0Next generation nanolithography based on Ru/Be and Rh/Sr multilayer optics
N. I. Chkhalo and N. N. Salashchenko
Citation: AIP Advances 3, 082130 (2013); doi: 10.1063/1.4820354 View online: http://dx.doi.org/10.1063/1.4820354
View Table of Contents: http://aipadvances.aip.org/resource/1/AAIDBI/v3/i8 Published by the AIP Publishing LLC.
Additional information on AIP Advances
Journal Homepage: http://aipadvances.aip.org
Journal Information: http://aipadvances.aip.org/about/journal
Top downloads: http://aipadvances.aip.org/features/most_downloaded
Information for Authors: http://aipadvances.aip.org/authors
ADVERTISEMENT
Gooifelloui
metals • ceramics • polymers • composites
70,000 products 450 different materials
small quantities fast
AIP ADVANCES 3, 082130 (2013)
(■) CrossMark
VHi «-dick for updates
Next generation nanolithography based on Ru/Be and Rh/Sr multilayer optics
N. I. Chkhalo and N. N. Salashchenko
Institute for Physics of Microstructures ofRAS, GSP-105, 603950, Nizhny Novgorod, Russia (Received 4 March 2013; accepted 12 August 2013; published online 29 August 2013)
A prospective move to 10.5 and 11.2 nm wavelengths, as an alternative to 6.7 and 13.5 nm, for next generation nanolithography is discussed. Ten-mirror optical systems based on Ru/Be, Mo/Be, Rh/Sr, Mo/Si, and La/B multilayers were compared for efficiency at their working wavelengths. It is shown that a transition to 10.5 nm and 11.2 nm may be a solution to the problem of increasing performance and resolution of a projection system. © 2013 Author(s). All article content, except where otherwise noted, is licensed under a Creative Commons Attribution 3.0 Unported License. [http://dx.doi.org/10.1063/L4820354]
I. INTRODUCTION
At present, various research groups and large chip-maker corporations are focusing on the development of extreme ultraviolet (EUV) lithography equipment for operation at X = 13.5 nm. In fact, half-pitch (hp) spatial resolution of 16 nm has already been achieved at this wavelength, using ASML's NXE:3100 platform.1 However, progress towards smaller critical dimensions (CDs), sub-10 nm, is impossible without increasing the numerical aperture of the projection lens to 0.7. This approach is problematic for several reasons. Firstly, the depth of focus (DoF) of the projection lens needs to be less than 30 nm. Secondly, the number of mirrors in the projection lens is increased by two, resulting in a performance drop of more than two times. Thirdly, the production of a high-aperture objective is a major challenge at this time.
An alternative approach is to move to shorter wavelengths. Performance calculations for a nanolithography system based on La/B multilayer optics at a wavelength of 6.7 nm (beyond extreme ultraviolet (BEUV)) have revealed the advantage of these optics over conventional Mo/Si optics for 13.5 nm, in terms of both efficiency and space resolution (see Table I and Ref. 2). Consequently, the volume of research into basic elements for BEUV lithography has recently increased significantly.3
This article provides a brief overview of the state of research in the field of optics based on La/B multilayer structures for BEUV lithography at a wavelength of 6.7 nm. Based on existing experimental data, it is concluded that the commercial prospects of this wavelength are largely dependent on the progress in La/B multilayer optics. In view of the existing risks in achieving high reflectance mirrors (greater than 70%), as well as more than two times lower sensitivity of the organic photoresist (because of the greater transmission at this wavelength compared with 13.5 nm), we believe that, in parallel with research in the field at 6.7 nm, it is time to start looking at other spectral ranges. The main problem of EUV lithography at 13.5 nm is its low fabrication rate; four wafers with 300 mm diameter can be fabricated per hour using the NXE:3100 machine1,4 and 43 per hour, as declared by ASML, using the NXE:3300B, which is scheduled for delivery in the second half of 2013. Taking this into account when choosing a new wavelength, and considering that a rate of more than 100 wafers per hour is desired, it is necessary to consider increasing both the spatial resolution and the performance of the system.
