Escolar Documentos
Profissional Documentos
Cultura Documentos
PHYSICAL CONSTANTS
Quantity
Symbol
Magnitude
1
Angstrom Unit
Avogadro's Number Bohr Radius Boltzmann's Constant Electron Charge
Electron Volt
A = 10- 4 fan x 10
23
lO" 8
cm =
10" 10
NAyo
aB
6.025
molecules/g-mole
0.53A
1.380
k
q
x 10" 23 joule/K
1.602 x 10- 19 x 10- 31 kg
= 8.62
x 10" 5 eV/K
kcal
19 1.602 x 10coul
1
eV
eV =
joule
= 23.1
9 ^
m
jttm
9.108
Micron
Permittivity of Free Space
10- 6
8.854
m=10- 4 cm
x 10- 14 farad/cm
~ 34 / 10
1
o
H<o
\
1
(4tt
x 10~ 9 )
h
h
6.625
hlltr
x 10
joule-sec
=
x
1.054 10 4
1.033
(torr)
= 300K)
kT/q
c
0.0259 volt
2.998
x 10
10
cm/sec
eV
1.239 fim
S.
M. SZE
of the Technical Staff
Member
Bell
Murray
New
Jersey
WILEY-INTERSCIENCE
Division of John
New York
London
Sydney
Toronto
PRESTON
PGLYT< HM1C
57825
m
G^n'Mi 33
Copyright
No part of this book may be reproduced by any means, nor transmitted, nor translated into a machine language without the written permission of the publisher.
All rights reserved.
10
y^^"^
Number: 69-16132
To
My
-:k->'
v-*r-~r
Preface
This book
is
concerned
specifically
for first-year graduate students in electrical engineering or applied physics and as a reference for scientists actively involved in solid-state device re-
search and development. It is assumed that the reader has already acquired an introductory understanding of solid-state physics and transistor theory such as is given in the standard texts Solid-State Physics by Dekker, Introduction to Solid-State Physics by Kittel (Wiley, 1966), Physics of Semiconductors by Moll, Introduction to Semiconductor Physics by Adler et al.
(Wiley,
Semiconductors by Smith, and Physics and Technology of Semiconductor Devices by Grove (Wiley, 1967). With this as a basis, the present book elaborates on device theory in more detail. semiconductor device is defined herein as a unit which consists, partially
1964),
or wholly, of semiconducting materials and which can perform useful functions in electronic apparatus and solid-state research. Since the invention of the transistor in 1948 the number and variety of semiconductor devices have
increased tremendously as advanced technology, new materials, and broadened comprehension have been applied to the creation of new devices. In a
book of this
length
it is
An
not possible to give a complete account of all of them. made to include most of the important de-
to present
them
in a unified
It is
hoped
that this
book
can serve as a basis for the understanding of other devices not included here and perhaps not even conceived of at the present time.
viii
Preface
This text began as a set of lecture notes for an out-of-hours course in " Semiconductor Device Physics " given at Bell Laboratories and later presented as a graduate course in "Selected Topics on Solid-State Devices" at
properties the principal source
For numerical data on semiconductor was the Data Sheets compiled by the Electronic Properties Information Center of Hughes Aircraft Company. More than two thousand papers on semiconductor devices, cataloged at the Murray Hill Library of Bell Laboratories, have provided background. For each chapter a
Stevens Institute of Technology.
brief historical review, as well as a general outline,
is
given in
its
introduction.
The physics of the device and its mathematical formulation are then presented in subsequent sections and are generally arranged in logical sequence without heavy reliance on the original papers. The literature and many illustrations used in the book are acknowledged in the reference lists at the close of each chapter. In Chapter 2 extensive data have been compiled and presented for the three most important semiconductors: germanium, silicon, and gallium
arsenide. These data are used in analyzing device characteristics throughout
this
book. In Chapters 3 through 14 each considers a specific device or a group is presented in such a way that it is self;
is
interested in bulk-effect devices, he can refer directly to the last chapter with-
out consulting any of those between Chapters 3 and 13. During the course of the writing I have been deeply grateful to
colleagues at Bell Laboratories
;
many of my
book could not have been written. I wish to express, in particular, my gratitude to Dr. R. M. Ryder for his constant encouragement and constructive criticism and to Drs. W. S. Boyle and F. M. Smits for providing a stimulating and challenging environment in which I have been welcomed, inspired, and abundantly assisted. I am
without their help
this
J.
M. Andrews, H.
J.
Boll, J. R. Brews, C.
Y. Chang,
DeLoach,
J.
J.
Gummel,
D. Kahng, S. Knight, M. Kuhn, H. S. Lee, T. P. Lee, M. P. Lepselter, W. T. Lynch, H. Melchior, T. Misawa, E. H. Nicollian, T. Paoli, H. C. Poon, R. J. Powell, J. E. Ripper, W. Rosenzweig, D. L. Scharfetter, L. S. Senhouse, J. Sevick, H. Thim, and T. H. Zachos for their helpful suggestions and discussions on one or more chapters of the book. I am especially indebted to Mrs. E. S. Blair for her lucid technical editing of the entire manuscript. Thanks are due also to Professor G. J. Herskowitz of Stevens Institute of Technology, who provided the opportunity for me to teach the course on which this text is based, and to Mrs. M. Neuberger of the Electronic Properties Information Center, Hughes Aircraft Company, who provided the up-to-date compilation of data sheets and kindly checked the tables used in Chapter 2. It is a pleasure to acknowledge Mr. G. P. Carey and Mr. A. Loya, who did most of the literature searching on semiconductor
F. Harper, E. Iwersen,
Preface
devices,
ix
and my students at Stevens Institute, in particular Mr. P. P. Bohn and Mr. H. A. Kruegle, who helped to correct many errors in the original class notes. I wish to thank Miss D. A. Williams who typed the final manuscript, Mrs. J. Hendricks, Miss J. McCarthy, and Mrs. E. H. Nevitt who typed various sections of the book in its revision stage, and the members of the Murray Hill Drafting Department of Bell Laboratories who furnished the more than five hundred technical drawings used in this book. The support of C. Y. Tung's Chair Professorship grant and the hospitality of the National Chiao Tung University provided the environment in which to read the proofs
of the book. Finally,
her assistance in
I
my
many
S.
M. Sze
Contents
Chapter
1.
Introduction
General Outline
of Semiconductor Devices
2. Classification 3. Specific
Remarks
PART
SEMICONDUCTOR PHYSICS
Physics and Properties of Semiconductors
Chapter 2
A
1.
Resume
11
11
Introduction
Crystal Structure
2.
12
17
3.
Energy Bands
Concentration at Thermal Equilibrium
Carrier Transport Phenomena Phonon Spectra and Optical, Thermal, and High-Field
Properties of Semiconductors
4. Carrier
5.
25 38
50
6.
7.
....
65
PART
II
p-n
3
JUNCTION DEVICES
77
77
.
.
Chapter
1.
Introduction
Basic Device Technology
.
2.
78
xi
xii
Contents
3. 4. 5.
6. 7.
84 96
109 127
131
8.
140
Chapter 4
1.
150
150
151
Introduction
of High Doping
2. Effects
3. 4. 5.
156
169
Due
to Effects of Doping,
....
172
190
193
Backward Diode
5
Chapter
(IMPATT
1.
Introduction
Static Characteristics
2. 3.
Basic
Dynamic
Characteristics
4. 5. 6. 7.
Noise Experiments
Chapter 6
1.
Junction Transistors
261
Introduction
Static Characteristics
261
2.
3.
4.
5.
6.
Unijunction Transistor
Chapter 7
1.
Introduction
2. 3.
Shockley Diode and Semiconductor-Controlled Rectifier Junction Field-Effect Transistor and Current Limiter
....
320 340
Contents
xiii
PART
III
INTERFACE
AND
THIN-FILM DEVICES
363
363
Chapter 8
1.
Metal-Semiconductor Devices
Introduction
2. 3.
Schottky Effect
364
368 378 393
4.
5. 6.
7.
8.
Energy Band Relation at Metal-Semiconductor Contact ... Current Transport Theory in Schottky Barriers Measurement of Schottky Barrier Height Clamped Transistor, Schottky Gate FET, and Metal-Semicondutor IMPATT Diode Mott Barrier, Point-Contact Rectifier, and Ohmic Contact Space-Charge-Limited Diode
. .
Chapter 9
1.
Introduction
2. Ideal 3.
Metal-Insulator-Semiconductor (MIS) Diode Surface States, Surface Charges, and Space Charges
444
4. Effects
of Metal
Work
MIS
5.
Characteristics
Electroluminescent
6.
MIS Diodes
Chapter 10
1.
Field Effects
505 505
Introduction
2.
3.
4. 5. 6. 7. 8.
506 512
515
524
.
IGFET IGFET
A Memory Device
Semiconductor Devices
Chapter
1.
11
Thin-Film Devices
567
Introduction
Insulated-Gate Thin-Film Transistors (TFT)
567
568
2.
xiv
3.
4.
Contents
Hot-Electron Transistors
Metal-Insulator-Metal Structure
587 614
PART
[f-
IV
OPTOELECTRONIC DEVICES
Optoelectronic Devices
625
Chapter 12
1.
Introduction
2. 3.
Electroluminescent Devices
Solar Cell
625 626
4.
Photodetectors
640 653
Chapter 13
1.
Semiconductor Lasers
687
687
688
Introduction
2.
3.
699
723
4.
Operation
5.
725
PART V
Chapter 14
1.
BULK-EFFECT DEVICES
Bulk-Effect Devices
731
Introduction
731
2.
3. 4.
5.
Bulk Differential Negative Resistance Ridley- Watkins-Hilsum (RWH) Mechanism Gunn Oscillator and Various Modes of Operation Associated Bulk-Effect Devices
732
741
749
778
Author Index
Subject Index
789
799
SEMICONDUCTOR DEVICES
SPECIFIC
REMARKS
Introduction
GENERAL OUTLINE
The contents of
Part
1
:
this
Part 2
Part 3
Part 4 Part 5
and
thin-film devices,
Part 1 in Chapter 2 is intended as a summary of materials properties, to be used throughout the book as a basis for understanding and calculating
device
characteristics.
Carrier distribution
emphasis on the three most important materials: Ge, Si, and GaAs. A compilation of the most recent and most accurate information on these semiconductors is given in the tables and illustrations of Chapter 2. Part 2, Chapters 3 through 7, treats of devices with one or more interacting p-n junctions. The classic moderate-field p-n junction of Chapter 3 is basic
briefly surveyed, with
both for
junction
and as a component of more involved devices. When the doped heavily enough on both sides so that the field becomes high enough for tunneling, one obtains the new features of tunnel diode behavior (Chapter 4). When the junction is operated in avalanche breakdown, under proper conditions one obtains an IMPATT diode which can generate microitself
is
Introduction
5).
Chapter 6 treats of the junction transistor, that two closely-coupled junctions, which is the single is, most important semiconductor device. Other junction devices are in Chapter 7; among these are the p-n-p-n triple-junction devices, and the junction field-effect transistor which utilizes the junction field to control a current flow which is parallel to the junction rather than perpendicular to it. Part 3, Chapters 8 through 11, deals with interfaces, or surfaces between semiconductors and other materials. Interfaces with metals, in particular the Schottky barrier, are in Chapter 8. The Schottky diode behavior is electrically similar to a one-sided abrupt p-n junction, and yet it can be operated as a
Metal-insulator-semi-
conductor devices and the related surface physics of the insulator-semiconductor interface are considered in Chapters 9 and 10. This knowledge of "surface states" is important not only because of the devices themselves but
and reliability of all other semiconductor devices. Chapter 11 considers some thin-film and hot-electron devices which also belong to the interface-device family. Part 4, Chapters 12 and 13, considers optoelectronic devices which can
also because of the relevance to stability
detect, generate,
and convert
We
and the solar cell in Chapter 12. Chapter 13 is devoted to one of the most important optoelectronic devices: the semiconductor laser. Part 5, Chapter 14, considers some so-called "bulk property" devices, primarily those concerned with the intervalley-transfer mechanism (Gunn oscillator and LSA oscillator). These devices do not depend primarily on
shall consider various photodetectors
fields,
so
prime
interest in
Chapter
14.
and physical
principles. It
independent of the other chapters, so that the instructor or the reader can select or rearrange the device chapters in accordance with his own
less
be more or
schedule.
A remark on notation
devices.
it is
necessary
more than once, with different meanings for different For example, the symbol a is used as the common-base current gain
and as the electron impact ionization
is
an
IMPATT
to find
Classification of
Semiconductor Devices
symbol
is
is
defined the
first
time
it
appears.
