MICRO: Optimizing polymers to increase pellicle lifetime and transmission for 157-nm lithography

The introduction of the 157-nm wavelength for next-generation optical lithography has created a need for new soft (polymeric) or hard (quartz) pellicle materials optimized for that wavelength. The design and development of ultratransparent fluoropolymers suitable for 157-nm soft-pellicle applications has produced several promising candidate materials with absorbances <0.03/µm to achieve pellicle transmissions >95%. In order to successfully fabricate 157-nm pellicles from these fluoropolymer materials, the materials must have the appropriate molecular weight, glass transition temperature, and mechanical strength and toughness so that thin polymer films can be spin coated, lifted, and adhesively mounted on pellicle frames.

Initial results from the Massachusetts Institute of Technology's Lincoln Labs demonstrated that commercial fluoropolymers used for pellicles at 248- and 193-nm wavelengths, such as Teflon AF and Cytop, rapidly burst under irradiation with 157-nm light because they lack sufficient mechanical integrity.¹ Consequently, an extesive program was initiated to develop and screen novel fluoropolymer candidates with the required combination of optical properties, film-formation characteristics, and mechanical and photochemical radiation durability to produce 157-nm pellicles. Such candidates must produce minimal outgassing. Moreover, they require the use of noncontaminating adhesives, pellicle frames, and gasket materials so that they are not susceptible to contamination.

To produce an optically transparent, radiation-durable 157-nm pellicle, the guiding principle is to minimize the optical absorption of all pellicle materials. Lowering optical absorption maximizes transparency and reduces the potential danger of radiation-induced photochemical darkening. Minimizing all contributors to optical absorption would enable the fabrication of a pellicle with an optimal radiation durability lifetime.

As part of the effort to produce pellicles suitable for 157-nm lithography, investigators at DuPont Central Research (Wilmington, DE) and DuPont Photomasks (Danbury, CT) have developed 12 families of experimental Teflon AF (TAFx) polymer materials with sufficient transparency to produce transmissions above 95%. However, upon irradiation, these polymers undergo photochemical darkening, reducing the 157-nm transmission of the material. Measurements of the photochemical darkening rate have enabled the investigators to estimate that useful pellicle lifetime corresponds to a 10% drop in 157-nm transmission. Increasing the lifetime of fluoropolymers requires the optimization of the materials, the pellicles, and the end-use application. This article describes that optimization program.

Vacuum ultraviolet (VUV) spectroscopy, including both transmission and reflectance measurements, has become an established technique for electronic structure studies of large band-gap, insulating materials such as polysilanes and fluoropolymers.² The development of new materials for use in 157-nm photolithography involves both VUV spectroscopy and VUV spectroscopic ellipsometry because of the high transparency of the pellicle film required, insufficient accuracy of simple transmission-based absorbance/micron measurements, and the need for critical information on the material's index of refraction for pellicle-fringe optimization.³ Ellipsometry-based film structure and optical models are required to determine the complex index of refraction and film thicknesses. Also, the rapid turnaround time possible from absorbance measurements is essential in the materials-screening phase, when novel materials are being developed.

Vacuum Ultraviolet Spectroscopy. The fluoropolymer material under discussion here was first measured with a VUV spectrophotometer using a laser plasma light source (LPLS) and a 1-m monochromator with Al/MgF₂- and iridium-coated optics.^4,5 The energy range of this windowless instrument is 1.7–44 eV (700–28 nm), which extends beyond the air cutoff of 6 eV and the window cutoff of 10 eV. The resolution is 0.2–0.6 nm, which corresponds to a 16-meV resolution at 10 eV and a 200-meV resolution at 35 eV.

