We present first damage threshold investigations on EUV mirrors and substrate materials using a table-top laser produced plasma source. A Schwarzschild objective with Mo/Si multilayer coatings for the wavelength of 13.5 nm was adapted to the source, generating an EUV spot of 5 µm diameter with a maximum energy density of ~6.6 J/cm2. Single-pulse damage tests were performed on grazing incidence gold mirrors, Mo/Si multilayer mirrors and mirror substrates, respectively. For gold mirrors, a film thickness dependent damage threshold is observed, which can be partially explained by a thermal interaction process. For Mo/Si multilayer mirrors two damage regimes (spot-like, crater) were identified. Fused silica exhibits very smooth ablation craters, indicating a direct photon-induced bond breaking process. Silicon shows the highest damage threshold of all investigated substrate and coating materials. The damage experiments on substrates (fused silica, silicon, CaF2) were compared to excimer laser ablation studies at 157 nm.
©2010 Optical Society of America
Damage studies on optics and laser crystals are performed ever since first lasers were able to generate higher fluences than optical elements could tolerate without being damaged. For optical materials of IR, VIS and UV lasers, the underlying material interaction mechanisms have been studied extensively . For several years now, also EUV and XUV sources are constantly under development, mainly employed for next-generation lithography (laser-generated or discharge plasma, nanosecond pulses), or fundamental research (free-electron-laser, femtosecond pulses) [2–4]. This wavelength regime requires new optical elements like grazing-incidence or multilayer mirrors. Grazing incidence mirrors are based on total external reflection, resulting in very low incidence angles with respect to the surface. In many cases thin high-Z metal layers (e.g. gold) are used, for instance as collector optics in EUV lithography, or for imaging and beam steering of free-electron-lasers, covering a broad XUV wavelength range. For more normal reflection angles, multilayer mirrors are employed, consisting of very thin alternating layers, especially Molybdenum and Silicon for the wavelength of 13.5 nm. These mirrors are currently being optimized in terms of thermal resistance and reflectivity close to the theoretical limit. However, damage threshold investigations for these optical elements are still lacking.
Previous work at the Laser-Laboratorium Göttingen was focused on the generation of high fluences at λ=13.5 nm, using a table-top EUV source. This setup was employed to investigate the EUV ablation process and the structuring of various polymers [5,6]. Recent changes led to the generation of energy densities up to 6.6 J/cm2 at pulse durations of 8.8 ns, which is sufficient to perform also damage experiments on EUV optics and substrates.
In this paper we present investigations on EUV induced damage thresholds of grazing incidence mirrors based on gold films, as well as Mo/Si multilayer mirrors. Furthermore, mirror substrate materials as fused silica, silicon and CaF2 were examined with respect to their single-pulse damage thresholds, corresponding radiation induced surface morphologies, and ablation rates. As to our knowledge, these results represent the first damage experiments on EUV optics using nanosecond radiation at a wavelength of 13.5 nm.
2. Experimental setup
2.1. EUV source
The experimental setup consists of a laser-based EUV/XUV source and a separate optics chamber adapted to this source (cf. Figure 1 ) [7–9]. EUV radiation is generated by focusing a Nd:YAG laser (Innolas, wavelength 1064 nm, pulse energy 700 mJ, pulse duration 8.8 ns) onto a target. For small EUV plasmas with high brilliance, a solid Au target (200 µm thick Au foil fixed on a copper rod) is used, yielding a plasma diameter of ~50 µm (FWHM).
