Researchers Seek Elusive Pellicle Solution
Readying extreme ultraviolet lithography (EUVL) for high volume manufacturing has presented many challenges along the road to creating next-generation semiconductors. Imec researchers believe they are a step closer in delivering a much needed pellicle solution.
Within a few years, the lithographic community needs a suitable pellicle to protect photomasks during extreme ultraviolet (EUV) exposures in high-volume manufacturing. While developing a new technology involves many vexing challenges, some seemingly innocuous aspects have far-reaching implications.
Developing a pellicle to protect photomasks is very challenging since EUV light is absorbed by most materials. Whatever pellicle is chosen must be greater than 90% transmissive at EUV's 13.5nm exposure wavelength. Materials that work for other lithographic processes do not work well in EUV. Thermal, chemical and mechanical requirements complicate pellicle development.
Emily Gallagher, principal engineer at imec, and her team are screening many candidate materials from both imec and partner organizations to assess suitability. In this article Gallagher explains the imec approach to pellicle development, and highlights carbon nanotubes as one of the promising solutions.
Why a pellicle?
Employing a pellicle is common practice in deep ultraviolet (DUV) lithography utilizing 193nm and 248nm wavelength exposures. The pellicle membrane protects the photomask from contamination. But the different exposure wavelength in EUV complicates pellicle choices.
"(The pellicle) is mounted a few millimeters above the surface of the photomask so that particles that land on it will be too far out of focus to print. For DUV, thin film pellicles can be fabricated from low-cost fluoropolymers. They are inexpensive and transmit over 99 percent of the light, ensuring that the imaging impact is minimized," Gallagher explained. "But extreme ultraviolet lithography (EUVL) is a different story. The 13.5nm EUV light is absorbed strongly by most materials, including fluoropolymers. Also, the pump-down sequences in the EUVL vacuum system and the high intensity of the EUV light source complicate the development of a suitable pellicle solution."
Gallagher added that the industry initially planned to introduce EUVL into manufacturing without any pellicle. The idea was to make the photomask handling and exposure vacuum chamber particle-free, eliminating the need for a protective membrane.
However, after defectivity assessments, the chip industry is convinced that a pellicle solution is mandatory for high-volume EUV exposure tools. The pellicle is currently considered a major area for EUVL development.
"While considerable progress has been made in improving the EUV source power, resist sensitivity and mask blank defects "“ areas that are also considered major challenges "“ a suitable pellicle for high power exposure (greater than 250 W) has not yet been found. That's why we initiated a project for pellicle membrane development as part of imec's Advanced Lithography Program. Our goal is to have a solution ready within two or three years from now, the time when the IC industry will need a pellicle for their high-volume EUV lithography tools," she said.
Optical, mechanical, chemical and thermal challenges
One of the most important requirements for the EUV pellicle is related to the transmission of EUV light. Gallagher noted that during EUV exposure, single-pass transmission (including light passing through the pellicle) must be at least 90 percent in order to ensure the productivity of the EUVL tool at a targeted source power of 250 W. Too low of a transmission would reduce the effective exposure power, hence the productivity of the tool (measured as wafers exposed per hour,) would be negatively impacted.
Finding a material with sufficient transmission qualities is very challenging since EUV light is absorbed by almost all materials. The pellicle also needs to be mechanically stable, which is difficult to achieve for membranes that are thin enough to meet light transmission requirements. In practice, thin pellicle membranes are mounted on a frame and fixed to the photomask. During use in the EUVL scanner vacuum chamber pellicles are subjected to handling and periodic pump-down/vent cycles, which enhance the risk for bulging and finally breaking membranes. Pellicle lifetime could also be affected by the presence of hydrogen radicals in the scanner since highly reactive hydrogen is used to keep the chamber sidewalls and optics clean, but it could also react with the pellicle material. Thermal considerations also complicate pellicle choices.
"We need to (also) take account of some thermal considerations. Some of the light that is not transmitted is absorbed, heating the membrane considerably. However, heat transfer options through evaporation or convection are minimal for a material in vacuum; heat conduction is very limited for a thin membrane. The only way to transfer heat is radiation. For some materials, additional emissivity layers will be needed to enhance radiation and decrease the peak temperature of the pellicle. These capping layers can also serve to protect the pellicle from hydrogen etching if selected appropriately," Gallagher said.
Imec's approach to pellicle development
To meet transmission requirements, it is logical to start with low absorption materials. This translates into selecting a material with a low EUV absorption coefficient k. Both silicon (Si) and carbon (C) fulfill this requirement.
"After choosing the right material, we can reduce the absorption even further by thinning and by reducing the density of the material. The latter can be done by either inducing voids (e.g. by etching into a continuous film), or by depositing a material that is inherently porous, like carbon nanomaterials," Gallagher explained. EUV pellicles have been in development for more than a decade. Reasonable options have already been proposed, but they all have their issues. Either the pellicle development is too immature (as is the case with graphene), or the pellicle membranes break at higher source powers (as is the case with poly-silicon or silicon nitride-based pellicles). A key element of imec's strategy is developing alternative solutions. "Building on existing solutions brings no additional value to our partners," Gallagher observed. One of many potential pellicle materials tested by imec are carbon nanotubes (CNTs). Carbon nanomaterials, like CNT films, can have an extremely low density, so the transmission is likely to be high.
"We also expect them to add to the membrane's mechanical strength. Another nice thing about carbon nanotubes is that their properties are tunable. This means that issues with the material can possibly be solved by careful engineering," she said.
Carbon nanotube pellicles can be fabricated using imec's 300mm process flow and can be scaled for larger wafer sizes. The pellicle process starts with a silicon wafer. Deposition, patterning and strip processes are deployed to create a flexible platform for membrane development. Silicon nitride (SiN) can serve several functions. It can either be thinned and used as a part of the final membrane stack, or it can be etched away so it is not part of the final membrane. The 300mm platform is compatible with scaling-up membrane sizes, targeting 10 x 10cm2 and even larger, which is important for developing 450 mm silicon wafer technologies. CNTs process
ed on this membrane platform successfully passed several tests that have been established at imec to evaluate potential pellicle solutions.
"The transmission of a stack of three to four layers of nanotubes (~50nm thick) was measured to be higher than 95%. The CNT-based pellicle was also subjected to mechanical testing. At imec, we built an experimental bulge tester that (can) apply a differential pressure across the membrane to test the point at which the pellicle bursts. This burst pressure is then compared to identical SiN membranes without additional layers. The measurements clearly show an improved durability when CNTs are added to the membrane.
"We also looked at the impact of hydrogen (H), which is particularly worrisome since hydrogen is known for etching carbon. The results from our hydrogen tester indeed showed that CNT films loose thickness after long exposure to hydrogen radicals. Fortunately, we can solve this issue by coating or encapsulating the CNTs with a suitable material that has minimal impact on EUV transmission, such as ruthenium or molybdenum," she explained.
Advancing EUVL pellicle development
Besides developing and evaluating its own pellicle solutions, imec is also screening pellicle materials developed by partners and other external suppliers. Gallagher said that EUVL pellicle development is a very complicated and challenging activity.
To accelerate development, imec believes it is important to have many companies exploring multiple paths in parallel, and to engage the EUVL community. For this reason, imec has made its testing processes available for partners including its in-house capabilities for optical, mechanical, chemical and thermal testing. This puts imec in a unique position.
"If we need a solution within two to three years, we must exploit this infrastructure to the maximum. With our carbon nanotube-based pellicle, we have a promising path forward, but we remain open to alternatives since enabling an industry solution constitutes a success for imec and for our partners," she concluded.