Proceedings Article | 15 October 2012
KEYWORDS: Oscillators, Data processing, Electrons, Manufacturing, Monte Carlo methods, Capacitors, Stochastic processes, Device simulation, Complex systems, Diffusion
This paper reports the study of a two-dimensional device-error-redundant single-electron (SE) circuit. The circuit is an
SE reaction-diffusion (RD) circuit that imitates the unique behavior of the chemical RD system and is expected to be a
new information processing system. The original RD system is a complex chemical system that is said to express selforganizing dynamics in nature. It can also be assumed to operate as parallel information processing systems. Therefore,
by imitating the original RD system for SE circuits, this SE-RD circuit can perform parallel information processing that
is based on a natural phenomenon. However, the circuit is very sensitive to noise because it is controlled by a very small
amount of energy. It is also sensitive to device errors (e.g., circuit parameter fluctuations in the fabrication process).
Generally, fluctuations caused by errors introduced in manufacturing the circuit components trigger incorrect circuit
operations, including noises. To overcome such noises, the circuit requires redundant properties for noise. To address
this issue, we consider mimicking the information processing method of the natural world for the circuit to obtain noise
redundancy. Actually, we previously proposed a unique method based on a model of neural networks with a stochastic
resonance (SR) for the circuit. The SR phenomenon, which was discovered in studies of living things (e.g., insects), can
be considered a type of noise-energy-harnessing system. Many researchers have proposed SR-based applications for
novel electronic devices or systems. In networks where SR exists, signals can generally be distinguished from noise by
harnessing noise energy. We previously designed SE-SR systems and succeeded in making an architecture for an SE
circuit that has thermal noise redundancy. At the time, we applied an SR model proposed by Collins to our circuit. Prior
to our current study, however, it had not yet been confirmed whether SE circuits have device-error redundancy. In this
study, we attempt to confirm this by using Monte Carlo simulation to study the characteristics of the abovementioned
SE-RD circuit. Simulation results indicate that the SE-RD circuit, which is based on an SR model, has not only deviceerror
redundancy but also thermal noise redundancy. The circuit is therefore expected to prove that the parameter
matching step in the circuit fabrication process can be omitted.
