Debate Heats Up Over Bigger Glass

New equipment and materials will be necessary and costly, but the upside is better resolution.

popularity

For more than two decades, photomask makers have been talking about moving to a new and larger mask size, sometimes called “bigger glass” by the industry.

Generally, the discussions about “bigger glass” have been all talk and no action. But now, some chipmakers are turning up the volume in the discussions and are pushing for a larger mask size. A larger mask size would require the photomask industry to develop new equipment and materials. The question is whether chipmakers can convince the industry to go down what will likely be a long and expensive path.

The main argument for moving to a larger mask size is to help boost the resolutions for extreme ultraviolet (EUV) lithography. In addition to a larger mask, the EUV scanner itself may require a new projection optics system that could support a higher numerical aperture (NA). In other words, the industry is talking about the need to develop basically a new EUV system even before the first-generation scanners have been proven in the field.

Years ago, the industry talked about moving from the current six-inch mask standard to a larger nine-inch reticle. Now, the industry is looking at several different mask sizes and formats—seven-inch square; nine-inch square; 300mm square; and a 300mm round-like wafer. Today’s six-inch masks are based on a square format.

So far, no decision has been made about moving to a new mask size. At present, there are two schools of thought. The first one believes the mask size will remain at six inches, because it is too expensive to move to “bigger glass” and there are still some uncertainties with EUV.

The other school of thought is that the industry will eventually move to a larger mask size, but it’s unclear who will pay for the R&D. If the industry moves to bigger glass, the current favorite is the nine-inch reticle, with 300mm in the running. “12-inch square is technically feasible,” said Franklin Kalk, executive vice president and chief technology officer at Toppan Photomasks, one of the world’s largest merchant mask makers. “It will easily take five years. It will be incredibly expensive, probably about double the mask cost-of-ownership. You can apply all this to nine-inch. It’s essentially identical.”

Today, it costs between $110 million to $140 million to set up a mask shop to produce 28nm photomasks. In comparison, it may require a staggering $1 billion to set up a leading-edge mask shop capable of producing a larger reticle size, said Brian Grenon, principal of Grenon Consulting. “When you go to nine-inch or 12-inch glass, you essentially make the merchant shops extinct as leading-edge mask shops,” Grenon said. “So what you are going to have is only about five or six players (in the leading-edge mask business). The players will essentially be those who can afford 450mm wafers and EUV.”

In other words, only a handful of large chipmakers with captive mask shops—such as Intel, Samsung, TSMC, and a GlobalFoundries venture—could afford the move to larger masks. But it’s unclear which chipmakers could afford the mask costs if the industry moves to bigger glass. “Mask costs today are in the $150,000 to $200,000 range for a critical level. Going to larger glass would increase that cost by at least 10X per critical level,” Grenon said.

Higher NA or not?
There are similarities and differences between moving to a larger mask size and the migration from 300mm to 450mm wafers. Both efforts are expensive. The difference is the decision to move to bigger glass largely depends on the progress of EUV. ASML Holding’s production-worthy EUV scanner, the NXE:3300B, is shipping this year. Supporting today’s six-inch masks, the tool has a NA of 0.33, a 4X magnification scheme and a resolution of 22nm (half-pitch).

At 10nm and beyond, chipmakers may need to use today’s EUV scanners with multiple pattering. Another way to extend EUV is to move to a higher NA, which would likely require a new projection lens. And to maintain the use of today’s six-inch masks, there are three high NA options—an NA of greater than 0.45 with a 5X magnification scheme; 0.5 NA with 8X; or 0.6 NA with 8X, according to EUV lens maker Zeiss.

“First of all, EUV will presumably be inserted at the 7nm logic node,” Toppan’s Kalk said. “The first-generation tool can resolve roughly 10nm. So, you need to improve the resolution. How do you do that? Higher NAs. You can retain six-inch as a mask format (with higher NA). But then, you effectively reduce the field size of the scanner. You are maybe printing a half-field or quarter-field size.”

To obtain the desired full-field size of 26 x 33 on the EUV scanner, the optimal solution is a tool with a 0.45 NA and 6X magnification, which, in theory, will enable 8.9nm resolutions, according to Zeiss. This solution, however, would require a nine-inch mask. “If you increase the mask size, you can preserve the scanner field size,” Kalk said.

By staying with a six-inch mask—and reducing the field size—chipmakers may have to resort to stitching techniques. This involves the process of exposing one part of a pattern with one mask and then exposing the next part with a second mask. Then, the masks are shuffled around and stitched together, which is a complex and slow process. “Even though (stitching) can be done, it is something we don’t want to do,” said Pawitter Mangat, senior manager and deputy director for EUV lithography at GlobalFoundries.

“Looking ahead, we need to understand how to extend EUV (to) 7nm, 5nm and beyond,” Mangat said. “For that, we probably need a higher NA. We also have choices of quadruple patterning and double patterning. That could be a nightmare for EUV with all of the mask blank costs and the infrastructure that will support it.”

So, in turn, the industry may have no choice but to move to bigger glass. Mask equipment vendors are willing to develop new tools for larger reticles—if someone foots some or all of the bill. “This change is very difficult,” said Banqiu Wu, principal member of the technical staff and chief technology officer for the Mask and TSV Etch Division at Applied Materials. “It involves different interests. Some companies can make a little change and they can handle it. For some companies, they have to invest a lot to do the whole thing.”

Litho and mask shop options
The probable timing for high NA EUV is the 5nm node. It’s hard to predict, but at 5nm, the industry could have several lithographic options—193nm immersion and multiple patterning; EUV and multiple patterning; high NA EUV; multi-beam; and nano-imprint. Directed self-assembly (DSA), a complementary scheme, could be part of the mix.

Still, even if the industry moves to high NA EUV, it’s unclear if photomask makers can afford to move to a larger mask size. “There are a lot of variables to sort out,” said Janice Golda, director of lithography capital equipment development at Intel. “It’s a cost and return-on-investment question. It’s also a timing question.”

Regarding the future of the mask size, Golda ruled out the idea of migrating to seven-inch reticles, but she said the nine-inch and 300mm round formats are viable options. “We need to de-couple the technology complexity from the mask size change,” she said. “We can start with higher NA EUV on six-inch masks and view nine-inch as a productivity improvement. I think there are different paths that we can take that would mitigate risk and potentially enable productivity if we were to go that way.”

By going with the 300mm round format, mask makers could re-use some tools that are already used in 300mm wafer fabs, Golda said. “We have probably 30 something different tools in the mask shop. If I don’t have to develop all 30, that’s going to take the development costs down,” she said. “So, we should consider 12-inch rounds. You can take advantage of the existing equipment infrastructure for defect metrology or multi-layer deposition.”

Looking at the problem from another perspective, GlobalFoundries’ Mangat said the industry must deal with another challenge: the EUV mask itself. An EUV mask must be defect-free in the flow, but the industry is struggling to achieve this goal. “We are still trying to get six-inch defectivity under control (for EUV masks),” Mangat said. “With EUV, going to a larger blank size is going to be a nightmare.”

All told, if the industry moves to a new reticle size, mask makers must help fund the development of new tools and materials. And the photomask industry must get on the same page and coordinate their efforts, which is easier said than done in the business. “Everyone must agree to do it,” Applied’s Wu said. “We will need a lot of cooperation and collaboration.”