MODELING AND SIMULATION OF

MICROELECTRONIC DEVICES

SKEL 4653

TOPIC 2 : CARRIER STATISTICS

BASED ON THE BOOK: NANOELECTRONICS -

QUANTUM ENGINEERING OF LOW-
DIMENSIONAL NANOENSEMBLES
Modeling and Simulation of Microelectronic Devices

Topic 2: Carrier Statistics

Lecture 1
 Two-Level System-Equilibrium
E2  E1

Exponential Distribution : P12  P21e k BT

Pauli exclusion principle inhibits

transition to already occupied
quantum state. Let f1(E1) be the
probability of occupation at level 1
and f2(E2) at level 2. Then
degeneracy demands :

P12  P12 f1 (1  f 2 )
D

P21  P21 f 2 (1  f1 )
D

Equilibrium: P12  P21D D

 Equilibrium Statistics
E2  E1 E1 E2
 f1 f2
e k BT
f1 (1  f 2 )  f 2 (1  f1 )  e k BT
e k BT

1  f1 1  f2
E EF
f 1
e k BT
 constant, say e k BT
 f (E)  ( E  EF )/ k BT
1 f 1 e
Lessons Learned:
1. Fermi Energy remains constant
throughout the system independent of
any pair of E1 and E2.
2. If 1 and 2 are ends of the mean free path
(mfp), the equilibrium is established by
equal number of velocity vectors
directed in either x,y,z directions
 Band Diagram in Equilibrium
 
E1 E2
EF v  0
EC Intrinsic Velocity
vi  0
nd nd
with with
2 2
v  vi v  vi

nd EV
nd   nd  
2
d  3(bulk 3D), 2 (2 D),1 (1D)
 Equilibrium to Nonequilibrium
  
E2  E1  q E    qE cos( ) E
f
EF  q E 

e k BT
 e k BT
E1
 E2 1 f
qE 1
f ( E, E)   
E  ( EF  q E  )

EF 1 e k BT

EC NEADF (Chapter 4)
nd   nd 

E EV
 DOS, Distribution, and Concentration

3/ 2
1  2m*

D3e ( E )  
dse
 E  Ec
2 2  
2

3/ 2
1  2m*

D3h ( E )  2 
dsh
 Ev  E
2   2

The Fermi-Dirac Probability Distribution
Function
1
f(E)  E  EF  / k B T
1 e
  E  EF  / k B T
e (Maxwell - Boltzman Approx )
 Maxwell–Boltzmann approximation
  E  EF  / k B T
f(E)e (Maxwell - Boltzman Approx )
Determine the energy at which the Boltzmann approximation may be
considered valid. Calculate the energy, in terms of kT and EF, at which
the difference between the Boltzmann approximation and the Fermi–
Dirac function is 5 percent of the Fermi function
The Temperture Dependence and Rescaling
for Bandstructure
 Carrier Concentration-3D

top 

n3   D3se ( E ) f ( E )dE  N c  1 c 

EC 2
EV

p3   D3sv ( E )( 1  f ( E ))dE  N v  1 v 

bottom  2
 Electron Concentration
Etop  3/2
1  2m *  1
  E  Eco 
1/2
n3  dE 2  2 
2    E  EF

1
Eco k BT
e
E  Eco
x  E  Eco  k BTx dE  k BT dx
k BT
E  EF 3 E  Eco  Eco  EF 3 EF 3  Eco
  x  3 3 
k BT k BT k BT
3/2 
1  2m *  3 1 1
    k BT  
3/2
n3  2  2  dx x1/2

2    2 3 0 e x 3  1

 
2
 N c 3 1/2 3 
 Effective Density of States
3
m *
k T 2
 2 
dse( h )3 B
N c( v )3
 2  2 
  3 3
 mdse( h )3
*
  T  3
2 2
N c( v )3  2.54  10 
25
   m
 mo  300 
 
 Fermi-Dirac Integral
1 
j
 j   
x

 ( j  1) 0 1  e x 
dx

E F  Ec
 Normalized Fermi Energy
k BT

 j    e M  B Approx ( Non deg enerate)



