o
. Thus, the motion is defined by
If we can make the semiconductor as nearly pure as possible and, if the energy gap is large compared to kT, the
free carriers generated from impurities are insignificant and the intrinsic concentration of EHPs generated
thermally is very small. If all of the carriers are generated thermally, the densities of electrons and holes are the
same and equal the intrinsic concentration, n
i
for
silicon at 300K
= = =
10 3
i
n p n 10 cm
where n and p are the concentrations (cm
3
) of free
electrons and holes. Intrinsic semiconductors have
so few EHPs that they arent very interesting so we
make them extrinsic by adding useful intentional
impurities dopants. The dopant atoms are relatively weakly bound and behave like a fairly mobile charged
particle bound to the fixed charge ionized impurity atom. This is essentially the same as Bohrs hydrogen atom
with corrections for the effective mass and permittivity. Note the inclusion of these corrections in the equation for
the binding energy in silicon.
The binding energies E
A
and E
D
, of typical Column III and Column
V dopants are ~ 50 meV. Since this energy is much smaller than the
silicon band gap of ~ 1.1 eV, we refer to these impurities as shallow
donors and acceptors. How they behave depends strongly on the
temperature. For low temperatures (kT << E
A
or E
D
), there is a fixed
charge from the donor or acceptor and the carrier (electron or hole) is bound to the fixed charge. For high
temperatures (kT >> E
A
or E
D
), there is a fixed charge from the
donor or acceptor but the carrier (electron or hole) is no longer
bound to the fixed charge and instead moves into the band
becoming a mobile carrier as shown schematically in the figure.
This process is called ionization. The sign of the fixed charge is
opposite the charge on the mobile carrier. An important parameter for doping is the concentration (cm
3
) of donors
or acceptors, given by N
D
or N
A
. These are chemical quantities of fixed impurities in the silicon lattice. Upon
ionization, each donor or acceptor generates a single mobile carrier. Thus, for silicon at room temperature (large
kT), the concentration of free carriers is equal to the chemical concentrations of fixed donors and acceptors:
material m
e
* m
h
*
r
Si 1.08 0.56 11.9
Ge 0.55 0.37 16.2
GaAs 0.067 0.48 13.18
material m
e
* m
h
*
r
Si 1.08 0.56 11.9
Ge 0.55 0.37 16.2
GaAs 0.067 0.48 13.18
E
C
E
V
E
G
E
C
E
V
E
G
0.050 eV
donors
0.050 eV
acceptors
low temperature high temperature
+

D D
N N n
+
+ and
A A
N N p
+
where n and p are the concentrations of electrons and holes. Note the charge sign on the ionized donors and
acceptors.
Density of States. The simplified energy band structure clearly indicates that there are no states in the gap and
some states within the bands. The density of states, different for electrons and holes, is given by
Note the square root (parabolic) dependence of density of states on energy. The relationship between the energy
band diagram described earlier and the density of states is shown in this figure. The number of conduction band
states per cm
3
lying between E and E+dE is given by
C
(E)dE
and the number of valence band states per cm
3
lying between E
and E+dE is given by
V
(E)dE.
The density of states function tells us how many states exist at a
given energy but gives us no information about how many of
those states are occupied. At finite temperatures, there is always
some probability that there is an electron or hole occupying the
states. The function that defines the probability that the states are
occupied by electrons is the Fermi function which is given by
where E
F
is the Fermi level. The probability that a state is occupied by a hole is 1f(E). The Fermi level is the
energy where the probability is exactly 0.5. The Fermi function is shown at T = 0 and T > 0 in the figures. The
region where E is more than 3kT above E
f
is called the Boltzmann tail and can be approximated by
f
(E E )/kT
f(E) e
.
E
F
E
f(E)
0
1
T = 0 K
E
F
E
f(E)
0
1
T = 0 K
E
F
E
f(E)
0
1
T > 0 K
E
F
E
f(E)
0
1
T > 0 K
(higher T)
The carrier distributions for electrons and holes in the conduction and valence bands is found by taking the
density of states for the bands and multiplying them by the probability that the states are occupied (the appropriate
Fermi function).
