High-index materials research key to extending immersion lithography
02/01/2008
EXECUTIVE OVERVIEW
The adoption of water-based immersion lithography into 45nm half-pitch processes is in full swing, but if optical lithography is to continue to dominate leading-edge semiconductors, some advances are needed. In this article, we explore extensions to immersion lithography including high-index immersion and multiple exposure lithography.
|
In the past few years, the lithography industry roadmap has undergone a metamorphosis from logic driven to memory driven. Traditionally, the companies at the leading edge of that roadmap bear the burden of underwriting the development of those new technologies, but pricing wars and the attendant drops in asking sales prices have forced the creation of novel business models to enable the continued development of lithography solutions. The resulting message to lithography tool makers is a cacophony of individual, specialized requests rather than a unified roadmap. When asking the question “will technology X be ready in time?” the questioner must also specify for which customer base: DRAM, Flash, or Logic.
The lithography landscape for the near future is now much more complicated. Complexity inevitably makes semiconductor manufacturers, lithography tool makers, and research consortia uncomfortable. It dilutes resources and reduces focus, forces groups with research dollars to make difficult decisions without adequate time or results, and often leads to “betting big” on one horse and hoping it wins (or at least finishes the race). In a word, complexity in a strategic roadmap is “bad.”
At the 4th International Symposium on Immersion Lithography in October 2007, semiconductor manufacturers from around the world stated that immersion lithography is currently being used as their leading-edge patterning technology [1-3]. Additional manufacturers have expressed plans to use it in upcoming process technologies [4]. Technical hurdles such as throughput, cost of ownership (CoO), and defectivity, all seem to have been overcome or driven to levels that are equivalent to dry 193nm lithography. In short, water-based immersion lithography has been accepted by the industry and is healthy and dominant.
With immersion technology essentially stable and robust, attention from the industry is now centered almost exclusively on how it will either evolve or work in tandem with complementary techniques to enable extension to the 32nm half-pitch node. The primary path being discussed is the use of immersion plus double-patterning to enable these tight pitches. A more attractive, but more challenging approach, is to use high-index immersion lithography to allow a single-exposure solution.
In both cases, though for different reasons, economics is the primary barrier. Any double- patterning solution will involve a significant CoO increase due to mask price increases and longer cycle times resulting from “extra” processing steps. High-index immersion is still in the research phase, and in order to achieve manufacturability (if this is possible at all) will require a modest capital investment to solve materials problems and commission first-generation tooling.
Is high-index immersion lithography going to become a reality? The answer to this question lies in the availability of both high-index lens and high-index fluid materials. For the former, the best candidate lens material (lutetium aluminum garnet, LuAG) is nearly achieving most of the specs required for production with the significant exception of absorbance [5]. The final decision for the success of the lens material will be made in the first half of 2008. The fluid situation is better, with organic materials of n(193nm) = 1.64 currently available. The combination of these fluid and lens materials would enable NA = 1.55 tools and, together with double patterning, would provide a 22nm half-pitch lithography solution.
The primary goal of Sematech’s Immersion Lithography Program is to enable, or unambiguously eliminate from consideration, immersion lithography solutions for future technology. Sematech has directed research projects in high-index lens, fluid, and resist development for high-index immersion throughout the year and has met with a broad spectrum of successes and disappointments. In the next section, these developments are discussed in detail.
High-index materials research
Sematech has led the world effort into high-index immersion materials research by supporting lens, fluid, and resist development teams across three continents. The top priority has been to determine if a suitable lens material can be manufactured with production-quality specifications. At the end of 2006, after a comprehensive NIST study, the “best” candidate material was identified as (LuAG) [6].
|
Schott Lithotec, a prominent German optical materials company, commenced a three-stage development program in 2006 (Table 1). The first phase of this work, the so-called “green” phase, was aimed at demonstrating feasibility of the materials by achieving an intermediate absorbance target of 0.05/cm. This phase was in progress when Sematech contracted with Schott to fund the second or “yellow” phase of development aimed at creating 80mm diameter single crystal boules of material simultaneously meeting intermediate absorbance, homogeneity, and stress birefringence targets. When key milestones from the green and yellow phase were completed, work would commence on the third or “blue” phase aimed at making production quality final lens elements.
|
Schott has made great progress in crystal growth techniques for LuAG. They have installed and created growth processes for both Czochralski and Bridgman furnaces. On the Czochralski side, Schott developed internal expertise in achieving “flat interface” growth first with YAG materials and then demonstrated the technique with LuAG. Their greatest success was in growing a 280mm × 80mm boule of LuAG with a flat interface. Additionally, their ability to achieve low stress birefringence and good homogeneity with YAG crystals is approaching the required specification (Table 2). Schott is confident this learning can be transferred quickly to LuAG.