Many
this
have the same or similar meanings consistently throughout summarized in Table 1.1 for convenient reference.
In the previous section we have classified semiconductor devices in accordance with material combinations (such as the semiconductor alone, or a combination of metal, semiconductor, and insulator) and material properties
(such as the optical property, or the intervalley transfer property). This
classification
is
used because
it
TABLE
LIST
1.1
OF BASIC SYMBOLS
Unit
Symbol
a
31
Name
lattice spacing magnetic induction velocity of light in free space
A
Weber/m 2
cm/sec
farad
C
2>
capacitance
electric
displacement
coul/cm 2
D
E Ec EF Eg
Ey
<S
diffusion coefficient
cm 2 /sec
eV eV eV eV eV
volt/cm
volt/cm volt/cm
energy
bottom of conduction band Fermi energy level energy gap band top of valence band
electric field
critical field
$c
m f
F{E)
h
hv
I
Ic
maximum
frequency
field
Hz(cps)
joule-sec
photon energy
current
collector current
eV
amp amp
amp/cm 2 amp/cm 2
joule/K
J
Jt
current density
kT L
eV
cm
mass mass
kg kg
or
nm
m
m*
TABLE
Symbol
n
I.I
(Cont.)
Name
refractive index
Unit
n
Hi
N NA Nc ND Nv
P
doping concentration
acceptor impurity density
effective density of states in
conduction band
band
cm -3 cm -3 cm -3 cm -3 cm -3 cm -3 cm -3 cm -3
dyne/cm 2 coulomb
charges/cm 2
P
<?
pressure
Qss
R
t
resistance
ohm
sec
time
absolute temperature
carrier velocity
T
V
Vsl
K
cm/sec cm/sec cm/sec
volts volts volts volts
scattering-limited velocity
Vi*
V vbl
Veb
emitter-base voltage
Vb
breakdown voltage
thickness
W wB
X
base thickness
x-direction
differential
cm cm
or
fira
or /u.m
operator
VT
o
Ss
Ei
temperature gradient
free space permittivity
K/cm
farad/cm farad/cm farad/cm
semiconductor permittivity
insulator permittivity
e s /e
or ,/e
dielectric constant
T
e
lifetime or
decay time
second
radian
angle
A
V
1*0
wavelength
frequency of light free-space permeability
electron mobility
(Am or
Hz
henry/cm
Hn
[Ap
hole mobility
resistivity
cm 2 /V-sec cm 2 /V-sec
ohm-cm
volts
P
<!>
<f>Bn
<}>Bp
</>,
Schottky barrier height on n -type semiconductor Schottky barrier height on p -type semiconductor
metal work function
angular frequency
(2-rrf
CO
or 2itv)
ohm
Hz ohm
Specific
Remarks
posed by Angello, 1
to classify
this section. This system can be used and future semiconductor devices, and can serve to organize creative thinking. The system starts by listing a complete set of semiconductor properties and a complete set of external influences (such as applied voltage) which can modify semiconductor attributes. Devices are then
be presented in
the present
and so on.
the
We
shall
start
and so on.
departure from bulk-effect devices, and the external influences will be applied
and so on. Finally, we shall consider multiple junction (or and the external influences will be applied accordingly. Table 1.2 presents some of the bulk-effect devices where the symbols for external influences are defined in Table 1.1. Table 1 .3 presents some important single-junction (or interface) devices. Table 1.4 presents some important
singly, in pairs,
interface) devices,
lists
exhaust
all
of effects and
in-
fluences, other
semiconductor devices
all
may
be classified similarly.
Table
1.5
shows
They
7,
We
may
include devices
from
all
three groups.
For example,
Chapter
we
shall consider the current limiter (bulk-effect), the field-effect diode (single-
junction),
they
SPECIFIC
It
REMARKS
may
book.
this
and the semiconductor-device and so fast-changing that today's concepts field in particular are so dynamic may be obsolete tomorrow. Remember these "famous last words" 2 ? " They'll never replace the smoke signal as the fastest means of communication." Chief White Cloud, "The telegraph is the ultimate in fast communication." Engineer, 1850's, " With the vacuum tube, we've reached the zenith in communication potenFirst of
Introduction
TABLE
1.2
BULK-EFFECT DEVICES
External
Influence
Number
of
Electrodes
Features
No.
Influence
Optical
filter
hv
optical
Lasers
/= aS
1
Resistor
voltage-controlled negative
resistance
Gunn
oscillator
2
current-controlled negative
resistance
Cryosar
/= COnSt.
for
<f
>
threshold
Current limiter
vr
hv,i
2 2
Seeback
effect ?
~ \JT
Photoconductor
Magnetoresistor
J=a{hv)
j=o{3>?)g
Hall
effect,
2
s,
ze
4
V = f{Jf,
#)
Hall generator
2
S,
<f,
2 2 2
*up
^*
J=a{J)S
Thomson
heat,
Thermistor
vr
~ VT
Strain gauge
s,p
3
jp
j
,
peizoresistance effect
2)
jp
Suhl
effect
"Transistors are the final step in the search for speedy, reliable means of
Engineer, 9 They can't possibly go beyond ary new concept." Engineer, ....
communication. "
1
50's,
this revolution-
1960's,
thus important for one to understand the fundamental physical processes and to equip himself with sufficient background in physics and mathematics to digest, to appreciate, and to meet the challenge of these dynamic fields. We should also be aware of the widespread use of digital computers in almost every field. A digital computer is basically a faster, more powerful,
It is
TABLE
1.3
Number
of
Electrodes
Features
No.
Influence
hv
photovoltaic effect
regular p-n junction (p-n)
Diode,
rectifier
Varistor
Varactor
Electroluminescent
p-n used as
light
source
diode
p-n with very high dopings on both sides
with moderately high dopings, direct band-gap,
reflection surfaces
limit voltage
Tunnel diode
Injection laser
by avalanche
Voltage regulator
breakdown or tunneling
microwave generation by impact-avalanche and
transit-time effect
IMPATT
diode
Field-effect diode
semiconductors with
different
Heteroj unction
(n-n, p-n,
band gaps
and p-p)
Schottky diode,
semiconductor
Mott diode,
point-contact
Thermocouple,
VT
Seeback
effect
thermoelectric
generator
r,VT
,hv
Peltier effect
Peltier refrigerator
J = f(S,hv)
Photodiode
Introduction
TABLE
1.4
Number
of
Electrodes
Features
Name
of Device
(where applicable)
No.
Influence
Shockley diode
metal-insulator-metal device
MIM
MIS
tunneling
diode
metal-insulator-semi-
conductor device
p-n-p and n-p-n
diode
Junction transistors
Semiconductor con3
trolled rectifier
(SCR)
1
JFET
or
2
and/or
2
IGFET
transistor
thin-film transistor with de3
TFT
Hot-electron
transistor
semiconductor-metal-semi3
,hv
Optical transistor
and more
nonlinear problems,
With this "sliderule" we can solve many we can handle millions of computations in a short time,
results prior to
many
theoretial results
and
illustrais
The reader
thus
important to point out that many of the devices, especially the microas IMPATT diodes and Gunn oscillators, are still under intensive investigation. Their large-signal behaviors and ultimate performances are by no means fully understood at the present time. The material
It is
presented in this
book
is
is
ci
U T3 o
&0
u
GO
<L>
03
C E
c E
!_
03
o
<a
X>
Ci
3 O
CO
F 3
TD
crj
O O
C E
<u 1>
o c
x:
h.
t>
^
c 3 c
C
03 l"
v
TJ <e
60 *> 77 <"
U
-
<u
a
x.
3 &
emp
C ^
E
essu
u
r-
<s
c
mag
<*
<+
13
4>
X!
U
c/5
X
LU
r^ ^
*
<U
<
-
O
>
n u. o
c
(Tl
c
(]>
u.
w
>
!D
C
<1>
it:
u >
2 SiS 3 w
CC
J-J
ta T3 T3 JD
T3
C o62
<L>
-1-.
13
Q
o o 3 C o o
X!
a.
L.
T3
8 S 7- D 'H
O
l-H
ITS
13
^!
O
o, 60
V Uh
1>
c o
CJ
<ij
'33
00
3
pq
0)
u C
xj
(J
O
(I) (Tl
L-
E c
i-
o,
60
03
GO oo
E 3
U,
O
>
T3
3
V,
60
U 3
rt
J
*
t oj
F
X)
o
1-1
W5
>->
pq
pq
> c C (J y 3 3 o
(Tl
60
UJ
X H
*
<u
10
Introduction
expected to enhance his understanding by consulting the original literature with an inquiring mind that will not blindly accept all statements. He is also
expected to attend technical conferences to exchange, to listen to, and to argue about technical ideas which may clarify misconcepts, inspire new ideas,
is
basically a technology.
We
not discussed in this book. This is because IC are primarily concerned with the physics and
operational principles of the individual devices which form the building blocks of integrated circuits. Only by understanding individual devices can we fully use them either as discrete components or as integrated systems.
REFERENCES
1.
S. J.
Angello, "Review of Other Semiconductor Devices," Proc. IRE, 46, 968 (1958).
2.
Compiled by H. C. Spencer.
PART
SEMICONDUCTOR PHYSICS
Resume
INTRODUCTION CRYSTAL STRUCTURE ENERGY BANDS CARRIER CONCENTRATION AT THERMAL EQUILIBRIUM CARRIER TRANSPORT PHENOMENA
PHONON
SPECTRA,
AND
OPTICAL,
THERMAL, AND HIGH-FIELD PROPERTIES OF SEMICONDUCTORS BASIC EQUATIONS FOR SEMICONDUCTOR DEVICE OPERATION
2
Physics and Properties of
Semiconductors
A Resume
INTRODUCTION
The
physics of semiconductor devices
It is
is
naturally dependent
on the physics
of semiconductors themselves.
summary of
The summary
on semiconductors
For detailed consideration of semiconductor physics the reader should 1 Madelung, 2 4 3 Moll, and Smith.
amount of information
and over thirty illustrations drawn from the experimental data are compiled and presented here. It is well known that, of all semiconductors, the elements germanium and silicon are the most extensively studied. In recent
four tables
years^ gallium arsenide has also been intensively investigated because of
its
from those of Ge and Si, particularly its direct band gap for laser application and its intervalley-carrier transport for generation of microwaves. Hence in this chapter emphasis will be placed on the above three most important semiconductors.
many
interesting properties
which
differ
11
12
A Resume
We
shall first consider the crystal structure in Section 2, since the electronic shall properties of solids are intimately related to their lattice structures.
We
review the Miller indices which define crystal planes in real space and designate momenta in energy-momentum space. In Section 3, we shall consider
energy bands, and the related parameters such as the effective mass m* and the energy band gap E It is found that for most semiconductors there is a
g
.
motion of an electron in a crystal and the free an electron with energy near a minimum in energymomentum space. Most of the electronic behaviors can be described in classical terms as though they were free electrons with an effective mass m*.
The energy gap corresponds to the energy difference between the top of the valence band which is completely filled at 0K and the bottom of the conduction band which is empty at 0K. At higher temperatures there are electrons in the conduction band and holes in the valence band. For most semiconductors the transport processes for electrons and holes are considered to
be independent of each other; and the total conduction current is simply the sum of the electron and hole current components. Carrier concentration and carrier transport phenomena, based on the
energy band theory, are considered in Sections 4 and 5 respectively. Other related properties of semiconductors under various external influences are discussed in Section 6. Finally, the basic equations for semiconductor device
operation are summarized in Section
7.
CRYSTAL STRUCTURE
For a
crystalline solid there exist three primitive basis vectors, a, b, c,
such that the crystal structure remains invariant under translation through any vector which is the sum of integral multiples of these basis vectors. In
other words, the direct lattice sites can be defined by the set
t }\ ,vv|
l
= mz + nb + pc
cells (direct lattices).
(1)
where m,
Figure
n,
1
great
many
of the important semiconductors are of diamond or zincblende lattice structures which belong to the tetrahedral phases, i.e., each atom is sur-
tetrahedron.
rounded by four equidistant nearest neighbors which lie at the corners of a The bond between two nearest neighbors is formed by two electrons with opposite spins. The diamond and the zincblende lattices can be considered as two interpenetrating face-centered cubic lattices. For the
diamond
lattice,
atoms are silicon; in a zincblende (GaAs), one of the sublattices is gallium and such as gallium aresnide
lattice,
such as
Crystal Structure
13
BODY-CENTERED CUBIC
(Na
,
PACE-
W, etc
CENTERED CUBIC
(At, Au, etc)
DIAMOND
(C.Ge.Si.etc)
ZINCBLENDE
(GaAs, GaP InSb.etc)
Fig.
