Organic growth on pattern side of reticles gives rise to new progressive mask defects
07/01/2005
Unexplained progressive mask defects continue to plague and puzzle semiconductor manufacturers. Even if new reticles are determined to be clean upon arrival from mask shops, some photomasks show catastrophic defect growth over the course of production use in fabs. Studies have identified a new wave of progressive defect growth consisting of organic contaminants on half-tone edges and chromium edges of masks. Sources of new defect mechanisms remain unresolved, but several candidates have been considered. While investigations continue, device makers are encouraged to carefully develop mask requalification inspection strategies to identify contaminated reticles before defects impact the performance of leading-edge products.
Progressive mask defects are an industry-wide mask reliability issue, which has been a problem for more than a decade and observed at almost every lithographic wavelength. Categories of defects that cause reticle-quality degradation over time are commonly known as crystal growth, haze, fungus, or precipitate. This problem is especially severe at 193nm lithography because exposure wavelengths are shorter than previous photolithography generations and photons are highly energized. The risk of progressive mask defects also has increased with the transition to 300mm wafers because photomasks are exposed over longer periods of time to cover larger substrates compared to 200mm wafers. Both embedded phase-shift masks (EPSM) and chrome-on-glass masks are affected by progressive defects, which are generally found on the patterned reticle surface underneath pellicles (on clear, half-tone, or chrome patterns), as well as on the backside surface of photomasks. Past cases have indicated that this problem mainly starts on the scribes and borders, with emerging semitransmissive contamination of ~100nm. These defects then propagate into the die area while growing in both size and opaqueness. Compositional analysis has shown that the majority of these defects are ammonium sulfate. However, since significant effort focused on the elimination of ammonium sulfate, a new trend has emerged.
Current studies show severe defect growth consists of organic contaminants (ammonium oxalate, cyanuric acid, etc.) on half-tone edges and on chromium edges. Sources for these progressive defect mechanisms are under investigation, though several candidates have been considered: maskmaking materials and process residues (mainly ammonium or sulfate ions); the fab environment; or the stepper environment. Controlling or balancing these sources may help to reduce the frequency at which these defects occur, but thus far ongoing efforts have been unable to eliminate the problems. Changes in photolithographic wavelengths, successive device shrinks, and new processes in maskmaking as well as IC fabrication often disrupt the fine balance among suspected defect sources, resulting in the return of catastrophic progressive defect growth. Due to ongoing uncertainty, strict mask quality monitoring in the fab is essential. The ideal reticle quality-control goal in a fab should be to detect any nascent progressive defects before they become yield-limiting.
Background
In 1953, under the tutelage of Nobel Laureate H.C. Urey, S.L. Miller conducted an experiment (now known as the Miller-Urey experiment) using a mixture of methane, water vapor, and ammonia in water in a closed reaction vessel. The vessel was then irradiated with an electric discharge. Analysis following the experiment indicated that amino acids and carboxylic acids (organic acids) formed at a high yield [1]. Since its inception, many variations of the experiment have been successfully attempted. In reality, this experiment is conducted thousands of times daily in wafer fabs worldwide. In these cases, the reactants are often the same as in the Miller-Urey experiment. The closed reaction vessel is the intrapellicle space on photomasks, and the energy source is 365, 248, or 193nm lasers.
 Figure 1. Five factors influence mask defect growth and its printability in fab. |
Historically, manufacturing experience shows that if a clean photomask is delivered to the fab, very little goes wrong with the quality of a mask during its lifetime in fab production unless there is mishandling, etc. In today’s advanced technology fabs, however, this concept is no longer valid. Industry data show that the occurrence of mask returns from fabs due to contamination defects is significant [2]. Progressive defect growth causing reticle degradation over time has become a common problem in the fabs using DUV lithography [3]. Five major factors influencing this problem are shown in Fig. 1. Traditional electrostatic discharge (ESD) defects and “migrating” defects (from noncritical to critical location on mask) continue to be a concern and should also be monitored.
Typically, most of the progressive mask defects reported to date have been on clear (or glass) areas of the mask (on the pattern side) [3-4]. In the past, work has been performed to understand backside glass defect growth as well [5]. Although these defects contained many different chemical compounds, ammonium sulfate was dominant. As a result of previous findings, much effort has been placed on eliminating sulfate and ammonium ions on mask surfaces [6-8]. Some success has occurred; however, changes to the respective processes have, in essence, changed the reactants in the intrapellicle space only to result in different reaction products. Different reactants will yield different progressive defects; this is consistent with the different variations of the Miller-Urey experiment.
