Citation
Thermally Stable Ohmic and Schottky Contacts to GaN

Material Information

Title:
Thermally Stable Ohmic and Schottky Contacts to GaN
Creator:
Voss, Lars Fredrik
Place of Publication:
[Gainesville, Fla.]
Publisher:
University of Florida
Publication Date:
Language:
english
Physical Description:
1 online resource (144 p.)

Thesis/Dissertation Information

Degree:
Doctorate ( Ph.D.)
Degree Grantor:
University of Florida
Degree Disciplines:
Materials Science and Engineering
Committee Chair:
Pearton, Stephen J.
Committee Members:
Abernathy, Cammy R.
Singh, Rajiv K.
Norton, David P.
Ren, Fan
Graduation Date:
5/1/2008

Subjects

Subjects / Keywords:
Annealing ( jstor )
Borides ( jstor )
Contact resistance ( jstor )
Electric potential ( jstor )
Electrons ( jstor )
High temperature ( jstor )
Nitrides ( jstor )
Surface temperature ( jstor )
Temperature profiles ( jstor )
Temperature resistance ( jstor )
Materials Science and Engineering -- Dissertations, Academic -- UF
contacts, gan, high, led, ohmic, schottky, semiconductor, stable
Genre:
Electronic Thesis or Dissertation
bibliography ( marcgt )
theses ( marcgt )
Materials Science and Engineering thesis, Ph.D.

Notes

Abstract:
Thermally stable Ohmic and Schottky contacts to GaN by Lars Voss, Materials Science and Engineering, University of Florida, Gainesville, FL. A study of Ohmic and Schottky contacts to Gallium Nitride was conducted in order to improve the thermal stability of said contacts for use in microelectronic and optical devices. Contacts were based on conventional schemes with an added layer in order to minimize intermixing between layers and improve the integrity of the contacts upon high temperature annealing and aging. Contact materials examined were of three types: nitride, boride, and Ir. The most successful schemes were based on borides and provided superior characteristics when subjected to both high temperature anneals as well as long term thermal aging. In addition, light emitting diodes were fabricated with conventional contact schemes and selected high temperature schemes. It was demonstrated that the high temperature schemes provided superior long term thermal aging characteristics compared to conventional schemes. ( en )
General Note:
In the series University of Florida Digital Collections.
General Note:
Includes vita.
Bibliography:
Includes bibliographical references.
Source of Description:
Description based on online resource; title from PDF title page.
Source of Description:
This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Thesis:
Thesis (Ph.D.)--University of Florida, 2008.
Local:
Adviser: Pearton, Stephen J.
Statement of Responsibility:
by Lars Fredrik Voss

Record Information

Source Institution:
UFRGP
Rights Management:
Copyright Lars Fredrik Voss. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Embargo Date:
7/11/2008
Classification:
LD1780 2008 ( lcc )

Downloads

This item has the following downloads:


Full Text





rough Ni surface after annealing at 6000C [44,108]. A way to prevent excessive intermixing and

contact morphology degradation is to use a high-melting-point diffusion barrier in the contact

stack [109-111]. TiB2, with a melting temperature of 30000C, reasonable electrical resistivity

(28 [tQ.cm) and thermal conductivity (26 W.m-.K-1), and heat of formation comparable to those

for silicides or nitrides [112], shows promise as a diffusion barrier.

In this letter, we report on the long-term annealing characteristics at 200-350C of

InGaN/GaN MQW-LEDs with TiB2- and Ir-based p-Ohmic contacts. This high-temperature

stress stimulates accelerated aging of GaN-based LEDs and gives an idea of the expected

reliability of the Ohmic contacts. The use of TiB2 or Ir as a diffusion barrier in Ni/Au-based

contacts is found to produce superior long-term stability of turn-on voltage, leakage current and

output power.

6.2 Experimental

The MQW-LED structures were grown by metal organic chemical vapor deposition on c-

plane sapphire substrates. The layer structure consisted of a low-temperature GaN buffer, 3 |tm-

thick n-GaN, 0.1 |tm n-AlGaN clad, three period undoped InGaN/GaN MQW active, 0.1 [tm p-

AlGaN clad, and 0.3 |tm p-GaN layer. 1 [tm-deep mesas were fabricated using C12/Ar

inductively coupled plasma etching. Ohmic contacts to n-GaN were formed by lift-off of e-beam

deposited Ti (20 nm) / Al(80 nm) / Pt(40 nm) / Au(80 nm) subsequently annealed at 8000C for 1

min in a flowing N2 ambient in a rapid thermal annealing (RTA) furnace. The first p-

metallization scheme investigated consisted of Ni (50 nm) / Au (80 nm) / Ir (50 nm) / Au

(80 nm). The second contact scheme was Ni (50 nm) / Au (80 nm) / TiB2 (50 nm) / Ti (20 nm)/

Au (80 nm). Ti was added to improve the adherence of Au on TiB2. The Ni/Au layers were

deposited by e-beam evaporation while the Ir/Au and TiB2/Ti/Au overlayers were deposited by















2000 6000

S4000 Sample #2
S0Annealed at 750C
o 2000

C/S 2000 amle Au C/S 0 Au
c/s 2000
Sample #1 u Ti l i AlNi
As deposited Au Zr Au
-4000 -2000 Au
0 % Atomic %
Atomic % Au Cl 44.9
Au3 52.1 Af01 18.9
Auc : -4000
u Cl 44.2 -4000 Au3 15.7
-6000 l 2 Zr2 14.0
01 1.5 A12 2.9
-6000 Til 2.2
0 Nil 1.4
-8000 c
500 1000 1500 2000 -8000 500 1000 1500 2000
Kinetic Energy (eV) Kinetic Energy (eV)


100 100
90 90
S80 Ni 80
.2 70 .2 70
7Ga An
S60 Zr Ga Au
,50r ,60
50o \ 50/
40 0 40
*30 zr
S30 30 Zr
O O


0 / 20

0 500 1000 1500 2000 2500 3000 0 1000 2000 3000 4000 5000
Sputter Depth (A) Sputter Depth (A)





Figure 3-14: Surface scans and depth profiles ofNi/Au/TiB2/Ti/Au Ohmic contacts on p-GaN as
a function of anneal temperature.


4UUU


OUUU

























Figure 6-4: Image of aged LEDs with (a) Ni/Au and (b) Ni/Au/TiB2/Ti/Au p-Ohmic contacts. In
(a), the picture was taken for a forward bias of 10 V (I = 80 pA), while the forward
voltage in (b) was 4.5 V (I= 300 LIA).









between the Ni and the GaN. Table 3-2 summarizes the near-surface composition AES data. The

increased oxygen concentration evident from this data most likely comes from oxidation of the

out-diffused Ti. The carbon signal in all cases comes from adventitious carbon on the surface.

After annealing at 7500C, the Ni shows significant movement through all of the overlying layers

and even the Zr diffuses out of the boride later, leaving the boron in its original location. Both Zr

and Au diffuse to the interface with GaN. The increased contact intermixing at higher annealing

temperatures did roughen the contact morphology as shown in the SEM pictures of the TLM

contact pads as a function of annealing temperature in Figure 3-15. After 8000C annealing the

morphology is degraded, but is still much smoother than conventional Ni/Au contacts under

these conditions. The darker appearance of the contacts after high temperature anneals is mainly

a result of the outdiffusion of Ti, which then oxidizes.

The same basic trends were observed with the ZrB2/Ti/Au structures. Figure 3-16 shows

the AES surface scans and surface scans of ZrB2/Ti/Au Ohmic contacts on p-GaN as a function

of anneal temperature. The as-deposited sample shows abrupt interfaces but after annealing at

9000C, the Zr shows a very broad distribution (more so than the B). The contacts after annealing

showed the presence of reacted islands, as shown in the elemental maps of Figure 3-17. The

islands contain both Au and Ga and show that the GaN epi layer has begun to dissociate at this

temperature, at least under the contact metallurgy. Note that the areas with Au and Ga overlap,

indicating that they are indeed forming the Au-Ga phases reported in the literature.

3.1.4.3 Gallium Nitride//Tungsten Boride-based contacts

Figure 3-17 shows the specific contact resistivity of the W2B/Ti/Au/p-GaN structure as a

function of annealing temperature, along with the sheet resistance of the p-GaN under the

contacts, extracted from the TLM measurements. The contacts are rectifying below an anneal

temperature of 5000C but transition to Ohmic behavior at this temperature. The specific contact









resolution spectra were acquired to determine the binding energy (i.e. chemical state) and

concentration of specific elements observed in the survey spectra. Charge correction was

performed by using the known position of the C-(C,H) line in the C Is spectra at 284.8 eV. The

SBH was determined from the binding energy of the Ga 3d core level EB and the energy

difference between that core level and the valence band maximum Evc according to

B = EB -Evc .[62,79]

The I-Vs were recorded over the temperature range 25-200C using a probe station and an

Agilent 4145B parameter analyzer. For the relatively high doped p-GaN samples investigated in

this work, TFE over a Schottky barrier is expected to be the dominant mechanism of forward

current flow [56,57,62,63,80]. We fit the forward I-V characteristics to the relation for TFE over

a barrier [81]


JF =0 exp E)(5.1)
l-Eo

where JF is the current density, e is the electronic charge, and Vis the applied voltage. In

Eq. (1), Eo = E00 coth(E00/kBT) is the characteristic energy related to the tunneling probability,

kB and Tbeing the Boltzmann's constant and absolute temperature, respectively. For p-GaN,

Eoo (eV) 7.75 x 10 12O N (c 3) [56,62] NA being the density of acceptors. The saturation

current density Jo is given by

A*T ;T^Eooq( V -f ) 05 e e (OB -f
J= kB cosh(Eoo/kBT) exp kBT E (5.2)

where A*=103.8 Acm-2K-2 is the effective Richardson's constant for p-GaN, [82] OB is the

barrier height, and -(EF-Ev)/e is the difference between the valence band maximum and the

position of the Fermi level. This parameter depends on the doping level, but, as will be shown









be sufficient. The desired flat band diagram for this situation is shown in Figure 2-3 [40]. It is

clear from this diagram that holes may pass easily between the metal and the contact at the

junction.

However, for some semiconductors it can be difficult to find materials systems that produce an

Ohmic contact through thermionic emission, especially for p-type semiconductors. A wide band

gap and large electron affinity, such as with GaN, effectively limits the choices to materials

which have a very high work function. A second approach for formation of Ohmic contacts to

semiconductors utilizes another form of carrier transport: tunneling. This type of transport is

characterized by the carriers tunneling through the potential barrier at the semiconductor-contact

junction. Equation 2-3 describes the specific contact resistance for this type of transport:


P = exp hND (2.3)


The relevant terms to consider in this equation are the barrier height and the doping density, ND

or NA for p-type semiconductors. It is obvious that in order to achieve a low contact resistance

the barrier height must be minimized and the doping density must be maximized. Barrier height

is determined by material choice. Doping density can be determined during growth of the

material, with ion implantation, or use of contacts which can increase doping through

interactions with the semiconductor.

2.3.1.1 Ohmic contacts to p-GaN

High quality, low resistance Ohmic contacts to p-GaN are difficult to achieve. This is due

to several factors: the relative dearth of appropriate metal systems due to the large work function

of GaN, low hole densities due to the difficulty in achieving a high doping concentration during

growth, the presence of compensating hydrogen interstitials, and the presence of compensating











Table 2-4: Properties of the nitrides
Properties TaN TiN ZrN
Melting point (K) 3633 3203 3253

Structure FCC FCC FCC
Thermal expansion 3.6 9.4 8.7
coefficients x 106 (/deg)
Work function (eV) 4.7 3.74 4.6
Lattice constant (A) 4.3 4.32 4.574
Thermal conductivity (W 57.5 19.2 20
m-1 K-1)
Electrical resistivity (tLQ- 15.52 25 13.6
cm)









applications. The promise of solid state lighting is one that grows more important with each

passing year, as the world's insatiable hunger for energy grows while reserves of conventional

fossil fuels struggle to keep up. Because production cannot indefinitely keep up with demand,

focus should and is on increasing the efficiency of consumption. Nearly 10% of world energy

consumption is used for lighting. Most of this is currently done with extremely inefficient

incandescent bulbs. While compact fluorenscent lighting has grown more popular and offers a

substantial improvement in energy efficiency, it cannot compete with that possible with the wide

spread use of LEDs. Further, CFBs present a hazard due to the non-trivial amount of mercury

present in each bulb. Because of this, any large scale use of the bulbs must go hand in hand with

proper disposal of them in order to prevent contamination of the environment. Solid state

lighting does not present this problem.

While many challenges still exist to realizing commercially viable solid state lighting

sources, including significant improvement in the efficiency of the green emitting alloys as well

as possible improvements in the phosphors used to emit the yellow and white light desired, it is

still important to develop improved contact structures in order to allow for longer life times and

operation in environments with higher temperatures. For p-type GaN especially, existing contact

schemes could be improved to offer more stable performance. Failure of GaN-based devices due

to contact failure should be minimized or eliminated.

The goal of this work is to develop improved contact schemes for both electronic and

optoelectronic applications that will offer acceptable specific contact resistances and barrier

heights which will be stable over a wide range of processing temperatures as well as during long

term high temperature aging. Chapter 2 deals with the basic properties of GaN as well as an

overview of the processing and characterization techniques that were utilized during this work.
















-- W2B/Pt/Au
4.5 WB/Pt/Au



4.0 -

-e-

3.5

Highly-doped Lightly-doped

3 .0 I 1I I 1 _
0.08 0.12 0.16 0.20 0.24

(eV)





Figure 5-6: Dependence of the apparent Schottky barrier height on the parameter defined as the
difference between the valence band maximum and the position of the Fermi level.
Low and high









size of the metallic atom in the nitride, with the largest increase seen with the smallest atom Ti

and vice versa for Ta, suggesting the atoms causing the spike are from the nitride. The reason

the smaller atoms cause a more pronounced increase is the greater ease of diffusion into the GaN

as well as a likely higher solid solubility.

Figure 3-2 shows the AES depth profiles for the TaN-based contacts under a variety of

conditions. Upon annealing at intermediate temperatures, most of the Ti has migrated to the

surface and oxidized while Ni and Au are both present at the GaN interface. In addition, the

nitride layer is present at the GaN surface, suggesting that the increase in sheet resistance is due

to incorporation of the metallic atoms from the nitride into the GaN. After annealing at 700 C

and aging at 200 C or annealing at 1000 C, there is a large amount of intermixing between the

metals and the GaN. Note that the upon annealing at 6000C the Ta and N peaks are still

correlated, indicating that TaN has not reacted a great deal with the GaN yet. However, in the

profile of the thermally aged contact there is no longer a correlation between the peaks of the Ta

and N. This is an indication that the nitride is breaking down during the thermal aging. In fact,

the profile of the aged contacts bears a much stronger resemblance to that of those annealed at

much higher temperatures. SEM micrographs are shown in Figure 3-3 for the TaN-based

contacts. The contacts show a smooth morphology upon deposition, with an increasingly rough

surface as anneal temperature increases due to the increased intermixing of the contacts as well

as oxidation of the surface. TiN and ZrN contacts display the same type of profile structure upon

annealing and after thermal aging.

As mentioned previously, the Ni/Au/GaN structure is preferred due to the greater

difference in electronegativities between Au and Ga than between Ni and Ga. By incorporating a

layer on top of the Au, the dynamics of this relationship may change. The electro-negativities of









The AES depth profiles from the as-deposited and 800 and 9000C annealed samples are

shown in Figure 3-12. The as-deposited sample shows sharp interfaces between the metals and

between the Ni and the GaN. After annealing at 8000C, the Ni shows significant movement

through all of the overlying layers and by 950C, most of the contact metallurgy is intermixed

and the Ti is essentially all removed to the surface. There is an accompanying decrease in

abruptness of the TiB2 interfaces with the metals on either side. The TiB2 appears to be a barrier

for Ti diffusion, so this probably excludes formation of TiN phases at the interface of the GaN to

account for the improved contact resistance.

3.1.4.2 Zirconium Boride-based contacts

ZrB2-based contacts were also fabricated. Figure 3-13 shows the specific contact

resistance of the ZrB2/Ti/Au and Ni/Au/ZrB2/ Ti/Au/p-GaN structures as a function of annealing

temperature, along with the sheet resistance of the p-GaN under the contacts, extracted from the

TLM measurements. The contacts are rectifying below an anneal temperature of 750C but

transition to Ohmic behavior for higher temperatures. The Ni/Au/ZrB2/Ti/Au contact resistance

improves with higher anneal temperature, at the expense of poorer morphology. The ZrB2/Ti/Au

contacts do not show low resistance values until 8000C and the value degrades at higher

temperatures. The associated sheet resistance generally decreases with temperature. The specific

contact resistance values obtained with both types of ZrB2-based schemes is comparable to that

achieved on the same samples with Ni/Au metallization annealed at 5000C but the former have

much higher thermal stability, as expected from the high melting temperature of the ZrB2.

Figure 3-14 shows the AES surface scans and depth profiles from the as-deposited

Ni/Au/ZrB2/ Ti/Au/p-GaN sample (left) and from the sample annealed at 7500C (right). Ti, Ni

and Zr are evident on the surface after annealing at 7500C and the concentration increased with

annealing temperature. The as-deposited sample shows sharp interfaces between the metals and














10-4


105 -


6
2 10-6 ....-2000C '
--1500C .
S...... 100 C
10- --- 50oc
-- 250C
10-8
-4.0 -3.5 -3.0 -2.5 -2.0 -1.5 -1.0 -0.5

Voltage (V)



Figure 5-8: Reverse current-voltage characteristic of W2B-based Schottky diodes as a function
of measurement temperature.










Table 3-1: Concentration of elements detected on the as-received surface (in atom%)


Filename Sample ID C(1) 0(1) Ti(1) Au(3)
Sensitivztyfactors [0.076] [0.212] [0.188] [0.049]
051829101 As deposited 36 1 nd 64
051829201 8000C annealed 38 31 19 12
051829301 9000C anneal 27 33 28 12
051829401 950C Anneal A 30 26 33 nd
051829401 950C AnnealB 28 26 35 nd

































2008 Lars Fredrik Voss









data was 4 times lower than that for 40 [tm-diameter diodes, as expected from the surface area

ratio. In addition, the density of acceptors obtained from the corrected C-Vs was 1.8x102 cm-3,

which is very similar to the value determined above from the I-Vs. Table 5-1 summarizes the

calculated barrier heights from XPS, IV with both thermionic and thermionic field emission, and

CV measurements.

Figure 5-8 presents the influence of the measurement temperature on the reverse-bias

current for the W2B-based diodes. All log I vs V curves measured between 25 and 200C exhibit

a linear region at high reverse bias and a steeper variation at lower reverse voltages. In addition,

as the measurement temperature increases, the reverse leakage current also increases. Given the

large barrier height typical for rectifying contacts to p-GaN, one can assume that thermionic

emission over the Schottky barrier makes only a negligible contribution to the reverse-bias

current flow. This assumption is further supported by the fact that all log Ivs Vcurves in Figure

5-8 are parallel, which is clearly inconsistent with the thermionic emission model. On the other

hand, in the tunneling approach, the reverse current density JL is given by the Fowler-Nordheim

tunneling expression J, = CE2 exp(Eb/D), where C=C(OB) and D=D(OB) are parameters

independent of the measurement temperature [91-93]. For tunneling, the reverse-bias current is

thus independent of the measurement temperature, which again is inconsistent with the

experimental data displayed in Figure 5-6. Leakage must therefore originate from other

mechanisms such as generation in the depletion layer or surface leakage. The large band gap of

GaN makes the intrinsic carrier concentration in a depletion region very small, suggesting that

contributions to the reverse-bias leakage from generation in the depletion layer are negligible.

However, given the large number of deep-level defects located at the surface vicinity [88] as a

result of both the high Mg doping [29,85,86] and the sputter-deposition process [29,87], one may









nitrogen vacancies. Despite this, several materials systems have been found to produce Ohmic

behavior. These are generally based on a high work function metal such as Ni, Pd, or Pt with an

overlayer of Au and have been found to achieve specific contact resistances in the range of 1 x

10-4 Q-cm2 for hole concentrations of 1017 cm-3 [44-48]. Ni/Au contacts are the most common.

The Ni serves to remove hydrogen from the near surface region of the p-GaN at elevated

annealing temperatures, as the Ni hydride is thermodynamically favorable. In addition, at these

elevated temperatures Au begins to diffuse in towards the GaN and it has been found that the Au

can form intermediate phases with the Ga in the form of AuGa and AuGa2 [49]. This can serve

to further increase the near surface doping through the formation of Ga vacancies in the GaN

which act as acceptors. However, at these anneal temperatures of> 500C, the Au and Ni may

completely exchange places at the GaN surface [44]. This leads to a severe roughening of the

contact surface and degraded performance and limits the temperature at which anneals may take

place as well as the lifetime of the contacts when operated at high temperature or, equivalently,

high powers. Because one of the most attractive features of GaN is its ability to operate reliably

under these conditions, it is important to find a contact system that will perform better.

2.3.1.1 Ohmic contacts to n-GaN

Ohmic contacts to n-GaN are more studied, better understood, and easier to achieve

compared to those to p-GaN. The most common materials system used is GaN//Ti/Al/X/Au,

where X is a material chosen to act as a diffusion barrier to separate the Al and Au. This is

necessary, as Al and Au will react to form A1Au4 [50]. This phase is viscous at low temperatures

and lateral flow can create problems, especially as the gate/source separation becomes small.

Because of this, there is great interest in the use of high temperature metals. The Ti is present at

the junction in order to form TiN through reaction with the GaN, which creates nitrogen









Table 3-4: Summary of specific contact resistances


Contact Scheme
Ni/Au/W2B/Ti/Au
W2B/Ti/Au
Ni/Au/CrB2/Ti/Au
Ni/Au/TiB2/Ti/Au
Ni/Au/ZrB2/Ti/Au
ZrB2/Ti/Au
Ni/Au/TaN/Ti/Au
Ni/Au/TiN/Ti/Au
Ni/Au/ZrN/Ti/Au
Ni/Au/Ir/Au
TaN/Ti/Au
ZrN/Ti/Au


Psc,minium (~-cm2)
3.5 x 10-5
1.69x 10-3
7x 10-5
1.93 x 10-4
1 x 10-4
1.8x 10-3
2.5 x 10-4
2.45 x 10-4
2x 10-4
2.3 x 10-4
3.7x 10-4
3.1x 10-4


Anneal temperature (C)
1000
800
1000
850
900
800
800
600
600
500
800
600


Failure temperature (C)
>1000
>900
>1000
>900
>900
>900
>1000
>1000
>1000
>500
>1000
>700



























Figure 6-1: Optical micrograph of an as-fabricated MQW-LED. The p-contact at the center of
the diode is 80 tm in diameter.




























-3 -2 -1 0


1 2 3 4


Voltage (V)






Figure 5-7: As-measured and after oxide correction dependence of C2 versus V of Au/Pt/W2B/p-
GaN Schottky diodes. The measurement frequency was set to 1 kHz.


3.5

3.0

2.5

2.0


0~


I I I I


As-measured
0 After oxide correction


0
0




<^-:::-'


0.5


0.0 L
-4


' '


1.5

1.0









Figure 3-8 displays the AES depth profiles for the W2B-based contacts as deposited,

annealed at 600 oC, annealed at 700 C and aged at 350 oC, and annealed at 1000 C. Upon

annealing at 600 oC, the Ni and Au show strong mixing at the GaN interface. However, the

boride barrier prevents most intermixing between the Ni/Au and the Ti/Au overlayer. The boride

has minimal reaction with the GaN surface and the surface had excellent morphology at 6000C.

Note that the boride peak remains strongly correlated, indicating the W and B atoms remain

together. This is the case even at 10000C anneals and after long term thermal aging at 350 C,

indicating that the boride is not chemically active in these contacts. At higher temperatures, the

AES profiles were not entirely reliable due to non-uniformity depending on where the profile

was taken. However, the IV characteristics were still Ohmic and the surface still had reasonable

morphology. At temperatures of 10000C and higher, the contact layers became totally

intermixed, although the peaks of W and B still correspond to one another. Similar results were

obtained with CrB2-based contacts.

3.1.3.2 Summary

These boride-based contacts display excellent thermal stability both upon high temperature

anneals and during long term thermal aging at very high temperature. This is due to the fact that

there is little intermixing between the boride and the GaN itself, due to the large change in Gibbs

Free Energy that would be necessary for any chemical reaction to occur. This is supported by

the depth profiles obtained from AES, which show that the boride has not broken down during

experiments.

3.1.4 Contact Resistance for Other Boride-based Contacts

3.1.4.1 Titanium boride-based contacts

Contacts with a TiB2 diffusion barrier for enhancing the thermal stability of Ni/Au Ohmic

contacts on p-GaN were fabricated. Figure 3-9 shows the specific contact resistivity of the















Temperature (C)


400 300 200


1.5


2.0


2.5


Figure 2-1 Intrinsic carrier concentration of GaN, GaAs, and Si


10201


100


25 0


1010




100


f?
u

o


om
u



ci


I I I


Si


GaAs -


GaN


10-20


3.0


3.5


1000 / T (K1)


4.0


' '


I
I


.









CHAPTER 4
OHMIC CONTACTS TO N-GAN

In this chapter, we report results on Ohmic contacts based on these materials to n-GaN.

Specific contact resistance has been studied as a function of annealing temperature, annealing

time, and thermal aging for TaN, TiN, and ZrN diffusion barriers. All nitrides show comparable

specific contact resistance to conventional Ti/Al/Pt/Au Ohmic contacts and display similar

stability when aged at 3500C in air.

4.1 Experiment

The samples consisted of 3 |tm thick Si-doped GaN grown by Metal Organic Chemical

Vapor Deposition on c-plane A1203 substrates. From Hall measurements, the electron

concentration was -3 x 107 cm3. Mesas 0.6 [m deep were formed by C12/Ar inductively

coupled plasma etching to provide for electrical isolation of the contact pads and to prevent

current spreading. Deposition of a Ti (200 A)/Al (800 A)/X (500 A)/Ti (200 A)/Au (800 A)

metallization was performed by sputtering, where X is TaN, TiN, or ZrN. Sputtering was

performed using a Ar plasma-assisted system at a pressure of 5 mTorr and rf (13.56 MHz)

powers of 150-350W. For comparison, a Ti (200 A)/Al (800 A)/Pt (400 A)/Au (800 A) was

deposited by e-beam evaporation. The contacts were patterned by lift off and annealed at 400-

10000C for 30-240s in a flowing N2 ambient in a rapid thermal annealing (RTA) furnace.

Specific contact resistances were obtained using the linear transmission line (L-TLM)

method. 100 |tm x 100 |tm pads with spacings of 80, 40, 20, 10, and 5 as well as 76, 36, 16, 6,

and 1 |tm were used. The total resistance is given by

RT= 2R + R,(L/W) (4.1)









The rest of this dissertation is devoted to the experiments and analysis. First, Ohmic contacts to

p-GaN using a variety of high temperature materials are discussed. The next section deals with

Ohmic contacts to n-GaN. A section on Schottky contacts to p-GaN follows. The final chapter

discusses the use of the Ohmic contacts to p-GaN in light emitting diodes.









Table 2-1: Bulk GaN properties


Property
Lattice parameters at 300 K (nm)

Density (g-cm-3)
Stable phase at 300K
Melting point (C)
Thermal conductivity (W/cm-K)
Linear thermal expansion coefficient

Static dielectric constant
Refractive index
Energy bandgap (eV)
Exciton binding energy (meV)
Electron effective mass
Hole effective mass


Value
ao = 0.3189 nm
Co = 0.5185 nm
6.095 g-cm-3
Wurtzite
2500
1.3, 2.2 +/- 0.2 for thick, free-standing GaN
Along ao = 5.59x10-6 K-
Along co = 7.75x10-6 K-
8.9
2.67 at 3.38 eV
Direct, 3.45
26
0.20
0.59









diffusion barrier, such as Ni. Figure 4-6 shows a profile for a Ti/Al/Ni/Au contact annealed at

500C. From the profiles, the nitrides appear to be breaking down somewhat, with TiN the least

affected, as the location of the Zr and Ta peaks do not correspond precisely to those of the N.

Little change in the contact structure is seen with long term aging even after greater than 20 days.

Because there is only minimal intermixing of the layers, little of the viscous A1Au4 phase has

formed compared to conventional contact schemes.

Figure 4-7 shows the dependence upon anneal time at 8000C. Upon annealing for up to

180 seconds, an increase in contact resistance is seen in each of the contacts. The standard

Ti/Al/Pt/Au contact exhibit the largest increase in contact resistance, while the contacts

containing TaN and TiN display more moderate increases. The ZrN-based contact displays only

a slight increase in contact resistance even upon annealing for 180 seconds.

Figure 4-8 displays the specific contact resistance as a function of long term thermal aging.

Samples placed on a hot plate at 3500C display all display a gradual increase in the specific

contact resistance throughout the period. Table 4-1 shows the percentage increase in specific

contact resistance for each scheme. Compared to the Ti/Al/Pt/Au contacts, the ZrN and TaN

contacts display better stability, with the specific contact resistance increasing only about 100%.

The TiN contacts displayed a larger increase.

4.3 Conclusions

Ohmic contacts to n-GaN were fabricated using a Ti/Al/X/Ti/Au scheme, where X is either

TaN, TiN, or ZrN. These contacts display similar specific contact resistances to the conventional

Ti/Al/Pt/Au contacts deposited on the same wafer, -10-4 ohm-cm2 over a wide range of

annealing temperatures. In addition, the TaN and ZrN contacts display only about half the

increase in specific contact resistance that is seen in the Ti/Al/Pt/Au contacts. The AES depth

profiles confirm that little intermixing has occurred during the aging process. Because of this,





















































Figure 3-15: Scanning electron microscopy images of Ni/Au/TiB2/Ti/Au contact pads on p-GaN
as-deposited (top) or after annealing at either 750 (center) or 8000C (bottom).









111. L. F. Voss, R. Khanna, S. J. Pearton, F. Ren, and I. I. Kravchenko, Appl. Surf. Sci. 253,
1255 (2006).

112. W. Zagozdzon-Wosik, D. Radhakrishnan, C. Dame, I. Rusakova, P. van der Heide, Z.-H.
Zhang, J. Bennett, L. Trombetta and P. Majhi, Rev. Adv. Mater. Sci. 8, 185 (2004).

113. Y. D. Qi, H. Liang, D. Wang, Z. D. Lu, W. Tang and K. M. Lau, Appl. Phys. Lett. 86,
101903 (2005).

114. X. Guo and E. F. Schubert, Appl. Phys. Lett. 78, 3337 (2001).

115. A. Chitnis, J. Sun, V. Mandavilli, R. Pachipulusu, S. Wu, M. Gaevski, V.Adivarahan, J. P.
Zhang, M. AsifKhan, A. Sarua and M. Kuball, Appl. Phys. Lett. 81, 3491 (2002).

116. P. G. Eliseev, P. Perlin, J. Furioli, P. Sartori, J. Mu, and M. Osinski, J. Electron. Mater. 26,
311 (1997).

117. G. Franssen, E. Litwin-Staszewska, R. Piotrzkowski, T. Suski, and P. Perlin, J. Appl. Phys.
94, 6122 (2003).

118. J. M. Shah, Y. L. Li, T. Gessmann and E. F. Schubert, J. Appl. Phys. 94, 2623 (2003).

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









74. B. Luo, F. Ren, R. C. Fitch, J. K. Gillespie, T. Jenkins, J. Sewell, D. Via, A. Crespo, A. G.
Baca, R. D. Briggs, D. Gotthold, R. Birkhahn, B. Peres, and S. J. Pearton, Appl. Phys. Lett.
82, 3910 (2003).

75. R. Khanna, S. J. Pearton, F. Ren, and I. Kravchenko, J. Electrochem. Soc. 11, G804 (2005).

76. R. Khanna, S. J. Pearton, F. Ren, I. Kravchenko, C.J. Kao and G.C. Chi, Appl. Phys. Lett.
87, 052110 (2005).

77. R. Khanna, S. J. Pearton, F. Ren, and I. Kravchenko, Appl. Surf. Sci. 252, 5814 (2006).

78. T. N. Oder, P. Martin, J. Y. Lin, H. X. Jiang, J. R. Williams, and T. Isaacs-Smith, Appl.
Phys. Lett. 88, 183505 (2006).

79. R. T. Tung, Mater. Sci. Rep. R35, 1 (2001).

80. L. Stafford, L. F. Voss, S. J. Pearton, J. -J. Chen, and F. Ren, submitted.

81. Michael Shur, Physics of Semiconductor Devices (Prentice-Hall, Engelwood Cliffs, NJ,
1990).

82. C. Merz, M. Kunzer, U. Kaufmann, I. Akasaki, and H. Amano, Semicond. Sci. Technol. 11,
712(1996).

83. H. Nakayama, P. Hacke, M.R.H. Khan, T. Dtechprohm, K. Hiramatsu, and N. Sawaki, Jpn. J.
Appl. Phys. 35, L282 (1996).

84. J. R. Waldrop and R. W. Grant, Appl. Phys. Lett. 68, 2879 (1996).

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

86. J. S. Kwak, O. H. Nam, and Y. Park, J. Appl. Phys. 95, 5917 (2004).

87. M.W. Cole, F. Ren, and S. J. Pearton, J. Electrochem. Soc. 144, 275 (1997).

88. K. Shiojima, S. Sugitani, and S. Sakai, Appl. Surf. Sci. 190, 318 (2002).

89. T. Okumura, Jpn. J. Appl. Phys. 24, L437 (1985).

90. M. Passlack, N. E. J. Hunt, E. F. Schubert, G. J. Zydzik, M. Hong, J. P. Mannaerts, R. L.
Opila, and R. J. Fischer, Appl. Phys. Lett. 64, 2715 (1994).

