Vol.2 No.3 2009
Research paper : Creating non-volatile electronics with spintronics technology (S. Yuasa et al.)−205−Synthesiology - English edition Vol.2 No.3 (2009) scanning display devices, and occasionally computing encryption algorithms and other such tasks, and thus not necessarily idle. I think you need to strengthen the explanation of the architecture for normally-off computers a little.Answer (Shinji Yuasa)My desktop personal computer is connected to a liquid crystal monitor, two hard disk drives, a DVD drive, LAN cables, and USB memory. I have Windows running on it, and under Windows I run word processing software, an Internet browser, a mail program, presentation authoring software, and a spreadsheet program all at the same time. Still, monitoring the CPU use shows that the use rate is normally only 1 to 4 %, and rarely exceeds 5 %. In other words, even though various kinds of processing go on in the background, it’s still true that most of the time the CPU is in the standby state. Current electronics that are based on volatility are designed to reduce power consumption by lowering the CPU clock frequency or the power supply voltage during times of low load, but those methods are naturally limited. We think that, in the long term, green IT technology implemented with normally-off technology is necessary.3 Switching speedQuestion and comment (Kazuhito Ohmaki)I’m not an expert in this, but my feeling is that switching speed will be a problem when using magnetic operations. When aiming for energy conservation by using magnetism, as with the TMR device, I wonder about the prospects for switching speed compared to current silicon technology for implementing normally-off computers. Would you comment on that?Answer (Shinji Yuasa)The magnetic switching speed is essentially fast, and can be faster than a few nanoseconds. Non-volatile memory that operates about as fast as the fast memory SRAM that is currently used in CPUs is feasible. It does not, however, represent a landmark speed increase over current silicon technology. In other words, while we can expect about the same operating speed as with current technology, the objective for non-volatility is landmark low power consumption. We added a brief explanation concerning operating speed to section 1.3.4 Theoretical backgroundQuestion and comment (Naoto Kobayashi)You have explained that what provided the opportunity for the first breakthrough (giant TMR using MgO) was the first-principle computation by Butler and Mathon, but I would like to have some explanation of the research background of their dealing with MgO. Was there any theoretical contribution from Japan regarding this?Answer (Shinji Yuasa)It’s not that Butler and Mathon were the first to predict the giant TMR effect, but that they were the first to choose the Fe/MgO/Fe structure as a representative example that strict first-principle computation was possible. With the conventional amorphous Al-O tunnel barrier, the unordered amorphous structure did not allow first-principle computation. The Fe/MgO/Fe structure, on the other hand, features good crystal lattice matching and the possibility of experimental implementation, and I heard that led them to consider the Fe/MgO/Fe structure. Thus, theoretical prediction of the giant TMR effect as a result of performing the Fe/MgO/Fe theoretical calculations could be called a kind of serendipity. In 2001, there were, unfortunately, no researchers who had performed those theoretical calculations in Japan. The computation itself is not particularly difficult, and I even remember it being called “a too-obvious theoretical calculation.” However, this was a “Columbus’s egg” kind of thing, and I think that to actually perform an obvious computation and present the results to experimenters is a praiseworthy achievement.5 A critical eye on technology seedsQuestion and comment (Naoto Kobayashi)The statement in the conclusion that “the potential of technology seeds” is very important is highly interesting. 1) It is very clear that this R & D was a technology seed of very high potential, but I wonder if that might be considered “serendipitous”. Or perhaps only the technology seeds that reach the final goal can be said to effectively have had potential. If that is not so, how should we cultivate the feeling or sense (Leo Esaki’s taste?) for looking critically at the potential in advance? 2) I think that whether the technology seed has potential or not is, not knowable until technological development has proceeded to some extent. Getting across the valley of death requires passing over a number of hurdles, but how can we judge the potential of technology seeds in those respective stages? In other words, how should we decide whether to continue technological development or abandon it when taking the “critical view”? Answer (Shinji Yuasa)Let me respond to questions 1) and 2) together. As is also described briefly in discussion 1, of the various requirements for practicality, whether or not product-level reliability, yield, and other such requirements can be satisfied cannot be known until the final stage of Full Research. Accordingly, we can also say that whether it is good technology that has potential or not cannot really be known until commercialization is attempted. Nevertheless, concerning the opposite judgment that this technology has weak potential, we believe that it is possible to make a decision before beginning R & D or in the first stage. Out of the many technology seeds that appear, I, myself, try to quickly terminate work on weak technology in an early stage. I think that recognizing technology as strong or weak requires a sense that allows phenomena to be analyzed logically and from many viewpoints in a broad field of view. This is probably the opposite of the ability to drill down deeply into a particular phenomenon. While both of those capabilities are essential to conducting Full Research, it is probably too much to expect both from a single researcher. I believe that researchers who are able to drill down into a single phenomenon are in the majority, and those capable of analyzing phenomena from diverse viewpoints in a wide field of view are a smaller group. If researchers lack a sense that allows them to judge the relative potential of technology, the research and development manager should be able to compensate for it. I do not know how we can cultivate a sense for judging the potential of technology seeds, but I should hope that those who become research and development managers would have such a sense.