Professional Documents
Culture Documents
(반도체공학) Fundamentals of Semiconductor Devices Solutions
(반도체공학) Fundamentals of Semiconductor Devices Solutions
(반도체공학) Fundamentals of Semiconductor Devices Solutions
1.2. Consider a lithium nucleus, of charge +3q. Calculate the first three
electron energies for an electron in a Li++ ion, using the Bohr model.
We repeat the analysis that we used for the hydrogen atom, except that
now the charge of the nucleus Q1 is equal to 3q = +3(1.6 × 10−19 )C . The
results of the key steps are
QQ −3q 2
F = 1 22 =
4πε 0 r 4πε 0 r 2
3q 2
EP (r ) = Evac −
4πε 0 r
mv 2 3q 2
− =0
r 4πε 0 r 2
mvn rn = nh
3q 2 ⎛1⎞
vn =
( 4πε 0 ) h ⎜⎝ n ⎟⎠
( 4πε 0 ) h 2
rn = (n 2 )
3mq 2
m32 q 4 ⎛ 1 ⎞
EK = ⎜ 2⎟
2 ( 4πε 0 ) h ⎝n ⎠
2 2
9mq 4 ⎛ 1 ⎞
E n = EPn + EKn = Evac − ⎜ 2⎟
2 ( 4πε 0 ) h ⎝n ⎠
2 2
1.3. Show that Equations (1.12) and (1.13) follow from (1.8) and (1.11).
mv 2 q2
Equation (1.8) is: − =0
r 4πε 0 r 2
q2
mvr = which from Equation (1.11) is
4πε 0 v
q2
mvr = = nh
4πε 0 v
q2
v= (Equation (1.13))
4πε 0 nh
nh nh ( 4πε 0 ) nh 4πε 0 n 2 h 2
r= = = (Equation (1.12)).
mv mq 2 mq 2
1.3. In each of the potential energy distributions in Figure 1P.1, sketch the
magnitude and direction of the force on the electron.
1.5. Consider the electron in the energy diagram of Figure 1P.2. Taking the
energy the electron has at Point A as Etotal, at each of the indicated positions,
find the total energy, the kinetic energy, the potential energy, and the
electron’s velocity. Indicate the direction of force (if any). Recall that total
energy is conserved. At point “D” the electron collides inelastically with
something (perhaps an atom in the crystal). After the collision, the electron’s
energy is equal to its potential energy, and its kinetic energy is zero. Its total
energy is much less than before the collision; where did the extra energy go?
The electron at Point A is at rest. Its total energy is Etotal, its kinetic energy
and velocity are zero, and its potential energy is equal to its total energy.
The force on the electron is to the right, because the slope of the potential
energy is negative.
At Point B, the total energy of the electron is the same (by conservation of
energy), but the potential energy is EP(B ) = Etotal − 5eV . The kinetic energy
is the difference between the total and the potential energies, so
EK (B) = Etotal − EP (B ) = Etotal − (Etotal − 5eV ) = 5eV . The velocity of the electron
There is no force on the electron because the slope of the potential energy
is zero, but the electron continues moving to the right because it still has
kinetic energy.
At Point D, the electron has made a collision. Its total energy is now 10eV
less than it was- the extra 10 eV of energy was transferred to the atom it
collided with. The potential energy is EP (D) = Etotal − 10eV . The electron’s
total energy is now Etotal(new) = Etotal (original ) − 10eV , which is the same as its
potential energy. Therefore, the kinetic energy of the electron is zero, and
so is the velocity. There is no force on the electron because the slope of
the potential energy is zero.
1.6. Find the kinetic energies in each of the following. Express all your answers
in electron volts.
(a) an electron in the first allowed energy states of the hydrogen atom
(according to the Bohr model, Eq. (1.14))
mq 4 1
We use the relation EK =
2 ( 4πε 0 )
2 2 2
h n
( 9.1E − 31) kg (1.6 E − 19 ) C 4
4
1
EK =
2 ( 4π 8.8E − 12 ) ( F / m ) (1.06 E − 34 ) J s 1
2 2 2 2 2 2 2
= 2.15 E − 18 J
= 13.6eV
(b) a free electron, initially at rest at the back of a cathode ray tube in
your television, accelerated through a potential of 10 kV to strike the phosphor
layer
An electron accelerated through 10kV acquires an energy of 10KeV. Since
this electron was initially at rest, its kinetic energy is now 10 KeV, or
10,000 × 1.6 × 10−19 J = 1.6 × 10−15 J .
(c) a tiny, drifting dust particle, of mass 1 µg and velocity a leisurely 1
mm/s.
1 1
We use EK = mv 2 = ⋅10−9 kg ⋅ (10−3 m / s ) 2 = 5 × 10−16 J = 3.1keV
2 2
Ge: This element is in Column IV of the periodic table. Each atom in the
crystal shares 4 electrons with each neighboring atom, and in return
“receives” 4 electrons. Because every atom is “giving” and “receiving” the
same number of electrons, the charge is evenly distributed between
atoms, and the bonding is covalent.
GaP: Gallium is in Column III and phosphorous is in Column V. Thus in
the bonding, indium has less positive charge than the phosphorus, so the
electrons spends slightly more time near the P atoms than the Ga. This
bonding is largely covalaent but with a slightly ionic flavor.
InGaAsP: Indium and gallium are in Column III and arsenic and
phosphorous are in Column V, so the bonding has the same character as
that of InP, mostly covalent but somewhat ionic.
HgCdTe: Mercury is in Column II, while cadmium and tellurium are in
column VI. This II-VI semiconductor material has a largely ionic but slightly
covalent bonding. The electrons spend quite a bit of their time nearer the
column VI atoms because of those atoms’ greater positive charge.
1.8. For each of the semiconductors below, draw (to scale) the energy band
diagrams:
Si: Eg=1.12 eV. χ=4.05 eV
Ge: Eg=0.67 eV, χ=4.0 eV
GaAs: Eg=1.43 eV, χ=4.07 eV
InP: Eg=1.35 eV, χ=4.35 eV
1.12. Suppose the electron in Figure 1.12 is traveling to the right at constant
energy. What happens to it as it approaches the surface of the material?
Explain your answer, using the energy diagram.
The electron travels at constant energy and thus at constant speed
(constant kinetic energy) in the region where EP is constant. When the
electron approaches the barrier, the potential energy increases. The
kinetic energy thus decreases, and the electron slows down. When it
reaches the edge, its total energy is equal to its potential energy so it
stops. There is a force to the left (gradient in the potential energy) so the
electron is accelerated back to the left. Thus the electron is reflected at
this barrier.
1.13. Show that Equation (1.38) is a solution to Equation (1.37). What is the
significance of the positive and negative signs of K?
2m0 (E − E0 )
Equation (1.38): ψ (x) = Ae jKx + Be − jKx . Letting K = 2 (Equation
h
1.40), the solution can be written as
2m0 (E − E0 ) ⎛ jx ⎞
2m 0 (E − E0 ) 2m 0 (E − E0 )
− jx
h2 h2
⎜ Ae + Be ⎟ =0
h
2
⎜ ⎟
⎝ ⎠
0=0
From Equation (1.40), K can be taken as either the positive or negative
root. If we choose the opposite sign, then from Equation (1.38),
ψ (x) = Ae jKx + Be − jKx , it would be the same as changing the sign of x. Thus
the sign of K reverses the direction of propagation of the wave.
p2
EK = 1eV = 1.6 × 10−19 J =
2m
kg ⋅ m
p = 2(9.11×10−31 kg )(1.6 × 10−19 J ) = 5.39 × 10−25
sec
λ = 0.0123nm
m = 1µ g = 10−9 kg
v = 1mm / s = 10−3 m / s
p = mv = 10−12 kg − m / s
λ = h / p = 6.63 × 10−22 m
λ=h/p=5.26×10-32m
2π
At the bottom of the parabola, K=0, and since λ = , λ is infinity. .
K
Note,however that for a finite crystal, the point at K=0 is not permitted
since the electron wave length cannot exceed the crystal dimension. Since
the kinetic energy is the total energy minus the potential energy (E0), the
kinetic energy is zero.
Kx
5π 4π 3π 2π π π 2π 3π 4π 5π
− − − − −
a a a a a a a a a a
a) Verify that it meets the required criteria:
i) E(K) is periodic in K-space with period 2π/a
ii) Equivalent extrema exist at K=0, ±2π/a, ±4π/a…
iii) Equivalent extrema exist at K=±π/a, ±3π/a, ±5π/a…
iv) The slope of the E-K curve is zero at K=0, ±π/a, ±2π/a, …
first
Brillouin
zone Kx
5π 4π 3π 2π π π 2π 3π 4π 5π
− − − − −
a a a a a a a a a a
c) Sketch the corresponding vg-K diagram.
1 dE
The group velocity vg = is proportional to the slope of E:
h dK
vg
5π 4π 3π 2π π π 2π 3π 4π 5π Kx
− − − − −
a a a a a a a a a a
first
Brillouin Kx
zone
5π 4π 3π 2π π π 2π 3π 4π 5π
− − − − −
a a a a a a a a a a
1.17. Explain the analogy between using a conducting layer thicker than the
skin depth to shield a region of space from electromagnetic waves, and the
ability of an electron to penetrate a potential barrier.
1.18. The infinitely thick potential barrier of Figure 1.18 (a) can be considered a
crude approximation to the potential barrier at the surface of a semiconductor
(see Figure 1.12).
(a) How, then, might you construct a thin potential barrier like that in
Figure 1.18 (b)? Thin potential barriers are used in a wide variety of
semiconductor devices, including tunnel diodes, contacts, and field
effect transistors.
One could place two pieces of semiconductor close together with a narrow
air gap (another insulator will do) between them.
(b) How would you construct a potential well (thin region of lower
potential energy bounded by region of higher potential energy?
Potential wells are widely used in lasers, photodectors, and
heterojunction bipolar transistors.
In this case one would use a thin layer of semiconductor with air on either
side.
1.19. a) From the Bohr model, what emission wavelength would you expect
for a transition in hydrogen from E2 to E1? Transitions ending at E1 are
collectively called the Lyman series, and are generally found in the ultraviolet
region of the spectrum.
λ= = = 122nm
E 10.2eV (1.6 ×10−19 J / eV )
1.21. In discussing Figure 1.19 (a), we pointed out that in a material with a
band gap of 2.5 eV, an electron near the top of the valence band could not
absorb a photon of energy 2.06 eV, since it would have to end up at a
forbidden energy state.
a) What about an electron deep in the valence band, more than 2.06 eV
below the band edge EV? Why is it unlikely for this electron to absorb the
photon?
The electron must have an empty state at 2.06eV above its current state.
If this electron is deep in the valence band, the desired new state will
exist, but it is most likely occupied, since it is in the valence band.
2.1. Of the two materials whose E-K diagrams are shown in Figure P2.1, which
will have the lowest effective mass for electrons? Which will have the lowest
effective mass for holes? Explain how you arrived at your conclusion.
E E
C C
E E
E E
V V
K K
−π/a π/a K
b) What is the effective mass for an electron near the bottom of the
conduction band?
−1
⎛⎜ d 2 E ⎞
m* = h 2
⎝ dK 2 ⎠
K =0
−1
⎡ d2 ⎤
⎢⎣ dK 2 ( C
m* = h 2
E + E1 sin 2 (Ka))
⎥⎦
K= 0
−1
⎡ d
= h 2 E1 (2asin(Ka)cos(Ka))⎤
⎣ dK ⎦
K =0
[
= h 2 E1 2a(−asin 2 (Ka) + acos 2 (Ka)) ]
−1
K= 0
2
h
=
2E1 a2
⎛⎜ 6.63 × 10 −34 Js ⎞
2
⎝ 2π ⎠
=
2(5eV ⋅1.6 × 10 J / eV )(5 × 10 m)
−19 −10 2
−32
= 2.8 × 10 kg = 0.03m0
c) What is the effective mass for holes near the top of the valence band?
[
= h 2 E2 2a(+asin 2 (Ka) − a cos2 (Ka)) ]
−1
K =0
−h
2
=
2E2 a 2
⎛ 6.63 × 10 −34 Js ⎞
2
−⎜
⎝ 2π ⎠
=
2(4eV ⋅1.6 × 10 J / eV )(5 × 10 m)
−19 −10 2
−32
= −3.5 × 10 kg = −0.04m0
2.3. Explain in your own words why the effective mass of an electron is
different than its mass in vacuum.
The effective mass includes the effects of the periodic potential of the
crystal, so that Newtonian or classical mechanics can be used. If a given
external force is applied, the effect on the electron is the sum of the
external force plus all the internal forces due to the varying potential
energy in the crystal. The effective mass can be used to predict the
behavior of an electron in the crystal in reaction to the external force.
2.4. For the material of Figure P2.2, there are two different effective masses.
Which effective mass will be the one displayed by most of the electrons in the
conduction band? Why?
E
V
Most of the electrons in the conduction band will be at the low energies
near the bottom of the band. Thus most of them will have the effective
mass corresponding to K=0 in this case.
EV
2 µm
x
a) Find the electric field, and express the result in V/m. What direction
does it point?
Electric field is voltage/distance. The potential energy changes by 1 eV
over a distance of 2µm. Since the voltage ∆V=-q∆EP, the electric potential
change is
⎛ −1eV ⎞
(
∆V = − q∆EP = 1.6 × 10−19 C ⎜) −19 ⎟ = 1V .
⎝ 1.6 × 10 eV / J ⎠
The electric field is
dV −1V
E =− = −6
= −5 × 105V / m .
dx 2 × 10 m
The field points to the left (uphill).
EC EC
0 2π/a 0 2π/a
K <100>
(a) GaAs
(b) Silicon
If K is to be conserved, the total K of the electron plus the photon must be
equal to the K of the electron after the absorption (the photon is
annihilated). Since photons have nearly zero K, the electrons end up at
(almost) the same value of K as when they start, or that transitions on the
E-K diagram must be vertical. For optical emission, electrons must be in
the conduction band and holes must be available in the valence band.
Since most electrons are found at the bottom of the conduction band, the
most probably transition is from there to the top of the valence band,
where the empty states are expected to be. For silicon, the bottom of the
conduction band is at a different value of K than the top of the valence
band, so emission is unlikely.
0.2 eV
E EC
1 eV
EV
0.1 eV
x
i) What is the potential energy of the electron?
The potential energy for an electron near the bottom of the conduction
band is EC.
b) What is the terminal velocity for a hole under the same conditions?
The force on the hole is the same, since the charge is the same (with a
sign change). The acceleration is different because the effective mass is
different:
F F 1.6 ×10−15 N
a= = = = 0.52 × 1016 m / s 2
m * mch *(m0 ) ( 0.34 ) ( 9.1× 10 kg )
−31
2.9 Near the bottom of the conduction band in a semiconductor , the electron
energy can be expressed as E = EC + AK where A is independent of K and is
2
positive.
a) Find an expression for the electron effective mass in terms of A.
−1
⎛ d 2E ⎞ 2
m *= h ⎜
⎝ dK 2 ⎟⎠
dE d 2E
The derivatives are = 2AK ; = 2A
dK dK 2
h2
Thus m *= .
2A
function problem210
Kx = 20;
Ky = 20;
A = 1;
x = -Kx:Kx;
y = -Ky:Ky;
for i= 1:2*Kx+1
for j = 1:2*Ky+1
mesh(x,y,E)
view(40,10);
d 2E d 2E
= 2 A = 2B = 4A . Thus the curvature is greater in the y-
dKx2 dK 2y
direction and the effective mass is smaller for an electron traveling in the y
direction.
(c) What determines which effective mass a given electron actually has in
this crystal?
2.11. For each of the following, identify whether the impurity will produce n-type or
p-type material, or be amphoteric:
a) Phosphorus replacing a gallium atom in gallium arsenide
2.12. Show that the difference between the intrinsic level and midgap is small,
using GaAs as an example. Compare the result to that of silicon.
The difference between midgap and the actual intrinsic level, from
Equation (2.72), is
E + EV 3 ⎛ m* ⎞
Ei − C + kT ln ⎜ dsh * ⎟
2 4 ⎝ mdse ⎠
E + EV
The term C is the average of these two energies, or midgap. The
2
offset of the intrinsic level from this is
(0.026eV )ln ⎛⎝
3 0.48 ⎞ −2
= 3.8 × 10 eV = 38meV
4 0.067 ⎠
While still small compared to the band gap, this is considerably larger than
the case for silicon (from Example 2.4 the result is 1 meV). The larger
offset in GaAs results from the large difference in the effective masses for
holes and for electrons.
2.13. What is the probability that a state at the conduction band edge of intrinsic
GaAs is occupied at room temperature?
1
The probability function is f (EC ) = EC − Ef , but since the GaAs is
1 + e kT
intrinsic, the Fermi level is well below the conduction band edge, so we
can use the Boltzmann approximation:
EC − Ef
−
f (EC ) = e kt
2.14. The donor ground states for tellurium in GaAs are 5.9 meV below the
conduction band edge. (There are two of them because of spin.) At room
temperature, what is the probability that a given ground state is occupied if
the Fermi level is 0.1 eV below EC ?
1 1
f ( ED ) = ED − E f
= 0.094
= 0.026
1+ e kT 1+ e 0.026
2.15. What is the probability of occupancy of a state 0.1 eV above the Fermi
level? Below the Fermi level?
The energy difference E-Ef is 0.1eV, which is greater than 3kT, so the
Boltzmann approximation is valid:
E− Ef −0.1
−
f (E ) = e kT
=e 0.026
= 0.021 or a 2.1% chance of occupancy.
For a state 0.1 eV below the Fermi level, E-Ef=-0.1, which is not greater
than 2.3kT. Therefore we must use the full expression
1 1
f (E) = E −E f = −0.1 = 0.98
1+e kT 1+e 0.026
, or 98 percent of these states are
occupied.
2.16. Verify the definition of non-degeneracy: show that if the Fermi level is
2.3 kT below the conduction band edge, the probability of occupancy of the
lowest energy state in the conduction band is 10%.
We have EC-Ef=2.3kT=2.3(0.026eV)=0.0598eV
1 1
f (E) = E−E f
= 0.0598
= 0.091 which is close to 10%.
1 + e kT 1 + e 0.026
2.17. Calculate the effective densities of states for electrons and holes at room
temperature for
⎝ m0 ⎠ ⎝ 300 ⎠
⎛ m*dsh ⎞ ⎛ T ⎞ 3/ 2
3/ 2
⎟ ⎜ ⎟
−3
NV = (2.54 × 10 )cm ⎜
19
⎝ m0 ⎠ ⎝ 300 ⎠
For silicon, m*dse = 1.09m 0 , thus
NC = (2.54 × 10 )cm (1.09) (1) = 2.89 × 10 cm
19 −3 3/ 2 3/ 2 19 −3
b) GaAs
2.18. Compare the effective density of states for electrons, NC and for holes NV
for GaAs.
