Optical Transitions in Semiconductors

In this lecture you will learn:

• Electron-photon Hamiltonian in semiconductors

• Optical transition matrix elements
• Optical absorption coefficients
• Stimulated absorption and stimulated emission
• Optical gain in semiconductors
• Spontaneous emission

ECE 407 – Spring 2009 – Farhan Rana – Cornell University

Semiconductors

850 nm
980 nm
InP
1300 nm
1550 nm

ECE 407 – Spring 2009 – Farhan Rana – Cornell University

1
 Interactions Between Light and Solids

The basic interactions between light and solids cover a wide variety of topics
that can include:

• Interband electronic transitions in solids

• Intraband electronic transitions and intersubband electronic transitions

• Plasmons and plasmon-polaritons

• Surface plasmons

• Excitons and exciton-polaritons

• Phonon and phonon-polaritons

• Nonlinear optics

• Quantum optics

• Optical spintronics

ECE 407 – Spring 2009 – Farhan Rana – Cornell University

Fermi’s Golden Rule: A Review

Consider a Hamiltonian with the following eigenstates and eigenenergies:

Hˆ o  m  E m  m  m  integer
Now suppose a time dependent externally applied potential is added to the
Hamiltonian:

Hˆ  Hˆ o  Hˆ  e  it  Hˆ  e it
Suppose at time t = 0 an electron was in some initial state k:  t  0    p
E
Fermi’s golden rule tells that the rate at which the electron absorbs
energy  from the time-dependent potential and makes a
transition to some higher energy level is given by:

2
 E m  E p   
2
W    m Hˆ   p
 m
The rate at which the electron gives away energy  to the time-
E
dependent potential and makes a transition to some lower energy
level is given by:

2 
 E m  E p   
2
W    m Hˆ   p
 m
ECE 407 – Spring 2009 – Farhan Rana – Cornell University

2
 Optical Transitions in Solids: Energy and Momentum Conservation
For an electron to absorb energy from a photon E
energy conservation implies:

   
 
E m kf  E n ki   Intraband

Final Initial Photon
energy energy energy
Interband
Momentum conservation implies: 
   k
 kf   ki  q

Final Initial Photon Intraband

momentum momentum momentum

Note that the momentum conservation principle is stated in terms of the crystal
momentum of the electrons. This principle will be derived later.

Intraband photonic transitions are not possible:

For parabolic bands, it can be shown that intraband optical transitions cannot satisfy
both energy and momentum conservation and are therefore not possible

ECE 407 – Spring 2009 – Farhan Rana – Cornell University

Electromagnetic Wave Basics

Consider an electromagnetic wave passing through a solid with electric field
given by:    
E r , t    nˆ Eo sinq . r   t 
The vector potential associated with the field is:
  
  Ar , t    E   E 
E r , t     Ar , t   nˆ o cosq . r   t  q
t 
 
 nˆ Ao cosq . r   t 

H
The divergence of the field is zero:
   
 . E r , t    . Ar , t   0

The power per unit area carried by the field is given by the Poynting vector:

      nEo2  2nAo2  2nAo2

S r , t   E r , t   H r , t   qˆ  qˆ I
2 oc 2 oc 2o c
The photon flux per unit area is:

Sˆ
 nAo2
F 
 2 oc
ECE 407 – Spring 2009 – Farhan Rana – Cornell University

3
 Photons and Photon Density

Photons and Photon Flux:

Energy in radiation fields can be added or subtracted in discrete units only, called
photons. Each photon has energy  .
n 2  2
So far a plane wave, whose intensity is = | Ao |
2oc
n 
the photon flow per unit area per second = | A0 |2
2oc
Photon Density:
The photon density (number of photons per unit volume) for a plane wave equals:

n 
| Ao |2
Photon flow per unit area per second 2oc
np = =
velocity of photons cn
1 2 2  1 2  1  
= n  o | Ao |2 =  r  o | Ao |2  r o | Eo |2
2  2  2 

ECE 407 – Spring 2009 – Farhan Rana – Cornell University

Photon Density in Dispersive Media

Phase Velocity and Group Velocity
In dispersive media, where the refractive index is frequency dependent, a distinction
needs to be made between the phase velocity v p of a wave and the group velocity v g
of a wave:
c c c ng  group index
vp  vg  
n n   dn d ng

The group velocity is the velocity at which the energy, and therefore the photons, travel.

