Semiconductor Optoelectronics (Farhan Rana, Cornell University)

Chapter 11

Basics of Semiconductor Lasers

11.1 Introduction
11.1.1 Introduction to Semiconductor Lasers:
In semiconductor optical amplifiers (SOAs), photons multiplied via stimulated emission. In SOAs
photons were confined in the dimensions transverse to the waveguide but were allowed to escape from the
end of the waveguide. We now consider optical cavities in which the photons are confined in all three
dimensions and kept inside the cavity for much longer durations allowing them to multiply via stimulated
emission and thereby generating a large population of photons inside the cavity. The simplest way to
realize an optical cavity is to take an SOA waveguide and coat both the input and output facets with a
high-reflectivity (HR) optical coating (SOAs generally have an anti-reflecting (AR) coating on the input
and output facets). The optical cavity thus obtained is called a Fabry-Perot cavity and is shown below.

R1 R
2

z=0 z=L
~
Suppose the modal gain is a g and the facet reflectivities are R1 and R 2 . If one follows a guided mode
through one complete roundtrip of the cavity, one finds that the change in optical power after one
complete roundtrip is,
R1R2e a g  2L
~ ~

When g ~ is small such that, R R e a g~ ~ 2L  1, any photons introduced into the cavity will
1 2
eventually leave the cavity (it will either be transmitted out of the cavity through either one of the two
facts of the cavity or it will leave the cavity by being absorbed in the cavity because of the modal loss ~ ).
The question that arises is what if the gain g ~ is made large enough such that R R e a g~ ~ 2L
1 2
approaches unity? In this case, the number of photons lost in one complete roundtrip from the facets or
due to the waveguide loss equals the increase in the number of photons in one roundtrip due to stimulated
emission. In other words, the cavity gain per roundtrip equals the cavity loss per roundtrip. The condition,
R1R 2e a g  2L  1
~ ~

is the lasing condition. When this condition is satisfied, a large photon population can build up inside the
cavity starting from spontaneous emission and we have a laser (light amplification from stimulated
emission of radiation).
 Semiconductor Optoelectronics (Farhan Rana, Cornell University)

The simplest way to analyze and understand laser dynamics is using rate equations. In this Chapter, we
will setup laser rate equations using the Fabry-Perot optical cavity as a model.

11.2 Photon Density Rate Equation

11.2.1 Laser Threshold Gain:
The value of the material gain that satisfies the lasing condition,
R1R 2e a g  2L  1
~ ~

~ . We can write the expression for the threshold gain as,

is called the threshold gain gth

1  1 
a g~th  log   ~
L  R1R2 

The threshold gain is function of the parameters of the optical cavity. The lasing condition states that
photons multiply via stimulated emission at the same rate inside the cavity as the rate at which they are
being lost from the cavity. There are two sources of loss; the loss from the facets (or the mirrors) of the
cavity and the intrinsic loss ~ from the waveguide. We can define a mirror loss ~m as follows,
1  1 

~m  log
L  R1R2 

The lasing condition then becomes,
 g~  ~  ~
a th m

11.2.2 Effective Mode Volume:

For the Fabry-Perot optical cavity, the volume of the active region is Va  WhL . Here, W is the width of
the active region and h is the height of the active region. The area of the active region is Aa  Wh . The
effective area of the mode is Aeff  Aa a and the effective volume of the mode is Vp  Aeff L .

We now generalize the definition of the mode volume to all types of optical cavities. For any optical
cavity, the optical mode confinement factor a for the active region is equal to the ratio of the mode
energy in the active region to the total mode energy,
M
  3
  o n n g E.E * d r
a  active  
M 3
  o n n g E.E * d r

Active region

Optical cavity

Va
 Semiconductor Optoelectronics (Farhan Rana, Cornell University)

Suppose the total number of photons in the cavity is N p and the average photon density in the active
region is n p , then,
n pVa
a 
Np
The effective mode volume Vp is defined by the expression,
V
Vp  a
a
This definition implies that the total number of photons N p can be written as,
N p  n pV p
Note that for the Fabry-Perot cavity,
V A L Aa
Vp  Aeff L  a  a  Aeff 
a a a

