INDUSTRY NEWS

Suppliers ponder whether chip technology map is road to ruin

On November 22, 1999, SIA published the latest version of The International Technology Roadmap for Semiconductors (ITRS). The next day executives from equipment and materials suppliers across the world gazed slack jawed at the document and asked themselves one question: "How are we gonna pay for this stuff?"

SEMI is in the process of trying to answer that question. The trade association, which represents more than 2300 tool and materials vendors, is heading up a global effort to assess the dent members will take in their wallets while trying to meet the R&D requirements set forth in the document. SEMI, along with its Japanese and European counterparts as well as SEMI/ Sematech, announced at Semicon Taiwan in September that they will conduct "a coordinated global assessment of the economic implications of the increased research and development" needs imposed by the ITRS.

The organizations have met twice in executive forums, with the next forum scheduled to meet this month. A meeting set for April will bring together executives from International Sematech and SEMI/Sematech, recently renamed the Semiconductor Industry Suppliers Association (SISA).

"We're attempting to develop a process to identify crucial technology areas in which the R&D costs make the return on investment prohibitive for a company or a group of companies," points out Stanley Myers, president of SEMI. Myers tapped James Greed, a member of SEMI's board of directors who retired recently as president at VLSI Standards, to act as coordinating director of the assessment.

Dave Anderson, director of supplier relations for International Sematech, says the first executive forum in June 1999 took place when the consortium's board of directors invited the presidents of the top global equipment manufacturers to discuss the upcoming challenges. At that meeting, he recalls, "Some of the suppliers said, 'If you really want this to be an executive forum, why not invite some of the other device makers?' Which we did--we invited to the second forum the top IC manufacturers worldwide. Micron, Samsung, and Chartered were invited, in addition to the Sematech member companies. The Japanese companies politely declined and said, 'Maybe next time.' We're trying to truly take it beyond the walls of Sematech."

Part of Sematech's efforts, Anderson says, involves "what we term 'global economic modeling' of the industry." The goal is to determine the "feasibility from an economic standpoint of achieving the roadmap. Do we have enough money to achieve what we want to achieve?"

The primary reason for the concerted study is the "acceleration of the roadmap," Myers says. "The fact that we also accelerated very rapidly on 300-mm development, and there's no business there at this stage" is a mighty factor as well. "You've got a number of our members who have spent a significant amount on R&D that they haven't been able to fundamentally pay off because there are no sales. As the roadmap accelerates, you're going to get more and more of that."

Photoresist manufacturers in particular are suffering because of the warp-speed nature of the ITRS timetable, Myers notes. It may take five to seven years to pay off the investment required to manufacture the future photoresists needed for one product generation. "But if you look at the roadmap, you're going to need three other generations in about the same time frame. It's those kinds of things we're looking for."

The ability of silicon houses to make specific types of 300-mm wafers is a particular concern within the overall push to the larger substrate, Myers indicates. "Those guys are sucking gas right now in that industry." He also cites chipmakers' needs for new process metals and their concerns related to etch processing. "I would say there's a half dozen things we would have to zero in on."

Anderson and Myers both cite photolithography as one area of intense industry focus. The other is metrology. With the former "there's more than one technology to look at. Is it going to be Scalpel? Is it going to be x-ray?" notes Myers, citing just two of the alternatives to optical steppers being bruited about in the industry.

"Generally speaking, we may need a new technology at each node in order to do that development," says Anderson. He added that theoretically the suppliers could increase the amount of developmental money they spend over that period. Doing that, however, "also shortens the return on the sales of the equipment."

"You know from past history that some of the areas in lithography don't have a reasonable business model," muses Paul Peercy, who recently stepped down as president of SEMI/Sematech to become dean of the college of engineering at the University of Wisconsin in Madison. Now serving as a consultant for the industry assessment, Peercy cites photomasks and mask repair as two areas that "take a significant amount of R&D investment, and the number of tools is just not that large."

The cash crunch is similar for interconnect technologies, Anderson asserts, as the industry "moves forward with different materials, whether they be low-k or high-k dielectrics, for example. It may require different deposition techniques or processes or tools at each new technology node."

"Is there enough money available to do the R&D in the various different areas?" asks Peercy. "It's interesting. I'm not convinced that we have proved that the industry generates enough revenue to sustain this technology acceleration indefinitely."

Myers says manufacturers of capital equipment spend about 15 to 20% of their revenues on research and development. For materials suppliers that figure is approximately 7%. The trade association president notes that chipmakers are projected to sell $250 billion worth of products by 2002 or 2003. "Of that total industry revenue x percent is going to be spent on materials, y percent is going to be spent on capital equipment, and z percent, say, is going to be spent on bricks and mortar. And when you get all of that done, there's going to be some number that says this is what the industry has got to spend on the technologies in that time frame. With that number we're going to be able to see where this equipment does not balance."

"Some percentage of that total is going to have to be spent to meet the new technology requirements of the roadmap," he continues. "The question is: What have you got to spend on your R&D in the end? We think that's going to be limited, and so we've got to be able to zero in on the technologies and pick the right ones in the future."

"Suppliers will continue to meet the needs of the industry," Peercy concurs. "The question we're asking is a longer-term question. That is, how do we include some of the economics in the roadmapping process so that perhaps it will guide the decisions earlier on which technology to back, which would then save money for the industry overall."

The former SEMI/Sematech president and Anderson agree about the particular pressures put on metrology tool providers. "Metrology tools are required for the very 'bleeding-edge' technologies," Anderson points out. "The metrology needs for R&D come in the very early stages of any given technology node, but the return on investment comes much later when that node comes into volume production. [The suppliers] have a particular problem in that their development cycle is very early in the process."

Warns Peercy: "Metrology in general across all of the roadmap areas is going to be a big issue." Spotting defects and deciding which ones are killers is another issue, he adds.

Is it possible that by the April meeting the industry experts will find some of these issues to be less dismaying than they initially appear? "It's very likely that that could happen, that we see some of the things are less daunting than originally thought," Anderson replies. "That generally happens in our industry. We find ways to make things work. What is most daunting, though, is that we are reaching the limits of device physics. Because of that it becomes a much more imposing barrier to us" than just scaling.

Asked whether he thinks some sort of cost sharing might be the solution, Anderson paused. "That's an awkward question," he answers amiably. "We haven't arrived at that point in the discussion yet."

Myers calls the discussions so far "very preliminary," adding colorfully that the process isn't "a building's-on-fire-and-the-kids-are-jumping-out-of-the-window type of thing. It depends on how fast we can move. Certainly the April meeting with Sematech will present a supplier's position on what we see as [workable]." He says once a process is in place it will be ongoing. "I would expect it to be an evergreen."

The overall objective, Myers says, is for both users and producers to try to continue down the cost-reduction path "on a 20-to-25, maybe 30%, learning curve. That's what we've done in the past as a subset of Moore's Law," making even more powerful and cheaper devices for all types of end products.

A sympathetic Anderson notes that suppliers have faced increasing pressures as they have taken on more financial and technical responsibility. "In the '70s equipment companies were guys who made vacuum chambers. In the '80s it was process development, and in the '90s it's process integration. More and more developmental responsibility has fallen on the suppliers' shoulders. That's the reason for bringing us together here. We can't go forward without each other."