In this paper, we propose an evolutionary rather than revolutionary (as is the 6.7 nm case) approach to changing the operation wavelength of a BEUV lithographer, analogous with the case of traditional ultraviolet lithography. We consider that wavelengths of 10.5, 10.8, and ~11 nm show most promise, and, except for 10.5 nm, have been discussed earlier.5,6 In our opinion, these
|@ © I
2158-3226/2013/3(8)/082130/9 3, 082130-1 © Author(s) 2013 Ic^n^
TABLE I. Comparison of the parameters of the optical systems and the most promising radiation sources for BEUV lithography.
X, nm Ion material MLS d, nm D 10 AX, nm AX/X,% CE, % (in band AX) E = Rm10CE
13.5 Sn Mo/Si 6.9 0.044 0.27 2 4.5 0.20
13.5 Xe Mo/Si 6.9 0.044 0.27 2 0.8 0.033
11.18 Xe Ru/Be 5.65 0.086 0.19 1.7 3.6 0.31
11.18 Xe Mo/Be 5.61 0.070 0.14 1.3 2.7* 0.19
10.8 Xe Rh/Sr 5.52 0.050 0.19 1.8 3.6 0.18
10.5 Cs Rh/Sr 5.38 0.055 0.14 1.3 ? ?
6.62 Tb La/B 3.32 0.15 0.045 0.6 1.7 0.255
Notation in the table: k: wavelength; MLS: multilayer structure; d: period of the MLS; Rm10: the calculated peak transmittance of the optical system of ten multilayer mirrors; Ak: spectral band width of the ten-mirror optical system; Ak/k: band width expressed in percentage; CE: conversion efficiency of the radiation source in the spectral region of the optical system band width, *: CE is corrected according to the lower Ak of the Mo/Be optics, compared with Ru/Be.
wavelengths are worthy of further attention, taking into account the state-of-the-art research in this area.
II. SELECTION OF THE WORKING WAVELENGTH
When choosing a working wavelength for (B)EUV lithography, both ultimate spatial resolution and performance of a lithographic unit, which depends on the power and spectral characteristics of the radiation source and reflectivity of the multilayer mirrors, are taken into account. The radiation power PW incident on the wafer is proportional to the (B)EUV source power PS, the integral of the product of the spectral power density of the source S(k), and the spectral dependence of the reflectance of the multi-mirror (for example, ten-mirror) system, which is the product of the reflection curves R(k) of each mirror.
Pw ^ Ps • j S(k) • R10(k)dk. (1)
In practice, for a correct comparison of the effectiveness E of various sources and optical systems in general, and taking into account the narrow bandwidth of the optical systems, instead of expression (1 ), the following ratio is used:
E = Rim • S(kw) • Ak = Rim • CE, (2)
where CE is the conversion efficiency, which is the fraction of the total energy supplied to the source that is irradiated by the source in the half-space in the spectral bandwidth of the optical system, for instance, the laser energy incident on the target; and kW is the working wavelength. Thus, CE depends on the properties of the source and the band-pass of the optical system.
Since the refractive index of all substances in this spectral region is close to unity and strong absorption takes place, the maximum reflectance of multilayer interference mirrors is achieved only in narrow spectral regions adjacent to the region of anomalous dispersion of the weakly-absorbing material (spacer), where the minimal absorption and maximum contrast of the refractive index at the boundaries exist. The choice of such materials in the range 4-14 nm is limited to Si (L-absorption edge 12.4 nm), Be (K-absorption edge of 11.1 nm), B (K-absorption edge of 6.6 nm), and C (Kabsorption edge 4.4 nm). Independently, materials such as Y (M-absorption edge below 9 nm) and Sr (M-absorption edge below 10 nm) can be used, with which it is possible to achieve reflectivity of more than 60 and 70%, respectively.
