QM to this point:
1) General:
Required by SWE
wherever V = const.
x t , ( ) A e
i k x t ( )
B e
i k x t ( )
+ = so long as: k
2 m E V
( )
h
bar
2
=
Also know that for these electron waves:
h
p
= "de Broglie wavelength"
From form of : k 2
2
h
bar
2
2m L
2
+ = put in k
n
L
= and this becomes:
E V
well
h
bar
2
k
2
2m
+ =
Semiconductor_Devices_7.mcd 1 5/28/2009
Solutions for large electron quantum well  stated in terms of acceptable quantized choices of E and k:
E
Near continuous single band of closely
spaced solutions to Schrdinger Equation
Band
k
V
3) Quantum well consisting of block of atoms: Added DIFFRACTION when = multiple of atomic spacing / 2
At these special values of (and k):
 Electron interacts so strongly with atoms that it slows down, stops, and reflects
 This opens up gaps in band we calculated from simple well w/o atoms
Led to solutions for large quantum well made up of atoms:
Semiconductor_Devices_7.mcd 2 5/28/2009
E
Band
Multiple bands, separated by bandgaps
Caused by DIFFRACTION when electron
wavelength matched atom spacing
Band
Band
k
a
o
a
o
V
Ability of electrons to diffract at these gaps (gaining or loosing momentum from atoms) led to:
E
E
More compact
way of describing =>
E vs. k combos
2
3
1 4
k k
2
a
o
a
o
a
o
2
a
o
Semiconductor_Devices_7.mcd 3 5/28/2009
4) Assert now (and can be proven) that most of our other results from simple quantum wells still ~ apply
If k's still ~ obey: k
well
n
L
= with n = +1, +2, +3 ..
Then for every positive value of k, there is an equal negative value of k
If p still ~ h
bar
k
Then possible positive momentum states are precisely balanced by negative momentum states
GIVES: A BAND FULL OF ELECTRONS CANNOT CONDUCT!
CONDUCTORS must have at least one partially filled band:
Empty band  nothing to conduct!
Partially filled band  CAN conduct
Semiconductor_Devices_7.mcd 4 5/28/2009
INSULATORS must have only empty or completely filled bands:
Completely empty band  nothing to conduct!
Completely filled band  can't conduct
Is there a third possibility? For instance, what the heck is a semiconductor?
E E
Empty band
Empty band
Insulator 1: Insulator 2:
Filled band
Filled band
For second insulator, what if separation is so small that room temperature heat can excite e

across gap?
Semiconductor_Devices_7.mcd 5 5/28/2009
1 empty band + 1 filled band 2 partially filled bands
Conducting!
Add
Small
e

Heat
Conducting!
Insulator Semiconductor
SEMICONDUCTOR = Insulator with small enough gap between highest filled band / lowest empty band
That at temperatures of interest (generally ~ 25C = 298K) electrons can jump the gap
That was simple... But now need to figure out how small a gap we need => Thermal promotion of electrons
Physical Rules Governing Electrons:
1) Maximum of one electron allowed per quantum state ("Pauli Exclusion Principle")
2) Electrons are indistinguishable (if swap two of them, there is no way to tell)
3) Electrons can freely exchange thermal energy with one another (through collisions)
4) All possible energy configurations of electrons are equally likely
Semiconductor_Devices_7.mcd 6 5/28/2009
With these rules, imagine a very simple QM system:
 Series of equally spaced energy states (with separation of A)  Four available electrons
 At T = 0 degrees K, electrons fall into four lowest states:
Energy State # Levels Energy
11
10
9
8
7
6
5
4
3
2
1
A
Probability state is filled
0 1
Now, raise the temperature to T
1
On average each electron gains thermal energy ~ k
B
T
1
(k
B
= "Boltzman's Constant" from Chem. class)
System gains energy ~ 4 k
B
T
1
Assume this gained system energy is equal to 6 A i.e. k
B
T
1
= 3/2 A
Semiconductor_Devices_7.mcd 7 5/28/2009
Possible arrangements of electrons (as they divide between themselves 6 A in total gained energy):
11
10
9
8
7
6
5
4
3
2
1
Probability that level n is filled = # Times n is filled in all of these alternatives
# of alternatives
# of alternatives = 9
Level Number of times it was filled Probability it is filled
11 0 0
10 1 1/9 = 0.11
9 1 1/9 = 0.11
8 2 2/9 = 0.22
7 3 3/9 = 0.33
6 3 3/9 = 0.33
5 4 4/9 = 0.44
4 4 4/9 = 0.44
3 5 5/9 = 0.55
2 6 6/9 = 0.66
1 7 7/9 = 0.77
I may have missed an
alternative or two
but let's plot anyway:
Semiconductor_Devices_7.mcd 8 5/28/2009
E Mathematical Function:
"FermiDirac Distribution Function"
(Can be proven it is exact w/ many electrons)
f
F
E ( )
1
1 e
E E
f
k
B
T
+
=
k
B
= Boltzman's constant
T = Temperature in degrees K (!)