11.2.3 Cavity Optical Gain:

Consider an optical cavity with N p photons. The rate of stimulated emission (number of stimulated
transitions per second per unit volume of the active region) is v gM gn p . If the cavity is a Fabry-Perot
cavity then the stimulated emission rate is usually written in terms of the waveguide group velocity v g
as, v g~n . Assuming a Fabry-Perot cavity, the total stimulated emission rate in the active region is,
g p
v g g~n pVa . But,
V
v g g~n pVa  v g g~n p a V p  a v g g~N p
Vp
Photon multiplication in the cavity due to stimulated emission can be expressed as,
dN p
 a v g g~N p
dt
The quantity a v g g~ is the cavity optical gain since it describes the rate of increase of the number of
photons inside the cavity due to stimulated emission.

11.2.4 Cavity Optical Loss and Cavity Photon Lifetime:

Cavity photon lifetime  p expresses the rate at which photons are lost from the cavity, for example, by
escaping from the end facets of a Fabry-Perot cavity or by getting absorbed in the cavity material
(excluding stimulated absorption in the active region). The rate of photon loss is, N p  p . The total rate
of change of photon number in the cavity is therefore,
dN p  1
  av g g~  N p
dt   p 
The first term inside the bracket on the right hand side describes optical gain and the second term
describes the optical loss. The condition for lasing is that the rate of increase of photon number due to
stimulated emission (optical gain) is equal to the rate of decrease of photon number due to loss. In other
words, the threshold gain must satisfy,
 Semiconductor Optoelectronics (Farhan Rana, Cornell University)

1
a v g g~th 
p
Using the expression obtained earlier for the threshold gain of a Fabry-Perot cavity,
a g~th  ~m  ~
we get an expression for the photon lifetime of a Fabry-Perot cavity,
1 vg  1 
 v g ~m  ~  log  v g ~
p L  RR 
 1 2 
For more complex optical cavities, numerical methods are generally employed to calculate photon
lifetimes, as we will see in the following Chapters. Once the photon lifetime has been determined, the
threshold gain can be calculated.

11.2.5 Spontaneous Emission:

The rate equation for the photon number,
dN p  1
 a v g g~  N p
dt   p 
does not consider the increase in the photon number due to spontaneous emission. Remembering that the
spontaneous emission rate equals the stimulated emission rate assuming one photon in the optical mode,
one can write,
dN p  1
 a v g g~  ~
N p  a v g g n sp
dt   p 
The spontaneous emission factor n sp is needed here since the gain g~ is related to the net stimulated rate
(stimulated emission rate minus the stimulated absorption rate). The above equation is usually expressed
not in terms of the total photon number N p but in terms of the average photon density n p in the active
region by dividing both sides by Vp ,
dn p 1 ~ n sp
 a v g g~   n p  a v g g
dt   p  Vp
The above equation is the main result that we will use throughout this Chapter.

11.3 Carrier Density Rate Equation

11.3.1 Carrier Density Rate Equation:
We assume that the active region of the laser is biased with a current source, as shown below.

R1 R
2
I
 Semiconductor Optoelectronics (Farhan Rana, Cornell University)

The rate equation for the carrier density is the same as the one used for SOAs,
dn i I
  Rnr n   Gnr n   Rr n   Gr n   v g g~ np
dt qVa
The non-radiative recombination-generation rates and the radiative recombination-generation rates due to
spontaneous emission into all the radiation modes are expressed as follows,
 
Rnr (n)  Gnr n   A n  ni   C n n 2  ni2  An  Cn 3
 
Rr (n)  Gr n   B n 2  ni2  Bn2
The gain is also a function of the carrier density and this dependence is usually approximated by a
logarithmic function,
 n 
g~n   g~o log 
 ntr 

11.3.2 Threshold Carrier Density:

The carrier density at which the gain g~ equals the threshold gain g~th is called the threshold carrier
density. If,
 n 
g~n   g~o log 
 ntr 
~ ~
then, nth  ntr e g th g o .

11.4 Laser Rate Equations

11.4.1 Laser Rate Equations:
We can now write down the laser rate equations for the photon density and the carrier density as follows,
dn p  1 nsp
  av g g~  n p  av g g~ (1)
dt   p  Vp

dn  i I
  Rnr n   Gnr n   Rr n   Gr n   v g g~ n p (2)
dt qVa
The above two coupled nonlinear equations can exhibit a wide variety of dynamics associated with the
operation of semiconductor lasers.