91. M. Lenzlinger and E. H. Snow, J. Appl. Phys. 40, 278 (1969).

92. Z. Weinberg, J. Appl. Phys. 53, 5052 (1982).

93. H. Zhang, E. J. Miller, and E. T. Yu, J. Appl. Phys. 99, 023703 (2006).









shown in Figure 6-4. While the brightness of MQW-LEDs with Ni/Au Ohmic contacts at 10 V

(I= 80 pA) was low, optical images of aged MQW-LEDs with boride-based Ohmic contacts

show much stronger emission, even at a lower applied voltage of 4.5 V (I= 300 pA).

Several mechanisms can be invoked to explain the drastic degradation of Ni/Au contacts

upon aging, including formation of an islanded contact morphology, formation of reactions with

the GaN resulting in a modification of the doping profile, and excessive intermixing of Ni and

Au leading to oxidation on the rough Ni surface and an increase of the series resistance

[44,109,119]. Although the dominant degradation mechanism remains unknown' the use of TiB2/Ti/Au

or Ir/Au overlayers o" the Ni/Au-baed Ohmic contacts clearly improves the long-term thermal stability of

InGaN/GaN MQW-LEDs.

6.4 Conclusions

In conclusion, TiB2 and Ir diffusion barriers in p-Ohmic contacts on GaN LEDs produced

less change in turn-on voltage, leakage current, and output power after long-term annealing at

2000C and 3500C for 45 days. These schemes look promising for high temperature applications

where improved stability over Ni/Au is mandatory.






Table 6-1: Influence of long-term aging at 2000C and 350C on the turn-on voltage and reverse
current of InGaN/GaN MQW-LEDs.


Contact to p-GaN Turn-on voltage (V) Reverse current ( -5V (A)
Day 0 Day 10 Day 45 Day 0 Day 10 Day 45
(@ 2000C ,@ 3500C (@ 2000C ,@ 3500C
Ni/Au 3.30.3 4.90.4 9.1+0.5 (21) x10-6 (62)x10-9 (53) x105
Ni/Au/TiB2/Ti/Au 3.50.2 4.1+0.5 4.00.2 (52)x 10-6 (21) x10-9 (2+1)x10-8
Ni/Au/Ir/Au 3.30.2 4.30.3 3.90.3 (3l)x 10-6 (21)x10-9 (32)x10-7









Table 5-1: Comparison of different barrier height calculations

XPS TE TFE
B = 2.7 eV B = 1.2 eV B = 3.8 e


n =4.4


V


Eo= 110 meV


C-V
= 4.1 eV



























0 100 200 300 400 500 600
Anneal temperature (oC)


700 800


Figure 5-9: Influence of the annealing temperature on the breakdown voltage.













6000


500 1000 1500 2000
Kinetic Energy (eV)


100
90 Au Ga
80
70
60
50
40
30
20
10 0

0 500 1000 1500
Sputter Depth (A)


-10000



100
90
gso
-80
o70
S60
550
040
.o 30
E


10
0


Same #6
Annealed at


MgAI


Atomic %
Cl 52.9
01 16.3
Zr2 12.0
Au3 8.9
Mg2 4.2
AI2 4.0
N1 1.7


500 1000 1500 2000
Kinetic Energy (eV)


500 1000 1500 2000
Sputter Depth (A)


Figure 3-16: Surface scans and depth profiles of ZrB2/Ti/Au Ohmic contacts on p-GaN as a
function of anneal temperature.
















Au Ir Au Ni (a)
? 80-
0
SGGa
S60-

o 40- N



o
0 1000 2000 3000
Depth (A)




100
A r (b)
Au
c 80-
0
Ga
C 60-
a Au
o 40- N

2 2Ni
E 20-
o
0
0 1000 2000 3000 4000 5000
Depth (A)




100-
F (c)
80-
o
C 60- Ir
a) /Ga
o 40- 0

E 20- Ni N
0

0 1000 2000 3000 4000
Depth (A)


Figure 3-23: Depth profiles for Ni/Au/Ir/Au contacts (a) annealed at 300 C (b) annealed at 500
C and (c) annealed at 700 C.












TaN TiN ZrN


TaN-based as deposited TiN-based as deposited










TaN-based 600C anneal TiN-based 600C anneal





S.0


ZrN-based as deposited










ZrN-based 600C anneal


TaN-based aged TiN-based aged ZrN-based aged

Figure 4-2: Scanning electron microscopy images of annealed contacts









structure Titanium / Aluminum / X / Titanium / Gold were fabricated using standard

semiconductor processing procedures, where X is either TaN, TiN, or ZrN. For comparison,

conventional contacts of the form Titanium / Aluminum / Platinum / Gold were fabricated as

well. The contacts were then subjected to a range of anneal temperatures between 500C and

1000C in a flowing N2 ambient for 60 s, as well as anneals at 8000C for up to 180 s, in order to

determine if the nitrides would allow for an increased thermal budget. Contacts were also aged

on a hot plate at 3500C and aged for a period of 24 days. Current-voltage measurements, Auger

Electron Spectroscopy, and Scanning Electron Microscopy were used to characterize the

contacts. Current-voltage measurements of the nitride contacts did not show any improvement

over the performance offered by the conventional Pt-barrier contacts in any of the experiments.

However, AES depth profiles revealed that the nitride based contacts displayed far less

intermixing than Ni-barrier contacts, even when the nitrides were annealed at higher

temperatures. Further, the profiles of thermally aged nitride contacts displayed no noticeable

difference from unaged contacts. Thus, while electrical performance of these contacts was

essentially unchanged from that of the conventional ones, they do offer the benefit of

dramatically reducing the amount of intermixing between layers even after being subjected to

harsh long term aging. This is significant, as one of the sources of failure for small gate width

GaN devices is lateral flow arising due to the mixing of the Al underlayer and Au overlayer,

leading to the formation of the viscous A1Au4 phase. This leads to short circuiting and thus

failure of devices.

Chapter five dealt with the fabrication of Schottky contacts to p-GaN. For this purpose,

W2B and W2B5 sputter targets were chosen. Nitrides were ignored due to the severe intermixing

expected between them and the GaN. It is also expected Ir would diffuse a great deal into the














E x10" x -3 -- *100

\"; -80 a



S-*- TiN
-"--TaN -60 =

1x104 -* N *
ZrN u
-v- TaN "40 4)
-0- TiN
<- ZrN \
0 /20


S1105 ,--- 0
500 600 700 800 900 1000
CO Anneal Temperature (OC)


Figure 3-1: Specific contact resistance and sheet resistance under the contact ofNi/Au/ X/ Ti/Au
contacts as a function of anneal temperature.









Table 4-1: Percent change in specific contact resistance during thermal aging
Contact Percent Change

Ti/Al/Pt/Au 236%
Ti/Al/TaN/Ti/Au 107%
Ti/Al/TiN/Ti/Au 330%
Ti/Al/ZrN/Ti/Au 091%













0
E lx10--

C


80 4




0

-5
0
0.
n,)







1x10"-
Q.
(0


60 80 100 120 140 160 180

Anneal Time (s)


Figure 4-7: Specific contact resistance as a function of anneal time.


-- Ti/Al/Pt/Au
-*- Ti/Al/TaN/Ti/Au
T Ti/AI/TiN/Ti/Au
-v- Ti/AI/ZrN/Ti/Au















co
0


c0
0
0


o
=l


100


80

60


40


20


0


0 1000 2000 3000
Depth (A)




Figure 4-6: Depth profile of Ti/Al/Ni/Au contact annealed at 5000C.









ACKNOWLEDGMENTS

First and foremost I would like to thank my advisor Prof. Stephen J. Pearton for his

guidance and for all of the support and opportunities he provided. I would also like to thank my

other supervisory committee members, Fan Ren, Cammy Abernathy, David Norton, and Rajiv

Singh, for their help and their time.

Thanks go to Dr. Patrick M. Lenahan for providing me with my first experience working in

the semiconductor field while I was an undergraduate at Penn State as well as all of his advice

and encouragement.

I thank the members of the Pearton, Ren, and Abernathy research groups with whom I

have had the opportunity to work, including Luc Stafford, Kelly Ip, Jon Wright, Wantae Lim,

Hungta Wang, Sam Kang, Travis Anderson, Soohwan Jang, Brent Gila, Jerry Thaler, Jennifer

Hite, Mark Hlad and many, many more. I would also like to thank Ivan Kravchenko for his

support in the UF Nanofabrication Facility.

Thanks also go to all the people at Sandia National Laboratories who gave me the

opportunity to work there and with them for two enjoyable summers. I want to say thank you to

my mentors, Randy J. Shul, Albert G. Baca, and JeffE. Stevens, my managers, Charles Sullivan

and Dale Hetherington, as well as all of the people I had the pleasure to work with while at

Sandia including Carlos Sanchez, David Torres, Melissa Cavaliere, Karen Cross, Mark

Overberg, Michael Cich, and many others.

Thanks also go to all of my family and friends as well.











10-2


1-3 .- ---
10 -



10-4



10 -



10-6 ,
0.0 0.5


1.0 1.5 2.0 2.5 3.0 3.5 4.0


Voltage (V)



Figure 5-2: Forward current-voltage characteristic ofW2B-based Schottky diodes as a function
of measurement temperature.









LIST OF TABLES


Table page

2-1 B ulk G aN properties .......................... .......................... .. .. .. ......... ........ 33

2-2 Properties of comm on semiconductors ..................................................... ... .......... 34

2-3 P properties of the b rides ..................................................................... .... .....................35

2-4 P properties of the nitrides ............................................................................. .............. 36

2 -5 P ro p erties o f Ir ..................................................................................... ........................3 7

3-1 Concentration of elements detected on the as-received surface (in atom%) .....................60

3-2 Concentration of elements detected on the as-received surfaces (in atom%) ....................61

3-3 Concentration of elements detected on the as-received surfaces (in atom%) ....................62

3-4 Sum m ary of specific contact resistances ........................................ ....................... 63

4-1 Percent change in specific contact resistance during thermal aging..............................94

5-1 Comparison of different barrier height calculations ........................................................ 112

6-1 Influence of long-term aging at 2000C and 350C on the turn-on voltage and reverse
current of InGaN/GaN MQW-LEDs. ................................... ............... 127











10-3


10-4


10-5


10-6


10.7
0


I W B/Pt/Au |I

12 /





S As-deposited
S -- 3000C anneal
/...... 6000C anneal

1 2 3 4 5 6
Voltage (V)


0 1 2 3 4 5 6
Voltage (V)




Figure 5-3: Forward current-voltage characteristic of W2B-based (top) and W2B5-based (bottom)
Schottky diodes as a function of annealing.


10-3


10-4


10-5


10-6


W2B/Pt/Au








;. /r As-deposited
S 3000C anneal
.- --A- o 6000C anneal









GaN at elevated annealing temperatures, also making it unsuitable for use as a Schottky contact.

Contacts were fabricated with the scheme X / Platinum / Gold, where X is the boride. Platinum

was chosen as the adhesive layer instead of Titanium in order to eliminate formation of TiNx

phases. Current-voltage measurements were used to evaluate the barrier height of the contacts.

Because IV curves measured at different temperatures were parallel, it was determined that the

current transport mechanism present was thermionic field emission. Unphysical barrier heights

of approximately 4 eV were observed for anneals up to 7000C, with good stability. A slight

decrease is seen at increased annealing temperatures. This likely was the result of either an

increased near surface defect concentration due to sputtering or due to incorporation of an

interfacial oxygen layer between the GaN and the contact. X-ray Photoelectron Spectroscopy

measurements of thin boride layers on GaN reveal a true barrier height of 2.85 eV, in close

agreement with that predicted by the Schottky-Mott model. Capacitance-voltage measurements

confirm the large barrier height in agreement with the IV results, but if an interfacial layer of 1-2

monolayers is included the corrected barrier height would be in agreement with XPS

measurements.

Chapter six deals with the application of boride and Ir-based Ohmic contacts to p-GaN for

light emitting diodes. Contacts to LEDs were fabricated with the structures Nickel / Gold,

Nickel / Gold / Titanium Boride / Titanium / Gold, and Nickel / Gold / Iridium / Gold. They

were then annealed at 5000C, in order to prevent damage of the LED structure as well as metal

spiking into the active layers. Initial IV and photoluminescence measurements revealed all

contacts to have similar properties and performance. Contacts were then aged initially at 2000C,

after which all displayed a slight degradation of performance. After further aging at 3500C, the









drop between pads. IV measurements will be the main technique employed to determine the

suitability of both Ohmic and Schottky contacts. The IV curves allow for the extraction of

contact resistance for the former and barrier height and reverse breakdown voltage for the latter.

IV measurements can be performed using either a circular or linear transmission line method.

The CTLM is useful for materials which are difficult to etch and therefore may demonstrate

problems with current spreading. In this study, linear TLM patterns will be used, as shown in

Figure 2-6. Data from these curves allow for the creation of a plot such as that shown in Figure

2-7. Characterization of the pads is then performed using the relation


R, =2Rc + Rs


Pc =RW2) (2.8a,b)



Figure 2-8 shows a schematic for Schottky contacts. Schottky pads are formed using an

inner Schottky contact with an outer Ohmic contact.

2.5.2 Capacitance-Voltage

Capacitance-voltage (CV) measurements will be employed on Schottky contacts in order

to further confirm results of IV measurements as well as to provide a deeper understanding of the

barrier. If an interstitial layer exists between the contact and the p-GaN, utilizing a CV

measurement can reveal this. This is a distinct possibility that could arise either from an

interstitial oxide layer forming or from defects at the surface possibly caused by sputtering

damage. Measurements were performed using an Agilent 4284A precision LCR meter in the

parallel mode. The modulation frequency was set to 1 kHz. Barrier height is determined using









performed at a range of temperatures from 300-10000C in order to determine how the properties

of the contacts are affected at a variety of annealing temperatures.

Following the development of these contacts, devices will be fabricated. While p-GaN is

unsuitable for use in HEMTs due to its low hole mobility, it is necessary for use in LEDs, LDs,

and HBTs. LEDs will be fabricated on wafers with the structure shown in Figure 2-5. These

will be compared to devices using more traditional metal schemes such as Ni/Au. They will then

be thermally aged in order to determine the performance relative to Ni/Au when maintained at an

elevated temperature.

2.5 Characterization Techniques

A variety of techniques will be utilized in order to determine device characteristics as well

as to observe the physical and chemical structure of the contacts. The techniques to be employed

are

1. Current-voltage measurements

2. Capacitance-voltage measurements

3. X-ray Photoelectron Spectroscopy

4. Auger Electron Spectroscopy

5. Scanning Electron Microscopy

2.5.1 Current-Voltage

Current-voltage (IV) measurements will be performed using an Agilent 4156

Semiconductor Parameter Analyzer. Two probe measurements will be performed when possible.

However, if problems which make it impossible to extract contact resistance arise due to high

sheet resistance and the resistance of the probes, a four probe measurement technique will be

employed. Using this method, current will be injected with two outer probes while the voltage

drop is measured between two inner probes. This allows for measurement of only the voltage

















090 Au3 1s8

80


Au3 1s8


rm Concentration (%)


6q
Gal Is7
50

40

30 \

20

10


0 500 1000 1500 2000 2500 3000
Sputter Depth (A)


oncertrdton (%)

12 Is1

Ai3Js4 / s


Ga ls2


M ls2


500 1000
Sputter Depth (A)


Doncertraton (%)


50 -Gl 1s7

Au3 1s8


30 T2 ls5 N1 s3




01 IQ

0 500 1000 1500 2000
Sputter Depth (A)


Figure 3-12: Depth profiles of Ni/Au/TiB2/Ti/Au Ohmic contacts on p-GaN as a function of

anneal temperature. The as-deposited sample is at top, that annealed at 800 pC at

center, and that at 900 oC at bottom.









2.1.2 Electronic Properties

The properties GaN possesses are advantageous for a variety of reasons. They, as well as

those of other common semiconductors, are displayed in Table 2-2 [40,41,42]. It has a high

breakdown field, greater than 50 times that of Si or GaAs, which allows for its use in high power

electronic applications. The high electron mobility for its 2D electron gas and saturation velocity

allows its use in high speed electronics. In addition, heterostructures such as AlGaN/GaN allow

for the manufacture of interesting and high speed devices such as HEMTs. Recent work

suggests that the 1.5 W/cm may be a lower limit for the thermal conductivity of GaN. GaN also

shows excellent resistance to irradiation.

2.1.3 Crystal Structure

The thermodynamically stable phase of GaN is the hexagonal wurtzite structure, referred

to as the a-phase. In addition, a metastable P-phase with a zinc blende structure exists. These

two phases only differ in the stacking sequence of Ga and N; their coexistence in epitaxial layers

is possible due to stacking faults [43]. The lattice parameters are a = 3.189 A and c = 5.185 A.

The wurtzite structure of Ga-face and N-face GaN is shown in Figure 2-2.

2.2 Properties of the Contact Materials to be Studied

2.2.1 Borides

The family ofborides to be studied includes the refractory compounds W2B, W2B5, CrB2, TiB2,

and ZrB2. Previous work has shown the promise of using refractory materials such as W as

contacts to p-GaN. These borides also possess many of the requisite characteristics for use as

electrical contacts for semiconductor devices. Table 2-3 summarizes known properties of the

borides to be studied. Each of the borides possesses a high melting temperature, good thermal

and electrical conductivities, high work functions, and large heats of formation. Each of these

properties is important when considering a material for use in electrical contacts. The high









these values are unphysical, being larger than the GaN band gap, and are inconsistent with both

our XPS measurements and the predictions of the Schottky-Mott model. One may argue that this

discrepancy results from the value of used in our calculations. However, it is shown in Figure

5-6 that over the wide range of values investigated, this parameter only slightly influences the

SBH. The discrepancy between XPS and I-Vmeasurements is more likely related to the presence

of an interfacial defect layer or due to the presence of an interfacial oxide layer. Indeed, it was

shown by Shiojima et al. [88] using high-temperature isothermal capacitance transient

spectroscopy that the defects induced by the high Mg doping in p-GaN are essentially located at

the surface vicinity and that this region acts as a series capacitance [88,89]. Therefore, when such

defect or oxide layer is present, the carriers have to tunnel through an additional barrier, thus

resulting in an apparent SBH given by 0, = 0 + A0 where is the true SBH between the

contact metal and p-GaN and where Ao is the increment of the SBH due to the presence of the

thin defect or oxide layer. Our separate capacitance-voltage measurements (C-V) have indeed

shown that by accounting for such defect layer, the SBH obtained from the C-V characteristics

becomes similar to that determined by XPS [80]. We have calculated a corrected junction

capacitance Coorr for W2B/p-GaN from the equivalent circuit using the simple relation

(1/Ccor) = (1/Cm) -(1/C,), where Cm and Cox are the as-measured and oxide capacitances

respectively. From our XPS measurements, the true SBH for W2B/p-GaN is O 2.7 eV. Using

our C-V measurements, Figure 5-7 shows that the value of Co required to obtain a SBH similar

to that determined by XPS is Cox-0.70 nF. Assuming a contact area of -2x109 m2 and a relative

permittivity of -10 for the oxide layer [90], we estimate an oxide layer thickness of -0.25 nm,

i.e. about one monolayer. For 20 [tm-diameter diodes, the value of Co required to fit the XPS









The I-V characteristics as a function of the measurement temperature for the W2B-based

diodes are shown in Figure 5-2. All log Ivs Vcurves measured between 25 and 2000C exhibit a

linear region at low forward bias (up to -2.0 V) and a less steep region at higher voltages.

Furthermore, all log Ivs Vcurves are parallel. As noted in refs. [56,57,62,63] for p-GaN samples

with relatively high doping levels (>1017 cm-3), this latter result is inconsistent with the

thermionic emission model but typical for carrier transport with a dominant tunneling

component. Note that Shirojima et al. [61] have reported accurate determination of the SBH

using the TE model. However, the acceptor concentration in their p-GaN samples was relatively

low (~1016 cm-3). In our case, the Mg concentration is expected to be in the 1019 cm-3 range, thus

confirming that the forward current cannot be analyzed using the thermionic emission model.

Figure 5-3 displays the I-V characteristics obtained from the diodes annealed at different

temperatures. The extracted characteristic energy related to the tunneling probability as a

function of annealing temperature is shown in Figure 5-4. For W2B, the value of Eo is seen to

remain stable in the 120 meV range and then increases beyond -5000C. This value of Eo

corresponds to an acceptor density of about 2x1020 cm-3, which is higher than that expected from

the Mg concentration alone (Eo~25meV for NA-1019 cm-3). This can probably be attributed to the

presence of acceptor-like deep level defects induced by both the high Mg doping [63,85,86] and

the sputter-deposition process [29,87]. In the case of W2B5, a slight decrease of Eo is observed at

intermediate annealing temperature which can probably be attributed to the annealing of sputter-

damage [29,87].

The influence of annealing on the SBH derived from the I-V measurements of W2B/Pt/Au

and W2B5/Pt/Au contacts is presented in Figure 5-5. Over the whole range of temperatures

investigated, the SBH only slightly decreases with increasing annealing temperature. However,









resistivity improves steadily with temperature and a minimum of 1.7 x10-3 Q.cm-2 was obtained

after annealing at 800 OC for 60 sees. Higher annealing temperatures produced sharp increases in

the sheet resistance of the GaN and irreproducible contact properties. The mechanism for the

increase in sheet resistance may include loss of N (producing nitrogen vacancy-related donors)

or a reduction in the soluble Mg concentration. However the contact morphology as determined

by SEM was similar over the entire annealing range used here. This specific contact resistance is

comparable to that achieved on the same samples with Ni/Au metallization annealed at 5000C.

Figure 3-18 shows the AES surface scans from the as-deposited sample and those

annealed at 500, 700 and 8000C.Ti is evident on the surface after annealing at 5000C and the

concentration increases with annealing temperature. The increased oxygen concentration evident

from the summary of the data in Table 3-3 most likely comes from oxidation of this out-diffused

Ti. The carbon signal in all cases comes from adventitious carbon on the surface.

The AES depth profiles from the as-deposited and annealed samples are shown in Figure

3-19. The as-deposited sample shows sharp interfaces between the metals and between the W2B

and the GaN. After annealing at 5000C, the Ti shows significant movement through the Au layer

and by 800C is essentially all removed to the surface. There is an accompanying decrease in

abruptness of the W2B/GaN interface, which suggests the onset of reactions that are responsible

for the improved contact resistance. Since the W2B appears to be a barrier for Ti diffusion, this

probably excludes formation of TiN phases. Cole et al.[29] reported the formation of the P-W2N

and W-N interfacial phases after annealing sputtered W on GaN [68]. The former appeared after

anneals at 6000C and the latter after 10000C. That work demonstrated the importance of these

interfacial phases in the resulting contact properties. The specific contact resistance obtained for

the W2B-based contacts is lower than reported previously for pure W on p-GaN [69], likely due











Metal p-Semiconductor


Figure 2-3: Flat band diagram for a p-type Ohmic contact












MARKET FORECAST FOR GAN-BASED




Revenue


5-
0 Electr m s

O0ptical


3 -









2001 2003 2005 2007 20D9


Source: Strategies Unlimited


Figure 1-1 Market forecast for GaN-based devices









2.4 Experiments

The goal of this work is to find suitable contact systems for use at high temperatures on n-

and p-GaN. Towards this end, the previously discussed materials will be investigated in order to

determine which of them allow good quality Ohmic and Schottky contacts. In addition, a study

will be made of devices which use p-GaN, mainly LEDs. The contacts will be aged at high

temperatures and compared with more conventional schemes. A variety of techniques will be

used to characterize the contacts and devices in order to determine their performance and to

better understand what is happening.

The main experiments that will be done are the processing of the Ohmic and Schottky

contacts for electrical characterization. The following schemes will be produced and analyzed:

1. GaN//Ni/Au/X/Ti/Au for p-type Ohmic contacts, where X indicates the refractory material to
be examined

2. GaN//X/Ti/Au for p-type Ohmic contacts

3. GaN//Ni/Ir/Au for p-type Ohmic contacts

4. GaN//Ti/Au/X/Ti/Au for n-type Ohmic contacts

5. GaN//X/Pt/Au for p-type Schottky contacts

In case (1), the Ni/Au should act as an Ohmic contact while the boride layer provides a

diffusion barrier to prevent breakdown of the metallization at elevated temperatures. The Ti

layer is present in order to promote adhesion of the Au to the rest of the contact. The Au

overlayer is to prevent oxidation and act as a current spreading layer. In case (2), it is hoped that

at least some of the borides may provide an Ohmic contact on p-GaN. Case (3) may provide

some insight into the importance of Au at the GaN surface for these types of contacts. Case (4)

is based on standard Ohmic contacts to n-GaN, with a nitride used as the diffusion barrier. Case









increase and relatively low output power can probably be attributed to enhanced current

crowding and self-heating effects [114,115]. These effects are expected to be particularly

important for the LED structures investigated in this work due to the high density of dislocations

in InGaN and GaN epilayers grown on sapphire as well as the low thermal conductivity of

sapphire [99-101]. As the dislocation density or LED junction temperature increases,

confinement in the MQW becomes less efficient, leading to premature saturation of the emission

power.

I-V characteristics from MQW-LEDs with Ni/Au, Ni/Au/TiB2/Ti/Au, and Ni/Au/Ir/Au

Ohmic contacts to the p-GaN layer are shown in Figure 6-3. For all diodes, the ideality factor

was >2, consistent with other data reported previously [95-101,116,117]. There is still no

agreement whether such high values result from tunnelling [101,117], recombination in the

depletion layer [117], or rectifying junctions present in heterostructure diodes [118]. The turn-on

voltage (voltage at which the current reaches 0.05 mA) and reverse leakage current (measured at

-5V) are given in Table 6-1. The uncertainty on each value was estimated from the deviation

between different devices on the same wafer. For as-fabricated devices, the turn-on voltage was

3.2-3.6V and the reverse leakage current was ~10-6 A. Such high leakage currents are commonly

observed in GaN-based LEDs grown on sapphire substrates and essentially result from the high

density of dislocations in such samples [101]. After 10 days aging at 2000C, an increase of the

turn-on voltage to -4.1 V and a decrease of the leakage current to -10-9 was observed for all

devices. After aging at 3500C for an additional 30 days, MQW-LEDs with TiB2- and Ir-based

Ohmic contacts show turn-on voltage and leakage current similar to those in the as-deposited

state whereas there was serious degradation in LEDs fabricated with Ni/Au contacts. This

reduction of injected current at a given forward voltage had a large impact on the EL output as





















4U- /
Ta Ti Ga
20-


0 1000 2000 3000 4000 5000 6000
Depth (A)
100-
(c)
80-

60-

40- Al
O G
20 u Ta

0
0 1000 2000 3000 4000 5000 6000
Depth (A)


100 1
(b)
80 U

60
N O Ga
40- \a
40

20 A Ti

0-
0 1000 2000 3000 4000 5000 6000 7000
Depth (A)


1000 2000 3000 4000 5000 6000
Depth (A)


Figure 4-3: Depth profiles of Ti/Al/TaN/Ti/Au contacts (a) as deposited (b) annealed at 6000C.
(c) annealed at 800C and (d) annealed at 800C and aged at 3500C.









Table 2-2: Properties of common semiconductors


Bandgap
(eV)
@300C
Electron
mobility
(cm2/V-s),
RT
Hole
mobility
(cm2/V-s),
RT
Saturation
velocity
(cm/s), 107
Breakdown
field (V/cm)
x 106
Thermal
conductivity
(W/cm)
Melting
temperature
(K)


Si
1.1
indirect

1400



600



1


0.3


1.5


1690


GaAs
1.4
direct

8500



400



2


0.4


0.5


1510


GaN
3.4
direct

1000 (bulk)
2000
(2DEG)

30



2.5


>5


1.5


>1700


A1N
6.2
direct

135



14



1.4


3000


6H-SiC
2.9
indirect

600



40



2


4


>2100





















E

LU


260
240
220
200
180
160
140
120
100
80
60
40
20
0


0 100 200 300 400 500 600

Anneal temperature (C)


700 800


Figure 5-4: Influence of the annealing temperature on the characteristic energy related to the
tunneling probability. Dashed and dotted lines correspond to the values ofEo for
NA-1019 and 5x1019cm-3 respectively.


- W2B/Pt/Au
--- W2B /Pt/Au


NA-5x1019 cm

N ~1019 cm-3
N-0 cm --


I I I I I I I I


' ' '


















I-
I-
0l


0.01

1E-3

1E-4

1E-5

1E-6

1E-7


1E-8-

1E -9 ,
-3 -2 -1 0 1 2 3

Voltage (V)




Figure 3-22: Current-voltage curves for Ni/Ir/Au contacts



































To my grandparents.









3 .1.3 .2 Su m m ary ............................................................................. 52
3.1.4 Contact Resistance for Other Boride-based Contacts ..........................................52
3.1.4.1 Titanium boride-based contacts .................... ......... .................52
3.1.4.2 Zirconium Boride-based contacts.............................................................54
3.1.4.3 Gallium Nitride//Tungsten Boride-based contacts.................. ............55
3.1.5 Iridium-Based Contacts .............................. ........57
3.1.5.1 Experim ent and discussion....................................... ......... ...............57
3 .1.5 .2 Su m m ary ...............................................................59
3 .2 C o n c lu sio n s ............................................................................................................... 5 9

4 O H M IC C O N TA C TS TO n-G aN ...................................................................... ..................89

4 .1 E x p e rim e n t ........................................................................................................................8 9
4.2 Results and D discussion .................................... ..... .......... .............. .. 90
4 .3 C o n c lu sio n s ................................................................................................................. 9 2

5 BORIDE-BASED SCHOTTKY CONTACTS TO p-GaN.................... ........................... 103

5 .1 In tro d u ctio n ............................................................................................................... 1 0 3
5.2 E xperim mental D details ............................................................................ .................... 104
5.3 R results and D discussion ..................................... .................. .. .......... 106
5.4 C onclu sions ...................................... .................................................... 110

6 BORIDE AND IR BASED CONTACTS FOR LIGHT EMITTING DIODES......................123

6 .1 In tro d u ctio n ............................................................................................................... 12 3
6.2 E xperim mental .......................................... .... .......... ................. ............ 124
6.3 R results and D iscu ssion ................................................................................ ..... ......125
6 .4 C o n c lu sio n s ............................................................................................................... 12 7

7 C O N CLU SIO N ......................................... .... .............. ................... ............ 132

L IST O F R E F E R E N C E S .................................................................................... ...................137

B IO G R A PH IC A L SK E T C H ......................................................................... .. ...................... 144

















6









to the lack of these phases and the resulting compensation of holes by nitrogen vacancies. The

absence of Ti diffusion through the W2B also suggests that an optimum metal scheme might

involve a thin sandwich of the boride layers around the Ti, with the Au overlayer to reduce the

sheet resistance. The reflectivity of pure W is around 60% at 460 nm, while that of Au is around

95% [70], so clearly the thickness of the boride layers should be small relative to the Au

thickness. The long-term stability and reliability of the boride-based contacts must still be

established on LEDs and laser diodes. In the former case, the reflectance of the multi-layered

contacts at the emission wavelength of the LEDs must also be established.

3.1.5 Iridium-Based Contacts

3.1.5.1 Experiment and discussion

Figures 3-21 and 3-22 show the IV plots for both types of contacts at different anneal

temperatures and 20 pm spacing. Only the Ni/Au/Ir/Au contacts annealed at 500 C were found

to have Ohmic characteristics. It was observed that these contacts displayed a specific contact

resistance of -2.3 x 10-4 Q-cm2, comparable to results previously achieved on similar wafers for

other schemes. Table 3-4 shows a summary of previous results achieved using a similar

approach with other thermally stable materials as well as the anneal temperatures above which

the contacts failed either due to unreasonable morphology or breakdown of Ohmic

characteristics. From the IV curves for the Ni/Au/Ir/Au contacts, we see a decrease in the

amount of current upon annealing at 300 C, which likely corresponds to indiffusion into the

GaN of Ni but at a temperature too low for the Ni hydride to form and increase the doping. At

500 C the contacts transition to Ohmic behavior due to increased near surface doping. At higher

temperatures the contacts revert to rectifying behavior. For the Ni/Ir/Au contacts, the as

deposited display the largest amount of current. At 300 C the amount of current decreases









CHAPTER 6
BORIDE AND IR BASED CONTACTS FOR LIGHT EMITTING DIODES

6.1 Introduction

InGaN/GaN multiple quantum well light-emitting diodes (MQW-LEDs) are commercially

available in a broad range of wavelengths for use in applications such as full color displays,

traffic signals, and exterior lighting. There is also interest in shorter wavelength LEDs with

AlGaN active regions to excite down conversion phosphors for white light [94]. Nevertheless, to

compete with fluorescent and other high-efficiency lighting sources, it is essential to drive GaN-

based LEDs at very high current densities to maximize light output. One drawback of the high

current densities is self-heating of the heterostructure which can produce Ohmic contact

degradation and generation of non-radiative recombination centers [95-98]. Cao et al. [99-101]

demonstrated improved heat dissipation and current spreading by reducing the density of

dislocations in the heterostructure through the use of free-standing GaN instead of the more

common sapphire substrates. Further reduction of self-heating effects and improved efficiency

and operating lifetime can be obtained through the use of low-resistance and thermally stable

Ohmic contacts to the p-GaN layer. Indeed, power dissipation by Joule heating across the p-

GaN/metal interface at high current injection levels was found to produce indiffusion of the

Ohmic contact elements along dislocations in the III-nitride-based epilayers, leading to an

electrical short of the pn junction. [102,103]

The conventional metallization schemes used for making Ohmic contacts to p-GaN for

InGaN/GaN LEDs are based on high work function metals such as Ni, Pd, Cr or Pt with an

overlayer of Au to reduce the sheet resistance. Specific contact resistances of 10-2-10-4 Q.cm2

are obtained by annealing at 450-650 C [106,107]. These contacts have stability problems at

high temperature. For example, the initial Au/Ni/GaN structure transforms to Ni/Au/GaN with a









GaN using the TE model, most of the work has demonstrated that TFE is the dominant transport

mechanism. Reported barrier heights on p-GaN range from -0.5 to 2.9 eV even for the same

metallization scheme because the mechanism of current flow is not definitively established [51-

65].

