At some point, the extendibility of deep-ultra violet optical lithography will decline and be followed by exposure technologies using fundamentally shorter wavelengths. The development of these new production tools is underway. Industry watchers anticipate their commercialization at the 45nm node. But this extreme shrinkage is not good news for all—in particular not for the makers of small-lot and quick-turn semiconductor devices. Therefore, Canon has begun work on two distinctly different solutions--- EUV and Maskless Lithography.

EUV Lithography

Canon started its work with extreme ultraviolet (EUV) lithography in the mid 1980's as part of its x-ray lithography program giving the company a head start in several fundamental aspects of EUV projection, including development of a platform that incorporates many of them.

It’s early work led to the use of multiplayer thin film coatings on reticles and to various reduction mirror designs. Its breakthroughs in platform development include a high-speed 300mm stage that operates in a vacuum, ultra high precision temperature control, environmental control, and wafer and reticle load locks.

Work still remains, in optical materials and lens design, contamination control within the optical path, and thermal management at high throughput. Currently, Canon is developing automated methods for polishing and coating the incredibly precise mirrors required for EUV lithography, and developing metrology and manufacturing tools and methods.

But, overshadowing the gamut of technical issues still facing EUV lithography is its cost of ownership upon reaching commercialization. Canon has a pragmatic roadmap. It regards EUV as a future solution, worthy of aggressive R&D, but not at the expense of continued work on F2 lithography systems and advanced KrF/ ArF lenses and platforms to support new reticle enhancement techniques. Canon's EUV lithography program is underway not only internally but also in close collaboration with industry consortia such as the EUVLLC and EUVIC. Canon is one of the ten Japanese companies forming the Extreme Ultraviolet Lithography System Development Association (EUVA).

The first full field Canon EUV scanner is targeted for 2005, with a second-generation system in 2007.

Maskless Lithography

The scanners and steppers currently being used to make semiconductor chips work by projecting a circuit patterns onto silicon wafers from a mask—a master image printed on a transparent reticle. But the demand for faster chips at a more competitive price that is constantly shrinking the design rules is also driving up mask costs. There are even predictions of $1-million and up mask sets in the sub 100nm future.

While high-volume chipmakers may be able to justify these costs, the makers of small-lot and quick-turn semiconductor devices, particularly system-on-a-chip (SOC) and ASICs, will have a great burden in not only chip development but production costs.

With this scenario in mind, Canon is developing a maskless solution that uses a new multiple e-beam system, called ML2, with correction lens array (CLA) that addresses making direct write practical and affordable.

Although it also utilizes an e-beam as a light source, the ML2 system avoids many of the limitations of traditional single-column electron projection lithography (EPL) tools, namely achieving high resolution and high throughput simultaneously without the need for a reticle.

Simulations and early proof-of-concept tests indicate the throughput rates of ML2 system are similar to EPL methods. Working with others who have expertise in this field, Canon expects to introduce the initial system in 2004, then a second-generation system in 2007.