Toppan talks about masks and its strategy .
Semiconductor Engineering sat down to talk about photomasks and lithography with Franklin Kalk, executive vice president of technology at Toppan Photomasks, a merchant photomask supplier. What follows are excerpts of that conversation.
SE: What’s hot in mask technology these days?
Kalk: It’s everything from the bleeding-edge like EUV to much more mature manufacturing. On the mature side, I liken it to the coming rise of IoT. IoT is still this sort of nebulous mass of potential and new applications. You don’t need 14nm to make IoT devices.
SE: Chips for the Internet of Things (IoT) involve 200mm fabs, right?
Kalk: 200mm fabs are enjoying a resurgence. They have legs and will have legs for a while. A lot of companies have 200mm wafer manufacturing capabilities. A lot of them are trying to find some capacity to buy and build this stuff.
SE: How is Toppan Photomasks responding to the demand for masks at the trailing edge?
Kalk: We certainly have manufacturing capacity for masks that would feed into the 90nm and up range. We welcome that. In China, this is where we have quarter-micron capabilities. We’ve been in China since in 1997. Fab 1, as we call it, is in Shanghai. It’s about a quarter-micron fab. We just built Fab 2 next to it and started that up a few months ago. It’s a 90nm line that can be extended to a 65nm line, with the addition of a couple of metrology tools. There is a lot of demand in China for that.
SE: What’s next for Toppan’s mask shop in China?
Kalk: There is room in that fab in the existing cleanroom to extend it to 65nm. But also, we can double the capacity we just installed. We also have expansion capacity for another equivalent size cleanroom in the same building. That’s not going to be a 65nm fab. That next step will probably be more like 40nm or 28nm capability. We have a lot of people asking us to do that now.
SE: Let’s move to the leading-edge. Where are we with EUV?
Kalk: The printing quality of EUV is not a question anymore. But you still have the same three hurdles. There is one around the source. I would call it the source plus the optics as you get into manufacturing. There is another one around the resists. That’s the classic trade-off between acuity or line-edge roughness, resolution and speed. And there’s the mask. That’s always been about defect management.
SE: What about the EUV source?
Kalk: It can operate at a reasonable level. But I wouldn’t call it at a high-manufacturing level. It’s 80 Watts for a duration of months. That’s a big deal. The collector optics still degrades fairly rapidly. But it’s not once a day.
SE: What about EUV mask blanks?
Kalk: The blanks are getting better, but they haven’t gotten tremendously better in the last couple of years. What’s really changed is how we manage the defects. The industry is making use of pattern shifts to account for defects. The density is low enough that you can use pattern shifts and have very high blank yield.
SE: Do we need actinic inspection for EUV masks?
Kalk: It’s one thing to have what I call a bridge strategy. In that case, you are doing low volume masks. You can use a bunch of techniques and piece them together in an approach that gives me good defect quality on the mask. At some point, however, you can’t do that. You will need actinic inspection. No one has a tool so far. We’ll probably have it. But I am not sure who the tool manufacturer will be. Given the cost of reducing one of those machines to practice, it’s going to end up being the cost of as a scanner. It’s going to cost that much.
SE: Do we need actinic inspection to put EUV into mass production?
Kalk: I don’t think that it will hold up introduction.
SE: What about EUV pellicles?
Kalk: I’m pretty optimistic the pellicle nut can be cracked in the next couple of years. It’s no more a science project. It’s an engineering program.
SE: Where does it make sense to use EUV?
Kalk: In 10nm logic, roughly speaking, a metal one layer is three masks. At 7nm, it’s four masks. If you can replace three of those four masks with EUV, that’s a real savings in terms of time and money. That’s a load in the fab.
SE: Toppan has developed an EUV mask based on a black border technology. What is that?
Kalk: Toppan has developed a structure. You etch down to the glass and then you etch the glass with a pattern. So it’s basically a diffraction pattern that is etched into the glass. It causes the out of band radiation to diffract out of the acceptance angle optics. So it makes it much darker than the standard black border. The idea is to reduce the out of band reflectivity. It can reduce it by 70%. It goes from around 4% to 5% reflectivity to more like 1%.
