Early results show much promise for immersion
05/01/2004
In the past year, a great deal of hope has been placed on immersion lithography technology to extend 193nm optical wavelength step-and-scan systems beyond 65nm processes for IC manufacturing. This article describes the results of a joint program to prove the feasibility of immersion technology in high-volume production.
Almost all leading-edge companies have added immersion to their lithography roadmaps. Immersion lithography involves adding a fluid between the bottom lens of an optical exposure tool and the wafer to increase the refractive index [1–4]. The higher refractive index provides better resolution by supporting a higher numerical-aperture (NA) lens value compared to conventional dry lithography tools. The higher refractive index also provides increased depth of focus (DOF) at the operating NA.
Although the concept sounds relatively simple, many technical issues must be resolved before immersion lithography moves into production [3, 4]. Over the past year, studies were conducted by Nikon Corp. and Tokyo Electron Ltd. (TEL) in a joint program to prove the feasibility of immersion for high-volume production. The most critical components of immersion were studied, including the fluid fill method, edge exposure capability, immersion projection optics, real-time focus control, water supply, polarization effect and polarized illumination, and the impact on resist.
Based on the results of the studies, the joint program defined methods to meet production requirements and concluded that immersion lithography using 193nm wavelength is feasible. Both Nikon and TEL are moving forward with product development of scanners and track systems, respectively, to meet future immersion requirements. Nikon has started construction of a full-field immersion scanner with NA = 0.85, and plans to ship production systems with NA≥1.0 in 2005.
Adding water
For 193nm immersion lithography, pure water looks to be the best liquid, with a refractive index of 1.44 and material properties that are required for immersion. Several methods have been proposed for putting the water between the lens and the wafer. Among them, the local fill method [3] has been considered most promising. In tests, a 1mm gap was filled with water. The wafer was moved back and forth to ensure constant water fill in the gap, as well as no leakage or water spotting (Fig. 1).
Local fill requires the least amount of change to existing dry 193nm scanners, and it has also worked with high-speed scanning stages. Nikon reported the possibility of this method by simulations in early 2003. After that, several designs were constructed and evaluated. By improving the nozzle design, we were able to prove the local fill method works with scan speeds of 500mm/sec, comparable to conventional dry systems. Therefore, overall system throughput is expected to be similar to new dry scanners.
One concern about the local fill method is the ability to do edge exposures while maintaining the local fill area and preventing water leakage onto the backside of the wafer or inside the wafer stage area. When the local fill nozzle moves to the edge of the wafer, the water could leak through the gap between wafer and stage. Additionally, since a vacuum is used to hold the wafer, water may be drawn to the wafer' backside. This problem could cause the wafer to stick on the wafer stage and require additional drying steps. To prevent this, we developed a method to protect the edge against water leakage. Evaluations have demonstrated the ability to perform edge-shot exposures with leakage protection and no water droplets on the backside of wafers.
Immersion projection optics
The table below shows Nikon projection-lens designs that meet our target for extremely low aberration levels and have a lens overall size compatible with standard exposure-tool platforms.
 |
Most of the designs listed have 4× reduction ratios. Generally, the 4× design is more difficult than the higher reduction-ratio designs, such as 5×, 6×, or 8×. We can attain an NA of 1.00 using 4× reduction with an all-refractive design and field size at 26×33mm, the current industry standard. If we adopt a catadioptric design, which requires the addition of mirrors into the optics, an NA of 1.20 is possible, while still maintaining 4× reduction and a full 26×33mm field.
On the other hand, if the reduction ratio is changed to 8×, we can adopt an all-refractive design to attain NA = 1.20. In this case, the field size is reduced to one-quarter of the standard 26×33 mm. The 8× reduction design has the advantage of using an all-refractive design and could also enable lower-cost 8× masks. The 8× reduction design will significantly reduce the system throughput, however, requiring 4× the number of exposures to cover a wafer. This throughput loss is unacceptable for high-volume production. Nikon's catadioptric lens designs allow 193nm argon-fluoride (ArF) immersion to be extended to the 32nm process node, using standard 4× reduction masks and a standard field size (Fig. 2).