Photon Density:
The photon density for a plane wave in a dispersive mdeium equals:
n 
| Ao |2
Photon flow per unit area per second 2oc
np = =
velocity of photons c ng
1 2  1   r   o  2
= nng o | Ao |2 = | Eo |
2  2  

ECE 407 – Spring 2009 – Farhan Rana – Cornell University

4
 Electron-Photon Hamiltonian in Solids
Consider electrons in a solid. The eigenstates (Bloch functions) and eigenenergies
satisfy:


Hˆ o  n ,k  E n k  n ,k
where:
 
Pˆ . Pˆ
ˆ
Ho 
2m

 Vlattice rˆ 
In the presence of E&M fields the Hamiltonian is:

ˆ
H

  
 
Pˆ  qA rˆ , t
2
 Vlattice rˆ


   
Ar , t   nˆ Ao cosq . r   t 
2m
 
Pˆ . Pˆ

2m


 Vlattice rˆ 
q ˆ  ˆ
2m
P . A r ,t 
2m
 
q  ˆ  q2  
A r , t . Pˆ 
2m
   
A rˆ , t . A rˆ , t    
 Hˆ o 
q ˆ  ˆ
2m
 
P . A r ,t 
q  ˆ
2m
A r , t . Pˆ

  Assume small

 Hˆ o 
q  ˆ
m
  
A r , t . Pˆ
 
Provided:  . Ar , t   0
 ˆ  ˆ
q Ao  e iq . r  i  t  e  iq . r  i  t  
 
 Hˆ o  nˆ  Pˆ
2m
ECE 407 – Spring 2009 – Farhan Rana – Cornell University

Optical Interband Transitions in Solids



Hˆ o  n ,k  E n k  n ,k E
   
e Ao  e iq . rˆ  i  t
e  iq . rˆ  i  t  
  ˆ
Hˆ  Hˆ o  P . nˆ
2m
Where:  
iq . rˆ
q Aoe 
Hˆ   Pˆ . nˆ
2m Interband
  
 iq . rˆ  k
q Aoe
Hˆ   Pˆ . nˆ 
2m ki
Suppose at time t = 0 the electron was sitting
 in the
valence band with crystal momentum k i :

 t  0    v ,k
i

The transition rate to states in the conduction band is given by the Fermi’s
golden rule:
 2  
   Ec kf   Ev k i    
2
W k i    c ,k Hˆ   v ,k
 kf f i

The summation is over all possible final states in the conduction band that have the
same spin as the initial state. Energy conservation is enforced by the delta function.

ECE 407 – Spring 2009 – Farhan Rana – Cornell University

5
 Optical Matrix Element
 2  
   Ec kf   Ev k i    
2
W k i    c ,k Hˆ   v ,k
 kf f i

2 2
2  qAo   ˆ   
     c ,k e iq . r Pˆ . nˆ  v ,k  Ec kf   Ev k i    
  2m  kf f i

E
Now consider the matrix element:

 ˆ 
 c ,k e iq . r Pˆ . nˆ  v ,k
f i
3  
   
  d r  c ,k f r
*
  e iq . r
Pˆ . nˆ  v ,k r  Interband
i
       
3   ik f . r *  
r  e iq . r Pˆ . nˆ e ik i . r uv ,k r 
 k
 d r e u c ,k f

 
    0 i
     
  d 3 r e  ik f r  e iq . r e ik i . r Pˆ  ki . nˆ uv ,k r 
.r * 
u c ,k f
i
      
  
  d 3 r e  ik f r  e iq . r e ik i . r Pˆ . nˆ uv ,k r 
.r * 
u c ,k f
i
Rapidly
varying in
Slowly space
varying in
space
ECE 407 – Spring 2009 – Farhan Rana – Cornell University

