Nitride semiconductors and

their applications

Part II: Nitride semiconductors

 Nitride papers published
1000
# papers published

100

10

1980 1985 1990 1995 2000 2005

Year
 Nitride-based semiconductors

• A III-V semiconductor in which N is one of

the elements. Examples include AlN, GaN,
InN, and alloys such as AlxGa1-xN.

• These materials have high melting points

(strong bond with N) and a wide span of
bandgaps (from 0.7 eV for InN to 3.4 eV for
GaN to 6.2 eV for AlN).
 Applications
• High-temperature, high-power electronics
• Ultraviolet (UV) radiation detectors
– feedback systems in furnaces and engines
– solar-blind missile early warning systems
– astronomical applications
• LEDs and LDs
 Blue LEDs and LDs
• Full color day-visible displays
• Energy efficient lighting
– Traffic lights, replace every 6 months, LEDs,
every 5-10 years (60,000 hours)
– Consumer lighting applications.
• Higher storage density on optical media
(>50 GB on a single DVD)
• Faster on-time
• Fluorescent photosensitizers accumulate
preferentially in cancerous cells.
 QuickTime™ and a
GIF decompressor
are needed to see this picture.
 Short history
1971 GaN LED demonstrated (Pankove)
1986 “High” quality GaN grown (Akasaki)
1988 P-type GaN grown (Akasaki); Nakamura starts work
on GaN
1990 Two-flow MOCVD system developed (Nakamura)
1991 High quality p-type GaN grown; first pn-junction
GaN LED created
1992 ZnSe-CdZnSe blue laser developed (3M)
1993 Commercial blue GaN LEDs introduced (Nichia)
1996 Room temperature nitride LDs developed (Nichia)
1999 Commercial nitride LDs introduced (Nichia)
2004 SONY markets blue laser DVD writers (23 GB/layer)
 Nitride problems

1. Inability to grow good quality crystals

2. Inability to grow p-type crystals
 Crystal quality
• Problem: No lattice matched substrates, high
growth temperature results in convection currents
– Sapphire is closest but is 15% off.
– SiC is too expensive
– MOCVD growth too fast for good control (few mm/min)
• Solution: Buffer layers, new growth system
– First grow GaN or AlN buffer layer
– Two-flow MOCVD system
– Still many many dislocations in material (1010 cm-2) but
dislocations don’t matter?
Two-flow MOCVD

Nakamura, Harada, and Seno, J. Appl. Phys 58, 2021 (1991)

 Buffer layer
Dislocation

High quality GaN

Low quality GaN (0.2 mm)

AlN or GaN buffer layer

(0.05 mm)

Substrate
 p-type GaN

• Problem: No one could dope GaN p-type

• Solution:
– At first, LEEBI (Low Energy Electron Beam
Irradiation)
– Later, annealing at 700°C in non H-containing gas
– Hydrogen was passivating the acceptors!
 Factors leading to success

• Small bureaucracy (N. Ogawa and S.

Nakamura)
• A 3.3 M$ USD gamble (1.5% of annual
sales)
• Large companies tend to be conservative,
both in funds and in research outlook
 Current research on nitrides

• Fundamental physics
• Improving crystal quality (still very poor)
• Ultraviolet lasers
• Lattice matching with quaternary alloys
(AlGaInN)
• Nitride heterostructures and accompanying
applications
 Nitride heterostructures

AlGaN GaN

Conduction band
Energy

AlGaN GaN
Valence band
 Heterostructure usefulness

Charge carriers are

spatially separated from
the (now) ionized impurity
atoms, leading to higher
carrier mobilities.

Electrons form a 2-
Dimensional Electron Gas
(2-DEG).
 Research questions

• Even undoped, carrier densities in AlGaN/GaN

heterostructures is 10 to 100 times larger than
those in similar (AlGaAs/GaAs) systems.
What is the source of these carriers?

• Carrier mobilities in AlGaN/GaN heterostructures

are 10 to 100 times lower than in the
AlGaAs/GaAs system.
What are the principle mechanisms limiting the
mobility?
 Origin of carriers

Current theory: surface donor defects on AlGaN

Concentration of surface defects and transfer of electrons to

GaN well is enhanced by strain-induced electric field.
 Pseudomorphic growth

AlGaN

GaN

lattice matched non-lattice matched

AlGaN/GaN interface
• GaN and AlN have ~2.5% lattice mismatch
• Grown on polar c-axis
• Spontaneous and induced piezoelectric fields are present
 Band structure
CB
• When AlGaN barrier is thick
enough to pull defect level above
bottom of GaN well, electrons
begin to transfer to well.
defect
level

• Formation energy of donor defects

is reduced because electrons can
drop to lower energy level by
transferring to GaN well

