Physics and Technology of Semiconductor Devices - As GROVE

<)
' O
X,'j
C)
c >
Physics and Technology of

Semiconductor Devices
A. S. G R O V E
Intel C o r p o r a t i o n , M o u n t a i n V i e w
University of C a l i f o r n i a , B e r k e l e y
V
JOHN WILEY & SONS
New Y o r k C h i c h e s t e r B r i s b a n e T o r o n t o S i n g a p o r e
( >
Preface
Copyright 1967 by John Wiley & Sons, Inc.
Reproduction or translation o any pan of this work beyond thai
permitted by Sections 107 or 108 of the 1976 United States Cop>riehi Act without the permission of the copyright owner is unlawful. Requests for permission or further information should be
addressed 10 the Permissions Department. John Wiley & Sons. Inc.
Library of Congress Catalog Card Number: 67-17340

Printed in the United States of America
23 24 25 26 27 2E 29 30
The 1950's was the decade in which semiconductor devices rose to

prominence and attained great industrial significance. I n that decade,
most o f the research and engineering w o r k was directed to germanium.
The 1960's can be considered as the decade in which silicon semiconductor
devices and integrated circuits made by the planar technology o v e r t o o k
germanium devices.
The purpose o f this b o o k is to provide an i n t r o d u c t i o n to the physics
and technology o f planar silicon devices, i.e., devices made by the planar
technology. T o be sure,-the physical principles u n d e r l y i n g the f a b r i c a t i o n
and the operation o f these devices do not differ from those underlying the
fabrication and the operation o f devices made from other semiconductors
by other technologies. However, problems viewed i n general terms tend
to be very difficult to solve. T o render a p r o b l e m tractable, we must
concentrate o n its most important features. W h a t is i m p o r t a n t for an
alloy germanium diode, for instance, may not be i m p o r t a n t for a planar
silicon diode. A s a result, approximations which describe well the characteristics o f the former often give results i n very p o o r agreement w i t h
observations on the latter. M o s t i f not all books i n the past have tacitly
concentrated on those factors that are i m p o r t a n t i n relation to g e r m a n i u m
alloy devices. I n this one. the emphasis is on those w h i c h are i m p o r t a n t
i n dealing w i t h planar silicon devices.
After a brief description o f the planar technology, the text is divided
into three parts. The first one deals w i t h those processes o f solid-state
VI
^retace
technology which are most intimately involved in determining the electrical

characteristics ot" semiconductor devices. The three processes discussed
vapor-phase g r o w t h , thermal o x i d a t i o n , and solid-state diffusionare the
means by which the semiconductor crystal is turned into a device o f
desired characteristics.
The second part deals with semiconductors and semiconductor
devices.
A summary o f the most important results o f the band theory o f solids is
followed by a discussion o f semiconductors under n o n - e q u i l i b r i u m conditions. Then p-n junctions, which are the b u i l d i n g block o f most
i m p o r t a n t semiconductor devices, are considered in detail followed by a
treatment o f j u n c t i o n transistors and j u n c t i o n field-effect transistors.
One o f the principal advantages o f the planar technology is that it
results i n excellent c o n t r o l o f the semiconductor surface. This c o n t r o l
has made the surface a designable part o f semiconductor devices and has
also led to the r a p i d development o f new types o f devices. One o f these,
the surface field-effect transistor, is already second in importance only to
the j u n c t i o n transistor among active devices. Thus the t h i r d part o f this
text considers surface effects and surface-controlled
devices. This part
includes discussion o f the theory o f semiconductor surfaces, surface effects
o n p-n junctions, the s'urface field-effect transistor, and the status o f the
silicon-silicon dioxide system.
The text is principally intended for use by senior undergraduate or
first-year graduate students in electrical engineering and in materials
science. For this reason I have attempted to keep the most i m p o r t a n t
phvsical principles always in the forefront and to use simple physical models
wherever possible. However, I have also attempted to make the b o o k
remain useful to the student when he becomes a practicing engineer or
scientist. To serve this purpose, principles are generally illustrated w i t h
actual numerical computations or w i t h experimental measurements.
T h u s , in a d d i t i o n to boosting the readers' confidence in these principles,
the illustrations also contain numbers which, at a later time, can be used
in design calculations.
The dual nature o f this book was also helped by the way it developed.
The material contained in it is based on a series o f lectures given at
F a i r c h i l d Semiconductor to scientists and engineers who were engaged i n
research, development, manufacturing, and application w o r k involving
semiconductor devices and integrated circuits, and o n a graduate course
I taught at the University o f California, Berkeley. The material, i n
manuscript f o r m , was then used in the Fairchild course in the f o l l o w i n g
vear, as well as by Professors C. A . Mead, California Institute o f Technology, and P. O. Lauritzen, University o f Washington, in their undergraduate electrical engineering courses.
I n the writing o f this book I benefited very significantly from the manv
suggestions o f the reviewers: D."J.' Fitzgerald, E. H . Snow, L . Vadasz,
and R. J. Whittier o f Fairchild Semiconductor; and Professors P. O.
Lauritzen, University o f Washington, C. A . M e a d . California Institute o f
Technology, J. L . M o l l , Stanford University, and W. G . O l d h a m , U n i versity o f California, Berkeley. I am further indebted to J. P. Bissell for
his assistance in evaluating device characteristics, and to Miss S. J.
Steele and M r s . D o r o t h y . U n r u h for their help w i t h the calculations
and illustrations. Finally, I w o u l d like to express my appreciation to the
management o f Fairchild Semiconductor for providing an environment i n
which the w r i t i n g o f this book was possible.
A . S. G R O V E
Palo Alto,
California
February 1967
Contents
Introduction:
T h e Planar Technology
1.
Chapter I
SOLID-STATE
TECHNOLOGY
Vapor-Phase Growth
1.1 Kinetics o f G r o w t h
1.2 Gas-Phase Mass-Transfer
1.3 Some Properties o f Gases
Chapter 2
7
10
13
18
Thermal Oxidation
22
2.1 Kinetics o f Oxide G r o w t h

2.2 Space-Charge Effects on O x i d a t i o n
23
31
Chapter 3
Solid-State Diffusion
35
3.1 F l u x
3 . 2 . T h e Transport Equation
3.3 Diffused Layers
3.4 Deviations f r o m Simple Diffusion Theory
3.5 Redistribution o f Impurities during Thermal O x i d a t i o n .
3.6 Diffusion through a Silicon D i o x i d e Layer (Oxide M a s k i n g )
3.7 T h e Redistribution o f Impurities i n E p i t a x i a l G r o w t h
.
ix
36
41
43
58
69
75
78
II. S E M I C O N D U C T O R S
Chapter 4
4.1
4.2
4.3
4.4
4.5
AND
SEMICONDUCTOR
E l e m e n t s of S e m i c o n d u c t o r
DEVICES
Physics
The Band Theory o f Solids

Electrons and Holes in Semiconductors
Fermi-Dirac Distribution Function . . .
I m p o r t a n t Formulas for Semiconductors
Transport o f Electrons and Holes
91
in E q u i l i b r i u m
91
95
98
100
106
Chapter 8
5.1
5.2
5.3
5.4
5.5
Semiconductors
Conditions
III. S U R F A C E
117
117
127
134
136
140
7.1
7.2
7.3
7.4
7.5
7.6
7.7
p-n Junctions
Electrostatics
Space-Charge Region for Step Junctions
Space-Charge Region for Linearly Graded Junctions
Space-Charge Region for Diffused Junctions
Capacitance-Voltage Characteristics
Current-Voltage Characteristics
Junction Breakdown
Transient Behavior
Chapter 7
244
248
253.
AND
SURFACE-CONTROLLED
DEVICES
T h e o r y of S e m i c o n d u c t o r Surfaces
263
9.1 Characteristics o f Surface Space-Charge RegjonsEquil i b r i u m Case

9.2 The Ideal M I S (or M O S ) Structure
9.3 Effect o f W o r k Function Difference, Charges, and States
on M O S Characteristics .
C h a p t e r 10
6.1
6.2
6.3
6.4
6.5
6.6
6.7
6.8
EFFECTS
. . .
under N o n - E q u i l i b r i u m
Injection
Kinetics o f the Recombination Process
L i f e t i m e i n Low-Level Injection
Surface Recombination
Origin o f Recombination-Generation Centers
Chapter 6
243
8.1 Principles o f Operation

8.2 Characteristics o f Junction Field-Effect Transistors
8.3 Modifications o f the Simple Theory
Chapter 9
Chapter 5
Junction Field-Effect T r a n s i s t o r s
149
. .
Junction T r a n s i s t o r s
Principles o f Transistor A c t i o n
'. . .
Currents Flowing in a Transistor; Current G a i n . . . .
Limitations and Modifications o f the Simple Theory . .
Base Resistance
M a x i m u m Voltage L i m i t a t i o n s
M i n i m u m Voltage L i m i t a t i o n s
Thermal L i m i t a t i o n
152
153
163
167
169
172
191
201
208
209
214
222
228
230
234
236
Surface Effects on p-n Junctions
11.1
11.2
11.3
11.4
Surface Field-Effect T r a n s i s t o r s
P r o p e r t i e s of the S i l i c o n - S i l i c o n
System
12.1 Fast Surface States

12.2 Space Charge w i t h i n the Oxide
290
296
298
305
311
311 -
Principles o f Operation
Characteristics o f Surface Field-Effect Transistors . . . .
Modification o f the Simple Theory
Other Types o f Surface Field-Effect Transistors . . . . . .
C h a p t e r 12
278
289
10.1 Characteristics o f Surface Space-Charge RegionsNonEquilibrium- Case

10.2 Gate-Controlled Diode Structure
10.3 Recombination-Generation Processes i n the Surface SpaceCharge Region
10.4 Field-Induced Junctions and Channel Currents
10.5 Surface Effects on Junction Breakdown Voltage. . . . .
C h a p t e r II
264
271
318
321
327
329
Dioxide
334
335
337
Contents
12.3
12.4
12.5
12.6
Surface-State Charge
. . . .
Barrier Energies
Surface M o b i l i t y
Conduction on Oxide Surfaces
12.7 Other Insulators
List of Symbols
357
Index
i m p u r i t y concentration gradient at the j u n c t i o n .
coefficient in the general relationship for the thermal o x i d a t i o n

o f silicon
cross-sectional area o f metallurgical p-n j u n c t i o n
depleted surface area
Aj
A
s
B
B
BV
BV
BV
CBQ
CE0
C
C
C*
C
C
B
C
C
C,
C
C
C
C.
C
G
Bllli
coefficient in the general relationship for the thermal o x i d a t i o n

o f silicon;, also referred to as parabolic rate constant
transistor base factor
j u n c t i o n breakdown voltage
collector-base j u n c t i o n breakdown voltage, w i t h emitter open
collector-to-emitter breakdown voltage, w i t h base open
concentration
capacitance per unit area
e q u i l i b r i u m concentration o f oxidant i n oxide film
b u l k concentration
concentration o f external doping i m p u r i t y at the surface o f an
epitaxial film
concentration of a species in the b u l k o f the gas
t o t a l gate capacitance
concentration o f the oxidant at the oxTde-silicon interface
concentration of the oxidant at the gas-oxide interface
capacitance per unit area o f oxide layer
surface concentration
capacitance per unit area o f the surface space-charge region
i m p u r i t y concentration w i t h i n the substrate in epitaxial g r o w t h
List 01 aymoois
total concentration o f moiecules in a gas

total concentration o f ionized impurities in a crystal
thickness
of
metallurgical
channel
of junction
tr
0m
o rnaut
field-effect
transistor
diffusion coefficient
effective diffusion-coefficient
2aseous diffusion coefficient,
diffusivity o f electrons
diffusivity o f impurities in oxide
diffusivity o f holes
density o f uniformly distributed surface recombinationgeneration centers (per unit area and energy)
concentration o f uniformly distributed b u l k recombinationgeneration centers (per unit volume and energy)
electric field
electron energy
transistor emitter factor
activation energy
electron energy at conduction band edge
critical electric field at breakdown
electron energy at the Fermi level
electron energy at quasi-Fermi level for electrons
electron energy at quasi-Fermi level for holes
width of the forbidden gap
electron energy at the intrinsic Fermi level
electric field at the semiconductor surface
energy level o f recombination-generation center
electron energy at valence band edge
flux
flux of electrons
flux of holes
probability o f occupation by an electron
m a x i m u m frequency o f operation
rate o f generation o f electron-hole pairs due to absorbed light
(per unit time and volume)
conductance o f the metallurgical channel o f a j u n c t i o n fieldeffect transistor
rate o f generation o f electron-hole pairs in dark (per unit time
and volume)
channel conductance
transconductance
transconductance in the saturation region
Henry's-law constant
gas-phase mass-transfer coefficient i n terms o f concentrations in

the solid
FE
K
H
common-base current gain

common-emitter current gain
small-signal common-emitter current gain
gas-phase mass-transfer coefficient i n terms o f concentrations in
the gas
gas-phase heat-transfer coefficient
current
base current
collector current
IcBO
A,
I
collector-base reverse current, w i t h emitter open

emitter-to-coilector reverse current, w i t h base open
drain current
diffusion current
drain current i n saturation region
emitter current
forward current
generation current
hole current
reverse current
recombination current
h
^liff
r
n c
k
k
k,
K
K
K
s
Boitzmann's constant
surface reaction rate constant
thermal conductivity
dielectric constant
dielectric constant o f oxide
dielectric constant o f semiconductor
channel length
diffusion length o f electrons
diffusion length o f holes
x v
List of Symbols
>
segregation coefficient.of i m p u r i t y at the oxide-silicon interface
m'
effective mass
multiplication factor
n
/I,
n
n
n
VD
7J
electron concentration at the surface
Nj
A'
concentration of acceptor impurities

effective density of states in the conduction band
fixed surface-state charge density per u n i t area
ideal gas-law constant
t o t a l recombination rate (per u n i t time and volume)

transistor recombination factor
d r a i n series resistance
Reynolds number
source series resistance
t h e r m a l resistance
R
*
Re
X.
*a
X
sheet resistance o f a square
'B
base spreading resistance
sc
series resistance o f the collector region

series resistance o f the emitter region
SE
surface recombination velocity
concentration of donor impurities
charge per unit area i n the semiconductor

Q
ss
electron concentration
intrinsic carrier concentration
concentration of electrons i n an 77-type semiconductor
concentration of electrons i n an 77-type semiconductor i n
equilibrium
concentration of electrons i n a p-type semiconductor
concentration o f electrons i n a p-type semiconductor, in equilibrium
n0
List of Symbols
A",,
density of surface recombination-generation centers per u n i t area
m a x i m u m surface recombination velocity
TV,
concentration o f bulk recombination-generation centers per u n i t
A',.
volume
effective density o f states i n the valence band
surface recombination velocity o f a surface w i t h o u t a

space-charge region
T
pressure
hole concentration
concentration of holes in an 77-type semiconductor
concentration of holes i n an 77-type semiconductor
librium
P
P
Pn
Pn*
Pv
Pvo
concentration of holes in ay>-type semiconductor
Ps
hole concentration at the surface

_ 1 S
magnitude o f electronic charge, 1.6 x 1 0
total amount o f impurities per unit area in the solid

B
onset o f
charge per
onset o f
charge per
charge per
charge per
strong inversion
unit area v, ;thin the surface depletion region at the
strong inversion, in e q u i l i b r i u m
unit area on the gate
unit area w i t h i n the oxicu
unit area due to electrons in *he inversion layer
temperature
j u n c t i o n temperature
time
time interval between collisions
response time o f field-effect transistor
t u r n - o f f time
transit time
gas velocity
velocity o f undisturbed gas
net rate o f carrier recombination-generation in the b u l k (per
u,
net rate o f carrier recombination-generation at a surface

unit time and area)
coulomb
charge per unit area w i t h i n the surface depletion region at the
Q
0~
O
h
off
u n i t time and volume)
total number o f impurities per u n i t area in the base region o f a

transistor
(!
in equi-
concentration o f holes i n a />-type semiconductor, i n e q u i l i b r i u m
q
0
coll
o
surface
'drift
average drift velocity

t h e r m a l velocity o f carriers
V
V
film
g r o w t h rate
voltage
average base voltage d r o p
drain voltage
d r a i n voltage at the onset o f saturation
(per
List of Symbols
xviii
forward bias'
flat-band voltage
gate voltage
applied j u n c t i o n voltage
Vr
Va
y.
voltage d r o p across oxide

polarizing voltage
reverse bias
turn-on or turn-off voltage
base w i d t h
w i d t h o f the depletion region w i t h i n the c y l i n d r i c a l region, at
w*
WE
WEB
K
w,
*4
x
tm*X
"^.iniax.o
breakdown
emitter depth
width o f the emitter-base j u n c t i o n space-charge region
width o f the lowly doped region i n an epitaxial device
depletion region w i d t h i n e q u i l i b r i u m
sample thickness
w i d t h o f surface depletion region
maximum w i d t h o f surface depletion region
m a x i m u m w i d t h o f surface depletion Fegion. in e q u i l i b r i u m
j u n c t i o n depth
oxide (insulator) thickness
*
Y
mole fraction
X
X
y.
J
y
5
K
>
density
resistivity
space-charge density
conductivity
capture cross-section
r
r,
r
~n
<p
<b
<b
offset time in the general relationship for the oxidation o f silicon

effective lifetime w i t h i n a reverse-biased depletion region
lifetime o f holes in an /i-type semiconductor
lifetime o f electrons in a p-lype semiconductor
electrostatic potential
built-in voltage o f metallurgical p-n j u n c t i o n
Fermi potential
4>
surface potential, i.e., total potential variation across surface

space-charge region
total potential variation across space-charge region
metal w o r k function
depletion region w i d t h o f metallurgical j u n c t i o n
p
p
0
1
rt
channel w i d t h
common-base current gain
common-base current gain in the reverse direction
transport factor
common-emitter current gain
emitter efficiency
average boundary layer thickness
permittivity o f free space (8.86 x 10~" f/cm o r 55.4 e / V
thermal diffusivity
viscosity
mobility
electron m o b i l i t y
hole m o b i l i t y
4>
<J>
5
u
metal-semiconductor work function difference

semiconductor w o r k function
Physics and Technology of

Semiconductor Devices
Introduction:
The
Planar Technology
The planar technology and semiconductor devices made by this technology were first described in I960. The planar technology has since
become the principal method of fabricating semiconductor devices a n d
integrated circuits, and has strongly contributed to the rapidity w i t h which
semiconductor devices have displaced older types o f electronic components and penetrated into entirely new electronic applications.
Like most important technological advances, the planar technology
evolved from several generations o f earlier ones. This evolution, as well
as the essentials o f the planar technology, are best illustrated by considering i t in comparison w i t h two o f the most important earlier semiconductor
device-fabrication techniques, the grown junction method and the 'alloys
junction method. These are illustrated in the figure.
1
11
I n the grown junction method, a semiconductor crystal is g r o w n out o f

a melt o f the semiconductor which is doped a certain type (taken to be
/7-type in this example). A t some point i n the growth-process, the doping
concentration in the melt is suddenly changed; for instance, by dropping
a pill containing donor-type impurities into the melt. As a result, the
rest o f the crystal w i l l be grown /i-type. W h e n the g r o w t h is completed,
the crystal is sectioned into little bars containing the p-n j u n c t i o n as
indicated by the dashed lines.
This method was extremely important in the first years following the
invention of the j u n c t i o n transistor. For instance, i t . y i e l d e d the first
Introduction.
T h e Planar Technology
7
diod-s w i t h which the theory of the current-yoltage characteristics of p-n
junctions was verified. The grown j u n c t i o n m e t h o d however, was n o t as
Suitable for mass p r o d u c t i o n as another method developed i n the early
years o f semiconductor device technology, the alloy j u n c t i o n m e t h o d .
Crystal
Pellet of p-type
impurity
SiO,
ILLLi U
p-type
melt
n-lype crystal
p-type impurities
Molten
boundary
Recrystallized
p-type r e g i o n _ ^
w
m
(a) Grown junction
method
The search for a method which gives superior control o f the l o c a t i o n o f

p-n junctions led to the development o f diffused junctions.*
Diffused
junctions are formed i n a manner similar to the alloying process i n that
the surface of the wafer is exposed to a source of a high concentration o f
opposite-type impurities contained, for instance, i n a gas. However, no
phase-formation takes place i n this case; the impurities penetrate the
semiconductor crystal by solid-state diffusion which can be controlled to
a very precise degree.
W i t h the a d d i t i o n a l discoverv that a t h i n laver of silicon dioxide can
effectively mask against the diffusion o f most i m p o r t a n t acceptor and
donor i m p u r i t i e s . a new degree o f precision was added i n c o n t r o l l i n g the
geometry o f diffused-junction semiconductor devices; the device geometry
n o w could be delineated b \ covering the semiconductor w i t h silicon
dioxide, and then exposing the semiconductor to the diffusing impurities
only in selected areas, defined bv photolithography, where the oxide laver
has been removed.
5
Melt changed
to n-type
However, as semiconductor devices reached out for new applications,

demands on their performance greatly increased.
These increasing
demands soon demonstrated the inherent limitations o f the a l i o } ' j u n c t i o n
method. For instance, we shall see later that the most i m p o r t a n t factor
determining the performance o f junction transistors is the distance separati n g two junctions. I n the alloy process, the location o f j u n c t i o n s was
determined by the extent to w h i c h the recrystallized regions penetrated
into the semiconductor. However, this penetration was always difficult to
control.
(b) Alloy junction

method
(c) Planar technology
I n the alloy junction method, a pellet containing acceptor-type impurities

( i n this example) is placed upon a wafer o f n-type semiconductor. The
wafer and the pellet are then heated to a temperature high enough so that
the pellet fuses or alloys-into the semiconductor wafer. W h e n the crystal
is cooled, a recrystallized region which is saturated with acceptor-type
impurities forms under the pellet. Thus a p-n j u n c t i o n results.
T h i s method has beenand still isemployed w i t h great success i n the
mass production o f diodes and transistors, m a i n l y made o f germanium.
The planar technology, schematically illustrated i n the last figure, c o m bines the advantages o f j u n c t i o n f o r m a t i o n by solid-state .diffusion and
the masking property o f silicon dioxide for precise definition o f device
geometry. I t also makes use o f the very i m p o r t a n t fact that the electrical
characteristics o f a silicon surface covered w i t h an oxide layer are superior
to those o f a bare surface.
Because o f this last feature, the sensitivity of semiconductor devices to
their environments has been greatly reduced with a corresponding increase
in the r e p r o d u c i b i l i t y and stability o f device characteristics. Recently,
thermally oxidized silicon structures have even made the construction o f
surface field-effect devices feasible for the first time since the conception
of such devices some t h i r t y years ago. M o s t i m p o r t a n t l y , the planar
technology has led to the development o f integrated circuits w h i c h ,
although only a few vears o l d . show indications o f influencing electronics
technology, design, and applicationpossibly to as large an extent as the
advent o f semiconductor devices did ten vears earlier.
Introduction
READING
REFERENCES
The earlier methods of junction formation are summarized (with manv references)
by \V. C . Dunlap, Section 7 in Handbook of Semiconductor Electronics. 2nd d., L . P.
Hunter, Editor, McGraw-Hill Book C o . , 1962.
A detailed description of the planar technology and its application to integrated
circuits is given by G . E . Moore, Chapter 5 in Microelectronics, E . Keonjian, Editor,
McGraw-Hill Book C o . . 1963.
REFERENCES
SOLID-STATE TECHNOLOGY
CITED
1. J . A . Hoerni. "Planar Silicon Transistors and Diodes," IRE Electron

Meeting, Washington, D . C . (1960).
PART I
Devices
2. G . K . Teal, M . Sparks, and E . Buehler. "Growth of Germanium Single Crystals

Containing P-N Junctions," Phys. Rev., 81, 637 (1951).
3. R . N . Hall and W. C . Dunlap, "P-M Junctions Prepared by Impurity Diffusion,"
Phys. Ree, 80. 467 (1950).
4. M. Tanenbaum and D . E . Thomas. "Diffused Emitter and Base Silicon Transistors,"
Bell System Tech. J.. 35, 1 (1956); C . A . Lee, " A High Frequency Diffused Base
Germanium Transistor," Bell Systems Tech. J.. 35. 23 (1956).
5. C . J . Frosch and L . Derrick, "Surface Protection and Selective Masking During
Diffusion in Silicon,"/. Electrochem. Soc, 104. 547 (1957).
Vapor-Phase G r o w t h
T h e r m a l Oxidation
KINETICS
GAS-PHASE
OF
SOME
GROWTH
MASS
PROPERTIES
TRANSFER
O F GASES
Vapor-phase g r o w t h techniques are employed in semiconductor technology for the deposition o f metals (e.g.. aluminum"), insulators (e.g..
S i 0 ) , and semiconductors (e.g., silicon). Of these processes, the most
important one f r o m the standpoint o f device fabrication is the g r o w t h o f
single crystal semiconductor films upon single crystal substrates o f the
same semiconductor. Such g r o w t h is called epitaxial (Greek for "arranged
upon").
2
The importance o f epitaxial g r o w t h in semiconductor device technology

is due to the ease w i t h which the i m p u r i t y concentration in the h i m can be
adjusted independently o f the impurities within the substrate by c o n t r o l l i n g
their concentration in a gas. Thus epitaxial growth can be used to f o r m
p-n junctions between the epitaxial film and the substrate. M o r e i m p o r tantly, it can be used to grow films of relatively l o w i m p u r i t y concentration
upon substrates w h i c h contain the same type o f i m p u r i t y I n much higher
concentrations. I n this manner the series resistance associated w i t h the
substrate can be reduced w i t h o u t otherwise changing the characteristic*
of semiconductor devices.
Because the epitaxial growth o f semiconductors is so i m p o r t a n t , it has
been studied i n great detail. There are several ways in which such g r o w t h
can be performed. I n this chapter we study the method which is most
widely used in the epitaxial g r o w t h o f silicon: the vapor-phase reduction
of silicon tetrachloride. ( F o r a review o f other tvpes o f vapor-phase
g r o w t h reactions, see the Reading References listed at the end o f this

chapter.)
T w o types o f reactors that have been employed i n the epitaxial g r o w t h
o f silicon are illustrated in Figure 1.1. Figure 1.1a shows the vertical
reactor used in the pioneering w o r k o f Theuerer.
Figure 1.1b shows the
horizontal reactor used in a more recent investigation by Shepherd.
The
latter is typical o f present-day industrial reactors i n which a film is grown
simultaneously on many wafers. I n b o t h reactors, a stream o f hydrogen
containing a certain concentration o f silicon tetrachloride enters the
reactor. The gas flows past silicon wafers resting on a susceptor w h i c h , i n
t u r n , is heated by inductive coupling to radio-frequency induction-coils.
I t is also k n o w n that an additional competing reaction takes place

simultaneously w i t h the g r o w t h reaction, given by
S i C l + Si(solid) =
2SiCl,.
(al Vertical reactor

Fig. 1.1
As a result, i f the S i C l concentration is very high, etching o f the silicon

w i l l take place even in the absence o f a significant concentration o f HC1
. i n the incoming gas stream. This is illustrated i n Figure 1.2 which shows
4
"' Horizontal reactor
Reactors employed in che epitaxial growth of silicon.
T h e reactor itself has quartz or glass walls. Since induction heating is

employed, the walls remain cold d u r i n g the g r o w t h process. This e l i m i nates deposition on the reactor walls and results in m i n i m u m c o n t a m i n a t i o n
f r o m the walls.
W h i l e vapor-phase deposition o f materials can take place at relatively
low temperatures, epitaxial g r o w t h itself requires relatively high deposition
temperatures, typically above 1000C in the case o f silicon. Epitaxial
g r o w t h needs high temperatures because the deposited atoms must find
their proper position w i t h i n the crystal lattice i n order to result i n a single
crystal film. As the temperature o f deposition is reduced, the m o b i l i t y o f
the deposited atoms decreases, and the resulting film becomes more and
more defective, eventually losing its crystalline nature.
The overall reaction which results in the g r o w t h o f silicon films is
.
S i C l + 2H., %yt Si(solid) + 4 H C 1 .

4
This reaction is reversible;
i.e., it can take place in either direction. I f
the carrier gas entering the reactor contains HC1. removal or

rather than growth o f silicon w i l l take place.
etching
Vlole fraction of S\C\
Fig. 1.2
Effect of SiCI, concentration on silicon deposition.
the effect o f the concentration o f S i C l i n the gas o n the reaction. (Concentrations in gases are most conveniently given i n terms o f mole fractions.
The mole fraction Yls defined as the ratio o f the number o f molecules o f a
given species to the t o t a l number o f molecules i n the gas.) I t is evident
that initially the deposition rate increases w i t h increasing concentration
of S i C l . As the concentration o f S i C l is increased, a m a x i m u m growth
rate is reached at a mole fraction o f about 0 . 1 . Further increase i n
concentration brings about a decrease i n the g r o w t h rate and eventually
even results in etching o f the silicon surface.
4
Silicon is usually g r o w n in the low concentration region, as indicated in

Figure 1.2. w i t h a typical growth rate o f ~ 1 u / m i n . I n this region the
Kinetics of G r o w t h
IO
reaction, which we assume this one to be, i t is the concentration o f the

reacting species.t
dependence o f film g r o w t h rate V on the mole fraction o f SiClj in the gas

mixture Y is approximately linear.
In steady state F = F = F. U s i n g this c o n d i t i o n , we can solve the

above two equations to obtain the surface concentration o f the silicon
tetrachloride at the gas-film interface, .
1
1.1
KINETICS
OF
GROWTH
Cr
We shall now study the kinetics o f film g r o w t h on the basis o f a very
simple m o d e l . The essentials o f this model are depicted i n Figure 1.3
where we show the concentration d i s t r i b u t i o n o f the silicon tetrachloride
i n the gas and we indicate the flux o f the silicon tetrachloride f r o m the
C\ =
(1.3)
i + kjffhe
This formula shows that the surface concentration w i l l go to zero i f
h kg. This c o n d i t i o n is c o m m o n l y referred to as mass-transfer
control.
On the other h a n d , iLh
k . the surface concentration approaches C .
This condition is c o m m o n l y referred to as surface-reaction
control.
We can n o w readily express the g r o w t h rate o f the silicon film by
writing
V=
=
(1.4)
A'i
k + h A,
G
-Film-
-Gas-
ili
where A'] is the number o f silicon atoms incorporated i n t o a u n i t v o l u m e

of the film. Its value for silicon is 5.0 x 1 0 - c m . N o t i n g that C =
YC
where C is the total number of molecules per cubic centimeter in the gas,
we get the expression for the g r o w t h rate,
- 3
Fig. 1.3
1=5-2
k + h
Model of the growth process. Direction of gas flow i> normal to plane of paper.
b u l k o f the eas to the surface o f the g r o w i n g film F and the flux corresponding to the silicon tetrachloride consumed i n the
F
film-growth
reaction
(Flux is defined as the n u m b e r o f atoms or molecules crossing a unit
area i n a unit time.)

W e approximate the flux F , by the linear f o r m u l a
G
where C
and C
(1.1)
C)
8
are the concentrations o f the SiClj (molecules per cubic
centimeter) in the b u l k o f the gas and at the surface, respectively, and h

is the gas-phase mass-transfer
coefficient.
cal reaction taking place at the surface o f the g r o w i n g film F is a p p r o x i 2
mated by the f o r m u l a
s
I ' = k
A'i
(1.2)
where k
h the chemical surface-reaction
rate constant.
These linear
approximations are analogous t o O h m ' s l a w : they describe a flux as being
proportional to a driving force.
I n the case o f mass transfer, the d r i v i n g
force is a concentration difference; i n the case o f a first-order chemical
[surface-reaction c o n t r o l ]
(1.6)
[mass-transfer c o n t r o l ] .
(1.7)
or by
-
C
V=h Y
G
(1.5)
T h e flux consumed by the chemi-
F, = k C
M'.
A',
Note the f o l l o w i n g features o f this equation. First, it predicts that the

film growth rate V is p r o p o r t i o n a l to the mole fraction } ' o f the reacting
species. As we have seen earlier, this is in agreement w i t h the experimental
observations for small values o f Y such as are encountered i n usual
practice. Second, the g r o w t h rate at a given mole fraction is determined
by the smaller o f h or k . This corresponds to the t w o l i m i t i n g cases o f
mass-transfer controlled and surface-reaction controlled conditions. I n
these l i m i t i n g cases the g r o w t h rate w i l l be given either by
G
F , = h {C
* Strictly speaking, we should include in our consideration the flux of the reaction
product HCI from the surface back to the bulk of the-gas. In this treatment we neglect
this flux for simplicity. This is equivalent to the assumption that the mass-transfer
coefficient of the reaction product is much larger than that of S i C l .
4
12
Gas-Phase Mass Transfer
13
7
li
e v , k T
/ V ; = 5 to 10 cm/sec and k = 10 cm/sec e~

. Numbers o f the
same order o f magnitude can be extracted f r o m Theuerer's data obtained
in the vertical reactor.
This model o f the g r o w t h process is quite oversimplified. I t does n o t .
consider the flux o f the reaction p r o d u c t ; the linear approximation that
describes the surface reaction is. as we have discussed earlier, valid only
for l o w values o f Y. A l s o , the temperature gradient present in a cold-wall
reactor is extremely steep. Because o f this, the properties o f the gas w i l l
vary radically between the heated substrate and the cold glass wall.
Nevertheless, this simple model still predicts the overall features observed
experimentally. I t describes the two regions o f the growth processthe
mass-tiansfer and the surface-reaction c o n t r o l l e d regionsand yields at
least an order o f magnitude estimate o f b o t h the chemical surface-reaction,
rate constant k and the gas-phase mass-transfer coefficient h from the
growth-rate data.
s
The temperature dependence o f the film g r o w t h rate observed e x p e r i mentally by Shepherd is shown by the points in Figure 1.4. I t is evident
that at low temperatures the g r o w t h rate follows an exponential law,
fee - V *
j
activation energy E is about 1.9 ev. Theuerer also
2
I n order to obtain a good quality epitaxial film, g r o w t h temperatures

have to be relatively h i g h . As a result, most industrial epitaxial processes
take place i n the mass-transfer controlled region, so i n the next section we
consider the mechanisms which determine gas-phase mass-transfer
coefficients.
1.2 G A S - P H A S E
MASS T R A N S F E R
I n Section 1.1 we approximated the flux f r o m the bulk o f the gas to the
surface o f the solid by the formula
F , = h (C
G
0.625
0.65
0.675
0.70
0.725
C ).
s
We n o w consider the factors which determine
h.
G
lo-VrrK-'i
Fig. 1.4
Temperature dependence of the film-growth rate. Points represent experi-
mental data.- curves were calculated based on Equation 1.5.
observed an exponential temperature dependence in this range corresponding to an activation energy o f 1.6 ev. A t high temperatures the
g r o w t h rate levels off, and becomes relatively temperature insensitive.
Since chemical reaction rate constants generally follow an exponential
temperature dependence while mass-transfer coefficients are relatively
insensitive to variations i n the temperature, the simple model explains the
general features o f the temperature-dependence data shown i n Figure 1.4.
By fitting E q u a t i o n 1.5 to these data, we can obtain theoretical curves such
as shown i n that figure. The best fit seems.to be obtained for the values o f
a.
Stagnant-film Model*
The simplest possible picture o f the mass-transfer process is shown i n

Figure 1.5. I n this picture we split the gas phase into two regions: one
which is well mixed a n d which flows past the solid surface w i t h a uniform
velocity U, and another, a stagnant film o f thickness <5 next to the solid.
Transport o f the active species across this stagnant film proceeds by
diffusion alone. Thus the flux F can be immediately written down as
x
Fx -
(1.8)
Vapor-Pha.se G r o w t h
14
where D is the diffusivity
o f the active species in the gas. This results in a
formula for the gas-phase mass-transfer coefficient.
(1.9)
hr.=
A l t h o u g h this stagnant-film m o d e l is quite arbitrary and fictitious, i t has

been successfully a p p l i e d to many problems i n v o l v i n g gas-phase mass
Gas-Phase Mass T r a n s f e r
15
possessed a finite velocity, the velocity gradient there w o u l d be infinitely

larse. Then according to Equation 1.10 an infinitely large frictional force
would act on the fluid, immediately bringing its velocity to zero. Thus the
velocity o f the fluid right next to a solid body must clearly be zero in order
to avoid the existence o f infinitely large velocity gradients.
The b o u n d a r y c o n d i t i o n o f zero velocity right next to the plate disturbs
the velocity d i s t r i b u t i o n . As the fluid moves along the plate, this disturbance w i l l spread further and further into the bulk o f the fluid. T o calculate
the thickness o f the disturbed region 6, we recall that it is the friction at
the wall that causes the deceleration of the fluid.
iM.v)
it = 0.991/
( BoundaryI layer
1
rf.v-H
-LFig. 1.6
/////y/
v///////////////""" >
Fig. 1.5
The "stagnant-film"
transfer since the 1930's.
o f
S^'P"*"
m a s s
t r a n s f e r
However, the thickness o f the stagnant
Development of a boundary layer in flow past a flat plate.
According to N e w t o n ' s second law.

film
F=ma.
(1.11)
being an arbitrary quantity, it has to be determined experimentally for any

given set of conditions.
b.
Boundary L a y e r T h e o r y
Consider n o w the shaded element shown in Figure 1.6. and take it to

represent a v o l u m e element whose size in the direction n o r m a l to the paper
is unity.
The p r i n c i p a l force acting on this element is F
dx.
The acceleration o f this element is
f r i c t j o n
F l u i d mechanics can provide a more realistic and useful estimate o f the

mass-transfer coefficient h .
c
Consider the geometrically simplest p r o b l e m :
a fluid flowing parallel to a flat plate o f length L . We assume that the

extent of the flowing fluid is infinitely large or. i n other words, that the
fluid is unbounded.
Far away from the plate the fluid flows w i t h a u n i f o r m
velocity U as shown i n Figure 1.6.
Right next to the plate the velocity o f
du
du dx
du
dt
dx dt
dx
u.
The mass o f this element is p6(x) dx. Here p denotes the density.
I f we combine the above with Newton's second law. we get
the fluid is zero. The reason f o r this is as follows. The frictional force per
u n i t area along the i - d i r e c t i o n acting on a fluid element next to the plate
is given by
-
F|rleUB ~ /< T
By
(1-10)
where ft is the coefficient of viscosity or simplv viscosity.

The plate is
stationary, i.e.. its velocity is zero. I f right next to the plate the fluid
ou du
- u = po(x)u .
by
dx
(1.12)
Replacing the differentials by the respective differences results in

/u
d{x)
a, it-
~ pdlx)l
x
(1.13)
Gas-Phase Mass Transfer
16
17
where n\pD ,
a dimensionless group called the Schmidt number, has
values between 0.6 and 0.8 for most gases practically independently o f
temperature. Thus the numerical factor by which our simple derivation
deviated from the result o f the more exact treatment is o f the order o f
unity.
We can now estimate the mass-transfer coefficient for an epitaxial
reactor such as employed i n Theuerer's and Shepherd's w o r k . Typical
velocities are 10 to 30 cm/sec, leading to values o f the Reynolds number
of ~ 2 0 . Using Equation 1.19, this leads to h ^ 5 cm/sec. This is i n
0
w h i c h can be rearranged to lead to an expression giving an estimate o f the

size o f the region o f disturbed velocity c o m m o n l y referred to as the
boundary layer
thickness:
af*)3S
(1.14)
y
fj
M o r e exact calculations by Blasius i n the 1930's. based o n the solution

o f the complete equations o f m o t i o n for a fluid, led to a f o r m u l a which
differs f r o m E q u a t i o n 1.14 only by a numerical coefficient whose value is
between i and 5. depending on the precise definition o f d. ( I n Figure 1.6
the dashed line indicates the l o c a t i o n o f those points where the velocity
reaches 9 9 % o f the velocity o f the free stream U. T h i s is one way o f
defining 6.)
T h e average boundary layer thickness 6 over the whole plate is given by
d = -
I d(x)
L
dx = iL^pJplJL
(1.15)
'
3=1
Jj
L Jo
or
;"
5 = . - k = .
(1.16)
3 v Re,,
This expression gives the average boundary layer thickness i n terms o f the
i'
0.2
0.4
0.6
0.3
z
Reynolds
number,
The Reynolds number is an extremely i m p o r t a n t dimensionless number

i n f l u i d dynamics.
I t represents the ratio o f the magnitude o f inertial
Fig. 1.7
p.i
effects to viscous effects i n fluid m o t i o n . Thus large Reynolds numbers

i m p l y small viscous effects, and vice versa.
N o w , i f we take the average b o u n d a r y layer thickness h for the thickness
o f the stagnant film b i n E q u a t i o n 1.9, we get an expression for the masstransfer coefficient h :
I
.
%.
w h i c h can be rearranged i n t o the dimensionless equation

-f- = *
T h i s can be compared w i t h the result o f a more exact calculation b y
Pohlhausen: "-
;
-2- -
R ^ -
(1-18)
1.0
(Gas flow rate)" (liter/min.)"
|
JvRT i>JZp~DZ
L
(1.19)
1.2
1.4
1.6
Effect of gas-flow rate on the film growth rate in a vertical reactor.
order-of-magnitude agreement w i t h the value we had obtained by fitting

our model to both Shepherd's and Theuerer's experimental data. (The
estimation o f viscositv, densitv, and diffusivity are discussed i n Section
1.3.)
Theuerer's data o f film g r o w t h rate as a function o f the square root o f
gas-flow rate i n the vertical reactor are shown in Figure 1.7. I t appears
that at l o w gas-flow rates the film g r o w t h rate V increases approximately
in p r o p o r t i o n to the square root o f the gas-flow velocity. A t high Hew
rates, V reaches a m a x i m u m ; i n this region V depends exponentially on
temperature, indicating that the g r o w t h rate has become surface-reaction
limited.
In contrast. Shepherd found no effect o f the gas-flow rate on the masstransfer coefficient in the horizontal reactor. T o explain this discrepancy,
we must recall that the length o f the effective flat plate (the susceptor) in
Shepherd's horizontal reactor is considerably longer than in Theuerer's
vertical reactor. In-fact, the average boundary-layer thickness for Shepherd's reactor can be estimated to be o f the order o f 2 to 3 cm, which is as
18
S o m e Properties of Gases
19
a.
larce as the radius o f the reactor. Thus, evidently, the unbounded fluid
approximation cannot be expected to yield reasonable results.
I n such a case we may use another simple approximation based on
Graetz's theory, developed i n 1885 for heat transfer i n a circular tube.
Because the transport o f heat and o f mass are described by the same
equations, Graetz's theory can be readily adapted to mass-transfer considerations as shown i n Figure 1.8. I t is evident that for both a parabolic
Ideal G a s L a w
The ideal gas law states that
P' mobr = RT = (Avogadro's number) kT
( ] 20)
where the gas constant R is conveniently given for such calculations'by
R = 8^ 1
(g-mole)K '
100
I I 1 1 111
I I I 1111
I I
1 1
I ]'_
and the B o l t z m a n n constant k is conveniently given by

c m 3 a
k = 1.37 x 1 0 - " tm_

(mo)ecule)K '
*^ZL
S,Vt
he
" - * ""* *
p
(1-21)
10
Rod-like flow
Parabolic velocity profile
I I ! ! I 11
10
M i l l
10
10
(Gas flow raid
Fig. 1.8
tube.
The Graetz-solution for mass transfer from the wall to the fluid in a circular
velocity profile and for a rod-like flow (uniform velocity profile) the masstransfer coefficient for l o w gas-flow rates becomes independent o f the flow
rate. The conditions in Shepherd's reactor correspond to an abscissa o f
the order o f 1. This leads to a mass-transfer coefficient of the same order
as the boundary layer considerations d i d . However, this theory predicts
the complete absence o f a gas flow-rate dependence o f the mass-transfer
coefficient, i n agreement w i t h Shepherd's observation.
1.3 S O M E
PROPERTIES
O F GASES
10-5
We now briefly review h o w certain properties o f gases, which

needed i n mass-transfer calculations, are obtained.
are
F
' g . 1.9
ensity and viscosity of H N , and O as a function of the abs

s
olute temperature.
20
Vapor-Phase Growth
Problems
21
The ideal gas l a w can also be rearranged to give directly the density o f a
6. Ibid., Chapter 14.
gas,
7. R. B. Bird. W. E . Stewart, and E . N. Lightfoot. Transport Phenomena. Wiley, 1960,

Chapter 21.
P = (molecular weight) .
RT
The
densities o f three gasesoxygen,
nitrogen,
(1.22)
3. W. H . McAdams, Heat Transmission, 3rd ed., McGraw-Hill Book C o . , 1954,
Chapter 9.
and
hydrogenare
9. R. C . Reid and T . K . Sherwood, The Properties of Gases and Liquids, McGraw-Hill

Book C o . , 1958.
s h o w n i n Figure 1.9 as functions o f the absolute temperature.
PROBLEMS
b.
Transport Properties
The
estimation o f viscosities a n d
diffusivities
o f gases
is
1.1
treated
lkT
Exponential temperature dependence is alternately represented bv the form

e~ i
or e~ ,
where E is the value of the activation energy in electron volts, and
A H is its value in Kcal/mole. Show that if 7 = 1 ev, the corresponding value
of A H is 23 Kcal/mole.
(
lBIRT
t h o r o u g h l y elsewhere.
T o provide a feeling for the orders o f magnitude
i n v o l v e d , we show the viscosity o f the same three gases as a f u n c t i o n o f

absolute temperature i n Figure 1.9.
1.2
Calculate the time required to grow a monolayer of silicon epitaxially under the
conditions of Figure 1.4, at 1200C.
1.3
Derive the expression for the film-growth rate V if deposition takes place through
the chemical reaction.
Diffusivities o f gases at r o o m temperature generally range between 0.1

and
1 crnr/sec.
Gaseous diffusivities
temperature as 7 ,
generally increase
with
absolute
w i t h in between 1.5 and 2.
S i X ^ Si(solid) + X .
in terms of the rate constants of the forward and reverse chemical reactions k
and kf and the gas-phase mass-transfer coefficients of S i X and of X . h and h .
Under what conditions does the resulting expression reduce to Equation 1.5?
s
READING
REFERENCES
A general disaissiatt.ofj.Lappr-ohase growth is given by C . P. Powell. J . H . Oxlev,

and J . M . B\ocher^VaporDeposuion,
Wiley, 1966.
1.4
qx
A tubular flow reactor is 2" in diameter. The flow meter indicates a gas-Row rate
of 1 liter/min. The entering gas mixture is 98% H , , 2 % S i C l by volume. T h e
reactor is heated to 1200C. Calculate:
(a) The Reynolds number based on the wafer size.
(b) The average boundary layer thickness over the first wafer.
(c) The concentration of S i C U (molecules/cm ) in the entering gas.
(d) The flux of S i C l molecules to the surface.
4
The epitaxial growtbAotêmiconductors is discussed in a series of papers in the

December 1963 issue of RCA Review; and in Volume 9 of a series of review-reports
on "Integrated Silicon Device Technology" by the Research Triangle Institute, A S D TDR-63-316 (1965).
Gas-phase mass transfer is treated by R. E . Trevbal, Mass-Transfer
Operations.
McGraw-Hill Book C o . , 1955; and by R . B. Bird, W. E . Stewart, and E . N . Lightfoot,
(e) The film growth rate, assuming the growth is mass-transfer limited.
(f) The concentration of S i C l in the vent gas. assuming that there are 20 wafers
in the reactor, and that the film growth rate is the same on all wafers and is
that calculated in (e).
4
Transport Phenomena, Wiley, 1960.
REFERENCES
CITED
1. H . C . Theuerer, "Epitaxial Silicon Films by the Hydrogen Reduction of S i C l , "

4
/. Etectrochem. Soc, 108, 649 (1961).

2. W. H . Shepherd. "Vapor Phase Deposition and Etching of Silicon." / . Etectrochem.
Soc,
112, 988 (1965).
3. A . S. Grove, "Mass Transfer in Semiconductor Technology," Ind. & Eng. Chem.,

58. 48 (1966).
4. R . E . Treybal. Mass-Transfer Operations. McGraw-Hill Book C o . . 1955. Chapter 3.
5. H . Schlichting, Boundary Layer Theory, 4th ed., McGraw-Hill Book C o . , 1960,
Chapter 7.
KINETICS
SPACE-CHARGE
OF OXIDE
23
Kinetics of O x i d e G r o w t h
GROWTH
EFFECTS
Thus some o f the silicon is used up in the g r o w t h o f an oxide film. I t

can be shown f r o m the densities and molecular weights o f silicon a n d
silicon dioxidef that i n the g r o w t h o f an oxide film whose thickness is
a: a layer o f silicon 0.45z thick is consumed.
I n this chapter we study the thermal o x i d a t i o n o f silicon i n some detail.

The thermal o x i d a t i o n o f silicon is performed in o x i d a t i o n furnaces or
reactors o f the type illustrated schematically in Figure 2 . 1 . The reactor
itself is tubular, usually made o f quartz or glass. I t is heated by a resistance
furnace t o temperatures i n the vicinity o f 1000C. A gas containing the
oxidizing m e d i u m (oxygen or water vapor) flows t h r o u g h the reactor and
Thermal Oxidation
O O O O o o o o o o o o o o
O , or 1-UO
Carrier gas
(I
-Vent
OOOOOOOOOI
Fig. 2.1
The growth o f a t h i n layer o f silicon dioxide on a silicon wafer is a basic

feature o f the planar technology. B o t h the precise control of the thickness
o f the oxide laver and a knowledge of the kinetics o f the oxidation process
are therefore o f obvious importance i n the fabrication o f planar devices.
In
addition, phenomena
involved i n the
oxidation process play an
i m p o r t a n t role in determining the electrical characteristics o f planar

silicon devices.
. . .
The mechanism o f o x i d a t i o n o f silicon is also o f general scientific
interest. The oxidation o f certain metals such as copper and aluminum
has been studied extensively i n the past i n connection w i t h corrosion
problems. Silicon now provides another example o f an element whose
oxidation kinetics and mechanism are relatively well understood.
Silicon dioxide ( S i O ) layers on silicon can be formed by various
methods. F o r example, they can be deposited through a vapor-phase
reaction, or they can be formed by electrochemical oxidation (anodization)- or by a plasma reaction. I n industrial practice, silicon dioxide
layers are most frequently formed by the thermal oxidation o f silicon
through the chemical reaction
:
Silicon wafers
-Resistance heater
Oxidation furnace.
past the silicon wafers. Gas-flow velocities are typically o f the order o f
1 cm/sec. A t typical o x i d a t i o n temperatures such a flow velocin corresponds to a Reynolds number Re % 10 to 20. (See Chapter 1.)
After the o x i d a t i o n step is completed, the oxide thickness can be determined by either precise measurement o f the weight gain o f the silicon
sample or by optical techniques. These t w o methods have been shown to
be in excellent agreement.
5
2.1
KINETICS
OF OXIDE
GROWTH
We shall study the kinetics o f the oxidation process on the basis o f the
simple m o d e l illustrated in Figure 2.2.
I t has been demonstrated by the use o f radioactive tracers and also bv
other means t h a t o x i d a t i o n o f silicon proceeds by the i n w a r d m o t i o n o f
the oxidizing species t h r o u g h the oxide layer rather t h a n by the opposite
process o f the o u t w a r d m o t i o n o f silicon to the outer surface o f the oxide.
(This forms an interesting contrast to the case o f copper whose o x i d a t i o n
proceeds by the o u t w a r d m o t i o n o f the metallic i o n . and also to the case
of anodic o x i d a t i o n o f silicon where silicon moves o u t w a r d . )
6
Si(solid) + 0
Si0 (solid)
2
or
Si(solid) + 2 H 0 S i 0 ( s o l i d ) + 2 H .
:
Some important properties of S i O , are listed in Table 4.1.
Thermal Oxidation
24
25
F o r the oxidizing species to reach the silicon surface it must go through

three consecutive steps:
1. I t must be transported f r o m the b u l k o f the gas to the oxide-gas
partial pressure o f that species in the surrounding gas. Thus we assume

that the concentration at the outer surface o f the oxide C, is p r o p o r t i o n a l
to the partial pressure o f the oxidant right next to the oxide surface p ,
i.e., C = Hp , where H is the Henry's law constant. Furthermore, we
denote the equilibrium concentration i n the oxide, i.e.,. the concentration
which w o u l d be in e q u i l i b r i u m w i t h the partial pressure in the bulk o f the
gas p by the symbol C*, i.e., C* = Hp .
s
interface.
2. I t must diffuse across the oxide layer already present.
3. I t must react at the silicon surface.
I f we recall that according to the ideal gas law (see Chapter 1) C

p jkT,
and C = pslkT, we can rewrite Equation 2.1 as
= h(C* -
C)
(2.2)
where h is the gas-phase mass-transfer coefficient in terms of concentrations

in the solid, given by h =
h \HkT.
The flux across the oxide F-, is taken to be a diffusive flux,
G
' ~
(2.3)
where D is the diffusivity o f the oxidizing species in the oxide layer.

Finally, the rate o f the reaction taking place at the oxide-silicon interface
is assumed to be p r o p o r t i o n a l to the concentration o f the oxidant at this
interface. Thus.
F
where k
= k C,
(2.4)
is the chemical surface-reaction rate constant for o x i d a t i o n . j
Using the condition o f steady state, F = F = F, = F , after some

algebra, we get the following expressions for the concentrations o f the
oxidant at the oxide-silicon interface C, and at the gas-oxide interface C,
x
Fig. 2.2
Model for the thermal oxidation of silicon.* Direction of gas flow is normal
(2.5)
to plane of paper.
T h e three fluxes corresponding to the three steps are equal in steadystate o x i d a t i o n . T h e y w i l l be a p p r o x i m a t e d as follows. As i n Chapter 1,
a linear a p p r o x i m a t i o n to the gas-phase flux Fi is obtained by taking the
flux o f oxidant f r o m the b u l k o f the gas to the oxide-gas interface to be
p r o p o r t i o n a l to the difference between the concentration o f the oxidant
i n the b u l k o f the gas C and the concentration o f the oxidant right next
to the oxide surface C ,
' , 1
and
in
C
D )'
k
"
( 2
k x
'
6 )
Here h
= h {C
G
C ).
s
(2.1)
is the mass-transfer coefficient which we discussed in the previous
I t is interesting to consider the two l i m i t i n g forms o f Equations 2.5 and

2.6. W h e n the diffusivity is very small, C 0 and C, - C*. I n the
opposite case, when the diffusivity is very large, C and C, w i l l be equal and
{
chapter.
W e now assume Henry's law to h o l d . Henry's law states that, i n equil i b r i u m , the concentration o f a species w i t h i n a solid is p r o p o r t i o n a l to the
Note the similarity between this model and the one used in the treatment of vaporphase growth in Chapter 1.
w i l l be given by C * / ( l + k / / i ) . T h e s e t w o l i m i t i n g cases are called the

diffusion-controlled
and reaction-controlled
cases, respectively.
The
distributions o f the.concentration o f the oxidizing species in the oxide layer
for these t w o l i m i t i n g cases are illustrated in Figure 2.3. (We are assuming
i n this figure that k is much smaller than h. W e shall see later that this is
indeed a realistic assumption.)
I n order to calculate the rate o f oxide g r o w t h we w i l l have to define
one more quantity, A ' j , which is the number o f oxidant molecules incorporated into a unit volume of oxide. There are 2.2 x 1 0 S i 0 molecules/
c m i n the oxide and we incorporate one 0 molecule into each S i 0
oxidation m o d e l may not have been valid. W e w i l l consider the significance

of x, i n this case i n more detail i n Section 2.2.
22
27
The s o l u t i o n o f the differential equation (2.7) leads to the general relationship f o r the oxidation o f silicon,'
x - + Ax
B
where A = 2D(
= Bit + t )
(2.8)
-)
(2.8a)
IDC*
Silicon
Oxide
Gas
B ^
k - D
(reaction controlled)
(2.8b)
x,
Ax,
and
t m
'-.
(2.8c)
E q u a t i o n 2.8 can be solved for the oxide thickness as a function o f t i m e ,
resulting i n
Ail
/ 1 + 2
A*lAS
1
(
(2 9)
'
This relationship is shown in Figure 2.4 along w i t h a large number o f

experimental data taken by various investigators under widely, v a r y i n g
conditions o f temperature, partial pressure o f oxidants, using either
oxygen or water vapor as oxidizing species.
Fig. 2.3
Distribution of the oxidizing species in the oxide layer for the two limiting
cases of oxidation.
Note the t w o l i m i t i n g cases o f this general relationship. F o r large times,

i.e., / A /4B, the so-called parabolic relationship, x - = B: is approached.
I n this relationship B is referred to as the parabolic
rate
constant.
For small times, i.e., (r + t ) A * j 4 B , we o b t a i n the linear
law,
x = 'BjA){t
+ t ) where B'jA is referred to as the linear rate constant and is
given by
2
molecule, whereas we incorporate t w o H , 0 molecules into each S i O ,

molecule'. Hence, Af, for oxidation i n d r y oxygen w i l l be 2.3 x 1 0 cm ,
whereas for oxidation in water vapor it w i l l be twice this number.
Thus the flux o f oxidant reaching the oxide-silicon interface is given by
s2
kgC*
1
dt
1 + ^
(2.7)
D-
We solve the above differential equation subject to the initial condition,

x (0) = x . Here z, is the thickness o f oxide layer g r o w n i n an earlier
oxidation step. T h i s general i n i t i a l c o n d i t i o n permits us to consider t w o
or more successive oxidation steps. The quantity x, can also be regarded
as the thickness o f oxide that is present at the end o f an initial period
of oxidation d u r i n g which the assumptions involved in this simple
o
i
A
J ^ - ^ l .
k + h h\
(2.10)
These t w o l i m i t i n g cases are also illustrated i n Figure 2.4.

As is evident f r o m Figure 2.4, the experimental results f o l l o w the predictions o f the simple model o f oxidation very well over a wide range o f
conditions. I n a d d i t i o n , the predicted effect o f temperature and pressure
on the coefficients A and B o f the general relationship have a l l been v e r i fied experimentally. I n particular, it is found that B is p r o p o r t i o n a l to the
partial pressure o f the oxidant in the gas, indicating that the assumption o f
I MCI mai WA.\jaLiu
wi W A I U B vjrowtn
2?
Henry's law was indeed justified. This implies the absence o f any dissociation effects at the gas-oxide interface. Thus, for o x i d a t i o n b o t h w i t h
oxygen a n d w i t h water vapor, the oxidizing species m o v i n g t h r o u g h the
oxide layer are apparently molecular.
The principal effect o f temperature on B should be reflected i n the
diffusivity D., .
Experimental measurements o f the temperature dependence o f the parabolic rate constant B are shown i n Figure 2.5. I t is seen
([
f - T
:
A
Fig. 2.4
4B
The general relationship for silicon oxidation and its two limiting forms.*
t h a t the dependence is exponential w i t h activation energies o f 0.71 ev

a n d 1.24 ev for the cases o f o x i d a t i o n i n oxygen and i n water vapor,
respectively. These activation energies are very close to the respective
activation energies o f the diffusivities o f water vapor and oxygen through
fused silica (bulk S i 0 ) . Using the k n o w n values o f these diffusivities,
we can calculate the equilibrium concentration o f the oxidizing species
i n the oxide C*. I t is found that these numbers (5 x 10 c m for oxygen
and 3 x 10 c m
for water) are i n good agreement w i t h independent
measurements o f the solubility o f oxygen and water in fused silica, at
atmospheric pressure.
2
16
19
- 3
- 3
The temperature dependence o f the linear rate constant B\A is shown i n

Figure 2.6. I t is evident that the linear rate constant B/A also depends o n
temperature exponentially with an activation energy o f about 2 ev for
both oxidants. The fact that the temperature dependence o f BlA is
exponential and that 2 ev is close to the bond-breaking energy o f silicon as
obtained by independent techniques, indicates that the rate constant
which dominates B/A is the one associated w i t h the reaction at the oxidesilicon interface, A: . (We can see f r o m E q u a t i o n 2.10 that o f k and h
the smaller w i l l determine the magnitude o f BlA.)
10
Confirmation o f the fact that it is the oxide-silicon interface reactionrate constant k that controls B/A, and not the gas-phase mass-transfer
s
Thermal Oxidation
30
rM
o
o
o
o
o
T-
31
of the model o f o x i d a t i o n discussed in this section. There remains one

i m p o r t a n t unexplained observation. I t is found that data obtained by
oxidation i n oxygen can only be brought into consistent agreement w i t h
theory i f we assume a fictitious initial condition o f z,- = 200 A . Careful
measurements o f oxide g r o w t h in dry oxygen at l o w temperatures indicate
that there is, i n fact, a rapid i n i t i a l phase o f oxidation. This is illustrated i n
Figure 2.7, where the i n i t i a l r a p i d oxidation phase is indicated as well as the
o
o
Space-Charge Effects on O x i d a t i o n
0
Fig. 2.7
20
40
60
80
100
Oxidation time (hr)
120
The rapid initial oxidation phase (O. at 700C).
subsequent linear g r o w t h . I t is evident that the linear p o r t i o n o f the oxide

thickness versus time curve extrapolates to approximately x = 200 A
rather than to zero. This has been found consistently through all experiments w i t h dry oxygen. However, such rapid initial phase (or finite x )
has never been observed i n o x i d a t i o n i n a wet ambient.
i
0.7
0.8
0.91.0
]000/Tikl
;
Fig. 2.6
1.1
1 2
The effect of temperature on the linear rate constant.'
coefficient h, is supplied by the realization that the crystallographic

orientation o f the single crystal silicon sample influences the magnitude o f
the linear rate constant B/A .
[ A l l data shown i n this chapter, are f o r
(111) orientation.] Also, estimates o f h based on boundary layer theory,
as outlined in Chapter 1, show t h a t h is about 1.000 times larger than
ks l(Ji
t h). Direct experimental evidence o f the absence of a gas-phase
transport limitation i n thermal o x i d a t i o n is supplied by the observation
t h a t a 50-fold variation i n gas-flow rate has no effect on the value o f B\A}
The experimental observations provide a very good overall verification
n
A possible explanation for this observation might be based on the role o f

space charges in the oxidation process.
2.2 S P A C E - C H A R G E
EFFECTS
ON
OXIDATION
So far we have treated the oxidation process as i f i t proceeded t h r o u g h

the m o t i o n o f an uncharged oxidizing species across the oxide. H o w ever, this may not be the case.
As we have stated earlier, it is k n o w n that oxidation proceeds t h r o u g h
the m o t i o n o f the oxidizing species acrcss the oxide, rather than t h r o u g h
Thermal Oxidation -
32
the m o t i o n o f silicon.
Other experiments studying the effect o f an electric
field applied across the oxide on the o x i d a t i o n rate have indicated that the
o x i d i z i n g species i n o x i d a t i o n i n 0.
is negatively c h a r g e d .
12
W e can n o w speculate about the reaction t a k i n g place at the gas-oxide

interface.
gas
Let us suppose, for instance, that molecular oxygen f r o m
gets adsorbed at the interface.
the
U p o n its entry i n t o the oxide, i t dis-
sociates i n t o a negatively charged oxygen molecule and
a positive hole
B o t h the oxygen i o n a n d
t o w a r d the silicon.
faster.
hole.
the hole begin to move across the oxide layer.
However, the
oxide m u s t still be zero.
The
net electric current flowing across
hole, w h i c h has
the
a higher m o b i l i t y , moves
I n effect, i t runs ahead o f the slower ion and drags the i o n w i t h i t .
Such a consideration is not restricted to the case of t h e r m a l o x i d a t i o n o f

silicon.
Whenever we have the c o u p l e d m o t i o n o f two charged species
t h a t have different mobilities, a b u i l t - i n electric field w i l l result. T h i s

aids the motion of the slower of the two
C a b r e r a and M o t t
have considered this process i n detail.

an
3. J . R . Ligenza, "Silicon Oxidation in an Oxvgen Plasma Excited by Microwaves,"

J. Appi. Phys., 36, 2703 (1965).
4. M. M. Atalia. ""Semiconductor Surfaces and Films; The Silicon-Silicon Dioxide
System," in Properties of Elemental and Compound Semiconductors, H . Gatos, E d . ,
Vol. 5, pp. 163-181, Interscicnce, 1960.
5. B. E . Deal, ""The Oxidation of Silicon in Dry Oxygen, Wet Oxygen, and Steam,"
J. Electrochem. Soc, 110, 527 (1963).
i n i t i a l r a p i d o x i d a t i o n phase.
7. J . R. Ligenza and W. G . Spitzer. "'The Mechanisms for Silicon Oxidation in Steam

and Oxygen," Phys. Chem. Solids, 14, 131 (1960).
3. N . Cabrera and N . F . Mott, ""Theory of the Oxidation of Metals," Rep. Progr.
Phys., 12,163 (1948).
9. E . A . Benjamini. E . F . Duffek, C . A . Mylroie, and F . Schulnburg, "The Mobi'e
Ionic Species During the Anodic Oxidation of Silicon in a Non-Aqueous Medium,"
Electrochemical Society October 1963 Meeting, New York.
field
10. L . Pauling, 77ie Nature of the Chemical Bond, 3rd ed., Cornell Univ. Press,
1960. p. 85. "
T h e i r theory
11. J . R . Ligenza. "Effect of Crystal Orientation on Oxidation Rates of Silicon in High

Pressure Steam," / . Phys. Chem., 65, 2011 (1961).
species.
predicts that so l o n g as the oxide thickness is smallsmaller than a critical

distancethere w i l l be
33
6. B. E . Deal and A . S. Grove, "'General Relationship for the Thermal Oxidation

of Silicon," J. Appi. Phys.. 36, 3770 (1965).
a c c o r d i n g to the reaction
0 ? O . ~ +
Problems
As
the
oxide
grows beyond this c r i t i c a l thickness, its rate o f g r o w t h w i l l slow d o w n .
12. P. J . Jorgensen. "Effect of an Electric Field on Silicon Oxidation," J. Chem. Phys..

37, 874(1962).
Estimates o f this critical thickness (called extrinsic Debye length) yield

150 A for silicon o x i d a t i o n i n 0
first
and 5 A for o x i d a t i o n i n H 0 .
;
Since the
value is in agreement w i t h the observed thickness below w h i c h
o x i d a t i o n rate is r a p i d in 0 - o x i d a t i o n , and the second is too small to

2
detectable experimentally, this mechanism may
account for
the
be
2.1
Show that when a silicon dioxide film of thickness x, is formed, a layer of silicon
of thickness 0.45x, is consumed.
2.2
A silicon sample has a surface area of 1 cm*.

(a) Calculate the weight of a 0.2-, thick oxide layer on this sample.
" (b) I f the oxide laver contains 5 x 1 0 excess oxygen molecules/cm , calculate
the total weight of the excess oxygen.
experi-
m e n t a l observations.
READING
PROBLEMS
the
ls
REFERENCES
A comprehensive treatment of the oxidation of metals is given by U . R . Evans,

The Corrosion and Oxidation of Metals, Edward .Arnold and C o . , London 1960.
2.3
Calculate (a) the flux of oxidant, and (b) the time required to form a monolayer
of SiOo in dry oxygen at 1200C, when the oxide thickness is 0.2 p.
The oxidation of silicon is reviewed in Volume 7 of a series of review-reports on

"'Integrated Silicon Device Technology" by the Research Triangle Institute, A S D TDR-63-316 (1965).
2.4
Using the general relationship. Equation 2.8. and the data given in the figures,
construct the oxide thickness versus time curves for (a) dry oxygen at 1200C; and
(b) water vapor at 1000C.
2.5
Using the curves obtained above, determine the oxide thickness:

(a) After oxidation for one hour in dry oxygen at 1200C.
(b) After oxidation for 75 minutes in water vapor at 1000C.
(c) After step (a) + step (b).
(d) Alter step (b) + step (a).
2.6
A silicon sample is covered with a 0.2-. thick oxide layer. What is the additional
time required to grow 0.1 ft more oxide in dry oxvgen at 1200C?
REFERENCES
CITED
1. E . L . Jordan, " A Diffusion Mask for Germanium," / . Electrochem. Soc,
108, 478
(1961).
2. P. F . Schmidt and W. Michel, "Anodic Formation of Oxide Films on Silicon,"
J. Electrochem. Soc,
104, 230 (1957).
FLUX
TRANSPORT
DIFFUSED
rate constant in epitaxial growth.
DEVIATIONS
DIFFUSION
EQUATION
LAYERS
FROM
SIMPLE
THEORY
REDISTRIBUTION
DURING
OXIDATION
DIFFUSION T H R O U G H
REDISTRIBUTION
IN
SiO, LAYER
EPITAXIAL
GROWTH
algebra.)
i
ii
The p r i n c i p a l a i m o f semiconductor technology is to control the type and

concentration o f impurities w i t h i n specific regions o f a semiconductor
crystal. The most practical wav of achieving this is through
solid-siaie
diffusion. A s a result, the diffusion of various impurities in semiconductors
has been studied very extensively. Such studies have been facilitated by the
fact that the electrical characteristics o f p-n j u n c t i o n s and other semiconductor devices formed by solid-state diffusion can be directly employed i n
the measurement o f the impurity concentrations. U s i n g such electrical
measurements, concentrations in the range o f 10 parts per b i l l i o n can be
determined relatively accurately.
I n this chapter, we discuss solid-state diffusion, w i t h emphasis on the
diffusion o f i m p u r i t i e s in s i l i c o n . . We begin by deriving the formula g i v i n g
the flux o f a charged species, and the transport equation which determines
their d i s t r i b u t i o n i n a solid. Then we discuss the method o f f o r m a t i o n and
evaluation o f diffused layers, using simple diffusion theory. D e v i a t i o n s
from this theory due to such factors as the two-dimensional geometry o f
planar semiconductor devices, the charged nature o f the impurities, and
external rate l i m i t a t i o n s are then considered.
The second h a l f o f the chapter contains a treatment o f other solid-state
diffusion problems encountered in the fabrication o f semiconductor
devices. These include the redistribution o f impurities d u r i n g t h e r m a l
oxidation, the diffusion o f impurities through a silicon dioxide laver, and
the redistribution o f impurities during epitaxial g r o w t h .
35
Solid-State
36
3.1
Diffusion
Flux
37
the probability o f a j u m p o f any o f these impurities to the valley at x ;

and (iii) the frequency o f attempted j u m p s v. Thus we can write
FLUX
The flux F o f any species is-derined as F = number passing through-unit
area i n unit time.

T o derive a formula for the flux F. let us consider the example o f the
m o t i o n o f positively charged impurities i n a crystal. The atoms o f
the crystal f o r m a series o f potential hills w h i c h impede "the m o t i o n o f
the charged impurities. This c o n d i t i o n is represented in Figure 3.1a. The
= [aC(x -
a)] exp
[W
\ai]
(3.1)
kT
where [aC(x a)] is the density per u n i t area o f particles situated i n the
valley at (x a), and the exponential factor is the probability o f a successful j u m p from the valley at (x a) to the valley at x. Note the lowering o f
the barrier due to the electric field S.
Similar formulas can be written f o r F,, F , and F
When these are
combined to give a f o r m u l a for the flux F at position x, w i t h the concentrations C(x a) approximated by C(x) a(dCjdx),
we obtain
3
F(x) = -[va-e-^ ^
f
dx
^
2kT
[ 2
a r e - " ] C sinh SSl.

