Copyright 2011 American Scientific Publishers All rights reserved Printed in the United States of America Journal of Nanoscience and Nanotechnology Vol. 11, 1 6, 2011 In-Situ Monitoring of Pattern Filling in Nano-Imprint Lithography Using Surface Plasmon Resonance Hong Hocheng 1, Wei-Hsuan Hsu 1, and Jow-Tsong Shy 2 1 Department of Power Mechanical Engineering; 2 Department of Physics, National Tsing Hua University, 101 Section 2 Kuang Fu Road, Hsinchu, TAIWAN 30013, Republic of China Nano-imprint lithography possesses the advantages of high throughput, sub-10-nm feature and low cost. In NIL, the mold filling is subjected to the applied imprinting pressure, temperature and time. Incomplete mold filling causes a detrimental effect on the final imprinted pattern dimensions. The monitoring system of imprinting is essential to control the imprinting parameters properly. Up to now, no high-sensitivity monitoring of filling rate and end point has ever been proposed. In this study, the authors apply the surface plasmon resonance to monitor the filling rate and end point during imprint process. The mold contains a layer of glass of high refractive index, a metal thin film and the pattern of low refractive index. In addition, the imprinted polymer is selected considering its refractive index, which should be lower than the glass layer of mold. When the filling rate varies, it will affect the SPR behavior, including the measurable reflectivity change and resonance angle shift. The analysis results reveal that the resonance angle is truly proportional to the filling rate. When the filling rate varies from 50% to 100%, the SPR angle shifts more than 5 degree. The analysis demonstrates this innovative method for monitoring of filling rate is effective with high sensitivity. Keywords: Nanoimprint, Imprint Lithography, Filling Rate, Monitoring, Surface Plasmon Resonance (SPR). 1. INTRODUCTION In the past decades of years, industrial technology which followed Moore s law has grown in semiconductor and integrated circuit. The crucial technology of photolithography has influenced the line width to be exposed. However, when the pattern feature goes below 32 nm, the photo-lithography process becomes difficult. The line widths are limited by the wavelength diffraction, posing a challenge in physics. The cost of manufacturing process and exposing equipment will increase when the pattern size decreases. For this reason, a low-cost and highthroughput manufacturing for nanostructures is essential. The semiconductor industry works hard exploring every kind of technology in order to lead the industry following the Moore s law. Using the 193 nm light source in coordination with immersion technology, optical proximity correction (OPC), phase shift mask (PSM) and double patterning technique can achieve the exposed pattern of the 45 nm and 32 nm node; nevertheless, the 22 nm node technology is still to be developed. There are two potential methods for 22 nm node technology; one is developing Author to whom correspondence should be addressed. shorter wavelength of the light source for exposure, and the other is nano-imprint lithography (NIL). 1 2 NIL allows for fast replication of nano-scale patterns. It is a promising technique for wide applications, including the electronic device, bio-sensor, micro-channel, highdensity memory disk and optical-device. Compared with other nano-scale fabrication techniques, it possesses the advantage of low cost and high throughput. To make this technology practical for industry, not only the high resolution but also be reliable patterning over a large area has to be achieved. NIL was reported fifteen years ago, 1 and since then there have been three aspects concerned, namely the printing in nano-scale crucial dimension, 3 fabricating structure with high aspect ratio, 4 and imprinting on large area. 5 6 However, these three conditions have barely been satisfied at the same time in the published references. It reveals that there are a lot of problems involved in NIL which need to be analyzed and solved. On the other hand, increasing uniformity of imprinted pattern 7 8 is another key point for successful reactive ion etching (RIE) after imprinting process. The uniform imprinted pattern means the residual layer thickness of imprinted pattern is equal across the patterned area. How to ensure the quality of imprinted J. Nanosci. Nanotechnol. 2011, Vol. 11, No. xx 1533-4880/2011/11/001/006 doi:10.1166/jnn.2011.3780 1
