We describe the theory and report the first experimental demonstration of Cavity Resonance Lithography (CRL); a double pattering (DP) technique that can generate patterns on a photoresist 1) with twice the spatial frequency of that of the diffraction limited lithography mask, and 2) at an offset distance that is in the far field of the mask. CRL requires only a single exposure and development step and does not require any additional processes. With commercially available photoresists (PR) and developers, we have recorded a 32.5 nm half-pitch pattern (which is well below the diffraction limit) at an offset distance of 180 nm (which is well beyond the evanescent decay length scales) using 193 nm illumination. We also discuss strategies to improve the minimum feature size and potential implementation schemes.
©2011 Optical Society of America
As the half-pitch of critical features on semiconductor ICs decrease below the diffraction limit of 37 nm for off-axis 193 nm exposure, the semiconductor industry has turned to various double patterning (DP) methods to enhance resolution while maintaining current 193 nm lithography tooling [1,2]. For example, LELE (Litho-Etch-Litho-Etch)  and LFLE (Litho-Freeze-Litho-Etch)  methods implement two lithography and etch steps, providing a doubling in the achievable spatial frequency. Because the second lithography patterns are adjacent to the first set of features, these techniques must meet very tight overlay requirements [5,6]. In SADP (Self-Aligned Double Patterning) techniques, the first lithographically patterned resist acts as self-aligning spacers, which are subsequently removed after depositions on these spacer side walls [7,8]. This method eliminates the overlay alignment issues inherent with the LELE and LFLE methods, but at the added cost of additional processing steps [7,8]. Double Exposure Lithography (DEL) utilizes a specially designed photoresist (PR) with a non-linear exposure response that allows for generation of spatially doubled patterns [9,10]. As a promising DP technique, DEL does not require either phase shifting of the mask pattern or any additional processing steps. DEL does however require the use of a special dual-tone resist that, as yet, is not commercially available . Similarly, DTD (Dual-Tone Development) schemes use a two-step development process (a positive-tone develop followed by a negative-tone develop), where regions with higher than the maximum and lower than the minimum threshold exposure doses are developed away, leaving only the areas that have been exposed with intermediate doses. Various challenges still remain in formulating and qualifying such specialized processes and resists and reducing the line edge roughness (LER) generated by two wet processing steps .
Recently, several research groups have exploited the concept of a negative index material to develop optical “superlenses” that replicated sub-diffraction limited features from a sub-diffraction limited mask pattern [12–14]. Unfortunately, at optical wavelengths, a true negative index material (i.e., one with negative values for both permittivity and susceptibility) cannot be easily constructed. Thus, all have demonstrated the use of a material with negative permittivity (e.g., by working in the vicinity of a surface plasmon resonance), and thus limiting the application space to the “near field” of the superlens. With this approach, sub-diffraction limited features can be imaged, but only on PR layers that are within the near field of the superlens.
In this report, we present a new Double Patterning technique called Cavity Resonance Lithography (CRL). CRL utilizes wave interferences and field resonances within an optical cavity formed by the mask and the substrate (with PR and index matching immersion fluid being the dielectric medium of the cavity) to create spatially doubled sub-diffraction limited patterns in the far field. CRL requires only a single exposure and development step, similar to conventional lithography steps, and does not require any additional processing steps. In addition, CRL methods can be implemented with currently available off-the-shelf PRs and developers, and do not require optical proximity corrections. We first introduce the working principle of CRL and then report on the first proof-of-concept experimental results.
2. Theory and simulations
Figure 1 illustrates the principle of Cavity Resonance Lithography (CRL). The schematic shown in Fig. 1(a) was used in the proof-of-concept demonstration, wherein we performed lithography directly on top of the mask (this was done for ease of fabrication). The experiment can also be performed with an immersion layer separating the mask and the PR layer. The mask pattern is a periodic grating whose pitch is designed such that only the 0th and the 1st order ( + 1 and −1) diffracted waves can propagate; higher order waves are evanescent and decay in the near field of the mask surface [14,15]. The first order diffracted waves pass through a thin metal layer and enter the dielectric (PR) cavity. Their planar wave vectors (kx) have the same periodicity as the mask gratings (kp), and the interference of the waves traveling in opposite directions (kx and k-x) generates field patterns with a spatial frequency (2kp) twice that of the mask gratings . Because CRL utilizes the propagating – rather than the evanescent – portion of the diffracted waves, the generated field patterns are not restricted to the near field [12–15]. This technique thus enables lithographic processing with a large off-set distance between the mask and the substrate.
