This action might not be possible to undo. Are you sure you want to continue?
Switching of bipolar devices
10 lectures – 5 classes
Dr. K. Fobelets
k.fobelets@imperial.ac.uk
Taken from www.necel.com/process/en/ux6b_90nm_bicmos.html
1 EE2  Semiconductor Devices 2012
The course aims:
(1) To review the operation of diodes and bipolar junction transistors.
(2) To extend knowledge on bipolar devices to include the influence of recombination.
(3) To investigate the physical mechanisms underlying the delays and speed limitations of the devices.
(4) To extract equivalent circuit models for the devices.
Objectives:
Students should be able to
(1) Explain qualitatively the mechanisms of electronic conduction in bipolar devices, and calculate relevant
quantities from given data.
(2) Calculate and explain DC currentvoltage behaviour of diodes and BJTs, given their geometry and
material properties.
(3) Explain the influence of excess minority carrier recombination of the performance of the devices.
(4) Explain the large signal behaviour of the devices from the internal storage of charge.
Recommended textbooks
EE1 background knowledge on semiconductors and semiconductor devices is required. The EE1 course
notes can be found on blackboard.
It is strongly recommended that you read these notes in order to understand the basic principles of
semiconductor devices.
“Solid State Electronic Devices”, B.G. Streetman & S. K. Banerjee, Prentice Hall International Editions, 6th
ed. This is a more expensive book but the course is based on this book. It contains a lot more than what is
taught in the 1
st
and 2
nd
year.
“Microelectronic Devices”, K.D. Leaver, IC Press, 2nd ed. This is a compact and relatively inexpensive
book which covers all the essentials.
Note that in popular books such as “Microelectronics circuits” by Sedra & Smith some very brief
descriptions of semiconductor device operation can be found. Whilst previous two books are very physics
oriented, this book is more oriented towards the circuit engineer and thus might give additional insight.
There are plenty of copies available of all these books in the Central Library.
An interesting online text books can be found on:
http://ecee.colorado.edu/~bart/book/book/contents.htm
http://www.ecse.rpi.edu/~schubert/CourseECSE2210MicroelectronicsTechnology2010/
2 EE2  Semiconductor Devices 2012
Course synopsis
1. Revision on semiconductor devices
Last year you studied the DC behaviour of metalsemiconductor, metaloxidesemiconductor and
semiconductorsemiconductor junctions. We looked at the simplified version of three devices: the pn
junction, the MOSFET and the BJT. We will start the course with reviewing the basic device DC operation
concepts of the pn diode and the BJT because this year we will focus on bipolar devices. For more extensive
information on MOSFETs please see the 3
rd
year Advanced Electronics Devices course
1
.
2. Long pn diode
In contrast to the “short” devices we’ve studied in the first year, this year we will introduce recombination
processes and check how they influence device behaviour. These devices will be referred to as “long”
devices.
Although an important characteristic of semiconductor devices is their fast response time, some delays do
occur. We will investigate the physics behind the delays and investigate the resulting switching
characteristics. The concept of a fast switching majority carrier device and a “slowly” switching minority
carrier device will be made clear.
3. The BJT
An npn or pnp junction is the basis for the bipolar junction transistor (BJT). Based on our knowledge of pn
junctions, the DC functioning of a BJT will be described. As in pn junctions, minority carriers will play an
important role in the conduction mechanism of the BJT. We will look explicitly at the short BJT and
discover that calculating the currents in this particular configuration is very straightforward. The BJT in
which recombination is taken into account offers some challenging maths but gives an insight into delays
and current gain. Switching of the BJT will also be studied and methods to reduce the delay discussed.
Why study semiconductor devices?
There can be lots of reasons to study devices. It can be as simple as being interested in the physics behind the
devices’ operation or it can give you the background to understand the aggressive downscaling effort of the
CMOS industry and maybe in the future you can join in the effort. Having knowledge of the operation of
1
On blackboard
3 EE2  Semiconductor Devices 2012
devices will enable you to understand both device and circuit modelling programs better and thus make
modelling and the resulting circuits more efficient.
Although the speed of the aggressively downscaled MOSFET
2
has removed the need for bipolar transistors
from logic circuits, microprocessors and memory chip, there are still some fields in which bipolar transistors
remain important. These applications mainly need ultrahigh speed in which minimizing power consumption
is not so important. Examples are emitter coupled logic, power switching, microwave power amplifiers, cell
phone amplifiers etc.
This course fills the semiconductor knowledge gap between the 1
st
year Semiconductor Devices course and
the 3
rd
year Advanced Electronics Devices course as it focuses more on bipolar devices than these other two
courses. Knowledge of bipolar transistors is also important for courses such as Microwave Technology, Radio
Frequency Electronics, Power Electronics, High Performance Analogue Electronics and Analogue Integrated
Circuits and Systems.
2
See the yr3 Advanced Electronics Devices course
4 EE2  Semiconductor Devices 2012
1 Semiconductors and junctions – a review
1.1 A semiconductor
• A semiconductor is a material that, at room temperature, has a conduction character that is lower than
metals but higher than insulators. This is because a semiconductor always has at least an intrinsic carrier
concentration: n
i
.
• There exist two types of charged carriers in a semiconductor that can both carry current: the electrons
with a negative charge and the holes with a positive charge:
eV C e q 1 10 6 . 1
19
t · t · t ·
−
. Only “free”
charges can carry currents. “Free” means not bonded to the atoms.
• The conduction in a semiconductor can be changed via doping. Doping is the introduction of foreign
atoms such as B or As in Si. Doping with donors: N
D
(donor concentration) gives an ntype semiconductor.
Thus the density of free electrons n
n
is larger than the density of free holes p
n
in an ntype material. Doping
with acceptors: N
A
(acceptor concentration) gives a ptype material. The density of free holes p
p
is larger than
the density of free electrons n
p
3
.
Important formulae that govern the conductivity of a semiconductor are:
The intrinsic carrier concentration:
i i
p n ·
(same density of free electrons and holes in an intrinsic semiconductor).
The extrinsic ntype semiconductor:
¹
¹
¹
'
¹
·
·
D
i
n
D n
N
n
p
N n
2
The extrinsic ptype semiconductor
¹
¹
¹
'
¹
·
·
A p
A
i
p
N p
N
n
n
2
With n, p respectively the electron and hole density ; N
D,A
respectively the donor and acceptor doping density
(concentration).
The total conductivity of a semiconductor is:
( )
p n tot
p n e µ µ σ + ·
With µ
n
, µ
p
the mobility of electrons, resp. holes. The value of the mobility is a function of the scattering
processes and the effective mass of the carriers:
m
qτ
µ ·
, with q the charge, τ the average time between
collisions (scattering), m the mass. Thus the hole and electron mobility are not necessarily the same. The
mobility gives an indication into how fast a device will work (when parasitic influences are negligible). The
mobility is the proportionality constant in the drift velocityelectric field relationship for small values of
electric field:
E v µ ·
, with v the drift velocity and E the electric field. Electrons and holes have opposite
velocities.
* Drift currents
In a homogeneously doped semiconductor or a semiconductor with a constant carrier density, applying an
electric field will cause drift currents to flow. Drift current can be carried by both electrons and holes.
3
The subscript indicates the material type. n is the symbol for electron concentration, p for the hole concentration. Units are cm
3
.
5 EE2  Semiconductor Devices 2012
Drift current in a semiconductor is given by:
( ) E p n eA AE I
p n tot tot
µ µ σ + · ·
With A the cross sectional area perpendicular to the current flow, E is the applied electric field. Remember
that the electric field can also be internal to the device structure.
* Diffusion currents
When carrier gradients exist in a semiconductor, diffusion currents will occur. Hole density gradients cause
hole diffusion currents, electron density gradients cause electron diffusion currents and gradients in both
carrier types will cause diffusion of both, creating a total diffusion current of
4
:
,
_
¸
¸
− ·
dx
dp
D
dx
dn
D eA I
p n tot
With D
n,p
the diffusion constant of electrons respectively holes, x is the direction of carrier propagation. The
Einstein equation gives the relationship between the diffusion constant and the mobility of the carrier:
e
kT D
·
µ
with, k the Boltzman constant, T the temperature in Kelvin.
* In the general case where both concentration gradients and electric fields are present the total current is the
sum of both drift and diffusion currents:
,
_
¸
¸
− + + ·
dx
dp
D
dx
dn
D E p E n eA I
p n p n tot
µ µ
This equation is normally referred to as the driftdiffusion equation of carriers and is the basic equation that
describes carrier movement in semiconductor devices.
E(x) J
n
drift
J
p
drift
n(x)
J
n
diff
p(x) J
p
diff
Figure 1: the direction of the current densities as a function of the electric field and the carrier gradient.
Note that the direction of the electron particle flux is opposite to the current flow. Particle flux is strictly
governed by electrostatics, the current is dependent on the sign of the charge q=±e.
The third important basic equation in semiconductor devices is the Poisson equation
5
:
( )
− +
− + − − · − ·
A D
N N n p
e x
dx
V d
ε ε
ρ ) (
2
2
With V the electrostatic potential, ρ the charge density as a function of x ; p & n the free hole, resp. electron
4
In one dimension.
5
In one dimension.
6 EE2  Semiconductor Devices 2012
density which are both a function of x as well as of V and
− +
A D
N N &
the concentration of ionised doping
atoms which are a function of x.
1.2 Energy band diagrams
The carriers in a semiconductor are quasi free because they are no longer covalently bonded to the atoms and
thus can move but at the same time they undergo interactions with the atoms in the regular lattice. As a
consequence of the mathematical description of the wave character of carriers in semiconductors, a region of
forbidden energies exists called the bandgap E
g
. In a perfect semiconductor there are no energy levels
available within the bandgap. This means that no electrons nor holes can reside in the energy region defined
by the bandgap.
Free electrons have energies in an allowed energy band higher in energy than the bandgap, the conduction
band E
c
, whilst free holes have energies in an allowed energy band lower in energy than the bandgap, the
valence band E
v
.
The amount of electrons available at each energy value is determined by the density of states g(E) and the
FermiDirac distribution function f(E) as illustrated in Figure 2. g(E) gives the distribution of energy levels
(states) as a function of energy. Note that g(E)=0 in the bandgap (no energy levels). f(E) gives the
probability of finding an electron at energy E. Thus 1f(E) gives the probability of finding a hole.
Figure 2: The energy band diagram with the position of the Fermi level E
F
as a function of the doping type
(left). The integration of the product of the density of states g(E) and the FermiDirac distribution function
f(E) gives the carrier distribution over the energy range above or below the bandgap (right).
7 EE2  Semiconductor Devices 2012
E
F
E
1/2 F(E) 1
E E
E
c
E
v
g(E)
carrier
concentration
a) intrinsic
electrons
holes
E
F
E
1/2
F(E)
1
E E
E
c
E
v
g(E)
carrier
concentration
b) ntype
electrons
holes
E
F
E
1/2 F(E) 1
E E
E
c
E
v
g(E)
carrier
concentration
c) ptype
electrons
holes
Extrinsic
semiconductor
Intrinsic
semiconductor
g(E) Density of states
F(E) FermiDirac distribution function
T>0K
Remember the relationship between the position of the Fermi level E
F
and the density of carriers in the
semiconductor (given by the FermiDirac or MaxwellBoltzman distribution functions):
( ) ( )
( ) ( )
kT
E E
c i
kT
E E
c
kT
E E
v i
kT
E E
v
c i c F
i v F v
e N n e N n
e N p e N p
− −
− −
· ·
· ·
and
and
with E
i
the intrinsic level
6
, N
c
the effective density of states in the conduction band and N
v
the effective
density of states in the valence band. The intrinsic level E
i
is an energy position within the forbidden gap that
is determined by the intrinsic carrier concentration.
* Fermi level constant
At equilibrium, when no external bias is applied, no discontinuity or gradient can arise in the position of the
Fermi level throughout the complete structure, otherwise a current (diffusion of high energy carriers to lower
energetic places) would flow. Thus:
0 ·
dx
dE
F
for V
ext
=0.
This gives the starting point for drawing the energy band diagram since E
c
E
F
and/or E
F
E
v
is known from
the doping concentration and E
g
is known and remains constant in a homojunction.
* Influence of the electric field on the band diagram
Applying an electric field will result in a discontinuity in the Fermi level (this discontinuity is directly
proportional to the applied electric field) and will result in band bending (see the tilt in potential energy E
c
in
Fig. 3).
Figure 3: Relationship between the electric field, the electrostatic potential and the potential energy.
dx
dE
e dx
dV
x E
c
1
) ( − ·
−
·
E(x): Electric field, V(x): electrostatic potential, E
c
(x): potential energy.
Electric field E(x) points “uphill”, electrons drift “downhill”. The reasoning is that electrons will flow from
v to v+ (electrostatic attraction). This means that electrons are losing potential energy since it is converted
in kinetic energy, thus E
c
at v needs to be higher than E
c
at v+. The electric field points from + to – and thus
up the potential hill.
6
The intrinsic level Ei is the Fermi level position for the case of an intrinsic semiconductor. The introduction of this level (not a
state!) makes some calculations more straightforward.
8 EE2  Semiconductor Devices 2012
e n
e r
g y
distance
E(x): electric field
E
c
E
v
v+ v
E
F
E
c
E
v
E
F+
Holes are flowing up the potential hill. You can check that this picture is completely consistent with the
carrier fluxes and current directions that have been discussed before. It is also comforting to see that these
directions are completely consistent with current directions used in the other courses such as Analysis of
Circuits and Analogue Electronics!
1.3. Junctions
* A junction is formed when 2 or more different materials are brought into close (atomic scale) contact.
Materials can include semiconductors of different doping type (homojunctions), different materials (e.g. Si
and SiGe) heterojunctions and metalsemiconductor junctions (e.g. Ohmic contact and Schottky contact).
* The work function φ
In order to draw an energy band diagram of a junction we need to draw the energy bands of each material
separately with respect of a local reference level that is defined for each material, the vacuum level E
vac
. The
position of the Fermi level with respect to the vacuum level, before contact of the material, is given by the
workfunction φ. Diffusion of electrons will appear from the material with the higher lying Fermi level to the
material with the lower lying Fermi level. This process will align the Fermi levels and create band bending in
the potential energy bands. The reason why band bending is created is due to the fact that no net current can
be flowing when no external bias is applied, thus the diffusion of carriers needs to be compensated. This is
done by drift of carriers which is caused by the buildup of an internal electric field that creates a drift current
opposite to the diffusion current and allows the system to reach equilibrium. Note that the workfunction of
the materials does not change when the junction is formed. As a result the internal electric field occurs in the
vacuum level across the junction – this is called the contact potential or builtin voltage.
* Band diagram of junctions
When two different conducting materials, initially uncharged, are brought into contact, the electrons close to
the junction redistribute themselves so that the Fermi energy is constant across the junction (in the
absence of an applied voltage).
Below we present different junctions with their energy band diagrams before and after contact, all without
external bias. Ensure that you understand how they are generated.
Metalsemiconductor junctions
Ohmic contact
metalsemiconductor junction
E
Fm
E
Fs
E
c
E
v
metal ptype
E
F


+
+
metal p
E
c
E
v
ohmic contact
9 EE2  Semiconductor Devices 2012
V
I
Figure 4: The Energy band diagram of a metal and a ptype semiconductor. Left: before contact, right: after
contact. For Ohmic contacts there should be an accumulation of majority carriers at the junction. Right:
The currentvoltage characteristic of the Ohmic contact is linear.
The currentvoltage characteristic of this junction is linear. The slope of the curve is proportional to the
conductivity of the semiconductor and the resistance of the Ohmic contact:
I R V
R R R
tot
tor semiconduc contact tot
·
+ · 2
Schottky contact
In fig. 5 a Schottky contact on an ntype semiconductor is given. Left (bottom) is the energy band diagram
when no voltage is applied. Notice the bands bend up (depletion of electrons closer to the junction). At the
right (top) a positive voltage is applied to the metal with respect to the semiconductor. This causes a
discontinuity of the Fermi level at the junction (E
F
goes down in energy due to positive voltage, and the
internal electric field across the semiconductor of the junction is reduced by the externally applied field and
thus the barrier lowers). The electrons in the ntype material try to cross the junction towards the metal but
are not allowed to go through the wide forbidden bandgap. Thus the only electrons that can contribute to the
current are the ones above the potential barrier and this is only a small amount thus the current flowing at
this voltage will be small.
Figure 5: The energy band diagram of a metal and a ntype semiconductor. Left: before contact, right: after
contact. For Schottky contacts there should be depletion of majority carriers at the junction. The current
voltage characteristics of the Schottky contact is exponential. A Schottky contact can be used as a rectifier.
It is a majority carrier device.
Semiconductorsemiconductor junction
pn junction and many more
10 EE2  Semiconductor Devices 2012
metal
l
semiconductor junction
E
Fm
E
F
s
E
c
E
v
metal ntype
E
F


+
+
metal n
E
c
E
v
Schottky
contact
I
V
E
cp
E
F
E
vp
E
g
ptype ntype
E
cn
E
F
E
vn
E
cp
E
F
E
vp
E
g
ptype ntype
E
cn
E
F
E
vn
eV
0
W
0
Figure 7: the energy band diagram of a pn junction. Left: before contact, right: after contact.
The work function difference between nSi and pSi causes a potential barrier between n and p Si when
forming the junction. Applying an external electric field will increase or decrease this potential barrier
respectively inhibiting and allowing current to flow.
* How to interpret an energy band diagram?
Energy band diagrams are great tools to predict the behaviour of the junctions under the influence of an
externally applied voltage. This is illustrated in fig. 8. The figure gives a Schottky contact on an ntype
semiconductor. The Schottky contact for the ntype semiconductor is characterised with a potential barrier
for electrons (depletion of minority carriers at the junction). When a positive voltage is applied to the metal
(V+) the Fermi level of the metal drops with a value equal to eV with respect to the Fermi level of the
semiconductor (which node we have grounded and thus it doesn’t change). This decreases the potential
barrier seen by the electrons in the semiconductor that are electrostatically attracted to the metal. However
not all electrons available in the semiconductor can flow through the junction to the metal due to remaining
potential barrier.
11 EE2  Semiconductor Devices 2012
Figure 8: Left bottom: energy band diagram of a metalsemiconductor contact under zero bias. Left top:
circuit connection. Right bottom: energy band diagram under bias. The density of electrons as a function of
energy is also given. Right top: circuit connection, electrons are flowing through the closed circuit.
The potential barrier (increase in E
c
towards the junction) blocks all carrier transport of carriers with
energies lower than the barrier. This is because these carriers would flow into the forbidden gap where there
are no energy levels to receive them. The carriers with energy higher than the barrier will be able to “flow”
across the junction and occupy free energy levels (states) at the other side of the junction. This will cause a
small current to flow. If the potential barrier is reduced more by applying a large V on the metal, then the
current will increase exponentially.
Thus the electrostatics imposed by the +V voltage attracts the electrons, however the quantum mechanics
presented by the energy band diagram shows that there is a potential barrier that blocks a certain amount of
available charges.
1.4 Conclusions
Semiconductors have two types of free carriers  electrons and holes  this gives the possibility for two types
of currents to flow in different junctions – drift and diffusion currents.
Energy band diagrams are a handy way to predict the expected magnitude of the current in junctions under
the influence of an external bias without the need for full scale device modelling.
12 EE2  Semiconductor Devices 2012
E
c
E
F
E
v
metal semiconductor
Energy
Density of electrons at each energy value. Lots of
electrons close to E
c
: n
2
> n
1
> n
3
but the density is
exponentially decreasing with increasing energy.
Potential barrier to electrons
intending to traverse to the
metal
E
Fm
V+
Forbidden band, no electrons can travel
through the wide forbidden band
distance
Only these electrons have
sufficient energy not to have to go
through the forbidden band. Few
electrons>small current.
E
c
n
1
n
2
n
3
Builtin voltage
Potential barrier
2. The pn diode
2.1 Review – the short pn diode
In EE1 the operation of the special case of a short diode was discussed. The definition of short diode is that
the width of the materials that make the pn junction, are smaller or equal to the diffusion length of the
minority carriers. As a consequence the number of minority carriers in the neutral region varies linearly,
with a slope determined by
1) the injection of minority carriers across the junction
2) the neutrality condition at the contacts (here we made the assumption that the contacts only impose the
bulk conditions on the carrier concentrations).
We found that the currents in a pn diode are governed by diffusion of minority carriers in the neutral regions
(the neutral regions are those where the sum of all charges is zero, thus outside the depletion region).
The assumptions that were made to derive the current are:
 the contacts are ideal and do not show any band bending
 the contacts can absorb and generate the necessary carriers to keep bulk condition carrier
concentrations fixed
 no voltage is dropped across the neutral regions
 the applied voltage drops across the depletion region only
 the amount of injected carriers across the junction due to the applied voltage is negligible
compared to the majority carrier concentration determined by the doping.
Under forward bias, electrons are injected from the nregion into the pregion and holes are injected from the
pregion into the nregion due to the fact that the externally applied potential lowers the energy barrier
between the n and p type region. This injection of carriers causes a minority carrier gradient of electrons in
the ptype region and a hole gradient in the ntype region. Gradients of carriers cause diffusion current.
Similar gradients occur in reverse bias but they are a lot smaller because they are minority carriers rather
than majority carriers that are injected across the barrier. The reverse bias leakage current is limited by the
amount of holes available and/or generated
7
in the ntype region and the amount of electrons available and/or
generated in the ptype region. Since under normal circumstances there is a limited number of minority
carriers available, the reverse leakage current is small.
* Band diagram of unbiased junction
7
See more on generation of carriers later in this chapter.
13 EE2  Semiconductor Devices 2012
E
cp
E
F
E
vp
E
g
ptype ntype
E
cn
E
F
E
vn
E
cp
E
F
E
vp
E
g
ptype ntype
E
cn
E
F
E
vn
eV
0
W
0
Figure 1: The energy band diagram of a pn junction. Left: before contact, right: after contact.
V
0
(also called V
bi
) is the builtin voltage, it represents the potential barrier that controls the amount of
carriers that diffuse. V
0
is generated by the internal electric field that is a consequence of carrier diffusion
upon contact.
,
_
¸
¸
·
2
0
ln
i
A D
n
N N
e
kT
V
* Space charge region or depletion region
p n p n



+
+
+
W
0
builtin
Efield
ve acceptor
ions
+ve donor
ions
depletion region
Figure 2: Left: p and n material before contact. Right: pn junction with 2 neutral regions (n & p) and the
space charge region. W
0
is the depletion width when no external bias is applied.
In the “depletion approximation” we assume that the depletion region is empty of moving carriers. It
contains 2 sides with fixed ionised charges of opposite charge at each side. The fixed ionised charges are the
ionised doping atoms (they have lost a hole or electron → N
A

resp. N
D
+
). The doping atoms are fixed in the
lattice and thus cannot move and as a consequence cannot carry current. This region is thus more resistive
than the neutral regions.
The charge separation (the ionised charges in the depletion region) is caused by the internal electric field.
The internal electric field is due to that builtin voltage V
bi
, the builtin voltage is a result of the contact
potential difference = work function difference between n and ptype Si.
* The pn junction under forward bias
14 EE2  Semiconductor Devices 2012
W
n
W
p
ptype ntype
depletion regions
neutral region neutral region
+ +
+ +
+ +
+ +
+ +
 
 
 
 
 
fixed donor ions
fixed acceptor ions
E
c
E
v
E
F
e (V  V)
eV
+ +
+
free holes
free electrons
0
V
Figure 3: Top: forward bias condition. Bottom: energy band diagram of the pn diode under forward bias
Forward bias lowers the potential barrier across the junction with a value proportional to the applied voltage,
therefore more carriers will have an energy higher than the barrier and cross the junction. Current is flowing
and is increasing exponentially.
* The pn junction under reverse bias
The reverse bias adds the absolute value of the applied voltage to the builtin voltage increasing the potential
barrier, therefore there are no carriers with sufficient energy to travel across the barrier. There is drift of
minority carriers across the junction. Since the number of minority carriers is low, the current is small and
determined by the minority carrier concentration rather than the electric field. Thus the offcurrent in a diode
is small and constant as a function of reverse bias voltage when no break down mechanisms occur.
E
c
E
v
E
F
eV + + +
free holes
free electrons
e (V + V)
0
W
n
W
p
ptype ntype
neutral region neutral region
+ + +
+ + +
+ + +
+ + +
+ + +
  
  
  
  
  
V
Figure 4: Top: reverse bias condition. Bottom: energy band diagram of the pn diode under reverse bias
15 EE2  Semiconductor Devices 2012
* Minority carrier distribution under bias and current.
It is important to remember that under certain bias conditions a certain amount of excess carriers is stored in
the n and p regions between the edge of the depletion region and the contact (neutral regions). These excess
carriers will cause the delays when switching pn diodes. More on this topic later.
Under forward bias the excess minority carrier concentration at the edges of the depletion width are given as
a function of the bulk minority carrier concentration and the applied voltage. The excess of carriers is
defined as the surplus (positive or negative) of minority carriers with respect to the initial minority carrier
concentration that is determined by the bulk material. The excess carrier concentration at the edge of the
depletion width (thus at x=w
n
or –w
p
) is given by:
edge region depletion at region type  n in ion concentrat hole excess
edge region depletion at region type  p in ion concentrat electron excess
0
'
0
'
n n n
p p p
p p p
n n n
− · ∆
− · ∆
With the injected concentrations given by:
,
_
¸
¸
· ·
kT
eV
n
n
p
p
e
p
p
n
n
0
'
0
'
(1)
with n’
p
and p’
n
respectively the minority electron concentration in the pregion at the depletion region edge
and the minority hole concentration in the nregion at the depletion region edge (injected minority carrier
concentration). n
p0
and p
n0
are the intrinsic minority carrier concentrations.
To derive the expression for the current, it is best to refer to the graphical picture of the device under bias as
given in fig. 5. Note that looking at forward bias only, is sufficient to derive the expression of the current in
both bias regions.
Figure 5: Top: Short pn diode under forward bias with exaggerated depletion width. Bottom: excess
minority carrier distribution of short diode. Shaded regions represent the amount of stored excess minority
carrier charge
8
. Note that at the contacts the excess carrier concentration is zero. This condition is imposed
by the contacts.
8
Note that the length of the ntype region is given as Lp. This seems contradictory but isn’t from a physics point of view.
Remember than in the ntype region a hole diffusion current exists. The boundary for the definition of short diode is that the
length of the region is exactly equal to the diffusion length of the minority carriers which is Lp. The choice of taking the
maximum value of the lengths for deriving the short diode current is that this gives exactly the same expression for the current as
in the case the short diode approximation cannot be applied (see next). It is a bit cheeky, but now you know why.
16 EE2  Semiconductor Devices 2012
x
w
p
w
n
0
n
p
p
n
x
n
p0
x
∆
n
p ∆p
n
p
n0
0
0
Short
diode
p’
n
n’
p
L
n
L
p
In a short diode (the lengths of each neutral region is smaller or equal to the minority carrier diffusion
length) the minority carrier concentration is decreasing linearly in each region. The excess carrier
concentration is given by the shaded triangles in the neutral regions (Fig.5). The density of minority carriers
is imposed by the boundary conditions: eq. (1) and bulk conditions kept at Ohmic contacts. Since gradients
in carrier concentration occur due to the injection process across the junction, diffusion of minority carriers
will happen.
Calculation of the diode currents in a short diode is easy. Remember the expression for the diffusion current:
( )
¹
¹
¹
¹
¹
¹
¹
¹
¹
¹
¹
¹
¹
¹
¹
¹
¹
¹
¹
¹
¹
'
¹
,
_
¸
¸
−
,
_
¸
¸
−
,
_
¸
¸
+
,
_
¸
¸
−
+
−
,
_
¸
¸
−
+
−
,
_
¸
¸
−
−
−
− −
−
,
_
¸
¸
−
·
,
_
¸
¸
,
_
¸
¸
,
_
¸
¸
,
_
¸
¸
1
1
1 1
'
'
0
'
0
'
0 0
0 0
0
0
0
0
kT
eV
s
kT
eV
p
p n
n
n p
p
kT
eV
p n
n
kT
eV
n p
p
n n
p
n
p p
n
p
n n
p
n
p p
n
n
p
p
n
tot
e I
e
L
D p
L
D n
eA
L
e
D p
L
e
D n eA
L
p p
D
L
n n
D eA
L
p p
D
L
n n
D eA
dx
dp
D
dx
dn
D eA
I
An interesting problem to think about is: what happens when one or both of the materials forming the pn
diode are shorter than the diffusion length (L
n
and L
p
) of the minority carriers (but longer than the depletion
width in each region)? Thus do the derivation of the short diode current for a length x
n
<L
p
for the nregion
and x
p
<L
n
for the pregion and compare the result taking into account the given inequalities.