3/ 2
⎛ m * kT ⎞
From Equations (2.61) and (2.62), N C = 2⎜ dse 2 ⎟ and
⎝ 2π h ⎠
3/ 2
⎛ m * kT ⎞
N V = 2⎜ dsh 2 ⎟ . Thus
⎝ 2πh ⎠
3/ 2
⎛ m* kT ⎞
2⎜ dse 2 ⎟ 3 3
NC ⎝ 2π h ⎠ ⎛ mdse
*
⎞ 2 ⎛ 0.067m0 ⎞ 2
= = ⎜ * ⎟ =⎜ ⎟ = 0.052
⎝ mdsh ⎠ ⎝ 0.48m0 ⎠
3/ 2
NV ⎛ m *dshkT ⎞
2⎜
⎝ 2πh 2 ⎟⎠
or the effective density of states in the conduction band is only about 5%
of that in the valence band.
⎛ E f −E ⎞ E f −E E f −E
⎜ ⎟ =1 −
kT kT kT
1 e e e
f p (E ) = 1 − E − E f ⎜ E f −E E ⎟ Ef − E E − E f E f −E
= 1 − E f −E
kT ⎜ ⎟
1+ e ⎝e kT
⎠ e kT
+e kT
e kT
e kT + 1
We find a common denominator:
⎛ E f −E ⎞ Ef − E E f −E E f −E
+ 1⎟ + 1− e k T
f p (E ) = 1⎜ E − E
kT kT kT
e e e
− E −E =
⎜ f ⎟ Ef − E
⎜ kT ⎟
f
⎝e + 1⎠ e kT + 1 e kT + 1
1
f p (E ) = E f −E
e kT
+1
− (EC − E f ) − (E f − EV )
2.20 Recall that the equations n0 = NC e and p0 = NV e
kT
are only kT
2.21. (a) Silicon is doped with ND=5×1015cm-3. Find n0 and p0 , and locate the
Fermi level. Draw the energy band diagram.
The material will be n-type. Since the number of donors is much greater
than the intrinsic concentration, and assuming all dopants to be ionized,
⎛ EC − E f ⎞
−⎜ ⎟
−3 ⎝ kT ⎠
then n0=ND. Thus n 0 = 5 × 10 cm = N C e 15
assuming non-degenerate
material (which we will check later).
⎛ND ⎞ ⎛ 5 × 1015 ⎞
So EC − E f = −kT ln ⎜ ⎟ = −0.026eV ln ⎜⎝ ⎟ = 0.22eV
⎝ NC⎠ 2.86 × 1019 ⎠
Since 0.22>2.3kT, the assumption of non-degeneracy is verified.
b) A new batch of silicon is doped with boron to NA=4×1015 cm-3. Find n0 and
p0 and locate the Fermi level. Draw the energy band diagram.
Again we assume non-degenerate material and complete ionization. thus
p0=4×1015, and
⎛N ⎞ ⎛ 4 × 1015 ⎞
E f − EV = −kT ln ⎜ A ⎟ = −0.026eV ln ⎜ = 0.23eV
⎝NV ⎠ ⎝ 3.1× 1019 ⎟⎠
E
C
E
i
E 1.12eV
f 0.23eV
E
V
(ni ) 2
n0 = = 2.9 × 104 cm -3
p0
2.22. Silicon is doped with NA=4×1017cm-3. Find n0 and p0. Draw the energy band
diagram indicating the position of the Fermi level.
ni2 (1.08 × 10 )
10 2
−3
n0 = = = 292cm
4 × 10
17
p0
Once the carrier concentrations are known, we can find the Fermi level
from the majority carrier concentration:
⎛ p ⎞ ⎛ 4 × 1017 ⎞
E f − EV = −kT ln ⎜ 0 ⎟ = − 0.026 ln ⎜ 19 ⎟
= 0.11eV
⎝ NV ⎠ ⎝ 3.1× 10 ⎠
To locate the Fermi level, we need to find n0 and p0. The intrinsic
concentration ni for GaAs is 2.2×106cm-3, so NA and ND are both >>ni.
⎛ n ⎞ ⎛ 6 ×1014 ⎞
EC − E f = −kT ln ⎜ 0 ⎟ = − 0.026eV ln ⎜ 17 ⎟
= 0.17eV
⎝ NC ⎠ ⎝ 4.4 × 10 ⎠
2.24. Ge is doped with 1015 Sb/cm-3. Assuming Sb to be a shallow donor, find n0 and
p0 at room temperature. Draw the energy band diagram.
N D = 1015 cm−3
Then
n0 = N D = 1015 cm -3
ni2 (1.65 × 1013 ) 2
p0 = = 15
= 2.71 × 1011 cm -3
n0 10
2.25. a) Calculate the intrinsic carrier concentration for Ge at room
temperature.
⎛ − Eg ⎞
We use the relationship ni = NC NV ⎜ e 2kT ⎟
⎝ ⎠
For Ge, Eg=0.67eV, resulting in
−3 −3
⎛ − 2(0.026
0.67V
⎞ −3
ni = 1.06 ×10 cm ⋅ 4.01× 10 cm ⎜ e
19 18 eV )
⎟⎟ = 1.64 × 10 cm
13
⎜
⎝ ⎠
b) Make a semilogarithmic plot of nI vs. temperature over the range 200-
500K. The intrinsic carrier concentration should be on the vertical
(logarithmic) axis. Add curves for silicon and GaAs.
Note that the densities of states are functions of temperature.
ni2 (1.08 × 10 )
10 2
−3
n0 = = = 1.3 × 10 cm
4
9 × 10
15
p0
To check the assumption of non-degeneracy, we find the Fermi level:
⎛ p ⎞ ⎛ 9 × 1015 ⎞
E f − EV = −kT ln ⎜ 0 ⎟ = (−0.026eV )ln ⎜ = 0.2eV , which is also
⎝ NV ⎠ ⎝ 3.1 × 1019 ⎟⎠
much greater than 2.3kT.
16 -3 15 -3
2.27. A sample of InP is doped with ND=10 cm and NA=10 cm .
a) Find n0, p0.
Assuming room temperature, all dopants are ionized. The material is net
n-type, so we solve for the electrons concentrations first. We have
n0= N D' =ND-NA=1016-1015=9.0×1015 cm-3.
b) Locate Ef.
c) Sketch the energy band diagram. Label both axes and identify EC, EV Ef, Ei,
and Evac on your drawing. Indicate the electron affinity and the band gap.
We look up the band gap for InP and find it to be 1.35 eV, and the electron
affinity χ is 4.35 eV. Thus the energy band diagram looks like:
3.2 × 10
15
n0
⎛ n ⎞ ⎛ 3.2 × 1015 ⎞
EC − E f = −kT ln ⎜ 0 ⎟ = −0.026eV ⋅ ln ⎜ = 0.23eV
⎝NC⎠ ⎝ 2.86 × 1019 ⎟⎠
2.29. How heavily would you need to dope silicon with donors to violate the
assumption of non-degenerate doping? How many acceptors would be needed
to just cause a degenerately doped type material?
The threshold of degenerate doping occurs when EC-Ef=2.3kT. At this
threshold, we will take the Boltzmann approximation to apply (just barely).
Thus
−(EC − EF )
E f − Ei
n0 = ni e kT
Ei − E f
p0 = ni e kT
n0 p0 = ni (Equation (2.66)).
2
Thus
n2
p0 = i = 1.2 × 1014 . Assuming all dopants are ionized, from Equation
n0
(2.90),
p0 + N D = n0 + N A
N A = p0 + N D − n0 = 1.2 × 1014 + 1015 − 106 = 1.1 × 1015
2.32. Manganese makes a donor trap state 0.53 eV below the conduction band
edge in silicon. If the silicon is doped with ND=4x1016 cm-3, what is the
probability of occupancy of the trap state? Assume the concentration of Mn is
small enough not to affect the overall doping and that the trap state is single-
valued.
We find the Fermi level:
⎛ n0 ⎞ ⎛ 4 × 1016 ⎞
EC − E f = −kT ln ⎜ ⎟ = −0.026 ln ⎜ 19 ⎟
= 0.17eV
⎝ NC ⎠ ⎝ 2.86 × 10 ⎠
The probability of occupancy of the Mn trap state is
1 1 1
f (E Mn ) = E Mn − E f = ( E C − E f ) −( E C − E Mn ) = 0.17− 0.53 ≈1
1+ e kT
1+e kT 1+e kT
2.33. A sample of Si doped with phosphorus has its Fermi level 0.2 eV below
EC and the donor ground state is 0.045 eV below EC. Find n0.
Since the donor level is well above the Fermi level, we can take all the
dopants to be ionized. Thus
(E C −E f ) 0.2 eV
− 19 − −3
n 0 = N C e kT = 2.86 × 10 e 0.026 eV = 1.3 × 10 cm
16
2.34. The probability that an energy state at the conduction band edge ( EC ) of
Si is occupied is 10-4.
a) Is this Si n-type, p-type, or intrinsic?
⎛ 1 ⎞
Thus EC − E f = kT ln ⎜ −4 − 1⎟ = 0.026eV ln ( 9,999 ) = 0.24eV . Since the
⎝ 10 ⎠
band gap of silicon is 1.12 eV, this place the Fermi level above midgap
and the material is n-type.
b) Find ND-NA.
Assuming ND-NA>>ni (which it will be since the Fermi level is well above
the intrinsic level),
( EC − EF ) 0.24
− −
−3
N D − N A = n0 = N C e kT
= 2.86 × 10 cm e
19 0.026
= 2.8 × 1015 cm −3
⎝ m0 ⎠ ⎝ 300 ⎠
2 2
m *
⎛ NC ⎞ ⎛ 7 × 1017 ⎞ 33
=⎜dse
⎟ = ⎜ 19 ⎟
= 0.09
m0 ⎝ 2.54 × 1019 ⎠ ⎝ 2.54 × 10 ⎠
or mdse = 0.09 m0
*
2.36. Given two semiconductors A and B, let them have the same density-of-
states effective masses. Let Eg = 1 eV for A and 2 eV for B. Find the ratios of the
intrinsic carrier concentrations.
From Equation (2.67),
E gA
Eg − E gB − E gA 2−1
− n N CN V e 2kT
n i = N CN V e 2kT
. Thus iA = EgB =e 2kT
=e 2(0.026)
= 2.2 × 108
niB −
2kT
N CN V e
In the p-region
⎝ 300 ⎠ ⎝ 300 ⎠
and similarly
3 3
⎛ 218 ⎞ 2 19 ⎛ 218 ⎞
NV = [ NV (T = 300 K ) ] ⋅ ⎜
2
−3
⎟ = 3.1× 10 ⎜ ⎟ = 1.92 × 10 cm
19
⎝ 300 ⎠ ⎝ 300 ⎠
Neglecting any changes in band gap with temperature, we find
⎛ Eg ⎞ 1.12 eV
− ⎜⎜ ⎟⎟ −
n = N C NV e
2
i
⎝ kT ⎠
(
= 1.77 × 10
19
)(1.92 × 10 )e 19 8.62 ×10 −5 eV / K (218K ) )
1.12 eV
( )(1.92 × 10 )e
−
= 1.77 × 10 19 19 1.88 ×10−2 e V
= 4.55 × 1012 cm−6
ni = 2.1 × 106 cm −3
Then
n 2 ( 2.1× 10 )
6 2
n0 = i = = 4.9 × 10−4 cm −3
p0 9 ×10 15
⎝ 300 ⎠ ⎝ 300 ⎠
⎛ Eg ⎞ 1.12e V
− ⎜⎜ ⎟⎟ −
( )(5.21× 10 )e eV / K (423K ) )
−5
⎝ kT ⎠
n = N C NV e = 4.79 × 10
2 19 19 8.62×10
i
= 1.14 × 10 26 cm −6
ni = 1.07 × 1013 cm−3
n0 = i = = 1.3 ×1010
p0 9 × 10 15
EC+kT/2 as indicated in Figure 2.17, and the average electron kinetic energy is
at 3kT/2. For intrinsic material, find the probability at a state at these two
For the average kinetic energy of 3kT/2, the probability of the state being
occupied is
⎛ 3(0.026) ⎞
⎜ + 0.56 ⎟
⎡ 3 kT ⎤
⎢ + ( EC − Ei ) ⎥ −⎜ 2 ⎟
⎣ 2 ⎦ ⎜ 0.026 ⎟
− ⎜ ⎟
f (E) = e kT
=e ⎝ ⎠
= 1 × 10−10
⎛ 0.026 ⎞
⎜ + 0.56 ⎟
KT −⎜ 2 ⎟
+ ( EC − E f )
− 2 ⎜ 0.026 ⎟
KT ⎜ ⎟
For E = EC + , fE = e =e = 2.7 × 10−10
KT ⎝ ⎠
2
For a state at 10kT above the conduction band edge, the probability of
finding an electron at such a high energy is
⎡⎣10 kT + ( EC − Ei ) ⎤⎦ ⎛ 10×.026 + 0.56 ⎞
− −⎜ ⎟
f (E) = e kT
=e ⎝ 0.026 ⎠
= 2 × 10−14
2.39. Calculate the electron concentrations in Si and GaAs if the Fermi level is
0 = e kT ⎢ E K ⎜ − ⎟ + ⎜ ⎟⎥
⎢ ⎝ kT ⎠ ⎝ 2 E ⎠ ⎥
⎣ K ⎦
⎛ 1 ⎞ ⎛ 1 ⎞
EK ⎜ − ⎟ = − ⎜ ⎟
⎝ kT ⎠ ⎝ 2 EK ⎠
⎛ 1 ⎞ 1
EK ⎜ 2 2 ⎟ =
⎝ k T ⎠ 4EK
(kT )
2
E2
K =
4
1
EK = kT
2
EK =
∫
0
∫ n(E )dE
∞
K K
0
Substituting in Equation (2.84) as in the example, yields
− EK
∞
∫ EK3/ 2e kT
dEK
EK = 0
−E K
∞
∫
1/ 2 kT
E e K dE K
0
The numerator and denominator are both in the form
x me ax m m −1 ax
∫ e dx = a − a ∫ x e dx
m ax
x
For us m=3/2 (in the numerator), x=EK,. and a=-1/kT.
∞
− EK
3/ 2 k T − EK
E e 3/ 2 ∞
− ∫
K 1/ 2 kT
EK e dEK
(−1 / kT ) (−1/ kT ) 0
EK = 0
−E K
∞
∫0
EK1/2 e kT
dEK
⎢ −1 / kT − −1/ kT ⎥ − −1 / kT ∫0 EK e dEK
1/ 2 kT
⎢⎣ ( ) ( )⎥ (
⎦
)
EK = − EK
∞
∫ K E 1/ 2 kT
e
0
dE K
−E K
[ ]
3 ∞
0 − 0 + kT ∫ EK1/2 e kT
dEK
EK = 2 0
−E
∞ K
∫ 0
E 1/K 2e kT
dEK
3
= kT
2
2.42. Suppose you were to dope some silicon with ND=NA=1016 cm-3.
a) Where do you expect the Fermi level to be?
Since the number of donor and acceptors is the same, we expect the
Fermi level to be at the intrinsic level.
If the Fermi level is at the intrinsic level, then it must the case that n0=p0
=ni.
N A + n0 = N D + p0
10 + n0 = 10 + p0
16 16
∴n0 = p0
We also know that n0 p0 = ni2 , thus n0=ni and p0=ni, as we predicted. What
is happening physically is that the electrons from the donor states are
2.43. Plot and compare ni for Si as a function of temperature from 250 K to 500
K as calculated from Equation (2.88) with that calculated from Equations (2.85),
(2.86), (2.63), and (2.64). Neglect the small temperature dependences of the
density of states effective masses.
Two ways: combining
3 3
19 ⎛ m dse ⎞ ⎛ T ⎞ 2
Eg *
− 2
n i = N CN V e , N C = 2.54 × 10 ⎜
kT
⎟ ⎝ ,
⎝ m0 ⎠ 300 ⎠
3 3
⎛ m* ⎞ 2 ⎛ T ⎞ 2 4.73 × 10−4 T 2
N V = 2.54 × 1019 ⎜ dsh ⎟ , and Eg = 1.17 −
⎝ m0 ⎠ ⎝ 300 ⎠ T + 636eV
p0 = + ( N D − N A ) + ni2 ⎤ 2
2
2 2 ⎣ ⎦
1
1014 − 1015 1 ⎡ 14
( ) + ( 4 × 10 ) ⎤2
2 2
= + 10 − 1015 15
2 ⎢
2⎣ ⎥⎦
−3
= 1.6 × 10 cm
15
and
ni2 ( 4 × 10 )
2
15
n0 = = = 1.6 × 1015 cm −3
p0 1.6 × 1015
Thus at this temperature the material is nearly intrinsic.
3.1. Calculate the speed of an electron in Si with kinetic energy 0.013 eV. Draw
an equilibrium energy band diagram for silicon and indicate where this
electron will be. Compare your calculated thermal speed to the typical drift
velocities cited in the text of 104 cm/s. How does it compare to typical
saturation velocities?
Even with this very small kinetic energy, the electron’s instantaneous
speed is still about 1000 times faster than the drift velocity. It is
comparable to the drift saturation velocity for silicon of 1×107cm/s.
EK=0.013eV
EC
Eg=1.12eV
EV
3.2. Calculate the resistivity for a uniformly doped silicon sample with 1017
p0 = = = 1.2 × 103 cm −3
n0 1017 cm −3
σ = qµn n + qµ p p
= 1.6 × 10−19 ( 630 ) (1017 ) + 1.6 × 10−19 ( 460 ) (1.2 × 103 )
= 10 ( Ω − cm )
−1
1
The resistivity ρ is ρ = = 0.1 Ω-cm .
σ
3.3. A lightly doped Si sample (ND = 1014 cm-3) is heated from 300 K to 400 K. Is
its resistivity expected to increase or decrease? Explain your answer. Repeat for
For the lightly doped sample, as the temperature is increased the intrinsic
concentration increases. It is still true that N D ni , though, and to
reasonable approximation n0 is constant. For this doping level, however,
lattice (phonon) scattering predominates and mobility decreases with
increasing temperature (Figure 3.8). Thus, the conductivity decreases or
the resistivity increases
or greater than that of intrinsic silicon? How about compared to GaAs? Why?
