0.15 µm: ArF or KrF?

Many device manufacturers have begun 0.15 µm development programs to meet the requirements of the recently published SIA Roadmap. One possible approach is early development with 248 nm exposure technology followed by volume manufacturing with 193 nm. A second is development and manufacturing solely on high NA 248 nm systems. Robert D. Allen of IBM (San Jose, Calif.) and Will Conley of Cypress Semiconductor (San Jose, Calif.) debate the pros and cons of each scenario.

Will: My preliminary data at 0.15 µm are very encouraging, using high NA DUV with mature 248 nm positive resists. I've seen greater than 0.7 µm of focus window and about 15% dose window with annular illumination. That's before OPC and phase shift. So why should I take a more expensive immature tool along with an immature photoresist?

Bob: I agree that DUV lithography has progressed in an amazing fashion in the past few years, thanks to better litho tools and tremendous progress in the development of high-performance chemically amplified (CA) resists. Our hope is that we'll get the same performance increase in the 193 nm lithography system (tool + resist process), only it can't take us 15 years to get there this time. Remember, IBM was working on first-generation DUV resists in the early 1980s. The worldwide development efforts that followed have lead to the ability to go subwavelength in resolution pretty easily. But, I would argue with your implied point that printing at such low k-factors is not without unforeseen consequences. Obviously, we have no choice but to do early development of 0.15 µm litho using DUV. Clearly, 193 nm is not ready yet in terms of resist or tool. But 0.15 µm with DUV, you're down in k factor and far below wavelength in resolution. This is something that hasn't been done before for manufacturing. Furthermore, your OPC requirements will be enormous; the nearly non-existent aerial image will create huge problems with image shortening. Will, your data suggest that environmental stability problems become even greater with further reduction in k factor.

Will: Oh, is it my turn now? Introducing 193 nm tools and resists doing lithography well below wavelength? That would be an industry first.

Bob: I know. Sounds pretty frightening, but the technological learning from DUV resist development translates so well to the 193 nm materials that I don't think that that's the problem. We have more difficult challenges, like developing the full set of desirable performance attributes in a single, new polymer package.

Will: Equipment suppliers say they will have a 0.70 NA tool ready in 1999. Many are taking similar tool platforms and adding a higher NA lens along with improvements in off-axis.

Bob: Don't you think that with higher NA lenses it will be more difficult to control aberrations and other lens-related parameters? What if your supplier tells you that we have problems with the glass, or we can't meet the lens specifications that are required? What's your strategy?

Will: There's risk in just about everything we do in the device business. I do agree that these problems can and will arise, however there has been tremendous progress over the past three years in DUV lens and resist technology. Today, 193 nm resists are nowhere near the stage of development that DUV was three years ago. There is a tremendous amount of work ongoing in OPC, and there have been a number of publications that have demonstrated significant improvements in process windows, even with currently available lenses.

Bob: But mask manufacturers are having a hard time making 0.25 m masks and are looking for help to develop 0.18 µm and sub-0.18 µm masks. Right now you’re asking these guys to print subresolution assist features and produce phase shifted features. The defect levels are and will continue to be out of control for a few years. Many of these guys are still using wet etch and are patterning with 15-year-old photoresist technology. Do you think that CA resists will be required to produce these subresolution assist features? I haven’t seen much work in this area.

Will: I agree that there are mask-making problems. However, these problems will be the same for 193 nm imaging. Furthermore, cost of current DUV resists can be down in the $1200-1400 range; 193 nm systems won’t be that low for a long time. Current 193 nm resists haven’t even demonstrated similar k factors to today's DUV resists. Have you even identified the polymer type you’ll use for 193 nm resists?

Bob: No, you’re correct, there is no one polymer, but at least three possibilities. For 193 nm there are several choices that include acrylic, alternating copolymers, cyclic olefin polymers and other even more exotic materials. There are three properties of these resists that need to be nailed: transparency, imaging and etch performance. We must work on transparency since the resist process will most likely use an anti-reflective underlay. We can easily achieve two of these in one resist composition, but achieving all three simultaneously is very tough. Just a few years ago, there was lots of excitement about 193 nm resist with decent imaging at 0.35 µm with 20% less etch resistance than APEX-E. Since then there's been an explosion of chemistries that can bring 193 nm imaging up to the level of DUV in a short period of time.

Will: I think etch resists of 193 nm have to be greater than that of 365 nm and 248 nm systems because of the thinner resist films needed to meet resolution and process window requirements.

Bob: I agree, but the same it true for DUV at 0.15 µm. Our early exploratory work in this area indicates that 193 nm resists are five to 10 times more transparent at 248 nm than are traditional phenolic DUV resists. This transparency translates into 90° wall angles and the potential to use greater thickness.

Will: But what about the supply of CaF2 for stepper and laser lens? There are only a few major suppliers, and the market is small. They will need to heavily invest to meet the possible demand.

Bob: Many of the stepper manufacturers have been working with CaF2 suppliers to address this problem, along with some activities at SEMATECH.

Will: We'll have to wait and see.