New Semiconductor Lithography Makes a Splash

Hank Hogan

Semiconductor manufacturers are about to be all wet. But that's a good thing. A new technique called immersion lithography could allow the industry to continue tracking Moore's law, which forecasts substantial annual decreases in state-of-the-art transistor sizes.

Figure 1. Next-generation lithography may be all wet. Immersion lithography uses the optical properties of a liquid, typically water, to image smaller features. This conceptual design shows the favored approach, which has been dubbed "splash and flash." Courtesy of Nikon Corp.

The name of the method stems from its key difference with existing photolithography. "The space from the bottom of the lens system to the [silicon] wafer is filled with a liquid, and so that becomes another optical element," said Walt Trybula, strategic planning manager for lithography at International Sematech in Austin, Texas. The research consortium has been active in assessing the promise and perils of immersion lithography since late 2002, and leading toolmakers -- such as Canon Inc. and Nikon Corp., both in Tokyo, and ASML of Veldhoven, the Netherlands -- have said they will make a decision on immersion lithography products early in 2004.

Figure 2. A side view of "splash and flash" reveals a layer of water between the wafer and the optics. Courtesy of Nikon and Tokyo Electron Ltd.

Everyone is interested in getting wet because staying dry may be too challenging and because it may be too difficult for the dry approach to keep up with Moore's law. Currently, the most advanced manufacturing process is based on ArF excimer lasers operating at 193 nm. A rule of thumb is that features can be imaged that are approximately a third of the wavelength of the illumination source, so that a 193-nm laser yields minimum features of 65 nm. The next step involves light sources at 157 nm, which will require new imaging optics and photoresists that may not be ready in time.

Figure 3. Immersion lithography has successfully imaged simple lines and spaces, shown here in scanning electron micrographs. Toolmakers such as Nikon and resist-makers such as Tokyo Ohka Kogyo Co. Ltd. expect to decide on the production of immersion lithography tools in early 2004. Courtesy of Nikon and Tokyo Ohka Kogyo.

That's where immersion lithography comes in. The wavelength of light in a liquid is equal to the wavelength in air divided by the refractive index of the liquid. For example, because the index of refraction of water is roughly 1.45, the wavelength for a 193-nm source in water is 135 nm. This thereby shrinks the minimum feature size to approximately 45 nm for such sources.

In practice, things are a bit more complicated. For example, if water is to be used, its optical properties must be well characterized and controlled. Working with the National Institute of Standards and Technology in Gaithersburg, Md., and in Boulder, Colo., the semiconductor industry is developing techniques to measure the index of refraction of water to parts per million.

Investigations have shown that using doubly deionized, degassed water eliminates bubbles and their impact on images. Other studies have shown that water's index of refraction can meet the tight specifications, provided that the temperature is controlled to 0.1 °C.

The thinking is that immersion lithography will follow a "splash-and-flash" approach in which the tool will step over the wafer, inject water into the field, expose it to light, remove the water and then move to the next field to repeat the process. This is similar to the current dry method and ensures that a known amount of liquid is in the optical path. Using this technique, Nikon has successfully imaged 45-nm features.

Taku Hirayama, project leader of immersion lithography materials at resist-maker Tokyo Ohka Kogyo Co. Ltd. of Kawasaki, Japan, noted that, for a resist company, the challenge is to develop materials that will minimize the dissolution of the resist components in water and that will decrease water uptake by the surface of the resist film. The solution may be a barrier layer, which the company already has demonstrated, but there may be other approaches. Such barrier layers cap the resist, minimizing the ability of the water to dissolve the resist and reducing the amount of water absorbed by the resist.

Akiyoshi Suzuki, general manager of a Canon optics research division, predicted a wet future. "If we conclude immersion lithography in 193 nm is possible, it will surely be used as a production technology in the 45-nm node, because it is cost-effective and is a very simple extension from the user's point of view," he said.