Optical Matrix Element

E
 
iq . rˆ

 c ,k e Pˆ . nˆ  v ,k 
f i
      
3  
 d r e  ikf . r *
u

c ,kf r  e iq . r e iki . r Pˆ . nˆ uv ,ki r 
  
e i ki q  kf  . r u *c ,kf r  Pˆ . nˆ uv ,k r 
3

    Interband
 
 d r 
k
i
R j j- th primitive cell
   
 

i k i q  k f . R j 3    
u *c ,kf r  Pˆ . nˆ uv ,k r 
 e d r i
Rj j- th primitive cell
   
 

i k i q  k f . R j 3    
u *c ,kf r  Pˆ . nˆ uv ,k r 
 e d r i
Rj any one primitive cell
3    
 N k   d r u *c ,kf r  Pˆ . nˆ uv ,k r 
i q , k f i
any one primitive cell

  k   Pcv . nˆ
i q , k f

ECE 407 – Spring 2009 – Farhan Rana – Cornell University

6
 Crystal Momentum Selection Rule
2 2
 2  qAo    
   Ec kf   Ev k i    
 ˆ
w  ki      c ,k e iq . r Pˆ . nˆ  v ,k
  2m  kf f i

2
2  qAo    
 Ec kf   Ev k i    
2
    Pcv . nˆ  k  
  2m  kf i q , k f
E
2
2    

 qAo 
 
2
 
Pcv . nˆ  Ec k i  q  Ev k i      
  2m 
Crystal Momentum Selection Rule:

We have the crystal momentum selection rule: 

Interband
   
kf  k i  q k

The wave vector of the photons is very small since

the speed of light is very large

Therefore, one may assume that:

  Optical transitions are
kf  k i vertical in k-space

ECE 407 – Spring 2009 – Farhan Rana – Cornell University

Transition Rates per Unit Volume

E
2
 2  qAo    
 
W k i  
  2m 
2
  
 Pcv . nˆ  Ec k i  Ev k i     
 

Generally one is not interested in the transition rate for any one W k i

particular initial electron state but in the number of transitions Interband

happening per unit volume of the material per second 

k

ki
The upward transition R rate per unit volume is obtained by
summing over all the possible initial states per unit volume
weighed by the occupation probability of the initial state and
by the probability that the final state is empty:
E

     
2   
R      W k i fv k i 1  fc k i
V ki 
R
If we assume, as in an intrinsic semiconductor, that the
valence band is full and the conduction band is empty of 
Interband
electrons, then: 
k
 
2 
R     
 W k i
V ki
2
2  qAo    

2
 
V ki   2m 

2
  
Pcv . nˆ  Ec k i  Ev k i     
ECE 407 – Spring 2009 – Farhan Rana – Cornell University

7
 Transition Rates per Unit Volume E

2
2  qAo    
R   
2

V k i   2m 

2
  
 Pcv . nˆ  Ec k i  Ev k i     
R
 2 
d 3 k i 2  qAo    
 2  3   2m 
P cv . ˆ
n
2
 E k
c i    
E k
v i     
Interband
 
FBZ 2     k

2 
2  qAo   d 3k i  
   Pcv . nˆ
2
2    
 E c k i  E v k i     
FBZ 2 
  2m  3

2 
2  qAo   d 3k  
   Pcv . nˆ
2
2   
 E c k  E v k    
FBZ 2 
  2m  3

The integral in the expression above is similar to the density of states integral

