BACUS attendees focus on mask costs

12/01/1998

BACUS attendees focus on mask costs

While the unknowns associated with future mask technology and costs linger, 248-nm DUV lithography strength was evident at the 1998 BACUS Symposium on Photomasks.

IBM assignee to SEMATECH, Wally Carpenter, reported, "Mask repair is a critical issue to be looked at immediately. We need more root cause analysis to determine what we can do to eliminate defects altogether. Automation may play a critical role." In addition, a SEMATECH survey shows with today`s capabilities "toward the 180-nm technology node, that mask-making repair falls apart, and at 130-nm increasingly everything is at risk," says Carpenter.

BACUS discussions involved the potential for maskmaking costs to soar. Carpenter said, "The device lines are going to find solutions to maskmaking issues and cost affordability with or without our assistance; I prefer that maskmakers find the solutions ourselves." He indicated that nonaffordable mask costs could lead the industry to alternate technologies; possibilities include direct write for gate features combined with conventional masking of less critical areas, larger magnification with field stitching on the wafer, or multiple masking. Carpenter stressed that mask industry players must partner with wafer fabs to learn defect reduction techniques.

A report from John Nistler of AMD slanted to US-based lithography and microprocessor production said, "4? DUV steppers are installed at every major semiconductor production fab. Pilot production is in the sub-0.2-?m gate region. Maskmakers haven`t really been pushed for smaller resolutions because many leading facilities have the ability to do off-axis illumination with steppers to avoid OPC down to 180 nm."

Nistler`s view is a long stretch for 248-nm DUV lithography because many "are hesitant to jump into 193-nm technology having just finished (suitable) 248-nm development work."

SEMATECH`s John Peterson, reporting on extending optical lithography to 70 nm, said, "While the SIA Roadmap says `193-nm lithography for 130-nm technology,` we think that we clearly have a 248-nm solution for all features except for maybe contacts." He sees that the extension of optical lithography involves integration of exposure tools and lens, mask and layout, resist, and processing if "we are to extend the resolutions that are historically beyond the typical k1 factors in production. Our goal is to use integration of the system and resist set-up to simplify the mask as much as possible, and really push lithography," he says.

Peterson said, "High transmission masks help, OPC and PSM must be thought of together, and unique illumination schemes take some load off masking. The resist threshold and its tuning is also important and will drive a lot of processing that will drive optical lithography."

Knowing that lens aberrations account for 20% in the linewidth budget, Peterson outlined optical lithography scenarios beyond 248 nm. "At 193 nm, if we can`t get rid of aberrations, we have to go to a ~0.8 NA stepper to realize the 70-nm technology node. If we can pound down aberrations, then we

probably can do 70 nm with a 0.7 NA stepper. Otherwise, we will have to go to 157 nm with a 0.7 NA stepper; we have no earthly idea right now if we could actually do something like this [157 nm]." Peterson noted that the possible "long ball" for lithography is interferometric imaging, the work of Steve Brueck at the University of New Mexico.

Peterson`s list of needs for extending optical lithography includes tunable illuminators, aberration-smart exposure tools, better resist systems, grid-based IC designs, high transmission ternary-simulated phase shift, alternating phase shift blanks, <2? of phase shift error, better proximity correction, and gridless lithography. "We have to really start thinking about these things and the way they will make our life easier. There are always going to be aberrations, but let`s build lenses to minimize them to meet our needs the best. In addition, cutting into quartz to create phase shifters is really not smart." For inspection in lithography, Peterson sees the need for "hierarchical inspection, a method that looks at the aerial image at the actinic wavelength, and we need to do phase and transmission defects." - P.B.