HOME | SEARCH | PACS & MSC | JOURNALS | ABOUT | CONTACT US
Large area interference lithography using a table-top extreme ultraviolet laser: a systematic study of the degree of mutual coherence
This article has been downloaded from IOPscience. Please scroll down to see the full text article. 2009 Nanotechnology 20 115303 (http://iopscience.iop.org/0957-4484/20/11/115303) The Table of Contents and more related content is available
Download details: IP Address: 140.105.5.88 The article was downloaded on 27/08/2009 at 10:11
Please note that terms and conditions apply.
IOP PUBLISHING
NANOTECHNOLOGY
Nanotechnology 20 (2009) 115303 (4pp)
doi:10.1088/0957-4484/20/11/115303
Large area interference lithography using a table-top extreme ultraviolet laser: a systematic study of the degree of mutual coherence P Zuppella1 , D Luciani1 , P Tucceri1, P De Marco1 , A Gaudieri1 , J Kaiser2 , L Ottaviano1 , S Santucci1 and A Reale1 1
Dipartimento di Fisica, Universit`a dell’Aquila, gc-LNGS INFN, Via Vetoio, 67100 L’Aquila, Italy 2 Institute of Physical Engineering, Brno University of Technology, Technicka 2, 61669 Brno, Czech Republic E-mail:
[email protected]
Received 15 December 2008, in final form 22 January 2009 Published 24 February 2009 Online at stacks.iop.org/Nano/20/115303 Abstract A prototype low cost table-top Ar capillary discharge laser source (1.5 ns pulse duration, λ = 46.9 nm) was successfully used to produce, by means of interference lithography (with a simple Lloyd mirror setup), large area (0.1 mm2 ) regular patterns from 400 nm down to 22.5 nm (half-pitch) on PMMA/Si (PMMA: polymethylmethacrylate) substrates. The experiments allowed a systematical investigation of the degree of mutual coherence of the source, giving a clear indication that the interference lithography can be pushed down to the ultimate resolution limit of λ/4.
sources has been recently addressed [6–8]. This technique is maskless, low cost (laboratory scale), and has the capacity to engrave (in few minutes) large areas of the photoresist ( mm2 ) with nanometer (100 nm) size ordered patterns. Capeluto et al [6] recently showed that this lithographic technique produces regular patterns of lines or dots on PMMA (typical period 55 nm) [6, 7]. Ritucci et al [8] showed a nice example of large area engineering at the nanoscale of the optical properties of LiF [9], and very recently we have also succeeded to complete a full lithographic process by metal deposition and lift off on a XIL patterned PMMA/Si(100) substrate [10]. The spatial and temporal coherence of the laser source are obviously parameters of primary importance for the lithography. In particular, when using the simplest optical setup for the interferometer (i.e. the Lloyd scheme, as in [8]) (see inset in figure 1) the beam is physically split into two halves, one directly beamed at the lithographic target, the second half is folded on the first by reflection of a mirror. Thus, the two parameters of spatial and temporal coherence come hardly into play for the quality of the IL. Good spatial coherence is required, as the interfering beams belong to two
1. Introduction The pivotal issue in nanoscale science and technology is nanofabrication: i.e. production of few tens nanometers size structures with engineered physical properties. To this aim, despite the conceptual beauty of promising bottom up approaches, mostly based on self-assembling and nanomanipulation processes, there is still enormous potential room of development in the use of lithographic techniques with their top-down approach. In this field, the microelectronics industry is today well into the nanoscale with 45 nm critical dimension. However, the tools of optical lithography, adopted by the integrated circuit industry, are very expensive and their use is justified only in high-product-value manufacturing context. Yet, when it comes to research issues and more small scale leading edge laboratory based nanoscience, alternative low cost, portable, and rather flexible nanolithographic strategies are most appropriate (i.e. nanoimprint lithography [1, 2], scanning probe lithography [3], electron or ion beam lithography [4, 5]). Beside those above mentioned techniques, the potential of x-ray interference lithography (XIL) combined with the use of small scale, low cost, x-ray coherent laboratory 0957-4484/09/115303+04$30.00
1
© 2009 IOP Publishing Ltd Printed in the UK
Nanotechnology 20 (2009) 115303
P Zuppella et al
