ISMI News

Willingness to “Own” Manufacturing Problems Can Benefit Materials Suppliers, International SEMATECH Expert Tells Industry Meeting at SEMICON West

San Francisco, CA (15 July 2004) – Semiconductor materials suppliers will need to ensure that the plethora of emerging process substances works seamlessly for IC manufacturers, an International SEMATECH executive told a suppliers’ conference here Tuesday.

Dr. Ken Monnig, associate director of SEMATECH’s Interconnect Division and an acknowledged interconnect expert, told the Chemical & Gas Manufacturer's Group (CGMG) that manufacturers are facing a dizzying array of new materials after 35 years of working essentially with silicon, oxygen, nitrogen and aluminum.

Consequently, the potential for processing problems has greatly increased, and IC producers are in need of quick resolutions, he said.

“The supplier that owns the problem in the field will keep the customer loyal,” Monnig said. “In other words, if a materials supplier works with the tool supplier to immediately source and correct the problem as it occurs in the factory, that supplier will become recognized as the ‘go-to’ authority. The manufacturer will not be left trying to guess whether the issue belongs to the tool, the material, or both.”

Monnig suggested that in this decade, which will continue to introduce low-k, high-k, substrate variations at unprecedented levels, materials and tool suppliers may want to form formal collaborations to ensure that chip-makers receive an integrated solution.

“The cost of keeping a near-future semiconductor fab idle for one hour is approaching $100,000,” Monnig noted. “For IC manufacturers, cost and risk are the first and second decision drivers in dealing with materials.”

The industry’s emphasis on new materials is less than 10 years old, Monnig said, and at sharp contrast to the recent past in which IC makers had only four materials to worry about. In that period, process materials became commodities and device manufacturers sought the lowest-cost providers.

That began to change in 1997, however, when IBM introduced copper interconnects along with tantalum and tantalum nitride as barriers. The drive to low-k and high-k brought a new family of materials, and the industry began transforming from a commodities field to the customized business of today.

Besides new approaches to problem-solving, the semiconductor business also faces shortened product development cycles, Monnig continued. “It typically takes 10 years to introduce a new material into manufacturing. But the ITRS says that a new material is needed practically every technology generation, which occurs every two to three years.

“This is another issue that might be successfully addressed by R&D collaboration,” he said.

In a related meeting moderated by Monnig on Monday, a panel of specialists agreed that the materials challenges inherent in moving to the 65 nm technology node are some of the most sobering the chip industry has yet faced.

“Before, we were using industrial chemicals and purifying them; now, we’re inventing all new molecules (that) take real time and real investment dollars,” said one panelist representing a chemical company. “It takes most of a decade to develop these new materials… and the new question has become: ‘Is it economically viable?’”

Other viewpoints included:

  • The transition from 90 nm to 65 nm will mark the end of traditional CMOS scaling, requiring manufacturers to play “more and more tricks as we put more and more stress on the fabs.”
  • Next-generation, low-k materials at 2.3 dielectric constant, with k-effective around 2.6, will be available at the 65 nm node, but probably no new, high-k gate stack materials.
  • The 65 nm node will be on schedule, a process that will engender both delays and success stories.

The experts appeared at a SEMICON West forum, “The Challenges of Migrating to 65 nm,” hosted by DuPont IC Fab Materials. Several speakers cited collaboration among materials suppliers and IC manufacturers as key to the cost-effective introduction of 65 nm.