An uncertain future for EUVL
02/01/2004
Solid State Technology asked industry experts to comment on the future status of EUV lithography.
Consensus, collaboration, and checkpoints are needed
Kevin Kemp, EUV program manager, International Sematech, Austin, Texas
I am frequently asked, "What is the future of EUV lithography? Will it happen? When will it happen?" Of course, there are no easy answers.
Extreme ultraviolet lithography (EUVL) is certainly the exposure technology of choice to replace optical lithography for volume semiconductor manufacturing at feature sizes <~50nm. This is because its extremely short wavelength provides the potential for high resolution at moderate k1 values for multiple technology generations, and because the overall system concept — scanning exposure, projection optics, mask format, and resist technology — is quite similar to that used for current optical technologies.
Remarkable progress has been achieved over the past few years in key aspects of EUVL technology. The Engineering Test Stand (ETS) program funded by EUV-LLC successfully demonstrated full-field scanned imaging in 2001. Source power has been increased by a factor of 10; EUV mask blanks are now available from commercial suppliers; and exposure toolmakers have announced schedules for alpha, beta, and production tools.
Nevertheless, significant critical issues remain to be solved. Source power needs to be increased by another factor of 10, and resists with adequate resolution and sensitivity must be developed to achieve 100 wafers/hr throughput. Mask blanks must be defect-free, and methods must be found to keep masks free of particles without the benefit of a pellicle. The lifetime of source and optics components must be improved, and equipment, mask, and consumables costs must yield a competitive cost-of-ownership. And customers must be convinced to accept the technology, which means they should have enough confidence to start placing orders for tools.
So what is the future of EUV lithography? Or more specifically, what is it going to take to make EUV happen in time for it to be useful? There are a few things that need to happen.
First, there needs to be greater consensus throughout the industry on the lithography roadmap for the next few generations. The promise of immersion lithography has opened the possibility for extending optical lithography even further than what was previously thought possible. However, the current debate between 193nm immersion and 157nm lithography has brought into question the timing for all of the future nodes. Nevertheless, as one observer put it, "All current scenarios start with 193nm and end with EUV." So whatever we do in the interim, let's make sure that the funding, resources, and effort remain focused on making EUV happen on time.
I also believe we need a greater level of collaboration between the various entities engaged in EUV development worldwide. Certainly, there is need for competitive development; the very best ideas and ultimately the most successful technologies will emerge from individuals pursuing the rewards of success in the marketplace. However, there are approximately 13 organizations involved in source development and pursuing at least four significantly different technical approaches. We need to do everything we can to identify the most promising approaches and combine efforts to ensure that the problem of adequate source power is solved. Fortunately, under the leadership of the International EUV Initiative, the community is starting to respond to this challenge, and a greater sense of collaboration and focus is starting to emerge.
Finally, I think the EUV development community must focus its attention not only on achieving the end goal of EUV in production at the 32nm node, but also have a plan to demonstrate success at two or three intermediate checkpoints along the way. These checkpoints should be designed as integrated technology demonstrations incorporating exposure tools and the various infrastructure components (masks, resists, and metrology), and agreed upon in advance with a specific set of targeted performance criteria — for example, source power, tool throughput, resolution, mask defectivity, and critical component lifetime. Like the EUV-LLC demonstration of full-field scanning capability, these integrated demonstrations will do more to drive progress and instill user confidence in EUV technology than any conference paper, supplier schedule, or technology roadmap. Ultimately, these demonstrations will provide the decision-makers in the industry with the information they will need to decide the true future of EUV lithography.
 Kevin Kemp |
For more information, contact Kevin Kemp, International Sematech, 2706 Montopolis Dr., Austin, TX 78756; e-mail [email protected].
EUV: Will it be ready?
Peter J. Silverman, Intel Fellow, Intel Corp., Santa Clara, California
Many next-generation lithography technologies have been pursued over the past two decades. Some, like x-ray lithography and e-beam projection lithography, have faded away as a consensus developed that they would not meet industry needs. Others, like 157nm lithography and 193nm immersion lithography, are still in commercial development, and new technologies such as laser direct-write and imprint lithography are regularly proposed. The semiconductor industry needs timely introduction of affordable lithography tools with the resolution capability required to continue Moore's Law, but is extreme ultraviolet (EUV) the right technology to meet the industry's needs?
The resolution of an optical system is given by the Raleigh formula:
Resolution = k1 × λ/NA
For simple optical lithography with binary masks, no OPC and standard resists, k1 is ~0.80. Introduction of phase-shift masks (PSM), complex OPC, and specialized resists has enabled production lithography at k1≈0.40, but at a high cost. Introduction of complex OPC and PSMs has driven up the cost of a mask set by 5× from the 500nm generation to the 90nm generation. As k1 is driven lower by the need for more resolution, the cost of masks will continue to increase at ~40% (or more) per generation. At the current rate, a 32nm-node mask set will cost ~$2.0 million. The only way to reverse the trend of unaffordable masks is to reduce mask complexity.
Lithography at the 90nm node uses 193nm, 0.75NA exposure tools for high-resolution layers. Further increases in numerical aperture, resist capability, and more complex masks will allow extension of 193nm technology through the 45nm node. Three technologies are under consideration for the 32nm node: 193nm immersion, 157nm, and EUV. The table shows the resolution capability for each technology.
 |
EUV is the only technology capable of resolution below 40nm. Furthermore, the projected cost-of-ownership (CoO) of EUV is lower than either 193nm immersion or 157nm lithography. In addition, the resolution capability of EUV limits the need for complex masks. EUV mask sets may cost as much as 30% less than comparable optical masks.
Leading-edge companies will start development of 32nm node technology in late 2006. Can EUV be ready on time? The EUV-LLC program was initiated in the mid-1990s to reduce the risks of EUV commercialization. By April 2001, all components of EUV lithography were successfully demonstrated: resist, masks, and a prototype 0.1NA, 25 × 32mm field EUV exposure tool. Design and fabrication of the prototype exposure tool required solving many engineering challenges, and its operation provided valuable learning in the challenges of EUV lithography, showing that there are no hidden technical surprises in EUV lithography.
The two most significant remaining challenges for EUV lithography are development of a high-power (>100W) EUV source and development of low-defect mask blanks. A global program of academic research is under way to improve EUV sources. Research is being done on EUV conversion efficiency, source debris mitigation, and thermal management.
There are also active programs to develop commercial extreme ultraviolet sources at many companies, including all the major suppliers of lithography lasers. International Sematech, ASET, and the commercial mask-blank suppliers have implemented aggressive programs to improve EUV mask blanks.
With the industry-wide commitment to EUV and the high level of commercial EUV development activity, EUV will be ready for the 32nm node.
For more information, contact Peter J. Silverman at [email protected].