Heterojunctions

Modulation-doped heterojunctions
• They are the interfaces between two
semiconductors of different gaps, and are one of
the most versatile building blocks of electronic
devices, especially those based on III-V
compounds
• Most studied heterostructure is n-type
AlxGa1−xAs and almost intrinsic or lightly doped
p-type GaAs hetrojunction.
• Similar to the MOS structure, an inversion layer
of electrons is formed in the GaAs close to the
GaAs–AlGaAs interface.
• Therefore the principle be very similar to MOS
structure
• Mainly used in higher frequencies than Si devices
due to the high mobility of electrons in GaAs.
• Most important device applications are based on a
Schottky structure of the type metal–AlGaAs–
GaAs
• The band diagram of the AlxGa1−xAs–GaAs
heterojunction
• We have an AlGaAs–GaAs heterojunction, where
the left material is GaAs doped with Al and the
right one is near-intrinsic GaAs. This structure is
called a modulation-doped heterojunction and
the method to produce it is known as
modulation doping
• Before the two semiconductors enter in contact,
the Fermi-level in n-type AlGaAs is close to the
conduction band, and for lightly p-doped GaAs is
located close to the middle of the gap
• The bands are flat because the materials are
electrically neutral and have uniform doping
• The barrier between the material in the
conduction band, ∆Ec, can be approximately
found following Anderson’s rule.
• According to this rule, when we join two
materials, the vacuum levels should line up. If
χA and χB are the electron affinities of the
AlGaAs and GaAs, respectively, we should have
∆Ec ≡ χA − χB
• Since the electron affinity of a semiconductor
is defined as the energy required for an
electron located at the bottom Ec of the
conduction band, to get out of the solid, i.e.
χ = Evac– Ec.
• According to this rule, one gets a value of ∆Ec of
0.35 eV for a doping x in AlxGa1−xAs around 0.3.
• When both materials in contact, some of the
electrons from the donors of the n-material will
cross the interface reaching the undoped GaAs.
• Similar to the p–n junction, an internal electric
field will be created and directed from the non-
neutralized donors in the AlGaAs to the
additional electronic charges in the GaAs.
• This field causes the band bending
• At equilibrium, the two Fermi levels line up,
the bands are bent like in the case of the p–n
junction, with the only difference that the
barrier ∆Ec is created.
• Far from the interface, the bottom Ec of the
conduction bands is flat and at the same
distance from the Fermi level EF
• The quantum well for the electrons produced at
the AlGaAs–GaAs interface has a shape close to a
triangle similar to MOS structure
• If we call z the direction perpendicular to the
interface, the electrons forming the 2D inversion
layer are free to move along the (x, y) plane, but
their energy for the motion along z is quantized
as in a potential well.
• An important aspect of this heterojunction is that
the charge carriers are located in a region
(mainly in the GaAs), spatially separated from
the AlGaAs semiconductor which originates the
free electrons
• For understanding, solution by Poisson’s
equation for the potential and Schrödinger’s
equation for the electron wave functions, but
they are complicated.
• So the problem has been solved, usually
taking some approximations, such as
assuming that the potential well is perfectly
triangular in shape
• The wells cannot be assumed to be of infinite
height, since Ec ≈ 0.3 eV. Detailed calculations
also allow calculation of the average width of
the well (40–80Å), the electron concentration
per unit area, ns ≈ 1012cm−2, and the energy ε1
of the first level ≈ 0.04 eV
 SiGe strained heterostructures
• At first SiGe heterojunctions did not attract
too much attention, because of the large
lattice constant difference between Si and Ge
which amounts to about 4% (≈ 0.2 Å)
• This means that a critical thickness should not
be surpassed otherwise the structure breaks
off.
• Also the energy gaps of silicon (Eg = 1.12 eV)
and germanium (Eg = 0.66 eV) are fairly small,
then the height of the barriers which appear
at the interface should always be small.
• But SiGe heterostructures are used in several
fields such as high frequency transistors and IR
photodetectors. Since SiGe are strained, the
degeneracy of the heavy and light hole bands is
lifted and the band structures show similar
features to those shown below
• Two typical examples of SiGe heterostructures
• First case, the substrate is <001> Si (Eg = 1.17 eV)
and the strained active layer Si0.7Ge0.3
(Eg = 0.78 eV).
• In this case the conduction band offset is smaller
comparing to the valence band offset. This
situation allows the formation of a 2D hole gas in
the SiGe alloy, with electron mobilities around
2m2V−1s−1. It is about half of that of a typical
MOSFET
• In second case, the situation is reversed and
the strained layer is Si. Here the discontinuity
in the conduction band is fairly large and the
electrons form a 2D gas, with free motion in
the plane of the interface.
• The silicon effective mass corresponding to
this motion is the low transversal one
(mT∗ ≈ 0.19m0), therefore yielding a high
mobility of around 20m2V−1s−1, which is
several times higher than that of the MOSFET.
• SiGe heterostructures have also found an
important application in the field of bipolar
silicon transistors
• One way to improve the efficiency of a bipolar
transistor is to use a narrow-bandgap material for
the base region, which improves the efficiency of
the Si emitter region.
• This can be achieved by reducing values of the
SiGe bandgap alloys, in comparison to silicon.
• The strain that appears in the heterojunction also
contributes to the decrease of the bandgap. The
large bandgap offset allows the fabrication of a
highly doped, low resistivity, base material, which
extends the performance of silicon transistors to
much higher frequencies.