In the case of TE, electrons or holes are excited over the potential barrier and the forward

current is given by the following expression:

2 e, eV
JF = AT2exp exp e 1(2.5)
SkT) nkkT)

where A*=103.8 Acm-2K-2 is the effective Richardson's constant for p-GaN, Tis the absolute

temperature, e is the electronic charge, OB is the SBH, ks is the Boltzmann's constant, and Vis

the applied voltage.

In the case of TFE, quantum tunneling of charge carriers through the barrier occurs and the

forward current can be described by


qJ e
JF = oexp qV (2.6)
Eo)

where the saturation current density Jo is given by

A'T \Eo q (] -V- 05 (_ q _
J = k cosh(Eoo/kBT) exp kT kT 7)


In Eq. (2.7), -(E-Ev)/q is the difference between the valence band maximum and the

position of the Fermi level and E0 = E00 coth (Eo00/kT) is the characteristic energy related to the

tunneling probability.









reasonably stable upon annealing to -600-7000C. While current-voltage characteristics as a

function of measurement temperature suggest thermionic field emission (TFE) over a Schottky

barrier as the most dominant mechanism of forward current flow, the bias and measurement

temperature dependence of the reverse-bias current indicates that leakage must originate from

surface leakage or generation in the depletion layer.

5.2 Experimental Details

The p-GaN samples were 1 [tm-thick Mg-doped GaN layers grown by metal organic

chemical vapor deposition (MOCVD) on 1 utm-thick undoped buffers on c-plane Al203

substrates. The hole concentration obtained from Hall measurements after acceptor activation

annealing was ~1017 cm-3. The surface was cleaned by sequential rinsing in acetone, ethanol and

10:1 H20:HC1 prior to insertion in the deposition chamber. Boride(500A)/Pt(200A)/Au(800A),

where the boride was either W2B or W2B5, was used as the Schottky metallization scheme. Au

was added to lower the contact sheet resistance while Pt is a diffusion barrier. All metals or

compounds were deposited by Ar plasma-assisted rf (13.56 MHz) sputtering at a pressure of 15

mTorr and rf powers of 250-400 W. The contacts were patterned by liftoff of lithographically-

defined photoresist and annealed at temperatures up to 7000C for 1 min in a flowing N2 ambient

in a RTA furnace. For Ohmic contacts, we used Pt/Au annealed at 5000C in 02 for 30s prior to

deposition of the Schottky metallization. Ring contact geometry for the diodes was employed,

with the Schottky contacts surrounded by the Ohmic contacts. The Schottky contact dots were

40 [tm in diameter, the surrounding was 60 [tm in inner diameter and 70 [tm in outer diameter.

XPS, current-voltage (I-V), and auger electron spectroscopy (AES) were used to

characterize boride-based rectifying contacts. XPS measurements were taken with a Physical

Electronics 5100LSci spectrometer with an aluminum x-ray source (energy 1486.6 eV). High-









18. R. Li, S. J. Cai, L. Wong, Y. Chen, K. L. Wang, R. P. Smith, S. C. Martin, K. S. Boutros, J.
M. Redwing, IEEE Electron Device Letters, 20, 323 (1999).

19. S. T. Sheppard, K. Doverspike, W. L. Pribble, S. T. Allen, J. W. Palmour, L. T. Kehias, T. J.
Jenkins, IEEE Electron Device Letters, 20, 161 -163 (1999).

20. D. Qiao, Z. F. Guan, J. Carlton, S. S. Lau, and G. J. Sullivan, Appl. Phys. Lett. 74, 2652
(1999)

21. V. Tilak, R. Dimitov, M. Murphy, B. Green, J. Smart, W. J. Schaft, J. R. Shealy and L. F.
Eastman, Mat. Res. Soc. Symp. Proc., 622, T7.4.1(2000).

22. S. -H. Lim, J. Washburn, Z. Liliental-Weber, and D. Qiao, Appl. Phys. Lett. 78, 3797 (2001).

23. M. AsifKhan, M. S. Shur and Q. Chen, Appl. Phys. Lett., 68, 3022(1996).

24. S. Murai, H. Masuda, Y. Koide, and M. Murakami, Appl. Phys. Lett. 80, 2934 (2002)

25. C. Lu, H. Chen, X. Lv, X. Xie, and S. N. Mohammad, J. Appl. Phys. 91, 9218 (2002)

26. J. S. Foresi and T. D. Moustakus, Appl. Phys. Lett., 62, 2859(1993).

27. M. W. Fay, G. Moldovan, P. D. Brown, I. Harrison, J. C. Birbeck, B. T. Hughes, M. J. Uren,
and T. Martin, J. Appl. Phys. 92, 94 (2002).

28. K. O. Schweitz, P. K. Wang, S. E. Mohney, and D. Gotthold, Appl. Phys. Lett. 80, 1954
(2002).

29. M. W. Cole, D. W. Eckart, W. Y. Han, R. L. Pfeffer, T. Monahan, F. Ren, C. Yuan, R. A.
Stall, S. J. Pearton, Y. Li and Y. Lu, J. Appl. Phys. 80, 278 (1996).

30. A. Zeitouny, M. Eizenberg, S. J. Pearton, and F. Ren, J. Appl. Phys. 88, 2048 (2000).

31. D. Selvanathan, F. M. Mohammed, A. Tesfayesus and I. Adesida, J.Vac.Sci.Technol. B22
2409(2004).

32. B. Luo, F. Ren, R. C. Fitch, J. K. Gillespie, T. Jenkins, J. Sewell, D. Via, A. Crespo, A.G.
Baca, R. D. Briggs, D. Gotthold, R. Birkhahn, B. Peres and S. J. Pearton, Appl. Phys. Lett.
82, 3910 (2003).

33. H. W. Jang and J. -L. Lee, J.Appl.Phys.93,5416(2003).

34. R. C. Fitch, J. K. Gillespie, N. Moser, T. Jenkins, J. Sewell, D. Via, A. Crespo, A.M.
Dabiran, P. P. Chow, A. Osinsky, J. R. LaRoche, F. Ren and S. J. Pearton, Appl.Phys.Lett.
84, 1495 (2004).

35. R. C. Fitch, J. K. Gillespie, N. Moser, T. Jenkins, J. Sewell, D. Via, A. Crespo, A.M.
Dabiran, P. P. Chow, A. Osinsky, J. R. LaRoche, F. Ren and S. J. Pearton,
J.Vac.Sci.Technol. B22, 619(2004).

















































Figure 3-25: Scanning electron microscopy images of Ni/Au/Ir/Au contacts


i -1
5/07/07 350.0'X 100.Omm
I Annealed at 300C













i -1
5/07/07 350.0 X 100.Omm
I Annealed at 700C


5/07/07 350.0 X 100.0mm
I Annealed at 500C

















100 1
90 Au pt W2B/Pt/Au as-deposited -
80
70
60
W
50 Ga
40 B
30-
20
0
10
0
0 50 100 150 200

Sputter depth (nm)


uu
90 Au Pt W2B5/Pt/Au as-deposited -
80
70 -
60 Ga
50 W
40 N
30 B
20-
10
O -
0
0 50 100 150 20

Sputter depth (nm)


0 50 100 150 200

Sputter depth (nm)


0 50 100 150

Sputter depth (nm)


Figure 5-10: Depth profiles of W2B/Pt/Au contacts and W2B5/Pt/Au rectifying contacts (a,b)

before and (c,d) after annealing at 6000C.









CHAPTER 5
BORIDE-BASED SCHOTTKY CONTACTS TO P-GAN

5.1 Introduction

The unique combination of wide bandgap, high breakdown field, high saturation velocity,

and ability to form high-quality heterostructures with good transport properties make GaN an

ideal candidate for several device applications including light emitting diodes (LEDs) for

displays, laser diodes for data storage, and high electron mobility transistors (HEMTs) for high-

power and high frequency electronics. An important aspect for improved reliability of GaN-

based devices is the development of more thermally stable Ohmic and Schottky contacts to both

n-type and p-type GaN. This aspect has been one of the major issues in the operating lifetime of

III-nitride lasers and LEDs and it is yet to be overcome completely. Another example is

AlGaN/GaN HEMTs in advanced power amplifiers and converters for which the development of

Ohmic and Schottky contacts ensuring long-time operation under uncooled and high temperature

conditions without metal spiking or loss of edge acuity remains a challenge. One approach that

has proven promising for contact formation on n-GaN is the use of very high melting

temperature metals like W and WSix [24,68,71-74]. Recently, other promising metallurgies

based on transition metal borides like W2B, W2B5, ZrB2, CrB2, and TiB2 have also been

proposed [75-78]. The interest in borides is due to their high melting temperature (e.g. 3200C

for ZrB2) and high thermodynamic stability. For p-GaN, most rectifying contact studies have

been performed with simple metals such as Ni, Pd, Ti, Cr, and Pt, [17, 54-66] and studies on

boride-based metallurgies are still in their infancy.

In this work, we report on the annealing and measurement temperature dependence of W2B

and W2B5-based rectifying contacts to p-GaN. The barrier heights determined by X-Ray

Photoelectron Spectroscopy (XPS) are in the range 2.7-2.9 eV. These contacts are found to be










E 1x10-3- 50


0 -40
o8
m 1x104 CrB )
30 a


S2- 20
1x10 5

0o 10Q

6
1x10-6 0 O
S0o 5 10 15 20 25
0.
CO Days



Figure 3-7: Specific contact resistance and sheet resistance under the contact as a function of
long term thermal aging at 3500C.









Ar plasma-assisted rf magnetron sputtering. For comparison, devices with Ni/Au contacts were

also fabricated using the same wafer. All contacts were patterned by liftoff of lithographically-

defined photoresist and annealed at 6000C for 1 min in a flowing N2 ambient. Prior to their

insertion in the e-beam chamber, the surface of both n- and p-type GaN were cleaned in a

1HC: 10H20 solution for 1 minute. Figure 6-1 shows a typical optical micrograph of as-

fabricated MQW-LEDs with Ni/Au/TiB2/Ti/Au Ohmic contacts. Similar contact morphology

was achieved using the other metallization schemes. All devices were first aged for a period of

10 days at 2000C on a heater plate in air. The samples were then removed from the heater block,

allowed to cool to room temperature, and characterized before being returned for further 35 days

aging at 3500C. MQW-LEDs were analyzed by luminescence-current-voltage (L-I-V)

measurements using a probe station and an Agilent 4145B parameter analyzer. The light output

power was measured using a Si photodetector located at -2 cm from the sample surface.

Electroluminescence (EL) spectra were acquired by a fiber optic spectrometer.

6.3 Results and Discussion

Figure 6-2 presents the 300 K L-I characteristics from each as-fabricated device. Typical

EL spectra measured for different forward currents are also shown in the inset. For an injected

current of 500 ptA, the EL peak wavelength of the InGaN/GaN MQW-LEDs is 459 nm with a

full width at half maximum (FWHM) of 26 nm. As the injected current increases, the peak

exhibits a blue shift while the FWHM remains fairly constant. This behavior results from the

decrease by carrier screening of the quantum-confined Stark effect induced by the built-in

piezoelectric field in the strained InGaN/GaN QWs [113]. The L-I characteristics are similar for

all metallization schemes investigated, showing that the TiB2/Ti/Au or Ir/Au overlayers did not

reduce the light output with respect to the more conventional Ni/Au contacts. The non-linear











E 1x10-3



C

) 10-4 __,
---- Ti/AI/Pt/Au
S-- -- Ti/AI/TaN/Ti/Au
Ti/Al/TiN/Ti/Au
C --v- Ti/AI/ZrN/Ti/Au
0


1x105
o.
80 2 4 6 8 10 12 14 16 18 20 22 24
Cn Days






Figure 4-8. Specific contact resistance as a function of long term thermal aging.









vacancies and thus increases the donor density. The Au overlayer is present in order to prevent

oxidation as well as to promote current spreading in the contact.

2.3.2 Schottky Contacts

The other type of electrical contact required for electronic devices are known as Schottky

contacts or diodes. Schottky contacts allow little or no current to flow through them until a

critical voltage is reached, above which large amounts of current flow. This is called the forward

biased region. In the reverse bias region, voltage is applied in the opposite direction and no

current flows initially. However, if the voltage is pushed high enough then breakdown occurs, a

huge amount of current flows through the contact and the contact is generally destroyed.

Ideal Schottky contacts are formed in the absence of any surface states. When a metal and

semiconductor are brought together, the Fermi levels must line up. In order for this to occur, a

charge exchange occurs between the two materials. When this occurs, a space charge or

depletion region is formed and a barrier to current flow is created. For p-type semiconductors,

the flat band diagram is shown in Figure 2-4.

The barrier height, 4bp, for a p-type semiconductor is given by the equation

q4bp = Eg -q (m -X) (2.4)

where Eg is the band gap, 4m is the metal work function, and x is the electron affinity of the

semiconductor. Invariably, however, surface states are present. This makes the prediction of

barrier height using the equation above inaccurate. In order to determine barrier height in this

case, it is necessary to fabricate a specimen to test.

The predominant carrier in Schottky contacts is the majority carrier unlike a pn junction

diode. Two mechanisms for current flow exist: thermionic emission (TE) and thermionic field

emission (TFE) or tunneling. While some authors have reported values for barrier heights on p-









LIST OF FIGURES


Figure page

1-1 M market forecast for GaN-based devices................................ ........................ ........ 17

2-1 Intrinsic carrier concentration of GaN, GaAs, and Si........................................................38

2-3 Flat band diagram for a p-type Ohmic contact ...................................... ............... 40

2-4 Flat band diagram for a p-type Schottky contact..........................................................41

2-5 LED cross section (a) before and (b) after processing........................................... 42

2-6 L inear transm mission line pattern .............................................................. .....................43

2-7 R resistance v s. pad spacing plot............................................................... .....................44

2-8 Schottky contact schem atic............................................ ................... ............... 45

3-1 Specific contact resistance and sheet resistance under the contact ofNi/Au/ X/ Ti/Au
contacts as a function of anneal temperature. .......................................... ............... 64

3-3 Scanning electron microscopy images of Ni/Au/TaN/Ti/Au contacts (a) as deposited
(b) annealed at 600 o C (c) annealed at 7000C and aged at 200 C until the contacts
became non-Ohmic and (d) annealed at 1000 C............. ................... ..... ...........66

3-6 Specific contact resistance versus measurement temperature. ........................................69

3-7 Specific contact resistance and sheet resistance under the contact as a function of
long term therm al aging at 350 C. ............................................ ............................. 70

3-8 Depth profiles of W2B-based contacts (a) as deposited (b) annealed at 600 C (c)
annealed at 700 C and aged at 3500C and (d) annealed at 1000 C..............................71

3-9 Specific contact resistivity of Ni/Au/TiB2/Ti/Au Ohmic contacts and p-GaN sheet
resistance under the contact as a function of annealing temperature.............................72

3-10 Secondary electron images of Ni/Au/TiB2/Ti/Au contact pads on p-GaN as-deposited
(top) or after annealing at either 800(center) or 9000C (bottom). ..................................73

3-11 Surface scans of Ni/Au/TiB2/Ti/Au Ohmic contacts on p-GaN as a function of anneal
temperature. The as-deposited sample is at top, that annealed at 8000C at center and
that at 9000C at bottom .......................................... .............................74

3-12 Depth profiles of Ni/Au/TiB2/Ti/Au Ohmic contacts on p-GaN as a function of
anneal temperature. The as-deposited sample is at top, that annealed at 800 pC at
center, and that at 900 oC at bottom ............................................................................ 75
















I I I I I I'
0- -- Contact Resistance
E Sheet Resistance
o80
S1E-3

O 60 D
(D

40 (D
__ 1E-4 -


0 -20



0 I I I I
o -----------------

700 750 800 850 900 950

Temperature



Figure 3-9: Specific contact resistivity ofNi/Au/TiB2/Ti/Au Ohmic contacts and p-GaN sheet
resistance under the contact as a function of annealing temperature.












Table 2-3: Properties of the borides


Properties
Melting Point
(C)
Structure
Thermal
expansion
coefficients x
106 (/deg)
Phonon
component of
heat
conduction
(wt/m-deg)
Elastic
modulus E x
10-6 (kg/cm2)
Characteristic
Temperature
(K)
Density of
electronic
state g x 10-21
(eV1 cm1)
Work
function (eV)
Heat of
Formation
(kcal/mol)
Lattice
constant (A)
Thermal
conductivity
(W/m-K)
Electrical
resistivity
(uOhm-cm)


TiB2
2980 -3225

Hexagonal
4/6


20.6


ZrB2
3040 -3200

Hexagonal
5.9


W2B
-2670


18.9


1100


4.76


4.19(?)


3.94(?)


71.4


3.028


3.169


Unknown 32


W2B5
-2385


Hexagonal


CrB2
2200

Hexagonal
10.5


10.4


54.6


3.18(?)


2.982


2.969









94. Introduction to Nitride Semiconductor Blue Lasers and Light-Emitting Diodes, edited by S.
Nakamura and S.F. Chichibu (Taylor and Francis, London, 2000).

95. M. Meneghini, L. -R. Trevisanello, U. Zehnder, T. Zahner, U. Strauss, G. Meneghesso and E.
Zanoni, IEEE Trans. Electron. Dev. 53, 2981 (2006).

96. S. Bychikhin, D. Pogany, L. K. J. Vandamme, G. Meneghesso, and E. Zanoni, J. Appl. Phys.
97, 123714 (2005).

97. M. Pavesi, F. Rossi, and E. Zanoni, Semicond. Sci. Technol. 21, 138 (2006).

98. F. Rossi, M. Pavesi, M. Meneghini, G. Salviati, M. Manfredi, G. Meneghesso, A. Castaldini,
A. Cavallini, L. Rigutti, U. Strass, U. Zehnder, and E. Zanoni, J. Appl. Phys. 99, 053104
(2006).

99. X. A. Cao and S.D. Arthur, Appl. Phys. Lett. 85, 3971 (2004).

100. X. A. Cao, S. F. LeBoeuf, M. P. D'Evelyn, S. D. Arthur, J. Kretchmer, C. H. Yan, and Z. H.
Yang, Appl. Phys. Lett. 84, 4313 (2004).

101. X. A. Cao, J. M. Teetsov, M. P. D'Evelyn, D. W. Merfeld, and C. H. Yan, Appl. Phys. Lett.
85, 7 (2004).

102. S. Nakamura, M. Senoh, S. Nagahama, N. Iwasa, T. Yamada, T. Matsushita, Y. Sugimoto
and H. Kikyou, Jpn. J. Appl. Phys., Part 2 36, L1059 (1997).

103. C. Y. Hsu, W. H. Lan and Y. S. Wu, Appl. Phys. Lett. 83, 2447 (2003).

104. A. P. Zhang, B. Luo, J. W. Johnson, F. Ren, J. Han and S. J. Pearton, Appl. Phys. Lett. 79,
3636 (2001).

105. T. Maeda, Y. Koide and M. Murakami, Appl. Phys. Lett. 75, 4145 (1999).

106. L. -C. Chen, J. -K. Ho, C. -S. Jong, C. C. Chiu, K. -K. Shih, F. -R. Chen, J. -J. Kai and L.
Chang, Appl. Phys. Lett. 76, 3703 (2000)

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

108. H. S. Venugopalan, S. E. Mohney, J. M. DeLucca, B. P. Luther and G. E. Bulman, J. Vac.
Sci. Technol. A 16, 607 (1998).

109. C. -F. Chu, C. C. Yu, Y. K. Wang, J. Y. Tsai, F. I. Lai and S. C. Wang, Appl. Phys. Lett.
77, 3423 (2000).

110. L. F. Voss, R. Khanna, S. J. Pearton, F. Ren and I. I. Kravchenko, Appl. Phys. Lett. 88,
012104 (2006).









CHAPTER 3
THERMALLY STABLE OHMIC CONTACTS TO P-GAN

3.1 Ohmic Contacts

In this section, two approaches to fabricating Ohmic contacts using the boride family of

materials will be discussed. The first is as an overlayer on conventional Ni/Au contacts and the

second as the actual contact material. Because these borides have a large work function, it is

expected that both approaches will be successful in yielding Ohmic behavior, provided that

sputter damage and oxygen incorporation into the borides during deposition does not

compromise the contacts, especially the latter.

3.1.1 Fabrication of Ohmic Contacts

The samples used were light-emitting diode wafers with a 0.1 um Mg-doped p-GaN layer

on top of a 0.1 um p-AlGaN layer followed by a 0.3 um InGaN/GaN superlattice on top of a n-

type GaN layer grown on 2 um-thick undoped buffers on c-plane A1203 substrates. Hole

concentrations in the p-GaN layer were 1017 cm-3 obtained from Hall measurements. Mesas for

the linear transmission line method (LTLM) method were etched using a C12/Ar inductively

coupled plasma for pad isolation and to minimize current spreading. Samples were dipped in 1

M KOH for 1 minute and then rinsed in acetone and ethanol prior to photolithography. Prior to

insertion into the evaporation chamber, the samples were exposed to a 30 s 02 RIE plasma in

order to remove any residual carbon and then dipped in 10:1 H20:HC1 for 5 minutes, which

serves to remove any native oxide. A Ni/Au contact of 500 A/800 A was deposited using an e-

beam evaporator in order to minimize surface damage. After unloading, the samples were put

into a sputtering system for deposition of the X/Ti/Au layer of thickness 500 A/200 A/800 A,

where X is the refractory metal used. Au is necessary to decrease the sheet resistance while the

Ti is necessary to prevent peeling of the Au from the surface due to the large lattice mismatch.


















Slope= Rs/W


L, L2
Distance L


Figure 2-7: Resistance vs. pad spacing plot









reaction with the GaN. This suggests that that the boride-based metallurgies may be promising

for rectifying contacts to p-GaN where thermal stability is a critical issue. The bias and

measurement temperature of the reverse leakage current indicates that leakage must originate

from surface leakage or generation in the depletion region through deep-level defects.









Ni/Au/TiB2/Ti/Au/p-GaN structure as a function of annealing temperature, along with the sheet

resistance of the p-GaN under the contacts, extracted from the TLM measurements. The contacts

are rectifying below an anneal temperature of 5000C but transition to Ohmic behavior for higher

temperatures. The specific contact resistivity improves steadily with temperature above 700C

and reaches a minimum of 2 x10-4 Q.cm-2 after annealing at 800 OC for 60 secs. Note that the

contact resistance stays at a similar value even for temperatures up to 9500C, even though the

higher annealing temperatures produced increases in the sheet resistance of the GaN and less

reproducible contact properties due to roughening of the contact morphology. The mechanism

for the increase in sheet resistance may include loss of N (producing nitrogen vacancy-related

donors) or a reduction in the soluble Mg concentration. This specific contact resistance is

comparable to that achieved on the same samples with Ni/Au metallization annealed at 5000C.

The incorporation of the TiB2 diffusion barrier clearly provides a much wider range of thermal

stability compared to the standard Ni/Au contacts.

Figure 3-10 shows SEM pictures of the TLM contact pads as a function of annealing

temperature. After 8000C annealing the morphology is degraded, but is still much smoother than

conventional Ni/Au contacts under these conditions. The darker appearance of the contacts after

high temperature anneals is mainly a result of the outdiffusion of Ti, which rapidly oxidizes.

Figure 3-11 shows the AES surface scans from the as-deposited sample (top) and those

annealed at 8000C (center) and 9000C (bottom). Ti is evident on the surface after annealing at

8000C and the concentration increases with annealing temperature. The increased oxygen

concentration evident from the summary of the data in Table 3-1 most likely comes from

oxidation of this out-diffused Ti. The carbon signal in all cases comes from adventitious carbon

on the surface.









3-13 Specific contact resistance of Ni/Au/ZrB2/Ti/Au and ZrB2/Ti/Au Ohmic contacts and
p-GaN sheet resistance under the contact as a function of annealing temperature............76

3-14 Surface scans and depth profiles of Ni/Au/TiB2/Ti/Au Ohmic contacts on p-GaN as a
function of anneal tem perature. ............................................... .............................. 77

3-15 Scanning electron microscopy images of Ni/Au/TiB2/Ti/Au contact pads on p-GaN
as-deposited (top) or after annealing at either 750 (center) or 8000C (bottom) ...............78

3-16 Surface scans and depth profiles of ZrB2/Ti/Au Ohmic contacts on p-GaN as a
function of anneal tem perature. ............................................... .............................. 79

3-17 Elemental maps obtained from scanning AES of ZrB2/Ti/Au Ohmic contacts pads on
p -G aN ................... ............................................................ ................ 8 0

3-18 Specific contact resistivity of W2B/Ti/Au Ohmic contacts and measured p-GaN sheet
resistance under the contact as a function of annealing temperature.............................81

3-19 AES surface scans of W2B/Ti/Au Ohmic contacts on p-GaN as a function of anneal
tem p eratu re. ............................................................ ................. 82

3-20 Depth profiles of W2B/Ti/Au Ohmic contacts on p-GaN as a function of anneal
tem p eratu re. ............................................................ ................. 83

3-21 Current-voltage curves for Ni/Au/Ir/Au contacts. .................................. .................84

3-22 Current-voltage curves for Ni/Ir/Au contacts ........................................ ............... 85

3-23 Depth profiles for Ni/Au/Ir/Au contacts (a) annealed at 300 C (b) annealed at 500
C and (c) annealed at 700 C ................................................ ............................... 86

3-24 Depth profiles for Ni/Ir/Au contacts (a) annealed at 300 C (b) annealed at 500 C
and (c) annealed at 700 oC ......................................... ..................... .. .....87

3-25 Scanning electron microscopy images of Ni/Au/Ir/Au contacts........................................88

4-1 Specific contact resistance as a function of anneal temperature........................................95

4-2 Scanning electron microscopy images of annealed contacts.............................................96

4-3 Depth profiles of Ti/Al/TaN/Ti/Au contacts (a) as deposited (b) annealed at 6000C.
(c) annealed at 8000C and (d) annealed at 8000C and aged at 350C. .............................97

4-4 Depth profiles of Ti/Al/TiN/Ti/Au contacts (a) as deposited (b) annealed at 6000C.
(c) annealed at 8000C and (d) annealed at 8000C and aged at 350C. .............................98

4-5 Depth profiles of Ti/Al/ZrN/Ti/Au contacts (a) as deposited (b) annealed at 6000C.
(c) annealed at 8000C and (d) annealed at 8000C and aged at 350C. .............................99









TABLE OF CONTENTS

page

A CK N O W LED G M EN T S ................................................................. ........... ............. .....

L IS T O F T A B L E S ................................................................................. 7

LIST OF FIGURES .................................. .. ..... ..... ................. .8

A B S T R A C T ................................ ............................................................ 1 1

CHAPTER

1 INTRODUCTION .................................................. .......... 13

2 BACKGROUND .................................. ... .. .... ...... ................. 18

2 .1 G allium N itride P roperties........................................................................ .................. 18
2.1.1 Fundam ental Properties ..............................................................................18
2.1.2 Electronic Properties ...................................................................... 19
2 .1.3 C ry stal Structure .............................. ........................................ ........... ... 19
2.2 Properties of the Contact Materials to be Studied ........................................................19
2 .2 .1 B o rid e s ................................. .......................................................... ............... 19
2 .2 .2 N itrid e s .......................................................................... 2 0
2 .2 .3 Iridiu m .............................................................................2 0
2.3 E electrical C contacts .................. ................. .............. ............... ...... .... 21
2 .3 .1 O hm ic C contacts ................ ........................................ .............. .. .... ...... .. 22
2.3.1.1 Ohm ic contacts to p-G aN .................................................................. 23
2.3.1.1 Ohm ic contacts to n-G aN .................................................................. 24
2.3.2 Schottky Contacts ................. ............... ..................... ............ 25
2 .4 E xperim ents ...............................................................................27
2.5 Characterization Techniques ................................................. ............................... 29
2.5.1 Current-V oltage ................................................... .............. .. ...... 29
2 .5.2 C apacitance-V oltage...................................................................... ...................30
2.5.3 X -ray Photoelectron Spectroscopy .................................... .................................. 31
2.5.4 Auger Electron Spectroscopy ................................................... ..................31

3 THERMALLY STABLE OHMIC CONTACTS TO p-GaN.................................................46

3.1 O hm ic C ontacts ......................................................... ................. 46
3.1.1 Fabrication of O hm ic Contacts......................................... .......................... 46
3.1.2 N itride-B asked C ontacts............................................................................ ...... 47
3.1.2.1 Experim ent and discussion.................................................. ...................47
3.1.2.2 Summary ....................... ..... ................. 50
3.1.3 Tungsten Boride and Chromium Boride-Based Contacts and Long Term
T herm al A going of B rides ........................................ ............................................50
3.1.3.1 Experim ent and discussion....................................... ......... ............... 50









Ta, Ti, and Zr are 1.5, 1.5, and 1.33, respectively. The values for Au, Ni, and Ga are 2.4, 1.8,

and 1.6, respectively. From this, it is apparent that the differences between Au and either Ti, Ta,

Zr, or Ga are roughly the same, indicating no preferred bonding to any of them. Various

equilibrium phases exist for Au-Ti, Au-Zr, and Au-Ta, but none for Au-N, supporting the

conjecture that Au is bonded to the metal. Thus, neither configuration is strongly preferred.

From the Au-Ni phase diagram, a miscibility gap exists with a maximum of 810.3 C at around

70 at% Au. At higher and lower Au concentrations, the temperature above which Au and Ni

intermix freely is lower; for example, at 10 at% Au, it is below 500 C and at 30 at% Au below

700 C. Since there is no driving force in terms of greater bond strength to cause Au and Ni to

switch places when using these diffusion barriers, the Ni and Au should each be present at the

surface of the GaN. It is expected that the Ni and Au would separate with slow enough cooling

but the layered structure of pure Ni/Au contacts may not form. The AES depth profiles confirm

this, displaying significant intermixing at the GaN interface. We also see that the boundary

between the GaN and the Ni-Au is not well-defined. This is consistent with the formation of the

Au-Ga phases. From the Au-Ga phase diagram, these phases have melting points of 461.3 C

and 491.3 C, respectively. These are lower than the temperatures required to obtain Ohmic

behavior, suggesting it is beneficial to have some Ni present at the GaN interface in order to

maintain contact integrity. [67]

Figure 3-4 shows specific contact resistance and sheet resistance under the contacts as a

function of thermal aging at 200 C. This simulates long term operation of these contacts in

devices. These contacts display poor stability upon aging at this temperature, lasting for nearly

10 days before the I-V characteristics are no longer completely linear. From the AES depth

profiles of the contacts that failed, there was significant in-diffusion of metals into the GaN. The









these contacts show promise for use in high temperature applications. While the contact

resistance does not seem to be better than that of conventional contacts after aging, there is far

less mixing between layers. This is a primary concern, due to the A1Au4 phase, for devices with

short spacings for high temperature applications. Further evaluation in devices such as HEMTs is

needed in order to determine to potential of these contacts. In addition, the effect of oxygen

incorporation in the nitride should be examined.















< 3.6
E 140

3.2- C)

=C
0 120

( 2.8
c --Contact Resistance *
100
S-*- Sheet Resistance
.* 2.4 -

O8 -80
S2.0 .

C 60
0 1.6

500 550 600 650 700 750 800

Temperature (oC)



Figure 3-18: Specific contact resistivity ofW2B/Ti/Au Ohmic contacts and measured p-GaN
sheet resistance under the contact as a function of annealing temperature.























































Figure 3-10: Secondary electron images of Ni/Au/TiB2/Ti/Au contact pads on p-GaN as-
deposited (top) or after annealing at either 800(center) or 9000C (bottom).


'104





10/03/05 10.OkV 250.0 X 100.0m
5 SOOC annealed
T-











10/03/05 10.OkV 250.0 XI 100.Omm
5 900C annealed









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

THERMALLY STABLE OHMIC AND SCHOTTKY CONTACTS TO GaN

By

Lars Fredrik Voss

May 2008

Chair: Stephen J. Pearton
Major: Materials Science and Engineering

This dissertation is focused on the development of Ohmic and Schottky contacts to both n-

and p-type Gallium Nitride for use in microelectronic and optical devices. The goal is to develop

low resistance contacts with greater thermal budgets and superior thermal aging characteristics to

those commonly in use today as well as to understand the mechanisms by which these contacts

may fail. In addition, p-type Ohmic contacts have been used to fabricate light emitting diodes

(LEDs) which display far superior aging properties than those made with conventional Ni/Au

contacts.

Ohmic contacts to p-GaN were fabricated using a variety of refractory materials. The

materials examined were of three basic types: boride, nitride, and the refractory metal Ir. The

boride family includes W2B, W2B5, CrB2, ZrB2, and TiB2. The nitrides examined were TaN,

TiN, and ZrN. Contacts based on these materials were fabricated using either a

GaN//Ni/Au/X/Ti/Au, GaN//X/Ti/Au, or GaN//Ni/X/Au scheme, where X is the refractory

material. Contact resistances as low as -1 x 10-4 Q/cm2 were consistently achieved after

annealing at temperatures from 500-10000C for 60 s in N2 using these materials for p-GaN with a

carrier concentration of-1 x 107 cm-3. In addition, high temperature thermal aging was









later, it only slightly influences the SBH of W2B and W2B5-based rectifying contacts to p-GaN.

Nakayama et al. [83] have suggested that a hole concentration of2x1017 cm-3 in p-GaN samples

leads to -=0.13 V. This hole concentration is very close to that achieved in the present study so

that we also assume = 0.13 V.

