Quantum Computing Progress Continues
John Baliga, Associate Editor -- Semiconductor International, 10/1/2000
 |
Other walls are not far from being reached. Recent work indicates gate dielectrics must be at least four atomic layers thick. Research going back to the late 1980s indicates that using less than ~100 electrons to form a signal pulse or storage bit can be unreliable,1 and current scaling trends will bring CMOS technology to that limit within a couple of decades.
Quantum computing research has been going on for years, initially motivated by the large number of computations possible. Now it is being viewed as a way to extend computing technology after CMOS runs out of steam.
Quantum computing uses quantum states of atoms, quantum wells or quantum dots as bits, rather than collections of electrons. Quantum-level interactions between the atoms, wells or dots are used to compute or communicate. No actual transfer of electrons occurs from one place to another.
Scientists from IBM Research, Stanford University and the University of Calgary have succeeded in making a five quantum bit (Q-bit) computer. It is actually a molecule specially designed so the nuclear spins of five fluorine atoms can interact with each other. The computer was programmed using rf pulses, and the Q-bit states were detected using nuclear magnetic resonance (NMR). The results were reported at the Hot Chips 2000 Conference at Stanford University, sponsored by the IEEE Computer Society.
Though input and output require a great deal of energy in this "computer," and it runs at only 215 Hz, it did demonstrate the feasibility of reducing the number of steps required to perform a specific task. It required only one step to solve an order finding problem, where a conventional approach usually requires as many as four. This may seem like a small improvement, but in a scaled-up system with a large input the difference can be very significant.
The researchers say the first applications for computing systems like these likely would be as coprocessors for specific applications, such as database searches and specific complex mathematical tasks.
It may be possible to use small devices, such as this research device, as multigate or multiterminal devices. Replacing transistors with multiterminal devices is one of the general approaches under examination for future computing systems.2 It may be that circuit topology as well as device technology will have to change in the next two or three decades.
GaN Growth Processes Improve
Gallium nitride, with a band gap of ~3.44 eV, can be used in LEDs and laser diodes in the blue and green parts of the spectrum, heterostructure devices, and high-temperature devices. Researchers at the Lawrence Berkeley National Laboratory have developed some new techniques for growing high-quality GaN and group III nitride films.
One method uses metallic gallium as a buffer layer to grow GaN films on lattice mismatched substrates, such as sapphire. The buffer is said to provide favorable growth conditions for GaN at higher temperatures for processes such as molecular beam epitaxy (MBE) and metal organic CVD (MOCVD), and it is expected to benefit any group III nitride film growth process.
The researchers also developed a method using bismuth as a surfactant to allow high-quality MBE film growth at lower temperatures. One advantage of the method is that the GaN film can grow as two-dimensional grains that can coalesce, rather than growing individual three-dimensional grains. Another advantage is an increase in the amount of p-type doping possible by using MOCVD.
REFERENCES
- R. W. Keyes, "Miniaturization of Electronics and its Limits," IBM Journal of Research and Development, vol. 32, no. 1 (1988).
- T. Shibata, T. Ohmi, "Neural Microelectronics," Technical Digest of the Int. Electron Devices Meeting, 1997, pp. 337-342.
