ASML Introduces 193nm Step and Scan

In light of recent announcements that 248 nm (DUV) lithography can print features as small as 100 nm (0.10 microns), 193 nm lithography has seemed to some to be a technology in search of a market. This wavelength, matching the emission of an ArF excimer laser, requires new lens designs, new lens materials, and new resist chemistries. It is, critics claim, far too expensive for economical production.

Supporters of 193 nm lithography point out that extending DUV is expensive, too. At 100 nm, less than half the exposure wavelength, DUV exposure requires both optical proximity correction (OPC) and phase-shifting masks (PSM). OPC adds sub-resolution features to the mask to correct for printing and etching errors, increasing the mask write time dramatically. PSMs, with two patterned layers instead of the usual one, exploit destructive interference between adjacent features to cancel diffraction effects (see a related article). Both techniques increase the mask cost.

According to Richard George, director of step and scan marketing at ASM Lithography (Veldhoven, The Netherlands), the cost of a complete DUV mask set for the 130 nm technology node could exceed US$500,000. While such costs are tolerable for high volume products like memories, they are prohibitive for many applications. For low volume logic products, George said, reticle costs would dominate the total wafer cost.

While 193 nm lithography still requires resolution enhancements, OPC alone is likely to be sufficient, George said. In addition, the shorter wavelength offers improved depth-of-focus, widening the process window.

George expects 193 nm lithography to be used for critical layers beginning with the 130 nm technology node, which he expects to ramp production beginning in 2001. Though this schedule coincides with current expectations for 300 mm production, George said, initial 193 nm production is likely to use 200 mm wafers. Few chipmakers are likely to introduce a new lithography technique and a new wafer size at the same time in the same fab. Instead, he expects the first 300 mm fabs to use i-line and 248 nm lithography to make 180 nm devices.

With this schedule in mind, last week ASML introduced the PAS 5500/900 . The system addresses many key issues in 193 nm lithography.

For example, quartz glass, used in lenses for 248 nm tools, strongly absorbs 193 nm radiation. Fused silica and CaF₂, the two primary alternatives, have problems as well. Fused silica is prone to compaction after prolonged radiation exposure, changing the optical properties of the lens. While more sensitive resists will help by reducing the exposure dose, lens lifetime remains a serious issue. Calcium fluoride is not prone to compaction, but it is inherently difficult to polish to the required tolerance: polish rate varies with crystalline direction. The 190 mm-diameter optical-grade CaF₂ blanks required by stepper lenses are in short supply, at least in part because there has been little demand for them (see a related article).

ASML's chosen projection lens, the Zeiss Starlith 900, uses both CaF₂ and fused silica elements. Zeiss uses automated polishing systems to vary the polish speed with position compensating for anisotropic polishing rates. CaF₂ elements in the illuminator provide superior transmission and radiation hardness. In addition to these advantages, George noted, the additional CaF₂ consumption will help suppliers (Bicron and Heraeus) justify additional capacity investments

Catadioptic lens designs, which use both refractive and reflective lens elements, require a working distance of only 1-2 mm between the imaging lens and the wafer. Resist outgassing can contaminate the lens. The Zeiss lens, with a purely refractive design, offers a much larger working distance and is mounted to allow easy access for cleaning.

Finally, George said, laser reliability is critical to a mature 193 nm technology, but currently available 193 nm lasers do not meet the performance benchmarks set for 248 nm lasers. Laser power and pulse frequency need to improve as well. George has seen significant improvements since 1996, and expects further progress from suppliers Lambda Physik and Cymer.

While George declined to discuss the post-193 outlook in detail, he noted that 193 nm lithography has printed features as small as 70 nm. The next wavelength transition, perhaps to 157 nm exposure, is likely to occur after the 100 nm technology node.

By Katherine Derbyshire