 Figure 2. Inspection results show a massive (>10μm in size) growth on half-tone material of a via-level mask. |
As features on photomasks become smaller, a new generation of defects has appeared. These new defects are growing at a significantly faster rate than previous generations, mostly on chrome or on the half-tone surface of masks. Figure 2 shows a typical defect growth of this new type at its most developed phase, based on a series of inspections performed on three photomask plates.
Analyzing growth defects
Masks containing defect growth were examined with a KLA-Tencor Corp. STARlight reticle contamination inspection system. Three masks were analyzed: one chrome-on-glass (COG) gate-level mask, and two EPSMs for contact and poly levels. On-chrome/on-half tone defect growth was identified. Chemical analysis was then performed on the defective masks, using Raman spectroscopy and time-of-flight secondary ion mass spectroscopy (ToF-SIMS)
Raman spectroscopy is a light-scattering technique often referred to as the “sister” or complementary technique to infrared spectroscopy. Raman spectroscopy provides vibration data about the analyzed compounds. The technique depends on a change in the induced dipole moment or polarization to produce Raman scattering. When a beam of photons strikes a molecule, the photons are scattered elastically (Rayleigh scattering) and inelastically (Raman scattering), generating Stokes and anti-Stokes lines. The defects were analyzed using a Renishaw Model 2000 Raman spectrometer equipped with a long working-distance objective, allowing analysis through the pellicle.
 Figure 3. Inspection results from the gate level chrome-on-glass mask. |
ToF-SIMS uses a low primary-ion fluence beam to probe the outer surface monolayers (i.e., a few angstroms). The incident beam generates secondary ions, which are analyzed by a mass spectrometer. The actual damage to the sample is very low, and thus the method is “quasi-nondestructive” or static in nature. This technique provides not only elemental information about the surface, but chemical as well, in the form of fragmentation patterns of molecular species. ToF-SIMS analysis was conducted on the pattern surface of the reticle on the haze defect. The instrument used in this work was a TOF-SIMS IV secondary ion mass spectrometer from ION-TOF GmbH.
Inspection and chemical analysis results
The three masks were inspected on the STARlight system using simultaneously transmitted and reflected light. These masks were not highly defective and the defects were still in a “growing state,” i.e., still small in size. Figures 3 and 4 show the different shapes and types of defects found on these three masks.
 Figure 4. Inspection results from the contact level embedded phase-shift mask. |
Raman spectroscopy was attempted on all three plates at defect co-ordinates provided; however, the size of the defects was too small to observe or obtain spectra for the first two masks. ToF-SIMS analysis was done on defect locations and on a reference area. The third mask (EPSM poly level) showed relatively strong spectra with Raman.
 Figure 5. a) Positive and b) negative ions detected on the COG plate with ToF SIMS: nondefective vs. defective area. |
Figure 5 shows ToF-SIMS results for positive and negative ions detected on the COC plate. This indicates that defective areas of this mask contain higher hydrocarbon concentration as compared to nondefective areas. In Fig. 6, results from the Raman spectroscopy clearly show peak at cyanuric acid (1728 and 705cm-1).
Summary
Each mask type suffered from different contamination levels. The COG mask had extremely high levels of hydrocarbon contamination. High levels of fluorine contamination existed on both COG plate and contact-level EPSM; both defect areas showed higher contamination levels than on reference areas. Specific discrete defects were not found with microscopy, Raman, or ToF-SIMS imaging. However, the poly-level EPSM showed cyanuric acid on the defective areas.
 Figure 6. Raman spectroscopy results from the poly-level EPSM showing that the defects are cyanuric acid. |
Contamination levels were higher than those detected on an unexposed mask. Ammonium ion contamination levels were consistent with historical data. Sulfate ion concentrations were lower than normally detected. Sources of contamination were most likely from outgassing of the reticle shipping box, the reticle cassette at exposure tool, the reticle exposure area, and the mask cleaning processes.
The haze cycle
The possible mechanisms of these progressive defects are not straightforward. Many possible intriguing mechanisms can be involved and linked. Compounds such as ammonium oxalate are most typical of defects seen on these three plates. Possible mechanisms that could be involved in progressive mask-defect formation include:
- Ammonium hydroxide and sulfuric acid form ammonium sulfate.
- Ammonium sulfate and calcium carbonate form ammonium carbonate.
- Ammonium carbonate is a mixture of ammonium bicarbonate and ammonium carbamate.
