AN INTRODUCTION TO NANOCOMPUTING - Bio-Inspired and Nanoscale Integrated Computing - page 9

−

−

−

−

−

−
−

−

−
−

−

−

−

−

− −

− − − − − −

−

−

−

− −

− −

−

− − −

− −

Channel length > 5 nm to 15 nm,
approximately

Channel length < 5 nm to 15 nm,
approximately

Figure 1.7. When transistors become very tiny, electrons can tunnel across the
channel when the gate tries to block current flow. If the channel length is small
enough, electrons will regularly tunnel in this way, and the gate would no longer
effectively control current flowing through the wire.

area interconnections require as long as we connect transistors with traditional
wires [8]. As we will see in the next few sections, there are many ideas in this topic
that reduce the limitations of wiring.
Second, variations during fabrication are now becoming a very significant
problem. Relative to such tiny transistors, variations in geometry or chemical
concentrations can easily change or break the behavior of the transistor. This
variability decreases the yield and reliability of devices. Fault-tolerant methodology
(Chapter 10) is desired for computing under unreliable conditions, and new
fabrication methods, such as self-assembly (Chapter 12), may be better for reliable
fabrication at the nanometer scale. Furthermore, reconfigurability (Chapter 5)
offers a way to keep a device useful by updating or fixing its functionality.
Finally, when transistors become very small (below 5 to 15 nm approxi-
mately), electrons will be able to tunnel in a different, much more challenging way:
electrons would be able to tunnel through the channel itself, even when the gate
tries to block current, defeating the purpose of a gate entirely (Fig. 1.7). It is
currently not clear how to overcome this upcoming problem, except to find a
better nanoscale device that can behave like a switch [9].

1.4. BEYOND THE TRANSISTOR: NANOSCALE DEVICES

In practice, the use of transistors has been so successful that so far it remains
unchallenged as the ''best way'' to use physics for computation. However, as
mentioned above, it is not clear that the transistor will continue to be the
best device to use as a switch at the nanometer scale. One major facet of
nanocomputing research is finding new devices that exhibit switching or other
behaviors that are useful for computing. Unlike the classical transistor, these