a table-top soft x-ray laser based on a 45 cm long Ar filled capillary. The capillary is pumped by an electrical discharge to create a plasma. In the plasma, Ne-like ions provide a 3p– 3s lasing transition at a wavelength of 46.9 nm [11]. The capillary length and the cross sectional size of the plasma (∼300 μm) cause a large spatial coherence of the emitted radiation. The temporal coherence, notable as well, is mainly determined by the Doppler effect produced by the fast motion of ions in the plasma (λ/λ ∼ 10−4 ). The laser, operating at 0.1 Hz, produces 1.5 ns long pulses each delivering an energy of approximately 150 μJ. The beam has an annular structure with a 6 mrad divergence. Interference fringes have been generated by using a Lloyd mirror (0.5 nm RMS Si reflecting surface) interferometer (lower right inset in figure 1(a)). In this scheme one half of the laser beam grazes the mirror at an angle α and then is reflected and overlaps on the sample surface with the undeflected half of the beam itself. The outcome is the formation of a sinusoidal interference pattern with a period p = λ/2 sin α . Thus, the obvious physical resolution limit of the XIL is pmin = λ/2. As substrates for lithography we used 20–30 nm PMMA (Sigma Aldrich 120 000 molecular weight) photoresist layers on Si(100) wafers. Substrates were placed orthogonal to the mirror at the distance of 80 cm from the source. The optimized substrate exposure took 15–25 laser shots (corresponding to a dose of 6–10 mJ cm−2 ). After developing, the sample morphology has been investigated with tapping mode atomic force microscopy (AFM, Digital Instruments D5000 microscope-Nanoscope IV controller using silicon tips, frequency range 310–370 kHz). The AFM image of figure 1(a) shows a small scale area of the overall nanopatterned surface (that extends up to a notable value of 200 × 200 μm2 ). The 100 nm grating of figure 1(a) consists of well-defined PMMA lines (bright) separated by dark regions (12 nm deeper in average). This modulation has been engraved with 25 laser shots. The squared flat shape of the minima of the average of 256 horizontal height line profiles reported in the upper right inset of figure 1(a) clearly indicates that the dark regions correspond to stripes of exposed bare Si substrate where the PMMA layer has been completely removed. This is also directly evidenced by the three dimension AFM image of figure 1(b) where the Z height scale has been reversed. The patterned exposure of the bare Si substrate produces samples ready to a subsequent full lithographic process like metal deposition and lift off as reported by us in [10]. It is also worth noting that, the peak– valley height of the patterned PMMA is always a fraction of the pristine (non laser exposed) PMMA thickness as, after developing, a 10–15 nm resist layer is uniformly removed due to exposure to the incoherent part of the light. Figure 2 shows an AFM image of a PMMA/Si lithographed (XIL) sample with a periodic pattern of quite well-defined PMMA lines. The lines are 22.5 nm apart. This is evidenced in the upper right inset of the figure that shows the average of 256 horizontal height profiles taken on the AFM image. The Si bare substrate is also likely exposed in this case, though the average height profile of figure 1(b) does not show a clear squared flat shape of the minima. This is likely due to the AFM artifact of tip convolution which
Figure 1. AFM (plan view in (a) and three-dimensional plot in (b)) images of patterned PMMA/Si(100) with 50 nm half-pitch. Lower left inset in (a): scheme of the Lloyd interferometer ((S) sample, (M) mirror). Upper inset in (a): average of 256 horizontal height line profiles taken on the corresponding AFM images. The vertical Z scale in (b) is reversed.
different spatial regions of the light wavefront. Moreover, due to the different optical path, as the reflected wave overlaps to the unreflected one after a temporal delay, interference occurs only if such delay is smaller than the temporal coherence length of the source. The smaller is the IL period desired, the larger has to be the primary beam to mirror angle, the longer is the optical path of the reflected beam, the longer is its delay. Even if 100 nm patterns have been successfully realized in this scheme [8], a question can then arise, whether a Lloyd setup can be still used up to the intrinsic resolution limit λ/4 of the system. In this work we report a systematic investigation of the mutual coherence degree (of the interfering beams) as a function of the IL period, down to the smallest half-pitch resolution achieved so far with table-top laser sources. Results show that our prototype table-top source is characterized by a temporal and spatial coherence that allow to use the very simple Lloyd setup down to the intrinsic resolution limit of the technique.
