the I-V data, but due to problems involving large ideality

factors for the contacts, the calculated barrier heights

were concluded to be inaccurate.

The charge transport mechanisms for Ni/Au, Pd/Au, and

Cr/Au contacts were determined using temperature dependent

I-V measurements. For these measurements, the samples were

mounted and Au wire bonded onto T05 headers which were

placed in a liquid nitrogen cooled sample holder. The I-V

characteristics were determined from -5 to +5 V using a

Hewlett-Packard 6111A DC power supply with a Hewlett-Packard

3455A digital voltmeter and the current was measured across

a IkQ resistor. The temperature ranged from 80-400 K and

was measured using a Pt resistance thermometer calibrated to

0.01 controlled by a Lake Shore Cryogenics DRC 82C

temperature controller. All measurements were taken at a

pressure of 10 mTorr.

The p-GaN samples were also characterized in terms of

their doping character using photoluminescence (PL) [Bru92]

and Hall effect measurements. The PL data was obtained

using excitation from a HeCd laser (X=325 run) at a

temperature of 22 K for an as-grown sample and also

following a 900C, 15 second RTA in N2. For the Hall effect

data, In pads were soldered onto the p-GaN and heat treated

at 400C for 3 minutes in N2 for better adhesion. The data

was taken with a current of 0.0001 A and a magnetic field

of 8 KGs. Following measurement, the In pads were etched

using a 10:1 DI:HCl solution and the same sample and

contacting process was used following the 900C heat


Surface Composition Analysis

Following heat treatment and electrical

characterization, the elemental surface composition was

determined by Auger electron spectroscopy (AES) [Hol80]

using a Perkin-Elmer PHI 660 scanning Auger microprobe

(SAM), or by secondary ion mass spectrometry (SIMS) [Ben87]

using a Perkin-Elmer PHI 6600. For contacts to p-ZnTe and

p-GaN, AES surface survey spectra were recorded over the

energy range of 50-2050 eV using a 5 keV, 30 nA electron

beam with a diameter of ~-1 un. For depth profiles, a 5 keV

argon ion beam was used to sputter the sample at a nominal

rate of -100A/min for 6 second intervals after which the

surface was scanned with the electron beam. SIMS depth

profiles of Au/p-ZnTe contacts were performed using a 5.5

keV, 53 nA Cs* ion beam with a 200 um x 200 pun raster size

using a quadruple mass spectrometer to analyze the sputtered

ions. For p-GaN, SIMS was performed only on the Ni/C/Au

contacts with a 3 or 5 keV, 30 nA Cs* ion beam. The beam

had a linear gating of 70% and a 300 um x 300um raster size.

Surface Morpholocv

Changes in surface morphology were investigated using

optical and scanning electron microscopy (SEM) [Gol92]. All

samples were viewed following heat treatments with an

Olympus BH-2 optical microscope with photographic

capabilities for any large changes in surface morphology.

SEM was performed on the Au/p-ZnTe samples using a JEOL 6400

scanning microscope with an accelerating voltage of 15 KV.

Interfacial Reaction Products

For Cr/Au contacts on p-GaN. cross-sectional

transmission electron microscopy (XTEM) [Edd74] and energy-

dispersive x-ray spectroscopy (EDS) [Bru92] were used to

determine interfacial reaction products formed in samples

subjected to a 900C, 15 second RTA. These data were

compared to as-deposited samples.

For XTEM, the sample preparation technique is extremely

important in obtaining high quality samples. For these

cross section experiments, two samples were first glued face

to face using G1 epoxy (Gatan). This structure was then

placed inside a stainless steel notched rod which was in

turn slid into a stainless steel outer tube with an inner

diameter just larger than the outer diameter of the rod. G1

epoxy was then applied to fix the samples in the notch and

to fill the area between the rod and the tube. The epoxy

was cured on a hot plate at 100C for 30 minutes. After

curing, samples were cut on a Accutom low speed saw using a

diamond blade to a thickness of =300 um. Figure 3.6 shows a

planar schematic view of the samples following the cut. The

samples were manually polished using SiC polishing paper

until one side was smooth, flipped over, and polished down

to a thickness of -100 um. A slotted Cu ring was then glued

to the sample using Ml epoxy (Micro Measurements Inc.) for

added stability. Next the samples were flattened to a

thickness of -30 um on a VCR Group Inc. Dimpler* using a

tool steel wheel and a 1 mm diameter cylinder. These

samples were ion milled in a Gatan Dual Ion Mill (Model 600)

with a voltage of 4-6 kV at imA with a final milling angle

of 120. After ion milling to produce a hole in the center

of the structure, the edges of the hole were electron

transparent. The samples were analyzed by amplitude

contrast imaging in a JEOL 200CX TEM using a two beam bright

field condition. Phase contrast imaging was used to obtain

high resolution images in a JEOL 4000FX high resolution TEM.