- Ammonium carbonate can react with ammonia gas to form ammonium oxalate.
- Ammonia gas and carbon dioxide can form ammonium carbamate.
- Carbon dioxide and water can form oxalic acid.
- Oxalic acid and ammonia gas can form ammonium oxalate.
- Ammonia gas, carbon dioxide, and water can form ammonium oxalate.
Conclusion
Crystal growth and haze formation on reticles continue to be significant sources of concern for the semiconductor industry. Possible sources, causes, and formation mechanisms are being investigated. Considerable work toward reducing sulfate ions from mask surfaces during the maskmaking process has occurred, mainly due to the knowledge that ammonium sulfate was the dominant defect growth during the last few years of ArF 193nm lithography. But even on low-sulfate masks (such as those discussed in this article), a new hydrocarbon-rich (organic) defect growth was observed. This new generation of organic growth needs to be taken seriously because its growth rate and opaqueness is much higher than with defects previously seen. It is important to note that the intrapellicle space acts as a reaction chamber and that merely changing cleaning processes does not eliminate the possibility of a different type of crystal growth.
Ideally, quality improvement would require eliminating the source of defect signatures. The next photomask incoming quality-control goal should be to detect defect growth using a high-resolution mask inspection system as the defects are just starting to form but are not yet yield-limiting. Masks should be monitored on an established frequency so problem masks can be removed from production and sent to rework prior to affecting device performance and fab yields. A carefully developed mask requalification inspection strategy should be implemented to optimize mean time to detect any defect growth showing this progressive growth signature.
Acknowledgments
STARlight is a registered trademark of KLA-Tencor Corp., and TOF SIMS IV is a trademark of ION-TOF GmbH.
References
- S. Miller, “A Production of Amino Acids under Possible Primitive Earth Conditions,” Science, Vol. 117, pp. 528-529, 1953.
- K. Kimmel, “Mask Industry Assessment: 2002,” Proc. SPIE, 22nd Annual BACUS Symp. on Photomask Technology, Vol. 4889, pp. 1-14, 2002.
- K. Bhattacharyya, W. Volk, B.J. Grenon, D. Brown, J. Ayala, “Investigation of Reticle Defect Formation at DUV Lithography,” Proc. SPIE, 22nd Annual BACUS Symp. on Photomask Technology, Vol. 4889, pp. 478-487, 2002.
- B.J. Grenon, C.R. Peters, K. Bhattacharyya, “Tracking Down Causes of DUV Subpellicle Defects,” Solid State Technology, p. 159, June 2000.
- B.J. Grenon, K. Bhattacharyya, W. Volk, K. Phan, A. Poock, “Reticle Surface Contaminants and their Relationship to Subpellicle Defect Formation,” Proc. SPIE, Metrology, Inspection, and Process Control for Microlithography XVIII, Vol. 5375, pp. 355-362, 2004.
- F. Eshbach, D. Tanzil, M. Kovalchick, U. Dietze, M. Liu, et al., “Improving Photomask Surface Properties through a Combination of Dry and Wet Cleaning Steps,” Proc. SPIE, Photomask and Next-Generation Lithography Mask Technology XI, Vol. 5446, pp. 209-217, 2004.
- H. Ishii, A. Tobita, Y. Shji, H. Tanaka, A. Naito, et al., “Root Cause for Crystal Growth at ArF Excimer Laser Lithography,” Proc. SPIE, Photomask and Next-Generation Lithography Mask Technology XI, Vol. 5446, pp. 218-224, 2004.
- C.H. Shiao, C.C. Tsai, T. Hsu, S. Tuan, D. Chang, et al., “Evaluation, Reduction, and Monitoring of Progressive Defects on 193nm Reticles for Low-k1 Process,” Proc. SPIE, Photomask and Next-Generation Lithography Mask Technology XI, Vol. 5446, pp. 225-230, 2004.
Kaustuve Bhattacharyya is a senior applications manager at KLA-Tencor Corp., 160 Rio Robles, San Jose, CA 95134; ph 518/526-4861, e-mail [email protected].
Benjamin Eynon is senior director of marketing for KLA-Tencor’s Reticle and Photomask Inspection Division.
Mark Eickhoff is a senior staff field applications engineer at KLA-Tencor.
Brian J. Grenon is a consultant at Grenon Consulting Inc., 92 Dunlop Way, Colchester, VT; ph 802/862-4551, fax 802/658-8952, e-mail [email protected].