Remember that there is also a majority carrier flow as the majority carriers injected across the junction have
to be resupplied
9
. Luckily the rule that the total current is constant through the device makes calculation of
the minority carrier diffusion current in each region of the pn junction sufficient for deriving the total current
in the pn junction. The magnitude of the diffusion current of the electrons in the ptype region is equal to the
magnitude of the electron drift current in the ntype region, the latter current is the resupply of the injected
carriers across the junction. This is similar for holes. This is illustrated in the picture below (fig. 6).
9
See first year course notes.
17 EE2  Semiconductor Devices 2012
Figure 6: The contribution of the minority and majority carrier current to the total current through the pn
junction. Majority carriers injected across the junction are resupplied via drift through the contact and
diffuse to the other contact through the minority carrier gradient region. The minority carrier gradients
determine the total current completely. The current remains constant in the space charge region because we
assume at the moment that there is no recombination no generation of carriers.
2.2 Generation and recombination – long pn diodes
In real diodes, the length of the materials forming the junction is not necessarily smaller or equal to the
diffusion length of the minority carriers. In the case of long diodes recombination of minority carriers will
happen whilst they are diffusion to the contact. As a result, the number of minority carriers varies non
linearly in the neutral regions of the device. In what follows we will study its impact on the device
characteristics.
* Generation – recombination
We have seen that due to thermal energy electronhole pairs are created, thus electrons and holes are freed
from the atoms and become free carriers that can be involved in the conduction processes. This process is
called generation of carriers and is happening continuously as long at the temperature is sufficiently high to
break the valence electronatom bonds. Of course other forms of energy can do the same. An example is
light: shining light on a semiconductor, with the correct energy, larger than the energy gap, can also release
electronhole pairs.
The opposite process is called recombination. In this case an electron and hole will recombine and thus get
fixed to the atom and become unavailable for conduction. This is a reduction of the number of free carriers.
As with generation, recombination is also continuously happening. Recombination can happen in two ways.
The first is direct recombination of electron and holes which means that excess electrons in the conduction
band recombine with holes in the valence band. Energy is lost by the electron and is given off as photons
(light). The second type is indirect recombination. In this case there are impurity centres or lattice defects in
the semiconductors, called traps – these will always be available as a 100% perfect semiconductor does not
18 EE2  Semiconductor Devices 2012
junction
Depletion region
current
p
+
region nregion
contact
contact
I
n
I
p
I
tot
Minority carrier diffusion
currents
Majority carrier drift
currents
Remain constant in
space charge region.
exist. The traps will cause some discrete energy levels to exist within the bandgap of the semiconductor
which can have empty states to which the electrons from the conduction band can fall in order to cross the
bandgap in steps. In this case, the energy released by the electrons causes heating of the lattice (Heating
causes lattice vibrations. These vibrations can be seen as particles called phonons with which the carriers in
the lattice can interact – scatter. They can also be regarded as waves with certain energy and wavelength).
The characteristic timescale involved in the recombination processes is called the carrier lifetime τ
n,p
. The
characteristic length scales are the diffusion lengths L
n,p
. These parameters give the statistically average time,
respectively distance that a carrier travels before it recombines. This is a statistical average, thus means that
not all carriers will recombine within that time scale, but the majority will such that at x=L
n,p
the number of
carrier have been reduced by a factor Exp(1). The values are dependent on the material, the quality of the
material, the doping density and the type of charged carrier. The relationship between the carrier lifetime
and diffusion length is given by the diffusion constant following:
p n p n p n
D L
, , ,
τ × ·
The intrinsic number of free electrons and holes is the same as carrier generation and recombination is in
equilibrium. In a pn diode under forward bias an excess minority carrier density exists in the neutral regions
of the device. When the voltage across the junction is not too high we can assume that this excess is a lot
smaller than the number of majority carriers in that region. It is the gradient of the excess minority carriers
that is significant. Recombination will happen between the majority carriers and the excess minority carriers.
Since the density of the excess carriers is small compared to the density of the majority carriers,
recombination processes can have a significant impact on the number of excess carriers, whilst being
insignificant for the majority carriers in comparison with the amount available. Recombination becomes
important for devices where the neutral regions are larger than the diffusion length
10
. If we put the pn diode
under reverse bias then there will be a reduction of minority carriers in the neighbourhood of the depletion
region edges (junction). In this case generation of minority carriers is larger in order to maintain equilibrium
conditions.
Thus the rate of recombination and generation will adapt itself to the availability of excess minority carriers
in the “neutral regions”
11
. When there is an excess of minority carriers, then the recombination rate will
increase proportional to this excess in order to drive the semiconductor back to equilibrium. Thus the
generation rate of minority carriers will then be smaller than the recombination rate. However when there is
a shortage of minority carriers then the generation rate increases to try to bring the minority carrier
concentration back to equilibrium conditions. In this case the generation rate of minority carriers is larger
than the recombination rate. Of course when minority carriers recombine they do so with majority carriers.
As a consequence the majority carrier recombination or generation rate is equal to that of the minority
carriers. We will assume that the injected minority carrier concentration is sufficiently small compared to the
majority carrier concentration such that the variation of majority carriers can be neglected.
12
10
It is indeed an approximation, although acceptable, in the short diode to assume no carriers recombine within the neutral region
as the diffusion length is a statistical average. Thus some do recombine but we neglect this.
11
What we call neutral is actually quasi neutral under the given assumption, that’s why it is written between quotes.
12
This assumption will actually break down the rule on charge neutrality in the “neutral region”. This is of course not completely
correct as charge neutrality is a fundamental rule. But the assumption will allow us to derive analytic expressions that still
describe the conduction through the device with reasonable accuracy.
19 EE2  Semiconductor Devices 2012
The minority carrier recombination rate in the “neutral region” can be described to acceptable approximation
by the simple equations:
p
n
p
n
p
n
p
U
n
U
τ
δ
τ
δ
·
·
(2)
Recombination in a depletion region or in those cases where the minority carrier concentration is of the same
order of magnitude as the majority carrier concentration have to be described by different equations than
those given above.
In order to calculate currents in the long diode, we need to know the variation of the minority carrier excess
in the neutral region. The first step is thus to derive the expression of this variation in the case recombination
happens
13
. In order to do that we take a volume of semiconductor material through which a hole diffusion
current is flowing as given in fig. 7. We look at the variation of the density of holes, as a function of time
and place, over an infinitesimal distance ∆x. This equation will give us the continuity equation for holes. It
gives the rate of change of the hole density while a current is flowing. It is also this equation that will be
important to analyse the transient processes in bipolar devices under large signal switching conditions.
Figure 7: Hole current entering and leaving a volume ∆x A.
p
n
p p
x x x
n
t x p
x
x x J x J
e t
t x p
τ
δ ) , (
) ( ) (
1 ) , (
−
∆
∆ + −
·
∂
∂
∆ + →
(3)
In words this equation means:
The rate of variation of the hole concentration = increase of the hole concentration in volume ∆x ×A per unit
time  loss of holes due to recombination.
The term δp
n
is the excess minority carrier concentration. Excess means the density of carriers larger than
the minority carrier concentration defined by the doping of the bulk material only. The second term is the
carrier recombination term given in eq. (2). For ∆x→0 the equation (3) becomes a differential equation:
p
p t x p
x
t x J
e t
t x p
t
t x p
τ
δ δ ) , (
) , (
1 ) , ( ) , (
−
∂
∂
−
·
∂
∂
·
∂
∂
(4a)
Similar for electrons:
n
n
t x n
x
t x J
e t
t x n
t
t x n
τ
δ δ ) , ( ) , ( 1 ) , ( ) , (
−
∂
∂
·
∂
∂
·
∂
∂
(4b)
13
Remember, without recombination it is linear.
20 EE2  Semiconductor Devices 2012
propagation direction, x
x x+∆ x
J
p
(x) J
p
(x+∆x)
A
These equations are referred to as the continuity equations for the holes resp. the electrons.
If we can assume that drift is negligible
14
then the current is solely diffusion and the current densities can be
written as:
x
p
eD J
x
n
eD J
n
p p
p
n n
∂
∂
− ·
∂
∂
·
δ
δ
(5)
Substituting (5) in (4)
15
gives the continuity equations of minority carriers taking recombination into
account:
p
p
n
n
p
x
p
D
t
p
n
x
n
D
t
n
τ
δ δ δ
τ
δ δ δ
−
∂
∂
·
∂
∂
−
∂
∂
·
∂
∂
2
2
2
2
(6)
This equation is of utmost importance for analysing the switching characteristics of bipolar devices.
* minority carrier variations and currents
In case we have steady state carrier injection as e.g. across a forward biased pn diode in DC, the time
derivative is zero and in steady state the continuity equations become:
2 2
2
2 2
2
p p p
n n n
L
p
D
p
x
p
L
n
D
n
x
n
δ
τ
δ δ
δ
τ
δ δ
· ·
∂
∂
· ·
∂
∂
(7)
Solving this equation with boundary conditions:
0 & 0 , @
& 0 @
· · ·
∆ · ∆ · ·
p n X X x
p p n n x
n p
δ δ
δ δ
(8)
These boundary conditions mean that x=0 is the point of steady state carrier injection (depletion region edge)
due to the applied DC voltage across the junction. The second boundary condition gives that at the contacts
the excess carrier concentration needs to be zero.
The solution of (7) for holes, which is a second order differential equation, can be written as the sum of two
exponentials:
,
_
¸
¸
− +
,
_
¸
¸
·
p p
L
x
C
L
x
C x p exp exp ) (
2 1
δ
or using a hyperbolic sine function:
,
_
¸
¸
+ ·
2 1
sinh ) ( C
L
x
C x p
p
δ
with C
1
and C
2
integration constants that must be determined by the boundary conditions (8). Using the sinh
function as possible solution, we find the expression
16
of the excess hole concentration variation as a
function of position:
14
This is a valid assumption as we are dealing with minority carriers and we have assumed that the injection rate is small. So the
absolute number of minority carriers is small and as a consequence the drift minority carrier current will be negligible.
15
Note that the subscript n and p are omitted. Remember that these equations are for the minority carriers!
16
Mathematica
TM
was used for this extraction.
21 EE2  Semiconductor Devices 2012
,
_
¸
¸
−
,
_
¸
¸
−
∆
·
p
n
p
n
L
X x
L
X
p
x p sinh
sinh
) ( δ
(9)
This is a rather complex equation, useful for numerical simulations only but not for pen and paper work.
Therefore we want to simplify this equation for two special cases: the short diode approximation and the
long diode approximation. For the short diode approximation we take the limit of X
n
<< L
p
for eq. (9) and for
the long diode approximation we take the limit of X
n
. → ∞
In the following we will evaluate what the errors are that we make by taking these approximations. For this
we need to calculate the current based on (9) and that based on the limits of short and long diode
approximations.
In order to calculate the current we have to take the first derivative of (9). This gives for the hole current:
,
_
¸
¸ ∆
·
p
n
p
p
p
L
X
L
p A eD
I coth
(10)
The following figures in Fig. 8 give the difference in excess minority carrier concentrations between the
short, long and correct diode descriptions.
22 EE2  Semiconductor Devices 2012
Figure 8: The excess minority carrier hole concentration in the nlayer under normalised injection
conditions for the different solutions of the continuity equation and for different relative differences between
contact position relative to the minority carrier diffusion length. The minority carrier diffusion length is
L
p
=200 nm, the distance to the contact varies: X
n
= (a) 20, (b) 200, (c) 400 and (d) 1000 nm.
For short X
n
(a), the short diode approximation describes the excess carrier concentration to excellent
accuracy. For long X
n
(d) the long diode approximation is excellent. In the other two situations some error is
made. When X
n
=L
p
, the excess carrier concentration is best described by the short diode approximation
(smallest difference in area underneath the curves and correct contact boundary condition). When X
n
>L
p
but
of the same order of magnitude, the linear approximation becomes less accurate and the exponential
equation more accurate. The error in the long diode approximation is in the nonzero excess at the contacts.
23 EE2  Semiconductor Devices 2012
(a)
(b)
(c)
(d)
N
o
r
m
a
l
i
s
e
d
m
i
n
o
r
i
t
y
c
a
r
r
i
e
r
c
o
n
c
e
n
t
r
a
t
i
o
n
Distance (nm)
N
o
r
m
a
l
i
s
e
d
m
i
n
o
r
i
t
y
c
a
r
r
i
e
r
c
o
n
c
e
n
t
r
a
t
i
o
n
Distance (nm)
N
o
r
m
a
l
i
s
e
d
m
i
n
o
r
i
t
y
c
a
r
r
i
e
r
c
o
n
c
e
n
t
r
a
t
i
o
n
Distance (nm)
N
o
r
m
a
l
i
s
e
d
m
i
n
o
r
i
t
y
c
a
r
r
i
e
r
c
o
n
c
e
n
t
r
a
t
i
o
n
Distance (nm)
Correct solution
Short diode approximation
Long diode approximation
When looking at the currents, we can conclude why the use of the approximations is acceptable in different
regions. In figure 9, the ratio of the correct current to the approximated current is given as a function of the
ratio of the lengths.
Figure 9: The ratio of correct current to approximated current. Blue: long diode approximation, pink: short
diode approximation.
Interesting observations are:
1. the error of using the short diode approximation becomes negligible small when X
n
<<L
p
. Obviously for
X
n
<L
p
the short diode approximation is best.
2. for X
n
=L
p
the error using the short or long diode approximation is exactly the same
17
. Thus we use the
short as it is easier.
3. for X
n
>L
p
the error made by using the long diode approximation decreases quickly and it is better to use
the long rather than the short diode approximation. We see that if the length of the diode layer is more than
twice that of the minority carrier diffusion length, the error becomes negligible.
* Simplified solution of eq. (7)
1. Short diode approximation
In this case, the Ohmic contacts are at a distance shorter than the minority carrier diffusion length. Thus X
p
<
L
n
and X
n
< L
p
, where X
p
is the position of the contact of the player and X
n
is the position of the contact in
the nlayer. In this case we neglect any recombination by setting τ
n,p
= in eq. (7). Then under the boundary ∞
condition (8), the excess minority carrier concentration varies linearly. This is what is referred to as the
SHORT diode approximation.
17
This is obvious from the equations derived in this case for the short and long diode approximations.
24 EE2  Semiconductor Devices 2012
0
1
2
3
4
5
6
7
8
9
10
0 1 2 3 4 5
X
n
/L
p
I
r
e
a
l
/
I
a
p
p
r
o
x
Ireal/Iexp
Ireal/Ilin
2. Long diode approximation
In this case, the Ohmic contacts are at a distance larger than the minority carrier diffusion length. Thus X
p
>>
L
n
and X
n
>> L
p
. In this case we assume that the contacts appear at a distance equal to infinity: X
p
= X
n
= . ∞
18
In that case the solution to (7) is an exponential. This is called the LONG diode approximation.
Solving (7) in the long diode approximation we impose that the excess minority carriers have completely
disappeared by recombination. The solution for the excess carrier concentration is then:
,
_
¸
¸
−
,
_
¸
¸
−
∆ ·
∆ ·
p
n
L
x
L
x
e p p
e n n
δ
δ
excess minority carrier concentration
19
(11)
With τ D L · the diffusion length, which is the average distance a carrier diffuses (travels) before
recombining. τ is the average time between scattering
20
events. Thus in the case of diffusion with
recombination the excess carrier concentration is varying exponentially in a long diode in contrast to the
short diode where the excess carrier concentration is varying linearly. This is illustrated in fig. 10.
Figure 10: the variation of the excess carrier hole concentration as a function of distance due to
recombination (full line).
n
0
(p
0
) is the carrier concentration determined by the doping of the bulk material. ∆n (∆p) is the excess carrier
concentration at x=0 resulting from the carrier injection process. x=0 is the edge of the depletion region. n δ
( p) is the xdependent excess carrier concentration. Note that we are talking about minority carriers! δ
As in the short diode, the current is still determined by the diffusion of the excess minority carriers in the
neutral region. The difference is that now this minority carrier excess is changing exponentially. As a
consequence the minority carrier diffusion current calculated for holes is:
,
_
¸
¸
−
∆ ·
p
L
x
p
p
p
e p
L
D
q x J ) (
(12)
Thus this current is decreasing for positive x (see fig. 10) because holes are disappearing by recombination!
But the total current needs to be constant. This can be obtained by remembering that total current consists of
18
Note that we are ignoring the signs.
19
Note that the variations of both minority carrier concentrations are calculated in the positive xaxis. Signs will change for pn
junctions.
20
Scattering is caused by interaction of the carriers with Si atoms, impurities (doping and others) and crystal imperfections.
25 EE2  Semiconductor Devices 2012
p(x)= p
0
+ ∆p e
(x/Lp)
∆p
p
0
x
X
n
L
p
Excess carrier
concentration is forced
to zero at contact.
hole and electron movement. Since we have neglected the variation of majority carriers, there must be a drift
current of majority carriers that is resupplying the recombined carriers and resupplying the carriers that are
injected across the junction. The sum of this majority carrier drift current and minority carrier diffusion
current is constant. Thus we see that if minority carrier diffusion current is decreasing when going away
from the point of carrier injection, the majority carrier drift current in decreasing when going towards to
point of injection. The sum of the two is constant: I
tot
. As a consequence it is the diffusion of minority
carriers that limits the total current, therefore the total current must be determined by the maximum electron
and hole diffusion currents that can happen under the given injection conditions (determined by the applied
voltage) The maximum diffusion current that can flow can be calculated at the point of injection at both
sides of the junction (edge of the depletion width in the pn diode) as there the gradient of the exponentials is
maximum. This gives the first approach of finding the expression of the current in a long pn diode.
Figure 11: Top: pn diode under forward bias with exaggerated depletion width. Bottom: excess minority
carrier distribution of long diode. Shaded regions represent the amount of stored excess charge.
Calculation of the diode currents in a long diode is based on calculating the maximum diffusion current at
the edge of the depletion region width, see fig. 11. This means that we determine the excess carrier gradient
at x=0 in each region:
( )
¹
¹
¹
¹
¹
¹
¹
¹
¹
¹
¹
¹
¹
¹
¹
¹
¹
'
¹
,
_
¸
¸
−
+
−
,
_
¸
¸
∆
+
∆
,
_
¸
¸
∆
−
∆ −
,
_
¸
¸
−
,
_
¸
¸
−
·
·
−
·
−
·
·
·
·
p
n n
p
n
p p
n
p
n
p
n
p
n
x
L x
n
p
x
L x
p
n
x
n
p
x
p
n
x
n
p
x
p
n
tot
L
p p
D
L
n n
D eA
L
p
D
L
n
D eA
dx
e p d
D
dx
e n d
D eA
dx
p d
D
dx
n d
D eA
dx
dp
D
dx
dn
D eA
I
p n
0
0
0
/
0
/
0
0
0
0
'
'
δ
δ
26 EE2  Semiconductor Devices 2012
x
w
p
w
n
0
n
p
p
n
x
n
p0
x
∆
n
p ∆p
n
p
n0
0 0
Long
diode
p’
n
n’
p
¹
¹
¹
¹
¹
¹
¹
¹
¹
¹
¹
¹
¹
'
¹
,
_
¸
¸
−
,
_
¸
¸
−
,
_
¸
¸
+
,
_
¸
¸
−
+
−
·
,
_
¸
¸
,
_
¸
¸
,
_
¸
¸
,
_
¸
¸
1
1
1 1
0 0
0 0
kT
eV
s
kT
eV
p
p n
n
n p
p
kT
eV
p n
n
kT
eV
n p
tot
e I
e
L
D p
L
D n
eA
L
e
D p
L
e
D n eA
I
Note that the result we have obtained for the long pn diode current is the same as the short diode. This is
only the case when for the short diode we take the lengths of the neutral regions to be exactly the same as the
diffusion lengths of the minority carriers. For shorter diodes, the values of L
n
and L
p
will have to be changed
by the actual widths of the neutral regions and the current will thus increase in shorter diodes.
Figure 12 gives the electron and hole components of the currents in a long pn diode. Each component is not
constant throughout the device but the sum of electron and hole current is constant. The majority carrier
component occurs as a result of recombination between minority and majority carriers, majority carriers
need to be resupplied, which causes the current. Compare this to the short diode where all current are
constant throughout the device. Thus our initial assumptions that there is no electric field
21
in the neutral
regions is not quite correct otherwise there would be no drift current for the resupply of carriers. The re
supply can be allocated to drift as the gradients of the majority carriers are so small that it would not be
sufficient for resupply. The electric field can be very very small as long as the number of carriers is large
(which is the case for majority carriers) to make a sufficiently large drift current. Thus although there is a
small electric field across the neutral regions it still is negligible compared to the electric field dropped
across the depletion region.
Figure 12: The variation of the magnitude of the hole and electron current in a long diode. Note I
p
is a
minority carrier diffusion current in the nregion and a majority carrier drift current in the pregion (similar
for I
n
).
There is another way to calculate the currents in the pn diode that will come in handy when discussing the
21
Nor a gradient in majority carriers
27 EE2  Semiconductor Devices 2012
p n
I
tot
I
p
I
n
I
n
I
p
I
tot
=I
n
+ I
p
switching behaviour of the pn diodes. To that aim we quickly look at this second method for calculating the
diffusion currents in a diode.
With reference to fig. 11 note that there is excess minority carrier charge in the neutral region (this is
indicated by the shaded regions). We call Q
n,p
the excess minority carrier electron, hole concentration in resp
the p, n type region. The density of the excess charge is given by the shaded area underneath the curves for
δn and δp. To calculate the excess charge we integrate the excess minority carrier functions:
∫
∫
∞
∞ −
·
− ·
0
0
dx p eA Q
dx n eA Q
n p
p n
δ
δ
(13)
Note that the boundary conditions chosen for this calculation are taken to be infinite. These boundary
conditions represent the position of the contacts under the long channel approximation
22
. Under the
assumption that the exponential variation of the excess carrier concentration is the solution of the steady
state continuity equation we have to make sure that the excess carrier concentration at the contacts is zero.
We know that current is a variation of charge as a function of time. In the case where recombination of the
excess carriers occurs we know that this recombined charge needs to be resupplied. Every τ
n,p
the charge
Q
n,p
disappears via recombination and this charge needs to be resupplied in the same time scale in order to
maintain steady state. The current that resupplies these recombined minority carriers is given by:
p
n p
p
p
p
n
p n
n
n
n
p eAL Q
I
n eAL
Q
I
τ τ
τ τ
∆
· ·
∆
· ·
(14)
Thus the current can be expressed as the stored minority charge in the neutral regions divided by the lifetime
of the minority carriers in this region. Filling in the expressions for the excess carriers and the diffusion
constants will give the same current equations as extracted before for the long diode.
2.4 Nonidealities in pn diodes
When measuring a Si pn diode and plotting the current on a log scale we get a graph as in fig. 13.
Region a
The assumption in the ideal pn diode is that no recombination happens in the depletion width, thus carriers
travel through unhindered. This is not completely correct, especially in pin
23
diodes or diodes with long
depletion widths. If the depletion width is larger than the carrier diffusion lengths, recombination of carriers
in transit will occur. This will affect the current at low current densities (small number of carriers in
transfer). More carriers need to cross in order to obey the relationship for injected carrier concentration as a
function of applied voltage.
Region b
This region behaves as ideal diffusion and happens for intermediate voltages.
22
This ensures a zero excess carrier concentration at the contact!
23
ptype region + intrinsic region + ntype region
28 EE2  Semiconductor Devices 2012
Region c
This is the region of high level injection. A huge amount of carriers is injected across the junction and makes
the number of minority carriers at the edge of the depletion region of the same order of magnitude as the
majority carriers. This brings changes to the majority carrier density which we have assumed to be constant
in the ideal case. In the high carrier injection regime the relationship for the injected carrier concentration
that we have derived in EE1 breaks down. Its effect is a lowering of the total current flowing.
Region d
At still higher voltages or for large ohmic contact resistances a large amount of the applied voltage is
dropped across the ohmic contact rather than across the depletion region and thus the intrinsic diode feels
less voltage applied, keeping the current lower than expected.
Figure 13: The difference between an ideal and a real IV curve of a Si pn diode.
Although all these effects can be calculated, there is no easy solution. Therefore an ideality factor n is
introduced in the formula for the current in a pn junction that takes into account these nonideal effects. Note
that the ideality factor n is called the emission coefficient in SPICE.
The expression for the current then becomes:
,
_
¸
¸
− ·
,
_
¸
¸
1
nkT
eV
s tot
e I I
With 1 ≤ n 2. n=1 in the ideal region and goes towards 2 in the recombination and high level injection ≤
region.
2.5 Switching delays in pndiodes
The switching characteristics of pn diodes are all governed by EXCESS MINORITY CARRIERS.
Under bias there exists an excess minority carrier concentration inside the neutral regions close to the
depletion region. The amount of which is a function of the applied bias – the larger the current that is
flowing the larger the excess charge. In order to change the current, the excess charge stored should change
before the voltage across the diode junction can change. This is inherently a capacitive effect.
We can look at two ways of switching. First (fig.15), one can switch the diode from on (V
applied
>0V, forward
bias) to off (V
applied
=0V, by opening the switch such that I=0A). We will see that removal of excess carriers,
when the diode switches from forward bias to open circuit, occurs via recombination only. This is thus
29 EE2  Semiconductor Devices 2012
Ln(I)
V
a)
b)
c)
ideal
real
d)
governed by the carrier lifetimes. The carrier lifetime is the average time an electron (resp. hole) spends in
the material before it recombines with a hole (resp. electron). τ
n
and τ
p
are the minority carrier life time of
electrons and holes respectively, thus it will take a time equal to τ
n
and τ
p
to reduce the excess carrier
concentration by a factor e.
Fig. 15 illustrates a switching experiment of a p
+
n diode (current mainly governed by holes) from forward to
zero bias via a switch that opens at t=0. Although the current drops immediately to zero, the voltage, v(t)
across the diode remains until all excess charge is removed. That means that there will be a voltage across
the load R
L
until excess charge removal is complete.
I Q
p
(t) δ p
I I τ
p
t t x
n
Current through diode. Decay of stored charge. Change of excess
in function of position.
Figure 15: Top: switching circuit. Bottom: Left: upon switchoff the current through the diode drops to zero
but the excess charge remains and decays exponentially via recombination only (middle). The state of the
diode cannot change before the charge is removed. Right: dashed lines give gradient of excess charge at the
edge of the depletion region. When the current is 0 the gradient is 0.
Repeating equation (4a):
p
p t x p
x
t x J
e t
t x p
τ
δ δ ) , (
) , (
1 ) , (
−
∂
∂
−
·
∂
∂
Reorganising the terms, multiplying by cross sectional area A and integrating over the neutral region:
A
t
t x p e
A
t x p e
A
x
t x J
contact contact
p
contact
p
∫ ∫ ∫
∂
∂
+ ·
∂
∂ −
0 0 0
) , ( ) , (
) , (
δ
τ
δ
Remember that the diffusion current is zero at the contact and maximum at the edge of the depletion region
(x=0) and that the integral of the excess minority carrier concentration in the neutral region multiplied by
unit charge e and cross sectional area A gives the total stored charge Q
p
in that region (see eq. 13).
This gives the expression for the time variation of the current as a function of the stored charge in the neutral
30 EE2  Semiconductor Devices 2012
R
L
e(t)
+

i(t)
v(t)
v
L
(t)
t=0
e(t)
region:
dt
t dQ t Q
t i
p
p
p
p
) ( ) (
) ( + ·
τ
(15)
Eq. (15) states that the instantaneous current in a switching device i
p
(t) is equal to the resupply of the excess
carrier concentration that is disappearing as a result of recombination (this is the DC component of the
current – 1
st
term RHS) plus the AC component that allows a buildup or depletion of charge as a function of
time (2
nd
term RHS).
In the special case of switching the p
+
n diode from ON to zero, a solution to equation (15) can be easily
found. For t≤0 the DC forward bias imposes a current I
0
through the device. Thus at t=0
24
the excess
minority carrier charge is given by: Q
p
=I
0
τ
p
. At t=0+
25
the current drops to zero (open circuit) and remains
zero, the charge cannot follow immediately. These form the boundary conditions to solve (15)
26
.
Thus (15) becomes:
dt
t dQ t Q
p
p
p
) ( ) (
− ·
τ
Integration over time gives:
( )
p
t
p
p
p
t
p
t
p
p
e C t Q
c
t
t Q
dt
t Q
t dQ
τ
τ
τ
/
0 0
) (
) ( ln
) (
) (
−
·
+ − ·
− ·
∫ ∫
With c and C integration constants.
Since Q
p
(t=0+)= I
0
τ
p
the integration constant C= I
0
τ
p
.
Thus the variation of the excess minority carrier charge as a function of time is given by:
p
t
p p
e I t Q
τ
τ
/
0
) (
−
·
(16)
The proof that the voltage remains across the diode goes as follows. We need to find the relationship
between the voltage across the diode and the minority carrier excess charge. This relationship, we all know
from the injection of carriers relationship (1):
,
_
¸
¸
−
,
_
¸
¸
· − · ∆ 1
) ( v
exp
0 0
'
kT
t e
p p p p
n n n n
(17)
With v(t) the voltage across the diode. Thus as long as ∆ p
n
is nonzero, there exists a voltage v(t) across the
diode.
