SI CHINA     SI JAPAN
Login  |  Register          Free Newsletter Subscription
Subscribe
Email
Print
Reprint
Learn RSS

Photomask Defectivity and Cleaning: A New Milieu

Along with almost everything else in semiconductor lithography, photomask defect detection and elimination are experiencing a revolution, as new defect types require more innovative cleaning tools and processes.

Franklin Kalk, Joseph Gordon and David Chan, Toppan Photomasks Inc., Round Rock, Texas -- Semiconductor International, 9/1/2007

Photomasks are conceptually simpler than wafers: fewer levels, fewer process steps, and larger main features. However, mask optical proximity correction (OPC) features are about the same size as a wafer's printed features. A single wafer defect affects one die, while a single mask defect affects every die. Mask defects are detected optically, so even minor optical fluctuations on the mask render it defective. Thus, the historical mask cleaning challenge has been to provide a mask perfectly free of microscopic contamination without altering its physical properties or damaging the masking layers.

ArF (193 nm) lithography tools have introduced a new defect class known as haze. All variants of haze share two attributes: they are often detected only after the mask is exposed in the wafer scanner, and the source is molecular contamination near the mask surface. Classic mask cleaning techniques, effective at removing microscopic contamination defects, can actually contribute to one type of haze. Eliminating haze requires a collaborative effort between mask manufacturers, wafer manufacturers and tool and materials suppliers.

Focusing primarily on haze, this article describes our most recent understanding of mask defects, their detection and elimination, and methods for maintaining a defect-free mask throughout its useful life in the wafer fab.

Mask defects

1. An example of a hard defect: an extension of the chromium-based absorber layer on the mask.
Mask defects are classified as hard or soft. Hard defects are known as pattern defects, because they manifest in the pattern imposed on the mask. Examples include extra or missing absorber material, quartz bumps, or subtle alteration of the phase, transmission or reflection properties of the films. Hard defects (Fig. 1 ) are detected by pattern inspection tools that compare the pattern to the mask design database (die-to-database inspection) or to another ostensibly identical pattern on the mask (die-to-die inspection). Defects can often be repaired by adding missing material or removing unwanted material with a variety of chemistries and localized excitation (laser, focused ion beam [FIB] or e-beam) or nanomachining techniques.

Soft defects are also known as contamination defects because they usually manifest as small particles that do not cause pattern damage, but can cause wafer print failures. Examples are resist particles left on the mask absorber films and particles that fall onto the mask after it is

2. Soft defects, such as this particle, are typically removed during the mask cleaning process.
patterned. Soft defects (Fig. 2 ) are usually removed by the mask cleaning process. Haze is a type of soft defect.

Mask cleaning challenges

The mask cleaning process must not alter the physical properties of the mask or damage the masking layers. In particular, it must not damage hard-defect repairs or subresolution assist features (SRAFs), nor should it alter the optical properties of the mask.

There are two main types of mask cleaning, loosely known as strip and final clean. Stripping removes the remaining photoresist after mask patterning. With positive resists, this can be accomplished with a blanket exposure, followed by conventional develop and rinse steps. For negative resists, no exposure is required before the develop and rinse steps. The second approach, which is effective for all organic materials, employs a mixture of concentrated sulfuric acid and 30% hydrogen peroxide (typically 2:1–4:1 ratio at >100°C, followed by a deionized [DI] water rinse,1 called Piranha etch). The two approaches may be combined for maximum stripping efficiency.

Final clean removes any resist or particle contaminants that remain on the mask after patterning, inspection and repair. Particle-removal efficiency (PRE) in this step is critical, as it is followed immediately by pellicle application. Similar to stripping, the final clean typically employs Piranha etch and water rinse to remove residual organic particulates. In the most efficient process, this is followed by an ammonium hydroxide megasonic spray to neutralize residual acid (thereby forming ammonium sulfate, an easily removable salt), a DI water rinse to remove the ammonium sulfate, and a spin-dry step.

Haze

With the introduction of the pellicle as a protective element against airborne particulates, the typical period between reticle cleanings increased from days to years. This set the stage for a new type of problem — haze or progressive growth defects.