2kT
( 3
. )
2
A n extremely i m p o r t a n t l i m i t i n g f o r m o f this equation is obtained for

the case when the electric field is relatively small, i.e., S kT/qa. I n this
case we can expand the cosh and the sinh terms i n the above equation.
N o t i n g that cosh (x) = I and sinh ( x ) == x for x
0, this results i n the
l i m i t i n g f o r m o f the flux equation for a positively charged species,
f ( x ) = - D + p.SC
ox
(3.3)
where
x ii
x ~ a
D s
Distance
(3.3a)
and
Ibl
Fig. 3.1
w/kT
a-e-*
Model of ionic motion within a crystal; potential distribution with and without
H a
(3.3b)
applied bias.
Note that the m o b i l i t y a and the diffusivity D are related by

height o f the potential barrier W is typically o f the order o f electron volts
in most materials. The distance between successive potential barriers a is
o f the order o f the lattice spacing w h i c h is typically several angstroms.
I f a constant electric field is applied, the potential d i s t r i b u t i o n as a funct i o n o f distance w i l l be tilted, as shown i n Figure 3.1b. This w i l l make the
passage o f positively charged particles to the right easier a n d i t w i l l make
their passage to the left more difficult. L e t us n o w calculate the flux F at
p o s i t i o n x. This flux w i l l be the average o f the fluxes at position {x a/2)
and at (x + a/2). I n t u r n , these two fluxes are given by F F, and
F F , respectively, as indicated i n Figure 3.1b.
y
Consider the component F

I t w i l l be given by the product o f (i) the
density per unit areaf o f impurities at the potential valley at (x a); (ii)
v
t This "area' is along a plane which is normal to the direction of the flux.
kT
>-
ft.
(3.4)
<?
This is the w e l l - k n o w n Einstein's
relationship.
It is customary to identify the c o n t r i b u t i o n to the flux which is proportional to the concentration gradient as the diffusion term, while the contrib u t i o n which is p r o p o r t i o n a l to the concentration itself is referred to as the
drift term.
A similar derivation can be made f o r the m o t i o n o f negatively charged
species. Such a derivation leads to an equation similar to E q u a t i o n 3.3
except that the sign o f the drift t e r m is negative. These expressions, as
well as other i m p o r t a n t formulas developed in this chapter, are summarized
in Table 3.3 at the end o f this chapter.
Solid-State
38
Diffusion
0.6
0.7
lOQO/TfKl
Fig.<3.2
Diffusivity of acceptor impurities in silicon.
Fig. 3.3
Experimentally measured values o f the diffusivities o f various i m p u r i ties i n silicon are shown i n Figures 3.2 t h r o u g h 3.4. Diffusivities in silicon
dioxide, in t u r n , are shown i n Figure 3.5. N o t e that the l o g a r i t h m o f the
diffusivity (or, equivalently, the l o g a r i t h m o f the square r o o t o f diffusivity
which we show here for later convenience) plotted versus the reciprocal o f
the absolute temperature gives a good straight line i n all cases. This i m plies that the temperature dependence of the diffusivity can be represented
by the equation
E ltT
D = D e- '
0
in agreement w i t h Equation 3.3a.
(3-5)
0.8
1000/TCK I
0.9
Diffusivity of donor impurities in silicon.
T o assign a physical meaning to the activation energy E we must k n o w

the particular atomistic mechanism o f diffusion. I t is k n o w n that i n semiconductors diffusion o f substitutional
impurities, i.e., impurities w h i c h
occupy sites i n the silicon lattice, usually proceeds" by the i m p u r i t i e s
j u m p i n g i n t o silicon vacancies in the lattice. Thus the activation energy
corresponds to the energy required to f o r m a silicon vacancy rather than to
the energy required to move the i m p u r i t y . Since silicon-to-silicon bonds
must be broken to form a vacancy, we m i g h t expect a relationship between
the energy gap o f a semiconductor (see Chapter 4) and the activation
a
Solid-State
40
Diffusion
The T r a n s p o r t Equation
energy o f the diffusivities o f substitutional impurities. I n fact, the activation energy for the diffusivities o f acceptor and d o n o r type impurities,
which are all substitutional, ail ranse between 3 and 4 ev in silicon, whereas
the similar activation energies for diffusion in germanium all range between
2 and 3 evthree to four times the energy gaps o f Si and Ge, respectivelv.
Other impurities occupy the space between the silicon atoms, instead o f
the lattice sites in thexrystal. Such interstitial impurities (e.g., manganese
and nickel in silicon) generally move much more rapidly than substitutional
impurities, as is evident from Figure 3.4. The atomistic mechanism o f the
41
bo
ooo o
o o
o
o
<no
o
o
CJ
10
1.5
2.0
2.5
3.0
3.5
lOOO/rCK)
Fig. 3.S
Diffusivities in amorphous S I O . .
diffusion o f gold i n silicon is quite complex; it involves the m o t i o n o f

interstitial gold, but is influenced also by the concentration o f vacancks.'
3.2
THE TRANSPORT EQUATION
Consider now a small element of a solid such as shown in Figure 3.6.

A n inventory o f the material contained w i t h i n the element bounded by
the planes at position x and at position x + A x yields
Increase i n the density o f material per unit area _ _
UXJO/TVKl
Fig. 3.4
Diffusivity of miscellaneous impurities in silicon.
U n i t time
T
^ in
t out,
provided material is neither formed nor consumed within this element.
Solid-State
42
Diffusion
I n terms of concentrations, we can write

Ax
= F(x) dt
F(x + Ax)
.(3.6)
where C is the average concentration w i t h i n the element.

F(x
+ Ax) -
and
F{x)
dx
so that
(3.7)
dt
ox
which is the general form, o f the transport
equation
i n one dimension.!
Unit cross-sectional
area
DIFFUSED
LAYERS
Solid-state diffusion is most frequently employed i n semiconductor

technology to f o r m diffused layers o f impurities. Typically, diffused layers
are formed i n a tworstep process. I n the first step, impurities are i n t r o duced i n t o the semiconductor to a depth of a few tenths o f a m i c r o n . T h i s
process is called the predeposition step. Once the impurities have been
introduced i n t o the semiconductor, they are then diffused deeper to provide
a suitable concentration d i s t r i b u t i o n w i t h o u t any more impurities being
added to the semiconductor. T h i s second step is called the drive-in
diffusion step.
a.
'Fix + A.xl
Rvl
43
T h e c o n d u c t i o n - o f heat in solids is described by an identical equation

except that the diffusivity D is replaced by the thermal diffusivity. k. T h u s
we can use the solutions o f heat conduction p r o b l e m s in solid-state
diffusion w o r k .
3.3
dF
Ax
Layers
N o w , i f we let
Ax 0 ,
CC(x),
Diffused
T h e P r e d e p o s i t i o n Step
The predeposition step is typically performed by placing the semiconductor sample i n a furnace t h r o u g h which flows an inert gas containing
the desired i m p u r i t i e s . The furnace and flow conditions are similar to
t h o s ; used in thermal oxidation (Chapter 2). The temperature usually
ranges between 800 and 1200C.
6
A.v
Fig. 3.6
The control element used in the derivation of the transport equation.
The impurities can be introduced into the carrier gas in several w a v s :

I f we n o w substitute our f o r m u l a for the flux o f a positively charged
species, Equation 3.3, and i f we assume that the diffusivity is independent
of the concentration, we arrive at a simpler f o r m o f the transport e q u a t i o n
dt
= D
dx
n
1
- p.S '
ox
[<o and D = const.].
I n most o f this chapter we concern ourselves w i t h problems
(3.8)
whose
solutions we can approximate by assuiiung that the electric field S = 0.

This results in the diffusion
equation,
2 == ^
dt
dx'
[ / = 0;
D const.].
(3.9)
S o l i d S o u r c e . I n one arrangement the i m p u r i t y is introduced i n t o the

carrier gas f r o m a solid source w h i c h evaporates. I f e q u i l i b r i u m is reached
at the source, the partial pressure o f the i m p u r i t y compound in the gas
w i l l equal its vapor pressure at the source temperature.
A solid source can be used in the predeposition o f most impurities. T h e
source material is usually an oxide o f the i m p u r i t y , e.g., P , 0 for phosphorus, B 0 for b o r o n . A s 0 for arsenic. S b 0 for antimony. The
oxide is carried to the semiconductor surface by the gas and then is reduced
at the surface. A n example o f this reducing reaction m i g h t be
5
P-ACgas) + Si(solid) P + SiO (solid).

=
t In three dimensions Equation 3.7 would become

dC
dt
-divF.
Thus an oxide layer is formed on the silicon surface d u r i n g predeposition.

L i q u i d S o u r c e . A liquid c o m p o u n d can also be used as a source o f
the impurities by passing the carrier gas through a flask containing the
Solid-btace
44
Uittusion
l i q u i d . The flask and the l i q u i d in i t are kept at a fixed temperature. The

gas w h i c h bubbles through the l i q u i d reaches e q u i l i b r i u m w i t h the liquid
w i t h the result that the partial pressure o f the i m p u r i t y c o m p o u n d i n the
gas leaving the flask w i l l equal its vapor pressure at the temperature o f the
l i q u i d . L i q u i d sources are available for several impurities, e.g.. P O C l for
phosphorus predeposition. and B B r for b o r o n predeposition.
3
C h e m i c a l T r a n s p o r t . A t h i r d method o f i n t r o d u c i n g an i m p u r i t y
i n t o the gas stream is t h r o u g h a chemical reaction t a k i n g place at the
source. A n example o f this chemical transport process is that o f gallium.
The source employed i n gallium predeposition can be g a l l i u m oxide, G a 0 .
I t has been f o u n d that gallium is introduced into the carrier gas only i n
the presence o f hydrogen. This specificity indicates t h a t a chemical
reaction takes place at the source. One reasonable reaction m i g h t be
;
Ga 0 (solid) + 3 H
2
2Ga(vapor) + 3 H . 0 .
F o r this reaction the partial pressure o f g a l l i u m vapor s h o u l d be given by

Paa. 3C I
v
J
Ph,o'
3
according to the law o f mass action. Experiments show that the gallium
concentration is, in fact, p r o p o r t i o n a l to the I.5-power o f the hydrogen-towater ratio, indicating that this reaction is the correct one.
I f there is no resistance to the transport o f the i m p u r i t y f r o m the gas to
the solid surface, the surface concentration C i n the solid w i l l be in equilibr i u m w i t h the partial pressure p o f the i m p u r i t y i n the gas. A c c o r d i n g l y ,
i f Henry's law is obeyed (see Chapter 2), we expect that
C = Hp.
s
There is, however, a m a x i m u m concentration o f any i m p u r i t y that can be

accommodated i n a solid at any given temperature. This concentration is
called the solid solubility o f the i m p u r i t y . Thus the above relationship
can h o l d only up to the l i m i t o f the solid solubility o f the i m p u r i t y i n the
semiconductor at the temperature o f the predeposition. Solid solubilities
o f i m p o r t a n t impurities i n silicon are shown i n Figure 3.7 as a function o f
temperature.
The relationship between surface concentration and p a r t i a l pressure is
illustrated by Figure 3.8 w h i c h shows the measured surface concentration
o f b o r o n i n silicon as a function o f the partial pressure o f boron-oxide i n
the carrier gas for an 1100C predeposition.
I t is seen t h a t the surface
concentration is indeed p r o p o r t i o n a l to the partial pressure i n the gas at
low concentrations, w i t h an eventual saturation expected at the solid
solubility l i m i t .
11
I n industrial practice, predeposition is usually performed with a partial

pressure o f the impurity in the carrier gas high enough so that the surface
concentration i n the semiconductor corresponds to the solid solubility o f
the i m p u r i t y . Under such conditions the surface concentration is independent o f the partial pressure o f the i m p u r i t y and w i l l be reproducible and
u n i f o r m even i f the partial pressure is not.
We can now proceed to the consideration o f the i m p u r i t y distribution i n
the semiconductor resulting after predeposition. Since during.predeposit i o n the surface concentration C o f the i m p u r i t y is kept constant, we must
solve the diffusion equation (3.9) subject to the'boundary conditions
s
C(0, 0 =
(3.10)
Solid-State
-46
Diffusion
47
1
Solid-solubiliiv of
3 x 10
iIr
2 0
B in Si ai 1I00-C
Line corresponds to Henry's law.

with fi 2 x 1 0 atm/cm .
: i
i.5
2 x icr-
p (Torri
s
Fig. 3.8 Surface concentration of boron in silicon as a function of the partial pressure
of B 0 in the ambient, at I IOOC.
n
and
C(co, r) = 0
(3.11)
C{x, 0) = 0.
(3.12)
and the initial condition
The solution satisfying the diffusion equation as well as the above conditions is the complementary error function,*
1
C(x.
t) = C
erfc
2,/Dr
(3.13)
w h c h is one o f the most i m p o r t a n t functions encountered in diffusion

theory. Some o f its properties are summarized in Table 3.1. I t is tabulated in Carslaw and Jaeger' as well as in many mathematics handbooks.
The concentration d i s t r i b u t i o n resulting i n a predeposition process
according to the above simple theory is shown in Figure 3.9 where we plot,
on b o t h linear and logarithmic scales, the normalized concentration
(normalized to the surface concentration, which is constant) as a function
o distance for three different values o f the diffusion length 2 \ Dl corresponding to three consecutive predeposition times. N o t e that as the
predeposition time progresses, the i m p u r i t y penetrates deeper and deeper
into the solid. This penetration can be represented by a useful quantity,
f
Xl/ll
2(0
\C{x,
Jo
Integration o f E q u a t i o n 3.33 vields
2(0 =
~_
\ "
t) dx.
y D~t C
(3.14)
(3.15)
Diffused Layers
49
48
TABLE
3.1
b.
Drive-In Diffusion
SOME PROPERTIES OF T H E ERROR FUNCTION
erf (x) =
V 77 Jo
As discussed above, predepositions usually result in surface concentrations which equal the solid solubility o f the impurity. It is often necessary to lower the surface concentration from this value. Furthermore, it is
often desirable to move the impurities deeper into the semiconductor
w i t h o u t at the same time increasing the total number o f impurities w i t h i n
the semiconductor.
da
erfc (x) == 1 erf (x)

en (0) = 0
Both purposes can be accomplished by a further high-temperature heat

treatment in a gas which does not contain any impurities. I n practice, this
additional heat treatment, the drive-in diffusion, is usually carried out i n
an oxidizing ambient.
A comparison o f the data shown i n Figures 3.2 and 3.3 w i t h those
shown in Figure 3.5 indicates that most acceptor- and donor-type i m p u r i ties diffuse much faster in silicon than i n silicon dioxide. ( G a l l i u m is a
notable exception.) Thus, i f the heat treatment is performed i n an o x i dizing ambient, the resulting oxide layer w i l l tend to seal these impurities
into the silicon; no more impurities w i l l be added nor will a significant
amount o f impurities escape because the oxide acts as an effective mask
against the diffusion o f these impurities.
erf ( x ) m i
i
erf (x) 35 = x
for
x 1
1 e~ '
ertc (x) 3 =
v rr x
d'erf(x)
2
for
erfc ( V ) ax' = x erfc x H
x 1
= ( 1 -<?*")
The concentration d i s t r i b u t i o n after drive-in diffusion is accordingly

given by that solution o f the diffusion equation which satisfies the boundary
conditions
erfc (x) rfx = =
T h i s expression can be interpreted simply as follows.

0(f)
represents the area
curves o f the top p o r t i o n o f Figure 3.9. These distributions can, i n t u r n ,

be a p p r o x i m a t e d by triangles whose height is C
-~2\
Dt.
This leads to Q(f) = \ Dt C ,
and whose base is
which is reasonably close to the
dx
The quantity
under one o f the concentration d i s t r i b u t i o n
exact f o r m . E q u a t i o n 3.15.
A n o t h e r quantity o f considerable importance i n determining the charac-
= 0
and
C(co,r) = 0
t r a t i o n dC'idx.
I t can be obtained by differentiating Equation 3.13. This
leads to
5C
HDl
Usins
the
asvmptotic
(3.16)
i-Dt
dx
a p p r o x i m a t i o n for
the
complementary
error
f u n c t i o n given'in Table 3.1, which is valid under most practical conditions,

this can be rewritten as
3C
I
(x.t)
2Dt
C(x, f ) .
(3.17)
(3.19)
which provide for a constant 0 being maintained in the semiconductor

during the drive-in diffusion step. The initial c o n d i t i o n is given by
C(x, 0) = C erfc
s
teristics o f semiconductor devices is the gradient o f the i m p u r i t y concen-
(3.18)
(o.n
-\
*
(-Ôpredep
(3.20)
since the impurity d i s t r i b u t i o n at the beginning o f the drive-in diffusion

step is the impurity d i s t r i b u t i o n resulting at the conclusion o f the predeposition step.
This is a difficult problem to solve. However, for most practical cases,
v Dt for drive-in diffusion is much larger than \ Dt for predeposition.
Thus we can regard the extent o f penetration o f the predeposited profile to
be negligibly small in comparison to that o f the final profile resulting after
the drive-in diffusion step. Accordingly, we can represent the predeposition profile mathematically "by a delta function.
50
Solid-State
Diffusion
Diffused
Layers
SI
The concentration d i s t r i b u t i o n after drive-in diffusion subject to this

approximation then w i l l be given b y
5
iD
C ,<)
= -S=e->"-' '.
{X
(3.21)
This is the well-known Gaussian d i s t r i b u t i o n . N o t e that in this solution

the t o t a l amount o f impurities in the solid, Q. is constant by virtue o f the
boundary condition E q u a t i o n 3.18. Since the impurities move i n t o
"the solid with the passage o f t i m e , i n order to keep the total O constant
the concentration near the surface must d r o p . Indeed, it is evident
from Equation 3.21 that the surface concentration is given by
C (0
s
= -$=
.
/ IT VI
(3.22)
Thus the concentration d i s t r i b u t i o n can also be represented by

: 1D
C(x. f) = C (t)e-* ' >.

s
(3.23.)
The distribution o f impurities for a Gaussian profile is shown in Figure

3.10 where we plot the concentration o f the impurities normalized by 0,
which is a constant, as a function o f distance for three increasing diffusion
lengths or three successive drive-in diffusion limes. N o t e that the slope o f
each impurity distribution at x = 0 is zero corresponding to the first
boundary condition, and that the diffusion profile spreads inward into
the semiconductor, w i t h the surface concentration dropping as the
diffusion time increases.
The gradient of the i m p u r i t y concentration i s o b t a i n e d by differentiating
Equation 3.23 and is
d
ox
C(x. f).
iz.o
(3.24)
2Dt
I t is important to note that for b o t h the complementary error-function

and the Gaussian distributions the concentration is a function o f a
normalized distance, xjl\' Di. T h u s , i f we normalize the concentration
w i t h the surface concentration, we can represent the distributions w i t h a
single curve each, valid for all times, as shown i n Figure 3.11. I t is i m p o r t ant to realize that i n the case o f the Gaussian distribution we n o w have
also introduced the time variable i n t o our other normalizing parameter,
the surface concentration.
Let us now consider the exact solution o f this problem w i t h o u t the
delta-function a p p r o x i m a t i o n to the initial d i s t r i b u t i o n . Such a s o l u t i o n
12
xl.ul
Fig. 3.10
The Gaussian: normalized concentration versus distance for successive times.
is shown i n Figure 3.12. in comparison to the corresponding d i s t r i b u t i o n

obtained w i t h the delta-function a p p r o x i m a t i o n . I n this figure the concentration normalized by the constant surface concentration present d u r i n g
the predeposition step is plotted as a function o f the distance normalized
w i t h the diffusion length o f the drive-in diffusion step. The parameter is
the ratio o f the diffusion length o f the predeposition step to that o f the
drive-in diffusion step. N o t e that for values o f this ratio as large as 0.25,
the delta-function a p p r o x i m a t i o n provides an excellent fit to the more
53
Fig. 3.12 Two-step diffusion: predeposition followed by drive-in diffusion. Solid lines
are exact final distributions.
Dashed lines correspond to case when predeposited
distribution is approximated by delta-function at x = 0.
12
XP*/**
Fig. 3.U
The diffusion functions: norma
lized concentration versus normalized distance.
etching the surface w i t h a solution w h i c h reacts selectively w i t h the t w o "

sides o f the j u n c t i o n , thus resulting i n a c o l o r variation between the
p and the n sides as illustrated i n Figure 3.13. T h i s method makes possible
the determination o f the j u n c t i o n depth X) w i t h about a 0.1-micron
accuracy using optical techniques.
13
exact results. Only when this ratio becomes o f the order o f 1, does thedelta-function a p p r o x i m a t i o n begin to lead to large errors.
The j u n c t i o n depth x as illustrated i n Figure 3.13, is denned by the

condition
C(* 0 = C .
(3.25)
c.
E v a l u a t i o n of Diffused L a y e r s
Routine evaluation o f diffused layers is relatively simple i f the diffused

laver forms a p-n j u n c t i o n w i t h the underlying substrate. The depth o f
this j u n c t i o n x, can be delineated by grooving into the semiconductor and .
Thus, i f the j u n c t i o n depth and the b u l k concentration C are k n o w n ,

the surface concentration C as well as the rest o f the impurity d i s t r i b u t i o n
can be calculated, provided we can safely assume that the concentration
distribution follows one or the other simple diffusion functions. A n
B
Solid-State
54
Diffusion
logC
Diffused
Layers
55
which we can rearrange to yield

C* -
Cj, + 4 " .
Q.u p
(3.28)
I f we take p = 85 and fi = 45 c m / v sec which correspond to the l i m i t i n g

mobility values reached at high i m p u r i t y concentrations in silicon, such
as we are likely to encounter i n the case o f diffused layers (see Chapter 4 ) ,
ri
Fig. 3.13
Determination of junction depth by grooving and staining.
i m p o r t a n t and convenient check on the validity o f this assumption is

provided by a measurement o f the average resistivity o f the diffused
layer. The reciprocal of the average resistivity p o f an n-type layer (taken as
example) is
l = W n ( x ) d x
p
Xj Jo
(3.26)
where we are essentially adding up the parallel conductances o f the elem e n t a l segments of the diffused layer as indicated i n Figure 3.14.
T o evaluate the integral appearing i n this equation, we must have
(i) a relationship between the m o b i l i t y and the i m p u r i t y concentration,
and (ii) the carrier concentration as a f u n c t i o n o f distance. F o r complete
ionization o f impurities (see Chapter 4), this w i l l be approximately the
same as the distribution o f the impurities.
W e n o w illustrate the procedure o f evaluating p by considering the
simplest o f all possible distributions: t h a t corresponding to a " b o x "
profile illustrated i n the inset to Figure 3.15. I n the case o f a "box'"
distribution the,integral in E q u a t i o n 3.26 can be evaluated w i t h ease and
results i n
\ = qu{C
P
C)
B
(3.27)
Ji (fl-cm i
Fig. 3.15
Surface concentration versus average resistivity"box" distribution.
we can obtain the relationships between surface concentration and

average resistivity for a " b o x " d i s t r i b u t i o n for both n- and />-type lavers.
These are shown in Figure 3.15.
The box d i s t r i b u t i o n , o f course, is not usually a realistic one. The
integral in E q u a t i o n 3.26 has also been evaluated numerically for the t w o
diffusion functions, the complementary error function and the Gaussian,
for the case o f s i l i c o n .
The resulting curves, analogous to the one in
Figure 3.15 for the "box*" d i s t r i b u t i o n , are shown in Figure 3.16. where
we again show surface concentration C as a function o f average resistivity
f . for different substrate concentrations C,.. Note that these curves,
which are based on computer calculations and take the cxperimentaliv
14
Fig 3.14
Determination of the average resistivity of diffused layers.
1 lO-cm)
Fig. 3.16c
tribution.
14
Surface concentration versus average resistivity p-Type layer, erfc dis-
/" lO-cml
(d)
Fig. 3.16b Surface concentration versus average resistivity"n-Type layer, Gaussian
distribution.
y
"
56
Fig. 3.l6d Surface concentration versus average resistivity"p-Type layer, Gaussian

distribution.
57
Solid-State
58
Diffusion
measured variation o f the electron and hole mobilities with i m p u r i t y concentration into account (see Chapter 4). have the same general tendency as
those obtained by the crude " b o x " profile
approximation.
The average resistivity o f diffused layers is
customarily measured by the four-point
probe
technique illustrated i n Figure 3.17. I n this
technique four probes are placed on the surface
of the semiconductor. Current is passed t h r o u g h
the t w o outermost probes and the voltagethe
IR dropdue to this current flow is measured
between the two middle probes. Thus any p r o b lems due to probe-to-semiconductor contact
resistances are eliminated. I n order to relate
the measurements o f the voltage d r o p V and
the current / to the average resistivity p o f the
diffused laver we must solve the electrostatic
Top view
problem which takes into account the spreading
Fig. 3.17 Measurement of
of the current from the t w o outside probes as
average resistivity using a
shown in the top view i n Figure 3.17. T h i s
four-point probe.
has been done and the result for layers whose
lateral dimensions are large and whose w i d t h x> is small in comparison to
the spacing of the probes i s
15
l c
77 V
P =
ln2 7
4.52
j
(3.29)
I '
Since a typical probe spacing is s = 1000 u. this condition is amply met

for diffused layers.t I f surface concentrations based on measurements o f
the average resistivity agree w i t h those obtained from j u n c t i o n depth
measurements, we can conclude that our diffusion profile does indeed
follow the assumed distribution.
3.4
DEVIATIONS
THEORY
FROM
SIMPLE
DIFFUSION
We now examine some o f the causes o f deviation of the i m p u r i t y

distributions from the simple theory considered in Section 3.3.
f Because semiconductor wafers are generally sliced 200 to 400. thick. Equation 3.29
can also be used to calculate wafer resistivity from four-point probe measurements bv
substituting the wafer thickness for .r,. If either the wafer thickness or ihe lateral
dimension of the semiconductor sample is comparable to the probe spacing, corrccuon
factors must be applied."
Deviations f r o m S i m p l e Diffusion T h e o r y
a.
T w o - D i m e n s i o n a l Diffused Layers
I n the planar technology, the diffusion process is usually carried out

through w i n d o w s cut in a diffusion mask (usually an S i O , layer). A s a
result, the i m p u r i t i e s w i l l diffuse parallel \o the surface as well as n o r m a l to
the surface. The resulting j u n c t i o n shape near the edge o f the w i n d o w is
observed experimentally to be approximately cylindrical.
17
The results o f detailed calculations of this two-dimensional diffusion

problem for the predeposition case are shown in Figure 3.18. Here we
plot the n o r m a l i z e d concentration distribution as a function o f distance
parallel to the surface from the edge o f the opening, and also as a f u n c t i o n
of distance n o r m a l to the surface. The latter is identical w i t h the c o m plementary error-function type o f distribution. The d i s t r i b u t i o n parallel
to the surface deviates from the error-function d i s t r i b u t i o n b u t this
deviation leads t o no more than about 15 or 2 0 % error in the location o f a
j u n c t i o n . T h u s , w i t h i n this error, the j u n c t i o n shape can indeed be
regarded as cylindrical. The calculations for the case o f a drive-in diffusion lead to similar conclusions.
b.
F i e l d - A i d e d Diffusion
W h e n d o n o r or acceptor impurities enter the silicon crystal, they

become ionized. Consequently, we have to consider the simultaneous
m o t i o n o f the ionized donors (or acceptors) and electrons (or holes).
We have already discussed the simultaneous m o t i o n o f t w o charged
species in S i 0 in connection w i t h oxidation kinetics i n Chapter 2. We
concluded then that whenever t w o charged species which have different
mobilities move simultaneously, a built-in electric field w i l l result, aiding
the m o t i o n o f the slower o f the t w o species.
2
Detailed consideration o f the problem o f the m o t i o n o f d o n o r or

acceptor ions i n a semiconductor leads to a modified formula for the
flux o f the ions.
_
_
16
F =
-D 1 +
.
I f g
v'.l + 4(n,/C)- 02-
(3.30)
where C is the concentration o f the ions and , denotes the concentration

of electrons or holes i n a semiconductor containing no donors or acceptors
at a l l , at the temperature o f diffusion. (This quantity, called the intrinsiccarrier concentration, is discussed in Chapter 4.)
E q u a t i o n (3.30i can also be written in the form
F
d
D
S ox
(3.3D
60
wnere
D
=D
61
Deviations from Simple Diffusion T h e o r y
We can now consider the two limiting values o f the effective diffusivitv
D,, . A t a given temperature, hence at a given value o f n , i f the impurity
concentration is relatively low. i.e.. C n D, = D. Thus, the built-in
field has no effect. I n the opposite case, when the impurity concentration
is high, i.e.. C n D
= 2D. Thus, the electric field can bring about
an effective d o u b l i n g o f the diffusion coefficient.
((
\ I -r
Mn.,0--
is the effective diffusivity o f the impurities, which incorporates the effect

o f the built-in electric field. Note that this effective diffusivity is a function
o f the concentration.
l
e t f
The effect o f the built-in field on the shape o f the concentration distrib u t i o n is illustrated in Figure 3.19 where the results o f numerical calcul a t i o n s based on the equations for field-aided diffusion, are shown for a
particular case. F o r comparison, we also show a complementary errorfunction distribution, which does not take the field-aiding effect into
consideration and which is fitted to the correct distribution at low concentrations. I t is apparent that the field-aiding effect brings about asmaller gradient near the surface. This is because the electric field aids
the m o t i o n o f the ions to the largest extent in the region o f highest impurity
concentration. Thus i n this region the concentration gradient does not
need to be so steep to maintain the same- flux flowing into the semiconductor.
18
O(etfc)
Parallel to s u r f a c e ( 2 )
x, \ Dc
19
Experimentally measured
concentration distributions o f b o r o n i n
silicon under various surface-concentration conditions are shown in
Fiaure 3.20. I t is evident that for the lowest surface concentration the
distribution closely follows the complementary error-function distribution. I t also appears that the d i s t r i b u t i o n begins to deviate from the
error-function shape as the surface concentration exceeds 10 c m . (Note
that n i n silicon at 1100C, the diffusion temperature, is approximately
10 c m . ) The intermediate profile could be accounted for in terms o f the
field-aided diffusion effect as shown in Figure 3.19. However, as the
surface concentration o f boron is increased further to exceed 10 c m ,
the profile begins to deviate from the complementary error-function distrib u t i o n to a much larger extent than could be accounted for i n terms o f the
field-aiding
effect. I t is evident that we must consider other causes o f
deviation in such a case.
19
- 3
19
- 3
20
c.
.0
Fig. 3.I8 Comparison of the concentration distributions{T) normal, and ( 5 ) parallel to

the surface, for conditions corresponding to predeposition."
- 3
Effect of L a t t i c e S t r a i n
One important anomaly that is observed when the surface concentration

exceeds a certain value is the c o m m o n l y k n o w n "emitter-push" effect.
( I t is alternately called the " e m i t t e r - d i p " effect or "cooperative diffusion"
effect.) This phenomenon is illustrated in Figure 3.21. I t consists o f the
enhanced diffusion o f a diffused layer i n regions where a second, highconcentration diffused layer penetrates the silicon. I t is often observed in
transistor structures which, by necessity, must employ two consecutive
62
Deviations f r o m S i m p l e Diffusion T h e o r y
63
* (/<)
Fig. 3.20 Experimentally measured distribution of boron in silicon for various surface
concentrations."
Fig. 3.19 Illustration of the effect of a built-in field on the concentration distribution
in a semiconductor."
diffusions w i t h the second one, the emitter, having a high surface concent r a t i o n (hence the name).
I n Figure 3.21 the b o r o n d i s t r i b u t i o n in silicon is shown both under a
high concentration phosphorus-diffused region and elsewhere in the
sample. I t is evident that the phosphorus diffusion "pushes" the b o r o n

d i s t r i b u t i o n ahead o f it. This brings about the irregular j u n c t i o n b o u n d a r y
illustrated i n the inset.
Several possible explanations o f this effect have been p r o p o s e d .
It
appears most likely that straining o f the lattice by the high concentration o f
impurities is involved. This p r o p o s i t i o n was supported by an e x p e r i m e n t
20
20
uixuaiun
d.
65
i neory
E x t e r n a l Rate L i m i t a t i o n of S o l i d - S t a t e Diffusion
I n our treatment o f the problem o f predeposition. we have assumed

that the surface concentration o f the i m p u r i t y C is a constant throuahout
the time o f predeposition. Initially, however, there are no impurities in the
semiconductor and therefore the surface concentration must also be zero.
Because physically the surface concentration c a n n o t j u m p instantaneously
from zero to its final constant value, there w i l l have to be a period "of
adjustment during which the simple boundary condition C(0. /) = C
will not hold. The duration o f this initial period is determined by the
relative rates o f (i) transport o f impurities from the bulk of the gas to the
surface, and (ii) solid-state diffusion f r o m the surface into the interior o f
the semiconductor. Thus an external l i m i t a t i o n to the rate o f transport o f
the impurities will influence the solid-state diffusion process.
s
We next consider the solution o f the predeposition problem subject to

such an external rate limitation. The initial c o n d i t i o n subject to which we
have solved the predeposition problem.
Ox.
0) = 0
(3.32)
as well as one o f the boundary conditions.

C(.r)-0
(3.33)
remain unchanged. However, we now replace the second boundary

condition, that of a constant surface concentration, with the condition
h[C
X I/II
Fig. 3.21
The "emitter-push" effect: anomalous diffusion due to a second, high con-
centration, diffusion.-
C(0, f ) ] -
-Dpox
(3.34)
(o-.o
which specifies that the flux to the surface via mass transfer in the gas w i l l
equal the flux away from the surface via solid-state diffusion. As" i n
Chapter 2, h denotes the gas-phase mass-transfer coefficient in terms- ofconcentrations in the solid.
The solution to this p r o b l e m
21
C*j
_
=
r r
x
r
e r f c
__ AhtlVDIY.
wi v Dty.
e
DtUziWD
lhtl\
, v 0.11/!
v on
M
2^'Dt
ht
x_
Dt
' 2 ' ] D i - l
e r f c
(3.35)
in w h i c h a pressure probe was placed on the silicon surface d u r i n g the
diffusion process. I t was found that under the pressure probe the i m p u r i ties d i d indeed penetrate deeper into the silicon than elsewhere.
T h i s effect leads to a very serious l i m i t a t i o n in the fabrication o f highfrequency transistors which, as we shall see later, require close spacing
between two diffusion fronts.
This solution consists o f a complementary error function from which a

correction term is subtracted.
As the parameter htjsi Dt becomes very
large, the subtracted term vanishes.
Calculations based on Equation 3.35
are shown in Figure 3.22. I t is evident that as /if/v Dt approaches approximately 10, the solution merges into the complementary error function.
66
Solid-State
Diffusion
67
Deviations from Simple Diffusion T h e o r y
Thus the dimensionless parameter htj\ Dt provides a simple estimate

of the period o f time during which this adjustment o f the surface concentration takes place. This time w i l l be o f the order o f / ^ 100(Z),/;-). A s
an estimate, we can take h = 5 cm/sec for the gas-phase mass-transfer
coefficient, f r o m Chapter 1. The mass-transfer coefficient in terms
o f concentrations i n the solid h = h IHkT.
I f we base the value
of the Henrv's law constant on the b o r o n data shown in Figure 3.8.
H = 2 x 1 0 atm/'cm , we get / i c ^ 2 x 1 0 cm/sec, or about 70 p/hr. F o r
D ^ 1 /v /hr, this leads to r ^ 0.02 hr or a little over a minute. Since predepositions are often o f onlv a few minutes d u r a t i o n , the influence o f gasphase rate l i m i t a t i o n may become appreciable.
G
25
- 6
The complementary p r o b l e m to that o f predeposition is the escape o f

impurities f r o m a semiconductor, often referred to as " o u t - d i f f u s i o n . "
The initial c o n d i t i o n for this p r o b l e m w i l l be a u n i f o r m d i s t r i b u t i o n corresponding t o the bulk concentration C .
B
C(x, 0) = C
(3.36)
and the boundary conditions w i l l be

C(cc. t) = C
(3.37)
and
;iC(0, t) = - D ^
(3.38)
CX 1(0.(1
Here we assume that the concentration o f the i m p u r i t i e s in the b u l k o f the

gas is zero. The solution o f this p r o b l e m is very similar to the one for the
case o f predeposition. I t is given b y
2 1
ll _
C
__
[ht:\DtY M!XDtHzl2\~Dt)
e r
2 /.Dr '
D: ' 2
I
I)J'
(3.39)
N o t e that this solution consists o f a simple error function to w h i c h a

term dependent upon the parameter hlj\
Dt is added.
A s this parameter
becomes large, the additional term becomes very small.

based on E q u a t i o n 3.39 are shown in Figure 3.23.
solution converges to the simple error function for htl
Experimental measurements
22
Calculations
I t is evident that the

Dt > 10.
o f the i m p u r i t y d i s t r i b u t i o n in a silicon
sample resulting after exposure to a hydrogen ambient are shown i n

Figure 3.24. I t is evident that most o f the experimental points lie below the
theoretical curve corresponding to htj\
Dt = 5. F o r the time and temper-
ature o f these experiments, this indicates that the value o f h in this case is
larger than about 1 0 / h r . which is consistent w i t h the above estimate o f
70^/hr.
n e a i s t n o u t i o n 01 impurities during
0.2
0.4
0.6
0.8
I nermal Oxidation
10
1.2
1.4
1.6
6?
1.8
2.0
.v. 2 Di
N
Fig. 3.24 Outdiffusion of impurities from silicon after exposure to H . ambient for 30
minutes ac I27SC.'
:::
3.S
REDISTRIBUTION O F IMPURITIES
DURING THERMAL OXIDATION
Whereas the previous considerations dealt w i t h deviations from the

boundary condition applied during the predeposition step, in this section
we consider a phenomenon which can cause deviations from the boundarv
condition applied in our treatment o f the drive-in diffusion step, BClox = 0
at x = 0.
To arrive at this boundary c o n d i t i o n we have reasoned that, because
most impurities diffuse much slower in silicon dioxide than in silicon, the
oxide layer grown d u r i n g the drive-in diffusion step will effectively seal the
impurities into the silicon. In reality, the situation at the g r o w i n g silicon
dioxide-silicon interface is considerably more complex.
It has been f o u n d - that impurities in the silicon will be redistributed
near a growing thermal oxide. This is due to several factors. I f any two
phasessolid, gas. or liquidare brought into contact, an i m p u r i t y contained in one o f the two phases w i l l be redistributed between the two u n t i l
equilibrium is reached. I n equilibrium, the ratio o f the concentrations in
the two phases will be a constant. The ratio o f the equilibrium concentration in the silicon to that in the silicon dioxide is denoted by the term
segregation coefficient and is defined as
3
E q u i l i b r i u m concentration o f i m p u r i t v in silicon
m -
.
E q u i l i b r i u m concentration o f impurity in SiO.-.
Another factor that influences the process o f impurity redistribution
during thermal oxidation-is that the i m p u r i t y may have a tendency to
70
escape through the oxide layer. I f the diffusivity o f the i m p u r i t y i n the

oxide is small, this factor w i l l be u n i m p o r t a n t ; however, i f the diffusivity
is relatively large, this factor can significantly affect the i m p u r i t y distrib u t i o n in the silicon.
A t h i r d factor entering into the redistribution process is the fact that the
oxide layer is growing and therefore the boundary separating the oxide
Oxide takes
Redistribution of I m p u r i t i e s during T h e r m a l
71
Oxidation
24
F o u r possible cases o f the redistribution process are illustrated i n

Figure 3.25. These cases can be considered in t w o groups: in one. the
oxide has a tendency to take up impurity, and in the other the oxide has a
tendency to reject the i m p u r i t y . I n each case the situation can be different
for slow and for fast diffusion in the oxide. I n Figure 3.25 an experimentally
observed example is listed for each of the four cases.
We n o w consider the starting premise and the results o f the theory o f the
redistribution process. This theory is based on a solution o f the diffusion
equation w i t h a m o v i n g boundary and describes the redistribution o f an
initially u n i f o r m d o p i n g concentration in the silicon. The boundary c o n d i tions applied to the problem are: (i) the concentration o f the i m p u r i t y at
the gas-oxide interface is a constant, C; (ii) far w i t h i n the silicon substrate the concentration approaches the bulk concentration C .
In
a d d i t i o n , we have to satisfy two matching conditions. The first is that the
concentrations at the t w o sides o f the oxide-silicon interface must be i n the
ratio prescribed by the segregation coefficient m, and the second is that, as
the oxide grows, the i m p u r i t y must be conserved at the m o v i n g oxidesilicon interface. I t is also assumed that the oxide growth w i l l proceed
p r o p o r t i o n a l l y to the square root o f the oxidation time, i.e.. x =
Bt
where B is the parabolic oxidation rate constant (see Chapter 2).
24
up impurity |m < 1)
(a) Diffusion in oxide slow

(boron)
(b) Diffusion in oxide fast

(boron with H , ambient)
Oxide rejects impurity |m >
]|
The solution giving the i m p u r i t v distributions in the silicon and in the

silicon dioxide laver is in terms o f error functions and complementary
error functions. I t is not reproduced here. The most i m p o r t a n t feature
of this solution is the resulting formula for the concentration o f the i m p u r ity on the silicon side o f the interface.
Cg
/i)
(cl Diffusion in oxide slow
(phosphorus)
(dl Diffusion in oxideJ'usi

(gallium!
l + (CJCJM
- f (1/m -
x ) - w exp
v
{o'BjAD)
erfc U
.
N
/3/4Dj
B -4D +
ijm
(3.40)
where
/. = r exp [(v.-r- -
1)5/4 .Dr ] erfc ( a ' / 4 i ) . / e r f ( V 5 / 4 > )

v
.Fig. 3.25
Four different cases of impurity redistribution in silicon due to thermal
oxidation."
and
and silicon phases is m o v i n g as a function o f time. The relative rate o f this

m o t i o n as compared to the diffusion rate is i m p o r t a n t in determining the
extent o f redistribution.
N o t e that even i f the segregation coefficient o f a n ' i m p u r i t v m = 1 ,
some redistribution o f the i m p u r i t y i n the silicon w i l l still take place. As
pointed out i n Chapter 2. the oxide layer w i l l fill up more space than the
silicon used in the o x i d a t i o n . Thus the same amount o f i m p u r i t v w i l l now
be distributed in a larger volume, b r i n g i n g about a depletion o f the impurity near the silicon surface.
D . D are diffusivities o f the i m p u r i t y in oxide and silicon, respectively,

and v. is the r a t i o o f the thickness o f silicon consumed d u r i n g o x i d a t i o n to
the oxide thickness. Its value is 0.45.
One interesting feature o f the'theory o f redistribution is that, as evident
from E q u a t i o n 3.40. the concentrations at the m o v i n g oxide-silicon interface will be independent of oxidation time. I n simple terms, both o x i d a t i o n
and diffusion proceed with v t i m e and. because o f this, the time variable
disappears f r o m the expression for surface concentration. Thus a steady
state is reached at the interface.
' =
0
sDJD.
E q u a t i o n 3.40 shows that, in line w i t h our earlier qualitative discussion,

the surface concentration i n the silicon is a function o f (i) the segregation
coefficient m. (ii) the relative rates o f diffusion in the silicon and in the
oxide DJD, and (in) the relative rates o f oxidation to diffusion. BID.
The role o f the first t w o o f these factors has already been illustrated i n
Figure 3.25. The t h i r d one is illustrated in Figure 3.26 for phosphorus
and in Figure 3.27 for b o r o n where the surface concentrations o f these
1
disturbance o f the originally u n i f o r m i m p u r i t y distribution. This, i n turn,

is given approximately by the diffusion length o f the impurity,
2 r.
v
Figure 3.28 shows the calculated b o r o n distribution profiles i n silicon

1
iliIr
900
I i i i
i ! i
1000
1100
Oxidation temperature C O
i
1200
Fig. 3.27 Surface concentration of boron in silicon after thermal oxidation. Calculated
using Equation 3.40. and m = 0.3.
900
1000
1100
Oxidation temperature i*Cl
Fig. 3.26 Surface concentration of phosphorus

Calculated using Equation 3.40, and m = 10.
1200
in silicon after thermal. oxidation.
impurities, calculated by using E q u a t i o n 3.40 and the experimentally

determined values o f the segregation coefficient m, are shown as a function
o f o x i d a t i o n temperature, for oxidation in 0 and H , 0 ambients. I t is
apparent that as the speed o f oxidation is increasedfor example, by
oxidizing i n water vapor rather than in oxygen at a given temperaturethe
extent o f redistribution w i l l be increased as evidenced bv the higher surface
concentration o f phosphorus and by the lower surface concentration o f
boron.
;
These t w o figures give the surface concentration as a function o f o x i d a t i o n temperature. The amount o f i m p u r i t y lost from the silicon as a
resulc o f redistribution is another matter. Whereas the surface concent r a t i o n depends on oxidation temperature as shown i n the above t w o
figures, the amount o f impurity lost w i l l depend also on the extent o f the
0.5
1.0
1.5
2.0
Distance from S i - S i O , interface un

Fig. 3.28 Calculated boron concentration distribution in silicon after oxidation at
various temperatures (O., x, = 0.2 u ) .
:s
74
Solid-State
Diffusion
Diffusion through a Silicon Dioxide Layer.
75
after oxidation at various temperatures, w i t h each oxidation resulting i n an

oxide layer 0.2 fi t h i c k .
N o t e that as the oxidation temperature is
decreased, the surface concentration o f boron decreases in accordance w i t h
Figure 3.27. However, because the diffusivitv o f b o r o n also decreases,
the extent o f the depression i n concentration w i l l also be smaller at the
lower temperatures. These calculations were fully verified by experimental
measurements.
25
25
Table 3.2 shows the segregation coefficients o f i m p o r t a n t acceptor and

donor impurities as determined by several experimental techniques. F o r
comparison, we also list values o f these segregation coefficients w h i c h were
predicted on the basis o f solubility thermodynamic considerations.
26
(
TABLE
3.2
SEGREGATION COEFFICIENTS OF IMPURITIES AT

Si/SIO. INTERFACE
m
Impurity
Thermodynamic
estimate
26
Based on
redistribution
experiments
Ga
B
>10
IfrMO
In
P, Sb, As
>10
>10
2 1
_0.3 '
0.1
2 5
Based on
oxide masking
experiments *
-20
~io-
24
10 -
25
c
So far we have considered only the relatively simple problem o f the
redistribution of an originally u n i f o r m i m p u r i t y concentration present in
the silicon. A n even more i m p o r t a n t practical problem is the redistrib u t i o n o f a predeposited layer d u r i n g the drive-in diffusion step. Because
the initial distribution is more complicated, this problem is more difficult to
study both experimentally and theoretically. The results o f a s t u d y o f the
redistribution of diffused boron layers after thermal oxidation is shown in
Figure 3.29. Here the theoretical line represents the results o f numerical
computations for an assumed segregation coefficient m = OA while the
experimental points were obtained by sectioning and successive measurements of the average resistivity o f the boron-diffused layer. The solution is
normalized to the surface concentration d u r i n g the predeposition step
which preceded the drive-in diffusion. I t is evident that the results shown
in this figure are in general agreement w i t h what we would expect from
the studies o f the redistribution o f a u n i f o r m l y doped silicon substrate.
27
Distance from Si-SiO_- interface i/ij

Fig. 3.29
3.6
Redistribution of diffused boron layer after thermal oxidation.
DIFFUSION T H R O U G H A SILICON
DIOXIDE LAYER (OXIDE MASKING)
The fact that the silicon dioxide layer is relatively impermeable to

several acceptor and donor i m p u r i t i e s has been an exceedingly i m p o r t a n t
factor in the development o f the planar technology. I n order to get a
more quantitative picture o f the masking phenomenon, we n o w c o n sider the predeposition o f an i m p u r i t y through a silicon dioxide laver.
We assume that the thickness o f this layer remains constant t h r o u g h o u t
the predeposition. The expected distribution o f the impurities after predeposition is illustrated schematically in Fieure 3.30.
6
: C i
\<
C
c
c
c
Solid-state
76
Diffusion
Uirtusion through a Silicon Dioxide Layer
77
which, when substituted i n t o E q u a t i o n 3.41. results in
T h e initial c o n d i t i o n for this diffusion problem is t h a t the concentration

o f the impurity is zero both i n the oxide and in the silicon. The boundary
conditions are (i) that the concentration at the outer surface o f the oxide
is a constant throughout the predeposition step (corresponding to the
constant C i n the case o f predeposition onto a bare semiconductor sample
considered earlier); and (ii) that the i m p u r i t y concentration vanishes deep
inside the semiconductor. I n a d d i t i o n , we must satisfy the t w o matching
X,-
1 X,
(3.42)
r v t
where
/ =
lyJD
arg
erfc (m
r)C,
2mrC.
9
Computer calculations- o f the j u n c t i o n depth x, as a function o f the

oxide thickness x, based o n the infinite series, are shown in Figure 3.31 for
Fig. 3.30
Schematic illustration of the impurity distribution resulting alter predeposi-
tion througn an oxide layer.
1
1
c o n d i t i o n s : (i) that the ratio o f concentrations at the oxide-silicon interface is determined bv the segregation coefficient in; and ( i i ) that the flux
o f impurities t h r o u g h this interface is continuous.
T h e solution o f this problem is i n the f o r m o f an infinite series.
0
approximate the infinite series by its first term alone. This results i n
Imr
erte
m + r
(3.41)
;Dj-
1
3
1
4
1
5
1
6
Fig. 3.3I Comparison between theoretical calculations and experimental results for
gallium predeposition at I IOOC.-'
23
H o w e v e r , for relatively small values o f x such that x \i DfD x , we can
C(x, t)
!
2
2 DfJ
V
where r = V DJD.
A particularly i m p o r t a n t question is the depth o f a j u n c t i o n formed in
the silicon under a protective oxide layer when the i m p u r i t y has diffused
t h r o u g h this [aver. The position o f this j u n c t i o n is defined by the c o n d i t i o n
the specific case o f the predeposition o f g a l l i u m at 1100C. The curves

were calculated for various assumed values o f the segregation coefficient
m. N o t e that regardless o f the value o f m all curves approach the
straight-line relationship. E q u a t i o n 3.42. for small values o f x,. The slope
o f this straight-line asymptote directly yields r. The shape o f the experimental Xj versus x relationship at relatively large values o f x in comparison w i t h the family o f theoretical curves enables the estimation o f the
segregation coefficient in. Segregation coefficients estimated in this
manner f r o m the experimental data on g a l l i u m and b o r o n predeposition
through an oxide film are included in Table 3.2.
0
29
3 0
A technologically i m p o r t a n t quantity, the m i n i m u m oxide thickness

required to prevent the f o r m a t i o n o f a j u n c t i o n in.the silicon, the masking
thickness, is readily calculated from E q u a t i o n 3.42 by setting x, = 0.
Solid-State
78
3.7
Diffusion
Cjix, 0) =
I n epitaxial growth the film contains either different impurities than the
substrate, or the same impurities but i n different concentrations. I t is
usually desirable that the d o p i n g concentration gradients between film and
substrate be sharp. However, epitaxial g r o w t h has to be performed at
-Film-
ICA
Gas
(3.44)
which designate that the i n i t i a l distribution i n the substrate is a constant

C
corresponding to the substrate d o p i n g concentration and that the
i m p u r i t y d i s t r i b u t i o n deep inside the substrate remains undisturbed as the
epitaxial g r o w t h proceeds.
at
- z = (h + K ) C ,
OX
X-mm X.mm Vl.
where h is the gas-phase mass-transfer coefficient in terms o f concentrations i n the solid, and V is the film-growth rate. E q u a t i o n 3.45 is a
statement o f the conservation o f impurities at the advancing film surface.
Because o f its importance we w i l l derive this b o u n d a r y c o n d i t i o n carefully.
Xj
0(1)
1) and
therefore the diffusion o f
Cj(a\ I) dx.
(3.46)
<
The time rate o f decrease o f 0(1) is. in t u r n , given bv
Since diffusion w i l l tend to even out
gradients at the interface between epitaxial film
and
substrate, the diffusion o f impurities d u r i n g epitaxial g r o w t h is o f obvious

practical importance.
The impurity distribution near the epitaxial film-substrate
consists of two components as illustrated i n Figure 3.32.
interface
One, whose
concentration is designated by d , is due to the i m p u r i t y diffusing out from

the substrate; the other, whose concentration is designated by C . is the
2
doping i m p u r i t y w h i c h is intentionally incorporated
into
the
do
di
-J-
r*
-
3C
oi
J-T.
VC,(x
t) = hCiLx,,
where we employed Leibnitz's rule o f differentiating integrals.
if they are impurities o f opposite types.

at the intersection
o f the t w o i m p u r i t y
concentration
The last
in the gas is zero.

Next, we note that because o f the diffusion equation we have
0
difference
In the latter case a p-n j u n c t i o n
(3.47)
a solid surface to a gaseous ambient, i f the concentration o f the i m p u r i t y
growing film and w h i c h , d u r i n g g r o w t h , diffuses inward into the substrate.

components i f they are impurities o f the same type and by their
1)
p o r t i o n o f the equation is the expression for the escape o f impurities f r o m
The total distribution o f impurities is determined by the sum o f these t w o
distributions.
The total a m o u n t o f substrate impurities per unit area contained w i t h i n

the semiconductor substrate and film is siven bv
Chapter
impurities will be relatively r a p i d .
will be formed
(3.45)
Impurity distribution in epitaxial growth.
elevated temperatures (see
external
substrate
(3.43)
S U D
d C - c o , r ) = sub
-D
C,(XJ 1 Impurin iffusiiie

from he substrate
x = u
the concentration
the
The second b o u n d a r y c o n d i t i o n is given by
C . l v . n External
doping impuriiv
Fig. 3.32
to
and the b o u n d a r v c o n d i t i o n .
Film surface (moving

boundary)
^*tr
79
S U D
Metallurgical
/ " interface
Initial
distribution
Redistribution of Impurities in Epitaxial G r o w t h .
The solid-state diffusion problem corresponding

i m p u r i t y is solved subject to the initial c o n d i t i o n
T H E REDISTRIBUTION O F IMPURITIES
IN E P I T A X I A L G R O W T H
I"Substrate-
The
-tr.
01
dx = D
I
ffn
7 dx = D
CX~
_ 3C,
3C,
ox
T-T,
OX
(3.48)
r=a:->
Thus, by c o m b i n i n g E q u a t i o n 3.48 with 3.47, we arrive at our b o u n d a r y

c o n d i t i o n . E q u a t i o n 3.45.
>
impurities in epitaxial G r o w t h
1.0
The solution-- o f this problem is fairly complicated, especially since this

last boundary c o n d i t i o n is specified at a moving boundary.
For
our
Vt
(3.49)
31
0.8
purposes, this solution can be represented as

lit
.1
0.6
0.4
ht
Vt
*- but)
w h i c h shows that the normalized concentration d i s t r i b u t i o n is a function
o f the dimensionless distance, an escape-rate parameter, and a g r o w t h rate parameter. F o r V = 0. the solution reduces to the externally ratel i m i t e d outdiffusion case which we discussed earlier in this chapter.
5\V
0.2
1
x ; 2 Dr
v
M o r e i m p o r t a n t l y , however, the solution has a very simple l i m i t i n g f o r m

for the case o f F7/V Dt 1 . I n that case the solution is given to a very
good a p p r o x i m a t i o n by
10-'f
d(x, t)
= i ertc
C b
(J.DO)
- v Dt
s u
10-21
regardless o f the value o f A.

Calculations based o n the complete solution, for the case htj\j Dt 1
and for various values o f theparameter Vtf\ Dt, are shown i n Figure 3.33.
10"
I t is evident that-tfre i m p u r i t y d i s t r i b u t i o n approaches the simple f o r m

given by Equation 3.50 for values o f Vtfx Dt larger than about 5.
typical epitaxial
film:thicknesses
Since
are o f the order o f about 10,u. whereas
10-->b
typical values ofW-X>^might-beof the order o f less than 1 u , typical values

o f the g r o w t h parameter Vtjs Dt w o u l d be larger than
finitely"
large for the purposes o f these calculations.
10-thus ' ' i n 10-s
Hence, the simple
-L
s o l u t i o n given by E q u a t i o n 3.50 provides a very good a p p r o x i m a t i o n for

practical epitaxial g r o w t h conditions.
Dt between 1 0
- 3
(The same conclusion was reached
Fig. 3.33
Distribution of the impurity diffusing from the substrate."
and the boundary conditions
and 10 .)
I t is easy to give an interpretation to this simple l i m i t o f the exact

solution. The film g r o w t h rate is,so high relative to the rate o f diffusion
f r o m the substrate that, insofar as the diffusion profile is concerned, the
f i l m grows to an " i n f i n i t e " thickness almost instantaneously. Thus the
concentration d i s t r i b u t i o n w i l l be very close to that w h i c h is obtained i n
the problem o f diffusion between two semi-infinite slabs Equation
3.50.
Cj(-co, t) = 0
C,(x ,
f
The diffusion o f the external d o p i n g i m p u r i t y is described by a s o l u t i o n

o f the diffusion equation subject to the initial c o n d i t i o n
If
(3.51)
t) =
(3.52)
C.
(3.53)
The first boundary c o n d i t i o n states that deep inside the substrate the
concentration o f the external doping i m p u r i t y vanishes. The second
specifies that the concentration o f the external doping i m p u r i t y at the
growing film surface is a constant C, (C, is determined by the concentration
o f the i m p u r i t y in the gas mixture). The solution to this p r o b l e m is also
fairly complicated. I t can be represented as
22
Cijx,
C.(x, 0) = 0
1
X/2y/Dt
on the basis o f calculations performed for values o f the escape rate p a r a m eter hti\
t)
Vt
-A
ily/Dt
' jDtJ
'
(3.54)
83
References C i t e d
In the above treatment we assumed that the i m p u r i t y d i s t r i b u t i o n i n the

I
epitaxial film is determined by the process o f solid-state diffusion
alone.
This is indeed the case p r o v i d e d precautions are t a k e n to eliminate c o n tamination both from
Substrate concentration
doped substrates.
31
the reactor and
been shown to be carried by the

growing epitaxial
d i s t r i b u t i o n o f the
f r o m the back-side o f the
heavily
I m p u r i t i e s from the back-side o f the substrates have

film
gas
on the f r o n t ,
3 2
stream and
incorporated into
the
resulting i n a m u c h more gradual
substrate i m p u r i t y t h a n solid-state diffusion
would.
This p h e n o m e n o n is often referred to as " a u t o d o p i n g . "
READING
REFERENCES
General treatments of solid-state diffusion are given by P. G . Shewmon, Diffusion

in Solids, McGraw-Hill Book C o . , 1963; R . M . Barrer, Diffusion In and Through
Solids, Cambridge Universitv Press, 1941; and W. Jost, Diffusion in Solids, Liquids,
Cases. Academic Press, 1952.
A treatise on Diffusion in Semiconductors is given bv B. I . Boltaks, Academic Press,
1963.
A review of diffusion data specifically dealing with silicon and of diffusion technology
is found in Volume 4 of a series of review-reports on "Integrated Silicon Device Technology," by the Research Triangle Institute, A S D - T D R - 6 3 - 3 1 6 (1964).
Diffusion data in glasses is reviewed bv R. H . Doremus, "Diffusion in Non-crvstalline
Silicates." Chapter 1 in Volume 2 of Modern Aspects of the Vitreous State. J . D .
Mackenzie. E d . . Butterworths. 1962.
An extensive compilation of the solutions of the diffusion equation under a wide
variety of conditions can be found in H . S. Carslaw and J . C . Jaeger, Conduction of
Heal in Solids, Oxford Universitv Press. 2nd ed., 1959.
x (_)
Fig. 3.34
For
Impurity distribution after epitaxial growth (antimony, 20 min. at I275C)".

!
Vti\
Dt >> 1 and
for distances x
2\jDi
this solution
The
0 =
C,.
combined solution describing the
in epitaxial growth can then be obtained

C(x, t) =
C,(x, 0
net
1. These diffusivities are based on Volume 4 of a series of review-reports on "Integrated

Silicon Device Technology," bv the Research Triangle Institute, ASD-TDR-63-316
(1964).
(3.55)
2. These diffusivities are based on:

a. D . E . Swets. R . W. Lee. and R . C . Frank. "Diffusion Coefficients of Helium in
Fused Quartz." J. Chem. Phys.. 34, 17 (1961).
b. R. W. Lee. R. C . Frank, and D. E . Swets, "Diffusion of Hydrogen and Deuterium
in Fused Quartz." J. Chem. Phys.. 36, 1062 (1962).
c. R . C . Frank. D . E . Swets. and R . W. Lee, "Diffusion of Neon Isotopes in Fused
Quartz," J. Chem. Phys.. 35, 1451 (1961).
d. F . J . Norton. "Permeation of Gaseous Oxvgen through Vitreous Silica,"
Nanirc. 171. 701 (1961).
e. T . Drury and J . P. Roberts. "Diffusion in Silica Giass Following Reaction with
Tritiated Water Vapor." Phys. Chem. Glasses. 4, 79 (1963).
d i s t r i b u t i o n o f impurities
from
C (x,
2
r)
(3.56)
where the positive sign is taken f o r i m p u r i t i e s o f the same type and
the
negative sign for impurities o f opposite type.

Extensive experimental measurements- have verified the solution given
above. A n example o f these results is shown i n Figure
3.34.
CITED
be
can
approximated simply by setting

C.(x,
REFERENCES
Solid-State
84
Diffusion
f. A . S. Grove. 0 . Leistiko. and C . T . Sah. 'Diffusion of Gailium through a

Silicon Dioxide Layer." J. Phys. Chem. Solids, 25. 985 (1964).
g. A . E . Owen and R . W. Douglas, "The Electrical Properties of Vitreous Silica."
J. Soc. Class Tech.. 43, 159 959).
h. M . O. Thurston and J . C . Tsai. "Diffusion of Impurities into Silicon through
an Oxide Layer," Signal Corps Contract, Final ReDOrt DA-36-039-SC 87426
(1961).
i. C . T . Sah, H . Sello, and D . A . Tremer, "Diffusion of Phosphorus in Silicon
Oxide Film." / . Phys. Chem. Solids. 11, 288 (1959).
j . S. Horiuchi and J . Yamaguchi. "Diffusion of Boron in Silicon through Oxide
Layer," Jap. J. Appl. Phys. 1. 314 (1962).
3. B . I . Boltaks. Diffusion in Semiconductors, Academic Press, 1963.
4. w . K Wilcox and T . J . LaChapclle, "Mechanism of Gold Diffusion into Silicon."
/. Appl. Phys., 35, 240 (1964).
5. H . S. Carslaw and J . C . Jaeger, Conduction of Heat in Solids, Oxford University
Press. 2nd ed., (1959).
6. C . J . Frosch and L . Derick, "Surface Protection and Selective Masking During
Diffusion in Silicon," J. Electrochem. Soc, 104, 547 (1957).
7. A n extensive review of silicon diffusion practices is given in Reference 1.
3. C . J . Frosch and L . Derick, "Diffusion Control in Silicon by Carrier Gas Composition." J. Electrochem. Soc,
105, 695 (1958).
9. G . E . Moore, unpublished.
10. F . A . Trumbore. "Solid Solubilities of Impurity Elements in Germanium and
Silicon." Bell System Tech. J.. 39, 205 (I960).'
11. T . H . Yeh and W. Armstrong, "Diffusion of Boron in Silicon," Electrochemical
Society Spring 1961 Meeting, Abstract No. 69. Indianapolis.
12. D . P. Kennedv and P. C . Murley. "Impurity Atom Distribution from a Two-Step
Diffusion Process." Proc. IEEE (Corresp.), 54, 620 (1964).
13. W . L . Bond and F . M . Smits. "The Use of an Interference Microscope for Measurement of Extremely Thin Surface Layers," Bell System Tech. J., 35, 1209 (1956).
Problems
85
20. K. H . Nicholas, "Studies of Anomalous

Solid-State Electron., 9, 35 (1966).
Diffusion of Impurities in Silicon."
21. F . M. Smits and R. C . Miller. "Rate Limitation at the Surface for Impurity Diffusion
in Semiconductors." Phys. Rev.. 104, 1242 (1956).
22. A . S. Grove, A . Roder, and C . T . Sah, "Impurity Distribution in Epitaxial Growth,"
J. Appl. Phys., 36, 802 (1965).
23. M. M. Atalla and E . Tannenbaum. "Impurity Redistribution and Junction Formation in Silicon by Thermal.Oxidation," Bell System Tech. J., 39, 933 (1960).
24. A . S. Grove, O. Leistiko, and C . T . Sah. "Redistribution of Acceptor and Donor
Impurities During Thermal Oxidation of Silicon," / . Appl. Phys., 35, 2695 (1964).
25. B. E . Deal. A . S. Grove. E . H . Snow, and C . T . Sah. "Observation of Impurity
Redistribution During Thermal Oxidation of Silicon Using the MOS Structure,"
J. Electrochem. Soc, 112, 308 (1965).
26. C . D. Thurmond, "Distribution Coefficients of Impurities Distributed Between Ge
or Si Crystals and Ternary Alloys or Surface Oxides," in Properties of Elemental
and Compound Semiconductors, H . C . Gatos, E d . . p. 121, Interscience, 1960.
27. T. Kato and Y . Nishi, "Redistribution of Diffused Boron in Silicon by Thermal
" Oxidation," Jap. J. Appl. Phys., 3, 377 (1964).
28. C . T. Sah. H . Sello, and D. A. Tremere. "Diffusion of Phosphorus in Silicon Oxide
Film." J. Phys. Chem. Solids, 11. 288 (1959).
29. A . S. Grove, O. Leistiko. and C . T . Sah. "Diffusion of Gallium Through a Silicon
Dioxide Layer." J. Phys. Chem. Solids, 25, 985 (1964).
30. S. Horiuchi and J . Yamaguchi. 'Diffusion of Boron in Silicon Through Oxide
Layer," Jap. J. Appl. Phys., 1, 314 (1962).
31. H . Basseches. S. K . Tung, R. C . Manz, and C . O. Thomas, "Factors Affecting the
Resistivity of Epitaxial Silicon Layers," Metallurgy of Semiconductor Materials,
15, 69 (1962).
32. B. A . Joyce. J . C . Weaver, and D . J . Maule, "Impurity Redistribution Processes
in Epitaxial Silicon Wafers." J. Electrochem. Soc, 112, 1100 (1965).
PROBLEMS
14. J . C . Irvin. "Resistivity of Bulk Silicon and of Diffused Layers in Silicon," Bell
System Tech. J., 41, 387 (1962).
15. L . B . Valdes, "Resistivity Measurements on Germanium for Transistors," Proc.
IRE.
3.1
Derive the transport equation in a manner similar to that employed in Section 3.2
for the case when the transported species is being consumed by a chemical reaction.
Assume that the rate of this reaction at any point is proportional to the concentration of the species.
3.2
Verify that Equations 3.13 and 3.21 satisfy the diffusion equation and the appropriate initial and boundary conditions.
3.3
A phosphorus predeposition into silicon containing 10 * acceptor atoms/cm

resulted in an average resistivity of 4 x I 0 ~ ' Q cm. Evaluate the phosphorus
surface concentration, assuming
(a) erfc
)
(b) Gaussian
distribution.
(c) "box"
J
Which is most realistic?
42. 420 (1954).
16. F . M . Smits, "Measurement of Sheet Resistivities with the Four-Point Probe,"

Bell System Tech. J., 37, 711 (1958):
17. D . P. Kennedv and R . R. O'Brien, "Analysis of the Impurity Atom Distribution
Near the Diffusion Mask for a Planar P-iV Junction." IBM Journal, 9, 179 (1965).
18. K . Lehovec and A . Slobodskoy, "Diffusion of Charged Particles into a Semiconductor under Consideration of the Built-in Field," Solid-State Electron., 3, 45
(1961).
19. S. Maekawa and T . Oshida. "Diffusion of Boron into Silicon." J. Phys. Soc Japan,
19, 253 (1964).
86
3.4
A p'n junction diode was fabricated as follows:

Starting material: n-type. 5 x 10" phosphorus/cm .
Predeposition: B C I . 990"C. 15 minutes.
Drive-in diffusion: Dry O . . 1200 C. 9 hours.
Measured V/I = 60 h
Measured x = 7.0 //
Calculate the impuritv concentration as a function of distance, the total amount
of diffused impurities per unit area (Q). and the impurity concentration gradient
at the junction.
3
3.5
A commercial diode is fabricated as follows:

Starting material: n-type. 4 x 10 phosphorus/cm .
Predeposition: B B r , , 1 ISO'C, 6 minutes.
Drive-in diffusion: Dry O , 1280 C, 16 hours.
Measured V/I = 0.8fi
Measured
= 37 p
Evaluate as in Problem 3.4.
Compare the two diodes.
1J
3.6
Derive Equation 3.30. In this derivation, assume that (i) the electric current is
zero, (ii) space-charge neutrality prevails, and (iii) the product of electron and hole
concentrations pn n,~.
3.7
Using the intermediate results of the above problem, estimate the built-in electric
fie;c near the surface associated with the predeposition of boron at 1200C, after
1 hour.
3.8
The different redistribution tendencies of impurities during thermal oxidation can

be employed in the fabrication of useful device structures. A s an example, oxidation of silicon initially containing both gallium and phosphorus in concentrations
of 2 x 10" c m and 1 x 10" c m , respectively, will lead to the formation of an
n-type skin under the oxide. If the oxidation is performed at 1200C in dry oxygen,
for 2 hours, calculate:
- 3
- 3
(a) The resulting impurity distributions.

(b) The depth of the junction, and the impurity concentration gradient at the
junction.
(c) The average resistivity of the n-type skin.
3.9
Estimate the thickness of the silicon dioxide laver required to mask against the
predeposition of boron at 1100C, for a predeposition time of 1 hour. Assume that
the solid solubility of boron in silicon dioxide is (1/m) times its solid solubility in
silicon. (Justify this assumption.)
3.10 A n epitaxial film doped to. contain 10" boron atoms/cm?, is grown upon four
different substrates simultaneously. Tbese substrates contain:
(a) 10 phosphorus atoms/cm
(b) 10" phosphorus atoms/cm
(c) 10" phosphorus atoms/cm
(d) 10" antimony atoms/cm .
Growth conditions are 20 minutes at 1250 C.
Calculate the junction depth for each wafer.
11
Problems
3.11 The base region of a pnp transistor is fabricated as follows:
Starting material: /Mype, 10" boron/cm
Epitaxial growth: 20 min. at 1250""C

Film growth rate: 1 ^/min.
Film doping: 5 x 10" boron/cm
Oxidation:
80 min. at f200"C, water vapor.
Base window opened.
Predeposition: P . O . 30 min. at 800C.