The EDS data was taken with a 400 KeV incident electron beam

with a spot size of =500A. Characteristic x-rays were


detected from 1-20 KeV using a Princeton Gamma Tech model

OS26-J029 detector.

p-ZnSe Barrier 20
p-ZnTe Well 17

11 A p-ZnTe
ZnSe/ZnTe MQW p-ZnTe
_________________ a^^~^ --^- A p-ZnTe
6 A p-ZnTe
5 A p-ZnTe
p-ZnSe 4 A p-ZnTe
2 Jm 3 A p-ZnTe

Figure 3.1. Schematic diagram of Au/p-ZnTe sample showing
ZnSe/ZnTe MQW.

Al Sample Holder


1 1 cm

4_ Ta Wire

Steel Shiel


3 cm

2" Au


Figure 3.2. Schematic of the sample holder and magnetron
source for the sputter system used to deposit Au conacts.

Circular Dot Contacts

Semiconductor Substrate

Semiconductor Substrate

ZnTe 0.8 mm
GaN 0.5 mm

(A) Top View

Metal Contacts

(B) Side View

Figure 3.3. Top (A) and side (B) views of dot contacts on
the semiconductor films showing electrical isolation of the
metal contacts.

0.4 mm
0.2 mm

Sample ---


Beam .. ..


Schematic of electron beam evaporation system.

Figure 3.4.

Electrical Connections

.-......-....-- ................ IEEE Connections

Figure 3.5. Hardward setup for current-voltage measurement

..................... ...... .................. ...


Stainless Steel

Stainless Steel

Metal Contact

GaN Samples

Figure 3.6. Planar view of Cr/Au on p-GaN sample used for



Formation and characterization of Au/p-ZnTe ohmic

contacts are discussed in this chapter. The sputter

deposited contacts were heat treated in forming gas and

current-voltage (I-V) measurements were used to determine

the effect of elevated temperatures on the electrical

properties. Heat treatments were identified which led to

the decrease in contact resistance and formation of the

ohmic contacts. The temperature stability of this contact

scheme was also studied. Analytical characterization

techniques, including Auger electron spectroscopy (AES) and

secondary ion mass spectrometry (SIMS) depth profiling, were

used to identify elemental surface composition, interfacial

reactions and compound formation induced by heat treatment.

These analytical results, along with surface morphological

changes determined from scanning electron microscopy (SEM)

were correlated with the electrical results to determine the

origin of the ohmic behavior and the eventual degradation of

the contact properties.


I-V Results

As-deposited I-V curves were non-linear and

characteristic of poor back-to-back Schottky contacts

(Figure 4.1). Upon heating to 150C for 15 minutes, there

was a slight reduction in breakdown voltage (VW) from -0.5

eV to -0.2 V, but the curves remained slightly non-linear.

The breakdown voltage was determined by extrapolating the

linear region of the I-V curve to its x-axis intercept. For

samples heat treated at 200C and 250C for 15 minutes,

linear I-V curves were measured with decreased contact

resistance. A maximum current density of 2.3 A/cm2 at 5 V

was observed following a 250C, 15 minute heat treatment.

Upon heating at 350C, the I-V curves remained linear but

the resistance was a factor of 25 higher than for samples

heated at lower temperatures for equivalent times,

indicating degradation of the contacts or increased bulk

resistance of the p-ZnTe (Figure 4.2). Increased time at

this temperature led to an increased contact resistance.

The changes that occurred upon heat treatment of the Au/p-

ZnTe contacts are summarized in Figure 4.3 showing that the

contacts were thermally stable only for T1250C.

AES depth profiles were collected for as-deposited

samples and after heat treating at 200, 250, or 350C for

90 minutes. The as-deposited profile (Figure 4.4) shows an

abrupt Au/p-ZnTe interface indicated by the rapid decrease

of the Au signal at -22 minutes. For samples heat treated

at 200 and 250C (Figures 4.5 and 4.6) the slope of the Au

signal versus sputter time was not as large as that of the

as-deposited case. The smaller slopes of the Au and Te

signals versus depth (at -11 minutes in Fig. 4.5 and -9

minutes in Fig. 4.6) indicated that Au had begun to diffuse

into the near surface region of the p-ZnTe leading to a

slight broadening of the metal/semiconductor interface.

Even with this limited diffusion, the interface remained

reasonably planar with no evidence of compound formation.

In Figure 4.7, the AES depth profile for the 350C heat

treatment would be consistent with an interface between Au

and ZnTe which was no longer planar. Au had diffused

throughout the ZnTe capping layer and into the underlying

ZnSe. The change of Au slope leading to a relatively flat

Au signal from -25-35 minutes of sputtering time suggests

formation of an interfacial phase or compound.

SIMS depth profiles were measured on samples as-

deposited and heat treated at 250 or 350C for 90 minutes.

The as-deposited profile (Figure 4.8) is consistent with a

very planar Au/ZnTe interface, similar to the AES depth

profile, with no indication of any Au diffusion into the

semiconductor. Upon heating at 250C for 90 minutes (Figure

4.9) Au had begun to diffuse into the ZnTe. At 5 minutes of

sputter time the Au signal was larger than the Te signal as

opposed to data in Figure 4.8 at -4 minutes of sputter time

where the Au signal was less than the Te signal strength and

remained lower throughout the profile. At 350C for 90

minutes (Figure 4.10), it is readily apparent that Au has

diffused through the entire ZnTe capping layer and into the

underlying ZnSe film.

