Issue



Phase-shifting methods at 65nm: A comparison of AAPSM and CLM


10/01/2003







Overview

Traditionally, the shrink of semiconductor IC devices has been enabled primarily through the reduction of the exposure wavelength. Optical lithography capabilities, however, have been extended to the sub-100nm regime through heavy utilization of reticle-based resolution enhancement techniques. With existing challenges (including the delay in introduction of 157nm lithography and other next-generation lithography approaches), most teams developing the 65nm node are relying on resolution enhancement techniques-based 193nm lithography.


Figure 1. Through-pitch MEEF as a function of different mask CD ranges for a) CLM and b) AAPSM.
Click here to enlarge image

null

There are many types and variations of resolution enhancement techniques (RETs) under consideration for sub-100nm lithography applications, although the alternating aperture phase-shift mask (AAPSM) has excellent imaging properties such as high image contrast, low mask error enhancement factor (MEEF), and tight critical dimension (CD) control [1, 2]. Dark-field AAPSM double-exposure technology has been successfully implemented in a production environment at the 130nm design node using 248nm lithography [3, 4] and is currently in a development stage for sub-100nm applications based on 193nm lithography.

CLM is a single-exposure technology that is based on a combination of a chrome-less shifter-shutter PSM and a strong off-axis illumination (OAI) [5]. The shifter-shutter PSM is also referred to as a binary halftone chrome-less PSM or a 100% transmission attenuated PSM. Stand-alone, this technology provides moderate resolution enhancement, but its performance is significantly improved when combined with high projection lens numerical apertures and optimized quasar or dipole OAI. In recent years, there have been a number of studies reporting on the progress of the development of the CLM technology with an outlook toward the 65nm design node [6, 7].

Imaging performance, MEEF, and the process window that results from using AAPSM and CLM are discussed; an overview of relative printability is also presented [8].

Inherent imaging performance

Aerial image properties of alternating aperture and shifter-shutter PSMs provide a useful insight into imaging capabilities and limitations of both technologies. A fully transparent, 180° phase-shifted CLM feature forms a dark line due to the proximity interaction of the feature's phase-edge images. Assuming CLM-optimized imaging conditions [5], features with edge separation of 150nm or larger form a separate dark line per edge, essentially acting as a phase-edge PSM. In order to form a single dark line, feature sizes have to be about 100nm or smaller. On the other hand, AAPSM has more conventional imaging properties, and forms a high-contrast aerial image regardless of the width of the chrome feature on the mask or even in its absence. Overall, image CDs scale proportionally with chrome linewidths on the mask, although minimum image dimensions start to saturate as mask feature size approaches zero. Unlike the CLM, whose performance benefits from strong OAI, AAPSM imaging is enhanced by high coherence illumination and does not require ultra-high lens numerical aperture (NA) for best results.


Figure 2. EL-DOF process window comparison between AAPSM and CLM for different pitch values. The CLM results were obtained at the following imaging conditions: 193nm exposure wavelength, 0.85NA, QASAR OAI (0.55/0.85/30°). AAPSM results were based on a 193nm exposure wavelength, 0.75NA, and 0.3s.
Click here to enlarge image

null

Aerial image properties as a function of pitch are indicative of imaging robustness of a given technology and the level of complexity needed for correction of optical proximity effects. Image CD at a given threshold and normalized image log-slope (NILS) are commonly used to quantify image fidelity. Assuming a 193nm exposure wavelength, a numerical aperture of 0.85, and a QASAR OAI (0.55/0.85/30°), CLM imaging exhibits a large, nonuniform CD variation through pitch. It amounts to about 30nm or more, depending on the mask feature size. Also, the image quality exhibits a rapid degradation as pattern pitches fall below 220nm. In contrast to the CLM, the AAPSM through-pitch CD uniformity is noticeably better. The pitch proximity effect is consistent as a function of mask feature dimension — through-pitch image CDs scale uniformly with mask feature CDs. Additionally, AAPSM provides an enhanced resolution of denser lines (pitches <250nm) even at a smaller 0.8 NA. Detailed CD and NILS quantitative results have been described in a previous publication [9].

