6.

772/SMA5111 - Compound Semiconductors

Lecture 20 - Laser Diodes 1 - Outline
• Stimulated emission and optical gain
Absorption, spontaneous emission, stimulated emission
Threshold for optical gain
• Laser diode basics

Lasing and conditions at threshold

Threshold current density

Differential quantum efficiency

• In-plane laser cavity design (as far as we get; to be cont. in Lect. 21)
Vertical structure: homojunction
double heterojunction
quantum well, wire, dot; quantum cascade
Lateral definition: stripe contact
buried heterostructure
shallow rib
End-mirror design: cleaved facet
etched facet
distributed feedback, Bragg reflector

C. G. Fonstad, 4/03 Lecture 20 - Slide 1

 Laser diodes: comparing LEDs and laser diodes

Light emitting diodes vs. Laser diodes

LEDs are based on spontaneous emission, and have

1. A broad output beam that is hard to capture and focus

2. A relatively broad spectral profile

3. Low to moderate overall efficiency
4. Moderate to high speed (≈ 1/tmin)
Laser Diodes are based on stimulated emission, and
have the opposite characteristics
1. Narrow, highly directed output
2. Sharp, narrow emission spectrum
3. High differential and overall efficiency

4. High to very high speed

Stimulated emission occurs when a passing photon triggers

the recombination of an electron and hole, with emission
of a second photon with the same frequency (energy),
momentum, and phase.
C. G. Fonstad, 4/03 Lecture 20 - Slide 2

 Laser diodes: achieving stimulated gain

To understand what is necessary to obtain net optical

gain, rather than net absorption, we consider optical
transitions between two levels in a solid (E1 and E2),
and we look at three transitions ocurring with the
absorption or emission of photons:
1. from E1 to E2 due to absorption
2. from E2 to E1 due to spontaneous emission
3. from E2 to E1 due to stimulated emission

We model the rate of each process using the Einstein A

and B coefficients, and then find when the probability
is higher that a photon passing will stimulate emission
than be absorbed.

C. G. Fonstad, 4/03 Lecture 20 - Slide 3

 Laser diodes: achieving stimulated gain, cont.

In a semiconductor we consider one state, E1, to be in the

valence band, and the other, E2 to be in the conduction
band:
E2 - - -
Ec

hn
hn hn hn

E1 + + +
Ev
Absorption Spontaneous Stimulated
emission emission

The rates these processes occur depend on the populations:

Absorption rate, Rab: photon pop. x E1 pop. x E2 empty state pop.

Spontaneous emission rate, Rsp: E2 pop. x E1 empty state pop.
Stimulate emission rate, Rst: E2 pop. x E1 empty state pop. x photon pop.
C. G. Fonstad, 4/03 Lecture 20 - Slide 4
 Laser diodes: achieving stimulated gain, cont

Absorption rate:
Rab = B12 ⋅ f1 ⋅ N v (E1 ) ⋅ (1- f 2 ) ⋅ N c (E 2 ) ⋅ r p ( E 2 - E1 )
where
B12: transition probability for absorption
N v: valence band density of states at E1
N c: conduction band density of states at E2
rp(E2-E1): density of photons with correct energy
fi: Fermi function evaluated at Ei

where

(
fi = 1 e
E i -E fi
)
+1
Efi: quasi-Fermi level for level i

Spontaneous emission rate:

Rsp = A21 ⋅ f 2 ⋅ N c (E 2 ) ⋅ (1- f1 ) ⋅ N v (E1 )

C. G. Fonstad, 4/03 Lecture 20 - Slide 5

 Laser diodes: achieving stimulated gain, cont

In the last equation we introduced:

A21: transition probability for spontaneous emission

Stimulated emission rate:

Rst = B21 ⋅ f 2 ⋅ N c (E 2 ) ⋅ (1- f1 ) ⋅ N v (E1 ) ⋅ r p ( E 2 - E1)
where
B21: transition probability for stimulated emission
Note, finally, that in these expressions the Fermi function is evaluated
either in the conduction band (i = 2) or valence band (i =1):

(
f1 = 1 e
E1 -E fv
)
+1 , (
f2 = 1 e
E 2 -E fc
)
+1
****************

The coefficients, A21, B12, and B21, are related, as we can

see by looking at thermal equilibrium, where

8p ro3 2 1
Rab = Rsp + Rst , E fv = E fc , r p (E i ) = 3 3 E i E / kT
hc (e -1) i

C. G. Fonstad, 4/03 Lecture 20 - Slide 6

 Laser diodes: achieving stimulated gain, cont

Proceeding in this we we find:

8p ro3 E i2
B12 = B21, and A21 = 3 3
B21
hc
**************

Now we are ready to find the condition for optical gain,

which we take as when the probability of stimulated

emission is greater than that for absorption. Looking

back at our equations, we find Rst > Rab leads to:

B21 ⋅ f 2 N c ⋅ (1- f1 ) N v ⋅ r p ( E 2 - E1 ) > B12 ⋅ f1N v ⋅ (1- f 2 ) N c ⋅ r p ( E 2 - E1 )

Canceling equivalent terms yields:
f 2 (1- f1 ) > f1 (1- f 2 )
and substituting the appropriate Fermi functions gives
us:
E fc - E fv > ( E 2 - E1 ) = hu ≥ E g

C. G. Fonstad, 4/03 Lecture 20 - Slide 7

 Laser diodes: achieving stimulated gain, cont.