Surface Morpholo-v

Changes in the surface morphology, which were

significant to thermal stability of the Au contacts, were

observed. The as-deposited Au film was specular reflective

to the eye and very smooth when examined by scanning

electron microscopy. After heat treating for 90 minutes at

T<250C the Au remained specular to the eye. After 250C

for 90 minutes, optical microscopy and SEM showed that the

ZnTe between the Au dot contacts became hazy (Figure 4.11).

An AES surface survey showed the presence of Au between the

dot contacts, creating an extended diffusion zone =100m

across the ZnTe surface. For the same sample at t&75

minutes, T=250C, there was no evidence of this diffusion.

At 350C for 90 minutes, a rough surface indicative of

further degradation was seen (Figure 4.12).


Ohmic contacts are characterized by their electrical

properties. This section will discuss three changes in the

I-V curves upon heat treatment and provide metallurgical

explanations for the changes. These changes were: 1)

decreased breakdown voltage at 150C, 2) increased current

levels and linear I-V with increased temperature above

150C, and 3) dramatic resistance increase and contact

degradation at 350C.

To understand the properties expected from Au/p-ZnTe

contacts, it must first be determined if the Fermi level is

pinned. From the Schottky model for a semiconductor with no

Fermi level pinning, Eq. 2.1 predicts the potential offset

in the valence band. For Au/p-ZnTe contacts with Eg=2.26 eV

[Mil72], XznT.=3.5 eV [Mil72], and 0Au=5.1 eV, Eq. 2.1 yields

a value of 0.66 eV. This is in good agreement with the

experimentally observed as-deposited value of breakdown

voltage =0.5 eV. This would only be expected if the Fermi

level was not pinned in p-ZnTe.

The decreased breakdown voltage upon heat treating to

150C for 15 minutes is attributed to decomposition of the

thin interfacial contamination layer, evident from the weak

oxygen signal in Figures 4.4-4.7. The increased current and

linear I-V curves after heating to 200C are attributed to

diffusion of Au into the underlying ZnTe. The analytical

results from the AES depth profiles (Fig. 4.5 and 4.6) show

a broadening of the Au/ZnTe interface consistent with Au

diffusion into the near-surface region of the p-ZnTe. The

SIMS depth profile for 250C (Fig. 4.9) also shows direct

evidence of Au diffusion into the ZnTe capping layer. Au

has been shown to be principally accommodated

substitutionally on Zn sites (AUz,) and to form a simple

acceptor level at E,+272 meV. [Mag80] Thus it is proposed

that the introduction of Au as an acceptor species in ZnTe

was responsible for the increased current after elevated

temperature treatments and therefore the ohmic contact.

If Au was acting as an acceptor in p-ZnTe and increased

the doping concentration of the near-surface region, the

main conduction mechanism in these Au/p-ZnTe contacts should

have been field emission. To evaluate whether field

emission or thermionic emission was dominant, thermionic

emission currents were modeled using Eq. 2.5-2.8. For these

calculations, the following values were used; 0B for Au on

p-ZnTe =0.73 eV, T= 300 K, area=0.005 cm2 (0.8 mm diameter),

NA=3x10'18 cm-'3, hole effective mass=0.60, and low frequency

dielectric constant e=9.4. [Wag92] The built in potential

of the junction (Vi) was calculated by determining the

position of the Fermi level and subtracting this value from

0.. Based on these calculations, the reverse bias breakdown

voltage for the Au/p-ZnTe contacts would be 18 V if

thermionic emission was dominant, rather than the observed

value of 0.5 eV. Thus thermionic emission was not

dominating current transport.

Therefore the properties of the Au/p-ZnTe contacts were

modeled based on field emission. From the equations for

field emission, taking the natural logarithm of Eq. 2.11 and

with some simple rearrangements:

ln(I)=ln(Is)+(-qV,/Eo). (4.1)

Based on this equation, a plot of the ln(I) vs. V should

produce a straight line if tunneling is dominant. Figure

4.13 shows a plot of ln(I) vs. V from 0-5 volts for an as-

deposited sample as well as samples heat treated at 150,

200, 250, or 350C for 15 minutes. As can be seen in all

cases, the plot of ln(I) vs. V was only linear for Vz2V,

indicating that tunneling was not dominant over the entire

range of reverse biases.

The barrier height for the metal/semiconductor contacts

can also be determined from the plot of ln(I) vs V. From

these curves the value of the saturation current Io can be

obtained by extrapolating the linear portion of the curve to

V=0. Knowing Io, Richardson's constant A, the diode cross

sectional area S, and the temperature T, the barrier height

(0.) can then be determined from:

I0=SAT'exp(-OB/kT). (4.2)

The value of barrier height obtained in this way is the zero

bias barrier height and includes the image force barrier

lowering A,. The values of 0B for Au/p-ZnTe contacts are

shown in Table 4.1. However, when the diode ideality

Table 4.1. Barrier heights for Au/p-ZnTe contacts.