1  1 j 1
 j     x dx 
j
(deg)
 ( j  1) 0  ( j  1) j  1
 Gamma Function

( j )   x j 1e  x dx 
0
 j  1! for integer j

( j  1)   x j e  x dx 
0
 j ! for integer j

1 3 1 1 1

          
2 2 2 2 2

Recursion : ( j  1)  j ( j )
 3D Non-Degenerate (ND) Electrons
EF 3  Eco

 j    e 3
e k BT

n3  Nc3 j 3   Nc3e 3

 Eco  EF 3 

 Nc3e kBT

n3 3
e
Nc3
3D Hole Concentration
 3D Non-Degenerate (ND) Holes
Evo  EF 3

 j    e
v 3
e k BT

 EF 3  Evo 

p3  Nv3e k BT
 Intrinsic Concentration-ND
EF3→ Ei3 and n3= p3= ni3

 Eco  Ei 3 

ni 3  N c 3e k BT
 Ei 3  Evo 

ni 3  N v 3 e k BT
Intrinsic Concentration-ND

Multiply the two ni Eqs.

Eliminate Ei
Ego

ni 3  N c 3 N v 3 e 2 k BT

Ego  Eco  Evo

Microelectronics I

Intrinsic semiconductor; A pure semiconductor with no impurity atoms

and no lattice defects in crystal
1. Carrier concentration(ni, pi)
2. Position of EFi

1. Intrinsic carrier concentration

Concentration of electron in in conduction band, ni

Concentration of hole in in valence band, pi

  ( EC  E Fi )    ( E Fi  Ev ) 
ni  pi  N C exp    N v exp  
 kT   kT 

2   ( EC  Ev )    Eg 
ni  N C N v exp    N c N v exp  
 kT   kT 
Independent of Fermi energy
 no and po in terms of ni
3/ 2
 2m kT 
*
  ( EC  E F )    ( EC  E F ) 
no  2 n
2
 exp    N C exp  
 h   kT   kT 

3/ 2
 2m kT 
*
  ( E F  Ev )    ( E F  Ev ) 
po  2  
p
exp   N exp  
 h 2   kT 
v
 kT 
 

  ( EC  E Fi )    ( E Fi  Ev ) 
ni  pi  N C exp    N v exp  
 kT   kT 
 n3 and p3 in terms of ni
 ( EC0  EF3 )   ( EF3  EV0 ) 
n3  Nc3 exp   p3  Nv3 exp  
 k BT   k BT 
 ( EC0  EFi3 )   ( EFi3  EV0 ) 
ni 3  Nc3 exp   pi3  Nv3 exp  
 k BT   k BT 

Divide no3/ni3 and po/ni3

 EF3  EFi3 
n3  ni 3 exp  
 k BT 
 EFi3  EF3 
p3  pi 3 exp  
 k BT 
Microelectronics I

2. Intrinsic Fermi level position, EFi

If EF closer to Ec, n>p

If EF closer to Ev, n<p

Intrinsic; n=p

EF is located near the center of the forbidden bandgap

  ( EC  E Fi )    ( EFi  Ev ) 
N C exp    N v exp  
 kT   kT 
3  m*p  Ec
EFi  Emidgap  kT ln * 
4 m 
 n Emidgap
Ev
Mp = m n EFi = Emidgap

Mp ≠ m n EFi shifts slightly from Emidgap

 Intrinsic Energy EF→ Ei
 ( EC0  EFi 3 )   ( EFi3  EV0 ) 
ni 3  pi3  Nc3 exp    Nv3 exp  
 k BT   k BT 
3/2
 ( EC0  2EFi 3  EV0 )  Nv3  m  *
exp    
dsh
*
 kBT  Nc 3  m  dse

3  mdsh
*

 EC0  2EFi 3  EV0  kBT ln  * 
2  mdse 
EC0  EV0 3  mdsh
*

EFi 3   kBT ln  * 
2 4  mdse 
Modeling and Simulation of Microelectronic Devices

Topic 2: Carrier Statistics

Lecture 2
 Degenerate versus Nondegenerate
n3
n3  N c3  1 c    1 c 
2 N c3 2
How to evaluate when u3 is given?