At the beginning of research into EUV lithography, the main choice was between wavelengths of 11.3 and 13.5 nm, Mo/Be and Mo/Si optics, respectively.7 The radiation sources discussed are Xe and Li multi-charge ions. Despite the fact that the Mo/Be mirrors had higher reflectance due to the narrower spectral bandwidth of the ten-mirror system compared with 13.5 nm, the choice was made in favor of 13.5 nm, with Mo/Si optics. As a consequence, due to the high conversion efficiency of the energy supplied to the source into "useful" light, CE & 4.5%, tin ions were used as the main source of EUV radiation for 13.5 nm lithography.
III. ISSUES OF MULTILAYER La/B OPTICS FOR BEUV AT 6.7 nm
Recent research in the field of radiation sources at 6.7 nm allowed us to identify the most promising materials (Gd and Tb), whose highly charged ions generate radiation in the vicinity of 6.7 nm, with a CE = 1.65% of CO2 laser radiation into BEUV, and a spectral band width of 0.6% in the half-space.8-10 This value is nearly three times less than the conversion efficiency of the tin source at 13.5 nm, however, the overall theoretical efficiency of the scanner at 6.7 nm exceeds the efficiency at 13.5 nm, see Table I. Furthermore, by optimizing the material composition of the targets and conditions for the creation of hot plasma, the researchers hope to increase the CE.11
The most significant problem is achieving the required values of reflection coefficients for the multilayer La/B mirrors. Systematic studies of La/B, La/B4C, La/B9C, and La2O3/B4C normal-incidence multilayer mirrors showed that La/B4C multilayer structures (MLSs) had the best reflectivity.12-15 To date, the maximum measured reflection coefficients were 44% at X = 6.7 nm,12 and48.9% at X = 6.68 nm.15 We see strong discrepancy between the theoretical calculation, a value of more than 80%, and the experimental results. The main reason for this discrepancy is the broadening of the transition boundaries in the multilayer structures, due to mixing of the materials.13,14 In Refs. 13 and 16 it is shown also that in La/B4C multilayers, lanthanum forms compounds with boron and carbon (LaB6 and LaC2). Therefore, the profile of the electron density within a period of the MLS is not symmetrical. The full mixing depth is up to 1.5 nm when La is deposited onto B4C (the equivalent roughness parameter in the attenuator Debye factor for the amplitude reflection coefficient is a 1 = 0.75 nm) and 0.5 nm (a2 = 0.25 nm) when B4C is deposited onto La.
Recently, a record peak reflectance of around 60% was achieved on a La/B4C/C MLS with carbon antidiffusion barrier layers.17 Application of the barriers succeeded in reducing the depth of mixing of the film materials. From analysis of the angular and spectral dependence of the reflection coefficients, the following values were obtained: a 1 = 0.6 nm and a2 = 0.25 nm. Fig. 1 shows the reflectance spectra of the ten-mirror optical systems, R10(X), calculated for the following cases: ideal (a 1,2 = 0) La/B MLS (upper curve), ideal La/B4C MLS (middle curve), and La/B4C MLS with experimentally obtained record values of a 1 = 0.6 nm and a2 = 0.25 nm. The number of periods of the calculated MLSs was N = 200. In the calculations, we used the optical constants from.18 As can be seen from the figure, the interfacial roughness (mixing of the materials) leads to a sharp fall of both the peak reflectance and the spectral bandwidth AX. In particular, the integral reflection coefficient R10(X)»AX of the real structure, currently with record-breaking performance, is inferior to the theoretical limit by 16 times.