Heat Treatment 0B (eV)
As-deposited 0.433
150C, 15 min 0.426
200C, 15 min 0.419
250C, 15 min 0.412
350C, 15 min 0.504

factor, n, was determined from the following equation:

I=Io[exp(qV/nkT)-l], (4.3)

values of =100 were obtained which violated the initial

assumption of the analysis. Therefore this reduction of I-V

data to 0. is erroneous and will not be continued.

With neither thermionic nor field emission dominating

conduction over the voltage range from 0-3 V, it is proposed

that thermionic field emission is the dominant mechanism.

Thermionic field emission is an intermediate conduction

mechanism which combines both thermionic and field emission

processes. In a case where the barrier height is too large

for thermionic emission to take place and the depletion

width is too deep for tunneling to occur, current can be

transported by the thermal excitation of holes to a level at

which the potential barrier is thin enough for tunneling to

occur. This process is known as thermionic field emission

and it dominates for mid-range temperatures and moderate

doping concentrations. In Chapter 2, the tunneling

parameter E00 was introduced as a measure of the relative

importance of thermionic versus field emission transport.

From Equation 2.13 for Au/p-ZnTe, E00 is 0.0135 eV. This

value is approximately the room temperature value of kT

(0.0258 eV) which indicates thermionic field emission should

dominate conduction. While this value of E00 is smaller

than kT, increasing the near-surface carrier concentration

by Au diffusion would cause the value of E00 to approach kT.

The non-linearity of the ln(I) vs V plots at V<3V would

discount pure field emission.

TFE dominance is also supported by the fact that Au is

a relatively deep acceptor in ZnTe (272 meV). This level is

too deep for the acceptors to be sufficiently activated at

room temperature, leading to the non-linear I-V curve for

as-deposited samples. Upon heat treatment of the Au/p-ZnTe

contacts, increased Au diffused into the p-ZnTe would in

turn increase the carrier concentration. The Au atoms then

can participate in conductivity through the process known as

hopping conduction. [Mil73] In this process, a charge

carrier, in this case a hole, can move from one deep defect

center to another by the tunnel effect without activation

into the valence band. This method of conductivity is

common where there is a high density of these centers with

no alternative transport mechanism [Mark95].

Based on the AES and SIMS depth profile data, no

compounds formed at Ts250C which could have affected

current transport, which lends support to the Au doping

postulate. Surface conductivity played no role in the I-V

data measured for the contacts heat treated for 15 minutes.

The surface diffusion, evident in Figure 4.11, occurred upon

heating for extended times (90 minutes) and thus had an

insignificant role in the I-V data measured after 15 minutes

when no visible surface diffusion was present.

The final change in the I-V curves for Au/p-ZnTe upon

heat treating was the dramatic increase in resistance upon

heating to 350C. As was shown in the I-V curves of Figure

4.3, the current passed following a 350C heat treatment was

approximately two orders of magnitude lower than for the

contacts heat treated at lower temperatures. It was also

shown that increased heat treatment time at this elevated

temperature caused an increased resistance. It is proposed

that widespread diffusion of Au along with interfacial phase

formation, as evidenced from AES and SIMS depth profiles,

limited the temperature stability of these contacts.

Au and Te form a congruently melting compound AuTe2

[Mas90] which is postulated to be responsible for the

contact degradation in these Au/p-ZnTe samples. It is

suggested that this AuTe2 phase is forming at the expense of

the ZnTe layer, causing increased resistance. As mentioned

previously, Au may be substitutionally incorporated into the

ZnTe lattice on a Zn site. With increased time at this

elevated temperature, more Au can be incorporated into the

ZnTe until a point at which AuTe2 forms in place of the

ZnTe. It is this total breakdown of the ZnTe, caused by Au

diffusion and formation of AuTe2, which is postulated to be

responsible for the degraded contact performance at 350C.

The I-V curves in Figure 4.2 show that as the heat treatment

time at 350C was increased, more degradation occurred which

supports this postulate since diffusion is both time and

temperature dependent.


Formation of ohmic contacts to nitrogen doped p-ZnTe by

sputter deposition of Au films was demonstrated. As-

deposited contacts were rectifying with a breakdown voltage

of =0.5 eV, close to the barrier height value of 0.66 eV

obtained from the band alignment. This strongly suggests

that the Fermi level was not pinned in ZnTe. Slightly

decreased breakdown voltages after heat treatments at 150C

were attributed to the breakup of interfacial contamination

layers. Heating for 15 minutes at Tk200C resulted in ohmic

behavior which was credited to creation of near-surface

acceptors due to doping by Au without interfacial phase

formation. This led to increased conductivity through the

near-surface layer, culminating in a maximum current density

of 2.3 A/cm2 at 5 V following 15 minutes at 250C. This

increased conduction was attributed to thermionic field

emission occurring in the contacts at V<3V, and field

emission for V>3V. Au acceptors with an energy of EF+272

meV were proposed to participate in current transport by

hopping conductivity between deep centers in the lattice.

