Î

REP5912 Advanced Theory of Solar Cells II September 2023

Lecture 9: Hetero-junctions
Lecturer: Dr P Dobreva Current state: continuation of notes

This is a brief non-exhaustive introduction to semiconductor-semiconductor heterojunctions.

ÏÄ
9.1 Basics
ˆ Homojunction—junction between different types of the same material— same band gap, like the p-
and n-doped silicon
ˆ Heterojunction—in this section of our course heterojunction will be understood to mean junction
between different semiconductors, semiconductors of different band gaps (gives rise to heterostructure
devices)
– can have double heterostructure device (two heterojunctions) or triple, etc.;
– common in multijunction solar cells (III-IV semiconductor solar cells), fig. 9.1;
ÍÀ
Figure 9.1: Source: https://pubs.rsc.org/en/content/articlehtml/2009/ee/b809257e

– crystal lattice mismatch and associated dangling bonds may arise, as shown in fig. 9.2

Figure 9.2: (a) Different lattices apart, (b) Dislocations at the interface of lattices, (c) strain at lattice
interface. Image source: the Sze textbook.

9-1
 9-2

Î
– lattice matching possible—see the relationship between lattice constant and bandgap for semi-
conductors in the diamond and zinc-blende crystal lattice configurations in fig. 9.3

ÏÄ
Figure 9.3: Band gap vs. lattice constant for III-V and II-VI compound semiconductors. The lines, con-
necting two binary compounds, represent a ternary compound in which the composition is varied, gradually
increasing the fraction of one element while decreasing the fraction of the other. For example, the line con-
necting ZnS to MgS, indicates a ternary compound M gx Zn1−x S; as the fraction of Mg gradually increases,
the fraction of Zn gradually decreases. Source: https://woodall.ece.ucdavis.edu/resources/

– at the junction—discontinuity in the conduction and valence band edges due to change in band
ÍÀ
gap. How do the bands align and bend?
– electrostatic properties and nature of the currents—different for electrons and holes;
– heterojunction benefits:
* greater part of the solar spectrum can be utilised—see the EQE graph fig. 9.4 for a four-
junction device;

Figure 9.4: EQE for four-junction solar cell.

 9-3

Î
* increase the light absorption properties;
* Voc may be increased—refer to the posted Fraunhofer paper and posted NREL paper (6
junctions)
→ Voc is determined by the split of the quasi-Fermi levels; upper limit of Voc is the Eg
Jph
→ Voc ≈ kT q ln J0 , J0 ∝ ni , ⇒ if ni decreases, then J0 decreases; large Eg ⇒ small ni

9.2 Formation, properties and carrier transport for heterojunc-

ÏÄ tions
ˆ Types of heterojunctions—determine how the bands will align when the junction is formed; before the
junction is formed, align the vacuum level
ˆ Heterojunctions need not be between differently doped semiconductors—can have n−n heterojunction,
p − p heterojunction (isotype heterojunctions) or p − n heterojunction (anisotype heterojunction)

Figure 9.5: Straddling gap—the band gap of one is completely contained in the band gap of the other
material—both, electrons and holes need energy to move from the smaller band gap to the larger band gap
material. Staggered gap—the band gaps overlap but one type of charge carrier needs energy to move from
ÍÀ
the smaller to the higher band gap material, while the other type of charge carrier loses excess energy when
crossing between the materials. Broken gap—the band gaps do not overlap at all and crossing of the charge
carriers from one to the other material is similar to the Staggered type only more pronounced.

