Formulating materials that are compatible with high-performance immersion lithography

Karen Petrillo and Kaushal Patel, IBM; and Mark Slezak and Gary Dabbagh, JSR Micro

The use of immersion lithography asa technique to enhance resolution and depth of focus was described as early as 1987, but it moved to the forefront for 45- and possibly 32-nm-node patterning options with the publication of the latest roadmap.^1,2 Recently, the technology's viability as a manufacturing tool for 90-nm devices as well was demonstrated by two firms. IBM built a critical level of a 64-bit power processor, while TSMC fabricated electrically functioning SRAM chips.^3,4 Both companies maintain that the process yield was overlay, resist materials, and water handling.

While it is clear that immersion lithography is moving rapidly toward the production line, many challenges related to resist materials and processes remain to be addressed to realize its potential advantages and to prevent defects. Potential defect sources associated with the process are numerous. Bubbles and particles may be created during imaging, material residues may remain following the removal of topcoats, and image collapse may be caused by swelling of the resist from water infusion.

One of the most important challenges to be overcome is the leaching of resist constituents into the immersion fluid. Resist extraction studies have shown that both the photoacid generator (PAG) and photogenerated acids are present in water after 30 seconds or less of contact.^5–8 Although recent modeling data suggest that low levels of PAG components may not contaminate the scanner lens, there are still concerns that PAG leaching will affect resist performance by changing dissolution rates and resist profiles and by degrading CD control.^9,10 Consequently, decreasing the amount of extractable products is a priority.

Currently, the application of a topcoat is the primary approach to preventing extraction products from entering the immersion fluid. Such coatings have been shown to be effective.¹¹ They can also be engineered to be antireflective and have the added advantage of preventing airborne contaminants from reaching the resist surface, thus mitigating postexposure delay effects.^12,13 Topcoat design is critical in other respects as well. Because the resist profile and performance must be maintained, topcoat-resist interactions must be avoided. The added layer of material also raises concerns about additional defects. Ultimately, the goal is to eliminate the need for topcoats by developing materials that resist leaching and contamination, but in the meantime topcoats are an important part of immersion processing.

To gain an understanding of how various resist chemistries and formulations behave under immersion conditions, the study reported in this article compared the lithographic performance of resists containing different polymer platforms, protecting groups, and formulations under immersion and dry process conditions. The compatibility of several developer-soluble topcoat materials with a variety of resists was also studied from the standpoint of profile control issues and defect reduction. In addition, parallel efforts focused on understanding and eliminating defect sources.

Experimental Procedure

Materials. The comparison of the lithographic performance of various resists was accomplished using polymers, PAGs, and formulations supplied by JSR Microelectronics (Sunnyvale, CA). Methacrylate polymer platforms with high-, low-, and hybrid-activation-level protecting groups were used in the immersion formulations, and the resists also contained two sulfonium PAGs in combination with one of four solvent mixtures. In all cases, the primary solvent was propylene glycol monomethyl ether acetate (PGMEA), while the different cosolvents used were gamma-butyolactone (GBL), ethyl lactate (EL), and cyclohexanone. The compatibility of several developer-soluble topcoat materials with a variety of resists was also studied using three materials supplied by JSR. The bake process for all topcoats was 90°C for 60 seconds.

Equipment. Coating was performed on one of two tracks. An ACT 12 track from TEL (Kumamoto, Japan) was used for experimental resist coatings in both tool- and hand-apply modes. Using a chemically filtered FOUP, the sample wafers were then transported to an immersion tool cluster for exposure, postexposure bake (PEB), and development. For comparison testing, a TEL Lithius track in a linked configuration with a 0.75-NA 1150i immersion scanner from ASML (Veldhoven, The Netherlands) was used with several commercially available photoresists, antireflective coatings, and topcoats plumbed in. Critical dimension (CD) measurements were done using a VeraSEM (Applied Materials; Santa Clara, CA), while an Applied Complus defect inspection tool was used for defect analysis in conjunction with an Applied G2 SEM Vision for E-beam inspection and EDX analysis. Leaching experiments were performed using an apparatus developed in-house that has been described elsewhere.5 The resulting samples were then analyzed for trace amounts of perfluorosulfonates using liquid chromatography/mass spectrometry/mass spectrometry by Exygen Research (State College, PA). The detection limit for this technique is 0.2 ppb.