Nondegenerate: u3   1    e 

2
Nondegenerate:   ln( u3 )
Degenerate u3   1    u3   1    0
2 2

fzero function of MATLAB

allows to find  when seed
value say  is given
 MATLAB function file for FDI
Fermi.m downloadable from Mathworks or
nanohub.org is a MATLAB function file for
half integer value of FDI.
Nanohub.org has another function file for
arbitrary index of FDI. It is called FDjx.m
These function files are useful in finding
average values of various quantities like
velocity and energy
Table of Fermi–Dirac Integrals (Page 1)
Table of Fermi–Dirac Integrals (Page 2)
Table of Fermi–Dirac Integrals (Page 3)
 Intrinsic concentration, effective densities of
states, and effective mass in silicon
Martin A. Green
Effective DOS in Silicon

http://scitation.aip.org/content/aip/journal/jap/67/6/10.1063/1.345414
Example-ni3 for Si at 300 K
*
mdse  1.09 mo *
mdsh  1.15mo E g  1.12eV
3/2 3/2
3  m 
*
 T 
N c 3  2.54 10 cm 
19

ds
   2.86 1019 cm 3
 mo   300 
N v 3  3.10 1019 cm 3
 1.12  3
ni 3  2.86 1019  3.10 1019 exp   cm
 2  0.0259 
 1.08 1010 cm 3

 ( EC  Ev )    Eg 
ni  N C N v exp    N c N v exp  
 kT   kT 
Effective DOS in GaAs
 Example: p-type GaAs
*
mdse  0.067 mo *
mdsh  0.48mo Eg  1.43 eV , N A  10  1016 cm 3 , N D  9 1016 cm 3

*
mdse  0.067 mo *
mdsh  0.48mo Eg  1.43 eV ,
N A  10 1016 cm 3 , N D  9 1016 cm 3
p  N A  N D  1.0 1016 cm 3
3/2 3/2
3  mds   T 
*
N c  2.54  10 cm 
19
    4.4 1017 cm 3
 o 
m  300 
N v  8.3 1018 cm 3

 Nv 
EF  Ev  k BT ln  
 po 
 8.30 1018 
 0.0259eV ln  16 
 10 
 0.17eV
 Minority and Majority
Carriers
1.43 eV

ni  4.4 1017  8.3 1019 e 20.0259 eV
cm 3
 2.2 106 cm 3

np  ni  n 
2 n

 2.2
2
i
 10 cm 6

3 2

p 1016 cm 3
4 3
 2.2  10 cm 0
 Uncompensated versus Compensated

3
N A  4 10 cm 16

3 3
N A  4  10 cm
16
N D  1  10 cm
17

n3  N D  N A  6  1016 cm 3
 Strong Degenerate 3Doping

2 3 2
n3  N c3  1 c   N c3
2  3/ 2
2

 
2
 3  n3  3  2 3 2 n3 3
EF  Eco    k BT 
 4 Nc  2m*n
1
1

 3 n3 2
 3


k F  3 n3
2
 3 vF 
m*
Microelectronics I

Charge-neutrality condition

no  N a  po  N d
Negative charges Positive charges

no  ( N a  pa )  po  ( N d  nd )
 Charge Neutrality
Negative charges
 
p3  N  n3  N
Positive charges

D A
2
n  
 N D  n3  N A
i
n3
n3  n3 ( N D  N A )  ni 3  0
2 2

2
ND  N A  ND  N A 
n3      ni
2

2  2 
2
N A  ND  N A  ND 
p3      n 2
i3
2  2 
 Charge Neutrality-Approximation

n3  N D  N A when N D  N A  ni

p3  N A  N D when N A  N D  ni

Perfect compensation : N A  N D
n3  p3  ni 3
n3 versus T
Modeling and Simulation of Microelectronic Devices

Topic 2: Carrier Statistics

Lecture 3
 Carrier Freeze-Out
 
N D  ND N A  N A ( partial Ionization )
N D 1
 ( EF  E D )
g D  2 for spin states
ND
1  gDe k BT

N A 1
 ( E A  EF )
NA
1  g Ae k BT

g A  2  2  4 for spin states l  h holes

Charge Neutrality:
 
p  n  N  NA  0 D
n3/ND versus Temperature
 Quasi-Low-Dimensional QWs
2

Ek  Ec 0 
2m *
2

kx  k y  kz
2 2
 (3D)
2

Ek  Ec 2 
2m2 *2
2

kx  k y
2
 (2 D)
 kx
Ek  Ec1  (1D)
2m *
Ecd Eco 0d d 3(bulk), 2(2D),1(1D)
03 0 02 0z 01 0z 0y
 Carrier Statistics in Low-D QW
nd  N d  d
1
d  d  3, 2, or 1
2
 m * k BT 
d /2
2 
N d  2  D 
2 
 2  2 D 
d 2m * k B T