11.4.2 Laser Output Power:

We also need expressions for the light coming out of the laser. Photons leave the cavity in two ways; they
can either escape from the end facets (or mirrors) or they can get absorbed by the cavity. Only the photons
that leave the cavity from the mirrors constitute useful output. We define an output coupling efficiency
o of the laser as,
~
o  ~ m ~
 
m
The output coupling efficiency o is equal to the fraction of the photons leaving the cavity from the end
facets (or mirrors). The total number of photons leaving the cavity per second is,
 Semiconductor Optoelectronics (Farhan Rana, Cornell University)

Np n pVp

p p
The number of photons leaving the cavity from the mirrors is therefore,
Np n pVp
o  o
p p
The output power P of the laser is.
Np n pVp
P  o   o 
p p

11.5 Operation of Semiconductor Lasers

11.5.1 Introduction:
The nonlinear laser equations can easily be solved numerically on a computer. However, more insight is
obtained using approximate analytical solutions in different regimes of operation. We need to solve
Equations (1) and (2) above in steady state for different values of the current bias. Steady state implies,
dn p dt  dn dt  0 . So the equations that need to be solved are,
nsp
av g g~
Vp
np  (3)
1
 av g g~
p
i I
 Rnr n   Gnr n   Rr n   Gr n   v g g~ np (4)
qVa

11.5.2 Regime I ( n  ntr ) – Laser below Threshold:

Suppose the current is switched on from zero and is small enough such that n  n tr and the gain is
negative. Any photons emitted spontaneously into the cavity will experience loss from the mirrors ( ~m ),
from the waveguide ( ~ ), as well as from the active region. Photons will therefore not last for long inside
the cavity and the photon density inside the cavity will be very small. In this regime, one can ignore the
stimulated emission term in Equation (4) and determine carrier density from the equation,
i I
 Rnr n   Gnr n   Rr n   Gr n  (5)
qVa
Once the carrier density has been determined using Equation (5) above for a given current bias, the
photon density can be determined using Equation (3) above.

11.5.3 Regime II ( ntr  n  nth ) – Laser below Threshold:

As the current is increased, the carrier density will increase as dictated by Equation (5) above. At some
point the carrier density will exceed the transparency carrier density and the gain will become positive.
Photons emitted spontaneously into the cavity will multiply via stimulated emission. Since n  n th , and
g~  g~ , the photon multiplication rate is not large enough to balance the photon loss rate from the
th
cavity and therefore the photon density in the cavity will be small. Again, one can ignore the stimulated
 Semiconductor Optoelectronics (Farhan Rana, Cornell University)

emission term in Equation (4) and determine carrier density from Equation (5) and once the carrier
density has been determined using Equation (5) for a given current bias, the photon density can be
determined using Equation (3).

11.5.4 Regime III ( n  nth ) – Laser near Threshold:

As the current is increased further, at some current value the carrier density, as predicted by Equation (5),
will equal the threshold carrier density n th . The current for which the carrier density, as predicted by
Equation (5), equals the threshold carrier density is called the threshold current and is given as,
i Ith
 Rnr nth   Gnr nth   Rr nth   Gr nth  (6)
qVa
When the carrier density n equals the threshold carrier density n , the gain g~ equals the threshold gain
th
g~th and the photon density, as given by Equation (3) is infinite because the denominator is zero. In fact,
if the carrier density were to exceed the threshold carrier density, the photon density, as given by
Equation (3), would be negative – an obviously unphysical result. What is happening is that as the current
is increased, and the carrier density increases and approaches the threshold carrier density, the gain
approaches the threshold gain. As the gain increases, the simulated emission rate increases and
approaches the rate at which the photons are lost from the cavity. Every photon in the cavity now has a
chance to multiply before it is lost from the cavity and so the steady state photon population inside the
cavity also increases. Equation (3), reproduced below,
nsp
av g g~
Vp
np 
1
 av g g~
p
predicts that as the gain g~ approaches the threshold gain g~ , and optical gain  v g~ approaches the
th a g
optical loss 1  p , the steady state photon density increases significantly because the denominator
approaches zero. When the photon density becomes very large, Equation (5) is no longer valid because
carrier recombination rate due to the stimulated emission cannot be ignored compared to the other non-
radiative and radiative recombination rates. Therefore, Equation (4) has to be used to calculate the carrier
density in steady state. Equation (4) shows that as the photon density increases significantly, the
stimulated emission rate av g g~n p also increases and keeps the carrier density, and therefore the gain,
from increasing as much with current as when the stimulated emission rate is ignored. In fact, because the
photon density increases drastically when the optical gain av g g~ gets close to the optical loss 1  p , the
increased stimulated emission rate never allows the carrier density to ever exceed the threshold carrier
density nth and, therefore, the gain never exceeds the threshold gain g~th . This gain saturation brought
about by a large photon density is needed to stabilize the photon density inside the cavity. If the optical
gain were to exceed the threshold gain then the photon multiplication rate due to stimulated emission
would exceed the photon loss rate and the steady state photon density would increase to infinity. In other
words, there would be no steady state and this is the reason why Equation (3) predicts a negative photon
density for av g g~  1  p .

