Editor's Notes

Recent important breakthroughs in a technology called superconductivity promise major advances in several fields. COMPUTE!'s Technical Editor, Ottis Cowper, explains the implications for computers in this guest editorial.—Richard Mansfield

Computer designers, like race-car designers, are constantly seeking ways to make their machines run faster. The faster a computer operates, the more information it can process in a given amount of time; hence, the more effective and powerful it is.

Information is stored and manipulated within a digital computer as collections of electronic pulses, so the fundamental speed limit for information flow is the speed at which electricity flows through the computer's circuitry—approximately the speed of light. However, there are a number of reasons why computers achieve only a fraction of this theoretical speed.

One limitation is inherent in the integrated circuit chips that make up the computer. These chips contain thousands of tiny electronic switches called gates. The gates can't switch off and on instantaneously; it takes a certain amount of time for a signal to make it through a gate. The more gates a signal has to flow through, the more delays it will encounter. In even the simplest of computers, a pulse may pass through dozens, or even hundreds, of gates between input and output, with each gate delaying the pulse a bit more.

The easiest way to make a given electronic gate switch faster is to operate it at a higher power, but this introduces other problems. At higher power, each gate must dissipate more energy as it is switched off or on—energy dissipated in the form of heat. High power can yield high speeds, but it can also cause chips to overheat and break down. This is especially a problem in high-density integrated circuits, which may pack 100,000 gates on a single chip.

Another way to improve computer speed is to construct chips from a material that switches faster at a given power level. Typical gate delays in silicon, the material most commonly used in computer chips, are measured in tens of nanoseconds (a nanosecond is 10^-9 second, one-billionth of a second). That might sound unimaginably fast, but in electronic terms, a nanosecond is a long time. An electrical pulse could travel 30 feet through a wire during the 30 nanoseconds spent waiting for a silicon gate to switch.

Currently being studied as an alternative to silicon as a semiconductor material is gallium arsenide (GaAs). A gate fabricated in GaAs can operate with a delay 100 times less than that of the equivalent silicon gate. Typical gate delays in gallium arsenide are measured in hundreds of picoseconds (a picosecond is 10^-12 second, one-trillionth of a second). However, for even faster speeds, a totally new technology is needed.

In the early 1960s, researcher Brian Josephson proposed a new type of gate, which later bore his name. A Josephson junction can operate (switch) in a picosecond or less, much faster than any other gate, even one made from gallium arsenide. However, even though the Josephson junction has been around for 20 years, it made its appearance in commercially available hardware only in the past year. The problem with developing practical Josephson junction devices is that they can be fabricated only from superconducting materials.

Superconductors were discovered in the early 1900s, when a Dutch researcher found that some common metals suddenly take on radically different properties when cooled to near absolute zero. (Absolute zero is 0° on the Kelvin temperature scale used by physicists; it's equivalent to -459° Fahrenheit.) The most notable property of a superconductor is that it exhibits essentially no resistance to the flow of electric current. (Superconductors have some unusual magnetic effects as well.)

For years, researchers have sought materials that would become superconducting at warmer temperatures. Until recently, the results weren't encouraging. The "warmest" superconducting alloy still required a temperature of about -420°. To achieve and maintain such low temperatures, the materials had to be immersed in liquid helium or liquid hydrogen, which is very expensive. Under these conditions, superconducting gates weren't practical for computing applications.

The past few months, however, have seen an avalanche of breakthroughs. Just a few weeks ago there was the astonishing announcement of the discovery of a material which superconducts at a balmy 9° F, a temperature well within the capabilities of conventional refrigeration equipment. Many researchers now believe that room-temperature superconducting materials are possible, and that the discovery of such materials is imminent.

At these temperatures, superconductors—and chips with blindingly fast Josephson-junction integrated circuits—can make the transition from the laboratory into practical applications. For example, freed from extreme power and cooling requirements, today's mainframe supercomputers could shrink to desktop size, with corresponding reductions in cost. This would bring the tremendous computing power of these machines to far more users, paving the way for breakthroughs in areas such as artificial intelligence, where the processing speed of the current generation of desktop computers is a major bottleneck.

We can expect to hear more about superconductors in coming months. This rapidly expanding technology has applications in areas other than just computers. For example, superconducting materials can be used to make extremely powerful electromagnets. A number of states are currently competing to house the Superconducting Supercollider, a federally funded project that promises to open new frontiers in high-energy physics. Using superconducting magnets, this unit will produce energy levels far in excess of those in existing accelerators, allowing scientists to probe deeper into the arcane world of subatomic particles. And superconducting magnets may make practical the proposed high-speed maglev (magnetic levitation) trains, which float just above their tracks on a powerful magnetic field. It appears that the superconductor will be one of the next high-tech items to make the leap from research labs into our everyday lives.