Drive-in diffusion: 50 min. at 1200C, dry oxyeen
Measured V\l = 1.4 0.
Measured x, = 4 p
Calculate:
s
(a) The oxide thickness over the base window and elsewhere.
(b) The impurity distribution after drive-in diffusion, including the distribution of
the substrate impurity.
(c) What is the total number of impurities in the base ( Q ) ? The gradient at the
collector-base n-p junction? The distance over which the acceptor impurity
concentration is uniform?
aa
TABLE
3.3
IMPORTANT FORMULAS IN DIFFUSION
Flux:
PART I I
dC
F = D - + fi<fC
ox
+ charge
dC
charge
F = D-
SEMICONDUCTORS A N D
fitTC
ox
SEMICONDUCTOR DEVICES ,
kT
D
Einstein's relationship:
Transport equation:
for S = constant,
+ charge
ix
ac
dF
dt
dx
ac
d-c
= D
dt
ox-
ac
-
Elements of Semiconductor
Physics
ix e
ox
ac
OC
charge
Diffused layers
3' c
Semiconductors under
Non-Equilibrium Conditions
p-n Junctions
= D r -+-,<<?
dt
dxox
1
.r
constant
Junction Transistors
C O , t) C o erfc
2 v Dt
Junction Field-Effect
Transistors
2(0 =
v Dt C<s
1
dC
dx
constant Q
v Dt
(x,()
C(x, t) =
Q
v-Dt
z-HDt
IDt
CO, t)
THE
0 -\
BAND
THEORY
ELECTRONS
IN
M
AND
O F SOLIDS
HOLES
SEMICONDUCTORS
FERMI-DIRAC
DISTRIBUTION
FUNCTION
" P h y S t e S
o"T
S e w n
e*^
"Tec"
ca/i d o c t o r
noiosi
t > e \ X e e s
IMPORTANT
FORMULAS
TRANSPORT
OF
AND
ELECTRONS
HOLES
4
Physics
I n this chapter we present and discuss some i m p o r t a n t results o f the

physics o f semiconductors. We consider only those aspects o f semiconductor physics that are essential for the treatment o f semiconductor devices
and surface phenomena.
We begin with a qualitative discussion of the band theory o f solids
and of intrinsic and extrinsic semiconductors. Then we discuss the
Fermi-Dirac distribution function and the concept o f Fermi level, and
present those i m p o r t a n t formulas o f semiconductor physics which we w i l l
use throughout this book. Finally, we discuss the transportdrift and
diffusionof electrons and holes in semiconductors.
4.1 T H E B A N D T H E O R Y
O F SOLIDS
- T h e most important result o f the application o f q u a n t u m mechanics to

the description of electrons in a solid is that the allowed energy levels o f
electrons w i l l be grouped into bands. The bands are separated by regions
which designate energies that the electrons in the solid cannot possess.
These regions are called forbidden gaps. The energy bands and a forbidden
gap are illustrated schematically in Figure 4.1._The electrons in the outermost shell o f the atoms comprising the solid, the valence electrons, are
91
Physics
shown here in their lowest energy states. The band o f these states is called
the valence band.
The phenomenon o f conduction is o f principal interest in the study o f
semiconductor physics. Conduction consists o f the m o t i o n o f electrons.
Thus conduction is possible only i f we can get electrons into motion. I n
terms o f energy considerations, this means that conduction is possible only
if we can impart kinetic energy to an electron. We can now examine
three different classes o f solidsmetals, insulators, and. semiconductors
in terms o f both an atomistic representation and the energy-band representation from the viewpoint o f whether or not it is possible to energize an
electron.
The
Band Theory of Solids
93
band are empty. Because small electric fields cannot raise the uppermost
electron in the valence band to the conduction band, it will be impossible
to impart energy to any o f the electrons shown in the band picture. For
this reason, silicon dioxide will be an insulatorit will not conduct
electricity.
The intermediate case o f a semiconductor, with silicon taken as an
example, is shown in Figure 4.2c. The bonds between neighboring silicon
Sea of conduction
electrons : free to move
8onds hard to break :

no conduction electrons
Some bonds are broken :

few conduction electrons
and holes result
3 0 0 S O O O S O O C
[
-Conduction band
.Al"
AI - \
+
Forbidden gap
0
o
0
0
o
0
DSi
30osiooosiooc
d'siCE
0 >0
3Si<5Z2>Si<I
Valence band
Valence electrons
Fig. 4.1
Schematic energy band representation for electrons in a solid illustrating the
energy bands and a forbidden gap.
I n metals the valence electrons are free and constitute a sea o f electrons
which are free to move upon the application o f even a small electric field.
The corresponding energy-band representation w i t h a l u m i n u m taken as
an example is shown i n the lower portion o f Figure 4.2a. The two bands
corresponding to the previous figure here overlap. Thus there is h o forbidden gap. As a consequence, it is possible to move the topmost electron
shown i n this figure to the next level and then to the next and then to the
next; in other words, it is possible to impart a kinetic energy to the
electrons. Hence, conduction is possible.
I f we now turn to the example o f an insulator such as silicon dioxide,
shown i n Figure 4.2b, we are faced with a different situation. The valence
electrons here form strong bonds between neighboring atoms. These
bonds are difficult to break and, because o f this, there w i l l be no free
electrons that could participate in conduction, I n terms o f the energyband picture, this means that there is a large forbidden gap between the
valence band and the next higher band, the conduction band. A l l levels in
the valence band are occupied by electron;,, all levels i n the conduction
(a)
Metal:
Bands overlap
and even infinitesimal
added energy loads to
conduction.
(b)
(c)
Insulator:
Semiconductor:
Separation between bands
moderatefew electrons
appear in conduction band,
equal number of holes in
valence band. Limited
conduction possible.
Very large reparation

between bands : conduction impossible.
Fig- 4.2 Schematic atomistic and energy band representation of a conductor, an insulator,
and an intrinsic semiconductor.
atoms are only moderately strong. Therefore, due to the thermal vibration
of the silicon atoms, some bonds will be broken at any temperature above
absolute zero. When a bond is broken, a free electron results which is
capable o f conducting electricity. Such an electron is called a conduction
electron. In addition, there is now a "deficit" where the electron had been
before the bond was broken. This deficit is referred to as a hole. Valence
electrons can j u m p from neighboring bonds into the position o f the hole
and therefore additional conduction can take place. I t is easy to think o f
E l e m e n t s of Semiconductor Physics
94
this a d d i t i o n a l conduction as the movement o f the positively charged hole

in the opposite direction.!
I n terms o f the band representation, the forbidden gap of a semiconductor is n o t as large as that o f an insulator. Because of this, some
electrons w i l l be able to make the j u m p f r o m the valence band to the
conduction band, leaving behind holes i n the valence band. U p o n the
application o f an electric field, the electrons i n the conduction band w i l l
be able to gain kinetic energy; hence, they w i l l be able to conduct electricity. A t the same time the holes i n the valence band w i l l also be able
to take on kinetic energy and conduct electricity.
95
Electrons and Holes in Semiconductors
,.. these electrons and holes have kinetic energies designated by the
difference between their energies and the respective band edges, as i l l u strated in Figure 4.3.
4.2
ELECTRONS AND HOLES

IN S E M I C O N D U C T O R S
In absolutely pure semiconductors, conduction electrons and holes

result only through the breakage o f bonds. Thus the concentration o f
electrons ?? equals the concentration o f holes p. These concentrations are
called the intrinsic carrier concentrations n o f the semiconductor. W e
would expect the intrinsic carrier concentration to be a function o f the
vibrational energy o f the lattice (hence the temperature) that is responsible
for the breakage o f bonds. The intrinsic carrier concentration should also
be a function o f the energy required to break a b o n d , which in terms o f the
band diagram corresponds to the w i d t h o f the forbidden gap. or energy
gap. &
Both of these dependences are borne out by the experimental data shown
in Figure 4.4. Here the intrinsic carrier concentrations o f three i m p o r t a n t
semiconductorsgallium arsenide, silicon, and germaniumare shown as
a function o f temperature. The energy gaps o f these semiconductors are
also indicated in the figure. I t is evident (i) that for any given semiconductor, H , increases very sharply w i t h increasing temperature; and (ii) that
at any given temperature,
decreases very sharplv w i t h increasing energy
gap. M o r e detailed examination o f these data indicates that both dependences can be summarized by the exponential temperature dependence,
n cc e~ '
where the activation energy E is approximately / 2 . This
important relationship w i l l be discussed further at a later stage.
i
.'"',
. ,
Kinetic energy of electron
Potential energy of electron
_ r-
"j
1
Kinetic energy of hole
t
Fig. 4.3
Potential energy
of hole
Illustration of the meaning of the energy band diagram.
The energy-band diagram, such as shown i n Figures 4.1 and 4.2.

indicates electron energies. Thus when the energy o f an electron is i n creased, the electron will take on a hight r position in the band diagram.
When we say that the energy of a hole i ' . increased, what we mean is that
the energy of tne other electrons in t h valence band is increased. Thus
some o f the valence electrons will take on higher positions in the band
diagram. Accordingly, the increase in the energy o f a hole is represented
by the hole moving downward i n the valence band. (Recall the "bubble
analogy".)
p
I t is i m p o r t a n t to note at this point that the lowest level i n the conduction band designates the energy o f a conduction electron which is at rest.
The energy o f an electron at rest is, o f course, the potential energy o f an
electron, so the lower edge o f the conduction band designates the potential energy o f an electron. Similarly, the upper edge o f the valence band
,. designates the potential energy o f a hole. I f an electron is at a higher
energy than the level ,, or a hole at a lower electron energy than the level
r
t The concept of a hole is analogous to that of a bubble in a liquid: although it is

actually the liquid that moves, it is much easier to talk of the motion of the bubble in
the opposite direction.
lkT
Let us now consider the case where an i m p u r i t y is incorporated i n t o

the single crystal semiconductor in concentrations which are much larger
than 7i,. I n particular, let us first consider the case that arises when we add
an i m p u r i t y which has five valence electrons, such as phosphorus, to silicon
which has four. This case is illustrated in Figure 4.5a. (Phosphorus
occupies a place in the fifth c o l u m n o f the periodic table o f the elements,
while silicon occupies a place i n the fourth c o l u m n , in accordance w i t h
the respective number-of their valence electrons.) The extra electron o f the
phosphorus atom cannot be accommodated i n the regular b o n d i n g arrangement o f the silicon lattice and, since it is out o f place, it is easier to
tear off. Thus the ionization energy o f phosphorus in silicon is much
smaller than the silicon energy gap. I n fact, this ionization energy is only
about 0.05 ev.
fclements
ot
Semiconductor
Physics
Electrons
and Holes
in
Semiconductors
97
c o n d i t i o n is called complete ionization.

Thus, under the condition o f complete ionization, we can write that the concentration of electrons n = N
where N denotes the concentration o f the donor impurities. (Column V
impurities in silicon are called donor impurities because they donate an
electron to the conduction band o f the silicon crystal.) This again is
illustrated in the energy-band representation o f Figure 4.5a where equal
concentrations o f electrons and donor ions are indicated. The donor ions
are denoted by positive charges slightly below the conduction band edge
in energy.
D
3 Si ( E 3 S i d
p|
(Jy
Extra electron :
easy to remove
3Si(IZE)P
5Si<E3SiG
fi\
jj
Extra hole :
easy to remove
DSicn-DBd
Ionization energy
E.
++-++++++- Donor ions

-t',
Acceptor
-/-'
ions
-,.r
lal
d-Typo semiconductor
/;,
--r^^/
T
o o o o
Ionization energy
lb)
p-Type semiconductor
Fig. 4.5 Schematic atomistic and energy band representation of extrinsic semiconductors.
A n analogous argument can be made for the case when an i m p u r i t y

which has three valence electrons (a c o l u m n I I I impurity), such as boron,
is introduced i n t o the silicon lattice. Because a column I I I i m p u r i t y has
one less electron than silicon, we can consider i t to carry a hole. This hole
can then be removed relatively easily, w i t h an ionization energy o f approximately 0.05 ev.
I f the ionization is complete, the concentration o f the holes p == N
where N denotes the concentration o f acceptor impurities. ( C o l u m n I I I
impurities i n silicon are called acceptor impurities because they can accept
an electron f r o m the valence band o f the crystal. This, o f course, is equivalent to supplying a hole to the valence band.) This situation is illustrated
in Figure 4.5b, b o t h i n the atomistic and in the energy-band representations. Since we are considering the case o f complete ionization, the
concentration o f holes in the valence band is shown to be equal to the
concentration o f acceptor ions in the crystal. The acceptor ions are denoted by negative charges slightly above the valence band edge in energy.
A
HtOO/Tl'K)
Fig. 4.4
Intrinsic carrier concentration of gallium arsenide, silicon, and germanium as
a function of temperature.
I n a silicon crystal at r o o m temperature there is usually enough

v i b r a t i o n a l energy available to supply this amount of energy and,
fore, c o l u m n V I m p u r i t i e s i n silicon w i l l generally be all ionized at
temperature, p r o v i d i n g an equal number o f conduction electrons,
they are present i n relatively h i g h - ( > 1 0
c m " ) concentrations.
1 8
lattice
thereroom
unless
This
Elements of Semiconductor Physics
98
I n the first of the above t w o cases, the concentration o f electrons is much

larger than that o f holes. Because current in such a case is carried predomi n a n t l y bv electrons, we say t h a t the conductivity type of the semiconductor
is 77 (for negative). I n the second case, the concentration of holes is much
lareer than the concentration o f electrons and current is predominantly
carried by holes. The c o n d u c t i v i t y type in such a case is p (for positive).
I n eeneral, both acceptor- and donor-type impurities may be present
simultaneously. The conductivity type of the semiconductor is then determined by that impurity which is present in the greater concentration. The
concentration of the corresponding majority carrier will then be given by
= N - Nj i f N > N ; and by p = N - N i f N > N .
D
4.3
FERMI-DIRAC D I S T R I B U T I O N
F e r m i - D i r a c Distribution Function
99
The F e r m i - D i r a c d i s t r i b u t i o n function is symmetrical around the

Fermi level E .
Thus, i f the number of energy states in the conduction
and valence bands is the same, and i f the number o f electrons in the conduction band and the number o f holes in the valence band is also the same,
the Fermi level must be located in the middle o f the energy gap. This is
approximately what happens i n an intrinsic semiconductor. The F e r m i
level in an intrinsic semiconductor is often referred to as the intrinsic
Fermi level and is denoted bv the svmbol
r
I n an 77-type semiconductor the concentration o f electrons in the conduction band is larger than in the intrinsic case. Since, however, the density
FUNCTION
I n the previous sections we have discussed various electronic energv

states in semiconductors: states i n the conduction and valence bands, and
states introduced into the forbidden gap by the addition o f donor or
acceptor impurities. We n o w discuss what determines the probability
that a given energv state is occupied by an electron.
The energy distribution o f electrons in a solid are governed by the laws
of Fermi-Dirac
statistics.
The principal resuh of these statistics is the
Fermi-Dirac
distribution function which gives the probability that an electronic state with energy E is occupied by an electron,
- j jjpggi
+
fti)
(a! Intrinsic.
(c) p-Type
Fig. 4.6 Illustration of the Fermi-Dirac distribution function for intrinsic, n- and p-type
semiconductor.
of energv states in the conduction band is the same as in the intrinsic case,
it follows that in an n-type semiconductor the F e r m i level, and with i t the
entire F e r m i - D i r a c d i s t r i b u t i o n function,, will be shifted upward in the
This function contains a parameter, E . which is called the Fermi level.

A rigorous definition o f the F e r m i level describes it as the chemical potential o f electrons in a solid. However, for our purposes it is sufficient to
note that the Fermi level is that energy at which the probability of occupation of an energy state by an electron is exactly one-half.
energy-band picture.
The Fermi-Dirac distribution function is illustrated in Figure 4.6a for

the case of an intrinsic semiconductor. A t the left side of this figure we
show the probability o f occupation o f states by electrons as a function o f
the energy o f the states. I n the conduction band there are a large number
o f states. However, the probability o f occupation of these states is small;
hence, there will be only a few electrons in the conduction band. By contrast, there are also a large number o f states i n the valence band. M o s t o f
these are occupied bv electrons since the probability o f occupation o f states
there is nearly unity: Thus there will be only few unoccupied electron
states, i.e., holes, in the valence band.
simpler formulas
(bin-Type
In contrast, in a /"-type semiconductor the F e r m i
level and the F e r m i - D i r a c distribution function w i l l b o t h be shifted d o w n w a r d . These t w o cases are illustrated in Fig. 4.6b and c.
For energies that are at least several kTunils above or below the F e r m i
level, the F e r m i - D i r a c d i s t r i b u t i o n function can be approximated by the
/() *
and
-*
for > E
.
/() =
(4.2)
.
1 -
{ E f
E ) l k T
for < E .
F
(4.3)
It is useful to regard the second term o f the last expression as the probability o f occupation o f a center located at energy E by a hole.
If. instead o f F e r m i - D i r a c statistics. B o l t z m a n n statistics had been
employed in the derivation o f the distribution function, these expressions
1 < J W
'
C l e m e n t s ot S e m i c o n d u c t o r
w o u l d have resulted directly.
Physics
Thus Boltzmann statistics provide a good
a p p r o x i m a t i o n for F e r m i - D i r a c statistics for energies at least several kT

away from
4.4
E.
F
IMPORTANT
FORMULAS
SEMICONDUCTORS
a.
Important
Formulas for Semiconductors

1 9
in
Equilibrium
101
concentration reaches about 1 0 c r r r , Equations 4.4 and 4.5 lose their

validity. Under such conditions we refer to the semiconductor as degenerate.
A n alternate set o f formulas which are more useful when symmetry in
the mathematical expressions is desirable can be derived from Equations
4.4 and 4.5. These are
FOR
IN E Q U I L I B R I U M
n = /Kr-E,mT
n
p =
E l e c t r o n and H o l e C o n c e n t r a t i o n s
faM*-*w*
( 4
6 )
( 4
7 )
where
I f the densities o f states i n the c o n d u c t i o n and valence bands are calculated on the basis o f q u a n t u m mechanics, and i f the probability o f
occupation of these states by electrons is taken into account, the electron
a n d hole concentrations in a semiconductor can be calculated. Such a
calculation yields the concentration o f electrons,
n = \pjf**r**M
(4.4)
,. i(E
+ E ) + { S T In ^
v
.'
(4.8)
is the intrinsic Fermi level. The intrinsic Fermi level is-thus seen to be i n
the vicinity o f the middle o f the gap, displaced from it by a term which is
usually very small. Consequently, for most purposes we can regard the
intrinsic Fermi level to be in the middle o f the energy gap.
and the concentration of'holes

(4.5)
p m Êr-E>VT
N
We can recognize the exponential factor i n the first o f these expressions

as the probability o F d c c u p a t i b n by an electron o f a state located at the
conduction-band e d g e _ . The-exponential factor in the second expression
is the probability o f occupation by a hole o f a state located at the valenceb a n d edge E . A c c o r d i n g l y - w e - a n assign a simple interpretation to the
pre-exponential factors: they designate the effective densities of states
i n the conduction and valence bands, respectively. B o t h ;V and N are
p r o p o r t i o n a l to T*-.
c
b.
T h e pn Product in E q u i l i b r i u m
It is evident from Equations 4.4 and 4.5 as well as from Equations 4.6
and 4.7, that the product o f electron and hole concentrations is independent o f the Fermi level; hence, it is independent o f the type o f the semiconductor as well as o f the individual electron and hole concentrations.
Thus,
The values o f the effective densities o f states for germanium, silicon,

a n d gallium arsenide at r o o m temperature (27C) are given in Table 4 . 1 ,
along w i t h other i m p o r t a n t properties o f these semiconductors and o f an
insulator, silicon dioxide.
Equations 4.4 and 4.5 show, in accordance w i t h the discussion in the
previous section, that as the F e r m i level E moves close to the conductionb a n d edae E , the concentration o f electrons increases and the concentrat i o n o f holes decreases. I n contrast, when the F e r m i level moves closer to
the valence-band edce E , the electron concentration decreases and the
hole concentration increases.
As the Fermi level approaches either the c o n d u c t i o n - or the valenceb a n d edge within a few kT units, the a p p r o x i m a t i o n s to the Fermi-Dirac
d i s t r i b u t i o n function that we use in the above formulas will become i n accurate.
For this and some a d d i t i o n a l reasons, when the impurity
F
pn = v
EalkT
= N N e- .
c
(4.9)
- This very important relationship can be derived on the basis o f the law
of mass action without even obtaining the individual carrier concentration
equations. I t always holds, provided the semiconductor is under equilibrium conditions. Thus we can employ it as the criterion for e q u i l i b r i u m
in semiconductors.
Equation 4.9 also indicates that the intrinsic carrier concentration
should depend on temperature approximately in an exponential manner
n cc e~ * ,
in agreement with the experimental results shown in
Figure 4.4. (The actual temperature dependence is somewhat stronger
because N and N themselves also increase with temperature.)
Bati
c.
Space-Charge N e u t r a l i t y
A semiconductor in which impurities are uniformly distributed w i l l

obey the condition o f space-charge neutrality, which means that the net
charge density p in any volume element o f the semiconductor will be zero. -
TABLE
IMPORTANT
4.1
PROPERTIES
OF GERMANIUM,
Atomic or molecular weigh!

Atoms or niolecules/cnv
Crystal structure
ARSENIDE,
A N D OF SILICON
DIOXIDE
Gc
Si
GaAs
SiO,
72.60
28.09
144.63
60.08
4.42 x
SILICON, GALLIUM
IO
22
5.00 x
10
2.21
22
IO
Diamond,
Dinmoncl,
Zinc-blende,
8 a(onis/iini(
8 a(oms/uni(
8 atoms/unit
IO
27X.
72
R a n d o m network of SiO,|
(ctrnhed in.
50% cova lent, 50% ionic
cell
cell
cell
2.3 x
22
AT
bonding
Lattice constant (A)
5.66
5.43
5.65
Density, p (g/cm")
5.32
2.33
5.32
2.27
Energy gap (cv)
0.67
I.I 1
1.40
~8
rn
3
n
Effective density of states

r
:l
conduction band A ( c m " )

r
valence band N
1.04
6.0 x IO'"
(enr")
2.8 x I 0
x'lO'"
4.7 x 10"
, n
7.0 x I 0 '
1.04 x 10'"
J
O-
c
Intrinsic carrier concentration
2.4 x
-1
/;,- (cm" )
10
1.45 x 10'"
9 x 10"
Lattice (intrinsic)
2
mobili!ics (cm /v sec)

electrons
3900
1350
.8600
holes
1900
480
250
Dielectric constant
16.3
1 1.7
12
Breakdown field (v//i)
-8
-30
-35
600
Meiling point ( " Q
937
1415
1238
1700
V;ipor pressure
(ton)
10
a( 880 C
IO"" nl 750 C
10 "'al 1250'C
10
Insulator; p > 10'" fl-cm

al 300 K .
3.9
1 al I050"C
al I 0 5 0 X
100 at 1220 C
10
at 1450 C
10 ' nl 1700 C
Specific heal, C
(Joulc/g"C)
0.31
0.7
0.35
1.0
0.6
1.5
0.81
0.014
0.36
0.9
0.44
0.006
5.9 x 10 !
0.5 x 10 "
Thermal conduclivily, /,,

(wn(l/cm'C)
T h e r m a l cliffusiviiy
(at)
= ts.
pC \sec 1
Linear cocfficicn! of
llicrmal expansion
5.8 x 10*
2.5 x
10'
! . J
M M | >M *
I-- <1 >X-Y.
Elements
of S e m i c o n d u c t o r
Physics
Important
F o r m u l a s f o r S e m i c o n d u c t o r s " in
Equilibrium
105
and
P p = ' V A - X
(4.15)
A g a i n , combining the above formulas for the concentration o f majority

carriers with the equilibrium relationship yields the concentration o f the
minority
carriers:
1
Pn =
- (4.16)
*-N -Nj
and
K = Z
N
"'"
N
(4.17)
I n the above formulas the subscripts refer to the type of the semiconductor (for example, n denotes the concentration o f electrons in an /i-type
semiconductor). The carrier whose concentration is the larger o f the two
is referred to as the majority carrier and the other one is referred to as the
minority
carrier.
Using Equations 4.12 and 4.13 together w i t h 4.6 and 4.7, we can calculate the position of the Fermi level w i t h i n the forbidden gap as a function
o f temperature, for a given acceptor or donor concentration. Such
calculations are shown in Figure 4.7 for silicon. Note that as the temperature increases, the Fermi level approaches the intrinsic Fermi level, i.e..
the semiconductor becomes intrinsic. (This figure also indicates the slight
variation o f the silicon energy gap with temperature.)
n
Fig. 4.7
400
200
300
Temperature ( K )
100
500
The Fermi level in silicon as a function of temperature, for various Impurity
concentrations.
T o obtain the net charge density, we add up all posittve and negative
charges. Thus, i f the donors and acceptors are fully ionized.
=q{p-n
+ N
N ).
A
10
(- )
Hence, the c o n d i t i o n o f space-charge neutrality implies that
We can combine the c o n d i t i o n o f space-charge neutrality w i t h the

e q u i l i b r i u m relationship, np -
This leads to the concentration o f
electrons i n an n-type semiconductor, in e q u i l i b r i u m .

(4.12)
and the concentration o f holes in a /7-type semiconductor
(4.13)
W e can see that when the magnitude o f the net i m p u r i t y concentration
| / y - 2VJ is much larger than the intrinsic carrier concentration n
D
(this is generally the case for silicon at r o o m temperature), the above

relationships simplify to
n ^ V - N
- (4-14)
n
Conversely, we can consider the concentration o f majority carriers as a

function o f temperature. Figure 4.8 shows experimental measurements
o f the electron concentration in rt-type silicon as a function o f temperature.
A t low temperatures the thermal energy in the crystal is not sufficient to
ionize all o f the donor impurities present. Thus the electron concentration
is less than the donor concentration. As the temperature is increased,
the c o n d i t i o n o f complete ionization, where the electron concentration
equals the donor concentration, is approached. As the temperature isfurther increased, the electron concentration remains practically constant
over a considerable temperature range. However, as the temperature is
increased even further, we reach the condition where the intrinsic carrier
concentration becomes comparable to the donor concentration. Beyond
this temperature the semiconductor becomes intrinsic. Thus it is evident
that a semiconductor may be intrinsic even i f it is doped with a relatively
high concentration o f donors or acceptors i f the temperature is high
enough so that the intrinsic carrier concentration exceeds the donor or
acceptor concentration. (This is evident also from the calculations shown
in Figure 4.7.) The temperature at which the semiconductor becomes
E l e m e n t s of Semiconductor Physics
106
T r a n s p o r t of Electrons and Holes
107
T o see what factors influence the drift velocity o f electrons, we can consider the f o l l o w i n g simplified picture. The magnitude o f the drift velocity
at a given time 1 after a collision will be given by r ( r ) = t>(0) + at where
r(0) is the drift velocity immediately upon collision. We will take i t as
zero. This is equivalent to assuming that the electrons suffer collisions
R a n d o m thermal
m o t i o n of
100
200
300
400
500
600
700
7TK)
Fig. 4.8
Electron concentration in n-type silicon as a function of temperature.
Electric field
intrinsic o f course depends on the concentration o f donors or acceptors

i n the semiconductor.
T h e above formulas pertaining to semiconductors i n thermal equilibrium
are summarized in Table 4.2 at the end o f this chapter.
4.5
a.
TRANSPORT
OF ELECTRONS
AND
Dritt c o m p o n e m
due t o electric
field
(bj
Combined motion
of electron in
electric f i e l d
HOLES
Drift
Let us consider an n-type semiconductor, w i t h u n i f o r m donor concent r a t i o n , i n the absence of an applied electric" field. The electrons in the
semiconductor will undergo a continual r a n d o m thermal m o t i o n interrupted by collisions, as illustrated in Figure 4.9a. The thermal motion
leads to no net displacement o f the electrons over a long enough period of
time.
I f an electric field is applied to the semiconductor sample, an additional
velocity component will be superimposed upon the thermal m o t i o n of
the carriers. This additional velocity component, called the drift velocity,
w i l l have a direction opposite to (for electrons) the electric field. The
displacement of the electron due to this component is shown in Figure
4.9b. W h e n we add these two components, we obtain the combined
displacement o f electrons, as illustrated in Figure 4.9c.
Fig. 4.9
Illustration of the motion of an electron in a crystal.
which completely randomize their m o t i o n . The magnitude o f the accelerat i o n a is given by Newton's second law as a = qtf/m*, where m* is the
effective mass o f the electrons in the semiconductor crystal. The effective
mass is a quantity which takes the place o f the mass o f a free electron m
in various calculations. I t essentially corrects for the effect o f the presence
of the semiconductor crystal lattice on the behavior o f the electron.
I f the time interval between collisions is f . then the average drift
velocity o f the electrons will be
c o l )
108
T r a n s p o r t of Electrons and
Physics
109
is the sum o f the probabilities o f collisions due to the various such scattering mechanisms, i.e.,
where
q'cou
L'clritt
Holes
(4.19)
41.
2m*
'coll
is the mobility o f the electrons.!

' Thts treatment assumes that the time interval between collisions f
is independent of the applied electric field. This is a reasonable assumption
onlv so lono as the d r i f t velocity is small i n comparison to the thermal
velocity o f carriers w h i c h is about 10' cm/sec for Si at r o o m temperature.
'coll, impurity
'coll, lattice
or
c o l l
10 X 1 0
^ - F ^
.
electrons
0.2
0.4
'.6
0.8
1.2
1.4
1.6
1.8
P-i
(4.20)
P-L
which correspond to the two most important scattering mechanisms,

impurity and lattice scattering.
I m p u r i t y scattering is due to the fact that when an electron travels past
a fixed charged particle, e.g., an ionized acceptor or donor, its path will be
deflected by the charge on that fixed particle. The probability o f i m p u r i t y
scattering will depend on the total concentration o f ionized impurities
C present in the crystal, i.e., the sum o f the concentrations o f negatively
and positively charged ions. The mobility dominated by impurity scattering fti can theoretically be shown to be p r o p o r t i o n a l to
T^-jCj.
Lattice scattering is due to the thermal vibration o f the atoms o f the
crystal lattice which disrupts the periodicity o f the lattice and thereby
impedes the m o t i o n o f electrons. Theoretical analysis shows that the
m o b i l i t y dominated by lattice scattering u is p r o p o r t i o n a l to 1/7**.
Experimentally measured' electron and hole mobilities in silicon at r o o m
temperature are shown in Figure 4.11 versus the total ionized i m p u r i t y
concentration C . (t is seen that the mobility reaches a m a x i m u m value
at low impurity concentrations corresponding to the lattice-scattering
l i m i t a t i o n , and that both electron and hole mobilities decrease w i t h i n creasing impurity concentration, eventually approaching a m i n i m u m value
at high concentrations. I t can also be seen that the mobility o f electrons
is larger than the mobility o f holes. This is the case in many semiconductors.
T
"
Z.
- f " 1
2x10*
c?' (v/cml
F i g . 4.10
Effect of e l e c t e e field o n t h e m a g n i t u d e of t h e d r i f t v e l o c i t y of c a r r i e r s i n
silicon.
As the drift velocity becomes comparable to the thermal velocity, its

dependence on electric field will begin to depart from the simple relationship given above. T h i s is illustrated by the experimental measurements
o f the d r i f t velocity o f electrons and holes in silicon as a function o f the
electric field, shown i n Figure 4.10. Evidently an initial straight-line
dependence is followed by a less r a p i d increase as the electric field is
increased. A t large enough fields, a m a x i m u m drift velocity seems to be
approached.
3
b.
Mobilities of E l e c t r o n s and H o l e s
The time interval between collisions is determined by the various mechanisms by which the electrons or holes can lose their acquired drift
velocity. The p r o b a b i l i t y o f a collision taking place in a unit time, l / f ,
c o l l
Experimental measurements o f the influence o f temperature on the

m o b i l i t y o f holes in silicon are shown i n Figure 4.12 for two different
i m p u r i t y concentrations. We can distinguish two regions: at low temperatures, i m p u r i t y scattering dominates and separate curves are observed
for the different doping concentrations. A t high temperatures, lattice
scattering dominates and the impurity concentration has little effect on
the mobility as evidenced by the merging of the curves. The mobility is
seen to decrease with increasing temperature in this range. Experimentally,
mobilities have been found to follow a 7"~-- dependence rather than
the theoretically predicted. 7" - dependence in the lattice scattering
ranee.
5
-1
This argument is actually rather oversimplified. A more careful analysis would lead
to a similar formula but without the factor 2 appearing in the denominator.
E l e m e n t i of S e m i c o n d u c t o r . Physics
I IO
T r a n s p o r t of Electrons and Holes
c.
1 0
II I
C o n d u c t i o n in Homogeneous S e m i c o n d u c t o r s ;
Resistivity
i IO
Let us n o w examine the representation o f the conduction process i n

terms of the band diagram, taking an n-type semiconductor as an example.
Figure 4.13a shows an n-type semiconductor sample w i t h no bias applied.
Figure 4.13b shows the same sample w i t h - f 2-volt bias applied to one o f
the terminals.! I n the lower half of the figure we show the corresponding
energy bands as a function o f position along the semiconductor sample.
JD
-Type
1-1
n-Type
',0'
10 '-
1 0
1 0 "
1 0
10'*
1 0
1 0
-*E
'
- 3
Tola! impurity concentralion C ( c m )

T
Fig. 4 . I I The effect of the total ionized impurity concentration on the mobility of
carriers In silicon at room temperature.' Also shown are the corresponding values of
difiusivity.
200
(a)
Fig. 4.13
150
lb)
Illustration of the conduction process in an n-type semiconductor.
The electrons in the conduction band will accelerate, as a result o f the

applied field. D u r i n g acceleration, they will neither gain nor lose a significant amount o f their total energy so they will proceed along a more or
less horizontal p a t h i n the energy band representation. W h i l e the t o t a l
energy of the electrons does not change significantly d u r i n g the acceleration
process, they w i l l lose potential energy, and w i l l increase their kinetic
energy at the expense o f this loss in potential energy. This is evidenced by
the fact that the electron trajectory takes the electron some distance above
the conduction-band edge (recall that the conduction-band edge represents
the potential energy o f the electron).
100
When an electron suffers a collision, it loses some or all o f its kinetic

energy and imparts it to the semiconductor lattice. I n this manner tl:c
kinetic energy o f the electrons will be convened to heat. After the electron
has lost some or all o f its kinetic energy, it will again begin to accelerate
100
Fig. 4.12
200
3U0
400
T CK)
500
600
700
Illustration of the effect of temperature on the mobility of carriers in silicon.
t We ignore any effects associated with the entry or exit of the electrons into and out of
the semiconductor, i.e., we will ignore metal-semiconductor contact effects.
1
I 12
Physics
113
and the same process will be repeated many times. This is how the conduction process appears in the band d i a g r a m . Conduction by holes can be
pictured by a similar but opposite process.
In this discussion we have considered a homogeneous semiconductor,
i.e., a semiconductor in which the i m p u r i t y concentration is spatially
uniform. I n order to keep the electron concentration spatially uniform
also, we had to indicate the Fermi level i n Figure 4.13 at the same distance
from the conduction band at all points. Thus" the Fermi level follows the
conduction-band edge in a parallel fashion.
The current flowing in the semiconductor sample w i l l be given by
I = q n Pfetn A = qnu
V
- A
(4.21)
where V is the voltage applied across the semiconductor sample which has
a length L and a cross-sectional area A and ,u is the electron mobility.
The resistance o f such a sample is given by
R =
- .
,4
Thus by comparison o f the two formulas we find that the resistivity p o f

the >i-type semiconductor sample is
1
P =
Impurity concentration ( c m " )
(4.22)
(4.23)
F i g . 4.14
R e s i s t i v i t y o f s i l i c o n at r o o m t e m p e r a t u r e as a f u n c t i o n o f a c c e p t o r o r d o n o r
impurity concentration.*
Figure 4.14. This figure is based on an extensive survey o f large numbers o f

measurements on samples which contain either acceptor or donor i m purities. *
Similarly, the resistivity o f a /'-type semiconductor sample is

Diffusion
1
P ~
(4.24)
qpvP
I n general, when both carriers are taken into consideration, the resistivity
is given by
p =
(4.25)
<f(rV' + , vP)
The mobilities, as discussed above, depend o n the total concentration o f
ionized impurities, hence, on the sum o f the acceptor and donor concentrations. The electron and hole concentrations, on the other hand, depend on
the difference o f the acceptor and d o n o r concentrations. Thus, in a general
case, the resistivity must be calculated by using the mobility data given i n
Figure 4.11 and carrier concentrations based on Equations 4.14 or 4.15.
However, i n the case when only one type o f i m p u r i t y is present in the semiconductor, the resistivity will become a simple function o f the concentration o f that i m p u r i t v .
The resistivity o f b o t h p- and n-type silicon at r o o m temperature as a
function o f acceptor or donor concentration, respectively, is shown i n
The discussion in this section has so far dealt only with the case when
the electron concentration w i t h i n the semiconductor sample is u n i f o r m
and the electrons move under the influence o f an electric field. I f the electron concentration is not uniform, electrons will also diffuse under the
influence o f the concentration gradient. As in the case o f ions, this w i l l
lead to an additional contribution to the expression for the flux. This
c o n t r i b u t i o n is p r o p o r t i o n a l to the concentration gradient, and the proportionality constant is the diffusivity o f electrons, D. The diffusivity o f
electrons, in turn, is related to the electron mobility by Einstein's relationship.
n
(4.26)
Similar considerations apply TdThe transport o f holes.

The diffusivity o f electrons and holes at r o o m temperature can be
obtained from Figure 4.11 where the right-hand axis is labeled in terms o f
diffusivities-.
Elements
I M
READING
Physics
Problems
level as a reference, label the electron energy E, and the electrostatic potential
<h corresponding to each level.
REFERENCES
4.3
The band theory of. and electronic conduction in, solids is discussed in detail in a
varietv of texts on solid-state and semiconductor physics. See. for instance. Chapters
10-13 in A . J . Dekker, Solid Swie Physics, Prentice-Hall. 1957: and Chapters 1-5 in
J. L . Moll, Physics of Semiconductors. McGraw-Hill Book Co.. 1964. A tutorial
"treatment of band theory is given bv F . Herman, "The Electronic Energy Band Structure
of Silicon and Germanium," Proc. IRE. 43,1703 (1955).
4.4
Prepare plots of the Fermi-Dirac distribution function at 78C, room temperature, and 500C. Compare the three curves.
4.5
A small concentration of minority carriers is injected into a homogeneous semiconductor crystal at one point. A n electric field of 10 v/cm is applied across the
crystal, and this electric field moves these minority carriers a distance of 1 cm in
a time 250//sec. Determine the drift velocity and the diffusion coefficient of the
minority carriers.
4.6
For an electron mobility of 500 cm /v sec, calculate the lime between collisions.
For anelectric field of 100 v/cm. calculate also the distance traveled by an electron
between collisions. Take m* ~ m in these calculations.
4.7
Calculate the electron and hole concentrations, the resistivity, and the position of
the Fermi level of a silicon crystal containing 1.1 x 10" boron atoms/cm and
9 x 10" phosphorus atoms/cm , at 2 7 C .
CITED
1. These intrinsic carrier concentrations are based on:

a. R . N . Hall and J . H . Racette. "Diffusion and Solubility of Copper in Extrinsic
and Intrinsic Germanium, Silicon, and Gallium Arsenide." J. Appl. Phvs.. 35,
379 (1964).
b. F . J . Morin and J . P. Maita, "EJec:rical Properties of Silicon Containing Arsenic
and Boron." Phys. Rev., 9fi, 28 (1954).
c. F . J. Morin and J . P. Maita, "Conductivity and Hall Effect in the Intrinsic Range
of Germanium," Phys. Rev., 94, 1525 (1954).
2. F . J . Morin and J. P. Maita, "Electrical Properties of Silicon Containing Arsenic
and Boron," Phys. Rev., 96, 28 (1954).
Draw the energy band diagram corresponding to the above cases with 20 volts
applied across the silicon sample. Keep the diagrams to scale. Calculate the
current flow in each case if the length of the sample is 100 fi, and its cross-sectional
area is 1 0 cm".
-3
For the electronic properties of semiconductors, see E . M . Conwell. "Properties of

Silicon and Germanium," Proc. IRE, 46, 1281 (1958); and O. Madelung. Physics of
III-V Compounds. Wiley, 1964.
REFERENCES
I 15
lf
4.8
Calculate the Fermi level of silicon doped with 10". 1 0 , and 10" donor atoms/cm
at room temperature, assuming complete ionization. Then, using the Fermi levels
you have found, check if this assumption is justified in each case. In these calculations, take the donor level to be 0.05 ev below the conduction band edge.
4.9
Give the equilibrium electron and hole concentrations, mobilities, and resistivity
for silicon at 27 C. for each of the following impurity concentrations:
(a) 3 x l O " boron/cm .
(b) 1.3 x 10" boron/cm -r 1.0 x 10" phosphorus/cm .
(c) 1.3 x 10" phosphorus/cm + 1.0 x 10" boron/cm .
(d) 3 x lO^phosphorus/cnv' -j- 1.0 x 10 ' gallium/cm -f i.O x 10 arsenic/cm .
C
3. E . J . Ryder, "Mobility of Holes and Electrons in High Electric Fields,"' Phys. Rev.,
90, 766'()953).
4. E . M . Conwell. "Properties of Silicon and Germanium." Proc. IRE. 46. 1281 (1958).
5. G . L . Pearson and J . Bardeen, "Electrical Properties of Pure Silicon and Silicon
Alloys Containing Boron and Phosphorus." Phys. Rev., 75, 865 (1949).
6. J . C . Irvin, "Resistivity of Bulk Silicon and of Diffused Layers in Silicon," Bell
System Tech. J., 41, 387 (1962).
l:
4.10 Repeat, at 300 C. Compare the two cases and discuss.

4.11 Gold in silicon introduces an acceptor level 0.54 ev below the conduction band
edge, and a donor level 0.35 ev above the valence band. What will be the state of
charge tor occupation) of the gold levels in silicon doped with:
(a) High concentration of donor atoms (with respect to the gold concentration) ?
(b) High concentration of acceptor atoms?
What is the effect of gold on the electron and hole concentrations, respectively?
PROBLEMS
4.1
(a) Verify the formula for the intrinsic Fermi level, Equation 4.8.
(b) Calculate the location of the intrinsic Fermi level of silicon at 78 C. 27C,
and 300 C. Is it reasonable to assume it is in the center of the forbidden gap?
C
,:
4.2
Draw the energy band diagram:

(a) at - 7 8 C , '
(b) at room temperature, and
(c) at 300 C for silicon doped with 10" boron atoms/cm . Show the band edges,
the intrinsic Fermi level, and the actual Ferm level. Using the intrinsic Fermi
C
4.12 Using the information given in the previous problem, determine the charge state
of the gold levels and the position of the Fermi levels in a silicon crystal containing
gold atoms only. Is the crystal p- or H-type ?
Elements of Semiconductor Physics
li
INJECTION
KINETICS OF T H E
RECOMBINATION
PROCESS
LIFETIME FOR LOW-LEVEL

INJECTION
SURFACE
ORIGIN OF RECOMBINATIONGENERATION CENTERS
RECOMBINATION
5
T A B L E 4.2
Semiconductors under
Non-Equilibrium Conditions
IMPORTANT FORMULAS IN SEMICONDUCTOR PHYSICS
Complete ionization of impuritiesThermal equilibrium

Charge neutrality
Equilibrium condition
p == q{p - n + N
p"
<V,i) = 0
= "t
Fermi-Dirac distribution function
Carrier concentrations in
non-degenerate semiconductors:
n e*
t
In the extrinsic case,

/ 7
" -
tfj
E.Zy> kT
Most semiconductor devices operate u n d e ^ n o n j ^ u ^ b r i u m conditions.

i.e., under conditions in which the carrier concentration product pn
differs f r o m its equilibrium value, nf. The performance of many semi-""
conductor devices is determined by their tendency to return to equilibrium.
In this chapter we derive and examine the quantities which characterize
this tendency: lifetime, diffusion length, and surface recombination velocity.
We begin by discussing the concept o f injection o f excess carriers. The
process o f return to equilibrium then is considered through examples o f
two types: transient cases, and steady-state cases in which the distribution
of excess carriers is non-uniform. We show that both types o f cases can be
characterized by the lifetime o f the excess carriers, their diffusion length,
and the surface recombination velocity. In order to relate these quantities
to characteristics o f the semiconductor, we then study the mechanism o f
recombination, both in the bulk and at the surface. Finally, we discuss the
physical origin o f bulk and surface recombination centers.
5.1
INJECTION
Let us consider non-equilibrium situations in which the condition

r.p = nf is violated. Accordingly, we can distinguish between two types
of deviation from equilibrium. I n the first type, when np > n,-, we talk
117
Qg^
Semiconductors under Non-Equilibrium Conditions
118
of injection of excess earners. I n the second, when np < n, . we talk o f

the extraction
of carriers from the semiconductor.
Injection
119
preserve space-charge neutrality. I n the example shown i n Figure 5.1b

we have increased the m i n o r i t y carrier concentration a h u n d r e d - m i l l i o n
fold, from 10 to 10 c m . A l t h o u g h at the same rime we have also added
approximately 1 0 c m
majority earners to the semiconductor, this
concentration o f excess electrons is negligibly small i n comparison to the
concentration o f electrons already present in the n-type semiconductor.
Thus, while the magnitude o f the change in the concentration o f electrons
is the same as the change in the concentration o f the holes, the percentage
change in electron concentration is negligible. T h i s c o n d i t i o n , i n which
the excess carrier concentration is negligibly small i n comparison to the
d o p i n g concentration, i.e., A n = Ap A ^ , is referred to as
low-level
injection.
J
12
1 2
a.
Injection Level
The meaning of injection level is illustrated in Figure 5.1. Here we

indicate the donor concentration N 1 0 c m and the majority and
m i n o r i t v carrier concentrations in an n-type semiconductor under equilibr i u m , and low- and high-level injection conditions.! A s we have seen in
1 6
- 3
- 3
- 3
For comparison, we also illustrate the case o f high-level injection in

Figure 5.1c. I n high-level injection the injected excess carriers are i n
larger, or at least in comparable, concentrations t o the concentration o f
the donor ions. Thus in this case the percentage change in m a j o r i t y carrier
concentration is no longer negligible. Even though high-level injection is
often encountered in semiconductor device operation, because o f the
complexities involved i n its treatment we shall consider only low-level
injection.
a, for Si at
300 K
b.
(a) Equilibrium
Fig. 5.1
(b) Low-level injection
(c) High-level injection
Illustration of the concentration of electrons and holes in an n-type semicon-
ductor in-equilibrium and under low- and high-level injection conditions.
Chapter 4. under equilibrium conditions the product o f majority and

m i n o r i t y carrier concentrations must equal n , . or approximately 10 c m
i n silicon at r o o m temperature. The majority carrier concentration approximately equals the donor concentration. Hence,
2
1 6
10 cm-
as shown in Figure 5.1a.

type
tv
and
Pm = 10 c m -
- 6
Whenever the carrier concentrations are disturbed f r o m their e q u i l i b r i u m

values they w i l l attempt to return to equilibrium. I n the case o f injection
o f excess carriers, return to e q u i l i b r i u m is t h r o u g h recombination of the
injected m i n o r i t y carriers with the majority carriers. I n the case o f extraction o f carriers, return to e q u i l i b r i u m is through the process o f generation
o f electron-hole pairs.
We now develop the parameter which characterizes the rate o f r e t u r n
to e q u i l i b r i u m . This parameter, as well as certain other concepts o f the
utmost importance in semiconductor device operation, can best be
developed by considering specific examples, starting with the simplest
case.
I n this notation the first subscript refers to the
o f the semiconductor.
quantity. Thus n,
20
Return to Equilibrium
and p
nB
The subscript o indicates an equilibrium

denote the electron and hole concentrations,
respectively, i n an n-type semiconductor in equilibrium.

Let us now consider the case when we somehow introduce excess carriers
o f both types into the semiconductor, in equal concentrations in order to
t Throughout this chapter we will use an n-type semiconductor as an example. All
results, nowever, are equally applicable to a />-type semiconductor.
D e c a y o f I n j e c t e d C a r r i e r s . Consider a uniformly, illuminated

semiconductor slice as shown in the inset to Figure 5.-2. We assume t h a t
the light is so energetic that i t creates electron-hole pairs w i t h i n the semiconductor. Furthermore, we assume that the light is absorbed u n i f o r m l y
throughout the semiconductor sample, resulting in a u n i f o r m rate o f
generation G _ o f eiectron-hole pairs per unit v o l u m e t h r o u g h o u t the
crystal. The resulting spatially uniform d i s t r i b u t i o n o f m i n o r i t y carriers
y
120
Semiconductors
under N o n - E q u i l i b r i u m
Injection
Conditions
121
This assumption has the correct feature that U = 0 in equilibrium. The

constant of proportionality | / r , will have to be determined f r o m a study
of the mechanism o f the recombination process. The constant r (or
r for a p-iypz semiconductor) is referred to as the lifetime o f the excess
minority carriers. The form o f this equation is similar to others we have
used earlier in the treatment o f solid-state processes. In all o f them we have
assumed that a rate is proportional to a driving forcea measure o f the
deviation from equilibrium.
....
.
.
Combining the above relationships, we get the differential equation
describing the minority carrier concentration in the semiconductor as a
function o f time.
U n i f o r m l y absorbed light
PL
dPj,
X
Fig. 5.2 Illustration of the steady-state minority carrier distribution in a uniformly
illuminated semiconductor sample, in the absence of surface recombination.
(5
.4)
I n steady state, dpjdt = 0. Thus the steady-state concentration o f

minority carriers during illumination is given by
in the semiconductor specimen is illustrated in Figure 5.2. Because o f the

increased generation rate resulting from illumination, the concentration
o f minority carriers w i l l build up to a steady-state level high enough to
make the rate o f recombination o f the carriers just equal to the rate o f
generation.
To calculate the-steady-state m i n o r i t y carrier concentration reached
under illumination we first note that the time rate o f increase o f minority
carrier concentration dpjdt equals the total generation rate o f minority
carriers minus the. total recombination rate, or
'
P L
P n o + ' , G
(5.5)
I f the light is now turned off, i.e., G = 0, the excess minority carrier
concentration will begin to decay. This decay is described by the solution
of the differential equation.
L
UZZJM
( 5
.6)
T,
dt
subject to the initial condition,
(5.1)
pM
. .
The solution is given by
dt
where Gr. is the rate o f generation due to the absorbed light, G, is the
rate o f seneratibn i n dark, due to thermal mechanisms, and R is the total
rate o f recombination ( a l l per unit time and unit volume). I t is convenient
to define the net rate o f recombination,
U = R G , and rewrite the
above equation to give
dt
= PL-
PnC) = ~P% + {p
(5-7)
pje-'l".
This solution is illustrated for various values o f the lifetime r

5.3.
(5.8)
p
in Figure'
tK
U.
(5.2)
dt
By this definition the net rate o f recombination U 0 in the steady state
if there is no injection by light or other means.
Before we can solve this equation, we need a relationship between U
and the m i n o r i t y carrier concentration. We assume the simplest possible
relationship, namely that U is p r o p o r t i o n a l to the excess m i n o r i t y carrier
concentration, or
1
(5.3)
U
(Pn
~
Pnoi-
ft M
Surface R e c o m b i n a t i o n in Steady S t a t e . Let us again consider

a semiconductor slice which' is uniformly illuminated as i n the previous
case. Let us further assume that the recombination rate o f the excess_
carriers is enhanced at o n e l i d e o f this slice, as illustrated in the inset to
Figure 5.4. Because o f the enhanced rateoT recombination at the surface
corresponding to the plane x 0. the concentration o f the excess m i n o r i t y
carriers will be lower at this plane than in the body o f the semiconductor
sample. _As a result, minority and majority carriers will flow to this surface
and recombine there. Because one electron recombines w i t h one hole, the
, flux o f holes ojhe_sjJrface^ _wilI precisely equal the flux o f ejectrons F .
T h u s there will be no net current flowing.
t
122
123
Injection
Semiconductors under N o n - E q u i l i b r i u m C o n d i t i o n s .
including the generation and recombination terms i n t o the balance over

the control-element in Fisure 3.6. The flux o f holes is siven by
P*
(5.10)
CX
where > and u , are the diffusivity and m o b i l i t y o f holes, respectively

(see Chapters 3 and 4).
This p r o b l e m would be greatly simplified i f we could neglect the drift
term. i.e.. the second term-of the fluv equation. T o show that for minority
zzr\ ie/'j in low-level injection the drift term is in fact negligible in comparison
to the diffusion term, we recall that the flux o f holes to the surface equals
the flux o f electrons, which is given by
P
4
5
6
7
t (arbitrary units)
(5.11)
r:
(5.9)
dx
(D,
=-
I n low-level injectior
eiven bv
where F. denotes the flux of holes.

Equation 5.9 can be derived by considerations exactly like those we
employed in the derivation of the transport equation i n Chapter 3, but
4>
D )(opJBx)
Mr,
(5.12)
Pv",,
: thus the drift term in E q u a t i o n 5.10 is
OPr.
P-v
Pr,
ox
u, r...
* fn
"n
v
ex
"*
2
n
*s .'
r
(5-13)
"
Since p n, this term is clearly negligible in comparison to the diffusion

term.
I n contrast, i t can be readily shown by using a similar argument that the
drift term for majority carriers is not negligible in comparison to the
diffusion term. Since diffusion and drift contributions to the m a j o r i t y
carrier flux are o f opposite sign, the net flux o f majority carriers to the
surface is given by their difference.
n
Uniformly absorbed light
Surface
recombination
F.
P.(x)
p6n
where D and /u are the diffusivity and m o b i l i t y o f electrons, respectively.

F r o m the equality o f the t w o fluxes F, = F, and from the requirement o f
space-charge neutrality due to which dnjox
=
dpjcx,
we o b t a i n the
expression for the electric field,
I D this problem b o t h minority and majority carrier distributions vary

spatially. Thus the hole distribution is described by a solution o f the
transport equation,
""
- D ~
ox
Fig. 5.3 The decay of excess minority carrier concentration as a function of time in
the transient case, for various values of the lifetime.
di
PL
Fig. 5.4 Illustration of the steady-state minority carrier distribution in a uniformly

illuminated semiconductor sample, vcith surface recombination.
A graphical illustration o f the relative magnitudes o f the four flux terms

i n l o w - and in high-level injection is given in Figure 5.5. I t is evident that
whereas i n high-level injection all four flux terms are o f comparable
magnitude, in low-level injection only the diffusion term provides an
i m p o r t a n t c o n t r i b u t i o n to the flux o f minority carriers.
A l t h o u g h the above argument is for a case with no current flowing, this
result is o f general validity and greatly simplifies the study o f the m o t i o n o f
m i n o r i t y carriers in low-level injection. Thus the transport o f m i n o r i t y
Semiconductors
124
under
Non-Equilibrium
Conditions
Injection
125
. All four terms

comparable
Majority carrier
diffusion term
Maiority carrier
drift term
Minority carrier
diffusion term
Minority carrier
drift term
10
fa) Low-level injection
Fig. S.3
(b) High-level injection
Illustration of the relative magnitudes of the diffusion and drift terms for the
case of steady-state surface recombination.
This solution is shown in Figure 5.6 for various relative values o f surface
recombination velocity s . I n the l i m i t when the surface r e c o m b i n a t i o n ^ ^
velocity s - 0, the solution reduces to the case illustrated in Figure 5.2. '
In the other limit, when s co, the minority carrier concentration at the
surface approaches its equilibrium value. The d i s t r i b u t i o n of minority
carriers for this case reduces to
s
carriers is described simply by the diffusion equation w i t h the generation
=
dt
O'Pn
G,
Pno
^0
(5.14)
" dx*
Returning now to the p r o b l e m o f surface recombination, in steady state

dpjdt
?.(*)
= 0. We then seek a solution to the resulting ordinary differential
IL
- PL ~ (PL ~ Pno)e-* .
(5.18)
. .
This d i s t r i b u t i o n is shown in Figure 5.7 for various values o f L.f
equation subject to the b o u n d a r y conditions

/>(>) = PL -
Pno + rfif,
fc
(5-15)
and
D
OPn
= s [p {0)
v
(5.16)
p ].
na
The second boundary c o n d i t i o n states that the m i n o r i t y carriers. which

reach the surface recombine there.
As in the case o f bulk recombination,
we assume that the r e c o m b i n a t i o n rate at the surface, is p r o p o r t i o n a l to \

the concentration o f excess m i n o r i t y carriers (p
p) there. The propor-
tionality constant s, w h i c h has the units centimeter per second is called
the surface recombination
velocity.
The solution o f this boundary value
problem is
* - IT
x
Pn( )
where L
= yjDT
PL
(PL
Pno)
is called the diffusion
-tIL,
S t e a d y - S t a t e I n j e c t i o n f r o m O n e B o u n d a r y . A case corresponding to a different physical situation is illustrated in Figure 5.8. Here

we illuminate the semiconductor sample from one side only, and in such a
way that all o f the light is absorbed in a very t h i n layer next to the semiconductor surface. Thus in this thin layer a large excess minority carrier
concentration is set up. The excess minority carriers diffuse inward i n t o
the undisturbed body of the semiconductor from this surface layer as
illustrated i n Figure 5.8. Since no net current is entering the semiconductor
sample, the majority carrier concentration d i s t r i b u t i o n must again be such
that the net flux o f electrons and o f holes w i l l be i n the same direction and
equal in magnitude.
f
(5.17)
length o f minority carriers.
It is interesting to note how similar this problem is to diffusion of impurities subject

to external rate-limitation which we considered in Chapter 3. Accordingly, these two
limiting cases could be called 'surface-recombination controlled' and 'diffusion controlled,' respectively.
-V)
126
Kinetics of the Recombination Process
127
which is shown in Figure 5.9 as a function o f distance, for three different

values o f the diffusion length L .
Note the complete analogy between the steadv-state d i s t r i b u t i o n o f
excess m i n o r i t y carriers as a function of distance, given by E q u a t i o n 5.22
and shown in Figure 5.9. and the time-wise decay o f the excess m i n o r i t y
carrier concentration, given by Equation 5.8 and shown in Figure 5.3.
I n each case there is a disturbance o f m i n o r i t y carrier concentration. I n
one case, the disturbance occurs at a given p o i n t i n space; in the other, at
a given'point in time. W i t h increasing distance, or w i t h a passage .of t i m e ,
p
l
"0
TO
20
30
40
50
60
,\ (arbitrary units)
70
80
90
I I I
I r
100
Fig. 5.7 The steady-state distribution of excess minority carriers for various values of
the diffusion length, for
oo.
The steady-state m i n o r i t y carrier distribution within the b u l k o f the

semiconductor sample must satisfy the equation
p
^' " _ * "

dxT
(5.19)
subject to the boundary conditions,

^ ( 0 ) = constant, function o f light intensity
(5.20)
X (arbitrar) units)
and
(5.21)
The solution is
P (z)
n
Puo + [p(0) -
zlLr
p o)e~
n
(5.22)
Fig. 5.9 The steady-state distribution of excess minority carriers for various values of
the diffusion length.
the e q u i l i b r i u m m i n o r i t y carrier concentration is then approached. I n

each case the rate o f approach to equilibrium is related to the same
parameter, the lifetime (or, equivalent))', the diffusion length) o f m i n o r i t y
carriers.
5.2
KINETICS
PROCESS
OF T H E
RECOMBINATION
In the previous section we have seen that the rate o f return o f the m i n o r i t y carrier distribution to equilibriumwhether such return takes place
with passing time or with increasing distancedepends on the rate conFig. 5.8 Illustration of the minority carrier distribution in a sample illuminated on one
side with non-penetrating light.
slant characterizing the recombination process 1/TJ,. This resulted
from
our assumption that the net rate o f recombination i n the bulk o f an - t y p e
semiconductor can be described by the linearized expression
U = ( I/T)(p
- p
J .
In this section we attempt to relate the lifetime - . ; to the physical^

characteristics o f the semiconductor-by. considering the mechanisms ot
the recombination process.
a.
Band-to-Band R e c o m b i n a t i o n
When electrons in the conduction band and holes in the valence band
recombine directly, we talk of a band-to-band recombination process which
is illustrated schematically in Figure 5.10.
Here we indicate generation
Altnougn lrtsome semiconductors, e.g., gallium arsenide, band-to-band

recombination can be very important, in germanium and in silicon the
detailed structure o f the energy bands makes this process extremely
unlikely. I n fact, early w o r k with both germanium.and silicon indicated
that the lifetime is extremely structure sensitive in such materials, i.e.. it
is sensitive to the method employed in the preparation o f the semiconductor
samples. We w o u l d expect the rate o f the band-to-band recombination
process to be dependent only on the band structure o f the semiconductor.
Thus the fact that the_method o f sample preparation has an effect on the
lifetime indicates that the recombination process involves imperfections
or impurities in the semiconductor whose concentration could indeed
depend on the fabrication technique.
We now consider the recombination-generation process taking place
through the action o f such imperfections.
b.
Light or .
heat
Light
F i g . 5.10
through I n t e r m e d i a t e Centers"
The band-co-band recombination process.
due to light absorbed in the semiconductor, which has a rate G , and the
net band-to-band recombination, which has a rate U. In such a recombination process the electron must lose energy of the order o f the energy gap.
This energy can be emitted in the form o f light or in the f o r m of heat.
W p would ftifflgct, the rate o f the band-to-band recombination process
L
>CJ
Recombination-Generation
to be p r o p o r t i o n a l to both the concentrations o f electrons and o f holes.
Imperfections w i t h i n the semiconductor can disrupt the perfect periodicity o f the crystal lattice, and as a result can introduce energy levels into the
forbidden gap much as donor and acceptor impurities do. These energy
levels then act as "stepping-stones" in the transition o f electrons and holes
between the conduction and valence bands. Because the probability o f
transitions depends on the size o f the step, imperfections can make such
transitions more probable and, therefore, can exert a drastic influence on
the lifetime in the semiconductor.
The theory o f the recombination-generation process taking place
through the action o f such intermediate energy-level
recombinationgeneration centers has been worked out by H a l l , and by Shockley and
Read. This theory has been remarkably successful in explaining a wide
variety o f phenomena in many semiconductors and semiconductor devices.
Because of this we now consider it in some detail.
The various steps that occur in the recombination and generation process
through intermediate-level centers are shown i n Figure 5.11. I n particular,
we illustrate the state o f the center before and after each of the four basic
processes takes place. The arrows in this figure designate the transition o f
the electron d u r i n g the particular process. This illustration is for the case
of a center w i t h a single energy level which can have two charge states:
negative and neutral.
1
Thus for an /t-type semiconductor,

R = *.Pn,
_ O
(5-23)
where x is a p r o p o r t i o n a l i t y constant.
p?
I n e q u i l i b r i u m , R = G,
= xn p .
no
no
I n low-level injection, the majority
carrier concentration does n o t change significantly, i.e., n
'po
-^\J:
ixffr-
U = R -
G, = y.n [p
h
no
- p ).
n0
= n.
no
Thus
(5-24)
Hence, the lifetime for the band-to-band recombination process becomes
y-n
Process (a) is the capture of an electron from the conduction band by the
center. Process (b) is the reverse processthe emission of an electron from
the center into the conduction band. Process (c) is the capture of a hole
Semiconductors under N o n - E q u i l i b r i u m Conditions
130
Kinetics of the Recombination P r o c e s s .
131
the center. The capture cross-section of the center a is a measure o f

h o w close the electron has to come to the center to be captured." " JWe
rnight expect that the capture cross section would be o f the order o f atomic.
dimensions, i.e.. o f the_order o f ) ~ " c m .
The rate o f electron emissionprocess ( b ) w i l l be p r o p o r t i o n a l to the
concentration o f centers which are occupied by electrons, i.e., K,f.
Thus,
i y |
C^Vb
r,
Before
C
After
(a)
Electron
capture
Before
After
(bl
Electron
emission
Befoie
After
Before
After
(d)
Hole
emission
ic)
Hole
capture
(~.
^
Fig. 5.11 Recombination and generation through intermediate centers. Arrows designate direction of electron transition.
f r o m the valence band by a center. This process can also be described as

the transition of an electron f r o m the center i n t o the valence band.
Finally, process (d) is the emission of a hole f r o m a center into the valence
band. This can also be described as the transition o f an electron from the
valence band to the center, leaving behind a hole i n the valence band.
Let us now consider the rates o f these individual processes. The rate o f
electron captureprocess (a)should be p r o p o r t i o n a l to the concentration o f free electrons i n the conduction band and also to the concentration of centers which are no: occupied by electrons. This is because only
one electron can occupy a given center; thus once a center is occupied by
an electron i t cannot capture another one. I f the concentration o f centers
i n the semiconductor is A ' , the concentration o f unoccupied centers is
given by N,(l f) where / designates the p r o b a b i l i t y o f occupation o f a
center by an electron. I n e q u i l i b r i u m ,
r. -
eNJ.
(5.29)
The p r o p o r t i o n a l i t y constant e . called the emission probability., is the

probability of a j u m p from an occupied center i n t o the conduction band.
Thus it depends on the density o f unoccupied states w i t h i n the conduction
band and also on the location o f the cemer w i t h i n the forbidden gap.
Intuitively, we w o u l d expect that i f the center is shallow, i.e., i f it is close
to the conduction-band edge, this j u m p - p r o b a b i l i t y and hence e would be
large, and vice versa. We will see later that the theory indeed bears out
this guess.
(a) be given by
The rate o f capture o f holesprocess (c)will by analogy to process
r = r/i<vA\/
(5.30)
V
*
^
i
*~
_
~%
Since capture o f holes by a center corresponds to the transition o f an

electron from a center to the valence band, this process is p r o p o r t i o n a l to
the concentration o f centers occupied bv electrons. A , f.
Conversely, the rate o f hole emissionprocess ( d ) w i l l be given bv
CT
T
.A '.
ii
-
rf
r , = e N,(]
-f)
(5.31)
is the Fermi level.

(5.27)
We designate the p r o p o r t i o n a l i t y constant by the p r o d u c t v ,.a.

t
! f t
i.e.,
aÂ (l - / ) .
fa
'
. ?
,,
C_7
'_
V^-'Ec-AXA-r _ lEr-E)/l:T_
ri
j The subscript ; has been traditionally employed to denote quantities pertaining to

recombination-generation centers, it originates from the term nap which we shall not
use in order to avoid confusion.
>
'
j C ~|
Here v is the thermal velocity o f the carriers, i & v'3A.-77w ^ 10 cm/sec

at r o o m temperature. The quantity c can be interpreted as follows. I n
order to be captured, the electron must come to the physical vicinity o f
we obtain
/e
= v.o N e-*>-E,mT
n
CCT
'
^cj
(5.28)
ih
: \Z
r . cc ,;A' (1 - / ) .
r = L
where e is the emission probability o f holes and depends on factors

analogous to those that enter into determining e.
First we would like to evaluate the emission probabilities. e and e .
T o do this, we consider the equilibrium case: w i t h o u t any external
generation mechanism acting, i.e.. when G = 0. I n this case, the rates o f
the t w o processes through which transition into and out o f the conduction
band takes place must be equal. Thus r, = r ,. Substituting the proper
equations for these rate's and recalling from Chapter 4 that the electron
concentration in equilibrium is given by
Jt
where E, is the energy level of the center and E

Thus the rate o f process (a) w i l l be given by
(5.32) <
m T
t Strictly speaking, we should base the value of the thermal velocity on the ejieciive
mass of the carrier whose capture we are considering rather than on the mass of a free
electron, m. However, for simplicity, we include the correction factor, 'wi/itri. with
the respective capture cross sections.
v
~"; )
"
r
i ~ T~
J
f6*x
*
*
^
'
'/'ff*"
'
'I
|32
Semiconductors
under
Non-Equilibrium
Conditions
Noce chat, in agreement w i t h our intuitive.argument, the emission probability o f electrons e indeed increases exponentially as the center energy
level E, approaches the conduction band edge E .
Similarly, i n e q u i l i b r i u m the two processes by which holes enter and
leave the valence band, processes (c) and (d), must have identical rates.
Thus r = r j . Again substituting the appropriate equations and noting
from Chapter 4 that the concentration o f holes in equilibrium is given by
Kinetics of the Recombination
Process
133
steady-state non-equilibrium conditions we can eliminate G

above expressions and write
between the
p -
-<Er-E,mT
Njie
Ê,-E kT
Flf
we o b t a i n
{E
E )lkT
, = v a N e- '- '
th
= v.^^
'-
'
(5.33)
f -
r.
(5.36)
By substituting the proper rates into Equation 5.36 we can solve for the
occupancy f a c t o r / o f the centers under a given non-equilibrium condition
in terms o f the electron and hole concentrations. Note that n e i t h e r ^ e "
expression f o r / , Equation 5.26, nor those for n a n d p are meaningful under
non-equilibrium conditions since the concept o f the Fermi level is only
valid in e q u i l i b r i u m . The electron and hole concentrations instead w i l l V
dependent on the injection level, i.e., on G , thereby m a k i n g / a l s o depend
on the injection level. Thus.f
L
A g a i n we note that the emission probability e increases exponentially as

the center level E approaches the edge o f the valence band E .
P
t g
on 4- , , V , - ' - - ' ' * r '

g
iE
o [n
E ,lkT
+ N e- '- ' ]
,E
+ o[p + i V e - ' - ^ > ' ' ]
'
'
or. in an alternate form,

f =
a [n
n
We
Fig. S.I2
"
<
* '
h
+ n , e * * " ] + a,[p + n^*'~*^ ]'
can substitute these values o f /
'
into the rates of the individual
processes and obtain the steady-state net rate o f recombination through
T r a n s i t i o n s t a k i n g place u n d e r n o n - e q u i l i b r i u m c o n d i t i o n s .
the action o f intermediate centers U.