Auger electron spectroscopy (AES) was used to analyze the depth profiles. A Physical

Electronics 660 Auger Microprobe Electron Beam at 10 keV, 0.3 utA, 300 from sample normal

was used for the data collection while for profiling the ion beam conditions were 3 keV Ar+,

2.0 utA, (4 mm)2 raster with sputter-etch rates of 80, 40, and 104 A/min for the boride, Pt and Au

layer respectively. A survey spectrum (a plot of the first derivative of the number of electrons

detected as a function of energy) was used to determine the composition of the outer few

nanometers of each sample. Quantifications were accomplished by using elemental sensitivity

factors.

5.3 Results and Discussion

Figure 5-1 presents an example of the Ga 3d core level and the valence band spectrum

collected on a p-GaN surface without a boride overlayer. From Figure 5-1, we find

Evc=17.54 eV, which is in reasonable agreement with the value of 17.8 eV reported previously.

[62,84] After W2B deposition, Figure 5-1 shows that the binding energy of the Ga 3d core level

is EB=20.25 eV, thereby yielding OB=2.71 eV. Similarly, we find EB=20.41 eV for W2B5, which

gives B=2.87 eV. These results are similar to the 2.68-2.78 eV values reported by Yu et al. [56]

for Ni/p-GaN. Based on the Schottky-Mott model [52], the SBHs for n-GaN and p-GaN should

add up to the GaN band gap 3.4 eV. For W2B and W2B5-based rectifying contacts to n-GaN,

Khanna et al. [76,77] have reported OB-0.55 eV, thereby suggesting ~B-2.85 eV for p-GaN. Our

XPS results are clearly in excellent agreement with the predictions of this model.












1.0
0.8
0.6
0.4
0.2
0.0
1.0-
0.8
0.6
0.4
0.2
0.0
1.0
0.8
0.6
0.4
0.2
0.0
-4


6 8 10


Figure 6-2 : L-I characteristics of MQW-LEDs with Ni/Au, Ni/Au/TiB2/Ti/Au, and Ni/Au/Ir/Au
p-Ohmic contacts. The inset shows emission spectra from as-fabricated LEDs at
various injection currents.


-Day 0
- Day 10 @ 2000C
-----Day 45 @ 3500C

------------------............---_


-2 0 2 4
Voltage (V)









CHAPTER 2
BACKGROUND

2.1 Gallium Nitride Properties

2.1.1 Fundamental Properties

The materials properties of GaN make it desirable for a variety of applications. For optical

electronics, GaN holds great promise. Because it is a direct band gap semiconductor, it is an

efficient light emitter. GaN fills the need for a blue and ultraviolet wavelength emitter that is left

by other materials due to its band gap of 3.475 eV. Its exciton binding energy is 28 meV. This

is not as high as another blue emitters, ZnO, but GaN technology is more developed. The first

GaN LED was introduced in 1993; however, further development of the GaN system is

necessary for better LEDs and LDs.

In addition to optical applications, GaN is robust and therefore a candidate for high

temperature and high power applications, with potentially superior electronic properties when

compared with SiC [38]. The reason for this is due to the relatively wide band gap when

compared to Si and GaAs. Figure 2-1 displays the intrinsic carrier concentration as a function of

temperature of GaN, Si, and GaAs. Because of the wide band gap, the intrinsic carrier

concentration is much lower; 1015 cm-3 is reached at 3000C for Si, 5000C for GaAs, and at

10000C for GaN. Because of this, GaN can be operated at much higher temperatures before

device performance degrades and breaks down.

Table 2-1 contains some basic physical properties of GaN [39,40]. A key problem in

device operation is the presence of the lattice mismatch of GaN with the commonly used

sapphire (13%) and SiC (3%) substrates, which results in a large number of lattice defects which

hinder electrical conduction.









Schottky contact


Ohmic contact
/


Semiconductor film


Figure 2-8: Schottky contact schematic










CM
E

0

0
c
Cu
C,


0
e-

0


1x10-2




1x10 3




1x10-4




1x10-5


600


800


1000
1000


vC Anneal Temperature (OC)





Figure 4-1: Specific contact resistance as a function of anneal temperature


-*- Ti/Al/Pt/Au
-*-- Ti/AI/ZrN/Ti/Au
Ti/Al/TiN/Ti/Au
-v- Ti/Al/TaN/Ti/Au

-0


1 ^\y
^1^ ==


































0


4000

3000

2000

1000
1 om

0

c/s
-1000

-2000

-3000

-4000

-5000

-6000



8000

6000

4000

2000

0

c/s -200


-400

-600


'F> Qj
Bi AiM


500 1000 1500 2000
Kiretic Energy (eV)













B AuAu





TII 27.8
Tl C1 27.2

B2 5.7


0



500 1000 1500 2000
Kinetic Energy (eV)


Figure 3-11: Surface scans of Ni/Au/TiB2/Ti/Au Ohmic contacts on p-GaN as a function of

anneal temperature. The as-deposited sample is at top, that annealed at 8000C at

center and that at 9000C at bottom .


50C 10o 1500
Kinetic Energy (eV)


-12000


x10


-1 5









CHAPTER 7
CONCLUSION

Improved device processing is necessary in order to realize the full potential of GaN based

electronics and optoelectronics. While improved material quality is critical, especially for p-

GaN as well as related alloys such as InGaN to increase the efficiency of green emitting devices,

it is not the only challenge. Developing reliable, stable, low resistance Ohmic and Schottky

contacts to both n- and p-type GaN is still a challenge and hence remains of interest to the GaN

community at large. Without contacts that can withstand elevated temperatures and other harsh

environments, many of the properties that make GaN a unique and desirable semiconductor

mean nothing.

The goal of this work was to develop such contacts to both n- and p-GaN. In order to

improve upon existing contact schemes, it is necessary to achieve at least comparable contact

resistances or barrier heights. In addition, the contacts should be able to withstand more stressful

processing conditions, such as elevated annealing temperatures, while maintaining predictable

and stable characteristics. Perhaps the most critical factor is the ability of the contacts to

withstand long periods in a harsh environment, simulated here by placement on a hot plate at

temperatures of 200C and 350 C. A final consideration is the amount of intermixing of the

contact layers, especially in Ohmic contacts to n-GaN, as large amounts of intermixing can lead

to undesirable phases, such as A1Au4 and issues with lateral flow. In addition, if any undesirable

phases may form at device operation temperatures, or if the contacts themselves interact with the

GaN at these temperatures, devices will prove unreliable during prolonged operation. To tackle

these issues, three separate materials systems were examined: borides, nitrides, and Ir.









In the case of Ir, this Ti layer is unnecessary. Contacts were also fabricated of the form Ni/Ir/Au,

in order to confirm the importance of the Au near the GaN surface, and of the form X/Ti/Au to

investigate the use of the refractory materials without the increased doping present due to the

Ni/Au layer. Sputtering was performed using an Ar plasma-assisted rf sputter at 5 mTorr and rf

(13.56 IMHz) power of 150 W. The sputter rates were 5 A/s for Au, 0.5 A/s for Ti, and were 0.8-

2 A/s for the refractory materials. Following lift-off, the contacts were annealed in a flowing N2

ambient for 60 s in a rapid thermal annealing furnace at temperatures up to 1000 oC.

3.1.2 Nitride-Based Contacts

3.1.2.1 Experiment and discussion

Ohmic contacts based on Ni/Au but with a nitride overlayer were fabricated in order to

improve the temperature stability. Figure 3-1 shows the specific contact resistivity of the nitride-

based contacts as a function of annealing temperature, as well as the sheet resistance of the p-

GaN under the contacts. At annealing temperatures of less than 500 C, contacts were rectifying.

The contacts transition to Ohmic behavior at higher temperatures. The specific contact

resistance was not a function of temperature in the range 25-175C, indicating that tunneling is

the dominant transport mechanism. The specific contact resistance is fairly stable on annealing

up to 1000 oC, with a minimum of 2 x 10-4 Q-cm2 at 6000C for the Ni/ Au/ ZrN/ Ti / Au

contacts. This indicates that there is minimal change in the doping of the near surface region of

the GaN at these elevated anneal temperatures. The formation of AuGa and AuGa2 occurs at

600C at the interface 49]. A sharp maximum in the sheet resistance is observed at 800 oC, with a

drop at higher temperatures. This is likely due to increased in-diffusion of metal to the GaN

without complete alloying, resulting in interstitial atom incorporation into the lattice. The drop

in sheet resistance at higher temperatures is an indication that these interstitials are increasingly

incorporated into the GaN. The magnitude of the increase in sheet resistance is related to the

















Valence band
Region (X5)


I I

24 22 20 18 16 4 2 0
Binding Energy (eV)




W W2 B, p-GaN
Ga 3d / --- W2BI p-GaN






,I II


II
26 24 22 20 18 18 14
Binding Energy (eV)




Figure 5-1: XPS spectra without (top) and with (bottom) a boride overlayer. The left-hand
spectrum in the top figure corresponds to the Ga 3d core level whereas the right-hand
panel presents the spectrum of the valence band region.









Table 3-2: Concentration of elements detected on the as-received surfaces (in atom%)


Sample ID C(1) 0(1) Ti(1) Ni(1) Zr(2) Au(3)
Sensitivity factors
0.076 0.212 0.188 0.227 0.043 0.049
As deposited 44 4 nd nd nd 52
Ni/Au/ZrB2/Ti/Au
annealed 750C 45 19 2 1 14 19
annealed 800C 42 22 9 1 16 10
As deposited 37 4 nd nd nd 59
ZrB2/Ti/Au
annealed 800C 43 23 nd nd 12 22
annealed 1000C 48 19 nd nd 14 19










E 1x10-2 -
S! 7000C
1 3- -*- 1000C
x 0
o 1x103


x -4
re







U, Measurement Temperature (_C)
Figure 3-6 Specific contact resistance versus measurement temperature.
o

E 1x10
o 0 25 50 75 100 125 150
co Measurement Temperature (C)
Figure 3-6: Specific contact resistance versus measurement temperature.




Full Text

PAGE 1

1 THERMALLY STABLE OHMIC AND SCHOTTKY CONTACTS TO GaN By LARS FREDRIK VOSS A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2008

PAGE 2

2 2008 Lars Fredrik Voss

PAGE 3

3 To my grandparents.

PAGE 4

4 ACKNOWLEDGMENTS First and forem ost I would lik e to thank my advisor Prof. Stephen J. Pearton for his guidance and for all of the support and opportunities he provided. I would also like to thank my other supervisory committee members, Fan Re n, Cammy Abernathy, David Norton, and Rajiv Singh, for their help and their time. Thanks go to Dr. Patrick M. Lenahan for provi ding me with my first experience working in the semiconductor field while I was an undergraduate at Penn State as well as all of his advice and encouragement. I thank the members of the Pearton, Ren, and Abernathy re search groups with whom I have had the opportunity to work, including Luc Stafford, Kelly Ip, Jon Wright, Wantae Lim, Hungta Wang, Sam Kang, Travis Anderson, Soohwa n Jang, Brent Gila, Jerry Thaler, Jennifer Hite, Mark Hlad and many, many more. I would al so like to thank Ivan Kravchenko for his support in the UF Nanofabrication Facility. Thanks also go to all the people at Sandi a National Laboratories who gave me the opportunity to work there and with them for two enjoyable summers. I want to say thank you to my mentors, Randy J. Shul, Albert G. Baca, and Jeff E. Stevens, my managers, Charles Sullivan and Dale Hetherington, as well as all of the pe ople I had the pleasure to work with while at Sandia including Carlos Sanchez, David Torre s, Melissa Cavaliere, Karen Cross, Mark Overberg, Michael Cich, and many others. Thanks also go to all of my family and friends as well.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS...............................................................................................................4 LIST OF TABLES................................................................................................................. ..........7 LIST OF FIGURES.........................................................................................................................8 ABSTRACT...................................................................................................................................11 CHAP TER 1 INTRODUCTION................................................................................................................... ..13 2 BACKGROUND..................................................................................................................... ..18 2.1 Gallium Nitride Properties................................................................................................ 18 2.1.1 Fundamental Properties.......................................................................................... 18 2.1.2 Electronic Properties..............................................................................................19 2.1.3 Crystal Structure..................................................................................................... 19 2.2 Properties of the Contact Materials to be Studied ............................................................19 2.2.1 Borides....................................................................................................................19 2.2.2 Nitrides...................................................................................................................20 2.2.3 Iridium.................................................................................................................. ..20 2.3 Electrical Contacts........................................................................................................ ....21 2.3.1 Ohmic Contacts...................................................................................................... 22 2.3.1.1 Ohmic contacts to p-GaN............................................................................. 23 2.3.1.1 Ohmic contacts to n-GaN............................................................................. 24 2.3.2 Schottky Contacts...................................................................................................25 2.4 Experiments......................................................................................................................27 2.5 Characterization Techniques............................................................................................ 29 2.5.1 Current-Voltage......................................................................................................29 2.5.2 Capacitance-Voltage...............................................................................................30 2.5.3 X-ray Photoelectron Spectroscopy......................................................................... 31 2.5.4 Auger Electron Spectroscopy................................................................................. 31 3 THERMALLY STABLE OHMIC CONTACTS TO p-GaN.................................................... 46 3.1 Ohmic Contacts............................................................................................................. ...46 3.1.1 Fabrication of Ohmic Contacts............................................................................... 46 3.1.2 Nitride-Based Contacts...........................................................................................47 3.1.2.1 Experiment and discussion........................................................................... 47 3.1.2.2 Summary......................................................................................................50 3.1.3 Tungsten Boride and Chromium Bori de-Based Contacts and Long Term Thermal Aging of Borides........................................................................................... 50 3.1.3.1 Experiment and discussion........................................................................... 50

PAGE 6

6 3.1.3.2 Summary......................................................................................................52 3.1.4 Contact Resistance for Othe r Boride-based C ontacts............................................ 52 3.1.4.1 Titanium boride-based contacts................................................................... 52 3.1.4.2 Zirconium Boride-based contacts.................................................................54 3.1.4.3 Gallium Nitride//Tungsten Boride-based contacts....................................... 55 3.1.5 Iridium-Based Contacts..........................................................................................57 3.1.5.1 Experiment and discussion........................................................................... 57 3.1.5.2 Summary......................................................................................................59 3.2 Conclusions...............................................................................................................59 4 OHMIC CONTACTS TO n-GaN.............................................................................................. 89 4.1 Experiment........................................................................................................................89 4.2 Results and Discussion..................................................................................................... 90 4.3 Conclusions.......................................................................................................................92 5 BORIDE-BASED SCHOTTKY CONTACTS TO p-GaN ...................................................... 103 5.1 Introduction............................................................................................................... ......103 5.2 Experimental Details......................................................................................................104 5.3 Results and Discussion................................................................................................... 106 5.4 Conclusions.....................................................................................................................110 6 BORIDE AND IR BASED CONTACTS FOR LIGHT EMITTING DIODES ...................... 123 6.1 Introduction............................................................................................................... ......123 6.2 Experimental............................................................................................................... ....124 6.3 Results and Discussion................................................................................................... 125 6.4 Conclusions.....................................................................................................................127 7 CONCLUSION........................................................................................................................132 LIST OF REFERENCES.............................................................................................................137 BIOGRAPHICAL SKETCH.......................................................................................................144

PAGE 7

7 LIST OF TABLES Table page 2-1 Bulk GaN properties........................................................................................................ ..33 2-2 Properties of common semiconductors.............................................................................. 34 2-3 Properties of the borides.................................................................................................. ..35 2-4 Properties of the nitrides................................................................................................. ...36 2-5 Properties of Ir...................................................................................................................37 3-1 Concentration of elements detected on the as-received surface (in atom %)..................... 60 3-2 Concentration of elements detected on the as-received surfaces (in atom%) .................... 61 3-3 Concentration of elements detected on the as-received surfaces (in atom%) .................... 62 3-4 Summary of specifi c contact resistances ........................................................................... 63 4-1 Percent change in specific cont act resistance during therm al aging.................................. 94 5-1 Comparison of different barrier height calculations ........................................................112 6-1 Influence of long-term aging at 200C and 350C on the turn-on voltage and reverse current of InGaN/GaN MQW -LEDs............................................................................... 127

PAGE 8

8 LIST OF FIGURES Figure page 1-1 Market forecast for GaN-based devices............................................................................. 17 2-1 Intrinsic carrier concentr ation of GaN, GaAs, and Si ........................................................38 2-3 Flat band diagram for a p-type Ohm ic contact.................................................................. 40 2-4 Flat band diagram for a p-type Schottky contact ...............................................................41 2-5 LED cross section (a) before and (b) after processing ....................................................... 42 2-6 Linear transmission line pattern......................................................................................... 43 2-7 Resistance vs. pad spacing plot.......................................................................................... 44 2-8 Schottky contact schematic................................................................................................45 3-1 Specific contact resistance and sheet resist ance under the contact of Ni/A u/ X/ Ti/Au contacts as a function of anneal temperature..................................................................... 64 3-3 Scanning electron microscopy images of Ni/Au/TaN/Ti/Au contact s (a) as deposited (b) annealed at 600 o C (c) annealed at 700oC and aged at 200 oC until the contacts became non-Ohmic and (d) annealed at 1000 oC...............................................................66 3-6 Specific contact resistance ve rsus m easurement temperature...........................................69 3-7 Specific contact resistance and sheet resistance under the contact as a function of long term thermal aging at 350oC......................................................................................70 3-8 Depth profiles of W2B-based contacts (a) as depo sited (b) annealed at 600 oC (c) annealed at 700 oC and aged at 350oC and (d) annealed at 1000 oC..................................71 3-9 Specific contact resistivity of Ni/Au/TiB2/Ti/Au Ohmic contacts and p-GaN sheet resistance under the contact as a function of annealing temperature................................. 72 3-10 Secondary electron images of Ni/Au/TiB2/Ti/Au contact pads on p-GaN as-deposited (top) or after annealing at either 800(center) or 900C (bottom)......................................73 3-11 Surface scans of Ni/Au/TiB2/Ti/Au Ohmic contacts on p-GaN as a function of anneal temperature. The as-deposited sample is at top, that annealed at 800C at center and that at 900C at bottom .....................................................................................................74 3-12 Depth profiles of Ni/Au/TiB2/Ti/Au Ohm ic contacts on p-GaN as a function of anneal temperature. The as-deposited sample is at top, that annealed at 800 pC at center, and that at 900 oC at bottom.................................................................................. 75

PAGE 9

9 3-13 Specific contact resistance of Ni/Au/ZrB2/Ti/Au and ZrB2/Ti/Au Ohmic contacts and p-GaN sheet resistance under the contact as a function of annealing temperature............ 76 3-14 Surface scans and depth profiles of Ni/Au/TiB2/Ti/Au Ohmic contacts on p-GaN as a function of anneal temperature.......................................................................................... 77 3-15 Scanning electron microscopy images of Ni/Au/TiB2/Ti/Au contact pads on p-GaN as-deposited (top) or after annealing at either 750 (cente r) or 800C (bottom)................. 78 3-16 Surface scans and depth profiles of ZrB2/Ti/Au Ohmic contacts on p-GaN as a function of anneal temperature.......................................................................................... 79 3-17 Elemental maps obtained from scanning AES of ZrB2/Ti/Au Ohmic contacts pads on p-GaN.................................................................................................................................80 3-18 Specific contact resistivity of W2B/Ti/Au Ohmic contacts and measured p-GaN sheet resistance under the contact as a function of annealing temperature................................. 81 3-19 AES surface scans of W2B/Ti/Au Ohmic contacts on p-Ga N as a function of anneal temperature.................................................................................................................... ....82 3-20 Depth profiles of W2B/Ti/Au Ohmic contacts on p-Ga N as a function of anneal temperature.................................................................................................................... ....83 3-21 Current-voltage curves for Ni/Au/Ir/Au contacts.............................................................. 84 3-22 Current-voltage curves for Ni/Ir/Au contacts.................................................................... 85 3-23 Depth profiles for Ni/Au/Ir/A u contacts (a) annealed at 300 oC (b) annealed at 500 oC and (c) annealed at 700 oC............................................................................................86 3-24 Depth profiles for Ni/Ir/Au contacts (a) annealed at 300 oC (b) annealed at 500 oC and (c) annealed at 700 oC.................................................................................................87 3-25 Scanning electron microscopy im ages of Ni/Au/Ir/Au contacts ........................................88 4-1 Specific contact resistance as a function of anneal tem perature........................................95 4-2 Scanning electron microscopy images of annealed contacts............................................. 96 4-3 Depth profiles of Ti/Al/TaN/Ti/Au contacts (a) as deposited (b) annealed at 600oC. (c) annealed at 800oC and (d) annealed at 800oC and aged at 350oC................................ 97 4-4 Depth profiles of Ti/Al/ TiN/Ti/Au contacts (a) as de posited (b) annealed at 600oC. (c) annealed at 800oC and (d) annealed at 800oC and aged at 350oC................................ 98 4-5 Depth profiles of Ti/Al/ ZrN/Ti/Au contacts (a) as de posited (b) annealed at 600oC. (c) annealed at 800oC and (d) annealed at 800oC and aged at 350oC................................ 99

PAGE 10

10 4-7 Specific contact resistance as a function of anneal tim e.................................................. 101 4-8 Specific contact resistance as a f unction of long term thermal aging.............................. 102 5-1 XPS spectra without (top) and with ( bottom ) a boride overlayer. The left-hand spectrum in the top figure corresponds to th e Ga 3d core level wh ereas the right-hand panel presents the spectrum of the valence band region.................................................. 113 5-3 Forward current-voltage characteristic of W2B-based (top) and W2B5-based (bottom) Schottky diodes as a function of annealing...................................................................... 115 5-4 Influence of the annealing temperature on the characteristic energy related to the tunneling probability. D ashed and dotte d lines correspond to the values of E0 for NA~1019 and 5 1019cm-3 respectively..............................................................................116 5-5 Influence of the annealing temperatur e on the apparent Schottky barrier height derived from IV m easurements........................................................................................ 117 5-6 Dependence of the apparent Schottky barrier height on the param eter defined as the difference between the valence band maximum and the position of the Fermi level. Low and high...................................................................................................................118 5-7 As-measured and after oxide correction dependence of C-2 versus V of Au/Pt/W2B/pGaN Schottky diodes. The measurement frequency was set to 1 kHz............................ 119 5-8 Reverse current-voltage characteristic of W2B-based Schottky diodes as a function of measurement temperature................................................................................................ 120 5-9 Influence of the annealing te m perature on the breakdown voltage................................. 121 5-10 Depth profiles of W2B/Pt/Au contacts and W2B5/Pt/Au rectifying contacts (a,b) before and (c,d) after annealing at 600 C....................................................................... 122 6-1 Optical micrograph of an as-fabricated MQW -LED. The p-contact at the center of the diode is 80 m in diameter.........................................................................................128 6-2 L I characteristics of M QW-LE Ds with Ni/Au, Ni/Au/TiB2/Ti/Au, and Ni/Au/Ir/Au p-Ohmic contacts. The inset shows emission spectra from as-fabricated LEDs at various injection currents................................................................................................. 129 6-3 Influence of long-term aging at 250C and 350C on the I-V characteristics of LEDs with (a) Ni/Au, (b) Ni/Au/TiB2/Ti/Au, and (c) Ni/Au/Ir/Au p-Ohm ic contacts............. 130 6-4 Image of aged LEDs with (a) Ni/Au and (b) Ni/A u/TiB2/Ti/Au p-Ohmic contacts. In (a), the picture was taken for a forward bias of 10 V ( I = 80 A), while the forward voltage in (b) was 4.5 V ( I = 300 A)............................................................................. 131

PAGE 11

11 Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy THERMALLY STABLE OHMIC AND SCHOTTKY CONTACTS TO GaN By Lars Fredrik Voss May 2008 Chair: Stephen J. Pearton Major: Materials Science and Engineering This dissertation is focused on the developmen t of Ohmic and Schottky contacts to both nand p-type Gallium Nitride for use in microelectr onic and optical devices. The goal is to develop low resistance contacts with greater thermal budgets and superior thermal aging ch aracteristics to those commonly in use today as well as to unde rstand the mechanisms by which these contacts may fail. In addition, p-type Ohmic contacts ha ve been used to fabricate light emitting diodes (LEDs) which display far superi or aging properties than those made with conventional Ni/Au contacts. Ohmic contacts to p-GaN were fabricated using a variety of refractory materials. The materials examined were of three basic types: bo ride, nitride, and the refractory metal Ir. The boride family includes W2B, W2B5, CrB2, ZrB2, and TiB2. The nitrides examined were TaN, TiN, and ZrN. Contacts based on these ma terials were fabricated using either a GaN//Ni/Au/X/Ti/Au, GaN//X/Ti/A u, or GaN//Ni/X/Au scheme, where X is the refractory material. Contact resistances as low as ~1 x 10-4 /cm2 were consistently achieved after annealing at temperatures from 500-1000oC for 60 s in N2 using these materials for p-GaN with a carrier concentration of ~1 x 1017 cm-3. In addition, high temperature thermal aging was

PAGE 12

12 performed on selected schemes as well as on devi ces in order to attempt to estimate long term performance. Nest, nitride-based Ohmic cont acts to n-GaN are examined. The nitride was used to replace the conventional Pt, Ni, or Mo diffusion barrier in Ti/A l based contacts, for a contact scheme of GaN//Ti/Al/X/Ti/Au. These contacts ac hieve a similar specific contact resistance of ~1 x 10-4 /cm-2 for samples with a carrier concentration of 1 x 1017 cm-3 as that achieved with a Ti/Au/Pt/Au contact. The contacts are also ex amined as a function of aging and are found to display less intermixing of layers than t hose fabricated with a Ni diffusion barrier. Schottky contacts to p-GaN were also fabricated using the family of boride based materials with the scheme GaN//X/Pt/Au. These were f ound to exhibit tunneling transport through the metal-semiconductor junction and barri er heights of >3.5 eV were de termined from the fit of the current-voltage (IV) curves. X-ray Photoelect ron Spectroscopy was used to determine a true barrier height for these borides on p-GaN of 2.7 eV, close to that expected from the SchottkyMott model. Capacitance-voltage (CV) measuremen ts confirm the IV barrier height, but reveal a thin interfacial layer likely arising from oxidation or defects at the surface.

PAGE 13

13 CHAPTER 1 INTRODUCTION Se miconductor technology and its rapid improvements in device speed, yield, and density have enabled much of the technology and advan cements the world has experienced over the past half century. Beginning with the first transistor deve loped in 1947 at AT&T Bell Labs and then in 1957 with the first integr ated circuit by Jack Kilby, the pace of improvement in semiconductors and integrated circuitry is unmat ched by any other techno logical development in human history, and few have been more influential. Today, what once seemed ridiculously optimistic is possible and billions of transistors are present in a single integrated circuit [1]. Silicon based technology has come to dominate an overwhelming share of the semiconductor market, and for good reason. The main advantage of Si-based technology is the overwhelming knowledge base that exists as well as the f act that is possesses a wide range of acceptable properties. It possesses a higher melting point th an Ge, from which the first devices were made, as well as a stable oxide which di splays low leakage current. Furt her, nearly all of its properties fall into the acceptable range necessary for use in nearly all applications, including a reasonable band gap, carrier mobility, and breakdown field. For certain areas of application, however, Si is unacceptable. For applications requiring higher speed electronics, GaAs dominates the mark et. This is due to a much higher electron mobility, greater than 6 times that of Si. In a ddition, Si is also inappr opriate for optoelectronic applications due to its direct band gap. GaAs-based devices are also useful in this area. However, for high power, high temperature electronics as well as emission of wavelengths in the blue to ultra-violet range, GaAss properties are insufficient. It is therefore necessary to develop fabrication techniques for alternatives to Si and GaAs [2,3]. For these applications GaN is the material of choice. Its wide, direct band gap of 3.4 eV allows for emission in the UV range,

PAGE 14

14 while the use of InGaN and AlGaN allow for tu nable emission over the range 1.92 to 6.2 eV. [437] Development of GaN-based electronics has been necessitated due to the limitations of SiC, another semiconductor used for high power and high temperature electronics. SiC MOSFETs are limited by primarily thermal oxides, as the ga te contact degrades and becomes leaky at higher operating temperatures. In addition, th e electron mobility is low, only ~400 cm2/V-s, resulting in low power-added efficiencies of less than 30% for operation between 1 a nd 5 GHz. Because of its nature, GaN is a superior choice due to its wi der bandgap as well as greater chemical stability, leading to improved performance at higher operating temperatures. GaN based electronics are already in pr oduction, most notably high electron mobility transistors (HEMTs). Other devi ces of interest include heterojunction bipolar transistors (HBTs) and metal oxide field effect transistors (MOSFETs). In order to fully realize the potential of GaN-based electronics, improvements in their pr ocessing must be achieved in addition to the significant improvements in material quality that are necessary. This includes improvements in high temperature processing and operation, one area of which is the development of both Ohmic and Schottky contacts which are able to withstand high temperatures without degradation of device characteristics or contact structure [4]. This becomes especially important as the quality of the materials improves and fa ilure is due to the contacts. While electronic applications for GaN are important, optoelectronic applications are currently dominant. As Figure 11 indicates, most of the current and projected market are for optical applications. The first bright blue LED using GaN to be put into production was demonstrated by Shuji Nakamura in 1993 at Nichia Corporation. Since that time, a great deal of progress has been made in the attempt to realize the full potential of GaN for light emitting

PAGE 15

15 applications. The promise of solid state lighti ng is one that grows more important with each passing year, as the worlds insatiable hunger fo r energy grows while reserves of conventional fossil fuels struggle to keep up. Because produc tion cannot indefinitely keep up with demand, focus should and is on increasing the efficiency of consumption. Nearly 10% of world energy consumption is used for lighting. Most of this is currently done with extremely inefficient incandescent bulbs. While compact fluorenscent lighting has grown more popular and offers a substantial improvement in energy efficiency, it cannot compete with that possible with the wide spread use of LEDs. Further, CFBs present a hazard due to the non-trivial amount of mercury present in each bulb. Because of this, any large scale use of the bulbs must go hand in hand with proper disposal of them in order to prevent co ntamination of the environment. Solid state lighting does not present this problem. While many challenges still exist to realiz ing commercially viable solid state lighting sources, including significant improvement in the efficiency of the green emitting alloys as well as possible improvements in the phosphors used to emit the yellow and white light desired, it is still important to develop improve d contact structures in order to allow for longer life times and operation in environments with higher temperatures For p-type GaN especially, existing contact schemes could be improved to offer more stable performance. Failure of GaN-based devices due to contact failure should be minimized or eliminated. The goal of this work is to develop impr oved contact schemes fo r both electronic and optoelectronic applications that will offer acceptable specific c ontact resistances and barrier heights which will be stable over a wide range of processing temperatures as well as during long term high temperature aging. Chapter 2 deals wi th the basic properties of GaN as well as an overview of the processing and char acterization techniques that were utilized during this work.

PAGE 16

16 The rest of this dissertation is devoted to the ex periments and analysis. First, Ohmic contacts to p-GaN using a variety of high temperature material s are discussed. The ne xt section deals with Ohmic contacts to n-GaN. A section on Schottky contacts to p-GaN follows. The final chapter discusses the use of the Ohmic contact s to p-GaN in light emitting diodes.

PAGE 17

17 Figure 1-1 Market forecas t for GaN-based devices MARKET FORECAST FOR GAN-BASED

PAGE 18

18 CHAPTER 2 BACKGROUND 2.1 Gallium Nitride Properties 2.1.1 Fundamental Properties The m aterials properties of GaN make it desirabl e for a variety of applications. For optical electronics, GaN holds great promise. Because it is a direct band gap semiconductor, it is an efficient light emitter. GaN fills the need for a bl ue and ultraviolet wavelength emitter that is left by other materials due to its ba nd gap of 3.475 eV. Its exciton binding energy is 28 meV. This is not as high as another blue emitters, ZnO, but GaN technology is more developed. The first GaN LED was introduced in 1993; however, fu rther development of the GaN system is necessary for better LEDs and LDs. In addition to optical applications, GaN is robust and therefore a candidate for high temperature and high power appli cations, with potentially superi or electronic properties when compared with SiC [38]. The reason for this is due to the relatively wide band gap when compared to Si and GaAs. Figure 2-1 displays th e intrinsic carrier concentration as a function of temperature of GaN, Si, and GaAs. Because of the wide band gap, the intrinsic carrier concentration is much lower; 1015 cm-3 is reached at 300C for Si, 500C for GaAs, and at 1000C for GaN. Because of this, GaN can be operated at much higher temperatures before device performance degrades and breaks down. Table 2-1 contains some basic physical prope rties of GaN [39,40]. A key problem in device operation is the presence of the lattice mismatch of GaN with the commonly used sapphire (13%) and SiC (3%) substrates, which results in a large number of lattice defects which hinder electrical conduction.

