ASML favors 'shower' immersion as 'bath' hits throughput
Peter Clarke
12/03/2003 10:57 AM EST
VELDHOVEN, The Netherlands -- As ASML Holding NV launches the first immersion lithography system, the Twinscan XT:1250i, executives are coming to the conclusion that 'showers' are better than 'baths'.
As with the cleaning of human beings, where showers use less water and offer higher throughput, so it is with lithography systems according to Bert Koek and Boudewijn Sluijk, two directors of product marketing at ASML.
While ASML has been developing an internal prototype system over the last seven months the company has tried both 'bath' and so-called 'shower' systems.
"Both options have their pros and cons," said Koek. "With the bath the water is contained and there should be no worries about operation at the edge of the wafer. But the stage becomes heavier and that means motion becomes slower and productivity lower. With the 'shower' method, the stage is lighter but containing the liquid is harder."
With the shower approach only a small bubble of water is required. The water sits immediately below the lens and above the die being exposed and moves with the lens.
Sluijk said that unfortunately the water is not moved across the surface of the wafer simply by surface tension. But it is possible to contain and move the water with blowers. What remains unclear is whether repeated blowing of the water while stepping and scanning can be done for hundreds of exposures without eventually creating disruptive micro-bubbles in the water.
In addition it is not clear if control of the water can be maintained to the edge of the wafer, Sluijk said.
"The shower head concept is still favorite although we are going to have to pay attention to yields at the wafer's edge," said Koek.
Koek said if ASML can't maintain yield to the wafer's edge it is unlikely TSMC and others will be satisfied. Koek said.
But the shower concept remains favorite because the weight of the bath and the water is a productivity killer, Koek said. "At ASML we have had a very active program to save stage weight. In the past we've gone to tremendous lengths just to shave a few milligrams off the weight of the stage. Adding the weight of a bath and one or two liters of water cuts right across that."
Other work in progress related to immersion lithography remain the detection of micrometer scale bubbles and improvements to the photoresists.
"Resists used underwater have different properties to when they are used in air," said Koek. "There are possibilities of adding a top-coat layer on to the resist to make it water tight but for production the manufacturers are saying they don't want that for reasons of complexity and productivity."
Koek said ASML is working with resist manufacturers who are expected to come up with novel resist formulations that work as well underwater as current resists work in air.
Although ASML has successfully made line and space structures at 90-nm critical dimensions using immersion lithography Sluijk admitted the company cannot detect bubbles below about 1-mm in size.
"At one to ten microns we don't know. We do have the capability to de-gas the liquid but what happens in high volume production? Just look how many bubbles you introduce when you stir a liquid."
Koek said ASML was looking at optical methods for detecting micro-bubbles but there is not much room to introduce an optical probe into the water bubble. "That's why we need more production-like experience," said Koek. |