Let us now consider the situation pertaining to non-equilibrium conditions, such as, for instance, the case o f a uniformly illuminated semiconductor w i t h a u n i f o r m generation rate per unit volume G The
transitions taking place under such conditions are illustrated i n Figure
5.12. N o t e that in a d d i t i o n to the processes (a),'(b), (c), and (d) o f Figure
5.11, electrons now also leave the valence band and enter into the conduction band through the action o f the illumination process. I n steady
state, t h e r a t e by w h i c h electrons enter the conduction band w i l l equal the
rate by which electrons leave the conduction band. Thus
U = r
a o v N,[pn
B
dt
(r f
r j - 0.
(5.34)
Similarly, in steady state the rate by w h i c h holes leave the valence band
equals the rate by which holes enter the valence band and. therefore.
G -(r
L
- r , ) = 0.
(5.35)
dt
Equations 5.34 and 5.35, o c p u r s e . reduce to the equilibrium case when
we set G == 07 ~(Note that steady state does not imply
equilibrium!)~r~6f~
L
th
-(E-/>;,i/*Ti
n?]
P
NJ
(5.39)
-tK,-E,)!lcT\
or, alternatively.
",a v, N.[pn
n
U =
nr]
(5.40)
To see the principal features o f this extremely important relationship,

let us consider a special case when the capture cross sections for electrons/
and for holes are equal, i.e., a = o = a. I n this case Equation 5.40
reduces to
U = ou, N
h
(5.41)
p -r 2n, cosh (
)
V
IcT I
We can identify the " d r i v i n g force" for recombination as pn n -, which
is in fact the deviation f r o m the equilibrium condition. The "'resistance''
t
* We assume here [hat the illumination does not change the rates of the emission and
capture processes other than through the changed carrier concentrations.
Lifetime in L o w - L e v e l Injection
134
for this recorrrbination process increases w i t h 7? and with p. thus that part
o f the resistance will be smallest when the sum (n + p) is at its m i n i m u m
value.
The third term in the denominator increases as , moves away from the
middle o f the energy gap , and approaches either the conduction-band
or the valence-band edge. I n such a case one o f the emission processes
becomes increasingly probable and this reduces the effectiveness o f the
recombination center. This is because after an electron is captured by
the center, a hole must be captured by i t next in order to complete
the recombination process. If, however, the energy level of the center is
very near the conduction-band edge, it will be more likely to re-emit
the captured electron into the conduction band, thereby preventing the
completion o f ths recombination process. ( A similar argument holds
for centers near the valence-band edge.)
This and other important formulas relating to semiconductors under

non-equilibrium conditions are summarized in Table 5.1 at the end o f
this chapter.
N o t e that the lifetime is independent of the concentration o f electrons.
This is because there is a great abundance of electrons in an n-type semiconductor. Thus, as soon as a hole is captured by a center, an electron
w i l l immediately be captured by the same center and the recombination
process thereby completed. I n other words, the raie-limiiing step in the
recombination process is the capture of the minority
carrier.
E
Thus a recombination center is most effective i f the two emission probabilities are about the same, i.e.. when its energy level is near the middle o f
the energy gap. In the next chapter we w i l l s e e that such_a center is also
most effective in general ingârr]tr_s.
We can summarize this argument in
simple terms by recalling what we have said earlier: that the centers
provide "stepping stones" between the conduction and valence bands. F o r
a stepping stone to be most effective it should halve the total distance
between the two bands.
LIFETIME
IN
F i g . 5.13
T h e c h a r g e - s t a t e of m i d g a p
equilibrium.
I n the various cases involving recombination o f injected excess carriers

i n an n-type semiconductor, we have assumed that the net recombination
rate per u n i t volume is given by the relationship U = (p , p ,o)' v
can now apply Equation 5.40 to an n-type semiconductor in low-level injection. Under these conditions. n p .
Furthermore. n
n^ '~ *
forcenters that are efficient recombination centers, i.e., that areôT'boo
near the conduction-band edge. Thus U can be approximated by
r
ri
B,il
V =
= o v h [p
v
t!l
p].
(D.42)
Accordingly the lifetime o f holes i n low-level injection in an n-type semiconductor is

V - r - t - .
av h
v
ih
/(,! = 0
<ai
11-Type semiconductor
LOW-LEVEL INJECTION
/<,i = 1
Let us n o w consider special forms o f Equations 5.39 and 5.40 applicable

to the specific cases we have studied in the earlier part of this chapter.
-5.3
(5.43)
135
fbl
f>-Type semiconductor
centers
in n- and p - t y p e s e m i c o n d u c t o r s ,
in
This can also be seen in another way. I f the recombination rate w i l l

be limited by the rate o f capture o f the injected holes, we w o u l d expect V
to equal the rale o f process (c). However, for the rate o f process (c) to
lead to Equation 5 . 4 2 , / m u s t approximately equal 1. This means that in
an n-type semiconductor in low-level injection most o f the centers w i l l be
occupied by electrons. ( O f course, this is w h a i we would expect in an
w-type semiconductor in e q u i l i b r i u m for centers near the middle of the
forbidden gap. since then E > , as illustrated in Figure 5.13a.) Thus,
on the average, the centers are occupied by electrons, and are waiting to
capture holes. When one does capture a hole, it immediately captures an
electron again and is occupied by an electron for a longer period o f time
until it again captures a hole, and so on.
F
A similar argument for a />-type semiconductor leads to the rate of

recombination o f excess m i n o r i t y carriers in low-level injection as
V = ^ ' / I A I K ~ ">J-
(5-44)
fk
136
Semiconductors
under
Surface
'
semiconductor is given by
-
{ 5
4 5
5.4
SURFACE
>
y .
layer o f thickness x near the surface o f the semiconductor sample, as

illustrated in Figure 5.14. I n such a case we expect that the recombination
rate U w i l l be enhanced near the surface vvith the result that the excess
carrier density w i l l be smaller there.
ferences i n excess carrier concentration i n this surface layer and in the

rest o f the semiconductor, carriers w i l l diffuse from the body o f the semi-
<(
> '
"
tirhe is given by
" (
U ~ W.V,(xilp.m - Pno\
t ^ * )
-PL
\ Y-
;
In
la)
(5.46)
\s
where />(0) designates the average m i n o r i t y carrier concentration in the
'
iv^
> )
*
'"5^>
,\
surface layer, and iV * is the concentration (per cubic centimeter) o f centers

(
there.
SJnce the flux o f ' m i n o r i t y carriers to this region must equal 6',..
lb)
Fig. S.14 Enhanced recombination race near a surface due co a high center density, and
the resulting distribution of excess minority carriers. (No surface space-charge region.)
we arrive at the c o n d i t i o n
tTMiNfxMO)
p ].
no
(5.47)
x=0
N o t e that the f o r m o f this c o n d i t i o n is identical to Equation 5.16:

IV
OX 'i=fl
By c o m p a r i n g Equations 5.47 and;'5.16 we see that the surface
recom-
(5.48)
gfk
0
<2/A->
'
I n the case o f real surfaces, an additional complication may enter into

the description o f the surface recombination process. If, for example,
negatively charged ions are present on the surface o f the -type semiconductor sample, as shown in Fig. 5.15. the carrier distribution w i t h i n
the sample w i l l be disturbed by these ions: .electrons w i l l J 3 e j ^ h j ^ d _ a w a y
f r o m the surface and holes will be attracted^ there. Consequently, spacecharge neutrality w i l l ^ S ) longer hold in a region o f thickness x near the
surface. This region V c a l l e d a surface space-charge region.i
4
b i n a t i o n velocity o f holes s is given by

1'
"r>*t"
I apro y
ooj
C^v
'rnorric/
I n an attempt to even out the dif-
conductor.
The fluxes o f carriers that flow to this region of enhanced recombination
rate supply all the carriers that recombine in this region. T h e total
number o f carriers recombining i n the surface layer per u n i t area and unit
en sjfc.- "+'Y >^
*i
'
RECOMBINATION
So far we have considered only recombination at centers distributed

u n i f o r m l y w i t h i n the semiconductor. We now consider what happens i f
we have an increased concentmUon__of recombination centers i n a thin
Thus it is evident that surface recombination can be considered as a

special case o f bulk recombination for a high density o f centers distributed
w i t h i n a very thin region near a surface.
C o m p a r i s o n o f Equation 5.44 w i t h the rate o f electron capture, process

(a), shows that the probability o f occupation o f a near-midgap center by
an electron i n a />-type semiconductor is approximately zero, just as we
w o u l d expect for a-/?-type semiconductor in e q u i l i b r i u m (see Figure 5.13b).
*
V
^
137
we replaced (iV^xj
with J V , , , the total number o f centers (per unit
surface area) within the boundary r e g i o n
'.
/i
Recombination
Conditions
w
Thus the lifetime ot electrons i n a p-i\pe
... ^
Non-Equilibrium
f/tjeVrvj
Surface space-charge regions are discussed in detail in Chapters 9 and 10.
Semiconductors
138
under
Non-Equilibrium
Conditions
Surface Recombination
139
are located at energy , = , (such centers w i l l , in fact, be the most

effective), and that the capture cross sections are equal, i.e., a, = a = a,
U becomes
/ ,~
n
V, =
P,,
ov N
lft
(5.50)
n, + V, + 2n,where s s av N
is the surface recombination
velocity of a surface
without a surface space-charge region, as given by E q u a t i o n 5.48.
T h e flux o f m i n o r i t y carriers reaching the surface must equal L\. I f
the recombination rate w i t h i n the surface space-charge region is n o t too
high, this flux can be approximated by the flux o f m i n o r i t y carriers
reaching the edge o f the surface space-charge region. Thus,
D
th
s1
Pn
= s
ox
(5.51)
, + p + In,
s
The above relationship is really a boundary c o n d i t i o n for the diffusion

p r o b l e m w h i c h describes the d i s t r i b u t i o n o f m i n o r i t y carriers w i t h i n the
b u l k o f the semiconductor sample. A boundary c o n d i t i o n is useful only i f
it is expressed i n terms o f concentrations and concentration gradients at the
boundary, i.e., at the plane x x i n this case. T h u s we must express the
right-hand side o f Equation 5.51 i n terms o f concentrations at the edge o f
the surface space-charge region rather than at the surface.
A',
-N*
(bl
Fig. 5.15 Enhanced recombination rate near a surface due to a high center density, and
the resulting distribution of excess minority carriers. (Surface space-charge region
induced by negative ions.)
I f we now shine light on the sample and generate electron-hole pairs

uniformly throughout its interior, the carriers will move to the surface to
recombine there as in the previous case. However, because of the presence,
of _the_ surface space-charge regjon,_aidditional care must be taken i n
establishing the.balanc_between the flux o f minority carriers, to the
swface and the surface recombination r a t e ^
The total number o f carriers recombining at the surface per unit area
and unit time will be
a-a-v.N [p,n,
n.e
] + Cp[P
ps"s = Px )n (x )
d
no
(5.52)
p (x )N
n
sample.
x=i
n, + p + 2/i,
[Pn( d)
~ Poi
(5.53)
or
(5.49)
by analogy to Equation"5.40, where n, and p denote the electron and.hole.

concentrations at the surface.- I f for simplicity we assume that the centers
=5=
D,
(E,-E,)/kT
where N
is the donor concentration in the semiconductor
N o t i n g that p N
= n,-, we obtain
n,*]
(,-E,)/):T
W e can accomplish this by assuming that the p r o d u c t o f electron and

hole concentrations will be constant throughout the surface space-charge
region even t h o u g h equilibriurrTdoes not p r e v a i l T h i s constam^jwniôf
^?Hi îi3I 2i.i L2A Such an assumption is consistently employed
in the treatment o f space-charge regions, and w i l l be discussed in the next I
chapter. I n the present case it leads to
"""^
dp
X
s[Pr,( d) ~ Pno]
ox
(5.54)
where
(5.55)
ss
Pi
2n,
Semiconductors
under
Non-Equilibrium
Conditions
O r i g i n of R e c o m b i n a t i o n - G e n e r a t i o n
Centers
141
140
is the surface recombination velocity. T h u s the surface recombination

velocity w i l l take o n a more complicated f o r m when a surface space-charge
region is present; i t w i l l depend not o n l y o n the density o f surface recombination centers b u t also o n the d o n o r concentration and on the amount
of surface charge which determines n a n d p . A s the charge o n the surface
is. varied, n, and p and consequently the surface recombination velocity
itself, w i l l vary.
The surface recombination velocity w i l l go through a m a x i m u m when
(/J, -+- p ) is at a m i n i m u m . T h i s w i l l happen when they are b o t h relatively
close to the intrinsic carrier concentration n
The m a x i m u m surface
recombination velocity w i l l then be given by
s
(5.56)
Smax
5.5
ORIGIN O F RECOMBINATIONGENERATION CENTERS
We have now seen h o w energy levels introduced into the forbidden gap
can facilitate therrecambmarion o f excess carriers w i t h i n a semiconductor
by acting as ''stepping stones'' between conduction band and valence band.
We have also shown:, that the theory o f the recombination-generation
process relatestheicharacteristic constants o f the recombination rate to
the density of recombination-generation centers. Thus we have seen that
the lifetime in low-level injection is given by
1
a.
Impurities
When introduced into a silicon sample, column I I I and c o l u m n V

impurities lead to energy levels w i t h i n the forbidden gap. Because these
elements are relatively similar to silicon (a column I V element), the energy
levels associated w i t h them w i l l be shallow, i.e., they w i l l be close to the
valence- and conduction-band edges. Thus column I I I and c o l u m n V
impurities become acceptors and donors in silicon.
There are other impurities which, in contrast with column I I I and V
elements, introduce energy levels nearer to the middle of the forbidden
gap. Such impurities, examples o f which are copper i n germanium and
gold in silicon, will therefore act as efficient recombination-generation
centers.
The properties o f gold in silicon have been studied extensively. I t is
known that there are actually two energy levels associated with each g o l d
a t o m : an acceptor level (which can be either neutral or negatively charged)
near the middle of the energy gap, and a donor level (which can be either
neutral or positively charged) about 0.2 ev below the middle. The theory
of the recombination mechanism taking place through the action o f such
multi-level centers is more complicated than the theory o f a single-level
center given i n this chapter; however, in low-level injection we may apply
the latter to describe gold in silicon as a reasonable approximation by
considering only the acceptor level in n-type, and the donor level in p-type
silicon.
3
The lifetime of silicon samples can be varied by the controlled a d d i t i o n

of gold. This is done by depositing gold onto the surface o f the sample and
then heating the sample at a certain temperature. After a period o f time t
such that t Wf-'\D, where W is the thickness of the sample and D is
the gold diffusivity, gold w i l l be uniformly distributed in the sample.in a
concentration corresponding to its solid solubility at that temperature.
(The diffusivity and solid solubility o f gold are shown as a function o f
temperature i n Figures 3.4 and 3.7, respectively.)
The experimentally observed relationship between gold concentration
and lifetime i n silicon is shown in Figure 5.16. I t is evident that the data
follow the inverse relationship indicated by Equation 5.57. The "constant
of proportionality corresponds to a capture cross section o f ~ 5 X 10~
cm . Several other investigators have reported values within an order o f
magnitude o f this one. Also shown in this figure are the respective
temperatures at which the silicon sample was saturated with gold.
As the gold concentration in the silicon sample is increased sufficiently
to become comparable to the concentration o f the donor or acceptor
impurity, another effect has to be taken into account. Each o f the deep
s
for the example o f an /i-type semiconductor, where N is the concentration

o f the centers per u n i t volume, and a, is their capture cross section f o r
holes. Similarly, by extending the theory to recombination taking place
at a surface, we have seen that the surface recombination velocity i n the
absence o f a surface space-charge region is given by
t
1S
for an /i-type semiconductor, where N,
is the density o f the centers per
unit surface area.

I n this section we n o w briefly consider some examples o f the physical
o r i g i n o f recombination-generation centers, both in the bulk and at the
surface.
142
l y i n g energy levels associated w i t h the gold atoms w i l l , in effect, remove

one "majority earner from the conduction-band in the case o f an n-type
semiconductor, or from the valence band i n the case o f a /"-type semiconductor. This effect is sometimes referred to as carrier removal. Thus.
Gold saturation temperature ("Cl
10"
o o c o o
o o o
o in o io c
o m o
m o o ,- .- r\j
co co cx> o-.
I
O r i g i n of Recombination-Generation C e n t e r s
b.
143
Radiation D a m a g e
A n o t h e r manner in which energy levels can be introduced into the

forbidden gap is by exposure to high-energy radiationelectrons, protons,
gamma rays, or neutrons. The high-energy particles can displace atoms
from their n o r m a l positions in the semiconductor lattice, resulting first
in the f o r m a t i o n o f a vacancy and an interstitial a t o m . These, in t u r n ,
1 0
" pn
rrr\
mr|iTTT|i
I I I
.i
rqirrr
1er
10"
1
r
o t
r
(7.
lt
with
A',
5 x 10"
ir.
15
y 10'-
enr
10"' -
10-eI
2
10'-
I I I
10
13
1 I I
Eleclron bombardmem
Fig. 5.17
silicon.
10
10*'
15
I I I
10'
I I I
10
17
A,. ( c m ~ l
Effect of high energy electron irradiation on the lifetime of holes in n-type
10
15
10"
TO
17
- 3
N, ( c m '
Fig. 5.16
Relationship between lifetime of holes and gold concentration in n-type
silicon.*
for example, in an n-type semiconductor the electron concentration w i l l

become naiN
N,. As a result, the resitivity of the sample w i l l
increase with the addition o f gold. A similar phenomenon takes place i n
the case o f a p-xype semiconductor/!"
D
t This simple description loses its validity once the concentration of deep-lying levels
approaches the donor or acceptor concentration. Exact calculations of the resistivity
of silicon as a function of gold concentration were oerformed by Boltaks el a l . Their
results are reproduced also in Bullis's paper.
w i l l rapidly f o r m more complex lattice defects which behave much like

impurities introduced into the semiconductor; they w i l l have energy levels
within the forbidden gap and can act as acceptors, donors, and recombination centers.
The i n t r o d u c t i o n o f recombination centers by high-energy electron
i r r a d i a t i o n is illustrated by the experimental data shown i n Figure 5.17.
Here the lifetime o f holes i n an n-type silicon sample is shown as a function
of the total number o f bombarding electrons h i t t i n g a unit area o f the surface o f the sample, N . I t is evident that as b o m b a r d m e n t proceeds, the
lifetime begins to decrease. I f we assume that the incident electrons create
recombination centers uniformly w i t h i n the semiconductor, the total
concentration o f recombination centers A', w i l l be given by
6
N, = N
t0
+ KN
(5.59)
144
Semiconductors
under
Non-Equilibrium
Conditions
Origin
of
Recombination-Generation
Centers
I4S
where N is the concentration o f recombination-generation centers before

irradiation, and K is the probability that an incident electron w i l l create a
new recombination center. M a k i n g use o f Equation 5.57, we then obtain
t0
1 2 2
r =
1 +
(5.60)
KNJN
t0
which predicts that the lifetime should decrease from its initial value r in
an inversely p r o p o r t i o n a l manner w i t h increasing radiation dose. This
relationship is borne o u t by the experimental observations shown in
Figure 5.17.
The radiation-induced centers w i l l bring about a decrease in the majority
carrier concentration (carrier removal) much the same way as gold does
in silicon. Thus the resistivity o f semiconductors will increase w i t h
increasing r a d i a t i o n dose.
A t the present time irradiation cannot be readily employed to c o n t r o l
the lifetime o f semiconductor devices as is possible w i t h impurities like
gold. This is because the radiation-induced centers have a tendency to
anneal out after a p e r i o d o f time, even at a relatively l o w temperature.
However, the influence o f radiation on the lifetime and resistivity o f
semiconductors is i m p o r t a n t because many types o f semiconductor devices
are intended for operation in nuclear reactor or space environments.
p 0
c.
Surface S t a t e s
The cause o f such a reduction is u n k n o w n at present; it is evidently

brought about by the presence o f even a very thin oxide layer such as forms
on both germanium and silicon surfaces upon exposure to r o o m air.
Surface recombination velocity s values observed on germanium and
silicon surfaces are o f the order o f 10 cm/sec; on thermally oxidized
silicon, 1 to 10 cm/sec.f U p o n irradiation, the magnitude of s as well as
the density o f surface states has been found to increase. Figure 5.18 shows
a comparison between measured values o f 5, and independently estimated
values o f the surface state density N o f thermally oxidized silicon samples.
T h e straight-line relationship observed is in agreement with Equation 5.58,
w i t h the slope corresponding to a capture cross section o f the same order
as that o f bulk recombination-generation centers.
. .
n
We have seen that a foreign element or even a lattice defect w i t h i n the

semiconductor crystal can introduce energy levels into the forbidden gap.
I t might be expected that the drastic irregularity associated w i t h a surface
of the crystal, where the semiconductor lattice is altogether terminated,
should also result in the i n t r o d u c t i o n o f a large density o f levels into the
forbidden gap. Such levels, the existence o f which has been predicted
theoretically by T a m m and by Shockley, are called surface states. I f
some o f these surface states should have energy levels near the middle o f
the forbidden gap, we w o u l d expect them to act as efficient surface
recombination centers.
7
Theoretical estimates o f the density o f surface states yield values o f the

same order as the density o f surface atoms, ~ l O c m . Such densities
have indeed been observed on very clean semiconductor surfaces obtained
by cleaving samples under high vacuum. However, germanium and
silicon samples after exposure to air for only a few minutes show surface
state densities o f the order o f only ~ l 0 c m ~ , and thermally oxidized
silicon surfaces can show densities yet another order o f magnitude smaller.f
l 5
1 1
_ i
READING
A rigorous and thorough treatment of the transport of excess carriers is given by

A . K . Jonscher, Principles of Semiconductor Device Operation, Wiley, 1960, Chapters
2 and 3.
f Actual measured values of the surface recombination velocity s will not necessarily be
the same as s, but will be related to it by Equation 5.55. Thus, depending on the particular semiconductor and the temperature (through n<), on the impurity concentration
in the semiconductor, ; V , and on the surface charge density (through n, + p,), s may
be smaller, larger, or the same as s,.
0
* Surface states are. discussed in further detail in Chapters 9 to 12.
REFERENCES
146
Problems
(a) The number of holes recombining at the surface, per unit surface area, in unit
time, and
fb) The number of holes recombining in the bulk within 3-difTusion lengths of the
surface, per unit surface area, in unit time.
Band-to-band recombination and recombination through multiple-level centers are

discussed in further detail in Chapter 6, J . L . Moll. Physics of Semiconductors, McGrawHill Book C o . . 1964.
A review of and extensive bibliography on "Recombination in Semiconductors" is
given by G . Bemski. Proc. IRE. 46,"990 (] 958).
147
5.4
(a) Estimate the electric field near the surface for the above problem.
(b) Estimate the percentage contribution of the drift-term to the flux of holes to
the surface.
5.5
A wafer of silicon doped with 2 x 10" donor atoms/cm has been saturated with
gold at 920"C. It then was oxidized and treated in such a way that it now has ] 0
surface recombination centers/cm .
(a) Calculate the bulk lifetime the diffusion length, and the surface recombination
velocity in the absence of a surface space-charge region, and also the maximum
surface recombination velocity.
(b) If the sample is illuminated with uniformly absorbed light which creates ) 0
carrier pairs/(cm sec), what will the hole concentration at the surface and the
hole flux to the surface be in the absence of a surface space-charge region?
Surface recombination is discussed in further detail by A . Many, Y . Goldstein, and

N . B . Grover, Semiconductor Surfaces, Wiley, 1965, Chapters 5, 7. and 9.
1C
REFERENCES
CITED
5. R i N. Hall. "Electron-Hole Recombination in Germanium," Phys. Rev., 8", 387

-{1952); W. Shocklev and W. T . Read, "Statistics of the Recombination of Holes
and Electrons," Phys. Rev., 87, 835 (1952).
17
2. D . T . Stevenson and R . J . Keyes, "Measurements of the Recombination Velocity

at Germanium Surfaces," Physica, 20, 104] (1954).
5.6
3. A review of the "Properties of Gold in Silicon" is given by W. M . Bullis, Solid-State

Electron., 9, 143 (1966).
What is the electron concentration and the resistivity of the sample in Problem 5.5
before and after gold diffusion?
5.7
Derive the expression for the recombination-generation rate, and indicate if net
recombination or generation of carriers lakes place, under the following conditions:
(a) In a region of a semiconducior sample which is completely depleied of carriers
(i.e., both n and p n,).
(b) In a region of a semiconductor sample which is depleted of minority carriers
only (for example.p p , . but n , = n ) .
(c) In a region of a semiconducior in which n = p = , where n
(These formulas will be used in Chapter 6 in the derivation of the curreni-voliage
characteristics of p-n junctions.)
4. A . E . Bakanowski and J . H . Forster. "Electrical Properties of Gold-Doped Diffused

Silicon Computer Diodes." Bell System Tech. J., 39, 7 (1960).
5. B . I. Boltaks, G . S. Kulikov. and R. Sh. Malkovich, "The Effect of Gold on the
Electrical Properties of Silicon," Soviet PhysicsSolid State, 2, 167 (1960).
n (
6. G . K . Wertheim, "Energy Levels in Electron Bombarded Silicon," Phys. Rev.. 105,

1730 (1957).
7. See. for instance, A. Many, Y . Goldstein, and N . B . Grover, Semiconductor Surfaces,
Wiley. 1965, Chapters 1 and 5.
5.8
In a region of semiconductor which is completely depleted of carriers (i.e., n and

p n,) electron-hole pairs are generated by alternate emission of electrons and of
holes by the centers. Derive the average lime that lakes place between such
emission processes. (This time is called the emission time constant of the centers.)
Estimate its value for centers with . = , . in silicon.
5.9
Considering a p-iype semiconducior, derive the energy level of those centers from
which a trapped electron is as likely to be re-emitted into the conduction band as it
is to recombine with holes. Will centers with energy above or below this level be
efficient recombination centers?
8. D . J . Fitzgerald and A . S. Grove, "Radiation-Induced Increase in Surface Recombination Velocity of Thermally Oxidized Silicon -Structures." Proc. IEEE, 54, 1601
(1966).
PROBLEMS
5.1
Calculate the resistivity of an n-type silicon sample under illumination. The sample
has a lifetime of 1 ftsec and its resistivity in dark is 10 fi cm. The light is absorbed
uniformly in the semiconductor, leading to the creation of 10" electron-hole
pairs/(cm sec). What part of the conductivity is due to:
(a) Majority carriers?
(b) Minority carriers?
3
5.2
Verify that Equation 5.17 satisfies the appropriate differential equation and
boundary conditions.
5.3
A sample of 1 O.cm n-lype silicon is illuminated. The uniformly absorbed light

creates 10 electron-hole pairs /(cm sec). The lifetime is 10 /isec; the surface
recombination velocity 100 cm/sec. Calculate:
17
5.10 Consider recombination-generation centers uniformly distributed in energy

throughout the forbidden gap. with density D, ( c m ev" ).
(a) By superposition of the effect of such centers, and assuming o = c = c,
derive the formula for the net recombination-generation rate U.
(b) Derive the expression for the generation rate in a region which is completely
depleted of carriers (i.e., n and p n,).
(c) Derive the expression for the lifetime in low-level injection.
Compare each of the above formulas with ihe corresponding ones for single
level centers located at , = , , and discuss the similarities and differences.
_ !

148
ELECTROSTATICS
SPACE-CHARGE
STEP
REGION
FOR:
JUNCTIONS
LINEARLY GRADED JU N C T I O N S
DIFFUSED
JUNCTIONS
CAPACITANCE-VOLTAGE
CHARACTERISTICS
CURRENT-VOLTAGE
CHARACTERISTICS
TABLE
IMPORTANT
UNDER
FORMULAS
FOR S E M I C O N D U C T O R S
CONDITIONS
Midgap recombination-generation centers, i.e., E
E q u a l capture cross-sections, i.e., o, - a
Net bulk recombination

rate per unit volume
Net surface recombina-
- T y p e semiconductor 1
1
U =-(p
= [p (0)
= s[n,(0)
v0
'hoi
av Nt
Lifetime
velocity
tll
S
S
" n
+ p
~ n, + p, + 2*i
+ 2<
J,
BEHAVIOR
,
n)
tion rate per unit area
Surface recombination
BREAKDOWN
/)-Tvpe semiconductor
po\
TRANSIENT
p-n Junctions
1
(J = - (n
.
- P-no)
J U N C T I O N
S.l
NON-EQUILIBRIUM
av N
tk
st
I n the preceding t w o chapters we have discussed the behavior o f h o m o geneous semiconductor samples b o t h under e q u i l i b r i u m and non-equilibr i u m conditions. M o s t semiconductor devices, however, incorporate b o t h
p- and /j-type regions, and i t is the junction between these regions that leads
to their useful electrical characteristics.
A planar silicon n~p junction diodej is illustrated schematically in
Figure 6.1a. Such a diode may be fabricated by first growing a relatively
lowly doped />-type epitaxial film upon a highly doped substrate o f the
same type; and then thermally oxidizing the surface o f the silicon.
Windows are then opened i n the silicon dioxide layer, and donor impuritiesare permitted to diffuse i n t o the silicon where the oxide layer had been
removed.{ This results i n the formation o f the n p j u n c t i o n illustrated i n
the figure. (The superscripts + or indicate regions o f relatively high
or l o w impurity concentrations, respectively.) M a n y such junctions are
formed simultaneously o n a wafer. After they are cut apart by scribing,
contacts are attached to b o t h the b o t t o m side and to the diffused p o r t i o n
o f each diode, and the diodes are inserted into a package.
+
f The term junction denotes the boundary between the two semiconductor regions,
whereas the term diode refers to the finished semiconductor device incorporating a
single junction. These terms sometimes are used interchangeably,
t Epitaxial growth, thermal oxidation, and solid-state diffusion are discussed in detail
in P a n I.
149
-p-n
ISO
Junctions
p-n
Junctions
151
as rectifiers, or as switches i n digital computers. However, the importance

of p-n j u n c t i o n s is broader than the use o f diodes alone w o u l d indicate.
T w o o f the most i m p o r t a n t semiconductor devices, j u n c t i o n transistors
and j u n c t i o n field-effect transistors (which f o r m the subject o f the next
Epitaxial film
( - 10;<)
(C,
Substrate
( - 100 tt)
u b
~ 1 0 " cm"*)
C -
10'
cm"
-10
8
(a) Actual structure
R
D
2
%
+
A typical planar n p junction diode.
Representative impurity concentrations
and dimensions are indicated.
B j H i EU m H M
H i a BH B B B
H
H
EH M B B
m
m m
H
4
6
For simplicity, we study an idealized one-dimensional model o f this

ti p j u n c t i o n , which is illustrated in Figure 6.1b. This model can be
considered as a section o f the planar diode along the dotted lines shown i n
Figure 6.1a. This idealized model also neglects the variation o f the
impurity concentration i n the />region due to the epitaxial construction.
Where this concentration v a r i a t i o n or the characteristic shape o f the
planar j u n c t i o n leads to deviations, we modify our discussion accordingly.
+
The most i m p o r t a n t characteristic o f p-n junctions is that they rectify,

i.e., they permit the passage o f electric current in only one direction. This
is illustrated in Figure 6.2 where we show the current-voltage characteristic
of a typical silicon p-n j u n c t i o n diode whose structural parameters are
approximately as indicated i n Figure 6.1. I t is evident that when a
-negative voltage is applied to the n-region, a current begins to flow at a
"very small applied voltage. I n contrast, when a positive voltage is applied
to the n-region. no current flow is perceived at the scale of this illustration.
Eventually, when a large enough positive voltage is applied to the n-region,
current does begin to flow again. This c o n d i t i o n is referred to as junction
breakdown.
The bias c o n d i t i o n corresponding to easy conduction is
called forward bias, and the bias c o n d i t i o n corresponding to no conduction
is called reverse bias. The current-voltage characteristic under forward
bias is also shown w i t h an expanded voltage scale in the lower half o f
^Figure 6.2.
Diodes themselves are very i m p o r t a n t and useful because o f their
special current-voltage characteristics; for instance, they can be employed
2 H
(b) Idealized one-dimensional model

Fig. 6.1
a B B U mmB B
1 mmB flll mmB B
mB B B E I mmmB
m B B B mB B
B B B f l B mB B
: -
8
10
BBBB
M
BEi t m
m
BU
i l ES H i
B B B
H B B Bfl B B m.
B B B B E E B B
80
40
M)
30
Vjtv)
HflB
Hhbb
1 M B
Ii Im
BPW
II
Bl B
H B
B
B
B
mm
flT^*lÊB - 0 . 4
ESC
Vj (v)
Fig. 6.2 The current-voltage characteristics of a silicon n*p diode.
parameters of this diode are approximately as indicated in Fig. 6.1.
The structural
two chapters), consist o f t w o p-n junctions in the vicinity o f each other.

Their characteristics w i l l be seen to be a direct consequence o f the characteristics o f single p-n j u n c t i o n s . I n addition, in Part I I I we w i l l see that the
analysis o f surface phenomena and o f surface-controlled devices can be
greatly simplified by relating i t to the analysis o f p-n j u n c t i o n s .
p-n
152
junctions
Accordingly, in this chapter we consider p-n junctions i n detail. W e

begin w i t h a brief review o f the relevant principles o f electrostatics, and
then apply these principles to the study o f the space-charge regions o f
diffused j u n c t i o n s - a n d their two l i m i t i n g forms: step junctions and
linearly graded junctions. W e then consider the capacitance-voltage and
the current-voltage characteristics o f p-n junctions, and j u n c t i o n breakd o w n . Finally, we give a brief discussion o f transient phenomena.
Y Comparison o f E q u a t i o n 6.3 w i t h 6.1 yields
a.
(6.5)
Poisson's E q u a t i o n
Poisson's equation states that
ELECTROSTATICS.
d d>
Connection between Band Structure

and E l e c t r o s t a t i c Q u a n t i t i e s
(6.6)
dx
Kt
or
T h e electric field S is defined as the force acting on a u n i t positive charge.
d%
A c c o r d i n g l y , the force acting on an electron, w h i c h has a charge q, is
dx'
-qS.
I n general, we k n o w that a force is equal to the negative o f the gradient
o f potential energy. Accordingly, the force acting on an electron equals
qg=
which provides the relationship between the electrostatic potential and the
potential energy o f an electron.
b.
6.1
E,
<P= -
(gradient o f potential energy o f electrons).
31.
(6.7)
or
dS
p_
(6.8)
dx
where p is the charge density per u n i t volume, conveniently expressed i n

semiconductor w o r k i n units o f e/cm where e is the magnitude o f theelectronic charge, 1.60 x 10~ coulombs; K is the dielectric constant,
and is the permittivity o f free space, 8.86 x 10~ f/cm = SSAejvp. i n
units convenient i n semiconductor w o r k .
3
W e saw in Chapter 4 that the potential energy o f an electron is represented by the lower edge o f the c o n d u c t i o n band, E . However, since we
are interested only i n the gradient o f the potential energy, we can also use
any part o f the energy bands that is displaced from the conduction-band
edge by a fixed amount. Thus we c o u l d equally well choose the gradient
o f the valence-band edge E or o f the intrinsic Fermi level E i n the above
expression rather than the gradient o f E . I n practice i t is frequency c o n venient to use E,. Accordingly, we have the expression
c
g = - grad Ei
(6.1)
19
14
I t is evident from E q u a t i o n 6.8 that the electric field is obtained by

integrating the charge d i s t r i b u t i o n as a function o f distance. Simple
examples o f such integrations are illustrated i n Figure 6.3.
I n Figure 6.3a w e s h o w a charge d i s t r i b u t i o n given by a delta function
containing a total charge per unit cross-sectional area o f 0. Integration
o f this charge d i s t r i b u t i o n results i n a step change i n electric field by the
amount QlKe , as shown i n the figure. I n Figure 6.3b we show a u n i f o r m
charge distribution over the distance L . I n this case the electric field i n - creases linearly w i t h distance, b u t the t o t a l increase i n the electric field is
again related only to the total charge contained i n the region, p L .
0
or, i n one dimension,

(6.2)
q dx
T h e electrostatic potential <j> is the quantity whose gradientis the negafive o f the electric field, i.e., i t is defined by the equation
g =
grad./.
or, i n one dimension,

i
" "
dd>
(6.3)
6.2
SPACE-CHARGE REGION
FOR STEP J U N C T I O N S
M a n y i m p o r t a n t characteristics o f p-n junctions are associated w i t h a

space-charge region (a region where p ^ 0), formed between the n-type
and /'-type regions. Accordingly, we n o w consider what happens when a n
p-n Junctions
154
P
155-
Space-Charge Region for Step Junctions
the j u n c t i o n as indicated by the arrows in Figure 6.4b. This electric field

is o f such direction as to keep the holes j n the /^-region and the electrons
in the ??-region. Thus, after a certain number of electrons and holes have
flowed from one region to the other, an electric field will be built up, preventing further net flow of the carriers.
This is the manner i n which e q u i l i b r i u m is reached. T o restate, i n
e q u i l i b r i u m the net flux o f both holes and o f electrons w i l l be zero; the
(Many electrons)
( M a n y holes)
la)
Before contact
(a)
(b)
Metallurgical junction
Fig. 6.3 Illustration of the consequences of Poisson's equation for two simple onedimensional cases.
n-type semiconductor and a />-type semiconductor are brought i n t o i n t i mate contact. Bringing separate pieces o f semiconductor into intimate
contact is not a practical way o f m a k i n g useful p-n j u n c t i o n devices b u t
we employ this scheme to illustrate the principles involved.
a.
(b)
After contact
Fig. 6.4 Schematic illustration of a p-n junction in equilibrium.
(a) n- and p-type semiconductors.
(b) Junction between the two.
Equilibrium Case
First we consider the case w i t h no bias applied to either semiconductor

region. The n-type and./>-type samples before contact are illustrated i n
Figure 6.4a. As we discussed i n Chapter 4, an n-type semiconductor is
characterized primarily by the fact t h a t i t contains a large concentration o f
electrons. Similarly, a />-type semiconductor sample is characterized by
t h e fact that i t , in t u r n , contains a large concentration o f holes. I f these
two samples are brought i n t o i n t i m a t e contact, as illustrated in Figure
6.4b, a flux o f electrons and o f holes w i l l flow i n such directions as to even
out the large concentration gradients existing between the t w o regions.
I f the electrons and holes were n o t charged, these fluxes w o u l d continue
u n t i l a uniform concentration o f b o t h species was established throughout
the combined semiconductor sample. Because they are charged, and because the semiconductor samples also c o n t a i n ionized i m p u r i t y ions, the
situation will be entirely different, as illustrated in Figure 6.4b. W h e n a net
;fiow o f electrons from the n-region to the />-region. and o f holes from the
p-region to the n-region takes place, a space charge due to donor and
acceptor ions is formed. Hence an electric field results in the vicinity o f
diffusion flux of each carrier at the p-n junction will be precisely equal and
opposite to the flux of that carrier due to the electric field.
Let us n o w use this criterion to obtain the c o n d i t i o n for e q u i l i b r i u m i n
quantitative terms. The net flux o f holes F is given by
v
(6.9)
dx
Substituting the expression for the hole concentration,
p =
lE
)lkT
n e >- r
(6.10)
and its derivative,

dp
-dE,
dx
dE
kT\-dx
and i n c o r p o r a t i n g Einstein's relationship p

holes is then given by
D.
dE*-
(6.11)
dx -
= qDJkT,
dE
the net flux o f
p-n Junctions
156-
T h u s the c o n d i t i o n o f zero net hole flux means that the F e r m i level must
be u n i f o r m as we pass f r o m the - r g i o n to the ^-region.
Similar con-
siderations applied to the electron flux lead to the f o r m u l a

I
F
kT
dE
(6.13)
dx
dx
Space-Charge Region for Step
-qN
Let us now consider what determines the total amount o f bending o f the
bands, i.e., the total amount o f variation o f either electron energy or
electrostatic potential, as we cross f r o m the /7-region to the ^-region. I t is
evident from Figure 6.5c that, since the Fermi level is constant, this variat i o n w i l l consist o f the sum o f the absolute values o f the two Fermi potentials 
-JE
JC
?Fv
T'-T-
Fig 6.5 Schematic illustration of the (a) charge, (b) electric field, and (c) potential
distribution within a step junction in equilibrium (depletion approxunacon).
T h u s in equilibrium
_
=
Et
the Fermi level must be constant
through-
9Fn
Ip-region
EF_
Et
(6.14)
Thus 6
is positive while <f> is negative. (Hence the need to take the
absolute value o f the latter.)
T o calculate the Fermi potentials, we consider that i n the neutral pregion (i.e., outside the space-charge region) p == N , while i n the neutral
rc-region n = N .
Then, f r o m Equation 6.10 we get
Fn
<p , = In ^
according to w h i c h the condition o f zero net electron flux leads to the same
out the
Fn
F p
(c)
requirement.
(a)
-v.
- < o
- -qi*A
157
small i n comparison to the i m p u r i t y concentration over most o f the spacecharge region. This is called the depletion
approximation.!
The charge d i s t r i b u t i o n in the semiconductor sample as a function o f
distance, subject to this simplifying assumption, is illustrated i n Figure
6.5a for the case o f u n i f o r m i m p u r i t y concentration in both p and n
regions. Such a j u n c t i o n is called an abrupt or step junction. The density
of space charge o n the n and p sides o f the metallurgical j u n c t i o n w i l l be
given by qN
and by qN , respectively. The widths o f the respective
portions o f the space-charge region are designated by x and z , w i t h the
t o t a l w i d t h (x + x ) being denoted by W. The field distribution, obtained
by integration o f the charge distribution, is illustrated in Figure 6.5b.
Integration o f the electric field distribution i n t u r n leads to the electron
energy E as a function o f distance as illustrated in Figure 6.5c. The cond u c t i o n - and valence-band edges, o f course, follow the variation o f the
intrinsic Fermi level.
D
\v
Junctions
(6T5)
Likewise,
semiconductor.*
T o calculate the characteristics o f the space-charge region formed near

the j u n c t i o n , we assume that i t consists entirely o f the charge o f the ionized
acceptors and d o n o r s . T h i s assumption is equivalent to saying that m o s t
o f the space-charge region w i l l be completely depleted o f carriers or, i n
other words, t h a t the carrier concentrations n and p w i l l be negligibly
t As mentioned in Chapter 4, the Fermi level can be considered as the chemical potential
of electrons. Since the condition for equilibrium in any system is that the chemical
potential should be constant as we pass from one part of the system to the other, the
constancy of the Fermi level in equilibrium follows.
l
q
A
n
(6.16)
The total potential variation <p i n e q u i l i b r i u m then will be the sum o f

T
the absolute values o f the Fermi potentials,
This quantity is often referred to as the built-in voltage i
o f a p-n j u n c t i o n .
t In this context, the terms space-charge region and depletion region are used interchangeably.
I5g.
p-n Junctions
The magnitude o f the Fermi potential o n either side o f a j u n c t i o n is

given as a function o f the magnitude o f the net i m p u r i t y concentration
C = Wo N \ i n Figure 6.6 for silicon at r o o m temperature, based on
B
S p a c e - C h a r g e Region for Step
Junctions
side o f the j u n c t i o n to the other
159
6 .
T
Equations 6.15 or 6.16. This enables the calculation o f <p for any given
step j u n c t i o n .
Because the electric field i n the neutral regions o f the semiconductor
must be zero, we can immediately see that the t o t a l charge per unit area
W=
/'-^
(6.21)
A p a r t i c u l a r l y i m p o r t a n t case o f step junctions w h i c h is often encountered i n practice is a step j u n c t i o n where the concentration o f the i m p u r i t y
on one side o f the j u n c t i o n is m u c h larger than that o n the other side o f the
j u n c t i o n , f o r example, N N .
This results i n the simpler f o r m o f
E q u a t i o n 6.21,
D
qNj
, A n alloy j u n c t i o n and a very " s h a l l o w " diffused j u n c t i o n are b o t h very

m u c h like one-sided step, junctions.
(We put the w o r d " s h a l l o w " i n q u o t a t i o n marks because what is " s h a l l o w " and w h a t is "deep" depends o n a
number o f factors as we shall see later.) The concentration d i s t r i b u t i o n f o r
an example o f a shallow diffused n~p j u n c t i o n is shown i n Figure 6.7a.
The calculations shown i n this figure are for a-complementary error-funct i o n type i m p u r i t y d i s t r i b u t i o n w i t h a relatively h i g h surface c o n c e n t r a t i o n ,
C 1 0 c m : l o w b u l k concentration, C = 1 0 c m , and a relatively
shallow j u n c t i o n depth o f x = 1 u.
20
- 5
1 4
- 3
Fig. 6.6 Fermi potential versus net impurity concentration (Silicon, 300K). <j> > 0 for
p-type region; $ < 0 for n-type region.
_
The charge d i s t r i b u t i o n due to such a concentration d i s t r i b u t i o n as w e l l

as the corresponding one-sided step-junction a p p r o x i m a t i o n are s h o w n i n
Figure 6.7.b, b o t h assuming complete depletion o f carriers. T h e corresponding electric field and potential distributions are shown i n F i g u r e 6.7c
and d.
o n either side o f the p-n j u n c t i o n must be precisely equal and opposite.
These distributions were calculated as follows. First, Poissons e q u a t i o n

was integrated to obtain the electric field d i s t r i b u t i o n . This yields
10
10'^
10
10
C
- 3
(cm !
r
Thus,
N ^
= N ^.
A
(6.18)
{x)
(6.19)
where the o r i g i n , x = 0, is taken to be at the j u n c t i o n .

A p p l y i n g the boundary c o n d i t i o n &(W) 0 yields
The m a x i m u m electric field in the p-n j u n c t i o n is then

<^max
Inspection o f Figure 6.5b shows that the t o t a l potential variationthe
m a x
q&
-Mo
(6.23)
(6.24)
c
and
M L ^ k
<f(0) -
cf(0) = c f
area under the field triangleis given by

<PT =
( - )
W e can n o w readily obtain the total depletion region w i d t h W of a step

junction as a function of the total electroslatic p o t e n t i a l variation from one
<?max( 1 - I
WJ
which is shown in Fiaure 6.7c.
(6.25)
Space-Charge Region for Step Junctions ".
161
Integrating once again to obtain the d i s t r i b u t i o n o f electron energy

yields
( 0 ) = q ^maxâ: -
+ constant.
(6.26>
T a k i n g as an arbitrary choice o f zero electron energy the electron energy

in the neutral ^-region or E^W) 0 and e m p l o y i n g E q u a t i o n 6.20 yields
or
-^J
1 -
(6.28)
where
nC
 \ =
Fv
Fn
(6-29)
T h i s distribution is shown i n Figure 6 . 7 d . |

Important
relationships
pertaining
to
one-sided step junctions
are
summarized i n Table 6.1 at the end o f this chapter.

b.
R e v e r s e Bias and F o r w a r d Bias
I f we apply a positive voltage V to the rc-region w i t h respect to the

/"-region, the total electrostatic potential v a r i a t i o n across the j u n c t i o n =4>
T
+ V.
(6.30)
As a result, the w i d t h o f the space-charge region on b o t h the rc-side and o n

the /7-side w i l l increase.
Id)
Fig. 6.7 The one-sided step-function approximation, (a) Concentration distribution for
a "shallow" diffusion, (b) Charge distribution, (c) Electric field distribution, (d) Potential distribution. Calculations are for silicon at 300K.
This situation is-illustrated i n Figure 6.8 where
t It is difficult to actually calculate the Fermi potential on the heavily doped side of
such a junction. In practice, we can assume that the Fermi level will just touch the
appropriate band edge so that the Fermi potential on the heavily doped side will be
just half of the energy gap, or 0.55 ev for silicon. A n alternative assumption, which
often leads to results in closer agreement with experiment, is to take (f> = 2<j> where
<f>r is the Fermi potential on the lowly doped side.
.
"B
Space-Charge Region for Linearly G r a d e d Junctions
163
manner. I n this case V is negative and therefore the t o t a l v a r i a t i o n o f

electrostatic potential across the j u n c t i o n 4> is smaller than i n e q u i l i b r i u m ,
bringing about a n a r r o w i n g o f the space-charge region. However, as we
shall see i n a later section, i n this c o n d i t i o n large currentsmany orders
of magnitude larger than i n the reverse-bias caseflow across the p-n
j u n c t i o n . Corresponding to these large currents, the carrier concentrations
F
1 1 11
11
1 11 j
1 11 1
1 l>"L
10
C
= lO^cm-J^--^''^
1
10 j_____-lO'*
10 "
10
1
10"-
we indicate the increase in b o t h the m a x i m u m electric field and the total

potential variation f r o m one side o f the j u n c t i o n to the other as well as
the widening o f the space-charge region.
Substitution o f E q u a t i o n 6.30 i n t o 6.21 or 6.22 yields the w i d t h o f the
space-charge region as a function o f applied reverse voltage V .
Calculations are shown i n Figure 6.9 for the one-sided step-junction approximat i o n w i t h substrate i m p u r i t y concentration as parameter. The depletion
region width is relatively constant while the reverse bias V is small in
comparison to the built-in voltage i . W h e n the applied reverse voltage
becomes large in comparison to the b u i l t - i n voltage, the depletion region
w i d t h increases w i t h the square r o o t o f the applied reverse voltage.
R
The quasi-Fermi levels associated w i t h the reverse-biased p-n j u n c t i o n ,

E
and E , are also indicated i n Figure 6.8. Their significance w i l l be
discussed i n relation to f o r w a r d currents i n Section 6.6.
The forward-bias c o n d i t i o n , i n which a negative voltage V is applied
to the n-region relative to the /^-region, can be considered i n a similar
Fn
Fp
^>
-^"^
y
^
III.
10"'
lp
i i l l
i
10"'
^-
I M !
i i 11
10
ir(v)
1 11 1
10
2
I I I
10
Fig. 6.9 Depletion region width versus applied reverse bias In the one-sided stepjunction approximation. (Silicon, 300 K.)
5
in the space-charge region w i l l become less and less negligible i n c o m p a r i son to the i m p u r i t y ion concentration so that the depletion a p p r o x i m a t i o n
loses its validity.
6.3
SPACE-CHARGE REGION
GRADED J U N C T I O N S
FOR
LINEARLY
Whereas the one-sided step-junction a p p r o x i m a t i o n provides an adequate description for alloy junctions and for " s h a l l o w " diffused j u n c t i o n s ,
in the case o f "deep" diffused junctions we must use another simple
a p p r o x i m a t i o n , called the linearly graded junction
approximation. I n
Figure 6.10 a complementary error-function d i s t r i b u t i o n is shown w i t h
the same surface concentration and the same substrate concentration as
the one in Figure 6.7. The only difference is that the diffusion i n this case
p-n Junctions
164
is continued long enough to make the j u n c t i o n depth 100 u instead o f the

1 ,u as i n Figure 6.7.|
While it is evident that the one-sided step-junction a p p r o x i m a t i o n could
n o t be applied to describe such a d i s t r i b u t i o n , i t also appears from F i g .
6.10 that the net i m p u r i t y d i s t r i b u t i o n , or the charge d i s t r i b u t i o n , can be
well approximated by a straight line t h r o u g h o u t the depletion region.
T h i s straight-line charge d i s t r i b u t i o n , when integrated by using Poisson's
4.0 x 10'"
Space-Charge Region for Linearly G r a d e d Junctions
where W
and
20
erfc. C = 1 0 c m
X) = 100 u
s
2.0
is the- w i d t h o f the zero-bias (equilibrium) depletion region,
= =t
~ dx
is the impurity concentration gradient at the junctionthe

teristic o f linearly-graded junctions.
0.9
3.0
165
'I
III
I I I I
I I II
principal charac-
I I I I
I III
10"
IT"
0.8
0.7
0.6
0.5
0.4
1-OrrA l
~0NT
0.3
10'
10
Jj
10'
1Q20
I I II
I
.
1Q21
a (cm."" )
10"
X in)
1.0
Fig. 6.11
(b)
Fig. 6.10 The linearly graded junction approximation.
(a) Concentration distribution for "deep" diffusion.
(b) Charge distribution.
Built-in voltage for linearly graded junctions. (Silicon, 300K.)
I n Figure 6.11 the built-in voltage o f linearly graded junctions in silicon

at r o o m temperature is shown as a function o f a. This figure was obtained
by solving the transcendental equation that results when W is eliminated
between Equations 6.31 and 6.33. Figure 6.12 shows the depletion
region width as a function o f applied reverse bias V for l-p. and 10-,u
deep diffused junctions, based on the linearly graded j u n c t i o n a p p r o x >
mation.
I t should not be inferred from these figures that the linearly graded
j u n c t i o n approximation is actually v a l i d throughout the range o f these
calculations. For instance, i t can be seen f r o m Figure 6.10 that, as the
depletion region w i t h W increases, the straight-line approximation begins
to deviate increasingly from the actual charge distribution. The linearly
graded j u n c t i o n approximation becomes rather meaningless for
0
equation, leads t o a parabolic electric field d i s t r i b u t i o n and a cubic potent i a l d i s t r i b u t i o n . I n particular, the w i d t h o f the depletion region w i l l be
given by
l2K,e <p
0
(6.31)
qa
the m a x i m u m electric field by
6
1.5
(6.32)
a n d the b u i l t - i n voltage by
2kT.
9B
aW
In-
(6.33)
2n
t This is an extremely deep diffusion used only to emphasize the difference between the
two cases.
(6.34>
p-n Junctions
167
Space-Charge_ Region for Diffused Junctions
6.4
SPACE-CHARGE REGION
DIFFUSED JUNCTIONS
FOR
W e have considered t w o l i m i t i n g cases o f diffused j u n c t i o n s : shallow

junctions that can be well represented by -the one-sided step-junction
a p p r o x i m a t i o n , and deep junctions which can be well represented by the
linearly graded j u n c t i o n approximation. Practical diffused p-n j u n c t i o n s
m a y be approximated by one or the other o f these, or they m a y be i n
between.
10
1 IT]
I I |
I I I.
-TTTT-
Diffused junction''
One-sided step junction
- Linearly graded junction
10
10
10"
I II ' I
10"
I I II
1
' M l
10
'.IVJ
I I I I
10
10
Fig. 6.13 Depletion region width versus applied reverse bias for a diffused junction:
comparison with the one-sided^tep-junction and linearly-graded junction approximations.
Erfc distribution, C 10" cm" , C = 10" cm" , z = 10 u.
3
Extensive numerical integration o f Poisson's equation for diffused j u n c tions has been performed by Lawrence and W a r n e r . W e n o w consider
some o f their results plotted in a form which enables easy comparison w i t h
the t w o simple closed-form approximations considered above. I n Figure
6.13 we show the depletion region width as a function o f applied reverse
bias for a fairly typical diffused j u n c t i o n (Silicon, r o o m temperature). T h e
points represent the results o f Lawrence and W a r n e r , while the lines correspond to the one-sided step-junction and the linearly graded j u n c t i o n
approximations, respectively. This figure demonstrates that a given p-n
j u n c t i o n cannot by itself be classified as either linearly graded or step;
the particular one considered i n this figure follows the linearly graded
j u n c t i o n a p p r o x i m a t i o n at low bias voltages and the one-sided step-junct i o n a p p r o x i m a t i o n at high bias voltages.
1
168
p-n Junctions
R l )
Fig. 6.14 Depletion region width versus applied reverse bias for junctions diffused into
substrates of various impurity concentrations: comparison with the one-sided stepjunction and linearly graded junction approximations. Erie distribution. C = 10" cm" ,
3
T h i s interesting fact is given a simple explanation i f we consider that the

relative steepness o f the i m p u r i t y concentration gradient should depend o n
the size o f the depletion region which, i n t u r n , depends on the bias voltage.
T h u s we have to compare the characteristic length o f the diffusion
the diffusion length 2-4 Dtwith
the characteristic length o f the spacecharge region W to determine whether a particular j u n c t i o n is i n the step
o r i n the linearly graded j u n c t i o n regime. This is further illustrated by the
calculations shown i n Figures 6.14 and 6.15 where Lawrence and Warner's
calculations are compared w i t h the t w o approximations for various substrate i m p u r i t y concentrations, for 1- and 10-/x j u n c t i o n depths, respectively. Wherever possible, the diffusion length 2-JDt corresponding to a
p a r t i c u l a r j u n c t i o n is indicated.
A useful rule-of-thumb criterion can be inferred f r o m these figures:
W h e n the depletion region w i d t h W is larger than the diffusion length
2-JDt
o f the j u n c t i o n - f o r m i n g i m p u r i t y diffusion, the one-sided step
j u n c t i o n a p p r o x i m a t i o n is better. Conversely, when the depletion region
w i d t h is smaller than the diffusion length, the linearly graded j u n c t i o n
a p p r o x i m a t i o n is preferable.
Fig. 6.1S Depletion region width versus applied reverse bias for junctions diffused into
substrates of various impurity concentrations: comparison with the one-sided stepjunction and linearly graded junction approximations. Eric distribution, C = 10" cm" ,
x, = 10 u.
3
6.5
CAPACITANCE-VOLTAGE
CHARACTERISTICS"
Capacitance per u n i t area is definedf as C = dQ\dV, where dQ is the

incremental increase i n charge per u n i t area upon an incremental change
i n the applied voltage dV.
Let us now calculate the capacitance o f a p-n j u n c t i o n with an arbitrary
impurity distribution, such as shown i n Figure 6.16. The charge and e l e c ~
trie field distributions designated by the solid line correspond to a voltageV applied to the n-region. I f this voltage is increased by an a m o u n t dV,
the charge d i s t r i b u t i o n and the electric field d i s t r i b u t i o n will b o t h change
to that indicated by the dashed line. The incremental charge dQ corresponds to the area between the two charge-distribution curves on either
side o f the depletion region, while the magnitude o f the change i n the
applied voltage dV is indicated by the cross-hatched area between the t w o
electric field d i s t r i b u t i o n curves.
The increase i n depletion layer w i d t h and the corresponding increase i n
charge on the /j-side o f the depletion region w i l l b r i n g about an increase i n
the electric field by an amount dS = dQjKfy.
(See Section 6.1.) T h e .
t T h i s d e f i n i t i o n y i e l d s t h e small-signal c a p a c i t a n c e .
168
p-n
Junctions
Capacitance-Voltage C h a r a c t e r i s t i c s
10-2-1
10
1
2
LJjJ
1 0 " '
M i l
1
I I I I.
10
fill
1 0
I
2
I I I
1 0
Fig. 6.IS Depletion region width versus applied reverse bias for junctions diffused into
substrates of various impurity concentrations: comparison with the one-sided stepjunction and linearly graded junction approximations. En'c distribution, C , = 10" cm" ,
x, = 10 ft.
3
T h i s interesting fact is given a simple explanation i f we consider that the

relative steepness o f the i m p u r i t y concentration gradient should depend o n
the size o f the depletion region which, i n t u r n , depends on the bias voltage.
T h u s we have to compare the characteristic length o f the diffusion
the diffusion length 2-4 Dtwith
the characteristic length o f the spacecharge region W to determine whether a particular j u n c t i o n is i n the step
o r i n ' t h e linearly graded j u n c t i o n regime. This is further illustrated by the
calculations shown i n Figures 6.14 and 6.15 where Lawrence and Warner's
calculations are compared w i t h the t w o approximations for various substrate i m p u r i t y concentrations, for 1- and 10-/* j u n c t i o n depths, respectively. Wherever possible, the diffusion length 2>/Dt corresponding to a
p a r t i c u l a r j u n c t i o n is indicated.
A useful rule-of-thumb criterion can be inferred f r o m these figures:
W h e n the depletion region w i d t h W is larger than the diffusion length
2-J Dt o f the j u n c t i o n - f o r m i n g i m p u r i t y diffusion, the one-sided step
j u n c t i o n a p p r o x i m a t i o n is better. Conversely, when the depletion region
w i d t h is smaller than the diffusion length, the linearly graded j u n c t i o n
a p p r o x i m a t i o n is preferable.
6.5
CAPACITANCE-VOLTAGE
CHARACTERISTICS"
Capacitance per u n i t area is defined! as C a dQ/dV, where dQ is the

incremental increase i n charge per u n i t area u p o n an incremental change
in the applied voltage dV.
Let us now calculate the capacitance o f a p-n j u n c t i o n with an arbitrary
impurity distribution, such as shown i n Figure 6.16. The charge a n d e l e c "
trie field distributions designated by the solid line correspond to a voltager
V applied to the n-region. I f this voltage is increased by an a m o u n t dV,
the charge d i s t r i b u t i o n and the electric field d i s t r i b u t i o n will b o t h change
to that indicated by the dashed line. The incremental charge dQ corresponds to the area between the two charge-distribution curves o n either
side o f the depletion region, while the magnitude o f the change i n the
applied voltage dV is indicated by the cross-hatched area between the t w o
electric field d i s t r i b u t i o n curves..
The increase i n depletion layer w i d t h and the corresponding increase i n
charge on the /i-side o f the depletion region w i l l b r i n g about an increase i n
the electric field by an amount dtf = dQjK, .
(See Section 6.1.) T h e .
t This definition yields the small-signal capacitance.
Q
!70
p-n
Junctions
Capacitance-Voltage
Characteristics
corresponding change i n the applied voltage dV is approximately (dS) W
w i t h i n the depletion region. The rearrangement o f these mobile carriers
w h i c h equals (dQlK e )
w i l l contribute an a d d i t i o n a l t e r m to the capacitance.
W. Thus, by the definition o f the capacitance per
u n i t area, we find that
Because o f the very simple relationship between the reverse-bias j u n c t i o n

C-m
(6.35)
W
This equation holds for
any arbitrary
impurity
capacitance a n d the w i d t h o f the depletion region, measurements o f j u n c t i o n capacitance can provide useful i n f o r m a t i o n about the nature o f the
distribution.
impurity distribution.
F o r example for a one-sided step j u n c t i o n , i f we
combine Equations 6.22 and 6.35 we obtain
2(v
* *
+ <p )
(6.36)
w h i c h can be rearranged to yield
k
C
= - ~ - r (V* + <PBY
qK e C
s
(6-37)
I t is evident t h a t by p l o t t i n g 1/C versus V , a straight line should result

i f the actual i m p u r i t y distribution can be a p p r o x i m a t e d by the one-sided
step-junction theory. I n such a case, the slope yields the i m p u r i t y concent r a t i o n i n the substrate C while the intercept yields the b u i l t - i n voltage
R
dS =
4>B3
Similarly, f o r linearly graded j u n c t i o n s , we can p l o t 1/C versus

V,
and obtain a and <p from slope and intercept i f a straight line results.
I n general, however, we can evaluate an arbitrary i m p u r i t y d i s t r i b u t i o n
w i t h fairly good accuracy from the capacitance-voltage characteristics.
This general m e t h o d is illustrated i n Figure 6.17. W e consider the simplest
case, that o f a shallow diffused n+p j u n c t i o n w i t h an arbitrarily v a r y i n g
i m p u r i t y d i s t r i b u t i o n on the lower doped side. A s before, the charge o n
the heavily doped side o f the depletion region w i l l change by dQ as the
applied reverse voltage is changed by dV, where dO q'N(W) dW'.
Correspondingly, the electric field w i l l change by an amount
R
Fig. 6.16 Illustration of the change in charge and electric-field -distribution resulting
from a change In applied reverse bias.
T h e capacitance per u n i t area o f a parallel-plate capacitor is given by the

dielectric constant divided by the separation between the plates. Thus the
p-n j u n c t i o n capacitor can be regarded as a parallel-plate capacitor where
the distance between the t w o platesthe distance between the regions
where the incremental charge appearsis j u s t the depletion resion w i d t h
W.
I n deriving E q u a t i o n 6.35 we tacitly assumed t h a t all the extra charge
t h a t appears i n the semiconductor u p o n the change i n applied voltage
appears as a change i n the charge contained w i t h i n the depletion region.
T h i s is certainly a good assumption i n the reverse-bias c o n d i t i o n . I n the
-forward-bias condition, however, large currents can flow across the deplet i o n region. Corresponding t o these large currents, there w i l l be a considerable charge due t o mobile carrierselectrons and holespresent
dQ
_ qN(W)
dW
The corresponding change i n applied voltage is approximately
dV = (d<?) r
5
2K,t
(6.38)
Substituting for W
f r o m E q u a t i o n 6.35 leads to an expression for the
i m p u r i t y concentration at the edge o f the space-charge region:
p-n Junctions
172
173
a diode with respect to the /^-region as shown in Figure 6.18, the electronhole pairs, once generated, w i l l be separated and their probability o f
recombination is diminished. This is the basic mechanism giving rise to
all types o f reverse currents observed in p-n j u n c t i o n diodes. A l l such
currents, accordingly, are due to electron-hole pairs generated
someplace
in the semiconductor.
Various contributions to the reverse current are
<-j L
L.
ft
Generation within
\
' space-charge region : I
"generation current"
J
j
K - (iVV
Generation within diffusion length of depletion region :

"diffusion current"
rJ
(ai
-qN(W)
JL
X
(b)
Fig. 6.17
Illustration of the concentration, charge, and electric field distribution leading
to the measurement of the impurity distribution.
Fig. 6.18 Illustration of the mechanisms giving rise to reverse current.

(a) n p Diode under 3-v reverse bias.
(b) Corresponding band diagram.
T
Thus measurements o f the capacitance per unit area C as a function o f

reverse voltage, w i t h appropriate differentiation, can provide the i m p u r i t y
d i s t r i b u t i o n directly. This method has been-used extensively i n semic o n d u c t o r w o r k . A m o n g other places, i t was employed i n the investigation
o f i m p u r i t y redistribution d u r i n g thermal oxidation a n d i m p u r i t y redist r i b u t i o n d u r i n g epitaxial g r o w t h w h i c h were discussed i n Chapter 3.
6.6
CURRENT-VOLTAGE
CHARACTERISTICS
distinguished by the region o f the semiconductor where an electron-hole

pair was generated. Thus, when an electron-hole pair giving rise to the
current is generated w i t h i n the reverse-biased depletion region o f the
j u n c t i o n , region 2 i n Figure 6.18, we refer to the resulting current component as the generation current. W h e n electron-hole pairs are generated i n the
neutral regions 1 and 3 i n Figure 6.18 and the carriers diffuse to the reversebiased j u n c t i o n , thereby leading to an additional current component, we
talk o f the diffusion current.
W e n o w consider these two current components separately.
av
R e v e r s e Bias
U n d e r e q u i l i b r i u m conditions electron-hole pairs are generated cont i n u a l l y everywhere w i t h i n the semiconductor. I n the absence o f an
applied voltage, the electron-hole pairs recombine ur.d therefore no current
flow results. However, when a positive voltage is applied to the n-reg.cn ot
Generation within Space-Charge Region: Generation C u r r e n t .

I n a reverse-biased depletion region, for reverse bias V kTjq, the concentrations o f carriers are reduced well below their e q u i l i b r i u m concentrations. This is because the electric field, which is increased above its
e q u i l i b r i u m value by the applied reverse voltage, sweeps the carriers out o f
R
p-n
7 i )
Junctions
the depletion regionit sweeps holes to the />-region and electrons to Jhe
Current-Voltage
Characteristics
175
I f the centers are indeed located near the intrinsic F e r m i level. r w i l l be

practically independent o f temperature. We would-then expect the generation-current c o m p o n e n t to have the same temperature dependence as .
The generation-current component is dependent on the magnitude o f the
applied reverse biasat higher biases W is larger, more centers are i n cluded w i t h i n the depletion region, and the generation current increases i n
p r o p o r t i o n to W.
0
n-region.
Because of the reduction in carrier concentrations, o f the four recombination-generation processes that take place t h r o u g h intermediate-level
centers (see Figure 5.11), only the emission processes
are i m p o r t a n t .
The
capture processes are n o t i m p o r t a n t because their rates are p r o p o r t i o n a l

to the concentrations of the free carriers w h i c h , as we said above, are verysmall i n a reverse-biased depletion region.
The only way the t w o emission processes can operate i n steady state is by
alternatine;
thus t h centers w i t h i n the depletion region alternately emit
electrons and holes. The rate o f generation o f electron-hole pairs i n such

a situation can be readily obtained f r o m E q u a t i o n 5.40 by setting p,
n
This leads to
= _ HL
W <"<
where T is denned as the effective lifetime
within a reverse-biased
(6.40)
depletion
region and is given by

T #
(6.41)
2o ov N
P
ih
Generation
outside the
Space-Charge
Region:
Diffusion
C u r r e n t . I n the neutral regions outside the space-charge region there
is no significant electric field present and the m i n o r i t y carriers move only by
diffusion. I f electron-hole pairs are generated in the n-region i n the v i c i n i t y
o f the space-charge region, those holes that reach the edge o f the spacecharge region by diffusion w i l l be swept toward the /j-region b y the i n creased electric field present w i t h i n the space-charge region. Conversely,
those electrons t h a t reach the space-charge region edge f r o m the n e u t r a l
z>-region w i l l be swept by the electric field toward the n-region. These
current components are referred to as the diffusion current. T o calculate
the magnitude o f the diffusion current we have to solve the steady-state
diffusion equation f o r m i n o r i t y carriers. For electrons in the jP-region,
this is
. T o visualize the most i m p o r t a n t features o f this expression, let us consider a

D
simpler case in which a, = a = o. F o r this case. Equation 6.40 reduces to

U
E\
2 cosh
'
(6-42)
N o t e that, in agreement with our argument i n Chapter 5, only those centers

whose energy level E, is near the intrinsic Fermi level , contribute significantly to the generation rate. The generation rate falls exponentially as
the center leveljmoves away from the m i d d l e o f the gap i n either direction.
" I n the particular case when , = , , T w i l l be the same as the lifetime T
of the carriers in a neutral material,
\lav N,.
th
One electron-hole pair generated provides one electronic charge to the

external circuit. Thus the magnitude o f the current due to generation
w i t h i n the depletion region w i l l be given by
- ^ ^
= 0.
(6.45)
po
,(>)
= n
pc
+ rG .
(6.46)
A t the depletion-region edge, for large enough reverse bias V kTiq the
m i n o r i t y carrier concentration w i l l be zero, since any m i n o r i t y carriers
reaching the d e p l e t i o n region w i l l be rapidly swept away by the field as we
have argued above. Thus,
B
= q\U\WAj,...
(6.43)
n,(0) = 0
= \q
(6.44)
if, for this discussion, we take our origin, x = 0. at the edge o f the deplet i o n region.
g e n
WAj
|o
where Aj
eeD
J
or
/
I n this equation, i n addition to the net thermal generation-recombination

t e r m U = (n n )j-,
we have also included a term G due to external
means o f generating carriers such as, for instance, by i l l u m i n a t i o n . T h i s
equation is the same as the one we used i n Chapter 5 in the case dealing
w i t h surface r e c o m b i n a t i o n . As i n that problem, far away from the spacecharge region the carrier concentration w i l l be given bv
P
kT
^
dx-
is the cross-sectional area o f the p-n j u n c t i o n .
(6.47)
The solution o f this boundary value problem is eiven by

"(*) = ( + T.CiXl -
xlL
e~ ")
(6.48)
177-
where L
= \ Dr
n
is "the diffusion length o f electrons in the ^-region.
The diffusion current due to electrons is then given by

''1!
/.tiff,, = ( - q ) i - D . x
ax
(n, + T , G ) .
Aj.
r
L >
qD
ii-o''
{ P
G L )
The generation rate per unit volume i n the neutral region depleted o f
m i n o r i t y carriers only, p \r
or n \r ,
can be drastically different from
the generation rate per unit volume w i t h i n the depletion region, TI,, 2r .This
can lead to very wide differences i n the relative importance o f the diffusioncurrent component and the generation-current component o f the reverse
current. The ratio o f the two currents, f o r r = r, = r is given by
n
(6.49)
L ,
Aj.
(6.50)
"va
df,n
I n the case o f no illumination,
A similar argument for the n-region gives the diffusion current due to holes
i n t h e /i-region reaching the depletion region edge,
W. = qD,
for the example o f electrons i n the ^-region. This result is the same as
Equation 6.51.
these reduce to
gen
-" = 2 - ^ ^ .
(6.53)
i f f ,
2 r
n
Ui,t,
= qD -fA
n
= q D
(6.51)
and
This ratio is evidently proportional to the intrinsic carrier concentration

n . As the temperature is increased, the diffusion current has an increasing
tendency to dominate. Between different materials, those w i t h smaller
band gap, hence larger n have a larger diffusion-current-to-generationcurrent ratio than those w i t h a larger band gap.
{
Uui,
= qDÂj
= qD,--^
A j,
(6.52)
respectively, where we have incorporated the e q u i l i b r i u m c o n d i t i o n . N o t e

the absence o f bias dependence in these expressionsas we have mentioned, they represent the case only for V' kT'q, f o r w h i c h the diffusion
current saturates.
The temperature dependence o f the diffusion current
is the same as that o f nf.
The diffusion-current components can be interpreted also i n the f o l l o w ing way. Those m i n o r i t y carriers that are generated w i t h i n a diffusion
length o f the edge o f the depletion region w i l l c o n t r i b u t e to the diffusion
current because they have a chance to reach the edge o f the depletion
region. Thus we w o u l d expect the diffusion current to be given by
R
i U t
= q [net generation rate per u n i t volume i n the neutral region]

x [diffusion length] A j .
T o estimate the net generation rate i n the neutral region w i t h o u t i l l u m i nation, we use E q u a t i o n 5.42 subject to the c o n d i t i o n that p p
or
n n . ( N o t e that this is quite different f r o m the c o n d i t i o n we applied
i n the reverse-biased depletion region where b o t h carriers were depleted
below n. Here we need to say only that the m i n o r i t y carrier concentration
near the edge o f the depletion region is l o w e r than i t w o u l d be i n equilibrium.) This results i n
n
B0
n0
These two tendencies are illustrated by the experimental data shown in

Figure 6.19. Here the reverse current-voltage characteristics o f german i u m , silicon, and gallium arsenide diodes are shown at various temperatures. I t is evident that the germanium and the silicon diodes exhibit both
the diffusion-current componentwhich is distinguished by its lack o f
voltage dependence-r-and the generation-current componentwhich is
distinguished, in contrast, by- its voltage dependencein some tempera- .
ture range. The only difference is the temperature at which transition
from one type o f characteristic to the other takes place. The gallium
arsenide diode does not reach this transition, but i t does show a tendency
toward it.
-
I n Figure 6.20, the reverse current at 1 v o l t reverse bias is plotted as a

function o f the reciprocal o f the absolute temperature for all three diodes.
Different types o f points are employed here to designate primarily generation-current type characteristics and p r i m a r i l y diffusion-current type
characteristics. A break is evident i n each set o f data, occurring at a higher
temperature for the higher band-gap materials, in agreement w i t h our
argument. The temperature dependences o f the generation and diffusion
current components evidently agree reasonably well w i t h the temperature
dependence o f n and o f T ^ , respectively, as expected from Equations 6.44
and 6.51, or 6.52.
2
Zdiff.ii = q
L Aj
n
p-n Junctions
I7B
100 ua
10 ua
1 ua
^ 2 5 0 " C ^
100 na
10 na
1 na
/ S
s \
150"C^/
1 oo ' c /
100 pa
'
50'C/V
/To
1 pa
10"
1
I O " 10"
2
100 ua
Reverse current-voltage characteristics at various temperatures. Germanium
10 ua
100 fja
100 na
1 ua
-?
/l75'C
'10 na
1 0 0 pa
/75*C
'
10 pa
1 pa
0
1 pa
IO"
-r
IO'
1
IO
1
1
1
10
r(vi
Fig. 6.19b
Silicon diode.
10
\A
\Ge
\
\
\ \
\\ \
'C
W.
1 na
100 pa
10 na
!
1
V \
T i
G a A s \ ISi
A 1
ft t
1 na
:
10
1
\
\
10 ua
100 na
1 nia
/275'Z
10
li
11
IM
I \
diode.
l'(vl
Gallium arsenide diode.
Fig. 6.19c
Fig. 6.19a
'C
1
IT
!
2
3
4
1000..TI K )
1
5
\
6
Fig. 6.20 Reverse current at

= 1 v as a (unction of temperature. Dots designate
primarily generation-current type,'circles designate primarily diffusion-current type
current-voltage characteristics. Dashed lines represent temperature dependence of
n, ; solid lines that of n,.
4
p-n
180
Junctions
Let us now consider the reverse current flowing in a diode under i l l u m i n a t i o n . I f the i l l u m i n a t i o n is strong, i.e., G n -r , p Jr ,
/ T we
can replace the t h e r m a l generation rate by (7/_, i n b o t h the generationcurrent and the diffusion-current components. This leads to
L
W o
= 1 L(L,
+ L+
mi
W)Aj.
Current-Voltage
Characteristics
181
- 0.35
1
Recombination within
\
space-charge region:
\
"recombination currant"
Recombination within a diffusion length"of depletion
region : "diffusion c u r r e n t " -
(6.54)
Because the sum o f the two diffusion lengths is usually larger than the
depletion region w i d t h , the photocurrent w i l l show little bias dependence.
(a)
100 un
10
-0.35 v
^ 5
J
L-
(b)
1 /m
10Q.na
L,
_*
10nu
L,
_>
1~im
L,
100:03-;
10-paDarx
1 pa
10"
10"' 1 0
_ 1
I
10
10
K|v)
Fig. 6.21
Reverse currenc of a silicon diode in the dark, and under illumination by white
light of various intensities.