Some
of the important unit cells (direct lattices) and their representative elements
or compounds.
the other
is arsenic. Gallium arsenide is a III-V compound, since it is formed from elements of Column III and Column V in the Periodic Table. Most
however, many semiconductors (including some III-V compounds) crystallize in the wurtzite structure which also has a tetrahedral arrangement of nearest neighbors. Figure 2 shows the hexagonal close-packed lattice as well as the wurtzite lattice which can be considered as two interpenetrating hexagonal closepacked lattices. We note that, as in the zincblende lattice, each atom of the
'
III-V
compounds
wurtzite lattice
is
at the
14
A Resume
Fig. 2
Hexagonal close-packed
lattice
lattice.
corners of a tetrahedron.
The
lattice constants
8 conductors are summarized in Table 2.1 along with their crystal structures. It is of interest to note that some compounds, such as zinc sulfide and cad-
mium
lattice
sulfide,
For a given
can crystallize in both zincblende and wurtzite structures. set of the direct basis vectors, one can define a set of reciprocal basis vectors, a*, b*, c* such that
a*
b x c
2n
a
b* b x c'
= 2n
c x a
,
a-bxc
c*
*~ axb = 2n
a
b x c
(2)
o o
"3
no
I/O
no
1!
^
II
c
c o
N O ON <* m m oo
vd
OO in
V0wrj"\0iflirtvo
II
II
fl-
ifi
in no m o ON Tf
in
^
II
no -* in ci
II
in no
<3
3 o 3
55
Diamond
c
H)
Diamond
Diamond Diamond
T3 "O XI
X)
"c3
o c
co
>>
N N N ^ N N N N N N
4>
DOOUDU
Zincblende
Zincblende Zincblende
Wurtzite
Wurtzite
Cubic
u u
3 3
antimonide
u E
4>
(Diamond)
phosphide
phosphide
antimonide
selenide
phosphide
antimonide
sulfide
sulfide
arsenide
nitride
IB
nitride
arsenide
Tin
Germanium
Carbon
Silicon
o c o o
ca
oxide
sulfide sulfide
CO
Aluminum
+-
Cadmium
Gallium
Cadmium
Cadmium
Gallium
Gallium
Gallium
Boron
Boron
Indium
Indium
Indium
Grey
Zinc
Zinc
Zinc
j/5
Element
or
Compound
UOmw
u O
y
1/3
a t3 t3
U U U N N N
c a c
X) X)
3 3
>
>
>
8
1)
> >
15
16
A Resume
vector
so that a
is
a*
= 2%
b*
= 0,
g
etc ;
lattice
given by
= ha* + kb* +
/c*
(3).
where h, k, and / are integers. It can be shown that the product g 1 = 2n x integer that each vector of the reciprocal lattice is normal to a set of planes in the direct lattice and that the volume V* of a unit cell of the reciprocal lattice is inversely pro; ;
(100)
(110)
(in)
Fig. 3
Miller indices of
some important
planes
in
a cubic crystal
where
is
the lattice
constant.
Energy Bands
17
where
Vc = a
b x
c.
by the
use of the Miller indices which are determined by first finding the intercepts of the plane with the three basis axes in terms of the lattice constants and
then taking the reciprocals of these numbers and reducing them to the smallest
three integers having the
same
ratio.
The
(hkl) as the Miller indices for a single plane or a set of parallel planes.
Miller indices of
Fig.
3.
The shown in
side of
Some
on the negative
the origin.
{hkl}: for planes of equivalent
(010), (001), (TOO), (010),
(100),
[hkl]
(hkl)>
[a x a 2 a c]
2)
Here
it is
For the two semiconductor elements, germanium and silicon, the easiest breakage, or cleavage, planes are the {111} planes. In contrast gallium arsenide, which has a similar lattice structure but also has a slight ionic component
in the bonds, cleaves
on {110}
planes.
The
by a
Wigner-Seitz
The Wigner-Seitz
cell is
from the chosen center to the example is shown 9> 10 a face-centered cubic structure. If one first draws lines from
sites.
all
a Wigner-Seitz
lattice
lattice
It
lattice
can be shown that 11 a face-centered cubic (fee) direct constant "a" has a body-centered cubic (bee) reciprocal
cell
with with
shown
lattice.
in Fig. 4(a)
is
of the reciprocal
Fig. 4(b).
lattice
of an fee direct
of the symbols
we can construct The result is shown in adopted from Group Theory. Some
Similarly
ENERGY BANDS
The band
structure of a crystalline solid,
is
(E-k) relationship
usually obtained
18
A Resume
(a)
Fig.
(a)
Brillouin
(b) Brillouin
zone for diamond and zincblende lattices. zone for wurtzite lattice. Also indicated are the most important symmetry
lines
points and
symmetry
such
as:
zone center L: 27r/o (, 4> i), zone edge along <111> axes (A) X: 27r/o (0, 0, 1), zone edge along <100> axes (A) K: 2tt/o (I, |, 0), zone edge along <110> axes (2). (After Brillouin, Ref. 9; and Cohen, Ref. 12.)
T:
One of the most important theorems basic to band structure is the Bloch theorem which states that if a potential energy PE(r) is periodic with the periodicity of the lattice, then n 13 the solutions ^> k (r) of the Schrodinger equation
equation of an approximate one-electron problem.
'
2 (- 2m V +
P(r) (f>k(r) )]
= Ek
(f)
k (r)
(4)
are of the
form
k (r)
Jk
"
Uk (r) =
Bloch function
(5)
Energy Bands
19
where Uk(r) is periodic in r with the periodicity of the direct lattice. From the Bloch theorem one can show that the energy "k is periodic in the reciprocal lattice, i.e., Ek = Ek+g where g is given by Eq. (3). Thus to label the
energy uniquely
ciprocal lattice.
it is
sufficient to use
cell
of the re-
the reciprocal
The standard convention is to use the Wigner-Seitz cell in lattice as shown in Fig. 4. This cell is called the Brillouin zone
9
or the
first
Brillouin zone.
It is
zone where any energy state can be given a label in the reduced zone schemes. The Brillouin zone for the diamond and the zincblende lattices is the same as that of the fee and is shown in Fig. 4(a). The Brillouin zone for the wurtzite
lattice is
tum k
shown
in Fig. 4(b).
[T =
2n/a(0, 0, 0)],
> axes (A) and their intersections with the zone edge [L = 2n/a(^, \, i)], the <100> axes (A) and their intersections ]X= 2nja(0, 0, 1)], and the <110> axes (I) and their intersections [K= 2n/aQ, f, 0)].
The energy bands of solids have been studied theoretically using a variety of computer methods. For semiconductors the two methods most frequently used are the orthogonalized plane wave method 14 15 and the pseudopotential
'
method. Recent results of studies of the energy band structures of Ge, Si, and GaAs are shown in Fig. 5. One notices that for any semiconductor there is a forbidden energy region in which no allowed states can exist. Above and below this energy gap are permitted energy regions or energy bands. The upper bands are called the conduction bands; the lower bands, the valence bands. The separation between the energy of the lowest conduction band and that of the highest valence band is called the band gap, E which is the g most important parameter in semiconductor physics. Before we discuss the
,
16
17
band structure, we shall first consider the simplified band shown in Fig. 6. In this figure the bottom of the conduction band is designated by E c and the top of the valence band by Ev The electron energy is conventionally defined to be positive when measured upwards, and the hole energy is positive when measured downwards. The band gaps of some important semiconductors are listed 18 in Table 2.2. The valence band in the zincblende structure consists of four subbands when spin is neglected in the Schrodinger equation, and each band is doubled
details of the
picture as
spin is taken into account. Three of the four bands are degenerate at k = (T point) and form the upper edge of the band, and the fourth one forms the bottom. Furthermore the spin-orbit interaction causes a splitting of the band at k = 0. As shown in Fig. 5 the two top valence bands can be approximately fitted by two parabolic bands with different curvatures: the heavy-hole band (the wider band with smaller d 2 E/dk 2 ) and the light-hole band (the narrower band with larger 8 2 E/dk 2 ). The effective mass, which is
when
^ '
<
I 1
O
"C
4>
!->
s 5 g
o
>o
</S
oo
VO rt
f-4 ""^
CJ\
</-)
r11
u->
11
>
."S
d o -H <
t-~ 11
Tt Tj^H ,
Q U
"*
in <D
COS
tn
"o
ffi
m S
O
o"
ii
ii
co
<-J
Wl
o"
ll
II
ll II
p
1
*5 ** ** s s s s
rs oo r- on Ov **
* d
odd d * t d d
V) lO
v~%
^H
VD
5
"S
> * s
c o
l-l
vq
p d
II II
t- 00 ^- VO
CJ 4>
Slit
* *
vo
co
O O
odd o o o odd
v-i
<N r~
W
++
.>
5 S S
C
j
t/5
c3 -3
<c
<
33
3 C3
Dfl
35
<u
33
rt
33
<< a a OO
V)
t/5
n n
ooo
o o o vo in
</-i
55
to
111
Z <
* 8 * f
Ct
8
w
00 **
88
V ON
o <o
o <N
Q O
rr
|^
V-)
r-
tj-
^h
VI
1? 2^
c o
88
00
i-h
(J n>
CT\
w->
CO
o <N
53
8 8 2O O Q OOO O O vo t oo oo rt m
tJr-~
a
nl
o^
-h v->
>/-i
>/->
00
(-;
*"i
o
*
^* co
VI
r~;
o oo
es
<ri
_^
o" -I r4
* ^
2> c w
rt
PO
vo rt vq >o
r-~
rJ
CO vq
co
i/o I' 2
VO
m ^ m m ^
^^
o o 3 TD C o o S
in
>_
T3
J2 "2
a o E
33
<u
5
y
S2
5 uo
33
._
<
o. PQ pq
oooo
rt rt rt
cti
!/3
<
Ph
00
(3
< C
Ph
CJ
e i
s
20
> J
1
>
1 1
'
o\ oo
r-
ie
VO
d d
VO fS
d d
w
ctf
** S
.
#.
'O
VJ
(/J
< a
VI
crt
</3
</>
ooo
a a
4)
a)
Ctf
Q
GO
4_>
o o o o t-
CO
"
4)
00
^3
II
8.5
w-
S
00
O ^ 2 e G
O CO
VO (N
Q O o O vo
J3
u 2 .*
GO
go
,4)
3
O.
*\
e3
W go
VO
C-J
in 00
t-;
d d
,2 c
r~;
v
en
d d
5T 4
i>
o .. M .-3 go
i
c
2 8
,s c S to T3 -a "5
in
<2
O C UN N
W3
S
&h Pi
3
>
00 8"H' a G KJ O aj .-
< g 3
||
>
>
_g
h a
1-
<u
++
s
cos
21
22
A Resume
VA
G
GaAs
f\\
CONDUCTION ELECTRONS
I
/
UPPER \ VALLEY \
~y^
1
E9
\/
,/
1
+~^^v_/
AE = 0.36
T~
LOWER VALLEY
Eg
\
[III].
[100]
[hi]*-
--[ioo]
k
[III].
[100]
MOMENTUM
Fig. 5
Si,
is
and (
bands.
(After
Cohen and
ELECTRON ENERGY
HOLE ENERGY
--
DISTANCE
Fig. 6
Energy Bands
23
defined as
m =
/d E\
'
(6)
Table 2.2 for some important semiconductors. The conduction band (Fig. 5) consists of a number of subbands. The bottom of the conduction band can appear along the <lll>-axes (A or L), along the <100>-axes (A or X), or at A: = (F). Symmetry considerations alone do not determine the location of the bottom of the conduction band.
is listed in
Ge
it is
1 1
>-axes,
energy surfaces
19
at
7.
A;
= 0. The
revolution along the <lll>-axes; the Brillouin zone boundaries are at the
middle of the ellipsoids. For Si there are six along the <100>-axes with the centers of the ellipsoids located at about three quarters of the distance from the Brillouin zone center. For GaAs the constant energy surface is a sphere at the zone center. By fitting experimental results to parabolic bands, we
one for GaAs, two for Ge, and two for * transverse to the symmetry axes. along the symmetry axes and These values also are given in Table 2.2.
obtain the electron effective masses
Si:
;
m*
z
At room temperature and under normal atmosphere, the values of the band gap are 0.66 eV for Ge, 1.12 eV for Si, and 1.43 eV for GaAs. The above values are for high-purity materials. For highly-doped materials the band gaps become smaller. (The band gaps of degenerate semiconductors are considered in Chapter 4.) Experimental results show that the band gaps of most semiconductors decrease with increasing temperature. The detailed
GaAs
in Ge, Si, and GaAs. For Ge there are eight <111> axes, and the Brillouin zone boundaries are at the middle of the ellipsoids. For Si there are six along <100> axes with the centers of the ellipsoids located at about three quarters of the distance from the Brillouin zone center. For GaAs the constant energy surface is a sphere at the zone center. (After Ziman, Ref. 19.)