Early cases of haze were either organic (caused by outgassing from pellicle materials) or ammonium sulfate crystals (caused by poor rinsing and neutralization of the sulfuric mask cleaning chemistry).2 These initial problems were quickly diagnosed and resolved. The migration from g-line to i-line lithography progressed with few haze-related issues and, for many years, masks were so reliable that indefinite reticle lifetime became the expectation.

As the industry moved to KrF (248 nm) lithography, progressive growth defects began to reappear.3 These defects were traced to the breakdown of the pellicle gasket and adhesive materials. At 248 nm, the photon energy surpasses common bond strengths found in organic materials, and many organic materials absorb strongly in the deep ultraviolet (DUV) range. It was discovered that the pellicle materials and their outgassed constituents were undergoing photodecomposition and subsequent reaction, finally precipitating in the form of haze. Changes to more durable pellicle-construction materials and elimination of specific volatile constituents mitigated haze formation. However, a new type of growth was observed at 248 nm, where large crystals grew on the fluoropolymer pellicle film. This growth was identified as oxalic-acid derivative. The source is still unclear, but it is interesting to note that oxalic acid is one of the major components of organic-fine particulates formed by natural atmospheric processes.

With the advent ArF lithography, the industry is again experiencing a significant increase in haze incidence and types. Industry surveys indicate that the incidence rate has increased by a factor of 10 between 248 and 193 nm.4 Early adopters of 193 nm lithography experienced reduced scanner lens transmission, increased flare and extremely short reticle lifetime caused by haze formation on optical surfaces.5,6 Even double carbon bonds in organic materials are easily broken at 193 nm (Table 1 ). In addition, oxygen is efficiently converted to ozone at 193 nm, and pellicle gasket materials are vulnerable to attack by this strong oxidizer.

For ArF lithography, ammonium sulfate is the most frequently reported form of haze. Organic haze, a class of sub-500 nm defects, occurs after reaction of volatile organics under the pellicle during the first few exposures of a mask. Cases of ammonium nitrate have also been reported at 193 nm, along with a few observations of other forms of haze.

Ammonium sulfate haze mitigation initially focused on improved cleaning to reduce sulfate and ammonia residual levels. Photomask manufacturers reduced these residual contaminants by two orders of magnitude in the past five years (Fig. 3 ). Despite these improvements, the problem persists, so the current trend is to eliminate sulfates from the mask manufacturing process.

3. Over the past five years, mask manufacturers have significantly reduced ammonium sulfate residual levels.

The reticle storage environment also impacts residual mask contaminant levels, even without photon exposure. These levels are quantified by accumulating residual contamination on witness plates stored at different points in a wafer fab and measured using ion chromatography. Sulfate and ammonia levels in the fab after one month are 3-4× those of plates leaving the mask shop (Fig. 4 ). These residuals were adsorbed from the air around the reticle (airborne molecular contaminants [AMCs]). Therefore, improvements in the reticle environment are required.

4. Residuals buildup on plates in a wafer fab caused by natural levels of sulfate and ammonia in the atmosphere.

Testing at our accelerated exposure test bed revealed multiple routes to ammonium sulfate formation.7-9 It can occur during both wafer exposure and storage, where ammonia and sulfate are present. Customer feedback indicates that implementing AMC filtration for storage and exposure significantly improves reticle lifetime and reliability. Atmospheric analogs provide models for photomask haze, especially for ammonium sulfate formation (Table 2 ).

