The world's dominant IC material, silicon, cannot do everything we want a semiconductor material to do.
However, for this discussion, the fact that Si wafers are of high quality, large and cheap is of great interest.
This is important for at least two reasons. First, nearly all of the electronic and photonic compound
semiconductor devices that comprise the current $20 billion per year market are fabricated on substrates
that are either very expensive or non-optimal for the epitaxy required to realize the device or an IC of
interest. A second reason is the integration of new functionality to current Si technology. Clearly, if many
of the current photonic applications already realized in current compound semiconductor technology could
be integrated into Si technology, some of the herculean efforts to continue following Moore's Law
(including trying to do it via nanotechnology) could be mitigated. This presentation examines some of the
basic materials science issues involved with heterogeneous integration of semiconductor materials. These
include those applications in which the active device region requires a high degree of crystal perfection and
those that do not. Epitaxy issues at the hetero-interface, heterovalent versus homovalent epigrowth, and
dislocation dynamics are presented. Notable historical examples are summarized, followed by examples of
current successful approaches including the materials science concepts used to achieve the results. A list is
made of some challenges that need to be solved in order to continue making future progress.
|