Why Mask Blanks Are Critical

Geoff Akiki, president of Hoya LSI at the Hoya Group, sat down with Semiconductor Engineering to talk about optical and extreme ultraviolet (EUV) lithography as well as mask blanks. What follows are excerpts of that discussion.

SE: Mask blanks are components that serve as the base or the substrate for a photomask. Why are they critical?

Akiki: If you look at Hoya, we’ve been positioned as leaders in mask blanks for a long time. We are a critical part of the general photomask industry. The blank is really a substrate. Optical blanks can be transmissive or refractive. You have phase-shift or binary here. In contrast, EUV masks are mirrors for X-rays. Both types help get those patterns on the wafers, which eventually become the circuits.

SE: For years, the industry has relied on mask blanks for traditional optical lithography. Optical blanks consist of an opaque layer of chrome on a glass substrate, right?

Akiki: It’s a fairly sophisticated piece of quartz measuring 6- x 6-inches with a certain thickness. Traditionally, it has been chrome. Chrome has been around forever. There are other films like molybdenum silicide or MoSi. Silicon nitride is coming up and replacing MoSi in some places. That’s the space for optical.

SE: Mask blanks play an important role, right?

Akiki: Mask blanks are a critical part of the supply chain. Optical blanks have been around a long time. But they’ve never really been a bottleneck. There have been some tight and longer lead times. But it’s really not a limiter. We have to develop and get them ready a couple years before the node ramps up. The masks have to be ready to do your development in the fab.

SE: That all changed with the advent of EUV and the associated blanks for the technology, right?

Akiki: EUV is even more critical because of the rapid ramp. I joined Hoya a little over three years ago. No one was producing EUV blanks in volume. And all of a sudden, people wanted to go into high gear. So we ramped them up. It was difficult. There were a lot of challenges in terms of defects, yield and control. We learned about them. Part of going through that pain early on and having to produce them in volume is you learn things. The first, second and third order things come out much more quickly. With optical blanks, there’s enough options and it’s well understood. It has been around a while. EUV is much less mature. And it’s critical you understand all the pieces of it and what your expectations should be.

SE: EUV and optical blanks are different, right?

Akiki: EUV is different. An EUV blank has 40 molybdenum and silicon pairs. The defects specs are quite difficult. The defects need to be near or at zero levels. Then, there is a demand from vendors to reduce the 3D effects. An EUV blank is very difficult. It can’t have any problems with the reflection of the X-rays. You must mitigate those defects. That takes time and effort.

Fig. 3: Cross-section of an EUV mask. Source: Luong, V., Philipsen, V., Hendrickx, E., Opsomer, K., Detavernier, C., Laubis, C., Scholze, F., Heyns, M., “Ni-Al alloys as alternative EUV mask absorber,” Appl. Sci. (8), 521 (2018). (Imec, KU Leuven, Ghent University, PTB)

SE: In the early days, the EUV blanks were beset with defects. Have those issues been resolved?

Akiki: They’ve pretty much been resolved for the current generation. We’re shipping reliable products with high yields. No one has 100% yields. But we are in a high production mode with high productivity. We’re ramping even more tools. We’re working on zero-defect blanks. But the defects are under control and we have our arms around them. But time doesn’t stand still, and neither do our customers. They want tighter specs. They want to see a higher output. One way you can manage your yield and defects is to take more time. But they want higher output with the tools running every second. And that’s something you have to balance. At the same time, they’re asking for new materials. But then, you have to start re-learning as you go forward. They want thinner blanks with a little bit different metallurgy mix.

SE: To make EUV mask blanks, you deposit alternating layers of materials on the substrate. Then, you need to inspect it. What types of deposition and inspection techniques do you use?

Akiki: We use ion beam deposition (IBD). We have some PVD depositions, but for the critical ones, we’re using IBD. For inspection, we have two capabilities–actinic and optical. Actinic blank inspection (ABI) is where the rubber hits the road. We use optical inspection. That helps with the characterization. It’s useful for other reasons like screening. But as we get more ABI tools and that capability, that’s really the go to in the standard you need. We’re pushing the capabilities there as people ask for even tighter specs.

SE: For some time, we’ve seen chip shortages and components in the market. What about the supply and demand for mask blanks?