2. Results and discussion The x-ray laser source, the interferometer setup, and the PMMA/Si(100) substrates preparation have been described in [10]. The main experimental specifications are summarized hereafter for the reader’s convenience. The source consists of 2
Nanotechnology 20 (2009) 115303
P Zuppella et al
photoresist layer to the radiation dose. In this assumption, if d is the thickness of the resist layer engraved by light exposure, then d = K I N , where I is the (average) intensity of the laser light at each shot, N is the number of PMMA exposure shots, and K is a parameter accounting for the PMMA sensitivity at the wavelength of radiation under consideration. In interference lithography, the intensity of the light at the sample surface is, indeed, modulated by the interferometer, thus there will be a maximum ( Imax ) and a minimum ( Imin ) intensity. Accordingly there will be (as observed) an height modulation d of the PMMA after light exposure and development. This value will be d = K (Imax − Imin )N . The values of Imax and Imin can be simply related to fundamental physical parameters according to the following formula:
Imax − Imin = 4 I γ R Figure 2. AFM scan of XIL patterned PMMA/Si(100) with 22.5 nm half-pitch. Upper inset: average of 256 horizontal height line profiles taken on the AFM image.
(1)
where I is the intensity of the undeflected beam, γ is the mutual√coherence √ degree of the light waves at the sample and R = R p + R s is a function of R p and R s , that are the reflectivities of the p and s beam polarization [8]. Accordingly,
becomes not negligible at length scales comparable to the tip apex size (10–20 nm). The height (peak–valley) modulation of the fringes is ∼3.5 nm. Such modulation has been obtained with 15 laser shots. Although a similar (19 nm half-pitch) ultimate resolution was already reported for XIL by using synchrotron radiation [12], nevertheless table-top XIL patterns with 22.5 nm half-pitch, to the best of our knowledge, have never been realized so far neither with a Lloyd nor with more sophisticated (Mach–Zehnder type) interferometer setups [13]. Let us discuss some important technicalities. As illustrated in detail in the following, the height modulation of the lithographed PMMA pattern is obviously dependent on the laser fluence on the sample. In our case the sample to source distance has been adjusted to 0.8 m. The choice of a Si(100) wafer as a mirror is an excellent compromise between cost and reflectivity (at least for lithography down to 40–50 nm [14]). Particular care was also paid in the optimization of the PMMA layer thickness. In view of technological applications (whether a subsequent metal deposition and lift off is foreseen) the lithographic process has to lead to the selective removal of the resist layer with exposure of the substrate. Thus the resist layer thickness has to be comparable to the penetration depth of the radiation used for lithography. For a 46.9 nm source such penetration depth is ≈20 nm in PMMA. This explain our use of ultrathin (20 nm thick) PMMA layers. All the above mentioned improvements lead to a noteworthy decrease of the number of laser shots required to significantly engrave the PMMA layer and, accordingly, to a reduced blurring of the XIL patterns due to mechanical vibrations of the XIL system.
d = 4 K I γ R N.
(2)
It is worthwhile to note that K , expressing the resist sensitivity, is depending only on the wavelength of the light, while γ and R depend on the period of the pattern to be engraved [14]. A convenient parameter to define is what will be called henceforth the lithographic efficiency of the laser source defined as E l = d/N . This parameter has been measured for XIL periods of 45 nm, 100 nm (data reported in figures 1 and 2) and also from AFM data of XIL patterned samples at 200, 400 and 800 nm (images not reported for brevity). According to the above formulae, the ratio of the lithographic efficiencies at two given XIL periods, p 1 and p 2 can be derived experimentally as: d( p1)N( p 2) E l ( p1)/E l ( p2) = (3) d( p2)N( p 1) but can also be written as
E l ( p 1)/E l ( p2) = (γ p1 /γ p2 )
R p1 R p2
(4)
where γ p1 and γ p2 are the mutual coherence degrees at p1 and R p2 period respectively, and the factor R pp12 can be estimated by known values of the reflectivities of the Si mirror [14]. Accordingly, one can derive as a function of the XIL period the ratio γ p1 /γ p2 . We have reported in figure 3 the values of γ p1 /γ800 and R p1 /R800 considering 800 nm as a reference period. The γ /γ800 curve in figure 3, nicely fitted by a parabolic function, shows that, while there is a significant loss (by a factor 5.3) of mutual coherence of the XIL setup when passing from 800 to 100 nm, by further reducing the XIL period to 45 nm the subsequent loss of coherence is only by a factor 1.3. No further significant losses of mutual coherence are expected extrapolating the curve to 23 nm corresponding to the pmin = λ/2 physical limit of the XIL system. Indeed, as indicated by the reflectivity graph in figure 3, below 200 nm,
2.1. Study of the mutual coherence degree More interestingly, our experimental results can also be discussed in terms of the coherence of the x-ray laser and of the physical parameters that determine the performance of the interferometer used. Given the low fluence of the primary x-ray beam (150 μJ), in this regime a very first reasonable assumption to make, is the linearity of the response of the 3
Nanotechnology 20 (2009) 115303
P Zuppella et al
maskless, and, last but not least, the simplicity of the optical interferometer scheme used, make it a very promising tool in many potential applications in nanotechnology. In particular, with this technique the low cost nanofabrication of devices becomes more viable.