The Figure below shows the carrier density and the photon density vs. the current for regimes I-III.
 Semiconductor Optoelectronics (Farhan Rana, Cornell University)

n np

nth

n
tr

Ith I I I
th

11.5.5 Regime IV ( n  nth ) – Laser above Threshold:

When the current is increased beyond the threshold current Ith , the photon density becomes so large that
the resulting increased recombination due to stimulated emission prevents the carrier density from
increasing beyond nth . The carrier density gets fixed at a value close to (but less than) nth . To find the
photon density when I  Ith , we start from Equation (4) and subtract Equation (6) to get,
i I  Ith 
 Rnr n   Gnr n   Rr n   Gr n   Rnr nth   Gnr nth   Rr nth   Gr nth 
qVa
 v g g~ np
Since for I  Ith , n  nth and g~  g~th  1 av g p , the term inside the curly brackets is close to zero,
and we get,
i I  Ith  np

qVa a p

 Np  npVp  i
I  Ith  
p
q
The above equation shows that the photon density increases linearly with the current when the current
exceeds the threshold current. The point where I  Ith is called the “threshold for lasing” or just as the
“laser threshold.” Above threshold, the carrier density n , and the optical gain g~ , remain fixed to their
values at threshold, and the photon density increases with the current and the corresponding increase in
the stimulated emission rate is just enough in order to maintain the carrier density at its threshold value as
the current is increased. The Figure below shows the carrier density and the photon density vs. the current
for a laser operating below and above threshold. The rapid buildup of photon population when I  Ith is
called lasing.
n np

nth

n
tr

Ith I I I
th
 Semiconductor Optoelectronics (Farhan Rana, Cornell University)

11.5.6 Laser Output Power above Threshold:

The output power of the laser is the power coming out from the two end facets of the laser cavity. Since,
Np n pVp
P  o   o 
p p
the laser power above threshold in terms of the current is,
Np 
P   o   oi I  Ith 
p q
The above expressions shows that if oi were equal to unity then every electron injected into the laser
per second above the threshold injection rate of Ith q would end up producing a photon in the laser
output. Above threshold, the laser is therefore a very efficient converter of electrical energy into optical
energy. This property of the laser is commonly expressed in terms of the differential slope efficiency
dS ,
dP 
dS   oi
dI q
or the differential quantum efficiency  dQ ,
q dP
dQ   oi
 dI

11.5.7 A Worked Example:

Consider an InGaAsP/InP laser (shown in the Figure below) with the following parameter values:

Top metal
Optical Power (mW)

Insulator

1.5 m
InP substrate

Current (mA)
(LEFT) Output facet of a 5 QW InGaAsP/InP laser for 1.55 m operation. (RIGHT) Measured LI
characteristics of the laser.

A multiple quantum well (MQW) active

region with 7 nm wells, 9 nm barriers p doped SCH SCH n doped
and 100 nm SCH regions.

