ECE 475 - Lecture 14

March 4, 2025

1 Heterojunctions

• A homojunction is a pn junction between two differently doped regions of the same semiconductor

– The bandgap does not change along the device

• A heterojunction is a pn junction between two different semiconductors, each with its own bandgap,

such that:

∆Eg = Eg,2 − Eg,1 = ∆EC + ∆EV

where ∆EC and ∆EV are the conduction and valence band offsets, respectively

• There are many different tpyes of possible heterojunctions, including ”Type I” and ”Type II”

alignments

Figure 1: Source: Kasap

– Type I is a straddled bandgap alignment heterojunction

1
 – Type II is a staggered lineup heterojunction

• By convention, capital letters (N or P) are used to indicate the doping of the wider bandgap

• Type I is the most common in optoelectronics, e.g. GaAs/Alx Ga1−x As, though some optoelectronic

systems are Type II

• Both Np and pP heterojunctions are widely used for LEDs and laser diodes

• The band edge profiles are a given heterostructure are determined by the doping levels, carrier

transport, and recombination around the junction, just like in a simple homojunction

– Once contact is made, the EF must be uniform through the device at equilibrium

Figure 2: Source: Kasap

• Np Type I Junction

– The potential barrier for hole injection, EV,2 − EV,1 from p to N is greater than the barrier

for electron injection from N to p EC,2 − EC,1

– Forward bias current is dominated by injection of electrons from N to p

– Holes are ”confined” to p by the offset ∆EV , therefore recombination is ”confined” to the

p-region with hν = Eg,2

• pP Type I Junction

– The potential barrier from EC,1 to EC,2 increases, preventing electrons in the CB of the p

side cannot overcome this barrier

2
 – If electrons are injected (from the right) into the p-region, they are confined by ∆EC to

stay/recombine there

– Current flows ”easily” via conduction of holes

2 Heterostructure High Intensity LEDs

• Most modern optoelectronic devices (including LEDs) employ heterojunctions

• By constructing LEDs from heterojunctions, a dielectric waveguide can be engineered within the

device to channel photons out of the recombination region

• For example, a basic double-heterostructure LED

– The narrow band-gap ”active region” (GaAs in the above figure) is typically ∼ 1 µm thick

and lightly doped

– The DH-LED offers numerous advantages compared to a basic homojunction LED

• Carrier confinement to the active region

3
 – Potential energy ”offsets” confine carriers, e.g. e− injected into the active region are inhibited

by ∆EC from flowing into P-AlGaAs region (above figure)

– Confinement of electrons and holes inside the narrow bandgap active layer increases ηIQE

∗ Enhances radiative recombination (R ∝ np), i.e. reduces τr

∗ Emitted photons have a lower chance of reabsorption because the active region is thin,

typically smaller than the absorption length, δ = 1/α

• The wider bandgap acts as a transparent ”window”

– Emitted photons mostly do not cause intraband absorption in the N+ AlGaAs surface layer

(figure above)

• Reduction of non-radiative recombination at surfaces

– In the homojunction LED, the top p-layer must be thin to minimize re-absorption. e− and

h+ can diffuse to the top surface and recombine non-radiatively via ”surface states”

• Waveguiding (edge-emitting LEDs)

– Fortuitously, lower Eg is almost always correlated with higher refractive index

– The implies the presence of an optical waveguide, which can be exploited to ”channel” inter-

nally emitted light to the edge of the chip, with relatively high ηEE

• Edge-emitting DH LEDs have been widely used for short-hand, relatively low bit-rate fiber com-

munications, e.g. < 100 Mbps over a few km

• This can also happen in surface-emitting diodes

4
 2.1 Notes

• Edge-emitting LEDs typically provide a higher intensity, and more directional beam, compared to

surface-emitting LEDs

– More power can be coupled into a fiber

– Surface emitting resonant-cavity LEDs are becoming more popular, as they offer similar ad-

vantages

3 Quantum Well High Intensity LEDs

• A quantum well can be considered as a heterostructure with an ultra-thin, < 50 nm, narrow

bandgap layer, Eg,1 , between two wider bandgap semiconductors (Eg,2 )

