Fab Tool Industry Has Lost Its Way

The relationship between chipmakers and fab tool vendors has always been a bit rocky, but the supply chain has generally worked. Chipmakers demand a tool for a particular application. Then, tool makers attempt to deliver the goods, and ask few, if any, questions.

Now, fab tool executives are beginning to ask some tough questions about the industry. And the tension is mounting between equipment makers and their customers.

Many executives wonder if the industry can afford to fund all of the costly chip projects on the table. Two projects–450mm technology and extreme ultraviolet (EUV) lithography—are the biggest question marks. Both 450mm and EUV are late and the returns are unclear.

The industry is also wondering if it can continue to march down the two-year process technology cycle without falling into bankruptcy. Many would like to slow down Moore’s Law, but not stop it all together.

Here’s the bigger question: Is the industry headed in the wrong direction? “We used to be driven by innovation and excitement. That was the value proposition,” said Terry Brewer, president of Brewer Science, a supplier of materials and equipment. “This industry has drifted. It has drifted away from innovation. This is my big concern: Can we innovate and is innovation still there?”

Instead of innovation, the industry’s focus has shifted towards the cost factor. “I hear very little about a value proposition,” Brewer said during a panel at SEMI’s recent Industry Strategy Symposium (ISS). “I see an industry where every other word is cost. I see an industry that scales for the purpose of lowering the cost of the chip.”

Brewer also indicated that the industry has become too pre-occupied with 450mm and EUV, thereby hampering innovation on other fronts. “Everyone is throwing all of their eggs in one of those baskets,” he said. “That’s getting kind of boring.”

In fact, the sky isn’t falling in lithography. “If you go talk to customers, their biggest issue is do we have EUV or not EUV? And oh my goodness, we can’t do double patterning because it’s too expensive,” he said. “Some also (say) that we are running into the end of lithography. That isn’t true. Whether we have EUV or not, we can drive (optical lithography) down quite a bit. We will drive that down further with different device designs and different materials. We have a long, long way to go.”

And over time, the industry has been moving away from traditional brute-force scaling. Instead, chipmakers are relying more on new design methodologies and materials. These technologies, in turn, will enable new design architectures based on finFETs, FDSOI and stacked-die.

“I think higher value things are those that improve performance, lower power and reduce cost,” said Mike Splinter, executive chairman at Applied Materials. “And since we have limited resources, we ought to press everything for those (metrics) and prioritize in what we do in our businesses.”

In fact, the industry may need to re-think its priorities. 450mm is a good example. For some time, Splinter has been an outspoken critic of 450mm. The Applied executive believes the transition from 200mm to 300mm fabs made sense, but he still questions the need for 450mm.

“The unit growth rate in the industry is dramatically different. In the 90s, we were seeing a tremendous growth rate of 25% to 30% unit growth every year. We really needed more wafers,” he said during the ISS panel. “Today, the growth rates are smaller, maybe 5% to 10%. There are very few fabs being built relative to that. The real issue is not CapEx. I don’t see the tremendous demand driver for the next wafer size.”

There are other concerns. “The other thing that I’m really worried about in the (450mm) wafer size is there are no memory manufacturers behind it. Most of the wafers are made by the memory manufacturers. And if we don’t have that volume, I am worried we won’t get the cost per wafer to come down as fast,” he said.

Splinter believes there is still a lot of room for improvement in 300mm. “The technology changes that are being thought about for 450mm are much easier to do on 300mm. I think we could change to notchless wafers and productivity. And let’s really concentrate our dollars there,” he said.

Still, there is a strong possibility that 450mm will happen. “If customers desire to make a risky move like this, I guess we have to weigh it against everything else that we do and make those trade-offs. Customers have a way of getting what they want,” he said.

As in the past, customers may get what they want. But at least suppliers are now beginning to ask the tough and hard questions. That’s a good start but will it be enough?