In addition, by considering the substantially lower sensitivity of existing photoresists and CE, we may assume that the recent overall productivity of lithography at a wavelength of 6.7 nm is around two orders of magnitude smaller than that at 13.5 nm. Secondly, a two-fold reduction in wavelength implies that the roughness and form accuracy requirements of the optical elements have to be increased by the same factor. To achieve a diffraction-limited image, the projection lens aberrations should be less than 0.3 nm. This calls for mirror form precision and integral surface roughness of better than 0.1-0.2 nm for roughness with lateral dimensions from 1 nm to 1 mm. These characteristics are at least twice the tolerance the optical elements industry currently has to offer, and achieving such quality will require significant new developments.
Summarizing the above, it should be noted that all of these shortcomings pose serious problems, and solutions are not currently forthcoming. Therefore, in our view, the time has come to consider an alternative wavelength. As was stated earlier, one has to seek not only an increase in spatial resolution, but also higher efficiency of the optical system and radiation sources.
FIG. 1. Spectral dependence of the reflection coefficients of ten-mirror optical systems, R (X), calculated for the following cases: ideal (a 1,2 = 0) La/B MLS (upper curve), ideal La/B4C MLS (middle curve), and La/B4C MLS with experimentally obtained record values of a 1 = 0.6 nm and a2 = 0.25 nm. The number of periods in the calculated MLSs was N = 200.
Wavelength,
FIG. 2. The spectral dependence of the reflection coefficients of Mo/Be and Ru/Be MLSs.
IV. MULTILAYER Be- AND Sr-BASED OPTICS FOR BEUV AT 11 nm
In early works, Mo/Be MLSs have been studied for potential use as mirrors for the spectral region around 11 nm. The first results showed a high reflection coefficient, Rexp = 70.1%; the theoretical limit is Rth = 75.6%.5 Peak reflectivity, even for experimental samples, exceeds the reflectance of the Mo/Si mirrors. However, their spectral bandwidth is greatly inferior. To increase the efficiency of Be-based MLSs, we propose an alternative; Ru/Be MLS. For comparison, Fig. 2 shows the calculated reflectance spectra of the Mo/Be and Ru/Be MLSs, with central wavelength of
FIG. 3. Spectral dependence of the calculated integral R10(k)»Ak and peak R10(k) reflectance of the ten-mirror system based on Ru/Be optics.
the reflection peak in the vicinity of 11.3 nm. As seen from the figure, the peak reflectance (77.8% against 75.9%) and the spectral bandwidth at half maximum (0.407 nm against 0.331 nm) of Ru/Be are markedly higher than those for Mo/Be optics. The total integral gain in the case of the ten-mirror system, is 1.7 times, see Table I.
Fig. 3 shows the spectral dependence of the calculated integral R10(X)»AX and the peak reflectance R10(X) of the ten-mirror system based on Ru/Be optics. These dependencies can be used to choose the optimal wavelengths of the BEUV radiation source. As follows from the figure, the integral reflection coefficient reaches a maximum at a wavelength of around 11.2 nm, and further varies slightly with increasing wavelength. It should be noted that in the case of narrow spectral line width, narrower than the bandwidth of the multi-mirror system, a high peak reflectance of the Ru/Be mirror is preserved down to 11.12 nm.
In the spectral range below 11 nm, the highest reflectivity is offered by MLS, using a weakly absorbing material Sr. Experimental data on the reflective characteristics of Mo/Sr mirrors, which showed a fairly high degree of perfection, has been reported.6 In particular, at a wavelength of 10.5 nm at normal incidence, a reflection coefficient of 48.3% was obtained, with a theoretical limit of 65%.
Calculation of the reflection coefficients of Rh/Sr MLSs showed that they have significantly higher peak and integral reflection coefficients than Mo/Sr MLSs. In particular, in the range 9.711.3 nm, reflectance is greater than 70%. As seen in Table I, the integral reflection (a product of reflectivity and spectral band-pass) of the ten-mirror system at 10.8 nm is almost the same as at 13.5 nm with Mo/Si optics.