E
C
E
V
E E
E
F
1  f(E)
f(E)
E
C
(E)f(E)
V
(E)[1f(E)]
energy band
diagram
density of
states
occupancy carrier
distributions
E
C
E
V
E
C
E
V
C
(E)
E
V
(E)
C
(E)
E
V
(E)
The carrier distributions will look different depending on
the position of the Fermi level. The total concentration of
carriers n and p is found by integrating the carrier
distributions over all available energies (the areas under
the curves in red and blue).
These functions cannot be solved analytically in closed
form usually a computer solution is required. In the
degenerate case when the Fermi level is either in the band or within 3kT of the band the exact solution is
required. If, however, the Fermi level is more than 3kT from the appropriate band edge, the Boltzmann tail can be
used as an approximation for the Fermi function. This is the nondegenerate case and is the usual case for silicon
electronic devices. Then we can define an effective density of states for the conduction and valence bands
and the solutions become
There is a set of alternative expressions that can be useful in some circumstances. Remember intrinsic
semiconductors have very few carriers. For the intrinsic case, n = p = n
i
and E
F
= E
i
. From this we can get
alternative solutions
Using the two forms for the solution, we can solve to find
G
E /2kT
i C V
n N N e
= and
2
i
np n = .
Charge neutrality. At reasonable temperatures, there are two kinds of charges fixed charges from donors (N
D
+
)
and acceptors (N
A

) and mobile charges (n and p, electrons and holes) from ionized carriers. For complete
ionization, the usual case at room temperature, N
D
+
= N
D
and N
A

= N
A
, so, from charge neutrality
Case 1: Intrinsic semiconductors
D A
N 0 and N 0
i
n p n
Case 2: Extrinsic semiconductors
D A i
N N n >>
ntype extrinsic doping
2
i
D A D
D
n
N N n N p
N
>>
ptype extrinsic doping
2
i
A D A
A
n
N N p N n
N
>>
Case 3: Compensated semiconductors
D A
N N 0
E
E
F
E
E
F
E
E
F
E
E
F
E
E
F
n
p
In a compensated semiconductor, N
D
and N
A
may be large but, if their difference is small, the material still
appears much like intrinsic material.
Calculations of E
i
and E
F
. In intrinsic materials, n = p and E
F
= E
i
.
The first term is midgap (0.55 eV for silicon) and the second term depends on the difference between electron
and hole effective masses. In silicon, the difference is not large.
For doped materials, solve these equations for E
F
E
i
.
ntype doping
2
i D
D A D F i
D i
n N
N N n N p E E kTln
N n
 
>> =

\ .
ptype doping
2
i A
A D A i F
A i
n N
N N p N n E E kTln
N n
 
>> =

\ .
3. Carrier Motion
Drift current. If a semiconductor with mobile carriers is inserted in an electric field, carriers will drift according
to their sign which results in a current density. For an arbitrary area A, the drift current is given by
drift d
J qnv =
where q is the fundamental charge, n is the
concentration of electrons, and v
d
is the drift velocity.
Current density has units of A/cm
2
. The drift velocity
as a function of electric field , is shown in the figure.
In the low field region (< 10
3
V/cm), the relationship is
linear. We can write this as
d
v =
where is the mobility (different for electrons and
holes) and has units of cm
2
/Vsec. Thus we get
n
drift n
J q n = and
p
drift p
J q p =
The maximum mobility is at low doping levels while
at higher levels scattering effects reduce the mobility.
In silicon the average atomic spacing is about 2.7
and the average range of a mobile electron or hole is ~ 200 .
The average spacing between impurities for a doping level of
10
15
cm
3
is ~ 1000 which means that mobile carriers will
rarely encounter another fixed charge. However, the spacing
between impurities at 10
17
cm
3
is ~ 200 . Thus, for doping
levels above 10
17
cm
3
, the mobile charges will encounter and
be affected by nearby fixed charges. The effect is always to
retard the motion and slow the carriers down reduced
mobility. Increased temperature also reduces carrier mobility.
From electrostatics, we can use the relationship between drift
current and electric field to find the resistivity of a
semiconductor.