Our conclusion is that we will have net optical gain, i.e.,

more stimulated emission than absorption, when we

have the quasi-Fermi levels separated by more than

the band gap. This in turn requires high doping and

current levels. It is the equivalent of population

inversion in a semiconductor: E - E > E

fc fv g
**************
Next we relate the absorption coefficient, a, to Rab, Rst,
and Rsp. A bit of thought shows us that we can say:
Rab (E ) > [ Rst (E ) + Rsp (E )] ª Rst (E ) Æ a (E ) > 0 Net loss
Rab (E ) < [ Rst (E ) + Rsp (E )] ª Rst (E ) Æ a (E ) < 0 Net gain
Rab (E ) = [ Rst (E ) + Rsp (E )] ª Rst (E ) Æ a (E ) = 0 Æ E = E fc - E fv
Notes: Spontaneous emission is negligible because it is randomly

directed. It starts the lasing process, but it does not sustain it.

The point at which a = 0 is called the transparency point.

C. G. Fonstad, 4/03 Lecture 20 - Slide 8

 Laser diodes: optical gain coefficient, g(E)

The negative of the absorption coefficient is defined as

the gain coefficient: g(E ) ≡ - a (E )

Writing the light intensity in terms of g(E) we have:

L(E, x) = Lo (E )e-a ( E )x = Lo (E )e g( E )x
**************
Stimulated recombination is proportional to the carrier
populations, and in a semiconductor one carrier is
usually in the minority and its population is the one
that changes significantly with increasing current
injection. If we assume p-type material, we have:
g > 0 Æ n > n tr
To first order, the gain will be proportional to this

population, to the extent that it exceeds the

transparency level:

g @ G(n - n tr )
C. G. Fonstad, 4/03 Lecture 20 - Slide 9

 Laser diodes: threshold current

We not look at a laser diode and calculating the threshold

current for lasing, and the light-current relationship

Consider the following cavity:

Lasing will be sustained when the optical gain exceeds the

optical losses for a round-trip in the cavity.
The threshold current is the current level above which
this occurs.
C. G. Fonstad, 4/03 Lecture 19 - Slide 18
 Laser diodes: threshold current, cont.

Track the light intensity on a full circuit, beginning with Io

just inside the facet at x = 0+, and directed to the right:
1. At x = 0 + , directed to the right, I(0 + ) = Io
2. At x = L-, directed to the right, I(L- ) = Ioe( g-a l ) L
3. At x = L-, directed to the left, I(L- ) = R 2 Ioe( g -a l ) L
4. At x = 0 + , directed to the left, I(0 + ) = R2 Ioe( g -a l ) 2L
5. At x = 0 + , directed to the right, I(0 + ) = R1 R2 Ioe( g-a l ) 2L

For sustained lasing we must have the intensity after a

full circuit (5) be equal to, or greater than, the initial

intensity (1):
R1 R2 Ioe( g-a ) 2L
≥ Io l

C. G. Fonstad, 4/03 Lecture 20 - Slide 11

 Laser diodes: threshold current, cont.

This leads us to identify the threshold gain, gth:

1 Ê 1 ˆ
gth ≡ a l + ln Á ˜
2L Ë R1 R2 ¯
To relate this threshold gain to current we recall that the
gain is proportional to the carrier population in excess of
the transparency value,
g ª G( n - n tr ) = G'G( n - n tr )
where G': the portion of G due to material parameters alone

G: the portion of G due to geometrical factors (i.e., the

overlap of the optical mode and the active medium)
and that the population will in general be proportional

to the current:

n ª KiD
where K: a proportionality factor that depends on the device
structure, which we will determine in specific
situations later
C. G. Fonstad, 4/03 Lecture 20 - Slide 12

 Laser diodes: threshold current, cont.

Writing g in terms of iD, and setting it equal to gth, yields:

1 Ê 1 ˆ
gth = G' G (KiD - n tr ) = al + ln Á ˜
2L Ë R1 R2 ¯
The diode current that corresponds to this threshold gain
is defined to be the threshold current, Ith:
1 Ê 1 È 1 Ê 1 ˆ˘ ˆ

iD @gth ≡ Ith = ÁÁ Ía l + ln Á ˜˙ + n tr ˜˜
K Ë G'G Î 2L Ë R1 R2 ¯˚ ¯

(This will take on more meaning as we look at specific laser

diode geometries and quantify the various parameters.)
*************

A final useful observation is that the mirror reflectivity

term in these equations can be viewed as a mirror loss
coefficient, am: 1 Ê 1 ˆ
am ≡ ln Á ˜
2L Ë R1 R2 ¯
C. G. Fonstad, 4/03 Lecture 20 - Slide 13

 Laser diodes: threshold current, cont.

