Stepping and scanning into the NA>1 immersion exposure era
08/01/2005
During the past two years, immersion 193nm lithography has made astonishing strides in moving from the conceptual stage in R&D to the development and testing of the first full-field scanners that use water between the wafer and exposure lens to improve depth-of-focus (DOF) and printing resolution. This article provides new details about the results from immersion ArF lithography compared to equivalent dry ArF scanners, and it outlines ASML’s roadmap for rolling out new production immersion systems with numerical-aperture optics >1.0NA.
The feasibility of scanning immersion exposure systems has been demonstrated on the first full-field immersion scanner to make its public debut in 2003. Based on ASML’s dual-stage Twinscan platform, the AT:1150i 0.75NA argon-fluoride (ArF) system produced first immersion images in October 2003, and results from a comparison of equivalent wet and dry 0.75NA scanners showed a substantial gain in DOF for dense patterns and so-called forbidden pitches [1]. During the first half of 2004, the scanner was tested by 16 leading-edge IC manufacturers, and all confirmed a significant benefit from immersion 193nm lithography [2, 3]. This successful proof of principal has helped to alter the industry’s lithography roadmap by inserting ArF immersion technology into the exposure solution set for the 65nm and 45nm process nodes, with potential extension to even smaller resolution.
The first commercial immersion system for production, the XT:1250i, uses a 0.85NA ArF lens and the Twinscan platform, which separates the exposure step and the wafer alignment-and-leveling step into two concurrent actions. The focus and alignment systems can be used in dry conditions, while wafer exposure is done under wet conditions. Fluid containment is based on the so-called showerhead configuration, with the fluid contained locally under the projection lens. Resolution capabilities support 70nm half-pitch resolution. In 2004, ASML shipped several systems to customers. This year, the company has shipped next-generation immersion systems, called the XT:1400i, with 0.93NA ArF lenses. ASML has also recently announced expected shipments of the 1.2NA XT:1700i systems in the first half of 2006. The XT:1700i targets resolution of <50nm half pitch. Meanwhile, new high-index fluids being researched may extend ArF lithography even further.
Evaluating immersion’s results
At this time, several XT:1250i 0.85NA immersion systems are operating in the field. The system’s lens is based on the dry Starlith1250 optics from Carl Zeiss SMT AG. Due to the specifics of the lens design, only minor readjustments and a special last lens element (modified in thickness to make up for the water film) are needed to make this an immersion lens. For this reason, the lens PMI data are well in line with the latest data obtained on dry 0.85NA lenses, demonstrating that immersion lenses are of the same high quality. As reported earlier this year, the system’s distortion, image plane variation (IPD), and astigmatism (AST) adhered well to dry system specifications; and 70nm imaging performance of both dense and isolated lines was achieved [4].
Although polarization is a resolution enhancement technology (RET) in its own right, the full potential of the technology is exploited in combination with immersion. Initial experiments combining immersion and polarization confirm the theoretical predictions. Both exposure latitude and DOF show significant improvements (see Fig. 1). The greatest advantage of polarized illumination is the reduction of mask-error enhancement factor, which goes side by side with exposure latitude enhancement.
The challenge for photoresist suppliers is adapting their ArF resist for water immersion. Although standard developer-soluble topcoats are available, the industry prefers to work without topcoat. It was demonstrated [5] that ArF photoresist also can be used without topcoat (see Fig. 2). The resist used came from Fujifilm. The current focus of resist development is on developing defect-free, immersion-compatible resists. Close cooperation between the exposure tool supplier, the resist and track suppliers, and the user is necessary for fast development cycles and to ensure an optimum process choice.
 Figure 2. Results of new ArF photoresist exposed without topcoat. The images show 90nm dense lines over 10% dose range. (Data courtesy of Fujifilm, published in [5]) |
An additional challenge for a wet system is to obtain the same overlay as a dry system. Although the dual-stage configuration of the Twinscan immersion system measures the alignment in the same way as a dry system, the fluid film between the lens and the wafer may put additional shearing forces on the wafer stage. The thermal impact of the fluid film on overlay performance also must be minimized. Consequently, steps have been taken to compensate for these immersion-specific phenomena in the XT:1250i. Figure 3 presents the overlay data obtained on an XT:1250i system. It shows that single machine overlay of <10nm can be achieved.
 Figure 3. Overlay measurements on nine wafers exposed on the XT:1250i system. |
null
Launch of 0.93NA immersion tool
Shipment of the XT:1400i to the field started during the first half of 2005. With 0.93NA, the new system shows a DOF gain of a factor of 2 due to immersion. Top-down SEM pictures confirm the DOF gain of the wet system compared with the dry version (see Fig. 4).
Figure 5. shows that also using polarization to enhance imaging, k factors as low as 0.265 can be printed with a large DOF. The data were obtained using an NA of 0.93, dipole illumination with an opening angle of 35°, and a sigma of 0.81-0.97. The polarization in the poles was chosen to maximize the contrast of the dense lines to be imaged. These results clearly show the extension possibilities of ArF lithography in combination with immersion and polarization.