Acknowledgments The authors acknowledge the EU support under the COST action MP0601 ‘Short wavelength laboratory sources’. Paola Tucceri and Domenico Luciani thank ‘Progetto INFN-LNGS P.O.R. Abruzzo 2000–2006’ and ‘Progetto regionale IN-COPOR C3/IC1E’. Figure 3. Open circles: plot of the measured coherence degree parameter γ of the XIL setup normalized to the γ800 value (measured at 800 nm). Full circles: corresponding plot of the overall normalized reflectivity R/R800 .
References [1] Chou S Y, Krauss P R and Renstrom P J 1995 Appl. Phys. Lett. 67 3114 [2] Chou S Y, Krauss P R and Renstrom P J 1996 Science 272 85 [3] Jaschke M and Butt H-J 1995 Langmuir 11 1065 [4] McCord M 2002 Introduction to Electron Beam Lithography (Proc. SPIE) vol VT11229 (Bellingham, WA: SPIE Optical Engineering Press) ISBN-13: 9780819433947 [5] Yao N 2007 Focused Ion Beam System: Basics and Applications (Cambridge: Cambridge University Press) pp 1–30 [6] Capeluto M G, Vaschenko G, Grisham M, Marconi M C, Luduea S, Pietrasanta L, Lu Y, Parkinson B, Menoni C S and Rocca J J 2006 IEEE Trans. Nanotechnol. 5 3 [7] Wachulak P W, Capeluto M G, Marconi M C, Patel D, Menoni C S and Rocca J J 2007 J. Vac. Sci. Technol. B 25 2094 [8] Ritucci A, Reale A, Zuppella P, Reale L, Tucceri P, Tomassetti G, Bettotti P and Pavesi L 2007 J. Appl. Phys. 102 034313 [9] Tomassetti G, Ritucci A, Reale A, Arrizza L, Flora F, Montereali R M, Faenov A and Pikuz T 2004 Appl. Phys. Lett. 85 4163 [10] Ottaviano L et al 2008 Plasma Sources Sci. Technol. 17 024019 [11] Ritucci A, Tomassetti G, Reale A, Flora F and Mezi L 2004 Phys. Rev. A 70 023818 [12] Solak H H, He D, Li W, Singh-Gasson S, Cerrina F, Sohn B H, Yang X M and Nealey P 1999 Appl. Phys. Lett. 75 2328 [13] Wachulak P, Grisham M, Heinbuch S, Martz D, Rockward W, Hill D, Rocca J J, Menoni C S, Anderson E and Marconi M 2008 J. Opt. Soc. Am. B 25 104 [14] http://www.cxro.lbl.gov
the drop of lithographic efficiency is, by far, mostly determined by the drop of reflectivity of the Si mirror. This bottleneck can be easily removed by the use, for example, of a (35 nm period, ratio of 0.52) multilayer Si–Sc mirror [14]. This mirror would exhibit a reflectivity exceeding by a factor of almost one order of magnitude the one of Si (at light incidence angles close to 45◦ ) [14]. Other main technical drawbacks can be easily overcome. The lithographic efficiency can be increased reducing the mechanical vibrations, by using other photoresists (like HSQ or an amplified photoresist) and minimizing the sample to source distance.
3. Conclusions With the use of a table-top low cost, portable, capillary discharge laser source (46.9 nm) we have successfully performed maskless XIL on PMMA targets engraving nanostructures down to 22.5 nm half-pitch, using a very simple Lloyd interferometer. We demonstrated that the source shows excellent properties of mutual coherence that will allow to scale the lithography down to the λ/4 physical limit (halfpitch). The features of this technique i.e. its low cost, the portability, the capability to engrave large areas, its being
4