If we ignore the distortion of the exponential variation of the excess charge at the depletion region edge
27
, we
24
An infinitesimal amount of time before switching.
25
An infinitesimal amount of time after switching.
26
Note that these switching equations can also be solved with the methods introduced in the Signals and Linear Systems course
this year.
27
Since the current is zero, the gradient at the depletion region edge of the minority carrier density needs to be zero too! So there
the line for the excess carrier concentration is horizontal and not exponential.
31 EE2  Semiconductor Devices 2012
I
r
→ gradient>0
e(t)
can still use the pure exponential relationship of the excess carrier concentration as given by eq. (11). The
stored charge at any time is then given by (13):
) (t p eAL Q
n p p
∆ ·
(18)
Substituting eqs. (16) in (18) gives
) (t p
n
∆
as an exponential function of time. Substituting this in (17) and
extracting v(t) gives:
,
_
¸
¸
− ·
−
1 ln ) ( v
/ 0
p
t
o n p
p
e
p eAL
I
e
kT
t
τ
τ
(19)
In order to improve switching speed in this situation, recombination centres are added to decreases the
carrier life time, τ
p
. On the other hand one can also make a p
+
n diode with a short p
+
region, smaller than the
recombination length in order to limit the excess carrier buildup.
Thus:
• the excess charge dies out exponentially via recombination as a function of time
• the voltage across the diode junction cannot change instantaneously but has to wait till the excess
charge has disappeared in order to establish the correct depletion layer for a reduced voltage.
Another way to improve the switching time is to switch the diode from ON to OFF by applying a negative
(reverse bias) voltage during off switching. This will give rise to the reverse recovery transient. In this case a
current will be flowing during the switching and thus the excess charges will be removed via a combination
of diffusion and recombination. The physics associated to this process is represented in the drawings below.
The magnitude of the switchoff voltage in the given case is equal to the magnitude of the on voltage. The
time period T has to be large enough to allow the diode to switch off completely in each cycle.
The second method to switch the diode is to drive the excess carriers from the neutral region via a current,
then charge will be removed via recombination and diffusion currents.
e(t)
+E
i
0 t
E
T
Figure 16: Circuit and input square wave. R is the load resistance.
32 EE2  Semiconductor Devices 2012
I
r
→ gradient>0
p
+
n
R
e(t)
Figure 17: Excess hole concentration distribution in the nregion as a function of time during the transient.
Note the change in the gradient of the excess minority carrier density at the moment of switchoff (below the
xaxis). At ON: the current is positive, at the moment of switch OFF the current becomes negative.
28
Figure 18: Top: the full line is the
diode currentvoltage
characteristic. The dashed line
with the arrows is the currenttime
diode characteristic whilst
switching. Bottom is the variation
of the voltage across the diode as a
function of time. The graphs at the
right are the currents and voltages measured across the diode as a function of time. t
sd
is the storage delay
time.
At t<0 the diode is ON and a current I
f
=E/R is flowing. The forward bias current will be mainly determined
by the bias circuit because the resistance of the forward biased diode is very small. As a consequence, the
voltage drop across the diode is negligible compared to that across R. Thus the load resistance will limit the
current. This current determines the slope of the excess carrier concentration at the depletion region edge.
We know that this slope is consistent with the diffusion of minority carrier holes in the ntype region from
28
Note that the on and off state excess carrier concentration is not drawn to scale. The on state excess carrier concentration is a lot
larger than the offstate excess carriers concentration.
33 EE2  Semiconductor Devices 2012
x
δ
p
I
f
gradient<0 →
I
r
→ gradient>0
t
I
V
I
f
I
r
t
v(t)
t
i(t)
E
I
f
t
I
r
v(t)
t
sd
t
sd
the depletion region edge to the contacts (similar for electrons in the ptype region). At t=0 the voltage
polarity is reversed whilst the magnitude of the voltage is the same (this equality is not necessary and is here
used for convenience). Then a negative current I
r
=E/R will be flowing that is consistent with this voltage.
This negative current is many times larger (in absolute value) than the reverse bias current that is normally
flowing through the pn diode in DC reverse bias. The reason why this happens is the following. When the
diode was initially forward biased it had a narrow depletion region (that only allows a small voltage drop
across it) and a relatively large density of stored minority carriers. When the external bias is reversed, we
know from the previous switching experiment that the excess minority carrier charge that is stored in the
neutral regions will not disappear immediately. As a consequence, the depletion region cannot increase and
the voltage across the diode can also not increase immediately. Remember that the depletion region is a
region without free charged carriers. Thus in order to deplete the neutral regions, we first have to deplete the
regions around the junction of the excess minority carrier charge. Once that is done, the region around the
junction can be depleted of the majority carriers. Doing this, forms the depletion region. Thus we can see
now that when the external bias goes negative, the voltage drop across the diode is still small and negligible
compared to the voltage drop across R. The current is thus determined by load resistance R and applied
voltage –E.
Since a negative current, I
r
is flowing, the slope of the excess carrier concentration at the depletion region
edge is exactly opposite to the one in forward bias. This indicates that minority carriers have to flow the
other way than under positive bias
29
. Thus minority carrier excess holes are diffusing from the ntype region
to the ptype region and electrons from p to n. This current removes the excess charge. The current is large
as long as it is determined by the external circuit and excess minority carriers are available. When the excess
is removed (mainly by diffusion but also by recombination) then the depletion region will be able to build
up. The current is decreasing as only the intrinsic minority carrier density remains to carry current.
Thus as long as excess carriers remain, the slope of δp at the depletion region edge remains constant. When
the excess is removed, the depletion width adapts and an increasing voltage will be dropped across the
increasing depletion width. The current through the circuit will then be more and more limited by the current
that is allowed to flow through the diode. The final current will be determined by the small amount of
minority carriers in the neutral region near the depletion region edge. This is maintained at steady state via
carrier generation processes. The OFF current is thus the leakage current through the reverse biased diode.
At t=0 the excess carrier concentration at the edge of the depletion region is p’
n
and at the end of the switch
cycle the excess carrier concentration at the edge of the depletion region should have become p”
n
.
,
_
¸
¸ −
,
_
¸
¸
·
·
kT
eE
n n
kT
eE
n n
e p p
e p p
0
' '
0
'
This is clearly illustrated in fig. 18. Important to note when you look at fig. 18 is the point where the diode
can finally go into reverse bias. This point is where v(t) – the voltage across the diode junction  becomes
equal to zero, which is consistent with the complete removal of excess charge, thus Q
p
=0. The time it takes
to remove all the excess charge is called the storage delay time t
sd
.
In order to find the variation of the charge as a function of time, eq. (15) has to be solved with the boundary
29
Remember that the direction of the diffusion current is given by the gradient of the minority carriers at each point x in the
device.
34 EE2  Semiconductor Devices 2012
conditions that for t<0 the charge is Q
p
=I
f
τ
p
and at t=0 the current becomes I=I
r
. This approach will allow
us to derive an expression for the storage delay time. The assumption that we are making is that the
condition I=I
r
is steady state until we reach t=t
sd
. This approach will not give the time it takes the diode to
buildup the depletion region from the “noexcess charge” condition. Solving (15) for a p
+
n diode
30
can be
done by using Laplace transforms or as follows, rewriting (15) :
p
p
p
p
t Q
t i
dt
t dQ
τ
) (
) (
) (
− ·
For 0 < t < t
sd
we know that i
p
remains constant at –I
r
, thus:
p
p
r
p
t Q
I
dt
t dQ
τ
) ( ) (
− − ·
Separation of variables:
p r p p
p dt
I t Q
t dQ
τ τ
− ·
+ ) (
) (
Integrate:
( )
( ) ( )
( )
( )
( )
,
_
¸
¸
− + + − ·
− ·
+
+
− · + − +
− · +
− ·
+
∫ ∫
p
r p F p r p p
p r p F p
r p p
p
r p F p r p p
p
t Q
I
r p p
t
p
t Q
Q r p p
p
t
I I I t Q
t
I I
I t Q
t
I I I t Q
t
I t Q
dt
I t Q
t dQ
p
f p
p
p
τ
τ τ τ
τ τ τ
τ
τ
τ τ τ
τ
τ
τ τ
τ
exp ) (
) (
ln
ln ) ( ln
) ( ln
) (
) (
) (
0
) (
) 0 (
or
( )
1
1
]
1
¸
,
_
¸
¸
−
+ + − ·
p
r f r p p
t
I I I t Q
τ
τ exp ) (
(20)
Eq. (20) is valid for 0 < t < t
sd
only.
The storage delay time t
sd
can then be extracted by stating that at t=t
sd
, Q
p
(t
sd
)=0. Thus:
( )
1
1
]
1
¸
,
_
¸
¸
−
+ + − ·
p
sd
r f r p
t
I I I
τ
τ exp 0
The storage delay time is then (under the assumption of quasi steady state) :
1
]
1
¸
+ ≅
r
f
p sd
I
I
t 1 ln τ
Switch from forward to reverse bias state:
• reverse current larger than diodes’ reverse saturation current
• transient current and voltage are delayed by change in stored charge
• this causes a storage delay time t
sd
during which the current is larger than the reverse bias current
30
Current determined by the hole current.
35 EE2  Semiconductor Devices 2012
associated to the DC characteristics
• delays in pn diodes are a consequence of excess minority carrier charge
As before, to improve the switching time in pn diodes one can introduce recombination centers such that the
lifetime of the carriers reduces which makes the excess charge disappear faster. Unfortunately this also
increases the reverse leakage current.
The delay time can also be reduced by using a short device layer.
Other solutions to obtain fast rectifying devices include the use of Schottky diodes. Here only majority
carriers need to be removed which is a fast process. In a Schottky diode the junction is formed by a metal
semiconductor contact, e.g a metal with a larger work function than the work function of ntype silicon. This
depletes the majority carriers (electrons) in the semiconductor at the junction. No depletion of carriers
occurs in the metal. In order to switch the Schottky diode from on to off, it is the majority carriers that move
back towards the junction. There are plenty of them and we do not have to wait for generation of these
majority carriers to establish resupply. Remember that the fast switching in the Schottky contact can be
exploited to improve the switching behaviour in bipolar junction transistors too.
The use of GaAs (high mobility carriers and shorter lifetime than Si) results in ultra high speed diodes.
Specially designed pn diodes which strongly conduct for a short period of time in reverse bias – the step
recovery diode – can switch in the picosecond range and are used for harmonic generators.
2.6 Small signal equivalent circuit of a pn junction diode
If diodes are to be used as components in circuits, their nonlinear characteristics can become cumbersome
for fast circuit simulation purposes. Very often in circuit operation a device is functioning around a DC point
(see analogue electronics course). The variations around this point are relatively small such that the device
acts as if it is linear around this point. Therefore for circuit design purposes an equivalent circuit can be used
with component values dependent on the chosen DC point but further giving a linear response. For the diode
we have seen that there is a resistive component and a capacitive component that are both a function of
voltage. The equivalent circuit is given in fig. 21.
R
s
r = 1/g
d
d
C
Figure 21: Small signal equivalent circuit of a pndiode. R
s
is the contact resistance, r
d
is the diode’s
differential resistance and C is the voltage dependent capacitance.
36 EE2  Semiconductor Devices 2012
The series resistance R
S
represents the nonideal characteristic of the Ohmic contacts. The capacitance C is
due to charging and discharging effects occurring during AC operation. There are two different situations.
One is in reverse bias where a depletion region with fixed ionised charges is builtup with an amount
dependent on the applied voltage
31
. The other is in forward bias and takes into account the variation of the
builtup excess charge as a function of voltage.
 Depletion capacitance in reverse bias (cfr. parallel plate capacitance)
 Diffusion capacitance in forward bias (current injection across junction)
Reverse bias: Depletion capacitance
( ) ( ) V V N N
N eN
A C
W
A
C
D A
D A
depl
depl
depl
− +
·
·
0
2ε
ε
ε
W
depl
is the depletion width, V
0
the builtin voltage and ε = ε
0
ε
r
.
Figure 22: The variation of the depletion capacitance as a function of bias across the pnjunction.
Forward bias: Diffusion capacitance
τ I
kT
e
C
dV
dQ
C
dV
dQ
C
diff
p
diff
diff
·
·
·
+
junction n p for
The differential resistance r
d
relates to the slope of the IV characteristic. Easily calculated as a differential
conductance:
I
kT
e
dV
dI
g
r
d
d
· · ·
1
2.7 Large signal equivalent circuit of a pn junction diode
While a small signal equivalent circuit does not consist of nonlinear elements because due to the small
variation in applied voltage the IV region used can be assumed linear. This is no longer the case for a large
31
Review the calculation of the depletion width in your first year EE1.3 Semiconductor Devices course.
37 EE2  Semiconductor Devices 2012
Depletion
Capacitance
V
reverse
forwar
d
signal equivalent circuit. Linearization is no longer possible because the whole nonlinear currentvoltage
region is used.
The large signal equivalent circuit for pn diodes is given in fig. 23.
Figure 23: large signal equivalent circuit of pn diode.
2.8 Conclusion
The characteristics of a pn diode, at low injection levels (meaning that the injected density of carriers across
the junction is much smaller than the majority carrier concentration available), are completely described by
the variation of the minority carriers at each side of the junction. The bias voltage across the junction defines
the excess carrier concentration at the edges of the depletion region while the Ohmic contacts impose bulk
conditions on the material. When recombination of carriers is taken into account, the excess carrier
concentration varies exponentially if we take the long diode approximation. If we ignore carrier
recombination completely – an assumption that can be made when the diode is sufficiently short – then the
excess carrier concentration varies exponentially. The maximum gradient of the excess carrier
concentrations determines the current in the pn diode. This maximum gradient occurs at the depletion region
edges. Due to the existence of excess carriers in the neutral regions of the diode, delays occur when
switching. The density of excess carriers and the bias voltage across the junction are intimately related, this
means that the voltage across the junction can only change when the density of excess carriers has changed.
When switching from on to off, the depletion region in the pn diode across which the largest part of the
voltage drops, can only extend when the excess carriers are removed. Switching times in devices where
minority carrier concentration play an important role are normally relatively large compared to switching
times in majority devices such as a Schottky diode.
38 EE2  Semiconductor Devices 2012
C
R
s
3. Bipolar Junction Transistor (BJT)
3.1 Review – the short BJT
The BJT is a device which has three ohmic contacts to the three different semiconductor layers that are
connected together. Basically, the BJT structure consists of a series connection of a pn and np or np and pn
diode where the nregions, respectively pregions, of both diodes overlap. This makes a pnp and npn BJT.
Having a series connection of two pn diodes, there are now two junctions to consider of which the
boundaries are not connected to an Ohmic contact. We will see that it is the voltage across each junction
separately that determines the currents through the BJT.
The BJT is another device in which the minority carriers will play an important role, similar to the pn diode.
As in the pn diode the voltage across each junction will determine the density of excess carriers at the edges
of the depletion regions of each junction. The calculation of the excess minority carrier concentration across
each junction is thus identical to that of the pn diode. As a consequence the currents that are flowing in the
different layers of the BJT will be determined by the maximum minority carrier concentration gradients,
which will occur near the depletion region edges.
Recombination of carriers in the base is very important as it will determine whether the material
combination will have amplification or not. Notwithstanding this, very often recombination is ignored and a
“short BJT” approximation is used. This will allow easy mathematical formulae to describe the currents and
gives relatively good approximations for the collector current.
3.2 Charge transport in BJTs (principle of operation)
In order to get an idea of how a BJT functions it is good to review the pn diode. Remember that in a reverse
biased pn junction, the current
,
_
¸
¸
− ·
,
_
¸
¸
1
nkT
eV
s tot
e I I
becomes
s tot
I I − ·
and is independent of the magnitude
of the reverse bias voltage V. I
s
is given by:
,
_
¸
¸
+ ·
p
p n
n
n p
s
L
D p
L
D n
eA I
0 0
. The magnitude of I
s
is determined by
the density of minority carriers, p
n0
and n
p0,
in each layer. In reverse bias it are minority carriers that are
crossing the junction, namely holes flow from the ntype material to the ptype and electrons from the ptype
to the ntype. As the number of minority carriers is restricted
32
, this magnitude determines the current.
p n
V
I
V
I
I
sat
Figure 1: Left: reverse biased pnjunction. Right: Currentvoltage characteristics.
32
Remember that the minority carriers need to be supplied by carrier generation.
39 EE2  Semiconductor Devices 2012
Thus if we decrease the doping density in each layer, the magnitude of the minority carriers will increase
and the reverse bias current in the pn diode will increase. Of course doping is not a good option for active
control of the offcurrent. Another possible way to control the current density is by shining light of the
correct frequency on the junction. This light will be absorbed and creates electronhole pairs (carrier
generation). The minority carriers near the depletion region edge – thus those that feel the electric field
across the depletion region – will be swept across the depletion region and thus increase the offcurrent. This
can be done in a controlled way by varying the intensity of the light. More intense light will generate more
carriers, and thus more minority carriers near the depletion region edges and thus the offcurrent will
increase with increasing light intensity. This is illustrated in fig. 2 where only an increased injection of holes
is considered.
p n
V
I
V
I
I
sat
hole injector
Figure 2: Left: Holes are injected in the ntype region near the depletion region edge of a reversed biased
pnjunction. Right: the offleakage current increases as the available minority carrier density increases due
to the hole injection process.
For an all electrical solution to the active control of the offcurrent in a reversed biased pn diode, we can use
a forward biased p
+
n junction (+: heavily doped pside). A heavily doped p layer in a pn junction will cause
the total current to be mainly dependent on holes. The reason can be easily seen from the gradients of the
minority carriers in each region: in the nregion the gradient is given by
,
_
¸
¸
· ≈ −
T
EB
n n n n
V
V
p p p p exp ' '
0 0
and
in the ptype layer the gradient is given by
,
_
¸
¸
· ≈ −
T
EB
p p p p
V
V
n n n n exp ' '
0 0
. Here V
EB
>0V (forward bias) and
e
kT
V
T
·
. Since N
A
>> N
D
we find that
0 0
p n
n p >>
and thus the hole minority carrier gradient, that defines
the hole diffusion current in the nregion is larger than the electron minority carrier gradient in the p
+
region.
Since I
p
>> I
n
we call a p
+
n junction a hole injector.
Thus if we put this hole injector sufficiently close to a reverse biased pn junction then the hole injector will
increase the amount of minority carriers (in this case only holes) that will flow across the junction and thus
the offcurrent increases in exactly the same way as given in fig. 2.
40 EE2  Semiconductor Devices 2012
p
+
n
V
EB
V
BC
I
C
I
E
I
B
I
C
V
BC
p
Emitter
Collector
holes injected holes collected
Base
W
b
I
E
Figure 3: Left: a combination of a forward biased p
+
n junction, acting as a hole injector and a reverse
biased np junction acting as a hole collector – the BJT. Right: the collector current (holes collected) as a
function of the emitter current (≈holes injected).
Fig. 3 gives the whole combination of pn and np junction forming a pnp bipolar junction transistor (BJT).
The heavily doped p+ layer of the injector diode is called the emitter (E), the layer that is in common for the
two diodes is the base (B) and the layer that will collect all the holes that are injected by the p
+
n diode (thus
injected by the EB junction) will be collected by the reverse biased np junction. The player of the hole
collecting diode is called the collector (C). If the base width is sufficiently small the collector current that
will flow as a function of reverse bias voltage across the BC junction is given in fig. 3 (right). Thus if the
emitter current – which is the current across the EB junction – increases, then the collector current is
increasing accordingly. The magnitude of the current is determined by the minority carriers that are injected
across the EB junction rather than the reverse bias voltage across the BC junction. Note that if we take the
approximation that the emitter current consists of a hole current I
p
only, and all the injected holes get through
the base, then the collector current is equal to the emitter current and there is no current amplification
between I
E
and I
C
.
The BJT thus operates by injection and collection of minority carriers. A good BJT is an asymmetric
device: the doping density in the emitter is higher than the doping density in the collector. From the physics
we can see that since we want to collect more holes, it needs to be holes that are injected, thus we need a
hole injector. Later we will see from the expression of the current amplification factor and the emitter
junction injection efficiency that high emitter doping gives good amplification and injection efficiencies.
This approach has given us an insight into the relationship between the emitter and collector current in a
common base configuration (as given in fig. 3). However what happens in between the common base point
(common between the emitter and collector)? Is there a current flowing from this point in or out of the base
region? The answer of course is yes. In order to understand this, we have to take a closer look at what all the
carriers are doing in the structure under the given biasing conditions.
The main carrier flow in a pnp BJT is holes, as seen above. In normal active mode operation of the BJT:
V
EB
>0V and V
BC
<0V in a pnp BJT, a large amount of holes are injected by the emitter into the base and
these are then collected by the collector. However we have seen in pn diodes that both hole and electron
currents exist at the same time but not necessarily in the same magnitudes. The magnitudes is determined by
the doping densities. We have also seen that recombination of carriers can occur. If recombination of holes
occurs in the base region then it can be expected that the collector current will decrease, as there will be less
holes available for collection at the collector side.
41 EE2  Semiconductor Devices 2012
As a consequence there will also be electron currents in the BJT. These electron currents have a direct
relationship to the base current in a pnp BJT. In fig. 4 all the different current components that can exist in a
BJT are given.
p
+
n
V
EB
I
E
I
B
I
C
V
BC
p
hole flow
electron flow
1
2
3
4
5
Figure 4: The BJT in common base configuration in forward active mode with the flow of both holes and
electrons (flux) indicated with arrows. The numbers indicate different physical events explained in the text
below.
At the emitterbase junction holes are injected into the base and electrons are injected from the base into the
emitter. This means that electrons are disappearing out of the base region. If there would be no resupply of
these disappearing electrons then the density of majority carriers in the base (electrons) would decrease. To
accommodate this decrease in majority carriers while maintaining charge neutrality, the depletion region at
the EB junction has to adapt. This means that the forward biased voltage across the EB junction decreases
and thus reduces the amount of injected carriers. Thus no resupply of electrons would lower the hole
current to zero. Therefore in order to maintain a collector current, the lost electrons from the base through
the forward biased EB junction has to be resupplied by an electron base current. This is indicated in the fig.
4 by arrow n
o
5.
Whilst the injected holes (hole flow in fig. 4) are diffusing through the base, there is a probability for these
holes to recombine with the electrons available in the base region. As a consequence of the recombination
process the number of holes and electrons in the base is decreasing. If the number of electrons in the base is
not resupplied the charge density relationship in the base would break down and hole current would
disappear. In order to maintain the hole current I
C
for the given biasing conditions, the base contact has to re
supply the electrons, this is given by arrow n
o
4. Arrow n
o
1 represent the injected holes lost for I
C
through
recombination in the base.
The thermally generated electrons and holes make up the reverse saturation current across the basecollector
junction. The reverse biased BC junction thus supplies electrons to the base. This reduces the amount of
electrons that the base current needs to resupply. This supply however is found to be dramatically smaller
than the resupply current and will therefore often be ignored. Arrow n
o
2 are the holes reaching the
collector.
From this diagram and the need to maintain the given steady state bias conditions, the base current I
B
can be
found to consist of three terms:
0
" '
CB B B B
I I I I − + ·
(1)
where
B
I '
= base current to compensate lost carrier injected into the emitter
B
I"
= base current to resupply the recombined charge
42 EE2  Semiconductor Devices 2012
0 CB
I
= reverse bias leakage current that injects carriers from collector into base (very small)
Thus the base current resupplies the majority carriers that are lost in the base (neglecting I
CB0
). If these
would not be resupplied, charge neutrality is not maintained unless the depletion regions and thus the
voltage drop across the different junctions adapt themselves and thus change the total current I
C
that is
flowing through the device.
Although this discussion of the different currents is done for the common base configuration, these currents
are exactly the same in the other BJT biasing conditions, common emitter or common collector. The
common base configuration is used because it gives a direct insight into the voltage across each junction EB
and BC separately, making the analysis the same as for the pn diodes we have discussed in chapter 2. Note
that the main difference in analysis is the boundary conditions in the base junctions. These boundary
conditions are imposed by the voltage across each junction rather than an ohmic contact as was the case for a
single pn junction. This means that the minority carrier excess in the base at the depletion region edges will
not be the bulk minority carrier concentration
33
but will be determined by the voltage across each junction
and thus determined by the formulae (chapter 1 eq.(1)):
,
_
¸
¸
· ·
kT
eV
n
n
p
p
e
p
p
n
n
0
'
0
'
(2)
The variation of the minority carrier concentration will then be dependent on whether recombination has to
be taken into account. Note that a BJT has a short base. However, when recombination is taken into account
in order to more accurately predict the current gain, the variation of the minority carriers in the base has to
be given as in eq. (9) of chapter 2.
To obtain a “good” BJT it is essential to limit the amount of carriers that are lost from the base via
recombination and via the current injected from the base into the emitter. In order to deliver the 1
st
requirement the base needs to be made sufficiently narrow and for the 2
nd
requirement, the emitter needs to
be heavily doped such that carrier injection from emitter to base is a lot larger than carrier injection from
base to emitter. This is the reason why BJTs are asymmetric – the emitter is more heavily doped than base
and collector.
3.3 Calculating the currents in a BJT
As for the pn diode, we can make the choice of including recombination or not. The choice will depend on
the accuracy that is required and the speed with which derivations and calculations have to be done. In a BJT
under normal active mode (EB forward biased and BC reverse biased) the short diode approximation, in
which recombination is neglected, is often sufficiently accurate to get an idea about device performance.
The derivation of the currents is easy for a BJT in the short diode approximation. It that case, both emitter
and base layers are thin. For a longer base or in case we want to take recombination into account, the
calculations become more complicated. In what follows we will explicitly derive the currents in the case
where recombination is not taken into account and we will look into some of the issues, such as boundary
conditions and shape of the minority carrier concentration in the base, that are a result of recombination.
33
As would be the case for an ohmic contact
43 EE2  Semiconductor Devices 2012
The short BJT approach
Assumptions:
• No recombination in the emitter and base region
34
• The reverse bias collector current into the base is negligible
35
• The depletion regions are neglected
In this case and under normal forward active mode, the currents in the emitter base diode determine all the
BJT currents completely. The holes that are diffusing through the base will be collected by the collector and
thus make the collector current. The electrons that are escaping from the base into the emitter have to be re
supplied by the base current, thus the electron current of the EB diode defines the base current. The emitter
current is the total current in the EB diode. This gives the well known result:
I
E
=I
B
+I
C
A good way to remember this is to draw a BJT symbol with the currents and remember the nodal analysis
techniques that you learned in the EE1 Circuit Analysis course.
Figure 5: Left: A pnp BJT symbol with the directions of the different currents. Right: an npn BJT symbol
with the direction of current.
Apply nodal analysis (sum of current into node = sum of currents out of node) in the base node of fig.5:
For pnp BJT: I
E
= I
C
+ I
B
For npn BJT: I
C
+ I
B
= I
E
34
Whether recombination exists in the collector or not is not important for the derivation of the expression of the current in this
case.
35
This can only be done for forward active mode, when the BJT is OFF, thus EB and BC junctions reverse biased then the reverse
bias leakage current is important from a viewpoint of power consumption in the OFF state.
44 EE2  Semiconductor Devices 2012
C
E
B
p
p
n
I
E
I
B
I
C
C
E
B
n
n
p
I
E
I
B
I
C
Figure 6: top: pnp BJT in forward active mode. Bottom: The minority carrier density variation in each layer
for the short BJT. The shaded regions are the depletion regions. Notice that the variation of the density of
the minority carrier concentrations is linear in each (short) layer.
In bipolar devices (thus both electrons and holes play a role in the transport processes) diffusion of minority
carriers are the basis for the calculation of the current. Diffusion happens because there is a carrier gradient.
Thus the first step in the derivation of the current in the BJT is to determine the minority carrier
concentration at each side of the junction for both junctions in the device. This is done in exactly the same
way as in the pn diodes. The result is shown in fig. 6.
In the emitter at the EB junction the injected minority carrier concentration (electrons coming from the base)
is
36
:
,
_
¸
¸
·
kT
eV
p p
EB
e n n
0
'
(3)
In the base at the EB junction the injected minority carrier concentration (holes coming from the emitter) is:
,
_
¸
¸
·
kT
eV
n n
EB
e p p
0
'
(4)
In the base at the BC junction the minority carrier holes are injected to the collector and thus we have a
reduction of the minority carrier concentration in the base at the reverse biased collector:
,
_
¸
¸ −
·
kT
V e
n n
BC
e p p
0
"
(5)
Remember that diffusion currents are directly related to the gradients of the minority carriers. The minority
carrier concentration at the ohmic contact of the emitter and collector imposes the bulk minority carrier
concentration
37
.
36
Note that the ‘ indicates forward bias and “ reverse bias.
37
Due to infinite generationrecombination rates in ohmic contacts, the contacts can supply and absorb all excess carriers and thus
keep the carrier concentrations constant at bulk conditions.
