Lecture Notes # 3 - sunlight.caltech.edu
Download
Report
Transcript Lecture Notes # 3 - sunlight.caltech.edu
Lecture Notes # 3
• Understanding Density of States
– Solve 1-D Schrödinger equation for particlein-a-box
– Extend to 3-D
– Invoke periodicity requirement
– Solve for density of states
Review of Quantum Mechanics
H Hamiltonian operator
H
Mathematicwavefunction depicting
natureof electronin space/time
Energyor allowable quantizedenergies
of thatelectron
• Often times you do not know or , but
you have boundary conditions and want to
solve for possible values of and a
functional form of
Review of Quantum Mechanics
• Most general case: Time independent
2 2
H
U x, y, z
2m
Kinetic E.
Potential E.
• How do we know first part is K.E?
1 2
m v classically
2
1 2 p2
mv
2
2m
pˆ i is quantum momentumoperator
KE
1 2 2 2
So,
mv
2
2m
Review of Quantum Mechanics
For r expik r
pˆ a
Momentum operator
Eigenvalue
pˆ i r i r iik r k r
r
a = Momentum Eigenvalue
p m v k
k
v
velocityof part icle
m
Particle in a 1-D Box
• Free electron floating around in vacuum
• Let’s impose some boundary – confine it to a
region of space, a box or a unit cell in 1-D
U
Confine it by setting U
outside the box and U = 0
inside the box
U
Confined
e-
U=0
0
L
x
• Since U(x) = 0 for 0<x<L, we can drop U(x) out
of the Hamiltonian, which becomes
2 2
H
2m
Particle in a 1-D Box
• Because U outside the box, we know the eCANNOT be there, so we get the boundary
condition:
x 0 x L 0
Because t heHamilt onian must give
back an eigenvalue, we must pick
f x where f x Af x
and f 0 f L 0
• We do not care what happens between 0 and L,
so the simplest solution is just:
n
A sin
x , n 1,2,3,4...
L
for x 0, sin(0) 0 OK
for x L, sin(nL) 0 OK
Particle in a 1-D Box
• Plug into the Schrödinger equation to make
sure H = E
2 d 2
n
A
sin
2
2m dx
L
x
2 d n
n
A cos
x
2m dx L
L
2
2m
n 2 2
n
x
2 A sin
L
L
1 n
n
A
sin
2m L
L
2
Energy, E
x
Particle in a 1-D Box
• Energy values are quantized since n
is an integer
• n=1 is lowest energy state, n=2 has
higher energy, etc.
n=4
E
n=3
9x
n=2
4x
n=1
18
0
16
14
L
• We can map out (x,n) vs. E
12
10
n
E x
2
2mL
2
2
2
8
• Allowed energy states
6
4
2
0
0
1
2
n
3
4
Particle in a 1-D Box
• Now let’s fill up the states with electrons.
Suppose we have N e- we want to pour into our
1-D box.
• For N e- you can calculate the energies since
we know we can have 2e-/n states (two spins).
• So N electrons fills nF= N/2 states.
• The highest energy state, nF, gives F, the Fermi
energy.
2
F
nF
2m L
Fermi-Dirac Distribution
• Fermi energy is well defined at
T = 0 K because there is no
n
E x
thermal promotion
2mL
• At high T, there is
F
thermalization, so F is not as
clear
2
2
18
16
14
2
12
2
10
8
6
4
2
0
0
1
2
n
3
• Officially defined as the energy where the probability of finding an
electron is ½
• This definition comes from the Fermi-Dirac Distribution:
1
f
exp / k BT 1
• This is the probability that an orbital (at a given energy) will be
filled with an e- at a given temperature
• At T=0, =F and = F, so f(F)=1/2
4
Particle in a 3-D Box
• Let’s confine e- now to a 3-D box
• Similar to a unit cell, but e- is confined
inside by U outside the box
• Schrödinger’s equation is now
2 2
2
2
2 2 2 x, y, z x, y, z
2m x y z
• You can show that the answer is:
nx n y nz
n x, y, z A sin
x sin
y sin
z
L L L
• We now have 3 quantum numbers nx, ny, and nz that are
totally independent
• (1,2,1) is energetically degenerate with (2,1,1) and (1,1,2)
Particle in a 3-D Box
• Now let’s repeat this box infinitely in each direction to get a
repeated “unit cell”
• What’s different about this situation?
– U(x,y,z)=0
– No region where U = infinity
• So, there’s really no reason that x 0 x L 0
sinceU x 0 U x L
• We don’t need those boundary conditions anymore
Periodic Boundary Condition
• For now, we don’t need such a strict boundary condition
• Make sure is periodic with L, which would make each 3D box identical
• Because of this, we’ll have a periodic boundary condition
such that
x L, y, z x, y, z
• Wave functions that satisfy this periodic B.C. and are
solutions to the Schrödinger equation are TRAVELING
WAVES (not a standing wave anymore)
Periodic Boundary Condition
• Bloch function
k r exp ik r
• Wave vector k satisfies
2
4
k x 0;
;
;...
