Message archive‎ > ‎

Research on Magnetic Computers / 磁性コンピュータの研究

posted Apr 11, 2016, 10:41 PM by PARC Osaka University   [ updated Apr 11, 2016, 10:47 PM ]

Most functions in the computers that we use in our daily lives are implemented with semiconductor devices. If we could replace these semiconductor devices with magnetic devices, we could build a computer that is highly resistant to radiation and that uses very little energy, requiring no power to preserve information. While such a computer may or may not be feasible, our lab was presented with the opportunity to conduct research on magnetic computers.

I believe it was in the year 2001. A former professor had just returned from a business trip overseas and informed us that they were studying the use of magnets in logical operations there. However, magnetic bodies was not this professor’s specialty, and it was apparent that he understood little of the research and could offer no further information than the two keywords “magnets” and “operations.” There were a few ideas at that time, albeit few, for using tiny magnetic bodies and magnetic wires to perform computations. However, I had drawn no connections between the two since these devices had a long way to go before being useful for calculations. At some point, it struck me that we may be able to perform computations with small magnets if we packed several of them close together. The outcome may have been different had we first studied the existing research.

Two magnets repel each other when their north poles or south poles are brought close together because the system is in a high-energy state due to the magnetic fields produced by both magnets. Consequently, the system attempts to lower this energy state by moving the magnets away from each other. If the north pole of one magnet is brought close to the south pole of the other, the energy of the system drops as the two poles near each other, causing the magnets to stick together. When a tiny magnetic body is formed on a silicon substrate, on the other hand, the system lowers energy by reversing the orientation of the magnet’s north and south poles, i.e., the direction of magnetization, since the magnet itself cannot move. This means that a certain action will lead to a certain result, which could be construed as a logical operation.

Initially we decided to apply for a patent on this concept in March 2002 (unexamined patent application No. 2003-280892). At that time, we had focused only on the concept of using magnetic bodies for logical operations and, looking back, the shape and system of these devices are now considerably different. We strayed from this research for a while after that, but in 2003 we had a visiting postdoctoral fellow from Bangladesh, named Anis, explore some structures for devices through micro magnetic simulations. Anis specialized in computer science and had no specific knowledge of magnetic bodies. Perhaps for the very reason of not being constrained by common knowledge of magnetic bodies, he referenced papers on logical operations using electric fields and conducted simulations on magnetic elements with similar structures. It may have been just good fortune, but after about a half year he discovered a system for implementing logical operations with magnetic elements. When four magnetic bodies are arranged in close proximity to each other and data is inputted into three of the bodies, the direction of magnetization in the remaining magnetic body indicates the result of the operation and, thus, the output of the system. This configuration was shown to perform logical NAND and NOR operations (S.A. Haque, M. Yamamoto, R. Nakatani, and Y. Endo, J. Magn.& Magn. Mater., 282, 380-384 [2004]).

These findings have led to the current research, which has progressed from simple logical operations to elements that transmit data to each other and elements that divert data. Thus, there is promise for building a magnetic computer by connecting these elements together (H. Nomura, M. Yamamoto and R. Nakatani, Appl. Phys. Exp., 4, 013004 [2001]).

The magnetic logic gates mentioned above possess an extremely strong resistance to radiation and use no power to preserve data, which should allow them to operate under the extreme conditions in space and around nuclear reactors, for example, environments for which semiconductor devices are unsuitable. So what is the point? Looking further into the future, the environment on earth will one day become inhabitable for humans. Like in the plot of a science fiction story, humans may have to resettle in another planetary system. However, the effects of radiation in space described above are actually much stronger than people think. It is unlikely that living organisms would be able to tolerate it. It has been suggested that zygotes could survive in such an environment, perhaps shielded by water.

At birth, humans have a very immature form and require a lengthy development period. Conversely, this lengthy development period is necessary for the growth of human intelligence. Even if the zygotes were incubated in space, they must be educated in order to be raised as intelligent beings. Some people have envisioned robots to fill the role of educators. Naturally, the robots could not be dependent on semiconductor devices since semiconductors are vulnerable to cosmic rays. The robots would need to possess a brain that functions using magnetic logic gates. While this may appear to be an extreme fantasy—we would need to launch numerous spacecraft controlled by magnetic computers with the hope that one lucky spacecraft would discover a star system in which humankind could survive, and would have to incubate cryopreserved zygotes that would then be educated by robots once they become young humans—this may be the only solution for the survival of humankind. Of course all of this ignores the significance of going to these length to sustain the existence of humankind.

April 12, 2016
Ryoichi Nakatani, Professor
Division of Materials and Manufacturing Science
Graduate School of Engineering, Osaka University




 とりあえず、この概念を2002年3月に特許として出願しました[特開2003-280892]。今から考えると、概念だけ考えたわけですから、形状、システムは今とはかなり違います。しばらく、その研究は放置していましたが、2003年にAnisというバングラデシュからきたポスドクにマイクロマグネティクス・シミュレーションを使って、素子としての構造を模索してもらうことにしました。彼は、コンピュータサイエンスが専門で、磁性体は非専門です。磁性体の常識にはとらわれませんから、電場によって演算する論文を見つけてきて、似たような構成の磁性体素子についてシミュレーションを行いました。幸運だったのかもしれませんが、半年くらいで演算する系が見つかります。4個の磁性体を近接し、情報を3個の磁性体に入れておくと、残りの1個の磁性体の磁化の向きが演算結果となり出力されそうです。その演算はNANDおよびNORの論理演算でした[S. A. Haque, M. Yamamoto, R. Nakatani and Y. Endo, J. Magn. & Magn. Mater., 282, 380-384 (2004)]。
 現在では、そこから始まった研究は、論理演算だけではなく、情報を素子間で伝搬したり、情報を分岐する素子にも発展しています。これらを接続すればコンピュータができあがりそうです[H. Nomura and R. Nakatani, Appl. Phys. Exp., 4, 013004 (2011)]。