Atmospheric haze is caused by light scattering from particulates, much as haze does on a lens or mask. Roughly 50% of the fine particulates in the atmosphere are ammonium sulfate, created from airborne sulfur dioxide (SO2) and ammonia.10 SO2 is released into the atmosphere from both natural (volcanic activity, phytoplankton emissions) and man-made (fossil fuel burning) sources. Ammonia is a natural biowaste product. In combination with energy from sunlight, ozone and water vapor, gaseous ammonium sulfate is formed in the upper atmosphere and then quickly precipitates to form particulates. These same AMCs are present around the mask, with 193 nm photons providing energy to drive ammonium sulfate formation. Atmospheric concentrations of SO2 and ammonia vary significantly both regionally and seasonally,11 so the challenge to control such species varies from fab to fab and over time. Ammonium nitrate is the second most common inorganic fine particulate in the atmosphere, and is formed in a similar manner. Of the organic fine particulates in the atmosphere, carboxylic acids are prevalent components, and oxalic acid is the most common of these.12

Organic haze is manifest as small defects occurring during the first exposures of a mask. The defect density is high initially, but decreases with additional exposure. The defect size is typically below what is lithographically printable, but it can become a yield detractor. Smog provides the best analog for organic haze. Volatile organic species in the air combine in the presence of ultraviolet (UV) radiation and ozone to form organic compounds that precipitate to form atmospheric particulates. In the case of the reticle, the sources of organic compounds are the pellicle and storage container materials, as well as process-related species occurring in the fab. As with ammonium sulfate, witness plate tests have confirmed significant increases in organic contamination levels on the photomask after storage in fab conditions without exposure. These tests also revealed outgassed materials from the pellicle and mask packaging. Package suppliers have reduced these levels, and further reduction is needed, especially to extend improvements to reticle stockers and containers.

Managing haze in a large wafer fab is a critical issue. Haze detection usually occurs during mask inspection or, in the worst case, as a result of a yield loss event. Because haze is endemic at 193 nm, many fabs have implemented a number of containment actions, such as time- or exposure-based mask inspection.13 Suspect reticles are returned to the mask shop for cleaning, presenting logistical issues for the fabs. To keep production lines operating, multiple reticles for the same layer are often made, adding cost and complexity.

Traditional mask inspection does not adequately characterize the haze types seen by the reticle user. Typically, the mask must be quickly cleaned, protected with a new pellicle and returned to the fab. This limits the ability to analyze defects, understand root causes and develop corrective actions. Because each haze type has a unique chemical makeup, several techniques are employed to determine the defect type (Table 3 ). Both time-of-flight (TOF)-SIMS and micro Raman spectroscopy provide spatial resolution in the micron/near-submicron range to help determine the defect chemistry. As smaller defects become critical, new techniques will be required to determine the molecular makeup.


Advanced mask cleaning

To prevent the formation of ammonium sulfate haze on ArF masks, the photomask industry is migrating from the traditional Piranha chemistry to sulfate-free resist strip and final clean processes, including ozonated water (10–100 ppm O3), hydrogenated water (1–10 ppm H2, with small amounts of NH4OH added to maintain pH), oxygen plasma ashing and 172 nm UV exposure. Figure 5 shows typical strip and clean flows for sulfur-based and sulfur-free processes.14,15 Most advanced mask strip/clean systems employ single-substrate, spin-spray architectures for maximum cleanliness.

5. Comparison of the traditional Piranha strip and clean (top) and sulfate-free strip and clean.

This migration poses significant technical challenges. Ozonated water (O3:H2O) removes organic contaminants by chemical reaction, but the strip rate is relatively slow. High temperature, normally employed to accelerate chemical reactions, simply decomposes the ozone. Thus, O3:H2O can remove thin organic layers, but is not suitable for bulk resist stripping, particularly for thick resist edge beads or resists hardened by dry etching. Extended O3:H2O exposure can damage the chromium layer's top antireflective layer, so the ozone concentration must be optimized to provide a reasonable resist strip rate without causing damage. To widen the operating window, the resist strip process may need a plasma resist strip to remove the bulk of the resist, leaving a thin resist layer for removal by O3:H2O (possibly with the assistance of 172 nm UV exposure). Pellicle replacement poses another thick organic removal problem, where pellicle adhesives can be quite tenacious and may require solvents or Piranha chemistry for complete removal.

While the slow organic removal rate of O3:H2O is a minor issue for final clean, other techniques (e.g., high-temperature DI water rinse, baking or 172 nm UV) may be employed to reduce sulfate levels. Hot DI water is the most effective of these alternative methods, although it is not as effective as replacing Piranha with O3:H2O.