PAGE 19

19 2.1.2 Electronic Properties The properties GaN possesses are advantageous f o r a variety of reasons. They, as well as those of other common semiconductors, are disp layed in Table 2-2 [40,41,42]. It has a high breakdown field, greater than 50 time s that of Si or GaAs, which allows for its use in high power electronic applications. The high electron mobility for its 2D elect ron gas and saturation velocity allows its use in high speed electronics. In a ddition, heterostructures su ch as AlGaN/GaN allow for the manufacture of interesting and high sp eed devices such as HEMTs. Recent work suggests that the 1.5 W/cm may be a lower limit for the thermal conductivity of GaN. GaN also shows excellent resist ance to irradiation. 2.1.3 Crystal Structure The therm odynamically stable phase of GaN is the hexagonal wurtzite structure, referred to as the -phase. In addition, a metastable -phase with a zinc blende structure exists. These two phases only differ in the stackin g sequence of Ga and N; their co existence in epitaxial layers is possible due to stacking faults [43]. The lattice parameters are a = 3.189 and c = 5.185 The wurtzite structure of Ga-face a nd N-face GaN is shown in Figure 2-2. 2.2 Properties of the Contact Materials to be Studied 2.2.1 Borides The fa mily of borides to be studi ed includes the refractory compounds W2B, W2B5, CrB2, TiB2, and ZrB2. Previous work has shown the promise of using refractory materials such as W as contacts to p-GaN. These borides also possess many of the requisite characterstics for use as electrical contacts fo r semiconductor devices. Table 2-3 summarizes known properties of the borides to be studied. Each of the borides possesses a high melting temperature, good thermal and electrical conductivities, high work functions, and large heats of formation. Each of these properties is important when considering a materi al for use in electrical contacts. The high

PAGE 20

20 melting temperature is necessary in order to ensure that the material is able to withstand the elevated annealing temperatures which are generally necessary for Ohmic contact formation. A high melting temperature is an indication of mate rial stability as well. A large electrical conductivity is necessary so as to achieve a low re sistance within the contact itself as well as to avoid too much self heating, which can lead to inconsistent performance and diffusion between contact layers during operation. A large therma l conductivity is necessary for this reason as well. For a p-type material, a la rge work function is desirable in order to achieve the proper band bending for thermionic carrier transpor t or at least to minimize the barrier for tunneling transport. Finally, a large heat of formation is desira ble in order to minimize reactions at the semiconductor-contact interface during operation. These materials have previously been examined for use in contacts to n-GaN and have displayed good characte ristics as a diffusion barrier in a Ti/Al-based contac t compared to more conventional metals such as Pt or Ni. 2.2.2 Nitrides The second class of m aterials to be studied is the nitrides. Th e materials chosen were TaN, TiN, and ZrN. Table 2-4 summarizes the properti es of these nitrides. They possess many of the same characteristics as the borid es and are therefore also worthwhile to study. These include a high melting temperature, good thermal and electri cal conductivities, a wo rk function, and large heats of formation. These nitrides have been stud ied for use in thin film resistors, wear-resistant coatings on tools, thermal printer heads, ga te electrodes, and diffusion barriers in Cu interconnection, which suggest their suitability as diffusion barrier s for GaN-based electronics. They will be examined in this case for Ohmic contacts to both nand p-type GaN. 2.2.3 Iridium The f inal material to be studied for use in contacts to p-GaN will be Iridium. Properties of Ir are summarized in Table 2-5. Note that Ir ha s the required high melting temperature, large

PAGE 21

21 work function, and reasonable thermal and electri cal conductivities. Ir has previously been studied for use as a contact to p-GaN. Here, it will be examined for use in Ni-based contacts to p-GaN as an intermediate layer. It has already been demonstrated that Ir acts a superior diffusion barrier in Ti/Al-based contact s to n-GaN compared to Ni. 2.3 Electrical Contacts High quality electrical contacts to sem iconductors are cri tical for the manufacture of all types of devices. While the requirements for spec ific devices and applications may differ, it is always necessary to connect these devices to the outside world. Generally, a contact that displays the lowest specific contact resistan ce with minimal drift during extended operation is the most desirable, although for certain applications other properties may be more critical. For instance, in light emitting devices it is often desi rable to have either transparent contacts or contacts with high reflectivities in the range of emission wavelengths in order to maximize light output. For high temperature applic ations it is critical that cont acts are stable during operation at high temperature; thus materials which displa y little intermixing and minimal reactions are necessary in addition to the requirement of good electrical and thermal conductivities. Electrical contacts to semiconductors are generally understood as th e junction between the semiconductor and a metal or other contact materi al, as well as any layers of metal or other material above this. The electrical properties of the contact are controlle d by the materials at the junction and it is often necessary to anneal the contacts at elevated temperatures in order to achieve the required properties th rough elimination of defects, re moval of compensators such as Hydrogen, or formation of low resistance intermetal lic phases. In addition, the layers above the junction itself can also influence the performance of the contact. It may be necessary to use an overlayer such as Au in order to promote current spreading so that the conduction is spread evenly over the area of the cont act as well as to prevent oxidation of the contact at room

PAGE 22

22 temperature. Other materials may be used as di ffusion barriers in order to minimize intermixing between various contact layers which may lead to the formation of undesirable secondary phases, segregation of layers when it is desirable that they be interm ixed, or even to prevent outdiffusion of the semiconductor to the surface of the c ontact. Two types of c ontacts exist, and it is important to understand what mate rials properties may be important in order to fabricate each. 2.3.1 Ohmic Contacts Ohm ic contacts are those which obey Ohms Law: V = IR(2.1) Note that this is a linear rela tionship and thus current is allowe d to flow into and out of the semiconductor without distortion of the signal. In general, the most critical component of an Ohmic contact is a low specific co ntact resistance. This will mi nimize the power consumption of the contact, which will also serve to minimize the internal heating of the contact and thus allow for prolonged operation with predic table performance. An ideal Ohmic contact would have no noticeable effect on device performance, as it would consume none of the power in the system nor would it cause any rise in temperature during operation. In theory, fabrication of Ohmi c contacts should be straightfo rward. For semiconductors with moderate doping, the dominant mechanism of current flow is generally thermionic emission, which is governed by Equation 2-2: kT q TqA kb c exp**(2.2) where k is Boltzmanns constant, q the electronic charge, A** is the Rich ardsons constant, T the temperature, and b the barrier height. In this model, el ectrons are excited over a barrier. It is clear, then, that the barrier height should be sm all in order to produce a low contact resistance. For p-type conduction, a metal with a work function larger than that of the semiconductor should

PAGE 23

23 be sufficient. The desired flat band diagram for th is situation is shown in Figure 2-3 [40]. It is clear from this diagram that holes may pass eas ily between the metal and the contact at the junction. However, for some semiconductors it can be di fficult to find materials systems that produce an Ohmic contact through thermionic emission, especia lly for p-type semiconductors. A wide band gap and large electron affinity, su ch as with GaN, effectively limits the choices to materials which have a very high work function. A second approach for formation of Ohmic contacts to semiconductors utilizes another form of carrier tr ansport: tunneling. This type of transport is characterized by the carriers t unneling through the potential barri er at the semiconductor-contact junction. Equation 2-3 describes the specific contact resistance for this type of transport: D bs cN m *2 exp (2.3) The relevant terms to consider in this equati on are the barrier height and the doping density, ND or NA for p-type semiconductors. It is obvious that in order to achieve a low contact resistance the barrier height must be minimized and the dop ing density must be maximized. Barrier height is determined by material choice. Doping density can be determined during growth of the material, with ion implantation, or use of contacts which can increase doping through interactions with the semiconductor. 2.3.1.1 Ohmic contacts to p-GaN High quality, low resistance Ohmic contacts to pGaN are difficult to achieve. This is due to several factors: the relative dearth of appropriate metal systems due to the large work function of GaN, low hole densities due to the difficu lty in achieving a high dopi ng concentration during growth, the presence of compensating hydrogen interstitials, and the pr esence of compensating

PAGE 24

24 nitrogen vacancies. Despite this, several ma terials systems have been found to produce Ohmic behavior. These are generally based on a high work function metal such as Ni, Pd, or Pt with an overlayer of Au and have been found to achieve specific contact resistances in the range of 1 x 10-4 -cm2 for hole concentrations of ~1017 cm-3 [44-48]. Ni/Au contacts are the most common. The Ni serves to remove hydrogen from the near surface region of the p-GaN at elevated annealing temperatures, as the Ni hydride is thermodynamically favor able. In addition, at these elevated temperatures Au begins to diffuse in towards the GaN and it has been found that the Au can form intermediate phases with th e Ga in the form of AuGa and AuGa2 [49]. This can serve to further increase the near surface doping thro ugh the formation of Ga vacancies in the GaN which act as acceptors. However, at these anneal temperatures of > 500oC, the Au and Ni may completely exchange places at the GaN surface [44] This leads to a severe roughening of the contact surface and degraded performance and limits the temperature at which anneals may take place as well as the lifetime of the contacts when operated at high temperature or, equivalently, high powers. Because one of the most attractive f eatures of GaN is its abili ty to operate reliably under these conditions, it is important to find a contact system that will perform better. 2.3.1.1 Ohmic contacts to n-GaN Ohmic contacts to n-GaN are more studied, better understood, and easier to achieve compared to those to p-GaN. The most common materials system used is GaN//Ti/Al/X/Au, where X is a material chosen to act as a diffusi on barrier to separate the Al and Au. This is necessary, as Al and Au will react to form AlAu4 [50]. This phase is viscous at low temperatures and lateral flow can create problems, especially as the gate/source separation becomes small. Because of this, there is great inte rest in the use of high temperature metals. The Ti is present at the junction in order to form TiN through reac tion with the GaN, which creates nitrogen

PAGE 25

25 vacancies and thus increases the donor density. The Au overlayer is present in order to prevent oxidation as well as to promote cu rrent spreading in the contact. 2.3.2 Schottky Contacts The other type of electrical contact required for electronic devices are known as Schottky contacts or diodes. Schottky contacts allow l ittle or no current to flow through them until a critical voltage is reached, above which large amount s of current flow. This is called the forward biased region. In the reverse bias region, vol tage is applied in th e opposite direction and no current flows initially. However, if the voltage is pushed high enough then breakdown occurs, a huge amount of current flows through the contact and the contact is generally destroyed. Ideal Schottky contacts are formed in the abse nce of any surface states. When a metal and semiconductor are brought together, the Fermi levels must line up. In order for this to occur, a charge exchange occurs between the two materi als. When this occurs, a space charge or depletion region is formed and a barrier to current flow is creat ed. For p-type semiconductors, the flat band diagram is shown in Figure 2-4. The barrier height, bp, for a p-type semiconductor is given by the equation qbp = g -q ( m ) (2.4) where Eg is the band gap, m is the metal work function, and is the electron affinity of the semiconductor. Invariably, however, surface states are present. This makes the prediction of barrier height using the equation above inaccurate. In order to de termine barrier height in this case, it is necessary to fabricate a specimen to test. The predominant carrier in Schottky contacts is the majority carrier unlike a pn junction diode. Two mechanisms for current flow exist: thermionic emission (TE) and thermionic field emission (TFE) or tunneling. While some authors have reported values for barrier heights on p-

PAGE 26

26 GaN using the TE model, most of the work has de monstrated that TFE is the dominant transport mechanism. Reported barrier heights on p-GaN ra nge from ~0.5 to 2.9 eV even for the same metallization scheme because the mechanism of cu rrent flow is not definitively established [5165]. In the case of TE, electrons or holes are exc ited over the potential barrier and the forward current is given by the following expression: *2expexpB F BBeeV JAT kTnkT (2.5) where A*=103.8 Acm-2K-2 is the effective Richardsons constant for p-GaN, T is the absolute temperature, e is the electronic charge, B is the SBH, kB is the Boltzmanns constant, and V is the applied voltage. In the case of TFE, quantum tunneling of char ge carriers through the ba rrier occurs and the forward current can be described by 0 0expFqV JJ E (2.6) where the saturation current density J0 is given by 0.5 00 0 00exp coshB B BBBBATEqV q q J kEkTkTkT (2.7) In Eq. (2.7), =( EFEV)/q is the difference between the valence band maximum and the position of the Fermi level and 00000cothBEEEkT is the characteristic energy related to the tunneling probability.

PAGE 27

27 2.4 Experiments The goal of this work is to find suitable cont act systems for use at high temperatures on nand p-GaN. Towards this end, the previously disc ussed materials will be investigated in order to determine which of them allow good quality Ohmi c and Schottky contacts. In addition, a study will be made of devices which use p-GaN, mainly LEDs. The contacts will be aged at high temperatures and compared with more conventional schemes. A variety of techniques will be used to characterize the contacts and devices in order to determine their performance and to better understand what is happening. The main experiments that will be done are the processing of the Ohmic and Schottky contacts for electrical characterization. The following sc hemes will be produced and analyzed: 1. GaN//Ni/Au/X/Ti/Au for p-type Ohmic contacts, where X indicates the refractory material to be examined 2. GaN//X/Ti/Au for p-type Ohmic contacts 3. GaN//Ni/Ir/Au for p-type Ohmic contacts 4. GaN//Ti/Au/X/Ti/Au for n-type Ohmic contacts 5. GaN//X/Pt/Au for p-type Schottky contacts In case (1), the Ni/Au should act as an Oh mic contact while the boride layer provides a diffusion barrier to prevent breakdown of the meta llization at elevated temperatures. The Ti layer is present in order to promote adhesion of the Au to the rest of the contact. The Au overlayer is to prevent oxidation an d act as a current spreading layer. In case (2), it is hoped that at least some of the borides may provide an Ohmic contact on p-GaN. Case (3) may provide some insight into the importance of Au at the GaN surface for these types of contacts. Case (4) is based on standard Ohmic contacts to n-GaN, with a nitride used as the di ffusion barrier. Case

PAGE 28

28 (5) will be used to fabricate Schottky contacts, as the Pt should not be as reactive as Ti should it reaches the GaN surface and is therefore used as the adhesion layer. Samples will be prepared on p-GaN and nGaN wafers grown in an MOCVD reactor. PGaN will be 3 m thick Mg-doped and grown on c-plane Al2O3. Hole concentration after activation obtained from Hall measurements is ~1 x 17 cm-3. The same doping will be used for both Ohmic and Schottky contacts. This is because it is extremely difficult to obtain a higher doping level for use in Ohmic contacts. At lo wer doping concentrations, it is not possible to obtain an Ohmic contact for use in Schottky contacts. For Ohmic contacts to n-GaN, epilayers 3 m thick Si-doped and grown on c-plane Al2O3 with an electron concentration of ~1 x 1017 cm-3 will be used. In order to fabricate the samples, photolit hography will be performed using a Karl Suss MJB3 aligner using positive photoresists. Samples for Ohmic contacts will then be etched in a Unaxis ICP tool in order to isolate devices and pr event current spreading. Prior to patterning for the metal layers, p-GaN samples will be dipped in a 1M KOH solution for one minute in order to remove surface oxide. Following patterning, su rface treatments may be necessary to improve performance. A 30 s RIE O2 plasma exposure to remove carbon followed by a 10% HCl dip for one to ten minutes immediately prior to insert ion into the deposition chamber will be employed to ensure a clean surface. Initial Ni/Au layers for contacts to p-GaN will be deposited using an e-beam evaporator in order to minimize surface damage. For deposition of all other layers, a Kurt Lesker sputterer will be employed. This is necessary, as the refractory materials cannot be evaporated. In addition, it is necessary to depos it all overlayers in the sputterer in order to minimize oxidation. Following lift off of the metal layers, rapid thermal annealing will be

PAGE 29

29 performed at a range of temperatures from 3001000C in order to determine how the properties of the contacts are affected at a variety of annealing temperatures. Following the development of these contacts, de vices will be fabricated. While p-GaN is unsuitable for use in HEMTs due to its low hole mobility, it is necessary for use in LEDs, LDs, and HBTs. LEDs will be fabricated on wafers w ith the structure shown in Figure 2-5. These will be compared to devices using more traditional metal schemes such as Ni/Au. They will then be thermally aged in order to determine the perfor mance relative to Ni/Au when maintained at an elevated temperature. 2.5 Characterization Techniques A variety of techniques will be utilized in or der to determine device characteristics as well as to observe the physical and chemical structure of the contacts. The techniques to be employed are 1. Current-voltage measurements 2. Capacitance-voltage measurements 3. X-ray Photoelectron Spectroscopy 4. Auger Electron Spectroscopy 5. Scanning Electron Microscopy 2.5.1 Current-Voltage Current-voltage (IV) measurements will be performed using an Agilent 4156 Semiconductor Parameter Analyzer. Two probe meas urements will be perfor med when possible. However, if problems which make it impossible to extract contact resistance arise due to high sheet resistance and the resistance of the probes, a four probe measurement technique will be employed. Using this method, current will be in jected with two outer probes while the voltage drop is measured between two inner probes. Th is allows for measurement of only the voltage

PAGE 30

30 drop between pads. IV measurements will be the main technique employed to determine the suitability of both Ohmic and Schottky contacts. The IV curves allow for the extraction of contact resistance for the former and barrier height and reverse breakdown voltage for the latter. IV measurements can be performed using either a circular or linear tr ansmission line method. The CTLM is useful for materials which are di fficult to etch and therefore may demonstrate problems with current spreading. In this study, linear TLM patterns will be used, as shown in Figure 2-6. Data from these curves allow for the creation of a plot such as that shown in Figure 2-7. Characterization of the pads is then performed using the relation S C C SC TR WR W L RRR222(2.8a,b) Figure 2-8 shows a schematic for Schottky cont acts. Schottky pads are formed using an inner Schottky contact with an outer Ohmic contact. 2.5.2 Capacitance-Voltage Capacitance-voltage (CV) measurements will be employed on Schottky contacts in order to further confirm results of IV measurements as well as to provide a deeper understanding of the barrier. If an interstitial la yer exists between the contact and the p-GaN, utilizing a CV measurement can reveal this. This is a distinct possibility that could arise either from an interstitial oxide layer forming or from def ects at the surface possibly caused by sputtering damage. Measurements were performed using an Agilent 4284A precision LCR meter in the parallel mode. The modulation frequency was set to 1 kHz. Barrier height is determined using

PAGE 31

31 lnBintABVEkTqg(2.9) where Vint is the extrapolated intercept volta ge of the reverse bias in the 1/ C2 versus V plot, EA 0.12 eV is the activation energy of Mg dopants in p-GaN, and g=2 is the degeneracy factor for acceptors. Note that in our case, the series resistance (~2 k ) and the junction conductance (~10-8-10-9) were low enough so that the measured capacitance corresponds to the proper junction capacitance. 2.5.3 X-ray Photoelectron Spectroscopy X-ray Photoelectron Spectroscopy (XPS) measurements will be further used to confirm the barrier height of Schottky cont acts. XPS functions by bombardi ng a specimen with soft x-ray radiation to examine core energy levels. When an x-ray collides with an atom, an electron is ejected from its shell. The elec tron is then collected and its energy measured. Each element has a characteristic binding energy for its core electro ns, and thus each element has a characteristic spectrum. Comparison of the spectrums of a bare specimen, such as GaN, and a specimen with a thin film coating, such as GaN//W2B, can yield the barrier height of the surface. The barrier height can be determined from the binding energy of the Ga 3d core level EB and the energy difference between that core le vel and the valence band maximum EVC according to BBVCEE (63,64)(2.10) 2.5.4 Auger Electron Spectroscopy Auger Electron Spectroscopy (AES) will be used to determine the elemental composition versus depth of the contacts. This will allow fo r an understanding of how diffusion plays a role in the behavior of the contacts. AES functions by bombarding the surface of a specimen with a

PAGE 32

32 beam of focused electrons. These electrons pene trate into the sample and collide with the atoms of the specimen. A core shell electron is then ejected, creating a vacancy. Following this, an electron from an outer shell relaxes and fills this lower energy state, releasing energy. This excess energy then results in the ejection of an el ectron from an outer shell. This Auger electron possesses an energy characteristic of the element from which it is ejected, allowing for the composition of the specimen to be determined. Using an ion gun in combination with the electron beam allows for information about th e composition versus depth to be collected.

PAGE 33

33 Table 2-1: Bulk GaN properties Property Value Lattice parameters at 300 K (nm) ao = 0.3189 nm co = 0.5185 nm Density (g-cm-3) 6.095 g-cm-3 Stable phase at 300K Wurtzite Melting point (oC) 2500 Thermal conductivity (W/cm-K) 1.3, 2.2 +/0.2 for thick, free-standing GaN Linear thermal expansion coefficient Along ao = 5.59x10-6 K-1 Along co = 7.75x10-6 K-1 Static dielectric constant 8.9 Refractive index 2.67 at 3.38 eV Energy bandgap (eV) Direct, 3.45 Exciton binding energy (meV) 26 Electron effective mass 0.20 Hole effective mass 0.59

PAGE 34

34 Table 2-2: Properties of common semiconductors Si GaAs GaN AlN 6H-SiC Bandgap (eV) @300oC 1.1 indirect 1.4 direct 3.4 direct 6.2 direct 2.9 indirect Electron mobility (cm2/V-s), RT 1400 8500 1000 (bulk) 2000 (2DEG) 135 600 Hole mobility (cm2/V-s), RT 600 400 30 14 40 Saturation velocity (cm/s), 107 1 2 2.5 1.4 2 Breakdown field (V/cm) x 106 0.3 0.4 >5 4 Thermal conductivity (W/cm) 1.5 0.5 1.5 2 5 Melting temperature (K) 1690 1510 >1700 3000 >2100

PAGE 35

35 Table 2-3: Properties of the borides Properties TiB2 ZrB2 W2B W2B5 CrB2 Melting Point (oC) 2980 ~3225 3040 ~3200 ~2670 ~2385 2200 Structure Hexagonal Hexagonal Hexagonal Hexagonal Thermal expansion coefficients x 106 (/deg) 4/6 5.9 10.5 Phonon component of heat conduction (wt/m-deg) 20.6 18.9 10.4 Elastic modulus E x 10-6 (kg/cm2) 5.6 4.3 2.5 Characteristic Temperature (K) 1100 765 726 Density of electronic state g x 10-21 (eV-1 cm-1) 4.5 4.76 54.6 Work function (eV) 4.19(?) 3.94(?) 3.18(?) Heat of Formation (kcal/mol) 71.4 76 31 Lattice constant (A) 3.028 3.169 2.982 2.969 Thermal conductivity (W/m-K) 26 80 Unknown 32 Electrical resistivity (uOhm-cm) 28 4.6 19 21

PAGE 36

36 Table 2-4: Properties of the nitrides Properties TaN TiN ZrN Melting point (K) 3633 3203 3253 Structure FCC FCC FCC Thermal expansion coefficients x 106 (/deg) 3.6 9.4 8.7 Work function (eV) 4.7 3.74 4.6 Lattice constant () 4.3 4.32 4.574 Thermal conductivity (W m-1 K-1) 57.5 19.2 20 Electrical resistivity ( cm) 15.52 25 13.6

PAGE 37

37 Table 2-5: Properties of Ir Properties Ir Melting point (K) 2719 Structure FCC Thermal expansion coefficients x 106 (/deg) 6.4 Work function (eV) 5.27 Lattice constant () 3.8 Thermal conductivity (W m-1 K-1) 147 Electrical resistivity ( cm) 4.71

PAGE 38

38 Figure 2-1 Intrinsic carrier concen tration of GaN, GaAs, and Si 1.52.02.53.03.54.0 10-2010-1010010101020 400300200100 0 GaN GaAs SiIntrinsic Conc. (cm-3)1000 / T (K-1) 25Temperature (oC)

PAGE 39

39 Figure 2-2: Stable wurtzite GaN crystal

PAGE 40

40 Figure 2-3: Flat band diagram for a p-type Ohmic contact

PAGE 41

41 Figure 2-4: Flat band diagram for a p-type Schottky contact

PAGE 42

42 (a) Sapphire GaN buffer n-GaN InGaN/GaN MQW p-GaN p-AlGaN n-AlGaN Sapphire GaN buffer n-GaN InGaN/GaN MQW p-GaN p-AlGaN n-AlGaN (b) Sapphire GaN buffer n-GaN InGaN/GaN MQW p-GaN p-AlGaN n-AlGaN p-contact n-contact Sapphire GaN buffer n-GaN InGaN/GaN MQW p-GaN p-AlGaN n-AlGaN p-contact n-contact Figure 2-5: LED cross section (a) before and (b) after processing

PAGE 43

43 Figure 2-6: Linear tran smission line pattern Metal Pads Semiconductor film L1L2L3W Metal Pads Semiconductor film L1L2L3W

PAGE 44

44 Figure 2-7: Resistance vs. pad spacing plot Distance LResistance R L1L2L3Slope=RS/W 2Rc Distance LResistance R L1L2L3Slope=RS/W 2Rc

PAGE 45

45 Figure 2-8: Schottky contact schematic Schottky contact Ohmic contact Semiconductor film Schottky contact Ohmic contact Semiconductor film

PAGE 46

46 CHAPTER 3 THERMALLY STABLE OHMIC CONTACTS TO P-GAN 3.1 Ohmic Contacts In this section, two approaches to fabricat ing Ohmic contacts using the boride family of materials will be discussed. The first is as an overlayer on conventional Ni/Au contacts and the second as the actual contact material. Because th ese borides have a large work function, it is expected that both approaches will be successf ul in yielding Ohmic be havior, provided that sputter damage and oxygen incorporation in to the borides duri ng deposition does not compromise the contacts, especially the latter. 3.1.1 Fabrication of Ohmic Contacts The samples used were light-emitting diode wafers with a 0.1 um Mg-doped p-GaN layer on top of a 0.1 um p-AlGaN layer followed by a 0.3 um InGaN/GaN superlattice on top of a ntype GaN layer grown on 2 um-thick undoped buffers on c-plane Al2O3 substrates. Hole concentrations in the p-GaN layer were 1017 cm-3 obtained from Hall measurements. Mesas for the linear transmission line method ( LTLM) method were etched using a Cl2/Ar inductively coupled plasma for pad isolation and to minimi ze current spreading. Samples were dipped in 1 M KOH for 1 minute and then rinsed in acetone a nd ethanol prior to photo lithography. Prior to insertion into the evaporation chamber, the samples were exposed to a 30 s O2 RIE plasma in order to remove any residual ca rbon and then dipped in 10:1 H2O:HCl for 5 minutes, which serves to remove any native ox ide. A Ni/Au contact of 500 / 800 was deposited using an ebeam evaporator in order to minimize surface damage. After unloading, the samples were put into a sputtering system for deposition of the X/Ti/Au laye r of thickness 500 /200 /800 where X is the refractory metal used. Au is n ecessary to decrease the sheet resistance while the Ti is necessary to prevent peeling of the Au fr om the surface due to the large lattice mismatch.

PAGE 47

47 In the case of Ir, this Ti layer is unnecessary. Contacts were also fabricat ed of the form Ni/Ir/Au, in order to confirm the importance of the Au n ear the GaN surface, and of the form X/Ti/Au to investigate the use of the refractory materials without the increased dop ing present due to the Ni/Au layer. Sputtering was perf ormed using an Ar plasma-assisted rf sputter at 5 mTorr and rf (13.56 MHz) power of 150 W. The sputter rates were 5 A/s for Au, 0.5 A/s for Ti, and were 0.82 A/s for the refractory materials. Following lift-off, the contacts were annealed in a flowing N2 ambient for 60 s in a rapid thermal ann ealing furnace at temperatures up to 1000 oC. 3.1.2 Nitride-Based Contacts 3.1.2.1 Experiment and discussion Ohmic contacts based on Ni/Au but with a nitr ide overlayer were fabricated in order to improve the temperature stability. Figure 3-1 show s the specific contact resistivity of the nitridebased contacts as a function of annealing temperature, as well as the sheet resistance of the pGaN under the contacts. At annealin g temperatures of less than 500 oC, contacts were rectifying. The contacts transition to Ohmi c behavior at higher temperatures. The specific contact resistance was not a function of temperature in the range 25-175oC, indicating that tunneling is the dominant transport mechanism. The specific c ontact resistance is fair ly stable on annealing up to 1000 oC, with a minimum of ~2 x 10-4 -cm2 at 600oC for the Ni/ Au/ ZrN/ Ti / Au contacts. This indicates that th ere is minimal change in the dopi ng of the near surface region of the GaN at these elevated anneal temper atures. The formation of AuGa and AuGa2 occurs at 600oC at the interface 49]. A sharp maximum in the sheet resistance is observed at 800 oC, with a drop at higher temperatures. Th is is likely due to increased in-diffusion of metal to the GaN without complete alloying, resulti ng in interstitial atom incorporat ion into the lattice. The drop in sheet resistance at higher temperatures is an indication that these interstitials are increasingly incorporated into the GaN. The magnitude of the increase in sheet resistance is related to the

PAGE 48

48 size of the metallic atom in the nitride, with the largest increase seen with the smallest atom Ti and vice versa for Ta, suggesting the atoms causing the spike are from the nitride. The reason the smaller atoms cause a more pronounced increase is the greater ease of diffusion into the GaN as well as a likely higher solid solubility. Figure 3-2 shows the AES depth profiles for the TaN-based contacts under a variety of conditions. Upon annealing at intermediate temper atures, most of the Ti has migrated to the surface and oxidized while Ni and Au are both pres ent at the GaN interface. In addition, the nitride layer is present at the GaN surface, sugges ting that the increase in sheet resistance is due to incorporation of the metallic atoms from the nitride into the GaN. After annealing at 700 oC and aging at 200 oC or annealing at 1000 oC, there is a large amount of intermixing between the metals and the GaN. Note that the upon annealing at 600oC the Ta and N peaks are still correlated, indicating that TaN ha s not reacted a great deal with the GaN yet. However, in the profile of the thermally aged c ontact there is no longer a correlat ion between the peaks of the Ta and N. This is an indication that the nitride is breaking down during the thermal aging. In fact, the profile of the aged contacts bears a much stronger resemblance to that of those annealed at much higher temperatures. SEM micrographs are shown in Figure 3-3 for the TaN-based contacts. The contacts show a smooth mor phology upon deposition, with an increasingly rough surface as anneal temperature incr eases due to the increased interm ixing of the contacts as well as oxidation of the surface. TiN and ZrN contact s display the same type of profile structure upon annealing and after thermal aging. As mentioned previously, th e Ni/Au/GaN structure is pr eferred due to the greater difference in electronegativities between Au and Ga than between Ni and Ga. By incorporating a layer on top of the Au, the dynamics of this relationship may change The electronegativities of

PAGE 49

49 Ta, Ti, and Zr are 1.5, 1.5, and 1.33, respectively. The values for Au, Ni, and Ga are 2.4, 1.8, and 1.6, respectively. From this, it is apparent that the differences between Au and either Ti, Ta, Zr, or Ga are roughly the same, indicating no preferred bonding to any of them. Various equilibrium phases exist for Au-Ti, Au-Zr, and Au-Ta, but none for Au-N, supporting the conjecture that Au is bonded to the metal. Thus neither configuration is strongly preferred. From the Au-Ni phase diagram, a miscibil ity gap exists with a maximum of 810.3 oC at around 70 at% Au. At higher and lower Au concentrat ions, the temperature above which Au and Ni intermix freely is lower; for exam ple, at 10 at% Au, it is below 500 oC and at 30 at% Au below 700 oC. Since there is no driving force in terms of greater bond strength to cause Au and Ni to switch places when using these diffusion barriers, the Ni and Au should each be present at the surface of the GaN. It is expected that the Ni and Au would separate with slow enough cooling but the layered structure of pure Ni/Au contacts may not form. The AES depth profiles confirm this, displaying significant intermixing at the GaN interface. We also see that the boundary between the GaN and the Ni-Au is not well-defined. This is consistent with the formation of the Au-Ga phases. From the Au-Ga phase diagra m, these phases have melting points of 461.3 oC and 491.3 oC, respectively. These are lower than th e temperatures required to obtain Ohmic behavior, suggesting it is beneficial to have so me Ni present at the GaN interface in order to maintain contact integrity. [67] Figure 3-4 shows specific contac t resistance and sheet resistance under the contacts as a function of thermal aging at 200 oC. This simulates long term operation of these contacts in devices. These contacts display poo r stability upon aging at this te mperature, lasting for nearly 10 days before the I-V characteristics are no l onger completely linear. From the AES depth profiles of the contacts that faile d, there was significant in-diffusi on of metals into the GaN. The

PAGE 50

50 temperature during aging is not sufficient for the formation of the Au-Ga phases and as a result the metal in-diffused to the GaN is in the form of interstitials, which hinder electrical conduction. In addition to this, the presence of the nitride on the surface of the GaN will lead to a breakdown the near surface GaN layer. Failure is attributed to the fact that the nitrides will intermix with the GaN. Since no change in chemical composition is expected to take place, only for Ga to exchange places with a metal such as Ti, there is no contribution to a change in the Gibbs free energy from formation or decomposition of comp ounds. In fact, it is expected that the intermixing may be favorable, as it will serve to increase the entropy of the system. From the equation G = H T S it is apparent that an increase in entropy leads to a lower free energy, and is therefore favorable. This leads to decomposition of the GaN. This is in agreement with the sheet resistance under the contacts increasing in the failed cont acts. It is also consistent with the finding in the AES profiles that the aged contacts display a breakdown in the nitride barrier layer. 3.1.2.2 Summary In conclusion, nitride-based contacts display low specific contact resistance and are stable upon anneals up to 1000 oC, with reasonable stability upon agi ng at elevated temperatures. Due to the nature of the contacts, they display extremely poor stability during aging and are therefore unsuitable for long term device operation. 3.1.3 Tungsten Boride and Chromium Boride-B ased Contacts and Long Term Thermal Aging of Borides 3.1.3.1 Experiment and discussion Figure 3-5 displays the specific contact resistan ce of the contacts as a function of anneal temperature as well as the sheet resistance under the contact. Th e contacts transition to Ohmic behavior at annealing temperatures of > 600oC. This likely correspond s to the creation of Ga

PAGE 51

51 vacancies due to the formation of AuGa and AuGa2 at the GaN surface and thus a localized increase in the hole concentration. This im plies that the current transport mechanism is tunneling. Figure 3-6 supports th is conclusion, as the contact resistance is not dependent on measurement temperature in the measured range 25-140 oC. The sheet resistance is also relatively stable as a function of annealing temperature, suggesting th at there is little intermixing between the boride and the GaN, even dur ing elevated anneal temperatures. Figure 3-7 displays the specific contact resistance of the W2B and CrB2 contacts as a function of long term thermal aging at 350oC. Both W2B and CrB2 display excellent stability for the duration of the aging, 23 days. While the nitrides provided good stability over the same range of anneal temperatures, the contacts fa iled after approximately 9 days at a lower temperature. This is attributed to the fact that the nitrides will inte rmix with the GaN. In the case of the borides, however, changes in the compounds must occur for significant intermixing to take place. This could take several forms, includ ing the reactions XB + GaN => XN + GaB or XB + GaN => XGa + BN. GaB phases are not present in phase diagrams [51]. Full thermodynamic evaluation of these is not possibl e, as most of the values fo r the Gibbs free energy are not available. However, it would be expected th at these reactions could only proceed at very elevated temperatures, as the compounds involve d are each very stable while any Ga-metal phases have relatively low melting te mperatures and thus can be assu med to be less stable [51]. For example, the maximum melting temperatur es of selected Ga-metal phases are 1670oC for GaTi, 1863oC for GaCr, and 1855oC for GaZr. These are much lower than the melting temperatures of either GaN or the borides or nitr ides. Thus, while a reaction of this nature may proceed during very high temperature anneals, during the thermal aging it will only proceed very slowly, if at all.