Fig. 7
24
A Resume
GaAs
Egm-EgW)-*^
^\U43)
1.2
Si
^(1.12)
0.9
MATERIAL Eg(0)
Ge
a
(XIO"
4
)
GaAS
1.522
1.16
5.8
300
1108
0.7
^^\(0.66)
1
Si
7.02
0.6
400
V
T(K)
Ge
i
0.741
1
4.56
,
210
1
800
Fig.
Energy band gaps of Ge, Si, and GaAs and Casey, Ref.
as a function of
temperature.
73).
variations of
band gaps
and GaAs
are
shown 4
in Fig. 8.
The band gap approaches 0.75, 1.16, and 1.52 eV three semiconductors at 0K. The variation of band gaps
with temperature can be expressed by a universal function E (T) = E (0) g g - ccT2 j(T + fi), where ^(0), a and p are given in Fig. 8. The temperature
coefficient, dEJdT, is negative for the above three semiconductors. There are some semiconductors with positive dEJdT, e.g. PbS (Table 2.2), which increases from 0.34 eV at 0K to 0.41 eV at 300K. Near room temperature, the band ~ gaps of Ge and GaAs increase with pressure 20 (dEJdP = 5 x 10 6 eV/kg/cm 2 6 2 for Ge, and about 9 x 10~ eV/kg/cm for GaAs) and that of Si decreases -6 with pressure (dE fdP= -2.4 x 10 eV/kg/cm 2 ).
,
25
The three basic bond pictures of a semiconductor are shown in Fig. 9. Figure 9(a) shows intrinsic silicon which is very pure and contains a negligibly small amount of impurities each silicon atom shares
;
its
four valence electrons with the four neighboring atoms forming four covalent bonds (also see Fig. 1).
Figure 9(b) shows schematically an -type silicon where a substitutional phosphorus atom with five valence electrons has replaced a silicon atom, and a negative-charged electron is "donated" to the
0::
(a) INTRINSIC Si
conduction band. The silicon is n type because of the addition of the negative charge carrier, and the phosphorus atom is called a "donor." Similarly as
shown
in Fig. 9(c),
three
valence electrons substitutes for a silicon atom, an additional electron is "accepted" to form four
covalent bonds around the boron, and a positive charged "hole" is created in the valence band. This is p type, and the boron is an "acceptor."
(1)
"'
:
Intrinsic
Semiconductor
We now consider the intrinsic case. The number of occupied conduction-band levels is given by
n
(b)n-TYPE
Si
WITH DONOR
f J
N(E)F(E) dE
EC
(7)
;::*
Ec is the energy at the bottom of the conduction band and Etop is the energy at the top. N(E) is the density of states which for low-enough
where
carrier densities and temperatures can be approximated by the density near the bottom of the con-
;, ::0:::
: :
(c)p-TYPE
Si
WITH ACCEPTOR
duction band:
N(E) = M,
where
J2(E-E cy 2
h-
0n de ) 3/2
Fig. 9
Three
basic
bond
(8)
pictures of a semiconductor:
(a)
intrinsic
is
the
with negligible
impurities;
(b) n-type with donor (phosphorus); and
(c)
effective
mde (m
*m 2 *m 3 *) x/3
p-type
with
acceptor
(9)
(boron).
26
A Resume
.
where
*,
m 2 *, m 3 *
in silicon
mde = (raj*m,* 2
1/3
)
F(E)
is
by
(10)
}
F(E)
=
1
exp
E-E
kT
where k is Boltzmann's constant, T the absolute temperature, and EF the Fermi energy which can be determined from the charge neutrality condition
[see Section 4(3)].
The
integral,
Eq.
(7),
can be evaluated to be
2
/EF
- Ec\
en)
band and
is
where
Nc
is
given by
N _J " \
f
Mc
,
(12)
)
is
and
F1/2
(r}
f)
is
shown 21
in Fig. 10.
For the
6-4-2
(E c
-E F )/kT =
Fig. 10
27
case of Boltzmann
kT below E c
in
nondegenerate
semiconductors,
the
approaches
Jne~ ns\2
and
Similarly,
iV c
exp(-^^.
2
r,
(13)
we can
obtain the hole density near the top of the valence band
P~
where
r N
{E V
-EF \
(14)
"T*4~^>
band and
is
Nv
is
given by
Nv =
where
J2nm dh kT\^
{
(15)
)
m dh is the density-of-state effective mass of the valence band: 4 dh = (m*V 2 + mth3/2 ) 213
(16)
where the subscripts refer to "light" and "heavy" hole masses discussed before, Eq. (6). Again under nondegenerate conditions
p^NyCxp^-^^y
(17)
For intrinsic semiconductors at finite temperatures there is continuous thermal agitation that results in excitation of electrons from the valence band to the conduction band and leaves an equal number of holes in the valence = n This process is balanced by recombination of the band, i.e., electrons in the conduction band with holes in the valence band.
n=p
The Fermi
level for
is
nondegenerate)
an intrinsic semiconductor (which by definition obtained by equating Eqs. (13) and (17):
is
Ec + E v
Hence the Fermi
level (;)
kT
Nv
Ec + E v
3kT
(m dh \
very
lies
band gap.
(13), (17),
The
(18):
and
(
1
np
or
.
2
.
nc Nv exp(
EJkT),
9)
= =
^Nc N v e- E l2kT
4.9 x 10
15
/^^\ \^ e -E
g iikT
(19a)
28
Resume
where Eg = (Ec Ev ), and m is the free-electron mass. The temperature dependence 22 23 of n is shown in Fig. 11 for Ge, Si, and GaAs. As expected, the larger the band gap the smaller the intrinsic carrier concentration.
'
(2)
When
400
300 273
1
10"
10
V Ge
10"
I
co
15
to
\si
10
Q
<r
3
io'
^aAs
CO
10
10"
_,
10'
10'
0.5
1.0
1.5
2.0
2.5
1
3.0
3.5
4.0
lOOO/TCK"
Fig. 11
Intrinsic carrier densities of Ge, Si,
and GaAs
as
(After Morin and Maita, Ref. 22; Hall and Racette, Ref. 23; and Neuberger, Ref. 18.)
29
neutral neutral
by an
electron,
and
positive if empty.
An
acceptor level
is
if filled
by an
electron.
The simplest calculation of impurity energy levels is based on the hydrogenatom model. The ionization energy for the hydrogen atom is
where
is
The
Ed
by the conductivity
_1
1
1
donor mass 4 of
electrons
m ce=3[
and
e
,/l
es
in Eq. (20)
*-ef(K
The
ionization energy for donors as calculated
for Ge, 0.025
eV
for
Si,
from Eq. (21) is 0.006 eV and 0.007 eV for GaAs. The hydrogen-atom
is
donors.
one plus an imaginary hole in the central force field of a negatively charged acceptor. The
filled
We
from the valence band edge) and about 0.05 eV for GaAs. The above simple hydrogen-atom model certainly cannot account for the details of ionization energy, particularly the deep levels in semi24-26 conductors. However, the calculated values do predict the correct order of magnitude of the true ionization energies for shallow impurities. Figure 12 shows 27 28 the measured ionization energies for various impurities in Ge, Si, and GaAs. We note that it is possible for a single atom to have many levels, e.g., gold in Ge has three acceptor levels and one donor level
calculated acceptor ionization energy (measured
is
0.015
eV
eV
for Si,
'
in the
Of
all
forbidden energy gap. 29 the methods of introducing impurities into a host semiconductor,
30
is
solid-state diffusion
method 31
is
still
in
its
infancy.)
The
D(T)
can be described by
(22)
D(T)
where
= D
exp( - AE/kT)
is
AE
is
is
GaAs
D(T) for Ge, Si, and The D(T) for GaAs larger concentrations D(T)
Li
sb
[&0095QO096
Ag
0.09 A
0.29
Au
0.04
A
0.20
0.14
0.28
0.31
A
0.35
GAP CENTER
_A
0-30
_J_
0.25
0.12
ooi
B
ooi aoi
JJ
go
in
0.07
OI3_
0.15 -
"A
0.09
0.04
6~o
7~
0.05
<L! 05
0.16
0.07
AC
ML
Be
zn
cd
Mn
Ni
Hg
Pt
cr
Sb
0.039
AS
Bi
Mn
Ag
pt
Hg
00 44 0.049
0.069
Q.33 0.37-
GAP CENTER
0.55
-
53
Q.52
0.37
0.40
0.35
D
0.045
B
"
AS.
Ga
in
co
zn
SHALLOW LEVEL
GAP CENTER
0.52
0.37 0.63 D
0.51
0.24
0.143
,
0.15
0.019 0_02l
O023
Li
0.023
0.023
Mg
cd
zn
Mn
Co
Ni
si
Fe
cr
Li
Measured ionization energies for various impurities in Ge, Si, and GaAs. The levels below the gap centers are measured from the top of the valence band and are acceptor levels unless indicated by D for donor level. The levels above the gap centers are measured from the bottom of the conduction band level and are donor levels unless indicated by A for acceptor level. The band gaps at 300K are 0.66, 1.12, and 1.43 eV for Ge, Si, and GaAs
Fig. 12
respectively.
(After Conwell, Ref. 27; and Sze and Irvin, Ref. 28.)
30
TEMPERATURE
900
1
(C)
800
1
700
1
600
1
1400
1
1200
1
H00
1
1000
1
1200
1
1100
1
1000
1
900
1
Cui
Ni.Cu
^Li
Li
^Li
Fe ^""^n.
'"-^^Cu
A^
^IHe
Zn
*>^A9
\Au
V^Au
"""*
\Mn(WITH
\
o
I0"
10
l\
yS
\Au
0\sb
10""
V\ai
\\
InTi
p\\
B
\^B
3^\
^\uVV
VSn
vv\B
\ Zn
^
\
\S
As
Ago
\\ as\
1
1
\ Cd \
1
\V\
2"S
>
0.8 .65 0.7
\
0.8
\
0.9
0.8
0.9
1.0
I.I
1.2
0.6
0.7
Si
GoAJ
1000/ T(K)
Fig. 13
Si,
and GaAs.
(After Burger and Donovan, Ref. 32; and Kendall, Ref. 33.)
31
32
A Resume
with
the
In
connection
which
is the maximum concentration of an impurity that can be accommo34 of some dated in a solid at any given temperature. The solid solubilities important impurities in Si are plotted in Fig. 14 as a function of temperature. This figure shows that arsenic or phosphorus should be used as the impurity
in
silicon while
(3)
The Fermi
lies
semiconductor
is
depicted in
distribution
Fig. 15(a) where, from left to right, are shown band diagram, the density of states N(E), the Fermi-Dirac
TEMPERATURE
Fig. 14 Solid solubility of various elements in
Ref. 34.)
Si as
CO
temperature.
a function of
(After
Trumbore,
33
function F(E), and the carrier concentrations. The shaded areas in the conduction band and the valence band are the same indicating the n = p = n for the intrinsic case.
When
Fermi
level
impurity atoms are introduced as shown in Fig. 15(b) and (c), the must adjust itself in order to preserve charge neutrality. Consider
N c expj-(E c -E F )/kT~|
(--ns)
N v expj-(E F -E v )/Kt1
( =
n.)
(b)
TYPE
0.5
1.0
N(E)
(c) p
F(E)
AND
TYPE
Fig. 15
Schematic band diagram, density of states, Fermi-Dirac distribution, and the carrier
concentrations for
(a)
intrinsic,
(b) n-type,
(c)
and
at
p-type
semiconductors
=n
2
t
for
all
three cases.
34
A Resume
the case as
where donor impurities with the concentraadded to the crystal. To preserve electrical neutrality we must have the total negative charges (electrons and ionized acceptors) equal to the total positive charges (holes and ionized donors), or for the
shown
3 )
in Fig. 15(b)
are
present case
n
= ND + +p
p
is
(23)
where n
in the
is
ND +
is
the
number of
1
N D + = ND
where g
is
.,
+
,
- exp r
(Ed
\
EF kT
(24)
is the ground state degeneracy of the donor impurity level which equal to 2 because of the fact that a donor level can accept one electron with either spin or can have no electron. When acceptor impurities of con-
centration
be written for the charge neutrality condition, and the expression for ionized
acceptors
Nr
=
1
-f-
-exp r
9
\
(25)
kT
where the ground state degeneracy factor g is 4 for acceptor levels. This is because in Ge, Si, and GaAs each acceptor impurity level can accept one hole of either spin and the impurity level is doubly degenerate as a result of the two degenerate valence bands at k = 0.