V. EFFICIENCY COMPARISON OF BEUV AT 11 nm AND EUV AT 13.5 nm
EUV lithography uses Mo/Si multilayer optics with theoretical ten-mirror system reflectance of around 4.4%. An optimal radiation source for this spectral range is tin plasma, whose conversion efficiency reaches a value of 4.5%. The overall efficiency corresponding to these optics and the CE of the tin source, as defined in (2), is 0.198, see Table I.
Fig. 4 shows the emission spectrum of xenon plasma at a temperature optimized for generation at 13.5 nm. The right-hand "bell" corresponds to the bandwidth of a ten-mirror system with Mo/Si MLSs. On the left is a similar curve for Rh/Sr MLSs. Circles mark the bandwidth of the same
FIG. 4. The emission spectrum of Xe plasma (line). The plasma temperature is optimized for maximum wavelength in the vicinity of 13.5 nm. Symbols denote the spectral band-pass of the ten-mirror systems: left: Rh/Sr; central: Ru/Be; and right: Mo/Si MLSs.
Wavelength,
FIG. 5. The same as in Fig. 4, but the plasma temperature is optimized to generate emission in the region of 11 nm.
optical system with Ru/Be multilayer mirrors. The spectral data was provided by K. Koshelev and M. Krivtsun (Institute for Spectroscopy of Russian Academy of Sciences, Troitsk), and are similar to those given previously.19 For this analysis, pairs of materials for MLSs were chosen so they had the highest theoretical reflectivity for the largest number of prospective wavelengths of plasma sources.
The conversion efficiency of the discharge energy into EUV radiation is about 0.8% in a 2% spectral band in the vicinity of 13.5 nm. As can be observed, in both cases the effectiveness of the source and the integral transmission of the optical system at ~11 nm are higher than at 13.5 nm. At the same time, the wavelengths are shorter, and thus the resolution is better, by about 20%. In practice, 13.5 nm lithography uses higher efficiency sources of tin ions (see Table I), with conversion efficiency up to 4.5%.20,21 However, by optimizing the temperature of the xenon plasma, see Fig. 5, the efficiency in the short wavelength range can be significantly increased.19,22
In contrast with 13.5 nm, there is no precise data on the conversion efficiency of Xe sources at 11 nm in the literature, however, some estimates are given. In particular, in Ref. 21 is stated the
conversion efficiencies of the Sn source at 13.5 nm and the Xe source at 11 nm are approximately the same. The measured intensity of the radiation in the region of 11 nm is 4-5 times higher than at 13.5 nm.19-22-25
Calculated spectra of Xe plasmas with different temperatures and electron densities have been reported in.26 They show that the position of the spectral maxima and the ratio of the radiation intensities at 11 nm and 13.5 nm vary. In addition, there are spectra with an intensity ratio of around four. An interesting result is the calculation of individual contributions to the spectra of ions with different charge states. In particular, it is shown that the largest contribution to the emission of 13.5 nm radiation is from Xe+10 ions. The spectrum has a maximum at 11 nm. The intensity ratio in this case is around six. The emission spectrum of the Xe+9 ions has a strong peak with a central wavelength of 11.3 nm, which, significantly, is around six times higher than the peak at 13.5 nm. This result allows: first, hope that the estimated conversion efficiency of the Xe source in the region of 11 nm is conservative and may even be increased by optimizing the parameters of the plasma (CEn = 0.8 x 4.5 = 3.6%, where 0.8 is the CE of the Xe source optimized for generation in the field of 13.5 nm and confirmed in a number of papers, for example,19 and 4.5 is the ratio of the intensities in the regions of 11 and 13.5 nm observed in the above-mentioned papers); and second, through optimization of the number of Xe+9 and Xe+10 ions in the plasma, the operating wavelength of the source can be better aligned with the spectral band-pass of the Ru/Be multilayer mirrors. The latter is very important, see Fig. 3, since beryllium-based mirrors have a short-wave boundary of the reflection coefficients, which is about XBe = 11.1 nm.
