Creation of molecular neural network based on electron transfer reaction
Research Aim
In order to realize molecular-scale electronics, a system such as the brain, which works cooperatively as a network of individual molecules at work—in other words, a neural network—is a more appropriate model than adapting current silicon devices to the workings of a single molecule. Neural networks are unlike current semiconductor devices that pursue high-speed, high-density, and high-reliability; instead, we will realize flexible information processing in which stochastic elements act collectively to create a network that, based on stochastic resonance, is tolerant to defects. This research is an attempt to use molecular architectonics to build a neural network that shows functions similar to that of the brain with molecular material.
Role and Need in the Group
A specific methodology is required to integrate single molecules and functions. We will build a prototype of a molecular neural network based on the functions of a single molecule that were revealed by this research area and present an architecture that should be sought by molecular electronics.
Research Content
Using molecular architectonics, we will incorporate functional molecules into the molecular network and construct a network that shows brain-like functions based on stochastic resonance. We will proceed with our research while considering in particular the following points:
1. Memory function: Incorporate a protein that is capable of multi-stage oxidation-reduction and expressing memory function based on the state transition of stochastic resonance.
2. Connect to the stochastic resonance network with negative resistance molecules, form a feedback circuit, and produce oscillation behavior.
3. Bias supply: Supply electric potential to the network by incorporating molecules with photovoltaic power.
Creation of molecular neural network based on electron transfer reaction
Research Aim
In order to realize molecular-scale electronics, a system such as the brain, which works cooperatively as a network of individual molecules at work—in other words, a neural network—is a more appropriate model than adapting current silicon devices to the workings of a single molecule. Neural networks are unlike current semiconductor devices that pursue high-speed, high-density, and high-reliability; instead, we will realize flexible information processing in which stochastic elements act collectively to create a network that, based on stochastic resonance, is tolerant to defects. This research is an attempt to use molecular architectonics to build a neural network that shows functions similar to that of the brain with molecular material.
Role and Need in the Group
A specific methodology is required to integrate single molecules and functions. We will build a prototype of a molecular neural network based on the functions of a single molecule that were revealed by this research area and present an architecture that should be sought by molecular electronics.
Research Content
Others
Members
Papers List
2014
Yoshiaki Hirano; Yuji Segawa; Takayoshi Kuroda-Sowa; Tomoji Kawai; Takuya Matsumoto
Appl. Phys. Lett., 104, 233104, , 2014/June/09
DOI: 10.1063/1.4882160