As computer chips get faster and feature sizes on integrated circuits get smaller, scientists and engineers must get smarter. Every chip generation requires advances in photolithography technology to manufacture the latest integrated circuits. Predictions are repeatedly made regarding the minimum feature size that lithography technologies can be pushed to produce — and every time these predictions are surpassed using clever new techniques. After 193 nm lithography came 193 nm immersion lithography, and after that came double-patterning.

Every new generation pushes the limits of lithography, but these advances are more like leaps than steps. The semiconductor industry is putting all its weight behind extreme ultraviolet (EUV) lithography — a lithography technology unlike any other. Indeed, it's the most challenging lithography technology ever developed, according to John Warlaumont, vice president of advanced technologies at SEMATECH, a global consortium of semiconductor manufacturers (see page 30). The fact that EUV is absorbed by all matter, including air, is one reason why this technology is so challenging. The other is that lenses can't operate in the EUV; instead, specially designed mirrors must be used to manipulate the EUV beam. There is also the challenge of constructing reliable and powerful sources of EUV light, and scientists and engineers are making significant progress. Many of the elements needed for EUV have already been developed, and the industry is optimistic that a stable source should be available by 2012 (see page 24).

At the same time, various maskless lithography technologies are gaining acclaim. Although they may not have the huge industry backing of EUV developers and large chipmakers, they are each overcoming their own challenges and addressing smaller markets. For example, nano-imprint lithography has come out of the lab and is being used for photonic applications (see page 27), and table-top lithography systems are finding fascinating applications in biotechnology (see page 22).

The big question is where to go after EUV, when feature sizes start to approach the atomic scale.