Although it has been stated for many years, it would appear that optical lithography for semiconductor manufacturing is reaching the end of physical limits in its ability to continually scale down device sizes. Nigel Farrar, VP of Lithography Applications and David Brandt Senior Director of EUV Product Planning for laser and light source company, Cymer answer questions from EuroAsia Semiconductor on the future of lithography
Q. Lithography has always been a key driver in semiconductor manufacturing but the optical variety is now reaching the natural limits leading to a greater dependence on every aspect of the lithographic process. How true has this been for Cymer?
A. Cymer has provided the leading edge light source technology to allow the industry transitions to KrF and ArF wavelengths and is now also showing industry leading performance in EUV source technology with a 100W Laser Produced Plasma (LPP). Although high index immersion extensions are uncertain at this time, Cymer technology developments will also support this approach if the industry chooses to follow this direction. Cymer has also supported the continued demand for tighter process control with technologies to provide tighter control over key light source parameters that are directly correlated to improved control of wafer process performance
Q. Finding a new type of lithography to replace ArF Immersion has proved difficult to develop to acceptable production quality. EUVL appears to be the most likely replacement. When do you think this transition will occur and what are the major challenges?
A. EUVL technology is currently targeted for insertion at the 22nm process node, since it appears that it will not be ready in time for the 32nm node. Pre-production scanners are expected to be ready for delivery in late 2009/early 2010. The industry has ranked the leading challenges as source power and lifetime, defect free masks and resist resolution and sensitivity. Continued development in tools, infrastructure and process development are all needed.
Q. How important is a stable light source?
A. A reliable and long lifetime high power source is currently the top challenge for EUVL. Cymer is pursuing Laser Produced Plasma (LPP) technology and this is the best technology for scaling power for future EUVL requirements.
Q. The semiconductor manufacturing industry is seeking an extension on immersion lithography as EUV will not be ready for the next technology node. Double patterning seems a likely candidate at present to continue device scaling. What are the significant challenges for double patterning?
A. The key challenges for Double Patterning are higher process costs, greater design complexity and tighter process budgets for CD and overlay.
Q. What are the specific challenges for light source manufacturers?
A. Light source technology can support the needs of double patterning in two main areas. One is higher power and higher availability to support improved system productivity and reduce process costs. For example, Cymer's GLX (Gas Lifetime eXtension) technology significantly improves availability by reducing the downtime associated with gas refills. The other area is greater source operating stability to reduce the light source contributions to the shrinking process budget. For example, Cymer's ABS (Advanced Bandwidth Stabilisation) technology and XLR laser pulse energy stability both reduce potential sources of CD variation due to the light source.
Q. Double patterning brings its own unique challenges and implications to the imaging process. What are the major implications and please expand on the challenges issues such as overlay and CD control?
A. For double patterning, final CD control is influenced by new contributions and interactions between process variables. The process design must be optimized for the particular product design and process flow. For example, CD control in some Double Patterning schemes is determined by the CD control for both imaging step plus the overlay between them. CD control can now also be influenced by process variables outside the lithography cell, related to film deposition and etch, which require additional control.
Q. What methods can be employed to maintain litho-cell productivity and improve process control?
A. There are many different varieties of Double Patterning and there is a lot of ongoing work to develop process integration schemes with improved cycle time and productivity. Similarly, the different schemes have different trade-offs between the variables that affect process control and these need to be optimised for particular product designs and productivity goals.
Q. Modelling software is advancing to meet the extreme needs of advanced processes such as Optical Proximity Control. What developments have occurred in this area to improve accuracy when process control is so vital?
A. As CD control budgets become tighter, many of the approximations that have been used to enable high speed simulations have to be reconsidered. This drives renewed attention to the trade off between speed and accuracy. Examples include the need to model mask topography, rather than using a thin mask approximation, using detailed illumination profiles, and including laser spectral bandwidth in imaging calculations. Cymer has led developments in modeling the effects of laser bandwidth on imaging and OPC and has recently described new techniques to simultaneously accelerate simulation speed and improve accuracy. Speed becomes increasingly important as 2D simulation and large area simulation is required.
Q. Do you have any guidance on when ArF Immersion will likely be replaced by EUV lithography? If at all?
A. Pre-production scanners are targeted to be available in later 2009/ early 2010, which is on track for insertion of EUV for 22nm node production.
Q. Will cost become a potential stumbling block to lithographic advancement?
A. Advanced lithography technology must continue to be cost effective and this has typically been achieved by continual improvements in productivity. High source power is a particularly important requirement for EUVL because it drives the high throughput that will make the technology cost effective compared to alternative solutions.