Thursday, September 21, 2000

Though the 50nm limit remains the subject of an interesting academic discussion, there remains a great deal of space between it, and even the newest 180nm processes available to the industry. Making progress in this regard, too, is no small feat. Lithographic techniques are used to print circuits on the fingernail-sized slips of silicon used in computer chips. Lithography employs focused beams of electromagnetic light as the medium for writing to a silicon wafer, projected through an intermediary mask etched with the desired pattern. As stands to reason, to write at a scale of 180nm, one requires a "pen" with a very fine point; one approximating the width of the line desired. To this effect, modern lithography employs ultra-violet light with a wavelength of about 200nm to do the job. To reach 50nm, though, will require lasers of much shorter wavelengths.

To this end, several research teams are developing their own methods of putting a finer point on the lithographic technique. A team at Intel has taken to extending the ultra-violet path into the Extreme Ultra Violet, or EUV, realm. Their technique involves a controlled supersonic xenon implosion as a method of generating an EUV beam with a 13nm wavelength. At this extreme frequency, though, focusing equipment must come in the form of a complex array of multi-layered mirror; no lens system exists that can focus these extreme frequencies. This method is fairly complicated, and it remains to be seen whether it will prove cost-effective.

A team at Lucent has taken to adapting another existing lithographic device - that used to produce optical masks. Masks themselves are drawn using electron beams. With their radii of about 2.83*10^-23, electrons are clearly small enough to be used to draw next generation circuits. The problem with using electrons, rather than photons, is that electrons are negatively charged particles with a strong inclination to repel one another. This hasn't proved much of a problem when drawing masks; it is a much slower process using a thin beam, and fewer particles. For quick, mass production of millions of chips, though, the technique would require many beams of a greater number of electrons, all of which would be strongly repelling each other. The Lucent team is then faced with a serious challenge to overcome the mutual repulsion generated by their beams.

IBM, for its part, is working on a system employing x-rays as their breakthrough-hopeful. To generate a 5nm beam, the IBM team uses a huge, bulk of a device called a synchrotron. In it, electrons are accelerated to the point where they begin to radiate x-rays at the desired wavelength. As they do, that energy is directed at the target; in this case the silicon wafer on which ICs are printed. Since the emitter, and mask are placed in close to the face of the wafer there is also no need for a complex focusing device. On the other hand, the mask itself must be scaled to the exact size of the circuit to be printed. With today's lithographic techniques, mask's are typically scaled to about 4 times the size of the target device. In contrast, printing up masks of suitable size for this process will possibly prove a bit tricky, in and of itself.

part 3: Leaping Leptons