How Low Can It Go? 248 nm Light Prints 100 nm Features
Just two months ago, researchers at the University of Texas at Austin surprised the lithography world by printing 80 nm features with 193 nm light (See Optical Lithography Lives On). Now, the latest advance may be even more significant.
Researchers from Photronics, MicroUnity, National Semiconductor and Sematech (Austin, TX), in work presented at the recent Photomask Japan '98 Conference, achieved good critical dimension (CD) control of 100 nanometer features using 248 nanometer deep ultraviolet (DUV) wavelength exposure on 4X lithography systems. The collaboration, part of Sematech's DELPHI project to determine the practical limits of optical microlithography, combined optical proximity correction (OPC) design features with alternating phase-shift photomasks (PSM).
DUV lithography is currently used for manufacturing of 0.25 micron (250 nm) features. Most observers expected that a shorter wavelength would be required for the 100 nm generation. (Expected to reach manufacturing in 2006.) Both 193 nm and x-ray exposure have been suggested, but neither has been demonstrated in volume production, at any design rule.
About a year ago, Sematech contracted MicroUnity to provide photomask design and process support to help address the proximity effect problems seen with phase-shift reticles. As features approach one another, their diffraction patterns change, causing CD's to vary significantly from the drawn design. These proximity effects can significantly limit yield, and worsen as circuit features shrink below the wavelength of light used to print them. Illumination with highly coherent light strengthens PSM effects, but also worsens the proximity effect.
Together, Sematech, National Semiconductor, Photronics and MicroUnity engineers designed test reticles with 180 degree phase-shifted structures, combined with MicroUnity's sub-resolution scattering bar features and fine selective biasing. The designs were drawn with a customized version of MicroUnity's MaskRigger software (See MicroUnity and DuPont Lower OPC Pain Threshold). Two types of simulations were done prior to reticle fabrication. First, the entire imaging process was simulated with PROLITH 3D, provided by FINLE Technologies of Austin, Texas. The program incorporates the latest resist models used at Sematech. These simulations fine tuned the placement and size of the OPC structures needed to obtain the desired image. Second, to minimize diffraction-related feature placement problems and to further improve process latitude, TEMPEST, a program from the University of California at Berkeley that simulates the electromagnetic field at the mask, was used to design the best mask topography for shaping the projected image. Finally, the alternating PSM topographical design was validated with non-OPC test masks designed by Benchmark Technologies (Lynnfield, MA) and manufactured by Photronics.
Upon completion of the design phase, Photronics fabricated the OPC reticles using their proprietary UltraRes process and phase-shift fabrication techniques, achieving resolution down to 0.25 micron on the reticle. Sematech then printed wafers using a 248nm exposure tool with a numerical aperture of 0.53, from Integrated Solutions, Inc. (Austin, TX). MicroUnity and SEMATECH then compiled and analyzed thousands of cross-section SEM measurements of the tiny resist features to determine optimal process conditions.
"We are extremely pleased with results on the printed layers," said John S. Petersen, Sematech Fellow and DELPHI Project Leader. "The data clearly indicates that OPC combined with alternating phase-shift can overcome proximity-related CD error, while taking full advantage of the resolution improvement of the PSM technique."
"This work presents the first clear picture of how to solve the proximity problems encountered with deep sub-wavelength phase-shift processing," according to Roger Caldwell, VP of silicon technology at MicroUnity. "The long-standing relationship we have with Photronics has been critical in our better understanding of the intricacies involved in the manufacture of complex OPC reticles."
"As a result of this collaboration, we are optimistic that our 248 nanometer deep ultraviolet capital infrastructure can survive a few more generations." said Robert Socha, a National Semiconductor senior engineer assigned to DELPHI.
A proof-of-concept demonstration is a long way from a manufacturable process. Still, this work radically changes the post-optical lithography picture. Most observers have always viewed 193 nm lithography as only a one or two generation solution. Further extension of 248 nm technology pushes the 193 nm insertion point back, and makes large investments in 193 nm more difficult to justify.
By Katherine Derbyshire