45 EE2  Semiconductor Devices 2012
x
p+ n p
V
EB
V
BC
x
e W
b
0 p
n
m i
n o
r i t
y
c a
r r i
e r
c o
n c
e n
t r a
t i o
n
p
n'
n
p"
p
n"
0 p
n
n
p'
Thus the electron current of the EB diode is given by:
,
_
¸
¸
,
_
¸
¸
≈
−
·
−
−
·
kT
eV
e
p n
e
p
kT
eV
p
n
e
p p
n n
EB
EB
e
x
An eD
x
n e n
A eD
x
n n
A eD I
0 0 0 0
0
'
The hole current crossing the EB junction is given by:
,
_
¸
¸
,
_
¸
¸
· ≈
−
·
kT
eV
b
n p
b
kT
eV
n
p
b
n n
p p
EB
EB
e
W
Ap eD
W
e p
A eD
W
p p
A eD I
0
0
" '
Note that n n
p p ' ' ' <<
and n n
p p '
0
<<
, therefore the formulae above can be simplified. The emitter current is
then the pn diode current: p n E
I I I + ·
The base current is:
n B
I I ·
and is thus the resupply of the electrons that have left the base via the EB
junction. The collector current is: p C
I I ·
. Thus all the injected holes by the EB junction are collected by the
BC junction. It is interesting to note that in the short diode approximation, the currents in the BJT are
completely determined by the emitterbase diode only!
Important parameters defining the quality of the transistor are:
The emitter efficiency:
n p
p
I I
I
+
· γ
, this parameter gives how much of the total input current (thus the
emitter current) is “useful” for the output (the collector current). Ideally we want γ =1 because in that case
I
n
=0 and thus no electrons would be lost out of the base via the EB junction. Thus the base current needed to
maintain the collector current would be equal to zero in this case.
The current transfer ratio:
E
C
I
I
· α
gives the ratio of the input current to the output current. Thus it is
preferable as close to 1 as possible. This means no amplification between emitter and collector current. Note
that γ=α when no recombination is taken into account. This will not be true for the case of recombination
because then I
C
is lower compared to I
p
injected across the EB junction.
The current amplification factor (gain):
B
C
I
I
· β
shows that we want I
B
as small as possible for a given I
C
.
Ideally we want β as large as possible.
Taking recombination into account
Assumptions:
• Recombination in the emitter, base and collector region
• The reverse bias collector current into the base is negligible (thus I
CB0
=0A)
• The depletion regions are neglected
The minority carrier variations are given in fig. 7. The first important observation is that the minority carrier
concentrations at the junctions are still the same as those given by equations (1) as they are determined by
the voltage across the junctions. The difference lies in the variation of the minority carrier concentration
within each region. In the emitter and the collector regions we can expect the variation of the minority
carrier density to be the same as in the long diode approximation since the boundary condition imposed by
46 EE2  Semiconductor Devices 2012
the contacts ensures that the excess minority carrier concentration in these layers go to zero at the contacts.
An exponential approximation to the minority carrier concentration can be used for the emitter and collector.
In the base however, the boundary conditions are imposed by the junction voltages. This means that the
exact solutions to the continuity equations need to be used. In this case, the sum of two exponentials as the
solution of the second order differential equations will be used.
The continuity equations are repeated:
0
0
2 2
2
2 2
2
· −
∂
∂
· −
∂
∂
p
n
L
p
x
p
L
n
x
n
δ δ
δ δ
(6)
Figure 7: top: pnp BJT in forward active mode. Bottom: The minority carrier density variation in each layer
for a BJT in which recombination is taken into account. The shaded regions are the depletion regions.
Notice that the variation of the density of the minority carrier concentrations is exponential in E and C
layer. In the base the variation of the minority carriers is a sum of an increasing and decreasing
exponential.
The boundary conditions for the base region (see fig. 7), which is surrounded by 2 junctions, in forward
active mode is V
EB
> 0V and V
BC
< 0V for a pnp BJT:
0 0
0 0
' ' ) ( @
exp ' ) 0 ( 0 @
n n n n b b
T
EB
n n n n n
p p p p W p W x
V
V
p p p p p x
C
E
− ≈ − · ∆ · ·
,
_
¸
¸
≈ − · ∆ · ·
δ
δ
(7)
47 EE2  Semiconductor Devices 2012
x
p+ n p
V
EB
V
BC
x
e W
b
0 p
n
m i
n o
r i t
y
c a
r r i
e r
c o
n c
e n
t r a
t i o
n
p
n'
n
p"
p
n"
0 p
n
n
p'
0 W
b
x
x=0 is the start point of the base region and W
b
is the width of the base
38
. Thus the solution of the differential
equation (6) with these boundary conditions is no longer simply a decaying exponential. The general
solution of a second order differential equation for the holes is:
,
_
¸
¸
−
+
,
_
¸
¸
·
p p
n
L
x
C
L
x
C x p exp exp ) (
2 1
δ
(8)
Where C
1
and C
2
are constants that are determined by the boundary conditions (7):
,
_
¸
¸
−
+
,
_
¸
¸
· − ·
+ · ·
p
b
p
b
n b n
n n
L
W
C
L
W
C p W p
C C p p
exp exp ) (
' ) 0 (
2 1
2 1
0
δ
δ
(9)
Substituting these boundary conditions in (8) gives the expression for the minority carrier variation in the
base when recombination is taken into account.
In order to calculate the currents in the case where recombination is taken into account, the gradients of the
minority carrier concentration have to be calculated at the edges of the depletion regions for each junction
because at these points the minority carrier gradients are maximum.
Thus the emitter current is given by:
0
0
·
·
+ − ·
x
p
n
x
n
p E
dx
n d
eAD
dx
p d
eAD I
δ
δ
39
and is the sum of the hole and the
electron current determined by the gradient of the minority carriers at the EB junction.
The collector current is given by:
b
W x
n
p C
dx
p d
eAD I
·
− ·
δ
and is the gradient of the hole minority carrier
concentration in the base at the BC junction.
The base current will now consist of 2 components as given by eq. (1) B B B
I I I " ' + ·
where we have now
assumed that
A I
CB
0
0
·
. Thus the base current consists of the electron current lost across the EB junction:
0
'
·
·
x
p
n B
dx
n d
eAD I
δ
and the current that needs to resupply the recombined electron population in the base. The
number of electrons that have recombined in the base is equal to the number of holes with which they have
recombined. We know the number of holes that have recombined in the base from the difference in the hole
minority carrier gradient between the emitter and the collector side. Thus this currentdifference that follows
from the gradient difference at the junction at both sides of the base must be the current that needs to be re
supplied via the base contact to compensate the electron charge loss in the base. Thus:
b
W x
n
p
x
n
p B
dx
p d
eAD
dx
p d
eAD I
· ·
+ − ·
δ δ
0
"
. This is illustrated in fig. 8 that focuses on the base region only. Note:
A is the cross sectional area.
38
Notice that in fig. 7 the depletion regions, that reduce the effective base width, have been taken into account.
39
Note that x=0 is the edge of the depletion region in each region E and B.
48 EE2  Semiconductor Devices 2012
Fig. 10: saturation region for
small –v
CE
(in red – online)
Figure 8: The base region of a pnp BJT with the variation of the excess minority carrier concentration
under forward active mode in the case recombination is taken into account. The dashed line (red) gives the
gradient of the minority carrier concentration at the emitter side, the full bold line (blue) is the gradient of
the minority carrier concentration at the collector side. The gradient at the emitter side is larger than that at
the collector side indicating that the hole current at the emitter side is larger than the hole current at the
collector side. This difference is cause by recombination of carriers in the base.
The full solution of the currents in the BJT in case recombination is taken into account is given by:
(10)
For the derivation, the interested reader can check the book of B. Streetman (see recommended textbooks).
An alternative way to calculate the base current associated with recombination is to look at the charge in the
base under steady state injection and collection (thus DC voltage across each junction). Due to the injection
process of holes at the EB junction into the base and the collection process at the collector, there will be a
hole charge Q
p
40
in the base that is consistent with this injection/collection process. The holes are diffusing
through the base with a characteristic time constant τ
t
, called the transit time. With
p
b
t
D
W
2
2
· τ
determined by
the diffusion process and the base width. Thus the steady state hole current is given by
t
p
p
Q
I
τ
·
. This hole
current is equal to the collector current I
C
.
On the other hand, we also know that the holes that are diffusing in the base recombine with a characteristic
time τ
p
, the hole lifetime in the base. Thus a charge Q
p
is disappearing per τ
p
and is thus resupplied by the
EB current. However the hole charge Q
p
recombines with the same amount of electron charges in the base.
Thus the amount of disappearing electrons in the base is Q
n
=Q
p
. Thus an amount Q
n
is disappearing per τ
p
and needs to be resupplied by the base current. Thus the base current that is needed to keep Q
n
=Q
p
is given
40
Note that the total charge in the base is given by the integration of the excess carrier concentration in the base region over the
width of the base.
49 EE2  Semiconductor Devices 2012
B E C
δ p
n
(x)
x
0
W
b
0
) 0 (
·
− ·
x
n
p p
dx
p d
eD I
δ
b
W x
n
p b p
dx
p d
eD W I
·
− ·
δ
) (
A
A
I
E
(x) e A D ≈
p
/L
p
(∆ p
E
ctnh(W
b
/L
p
)  ∆p
C
csch(W
b
/L
p
) )
I
C
(x) e A D ≈
p
/L
p
(∆p
E
csch(W
b
/L
p
)  ∆p
C
ctnh(W
b
/L
p
) )
I
B
(x) e A D ≈
p
/L
p
((∆p
E
+ ∆p
C
) tanh(W
b
/2L
p
) )
Fig. 10: saturation region for
small –v
CE
(in red – online)
by
p
p
p
n
n
Q
Q
I
τ τ
· ·
. In the case that γ =1, thus no loss of electrons from the base into the emitter, thus I’
B
=0A,
the current I
n
is equal to the base current I
B
= I’’
B
. Thus the currents and current amplification factor are given
in function of charge and time constants:
t
p
B
C
p
p
B
t
p
C
I
I
Q
I
Q
I
τ
τ
β
τ
τ
· · ⇒
·
·
"
"
(11)
Since the base transit time is a lot smaller than the recombination time, the current amplification factor β is
large. Note that eq. (11) only takes the recombination component into account in the current amplification
factor. In order to add the loss of carrier from base into emitter I’
B
has to be added to of I’
B
in (11).
3.4 Switching of the BJT
To study the switching delay in a BJT, we need to apply a biasing circuit. In common emitter configuration
the BJT biasing circuit is as given in fig. 9. A square wave voltage is applied to the base that drives the pnp
BJT from off to on and vice verse. The variation of the collector current is measured as a function of time
and the waveforms of the applied base voltage and the collector current are recorded on oscilloscope (see fig.
9).
Figure 9: Left: simple switching circuit for the pnp BJT in common emitter forward active mode. Right top:
switching voltage on the base. Right bottom: Output current i
C
as a function of time during switching.
Switching cycle
First the BJT is switched from off (both junctions reversed biased) to on, with a base current that drives the
BJT into oversaturation (both junctions forward biased, see fig.10). Then in the second stage the BJT is
driven off from oversaturation into reverse bias across both junctions. We use a pnp BJT in order to avoid
confusion between carrier flux and current. The hole current and hole flux are in the same direction, while
the electron current and electron flux are opposite due to the negative charge of the electrons.
Switchon:
50 EE2  Semiconductor Devices 2012
e
s
i
B
i
C
i
E
R
L
E
CC
R
S
p
n
p
+
i
C
I
Cmax
t
e
s
t
E
s
E
s
i
C
v
CE
Fig. 10: saturation region for
small –v
CE
(in red – online)
The circuit in fig. 9 can be solved by using the load line technique (see EE1 Analysis of Circuits). Therefore
sketch the currentvoltage characteristic of the ideal transistor in the given configuration for the left part of
the circuit in fig 9 for different values of the control current. The base controls the collector current with a
control base current: I
B
. The output characteristics of a commonemitter transistor are given in fig. 10. Each
single line in the graph for the BJT is for one value of I
B
. The lowest characteristic is for the smallest base
current. For increasing I
B
, I
C
increases. The constant current value of I
C
(active BJT mode) is related to
I
C
=βI
B.
The shaded region on the graph is the saturation region. In this region both emitterbase and base
collector junction are forward biased. Lots of minority carrier charge is built up in the base under saturation
and oversaturation condition.
The expression for the load line (purple online) in fig. 11 is:
L
CC
ce
L
c
R
E
v
R
i + −
−
· ) (
1
We can now draw the BJT characteristic and the load line on
the same graph (fig. 11) and indicate the OFF and ON point on
the graph. When the transistor is OFF the voltage across it is –
v
CE
=E
CC
. And the current is leakage current through the reverse
biased diodes only. When the transistor is ON the voltage
across the BJT goes to zero but doesn’t quite get there (see fig.
11). The current through the BJT is now limited by the load
line and the condition of I
c
=β I
B
is no longer valid.
When switching from OFF to ON the collector current through
the BJT needs to follow the load line. Each crosssection
between the BJT characteristics and the load line is a solution.
The base current is increasing, thus the output current, i
C
through the BJT increases too, until the BJT goes into
saturation at a time t
s
indicated on the graph. The voltage
across the basecollector junction is determined by the cross
point between load line and BJT characteristics, e.g. at t=t
1
the
voltage across the BJT is V
1
. The basecollector voltage goes
from reverse bias into forward bias (saturation) at this time.
The maximum collector current that can flow through the
circuit is limited by the load circuit and can be approximated
by:
L
CC
c
R
E
i ·
When the base current is increased further via the control
circuit, the collector current cannot further increase than this
maximum. The relationship I
C
=βI
B
breaks down at that point.
We have to look at the minority carrier concentration to find out what is happening when the base current
increases further.
When the baseemitter junction of the pnp BJT is switched from reverse to forward bias, holes are injected
from E into B and an excess hole charge is built up in the base. The amount of excess charge in the base is
51 EE2  Semiconductor Devices 2012
i
C
v
CE
E
CC
/R
L
E
CC
t
1
t
s
i
csat
≠β I
B
i
csat
=βI
B
i
c
=βI
B
V
1
V
ON
Fig. 11: ON switching sends the BJT from OFF
to saturation (I
B
sufficiently large). The load
line limits the maximum current that can flow in
saturation at t
s
.
determined by the magnitude of the base current. The collector current in a BJT, under the assumptions that
I
BC0
=0, is determined by the minority carrier concentration variation in the base, thus by the gradient, at the
BC junction.
The minority carrier variation in the base for the pnp BJT in OFF state is given in fig. 12.
Note that there
is a gradient of
minority
carriers near the
junctions,
otherwise there
would be no offleakage current. This gradient is nonlinear when recombination is taken into account. Both
junctions are reverse biased when the BJT is OFF:
0 & 0 < <
bc eb
v v
.
When the base current drives the BJT into saturation, the minority carrier variation in the base for the pnp
BJT at the starting point (onset) of saturation becomes as given in fig. 13.
At saturation,
the voltage
52 EE2  Semiconductor Devices 2012
5. Ic remains constant until the excess charge
built up above saturation is removed. After that
the BC junction goes into reverse bias and IC
follows the load line.
p(x)
x
W
b
p
n0
E B C
0
p”
nE
p
n0
p”
nC
Fig. 12: Minority carrier
concentration in the base of
BJT in OFF state.
Recombination is taken into
account.
p(x)
x
W
b
p
n0
E B C
0
p’
nE
p
n0
Fig.13: Minority carrier
concentration in the base of
BJT at onset of saturation.
Recombination is taken into
account.
condition across each junction:
0 & 0 · >
bc eb
v v
(thus EB forward biased and no bias across BC). After the
BJT saturates, the collector current is constant because the load line limits the maximum collector current. At
this moment the charge in the base is called the saturation charge Q
s
.
If the emitterbase bias is such that the base current drives the transistor into oversaturation, then the
associated charge buildup in the base is larger than the saturation charge: s p
Q Q >
where B p p
I Q τ ·
, then
B C
I I β ≠
.
At saturation Q
b
=Q
s
(Q
b
=base minority carrier charge) but beyond saturation s p
Q Q >
and the basecollector
voltage becomes forward biased → oversaturation. The minority carrier variation in the base for the pnp BJT
in forward active mode oversaturation is given in fig. 14:
Thus the
minority carrier
charge in the
base buildsup
further but does
not have any
more influence
on the collector current. Driving the transistor ON using oversaturation ensures fast switchon times,
however this will create a large density of store minority carrier charge in the base that will need to be
removed before the BJT can switch OFF (similar to the pn diode!)
Switchoff
Now the pnp BJT will be switched OFF by applying a reverse bias voltage on the EB junction. This removes
charge out of the base by a reverse pn diode current, the EB junction remains forward biased and the BC
junction too until all excess charge larger than the saturation charge is removed. Thus the collector current
remains constant until the excess charge s p
Q Q Q − ·
is removed. The basecollector voltage can only return
to reverse bias when this excess charge is removed. When the charge in the base has reached Q= s
Q
the BC
junction first goes to zero bias, then with further reduction of the charge in the base, the BC junction goes
into reverse bias and the collector current can decrease following the load line. The baseemitter junction is
still forward biased until all remaining minority carrier charge is removed from the base, after which the EB
53 EE2  Semiconductor Devices 2012
5. Ic remains constant until the excess charge
built up above saturation is removed. After that
the BC junction goes into reverse bias and IC
follows the load line.
p(x)
x
W
b
p
n0
E B C
0
p
n0
p’
nC
p’
nE
Fig. 14: Minority carrier
concentration in the base of
BJT in saturation.
Recombination is taken into
account.
junction can go into reverse bias. The exponential reduction of the collector current is caused by the base
emitter diode switchoff.
A summary of what is happening with the different parameters in the BJT is given in fig. 15. In this case the
short diode approximation is used for the minority carrier gradient in the base.
Derivation of the turnon and turnoff time gives:
54 EE2  Semiconductor Devices 2012
e
s
i
B
i
C
i
E
R
L
E
CC R
S
i
B
Q
B
i
C
δp
x
W
b
p
n
t
0
∆p
E
0
t
s
∆p
E
∆p
C
t
2
I
B
Q
s
t
sd
I
C
I
C
E ≈
CC
/R
L
t
E
s

E
s
t
1
t
s
t
2
t’
s
∆p
E
t
1
I
B
I
B
E ≈
s
/R
S
i
C
v
CE
E
CC
/R
L
E
CC
t
1
t
s
i
csat
≠β I
B
i
csat
=βI
B
i
c
=βI
B
1. Magnitude of the base current
is large and drives the BJT into
oversaturation.
2. Switching off the
BJT by driving the
base into reverse bias
3. Negative base
current extracts
the carriers from
the base.
5. Ic remains constant until the excess charge
built up above saturation is removed. After that
the BC junction goes into reverse bias and IC
follows the load line.
6. BC goes reverse
bias
4. Indicates negative
current through EB
junction when
driving off
7. When excess charge in
the base is removed, the EB
junction goes to reverse
bias.
Fig. 15: Top left pnp BJT with bias circuit in EB and EC. Bottom left: the different DC characteristics of the BJT in
common emitter configuration. The output current i
c
is controlled by the base current which is controlled by the EBdiode
bias circuit. The EC bias circuit forms the load line. Each point of the cross section of BJT characteristic and load line is a
possible solution to the circuit.
Middle from top to bottom: i
b
: EB drive current, Q
b
: charge in the base, ic output current. All curves are draw for 0 ON →
immediately followed at t=t
2
to OFF. The BJT is driven into saturation in this example.
Right: Minority carrier concentration in the base of BJT at different characteristic times. Recombination is ignored in the
drawings.
t=t
s
is the onset of saturation when switching on. t
2
=point of switchoff of EB junction. t'
s
is the time it takes to remove the
excess stored charge Q
B
excess. This time is referred to as storage delay time t
sd
.
Q
B
:excess
In order to calculate the switching dynamics in a p
+
np BJT, the following differential equation needs to be
solved
41
(see eq. (15) in the diode chapter):
dt
t dQ t Q
t i
p
p
p
p
) ( ) (
) ( + ·
τ
(12)
This differential equations can be solved using Laplace transforms, but in the bipolar devices’ switching
experiments an easier way exists because in large signal switching, steady state conditions occur before and
after switching on the current that controls the switching dynamics (in a BJT that is the base current
determined by the EB junction bias circuit).
Turn ON: from 0 bias to ON (ON is normally in saturation)
Switching happens at t=0s.
Boundary conditions:
For t<0 i
p
(t=0)=0 (switching from zero bias in current and voltage), then Q
p
(t=0)=0.
For t≥0 i
p
(t)=I
B
(EB bias circuit determines the current that is flowing, the potential drop across the diode is
small and the depletion region needs to be removed, therefore I
B
=E/R
s
). Q
p
(t) cannot change immediately,
the variation of this charge is described by the differential equation (12).
Under the given conditions, we see that after switching a steady state current condition is reached, then eq.
(12) can be rewritten for the time variation of Q
p
(t) after t=0 taken this steady state condition into account:
( ) ( )
,
_
¸
¸
,
_
¸
¸
−
− ·
,
_
¸
¸
−
− · −
−
·
,
_
¸
¸
−
−
−
· − − −
−
·
−
−
·
−
+ ·
∫ ∫
·
p
p B p
p
p B p B p
p p B
p B p
p
Q
p B p
t Q
p B p
t
p
t Q
t Q p B p
p
p p B p
p
p
p
p
B
t
I t Q
t
I I t Q
t
I
I t Q
t
I t Q I t Q
dt
I t Q
t dQ
dt
I t Q
t dQ
dt
t dQ t Q
I
p p
p
p
τ
τ
τ
τ τ
τ τ
τ
τ
τ τ
τ τ
τ τ
τ
exp 1 ) (
exp ) (
) (
ln
) ( ln ) ( ln
) (
) (
) (
) (
) ( ) (
) 0 ( ) (
0
) (
) 0 (
(13)
Equation (13) gives the variation of the charge in the base as a function of time when the EB junction is
switched from 0 to forward bias I
B
. If switching would have happened from reverse bias to forward bias then
there would be a delay of time t
r
before the charge can be build up in the base. This delay t
r
is related to the
removal of the depletion region
42
that would exist if the diode would have been in reverse bias before t=0s.
41
Note that we look at the charge related to the maximum current component only. This is a good approximation for the total
current variation for a p
+
np BJT in forward active mode.
42
An easy way to describe this delay time is to approximate it by the RC time constant imposed by the capacitance of the
depletion width and the resistance of the bias circuit.
55 EE2  Semiconductor Devices 2012
If the EB drive current I
B
is smaller than that necessary to drive the BJT into saturation, then the collector
current would be given by:
,
_
¸
¸
,
_
¸
¸
−
− · ·
p t
p B
t
p
C
t
I t Q
t i
τ τ
τ
τ
exp 1
) (
) (
(14)
With τ
t
the base transit time (is the time it takes the holes to travel through the width of the base).
However when the EB drive current I
B
is larger and drives the BJT into saturation, then the load line (EC
bias circuit) will limit the current that can flow in the output (see fig. 15). At that moment t=t
s
(fig. 15) the
charge in the base is equal to the saturation charge Q
B
sat
and the collector current i
c
(t) cannot further follow
the base charge given in (13).
The maximum collector current is then
43
:
L
CC
C
R
E
t i ≈ ) (
(15)
Equations (14) and (15) allow us to extract the time delay to saturation because at t=t
s
eq.(14)=eq.(15)
,
_
¸
¸
−
·
· ·
,
_
¸
¸
,
_
¸
¸
−
−
p B
t C
p s
L
CC
C
p
s
t
p B
I
I
t
R
E
I
t
I
τ
τ
τ
τ τ
τ
max
max
1
1
ln
exp 1
t
s
is small for small τ
p
and for I
C
small compared to βI
B,
t
p
τ
τ
β ·
. Thus the switchon time to saturation is fast
if we drive the BJT into oversaturation (large I
B
, I
C
limited by external circuit).
Turn OFF: from ON (saturation) bias to OFF
Switching happens at t=0s.
The turnoff time is based on two factors:
1. the removal of the charge in excess of the saturation charge. This is when the collector current
remains constant. This time is called the storage delay time in the BJT.
2. the removal of the excess charge that prevents the EB junction to turn into reverse bias. This is
similar to the pn diode driveoff delay time.
Two different techniques can be applied to turn the BJT off. One is to switch to 0 bias and current conditions
(open circuit) and the other one is by driving the EB diode into reverse bias. The second one will give much
faster switching times as we have seen for the pn diode switching. For the calculations the boundary
conditions to solve eq. (12) will change.
ON (saturation) bias to OFF (zero)
Boundary conditions:
For t<0 i
p
(t=0)=I
B
(the BJT is ON), then Q
p
(t=0)= I
B
τ
p
.
For t≥0 i
p
(t)=0 (imposed by bias circuit).
43
Slightly smaller because the voltage drop across the BJT in the ON state is not 0 (it is around 0.2V for Si BJTs).
56 EE2  Semiconductor Devices 2012
The calculation method is the same as for the ON switching but the t=0 boundary conditions have changed:
the start value of the base charge is different. After switching the steady state control current is zero.
,
_
¸
¸
−
·
+ ·
p
p B p
p
p
p
t
I t Q
dt
t dQ t Q
τ
τ
τ
exp ) (
) ( ) (
0
Although the base charge is decreasing exponentially, the collector current first remains constant. This is
because first the excess saturation charge needs to be removed before the collector current can change its
value and follow the variation of the base charge. Thus initially i
c
(t) is determined by the load line:
L
CC
C
R
E
t i ≈ ) (
for t < t
sd
(16)
,
_
¸
¸
−
· ·
p t
p B
t
p
c
t
I t Q
t i
τ τ
τ
τ
exp
) (
) (
for t ≥ t
sd
(17)
The storage delay time associated to the removal of the excess base charge above saturation t
sd
can then be
found by equating eq. (16) and (17) at t = t
sd
The storage delay time is then:
,
_
¸
¸
·
C
B
p sd
I
I
t
β
τ ln
Thus t
sd
is small for small τ
p
and for βI
B
small compared to I
C
. Thus the switchoff time is fast if we prevent
the BJT to go into oversaturation. This requirement is opposite to the ON requirement. However we can
decrease this OFF delay by driving the EB diode into reverse bias rather than waiting for recombination to
remove the charges as what is done in the ON→0 bias switch.
ON (saturation) bias to OFF (reverse bias)
Boundary conditions:
For t<0 i
p
(t=0)=I
B
(the BJT is ON), then Q
p
(t=0)= I
B
τ
p
.
For t≥0 i
p
(t)=I
B
(I
B
=E/R
s
imposed by bias circuit)
44
.
,
_
¸
¸
−
,
_
¸
¸
−
·
+ · −
1 exp 2 ) (
) ( ) (
p
p B p
p
p
p
B
t
I t Q
dt
t dQ t Q
I
τ
τ
τ
L
CC
C
R
E
t i ≈ ) (
for t < t
sd
(25)
,
_
¸
¸
−
,
_
¸
¸
−
· · 1 exp 2
) (
) (
p t
p B
t
p
c
t
I t Q
t i
τ τ
τ
τ
for t ≥ t
sd
(26)
44
Note that the amplitude of the switch off voltage is equal to that of the switch on voltage. This is not necessary.
57 EE2  Semiconductor Devices 2012
The storage delay time is then:
1
1
1
1
1
]
1
¸
,
_
¸
¸
+
·
t Csat
p B
t Csat
p B
p sd
I
I
I
I
t
τ
τ
τ
τ
τ
2
1
2
1
The addition of the term:
,
_
¸
¸
+
t Csat
p B
I
I
τ
τ
2
1
2
1
in the denominator will decrease the storage delay time compared
to the previous case.
3.5 BJT – Schottky diode clamp
It is clear from the explanations and calculations that in order to obtained fast switchon, the BJT has to be
driven into oversaturation with a large forward bias base current. However this condition is in conflict with a
fast switchoff time for which the stored minority carrier charge in the base has to be as low as possible, thus
no oversaturation. In order to be able to switchon fast and also switchoff fast without being hindered by
the excess oversaturation charge in the base, a clever technique has been invented to prevent the charge
buildup of the base region after saturation, this is done by introducing a Schottky diode clamp. Remember
that a Schottky diode is a majority carrier device that does not store minority carriers and thus switches
faster than a pn junction. Remember also that the onvoltage of a Schottky diode is smaller than that of a pn
diode. The onvoltage of the Schottky diode is determined by the metalsemiconductor work function
difference
45
. The onvoltage of a pn diode is determined by the builtin voltage. Fig. 16 gives a BJT with
Schottky diode clamp.
Figure 16: BJT with a Schottky diode clamp. Left: npn, righ: pnp.
Introducing a Schottky diode (metalsemiconductor junction) keeps the charge in the base region at the value
given at time t
s
. This is a result of the fact that the Schottky diode switches on (goes into conduction mode)
at lower voltages than a pn junction made of the same material. Therefore the Schottky diode will fix the
basecollector voltage at the Schottky onvoltage << 0.7V.
