In contrast to the silicon industry, where process-, device-, and interconnect-
simulation tools form a continuous virtual workbench from material analysis to
chip design, III-V simulation mainly is focused on device and circuit aspects.
The latter is accompanied by few examples for MESFET technology simulation
tools developed in parallel to SUPREM e.g. [44]. For heterojunction
devices, inclusively SiGe HBTs, due to the extensive number of process steps,
device simulation is focused on process control and inverse modeling e.g. of
geometry.
A common feature is the lack of a rigorous approach to III-V group
semiconductor materials modeling. As an example, modeling of AlGaAs, InGaAs, or
even InAlAs and InGaP is restricted to slight modifications of the GaAs
material properties. Another common drawback is the limited feedback from
technological state-of-the-art process development to simulator
development. Critical issues concerning simulation of heterostructures are
mostly not considered, such as interface modeling at heterojunctions and
insulator surfaces, as well as hydrodynamic and high field effects modeling -
carrier energy relaxation, impact ionization, gate current modeling,
self-heating effects, etc.
The two-dimensional device simulator PISCES [42], developed at the
Stanford University, incorporates modeling capabilities for GaAs and InP based
devices. One of its many modifications G-PISCES from Gateway Modeling
[44] has been extended by a full set of III-V models. Examples of MESFETs,
HEMTs, and HBTs for several material systems, e.g. InAlAs/InGaAs,
AlGaAs/InGaAs, AlGaAs/GaAs, and InGaP/GaAs HBTs are demonstrated. Disadvantage
of this simulator is the lack ET or HD transport model, necessary to model
high-field effects, in comparison to the original version of PISCES.
The device simulator MEDICI from Avant! [45], which is also based on
PISCES, offers simulation capabilities for SiGe/Si HBTs and AlGaAs/InGaAs/GaAs
HEMTs. Advantages of this simulator are HD simulation capabilities and the
rigorous approach to generation/recombination processes. In addition, recently
an option treating anisotropic properties was announced. Next to III-Vs
materials modeling this simulator has drawbacks in the interface modeling and
in the capabilities of mixed-mode device-circuit simulation. However, it has
been successfully used for the simulation of AlGaAs/GaAs HBTs [46].
The two- and three-dimensional device simulator DESSIS from ISE
[40] has demonstrated a rigorous approach to semiconductor physics
modeling. Some critical issues, as the above stated extensive trap modeling,
are solved. Recently, first steps in direction of interface and III-V modeling
have been reported [47].
Using a simplified one-dimensional current equation quasi-two-dimensional
approaches are demonstrated, formerly by the University of Leeds
e.g. [48]. This approach has also been verified for a number of
examples and for gate-lengths down to 50 nm [49]. It is available as a
submodule of Agilents Advanced Design System (ADS) delivering an interface to
a microwave circuit simulator. The emphasis is put on the extraction of compact
large-signal models. Examples of S-parameter simulations of AlGaAs/GaAs HEMTs
have been presented. This tool combines the advantages of a full HD transport
model combined with Schrödinger solution, but has the drawback of
the simplified one-dimensional current equations.
A similar quasi-two-dimensional tool is Fast Blaze from Silvaco, also based on
code from Leeds, which together with the two-dimensional ATLAS [50] has
claimed the simulation of AlGaAs/GaAs and pseudomorphic AlGaAs/InGaAs/GaAs
HEMTs. In addition, simulations of SiGe HBTs were announced, based on simulator
originally developed at the University of Illmenau, PROSA [51].
In the latter no materials interfaces are considered.
A drawback of most simulators, similarly to III-V modeling, is that the
modeling of SiGe is performed by slight modifications of the properties of
Silicon. However, several authors
made use of them e.g. [52,53,54]. A considerable advance in the SiGe HBT simulation was achieved with SCORPIO [55].