Above threshold, essentially all of the additional excitation

fuels stimulated recombination, and n' stays fixed at its
threshold value. So too does Efn - Efp, which implies
that the junction voltage is also pinned.
The current-voltage and power-current characteristics of
a laser diode thus have the following forms:
iD Popt

µ 1/Rseries µ hed

Ith
vAB Ith iD
current-voltage power-current
C. G. Fonstad, 4/03 Lecture 20 - Slide 14

 Laser diodes: output power, Popt, and external differential

quantum efficiency, hed

To calculate the optical output power, Popt, we begin with

several points:
First, we recall that a particle flux can be written in terms
of a particle density times their velocity. This holds for
photons as well, and the velocity is the mode, or "group"
velocity:
Fph (x, y,z) = v g N ph ( x, y,z)
Second, we recall that the rate of change of a photon pop-
ulation with time at a given point, is the photon flux
times the absorption coefficient, a:
∂N ph (x, y, z)
= - a v g N ph (x, y, z)
∂t
Finally, we note that the output power will be the flux of

photons emitted times the energy per photon, hn.

Pout = hn Fph,out,tot
C. G. Fonstad, 4/03 Lecture 20 - Slide 15

 Laser diodes: Popt and hed, cont.

We next move inside the laser diode and look at the

photon population there. The total number of photons

inside the laser will be

N ph,in,tot = Ú N ph,in (x, y, z) dx dy dz

This photon population is decreasing because photons are
being absorbed internally and emitted from the ends of
the cavity at a rate given by the photon flux times the
effective absorption coefficient:
Loss : Ú (a l + a m ) v g N ph,in (x, y, z) dx dy dz = (a l + a m ) v g N ph,in,tot
Note that the loss out the end mirrors, am, is the output we are looking to calculate!
and it is increasing because the diode current exceeds
the threshold current, and the gain exceeds the thresh-
hold gain:
Photon generation : Ú g v g N ph,in (x, y, z) dx dy dz =
(iD - I th )
hi
q
C. G. Fonstad, 4/03 Lecture 20 - Slide 16

 Laser diodes: Popt and hed, cont.

In the last equation, hi, is the current utilization efficiency,
the fraction feeding stimulated recombination.
In the steady state the loss equals the generation:
( iD - I th )
(a l + a m ) v g N ph,in,tot =
hi
q
Thus the total photon population inside the laser diode is:

N ph,in,tot =
(iD - I th ) h
q (a l + a m ) v g
i

The total photon flux emitted from the laser diode output
mirrors is the portion of the "loss" due to am:

Popt = hn a m v g N ph,in,tot = hn
(iD - I th ) am
hi
q (a l + a m )

C. G. Fonstad, 4/03 Lecture 20 - Slide 17

 Laser diodes: Popt and hed, cont.

We next introduce the extraction efficiency, he:

am
he ≡
(a l + a m )
With this we write the output power as:

Popt = hn
(iD - I th )
he hi
q
Note that this is the total output power from both ends of
the laser.
If the two end-faces have equal reflectivities, then half the power will
come out each end.
If one end is much more highly reflecting than the other, then little
power will come out it, and all of the power will come out the lower
reflectivity facet.
If the reflectivities of the ends differ, but not by a large amount, then
the division of output is complicated to calculate.
Next we turn to the external differential quantum efficiency.

C. G. Fonstad, 4/03 Lecture 20 - Slide 18

 Laser diodes: Popt and hed, cont.

The external differental quantum efficiency is defined as
the ratio between the number of photons emitted per unit
time, divided by the number of carriers crossing the diode
junction per unit time:
D ( # of photons out/unit time)
hed =
D (# of carriers across junction/unit time)
In terms of the output power and diode current this is:

D (Popt / hn ) q dPopt
hed = =
D (iD /q) hn diD

Using the result on the previous foil, we find:

hed = he hi

C. G. Fonstad, 4/03 Lecture 20 - Slide 19

 Laser diodes: device design and optimization

With this general background, we now go to the white

board and turn to looking at specific device design and

the evolution of diode laser structures over time.

• We will begin looking at evolution of active region design

and the vertical device structure.

• Next we turn to lateral definition of the cavity.

• Finally we look at defining the ends of the cavity.

• This initial discussion will focus on in-plane lasers. After

this we will turn to vertical cavity devices.

• At the very end we will discuss laser diode modulation.

C. G. Fonstad, 4/03 Lecture 20 - Slide 20