Lithography: The challenges in materials design for 157nm photoresists
03/01/2000
Part two in a series
Kyle Patterson, Mark Somervell, C. Grant Willson, University of Texas, Departments of Chemistry and Chemical Engineering, Austin, Texas
Developing 157nm lithography capability in time for production requires the solution to several daunting materials challenges. Among the most difficult of these is finding a suitable 157nm photoresist. The transitions from i-line to DUV and from DUV to 193nm both required development of new materials because the old ones were opaque at the shorter wavelength. Again, researchers are learning that all existing imaging materials were opaque at 157nm and that they must search for materials capable of balancing the requirements of transparency, etch resistance, and developer solubility. Encouraging progress is being made, but much remains to be done in a very short time.
Existing DUV lithographic technologies and associated optical extensions combined with early 193nm-based systems are expected to provide sufficient resolution to enable the 130nm technology node defined on the 1999 International Technology Roadmap for Semiconductors (ITRS). However, there is no mature lithographic technology to support manufacturing 100nm device generations.
While the transition to one or more of the so-called next-generation lithographic techniques (e.g., projection e-beam lithography such as SCALPEL and Prevail, x-ray lithography, and extreme ultraviolet) will eventually be necessary, they each pose an entirely new set of engineering and infrastructure challenges. The extension of optical lithography to 157nm has therefore garnered widespread support within the past year as it offers the prospect of improved resolution based on experience gathered from decades of optical lithography.
According to the 1999 ITRS, early 100nm-technology node pilot line fabrication should begin the first quarter of 2002 and volume production capability should be in place by the first half of 2005. Considering that the development cycle for DUV photoresist materials spanned at least 10 years and that 193nm resist development took ~8 years (1991-1999), 157nm photoresist development efforts are faced with an extremely abbreviated and challenging development schedule.
Associated areas report progress
Fortunately, efforts to develop 157nm lithography are receiving widespread support. In addition to photoresist development, research in other critical areas is beginning to show significant progress. For example, calcium fluoride (CaF2) is the leading candidate to replace fused silica (absorbance of fused silica increases dramatically at wavelengths below 180nm [1]) as the lens material for 157nm lithography. The quality of CaF2 samples now available for the design of optical elements is extremely encouraging. In another example, several companies have reported recent developments in fluorinated fused silica and have fabricated mask substrates that appear to have the requisite properties for use at 157nm (the use of CaF2 as a mask blank substrate material presents a long and daunting list of challenges). In addition, recent reports on fluoropolymer materials from DuPont have been very promising as a new pellicle material offers sufficient transparency at 157nm, which was considered a potential showstopper at one time. These new materials offering sufficient mechanical properties have been recently reported with absorbance values <0.001µm-1 at 157nm.
An infrastructure for 157nm lithography developmental studies is also beginning to take shape. As you read this article, a 157nm microstepper built by Exitech Ltd. and based on a 0.6NA catadioptric Tropel lens is coming on line at International Sematech (the end of 1Q00). Offering a relatively small 1.5 x 1.5mm field size, the microstepper is designed for early material testing and photoresist evaluations. It will also allow for testing important process-related issues, such as nitrogen purging of the beam path and wafer plane, a technique that is required for the prevention of parasitic absorbance from airborne contaminants such as water vapor that absorb strongly at 157nm.
Single-layer photoresists
A single-layer photoresist process for imaging at 157nm is the clear favorite among potential industry consumers. Important details are beginning to emerge to guide the successful design of single-layer 157nm photoresists. Fundamentally, the design of materials for 157nm photoresists is identical to the approach used in previous lithographic generations. It must adhere to the set of requirements common to all chemically amplified photoresist materials. Specifically, potential materials designed for use in a single-layer photoresist must offer three fundamental characteristics:high transparency at the exposure wavelength; sufficient resistance to plasma-etching processes; and functional units capable of undergoing efficient photochemical transformations that change their solubility in developer solutions.
 |
Figure 1. Typical for developing any new advanced photoresist material, the design of a 157nm photoresist can be broken down into these functional subunits.