Along with the interference of planar components, phase-matched multiple reflections by the waves in z-direction induce a cavity resonance that greatly increases the field intensity and the image contrast. In this particular configuration, where the top reflector is a PR/Air interface (Fig. 1(b)), the diffracted angle θi is larger than the critical angle, θc, thus resulting in total internal reflection.
The resonance condition for Fabry-Perot type cavities can be written as:
where, m is the integer order, k is the wave vector inside the cavity, θi is the incident angle, φd is the phase change due to reflection at the PR/Air interface, and φm is the phase change due to reflection at the PR/metal interface. Rewriting Eq. (1) for d:
Equations (3) and (4) assume infinitely thick layers, and are approximations. Better estimates for the resonance condition can be obtained with FEM (Finite Element Method) simulations. Accordingly, we have performed FDFD (Finite Difference Frequency Domain) numerical studies with a commercial tool (COMSOL Multiphysics). In our simulations, we used exposure wavelength of λ0 = 193 nm, and the mask comprises a 25 nm thick tungsten (W) layer on a quartz substrate patterned with 1-dimensional infinite array of 40 nm wide openings on a 130 nm pitch. A 10 nm thick Al layer was used as the cavity ‘mirror’ layer which was separated from the mask by a 20 nm thick PMMA (Poly(methyl methacrylate) spacer/planarization layer (Fig. 2(a) ). We used published material properties for W and Al , and the refractive index of the PR cavity () was set to the experimentally characterized value of 1.57 . We also used the same value for the refractive index of PMMA.. We limited the size of our simulation mesh by using two mask periods with periodic boundary condition (so as to effectively simulate an infinite grating) combined with open boundaries for the Air and Quartz regions (top and bottom respectively).
Figures 2(a-c) summarize the simulated E-field intensity plots for the first three resonant cases occurring at PR thicknesses d = 200 nm, 388 nm, and 577 nm, corresponding to m = 1, 2, and 3, in Eq. (1), respectively. All three cases clearly show E-field patterns with 32.5 nm half-pitch, corresponding to a spatial doubling of the 65 nm half-pitch mask pattern. Figure 2(d) shows the maximum E-field intensity plot with varying cavity thickness, d, where the values of d for each of the three resonant peaks (blue solid), which are about an order of magnitude higher than that at off-resonance, corresponds to the thickness of Figs. 2(a), (b), and (c) respectively.
We have also simulated for cases with the Al layer (solid blue) and without the Al layer (dotted blue), the latter case corresponding to a simple arrangement of PR coated on a bare mask. Surprisingly, even without the reflective Al layer, the W mask itself acts as a decent ‘mirror’ for the cavity and demonstrates a degree of resonant peaks. The slight difference in the thickness (d) necessary for peak resonance for the two cases is a result of the differences in the phase shifts upon reflection for the two cases. We have also plotted the maximum contrast of the field intensities with the Al layer (solid red) and without the Al, layer (dotted red). For both arrangements, the contrast ranged from a low of 0.7 (off-resonance) to a high, at resonance, of nearly 1.0.
We have conducted a proof-of-concept lithography experiment (Fig. 3 ) utilizing an Al layer “mirror”. We fabricated a lithography mask by sputter depositing a 25 nm thick W film on a UV-transparent quartz wafer, followed by E-beam lithography using PMMA resist (495 A1, Microchem Corp.). We patterned linear openings in the PMMA on a 150 nm pitch, and transferred the pattern to the W film using RIE (Reactive Ion Etching, Oxford Instrument) with SF6: O2 = 4:1 ratio providing a directional W etch at about 1 nm/s. We planarized the mask with a 20 nm PMMA layer, and a 10 nm thick Al layer was sputter deposited. Finally, a 100 nm thick PR (positive tone 193 nm, TOK TArF-7a-144) was spin-coated directly on top of the Al layer. A deuterium lamp (Newport Corp.) was used to expose the PR from the substrate side as shown in Fig. 3(a). For a 150 nm pitch mask pattern, a PR thickness of d = 100 nm provides a reasonable cavity resonance with doubled spatial frequency, as evident from the simulation results in Fig. 3(b). The doubled spatial frequency (75 nm half-pitch) was successfully recorded by the PR as shown by the Atomic Force Microscopy (AFM) image in Fig. 3(c). A FFT (Fast Fourier Transform) analysis of the AFM image also confirmed the periodicity by clearly showing a peak at the 37.5 nm half pitch (Fig. 3(d)). To our knowledge, this is the first experimental demonstration of sub-diffraction limited spatial frequency doubling using the CRL method.