The estimated performance of the lithographic facility, including reflectance of Ru/Be mirrors and CE = 3.6%, is 0.310, which exceeds by half the performance at 13.5 nm (see Table I).
The wavelength of 10.8 nm coincides with the peak of the unresolved transition array (UTA) spectrum of xenon, shown in Fig. 5. As was described earlier, at this wavelength the highest reflectivity is offered by Rh/Sr MLSs. As seen from Table I, if CE = 3.6% we see that the efficiency of the machine at this wavelength is about 10% lower than the machine at 13.5 nm.
An interesting wavelength for BEUV lithography is 10.5 nm. At this wavelength, the reflection coefficients of Rh/Sr multilayer optics are comparable with the reflection of Mo/Si MLSs at 13.5 nm, but this wavelength coincides with the center of the UTA emission of multiple charged ions of cesium. Given the form of the cesium spectrum in the vicinity of 10 nm,27 one may expect higher source efficiency. So, the application prospects for this wavelength for BEUV lithography depend on the conversion efficiency of the source.
VI. CONCLUSION
Presented data indicates that a critical issue for the future of BEUV lithography at 6.7 nm is to achieve high reflectivity multilayer mirrors. Currently, the achieved integral reflectivity of a ten-mirror optical systems is inferior to the theoretical limit, by 16 times, and the prospects for an increase in reflectivity of more than an order of magnitude remain uncertain.
For minimizing the risks we believe that one has to think about alternatives to BEUV lithography at 6.7 nm. The most interesting wavelengths are 10.5 and 10.8 nm using Rh/Sr multilayer optics, and ~11 nm using Ru/Be MLSs. It should be noted that, initially, the situation with multilayer mirrors at these wavelengths was much better than for X & 6.7 nm. In particular, Mo/Be MLSs were manufactured and investigated as early as 1998. At 11.3 nm at normal incidence, a reflection coefficient of around 70.1% was measured,5 with the theoretical maximum being 75.6%. For a Mo/Sr mirror, a reflection coefficient of 48.3% was obtained at 10.5 nm,6 which, although lower than the theoretical limit of 65%, looks promising.
Prospects of BEUV lithography on Sr-based optical solutions depend on two main issues. First, the problem of degradation of the mirror reflectivity due to oxidation has been observed on Mo/Sr MLSs.6 The second is that the most promising wavelength for this optics is 10.5 nm. This wavelength coincides with the center of the UTA emission of multiple charged ions of cesium. Given the form of the cesium spectrum in the 10 nm range,27 one may expect high source efficiency. So, application prospects for this wavelength for BEUV lithography depend on the conversion efficiency of the Cs source, which should be measured.
The most realistic wavelength for next generation nanolithography in the BEUV range appears to be ~11 nm, at which the highest reflectivity (given the ~78% theoretical limit) is offered by the Ru/Be MLS. The advantage of this wavelength in terms of improving the projection scheme resolution is certain. Based on the experimental Mo/Be MLS reflection data, we may, with high probability, expect the optical system transmission to be higher than that at 13.5 nm. Currently, there is no precise data concerning the conversion efficiency of Xe sources optimized for X « 11 nm. The experimental data in the literature allow the authors to hope that they are close to the optimum CE of the Sn source at 13.5 nm. The operating wavelength, between 11 and 12 nm, is entirely dependent on optimization of the ion composition of the Xe source. For example, Xe+10 ions emit radiation around 11 nm, while Xe+9 emit in the vicinity of 11.3 nm.