In order to achieve high EUV fluences, a Schwarzschild objective with a magnification of 0.102 at a maximum numerical aperture of 0.4 was employed . It consists of two spherical, annular mirror substrates coated with a Mo/Si multilayer system (reflectivity R ~0.65 per mirror at 13.5 nm). A plane mirror (Au coated Silicon wafer, R ~0.675 for grazing angle of 10 degree)  is positioned between source and objective in order to protect the Mo/Si multilayers against contamination from debris of the laser plasma. Due to a significant reflectivity of the Mo/Si mirrors also at wavelengths above 100 nm, a thin zirconium foil fixed on an etched steel mesh can be inserted in the beam path, blocking effectively all radiation above 25 nm . The transmission of this zirconium filter at λ=13.5 nm is ~ 0.175, as measured with a separate setup accomplishing EUV reflectivity, transmission and absorption measurements .Table 1 compiles selected properties of the employed EUV source and optics system as used in the experiments. Using a spherical laser focus on the gold target, EUV energy density saturates at 0.72 J/cm2 due to overheating of the plasma by the incoming laser energy . Further enhancement of the fluence was achieved by use of a line focus rather than a spherical focus on the solid target: The laser energy was dispersed in one direction with the help of cylindrical lenses, leading to a maximum EUV fluence of 1.16 J/cm2 (6.6 J/cm2 without Zirconium filter) in the focus of the EUV objective.
Due to the numerical aperture of this objective, the incidence angles on the sample range from 12.7 deg to 26.6 deg (mean angle ~ 20 deg). Therefore, theoretical penetration depths are calculated for this incidence angle.
3.1. Damage test of grazing incidence gold mirrors
In order to investigate the EUV damage resistance of grazing incidence metal mirrors, a 69 nm thick gold layer on a BK7 glass substrate was tested with the described setup. A 1-on-1 damage test according to ISO11254  was performed, irradiating 10 positions with single EUV pulses at constant fluence. The number of damaged positions, normalized to the total number of irradiated sites, corresponds to the damage probability, which has to be determined for different EUV energy densities . The result is displayed in Fig. 2 , showing threshold energy densities of 420 mJ/cm2 for surface damage and 650 mJ/cm2 for complete film removal, respectively. Obviously, from the damage morphology (AFM images in Fig. 2, right) a thermally induced damage mechanism can be deduced.
The experiment was repeated for samples with different film thicknesses. As displayed in Fig. 3 (left), a linear relation between damage threshold fluence and the thickness of the Au film (measured by AFM) could be observed. This linear behaviour is already well-known for excimer laser ablation (248 nm) of metal films  and can be generally explained by a photo-thermal model: a surface modification results if the temperature of the film rises up to a value where melting (visible damage) or evaporation (film removal) occurs. For thin films (thickness d < thermal diffusion length of film Lth.f) with rapid heat diffusion, the threshold fluence is linear to the film thickness, reaching a plateau for d ≥ Lth,f.
From this thermal model, the theoretical threshold fluence can be calculated . The resulting graph for melting of a gold layer on fused silica is displayed in Fig. 3 (right, curve (a)), taking into account a complete conversion of radiative energy into heat, as well as 100% absorbance. Bulk material constants were used for this calculation.
Obviously, the slope of the theoretical curve is much lower than that of the measured data. The main reason for this difference can be attributed to the small spot size: the radiative energy is applied on a very small area (approximately 8.4 µm2) of the gold film (Table 1); thus, within the EUV pulse duration, the heat generated by the absorbed EUV photons spreads out into the surrounding film material. For the applied nanosecond EUV pulse (27 ns, 10%) the corresponding heat diffusion length is 2.64 µm, which is larger than the radius of the EUV focus (Table 1). Taking heat dissipation and focus size into account, a 7.3 times higher EUV energy density has to be applied to attain the melting (or evaporation) temperature in the focus (Fig. 3, right, (b)).
Nevertheless, the measured thresholds are still ~2.3 times larger than the calculated values (Fig. 3 left (b), compared to the best fit (c)). The reason for this discrepancy might be a dissipation of radiative energy into alternative, non-thermal loss channels (e.g. photo-electrons, photo-fluorescence, scattered light, excitation of core-level electrons), being quite conceivable in regard of the photon energy of 91.85 eV. Another reason could be the fact that the thermal film constants (e.g. heat conductance, specific heat) to be applied for the theoretical curve are expected to be different from the used bulk material data.
