Techniques for removal of contamination from EUVL mask without surface damage Sherjang Singh a*, Ssuwei Chen a, Tobias Wähler b, Rik Jonckheere c Ted Liang d, Robert J. Chen d, Uwe Dietze a a HamaTech APE USA, 1909 W. Braker Lane, Austin, TX 78758, USA b HamaTech APE, Ferdinand-von-Steinbeis-Ring 10, 75447 Sternenfels, Germany c IMEC vzw, Kapeldreef 75, B3001 Leuven, Belgium d Intel Corp., SC1-03, 2200 Mission College Blvd, Santa Clara, CA 95054, USA ABSTRACT Mask defectivity is an acknowledged road block for the introduction of EUV lithography (EUVL) for manufacturing. There are significant challenges to extend the conventional methods of cleaning developed for standard 193nm optical photomask to meet the specific requirements for EUV mask structure and materials. In this work, the use of UV activated media for EUV mask surface cleaning is evaluated and the effects on Ru capping layer integrity are compared against conventional cleaning methods. Ru layer surface is analyzed using roughness measurements (AFM) and reflectivity changes (EUV-R and optical). Key words: Capping layer integrity, contamination removal, surface restoration, EUV mask cleaning 1. INTRODUCTION With EUVL approaching high volume manufacturing (HVM) [1], mask defectivity remains one of the obstacles to commercial viability [2,3]. Key to overcoming this is development of a mask clean process that is effective for defect removal and preserves the integrity of the mask surface. Damage to the Ru capping surface degrades EUV reflectivity which can lead to critical dimension (CD) shift and non-uniformity [4] It has been shown that conventional 193i reticle cleaning processes will not be able to meet EUV reticle requirements. DIO 3 (ozonated water) for example significantly degrades the Ru surface due to galvanic corrosion and oxidation [5]. Standard SPM (H 2 SO 4 + H 2 O 2 ) and SC1 (NH 4 OH+H 2 O 2 +DI), while considered safe for Ru integrity [6], contributes a large number of process adders [7]. Furthermore, residual molecular contamination from acid- based processes has been identified as a major source of haze formation (progressive defects) in 193i lithography [8], and is expected to present the same problem for EUVL [9]. For these reasons, alternative cleaning techniques must be developed. viously, HamaTech demonstrated techniques for organic removal, surface preparation and residual ion removal without surface damage on the 193i masks [10-12]. These new techniques are based on POU UV exposure of the wet cleaning chemistry and the mask surface simultaneously. In this work, the effect of these techniques is evaluated on the surface integrity of EUVL mask capping layer (Ru). 2. EXPERIMENTAL 2.1 Materials To investigate the effects of various cleaning steps on surface integrity, as-deposited Ru capped EUV ML blanks were used. The ML blanks used were deposited with 2.5nm Ru on 4.1nm Si-top Mo/Si multilayer over a quartz substrate. As an exception, for studying carbon removal, a 3 nm carbon layer was intentionally deposited using magnetron sputtering on a blank that had 2.5nm Ru layer deposited on 8nm Si-top Mo/Si multilayer. To study particle removal efficiency, particles were deposited on the surface by dispensing a solution of sulfuric acid and hydrogen peroxide followed by a DI-water rinse. Such a mixture is known to be contaminated with a variety of inorganic (such as metallic) and organic particles. This was intended to create process-caused adders normally observed from conventional SPM-based cleaning. Extreme Ultraviolet (EUV) Lithography, edited by Bruno M. La Fontaine, Proc. of SPIE Vol. 7636, 76360Y 2010 SPIE CCC code: 0277-786X/10/$18 doi: 10.1117/12.847026 Proc. of SPIE Vol. 7636 76360Y-1