3.5. Find the electron and hole concentrations and the conductivity of silicon
doped with ND=1016 cm-3 and NA=1019 cm-3 . Be sure to take into account the
σ = qµn n + qµ p p .
This material is p-type and degenerately doped. The electrons are minority
carriers, and their mobility is 250 cm2/V-s, from Figure 3.4 at NA=ND=1019
cm-3. The holes are majority carriers, and µp ≈60 cm2V-1s-1.
⎛ ∆EG* ⎞
⎛ 0.085 ⎞
N 1 ⎜⎜⎝ ⎟
1019
e⎝ 0.026 ⎠ (1.08 × 1010 ) = 99cm −3
kT ⎟⎠ 1 ⎜ ⎟ 2
n0 = A e n = 19
2
−3
10 3.1×10 cm
i 19
p0 NV
σ = qµn n + qµ p p
= 1.6 ×10−19 ⎡⎣( 250cm 2 / V − s )( 99cm −3 ) + ( 60cm 2 / V − s )(1019 cm −3 ) ⎤⎦
= 96 ( Ω − cm )
−1
0.1 µm ×1 µm. The length of the path is 0.1 µm. If the material is doped n-type with
ND=1018 cm-3, what is the current in the sample? What is the current density? These
The cross sectional area is A = 0.1µm × 1.0µm = 10−9 cm2 , and the length is
L=0.1µm=10-5 cm..
Thus
V 2.5V
= 37 ( Ω − cm ) ⋅
−1
J = σE = σ = 9 × 106 A / cm2
L 0.1×10−4 cm
3.7. Calculate the electron drift velocity for ND -NA = 1016 cm-3 in a bar of Si of
I 50 × 10−3 A
From Equation (3.2), I=JA, or J = = = 5 A / cm 2 .
A ⎛ 1cm ⎞
2
1mm 2 ⎜ 2 ⎟
⎝ 100mm ⎠
J n ( drift ) 5 A / cm 2
v= = = 3.1× 103 cm / s
qn 1.6 × 10−19 C (1016 cm −3 )
3.8. Explain how an electron can "collide" with a crystal defect. Hint: the
poor quality crystal will have lower carrier mobilities than a good one.
The defect can create a potential barrier well or trap. This will scatter
electrons and holes the same way a charged impurity does.
doping level N, the electron mobility is larger in p-type Si than in n-type Si.
3.10. With reduced N in problem 3.9 above, the electron mobilities approach
As the doping becomes less, the electrons are less affected by the
impurity miniband, and thus see the same scattering mechanisms as the
electrons in p-type material do.
3.11. Compare the mean free time between collisions for electrons and for
holes in intrinsic GaAs. How do these values compare to those for silicon?
⎛ 1m 2 ⎞
( 0 )⎜ × ⋅ ×
10 4 cm2 ⎠ 0.067 (9.1× 10 kg )(0.8m / V ⋅ s)
⎟
3 2
0.067m 8 10 cm / V s −31
⎝
2
mceµn
*
tn = = =
q 1.6 × 10−19 C 1.6 × 10 −19 C
= 0.3ps
where we obtained the value of µn from Figure 3.7 and mce* from the
table inside the book’s cover. This is comparable to the case for silicon
(0.2ps).
For holes,
⎛ 1m 2 ⎞
*
µp
( 0.34 m0 )⎜ 4 × 10 2
cm 2
/ V ⋅ s × ⎟
mch ⎝ 104 cm 2 ⎠
tp = = −19
= 7.7 × 10−14 s = 0.077 ps
q 1.6 × 10 C
compared to 0.1 ps in silicon.
3.12. For the case of Problem 3.11, find an average gain in kinetic energy
between collisions for electrons and holes for an applied field of 100 V/cm.
⎛ V ⎞ ⎛ 100cm ⎞
qE
(1.6 ×10 C ) ⎜100
−19
⎝
⎟⎜
cm ⎠ ⎝ m ⎠
⎟
vdp = tp = 7.70 × 10−14 s
*
mch ( 0.34 ) ( 9.1×10 kg )
−31
= 4.0 × 102 m / s
(0.34)(9.1 × 10−31 kg )
1 * 2
( 4.0 × 106 m / s )
2
EK = mch vdp =
2 2
⎛ 1J ⎞
= 2.5 × 10−26 J ⎜ −19 ⎟
⎝ 1.6 × 10 eV ⎠
= 0.15µ eV
3.13. Estimate the saturation velocity of electrons in intrinsic GaAs. How does
From Figure 3.9, we see that the saturation velocity for GaAs peaks at
about this value, 2×107 (a very good match), but then decreases and
levels off at about 6×106. The reduction is due to scattering into the other
conduction band minima, invalidating our assumption that the electron
remains in a region in which the concept of effective mass is valid- to go
from one such region to another involves crossing through a region in
which this assumption is not true.
3.14. The electron velocity in Si has its saturation value ( v sat≈ 1 x 107 cm/s)
over the range of 5 x 104 V/cm to 2 x 105 V/cm. Plot the mobility-field (µ - E) and
Mobility µ (cm2/V·s )
dn(x)
Jn(diff ) = qDn . Since the electron concentration is dropping linearly,
dx
the expression for n(x) is
103 cm−3
n(x) = − −4 x , and the derivative is
10 cm
dn −7 −4
= −10 cm .
dx
At NA =1018cm-3, the diffusion constant is, from Figure 3.11 (electrons are
minority carriers since the material is p-type), Dn=9cm2 /s. Thus
dx ⎝ s ⎠
The minus sign indicates that the current is flowing from a region of higher
concentration to a region of lower concentration.
3.16. A material is doped such that the electron concentration varies linearly
across the sample. The sample is 0.5 µm thick. The donor concentration varies
∆N D 1 × 1016 cm −3
n0 = 0 + x= −4
x = 2.0 × 1020 cm −4 ( x )
∆x 0.5 × 10 cm
ni2 ni2
p ( x) = =
n ( x ) 2.0 × 1020 ( x)
For this doping range, the mobility µn is fairly constant, and thus so is Dn.
⎝ V ⋅ s ⎠
c) Find the hole diffusion current density at x-0 and x =0.5 µm. Can the
The hole diffusion coefficient is, from Figure 3.11 for minority carriers,
about Dp =12 cm2/V-s .
which is negligible.
⎛ n ( x) ⎞ ⎛ 2.0 × 1020 x ⎞
EC ( x) − E f = − kT ln ⎜ ⎟ = − 0.026 eV ln ⎜ 19 ⎟
⎝ NC ⎠ ⎝ 2.86 × 10 ⎠
e) At equilibrium, the total current must be zero. Show that there must
We have
Also J=0 at equilibrium. The diffusion currents are not zero; therefore the
drift current is also not zero. The drift current is proportional to the electric
field E, and since q, n, p, and the µ’s cannot be zero, the electric field
cannot be zero.
photon of λ=600 nm (red) by the following materials: Si, Ge, GaAs, InAs, SiC,
GaN, CdS.
3.18. Explain physically why the carrier lifetimes are much less in direct gap
3.19. Light of hν=1.5eV (this in the near infrared, at λ=826 nm) at a power level
lifetime be 10 ps.
assuming the GaAs sample is 1 mm thick, what is the optical generation rate?
d) What are the equilibrium electron and holes densities (in the dark)?
e) What are the excess carrier concentrations when the light is on?
dn
Since this is steady state, =0. From Equation (3.63) we have
dt
∂n 1 ⎛ ∂J n ⎞ ⎛ ∆n ⎞
= ⎜ ⎟ + ⎜ Gop − ⎟=0
∂t q ⎝ ∂x ⎠ ⎝ τn ⎠
If the photons are uniformly absorbed, there is no diffusion current, and
∆n = Gopτ n = 4.2 × 1017 cm =3 s −1 (10 × 10−12 s ) = 4.2 × 106 cm −3
f) What are the recombination rates for electrons and holes when the light is
Since the material is intrinsic, n0=p0=ni, and since the excess electrons
and holes are created in pairs, ∆n=∆p, so
p=p0+∆p=n0+∆n=n=6.4×106 cm-3 .
h. How much does the conductivity of this sample change compared with its
dark value?
We know that
σ = q ( µn n + µ p p ) = q ( µn n0 + µ p p0 ) + q ( µn ∆n + µ p ∆p )
The mobilities for intrinsic GaAs are, from Figure 3.7, µn=8000 cm2/V-s
and µp=400 cm2/V-s. In the dark, the conductivity is
σ = q ( µn n + µ p p )
= 1.6 × 10−19 C ⎡⎣( 8000cm 2 / V ⋅ s )( 6.4 × 106 cm −3 ) + ( 400cm 2 / V ⋅ s )( 6.4 × 106 cm −3 ) ⎤⎦
= 8.6 × 10−9 ( Ω − cm )
−1
showing that the conductivity nearly triples when the light is on in this
case.
shifted further into the infrared, at E=hν=1 eV (λ=1240 nm). What is the
It is zero. These photons have energy smaller than the band gap and
cannot be absorbed.
3.20. Solve Equation (3.72) to prove that Equation (3.73) is a solution under the
Equation (3.72) is
d∆n ∆n
= Gop −
dt τn
Recognizing that
dx 1
∫ = ln (a + bx ),
a + bx b
d∆n ⎛ ∆n ⎞
∫1− ∆n = Gop τ n ln ⎜
⎝
1−
Gopτ n ⎟⎠ ∫ op
= G dt
Gopτ n
Rise Time:
Using the initial condition at t=0, ∆n=0 and solving for ∆n, gives after a bit
of algebra,
⎛ − ⎞
t
∆n(t ) = Gopτ n ⎜ 1 − e ⎟
τn
⎝ ⎠
which is Equation (3.73).
Fall time:
J
10−3 ×1cm 2
N= s ⋅ cm 2
= 2.5 × 1015 photons / s
( 2.48eV / photon ) (1.6 ×10 J / eV )
−19
surface, what are the excess carrier concentrations ∆n and ∆p in this region?
and ∆p = ∆n .
more electrons and holes in that part than elsewhere. What might be another
way to produce more electrons and holes in one location, without applying an
One could heat one end of the bar, increasing ni locally. Thus electrons
would be diffusing from the warm end (higher concentration) to the cool
end, producing a net electron diffusion current. But, holes will be diffusing
in the same direction as the electrons, so the hole diffusion current is
going in the opposite direction. The net diffusion current is zero.
position.
The electron current flows toward the center of the bar, and the hole
diffusion current flows away from the center.
c) Explain why a plot of the voltage versus position along the length of the
3.24. a) Find the conductivity of a sample of GaAs doped with NA=1017 cm-3.
ni2 (2.2 × 10 cm )
6 −3 2
−5 −3
n0 = = 17 −3 = 4.8 × 10 cm →n0=4.8×10-5 cm-3
p0 10 cm
µn≈5300 cm2/V·s and µp≈230 cm2/V·s. Plugging into the expression for σ0,
we have
[ ]
σ 0 = 1.6 × 10 −19 C (5300cm2 / V ⋅ s)(4.8 × 10−5 cm−3 )+ (230cm 2 / V ⋅ s)(1017 cm −3 )
= 3.68(Ω − cm )−1
When electrons are generated optically, for each electron that is created
there is a hole created too. Therefore the excess hole concentration is
∆p=1016 cm-3 also.
σ = q ( µn n + µ p p ) = q ( µn (n0 + ∆n) + µ p ( p0 + ∆p ) )
= 1.6 ×10−19 C ⎡⎣5300cm 2 / V ⋅ s ( ≈ 0 + 1016 cm −3 ) + 230cm 2 / V ⋅ s (1017 + 1016 cm −3 ) ⎤⎦
illumination, ∆n=∆p = 5 x 1016cm-3. Find the quasi Fermi levels for electrons
and holes. Compare these locations to the Fermi level when the light is off.
⎛N ⎞
From Equation (3.86), E fn = EC − kT ln⎛⎝ C ⎞⎠ and E fp = EV + kT ln⎜ V ⎟ . We
N
n ⎝ p ⎠
know that n=n0+∆n. The material is n-type, so n0 =ND-NA =10 . Since this
17
ni2 (1.08 × 10 )
10 2
−3
p0 = = = 1.2 × 10 cm
3
17
n0 10
The light moved the quasi-Fermi level for electrons (majority carriers) by
0.01eV.
⎛N ⎞ ⎛ 3.1 × 1019 ⎞
E fp − EV = kT ln ⎜ V ⎟ = 0.026 eV ln ⎜ 16 ⎟
= 0.167eV
⎝ p ⎠ ⎝ 5x10 ⎠
With no light, the Fermi level was 0.147 eV away from the conduction
band edge, and with light the quasi Fermi level for holes is close to the
valence band edge- it has moved by 0.8eV.
3.26. A sample of InP is doped such that Ef –EV=0.2 eV. It is also illuminated
such that ∆n=∆p=103cm-3. Find the quasi-Fermi levels and sketch the energy
ni2 (1.97 × 10 )
7 2
−3
n0 = = = 0.12cm
3.1 × 10
15
p0
The total carrier concentrations are:
n = n0 + ∆n = 0.12 + 103 ≈ 103 cm −3 ; p = p0 + ∆p = 3.1× 1015 + 103 ≈ 3.1× 1015 cm −3
From which we can find the quasi -Fermi levels:
⎛ n ⎞ ⎛ 103 ⎞
EC − E fn = − kT ln ⎜ ⎟ = −0.026eV ⋅ ln⎜ 18 ⎟ = 0.91eV , and Efp should
⎝ NC ⎠ ⎝ 2.07 × 10 ⎠
be the same as Ef since the majority carrier concentration is almost
unaffected.
When the excess carrier concentrations are 1010, n=n0+∆n≈1010cm-3, and
0.91 eV 0.5 eV
EC
1.43 eV Efn (δn=1010 )
Ei Efn (δn=103 )
E f =Efp
EV
0.2 eV
4.1. Consider a p-type Si sample of length 0.1 µm in which the net doping varies
x − x0 ⎛ N ' (x )⎞
Following the example of Section 4.2.2, = ln⎜ A' 0 ⎟
λ ⎝ N A (x ) ⎠
and λ =
( x − x0 ) = ( 0.1×10−4 cm ) = 2.2 ×10−6 cm
⎛ N' (x ) ⎞ ln (100 )
ln ⎜ A' 0 ⎟
⎝ NA ( x) ⎠
kT 1 ⎛ 1 ⎞
E =− = −0.026V ⎜ ⎟ = −12kV / cm
q λ −6
⎝ 2.2 × 10 cm ⎠
4.2. Consider the graded-composition Si:Ge alloy discussed in Section 4.3. The
equilibrium but there is a field for electrons in the conduction band. Since at
equilibrium there is no net current, the drift electron current must be offset by
an opposing diffusion current. Identify and explain the source of the varying
ni2
constant, n 0 = is not. At the end where the band gap is smaller, ni is
p0
4.3. Find the time required for an electron to traverse the p-region of Example
The electric field is 20 kV/cm. at this field the electrons travel near their
∆x 0.14 ×10−4 cm
Then t = = = 1.6 ps .
v 8.5x106 cm / s
4.4. A graded alloy is manufactured in the AlGaAs system. At x=0, the material
Find the effective electric fields for electrons and holes, and find the true
electric field.
We first draw the energy band diagram by drawing the Fermi level as a
constant, and then constructing the energy band diagram of the two
GaAs end, EC ≅ 1.43/2=0.715 eV above Ef, and Evac is 4.07 eV above EC.
At the AlGaAs end, the band edges are each ≈ 1.92/2=0.96 eV from the
4.07 eV 3.4 eV
EC
0.715 eV 0.96 eV
Ef
EV
0 .0 0 .5 1 .0 1 .5 2 .0
distance (um)
The effective electric field for electrons is, from Equation (4.27),
= 1.2kV / cm
= −1.2kV / cm
EC
0.2 eV
Ef
Ei
EV
0.2 eV
1 µm
(below)
Ef
Ei
ξ
EV
Fp diffusion
Jp diffusion
Fp drift
Jp drift
4.6. For the sample whose energy band diagram is shown in Figure P4.2:
EC
0.2 eV
Ef
Ei
EV
0.1 eV
1 µm
(below)
i. Hole diffusion
ii. Hole diffusion current
iii. Hole drift
iv. Hole drift current
v. Electron diffusion
vi. Electron diffusion current
vii. Electron drift
viii. Electron current
ix. Electric field for electrons
x. Electric field for holes
on the right hand end the material is more heavily n-type, so the hole
concentration should be small, but on the other hand the band gap is
small. The relative size of the “sail” can be estimated by the distance to
4.7. Draw the energy band diagram for GaAs that is doped such that EC-Ef=0.2+
(below)
GaAs to Al0.4Ga0.6As over a length of 10 µm. The sample is not doped. Find the
value of the built-in electric field for electrons and holes. The band gap of
AlyGa1-yAs is given by Eg = 1.43 + 1.425 y (in eV) for y<0.43. Note the “y” here is
The conduction band edge moves upward by half this amount (since the
material is intrinsic) and the valence band moves down by half this
amount. The electric field for electrons is thus 0.285 V/10 µm=285 V/cm
(the conduction band edge has a positive slope) and the field for holes is –
0.285eV E
C
1.43 eV
1.43+0.57=2.0eV
E
f
EV
0.285eV
10 µm
an InP substrate. To make sure the layers will have the same lattice constant as
the substrate, the gallium fraction x is kept equal to 0.47y. The arsenic fraction
is varied among the layers from y=0 to y=0.5. The band gap varies as
Eg = 1.35 − 0.72y + 0.12y 2 eV. Assuming for convenience that the effective
masses are constant at mdse and mdsh , find the ratio of ni(y=0) to ni(y=0.5) from
* *
The layers vary from pure InP to Ga0.235In0.765As0.5P0.5. The band gap of
pure InP is 1.35 eV, and the band gap of the quaternary layer is
Since the effective masses are assumed to be equal, then NC and NV are
So there are 570 times as many intrinsic carriers in the narrower band gap
alloy.
