Solving the 193nm optics contamination puzzle

12/01/2002

To those dedicated to the control of molecular contamination, 193nm lithography presents its own set of demanding challenges. A whole new group of chemical contaminants that threaten lithography processes are introduced, and the entire molecular contamination equation is changed. Early feedback suggests that traditional thinking on thresholds must be cast aside and new sources of molecular contaminants must be examined.

The looming optics threat

Ever try putting together a 5000-piece jigsaw puzzle with 4999 pieces? Preventing optics contamination in 193nm lithography is a similar endeavor. Following years of experience in chemically filtering wafer chamber air to protect sensitive photoresists, the industry has come to understand much about molecular contamination. Unfortunately, now there is a new problem of optics contamination, and all the data needed to complete the picture is not at hand.

Refractory contaminants — those that contaminate ArF optics as residue or ash — can be controlled, although proper filtration and monitoring are necessary. Acid levels as high as 500 parts-per-trillion (ppt) and sulfur oxide at 100ppt have been seen in the ambient at some sites and have been effectively removed by chemical air filtration designed to protect optics.

Condensable organics usually gauge at an average of 25 parts-per-billion (ppb) using conventional methods of measurement. Occasionally, silicon-containing organics have been measured as high as 70ppb in the fab where total organics were over 500ppb. Chemical filters have been successfully designed to remove the heavier, lens-damaging organic compounds from air delivered into the exposure tool. To ensure that optics are protected in either situation, monitoring filter performance and analyzing performance data are crucial to making timely and appropriate filter changes.

Resist complacency

Here's the bad news: When it comes to optics protection, complacency is the enemy. The industry has become so experienced in protecting resists from molecular bases that the optics contamination issue has taken some lithographers by surprise. Resist protection from molecular bases is well recognized as a necessity, even as resists become more environmentally stable. The industry has been developing methods of protecting resists from molecular bases (ammonia, NMP, amines) for more than 10 years.

The earliest methods of resist protection that measured wafer chamber air quality using witness wafers to determine filter condition have given way to modern ones that allow in situ, mid-filter monitoring for the presence of a few ppb of total molecular bases. Mid-filter monitoring of molecular bases employing a serial filter cell layout enables quick, reliable diagnosis of the filter condition when trying to identify which one of a hundred factors may be causing a CD shift. Monitoring for bases has also enabled forecasting of filter degradation rates, allowing more complete use of the filter. Filter technology has matured to extend filter life to three years and longer, so monitoring for bases has become popular. With this maturity, many production 248nm operations measure just 1ppb of ammonia as an indictor of when to change the filter.

Molecular base filtration for resists has also done a reasonable job of protecting 248nm optics from the refractory contaminants that could coat the lens. And because resists for 193nm processes have evolved from their mature 248nm ancestors, the molecular contamination challenges related to 193nm resist protection are not expected to upset current solutions. Because 193nm optics are far more susceptible to the new range of refractory contaminants, however, the challenge is to add organics and refractory contaminant measurement and control to the exposure tool filtration requirements.

The wild card

The industry is now dealing with a wild card: condensable organic contaminants. These organics call into question traditional filter-related cost-of-ownership thinking and operational practice. Take, for example, the parameters for protecting a lithography process from ammonia and base concentrations. Few fabs have high ammonia concentrations (more than 100ppb); most tend to be uniformly low, with the average fab levels at 21ppb and most new fabs below 15ppb. Total variation within a fab is typically 5–40ppb with a few rare spikes >300ppb.

Such is not the case with organics. Organic concentrations may vary from 10–500ppb and, in general, may fluctuate more widely fab-to-fab and within each fab, if only because they have not been targeted for measurement and control to the degree molecular bases have. Solvents such as isopropyl alcohol used for tool wipe down or chemicals used to clean floors can cause a large variation in total organics.

Then there are the undetermined sources of refractory contaminants. Construction materials of the tool itself and related components were a huge issue in the early days of chemically amplified resists, and are again an area of concern for optics. The close proximity of resists to the bottom lens element is a well-documented threat to optics. Resist outgassing of

organics in general, and particularly those containing refractory components, directs the attention of the resist manufacturer to doing his best to mitigate this effect, and the tool manufacturer to protecting the bottom lenses. Servicing and handling protocols need attention; at the same time, filtering the wafer chamber air may only be part of the solution. The battle is ongoing, costly, and competitive, putting demands on all parties involved.

The ramp of 193nm lithography presents a challenge and an opportunity to lithographers. Many optics-threatening pollutants will only be identified and properly characterized in production fab environments. The rate at which these contaminants and new sources are identified, measured, and effectively controlled will accelerate with the accumulation of good data.

Work is underway, of course, at the OEM exposure tool companies, as well as some development and pilot production fabs. But the real experience and data collection will occur as 193nm tools ramp in high-volume fabs. The more data we can collect now and during these early ramps, the faster the solutions will be identified and ultimately mature.

John Higley
Extraction Systems Inc.
Franklin, Massachusetts