After being measured with the VUV-LPLS, comparison measurements were performed with a Lambda 9 spectrophotometer from Perkin Elmer Instruments (Norwalk, CT). The spectrophotometer was equipped with transmission and reflectance attachments and used xenon and deuterium lamps to cover the UV-visible-near infrared (UV-vis-NIR) spectral regions. Wavelength resolution was 2 nm at 5.0 eV and energy resolution was 40 meV at 5.0 eV. The transmission was measured over the complete energy range of the instrument to check the accuracy of the VUV transmission spectra. The results closely agreed with the data from the VUV-LPLS spectrophotometers. With the aid of a multiplicative scaling factor on the VUV transmission profile, the data were spliced together in the overlapping wavelength region to bring them into agreement with the UV-vis-NIR results. Once the transmission spectra of the substrate and the film coated on that substrate had been determined, film transmission (T) and optical density (OD) were determined using the following equations:

Absorbance (A) values in units of 1/micron (base 10) were obtained by dividing optical density by film thickness:

Vacuum Ultraviolet Spectroscopic Ellipsometry. Another instrument used to measure the fluoropolymer material was the Model VU-302 variable-angle spectroscopic ellipsometer (VUV-VASE) from J. A. Woollam (Lincoln, NE).⁶ The tool covers the wavelength range of 142–1700 nm and has an angle range of 15°–90°. It is based on a rotating analyzer VASE instrument that covers the UV-vis-NIR spectral range and incorporates a computer-controlled MgF2 Berek waveplate as a compensator to improve ellipsometric phase difference (D) measurement accuracy. The introduction of a retarding element also allows the ellipsometer to distinguish unpolarized light, improving results when dealing with thicker films by carefully treating nonidealities such as instrument bandwidth and sample nonuniformity. Purging the entire system with dry nitrogen gas can prevent the absorption of VUV light by ambient oxygen and water vapor.

The instrument functions by permitting light from both the deuterium and the xenon lamp to pass through a double-chamber Czerny-Turner–type monochromator to provide wavelength selection and stray-light rejection. Computer-controlled slit widths can adjust the bandwidth to ensure that the data reflect adequate spectral resolution of optical features such as closely spaced fringes, which arise in very thick films. A photomultiplier tube is utilized for signal detection in the ultraviolet range while a stacked Si/InGaAs photodiode detector is used for longer wavelengths.

The ellipsometric analysis discussed here focused on a pellicle made of the fluoropolymer TAFx3P-10517. The polymer films were prepared by spinning polymer solution onto various substrates. VUV spectroscopy tests used CaF₂ substrates from Corning (North Brookfield, MA), while VUV ellipsometry tests used pellicle membrane samples. Film thicknesses were measured using an F20 thin-film measurement system from Filmetrics (San Diego, CA). The ellipsometry and transmission data from the pellicle membrane, shown in Figures 1 and 2, respectively, were fitted to an optical model of the pellicle to determine the pellicle's membrane thickness, roughness, thickness nonuniformity, and complex refractive index⁷:

In this equation is the complex index of refraction, n is the (real) normal index of refraction, and k is the imaginary extinction coefficient. The model chosen for these films was the Tauc-Lorentz model, which remained Kramers-Kronig consistent and helped describe the onset of film absorption.⁸

Figure 2: The transmission of a TAFx3P-10517 pellicle and the fitted results from the optical model.

The analysis focused on pellicle membranes with thicknesses on the order of 1 µm. Such thicknesses make the film highly sensitive to optical absorption and give them a high extinction coefficient because the path of light through the material is long. At the same time, these pellicle membranes exhibit roughness and thickness nonuniformity and are quite susceptible to the effects of finite spectrometer bandwidth. Thickness nonuniformity and a finite bandwidth can depolarize the measurement beam. Ellipsometers with a retarding element can measure the resulting percent of depolarization, enabling the quantification of these nonideal effects and the modeling of their behavior during experimental data analysis.⁹ The roughness of these pellicle membranes was 0.4 nm, the thickness nonuniformity was 2.8%, and the light bandwidth was 1 nm.

Once the extinction coefficient (k) had been determined, the optical absorption parameters a and A could be arrived at. The absorption coefficient (a) is calculated on a natural logarithm basis while the absorbance/micron, A, is determined from the base-10 logarithm of the optical density:

Both k and a are inherent optical properties of the material, while the absorbance/micron is based only on transmission measurements. Effects arising from the index mismatch between the film and substrate, thin-film (or Fresnel) interference effects, and film nonuniformity are negligible. The equation

Figure 3 shows the optical absorbance/micron of two commercial amorphous fluoropolymers, such as Teflon AF and Cytop. The 157-nm transmission of a pellicle membrane made from Teflon AF would be ~45%, while a pellicle membrane made from Cytop would be ~2.5%. Studies indicate that these fluoropolymers rapidly become perforated under 157-nm laser irradiation and have physical lifetimes of ~1 J. It is estimated that a 10% drop in 157-nm transmission for Teflon AF corresponds to a pellicle lifetime of ~0.001 J.