L
L
• Etc. for ky and kz
• Quantum numbers are components of k of the form 2n/L
where n=+ or - integer
expik x x L expi 2nx L / L
expi 2nx / Lexpi 2n
expi 2nx / L
expik x x
• Periodicity satisfied
Back to Schrödinger Equation
• Substitute
k r exp ik r
into
2 2
2
2
2 2 2 k r k k r
2m x y
z
gives
2 2 2 2
k
k
k x k y2 k z2
2m
2m
• Important that kx can equal ky can equal kz or NOT
pˆ i
• The linear momentum operator
for k r exp ik r
pˆ k r i k r k k r
so theplane wave k r is an eigenfunction
of linear moment umwith an eigenvalueof
k , and thepart iclevelocityin orbit alk is
k
v
m
Fermi Level in 3-D
kx
• Similarly, can calculate a Fermi level
F
Fermi level
2
2
kF
2m
kF
kz
Vector in 3-D space
• Inside sphere k<kF, so orbitals are filled.
k
k>kF, orbitals are empty
• Quantization of k in each direction leads
to discrete states within the sphere
• Satisfy the periodic boundary conditions
at ± 2/L along one direction
This is a sphere only if
• There is 1 allowed wave vector k, with NOTE:
k =k =k . Otherwise, we have
an ellipsoid and have to
distinct kx, ky, kz quantum #s for the
recalculate everything. That can
3
volume element (2/L) in k-space
be a mess.
Sphere: GaAs (CB&VB), Si (VB)
• So, sphere has a k-space volume of
Ellipsoid: Si (CB)
y
x
4
V k F3
3
y
z
Number of Quantum States
• Number of quantum states is
4 k F3
totalvolume
3 3
volumeof 1 allowed quantizedstate 2
L
• Since there are 2 e- per quantum state
4 k 3
3 F L3 3
N 2
2 kF
3
2
3
L
V
N 2 k F3
3
Solve for k F ,
3 N
k F
V
2
1
3
• Depends on e- concentration
Density of States
• Plug kF into
2 2
F
k
2m
3 N
F
2m V
2
2
2
3
• Relates Fermi energy to electron concentration
• Total number of electrons, N:
3
V 2 m 2
N 2 2
3
• Density of states is the number of3 orbitals per unit energy
dN
V 2m 2 1 2
D
2 2
d 2
Relate to the surface of the sphere. For the next incremental growth in the sphere,
how many states are in that additional space?
Density of States
• Divide by V to get N/V which is electron density (#/cm3)
• Volume density of orbitals/unit energy for free electron gas
in periodic potential
3
1 2m 2 1 2
D
2
2
2
Starting point energy
Effective mass of e-
1 2m
D 2 2
2
*
e
3
2
E ECB 2
1
CB
Effective mass of h+
1 2m
D 2 2
2
*
h
3
2
EVB E 2
1
Start from VB and go down
VB
Concentration of Electrons
n
D E f E dE fun
e
e
EC
“EF”
3
m kT 2
exp EC / kT
n 2
2
2
*
e
m kT
N C 2
2
2
*
e
3
2
Effective density of states in CB
n NC exp EC EF / kT
Writing it with a minus sign indicates that as E difference between EC
and EF gets bigger, probability gets lower
Concentration of Holes
EV
p Dh E f h E dE
f h 1 f e
3
m kT 2
expEV / kT
p 2
2
2
*
h
m kT
NV 2
2
2
*
h
3
2
Effective density of states in VB
p NV exp EF EV / kT
Intrinsic Carrier Concentration
EC EF EF EV
n p NC NV exp
kT
EC EV
n p NC NV exp
kT
EC EV EG , theband gap
n p NC NV exp EG / kT
Entropy term
Enthalpy term
n p ni2
ni intrinsiccarrierconcentration
Constant for a given temperature.
Intrinsic = undoped
Intrinsic Carrier Concentration
EG
ni N C NV exp
2kT
ni constantfor a given T means that
n p is constant,too,which holds under
equilibrium and away fromit
•
•
•
•
ni
Ge: 2.4 x 1013 cm-3
Si: 1.05 x 1010 cm-3
GaAs: 2 X 106 cm-3
At 300 K
At temperature T, n = p by conservation
Add a field and np = constant, but n does not equal p
As n increases, p decreases, and vice versa
Useful to define Ei, which is Ei = EF when it is an intrinsic
semiconductor (undoped), so n = p = ni
EC Ei
Ei EV
ni NC exp
NV exp
kT
kT
Intrinsic Fermi Level
EC Ei
Ei EV
ni NC exp
NV exp
kT
kT
NV
EC Ei Ei EV
ln
kT
kT
NC
NV
Ei EC EV Ei kT ln
NC
NV
2 Ei EV EC kT ln
NC
NV
2Ei EV EC EV kT ln
NC
EG kT NV
Ei EV ln
2
2 NC
Intrinsic Fermi Level
EG kT NV
Ei EV ln
2
2 NC
• This says that the intrinsic Fermi level (relative to the
valence band) is about mid-gap ± the (kT/2)ln(NV/NC)
scaling factor
kT NV
ln
2 NC
EG (eV)
-13meV for Si
1.12
35 meV for GaAS 1.42
0.67
- 7 meV for Ge
• So Ei for Si and Ge is slightly below mid-gap. Ei for GaAs is
slightly above. It is minor compared to EG, but just so you
know
• All of this has been intrinsic with no dopants
Dopants
• Let’s consider adding dopants
n-type
carriersn ni N D
ED
ECB
Ei
EVB
Depends
on EG, T
Depends on D, T, and
dopant density
D ECB ED
At T=0, all donor states are filled. Hence, n = 0.
But at room temperature in P doped Si, 99.96%
of donor states are ionized.
At mid temperature, if N D ni , thenn N D
At high temperature, such that
ni ND , thenn ni
Dopants
n p ni2 still holds, just substitute
n ni N D
ni2
p
at room temperatu
re, n N D for typical dopantdensities
ni N D
Say N D 1016 cm3
N D 1016 cm3
n 1016 cm3
n NC exp EC EF / kT
n 1016 cm3
19 3 exp EC EF / kT
N C 10 cm
about 60 mV/decade
3 decades x 60 mV 180 mV 180 meV down