Laser length = L = 500 m

 Semiconductor Optoelectronics (Farhan Rana, Cornell University)

Active region width = W = 1.5 m

Active region height (all 5 quantum wells) = h = 0.035 m = 35 nm
Facet reflectivities = R1 = R2 = 0.3
Transparency carrier density = ntr = 1.75x1018 1/cm3
Active region mode confinement factor (for all 5 quantum wells) = a = 0.07
Waveguide group velocity = vg = c/3.4
Waveguide modal loss = ~ = 15 1/cm
A=0
B = 10-9 cm3/s
C = 5x10-29 cm6/s
g~o =1500 1/cm
Current injection efficiency =  i = 0.85

Using the above parameters we can calculate the remaining laser parameters as follows. The effective
mirror loss is,
1  1 
~m  log  24 1/cm
L  R1R2 
The output coupling efficiency is,
~
 o  ~ m ~  0.62
 
m
The photon lifetime in the cavity is,
1 vg  1 
 v g ~m  ~  log   v ~
g
p L  RR 
 1 2 
  p  2.9 ps
The threshold gain is,
1
 v g~ 
a g th
p
 g~th  558 1/cm
Since,
 n 
g~  g~o log 
 ntr 
we can calculate the threshold carrier density,
~ ~
nth  n tr e gth go  2.59  1018 1/cm3
From the threshold carrier density we can obtain the threshold current,
 i I th
 R nr nth   Gnr nth   R r nth   Gr nth   Anth  Bnth
2 3
 Cnth
qVa
I th  37.5 mA
The calculated threshold current value compares favorably with the observed value in the Figure above.
 Semiconductor Optoelectronics (Farhan Rana, Cornell University)

11.6 Laser Stability and Relaxation Oscillations

11.6.1 Introduction:
Suppose a laser is biased with current I above threshold. The steady state values of carrier and photon
densities are n and n p , respectively. Since I I th , n  n th . Suppose at time t  0 , the carrier density is
suddenly increased from n to n  n . The new value of the carrier density is not the steady state value
and we would like to see how, if at all, the steady state value is recovered. The carrier density n  n is
greater than n th , and, consequently, the gain g~ is greater than the threshold gain g~th . In steady state, g~
can never be greater than g~th , but this restriction does not hold in non-steady state situations.
Alternatively, we could have perturbed the photon density at time t  0 to n p  n p . If the carrier or
photon densities in a laser are disturbed (by some means) from their steady state values then it is
important to know if these quantities return to their steady state values. If they do, the laser is stable. If
they don’t, the laser is unstable. Studying the recovery dynamics associated with such carrier density or
photon density perturbations tell a lot about the underlying laser physics.

We start from the laser rate equations,

dn i I
  Rnr n   Gnr n   Rr n   Gr n   v g g~ np
dt qVa
dn p  1 ~ n sp
 a v g g~   n p  a v g g
dt   p  Vp
and make the substitutions,
n  n  n t 
n p  n p  n p t 
The laser rate equations contain many nonlinear terms and we will linearize each of these terms around
their steady state values and keep terms to only first order in the perturbed quantities, n t  and n p t  .
This procedure yields,
dg~
v g g~ n p  v g g~ n p  vg n p nt   v g g~n p t 
ss dn ss
nt  n p t 
 v g g~ n p  
ss  st a p
Rnr n   Gnr n   Rr n   Gr n  Rnr n   Gnr n   Rr n   Gr n ss 
d
Rnr n   Gnr n  Rr n   Gr n  nt 
dn ss
nt 
 Rnr n   Gnr n   Rr n   Gr n  ss 
r
dg~
av g g~ n p  a v g g~ n p  av g n p nt   av g g~n p t 
ss dn ss
nt  n p t 
 av g g~ n p  a 
ss  st p
 Semiconductor Optoelectronics (Farhan Rana, Cornell University)

We have introduced two new quantities in the above equations; the differential stimulated emission time
 st and the differential recombination time  r , defined as follows,
1 d
 Rnr n  Gnr n  Rr n  Gr n
 r dn ss
1 ~
dg
 vg np
 st dn ss
The laser rate equations result in the following linear coupled differential equations for the perturbations,
  I I  1 
      
d  n t      r  st  a p   n t  
  
dt n p t   a  n p t 
 0 
  st 
We have ignored the perturbation in the spontaneous emission term in the laser rate equations since it is
much smaller than the perturbation in the stimulated emission term.