3.6 Small signal equivalent circuit of a BJT
45
To first approximation. See the Advanced Electronics Devices course for deviations from this approach.
58 EE2  Semiconductor Devices 2012
B
C
E
B
C
E
Figure 17: Hybrid π equivalent circuit.
Figure 17 gives the complete intrinsic small signal equivalent circuit. The EB diode is represented by the
differential resistance in parallel with the depletion and diffusion capacitance. The depletion capacitance can
be neglected as for amplification the BJT is used with EB forward biased. Between base and collector we
find the depletion capacitance of the reverse biased BC junction. This capacitance is also referred to as the
Miller capacitance and introduces an unwanted feedback network from the output C to the input B. g
m
is the
transconductance, it is the derivative of the output current i
C
to the control voltage v
be
(similar to the
transconductance in a MOSFET, see yr.1). R
o
is the output resistance. This resistance is associated to base
with modulation which imposes an increase of the collector current as a result of the reverse bias BC
junction depletion width extending into the base. This growing depletion width with increasing reverse bias
reduces the effective base width (= undepleted base region). This is associated to the early voltage V
A
.
B
C
E
r
ce v
be
r
be
g
m
v
be
βi
b
or
i
b i
c
Figure 18: Simplified small signal equivalent circuit. Junction capacitances are ignored.
Smallsignal current generator β (alternative symbol h
fe
)
β =
t
p
b
c
B
C
i
i
dI
dI
τ
τ
· ·
For a short pnp BJT:
b D n
e A p
b p n
e n p
n
p
W N D
x N D
W n D
x p D
I
I
· · ·
0
0
β
for high gain we require
(i) small base width W
b
(ii) large emitter doping, compared to base doping.
Transconductance
Input resistance, r
be
=r
π
59 EE2  Semiconductor Devices 2012
B
C
E
R
π
C
j,BE
C
d,BE
C
j,BC
R
o g
m
v
be
npn
v
b
e
Output resistance, r
ce
r
0
The baseemitter capacitance is due to the diffusion and depletion capacitance of the baseemitter diode. The
basecollector capacitance is due to the basecollector depletion capacitance.
The cutoff frequency of the BJT can be derived from the small signal current gain:
The transit time of the carriers from emitter to collector:
Note: the delay time through the basecollector depletion region is ignored.
3.6 Large signal equivalent circuit of a BJT
The coupled diode model – EbersMoll model.
This is a large signal model of the BJT that is frequently used in simulation tools such as SPICE.
Take a p
+
np transistor, the coupled diode models separates the operation of the BJT into two diodes, the EB
and the BC diodes. Extraction of currents is based on the evaluation of the hole distribution in the base of the
transistor for these two diodes. The operation is divided into “normal operation” and “inverted operation”,
annotated with N and I respectively.
In fig. 19 a pnp BJT is given and normal and inverted active mode defined.
60 EE2  Semiconductor Devices 2012
N : E B C
I : C B E
p
+
n p
I
E
I
C
I
B
Figure 19: definition of normal and inverted mode operation.
The charge distribution and currents in the base (plots are draw without taking recombination effects into
account) can be drawn for each operation mode separately and then combined for the complete operational
picture. This is given in fig. 20.
Figure 20: Left: components due to injection and collection in normal operation mode. Middle: components
due to injection and collection in reverse operation mode. Right: the gradient of the hole density in the
narrow base (fat line) is the sum of the dashed lines due to combined injection in normal and inverted mode.
Ebers – Moll equations
I
E
= I
EN
+ I
EI
= I
ES
(exp[(eV
EB
/kT)] – 1) – α
I
I
CS
(exp[(eV
CB
/kT)] – 1)
I
E
= EB diode current + fraction of CB diode current that reaches the EB junction
(27)
I
C
= I
CN
+ I
CI
= α
N
I
ES
(exp[(eV
EB
/kT)] – 1) – I
CS
(exp[(eV
CB
/kT)] – 1)
I
C
= fraction of EB diode + CB diode current
current that reaches the
CB junction
With:
• α: ratio of collected current to injected current
• I
ES
: emitter saturation current in normal mode (V
CB
=0, ∆p
C
=0)
• I
CS
: collector saturation current in inverted mode (V
EB
=0, ∆p
E
=0)
Or (27) can be rewritten as:
61 EE2  Semiconductor Devices 2012
δp
W
b
∆p
E
δp
W
b
∆p
C
+
δp
W
b
∆p
C
=
∆p
E
δp
W
b
I
EN
δp
W
b
∆p
C
+
I
CN
I
EI
I
CI
δp
=
I
EN
I
CN
+
I
EI
+
I
CI
I
E
= I
E0
(exp[(eV
EB
/kT)] – 1) + α
I
I
C
(28)
I
C
= α
N
I
E
– I
C0
(exp[(eV
CB
/kT)] – 1)
Where:
I
E0
= (1 – α
N
α
I
) I
ES
is the magnitude of the E saturation current with collector junction open (I
C
=0).
I
C0
= (1 – α
N
α
I
) I
CS
is the magnitude of the C saturation current with emitter junction open (I
E
=0).
This gives a description of the BJT in terms of simple diode characteristics + an extra current generator.
Based on (28) a large signal equivalent circuit can be proposed for the BJT that is valid in all operation
regimes.
Equivalent circuit:
I
E0
(exp[(eV
EB
/kT)] – 1) α
N
I
E
E C
I
E
I
B
I
C
α
I
I
C
I
C0
(exp[(eV
CB
/kT)] – 1)
B
Figure 21: Equivalent circuit derived from the EbersMoll equations.
An interesting feature of the EbersMoll equations is that emitter and collector current are described in
terms that resemble the diode equation + an extra term that provides the coupling between the emitter and
collector properties.
3.7 BJT amplification
In bipolar transistors the input impedance is low, such that a small input voltage change can result in a high
current change.
Small changes in base current result in large changes in collector current.
Region: active
* BJT characteristics (sketches)
active active
62 EE2  Semiconductor Devices 2012
I
C
I
C
I
E
I
B
cutoff cutoff
V
CB
V
CE
saturation saturation
Common base connection Common emitter connection
Figure 22: Bottom: The shaded arrows indicate the biasing connections in the common base (left) and the
common emitter (right) configuration. Top: the current–voltage characteristics associated with the two
configurations.
Note that in common base, the emitter current controls the collector current, whilst in a common emitter
configuration the base current controls the collector current. Thus common base does not give current
amplification while common emitter does.
3.8 Conclusion
The major time delay in a BJT is caused by the storage of excess minority carrier charges in the base of the
transistor. For fast onswitching a large IB is used to drive the transistor in oversaturation. For fast off
switching the baseemitter junction is reverse biased. The conditions for fast on and off switching cannot be
met simultaneously. In order to improve this, a Schottky diode clamp can be used between the base and the
collector. A popular large signal equivalent circuit for the BJT is provided by the EbersMoll equations that
allow the BJT to be modelled by two diodes and two current sources. This equivalent circuit is valid in all
regions of the BJT operation.
Whilst the MOSFET is used for integrated logic applications (fast large signal switching and very low off
currents), the BJT is mainly used for amplification (very large current gain). Although, if power
consumption is not an issue, BJT logic can be used to improve the speed performance. Due to the current
aggressive downscaling of the gate length of the MOSFET, the switching speed of the MOSFET has become
comparable to the switching speed of the BJT, but when carefully designed, at much lower power
consumption.
63 EE2  Semiconductor Devices 2012
The course aims: (1) To review the operation of diodes and bipolar junction transistors. (2) To extend knowledge on bipolar devices to include the influence of recombination. (3) To investigate the physical mechanisms underlying the delays and speed limitations of the devices. (4) To extract equivalent circuit models for the devices. Objectives: Students should be able to (1) Explain qualitatively the mechanisms of electronic conduction in bipolar devices, and calculate relevant quantities from given data. (2) Calculate and explain DC currentvoltage behaviour of diodes and BJTs, given their geometry and material properties. (3) Explain the influence of excess minority carrier recombination of the performance of the devices. (4) Explain the large signal behaviour of the devices from the internal storage of charge. Recommended textbooks EE1 background knowledge on semiconductors and semiconductor devices is required. The EE1 course notes can be found on blackboard. It is strongly recommended that you read these notes in order to understand the basic principles of semiconductor devices. “Solid State Electronic Devices”, B.G. Streetman & S. K. Banerjee, Prentice Hall International Editions, 6th ed. This is a more expensive book but the course is based on this book. It contains a lot more than what is taught in the 1st and 2nd year. “Microelectronic Devices”, K.D. Leaver, IC Press, 2nd ed. This is a compact and relatively inexpensive book which covers all the essentials. Note that in popular books such as “Microelectronics circuits” by Sedra & Smith some very brief descriptions of semiconductor device operation can be found. Whilst previous two books are very physics oriented, this book is more oriented towards the circuit engineer and thus might give additional insight. There are plenty of copies available of all these books in the Central Library. An interesting online text books can be found on: http://ecee.colorado.edu/~bart/book/book/contents.htm http://www.ecse.rpi.edu/~schubert/CourseECSE2210MicroelectronicsTechnology2010/
2
EE2  Semiconductor Devices 2012
Course synopsis 1. Revision on semiconductor devices Last year you studied the DC behaviour of metalsemiconductor, metaloxidesemiconductor and semiconductorsemiconductor junctions. We looked at the simplified version of three devices: the pn junction, the MOSFET and the BJT. We will start the course with reviewing the basic device DC operation concepts of the pn diode and the BJT because this year we will focus on bipolar devices. For more extensive information on MOSFETs please see the 3rd year Advanced Electronics Devices course1. 2. Long pn diode In contrast to the “short” devices we’ve studied in the first year, this year we will introduce recombination processes and check how they influence device behaviour. These devices will be referred to as “long” devices. Although an important characteristic of semiconductor devices is their fast response time, some delays do occur. We will investigate the physics behind the delays and investigate the resulting switching characteristics. The concept of a fast switching majority carrier device and a “slowly” switching minority carrier device will be made clear. 3. The BJT An npn or pnp junction is the basis for the bipolar junction transistor (BJT). Based on our knowledge of pn junctions, the DC functioning of a BJT will be described. As in pn junctions, minority carriers will play an important role in the conduction mechanism of the BJT. We will look explicitly at the short BJT and discover that calculating the currents in this particular configuration is very straightforward. The BJT in which recombination is taken into account offers some challenging maths but gives an insight into delays and current gain. Switching of the BJT will also be studied and methods to reduce the delay discussed. Why study semiconductor devices? There can be lots of reasons to study devices. It can be as simple as being interested in the physics behind the devices’ operation or it can give you the background to understand the aggressive downscaling effort of the CMOS industry and maybe in the future you can join in the effort. Having knowledge of the operation of
1
On blackboard 3
EE2  Semiconductor Devices 2012
Examples are emitter coupled logic. Radio Frequency Electronics. microprocessors and memory chip. These applications mainly need ultrahigh speed in which minimizing power consumption is not so important.Semiconductor Devices 2012 . Power Electronics. there are still some fields in which bipolar transistors remain important. Knowledge of bipolar transistors is also important for courses such as Microwave Technology. power switching. microwave power amplifiers. 2 See the yr3 Advanced Electronics Devices course 4 EE2 . This course fills the semiconductor knowledge gap between the 1st year Semiconductor Devices course and the 3rd year Advanced Electronics Devices course as it focuses more on bipolar devices than these other two courses. High Performance Analogue Electronics and Analogue Integrated Circuits and Systems. cell phone amplifiers etc. Although the speed of the aggressively downscaled MOSFET2 has removed the need for bipolar transistors from logic circuits.devices will enable you to understand both device and circuit modelling programs better and thus make modelling and the resulting circuits more efficient.
3 The subscript indicates the material type. Thus the hole and electron mobility are not necessarily the same. * Drift currents In a homogeneously doped semiconductor or a semiconductor with a constant carrier density. m the mass. ND.6 10 C = ±1eV . The mobility is the proportionality constant in the drift velocityelectric field relationship for small values of electric field: v = µE . τ m the average time between collisions (scattering).1 Semiconductors and junctions – a review 1. resp. The density of free holes pp is larger than the density of free electrons np3. with v the drift velocity and E the electric field.A respectively the donor and acceptor doping density (concentration). Only “free” charges can carry currents. p respectively the electron and hole density . nn = N D 2 The extrinsic ntype semiconductor: p = ni n N D 2 n n p = i NA The extrinsic ptype semiconductor pp = NA With n. Doping with donors: ND (donor concentration) gives an ntype semiconductor. Thus the density of free electrons nn is larger than the density of free holes pn in an ntype material. This is because a semiconductor always has at least an intrinsic carrier concentration: ni. • The conduction in a semiconductor can be changed via doping. holes. n is the symbol for electron concentration. • There exist two types of charged carriers in a semiconductor that can both carry current: the electrons −19 with a negative charge and the holes with a positive charge: q = ±e = ±1. Doping is the introduction of foreign atoms such as B or As in Si. Doping with acceptors: NA (acceptor concentration) gives a ptype material. The value of the mobility is a function of the scattering processes and the effective mass of the carriers: µ = qτ . The total conductivity of a semiconductor is: σ tot = e( nµ n + pµ p ) With µ n. at room temperature. The mobility gives an indication into how fast a device will work (when parasitic influences are negligible).1 A semiconductor • A semiconductor is a material that. Units are cm3. 5 EE2 . Important formulae that govern the conductivity of a semiconductor are: The intrinsic carrier concentration: ni = pi (same density of free electrons and holes in an intrinsic semiconductor). Drift current can be carried by both electrons and holes. µp the mobility of electrons. has a conduction character that is lower than metals but higher than insulators. Electrons and holes have opposite velocities.Semiconductor Devices 2012 . applying an electric field will cause drift currents to flow. with q the charge. p for the hole concentration. “Free” means not bonded to the atoms.
ρ the charge density as a function of x . * In the general case where both concentration gradients and electric fields are present the total current is the sum of both drift and diffusion currents: dn dp I tot = eA nµ n E + pµ p E + Dn − Dp dx dx This equation is normally referred to as the driftdiffusion equation of carriers and is the basic equation that describes carrier movement in semiconductor devices. diffusion currents will occur. T the temperature in Kelvin. The third important basic equation in semiconductor devices is the Poisson equation5: ρ ( x) d 2V e + − =− = − p − n + ND − NA 2 ε ε dx ( ) With V the electrostatic potential. electron 4 5 In one dimension. resp.Semiconductor Devices 2012 . E(x) Jndrift Jpdrift n(x) Jndiff p(x) Jpdiff Figure 1: the direction of the current densities as a function of the electric field and the carrier gradient. p & n the free hole. In one dimension. x is the direction of carrier propagation. electron density gradients cause electron diffusion currents and gradients in both carrier types will cause diffusion of both.Drift current in a semiconductor is given by: I tot = σ tot AE = eA( nµ n + pµ p ) E With A the cross sectional area perpendicular to the current flow. 6 EE2 . Particle flux is strictly governed by electrostatics. * Diffusion currents When carrier gradients exist in a semiconductor. creating a total diffusion current of4: dn dp I tot = eA Dn − Dp dx dx With Dn. Remember that the electric field can also be internal to the device structure. Hole density gradients cause hole diffusion currents. The Einstein equation gives the relationship between the diffusion constant and the mobility of the carrier: D kT = µ e with. Note that the direction of the electron particle flux is opposite to the current flow. k the Boltzman constant.p the diffusion constant of electrons respectively holes. E is the applied electric field. the current is dependent on the sign of the charge q=±e.
In a perfect semiconductor there are no energy levels available within the bandgap. The integration of the product of the density of states g(E) and the FermiDirac distribution function f(E) gives the carrier distribution over the energy range above or below the bandgap (right). whilst free holes have energies in an allowed energy band lower in energy than the bandgap. 1. The amount of electrons available at each energy value is determined by the density of states g(E) and the FermiDirac distribution function f(E) as illustrated in Figure 2. a region of forbidden energies exists called the bandgap Eg. This means that no electrons nor holes can reside in the energy region defined by the bandgap. f(E) gives the probability of finding an electron at energy E. Thus 1f(E) gives the probability of finding a hole. Note that g(E)=0 in the bandgap (no energy levels). Free electrons have energies in an allowed energy band higher in energy than the bandgap. g(E) gives the distribution of energy levels (states) as a function of energy. the conduction band Ec.+ − density which are both a function of x as well as of V and N D & N A the concentration of ionised doping atoms which are a function of x. the valence band Ev. 7 EE2 .2 Energy band diagrams The carriers in a semiconductor are quasi free because they are no longer covalently bonded to the atoms and thus can move but at the same time they undergo interactions with the atoms in the regular lattice.Semiconductor Devices 2012 . As a consequence of the mathematical description of the wave character of carriers in semiconductors. E E E T>0K Intrinsic semiconductor Ec EF Ev a) intrinsic g(E) E Ec EF Ev b) ntype g(E) 1/2 E 1 F(E) E E 1/2 1 F(E) E electrons holes carrier concentration electrons holes carrier concentration electrons Extrinsic semiconductor Ec EF Ev c) ptype E holes g(E) 1/2 1 F(E) carrier concentration g(E) Density of states F(E) FermiDirac distribution function Figure 2: The energy band diagram with the position of the Fermi level E F as a function of the doping type (left).
Semiconductor Devices 2012 6 .needs to be higher than Ec at v+. − dV 1 dE c = − dx e dx E(x): Electric field. The reasoning is that electrons will flow from v. Nc the effective density of states in the conduction band and Nv the effective density of states in the valence band. 3).Remember the relationship between the position of the Fermi level EF and the density of carriers in the semiconductor (given by the FermiDirac or MaxwellBoltzman distribution functions): p = Nve n = Nce ( Ev − EF ) kT ( E F − Ec ) and p i = N v e and ni = N c e ( Ev − Ei ) kT kT ( Ei − Ec ) kT with Ei the intrinsic level6. otherwise a current (diffusion of high energy carriers to lower energetic places) would flow. * Influence of the electric field on the band diagram Applying an electric field will result in a discontinuity in the Fermi level (this discontinuity is directly proportional to the applied electric field) and will result in band bending (see the tilt in potential energy Ec in Fig. dx This gives the starting point for drawing the energy band diagram since EcEF and/or EFEv is known from the doping concentration and Eg is known and remains constant in a homojunction. electrons drift “downhill”. The introduction of this level (not a state!) makes some calculations more straightforward. Thus: dE F = 0 for Vext=0. v+ en er gy Ec EF+ Ev distance E(x): electric field vEc EFEv Figure 3: Relationship between the electric field. The intrinsic level Ei is the Fermi level position for the case of an intrinsic semiconductor. thus Ec at v. This means that electrons are losing potential energy since it is converted in kinetic energy. 8 EE2 . no discontinuity or gradient can arise in the position of the Fermi level throughout the complete structure. when no external bias is applied. * Fermi level constant At equilibrium. Ec(x): potential energy. The intrinsic level Ei is an energy position within the forbidden gap that is determined by the intrinsic carrier concentration. E ( x) = Electric field E(x) points “uphill”. the electrostatic potential and the potential energy.to v+ (electrostatic attraction). The electric field points from + to – and thus up the potential hill. V(x): electrostatic potential.
+ metal . Below we present different junctions with their energy band diagrams before and after contact. initially uncharged. Materials can include semiconductors of different doping type (homojunctions). are brought into contact. the vacuum level Evac. You can check that this picture is completely consistent with the carrier fluxes and current directions that have been discussed before.3. Ensure that you understand how they are generated. before contact of the material. Si and SiGe) heterojunctions and metalsemiconductor junctions (e. This process will align the Fermi levels and create band bending in the potential energy bands. Diffusion of electrons will appear from the material with the higher lying Fermi level to the material with the lower lying Fermi level. It is also comforting to see that these directions are completely consistent with current directions used in the other courses such as Analysis of Circuits and Analogue Electronics! 1. Junctions * A junction is formed when 2 or more different materials are brought into close (atomic scale) contact. * Band diagram of junctions When two different conducting materials. Note that the workfunction of the materials does not change when the junction is formed. different materials (e. This is done by drift of carriers which is caused by the buildup of an internal electric field that creates a drift current opposite to the diffusion current and allows the system to reach equilibrium. Metalsemiconductor junctions Ohmic contact metalsemiconductor junction metal ptype Ec EFm EFs Ev .g. the electrons close to the junction redistribute themselves so that the Fermi energy is constant across the junction (in the absence of an applied voltage).Holes are flowing up the potential hill. The position of the Fermi level with respect to the vacuum level. The reason why band bending is created is due to the fact that no net current can be flowing when no external bias is applied. thus the diffusion of carriers needs to be compensated.Semiconductor Devices 2012 . As a result the internal electric field occurs in the vacuum level across the junction – this is called the contact potential or builtin voltage.g. * The work function φ In order to draw an energy band diagram of a junction we need to draw the energy bands of each material separately with respect of a local reference level that is defined for each material.+ p Ec EF Ev ohmic contact V I 9 EE2 . is given by the workfunction φ. Ohmic contact and Schottky contact). all without external bias.
+ F E s Ev I Ec EF Ev Schottky contact V Figure 5: The energy band diagram of a metal and a ntype semiconductor. Left: before contact. Left: before contact. right: after contact. At the right (top) a positive voltage is applied to the metal with respect to the semiconductor. The currentvoltage characteristic of this junction is linear. Thus the only electrons that can contribute to the current are the ones above the potential barrier and this is only a small amount thus the current flowing at this voltage will be small. For Ohmic contacts there should be an accumulation of majority carriers at the junction. right: after contact. It is a majority carrier device. A Schottky contact can be used as a rectifier. The slope of the curve is proportional to the conductivity of the semiconductor and the resistance of the Ohmic contact: Rtot = 2 Rcontact + Rsemiconductor V = Rtot I Schottky contact In fig. and the internal electric field across the semiconductor of the junction is reduced by the externally applied field and thus the barrier lowers).Figure 4: The Energy band diagram of a metal and a ptype semiconductor. This causes a discontinuity of the Fermi level at the junction (EF goes down in energy due to positive voltage. Right: The currentvoltage characteristic of the Ohmic contact is linear.+n metal Ec . The currentvoltage characteristics of the Schottky contact is exponential. 5 a Schottky contact on an ntype semiconductor is given. Left (bottom) is the energy band diagram when no voltage is applied.Semiconductor Devices 2012 . For Schottky contacts there should be depletion of majority carriers at the junction. Notice the bands bend up (depletion of electrons closer to the junction). Semiconductorsemiconductor junction pn junction and many more 10 EE2 . metal EFm metal semiconductor junction l ntype . The electrons in the ntype material try to cross the junction towards the metal but are not allowed to go through the wide forbidden bandgap.
When a positive voltage is applied to the metal (V+) the Fermi level of the metal drops with a value equal to eV with respect to the Fermi level of the semiconductor (which node we have grounded and thus it doesn’t change). This is illustrated in fig.Semiconductor Devices 2012 . right: after contact. * How to interpret an energy band diagram? Energy band diagrams are great tools to predict the behaviour of the junctions under the influence of an externally applied voltage. Applying an external electric field will increase or decrease this potential barrier respectively inhibiting and allowing current to flow. The Schottky contact for the ntype semiconductor is characterised with a potential barrier for electrons (depletion of minority carriers at the junction). The work function difference between nSi and pSi causes a potential barrier between n and p Si when forming the junction. Left: before contact. 11 EE2 . This decreases the potential barrier seen by the electrons in the semiconductor that are electrostatically attracted to the metal. 8. The figure gives a Schottky contact on an ntype semiconductor. However not all electrons available in the semiconductor can flow through the junction to the metal due to remaining potential barrier.W0 Ecp Eg EF Evp ptype ntype Evn Ecn EF Ecp EF Evp ptype ntype Evn Eg eV 0 Ecn EF Figure 7: the energy band diagram of a pn junction.
The carriers with energy higher than the barrier will be able to “flow” across the junction and occupy free energy levels (states) at the other side of the junction. If the potential barrier is reduced more by applying a large V on the metal. n3 n1 n2 Only these electrons have sufficient energy not to have to go through the forbidden band. no electrons can travel through the wide forbidden band Ev Figure 8: Left bottom: energy band diagram of a metalsemiconductor contact under zero bias.Energy metal semiconductor V+ Builtin voltage Potential barrier Ec EF EFm Ec Density of electrons at each energy value. This will cause a small current to flow. Right bottom: energy band diagram under bias.this gives the possibility for two types of currents to flow in different junctions – drift and diffusion currents. Right top: circuit connection. Left top: circuit connection.4 Conclusions Semiconductors have two types of free carriers . This is because these carriers would flow into the forbidden gap where there are no energy levels to receive them. Potential barrier to electrons intending to traverse to the metal distance Forbidden band. The potential barrier (increase in Ec towards the junction) blocks all carrier transport of carriers with energies lower than the barrier. Lots of electrons close to Ec : n2 > n1 > n3 but the density is exponentially decreasing with increasing energy.Semiconductor Devices 2012 . Thus the electrostatics imposed by the +V voltage attracts the electrons. 12 EE2 .electrons and holes . however the quantum mechanics presented by the energy band diagram shows that there is a potential barrier that blocks a certain amount of available charges. electrons are flowing through the closed circuit. then the current will increase exponentially. The density of electrons as a function of energy is also given. Energy band diagrams are a handy way to predict the expected magnitude of the current in junctions under the influence of an external bias without the need for full scale device modelling. Few electrons>small current. 1.
* Band diagram of unbiased junction 7 See more on generation of carriers later in this chapter.2. The reverse bias leakage current is limited by the amount of holes available and/or generated7 in the ntype region and the amount of electrons available and/or generated in the ptype region. Since under normal circumstances there is a limited number of minority carriers available.1 Review – the short pn diode In EE1 the operation of the special case of a short diode was discussed. the reverse leakage current is small. electrons are injected from the nregion into the pregion and holes are injected from the pregion into the nregion due to the fact that the externally applied potential lowers the energy barrier between the n and p type region. are smaller or equal to the diffusion length of the minority carriers. The definition of short diode is that the width of the materials that make the pn junction. The assumptions that were made to derive the current are: the contacts are ideal and do not show any band bending the contacts can absorb and generate the necessary carriers to keep bulk condition carrier concentrations fixed no voltage is dropped across the neutral regions the applied voltage drops across the depletion region only the amount of injected carriers across the junction due to the applied voltage is negligible compared to the majority carrier concentration determined by the doping. We found that the currents in a pn diode are governed by diffusion of minority carriers in the neutral regions (the neutral regions are those where the sum of all charges is zero. This injection of carriers causes a minority carrier gradient of electrons in the ptype region and a hole gradient in the ntype region. 13 EE2 . Gradients of carriers cause diffusion current.Semiconductor Devices 2012 . As a consequence the number of minority carriers in the neutral region varies linearly. The pn diode 2. Similar gradients occur in reverse bias but they are a lot smaller because they are minority carriers rather than majority carriers that are injected across the barrier. Under forward bias. thus outside the depletion region). with a slope determined by 1) the injection of minority carriers across the junction 2) the neutrality condition at the contacts (here we made the assumption that the contacts only impose the bulk conditions on the carrier concentrations).
+ depletion region n p n p ve acceptor builtin +ve donor ions Efield ions Figure 2: Left: p and n material before contact. W0 is the depletion width when no external bias is applied. Right: pn junction with 2 neutral regions (n & p) and the space charge region. ND+). The doping atoms are fixed in the lattice and thus cannot move and as a consequence cannot carry current. The charge separation (the ionised charges in the depletion region) is caused by the internal electric field. V0 (also called Vbi) is the builtin voltage. It contains 2 sides with fixed ionised charges of opposite charge at each side. This region is thus more resistive than the neutral regions.+ .+ . In the “depletion approximation” we assume that the depletion region is empty of moving carriers.W0 Ecp Eg EF Evp ptype ntype Evn Ecn EF Ecp EF Evp ptype ntype Evn Eg eV 0 Ecn EF Figure 1: The energy band diagram of a pn junction.resp. Left: before contact. right: after contact.Semiconductor Devices 2012 . kT N D N A V0 = ln e ni2 * Space charge region or depletion region W0 . The internal electric field is due to that builtin voltage Vbi. V0 is generated by the internal electric field that is a consequence of carrier diffusion upon contact. * The pn junction under forward bias 14 EE2 . it represents the potential barrier that controls the amount of carriers that diffuse. The fixed ionised charges are the ionised doping atoms (they have lost a hole or electron → NA. the builtin voltage is a result of the contact potential difference = work function difference between n and ptype Si.
