Device variability and circuit redundancy in signal processing based on nanoswitches

A Probabilistic Finite State Logic Machine Realized Experimentally on a Single Dopant Atom

Exploiting the potential of nanoscale devices for logic processing requires the implementation of computing functionalities departing from the conventional switching paradigm. We report on the design and the experimental realization of a probabilistic finite state machine in a single phosphorus donor atom placed in a silicon matrix electrically addressed and probed by scanning tunneling spectroscopy (STS). The single atom logic unit simulates the flow of visitors in a maze whose topology is determined by the dynamics of the electronic transport through the states of the dopant. By considering the simplest case of a unique charge state for which three electronic states can be resolved, we demonstrate an efficient solution of the following problem: in a maze of four connected rooms, what is the optimal combination of door opening rates in order to maximize the time that visitors spend in one specific chamber? The implementation takes advantage of the stochastic nature of electron tunneling, while the output remains the macroscopic current whose reading can be realized with standard techniques and does not require single electron sensitivity.

Sub-threshold signal processing in arrays of non-identical nanostructures

Weak input signals are routinely processed by molecular-scaled biological networks composed of non-identical units that operate correctly in a noisy environment. In order to show that artificial nanostructures can mimic this behavior, we explore theoretically noise-assisted signal processing in arrays of metallic nanoparticles functionalized with organic ligands that act as tunneling junctions connecting the nanoparticle to the external electrodes. The electronic transfer through the nanostructure is based on the Coulomb blockade and tunneling effects. Because of the fabrication uncertainties, these nanostructures are expected to show a high variability in their physical characteristics and a diversity-induced static noise should be considered together with the dynamic noise caused by thermal fluctuations. This static noise originates from the hardware variability and produces fluctuations in the threshold potential of the individual nanoparticles arranged in a parallel array. The correlation between different input (potential) and output (current) signals in the array is analyzed as a function of temperature, applied voltage, and the variability in the electrical properties of the nanostructures. Extensive kinetic Monte Carlo simulations with nanostructures whose basic properties have been demonstrated experimentally show that variability can enhance the correlation, even for the case of weak signals and high variability, provided that the signal is processed by a sufficiently high number of nanostructures. Moderate redundancy permits us not only to minimize the adverse effects of the hardware variability but also to take advantage of the nanoparticles' threshold fluctuations to increase the detection range at low temperatures. This conclusion holds for the average behavior of a moderately large statistical ensemble of non-identical nanostructures processing different types of input signals and suggests that variability could be beneficial for signal processing. We demonstrate also that circuits composed of coupled non-identical nanoparticles can act as elementary nano-oscillators that show synchronization properties for sub-threshold stimuli. The results obtained should be of conceptual interest for the design of reliable signal processing schemes with non-identical nanostructures.

Reliable signal processing using parallel arrays of non-identical nanostructures and stochastic resonance

In the stochastic resonance (SR) phenomena, the response of a non-linear system to a weak periodic input signal is optimised by the presence of a particular level of noise which enhances signal detection. We explore, theoretically, the influence of thermal noise in arrays of metal nanoparticles functionalised with organic ligands acting as tunnelling junctions, with emphasis on the interplay between the SR phenomena and the nanostructure variability. In this system, the transference of a reduced number of electrons may suffice to implement a variety of electronic functions. However, because nanostructures are expected to show a significant variability in their physical characteristics, it is important to study the relation between the diversity-induced static noise and the dynamic noise caused by thermal fluctuations. We consider an ideal model based on the Coulomb blockade and tunnelling effects that includes the stochastic nature of electron transference due to thermal noise together with the nanostructure variability found in experimental distribution functions. The correlation between the input (potential) and the output (current) signals, as well as the absolute value of the current and its time fluctuations, are analysed as a function of the temperature and the number of nanostructures. Extensive kinetic Monte Carlo simulations suggest that the interplay between thermal noise and variability could permit reliable processing of weak signals with many non-identical nanostructures.