( EF  Ec ) d
d 
kBT
 Fermi Energy
Degenerate and Nondegenerate
Fd
 Intrinsic Velocity
1 2( E  Ec ) 2 k BT
E  Ec  m * v 2 v  vth x1/2 vth 
2 m* m*
nd  N d  d  2  d 
2
 d 1 d   d 1 d 
vid  vthd 2
 vthd 2
 d  2 d  nd / N cd
2

 d 1
 
 2 
vthd  vth
d 
 
2
 Intrinsic Velocity
Nondegenerate Approximation
 d 1   d 1 
   
2 2 2
 vth    
k T
vid  vthd B

d  m* d 
   
2 2
2 8 kBT
vi 3  vth 3  vth 
 m*
  kBT
vi 2  vth 2  vth 
2 2m *
1 2 kBT
vi1  vth1  vth 
 m*
 Intrinsic Velocity
Degenerate Approximation
d d 2( EF  Ec ) d
vid  vFd 
d 1 d 1 m*
 1

d     d  2  nd  d 
vid  2     

d 1  m *   2  2 
 
1
d    d  2  nd  d
vid  2     
d 1 m *   2  2
 Derivation
Intrinsic Velocity Degenerate Approximation
d d 2( EF  Ec )d
vidDeg  vFd 
d 1 d 1 m*
1/2
d d  2  ( EF  Ec )d
 vFd  η  k T where d 
d 1 d  1  m* d B  k BT

Fermi energy is a strong function of carrier concentration

1 η j 1 η j 1
 j  η  
  j  1 j  1   j  2 

For dimensional structure, replace j with  d  2  2

1 ηd /2 ηd 2
 d  2 2  ηd   
  j  1 j  1   d 2  1
 Derivation
It is shown that carrier statistics
nd  N cd  d 2 2  ηd 
ηd 2
 N cd
  d 2  1
nd  d 
ηd 2     1
N cd  2 
2/ d d /2
 n d   mn* k BT 
η   d    1  where N cd  2 2 
 N cd  2   2   
2/ d
 
 
 nd d  Intrinsic Velocity Degenerate Approximation
 d /2
   1  1/2
  m k BT 
* 2  d d  2 
vidDeg  vFd  η  k T
 2  2  2   d 1 d  1  m* d B 
   
2/ d
 nd  d  2   1
η    
 2  2   m*k BT 
 2 
 2  
 Derivation
Next we replace η in vidDeg
2/ d
2  n  d  2  2  2 2
v 2
 ηd  k BT   d    k T
Fd
m*  2  2  m*k BT m* B
2/ d
 n  d  2  22   2
  d  
 2  2  m*2

d d 2( EF  Ec )d
vidDeg  vFd 
d 1 d 1 m*
1/2
d   nd  d  2   22   2 
2/ d
d
 v    
d  1 Fd d  1   2  2   m*2 
 
1/2 1/ d
d  22   2   nd  d  2  
     
d  1  m*2   2  2 

1
d    d  2  nd d
vidDeg  2     2
d 1 m*
  2  
Intrinsic Velocity Degenerate
Approximation
3 1/3
vi 3  3 n3 
2

4m *


8 n2 / 9
1/2
vi 2 
m*


vi1  n1
4m *
Intrinsic Velocity in Low-D Systems
8
3D
7 2D
1D
6

5
(vi/vth)d

1 -2 -1 0 1
10 10 10 10
(n/N)
d
 Mean Energy and rms velocity
 d  Fd 
d
Emd  k BT 2
2  d  2  Fd 
2

 d  Fd 
d k BT
vrms d  2
m *  d  2  Fd 
2
 Mean Energy
Nondegenerate Approximation
3 3k BT
Em 3  k BT vrms 3 
2 m*
2 2 k BT
Em 2  k BT  k B T vrms 2 
2 m*

1 k BT
Em1  k BT vrms1 
2 m*
 Intrinsic Energy
Degenerate Approximation
2/ d
d d 2
d  d  
Emd  ( EF  Ec ) d  2     nd / 2 
d 2 d  2 m*  2  2  
2/3
6   3  
2
Em 3   n3 
5 m*  8 
 2
Em 2  n2
2 m*