PAGE 52

52 Figure 3-8 displays the AE S depth profiles for the W2B-based contacts as deposited, annealed at 600 oC, annealed at 700 oC and aged at 350 oC, and annealed at 1000 oC. Upon annealing at 600 oC, the Ni and Au show strong mixing at the GaN interface. However, the boride barrier prevents most in termixing between the Ni/Au and the Ti/Au overlayer. The boride has minimal reaction with the GaN surface and the surface had excellent morphology at 600C. Note that the boride peak remains strongly co rrelated, indicating the W and B atoms remain together. This is the case even at 1000oC anneals and after long term thermal aging at 350 oC, indicating that the bo ride is not chemically ac tive in these contacts. At higher temperatures, the AES profiles were not entirely reliable due to non-uniformity depending on where the profile was taken. However, the IV characteristics we re still Ohmic and the su rface still had reasonable morphology. At temperatures of 1000C and higher, the contact layers became totally intermixed, although the peaks of W and B still co rrespond to one another. Similar results were obtained with CrB2-based contacts. 3.1.3.2 Summary These boride-based contacts display excellent thermal stability both upon high temperature anneals and during long term thermal aging at very high temperature. This is due to the fact that there is little intermixing between the boride and the GaN itself, due to the large change in Gibbs Free Energy that would be necessary for any chem ical reaction to occur. This is supported by the depth profiles obtained from AES, which s how that the boride has not broken down during experiments. 3.1.4 Contact Resistance for Other Boride-based Contacts 3.1.4.1 Titanium boride-based contacts Contacts with a TiB2 diffusion barrier for enhancing the thermal stability of Ni/Au Ohmic contacts on p-GaN were fabricat ed. Figure 3-9 shows the specifi c contact resistivity of the

PAGE 53

53 Ni/Au/TiB2/Ti/Au/p-GaN structure as a function of a nnealing temperature, along with the sheet resistance of the p-GaN under the contacts, extr acted from the TLM measurements. The contacts are rectifying below an anneal temperature of 500C but transition to Ohmic behavior for higher temperatures. The specific contact resistivity improves steadily with temperature above 700C and reaches a minimum of ~2 x10-4 .cm-2 after annealing at 800 C fo r 60 secs. Note that the contact resistance stays at a similar value even for temperatures up to 950C, even though the higher annealing temperatures produced increases in the sheet resistance of the GaN and less reproducible contact properties due to roughening of the contact morphology. The mechanism for the increase in sheet resistance may incl ude loss of N (producing nitrogen vacancy-related donors) or a reduction in the solu ble Mg concentration. This sp ecific contact resistance is comparable to that achieved on the same sample s with Ni/Au metallization annealed at 500C. The incorporation of the TiB2 diffusion barrier clearly provides a much wider range of thermal stability compared to the standard Ni/Au contacts. Figure 3-10 shows SEM pictures of the TLM contact pads as a function of annealing temperature. After 800C annealing the morphology is degraded, but is still much smoother than conventional Ni/Au contacts under th ese conditions. The darker appearance of the contacts after high temperature anneals is mainly a result of the outdiffusion of Ti, which rapidly oxidizes. Figure 3-11 shows the AES surface scans from the as-deposited sample (top) and those annealed at 800C (center) and 900C (bottom). Ti is evident on the surface after annealing at 800C and the concentration increases with a nnealing temperature. The increased oxygen concentration evident from the summary of th e data in Table 3-1 most likely comes from oxidation of this out-diffused Ti. The carbon signal in all cases comes from adventitious carbon on the surface.

PAGE 54

54 The AES depth profiles from the as-deposit ed and 800 and 900C annealed samples are shown in Figure 3-12. The as-deposited sample shows sharp interfaces be tween the metals and between the Ni and the GaN. After annealing at 800C, the Ni shows significant movement through all of the overlying layers and by 950C, mo st of the contact meta llurgy is intermixed and the Ti is essentially all removed to the surface. There is an accompanying decrease in abruptness of the TiB2 interfaces with the metals on either side. The TiB2 appears to be a barrier for Ti diffusion, so this probably excludes formati on of TiN phases at the interface of the GaN to account for the improved contact resistance. 3.1.4.2 Zirconium Boride-based contacts ZrB2-based contacts were also fabricated. Figure 3-13 shows th e specific contact resistance of the ZrB2/Ti/Au and Ni/Au/ZrB2/ Ti/Au/p-GaN structures as a function of annealing temperature, along with the sheet resistance of the p-GaN under th e contacts, extracted from the TLM measurements. The contacts are rectifying below an ann eal temperature of 750C but transition to Ohmic behavior for hi gher temperatures. The Ni/Au/ZrB2/Ti/Au contact resistance improves with higher anneal temperature, at the expense of poorer morphology. The ZrB2/Ti/Au contacts do not show low resist ance values until 800C and the value degrades at higher temperatures. The associated sheet resistance gene rally decreases with te mperature. The specific contact resistance values obtai ned with both types of ZrB2-based schemes is comparable to that achieved on the same samples with Ni/Au metallization annealed at 500C but the former have much higher thermal stability, as expected from the high melting temperature of the ZrB2. Figure 3-14 shows the AES surface scans and depth profiles from the as-deposited Ni/Au/ZrB2/ Ti/Au/p-GaN sample (left) and from the sa mple annealed at 750C (right). Ti, Ni and Zr are evident on the surface after annealing at 750C and the concentration increased with annealing temperature. The as-deposited sample shows sharp interfaces be tween the metals and

PAGE 55

55 between the Ni and the GaN. Table 3-2 summar izes the near-surface composition AES data. The increased oxygen concentration evid ent from this data most likely comes from oxidation of the out-diffused Ti. The carbon signal in all cases comes fr om adventitious carbon on the surface. After annealing at 750C, the Ni shows significa nt movement through all of the overlying layers and even the Zr diffuses out of the boride later, le aving the boron in its orig inal location. Both Zr and Au diffuse to the interface with GaN. The increased contact intermixing at higher annealing temperatures did roughen the contact morphology as shown in the SEM pictures of the TLM contact pads as a function of annealing temperature in Figure 3-15. After 800C annealing the morphology is degraded, but is still much sm oother than conventiona l Ni/Au contacts under these conditions. The darker appearance of the c ontacts after high temperature anneals is mainly a result of the outdiffusion of Ti, which then oxidizes. The same basic trends were observed with the ZrB2/Ti/Au structures. Figure 3-16 shows the AES surface scans and surface scans of ZrB2/Ti/Au Ohmic contacts on p-GaN as a function of anneal temperature. The as-deposited sample shows abrupt interfaces but after annealing at 900C, the Zr shows a very broad distribution (more so than the B). The c ontacts after annealing showed the presence of reacted islands, as s hown in the elemental maps of Figure 3-17. The islands contain both Au and Ga a nd show that the GaN epi layer ha s begun to dissociate at this temperature, at least under the contact metallurgy. Note that the areas with Au and Ga overlap, indicating that they are indeed forming th e Au-Ga phases reported in the literature. 3.1.4.3 Gallium Nitride//Tungsten Boride-based contacts Figure 3-17 shows the specific co ntact resistivity of the W2B/Ti/Au/p-GaN structure as a function of annealing temperat ure, along with the sheet resi stance of the p-GaN under the contacts, extracted from the TLM measurements. The contacts are rectifying below an anneal temperature of 500C but transition to Ohmic behavior at this te mperature. The specific contact

PAGE 56

56 resistivity improves steadily with temperature and a minimum of 1.7 x10-3 .cm-2 was obtained after annealing at 800 C for 60 secs. Higher anne aling temperatures produ ced sharp increases in the sheet resistance of the GaN and irreproduci ble contact properties. The mechanism for the increase in sheet resistance may include loss of N (producing nitrogen vacancy-related donors) or a reduction in the soluble Mg concentration. However the contact morphology as determined by SEM was similar over the entire annealing range used here. This specific contact resistance is comparable to that achieved on the same sample s with Ni/Au metallization annealed at 500C. Figure 3-18 shows the AES surface scans from the as-deposited sample and those annealed at 500, 700 and 800C.Ti is evident on the surface after anne aling at 500C and the concentration increases with annealing temperat ure. The increased oxygen concentration evident from the summary of the data in Table 3-3 most likely comes from oxidation of this out-diffused Ti. The carbon signal in all cases comes fr om adventitious carbon on the surface. The AES depth profiles from the as-deposited and annealed samples are shown in Figure 3-19. The as-deposited sample shows sharp in terfaces between the metals and between the W2B and the GaN. After annealing at 500C, the Ti shows significant movement through the Au layer and by 800C is essentially all removed to the surface. There is an accompanying decrease in abruptness of the W2B/GaN interface, which suggests the onset of reactions that are responsible for the improved contact resistance. Since the W2B appears to be a barrier for Ti diffusion, this probably excludes formation of TiN phases. Cole et al.[29] reported the formation of the W2N and W-N interfacial phases after annealing sputte red W on GaN [68]. The former appeared after anneals at 600C and the latter after 1000C. That work demons trated the importance of these interfacial phases in the resulting contact properties. The specific contact resistance obtained for the W2B-based contacts is lower than reported previously for pure W on p-GaN [69], likely due

PAGE 57

57 to the lack of these phases and the resulting compensation of holes by nitrogen vacancies. The absence of Ti diffusion through the W2B also suggests that an optimum metal scheme might involve a thin sandwich of the bor ide layers around the Ti, with the Au overlayer to reduce the sheet resistance. The reflectivity of pure W is around 60% at 460 nm, while that of Au is around 95% [70], so clearly the thickne ss of the boride layers should be small relative to the Au thickness. The long-term stability and reliability of the boridebased contacts must still be established on LEDs and laser diodes. In the former case, the reflectance of the multi-layered contacts at the emission wavelength of the LEDs must also be established. 3.1.5 Iridium-Based Contacts 3.1.5.1 Experiment and discussion Figures 3-21 and 3-22 show the IV plots for both types of contacts at different anneal temperatures and 20 m spacing. Only the Ni/Au/Ir/Au contacts annealed at 500 oC were found to have Ohmic characteristics. It was observed that these contacts displa yed a specific contact resistance of ~2.3 x 10-4 -cm2, comparable to results previously achieved on similar wafers for other schemes. Table 3-4 shows a summary of previous results achieved using a similar approach with other thermally stable materials as well as the anneal temperatures above which the contacts failed either due to unreas onable morphology or breakdown of Ohmic characteristics. From the IV curves for the Ni/Au/Ir/Au contacts, we see a decrease in the amount of current upon annealing at 300 oC, which likely corresponds to indiffusion into the GaN of Ni but at a temperature too low for the Ni hydride to form and increase the doping. At 500 oC the contacts transition to Ohmi c behavior due to increased near surface doping. At higher temperatures the contacts revert to rectifying behavior. For the Ni/Ir/Au contacts, the as deposited display the largest amount of current. At 300 oC the amount of current decreases

PAGE 58

58 again, only to increase at 500 oC due to removal of H by Ni. At 700 oC it begins to decrease again. Figures 3-23 and 3-24 show the Auger electron spectroscopy (AES) depth profiles for the Ni/Au/Ir/Au and Ni/Ir/Au contacts, respectively, at several different anneal temperatures. At 300oC, the contacts maintain well defined layers, with a distinct lack of intermixing. At this temperature, the contacts do not display Ohmi c characteristics because the Ni has not depassivated the H-Mg complexes. At 500 oC, the Ni/Au/Ir/Au contacts display significant interaction between the GaN and the Ni/Au underlayer. This likely corresponds to the formation of the Au-Ga phases, which serve to increase the doping. The Ni/I r/Au contacts show that the Ni has diffused towards the Au, but the Au has not diffused in towards the GaN. With little or no Au at the surface of the GaN, the hole concentrati on is not expected to increase as much as with the Ni/Au-based contact. This is the reason why Oh mic behavior is not observed in the Ni/Ir/Au contacts. In addition, we can observe that the Au seems to be essentially insoluble in the Ir at the moderate temperature anneals. At 700 oC, we can see that the layer structure has broken down and the metals are completely intermixed. Figure 3-25 displays scanning electron micr oscopy (SEM) images of the Ni/Au/Ir/Au contacts as deposited and annealed at several temperatures. The surface at 500 oC displays good morphology while that of the 700 oC anneal shows signs of the br eakdown of the layer structure. The failure of any contacts to display Ohmic behavior at high anneal temperatures is likely due to the nature of the Ir. Ir is expected to be stable on the GaN surface because it is not expected to form extensive secondary phases with the Ga or the N. While IrN has been recently been observed, it has only been synthesized at very high pressures. For the high temperature anneals, the Ir diffuses into the GaN. Howeve r, because no nitride can be formed under these

PAGE 59

59 conditions, the Ir must occupy interstitial sites and cannot be incorporat ed into the crystal structure of the GaN itself. Because the Ir atoms are very large a nd are likely sitting in interstitial sites, they can cause significant lattice distortions and it is expected that the crystallinity of the GaN would be significantly decreased. These distortions, as well as the Ir atoms themselves, would act as scattering sites for ch arge carriers. It is in this manner that the contacts fail when annealed at high temperatures. Thus, even when Au arrives at the surface in the Ni/Ir/Au contacts at 700 oC, the Ir has already caused enough disruption to the GaN that it does not matter. This same lack of chemical interaction of the Ir on the GaN may allow for improved long term operation of LEDs using the Ni/Au/Ir/Au contacts. Because the Ir is not expected to interact chemically with the GaN, at the lower temperatures which were examined during aging, and which would be present during long term ope ration, the GaN itself is not expected to decompose over a long period of time. 3.1.5.2 Summary Ir diffusion barriers for contacts to p-GaN have been fabricated. The importance of Au being in contact with the GaN for the format ion of low resistance Ohmic contacts has been confirmed, as the Ni/Ir/Au contacts do not display Ohmic characteristics. 3.2 Conclusions Ohmic contacts have been fabricated to p-GaN using three classes of materials. It has been determined that boride based contacts are superior to either Ir or nitride based ones due to both their ability to withstand a wide range of annealing temperatures and to withstand long term thermal aging. From these results, it can be not ed that it is important to choose materials for contacts which will not display a great deal of either chemical reaction or indiffusion.

PAGE 60

60 Table 3-1: Concentration of elements de tected on the as-received surface (in atom%) Filename Sample ID C(1) O(1) Ti(1) Au(3) Sensitivity factors [0.076] [0.212] [0.188] [0.049] 051829101 As deposited 36 1 nd 64 051829201 800C annealed 38 31 19 12 051829301 900C anneal 27 33 28 12 051829401 950C Anneal A 30 26 33 nd 051829401 950C Anneal B 28 26 35 nd

PAGE 61

61 Table 3-2: Concentration of elements dete cted on the as-received surfaces (in atom%) Sample ID C(1) O(1) Ti(1) Ni(1) Zr(2) Au(3) Sensitivity factors 0.076 0.212 0.188 0.227 0.043 0.049 As deposited Ni/Au/ZrB2/Ti/Au 44 4 nd nd nd 52 annealed 750C 45 19 2 1 14 19 annealed 800C 42 22 9 1 16 10 As deposited ZrB2/Ti/Au 37 4 nd nd nd 59 annealed 800C 43 23 nd nd 12 22 annealed 1000C 48 19 nd nd 14 19

PAGE 62

62 Table 3-3: Concentration of elements dete cted on the as-received surfaces (in atom%) Sample C(1) O(1) Ti(2) Au(3) [0.076] [0.212] [0.296] [0.049] As deposited 41 nd nd 59 500C annealed 43 24 5 28 700C annealed 44 34 7 15 800C annealed 38 34 15 13

PAGE 63

63 Table 3-4: Summary of specific contact resistances Contact Scheme sc,minium ( -cm2) Anneal temperature (oC) Failure temperature (oC) Ni/Au/W2B/Ti/Au 3.5 x 10-5 1000 >1000 W2B/Ti/Au 1.69 x 10-3 800 >900 Ni/Au/CrB2/Ti/Au 7 x 10-5 1000 >1000 Ni/Au/TiB2/Ti/Au 1.93 x 10-4 850 >900 Ni/Au/ZrB2/Ti/Au 1 x 10-4 900 >900 ZrB2/Ti/Au 1.8 x 10-3 800 >900 Ni/Au/TaN/Ti/Au 2.5 x 10-4 800 >1000 Ni/Au/TiN/Ti/Au 2.45 x 10-4 600 >1000 Ni/Au/ZrN/Ti/Au 2 x 10-4 600 >1000 Ni/Au/Ir/Au 2.3 x 10-4 500 >500 TaN/Ti/Au 3.7 x 10-4 800 >1000 ZrN/Ti/Au 3.1 x 10-4 600 >700

PAGE 64

64 5006007008009001000 1x10-51x10-41x10-3 0 20 40 60 80 100 TaN TiN ZrN Specific Contact Resistance(-cm2)Anneal Temperature (C) Sheet Resistance (/) TaN TiN ZrN Figure 3-1: Specific contact resistance and sheet re sistance under the contact of Ni/Au/ X/ Ti/Au contacts as a function of anneal temperature.

PAGE 65

65 Figure 3-2: Depth profiles of Ni/Au/TaN/Ti/Au contac ts (a) as deposited (b) annealed at 600 o C (c) annealed at 700oC and aged at 200 oC until the contacts became non-Ohmic and (d) annealed at 1000 oC. 05001000150020002500300035 0 0 10 20 30 40 50 60 70 80 90 100 (a) Au Ga Ta Ti Ni Au NAtomic Concentration (%)Depth ()N 010002000300040 0 0 10 20 30 40 50 60 70 80 90 100 (c)O Au Ga Ta Ti NiAtomic Concentration (%)Depth ()N 010002000300040 0 0 10 20 30 40 50 60 70 80 90 100 (d)O Au Ga Ta Ti NiAtomic Concentration (%)Depth ()N010002000300040 0 0 10 20 30 40 50 60 70 80 90 100 O Au Ga Ta Ti Ni Au NAtomic Concentration (%)Depth ()N(b)

PAGE 66

66 Figure 3-3: Scanning electron micr oscopy images of Ni/Au/TaN/Ti/Au contacts (a) as deposited (b) annealed at 600 o C (c) annealed at 700oC and aged at 200 oC until the contacts became non-Ohmic and (d) annealed at 1000 oC.

PAGE 67

67 0246810 1x10-51x10-41x10-3 0 20 40 60 80 100 120 140 160 TaN TiN ZrN Specific Contact Resistance (-cm2)Day Sheet Resistance ( TaN TiN ZrN Figure 3-4: Specific cont act resistance of Ni/Au/X/Ti/Au cont acts as a function of aging at 200 oC.

PAGE 68

68 Figure 3-5: Specific contact resistance and sheet resistance under the contact versus anneal temperature.6007008009001000 1x10-61x10-51x10-41x10-31x10-2 0 5 10 15 20 25 30 Specific Contact Resistance (-cm2)Anneal Temperature (oC) CrB2 W2B Sheet Resistance ()

PAGE 69

69 Figure 3-6: Specific contact resistance versus measurement temperature. 0255075100125150 1x10-61x10-51x10-41x10-31x10-2 Specific Contact Resistance (-cm2)Measurement Temperature (oC) 700oC 1000oC

PAGE 70

70 Figure 3-7: Specific contact resistance and sheet resistance under the contact as a function of long term thermal aging at 350oC. 0510152025 1x10-61x10-51x10-41x10-3 0 10 20 30 40 50 Specific Contact Resistance (/cm 2 )Days CrB2 W2B Sheet Resistance ()

PAGE 71

71 Figure 3-8: Depth profiles of W2B-based contacts (a) as depo sited (b) annealed at 600 oC (c) annealed at 700 oC and aged at 350oC and (d) annealed at 1000 oC 0 20004000 0 20 40 60 80 100 Ti Ni (b) N O W B Au Atomic Concentration (%)Depth () 01000200030004000 0 20 40 60 80 100 Au Au W Ga Ni Ti N B (a) Atomic Concentration (%)Depth () 0500100015002000 0 20 40 60 80 100 (b) N Ga W B Au Atomic Concentration (%)Depth () 0 10002000 0 20 40 60 80 100 Ni Ti (b) N Ga W B Au Atomic Concentration (%)Depth ()

PAGE 72

72 700750800850900950 1E-5 1E-4 1E-3 Contact Resistance Contact Resistance (ohm-cm2)Temperature0 20 40 60 80 100 Sheet Resistance Sheet Resistance Figure 3-9: Specific contact resistivity of Ni/Au/TiB2/Ti/Au Ohmic contacts and p-GaN sheet resistance under the contact as a function of annealing temperature.

PAGE 73

73 Figure 3-10: Secondary electron images of Ni/Au/TiB2/Ti/Au contact pads on p-GaN asdeposited (top) or after annealing at either 800(ce nter) or 900C (bottom).

PAGE 74

74 Figure 3-11: Surface scans of Ni/Au/TiB2/Ti/Au Ohmic contacts on p-GaN as a function of anneal temperature. The as-deposited sample is at top, that annealed at 800C at center and that at 900C at bottom 50 0 1000 1500 2000 -6000 -5000 -4000 -3000 -2000 -1000 0 1000 2000 3000 4000 Kinetic Energy (eV ) c/ s Ci O S Au Au Au Au Au Au Au Atomic % Au 3 60 0 C1 36.2 S1 2.4 O1 1.4 500 1000 1500 2000 -12000 -10000 -8000 -6000 -4000 -2000 0 2000 4000 6000 8000 Kinetic Energy (eV ) c/s C O Ti Ti Au Au Au S Bi B Atomic % C1 38.1 O1 30. 6 Ti1 18.7 Au3 7.7 B2 4.5 S1 0.4 500 100 0 1500 2000 1.5 -1 0.5 0 0.5 1 x 10 4 K inetic Energy (eV) c/ s C O B B Ti Au Au Ti Au Atomic % O1 33. 5 Ti1 27.8 C1 27.2 Au3 5.8 B 2 5 7

PAGE 75

75 Figure 3-12: Depth profiles of Ni/Au/TiB2/Ti/A u Ohmic contacts on p-GaN as a function of anneal temperature. The as-deposited sample is at top, that annealed at 800 pC at center, and that at 900 oC at bottom. 0 500 1000 1500 2000 0 10 20 30 40 50 60 70 80 90 100 Sputter Depth ( ) A tomic Concentration (%) Ti1.ls1 O1.ls2 N1.ls3 Ti2.ls 5 Ga1.ls 7 A u3.ls 8 B2.ls 9 0 500 1000 1500 0 10 20 30 40 50 60 70 80 90 100 Sputter Depth ( ) A tomic Concentration (%) Ti2.ls1 N1.ls2 B2.ls3 A u3.ls4 Ni1.ls 5 Ga1.ls6 0 500 1000 1500 2000 2500 3000 0 10 20 30 40 50 60 70 80 90 100 Sputter Depth ( ) A tomic Concentration (% ) Ti1.ls1 O1.ls2 N1.ls3 Ti2.ls 5 Ni1.ls6 Ga1.ls7 A u3.ls8 B2.ls 9 A u3.ls8

PAGE 76

76 750 800 850 900 10-410-310-2 Contact Resistance Ni/Au/ZrB2/Ti/Au ZrB2/Ti/Au Contact Resistance (-cm)Temperature (oC)3 4 5 6 7 8 9 10 11 12 Sheet Resistance Ni/Au/ZrB2/Ti/Au ZrB2/Ti/Au Sheet Resistance () Figure 3-13: Specific contact resistance of Ni/Au/ZrB2/Ti/Au and ZrB2/Ti/Au Ohmic contacts and p-GaN sheet resistance under the contact as a function of annealing temperature.

PAGE 77

77 Figure 3-14: Surface scans and depth profiles of Ni/Au/TiB2/Ti/Au Ohmic contacts on p-GaN as a function of anneal temperature. 500 1000 1500 2000 -8000 -6000 -4000 2000 0 2000 4000 Kinetic Energy (eV) c/s C O S Au Au Au Au Au Au Au A tomic % A u3 52.1 C1 44.2 S1 2.2 O1 1.5 Sample #1 As deposited 500 1000 1500 2000 -8000 -6000 -4000 -2000 0 2000 4000 6000 8000 c/s C O Al Al Ni Au Au Ti Au Au Z r Z r A tomic % C 1 44.9 O 1 18.9 A u3 15.7 Z r2 14.0 A l2 2.9 T i1 2.2 N i1 1.4 Sample #2 Annealed at 750C Kinetic Energy (eV) 0 500 1000 1500 2000 2500 3000 0 10 20 30 40 50 60 70 80 90 100 Sputter Depth ( ) Atomic Concentration (%) C O Au Z r B Ni Ti Ga Au 0 1000 2000 3000 4000 5000 0 10 20 30 40 50 60 70 80 90 100 Atomic Concentration (%) C Ti O Z r Ni Ga A u B Sputter Depth ( )

PAGE 78

78 Figure 3-15: Scanning electron mi croscopy images of Ni/Au/TiB2/Ti/Au contact pads on p-GaN as-deposited (top) or after annealing at either 750 (center) or 800C (bottom).

PAGE 79

79 Figure 3-16: Surface scans and depth profiles of ZrB2/Ti/Au Ohmic contacts on p-GaN as a function of anneal temperature. 500 1000 1500 2000 -5000 0 c/s C O S Au Au Au Au Au Au Au A tomic % A u3 59.1 C1 36.4 S1 3.1 O1 1.3 Sample #4 As deposited 5000 -10000 Kinetic Energy (eV) 500 1000 1500 2000 -6000 0 2000 4000 c/s C N O Mg Al Al Z r Z r Au Au Au A tomic % C1 52.9 O1 16.3 Zr2 12.0 A u3 8.9 M g2 4.2 A l2 4.0 N 1 1.7 Sample #6 Annealed at 900C -8000 -10000 6000 Kinetic Energy (eV) -2000 -4000 0 500 1000 1500 0 10 20 30 40 50 60 70 80 90 100 C Ti O Z r Ga Au B Atomic Concentration (%)Sputter Depth ( ) 0 500 1000 1500 2000 0 10 20 30 40 50 60 70 80 90 100 Atomic Concentration (%) Au B Ga Z r O Ti C Sputter Depth ( )

PAGE 80

80 Figure 3-17: Elemental maps obtained from scanning AES of ZrB2/Ti/Au Ohmic contacts pads on p-GaN. 0.0 130400 0.0 25900 Au3 Ga1 0.0 47.0 Ga1+Au3 151.0 191.0 SEM

PAGE 81

81 500550600650700750800 1.6 2.0 2.4 2.8 3.2 3.6 Contact Resistance Contact Resistance (10-3 -cm2)Temperature (oC)60 80 100 120 140 Sheet Resistance Sheet Resistance (/ Figure 3-18: Specific contact resistivity of W2B/Ti/Au Ohmic contacts and measured p-GaN sheet resistance under the contact as a function of annealing temperature.

PAGE 82

82 Figure 3-19: AES surface scans of W2B/Ti/Au Ohmic contacts on p-GaN as a function of anneal temperature. 500 1000 1500 2000 -6000 -5000 -4000 -3000 -2000 -1000 0 1000 2000 3000 4000 051535101 .spe Kinetic Energy (eV)c/s C N Au Au Au Au Au Au AuAtomic % Au3 58.7 C1 37.5 N1 3.9Sample #1 As deposited 500 1000 1500 2000 -10000 -8000 -6000 -4000 -2000 0 2000 4000 6000 8000 051535201.spe Kinetic Energy (eV) c/s C O Ti AuAtomic % C1 43.3 Au3 27.9 O1 24.0 Ti2 4.8Sample #2 500C annealed Surface survey 500 1000 1500 2000 -1.5 -1 -0.5 0 0.5 1 x 10 4 051535301.spe Kinetic Energy (eV)c/s C O Ti Au Au Au S TiAtomic % C1 44.0 O1 33.7 Au3 14.7 Ti2 7.1 S1 0.5Sample #3 700C annealed Surface survey 500 1000 1500 2000 -1.5 -1 -0.5 0 0.5 1 x 104 051535401.spe Kinetic Energy (eV)c/s C O S Ti Ti Au AuSanmple #4 800C annealed Surface surveyAtomic % C1 38.1 O1 33.6 Ti2 15.1 Au3 12.7 S1 0.5

PAGE 83

83 Figure 3-20: Depth profiles of W2B/Ti/Au Ohmic contacts on p-GaN as a function of anneal temperature. 0 500 1000 1500 2000 0 10 20 30 40 50 60 70 80 90 100 051535103.pr o Sputter Depth ( )Atomic Concentration (%) C1.ls1 O1.ls2 N1.ls3 Ti2.ls5 W2.ls6 B1.ls7 Au3.ls8 Ga1.ls9 0 500 1000 1500 2000 2500 0 10 20 30 40 50 60 70 80 90 100 051535203 .pro Sputter Depth ( )Atomic Concentration (%) C1.ls1 N1.ls2 O1.ls4 Ti2.ls5 W2.ls6 B1.ls7 Au3.ls8 Ga1.ls9 0 500 1000 1500 2000 2500 0 10 20 30 40 50 60 70 80 90 100 051535303.pro Sputter Depth ( )Atomic Concentration (%) C1.ls1 N1.ls2 O1.ls4 Ti2.ls5 W2.ls6 Au3.ls7 Ga1.ls8 B1.ls9 0 500 1000 1500 2000 2500 0 10 20 30 40 50 60 70 80 90 100 051535403.pro Sputter Depth ( )Atomic Concentration (%) C1.ls1 N1.ls2 O1.ls4 Ti2.ls5 W2.ls6 Au3.ls7 Ga1.ls8 B1.ls9

PAGE 84

84 -3-2-10123 1E-10 1E-9 1E-8 1E-7 1E-6 1E-5 1E-4 1E-3 0.01 0.1 As dep 300oC 500oC 700oC Current (A)Voltage (V) Figure 3-21: Current-voltage cu rves for Ni/Au/Ir/Au contacts.

PAGE 85

85 -3-2-10123 1E-9 1E-8 1E-7 1E-6 1E-5 1E-4 1E-3 0.01 As dep 300oC 500oC 700oC Current (A)Voltage (V) Figure 3-22: Current-voltage cu rves for Ni/Ir/Au contacts

PAGE 86

86 0100020003000 0 20 40 60 80 100 Au N Ga Ni Ir Au Atomic Concentraton (%)Depth ()(a) 010002000300040005000 0 20 40 60 80 100 Au N Ga Ni Ir Au Atomic Concentraton (%)Depth ()(b) 01000200030004000 0 20 40 60 80 100 O N Ga Ni Ir Au Atomic Concentraton (%)Depth ()(c) Figure 3-23: Depth profiles for Ni/Au/ Ir/Au contacts (a) annealed at 300 oC (b) annealed at 500 oC and (c) annealed at 700 oC.

PAGE 87

87 0 10002000 0 20 40 60 80 100 O N Ga Ni Ir Au Atomic Concentraton (%)Depth ()(a)0 1000 2000 0 20 40 60 80 100 N Ga Ni Ir Au Atomic Concentraton (%)Depth ()(b)0100020003000 0 20 40 60 80 100 N Ga Ni Ir Au Atomic Concentraton (%)Depth ()(c) Figure 3-24: Depth profiles for Ni/I r/Au contacts (a) annealed at 300 oC (b) annealed at 500 oC and (c) annealed at 700 oC.