Good source of semiconductor properties info is http://www.ioffe.ru/SVA/NSM/Semicond/

– the GaAs/AlGaAs heterojunction is straddling gap type; band gaps at 300 K: EgGaAs = 1.42 eV ,
EgAlGaAs = 2 eV
– the InP/InSb heterojunction is staggered gap type; band gaps at 300 K: EgInP = 1.27 eV , EgInSb =
0.17 eV
– the GaSb/InAs heterojunction is broken gap type; band gaps at 300 K: EgGaSb = 0.68 eV , EgInAs =
0.36 eV
ˆ The band diagrams of two semiconductors (differently doped) that will form straddling heterojunction
are shown in fig. 9.6 before they are joined.
ˆ Electron affinities, work functions and band gap considerations for band alignment—Anderson’s elec-
tron affinity rule (simplest sort of theory; not always working well)
ˆ In our example, fig. 9.6, the two semiconductors are assumed to have different band gaps, different
electron affinities, different doping and different work functions
– the larger band gap (Eg1 ) material is p-doped and the smaller band gap (Eg2 ) material is n-doped
 9-4

Î
ÏÄ
Figure 9.6: Band diagram—two different semiconductors before the junction is formed. The conduction
bands are offset by ∆Ec and the valence bands are offset by ∆Ev .
ÍÀ
Figure 9.7: Heterojunction. If the conduction bands are aligned (after the respective band bending), then the
valence bands can’t be aligned—discontinuity in the valence bands (indicated by the spike). Discontinuity
can occur also in the conduction bands, or even in both—conduction and valence bands. The discontinuity
has an effect on the charge transport mechanisms across the junction.
 9-5

Î
ÏÄ
ÍÀ
Figure 9.8: Electrostatics of the junction
 9-6

Î
– the electron affinities compare as follows, χ1 < χ2
– band offsets
* conduction band offset, ∆Ec (by Anderson rule, eq. (9.1))

∆Ec = χ2 − χ1 (9.1)

* valence band offset, ∆Ev ;

ÏÄ ∆Ev = Ev1 − Ev2 = (χ1 + Eg1 ) − (χ2 + Eg2 ) = Eg1 − Eg2 + χ1 − χ2 = ∆Eg − ∆Ec

∆Ev = ∆Eg − ∆Ec

– when the junction forms (at equilibrium)—equilibration of the energy per charge—Fermi levels
line up—formation of depletion regions fig. 9.7;
– built-in voltage, Vbi ; the built-in voltage essentially equals the difference in the Fermi levels before
the materials are joined; in our case,

The built-in voltage can also be given by

qVbi = |EF1 − EF2 |

qVbi = |Φ1 − Φ2 |

where Φ2 , Φ2 are the respective work functions.

– bending of the bands—determined by the depletion region—similar
in metal-semiconductor junction; since n = Nc exp − EckT −EF

to the bending reasoning
and the higher energy electrons
in material 2 move to material 1 (in our example) at the time of junction formation, then the
depletion region that forms at the junction has positive charge in mat. 2 and negative charge in
ÍÀ
mat. 1, as shown in fig. 9.7. Also, to indicate the depletion from electrons in mat.2, (Ec − EF )
has to increase, i.e. Ec bends up in mat. 2 while it
bends down in mat.1. Similar reasoning goes
−Ev
using the hole concentration p = Nv exp − EFkT and bending of the valence bands.
– ⇒ establishment of electric fields and potential differences (pot. barriers);
– The electrostatics of the junction—the fixed ions concentrations on eihter side of the junction,
the electric fields that are established on either side and the potentials are shown in fig. 9.8
Electrostatics at the junction

dE ρ(x) dV (x) d2 V (x) ρ(x)

= , E(x) = − , =−
dx ε dx dx2 ε

Solving Poisson’s equation for each side of the junction, we get E1 , V1 , E2 and V2

−qNA x qNA x2
E1 (x) = , V1 (x) = , −xp ≤ x < 0
ε1 2ε1

+qND x qND x2
E2 (x) = , V2 (x) = , 0 < x ≤ xn
ε2 2ε2
– the vacuum level follows the electrostatic potential

qVbi = qV1 + qV2 = |Φ2 − Φ1 |

 9-7

Î
– From the condition for continuity of the electric displacement vectors D1 = ε1 E1 and D2 = ε2 E2 ,
and the balance of the positive and negative charge in the depletion regions, qNA xp = qND xn ,
and the total built in voltage, the widths of the regions can be determined
s
2ε1 ε2 ND Vbi
xp =
qNA (ε1 NA + ε2 ND )

s
2ε1 ε2 NA Vbi

ÏÄ xn =
qND (ε1 NA + ε2 ND )

ˆ Under bias—similar behaviour to homojunctions, figs. 9.9 and 9.10.