SE: Toppan Photomasks and GlobalFoundries have a joint photomask venture called the Advanced Mask Technology Center (AMTC). It is based in Dresden, Germany. Toppan also has a mask venture with IBM. What is the status of these ventures?
Kalk: The goal of AMTC is the development and reduction of practice of leading-edge mask technology. Just in the past year, GlobalFoundries purchased IBM’s Microelectronics business. Coincidentally, Toppan and IBM had a joint development program on advanced mask development, based in Burlington, Vermont. So, now we have three R&D groups–IBM in Burlington; AMTC in Dresden; and Toppan in Osaka, Japan.
SE: What’s next?
Kalk: They are all working together. Those three are morphing into a more cohesive bunch. So, instead of having two separate efforts, we are putting the AMTC and the Burlington efforts together.
SE: I assume Toppan will keep R&D in Osaka. Will you combine R&D in either Dresden or Burlington? Or will you keep both of locations?
Kalk: There is R&D in all three facilities. The one in Osaka is much more focused on certain types of R&D. They are also doing manufacturing development. Burlington is much more what I call elemental technologies. They have materials expertise. AMTC has good history with EUV. We’ll focus on the EUV stuff at AMTC.
SE: What about manufacturing?
Kalk: For leading-edge, it’s in Osaka, Japan. Burlington as well Dresden will do manufacturing.
SE: What technologies is Toppan ramping up today?
Kalk: Toppan has 14nm in manufacturing. 10nm is in a very advanced stage of development. 7nm is not an infant. But it’s barely a toddler.
SE: What are the capital spending trends for mask makers?
Kalk: The biggest CapEx items tend to be inspection tools. And then the writing tools are next. Luckily, those tools can be used over multiple technology cycles. With the advent of EUV, though, things change a lot. You need an AIMS tool that can work on an actinic basis. We are not talking about $5 million or $10 million for an AIMS tool. We’re talking more like $50 million type stuff. It’s incredibly expensive. We’re seeing a hockey stick in terms of CapEx when we look at EUV, depending on what tools are actually placed into the mask fab. It gives us pause.
SE: It appears the captive mask shops will ramp EUV masks first. What about the merchant players?
Kalk: We know we aren’t going to be the very first manufacturers of high-volume EUV masks. Eventually, the industry will ask us to do it. Then, we must face the equation of how much CapEx we want to spend.
SE: Can you build EUV masks?
Kalk: We can build EUV masks in Osaka, Burlington and Dresden. In Dresden, we’ve been building them since 2006 or 2007. We’ve produced a fair number. At this point, we have a pretty solid process.
SE: The captive mask shops appear to be doing a lot of leading-edge technology today. They seem to be taking share away from merchants, right?
Kalk: That has been going on for maybe 10 years now. Fifteen years ago, merchants were 65% of the worldwide business in masks by dollar volume. Now, captives are well over 50%. The technology demands are so strong right now. If you’re making wafers at 14nm, it’s almost imperative that you have mask making capabilities now. It’s a strategic asset. That doesn’t mean they don’t work with merchant mask manufacturers like Toppan. GlobalFoundries is a good example. Samsung also outsources a fair amount of work to the merchants.
SE: How do the merchant mask makers fit in the ecosystem today?
Kalk: We provide several benefits. First, we have very good development capabilities. We also provide what could be considered an earthquake strategy. If someone has an earthquake and their mask shop goes down, what do they do? We also provide surge capacity. If you can buy surge capacity on demand, then that’s a good thing. That’s a big part of our relationship with GlobalFoundries. They have internal capability. But they can rely on us to produce what they need and when they need it.
Related Stories
Survey: Mask Complexity To Increase
Gaps Remain For EUV Masks
Mask Supply Chain Preps For 10nm
7nm Lithography Choices
 |
 |
 |
 |
 |
 |
 |  |
At 7 nm, you would replace the three immersion masks with two EUV masks, because the H-V asymmetry at 32 nm pitch (or less) is too much, on the (NXE:33×0) tools currently being worked on. I would not call that a good bargain, it’s the same as adding double patterning cost to EUV. No one wants that.