Updated: Are Designers Conscious of Yield?

Laura Peters, Lead Technical Editor -- Semiconductor International, 8/3/2007

Design for manufacturing (DFM) means design for money and design for profitability. At the 90 and 65 nm nodes, yield predictability can be 90%. “Predicting yield is our economic model for the foundry,” said Dan Armbrust, vice president of 300 mm semiconductor operations for IBM’s Systems and Technology Group (East Fishkill, N.Y.). Armbrust was one of three panelists discussing the state of DFM at a yield panel at SEMICON West. He was joined by Anantha Sethuraman, vice president of DFM solutions at Synopsys (Mountain View, Calif.), and Chris Progler, CTO of Photronics (Brookfield, Conn.).

Listen to the Yield Panel (Runtime: 57:15)

Some of the most interesting findings reported by the panelists had to do with the status of 90 and 65 nm designs. “65 nm tapeouts are getting stronger, even though there is a lot of new learning happening at 65,” Sethuraman said. Part of the reason DFM has advanced as quickly as it has is the result of manufacturing and design people truly getting together to solve the problems. “If you go to a typical DFM meeting today, 75% of the people in the room are manufacturing people,” he said.

But there are still serious DFM issues. For instance, knowing whether a defect will truly impact yield is quite a challenge. Tying defects to actual electrical problems on the chip is difficult. Of course, a good percentage of defects will not impact yield, while others will. This varies from one chip to another. “From a maskmaker’s perspective, our studies show that 10-20% of defects will not impact yield, but this will not be true for another device,” Progler said.

Armbrust also talked about where some new technologies might come into play. For instance, he said double patterning will happen at the 32 nm node. He also expressed confidence that immersion lithography was coming along well, and is already being used by IBM. However, he hinted that additional investment may be needed to make double patterning a production-worthy technology. It has to happen because extreme ultraviolet (EUV) lithography will not be ready in time.

Armbrust emphasized that the power issues at 32 nm mean that you can shrink to 32 nm, but the performance simply is not there without excellent power management. “At 32 nm, you’re almost going backwards in terms of performance, so we’ll see some important developments here.”

Sethuraman addressed the predictability of yield. Because a manufacturer has “n-1” manufacturing data, this can be used to predict yield with a high accuracy. “But accuracy is a moving target,” he said.

Progler talked about the fact that the mask is the first time a design makes its transition from software to hardware. “You can measure many things on these first masks, and I don’t think this ability is leveraged as heavily as it could be,” he stated. At the same time, he said that the most successful designs today are more memory-like. By altering the design to give it more regularity, the chance of successful yield is going to be higher. The most extreme example of this is NAND flash memory, he said, which is very regular in structure. New developments in the memory space, Progler said, are likely in the next several years.

Talking DFM were (left to right) moderator Ron Wilson of EDN, Dan Armbrust of IBM, Anantha Sethuraman of Synopsys, Chris Progler of Photronics and Peter Singer of Semiconductor International.

Asked to give a 10- or even a 50-year outlook for the industry, Progler stated that tech companies do not have a good track record of predicting the future with the incredible pace at which technology moves today. “Our fathers didn’t know about the Internet,” Sethuraman said. The battery and power issues clearly need to be addressed for mobile applications. “Otherwise, you have a time to recharge issue,” he added. People have already talked about this issue with the new iPhone.