Absorbance, however, continues to be the greatest issue with LuAG development. Improvement in this area has been governed by the “peeling of the onion” of materials impurities. To grow LuAG, Schott requires extremely pure LuO2 and Al2O3. Significantly, key impurities that absorb strongly at 193nm must be removed to better than parts-per-million levels. Progress in eliminating contaminants has been slow but steady. Crucibles, furnace components, and raw materials have each undergone steady improvement. The single largest contributing factor is impurity in the Al2O3.
The difficulty in meeting the absorbance targets set for the green phase is leading to a delay in the start of the blue phase. The best case delivery of a first production quality last lens element (LLE) to lithography tool makers is now 4Q09, provided that the raw materials purity problem is solved in the first half of 2008.
High-index fluid development
High-index fluid availability is encouraging as Generation 2 (G2) fluids with n(193nm) = 1.64 are achieving increasingly longer lifetimes in test setups resembling normal manufacturing conditions [7, 8]. Generation 3 (G3) fluids with n(193 nm) = 1.8 are, however, still entirely in the research domain. Sematech’s research programs have been leading the way in this area. Considerable progress has been made using an approach that adds nanoparticles of metal oxide to water to form a nanocomposite fluid [9]. Another approach involving the synthesis of a highly dense aromatic carbon material is also being investigated [10].
Sematech’s sponsored research at Columbia and SUNY Albany has evaluated a number of groups of organic materials and eliminated them from consideration for G3 fluids. Organic compounds using sulfur, phosphoric, or ionic salts as routes to increase the index of refraction have been eliminated due to prohibitively strong absorbance at 193nm. A detailed study of aromatic carbons sponsored by Sematech at Columbia University in 2006 [11] showed a strong correlation between refractive index and density in aromatic carbon compounds. This suggested that the best chance for a high-refractive index organic solution would be to synthesize cubane, the densest aromatic compound.
A year-long project at Columbia has resulted in the creation of about 10mL of methylcubane, a liquid cubane derivative. Unfortunately, the measured absorbance of highly purified methylcubane showed this material to be too dark for use as an immersion fluid. Although it is not possible to “prove” a negative, the weight of evidence from Sematech’s comprehensive study of candidate immersion fluids strongly suggests that there is no single-compound organic solution to the G3 fluid problem.
A more promising research avenue is being pursued through Sematech-sponsored work at Cornell. There, researchers have made a breakthrough in obtaining 3-5nm spherical nanoparticles of a metal oxide with an index of refraction of 2.9 at 193nm, meaning that if mixed with water, it is possible to create an aqueous solution with n(193 nm) = 1.8 with ~37 volume % (vol %) of nanoparticles. The volume fraction required to achieve a more modest G2-like index is ~22 vol %.
Solutions with 10 vol % have been achieved. Problems facing the research team include increasing the loading to 37%, eliminating organic contaminants remaining from the synthesis, improving the monodispersity of the nanoparticle yield, and reducing the diameter of the nanoparticles. Furthermore, Sematech is attempting to increase the quantity of nanoparticles grown in order to support an environmental health and safety study of the nanoparticles themselves, including techniques for remediation from organics (such as photoresists) or aqueous solutions containing these nanoparticles.
The use of nanoparticles in fluid systems is the best remaining approach for yielding a Generation 3 fluid. Perhaps of even greater significance, however, is the potential for making these high-index immersion fluids, both G2 and G3, using aqueous solutions. If these can be made to work, the move to high-index immersion will be seen by lithography tool makers and semiconductor manufacturers alike as truly an incremental and evolutionary change from today’s water-based systems.
In all cases, high-index immersion fluids will almost certainly be much more viscous than water [12, 13]. The implication for lithography tool manufacturers is that new fluid handling solutions will be required to maintain both throughput and defect performance. Fortunately, ASML, Canon and Nikon have discussed novel approaches to the so-called “low contact angle fluid” regime [14-16]. Nikon’s “local fill” nozzle approach does not rely on surface tension to maintain throughput, and its “large puddle” concept would also permit continued high throughput operation. ASML’s “wet wafer” strategies take a different approach. In this method, high-index fluid material is allowed to remain in a “trail of droplets” behind the puddle and is subsequently “cleaned” from the wafer prior to post-exposure bake. These ideas are in the development phase, but tool makers have stated clearly that high-index immersion fluids will not force a reduction in tool throughput relative to water-based immersion tools.
High-index resist development
The final component of the high-index materials program at Sematech is research into creating resist systems with index greater than 1.9. This target value is arbitrary as the higher the n(193nm) of the resist, the better the improvement in overall depth of focus that resist process will show. Also, it is critical to note that high-index resists are only strictly necessary in the case where a G3 immersion fluid is in use.