As allowable defect size decreases with each technology node, increased force is required to remove the minimum defect. Because SRAFs, notably scatter bars, are approximately the same size as the maximum allowable defect, it is important to avoid SRAF damage while maximizing PRE. Megasonic-assisted cleaning (primarily spray in advanced mask cleaners) is often used to increase the PRE, and megasonic assist can damage small features. Higher megasonic frequency (typically ~3 MHz) produces smaller cavitation artifacts and hence smaller collapse forces than lower frequency (~1 MHz) agitation, but PRE can be compromised. Cleaning equipment suppliers have advanced several techniques to maintain PRE with minimal SRAF damage (dual frequencies, backside megasonic application and tilted spray angles).

The use of single-substrate spin/spray cleaning systems poses other challenges not found with tank-based cleaners. The edge of the mask may harbor organic contaminants, which can be difficult to remove because of low chemical exposure. One solution is to combine polyvinylacetate (PVA) brushes with the O3:H2O, or direct the plasma to these areas during ashing. While the backside of the plate is not intentionally exposed to the cleaning chemistry, modern systems flip the plate to clean both sides. It is also possible to rinse the backside with hot DI water.

Shared issues, shared answers

The challenges of preparing and maintaining a defect-free photomask are considerable. The as-shipped mask surface must be chemically pure enough to negligibly contribute to haze formation. Beyond the mask manufacturing facility, the wafer fab must have equipment and control systems capable of maintaining the mask in a pristine state over its useful life. Such systems are still not fully in place in the typical wafer fab; indeed, commercial solutions to overcome some of these challenges have not yet been developed.

There are multiple contributors and mechanisms that can lead to haze formation on the photomask. Not only must the reticle be protected from particulates, it must be protected from naturally occurring AMCs. Most of the lithography supply chain shares responsibility to develop and implement haze mitigation procedures. Besides mask and wafer manufacturers, suppliers of mask packaging and storage and lithography equipment can contribute to, and benefit from, successful solutions. Some semiconductor manufacturers have already taken a comprehensive approach to managing haze by collaborating with a variety of suppliers, and have realized great improvements in mask life.13

To address the challenges facing them, mask cleaning equipment suppliers and mask manufacturers must also work closely together. To maximize PRE while minimizing pattern damage, physical property damage and residual molecular contamination requires advanced cleanroom, inspection and analytical facilities that are not generally available at cleaning equipment suppliers and, in fact, exist only at a few leading photomask manufacturers. Thus, successful partnering is critical to continued successful mask cleaning solutions.

The immediate mask environment must now be controlled throughout the mask's life, both in the scanner and in the wafer fab reticle stocker. This requires a low outgassing environment in the facilities and containers used for reticle storage and transport. Monitoring systems should ensure that AMC excursions are detected before mask contamination occurs. Encouragingly, commercial mask storage and AMC control offerings are beginning to appear.16,17 These include low outgassing materials, improved AMC filtration and purging performance, and monitoring of the storage atmosphere, which promise to reduce both inorganic and organic contamination to acceptable levels.

Historically, AMC filtration in scanners was focused on protecting chemically amplified resists, but the introduction of ArF scanners has shown that the mask environment must be as tightly controlled. Similarly, mask inspection equipment historically offered no AMC protection to the mask. The recent introduction of 257 nm mask inspection equipment presented a haze challenge similar to the one accompanying KrF scanners, but with an added factor: The mask inspection tool's extremely high photon fluence can precipitate haze in a very short time without environmental and beam exposure controls. The next generation of mask inspection equipment, which will use sub-200 nm wavelength, will require strengthened environmental controls to prevent mask and optical system degradation.

To manufacture and maintain a clean, defect-free maskrequires cooperative work all along the supply chain. This approach is more cost-effective than separate (and inevitably redundant) efforts.