Since the number of minority carriers is low. the current is small and determined by the minority carrier concentration rather than the electric field.V depletion regions Wp ++ ++ ++ ++ ++ Wn free electrons e (V0 . Bottom: energy band diagram of the pn diode under forward bias Forward bias lowers the potential barrier across the junction with a value proportional to the applied voltage. therefore more carriers will have an energy higher than the barrier and cross the junction. V ptype neutral region Wp + + + + + ++ ++ ++ ++ ++ Wn ntype neutral region Ec free electrons e (V0 + V) EF Ev eV + + + free holes Figure 4: Top: reverse bias condition.V) ptype neutral region ntype neutral region fixed acceptor ions fixed donor ions Ec EF Ev eV + + + free holes Figure 3: Top: forward bias condition. * The pn junction under reverse bias The reverse bias adds the absolute value of the applied voltage to the builtin voltage increasing the potential barrier. therefore there are no carriers with sufficient energy to travel across the barrier. Current is flowing and is increasing exponentially. Thus the offcurrent in a diode is small and constant as a function of reverse bias voltage when no break down mechanisms occur. Bottom: energy band diagram of the pn diode under reverse bias 15 EE2 . There is drift of minority carriers across the junction.Semiconductor Devices 2012 .
Semiconductor Devices 2012 8 .type region at depletion region edge eV With the injected concentrations given by: n 'p n p0 ' pn = = e kT pn0 (1) with n’p and p’n respectively the minority electron concentration in the pregion at the depletion region edge and the minority hole concentration in the nregion at the depletion region edge (injected minority carrier concentration). These excess carriers will cause the delays when switching pn diodes. It is a bit cheeky. It is important to remember that under certain bias conditions a certain amount of excess carriers is stored in the n and p regions between the edge of the depletion region and the contact (neutral regions). To derive the expression for the current. Remember than in the ntype region a hole diffusion current exists. Under forward bias the excess minority carrier concentration at the edges of the depletion width are given as a function of the bulk minority carrier concentration and the applied voltage. The boundary for the definition of short diode is that the length of the region is exactly equal to the diffusion length of the minority carriers which is Lp. The excess carrier concentration at the edge of the depletion width (thus at x=wn or –wp) is given by: ∆n p = n 'p − n p 0 excess electron concentration in p . wp 0 wn x np Short diode n’p ∆ np np0 ∆pn pn0 0 pn p’n x Ln 0 Lp x Figure 5: Top: Short pn diode under forward bias with exaggerated depletion width.* Minority carrier distribution under bias and current. it is best to refer to the graphical picture of the device under bias as given in fig. The excess of carriers is defined as the surplus (positive or negative) of minority carriers with respect to the initial minority carrier concentration that is determined by the bulk material. Shaded regions represent the amount of stored excess minority carrier charge8. More on this topic later. The choice of taking the maximum value of the lengths for deriving the short diode current is that this gives exactly the same expression for the current as in the case the short diode approximation cannot be applied (see next). np0 and pn0 are the intrinsic minority carrier concentrations. Note that looking at forward bias only. 5. is sufficient to derive the expression of the current in both bias regions.type region at depletion region edge ' ∆p n = p n − p n 0 excess hole concentration in n . Note that at the contacts the excess carrier concentration is zero. This seems contradictory but isn’t from a physics point of view. Note that the length of the ntype region is given as Lp. This condition is imposed by the contacts. but now you know why. Bottom: excess minority carrier distribution of short diode. 16 EE2 .
The magnitude of the diffusion current of the electrons in the ptype region is equal to the magnitude of the electron drift current in the ntype region. The excess carrier concentration is given by the shaded triangles in the neutral regions (Fig. 9 See first year course notes.Semiconductor Devices 2012 .5). 17 EE2 . Calculation of the diode currents in a short diode is easy. Since gradients in carrier concentration occur due to the injection process across the junction. This is illustrated in the picture below (fig. Remember that there is also a majority carrier flow as the majority carriers injected across the junction have to be resupplied9. diffusion of minority carriers will happen. 6). (1) and bulk conditions kept at Ohmic contacts.In a short diode (the lengths of each neutral region is smaller or equal to the minority carrier diffusion length) the minority carrier concentration is decreasing linearly in each region. The density of minority carriers is imposed by the boundary conditions: eq. Remember the expression for the diffusion current: dn p dp − Dp n eA Dn dx dx n' p − n p 0 p' − p − D p n n0 eA Dn 0 − ( − Ln ) 0 − Lp n' p − n p 0 p'n − pn0 eA D + Dp n Ln Lp eV eV = e kT − 1 e kT − 1 eA n p 0 Dn + p n0 D p Ln Lp n D p n 0 D p eV eA p 0 n + e kT − 1 Ln L p eV kT − 1 I s e I tot An interesting problem to think about is: what happens when one or both of the materials forming the pn diode are shorter than the diffusion length (Ln and Lp) of the minority carriers (but longer than the depletion width in each region)? Thus do the derivation of the short diode current for a length x n<Lp for the nregion and xp<Ln for the pregion and compare the result taking into account the given inequalities. Luckily the rule that the total current is constant through the device makes calculation of the minority carrier diffusion current in each region of the pn junction sufficient for deriving the total current in the pn junction. the latter current is the resupply of the injected carriers across the junction. This is similar for holes.
In this case an electron and hole will recombine and thus get fixed to the atom and become unavailable for conduction. the length of the materials forming the junction is not necessarily smaller or equal to the diffusion length of the minority carriers. An example is light: shining light on a semiconductor. the number of minority carriers varies nonlinearly in the neutral regions of the device. Figure 6: The contribution of the minority and majority carrier current to the total current through the pn junction. recombination is also continuously happening. As a result. The minority carrier gradients determine the total current completely. with the correct energy. called traps – these will always be available as a 100% perfect semiconductor does not 18 EE2 . thus electrons and holes are freed from the atoms and become free carriers that can be involved in the conduction processes. This is a reduction of the number of free carriers. larger than the energy gap.Semiconductor Devices 2012 . 2. In this case there are impurity centres or lattice defects in the semiconductors. In the case of long diodes recombination of minority carriers will happen whilst they are diffusion to the contact. The second type is indirect recombination. can also release electronhole pairs. This process is called generation of carriers and is happening continuously as long at the temperature is sufficiently high to break the valence electronatom bonds. As with generation.current p region Itot Ip In + Depletion region nregion junction contact Minority carrier diffusion currents Majority carrier drift currents contact Remain constant in space charge region. The opposite process is called recombination. The first is direct recombination of electron and holes which means that excess electrons in the conduction band recombine with holes in the valence band. Majority carriers injected across the junction are resupplied via drift through the contact and diffuse to the other contact through the minority carrier gradient region. Energy is lost by the electron and is given off as photons (light). Recombination can happen in two ways.2 Generation and recombination – long pn diodes In real diodes. The current remains constant in the space charge region because we assume at the moment that there is no recombination no generation of carriers. Of course other forms of energy can do the same. * Generation – recombination We have seen that due to thermal energy electronhole pairs are created. In what follows we will study its impact on the device characteristics.
Thus some do recombine but we neglect this. The values are dependent on the material. It is the gradient of the excess minority carriers that is significant. whilst being insignificant for the majority carriers in comparison with the amount available. the energy released by the electrons causes heating of the lattice (Heating causes lattice vibrations. This is a statistical average. The relationship between the carrier lifetime and diffusion length is given by the diffusion constant following: Ln. This is of course not completely correct as charge neutrality is a fundamental rule. 12 This assumption will actually break down the rule on charge neutrality in the “neutral region”. Since the density of the excess carriers is small compared to the density of the majority carriers.12 It is indeed an approximation.p the number of carrier have been reduced by a factor Exp(1).Semiconductor Devices 2012 10 . When the voltage across the junction is not too high we can assume that this excess is a lot smaller than the number of majority carriers in that region. p The intrinsic number of free electrons and holes is the same as carrier generation and recombination is in equilibrium. Thus the rate of recombination and generation will adapt itself to the availability of excess minority carriers in the “neutral regions”11. But the assumption will allow us to derive analytic expressions that still describe the conduction through the device with reasonable accuracy. that’s why it is written between quotes. in the short diode to assume no carriers recombine within the neutral region as the diffusion length is a statistical average. These parameters give the statistically average time. 19 EE2 . Recombination becomes important for devices where the neutral regions are larger than the diffusion length10. We will assume that the injected minority carrier concentration is sufficiently small compared to the majority carrier concentration such that the variation of majority carriers can be neglected. In this case generation of minority carriers is larger in order to maintain equilibrium conditions. Thus the generation rate of minority carriers will then be smaller than the recombination rate. although acceptable. thus means that not all carriers will recombine within that time scale. The traps will cause some discrete energy levels to exist within the bandgap of the semiconductor which can have empty states to which the electrons from the conduction band can fall in order to cross the bandgap in steps. However when there is a shortage of minority carriers then the generation rate increases to try to bring the minority carrier concentration back to equilibrium conditions. In a pn diode under forward bias an excess minority carrier density exists in the neutral regions of the device. If we put the pn diode under reverse bias then there will be a reduction of minority carriers in the neighbourhood of the depletion region edges (junction).p. When there is an excess of minority carriers. Recombination will happen between the majority carriers and the excess minority carriers. These vibrations can be seen as particles called phonons with which the carriers in the lattice can interact – scatter. The characteristic timescale involved in the recombination processes is called the carrier lifetime τ n. respectively distance that a carrier travels before it recombines. As a consequence the majority carrier recombination or generation rate is equal to that of the minority carriers. They can also be regarded as waves with certain energy and wavelength). 11 What we call neutral is actually quasi neutral under the given assumption. The characteristic length scales are the diffusion lengths Ln. the quality of the material. In this case the generation rate of minority carriers is larger than the recombination rate. p = D n. In this case. Of course when minority carriers recombine they do so with majority carriers. the doping density and the type of charged carrier. then the recombination rate will increase proportional to this excess in order to drive the semiconductor back to equilibrium.p.exist. recombination processes can have a significant impact on the number of excess carriers. p ×τ n. but the majority will such that at x=Ln.
loss of holes due to recombination. as a function of time and place. Excess means the density of carriers larger than the minority carrier concentration defined by the doping of the bulk material only. without recombination it is linear. t ) ∂t = x → x + ∆x 1 J p ( x) − J p ( x + ∆x) δp n ( x. In order to do that we take a volume of semiconductor material through which a hole diffusion current is flowing as given in fig.Semiconductor Devices 2012 . t ) δp ( x. In order to calculate currents in the long diode. t ) ∂δn( x. For ∆x→0 the equation (3) becomes a differential equation: ∂p( x. The first step is thus to derive the expression of this variation in the case recombination happens13. x x x+∆ x Figure 7: Hole current entering and leaving a volume ∆x A. ∂p n ( x. t ) − e ∆x τp (3) In words this equation means: The rate of variation of the hole concentration = increase of the hole concentration in volume ∆x ×A per unit time . t ) ∂δp( x. The second term is the carrier recombination term given in eq. We look at the variation of the density of holes. 7. t ) δn( x. over an infinitesimal distance ∆x. t ) = = − ∂t ∂t e ∂x τn 13 (4a) Similar for electrons: (4b) Remember. we need to know the variation of the minority carrier excess in the neutral region. This equation will give us the continuity equation for holes. t ) 1 ∂J n ( x. It gives the rate of change of the hole density while a current is flowing. (2). It is also this equation that will be important to analyse the transient processes in bipolar devices under large signal switching conditions. A Jp(x) Jp(x+∆x) propagation direction.The minority carrier recombination rate in the “neutral region” can be described to acceptable approximation by the simple equations: δn p Un = τn δp Up = n τp (2) Recombination in a depletion region or in those cases where the minority carrier concentration is of the same order of magnitude as the majority carrier concentration have to be described by different equations than those given above. t ) = = − ∂t ∂t e ∂x τp ∂n( x. 20 EE2 . The term δpn is the excess minority carrier concentration. t ) − 1 ∂J p ( x.
Remember that these equations are for the minority carriers! 16 MathematicaTM was used for this extraction.Semiconductor Devices 2012 14 . which is a second order differential equation. 21 EE2 . we find the expression16 of the excess hole concentration variation as a function of position: This is a valid assumption as we are dealing with minority carriers and we have assumed that the injection rate is small.These equations are referred to as the continuity equations for the holes resp. Xn δn = ∆n & δp = ∆p δn = 0 & δp = 0 (8) These boundary conditions mean that x=0 is the point of steady state carrier injection (depletion region edge) due to the applied DC voltage across the junction.g. If we can assume that drift is negligible14 then the current is solely diffusion and the current densities can be written as: J n = eD n ∂δn p ∂x ∂δp n J p = −eD p ∂x (5) Substituting (5) in (4)15 gives the continuity equations of minority carriers taking recombination into account: ∂δn ∂ 2δn δn = Dn − ∂t ∂x 2 τ n ∂δp ∂ 2δp δp = Dp − ∂t ∂x 2 τ p (6) This equation is of utmost importance for analysing the switching characteristics of bipolar devices. across a forward biased pn diode in DC. Using the sinh function as possible solution. the electrons. 15 Note that the subscript n and p are omitted. can be written as the sum of two exponentials: δp( x) = C1 exp x + C 2 exp − x or using a hyperbolic sine function: δp( x) = C1 sinh x + C 2 Lp Lp Lp with C1 and C2 integration constants that must be determined by the boundary conditions (8). the time derivative is zero and in steady state the continuity equations become: ∂ 2δn δn δn = = 2 2 Dnτ n Ln ∂x ∂ 2δp δp δp = = 2 2 D pτ p L p ∂x (7) Solving this equation with boundary conditions: @x=0 @ x = X p. So the absolute number of minority carriers is small and as a consequence the drift minority carrier current will be negligible. The solution of (7) for holes. * minority carrier variations and currents In case we have steady state carrier injection as e. The second boundary condition gives that at the contacts the excess carrier concentration needs to be zero.
Semiconductor Devices 2012 . (9) and for the long diode approximation we take the limit of Xn → ∞. Therefore we want to simplify this equation for two special cases: the short diode approximation and the long diode approximation. For the short diode approximation we take the limit of Xn << Lp for eq. In order to calculate the current we have to take the first derivative of (9). 22 EE2 . This gives for the hole current: Ip = eD p A∆p Lp X coth n Lp (10) The following figures in Fig. In the following we will evaluate what the errors are that we make by taking these approximations. useful for numerical simulations only but not for pen and paper work. 8 give the difference in excess minority carrier concentrations between the short.δp( x) = ∆p − Xn sinh Lp x − Xn sinh Lp (9) This is a rather complex equation. For this we need to calculate the current based on (9) and that based on the limits of short and long diode approximations. long and correct diode descriptions.
For short Xn (a).Semiconductor Devices 2012 . the distance to the contact varies: Xn= (a) 20. The minority carrier diffusion length is Lp=200 nm. The error in the long diode approximation is in the nonzero excess at the contacts. In the other two situations some error is made. the short diode approximation describes the excess carrier concentration to excellent accuracy. When Xn=Lp. the excess carrier concentration is best described by the short diode approximation (smallest difference in area underneath the curves and correct contact boundary condition). the linear approximation becomes less accurate and the exponential equation more accurate. When Xn>Lp but of the same order of magnitude. 23 EE2 . For long Xn (d) the long diode approximation is excellent. (b) 200. (c) 400 and (d) 1000 nm.Normalised minority carrier concentration (a) (b) Normalised minority carrier concentration Correct solution Short diode approximation Long diode approximation Normalised minority carrier concentration Distance (nm) Normalised minority carrier concentration Distance (nm) (c) (d) Distance (nm) Distance (nm) Figure 8: The excess minority carrier hole concentration in the nlayer under normalised injection conditions for the different solutions of the continuity equation and for different relative differences between contact position relative to the minority carrier diffusion length.
We see that if the length of the diode layer is more than twice that of the minority carrier diffusion length. This is what is referred to as the SHORT diode approximation. 10 9 8 7 Ireal/Iapprox 6 5 4 3 2 1 0 0 1 2 Xn/Lp 3 4 5 Ireal/Iexp Ireal/Ilin Figure 9: The ratio of correct current to approximated current. Obviously for Xn<Lp the short diode approximation is best. the error of using the short diode approximation becomes negligible small when Xn<<Lp. for Xn>Lp the error made by using the long diode approximation decreases quickly and it is better to use the long rather than the short diode approximation. (7). pink: short diode approximation. for Xn=Lp the error using the short or long diode approximation is exactly the same 17.Semiconductor Devices 2012 .p = ∞ in eq. Thus Xp < Ln and Xn < Lp. the Ohmic contacts are at a distance shorter than the minority carrier diffusion length. the error becomes negligible. Blue: long diode approximation. where Xp is the position of the contact of the player and Xn is the position of the contact in the nlayer. 3. In this case we neglect any recombination by setting τ n. 17 This is obvious from the equations derived in this case for the short and long diode approximations.When looking at the currents. In figure 9. 2. Thus we use the short as it is easier. Then under the boundary condition (8). (7) 1. we can conclude why the use of the approximations is acceptable in different regions. the ratio of the correct current to the approximated current is given as a function of the ratio of the lengths. the excess minority carrier concentration varies linearly. * Simplified solution of eq. 24 EE2 . Interesting observations are: 1. Short diode approximation In this case.
20 Scattering is caused by interaction of the carriers with Si atoms. the Ohmic contacts are at a distance larger than the minority carrier diffusion length. p(x)= p0+ ∆p e(x/Lp) Excess carrier concentration is forced to zero at contact. Thus in the case of diffusion with recombination the excess carrier concentration is varying exponentially in a long diode in contrast to the short diode where the excess carrier concentration is varying linearly. In this case we assume that the contacts appear at a distance equal to infinity: Xp = Xn = ∞. As a consequence the minority carrier diffusion current calculated for holes is: J p ( x) = q Dp Lp ∆p e −x L p (12) Thus this current is decreasing for positive x (see fig. Note that the variations of both minority carrier concentrations are calculated in the positive xaxis. Solving (7) in the long diode approximation we impose that the excess minority carriers have completely disappeared by recombination. 10) because holes are disappearing by recombination! But the total current needs to be constant. n0 (p0) is the carrier concentration determined by the doping of the bulk material. ∆n (∆p) is the excess carrier concentration at x=0 resulting from the carrier injection process. which is the average distance a carrier diffuses (travels) before recombining. δn (δp) is the xdependent excess carrier concentration.Semiconductor Devices 2012 18 19 . This is called the LONG diode approximation. This can be obtained by remembering that total current consists of Note that we are ignoring the signs. The difference is that now this minority carrier excess is changing exponentially. 10. This is illustrated in fig. Thus Xp >> Ln and Xn >> Lp. Note that we are talking about minority carriers! As in the short diode. 25 EE2 .18 In that case the solution to (7) is an exponential. ∆p p0 Lp Xn x Figure 10: the variation of the excess carrier hole concentration as a function of distance due to recombination (full line).2. The solution for the excess carrier concentration is then: δn = ∆n e δp = ∆p e −x L n −x L p excess minority carrier concentration19 (11) With L = Dτ the diffusion length. τ is the average time between scattering20 events. x=0 is the edge of the depletion region. the current is still determined by the diffusion of the excess minority carriers in the neutral region. Signs will change for pn junctions. Long diode approximation In this case. impurities (doping and others) and crystal imperfections.
Thus we see that if minority carrier diffusion current is decreasing when going away from the point of carrier injection. As a consequence it is the diffusion of minority carriers that limits the total current. 11. This gives the first approach of finding the expression of the current in a long pn diode. there must be a drift current of majority carriers that is resupplying the recombined carriers and resupplying the carriers that are injected across the junction. The sum of the two is constant: Itot. therefore the total current must be determined by the maximum electron and hole diffusion currents that can happen under the given injection conditions (determined by the applied voltage) The maximum diffusion current that can flow can be calculated at the point of injection at both sides of the junction (edge of the depletion width in the pn diode) as there the gradient of the exponentials is maximum. wp 0 wn x Long diode np n’p ∆ np np0 pn p’n ∆pn pn0 0 x x 0 Figure 11: Top: pn diode under forward bias with exaggerated depletion width. Since we have neglected the variation of majority carriers. Bottom: excess minority carrier distribution of long diode. see fig. Shaded regions represent the amount of stored excess charge. This means that we determine the excess carrier gradient at x=0 in each region: dn p dp eA Dn − Dp n dx x = 0 dx x = 0 dδn p dδpn eA Dn − Dp dx x = 0 dx x = 0 − x / Lp d ( − ∆n p ) e − x / Ln d∆pn e I tot = eA Dn − Dp dx dx x =0 x =0 ∆n p ∆p eA Dn + Dp n Ln Lp n' p − n p 0 p' − p eA Dn + Dp n n0 Ln Lp 26 EE2 . Calculation of the diode currents in a long diode is based on calculating the maximum diffusion current at the edge of the depletion region width.hole and electron movement. The sum of this majority carrier drift current and minority carrier diffusion current is constant.Semiconductor Devices 2012 . the majority carrier drift current in decreasing when going towards to point of injection.
Note I p is a minority carrier diffusion current in the nregion and a majority carrier drift current in the pregion (similar for In). Each component is not constant throughout the device but the sum of electron and hole current is constant. p n Itot Ip Itot=In + Ip In In Ip Figure 12: The variation of the magnitude of the hole and electron current in a long diode. Compare this to the short diode where all current are constant throughout the device. The resupply can be allocated to drift as the gradients of the majority carriers are so small that it would not be sufficient for resupply. This is only the case when for the short diode we take the lengths of the neutral regions to be exactly the same as the diffusion lengths of the minority carriers. majority carriers need to be resupplied. The electric field can be very very small as long as the number of carriers is large (which is the case for majority carriers) to make a sufficiently large drift current. which causes the current. There is another way to calculate the currents in the pn diode that will come in handy when discussing the 21 Nor a gradient in majority carriers 27 EE2 . Figure 12 gives the electron and hole components of the currents in a long pn diode. Thus although there is a small electric field across the neutral regions it still is negligible compared to the electric field dropped across the depletion region.I tot eV eV kT e −1 e kT − 1 + pn 0 D p eA n p 0 Dn Ln Lp eV n p 0 Dn pn 0 D p kT e − 1 = eA + Ln L p eV I s e kT − 1 Note that the result we have obtained for the long pn diode current is the same as the short diode. For shorter diodes. Thus our initial assumptions that there is no electric field 21 in the neutral regions is not quite correct otherwise there would be no drift current for the resupply of carriers.Semiconductor Devices 2012 . the values of Ln and Lp will have to be changed by the actual widths of the neutral regions and the current will thus increase in shorter diodes. The majority carrier component occurs as a result of recombination between minority and majority carriers.
13. The current that resupplies these recombined minority carriers is given by: eALn ∆n p Q In = n = τn τn (14) Q p eAL p ∆p n Ip = = τp τp Thus the current can be expressed as the stored minority charge in the neutral regions divided by the lifetime of the minority carriers in this region. Filling in the expressions for the excess carriers and the diffusion constants will give the same current equations as extracted before for the long diode. In the case where recombination of the excess carriers occurs we know that this recombined charge needs to be resupplied. Region a The assumption in the ideal pn diode is that no recombination happens in the depletion width. hole concentration in resp the p.4 Nonidealities in pn diodes When measuring a Si pn diode and plotting the current on a log scale we get a graph as in fig.p the charge Qn. Under the assumption that the exponential variation of the excess carrier concentration is the solution of the steady state continuity equation we have to make sure that the excess carrier concentration at the contacts is zero. n type region. The density of the excess charge is given by the shaded area underneath the curves for δn and δp.p the excess minority carrier electron.p disappears via recombination and this charge needs to be resupplied in the same time scale in order to maintain steady state.Semiconductor Devices 2012 . These boundary conditions represent the position of the contacts under the long channel approximation22. To that aim we quickly look at this second method for calculating the diffusion currents in a diode. 11 note that there is excess minority carrier charge in the neutral region (this is indicated by the shaded regions). This is not completely correct. thus carriers travel through unhindered. If the depletion width is larger than the carrier diffusion lengths. especially in pin 23 diodes or diodes with long depletion widths. We know that current is a variation of charge as a function of time. This will affect the current at low current densities (small number of carriers in transfer). recombination of carriers in transit will occur. More carriers need to cross in order to obey the relationship for injected carrier concentration as a function of applied voltage. 2. To calculate the excess charge we integrate the excess minority carrier functions: Qn = −eA ∫ δn p dx −∞ 0 Q p = eA∫ δp n dx 0 ∞ (13) Note that the boundary conditions chosen for this calculation are taken to be infinite. We call Qn. Region b This region behaves as ideal diffusion and happens for intermediate voltages. With reference to fig. Every τ n. 22 23 This ensures a zero excess carrier concentration at the contact! ptype region + intrinsic region + ntype region 28 EE2 .switching behaviour of the pn diodes.
In the high carrier injection regime the relationship for the injected carrier concentration that we have derived in EE1 breaks down.Region c This is the region of high level injection. This brings changes to the majority carrier density which we have assumed to be constant in the ideal case. First (fig. Therefore an ideality factor n is introduced in the formula for the current in a pn junction that takes into account these nonideal effects. there is no easy solution. Note that the ideality factor n is called the emission coefficient in SPICE. We can look at two ways of switching.5 Switching delays in pndiodes The switching characteristics of pn diodes are all governed by EXCESS MINORITY CARRIERS. This is thus 29 EE2 . forward bias) to off (Vapplied=0V. A huge amount of carriers is injected across the junction and makes the number of minority carriers at the edge of the depletion region of the same order of magnitude as the majority carriers. We will see that removal of excess carriers. keeping the current lower than expected. n=1 in the ideal region and goes towards 2 in the recombination and high level injection region. Ln(I) ideal real c) b) a) d) V Figure 13: The difference between an ideal and a real IV curve of a Si pn diode. when the diode switches from forward bias to open circuit. Region d At still higher voltages or for large ohmic contact resistances a large amount of the applied voltage is dropped across the ohmic contact rather than across the depletion region and thus the intrinsic diode feels less voltage applied. The amount of which is a function of the applied bias – the larger the current that is flowing the larger the excess charge. 2. one can switch the diode from on (V applied>0V. Its effect is a lowering of the total current flowing. occurs via recombination only.15). by opening the switch such that I=0A).Semiconductor Devices 2012 . In order to change the current. The expression for the current then becomes: I tot eV = I s e nkT − 1 With 1 ≤ n ≤ 2. the excess charge stored should change before the voltage across the diode junction can change. This is inherently a capacitive effect. Under bias there exists an excess minority carrier concentration inside the neutral regions close to the depletion region. Although all these effects can be calculated.
Semiconductor Devices 2012 . This gives the expression for the time variation of the current as a function of the stored charge in the neutral 30 EE2 . τ n and τ p are the minority carrier life time of electrons and holes respectively. t ) A+ ∫ A τp ∂t 0 Remember that the diffusion current is zero at the contact and maximum at the edge of the depletion region (x=0) and that the integral of the excess minority carrier concentration in the neutral region multiplied by unit charge e and cross sectional area A gives the total stored charge Qp in that region (see eq. Repeating equation (4a): ∂δp( x. Although the current drops immediately to zero. Change of excess in function of position. 15 illustrates a switching experiment of a p+n diode (current mainly governed by holes) from forward to zero bias via a switch that opens at t=0. The state of the diode cannot change before the charge is removed.governed by the carrier lifetimes. multiplying by cross sectional area A and integrating over the neutral region: contact ∫ 0 − ∂J p ( x. v(t) across the diode remains until all excess charge is removed. t ) = − ∂t e ∂x τp t Reorganising the terms. electron). Right: dashed lines give gradient of excess charge at the edge of the depletion region. t ) eδp( x. t ) δp ( x. i(t) + e(t) t=0 v(t) vL(t) RL I I I τp Qp(t) δ p t xn Current through diode. That means that there will be a voltage across the load RL until excess charge removal is complete. t ) − 1 ∂J p ( x. Fig. 13). Figure 15: Top: switching circuit. thus it will take a time equal to τ n and τ p to reduce the excess carrier concentration by a factor e. Bottom: Left: upon switchoff the current through the diode drops to zero but the excess charge remains and decays exponentially via recombination only (middle). The carrier lifetime is the average time an electron (resp. t ) ∂x A= contact ∫ 0 contact ∂ eδp( x. the voltage. hole) spends in the material before it recombines with a hole (resp. Decay of stored charge. When the current is 0 the gradient is 0.