Figure 3: Optical VUV absorbance of commercial Teflon AF and Cytop polymers.

Extending pellicle lifetime requires the development of novel fluoropolymers with substantially less optical absorption at 157 nm. The researchers have uncovered twelve fluoropolymer families with optical absorbances in the range of 0.03/micron, the target value established for viable 157-nm pellicle materials. Table I lists these pellicle polymers and Figure 4 shows their optical absorbance.

Polymer Family	Absorbance/µm
TAFx24P	0.007
TAFx27P	0.008
TAFx3P	0.009
TAFx1P	0.012
TAFx2P	0.014
TAFx4P	0.015
TAFx5P	0.016
TAFx7P	0.016
TAFx20P	0.028
TAFx6P	0.03
TAFx28P	0.03
TAFx21P	—
Table I: Optical absorbance of experimental Teflon AF polymer families.

The development of a 157-nm pellicle requires a complete system design in which the important role of adhesives is considered from the beginning. A pellicle adhesive attaches the pellicle membrane to the frame, while a pellicle gasket attaches the pellicle frame to the photomask. The researchers have developed pellicle and frame adhesives for use in 157-nm pellicles with the required mechanical performance and low optical absorbances needed to minimize the impact of outgassing on the pellicle and photomask transmission.

A program for fabricating viable 157-nm pellicles also must optimize pellicle properties and materials. The physical parameters of the thin-film membrane, such as membrane thickness, must be optimized. These parameters produce the tuned interference fringes in the optical behavior of the membrane. Pellicles must have sufficient optical absorption and radiation durability; proper polymer molecular weight and spinning-solution viscosity needed for membrane formation; and low levels of organic, inorganic, and particulate contaminants. In addition, the degree of surface roughness and thickness nonuniformity must be optimized. Finally, steps also must be taken to optimize the pellicle's adhesive, frame gasket, mounting, and purging capability.

Careful consideration must also be paid to controlling the end use and environmental aspects of the 157-nm pellicle application.¹⁰ For example, adsorbed oxygen, water, and hydrocarbons can reduce 157-nm transmission. Test results show that active cleaning processes such as 157-nm laser cleaning or lamp cleaning can improve 157-nm transmission.¹

Teflon AFx2P and AFx3P Pellicles. Pellicle thickness directly affects transmission at the lithographic wavelength. Ellipsometric analysis of pellicles made from TAFx2P-10520 (one of the polymer materials uncovered by the research team) revealed that a pellicle with a 1.67-µm-thick membrane has a 157-nm transmission of ~86%. A TAFx2P-10520 pellicle with a membrane thickness of 0.8 µm has a 157-nm transmission of ~92.5%. The thicknesses of these pellicles were not optimized to establish an interference fringe peak maximum of exactly 157 nm. These values are the approximate percentage of 157-nm transmission obtained from the average of the fringe peak maxima above and below the 157-nm profile.

The fundamental absorption of the polymer material is another property that directly determines 157-nm pellicle transmission. Figure 5 shows the optical absorption coefficient (k) in units of 1/cm (base e) for the TAFx2P-10520 pellicle and the TAFx3P-10517 pellicle, a variant of another fluoropolymer. TAFx3P-10517 has a much lower 157-nm absorption coefficient than TAFx2P-10520 and thus should produce a pellicle with higher 157-nm transmission. Figure 5 also shows that TAFx3P-10517 exhibited an absorption peak centered at ~190 nm. That undesirable absorption peak was eliminated in the TAFx3P-10515, as shown in Figure 6.

Figure 6: Optical absorption coefficient in units of 1/cm (base e) for TAFx3P-10517 and TAFx3P-10515.