We can conclude that the interaction of EUV radiation with thin gold films is primarily governed by thermal effects, showing a linear relation between ablation threshold and film thickness. Additional non-thermal loss channels will be the topic of future investigations.
3.2. Damage to Mo/Si multilayer optics
Single-pulse damage experiments were also conducted on Mo/Si mirrors. Figure 4 shows the 1-on-1 damage probability plot obtained for a system consisting of 60 Mo/Si double layers, designed for a maximum reflectivity at 20° incidence angle (without interdiffusion barriers and additional capping layers).
At fluences larger than 800 mJ/cm2 spot-like damage occurs (Fig. 4 right, DIC image 1), merging into crater damage at 1.7 J/cm2. In all cases small craters with depths of 40-70 nm are generated, mostly with a small bump in the center. This is similar to experiments performed with EUV femtosecond pulses at FLASH (Free electron laser Hamburg), where very similar structures at depths of ~60 nm were observed . In case of spot-like damage, several micrometer-sized pits are detectable, independent from the plasma intensity. This effect may be attributed to inhomogenities or defects in the multilayer structure, leading to locally higher absorption and therefore lower damage threshold.
At higher fluences, the irradiated multilayer surface is damaged completely, resulting in a single crater. From the surface morphology the damage mechanism is apparently driven by thermal heating of the multilayer structure. Therefore, as observed before for the gold layers, the thermal diffusivity within the pulse duration might play a significant role for the determination of the effective threshold fluence. According to Ref . compaction of the multilayer mirror due to intermixing of Molybdenum and Silicon might also be involved in the damaging process.
4. Damage to mirror substrate
The majority of EUV optical elements consist of thin films or multilayers deposited on a substrate (e.g. fused silica or silicon); thus, it is worthwhile to investigate also substrate damage by EUV pulses. Since the fraction of photons penetrating through the coating and reaching the substrate is usually quite low, especially long-term degradation effects like oxidation or compaction have to be considered . However, also single-pulse damage thresholds have to be known in order to make sure that the substrate does not limit the usable fluence range. As an example, fused silica, silicon and calcium fluoride crystals (CaF2) were investigated.
4.1. Fused silica
For determination of damage threshold for fused silica, a Suprasil 1 sample (Heraeus, 2 mm thickness) was used. The sample was cut into 1 x 1 cm2 pieces and cleaned with ethanol.
In Fig. 5 (left) measured ablation depths are displayed as a function of the number of applied EUV pulses for selected fluences. Obviously, a linear behaviour is observed for low fluences. However, at fluences >5.4 J/cm2, the crater depths for 10 pulses are deeper than expected from the linear fit for lower pulse numbers, indicating the beginning of incubation effects. From the linear fit of the crater depths as measured with AFM, the ablation rate (depth per pulse) is determined (Fig. 5, right). Similar to excimer laser studies [18,19], the ablation rate d may be fitted by a logarithmic function according to d = αeff−1 ln (H / Ht). A corresponding best-fit yields an effective absorption coefficient αeff = 0.0455 nm−1 and an ablation threshold fluence Ht = 3.22 J/cm2.
It is important to note that the absolutely smooth crater profile (see AFM micrograph) is comparable to laser ablation reported for 157 nm radiation with nanosecond pulses . As for 157 nm, this might be an indication for a photon-induced direct bond-breaking process, in contrast to a more thermally induced mechanism at wavelengths > 190 nm . In Table 2 the fused silica ablation characteristics are compiled for 13.5 nm and 157 nm, respectively. Ablation rates are in the range of 2-15 nm/pulse. However, the determined threshold fluence for EUV is almost 3 times higher than for 157 nm (1.1 J/cm2) .