4.10. Consider the Si:Ge alloy discussed in Example 4.2. If the acceptor
concentration varies exponentially from 1018 cm-3 to 2×1016 cm-3, show that the
The electric field due to the grading of the composition was shown to be
20kV/cm in the example. Therefore it remains to find the electric field due
⎛ N (x = 0)⎞ ⎛ 1 × 1018 ⎞
E f − EV (x = 0) = − kT ln ⎜ A ⎟ = −0.026ln ⎜ ⎟ = 0.089eV . This is
⎝ NV ⎠ ⎝ 3.1 × 1019 ⎠
greater than 2.3kT so we can use this result. For the alloy end,
⎛ N (x = 0.05)⎞ ⎛ 2 × 1016 ⎞
E f − EV (x = 0.05 µm) = −kT ln⎜ A ⎟ = −0.026eV ln⎜⎝ ⎟ = 0.19eV
⎝ NV ⎠ 3.1 × 1019 ⎠
The slope in the valence band edge is therefore
∆EV 0.089 − 0.19
= = −2.02eV / µ m or 0.1eV in 0.05 µ m . This results in an
∆x 0.05µ m
effective electric field for holes of -20kV/cm. The resulting energy band
diagram, then, reflects the negative slope in the conduction band edge
due to graded composition and the negative valence band slope due to
the doping. Since, however, the band gap must be maintained at
∆Eg = 0.1eV / 5µm , as the valence band slopes the conduction band must
also slope to compensate.
We draw the energy band diagram by drawing Ef as constant first. Then
we add the valence band edge, with its slope of –2.02 eV/µm. Finally, we
add the conduction band edge such that the band gap decreases by 0.1
eV/µm. The total slope of the conduction band edge is thus equal to the
slope of EV and the change in Eg:
Si Si Ge0.13
0.87
∆EC =-0.2 eV E
E =1.12 eV C
g E =-1.02eV
g
Ef
E
∆EV =-0.1 eV V
enough knowledge to deduce some things from the energy band diagram,
electron drift
electron diffusion
region of electric field
EC
qV bi
Ei
Ef
0.2 eV 0.3 eV
EV
Eg =1eV
hole diffusion
hole drift
c) Sketch the electron and hole concentrations. Are the directions of the drift
equilibrium?
The electron hole and electron drift and diffusion compensate for each
Ei
EC
Ef Ef
δn=0.147 eV EV
Ei
δp=0.209 eV
EV
distance (x)
b) Find nn0, pn0, np0, and pp0. Sketch the carrier concentrations.
nn0 = N D = 1017 cm −3
ni2 (1.08 × 10 cm )
10 −3 2
−3
= = = 1.17 × 10 cm
3
pn0 17 −3
nn0 10 cm
−3
p p0 = NA = 10 cm
16
(1.08 × 1010 cm −3 )
2
ni2
n p0 = = = 1.17 × 10 4 cm−3
p p0 1016 cm −3
or Vbi =0.764V
qVbi=0.955 eV
EC
electron energy
Ei
EC
Ef
Ei δn=0.165 eV E V= E f
δp=0 eV
EV
distance (x)
5.4. Fill in the missing steps to derive Equation 5.35 for the junction width on
the n side of the junction.
V jn N A'
= . From this last equation, and Vj=Vjn +Vjp we can write
V jp ND'
⎛ Vp⎞ ⎛ N' ⎞
V j = V jn + V jp = V jn ⎜1 + jn ⎟ = V jn ⎜1 + D' ⎟
⎜ V ⎟
⎝ j ⎠ ⎝ NA ⎠
Vj 2εV jn
Rearranging, V = n
and substituting this into wn = to get
⎛ N D' ⎞
j
qN D'
⎜ 1 + ⎟
⎝ N A' ⎠
2ε V j
wn =
⎛ N' ⎞
qN D' ⎜ 1 + D' ⎟
⎝ NA ⎠
5.5. A step pn junction diode is made in silicon with the n-side having
N D' =2×1016 cm-3 and on the p-side the net doping is N A' =5×1015 cm-3.
a) Draw, to scale, the energy band diagram of the junction at
equilibrium.
We begin by finding the locations of the Fermi levels on each side of the
junction. On the n side,
⎛n ⎞ ⎛ N' ⎞ ⎛ 2 × 1016 ⎞
EC − E f = − kT ln⎜ 0 ⎟ = − kT ln⎜ D ⎟ = −0.026eV ln⎜ = 0.19eV
⎝ NC ⎠ ⎝ NC ⎠ ⎝ 2.89 × 1019 ⎠
0.31eV
1.12eV 0.37eV
b) Find the built-in voltage, and compare to the value measured off your
drawing in part (a).
⎛ ⎞
kT ⎛ N D' N A' ⎞ ⎜ (2 ×1016 cm −3 ) ⋅ (5 ×1015 cm −3 ⎟
Vbi = ln ⎜ ⎟ = 0.026V ln ⎜ = 0.71V
q ⎝ ni2 ⎠
⎝ (1.08 × 10 10
cm )
−3 2 ⎟
⎠
which agrees reasonably well with the result above. The difference results
from round-off errors.
d) Find the width of the n-side of the depletion region and the p-side of
the depletion region, and the voltage dropped across each side of the transition
region.
2εV j 2(11.8)(8.85 × 10−14 F / cm)(0.71V )
wn = =
⎛ N' ⎞ ⎛ 2 × 1016 cm −3 ⎞
qN D' ⎜1 + D' ⎟ 1.6 × 10−19 C (2 × 1016 cm −3 ) ⎜ 1 + −3 ⎟
⎝ 5 × 10 cm ⎠
15
⎝ NA ⎠
= 9.6 × 10−6 cm = 0.096 µ m
shown below:
3
30x10
ξmax=29000V/cm
Electric Field (V/cm)
25
20
15
10
0
0.0 0.1 0.2 0.3 0.4
Distance fromn x(µm)
Wn =0.098µm W =0.38µm
p
2ε
When x=0.098µm, we switch to
qN 'A
( ) qN 'A
(0.48µm − x )2 − 0.71V . The plot
2
V(x) = + x − x + V(x ) = +
2ε p p
2ε
becomes
Voltage (volts)
0.2
0.0
-0.14 -0.2
-0.4
-0.6
-0.71
-0.8
-1.0
-0.2 0 0.2 0.4 0.6 0.8
xn xo xp
Distance (µm)
Note that this is identical to the shape of the conduction band edge since
for electrons, EC=EP.
h) Draw the energy band diagram for Va=0.5 V.
The new junction voltage is Vj = Vbi − Va = 0.71− 0.5 = 0.21V
W=136µm
qV =5.71eV
j
In the diagrams below, the vertical axis is to scale but the junction widths
are not.
Equilibrium
V j =V -V =0.71-0.5V
bi a
Vj =Vbi -V =0.71-(-5)
a
Reverse Bias
5.6. Consider the equilibrium energy band diagram for a pn junction diode
shown in Figure P5.1 (a).
The junction voltage is 5 V, and the built-in voltage is 0.5 volts, so the
applied voltage is –4.5 V.
5.7. A pn junction is formed in silicon between n-type ( N D' =1018 cm-3) and p-
type ( N A' =1017 cm-3). Find, for equilibrium,
a) wn , wp and w
The junction voltage at equilibrium is
⎝ NA ' ⎠ ⎝ 10 ⎠
= 1.02 × 10 −6 cm = 0.0102 µm
wp = w − wn = 0.112µ m − 0.0102µ m = 0.102 µ m
5.9. A silicon diode has N D' =1017 on the n side and N A' -=1016 on the p side. It is
forward biased at Va=0.5 V.
Dn ⎛ qVkTa ⎞
We use Jn (x p ) = q np0 ⎜ e − 1⎟ . We therefore need to find Dn, Ln, and
Ln ⎝ ⎠
np0, At xp, electrons are minority carriers. Thus from Figures 3.11 and
3.23, Dn≈30cm2/s and Ln≈400 µm.
From this we can find
(1.08 × 1010 cm −3 )
2
ni2
We find npo using n p0 = = 16 −3 = 1.17 × 10 4 cm−3 .
p p0 10 cm
Thus
cm 2
⎛ ⎞ 30
s (1.17 × 10 4 cm−3 )⎡ e 0.026V − 1⎤
qVa 0.5V
Dn
Jn (x p ) = q np0 ⎜ e kT − 1⎟ = 1.6 × 10 −19 C ⋅ ⎢ ⎥
Ln ⎝ ⎠ 0.04cm ⎣ ⎦
= 1.4 × 10 −12 [2.2 × 108 − 1]= 3.1 × 10 −4 A / cm 2
We find µp from Figure 3.22 for a doping of 1017, and find µp=220 cm2/Vs.
For holes in n-type silicon doped at 1017 cm-3, Dp≈12 cm2/s and Lp≈70
µm=0.007cm, from Figures 3.11 and 3.23 respectively. The minority
carrier density is
ni2 (1.08 × 10 cm )
10 −3 2
−3
= = = 1.17 × 10 cm
3
pn0 17
nn0 10
and
cm2
⎛ qVa ⎞ 12
s (1.17 × 10 3 cm −3 )⎡ e 0.026V − 1⎤
0.5V
Dp
J p (xn ) = q pn0 ⎜ e kT − 1⎟ = 1.6 × 10 −19 C ⋅ ⎢ ⎥
Lp ⎝ ⎠ 0.007cm ⎣ ⎦
= 7.2 × 10 −5 A / cm 2
where the plus sign reflects that this current flows from p to n. For
simplicity we have assumed that recombination has its maximum rate in
the entire depletion region. We will need to calculate the junction width w
and to do that we first need Vbi:
⎡ ⎤
kT ⎛ N D' N A' ⎞ ⎢ 10161017 ⎥ = 0.774V
Vbi = ln ⎜ ⎟ = 0.026V ln
q ⎝ ni2 ⎠ ⎢ (1.08 × 1010 cm −3 )2 ⎥
⎣ ⎦
= ⎝ cm ⎠ = 0.20µ m
(1.6 × 10 C )(1016 cm−3 )(1017 cm−3 )
−19
To compare this to the diffusion current, we take the ratio and find
(2)(11.8)⎛⎝8.85 × 10−14
F ⎞ 17
cm ⎠ (10 + 1016 cm −3 )(0.774 − (−5)V
= = 0.91µm
(1.6 × 10 −19
C)(1016 cm −3 )(1017 cm −3 )
JG =
− qni w
=−
(1.6 × 10 −19 C )(1.08 × 10−10 cm −3 )(0.91 × 10−4 cm) −10
= −7.4 × 10 A / cm
2
2(10 s)
−4
2τ 0
−3
nn0 = N D = 5 × 10 cm
' 17
n 2 (1.08 × 10 cm )
10 −3 2
−3
= i = = 2.33× 10 cm
2
pn0 −3
5 × 10 cm
17
nn0
p p0 = 1017 cm −3
(1.08 × 1010 )
2
ni2
n p0 = = 17 = 1.17 × 103 cm −3
p p0 10
⎛ qV ⎞ ⎛ 0.5 ⎞
∆np (x p ) = n p0 ⎜ e kT − 1⎟ = 1.17 × 103 cm−3 ⎜ e 0.026 − 1⎟ = (1.17 × 10 3 )(2.2 × 108 )= 2.6 × 1011 cm −3
a
⎝ ⎠ ⎝ ⎠
⎛ qV a
⎞ ⎛ 0.5 ⎞
∆pn (x n ) = pno ⎜ e kT − 1⎟ = (2.33 × 102 )⎜ e 0.026 − 1⎟ = (2.33 × 102 )(2.2 × 10 8 ) = 5.1× 1010 cm −3
⎝ ⎠ ⎝ ⎠
Jndiff
xp
x
c) Since the total current is constant, and since the difference between
the total current and the minority carrier diffusion current is due to drift of
majority carriers, sketch the majority carrier drift current as a function of
distance on the p-side.
Jp(drift)
Jndiff
xp
x
Equation 5.46:
⎛ EC − E f ⎞
−⎜⎜ ⎟⎟
⎝ kT ⎠
n0 = N C e
−
[EC (x )− E f ]
nn0 (x) = NC e kT
−
[EC (x )− E f −EC n + EC n ]
Thus = NC e kT
−
[(E cn −E f )] [− ECn +E C (x ) ]
−
= NC e kT
e kT
−
[(E cn −E f )]
We recognize nn0 = NC e kT
and thus
−
[EC (x )− EC n ]
nn0 (x) = nn0e kT
which is Equation 5.93.
b) Derive Equation 5.94.
We multiply together Equations 5.93
[
− EC (x )− EV ( x)− E Cn + EVp ]
=N N e
' ' kT
D A
−[EC (x )− EV ( x) ] − − E Cn + E Vp [ ]
=N N e
' ' kT kT
D A e
[ ] [E Cn − E Vp ]
− Eg
= N N e kT e kT
' '
D A
[
− EVp − E Vn ]
=N N e'
D
'
A
kT
[ ]
− qV j
=N N e'
D
'
A
kT
−
[qVj ]
n(x)p(x) = N N e '
D
'
A
kT
qV bi
−
[qVj ]
=ne i
2 kT
e kT
qV bi
−
[q (Vbi −Va )]
=ne i
2 kT
e kT
qV a
=ne i
2 kT
5.12. Consider an n+p junction under reverse bias of 5 V. Let N A' =5×1017 cm-3
and the junction area be 75 µm2.
a) Find the reverse current due to diffusion.
From Equation 5.77, we have
⎛ qVa ⎞ ⎛ 0.026
−5
⎞
J = J 0 ⎜ e − 1⎟ = J 0 ⎜ e
kT
− 1⎟ ≈ − J 0
⎝ ⎠ ⎝ ⎠
⎛ Dn n p0 ⎞
J 0 = q⎜ −19 (
⎢ )(
⎡ 10cm2 ⋅ s 2.33 × 10 2 cm−3 )⎤⎥ = 8.3 × 10
−14
⎟ = 1.6 × 10 C ⎢ ⎥
A / cm2
⎝ Ln ⎠ ⎣
0.0045cm
⎦
The current is Idiff=J0A=(-8.3×10-14A/cm2)(75×10-8 cm2)=-6.20×10-20A.
b) Find the reverse current due to generation.
From Equation (5.92),
1
JG = −
( '
)
ni ⎡⎢ qε N D + N A (Vbi − Va )⎤⎥
' 2
ni ⎡ qε (Vbi − Va )⎤ 2
JG = − ⎢ ⎥
τ 0 ⎢⎣ 2N A' ⎥⎦
From Figure 5.10, Vbi=0.98V for the lightly doped side at 1017. Since most
of the junction is on the lightly doped (p) side, then for τ0 we should use
the electron (minority carrier ) lifetime for τ0, which from Figure 3.21 is
about 23 µs.
Thus
n ⎡ qε (Vbi − Va )⎤ 2
JG = − i ⎢ ⎥
τ 0 ⎢⎣ 2N A' ⎥⎦
=−
(1.08 × 1010 cm −3 ) (1.6 × 10 C)(11.8)(8.85 × 10
−19 −14
)
F / cm (0.98 − (−5))
23µs 2(5 × 10 cm )
17 −3
= 4.7 × 10 −10 A / cm 2
and IG=J0A=(-4.7×10-10A/cm2)(75×10-8 cm2)=3.5×10-16A.
5.13. Note that diffusion coefficient, diffusion length and lifetime depend on
doping. (Equations 3.18, 3.42, 3.82 and 3.76.) Consider a one-sided n+p junction
in silicon.
a) Plot the reverse diffusion current density as a function of doping
for Va =-5 V.
Equation 3.18: µn (N A )= 232 +
1180
0.9
⎛ NA ⎞
1+ ⎜ ⎟
⎝ 8 × 1016 ⎠
Dn kT
Equation 3.42: =
µn q
[ ]
−1
−12 −32
Equation 3.76: τ n = 3.45 × 10 N A + 9.5 × 10 N A
2
⎛ Dn n p 0 ⎞
J 0 = q⎜ ⎟ since pn0 is negligibly small.
⎝ Ln ⎠
JG = −
( '
)
ni ⎡⎢ qε N D + N A (Vbi − Va )⎤⎥
' 2
n ⎡ qε (Vbi − Va )⎤ 2
JG = − i ⎢ ⎥ . We also require Vbi as a function of doping, which
τ 0 ⎢⎣ 2N A' ⎥⎦
Using τn for τ0 since most of the junction is on the lightly doped side, and it
is minority carriers that are of interest. The magnitudes of both currents
are plotted below
-7
10
Current density (A/cm2)
-8
10
-9 generation
10
-10
10
-11
10
diffusion
-12
10
-13
10
14 15 16 17 18
10 10 10 10 10
5.14. A Si junction has N D' =1015 cm-3 and N A =1017 cm-3. The junction area is 100
'
µm2.
a) Find Vbi.
⎛ 15 −3 ⎞
kT ⎛ N 'A N D' ⎞ ⎜ 10 cm ⋅1017 cm−3 ⎟
Vbi = = 0.026V ⋅ ln = 0.71V
q ⎜⎝ n2i ⎟⎠
ln
⎜ 1.08 × 1010 cm −3 2 ⎟
⎝ ⎠ ( )
b) Find I0.
⎛ qVa ⎞
Neglecting generation and recombination, we have I = I0 ⎜ e kT − 1⎟
⎝ ⎠
(1.08 × 1010 cm −3 )
2
ni2
n p0 = = 17 −3 = 1.17 × 103 cm −3
p p0 10 cm
ni2 (1.08 × 10 cm )
10 −3 2
−3
p n0 = = = 1.17 × 10 cm
5
15 −3
nn 0 10 cm
From Figure 3.11 for minority carriers,
Dn=20 cm2/s
Dp=13 cm2/s
w=
(
2ε (Vbi − Va ) N 'A + N D' )
' '
qN N
A D
and JG =
(1.6 × 10 −19
C)(1.08 × 1010 cm −3 )(2.7 × 10 −4 cm )
= 2.3 × 10
−10
A / cm
2
−3
2(10 s)
The generation current is IG=JG·A=2.3×10-16 A.
The total reverse current at –5 volts is
I0+IG=3.1×10−19+2.3×10-16= 2.3×10-16A.
At zero volts, the device is in equilibrium and the current is zero.
5.15. Plot the small signal resistance Rp as a function of applied voltage for Va
positive. Let the reverse leakage current be I0=10-15 A. Use a logarithmic scale
for the resistance axis. Comment on your graph in view of your expectations of
the resistance of a diode.
The differential resistance
−1 −1
⎡ dI ⎤ 1 ⎧⎪ qI ⎛ qVa ⎞ ⎫⎪
Rp = ⎢ ⎥ = = ⎨ 0 ⎜ e kT − 1⎟ ⎬
⎣ dVa ⎦ ⎧⎪ ⎡ qV ⎤ ⎫⎪ ⎪ kT ⎝ ⎠ ⎪⎭
− 1⎥ ⎬ ⎩
d a
⎨ o⎢
I e kT
dVa ⎪⎩ ⎣ ⎦ ⎪⎭
Plotting this with I0=10-15 A and kT/q=0.026 V, we obtain
5.16 Consider a symmetrical junction in GaAs in which N D' = N 'A =1016 cm-3.
a) Find Vbi.