At t=0+25 the current drops to zero (open circuit) and remains zero.Semiconductor Devices 2012 24 25 . 26 Note that these switching equations can also be solved with the methods introduced in the Signals and Linear Systems course this year. the charge cannot follow immediately. Thus at t=024 the excess minority carrier charge is given by: Qp=I0 τ p. there exists a voltage v(t) across the diode. Thus (15) becomes: Q p (t ) dQ p (t ) =− τp dt Integration over time gives: t t dQ p (t ) dt = −∫ ∫ Q p (t ) 0 τ p 0 ln ( Q p (t ) ) = − Q p (t ) = C e t +c τp −t / τ p With c and C integration constants. For t≤0 the DC forward bias imposes a current I0 through the device. a solution to equation (15) can be easily found. the gradient at the depletion region edge of the minority carrier density needs to be zero too! So there the line for the excess carrier concentration is horizontal and not exponential. we all know from the injection of carriers relationship (1): e v (t ) ' ∆p n = p n − p n 0 = p n 0 exp − 1 (17) kT With v(t) the voltage across the diode. Since Qp(t=0+)= I0 τ p the integration constant C= I0 τ p. 31 EE2 . 27 Since the current is zero. Thus as long as ∆ pn is nonzero. we An infinitesimal amount of time before switching. We need to find the relationship between the voltage across the diode and the minority carrier excess charge. This relationship. In the special case of switching the p+n diode from ON to zero. An infinitesimal amount of time after switching. Thus the variation of the excess minority carrier charge as a function of time is given by: −t / τ Q p (t ) = I 0τ p e p (16) The proof that the voltage remains across the diode goes as follows. These form the boundary conditions to solve (15)26. If we ignore the distortion of the exponential variation of the excess charge at the depletion region edge27. (15) states that the instantaneous current in a switching device ip(t) is equal to the resupply of the excess carrier concentration that is disappearing as a result of recombination (this is the DC component of the current – 1st term RHS) plus the AC component that allows a buildup or depletion of charge as a function of time (2nd term RHS).region: i p (t ) = Q p (t ) τp + dQ p (t ) dt (15) Eq.
e(t) R e(t) +E p+ n i 0 E t T Figure 16: Circuit and input square wave. R is the load resistance. The second method to switch the diode is to drive the excess carriers from the neutral region via a current. The time period T has to be large enough to allow the diode to switch off completely in each cycle. 32 EE2 . Thus: • the excess charge dies out exponentially via recombination as a function of time • the voltage across the diode junction cannot change instantaneously but has to wait till the excess charge has disappeared in order to establish the correct depletion layer for a reduced voltage. The magnitude of the switchoff voltage in the given case is equal to the magnitude of the on voltage.Semiconductor Devices 2012 . Substituting this in (17) and extracting v(t) gives: v(t ) = kT I 0τ p −t / τ ln e p − 1 e eAL p p n o (19) In order to improve switching speed in this situation. τ p. This will give rise to the reverse recovery transient. (16) in (18) gives ∆p n (t ) as an exponential function of time. (11). The stored charge at any time is then given by (13): Q p = eAL p ∆p n (t ) (18) Substituting eqs. The physics associated to this process is represented in the drawings below. Another way to improve the switching time is to switch the diode from ON to OFF by applying a negative (reverse bias) voltage during off switching. recombination centres are added to decreases the carrier life time. then charge will be removed via recombination and diffusion currents. On the other hand one can also make a p+n diode with a short p+ region. In this case a current will be flowing during the switching and thus the excess charges will be removed via a combination of diffusion and recombination.can still use the pure exponential relationship of the excess carrier concentration as given by eq. smaller than the recombination length in order to limit the excess carrier buildup.
Bottom is the variation of the voltage across the diode as a function of time. The forward bias current will be mainly determined by the bias circuit because the resistance of the forward biased diode is very small. At ON: the current is positive. the voltage drop across the diode is negligible compared to that across R.Semiconductor Devices 2012 28 . Thus the load resistance will limit the current. The on state excess carrier concentration is a lot larger than the offstate excess carriers concentration. Note the change in the gradient of the excess minority carrier density at the moment of switchoff (below the xaxis). at the moment of switch OFF the current becomes negative. At t<0 the diode is ON and a current If=E/R is flowing. The dashed line with the arrows is the currenttime diode characteristic whilst switching.δ p If → gradient<0 Ir → gradient>0 x t Figure 17: Excess hole concentration distribution in the nregion as a function of time during the transient. t sd is the storage delay time. The graphs at the right are the currents and voltages measured across the diode as a function of time. This current determines the slope of the excess carrier concentration at the depletion region edge. 33 EE2 .28 If I If V v(t) Ir t v(t) t Ir tsd i(t) tsd t E Figure 18: Top: the full line is the diode currentvoltage characteristic. We know that this slope is consistent with the diffusion of minority carrier holes in the ntype region from Note that the on and off state excess carrier concentration is not drawn to scale. As a consequence.
thus Q p=0. As a consequence. The current is decreasing as only the intrinsic minority carrier density remains to carry current. This is maintained at steady state via carrier generation processes. When the diode was initially forward biased it had a narrow depletion region (that only allows a small voltage drop across it) and a relatively large density of stored minority carriers. The time it takes to remove all the excess charge is called the storage delay time tsd. we know from the previous switching experiment that the excess minority carrier charge that is stored in the neutral regions will not disappear immediately. The reason why this happens is the following. Remember that the depletion region is a region without free charged carriers. forms the depletion region.the depletion region edge to the contacts (similar for electrons in the ptype region).Semiconductor Devices 2012 29 . The final current will be determined by the small amount of minority carriers in the neutral region near the depletion region edge. Ir is flowing. 18. the slope of δp at the depletion region edge remains constant. Thus as long as excess carriers remain. which is consistent with the complete removal of excess charge. This point is where v(t) – the voltage across the diode junction . This indicates that minority carriers have to flow the other way than under positive bias29. Then a negative current Ir=E/R will be flowing that is consistent with this voltage. we first have to deplete the regions around the junction of the excess minority carrier charge. Once that is done. When the excess is removed (mainly by diffusion but also by recombination) then the depletion region will be able to build up. Since a negative current. This negative current is many times larger (in absolute value) than the reverse bias current that is normally flowing through the pn diode in DC reverse bias. The OFF current is thus the leakage current through the reverse biased diode. The current is thus determined by load resistance R and applied voltage –E. Doing this. Thus minority carrier excess holes are diffusing from the ntype region to the ptype region and electrons from p to n. the depletion region cannot increase and the voltage across the diode can also not increase immediately. When the excess is removed. In order to find the variation of the charge as a function of time. eq. At t=0 the excess carrier concentration at the edge of the depletion region is p’ n and at the end of the switch cycle the excess carrier concentration at the edge of the depletion region should have become p”n. The current through the circuit will then be more and more limited by the current that is allowed to flow through the diode. the region around the junction can be depleted of the majority carriers.becomes equal to zero. The current is large as long as it is determined by the external circuit and excess minority carriers are available. This current removes the excess charge. 34 EE2 . 18 is the point where the diode can finally go into reverse bias. (15) has to be solved with the boundary Remember that the direction of the diffusion current is given by the gradient of the minority carriers at each point x in the device. When the external bias is reversed. At t=0 the voltage polarity is reversed whilst the magnitude of the voltage is the same (this equality is not necessary and is here used for convenience). the slope of the excess carrier concentration at the depletion region edge is exactly opposite to the one in forward bias. the voltage drop across the diode is still small and negligible compared to the voltage drop across R. ' pn = pn0 e eE kT −eE kT ' p n' = p n 0 e This is clearly illustrated in fig. the depletion width adapts and an increasing voltage will be dropped across the increasing depletion width. Important to note when you look at fig. Thus we can see now that when the external bias goes negative. Thus in order to deplete the neutral regions.
rewriting (15) : dQ p (t ) dt dQ p (t ) dt = i p (t ) − Q p (t ) τp For 0 < t < tsd we know that ip remains constant at –Ir. (20) is valid for 0 < t < tsd only. Solving (15) for a p+n diode30 can be done by using Laplace transforms or as follows. 35 EE2 .conditions that for t<0 the charge is Qp=If τ p and at t=0 the current becomes I=Ir. thus: = −I r − Q p (t ) τp dt τp dt 0τp t t Separation of variables: dQ p (t ) Q p (t ) + τ p I r =− Integrate: Q p (t ) Q p (0) ∫ dQ p (t ) Q p (t ) + τ p I r = −∫ ln Q p (t ) + τ p I r ( ( )τ ) Q p (t ) pI f =− τp ln Q p (t ) + τ p I r − ln τ p I F + τ p I r = − ln ( ) t ( Q (t ) + τ I ) t (τ I + τ I ) = − τ p p r p F p r τp p t Q p (t ) = −τ p I r + τ p I F + τ p I r exp − τp ( ) or − t Q p (t ) = τ p − I r + I f + I r exp τ p ( ) (20) Eq. Qp(tsd)=0. The storage delay time tsd can then be extracted by stating that at t=tsd. Thus: −t 0 = τ p − I r + I f + I r exp sd τp ( ) The storage delay time is then (under the assumption of quasi steady state) : If t sd ≅ τ p ln 1 + Ir Switch from forward to reverse bias state: • reverse current larger than diodes’ reverse saturation current • transient current and voltage are delayed by change in stored charge • this causes a storage delay time tsd during which the current is larger than the reverse bias current 30 Current determined by the hole current.Semiconductor Devices 2012 . This approach will not give the time it takes the diode to buildup the depletion region from the “noexcess charge” condition. This approach will allow us to derive an expression for the storage delay time. The assumption that we are making is that the condition I=Ir is steady state until we reach t=tsd.
36 EE2 .associated to the DC characteristics • delays in pn diodes are a consequence of excess minority carrier charge As before.Semiconductor Devices 2012 . The use of GaAs (high mobility carriers and shorter lifetime than Si) results in ultra high speed diodes. The delay time can also be reduced by using a short device layer. The equivalent circuit is given in fig. Here only majority carriers need to be removed which is a fast process.g a metal with a larger work function than the work function of ntype silicon. Therefore for circuit design purposes an equivalent circuit can be used with component values dependent on the chosen DC point but further giving a linear response. There are plenty of them and we do not have to wait for generation of these majority carriers to establish resupply. their nonlinear characteristics can become cumbersome for fast circuit simulation purposes. Rs is the contact resistance. to improve the switching time in pn diodes one can introduce recombination centers such that the lifetime of the carriers reduces which makes the excess charge disappear faster. it is the majority carriers that move back towards the junction. Unfortunately this also increases the reverse leakage current. e. Rs r d = 1/gd C Figure 21: Small signal equivalent circuit of a pndiode. This depletes the majority carriers (electrons) in the semiconductor at the junction. No depletion of carriers occurs in the metal. 2. 21. For the diode we have seen that there is a resistive component and a capacitive component that are both a function of voltage. In order to switch the Schottky diode from on to off. Remember that the fast switching in the Schottky contact can be exploited to improve the switching behaviour in bipolar junction transistors too. rd is the diode’s differential resistance and C is the voltage dependent capacitance. Specially designed pn diodes which strongly conduct for a short period of time in reverse bias – the step recovery diode – can switch in the picosecond range and are used for harmonic generators. In a Schottky diode the junction is formed by a metalsemiconductor contact. The variations around this point are relatively small such that the device acts as if it is linear around this point.6 Small signal equivalent circuit of a pn junction diode If diodes are to be used as components in circuits. Other solutions to obtain fast rectifying devices include the use of Schottky diodes. Very often in circuit operation a device is functioning around a DC point (see analogue electronics course).
parallel plate capacitance) Diffusion capacitance in forward bias (current injection across junction) Reverse bias: Depletion capacitance εA C depl = Wdepl C depl = εA eN A N D 2ε ( N A + N D )(V0 − V ) Wdepl is the depletion width. 37 EE2 .Semiconductor Devices 2012 . One is in reverse bias where a depletion region with fixed ionised charges is builtup with an amount dependent on the applied voltage31.3 Semiconductor Devices course. The capacitance C is due to charging and discharging effects occurring during AC operation. This is no longer the case for a large 31 Review the calculation of the depletion width in your first year EE1. Forward bias: Diffusion capacitance dQ C diff = dV dQ p C diff = for p + n junction dV e C diff = Iτ kT The differential resistance rd relates to the slope of the IV characteristic. Easily calculated as a differential 1 dI e = gd = = I conductance: rd dV kT 2.The series resistance RS represents the nonideal characteristic of the Ohmic contacts. The other is in forward bias and takes into account the variation of the builtup excess charge as a function of voltage.7 Large signal equivalent circuit of a pn junction diode While a small signal equivalent circuit does not consist of nonlinear elements because due to the small variation in applied voltage the IV region used can be assumed linear. Depletion Capacitance reverse forwar d V Figure 22: The variation of the depletion capacitance as a function of bias across the pnjunction. There are two different situations. V0 the builtin voltage and ε = ε 0 ε r. Depletion capacitance in reverse bias (cfr.
The density of excess carriers and the bias voltage across the junction are intimately related.8 Conclusion The characteristics of a pn diode. this means that the voltage across the junction can only change when the density of excess carriers has changed. The large signal equivalent circuit for pn diodes is given in fig. 2. Rs C Figure 23: large signal equivalent circuit of pn diode. When recombination of carriers is taken into account. When switching from on to off. The maximum gradient of the excess carrier concentrations determines the current in the pn diode. Switching times in devices where minority carrier concentration play an important role are normally relatively large compared to switching times in majority devices such as a Schottky diode. the excess carrier concentration varies exponentially if we take the long diode approximation. The bias voltage across the junction defines the excess carrier concentration at the edges of the depletion region while the Ohmic contacts impose bulk conditions on the material. are completely described by the variation of the minority carriers at each side of the junction. Due to the existence of excess carriers in the neutral regions of the diode. 23.Semiconductor Devices 2012 . If we ignore carrier recombination completely – an assumption that can be made when the diode is sufficiently short – then the excess carrier concentration varies exponentially. delays occur when switching. the depletion region in the pn diode across which the largest part of the voltage drops. at low injection levels (meaning that the injected density of carriers across the junction is much smaller than the majority carrier concentration available). 38 EE2 . can only extend when the excess carriers are removed.signal equivalent circuit. This maximum gradient occurs at the depletion region edges. Linearization is no longer possible because the whole nonlinear currentvoltage region is used.
pn0 and np0. Basically. the BJT structure consists of a series connection of a pn and np or np and pn diode where the nregions. The BJT is another device in which the minority carriers will play an important role. Recombination of carriers in the base is very important as it will determine whether the material combination will have amplification or not. Having a series connection of two pn diodes.2 Charge transport in BJTs (principle of operation) In order to get an idea of how a BJT functions it is good to review the pn diode. Remember that in a reverse eV nkT − 1 becomes I tot = − I s and is independent of the magnitude biased pn junction. The magnitude of Is is determined by Lp the density of minority carriers. As a consequence the currents that are flowing in the different layers of the BJT will be determined by the maximum minority carrier concentration gradients. in each layer. As in the pn diode the voltage across each junction will determine the density of excess carriers at the edges of the depletion regions of each junction. this magnitude determines the current. of both diodes overlap. Bipolar Junction Transistor (BJT) 3. We will see that it is the voltage across each junction separately that determines the currents through the BJT. In reverse bias it are minority carriers that are crossing the junction. very often recombination is ignored and a “short BJT” approximation is used. V I I p n Isat V Figure 1: Left: reverse biased pnjunction. namely holes flow from the ntype material to the ptype and electrons from the ptype to the ntype. Right: Currentvoltage characteristics. 3. Is is given by: I s = eA n p 0 Dn Ln + p n0 D p . 39 EE2 .Semiconductor Devices 2012 . the current I tot = I s e of the reverse bias voltage V. similar to the pn diode. Notwithstanding this. respectively pregions. This will allow easy mathematical formulae to describe the currents and gives relatively good approximations for the collector current.3. The calculation of the excess minority carrier concentration across each junction is thus identical to that of the pn diode. 32 Remember that the minority carriers need to be supplied by carrier generation. This makes a pnp and npn BJT. there are now two junctions to consider of which the boundaries are not connected to an Ohmic contact. As the number of minority carriers is restricted32.1 Review – the short BJT The BJT is a device which has three ohmic contacts to the three different semiconductor layers that are connected together. which will occur near the depletion region edges.
we can use a forward biased p+n junction (+: heavily doped pside). 2. Since NA >> ND we find that p n0 >> n p0 and thus the hole minority carrier gradient. Right: the offleakage current increases as the available minority carrier density increases due to the hole injection process. Of course doping is not a good option for active control of the offcurrent. and thus more minority carriers near the depletion region edges and thus the offcurrent will increase with increasing light intensity. Here VEB>0V (forward bias) and T kT VT = . The minority carriers near the depletion region edge – thus those that feel the electric field across the depletion region – will be swept across the depletion region and thus increase the offcurrent. More intense light will generate more carriers. the magnitude of the minority carriers will increase and the reverse bias current in the pn diode will increase. Another possible way to control the current density is by shining light of the correct frequency on the junction. This light will be absorbed and creates electronhole pairs (carrier generation). 40 EE2 .Thus if we decrease the doping density in each layer. 2 where only an increased injection of holes is considered. This can be done in a controlled way by varying the intensity of the light.Semiconductor Devices 2012 . Since Ip >> In we call a p+n junction a hole injector. This is illustrated in fig. V I I p n V Isat hole injector Figure 2: Left: Holes are injected in the ntype region near the depletion region edge of a reversed biased pnjunction. Thus if we put this hole injector sufficiently close to a reverse biased pn junction then the hole injector will increase the amount of minority carriers (in this case only holes) that will flow across the junction and thus the offcurrent increases in exactly the same way as given in fig. The reason can be easily seen from the gradients of the V EB minority carriers in each region: in the nregion the gradient is given by p ' n − p n0 ≈ p ' n = p n0 exp V and T V EB in the ptype layer the gradient is given by n' p −n p0 ≈ n' p = n p0 exp V . For an all electrical solution to the active control of the offcurrent in a reversed biased pn diode. that defines e the hole diffusion current in the nregion is larger than the electron minority carrier gradient in the p+ region. A heavily doped p layer in a pn junction will cause the total current to be mainly dependent on holes.
a large amount of holes are injected by the emitter into the base and these are then collected by the collector.Semiconductor Devices 2012 . The player of the hole collecting diode is called the collector (C). However we have seen in pn diodes that both hole and electron currents exist at the same time but not necessarily in the same magnitudes. then the collector current is increasing accordingly. The magnitude of the current is determined by the minority carriers that are injected across the EB junction rather than the reverse bias voltage across the BC junction. then the collector current is equal to the emitter current and there is no current amplification between IE and IC. However what happens in between the common base point (common between the emitter and collector)? Is there a current flowing from this point in or out of the base region? The answer of course is yes. 3 gives the whole combination of pn and np junction forming a pnp bipolar junction transistor (BJT). The heavily doped p+ layer of the injector diode is called the emitter (E). The BJT thus operates by injection and collection of minority carriers. Thus if the emitter current – which is the current across the EB junction – increases.IE VEB IB p+ Emitter holes injected n Base IC VBC IC p Collector VBC IE Wb holes collected Figure 3: Left: a combination of a forward biased p+n junction. We have also seen that recombination of carriers can occur. The magnitudes is determined by the doping densities. it needs to be holes that are injected. 41 EE2 . the layer that is in common for the two diodes is the base (B) and the layer that will collect all the holes that are injected by the p +n diode (thus injected by the EB junction) will be collected by the reverse biased np junction. as there will be less holes available for collection at the collector side. 3 (right). In normal active mode operation of the BJT: VEB>0V and VBC<0V in a pnp BJT. If recombination of holes occurs in the base region then it can be expected that the collector current will decrease. In order to understand this. Later we will see from the expression of the current amplification factor and the emitter junction injection efficiency that high emitter doping gives good amplification and injection efficiencies. This approach has given us an insight into the relationship between the emitter and collector current in a common base configuration (as given in fig. A good BJT is an asymmetric device: the doping density in the emitter is higher than the doping density in the collector. The main carrier flow in a pnp BJT is holes. acting as a hole injector and a reverse biased np junction acting as a hole collector – the BJT. Fig. If the base width is sufficiently small the collector current that will flow as a function of reverse bias voltage across the BC junction is given in fig. 3). we have to take a closer look at what all the carriers are doing in the structure under the given biasing conditions. and all the injected holes get through the base. Note that if we take the approximation that the emitter current consists of a hole current Ip only. as seen above. thus we need a hole injector. Right: the collector current (holes collected) as a function of the emitter current (≈holes injected). From the physics we can see that since we want to collect more holes.
These electron currents have a direct relationship to the base current in a pnp BJT. This reduces the amount of electrons that the base current needs to resupply. Thus no resupply of electrons would lower the hole current to zero. This means that the forward biased voltage across the EB junction decreases and thus reduces the amount of injected carriers. the lost electrons from the base through the forward biased EB junction has to be resupplied by an electron base current. The reverse biased BC junction thus supplies electrons to the base. the base current IB can be found to consist of three terms: I B = I ' B + I " B − I CB 0 (1) where I ' B = base current to compensate lost carrier injected into the emitter I" B = base current to resupply the recombined charge 42 EE2 . Whilst the injected holes (hole flow in fig. If the number of electrons in the base is not resupplied the charge density relationship in the base would break down and hole current would disappear. This supply however is found to be dramatically smaller than the resupply current and will therefore often be ignored. In fig. This is indicated in the fig.As a consequence there will also be electron currents in the BJT. Arrow n o 2 are the holes reaching the collector. To accommodate this decrease in majority carriers while maintaining charge neutrality. From this diagram and the need to maintain the given steady state bias conditions. this is given by arrow no 4. the depletion region at the EB junction has to adapt. there is a probability for these holes to recombine with the electrons available in the base region. 4) are diffusing through the base. In order to maintain the hole current IC for the given biasing conditions. IE VEB IB 5 4 hole flow p+ 1 n IC electron flow 3 VBC 2 p Figure 4: The BJT in common base configuration in forward active mode with the flow of both holes and electrons (flux) indicated with arrows. Therefore in order to maintain a collector current. This means that electrons are disappearing out of the base region. 4 by arrow no 5. At the emitterbase junction holes are injected into the base and electrons are injected from the base into the emitter. As a consequence of the recombination process the number of holes and electrons in the base is decreasing. If there would be no resupply of these disappearing electrons then the density of majority carriers in the base (electrons) would decrease. The numbers indicate different physical events explained in the text below. The thermally generated electrons and holes make up the reverse saturation current across the basecollector junction. the base contact has to resupply the electrons. Arrow no 1 represent the injected holes lost for IC through recombination in the base.Semiconductor Devices 2012 . 4 all the different current components that can exist in a BJT are given.
Although this discussion of the different currents is done for the common base configuration. However. both emitter and base layers are thin. In order to deliver the 1st requirement the base needs to be made sufficiently narrow and for the 2nd requirement. The derivation of the currents is easy for a BJT in the short diode approximation. 3.(1)): eV ' n 'p pn = = e kT (2) n p0 pn0 The variation of the minority carrier concentration will then be dependent on whether recombination has to be taken into account. Note that the main difference in analysis is the boundary conditions in the base junctions. In what follows we will explicitly derive the currents in the case where recombination is not taken into account and we will look into some of the issues. the emitter needs to be heavily doped such that carrier injection from emitter to base is a lot larger than carrier injection from base to emitter.Semiconductor Devices 2012 . charge neutrality is not maintained unless the depletion regions and thus the voltage drop across the different junctions adapt themselves and thus change the total current IC that is flowing through the device. 33 As would be the case for an ohmic contact 43 EE2 . The common base configuration is used because it gives a direct insight into the voltage across each junction EB and BC separately.3 Calculating the currents in a BJT As for the pn diode. Note that a BJT has a short base. This is the reason why BJTs are asymmetric – the emitter is more heavily doped than base and collector. is often sufficiently accurate to get an idea about device performance. the variation of the minority carriers in the base has to be given as in eq. such as boundary conditions and shape of the minority carrier concentration in the base. This means that the minority carrier excess in the base at the depletion region edges will not be the bulk minority carrier concentration33 but will be determined by the voltage across each junction and thus determined by the formulae (chapter 1 eq.I CB 0 = reverse bias leakage current that injects carriers from collector into base (very small) Thus the base current resupplies the majority carriers that are lost in the base (neglecting ICB0). we can make the choice of including recombination or not. It that case. To obtain a “good” BJT it is essential to limit the amount of carriers that are lost from the base via recombination and via the current injected from the base into the emitter. For a longer base or in case we want to take recombination into account. the calculations become more complicated. these currents are exactly the same in the other BJT biasing conditions. In a BJT under normal active mode (EB forward biased and BC reverse biased) the short diode approximation. These boundary conditions are imposed by the voltage across each junction rather than an ohmic contact as was the case for a single pn junction. The choice will depend on the accuracy that is required and the speed with which derivations and calculations have to be done. that are a result of recombination. when recombination is taken into account in order to more accurately predict the current gain. common emitter or common collector. If these would not be resupplied. making the analysis the same as for the pn diodes we have discussed in chapter 2. (9) of chapter 2. in which recombination is neglected.
Right: an npn BJT symbol with the direction of current. Apply nodal analysis (sum of current into node = sum of currents out of node) in the base node of fig. thus the electron current of the EB diode defines the base current. 35 This can only be done for forward active mode. This gives the well known result: IE=IB+IC A good way to remember this is to draw a BJT symbol with the currents and remember the nodal analysis techniques that you learned in the EE1 Circuit Analysis course. when the BJT is OFF. thus EB and BC junctions reverse biased then the reverse bias leakage current is important from a viewpoint of power consumption in the OFF state.5: For pnp BJT: IE = IC + IB For npn BJT: IC + IB = IE Whether recombination exists in the collector or not is not important for the derivation of the expression of the current in this case. The holes that are diffusing through the base will be collected by the collector and thus make the collector current.The short BJT approach Assumptions: • No recombination in the emitter and base region34 • The reverse bias collector current into the base is negligible35 • The depletion regions are neglected In this case and under normal forward active mode. The emitter current is the total current in the EB diode. 44 EE2 . the currents in the emitter base diode determine all the BJT currents completely. The electrons that are escaping from the base into the emitter have to be resupplied by the base current.Semiconductor Devices 2012 34 . p IB B n C IC IE p E IB B p n C IC IE n E Figure 5: Left: A pnp BJT symbol with the directions of the different currents.
VEB VBC p+ n p x xe Wb mi no rit y ca rri er co nc en tra tio n 36 37 p' n n' p p"n n p0 n" p n p0 Figure 6: top: pnp BJT in forward active mode.Semiconductor Devices 2012 . Diffusion happens because there is a carrier gradient. The shaded regions are the depletion regions. Note that the ‘ indicates forward bias and “ reverse bias. Notice that the variation of the density of the minority carrier concentrations is linear in each (short) layer. The minority carrier concentration at the ohmic contact of the emitter and collector imposes the bulk minority carrier concentration37. the contacts can supply and absorb all excess carriers and thus keep the carrier concentrations constant at bulk conditions. 45 EE2 . 6. The result is shown in fig. Due to infinite generationrecombination rates in ohmic contacts. This is done in exactly the same way as in the pn diodes. Bottom: The minority carrier density variation in each layer for the short BJT. In bipolar devices (thus both electrons and holes play a role in the transport processes) diffusion of minority carriers are the basis for the calculation of the current. Thus the first step in the derivation of the current in the BJT is to determine the minority carrier concentration at each side of the junction for both junctions in the device. In the emitter at the EB junction the injected minority carrier concentration (electrons coming from the base) is36: n' p = n p 0 e eVEB kT (3) In the base at the EB junction the injected minority carrier concentration (holes coming from the emitter) is: p' n = pn0 e eVEB kT (4) In the base at the BC junction the minority carrier holes are injected to the collector and thus we have a reduction of the minority carrier concentration in the base at the reverse biased collector: p" n = p n0 − e VBC kT e (5) Remember that diffusion currents are directly related to the gradients of the minority carriers.
IC The current amplification factor (gain): β = shows that we want IB as small as possible for a given IC. this parameter gives how much of the total input current (thus the I p + In emitter current) is “useful” for the output (the collector current).Semiconductor Devices 2012 . Ideally we want γ =1 because in that case In=0 and thus no electrons would be lost out of the base via the EB junction.Thus the electron current of the EB diode is given by: I n = eDn A n p 0 − n' p 0 − xe = eDn A n p0 e eVEB kT − n p0 xe eVEB ≈ eDn An p 0 xe e eVEB kT The hole current crossing the EB junction is given by: p ' − p" n p e kT I p = eD p A n ≈ eD p A n 0 Wb Wb = eD p Ap n 0 Wb e eV EB kT Note that p' ' n << p' n and p n0 << p' n . base and collector region • The reverse bias collector current into the base is negligible (thus ICB0=0A) • The depletion regions are neglected The minority carrier variations are given in fig. The first important observation is that the minority carrier concentrations at the junctions are still the same as those given by equations (1) as they are determined by the voltage across the junctions. Thus it is IE preferable as close to 1 as possible. This will not be true for the case of recombination because then IC is lower compared to Ip injected across the EB junction. therefore the formulae above can be simplified. The difference lies in the variation of the minority carrier concentration within each region. In the emitter and the collector regions we can expect the variation of the minority carrier density to be the same as in the long diode approximation since the boundary condition imposed by 46 EE2 . Taking recombination into account Assumptions: • Recombination in the emitter. Thus all the injected holes by the EB junction are collected by the BC junction. Thus the base current needed to maintain the collector current would be equal to zero in this case. This means no amplification between emitter and collector current. The emitter current is then the pn diode current: I E = I n+ I p The base current is: I B = I n and is thus the resupply of the electrons that have left the base via the EB junction. It is interesting to note that in the short diode approximation. The collector current is: I C = I p . 7. IB Ideally we want β as large as possible. IC The current transfer ratio: α = gives the ratio of the input current to the output current. the currents in the BJT are completely determined by the emitterbase diode only! Important parameters defining the quality of the transistor are: Ip The emitter efficiency: γ = . Note that γ=α when no recombination is taken into account.