Author Information
Franklin Kalk, CTO of Toppan Photomasks since April 2004, leads the company's global R&D and company-wide programs aimed at improving the quality and yield of photomask products. Kalk has a doctorate in optics from the University of Rochester (N.Y.) and a B.S. in physics from Emory University (Atlanta).
Joseph Gordon is a senior member of Core R&D Technical Staff at Toppan Photomasks, with over thirty years in the photomask industry. He has held positions in engineering, R&D and manufacturing management. He has a B.S. in physics from UCLA and a Ph.D. from the University of California, Berkeley.
David Chan is a member of the technical staff at Toppan Photomasks, with over 25 years of experience in the semiconductor industry. He has an M.S. in chemistry from Memphis State University, a Ph.D. in chemistry from the University of Southern California (Los Angeles), and an M.S. in electrical engineering from the University of Texas in Austin.


References
  1. W. Kern, "Overview of Wafer Cleaning Technology," in Kern, ed. Handbook of Semiconductor Wafer Cleaning Technology, Noyes, Park Ridge, N.J., 1993, p. 19.
  2. K. Bhattacharyya et al., "Investigation of Reticle Defect Formation at DUV Lithography ," Proc. SPIE, 2002, Vol. 4889, p. 478.
  3. K. Bhattacharyya et al.,"Investigation of a New Generation of Progressive Mask Defects on the Pattern Side of Advanced Photomasks ," Proc. SPIE, 2005, Vol. 5853, p. 100.
  4. G. Shelden and P. Marmillion, "Mask Industry Assessment: 2006 ," Proc. SPIE, October 2006, Vol. 6349.
  5. A. Grayfer, O. Kishkovich, D. Ruede, "Protecting DUV Optics from Airborne Molecular Contamination," Microlithography World, February 2002, Vol. 11, No. 1, p. 20.
  6. Ekberg et al., "Meeting Airborne Molecular Contamination Challenges in Laser Pattern Generators," Micro, August 2004.
  7. J. Gordon et al., "Use of Excimer Laser Test System for Studying Haze Growth ," Proc. SPIE, 2005, Vol. 5992.
  8. J. Gordon et al., "Influence of Environmental Components on Haze Growth ," Proc. SPIE, 2007, Vol. 6607.
  9. E. Johnstone et al., "193 Haze Contamination: A Close Relationship Between Mask and Its Environment ," Proc. SPIE, 2003, Vol. 5256, p. 440.
  10. A. Gelencser and Z. Vargas, "Evaluation of the Atmospheric Significance of Multiphase Reactions in Atmospheric Secondary Organic Aerosol Formation," Atmospheric Chemistry and Physics Discussions, 2005, Vol. 5, p. 4407.
  11. The US Environmental Protection Agency, Clean Air Status and Trends Network Annual Report , 1999, p. 79.
  12. K. Crahan et al., "Oxalic Acid Production in the Coastal Marine Atmosphere ," 84th AMS Annual Meeting, January 2004.
  13. S. Gough et al., "Reticle Haze: An Industrial Approach ," Proc. SPIE, 2007, Vol. 6533.
  14. S. Hoyeh et al., "Non-Chemical Cleaning Technology for Sub-90nm Design Node Photomask Manufacturing ," Proc. SPIE, 2006, Vol. 6349.
  15. S. Anzai et al., "Sulfate-Free Photomask Cleaning Technology ," Proc. SPIE, 2006, Vol. 6283.
  16. Entegris Clarilite and RSPX.
  17. Particle Measuring Systems IMS.
Email
Print
Reprint
Learn RSS

Related Content

Related Content

By This Author

There are no other articles written by this author.

SPONSORED LINKS



 
Advertisement
SPONSORED LINKS

More Content

  • Blogs
  • Podcasts

Blogs

Podcasts

Advertisements





NEWSLETTERS
Plug in and get the latest SI news, trends and industry updates delivered free, directly to your inbox!

SI NewsBreak and Special Reports (Weekdays)
Wafer Processing Report (Monthly)
Lithography Report (Monthly)
Metrology Report (Monthly)
Clean Processing Report (Monthly)
Packaging Report (Twice Monthly)
©2008 Reed Business Information, a division of Reed Elsevier Inc. All rights reserved.
Use of this Web site is subject to its Terms of Use | Privacy Policy
Please visit these other Reed Business sites