Such adders are believed to be more difficult to remove than those from handling and storage. To ensure stronger adhesion between the surface and the particles, the particles were allowed to age for one week. 2.2 Experimental All tests were performed using the POU UV exposure method. For the surface preparation test, DI-water was used as the medi For organic removal and ion removal, DIO 3 and hot DIW (DI-water) were used respectively. All the tests were also performed with conventional cleaning technologies for comparison. Other details on the tests are summarized in table 1. Test Cleaning Step Media Concentration/ UV Process Time No. Condition Activation 1 DIW Yes Surface paration n/a 2 N2 Gas 172nm VUV 10 T o 3 Yes High 4 Organic Removal DIO 3 No 5 T o 5 Low Yes 9 Temp = T o C Ion Removal W 10 Temp = (T-10) o C Off 5 T o 11 Carbon Removal DIW n/a 2 T o Table 1: Table showing process condition summary of various tests performed for different cleaning steps using UV activation method and its comparison with conventional method. Process times longer than baseline (T o ) were used for accelerated effects. 2.3 Characterization Methods Ru integrity was characterized with an atomic force microscope (AFM) to determine surface roughness, a EUVreflectometer (EUV-R) was used to measure reflectivity at EUV wavelength, and an optical reflectance instrument was utilized for reflectivity measurements above 190nm. In addition, a Lasertec M1350 blank inspection system was used to evaluate the effectiveness of defect removal. Several sites on each sample were measured before and after each experiment. There was a 1 measurement uncertainty in AFM surface roughness (corresponding to the noise floor of the instrument) and a 0.25 % measurement error in absolute EUV-reflectivity. 3. RESULTS & DISCUSSION 3.1 Effect of DIO 3 photolysis on surface integrity (Organic Removal Technique) Figures 1 (a, b), shows a comparison of the effect of POU (Point Of Use) UV photolyzed DIO 3 process and a conventional DIO 3 process (no UV) on surface roughness and EUV-reflectivity. 2 1 Relative Change in Roughness for 5 X process (Initial roughness same for all) Off Low Conc. 2.5 2.0 Absolute Change in EUV Reflectivity (5 X process) Off Low Conc. Figure 1: Plot showing comparison between effect of photolyzed DIO 3 process and conventional DIO 3 process on a) Ru surface roughness and b) EUV reflectivity. Photolyzed DIO 3 process shows negligible change in surface roughness and EUV reflectivity whereas conventional DIO 3 process damages the surface significantly. Proc. of SPIE Vol. 7636 76360Y-2
Results for the conventional DIO 3 process support the literature findings [5], i.e., there is significant surface degradation. However, photolyzed DIO 3 process does not produce significant change in surface roughness and EUV reflectivity. This is attributed to the fact that DIO 3 photolysis decomposes ozone to produce hydroxyl radicals, which react primarily under the hydrogen abstraction mechanism [13,14]. viously, it was observed that organic removal rates were an order of magnitude faster with this method compared to conventional DIO 3 processes [10]. Figure 2 shows the optical reflectivity of the Ru surface before and after the same DIO 3 tests as discussed above. Optical reflectivity results correlate well with the changes in surface roughness and EUV-R in Figure 1. Surface treatment with conventional DIO 3 process results in a significant drop in optical reflectivity, whereas the photolyzed DIO 3 process shows no change. 8 6 DIO3 (UV on) 8 6 c. Figure 2: and post Optical reflectivity plots for a) High conc. DIO 3 (UV on) b) High conc. DIO 3 (UV off) c) Low conc. DIO 3 (UV on). Conventional DIO 3 process shows a signficant drop in optical reflectivity. DIO3 (UV off) 8 6 Low Conc. DIO3 (UV on) 3.2 Effect of UV activated DI-water (Surface paration Technique) viously, it was reported that UV-exposed DI water is equally effective as conventional dry 172nm UV treatment in preparation of the mask surface for wet cleaning [12]. Here, the technique of exposing a Ru blank to DI-water and UV (wavelength > 200nm) simultaneously at POU is evaluated and the effect on surface roughness and reflectivity is determined. Figure 3 compares the surface impact of this process with a traditional dry 172nm UV exposure under N 2. 