11.6.2 Relaxation Oscillations:

The coupled differential equations for the perturbations in the carrier and photons densities constitute a
second order linear system much like the equations for the current and the voltage in a LCR circuit. The
coupled equations give the following identical second order differential equations for the perturbations,
d 2 n p t  dn p t 
 2
 R n p t   0
2 dt
dt
d n t 
2
dn t 
  R2
n t   0
2 dt
dt
where the relaxation oscillation frequency  R and the damping constant  are defined as,
1 dg~ n p
R   vg
 st  p dn ss  p
I I
  
r  st

The solutions of the above second order equations have the form,
   2  2 
 t  2     2    
n t   e 2 A cos R    t  B sin R    t 
  2   2  
    
   2  2 
 t   
  2    
n p t    e
2 C cos R2
   t  D sin  R    t 
  2   2  
    
The constants A, B, C, and D can be chosen to satisfy the initial conditions. Suppose, n p t  0   0
then we get,
 Semiconductor Optoelectronics (Farhan Rana, Cornell University)

 nt  0
A  n t  0  B
2
 
2  R2  
2
a n t  0 
C 0 D
2
 
 st R2  
2
The Figure below shows the time evolution of the carrier density and photon density perturbations
vertical scale is normalized). The perturbations are damped and the steady state is stable because any
disturbances or perturbations decay with time. The decay is not monotonic but involves damped carrier
and photon density oscillations that are 90-degrees out of phase. These oscillations are called relaxation
oscillations. If the second order system is critically damped or over damped (i.e.,  R   2 ) then the
perturbations will decay monotonically without any relaxation oscillations.

np(t)

n(t)

11.7 Direct Current Modulation of Lasers

11.7.1 Introduction:
Consider the LI (light vs. current) curve of a semiconductor laser. Suppose the laser is biased with current
I and the corresponding output power is P .

Power

I I Current
th
Now suppose the current is varied such that,
I t   I  I t 
 Semiconductor Optoelectronics (Farhan Rana, Cornell University)

The output power can be written as,

P t   P  P t 
This current modulation is used in optical communication systems to transfer information from the
electrical domain to the optical domain. The question we need to answer here is how fast can a laser be
modulated? The answer can be obtained from the laser rate equations. We assume that,
n t   n  n t 
n p t   n p  n p t 
As in the previous Section, we linearize the laser rate equations and obtain,
  I I  1 
      
d  n t      r  st   a p   n t    i I t   1
   
dt n p t   a   n p t  qVa 0
 
 0 
  st 
We assume that the time varying part of the current, carrier density, the photon density, and the output
power are sinusoidal,

I t   I  Re I f e  i 2ft 

n t   n  Re n f e  i 2ft 

n p t   n p  Re n p f e  i 2ft


P t   P  Re P f e  i 2ft 
The solution is,
  i 2f 
 n f   I f   2 
n f    i  H f 
 p  qVa  R 
 a p 
 
The solution above relates the small signal carrier and photon densities to the small signal current. The
change in the output power is,

P f    o i H f  I f 
q
The modulation response function H f  is,
 R2
H f  
 2
R

 2f 2  i 2f 
Note that since H f  0   1 ,

P f  0    o i I f  0 
q
This low frequency result could also have be obtained directly from the relation,

P   o i I  I th 
q
Also note that, n f  0   0 . This result is to be expected since the carrier density in steady state above
threshold does not vary with current, and remains fixed at a value close to n th .

11.7.2 High Frequency Response of Lasers:

The relation,
 Semiconductor Optoelectronics (Farhan Rana, Cornell University)


P f    o i
H f  I f 
q
shows that the frequency response of the laser power to current modulation is governed by the function
H f  . In optical communication systems, a detector at the receiving end converts the modulated light
back into current. A schematic of a communication link is shown in the Figure below. Assuming the
frequency dependent detector responsivity to be Rf  , the RF current at the output of the link is related to
the RF current at the input to the link by the relation,

Iout f   R f P f   R f o i H f  I in f 
q
The ratio of the RF power at the output to the RF power at the input is called the link loss and its
expression is (assuming no optical losses in the fiber and no coupling losses),
Iout f 
2 2

 R f o i H f 
I in f 
2 q

The ratio is proportional to H f  .