Rewriting the neutrality condition of Eq. (23),
we
obtain
K exp(
1
(E v
EF
2 expl
kT
(26)
For a
set
of given
Nc ND Nv Ec ED Ev
,
,
and
T, the
Fermi
level
EF
can
to
(26).
16.
An
For
elegant graphical
method 70
ND
10
16
cm -3 T = 300K)
,
EF
is
illustrated in Fig.
band
edge and adjusts itself so that almost all the donors are ionized. For another temperature, one can first evaluate the values of JVC and v which are pro3/2 then obtain from Fig. 11 the value of n^T) which deterportional to r mines the intercept of the lines n(EF ) and p(EF); a new Fermi level is thus
35
Fig. 16
obtained.
As
the temperature
is
lowered
sufficiently, the
Fermi
level rises
semiconductors) and the donor level is with electrons. The approximate expression for the electron
level (for -type
density
is
then
2N
)N C expi-EJkT)
(27)
for a partially
NA >Nc exp(-EJkT)
36
Semiconductors A Resume
where
Ed =
(Ec
ED ),
or
n*-j= WoNc)
for
1 '2
exp(-EJ2kT)
example
is
(28)
typical
where n
is
peratures
we have
p P ND
is
Eq. (28), depending on the compensation conditions. The electron density, however, remains essentially constant over a wide range of temperatures
in Fig. 17).
shown
is
in Fig.
18.
The dependence of
the
temperature
(cf.
Fig. 8)
is still given by Eq. (19) which is called the mass-action law and the product is independent of the added impurities. At relatively elevated temperatures, most of the donors and acceptors are ionized, so the neutrality condition can be approximated by
When
NA =p + ND
(29)
10
F
: "
1
|
i
.J 5
10
cm"3
INTRINSIC
RANGE
=
SLOPE
16
'e
io
r
-
|l
SATURATION RANGE
ID
10
s^
FREEZE-OUT RANGE
4
H
io'
b-
r
-
\
i
n,-
-(p))
12
20
1
1000/T
Fig. 17
( K"
Si
concentration of 10 l5 cm
(After Smith, Ref. 4.)
-3
.
37
100
200
300
T (K)
400
500
600
Fig. 18
Si as
a function of
We
can combine Eqs. (19) and (29) to give the concentration of electrons and holes in an -type semiconductor:
n no
= K(ND -
NA
n
2
t
+ J(ND -
NA 2
)
+ Anh
and
*ND
Pno
if
\ND
-NA \^fii
,
ND
$>NA
(30)
ni /n no
jND
(31)
and
EC
or from Eq. (18),
-E F =
kT In
(32)
Ep-E^kTln^.
(33)
38
A Resume
is
The concentration of
given by
holes
and
electrons
on a p-type semiconductor
P Po = KWa ~ *d) +
wNA
n po
if
\NA
i
-ND \>n
,
and
NA >ND
(34)
(35)
n 2 /p po
i
^n 2 /NA
and
EF
or
-E v = kT\n^,
(36)
-EP = kT]n^.
(37)
In the above formulas the subscripts n and p refer to the type of semiconductors, and the subscripts "o" refer to the thermal equilibrium condition.
For -type semiconductors the electron is referred to as the majority carrier and the hole as the minority carrier, since the electron concentration is the
larger of the two.
The
CARRIER TRANSPORT
(1)
PHENOMENA
Mobility
electric field the drift velocity (vd) is proportional to the electric
At low
(fi
is
in
cm 2 /volt-sec),
nS.
(38)
For nonpolar semiconductors such as Ge and Si there are two scattering mechanisms which significantly affect the mobility, namely scattering due to acoustic phonons and to ionized impurities. The mobility due to acoustic 37 phonon, fi is given by
t ,
where
x t
is
Eds
the average longitudinal elastic constant of the semiconductor, band per unit dilation of the lattice,
mass.
From
it is
expected
Carrier Transport
Phenomena
39
effective
mass.
to ionized impurities,
/z f
is
given by 38
(40)
r is the ionized impurity density, and s s the permittivity. The mobility expected to decrease with the effective mass but to increase with the temperature. The combined mobility which consists of the above two mechanisms is is
where
fi
=(-+-)
(41)
For polar semiconductors such as GaAs there is another important scattering mechanism: optical-phonon scattering. The combined mobility can be approximated by 39
ji~(/*)- 3/2 r 1/2
.
(42)
which
In addition to the above scattering mechanisms there are other mechanisms also contribute to the actual mobility: for example (1) the intravalley scattering in which an electron is scattered within an energy ellipsoidal
and only long-wavelength phonons are involved; and (2) the interwhich an electron is scattered from the vicinity of one minimum to another minimum and an energetic phonon is involved. The measured mobilities of Ge, Si, and GaAs versus impurity concentrations at room temperature are shown 2 8 25 40 in Fig. 19. We note that,
(Fig. 7)
valley scattering in
>
>
room temperature
most impurities are ionized) the mobility decreases as predicted by Eq. (40). Also as m* increases, p decreases, thus for a given impurity concentration
the electron mobilities for these semiconductors are larger than the hole mobilities (the effect masses are listed in Table 2.2).
p-type silicon
on mobility is shown 41 in Fig. 20 for w-type and samples. For lower impurity concentrations the mobility decreases with temperature as predicted by Eq. (39). The measured slopes, however, are different from (-3/2) owing to other scattering mechanisms. It is found that for pure materials near room temperature the mobility varies 166 as T~ and T~ 233 for n- and p-type Ge respectively; as T~ 25 and 1 T~ 2J for n- and /Hype Si respectively; and as and T~ 2A for n- and
The temperature
effect
^p-type
GaAs
respectively.
40
A Resume
carrier
equili-
Another important parameter associated with mobility is the and D p for holes). In thermal 4 brium the relationship between D n and fi n (or D p and (i p) is given by
diffusion coefficient (Z> for electrons
::::
t= 300K=:-
MP
Ge^(
H-n
o
UJ
to
10'
I0<
o >
Mn
>i
3 =s-
rSi
L-J.
Mm
j:::^
ffi
2
io
/xp
"^-;-
i^4
3oAs-(
Mn
10*
zk~~~"~
:
/xp
"*""*--.
.
10
14
10
15
.16
10
17
10
18
10
19
Drift mobility of
Ge and
Si
vs.
impurity con-
centration.
(After Sze and Irvin, Ref. 28; Prince, Ref. 40; and Wolfstirn, Ref. 35.)
41
</>
4000
> 2000
Si
i
>-"
1000
14
800 600
-s=rN=iO
'.
_|6"""
1_l
ffi
400
N=
IC
r-
o
?
.
200
N
100
=
o at
\-
N = I0
I0
19
-50
50
(0)
100
150
200
400
200
Si
N=l0i r*--.
y* I0 14
N=I0
100
80 60
r-
fir-
40
20
10
N=l
I9
50
50
100
150
200
TEMPERATURE
(b)
Fig.
CO
20
in Si as
concentration.
(After Gartner, Ref. 41.)
where
F1/2
integrals.
"-(tK
and
similarly
(44a)
D p=\-rW
f?>
(44b)
42
SemiconductorsA Resume
as the Einstein relationship. At room and values of D's are readily obtainable from the mobility results shown in Fig. 19. The mobilities discussed above are the conductivity mobilities, which have been shown to be equal to the 27 They are, however, different from the Hall mobilities to be drift mobilities.
are
known
= 0.0259
volts,
(2)
Resistivity
The
electric field
(/)
(45)
& = pJ.
Its reciprocal
value
is
the conductivity,
i.e.,
l/p,
and
(46)
J=aS.
For semiconductors with both electrons and holes
1
1
as carriers,
we
obtain
a
If n
q(fin
+ n p p)
(47)
>p
as in -type semiconductors,
p~
or
(48)
q\x n
<j~q\i n
n.
is
(48a)
the four-point
in Fig. 21.
SEMICONDUCTOR SAMPLE
Fig. 21
resistivity.
(UJO-WHO)
A1IAUSIS3U
iZ
43
44
A Resume
current source
is passed through the outer two probes and the voltage is measured between the inner two probes. For a thin slice with its thickness much smaller than its diameter, and the probe spacing S ^> W, the resistivity is given by
_ ~
The measured resistivity 28 A4A5 is shown
~.
4 54
-
\\n2}
(at
IW~ W
F\
(49)
\l)
centration
300K) as a function of the impurity con22 for Ge, Si, and GaAs. Thus we can obtain the impurity concentration of a semiconductor if its resistivity is known. The impurity concentration may be different from the carrier conin Fig.
centration.
For example,
(25), Figs. 12 -3 16
17
cm
-3
gallium acceptor
[from Eq.
is
and 18];
only 7.7 x 10
cm
46
To measure
common method
where an electric magnetic field, along the z-axis. Consider field is applied along the x-axis and a downward a p-type sample. The Lorentz force (qvx x x ) exerts an average and the downward-directed current will cause a piling up force on the holes, of holes at the bottom side of the sample, which in turn gives rise to an
uses the Hall effect.
The
basic setup
47
is
shown
in Fig. 23
&
electric field
Sy
Since there
is
direction in the
field
Fig. 23
Basic setup to
effect.
Carrier Transport
Phenomena
is
45
This Hall
field
given by
(50)
y = {Vy lW) = R H Jx @ z
where
Ru
is
and
is
given by
R" =
r
P~b 2 n
2
.
q(FTbny'>
<t >/<t>
2
bs
^
~ E 312
(51)
(52)
The parameter
is
the
mean
x
free time
e.g.,
~ E~ 1/2
for
phonon
scattering,
for
where a and s are constants. From Boltzmann's distribution for nondegenerate semiconductors
the average value of the rath
= aE~ s
power of
t is
m <t >
=
'o
E 3/2 exp(-//cT) dE
/
E3
Jo
'
exp(-E/kT) dE
(53)
<r >
= =
a (kT)-
T(i -
2s)/r(l)
(54a)
and
<t>
2
[a(/cT)-T(l
5)/r(l)]
(54b)
where T()
is
the
gamma
T(n)=
function defined as
xn
e- x dx,
{T{\)
Jn).
We
obtain from the above expression r = 3n/S = 1.18 for phonon scattering and 3157r/512 = 1.93 for ionized impurity scattering. The Hall mobility \i H is defined as the product of Hall coefficient and
conductivity
Ii
h =\Rho\.
ju
(55)
(or
jj,
p)
From
Eq. (51),
Pp
\qnj
(56a)
and
if
p>n
R H = r()
(56b)
46
A Resume
Thus the
carrier concentration and carrier type (electron or hole) can be obtained directly from the Hall measurement provided that one type of
carrier dominates.
In the above consideration the applied magnetic field is assumed to be is no change in the resistivity of the sample. However,
fields, significant
is
observed
field is
zero magnetic
given by
Po
r 2 (i)r(i - 3 S ) r 3 (f-s)
fi n
+ n p 3 pi Hn + n P P
3
T(i)T(i-2sy p n H P V r 2 (f-s) J ln n n + n p p]
It is
(57)
proportional to the square of the magnetic field component perpendicular similar /' to the direction of the current flow. For n > p, (Ap/p ) for the case p > n. result can be obtained
(3)
Recombination Processes
the
Whenever
distributed,
thermal-equilibrium condition
of a physical system
is
i.e.,
pn
2
,
pn
2
t
.
shown
bination where an electron-hole pair recombines. This transition of the electron from the conduction band to the valence band is possible by the emission of a photon (radiative process) or by transfer of the energy to
another free electron or hole (Auger process). The latter process is the inverse process to impact ionization, and the former is the inverse process to direct optical transitions, which are important for most III-V compounds
with direct energy gaps.
Figure 24(b) shows single-level recombination in which only one trapping energy level is present in the band gap, and Fig. 24(c) multiple-level recombination in which more than one trapping level is present in the band gap.
The
single-level
capture, electron emission, hole capture, and hole emission. -1 -3 48 sec (in unit of cm tion rate, ), is given by
U=
n
"<
Vp<rnVth(pn
)N
exp
HH
(E
t
t
-E \l
J
ni
exp
E -Ei\t _
(58)
r ~tt)
ENERGY TRANSFERRED
TO FREE ELECTRON OR HOLE (AUGER PROCESS)
-m*
PHOTON
EMISSION OF
(RADIATIVE)
(Q)
BAND - TO - BAND
/4/
D
// / /"7
/
/ / /-*
E,
& / V
D
'
(i) (2)
D
(3) (4)
/ / / y / / / / / / / /-?