Experiments. The development of high-performance immersion-compatible materials requires a four-step approach. The first step involves protecting the immersion lens from leaching materials by defining component-leaching specifications that are conservative enough to ensure a stable and lasting lens quality but not so stringent as to prevent valuable experiments from being performed. Such specifications were determined by evaluating previous stray-light test results with respect to the number of wafers run in the lithography tool during a specific period of time in conjunction with the experimental results of resist-leaching tests.

The second step was to identify developer-soluble topcoats that are effective as leachant barriers. Several materials using tetramethylammonium hydroxide (TMAH) as the developer showed excellent barrier capabilities, and one was chosen for use in evaluations with experimental resist formulations.

Step three involved optimizing commercial resists and processes for use with metal- and gate-level applications so that defectivity learning related to tooling, materials, and processes could proceed in parallel with new resist studies.

In step four, materials research and development focused on 45-nm-node lithography without topcoats, comparing results for various polymers, protection levels, and solvents.

Leaching Evaluations

Setting a Specification. The initial component-leaching specifications used in this study were determined by examining historical data on the number of wafers previously exposed on the lithography tool, the resist and topcoat combinations used during those exposures, and the stray-light data collected for the same period of time. If extractable materials migrating to the lens surface had either been deposited on or attacked the lens surface during those runs, the stray-light tests would have provided evidence of degradation. The historical data showed that the level of stray light had been constant during the exposure of >11,000 wafers, indicating that the level of leaching in the resist and resist-topcoat combination was sufficiently low to prevent lens degradation.

When the amount of extractable components in resist-film samples was quantified, it was found that the level of extricated perfluorosulfonates was ~12 ppb when the resist was used without a topcoat and <0.2 ppb when a topcoat was used. Since the exact ratio of wafers with resist alone compared with wafers with resist and a topcoat was unknown, a conservative approach to specifying leaching levels was employed. The initial acceptance level of extractable perfluorosulfonates was set at 12 ppb. If an experimental resist film exhibited >12 ppb of extractable material, a topcoat was required to bring the level down to 5 ppb or less. For resists in the 5–12-ppb range, it was considered safe to expose a limited number of wafers. Additional information revealing the effects of water-soluble resist components on lens performance is expected through experimentation and modeling and will be taken into account in future specification adjustments.

Extraction Tests. The in-house technique employed for extracting resist components from coated films involves depositing a known volume of water on the resist film and subsequently collecting the water for analysis after a specific contact-residence time. When 10 different commercially available resists were evaluated for leaching, significant variations of extracted perfluorosulfonates were observed. Figure 1a shows the variation among these samples, which yielded extraction levels ranging from approximately 10 to >70 ppb.

Figure 1: Comparison of the results of PAG extraction tests of 10 commercially available resist films: (a) perfluorosulfonate extraction levels of resists without topcoats and (b) perfluorosulfonate levels of resists used with topcoat TC-2 (green bars). Topcoat efficiency is represented by the line.

The amount of variation was not surprising because the resists evaluated contained a cross section of polymers, protecting groups, PAGs, and solvents. With the application of topcoat TC-2, there was a significant reduction in perfluorosulfonate extraction, as shown in Figure 1b. This topcoat's efficiency at reducing leaching varied with the resist, but it was >90% in all cases. Many topcoated samples had levels below 1 ppb, and all were below 3 ppb. With respect to leaching specifications, all of these resist and topcoat combinations would be considered acceptable on the immersion scanner.

Figure 1 also reveals inconsistencies in the topcoat's ability to prevent leaching from sample to sample. While resist C had the highest level of perfluorosulfonate extractables when used without a topcoat, there was a 99% reduction in extractable components when the topcoat was used. In comparison, while the untopcoated resist H had fewer extractables (approximately 30 ppb) than resist C, that level was reduced by only 90% when it was used with the topcoat, resulting in a resist-topcoat stack with a perfluorosulfonate content of 3 ppb, the highest of any topcoated resist. These inconsistencies must be understood in order to progress toward more-efficient topcoats and ultimately a topcoat-free process.

A matrix of experimental polymers with varying activation levels and cosolvents, along with a commercially available resist, was also evaluated in terms of perfluorosulfonate leaching. In this test, PAG chemistry and loading were kept constant for all polymers, and postapplication bake (PAB) temperature was set at 110°C for 1 minute. The two high-activation-level polymers and the two hybrid-activation-level polymers had similar leaching values, while two of the three low-activation-level polymers showed significantly greater amounts of extractable materials. A comparison of PAG extraction with the hydrophobicity of the polymer as determined by contact angle revealed no correlation.