2 1 Relative Change in Roughness for 10 X process (Initial roughness same for all) DI-Water + POU UV N2 Gas + 172nm VUV 2.5 2.0 Absolute change in EUV Reflectivity (10 X process) DI-Water + POU UV N2 Gas + 172nm VUV Figure 3: Plot showing comparison between effect of new surface preparation process and 172nm UV process on a) Ru surface roughness and b) EUV reflectivity. UV-exposed DI process shows no change in roughness and EUV reflectivity. Whereas the 172nm UV process results in a slight increase in roughness and small drop in EUV reflectivity, the new process shows no change in either. Optical reflectance changes were equivalent for both processes (data not shown). 3.3 Effect of UV activation with Hot DI-water (Ion Removal Technique) Ion removal with conventional W rinse is achieved at a cost to cleanliness. This process is known to add particles due to contamination from the quartz heater system when used set at a high enough temperature [15]. Lowering the heater temperature setpoint significantly reduces adders. viously, the use of UV exposed DI water to lower the heater temperature while preserving the ion removal capability was investigated [16]. With this method, heater temperature could be lowered by 10 o C and remain equally effective. In fact, due to the UV activation, this process was found to be 77% more effective at a lower heater temperature than the conventional W process. Proc. of SPIE Vol. 7636 76360Y-3
Figure 4 shows the effect of this method on Ru surface roughness and EUV reflectivity and is compared with regular W process. Results show that both the processes are equally safe for Ru capped multi-layers. There is no significant change observed in surface roughness and EUV reflectivity. Optical reflectance data also remained unchanged (data not shown). 2 1 Relative Change in Roughness for 5 X process (Initial roughness same for all) UV Off UV 2.5 2.0 Absolute change in EUV Reflectivity (5 X process) UV Off UV Figure 4: Plot showing comparison between effect of UV assisted process (at T-10 o C) and regular (at T o C) process a) Ru surface roughness and b) EUV reflectivity. Both processes show no effect on surface. 3.4 Effect of full cleaning process A complete cleaning sequence of POU UV DIW + POU UV photolyzed DIO 3 + megasonic cleaning was performed on a Ru capped EUVL blank. Figure 5.a shows that the surface roughness change is negligible for the full process. Figure 5 (b&c) indicate that there is no change in EUV and optical reflectivity after the same process. Thus it is demonstrated that this process is capable of preserving the Ru surface integrity. 2 1 Relative change in Roughness, 1 X process (Initial roughness pristine) Full Cleaning Process 2.5 2.0 Absolute change in EUV-R (1 X process) Full Cleaning Recipe Optical Reflectivity Change for 1 X process c. Figure 5: Plot showing effect of a full cleaning process on a) Ru surface roughness b) EUV reflectivity c) Optical reflectivity. The full cleaning process is capable of preserving the Ru layer integrity. 8 6 Full cleaning process Furthermore, particle removal tests indicate greater than 95% removal effectiveness on deposited defects (Figure 6). 3.5 Carbon Removal Figure 7a shows the reflectivity measurements on Ru blank before carbon deposition, after carbon deposition and after POU UV activated DIW cleaning process. Figure 7.b displays the EUV reflectivity spectra for these three conditions. The reflectivity is dropped after carbon deposition by ~ 1.9%. The POU UV DIW treatment increases the reflectivity by ~1.5 %. The reflectivity is partially restored because the cleaning process was carried for a limited time; longer exposures are needed for complete removal. Proc. of SPIE Vol. 7636 76360Y-4