ELECTRON
EMISSION
I
ELECTRON CAPTURE
HOLE
CAPTURE
HOLE EMISSION
(b)
SINGLE
LEVEL
D
/ //
? 6 i a n
// /
(c)
A/
LEVEL
I',
MULTIPLE
24 Recombination processes band-to-band recombination (radiative or Auger process) (b) single-level recombination
Fig.
(a)
(c)
multiple-level recombination.
47
48
A Resume
E
t
where a p and a are the hole and electron capture cross section
v th the carrier
respectively,
the
U=
0.
2 obvious that under a thermal equilibrium condition, pn = n Furthermore, under the simplified condition that a = a p a, Eq. (58)
reduces to
U=
av th
+p+
2n cosh
t
P^)
E E
t t
.
(59)
The recombination
tive
rate approaches a
maximum
i.e.,
effec-
recombination centers are those located near the middle of the band gap. Under low injection conditions, i.e., when the injected carriers (Am =Ap) are much fewer in number than the majority carriers, the recombination
process
may
U=
Pn
( 60 )
where pno is the equilibrium minority carrier concentration, p = Ap + pn0 and x p is the minority carrier lifetime. In an H-type semiconductor, where n n no =the equilibrium majority carrier concentration, and n > n and p, Eq. (58) becomes
, {
U = ap v N
th
(pn
- pno
).
(61)
carrier lifetime (hole
lifetime) in
Comparison of Eqs. (60) and (61) yields the minority an H-type semiconductor, and
1
ap vth N
(62)
t
(63)
WthNt
For multiple-level traps it has been shown that the recombination processes have gross qualitative features that are similar to those of the single-level case. The details of behavior are, however, different, particularly in the highinjection-level condition (i.e., where An = Ap a5 majority carrier concentration) where the asymptotic lifetime is an average of the lifetimes associated
with
all
levels.
Carrier Transport
Phenomena
49
The
by the use of
in Fig. 12 that
solid-state diffusion
and (63), have been verified experimentally and high-energy radiation. It can be seen
many
is
band
impurities have energy levels close to the middle of the may serve as efficient recombination centers.
silicon.
typical
example
gold in
29
It is
carrier
to 10
lifetime decreases linearly with the gold concentration over the range of 10 14 17 cm -3 where t decreases from about 2 x 10~ 7 second to 2 x 10~ 10
is
important in some switching device applications when a desirable feature. Another method of changing the minority carrier lifetime is by high-energy particle irradiation which causes displacement of host atoms and damage to lattices. These, in turn, introduce energy levels in the band gap. For example, 18 in silicon electron
a short lifetime
is
irradiation
an acceptor level at 0.4 eV above the valence band and a donor level at 0.36 eV below the conduction band; neutron irradiation creates an acceptor level at 0.56 eV; deuteron irradiation gives rise to an interstitial state with an energy level 0.25 eV above the valence band. Similar results are obtained for Ge, GaAs, and other semiconductors. Unlike the solid-state
gives rise to
diffusion, the radiation-induced trapping centers
relatively
may
be annealed out at
low temperatures.
The minority carrier lifetime has generally been measured using the photoconduction effect 49 (PC) or the photoelectromagnetic effect 50
basic equation for the
(PEM ). The
PC effect
Jpc
is
given by
= q{Hn + Hp)An
(64)
where JPC
is
is the incremental current density as a result of illumination and S the applied electric field along the sample. The quantity An is the incremental
number of electron-hole pairs per the illumination which equals the product of the generation
from photon (G) and the
lifetime (t) or
volume created by
rate of electron-hole
pairs resulting
is
An =
xG.
The
lifetime
thus given by
An
J pr Gq(n n
+ u)
(65)
A measurement setup will be discussed in Section 7. For the PEM measure the short-circuit current, which appears when a constant
field
effect
we
magnetic
z is
current density
The
J pem
q(M
+ HP)@Z - (tG)
(66)
50
A Resume
where
L=
J Dx
is
MjDGq{[i n +
The
2
lifetime
is
given by
(JpEM i@ z f.
(67)
fi p ).
PHONON
high electric
section
fields
on
we
(1)
Phonon Spectra
known
that for a one-dimensional lattice with only nearestand 2 placed alternately, the t
It is well
neighbor coupling and two different masses, 3 frequencies of oscillation are given by
V+ +
,
|_
af
/I ( +
'
\
)
Wi
m 2/
af
+ V v \^i m
// A /(
L
.
1/2
a 4 sin
2 2,
^ ^)//^ m 2
(68)
if
the force constant, q the wave number, and a the lattice spacing. to be proportional to q near q = 0. This is the acoustic branch, since it is the analogue of a long- wavelength vibration of the lattice, and since the frequency corresponds to that of sound in such a medium. The
where a f
is
frequency v+ tends to
2a r
.
( + m \m
x
1/2
as q approaches zero. This branch is separated considerably from the acoustic mode. This is the optical branch, since the frequency v+ is generally in the
optical range.
different
For the acoustic mode the two sublattices of the atoms with masses move in the same direction, while for the optical mode, they
lattice
move
in opposite directions.
For a three-dimensional
cell
such as a simple
cubic, body-centered, or face-centered cubic lattice, there are only three acoustic modes. For a three-dimensional lattice with two atoms per unit cell such as Ge, Si, and GaAs, there are three acoustic modes and three optical
modes. Those modes with the displacement vectors of each atom along the
51
thus
wave vector are called the longitudinally polarized modes we have one longitudinal acoustic mode (LA) and one longitudinal optical mode (LO). Those modes with their atoms moving in the planes normal to the wave vector are called the transversely polarized modes. We have two transverse acoustic modes (TA) and two transverse optical modes
direction of the
(TO).
The measured
to
q.
results
#'s,
51
53
for Ge,
Si,
in Fig. 25.
We
with
LA and TA
The Raman phonon energy is that of the longitudinal optical phonon at Their values are 0.037 eV for Ge, 0.063 eV for Si, and 0.035 eV for GaAs. These results are listed in Table 2.3 along with other important properties of Ge, Si, and GaAs (in alphabetical order).
0.
(2)
Optical Property
Optical measurement constitutes the most important means of determining the band structures of semiconductors. Photon-induced electronic
Ge
TO
Si
LOnO.05
TO LO
LA/
LA/
TA
TA__
[ioo]
[00,]
'"
q/q MAX
Measured phonon spectra in Ge, Si, and GaAs. TO for transverse optical modes, mode, TA for transverse acoustic modes, LA for longitudinal acoustic mode. (After Brockhouse and Iyengar, Ref. 51 Brockhouse, Ref. 52; and Waugh and Dolling, Ref.
LO
53.)
52
A Resume
between different bands which lead to the determination band gap; or within a single band such as the free-carrier absorption. Optical measurement can also be used to study lattice vibrations. The transmission coefficient T and the reflection coefficient R are the two important quantities generally measured. For normal incidence they are given by
of the energy
(1
- R 2 )exp(-4nx/X)
'
l-U 2 exp(-87tx/A)
(1
(69)
- nf +
where X is the wavelength, n the refractive index, k the absorption constant, and x the thickness of the sample. The absorption coefficient per unit length
a
is
defined as
- 4nk =
nn (71)
By analyzing
the
observations of
T X or R X data at normal incidence, or by making or T for different angles of incidence, both n and k can be
obtained and related to transition energy between bands. 4 Near the absorption edge the absorption coefficient can be expressed as
a
(Av
- Egy
(72)
is the photon energy, E is the band gap, and y is a constant which g equals 1/2 and 3/2 for allowed direct transition and forbidden direct transition
where hv
respectively [with
k min
= k max
as transitions (a)
(c)
phonons must be incorporated. In addition y equals 1/2 for allowed indirect transitions to exciton states where an exciton is a bound electron-hole pair with energy levels in the band gap and moving through the crystal lattice as
a unit.
(hv
Eg ) become
com-
Coulomb
interaction be-
tween the
and electron must be taken into account. For hv absorption merges continuously into that due to the higher excited
free hole
< Eg the
states of
the exciton.
When hv > Eg
higher energy bands will participate in the transistructures will be reflected in the
tion processes,
absorption coefficient.
coefficients near
edge (band-to-band transition) for Ge, Si, and GaAs are 54" 56 in Fig. 27. The shift of the curves towards higher photon plotted
53
CONDUCTION BAND
Fig.
(a)
26 and
is
Optical transitions.
(b) are direct transitions.
(c)
is
(3)
Thermal Property
a temperature gradient
is
When
an applied
one dimension)
is
3EF
q dx
~8x) (73)
where a is the conductivity, EF the Fermi energy, and the differential thermoelectric power. For a nondegenerate semiconductor with a mean free ~s time t ~ as discussed previously, the thermoelectric power is given by
&
k{
4
2-5 + In|
(v)]^-^"Vn
ln
(y).
PM,
(74)
PP,
where k
is
and
A^c
and
states in the
This means that the thermoelectric power is negative for -type semiconductors, and positive for /?-type semiconductors. This fact is often used in
54
A Resume
^ of/?-type
and
1.7
determining the conduction type of a semiconductor. The thermoelectric power can also be used to determine the position of the Fermi level relative
to the
band
edges.
At room temperature
resistivity: 1
the thermoelectric
results (except a
power
silicon increases
with
mV/K
mV/K
for a
k,
Another important quantity in thermal effect is the thermal conductivity, ~s for both electrons and holes, is given by which, if t ~ E
KL
+
(nn
+ pn?
(75)
9 10
Fig. 27
Measured absorption coefficients near and above the fundamental absorption edge Si, and GaAs. (After Dash and Newman, Ref. 54; Philipp and Taft, Ref. 55; and Hill, Ref. 56.)
55
1000
1
-"
100
/
//
ys' /
1
/
\diamond
VTYPE1I)
<"~f ^n\ \
"^
/
/ Go As I
1 / *
/
i
\
'Si
Cu|
\\
1 I
1
1
1
\\
Si\
w
1
VN 4Ge\
Ge/
GoAs^' \
10
100
300
1000
T CK)
Fig. 28 Measured thermal conductivity vs. temperature for pure Ge, Si, GaAs, Cu, and diamond type II. The thermal conductivity is lower than indicated here for samples with
(After Carruthers et
Ref. 61.)
al.,
White,
Ref. 60;
and Berman et
al
The
first, second, and third term on the right-hand side of the above equation represent the lattice contribution, the electronic contribution, and the contributions due to mixed conduction respectively. The contributions of conduc-
term, however,
tion carriers to the thermal conductivity are in general quite small. The third may be quite large when > kT. It is expected that the
Eg
thermal conductivity will first increase with T at low temperatures and then decrease with temperature at higher temperatures.
conductivity
is
57-59
and GaAs
shown
room temperature
values
56
A Resume
ties
shown in Fig. 28 are the thermal conductividiamond type II. Copper is the most commonly used metal for thermal conduction in p-n junction devices; diamond type II has the highest room-temperature thermal conductivity known to date and is useful as the thermal sink for junction lasers and IMPATT oscillators to be
are listed in Table 2.2. Also
60 61
'
for
Cu and
for
discussed later.
(4)
High-Field Property
in
is
As discussed
semiconductor
constant
is
fields,
field,
which
is
independent of the
When
and
in
some
At
still
larger fields,
impact ionization
We
is
carriers
zero.
both emit and absorb phonons, and the The energy distribution at thermal
field the carriers
and lose it to phonons by emitting more phonons than are absorbed. At reasonably high fields the most frequent scattering event is the emission of optical phonons. The carriers thus, on the average, acquire more energy than they have at thermal equilibrium. As the field increases, the average energy of the carriers also increases, and they
acquire energy from the field acquire an effective temperature
ture T.
Te
which
is
From
the rate equation such that the rate at which energy fed from
the field to the carriers must be balanced by an equal rate of loss of energy
to the lattice,
we
obtain for
Ge and
1
Si:
3tt
T
and
+
8
kA
\
l/2\
(76)
where ^
sound.
is
the electric
field,
and
the velocity of
When
<^
we have
stant value at
low
fields,
and Eqs.
and
(77) reduce to
2
3tc/Mo*\
32
(78)
57
and
i>-mn
When
Finally at sufficiently high fields the drift velocities for
scattering-limited velocity:
(79)
the field is increased until n g ~ SCJ3, the carrier temperature has doubled over the crystal temperature and the mobility has dropped by 30%.