Vbi =
kT ⎛ N 'A N D' ⎞
ln ⎜
(
= (0.026V )⋅ ln ⎢
⎢ )
⎡ 1016 cm −3 2 ⎤
⎥ = 1.16V
⎟ 2⎥
q ⎝ ni ⎠ ( )
2
⎢⎣ 2.2 × 10 ⎥⎦
6
ρL
From Chapter 3, we have R = , so we need to find the resistivity ρ. The
A
current is carried by the majority carriers in the bulk, and their numbers
are the equilibrium concentrations.
On the n-side, we have µn=7000 cm2/v·s from Figure 3.7 at a doping of
1016.
ρ pL (2Ω ⋅ cm)(2 × 10 −4 cm )
Rs(p = = = 800Ω
side )
A 50 × 10−8 cm2
Current (Amperes)
10-2
10-3
10-4
10-5
10-6
10-7
10-8
10-9
0.4 0.5 0.6 0.7 0.8 0.9 1.0
Voltage (V)
5.18. A silicon pn junction has N D' =1017 cm-13 and N 'A =1018 cm-3. The junction
area is 100 µm2.What is the junction capacitance at Va=-5V?
qε N D' N 'A
The junction capacitance is given by C j = A
( )
2 N 'D + N A' (Vbi − Va )
Thus
= 0.036pF
5.19. A junction has N D' =1018 cm-3 and N 'A =1016 cm-3. Both sides are long, and
the fraction of reclaimable charge is 0.5. Compare the magnitudes of the
junction capacitance and the stored charge capacitance at Va=-5 V, 0 V, and +0.5
qε N D' N 'A
Cj = A
( )
2 N 'D + N A' (Vbi − Va )
1
= 2.9 × 10 −14 F
(0.92 − V )
a
Cj(-5V)=1.2×10-14F
Cj(0V)=3.0×10-14F
Cj(+0.5V)=4.5×10−14F
and
⎛D Dp ⎞ 30
I0 = Aq ⎜ n n p0 + pn 0 ⎟ = 100 × 10 −8 ⋅1.6 × 10 −19 ⋅ ⋅1.17 × 104 = 1.4 × 10 −18 A
⎝ Ln Lp ⎠ 0.04
where Dn and Ln were obtained from Figure 3.11 and 3.23 respectively,
and we have neglected the second term in the parentheses since
pn0<<np0.
To compare, we have
V=-5 V Csc≈0 Cj(-5V)=1.2×10-14 F
V=0 Csc=0 Cj(0V)=3×10-14 F
V=+0.5 Csc=1.9×10-14 F Cj(+0.5V)=4.5×10-14 F
EC
EV
a) Identify the emitter, base, and collector on the diagram.
Since the EB junction is forward biased, it must be the one on the left, and
since the CB junction is reverse-biased, it must be the one on the right.
Since they share the base as a common region, that must be the center
region.
b) Is this an npn or a pnp transistor?
npn
c) At the EB junction, are electrons injected or extracted?
Anderson& Anderson 28 February 17, 2007
Solutions Chapter 5
The EB junction is forward biased, and electrons are injected into the
base.
d) At the CB junction, are electrons injected or extracted?
Extracted by the reverse bias.
e) Sketch the electron concentration distribution you expect across the
base.
carrier
carriers concentration
injected distribution
carriers
extracted
EC
EV
f) This thin base region is the source of the term “short base diode.” The
goal in designing a transistor is to ensure that every electron injected into the
base from the emitter eventually ends up in the collector (so that IE≈IC).
Explain how this short-base structure helps. Hint: What else could happen to
the electrons?
The base is so short that electrons are extracted by the collector before
they have a chance to recombine in the base.
5.21. In Figure 5.39, explain how the diode current can be negative when the
applied voltage is zero.
Although the applied voltage is zero, there is still some residual charge
across the internal capacitance of the diode. The negative current is a
result of this capacitance discharging.
⎛ I ⎞
5.22. From Equation (5.120), the storage time is given by t s = τ n ln ⎜ 1+ F ⎟ . From
⎝ IR ⎠
Figure 5.39 (obtained from a SPICE plot), IF=4.2 mA and IR= -0.6 mA. Estimate
the electron lifetime in the p-type region.
3
ID
2
ts≈0.2 µs
Thus, from Equation (5.120),
0.2 × 10 −6 s
τn = = 96ns
⎛ 4.2 × 10−3 ⎞
ln ⎜ 1+
⎝ −0.6 × 10−3 ⎟⎠
⎡ 1
⎤
2kT ⎢ a ⎛ 12εV bi ⎞ 3 ⎥
Vbi = ln ⎢ ⎜ ⎟
q 2n ⎝ qa ⎠ ⎥
⎢⎣ i ⎥⎦
kT 1 −0.026V
From Equation (4.22), E p = − = = −0.72V / µ m = −7.2kV / cm
q λ 0.036 µ m
We first find the grading constant a. One way to obtain it is express the
quantity NA-ND as a function of x, which we do using Equation (4.16):
− (x − x 0 )
(x − 0.2)
λp 18 − 0.036
N A (x) = N A (x 0 )e = 10 e
− (x − x 0 )
(x −.3)
λp −
N D (x ) = N D (x 0 )e = 6 × 1016 e −0.27
1
⎡12 (11.8 ⋅ 8.85 × 10−14 F / cm ) 0.83V ⎤ 3
1
⎡12εVbi ⎤ 3
w=⎢ ⎥ =⎢ ⎥ = 1.9 × 10−5 cm = 0.19µ m
⎣ qa ⎦ ⎢⎣ (1.6 ×10 C )(10 cm ) ⎥⎦
−19 22 −4
f) For the junction width you found in (e), comment on the validity of
the linear approximation used over this distance.
From the figure above in part (c) we find that over a distance of 0.2 µm on
either side of the junction, the approximation is poor.
6.3. In section 6.2.2, it was claimed that hyper-abrupt junctions exhibit a large
fractional change in junction capacitance with applied voltage. Explain
physically why we should expect this to be the case.
6.4. Figure P6.2 shows the equilibrium energy band diagram for a
heterojunction between n-type Semiconductor A (band gap 1.5 eV) and p-type
Semiconductor B (1.0 eV) at equilibrium.
χ=4.0 eV
quantum well
χ=3.5 eV
EC
Eg=1.0eV
Ef
EV
Eg=1.5 eV
over the barrier to the left or to the right, or it can recombine with a hole.
At equilibrium, it becomes:
The discontinuity in EV is
6.6. Repeat the previous problem only now let the GaAs be n-type with δn=0.1
eV and let the AlGaAs be p-type with δp=0.15 eV.
The discontinuity in EV is
Now the glitch is in the valence band, and the barrier for holes is smaller
6.7. Consider the Type I Np heterojunction of Figures 6.8 and 6.9 in which the
net doping N D' and N A' are uniform on the N and p sides respectively. Let εn be
the permittivity on the N side and εp that on the p-side. Solve Poisson's
equation to find the depletion widths wn and wp on the n and p sides and the
QV
E ( x) = ∫ dx
ε
x
qN D' N D'
On the N side, E n = ∫ dx ' = q ( x − xn ) for xn<x<x0
xn
εn εn
xp
−qN A' N A'
On the p side, E p = ∫ εp
dx ' = q
εp
(x p − x ) for x0<x<xp
x
ε nE n ( x0 ) = ε pE p ( x0 )
Therefore
(
qN 'D (x 0 − x n ) = qN A' x p − x 0 or )
N D' w n = N A' w p
qN 'A
( )
2
p side: Vp (x ) − Vp (x p ) = − x −x
2ε p p
qN D' qN D' 2
(x − x n ) =
2
V =n
w
j
2ε n 0 2ε n n
qN 'A
( ) qN A' 2
2
and on the p side V j = x − x0 =
p
w
2ε p p 2ε p p
Substituting in N D w n = N A w p , we find
' '
2
qN D' 2 qN A' ⎛ N D' ⎞ 2
Vbi = w + ⎜ ' ⎟ w n or
2ε n n 2ε p ⎝ NA ⎠
Vbi ⎛ 2ε nε p N A' ⎞
wn = ⋅⎜
qN D' qN D'2 ⎜ 2ε ε N ' ⎟⎟
+ ⎝ n p A⎠
2ε n 2ε p N A'
2ε nε p N A' Vbi
=
qN D' ( ε p N A' + ε n N D' )
Similarly
2ε n ε p N 'DV bi
wp =
(
qN A' ε p N A' + ε n N D' )
And the total junction width is
=
qN D' N A' ( ε n N D' + ε p N A' )
6.8. Show that the junction capacitance per unit area for Problem 6.7 can be
written as
dQV ⎡ qε nε p N A' N D' ⎤
Cj= = A ⎢ ⎥.
dVa
⎣ (
⎢ 2 ε n N D' + ε p N A' (Vbi − Va ) ⎥
⎦)
From Equation (5.108), we have
⎡ d ⎤ ⎡ d ⎤
dQV
= A⎢ ( qN D' wn ) ⎥ = A ⎢ ( qN A' wp )⎥
dVa ⎣ dVa ⎦ ⎣ dVa ⎦
d ⎡ ' 2ε nε p N A (Vbi − Va ) ⎤
'
= A ⎢ qN D ⎥
dVa ⎢ qN D' ( ε p N A' + ε n N D' ) ⎥
⎣ ⎦
⎡ 2ε nε p N A' ⎤ d 1
⎢
= A qN D
'
+⎥ (Vbi − Va ) 2
⎢ qN D' ( ε p N A' + ε n N D' ) ⎥ dVa
⎣ ⎦
⎡ q 2ε nε p N A' N D' ⎤ ⎛ 1 ⎞ 1
( )
−
= A⎢ ⎥ − −
(ε p N A' + ε n N D' ) ⎥⎦ ⎜⎝ 2 ⎟⎠ bi a
V V 2
⎢
⎣
⎡ qε nε p N A' N D' ⎤
= A⎢ ⎥
⎢ 2 ( ε p N A' + ε n N D' ) (Vbi − Va ) ⎥
⎣ ⎦
6.9. For the Si:Ge Nn junction of Figure 6.17, sketch the energy band diagram
that you would expect using the simple electron affinity model (i.e., ignoring
tunneling-induced dipoles and interface states). Discuss the difference in
current flow that would result from the two energy band diagrams (simple
model, and considering the effects of the presence of dipoles). Let EC-Ef=0.1
eV for silicon and 0.15 eV for Ge.
∆EC
EC
∆EV
EV
silicon germanium
Current would flow easily in both directions in the ideal case since there is
no barrier to overcome as there is in Figure 6.17. Although that spike may
be thin enough to support tunneling, the current across the junction would
not be as great as in the ideal case.
Evac
∆EC
EC
EV
silicon germanium
The possibility for electrons in the valence band to tunnel into the
forbidden band exists, and we can expect similar band alterations as a
result.
6.11. Consider an n-GaAs Schottky barrier diode of area 10 µm2 and potential
barrier EB (0) = 1eV .
a. Find the value of I0 from Equation (6.15).
J0 = e
2π 2 h 3
(1.6 × 10 C)(0.067 ⋅ 9.1× 10−31 kg )(1.38 × 10 −23 J/ K ⋅ 300K )
−19 2
exp ⎛ −
1eV ⎞
=
2π 2 (1.06 × 10 −34 J⋅ s ) ⎝ 0.026eV ⎠
3
The equation for the current is Equation (6.14) (but note footnote in text):
⎡ qV a ⎤
I = JA = AJ0 ⎢e nkT − 1⎥ . With J0=1.4×10-11A/cm2 and n=1.3, we obtain the
⎣ ⎦
plot below.
qVbi
Evac
E vac
EC
Ef Ef
E
V
Metal Semiconductor
E
C
Evac
E vac
EC
Ef Ef
E
V
Metal Semiconductor
E
C
Forward bias
Evac
E vac
EC
Ef
Ef
Metal E
V
Semiconductor
E
C
Reverse bias.
6.13. Consider an n-Si:Al Schottky barrier diode for which N D' =1017cm-3 and
the measured built-in voltage is 0.70 V. Let the junction area be 10µm2.
a) Find the depletion width w. Note that this junction can be treated as a
one-sided junction.
We use Equation (5.40) for a p+n junction because the depletion region is
entirely in n-type material:
1
⎡ 2(11.8)(8.85 × 10 −14 F / cm)(0.7)⎤ 2
1
⎡ 2εV j ⎤ 2
−6
w=⎢ ' ⎥ =⎢ ⎥ = 9.6 × 10 cm = 0.096 µm
⎣ D⎦
qN ⎢⎣ (1.6 × 10 −19
C)10 17
cm−3
⎥⎦
( )
1
⎡ 2ε (Vbi − Va ) ⎤ 2 ⎡ 2(11.8) 8.85 × 10 F / cm ( 0.7 − (−5) ) ⎤ 2
−14
w=⎢ ⎥ =⎢ ⎥
⎣ qN D' ⎦ ⎢⎣ ( )
1.6 × 10−19 C 1017 cm −3 ⎥⎦
= 2.7 × 10−5 cm = 0.27 µ m
6.14. A heterojunction is formed between n-type GaAs of N D' =1016 cm-3 and p-
type germanium of N A' =1017 . The measured discontinuities in the band edges
are ∆EC=0.27 eV and ∆EV=0.49 eV.
a) Find the barrier to electrons in eV.
⎛ N D' ⎞ ⎛ 1016 ⎞
In the GaAs, EC − E f = δ n = − kT ln ⎜ ⎟ = −0.026eV ln ⎜ 17 ⎟
= 0.098eV
⎝ NC ⎠ ⎝ 4.4 × 10 ⎠
The energy difference between the conduction band edge in the Ge and
the Fermi level is Eg-δp=0.67-0.096=0.57eV. The barrier is this height,
minus the difference between the conduction band edge in the GaAs and
the Fermi level, or the barrier to electrons=0.57-0.098≈0.47 eV.
qV bi
E
vac
χ=4.07
χ=4.0
barrier for
electrons
0.47eV E
C
0.57eV 0.096eV
Ef
EV
0.098eV ∆E =0.27eV
C
∆E =0.49eV barrier
V for holes =1.24 eV
From the figure, Vbi = χ GaAs + 0.47 − χ Ge = 4.07 + 0.47 − 4.0 = 0.54eV
7.1. In any circuit, the transistor operation can be understood by examining the
superposition of the transistor characteristics and the conditions imposed by the
circuit (the load line). What differentiates, then, a digital transistor circuit from an
analog one?
The difference is in how the circuit is operated. In a digital circuit, the gate
voltage is varied between two values, one that produces cutoff and one in
the sublinear region. In an analog circuit, the gate voltage is normally
varied within the current saturation range, producing proportional changes
in the output (drain) voltage.
7.2. In modern FETs, the gate is usually degenerately doped silicon, whose Fermi
level is essentially at the bottom of the conduction band edge (for an NFET) or at the
top of the valence band edge (for a PFET).
a) Draw an energy band diagram similar to Figure 7.5b, except making the
transistor a PFET.
b) Suppose the device is an NFET, but the gate is made of metal (as was done
in the early days). Draw the energy band diagram. Take ΦM<ΦS.
The energy band diagram is:
We require that the surface be inverted by qφf. That means the band
bending in the semiconductor must be 2φf. Since that much is also
dropped across the oxide, the built-in voltage must be 4φf.
Evac
qφs=2qφf
Vbi
qφ ox=2qφf
E vac
ΦS χS
Φ G=χ G
EC
qφf
Ei
EC=Ef Ef
qφ f
EV
qφs=2qφf
EV
7.6. Consider two silicon MOSFETs, one n channel and the other p channel, with
substrate dopings of 1016cm-3. The NMOS has an n+ gate and the PMOS has a p+
7.7. In the MOS process, structures like the gate of a transistor are used to make
capacitors as well. If the oxide thickness is 4 nm, what area is needed to achieve a
capacitance of 1 pF? The permittivity of silicon dioxide is 3.9ε0.
The capacitance per unit area is
Cox =
'
=
(
ε ox (3.9 ) 8.85 × 10 F / cm
−14
) = 8.63 × 10−7 F / cm2
t ox −7
4 × 10 cm
So the area required to obtain 1 pF is
C 1 ×10 −12 F −6
A= ' = = 1.16 × 10 cm .
2
−7
C 8.63 × 10 F / cm 2
7.8. In MOS processing, the W/L ratio is often intentionally made different from
transistor to transistor. Plot the I-V characteristics for the device of Figure 7.19 with
the W/L ratio changed to 10 instead of 5. Compare this to the results for W/L=5.
Using Equation (7.19), we have, for VDS<VDSsat ,and VGS>VT,:
⎛W ⎞ ⎡ V2 ⎤ ⎛ cm 2 ⎞ ⎡ VD2 ⎤
⎝L⎠
( )
I D = ⎜ ⎟ Cox’ µ ⎢(VGS − VT ) VDS − DS ⎥ = 10 8.63 × 10−7 F / cm 2 ⎜ 500 ⎟ ⎢( G
V ⋅s ⎠⎣
V − VT ) D
V − ⎥
⎣ 2 ⎦ ⎝ 2 ⎦
⎡ V2 ⎤
= 4.31 × 10−3 ⎢(VGS − VT ) VDS − DS ⎥
⎣ 2 ⎦
We will choose VGS-VT to be 1V, 2V, 3V, and 4 V. This only applies up to
VDS=VDSsat for each curve, or up to VDSsat=VGS-VT= 1, 2, 3, and 4
respectively. After that, the current is constant for each curve. The
resulting plot is
7.10. Plot the ID-VDS characteristics for an NFET, using the long channel model, for
which W=10µm , L=1µm, tox=4 nm, VT =0.25V, and the channel length modulation
parameter is λ=0.04V-1. Use VGS= 1V, 2V, 3V, and 4V. Find the output conductance
in saturation for VGS=3V.