E
F
= parameter = "Fermi Energy"
Probability state is filled
0 1
What is significance of the parameter E
f
("Fermi Energy")?
f
F
E E
F
=
( )
1
1 e
E
F
E
F
k
B
T
+
=
1
1 e
0
k
B
T
+
=
1
1 e
0
+
=
1
1 1 +
=
1
2
=
E
F
("Fermi Energy") is thus the Energy where the probability of state being filled falls to 1/2
Above E
F
, probability state is filled < 50% Below E
F
, probability state is filled > 50%
Semiconductor_Devices_7.mcd 9 5/28/2009
E
Plot of f
F
(E) for many electron system:
Higher the temperature:
More electrons from below E
F
are thermally promoted to levels above E
F
T = hot
T = warm
E
F T = 0 K
P
0 1
Actually does square up at low temperatures (as indicated by blue curve):
f
F
( E > E
F
, small T) =
1
1 e
E E
F
( )
k
B
T
+
1
1 e
positive
small
+
=
1
1 e
large
+
=
1
1 very_large +
= very_small =
Semiconductor_Devices_7.mcd 10 5/28/2009
f
F
( E < E
F
, small T) =
1
1 e
E E
F
( )
k
B
T
+
1
1 e
negative
small
+
=
1
1 e
very_negative
+
=
1
1 very_small +
= 1 =
Important features of this Fermi Function f
F
(E):
1) f
F
(E) = probability that a state is filled with an electron
thus 1f
F
(E) = probability that a state is empty of electron
E E
1f
F
(E)
1f
F
(E)
E
F
E
F
f
F
(E)
f
F
(E) 1f
F
(E)
0 0.5 1
0 0.5 1
Semiconductor_Devices_7.mcd 11 5/28/2009
2) Not only complementary, also symmetric about E
F
: f
F
E
F
+
( )
1 f
F
E
F
( )
=
OR:
E
F
+ f
F Equal widths at equal offsets from E
F
E
F
E
F
1 f
F
f
F
E ( )
1
1 e
E E
F
k
B
T
+
=
3) Far from E
F
, f
F
and 1f
F
=> "Boltzman Functions" (i.e. simple exponentials)
a) For E  E
F
>> k
B
T (E far above E
F
):
E E
F
k
B
T
Large =
f
F
E ( )
1
1 e
Large
+
=
1
e
Large
= e
Large
= e
E E
F
( )
k
B
T
= f
F
( E  E
F
>> k
B
T ) => e
E E
F
( )
k
B
T
Semiconductor_Devices_7.mcd 12 5/28/2009
b) For E E
F
 << k
B
T (E far below E
F
):
E E
F
k
B
T
Large = and thus: e
E E
F
k
B
T
small =
1 f
F
E ( ) 1
1
1 e
E E
F
k
B
T
+
=
1 e
E E
F
k
B
T
+ 1
1 e
E E
F
k
B
T
+
=
e
E E
F
k
B
T
1 e
E E
F
k
B
T
+
=
small
1 small +
= small = e
E E
F
k
B
T
=
1 f
F
( E  E
F
<< k
B
T ) => e
E E
F
( )
k
B
T
Or, in the plot:
f
F
= Decaying exponential
We WILL use this information
in a lecture or two from now !
1  f
F
= Decaying exponential
Semiconductor_Devices_7.mcd 13 5/28/2009
ONE BIG QUESTION WE HAVE NOT ANSWERED: How do we determine the position of E
F
?
Will get this form: Conservation of Electrons
Number promoted from below E
F
= Number added above E
F
But Before we can apply this, we must recall that: f
F
(E) is NOT the number of electrons at E
INSTEAD: f
F
(E) is the probability that a given single quantum state at E is occupied
Number of electrons at E: N (E) = ( # of available quantum states at E ) x f
F
(E)
Analogous to: # of cars in parking lot = ( # of parking places ) x ( Probability a place is filled )
Missing quantity is g(E) from lecture 5: g(E) = ( # of quantum states ) / volume / energy interval at E
Where from lecture 5: g E ( )
4 2 m ( )
3
2
h
3
E V
well
( )
1
2
=
Semiconductor_Devices_7.mcd 14 5/28/2009
For the moment, let's say (incorrectly): g(E) is constant within a band / g(E) = 0 between bands Then:
Number of electrons at E = g(E) x f
f
(E)
E E E
Band 2
Band 1
g E ( ) 0 1 N E ( )
f
f
E ( )
At each E, I multiplied g(E) (# of available states) times f
f
(E) (probabiltiy that state is filled) to get N(E)
OK, but in reality, g(E) does not have the form above. Instead we must deal with
V
well
was bottom of the quantum well
=> Bottom of one of our bands
g E ( )
4 2 m ( )
3
2
h
3
E V
well
( )
1
2
=
Semiconductor_Devices_7.mcd 15 5/28/2009
More realistic diagram of bands / g(E) (number of available quantum states / "parking places" ) would look like:
Bands as simple
boxes w/ electrons
E
E
g(E) o (E  E
C
)
1/2
E
C
E
V
g(E) o (E
V
 E)1/2
k g E ( )
a
o
a
o
Using my new symbols E
C
and E
V
for the bandedges
The top g(E) function (in green) is a straight translation of the lecture 4 formula
The bottom g(E) function (in red) does not seem to make sense:
It is as if I just flipped the curve for the top band upside down!
Semiconductor_Devices_7.mcd 16 5/28/2009
Yes, that is what we are going to do! It's based on the E vs. k diagram:
E
E
Few electrons
More Realistic =>
Few missing electrons
k k
Will treat missing electrons in lower band as "positive holes", and just flip the electron physics upside down
Semiconductor_Devices_7.mcd 17 5/28/2009