Ge and
Si
approach a
Uw
8p
3ttWq
10
cm/sec
Table
(79a)
where
Ep is
the optical
phonon energy
(listed in
2.3).
TABLE
PROPERTIES OF Ge,
Properties
Si,
2.3
AND
Ge
4.42
Si
GaAs
2.21
Atoms/cm
x 10 22
5.0
10 22
10 22
Atomic Weight
72.6
28.08
144.63
Breakdown Field
(F/cm)
Crystal Structure
~10 5
Diamond
5.3267
~3 x
10 5
~4 x
10 5
Diamond
2.328
11.8
Zincblende
5.32
10.9
Density (g/cm 3 )
Dielectric Constant
Effective Density of
16
1.04
x 10 19
2.8
x 10 19
4.7
x 10 17
States in Valence
Band, AVCcm" 3 )
Effective
6.1
x 10 18
1.02
x 10 19
7.0
x 10 18
Mass m*/m
w,*
Electrons
Holes
Electron Affinity, x(V)
= 1.6 w *= 0.082 mf = 0.97 m * = 0.19 w& = 0.04 mth = 0.3 mf =0.16 /?& = 0.5
t t
fc
0.068
0.12, 0.5
4.0
4.05
4.07
Energy
Gap
(eV) at
300K
0.66
1.12
1.43
58
A Resume
TABLE
Properties
2.3 (Cont.)
Ge
Si
GaAs
Intrinsic Carrier
Concentration
(cm" 3 )
Lattice Constant (A)
2.4
x 10 13
1.6
x IO 10
1.1
x 10 7
5.65748
5.43086
5.6534
Linear Coefficient of
Thermal Expansion
AL/LA7TC-
5.8
x 10- 6
937
2.6
x 10" 6
1420
5.9
x 10- 6
1238
Minority Carrier
Lifetime (sec)
io- 3
2.5
x 10" 3
~10~ 8
Mobility (Drift)
(cm 2 / F-sec)
jjl
(electrons)
/x p
(holes)
3900 1900
1500
600
8500 400
0.037
0.063
0.035
Heat
0.31 0.7
(Joule/gC)
0.35
0.64
1.45
0.46
0.36
0.9
0.44
1 at 1050C 100atl220C
Vapor Pressure
(torr)
Work
Function (V)
4.7
velocity-field, relationship is
more complicated.
We
must
GaAs as shown in Fig. 5. There is a highmobility valley (n 4000 to 8000 cm 2 /V-sec) located at the Brillouin zone 2 center, and a low-mobility satellite valley (/z 100 cm /V-sec) along the
consider the band structure of
eV higher
in energy.
The
is
effective
0.068
in the
in the
(12).
As the
59
GaAs. The
will
intervalley transfer
mechanin
Chapter
14.
The experimentally deduced drift velocity is plotted 62 63 in Fig. 29 for Ge, Si, and GaAs. We note that the drift velocities for the three semiconducapproach 10 7 cm/sec. For GaAs there is a region of differential negative mobility between 3 x 10 3 to 2 x 10 4 V/cm. We next consider impact ionization. When the electric field in a semiconductor is increased above a certain value, the carriers gain enough energy that they can excite electron-hole pairs by impact ionization. The electrontors all
is
given by
G=
a n nii n
at
pfi
(80)
(After Seidal and Scharfetter, Ref. 62; Norris and Gibbons, Ref. 62a; Duh and Moll, Ref. 62b; and Ruch and Kino, Ref. 63.)
60
A Resume
Get<pyy
{/ /
Si (*)/
/3e(<n)
/ 5
io'
/
V
/
GaAs/
fVV/
/si( p
j
/
'
~-
GaP/
f
evvy
1.5
4
(x IO 5
10
ELECTRIC FIELD
Fig. 30
V/cm)
Ge,
Measured ionization coefficient for avalanche multiplication vs. electric field for GaAs. and GaP. (After Miller, Ref. 64; Lee et al., Ref. 65; Logan and Sze, Ref. 66; and Logan and White,
Si,
Ref. 67.)
where
cc is
number of electron-hole
by an electron per unit distance traveled. Similarly a p is the analogously defined ionization rate for holes. Both a and a p are strongly dependent on the electric field. Figure 30 shows the experimental results of 64-67 the ionization rates for Ge, Si, GaAs, and GaP. These results are all obtained by the use of photomultiplication measurements on p-n junctions. It
pairs generated
6
is
of
Semiconductors
61
from the
or
"'MfJ
A
(81)
where a and
fit
<? are constants, and n, in general, varies from 3 to 9. better to the ionization rates over a wider range of field can be given by the
following expression
a
^exp[-(Z)/(f) m ].
(82)
values of A, b, and m are given in Table 2.4. has been shown that the dependence of the ionization rate on the electric field and on the lattice temperature can be expressed in terms of a modification of Baraff's three-parameter theory. 68,69 The parameters are: /? the ionization threshold energy; A, the optical phonon mean free path; and (Ep), the average energy loss per phonon scattering. The values of E for
The
It
best
fit
and A and (E
> are
given by
tanh(A.)
(83)
<>
= p
tanh^)
(84)
and
= <Ep>
^o
Ep
(85)
Ep is the optical phonon energy (listed in Table 2.4), and A is the high-energy low-temperature asymptotic value of the phonon mean free path.
where
Baraff's result is shown in Fig. 31 where od is plotted against EjjqSA, with the ratio of average optical-phonon energy to ionization threshold energy, <.Ep}/Ej as a parameter. Since for a given set of ionization measurements the values of Ej( = $E ), ct, and its field dependence are fixed, one can thus obtain g the optical-phonon mean free path, A, by fitting the ionization data to the
,
Baraff plot.
(
A typical result is shown in Fig. 31 for Ge at 300K. The value of Ep>IEi is 0.022. We obtain 64 A for the room-temperature electron-phonon
mean free path, and 69 A for the hole-phonon mean free path. Similar results have been obtained for Si, GaAs, and GaP. From the room temperature data one can obtain the value of A from Eq. (84). The average value of A and A (at 300K) are listed in Table 2.4. Once we know the value A we can predict
<
Oh
o in o
in
in
o
X
Tf
O
X
oo
fS
-H
^r
< O
V)
m m O
o"
>n
V)
o ~H
X
en
<o
1
o
'
-H in
-H 00 in
(S
f-
IT)
<1>
in
O *
X
in
VD
o
X
VO
"o DC
<* ro ^O
-H
-W
in in
CW
o o
c o
4)
V0
>c
>n
in
o
x
oo CO
O
X
in f-
-H es
V>
-H VO r-
1)
O
X
o
X
00
"o
a
r-
O
-H
in
o
-H
in o
r~
O d
c o ~ o u
-H
O
X
in in
\o
X
v>
in
?
E o S
-5"
O o 3 a o o
's
C/3
S-i
^ <
>
o ro
^<
2
o
<-<
^^
c
2
^ b
E o
1
o
1
5
ctf
II
62
63
2x10"'.
Fig. 31
vs.
Ej/qtfA
<Ep>\.i
results.
plot product of ionization rate and optical phonon mean free path where E, = f Eg and g is the electric field. The running parameter is the where < p > is the average optical phonon energy. The solid curves are
Baraff's
(aA)
ratio
theoretical
= 64A
the values of A at various temperatures and from the temperature dependence of <}, Eq. (84), we can choose the correct Baraff plot. The
;
theoretical
predicted electron ionization rates in silicon as obtained from the above approach are shown in Fig. 32. Also shown are the experimental results
at
64
A Resume
I0 :
S^ ^o**
-
\\
\
\ ^ \\\ f\ \\a
v
X \
\\
s
\
Si
t=ioo , k
EXPERIMENTAL
O n
100 K
\\\ V'\ \
\
- 2I3K - 300K
\
\
\300\
\
A
-
THEORETICAL
I00K 2I3"K
\ \ \
\
\
\
\
5xl0 6
400K
400^
| |
\/e
(v/cm)"
Fig. 32
vs.
Si.
The agreement
is
satisfactory.
To
facilitate
numerical analysis, the curves as shown in Fig. 31 can be represented by the 69 following approximation:
2
[(11. 5r
1.17r
2
4 2 3.9 x 10~ )x
od
exp
{
(86)
45
where
r
= <EP >/Ej
and
= E^qSk.
The
+2%
<r<
0.06 and 5
<x<
16.
Basic Equations 70
The basic equations for semiconductor device operation are those which describe the static and dynamic behaviors of carriers in semiconductors under the influence of external fields which cause deviation from the thermalequilibrium conditions. The basic equations can be classified in three groups: the Maxwell equations, the current density equations, and the continuity equations.
A.
m Vx<f=_^-.
+
"a7"
(87)
Jcond
Jtot
'
88 >
S>
=
'
p(x, y, z),
(89)
^ = 0,
-
(90)
(91)
= HoJ?,
^0>
e,(*
t')S(r,
O dt'
(92)
ao
where Eq. (92) reduces to 3) = zs g under static or very low frequency conditions, and Q) are the electric field and displacement vector respectively, jT and are the magnetic field and induction vector respectively, e s and ^ujare the permittivity and permeability respectively, p(x, y, z) the total electric
charge density, J cond the conduction current density, and J tot the total current density (including both conduction and convection current components and
the above six equations the most important is the (89), which determines the properties of the p-n junction depletion layer (to be discussed in the next chapter).
Jtot
0)-
Among
66
A Resume
(93)
B.
qH p p-qD p Vp,
J
(94)
(95)
+ JP
Where
and J p are the electron current density and the hole current density component due to field and the diffusion component due to carrier concentration gradient. The values of the electron and hole mobilities (// and pc p) have been given in Section 5(1). For nondegenerate semiconductors the carrier diffusion constants (D and D p ) and
J
(D n =
(kT/q)n,
etc.).
,.
nS
+ lD n - = Wn {nS +
I
dn
kT -),
dn\
,n _ (93a)
.
J,
_,,,,-, B, J! -,,(,*
dp
-_-).
kT
dp\
: nA (94a)
.
The above equations do not include the effect due to externally applied magnetic field. With an applied magnetic field, another current of J tan 0 and J pl tan 6 P should be added to Eqs. (93) and (94) respectively, where J x is the current component of J perpendicular to the magnetic field and
tan 9 n
=qnnR H \Jf\
RH
is
(which has negative value because the Hall coefficient similar results are obtained for the hole current.
;
C.
Continuity Equations
^ = G-/ + -V.J q
at
(96)
S J.
t
Gp
-U,-U. 3p
(97)
where G and G p are the electron and hole generation rate respectively -1 -3 sec (cm ) due to external influence such as the optical excitation with high energy photons or impact ionization under large electric fields. Un is the electron recombination rate in /?-type semiconductors. Under low injection
conditions
(i.e.,
when
is
much
less
brium
(n p
by the expressions
n po )fr where n p is the minority carrier density, n po the thermalequilibrium minority carrier density, and t the electron (minority) lifetime. There is a similar expression for the hole recombination rate with lifetime x p
.
67
in pairs
If the electrons
with no
trapping or other
xp
For the one-dimensional case under a low-injection condition, Eqs. and (97) reduce to
dn P
(96)
r r
np
n po
di
dn p
2 n d np
d Pn
Pn
Pno
dS
8 Pn
2 ^ d Pn
(2)
Simple Examples
Consider an -type sample, as illuminated with light and where the electronhole pairs are generated uniformly throughout the sample with a generation rate G. The boundary conditions are S = 0, and dpjdx = 0. We have from Eq. (97a):
A.
shown
in Fig. 33(a),
??
dt
Pn~ r, = g-
( 98 )
At steady
state,
dpjdt
0,
and
Pn
If at
=Pno +
G=
constant.
(99)
t = 0, the light is suddenly turned off, the boundary conditions are p H {0) = Pno + Xp G as given in Eq. (99) and oo) = p no p n (t The differential equation is now
d Pn
jt=
and the solution
is
+
^P
Pn-Pno
/f
(
vr~
Ge- tlx ".
io )
Pnit)
Pno
(101)
The variation of p n with time is shown in Fig. 33(b). The above example presents the main idea of the Stevenson-Keyes method for measuring minority carrier lifetime. 49 A schematic setup is shown in Fig. 33(c). The excess carriers generated uniformly throughout the sample by the light pulses cause a momentary increase in the conductivity. The in-
when a constant passed through it. The decay of this photoconductivity can be observed on an oscilloscope and is a measure of the lifetime. (The pulse width must be much less than the lifetime.)
current
is
68
A Resume
hv
(b)
LIGHT
PULSE
il-UU
I
>
>
'>
"
K
I
(c)
rh
Fig. 33
(a)
(b)
(c)
Stevenson-Keyes experiment.