PAGE 88

88 Figure 3-25: Scanning electron microsc opy images of Ni/Au/Ir/Au contacts

PAGE 89

89 CHAPTER 4 OHMIC CONTACTS TO N-GAN In this chapter, we report results on Ohm ic contacts based on these materials to n-GaN. Specific contact resistance has been studied as a function of annealing temperature, annealing time, and thermal aging for TaN, TiN, and ZrN diff usion barriers. All nitrides show comparable specific contact resistance to conventional Ti /Al/Pt/Au Ohmic contacts and display similar stability when aged at 350C in air. 4.1 Experiment The samples consisted of 3 m thick Si-doped GaN grown by Metal Organic Chemical Vapor Deposition on c-plane Al2O3 substrates. From Hall measurements, the electron concentration was ~3 x 1017 cm-3. Mesas 0.6 m deep were formed by Cl2/Ar inductively coupled plasma etching to provide for electrical isolation of the contact pads and to prevent current spreading. Deposition of a Ti (200 )/Al (800 )/X (500 )/Ti (200 )/Au (800 ) metallization was performed by sputtering, where X is TaN, TiN, or ZrN. Sputtering was performed using a Ar plasma-assisted system at a pressure of 5 mTorr and rf (13.56 MHz) powers of 150-350W. For comparison, a Ti (200 )/Al (800 )/Pt (400 )/Au (800 ) was deposited by e-beam evaporation. The contacts were patterned by lift off and annealed at 4001000C for 30-240s in a flowing N2 ambient in a rapid therma l annealing (RTA) furnace. Specific contact resistances were obtained using the linear transmission line (L-TLM) method. 100 m x 100 m pads with spacings of 80, 40, 20, 10, and 5 as well as 76, 36, 16, 6, and 1 m were used. The total resistance is given by RT = 2 RC + Rs ( L / W ) (4.1)

PAGE 90

90 where RC is the contact resistance, RS the sheet resistance, L the pad spacing, and W the pad width. A plot of RT vs pad spacing provides both RC, from the y-intercept, and RS, from the slope. The specific contact resistance is then calculated according to the equation c= RC 2W2/ Rs(4.2) All current-voltage plots were measured us ing an Agilent 4145B Parameter Analyzer. Auger electron spectrosc opy (AES) depth profiling of the contacts was peformed with a Physical Electronics 660 scanning Auger micr oprobe. The electron beam conditions were 10 keV, 1 A beam current at 30 from the sample nor mal. For depth profiling, the ion beam conditions were 3 keV Ar+, 2.0 A, (3 mm)2 raster. Before AES pr ofiling, scanning electron microscopy images (SEMs) were acquired from the samples. The SEMs were obtained at magnifications of 125x and 1000x and were used to locate and document analysis area locations and the surface morphology. Quantification of elements was achieved using the elemental sensitivity factors. 4.2 Results and Discussion Figure 4-1 shows the contact resistances as a fu nction of anneal temperature for the nitridebased Ohmic contacts as well as the conventio nal Ohmic contact. As deposited, the contacts display slight curvature in the current-voltage curves. Upon annealing at 600-1000C, the IV characteristics transform to Ohmic behavior. Th e specific contact resistances are stable up to 1000C anneals, except for the TiN, which show s increased specific contact resistivity at 1000C. The minimum specific contact resistance achieved was ~6 x 10-5 .cm2 for the TiNbased contacts at an anneal of 800C. This is likely due to diffusion of additional Ti to the surface to form the TiN phase. The ZrN-based c onract appears to be the most stable with increased annealing temperature, maintain ing a fairly constant value of ~1 x 10-4 for the range

PAGE 91

91 600-900C. These specific contact resi stance values are expected for the moderately low doped samples used in this study. This is because the current transport mechanism for Ti/Al-based Ohmic contacts to n-GaN is tunneling and is governed by the relation (4.3) Obviously, this relation shows a strong de pendence on the doping of the n-GaN. For doping levels an order of magnitude higher, 3 x 1018 cm-3, we would expect a decrease in the contact resistance of approximately an order of magnitude. Figure 4-2 shows the SEM images of the contacts as a function of anneal temperature. The as-deposited samples for TiN and ZrN show a smooth morphology while bubbles are present on the TaN as-deposited sample, possibly due to trapped sputter gases. At 600oC and 800oC anneals the contacts maintain a reasonable morphology. Furthermore, there appears to be little degradation of the surfac e even after aging at 350oC. All contacts show good edge acuity, a concern for devices with sm all contact separations. Figures 4-3,4, and 5 show the AES depth prof iles of the nitride-based contacts. As deposited, all contacts contain a significant amount of oxygen incor poration in the Ti and nitride layers. This is due to the relatively high base pressure of the sputte ring system of ~2 x 10-6 Torr. Note that despite this, the contact resistances are not adversely affected when compared to the Ti/Al/Pt/Au contacts, which were deposited in an electron beam evaporati on operating at a base pressure of ~3 x 10-7 Torr. As deposited, the metallic in terfaces are sharp and there is little intermixing between the layers. Upon annealing at 600C, Al has outdiffused to the surface and some Au has indiffused to the interface. At higher temperatures, the amount of interdiffusion increases. The amount of interdiffusion is much less than that seen with a standard metallic

PAGE 92

92 diffusion barrier, such as Ni. Fi gure 4-6 shows a profile for a Ti /Al/Ni/Au contact annealed at 500oC. From the profiles, the nitrides appear to be breaking down somewh at, with TiN the least affected, as the location of the Zr and Ta peaks do not correspond precisely to those of the N. Little change in the contact structure is seen with long term aging even after greater than 20 days. Because there is only minimal intermixing of the layers, little of the viscous AlAu4 phase has formed compared to conventional contact schemes. Figure 4-7 shows the dependence upon anneal time at 800C. Upon annealing for up to 180 seconds, an increase in contact resistance is s een in each of the contacts. The standard Ti/Al/Pt/Au contact exhibit the largest increas e in contact resistance, while the contacts containing TaN and TiN display more moderate in creases. The ZrN-based contact displays only a slight increase in contact resistance even upon annealing for 180 seconds. Figure 4-8 displays the specific co ntact resistance as a function of long term thermal aging. Samples placed on a hot plate at 350C display all display a gra dual increase in the specific contact resistance throughout the period. Table 4-1 shows the pe rcentage increase in specific contact resistance for each scheme. Compared to the Ti/Al/Pt/Au contacts, the ZrN and TaN contacts display better stability, with the spec ific contact resistance in creasing only about 100%. The TiN contacts displayed a larger increase. 4.3 Conclusions Ohmic contacts to n-GaN were fa bricated using a Ti/Al/X/Ti/A u scheme, where X is either TaN, TiN, or ZrN. These contacts display similar specific contact resistances to the conventional Ti/Al/Pt/Au contacts deposited on the same wafer, ~10-4 ohm-cm2 over a wide range of annealing temperatures. In a ddition, the TaN and ZrN contacts display only about half the increase in specific contact resistance that is seen in the Ti/Al/Pt/Au contacts. The AES depth profiles confirm that little intermixing has occu rred during the aging process. Because of this,

PAGE 93

93 these contacts show promise for use in high temperature applications. While the contact resistance does not seem to be better than that of conventional contacts af ter aging, there is far less mixing between layers. This is a primary concern, due to the AlAu4 phase, for devices with short spacings for high temperatur e applications. Further evaluation in devices such as HEMTs is needed in order to determine to potential of th ese contacts. In addition, the effect of oxygen incorporation in the nitride should be examined.

PAGE 94

94 Table 4-1: Percent change in specific contact resistance during thermal aging Contact Percent Change Ti/Al/Pt/Au 236% Ti/Al/TaN/Ti/Au 107% Ti/Al/TiN/Ti/Au 330% Ti/Al/ZrN/Ti/Au 091%

PAGE 95

95 Figure 4-1: Specific contact resistance as a function of anneal temperature 6008001000 1x10-51x10-41x10-31x10-2 Ti/Al/Pt/Au Ti/Al/ZrN/Ti/Au Ti/Al/TiN/Ti/Au Ti/Al/TaN/Ti/Au Specific Contact Resistance (-cm 2 )Anneal Temperature (C)

PAGE 96

96 TaN TiN ZrN TaN-based as deposited TiN-based as deposited ZrN-based as deposited TaN-based 600oC anneal TiN-based 600oC anneal ZrN-based 600oC anneal TaN-based 800oC anneal TiN-based 800oC anneal ZrN-based 800oC TaN-based aged TiN-based aged ZrN-based aged Figure 4-2: Scanning elect ron microscopy images of annealed contacts

PAGE 97

97 Figure 4-3: Depth profiles of Ti/Al/TaN/Ti/Au contacts (a) as deposited (b) annealed at 600oC. (c) annealed at 800oC and (d) annealed at 800oC and aged at 350oC. 0100020003000400050006000 0 20 40 60 80 100 (d) Ti Au N Ti O Au Ga O N TaAtomic Concentratioon (%)Depth ()Al0100020003000400050006000 0 20 40 60 80 100 (a) N Ti C O Au Ga O Ti N TaAtomic Concentratioon (%)Depth ()Al0100020003000400050006000 0 20 40 60 80 100 N Ti O Au Ga O N TaAtomic Concentratioon (%)Depth ()Al (c)01000200030004000500060007000 0 20 40 60 80 100 (b) Ta Ga O N Al N Al AuAtomic Concentration (%)Depth () Ti

PAGE 98

98 Figure 4-4: Depth profiles of Ti/Al/TiN/Ti/Au contacts (a) as deposited (b) annealed at 600oC. (c) annealed at 800oC and (d) annealed at 800oC and aged at 350oC. 0100020003000400050006000 0 20 40 60 80 100 (a) Ti C O Au Ga O Ti N NAtomic Concentratioon (%)Depth ()Al0100020003000400050006000 0 20 40 60 80 100 (b) N Ti O Au GaAtomic Concentratioon (%)Depth ()Al01000200030004000500060007000 0 20 40 60 80 100 (c) Ti O Au Ga O Ti N NAtomic Concentratioon (%)Depth ()Al01000200030004000500060007000 0 20 40 60 80 100 (d) Ti O Au Ga O Ti N NAtomic Concentratioon (%)Depth ()Al

PAGE 99

99 Figure 4-5: Depth profiles of Ti/Al/ZrN/Ti/Au contacts (a) as deposited (b) annealed at 600oC. (c) annealed at 800oC and (d) annealed at 800oC and aged at 350oC. 0100020003000400050006000 0 20 40 60 80 100 (a) N Ti C O Au Ga O Ti N ZrAtomic Concentratioon (%)Depth ()Al010002000300040005000 0 20 40 60 80 100 (c) N Ti O Au Ga Ti N ZrAtomic Concentratioon (%)Depth ()Al0100020003000400050006000 0 20 40 60 80 100 (d) N Ti O Au Ga Ti N ZrAtomic Concentratioon (%)Depth ()Al0100020003000400050006000 0 20 40 60 80 100 (b) Al Au N Ti C O Au Ga O Ti N ZrAtomic Concentratioon (%)Depth ()Al

PAGE 100

100 Figure 4-6: Depth profile of T i/Al/Ni/Au contact annealed at 500oC.0100020003000 0 20 40 60 80 100 Depth ()NAtomic Concentration (%)Au Ni Ga Ti Al O Ni-500oC

PAGE 101

101 Figure 4-7: Specific contact resistan ce as a function of anneal time. 6080100120140160180 1x10-51x10-41x10-3 Ti/Al/Pt/Au Ti/Al/TaN/Ti/Au Ti/Al/TiN/Ti/Au Ti/Al/ZrN/Ti/Au Specific Contact Resistance (-cm 2 )Anneal Time (s)

PAGE 102

102 Figure 4-8. Specific contact resistance as a function of long term thermal aging. 024681012141618202224 1x10-51x10-41x10-3 Ti/Al/Pt/Au Ti/Al/TaN/Ti/Au Ti/Al/TiN/Ti/Au Ti/Al/ZrN/Ti/Au Specific Contact Resistance (-cm2)Days

PAGE 103

103 CHAPTER 5 BORIDE-BASED SCHOTTKY CONTACTS TO P-GAN 5.1 Introduction The unique combination of wide bandgap, hi gh breakdown field, high saturation velocity, and ability to form high-quality heterostructur es with good transport properties make GaN an ideal candidate for several device applicati ons including light emitting diodes (LEDs) for displays, laser diodes for data storage, and hi gh electron mobility transistors (HEMTs) for highpower and high frequency electronics. An important aspect for improved reliability of GaNbased devices is the development of more therma lly stable Ohmic and Schottky contacts to both n-type and p-type GaN. This aspe ct has been one of the major issues in the operating lifetime of III-nitride lasers and LEDs and it is yet to be overcome completely. Another example is AlGaN/GaN HEMTs in advanced power amplifiers and converters for which the development of Ohmic and Schottky contacts ensuring long-time operation under uncooled a nd high temperature conditions without metal spiking or loss of edge acuity remains a challenge. One approach that has proven promising for contact formation on n-GaN is the use of very high melting temperature metals like W and WSix [24,68,71-74]. Recently, other promising metallurgies based on transition metal borides like W2B, W2B5, ZrB2, CrB2, and TiB2 have also been proposed [75-78]. The interest in borides is due to their high melting temperature (e.g. 3200C for ZrB2) and high thermodynamic stability. For p-GaN, most rectifying contact studies have been performed with simple metals such as Ni Pd, Ti, Cr, and Pt, [17, 54-66] and studies on boride-based metallurgies are still in their infancy. In this work, we report on the annealing and measurement temperature dependence of W2B and W2B5-based rectifying contacts to p-GaN. The barrier heights determined by X-Ray Photoelectron Spectroscopy (XPS) are in the rang e 2.7-2.9 eV. These contacts are found to be

PAGE 104

104 reasonably stable upon annealing to ~600-700C. While current-voltage characteristics as a function of measurement temperature suggest thermionic field emission (TFE) over a Schottky barrier as the most dominant mechanism of fo rward current flow, the bias and measurement temperature dependence of the reve rse-bias current indicates that leakage must originate from surface leakage or generation in the depletion layer. 5.2 Experimental Details The p-GaN samples were 1 m-thick Mg-doped GaN layers grown by metal organic chemical vapor deposition (MOCVD) on 1 m-thick undoped buffers on c-plane Al2O3 substrates. The hole concentration obtained fr om Hall measurements after acceptor activation annealing was ~1017 cm-3. The surface was cleaned by sequent ial rinsing in acetone, ethanol and 10:1 H2O:HCl prior to insertion in the depositio n chamber. Boride(500)/Pt(200)/Au(800), where the boride was either W2B or W2B5, was used as the Schottky metallization scheme. Au was added to lower the contact sheet resistance while Pt is a diffusion barrier. All metals or compounds were deposited by Ar plasma-assisted rf (13.56 MHz) sputtering at a pressure of 15 mTorr and rf powers of 250-400 W. The contacts were patterned by liftoff of lithographicallydefined photoresist and annealed at temperat ures up to 700C for 1 min in a flowing N2 ambient in a RTA furnace. For Ohmic contacts, we used Pt/Au annealed at 500C in O2 for 30s prior to deposition of the Schottky metallization. Ring contact geometry for the diodes was employed, with the Schottky contacts surrounded by the Ohmic contacts. The Schottky contact dots were 40 m in diameter, the surrounding was 60 m in inner diameter and 70 m in outer diameter. XPS, current-voltage (I-V), and auger electron spectroscopy (AES) were used to characterize boride-based rectifyi ng contacts. XPS measurements were taken with a Physical Electronics 5100LSci spectrometer with an al uminum x-ray source (energy 1486.6 eV). High-

PAGE 105

105 resolution spectra were acquired to determine the binding energy (i.e. chemical state) and concentration of specific elements observed in the survey spectra. Charge correction was performed by using the known posit ion of the C-(C,H) lin e in the C 1s spectra at 284.8 eV. The SBH was determined from the binding energy of the Ga 3d core level EB and the energy difference between that core le vel and the valence band maximum EVC according to BBVCEE .[62,79] The I-Vs were recorded over the temperature range 25-200C using a probe station and an Agilent 4145B parameter analyzer. For the relativ ely high doped p-GaN samp les investigated in this work, TFE over a Schottky barrier is expect ed to be the dominant mechanism of forward current flow [56,57,62,63,80]. We fit the forward I-V characteristics to the relation for TFE over a barrier [81] 0 0expFeV JJ E (5.1) where JF is the current density, e is the electronic charge, and V is the applied voltage. In Eq. (1), 00000cothB E EEkT is the characteristic energy re lated to the tunneling probability, kB and T being the Boltzmanns constant and absolute temperature, respectively. For p-GaN, 12 3 007.7510AEeV Ncm, [56,62] NA being the density of acceptors. The saturation current density J0 is given by 0.5 00 0 00 0exp coshB B BBBATEqV e e J kEkTkTE (5.2) where A*=103.8 Acm-2K-2 is the effective Richardsons constant for p-GaN, [82] B is the barrier height, and =(EF-EV)/e is the difference between the valence band maximum and the position of the Fermi level. This parameter depends on the doping level, but, as will be shown

PAGE 106

106 later, it only slightly influences the SBH of W2B and W2B5-based rectifying contacts to p-GaN. Nakayama et al. [83] have suggested that a hole concentration of 21017 cm-3 in p-GaN samples leads to =0.13 V. This hole concentration is very clos e to that achieved in the present study so that we also assume =0.13 V. Auger electron spectrosc opy (AES) was used to analyze th e depth profiles. A Physical Electronics 660 Auger Microprobe Electron Beam at 10 keV, 0.3 A, 30 from sample normal was used for the data collection while for profiling the ion beam conditions were 3 keV Ar+, 2.0 A, (4 mm)2 raster with sputter-etch rates of 80, 40, and 104 /min for the boride, Pt and Au layer respectively. A survey spectrum (a plot of the first derivative of the number of electrons detected as a function of energy) was used to determine the composition of the outer few nanometers of each sample. Quantifications we re accomplished by using elemental sensitivity factors. 5.3 Results and Discussion Figure 5-1 presents an example of the Ga 3d core level and the valence band spectrum collected on a p-GaN surface without a borid e overlayer. From Figure 5-1, we find EVC=17.54 eV, which is in reasonable agreement with the value of 17.8 eV reported previously. [62,84] After W2B deposition, Figure 5-1 shows that the binding energy of the Ga 3d core level is EB=20.25 eV, thereby yielding B=2.71 eV. Similarly, we find EB=20.41 eV for W2B5, which gives B=2.87 eV. These results are similar to the 2.68-2.78 eV values reported by Yu et al. [56] for Ni/p-GaN. Based on the Schottky-Mott model [52], the SBHs for n-GaN and p-GaN should add up to the GaN band gap 3.4 eV. For W2B and W2B5-based rectifying c ontacts to n-GaN, Khanna et al. [76,77] have reported B~0.55 eV, thereby suggesting B~2.85 eV for p-GaN. Our XPS results are clearly in excellent agreem ent with the predictions of this model.

PAGE 107

107 The I-V characteristics as a function of the measurement temperature for the W2B-based diodes are shown in Figure 5-2. All log I vs V curves measured between 25 and 200C exhibit a linear region at low forward bi as (up to ~2.0 V) and a less steep region at higher voltages. Furthermore, all log I vs V curves are parallel. As noted in refs. [56,57,62,63] for p-GaN samples with relatively high doping levels (>1017 cm-3), this latter result is inconsistent with the thermionic emission model but typical for car rier transport with a dominant tunneling component. Note that Shirojima et al. [61] have reported accurate determination of the SBH using the TE model. However, the acceptor conc entration in their p-GaN samples was relatively low (~1016 cm-3). In our case, the Mg concentration is expected to be in the 1019 cm-3 range, thus confirming that the forward current cannot be an alyzed using the thermionic emission model. Figure 5-3 displays the I-V characteristics obtained from th e diodes annealed at different temperatures. The extracted characteristic ener gy related to the tunneling probability as a function of annealing temperature is shown in Figure 5-4. For W2B, the value of E0 is seen to remain stable in the 120 meV range and then increases beyond ~500C. This value of E0 corresponds to an acceptor density of about 21020 cm-3, which is higher than that expected from the Mg concentration alone (E0~25meV for NA~1019 cm-3). This can probably be attributed to the presence of acceptor-like deep level defects induced by both the high Mg doping [63,85,86] and the sputter-deposition proce ss [29,87]. In the case of W2B5, a slight decrease of E0 is observed at intermediate annealing temperatur e which can probably be attributed to the annealing of sputterdamage [29,87]. The influence of annealing on the SBH derived from the I-V measurements of W2B/Pt/Au and W2B5/Pt/Au contacts is presented in Figure 5-5. Over the whole range of temperatures investigated, the SBH only slightly decreases with increasing annealing temperature. However,

PAGE 108

108 these values are unphysical, being larger than th e GaN band gap, and are inconsistent with both our XPS measurements and the pred ictions of the Schottky-Mott mode l. One may argue that this discrepancy results from the value of used in our calculations. However, it is shown in Figure 5-6 that over the wide range of values investigated, this parameter only slightly influences the SBH. The discrepancy between XPS and I-V measurements is more likely related to the presence of an interfacial defect layer or due to the pres ence of an interfacial oxide layer. Indeed, it was shown by Shiojima et al. [88] using high-temperature isothe rmal capacitance transient spectroscopy that the de fects induced by the high Mg doping in p-GaN are essentially located at the surface vicinity and that this region acts as a series capacitance [88,89]. Therefore, when such defect or oxide layer is present, the carriers have to tunnel through an additional barrier, thus resulting in an apparent SBH given by 0 BBB where 0 B is the true SBH between the contact metal and p-GaN and where B is the increment of the SBH due to the presence of the thin defect or oxide layer. Our sepa rate capacitance-voltage measurements (C-V) have indeed shown that by accounting for such def ect layer, the SBH obtained from the C-V characteristics becomes similar to that determined by XPS [80]. We have calculated a corrected junction capacitance Ccorr for W2B/p-GaN from the equi valent circuit using the simple relation 111corr m oxCCC where Cm and Cox are the as-measured and oxide capacitances respectively. From our XPS meas urements, the true SBH for W2B/p-GaN is 0~2.7eVB. Using our C-V measurements, Figure 5-7 shows that the value of Cox required to obtain a SBH similar to that determined by XPS is Cox~0.70 nF. Assuming a contact area of ~2 10-9 m2 and a relative permittivity of ~10 for the oxide layer [90], we estimate an oxide layer thickness of ~0.25 nm, i.e. about one monolayer. For 20 m-diameter diodes, the value of Cox required to fit the XPS

PAGE 109

109 data was 4 times lower than that for 40 m-diameter diodes, as expected from the surface area ratio. In addition, the density of acceptors obtained from the corrected C-Vs was 1.8 1020 cm-3, which is very similar to the value determined above from the I-Vs. Table 5-1 summarizes the calculated barrier heights from XPS, IV with both thermionic and thermionic field emission, and CV measurements. Figure 5-8 presents the influence of the m easurement temperature on the reverse-bias current for the W2B-based diodes. All log I vs V curves measured between 25 and 200C exhibit a linear region at high reverse bias and a steeper variation at lowe r reverse voltages. In addition, as the measurement temperature increases, the re verse leakage current also increases. Given the large barrier height typical for rectifying contacts to p-GaN, one can assume that thermionic emission over the Schottky barrier makes only a ne gligible contribution to the reverse-bias current flow. This assumption is furt her supported by the f act that all log I vs V curves in Figure 5-8 are parallel, which is clearly inconsistent with the thermionic emission model. On the other hand, in the tunneling approach the reverse current density JL is given by the Fowler-Nordheim tunneling expression 2expLbbJCEED where C=C(B) and D=D(B) are parameters independent of the measurement temperature [91-93] For tunneling, the re verse-bias current is thus independent of the measurement temperature, which again is inconsistent with the experimental data displayed in Figure 5-6. Leakage must therefore originate from other mechanisms such as generation in the depletion layer or surface leakage. The large band gap of GaN makes the intrinsic carrier c oncentration in a depletion regi on very small, suggesting that contributions to the reverse-bias leakage from generation in the depletio n layer are negligible. However, given the large number of deep-level de fects located at the surf ace vicinity [88] as a result of both the high Mg doping [29,85,86] and the sputter-deposition process [29,87], one may

PAGE 110

110 expect these defects to act as generation centers when the carrier density is below its equilibrium value as in the reverse-bias regime of a Schottky diode. The breakdown voltage, VB, of the boride-based rectif ying diodes as a function of annealing temperature is shown in Figure 5-9 VB shows the opposite tren d to the characteristic energy related to the tunneling probability displayed in Figure 5-4, decreasing where E0 increases. As E0 is directly related to the density of acceptor-like deep-level defects, this observation is consistent with our expectations that leakage esse ntially results from a defectmediated mechanism. As observed for boride-base d rectifying contacts to n-GaN [75,76], the improvement of breakdown voltage at intermedia te annealing temperatures is likely to result from the annealing of sputter-damage in the near-surface region of GaN. At high annealing temperature, VB is seen to decrease, indicating that the diodes become very leaky. AES depth profiles of the as-deposited and a nnealed contacts are shown in Figure 5-10. For both diodes, the as-deposited layers exhibit relatively sharp interfaces, consistent with the featureless surface morphology observed with both optical and scanning electron microscopy. After high-temperature annealing, W2B-based contacts show the onset of metallurgical reaction in the contact scheme whereas W2B5 presents significant interdiffusion of layers. 5.4 Conclusions In summary, our XPS measurements have shown that W2Band W2B5-based rectifying contacts to p-GaN produce barrier he ight in the 2.7-2.9 eV range, in excellent agreement with the predictions of the Schottky-Mott model. By comparison, the SBH determined from the I-Vs using the TFE model are unphysically large and much higher than those deduced from XPS due to the presence of a defect layer acting as an add itional barrier to carrier transport. The apparent SBH only slightly decrease s upon annealing to 600-700 C due to the onset of metallurgical

PAGE 111

111 reaction with the GaN. This suggests that that the boride-based metallurgies may be promising for rectifying contacts to p-GaN where thermal stability is a critical issue. The bias and measurement temperature of the reverse leakage current indicate s that leakage must originate from surface leakage or generation in the de pletion region through deep-level defects.

PAGE 112

112 Table 5-1: Comparison of differe nt barrier height calculations XPS TE TFE C-V B = 2.7 eV B = 1.2 eV n = 4.4 B = 3.8 eV E0= 110 meV B = 4.1 eV

PAGE 113

113 24222018164 0 2 17.54 eV Valence band region (X5) Ga 3d Intensity (a.u.)Binding Energy (eV) 26242220181614 2 W2B5/ p-GaN W2B / p-GaN Ga 3d Intensity (a.u.)Binding Energy (eV) Figure 5-1: XPS spectra without (top) and with (bottom) a boride overlayer. The left-hand spectrum in the top figure corresponds to th e Ga 3d core level wh ereas the right-hand panel presents the spectrum of the valence band region.

PAGE 114

114 0.00.51.01.52.02.53.03.54.0 10-61x10-51x10-410-310-2 Current (A)Voltage (V) 200oC 150oC 100oC 50oC 25oC Figure 5-2: Forward current-vo ltage characteristic of W2B-based Schottky diodes as a function of measurement temperature.

PAGE 115

115 0123456 10-710-61x10-51x10-410-3 W2B/Pt/Au Current (A)Voltage (V) As-deposited 300oC anneal 600oC anneal 0123456 10-710-61x10-51x10-410-3 W2B5/Pt/Au Current (A)Voltage (V) As-deposited 300oC anneal 600oC anneal Figure 5-3: Forward current-vo ltage characteristic of W2B-based (top) and W2B5-based (bottom) Schottky diodes as a function of annealing.

PAGE 116

116 0100200300400500600700800 0 20 40 60 80 100 120 140 160 180 200 220 240 260 NA~5x1019cm-3 NA~1019cm-3 W2B/Pt/Au W2B5/Pt/Au E0 (meV)Anneal temperature (oC) Figure 5-4: Influence of the annealing temperat ure on the characteristic energy related to the tunneling probability. Dashed and dotte d lines correspond to the values of E0 for NA~1019 and 51019cm-3 respectively.

PAGE 117

117 0100200300400500600700800 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 W2B/Pt/Au W2B5/Pt/Au B (V)Anneal temperature (oC) Figure 5-5: Influence of the annealing temper ature on the apparent Sc hottky barrier height derived from IV measurements.

PAGE 118

118 0.080.120.160.200.24 3.0 3.5 4.0 4.5 5.0 Lightly-doped Highly-doped W2B/Pt/Au W2B5/Pt/Au B (eV) (eV) Figure 5-6: Dependence of th e apparent Schottky barrier height on the parameter defined as the difference between the valence band maximum and the position of the Fermi level. Low and high

PAGE 119

119 -4-3-2-1012345 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 Voltage (V)1/C2 (1021 F-2) As-measured After oxide correction Figure 5-7: As-measured and afte r oxide correction dependence of C-2 versus V of Au/Pt/W2B/pGaN Schottky diodes. The measurement frequency was set to 1 kHz.

PAGE 120

120 -4.0-3.5-3.0-2.5-2.0-1.5-1.0-0.5 10-810-710-61x10-51x10-4 Current (A)Voltage (V) 200oC 150oC 100oC 50oC 25oC Figure 5-8: Reverse currentvoltage characteristic of W2B-based Schottky diodes as a function of measurement temperature.

PAGE 121

121 0100200300400500600700800 0 1 2 3 4 5 6 7 8 9 W2B/Pt/Au W2B5/Pt/Au VB (V)Anneal temperature (oC) Figure 5-9: Influence of the annea ling temperature on the breakdown voltage.

PAGE 122

122 0 50 100150200 0 10 20 30 40 50 60 70 80 90 100 W2B/Pt/Au as-deposited Ga N O Pt Au W B Atomic concentration (%)Sputter depth (nm)(a) 0 50 100150200 0 10 20 30 40 50 60 70 80 90 100 W2B5/Pt/Au as-deposited Ga N O Pt Au W B Atomic concentration (%)Sputter depth (nm)(b) 0 50 100150200 0 10 20 30 40 50 60 70 80 90 100 W2B/Pt/Au annealed @700oC Ga N O Pt Au W B Atomic concentration (%)Sputter depth (nm)(c) 0 50 100150200 0 10 20 30 40 50 60 70 80 90 100 W2B5/Pt/Au annealed @600oC Ga N O Pt Au W B Atomic concentration (%)Sputter depth (nm)(d) Figure 5-10: Depth profiles of W2B/Pt/Au contacts and W2B5/Pt/Au rectifying contacts (a,b) before and (c,d) after annealing at 600 C.

PAGE 123

123 CHAPTER 6 BORIDE AND IR BASED CONTACTS FOR L IGHT EMITTING DIODES 6.1 Introduction InGaN/GaN multiple quantum well light-emitting diodes (MQW-LEDs) are commercially available in a broad range of wa velengths for use in applications such as full color displays, traffic signals, and exterior ligh ting. There is also interest in shorter wavelength LEDs with AlGaN active regions to excite down conversion phosphors for white light [94]. Nevertheless, to compete with fluorescent and other high-efficiency lighting sources, it is essential to drive GaNbased LEDs at very high current densities to ma ximize light output. One drawback of the high current densities is self-heating of the he terostructure which can produce Ohmic contact degradation and generation of non-radiative recombination centers [95-98]. Cao et al. [99-101] demonstrated improved heat dissipation and cu rrent spreading by reduc ing the density of dislocations in the heterostructure through the use of free-stan ding GaN instead of the more common sapphire substrates. Further reduction of self-heating effects and improved efficiency and operating lifetime can be obt ained through the use of low-resistance and thermally stable Ohmic contacts to the p-GaN layer. Indeed, pow er dissipation by Joul e heating across the pGaN/metal interface at high current injection le vels was found to produce indiffusion of the Ohmic contact elements along disl ocations in the III-nitride-bas ed epilayers, leading to an electrical short of th e pn junction. [102,103] The conventional metallization schemes used for making Ohmic contacts to p-GaN for InGaN/GaN LEDs are based on high work function me tals such as Ni, Pd, Cr or Pt with an overlayer of Au to reduce the sheet resistance. Specific contact resistances of ~10-2-10-4 .cm2 are obtained by annealing at 450-650 C [106,107]. These contacts have stability problems at high temperature. For example, the initial Au/Ni/ GaN structure transforms to Ni/Au/GaN with a

PAGE 124

124 rough Ni surface after annealing at 600C [44,108] A way to prevent exce ssive intermixing and contact morphology degradation is to use a high -melting-point diffusion barrier in the contact stack [109-111]. TiB2, with a melting temperature of ~3000C, reasonable electrical resistivity (28 .cm) and thermal conductivity (26 W.m-1.K-1), and heat of formation comparable to those for silicides or nitrides [112], s hows promise as a diffusion barrier. In this letter, we report on the long-term annealing characteristics at 200-350C of InGaN/GaN MQW-LEDs with TiB2and Ir-based p-Ohmic contacts. This high-temperature stress stimulates accelerated aging of GaN-ba sed LEDs and gives an idea of the expected reliability of the Ohmic contacts. The use of TiB2 or Ir as a diffusion barrier in Ni/Au-based contacts is found to produce superior long-term stability of turn-on voltage, leakage current and output power. 6.2 Experimental The MQW-LED structures were grown by metal organic chemical vapor deposition on cplane sapphire substrates. The layer structure consisted of a low-temperature GaN buffer, 3 mthick n-GaN, 0.1 m n-AlGaN clad, three period undoped InGaN/GaN MQW active, 0.1 m pAlGaN clad, and 0.3 m p-GaN layer. 1 m-deep mesas were fabricated using Cl2/Ar inductively coupled plasma etchi ng. Ohmic contacts to n-GaN were formed by lift-off of e-beam deposited Ti (20 nm) / Al(80 nm) / Pt(40 nm) / Au (80 nm) subsequently annealed at 800C for 1 min in a flowing N2 ambient in a rapid thermal ann ealing (RTA) furnace. The first pmetallization scheme investigated consisted of Ni (50 nm) / Au (80 nm) / Ir (50 nm) / Au (80 nm). The second contact scheme was Ni (50 nm) / Au (80 nm) / TiB2 (50 nm) / Ti (20 nm) / Au (80 nm). Ti was added to im prove the adherence of Au on TiB2. The Ni/Au layers were deposited by e-beam evaporat ion while the Ir/Au and TiB2/Ti/Au overlayers were deposited by

PAGE 125

125 Ar plasma-assisted rf magnetron sputtering. For comparison, devices with Ni/Au contacts were also fabricated using the same wafer. All cont acts were patterned by liftoff of lithographicallydefined photoresist and annealed at 600C for 1 min in a flowing N2 ambient. Prior to their insertion in the e-beam chamber, the surface of both nand p-type GaN were cleaned in a 1HCl:10H20 solution for 1 minute. Figure 6-1 show s a typical optical micrograph of asfabricated MQW-LEDs with Ni/Au/TiB2/Ti/Au Ohmic contacts. Similar contact morphology was achieved using the other metallization schemes. All devices were first aged for a period of 10 days at 200C on a heater plate in air. The sa mples were then removed from the heater block, allowed to cool to room temperature, and charac terized before being returned for further 35 days aging at 350C. MQW-LEDs were analy zed by luminescence-current-voltage (L-I-V) measurements using a probe station and an Agilent 4145B parameter analyzer. The light output power was measured using a Si photodetector located at ~2 cm from the sample surface. Electroluminescence (EL) spectra were acq uired by a fiber optic spectrometer. 6.3 Results and Discussion Figure 6-2 presents the 300 K L-I characteristics from each as-fabricated device. Typical EL spectra measured for different forward currents are also shown in the inset. For an injected current of 500 A, the EL peak wavelength of the InGaN/GaN MQW-LEDs is 459 nm with a full width at half maximum (FWHM) of 26 nm. As the injected curren t increases, the peak exhibits a blue shift while the FWHM remains fa irly constant. This behavior results from the decrease by carrier screening of the quantum-confined Stark effect induced by the built-in piezoelectric field in the strain ed InGaN/GaN QWs [113]. The L-I characteristics are similar for all metallization schemes investigated, showing that the TiB2/Ti/Au or Ir/Au overlayers did not reduce the light output with respect to the mo re conventional Ni/Au c ontacts. The non-linear

PAGE 126

126 increase and relatively low output power can pr obably be attributed to enhanced current crowding and self-heating effects [114,115]. These effects are expected to be particularly important for the LED structures investigated in th is work due to the high density of dislocations in InGaN and GaN epilayers grown on sapphire as well as the low th ermal conductivity of sapphire [99-101]. As the dislocation density or LED junction temperature increases, confinement in the MQW becomes less efficient, l eading to premature satu ration of the emission power. I-V characteristics from MQW-LEDs with Ni/Au, Ni/Au/TiB2/Ti/Au, and Ni/Au/Ir/Au Ohmic contacts to the p-GaN layer are shown in Figure 6-3. For all diode s, the ideality factor was >2, consistent with other data reporte d previously [95-101,116,117]. There is still no agreement whether such high values result from tunnelling [101,117], recombination in the depletion layer [117], or rectifyi ng junctions present in heterost ructure diodes [118]. The turn-on voltage (voltage at which the current reaches 0. 05 mA) and reverse leakage current (measured at -5V) are given in Table 6-1. The uncertainty on each value was estimated from the deviation between different devices on the same wafer. Fo r as-fabricated devices, the turn-on voltage was 3.2-3.6V and the reverse leakage current was ~10-6 A. Such high leakage currents are commonly observed in GaN-based LEDs grown on sapphire substrates and esse ntially result from the high density of dislocations in such samples [101]. After 10 days agi ng at 200C, an increase of the turn-on voltage to ~4.1 V and a decr ease of the leakage current to ~10-9 was observed for all devices. After aging at 350C for an additional 30 days, MQW-LEDs with TiB2and Ir-based Ohmic contacts show turn-on volta ge and leakage current similar to those in the as-deposited state whereas there was serious degradation in LEDs fabricated with Ni/Au contacts. This reduction of injected current at a given forward voltage had a large impact on the EL output as

PAGE 127

127 shown in Figure 6-4. While the brightness of MQW-LEDs with Ni/Au Ohmic contacts at 10 V (I = 80 A) was low, optical images of aged MQW-LEDs with boride-based Ohmic contacts show much stronger emission, even at a lower applied voltage of 4.5 V (I = 300 A). Several mechanisms can be invoked to explain the drastic degradation of Ni/Au contacts upon aging, including formation of an islanded contact morphology, formation of reactions with the GaN resulting in a modification of the doping profile, and excessive intermixing of Ni and Au leading to oxidation on the rough Ni surface and an increase of th e series resistance [44,109,119]. Although the dominant degradation mechanism remains unknown, the use of TiB2/Ti/Au or Ir/Au overlayers on the Ni/Au-based Ohmic contacts clearly improves the long-te rm thermal stability of InGaN/GaN MQW-LEDs. 6.4 Conclusions In conclusion, TiB2 and Ir diffusion barriers in pOhmic contacts on GaN LEDs produced less change in turn-on voltage, leakage current, and output power after long-term annealing at 200C and 350C for 45 days. These schemes look pr omising for high temperature applications where improved stability over Ni/Au is mandatory. Table 6-1: Influence of long-term aging at 200 C and 350C on the turn-on voltage and reverse current of InGaN/GaN MQW-LEDs. Turn-on voltage (V) Reverse current @ -5V (A) Contact to p-GaN Day 0 Day 10 @ 200C Day 45 @ 350C Day 0 Day 10 @ 200C Day 45 @ 350C Ni/Au 3.30.3 4.90.4 9.10.5 (2)10-6 (6)10-9 (5)10-5 Ni/Au/TiB2/Ti/Au 3.50.2 4.10.5 4.00.2 (5)10-6 (2)10-9 (2)10-8 Ni/Au/Ir/Au 3.30.2 4.30.3 3.90.3 (3)10-6 (2)10-9 (3)10-7

PAGE 128

128 Figure 6-1: Optical micrograph of an as-fabricat ed MQW-LED. The p-cont act at the center of the diode is 80 m in diameter.