MEEF is an important lithographic parameter, which is directly linked to the process performance. Through-pitch MEEF has been evaluated for CLM and AAPSM technologies and the results are captured in Fig. 1. In order to verify the sensitivity of MEEF to mask feature size, the evaluated mask CDs varied from 50nm to 90nm, which is a relevant feature size range for the 65nm technology node. CLM results (Fig. 1a) show a very inconsistent MEEF behavior. It varies significantly through pitch and exhibits a strong dependence on the mask feature size. MEEF properties of AAPSM as a function of pitch are very different from those of the CLM. MEEF variation of AAPSM across pitch and as a function of mask feature size is very uniform and consistent (Fig. 1b). Almost insensitive to chrome feature dimensions varying from 50nm to 90nm, excellent AAPSM MEEF characteristics allow not only a superior CD control, but also improve the effectiveness of the optical proximity correction (OPC).

In addition to the previously discussed parameters, exposure latitude depth-of-focus (EL-DOF) process windows have also been evaluated for different pitches ranging from dense to isolated (Fig. 2). Results show that AAPSM yields larger common exposure latitude than the CLM, even though the AAPSM imaging was based on a smaller NA. Overall, AAPSM process windows are significantly larger for dense and isolated features in comparison to the CLM results, while they are comparable for semi-dense patterns. Also, it is important to note that the CLM process window calculated for 160nm pitch is not adequate in size for a manufacturable solution due to the lack of sufficient resolution. In the case of isolated lines, CLM has a limited focus latitude of about 200nm at 10% EL. However, this issue can be potentially addressed through the application of sub-resolution scattering bars. AAPSM results suggest that an overlap process window of about 300nm of DOF at 10% EL might be attainable at the 65nm technology node.

Test cell printability

The test cell (Fig. 3a) contains a 65nm poly-Si layer with a variety of features with pitches ranging from 160nm to nearly isolated, and different configurations of corners and line-ends. Synopsys' OPC conversion tool was utilized to generate the CLM and AAPSM test layouts. Figure 3b shows the resulting CLM layout that contains model-based OPC, and sub-resolution scattering bars, which were applied using a rule-based approach. The scattering bar size used was 40nm and the placement distance from the main feature was set to 120nm. In order to evaluate the AAPSM dark-field double-exposure technology, the test layout was separated into two parts for the PSM and the binary trim exposures. The layout transformation was performed using Synopsys' FullPhase technology, which allows phase shifting of all of the poly-Si features, including gates and field poly patterns [10, 11]. These results were obtained using a prototype tool that is currently under development. Figure 3c shows the PSM layout, and the trim mask is depicted in Fig. 3d. Model-based OPC was applied to both PSM and trim layouts. Overall, OPC methodology used for both CLM and AAPSM layouts was based on a relatively simple optimization scheme and would require further tuning to obtain best performance.


Figure 3. a) A 65nm logic test cell converted into b) a CLM and c) a FullPhase mask that consists of a full-field alternating-aperture PSM, and d) a binary trim mask.
Click here to enlarge image

null

Aerial simulation results are captured in Fig. 4 (AAPSM — 4a and 4b; CLM — 4c and 4d). Figures 4a and 4c represent results at best focus, while Figs. 4b and 4d are obtained at 100nm defocus. Consistent with previously described results, both AAPSM and CLM provide sufficient resolution for semi-dense and isolated features at best focus and 100nm defocus conditions. Scattering bars benefit the through-focus behavior of isolated and semi-isolated CLM patterns. AAPSM results show a higher across-pitch image contrast, however, which is especially apparent for dense features with a 160nm pitch size. The CLM resolution capability at 160nm pitch is clearly insufficient, while AAPSM yields a satisfactory performance even at a smaller NA. OPC performance is quite good for both technologies, but, undoubtedly, it could be further improved through additional optimization of OPC settings, as well as development of other necessary OPC capabilities and features not available at the present time.