1
J = =
where is the conductivity and is the resistivity and the
total drift current is given by
n p
drift drift drift n p
J J J q( n p) = + = +
so
n p
1
q( n p)
=
+
.
Recall that for ntype material, n ~ N
D
>>N
A
and for ptype material, p ~ N
A
>>N
D
.
Diffusion current. In addition to drift current, we need to account for uneven distributions of carriers. Any
unbalance in distribution is driven to become balanced. The imbalance leads to a flux, F of particles (units of
1/cm
2
sec) that depends on the concentration difference and is characterized by a diffusion coefficient D (units of
cm
2
/sec). Adding the charge to make it an electrical current density gives
n
diffusion n
p
diffusion p
n
J qD
x
p
J qD
x
.
The total current density simply sums all the contributions electrons and holes and drift and diffusion. Thus,
n n
n drift diffusion n n
p p
p drift diffusion p p
n
J J J q n qD
x
p
J J J q p qD
x
= + = +
= + = +
n p
J J J = +
At equilibrium, you can derive from these equations, the Einstein relationships.
and
p
n
n p
D
D kT kT
q q
= =
Carrier generation. Where do carriers come from?
E
C
E
V
E
C
E
V
Impact ionization in a large electric field. Carriers gain kinetic energy from
the electric field and become hot carriers. At high enough energy, the hot
carrier can collide with the lattice and give up enough energy to create another
EHP. Then all three carriers continue to gain kinetic energy from the field.
E
C
E
V
E
G
E
C
E
V
E
G
Shallow impurities binding energies < 3kT. These are usually good
impurities introduced intentionally. Carriers are easily ionized into the
conduction or valence band and only cooling to very low temperatures can
freeze out the carriers.
Deep levels binding energies >>3kT. These are usually bad impurities
introduced unintentionally. Carriers are not easily ionized into the conduction
or valence band and cooling only a little can freeze out the carriers.
E
C
E
V
E
G
E
C
E
V
E
G
Bandtoband generation takes a lot of thermal energy but is quite easy with
incident light having energy above the band gap energy (photogeneration).
Minority carriers. If carriers are generated or injected into a semiconductor material, they are in an excited state
and cannot remain that way. If the carrier concentrations at equilibrium are n
o
, p
o
and the carrier concentrations
under excitation are n, p, the excess carrier concentrations are
o
o
n n n
p p p
=
=
and, since they are electron hole pairs (EHP, n = p. These carriers are subject to the statistics of a large number
of particles and have a characteristic lifetime in the excited state. These lifetimes are
n
and
p
. Consider an ntype
sample and recall that
2
i
D
D
n
n N p
N
If N
D
= 10
15
cm
3
, we can calculate that p 10
5
cm
3
. This is a very small value and is the reason we refer to these
as minority carriers. At low injection levels, n = p<< n
o
, which means that n n
o
and p>> p
o
. This leads to
the important conclusion that it is the minority carriers that matter. Once the source of excess carriers is shut off,
the excess concentration decays according to
and
p n
p p n n
t t
= =
.
So what happens if we create a localized pulse of excess carriers? They will spread out by diffusion (D
p
, D
n
) and
eventually recombine (
n
,
p
). The average distance a carrier will diffuse before it recombines is called the
minority carrier diffusion length and is given by
and
p p p n n n
L D L D = = .
4. PN Junctions
Qualitative behavior. Putting two pieces of semiconductor together of opposite type (doping) will create a pn
junction. To understand how the energy bands look for a pn junction, recall that fixed charges (ionized donors and
acceptors) cannot move physically while mobile carriers (electrons and holes) can move within the bands. In
addition, the Fermi level must be flat if there is no applied voltage at the junction. Before the junction is formed,
the bands look like
E
C
E
V
E
i
E
F
and after junction formation the bands look like this
E
C
E
V
E
i
E
F
qV
bi
qV
bi
The Fermi level is flat so the other bands must bend to provide continuity. The shape of the bends depends on the
doping levels on either side of the junction. A builtin voltage V
bi
accounts for the step from the Fermi level
difference. In order to provide charge neutrality, mobile charges move out of the junction region leaving behind a
depletion region. There are fixed charges in this region but no mobile carriers. Note that energy increases upward
in the figure for electrons and downward for holes.
