by Ed Korczynski, senior technical editor, Solid State Technology
During SEMICON West 2006, a group of the industry’s top senior technologists came together to discuss the limits and possibilities for semiconductor material development. The technical seminar and panel discussion, sponsored by DuPont Semiconductor Materials, and moderated by this author, addressed material issues in the major International Technology Roadmap for Semiconductors (ITRS) initiatives.
In last week’s WaferNEWS, Part One of this report discussed the atomic limits of semiconductors materials engineering. But dealing with atomic layers, and delivering very small amounts of material, may also require fundamental changes in business models. Materials development costs are expensive (basic materials research, as well as process integration into manufacturing lines), and if suppliers are still paid per gram of material when critical amounts delivered are now sub-gram, materials providers may need to somehow be subsidized.
Part of the problem is that suppliers haven’t had a long-term R&D focus beyond a few generations, according to Larry Thompson, president of Intellectual Property Services & Solutions LP (IPSS), and former director of advanced lithography and chemical engineering at Bell Labs. Equipment and materials suppliers have borne the cost of development for several decades, focusing on a few technology generations at a time, but very long-range research has relied on groups such as IBM, Bell Labs, and other laboratories. “I think the equipment and materials companies have never had to look at three or four generations out. They look, at most, one generation out,” he said.
Part of the problem is that the industry is now pushing for an accelerated development roadmap of ~18 months, noted Robert Havemann, VP of process integration, Novellus Systems. “Nobody has time to really look ahead down the road,” he said. “We’re just looking for what we need to do to get to the next node.” Places like Bell Labs could afford to dedicate their work toward longer-range visions and revolutionary technologies, he said. But suppliers dealing with shrinking margins in an era of commoditization find there’s not enough money to spend on long-range research projects. “It’s kind of a vicious cycle,” he said.
Another problem is the market’s push toward supply commoditization, observed Havemann. “IDMs want a level playing field, so they can go to supplier A or B or C, get exactly the same products and drive competition and price,” he said. The consortia formed from IDMs now dominate our industry, and “their choices have huge effects on the market,” he said. “If your material wins, then you may have the world market or the majority, while if you lose then maybe you go out of business.” Thompson invoked the painful reminder of 157nm lithography as an example of the potential consequences of pursuing such a business model at the behest — and whims — of IDM customers. “They spent hundreds of millions, if not a billion dollars, between the materials for the optics and the machine,” he said. “To have it in a two-month period jerked out from under them is a totally unrealistic way to do business. If that happens again, you may not have even a commodity lithography tool company left. So, don’t eat your own lunch as an industry.”
So how can suppliers convince IDM customers that their complex engineered substrates with atomic-level precision are anything but commodities? Getting engaged early with a particular consortium to develop critical materials that can be integrated into the production line is one avenue, negotiating deals on IP exclusivity for materials, processes, or a new tool concept, pointed out Havemann. But eventually the same tools and processes will be used by chipmakers worldwide, often in a matter of months. “You can’t run a business just with one customer, even with one of the largest semiconductor companies in the world. So, you’re going to have to sell to more than one customer, and when it hits the worldwide market this commoditization starts to occur,” he said.
Raj Jammy, director for front end processes at SEMATECH, suggested that chipmakers have a better vision of what the problem set is than materials and tools suppliers, although toolmakers are participating a little more nowadays (citing his experience as a SEMATECH assignee from IBM). “Today an IDM would typically say, ‘Oh, you have a tool that does ALD high-k? Well, have you checked the electrical properties, and what is the history of the CV?'” Havemann agreed that equipment suppliers are getting involved earlier in the development process with consortia, supplying data that used to be provided by customers in early phases of development. — E.K.
[Editor’s Note: Multimedia presentations from the DuPont-sponsored panel are available online — click here.