by Pete Singer, Editor-in-Chief, Solid State Technology

May 20, 2008 – Applied Materials is urging the semiconductor industry to rethink how it defines productivity. Taking a page out of the Toyota playbook, AMAT’s Iddo Hadar said the focus should not be on improving productivity but eliminating waste, including what he calls “redistributed” waste. Hadar, the company’s CTO, CMO, and strategy officer for foundation engineering at the silicon systems group, spoke this morning at The ConFab conference in Las Vegas.

The industry is well known for its use of scaling to effectively increase the functionality of a square area of silicon. Geometries have shrunk, yield has been improved, and more functions have been integrated into a single chip. “Productivity in the way an economist would look at it would be to take all the things that go into production compared to the output — that’s the productivity,” Hadar said. “In our industry, just like in other industries, there’s the pressure between maximizing the output and minimizing the waste in what it takes to produce it. Historically, we’ve working a lot on maximizing the output. The model seems to have been working very well. We were driving scaling, making sure we were getting as many transistors per piece of silicon as we could, we were trying to make sure we crank out as much silicon per hour as we could, we push those transistors to work fast and improve the yield.”

From a waste perspective, this all means the waste of silicon has been greatly reduced. But the problem, said Hadar, is that this has been achieved by redistributing waste into other areas. “The two areas where the industry has generated a lot of waste was in the area of cycle time as well in the use of natural resources,” Hadar said. “We feel that these are the two main levers that the next generation factory needs to focus on to see that the waste is being controlled and managed. The idea is we look at the waste that is being generated in this area and find the best way to eradicate this waste.”

Ways to reduce such waste, he said, is eliminating several factors: time wafers spend waiting without action, time wafers spend waiting for other wafers to finish processing, processes that are added and then subsequently removed, and inconsistent wafer-to-wafer process results. “A wafer spends most of the time in the fab sitting and waiting. No process, no metrology, no motion.. just sitting and waiting. There are elements of this that are driven by equipment variability, significant elements driven by the impact of batch processing throughout the process flow, and there’s a significant impact of large lot sizes,” Hadar said.

In terms of process equipment, waste can be reduced by eliminating equipment idle time, equipment variability, and the gap between facilities/consumables needed for wafer processing and actual resources.

Hadar also touched on the need for sustainability, which he equated to energy waste. Contributors to energy waste include sub-optimized tools, inefficient fab architectures and the use of fossil fuel. “Theoretically, the carbon footprint of a fab could be an order of magnitude smaller than it is today,” he said.

Eliminating waste will require a coordinated industry effort, including a roadmap and prioritizing agility and sustainability, “just like scaling,” Hadar said. Agility will come in the form of small lot manufacturing, better equipment predictability (i.e., low variability) and universal single-wafer processing.

Right now the industry is being held hostage by “mental models,” when what we need to do is redefine productivity and recognize that extrapolating past methods won’t help tackle future challenges. Multiple sources of waste in the fab are beyond the classical focus on scale-down and scale-up, Hadar noted, whereas a coordinated move along all of them could eradicate waste in the next generation fab. “The first step we need to take is realizing that we need to align our definitions, we need to align our definition of success in the metrics that we use to articulate and that would help us focus our resources and investments into the areas that we have left behind — the areas of waste generation and waste redistribution where we believe there are significant improvements to be realized,” Hadar said.

He called for development of a common roadmap to tackle the key sources of waste in the fab, similar to how the ITRS has been used to tackle scaling challenges: Such a roadmap would:

– Leverage solid economic rationale, technical knowledge, and operational insights to evaluate and prioritize potential initiatives; set common vision and clear goals;
– Recognize and capture the value of agility — cycle time reduction — to expand the value of the 300mm fab investment; and
– Pursue sustainability and dramatic reduction in environmental impact

Hadar presented a perspective on waste (previously presented at ISSM in Tokyo last October), broken into four quadrants representing the factory, unit, wafer and equipment view (see figure). The goal in each area is to drive the waste to zero.

The focus in Quadrant 1 (lower left) is eliminating waste of silicon. Productivity solutions in this quadrant are well understood, and the waste perspective provides little new information, he said.

In Quadrant 2 (lower right), the waste perspective means reducing the input required to achieve the desired function. In this light, Hadar said equipment suppliers’ focus on outsourcing, consolidation, and streamlined operations clearly benefits industry productivity, as do the more traditional equipment design improvements.

Analysis of Quadrant 3 (upper left) from the waste perspective reveals fundamental manufacturing inefficiencies related to waiting time of wafers in the factory.

Waste in Quadrant 4 (upper right) stems from inefficient use of factory resources. In many fabs, production equipment often remains idle — even when WIP is available for processing. Furthermore, equipment variability causes unpredictable frequency and duration of maintenance events, resulting in transient queues in different areas of the fab. Many factories are forced to purchase additional equipment capacity to combat the effects of variability.

The semiconductor industry has systematically pushed waste into Quadrants 3-4, while making improvements in Quadrants 1-2, Hadar noted.

The 300mm Prime program represents an opportunity to create step-change improvements in 300mm fab productivity, Hadar said. If the semiconductor industry focuses its combined resources on these initiatives, with an unrelenting commitment to pushing process technology forward, he asserted, fab productivity can be driven for many years to come. — P.S.