Although the membrane of the pellicle made from TAFx3P-10515 was only 0.77 µm thick, the pellicle's 157-nm transmission was only ~92%, which was a result of the pellicle's thickness nonuniformity. That nonuniformity produced tuned interference fringes whose amplitude damped with decreasing wavelengths, as presented in Figure 7. This effect corresponded to the superposition of pellicles with varying thicknesses, in which the peak fringe transmission was reduced because an adjacent area of the membrane with a different thickness had a lower transmission and, therefore, a different wavelength for the fringe maxima. The effect of thickness nonuniformity differs from that of surface roughness. Surface roughness leads to a damping of the pellicle fringes at all wavelengths. Reducing the thickness nonuniformity of a membrane in a TAFx3P-10515 pellicle would increase 157-nm transmission to ~97%.

The PCD process can be considered a nucleation and growth mechanism that begins when a preexisting site (a PCD nucleation site or nuclei) absorbs 157-nm photons in the pellicle membrane and the absorbed energy produces absorbing chromophores. The absorption of photons leads to the formation of a transient excited-state formation that, upon decay, results in a final PCD absorbing species. This species can lead to the absorption of subsequent 157-nm photons.

When nucleation and PCD growth occur, the first line of defense is to reduce the initial number of absorbed 157-nm photons in the material (the nuclei) that produce the initial photochemical darkening. The second is to decrease the growth rate of the darkening. To decrease the nucleation rate in this study, all of the intrinsic and extrinsic absorptions in the polymers were reduced. An effort also was made to reduce the growth rate of PCD. By reducing the nucleation rate and the PCD growth rate, 157-nm radiation durability lifetime was increased.

The PCD Process. Investigators can dynamically examine the PCD process by measuring the 157-nm transmission of a polymer sample as a function of irradiation dose.¹¹ The time-dependent data derived from such an examination is essential to understanding and developing kinetic models of the process. Measuring PCD enables investigators to measure and analyze the rate of transient photochemical darkening accurately. They can then model this PCD data to determine the linear photochemical darkening rate per incident dose (PCDi) in units of induced absorbance/micron/J.

Using in situ power meters to determine the 157-nm transmission of a TAFx3P-1083 sample under irradiation, a value for the decrease in transmission as a function of increasing dose was obtained, as shown in Figure 8. An initial transient increase in transmission was observed in phase 1 of the PCD process, followed by a plateau in the transmission profile in phase 2. In phase 3, the sample darkened and 157-nm transmission decreased. The initiation and plateau phases of the process corresponded to the initiation phase of photochemical darkening. Once the transmission had begun to decrease, the absorbance of the sample could be determined from the sample's known thickness. By subtracting the initial absorbance, the induced absorption could be plotted as a function of increasing dose.

The PCDi rate of the phase 3 darkening process was determined from the induced absorption rate—that is, the linear increase in the induced absorption for a given incident dose, starting after the phase 2 plateau. This procedure was used to calculate the PCD rate corresponding to a linear model of the material-darkening process that occurs in phase 3. Figure 9 shows induced absorption versus incident dose for four different TAFx polymers. Table II lists the linear PCDi rates fitted to the phase 3 darkening.

Pellicle Lifetime. Pellicle lifetime (DT) can be estimated from the PCDi rate and the transmission change over DT. During a pellicle's lifetime, 157-nm radiation leads to induced absorption that decreases pellicle transmission. When the transmission drop corresponds to DT, the pellicle's lifetime has been exceeded. Pellicle lifetime is calculated with the following equation:

Upon irradiation by 157-nm radiation, existing commercial fluoropolymers such as Teflon AF (A = 0.43/micron) or Cytop (A = 1.9/micron) were ablated and their membranes physically thinned to bursting. Perforation occurred after an ~1-J dose of 157-nm radiation. Without measuring this very short lifetime experimentally, the investigators estimated that the 157-nm lifetime of these commercial polymers in the event of a 10% transmission drop was ~0.001 J.

The experimental Teflon AFx materials developed for this program have much lower optical absorbances (<0.03/micron) than previous fluoropolymers. Furthermore, these experimental materials exhibit much greater mechanical and physical integrity under irradiation. Nevertheless, the experimental Teflon AFx materials experience PCD under 157-nm irradiation, which decreases their initially high transmission.

A correlation was found between the materials' PCDi rate and their initial absorbance. In addition, the materials with a lower absorbance also experienced lower rates of photochemical darkening. Table II shows that TAFx2P had the highest absorbance and the shortest lifetime while TAFx3P-1402 had the lowest absorbance and the longest lifetime. The differences between these materials' physical properties corresponded to the differences between pellicles, indicating that successive material improvements have been achieved. Because of these improvements, pellicles' 10% linear lifetime has increased from 2.7 to 5.8 J.