Figure 4. 2-D images of AAPSM test cell at a) best focus and at b) 100nm defocus. c) CLM images at best focus and d) 100nm defocus. Imaging conditions used for the AAPSM are 193nm exposure wavelength, 0.8NA, and 0.35s for both PSM and trim binary exposures; image threshold at 30% for the PSM, and 40% for the binary mask. The CLM results are based on a 193nm exposure wavelength, 0.85NA, QASAR OAI (0.6/0.9/30°), and 30% threshold level.
Click here to enlarge image

null

Conclusion

An evaluation of AAPSM and CLM technologies in light of performance requirements for the 65nm technology node based on 193nm lithography shows that AAPSM has superior performance over CLM combined with a "strong" OAI and a higher lens NA. AAPSM demonstrated a higher through-focus image quality (NILS), the capability to resolve smaller pitches than the CLM, and a better through-pitch CD uniformity.

Also, AAPSM results showed a lower and a more consistent MEEF and a larger through-pitch overlap process window. Although the CLM has a sufficiently good performance for semi-dense patterns, it does not have a resolution capability for dense pitches <200nm; large variations of MEEF as a function of pitch and its great sensitivity to the mask feature size will complicate CD control. Additionally, CLM has a "difficult-to-correct" pitch proximity signature that requires a complicated scheme of proximity correction including model-based OPC and scattering bars.

Armen Kroyan, Hua-Yu Liu, Synopsys Inc., Mountain View, California

Acknowledgments

IC WorkBench and FullPhase are trademarks of Synopsys.

References

  1. M.D. Levenson et al., "Improving Resolution in Photolithography with a Phase-Shifting Mask," IEEE Transactions on Electron Devices, Vol. ED-29, No. 12, pp. 1828–1836, 1982.
  2. H-Y. Liu et al., "The Application of Alternating Phase-Shifting Masks to 140nm Gate Patterning: Line Width Control Improvements and Design Optimization," SPIE Proc., Vol. 3236, 1997.
  3. G.P. Watson et al., "A 2-Million Transistor Signal Processor with 120nm Gates Fabricated by 248nm Wavelength Phase-Shift Technology," MNE '99.
  4. M.E. Kling et al., "Practical Extension of 248nm DUV Optical Lithography Using Trim-Mask PSM," Proc. SPIE, Vol. 3679, pp. 10–17, 1999.
  5. J.F. Chen et al., "Binary Halftone Chromeless PSM Technology for l/4 Optical Lithography," Proc. SPIE, Vol. 4346, pp. 515–533, 2001.
  6. J.S. Petersen et al., "Development of a Sub-100nm Integrated Imaging System Using Chromeless Phase-Shifting Imaging with Very High NA KrF Exposure and Off-Axis Illumination," Proc. SPIE, Vol. 4691, pp. 515–529, 2002.
  7. D. Van Den Broeke et al., "Tuning MEEF for CD Control at 65nm Node Based on Chromeless Phase Lithography," Proc. SPIE, Vol. 4889, pp. 579–591, 2002.
  8. Modeling is performed using Synopsys' IC WorkBench lithography simulator.
  9. A. Kroyan et al., "Resolution Enhancement Technology Requirements for 65nm Node," Proc. SPIE, Vol. 5042, 2003.
  10. .C. Pierrat et al., "New Alternating Phase-Shifting Mask Conversion Methodology Using Phase Conflict Resolution," Proc. SPIE, Vol. 4691, pp. 325–335, 2002.
  11. C. Pierrat et al., "Full Phase-Shifting Methodology for 65nm-Node Lithography," Proc. SPIE, Vol. 4889, pp. 558–567, 2002.

Armen Kroyan received his PhD in electrical engineering from Rice University in 1999, specializing in the area of optical resolution enhancement techniques for photolithography. He is a senior engineer in lithography technology at Synopsys, 700 E. Middlefield Rd., Mountain View, CA 94043; ph 650/584-2659, fax 650/584-4016, e-mail [email protected].

Hua-Yu Liu manages the mask synthesis technical marketing and lithography application group for Synopsys. She has worked on advanced lithography including electron-beam resist, process and exposure systems, various resolution enhancement technologies, and process control for Hewlett-Packard Laboratories and Numerical Technologies. She has also authored and co-authored more than 50 technical papers.