As suggested by our simulation results, the W mask itself can also act as a mirror for the resonant cavity (Fig. 2(d)). In order to prove this, we conducted a lithography experiment without the Al layer. As shown in Fig. 4(a) , we spun coated the PR directly on top of the mask (patterned with a linear grating with a 130 nm pitch). The optimal thickness of the PR layer was previously determined via simulations (Fig. 2(d)) to be 180 nm. The simulated E-field intensity pattern for this case is depicted in Fig. 4(b). A spatially doubled E-field pattern is evident inside the PR. The intensity contrast across the white dotted line was calculated to be 0.97. The lithography result, recorded via an AFM, is depicted in Fig. 4(c). In this case, the image contrast is not as high as in Fig. 3(c), which is consistent with the weaker cavity resonance expected without the Al layer, but the FFT analysis of the image (Fig. 4d) clearly indicates the presence of the 32.5 nm half pitch pattern.
4. Discussion and conclusion
We have reported an experimental demonstration of Cavity Resonance Lithography, a single exposure/develop lithography method that doubles the spatial frequency of a diffraction limited mask pattern and records in the far field. In our demonstration, we have used a broad band source (Deuterium lamp) with incoherent illumination (condenser output). The fidelity and the LER (Line Edge Roughness) of the recorded images can be significantly improved through the use of a single wavelength and a coherent light source. The offset distance (distance between the mask and the substrate) can be hundreds of microns, or even several millimeters given a sufficiently large mask pattern area. The image contrast decreases towards the edges of the patterned area as the diffracted waves “walk” out of the cavity, but this effect can be minimized through the use of “stable” cavity configurations . Precisely controlling the off-set distance and generating high field intensity will enable shorter exposure time as well.
For practical lithography applications, the PR should be coated on a separate substrate and immersion lithography techniques must be integrated with our CRL methods (Fig. 5(a) ). In this case, the mask and the substrate act as cavity mirrors, and the PR/index matching immersion fluid acts as the dielectric medium. CRL requires that the two cavity surfaces (mirrors) are parallel and free of severe topographical irregularities; such conditions can be readily achieved with any polished substrate surface and existing SOA litho tools. The minimum feature size (Smin) that can be recorded on the PR layer can be represented by Smin.CRL = λ/4n, where λ is the exposure wavelength in vacuum and n is the refractive index of the dielectric medium. By contrast, the diffraction limit is given by Smin,diff = λ/2n. In our demonstration, Smin,CRL is 31 nm with = 1.57, and we have achieved a result of 32.5 nm. The refractive index of available immersion fluids are approaching = 1.7  and that of existing PRs are already approaching = 1.8 . With the incorporation of these materials, the limit of Smin,CRL becomes approximately 28 nm. For example, Fig. 5(b) shows a simulation result with a 120 nm pitch mask pattern and PR/immersion fluid index of 1.7, which shows the possibility of generating 30 nm in half-pitch resolution using the CRL method. This resolution can be achieved with a single exposure and develop step, and without any constraints on the distance between mask and photoresist layers. These benefits along with the ability of the CRL method to be combined with other current double patterning methods, we believe that the CRL method offers significant potential to meet semiconductor lithography requirements.
The CRL theory is dependent on the phase shift upon reflection via Eq. (3) (this being a function of polarization). Thus, for masks with 2-dimensional patterns, the offset distance corresponding to maximum E-field intensity for the two polarization directions may be different. Fortunately, our simulations also suggest that the E-field contrast remains substantially independent of offset distance (see Fig. 2d for an example). Thus, it should be possible to create a sub-diffraction limited, 2-dimensional pattern with our technique. Also, our resonant cavity should have a low finesse number, which corresponds to the relaxed resonance condition (corresponding to the breath of the E-field maxima, see Fig. 2d) under which our methods work. Even at these relaxed resonance conditions, the buildup in the E-fields for the first order diffracted waves compared to the zeroth order transmission is substantial. Thus, we do not see a need to block the zeroth order transmission.
In these proof-of-concept demonstrations, we designed our masks to provide for the propagation of only the first order diffracted waves, (higher order waves are evanescent). However, with respect to real world implementations, a mask pattern might comprise features for which diffraction orders greater than 1 are propagated in the medium. This would complicate the pattern on the photoresist, but it also affords the opportunity to create non-periodic patterns. This is an avenue that we have not yet explored. Our methods do not require a decaying evanescent wave, and thus are not limited by near-field distances. Given that a resonant cavity can be defined at large offset distances, we anticipate that our methods can be applied to projection lithography as well. Finally, for the same reasons, we anticipate that the costs of the CRL mask would compare favorably with currently used optical proximity correction masks.
This Work was supported by the Defense Advanced Research Projects Agency (DARPA, Contract no. W31P4Q-09-C-0262, Dr. Devanand Shenoy) under the Small Business Innovation Research (SBIR) program.
