AUSTIN, Texas The competition to develop a next-generation lithography solution drew nearly dead even this week, as two factions of the scanning e-beam camp and one high-profile backer of a competing extreme-ultraviolet (EUV) scheme floated announcements of broadened support.
International Sematech members who gathered here this week for a scheduled meeting reserved judgment on the dueling announcements, stating that the consortium will continue to evaluate the competing approaches but that it remains concerned about "the technical risks and costs for all of the next-generation lithography technologies at and below the 70-nanometer node."
Earlier this week, Lucent Technologies said it had signed two development partners for the Scalpel scanning e-beam system under development at Bell Labs' Murray Hill, N.J., facility: Applied Materials Corp. and Netherlands-based ASM Lithography.
A day later, in Tokyo, equally powerful partners announced a variation on the theme. Using e-beam technology licensed from IBM Corp., lithography leader Nikon Corp. divulged plans to develop a scanning e-beam lithography system that will be ready for prototype use in 2003 and for the factory floor by 2006.
Separately, equipment maker Silicon Valley Group rolled out production versions of its latest optical lithography gear, along with a fresh road map that views EUV as the natural successor to today's optical systems.
Lucent's Scalpel technology and the IBM approach endorsed by Nikon both compete with the EUV technology, which is being developed at the EUV LLC consortium (Livermore, Calif.). While Scalpel differs in certain respects from the IBM approach, several sources said the possibility exists that the Scalpel and Nikon/IBM schemes could converge around a common standard that would make it possible to use the same masks in either type of machine.
In fact, the Nikon/IBM work could represent an intellectual-property front: a means of developing a patent-bargaining position that would bring Nikon into the Scalpel fold with IP of its own as a trump card.
As one source at Sematech noted, lithography experts had expressed concern the Scalpel approach essentially stood alone, with one primary sponsor, though last week's announcement remedied that situation. The EUV approach, meanwhile, has been actively pursued by such heavyweights as Advanced Micro Devices Inc., Intel Corp. and Motorola Inc. in addition to the Silicon Valley Group.
Credibility is important as Sematech attempts to narrow the next-generation lithography candidates to one. That move could come over the next year and would likely channel scarce resources into either Scalpel or EUV.
In early December, the 110 participants in a Sematech-sponsored Next Generation Lithography Workshop in Broadmoor, Colo., voted to encourage Sematech to concentrate its resources on EUV and Scalpel. The apparent momentum behind EUV, which got more votes than Scalpel, makes this week's announcements all the more critical.
Shortly after the Lucent and IBM/Nikon announcements, technology managers from the 14 member companies of International Sematech met in Austin, Texas. By midweek, the consortium's engineering steering committee issued this terse statement: "We are concerned about the technical risks and costs for all of the next-generation lithography technologies at and below the 70-nm node. International Sematech will focus its programs on the technical risks, the costs and the capabilities of EUV and Scalpel. In addition, International Sematech continues to support a study of 157-nm optical lithography, at or below 100 nm."
In Tokyo, Nikon Corp. president Shoichiro Yoshida made somewhat similar remarks in describing Nikon's technology road map. "It is already an accepted scenario that ArF comes after KrF," Yoshida said, referring to the 193-nm emitting argon fluoride excimer-laser systems that will supplant the 248-nm krypton fluoride-based scanners now in use. "After ArF, three technologies-157-nm F2 [fluoride excimer lasers], e-beam and EUV-are standing in the wings. Nikon has been working on optical steppers, but from now on we intend to work on e-beam as well."
An IBM-Nikon blend
Akikazu Tanimoto, a general manager at Nikon, said the company's e-beam scanner will combine IBM's fundamental e-beam technology with Nikon's stage and vacuum technology. Nikon will sell and support the systems worldwide, with beta tests expected in 2003 and volume production several years later.
Nikon also is engaged in a cooperative national research project into EUV technology that's sponsored by Japan's Ministry of International Trade and Industry. For the 157-nm optical-lithography generation, Nikon is developing mirrors and lens materials for the optical element, Tanimoto said.
Asked whether Nikon's e-beam system will be compatible with the Lucent scheme, Tanimoto said, "Our technology and Scalpel are similar in principle, but there are practical differences, such as the width of the electron-beam stripe, the deflection of the beam and the applied current."
A business-development manager at Lucent noted that very preliminary discussions toward some level of cohesion took place some time ago among Lucent, IBM and Nikon. Lucent's goal is to make it possible to take a mask out of a Scalpel system manufactured by Applied Materials, for example, and put it in to a Nikon system, much as reticles can be switched among optical steppers.
At present, three to ten 8-inch (200-mm) wafers can be exposed per hour by a direct-write e-beam system. At the 70-nm technology node, Tanimoto said, "we have a target of 40 wafers per hour for 200-mm and 20 wafers per hour for 300-mm,"or 12-inch, wafers.
Tanimoto said it's too early to tell whether 157-nm F2 lasers or e-beam scanners will be first to provide 100-nm capability to the market. "E-beam may come first, but there the difficulty is in the reticle technology. Once the technology is established, F2 will have better throughput than e-beam," he said.
At Bell Labs, Scalpel project leader Lloyd Harriott said his team is working to automate 8-inch-wafer handling and improve the stage alignment of the Scalpel prototype. That "proof of lithography" system will form the basis of the development work done by researchers from Bell Labs, Applied Materials and ASML.
E-beam sources pose special problems. Electron optics are at a more primitive stage than photon-based optics, and that makes it impossible to rely solely on optical-style focusing systems.
"Part of the development work will be to design a high-throughput Scalpel system over the next couple of years. We are still targeting a system that could be in use for 90- to 100-nm lithography; the main difference is the emphasis on throughput, stage performance, the size of the scan of the field, the current required in the beam and so on," Harriott said.
Progress in the mask area is most important, Harriott added. "Fundamentally, for logic, the main reason to use Scalpel is the masks. You don't need to use phase shifting to achieve good results."