Figure 6: a) Lasertec defect map of deposited particles on the Ru surface. b) Lasertec defect map of Ru surface post cleaning process showing un-removed deposited particles. Absolute EUV-R (%) 52.5 52.0 5 5 5 5 49.5 49.0 EUV reflectivity after different treatments Clean Blank (no carbon) 3nm Carbon Deposition Cleaning with UV-DI EUV Reflectivity (%) 60 50 40 30 20 10 0 After UV-DI Clean Carbon Deposition No carbon 13 13.5 14 EUV Reflectivity (%) 54 53 52 51 50 49 48 47 46 13 13.5 14 Figure 7. ) Plot showing EUV reflectivity for a pristine blank, EUV reflectivity after carbon deposition and after cleaning with POU UV DIW process. b) shows the EUV reflectivity spectra for the three different conditions. POU UV DIW process demonstrates that the reflectivity can be restored after cleaning, signifying carbon removal. 4. CONCLUSIONS Conventional DIO 3 cleaning significantly increases surface roughness and decreases EUV reflectivity. In comparison, there is no change in either with the DIO 3 photolysis process proposed here. Surface preparation with >200nm wavelength POU UV exposure and DIW is also demonstrated to be equal or better than the conventional dry 172nm UV exposure. Lastly, POU UV activated warm DIW is shown to be as safe for the Ru surface as conventional W Proc. of SPIE Vol. 7636 76360Y-5
alone. Together, the complete cleaning process yields >95 % PRE with negligible surface damage as determined by surface roughness and EUV/optical reflectivity. ACKNOWLEDGEMENT Authors would like to acknowledge Kurt Ronse (IMEC), Bart Baudemprez (IMEC), Dr. Rahim Farouhi (n&k Technology Inc.), and Dr. Peter Dress (HamaTech APE) for their valuable support. Other co-workers at HamaTech APE, IMEC and Intel Corp. are also acknowledged for their help during various stages of this study. REFERENCES 1. Levinson, H.J., Extreme ultraviolet lithography s path to manufacturing, J. Micro/Nanolith. MEMS MOEMS, 8 (4), 041501 (2009) 2. Liang T. et al, EUV Mask Pattern Defect Printability, BACUS News, 22 (10), pp. 1-11(2006), 3. Jonckheere, R. et al, Investigation of EUV Mask Defectivity via Full-Field Printing and Inspection on Wafer, SPIE Proc. Vol. 7379-26 (2009) 4. Jonckheere, R et al, Investigation of mask defectivity in full field EUV lithography, SPIE Proc. Vol. 6730-37 (2008) 5. Shimomura, T. et al, Chemical durability studies of Ru-capped EUV mask blanks, SPIE Proc. Vol. 7122, pp. 712226 (2008) 6. Yan, P.Y. et al, Characterization of ruthenium thin films as capping layer for extreme ultraviolet lithography mask blanks, J. Vac. Sci. Technol. B, Vol. 25 (6), pp. 1859-1866 (2007) 7. Kapila, V. et al, A method to determine the origin of remaining particles after mask blank cleaning, SPIE Proc. Vol. 6730, pp. 67304L (2007) 8. Kalk, F. et al, Photomask Defectivity and Cleaning: A New Milieu, Semiconductor International (2007) 9. Sengupta, A. et al, Is HVM EUV Mask Micro-contamination a Significant Risk to Mask Lifetime, Sematech - 6th Annual Mask Cleaning Workshop, Monterey, CA (2009) 10. Singh, S. et al, Study on surface integrity in photomask resist strip and final cleaning processes, SPIE Proc. Vol. 7379-12 (2009) 11. Singh, S. et al, Automated imprint mask cleaning for Step-and-Flash Imprint Lithography, SPIE Proc. Vol. 7271-88 (2009) 12. Singh, S. et al, An advanced method to condition and clean photomask surfaces without damage Sematech - 6th Annual Mask Cleaning Workshop, Monterey, CA (2009) 13. Ledakowicz, S. et al, Oxidation of PAHs in water solution by ozone combined with ultraviolet radiation International Journal of Photoenergy, Vol. 3, 95-101, (2001) 14. Legrini, O. et al, Photochemical Processes for Water Treatment Chem. Rev., 93, pp. 671-698, (1993) 15. Shimomura, T. et al, 50nm particle removal from EUV mask blank using standard wet clean, Sematech - 6th Annual Mask Cleaning Workshop, Monterey, CA (2009) 16. Singh, S. et al, serving the mask integrity for the lithography process Photomask Japan 2010, 17th International Symposium on Photomasks and NGL Mask Technology, Paper No. 00074, (2010) Proc. of SPIE Vol. 7636 76360Y-6