In another set of tests, perfluorosulfonate extraction from polymer F was evaluated using various solvent systems. Again, the PAB temperature was kept constant at 110°C for 1 minute. The solvent systems evaluated included PGMEA with cosolvents of GBL, EL, and cyclohexanone. The results indicated that the use of the cosolvents resulted in minor variations in contact angle and a slight indication that increasing hydrophobicity yields lower leaching levels. These results suggest that the most important factors in controlling leaching are the polymer and its activation level. Once an optimized polymer and protecting group are chosen, additional reductions in leaching may be possible with the judicious choice of cosolvents that increase hydrophobicity.

Topcoat Performance

For use in immersion lithography, topcoats must meet multiple requirements, including efficiency as a barrier against leaching, low defectivity, and transparency at 193 nm. In addition, they must not intermix with the photoresist, affect resist profiles, or result in poorer resist performance than dry lithography. These characteristics must be combined with high dissolution rates in TMAH developers and insolubility in water.

To determine topcoat efficiency as a leaching barrier, Figure 2 compares three formulations used with experimental polymer A, which had exhibited a perfluorosulfonate content of ~24 ppb in previous tests without a topcoat (see Figure 1a). TC-1, an alpha material that had earlier provided sufficient barrier characteristics to warrant further study, exhibited a leaching efficiency of approximately 95%; TC-2, a second-generation topcoat, exhibited superior leachant-barrier characteristics with an efficiency of >98%; and TC-3, an experimental material under development, decreased PAG extraction into water by 96%.

Figure 2: Comparison of the results of PAG extraction tests using resist A and three topcoats. Perfluorosulfonate content is represented by the green bars and topcoat efficiency is represented by the line.

An examination of resist profiles revealed that although TC-1 worked well with some resists, it caused tee-topping in others. It was hypothesized that such profile changes would be avoided by increasing the acidity of the topcoat. When an acid moiety was incorporated into the polymer of TC-2, the profile changes were eradicated, as shown in Figure 3.

In order to develop suitable topcoats for immersion lithography, it is important to understand how materials and material-tooling interactions contribute to defect levels. The effects of the three topcoats on defects in blanket films were first evaluated without imaging or exposure to the immersion fluids. As seen in Figure 4, defect levels varied after PAB, with TC-2 yielding the lowest number of defects. After a PEB cycle was performed under conditions compatible with the underlying resist, the topcoats were removed during a 60-second develop step and another set of defect counts was taken. In that test, TC-1, the alpha topcoat material, yielded a large number of defects, most of which were polymer blobs. Both TC-2 and TC-3 caused significantly fewer defects than TC-1, with TC-3 displaying the fewest.

Figure 4: Comparison of defect counts in blanket films using three topcoats. Counts were measured both prior to exposure to immersion fluids (shown as bars) and after PEB and developer (shown as points).

Blanket-film studies cannot give a complete picture of defect generation in immersion lithography. Patterned films are required to learn about material and tooling interactions. Therefore, the next step in this study was to evaluate defect levels using patterned films and various resists and topcoats.

Tests were performed using resist B with topcoat TC-2. Imaging involved a metal-level mask from a 90-nm-node product. Figure 5 shows the resulting defect counts by category from four imaged wafers. There were approximately 20 total counts on each wafer, most of which were particles. Such particles had not been present on the blanket films and were found on top of the resist after topcoat removal and development, suggesting that they were tooling-related. The particles could have been generated from a combination of sources, including the water supply, degraded track or scanner components, or contaminated process bowls or supply lines. Many postlithography particle defects are closely related to missing-pattern defects because a similar generation process is involved. A particle that blocks the light during exposure might be removed later either by the immersion head or during the topcoat removal and development process, leaving a missing pattern.

Figure 5: Characterization of patterned defects on four different wafers using resist B with TMAH-soluble topcoat TC-2.

In other tests, immersion bubble defects had been abundant when a hydrophobic solvent–soluble topcoat was used, but these defects were much less frequent when the more-hydrophilic topcoat TC-2 was used. In the test illustrated in Figure 5, no immersion bubbles were observed. Micrographs of various defects are shown in Figure 6. EDX analysis of the particle defect in Figure 6a revealed carbon, oxygen, and silicon, making it indistinguishable from the resist and substrate. The missing-pattern defect in Figure 6b was formed when a large particle masked the resist during exposure but was later removed during the developing process. In some cases, particles are not removed during development, as illustrated by the defect in Figure 6c. EDX analysis of this defect revealed carbon, oxygen, silicon, and fluorine.