 2k 2


Suppose: Ec k  Ec 
2me
 32
d 3k
  
 1  2me 
Then: g3D E   2    E c k  E    E  Ec
FBZ 2 
3
2 2   2 

ECE 407 – Spring 2009 – Farhan Rana – Cornell University

Transition Rates per Unit Volume

2 
2  qAo   d 3k  
 Ec k   Ev k    
2
R      Pcv . nˆ 2 
  2m  FBZ 2  3
Joint density of
states
 2k 2  2k 2




Suppose: Ec k  Ec  Ev k  Ev  Reduced
2me 2mh effective mass
2k 2  1 1   2k 2
 
 
Then: E c k  E v k  E c  Ev      E g 
2  me mh  2mr
2 32
2  qAo   2 1  2m r 
R      Pcv . nˆ     E g E
  2m  2 2   2 

R   Joint density of states

R

Interband

k

Eg 
ECE 407 – Spring 2009 – Farhan Rana – Cornell University

8
 Transition Rates per Unit Volume
2 32 E
2  qAo   2 1  2m r 
R      Pcv . nˆ     E g
  2m  2 2   2 
R
The Intensity or the power per unit area and the photon
density of the E&M wave is: 
Interband

n 2  2 1 2  
I= | Ao | np = nng  o | Ao |2 k
2 oc 2 

The transition rate can be written as:

2 32
q  np   2 1  2mr 
R        Pcv . nˆ     E g
m   nn   2 2   2 
 g o

2 32
 q    o c I   2 1  2m r 
R        Pcv . nˆ     E g
 m    2n  2 2   2 

The transition rate depends on the photon density or the wave intensity

ECE 407 – Spring 2009 – Farhan Rana – Cornell University

Momentum Matrix Elements

In bulk III-V semiconductors with cubic symmetry, the average value of the
momentum matrix element is usually expressed in terms of the energy Ep as,
 2 mE p
Pcv . nˆ 
6
Note that the momentum matrix elements are independent of the direction of light
polarization!

Parameters at 300K GaAs AlAs InAs InP GaP

Ep (eV) 25.7 21.1 22.2 20.7 22.2

Transition Rates per Unit Volume and the

Absorption or Loss Coefficient
Because of transitions photons will be lost from the E&M wave traveling inside the
solid. This loss will result in decay of the wave Intensity with distance travelled:

 I  x   e  x I  x  0 
I x  I x  R I  x  x 
   I  x 
dx
 I  x  x   I  x    x I  x 
x x  x
Loss coefficient or absorption coefficient
ECE 407 – Spring 2009 – Farhan Rana – Cornell University

9
 Transition Rates per Unit Volume and Loss Coefficient
E
 I  x  x   I  x    x I  x 
The wave power loss (per unit area) in small distance x is
R
x I(x)

I x  I  x  x 
Interband
R 
k

x x  x
The wave power loss in small distance x must also equal:  R x

Therefore:
 R x   x I
 R   q
2
  2 1  2mr 
32
       Pcv . nˆ  2    E g
I  m   o n c 2
2   
Values of () for most semiconductors can range from a few hundred cm-1 to
hundred thousand cm-1

ECE 407 – Spring 2009 – Farhan Rana – Cornell University

Loss Coefficient of Semiconductors

ECE 407 – Spring 2009 – Farhan Rana – Cornell University

10
 Direct Bandgap and Indirect Bandgap Semiconductors

1  1 1 
  
mr  mh me 

If me approaches -mh ,
then mr becomes very
large
32
Then,   mr 
also becomes very
large

Direct bandgap Direct bandgap

(Direct optical transitions) (Indirect phonon-assisted transitions)

ECE 407 – Spring 2009 – Farhan Rana – Cornell University

Stimulated Absorption
Throwing back in the occupation factors one can write more generally:
2 
d 3k
2  qAo 
      
    
R      Pcv . nˆ
2
2  f k 1  fc k  E c k  Ev k  
3 v
 
  2m  FBZ 2 

Stimulated Absorption:
The process of photon absorption is called stimulated absorption because, quite
obviously, the process is initiated by the incoming radiation or photons