10
* (/H
T h i s is illustrated i n Figure 6.21 where the reverse current o f the silicon

diode is shown i n the d a r k and u n d e r i l l u m i n a t i o n by white light o f various
intensities.
b.
15
(cl
Fig. 6.22 III ustration of the mechanisms giving rise to forward current
(a) n p diode under 0.3S-V forward bias(b) Corresponding band diagram.
(c) Corresponding distribution of electrons and holes.
Dashed lines indicate equilibrium'electron and hole distributions.
T
Forward Bias
I f a negative voltage is applied to the n-region relative to the ^-region,

as s h o w n i n Figure 6.22, the t o t a l v a r i a t i o n o f the electrostatic p o t e n t i a l
across the j u n c t i o n , d> , is reduced. I n equilibrium, the c o n d i t i o n o f zero
current flow was. established by a precise balance between the diffusion
T
and the drift currents o f each carrier across the p-n j u n c t i o n . Since under
forward-bias conditions the t o t a i electrostatic potential variation, and
w i t h i t the electric field across the space-charge region, is reduced, the drift
current component o f each carrier is also reduced and the balance between
p-n Junctions
182
drift and diffusion contributions to the c u r r e n f i s disturbed, resulting in a

net current flow.
I t is i m p o r t a n t to realize that although the potential variation across the
j u n c t i o n is reduced, it w i l l not vanish. As a consequence, the electric field
i n the space-charge region w i l l still be directed t o w a r d the />-region. Thus
the current flow w i l l be against the direction of the electric field. This is
possible because o f the huge electron and hole' concentration gradients
existing near the j u n c t i o n .
A n alternative way o f considering the forward-bias case is as follows.
-With a negative potential applied to the 77-region relative to the ^-region,
excess electrons w i l l be pushed i n t o the n-region and excess holes into the
/^-region. Thus the electron and hole concentrations w i l l b o t h be above
-their respective e q u i l i b r i u m values and therefore the "pn" product w i l l
exceed H, throughout portions o f the semiconductor. Under such injection
conditions the carrier concentrations w i l l attempt to return to their equil i b r i u m values by recombination. I n the steady state, electrons and holes
disappearing through recombination w i l l be replenished by more electrons
and holes coming i n t h r o u g h the contacts to the n and the p regions,
respectively. This gives rise to a forward current.
A c c o r d i n g to this second picture, the forward current is due to the
recombination of electron-hole pairs i n much the same way as the reverse
current, which was considered i n the previous section, is due to the generat i o n o f electron-hole pairs in various regions o f the semiconductor. Thus
it should be possible to derive the magnitude o f the forward current byconsidering the rates o f recombination i n the various regions o f the semiconductor.
As i n the case o f reverse current, we again distinguish between three
-regions, illustrated i n Figure 6.22a: the neutral regions 1 and 3 adjacent
to the space-charge region and the space-charge region 2 itself. The
magnitude of the t o t a l f o r w a r d current w i l l be given by the flux of electrons
entering the ?j-region t h r o u g h the left-hand contact. T o calculate this flux
_ o f electrons, we will add up the three parts o f the flux which are consumed
by recombination w i t h holes i n the three regions mentioned above. Thus
the flux o f electrons entering t h r o u g h the contact will be given by
Current-Voltage
Characteristics
183
The current Components 1 and 3 are called the diffusion current as i n the
reverse-bias case; the current component 2 is called the
recombination
current.
Recombination outside the Space-Charge Region:
Diffusion
Current.
L e t us first consider the flux of electrons F injected i n t o the
neutral /> reeioni.e., current component 3. Since electrons i n the pregion are m i n o r i t y carriers, i n low-level injection they w i l l move away
f r o m the edge o f the space-charge region b y d i f f u s i o n o n l y (see Chapter 5).
I n order to calculate their flux we first must o b t a i n their d i s t r i b u t i o n i n
the/>region.
~ - - P '
T h i s d i s t r i b u t i o n i n the steady state w i l l be obtained by solving the

diffusion equation under steady-state conditions (Chapter 5),
n
^
dx
^
z.
'
(6.55)
The first b o u n d a r y condition requires that the concentration o f excess

electrons far away f r o m the edge o f the depletion region should vanish, i.e.,
M,(CO) = n .\
x
(6.56)
po
The second boundary condition is that the concentration o f injected

electrons at the edge o f the space-charge region, where we choose o u r
origin, x = 0, should equal some constant value nÔ).
This p r o b l e m is identical to the case of injection f r o m one b o u n d a r v
considered i n Chapter 5. Thus the solution is also identical to the s o l u t i o n
i n the earlier case and is given by
"(*) = " + M O ) -
xlL
n }e- .
V0
(6.57)
The flux o f electrons entering the /-rgion is then given bv

F
= - D
*
dx
- j / f f f l - . V .
(6
.58)
N e x t , we consider the flux o f electrons consumed by r e c o m b i n a t i o n w i t h

holes i n region 1. F o r every electron recombining i n this region, a hole
must be injected i n t o the ?i-region f r o m the space-charge region. T h u s we
can evaluate this c o n t r i b u t i o n to the current by calculating the flux o f
holes F entering the n-region. This can be done i n a manner entirelyanalogous to the above calculation o f the flux o f electrons entering the
/>region and results i n
t
= (number of electrons recombining w i t h holes i n neutral n-region
e r
time)
4- (number o f electrons recombining w i t h holes in space-charge

region (5) per u n i t time)
4- (number o f electrons recombining w i t h holes i n neutral
^-region (3) per unit time)
F , P , f "
- r
(6.59)
t This condition assumes that the contact to the /"-region is "infinitely" far from the
junction.
v_urrenc- oitage \ . i i i r a c u : r u u
where p.,(0) is the hole concentration at the edge o f the space-charge

region, on the rc-side. I n each case, x designates the distance from the edge
of the space-charge region, as shown i n Figure 6.22.
In order to evaluate these fluxes fully, we need to determine /i.,(0)-and
p^{0).
F o r this, we must resort to an i m p o r t a n t simplifying assumption:
we assume that quasi-equilibrium holds, i.e., that the "pn" product is a
constant throughout the forward-biased space-charge region. This constant, however, w i l l be larger than n , the value o f the "pn"' product i n
equilibrium.
A n alternative way o f stating the assumption o f quasi-equilibrium is in
terms o f quasi-Fermi levels. The quasi-Fermi level for either carrier is
defined as that quantity which, when substituted into the place of the Fermi
level, gives the concentration of that carrier under non-equilibrium
conditions.
Thus the concentration o f electrons under n o n - e q u i l i b r i u m conditions is
given b y
n = e "(6.60)
s
[ E
E i ) l k T
n i
where E
is the quasi-Fermi level for electrons; and, likewise, the concentration o f holes under non-equilibrium conditions is given by
Fn
. <E.-E),kT
6.6:1)
n e
where E is the quasi-Fermi level for holes.

As a consequence o f this definition, the fluxes o f electrons and o f holes
F
now
become
1
dE
u^n
q
dx
F=
n
Fn
(6.62)
IOJ
In accordance w i t h this result.- we show the quasi-Fermi level for electrons E and that for holes E
in the place o f the Fermi level in the
respective neutral regions o f Figures 6.18 and 6.22. I n both figures, the
band diagram, and consequently the quasi-Fermi level for the majority
carriers is shown horizontal w i t h i n the neutral regions because both bias
situations correspond to relatively small current densities.
F
Fp
Because the electrostatic potential v a r i a t i o n across the j u n c t i o n is

altered i n the amount o f the applied voltage Vj, i t follows that the quasiFermi level for electrons in the neutral n-region w i l l also be displaced from
the quasi-Fermi level for holes in the neutral /j-region by the applied
voltage, or
(6.65)
as indicated in Figures 6.18 and 6.22.
We can now employ this exceedingly i m p o r t a n t result in conjunction
w i t h the assumption o f quasi-equilibrium, i.e., pn = constant throughout
the space-charge region, to calculate the concentration o f minority carriers
at the edges o f the space-charge region under forward-bias condition,
1,(0) and p{0). I t follows from Equations 6.60 and 6.61 that, in quasiequilibrium, (E
E ) = qVj t h r o u g h o u t the space-charge region
also, i.e., both quasi-Fermi levels are extended horizontally across the
space-charge region. This is indicated in Figure 6.22.
Fn
Fp
Thus the assumption of quasi-equilibrium is equivalent to assuming that

the quasi-Fermi levels are constant across the space-charge region.
T h e m i n o r i t y carrier concentrations at the respective edges o f the spacecharge regions are consequently given by
and
rilkT
(0) =
1
dE
= n e"^
lE
(6.66)
P0
and
E
mT
' p ( 0 ) = n e >- '*
dx
= p
I n the neutral n-region, electrons are transported principally by drift.
n o
e^
V f [ l k T
(6.67)
Thus the minority carrier concentrations at the respective edges of the spacecharge region will be enhanced by the exponential factor e ^ '
.
The
carrier concentrations i n a forward-biased one-sided n p j u n c t i o n ,
calculated on the basis o f this assumption, are shown i n Figure 6.22c.
v ,
Thus the flux o f electrons i n the /i-region is given simply by

l
= - - p n =
q
n
- i ^ n ^ i .
q
dx
l l k T
(6.64)
N o t i n g now that n = n -/N

and p
= n^lNr,,
respective diffusion current components become
vo
C o m p a r i s o n o f E q u a t i o n 6.64 w i t h 6.62 shows that dE Jdx

=
dEjdx,
i.e.,_that the quasi-Fermi level follows the energy band variation i n a
parallel fashion. A similar argument leads -to an identical conclusion
regarding the quasi-Fermi level o f holes i n the neutral /j-region. T h u s ,
in a neutral region, the quasi-Fermi level for the majority carriers will behave
exactly as the Fermi level does in equilibrium.
no
-qD
[ e
the equations for the
Um.n
Um.,
= -qD,
l",l/w _
and
-~*f-
kT
[e^>
l ]
(6.69)
186
p-n Junctions
187
Comparison of these equations w i t h equations 6.51 and 6.52 shows that

the diffusion current component under f o r w a r d bias \ V \ kTjq equals
t h e diffusion current component under reverse bias, multiplied by the
exponential factor
Here, the carrier concentrations w i l l be given by
Recombination within the Space-Charge Region: Recombinat i o n C u r r e n t . The electrons lost by recombination in the space-charge
region 2 give rise to a current component
and, therefore. C '
F o r \V \
rw
free = -qAj\
U dx.
(6.70)
m a x
l ' i w
by
kTlq,
&m = * < x c A r , n ,
Jo
(6.76)
, | F
t t
'
, / U T
(6.78)
or
This integral is complicated because U depends o n n and p. and n and p i n

turn depend on the distance a: i n a complicated manner. However, the
i m p o r t a n t features o f the result can be illustrated in. an approximate
fashion.
W e saw i n Chapter 5 that the most effective recombination centers are
those which are located near the middle o f the energy gap. Thus we now
consider centers w i t h energy level E, = E . Furthermore, for simplicity, we
take c = a = a. I n such a case the recombination rate U is given by
^-- e"
2T
'
(6.79)
Note that the m a x i m u m recombination rate w i t h i n the space-charge

region under f o r w a r d bias equals the generation rate under reverse bias
multiplied by the exponential factor e l > l ' .
We can n o w approximate the recombination current component /
by
I
2 i r
r e c
irec = ~ i q ~
We'^Âj
(6.80)
V-ov^N,
which amounts to t a k i n g U = U . throughout the space-charge r e g i o n . j

As in the reverse case, i t is interesting to compare the magnitudes o f the
diffusion and the recombination current components. U s i n g the electron
diffusion current component, this ratio is given by
mi7
(6.71)
n + p +
Because o f the assumption of quasi-equilibrium, the product o f electron
and hole concentrations t h r o u g h o u t the space-charge region is given by
pn = ^ V
'
(6.72)
irec
Thus the recombination rate is

U = av^N,-^
Jdirt.r. _
n + p + 2n,-
(6.73)
F o r a given forward bias V , V w i l l have its m a x i m u m value at that

- l o c a t i o n i n the space-charge region where the sum o f the electron and hole
concentrations (p + n) w i l l be at its m i n i m u m value. Since the product of
these concentrations pn = constant, the c o n d i t i o n d(p + n) = 0 leads to
F
dp = -dn
m 2 dp
P
(6.74)
or
p-n
w,- L
"
r | / r
f
As in the reverse-bias case, this ratio depends on the temperature and the
band gap o f the material t h r o u g h n,-. I n addition, it depends on the f o r w a r d
voltage t h r o u g h the exponential factor. For silicon at r o o m temperature,
at small f o r w a r d voltages the recombination current generally dominates
and at f o r w a r d voltages larger than about 0.4 to 0.5 v o l t the diffusion
current usually dominates.
B o t h current components can also be observed i n other materials, but
the transition f r o m one characteristic to the other depends on the temperature and the band gap o f the material. This is illustrated i n Figure 6.23
where we "show the f o r w a r d current o f germanium, silicon, and g a l l i u m
arsenide diodes as a function o f forward voltage. W e can distinguish the
(6.75)
as the c o n d i t i o n for this m i n i m u m . T h i s c o n d i t i o n exists at that location

within the space-charge region where the intrinsic Fermi level , is half
\va\ oetween the quasi-Fermi level for electrons E and for holes
E.
T
Fv
t Since U decreases exponentially with distance away from the point where it has its
maximum value, a better approximation would be to multiply U
by the volume of
the space-charge region within which the potential changes by an amount kTjq; that is,
by
(kTiqt)Aj.~
mKX
<_urrent-Yoltage
Characteristics
18?
recombination current component from the diffusion current component

by their different voltage dependences which correspond to slopes o f
q\1kT
and qjkT, respectively, in such a semi-logarithmic representation.
These slopes, for r o o m temperature, are shown by the dashed lines. We
can observe the change in the slope as the diffusion current begins to
dominate with increasing forward bias in the case o f the silicon diode.
-
0.2
0.4
0.6
0.8
1.0
WW
Fig. 6.23c
,ir
Gallium arsenide diode. Dashed lines indicate slopes of t t *
and > l'"'

r
dependences at 25C.
This comparison is particularly clear i n Figure 6.24 which shows the three
room-temperature characteristics together and over a larger current rangeA simple empirical representation o f the f o r w a r d current-voltage characteristics is given by the formula
I cc
F
eIPi.|/i**
(6.82)
where the empirical factor m = 1 for pure diffusion c u r r e n t and m = 2

for pure recombination current, provided the above simple theory is
applicable. W h e n b o t h currents are comparable, m w i l l vary between 1
and 2.
A more gradual increase i n current at high f o r w a r d voltages is evident
in all three cases shown i n Figure 6.24. T h i s phenomenon is associated
/ 1*3.
(e
m
) -
n<
p-n Junctions
190
100ma
Junction B r e a k d o w n
I9I
when W L,
the m i n o r i t y
straight-line d i s t r i b u t i o n ,
B
10 m a
n (A
9
1 ma
= n
vo
carrier d i s t r i b u t i o n w i l l
approach
+ [n(0) - n ] ( l - - ^ - J
the
(6.83)
in contrast to the exponential distribution, E q u a t i o n 6.57,. w h i c h holds

when W L. A s a result, the diffusion length L or L is replaced by
W in our formulas for the f o r w a r d bias diffusion current components,
Equations 6.68 and 6.69, as can be readily verified.
B
100/ja
10 ua
. . . . . .
6.7
1 ua
10 n
n a
0.2
0.4
0.6
0.8
1.0
1.2
Fig. 6.24 Comparison of the forward current-voltage characteristics of the germanium,

silicon, and gallium arsenide diodes at 25C. Dashed lines Indicate slopes of e' r
and
e ^ ' V " dependences.
,r
a.
Z e n e r B r e a k d o w n (Tunneling)
ilkT
w i t h the onset o f high-level injection i n w h i c h the theory presented i n this

chapter must be modified. Because high-level injection phenomena are
particularly i m p o r t a n t i n the operation o f j u n c t i o n transistors, we w i l l
discuss them i n somewhat greater detail i n the next chapter.
c.
As we discussed earlier, the principal characteristic o f a p-n j u n c t i o n

diode is t h a t i t rectifies: i t permits significant conduction i n one d i r e c t i o n
o n l y . A c c o r d i n g l y , when a diode also begins to conduct i n the other direct i o n , its rectifying properties are lost. I t is well k n o w n that all diodes w i l l
conduct i n the reverse direction i f a sufficiently large reverse voltage is
applied to them. This phenomenon is called junction breakdown. W e shall
now discuss the t w o mechanisms o f j u n c t i o n b r e a k d o w n .
100 na
JUNCTION BREAKDOWN
Diodes of F i n i t e L e n g t h ( N a r r o w Base Diodes)

j
I n deriving the d i s t r i b u t i o n o f injected m i n o r i t y carriers we have assumed t h a t the contact to the y>region is " i n f i n i t e l y " far f r o m the edge o f
the depletion region. I n m a n y i m p o r t a n t cases this assumption is n o t
justified. The p r o x i m i t y o f the contact t o the /(-region w i l l change the
boundary condition and thereby change the distribution o f the injected
m i n o r i t y carriers.
The distribution o f injected carriers i n a case when the separation
W between a contact and the injecting j u n c t i o n is smaller than the d i f f u sion length L (for the case o f a jD-type substrate) is o f paramount i m p o r tance in transistor operation, and is therefore discussed i n detail i n the
next chapter. For the present we shall merely note that in the l i m i t i n g case
I f a high field exists w i t h i n a semiconductor, f o r instance, w i t h i n a reverse-biased depletion region, the covalent bonds between n e i g h b o r i n g
silicon atoms w i l l be distorted as illustrated schematically i n Figure 6.25.
I f the field becomes high enough, some bonds w i l l be " t o r n " apart, resulting i n c o n d u c t i o n electrons and holes: I n the band-diagram representation,
this corresponds to a transition o f a valence electron f r o m the valence b a n d
to the c o n d u c t i o n band. This process, i n w h i c h an electron penetrates
through the energy gap, is called tunneling.
T u n n e l i n g can take place only i f the electric field is very h i g h . E x p e r i mentally i t is f o u n d that the critical field < ?
at w h i c h tunneling becomes
probable, i.e., at which Zener breakdown commences, is approximately
10 v/cm or 100 v / > . "
5
cril
A l t h o u g h Zener breakdown can be observed i n certain p-n j u n c t i o n s ,

. most often another breakdown mechanism precedes i t . This process is
called avalanche breakdown.
b.
Avalanche Breakdown
The avalanche or impact ionization breakdown process is illustrated

schematically i n Figure 6.26. Let us consider this figure step by step. Step 1
Junction Breakdown
193
is the generation o f an electron-hole pair by thermal means, e.g., w i t h the

a i d o f an intermediate level center. I f the field w i t h i n the depletion region
is not too high, this process w i l l simply lead to the regular reverse current.
However, i f the electric field is high enough, the generated electron and
hole w i l l gain enough kinetic energy before colliding with the lattice so
that they w i l l be able to shatter silicon-to-silicon bonds leading to the
formation o f other electron-hole pairs. I n step 2 and 2', respectively, both
the electron and the hole acquire kinetic energy. Step 3 designates the
_
i i|
I I I|
>
io
Zener
1 1
, ^<
10
10
Avalanche
1 4
I II
10'
I
5
I I I i
10'
1
6
! II 1
10"
1 1 1
1
10'
I I I
10'
c (om-'i
Fig. 6.27
s-
The critical field for avalanche and Zener breakdown in silicon. *
impact o f the electron; the fast-moving electron transfers its kinetic

energy to an electron i n the valence band, thereby bringing i t into the.
conduction band. The corresponding process, the impact o f the energetic
hole, is denoted by process 3'. I n processes 3 and 3', two new electron-hole
pairs were created. A l l these electrons and holes now begin to accelerate
i n the high field o f the depletion region, as indicated i n the figure. I n t u r n ,
they w i l l again be able to ionize and create other electron-hole pairs i n a
like manner, and so o n . This process is called the avalanche process.
I f the electric field w i t h i n the depletion region is not high enough, the
carriers cannot acquire sufficient kinetic energy for ionization before
suffering a small collision w i t h the lattice. Thus there is a critical field
<^
at w h i c h the avalanche process w i l l take place.
Critical field values calculated f r o m measured avalanche b r e a k d o w n
voltages i n silicon are shown i n Figure 6.27 as a function o f the i m p u r i t y
concentration o f the substrate o f one-sided step junctions. Also shown is
the critical field for Zener b r e a k d o w n . I t is evident that Zener breakdown
w i l l take place only i n materials i n v o l v i n g high doping concentrations.
Even before breakdown actually takes place, there w i l l be some multiplication o f carriers w i t h i n the depletion region. I f without any breakdown-producing mechanism the reverse current is I
, then the actualc r l t
p-n
194
Junctions
reverse current will be
Junction
Breakdown
I9S
6
IR
MI
(6.84)
RO
where M is the multiplication factor.

B r e a k d o w n o f the j u n c t i o n occurs
when M * co.
I t has-been found that a good empirical representation o f M as a funct i o n o f reverse voltage V for germanium is given by
7
M =
i -
direct experimental observations o f the breakdown voltage o f one-sided

step j u n c t i o n s i n silicon (points). I t is evident that i n all three cases the
breakdown voltage approximately follows the inverse p r o p o r t i o n a l i t y t o
substrate i m p u r i t y concentration predicted by E q u a t i o n 6.87, especially
i n the high-voltage range. Furthermore, i t is evident that a material w i t h
a larger band gap has higher breakdown voltages. This is i n agreement
w i t h what we w o u l d expect on the basis o f the simple picture o f the
(6.85)
(V iBvy
R
where n is usually between 3 and 6, its value depending on the semicond u c t o r a n d the type o f the substrate, and BV is the breakdown voltage o f
the j u n c t i o n . A l t h o u g h this relationship does n o t h o l d well i n the case o f
- s i l i c o n , - i t can -nevertheless be used to illustrate the qualitative features
o f the multiplication process below avalanche breakdown.
c.
Breakdown Voltage of p-n Junctions
W e now calculate the breakdown voltage o f p-n junctions based on the

-condition that breakdown occurs when the m a x i m u m field in the depletion
region < f
reaches the critical field < f . This is only an approximate
criterion that can be improved by taking i n t o account the i n d i v i d u a l
ionization rates of electrons and holes, w h i c h have been determined
experimentally. However, this approximate criterion leads to simple
results which are i n reasonable agreement w i t h experiment.
m a x
crlt
F o r a one-sided step j u n c t i o n , equating the m a x i m u m field w i t h the

critical field results in
<f
m a x
=*= 2
*
= c?
j2K,e V
0
(6.86)
crit
where the reverse voltage V is now, by definition, the breakdown voltage

BV o f the junction. This leads to the f o r m u l a for the breakdown voltage
o f one-sided step junctions:
10 "
10
1 5
10
1 6
10"
C ( c m " )
10'
10
1 9
Fig. 6.28 Breakdown voltage of one-sided step-junctions. Points represent direct

experimental measurements in silicon;' lines were calculated based on ionization rates.
avalanche process discussed above: the carriers must have a higher k i n e t i c

energy to be able t o ionize across a larger band gap, hence higher breakd o w n voltages result.
Proceeding i n a similar fashion for a linearly graded j u n c t i o n and setting
B V =
^ "
2qC
(6.87)
>
qa
leads to the b r e a k d o w n voltaee

3
BV =
!32K d?
s
CTii
(6.89)
9qa
F o r a constant critical field, this shows that the breakdown voltage is
inversely proportional to the substrate i m p u r i t y concentration.
Breakdown -voltages o f one-sided step junctions i n three different
materials are shown i n Figure 6.28. The curves were calculated by using
the experimentally measured i n d i v i d u a l i o n i z a t i o n rates. Also shown are
Thus i n the case o f linearly graded junctions the i m p o r t a n t parameter is

the concentration gradient at the j u n c t i o n ,
dx
p-n Junctions
196
197
Junction Breakdown
both the substrate impurity concentration C and the impurity concent r a t i o n gradient at the j u n c t i o n a. It is evident that as the gradient a is
increased the step-junction limit is approached, whereas for small gradients the linearly graded j u n c t i o n l i m i t holds. The transition point between
the two types o f behavior depends on the substrate impurity concentration.
I n all our discussions o f p-n junctions so far we have consistently treated
w
10'
10
10'
10
2 0
10 '
10
2 2
10"
10"
iilcm" )
Fig. 6.29
Breakdown voltage of linearly graded junctions.
Curves calculated o n the basis o f ionization rates are shown i n Figure 6.29.
I t is evident that the breakdown voltage is approximately p r o p o r t i o n a l to
V l / f l in agreement with Equation 6.89. Thus the simple theory also provides a reasonable description o f the breakdown voltage in linearly graded
junctions.
9
C o m p u t e r calculations o f the breakdown voltage o f diffused p-n j u n c tions, based on ionization rates, are shown i n Figure 6.30 as a function o f
1ol
10'
10
3
L
i ni|
i ii|
I ill
i Im
C(cm~ )
i i iij^
3x10"
10
10'
3 x 10
1 5
10'
3 x 10'
10'
10>
3 x 10'
10'
~i
10'
7
i ill i i ill i
10'
10'
8
n i l I i ||l i
10
10 '
a (cm" )
2
i i l l i i il I l IT
10
10
10
2 2
2 3
2 4
Fig. 6.30
Breakdown voltage of diffused silicon junctionserfc distribution.'
l
10
I
1 5
i l l
10'
C (cm" )
Fig. 6.31
Breakdown voltage of planar silicon one-sided step-junctions showing the
effect of junction curvature.
I
4
10
a p-n j u n c t i o n as i f it was a perfect geometric plane. A planar junction,

i.e., a j u n c t i o n formed by the planar technology, however, is far from being
a plane; as we saw i n Chapter 3, diffusion from an oxide w i n d o w proceeds approximately as far along the surface as into the depth o f the semiconductor. Because o f this, the j u n c t i o n geometry is approximately
circular-cvlindrical near the w i n d o w region, as shown in the inset to
Figure 6.31. The field and potential distributions i n the cylindrical part
o f the p-n j u n c t i o n are different f r o m those in the plane portion o f the
junction.
I n general, the cylindrical part o f the j u n c t i o n constitutes a relatively
small percentage o f the total j u n c t i o n area, and therefore in those characteristics which are cumulative (such as capacitance and current) the error'
p-n Junctions
198
Junction B r e a k d o w n
199
introduced by neglecting these regions is usually relatively small. Breakd o w n , however, will commence at any region, no matter how small, where
the m a x i m u m field reaches the critical field o f the material. I n a planar
p-n j u n c t i o n the fields w i l l be increased i n the cylindrical region because o f
the finite radius of curvature. Thus the field here w i l l reach the critical
field at a lower reverse voltage than that corresponding to the b r e a k d o w n
voltage o f the plane region. Hence, this phenomenon will result i n a
lowering o f the breakdown voltage.
10
Calculations o f the breakdown voltage o f cylindrical p-n j u n c t i o n s ,

based on the critical-field criterion, are shown i n Figure 6.31 w i t h j u n c t i o n
depth x (which is taken to be the radius o f the cylindrical p-n j u n c t i o n )
as parameter. I t is evident that for practical values o f the j u n c t i o n depth
the breakdown voltage is significantly reduced, especially for l o w substrate
-impurity concentrations.
t
The effect o f both substrate concentration and j u n c t i o n depth can be

combined into a dimensionless parameter W*jx, which gives a relative
measure o f the amount o f field c r o w d i n g that occurs in the cylindrical
region o f the j u n c t i o n . This parameter is the ratio o f the depletion region
w i d t h w i t h i n the cylindrical region, at breakdown. W*, to the radius o f
curvature, x . The breakdown theory o f cylindrical p-n junctions can be
rearranged to yield the reduction o f breakdown voltage as a function o f
this dimensionless quantity. This theoretical curve is shown i n Figure 6.32
Fig. 6.33 Idealized charge and electric field distribution at breakdown of diodes constructed by epitaxial techniques.
11
along w i t h numerous experimental d a t a taken on planar p-n j u n c t i o n s

and compared to their plane counterparts.
A n o t h e r b r e a k d o w n voltage l i m i t a t i o n may be encountered i n diodes
constructed by epitaxial techniques such as the one illustrated i n F i g u r e
6.1a. This type o f construction is often employed i n order to m i n i m i z e the
series resistance due to the substrate. I n a structure o f this k i n d , i f the
epitaxial film is relatively t h i n , the reverse-biased depletion region m a y
reach the highly doped substrate. This is called the reach-through c o n d i t i o n , illustrated i n an idealized manner i n Figure 6.33.
A n approximate calculation o f the area under the idealized electric field
distribution, when the m a x i m u m field in the depletion region reaches the
critical field, leads to the reach-through limited b r e a k d o w n v o l t a s e :
B V = g
*
2K*
11
(6.90)
where W
is the w i d t h o f the l o w l y doped region. Because o f the redist r i b u t i o n o f the substrate i m p u r i t y d u r i n g epitaxial g r o w t h and subsequent high temperature heat treatments, H '
w i l l be smaller t h a n the
distance to the metallurgical interface (see Chapter 3).
Calculations o f the reach-through limited b r e a k d o w n voltage based o n
ionization rates are shown in Figure 6.34. N o t e t h a t for the l i m i t o f large
epi
Fig. 6.32 The reduction in breakdown voltage due to the curved shape of p-n junctions
formed by the-planar process. The theoretical curve represents the ratio of breakdown
voltages of cylindrical and plane junctions. The experimental points represent the ratio
of measured breakdown voltage of planar diodes to the corresponding breakdown voltage
o' plane diodes.
e p i
12
p-n Junctions
200
1000
900
800
700 -
201
already flows in this diode at reverse voltages well below the avalanche
breakdown voltage.
The disastrous implications o f this phenomenon on the rectifying
properties o f diodes are self-evident. The mechanisms giving rise to it.
however, are not nearly as clear. I t has been s h o w n that precipitates o f
such metals as copper or iron w i t h i n the silicon crystal can lead to an
increased incidence o f soft j u n c t i o n s ; and it has also been demonstrated
that soft junctions can be "hardened" by a treatment similar to the
predeposition o f phosphorus. (Such a treatment is called gettering.)
It
has been suggested that the excess current is due to localized breakdown
in small high-field regions around metallic precipitates present w i t h i n the
crystal. The treatment w i t h phosphorus is supposed to remove the
metallic precipitates, thereby restoring the hard reverse current-voitagecharacteristics.
40
13
600
500
Transient Behavior
400
>
13
200
100
10'
6.8
2
Impurity concentration in film Icra" J

Fig. 6.34
Reach-through limited breakdown voltage of diodes constructed by epitaxial

1
techniques. -
epitaxial thickness the regular j u n c t i o n breakdown is reached, but that as

the epitaxial film thickness is decreased, the breakdown voltage is also
decreased.
W e conclude our discussion o f breakdown in p-n junctions by consideri n g a very important anomalous phenomenon called soft
breakdown.
for want o f a better name. A " s o f t " reverse current-voltage characteristic
is illustrated in Figure 6.35 i n comparison w i t h the corresponding " h a r d "
b r e a k d o w n characteristic. I t is evident that a large excess reverse current
100
Hard characteristic
0
Fig. 6.3S
20
40
60
Illustration of the reverse current-voltage characteristics of a diode with
"soft" breakdown.
TRANSIENT BEHAVIOR
o d 'u
So far in this chapter we have dealt only with the d-c characteristics o f
p-n junctions. When diodes are used in switching applications, they are
alternately forward and reverse biased. Their performance in such applications is determined by the speed w i t h which they can assume a new bias
condition. Accordingly, we now briefly consider the factors determining
the transient behavior o f p-n junctions. I n particular, we estimate the
time required to t u r n a diode off, i.e., to bring it from a forward-biased
condition into a reverse-biased c o n d i t i o n .
In Figure 6.36a we indicate the d i s t r i b u t i o n o f minority carriers in the
lowly doped region o f a p'n diode under forward bias conditions (t < 0),
and at various times after the bias is reversed. A t / = 0, when the applied
voltage is reversed, the current is also reversed. Initially, the current
flowing in the reverse direction is large because o f the presence o f the
excess minority carriers w i t h i n the n-region. The current continues to
remove these minority carriers, and therefore the concentration o f minority
carriers in the n-region decreases u n t i l it finally reaches the distribution
corresponding to the reverse bias c o n d i t i o n . This process is illustrated i n Figure 6.36b where we show the forward and reverse currents as
a function o f time.
A rough estimate o f the time required for this transient period to take
place can be obtained as follows.. Under forward-bias conditions, there
is a certain charge density per unit area present w i t h i n the -region, due
to the injected minority carriers. This charge density 0
is given by
B
Q^\qpMX
= ^
(6.91)
p-n
202
Junctions
where we have approximated the m i n o r i t y carrier distribution curve by a

triangle o f base X. i n the case o f diodes o f finite length (also called
narrow-base diodes), the actual d i s t r i b u t i o n i n fact follows a straight line
and X is simply given by the separation between the contact and the
injecting j u n c t i o n W . When the contact is infinitely far away, i.e.,
W L , the distribution is exponential and X can be approximated by
the diffusion leneth / _ . .
B
T r a n s i e n t Behavior
-203
and
1/ lr
1
' " ^ - T
\*
h f
-^JK.ave-
forHiL,.
(6.95)
Up
T h u s the turn-off time w i l l depend on the ratio o f f o r w a r d and reverse

currents, w h i c h are determined b y the external c i r c u i t y and o n the
10
I 11
I I|
I I |
M L
I I I
P(0)
10"
(a)
'<*r
*
2
10-
1 111 1
1 .
TO"
* '
*i
(b)
Fig. 6.36 (a) Schematic distribution of minority carriers in the lowly doped region of a
p-'n diode at various times after applied bias is reversed,
(b) The corresponding current-time relationship.
I f the average reverse current flowing

-7
, the turn-off time w i l l be given by
d u r i n g the turn-off period is
/
10"
I ' 'I
10"
Fig. 6.37
1 1 11
- 1
10
The turn-off time oip-n
diodes.
(6.92)
II
Combining Equation 6.91 w i t h 6.92 leads to the estimate o f the turn-off
time, ..
(6.93)
VIi ? . a v e ' lOTi

-"p
J
or
loir
1 1 11
10
MM
K s v e
(off
1
J
for W L
B
(6.94)
characteristic t i m e constant o f the diode. This characteristic time constant

is the lifetime o f m i n o r i t y carriers i f the contact is infinitely far away, and
i t is the diffusion time constant W y / D i f the contact is much closer t h a n
a diffusion length.
p
T h i s simple analysis gives only a rather p r i m i t i v e description o f the

transient behavior o f the diode. The turn-off time can be calculated i n a
t The external circuit actually determines the maximum value of the reverse current
rather than its average.
p-n Junctions
202
where we have approximated the m i n o r i t y carrier distribution curve by a

triangle o f base X. i n the case o f diodes o f finite length (also called
narrow-base diodes), the actual d i s t r i b u t i o n i n fact follows a straight line
and X is simply given by the separation between the contact and the
injecting j u n c t i o n W . When the contact is infinitely far away, i.e.,
W L , the distribution is exponential and X can be approximated by
the diffusion leneth / _ . .
B
T r a n s i e n t Behavior
-203
and
'
o 7 ~ ^ I n
^rW <[<L .
B
(6.95)
T h u s the turn-off time w i l l depend on the ratio o f f o r w a r d and reverse

currents, w h i c h are determined b y the external c i r c u i t , !
d
the
a
I I 11
I M |
I i|
l | i _
1 II
A < 0
P(0)
fir
10"
(a)
It
*
'</r
t, I ; h U '<
1CT
1 111 1
1 .
-icr
]
+
e "
*i
(b)
Fig. 6.36 (a) Schematic distribution of minority carriers in the lowly doped region of a
p-*n diode at various times after applied bias is reversed,
(b) The corresponding current-time relationship.
/
I f the average reverse current flowing

~I
, the turn-off time w i l l be given by
R
d u r i n g the turn-off period is
icr
II I I
io
1 1 11
- 2
io
1
_ i
1 1 11
io
a v e
Fig. 6.37
r o ( r
flafk.
'
(6
-"JJ.ave
Combining Equation 6.91 w i t h 6.92 leads to the estimate o f the turn-off

time, ..
(6.93)
V7
/ i n
or
loir
for W L
B
The turn-off time of p-n diodes.
.92)
(6.94)
characteristic t i m e constant o f the diode. This characteristic time constant

is the lifetime o f m i n o r i t y carriers i f the contact is infinitely far away, and
i t is the diffusion time constant W ID
i f the contact is much closer t h a n
a diffusion length.
2
T h i s simple analysis gives only a rather p r i m i t i v e description o f the

transient behavior o f the diode. T h e turn-off time can be calculated i n a
t The external circuit actually determines the maximum value of the reverse current
rather than its average.
p-n Junctions
204
more exact manner by solving the time-dependent diffusion problem

describing the d i s t r i b u t i o n o f m i n o r i t y carriers. T h i s has been done i n
b o t h the general case and its two limits corresponding to wide- and
narrow-base diodes.
Some results o f such calculations are shown in
Figure 6.37 w h i c h gives the turn-off time as a function o f the r a t i o o f the
f o r w a r d current to the m a x i m u m reverse current for the wide- and narrowbase limits as well as for the intermediate case o f W = L .
14
Problems
14. R . H . Kingston. "Switching Time in Junction Diodes and Junction Transistors "
Proc. IRE. 42, 829 M954); M . Byczkowski and J . R . Madigan. "Minority Carrier
Lifetime in P-N Junction Devices," J. Appl. Phys., 28, 878 (1957); A . S. Grove and
C . T . Sah. "Simple Analytical Approximations to the Switching Time in Narrow
Base Diodes;" Solid-State Electron., 7, 107 (1964).
. . .
6.1
REFERENCES
CITED
1. H . Lawrence and R . M . Warner, "Diffused Junction Depletion Layer Calculations,"

Bell System Tech. J., 39, 389 (1960).
4. A more detailed discussion of breakdown mechanisms in p-n junctions is given in

Chanters 11 and 12, J . L . Moil, Physics of Semiconductors, McGraw-Hill Book C o . ,
1964.
6.2
6.4
Derive an expression for the width of the zero-bias depletion region resulting if
a contact is formed between a metal and an n-type semiconductor. Assume that
at the metal-semiconductor interface the conduction band of thesemiconductorwiil
be fixed at an energy <t> above the Fermi-level of the metal where <D > ( E )
in the bulk of the semiconductor. Compare with the expression for a one-sidec*
step junction. Discuss what happens if c> < (E , ) in the bulk of the semiconductor.
F
6.5
A diode is made as follows:

Starting material: 0.175 O cm n-type silicon.
Boron predeposition: results in Q = 10" boron atoms/cm .
Drive-in diffusion: 1 hour at 1200-C.
(a) If the junction area of the diode is 10~ c m , calculate its capacitance at zero
and at 10 v reverse bias, using the one-sided step junction and the linearly
graded junction approximations. (Is the latter meaningful?) Which approximation is besr in each case and why?
(b) Define a voltage range where one or the other approximation can be expected
to lead to good results.
2
9. D . P. Kennedy and R . R . O'Brien, "Avalanche Breakdown Characteristics of a

Diffused P-N Junction." IRE Trans. Electron Devices ED-9, 478 (1962).
10. H . L . Armstrong, " A Theory of Voltage Breakdown of Cylindrical P-N Junctions,
with Applications." IRE Trans. Electron Devices ED-4, 15 (1957).
13. A . Goetzberger and W. Shockley, "Metal-Precipitates in Silicon P-N Junctions,"

/. Appl. Phys., 31, 1821 (1960).
Using the potential distribution based on the depletion approximation, check the
validity of the depletion approximation by estimating the width of the region
where the carrier concentration is not negligible. Compare this width with the
total width of the depletion region under various bias conditions.
8. S. M . Sze and G . Gibbons. "Avalanche Breakdown Voltages of Abrupt and Linearly

Graded P-N Junctions in Ge, Si, G a A s and GaP," Appl. Phys. Lett., 8, 111 (1966).
12. R . J . Whittier, unpublished.
6.3 Derive expressions for the electric held, the potential distribution, the built-in
voltage, and the depletion region width for a linearly graded junction in equilibrium
7. S. L . Miller, "Avalanche Breakdown in Germanium," Phys. Rev., 99, 1234 (1955).
11. O. Leistiko and A . S. Grove, "Breakdown Voltage of Planar Silicon Junctions,"

Solid-State Electron., 9, 847 (1966).
(a) Using the potential distribution based on the depletion approximation, derive
the electron and hole distributions as a function of distance within the depletion
region of a one-sided step junction, in equilibrium.
(b) Derive expressions for the electron and hole diffusion and drift fluxes crossing
the plane where , - E , in equilibrium.
r
(c) Calculate the current due to each of the above flux components if the bulk
impurity concentration is 10" c m - ' , and the junction area is 1 0 c m . Compare these magnitudes with the forward current data shown in Figure 6.2 and
discuss.
-3
5. A . G . Chynoweth. W. L . Feldmann, C . A . Lee, R . A . Logan, G . L . Pearson, and

P. Aigrai'n. "Internal Field Emission at Narrow Silicon and Germanium P-N
Junctions," Phys. Rev.. 118, 425 (1960).
6. S. L . Miller. "Ionization Rates for Holes and Electrons in Silicon," Phys. Rev., 105,
1246 11957).
PROBLEMS
2. J . Hilibrand and R . D . Gold, "Determination of the Impurity Distribution in

Junction Diodes from Capacitance-Voltage Measurements," RCA Rev., 21, 245
(1960).
.3. The theory of the current-voltage characteristics of p-n junctions was established
bv W. Shockley. "The Theory of P-N Junctions in Semiconductors and P-N Junction
Transistors," Bell Svstem Tech. J., 28, 435 (1949). This theory was then extended by
C . T . Sah. R . N . Noyce, and W. Shockley, "Carrier Generation and Recombination
in P-N Junctions and P-N Junction Characteristics," Proc. IRE, 45, 1228 (1957).
A review of "The Evolution of the Theory for the Voltage-Current Characteristics
of P-N Junctions" is given by J . L . Moll, Proc. IRE. 46. 1076 (1958).
20S
6.6
Assuming that both electrons and holes within a reverse-biased depletion region
move with a constant maximum drift velocity, o i = 10 cm/sec, construct the
electron and hole distribution corresponding to Figure 6.18. Assume that the
bulk impurity concentration is 10" cm"' and that the lifetime is 1 usee. Do this:
(a) For a silicon diode at roomjemperature.
(b) For a germanium diode at room temperature.
7
l m
p-n
206
Junctions
6.7
Derive and calculate the small-signal conductance dl\dV

at V = JO v for the
diodes of the previous problem, at room temperature. Comparewith the data
shown in Figure 6.19b.
6.8
Using the potential distribution based on the depletion approximation, derive the
electron and hole concentration distribution within the space-charge region of a
one-sided step-junction under small forward bias. Compare your result with the
calculations shown in Figure 6.22.
6.9
Derive an expression giving forward current of a diode as a function of the difference

of the applied voltage and the built-in voltage of the junction.
6.10 By equating the expressions giving the electron and hole fluxes in terms of the
gradients of the respective quasi-Fermi levels to the formula giving the forward
current, arrive at an order-of-magnitude justification of the assumption of quasi
equilibrium.
6.11 Express the ratio of the diffusion current component to the recombination current
component as a function of forward current instead of forward voltage. Using
this expression, compare diodes made of germanium, silicon, and gallium arsenide
with respect to the relative importance of the two components. Compare your
conclusion with the data shown in Figure 6.24.
6.12 Derive the small-signal conductance dl \d\'
of a p-n junction diode, considering
both components of the forward current. Calculate its value for the diodes
-described in Problem 6.6 at Vj 0, and at V = 0.4 v. at room temperature. Compare with the experimental measurements shown in Figure 6.24. Also, compare
with the results of Problem 6.7.
r
TABLE
IMPORTANT
FORMULAS
6.1
F O R O N E - S I D E D STEP
Built-in voltage
JUNCTIONS
kT
C
<t>B = 2 In
B
n,
llK,* [4>
0
Depletion region width

^
max
Capacitance per unit area
: f o r w a r d ]
C -
\Vj\\
lC
+ : reverse j
where
M a x i m u m electric field
bias
...
K-
(t
Reverse current
6.13 Derive the forward voltage as a function of temperature at a given current densitv.
Comment on the use of a p-n junction as a thermometer.
'pen = \q
7 WAj
6.14 The difference in the breakdown voltage of those planar and plane diffused p-n
junctions which are well described bv the linearly graded junction approximation
is much less than the difference in cases where the junctions follow the one-sided
step-junction approximation. Give a simple explanation.
F o r w a r d current
6.15 Based on the simple picture of the avalanche process given in the text, propose
a qualitative argument giving the sign of the temperature dependence of the
avalanche breakdown voltage.
6.16 Derive an expression giving the time required to turn a diode on, i.e., to brine it
into forward bias from a reverse-bias condition.
IF
Auff-
A-ec + ^diff
-qD^-e"-r\lkT
(~ L
B
A v a l a n c h e breakdown
voltage
ny
<
'^' J ^"it
2qC
B
PRINCIPLES
ACTION
OF TRANSISTOR
C U R R E N T S F L O W I N G IN A
TRANSISTOR; C U R R E N T GAIN
LIMITATIONS AND MODIFICATIONS

O F T H E SIMPLE T H E O R Y
BASE
Principles of T r a n s i s t o r Action
209
RESISTANCE
MAXIMUM
MINIMUM
VOLTAGES
VOLTAGES
THERMAL
LIMITATIONS
w i t h i n each o f the three regions o f the transistor. Where the concentration

variation or the characteristic shape o f planar j u n c t i o n s leads to significant
deviations, we modify our discussion accordingly.
We begin w i t h a qualitative discussion o f the principles o f transistor
action. We then consider the various current components which flow
w i t h i n a transistor and the factors influencing the current gain by using a
A,
IO- cm-
7
The single most important solid-state device is the j u n c t i o n transistor.

Its invention brought about an unprecedented g r o w t h o f research and
development w o r k in solid-state physics and engineering. I t is the active
device on which most discrete and integrated solid-state circuits are based.
I n this chapter we examine the basic characteristics o f j u n c t i o n transistors.
A planar silicon pnp transistor is illustrated schematically in Figure 7.1a.
The fabrication o f such a transistor begins the same way as the fabrication
o f planar diodes: usually a lightly doped epitaxial film is grown upon a
heavily doped substrate of the same type, the surface o f the wafer is
oxidized, windows are opened in the oxide, and an i m p u r i t y is permitted
to diffuse t h r o u g h these windows to form a j u n c t i o n w i t h i n the epitaxial
f i l m . A t this point, however, the surface o f the silicon is oxidized again,
windows are opened at different places, and another i m p u r i t y , o f the same
type as the substrate, diffuses into the silicon to form a second j u n c t i o n .
W i n d o w s are then opened and metallic contacts are deposited upon both
diffused regions. The wafer is then cut apart by scribing, leads are
attached to all three regions o f the transistor, and the transistor is packaged.
F o r simplicity, we study an idealized one-dimensional model o f this
planar transistor analogous to the idealized model o f the p-n j u n c t i o n
diode o f the previous chapter. This one-dimensional model, shown i n
Figure 7.1b, can be considered as a section of the planar transistor along
the dashed lines. It neglects the variation o f the i m p u r i t y concentration
208
(b)
Fig. 7.1 (a) A typical planar pnp transistor. Representative impurity concentrations
and dimensions are indicated.
(b) Idealized one-dimensional transistor model.
fjj): Emitter
: Base
: Collector
simple theory. The various ways this theory must be modified undercertain conditions are discussed next. T h e n we consider base resistance,
the m a x i m u m and m i n i m u m voltage limitations o f transistors, and thermal
limitations.
7.1
PRINCIPLES O F T R A N S I S T O R
ACTION
The energy bands pertaining to the idealized pnp transistor o f Figure

7.1b are shown in Figure 7.2. Figure 7.2a shows the energy bands under
e q u i l i b r i u m conditions, i.e.,^when a l l three leads labeled (E), (B), a n d ( C ) , .
210
for emitter, base, and collector, are connected together. U n d e r such

conditions, as we saw i n Chapter 6. the electrostatic potential varies from
each semiconductor region to the next in such a way as to balance out the
diffusion flux of electrons and holes due to their concentration gradients.
When external voltages are applied to the various regions o f the
transistor, the junctions may become f o r w a r d or reverse biased. The
Emitter
Base
Collector
Principles of Transistor
Action
j u n c t i o n w i l l n o w flow i n it. This is the p r i n c i p a l feature o f transistor

action: a large current flows in a reverse-biased junction due to the existence
of a forward-biased
junction in its vicinity.
I t should be realized that not all injected holes w i l l reach the collectorbase depletion region but that some w i l l recombine w i t h electrons en
route t h r o u g h the base. Also, when the emitter-base j u n c t i o n is f o r w a r d
biased, some electrons w i l l be injected into the emitter along w i t h the
injection o f holes i n t o the base. Finally, electrons and holes w i l l recombine
i n the space-charge region o f the emitter-base j u n c t i o n . D u e to these
processes, electrons w i l l flow into the base t h r o u g h the base lead. T h i s
electron flow, o f course, corresponds to a current flowing out the base lead
as indicated i n Figure 7.3.
p*
Emitter
n
Base
P
Collector
lc
y BE = - o . 5 v
Fig. 7.3
Currents flowing In a pnp transistor.
As we have discussed above, the total emitter current I consists o f

those holes w h i c h reach the collector, l . and o f the electrons which flow
into the transistor t h r o u g h the base lead, J . Thus,
E
V - / C
Fig. 7.2 Energy band diagram for a pop transistor.

(a) Equilibrium case.
(b) Under bias.
-energy bands under the most i m p o r t a n t type o f bias condition are shown
i n Figure 7.2b. U n d e r this c o n d i t i o n , the j u n c t i o n between emitter and
base is forward biased and the j u n c t i o n between base and collector is
reverse biased.
D u e to the forward biasing o f the emitter-base j u n c t i o n , large numbers
o f holes w i l l now be injected i n t o the n-type base. I f the two junctions are
sufficiently close to each other, most o f these holes will reach the collectorbase j u n c t i o n where they w i l l be swept across by the electric field. Thus
- t h e y w i l l be collected in the />-type collector. I t should be noted that even
though the collector-base j u n c t i o n is reverse biased, a large current which
approximately equals the forward-biased current o f the emitter-base
/J.
(7.1)
T w o quantities o f great importance i n the characterization o f transistors

are the so-called common-base current gain a, also referred to as h ,
defined by
F B
o. = h
and the so-called common-emitter
defined by
f,
(7.2)
current gain (3, also referred to as h
fi = h
f.
F E
(7.3)
I t is evident from E q u a t i o n 7.1 that these are related to each other by

? =
.
(7.4)
1 R
U n d e r n o r m a l operating conditions such as we discussed above, cx w i l l
r-nnciples of T r a n s i s t o r
Action
213
60 times larger. The relevant small-signal
current gain h
ft
is defined by
dl
h,.
(7.5)
dK
I t can be readily shown from the two definitions, Equations 7 3 and 7 5

that the relationship between the small-signal current gain h and the d-c'
current gain h
is given by
F
FE
lq
h
dl
F E
Thus, for a transistor whose current

collector current. h = h .
(.7.6)
dh FE
c
ain h
is independent o f the.
'
F E
We now consider the behavior o f our transistor in a simple circuit shown

in Figure 7.5a. F o r a given input voltage V , a certain d-c base current I
BB
Emitter
lai
F i g . 7.4
H '
Base
Collector
ibi
C u r r e n t - v o l t a g e c h a r a c t e r i s t i c s of a s i l i c o n pno t r a n s i s t o r .
for i l l u s t r a t i o n t h r o u g h o u t this chapter.
Input
circuit
T h i s d e v i c e is used
OutDUt
circuit
Its s t r u c t u r a l p a r a m e t e r s a r e a p p r o x i m a t e l y as
applied
I n d i c a t e d i n Fig. 7.1a.
(a)
C o m m o n base c o n f i g u r a t i o n .
(b)
Common emitter configuration.
l.ll
always be smaller than unity, although its value in a good transistor w i l l

approach u n i t y very closely. Accordingly, /? will be large in a good
transistor.
The current-voltage characteristics o f a planar silicon pnp transistor
whose structural parameters are approximately as indicated i n Figure 7.1a
are shown i n Figure 7.4. T h i s device w i l l be used for illustration t h r o u g h o u t
this chapter. The characteristics are shown in two biasing configurations.
I n one, the voltages are applied w i t h respect to the base. This is referred
to as the common-base mode. I n the other configuration, the voltages are
applied w i t h respect to the emitter. This is referred to as the commonemitter mode.
I t is evident f r o m Figure 7.4a that a is i n fact very close to unity, a n d
f r o m Figure 7.4b that p an 60 at I = 4 ma, at low values o f
V .
Focusing our attention on the common-emitter configuration, we can also
see that a change i n I brings about a change in I which is also about
1
CE
6(
\r
O
u
T-
3 2
1
0
0.1 r 0.08
0.06
3 0.04
S 0.02
m
0
g
S
A /
c =
/AU
h [.
FE
i.
Ibl
F i g . 7.S
I l l u s t r a t i o n o f t h e use o f t h e t r a n s i s t o r as an a m p l i f i e r .
Junction
214
Transistors
and collector current I flow i n the transistor. I f a small a-c signal is

n o w superimposed on the i n p u t voltage, the base current w i l l vary as a
function of time as illustrated i n Figure 7.5b. This v a r i a t i o n , i n t u r n ,
brines about a corresponding a-c v a r i a t i o n i n the output current I w h i c h
however is h times larger t h a n the i n p u t current v a r i a t i o n . T h u s the
transistor amplifies the i n p u t signal.
c
Currents
Flowing in a Transistor; C u r r e n t
Gain
215
can separate the electron current at x = 0 i n t o t w o components: c o m ponent 1 is due to electrons which are injected i n t o the />-type emitter
region, w h i l e component 2 is due to electrons which are injected i n t o the
emitter-base space-charge region where they recombine w i t h holes. A s
we n o w proceed t o w a r d the collector region, the fraction o f the current
carried by holes decreases because some o f the holes injected to the n-type
A transistor can also be used as a switching device. By c o n t r o l l i n g a

small current (the base current), a larger one (the collector current) can
be turned on and off.
H'r.
-w
- . H i *
The first of these applications, as an amplifier, involves small-signal a-c

phenomena. The second, as a switch, involves large-signal transient
phenomena. However, b o t h o f these are consequences o f the d-c characteristics and limitations o f the j u n c t i o n transistor, to which we devote the
-remainder of this chapter.
o*-
o-
1m
A t this point we should define the established terminology o f j u n c t i o n

transistor parameters.
Since the transistor is a three-terminal device,
currents and voltages are often specified w i t h three-letter subscripts. The
first two o f these letters designate the t w o terminals between w h i c h the
current or voltage is measured. The t h i r d letter designates the state of
the t h i r d terminal w i t h respect to the second. F o r example,
BV
designates the breakdown voltage between collector and base w i t h the
-emitter-base j u n c t i o n open. BV
designates the breakdown voltage
between collector and emitter w i t h base shoned to emitter. 1 BO > d ICES
designate the corresponding leakage currents, and so on.
CB0
CES
ar
7.2
CURRENTS FLOWING
CURRENT GAIN
IN A
TRANSISTOR;
Neutral E-B
emitter spacecharge
region
Fig. 7.6
Neutral
base
C-B
soacecharge
region
Neutral
collector
The variation of hole current as a function of position tor a pnp transistor.
base recombine w i t h electrons. The fraction o f hole current consumed by

r e c o m b i n a t i o n i n the neutral base region is designated
a.
Current Components
as the c u r r e n t
component 3. T h e sum o f the three components is the base current.
The various current components involved i n transistor action are

illustrated in Figure 7.6 where the current carried by holes l is shown as a
function o f the distance f r o m the emitter t h r o u g h the base to the collector
region, for a pnp transistor. I n the p* emitter most of the current is
carried by holes so that the hole current l equals the total emitter current
I.
As we proceed t o w a r d the collector, an increasing fraction o f the total
current is carried by electrons.
v
T h e fraction o f the total current carried by holes w i l l not chanse-any

more
beyond
the
base-collector
depletion-region boundary.
This
is
because i n a reverse-biased depletion region the recombination process is

negligible.
Similarly, in the neutral collector region the hole current is a
m a j o r i t y carrier current and it is n o t diminished by r e c o m b i n a t i o n .
Let us n o w consider these current components i n a quantitative manner."
Let us consider the boundary between the emitter-base depletion region

a n d the n-type base, designated by the plane x = 0. A t this boundary, we
show the fraction o f the emitter current due to holes diffusing into the
base Iam.B- The rest o f the emitter current is carried by electrons. We
I n low-level injection, the transport o f the m i n o r i t y carriers injected i n t o

the base can be described entirely by diffusion as we saw i n Chapter 5. T o
o b t a i n their d i s t r i b u t i o n , we must solve the steadv-state diffusion e q u a t i o n :
_
d~v.
o. D--
^ - u . . c . i ^ riowing i n a
subject to the boundary conditions
I ransistor; Currenc Gain
|7
s e . W J J - I n good transistors W L , , hence we can approximate the

distribution o f m i n o r i t y carriers within the base bv the straight-line
distribution. Equation 7.11. This yields
b a
P(0) = p . e '
7.8)
i/*r
and
(7.9)
Amr.B
where the e q u i l i b r i u m m i n o r i t y carrier concentration w i t h i n the base is

iV
denoting the d o n o r concentration w i t h i n the base
qD
(7.12)
DB''B
nR
region (assumed to be u n i f o r m in our idealized model). T h e first o f these

conditions states that the concentration o f the m i n o r i t y carriers at the edge
o f the emitter-base j u n c t i o n space-charge region is increased above their
y
lkT
e q u i l i b r i u m value by the exponential factor tf " , as discussed in
Chapter 6. The second boundary c o n d i t i o n states that the concentration
of m i n o r i t y carriers at the edge o f the base-collector j u n c t i o n space-charge
region w i l l be zero. This is because this j u n c t i o n is reverse biased; the
electric field w i l l immediately sweep across any m i n o r i t y carriers that
arrive there.
where D denotes the diffusivity o f holes in the base region. The diffusion
current o f electrons injected into the emitter (component 1 o f Figure 7.6)
is given by
B
UM,E
n~
_ ,
qD
(7.13)
nE
A.*
where we have assumed that the emitter depth W is much smaller t h a n

the diffusion length o f electrons in the emitter region. N
denotes the
E
AE
T h e s o l u t i o n is
1.0 E-
sinh
sinh
Pn(*)
Pno
[P(0) -
P ]
0
sinh
sinh
W
sinh
=
for V
P(0)
EB
kTlq.
(7.10)
sinh
Calculations based on this s o l u t i o n are shown in Figure 7.7 where the
normalized m i n o r i t y carrier concentration is shown as a f u n c t i o n o f distance for a fixed diffusion length L = 10 p. and various values o f the base
w i d t h W . I t is evident that for W L the d i s t r i b u t i o n approaches the
simple exponential d i s t r i b u t i o n discussed in the previous chapter, E q u a t i o n
6.57. I n the other extreme, where W L, the d i s t r i b u t i o n approaches
the simple straight-line f o r m given by
B
P<*)-P.(0)(l-^J.
T h i s l i m i t i n g straight-line d i s t r i b u t i o n
connection w i t h diodes o f finite length.
W e can now write d o w n the various
current I . By referring to the plane x
w i l l , first o f all, consist o f the diffusion
E
was discussed i n Chapter
(7.11)
Fig. 7.7
width.
10
12
14
16
18
20
xliO
The distribution of injected minority carriers for various values of the base
acceptor concentration i n the emitter, assumed to be u n i f o r m i n our

idealized model; and D
is the diffusivity o f electrons i n the emitter.
Finally, the current carried by electrons injected into the emitter-base
space-charge region where they recombine w i t h holes (current component
2 i n Figure 7.6) is given by
_ , \JCC ?nE
6 in
l2kT
/rec = . * W e^" Aj
EB
components o f the total emitter

= 0 in Figure 7.6, we see that it
current of holes injected into the
where W
junction.
EB
(7.14)
is the w i d t h o f the space-charge region o f the emitter-base
'
ii
218
A n efficient emitter is one i n which the components 1 and 2 o f Figure

7.6 are small. Thus we w i l l define the emitter efficiency by
fraction o f the injected current that does reach the collector-base depletion
region is called the transport factor and is defined by
Diffusion current i n base

y =
_ Hole current reaching collector
T o t a l emitter current
H o l e current injected i n t o base
or
I il!,B
(7.15)
I diff.B + 1diff.E
Using Equations 7.12 to 7.14, this becomes
_
1
1 +
(7.16)
D,
Wink.
in,e
for the case o f apnp transistor. I n Figure 7.6, this v a r i a t i o n o f hole current
across the base region is indicated by the decrease o f hole current m a r k e d
by 3.
The transport, factor can be calculated by using the solution o f the
m i n o r i t y carrier distribution w i t h i n the base region, E q u a t i o n 7.10. By
the definition o f the transport factor,
dp.
QVjsa/ikT
dz
which can be rearranged to vield

(7.17)
qBA.
2
x=W
(7.19)
ay*
219
C u r r e n t s F l o w i n g in a T r a n s i s t o r ; C u r r e n t G a i n
dx
Using E q u a t i o n 7.10, this can be shown to lead t o
Jdltf.B
where
1
B
*r =
cosh
D,
xY
L
v
=
R
ÂE E
(7.18)
D TlE
(7.20)
where L
is the diffusion length o f m i n o r i t y carriers in the base region.
F o r good transistors, W is m u c h smaller t h a n L . Therefore, to a
good a p p r o x i m a t i o n ,
v
WEB
The first o f these three quantities is the base factor which depends on the
total number o f impurities i n the base region. The quantity , the emitter
factor, depends on the t o t a l number o f impurities in the emitter region.
Finally, the quantity R, the recombination factor, is characteristic of the
.recombination rate i n the emitter-base j u n c t i o n space-charge region.t
These and other i m p o r t a n t formulas relevant to junction transistors are
.summarized in Table 7.1 at the end o f this chapter.
By the definition o f the emitter efficiency, the current carried by minority
carriers injected i n t o the base is given by yI .
W h a t is o f principal^
importance is the fraction o f this current which reaches the collector-base
depletion region and is collected there. The quantity designating the
'
D-C C u r r e n t Gain
As we have discussed, the t w o figures-of-merit w h i c h are used to
describe the performance o f transistors in the common-base and c o m m o n emitter configurations are the current gains a and /?, respectivelv.
By its definition,
a s
FB
= = ya .t
T
(7.22)
t The relationship a = l \ l = ya. holds only if the reverse-bias leakage current of

the collector-base junction l
is negligible. I f this is not the case, the relationship
becomes
c
R =
!
W
"o
A.
+ s.
Aj
wnere s = <ri;,A and A, is the depleted surface area.
c s o
t This treatment neglects surface recombination. As discussed in Chapter 10, its effect
on y can be approximated by setting
oc m y a
Correspondingly,
220
C u r r e n t s Flowing in a Transistor; C u r r e n t G a i n
A c c o r d i n g l y , the c o m m o n emitter current gain will be given by
factor x - 1; thus / , = f .
I t is evident that in the absence o f
recombination w i t h i n the emitter-base space-charse region (R = 0), the
current gain is independent of the collector current." The larger the
recombination rate R, the more the current gain drops at low currenr
levels.
T
ya.
P =
FE
(7.23)
7*2
1 -
Thus i t is clear that we w o u l d like the product o f the transport factor

OL and the e m i t t e r efficiency y to approach unity as closely as possible.
221
1 acr
0.2
0.3
0.4
0.5
10
100 ma
= 0
10
10
x
y/
is
_
10"
y
1 ma
^y
y
/
\&y
y
7
100 u a b -
\vy
10~
TO"
i
10
y
- 5
10
- 4
mahr
y
3
.r
\
:
10" 10"
10"
J (amp/cm )
i
10
10
10
10ua
Fig. 7.8 Calculated common-emitter current gain as a function of collector current

density for various space-charge region recombination rates, for a silicon transistor.
Recombination in neutral base region is neglected, i.e., a , I. ( = S X I 0 sec/cm- ;
B = 1 X I 0 sec/cm )
M
1 /ial
l s
100 na
F o r a transistor w i t h a large p\ we can write

10 na
1
N W
DB
nE
N W
AE
Injection
into
emitter
W lr
EB
iVEBi*kT
1 na
Recombination
w i t h i n E-B
space-charge
region
100 pa
(7.24)
0.1
0.6
0.7
0.8
0.9
vzb (v)
Fig. 7.9 Collector and base currents as a function of emitter-base forward bias for the
pnp transistor. (V = 0)
CB
where the o r i g i n o f each o f the three terms is indicated.

I n many applications i t is i m p o r t a n t not only that /3 be large but also
that i t should not v a r y w i t h the current level. The nature o f the variation
o f the current gain w i t h collector current is illustrated i n Figure 7.8. Here
calculations based o n Equations 7.17 and 7.23 are shown. These calculations were made by using emitter and base factors a p p r o x i m a t i n g those o f
the pnp transistors o f Figure 7.1, for various values o f the recombination
parameter, R. I n these calculations we also assumed that the transport
Experimental measurements of the collector and base currents o f the

pnp transistor are shown i n Figure 7.9 as a function o f emitter-base
forward bias. Several features should be noted i n this figure. First, the
collector current follows the voltage dependence predicted for the injected
current, E q u a t i o n 7.12, over eight decades o f current. This agreement is an
indication o f the validity o f the injected carrier concentration formulasgiven i n Chapter 6. O n the other hand, the base current as a function o f
222
emiuer-base forward voltage follows

r
I -cc
!mkT
e" "
w i t h m ^ 1.7 in the lower current range. This is an indication o f the fact

that most o f the base current at l o w current levels is due to recombination
i n the emitter-base space-charge region.
Limitations and Modifications of the Simple T h e o r y
223
I f a transistor is to be useful, i t is expected that u p o n a change i n

emitter-base f o r w a r d bias the coDector current w i l l change also. I n order
to change the collector current, the m i n o r i t y carrier d i s t r i b u t i o n i n the
base must be altered as illustrated i n Figure 7.11. W e w i l l estimate the
time required for such a rearrangement by calculating the time needed for
holes to travel across the base region. (This calculation is a good example
of similar transit-time calculations in other physical systems.)
100
Emitter
Fig. 7.11
100pa
1na
10na
100na
Distance
The small-signal variation of minority carrier distribution in the base region.
1jta 10;ia 100//a Ima 10ma lOOma 1a
'c
Fig. 7.10
Collector
Common-emitter current gain versus collector current for the pnp transistor.
The distance traveled by a hole i n a lime T s j g r y e n . b y

dx = v(xJaT~
- " T h e common emitter current gain for the same device is shown i n Figure
7.10 as a function o f collector current. N o t e that the variation o f h
w i t h collector current is quite similar to the theoretical calculations shown
i n Figure 7.8. except for the decrease o f h
at high current levels. W h e n
the emitter j u n c t i o n area, ~ 1 0 c m , is taken i n t o account, a reasonable
fit is obtained between theory and experiment for 7^ ^ 10 cm/sec.
FE
(7.25)
where v(x) is the velocity o f a hole. Thu's-the^ransfr-time for holes across

the base region o f a pnp transistor w i l l be given by
dx
(7.26)
FB
- 4
A l s o shown in this figure is the experimentally measured small-signal

current gain h o f this device.
jc
7.3
a.
d(z)
LIMITATIONS AND MODIFICATIONS

OF T H E SIMPLE T H E O R Y
Transit-Time Limitation
A l t h o u g h i t is beyond the scope o f this b o o k to discuss the frequency

response o f transistors, we shall develop a simple estimate of the m a x i m u m
frequency up to which a transistor can be expected to be useful. This
l i m i t a t i o n is provided by the time required for the rearrangement o f
m i n o r i t y carriers i n the base region.
The velocity o f the holes is related to the hole current and the hole
distribution i n the base by
/, = qu{x)p(x)Aj.
(7.27)
Using a straight-line hole d i s t r i b u t i o n , we can then readily show that the
transit time across the base w i l l be
w
(7.28)
2D vB
The frequency l i m i t a t i o n corresponding to this time l i m i t is given a p p r o x i mately by the reciprocal o f the transit t i m e . | E x p e r i m e n t a l measurements
of the small-signal common-emitter current gain h o f the pnp transistor
as a function o f frequency are shown in Figure 7-.12. N o t e that h drops
u
t More rigorous considerations yield i/2-n!,, for this frequency limitation.
224
L i m i t a t i o n s and Modifications of the Simple T h e o r y

I
below unitv at a frequency o f the same order o f magnitude as given by the
225
above c r i t e r i o n .
It is interesting to compare the transit time with the base transport
factor x given earlier. I t is evident from Equations 7.21 and 7.28 that
the base transport factor is related to the ratio o f the transit time across
the base to the lifetime by
a
= l - - .
(7.29)
T
Collector
10*
10
1 5
10
1 2 1 4 1 6 1 8
Fig. 7.13 The impurity distribution in the pnp transistor.

diffusion conditions.)
20
(Estimated on the basis of
Physically, this is very suggestivethe probability o f recombination o f

an injected hole i n the base region w i l l indeed depend on this ratio.
Emitter
b.
i Base \
(epitaxial film)
G r a d e d B a s e Regions
So far we have based o u r discussion on the idealized transistor model

i n w h i c h the d i s t r i b u t i o n o f the impurities in the base region was assumed
to be u n i f o r m . I n double-diffused transistors, the distribution o f impurities
i n the base region is not u n i f o r m , but is quite strongly graded. The
estimated d i s t r i b u t i o n o f impurities in our pnp transistor is shown i n
Figure 7.13. N o t e the very significant variation o f i m p u r i t y concentration
across the base region. T h i s i m p u r i t y concentration profile w i l l b r i n g about
a similar v a r i a t i o n o f m a j o r i t y carrier concentration i n the base region.
However, i n e q u i l i b r i u m no current flows so that an electric field must
exist i n the neutral base region. This electric field w i l l counterbalance
the diffusion current due to the concentration gradient o f m a j o r i t y carriers
existing there. (This is also evident f r o m the corresponding band diagram
shown i n Figure 7.14.) I f m i n o r i t y carriers are now injected into the base,
their m o t i o n w i l l be affected by the finite electric field which is present i n
the neutral base region.
Collector
Fig. 7.14
(substrate)
The corresponding band diagram, in equilibrium.
I n the case o f pnp transistors, because o f the i m p u r i t y concentration

gradient, the electrons within the base tend to diffuse t o w a r d the collector.
Thus an electric field must be present, pushing the electrons t o w a r d the
emitter-base j u n c t i o n . This same electric field w i l l then be o f such direction
as to aid the m o t i o n o f injected holes. Thus the injected minority
carrierswill now move not only by diffusion but also by drift due to the existence o f
this electric field. As a result, the transit time across the base w i l l decrease
and the upper frequency limitation o f the transistor associated w i t h this
transit time w i l l increase. Correspondingly, the transport factor w i l l also
increase.
2
T o provide an indication o f the effect o f this b u i l t - i n electric field, i n

Figure 7.15 theoretical calculations o f the transit time are shown as a
function o f the ratio o f the concentration o f the base i m p u r i t y at the
3
Junction Transistors -
226
emitter-base junction to that at the collector-base j u n c t i o n . I n these calculations, the i m p u r i t v d i s t r i b u t i o n i n the base region was approximated
bv an erfc-type d i s t r i b u t i o n . The transit time for the pnp transistor,
corrected by using Figure 7.15, is also indicated i n Figure 7.12.
3
The calculations o f M o l l and Ross, on which Figure 7.15 was based,

also show that i n the derivation o f the currents flowing in a transistor, the
quantity [N W )
i n the denominator o f E q u a t i o n 7.12 and i n subsequent
DB
1.0.
0.6
227
However, i t is evident that as the collector-base j u n c t i o n is reverse biased,

the w i d t h o f the collector-base depletion region w i l l increase and consequently the width of the neutral base region W will be reduced.
Thus
the gradient o f the injected m i n o r i t y carriers i n the neutral base region w i l l
become steeper, and therefore the collector current w i l l increase. T h e base
current, however, will not change significantly since i t is p r i m a r i l y due to
phenomena occurring near the emitter-base j u n c t i o n .
B
The E a r l y effect is, o f course, much more pronounced i f the d o p i n g

concentration i n the base region is relatively light as compared to the doping concentration in the collector region.
0.4
0.2
d.
10
10'
10
H i g h C u r r e n t Effects
The simple theory developed in Section 7.2 predicts that, as the current
level is increased, h
should reach a constant, current-independent level.
T h i s theory does not consider deviations from the c o n d i t i o n o f low-level
injection.
F E
A (0).'A (H )
F i g . 7.15
Theory
10
and Modifications of t h e "Simple
Thus as the collector-base j u n c t i o n reverse bias is increased, the current

gain w i l l increase. This phenomenon, first pointed out by E a r l y and since
then c o m m o n l y referred to as the Early effect, is clearly evidenced i n the
characteristics o f Figure 7.4b. These characteristics are traced under
constant base current conditions. I t is evident that as the c o l l e c t o r - t o emitter bias and therefore the collector-base reverse bias is increased, the
collector current for a given base current increases as represented by the
u p w a r d t i l t i n g o f the characteristics.
0.8
-If
Limitations
R e d u c t i o n i n base t r a n s i t t i m e d u e t o t h e g r a d e d i m p u r i t y
concentration
(7.30)
A c t u a l l y , at high current levels, the injected carrier concentration may

greatly exceed the doping concentration in the base region, i.e.. p
K
may h o l d . I n such a case, in order to maintain charge neutrality w i t h i n the
base, the electron and hole concentrations there must become equal. T h u s
the majority carrier concentration will also increase as the minority
carrier
concentration is increased. Such condition is referred to as modulation of
the conductivity o f the semiconductor.
Furthermore, N
itself is replaced by its value at the emitter-base j u n c t i o n ,
^*x>i?(0)- Otherwise, the equations remain unchanged. W i t h these changes,
results based on the idealized transistor model can be readily adapted to
the consideration o f double-diffused transistors.
I n the conductivity-modulated condition the semiconductor i n effect

becomes more heavily " d o p e d " as the injection level is increased. A s a
result, the rate o f increase o f the injection level w i t h increasing f o r w a r d
bias slows d o w n . Thus the collector current w i l l no longer f o l l o w the
simple exponential law,
across t h e base r e g i o n .
N {0)
and Nj,(W )
s
a r e t h e b a s e - i m p u r i t y c o n c e n t r a t i o n s at t h e
e m i t t e r - b a s e and c o l l e c t o r - b a s e j u n c t i o n s , r e s p e c t i v e l y .
ri
equations is replaced by the t o t a l number o f impurities i n the base per

u n i t area Q , given b y
B
N {x)
DB
dx.
Jo
DB
c.
E a r l y Effect
The effect o f the collector-base reverse bias on the c o m m o n emitter

current gain h
is n o t explicit i n the equations o f the previous sections.
F E
but w i l l a p p r o a c h
cc
cc
DB
lkT
e '*'' ,
q V i :
e
k T
> - .
The slower increase o f injected current w i t h f o r w a r d bias at h i g h current
junction
228
i uiiiiûr ^
levels is indeed evident in Figure 7.9. I t is also evident i n the forward

current-voltaae characteristics o f the germanium, silicon, and gallium
arsenide diodes shown in Figure 6.24.
Corresponding to the increase i n majority carrier concentration in the
base region, the emitter efficiency w i l l decrease bringing about a reduction
i n current gain at high current levels. A reduction i n current gain is i n
fact observed at high collector currents i n Figure 7.10.
6
A n o t h e r i m p o r t a n t deviation f r o m the low-level theory is due to the fact

t h a t the simple theory of p-n junctions w h i c h we have discussed i n Chapter
6 a n d o n w h i c h the above treatment o f transistors has been based is
founded on the basic underlying assumption that the semiconductor is
divided into depletion regions i n w h i c h the carrier concentrations are
smaller than the i m p u r i t y concentration, and into neutral regions where
space-charge neutrality approximately prevails. A t high current levels,
significant space charge may be present everywhere in the semiconductor,
w i p i n g out a meaningful distinction between depletion regions and neutral
regions. Thus, the analysis o f semiconductor devices must be modified
for such a c o n d i t i o n . This is beyond the scope o f the present discussion.
Dase
resistance
the center line under the emitter region, and lowest near the base contact.
This voltage build-up is i m p o r t a n t because it results in a variation o f the
emitter-base forward bias as a function o f distance y, leading to a higher
forward bias near the edges o f the emitter region than i n the central section.
This, i n turn, results i n a higher current density near the edges o f the
emitter. This condition is referred to as current crowding. Thus we want
both to calculate and to c o n t r o l the. base spreading resistance through
which the base current must flow.
The average magnitude o f the base voltage d r o p is given by
1
y
f'
= 77
(7.31)
v {y) dy
B
\L Jo
i f the voltage drop outside the n a r r o w region between'the emitter and the
collector can be neglected. T h e n the base spreading resistance can be
defined by
(7.32)
P V . . T * :
The calculation o f the average base voltage d r o p V is dependent o n the

particular transistor geometry. Such a calculation can be illustrated by
considering the simple stripe geometry shown i n Figure 7.16. The voltage
d r o p over an element dy here is given by
B
7.4
BASE
RESISTANCE
W e have seen that due (i) to injection o f minority carriers into the emitter
region, (ii) to recombination i n the emitter-base space-charge region, and
(iii) to recombination in the base region a current w i l l flow to the base lead.
T h i s current flows in a direction transverse to the direction o f the normal
transistor current flow as s h o w n i n Figure 7.16. As a result, a voltage drop
w i l l b u i l d up in the base region along the path o f the base current flow. I n
the case o f the pnp transistor shown here, the potential w i l l be highest at
-dV
= e*-gLl (y)
(7.33)
where Z is the length o f the stripe, and p is the average resistivity o f the
base region. The base current at p o i n t y is assumed to be given by
B
h(y)
= ihf -
(7.-34).
Here I is the total base current. (This assumes no current crowding.)