Real-time focus control
One question commonly asked about immersion systems is whether or not it is possible to perform real-time focus control. Precise focus control is critical to ensure the best possible CD control. Conventional focus-control systems use optical beams (visible to infrared) passing through the space between the bottom lens and the wafer. The light reflected off the wafer surface is measured to control focus in real time. Since this space will be filled with water in immersion scanners, the focus-control design needed to be reviewed.
After careful studies and experimentation, it was found that optical, real-time focus techniques were still the best method for focus control in an immersion system. This method has the advantage of detecting the focus condition even if some disturbance occurred in the water-filled space between the lens and wafer.
In this configuration, focus-sensing beams pass through the water. We have a pair of optical windows at the water boundary; inside the windows, space is filled with water, while it is completely dry outside the windows.
 Figure 3. The focus beam reflectivity at a) the air esist surface and b) the water esist surface are the same when the angle is slightly adjusted. |
With immersion lithography, reflectivity of the focus beam from the resist surface (resist to water interface) is a concern. The reflectivity at the water-resist interface is generally lower than at an air-resist interface. Results of our analysis are shown in Fig. 3. Figure 3a shows the incidence angle dependence of reflectivity in the case of a dry configuration. In this case, reflectivity of ~55% at the resist surface is obtained by setting a high incidence angle. For the water-resist case (Fig. 3b), we found that almost the same reflectivity is possible by slightly increasing the incidence angle. The accuracy of the focus measurement mainly depends on this reflectivity; therefore, we can expect the same high level of accuracy and real-time focus-control performance with an immersion system as on advanced dry scanners.
Water temperature and bubbles
Water supply is another essential element for an immersion exposure tool. There are two important requirements for the water: temperature stability and bubble/particle suppression [3, 4]. A water-supply design being developed for the 0.85NA immersion system, planned for completion this year, has demonstrated temperature control and stability within ±0.01K when room temperature is approximately 23°C. Therefore, the required stability has already been satisfied.
Recent results show an improvement in the reduction of bubble (or particle) count to a target value that is more than sufficient to satisfy requirements for full-field exposure tools. In one water-supply system, a maximum bubble/particle count was recorded at six in a 10ml water sample. This was improved to three using a second design. The average bubble count was lowered to 0.024/cm3 for the output water in the second water-supply system. This bubble/particle suppression level is more than sufficient to satisfy the requirements for full-field exposure tools.
Solving polarization effects
Another area of concern is problems caused by polarization of light. Ultrahigh-NA systems — dry or wet — require polarization control in the illumination path to achieve optimal resolution. Conventional illumination uses randomly polarized light, in which two possible polarized lights are mixed (TE and TM polarization). The TE-polarized (or s-polarized) light, with an electric vector parallel to the pattern direction, makes high-contrast images, independent of the incidence angle of light to the wafer (Fig. 4). However, TM-polarized (or p-polarized) light, with an electric vector perpendicular to the pattern direction, creates lower-contrast images. When the incidence angle to the wafer increases, the TM-polarized light focused onto the wafer moves in different directions, resulting in a low-contrast interference. This image-contrast degradation by TM polarization becomes significant when the projection-lens NA>1.0.
 Figure 4. a) TE and b) TM polarization conditions for high-NA imaging. Imaging contrast by TM polarization degrades when NA approaches 1.0. |
To overcome this polarization effect, control is required to enhance TE-polarized light and to suppress TM-polarized light. Immersion exposure systems will have ultrahigh-NA systems ≥1.0, and polarization control will be a necessity. To gain control, a polarizer could simply be inserted in the illumination path. That solution reduces the illumination power, however, which lowers system throughput. By using a proprietary optics design, Nikon was able to create a polarization-control system without illumination power loss, and thus, without impact on throughput.
Polarization control is effective with the combination of off-axis illumination. One example is a combination of 3/4 annular illumination with azimuthal polarization (Fig. 5). With this illuminator, we can expect 65nm line and spacing (L/S) imaging with an attenuated phase-shifting mask on an immersion system with NA = 1.00. With this illumination setting, mask patterns of any direction are acceptable.