Diffusion of Au was extensive and resulted in AuTe2 compound

formation at 350C, leading to severe degradation of the

contact properties.

1.5e-2 -

1.Oe-2 -

5.0e-3 -

0.0e+0 -

-5.0e-3 -

-1.Oe-2 -

-1.5e-2 -

* As-deposited
0 150oC, 15 min
* 200C, 15 min
o3 2500C, 15 min

tI I I I 1I
6 -4 -2 0 2 4 6

Voltage (V)

Figure 4.1. I-V curves for Au/p-ZnTe.

6.0e-4 -

4. Oe-4 -

2. Oe-4 -

O.Oe+O -

-2. Oe-4 -

-4.Oe-4 -



Voltage (V)

Figure 4.2. I-V curves for Au/p-ZnTe heat treated at
350C in forming gas.

0 15 minutes
0 30 minutes
* 45 minutes
o 60 minutes
A 75 minutes
A 90 minutes


1. Oe-2 ...

5.0e-3 -

O.Oe+0 -

-5.0e-3 -

-1.Oe-2 -


* As-deposited
0 : 250C, 15 min
* 250C, 90 min
3 3500C, 15 min

i i I I I i
-6 -4 -2 0 2 4



Figure 4.3. I-V curves for Au/p-ZnTe.


3.0e+4 -
./"-*"*--*'-'.--. ....".."'"'''" '"\ Au

2.5e+4 -

S2.0e+4 -
4 Te

S1.5e+4 -
c Zn

1.Oe+4 -
/ I-/ -Se

5.0e+3 -

O.Oe+O -I .A
0 5 10 15 20 25 30 35 40 45 50

Sputter Time (Min)

Figure 4.4. AES depth profile of Au/p-ZnTe, as-deposited.





1.0e+4 -

5.0e+3 /e

0.Oe+O ,_- .., ,AA-
0 5 10 15 20

Sputter Time (Min)

Figure 4.5. AES depth profile for Au/p-ZnTe sample heat
treated at 200C for 90 minutes.




1. Oe+4 j A.
A Zn
r" iW i"" L iL.i,.
5.0o+3- J, y ^^i


O.Oe+O _r% In c A, .
0 5 10 15 20 25
Sputter Time (Min)

Figure 4.6. AES depth profile for Au/p-ZnTe sample heat
treated at 250C for 90 minutes.

4e+4 -

3e+4 -

2e+4 -

le+4 -

Oe+0 -

Sputter Time (Min)

Figure 4.7. AES depth profile for Au/p-ZnTe sample heat
treated at 350C for 90 minutes.


Zn v '

/ Se
" .-* ...-... F -

/ Te / "'

"'-... .. c...


le+5 -

.. ....... ................

Sle+4- \
". ... ZnSe
U *. \ .... 4__
-^ - -

*^ / "" **
w le+3 "..
.I .... .
I "'" ....-**: '^ ,,
I "' " .^
I-. I
le+2 \ZnTe

j I .
\1 \

0 2 4 6 8 10 12 14

Sputter Time (Min)

Figure 4.8. SIMS depth profile for Au/p-ZnTe sample,








0 5 10 15

Sputter Time (Min)

Figure 4.9. SIMS depth profile for Au/p-ZnTe sample heat
treated at 250C for 90 minutes.






le+O _/__

-k lv vr j. ,
le+O -- 1 A v i i vi ^.y v
0 5 10 15 20
Sputter Time (Min)

Figure 4.10. SIMS depth profile for Au/p-ZnTe sample heat
treated at 350C for 90 minutes.


Dot Contact-


Figure 4.11. SEM micrograph of Au/p-ZnTe sample heat
treated at 250C for 90 minutes showing an extended reaction
zone surrounding the Au contact.

Figure 4.12. SEM micrograph of the surface of a Au dot on a
Au/p-ZnTe sample heat treated at 350'C for 90 minutes
showing roughening of the Au contact.


-2 -

-4 -


-- -8

14. -10-

p As-deposited
0-- 150oC, 15 min
-12 --a- 200*C, 15 min

---- 250oC, 15 min
S350C, 15 min
-14 -

-16 -II

0 1 2 3 4 5

Voltage (V)

Figure 4.13. Plot of Ln(I) vs. V for Au/p-ZnTe.



The formation and characterization of electrical

contacts to p-GaN are described below. The contact

metallizations consisted of a single element (Au), bi-layer

contacts (Pd/Au, Ni/Au, and Cr/Au), and a three layer

contact scheme (Ni/C/Au). The elements are listed in the

order in which they were deposited, i.e. the first element

was deposited on GaN. As for Au/p-ZnTe contacts discussed

in Chapter 4, the effects of elevated temperature treatments

for various times on the electrical properties of the p-GaN

contacts were determined using I-V measurements. Analytical

characterization of the contacts included AES and SIMS depth

profiles, along with energy-dispersive x-ray spectroscopy

(EDS) and cross-sectional transmission electron microscopy

(XTEM) to determine composition and interfacial reaction

products. Photoluminescence (PL) and Hall measurements were

also used to determine doping character of the material.