Figure 9.9: Forward biased heterojunction of anisotype.

ÍÀ
Figure 9.10: Band diagram of forward biased heterojunction; the splitting of the Fermi level into quasi-Fermi
levels by the applied bias is shown.

The widths of the SCR on either side of the heterojunction shrink with the application of forward bias
Va and are obtained by solving the Poisson equation for either side of the junction.

s
2ε1 ε2 ND (Vbi − Va )
xp =
qNA (ε1 NA + ε2 ND )

s
2ε1 ε2 NA (Vbi − Va )
xn =
qND (ε1 NA + ε2 ND )
 9-8

Î
The difference is that the applied voltage Va is split between the two sides, Va = Va1 + Va2 , with each
side supporting only part of it; the relative voltage supported by each side is
V1 − Va1 ε2 ND
=
V2 − Va2 ε 1 NA
where V1 and V2 are the built-in voltages on either side of the heterojunction and Va1 and Va2 are the
parts of the applied voltage that each side supports.
ˆ Carrier transport—drift, diffusion, tunnelling might be possible. The current is obtained by solving
the continuity equation and applying suitable boundary conditions.

ÏÄ – electrons from n-region must overcome increased (by ∆Ec ) potential barrier; similarly, for holes
from the p-region; this might reduce the photocurrent
– the spike acts as a deterrent to the diffusion of holes from the n- to the p-region (potentially
reducing the saturation current); it also adds opportunity for tunnelling
– Simplification of the current transport can be obtained by assuming graded junction with ∆Ec
and ∆Ev becoming smooth transitions at the junction and the depletion regions. Similar to carrier
transport in homojunctions, the electron and hole diffusion current densities are

Jn =

Jp =

J = Jn + Jp =

qDn1 n2i1
Ln1 NA

qDp2 n2i2
L p 2 ND

Ln1 NA


+

exp

exp
qVa
kT

qVa
kT

Lp2 ND
−1


−1


Thanks to the simplifying assumptions of graded heterojunction, ∆Ec and ∆Ev are not present
in these expressions.
The total current is the sum
qDn1 n2i1 qDp2 n2i2

exp
qVa
kT
−1

ÍÀ
The ratio of the diffusion currents of each side is
E
Jn qDn1 n2i1 Lp2 ND g1
Dn1 Lp2 ND Nc1 Nv1 exp − kT ND ∆Eg
= = ≈ exp −
Jp qDp2 n2i2 Ln1 NA Eg 2
Dp2 Ln1 NA Nc2 Nv2 exp − kT N A kT
and decreases with increasing difference in the band gaps. Therefore, the diffusion across the
junction (which, recall from homojunctions, is the basis of the saturation current; the photogen-
earted current is drift current) decreases with increasing difference between the band gaps, ∆Eg ,
of the two materials. It is evident from this expression that the diffusion can be controlled also
by controlling the ratio of the doping concentrations.
ˆ If the lattices are not well matched, then energy states will be available within the band gap at the
interface and they will act as recombination centres. The interface states can affect the Fermi level
and introduce additional band bending themselves.

An illustration of the heterojunction of CdTe solar cell, figs. 9.11 and 9.12. The CdTe layer is p-doped and
the CdS layer is the n-doped emitter (CdTe doesn’t take readily n-doping). (The images are from the article
Fritsche J et. al., nterface Engineering of Chalcogenide Semiconductors in Thin Film Solar Cells: CdTe as
an Example, DOI:10.1557/PROC-668-H6.6)

9.3 Deposition processes

 9-9

Î
ÏÄ
Figure 9.11: Structure (layers) of a CdTe solar cell,
ÍÀ
Figure 9.12: Band diagram of CdTe solar cell.
 9-10

Î
ÏÄ Figure 9.13: Molecular Beam Epitaxy (MBE). Source: Wikipedia
ÍÀ
Figure 9.14: Metal-organic Chemical Vapour Deposition (MOCVD). Source: Wikipedia