 2 2
Em1  n12
24 m *
 RMS Velocity
Degenerate Approximation
2 d
vrms d  ( EF  Ec ) d
m* d  2
1/2 1/ d
  d  d  d  
  4   2   2  nd / 2 
m* d  2     
1/3
  12 
1/ 2
3     n1
vrms 3     n3  vrms1 
m* 5   8  m* 2 3


 n2 
1/2
vrms 2 
m*
 Unidirectional Transport
As we will find out that in a high electric field, the
Fermi energy is lifted in the degenerate regime because
of Pauli Exclusion Principle. The velocity vectors
transform to an ordered phase with lifting of Fermi
energy as electrons cannot go to already occupied states.
The unidirectional intrinsic velocity is appropriate for
2n electrons as only the half of k-space is occupied. As
degeneracy advances from nondegenerate level to
extreme degenerate level, intrinsic velocity changes
from vid to vud that is reflected as geometic mean vid vud
in the definition of intrinsic temperature. Intrinsic
temperature Ti is higher than the lattice tempertaure
ultimately becoming proportional to the Fermi Energy.
Modeling and Simulation of Microelectronic Devices

Topic 2: Carrier Statistics

Examples
 Example 1
The Fermi energy versus carrier concentration indicates that
electron gas is strongly degenerate when the Fermi energy is 3kBT
above the edge of the conduction band.
a) Find the probability of finding the electron at the conduction
band edge Ec.
b) Find the probability of finding electron at the Fermi energy
EF.
c) What is the carrier concentration (in m−3) of electrons in
GaAs at room temperature at the threshold of being strongly
degenerate?
d) Sketch a band diagram showing the intrinsic Fermi energy,
Fermi energy, and bandgap with appropriate values for GaAs
at room temperature.
 Example 1a
The Fermi energy versus carrier concentration indicates that
electron gas is strongly degenerate when the Fermi energy is 3kBT
above the edge of the conduction band.
a) Find the probability of finding the electron at the conduction
band edge Ec.
 Example 1b
The Fermi energy versus carrier concentration indicates that
electron gas is strongly degenerate when the Fermi energy is 3kBT
above the edge of the conduction band.
b) Find the probability of finding electron at the Fermi energy EF.
 Example 1c
The Fermi energy versus carrier concentration indicates that
electron gas is strongly degenerate when the Fermi energy is 3kBT
above the edge of the conduction band.
c) What is the carrier concentration (in m−3) of electrons in
GaAs at room temperature at the threshold of being strongly
degenerate?
Table of Fermi–Dirac Integrals (Page 3)
 Example 1d
The Fermi energy versus carrier concentration indicates that
electron gas is strongly degenerate when the Fermi energy is 3kBT
above the edge of the conduction band.
d) Sketch a band diagram showing the intrinsic Fermi energy,
Fermi energy, and bandgap with appropriate values for GaAs
at room temperature.

0.71 eV
1.42 eV
0.71 eV
 Example 2
Following the same procedure as for the carrier concentration,
derive the expression for the intrinsic velocity for d=1, 2, and 3.
Discuss its degenerate and nondegenerate limits.

The general intrinsic velocity expression is calculated as,

æd + 1ö÷ Áæ ö (ηd ) Áæd-1ö (ηd )

Γ ççç ÷÷ çç ÷÷÷
d -1 çç ÷÷
è 2 ø è 2 ø ç è 2 ÷ø
ç
νid = vth = vthd
æd ö÷ Áæd-2ö (ηd ) Áæd-2ö (ηd )
Γ ççç ÷÷ çç ÷
÷ çç ÷÷
è 2ø çè 2 ø÷ èç 2 ÷ø

1
d    d  2  nd  d
vid  2     
d 1 m *   2  2
 Example 2
Following the same procedure as for the carrier concentration,
derive the expression for the intrinsic velocity for d=1, 2, and 3.
Discuss its degenerate and nondegenerate limits.
 Example 3
The energy distribution per unit volume per unit energy of
electrons nE (E) in the conduction band is given by

where A is a constant.
a) Show that the maximum in the energy distribution is at
E − Ec = 1/2 kBT. Hint: nE is maximum when dnE/dE = 0.
 Example 3

a) Show that the maximum in the energy distribution is at

E − Ec = 1/2 kBT. Hint: nE is maximum when dnE/dE = 0.
 Example 4
Two semiconductors A and B have the same DOS effective
masses. The bandgap Eg = 1 eV for A and 2 eV for B. Find the
ratio niA/niB of the intrinsic carrier concentrations.
 Example 5
Silicon is doped with ND = 5 × 1015 cm−3. Find n0 and p0 and locate the Fermi
level. Draw the energy band diagram with numerical values indicated with
respect to each band edge (conduction and valence band edge).
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, then n0 =
ND. Thus, assuming ND material (which
we will check later). So
 Example 6
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 with numerical values indicated with respect to each
band edge (conduction and valence band edge).