2

Iout(f)
Iin(f)
P(f)
Load
I V
Laser Fiber Detector

An optical communication link using a direct

current modulated semiconductor laser

1 1
10 10

Increasing
current
|H(f)| 2

|H(f)| 2

0 0
10 10

-1
10 -2 -1 0 1 10 -2
-1
10 10 10 10 10 10
-1
10
0
10
1
10
2
2f/R 2f (rad/s)

The Figure above plots H f  as a function of frequency. The peak of H f 

2 2
occurs at a frequency f
equal to approximately  R 2 (provided  R   ). The peak is called the relaxation oscillation peak.
The frequency f3dB at which H f  decreases from its value at zero frequency by 3 dB (by a factor of 2)
2

is approximately equal to 1  2 R 2 (provided  R   ). f3dB is the maximum frequency at

which the laser power can be current modulated. The relaxation oscillation frequency  R 2 therefore
 Semiconductor Optoelectronics (Farhan Rana, Cornell University)

sets the scale for the maximum frequency at which a laser can be current modulated. The relaxation
oscillation frequency, given by the expression,
1 dg~ n p
R   vg
 st  p dn ss  p
increases as the square-root of the steady state photon density. Therefore, increasing the current will
increase the frequency f3dB . However, this trend does not continue to very high current levels. This is
because the damping constant, given by,
I I I dg~
     vg np
 r  st  r dn ss
increases linearly with the photon density (faster than the relaxation oscillation frequency). When the
laser current is slightly larger than the threshold current,  R   because 1  p  1  st ,1  r . When
the current is increased, f3dB also increases. As the current is increased to larger values, at some point the
relaxation oscillation frequency  R becomes equal to  2 . When this happens, the relaxation
oscillation peak in H f 
2
disappears and f3dB equals R 2  2 2 p . If the current is increased
beyond this point, the frequency f3dB decreases instead of increasing. The maximum value of f3dB is
therefore related to the inverse photon lifetime in the cavity,
2
f3dB max 
2 p
The inverse photon lifetime sets the upper limit on the modulation speed of semiconductor lasers.

11.8 Band Diagrams and Circuit Models

11.8.1 Band Diagram:
In a laser, one must have the following condition satisfied,
E fe  E fh  qV    E g
The band diagram is shown in the Figure below.

Efe

qV

E
fh

In the active region, the electron density n is a function of the Electron Fermi level, the hole density p is
a function of the hole Fermi level and quasineutrality implies,
n E fe   p E fh 
The above relation, together with the condition, E fe  E fh  qV , uniquely determines the carrier density
in the active region as a function of the voltage V across the junction.
 Semiconductor Optoelectronics (Farhan Rana, Cornell University)

11.8.2 Electrical Impedance of the Active Region:

Consider a laser operating in steady state. The impedance Z f  of the active relates the small signal
circuit current I f  to the small signal voltage V f  across the junction,
V f   I f  Z f 
A small change in the carrier density n f  can be related to a small change in the voltage V f  as
follows,
n
n f   V f 
V ss
The laser rate equations in Section 11.7 give,
 i 2f I f 
n f   H f   i
2 qVa

R
The above three Equations give,
V f   i 2f 1 1
Z f    H f   i
I f  R2 qVa n V ss
The impedance of the active region is proportional to the modulation response function H f  . At low
frequencies, when H f   1, the impedance of the active region is inductive and approaches zero as the
frequency approaches zero.

11.8.3 Total Electrical Impedance:

Consider the laser connected as shown below. The total laser impedance ZT f  consists of the active
region impedance Z f  in series with a resistor (representing the resistance of the top quasineutral region
and the top contact) and also the capacitance between the top metal contact and the substrate,
1
ZT f   Z f   R 
 i 2 f C

Iin(f)

I Laser
Laser

11.8.4 Circuit Parasitics and the High Frequency Current Modulation Response:
The current in the active region I f  is related to the current I in f  as,
I in f 
I f  
1  i 2 f C Z f   R 
The laser current modulation response is therefore more accurately given by the expression,
   1 
P f    o i H f  I f    o i H f    I in f 
q q  1  i 2 f C Z f   R  
At high frequencies, the capacitance can short out the active region and the decrease the laser current
modulation response. Careful attention must therefore be paid to circuit level parasitics when optimizing
lasers for high speed applications.
 Semiconductor Optoelectronics (Farhan Rana, Cornell University)