For positive-tone photoresists, this is typically accomplished by designing the matrix resin with an acidic functionality that is masked by an acid labile protecting group. Strong acids generated within the exposed regions of the photoresist film catalyze the thermolysis of the protecting groups, rendering these regions soluble in the aqueous base developer solutions [2]. Beyond this, there is a long list of other characteristics the new photoresist materials must have, including accessibility through reasonably simple synthetic schemes, glass transition temperatures above typical processing temperatures, acceptable shelf storage lifetime, and minimum toxicological risk.
It is often convenient to adopt a modular approach in the design of new photoresist materials. This approach would allow the polymer design to be broken down into several functional subunits that may be addressed in a somewhat independent fashion (Fig. 1), as we discuss below within the context of 157nm photoresist work. It is important to note that transparency requirements must be considered in the design of all subunits.
Backbone
The backbone subunit is responsible for tethering the monomer units together and providing the basic mechanical properties for the bulk polymer. The simplest and most common backbone structures based on methylene units are found in photoresist materials such as styrenes and acrylates. Unfortunately, polyethylene, which is comprised exclusively of methylene units, absorbs strongly in the vacuum-UV (VUV) spectral region [3]. To identify possible alternative polymer structures that offer reduced absorbance at 157nm, researchers at MIT Lincoln Labs have begun to survey the transparency of various polymer materials at 157nm. The goal of their research is to establish structure-property relationships between functional groups and the associated absorbance contribution at 157nm.
An important trend uncovered in the course of the work at Lincoln Labs indicates that the transparency of polyethylene can be greatly improved through the use of appropriately placed electron-withdrawing groups, such as oxygen and fluorine atoms. As an example, Teflon AF — a highly-fluorinated analog of polyethylene — is relatively transparent at 157nm. Another analog, poly(vinyl alcohol), also offers improved transparency at 157nm by incorporating electron-withdrawing hydroxyl groups on alternating carbons along the polyethylene backbone.
Another important finding of the Lincoln Labs studies is that silicon-based polymers such as siloxanes and silsesquioxanes offer high transparency at 157nm [4]. This information provides the basis for initial designs of the 157nm backbone.
Etch resistance
The etch-resistant subunit is responsible for decreasing the rate at which the polymer is destroyed by a plasma. The degree to which a photoresist can withstand etch image transfer processes dictates the minimum thickness of the photoresist layer required to protect underlying structures. Empirical relationships such as the Onishi parameter [5] tell us that dry-etch resistance is improved by increasing the relative amount of carbon in a polymer architecture.
In i-line and DUV photoresist materials, aromatic groups offer the requisite amount of carbon to give adequate dry-etch resistance. Since aromatic groups are too strongly absorbing to be incorporated in the design of 193nm materials, alicyclic groups were introduced to replace the benzene rings. These structures provide high carbon content through the use of multiple rings rather than strongly absorbing double bonds.
A similar approach is being taken in the design of 157nm photoresist materials (Fig. 2). However, the absorbance contribution of these structures must be minimized through selective incorporation of electron-withdrawing groups, something that we are diligently engaged in at our lab.
Developer solubility
Solubility in tetramethylammonium hydroxide (TMAH) developer solutions is achieved by designing matrix resins that include acidic functional groups. DUV photoresists are primarily based on poly(p-hydroxystyrene) (PHOST). The hydroxyl group in these resins is sufficiently acidic to render the polymer soluble in basic TMAH solutions. However, not all hydroxyl groups are so acidic.
 |
Figure 2. To give enough carbon content for dry-etch resistance, 248nm photoresists use aromatic benzene rings and 193nm photoresists use alicyclic groups. The structure that will provide dry-etch resistance in 157nm resists is still in question. The answer may be found in alicyclic groups with strategically placed electron-withdrawing groups.
The acidity of phenolic hydroxyl groups such as those on PHOST is explained by a concept from organic chemistry known as resonance stabilization of the anion. The anion can be made more stable by distributing the charge over several atoms in the molecule. In the case of PHOST, the negative charge can be distributed throughout the p-electron system of the adjacent aromatic ring (Fig. 3).