Temperature dependent I-V data were used to determine the

dominant conduction mechanisms for the Ni/Au, Pd/Au and

Cr/Au contacts. In this chapter the results from I-V

measurements, analytical characterization, and temperature

dependent I-V are given for all metal contacts, followed by

a discussion of these results. Finally, the results are




I-V resulta

Au contacts (2000A) were sputter deposited onto p-GaN

which had a resistivity of 7 Qcm and a free hole

concentration NA=lxl01 cm=3. As deposited the I-V curves

were rectifying with a breakdown voltage of =2.5 V (Figure

5.1). The barrier height (0.) was calculated using the

method described in chapter 4 to be 0.62 eV, but again the

ideality factor was =100, therefore no further 0. data will

be reported. Instead only the breakdown voltage will be

determined using the procedure described earlier. When this

same sample was heat treated at 400C for 5 minutes in

forming gas, the breakdown voltage was reduced to =0.8 V.

After 600"C for 15 minutes, the breakdown voltage decreased

only to =1.5 V. The I-V curves remained rectifying for all

heat treatments.

The Au and Ga signals in AES depth profiles from

contacts as-deposited or heat treated at 400C for 15

minutes in forming gas can be seen in Figure 5.2. For the

as-deposited sample, there is a very sharp interface between

the Au and GaN with no indication of GaN decomposition. At

400C for 15 minutes, the slope of the Au signal is

identical to that of the as-deposited profile. The

interface is still planar with no evidence of GaN

decomposition or Au diffusion into the GaN matrix to the

level detectable by AES.


I-V results

Contacts of Pd/Au (500/1000A) on p-GaN (NA=9.8x1016 cm-3)

were also studied. For as-deposited contacts, the I-V

curves were rectifying (Figure 5.3) with a breakdown voltage

of =3 V. Heating at 200C for 5 minutes in flowing N2

caused the contacts to became more rectifying and pass less

current with an increased breakdown voltage. The I-V curves

were unchanged by increasing the time at 200C up to 30

minutes. After 400C for 5 minutes, the current was

slightly greater than for as-deposited samples. Heating for

times up to 45 minutes at 400C led to a 60% increase in

current and a reduction in breakdown voltage from 2 to 1.5

V. The I-V data from samples annealed at 600C for 5

minutes were similar to those after 400C for 45 minutes.

After an RTA at 900C for 15 seconds in N2, the contacts


still rectifying, but the breakdown voltage was decreased to

=1 V with 5 mA of current at 5 V.

As-deposited (Figure 5.4), the Au/Pd interface was

broadened and at the Pd/GaN interface there was a slight Pd

tail into the GaN, but no evidence of Ga out-diffusion into

the overlaying films following deposition. At 200C for 5

minutes (Figure 5.5) there was an increased broadening of

the Au/Pd interface with a small Pd signal continuous

throughout the Au layer to the surface. The AES survey

spectrum (Figure 5.6) showed a Pd transition at 327 eV,

indicating that Pd diffused through the Au layer. The Pd

signal also tailed into the GaN, but there was no evidence

of Ga out-diffusion. After 400C for 5 minutes (Figure

5.7), Pd continued to diffuse through the Au capping layer,

but did not build up at the surface. Thus there were

constant levels of Pd and Au throughout the first layer,

followed by a Pd-rich layer underneath. The Pd/GaN

interface has broadened, but there was still no evidence of

dissociation of the GaN. In Figure 5.8, the AES depth

profile after an RTA of 900*C, 15 seconds shows that Pd had

completely dissolved into the Au layer forming a Au:Pd

region contacting the underlying GaN. The depth profile

suggests that dissociation of GaN occurred because a Ga

signal was detected throughout the contact layer up to the

surface. The long Pd and Au tails into the GaN suggest

diffusion into the GaN. Following this RTA heat treatment,

the contacts exhibited both dark regions rich in Pd and Au

(Figure 5.9), and light regions rich in Pd, Au, Ga, and N

(Figure 5.10).


I-V results

Evaporated Ni capped with sputtered Au (500/2400A)

contacts deposited on the same substrate as the Au contacts

were rectifying as-deposited with a lower breakdown voltage

(=0.5 V) and larger current (=0.5 mA @ 5 V) than the Au/p-

GaN contacts (Figure 5.11). The I-V curves remained

rectifying up to 200C, but after a 400"C, 5 minute

treatment, they became nearly linear with -0.6 mA passed at

5 V. Upon heating this sample to 600C for 5 minutes, the

contacts degraded as reflected by lower current levels

(Figure 5.11).