Substitution o f Equation 7.34 into 7.33 and integration leads to
B
_
V
1 opL
= I
12 W Z
(7-35)
The base spreading resistance is obtained by dividing the voltage drop by

I , and is
B
Fig. 7.16 Illustration of the calculation of the base spreading resistance for a stripe
geometry.
OB
230
The reciprocal of the average resistivity o f the base region is given by
PB
Pnl^ni*)
~ Â( )]
d-
- M a x i m u m Voltage Limitations
c o n d i t i o n , we estimate the area under the electric field d i s t r i b u t i o n curve

shown i n Figure 7.17. This yields
(7-37)
BV
A quick estimate of the average resistivity can be obtained by m u l t i p l y i n g
where Q
the average base i m p u r i t y concentration Q \W
and N
responding t o this concentration.
w i t h the m o b i l i t y cor-
231
= QQB
K,
(7.39)
2NjJ
is the t o t a l number o f impurities per u n i t area i n the base region

is the acceptor concentration i n the collector.
Because the base
Thus,
Hi-*
m q,u ^
(7-38)
PB
W W
mm. WE
where /xis the m o b i h t y value corresponding to the i m p u r i t y concentration
-Q^W-B-
7.5
MAXIMUM VOLTAGE
a.
LIMITATIONS
C o m m o n Base Configuration
T h e m a x i m u m voltage that can be applied to a transistor i n the c o m m o n base configuration, BV
CB0
avalanche
breakdown
(see Figure 7.4a'). is usually determined by the
voltage o f the collector-base j u n c t i o n .
A v a l a n c h e breakdown inp-n j u n c t i o n s has been discussed in the previous

chapter.
T h a t discussion applies w i t h o u t m o d i f i c a t i o n to the c o m m o n -
base breakdown in transistors i n such a case.
b.
C o m m o n - E m i t t e r Configuration
Fig. 7.17
T h e common-emitter b r e a k d o w n voltage BV
is measured w i t h the
base shorted to the emitter d u r i n g the measurement. The b r e a k d o w n
Illustration of the "punch-through" condition.
CES
-observed under this c o n d i t i o n w i l l usually be the avalanche breakdown o f

the collector-base j u n c t i o n . I n that case, BV
= BV .
However,
an interesting departure f r o m this can take place i f the space-charge
region on the base side o f the collector-base j u n c t i o n reaches the emitterbase j u n c t i o n before the collector-base j u n c t i o n can avalanche.
This
situation, w h i c h is referred to as the punch-through c o n d i t i o n , is illustrated
i n Figure 7.17. Once the collector-base depletion region reaches the
emitter-base j u n c t i o n , the t w o />regionsthe emitter and the collector
are connected w i t h a continuous depletion region. A current can now
CES
CB0
. . f l o w - h e n c e ' " b r e a k d o w n " takes place even i n the absence o f any avalanche
process.
To
calculate the voltage required to b r i n g about the punch-through
region o f diffused transistors is relatively heavily doped, the p u n c h - t h r o u g h

voltage is usually high i n diffused transistors, and therefore-the voltage
l i m i t a t i o n is usually due to the avalanche b r e a k d o w n o f the collector-base
junction.
The m a x i m u m voltage i n the common-emitter configuration w i t h the
base lead open, BV ,
is o f p a r t i c u l a r importance (see Figure 7.4b). T o
calculate this voltage, we must first consider the currents flowing i n the
transistor i n such a two-terminal operation. W h e n a certain voltage
V
is applied to the collector w i t h respect to the emitter, w i t h the base floating,
as shown i n Figure 7.18, the base region w i l l acquire a potential that is
intermediate between the emitter potential and the collector p o t e n t i a l .
As a result, the emitter-base junction will be very slightly forward
biased.
I n such a case, the collector current w i l l consist o f the reverse-biased
CE0
CE
junction
232
I ransistors
M a x i m u m Voltage Limitations
generation current o f the collector-base j u n c t i o n . l., ,
carried by those injected carriers which reach the collector-base j u n c t i o n .

V
I f the emitter current is I ,
the latter component will be given by
plus the current
en
Substituting
= BVCEO,
CB
yj. l .
E q u a t i o n 7.44 i n t o the
a d s to the relationship,
c o n d i t i o n , yt M
= 1 when
le
T K
However, the current flowing t h r o u g h the emitter-base j u n c t i o n and the

collector-base j u n c t i o n must be the same.
BV, CEO
Hence, in the open-base con-
BV, CBO
figuration,
Is
ic = y-TE
+ W
T h u s the above
10
Fig. 7.18
= 7*T CEO
[
(7.45)
/N
<i FE
We have seen i n Chapter 6 that the b r e a k d o w n voltage o f planar j u n c tions is determined by the electric field i n the rounded, nearly cylindrical
relation leads to
IcEO
y%=*
( - )
The generation current is j u s t the leakage current o f the collector-base

j u n c t i o n measured under open emitter conditions I
<yi -
+ CBC-
( -
4 1
Illustration of current flow in the open-base condition.
So far we have neglected the role o f multiplication in the collector-base

j u n c t i o n . I t was pointed out i n Chapter 6 that near b r e a k d o w n the current
n o r m a l l y flowing in a reverse biased p-n j u n c t i o n is multiplied by the factor
M because o f the incipient avalanche process. Thus, near breakdown, the
above relationship w i l l lead to
= (y^cso
ICEO
10=1
+ ICBO)M
(7.42)
or
=
( -)
1
ya M
This relationship shows that the leakage current in the common-emitter
configuration is larger than the leakage current i n the common-base configuration by the factor 1/(1 - ya M).
As a result, this current w i l l start
increasing rapidly when yat. M-+
1, rather than when M co as was the
case i n the common-base configuration. Thus the breakdown voltage in
the common-emitter
configuration will be lower.
T o estimate the reduction i n b r e a k d o w n voltage, we can use the empirical f o r m u l a for the m u l t i p l i c a t i o n factor discussed i n Chapter 6,
i i I
10
FE
i CEO
at I
1 0
e
" 1 ma, V
CE
i
1 0
10
Fig. 7.19 Open-base breakdown voltage of npn transistors as a function of commonemitter current gain.
3
(7.44)
region o f the j u n c t i o n . I n Chapter 10 we w i l l see that charges near the

surface can increase the electric field and further lower the breakdown v o l tage o f the j u n c t i o n . Thus i t is i m p o r t a n t to consider which i f any o f these
factors w i l l also have an effect on the common-emitter breakdown voltage
CEOT
above discussion applies o n l y to the region o f the transistor
under the emitter, i.e., to the section indicated in Figure 7.1a, because
injection i n other regions o f the transistor, e.g., near the surface, or near
the corner region, is much less efficient. Thus the correct breakdown
voltage BV
that must be used i n E q u a t i o n 7.45 is that o f a truly plane
j u n c t i o n which is often higher than the actual B V
o f a planar transistor.
n
SV
CB0
BV >
CB0
where BV
CB0
is the true b r e a k d o w n voltage of the collector-base j u n c t i o n ,
and n is between 3 and 6.
T h i s is illustrated by the experimental measurements* shown i n Figure

7.19, w h i c h show the breakdown voltage o f planar npn transistors in the
common-emitter mode. N o t e that the power-law relationship to" the current gain is observed and that the straight line extrapolates to the true
234
Saturation
plane breakdown-voltage value.

transistors, yielding n ^ 6.
Similar results were obtained w i t h pnp

= 0.18 ma
It is interesting to point out that since the current gain h

itself is a
function of collector current, the reduction i n b r e a k d o w n voltage as given
bv Equation 7.45 w i l l also depend on the collector current level. Thus
the breakdown characteristic in the common-emitter mode will exhibit a
switchback as shown i n the example i n Figure 7.4b due to the v a r i a t i o n o f
h
as a function of collector current.
F E
F E
7,6
MINIMUM
VOLTAGE
LIMITATIONS
^When a transistor is operated i n the common-base configuration, the

same collector current w i l l flow when there is zero bias applied across the
collector-base j u n c t i o n as when this j u n c t i o n is reverse biased. Thus
in this mode of operation the transistor imposes no lower l i m i t u p o n the
o u t p u t voltage. The situation, however, is different i n the c o m m o n emitter configuration.
if
0
Fig. 7.20
-0.5
-1
Collector current of the pnp transistor in saturation.
I
'
<
" A constant base current approximately corresponds to a constant

emitter-base forward bias, e.g., 0.5 v. As the collector-to-emitter voltage
is n o w reduced to this value, the collector-base j u n c t i o n cannot be reverse
.biased any more. Further reduction in V
w i l l actually cause the collector-base junction to be forward biased, injecting carriers in the opposite
direction to those injected from the emitter. T h u s two forward-biased
p-n
junctions now face each other. As a result, the measured net collector
current, which is the difference o f the t w o injected currents, will decrease
as V
is decreased as shown i n Figure 7.20. This condition is called
saturation.
i
f
CE
CE
Saturation was considered by Ebers and M o l l w h o treated the t w o

forward-biased j u n c t i o n s independently and calculated the net collector
current flowing through the transistor by superposition. The i n d i v i d u a l
and net carrier distributions due to forward-biasing o f just one and o f
b o t h junctions is illustrated in Figure 7.21. N o t e the reduction in collector
current as evidenced by the decreased magnitude o f the slope o f the net
m i n o r i t y carrier distribution in Figure 7.21b.
T o simplify the analysis, the recombination current components
associated w i t h the t w o depletion regions are neglected and the transport
factor a. is taken to be unity. The first assumption is justified at relatively
h i g h current levels; the second focuses attention on the effect o f the different emitter efficiencies o f the t w o j u n c t i o n s . This effect is particularly
i m p o r t a n t in double-diffused transistors where injection from the collector
to the base is an inefficient process due to the higher doping concentration
:>
(b)
"
Distance
Fig. 7.2I The minority carrier distribution within the base region.
(a) Emitter-base junction forward biased, collector-base junction reverse biased
(b) Both junctions forward biased:, saturation.
23S
Junction
236
i n the base.
Finally, equal j u n c t i o n areas are assumed.
Transistors
The
result,
after lengthv algebra, is
1 kT
VCE
In
1 +
/ (l
c
X)
\h SE\
\Ic scl
(7.46)
U
where r
and r
are the series resistances o f the emitter and the collector
regions, and x is the common-base current gain in the reverse direction,
i.e., using the collector as the e m i t t i n g j u n c t i o n . (<x is generally much
smaller than a for double-diffused transistors.) The positive sign is taken
for npn transistors, the negative for pnp.
S E
Thermal
Limitation
237
header through the silicon. F r o m the header directly under the j u n c t i o n ,

which has a temperature designated by 7\, the heat flows through the
header material to other parts o f the package, e.g.. the cap, which are
directly in contact w i t h the surrounding ambient. The cap w i l l have a
temperature r
which is lower than T . Finally, the heat will be conducted away from the transistor package to the surrounding ambient.
The total resistance to the flow o f heat can be pictured as a series combination o f the i n d i v i d u a l resistances o f the silicon, o f the header, and of
the air (in the absence o f a heat sink), as illustrated by the simple circuit
analogue shown i n Figure 7.23. I n this circuit, we also show what happens
c a p
N o t e the appearance o f the term I I { I B I F E )

Equation 7.46. O u t o f
saturation, where the current gain h
is denned, this term is unity. Thus
the argument o f the l o g a r i t h m is zero and therefore V
< co as i t
indeed does on the characteristics (see Figure 7.20). I n saturation, however, this term is no longer unity, thereby leading to a finite value o f
V the
m i n i m u m voltage at which the transistor can be operated i n
the common-emitter mode.
C
Junction
Air
(high
resistance)
wv*w%,fVA
Si
Header
CE
Heat sink
(low resistance)
CE
7.7
THERMAL
LIMITATION
Carriers accelerated i n the large electric field o f the collector-base

depletion region (see Figure 7.2b) suffer many collisions w i t h the semicond u c t o r lattice, thereby converting their kinetic energy into heat. The
power dissipated this way, IQV ,
must be transported away f r o m the
collector-base j u n c t i o n i f the temperature o f the j u n c t i o n is to remain at a
steady-state value. The path through w h i c h this heat must be conducted
away is illustrated schematically in Figure 7.22. First, the heat is transp o r t e d f r o m the j u n c t i o n , w h i c h has a temperature Tj, to the supporting
CB
Fig. 7.23
Circuit analogue for the heat-flow problem.
if the transistor package is placed in contact w i t h a heat sink, such as a

metal block. A heat sink w i l l provide a small shunt resistor in parallel
to the resistance due to the air.
Let us consider the individual resistances appearing in the case o f an
air-cooled device. The resistance for thermal flow is obtained from
Fourier's law o f heat conduction,
P = k,
A
ox
(7.47)
w h i c h states that the heat flow rate P is p r o p o r t i o n a l to the temperature

gradient.! The p r o p o r t i o n a l i t y constant k
is the thermal c o n d u c t i v i t y ;
A is the cross-sectional area available for the flow o f heat. I n steady state
this becomes
t h
P = k
^ A
(7.48)
where L is the length o f the path o f heat conduction. This can be recog
nized as the equivalent o f Ohm's law for the conduction o f heat. Thus the
7
Header
Fig. 7.22
Illustration of the flow of heat in an encapsulated transistor.
- Note the analoev to diffusion where the flux is proportional to the concentration
sradient.
238
thermal resistance
Problems
239
o f a material w i l l be given by
REFERENCES
*
= A
A c c o r d i n g l y , the i n d i v i d u a l resistances that we m u s t combine i n series are

Rth.si
LHH
CITED
4 9
< - >
the silicon
(7.50)
= -=
for the header
(7.51)
= r ~ "
f o r
1. Transistor action was first described by J . Bardeen and W. H . Brattain, "The Transistor, A Semiconductor Triode," Phys. Rev., 74, 230 (1948). The theory of current
flow in junction transistors was established by W. Shockley, "The Theory of P-N
Junctions in Semiconductors and P-N Junction Transistors," Bell System Tech. J.,
28, 435 (1949). This theory was then extended by C . T . Sah, R . K . Noyce, and W .
Shockley, "Carrier Generation and Recombination in P-N Junctions and P-N
Junction Characteristics," Proc. IRE, 45, 1228 (1957).
2. H . Kroemer, "Der Drifttransistor," Naturwiss, 40, 578 (1953).
t h e
a i r
( 7
5 2 )
-where A, is the gas-phase heat-transfer coefficient, analogous to the massh
transfer coefficient discussed i n Chapter 1. T y p i c a l values for the
quan-
thickness o f wafer ^
^ j u n c t i o n area, ~ 1 0 ~
SI
4. J . M . Early, "Effects of Space-Charge Layer Widening in Junction Transistors,"

Proc. IRE, 40, 1401 (1952).
5. R . N . Hall, "Power Rectifiers and Transistors," Proc. IRE, 40, 1512 (1952).
tities involved here are :

LS =
3. J . L . Moll and I. M . Ross, "The Dependence of Transistor Parameters on the Distribution of Base Layer Resistivity," Proc. IRE, 44, 72 (1956).
10~ cm,
3
cm
6. W. M . Webster, "On the Variation of Junction-Transistor Current Amplification

Factor with Emitter Current." Proc. IRE. 42, 914 (1954).
'
for a medium-sized device,

7. See. for instance, C . T . Kirk, "A Theory of Transistor Cutoff Frequency (f ) Fallon"
at High Current Densities." IRE Trans., Electron Devices E D - 9 , 164 (1962); A .
van der Ziel and D . Agouridis. "The Cutoff Frequency Falloff in U H F Transistors
at High Currents." Proc. IEEE (Correspondence). 54, 412 (1966).
T
* . t t S * 1 . 5 w a t t / C C cm)
hence, R
TH
5C/watt;
S i
1 cm,
8. W. W. Hooper, unpublished.
cm ,
fc ,
4 w a t t / ( C cm)
jR
5C/watt;
a t
hence,
at 10"
t t J k
* , S
hence, R
IHML
T h u s the
10-
to J O
- 2
9. J . J . Ebers and J . L . Moll, "Large Signal Behavior of Junction Transistors," Proc.

IRE, 42, 1761 (1954).
PROBLEMS
watt/( C cm ),
overall t h e r m a l resistance w i l l be
device is air-cooled.
7.1
Derive the expression giving the small-signal current gain h, Equation 7.6.
7.2
(a) Verify that Equation 7.10 satisfies the transport equation and the boundary
conditions.
10 to 10 C,/watt.
10
to
1 0 C / w a t t i f the
(b) Verify the exact and approximate expressions given for a .
T h i s resistance can be reduced to about 1 0 C / w a t t
by proper heat sinking.
7.3
If h
rT
for the case when a. = 1 is denoted by h ,.,

T
y 1 is denoted by h ^^.
i.
hrt
REFERENCES
and h
rE
for the case when
show that, in general,
rL
READING
1
h,
r y
.1
hf , .
aj
Junction transistors are the subject of several texts. Among these are:
R. D . Middlebrook, An Introduction to Junction Transistor Theory, Wiley, 1957.
W. W. Grtner, Transistors. Van Nostrand. I960:
A . B . Phillips, Transistor Engineering. McGraw-Hill Book Co., 1962;
7.4
When a transistor is irradiated with large doses of fast neutrons or high-energy

electrons, it is often found that the lifetime decreases with irradiation dose approximately as T = KI4>. where 6 is the dose and A' is an empirical constant (see Chapter
5). Assuming that the emitter efficiency is unity, derive a relationship giving
(a) base current, and (b) h as a function of the dose, and their respective values
before irradiation.
F[
S. Hakim, Junction Transistor Circuit Analysis, Wiley. 1962.
Junction
240
Transistors
7.5
If the total charge due to injected excess minority carriers within the base is Q, and
if the emitter efficiency
= 1, express the base current, collector current, and the
current gain in terms of Q.
7.6
(a) Estimate the magnitude of the electric field near the emitter-base junction, in
equilibrium, for the pnp transistor used for illustration in this chapter.
(Hint: Fit an exponential profile to the base impurity distribution shown in
Figure 7.13.)
(b) Calculate the fraction of the injected minority-carrier flow that is due to
drift when the emitter-base junction is forward biased. Assume that the
electric field is unchanged. (.Under what conditions is this assumption justified?)
(c) Recalling that a flux is the product of concentration and velocity, use your
result to estimate the increase of the velocity of the injected minority carriers
due to the built-in field. Compare your result with Figure 7.15.
Problems
241
transistor which is avalanche-breakdown limited and one which is punch-through

limited. C a n this test be used to distinguish the two cases?
7.14 For transistor " A " ,
BV
BV
BV o
BV =
0B0
CS
CB
BB0
For transistor " B " ,

BV o
BV
BV
BV
CB
CBS
CE0
SB0
7.7
7.S
(a) Using the collector current versus emitter-base voltage data for the pnp
transistor, and assuming a lifetime of 1 .usee, calculate the total number of
impurities per unit area in the base,
(b) Using the maximum value of the current gain, calculate the total number of
impurities per unit area in the emitter.
Compare both numbers with the corresponding estimates based on the impurity
concentration distribution given.
Derive an expression showing the effect of the collector-base junction reverse bias
on the transit time across the base and on h , assuming that the collector-base
junction is linearly graded.
= 105 v.
= 105 v.
= 96 v.
9v.
= 75 v.
= 69 v.
m 69 v.
=
6 v.
What mechanism limits the maximum voltage of these two transistors?

7.15 Draw the energy band diagram for a pnp transistor in saturation.
7.16 The Ebers-Moll analysis makes three important simplifying assumptions: It
neglects the recombination current components associated with the two depletion
regions; it assumes that both transport factors are unity; and it assumes that the
areas of the emitter and collector junctions are- equal. Discuss the validity and
importance of these assumptions, and indicate qualitatively the direction of any
errors they may introduce with respect to:
FB
7.9
The collector current is transported across the reverse-biased depletion region of

the collector-base junction by drift.
(a) Assuming that the velocity of the carriers is their maximum drift velocity,
show that the concentration of injected carriers across the base-collector
depletion region is constant.
(b) Sketch the electric field distribution within the collector-base junction depletion
region for increasing current densities, assuming the base is much more heavily
doped than the collector.
(c) A t what current density does the electric field approach a constant value?
(d) What will happen if the current density is further increased?
(a) Alloy germanium transistors.

(b) Planar, double-diffused silicon transistors.
7.17 Calculate the overall thermal resistance of the device discussed in Section 7.7, alsotaking into account conduction of heat through the leads. Assume a transistor,
with a copper base and a copper emitter lead, each 2 mil in diameter, and 5 mm
long. Do you expect the two leads to be equally effective as heat-conduction
paths? Justify.
7.18 A single pulse of current is imposed upon a silicon transistor of dimensions as
given in Section 7.7. How long a duration must the current pulse have for the
transistor chip to achieve its new steady-state temperature? In this calculation
neglect heat conduction away from the chip.
7.10 Calculate the current density at which the base becomes conductivity modulated
for a pnp transistor. Compare this current density with that obtained in Problem
7.9c. Which condition sets in first in a double-diffused transistor? In an alloy
transistor ?
7.11 One criterion of the onset of current crowding is when the transverse base voltage
drop exceeds kT/q. Estimate the corresponding collector current level for a pnp
transistor which has an h
of 50, an impurity distribution as given in Figure
7.13, and a stripe geometry with Z = 40 mil, L ' 0.5 mil.
FB
7.12 I n a transistor in which the value of BV o is determined by "'punch-through",

\BV \
= \BV \
+ \BV \.
Justify this relationship. (Hint: Consider what
happens at the emitter-base junction as "punch-through" is obtained.)
rB
CB0
CB0
Bao
7.13 A reverse bias V is applied to the base region with respect to the emitter. What
effect will this bias have on the voltage applied to the collector, also with respect
to the emitter, at which large reverse current begins to Sow? Consider both a
H
o, fa
242
PRINCIPLES
OF
OPERATION
CHARACTERISTICS
TABLE
7.1
MODIFICATIONS
IMPORTANT FORMULAS FOR JUNCTION TRANSISTORS
npn
pnp
I '
Pmh
FB
F E
P - y
J
1
1 - a
a = ystji [See footnote on page 219.]
B
OF THE
THEORY
Je
a=h ^
Current gain
SIMPLE
Junction Field-Effect
Transistors
Transport factor
1
V
2 A/
W
R -
Emitter efficiency
.
+s
A'M^JJ
The j u n c t i o n field-effect transistor, often caheddaeld-effect transistor,

proposed by Shockley in 1952 and firstêmOTBirated by Dacey and
Ross. is a device based on an entirely different.physlcalprinciple than the
j u n c t i o n transistor. While the j u n c t i o n transistor -operates t h r o u g h the
transport o f injected m i n o r i t y carriers, i n a junction.field-eflect transistor
the depletion region o f reverse-biased p-n j u n c t i o n s is used to modulate
the cross-sectional area available for current flow. The current is transported by carriers of one polarity o n l y ; hence, i t is usual to refer to the
field-effect transistor as a unipolar device i n contrast to the j u n c t i o n
transistor w h i c h is a bipolar device since i t involves b o t h types o f carriers.
1
A . *
-DE
* v
Transit time
r~2D
'
rB
Base resistance
r B
Leakage currents
"CEO
..
f o r
7 "w~ ^
'"~2Z)
s t r i p e
e o m e t r
N 2 3
>'
The field-effect transistor i n its most c o m m o n f o r m is illustrated i n

Figure 8.1. A n n-type layer is g r o w n epitaxially on a heavily doped />-type
substrate. T h e n , using regular planar technology, a p-xypt gate j u n c t i o n
is formed by diffusion from the t o p . Finally, contact is made to the pregions and also to the /i-region at either side o f the t o p gate resulting i n a
source and a drain contact.! W i t h such fabrication techniques, the t o p and
b o t t o m gate junctions are approximately symmetrical. A c c o r d i n g l y , i n the
f o l l o w i n g discussion we concentrate our attention on symmetrical fieldeffect transistors.
0230
w
1 yo. M
T
Maximum
voltages
DV Q
CE
f
Minimum
voltage
C E
Ti
(sat)-.
In
9
-pnp
/ c
, A H
- + |VSEMVSC|J
K>
+npn
-f In order to insure good contact to the n-region, it is usuallv necessarv to increase its
surface concentration. This can be accomplished by an additional n-type diffusion at
the source and drain regions. These are omitted in Figure 8.1.
243
244
Junction
Field-Effect
Transistors
P r i n c i p l e s of
Operation
24S
of n-type material. Accordingly, the device shown in Figures 8.1 and 8.2
is referred to as an /i-channel j u n c t i o n field-effect transistor.
The resistance o f the channel is given by
/.
Safari- W< ^ \*J '
R =
q/x N Z(d
n
2 HO
(3-11
where N is the d o n o r concentration i n the channel region, L , Z , a n d d

designate the length, width, and thickness o f the channel, respectively
(see Figure 8.1), and W is the width o f the depletion region o f the top and
b o t t o m gates.
I n the case shown i n Figure 8.2a, when no gate voltage is applied and
V is small. W is the zero-bias depletion region o f the gate j u n c t i o n s . I t
is evident from Figure 8.2a that these depletion regions restrict the current
flow to a smaller cross-sectional area than w o u l d exist w i t h o u t them.
For any given d r a i n voltage, the voltage along the channel w i l l increase
from zero at the source to V at the drain. Thus both gate junctions w i l l
become increasingly reverse biased as we proceed from the source to the
drain. So long as V is much less than the built-in voltage o f the gate
junctions d> , the depletion region w i d t h W w i l l remain practically independent o f V and the channel will act as a resistor. However, as V is
increased, the average cross-sectional area for current flow is reduced
because o f the increasing reverse bias o f the gate junctions near the drain
area. Thus the channel resistance R w i l l also increase. As a result, the
current-voltage characteristics will begin to fall below the i n i t i a l resistor
line. This tendency is evident i n the experimental measurements shown i n
Figure 8.3 where d r a i n current is plotted as a function o f d r a i n voltage for
a silicon j u n c t i o n field-effect transistor, used for illustration t h r o u g h o u t
this chapter. The top curve is for V = 0.
D
m
Fig. 8.1
n-Channel junction field-effect transistor fabricated by planar-epitaxial methods.
We begin by discussing the physical principles underlying the operation

o f the j u n c t i o n field-effect transistor. Then we derive the most i m p o r t a n t
characteristics o f such devices, including the current-voltage relationship,
the channel conductance, and the transconductance, a n d discuss the
factors affecting the gate leakage current. Finally, we discuss the various
modifications o f the simple theory o f field-effect transistors.
8.1
PRINCIPLES O F
OPERATION
I n Figure 8.2, we illustrate the conditions that prevail when the gate-tosource potential V = 0. I f a small positive voltage V is applied to the
drain, electrons w i l l flow f r o m source to d r a i n ; hence, a current w i l l flow
from drain to source t h r o u g h the rc-type region enclosed between the two
depletion regions. Such a region is c o m m o n l y referred to as a channel
G
As the drain voltage V is further increased, the depletion region widthalso increases near the drain until eventually the two depletion regions_
touch, as indicated i n Figure 8.2b. This happens when
D
Channel
W = - . \
2
(8.2)
]
fc = N ^
Using the one-sided stepjhjnction formulas, we can readily obtain the

corresponding value o f the drain voltage ^ s a t *
(a)
to
MV i
P e r a t i
(o
i U n C T i
smaH- T "
(a) V is small; channel resistance is constant.

(b) V = V
; onset of saturation.
(c) V > V ;
further increase in drain current.
S t
o r for
= 0
^sat = ^
"
[V -0]
o
(8.3)
D J d t
o n i
where i
is the b u i l t - i n voltage o f the gate junctions.
A t this drain voltage, the source and the drain are completely separated
B
246
Principles of O p e r a t i o n
by a reverse-biased depletion region which n o r m a l l y does not conduct

because there are very few carriers i n i t . However, we have already seen
in Chapter 6 that i f carriers are created w i t h i n a reverse-biased depletion
region by thermal generation or by avalanche b r e a k d o w n , a current w i l l
flow across i t . We have also seen in Chapter 7 that large currents can flowacross a reverse-biased depletion region i f carriers are injected into itas
?47
the point X w i l l still remain the same. V

T h u s the number o f carriers
arriving f r o m the source to this p o i n t , and hence the current flowing from
source to d r a i n , w i l l remain unaltered since the p o t e n t i a l d r o p in the channel
from source to the point X remains unaltered." T h u s for drain voltages
larger than V
the current will not change but will remain at the value
]
, as is evident f r o m the experimental data s h o w n i n Figure 8.3. T h i s
phenomenon is called saturation because the current saturates w i t h
increasing d r a i n voltage.
W h e n a gate voltage is applied to b o t h /^-regions o f such a p o l a r i t y as to
reverse bias the gate-to-channel j u n c t i o n s (i.e.. negative for an ??-channel
device), the depletion regions w i l l , o f course, become w i d e r . ; T h u s for
small values o f the drain voltage V the channel w i l l again act as a
resistor, b u t its resistance w i l l be larger because the cross-sectional area
available for current flow will have decreased due to the increased w i d t h
W of the depletion regions. This is evident f r o m the experimental data
corresponding to V = 1 v, 2 v. etc. i n F i g u r e 8.3.
As V is increased, the resistance o f the channel w i l l increase. When
V reaches a large enough value, the depletion regions w i l l again t o u c h
near the drain region. This w i l l take place when V VJJ
where
DSIV
DSIX
D s & t
UV
Fig. 8.3 Current-voltage characteristics of a silicon n-channel junction field-effect

..transistor. This device is used for illustration throughout this chapter. Its structural
parameters are: ZIL = 170, d = 3 ft, N = 2.5 x 10" cm" .
1
in the collector-base depletion region o f a transistor when the emitter-base

junction injects m i n o r i t y carriers i n t o i t , or in the case o f punch-through
breakdown.
The. situation existing in a field-effect transistor after the depletion
regions have met is quite similar. A current w i l l flow in this case across
the combined depletion regions separating the source from the d r a i n .
This current flows because o f carriers injected i n t o the depletion region
i f r o m the channel at the p o i n t where the depletion regions touch, marked
. by X in Figure 8.2b. A s in the case o f the current flowing in the depletion
region o f the collector-base j u n c t i o n o f a transistor, the current flowing
across the depletion region beyond the p o i n t X will be limited by the n u m ber o f carriers arriving at this p o i n t . The magnitude o f this current, i n
t u r n , depends on.the voltage d r o p from the-source to the p o i n t X.
This
voltage drop, however, is exactly V
since this is the reverse bias that is
required for the two depletion regions to touch.
D&!LX
I f the drain voltage V is increased beyond V . the depletion region

near the drain w i l l merely thicken and the p o i n t X will move slightlv
t o w a r d the source, as indicated in Figure 8.2c. However, the voltage at
N
LITMX
which is the voltage required for the onset o f saturation in the presence o f
a gate voltage. I t is evident that the a p p l i c a t i o n o f a gate voltage V
lowers the drain voltage required for the onset of saturation by an equal
amount.^ (Recall that V is negative for an n-channel device.)
As the d r a i n voltage is further increased beyond V ^.
the p o i n t X
where the depletion regions touch w i l l again merely move slightly t o w a r d
the source but the voltage d r o p f r o m the source to the p o i n t X w i l l remain
the same, V
. Accordingly, the drain current w i l l also remain at its
value corresponding to the onset o f saturation J
This value, however,
will be lower than in the V = 0 case because the voltage d r o p f r o m source
to p o i n t X. K i , , itself is lower. This is evident from the experimental
data in Figure 8.3.
Thus we can distinguish t w o different regions o f the current-voltage
relationship o f field-effect transistors. I n one region, when V is small,
the cross-sectional area o f the channel is practically independent o f V
G
DS
D S & V
s a t
t This assumes that the movement of the point X toward the source is negligible. The
validity of this assumption will be discussed in a later section.
t Usuallv the same bias is applied to both gates.
I The above as well as other formulas are summarized at the end of this chapter in
Table 8.1 for both n- and />-channel devices.
Junction
248
Field-Effect
Transistors
and the current-voltage characteristics are essentially ohmic or linear. We

refer to this l i m i t i n g region o f operation o f field-effect transistors as the
linear region. I n the other extreme, for V > V'yj , the current saturates
at /
. We w i l l refer to this region o f operation o f field-effect transistors
as the saturation
region.
n
Characteristics of Junction-Field-Effect
Transistors
249
to the fundamental equation o f field-effect transistors:
sat
n < a l
= G. V
- J J {{V
+ 4B ~
'.;)'- ~
WW
(3.7)
where
8.2
a.
CHARACTERISTICS OF JUNCTION
FIELD-EFFECT TRANSISTORS
Current-Voltage Relationship
is the conductance o f the metallurgical channel, i.e., the conductance o f

the rt-type layer between the two ^-type regions discounting the presence
Let us now consider a j u n c t i o n field-effect transistor before the onset o f

saturation as s h o w n i n Figure 8.4. The voltage drop across the elemental
section o f the channel is given by
dV = I dR
D
=
q,u N Z[d
n
(8.5)
lW{y)]
Fig. 8.4 The elemental section of the channel employed In the derivation of the currentvoltage characteristics of junction field-effect transistors.
i n analogy to E q u a t i o n 8.1 except that L is replaced by dy.

region w i d t h at distance y from the source is given by
W{ij)
llK^[V(y)
+ <j> B
The depletion
V]
(8.6)
Substituting E q u a t i o n 8.6 into 8.5 and integrating between the source

(where y = 0 and V = 0 | ) and the drain (where y = L and V = V ) leads
D
t We define the potential at the source as our ground potential. This analysis neglects
the series resistance between source contact and the beginning of the channel (y 01,
and between drain contact and the end of the channel (>j L).
250
o f the depletion regions altogether. I n Figure 8.5a we show the experimentally measured current-voltage characteristics o f the field-effect t r a n sistor used in this chapter as an example. Below i t . in Figure 8.5b. we
show the family o f characteristics calculated from Equation 8.7. using the
k n o w n values o f structural parameters. E q u a t i o n 8.7 is valid only below
saturation. Thus the curves shown were calculated for 0 < V < ^ s a t
using Equation 8.7;, beyond V
the current was taken to be constant,
i n line w i t h our earlier argument.
251
Characteristics of Junction Field-Effect T r a n s i s t o r s

10X10"
DB&1
O f particular importance is the f o r m o f the above equation i n the

linear and i n the saturation regions. F o r small drain voltages, i.e., for
V <f> V , the bracketed terms can be expanded leading to the
simple f o r m u l a ,
D
, 8K,o(ifi -
V)
0
[Linear region].
T h e o r y : Eq. 8.11
(no series resistances)
(8.9)
qN d*
D
I t is evident that this expression gives the current-voltage relationship o f

a resistor whose resistance increases w i t h gate voltage.
The channel conductance g is given by
31*
Experiment
(8.10)
:onsi
I n the linear region, g is given from E q u a t i o n S.9 as

f
kK.e (d>
qhi dD
IVfl
-5
I n Figure 8.6, we show the experimentally measured channel conductance

i n the linear region o f our device i n comparison w i t h calculations based on
E q u a t i o n 8.11 for this device. (Also shown are curves which were corrected
for the presence o f series resistances between source and drain contacts
and the channel region. This correction w i l l be discussed later.) As the
gate bias is increased, the conductance decreases u n t i l finally, at a certain
value o f gate voltage, the conductance vanishes altogether. This voltage
is called the turn-off voltage and it corresponds to the reverse bias -that
must be applied to the t w o gate junctions in order to deplete all o f the
channel region. A c c o r d i n g l y , the turn-off voltage can be calculated from
the condition W = d!2 to be
V
(8.12)
- 6
- 7
Fig. 8.6 Channel conductance of the field-effect transistor as a function of gate voltage
in the linear region (V small).
r
The drain current i n the saturation region I ,i

calculated f r o m
E q u a t i o n 8.7 by evaluating the drain current at the l i m i t o f the applicability o f this equation, i.e., at V = V .
Substitution o f E q u a t i o n 8.4
into 8.7 yields the saturation current,
Dst
DS&X
2 /8K,e,(*
f nsai C
V)
(;
its
VG)
qNod-
1<JAV
2
f l
8K.e
[Saturation r e g i o n ] .
(8.13)
252
b.
Transconductance
A n i m p o r t a n t property o f field-effect transistors is the transconductance

defined by
dI
(8.14)
The transconductance represents the change o f drain current at a given

d r a i n voltage upon a change in gate voltage. I t can be readily evaluated
by differentiating E q u a t i o n 8.7. This yields
8K ,
^TTT,
qN dS
om
<
[ V D
V
J > B - V G -
VVJJ
VQ\
(8-15)
The transconductance i n the linear region can be obtained by expanding

the bracketed terms. This yields
7 - ^ = ^ = =
[Linear region].
(8.16)
Conversely, the transconductance i n the saturation region can

calculated by inserting V = V
into Equation 8.15. This yields
D
= G
be
Dsat
[Saturation region].
(8.17)
qN d'
D
C o m p a r i s o n w i t h E q u a t i o n 8.11 shows that the transconductance

in the
saturation region exactly equals the conductance in the linear region. This
i m p o r t a n t result is given experimental verification by a comparison o f the
transconductance data shosvn in Figure 8.7 with the conductance measurements shown i n Figure 8.6.
c.
G a t e Leakage C u r r e n t
Because the gate is reverse biased w i t h respect to the channel, the

current flowing to the gate terminal is very small; i t is the reverse-bias
leakage current o f a p-n j u n c t i o n . Thus the gate impedance, i.e., the input
impedance o f a field-effect transistor, is very high. This high i n p u t
impedance is one o f the very i m p o r t a n t distinguishing characteristics o f
field-effect transistors as compared to j u n c t i o n transistors.
The ordinary room-temperature leakage currents o f reverse-biased
silicon p-n junctions p r i o r to breakdown are o f the order o f picoamps to
nanoamps. However," surface effects can lead to drastic increases i n
Fig. 8.7 Transconductance of the field-effect transistor as a function of gate voltage

in the saturation region. (V >
V ).
fl
Mt
j u n c t i o n leakage currents as discussed i n Chapter 10. I n the same.manner,

they can lead to similar increases i n the gate leakage current o f j u n c t i o n
field-effect transistors. Thus surface effects can degrade the high i n p u t
impedance o f field-effect transistors.
8.3
a.
M O D I F I C A T I O N S O F T H E SIMPLE
THEORY
G r a d e d C h a n n e l Regions
I n Section 8.2 we considered a field-effect transistor i n which the

impurity concentration in the channel region was u n i f o r m ; a n d - w e
J u n c t i o n . Field-Effect
254
Transistors
Modifications of the Simple
Theory
25S
employed the one-sided step-junciion a p p r o x i m a t i o n . This is a reasonably

good approximation for field-effect transistors fabricated i n the manner
described in the beginning o f the chapter since the junctions are usually
shallow. However, i f the diffusions are deeper and the channel is located
further away from the surface, we may approach the range o f validity o f
the linearly graded j u n c t i o n a p p r o x i m a t i o n .
of operation o f the field-effect transistor is then given by the

corresponding to this charging time constant.
A l s o , field-effect transistors can be made entirely by diffusion techniques,

w i t h o u t recourse to epitaxial g r o w t h . Thus, for instance, they can be made
by successive diffusions o f impurities o f opposite type much the same way
as the emitter and base regions o f a double-diffused j u n c t i o n transistor
are formed. I n such a case the i m p u r i t y d i s t r i b u t i o n in the channel region
i s .similar to the i m p u r i t y d i s t r i b u t i o n in the base o f a transistor and is
extremely graded.
A simple estimate o f the upper l i m i t off can be obtained f r o m the ratio

of the m a x i m u m transconductance to the m i n i m u m gate capacitance. The
m a x i m u m transconductance, f r o m the previous section, is G .
The
m i n i m u m gate capacitance is obtained when the depletion region reaches
its m a x i m u m w i d t h , which is one-half o f the channel depth d. T h u s ,
/. < * f ^ f ?
c.
W h e n the gate voltage is changed by an amount A f p . the w i d t h o f the

depletion regions changes. As a result, the d r a i n current will also change.
Some o f the additional d r a i n current w i l l be used to make up the change
i n charge contained w i t h i n the depletion regions o f the gate junctions.
The response time o f the field-effect transistor ; can then be defined as
that time in which the change i n the drain current makes up the change i n
the total charge on the gate, that is,
= AQ A .
(8.18)
Thus i t follows that this response time is given by

t
_AO A
a
_AO A AV
AI
G
_ C
AV
Here C
is the xoial gate capacitance o f the device, given by
Z
W
(8-22)
Note that this l i m i t i n g frequency is p r o p o r t i o n a l to the m o b i l i t y . Because

the m o b i l i t y o f electrons in silicon is approximately twice the m o b i l i t y
of holes, an ??-channel device promises twice as h i g h a frequency range as a
/i-channel device for the same geometric factors and doping levels.
C u t - O f f F r e q u e n c y for T r a n s c o n d u c t a n c e
(8.21)
tM
-I
H o w e v e r , i t has been shown theoretically that the general characteristics

o f the j u n c t i o n field-effect transistor are quite independent of the particular
i m p u r i t y distribution. Thus the simplest analysis, that for the u n i f o r m
i m p u r i t y distribution presented i n the previous section, provides most o f
the general features o f the analysis for an arbitrary distribution to a fairly
good a p p r o x i m a t i o n .
b.
1 - =
frequency
where IK is the average depletion region w i d t h . The m a x i m u m frequency
S o u r c e - t o - D r a i n Resistance in S a t u r a t i o n
We have already seen i n the previous section that in the saturation

region the potential at the end o f the channel, at the point X o f Figure
8.2. w i l l be fixed at precisely the value o f I ' ^ a , corresponding to the applied
gate voltage. This is so because the point A itself is where the t w o depletion
regions j u s t touch. Hence, the reverse bias across the gate j u n c t i o n s at
this p o i n t is fixed by the c o n d i t i o n that W = d/2 there.
1
As the drain voltage is increased further, the reverse bias between gate
and the drain region itself is also increased; hence, the w i d t h o f the
depletion region near the drain w i l l also increase. As a result, the p o i n t X
will move toward the source as indicated i n Figure 8.2c. The voltage at the
point X remains at the same value b u t the length L from the source to the
p o i n t X shortens: thus it is evident that the drain current w i l l increase at
a given gate voltage as the drain voltage is increased. This results i n an
u p w a r d t i l t i n g of the current-voltage characteristics beyond saturation
which-is particularly prominent for devices w i t h small channel length L .
This phenomenon is quite analogous to the Early effect discussed i n
connection w i t h j u n c t i o n transistors. I n b o t h cases, the increase i n current
takes place because the current path is shortened by the w i d e n i n g o f a
reverse-biased depletion region.
t More rigorous considerations yield 1/2CT/ for this frequency limitation.
Junction r.teld-fcrtect I ransistors
2So
d.
n^terences v-itea
257
zero, as assumed in the previous treatment, but w i l l have some finite value
K . Thus the effective gate voltage w i l l be
Effect of S e r i e s Resistance
I n the above calculations we considered only the resistance o f that

p o r t i o n o f the channel w h i c h can be modulated by the application o f a
reverse bias to the gate. I n reality, there are series resistances present,
b o t h near the source and near the d r a i n , which interpose an IR drop
between the source and d r a i n contacts and the channel. These series
resistances are illustrated schematically i n Figure 8.8.
= ^ .
C
a p p l
(8.25)
As a result, the observed transconductance is given by

/ . \
dI
dI
gjobs) = - =
^G.appl
d[V 4-.>7J
which, i n turn, yields
D
(8.26)
dI
dI
and hence,
S
gjobs) =
channel
(8.28)
1 + R,gm
^C.ppl
Fig. 8.8 Illustration of the series resistances due to the unmodulated portions of the
channel near source and drain.
The effect o f these series resistances o n the channel conductance i n the

linear region can be readily calculated by noting that
This last equation shows that the observed transconductance i n the.

saturation region w i l l be reduced due to the presence o f a series resistance
near the source from that attainable i n the absence o f such a series
resistance. This reduction is shown i n Figure 8.7 for various values o f R,.
The series resistance near the d r a i n w i l l act i n a different manner.
Because o f the IR d r o p across this resistance, the drain voltage required to
bring about saturation o f the d r a i n current w i l l be larger than w i t h o u t i t .
However, since beyond that voltage, i.e., f o r V > K
, the magnitude
of V has no significant effect o n the d r a i n current, the drain series
resistance will have no further effect either.
D
D s a t
=
g(obs)
- + R, + R
g
(8.23)
where e is the true channel conductance while g(obs) is the conductance

observed experimentally. R, and R are the series resistances near the
source and the drain, respectively, as indicated i n Figure 8.8. Thus,
READING
REFERENCES
g(obs) =
(8.24)
J. T . Wallmark, ' T h e Field-Effect TransistorA Review," RCA Rev., 24, 641 (1963).
L . J . Sevin, Field-Effect Transistors, McGraw-Hill Book Co., 1965.
1 + ( * , + *<)*
w h i c h shows that the observed conductance will be reduced due to the two
series resistances. This reduction was taken into account in Figure 8.6
for various series resistance values. The value (R, 4- Rf)
100 Q, which
provides a good fit w i t h the experimental data, is in fact a very reasonable
one for this particular device geometry.
Let us now consider the effect o f the series resistance near the source
region, R , on the transconductance i n the saturation region. Because o f
this resistance, the potential at the beginning o f the channel will not be
s
REFERENCES
CITED
1. W. Shockley, -'A Unipolar -Field-Effect' Transistor," Proc. IRE, 40, 1365 (1952).
2. G . C . Dacey and I . M . Ross, "Unipolar 'Field-Effect' Transistor," Proc. IRE, 41,
970 (1953); G . C . Dacey and I . M . Ross. "The Field-Effect Transistor," Bell System
Tech. J., 34, 1149 (1955).
3. R . S. C . Cobbold and F . N . Trofimenkolf, "Theory and Application of the FieldEffect Transistor," Proc. IEE, 111, 1981 (1964).
258
PROBLEMS
8.1
Prepare a lable comparing the electrical characteristics o f junction transistors and

junction field-effect transistors.
8.2
Rearrange the current-voltage relationship, and the expressions for conductance

and transconductance by employing the expression for the turn-off voltage V ,
Equation S.12.
T
8.3
Rederive the current-voltage relationship, Equation 8.7, in the presence o f a

constant (unmodulated) series resistance near the source. R and near the drain.
R . Discuss the effect o f each. I n particular, show that V
is independent o f
R.
t
LSEX
8.4 - Derive an expression for the drain conductance g dIj,ldV at a given V , i n the
saturation region. Assume that this conductance is due to the thickening o f the
depletion region near the drain, and approximate the latter by the one-dimensiona!
one-sided step-junction formula.
p
8.5
Derive an expression giving the electric field along the channel, and the carrier
velocity in the channel. Examine the validity of the use o f a carrier velocity which
is proportional to the electric field. Discuss aualitativelv how the current-voltage
characteristics may be affected by the real carrier velocity versus electric field
relationship.
8.6
Derive expressions giving the variation o f the conductance in the linear region
and o f the transconductance in the saturation region w i t h temperature, at a fixed
gate voltage.
8.7
I t is often stated that junction neid-eneci transistors are insensitive to the effect
of irradiation because their operation is independent o f the lifetime. Recalling
the "'carrier-removal" effect (Chapter 5). derive an expression giving the channel
conductance in the linear region as a function of radiation dose, for low radiation
doses. Usingthe data in Figure 5.17. estimate the electron dose at which the channel
conductance of the field-effect transistor used in the present chapter for illustration
is reduced by 1 0 % .
PART
111
SURFACE EFFECTS A N D
SURFACE-CONTROLLED DEVICES
Theory of Semiconductor
Surfaces
Surface Effects on p-n
Junctions
Surface Field-Effect
Transistors
Properties of the Si-SiCh

System
CHARACTERISTICS
SPACE-CHARGE
EQUILIBRIUM
THE
IDEAL
W O R K
SURFACE
CASE
MIS
STRUCTURE
F U N C T I O N
CHARGES,
OF
REGIONS-
A N D
DIFFERENCE;
STATES
9
T h e o r y of Semiconductor
Surfaces
I n Part I I we have considered semiconductor'devices w i t h the i m p l i c i t

assumption that their characteristics are determined strictly by b u l k
phenomena. I n realitv. surface effects-of-ten^ completely dominate the
characteristics o f semiconductor devices. I n fact, as-we had mentioned i n
the I n t r o d u c t i o n , one o f the principal reasons for the success o f the planar
technology is that planar junctions are covered by a thermally g r o w n
silicon dioxide layer. This reducesbut by no means eliminatesmany
surface effects and therefore results in better c o n t r o l o f device characteristics.
The properties o f the thermally oxidized silicon interface have been
studied very extensively; perhaps more extensively t h a n any other system
in the l o n g history o f semiconductor surface research. Because o f this,
and because o f their great importance in semiconductor device technology,
we w i l l n o w consider surface effects in some detail.
Surface effects on p-n junctions are p r i m a r i l y due to the fact that ionic
charges outside the semiconductor surface will induce an image charge i n
the semiconductor and thereby lead to the f o r m a t i o n o f surface spacecharge regions. This is illustrated schematically i n Figure 9.1 where we
show the idealized plane p-n j u n c t i o n structure t h a t was the basis o f our
previous studies. I f a surface space-charge region is formed, it w i l l modify
the j u n c t i o n space-charge region and can lead to changes in j u n c t i o n
characteristics.
263
T h e o r y of Semiconductor Surfaces
264
Surface space-charge
region^
(af
Fig. 9.1
Characteristics of Surface Space-Charge Regions
265
region in the case o f the metal-insulator-semiconductor capacitor w i l l be

in thermal equilibrium; i.e.. np = n?, and the Fermi level w i l l be constant
throughout the surface space-charge region.t I n this chapter we w i l l
restrict our attention to this case. I n order to treat characteristics o f
surface space-charge regions associated w i t h p-n junctions under bias,
such as i n case (b), we w i l l have to extend our consideration to nonequilibrium conditions. This w i l l be done i n the next chapter.
Ibl
Illustration of the role of surface effects in semiconductor devices.
u-Type inversion layer
(a) Idealized plane p-n junction.

(b) Same, with surface space charge induced by positive ions.
I n this chapter we w i l l first study the characteristics o f surface spacecharge regions i n thermal e q u i l i b r i u m . After that we w i l l study the metalinsulator-semiconductor ( M I S ) structure,! which has proved to be
extremely useful i n the s t u d y - o f semiconductor surfaces.
We. w i l l first
consider its characteristics in the ideal case; then we w i l l extend our
consideration to include the effect o f metal-semiconductor w o r k - f u n c t i o n
differences, interface and oxide charges, and interface states.
1
9.1
C H A R A C T E R I S T I C S O F S U R F A C E SPACEC H A R G E R E G I O N S E Q U I L I B R I U M CASE-
The three experimental structures customarily employed i n the study

o f surface effects and o f the characteristics o f surface space-charge regions
are illustrated in Figure 9.2. These structures are (a) the metal-insulatorsemiconductor capacitor structure; (b) the gate or field-plate c o n t r o l l e d
p-n j u n c t i o n ; and (c) the metal-insulator-semiconductor surface field-effect
transistor. I n all three cases we illustrate the case o f a p-type substrate,
w i t h a large positive voltage applied to the field plate.
A n i m p o r t a n t difference exists between the first structure and the
second t w o . I n the case of a metal-insulator-semiconductor capacitor,
no d-c current flow is possible across the space-charge region. This is
because any such current flow w o u l d have to go through the insulator. I n
contrast, current flow across the surface space-charge region is possible i n
structures (b) and (c) because i n these cases a contact is made to the surface
space-charge region through the n~ regions. Thus, the surface space-charge
(a)
(b)
(c)
Fig. 9.2 The experimental structures used to study surface effects.

(a) The metal-insulator-semiconductor capacitor structure.
(b) The gate-controlled p-n junction.
(c) The metal-insulator-semiconductor surface field-effect transistor.
In all cases V 0.
The energy band diagram o f a metal-oxide-semiconductor structure

with a p-type semiconductor, is shown i n Figure 9.3 for three bias conditions. Regardless o f the value o f the gate voltage, the Fermi level i n the
semiconductor remains constant since e q u i l i b r i u m holds. I n Figure 9.3a,
we show the conditions corresponding to a negative voltage applied to the
metal gate. This negative potential w i l l attract a positive charge i n the
semiconductor which, i n the case o f a p-type semiconductor, will consist
o f an enhanced concentrationaccumulation^-of holes near the oxidesilicon interface. The corresponding charge distribution is shown i n the
lower half of the figure.!
I f a small positive voltage is applied to the gate, a negative charge-will
t It should be noted that deviations from the equilibrium condition are possible also
in the case of the simple MIS structure. Such deviations can take place if the measurements are made under transient conditions, or if the insulator is imperfect and current
can leak across it. Although both of these conditions can be observed experimentally,
we will not consider them in this discussion.
X The conduction band of the silicon-dioxide layer is also indicated in this figure. This
conduction band is discussed in greater detail in Section 9.3.
3
* Because in most of the experimental studies the insulator has been silicon dioxide,
the term 'metal-oxide-semiconductor (MOS) structure' will be used interchangeably with
'metal-insulator-semiconductor (MIS) structure".
T h e o r y of Semiconductor
~266
Surfaces
be induced i n the semiconductor (Figure 9.3b). T h i s , at first, w i l l be due

t o holes being pushed away from the vicinity o f the interface, leaving
behind a depletion region consisting o f uncompensated acceptor ions as
shown i n the lower half o f the figure. The charge per u n i t area contained
i n the semiconductor Q w i l l then be given by the charge contained w i t h i n
s
this depletion reeion,

Q. -
-qN x
A
(9.1)
it
Characteristics of Surface Space-Charge Regions -
of the charge Q due to the electrons i n a very n a r r o w -type inversion

layer.i
Once an inversion layer is formed, the w i d t h o f the surface depletion
region reaches a m a x i m u m . This is because once the bands are pulled
down far enough for strong inversion to occur, even a very small increase
i n band bending, corresponding to a very small increase i n depletion
region w i d t h , w i l l result i n a very large increase i n the charge contained
w i t h i n the inversion layer. Thus, under such strong inversion conditions
the charge per u n i t area induced i n the semiconductor w i l l be given by
n
where x is the w i d t h o f the surface depletion region.
2, = 2* - qXjtemMx
-Type
' semiconductor
r
Oxide
V < 0.
c
E.
E
>
g g y a " c i * E,. o
"^Holes
r
where x
denotes the maximum width of the surface depletion region.
For most purposes, we are interested o n l y i n the cases o f depletion and
inversion. These cases can be described to an excellent a p p r o x i m a t i o n by
the depletion approximation that we have already employed i n our study
of p-n j u n c t i o n s .
When the semiconductor is depleted and the charge w i t h i n the semiconductor is given by E q u a t i o n 9.1, integration o f Poissons equation
yields the d i s t r i b u t i o n o f the electrostatic potential i n the surface depletion
region as
(9-2)
i m & x
Electrons
(metal)
267
(metal
i> = f [l
- - J
(9.3)
where the surface potential <b , w h i c h designates the t o t a l bending o f the

energy bands f r o m the bulk o f the semiconductor to the surface, is given by
s
Plx)
aK
(a) Accumulation of
majority carriers
near surface.
(b) Depletion of
majority carriers
from surface.
(c) Inversion: accumulation of

minority carriers near surface,
Fig. 9.3 Energy bands and charge distribution In an MOS structure under various bias
conditions, in the absence of surface states and work function difference.
-X "
We should note that this potential d i s t r i b u t i o n is identical w i t h the one we

derived i n Chapter 6 for the case o f a one-sided step j u n c t i o n . Indeed,
we w i l l see later that the theory o f surface space-charge regions for the
depletion and inversion cases is analogous to the theory o f one-sided step
junctions i n almost every detail.
The d i s t r i b u t i o n given by Equation 9.3 is shown i n Figure 9.4 i n c o m parison to exact numerical calculations o f the p o t e n t i a l d i s t r i b u t i o n i n the
surface space-charge region for a strongly inverted surface. The agreement
is clearly reasonable, except very near the surface. ( A l l calculations i n
this chapter are for silicon at 3 0 0 K . unless otherwise specified.)
4
I f we increase the positive potential applied to the gate, the w i d t h o f the

surface depletion region w i l l first increase. Correspondingly, the total
electrostatic potential v a r i a t i o n i n the silicon, as represented b y the bending
o f the energy bands, w i l l also increase as shown i n Figure 9.3c. However,
as the bands are bent further, the conduction band w i l l eventually come
close to the F e r m i level. W h e n this happens, the concentration o f electrons
near the interface w i l l suddenly increase very sharply. After this, most o f
the additional negative charge induced i n the semiconductor w i l l consist
The inversion laver is narrow because the minoritv carrier concentration drops to
~ 1 0 % of its value at .the interface over a distance 3A7"/o<f, where {., is the electric
field in the semiconductor, at the interface. This distance. 10 to 100 A . is much smaller
than the width of the surface depletion region.
Theory
268
Surfaces
C h a r a c t e r i s t i c s oi c u r i a t e o u a t e - v . i w t r g c
rveg.o,,
of the zero-bias depletion region o f a one-sided step j u n c t i o n (Chapter 61.

The two expressions are similar with cAînv) = 2d> taking the place o f the
built-in voltage (p . This analogy is indeed reasonable: when the surface
is inverted, a very t h i n n-type layer is formed, separated from the p-type
substrate by a depletion region. The one difference between the n^p stepj u n c t i o n case considered earlier and the inversion layer-substrate j u n c t i o n
that we are dealing w i t h here is that, i n the former, the rt-type conductivity
was brought about by a metallurgical process: i.e., the electrons were
F
. !
|
Depletion approximation, with tp =

s
26,
"
N
Computer calculations'*
= 10
1J
cm"
<Kv)
Fig. 9.4
Potential distribution in the surface depletion region.
One particularly interesting question concerns the p o i n t after which the

charge due to electrons i n the inversion layer becomes significant. A g o o d
approximate c r i t e r i o n for the onset o f this c o n d i t i o n of strong
inversionj
is that the electron concentration (per unit volume) near the surface should
exceed the concentration o f the substrate impurity ions, i.e.,
I
1
J*
10'
n, =
26 .
(9.6)
A s we discussed earlier, the surface depletion region w i d t h reaches a

m a x i m u m when the surface becomes strongly inverted. Accordingly, the
m a x i m u m w i d t h o f the surface depletion region can be estimated using the
depletion a p p r o x i m a t i o n , corresponding to the value o f the surface
potential at the onset o f strong inversion, ^ . ( i n v ) as denned above. T h i s
yields
/2iC e <k('nv)
s
Xdmai =
"
,10-
3
-
5 x1
(9.5)
for the case o f a p-type semiconductor. Under this c o n d i t i o n , the F e r m i

level at the surface w i l l be above the intrinsic F e r m i level by as m u c h as
it is below the intrinsic Fermi level i n the b u l k . A c c o r d i n g l y , the t o t a l
band-bending at the onset o f strong inversion w i l l be
d> (mv)
Depletion approximation
Computer calculations^
(9.7)
N o t e the similarity between this expression and that giving the w i d t h
Fig. 9.S
(cm- )
Surface depletion region width versus charge induced in the semiconductor.
introduced into the semiconductor by the i n t r o d u c t i o n o f donor ions. I n

contrast, in the present case the rc-type layer is induced by the electric
field applied with the field plate. Thus this j u n c t i o n , rather than being a
metallurgical j u n c t i o n , is a field-induced
junction.
We can readily calculate the depletion region w i d t h x as a function o f
the charge per unit area induced in the silicon O . Such calculations basedboth on the depletion approximation and o n more exact numerical
computations are shown in Figure 9.5. N o t e that, i n accordance w i t h
our earlier qualitative argument, the depletion region width first increases
as an increasing amount o f charge is induced i n the silicon. The rate o f
increasethe slope o f the straight linesis o f course dependent on the
impurity concentration. Eventually, when the surface is inverted and a
field-induced j u n c t i o n is formed, the surface depletion region reaches a
maximum width. This m a x i m u m surface depletion region w i d t h x
is
only a function o f the impurity concentration, the same way as the zerobias depletion region width o f a one-sided step j u n c t i o n is only a function
of the substrate impurity concentration. T h e relationship between x
d
i m A X
t Strictly speaking, the surface becomes inverted when the minority carrier concentration at the surface equals the majority carrier concentration at the surface, i.e.,
n, = p, Hi. However, this is not a useful criterion because the corresponding charge
of minority carriers per unit area is immeasurably small.
d m i x
- T h e o r y of Semiconductor
Surfaces
iopr
r
r
The
Ideal M I S (or M O S ) S t r u c t u r e
27I
and i m p u r i t y concentration, based on both the depletion a p p r o x i m a t i o n

and numerical computations, is shown i n Figure 9.6.
A n i m p o r t a n t quantity is the charge per unit area contained w i t h i n the
surface depletion region at and after the onset o f strong inversion. T h i s
is given by
QB -qKj*mMx
(9-8)
for a p-type semiconductor. A c c o r d i n g to E q u a t i o n 9.2. the t o t a l charge

in the silicon after strong inversion w i l l then be given by
Q. = Qr, + Q
(9.9)
Q.
(9.10)
or
Q = Q. -
Fig. 9.6
The maximum width of the surface depletion region as a function o
This relationship is shown in Figure 9.7 for three different d o p i n g concentrations. A g a i n , the depletion a p p r o x i m a t i o n is compared w i t h exact
numerical computations. The agreement i n this and i n the previous
figures demonstrates that the depletion a p p r o x i m a t i o n provides an
excellent description o f the surface space-charge region i n the depletion
and inversion regimes.
Impurity concentration.
9.2
a..
T H E IDEAL
MIS ( O R M O S )
STRUCTURE
Capacitance-Voltage Characteristics
I n the absence o f any contact potential or w o r k - f u n c t i o n differences

between metal and semiconductor (we w i l l discuss these i n Section 9.3).
any applied gate voltage w i l l partly appear across the oxide and p a r t l y
across the silicon. Thus.
Vc=K+
(9.11)
.where V and g> denote the potential v a r i a t i o n across oxide and semiconductor, respectively.
I n the absence o f any charges located at the interface between the oxide
and the semiconductor, Gauss' law requires that the electric displacement
should be continuous at this interface or
0
KS
B
Fig. V.7 The magnitude of the charge of the minority carriers within the inversion
layer as a function of the magnitude cf the total charge induced within the semiconductor.
= K4,
at the interface.
(9.12)
I f there are no charges present w i t h i n the oxide, the electric field i n it is

uniform and is given by
V.
..
Theory
Surfaces
where x, is the oxide thickness. The electric field at the silicon surface,
f r o m Gauss' law, equals
e
= -
(9.i4)
A c o m b i n a t i o n o f these three equations yields the voltage drop across the

oxide as
1/ = where C s K jx
0
When the gate voltage is zero or negative, no depletion region exists.