• Electrons and holes will be confined to the narrow bandgap layer, i.e. within the quantum well, by

band-gap offsets (potential energy steps), ∆EC and ∆EV

• This confinement to a small dimension (along x) means that we can treat an electron within the

well as confined in x, but free in yz

• The allowed energy levels of this electron confined to a small space in x are defined by the

Schrödinger Equation

3.1 Two main effects occur as follows

• Quantum confinement alters the E − k relationship and the allowed energy levels

5
– Using an ”infinite well” approximation, i.e. let ∆EC /∆EV → ∞ The potential experienced

by the electrons within the quantum well are:



∞

 x≤0



V (x) = 0 0≤x≤d





∞ 100 ≤ x


– Therefore the 1D Schrödinger Equation for the electron within the well is

ℏ2 d2 ψ(x)
− = Eψ(x)
2m dx2

that has solutions ψ(x) = A sin(kx) + B cos(kx).

We can apply the boundary conditions that ψ(0) = ψ(d) = 0. When x = 0, sin(0) = 0 and

cos(0) = 1; therefore, B must be equal to zero giving

ψ(x) = A sin(kx)

To solve for k, we differentiate the wavefunction with respect to x:

d2 ψ(x)
= −k 2 A sin(kx)
dx2
ℏ2
− (−k 2 A sin(kx)) = EA sin(kx)
2m r
8π 2 mE
k=
h"2 #
r
8π 2 mE
ψ(x) = A sin x
h2

When x = d, ψ(d) = 0

"r #
8π 2 mE
0 = A sin d
h2

6
 which can only be true when

r
8π 2 mE
d = qπ
h2

where q = 1, 2, 3, .... The quantized energy levels of the quantum well are then simply

q 2 h2
Eqn =
8me d2

– For electrons

ℏ2 ky2 ℏ2 kz2
E = EC + Eqn + +
2m∗e 2m∗e
| {z }
Kinetic energy associated
with ”free” electron
motion in y − z plane

where

q 2 h2
Eqn =
8m∗e d2

and d is the confinement length [m]

7
 – For holes

ℏ2 ky2 ℏ2 kz2
E = EV − Eqn − −
2m∗h 2m∗h

where

q 2 h2
Eqp =
8m∗h d2

– The minimum emission frequency is correspondingly modified

”Bulk” (Thick) Active Layer Quantum Well (Thin) Active Layer

hνmin = Eg hνmin = Eg + E1n + E1p

• Confinement alters the density of states function

– For example, for electrons 

∗
 qme


πℏ2 d E > EC + Eqn
g(E) =

0
 E < EC + Eqn

where q = 1, 2, 3, ... g(E) is a ”staircase” function

• Quantization greatly modifies the rate (probability) of transitions at the ”emission onset” frequency,

i.e. ν = νmin

”Bulk” (Thick) Active Layer Quantum Well (Thin) Active Layer

No quantum confinement Quantum confinement

m∗
g(EC ) = 0 g(EC + E1n ) = e
πℏ2 d ̸= 0

m∗
g(EV ) = 0 g(EV − E1n ) = h
πℏ2 d ̸= 0

Nothing Significantly impacts τr and improves ηIQE

• Single quantum wells (SQWs) have limited efficiency, especially at high currents

– Only a limited number of carriers can be confined to this well simultaneously before they short

to ”overflow”

• Multiple quantum level devices partly alleviate these problems and enable higher light output

8
 Notes

• In MQW LEDs, the wells are ”mutually coupled” so that energy quantization and density of states

modifications are intermediate to the ”bulk” and the ”SQW” active region cases

• Modern high-intensity UV/violet/blue, LEDs are typically based on Inx Ga1−x N (Eg,1 )/GaN (Eg,2 )

MQWs, where

Eg,2 = 3.4 eV

Eg,1 = 2.7 eV