When choosing between ~11 nm and 13.5 nm, it is necessary to take into account the fact that the Xe source produces substantially less debris than Sn, which may also have a positive impact on its effectiveness. Since the proposed wavelengths are close to the L-absorption edge of silicon, introduction of even a small amount of silicon atoms in a polymer molecule, which is quite natural, leads to a sharp absorption of radiation at the working wavelength. Thus, in spite of the shorter wavelength, photoresist sensitivity may even increase compared with that at 13.5 nm, whereas the sensitivity to mirror contamination by hydrocarbons decreases in this case by almost 1.5 times. Because of the lower absorption in this range, one can expect higher transmission of the spectral purity filters and the protecting mask pellicles, resulting in additional lithography performance at ~11 nm, compared with 13.5 nm.
In earlier analysis, the safety considerations required when working with beryllium were stressed. But, in our opinion, if for technical or any other reasons one needs to use dangerous materials, for instance nuclear or poisonous materials, they can be used by applying some additional safety measures. In practice, beryllium is widely used, for instance, as windows in conventional X-ray tubes for technical, medical, and scientific applications.
VII. SUGGESTIONS FOR FURTHER WORK
The first step toward implementation BEUV lithography in the vicinity of 11 nm is research into the field of highly reflective Ru/Be MLSs and their optimization, and measurement of the conversion efficiency of Xe sources in the region of 11 nm. From an optical point of view, the optimal wavelength chosen should be above 11.2 nm.
Regarding BEUV lithography bellow 11 nm the first steps are measurement of the conversion efficiencies of Xe at 10.8 nm and Cs at 10.5 nm sources, and researches into the field of Rh/Sr MLSs.
These investigations require no significant capital investment, and will give a definite answer to the question of the advisability of deploying a full-scale study into the field of BEUV lithography around 11 nm.
ACKNOWLEDGMENTS
This work was supported in part by ASML and Phystex, the Netherlands, by the RFBR grants 1102-00597, 11-02-00589, 11-02-00961, 11-02-97109, and 13-02-00377, and by the Federal Program "Scientists and science educators of innovative Russia in 2009-2013."
1 J. V. Hermans, D. Laidler, P. Foubert, K. D'have, S. Cheng, M. Dusa, and E. Hendrickx, Proc. SPIE 8322, 832202 (2012).
2 D. Glushkov, V. Y. Banine, L. A. Sjmaenok, N. N. Salashchenko, and N. I. Chkhalo, Multilayer mirror and lithographic apparatus, WO 10091907 A1. 19.08.2010.
3 Ch. Wagner and N. Harned, Nature Photon 4, 24 (2010).
4O. Wood, J. Arnold, T. Brunner, M. Burkhardt, J. H.-C. Chen, D. Civay, S. S.-C. Fan, E. Gallagher, S. Halle, M. He etal.,
Proc. SPIE 8322, 832203 (2012). 5C. Montcalm, S. Bajt, P. Mirkarimi, E. Spiller, F. Weber, and J. Folta, SPIE 3331, 42-51 (1998). 6B. Sae-Lao and C. Montcalm, Optics Letters 26(7), 468-470 (2001). 7 V. Banine, J. Benschop, M. Leenders, and R. Moors, Proc. SPIE 3997, 126 (2000). 8S. S. Churilov, R. R. Kildiyarova, A. N. Ryabtsev, and S. V. Sadovsky, Phys. Scr. 80, 045303 (2009). 9 T. Otsuka, D. Kilbane, J. White, T. Higashiguchi, N. Yugami, T. Yatagai, W. Jiang, A. Endo, P. Dunne, and G. O'Sullivan, Appl. Phys. Lett 97, 111503 (2010).
10D. B. Abramenko, V. S. Bushuev, R. R. Gayazov, V. M. Krivtcun, M. V. Spiridonov, A. M. Chekmarev, O. F. Yakushev, and K. N. Koshelev, Proc. 17 Int. Symp. "Nanophysics & Nanoelectronics," Nizhny Novgorod, 11-15 March 2013, Russia, V. 1,p. 261.