Figure 7: Depth of focus on a line-and-space feature processed using (a) immersion scanner and (b) dry scanner.

Resist Performance. The major advantage of immersion lithography over dry lithography is that it improves the depth of focus. Figure 7 illustrates the 2X improvement achieved in focus latitude for a 90-nm line-and-space feature when a commercially available resist was used with TC-2 topcoat. By verifying immersion performance, the study progressed toward developing new resist polymer formulations for 90-nm metal-level applications. An evaluation of resist performance was conducted using the seven polymers and four solvents studied in the leaching tests. Bake conditions typical for the polymers' activation levels were chosen, and the evaluation criteria concentrated on mask error factor (MEF) performance for dense and semi-isolated features and common process windows for dense features and semi-isolated and isolated spaces and lines.

Table I shows the performance of the seven polymers with solvent system C (PGMEA and GBL) compared with the commercially available resist G. Figure 8 presents micrographs of features created using polymer A. Without undergoing optimization, several of the experimental materials exhibited excellent performance characteristics that were on a par with the commercially available resist. With some fine-tuning, it is expected that the experimental formulations will perform better than commercial resists formulated for dry lithography.

Figure 8: SEM cross sections of features formed using polymer A, with various pitches and CD targets.

Resist performance was also evaluated using different solvent systems with polymer F. The results, shown in Table II, indicate that while differences in MEF and process window performance did arise, those differences were difficult to distinguish from problems encountered during coating. When solvents B and D were used, there were problems with severe pullback on the antireflective coating, while with solvent C topcoat uniformity was a serious issue. A detailed evaluation is needed to better understand these phenomena.

Table II: Comparison of the performance of polymer F with various cosolvents.

A similar comparison focusing on MEF performance and process windows was performed using a gate-level mask. Those results are presented in Table III. As was the case in the evaluation of metal-level materials, the bake processes were not optimized, leaving room for improvements. In addition, photospeeds were slower than anticipated, which may be improved with some formulation adjustments. Generally, MEF was higher for the low-activation-level polymers than for those with high or hybrid activation levels. In contrast to dry lithography, the process window enhancement in immersion lithography was less substantial at the gate level than at the metal level, with the formulation containing polymer A being comparable to dry exposures on a 90-nm product.

Conclusion

Immersion lithography promises to achieve a substantial improvement in the depth of focus. To move the technique into manufacturing, however, it will be necessary to develop materials and processes that will not degrade lens performance over time as a result of component extraction in the immersion fluids. In the study reported in this article, historical data were used to determine a specification for leaching components. Then several commercially available resists and six experimental polymers with various activation levels and cosolvents were evaluated in terms of extractables. The results of that evaluation indicated that component extraction was lowest for the polymers with high activation levels.

The polymers' leaching performance was also evaluated when a topcoat was applied to help contain the extractables within the film. Tests were run using three TMAH-soluble topcoats, all of which were found to be >90% efficient as leachant barriers.

Because the experimental topcoat TC-2 generated low levels of defects and did not affect the resist profiles, it was used in subsequent evaluations of several polymer platforms for metal- and gate-level applications. Lithography characterization included MEF and process window analysis. Test results revealed several promising candidate materials. Process and formulation optimization are needed to enable these candidates to outperform current dry lithography materials. Moving forward to the 45-nm node, the ultimate goal is a topcoat-free immersion process. Further studies including the use of PAGs and additives will be needed to reach that important milestone.

Acknowledgments

This article is a revised version of a paper originally presented at the SPIE International Symposium on Microlithography, held February 28–March 4, 2005, in San Jose. Used with permission. The authors would like to thank Rex Chen, Wenjie Li, Ranee Kwong, Peggy Lawson, Rao Varanasi, Chris Robinson, Steven Holmes, Dario Gil, and Kurt Kimmel of IBM. They also wish to acknowledge Takashi Chiba and Tsutomu Shimokawa of JSR Micro. Thanks also go to Carl Larson and Greg Wallraff for providing the leaching apparatus and procedures and to Robin Keller for her SEM support. The entire IBM Albany team also provided valuable process and SEM support. In particular, the authors would like to thank Jon Orth, Dan Kraft, Ed Couillard, Rich Conte, and Carol Boye. Finally, they gratefully acknowledge the help of Mike Della Selva, John Weeks, Lior Huli, Darren Brookhart, and Jerry Goldberg from the SUNY Albany facilities staff.