Aromatic rings are very strongly absorbing at 193nm. The base solubility of 193nm resists results from incorporating carboxylic acid functionalities that offer stabilization through resonance as well. In this case, the negative charge is stabilized by distribution between the two equivalent oxygen atoms (Fig. 4). As a result, carboxylic acids are sufficiently acidic but much more transparent at 193nm.
 |
Figure 3. The acidity of PHOST is the result of distributing the negative charge on the anion through the adjacent aromatic ring.
Unfortunately, both carboxylic acids and phenols are strongly absorbing at 157nm. Therefore, new acidic functional groups must be found for designing 157nm photoresist materials. Work at MIT has shown that materials containing any type of p-system are strongly absorbing at 157nm. Simply stated, this makes resonance stabilization unavailable for use in 157nm photoresist materials, so some different method of creating acidic functional groups must be utilized.
 |
Figure 4. The negative charge on the carboxylate anion is stabilized by distribution between the two equivalent oxygen atoms.
Negatively charged atoms can also be stabilized by adjacent functional groups that "pull" charge away through an effect known as inductive stabilization (Fig. 5). Since inductive effects do not require p-electron systems, they hold promise for creating base-soluble groups that can be used in 157nm matrix polymer designs. For example, the hexafluoroisopropanol group is a hydroxyl group that is fairly acidic through inductive stabilization of the conjugate base. The two trifluoromethyl groups adjacent to the alcohol moeity result in acid strengths near that of phenol [6].
Inductively stabilized base soluble groups have been previously used in photoresists. The use of a hexafluoroisopropanol functionality first appeared in a DUV photoresist formulation [7]. The researchers prepared both polystyrene-based matrix polymers and dissolution inhibitors containing these groups to achieve base solubility. This same functional group has also been incorporated into an alicyclic polysulfone matrix polymer for use as a 193nm photoresist [8].
The successful use of this functional group in earlier photoresist polymers combined with its high fluorine content make it an ideal candidate for use in 157nm designs. Our research group has measured the absorbance of analogous structures. They were found to be among the most transparent materials reported at this wavelength.
Protecting groups
Most common protecting groups from earlier photoresists, such as tert-butyl carbonates and t-butyl esters, cannot be utilized in 157nm photoresist materials since they contain carbonyl groups. Acetal protecting groups contain no carbonyls and currently are the best choice from a transparency standpoint. Unfortunately, concerns have been raised over potential outgassing problems stemming from the use of acetals in DUV photoresists. These concerns will have to be addressed before acetals can be used successfully in 157nm photoresists. Thus, part of our research is focused on the design of new, mass persistent protecting groups that modify solubility without production of volatile side products.
Multilayer imaging
Multilayer imaging has long been studied as an alternative to single-layer imaging. When research began on 193nm lithographic systems, many felt that a multilayer technique would prevail as the dominant lithographic technology. This initial momentum arose from the lack of transparent materials at 193nm.
In multilayer imaging systems, only the top layer of a photoresist film (~100nm or less) is imaged. This image is subsequently transferred through a thicker resist layer by an anisotropic etch process. Since only the top of the film is exposed, materials that are nearly opaque to the exposure wavelength can be used in these systems. Multilayer systems typically offer the advantages of enhanced depth of focus and insensitivity to underlying topography. Researchers have studied a wide variety of such systems.
 |
Figure 5. The anion of the bis(trifluoromethyl) carbinol is stabilized by the inductive effect of the electronegative fluorine atoms.
The push to use these systems at 193nm lost most of its momentum as researchers found transparent materials that could be tailored as single-layer photoresists. Thus, most researchers lost interest in working on problems associated with the complexity of multilayer systems when it became apparent they could use single-layer processes. Thus, there was diminished interest in bilayer systems and liquid phase silylation processes such as the "Carl Process" as potential 193nm technologies.