Sematech-sponsored research at the University of Queensland [10, 17] has shown n(193nm) = 1.9 material using sulfur addition to the polymer backbone. Unfortunately, above about n(193nm)=1.76 such materials become too absorptive and no longer image well. The conclusion from this work is that sulfur-based approaches will not yield a viable high-index resist.
In a happy coincidence, however, the nanoparticles from the Sematech sponsored project at Cornell University originally created to enable G3 fluids may, in fact, provide the answer to the high-index resist problem as well. An organic film containing these nanoparticles has been spun by researchers at IBM’s Almaden Research Center with n(193nm) = 1.9 and measured an absorbance of 2.5/µm. [26] The practical limit for a photoresist is ~4/µm, so the result indicates this approach is viable. Although this film is not a complete resist system, the basic result is extremely encouraging and suggests that imaging nanocomposite resist systems is a viable approach.
Extending optical lithography through multiple exposures
The lithography solution for the 32nm half-pitch node is double patterning plus water immersion. The reason is simply that this combination of technologies is the only solution that exists today. Can semiconductor companies afford the double-patterning solution? The answer is that companies including Samsung [18] and Intel [4] are using the technology right now. Numerous other companies have presented various double-patterning process flows that are in various stages of development. Combined with progress in lithography tool overlay capability [19-21], the future for double-patterning technology seems bright.
Work is also proceeding to find materials that would enable “two exposures/one etch” lithography, so-called “double-exposure materials.” Fuji [22] and Sematech [23, 24] both have active research programs investigating novel approaches to create nonlinear photoresist systems. A number of other resist companies are investigating “resist freezing” techniques as well.
The Sematech double-exposure materials program is a collaborative effort between the University of Texas, Columbia University, and a number of resist manufacturers. The project has produced seminal work in categorizing the types of approaches that are available for “two exposures/one etch” lithography [25]. The project has identified two approaches for concerted study: intermediate stage two-photon materials and optical threshold materials employing a photo-activated diffusion valve. Both concepts are novel and are in the early research phases at this time [24]. Because of the radical nature of the chemistries involved, it will likely be some time before high quality imaging is available from these materials.
The importance of double-exposure material research is the potential for cost reduction over double-patterning solutions that rely on “two exposures/two etch” process flows. The cycle time added by sending a wafer out for etch can be significant. For example, a typical 32nm process might contain n “critical layers” that must make use of double-patterning; this becomes 2n layers in the “double-patterning” process flow. If each new layer requires a metrology, etch, clean, inspection, and possibly deposition step prior to returning to lithography for the “second” patterning step, it is reasonable that such processing might occupy as much as one to two days of cycle time, if not considerably more. The conclusion is that critical layer double-patterning might add as much as a week (for n = 4) of cycle time to a process flow, a 10-20% increase compared with typical processes.
Another key challenge for double patterning is the need to realign the wafer, a task that adds to the complexity of maintaining overlay in an already difficult environment. Double-exposure material solutions allow the wafer to remain chucked for both mask exposures, eliminating a large source of alignment error.
Immersion lithography strategic outlook
Immersion lithography and double patterning are the highest resolution tools currently available to lithographers. At the 4th International Symposium on Immersion Lithography, many attendees expressed grave concerns about the timing of various solutions for the 32nm half-pitch node and beyond. Timing, however, means different things to different companies. DRAM, Logic and NAND device makers have diverged in their roadmaps due to a variety of technical and economic considerations.
The critical question of our time is, “What technology will be used for the next technology node?” For leading-edge resolution companies, i.e. NAND makers, the answer for the 32nm node is double patterning plus water immersion because it is available. With the early adopters from the NAND community choosing immersion plus double patterning, the question of extending to the 22nm half-pitch node becomes critical.
|
According to Table 3, a double-patterning + high-index solution could be obtained for 22nm half-pitch with the existence of a 1.55 NA high-index tool. Such a tool would require a LuAG lens and a G2 fluid. As mentioned previously in this report, a decision on LuAG capability will be made in 1H08. At present, however, no lithography toolmaker has created a full-size high-index immersion engineering tool that incorporates a G2 fluid with a high-index-ready projection optic.
If LuAG proves to be a viable lens material, the next step for the immersion lithography community is to create a test machine that incorporates the high-index lens and fluid materials and operates at high scan speed and low defectivity. This proof-of-concept machine could be funded by a semiconductor manufacturer, a consortium, or a lithography tool maker. Alternatively, some combination of stakeholders might be possible.
Conclusion
Immersion lithography extensions, including high-index immersion, provide opportunities for patterning solutions to the 32nm, 22nm, and perhaps even the 16nm half-pitch nodes. These solutions are not automatic or even guaranteed and will require research and development to become reality.