Polymer	PCDi Rate Absorbance/µm/J	10% DT Linear Lifetime (J)
TAFx2P-1078	0.022	2.7
TAFx3P-1083	0.016	3.7
TAFx3P-1401	0.018	3.2
TAFx3P-1402	0.010	5.8
Table II: PCDi rates and linear pellicle lifetimes for 10% transmission drop for different TAFx polymers.

Conclusion

The successful fabrication of 157-nm pellicles requires that fluoropolymer materials generate thin polymer films that can be spin-coated, lifted, and adhesively mounted to pellicle frames—the processes that produce freestanding pellicle membranes of micron-scale thickness. These materials must have proper molecular weight, glass transition temperature, and mechanical strength and toughness. Various ultratransparent TAFx polymer families with these physical properties have been developed.

Upon irradiation, however, these 157-nm pellicle polymers undergo photochemical darkening, which reduces their 157-nm transmission. PCD measurements enable investigators to estimate pellicle lifetime corresponding to a 10% drop in 157-nm transmission. Increasing fluoropolymers' 157-nm lifetime requires the simultaneous optimization of polymer materials, pellicles, and the end-use application. Similar optimizations have played an essential role in achieving desired radiation durability lifetimes for pellicles successfully developed for use with 248-nm krypton fluoride and 193-nm argon fluoride lithography.

The authors wish to thank David J. Jones, M. F. Lemon, Fredrick C. Zumsteg, Kenneth G. Sharp, and Weiming Qiu of DuPont Central Research for their contributions to the work presented in this article. They would also like to acknowledge the assistance of Vlad Liberman, Rod Kunz, and Mordechai Rothschild of MIT's Lincoln Labs for the PCD characterizations and their work on photochemical darkening. We thank the following for their assistance with this work: Gregg L. McCauley, Michael Crawford, Robert J. Smalley, William Wheeler and Dick Moore. We also acknowledge the assistance of Daniel Miller of Sematech for performing some VUV VASE measurements.

Roger H. French, PhD, is a senior research associate in materials science at DuPont Central Research (Wilmington, DE). He also is an adjunct professor of materials science at the University of Pennsylvania (Philadelphia). His current work focuses on materials for 157-nm pellicles, photoresists, and photomasks. French holds six patents and has published more than 90 papers. He received a PhD in materials science from the Massachusetts Institute of Technology (Cambridge), where he was involved in vacuum ultraviolet spectroscopy from 80 to 800 nm. (French can be reached at 302/695-1319 or roger.h.french@usa.dupont.com.)

Robert C. Wheland, PhD, is a senior research fellow in materials science and engineering at DuPont Central Research, where he has been for 29 years. His research is in the field of fluoropolymers and nylon amidation, and currently he is focusing on materials for 157-nm pellicles. Wheland holds more than 30 patents. He received a PhD in chemistry from Harvard University (Cambridge, MA), where he was involved with olefin cycloaddition reactions. (Wheland can be reached at 302/695-2272 or robert.c.wheland@usa.dupont.com.)

Joseph Gordon is in charge of pellicle technology and R&D at DuPont Photomask (Danbury, CT). Previously, he developed mask blanks and pellicles at Tau Laboratories before that company was acquired by DuPont in 1986. Gordon holds several patents on photomask material technology for pellicles and blanks. He is the author or coauthor of articles on pellicle technology. He received a BS in physics from the University of California, Berkeley. (Gordon can be reached at 203/730-5112 or joseph.gordon@photomask.com.)

Edward Zhang is a research and development engineer in the pellicle division of DuPont Photomasks (Danbury, CT). He is responsible for the development of next-generation lithography pellicles, fabrication processes, and defect elimination. He received an MS in chemical engineering from the University of Alabama in Huntsville. (Zhang can be reached at 203/730-5029 or edward.zhang@photomask.com.)

MicroHome | Search | Current Issue | MicroArchives
Buyers Guide | Media Kit

Questions/comments about MICRO Magazine? E-mail us at cheynman@gmail.com.