11 T. Otsuka, B. Li, C. O'Gorman, T. Cummins, D. Kilbane, T. Higashiguchi, N. Yugami, W. Jiang, A. Endo, P. Dunne, and G. O'Sullivan, Proc. SPIE 8322, 832214 (2012).
12 S. S. Andreev, M. M. Barysheva, N. I. Chkhalo, S. A. Gusev, A. E. Pestov, V. N. Polkovnikov, N. N. Salashchenko, L. A. Shmaenok, Y. A. Vainer, and S. Yu. Zuev, NIM A 603(1-2), 80 (2009).
13 T. Tsarfati, R. W.E. van de Kruijs, E. Zoethout, E. Louis, and F. Bijkerk, Thin Solid Films 518, 1365 (2009).
14 S. S. Andreev, M. M. Barysheva, N. I. Chkhalo, S. A. Gusev, A. E. Pestov, V. N. Polkovnikov, D. N. Rogachev, N. N. Salashchenko, Yu. A. Vainer, and S. Yu. Zuev, Technical Physics 55(8), 1168 (2010).
15 Y. Platonov, J. Rodriguez, M. Kriese, E. Gullikson, Te. Harada, T. Watanabe, and H. Kinoshita, Proc. of SPIE 8076, 80760N-2 (2011).
16 T. Tsarfati, E. Zoethout, E. Louis, R. W. E. van de Kruijs, A. Yakshin, S. Müllender, and F. Bijkerk, Proc. of SPIE, 7271 (2009).
17 N. I. Chkhalo, S. Künstner, V. N. Polkovnikov, N. N. Salashchenko, F. Schäfers, and S. D. Starikov, Appl. Phys. Lett 102, 011602(2013).
18 B. L. Henke, E. M. Gullikson, and J. C. Davis, Atomic Data and Nuclear Data Tables 54(2), 181-342 (1993).
19 U. Stamm, I. Ahmad, I. Balogh, H. Birner, D. Bolshukhin, J. Brudermann, S. Enke, F. Flohrer, K. Gabel, S. Götze, G. Hergenhan, J. Kleinschmidt, D. Klopfel, V. Korobotchko, J. Ringling, G. Schriever, C. D. Tran, and C. Ziener, Proc. SPIE 5037, 119-129 (2003).
20T. Otsuka, D. Kilbane, J. White, T. Higashiguchi, N. Yugami, T. Yatagai, W. Jiang, A. Endo, P. Dunne, and G. O'Sullivan,
Appl. Phys. 97, 111503 (2010). 21V. Y. Banine, K. N. Koshelev, and G. H. P. M. Swinkels, J. Phys. D: Appl. Phys. 44, 25300 (2011).
22 R. Rakowwski, A. Bartnik, H. Fiedorowicz, F. de Gaufridy de Dortan, R. Jarocki, J. Kostecki, J. Mikolajczyk, L. Rye, M. Szczurek, and P. Wachulak, Appl. Phys. B 101, 773-789 (2010).
23 B. A. M. Hansson, L. Rymmell, M. Berglund, and H. M. Hertz, Microelectronic Engineering 53, 667 (2000).
24 H. Fiedorowicz, A. Bartnik, R. Jarocki, J. Kostecki, J. Mikolajczyk, R. Rakowwski, M. Szczurek, and P. Wachulak, Proc. SPIE 5037,389(2003).
25 S. Miyamoto, A. Shimoura, S. Amano, K. Fukugaki, H. Kinugasa, T. Inoue, and T. Mochuzuki, Appl. Phys. Lett. 86, 261502 (2005).
26 E. R. Kieft, K. Garloff, A. M. van der Mullen, and V. Banine, Phys. Rev. E 72, 036402 (2005).
27 B. Li, A. Endo, C. O'Gorman, T. Cummins, T. Donnelly, D. Kilbane, W. Jiang, T. Higashiguchi, N. Yugami, P. Dunne, and G. O'Sullivan, Proc. SPIE 8139, 81390P (2011).
