Sei sulla pagina 1di 24

NCN

www.nanohub.org

EE‐606: Solid State Devices
EE‐606: Solid State Devices
Lecture 9: Fermi‐Dirac Statistics
Muhammad Ashraful Alam
alam@purdue.edu

Alam  ECE‐606 S09 1
Carrier Density

Carrier number = Number of states x filling factor

Chapters 2‐3 Chapter 4

Alam  ECE‐606 S09 2
Outline

1) Rules of filling electronic states 
Rules of filling electronic states
2) Derivation of Fermi‐Dirac Statistics: three techniques
3)) Intrinsic carrier concentration
4) Conclusion

Reference: Vol 6 Ch 4 (pages 96‐105)


Reference: Vol. 6, Ch. 4 (pages 96 105) 

Alam  ECE‐606 S09 3
E‐k diagram and Electronic States

Energy‐Band Density of States

E m* 2m*
E E −Ec
2π 2 =3
E3
E2
E1

π π

a a
k g(E)

Alam  ECE‐606 S09 4
Rules for filling up the States

E3
E2
E1

‰ Pauli Principle:  Only one electron per state

‰ Total number of electrons is conserved NT = ∑N i i

‰ Total energy of the system is conserved ET =∑ EN
i i i

Alam  ECE‐606 S09 5
Outline

1) Rules of placing electronic states 
R l f l i l t i t t
2) Derivation of Fermi‐Dirac Statistics: three techniques
3) Intrinsic carrier concentration
4) Conclusion

In 1926, Fowler studied collapse of a star to white dwarf by F‐D statistics, 
before Sommerfeld used the F‐D statistics to develop a theory of 
electrons in metals in 1927. Wikipedia has a nice article on this topic. 

Alam  ECE‐606 S09 6
Illustrative Example: 3 Energy Levels

NT=5 and ET=12 NT = ∑ i N i ET = ∑ i Ei N i

E=4
EE=2
2
E=0

2! 5! 7! 2! 5! 7! 2! 5! 7!
W203 = • • W122 = • • W041 = • •
1!2! 0!5! 3!4! 1!1! 2!3! 5!2! 0!2 ! 4!1! 6!1!
= 35 = 420 = 35

Alam  ECE‐606 S09 7
Occupation Statistics

E=4
E=2
E=0
W203 = 35 W122 = 420 W041 = 35

Choose the most 
W ((E)

probable configuration
probable configuration.

2,0,3          1,2,2           0,4,1

Alam  ECE‐606 S09 8
Occupation Statistics

2
E=4 f 3* =
7 E
E=2 f 2* =
2
5
E=0 1
f1* =
2 W122 = 420 f(E)
W ((E)

2,0,3 1,2,2 0,4,1

Alam  ECE‐606 S09 9
For N‐states

Ni

Si
lnW

2,0,3  1,2,2   0,4,1

Si !
Wc = ∏ R ll W203 =
2!

5!

7!
i ( Si − N i )! N i ! Recall.  1!2! 0!5! 3!4!

ln(W ) = ∑ [ ln Si !− ln( Si − N i )!− ln N i !]


i Stirling approx.
 ∑ [ Si ln Si − Si − ( Si − N i ) ln( Si − N i ) + ( Si − N i ) − N i ln N i + N i ]
i

= ∑ [ Si ln Si − ( Si − N i ) ln( Si − N i ) − N i ln N i ]
i
10
Optimization with Lagrange‐Multiplier
Choose the most 
h h
probable configuration.

∂ ln W
δ ln(W ) = ∑ dN i lnW
i ∂N i
configurations
⎡ ⎛ Si ⎞ ⎤
 ∑ ⎢ln ⎜ − 1⎟ dN i ⎥ − α ∑ dN i − β ∑ Ei dN i
i ⎢
⎣ ⎝ Ni ⎠ ⎥⎦ i i

⎡ ⎛ Si ⎞ ⎤ ET = ∑ i Ei N i
= ∑ ⎢ln
l ⎜ − 1⎟ − α − β Ei ⎥dN i
i ⎢
⎣ ⎝ Ni ⎠ ⎥⎦
NT = ∑ i N i

Alam  ECE‐606 S09 11
Final steps …

⎡ ⎛ Si ⎞ ⎤
⎢ln ⎜ − 1⎟ − α − β Ei ⎥ = 0
⎢⎣ ⎝ N i ⎠ ⎥⎦ E

Ni 1 EF
f (E) ≡ = fmax(E)=1
Si 1 + eα + β E f(E)
1
At E = EF , f ( EF ) ≡ ⇒ α + β EF = 0
2
Ni 1 1
f (E) = = = ( E − EF ) / k B T
S i 1 + e β ( E − EF ) 1+ e
1
A E → ∞, f Boltzman ( E ) = Ae
At A − E / kB T ⇒ β=
k BT
Alam  ECE‐606 S09 12
Derivation by Detailed Balance

E=4
E=2
E 0
E=0

⎧⎪ f 0 ( E1 ) f 0 ( E2 )[
)[1 − f 0 ( E3 )][1
)][ − f 0 ( E4 )] − ⎫⎪
A⎨ ⎬=0
⎪⎩ f 0 ( E3 ) f 0 ( E4 )[1 − f 0 ( E1 )][1 − f 0 ( E2 )]⎪⎭

