On the opening day of the 2011 SPIE Advanced Lithography exhibition, semiconductor metrology and process optimization leader KLA-Tencor held a Lithography Users’ Forum to present information on upcoming challenges and current progress on lithography semiconductor manufacturing.
With IDM and foundry semiconductor capital expenditures predicted to rise significantly for 2011 over 2010, KLA-Tencor’s chief marketing officer Dr. Brian Trafas noted the wide range of patterning error sources faced by advanced lithography. Reticle errors, litho and etch cell errors, and wafer issues – especially flatness – present issues that require leading-edge solutions.
Citing six areas for specific attention, Trafas noted extreme UV, pitch splitting, source mask optimization, ebeam direct write, nano-imprint, and quadruple pitch-splitting as areas where accurate and effective metrology will enable high-volume manufacturing success.
For a foundry perspective, KLA-Tencor presented Dr. Burn Lin, the vice president of nanopatterning technology of TSMC and his perspectives on the scope and limit of Moore’s Law in lithography research and practice.
Dr. Lin said that after decades of pushing the envelope of die shrinks enabled through advancing lithography, the industry is coming to the widespread understanding that Moore’s Law will slow due to device limits, lithography limits, and economic limits.
For each approach and potential solution, said Lin, there are now several significant challenges to be overcome. For example, pitch-splitting faces hurdles of cost, design rule restrictions, processing complexity, and the requirement of overlay accuracy. EUV faces difficulty in laser power requirements, resist sensitivity, and mask flatness – the last of these requiring about an order of magnitude improvement over current commercially-available options.
An approach that shows promise – reflective ebeam lithography, or REBL – has its own issues, including throughput and the massive data processing that is required.
In the end, noted Lin, progress with be defined by a relatively straightforward lithography decision tree. If these new technologies can overcome their development and production issues, and if they can work in high-volume manufacturing, then TSMC will invest in them – IF the customers perceives enough incremental value to pay for the incremental cost. This incremental cost is rising significantly for advances in the 2X and 1X nodes, and now many challenges must be overcome to bring any of these approaches to market.
Lin’s presentation showed the need for ongoing and expensive lithography research to drive solutions at advanced geometries, and it also highlighted the need for – and the value of – advanced metrology and process control at these geometries, neatly supporting the importance of the work being done at KLA-Tencor today.