The situation today for immersion lithography is clear: if the high-index lens material absorbance dilemma is solved by 1H08, then development scanners with NA = 1.55 will become available in 2011. For semiconductor manufacturers with aggressive pitch plans, high-index immersion lithography will not be available in time for 32nm node single-exposure solutions. For those companies planning to use double-patterning for the 32nm node, the probable insertion point for high-index immersion lithography will be as a 22nm half-pitch node double-patterning solution. For companies with less aggressive pitch roadmaps, 32nm single-exposure solutions may be an option.
Lithography history for the past few decades has been filled with declarations of the impending end of optical lithography. Sematech continues to believe that high-index lithography and other extensions to optical lithography merit full support from the semiconductor industry as the alternatives for the 22nm half-pitch node and beyond remain high risk.
Acknowledgments
The author wishes to acknowledge the critical inputs of Andrew Grenville, Michael Lercel, Chris Van Peski, and Paul Zimmerman. Finally, Jeff Byers’ contribution cannot be overstated in enabling Sematech’s double-exposure program. He will be sorely missed.
References
- L.J. Chen, et al., “Immersion Effect on Across-wafer Focus Performance,” 4th International Symposium on Immersion Lithography, Keystone, CO, 2007.
- T. Kono, et al., “Implementation of Immersion Lithography to NAND/CMOS Device Manufacturing,” ibid.
- J-Y. Toon, et al., “Studies on the Defectivity of 193nm Immersion Lithography for Sub-50nm Node and Beyond,” ibid.
- Y. Borodovsky, “Power of One,” 2007 MNE Conference, Copenhagen, Denmark.
- L. Partier, et al., “Development Progress of High-refractive LuAG for Hyper NA Immersion Systems,” 4th International Symposium on Immersion Lithography, Keystone, CO, 2007.
- J. Burnett, et al., “High-index Materials for 193nm Immersion Lithography,” 3rd International Symposium on Immersion Lithography, October 2006.
- R. French, et al., “Cost-effective Single-exposure Immersion Lithography with Second-generation Immersion Fluids for Numerical Apertures of 1.55 and 32nm Half Pitches,” 4th International Symposium on Immersion Lithography, Keystone, CO, 2007.
- T. Furukawa, et al., “High-refractive Index Fluids and Top Coat Design for Next-generation ArF Immersion Lithography,” ibid.
- E. P. Giannelis, et al., “High-refractive Index Nanoparticle Fluids for 193nm Immersion Lithography,” ibid.
- P. A. Zimmerman, “Challenges for Development of High-index Fluids and Resists for Generation-3 193nm Immersion Lithography,” ibid.
- J. Lopez Gejo, et al., “Outlook for Potential Third-generation Fluids,” Proc. of SPIE, Vol. 6519.
- J. Hoffnagle, et al., “Fundamental Analysis of Nanoparticle-based High-index Immersion Fluids,” 4th International Symposium on Immersion Lithography, Keystone, CO, 2007.
- D. Sanders, et al., “New Materials for Surface Energy Control of 193nm Photoresists,” ibid.
- H. Sewell, et al., “High-n Immersion Lithography,” ibid.
- K. Sakai, et al., “Feasibility Study of Immersion System Using a High-index Fluid,” ibid.
- Y. Ohmura, et al., “Development Progress of High-index Immersion Lithography,” ibid.
- A. K. Whittaker, et al., “High-refractive Index Resists for Generation 2 and 3 193i: Where Are We and Where Can We Go?” ibid.
- H.C. Kim, Samsung comments during Double Patterning Panel Discussion, ibid.
- C. Wagner, et al., “Immersion for 40nm Production with 1.35NA,” ibid.
- A. J. Hazelton, et al., “Immersion Lithography in Mass Production: Latest Results for Nikon Immersion Exposure Tools,” ibid.
- M. Kobayashi, et al., “FPA-7000AS7: An Immersion Exposure System for 45nm Node Mass Production,” ibid.
- S. Tarutani, et al., “Development of Materials and Processes for 32nm Node Immersion Lithography Process,” ibid.
- J. Byers, et al., “Double-exposure Materials: Simulation Study of Feasibility,” Photopolymers Japan 2007.
- J. Byers, et al., “Update on Double-exposure Materials Development,” 4th International Symposium on Immersion Lithography, Keystone, CO, 2007.
- G. Willson, et al., “Double Exposure vs. Double Patterning,” NEO Litho Workshop, September 2007.
- R. Sooriyakumaran, et al., “High-refractive Index Polymer Pplatforms for Use with 193nm Immersion Lithography,&rdwquo; 4th International Symposium on Immersion Lithography, Keystone, CO, 2007.
Bryan J. Rice received his BS in physics and MS in computer science from The Georgia Institute of Technology in 1991 and his PhD in nuclear physics from Duke Universtiy in 1998. He is the immersion lithography program manager and an Intel assignee to Sematech, Austin, TX, US; ph 503/703-6336, e-mail [email protected].