Thus the depletion approximation, and w i t h i t the above formula, lose
their meaning. However, the capacitance can be readily obtained by
considering that when the surface is accumulated, the semiconductor
merely acts as a resistor i n series w i t h the oxide capacitance. Thus the
measured capacitance w i l l be simply C.
I n the other extreme, when strong inversion sets i n , the w i d t h o f the
depletion region w i l l not increase with- further increase in gate voltage.
(9.15)
Q, = - = i
C
C.
is the capacitance per unit area o f the oxide layer.
The "ate voltage then w i l l be related to the characteristics o f the surface
if
space-charge region by
V=
+ <!>,.
(9.16)
0.8
0.7
0.6
The simplest measurable electrical characteristic o f an M O S structure is

its small-signal capacitance.
If Q
is the charge per u n i t area on the gate;

N
= 1.45 x 1 0
x = 0.2 n
this capacitance is given by

r as
_
V
= dV
'
\
^
0.5
1
-2
'
4
l
6
l
8
,
10
The capacicance-voltage characteristics of an MOS structure.
(9.18)
= _ <Q =
dd>
This w i l l take place at a gate voltage corresponding to a charge per u n i t

area- Q induced in the silicon and a surface potential <p\,(inv) = 2<p .
Substitution of these quantities into E q u a t i o n 9.16 yields the gate voltage
at the onset o f strong inversion, referred to as the turn-on voltage for
reasons that will become clear later,
s
where
c
cm'
Fig. 9.8
16
Computer calculations
(9 17)
_JQ ^_
0.9
(9.19)
is the capacitance per u n i t area o f the surface space-charge region i n the

semiconductor. Thus, the capacitance of the M O S structure is given b y
the series c o m b i n a t i o n o f C, a n d C . By eliminating *, we obtain the
f o r m u l a for the capacitance o f an M O S structure:
s
(9.20)
qN K,x
A
which predicts that the capacitance w i l l fall w i t h the square root o f the
gate voltage while the surface is being depleted.
+ cifmv).
(9.21)
Thus the capacitance w i l l level off and remain constant at a value given by
Equation 9.20 for V = V . The capacitance-voltage characteristic o f a
particular metal-silicon dioxide-silicon structure is illustrated i n Figure 9.8
based on both the depletion a p p r o x i m a t i o n and exact computer calculations. The depletion approximation evidently shows all the correct
features o f the more exact theory.
It should be pointed out that while we have used a p-type substrate i n
the preceding discussions, all o f the above considerations are equally valid
G
_C
= - ~
- T h e o r y of Semiconductor Surfaces
274
to the case of an n-type substrate w i t h proper changes i n symbols (e.g.,

0 to Q ) and signs. I n particular, the capacitance-voltage characteristics
w i l l have identical shapes b u t w i l l be m i r r o r images o f each other.
Table 9.1 at the end o f this chapter summarizes the i m p o r t a n t formulas
involved i n the analysis o f surface space-charge regions for both p- and
n-type semiconductors.
n
b.
Frequency Effects
I n the preceding calculation o f the capacitance we had assumed that

u p o n a change i n gate voltage all o f the incremental charge appears at the
edge o f the depletion region. T h i s consideration led to the simple formula
for the capacitance o f the semiconductor space-charge region,
r
^ 2
CiVg
The
Ideal MIS (or M O S ) Structure
275
This is, i n fact, what happens when the measurement frequency is high.
I f however the measurement frequency is l o w enough that r e c o m b i n a t i o n generation rates can keep up w i t h the small signal variation, then the
recombination-generation mechanism w i l l lead to charge exchange w i t h
the inversion layer i n step w i t h the measurement signal. I n that case, the
capacitance measured w i l l approach that o f the oxide layer alone.
T o understand this better, let us consider what happens when a positive
voltage applied to an M O S structure is increased by a small a m o u n t .
Instantaneously, as the voltage is increased, more negative charge is
induced i n the silicon. A t high frequencies, holes w i l l be pulled o u t o f the
p-type semiconductor and the w i d t h o f the depletion region w i l l increase
slightly, as shown i n Figure 9.9a. I f however electron-hole pairs can be
generated fast enough, i.e., before the voltage is reduced again, the
generated holes will replenish the holes pulled out from the edge o f the
depletion region and the extra electrons w i l l appear in the inversion layer.
Thus the incremental negative charge brought about by the incremental
increase i n gate voltage w i l l appear at the oxide-sDicon interface, as
illustrated in Figure 9.9b. As a result, the capacitance measured w i l l be
that o f the oxide layer alone, C.
As a consequence, the capacitance-voltage characteristics o f metaloxide-semiconductor structures are frequency dependent. Experimental
measurements on an M O S structure corresponding to the computations
given i n Figure 9.8 are shown i n Figure 9.10, for various measurement
-dQc
CJVe
(b)
-20
-15
-10
-5
10
15
20
-dQc
Fig. 9.9
Illustration of the variation of charge distribution in an MOS structure at
(a) high frequencies; (b) low frequencies.
Fig. 9.10 The effect of measurement frequency on the capacitance-voltage characteristics of MOS structures.'
T h e o r y of S e m i c o n d u c t o r -
276
Surfaces
frequencies. In agreement w i t h the above argument, it is evident that at

low frequencies the capacitance approaches the oxide capacitance at the
onset o f strong inversion.
The transition from the " h i g h - " to the "'low--' frequency type capacitancevoltage characteristics can also be brought about at a given measurement
frequency by an increase in the carrier recombination-generation rate, for
instance by an increase in temperature or by i l l u m i n a t i o n .
3
Low-frequency type characteristics can also be observed at relatively

high measurement frequencies due to two-dimensional effects, such as the
existence o f an inversion layer beyond the metal field plate. This phenomenon w i l l be discussed i n Chapter 12. Also, in gate-controlledp-n junctions
and i n M O S surface field-effect transistors, where contact is made to the
inversion layer t h r o u g h metallurgical junctions (see Figure 9.2), m i n o r i t y
carriers can be supplied to the inversion layer through this external contact.
Thus low-frequency type characteristics will prevail up to much higher
frequencies because the supply o f m i n o r i t y carriers is no longer dependent
on a generation-recombination process.
The
Ideal MIS (or M O S )
Structure
Combining Equation 9.25 w i t h 9.16 and 9.21 yields

g =-n ClV
n
V ).
(9.26)
Thus, i f the electron m o b i l i t y /< is independent o f the gate voltage, a"

straight line should result i f conductance is plotted as a function o f gate
'
1x10""
0.8
S
1.3 x 10' e r a
.x. = 0.26 u
= 3.6 x 10" cm
fheory (ji = 660 c m / v sec)

n
c.
Channel Conductance
Using the M O S surface field-effect transistor, structure (c) o f Figure 9.2,

an important complementary measurement can be performed. This
measurement pertains to the conductance o f an inversion layer or channel
induced by a positive gate voltage ( i n the case o f a />-type substrate). T h i s
conductance is given by
Z
*'o(x) dx
LJ
(9.22)
where Z is the w i d t h o f the conducting channel ( i n a direction n o r m a l to

Figure 9.2), L is the distance o f separation between the two n^ regions,
and a(x) is the conductivity o f the inversion layer, given by
-4
-3
-2
-1
V
a(x) = wA*l
(9-23)
The point x = x denotes the distance below the surface where the semiconductor is j u s t intrinsic. Thus the channel conductance becomes
Fig. 9.11
J
1
I
2
\y)
The channel conductance and gate capacitance of an MOS transistor.
g = - 7 , " Q
(9-24)
voltage. Pertinent experimental results are shown in Figure 9.11 along

w i t h the corresponding capacitance-voltage measurements. Note that i n
accordance with the above argument, the conductance is indeed proportional to the quantity {V V ). The voltage at which the conductance
begins to increase is V , which explains why this voltage is called the
turn-on voltage. Note that the onset o f conductance closely corresponds
to the rise in capacitance.
0
where O is the charge density per unit area due to mobile carriers i n the
inversion layer. C o m b i n i n g E q u a t i o n 9.24 w i t h 9.10 yields
n
8=
- f / U Q . - G * ] .
id
(9-25)
This simple derivation o f the channel conductance applies only when the
voltage drop V between the two n regions is small i n comparison to-.
r
278
Effects on M O S Characteristics
(V
V ). When this c o n d i t i o n is not met. as is often the case i n practical
operation of M O S transistors, the expression for the conductance along
the surface takes on a more complicated f o r m . This will be considered in
detail in Chapter 11.
G
279
9.3
a.
EFFECT O F W O R K F U N C T I O N
CHARGES, A N D STATES
O N MOS C H A R A C T E R I S T I C S
Vc = o
I
<tv
DIFFERENCE,
^ 7
W o r k Function Difference
The electron energies at the F e r m i level i n the metal and i n the semiconductor of an M O S structure w i l l , i n general, be different. Such an
energv difference is usually expressed as a difference i n work
functions,
which is the energy required to remove an electron from the F e r m i level i n
a given material to vacuum. W h e n the.metal o f an M O S structure is
shorted to the semiconductor, electrons will flow from the metal to the
semiconductor or vice versa u n t i l a potential will be built up between the
two which will counterbalance the difference i n w o r k functions. When
equilibrium is reached, the F e r m i level i n the metal is lined u p w i t h the
Fermi level in the semiconductor. Therefore, there will be an electrostatic
potential variation from one region to the other, as illustrated i n Figure
9.12a for the case o f an a l u m i n u m / s i l i c o n dioxide/p-type silicon sample.
I n this figure we also show the conduction band o f the oxide. The energy
required to move an electron f r o m the metal Fermi level into the conduct i o n band of the oxide is called the metal-oxide barrier energy. The energy
required to move an electron f r o m the silicon valence band to the conduct i o n band o f the oxide is the silicon-oxide barrier energy. These barrier
energies which are related to the respective w o r k functions modified by the
presence o f the oxide can be independently measured by i l l u m i n a t i n g
M O S structures with light o f increasing frequency until the p h o t o n energy
w i l l be high enough to excite electrons into the conduction band of the
oxide, thereby resulting i n electronic conduction between metal and
semiconductor. The barrier energies in the M O S svstem are discussed
further in Chapter 12.
T o derive the effect o f a w o r k function difference on M O S characteristics,
it is easiest to consider the c o n d i t i o n i n which just enough gate voltage is
applied to counterbalance the w o r k function diffeience and a flat-band
condition is maintained in the semiconductor, as illustrated i n Figure
9.12b. The gate voltage required to b r i n g about the fiat-band condition is
called the fat-band voltage
V .
r B
Va|'
-,
; AI ;
Si0
Si0
m m '
2
- p-Type Si
''////////.
'/A
la)
(b)
Fig. 9.12 The effect of metal-semiconductor work function difference on the potential
distribution in an MOS structure.
(a) Conditions for V = 0.
(b) Flat-band condition.
c
The flat-band voltage is clearly the difference o f the modified w o r k

functions indicated i n Figure 9.12b.
V FB
$ '
(9.27)
We w i l l discuss the role o f the flat-band voltage on the M O S characteristics later i n this section.
b.
C h a r g e s in the Insulator
Consider a sheet charge per unit area 0 w i t h i n the insulator o f a metalinsulator-semiconductor structure as shown i n Figure 9.13. Under the
conditions o f zero gate voltage (Figure 9.13a), this sheet charge w i l l
induce an image charge partly in the metal and p a r t l y i n the semiconductor.
The resulting field distribution is indicated i n the lower part o f Figure
9.13a. neglecting work function differences and electrostatic potential
variation i n the semiconductor.
I n order to b r i n g about a flat-band c o n d i t i o n (i.e.. no charge induced
in the semiconductor), we have to apply a negative voltage to the metal, as
shown i n Figure 9.13b. W i t h increasing negative voltage, we are p u t t i n g
more negative charge on the metal and thereby shift the electric field
distribution downwards until the electric field reaching the silicon surface
becomes zero. Under this condition the area contained under the electric
field d i s t r i b u t i o n is the flat-band voltage V .
O n the basis o f this figure
r B
T h e o r y of S e m i c o n d u c t o r
280
and Poisson's equation, V
FB
Surfaces
J -
u w r ^ w i M U U
can be shown by superposition o f individual elements o f such a cHarce

distribution that the corresponding flat-band voltage is given by
is given by
t n e c t s on n u j
2 .
(9.23)
Thus the fiat-band voltage not only depends on the density o f the sheet
charae 0 but also on its location within the insulator. W h e n the sheet
Metal
Insulator
Fig. 9.13
Semiconductor
The effect of a sheet charge within the insulator.

0
charge is next to the metal, i t wiT induce no image charge i n the silicon
and, therefore, have no effect o n the semiconductor surface. I n the other
extreme, when the sheet charge is located next to the semiconductor, i t
w i l l exert its m a x i m u m influence, and lead to a flat-band voltage o f
x,0
Ke
C.
p(x) clx.
(9.30)
Such space charges i n the insulator can be due to several causes such as
ionic contamination or traps ionized as a result o f exposure to ionizing
radiation. We w i l l discuss both o f these i n detail i n Chapter 12.
Metal
Insulator
Semiconductor
Fig. 9.14 The effect of an arbitrary space-charge distribution within the insulator.
(Flat-band condition.)
(a) Conditions for V = 0.

(b) Flat-band condition.
FB
f " x
- C, Jo -x
Semiconductor
Metal
F B =
(9.29)
We w i l l see i n Chapter 12 that there is indeed such a sheet charge associated

w i t h the silicon dioxide-silicon interface. Its density is denoted by Q.
The more general case o f an arbitrary space-charge d i s t r i b u t i o n w i t h i n
the insulator is illustrated i n Figure" 9.14 for the flat-band c o n d i t i o n . I t
Space charge or interface charge i n the insulator, and a metalsemiconductor w o r k function difference w i l l both lead to a translation o f
the flat-band point from V = 0 along the voltage axis. Experimental
observation o f the flat-band voltage shift is illustrated i n Figure 9.15.
N o t e that the entire capacitance-voltage characteristic is translated i n a
parallel manner along the voltage axis by the same amount as the flat-band
point. The total translation o f the flat-band voltage or o f any other well
defined point on the capacitance-voltage (or, alternatively, the channelconductance versus gate voltage) characteristic w i l l then be given- by
G
F B = Gj/s -
- f
C Jo
0
1 0
- p{x) dx.
x
(9.31)
282
Effects on M O S Characteristics
83
these states w i l l be lifted far above the Fermi level. Thus their p r o b a b i l i t y
of being occupied by electrons w i l l be small and most o f them w i l l be
unoccupied. A s the surface is depleted and inverted, the states are pulled
well below the F e r m i level and their probability o f occupation by electrons
Thus direct comparison between theory and experiment is possible by

p l o t t i n g experiments as a function o f (V V ) rather than as a function
of V alone. Experimental data are presented i n this manner for thermally
oxidized silicon i n Figures 9.16 and 9.17. These data show the effect o f
substrate i m p u r i t y concentration and o f oxide thickness on the shape o f
the characteristics in comparison w i t h the exact theoretical computations.
I t is evident that a uniform translation along the voltage axis brings about
a very good agreement between theory and experiment i n all cases.
c
FB
c.
Surface States
I n Figure 9.3 we have shown how the application o f a potential to the

gate o f an M O S structure will result in a movement o f the energy bands
relative to the F e r m i level near the surface. I f there are states w i t h i n the
forbidden gap concentrated at the surface, such as the surface recombination-generation centers discussed i n Chapter 5. the probability o f the
occupation of these surface states will change as a result o f the variation
in band bending. This w i l l happen because as the bands are moved up
and down at the surface, the p r o x i m i t y o f the energy level o f the surface
states to the Fermi level will change. This effect is illustrated i n Figure
9.18 where we show the charge condition o f surface states w i t h energy
level near the middle o f the forbidden gap. As the surface is accumulated.
-10
-5
5
V
10
-
15
20
25
V (vj
F i g . 9.16 T h e effect of s u b s t r a t e i m p u r i t y c o n c e n t r a t i o n o n t h e c a p a c i t a n c e - v o l t a g e
characteristics of M O S s t r u c t u r e s . '
will approach unity. T r a d i t i o n a l l y , such states whose charge can be

readily exchanged with the semiconductor are.called fast surface states.
Thus the charge i n the fast surface states w i l l vary w i t h the band bending
or surface potential tjt . O w i n g to this change, the capacitance-voltage or
conductance-voltage characteristics will be displaced from the theoretical
characteristics br an amount which itself varies with the surface
potential.
A
Theory
.-Experiment.^
1 No fast
\
\ surface
\
\ states
\
U" 0.8
0.7
Experiment,
Fast surface states 1
1
1
1
20
-15
-10
r*1
-5
10
Fig. 9.19 The effect of fast surface states on the capacitance voltage characteristics of
MOS structures.
7
The result may appear either as steps or kinks in the characteristics or, for
the case o f a c o n t i n u u m o f states i n the forbidden gap, as a gradual
distortion o f the characteristics. A n experimental illustration o f the effect
of fast surface states on M O S characteristics is shown i n Figure 9.19.
,V, 0.2l
1.45 x 1 0
16
cm"
READING
Theory
Experiment (recorder tracing)
A more comprehensive treatment of the theory of semiconductor surfaces along with" '
a bibliography of the earlier work on the subject can be found in A . Many, Y . Goldstein,
and N . B. Grover, Semiconductor Surfaces, Wiley, 1965.
0.1
-10
-5
10
15
20
25
REFERENCES
V , (v)
r
i . 9.17 The effect of oxide thickness on the capacitance-voltage characteristics
of
Fig.
f
MOS structures.
(a)
lb)
9 18 The charge condition of a particular set of fast surface states with energy
near the middle of the forbidden gap as a function of surface potential vanation.
r i g
(a) Surface-accumulated: states are unoccupied.

(b) Surface inverted: states occupied.
REFERENCES
CITED
1. The MOS structure was first proposed as a voltage variable capacitor by J . L . Moll,
"Variable Capacitance with Large Capacity Change," Wescon Convention Record,
Part 3, p. 32 (1959); and by W. G . Pfann and C . G . B. Garrett, '"Semiconductor
Varactors Using Surface Space-Charge Layers," Proc. IRE (Correspondence), 47,
2011 (1959). Its characteristics were then analyzed in detail by D . R . Franld, ."Some
Effects of Material Parameters on the Design of Surface Space-Charge Varactors,"
Solid-State Electron., 2, 71 (1961); and by R . Lindner, "Semiconductor Surface
Varactor," Bell System Tech. J., 41, 803 (1962). It was first employed in the study of
thermally oxidized silicon surfaces by L . M . Terman. "An Investigation of SurfaceStates at a Silicon/Silicon Dioxide Interface Employing Metai-Oxide-Silicon Diodes,"
Solid-State Electron., 5, 235 (1962); and by K . Lehovec and A. Slobodskoy, "Field
Effect-Capacitance Analysis of Surface States on Silicon," Phys. Stat. Solidi, 3, 447
(1963).
2. The theory of surface space-charge regions was developed by W. L . Brown, "jV-type
Surface Conductivity on iMype Germanium." Phys. Rev., 91, 518 (1953); C . G . B .
Garrett -and W. H . Brattain, "Physical Theory of Semiconductor Surfaces," Phys.
Theory of Semiconductor Surfaces
286
Rev., 99, 376 (1955); R . H . Kingston and S. F . Neustadter, "Calculation of the

Space Charge. Electric Field, and Free Carrier Concentration at the Surface of a
Semiconductor," J. Appl. Phys., 26, 718 (1955).
3. A . S. Grove, B . E . Deal, E . H . Snow, and C . T . Sah, "Investigation of Thermally

Oxidized Silicon Surfaces Using Metal-Oxide-Semiconductor Structures," SolidStale Electron., 8, 145 (1965).
4. C . E Young,. "Extended Curves of the Space Charge, Electric Field, and Free
Carrier Concentration at the Surface of a Semiconductor, and Curves of the Electrostatic Potential Inside a Semiconductor," J. Appl. Phys., 32, 329 (1961).
5. A . S. Grove, E . H . Snow, B . E . Deal, and C . T . Sah, "Simple Physical Model for the
Space-Charge Capacitance of Metal-Oxide-Semiconductor Structures," J. Appl. Phys.
' 35, 2458 (19~64).
6. E . H . Nicollian and A . Goetzberger, "Lateral A C Current Flow Model for MetalInsulator-Semiconductor Capacitors," IEEE Trans. Electron Devices, ED-12, 108
(1965); S. R . Hofstein and G . Warfield, "Physical Limitations on the Frequency
Response of a Semiconductor Surface Inversion Layer," Solid-Staie Electron., 8,
321 (1965).
7. B . E . Deal. M . Sklar. A . S. Grove, and E . H . Snow. "Characteristics of the SurfaceState Charge (Q) of Thermally Oxidized Silicon," J. Electrochem. Society, 114,
(March 1967).
Verify the expression given for the capacitance of the surface space-charge region,
Equation 9.19.
9.2
Derive an expression for the surface field required:

(a) T o make the surface exactly intrinsic.
(b) T o bring about strong inversion.
9.3
Calculate and plot:

(a) the charge in the semiconductor,
(b) the surface field, and
(c) the surface potential
at the onset of strong inversion, as a function of the impurity concentration in the
semiconductor.
9.4
Derive an expression giving the change of the turn-on voltage of an M O S structure

as a function of temperature.
9.5
(a) Derive a formula giving the ratio of the minimum high frequency capacitance
of an M O S structure C
to the oxide capacitance C as a function of oxide
thickness and substrate impurity concentration,
(b) The impurity concentration in the silicon in a region near the oxide-silicon
interface may be different from the concentration deep in the substrate; for
instance, due to redistribution during oxidation. In such a case, what impurity
concentration would be inferred from the C / C ratio?
min
m i D
9.6
9.8
9.9
Exact calculations of the capacitance C of an M O S structure at flat band show

it to be less than the capacitance of the oxide layer C (see. for example. Figure
9.8). Give a physical explanation for this fact.
ISf
/Jfe
"
We have seen that the flat-band voltage V depends on the metal-semiconductor

work function difference as well as on the surface-state charge density. Devise a
way by which these two factors can be determined from measurements of the fiatband voltage of M O S capacitors having various oxide thicknesses.
rB
Calculate the change in flat-band voltage corresponding to:

(a) A uniform positive charge distribution in the oxide.
(b) A triangular distribution, which is high near the metal and zero near the silicon.
(c) A triangular distribution which is high near the silicon and zero near the metal.
Let the total density of ions be 1 0 cm" in all three cases and consider a 0.2 p.
thick oxide film.
1!
9.10 Derive the charge per unit area Q in fast surface states as a function of surface
potential for:
(a) Single-level centers located at E, = , , with density A' ( c m ) .
(b) Centers uniformly distributed in energy throughout the forbidden gap, with
density D ( c m ev- ).
Assume that the surface states are acceptor type. i.e.. negative when occupied by
an electron, neutral otherwise.
-1
_ !
^
w
dSt
"
e
gr
*
f
9.11 Derive the capacitance-voltage characteristics of an M O S structure in the presence

of single-level and uniformly distributed fast surface states as described in Problem
9.10. ~
PROBLEMS
9.1
Problems
237
9.7 Derive the transition frequency between "low"' and ""high" frequency type capacitance-voltage characteristics by equating the generation rate within the surface
depletion region (see Chapter 5) to the charging current.
f(
r
'
f
f
f
f
r
r
f
<
I
r
..
CHARACTERISTICS OF SURFACE
SPACE-CHARGE
REGIONS
NON-EQUILIBRIUM
CASE
THE GATE-CONTROLLED
STRUCTURE
RECOMBINATION-GENERATION
IN T H E S U R F A C E S P A C E - C H A R G E
REGION
FIELD-INDUCED
AND C H A N N E L
SURFACE EFFECTS O N
JUNCTION
BREAKDOWN
VOLTAGE
DIODE
JUNCTIONS
CURRENTS
10
Surface Effects on p-n
Junctions
Perhaps the principal motivation for studying the properties of semiconductor surfaces is that surface effects can dominate the characteristics
of p-n junctions and transistors. I n Chapter 9 we saw how the metalinsulator-semiconductor capacitor structure can be employed to study the
characteristics o f surface space-charge regions. In this chapter we now
show how surface space-charge regions can affect the characteristics o f
p-n junctions.
We begin with an extension of the theory o f surface space-charge regionsto non equjlibjjujTtconditions such as are encountered" in the vicinity o f
a biased p-n junction. We then develop the characteristics o f the gatecontrolled diode structure which takes the place o f the metal-insulatorsemiconductor structure as the principal experimental tool. We next
consider the recombination-generation processes taking place in the surface
space-charae reaion. and their influence on the current gain o f transistors.
Following this, we study breakdown phenomena associated with fieldinduced junctions and show how they can lead to catastrophic changes in
both reverse currents o f diodes and in current gain o f transistors. Finally,
we consider how surface fields can influence the breakdown voltage o f
planar junctions.
;
289
S u r f a c e E f f e c t s o n f>-n J u n c t i o n s
290
10.1
C H A R A C T E R I S T I C S O F S U R F A C E SPACECHARGE REGIONSNON-EQUILIBRIUM
CASE
1
A l t h o u g h the gate-controlled diode structure shown i n Figure 10.1a

refers to a diode having a /"-type substrate, the following discussion is
equally valid for a diode w i t h an n-type substrate i f appropriate changes i n
Characteristics of Surface Space-Charge
Regions
291
equilibrium conditions (i.e.. when the applied j u n c t i o n voltage Vj = 0)

are illustrated i n Figure 10.2. I n this energy b a n d representation electron
energy, as represented by the conduction and valence bands, is shown as a
function of the two directions x and y corresponding to the axes o f Figure
10.1. I n the absence o f surface fields the energv bands do not vary i n the
i direction. The only variation is in the y d i r e c t i o n and is due to the
built-in voltage V
is
applied to the gate, the conduction band at the surface is brought close to
the Fermi level and the surface o f the p region becomes inverted.
(V ,
the turn-on voltage, denotes the gate voltage necessary to cause inversion
of the p surface.) The energy bands corresponding to this case are
illustrated i n Figure 10.2b. A field-induced /unction n o w exists between
the '?-type inversion layer and the u n d e r l y i n g p-xype silicon. Since bias is,
not applied across this j u n c t i o n , i t is in e q u i l i b r i u m and is characterized bv
the same Fermi level as the metallurgical j u n c t i o n . The" surface spacecharge region under these conditions is described by the equilibrium theory,
discussed in the previous chapter.
B
T w o aspects of the e q u i l i b r i u m theorv o f surface space-charge regions

should be recalled at this p o i n t . First, the t o t a l bending o f the bands, as
designated by the surface potential <f> is l i m i t e d t o some value less than the
band gap of the semiconductor, i.e.. about 1 v o l t for Si. M o r e specifically,
CD corresponding to strong inversion is given t o a good a p p r o x i m a t i o n bv
c/>,(mv) = 2<f> where d> is the Fermi p o t e n t i a l o f the substrate. This
value is the built-in voltage o f the field-induced j u n c t i o n . Second, the
width o f the surface depletion region x first increases w i t h increasing V
and then reaches a m a x i m u m value i "
when the surface becomes inverted.
This width is the zero-bias depletion region w i d t h o f the
field-induced
junction.
s
Ff
Fl
d m a x
(b)
F i g . 10.1 (a) The gate-controlled diode structure.
(b) Idealized representation of the portion of the surface enclosed by the dashed frame
in ().'
the signs of the applied voltages are made. A l l voltages in this discussion
are applied with the substrate held at ground potential. F o r simpjicity^
ih_7r-type region is assumed J o be_much more heavily doped than the
substrate. The condition of the surface o f the substrate w i t h i n the dashed
frame in Figure 10.1a is first examined at different values o f gate voltage
V
and reverse voltage V , in the absence o f surface states or w o r k
function difference. This region is shown i n a more idealized form in
Figure 10.1b.
c
The idealized j u n c t i o n and the energy bands pertaining to thermal
Figure 10.3 illustrates the n o n - e q u i l i b r i u m c o n d i t i o n which exists when

a reverse bias V is applied to the j u n c t i o n . F i g u r e 10.3a shows the case
in the absence o f surface fields. I n Figure 10.3b. a positive gate voltage is
applied, but it is not large enough to invert the surface o f the /^-region.
This is denoted bv the condition V < V (Vjf).
where V (V }
'
gate
voltage necessary to cause inversion o f the p region i n the presence o f a
reverse bias V .
This voltage is larger than in the zero junction bias case.
This is because application o f the reverse bias lowers the quasi-Fermi level
for electrons so that even i f the bands at the surface are bent as deeplv as
in the equilibrium case shown in Figure 10.2b. the band bending is still
insufficient to bring the conduction band near enouch to the quasi-Fermi
level for electrons to cause inversion. As a result, the surface is only
depleted.
s
292

Characteristics of Surface Space-Charge
Regions
293
I n the case illustrated in Figure 10.3c, the voltage applied to the gate is
large enough to overcome the influence o f the reverse bias and an inversion
layer is formed at the surface o f the p reg.on. I n terms o f the band diagram,
the bands are now bent deeply enough to bring the conduction band near
the quasi-Fermi level for electrons. Once an inversion layer is formed, i t
constitutes a region o f high conductivity which is at essentially the same
potential as the n region. The surface potential <p, at the onset o f strong
inversion is now given, to a good approximation, by
Wmv)
= V
+ 2<p .
Fp
(a) No surface held

Fig. 10.3
p-n Junction
(b.i Depletion
(c) Inversion
under che influence
fieldsreverse
bias
of a reverse bias V and smaller in the presence o f a f o r w a r d bias

applied to the p-n j u n c t i o n .
R
case
Otherwise, the description o f the surface space-charge region will be

unaltered. The electrostatic potential variation i n the depletion region
w i l l be given by
(10.1)
= ^(1
As i n the e q u i l i b r i u m case, the surface depletion region reaches a m a x i m u m

width x
at inversion. This width, however, is now a function o f the
reverse bias V , and is in fact the reverse-bias depletion region w i d t h o f
the field-induced j u n c t i o n formed between the -type inversion layer and
the underlying p region.
The most i m p o r t a n t features of the above discussion are summarized i n
Figure 10.4, where the charge distribution and energy bands of an inverted
/'-type substrate are shown as a function o f the distance x from the surface,
both for the e q u i l i b r i u m case (corresponding to Figure 10.2b), and for the
reverse biased case (corresponding to Figure 10.3c).
The characteristics o f the surface space-charge region wiil be given by
the depletion a p p r o x i m a t i o n in a manner analogous to the e q u i l i b r i u m
case considered i n the previous chapter. The only difference is that the
c o n d i t i o n o f the onset o f strong inversion corresponding to the criterion
n, = / V j will now be c6,(inv) = Vj + 2<h because o f the splitting o f the
quasi-Fermi levels brought about by the applied bias. Thus the surface
potential <_> at the onset o f strong inversion will be larger in the presence
d m i K
of surface
- ~
(10.2)
where the surface potential i which designates the total bendtn- o f the
energy bands, is given by
(10.3)
2K,e
The maximum width o f the depletion region w i l l now be given by

2K e {V
!
2i J
F
(10.4)
where we have used Equation 10.1 for <p,.(inv).
The charge per unit area induced i n the silicon prior to the onset o f
strong inversion is given by
2, =
-qN x
A
(10.5)
while after strong inversion it is given by
Fp
where
Qn + Q
(10.6)
Z'Vj-Trfinax.
(10.7)
QB =
294
Characteristics of Surface Space-Charge Regions
Calculations based on these relationships are shown i n Figure 10.5

where we show the surface depletion region width x as a function o f the
charge induced i n the silicon Q, for various values o f applied j u n c t i o n
bias. ( A l l calculations i n this chapter are for silicon at 3 0 0 K unless
m 10' C m -
10 v
/
P
29S
S
6
4
J
*
0.6
/ l > = -0.1: -0-2v

.
. 1,
1
f Computer calculations"
t
0
r, - o
5 x 10 "
|fi,'flj ( c m - )
(a)
Fig. 10.5 Surface depletion region width versus charge induced in the semiconductor
under various junction bias conditions.
(W
Fig. 10.4 Idealized charge distribution and energy band variation of an inverted p-region
in a direction normal to the surface, some distance away from the junction.
(a) Equilibrium case (V, = 0), also shown in Fig. 10.2b.
(b) Reverse bias case ( ^ ^ 5 v), also shown in Fig. 10.3c.
1
otherwise specified.) The lines are based on the two limits o f the depletion
a p p r o x i m a t i o n ; the points are based on more exact computer calculations.
I n Figure 10.6 we show the m a x i m u m width o f the surface depletion
region ^
as a function o f the applied reverse bias, for various substrate
i m p u r i t y concentrations. N o t e the similaritv between this figure and
Figure 6.9 which shows depletion region width versus applied reverse bias
for a one-sided step j u n c t i o n .
d m a x
Fig. 10.6 The maximum width of the surface depletion region as a function of the
magnitude of the junction reverse voltage. The parameter is substrate impurity concentration.
Gate-Controlled
Diode
Structure
277
296
0'
V , = 10'" cm"
til
i> = - 0 . 2 V
Calculations of the gate-to-substrate capacitance-voltage characteristics

of a gate-controlled diode are shown in Figure 10.8. Note that prior to
the onset o f strong inversion the characteristics follow the depletion
a p p r o x i m a t i o n ; then, at the onset o f strong inversion, in the low frequency
case shown here, the capacitance rapidly increases to the oxide capacitance.
(As discussed in Chapter 9, in the case o f gate-controlled diodes, minority
_carriers can be supplied to the inversion layer through the external circuit.
////'
I I
/
W ~ ~ ]
///
7rr7 r
4
1
1 /
/ / //
Depletion
approximation
/' <
()1
0
/ / //
^ a L _ J t !
2
4
6
3
^rr
10
QJq\
Fig. 10.7
,
12
Computer
_
calculations'
<
i
i
I
I
I
14
16
18 20 x 1 0 "
|cm" )
The magnitude of the charge of minority carriers within che inversion layer
as a function of the magnitude of the total charge induced wichin che
semiconductor,
under various junction bias conditions.
The charge due to electrons in the inversion layer is shown as a function

o f the total charge induced in the silicon for various values o f forward and
reverse bias in Figure 10.7.
10.2
GATE-CONTROLLED
DIODE
STRUCTURE
Fig. 10.8 Theoretical low frequency gate-co-substrate capacitance-voltage characteristics of a gace-controlled diode structure under various applied junction bias conditions.
Values of surface potential A, are indicated.
1
The relationship between the potential applied to the gate and the characteristics o f the surface space charge can be calculated exactly as in the case
o f the metal-insulator-semiconductor structure. Thus the gate voltage is
given by
V
= V + <p
(10.8)
= - + CP
C
(.10.9)
or
V
and the gate-to-substrate capacitance
For this reason, the " l o w frequency" type characteristics will be observed
to m u c h higher frequencies than in the case o f M O S capacitors.)
The point of onset o f strong inversion is considerably displaced from
the equilibrium case as the reverse bias is increased. Its location, the
turn-on uoltage, can be calculated from Equation 10.9 by substituting the
conditions for strong inversion. Q = O and <p. = 2(p
+ Vj. This
vields
s
per unit area, as before, will be
Fs
OR
given by the series c o m b i n a t i o n o f the oxide and semiconductor capaci-
+ :<p
F p
Vj.
(10.11)
tances,
C =
(10.10)
Because 0 itself is a function o f Vj, the change in turn-on voltage w i t h

applied junction bias V can be considerably larger than the j u n c t i o n
bias itself. This is borne out by the characteristics shown in Figure 10.8.
B
Co
4-
c,
0\sy
contribute to the generation current (Figure 10.10a). When the surface

under the gate is inverted, centers within the surface depletion region, i.e.,
the depletion region o f the field-induced j u n c t i o n , also contribute to the
generation current which is, therefore, larger than in the first case (Figure
10.10c). This c o n t r i b u t i o n to the generation current is related to the
width of the surface depletion region x .
Thus, .when the surface is
depleted and x is increasing with increasing gate voltage, this current
Gaie
SiO
Surface
r
depletion-*J
region
layer
Depletion
region of
metallurgica'
junction
ICj
Fig. 10.9 Direct observation of the surface space-charge region of a reverse biased
gate-controlled n p diode, cleaved through the junction.
T
As in the case o f the simple metal-insulator-semiconductor capacitor

structure, the effect o f a w o r k function difference or of charges in the oxide
w i l l be a displacement o f the characteristics along the voltage axis by the
flat-band voltage.
The theory o f surface space-charge regions under n o n - e q u i l i b r i u m
conditions has been verified by extensive electrical measurements on gatecontrolled diodes. Direct observations of the shape o f surface depletion
regions have also tended to verifv the qualitative features o f this theory.
A n example of such an observation made by a scanning electron m i c r o probe, is shown i n Figure 10.9.
3
Fig. 10.10 Illustration of the effect of variation in the nature of the surface space-charge
region on the reverse current of an n~p diode at a fixed reverse voitage.
1
10.3
RECOMBINATION-GENERATION
PROCESSES
IN T H E SURFACE
SPACE-CHARGE
REGION
0 f
We have seen in Chapter 6 that the r o o m temperature reverse current o f

silicon p-n junctions is due to electron-hole pairs generated through the
action o f recombination-generation centers within the depletion region.
Thus we expect that the magnitude of the reverse current depends O D t n e "
total number o f such centers included within the j u n c t i o n depletion region.
W i t h this in m i n d , let us now consider the reverse current versus gate
voltage characteristics o f a a t e - c o n t r o l l e d n~p junction depicted in Figure
IO.IOT
When the surface under the gate is accumulated, only those centers
which are within the depleiion region of the metallurgical p-n j u n c t i o n
component increases as indicated by the dashed line i n Figure 10.10. Once

the surface is inverted. x reaches its m a x i m u m value and hence there w i l l
be/'no^further increase' in this current component. However, while the
surface is depleted.."recombination-generation
centers at the oxide-silicon
interface provide vet another contribution to the total generation current.
This c o n t r i b u t i o n will result in a peak in the reverse current versus gate
voltage characteristics (Figure 10.10b).
This picture is given verification bv the measurements shown in Figure
10.11. Here we show both the reverse current and the gate-to-substrate
capacitance as a function o f gate voltage for various values o f reverse
voltage as a parameter. This figure shows that the reverse current increases
at the same gate voltage regardless of the value o f the reverse bias, corresponding to the depletion of the surface. The current decreases again,
although not to its original value, when the surface becomes inverted as
e
\
v
Surface
Effects
on
p-n
Recombination-Gene rati on
Junctions
Processes
301
300
Finally,
^vhich corresponds to the surface generation component.
In the above equations, A
denotes the area of the metallurgical
junction and A, the area of the substrate under the sate; U
and
U
are the carrier generation rates per unit volume in the reverse biased
depletion regions of the metallurgical and field-induced junctions, respectively. U. is the carrier generation rate per unit surface area at the oxidesilicon interface when this interface is completely depleted of both electrons
and holes, i.e., when the electron and hole concentrations at the surface
n and p are much smaller than the intrinsic carrier concentration
(a
condition that is met when the junction is reverse biased).
We saw in Chapter 6 that the generation rate per unit volume in a
reverse-biased depletion region is given by
i t J
nJ
FtJ
i m
u = --'
~>
o
'
(10.15)
with T , = \i<yv N for centers with energy level E =

Here ;V. is^he
concentration (per unit volume) of bulk recombination-generation centers,
and a is their capture cross section, assumed to be the same for electrons
and holes for simplicity.
The carrier generation rate of a completely depleted surface per unit
area is given by
U, \n,s
(10.16)
th
with 5, = tjv iN on the basis of Equation 5.50, for centers with energy
level E. = E. where N is the density (per unit area) of surface recombination-generation centers and a is their capture cross section, assumed
to be the same for electrons and holes.
Thus the three current components become
0 fS
t
si
verified by the observation that the point of decrease in current coincides

closely with the point of the capacitance rise.
Thus the generation current may consist of one or more of the following
three components, depending on the nature of the surface space-charge
region:
M3*M^T*frWs<!t*>l ^ < * K " * i
WW/
which corresponds to generation within the depletion region of the

metallurgical junction;
KJ J.-/ J
c
fen.F!J
^CeA
F U
i m i K
l%en,F[j
- H
itta*Jc.A
and
\^c' <
MHJ,
oFU
(10.17)
(10.18)
I I
\\r
(10.13)
which corresponds to generation within the depletion region of the fieldinduced junction. This equation applies only after the surface is inverted.
When the surface is depleted, but not inverted, 2
is replaced by x .
i m a x
The assumed condition of E. , is in good agreement with experimental measurements of the temperature dependences of the three current
components, all of which have the same temperature dependence as n. .
1
s. "I 4-11
/Ci
)Q.
fo
302
Surface Effects
o n p-n
Junctions
Recombination-Generation Processes
.JQJ
However, it is important to note that a continuum o f states distributed in

energy would lead to similar results since only those statesjjiaLare w i t h i n
approximately one -7" from the middie o f the energy.gap are efficient genera t i o n centers. For uniform distributions o f states of density D . ( c m e v )
and D ( c m e v ) in the bulk and at the surface, respectively, the effective
lifetime r, is
- 3
- 2
_ 1
_1
sl
T, =
'
OL (7rkTD )
th
(10.20)
and the surface recombination velocity s is

0
s, ov^vkTD*).
(10.21)
Such states would also lead to current components which have the same
temperature dependence as n,. Thus single-level states located at energy
, and a continuum o f states will have the same effect.
I t is evident from Equations 10.17. 10.18. and 10.19 that, whereas b o t h
bulk generation current components should depend on the magnitude
of the reverse bias through W and z
. the surface generation current
should be independent o f the reverse bias. The measurements shown i n
Figure 10.11 approximately bear out this conclusion.
Typical values of s on thermallv oxidized silicon surfaces are o f the order
of 1 to 10 cm/sec. Typical values o f T , are of the order o f 1 to 10 usee.
For junction and surface areas o f ~ ) 0 c m . these magnitudes correspond to reverse current variations o f the order of tens o f picoamps as
shown in Figure 10.11. A l t h o u g h these currents are small enough to be
almost inconsequential from a practical standpoint, their significance
becomes great under forward-bias conditions.
r t m a v
_ 3
: !
We have seen in Chapter 6 for the case o f bulk current components that
the recombination current in the forward-bias condition will be given
approximately by
(v)
Fig. 10.12 Base current as a function of gate voltage, with emitter-base forward bias
as parameter, for the transistor shown in the inset.' Surface concentration of base
region is ~ 2 . 5 x 10" cm" .
1
]rrU
:T
i r e c ^ he. e *' .
(10.22)
A similar relationship also holds approximately for the

field-induced
j u n c t i o n and the surface components. Thus the m a x i m u m recombination
current will be given approximately by
Iiei: =
iqn
the forward recombination current will also varv bv approximate)}' a

factor o f two. Since, in the case o f a transistor, all of the recombination
current w i l l appear as base current, this means that the common-emitter
current gain h
o f a transistor will vary in the same p r o p o r t i o n . This
effect is o f extreme importance in c o n t r o l l i n g the current gain o f transistors,
especially at relatively l o w collector-current levels.*
F
+ *.) A
V!
-W
(10.23)
This expression shows that the recombination current under f o r w a r d

bias will be affected by the field-induced junction and the surface c o m ponents in the same percentage manner as the generation current under
reverse bias. If, for example, the reverse current varies by a factor o f t w o .
Recall thai because recombination currents increase with the factor exD q)VA!3kT,
whereas the collector current increases with the factor exp o ' F' '/AT", the recombination
curreni components will become less and less imponani as the current level is increased.
(See Chapter 7.)
Surface
E f f e c t s o n p-n
Junctions
Field-Induced Junctions and Channel
Currents
305
304
its original value. These phenomena are in general agreement with the
above discussion. (For the moment we will ignore the variation o f base
current for large negative voltages.)
10.4
FIELD-INDUCED J U N C T I O N S A N D
C H A N N E L CURRENTS
Let us now focus our attention in further detail on the case when the
surface o f the substrate is inverted so that a field-induced j u n c t i o n exists
in parallel w i t h the metallurgical j u n c t i o n , as illustrated in Figure 10.14.
F i g . 10.14
Field-induced junction
F i g . 10.15
in
with
t h e b r e a k d o w n of t h e f i e l d - i n d u c e d
parallel
junction.
a metallurgical
I l l u s t r a t i o n o f che c o n s e q u e n c e o f
junction,
Increasing n u m b e r s c o r r e s p o n d t o i n c r e a s i n g l y
negative gate v o l t a g e s .
Experimental measurements on a transistor whose emitter-base j u n c t i o n is controlled by a field plate are shown in Figure 10.12 where we show
the base current as a function o f gate voltage, for various values o f emitterbase forward bias. T h e corresponding collector currents, which d i d not
vary with sate voltage, are also indicated. As the gate voltage vanes from
negative to positive values, the base current goes t h r o u g h a peak and then
it decreases again a l t h o u g h not to its original value. The corresponding
variation in the surface space-charge region associated w i t h the emitterbase j u n c t i o n is indicated i n the insets. The current gain h
= / / / is
shown as a function o f gate voltage i n Figure 10.13. Where the base current had a peak, h
has a d i p . After that it recovers p a r t i a l l y , but not to
F
This field-induced junction will have a breakdown voltage o f its o w n . I n

particular, in many cases the field-induced j u n c t i o n w i l l break d o w n at
some voltage which is lower than the breakdown voltage o f the metallurgical j u n c t i o n . The reverse current-voltage characteristics in such a case
are illustrated in Figure 10.15. When the reverse voltage exceeds the
breakdown voltage of the field-induced j u n c t i o n BV ,
current begins
to flow.
This current flows along the inversion layer to the diffused region, and
it w i l l saturate upon further increase in reverse voltage.f This type o f
current-voltage characteristic is often referred to as a channel characteristic. The level at which the reverse current saturates depends on the conductance o f the inversion layer. The larger this conductance, the higher
the saturation current level, as shown in Figure 10.15.
FIJ
* Such a phenomenon of current saturation is generally associated with current flow

through conducting channels. It is discussed in Chapters 3 and 11.
306
initial reverse current-voltage characteristics o f a p'n gate-controlled

diode for V = 1 0 0 volts. W i t h the exception o f a slight variation in
the breakdown voltage, there is very little effect o f gate voltage on the
" characteristics. A defect was then introduced under the gate by an
electrostatic discharge across the gate. The resulting characteristics for
V * 1 0 0 volts are shown by the solid lines. W h e n V = + 1 0 0 volts,
no field-induced j u n c t i o n exists and, therefore, the defect has no influence
on the j u n c t i o n characteristic. W h e n V = 100 volts, however, a fieldinduced j u n c t i o n does exist and a huge excess current flows due to the
-presence o f the defect in the field-induced j u n c t i o n .
1 ma
100 v
A conceptually very similar case involving a l o w value o f the b r e a k d o w n

voltage o f the field-induced j u n c t i o n is encountered i f the field-induced
junction is formed over the heavily doped r e g i o n o f the junction rather
than over the substrate, as illustrated i n Figure 10.17. Because the charges
present i n the silicon dioxide layer generally tend to be positive (see Chapter 12). this phenomenon is more likely to be encountered in practice over
p~ regions than over rr regions.
100 pa
10 pa
1 pa L
0.01
;
Inmal
Atiei imroducnon
of defect I
!
0.1
1
10 100 1000
Because the
field-induced
j u n c t i o n is formed over material o f high
400
Fig 10.16 The effect of the introduction of a low-breakdown producing defect on the
reverse /-V characteristics of a gate-controlled p-n diode, such as shown in Fig. I 0 . K
I f there is a defect present w i t h i n the field-induced j u n c t i o n shown i n

Fieure 10.14 which lowers the breakdown voltage of the field-induced
j u n c t i o n . ! a large excess reverse current will be observed even at relatively
l o w reverse biases. This effect is demonstrated in Figure 10.16 where we
show the reverse current-voltage characteristics on log-log scales in order
to illustrate the typical magnitudes involved. The-dashed lines show the
300
t Such a defect may be analogous to those which lead to the "soft" reverse current-
200
voltaee characteristics discussed in Chapter 6.

T'
c
100
Field-induced
junction
T3?^y\V^ Metallurgical
Metallur
junction
S^^S^S^
Fig. I0.IB The consequence of the breakdown of the field-induced junction shown in
E - I0.I7. Increasing numbers correspond to increasingly positive gate voltages.'
ri
Fig. 10.17
Field-induced junction formed over the heavily doped region.
308
Field-Induced Junctions and Channel C u r r e n t s
309
doping concentration, BV
will be l o w and the channel current w i l l
commence at a small value o f the reverse voltage. A n experimentally
observed" set o f such reverse current-voltage characteristics is shown in
Figure 10.18 for increasingly positive gate voltages. Note that channel
current begins to flow when the reverse voltage exceeds BV ,
which in
this case is approximately I volt.
The breakdown characteristics o f field-induced junctions, i.e., the characteristics o f channel currents well below saturation, are strongly affected
by the surface concentration o f the
region. A t low surface concentrations the breakdown mechanism is avalanche breakdown; at high surface
concentration it is Zener or tunneling breakdown. Zener breakdown has
the interesting property o f being approximately symmetrical near the zerobias point, i.e., i t leads to a large excess current also flowing in the forward
direction. The f o r w a r d and reverse current-voltage characteristics o f a
FlJ
FIJ
field-induced j u n c t i o n formed over a p region o f high surface concentration are shown i n Figure 10.19. These characteristics are evidently quite
symmetrical.
Such large excess forward currents are exceedingly important in the case
of the emitter-base junction o f transistors. Because such excess currents
do not contribute to transistor action, they will appear as base current.
This is illustrated in Figure 10.20 where we show the base current as a
function o f gate voltage o f a transistor similar to that used in Fig. 10.12
w i t h the exception that the surface concentration o f the base region is
higher than i n the previous case. It is evident that i n addition to the peak
4
I
100/iap
30
60
40
ChanneI
current.
Reverse
\ 0 /
77 0
Y
I
0.2
V
0.4
1
I
0.6
0.3
Forward
1 v
1v
Fig. 10.19 Forward and reverse channel current-voltage characteristics of a p*n junction
with high boron surface concentration ( C , = I x 10" cm"'). Also shown are the onginal
diode characteristics.
6
Fig. 10.20 3ase current as a function of gate voltage, with emitter-base forward bias
as parameter, for the cransistor shown in che inset." Surface concentration of che base
region is 4 x I 0 cm" .
1
1S
310
due to surface recombination, the base current drastically increases for

both sufficiently large negative and positive gate voltages. As indicated i n
the insets, these conditions correspond to the formation and Zener breakdown of field-induced junctions over the emitter and the base regions,
respectively.
The drastic effect o f such breakdown currents on the current gain of the
transistor is shown i n Figure 10.21. Note that this phenomenon can
influence the current gain to much higher collector-current levels than
surface recombination.
Junction Breakdown Voltage
10.5
31 I
SURFACE EFFECTS O N J U N C T I O N
BREAKDOWN VOLTAGE
In the preceding discussions we saw that the shape o f the depletion region
near the surface can be varied between the t w o extremes o f f o r m i n g a fieldinduced j u n c t i o n on the heavily doped side or o n the substrate. A c c o r d ingly, the field distribution near the surface w i l l also vary. U n d e r many
conditions the electric field near the surface w i l l be higher than i n the b u l k
and, therefore, breakdown w i l l occur near the surface at a lower reverse
bias than that corresponding to the breakdown voltage in the b u l k .
The type o f variation expected is illustrated i n Figure 10.22 for an
n~p diode. For a verv large negative gate voltage, a field-induced j u n c t i o n
will be formed over the n~ region (Figure 10.22a). Near the
"comer"
region the depletion region will be relatively n a r r o w and, therefore, the
electric field intensity there w i l l be high. ( I n this figure the arrows designate
only the direction o f the electric field.) Thus j u n c t i o n breakdown w i l l occur
in tVis region at a relatively low value o f the reverse bias. I n the other
extreme, shown i n Figure 10.22c, the gate voltage is the same as the junction-voltage. A surface depletion region o f a p p r o x i m a t e l y the same w i d t h
as the j u n c t i o n depletion region w i l l be formed a n d , therefore, the electric
field intensity near the corner w i l l be reduced below even that obtained in
an undisturbed planarp-n junction (i.e.. with n o surface fields). We would
then expect the breakdown voltage to be high. I n an intermediate case, for
instance, when the gate is kept at the same p o t e n t i a l as the substrate, the
field near the corner region w i l l be somewhat increased over the undisturbed case as shown in Figure 10.22b.
Experimental observations o f the breakdown voltage o f n~p diodes as a
function of the gate voltage are shown i n Figure 10.23 for three different
substrate i m p u r i t y concentrations. I n each case the. breakdown voltage
starts out from a relatively l o w value at negative gate voltages, increases as
the gate voltage is made more positive and approaches the plane rather than
planar value of the breakdown voltage i n each case. The gate voltage that
must be applied in order to approach the plane b r e a k d o w n voltage values
is approximately that required to bring about the f o r m a t i o n o f a fieldinduced j u n c t i o n over the substrate, i.e., the t u r n - o n voltage. Calculated
turn-on voltage values are designated by the cross-hatched line segments
intersecting the data points. ( I t should be kept i n m i n d that the turn-on
voltage is a function of the reverse voltage. T h i s was taken i n t o account
in the calculation of these line segments.)
Fig.
10.20.
312

I I m - UK) v
I
1
^ ^ ^ - " C o r n e r " region
"
\\ \\
V = 0
(a)
-W,
'Corner" region
= 0
Fig. 10.22 Illustration of the effect of gate voltage on the shape of the depletion region
and the breakdown voltage. V ^, Vm and V ^ correspond to the breakdown voltages
under the indicated gate-voltages.
a
I n the intermediate range, the breakdown voltage variation follows

BV = mVr; + constant
(10.24)
where m e 1.
To explain this type o f variation, we consider the potential distribution
w i t h i n the depletion region o f an idealized gate-controlled diode calculated
from numerical solution of Poisson's and Laplace's equations in the
depletion region and i n the oxide, respectively, shown i n Figure 10.24. I n
all four cases the electric field intensity in the corner is evidently larger than
elsewhere i n the depletion region. I t might be expected that, i f x W,
this electric field w i l l be more closely related to the field across the oxide
{V
V )jx,
than to the field across the depletion region. Thus the
electric field at the corner is given bv
0
(10.25)
lere -x is some geometric correction factor. I t then follows that the condition of breakdown
t.
w i l l lead to a relationship o f the
314
Problems
315
same f o r m as Equation 10.24. i.e..

BV=
c +
Ii!i'.
a
This relationship is in reasonable agreement

observations
as well as with
the corresponding
with
the
o.26)
experimental
estimates based on the
numerical calculations.
The electric field near the corner is related to the electric field across the
oxide only so l o n g as the oxide thickness is m u c h smaller than the depletion
region w i d t h .
As the
substrate i m p u r i t y
concentration
or
the
oxide
thickness is increased, there w i l l be an increasing d e v i a t i o n from this cond i t i o n . T h i s is borne out by the experimental data s h o w n i n Figure 10.23:
as the i m p u r i t y concentration is increased, the effect o f gate voltage o n the
breakdown voltage becomes increasingly smaller.
'
I
REFERENCES
CITED
1. A . S. Grove and D . J . Fitzgerald. "Surface Effects on P-Af JunctionsCharacteristics

of Surface Space-Charge Regions under Non-Equilibrium Conditions," Solid-State
Electron., 9, 783 (1966)"
2. Thefieldplate or gate-controlled p-n junction was employed in thestudv of germanium
surface properties by J . H. Forsier and H . S. Veloric. "Effeci of Variations in Surface
Potential on Junction Characteristics.'" J. Appl. Phvs.. 30, 906 (1959). This structure
was used to study surface effects on silicon p-n junctions bv C . T . Sah, 'Effect of
Surface Recombination and Cnannel on P-N Junction and Transistor Characteristics," IRE Trans. Electron Devices, E D - 9 , 94 (1962).
*"
|._.
3. N . McDonald and T . Everhart, unpublished.
1 ~
4. V . G . K . Reddi and C . A . Bittmann. "Second Quarterly Report on Micropower

Functional Electronic Blocks," Contract A F 33 (615)-30] 6 (1966).
5. D . J . Fitzgerald and A . S. Grove. "Mechanisms of Channel Current Formation in

Silicon PS' Junctions."' Physics of Failure in Electronics, Volume 4, p. 315, Rome
Air Development Center (1966).
6. A . S. Grove and D . J . Fitzgerald. "The Origin of Channel Currents Associated with
P~ Regions in Silicon," IEEE Trans. Electron Devices. E D - 1 2 , 619 (1965).
7. A . S. Grove, O. Leistiko. and W. \V. Hooper. "Effect of Surface Fields on the Breakdown Voltage of Planar Silicon PS' Junctions." IEEE
Trans. Electron Devices
ED-14 .157 (March. 1967).
PROBLEMS
10.1
Calculate and plot the turn-on voltage V as a function of reverse bias Vg for
two gate-controlied diodes having substrate impurity concentrations of 10" and
10" cm" , respectively, and an oxide thickness of 1 u. Take V = 0.
T
rt
Surface
316
10.2
E f f e c t s o n p-n J u n c t i o n s
-3
A gate-controlled p'n diode has a metallurgical junction area of 1 0 cm-. The

gate overlaps the ^-region over an area of 10~ cm-. The substrate impuntv
concentration is 10 c m " , the junction depth is 5 ft, the oxide thickness is 0.2 ft,
the lifetime T = I usee, the surface recombination velocity 5 , = 5 cm/sec. The
flat-band voltage is V- = 2 v. Calculate:
fa) The gate voltages at which the surface of the substrate is intrinsic, and is
strongly inverted, respectively, for V = 0.
(b) The room-temperature reverse current at K , = - I v. for Vr, = 20. 0, and
PRINCIPLES
CHARACTERISTICS
18
OPERATION
MODIFICATIONS
SIMPLE
THEORY
OTHER
TYPES
FIELD-EFFECT
OF
OF
OF
THE
SURFACE
TRANSISTORS
4-20 v.
(c) The forward current at V = 0.4 v, at the same values of the gate voltage.
Sketch the reverse and forward currents as a function of gate voltage.
(d) The breakdown voltages at the above values of the gate voltage, and also in
r
the absence of a gate.

10.3
For the diode of the previous problem, calculate the junction capacitance Cj at
V , = 0, and at
= 2 v for K , = 20, 0, and 4-20 v. Sketch Cj versus gate
voltage. Also calculate the capacitance between the gate and thep~ region, for
Vj = 0, for the above values of the gate voltage.
Surface Field-Effect
Transistors
10.4
Derive an expression giving the maximum variation of the current gain h with
gate voltage, for a transistor with a gace over the emitter-base junction, as a
function ofcollector current. (Neglect excess currents due to the Zener mechanism.)
10.5
A defect under the gate of a gate-controlled diode lowers the breakdown voltage
of the field-induced junction to 2 v. What gate voltage must be applied in order
to bring about an excess current due to this defect? The substrate impurity
concentration is 10" c m ' , the oxide thickness is I it. Take V
= 0.
rE
eB
10.6
Calculate and plot the surface field at the onset of strong inversion as a function
of reverse bias for a gate-controlled diode which has a substrate impurity concentration of 10 c m . Assuming the field-induced junction formed upon
inversion is plane, estimate its breakdown voltage. Compare this breakdown
voltage with that of a one-sided step junction formed within the same substrate.
18
10.7
- 3
13
A planar p*n diode (no gate) contains 2 x 10 positive charges/cm- within its
oxide. The substrate impurity concentration is 10 cm" , the oxide thickness is
1 ,a. Calculate its approximate breakdown voltage if the junction depth is (a) 5 ft
and (b) 50 ft.
14
The last semiconductor device that we consider here was perhaps the
first one to be conceived. The principle o f the surface field-effect transistor
dates back to the early 1930's when L i l l i e n f e l d i n the United States a n d
HeiF in England proposed to use the surface field-effect to achieve a solidstate amplifier. It was subsequently actively investigated by the BeLT"
Laboratories group in the late 1940's. The more or less accidental discovery o f the bipolar transistor then gave new direction to solid-state
device research and development for more t h a n a decade.
The advent o f thermally oxidized silicon b r o u g h t with it an increase in
the feasibility o f fabricating the surface field-effect transistor. I n 1960,
K a h n g and A t a l l a used a thermally oxidized silicon structure i n a surface
field-effect transistor. The ensuing years b r o u g h t about an exceedingly
intense activity in this field. This activity, o n the one hand, led to a high
level o f knowledge and understanding o f the thermally oxidized silicon
surface (see Chapter 12). On the other hand, this activity is responsible
for the fact that the MOS transistor, i.e., the surface field-effect transistor
using a thermally grown silicon dioxide layer, has become potentially the
second most important device next to the b i p o l a r transistor. I n fact, in
many integrated circuit applications, the M O S transistor may eventually
become the more important one.
1
A l t h o u g h various kinds of surface field-effect transistors can be and

have been fabricated, the discussion in this chapter deals w i t h the
317
318
n-channel device which we have already considered i n Chapter 9 i n connect i o n with studies o f semiconductor surfaces This device is illustrated i n
Fieure 11.1. I t consists o f a />-type silicon substrate i n t o which t w o n*
regions, the source and the drain, are diffused. The region between the
source and the drain is under the influence o f a metal field plate or gate.
I f a large positive voltage is applied to the gate, the surface o f the underlying /"-type silicon can be inverted and a conductive n-type clutnnel can be
induced connecting source and drain. The conductivity o f this channel
can then be modulated by varying the gate voltage.
Source
Principles of O p e r a t i o n
319
where Q is the charge density per unit surface area o f electrons i n the
inversion layer. The magnitude o f 0 w i l l depend on the silicon surface
field: hence, i t will depend on the potential difference between the gate
and the inversion layer.
n
As the d r a i n voltage is increased, the average potential difference f r o m

gate to the n-type inversion layer w i l l decrease. A s a result, Q w i l l also
decrease and the resistance o f the channel increase. Thus the d r a i n current
versus d r a i n voltage characteristic w i l l begin to bend d o w n w a r d f r o m the
i n i t i a l resistor line. This is evident i n the experimental current-voltage
Ti
Drain
U>)
Substrate or body
(b)
(cj
Fig. 11.2 Illustration of the operation of a surface field-effect transistor for V > V .
(a) V^, is small; channel resistance is constant.
(b) V = V , ; onset of saturation.
c
Fig. I I.I
n-Channel surface field-effect transistor.
We
first
consider the principles of the operation o f such a
field-effect transistor.
surface
field-r.fleet
surface
a s l
(c) V > V^sa,; no further increase in drain current.

p
Then we discuss some i m p o r t a n t characteristics o f
transistors:
the current-voltage
characteristics,
the
transconductance, and the gate leakage current. Finally, we consider the

modifications o f the simple theory.
Insofar as possible, we attempt to
m a i n t a i n a parallelism between the treatment o f the surface
field-effect
transistor and its sister device, the j u n c t i o n field-effect transistor discussed

i n Chapter 8.
11.1
As the d r a i n voltage V is increased still further, the voltage d r o p across

the oxide near the drain is further reduced u n t i l eventually it falls below the
level required to maintain an inversion layer. The drain voltage at w h i c h
this happens w i l l be denoted by the symbol V .
At this drain voltage,
the channel near the drain disappears. The surface w i l l be merely depleted
and no longer inverted here, as illustrated i n Figure 11.2b.
D
PRINCIPLES
OF
OPERATION
DSM
Let us consider the situation when a large enough gate voltage is applied
to induce an -type inversion layer between the source and the drain regions,
as shown in Figure 11.2. The cases o f small and large drain voltages are
considered separately.
characteristics o f an ^-channel M O S transistor, which we w i l l use for

illustration throughout this chapter, shown for various values o f the gate
voltage i n Figure 11.3. The characteristics indeed start out in a straightline fashion and begin to bend as the d r a i n voltage is increased.
F o r small drain voltages, the channel
induced
between source and drain essentially behaves like a resistor. Its resistance,
The potential at the end o f the inversion layer, at the p o i n t X in Figure

11.2b, w i l l be.that value for which the gate voltage V can still maintain
an inversion layer. By our above definition, this value is V
Once the
drain voltage exceeds V
the potential at the end of the inversion laver.
at the point X, will remain constant, independent of any further increase in
the drain voltage, although the p o i n t X w i l l move somewhat t o w a r d the
source as illustrated in Figure 11.2c.
G
Ds3lV
DSIV
as shown in Chapter 9, is
L
Surface
320
Field-Effect
Transistors
Characteristics of Surface
Field-Effect
Transistors
321
JUNCTION
The current now is due to the carriers that flow d o w n the inversion layer
and are injected into the depletion region near the drain. The magnitude
o f this current w i l l n o t change significantly with increasing d r a i n voltage
since it depends o n the p o t e n t i a l d r o p f r o m the beginning o f the inversion
layer to the end o f the inversion layer ( p o i n t X) and this potential d r o p
remains unaltered.t Thus, for drain voltages larger than VQ
the current
will not change substantially and will remain at the value fr> , as is evident
f r o m the experimental data s h o w n i n Figure 11.3.
FET
SURFACE
FET
The gate terminal is electrically isolated from channel by:

reverse-biased depletion region.
insulator.
The magnitude of the conducting charge is modulated bv:
%3A
the width of a reverse biased

depletion region.
sax
the incident surface field,
For small drain voltages, the channel is essentially ohmic. Increasing drain
voltage reduces the average magnitude of the conducting charge, thereby
reducing the channel conductance.
When drain voltage exceeds a certain value, the potential drop from source
to end of the channel remains at the fixed value P ^ , . Hence, the current
flow also remains at a fixed value Iryî for drain voltages V > Vo^ .'
Saturation region
a n
ÊHM d - therefore,
depend on the gate voltage applied.
Thus, the two devices are similar i n a l l respects except i n the physical
mechanism responsible for varying the magnitude o f the conducting charge.
Fig. 11.3 Current-voltage characteristics of a silicon n-channel MOS transistor. This

device is used for illustration throughout this chapter. Its structural parameters are:
Z/L = 25, x, = 0.08 (i, N , = 3 X . I O cm" , V = - 2 . 0 v.
1 8
FB
I f the gate voltage is increased, the conductance for small values o f d r a i n

voltage w i l l be larger and the d r a i n voltage at which the current saturates
V ,t i l l I
be larger. A s a result, the saturated current w i l l also have a
larger magnitude, as is evident i n Figure 11.3.
w
Dsa
1.2
a-
CHARACTERISTICS O F SURFACE
C u r r e n t - V o l t a g e Relationship*
idera
re
m
lgUie
ec
t
o
n
!
t
section ot the channel is given by
^;!;T
tT T
U r f a C e
fidd effeCt
dV =I dR
D
i t a
d r 0
e r a t i n
the linear
elemental
I p dy
(11.2)
ZPnQ'j)
Thus we can distinguish t w o regions o f operation o f the surface fieldeffect transistor. A t l o w d r a i n voltages, the current-voltage characteristics
are nearly ohmic or linear (linear region), while at high d r a i n voltages the
current saturates w i t h increasing d r a i n voltage (saturation region).
These
two regions, the shape o f the characteristics, and the manner i n w h i c h
saturation sets i n are a l l reminiscent o f j u n c t i o n field-effect transistors.
I t is instructive to compare the most important operating principles o f
the j u n c t i o n and surface field-effect transistors.
t This assumes that the movement of the point -X toward the source is negligible. The
validity of this assumption will be discussed in a later section.
Fig. 11.4 The elemental section of the channel emoloved in rh J i

current-voltage characteristics of surface H e l d - e t T e L " S o r s
'
V a t ,
**
322
in analogy to Equation 11.1-except that L is replaced by dy. (Note that

this equation is identical to the corresponding equation for the j u n c t i o n
field-effect transistor except for Q(y) replacing qN [d 2H\y)].
However, both quantities represent the charge density o f electrons per
unit surface area.)
A t a distance y from the source, the total charge induced i n the silicon
O w i l l partlv consist o f charge in the inversion l a y e r 0 * and partly o f
charge i n the surface depletion region due to the ionized acceptor ions,
2 .*Thus,
- S . ' J A ^ W .
D
C,(yj
C.(y)
+ Qabtft
(11-3)
C h a r a c t e r i s t i c s of Surface Field-Effect Transistors
323
inversion laver and the substrate becomes increasingly reverse biased as

we proceed f r o m source to d r a i n . Hence, both the energy band bending
<p, and the charge w i t h i n the surface depletion region 0
increase f r o m
source to d r a i n .
B
C o m b i n i n g Equations 11.2. 11.5, 11.6. and 11.7 and integrating between

the source, where y = Ot and V = 0, and the d r a i n , where y = L and
V = Fjr,, yields
1.-1**.
ll
1(7 ~ V-FB ~ 2>
Fp
F r o m Equation 9.16 we have
2 \'2K,e qN
,. i
(11.8)
. .
y/
= - & + 4,
FS
(ii.4)
which can be rewritten in the form
which relates the voltage applied to the field plate o f an M O S structure to

the charge induced i n the silicon Q and to the surface potential <p -wnich
denotes the total bending o f the energy -bands. We also include here the
flat band voltage V
which is due partly to the presence o f charges i n the
insulator or at the interface, and partly to a finite metal-semiconductor
w o r k function difference as discussed in Chapter 9. C = K i lx
is the
capacitance of the oxide layer per unit area.
Combining Equations 11.3 and 11.4 yields the charge density i n the
inversion layer.
s
III
FJ>
FB
QM
= - [v
F B
Uy)}C
c B
Q (y).
K,x
3 A.pS^niax.o
(11.9)
where
/^CoQVJ.
qN
t'rfmax.i/ A /
( l 1.5)
[(4fr-R2^)*-(2^,)N
(n.6)
is the w i d t h o f the depletion region of afield-induced j u n c t i o n in equilibrium.

The current-voltage characteristics of the ;?-channel M O S transistor are
shown in Figure 11:5 for various gate voltages applied. The corresponding
calculated characteristics are also shown in this figure, based or. E q u a t i o n
11.8. T h i s equation is valid only below saturation. Thus the curves shown
were calculated for 0 < V < J'josat- Beyond V
the current was taken
to be constant, as discussed i n Section 11.1.
where <j> is the F e r m i potential o f the substrate, and V(y) is the reverse
bias between the elemental section o f the channel and the substrate. The
charge within the surface depletion region 0 is given by
It is interesting to consider the two l i m i t i n g forms o f E q u a t i o n 11.8.

For very small drain voltages, i.e., V 2 i
an expansion - o f the
bracketed terms leads to the formulas,
Since we have assumed that a conducting inversion layer exists, the

surface potential <p w i l l be given approximately by the condition o f strong
inversion i n the presence o f an applied voltage. F r o m Equation 10.1, this
is
t
-, '
Uv)
Fr
0 (y)
B
= -qh'jXcnJo)
= -^'2K,e qK [V{y)
0
+ 2$ ),
Fv
DsKl
F j r
(11.7)
from Equations 10.7 and 10.4.

Since the voltage V(y) increases from the source toward the drain due to
the IR drop along the channel, the field-induced j u n c t i o n between the /i-type
t In the derivation of the current-voltage characteristics of surface field-effect transistors
some authors either neglect Q (y) in Equation 11.3 or treat it as a constant. This
approximation can lead to significant error in practical devices.
s
^FB
26 Ft
0&.
[ L i n e a r region]
(11.10)
t We define the potential at the source as our ground potential. In this treatment we
only cons,der the case in which no bias is applied to the substrate relative to the source
T M ,? , r
P
summarized at the end of this chapter in
Table 11.] for botn n- and ^-channel devices.
n
t h e r
a r e
Surface
Field-Effect
Transistors
324
Characteristics of Surface
Field-Effect
Transistors
32S
2 x 10 " r
1.5
Theory.
/t = 450 c n r / v sec
Vwm = - 2 V
1.0
n 1 i l
V =6v
a
0.5
-
5v
(b)
Experiment
4v
3v
l/s
u/^
J
; v
31 4
L
5
k
7
:6
Vc-M
Fig. 11.6 Channel conductance of the MOS transistor as a function of gate voltage, in
the linear region (V small).
D
Fig. I I.S
Current-voltage characteristics of the n-channel MOS transistor.
(a) Experiment (same as Fig. 11-3).

(b) Theory (Equation 11.3).
where 0 . = - ^ K ^ N ^ M
* the charge density
w i t h i n the surface depletton region, in equilibrium. Thus a n ohmic charac
B
teristic results. The channel conductance
The channel conductance of the M O S transistor i n the linear region is

shown i n Figure 11.6 as a function o f gate voltage. The reduction i n slope
at high gate voltages is due to a decrease i n the m o b i l i t y o f electrons o w i n g
to increased surface scattering (see Chapter 12).
The above derivation loses its validity when the inversion layer disappears near the drain, i.e., when V = V . The c o n d i t i o n Q (L) =_ 0,
inserted into E q u a t i o n 11.5, yields
D
dip
(11.11)
^ -K,e qNJV
Ds&t
VJML
in the linear region is
g = f,
U n
[Linear region]
C (F - V)
0
+ 2 ]
rv
Dslit
+ 2<p
Fv
- V
+ V
(11.14)
(11.12)
Li
where we have also used Equations 11.6 and 11.7, evaluated at V = ^ i f

Solving E q u a t i o n 11.14 for V
leads to
s a
where the turn-on voltage V , i.e., the voltage that must be applied to the
B s i t
gate i n order to induce a conducting channel, is given by

FB
IT
FB
F P
(11.13)
c
'
^Dsat
Fv
K N
fM
1+
2c;-{v
V )
FB
VQ
(11-15)
326
KDS;
G -
F B
Modification of the Simple T h e o r y
327
2i FT
FB
FT
(11.16)
N o t e that when the oxide thickness x is small in comparison t o the w i d t h

of the surface depletion region x ,
this expression reduces to the simple
form,
c
d!ntLX
K M -
V*m at y G -
hr,
io W . J -
P!.lf)
Substitution o f
i n t o the current-voltage relationship, E q u a t i o n 11.8
or 11.9, gives the magnitude o f the saturation current, V x , .
sat
b.
Transconductance
As for j u n c t i o n field-effect transistors, the transconductance is defined

by
(11.18)
8
dV r ^ - c o n s t
c
Differentiating E q u a t i o n 11.9 leads to
ta
J (v)
C
Fig. 11.7
im for V
<
V S I T
r P.C.VD
(11.19)
c.
The transconductance in- the saturation range can be calculated by

setting V = V .
This leads t o
D
DS&L
Transconductance of the MOS transistor as a function of gate voltage, in the
saturation region (V^ > V
M a l
).
Gate Leakage Current
I n the surface field-effect transistor, the gate is insulated from the channel region by a silicon dioxide layer.
Because a t h e r m a l l y grown silicon
dioxide layer is an excellent insulator, the leakage current between gate
and channel region is extremely small, less than 1 0

gmsat = 7 r'nCoj f'c I V . B
2<f>
- 1 1
a m p i n a typical case.
Thus the surface field-effect transistor features an exceedingly high i n p u t
Fv
resistance, several orders o f magnitude higher even t h a n j u n c t i o n
field-
effect transistors.
[Saturation region].
(11.20)
11.3
The transconductance o f the M O S transistor in the saturation region is
shown as a function o f gate voltage in Figure 11.7, in comparison t o calculations based on E q u a t i o n 11.20. A s in the case o f the conductance i n
the linear region, a decrease is observed at large gate voltages due to a
reduction i n surface m o b i l i t y .
a.
MODIFICATION
O F T H E SIMPLE
THEORY
Effect of Fast Surface States
We have already seen i n Chapter 9 that metal-semiconductor
work
function difference or charges w i t h i n the oxide or at the interface i n t r o duce a displacement of M O S transistor characteristics along the voltage
329
Surface Field-Effect Transistors
328
axis by an a m o u n t V .
I f . i n a d d i t i o n , there are fast surface states present, the charge i n these surface states w i l l also result in an additional
displacement along the voltage axis. T h i s c o n t r i b u t i o n to the displacement, however, varies w i t h the surface potential (and hence w i t h gate
voltage), because the p r o b a b i l i t y o f the occupancy o f a given surface state
itself varies w i t h the surface potential. Thus the displacement o f the
characteristics along the voltage axis due to fast surface states w i l l not be
parallel.
M o r e i m p o r t a n t , however, is the f o l l o w i n g : When the carrier concent r a t i o n near the surface is changed by changing the gate voltage, some o f
the a d d i t i o n a l carriers induced near the surface w i l l enter into surface
states and therefore w i l l n o t c o n t r i b u t e to an increase i n channel conductance. As a result, the observed transconductance will be smaller than the
theoretical transconductance. This w i l l be so at frequencies l o w enough
that the surface states can be charged and discharged i n phase w i t h the
measurement signal. A t higher frequencies, when the surface states cannot
respond rapidly enough, the observed transconductance w i l l approach the
transconductance w i t h o u t surface states.
FB
For this effect to become i m p o r t a n t , the charge v a r i a t i o n i n surface

states must be comparable to the magnitude o f the conducting charge i n
the channel which is generally o f the order o f 1 0 c m . Because i n
thermally oxidized silicon surfaces the surface state density can be reduced
to 10 to I 0 c m ~ (see Chapter 12), fast surface states do n o t pose a
significant problem i n device operation. This has been a very i m p o r t a n t
factor i n making the construction and operation o f M O S transistors
feasible.
u
1 0
- 2
for V < V af
Note that a high mobility is desirable for a large cut-off
frequency as in the case o f junction field-effect transistors.
D
c.
Ds
S o u r c e - t o - D r a i n Resistance in S a t u r a t i o n
We have seen i n the previous section that after saturation the potential
at the end o f the inversion layerat the p o i n t Xwill
be fixed at the value
Vp-sat- As the drain voltage is increased further, the reverse bias across the
drain j u n c t i o n w i l l increase. As a result, the depletion region separating
point X from the d r a i n w i l l widen and the p o i n t X w i l l move t o w a r d the
source, as indicated i n Figure 11.2c. Thus the effective channel length w i l l
become shorter and, as a result, the drain current w i l l increase slightly
with increasing drain voltage, resulting i n an u p w a r d tilting o f the currentvoltage characteristics beyond saturation. This phenomenon is particularly
prominent for devices w i t h small channel length L .
d.
Effect of Series Resistance
The same considerations that were applied in the case o f j u n c t i o n

field-effect transistors lead to the conclusion that the effect o f unmodulated
series resistances near the source and the drain w i l l be to lower the observed
channel conductance and transconductance as compared to their values
w i t h o u t series resistances according to the formulas
o
2
g(obs) =
1 -f- (R, +
surface field-effect transistor w i l l be given by

/ . - J = t
,
1 +
j u n c t i o n field-effect transistor, the m a x i m u m frequency o f operation o f a
(H-21)
where C is the total gate capacitance.