Another example is a combination of dipole with linear polarization (Fig. 6), which shows simulation results for L/S patterns. An advantage can be seen for all three NA values, in terms of both better resolution and DOF.
Resist test results
Since early 2003, experiments have been conducted [3, 4] with various photoresist manufacturers to evaluate the performance of the current 193nm resist formulations. The results show that no significant change to the resist is required, and there seem to be no fundamental limitations to the resist materials for immersion applications.
 Figure 7. SEM pictures show 193nm immersion exposure results with half-pitches of a) 65nm, b) 50nm, and c) 45nm. |
Figure 7 shows SEM pictures for 65nm to 45nm half-pitch L/S structures. These 193nm immersion results, conducted with Tokyo Ohka Kogyo Co. Ltd. (TOK), were also validated in collaborative evaluations of resists with suppliers JSR Micro Inc. and Clariant International Ltd.
What's next
The studies conducted with TEL proved feasibility for immersion at the end of 2003. Nikon has now moved to full product development of an immersion system, a full-field exposure tool with NA = 0.85 (completion planned for 3Q04). This system will be based on the company's S307E, which is a dry 193nm scanner, and will be used as an engineering evaluation tool (EET). Although the NA will be the same as the dry ArF system and the resolution will not change dramatically, we can expect improvements in the DOF. Chipmakers will be able to use this tool for evaluating immersion exposures and developing immersion processes. We also intend to continue our collaboration with resist suppliers on this tool.
Nikon also plans to ship production systems with NA≥1.0 in 2H05. The production model will use a standard 4× reduction ratio and a 26×33mm field size. An NA≥1.0 will be required for production systems to be used for applications below 65nm. As a next step, Nikon plans to increase the NA to ≥1.20 to meet production requirements for 45nm devices.
Conclusion
Based upon engineering results, Nikon is developing an ArF immersion system for volume production. Although some engineering work is still necessary, we have moved to full product development with an EET planned for this year, and a production system with NA≥1.0 planned for next year. As always, we should be careful about any new technology, where one small item may have significant impact on a program. There are areas requiring further investigation, such as hydrodynamic vibration in the nozzle, bottom lens protection, defects caused by nanobubbles, and chemical contamination in the water. Our decision is to construct the full-field immersion exposure tool and investigate the remaining issues during system development.
Acknowledgments
The authors would like to thank TOK, JSR, and Clariant for collaboration on the resist evaluation experiments, as well as their Nikon and TEL colleagues, who are engaging in the immersion exposure-tool study and development program.
References
- M. Switkes, M. Rothschild, J. Vac. Sci. Technol. B 19, 2353, 2001.
- B.J. Lin, Proc SPIE 4688, 11, 2002.
- S. Owa, H. Nagasaka, Proc SPIE 5040, 724, 2003.
- S. Owa, H. Nagasaka, J. Microlithography, Microfabrication and Microsystems 3 (1), 97, 2004.
Soichi Owa received his PhD in physics from the U. of Tokyo, and is a manager in the 1st development department at the development headquarters of Nikon Corp.'s Precision Equipment Co., 201-9, Oaza-Miizugahara, Kumagaya-city, Saitama 360-8559 Japan; ph 81/48-533-4918, fax 81/48-533-7458, e-mail [email protected].
Hiroyuki Nagasaka received his MS in physics from the U. of Tokyo, and is an engineer in the 1st development department of Nikon's Precision Equipment Co.
Yuuki Ishii received his MS in physics from the Tokyo U. of Science, and is a staff manager in the 1st development department of Nikon's Precision Equipment Co.
Osamu Hirakawa received his BS in mechanical engineering from Kyushu Institute of Technology, and is director of development and technology in the clean track business unit of Tokyo Electron.
Taro Yamamoto received his MS in mechanical engineering from Kumamoto U., and is an engineer in the development and technology operation of the clean track business unit at Tokyo Electron.