III-V heterostructure-based transistors and integrated circuits

The behavior of semiconductor electronic components is largely conditioned by the nature of the interfaces or junctions separating their various constituent parts (metallic, dielectric or semiconductor) and by the way in which carriers, electrons or holes, pass along or through these interfaces. The semiconductor/semiconductor junctions used in components and circuits based on field-effect transistors or bipolar transistors, which dominated microelectronics until the late 1980s, have long been homojunctions separating two regions of different doping on the same semiconductor host. In practice, the host semiconductor is generally silicon (NMOS, bipolar CMOS and BiCMOS series) and very rarely gallium arsenide (GaAs MESFET series).

During the 1980s, steady progress in materials development, manufacturing technology and the physics of complex semiconductor structures led to the emergence of a new generation of microelectronic devices known as heterojunction devices. Heterojunctions are junctions where two different semiconductors are juxtaposed. They are most often in crystal lattice agreement or near-crystal lattice agreement (as in the GaAlAs/GaAs junctions of heterojunction field-effect transistors, known as HEMTs, and heterojunction bipolar transistors, known as HBTs (§ 2.1 and 4.1)). But these heterojunctions can also be in slight mesh detuning (of the order of 1%, as in the case of the GaAlN/GaN junctions of GaN HEMTs (§ 3.1) or Ga _0.8 In _0.2 As/GaAs junctions of so-called pseudomorphic GaAs HEMTs (§ 2.4.1), or even in greater detuning, beyond what the limited elasticity of the crystal lattice can withstand, and thus involving highly dislocated crystal zones [case of the so-called metamorphic InP/GaAs structures (§ 2.4.3) and gallium nitride-based heterostructures (§ 3.1), where differences in crystal mesh size are of the order of a few percent between the starting substrate and the material that forms the active core of the component].

In all cases, the ability to combine semiconductors with different energy band structures within the same component provides additional degrees of freedom for the development of new components with enhanced performance or original functionality. Indeed, in addition to the applied fields and doping gradients that control electron and hole transport in ordinary semiconductor homostructure components, the ability to vary the bandgap energy in the case of a heterojunction enables abrupt spatial variations in potentials and fields [cf. the HEMT potential well used to separate carriers and donors (§ 2.1)]. What's more, these variations can be different for electrons and holes, introducing a kind of filtering into the transport of these two types...