References

1. BJ Lin, "The Future of Subhalf-Micrometer Optical Lithography," Microelectronics Engineering 6, no. 31 (1989): 31–51.

2. The International Technology Roadmap for Semiconductors (San Jose: Semiconductor Industry Association, 2004); available from Internet: http://public.itrs.net.

3. P Clark, "IBM Used Albany Immersion Tool to Make Power Processor," EE Times [on-line] (2 December 2004 [cited 1 June 2005]); available from Internet: www.eet.com/article/showArticle.jhtml?articleId=54800014&sub_taxonomyID=.

4. "TSMC Achieves 90nm ICs Using Immersion Litho," Electronic News [on-line] (22 December 2004 [cited 1 June 2005]); available from Internet: www.reed-electronics.com/electronicnews/article/CA489760?spacedesc=latestNews.

5. W Hinsberg et al., "Liquid Immersion Lithography: Evaluation of Resist Issues," in Proceedings of SPIE Microlithography vol. 5376 (Bellingham, WA: SPIE, 2004), 21.

6. C Taylor et al., "Implications of Immersion Lithography on 193-nm Photoresists," in Proceedings of SPIE Microlithography vol. 5376 (Bellingham, WA: SPIE, 2004), 34.

7. RJ LeSuer et al., "Using Scanning Electrochemical Microscopy to Probe Chemistry at the Solid-Liquid Interface in Chemically Amplified Immersion Lithography," in Proceedings of SPIE Microlithography vol. 5376 (Bellingham, WA: SPIE, 2004), 115.

8. S Robertson et al., "Characterization and Meaningful Quantification of Resist Component Leaching into Immersion Fluid" (paper presented at the International Symposium on Immersion and 157nm Lithography, Vancouver, BC, Canada, August 2–5, 2004).

9. G Nellis et al., "Contamination Transport from Wafer to Lens" (paper presented at the International Symposium on Immersion and 157nm Lithography, Vancouver, BC, Canada, August 2–5, 2004).

10. S Kishimura, M Endo, and M Sasago, "Resist Interaction in 193-/157-nm Immersion Lithography," in Proceedings of SPIE Microlithography vol. 5376 (Bellingham, WA: SPIE, 2004), 44.

11. R Dammel et al., "The PAG Leaching Phenomenon in 193nm Immersion Lithography" (paper presented at the International Symposium on Immersion and 157nm Lithography, Vancouver, BC, Canada, August 2–5, 2004).

12. M Slezak, "45nm Node Materials Solutions and Progress" (paper presented at the Anti-Reflective Coatings Symposium, Albany, New York, October 27–28, 2004).

13. W Hinsberg et al., "Influence of Water Immersion on Properties of Lithographic Materials" (paper presented at the International Symposium on Immersion and 157nm Lithography, Vancouver, BC, Canada, August 2–5, 2004).

Karen Petrillo is an IBM development engineer in the lithography processes area working at Albany Nanotech in Albany, NY. She has written numerous conference papers and holds 24 patents. She received a BA from SUNY Oswego, NY, and an MS in quality assurance from California State University at Dominguez Hills. (Petrillo can be reached at 518/487-6455 or kep@us.ibm.com.)

Kaushal S. Patel, PhD, is a development engineer in the lithography materials area at IBM in Hopewell Junction, NY. He has published several papers for journals and conferences and holds one patent. He received a BTech degree in chemical engineering from the Indian Institute of Technology in Mumbai and a PhD in chemical engineering from the Georgia Institute of Technology in Atlanta. (Patel can be reached at 845/894-3373 or kaushalp@us.ibm.com.)

Mark Slezak is a technical manager at JSR Microelectronics in Sunnyvale, CA. He has published 10 papers and holds one patent. He received a BS in engineering from the California Polytechnic State University in  San Luis Obispo. (Slezak can be reached at 408/543-8835 or mslezak@jsrmicro.com.)

Gary Dabbagh is an applications engineer at JSR Micro. Previously, he was a member of the technical staff at Bell Labs–Lucent Technologies. He has published more than 70 technical papers and holds one patent. He received a BA in chemistry from Hunter College in New York City and an MS in organic chemistry from Rutgers University in New Brunswick, NJ. (Dabbagh can be reached at 908/790-9327 or gdabbagh@jsrmicro.com.)