Information Technology Reference
In-Depth Information
11.1. INTRODUCTION
Silicon metal-oxide-semiconductor field-effect transistors (MOSFETs), the build-
ing elements of modern computing, have already entered the nanometer era.
Nonstandard MOSFET and non-MOSFET nanometer devices have received
intense research and are predicted to dominate the landscape of computing in
10 years. Molecular computing seeks to build computational systems wherein
individual or small collections of molecules serve as discrete device components or
play a significant role in them. Physicists and chemists have bravely ventured into
measuring and modeling electronic properties of molecules as well as synthesis and
assembly of molecules of desirable electronic properties. Typical molecules used in
these molecular switches are orders of magnitudes smaller than the state-of-the-art
silicon MOSFET. Such an advantage in size can potentially lead to computing
systems with much higher density and performance yet lower power and lower
cost than what promised by silicon MOS technologies. Inspired by the vision of
molecular computing, computer engineering researchers have explored its circuit,
logic, architecture, and even system-level design.
While molecules are approximately one million times smaller than their present
day solid-state counterparts, this small size brings with it a new set of challenges.
First, although molecules can be synthesized in large quantities relatively easily,
they can be difficult to arrange on a surface or in a three-dimensional array such
that each molecule is addressable. It is equally difficult to ensure that every
molecule stays in place. Second, although individual molecules have been shown to
be switching, it is very difficult, if possible at all, to interconnect them or selectively
interface them with microscale 1 input/output. This is particularly true for three
terminal molecules because it is far more difficult to bring three probes into close
proximity than to bring two probes into near contact.
These challenges invite a rethinking of computing design and implementation.
Present day computing is based on a top-down process in which predesigned
patterns are fabricated exactly as specified and small features, such as transistors,
are precisely etched into silicon using resists (chemicals) and light (photolitho-
graphy). Such a what-you-design-is-what-you-get is no longer true in the nan-
ometer domain. Uncertainty in conventional silicon MOSFET fabrication has
already made it impossible to fabricate precisely as specified, leading to great
variations in performance and power characteristics in circuits of the same design
and fabrication process. Thus various design methodologies have been introduced
to keep the top-down process alive in silicon MOSFET-based computing.
Unfortunately, the top-down process alone is unlikely to work for molecular
computing because it is extremely difficult to position and interconnect molecules,
as highlighted above. The fabrication of molecular switches is essentially a
bottom-up process, i.e., self-assembly of molecules into higher ordered structural
units. Self-assembled structures form the basis of
the first generation test
1 We use microscale to refer to what can be patterned using the state-of-the-art photolithography. We
use nanoscale to refer to things that are too small for the state-of-the-art photolithography.
 
Search WWH ::




Custom Search