B.
Steady-State Injection
From One
Side.
other simple example where excess carriers are injected from one side
we note
that for hv
3.5
eV
by
l/e in a
69
INJECTING'
SURFACE
\U/r
(b)
Fig. 34
(a)
one
side.
semi-infinite sample
(b)
W
near the surface. The
At steady
differential
equation
from Eq.
(97a),
d 2* =
dt
= -
Pn
Pno
+D
82p"
dx'
(102)
The boundary conditions are pn (x = 0) = (0) = constant value, pn pn (x -> co) = pno The solution of pn (x) which is shown in Fig. 34(a), is
.
and
Pn(x)
= pno +
L Pn (0)
Pno -]e-
xlL "
(103)
where
Lp = ^[Dpt~ p
The maximum
values of
L and
70
A Resume
but only of
L n ( = ^/D n x n)
If the
cm
is
,
in
germanium and
silicon,
the order of 10
cm
in gallium arsenide.
no
then
changed so that all excess carriers at we obtain from Eq. (102) a new
solution,
W -x
sinh
Pnfr)
Pno
[P(0)
Pno]
(104)
sinh
E
density at x
The above
result
is
shown
in Fig. 34(b).
The current
is
JP
= -4 D p
D>(0)
Dp
PnoJ
(105)
sinh
localized
light
It will
in junction transistors
is
C.
Diffusion.
When
d pn
(106)
no
field
8=
0, the solution is
given by
P(x,
N
y/4nD p t
exp
4D p t
tp
Pn
(107)
where
solution
diffused
number of electrons or holes generated per unit area. The is shown in Fig. 35(b), from which it can be seen that the carriers away from the point of injection, and that they also recombined.
is
the
If an electric field is applied along the sample, the solution is in the form of Eq. (107) but with x replaced by (x - n p t) as shown in Fig. 35(c). This means that the whole " package " of excess carrier moves toward the negative end of the sample with the drift velocity \i p $. At the same time, the carriers
outward and recombine as in the field-free case. The above example is essentially the celebrated Haynes-Shockley experiment 71 for the measurement of carrier drift mobility in semiconductors. With known sample length, applied field, and the time delay between the applied
diffuse
71
hi/
'-^
n-TYPE
(a)
PULSE GENERATOR
(5.
(c)
Fig. 35
(a)
Transient and steady-state diffusion Haynes-Shockley experiments (After Haynes and Shockley, Ref. 71.)
(b)
(c)
electric pulse
oscilloscope),
one
mobility
(/i
xjSt).
Surface Recombination. 36 When one introduces surface recombination at one end of a semiconductor sample (Fig. 36), the boundary condition at x = is given by
D.
=
ox
QSJp(0)
- pno ]
(108)
72
A Resume
SURFACE RECOMBINATION
kv
\\<\\\\)\
I
-n-TYPE
M
(0)
Fig. 36
is
Surface recombination at x
= 0.
The minority
which
which reach the surface recombine has the dimension cm/sec is defined as the surface recombination velocity. The boundary condition at x = oo is the same as that for example A. The differential equation is
states that the minority carriers
Sp which
Pn~
dt
Pn
'
d Pn
p
dx
(109)
The
boundary conditions
is
Pn(x)
iP
Tp
S p exp(-x/L p )
-kp
~^~
(110)
Tp
which is plotted in Fig. 36 for a finite Sp When Sp - 0, p(x) -> p n0 + x p G, the same as obtained previously (example A); when Sp ^> co, p n (x)-^p no + x p G[l -exp(-x/L p)], and the minority carrier density at the surface approaches its thermal equilibrium value p no Analogous to the low-injection bulk recombination process, in which the reciprocal of the minority carrier the surface recombination velocity is lifetime (1/r) is equal to a p v th
.
.
given by
S p = o p v th Nst
where Nst is the number of surface trapping centers per unit area boundary region.
(111)
at the
References
73
REFERENCES
1.
W.
C. Dunlap,
An
Introduction to Semiconductors,
John Wiley
&
Sons
New York
(1957).
2.
3.
4.
5.
See for example, C. Kittel, Introduction to Solid State Physics, John Wiley New York, 2nd ed. (1956), 3rd ed. (1966).
C. Hilsum and A. C. Rose-Innes, Semiconducting
III-
&
Sons
6.
New York
7.
V Compounds, Pergamon
2,
Press
(1961).
R. K. Willardson and A. C. Beer, Ed. Semiconductors and Semimetals, Vol. of III-V Compound, Academic Press, New York (1966).
Physics
8.
W.
B. Pearson,
Alloys,
Pergamon Press
9.
L. Brillouin,
Wave Propagation
New York
10. 11. 12.
(1953).
F. Seitz, The
J.
Modem
New York
(1940).
M. Ziman,
Principles
M. L. Cohen, "Pseudopotential Calculations for II-VI Compounds," a chapter of II-VI Semiconducting Compounds, Edited by D. G. Thomas, W. A. Benjamin Inc New York, p. 462(1967).
C. Kittel,
13.
&
Sons,
New York
(1963).
14.
C. Herring,
for Calculating
Wave Functions
Silicon
in Crystals "
Phys
& Germanium
Phys Rev
'
" Proc
Scheme
for
Energy Bands
in Solids,"
98 993
"
'
(1955).
16.
J.
C. Phillips, "Energy-Band Interpolation Scheme Based on a Pseudopotential " Phys. Rev., 112, 685 (1958).
17.
M.
L.
Cohen and
T.
Factors for Fourteen Semiconductors of the Phys. Rev., 141, 789 (1966).
18.
K. Bergstresser, "Band Structures and Pseudopotential Form Diamond and Zinc-blende Structures "
Hughes
18a.
Biblio-
M. Neuberger, Silicon Data Sheets DS-137 (May 1964) and Silicon Bibliographic Supplement, DS-137 (Supplement) (July 1968).
74
18c.
SemiconductorsA Resume
Arsenide Data Sheets, DS-144 (April 1965) and Gallium Supplement, DS-144 (Supplement 1) (Sept. 1967). Arsenide Bibliographic
19.
J.
Electrons and Phonons, Oxford at the Clarendon Press (1960). Ed., Solids Under Pressure,
20.
Inc.,
21.
S.
Blakemore,
Elec.
22.
Commun.,
Morin and
F.
J.
Boron," Phys. Rev., 96, 28 (1954); also "Conductivity Range of Germanium," Phys. Rev., 94, 1525 (1954).
23.
and Hall
R. N. Hall and
Intrinsic
J.
Germanium,
H. Racette, "Diffusion and Solubility of Copper in Extrinsic and Silicon, and Gallium Arsenide," J. Appl. Phys., 35, 379 (1964).
18, 253 (1964).
24. 25.
P.
" Pseudopotential Theory of Exciton and Impurity J. Hermanson and J. C. Phillips, States," Phys. Rev., 150, 652 (1966).
J.
26.
Callaway and A.
J.
in
E.
M. Conwell,
Germanium, Part
II," Proc.
IRE,
46, 1281
(1958).
28.
S.
M. Sze and
Si at
J.
and
29.
30.
GaAs, Ge,
W. M.
W.
Bullis, "Properties of
Gold
Academic Press
(1952).
30a. P.
G. Shewmon, Diffusion
I.
in Solids,
New York
(1963).
30b. B.
30c.
Boltaks, Diffusion in
(1963).
H.
S.
Carslaw and
J.
Press (1959).
31.
"Ion Implantation in Semiconductors Part Distribution Theory and Experiments," Proc. IEEE, 56, 295 (1968).
For a review
see
J.
F. Gibbons,
I,
Range
32.
Prentice-Hall Inc.,
Englewood
Cliffs,
New
Jersey (1967).
33.
D. L. Kendall,
in
Diffusion in III-
InSb, Rep. No. 65-29, Dept. of Material Science, Stanford University, Stanford
(August 1965).
34.
System Tech.
J.,
35.
K. B. Wolfstirn, "Holes and Electron Mobilities in Doped Silicon From Radio Chemical and Conductivity Measurements," J. Phys. Chem. Solids, 16, 279 (1960).
A.
S.
36.
Grove, Physics and Technology of Semiconductor Devices, John Wiley and Sons,
(1967).
New York
37.
J.
in
Nonpolar
References
38.
75
E. Conwell and V. F. Weisskopf, "Theory of Impurity Scattering in Semiconductors " Phys. Rev., 77, 388 (1950).
39.
in
40.
M.
I,
(1953).
41.
W. W.
Co.
Van Nostrand
42.
L. B. Valdes, "Resistivity
42, 420 (1954).
Measurement on Germanium
IRE
43.
F.
M.
Smits,
"Measurement of Sheet
44.
D. B.
Cuttriss, "Relation
in Diffused
45.
C. Irvin, "Resistivity of Bulk Silicon and of Diffused Layers in Silicon," Bell Syst Tech. J., 41, 387 (1962).
J.
46.
E.
H. Hall, "On a
New Action
Am
287(1879).
47.
Math '' 2
'
L. J. Van der Pauw, "A Method of Measuring Specific Resistivity and Hall Effect of Disc or Arbitrary Shape," Philips Research Reports, 13, No. 1, p. 1-9 (Feb. 1958).
48.
in
C. T. Sah, R. N. Noyce, and W. Shockley, "Carrier Generation and Recombination p-n Junction and p-n Junction Characteristics," Proc. IRE, 45, 1228
(1957).
48a. R.
in
(1952).
"
87 387
'
48b.
W. Shockley and W.
T.
in
Germanium and
50.
W. W. Gartner, "Spectral Distribution of the Photomagnetic Electric Effect," Phys. Rev., 105, 823 (1957). And also see other references listed in Ref. 41.
B.
51.
N. Brockhouse and
P.
53.
J.
L. T.
Waugh and G.
Dolling, "Crystal
W.
C. Dash and R. Newman, " Intrinsic Optical Absorption in Single-Crystal Germanium and Silicon at 77K and 300K," Phys. Rev., 99, 1151 (1955).
55.
in the Region 1 to 10 eV," Phys. Rev., 113, 1002 (1959), also "Optical Constants of Silicon in the Region 1 to 10 eV," Phys. Rev. Letters, 8, 13 (1962). Hill,
56.
D. E.
A866
(1964).
76
57.
J.
A Resume
Conductivity of
A. Carruthers, T. H. Geballe, H. M. Rosenberg, and J. M. Ziman, "The Thermal Germanium and Silicon between 2 and 300K," Proc. Roy. Soc,
(1957).
in Silicon," Proc. Inter-
Academic
Press,
New York
59.
M. G. Holland, "Phonon
Thermal Conductivity
Electrical Conductivity of
Copper
at
Low Tempera-
61.
R. Berman,
"The Thermal Conductivity of Diamond Roy. Soc. (London), Ser. A, 200, 171 (1953).
Low
Temperatures," Proc.
62.
T. E. Seidel
Field
and D. L. Scharfetter, "Dependence of Hole Velocity upon and Hole Density for p-type Silicon," J. Phys. Chem. Solids, 28, 2563
Electric
(1967).
62a. C. B. Norris
in Si
and J. F. Gibbons, " Measurement of High-Field Carrier Drift Velocities by a Time-of-Flight Technique," IEEE Trans. Electron Devices, ED-14, 38
(1967).
62b. C. Y.
Duh and
J.
IEEE
63.
J.
G. Ruch and G. S. Kino, "Measurement of the Velocity-Field Characteristics of Gallium Arsenide," Appl. Phys. Letters, 10, 40 (1967).
S. L. Miller,
64.
65.
Kleimack, and
W. Wiegmann,
"Ioniza-
ation Rates of Holes and Electrons in Silicon," Phys. Rev., 134, A761 (1964).
66.
"Avalanche Multiplication in Ge and GaAs/?-/; Junctions," on the Physics of Semiconductors, Kyoto, and J. Phys. Soc. Japan Supplement Vol. 21, p. 434 (1966).
R. A. Logan and
S.
M.
Sze,
67.
R. A. Logan and H. G. White, "Charge Multiplication Appl. Phys., 36, 3945 (1965).
in
GaP
p-n Junctions,"
J.
68.
in
69.
C. R. Crowell and S.
in
M.
Sze,
242 (1966).
70.
W.
J.
Semiconductors, D.
Inc.,
Princeton (1950).
71.
R. Haynes and
W.
Shockley,
"The
trons in
72.
Y.
P. Varshni,
Gap
in Semiconductors,"
M.
B. Panish
and H. C. Casey,
J.
Jr.,
in