 Figure 5. Top-down SEM pictures of 55nm dense lines exposed with the XT:1400i immersion system, with an 810nm focus range. |
null
1.2NA systems come in 2006
ASML plans to start shipping a fourth-generation immersion system in the first half of 2006. The NA of the XT:1700i system will be 1.2 to support <50nm resolution. The tool will be used down to 45nm half-pitch resolution in flash-memory IC manufacturing. One of the challenges for this system was the lens design. The high NA value and the required tight aberration tolerances tend to make the lenses exponentially more complex. To enable 4× magnification and 26×33mm2 field size at 1.2NA, a catadioptric lens is needed to maintain a reasonably low lens complexity, as well as manageable lens diameters. The increase in lens NA makes polarized illumination very beneficial [6, 7]. For the NA<1 systems, polarization will predominantly be used in R&D, but for the 1.2NA region, polarization will be introduced for high-volume manufacturing.
 Figure 6. a) Representation of a catadioptric lens and b) close-up of a mirror. |
Figure 6 shows the combination of lens elements and mirrors in a catadioptric lens. Carl Zeiss has obtained the required mirror-manufacturing accuracy from aerospace applications and, in particular, by building EUV test tools. Besides the fact that lens-element diameters are smaller than the ones from the XT:1400i, the incidence angles of the optical rays on the lens elements also are smaller. This allows proven polishing and coating technology to be used. The designers focused on finding a solution relying on one optical axis to ease lens adjustment and integration into the scanner body. As a result, standard (refractive lens) mounting technology can be used, making it hard to distinguish this catadioptric lens from its refractive predecessors.
Further extension of immersion
The theoretical limit in NA is the index of water when it is the immersion fluid. The practical limit for lens design is estimated to be ~1.3NA. This would result in 40nm half-pitch resolution with k1 of 0.27. The idea of increasing NA further with high-index fluids has attracted substantial interest [8]. For 193nm lithography, several fluids have been identified with indices varying from 1.53 to 1.64.
In addition to high-index fluids, high-index glass materials are required to enable superhigh-NA lens designs [9]. In time, availability of appropriate fluids and lens materials will determine whether this technology extension will be viable, or whether alternatives such as EUV will take over the lithography roadmap.
Conclusion
Revolutionary steps in 193nm lithography development have been made with the introduction of immersion technology. In less than two and a half years, three generations of immersion tools have been built, tested, and shipped. The DOF increase relative to dry systems has been demonstrated. The system exposes edge dies, and intrafield and full-wafer CD uniformity results show that immersion lithography can deliver the high levels of printing accuracy needed in volume production. Increased processing defects induced by immersion were one of the concerns, but significant progress has been made on particle reduction and bubble elimination.
Next year, a fourth-generation immersion system will be introduced. With 1.2NA, immersion, and polarized illumination, the NA>1 era will begin for expected use in flash-memory chip production down to 45nm resolution.
Using water as an immersion fluid, the resolution limit for ArF is estimated to be 40nm. Research has begun on high-index fluids and glass materials. Timely availability will determine whether the application will be viable and whether ArF immersion will extend below 40nm.
Acknowledgments
Twinscan is a trademark of ASML. Starlith is a trademark of Carl Zeiss SMT AG. This article is based on the work of ASML and Carl Zeiss SMT project teams using Twinscan immersion systems. The authors thank everyone involved.
References
- M. Van Bavel, et al., “Low-k1 Imaging for Contacts and Lines Using Immersion ArF,” Solid State Technology, Vol. 48, No. 2, p. 39, Feb. 2005.
- D. Gill, et al., “Immersion Lithography: New Opportunities for Semiconductor Manufacturing,” J. Vac. Sci. Technol. B 22(6), Nov. 2004.
- J.H. Chen, et al., “Characterization of ArF Immersion Process for Production,” SPIE Vol. 5754, 2005.
- J. Mulkens, et al., “Immersion Lithography Exposure Systems: Current Capabilities and Tomorrow’s Expectations,” to be published in SPIE Microlithography 2005.
- B. Streefkerk, et al., “Extending Optical Lithography with Immersion,” Proc. SPIE Vol. 5377, 2004.
- D. Flagello, et al., “Challenges with Hyper NA Polarized Light Lithography,” to be published in SPIE Microlithography 2005.
- C. Kohler, et al., “Imaging Enhancements by Polarization Illumination: Theory and Practice,” Proc. SPIE Vol. 5754, 2005.
- B. Smith, et al., ISMT Symp. on Immersion Lithography, 2004.
- J. Burnett, et al., ISMT Symp. on Immersion Lithography, 2004.
Ron Kool received his PhD in mechanical engineering from Delft U. of Technology. He is a director of product marketing at ASML Netherlands B.V., De Run 6501, 5504DR Veldhoven, The Netherlands; ph 31/40-268-4017, fax 31/40-268-4477, e-mail [email protected].
Christian Wagner received his PhD in quantum optics from Universität München. He is a product manager at ASML.
Reiner Garreis received his PhD in theoretical physics from the U. of Karlsruhe. He is a senior manager for the lithography optics division of Carl Zeiss SMT AG.