U s i n g E q u a t i o n 11.19 a n d C = Q Z L leads to
(11-23)
[Saturation region]
(11-24)
and
C u t - O f f F r e q u e n c y for T r a n s c o n d u c t a n c e
By exactly the same argument as we have used i n Chapter 8 for the
[Linear region]
RJg
gmsat(obs) =
b.
R,g
maAt
where g and ^
denote the channel conductance and transconductance
in the absence o f series resistances.
m s a t
Because in a surface field-effect transistor o f the type illustrated i n

Figure 11.1 all o f the channel region is under the influence o f the gate, the
series-resistance effect is considerably less i m p o r t a n t than in j u n c t i o n
field-effect transistors.
11.4
'finis
Jo
t More rigorous considerations yield gJ2^C
for this frequency limitation.
O T H E R TYPES O F SURFACE
(11.22)
The preceding discussion so far was for an n-channel device o f the type
where we have to apply a positive voltage to the gate i n order to bring
330
j ; the onset o f channel conduction. " B y complete analogy, the discussion could have been applied to a/>-channel device where a negative gate
voltage would have been required to bring" about channel conduction.
Because with zero gate voltage applied neither o f these devices conducts,
such devices are called normally "off" M O S transistors.
By contrast, a normally "on" device can be made by suitable control o f
the flat-band voltage. Thus, i f V
is sufficiently negative i n the case o f an
n-channel device or positive i n the case o f a />-channel device, a conducting
channel exists even w i t h zero gate voltage applied. I n such a case, the
channel conductance can be b o t h increased and decreased w i t h suitable
variation o f the gate voltage.
FB
The n o r m a l l y " o f f * n- and jP-channel devices as well as a n o r m a l l y " o n "

n-channel device are illustrated i n F i g . 11.8. I n all three devices the cond u c t i n g channel is induced by surface fields, whether the surface fields are
due t o a gate voltage or to charges w i t h i n the oxide. Thus we may call
these devices field-induced clmnnel devices. I t is also possible to b u i l d - i n a
channel metallurgical!)'. T w o examples o f such metallurgical
channel
devices are shown i n Figure 11.9. The first o f these, illustrated i n Figure
11.9a, is made by epitaxially g r o w i n g an ??-type film on a /"-type substrate.
The conductivity between the source and drain regions is then modulated
6
W
.
'
(b)
Fig. 11.9 Surface field-effect transistors with metallurgical channels.
(a) Epitaxially grown n-channel device.
(b) Sllicon-on-sapphire n-channel device.
by the gate voltage. A positive gate voltage attracts more electrons to the
channel, while a negative gate voltage depletes the surface and thereby
modulates the cross-sectional area available f o r current flow. I n this
mode of operation, this surface field-effect transistor works very m u c h like
a junction field effect transistor. A similar device made by g r o w i n g n-type
silicon on an insulating sapphire substrate is illustrated i n Figure 11.9b.
7
READING
REFERENCES
MOS transistors are reviewed in Field Effect Transistors, Physics, Technology and
Applications, J . T . Wallmark and H . Johnson, E d s . , Prentice-Hall, 1966.
REFERENCES
CITED
1. J . E . Lilienfeld, U . S . Patent 1,745,175 (1930).

2. 0 . Heil, British Patent 439,457 (1935).
3. W. Shockley and G . L . Pearson, "Modulation of Conductance of Thin Films of
Semiconductors by Surface Charges," Phys. Pet.. 74, 232 (1948).
4. D . Kahng and M . M . Atalla. "Silicon-Silicon Dioxide Field Induced Surface Devices,"
I R E Solid-State Device Research Conference, Carnegie Inst, of Tech., Pittsburgh,
1960.
(c)
Fig. 11.8
Surface field-effect transistors with field-induced channels.
(a) Ncrrr.sliy "off" n-channel device.

(b) Normally "off" p-channel device.
(c) Normally "on" n-channel device.
5. This treatment follows that given by H . K . J . Ihantola, "Design Theory of a Surface

Field-Effect Transistor," Stanford Electronics Laboratories Technical Report No.
1661-1 (1961). and by H . K_ J . Ihantola and J . L . Moll. "Design Theory of a
Surface Field-Effect Transistor," Solid-State Electronics, 7, 423 (1964).
The characteristics of such devices were also studied bv S. R . Hofstein and.F. P.
Heiman, "The Silicon Insulated-Gate Field-Effect Transistor." Proc. IEEE. 51. 1190
(1963); and.by C . T . Sah, "Characteristics of the Metal-Oxide-Semiconductor
332
Transistors." IEEE Trans. Electron Devices. ED-U, 324 (1964). For a discussion of
a somewhat different type of surface field-effect transistor structure, see, for instance,
H. Borkan and P. K . Weimer. " A n Analysis of the Characteristics of Insulated-Gate
Thin-Film Transistors," RCA Rev., 24, 153 (1963).
6. V. G . K . Reddi, "Tunable High-Pass Filter Characteristics of a Special MOS

Transistor. IEEE Trans. Electron Devices. ED-12, 581 (1965).
rsi
-a-
o
p
o
ce
O
h-
7. C . W. Mueller and P. H . Robinson, "Grown-Film Silicon Transistors on Sapphire,"

Proc. IEEE.
52, 1487 (1964); F . P. Heiman, "Thin-Film Silicon-on-Sapphire
Deep-Depletion M O S Transistors," and S. R . Hofstein. ".An Analysis of DeepDepletion Thin-Film MOS Transistors," IEEE Trans. Electron Devices, ED-13,
(December 1966).
o
u
u
11.3
Rederive the current-voltage characteristics, neglecting the charge Q within the

surface depletion region. Compare with Equation U.S. Under what condition
is this a reasonable approximation? Also, derive the conductance in the linear
region and the transconductance in saturation, neglecting GaDerive an expression giving the electric field along the channel. Compare its
magnitude for the device used for illustration in this chapter with the magnitude
of the surface field under various operating con4itions.
11.5
11.6
11.7
5?
The current-voltage characteristics of surface field-effect transistors are often

examined in a two-terminal mode, with the drain and gate connected together, and
the source and substrate grounded. Derive the current-voltage characteristics
in this configuration. Under what conditions is it possible to infer the turn-on
voltage V from these characteristics?
A surface field-effect transistor can be operated as a four-terminal device. For
example, the source mav be reverse biased with respect to the substrate.
(a) Derive the current-voltage characteristics, the conductance, and transconductance in such a case two ways: with substrate held at ground potential,
and with source held at ground potential.
(b) What is the turn-on voltage (with respect to ground) in these cases? Compare
with results obtained in Chapter 10.
(c) Derive the transductance with respect to the source-substrate voltage.
In many applications it is important to have a device with a square-law characteristic, i.e., / cc y . Examine the suitability of the surface field-effect transistor
from this standpoint. In particular, consider the best mode of operation and
the optimum choice of material and structural parameters for this purpose.
Derive the dependence of the channel conductance in the linear region on temperature, at a given gate voltage. Assume that the flat-band voltage is independent
of temperature, and that the inversion layer mobility is inversely proportional to
the absolute temperature.
Derive an expression for the channel conductance in the linear region in the
presence of surface states uniformly distributed in energy throughout the forbidden gap, with density D ( c n r ' e v - ) . Compare the effect of such surface
states with the effect of an inversion layer mobility which is a function of the
surface field.
1
st
VI
N M N I-J
>
N
ft
11.4
.o
I
ll.2
PROBLEMS
11.1
-9-
I"
S 2
-C
VI
VI
Ci
-9CN
41c
l-J
<
Z)
a>
z
z
- <
Ô
Ui X
_l u
-3-
.-
< Q
O
-9(N
C
-J
=>
X
oC
I
-9-
o"
A
c
o
N H
u.
II
t-
s?
>
<N
I n
c
o
VI
k"
.o
1I
XI
to
<
XI
u_
VI
VI
rN
1
a
sf
a o
II
N M
SO
II
c5
ii
z
<
H
m
'_)
<-n
*
u
u
o
o ta
ed
Q
333
<u
3
o-
Sg
"3
t=
EU
FAST S U R F A C E
STATES
SPACE
WITHIN
CHARGE
SURFACE-STATE
BARRIER
SURFACE
335
CHARGE
ENERGIES
MOBILITY
CONDUCTION
SURFACES
OTHER
Fast Surface States

OXIDE
ON
OXIDE
INSULATORS
12
Silicon
Silicon
(a) Fast surface states
(b) Mobile impurity ions
Properties of t h e SiliconSilicon D i o x i d e System

Si0
I n the previous three chapters we have discussed the theory of semiconductor surfaces, the effect o f surfaces on p-n junctions, and surface fieldeffect transistors. A l t h o u g h most o f that discussion is o f general validity,
its usefulness depends on the availability o f specific knowledge o f a particular interface system.
As a result o f extensive studies using M O S capacitors, gate-controlled
diodes, and M O S surface field-effect transistors, the silicon-silicon dioxide
system has become a very well characterized interface; A detailed understanding o f many o f its features is still lacking, but our empirical k n o w l edge o f and our ability to c o n t r o l the properties o f this interface are quite
extensive.
I n this chapter, we present a summary o f the electrical characteristics o f
the silicon-silicon dioxide system. We begin w i t h a discussion o f the
various types of charges and states associated w i t h this system. These are
summarized in FigUTe 12.1. They include fast surface states located at the
oxide-silicon interface. I n a d d i t i o n , there may be space charges present
w i t h i n the oxide layer due to mobile impurity ions, e.g., sodium ionic cont a m i n a t i o n , or due to traps ionized by i r r a d i a t i o n . Finally, there is a fixed
surface-state charge located at the interface between oxide and silicon.
F o l l o w i n g this, we discuss measurements o f the barrier energies associated with the ineial-oxide-silicon system and the i n f o r m a t i o n they yield
about the band structure o f this system. We then discuss the mobility o f
334
Si0
+ + + + + +
+ + + +
+
(3> (+) ffl @ @
Silicon
Silicon
(
(c) Traps ionized by radiation
Fig. 12.1
(d) Surface-state charge
Charges and states associated with the Si/SiO. system.
electrons and holes i n silicon near the interface. The conduction process
that may take place on the outer surface o f the oxide is considered next.
Finally, we briefly review what is k n o w n about other insulators used i n
M I S structures on silicon. "
12.1
FAST
SURFACE
STATES
1
I t has been predicted theoretically, by T a m m a n d by Shockley, that

because o f d i s r u p t i o n of the periodicity o f the lattice at a surface, a h i s h
density o f states w i l l be introduced into the forbidden gap near a semiconductor surface. Such states have traditionally been called fast
surface
states. Theoretically, i t is believed that there should be o f the order o f one
fast surface state for every surface atom, resulting i n a density o f about
1 0 c m . I n fact, experimental observations w i t h surfaces obtained
by cleaving under high vacuum bear out this p r e d i c t i o n . I n contrast,
etched germanium and silicon surfaces, which are always covered bv
a very t h i n oxide layer, show fast surface-state densities o f the order o f
15
- 2
Properties
336
10
12
- 2
to 10 c m .
of the Silicon-Silicon
Dioxide
System
Space C h a r g e w i t h i n
I n i t i a l investigations o f thermally oxidized silicon
surfaces have found.such surfaces to c o n t a i n fast surface states i n densities
337
- 2
_ 1
st
1 0
Oxide
contribute effectively to surface recombination and generation. Calculations show " that N i n Equation 12.1 is then replaced by -kTD,
where
D is the density ( c m e v ) o f the u n i f o r m l y distributed centers.
I n the case o f samples which were not annealed, the existence o f large
densities o f single-level surface states slightly below the conduction band
edge and slightly above the valence band edge have also been observed.
Because of the location o f these states, they do n o t significantly affect the
characteristics o f semiconductor devices. ( I n fact, for the same reason a
special technique had to be developed for their measurement. )
The relatively low density o f fast surface states at thermally oxidized
silicon surfaces has been an i m p o r t a n t advantage o f the planar technology.
o f the same order.

We saw i n Chapter 9 that fast surface states result i n a deviation i n the
shape o f the capacitance-voltage or channel conductance-voltage characteristics o f metal-insulator-semiconductor structures f r o m the ideal
theoretical curve shapes. I n fact, the experimental characteristics o f
aluminum-silicon dioxide-silicon structures fabricated by thermal oxidat i o n , including a low-temperature annealing step, follow the theoretical
curve shape very closely although they are displaced f r o m i t by a constant amount.
The degree o f parallelism o f experimental and theoretical characteristics
indicates that the t o t a l density o f surface states i n the middle 0.7-ev
p o r t i o n o f the silicon energy band is less than 5 x 1 0 c m . Measurements by different methodsusing the a-c conductance o f M O S capacit o r s , and the variation o f the t u r n - o n voltage o f M O S transistors w i t h
temperature yield results consistent w i t h this upper b o u n d .
I n Chapter 10 we saw that the reverse current versus gate voltage characteristic o f gate-controlled diodes directly yields the surface recombination
velocity 5,, which i n t u r n is related to the fast surface-state density N by
the relation (see also Chapter 5),
the
12.2 S P A C E C H A R G E W I T H I N T H E O X I D E
- 2
- a.
Ionic Contamination
st
s = <yv N, .
0
tli
(12.1)
8a
Such measurements have yielded values o f s between 1 and 10 cm/sec.

Comparison o f values o f s w i t h the corresponding values o f the fast
surface-state density. N, o n structures where they were both increased by
i r r a d i a t i o n verified E q u a t i o n 12.1, and yielded values for the capture
cross section, a, o f 1 0
to 1 0 ~ c m . W i t h such capture cross sections,
the p r e i r r a d i a t i o n values o f s correspond to fast surface-state densities o f
1 0 - 1 0 c m , w h i c h are consistent w i t h the upper b o u n d obtained by
the other methods.
The above values o f fast surface-state density were all obtained on
aluminum-thermally g r o w n silicon dioxide-silicon structures which were
subjected to a heat treatment step at about 500 C. Such a procedure has
been s h o w n to lead to an order-of-magnitude r e d u c t i o n i n the density o f
fast surface states.
F o r simplicity, this discussion has been i n terms o f single-level surface
states. A c t u a l l y , i t appears that the fast surface states are more or less
u n i f o r m l y d i s t r i b u t e d i n energy over the center p o r t i o n o f the energy
gap. A s we have seen i n Chapter 10, i n such a case only those centers
w h i c h are w i t h i n a few IcT i n energy o f the middle o f the energy gap
'1
"W
ab
- 1 6
1 5
10
- 2
A major difficulty encountered w i t h early M O S devices was that the

flat-band voltage was unstable, i.e., i t was subject to drift under bias at
elevated temperatures. A n example o f this drift behavior is illustrated i n
Figure 12.2. Here the capacitance-voltage characteristics observed
initially are marked by ( I ) , while those observed after 30 minutes at 127C,
w i t h V = + 1 0 v applied, are marked by (2). The characteristics could
be recovered; those observed after partial recovery obtained by heating
for 30 minutes at the same temperature w i t h the gate shorted to the substrate are marked by (3).
a
338
Properties of the Silicon-Silicon Dioxide
System
This drift process was shown to be due to the rearrangement o f an ionic

space charge distribution w i t h i n the oxide. We have seen i n Chapter 9
that a c o n t r i b u t i o n to the flat-band voltage,
9
fix) dx,
C.
Space Charge w i t h i n the O x i d e
339
flowing to the gate d u r i n g drift is identical w i t h the charge given by

AV C .
F i n a l l y , i t is i n agreement w i t h the results o f experiments
i n which part o f the oxide was gradually removed after drifting. Such
experiments showed that after drifting" all o f the positive charge was
located next to the silicon, as shown i n the middle p o r t i o n o f Figure 12.3.
FB
results when a space charge o f distribution p(x) is present i n the oxide.

Thus the drift o f the capacitance-voltage characteristics can be due simply
I t was i n f e r r e d b o t h from the rate o f the drift process and f r o m experiments i n which the oxide layer was intentionally contaminated w i t h s o d i u m
chloride that the uncontrolled drift i n M O S structures was due to trace
contamination by sodium. Subsequently, this has been verified by r a d i o active tracer measurements.
Such measurements have also shown t h a t
10
I
Initial
Afler recovery
After drift
with \' > 0
e
Fig. "12.3
process.'
Air/SiOj.1
Charge distributions pictured to correspond to various stages of the drift
.2
.3
.4
.5
.6
Distance from air interface (u)

Fig.
to the rearrangement o f the d i s t r i b u t i o n p(x) w i t h i n the oxide w i t h o u t any

change i n the total density o f ions per unit area,
p(x)
w i t h i n the oxide. T h i s rearrangement is illustrated in Figure 12.3 where

we show the initial charge d i s t r i b u t i o n (where all o f the positive ionic space
charge is pictured to be next to the metal, exerting no influence on the silicon), the d i s t r i b u t i o n after positive bias drift (resulting i n all o f the oxide
space charge being located next to the silicon and therefore i n d u c i n g its
image charge in the silicon), and finally the distribution obtained after the
recovery o f the ionic space charge back to the vicinity o f the metal
electrode.
T h i s picture of a rearrangement o f an ionic charge w i t h i n the oxide is i n
agreement w i t h all o f the experimental observations. I t fits the fact that
the m a x i m u m change i n flat-band voltage is independent o f the temperature or the magnitude o f the positive bias applied during the experiment
even though the rate o f the process itself is dependent on b o t h . I t is i n
agreement w i t h the observation that the time-intesral o f the current
12.4
Sodium concentration distribution in the oxide after drift.
10
the distribution o f the sodium ions i n the oxide at various stages o f the
drift process indeed corresponds to_the postulated distributions shown i n
Figure 12.3.
A n example o f the sodium distribution i n an oxide layer which was not
intentionally contaminated with sodium, i n an intermediate stage o f the
drift process, is shown i n Figure 12.4. I t is interesting to compare this
figure w i t h the results o f numerical c o m p u t a t i o n s
dealing w i t h the
transient i o n t r a n s p o r t problem i n the oxide layer. The results o f such
computations are shown i n Figure 12.5 for the particular case o f + 1 0 volts
applied to the gate at 127C for various lengths o f time. N o t e that
the beginning and the final distributions correspond to the first andsecond
sketches shown i n Figure 12.3. Between these extremes, i t is evident that
the sodium moves across the oxide w i t h a U-shaped d i s t r i b u t i o n p r e v a i l i n g
i n the intermediate stages, i n qualitative agreement w i t h the experimental
results shown i n Figure 12.4.
11
The mobile s o d i u m i m p u r i t y , being due to an external c o n t a m i n a t i o n ,

could be eliminated by appropriate precautions i n the device fabrication
procedure. Once this contamination was eliminated, M O S devices became
stable even under bias at elevated temperatures.
3,12
Surface-State
Properties of the Silicon-Silicon Dioxide System
340
10x 10
10"'
c_
t =
1 0 -
k i o -
IO"
ir
= 3 min.
10-3
0.2
0.4
0.6
0.8
1.0
10"
0.2
0.4
0.6
0.8
.v-'.v.
x/x.
1
1
->
10"'
'
g l 0 "
I
1.0
10
10"
Fig. I2.6
" 1 0 -
10~ 0
J
0.2
0.4
0.6
0.8
1.0
10"*
/
0
r
0.2
0.4
l
0.6
0.8
1.0
5 l 0 "
14
0.2
0.4
0.6
0.8
1.0
10-
10""
.X/.X,
Fig. 12.5
14
A n idealized model for the build-up o f such a space charge i n the oxide
under ionizing radiation exposure is as f o l l o w s .
D u r i n g irradiadon,-
10"
The build-up of excess charge as a function of Irradiation time.
voltage applied d u r i n g irradiation. I t is evident that the excess charge

0/ induced i n the silicon by the space charge i n the oxide first increases
w i t h time and then reaches a steady-state or saturation value which, i n
t u r n , is dependent o n the gate bias applied d u r i n g irradiation.
_i
.v. x.
v v..
20
Time (min.)
t = 5 min.
IO"'
b.
341
1 2
10"* 0
time.
Charge
0.2
0.4
0.6
0.8
1.0
xix.
Theoretical sodium concentration distribution In the oxide as a function of
11
Radiation-Induced Space Charge
A positive space-charge b u i l d - u p i u silicon dioxide films has been

observed to result f r o m exposure to ionizing radiation o f various kinds,
i n c l u d i n g X - r a y , gamma ray, l o w - and high-energy electron i r r a d i a t i o n .
The physical o r i g i n o f this charge is entirely different f r o m the ionic contamination discussed above.
Experimental results "pertaining to X-irradiated silicon dioxide layers
are shown i n Figure 12.6 as a function o f time, for various values o f gate
13
14
(b)
fc)
Intermediate
Final steady state
Fig. 12.7 Idealized model for the space-charge build-up as a function of time during
irradiation of an MOS structure under a positive gate bias.
l n i t l a l
11
Properties of the Silicon-Silicon Dioxide System
342
:
s ;c'_ron-hole pairs w i l l be generated i n the oxide. I f there is no electric

fieici present i n the oxide, the electrons and holes will recombine. resulting
i n no net charge b u i l d i n g u p i n the oxide. However, i f there is an electric,
field present i n the oxide, this field w i l l tend to separate the electrons and
holes. I n particular, a field corresponding to a positive gate voltage w i l l
tend to p u l l electrons t o w a r d the gate electrode. I f no electrons can enter
f r o m the silicon into the oxide, t r a p p i n g o f the holes near the oxide-silicon
interface w i l l result i n a gradual b u i l d - u p o f a space charge as shown i n
parts (a) and (b) o f Figure 12.7. D u e to the increased electric fields, an
increasing fraction o f the t o t a l applied voltage will be dropped across this
space-charge region. Thus, as the space charge grows, eventually the field
i n the rest o f the oxide layer is brought to zero. This results i n a steadystate situation, depicted i n Figure 12.7c.
T h e radiation-induced oxide space charge can be annealed out relatively
" r a p i d l y by heating above 3 0 0 C .
12.3
SURFACE-STATE
CHARGE
3
Experiments w i t h M O S structures showed the unexpected existence o f

a fixed charge, apparently located near the interface. This charge results
i n a parallel translation o f the capacitance-voltage (or channel conductancevoltage) characteristics along the voltage axis. I t is called the surface-state
charge, and its density per unit area is designated by the symbol 0 .
The characteristics o f the surface-state charge have since been experimentally studied very extensively.
The surface-state charge has the f o l l o w i n g properties:
343
Surface-State C h a r g e
rough agreement w i t h the order o f variation o f the linear o x i d a t i o n rate

constant BjA for these three orientations (see Chapter 2).
The effect o f oxidation and annealing conditions is summarized i n
Figure 12.8. Here we show the surface-state charge density O resulting
after oxidation i n dry oxygen or wet oxygen, and after heat treatment o f
both dry and wet oxides i n a nitrogen ambient. T h e results presented here
are extremely reproducible. A further feature o f these results is t h a t the
only relevant heat treatment is the final one. Regardless o f the previous
history of a sample, the final heat treatment w i l l determine the value o f
<2 , provided only that sufficient time is allowed f o r the sample to reach
steady state under the given conditions.
ss
SS
The above results p o i n t to the role o f excess ionic silicon in the oxide i n
the origin o f the surface-state charge.
I t appears that the surface-state
charge is due t o excess ionic silicon present i n the oxide d u r i n g o x i d a t i o n ,
waiting to react w i t h the oxidizing species that has diffused across the
oxide during the oxidation process, as illustrated i n Figure 12.9.
15
A n additional indication o f the role o f excess silicon i n the oxide has

been provided by a series o f experiments i n w h i c h the value o f <2 was
reproducibly and markedly increased by heating samples under negative
gate voltage conditions. (Note that this c o n d i t i o n is the opposite o f t h a t
required to b r i n g about a similar effect due to i o n i c contamination.) T h e
rate of this process is about four orders o f magnitude smaller than the
rate o f sodium rearrangement. Such results are shown in Figure 12.10,
S !
15
1. I t is fixed; i t cannot be charged or discharged over a wide v a r i a t i o n

o f bending o f the silicon energy bands.
2. I t is unchanged under conditions that would lead to the m o t i o n o f
sodium ions i n the oxide, or to the annealing o f radiation-induced space
charges.
3. I t is located w i t h i n 200 A o f the oxide-silicon interface.
4. Its density 0 is n o t significantly affected by the oxide thickness, or
by the type or concentration o f impurities i n the silicon.
5. O
strong f u n c t i o n o f the o x i d a t i o n and annealing conditions, _
and the orientation o f the silicon crystal.
S !
l S
ss
16
The orientation effect can be summarized in an approximate fashion

as follows: the r a t i o o f surface-state charge density 0 values under a
given oxidation c o n d i t i o n for silicon oriented along the (111). (110). and
(100) directions w i l l be i n the r a t i o o f approximately 3 : 2 : 1 . This is i n
Temperature (C|
Fig. 12.8 Dependence of the surface-state charge density on the ambient and temperature of the final heat treatment.
15
\
Barrier
Energies
345
where we show the final steady-state value o f 0 as a function o f the

applied field across the oxide. The final surface-state charge density is
evidently proportiona l to both the applied field and the initial surfacestate charge density.
12.4
BARRIER
ENERGIES
As discussed in Chapter 9, a difference between the metal and semiconductor w o r k functions results in a shift o f the flat-band voltage o f an
M O S structure. This shift provides one m e t h o d o f the evaluation o f this
w o r k function difference as illustrated i n Figure 12.11. I n this figure, we
0
1200_
-1
Dry 0 ,
(100)
-2
surface-state c h a r g e .
--3
15
-4
\ r
-5
-6
-7
-8
\ 9 2 0 Dry O ,
-9
-10
0
i
0.1
I
0.3
i
0.2
(111)
I
0.4
1
0.6
I
0.5
0.7
Fig. 12.11 Determination of metal-semiconductor work function difference from MOS

flat-band voltage measurements.
[Aluminum-SIO.-p type Si {N = 10" cm" )]
15
plot the flat-band voltage V

as a function o f oxide thickness for samples
prepared under the same conditions. Since the flat-band voltage is given"
FB
FB
MS -
ss
K
n
the fact that a straight line is obtained indicates that the surface-state
charge density 0 is independent of the oxide thickness i n agreement w i t h
our previous discussion. The intercept o f the lines corresponds to the
w o r k function difference, which is o f the order o f 1 volt for the present case
o f an aluminum-silicon dioxides-type silicon (N = 10 c m ) sample.
A m e t h o d o f evaluating the individual barrier energies is based on
the measurement o f the photoelectric thresholds at the metal-oxide and
16
- 3
Properties of the Silicon-Silicon Dioxide S y s t e m
346
Conduction on O x i d e Surfaces
347
i
17-19
tin- oxide-silicon i n t e r f a c e s .
I n this method, the M O S structure is
illuminated with light o f increasing energy under a given bias c o n d i t i o n .
W h e n the photon energy becomes sufficiently large to raise electrons f r o m
the filled valence band to the conduction band o f the oxide, conduction
begins across the oxide. The p h o t o n energy then corresponds to the barrier
energy o f the particular interface studied.
Barrier energies determined by b o t h the photoelectric response m e t h o d
and by the M O S method are tabulated i n Table 12.1. We also show for
METAL-SIO,
AND
TABLE
12.1
Si-Si0
BARRIER
ENERGIES
AS
i i
'
_
>^Holes
Electrons
-
Electrons
> Experiment
Holes
,
i !
10"
\QM
}
IO"
10
10
"*o.
N. T h e o r y N .
_i.
> i i
N
10'
(cm" )
i
10'
Fig. 12.12 Comparison between typical experimental inversion layer mobilities and
theory for diffuse surface scattering.
20
M E A S U R E D BY P H O T O E L E C T R I C E X C I T A T I O N A N D A S
INFERRED
FROM
MOS F L A T - B A N D
VOLTAGES.
S H O W N ARE THE CORRESPONDING V A C U U M

FUNCTION
Metal
<t> (Photo)
Mg
Al
Ni
Cu
Ag
Au
Si
2.25
3.2
3.7
3.8
4.15
4.1
4.35
ALSO
WORK-
VALUES."
(MOS)
corresponding b u l k mobilities. A t higher surface fields, the m o b i l i t y

begins to decrease slightly. These results are i n disagreement w i t h the
theory o f diffuse surface scattering, the predictions o f which are also
illustrated in Figure 12.12.
21
Vacuum
work function
The temperature dependence o f the inversion-layer m o b i l i t y i n the

constant-mobility region is f o u n d to b e fx cc J - . This is the temperature dependence predicted for b u l k m o b i l i t y ..by -the - theory o f lattice
scattering (see Chapter 4) w h i c h , however, had 4ieyef -been previously
observed.
"20
2.4
3.2t
3.6
3.8
4.2
4.1
3.70
4.20
4.74
4.52
4.31
4.70
5.15
t Arbitrarily chosen as reference.
comparison the vacuum w o r k function values o f the respective materials.

N o t e that the barrier energies f o l l o w r o u g h l y the same order as the w o r k
functions. The presence o f the oxide merely seems to result i n a l o w e r i n g
of the barrier energy as compared to v a c u u m by 0.5 to 1 ev.
12.5
SURFACE
MOBILITY
The mobilitv o f electrons and holes i n inversion layers has been studied
by using channel conductance measurements on M O S transistor structures.
F r o m the slope o f the conductance versus gate voltage plots, the
m o b i l i t y o f carriers i n the inversion layer was evaluated. A typical set o f
results for both electrons and holes is shown i n Figure 12.12. Over a fairly
large range o f variation o f the charge per u n i t area induced i n the silicon
Q (or equivalent]}', o f the surface field t Q the inversion-layer m o b i l i t y is
seen to be relatively constant, having a value approximately half that o f the
20
- 1
12.6
CONDUCTION
ON
OXIDE
SURFACES
22
Early investigators o f oxidized silicon surfaces have suggested that

conduction on the outer surface o f t h e oxide may play an i m p o r t a n t role
i n determining the characteristics o f underlying silicon devices. Tnis oxide
surface conduction process is illustrated schematically in Figure 12.13
where we show a gate-controlled n+p diode structure w i t h the gate biased
positively. Because o f conduction o n the outer surface o f the oxide, its
potential w i l l approach that o f the metal gate. This is indicated i n the
upper p o r t i o n o f the figure where we show the potential o f the oxide
surface for increasing times. I n essence, the "gate" w i l l be extended to a
larger area. This i n t u r n w i l l result i n the f o r m a t i o n o f a field-induced
junction, under this "extended gate." (The g r o w t h o f the field-induced
j u n c t i o n can actually be used to study the rate o f the surface c o n d u c t i o n
process. )
23
It can be s h o w n " that the surface conduction process is described bv an

equation o f the same form as the diffusion equation, w i t h the diffusivity
replaced by ljR-C
where R- is the sheet resistance o f a square-shaped
portion o f the outer surface o f the oxide and C is the oxide capacitance.
c
348
System
Conduction on O x i d e Surfaces
Accordingly, the oxide-surface potential in the case shown in-Figure 12.13

w i l l follow an erfc-type d i s t r i b u t i o n . Rate studies o f the charging process
can thus be used to determine the sheet resistance of the oxide surface R .
G
This picture o f the oxide surface conduction has been verified i n detailby an experimental technique which gives the potential o f the oxide surface
directlv. Using glass slides and this potential measuring technique, the
-49
conduction process tends to increase the inversion layer area. Thus as this
process continues, the silicon capacitance i n parallel with the oxide
capacitance will grow w i t h time. Correspondingly, the series combination
o f the two capacitances w i l l approach the capacitance o f the oxide layer
Fig. 12.13 Creation of a field-induced junction by charging outer surface of oxide

from a metal gate.1
25
sheet resistance o f glass surfaces has been studied as a function o f the

h u m i d i t y o f the atmosphere, a factor w h i c h affects the surface conduction
process very drastically. T h e results are shown i n Figure 12.14 along w i t h
data obtained on thermally oxidized silicon surfaces by measuring the rate
o f g r o w t h o f field-induced j u n c t i o n s . The two sets o f data are i n reasonably good agreement a n d show the tremendous influence o f h u m i d i t y i n
excess o f 4 0 % on the sheet resistance. Since the time constant o f the oxide
surface conduction process is given by ~R C ,
the oxide charging process
w i l l proceed very rapidly i n a h u m i d atmosphere, and very slowly i n a d r y
atmosphere.
A n interesting consequence o f the oxide surface conduction process is
its effect o n M O S characteristics as illustrated i n Figure 12.15. The surface
23
20
40
60
80
100
Relative humidity (%)

Fig.
12.14
Sheet resistance of glass plate and of thermal SIO, vs relative humidity."-"
alone because o f the increasingly large silicon capacitance up to much

higher measurement frequencies than otherwise.
Thus, as indicated i n
Figure 12.15, the shape o f the M O S capacitance-voltage characteristics
w i l l approach the low-frequency type characteristics.
28
3SX
Properties of the Silicon-Silicon Dioxide S y s t e m
O t h e r Insulators
351
10
radioactive tracers
showed that this effect is due to a m u c h higher
solubility o f sodium in the phospho-silicate glass than i n the silicon
dioxide. Thus sodium tends to segregate i n the phospho-silicate glass
layer and, therefore, it is not available for drift across the underlying silicon
dioxide layer.
Oxide
10
Experimental measurements
o f the sodium d i s t r i b u t i o n w i t h i n a
phospho-silicate glass/silicon dioxide double layer system are shown i n
Figure 12.16. N o t e the three orders o f magnitude enhancement o f the
sodium concentration i n the phospho-silicate glass layer relative to the
underlying silicon dioxide film.
28
I t has also been f o u n d that, independently o f sodium c o n t a m i n a t i o n ,

there is a dipole-type p o l a r i z a t i o n process associated w i t h the phosphosilicate layer itself. This dipole-type polarization can be shown to yield a
shift i n the flat band voltage, the steady-state value o f which is given by
1
Fig.
12.15
-t'o
The effect of oxide surface conduction on MOS capacitance-voltage charac-
&V =
- ^ V
PB
(12.2)
teristics.
12.7
OTHER
INSULATORS
I n order to fully utilize the potential capabilities o f metal-insulatorsemiconductor structures, certain combinations o f insulators may be more
advantageous than the metal-silicon dioxide-silicon system itself. O f the
numerous possibilities, several have already been studied.
I t has been f o u n d that i f a phospho-silicate glass, i.e., S i 0 rich i n
P 0 , is present on the outside o f a silicon dioxide layer, the instability o f
contaminated M O S devices is substantially reduced. Experiments w i t h
27
P;Oj-SO;
10
1 6
0
Fig.
12.16
.1
.2
.3
.4
.5
.6
.7
Distance from air interface (/d
The effect of a P.O -rich layer on the sodium concentration distribution
within the oxide.
10
where the p r o p o r t i o n a l i t y constant a depends o n the phospho-silicate

glass-to-oxide thickness ratio, x /x .
This equation shows that the shift i n
the flat-band voltage can take place i n either d i r e c t i o n , depending on the
p o l a r i t y o f the gate voltage applied during the p o l a r i z a t i o n process V .
c
Experimental measurements obtained w i t h a sample which consisted

o f a thick phospho-silicate glass layer on a t h i n silicon dioxide layer are
shown i n Figure 12.17. N o t e the displacements o f the M O S characteristics
in either d i r e c t i o n from the original one, w h i c h is denoted by V 0.
A n o t h e r double-layer dielectric system, w h i c h has been studied i n d e t a i l ,
is the lead-silicate glass/silicon dioxide double l a y e r .
Thi:- aouble-layer
system also showed considerable polarization effects w h i c h were due to the
rearrangement o f lead ions. These ions are originally u n i f o r m l y d i s t r i buted along w i t h their negative anions i n the lead-glass f i l m . I f a p o l a r i z i n g
voltage is applied, the lead ions can be rearranged, while the anions, w h i c h
are much less mobile, cannot. Because o f this, a space charge w i l l b u i l d
u p i n the lead-glass layer, w h i c h leads to an induced image charge i n the
silicon and, correspondingly, to a shift i n the fiat-band voltage. T h e
steady-state value o f this shift is given by
P
29
Thus this type o f polarization w i l l also be symmetrical and its d i r e c t i o n

will depend on the polarity o f the applied p o l a r i z i n g voltage.
The drift mechanisms associated w i t h these double-layer dielectrics are
illustrated in comparison to the sodium drift mechanism i n silicon dioxide
352
System
References Cited
353
i n Figure 12:18. I n this figure, we show the initial charge d i s t r i b u t i o n in

the insulator, and the charge d i s t r i b u t i o n after a negative a n d a positive
p o l a r i z i n g voltage.
Case (a) corresponds to sodium contamination in thermally g r o w n silicon dioxide. I n i t i a l l y , the positive sodium ions are at the outer interface.
T h e i r d i s t r i b u t i o n is not changed substantially when a negative voltage is
applied to the sample, but the sodium ions can be driven across to the
oxide-silicon interface w i t h the application o f a positive gate voltage.
Case (b) depicts an insulator containing an initially u n i f o r m d i s t r i b u t i o n o f
m o b i l e positive ions compensated by immobile negative ions corresponding to the case o f lead glass. This d i s t r i b u t i o n can be disturbed by either
negative or positive gate voltage. The positive ions w i l l pile up near the
negative terminal i n either case, leaving behind a negative space charge
near the positive terminal. The combined space charges w i l l then induce
an image charge i n b o t h the metal and the silicon as indicated i n the figure.
Case (c) depicts p o l a r i z a t i o n due to orientation o f dipoles, a n example o f
w h i c h is p r o v i d e d by the phospho-silicate glass. A. u n i f o r m d i p o l a r polari z a t i o n can be considered to be equivalent to equal and opposite surface
charges appearing o n the two faces o f the dielectric. These then induce an
opposite image charge i n the metal and i n the semiconductor as illustrated.
A n o t h e r silicon c o m p o u n d , silicon nitride ( S i N ) , has been studied
b o t h as a possible insulator i n M I S structures and in conjunction w i t h
silicon d i o x i d e .
Silicon n i t r i d e has the attractive feature that the diffusivity o f various impurities i n i t i n particular, that o f sodium are
3
la)
31
32
33
(bl
Fig. 12.18 The charge distributions in an MOS structure initially, and after drifting
to saturation under negative and positive gate bias. Three cases are shown:
(a) Corresponds to a thermally produced oxide contaminated with mobile positive ions
at the outer surface.
(b) Corresponds to a glass containing an initially uniform distribution of mobile positive
Ions compensated by immobile negative Ions.
(c) Corresponds to a glass with a uniform dipolar polarizabillty. A uniform dipolar
polarization is equivalent to equal and opposite surface charges on the two faces of
the dielectric.
30
m u c h lower than i n silicon dioxide. Thus surface field-effect transistors

made w i t h silicon nitride would be less susceptible to the ionic cont a m i n a t i o n problem described earlier than those made w i t h silicon dioxide..
However, i t appears that silicon nitride is inferior to silicon dioxide in
other ways: i t is much more conductive, and contains large densities o f
traps which lead to instability o f the flat-band voltage o f M I S devices. '
31
REFERENCES
-20
-15
-10
10
15
CITED
20
V (v)
c
Fig. 12.17
34
Polarizacion phenomenon associated with phospho-silicate layers.-'
1, A.. Many, Y . Goldstein, and N. B . Grover, Semiconductor Surfaces, Wiley, 1965,

Chapters 5 and 9.
354
Properties of the Silicon-Silicon
Dioxide
System
355
2. V. M . Atalla, E . Tannenbaum, and E . J . Schreibner, "Stabilization of Silicon

Surfaces by Thermally Grown Oxides," Bell System Tech. J., 38, 749 (1959).
19. B. E . Deal, E . H . Snow, and C . A . Mead. "Barrier Energies in Metal-Silicon

Dioxide-Silicon Structures." J. Phys. Chem. Solids, 27, 1873 (1966).
3. A . S. Grove, B . E . Deal, E . H . Snow, and C . T . Sah, "Investigation of Thermally

Oxidized Silicon Surfaces Using Metal-Oxide-Semiconductor Structures," SolidState Electronics, 8, 145 (1965).
20. O. Leistiko. A . S. Grove, and C . T . Sah. "Electron and Hole Mobilities in Inver- sion Layers on Thermally Oxidized Silicon Surfaces," IEEE
Trans. Electron
Devices.' ED-12, 248 (1965)'.
4. E . H . Nicollian and A . Goetzberger, " M O S Conductance Technique for Measuring

Surface State Parameters," Appl. Phys. Letters, 7, 216 (1965).
21. J . R . Schrieffer, "Effective Carrier Mobility in Surface-Space Charge Layers,"

Phys. Rev., 97, 641 (1955).
5. L . Vadasz and A . S. Grove, "Temperature Dependence of M O S Transistor Characteristics Below Saturation," IEEE Trans. Electron Devices ED-13,863
(December,
.1966).
22. M . M. Atalla. A . R . Bray, and R . Lindner. "Stability of Thermally Oxidized

Silicon Junctions in Wet Atmospheres," Proc. IEE, 106, P a n B, Supplement No. 17,
1130 (1960).
6a. A . S. Grove and D . J . Fitzgerald, "Surface Effects on P-//Junctions: Characteristics

of Surface Space-Charge Regions under Non-Equilibrium Conditions," Solid-Stale
Electronics, 9, 783 (1966).
6b. D . J . Fitzgerald and A . S. Grove, "Radiation-Induced Increase in Surface Recom- bination Velocity of Thermally Oxidized Silicon Structures," Proc. IEEE (Correspondence) 54, 1601 (1966).
7. G . Cheroff, F . Fang, and F . Hochberg, "Effect of Low Temperature Annealing on
the Surface Conductivity of Si in the Si-SiO.-Al System," IBM Journal, 8, 416
(1964); P. Balk, "Effects of Hydrogen Annealing on Silicon Surfaces," Electrochemical Society Spring Meeting, San Francisco, May 1965, Abstract 109.
23. E . H . Snow, in "A Study of Failure Mechanisms in Silicon Planar Transistors,"

Technical Documentary Report Dec. 1965, R A D C Contract A F 3 0 (602)-3776.
24. W. Shockley, W . W. Hooper, H . J . Queisser, and W . Schroen, "Mobile Electric
Charges on Insulating Oxides with Application to Oxide Covered P-N Junctions."
Surface Science, 2, 277 (1964).
25. W. Schroen and W. W. Hooper, in "Failure Mechanisms in Silicon Semiconductors,"
Final Report R A D C A F 3 0 (602)-3016 (1964).
-8. P. V . Gray and D . M . Brown, "Density of SiO.-Si Interface States," Appl. Phys.
Letters, 8,31 (1966).
26. E . H . Nicollian and A . Goetzberger, "Lateral A C Current Flow Model for MetalInsulator-Semiconductor Capacitors," IEEE Trans. Electron Devices. ED-12, 108
(1965); S. R . Hofstein and G . Warfield, "Physical Limitations on the Frequency
Response of a Semiconductor Surface Inversion Laver," Solid-Slate Electronics, 8,
321 (1965).
9. E . H . Snow. A . S. Grove, B . E . Deal, and C . T . Sah, "Ion Transport Phenomena in

Insulating Films," J. Appl. Phys., 36, 1664 (1965).
27. D . R . K e r r . J . S. Logan. P. J . Burkhardt. and W. A . Pliskin, "Stabilization o f S i O

Passivation Layers with P 0 ."" IBM Journal. 8, 376 (1964;.
10. E . Y o n , W . H . K o , and A . B. Kuper. "Sodium Distribution in Thermal Oxide on

Silicon by Radiochemical and M O S Analysis," IEEE
Trans. Electron Devices,
ED-13, 276 (1966).
28. E . H . Snow and B. E . Deal, "Polarization Phenomena and Other Properties of

Phosphosilicate Glass Films on Silicon," J. Electrochem. Soc. 113, 263 (1966).
11. D . A . Tremere. unpublished.
29. E . H . Snow and M . E . Dumesnil, "Space Charge Polarization in Glass Films."

J. Appl. Phys., 37, 2123 (1966).
12. P. Lamond, J . Kelley, and M . Papkoff, "Stable M O S Transistors," Electro-Technology, Dec. 1965, p. 40.
30. B. E . Deal, E . H . Snow, and A . S. Grove, "Properties of the Silicon DioxideSilicon System," SCP and Solid State Technology, 9, 25 (1966).
13. For a review of "Surface Effects of Radiation on Semiconductor Devices," see

J . P. Mitchell and D . K . Wilson, Bell System Tech. J., 46, 1 (1967).
31. S. M . H u , "Properties of Amorphous Silicon Nitride Films," J. Electrochem.

113. 693 (1966).
14. A . S. Grove and E . H . Snow, "A Model for Radiation Damage in Metal-OxideSemiconductor Structures," Proc. IEEE (Correspondence), 54, 894 (1966).
32. G . H . Schneer, W. vanGelder, V . E . Hauser. and P. F . Schmidt, " A Silicon Nitride

Junction Seal on Silicon Planar Transistors," Paper 5.3 at the I E E E Electron
Devices Meeting, Washington, October, 1966.
15. B , E . Deal, M . Sklar, A . S. Grove, and E . H . Snow, "Characteristics of the SurfaceState Charge (Q) of Thermally Oxidized Silicon," J. Electrochemical Soc, 114,
266 (March 1967).
16. P. Balk, P. J . Burkhardt. and L . V. Gregor, "Orientation Dependence of Built-in
Surface Charge on Thermally Oxidized Silicon," Proc. IEEE (Correspondence), 53,
2133 (1965).
17. R . Williams, "Photoemission
Phys. Rev.. 14C. .569.(1965).
of Electrons from Silicon into Silicon Dioxide,"

.
18. A . M . Goodman, "Photoemission of Electrons from Silicon and Gold into Silicon
Dioxide," Phys. Rev.. 144. 588 (1966).
Soc,
33. J . V. Dalton, "Sodium Drift and Diffusion in Silicon Nitride Films," J. Electrochem.
Soc, 113, I65C (1966).
34. See papers (Abstract Nos. 146-163) given at the Electrochemical Society Fall
Meeting, Philadelphia, October, 1966.
Index
V*
C
v5*
Acceptors, 97
Accumulation, 265
Activation energy, epitaxial growth, 12
intrinsic carrier concentration, 95
oxidation, 23, 29
solid-state diffusion, 39
Alloy junction, 2
Amplifier, transistor, 213
Anodization, 22
Atomic or molecular weight, of Ge, Si,
GaAs, and S I O T (Table), 102
Autodoping, 83
Avalanche breakdown. 191-194
Average resistivity of diffused layers,
54-57
Band theory, 91-95
Band-to-band recombination-, 128
Barrier energies in M O S structures,
345-346
Base, factor, 218
of junction transistors, 210
resistance, 228-230
Base-width modulation (Early effect), 226
Bias, forward, 150, 161, 180-191
reverse, 150, 161, 172-180
Bipolar transistor [see Junction
transistors)
Boltzmann distribution, 99
Boltzmann statistics, 99, 100
Boundary layer, theory, 14-18
thickness, 16
"Box" impurity distribution, 54
Breakdown, avalanche, 191-194
critical field, for Ge, G a A s , Si, and
S i 0 (Table), 103
2
Breakdown, critical field

for Si, 193
'
of p-n junctions, 150, 191-201
soft, 200
Zener (tunneling), 191
Breakdown voltage, common-base, 230
common-emitter, 230-234
of diffused junctions, 196
of field-induced junctions, 305
of linearly graded junctions, 195
of one-sided step junctions, 194
of planar junctions, 197
reach-through limited, 199
surface effects on, 311
Bubble analogy, 94
Built-in electric field, 32, 59-63, 224
Built-in voltage, of field-induced
junctions, 269, 291
of step junctions. 157
Capacitance, 169
of one-sided step junctions, 171
small-signal, 169
Capacitance-voltage characteristics, of
MOS structures, 271-276 frequency effects on, 274
of p-n junctions, 169-172
Capture cross section, 131
of gold, 141
of surface states, 145
Carrier concentration, intrinsic, 96, 101
Carrier removal, 142, 144
Centers, recombination-generation, 129
Channel, characteristic, 305
357
1
C
.358
Channel, conductance, junction
field-effect transistors, 245. 250
surface field-effect transistors, 276,
318,324 .
currents, 305
junction field-effect transistors, 244
metallurgical. 330
surface field-effect transistors, 318
Charges in insulator, 279-282, 337-341
Chemical potential, of electrons, 98, 156
Chemical reaction, first-order, 10
Chemical surface-reaction rate constant.
10, 25
Chemical transport, 44
Collector of junction transistors, 210
- Collisions, time interval between, 107
Common-base mode, 212
breakdown voltage. 230'
current gain, 211, 219
Common-emitter mode. 212
breakdown voltage, 230-234
current gain, 211, 220
Complementary error function. 46
Complete ionization, 97
Concentration gradient, at junction. 48.
50, 165
Concentrations, electron. 100
hole, 100
intrinsic carrier, 96. 101
Conduction. 92
band, 92
band edge. 94
electron, 93
on oxide surfaces, 347-350
Conductivity type, 98
Conductivity modulation. 227
-Gontinuum. of recombinationgeneration centers, 302
of surface states, 302
Cooperative diffusion, 63
Critical field, for junction breakdown.
191, 193
Crystal structure, of Ge, -Si. and G a A s
(Table). 102-103
Current crowding, in junction transistors.
229
Current gain, of junction transistors. 219
common-bast. 2' :
common-emitter. 211
effect of collector current on. 220
Index
Current gain, of junction transistors,
effect of reverse bias leakage
current on. 219
small-signal, 213
surface effects on. 303. 309
Current-voltage characteristics, of
gate-controlled diodes. 298-304
of junction field-effect transistors.
248-251
of p-n junctions. 172-191
of surfacefield-effecttransistors. 321 -326
Cut-off frequency, of junction
field-effect transistors. 254
of surface field-effect transistors, 328
of junction transistors, 223
Decay, of injected carriers. 119-121
Degenerate semiconductors, 101
Density, of gases. 20
of Ge, Si, G a A s , and Si02 (Table), 102
of states, effective, for Ge, Si, and
GaAs (Table). 102
Depletion approximation. 157. 267
Depletion region, 156
surface, 266, 293
maximum width of. 268. 293
width or. 159, 163. 166-169
Dielectric constant, for Ge, Si, GaAs.
and S i 0 (Table). 103
Diffused junction. 3
breakdown voltage of. 196
space-charge region for, 167-169
Diffused layers. 43-58
evaluation of, 52-58
two-dimensional. 59
Diffusion, drive-in, 49-52
equation, 42
field-aided. 59-63
formulas (Table), 88
of electrons and holes, 113
of gold in silicon, 41
rate limitation, external, 65-69
solid-state: 35-83
space-charge effects on. 59-63
through silicon dioxide layers, 75-77
Diffusion current, forward bias.
183-186
reverse bias, 173, 175-177
Diffusion length, of impurities. 46
of minority carriers, 124
2
Index
Diffusivity, 37
in gases. 20
in silicon, 38-40
in silicon dioxide, 41
of electrons, 113
of holes, 113
thermal, 43
Diodes (see p-n Junctions)
Diodes, finite length (narrow base).
190
Distributed surface states or
recombination centers, 302
Distribution function, Boltzmann. 99
Fermi-Dirac, 98
Drain, junction field-effect transistors.
243
surface field-effect transistors. 318
Drift, 106
of ions in M O S structures, 337-340
velocity, 106
Drive-in diffusion, 43, 49-52
Donors, 97
Early effect. 226
Effective densities, of states. 100
for Ge, Si, and G a A s (Table). 102
Effective mass, 107
Einstein's relationship, 37. 113
Electric field. 152
built-in, 32, 59-63, 224
Electrochemical oxidation. 22
Electron, capture, 129
chemical potential of. 98
concentration, 100, 104
conduction, 93
diffusivity of, 113
emission. 129
emission probability. 131
mobility of. 108
potential energy of, 152
transport of, 106-114
Electrostatic potential, 152
Emission probability, of electrons, 131
or holes, 132
Emitter, efficiency. 218
factor. 218
of junction transistors. 2)0
Emitter-dip effect. 63
Emitter-push effect, 63
Energy bands. 91
359
Energy gap. for Ge, Si, GaAs. and S i O ;
(Table). 102
Epitaxial diodes, breakdown voltage
limitations. 199
Epitaxial growth, 7-20
redistribution of impurities in.
78-83
Equilibrium criterion. 101
Error function, complementary. 46
properties of (Table), 48
Etching, 8
External rate limitation, on diffusion.
65-69
Extraction of carriers, 118
Fast surface states, 283. 335-337
in thermally oxidized silicon. 336
Fermi-Dirac distribution, 98
Fermi-Dirac distribution function. 98
Fermi-Dirac statistics. 98
Fermi level. 98
as a function of temperature. 104
in equilibrium, 156
intrinsic. 99. 101
Fermi potentials. 157
Field-aided diffusion, 32. 59-6?. 224
Field-effect transistors (.tee Junction
field-effect transistors and Surface
field-effect transistors)
Field-induced channel surface field-effect
transistors, 330
formulas for (Table). 3?"
Field-induced junction. 269. 2yi. 305
breakdown voltage of. 305
Field-plate controlled p-n junction. 264
Film growth rate, 11, 17. 79
temperature dependence of. 12
Finite length (narrow base) diodes. 190
First-order chemical reaction, 10
Flat-band voltage, 278, 281
Flux. 10. 24. 36
Forbidden gap, 91
Formulas, for field-induced channel
for junction field-effect transistors. 259
for junction transistors. 242
for p-n junctions, 207
for surface space-charge regions. 288
in solid-state diffusion. 88
semiconductor physics. 116
Index
360
Formulas, semiconductors in nonequilibrium. 148
Forward bias. 150. 161
Forward current, diffusion. 183-186
recombination. 186-190
temperature dependence of. 188-189
Forward current-voltage characteristics,
empirical representation. 189
for Ge, Si. and G a A s junctions.
188-189
Fourier's law of heat conduction, 237
Four-point probe technique, 58
Frequency effects, on MOS capacitancevoltage characteristics, 274
Frequency limitation, of junction
of-junction transistors, 223
of surface field-effect transistors, 328
Furnace, oxidation, 23
Gallium arsenide, intrinsic carrier
concentration. 96
properties of (Table), 102-103
Gallium arsenide p-n junctions, forward
current. 189
reverse current. 179
Gas constant. 19
Gases, properties of, 18-20
Gas-phase mass transfer. 13
coefficient, 10, 14, 16, 18. 24
in terms of concentrations in the
solid, 25. 65, 67, 79
Gate, junction field-effect transistors.
243
Gate-controlled diode. 264. 290, 296
current-voltage characteristics of, 298
Gate leakage current, junction field-effect
transistors, 252
Gaussian impurity distribution, 50
Generation, of electron-hole pairs, 119
Generation current-, 173, 300
Generation rate, in reverse-biased
depletion region, 174, 301
surface. 301
Germanium, intrinsic carrier
concentration, 96
properties of (Table), 102-103
surface recombination velocity, 145
surface-state density, 144
Germanium p-n junctions, forward

current, 188
Gettering, 201
Gold, capture cross section. 141
diffusion in silicon. 41
effect on resistivity, 142
Graded base region. 224
Graded channel regions, junction
Gradient of impurity concentration, 48,
50
Grooving, 52
Grown junction. 1
Growth rate, epksxial film. I i . 17, 79
Henry's law, 24, 44
constant. 25, 67
High-current effects, 227
High-level injection. 119. 227
Hole, capture. 129
concentration, 100. 104
diffusivity of. 113
emission. 130
emission probability, 132
transport of. 106
Horizontal epitaxial reactor, 8
Ideal gas law, 19
Illumination, effect on junction reverse
current. 180
Impurity, concentration gradient.
48. 50
interstitial, 40
scattering, 109
substitutional, 39
Injected carriers, decay of, 119
Injection, 117 '
from a boundary, 125
high-level, 119, 227
low-level, 119
Instabilities in MOS structures, 337
Insulator, 92
charges in, 279-282, 337-341
lead-silicate glass, 351
phospho-silicate glass, 350
silicon nitride. 352
Integrated circuits, 3
Intermediate centers, recombinationgeneration, 129-134 -
Index _
Interstitial impurities, 40
Intrinsic carrier concentration, 96, 101
Intrinsic Fermi level, 99, 101
Inversion layer, 267
mobility, 346-347
temperature dependence of, 347
loniacontamination of oxide, 337-340
Ionization energy, 95
Ionizing radiation, effect on fast surface
states, 145
effect on oxide space charge, 340-341
Junction field-effect transistors, 243-259
channel, 244
channel conductance, 245, 250
comparison to surface field-effect
transistors, 320-321
current-voltage characteristics, 248-254
cut-off frequency, 254
.. drain, 243
drain-current saturation, 251
formulas for (Table), 259
frequency limitation, 254
gate. 243
gate leakage current. 252
graded channel regions, 253
linear region, 248
principles of operation, 244-248
saturation, 247
saturation region, 248
series resistance, effect of, 256
source, 243.
source-to-drain resistance, in
saturation, 255
transconductance, 252
turn-off voltage, 250
Junction transistors, 208-242
base factor, 218
base resistance, 228-230
base-width modulation, 226
breakdown voltage, 230-234
common-base breakdown voltage. 230
common-emitter breakdown voltage,
230-234
conductivity modulation, 227
current components, 214
current crowding, 229
current gain, 219-222
Early effect, 226
- -
Jo I
Junction transistors, emitter efficiency, 218

emitter factor, 218
frequency limitation. 223
graded base region. 224
high-current effects, 227
maximum voltage limitations.
230-234
minimum voltage limitations, 234-236
one-dimensional model. 209
planar. 208
punch-through condition. 230
recombination factor, 218
saturation. 234
surface effects, on current gain, 303,
309
i
surface recombination, effect of, 218,
303
switch, 214
terminology, 214 .
thermal limitation, 236-238
thermal resistance, 238
transit-time limitation, 222
transport factor, 219
Kinetics, of oxidation. 23-31
of recombination process. 127-134
of vapor-phase growth. 10-13
Lattice, constant for Ge. Si, GaAs. and
S i O i (Table), 102
mobilities for Ge. Si, and G a A s
(Table), 103
scattering, 109
strain, effect on diffusion, 63
Lead-silicate glass, 351
Lifetime, band-to-band recombination,
128
effect of radiation damage on, 143
in low-level injection. 134-136
of excess minority carriers. 121
within a reverse-biased depletion region, 174
Linear coefficient of thermal expansion
for Ge, Si, GaAs, and SIO2

(Table), 103
Linear region, junction field-effect
transistors, 248
surface field-effect transistors, 320
Linear oxidation law, 27
362
Index
Linear oxidation rate constant, 27

-temperature dependence of, 30
Linearly graded junction, 163-166
breakdown voltage, 195
built-in voltage, 165
capacitance, 171
maximum electric field, 164
space-charge region, 163-166
Liquid source, 43
Low-level injection, 119
lifetime in, 134-136
Majority carrier, 98, 105
concentration, as a function of
temperature, 105
Masking,'75-77
thickness, 77
Mass. effective. 107
Mass-transfer coefficient. 10, 14, 16. 18.
24, 67, 79
in terms of concentrations in the
solid, 25. 67, 79
"Mass-transfer control, 11, 26
Maximum electric field, in linearly
graded junctions, 164
in step junctions, 158
Maximum surface recombination
velocity, 140
Maximum voltage limitations, of
junction transistors. 230-234
Metal. 92
Melting point, f o r G e , Si, GaAs~, and
S i 0 (Table), 103
Metal-insulator-semiconductor
structures (see M O S structures)
.Metal-insulator-semicpnductor surface
field-effect transistors (see
Surface field-effect transistors)
Metallurgical channel surface field-effect
transistors, 330
Metal-oxide-semiconductor structures
(see M O S structures)
Metal-oxide-semiconductor surface
field-effect transistors (see Surface
Minimum voltage limitations of junction
transistors. 234-236
Minority carriers. 105
concentrations under forward bias. 185
MIS structures (see MOS structures)
2
MIS transistors (see Surface field-effect

transistors) - Mobility. 37
in silicon, 109
inversion layer, 347
lattice, for Ge, Si. and G a A s (Table).
103
of electrons. 108
of holes, 108
surface. 346
Molecular or atomic weight, Ge, Si,
G a A s , and S i O ; (Table), 102
Mole fraction, 9
MOS structures, 264
capacitance-voltage characteristics. 271
channel conductance. 276
drift (instability). 337
frequency-effects. 274
turn-on voltage, 273
MOS transistors (see Surface
Multiplication, 193
Multiplication factor. 194
Narrow-base diodes. 190
Non-equilibrium conditions. 117-148
Normally "off" surface field-effect
transistors. 330
Normally "on" surface field-effect
transistors, 330
np product (see pn product)
;i-type conductivity, 98
One-sided step junction, 159
capacitance, 171
Orientation effect, on oxidation rate. 30
on surface state charge, 343
Out-diffusion, 67
Oxidation, electrochemical, 22
furnace. 23
kinetics of. 23
rate constants, 27
space-charge effects on. 31
thermal. 22
Oxide, diffusivities in. 41
masking. 75-77
Index
363
Oxide, properties of, 334-355

space charge. 337-341
surface conduction. 347-350
Q of diffused layer, 47, 50

Q of transistor base. 226
Q . 342
Quasi-equilibrium. 184-185
Quasi-Fermi levels, 162, 184-185
Parabolic oxidation law, 27

rate constant, 27
temperature dependence of rate
constant, 29
Phospho-silicate glass, 350
Photocurrent, junction, 180
Planar junction, 149
Radiation damage, effect of, on fast

surface states, 145
on lifetime, 143
Radiation-induced space charge in

Planar technology, 3
Planar transistors, 208
Plasma oxidation, 22
p-n junctions, 149-207
alloy, 2
breakdown, 150, 191-201
built-in voltage, 157, 165
capacitance, 169-172
concentration gradient at, 165
current-voltage characteristics, 172-191
curvature, effect on breakdown of.
197
depth, 52
diffused, 3
field-induced, 269. 291. 305
forward current, temperature
dependence of. 188-189
gate-controlled. 264, 290. 296
grown. 1
one-dimensional model. 150
photocurrent, 180
reverse current, temperature
dependence of. 179
surface effects on, 289
transient behavior, 201-204
pn product, in equilibrium, 101
in quasi-equilibrium. 184
in space-charge region. 139, 184
Poisson's equation, 153
Polarization in insulators. 351-353
Potential energy of electrons, 152
Predeposition, 43-48
Properties of gases. 18-20
Properties of G e , Si, GaAs, and S i 0
(Table), 102-103
/7-type conductivity. 98
Punch-through condition in junction
transistors. 230
oxides, 340
Rate constant, chemical surface reaction.
10, 25
oxidation, 27
temperature dependence of,' 29-30
Reach-through limited junction
Reactor, horizontal. 8
vertical, 8
Recombination, band-lo-band. 128
centers. 129
continuum of, 302
origin of. 140-145
current. 183. 186-190
factor, junction transistors, 218
in surface space-charge region. 298-304
kinetics of, 127-134
through intermediaie centers, 1 2 9 - 3 4
Rectification. 150
Redistribution of impurities, during
thermal oxidation. 69-75
in a predeposited l a y c . 74
in epitaxial growth, 78-?3
Resistivity, 111-113
average, of base region. 230
of diffused layers, 54
Reverse bias. 150, 161-163, 172-180
leakage current, effect on current gain,
219
Reverse current, diffusion. 173. 175-180
generation, 173-175
of G e . Si. and G a A s junctions.
178-179
temperature dependence, 179
under illumination. 180
Reynolds number, 16. 17, 23
Saturation, junction transistors. 234
Saturation current, in junction
Index
364
Saturation current, in surface field-erfect
transistors. 326
Saturation region, junction field-effect
transistors. 248
Scattering mechanisms. 109
Schmidt number, 17
Segregation coefficient, 69, 74, 77
Semiconductor. 93
degenerate. 101
physics. 91-116
surfaces, 263-288
under non-equilibrium conditions,
117-148
Series resistance effect, on junction
field-effect transistors, 256
on surface field-effect transistors, 329
Shockley-Read-Hall theory. 129
Silicon, diffustvities in, 38-40
intrinsic carrier concentration. 96
properties (Table), 102-103
surface recombination velocity. 145
surface state densities in, 144
Silicon dioxide. 22
diffusivities in, 41
masking, 75-77
properties of, 102-103. 334-355
Silicon nitride. 352
Silicon p-n junctions, forward current,
188
reverse current, 178
Silicon-silicon dioxide system, properties
of. 334-355
Silicon tetrachloride. 7, 8
Small-signal capacitance. 169
Small-signal current gain, of junction
transistors, 213
Sodium contamination, in oxides, 339
Soft breakdown, 200
Solid solubility, 44
Solid source. 43
Solid-state diffusion. 35-88
Source, junction field-effect transistors,
-43
Source-to-drain resistance in saturation,
junction field-effect transistors. 255
Space-charge, radiation induced. 340

within insulator. 280
within oxide. 337-341
Space-charge effects, on diffusion. 59-63
on oxidation. 31
Space-charge neutrality. 101 .
Space-charge region, for diffused
junctions, 167-169
for linearly graded junctions. 163-166
for step junctions, 153-163
Specific heat of Ge, Si. GaAs. and
S i 0 (Table). 103
Step junction. 157
built-in voltage, 157, 161
maximum electric field. 158
one-sided, 159
space-charge region for, 153-163
Stagnant-film model, 13
Strong inversion. 267-268, 292
Substitutional impurities, 39
Surface conduction on oxides.
347
Surface depletion region. 266
width. 266. 267
Surface effects, on junction breakdown
voltage, 311
on p-n junctions, 289-316
on transistor current gain. 303, 309
Surface field-effect transistors, 264.
317-333
channel. 318
channel conductance, 324
comparison to junction field-effect
transistors, 320-321
current-voltage characteristics. 321-327
drain, 318
drain-current saturation, 320
effect of surface states, 327
field-induced channel, 330
frequency limitation, 328
gate leakage current, 327
linear region, 320
metallurgical channel, 330 .
normally "off", 330
normally "on", 330
other types. 329-331
principles of operation. 318-321
2
Index
Surface field-effect transistors,

saturation, 320
saturation region. 320
series resistance, effect of. 329
source, 318
source-to-drain resistance in
saturation, 329
surface scattering, 325
transconductance, 326
turn-on voltage, 324
Surface generation, current. 301
rate. 301
Surface mobility in silicon. 346
Surface potential, 267. 293
Surface-reaction control, 11, 26
Surface-reaction rate constant. 10
Surface recombination. 121, 136-140
centers, 301
effect on emitter efficiency, 218
Surface recombination velocity; 124.
136. 302
maximum, 140
of a surface without a surface
space-charge region, 139
on Ge and Si surfaces, 145
on thermally oxidized silicon. 145, 336
Surface scattering, effect on surface
field-effect transistors, 325
Surfaces, semiconductors. 263-288
Surface space-charge region. 137, 263
equilibrium. 264-271
non-equilibrium case, 290-296
. recombination-generation within.
298-304
Surface-state charge in thermally
oxidized silicon, 342
Surface states, 144, 282, 335-337
continuum, 302
effect on surface field-effect transistors.
327
in germanium, 144
in thermally oxidized silicon. 144, 336
single level, 336
uniformly distributed. 336
Switch, transistor. 214
Temperature dependence, of film
growth rate, 12
of junction forward current, 188-189
365
Temperature dependence, of junction

of linear oxidation rate constant. 30
of parabolic oxidation rate constant.
29
Terminology of junction transistors. 214
Thermal conductivity, for Ge. Si. GaAs.
and S i 0 (Table), 103
Thermal diffusivity, 43
for Ge, Si. G a A s , and S i 0 (Table), 103
Thermal limitation of transistors. 236-238
Thermally oxidized silicon. 334-355
Thermal oxidation. 22-34
redistribution of impurities in, 69-75
Thermal resistance.. 238
Thermal velocity of carriers. 108. 130
Transconductance, junction field-effect
transistors, 252
Transient behavior- of p-n junctions.
201-204
Transistors {see Junction transistors)
Transit time limitation. 222
Transport equation. 41
Transport factor, junction transistors.
219
Transport properties, of gases, 20
Trap, 130
Tunneling (Zener breakdown). 191
Turn-off time, for p-n junctions. 202
Turn-off voltage, junction field-effect
transistors. 250
Turn-on voltage, of MOS structures,
273, 297
of surface field-effect transistors. 324
Two-dimensional diffused layers, 59 ..
2
Uniform distributions, of recombination

centers. 302
of surface states. 302
Unipolar transistors. 243
Valence band. 92
edge, 94
Valence electrons, 91
Vapor-phase growth, 7-21
kinetics of, 10-13
Vapor pressure for Ge. Si, G a A s . and
S i 0 (Table). 103
Velocity of carriers, drift, 106
2
Velocity of carriers, thermal, .108

Vertical reactor. 8
Viscosity. 14
of gases, 20
Work function difference, effect on

MOS structures, 278, 345
.*
Zener breakdown (tunneling), 191
r
A ,
W
^
!
i
: I
\ P

Physics and Technology of Semiconductor Devices - As GROVE

Hochgeladen von

Dokumentinformationen

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

Physics and Technology of Semiconductor Devices - As GROVE

Hochgeladen von

Copyright:

Verfügbare Formate

<)

Physics and Technology of

Library of Congress Catalog Card Number: 67-17340

The 1950's was the decade in which semiconductor devices rose to

technology which are most intimately involved in determining the electrical

2.1 Kinetics o f Oxide G r o w t h

The Band Theory o f Solids

9.1 Characteristics o f Surface Space-Charge RegjonsEquil i b r i u m Case

8.1 Principles o f Operation

Surface Effects on p-n Junctions

12.1 Fast Surface States

10.1 Characteristics o f Surface Space-Charge RegionsNonEquilibrium- Case

12.7 Other Insulators

i m p u r i t y concentration gradient at the j u n c t i o n .

coefficient in the general relationship for the thermal o x i d a t i o n

coefficient in the general relationship for the thermal o x i d a t i o n

total concentration o f moiecules in a gas

gas-phase mass-transfer coefficient i n terms o f concentrations in

common-base current gain

gas-phase heat-transfer coefficient

collector-base reverse current, w i t h emitter open

segregation coefficient.of i m p u r i t y at the oxide-silicon interface

electron concentration at the surface

concentration of acceptor impurities

fixed surface-state charge density per u n i t area

ideal gas-law constant

t o t a l recombination rate (per u n i t time and volume)

sheet resistance o f a square

base spreading resistance

series resistance o f the collector region

surface recombination velocity

concentration of donor impurities

charge per unit area i n the semiconductor

density of surface recombination-generation centers per u n i t area

m a x i m u m surface recombination velocity

concentration o f bulk recombination-generation centers per u n i t

surface recombination velocity o f a surface w i t h o u t a

concentration of holes in ay>-type semiconductor

hole concentration at the surface

magnitude o f electronic charge, 1.6 x 1 0

total amount o f impurities per unit area in the solid

velocity o f undisturbed gas

net rate o f carrier recombination-generation in the b u l k (per

net rate o f carrier recombination-generation at a surface

charge per unit area w i t h i n the surface depletion region at the

u n i t time and volume)

total number o f impurities per u n i t area in the base region o f a

concentration o f holes i n a />-type semiconductor, i n e q u i l i b r i u m

average drift velocity

voltage d r o p across oxide

offset time in the general relationship for the oxidation o f silicon

surface potential, i.e., total potential variation across surface

depletion region w i d t h o f metallurgical j u n c t i o n

metal-semiconductor work function difference

Physics and Technology of

I n the grown junction method, a semiconductor crystal is g r o w n out o f

The search for a method which gives superior control o f the l o c a t i o n o f

However, as semiconductor devices reached out for new applications,

(b) Alloy junction

(c) Planar technology

I n the alloy junction method, a pellet containing acceptor-type impurities

1. J . A . Hoerni. "Planar Silicon Transistors and Diodes," IRE Electron

2. G . K . Teal, M . Sparks, and E . Buehler. "Growth of Germanium Single Crystals

The importance o f epitaxial g r o w t h in semiconductor device technology