PAGE 129

129 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 -4-20246810 0.0 0.2 0.4 0.6 0.8 1.0 (a) Day 0 Day 10 @ 200oC Day 45 @ 350oC(b) (c) Current (mA)Voltage (V) Figure 6-2 : L-I characteristics of MQW-LEDs with Ni/Au, Ni/Au/TiB2/Ti/Au, and Ni/Au/Ir/Au p-Ohmic contacts. The inset shows emissi on spectra from as-fabricated LEDs at various injection currents.

PAGE 130

130 0.00.51.01.52.0 0.0 0.1 0.2 0.3 0.4 0.5 0.6 440460480500 0 4 8 12 16 20 Intensity (a.u)Wavelength (nm) 1.5 mA 1 mA 750 A 500 A Luminescence (W)Current (mA) Ni/Au Ni/Au/TiB2/Ti/Au Ni/Au/Ir/Au Figure 6-3: Influence of long-term aging at 250 C and 350C on the I-V ch aracteristics of LEDs with (a) Ni/Au, (b) Ni/Au/TiB2/Ti/Au, and (c) Ni/Au/Ir/Au p-Ohmic contacts.

PAGE 131

131 (a) (b) Figure 6-4: Image of aged LEDs with (a) Ni/Au and (b) Ni/Au/TiB2/Ti/Au p-Ohmic contacts. In (a), the picture was taken for a forward bias of 10 V (I = 80 A), while the forward voltage in (b) was 4.5 V (I = 300 A).

PAGE 132

132 CHAPTER 7 CONCLUSION Im proved device processing is necessary in orde r to realize the full potential of GaN based electronics and optoelectronics. While improved material quality is critical, especially for pGaN as well as related alloys such as InGaN to increase the efficiency of green emitting devices, it is not the only challenge. Developing relia ble, stable, low resistance Ohmic and Schottky contacts to both nand p-type GaN is still a challenge and hence remains of interest to the GaN community at large. Without contacts that can withstand elevated temperatures and other harsh environments, many of the properties that ma ke GaN a unique and desirable semiconductor mean nothing. The goal of this work was to develop such cont acts to both nand p-GaN. In order to improve upon existing contact schemes, it is necessary to achieve at least comparable contact resistances or barrier heights. In addition, the contact s should be able to withstand more stressful processing conditions, such as elevated anneali ng temperatures, while maintaining predictable and stable characteristics. Perhaps the most cr itical factor is the abil ity of the contacts to withstand long periods in a harsh environment, simulated here by placement on a hot plate at temperatures of 200oC and 350 oC. A final consideration is the amount of intermixing of the contact layers, especially in Ohmic contacts to n-GaN, as large amounts of intermixing can lead to undesirable phases, such as AlAu4 and issues with lateral flow. In addition, if any undesirable phases may form at device operation temperatures, or if the contacts themselv es interact with the GaN at these temperatures, devices will prove unreliable during prolonged operation. To tackle these issues, three separate materials system s were examined: borides, nitrides, and Ir.

PAGE 133

133 The first section of this dissertation involved the fabrication of Ohmic contacts to p-type GaN. All three material systems mentioned we re examined for this use. Contacts were fabricated of the following structures: 1. Nickel / Gold / X / Titanium / Gold, where X is a chosen boride or nitride 2. X / Titanium / Gold, where X is a chosen boride of nitride 3. Nickel / Gold / Ir / Gold 4. Nickel / Ir /Gold Each of these was then subjected to a nnealing at temperat ures ranging from 300oC to 1000oC in a flowing N2 ambient for 60 s. Schemes 1 and 3 consistently produced Ohmic contacts in the range of 500-1000 oC, 2 only inconsistently at specific temperatures, 3 at only 500oC and 4 not at all. Schemes 1 and 3 produced specific contact resistance values similar to those reported for Nickel / Gold contacts in the literature. Contacts of scheme 1 were subjected to long term thermal aging on a hot plate at both 200oC and 350oC. Nitride-based contacts failed early in aging, however boride based contacts displayed stable specific contact resistances throughout. Auger Electron Spectroscopy showed breakdown of the nitride structure during aging due to severe intermixing with the GaN. The borides showed minimal intermixing with the GaN, accounting for their stability. AES depth profiles of scheme 3 revealed severe intermixing of the contacts with the GaN at anneal temperatures above 500oC, accounting for their failure at elevated anneal temperatures. Contacts fabri cated with scheme 4 were not Ohmic due to the absence of Au at the surface to promote increased hole concentration as well as severe intermixing at high temperatures. The second section of the work dealt with the examination of nitride based contacts to nGaN. Contacts using the borides and Ir have previously been examined. Contacts of the

PAGE 134

134 structure Titanium / Aluminum / X / Titanium / Gold were fabricated using standard semiconductor processing procedures, where X is either TaN, TiN, or ZrN. For comparison, conventional contacts of the form Titanium / Aluminum / Platinum / Gold were fabricated as well. The contacts were then subjected to a range of anneal temperatures between 500oC and 1000oC in a flowing N2 ambient for 60 s, as well as anneals at 800oC for up to 180 s, in order to determine if the nitrides would allow for an increased thermal budget. Contacts were also aged on a hot plate at 350oC and aged for a period of 24 days. Current-voltage measurements, Auger Electron Spectroscopy, and Scanni ng Electron Microscopy were used to characterize the contacts. Current-voltage measurements of th e nitride contacts did not show any improvement over the performance offered by the conventional Pt-b arrier contacts in any of the experiments. However, AES depth profiles revealed that th e nitride based contac ts displayed far less intermixing than Ni-barrier contacts, even wh en the nitrides were annealed at higher temperatures. Further, the profiles of therma lly aged nitride contacts displayed no noticeable difference from unaged contacts. Thus, while electrical performance of these contacts was essentially unchanged from that of the conve ntional ones, they do offer the benefit of dramatically reducing the amount of intermixing between layers even after being subjected to harsh long term aging. This is significant, as on e of the sources of failure for small gate width GaN devices is lateral flow arising due to th e mixing of the Al underlayer and Au overlayer, leading to the formation of the viscous AlAu4 phase. This leads to short circuiting and thus failure of devices. Chapter five dealt with the fabrication of Sc hottky contacts to p-GaN. For this purpose, W2B and W2B5 sputter targets were chosen. Nitrides we re ignored due to the severe intermixing expected between them and the GaN. It is also expected Ir would diffuse a great deal into the

PAGE 135

135 GaN at elevated annealing temperatures, also ma king it unsuitable for use as a Schottky contact. Contacts were fabricated with the scheme X / Plat inum / Gold, where X is the boride. Platinum was chosen as the adhesive layer instead of T itanium in order to eliminate formation of TiNx phases. Current-voltage measurements were used to evaluate the barrier he ight of the contacts. Because IV curves measured at different temperat ures were parallel, it was determined that the current transport mechanism present was thermioni c field emission. Unphysical barrier heights of approximately 4 eV were observed for anneals up to 700oC, with good stability. A slight decrease is seen at increased annealing temperat ures. This likely was the result of either an increased near surface defect concentration due to sputtering or due to incorporation of an interfacial oxygen layer between the GaN and the contact. X-ray Photoelectron Spectroscopy measurements of thin boride la yers on GaN reveal a true barrier height of 2.85 eV, in close agreement with that predicted by the Schottky-Mo tt model. Capacitance-voltage measurements confirm the large barrier height in agreement with the IV results, but if an interfacial layer of 1-2 monolayers is included the corr ected barrier height would be in agreement with XPS measurements. Chapter six deals with the application of bor ide and Ir-based Ohmic contacts to p-GaN for light emitting diodes. Contacts to LEDs were fabricated with the structures Nickel / Gold, Nickel / Gold / Titanium Boride / Titanium / Gold, and Nickel / Gold / Iridium / Gold. They were then annealed at 500oC, in order to prevent damage of the LED structure as well as metal spiking into the active layers. Initial IV a nd photoluminescence measurements revealed all contacts to have similar propert ies and performance. Contacts were then aged initially at 200oC, after which all displayed a slight degradation of performance. After further aging at 350oC, the

PAGE 136

136 Ni/Au contacts had degraded severely while the TiB2 and Ir-based contacts maintained good performance. This increased device stability confirms the earlier work. In conclusion, contacts to nand p-GaN were fabricated. Nitride diffusion barrier contacts to n-GaN show much less intermixing that that present with a Ni di ffusion barrier, although no improvement in the IV characterist ics either as a functi on of annealing or duri ng long term aging. Ohmic contacts to p-GaN using th e borides displayed both the abil ity to withstand a wide range of anneal temperatures, indica ting their increased thermal budget as compared to conventional Ni/Au contacts, as well as far superior long term aging characteristics due to the stability of the borides on GaN. This has implications for processing of devices using p-GaN which may be used at elevated temperatures. While the specific Ohmic contacts may not be completely suitable for immediate use in light emitting a pplications due to thei r poor reflectivity and opacity, they show great promise. Development of transparent contacts using very thin layers of these borides may be possible, as are improveme nts in the reflectivity by experimenting with different metallization schemes and low anneal temperatures. In addition, if MOSFETs are to be achieved using GaN, low resistance, stable Ohmi c contacts will be necessary and these may fit the requirements.

PAGE 137

137 LIST OF REFERENCES 1. S. A. Campbell, The Science and Engineer ing of Microelectronics Fabrication 2nd ed., Oxford University Press, Oxford, UK, 2001. 2. J. C. Bean, in High-Speed Semiconductor Device s, edited by S.M. Sze, John Wiley & Sons, New York, NY, 1990. 3. S. M. Sze, Physics of Semiconductor Devices, John Wiley & Sons, New York, NY, 1981. 4. D.C. Look, Mater. Sci. Eng. B80, 383 (2001). 5. J. Sun, H. Fatima, A. Koudymov, A. Chitnis, X. Hu, H.-M. Wang, J. Zhang, G. Simin, J. Yang, and M.A. Khan, IEEE Electron Dev. Lett., 24, 375 (2003). 6. A. P. Zhang, L. B. Rowland, E. B. Kaminsky, J. W. Kretchmer, R. A. Beaupre, J. L. Garrett J. B. Tucker, B. J. Edward, J. Foppes and A. F. Allen, Solid-State Electron 47 821(2003). 7. A. P. Zhang, L. B. Rowland, E. B. Kaminsky, V. Tilak, J. C. Grande, J. Teetsov, A. Vertiatchikh and L. F. Eastman, J. Electron. Mater. 32 388(2003). 8. V. Kumar, L. Zhou, D. Selvanathan, and I. Adesida, J. Appl. Phys. 92, 1712 (2002) 9. A. Motayed, R. Bathe, M.C. Wood, O.S. Diouf R. D. Vispute, and S. Noor Mohammad, J. Appl. Phys. 93, 1087 (2003) 10. J. Burm, K. Chu, W. A. Davis, W. J. Schaff, L.F. Eastman, and T. J. Eustis, Appl. Phys. Lett. 70, 464 (1997) 11. M. E. Lin, Z. Ma, F. Y. Huang, Z. F. Fan, L. H. Allen, and H. Morko, Appl. Phys. Lett. 64, 1003 (1994). 12. B. P. Luther, S. E. Mohney, T. N. Jackson, M. Asif Khan, Q. Chen, and J. W. Yang, Appl. Phys. Lett. 70, 57 (1997). 13. Z. Fan, S. N. Mohammad, W. Kim, Aktas, A. E. Botchkarev, and H. Morko, Appl. Phys. Lett. 68, 1672 (1996) 14. S. Ruvimov, Z. Liliental-Weber, J. Washburn, D. Qiao, S. S. Lau, and Paul K. Chu, Appl. Phys. Lett. 73, 2582 (1998). 15. A. N. Bright, P. J. Thomas, M. Weyland, D. M. Tricker, C. J. Humphreys, and R. Davies, J. Appl. Phys. 89, 3143 (2001). 16. N. A. Papanicolaou, M. V. Rao, J. Mittereder and W. T. Anderson, J. Vac. Sci. Technol. B 19, 261 (2001). 17. Q. Z. Liu and S. S. Lau, Solid-S tate Electronics, 42, 677-691 (1998).

PAGE 138

138 18. R. Li, S. J. Cai, L. Wong, Y. Chen, K. L. Wang, R. P. Smith, S. C. Martin, K. S. Boutros, J. M. Redwing, IEEE Electron De vice Letters, 20, 323 (1999). 19. S. T. Sheppard, K. Doverspike, W. L. Pribble, S. T. Allen, J. W. Palmour, L. T. Kehias, T. J. Jenkins, IEEE Electron Devi ce Letters, 20, 161 (1999). 20. D. Qiao, Z. F. Guan, J. Carlton, S. S. Lau, and G. J. Sullivan, Appl. Phys. Lett. 74, 2652 (1999) 21. V. Tilak, R. Dimitov, M. Murphy, B. Green, J. Smart, W. J. Schaft, J. R. Shealy and L. F. Eastman, Mat. Res. Soc. Symp. Proc., 622, T7.4.1(2000). 22. S. -H. Lim, J. Washburn, Z. Liliental-Weber, and D. Qiao, Appl. Phys. Lett. 78, 3797 (2001). 23. M. Asif Khan, M. S. Shur and Q. Chen, Appl. Phys. Lett., 68, 3022(1996). 24. S. Murai, H. Masuda, Y. Koide, and M. Murakami, Appl. Phys. Lett. 80, 2934 (2002) 25. C. Lu, H. Chen, X. Lv, X. Xie, and S. N. Mohammad, J. Appl. Phys. 91, 9218 (2002) 26. J. S. Foresi and T. D. Mousta kus, Appl. Phys. Lett., 62, 2859(1993). 27. M. W. Fay, G. Moldovan, P. D. Brown, I. Harri son, J. C. Birbeck, B. T. Hughes, M. J. Uren, and T. Martin, J. Appl. Phys. 92, 94 (2002). 28. K. O. Schweitz, P. K. Wang, S. E. Mohney, and D. Gotthold, Appl. Phys. Lett. 80, 1954 (2002). 29. M. W. Cole, D. W. Eckart, W. Y. Han, R. L. Pfeffer, T. Monahan, F. Ren, C. Yuan, R. A. Stall, S. J. Pearton, Y. Li and Y. Lu, J. Appl. Phys. 80, 278 (1996). 30. A. Zeitouny, M. Eizenberg, S. J. Pearton, and F. Ren, J. Appl. Phys. 88, 2048 (2000). 31. D. Selvanathan, F. M. Mohammed, A. Tesfayesus and I. Adesida, J.Vac.Sci.Technol. B22 2409(2004). 32. B. Luo, F. Ren, R. C. Fitch, J. K. Gillespie, T. Jenkins, J. Sewell, D. Via, A. Crespo, A.G. Baca R. D. Briggs, D. Gotthold, R. Birkhahn, B. Peres and S. J. Pearton, Appl. Phys. Lett. 82, 3910 (2003). 33. H. W. Jang and J. -L. Lee, J.Appl.Phys.93,5416(2003). 34. R. C. Fitch, J. K. Gillespie, N. Moser, T. Jenkins, J. Sewell, D. Via, A. Crespo, A.M. Dabiran, P. P. Chow, A. Osinsky, J. R. LaRoch e, F. Ren and S. J. Pearton, Appl.Phys.Lett. 84, 1495 (2004). 35. R. C. Fitch, J. K. Gillespie, N. Moser, T. Jenkins, J. Sewell, D. Via, A. Crespo, A.M. Dabiran, P. P. Chow, A. Osinsky, J. R. LaRoche, F. Ren and S. J. Pearton J.Vac.Sci.Technol. B22, 619(2004).

PAGE 139

139 36. D. Selvanathan L. Zhou, V. Kumar, I. Adesida and N. Finnegan, J.Electron.Mater.32 335(2003). 37. X. A. Cao, S. J. Pearton, G. Dang, A. P. Zh ang, F. Ren, and J. M. Van Hove, Appl. Phys. Lett. 75, 4130 (1999). 38. H. Morkoc, S. Strite, G. B. Gao, M. E. Lin, B. Svedlov, and M. Burns, J. Appl. Phys. 76 (2), 1363 (1994). 39. S. Nakamura and G. Fasol, The Blue Laser Diode ~Springe r, Heidelberg,1997. 40. S. M. Sze, Physics of Semiconductor De vices, John Wiley & Sons, New York (1981). 41. http://www.onr.navy.mil/sci_tech/informa tion/312_electronics/ncsr/properties.asp 42. D. W. Palmer, www.semiconductors.co.uk, 2004a. 43. O. Ambacher, J. Phys. D: Appl. Phys. 31, 2653 (1998). 44. H. Omiya, F. A. Ponce, H. Marui, S. Tanaka, and T. Mukai ,Appl. Phys. Lett. 85, 6143 (2004). 45. M. R. Park and W. A. Anderson, Mat. Res. Soc. Symp. Proc. 680E, E6.5.1 (2001). 46. C. H Lin, D. L. Hibbard, A. Au, H. P. Lee, Z. J. Dong, F. J. Szalkowski, J. Chen, and C. Chen, Mat. Res. Soc. Symp. 639, G4.8.1 (2001). 47. T. Gessmann, Y.-L. Li, E.L. Waldron, J.W. Graff, E.F. Schubert, J.K. Sheu, Appl. Phys. Lett. 80, 986 (2002). 48. J. K. Kim, J.-L. Lee, J. W. Lee, H. E. Shin, Y. J. Park, and T. Kim, Appl. Phys. Lett. 73, 2953 (1998). 49. V. R. Reddy, Mater. Chem.Phys. 93 286 (2005). 50. D. Selvanathan, L. Zhou, V. Kumar, I. Adesida and N. Finnegan, J.Electron. Mater. 32, 335 (2003). 51. G. -T. Chen, C. -C. Pan, C. -S.Fang, T. -C. Huang, J. -I.Chyi, M. -N. Chang, S. -B. Huang and J. -T.Hsu, Appl. Phys. Lett. 85, 2797 (2004). 52. K. A. Rickert, A. B. Ellis, J. K. Kim, J. L. Lee, F. J. Himpsel, F. Dwikusuma, and T. F. Kuech, J. Appl. Phys. 92, 6671 (2002). 53. Q. Z. Liu and S. S. Lau, Solid -State Electron. 42, 667 (1998). 54. M. W. Wang, J. O. McCaldin, J. F. Swenberg T. C. McGill, and R. J. Hauenstein, Appl. Phys. Lett. 66, 1974 (1995)

PAGE 140

140 55. T. Mori, T. Kozawa, T. Ohwaki, Y. Taga, S. N aagai, S. Yamasaki, S. Asami, N. Shibata, and M. Koike, Appl. Phys. Lett. 69, 3537 (1996). 56. L. S. Yu, D. Qiao, L. Jia, S. S. Lau, Y. Qi and K. M. Lau, Appl. Phys. Lett. 79, 4536 (2001). 57. L. S. Yu, L. Jia, D. Qiao, S.S. Lau, J. Li, J.Y. Lin, and H.X. Jiang, IEEE Trans. Electron. Devices 50, 292 (2003). 58. X. A. Cao, S. J. Pearton, G. Dang, A. P. Zh ang, F. Ren, and J. M. Van Hove, Appl. Phys. Lett. 75, 4130 (1999). 59. P. J. Hartlieb, A. Roskowski, R. F. Davis, W. Platow, and R. J. Nemanich, J. Appl. Phys. 91, 732 (2002). 60. D. L. Hibbard, R. W. Chuang, Y. S. Zhao, C. L. Jensen, H. P. Lee, Z. J. Dong, R. Shih, and M. Bremser, J. Electron. Mater. 29, 291 (2000). 61. K. Shiojima, T. Sugahara, and S. Sakai, Appl. Phys. Lett. 74, 1936 (1999). 62. Y. J. Lin, Appl. Phys. Lett. 86, 122109 (2005). 63. K. Shiojima, T. Sugahara, and S. Sakai, Appl. Phys. Lett. 77, 4353 (2000). 64. T. Sawada, Y. Ito, N. Kimura, K. Imai, K. Suzuki, and S. Sakai, Appl. Surf. Sci. 190, 326 (2002). 65. C. K. Tan, A. Abdul Aziz, and F. K. Yam, Appl. Surf. Sci. 252, 5930 (2006). 66. Y. J. Lin and C. W. Lin, J. Electron. Mat. 33, 1036 (2004). 67. T. B. Massalski (Ed.), Binary alloy phase diagrams (ASM International, Materials Park, Ohio, 1990). 68. X. A. Cao, S. J. Pearton, F. Ren, and J. R. Lothian, Appl. Phys. Lett 73, 942 (1998). 69. X. A. Cao, F. Ren, J. R. Lothian, S. J. Pearton, C. R. Abernathy, J. C. Zolper, M. W. Cole, A. Zeitouny, M. Eizenberg, and R. J. Shul, MRS Internet Journal of Nitride Semiconductor Research. 4 (1999). 70. Handbook of the Physiochemical Properties of the Elements, edited by G.V. Samsonov, IFIPlenum, New York, (1968). 71. M. W. Cole, F. Ren, and S. J. Pearton, J. Electrochem. Soc. 144, L275 (1997). 72. A. Zeitouny, M. Eizenberg, S. J. Pearton, and F. Ren, J. Appl. Phys. 88, 2048 (2000). 73. J. Kim, F. Ren, A. G. Baca, and S. J. Pearton, Appl. Phys. Lett. 82, 3263 (2003).

PAGE 141

141 74. B. Luo, F. Ren, R. C. Fitch, J. K. Gillespie, T. Jenkins, J. Sewell, D. Via, A. Crespo, A. G. Baca, R. D. Briggs, D. Gotthold, R. Birkhahn, B. Peres, and S. J. Pearton, Appl. Phys. Lett. 82, 3910 (2003). 75. R. Khanna, S. J. Pearton, F. Ren, and I. Kravchenko, J. Electrochem. Soc. 11, G804 (2005). 76. R. Khanna, S. J. Pearton, F. Ren, I. Kravchenko C.J. Kao and G.C. Chi, Appl. Phys. Lett. 87, 052110 (2005). 77. R. Khanna, S. J. Pearton, F. Ren, and I. Kravchenko, Appl. Surf. Sci. 252, 5814 (2006). 78. T. N. Oder, P. Martin, J. Y. Lin, H. X. Ji ang, J. R. Williams, and T. Isaacs-Smith, Appl. Phys. Lett. 88, 183505 (2006). 79. R. T. Tung, Mater. Sci. Rep. R35, 1 (2001). 80. L. Stafford, L. F. Voss, S. J. Pearton, J. -J. Chen, and F. Ren, submitted. 81. Michael Shur, Physics of Semiconductor Devi ces (Prentice-Hall, Engelwood Cliffs, NJ, 1990). 82. C. Merz, M. Kunzer, U. Kaufmann, I. Akasaki, and H. Amano, Semicond. Sci. Technol. 11, 712 (1996). 83. H. Nakayama, P. Hacke, M.R.H. Khan, T. Dtec hprohm, K. Hiramatsu, and N. Sawaki, Jpn. J. Appl. Phys. 35, L282 (1996). 84. J. R. Waldrop and R. W. Grant, Appl. Phys. Lett. 68, 2879 (1996). 85. J. S. Kwak, O. H. Nam, and Y. Pa rk, Appl. Phys. Lett. 80, 3554 (2002). 86. J. S. Kwak, O. H. Nam, and Y. Park, J. Appl. Phys. 95, 5917 (2004). 87. M.W. Cole, F. Ren, and S. J. Pearton, J. Electrochem. Soc. 144, 275 (1997). 88. K. Shiojima, S. Sugitani, and S. Sakai, Appl. Surf. Sci. 190, 318 (2002). 89. T. Okumura, Jpn. J. Appl. Phys. 24, L437 (1985). 90. M. Passlack, N. E. J. Hunt, E. F. Schubert, G. J. Zydzik, M. Hong, J. P. Mannaerts, R. L. Opila, and R. J. Fischer, Appl. Phys. Lett. 64, 2715 (1994). 91. M. Lenzlinger and E. H. Snow, J. Appl. Phys. 40, 278 (1969). 92. Z. Weinberg, J. Appl. Phys. 53, 5052 (1982). 93. H. Zhang, E. J. Miller, and E. T. Yu, J. Appl. Phys. 99, 023703 (2006).

PAGE 142

142 94. Introduction to Nitride Semiconductor Blue Lase rs and Light-Emitting Diodes, edited by S. Nakamura and S.F. Chichibu (Taylor and Francis, London, 2000). 95. M. Meneghini, L. -R. Trevisanello, U. Zehnder, T. Zahner, U. Strauss, G. Meneghesso and E. Zanoni, IEEE Trans. Electron. Dev. 53, 2981 (2006). 96. S. Bychikhin, D. Pogany, L. K. J. Vandamme, G. Meneghesso, and E. Zanoni, J. Appl. Phys. 97, 123714 (2005). 97. M. Pavesi, F. Rossi, and E. Zanoni, Semicond. Sci. Technol. 21, 138 (2006). 98. F. Rossi, M. Pavesi, M. Meneghini, G. Salviati M. Manfredi, G. Meneghesso, A. Castaldini, A. Cavallini, L. Rigutti, U. Strass, U. Zehnder, and E. Zanoni, J. Appl. Phys. 99, 053104 (2006). 99. X. A. Cao and S.D. Arthur, Appl. Phys. Lett. 85, 3971 (2004). 100. X. A. Cao, S. F. LeBoeuf, M. P. D'Evelyn, S. D. Arthur, J. Kretchmer, C. H. Yan, and Z. H. Yang, Appl. Phys. Lett. 84, 4313 (2004). 101. X. A. Cao, J. M. Teetsov, M. P. D'Evelyn, D. W. Merfeld, and C. H. Yan, Appl. Phys. Lett. 85, 7 (2004). 102. S. Nakamura, M. Senoh, S. Nagahama, N. Iwasa, T. Yamada, T. Matsushita, Y. Sugimoto and H. Kikyou, Jpn. J. Appl. Phys., Part 2 36, L1059 (1997). 103. C. Y. Hsu, W. H. Lan and Y. S. Wu, Appl. Phys. Lett. 83, 2447 (2003). 104. A. P. Zhang, B. Luo, J. W. Johnson, F. Ren, J. Han and S. J. Pearton, Appl. Phys. Lett. 79, 3636 (2001). 105. T. Maeda, Y. Koide and M. Murakam i, Appl. Phys. Lett. 75, 4145 (1999). 106. L. -C. Chen, J. -K. Ho, C. -S. Jong, C. C. Chiu, K. -K. Shih, F. -R. Chen, J. -J. Kai and L. Chang, Appl. Phys. Lett. 76, 3703 (2000) 107. D. Qiao, L. S. Yu, S. S. Lau, J. Y. Lin, H. X. Jiang and T. E. Haynes, J. Appl. Phys. 88, 4196 (2000). 108. H. S. Venugopalan, S. E. Mohney, J. M. DeLucca, B. P. Luther and G. E. Bul man, J. Vac. Sci. Technol. A 16, 607 (1998). 109. C. -F. Chu, C. C. Yu, Y. K. Wang, J. Y. Tsai, F. I. Lai and S. C. Wang, Appl. Phys. Lett. 77, 3423 (2000). 110. L. F. Voss, R. Khanna, S. J. Pearton, F. Ren and I. I. Kravchenko, Appl. Phys. Lett. 88, 012104 (2006).

PAGE 143

143 111. L. F. Voss, R. Khanna, S. J. Pearton, F. Ren, and I. I. Kravchenko, Appl. Surf. Sci. 253, 1255 (2006). 112. W. Zagozdzon-Wosik, D. Radhakrishnan, C. Darne, I. Rusakova, P. van der Heide, Z.-H. Zhang, J. Bennett, L. Trombetta and P. Majh i, Rev. Adv. Mater. Sci. 8, 185 (2004). 113. Y. D. Qi, H. Liang, D. Wang, Z. D. Lu, W. Tang and K. M. Lau, Appl. Phys. Lett. 86, 101903 (2005). 114. X. Guo and E. F. Schubert, Appl. Phys. Lett. 78, 3337 (2001). 115. A. Chitnis, J. Sun, V. Mandavilli, R. Pachi pulusu, S. Wu, M. Gaevski, V.Adivarahan, J. P. Zhang, M. Asif Khan, A. Sarua and M. Kuball, Appl. Phys. Lett. 81, 3491 (2002). 116. P. G. Eliseev, P. Perlin, J. Furioli, P. Sartor i, J. Mu, and M. Osinski, J. Electron. Mater. 26, 311 (1997). 117. G. Franssen, E. Litwin-Staszewska, R. Piotrz kowski, T. Suski, and P. Perlin, J. Appl. Phys. 94, 6122 (2003). 118. J. M. Shah, Y. L. Li, T. Gessmann and E. F. Schubert, J. Appl. Phys. 94, 2623 (2003). 119. S. H. Wang, S. E. Mohney, and R. Birkhahn, J. Appl. Phys. 91, 3711 (2002).

PAGE 144

144 BIOGRAPHICAL SKETCH Lars Voss was born in Pittsburgh, Pennsylvania in 1982. He spent m ost of his life in Erie, Pennsylvania, until graduating from McDowell Senior High School in 2000. He then enrolled at The Pennsylvania State University where he ea rned a Bachelor of Science in Engineering Science, with a minor in electronic and photonic ma terials, in May 2004. While there, he had the opportunity to work for Prof. Paul Koch in plas tics engineering technology at Penn State Erie during the summer and with Prof. P.M. Lenaha n in his Semiconductor Sp ectroscopy Laboratory. After his undergraduate education, Lars enrolled at th e University of Florida in the Materials Science and Engineering Departme nt and began his graduate study under Prof. Stephen J. Pearton. During the summers of 200 6 and 2007, he worked as an intern at Sandia National Laboratories under the direction of Drs. Albert G. Baca and Randy J. Shul.