Akiki: Let’s start with optical. We’ve invested a lot of years in optical. Our capacity is in good shape. Over the years, we’ve been pretty good about adding capital and building to the demand. Is there a surge? There is. There’s three segments here–low end, medium range and advanced stuff. Our demand is driven by tape outs, not wafers. At the leading edge, there are fewer designs. Some of those layers are being siphoned off by EUV. So at the very edge at least from a volume standpoint, you’re seeing intense demand, but maybe lower numbers. People want them and they’re tough to make. The number of tape outs have come down. The number of layers has gone up, but some of those gets siphoned up for EUV.

SE: What about EUV blanks?

Akiki: EUV is somewhat different. First, we’re the clear leaders. We have been pushing the output. Demand is quite strong. However, it’s the early ramp on demand with incredible growth rates that you know won’t sustain through time. They have to level out. It’s obvious to us it is not going to stay at the 35% to 50% growth rate a year. But we continue to see growth. We continue to see our capacity utilization being very high. We continue to invest. We see a healthy period, at least for the next two or three years.

SE: Are we going to migrate from binary EUV masks for 0.33 NA EUV to high-k and/or phase-shift masks for high-NA EUV?

Akiki: It’s never that simple. In general, there’s a couple things we believe pretty strongly. One, ruthenium and tantalum will be around for a long time. That’s what we have today for EUV. That’s the ruthenium cap layer and the tantalum absorber. We will have those binary baseline things around for a while, just like we still have chrome in optical. In addition, we will start extending the technology just like we did for optical. Phase-shift masks are going to come in. We may have some double patterning. For high-NA, though, people think of that as a switch that drags all these other things. That will also come in phases. So we will push 0.33 NA for a while. When you get to 0.55 NA EUV, you start with what you know before you click into even more radical things. Right now, we’re using a MoSi multilayer structure. Everyone is looking at ruthenium in the future. How long is that going to take to get to an alternative structure? I don’t think it’s going to be a year or two. It’s going to be three, four or five years. That’s because you are going to have to relearn everything like defects, yield learning and control of your deposition. So, you’re always going to prioritize what you’ve have. So if you can make what you’ve have work, that’s probably the way you want to go.

SE: What about the absorber for high-NA EUV?

Akiki: If you are a lithographer, you will say: ‘I like these properties. Just give me these materials and thicknesses.’ But you need to look at the pieces. Can you etch it? Can you get to the line you want? The theoretical and practical now start to play in. Nickel is a good case. It’s a high-k material that is difficult to etch. There are some great papers with ‘n’ and ‘k’ curves. That’s one way to look at it. But if you’re going to implement this in manufacturing, you must make it fast at high yield. Those things are still being debated. Right now, it’s the R&D guys driving a lot of this. When the manufacturing guys get hold of it, they will have a different push-pull. Once again, a lot focus is on the R&D pieces. The real trick here will be integration and making it work in manufacturing, getting it out the door as a product. For instance, you have things like flatness, which we spend a lot of time worrying about. You have defects, which we all talk about. Researchers want to talk about high-k and low-n. That’s very important. There are many other facets that come into play. I’ve talked about etch. All those things have to line up. In a sense, the selection of all these things is like trying to tune a process window. It’s what gets you to the thing that’s usable at the end, not under ideal conditions.

SE: So the industry is exploring several mask blanks technologies in the future, right?

Akiki: There are several approaches as you go forward, whether it’s phase-shift, low-n or high-k. People have their favorite approach based on the capabilities they’re most comfortable with or they have available. For us, it’s a much wider space we operate in. First, we have the broadest capabilities in the industry. And we have real world problems that we’re trying to solve in this integrated view. Here’s the key–the less you change, the better. So if you can keep the material and tune it, that will be a preferred approach, if it works. If it doesn’t quite hit the cliff, then you need to get something new.

SE: In the future when high-NA appears, do you envision supporting a number of EUV mask blank technologies, such as binary, high-k and PSM?

Akiki: I look at optical. We still have hundreds of part numbers, different kinds of resists and metallurgies. EUV will have a broad menu. Certainly, customers will want their flavors. Some of these things will be niche-oriented products. For instance, high-k masks could be a niche application. Some of the others as well. We’re used to that diversity. As you go through time, customers probably will want more and more customization. At the same time, it’ll be harder and harder to justify that.