Stimulated
absorption


k

ECE 407 – Spring 2009 – Farhan Rana – Cornell University

11
 Stimulated Emission
An incoming photon can also cause the reverse process in which an electron makes
a downward transition. This reverse process is initiated by the term in Hamiltonian
that has the e it time dependence:   ˆ
q Ao e iq . r ˆ
Hˆ   P . nˆ
Hˆ  Hˆ o  Hˆ  e  it  Hˆ  e it 2m  
ˆ
q Ao e iq . r ˆ
Hˆ   P . nˆ
2m
Following the same procedure as for stimulated absorption, one can write the rate
per unit volume for the downward transitions as:
2 
d 3k
2  qAo 
      
    
R      Pcv . nˆ
2
2  fc k 1  fv k  Ec k  Ev k    
FBZ 2 
  2m  3

Stimulated Emission:
In the downward transition, the electron
gives off its energy to the electromagnetic
  field, i.e. it emits a photon! This process of
 photon emission caused by incoming
Stimulated k radiation (or by other photons) is called
emission  stimulated emission.

ECE 407 – Spring 2009 – Farhan Rana – Cornell University

Stimulated Absorption and Stimulated Emission

The net stimulated electronic transition rate is the difference between the
stimulated emission and stmulated absorption rates:
2 
d 3k
2  qAo 
      
    
R    R      Pcv . nˆ
2
2  fv k  fc k  Ec k  Ev k    
  2m  FBZ 2 3

And the more accurate expression for the loss coefficient is then:

 R    R    R    R  
     R    R    v g    n p
I np v g
2 
d 3k

      
    
q
  Pcv . nˆ
2
2  fv k  fc k  Ec k  Ev k    
 m   o n c 
FBZ 2 3

E E

Stimulated
Stimulated
absorption
 emission  
 
k k


ECE 407 – Spring 2009 – Farhan Rana – Cornell University

12
 Spontaneous Emission
Spontaneous Emission:
Electrons can also make downward transitions even in the absence of any incoming
radiation (or photons). This process is called spontaneous emission.
E Single radiation mode
Mode volume = Vp

 Semiconductor

Spontaneous k
emission
Temperature = T

Spontaneous Emission Rate:

Consider an optical cavity with a single mode of radiation at temperature T

The number of photons in the radiation mode is given by the Bose-Einstein factor:
1
Np 
e  KT
1

ECE 407 – Spring 2009 – Farhan Rana – Cornell University

Spontaneous Emission Rate

In thermal equilibrium, the emission and absorption

rates must be in balance:

R    R    0
2 
q    Np   d 3k

fc k   fv k   Ec k   Ev k      0
2    
     Pcv . nˆ 2 
 m    nng  o   V p  FBZ 2 
3
   
  
 fc k  fv k  0 for Ec k  Ev k   
The above condition does not hold in thermal equilibrium since: R    R  

Assume another emission mechanism such that: R    R sp    R    0

Also assume:
2 
  d 3k

      
   
R sp   
q
  
  nn  
 Pcv . nˆ
2
2  3 c
f k 1  fv k  Ec k  Ev k    
m  g o FBZ 2 
2 
 d 3k

      
    
q
 R sp    C   

 Pcv . nˆ

2
2  fc k 1  fv k  Ec k  Ev k    
 m    nng  o  FBZ 2 
3

ECE 407 – Spring 2009 – Farhan Rana – Cornell University

13
 Spontaneous Emission Rate
In thermal equilibrium, the emission and absorption
rates must be in balance:

R    R sp    R    0
 Np 
       
     

V  c
 f k  fv k  C fc k 1  fv k  0 for Ec k  Ev k    
 p
 
 v

f k  fc k   
CV p 


for Ec k  Ev k   
 
fc k 1  fv k   Np

 
 
 e Ec k  Ev k  KT  1  CV p e  KT  1 for Ec k  Ev k  
 
 
1
C 
Vp
The spontaneous emission rate is therefore:
2 
q   1   d 3k

      
   
R sp      


 
 Pcv . nˆ

2
2  fc k 1  fv k  Ec k  Ev k    
 m    nng  o   V p  FBZ 2 
3

The rate of spontaneous emission of a photon into a radiation mode is the same
as the rate of stimulated emission into the same radiation mode provided one
assumes that there is only one photon inside the radiation mode!