In-Situ Monitoring of Pattern Filling in Nano-Imprint Lithography Using Surface Plasmon Resonance Hocheng et al. pattern and increase the throughout of imprinting process are essential for industry. The mold filling of NIL is subjected to the applied imprinting pressure, temperature, and time. Incomplete mold filling has a detrimental effect on the final imprinted pattern dimensions. In order to avoid above problem, the imprinting time will be increased. However, the excessive imprinting time will decrease the throughput and cause the chemical change of polymer. The quality of imprinted pattern is concerned with the formation. In observation of the form by Scanning Electron Microscopy (SEM), the wafers must be destroyed to measure the cross-section profile. Although the SEM can provide feature shape data in nano-scale, the wafers still have to be taken off-line for measurement. Therefore, this kind of measuring technique can not improve yield without reducing productivity. An in-situ monitoring system is essential to measure the formation of imprinted pattern during imprinting process. There are two monitoring methods have been proposed in the published reference; one is time-resolved diffractive scatterometry, 9 10 and the other is capacitance. 11 There methods are innovative while disadvantages as for either non-periodic structure or signal robustness, hence limit their wide applications. Up to now, no high-sensitivity monitoring of filling rate and end point has been proposed in the published reference. This paper is devoted to design and set up a monitoring system for the filling rate to control the process parameter and enhance yield. 2. PRINCIPLE AND METHOD In this study, the authors apply the surface plasmon resonance (SPR) to monitor filling rate during imprint process. A prism needs to be placed on the backside of imprinting mold to couple the monitoring light, and a special mold structure needs to be designed for total reflection. The mold contains a high refractive index layer of glass, a metal thin film (gold or silver) and a low refractive index pattern. Figure 1 is the schematic illustration of the proposed system. In addition, the imprinted polymer is selected considering its refractive index, which should be lower than the glass layer of mold. The commonly used PMMA is an example. When the filling rate varies, it will change the resonance behavior, including the measurable reflectivity and resonance angle, which is defined as the incidence angle of the minimum reflectivity. The most common application of SPR is biosensor. 12 The practicability and feasibility of SPR for monitoring has been proved. In this case, SPR measurements are not only sensitive to changes in the refractive index of the medium surrounding the sensor, but also to the thickness of the sensor layer. Therefore, the surrounding location Fig. 1. Schematic illustration of the monitoring system. of the gold film will affect the SPR behavior greatly. For this reason, the SPR behavior of the proposed monitoring method will be influenced by refractive index of mold, polymer/resist and substrate. 3. THEORETICAL ANALYSIS BY SPR METHOD In this research, the simulation tool, PCGrate, is used to analyze the SPR behavior in varying dimensions of mold, thickness of imprinted polymer, wavelength of monitoring light source and substrate material. 3.1. Effect of Monitoring Light Source and Substrate Material In an optical system, the light source will affect the resolution and inaccuracy. The wavelengths of 632.8 nm (He Ni Laser) and 670 nm (Laser diode) are selected to analyze. Fig. 2. Simulation model and parameters for analyzing the effect of light source and substrate material. 2 J. Nanosci. Nanotechnol. 11, 1 6, 2011
Hocheng et al. In-Situ Monitoring of Pattern Filling in Nano-Imprint Lithography Using Surface Plasmon Resonance Therefore, the 670 nm light source is chosen for the experiment. 3.2. Effect of Mold Structure and Initial Polymer Thickness Fig. 3. Effect of light source and substrate material on reflectivity spectra. On the other hand, the light absorption of substrate material needs to be understood in the proposed monitoring method. It will affect the applied field of the proposed monitoring method. The substrate of silicon and quartz are selected. The light absorption of silicon substrate needs to be considered; nevertheless, the absorption of quartz can be ignored. The schematic representation of simulation model and parameters are shown in Figure 2. In this simulation, PMMA is used for imprinted polymer and the mold cavity is ignored for simplified analysis. Figure 3 shows the simulation results. The results reveal that the effect of substrate absorption is unclear when the distance between gold layer and substrate is 500 nm. The results demonstrate that the application of proposed monitoring method is not limited by substrate material. On the other hand, the reflectivity spectra are different from 632.8 nm and 670 nm light source. The results reveal that the longer wavelength can induce a smaller resonance angle and a sharper reflectivity curve. 670 nm-wavelength light source almost approach the limit of visible light. The mold for this investigation contains a high refractive index layer of glass, a metal thin film and a low refractive index pattern. However, the thickness and form of designed mold will affect the SPR behavior which includes the thickness of gold, residual thickness of silicon oxide layer and pattern structure (periodicity and aspect ratio). On the other hand, the effect of the initial polymer thickness for imprinted on SPR behavior need to be realized. In NIL, the initial polymer thickness is selected according to the scale of mold structure. The typical polymer thickness in NIL is between 100 nm and 500 nm. In order to realize the effect of above mention on SPR behavior, the different mold structure and initial polymer thickness will be analyzed during the imprint process. The schematic representation of simulation model and parameters are shown in Figure 4 and Table I, respectively. 