The effects of Ni layer thickness and Ni to Au

thickness ratios are shown in Figure 5.12. Data are shown

from as-deposited and heat treated (400C, 5 min. in forming

gas) samples (p=17 acm) with evaporated 200/2000A or

500/1000A Ni/Au contacts. The as-deposited samples were

very similar in both current magnitude (0.1 mA @ 5 V) and

breakdown voltage (=2.2. V). After heat treatment, the

breakdown voltage was drastically reduced to =0.5 V even

though the I-V curves remained slightly rectifying. There

was very little difference in the I-V data for different

metal thicknesses and thickness ratios after the heat


The effects of heat treatment ambient were also

investigated for Ni/Au contacts. Initial heat treatments

were performed in a forming gas environment, but this was

changed to N2 due to reports of hydrogen passivating Mg

acceptors in p-GaN [Pea96a]. Use of N2 also reduced the

probability of N vacancy formation in GaN which may lead to

semi-insulating films. The difference in current levels

obtained was negligible between the different ambients, but

it was determined that N2 were preferable to prevent any

potential acceptor passivation, especially at elevated



An AES depth profile for an (500/1000 A) contact can

be seen in Figure 5.13. There is a planar Ni/Au interface

with no indication of intermixing upon deposition. The

Ni/GaN interface appears less sharp with a slight Ni tail

into the GaN indicative of near-surface diffusion, but this

could result from loss of resolution at greater sputter

depths [Hol8l]. There is, therefore, no evidence of GaN

dissociation upon deposition except for the rather low

breakdown voltage for as-deposited films. The depth profile

for a sample heated to 200C for 5 minutes in flowing N2 was

similar to the as-deposited profile. After 400C for 5

minutes (Figure 5.14) the Ni/GaN interface was slightly more

diffuse, and Ni diffused through the Au capping layer to the

surface of the contact. For heat treatments of 600*C

(Figure 5.15), dramatic evidence of GaN dissociation is

obvious. Ga was detected at the Au/ambient surface, however

Ni also diffused through the Au capping layer to the surface

and formed a NiO, layer. There was also an increased C

signal at the metal/p-GaN interface.


I-V results

As will be discussed, the increased current for heat

treated Ni/Au contacts could have resulted from C doping of

the GaN surface. Therefore, the next contact scheme

investigated was an evaporated carbon layer intermediate

between the Ni and Au layers to give a Ni/C/Au contact

(500/100/lOOOA) on p-GaN with a carrier concentration of

2x1017 cm-3. I-V data for the Ni/C/Au contacts were

rectifying as-deposited and remained rectifying with all

heat treatments up to 400C for 5 minutes (Figure 5.16).

The contacts exhibited a decrease in current vs. voltage

upon heating to 200C for 5 minutes. This was then followed

by increased current at 400C for 5 minutes. For all heat

treatment conditions, the currents were lower and the

potential offsets higher than in the Ni/Au contacts under

similar processing conditions.


The AES depth profile from an as-deposited Ni/C/Au

contact (Figure 5.17) showed more diffuse Au/C, C/Ni and

Ni/GaN interfaces than for a Au/Ni contact. The peak of the

carbon signal occurred before the Au signal had decreased to

zero and there was a very long C tail into the Ni layer.

The Ni/GaN interface was also very broad. The depth profile

after heat treatment at 400C for 30 minutes (Figure 5.18)

was similar to the as-deposited profile with no evidence of

Ni diffusion to the surface or of GaN dissociation.

In the SIMS depth profiles, there were well defined

interfaces between all layers in an as-deposited sample

(Figure 5.19) with a noticeable C tail into the Ni region of

the sample. After heating to 400"C for 30 minutes in

forming gas (Figure 5.20), the interfaces were less well

defined and the C signal at the Au/Ni interface broadened.

As in the Ni/Au contacts, there was evidence of Ni diffusion

through the Au contact to the surface and the Ni signal

tailed into the p-GaN substrate with no evidence of Ga at

the ambient/Au surface.


.I-V results

I-V results from as-deposited Cr/Au (500/1000A)

contacts were once again rectifying, but with a breakdown

voltage =1-1.5 V (Figure 5.21). Annealing at 200C for 5

minutes in flowing N2 resulted in rectifying contacts with

lower transported current, similar to the other contact

schemes. At 400C for 5 minutes, the breakdown voltage was

-1 V. Following an RTA at 900*C for 15 seconds, the I-V

curves were linear with =13 mA @ 5 V. For this sample an

upper limit on the specific contact resistance has been

calculated to be 54.3x101 Qcm2 by assuming a bulk resistance

of zero for the underlying p-GaN.

In the as-deposited AES depth profile (Figure 5.22),

the Cr/Au interface was sharp while the Cr/GaN interface

showed a Cr tail into the GaN, but no evidence of

dissociation of GaN. In contrast to the AES data from the

other metallization schemes, there were small but detectable

carbon concentrations at the Cr/Au and Cr/GaN interfaces.

After 200C for 5 minutes in N2 (Figure 5.23), the depth

profile was similar in all aspects to the as-deposited

profile. At 400C for 5 minutes, Cr diffused through the Au

capping layer to the surface and formed an oxide layer

similar to the Ni reaction observed in the Ni/Au contacts

(Figure 5.24 vs Figure 5.14). Following this heat

treatment, an increased C signal was only detected at the

Cr/GaN interface. After a 900C, 15 second RTA (Figure

5.25), there was massive Cr diffusion through the Au capping

layer to the surface and also into the GaN layer. GaN

dissociation was evident from the large Ga signal in the Au

region (-8-15 minutes sputter time) without a corresponding

N signal and the small Ga signal at the surface. There were

increased Cr, nearly constant N and reduced Ga signals in

the region of the original Cr layer.