E
C
E
V
E
i
E
F
pdepletion region ndepletion region
In the ndepletion region, mobile majority carriers (electrons) move to the right. This depletes the region of
mobile charges while positive fixed charges (ionized donors) remain behind. In the pdepletion region, mobile
majority carriers (holes) move to the left which depletes the region of mobile charges while negative fixed
charges (ionized acceptors) remain behind.
Builtin potential. From electrostatics we know that the voltage profile V(x), electric field (x), and charge
density (x) are related.
r o
dV d
dx dx
= =
.
n n n
dn
J 0 q n qD
dx
= = +
solving these together will give the builtin voltage
A D
bi 2
i
N N kT
V ln
q n
(
=
(
.
As a numerical example choose midrange doping i.e. N
A
= N
D
= 10
16
cm
3
. Then V
bi
= 0.7 V. This is the classic
value used in circuit designs for the silicon pn junction builtin voltage.
Step junction. Consider the step junction doping profile shown
in (a) in the figure. This doping profile yields the charge
density profile shown in (b). We can integrate Poissons
equation
r o
d
dx
to solve for the linear electric field shown in (c). Integration of
this profile gives the potential profile shown in (d). which is
given by these equations
2 A
p
r o
2 D
bi n
r o
qN
V(x) (x x) x 0
2
qN
V(x) V (x x) x 0
2
= + <
= >
from continuity at x = 0 and recognizing from charge neutrality that N
A
x
p
= N
D
x
n
, we can solve for x
n
and x
p
and
the total depletion width w = x
n
+ x
p
.
N
D
N
A
x
N
D
N
A
x
x
V
x
V
x
N
D
N
A
qN
D
qN
A
x
p
x
n
x
p
x
n
x
p
x
n
a] doping profile b] charge density
c] electric field d] potential
1
2
r o A D
n p bi
A D
2 N N
w x x V
q N N
(   +
= + =
( 
\ .
In order to include an applied bias, we recognize that a forward bias works to reduce the
builtin potential. This adds a simple voltage term by subtraction.
1
2
r o A D
n p bi
A D
2 N N
w x x (V V)
q N N
(   +
= + =
( 
\ .
We can see the effect of applied voltages forward and reverse biases shown in the
figure in red and blue. These changes have a pronounced effect on the energy band
diagrams and introduce the concept of quasiFermi levels E
Fn
and E
Fp
which are non
equilibrium values of the Fermi level on each side of the junction. Forward bias reduces
the voltage barrier and reverse bias increases the voltage barrier.
Currentvoltage characteristic. The currentvoltage (IV) characteristic of a real pn junction diode is shown in the
figure. Note that the voltage scales are different in forward and reverse bias. In addition to rectification and the
builtin voltage, there is a slope in the curves that arise from series resistance, there are leakage currents, and at
large enough reverse bias there is breakdown. The primary shape of the curve is obtained from the ideal diode
equation (the Shockley equation).
qV/kT
o
I I (e 1) =
2 2
p
n i i
o
n A p D
D
D n n
I qA
L N L N
(
= +
(
(
E
C
E
V
E
i
E
F
zero bias
E
C
E
V
E
i
E
Fn
reverse bias
E
Fp
E
C
E
V
E
i
E
Fn
reverse bias
E
Fp
E
C
E
V
E
i
E
Fn
forward bias
E
Fp
E
C
E
V
E
i
E
Fn
forward bias
E
Fp
x
V
x
x
p
x
n
x
p
x
n
x
qN
D
qN
A
x
p
x
n
x
V
x
V
x
x
p
x
n
x
p
x
n
x
qN
D
qN
A
x
p
x
n
V > 0
V < 0
Junction breakdown. Reverse breakdown takes place in different ways in a pn junction diode. One of these is
avalanche breakdown. If the reverse bias is large enough there is a very large electric field in the depletion region
of the diode. Carriers that enter into this region are accelerated by the electric field and can gain significant kinetic
energy and undergo impact ionization described earlier. This effectively multiplies the current with a
multiplication factor M, which can get large quickly resulting in a rapidly growing
reverse current under reverse bias and blows up when the reverse bias equals the
breakdown voltage.
m
o
BR
I
1
M
I
V
1
V
= =
(
(
Zener breakdown is based on quantum mechanical tunneling. In this process,
an energy barrier that is thin enough (~ 100 ) will allow a significant
probability that a carrier can cross the barrier. This can happen under reverse
bias when the doping on both sides of the junction is heavy and the depletion
region is thin.