Bottom: The minority carrier density variation in each layer for a BJT in which recombination is taken into account.the contacts ensures that the excess minority carrier concentration in these layers go to zero at the contacts. the boundary conditions are imposed by the junction voltages. In the base the variation of the minority carriers is a sum of an increasing and decreasing exponential. In the base however. The boundary conditions for the base region (see fig. In this case. 7).Semiconductor Devices 2012 . in forward active mode is VEB > 0V and VBC < 0V for a pnp BJT: V @x=0 δp n (0) = ∆p nE = p ' n − p n0 ≈ p n0 exp EB V (7) T @ x = Wb δp (Wb ) = ∆p nC = p ' ' n − p n0 ≈ − p n0 47 EE2 . the sum of two exponentials as the solution of the second order differential equations will be used. The continuity equations are repeated: ∂ 2 δn δn ∂x 2 ∂ 2 δp ∂x 2 − L2 n =0 − δp L2 p (6) =0 VEB VBC p+ n p x xe Wb mi no rit y ca rri er co nc en tra tio n p' n n' p p"n n p0 n" p x 0 Wb n p0 Figure 7: top: pnp BJT in forward active mode. Notice that the variation of the density of the minority carrier concentrations is exponential in E and C layer. An exponential approximation to the minority carrier concentration can be used for the emitter and collector. which is surrounded by 2 junctions. The shaded regions are the depletion regions. This means that the exact solutions to the continuity equations need to be used.
x=0 is the start point of the base region and Wb is the width of the base38. Thus the solution of the differential equation (6) with these boundary conditions is no longer simply a decaying exponential. The general solution of a second order differential equation for the holes is:
x δp n ( x) = C1 exp Lp + C 2 exp − x Lp
(8)
Where C1 and C2 are constants that are determined by the boundary conditions (7):
δp n (0) = p' n = C1 + C 2
W δp n (Wb ) = − p n0 = C1 exp b Lp + C 2 exp − Wb Lp
(9)
Substituting these boundary conditions in (8) gives the expression for the minority carrier variation in the base when recombination is taken into account. In order to calculate the currents in the case where recombination is taken into account, the gradients of the minority carrier concentration have to be calculated at the edges of the depletion regions for each junction because at these points the minority carrier gradients are maximum.
n Thus the emitter current is given by: I E = −eAD p dx
dδp
+ eADn
x =0
dδn p dx
x =0
39
and is the sum of the hole and the
electron current determined by the gradient of the minority carriers at the EB junction.
n The collector current is given by: I C = −eAD p dx
dδp
x =Wb
and is the gradient of the hole minority carrier
concentration in the base at the BC junction. The base current will now consist of 2 components as given by eq. (1) I B = I ' B + I " B where we have now assumed that I CB 0 = 0 A . Thus the base current consists of the electron current lost across the EB junction:
I ' B = eADn dδn p dx
x =0
and the current that needs to resupply the recombined electron population in the base. The
number of electrons that have recombined in the base is equal to the number of holes with which they have recombined. We know the number of holes that have recombined in the base from the difference in the hole minority carrier gradient between the emitter and the collector side. Thus this currentdifference that follows from the gradient difference at the junction at both sides of the base must be the current that needs to be resupplied via the base contact to compensate the electron charge loss in the base. Thus: dδ p n dδ p n I " B = −eAD p + eAD p . This is illustrated in fig. 8 that focuses on the base region only. Note: dx x = 0 dx x =W
b
A is the cross sectional area.
38 39
Notice that in fig. 7 the depletion regions, that reduce the effective base width, have been taken into account. Note that x=0 is the edge of the depletion region in each region E and B. 48 EE2  Semiconductor Devices 2012
E
B
I p (0) = −eD p
δ pn(x)
dδpn A dx x =0
C
I p (Wb ) = −eD p
dδpn A dx x =Wb
x 0 Wb Figure 8: The base region of a pnp BJT with the variation of the excess minority carrier concentration under forward active mode in the case recombination is taken into account. The dashed line (red) gives the gradient of the minority carrier concentration at the emitter side, the full bold line (blue) is the gradient of the minority carrier concentration at the collector side. The gradient at the emitter side is larger than that at the collector side indicating that the hole current at the emitter side is larger than the hole current at the collector side. This difference is cause by recombination of carriers in the base. The full solution of the currents in the BJT in case recombination is taken into account is given by: (10) IE (x) ≈ e A Dp/Lp (∆ pE ctnh(Wb/Lp)  ∆pC csch(Wb/Lp) ) IC (x) ≈ e A Dp/Lp (∆pE csch(Wb/Lp)  ∆pC ctnh(Wb/Lp) ) IB (x) ≈ e A Dp/Lp ((∆pE + ∆pC) tanh(Wb/2Lp) ) For the derivation, the interested reader can check the book of B. Streetman (see recommended textbooks). An alternative way to calculate the base current associated with recombination is to look at the charge in the base under steady state injection and collection (thus DC voltage across each junction). Due to the injection process of holes at the EB junction into the base and the collection process at the collector, there will be a hole charge Qp40 in the base that is consistent with this injection/collection process. The holes are diffusing through the base with a characteristic time constant τ t, called the transit time. With τ t =
Wb2 2D p
determined by
Q
p the diffusion process and the base width. Thus the steady state hole current is given by I p = τ . This hole t
current is equal to the collector current IC. On the other hand, we also know that the holes that are diffusing in the base recombine with a characteristic time τ p, the hole lifetime in the base. Thus a charge Qp is disappearing per τ p and is thus resupplied by the EB current. However the hole charge Qp recombines with the same amount of electron charges in the base. Thus the amount of disappearing electrons in the base is Qn=Qp. Thus an amount Qn is disappearing per τ p and needs to be resupplied by the base current. Thus the base current that is needed to keep Qn=Qp is given
Note that the total charge in the base is given by the integration of the excess carrier concentration in the base region over the width of the base. 49 EE2  Semiconductor Devices 2012
40
p n by I n = τ = τ . In the case that γ =1, thus no loss of electrons from the base into the emitter, thus I’B=0A, p p
Q
Q
the current In is equal to the base current IB= I’’B. Thus the currents and current amplification factor are given in function of charge and time constants:
τp τt I ⇒β = C = Qp I "B τ t I "B = τp
IC =
Qp
(11)
Since the base transit time is a lot smaller than the recombination time, the current amplification factor β is large. Note that eq. (11) only takes the recombination component into account in the current amplification factor. In order to add the loss of carrier from base into emitter I’B has to be added to of I’B in (11). 3.4 Switching of the BJT To study the switching delay in a BJT, we need to apply a biasing circuit. In common emitter configuration the BJT biasing circuit is as given in fig. 9. A square wave voltage is applied to the base that drives the pnp BJT from off to on and vice verse. The variation of the collector current is measured as a function of time and the waveforms of the applied base voltage and the collector current are recorded on oscilloscope (see fig. 9). es iC Es iB n RS
e
s
p p+
RL ECC
t Es iC ICmax
iE
t
Figure 9: Left: simple switching circuit for the pnp BJT in common emitter forward active mode. Right top: switching voltage on the base. Right bottom: Output current iC as a function of time during switching. Switching cycle First the BJT is switched from off (both junctions reversed biased) to on, with a base current that drives the BJT into oversaturation (both junctions forward biased, see fig.10). Then in the isecond stage the BJT is C driven off from oversaturation into reverse bias across both junctions. We use a pnp BJT in order to avoid confusion between carrier flux and current. The hole current and hole flux are in the same direction, while the electron current and electron flux are opposite due to the negative charge of the electrons. Switchon:
50
EE2  Semiconductor Devices 2012 vCE Fig. 10: saturation region for small –vCE (in red – online)
11: ON switching sends the BJT from OFF 11). The maximum collector current that can flow through the circuit is limited by the load circuit and can be approximated by: E ic = CC RL When the base current is increased further via the control circuit. Therefore sketch the currentvoltage characteristic of the ideal transistor in the given configuration for the left part of the circuit in fig 9 for different values of the control current.Semiconductor Devices 2012 . the collector current cannot further increase than this maximum. e. When the baseemitter junction of the pnp BJT is switched from reverse to forward bias. The base controls the collector current with a control base current: IB. thus the output current. The lowest characteristic is for the smallest base current. iC through the BJT increases too. IC increases.The circuit in fig. Each crosssection between the BJT characteristics and the load line is a solution. line limits the maximum current that can flow in When switching from OFF to ON the collector current through saturation at ts. And the current is leakage current through the reverse biased diodes only. The voltage across the basecollector junction is determined by the crosspoint between load line and BJT characteristics. icsat≠β IB The expression for the load line (purple online) in fig. The current through the BJT is now limited by the load to saturation (IB sufficiently large). at t=t1 the voltage across the BJT is V1. When the transistor is OFF the voltage across it is – t1 vCE=ECC. For increasing IB. The relationship IC=βIB breaks down at that point. The base current is increasing. The load line and the condition of Ic=β IB is no longer valid. Lots of minority carrier charge is built up in the base under saturation and oversaturation condition. 10. Each single line in the graph for the BJT is for one value of IB. The shaded region on the graph is the saturation region.g. The output characteristics of a commonemitter transistor are given in fig. ECC Fig. The constant current value of IC (active BJT mode) is related to IC=βIB. When the transistor is ON the voltage vCE V1 VON across the BJT goes to zero but doesn’t quite get there (see fig. 9 can be solved by using the load line technique (see EE1 Analysis of Circuits). holes are injected from E into B and an excess hole charge is built up in the base. the BJT needs to follow the load line. until the BJT goes into saturation at a time ts indicated on the graph. We have to look at the minority carrier concentration to find out what is happening when the base current increases further. In this region both emitterbase and basecollector junction are forward biased. 11 is: iC −1 E icsat=βIB ic = (−vce ) + CC ECC /RL RL RL ts We can now draw the BJT characteristic and the load line on the same graph (fig. The amount of excess charge in the base is 51 EE2 . 11) and indicate the OFF and ON point on ic=βIB the graph. The basecollector voltage goes from reverse bias into forward bias (saturation) at this time.
otherwise there would be no offleakage current. Recombination is taken into account. 12: Minority carrier concentration in the base of BJT in OFF state.Semiconductor Devices 2012 .determined by the magnitude of the base current. Recombination is taken into account. This gradient is nonlinear when recombination is taken into account. 13. is determined by the minority carrier concentration variation in the base. at the BC junction. the minority carrier variation in the base for the pnp BJT at the starting point (onset) of saturation becomes as given in fig. under the assumptions that IBC0=0. At saturation. The minority carrier variation in the base for the pnp BJT in OFF state is given in fig.13: Minority carrier concentration in the base of x BJT at onset of saturation. the voltage 52 EE2 . The collector current in a BJT. thus by the gradient. p(x) E B C pn0 p”nE 0 pn0 p”nC Wb x Fig. p(x) E p’nE B C pn0 0 pn0 Wb Fig. When the base current drives the BJT into saturation. Both junctions are reverse biased when the BJT is OFF: v eb < 0 & vbc < 0 . Note that there is a gradient of minority carriers near the junctions. 12.
the collector current is constant because the load line limits the maximum collector current. after which the EB 53 EE2 . then with further reduction of the charge in the base. 14: Minority carrier concentration in the base of BJT in saturation. Thus the collector current remains constant until the excess charge Q = Q p − Qs is removed.condition across each junction: v eb > 0 & vbc = 0 (thus EB forward biased and no bias across BC). Driving the transistor ON using oversaturation ensures fast switchon times. At this moment the charge in the base is called the saturation charge Qs. The basecollector voltage can only return to reverse bias when this excess charge is removed. 14: p(x) E p’nE B C p’nC pn0 0 pn0 Wb x Fig. After the BJT saturates. however this will create a large density of store minority carrier charge in the base that will need to be removed before the BJT can switch OFF (similar to the pn diode!) Switchoff Now the pnp BJT will be switched OFF by applying a reverse bias voltage on the EB junction. then I C ≠ βI B . the EB junction remains forward biased and the BC junction too until all excess charge larger than the saturation charge is removed. Thus the minority carrier charge in the base buildsup further but does not have any more influence on the collector current. the BC junction goes into reverse bias and the collector current can decrease following the load line. then the associated charge buildup in the base is larger than the saturation charge: Q p > Qs where Q p = τ p I B . The baseemitter junction is still forward biased until all remaining minority carrier charge is removed from the base. The minority carrier variation in the base for the pnp BJT in forward active mode oversaturation is given in fig. When the charge in the base has reached Q= Qs the BC junction first goes to zero bias. This removes charge out of the base by a reverse pn diode current. At saturation Qb=Qs (Qb=base minority carrier charge) but beyond saturation Q p > Qs and the basecollector voltage becomes forward biased → oversaturation. If the emitterbase bias is such that the base current drives the transistor into oversaturation. Recombination is taken into account.Semiconductor Devices 2012 .
Middle from top to bottom: ib: EB drive current. 4. Negative base current extracts the carriers from the base. This time is referred to as storage delay time tsd. t's is the time it takes to remove the excess stored charge QB excess. Magnitude of the base current is large and drives the BJT into iB oversaturation. Ic remains constant until the excess charge built up above saturation is removed. t=ts is the onset of saturation when switching on. t2=point of switchoff of EB junction. Indicates negative current through EB junction when δp driving off t2 ts t1 ∆pC 0 t0 Wb t1 ts t2 t’s pn tsd 6.Semiconductor Devices 2012 . Bottom left: the different DC characteristics of the BJT in common emitter configuration. 2. ic output current. A summary of what is happening with the different parameters in the BJT is given in fig. 15: Top left pnp BJT with bias circuit in EB and EC. ∆pE ∆pE ∆pE 7. Qb: charge in the base. The EC bias circuit forms the load line. In this case the short diode approximation is used for the minority carrier gradient in the base. 15.junction can go into reverse bias. Derivation of the turnon and turnoff time gives: 54 EE2 . BC goes reverse bias IC≈ECC/RL x ECC 5. The BJT is driven into saturation in this example. Fig. After that the BC junction goes into reverse bias and IC follows the load line. Switching off the IB BJT by driving the base into reverse bias RL IB ECC QB e E s Es iC ECC /RL ts t1 ic=βIB vCE iC IC s iC IB≈Es/RS iB RS iE t icsat≠β IB icsat=βIB QB:excess Qs 3. The exponential reduction of the collector current is caused by the baseemitter diode switchoff. the EB junction goes to reverse bias. Right: Minority carrier concentration in the base of BJT at different characteristic times. All curves are draw for 0 → ON immediately followed at t=t2 to OFF. 1. Recombination is ignored in the drawings. Each point of the cross section of BJT characteristic and load line is a possible solution to the circuit. The output current ic is controlled by the base current which is controlled by the EBdiode bias circuit. When excess charge in the base is removed.
Turn ON: from 0 bias to ON (ON is normally in saturation) Switching happens at t=0s. (15) in the diode chapter): Q p (t ) dQ p (t ) i p (t ) = + (12) τp dt This differential equations can be solved using Laplace transforms. 42 An easy way to describe this delay time is to approximate it by the RC time constant imposed by the capacitance of the depletion width and the resistance of the bias circuit. therefore IB=E/Rs). This is a good approximation for the total current variation for a p+np BJT in forward active mode. 55 EE2 . the potential drop across the diode is small and the depletion region needs to be removed. the following differential equation needs to be solved41 (see eq. Qp(t) cannot change immediately. but in the bipolar devices’ switching experiments an easier way exists because in large signal switching.Semiconductor Devices 2012 41 .In order to calculate the switching dynamics in a p+np BJT. Boundary conditions: For t<0 ip(t=0)=0 (switching from zero bias in current and voltage). (12) can be rewritten for the time variation of Qp(t) after t=0 taken this steady state condition into account: Q p (t ) dQ p (t ) IB = + τp dt dQ p (t ) Q p (t ) − I Bτ p Qp (t ) = − dt τp =∫ 0 t Q p (t =0 ) ∫ dQ p (t ) Q p (t ) − I Bτ p − dt τp Q p (0) ln ( Q p (t ) − I Bτ p ) Q p (t ) − I Bτ p ln −I τ B p Q p (t ) − ln ( Q p (t ) − I Bτ p ) = −t τp (13) −t = τ p −t Q p (t ) − I Bτ p = − I Bτ p exp τ p − t Q p (t ) = I Bτ p 1 − exp τ p Equation (13) gives the variation of the charge in the base as a function of time when the EB junction is switched from 0 to forward bias IB. then eq. Under the given conditions. we see that after switching a steady state current condition is reached. For t≥0 ip(t)=IB (EB bias circuit determines the current that is flowing. then Qp(t=0)=0. Note that we look at the charge related to the maximum current component only. If switching would have happened from reverse bias to forward bias then there would be a delay of time tr before the charge can be build up in the base. steady state conditions occur before and after switching on the current that controls the switching dynamics (in a BJT that is the base current determined by the EB junction bias circuit). This delay tr is related to the removal of the depletion region42 that would exist if the diode would have been in reverse bias before t=0s. the variation of this charge is described by the differential equation (12).
Turn OFF: from ON (saturation) bias to OFF Switching happens at t=0s. ON (saturation) bias to OFF (zero) Boundary conditions: For t<0 ip(t=0)=IB (the BJT is ON). This is when the collector current remains constant. Two different techniques can be applied to turn the BJT off. This is similar to the pn diode driveoff delay time. IC limited by external circuit). the removal of the charge in excess of the saturation charge.2V for Si BJTs). For the calculations the boundary conditions to solve eq. then Qp(t=0)= IB τ p. β = . The maximum collector current is then43: E iC (t ) ≈ CC (15) RL Equations (14) and (15) allow us to extract the time delay to saturation because at t=ts eq. the removal of the excess charge that prevents the EB junction to turn into reverse bias. This time is called the storage delay time in the BJT. 56 EE2 . One is to switch to 0 bias and current conditions (open circuit) and the other one is by driving the EB diode into reverse bias.Semiconductor Devices 2012 .If the EB drive current IB is smaller than that necessary to drive the BJT into saturation. then the collector current would be given by: Q p (t ) I Bτ p 1 − exp − t i C (t ) = = (14) τ τt τt p With τ t the base transit time (is the time it takes the holes to travel through the width of the base). 2. then the load line (EC bias circuit) will limit the current that can flow in the output (see fig.(14)=eq. For t≥0 ip(t)=0 (imposed by bias circuit). (12) will change. The second one will give much faster switching times as we have seen for the pn diode switching.(15) I Bτ p 1 − exp − t s = I max = E CC C τ τt RL p 1 t s = τ p ln max 1− IC τ t I Bτ p τp ts is small for small τ p and for IC small compared to βIB. Thus the switchon time to saturation is fast τt if we drive the BJT into oversaturation (large IB. 43 Slightly smaller because the voltage drop across the BJT in the ON state is not 0 (it is around 0. The turnoff time is based on two factors: 1. At that moment t=ts (fig. However when the EB drive current IB is larger and drives the BJT into saturation. 15) the charge in the base is equal to the saturation charge QBsat and the collector current ic(t) cannot further follow the base charge given in (13). 15).
This requirement is opposite to the ON requirement. Thus initially ic(t) is determined by the load line: E iC (t ) ≈ CC for t < tsd (16) RL ic (t ) = Q p (t ) = I Bτ p −t exp τ p for t ≥ tsd τt τt (17) The storage delay time associated to the removal of the excess base charge above saturation tsd can then be found by equating eq. However we can decrease this OFF delay by driving the EB diode into reverse bias rather than waiting for recombination to remove the charges as what is done in the ON→0 bias switch. the collector current first remains constant. Q p (t ) dQ p (t ) 0= + τp dt −t Q p (t ) = I Bτ p exp τ p Although the base charge is decreasing exponentially. For t≥0 ip(t)=IB (IB=E/Rs imposed by bias circuit)44. After switching the steady state control current is zero. then Qp(t=0)= IB τ p. This is because first the excess saturation charge needs to be removed before the collector current can change its value and follow the variation of the base charge.The calculation method is the same as for the ON switching but the t=0 boundary conditions have changed: the start value of the base charge is different. Thus the switchoff time is fast if we prevent the BJT to go into oversaturation. This is not necessary.Semiconductor Devices 2012 . 57 EE2 . (16) and (17) at t = tsd The storage delay time is then: βI t sd = τ p ln B I C Thus tsd is small for small τ p and for βIB small compared to IC. Q p (t ) dQ p (t ) − IB = + τp dt −t Q p (t ) = I Bτ p 2 exp − 1 τ p E iC (t ) ≈ CC RL ic (t ) = Q p (t ) = I Bτ p 2 exp − t − 1 τ τt p for t < tsd (25) τt for t ≥ tsd (26) 44 Note that the amplitude of the switch off voltage is equal to that of the switch on voltage. ON (saturation) bias to OFF (reverse bias) Boundary conditions: For t<0 ip(t=0)=IB (the BJT is ON).
Fig. 3. However this condition is in conflict with a fast switchoff time for which the stored minority carrier charge in the base has to be as low as possible. This is a result of the fact that the Schottky diode switches on (goes into conduction mode) at lower voltages than a pn junction made of the same material. Remember also that the onvoltage of a Schottky diode is smaller than that of a pn diode.6 Small signal equivalent circuit of a BJT 45 To first approximation. Left: npn. Therefore the Schottky diode will fix the basecollector voltage at the Schottky onvoltage << 0. Remember that a Schottky diode is a majority carrier device that does not store minority carriers and thus switches faster than a pn junction. 58 EE2 . Introducing a Schottky diode (metalsemiconductor junction) keeps the charge in the base region at the value given at time ts. this is done by introducing a Schottky diode clamp.7V. the BJT has to be driven into oversaturation with a large forward bias base current. The onvoltage of the Schottky diode is determined by the metalsemiconductor work function difference45. a clever technique has been invented to prevent the charge buildup of the base region after saturation. thus no oversaturation. The onvoltage of a pn diode is determined by the builtin voltage. C C B E B E Figure 16: BJT with a Schottky diode clamp. righ: pnp. 16 gives a BJT with Schottky diode clamp.Semiconductor Devices 2012 . In order to be able to switchon fast and also switchoff fast without being hindered by the excess oversaturation charge in the base.The storage delay time is then: I Bτ p t sd = τ p I Bτ p 1 1 I Csatτ t + 2 2 I τ Csat t 1 1 I Bτ p The addition of the term: + 2 2 I τ in the denominator will decrease the storage delay time compared Csat t to the previous case.5 BJT – Schottky diode clamp It is clear from the explanations and calculations that in order to obtained fast switchon. 3. See the Advanced Electronics Devices course for deviations from this approach.
Transconductance Input resistance. Ro is the output resistance.B vb e Cj. see yr.BE Rπ Cd. This resistance is associated to base with modulation which imposes an increase of the collector current as a result of the reverse bias BC junction depletion width extending into the base. This growing depletion width with increasing reverse bias reduces the effective base width (= undepleted base region). The EB diode is represented by the differential resistance in parallel with the depletion and diffusion capacitance. compared to base doping. Between base and collector we find the depletion capacitance of the reverse biased BC junction.Semiconductor Devices 2012 . Junction capacitances are ignored. it is the derivative of the output current iC to the control voltage vbe (similar to the transconductance in a MOSFET.1).BE gmvbe npn Ro C E Figure 17: Hybrid π equivalent circuit. The depletion capacitance can be neglected as for amplification the BJT is used with EB forward biased.BC Cj. ib B ic C βib vbe rbe or rce gmvbe E Figure 18: Simplified small signal equivalent circuit. rbe=rπ 59 EE2 . Figure 17 gives the complete intrinsic small signal equivalent circuit. gm is the transconductance. Smallsignal current generator β (alternative symbol hfe) dI C ic τ p = = β= dI B ib τ t I p D p p n 0 xe D p N A xe = = For a short pnp BJT: β = I n Dn n p 0Wb Dn N DWb for high gain we require (i) small base width Wb (ii) large emitter doping. This capacitance is also referred to as the Miller capacitance and introduces an unwanted feedback network from the output C to the input B. This is associated to the early voltage VA.
In fig. annotated with N and I respectively. IE IB p+ n IC p N I : : E C B B C E 60 EE2 . the EB and the BC diodes.Output resistance. The cutoff frequency of the BJT can be derived from the small signal current gain: The transit time of the carriers from emitter to collector: Note: the delay time through the basecollector depletion region is ignored. 19 a pnp BJT is given and normal and inverted active mode defined. Extraction of currents is based on the evaluation of the hole distribution in the base of the transistor for these two diodes. This is a large signal model of the BJT that is frequently used in simulation tools such as SPICE. Take a p+np transistor. The operation is divided into “normal operation” and “inverted operation”. 3. The basecollector capacitance is due to the basecollector depletion capacitance.Semiconductor Devices 2012 .6 Large signal equivalent circuit of a BJT The coupled diode model – EbersMoll model. rce r0 The baseemitter capacitance is due to the diffusion and depletion capacitance of the baseemitter diode. the coupled diode models separates the operation of the BJT into two diodes.
Semiconductor Devices 2012 . This is given in fig. The charge distribution and currents in the base (plots are draw without taking recombination effects into account) can be drawn for each operation mode separately and then combined for the complete operational picture. Right: the gradient of the hole density in the narrow base (fat line) is the sum of the dashed lines due to combined injection in normal and inverted mode. Ebers – Moll equations IE = IEN + IEI = IES (exp[(eVEB/kT)] – 1) – αI ICS (exp[(eVCB/kT)] – 1) IE = EB diode current + fraction of CB diode current that reaches the EB junction (27) IC = ICN + ICI = αN IES (exp[(eVEB/kT)] – 1) – ICS (exp[(eVCB/kT)] – 1) IC = fraction of EB diode + CB diode current current that reaches the CB junction With: • α: ratio of collected current to injected current • IES: emitter saturation current in normal mode (VCB=0. ∆pE=0) Or (27) can be rewritten as: 61 EE2 . Middle: components due to injection and collection in reverse operation mode. ∆pE δp ∆pC Wb δp IEN ICN Wb ∆pC IEI + ICI Wb = IEN + IEI ICN + ICI + δp Wb = δp δp ∆pE δp ∆pC Wb Figure 20: Left: components due to injection and collection in normal operation mode.Figure 19: definition of normal and inverted mode operation. 20. ∆pC=0) • ICS: collector saturation current in inverted mode (VEB=0.
An interesting feature of the EbersMoll equations is that emitter and collector current are described in terms that resemble the diode equation + an extra term that provides the coupling between the emitter and collector properties. Based on (28) a large signal equivalent circuit can be proposed for the BJT that is valid in all operation regimes.Semiconductor Devices 2012 . IC0 = (1 – αN αI) ICS is the magnitude of the C saturation current with emitter junction open (IE=0).IE = IE0 (exp[(eVEB/kT)] – 1) + αI IC (28) IC = αN IE – IC0 (exp[(eVCB/kT)] – 1) Where: IE0 = (1 – αN αI) IES is the magnitude of the E saturation current with collector junction open (IC=0). Equivalent circuit: IE0 (exp[(eVEB/kT)] – 1) αN IE E IE αI IC IB C IC IC0 (exp[(eVCB/kT)] – 1) B Figure 21: Equivalent circuit derived from the EbersMoll equations. Small changes in base current result in large changes in collector current. such that a small input voltage change can result in a high current change. 3.7 BJT amplification In bipolar transistors the input impedance is low. This gives a description of the BJT in terms of simple diode characteristics + an extra current generator. Region: active * BJT characteristics (sketches) active 62 active EE2 .
if power consumption is not an issue. the emitter current controls the collector current. Although. The conditions for fast on and off switching cannot be met simultaneously.Semiconductor Devices 2012 . In order to improve this.IC IE IC IB cutoff VCB saturation Common base connection cutoff VCE saturation Common emitter connection Figure 22: Bottom: The shaded arrows indicate the biasing connections in the common base (left) and the common emitter (right) configuration. For fast onswitching a large IB is used to drive the transistor in oversaturation. Whilst the MOSFET is used for integrated logic applications (fast large signal switching and very low offcurrents). the switching speed of the MOSFET has become comparable to the switching speed of the BJT. For fast offswitching the baseemitter junction is reverse biased. a Schottky diode clamp can be used between the base and the collector. A popular large signal equivalent circuit for the BJT is provided by the EbersMoll equations that allow the BJT to be modelled by two diodes and two current sources. BJT logic can be used to improve the speed performance. Due to the current aggressive downscaling of the gate length of the MOSFET. 3. Top: the current–voltage characteristics associated with the two configurations. Thus common base does not give current amplification while common emitter does. whilst in a common emitter configuration the base current controls the collector current. but when carefully designed. Note that in common base. 63 EE2 . at much lower power consumption.8 Conclusion The major time delay in a BJT is caused by the storage of excess minority carrier charges in the base of the transistor. This equivalent circuit is valid in all regions of the BJT operation. the BJT is mainly used for amplification (very large current gain).
This action might not be possible to undo. Are you sure you want to continue?
We've moved you to where you read on your other device.
Get the full title to continue listening from where you left off, or restart the preview.