Moore's Law states that the number of transistors in a microchip doubles every 18 to 24 months. However, it seems apparent that Moore's Law cannot be maintained forever. One of the most significant barriers to continued miniaturization is the increasing heat generated by cramming ever more circuit elements into a given amount of space. Today's microprocessors can reach 200° Fahrenheit; if they get much hotter, they start failing.
One solution is to make chips from a material that is more heat resistant than that of conventional semiconductors. One material under serious consideration is diamond, a form of pure carbon. In addition to being the hardest known substance, diamond exhibits the highest thermal conductivity, allowing it to remain undamaged even in the presence of temperatures that would melt silicon. Of course, natural diamonds are much too expensive for such applications, but two companies have perfected different techniques for creating diamonds in the laboratory.
Gemesis (www.gemesis.com) simulates the conditions under which natural diamonds are formed. Using a technique first developed in Russia, a diamond seed is placed at one end of a ceramic growth chamber, while graphite (also pure carbon) is placed at the other end with metal solvents in between. The chamber is then compressed within a spherical apparatus, achieving a pressure of 58,000 atmospheres at the center. At the same time, an electric current heats the graphite, causing it to atomize. The carbon atoms are drawn to the cooler end of the chamber, where they bond to the diamond seed. Three days later, a sizable diamond emerges.
Apollo Diamond (www.apollodiamond.com) uses an approach called chemical vapor deposition (CVD). Diamond wafers are placed in a low-pressure chamber filled with a special gas, which is ionized into a plasma. That allows carbon atoms to precipitate out of the plasma cloud and deposit onto the wafer seeds, building up layers of diamond at a rate of half a millimeter per day. That technique, more than a decade old, has been used to coat surfaces with microscopic diamond crystals, but no one has been able to create large single crystals until now, thanks to Apollo's discovery of the correct combination of temperature, pressure, and gas composition.
Because of these advances, the projected cost of cultured diamonds is around $5 per carat, meaning that cost is no longer a roadblock to using diamond microchips. In addition, the crystal structure of cultured diamonds is much more consistent than natural stones, making them better suited for large-scale computer applications.
The biggest obstacle is controlling the diamond's electrical characteristics. Natural diamonds are not conductive, but both Gemesis and Apollo have discovered a way to “dope” the crystal lattice with boron atoms, which forms a p-type semiconductor with an excess of “holes” compared with the density of electrons. In June of this past year, scientists from the U.S. Navy, France, and Israel announced that they had found a way to invert the polarity of boron, allowing it to form an n-type semiconductor, which is the opposite of p-type. As a result, transistors based on p-n junctions are now possible, and diamond microchips are not far behind.
Japan is allocating $6 million per year to develop this technology; in fact, Nippon Telegraph and Telephone (NTT) recently demonstrated a prototype chip operating at 81 GHz. What does this have to do with music? Anything that advances computer technology in general will apply to electronic music, so we could start seeing diamond microchips in music products soon. However, I wouldn't recommend giving one to your fiancée instead of an engagement ring.