WCox’
µ⎡ VDS2
⎤
ID = (
⎢ GS
V − VT ) VDS − ⎥
L ⎣ 2 ⎦
⎛ cm 2 ⎞
(
(10µ m ) 8.6 × 10−7 F / cm ⎜ 500 ) ⎟
V ⋅s⎠ ⎡ VD2 ⎤
⎝
= ⎢(VG − VT ) VD − ⎥
1µ m ⎣ 2 ⎦
⎡ V2 ⎤
= 4.3 × 10−3 ⎢(VG − VT ) VD − D ⎥
⎣ 2 ⎦
WCox’ µ ⎛ VDSsat
2
⎞
I Dsat = ⎜ ⎟
L ⎝ 2 ⎠
7.11. An enhancement NFET with the characteristics in Table 7.1 has a threshold
voltage of VT=1V, a channel length of 1 µm and a width of 5 µm. Considering
velocity saturation, with vsat=5 x 106 cm/s, find the current ID for
a) VGS=0 V, VDS= 1V
Here VG<VT, so the device is below threshold, so the current is essentially
zero.
b) VGS=2V, VDS=1 V
From Equation (7.68),
c) VGS=3V, VDS=1 V
⎡ 1
⎤
vsat ⎢⎛ 2 µlf (VGS − VT ) ⎞ 2 ⎥
VDSsat = L ⎜1 + ⎟⎟ − 1⎥
µlf ⎢⎜⎝ vsat L ⎠
⎢⎣ ⎥⎦
⎡ 1
⎤
⎢⎛ ⎛ cm 2 ⎞ ⎞2 ⎥
⎟ ( 3 − 1)
(5 ×106 cm / s )⎢⎜
2 ⎜ 500
V ⋅s ⎠
⎟ ⎥
= −4 ⎜(
10 cm ⎢ 1 + ⎝
) ⎟ − 1⎥
500
cm 2 ⎢⎜
⎜
(
5 ×106 cm / s 10−4 cm )( ) ⎟
⎟⎟
⎥
V ⋅s ⎢⎜ ⎥
⎢⎣⎝ ⎠ ⎥⎦
= 1.2V
This is greater than our VDS, so now the device is not saturated. The
current is
7.12. An NFET is made with tox=4 nm, L=1µm, W=2 µm, VT=1V, and
µlf = 500cm2 / V ⋅ s .
If the simple model is used, what should the width of the PFET be to get the
same characteristics (apart from polarity). Let the low field mobility for holes be
200 cm2/V·s.
WCox' µ
(VGS − VT ) .
2
From the long channel model (Equation (7.25)), I Dsat =
L
Thus
Wn Cox' µn
(VGS − VT )
2
I Dsatn L Wµ
= =1= n n
I Dsatp W p Cox µ p
'
Wp µ p
(VGS − VT )
2
L
W µ ⎛ 500⎞
Thus W p = n n = 2µm = 5µ m
µp ⎝ 200⎠
7.13. An NFET and a PFET are made on the same chip, using the same process. The
NFET has Cox' =8.6×10-7F/cm2, tox=4 nm, L=0.2 µm, W=15µm, VT=1.5 V, and µlf=500
cm2/V·s. If the PFET is identical except for its mobility (200cm2/Vs) and width, W.
a) What should W be for the PFET to make the characteristics the same as
for the NFET, as predicted by the simple model?
W ⎡ V ⎤
I DNFET = NFET Cox’ µlfNFET ⎢(VGS − VT ) − DS ⎥ VDS
LNFET ⎣ 2 ⎦
= WNFET µlfNFET [ X ] = 15 × 500 [ X ]
⎡ V ⎤
where X = ⎢(VGS − VT ) − DS ⎥ VDS is a constant that is the same for both the
⎣ 2 ⎦
NFET and the PFET.
⎝ ⎠ 1 1 −1 ⎤
2
= ⎢ ( )( ) ⎥
0.2 × 10−4 cm ⎣ 2⎦
= 16mA
c) If velocity saturation is considered, How different are VDSsat and IDsat for
the NFET and the PFET compared to the simple model results? Express your result
as a ratio (e.g., IDsatNFET/IDsatPFET and VDSsatNFET/VDSsatPFET).
For the NFET, we have W=15 µm and L=0.2 µm, with µlf=500 cm2/V·s
⎡ 1
⎤
vsat ⎢ ⎛ 2 µ (V − V ) ⎞ 2
⎥
VDSsatNFET = L ⎢⎜ 1 + ⎟⎟ − 1⎥
lfNFET GS T
µlfNFET ⎝ ⎜ vsat L
⎢⎣ ⎠ ⎥⎦
⎡ 1
⎤
⎢ ⎛ ⎛ cm 2
⎞ ⎞ 2
⎥
⎢ ⎜ 2 ⎜ 500 ⎟ (1V ) ⎟ ⎥
4 × 10 cm / s
6
⎜1 + ⎝ V ⋅s⎠ ⎟ − 1⎥
= 0.2 × 10 −4
cm ⎢
cm 2 ⎢⎜ 4 × 106 cm ⋅ 0.2 × 10−4 cm ⎟ ⎥
500 ⎢ ⎜⎜ ⎟⎟ ⎥
V ⋅s s
⎢⎣⎝ ⎠ ⎥⎦
⎡ 1
⎤
= 0.16 ⎢(1 + 12.5 ) 2 − 1⎥
⎣ ⎦
= 0.43V
For the PFET,
µlfPFET ⎝ ⎜ vsat L
⎢⎣ ⎠ ⎥⎦
⎡ 1
⎤
⎢ ⎛ ⎛ cm ⎞ 2
⎞ 2
⎥
⎢ ⎜ 2 ⎜ 200 ⎟ (1V ) ⎟ ⎥
4 × 10 cm / s
6
⎜1 + ⎝ V ⋅s⎠ ⎟ − 1⎥
= 0.2 × 10 −4
cm ⎢
cm 2 ⎢⎜ 4 × 106 cm ⋅ 0.2 × 10−4 cm ⎟ ⎥
200 ⎢ ⎜⎜ ⎟⎟ ⎥
V ⋅s s
⎢⎣⎝ ⎠ ⎥⎦
⎡ 1
⎤
= 0.16 ⎢(1 + 5 ) 2 − 1⎥
⎣ ⎦
= 0.58V
V 0.43V
The ratio of these is DSsatNFET = = 0.74
VDSsatPFET 0.58V
7.14. A good way to check the validity of a derivation is to verify that it reduces to
the expected result for a particular known case. For example, in the simple model
we neglected velocity saturation, and in the later model we took it into account.
Since the high field that causes velocity saturation occurs in shorter channel devices,
we would expect that the later model would reduce to the simple model for long-
channel devices.
a) Show that in the limit of large L, VDSsat as given by Equation (7.68)
reduces to VDSsat = (VGS - VT) as given by Equation (7.26) for the simple (long
channel) model .
You may find the following information useful:
For x < 1
2!
n(n + 1)x 2
(1 ± x)−n = 1 m nx + m ...
2!
Expanding Equation (7.68) using the relation given yields
⎡ 1
⎤
vsat L ⎢ ⎛ 2 µ lf ( GS
V − VT ) ⎞ ⎥
2
VDsat = ⎢ ⎜⎜1 + ⎟⎟ − 1⎥
µlf ⎝ vsat L ⎠
⎢⎣ ⎥⎦
= ⎜1 + + ⎜ − 1⎟ ⎨ ⎬ + H .O.T ⎟ − 1⎥
µlf ⎢⎜ 2 vsat L 2 ⎝ 2 ⎠ ⎩⎪ vsat L ⎭⎪ ⎟ ⎥
⎣⎝ ⎠ ⎦
If the length L is large, then the term
2µlf (VGS − VT )
becomes small when raised to higher powers and the
vsat L
squared and higher order terms can be neglected.
Thus
v L ⎡⎛ 1 2µlf (VG − VT ) ⎞ ⎤
VDsat = sat ⎢⎜ 1 + ⎟⎟ − 1⎥
µlf ⎢⎣⎜⎝ 2 vsat L ⎠ ⎥⎦
v L ⎛ µlf (VG − VT ) ⎞
= sat ⎜ ⎟⎟
µlf ⎜⎝ vsat L ⎠
= (VG − VT )
b) Show that in the limit of large L, IDsat as given by Equation (7.61)
reduces to that of Equation (7.25) for the simple model.
WCox’ µlf ⎡ 2
VDSsat ⎤
I Dsat =
µlf VDSsat ⎣ ⎢ (V GS − VT ) DSsat
V − ⎥
2 ⎦
L+
vsat
µlf VDSsat
If L becomes large compared to , this term can be neglected in the
vsat
denominator and this expression reduces immediately to Equation (7.25):
WCox' µlf ⎡ 2
VDSsat ⎤
I Dsat = (
⎢ GS
V − V T ) V DSsat − ⎥
L ⎣ 2 ⎦
To see whether L is in fact large compared to the neglected term, we
choose some typical numbers: µlf=500cm2/V·s, VDSsat≈2V, and vsat=4×106
cm/s. Then
µlf VDSsat ( 500 )( 2 )
= = 2.5 × 10−4 cm = 2.5µ m .
vsat 4 × 106
For long channel devices this length is easily exceeded.
Equation (7.72):
⎛ V ⎞
WCox' µlf ⎜ VGS − VT − DS ⎟ (VDS − 2 I D RS )
ID = ⎝ 2 ⎠
, which applies only for
⎡ µlf (VDS − 2 I D RS ) ⎤
L ⎢1 + ⎥
⎣ Lvsat ⎦
VDS<VDSsat.
Multiplying both sides by the denominator:
⎧⎪ ⎡ µlf (VDS − 2 I D RS ) ⎤ ⎫⎪ ⎛ VDS ⎞
I D ⎨ L ⎢1 + ⎥⎬ = WCox µlf ⎜ VGS − VT − ⎟ (VDS − 2 I D RS )
'
⎪⎩ ⎣ Lvsat ⎦⎭⎪ ⎝ 2 ⎠
and multiplying through
µlf (VDS − 2 I D RS ) ⎛ V ⎞ ⎛ V ⎞
ID L + ID = WCox' µlf ⎜ VGS − VT − DS ⎟ VDS − WCox' µlf 2 RS I D ⎜ VGS − VT − DS ⎟
vsat ⎝ 2 ⎠ ⎝ 2 ⎠
Neglecting the ID2 term and collecting all terms in ID on the left,
µlf VDS ⎛ V ⎞ ⎛ V ⎞
ID L + ID − WCox' µlf 2 I D RS ⎜ VGS − VT − DS ⎟ = WCox' µlf ⎜ VGS − VT − DS ⎟ VDS
vsat ⎝ 2 ⎠ ⎝ 2 ⎠
Dividing by the coefficient of ID:
⎛ V ⎞
WCox' µlf ⎜ VGS − VT − DS ⎟ VDS
ID = ⎝ 2 ⎠
µlf VDS ⎛ V ⎞
L+ − WCox' µlf 2 RS ⎜ VGS − VT − DS ⎟
vsat ⎝ 2 ⎠
Take out the L in the denominator, and multiplying the third and fourth
terms in the denominator by vsat/vsat:
⎛ V ⎞
WCox' µlf ⎜ VGS − VT − DS ⎟ VDS
ID = ⎝ 2 ⎠
⎡ µlf VDS WCox' µlf 2 RS vsat (VGS − VT ) V ⎛ WCox' µlf 2 RS vsat ⎞ ⎤
L ⎢1 + − − DS ⎜ ⎟⎟ ⎥
⎣⎢ Lvsat Lvsat 2 ⎜⎝ Lvsat ⎠ ⎦⎥
Collecting terms in VD
Since VDS=50 mV and L=0.5µm, the average field in the channel is 0.1
V/µm=1kV/cm and the device is not in the velocity saturation region. Thus
the expression (Equation (8.4))
WCox' µ0VDS ⎛ VDS ⎞
ID = ⎜ VGS − VT − ⎟
L (1 + θ (VGS − VT ) ) ⎝ 2 ⎠
can be used.
The threshold is extrapolated from the linear region, and using Equation
(8.8),
V 0.05
VT + DS = 0.5V = VT +
2 2
Therefore VT=0.475V.
8.2. A particular MOSFET process produces C'B =10-7F/cm2 and I0 = 4 × 10−20 A , and
a threshold voltage of VT=0.5V. For gate oxide thicknesses of 6.5 nm and 4 nm,
find n and S. Which device is better, and why?
8.4. A CMOS inverter drives a load that consists of the gate of another FET.
The gate area is 0.2µm ×5µm. Find the dynamic power dissipation if the clock
frequency is 350MHz, the supply voltage is VDD=2.5V, and the oxide thickness
is 5 nm. If an MSI circuit (medium scale integrated circuit) has 1000 transistors,
what is the power consumption just due to switching? (Neglect feedthrough
current).
The load capacitance is
εA (3.9)(8.85 × 10 F / cm )(0.2 × 10 )(5 × 10 )
−14 −4 −4
CL = Cox A =
'
= −7 = 6.9 × 10 −15 F
tox 5 × 10 cm
Pdynamic = CL VDD f = (6.9 × 10 F )(2.5V ) (350 × 10 6 s−1 ) = 15 µW per transistor.
−15
2 2
Wp
8.5. A CMOS inverter has = 1.5 . The channel lengths are L=1µm, and
Wn
Wn=10µm. Find the propagation delay time if the load capacitance is 1pF. Let
tox=4 nm, VDD=2.5V, and VGS-VT=2V.
At a channel length of 1µm, this value of Wn/Wp equalized the saturation
currents, and IDsat(n)=IDsat(p).= Cox' Wvsat(VGS-VT-VDSsat)
For tox=4 nm, Cox' =8.63×10-7F/cm.
We can calculate either VDSsat to find IDsat since the IDsat’s have been
equalized:
⎡ 1
⎤
vsat ⎢ ⎛ 2 µ (V − V ) ⎞ 2
⎥
VDSsat ( n ) = ⎜1 + ⎟⎟ − 1⎥
lfn GS T
L
µlfn nFET ⎢⎜⎝ v sat LnFET ⎠
⎢⎣ ⎥⎦
⎡ 1
⎤
=
4 × 10 cm / s −4 ⎢
6
10 cm ⎢⎜1 +
⎛ 2 ( 500 cm 2
/ V ⋅ s ) ( 2V ) ⎞ 2
⎥
⎟ − 1⎥ = 1.16V
⎢⎝ ( 4 × 10 cm / s )(10 cm ) ⎠
500cm 2 / V ⋅ s ⎜ 6 −4
⎟
⎥
⎣ ⎦
td = ⎢ L DD + L DD ⎥ = ⎢ 2 × ⎥ = 0.5ns
2 ⎢⎣ 2 I Dsatn 2 I Dsatp ⎥⎦ 2 ⎢
⎣ 2 ( 2.5 × 10 −3
A ) ⎥⎦
8.6. For the transistor whose ID-VDS characteristics are shown in Figure P8.2,
find the small signal parameters:
∂I D
We have g d = . Finding the slope we have 360µA/0.3V=0.0012 S
∂VDS
b) gd for VDS>VDSsat, VGS=1.1V
In this region the slope is zero, and so is the output conductance.
8.7. A CMOS circuit must operate at 500 MHz. What is the maximum channel
length that can be tolerated, neglecting short-channel effects?
From Figure 8.16, the maximum channel length is about 3.3 µm (we use
the PFET line because the PFETs are slower, assuming equal channel
lengths, so they represent the worst case).
8.8 Practice using circuit models. Suppose a FET was found to have a finite
input resistance, Rin. How would you include this in the model of Figure 8.13
(a)? How would you include the channel resistance?
The input is between the gate and source, so we need to add a resistor
from gate to source. The channel resistance extends between the source
and the drain.
8.9. A designer wishes to design a for a 0.3µm process (meaning that the FETs
have L=0.3 µm). Furthermore, the current fabrication line is limited to tox=4
nm. What should the junction depth xj, be such that the short-channel effects
can be neglected? Let VDD=3.3V, and let N'A in the channel be 1017 cm-3. If the
channel doping is reduced, what is the effect on the allowable junction depth?
If the supply voltage is reduced, what is the effect?
⎝ 0.9 ⎠
The depletion region width at the source end is that of an n+p junction
under zero bias:
⎡ (3.1 × 1019 )(1017 )⎤
1
⎡ 2εVbi ⎤ 2 kT ⎛ NV N A ' ⎞ ⎢ ⎥ = 0.98V
wS = ⎢ ⎥ and Vbi = q ln ⎜⎝ n2 ⎟⎠ = 0.026 ln ⎢
⎣ ( ) ⎥⎦
10 2
⎣ A ⎦
qN ' i 1.08 × 10
so
2 (11.8 ) ( 8.85 × 10−14 ) ( 0.98 )
1
⎡ 2εVbi ⎤ 2
wS = ⎢ ⎥ = = 0.11µ m
⎣ qN A ' ⎦ 1.6 ×10−19 (1017 )
At the drain end, the junction is the same but it has a junction voltage
Vj=Vbi-Va=0.98 + 3.3=4.28V (the junction is reverse biased). Thus
2 (11.8 ) ( 8.85 × 10−14 ) ( 4.28 )
1
⎡ 2εV j ⎤ 2
wD = ⎢ ⎥ = = 0.24µ m
⎣ qN A ' ⎦ 1.6 × 10−19 (1017 )
γ 0.037
then x j = = = 0.075µ m
tox ( wS + wD ) 4(0.11 + 0.24) 2
2
If the doping in the channel is reduced, the depletion regions get wider
and the junction depth needs to be reduced.
If the supply voltage is reduced, then wD will be smaller and the junction
can be a little deeper.
8.11. (a) Draw the energy band diagram for the HFET, perpendicular to the
gate, for the case when the transistor is strongly on . Indicate the polarity of
the voltage applied to the gate.
channel enhanced
EC
ED
∆EC
Ef
EA
Efm
EV
∆EV
semi-
p-GaAs insulating
GaAs
metal n+ AlGaAs
undoped AlGaAs
b) Repeat for the depletion case (no channel). Now what is the polarity
of the gate voltage?
EC
∆EC ED
Efm
Ef
EA
EV
∆EV
metal n+ AlGaAs
semi-
undoped AlGaAs p-GaAs insulating
GaAs
Note that since the undoped layer is only slightly conductive, some of the
applied voltage is dropped across this layer. The size of ∆EC is determined
by the materials and not the voltages, so these steps remain the same.
c) Draw the energy band diagram along the channel for the equilibrium case
and when the channel is conducting.
We observe that the source, drain, and channel are all GaAs- no
heterojunction in this direction. The energy band diagram in this plane
looks the same as for a conventional MOSFET:
8.14. Draw the energy band diagram at the source and the channel end of the
JFET under saturation (e.g., the same type of figure as Figure 8.29 for the
MESFET. Point out the similarities and differences in operation of the two
different types of devices.
8.15. An n-channel JFET has a channel doping of 1016 cm-3. Let the thickness of
the n layer (a) be 1 µm, the length of the channel be 4 µm, and the width be 40
µm.
a) What is the threshold voltage?