3.2.1. Thickness of Gold Layer In the SPR-based biosensing applications, the depth of the reflectivity dip depends on the thickness of the gold film. 13 The optimum coupling thickness of gold is about 50 nm. The thickness of gold film affects the performance of exciting surface plasmon. In the proposed monitoring method, the optimum thickness of gold film needs to be analyzed. Figure 5 shows the relationship between the reflectivity and incidence angle in different gold thickness at full filling condition. The angular reflectivity spectra exhibit distinct dips that are associated with the transfer of energy from the incident light wave into a surface plasmon and its subsequent dissipation in the gold film. The strongest excitation of the surface plasmon occurs in 50 nm gold film, and this result is similar to the published reference. 13 Fig. 4. Simulation model and parameters for analyzing the effect of the mold structure. J. Nanosci. Nanotechnol. 11, 1 6, 2011 3
In-Situ Monitoring of Pattern Filling in Nano-Imprint Lithography Using Surface Plasmon Resonance Hocheng et al. Table I. Parameters in analysis. Feature size and light source of simulation model (Unit: nm) Thickness of gold layer 40, 50 a, 60 and 70 Residual thickness of oxide layer 20, 100 a and 180 Height of mold cavity 200 a, 400 and 600 Width of mold cavity 500 a Period of mold cavity 1000 a, 1500 and 2000 Initial thickness of PMMA 100, 300 and 500 a Wavelength of light source 670 a The parameters marked with a superscript a are used to analyze the effect of other feature size on SPR behavior. Thin residual silicon oxide can produce large resonance angle shift; nevertheless, the curve of reflectivity spectra is not sharp enough to ascertain the resonance angle. Increasing the residual oxide layer produces sharp reflectivity spectrum. Figure 7 is the relationship between resonance angle and filling rate at different residual silicon oxide thickness. (a) 3.2.2. Residual Thickness of Silicon Oxide Layer In the proposed mold structure, the silicon oxide is prepared to fabricate the mold cavity. In order to protect the gold film, the residual silicon oxide layer is retained in the proposed mold, as shown in Figure 4. The residual silicon oxide layer is near the gold layer which is possible to seriously affect the performance of exciting surface plasmon. Therefore, the influence of residual silicon oxide on the SPR behavior needs to be analyzed. The results agree with above forecast. The residual thickness of silicon oxide layer seriously affects the performance of exciting surface plasmon. The reflectivity spectra are shown in Figures 6(a c). In 20 nm-thickness residual silicon oxide, the filling ratio increases from 0% to 100% which can obtain 15.729 degree resonance angle shift. The proposed monitoring method has large resonance angle shift which means the method has high-sensitivity for monitoring filling rate. Figure 6(b) is the result of 100 nm-thickness residual oxide. As the filling ratio increases from 0% to 100%, the resonance angle shift increases from 58.438 degree to 65.071. The resonance angle shift is 6.633 degree. As the thickness of residual silicon oxide increases from 100 nm to 180 nm, the resonance angle shift decreases from 6.633 degree to 2.162 degree, as shown in Figure 6(c). (b) (c) Fig. 5. Effect of gold thickness on reflectivity spectra. Fig. 6. Effect of residual silicon oxide thickness on reflectivity spectra. 4 J. Nanosci. Nanotechnol. 11, 1 6, 2011
Hocheng et al. In-Situ Monitoring of Pattern Filling in Nano-Imprint Lithography Using Surface Plasmon Resonance When the cavity filling is near the end, the similar resonance angles can be observed at different residual silicon oxide thickness, and the effect of residual silicon oxide thickness on resonance angles is minor. It means that the signal of end point will not be affected by residual oxide thickness which is advantageous for monitoring end point. The proposed approach is extremely accurate and far superior to general SPR bio-sensor of which the resonance 14 15 angle shift is about 0.4 degree. 