This difference might be explained by the absorption processes at the two wavelengths. For 157 nm radiation, the photon energy of 7.9 eV is smaller than the bandgap energy of 9.3 eV , resulting in relatively weak absorption for low fluences. Studies  indicate that formation of excited-state and photo dissociation of Si-O bonds, resulting in creation of SiOx (x < 2) during exposure), seems to be responsible for the strong decrease of attenuation length and therefore to an increase in the absorption coefficient by a factor of 17,000 . In contrast, 13.5 nm radiation is mainly absorbed by excitation of core electrons, since the photon energy (91.85 eV) is close to the K-edge of silicon and the L-edge of oxygen, respectively . Here, photo-fluorescence, the generation of photoelectrons or electron-collision induced ionization occurs, leading not necessarily to bond breaking or damage of fused silica. Additionally, the photon energy of 13.5 nm radiation (91.85 eV) is slightly lower than the absorption edge energy of the L-shell (L2: 99.8 eV; L3: 99.2 eV) , therefore the main part of the incident radiation energy will be absorbed by the oxygen atoms. The generation of SiOx wrong bonds would possibly not lead to a significant increase of absorption for 13.5 nm radiation as established for 157 nm, resulting in an effective absorption coefficient closer to the theoretical one (cf. Table 2).
Moreover, although the crater morphology in fused silica indicates a more photochemical dominated ablation process, also thermal diffusivity of the material in correlation with the micron sized EUV spot might influence the threshold fluence value. As shown above for the ablation of gold films, this would increase the threshold fluence.
The observed ablation parameters do not agree well with the work of Makimura , who found rates of 40 nm/pulse at an EUV fluence of roughly 0.1 J/cm2. However, this much higher rate may result from the fact that the experiments were performed with a broad-band emitting EUV source focused by a grazing incidence mirror, leading to a broad wavelength spectrum on the sample. Thus, due to the nearby absorption edges of silicon, the penetration depth is much smaller than in our experiments, and silicon might be the main absorbing species.
The EUV damage behaviour of standard silicon wafers (Crystec, thickness 625 µm, orientation (100), single side polished) was investigated. Before irradiation, wafers were cut into 10 mm x 10 mm pieces and cleaned in ethanol. 1-on-1 damage threshold values were determined as described before for the gold and Mo/Si coatings. The result is shown in Fig. 6 together with corresponding AFM micrographs.
From the surface morphology, two damaging processes can be separated. Surface damage starts at 4.1 J/cm2 (Fig. 6, squares), leading to a change in surface roughness and partial material removal within the first 1-2 nm of the silicon substrate only. At fluences above 5 J/cm2 ablation craters with depths of the order of 100 nm are formed (Fig. 6, circles).
The surface change within the first nanometers can be explained by a roughening and partial removal of the SiOx layer present on the Si surface due to natural oxidation. Such a 3-7 nm thin layer of silicon oxide  has a ~6 times higher absorption for 13.5 nm radiation than that of the silicon substrate (attenuation lengths SiO2: 93 nm; Si: 553 nm) . The damage threshold of 4.1 J/cm2 is higher than that of bulk fused silica measured before (SiO2: 3.22 J/cm2, Tab. 2), which might be explained by lower absorption in the silicon oxide layer than in bulk SiO2 and a lower concentration of the absorbing oxygen in the film (SiOx, x<2), respectively. Moreover, assuming a thermally supported ablation process, the higher damage threshold can also be caused by the much better thermal conductance of silicon as compared to fused silica.
In contrast to fused silica, the observed crater morphology with depths of more than 100 nm and molten material at the rim indicates a thermally induced damaging mechanism for silicon. Craters are generated at fluences above 5 J/cm2 EUV, which is close to the upper limit of our setup; thus, a correct determination of the damage plot up to 100% damage probability was not possible.
4.3. Calcium fluoride
For calcium fluoride (CaF2) the EUV damage morphology is different from that of fused silica. Above the threshold, small crystallites can be observed at the rim of the irradiated area, with a deeper structure in the center of the damaged site (Fig. 7 , inset). This irregular damage profile leads to fluctuations in the determination of the crater depth, and thus a large scatter of the ablation rates (cf. Figure 6). The evaluated ablation rates are apparently more than 4 times higher (up to 60 nm per pulse) than in the case of fused silica.
