Finding NEMO

IBM researchers have found a way to extend a key chip-manufacturing process to generate smaller chip circuits, potentially postponing the semiconductor industry's conversion to an expensive alternative.

IBM

researchers have found a way to extend a key chip-manufacturing process to generate smaller chip circuits, potentially postponing the semiconductor industry's high-risk conversion to an extremely expensive alternative.

IBM scientists have created the smallest, high-quality line patterns ever made using deep-ultraviolet (DUV, 193-nanometre) optical lithography -- a technology currently used to essentially "print" circuits on chips. The distinct and uniformly spaced ridges are only 29.9 nanometres wide. This is less than one-third the size of the 90-nanometre features now in mass production and below the 32 nanometres that industry consensus held as the limit for optical lithography techniques.

For decades, the semiconductor industry has relied on continually shrinking circuits to drive increases in the performance and function of chips and the products that use them. But as chip features now approach the fundamental scale limits of individual atoms and molecules, the future of this trend of improvement, known as Moore's Law, is being threatened. IBM's new result indicates that a "high-index immersion" variant of DUV lithography may provide a path for extending Moore's Law further, thus buying the industry time.

"Our goal is to push optical lithography as far as we can so the industry does not have to move to any expensive alternatives until absolutely necessary," said Dr. Robert D. Allen, manager of lithography materials at IBM's Almaden Research Center. "This result is the strongest evidence to date that the industry may have at least seven years of breathing room before any radical changes in chip-making techniques would be needed."

The record-small pattern of well-defined and equally spaced 29.9-nanometre lines and spaces was created on a lithography test apparatus designed and built at IBM Almaden, using new materials developed by its collaborator, JSR Micro, Sunnyvale, California. The first technical details will be presented this week at the SPIE Microlithography 2006 conference being held in San Jose, California.

"We believe that high-index liquid imaging will enable the extension of today's optical lithography through the 45- and 32-nanometre technology nodes," said Mark Slezak, technical manager of JSR Micro. "Our industry faces tough questions about which lithography technology will allow us to be successful below 32 nanometres. This new result gives us another data point favoring the continuation of optical immersion lithography."

How it works

Microelectronic chips are made by a process called photolithography. Photolithography transfers the various circuit design patterns onto a silicon wafer by projecting a uniform beam of laser light through a shadow mask and then focusing it onto a photosensitive "photoresist" material that coats the silicon wafer. Subsequent development, etching, and materials deposition steps form the circuit features. Making a typical computer processor or memory chip may require dozens of photolithography cycles.

Over the years, the industry has created smaller circuit features -- which typically lead to smaller, faster and cheaper electronics -- by using ever-shorter wavelengths of light, stronger lenses and -- most recently -- inserting between the final lens and the silicon wafer a liquid, currently water, that enables even finer resolution.

Until now, it was not known if the industry could continue to adapt this optical immersion technique to produce sharp features smaller than 32 nanometres. New materials required to make such small features were thought to be incompatible with each other or capable of yielding only indistinct, blurred patterns. As a result, in recent years contingency plans are being explored for switching sometime in the future to a radically different but much more expensive (and still unproven) manufacturing method that uses soft-x-rays (also known as EUV, for extreme ultraviolet light) and exotic mirrors rather than laser light and lenses.

As part of its efforts to extend current optical lithography techniques, IBM developed an industry-leading interference immersion lithography test apparatus, called NEMO. IBM's NEMO tool uses two intersecting laser beams to create light-and-dark interference patterns with spacings closer than can be made with current chip-making apparatus. As a result, NEMO is ideal for researching, testing and optimizing the various high-index fluids and photoresists being considered for use in those future DUV systems that would create such fine features. Now that IBM's new result shows a path for extending optical lithography, high-index lens materials must be developed to enable its commercial viability.

When light passes through a transparent material, it slows down in proportion to the material's "refractive index." Light passing through a higher-index material has a shorter wavelength and can thus be focused more tightly. Resolution in immersion lithography is limited by the lowest refractive index of the final lens, fluid and photoresist materials. In IBM's NEMO experiments, the lens and fluid had indices of about 1.6, and the photoresist's index of refraction was 1.7. Future research is aimed at developing lens, fluid and photoresist materials with indices of refraction of 1.9, which would enable even smaller features to be imaged.