Semi meeting: slow 300-mm plans
04/01/1998
Semi meeting: chipmakers slow 300-mm plans
IBM announced that it is backing off its 300-mm schedule, and the International 300mm Initiative (I300I) will accelerate equipment demonstrations in 1998, after a slower-than-hoped pace in 1997. The announcement came at a SEMI-sponsored meeting in Burlington, MA.
IBM waits for others.
In a sharp change from its role as pioneer of 200-mm wafer development, IBM "doesn`t want to be first, or even second" in the move to 300 mm, and has backed off from an 18-nm (0.18 ?m) development plan announced last November, noted Tim Ravey, manager of the IBM Semiconductor Research and Development Center`s 300-mm Project. IBM now plans to begin working on 300-mm substrates for DRAMs at 15 nm, and for logic at 13 nm, but only "after others have debugged the integrated equipment set and factory interfaces," he said.
Ravey said the main reason for the slowdown in IBM`s schedule is that "the equipment simply is not ready. The manufacturers have done a great job so far, but there`s a long way to go. Our leading equipment suppliers are struggling with their investment position." He explained that IBM wants tools capable of producing 15-nm geometries on delivery, with extendibility to 13 nm and beyond. Compliance with standards and guidelines is mandatory, he added, and there must be "substantial progress toward meeting productivity goals."
He noted that "some film units, both in deposition and etch, are actually coming out better than the 200-mm tools," which he attributed to more sophisticated modeling techniques, as well as a ground-up redesign of systems for 300 mm.
"Lithography is the exception," continued Ravey. Although one supplier, Canon, has shipped a handful of 300-mm exposure tools, "they`re not 18 nm, and certainly not 13 nm; they`re more like 22 nm, and they`re steppers," he said. "The only company that has been in the scanning business for any length of time, SVG Lithography, can`t sell a 300 mm tool." He suggested that the transition to scanning lithography has been "more difficult than expected" for Nikon, ASM Litho, and Canon, and said the four litho developers "are really struggling."
Ravey acknowledged that the oft-discussed 1.3? tool cost multiplier for equivalent output at 300 mm compared to 200 mm will probably not be met in the first round of tools, but pointed out that some 300-mm equipment is actually coming in lower because of faster output. He said the ratio of tool footprint to output should be equivalent to the 200-mm generation, and that availability should be 90%.
At the SEMI meeting, Kristen Robak, 300-mm product manager for Eaton`s Semiconductor Equipment Operations, acknowledged that "Eaton doesn`t expect its 300-mm implanters to meet the 1.3? target," but that customer-dictated multiple-generation capabilities and automation requirements are contributors. Moreover, she said, in a comparison of high-energy implanters for a DRAM process scenario, the new line of tools will still provide a 30% die cost advantage over 200-mm tools even at 1.6? the cost of 200 mm.
Ravey cited another area for improvement - monitor and test wafer usage. IBM has a goal of using only one-fourth the current number of test wafers, and Ravey said the International 300mm Initiative "needs to do some work to get people focused on that." While there has been talk about using in situ sensors to eliminate monitor wafers, Ravey said, "I don`t see that happening. There`s not much in-house expertise at the equipment companies."
Responding to audience questions about single-wafer processing at 300 mm, Ravey said he saw no sign of batch wet processes being displaced by single-wafer spray processes at 300 mm, although a move to smaller wet benches is likely. "We`re moving away from the 30- or 40-ft decks to units with smaller footprints and lower throughput, to provide self-sufficient areas. We`ll have a cleaner in lithography, one in etch, and so on." Also, he added, while there may be adoption of single-wafer thermal processing tools at 300 mm, "for the most part, [batch] furnaces will still be around."
I300I won`t finish tool tests in 1998.
For the I300I, 1998 will be a time of accelerating equipment demonstrations from a slower-than-hoped pace in 1997. By the end of the year, says VP Frank Robertson, 90% of a full 300-mm fab equipment set will have been evaluated, with the remaining tools slated to be available and demonstrated early in 1999.
In part because of the equipment evaluation situation, chipmakers have in recent weeks substantially backed off on their plans for 300-mm adoption in 1998; DRAM overcapacity and the Asian financial turmoil have also contributed to the shift. George Lee, director of SEMI`s 300-mm Initiative, told the audience at the Massachusetts meeting that he now expects to see only one or two pilot lines set up this year, down from a prediction of five a month ago. Moreover, 1999 is now seen bringing just seven pilot lines instead of the 13 forecast in January, and broad full-flow manufacturing is not expected to begin until 2001.
"There is really only one project that can step forward and say, `this is a solid go`," said Lee, referring to the Siemens-Motorola line in Dresden, Germany, which has backing from the German government. He added, "I think the forecast of seven pilot lines for 1999 will hold through the first half of this year. There will be better visibility then. A lot will depend on the Asian financial crisis. The financial stability of South Korean manufacturers will have an impact on orders."
Meanwhile, equipment makers are trying to balance the ever-changing market picture with the drive to get tools out and evaluated, and in some cases shipment schedules have slowed. "Certainly, we were not able to do as many demos in 1997 as we wanted - we did all that could be done," said Robertson. Selete, I300I`s counterpart in Japan, has also had difficulty getting evaluation units from toolmakers. This slippage has resulted in some straining of resources at I300I. "There are a lot more suppliers knocking on our door than we have resources to do demos for," said Robertson.
In a shift from last year`s policy, all tools accepted for demonstration this year must be "beta-level," essentially the same machine as would go into a manufacturing facility after refinement. Tools must include integrated front-opening unified pod (FOUP) handlers, E-15 compatible load ports, compliance with SEMI safety standards, and GEM-compliant software ready to interface with CIM systems. In addition, says Robertson, "the technology target is now firmly at 18 nm, up from 25 nm last year."
Lithography is the biggest laggard, with demonstration systems not seen arriving until 1999. There are also a number of areas where some processes are in place but a full complement of tools is not ready, added Robertson. These include ion implant, metal CMP, and patterned wafer metrology equipment. - P.N.D.