From the logarithmic fit a threshold fluence of 2.21 J/cm2 can be determined, which is lower than that for fused silica. An effective absorption length of 62.9 nm was calculated, which is slightly higher than the theoretical absorption length of 48.1 nm . Apart from the observed surface damage also a coloration of the material can be observed by optical microscopy. Starting already at fluences > 1.2 J/cm2 a darkening of CaF2 occurs without any change of the surface morphology. This is probably caused by EUV-generated colour centres, as demonstrated before for LiF crystals .
In this paper we have presented first damage threshold investigations on EUV optical materials using a nanosecond table-top EUV source. With the help of an integrated Schwarzschild objective, a micro-focus of ~ 2 µm x 5µm in size and a maximum energy density of 6.6 J/cm2 at 13.5 nm wavelength could be generated, offering the possibility for 1-on-1 damage tests on EUV mirrors and substrates.
Single-pulse damage experiments on EUV optics are presented, in particular grazing incidence gold mirrors and Mo/Si multilayer mirrors. The results are compiled in Fig. 8 . For gold films, a linear behaviour between threshold fluence and film thickness could be determined. This indicates a thermally dominated process, substantiated by AFM images of the mirror surface. For Mo/Si multilayer mirrors two damage regimes are observed: At 0.8 J/cm2 spot-like damage occurs, merging into crater damage at around 1.7 J/cm2. Fused silica shows a higher damage threshold (3.22 J/cm2) than the reflective coatings, leading to very smooth ablation craters. For silicon the highest damage threshold fluences (4.1 – 5 J/cm2) were determined, mainly caused by the low absorption of 13.5 nm radiation. The observed surface damage with a depth of 1-2 nm may be attributed to a thin natural oxide layer. For CaF2, surface damage starts at 2.5 J/cm2, indicating a recrystallization process on the surface.
Our experiments demonstrate that single-pulse EUV damage thresholds of substrate materials are higher than those of thin reflective coatings, as well known already for higher UV and visible wavelengths. Thus, metal or multilayer films immediately exposed to EUV radiation will clearly be the limiting factor for high fluence applications of 13.5 nm radiation.
Further studies on EUV material interactions are being planned, comprising also multiple-pulse damage effects.
The financial support by the “Deutsche Forschungsgemeinschaft” within the Sonderforschungsbereich 755 “Nanoscale Photonic Imaging” is gratefully acknowledged.
References and links
1. Proceedings of SPIE, Boulder Damage Symposium (1996–2009).
2. W. Banqiu and K. Ajay, “Extreme ultraviolet lithography: A review,” J. Vac. Sci. Technol. B 25(6), 1743–1761 (2007). [CrossRef]
3. D. Hiroyuki, “Review of soft x-ray laser researches and developments,” Rep. Prog. Phys. 65(10), 1513–1576 (2002). [CrossRef]
4. J. R. Schneider, “Properties and scientific perspectives of a single pass X-ray free-electron laser”, Nucl. Instr. and Meth. in Phys Res. A 398(1), 41–53 (1997). [CrossRef]
5. F. Barkusky, C. Peth, A. Bayer, and K. Mann, “Direct photo-etching of poly(methyl methacrylate) using focused extreme ultraviolet radiation from a table-top laser-induced plasma source,” J. Appl. Phys. 101(12), 124908 (2007). [CrossRef]
6. F. Barkusky, A. Bayer, C. Peth, and K. Mann, “Direct photoetching of polymers using radiation of high energy density from a table-top extreme ultraviolet plasma source,” J. Appl. Phys. 105, 014906 (2009). [CrossRef]
7. S. Kranzusch and K. Mann, “Spectral characterization of EUV radiation emitted from a laser-irradiated gas puff target,” Opt. Commun. 200, 223 (2001). [CrossRef]
8. C. Peth, S. Kranzusch, K. Mann, and W. Viöl, “Characterization of gas targets for laser produced extreme ultraviolet plasmas with a Hartmann-Shack sensor,” Rev. Sci. Instrum. 75(10), 3288 (2004). [CrossRef]
9. S. Kranzusch, C. Peth, and K. Mann, “Spatial characterization of extreme ultraviolet plasmas generated by laser excitation of xenon gas targets,” Rev. Sci. Instrum. 74(2), 969 (2003). [CrossRef]
10. F. Barkusky, C. Peth, K. Mann, T. Feigl, and N. Kaiser, “Formation and direct writing of color centers in LiF using a laser-induced extreme ultraviolet plasma in combination with a Schwarzschild objective,” Rev. Sci. Instrum. 76, 105102 (2005). [CrossRef]
11. Center for X-Ray Optics, Berkeley Lab, Homepage, “http://www-cxro.lbl.gov/
12. A. Bayer, F. Barkusky, J.-O. Dette, S. Döring, B. Flöter, C. Peth, and K. Mann, “Material analysis with EUV/XUV radiation using a broadband laser plasma source and optics system,” Proc. SPIE 7360, 736004 (2009). [CrossRef]
13. International Organization for Standardization, “Lasers and laser-related equipment – Determination of laser-induced damage threshold of optical surfaces – Part 1: 1 on 1 test“, ISO 11254-1.