Since the gate is degenerately doped, we can assume that the depletion
region is entirely on the n side (channel). We find that the depletion width
is equal to the channel width when w = 1µm , or Vj=Vbi-Va=6.5V from Figure
5.12. Since from Figure 5.10 the built-in voltage is 0.92 V, the applied
voltage at threshold is VT=Va=Vbi-Vj=0.92-6.3=-5.4 volts.
⎧ ⎡ 3
⎤⎫
qW µlf N D' a ⎪ 2 ⎢ ⎛ V − V ⎞ 2
⎥⎪
= ⎨VGS − VT − (Vbi − VT ) ⎢1 − ⎜ ⎟ ⎥⎬
bi GS
I Dsat
L ⎪ 3 ⎝ Vbi − VT ⎠ ⎥ ⎪
⎩ ⎣⎢ ⎦⎭
From Figure 3.4 the low-field mobility at this doping (the electrons are
majority carriers in this case) is about 1100 cm2/V·s.
⎛ cm 2 ⎞
(1.6 × 10−19 C)(40µm )⎜⎝1100 V ⋅ s ⎟⎠ (1016 cm−3 )(10 −4 cm)
I Dsat =
4 µm
⎧ ⎡ 3 ⎤⎫
⎪ ⎢ ⎛ 0.92 − (−2 ) ⎞ 2 ⎥⎪
× ⎨−2 − (−5.3) − [0.92 − (−5.3)]⎢1− ⎜
2
⎟ ⎥⎬
⎪ 3 ⎢ ⎝ 0.92 − (−5.3)⎠ ⎥⎪
⎩ ⎣ ⎦⎭
−3
= 1.76 ×10 (0.42)
= 0.85mA
9.1. For each of the transistors in Figure P9.1, indicate the mode of operation
(forward active, cutoff, saturation, etc.)
9V 9V
4.9 V 5.0V
4.8V 0.2 V
4.1V
(a) (b)
9V
0.7 V 0.7 V
-5 V 0.5V
(c) (d)
-9 V
a) This is an npn transistor. The collector voltage is higher than the base
voltage so the CB junction is reverse biased. The base voltage is higher
than the emitter voltage so that junction is forward biased. Thus this
transistor operates in forward active mode.
c) This one is pnp. The EB junction is forward biased and the CB junction
is reverse biased, so this one is in forward active mode.
transistor
9.3. Draw the energy band diagram for a pnp transistor at equilibrium and
under forward active bias.
The drawing should reflect the relative dopings as well:
-W E WBM x
EC E
C
E ,E E ,E V
f V f
equilibrium
E
C
Fn In
E ,E V
f
Fp
Ip active mode bias
9.4. From Equation (9.13), α=γαTM, where M is the carrier multiplication factor
in the base collector junction. For small base-collector voltage, M=1 and α=γαT
α
and β = . Show that for avalanche breakdown,
1−α
1
M = 1+ .
β
Solution:
α αM
β= = 0
1 − α 1 − α0M
For breakdown, β = ∞ or α M = 1
Anderson & Anderson 3 February 17, 2007
Solutions Chapter 9
β βM
But α = and α M = =1
1+ β 1+ β M
Solving for M,
1
M = 1+
β
9.5. Using the prototype model, (i.e. ignore the apparent band-gap narrowing
and its effect on injection efficiency) find α and β for an npn BJT with
N DE = 1019 cm −3 , N AB = 2 × 1017 cm −3 , and N DC = 1017 cm−3 . Indicate clearly all your
steps. Let VBE=0.8V VCB=2.0 V. The metallurgical widths are WEM = 0.2 µm and
WBM = 0.2 µm.
We can find the minority carrier diffusion constants from Figure 3.11:
DpE=3.8 cm2/s and DnB=15 cm2/s
The built-in voltages for the two junctions are from Equation (5.13):
⎡ ( 3.1 × 1019 )( 2 × 1017 ) ⎤
kT ⎡ NV N AB ⎤ ⎢ ⎥ = 1.00V (one-sided
Vbi ( EB ) = ln ⎢ ⎥ = 0.026 ⋅ ln ⎢
q ⎣ ni2 ⎦
⎣ (1.08 × 10 ) ⎥⎦
10 2
junction) and
⎡ (1017 )( 2 × 1017 ) ⎤
kT ⎡ N DC N AB ⎤
Vbi ( CB ) = ln ⎢ ⎥ = 0.026 ln ⎢⎢ ⎥ = 0.85V
⎣ (1.08 × 10 ) ⎦
q ⎣ ni2 ⎦ 10 2 ⎥
If the forward bias on the EB junction is 0.8V, then the junction voltage is
VJ(EB)=1.00-0.8=0.2V, and the junction on the base side of the emitter
junction extends, from Equation (5.39)
1
⎡ 2ε V j ⎤ 2 ⎡ 2 × 11.8 × 8.85 × 10−14 F / cm × 0.2V ⎤
1/ 2
For a CB junction with 2V reverse bias, the transition region in the base is,
from Equation (5.36)
⎢ ⎥
2ε V j ( BC )
wp ( BC ) =⎢ ⎥
⎢ ⎛ N ⎞⎥
⎢ qN AB ⎜1 + AB ⎟⎥
⎢⎣ ⎝ N DC ⎠ ⎥⎦
1
⎡ ⎤2
=⎢
( )
⎢ 2 (11.8 ) 8.85 × 10−12 F / m ( 0.85 − (−2) ) V ⎥
⎥
⎢ −3 ⎛ 2 × 1017 cm −3 ⎞⎥
⎢ (1.6 × 10 C )( 2 × 10 m ) ⎜1 +
−19
⎟⎥
23
⎣⎢ ⎝ 1017 cm −3 ⎠ ⎦⎥
= 0.079 µ m
The remaining base width is then
WB = WBM − wp ( EB ) − wp ( BC ) = 0.2 − 0.036 − 0.079 = 0.085µ m
We take the emitter width to be equal to the emitter metallurgical width, or
WE=WEM.
From Figure (3.23), LnB=85µm for a doping of NAB=2x1017cm.
To find α, we must find γ and αT:
1 1
γ ≈ = = 0.998
N AB D pE WB 2 × 10 3.8 0.085
17
1+ 1+
N DE DnE WE 1019 15 0.2
WB2 (0.085µ m) 2
and αT = 1 − = 1 − = 0.9999995 ≈ 1
2 ( 85µ m )
2
2 L2nB
In this case α≈γ=0.998
α 0.998
We find β = = = 500 .
1 − α 1 − 0.998
9.6. For the device of Example 9.1, estimate the emitter-base junction width, the
width appearing on the emitter side and that appearing on the base side.
Assume that the junction is forward biased with V j = Vbi − Va = 0.2V . Draw the
emitter and base to scale and indicate the locations of the edges of the
transition region.
x=0 Wn Wp
EB junction CB junction
9.7. Derive the simple model expression for β for a pnp transistor with a non-
degenerate emitter (the pnp equivalent of Equation (9.42).
For a pnp transistor, the injection efficiency would be given by
1
γ= I . Following the derivation in the text for an npn, we would have
1+ nE
I pE
for the forward injection of holes:
dpB qAE DpB[p B (0 + ) − 0 ] qAE DpB pB (0+ )
I pE = − qAE DpB = =
dx x = 0 + WB WB
and for the back injection
dnE qAE DnE [nE (0 − ) − 0] qAE DnE nE (0− )
InE = −qAE DnE = =
dx x =0 − WE WE
and
Anderson & Anderson 6 February 17, 2007
Solutions Chapter 9
1 1
γ= − = −
WB qAE DnEnE (0 ) n (0 ) D W
1+ +
1 + E + nE B
qAE DpB pB (0 ) WE pB (0 ) DpB WE
For the excess carriers we have
qV
+ ni2 kTa
pB (0 ) = ' e
N DB
qVa
n2
nE (0 − ) = i' e kT
N AE
nE (0− ) N 'DB
Thus =
pB (0 + ) N 'AE
and
1
γ= '
N DnE WB
1+ DB
N 'AE DpB WE
From Figure (2.25), the apparent band gap narrowing in the emitter for n-
type material doped to 5x1019 cm-3 is 0.08 eV.
9.9. In Example 9.2, we neglected the band gap narrowing in the base. Find the
current gain β, this time taking ∆EgB
*
into account. Draw the equilibrium
energy band diagram, and indicate the true band edges and apparent band
edges in both the emitter and the base. What is the effective barrier height for
electrons? Holes?
1
From Equation (9.40), we have γ = ∆E *
'
N AE gBE
D pE WB
1+ e kT
NC DnB WE
γ
where ∆EgBE
*
= EgB
*
− EbE
*
. Since β ≅ , following Example 9.3 we find
1− γ
β
γ= .
1+ β
∆EgBE *
N' D W 1
Letting Z = AB pE B e kT , we can write γ = , and
N C DnB WE 1+ Z
1 1
β ≅ 1+ Z = 1+ Z =
1
1 Z Z
1−
1+ Z 1+ Z
∆EgBE
*
N DnB WE −
and β ≅ 'C e kT
.
N AB D pE WB
Anderson & Anderson 8 February 17, 2007
Solutions Chapter 9
barrier to qV
BE
electrons E
g
EC
∆E*
gE Ef
∆EgB
* E
g
E*
gE Eg qVBE barrier
to holes
EV
E B C
9.10. Suppose that a degenerate Si layer of ND=1020 cm-3 and NA=0 is epitaxially
deposited onto the emitter of the transistor of Section 9.4. Explain how such a layer
could increase β slightly. Both emitter layers together are still much shorter than
the diffusion length of either electrons or holes in the emitter. [Hint: There is a small
amount of bandgap narrowing due to the acceptors in the base (and the
compensated acceptors in the emitter)].
We draw the energy band diagram to show the effect of the surface layer:
∆E*g
(due to nB
donors)
E
C
p
E
∆E*g
(due to EV
acceptors)
We see that because there are substantial acceptors in the original emitter
and base layers, there is a small amount of band gap narrowing- affecting
the position of the valence band edge. This effect will not be present in the
surface layer, creating a small energy barrier for holes in the emitter. This
will tend to reduce the back injection of holes from the base by a small
amount, with an attendant increase in β.
N 'DE DnB WB
For an npn transistor β =
N'AB DpE WE
N 'AE DpB WB
For a pnp transistor β =
N 'EB DnE WE
If the transistors are made identically, the first ratio and the third ratio
would be the same The diffusion constants, however, will be different.
Since hole diffuse more poorly than electrons, the second ratio will always
be smaller in the pnp and thus its β is lower.
9.13. If a pnp transistor is considered to have all step junctions, with the
properties in the table below, find:
'
N AE =5×1019 cm-3 '
N DB =4×1017 cm-3 '
N AC =1017 cm-3
WEM=0.13 µm WBM=0.20 µm
VEB=0.8V (forward bias) VCB=2 V (reverse bias)
e) WB
Anderson & Anderson 10 February 17, 2007
Solutions Chapter 9
The emitter-base junction has a junction voltage Vj=Vbi-Va=1.02-
0.8=0.22V. We can find
2εV j 2 (11.8 ) ( 8.85 × 10−14 ) ( 0.22 )
wn ( EB ) = = = 0.027 µ m µm
qN D' 1.6 × 10−19 ( 4 × 1017 )
For the collector- base junction,
1 1
⎡ ⎤ 2
⎡ ⎤2
⎢ ⎥
wn ( CB ) = ⎢ 2ε (VbiCB − VaCB ) ⎥
= ⎢ ( )
⎢ 2 (11.8 ) 8.85 × 10−14 F / cm ( 0.87 − (−2 ) V ⎥
⎥
⎢ ⎛ '
⎞⎥ ⎢ ⎛ 17
⎞ ⎥
⎢ (1.6 × 10 C )( 4 × 10 cm ) ⎜1 +
N AC −19 −3 10
⎢ qN DB ' ⎜1 + ' ⎟ ⎥ 17
17 ⎟ ⎥
⎢⎣ ⎝ N DB ⎠ ⎥⎦ ⎣⎢ ⎝ 4 × 10 ⎠ ⎦⎥
= 0.087 µ m
The final base width is
WB = WBM − wnEB − wnCB = 0.20 − 0.027 − 0.087 = 0.086µ m
f) Neglecting band gap narrowing, find γ
1 1
γ= =
N '
D W 4 × 10 6.1 0.08
17
1 + DB ⋅ nE ⋅ B 1 + ⋅ ⋅
'
N AE D pB WE 5 × 1019 10 0.13
= 0.997
g) Find αT
2
W2 1 ⎛ 0.086 µ m ⎞
αT = 1 − B2 = 1 − ⎜ ⎟ = 0.999996 ≈ 1
2 L pB 2 ⎝ 30 µ m ⎠
i) Find β.
α 0.997
β= = = 332
1 − α 1 − 0.997
9.14. Repeat problem 13, but this time account for band gap narrowing in the
emitter.
∆E *
NV D pB WE − kTg
β= ' e
N DB DnE WB
β= ⋅ ⋅ e = (206)(0.013) ≈ 2.7
4 × 1017 6.1 0.08
∆E gBE
*
N D W −
Equation (9.43) is β = 'C nB E e kT
. From Figure 3.11, DnB=13 and
N AB D pE WB
DpE=3.6 cm2/s.
The band gap narrowing in the emitter is 93meV and in the base, 32 mV
from Figure 2.25.
− 0.032)
2.86 × 1019 13 0.15 − (0.0930.026
Thus β = e = 30
5 × 1017 3.6 0.1
9.16. Consider the graded base transistor whose SIMS profile is shown in
Figure P9.3.
β (η ) η
From Equation (9.60), = . We therefore first find β(0). The
β (0) 1 − e −η
emitter is degenerately doped, with average N DE'
≈ 1019 cm-3. From Figure
3.11, we find for the emitter that DpE=13 cm2/s, and using an average
doping in the base of 5×1016 we obtain DnB≈30cm2/s. From Figure P9.3,
we find WE=0.20µm, and WB=0.4-0.23=0.17µm. Thus
⎡ η ⎤ ⎛ 3 ⎞
β (η = 3.0 ) = β ( 0 ) ⎢ −η ⎥
= 217 ⎜ −3 ⎟ = 685
⎣1 − e ⎦ ⎝1− e ⎠
9.17. A graded based npn transistor has its emitter degenerately doped to 1019 cm-3,
its base has a doping concentration of 5×1017 cm-3 at the emitter edge and a grading
parameter η=3. The emitter width is WE=0.2µm and base width is WB=0.15 µm.
Find β.
The β for a prototype transistor is given by Equation (9.43):
∆E*
N D W − g
β = 'C nB E e kT . We find DnB=9 cm/s and DpE=4 cm2/s from Figure
N AB D pE WB
3.11. From Figure 2.25, the band gap narrowing ∆Eg*=0.05eV, yielding for
the prototype transistor
∆E *
N C DnB WE − kTg 2.86 × 1019 9 (0.2) ⎛ −0.05 ⎞
β= ' e = exp ⎜ ⎟ = 25
N AB D pE WB 5 × 10 17
4 (0.15) ⎝ 0.026 ⎠
Using Equation (9.60), we take the grading in the base into account and
obtain
η ⎡ 3 ⎤
β (η ) = β ( 0 ) = 25 ⎢ = 25 [3.2] = 79
⎣1 − e ⎥⎦
−η −3
1− e
9.18. An npn BJT is operating in the forward active region. Assuming the
default SPICE values IS=10-16A, βF=100, find the terminal currents IC, IB, and IE
for VBE=0.7V.
⎛ 1 ⎞
qVBE
⎛ 1 ⎞ −16 0.026
0.7
From Example 9.5, I E = ⎜ 1 + ⎟ ISe = ⎜1 + ⎟ (10 ) e = 50 µ A
kT
⎝ βF ⎠ ⎝ 100 ⎠
qVBE 0.7
IC = I S e kT
= 10−16 e 0.026 = 49 µ A
IC 49 × 10−6
IB = = = 490nA
βF 100
9.19. Typical forward β’s are on the order of 100, while typical reverse β’s are
on the order of 0.1. Find the corresponding αF and αR.
From Equations (9.66),
Then
R 2L
IB
PB h 3 =RL
RB = 2 = 2
IB IB 3h
9.22. For the transistor whose IC-VCE curves are shown in Figure P9.4, find the
Early voltage.
VBE=0.75 V, how small can the base width WBM be to keep punch-through
above 12 V?
Since band-gap narrowing has little effect on junction width, we ignore it.
The built-in voltage is
qVbi EB = Eg − δ pB
⎛ NV ⎞ ⎛ 3.1 × 1019 ⎞
and δ p = E f − EV = kT ln ⎜ ⎟ = 0.026ln ⎜⎝ ⎟ = 0.11eV
⎝ NA ⎠ 5 × 1017 ⎠
Thus VbiEB = 1.12 − 0.11 = 0.91V
The junction voltage is therefore VJEB=VbiEB-VEB=0.91-0.75=0.16V. From
Figure 5.12, the junction width (which appears almost entirely in the base)
is about 0.025.µm.
9.24. As indicated in Section 9.11, to avoid excessive base push-out, JC is less than
about 0.25 mA/µm2. Consider a BJT with an emitter width (L) of 0.5µm and a base
sheet resistance of 10kΩ / . For β = 100 , find IB and the lateral voltage drop in the
intrinsic base region and discuss the current crowding effect for this device.
JC 0.25 ×10−3 mA / µ m 2
JB = = = 2.5µ A / µ m 2
β 100
I B = J B Lh
RL
RB =
4h
R L J B LhR L J B R L2
VB = I B RB = I B = =
4h 4h 4
104 × (0.5) 2
−6
VB = 2.5 × 10 = 1.56mV
4
9.25. For the transistor whose IC-VCE curves are shown in Figure P9.5, explain
why the lines are closer together as IB increases.
There are two reasons, both relating to high injection. At high injection, the
back-injection of holes into the emitter (for an npn transistor) becomes
significant (cannot be neglected with respect to the majority carrier
1
concentration in the emitter) reducing the injection efficiency γ = .
I pE
1+
I nB
The second reason is the base push-out or Kirk effect. The concentration
of minority carriers in the base becomes so high they cannot all be
extracted at the collector edge of the base region, so the base width is
effectively increased, which lowers β from Equation 9.32.
9.26. Find β for the device whose Gummel plot is shown in Figure P9.6.
Ι C (µA)
100 µA
Current
Ι Β (µA)
10 µA
1 µA
100 nA
10 nA
0 0.2 0.4 0.6 0.8 1.0
VBE (Volts)
We choose a convenient spot in the range where IC and IB are parallel. At the
point chosen, IC=100µA and IB=0.9µA. Since β=IC/IB, β=100/0.9=111.