1
E1 + E2 = E3 + E4 Only solution is …. f0 ( E ) =
1 + e β ( E − EF )

‰ Pauli Principle, energy, and number conservation all satisfied


Alam  ECE‐606 S09 13
Derivation by Partition Function

e − β ( Ei − Ni EF ) e − β ( Ei − Ni EF )
Pi = ≡
E=4
E=2
∑e − β ( Ei − N i EF )

i
Z
E=0 β = 1 k BT

state Ei Ni Pi
− β ( 0 − 0× EF )
0 0 0 e Z
Ei − β ( Ei −1× EF )
1 1 1 e Z

Alam  ECE‐606 S09 14
Derivation by Partition Function
1
state Ei Ni Pi
− β ( 0 − 0× EF )
0 0 0 e Z
− β ( Ei −1× EF )
1 1 1 e Z

P1
Probability that state is filled ….
f (E) =
P0 + P1
E

f(E)
e−( Ei − EF ) / kBT / Z
=
EF

1 1 / Z + e−( Ei − EF ) / kBT / Z
1/2
1
=
f(E))

E 1 + e( Ei − EF ) / kBT
1/2
1

EF
Alam  ECE‐606 S09 15
Few comments on Fermi‐Dirac Statistics
‰ Applies to all spin‐1/2 particles

‰ Information about spin is not explicit; multiply DOS by 2. 
Information about spin is not explicit; multiply DOS by 2
May be more complicated for magnetic semiconductors. 

‰ Coulomb‐interaction among particles is neglected,
C l bi t ti ti l i l t d
Therefore it applies to extended solids, not to small molecules 

Lx 16
Alam  ECE‐606 S09
Outline

1) Rules of placing electronic states 
Rules of placing electronic states
2) Derivation of Fermi‐Dirac Statistics: three techniques
3) Intrinsic carrier concentration
I i i i i
4) Conclusion

Alam  ECE‐606 S09 17
Carrier Distribution 

DOS F‐D concentration


E 1− f (E) g c ( E ) f ( E ) dE
Etop
E n=∫
t

Ec

Ec gc ( E ) gc ( E ) f ( E )
EF
gυ ( E ) gυ ( E ) ⎡⎣1 − f ( E ) ⎤⎦

f (E) 1
gυ ( E ) ⎡⎣1 − f ( E ) ⎤⎦ dE

p=∫
Ebot

18
Electron Concentration in 3D solids

g c ( E ) f ( E ) dE
Etop
n=∫
Ec

Etop m*n 2m*n ( E − EC ) 1


=∫ 2× β ( E − EF )
dE
Ec 2π =2 3
1+ e
n ( E − EC )
* *
∞ m 2 m 1
∫
n
β ( E − Ec ) β ( Ec − EF )
dE
Ec π = 2 3
1+ e e
2
= NC F1 2 (ηc ) ηc ≡ β ( EF − EC )
π
32 ∞
⎛ 2π m*n β ⎞ ξ dξ
NC ≡ 2 ⎜ ⎟ F1 2 (η ) = ∫ ξ −η
⎝ h 2
⎠ 0
1 + e
Alam  ECE‐606 S09 19
Boltzmann vs. Fermi‐Dirac Statistics
2
n = NC F1 2 (ηc ) → N C eηc if − ηc ≡ β ( EC − EF ) > 3
π

eηc
F1 2 (ηc )

gc ( E ) f ( E )
EF
gυ ( E ) ⎡⎣1 − f ( E ) ⎤⎦

Alam  ECE‐606 S09 20
Effective Density of States

2
F1 2 (ηc ) → N C e
− β ( Ec − EF )
n = NC if Ec − EF > 3β
π

NC
gc ( E ) f ( E )
EF EF
gυ ( E ) ⎡⎣1 − f ( E ) ⎤⎦ NV

As if all states are at a single level EC

Alam  ECE‐606 S09 21
Law of Mass‐Action

− β ( Ec − EF )
n = NC e

+ β ( Ev − EF )
p = NV e EF

− β ( Ec − Ev )
n × p = N C NV e
− β Eg
= N C NV e

Alam  ECE‐606 S09 22
Fermi‐Level for Intrinsic Semiconductors

n = p = ni
− β Eg
n = N C NV e
2
i
EF
− β Eg 2
ni = N C NV e
EF ≡ Ei

n = p ⇒ NC e
− β ( Ec − Ei )
= NV e
+ β ( Ev − Ei ) E
3
EG 1 NV
Ei = + ln
2 2β NC 2
k
Alam  ECE‐606 S09 23
Conclusions

We discussed how electrons are distributed in electronic 
states defined by the solution of Schrodinger equation
states defined by the solution of Schrodinger equation. 

Since electrons are distributed according to their energy,
irrespective of their momentum states, the previously 
developed concepts of constant energy surfaces, density of 
states etc. turn out to be very useful.
y

We still do not know where EF is for general semiconductors … 
If we did we could calculate electron concentration
If we did, we could calculate electron concentration. 

Alam  ECE‐606 S09 24

Potrebbero piacerti anche