14. R. C. Spitzer, R. L. Kauffman, T. Orzechowski, D. W. Phillion, and C. Cerjan, “Soft x-ray production from laser produced plasmas for lithography applications,” J. Vac. Sci. Technol. B 11(6), 2986–2989 (1993). [CrossRef]
15. E. Matthias, M. Reichling, J. Siegel, O. W. Käding, S. Petzoldt, H. Skurk, P. Bizenberger, and E. Neske, “The Influence of Thermal Diffusion on Laser Ablation of Metal Films,” Appl. Phys., A Mater. Sci. Process. 58, 129–136 (1994). [CrossRef]
16. E. Louis, A. R. Khorsand, R. Sobierajski, E. D. van Hattum, M. Jurek, D. Klinger, J. B. Pelka, L. Juha, J. Chalupsky, J. Cihelka, V. Hajkova, U. Jastrow, S. Toleikis, H. Wabnitz, K. I. Tiedtke, J. Gaudin, E. M. Gullikson, and F. Bijkerk, “Damage studies of multilayer optics for XUV free electron lasers,” Proc. SPIE 7361, 73610 (2009). [CrossRef]
17. B. Schäfer, J. Gloger, U. Leinhos, and K. Mann, “Photo-thermal measurement of absorptance losses, temperature induced wavefront deformation and compaction in DUV-optics,” Opt. Express 17(25), 23025–23036 (2009). [CrossRef]
18. P. R. Herman, R. S. Marjoribanks, A. Oettl, K. Chen, I. Konovalov, and S. Ness, “Laser shaping of photonic materials: deep-ultraviolet and ultrafast lasers,” Appl. Surf. Sci. 154–155(1-4), 577–586 (2000). [CrossRef]
19. J. Ihlemann, M. Schulz-Ruhtenberg, and T. Fricke-Begemann, “Micro patterning of fused silica by ArF- and F2-laser ablation,” J. Phys: Conference Series 59, 206–209 (2007). [CrossRef]
20. K. Sugioka, S. Wada, Y. Ohnuma, A. Nakamura, H. Tashiro, and K. Toyoda, “Multiwavelength irradiation effect in fused quartz ablation using vacuum-ultraviolet Raman laser,” Appl. Surf. Sci. 96–98, 347–351 (1996). [CrossRef]
21. D. Attwood, Soft X-Rays and Extreme Ultraviolet Radiation: Principles and Applications (Cambridge University Press, ISBN 0–521–02997-X).
22. T. Makimura, Y. Kenmotsu, H. Miyamoto, H. Niino, and K. Murakami, “Ablation of silica glass using pulsed laser plasma soft X-rays,” Surf. Sci. 593(1-3), 248–251 (2005). [CrossRef]
23. J. Albers, “Grundlagen integrierter Schaltungen”, Hanser Fachbuchverlag, ISBN: 3-446-40686-7 (2007).