Navigation and service

Quantum Theory of Materials seminar, PGI-1/IAS-1

Prof. Tomonobu Nakayama

How can we let materials think?

18 Sep 2018 14:00
Seminar room N-142

NIMS, Tsukuba, Japan


Computers, the most important tools for our society at present, are now consuming a lot of energy. We, however, need more powerful computers to solve complex or simply large-scale problems in the coming era of “big data”. Even using huge energy, there exists difficult problems to be solved by present computers. To overcome these issues, researchers are working hard to establish new types of computing such as quantum computing, neuromorphic computing, and other kinds of so-called “natural computing”. In this presentation, I propose the use of complex networks for future neuromorphic computation and discuss how and why materials can think.

We prepared inorganic/organic neuromorphic nanowire networks made of doped poly-aniline nanowires (PANI-NWs) and sliver nanowires (Ag-NWs) by wet-chemical methods and by drop-casting or spin-coating them onto insulating substrates. In the case Ag nanowires, about 1-nm thick insulating layer of polyvinylpyrrolidone (PVP) was formed over the surface of each nanowire. Here, it is important to point out that these networks are not at all designed regarding their network structure as readily seen in Figs. 1(a) and 1(b). Then, we used multiple-probe scanning probe microscope (MP-SPM) and related techniques to investigate emerging properties of those nanowire networks.

Prepared networks exhibited “small-world” characteristics, indicating that there were multiple ways to efficiently connect two separated sites in the network. This would be a “motivation” for materials to think “Which way is more efficient to transport electrons?”. Also, the Ag nanowire network showed interesting behaviour. Because of the thin insulating PVP layer, the resistance of the Ag-NW network was very high at the beginning. However, the resistance was orders of magnitude lowered by applying appropriate voltages across the network and the low-resistance state disappeared after some retention time. Recently, we are able to perform training of the network to organize connectivity inside complex network as if our brain learns and memorize information.


Dr. Shigeru Tsukamoto
Phone: +49 2461 61-6449
Fax: +49 2461 61-2850