9.27. Figure P9.7 shows some experimental data for the current gain
(normalized to that at 300 K) of a bipolar junction transistor as a function of
temperature. Also included is a straight-line approximation to the linear
region. From this data, estimate the value of ∆Eg* .
β= ' e = β 0e kT
N AB DPEWB
where β0 is a constant (the β that would be predicted if bandgap
narrowing were ignored). We can rearrange the above equation to the
form
β −∆Eg ⎛ 1 ⎞
*
ln ≅ ⎜ ⎟
β0 k ⎝T ⎠
or
β
d ln
β0
∆Eg* = − k
1
d
T
⎛ β ⎞
∆ ln ⎜ ⎟
⎝ β 0 ⎠ = −k ln ( 71)
∆Eg* (eV ) = − k −3 −3
= − ( 8.62 ×10−5 eV / K ) ( −640 K )
⎛ ⎞
1 3.3 × 10 − 10 × 10
∆⎜ ⎟
⎝T ⎠
= 0.055eV
a) To find β, we use Equation (9.43) and from Figure 2.25, ∆Eg* = 50meV .
Then
−∆Eg*
N D W
β = 'C nB E e kT
N AB D pE WB
−∆Eg*
N D W
β = 'C nB E e kT
N AB D pE WB
2.86 × 1019 13 0.15 −0.026
0.05
= e
2 × 1017 3.8 0.13
= 83
Since I C = β I B ,
c) rπ
kT 0.026
rπ = = = 1300 Ω
qI B 20 × 10−6
d) gm
β 83
gm = = = 64mS
rπ 1300
e) Cπ
For a BJT operating in the active mode, CscEB C jEB and we can
approximate Cπ ≅ CscBE . From Equation 10.30 for constant base doping,
the stored charge capacitance is
f) Cµ
⎛ ⎞
kT ⎛ N A' N D' ⎞ ⎜ 2x1017 ⋅ 5 × 1016 ⎟ = 0.83V
Vbi = ln ⎜ ⎟ = 0.026 ln
⎝ ( )
q ⎝ ni2 ⎠ ⎜ 1.08 × 1010 2 ⎟
⎠
g) fco
The cutoff frequency for this transistor is, from Equation (10.37),
1 1
f co = = = 462MHz
2π rπ ( Cbe + Cµ ) 2π (1300Ω ) ( 2.4 × 10−13 + 2.5 × 10−14 F )
h) fT
ic
10.2. For the transistor of Problem 10.1, plot as a function of frequency.
ib
What is the unity-gain frequency?
ic β 83
We use = = and plot against frequency.
ib ⎛ f ⎞
2
⎛ f ⎞
2
1+ ⎜ ⎟ 1+ ⎜ 6 ⎟
⎝ f co ⎠ ⎝ 462 × 10 ⎠
10.3. An npn BJT with uniformly doped emitter, base, and collector has β=95,
base width WB of 0.15 µm, an electron diffusion coefficient of 10 cm2/s, an rπ of
1000 Ω, and a CB junction capacitance of 0.05 pF. What is its cut-off frequency?
( 0.15 ×10−4 ) ( 95 )
2
WB2 β
Csc ( BE ) = = = 0.71× 10−12 pF
3DnB rπ 3 (10 )(1000 )
1 1
f co = = = 210MHz
2π rπ ( CscBE + Cµ ) 2π (1000 ) ( 0.71`×10−12 + 0.05 ×10−12 )
10.4. If the base in the above problem is actually graded in doping, with
a grading parameter η = 2,
a) What is the field in the base?
kT η 0.026 × 2
From Equation (9.54), E = = = 0.35V / µ m = 3.5 kV / cm
q WB 0.15µ m
β (η ) η 2
= = = 2.3
β (0) 1 − e −η
1 − e−2
β (η ) = 2.3β (0) = 2.3 × 95=219
where we have used Equation (10.30), and the resulting cut-off frequency
is
1 1
f co = = = 300 MHz
2π rπ ( CscBE + Cµ ) 2π (1000 ) ( 0.48 × 10−12 + 0.05 × 10−12 )
300
which is an improvement of a factor of = 1.43
210
10.5. Two transistors that are otherwise identical differ in that one has a
uniform base and the other has a graded base with η=6. Assume the field in
the base is small enough that the diffusion constant has its low-field value.
a) What is the improvement in β for the graded device?
b) What is the change in the stored charge capacitance if both transistors are
operated at the same value of IC?
Anderson & Anderson 5 February 17, 2007
Chapter 10: Solutions
From Equation (10.29), we have
δ WB2 β DC δ WB2 qI C
CscBE = =
2 DnB rπ 2 DnB kT
If all other parameters are the same, using Figure 10.8 to find the values
of δ, we have
10.6. Equation (10.39) indicates that the electron velocity in the base increases
as ∆n( x) decreases (InB constant). Explain the physics of this.
The current is constant across the base. Since the current is a function of
drift plus diffusion and since E is constant, so is the force on the electrons
or the electron velocity due to the field is also constant. However, near the
collector where ∆n is small, diffusion predominates and the velocity due to
diffusion must increase.
10.7. Find the base transit time tT for an npn prototype BJT with WB = 0.05 µm
and base doping of 5 × 1017 cm-3.
W B2
From Equation (10.48), t T = . From Figure 3.11, the majority carrier
2Dn
diffusion length (electrons are majority carriers in the base) for a doping
concentration of 5×1017 Dn = 11cm 2 / s . Thus
W B2 (0.05 × 10 cm)
−4 2
tT = = = 1.1ps
2Dn ⎛ cm2 ⎞
2⎜ 11 ⎟
⎝ s ⎠
10.8. Repeat the above problem for a pnp prototype BJT with WB = 0.05 µm
and base doping of 5 × 1017 cm--3.
Now the carriers of interest are the holes, and Dp=5cm2/s, and
10.9 Find the transit time tBC across the base-collector transition region for an
npn prototype BJT with base doping of 5 × 1017cm-3 , collector doping of 5 ×
1016 cm -3 and collector-base voltage of 2.5 V.
In the high field of the base-collector transition region, the electrons travel
w
at their saturation velocity. From Equation 10.50, we have t BC = BC . The
vsat
junction width at 2.5 V is given by Equation (5.91),
1
⎡ 2ε (Vbi − Va ) ( N A' + N D' ) ⎤ 2
wBC =⎢ ⎥ .
⎢⎣ qN A' N D' ⎥⎦
⎛ ( 5 × 1017 )( 5 × 1016 ) ⎞
kT ⎛ N A' N D' ⎞
Vbi = ln ⎜ ⎟ = 0.026 ln ⎜⎜ ⎟ = 0.86V
⎝ (1.08 × 10 )
q ⎝ ni2 ⎠ 10` 2 ⎟
⎠
1
⎡ 2 (11.8 ) ( 8.85 × 10 ) ( 0.86 − ( −2.5 ) ) ( 5 × 10 + 5 × 10
−14 17 16
) ⎤⎥ 2
wBC = ⎢
⎢⎣ (1.6 × 10−19 )( 5 × 1017 )( 5 × 1016 ) ⎥⎦
= 3.1 × 10−5 cm = 0.31µ m
W B2
From Equation (10.48), t T = , thus if the base width is cut in half, the
2Dn
transit time decreases by a factor of 4.
10.11. An npn transistor's base region is doped to 1018 cm-3. How thick would
the base region need to be for the base transit time to be equal to 1/100 of the
electron lifetime in the base? If such a transistor were manufactured, describe
the effect on β and the operating frequency?
From Figure 3.21, the electron lifetime at the doping concentration is τn=1
µs.
W B2
From Equation (10.48), t tb = = 0.01(τ n ) = 10ns . From Figure 3.11, Dn
2Dn
(the electrons are minority carriers in the base) =9 cm2/V·s.
⎛ cm 2 ⎞
W B = 2Dn [0.01(τ n )] = 2⎜ 9 ⎟ (10 × 10 s ) = 4.2 × 10 cm = 4.2 µm .
−9 −4
⎝ s ⎠
This is a very thick base, and the unity current gain frequency from
1 1
Equation 10.49 is fT = = = 16 MHz
2π tT 2π (10 × 10−9 )
10.12. For high speed BJTs, the double poly self-aligned transistor is preferred
to the conventional transistor discussed in Chapter 9. For similar emitter, base
and collector dopings compare the following parameters and explain your
reasoning.
(a) Collector-base junction capacitance.
This will be smaller for the double poly self-aligned transistor because the
collector and base areas can be smaller.
⎛ qVa ⎞
J = J 0 ⎜⎜ e kT − 1⎟⎟ . Since the voltage is the same, the ratio of the currents is
⎝ ⎠
the same as the ratio of their reverse leakage currents
I pn 10 AJ pn 10 J 0 pn 10−10
= = = −5
= 10−5 . This speeds up the output
I Schottky AJ Schottky J 0 Schottky 10
transition low-to-high time since the stored minority charge because the
forward biased C-B junction is reduced by a factor of 105.
10.14. From Equations (10.43) and (10.23), find an expression for the base
transit time in a graded base transistor as a function of η.
QB W B2 ⎛ η − 1 − e−η ⎞
tT = ⎜ ⎟
IC DnB ⎝ η2 ⎠
1 1
TRF = = = 17 ps
vRF 60 × 109
For the optical signal, the wavelength is about 500 nm, from Figure 11.1.
The period is
1 λ 500 × 10−9
Top = = = = 0.0017 ps
vop c 3 × 108
17
The RF period is = 10000 times longer, so 10,000 cycles of optical
0.0017
light fit into one cycle of the RF. Therefore even the fastest detectors can
b) At what depth is the incident flux (neglecting Fresnel loss) reduced to 10%
of its value at the surface? 1%?
To find the absorption coefficient from Figure 11.4, we must convert this
1.24 1.24
E(eV) = = = 1.77eV
λ ( µm) 0.700
FL ( x) − ( 2.1×104 cm −1 ) x
= 0.1 = e
FL (0)
ln(0.1) −4
x10% = − = 1.1× 10 cm = 1.1µm
2.1 × 10 4 cm −1
ln ( 0.01)
x1% = − = 2.2 × 10−4 cm = 2.2 µ m
2.1 × 104 cm −1
c) The color is changed to orange. Now how deep does the light penetrate (to
the 10% level)?
From Figure 11.1, orange light has a wavelength of about 600 nm, or a
ln (0.1)
x10% = − = 0.43µm
5.3 × 10 4 cm −1
11.3. a) Calculate the Fresnel reflection at normal incidence for light going from
air to glass (n=1.5).
b) Explain why you can see into a store window and see your reflection at the
same time, but at night looking out your window from a lighted room you can only
see your reflection?”
When you look into a store window, you see 4% of the sunlight reflected
from you and off the surface of the glass, but you also see the light from
the merchandise being transmitted through the window. To your eye both
4% reflection of the room light is the only thing you can see.
1 dJn (x) ∆n
+ GL (x) − p = 0
q dn τn
But we know
GL ( x) = α (1 − R ) FL ( x) = α (1 − R ) FLi e −α x
dn
Jn = qDn
dx
Using n=np,
dJ n d 2 np
= qDn
dx dx 2
Therefore
1⎛ d 2 np ⎞ ∆n p
⎟⎟ + α (1 − R ) FLi e −
−α x
⎜⎜ qDn =0
q⎝ dx 2 ⎠ τn
1⎡ ⎛ d 2 n p 0 d 2 ∆n p ⎞⎤ ∆n p
⎟⎟ ⎥ + α (1 − R ) FLi e −
−α x
⎢ qDn ⎜ + =0
q ⎢⎣ ⎜ dx 2 dx 2 τn
⎝ ⎠ ⎥⎦
1⎡ d 2 ∆n p ⎤ ∆n p
⎢ qDn 2 ⎥
+ α (1 − R ) FLi e −α x − =0
q ⎢⎣ dx ⎥⎦ τn
RL=10 kΩ
Va
b) On your graph, also plot the load line for Va =+5V and Va=-5 V
c) Find the current flowing through the circuit for each load line and plot it
against IL. Recalling that the photocurrent IL is proportional to the intensity of the
light, under which bias regime should one operate photodiodes if one wants the
output current to be proportional to intensity?
Deleted: 2/26/2004
(1 − n ) = (1 − 3.65)
2 2
(−2.65) 2
The Fresnel reflection is R = = = 0.32
(1 + n ) (1 + 3.65)2
2
4.652
1.24 1.24
E (eV ) = = = 1.55eV . From Figure 11.4, we find the
λ ( µ m) 0.8µ m
absorption coefficient is α=9000 cm-1. After the surface layer, the flux
remaining is
( )(
− 9×103 cm−1 0.3×10−4 cm )
FL ( xn ) = FL (0.3µ m) = (1 − R ) FLi e −α xn = (0.68)( FLi )e
= 0.52 FLi
FL (0.3µ m) ⎡1 − e
( ) (1.5×10 ) ⎤ = 0.52 F 1 − 0.26 = 0.38F
−4
Li ( )
− 9×103
⎢⎣ ⎦ ⎥ Li
The responsivity is
11.7. For the following solar cell whose I-V characteristics are shown in Figure
P11.2, find Isc, Voc, and η. The incident power is 15 mW.
From the graph we find ISC = -4mA, VOC = 0.95V, Im=3.5 mA, and Vm=0.83
Im Vm 3.5 × 0.83
FF = = = 76%
IscVoc 4.0 × 0.95
PLi 15 ×10−3
11.8. Which absorbs more of the total solar spectrum, GaAs or Ge?
From Figure 11.8, we see that Ge has a small bandgap, and thus
can absorb all wavelengths below about 1.8µm, whereas GaAs has
spectrum.
EC
EV
From the Golden Rule, the total energy absorbed from the photon is
1.24 1.24
E (eV ) = = = 2.48eV
λ ( µ m) 0.5
If both the electron and hole relax to the band edges, the energy different
between them now is just 1.12 eV. Therefore the total energy given up as
heat is
11.10. If the doping in the p region of an n+p photodiode is decreased, one would
expect the diffusion length in a solar cell to increase. Verify (or contradict) this by Deleted: 2/26/2004
The parameter that is affected is the diffusion length Ln, which is increased
ηQ (λ = 1µm) = n = = 64%
(αLn + 1) (10 −2 µm−1 )(400µm) + 1
and
ηQ (λ = 0.5µm) = = = 53%
(αLn + 1) (1µm−1 )(400µm) + 1
We see that the lighter doping made a significant change in the efficiency
11.11. If a GaAs photodiode has a junction depth of 0.2 µm, and if light absorbed in
the surface layer is considered lost to surface recombination, what is the total
fractional loss in photons in the surface layer? Let the photon energy be 1.4 eV, and
repeat for Eph=1.8 eV. Where on the spectrum are these two energies?
At Eph=1.4 eV, α=4000 cm-1 (From Figure 11.4), and at Eph=1.8 eV,
α=30,000 cm-1
At 1.8 eV,
FL ( xn ) −( 3×104 )( 0.2×10−4 )
= e−α xn = e = 0.55 is kept or 45% is lost.
FL (0)
Figure 11.1 shows that 1.4 eV is in the near infrared, and 1.8 eV is in the
red.
Deleted: 2/26/2004
The field lines are concentrated at the corners. Field lines must start and
end on charges, but if the number of positive and negative charges have
to be equal, the field lines have to be closer (higher field strength) in the
corners.
b) Draw the energy band diagram for a PIN diode under high reverse bias,
and indicate the breakdown mechanism.
c) Draw the energy band diagram for a p+-p—-i-n junction under the same
reverse bias. Explain why this structure will break down at higher voltages.
The p- region drops some of the voltage, so that the field in the intrinsic
region is reduced.
wavelength:
Deleted: 2/26/2004
concentration of 0.65.
b) If you also take into account the fact that the electrons are concentrated a
little above EC and the holes are concentrated a little below EV, how much does that
change the bandgap you would choose?
The peaks of the electron and holes distributions are about 1/2 kT away
from the band edges. If this adds a total of kT=0.026 eV to the emission
energy, that is still a small change and we can neglect this effect.
11.14. Recall that the peak of the electron distribution with energy is about 1/2 kT
above EC, and the peak of the holes distribution is about 1/2 kT below EV If each
distribution is approximated as having an overall width of kT, estimate the spectral
width of the emission. Assume the material emits at 1.3 µm. Hint: to find ∆λ, use
E=hc/λ, and take the derivative dE/dλ to obtain an expression for ∆λ in terms of ∆E.
From the figure, we could estimate the range of possible energy emissions
have
−1
∆E = hc ∆λ = −
1
(6.6 × 10 −34
J ⋅ s)(3 × 108 m / s)∆ λ = 0.052eV
λ (1.3 × 10 m)
2 −6 2
∆λ =
(0.052eV ⋅ 1.6 × 10 −19
J / eV )
= 71nm
−13
1.17 × 10 J/m
Deleted: 2/26/2004
Since the nitrogen is not a donor, it will not be ionized in the lattice. Since
it is electrically neutral it will not attract electrons, (unless the electron gets
11.16. Optical fiber manufacturers battled the OH- ion (resulting from water) for
years. These ions, when incorporated into the glass, produce a strong absorption at
1.4 µm (see Figure 11.16). They have finally managed to nearly eliminate it. Is there
a similar absorption in the earth’s atmosphere?
A look at the solar spectrum in Figure 11.8 shows almost no light at λ=1.4
From Figure 11.17, considering only direct gap materials, the possibilities
substrates.
Deleted: 2/26/2004
We can follow the procedure in Example 11.7, but it is quicker to take the
derivative:
−1
dq = 2 nd∆λ and recognize that the change in q from one mode to the
λ2
−1 −6
dq = 1 = 2(3.5)(100 × 10 m)∆ λ
(900 × 10 m)
−9 2
or
∆λ =
(900 × 10 ) = 1.16nm where we have neglected the minus sign
−9 2
2(3.5)(100 × 10 )
−6
because we are interested in the separation, not the direction. There are
11.19. A diode begins to lase when the gain in the cavity exceeds the losses. One
source of loss is the partially reflective mirrors at either end of the cavity. Some
percentage of the light power is lost each time the light strikes one of the mirrors.
How would the power-current curve of a laser be changed if coatings are added to
the facets to increase the reflectivity?
We would expect it takes less current for the gain to exceed the loss since
Deleted: 2/26/2004
Deleted: 2/26/2004
eV, so if the semiconductor has a band gap larger than 2.07 eV, this
photon cannot be absorbed at all. For those whose band gap is correct,
we look for indirect vs. direct gap materials. Indirect materials have a low
Ge: high: indirect, the band gap is small enough. However there is
Deleted: 2/26/2004