GaN AlGaN
 Electron transfer
14
T=4K
2-DEG density (10 cm )

Al fraction
-2

12 25% Al
25%
12

10
15%
8 5%
6 15% Al

4
5% Al GaN AlGaN
2

0
0 100 200 300 400 500
)
Barrier width (
 Comparison with experiment
20
14

cm -2)
cm -2)

12
15

12
12

10

2 DEG density (10

2 DEG density (10

10 8
6
5 4
31 nm barrier 27% Al
2
T = 13K T = 13 K
0 0
5 10 15 20 25 30 35 0 100 200 300 400 500
Barrier Al composition (%) )
Barrier thickness (

Smorchkova et al., J. Appl. Phys 86, 4520 (1999)

 Transport properties
Semiconductor with no applied field:
• Electrons move randomly with an average velocity
of zero
• Mean time between electron-electron collisions is t.

With an applied field:

• Electrons have an acceleration of a = eE/m*
• Average velocity of electrons is at, parallel to field. vavg
m=
E
The mobility is a measure of how easily charge carriers
respond to an applied electric field.
Limiting factors

Scattering mechanisms
• Coulomb fields
• Phonons
• Alloy disorder
 Coulomb scattering

Electrons are affected by the long-range Coulomb

fields of randomly distributed ionized donor atoms.
• Thicker barriers move ionized surface donors
further away from carriers.
• Large 2-DEG densities screen the effect of these
Coulomb fields.
 Phonon scattering
Phonons are lattice vibrations in a crystal.
Acoustic phonons
• Both types of atoms move “in-phase”
• Low energy vibrations

Optical phonons
• Atoms of different types move “out-of-phase”
• High energy vibration
 Phonon scattering

• Phonons scatter carriers by creating small

fluctuating dipoles between atoms
(piezoelectric mode).

• Phonons scatter carriers by disturbing the

periodicity of the crystal lattice
(deformation potential mode).
 Alloy disorder

• Electron wavefunction
penetrates into AlGaN
barrier.
• Al and Ga atoms are
distributed randomly in
AlGaN
• Randomly varying potential
scatters electrons.
 2-DEG mobilities
106
2-DEG mobility (cm2/Vs) Phonon
Coulomb

105
Alloy

104
total 4K
Al 0.15Ga0.85N/GaN
0 100 200 300 400 500
Barrier thickness (Å)
 Improving the mobility
Strategies:
• Reduce 2-DEG density
– Smaller Al alloy fractions
– Thinner barriers
(Highest mobility heterostructures have Al fractions
of ~10% and barrier thicknesses of ~130 Å)

• Reduce alloy disorder scattering

– AlN spacer
 AlN spacer

GaN AlN AlGaN

 Conclusions
• Nitride-based semiconductors are a promising field
for a wide variety of new technological applications.

• 2-DEG mobilities are limited by two factors:

– Coulomb scattering (N < 2 x 1012 cm-2)
– Alloy disorder scattering (N > 4 x 1012 cm-2)
– We predict maximum low temp mobilities of 105 cm2/V s
without a AlN spacer.

• Using a AlN spacer seems a promising way to

improve the conductivity of nitride heterostructures.
 Subband structure
Confining potential results in
quantized energy levels.
4 4
2    2
 
2
 2   V z  E
2m *

Trial wavefunction

b 2 3 / 2 bz / 2
 0   x,y z e
6
Energy (meV) Subband structure

100

E1
EF

E
0
10 11
10 1012 1013
2-DEG density (cm -2)
 Comparison with experiment
3 104
2-DEG mobility (cm /Vs)
2

104
8 103
6 103

13 K
4 103
Al 0.05Ga0.95N/GaN
0 100 200 300 400 500
Barrier thickness (Å)
 Comparison with experiment
4
3 10
T = 13 K
2-DEG mobility (cm /Vs)

31 nm barrier

2-DEG mobility (cm /Vs)

T = 13 K 2 104 27% Al
2

2
4
2 10

4
4 1 10
1 10

0
0 10 0 100
5 10 15 20 25 30 35 0 100 200 300 400 500
Barrier Al composition (%) )
Barrier thickness (

Smorchkova et al., J. Appl. Phys 86, 4520 (1999)

 Comparison with experiment
4
3 10
T = 13 K
2-DEG mobility (cm /Vs)

31 nm barrier

2-DEG mobility (cm /Vs)

T = 13 K 2 104 27% Al
2

2
4
2 10

4
4 1 10
1 10

0
0 10 0 100
5 10 15 20 25 30 35 0 100 200 300 400 500
Barrier Al composition (%) )
Barrier thickness (

Smorchkova et al., J. Appl. Phys 86, 4520 (1999)