References and links
1. P. Zimmerman, “Double patterning lithography: double the trouble or double the fun?” SPIE Newsroom (2009), http://spie.org/documents/Newsroom/Imported/1691/1691_5999_0_2009-06-24.pdf.
2. E. S. Putna, T. R. Younkin, M. Chandhok, and K. Frasure, “EUV lithography for 30nm half pitch and beyond: exploring resolution, sensitivity, and LWR tradeoffs,” Proc. SPIE 7273, 72731L (2009). [CrossRef]
3. C.-M. Lim, “Positive and negative tone double patterning lithography for 50nm flash memory,” Proc. SPIE 6154, 615410 (2006). [CrossRef]
4. M. Hori, T. Nagai, A. Nakamura, T. Abe, G. Wakamatsu, T. Kakizawa, Y. Anno, M. Sugiura, S. Kusumoto, Y. Yamaguchi, and T. Shimokawa, “Sub-40-nm half-pitch double patterning with resist freezing process,” Proc. SPIE 6923, 69230H (2008). [CrossRef]
5. W. H. Arnold, “Toward 3nm overlay and critical dimension uniformity: an integrated error budget for double patterning lithography,” Proc. SPIE 6924, 692404 (2008). [CrossRef]
6. J. Finders, M. Dusa, B. Vleeming, H. Megens, B. Hepp, M. Maenhoudt, S. Cheng, and T. Vandeweyer, “Double patterning for 32nm and below: an update,” Proc. SPIE 6924, 692408 (2008). [CrossRef]
7. A. K. Raub, D. Li, A. Frauenglass, and S. R. J. Brueck, “Fabrication of 22 nm half-pitch silicon lines by single-exposure self-aligned spatial-frequency doubling,” J. Vac. Sci. Technol. B 25(6), 2224–2227 (2007). [CrossRef]
8. C. Bencher, Y. Chen, H. Dai, W. Montgomery, and L. Huli, “22 nm half-pitch patterning by CVD spacer self alignment double patterning (SADP),” Proc. SPIE 6924, 69244E (2008). [CrossRef]
9. S. Lee, J. Byers, K. Jen, P. Zimmerman, B. Rice, N. J. Turro, and C. G. Willson, “An analysis of double exposure lithography options,” Proc. SPIE 6924, 69242A (2008). [CrossRef]
10. X. Gu, A. J. Berro, Y. Cho, K. Jen, S. Lee, T. Ngai, T. Ogata, W. J. Durand, A. Sundaresan, J. R. Lancaster, S. Jockusch, P. Zimmerman, N. J. Turro, and C. G. Willson, “Fundamental study of optical threshold layer approach towards double exposure lithography,” Proc. SPIE 7273, 72731C (2009). [CrossRef]
11. C. Fonseca, M. Somervell, S. Scheer, W. Printz, K. Nafus, S. Hatakeyama, Y. Kuwahara, T. Niwa, S. Bernard, and R. Gronheid, “Advances and challenges in dual-tone development process optimization,” Proc. SPIE 7274, 72740I (2009). [CrossRef]
15. H. Lee, Y. Xiong, N. Fang, W. Srituravanich, S. Durant, M. Ambati, C. Sun, and X. Zhang, “Realization of optical superlens imaging below the diffraction limit,” N. J. Phys. 7, 255 (2005). [CrossRef]
16. T. E. Jewell and D. L. White, “Spatial frequency doubling lithography (SFDL) of periodic structures for integrated optical circuit technology,” J. Lightwave Technol. 7(9), 1386–1393 (1989). [CrossRef]
17. M. Born and E. Wolf, Principles of Optics: Electromagnetic Theory of Propagation, Interference and Diffraction of Light, 7th ed. (Cambridge University Press, 1999).
18. E. D. Palik, Handbook of Optical Constants of Solids (Academic Press, 1997).
19. The refractive index of 200 nm thick PR coated on a bare Si wafer was measured at λ = 193 nm using ellipsometry; J. A. Woollam Co., Inc.
20. A. Yariv, Optical Electronics (Holt McDougal, 1984)
21. M. K. Yang, S. G. Kaplan, R. H. French, and J. H. Burnett, “Index of refraction of high-index lithographic immersion fluids and its variability,” J. Micro/Nanolithogr. MEMS MOEMS 8(2), 023005 (2009). [CrossRef]
22. K. Matsumoto, E. A. Costner, I. Nishimura, M. Ueda, and C. G. Willson, “High index resist for 193 nm immersion lithography,” Macromolecules 41(15), 5674–5680 (2008). [CrossRef]