Junction capacitance. A pn junction under zero or reverse bias behaves like a
parallel plate capacitor with the dielectric constant of the semiconductor.
E
C
E
V
E
i
E
F
depletion region
E
C
E
V
E
i
E
F
depletion region
metal plates
semiconductor dielectric
One important difference is that the depletion width varies with applied bias, so the capacitance for a pn junction
diode is a voltagedependent capacitance. The capacitance of a parallel plate capacitor is given by the simple
formula
r o
A
C
w
=
where
1
2
r o A D
bi
A D
2 N N
w (V V)
q N N
(   +
=
( 
\ .
.
1 The surface of a Si wafer is a (100) plane.
(a) Determine the number of atoms per cm
2
at the surface of the wafer.
(b) Repeat, this time taking the surface of the Si Wafer to be a (111) plane.
2 A crystalline plane has intercept of 1a, 3a, and 1a on the x, y, and z axes, respectively. a is the cubic cell
side length.
(a) What is the Miller index notation for the plane?
(b) What is the Miller index notation for the direction normal to the plane?
3 Semiconductor A has a band gap of 1 eV, while semiconductor B has a band gap of 2 eV. What is the
ratio of the intrinsic carrier concentrations in the two materials (n
iA
ln
iB
) at 300 K. Assume any differences
in the carrier effective masses may be neglected.
E
L
z
E
L
z
E
C
E
V
E
i
E
Fn
E
Fp
E
C
E
V
E
i
E
Fn
E
Fp
4 Determine the equilibrium electron and hole concentrations inside a uniformly doped sample of Si under
the following conditions:
(a) T = 300 K, N
A
<< N
D
= 10
15
/cm
3
.
(b) T = 300 K, N
A
= 10
16
/cm
3
, N
D
<< N
A.
(c) T = 300 K, N
A
= 9 x 10
15
/cm
3
, N
D
= 10
16
/cm
3
.
For each of these conditions, determine the position of E
i
, compute E
F
E
i
, and draw a carefully
dimensioned energy band diagram for the Si sample.
5 The equilibrium and steady state conditions before and after illumination of a semiconductor are
characterized by the energy band diagrams shown in Fig. P3.24. T = 300 K, n
i
= 10
10
/cm
3
,
n
= 1345
cm
2
/Vsec,
p
= 458 cm
2
/Vsec. From the information provided, determine
(a) n
0
and p
0
, the equilibrium carrier concentrations.
(b) n and p under steady state conditions
(c) N
D
(d) What is the resistivity of the semiconductor before and after illumination?
6 More resistivity Questions
(a) A silicon sample maintained at room temperature is uniformly doped with N
D
= 10
16
/cm
3
donors.
Calculate the resistivity of the sample. Compare your calculated result with the p deduced from
GT1.
(b) The silicon sample of part (a) is compensated by adding N
A
= 1016/cm3 acceptors. Calculate
the resistivity of the compensated sample. Exercise caution in choosing the mobility values for
this part of the problem.
(c) Compute the resistivity of intrinsic (N
A
= 0, N
D
= 0) silicon at room temperature. How does your
result here compare with that for part (b)?
7 A Si step junction maintained at room temperature under equilibrium conditions has a pside doping of
N
A
= 2x10
15
/cm
3
and an nside doping of N
D
= 10
15
/cm
3
. Compute
(a) V
bi
.
(b) x
p
, x
n
, and W.
(c) at x = 0.
(d) V at x = 0.
(e) Make sketches that are roughly to scale of the charge density, electric field, and electrostatic
potential as a function of position.
(f) What is the junction capacitance at a bias of 5 V?