XTEM and EDS were performed on as-deposited Cr/Au

contacts and following the 900C, 15 sec RTA anneal. For

the as-deposited sample, the micrograph in Figure 5.26 shows

a low magnification (x 73K) image of the contact/GaN

structure. The polycrystalline Au and Cr metal layers are

fairly uniform in thickness across the contact. In this

figure the appearance of the Au layer being <1000A thick is

attributed to the ion milling process which removed part of

the Au capping layer. The ring structure of the selected

area diffraction pattern (SADP) in Figure 5.27 is indicative

of polycrystalline material with rings from both Au and Cr

observed as shown. EDS data (Figure 5.28) were consistent

with discrete metal layers. After the RTA anneal, the

metal/GaN interface was no longer planar as shown by Figure

5.29. In the thicker regions of the metal contact in Figure

5.29, EDS spectra (Figure 5.30) show the presence of both Au

and Cr. While exact compositions can not be obtained from

this data, the relative intensities of the Au and Cr peaks

can be used to ascertain that Cr has diffused through the Au

capping layer and there are regions where both Au and Cr are

present which is consistent with intermixing. This Cr

diffusion is also supported by the depth profile (Figure


In Figure 5.31 a high resolution XTEM of the

ambient/contact surface is shown. The presence of Ga at

this surface, as determined by EDS (Figure 5.32), indicated

Ga out-diffusion from the GaN and segregation to the

surface. SADP's taken from the regions indicated in Figure

5.31 show a polycrystalline phase containing Au and Cr

(Figure 5.33), and a multi-crystal phase which includes Ga

along with Au and Cr (Figure 5.34).

Temperature DeDendent I-V

Temperature dependent I-V measurements were used to

determine the conduction mechanisms dominating in the Ni/Au,

Pd/Au and Cr/Au contacts. The I-V data for a Ni/Au

(500/2500A) sample heated to 600*C (873K) for 5 minutes in

flowing N2 are shown in Figure 5.35. The contacts were

rectifying at all temperatures between 80-400 K. The

current increased and the reverse bias breakdown voltage

decreased with increasing temperature.

The temperature dependent I-V data for a Pd/Au

(500/1000A) sample heated to 6000C (873K) for 15 minutes in

flowing N2 are shown in Figure 5.36 for temperatures 90-390

K. As for Ni/Au contacts, the current increased and the

reverse bias breakdown voltages decreased with increasing

temperature. The current increase was much more dramatic

for Pd/Au versus Ni/Au contacts, but all contacts remained


For Cr/Au (500/2500A) contacts annealed at 400*C (673K)

for 5 minutes in flowing N2, the 80-400 K temperature

dependent I-V data can be seen in Figure 5.37. The I-V

curves were slightly rectifying at all measurement

temperatures. Unlike Pd/Au and Ni/Au contacts, there was

very little increase in current with increased temperature.

The reverse bias breakdown voltage also remained nearly

steady over this temperature range.


Metal Contact Schemes

As previously mentioned in Chapter 1, GaN has a direct

band gap of 3.4 eV and an electron affinity between 3.3-4.1

eV. Plugging these values into Equation 2.1 to find the

potential offset in the absence of Fermi level pinning leads

to values of 6.7-7.5 eV required for metal work functions to

decrease the potential offset at the semiconductor valence

band to zero. This becomes important due to the fact that

metal work functions do not have values greater than =5.5 eV

[Sze8l]. Thus in the absence of interfacial reactions,

there will always be a potential offset between metals and

p-GaN. Because of this, the interfacial reactions between

the metal contacts and p-GaN upon heat treatment are of

utmost importance for formation of low resistance ohmic


Thus there was a need to discover a metal contact

scheme that would react with the underlying GaN either as-

deposited or upon heat treatment to form ohmic contacts.

These metal schemes could provide low resistance contacts in

one of two ways. First, the formation of a large work

function compound or phase would decrease the potential

offset at the valence band, and second the doping

concentration in the near-surface region of the

semiconductor could be increased by incorporation of

acceptor dopant species from either the contact layer or

interfacial contamination.


The work function of Au is 5.1 eV [Sze8l], so the

calculation reported above predicted that the as-deposited

contacts would be rectifying, as was the case. Au has been

reported to exhibit limited reaction with GaAs as a single

element contact and increased reactivity upon incorporation

in a contact scheme such as Au/Ge/Ni or Au/Zn [Bras83]. The

ohmic behavior is then observed through a solid state

regrowth process in the manner described by Sands et al.

[San88], and Holloway et al. [Hol97]. As seen in Figure 5.2

for the Au/GaN contacts, there was no evidence of Au:Ga

phase formation upon deposition or following heating at

400C. Thus Au on GaN is non-reactive, leading to the