Silicon Quantum Computer: Toward the Ultimate Semiconductor Device

Kohei M. Itoh (Associate Professor, Department of Applied Physics and Physico-Informatics)

Silicon semiconductor technology is what supports our advanced information society. Its performance has been improved through "miniaturization (microfabrication)." Making them smaller increases their operating speed, enabling easily portable computers like electronic organizers and IC cards. Figure 1 shows the trend forecast for silicon semiconductor performance based on Moore's Law, where the vertical axis is transistor size and the horizontal axis is the year. While one might want to say that since Moore's Law has held for over 30 years, it will continue to do so, by 2030, we are projected to be processing bit information (0 or 1) at the size (volume) of a single atom—0.1 nanometers!

Therefore, we have chosen to research information processing using individual silicon atoms. Natural silicon is composed of three stable isotopes: Si-28, Si-29, and Si-30. Among these, only Si-29 is a "magnet" possessing a nuclear spin. Furthermore, let's skillfully introduce phosphorus atoms into the silicon. Doing so gives each phosphorus atom one extra electron, which also acts as a "magnet." The phosphorus nucleus is also a "magnet." Here, just as we define north on a compass as 0 and south as 1, if we define the upward orientation of these nuclear and electron magnets as "0" and the downward orientation as "1," and can manipulate them at will, we can create a computer where each bit of information is carried by a single atom and a single electron. (A conceptual diagram is shown in Figure 2.)

This is easier said than done; it is a monumental task. First, all phenomena at the atomic level obey "quantum mechanics" (experimentally discovering something that doesn't would be a Nobel Prize-worthy discovery!). Consequently, the atomic and electron magnets exist not in a digital state of "0 or 1," but in a strange state of being "both 0 and 1 at the same time." The development of quantum computers, which aims to use these mysterious "quantum bits" to make possible what is impossible for modern computers, is at the forefront of basic science.

So, what exactly constitutes this forefront? In our research, we first consider what kind of device we want to build (the architecture). We think about how to build with atoms as our blocks, how to load quantum information, how to perform computations, and how to read out the results. This requires knowledge from the forefront of computer science. Next, we develop the technology to arrange individual atoms according to the design. Figure 3 shows a snapshot of silicon atoms (the individual dots in the photo) that we have arranged in our laboratory. Since we handle individual atoms, this is the forefront of nanotechnology, requiring knowledge of physics and materials science, as well as advanced experimental skills. For computation, we manipulate the quantum bits (the orientation of the magnets) at will. Here, we utilize experimental physics techniques such as nuclear magnetic resonance and electron spin resonance. The final readout requires ultimate measurement technology to detect the extremely weak electromagnetic forces of individual magnets. Figure 2 shows a single particle of light (a photon) being emitted from a phosphorus atom. The polarization of this photon carries quantum information and is also useful for reading out the orientation of the phosphorus nuclear magnet. But measuring the polarization of a single particle of light sounds difficult, doesn't it?

Ultimately, the greatest pleasure in quantum computer research lies in fusing the frontiers of physics, chemistry, materials science, and information science, and working with leading experts from different fields around the world to make the impossible possible. Naturally, as this is also the forefront of engineering, the phenomena elucidated here will likely lead to unforeseen applications in the future. The accumulation of these efforts will be rewarded in 20 years in the form of a useful quantum computer. And as computer science develops concurrently, it is expected that simulations impossible with classical mechanics will become possible. All simulations, such as those for understanding the origin of the universe, chemical reactions, new drug development, heat conduction, magnetism, and device characteristics, are currently performed using computers based on classical mechanics. However, fundamentally, it is unnatural to calculate the behavior of a collection of atoms (i.e., quantum systems) using classical mechanics. That is why the late Professor Feynman, of the bestseller "Surely You're Joking, Mr. Feynman!," proposed, "Let's do simulation with quantum mechanics!" However, at that time (1982), the technology to handle individual atoms, electrons, and photons did not exist. Things are different now. That is precisely why it is so exciting to tackle quantum computer research by bringing together the collective wisdom of science and engineering. "Someday, we want to build a room-temperature silicon classical-quantum hybrid computer like the one shown in Figure 4!" This is our dream. Everyone says it's impossible, but someone has to do it for the sake of the next 30 years.