3.2.3. Period and Aspect Ratio of Mold Cavity The structure of mold cavity is not a designed parameter in the proposed monitoring method. The form, period and aspect ratio of mold cavity is designed for the various applications. Therefore, the structure of mold cavity has any influence on SPR behavior; it will reduce the value of the proposed idea and limit the applied field. Therefore, the effect of mold cavity on SPR behavior must to be investigated. Figure 8 depicts the effect of period of mold cavity on resonance angle during imprinting process. One can discover the common feature as follows. At 0%-filling rate, the resonance angle of 1000, 1500 and 2000 nm periods are 58.45, 59.62 and 60.68 degree, respectively. The period of mold cavity impacts significantly on resonance angle at initial filling stage. When the filling rate change from 50 to 100%, the variation of resonance angle is approach linear response. One the other hand, the variation of resonance angle is apparent when the filling rate nears the end which can demonstrate that the proposed idea has the high-sensitivity for monitoring the end point of filling. In the end of cavity filling, the resonance angle of 1000, 1500 and 2000 nm periods are 65.07, 64.94 and 64.88 degree, respectively. It means that the effect of cavity period on resonance angle is insignificant during final filling stage which is advantageous for a monitoring system. Fig. 8. Effect of cavity period on resonance angle during imprint process. Figure 9 depicts the change of resonance angle at different aspect ratio during imprint process. When the aspect ratio of mold cavity increases, the variation of the resonance angle is unapparent at initial filling stage. The reason is that when the aspect ratio of mold cavity increases, it will increase the distance between imprinted polymer and gold layer which will decrease the influence on SPR behavior. At the final filling stage, resonance angle is close to the same at different aspect ratio of mold structure which is advantageous for a monitoring system. 3.2.4. Effect of Initial Polymer Thickness Figure 10 shows the effect of initial polymer thickness on resonance angle during imprint process. One can discover that the results of 500 and 300 nm polymer thickness are exactly similar, and the result of 100 nm polymer thickness has little diversity with other. It means that the SPR Fig. 7. Relationship between resonance angle and filling rate at different residual oxide thickness. Fig. 9. Effect of cavity aspect ratio on resonance angle during imprint process. J. Nanosci. Nanotechnol. 11, 1 6, 2011 5
In-Situ Monitoring of Pattern Filling in Nano-Imprint Lithography Using Surface Plasmon Resonance Hocheng et al. reveal the filling rate varies from 50% to 100%, the SPR angle shifts more than 5 degree. The analysis demonstrates this innovative method for monitoring of filling rate and end point is effective with high sensitivity. This emerging application of SPR promises the enhanced nano-imprint performance in practice. Acknowledgment: The authors are grateful to Mechanical Industry Research Laboratories of Industrial Technology Research Institute for the simulation software, PCGrate. References and Notes Fig. 10. Effect of initial polymer thickness on resonance angle during imprint process. behavior will be affected by substrate absorption when the initial thickness of polymer layer is decreased. When the dimensions of pattern structure are reduced, the selected initial polymer will be decreased for obtaining thin residual layer. Therefore, it is necessary to consider the effect of substrate absorption on SPR behavior. One can increase the residual thickness of silicon oxide layer to avoid above problem. 4. CONCLUSIONS In NIL, the transferred pattern is defined by the imprinting process. The non-ideal contact behavior between mold and substrate causes incomplete filling. For this reason, the monitoring system is essential to avoid the above problems. A successful monitoring method has to detect the filling rate during imprinting process. The simulation results 1. S. Y. Chou, P. R. Krauss, and P. J. Renstrom, Appl. Phys. Lett. 67, 3114 (1995). 2. S. Y. Chou, P. R. Krauss, and P. J. Renstrom, Science 272, 85 (1996). 3. M. D. Austin, H. Ge, W. Wu, M. Li, Z. Yu, D. Wasserman, S. A. Lyon, and S. Y. Chou, Appl. Phys. Lett. 84, 5299 (2004). 4. Y. Hirai, S. Yoshida, N. Takagi, Y. Tanaka, H. Yabe, K. Sasaki, H. Sumitani, and K. Yamamoto, Jpn. J. Appl. Phys. 42, 3863 (2003). 5. C. Perret, C. Gourgon, F. Lazzarino, J. Tallal, S. Landis, and R. Pelzer, Microelectron. Eng. 73 74, 172 (2004) 6. S. Landis, N. Chaix, C. Gourgon, C. Perret, and T. Leveder, Nanotechnology 17, 2701 (2006). 7. H. Hocheng and W. H. Hsu, Jpn. J. Appl. Phys. 46, 6370 (2007). 8. H. Hocheng and W. H. Hsu, J. Nanosci. Nanotechnol. 9, 4267 (2009). 9. Q. Xia, Z. Yu, H. Gao, and S. Y. Chou, Appl. Phys. Lett. 89, 073107 (2006). 10. Z. Yu, H. Gao, and S. Y Chou, Nanotechnology 18, 065304 (2007). 11. H. Hocheng and C. C. Nien, Jpn. J. Appl. Phys. 45, 5590 (2006). 12. J. Homola, S. S. Yee, and G. Gauglitz, Sens. Actuator B-Chem. 54, 3 (1999). 13. J. Homola, Surface Plasmon Resonance Based Sensors, Springer, Berlin (2006). 14. B.-K. Oh, Y.-K. Kim, W. Lee, Y. M. Bae, W. H. Lee, and J.-W. Choi, Biosens. Bioelectron. 18, 605 (2003). 15. W. Lee, D.-B. Lee, B.-K. Oh, W. H. Lee, and J.-W. Choi, Enzyme Microb. Technol. 35, 678 (2004). Received: 31 July 2009. Accepted: 19 January 2010. 6 J. Nanosci. Nanotechnol. 11, 1 6, 2011