While the IC will continue to be the driving force behind a great deal of our future technological, economic, and cultural change, the people behind its continued evolution are increasingly faced with problems arising from the miniaturization of the IC's many components.

In 1965, Gordon Moore observed that transistor density in ICs had increased by a factor of two during the time his research & development team had been working with the devices. This led him to project, offhand, that transistor technology would continue to shrink by a similar factor every year for the next 10 years. In 1975, to his surprise, Moore looked back, and found that his prediction had been absolutely correct. Based on available data, he then went on to predict that the trend would continue at a slower pace, with transistor density doubling every 2 years. In fact, transistor doubling did slow down, with the density doubling every 18 months. Since then, Moore's Law, as it's been dubbed, has been applied to a number of features related to computer chips, and other ICs. Moore's Law is thus said to predict that line width, clock frequency, and memory density will double every 4 years, 2 years, and 18 months respectively.

Over the years, though, researchers in the field have come to worry about the effective limits to Moore's Law. That is, when will Moore's Law run up hard against physical laws that prevent further miniaturization? Even before then, will there come a time when the economic, and technological costs of the doubling of transistors, and of clock frequencies will outweigh the benefits? Years ago, when researchers had started experimenting with devices based on a 10 micron process, many speculated the costs would outweigh the benefits by the time devices were being developed on a 200nm (nanometer) process. Today, in 1999, manufacturers such as AMD, IBM, Intel, and Motorola are mass-producing chips based on a 180nm (0.18 micron) process that surpass former expectations. So far then, innovative R&D teams have happily found ways to extend the practical viability of Moore's Law. Today, it is predicted that circuits can be scaled down to a 50nm process before meeting any firm limits.

The question then becomes whether or not it will be possible to adapt current technology to work towards and past the 50nm limit, some 10 years from now, or whether fundamentally new technologies will be introduced to continue to improve microprocessor technology. Indeed, the question even arises whether it will even be necessary to "beat" the 50nm barrier. Given some approaches being worked out in the lab, it may be a long while before we'll miss the steady sound of shrinking line widths. Just in case we don't, though, others are working on improving the lithography techniques used to print up the chips and wafers that drive the computer, and electronics industries.

part 2: Crossing The Line