Top-surface-imaging (TSI) processes using vapor phase silylation, while less complex in principle, faced an additional problem; TSI almost universally yielded features with a high degree of line-edge roughness. Thus, multilayer imaging systems have fallen from favor and single-layer systems have become the choice of almost all manufacturers for 193nm devices.
Now at 157nm, optical transparency is at a premium and finding materials that are sufficiently transparent for use as single-layer materials and that meet all of the other requirements is a significant challenge. As a result, thin, multilayer imaging techniques have re-emerged as leading alternatives to single-layer resists. Research on bilayer and the CARL process is continuing for 157nm at International Sematech and Infineon. Furthermore, our research group has recently shown a TSI system using vapor phase silylation that can print high-resolution features with very little line-edge roughness. Based on the results, vapor-phase TSI is a promising candidate for a 157nm imaging process.
Thin-layer resists
In the end, it may be possible to use existing photoresist materials (193nm or DUV resists) for 157nm applications. In fact, early qualification of exposure systems will most likely be done with thin films of existing commercial photoresists. Such strongly absorbing materials dictate the use of very thin films to maintain an acceptable optical density. Unfortunately, the observed film defect density grows rapidly with decreasing film thicknesses, to the point that film thicknesses required to use any existing photoresist materials at 157nm now lead to unacceptable defect levels [4]. Improved coating techniques must be developed if existing resist materials are to be converted for use at 157nm.
Conclusion
The basic information required to design 157nm photoresists is beginning to emerge. However, there is still a large amount of fundamental materials research to be done before a single-layer system with the requisite combination of etch resistance and transparency can be produced. In addition, studies relating absorbance contributions at 157nm to various structural units need to be undertaken. Photoresist solvents that are free from carbonyls and effective at dissolving polymers with increased levels of fluorine must be identified and evaluated. The influence of increased fluorine content upon resist etch performance must be established. While there is much that must be accomplished in record time, the rate of progress is very promising.
References
- T.M. Bloomstein, et al., "Lithography with 157nm lasers," J. Vac. Sci. Technol. B, 13(6), 2112, 1997.
- L.F. Thompson, et al., ed., Introduction to Microlithography, Second Edition, American Chemical Society, Washington, DC, 1994, pp. 212-232.
- E.D. Palik, ed., Handbook of Optical Constants of Solids, Academic Press, San Diego, pp. 657-687, 1991.
- R.R. Kunz, et al., "Outlook for 157nm resist design," Proc. SPIE, 3678, 13, 1999.
- H. Gokan, S. Esho, Y. Ohnishi, "Dry-etch Resistance of Organic Materials," J. Electrochem. Soc., 130, 143, 1983.
- J.R. Grandler, W.P. Jencks, J. Amer. Chem. Soc. 104, 1937, 1982.
- K.J. Przybilla, H. Roschert, G. Pawlowski, "Hexafluoroacetone in Resist Chemistry: A Versatile New Concept for Materials for Deep UV Lithography," Proc. SPIE, 1672, 500, 1992.
- H. Ito, et al., "Micro- and Nanopatterning Polymers," ACS Symposium Series 706, ed. H. Ito et al., American Chemical Society, Washington, DC, Chapter 16, p. 208, 1998.
Kyle Patterson received his BS in chemistry at Texas Christian University. His current research topics include the design of 193nm and 157nm photoresists. Patterson is working toward a PhD at the University of Texas, Department of Chemistry, Mail Code A5300, Austin, TX 78712; ph 512/471-4781, fax 512/471-7222, e-mail kyle@willson. cm.utexas.edu.
Mark Somervell received his BChE at the Georgia Institute of Technology and his MS at the University of Texas at Austin. His research topic is the development of materials for improved top surface imaging applications. Somervell is working toward a PhD at the University of Texas, Department of Chemical Engineering.
C. Grant Willson received his PhD at the University of California, Berkeley. He directed organic materials research at the IBM Almaden Research Center for several years. Willson currently holds the Rashid Engineering Chair and is a professor of chemistry and chemical engineering at the University of Texas at Austin, Departments of Chemistry and Chemical Engineering.