11.8.5 Laser Structures for High Speed Operation

Few laser structures for high speed operation are shown in the Figure below.

p-InGaAsP n-InP
metal
metal
p-InP SI-InP
p-InP
n-InP SI-InP
active p-InP

n-substrate n-substrate

Buried heterostructure laser Semi-insulating InP based

with regrown reversed biased mushroom structure (25 GHz)
junctions (10-15 GHz)
polyimide
substrate contact
n-InP
metal
metal
p-InP p-InP
SI-InP SI-InP
active
active
n-substrate
n-substrate
Co-planar strip line laser with mushroom Polyimide planarized laser
structure (31 GHz) (19 GHz)

11.8.6 Laser Packaging for Different Applications:

Fiber

Economical
packages
RF-in

Packaging for high speed fiber optic

communication applications

Wire bonding
pad

Top metal
contact

Unpackaged laser chip High power Laser package

 Semiconductor Optoelectronics (Farhan Rana, Cornell University)

11.9 Cavity Modes and Multimode Lasing

11.9.1 Cavity Modes:
Consider the Fabry-Perot cavity, shown below.

R1 R
2

z=0 z=L
The mode field propagating in the forward direction can be written as,

 
E ( x, y , z)  xˆE x ( x, y )  yˆE y ( x, y )  zˆE z ( x, y ) e iz
If the transverse dimensions of the waveguide are chosen to be small enough, the waveguide will support
only a single transverse mode. Suppose this mode is HE00. As discussed in earlier Chapters, for this
transverse mode different values of the propagation vector  correspond to different longitudinal modes
of the cavity. The spacing (in frequency) between adjacent longitudinal modes can be very small for long
optical cavities, as we will see now. Since the cavity is closed at both ends periodic boundary conditions
cannot be used to determine the density of modes. Suppose the complex field amplitude reflection
2 2
coefficients at the two facets are r1 and r2 , respectively, and R1  r1 and R 2  r 2 . We can write
the reflection coefficients in terms of an amplitude and phase,
r1  R1e i1 r 2  R 2 e i 2
For any cavity mode, the change in phase in one complete cavity roundtrip must be an integral multiple of
2 ,
2L  1  2  p 2  p  integer
For adjacent modes we have,
2L  1  2  2
Dividing and multiplying by  , the frequency spacing between adjacent modes, gives,
    
2 L  1  2    2
    

  
L 1    
  1 2
v g 2    
The reflection phases are usually weak functions of the frequency and for long cavities we have,
 c
 
L ng
The frequency spacing between adjacent cavity modes is called the free spectral range of the cavity. It is
more commonly expressed as wavelength spacing,
 Semiconductor Optoelectronics (Farhan Rana, Cornell University)

2
 
2ng L
For example, in a 500 m long, 1.55 m Fabry-Perot laser cavity with a modal group index of 3.5, the
wavelength spacing between adjacent cavity longitudinal modes is 0.69 nm.

11.9.2 Multimode Lasing:

Since the gain bandwidth of semiconductors is typically in the 10-50 nm range, there can be many modes
within the gain bandwidth. This situation is depicted in the Figure below.

Cavity modes

Consider a laser cavity below threshold. As the current is increased, the carrier density increases and,
consequently, the gain increases and at some value of the current the peak gain g~ will equal the threshold
gain g~ and the cavity modes near the gain peak will start to lase. If the current is increased further, the
th
power in the lasing modes will increase but the carrier density will remain at the value equal to nth .
Therefore, the gain spectrum will also remain fixed and independent of the current. The modes away from
the gain peak will never acquire enough gain to lase no matter how much the current is increased. The
lasing spectrum is therefore narrow but still several modes near the gain peak lase simultaneously. In a
typical Fabry-Perot laser the number of lasing modes can be anywhere from just a few to as many as ~50.
This multimode lasing behavior is not suitable for many laser applications, such as optical
communications and spectroscopy. In the following Chapters we will discuss strategies to realize single
frequency lasers.