MOLECULAR COMPUTING: INTEGRATION OF MOLECULES FOR NANOCOMPUTING - Bio-Inspired and Nanoscale Integrated Computing - page 297

molecular devices that have been made to date. However, the bottom-up process
alone is unlikely to produce large-scale, integrated, heterogeneous structures for
computing because self-assembly usually leads to homogeneous structures. There-
fore, a hybrid or meet-in-between paradigm that combines the strength of both
bottom-up and top-down processes is more likely to address the design and
fabrication of molecular computing. In such a paradigm, nanoscale elements,
including molecules, are fabricated through a bottom-up process into microscale
units, which are patterned and programmed to achieve large-scale integration
through a more conventional top-down process.
In addition to building computing circuits directly, molecules can be em-
ployed to change the properties of semiconductor devices for computing. It has
been demonstrated that a monolayer of molecules can dramatically change the
threshold voltage, thus leakage current and delay, of silicon MOSFETs, providing
a new mechanism to counter process variations and control leakage power
consumption, which have haunted deep-submicron silicon MOSFET circuits.
In this chapter we will review recent progress in molecular computing, in
particular, our own endeavors in addressing its challenges. While it is impossible
to survey such a rapidly developing research field, we seek to address research in
building and characterizing molecular devices based on bundles of molecules,
circuit and architecture solutions for molecular computing, and molecular graft-
ing for silicon-based computing, with a focus on the integration of molecules into
large-scale computing systems. While many assumptions and speculations can be
made in molecular computing research, we strive to achieve a balance between
fabrication and theoretical design. We refer readers to monographs on molecular
computing [1, 2] for more comprehensive treatments. In particular, the chapter
does not cover the large body of literature on defect tolerance in molecular
computing [3-8] or devices and characterization based on a single molecule [9],
notably work in single-molecule (including carbon nanotube) transistors [10-12]
and molecular wires [13].
11.2. SWITCHING AND MEMORY IN MOLECULAR BUNDLES
While single-molecule switching and memory devices are desirable, positioning
and interconnecting such devices still pose a great research challenge. Therefore,
bundles of hundreds or thousands of molecules, or molecular bundles, have been
employed as switches. Many works that claim single-molecule devices are indeed
based on measurement on molecular bundles.
11.2.1. Self-Assembly and Molecular Ordering
The assembly of massive numbers of molecules onto a predefined surface under-
pins most molecular switches fabricated and characterized so far [14-17]. It is also
critical to molecular grafting of semiconductor surface, a complementary method,
in some cases of nanodevices, and to impurity doping as will be addressed