ECE 407 – Spring 2009 – Farhan Rana – Cornell University

Optical Gain in Semiconductors

2 
d 3k

fv k   fc k   Ec k   Ev k    
q     
    
2
 Pcv . nˆ 2 
 m   o n c FBZ 2 3

E
Note that power decays as: I  x   e  x
I  x  0

What if  were to become negative? Optical gain !!

    0  R    R   

A negative value of  implies optical gain (as opposed to k
optical loss) and means that stimulated emission rate
exceeds stimulated absorption rate

Optical gain is possible if: R    R  

   
 
fv k  fc k  0 for Ec k  Ev k    
   
 
 fc k  fv k  0 for Ec k  Ev k    
ECE 407 – Spring 2009 – Farhan Rana – Cornell University

14
 Optical Gain in Semiconductors
Optical gain is only possible in non-equilibrium situations when the electron and
hole Fermi levels are not the same

Suppose:
   
  1 E
fc k  f Ec k  Efe  
1  e Ec k  E fe  KT
   
  1
fv k  f Ev k  Efh 
1  e Ev k  E fh  KT


  Efe
 
Ec k  Ev k  

Then the condition for optical gain at frequency  is: 

k
 
     
f Ec k  Efe  f Ev k  Efh  0
Efh
 
 
 Efe  Efh  Ec k  Ev k
 Efe  Efh  

The above is the condition for population inversion (lots of electrons in the
conduction band and lots of holes in the valence band)

ECE 407 – Spring 2009 – Farhan Rana – Cornell University

Optical Gain in Semiconductors

The loss coefficient is a function of frequency and electron and hole densities:
2 
d 3k

      
    
q
      Pcv . nˆ
2
2  fv k  fc k  Ec k  Ev k    
 m   o n c FBZ 2 
3

   E

Efe
Net Loss
0 
Net Gain  k
Eg Efh
Increasing electron-hole
density (n = p)

Optical gain for frequencies for which:

E g    Efe  Efh

ECE 407 – Spring 2009 – Farhan Rana – Cornell University

15
 Photon Density of States
Consider a medium of volume Vp    
   A e iqr it Ao e iq r it 
The radiation modes are plane waves: A(r , t ) = nˆ  o e  e 
qc  2 Vp 2 Vp 
The wave dispersion is: 
n 
  3
In q-space, in a volume of size d 3q there are 2  V p d 3q 2  different plane
wave modes

   dn 
The dispersion relation gives:
n    d  d  dq c  d  v g dq

Since:
 2
d 3q q 2dq   n  d
2  Vp  Vp  Vp    Vp g p   d
2 3 2   c  vg

there are g p   different radiation modes per unit volume per unit frequency interval

2
 n 1
g p       Photon density of states
  c  vg

ECE 407 – Spring 2009 – Farhan Rana – Cornell University

Total Spontaneous Emission Rates in Semiconductors

The spontaneous emission rate into a single radiation mode is:

2 
 1   d 3k

      
   
q
R sp         P . nˆ
  nn    V  cv
2
2  fc k 1  fv k  Ec k  Ev k    
m  g o p FBZ 2  3

The spontaneous emission rate (per unit volume per second) into all radiation modes is:

 E
RTsp   R sp   g p   d
0
3
It can also be written as:
Efe
 g p    2
RTsp   v g     qV  KT d qV
0 e 1 
k
Efh
1

Spontaneous emission into different

radiation modes

ECE 407 – Spring 2009 – Farhan Rana – Cornell University

16