According to the semiconductor road map for technology, 157-nm lithography will make its production debut at the beginning of 2003. Two years ago, optical materials for 157-nm lenses, reticle substrates, and pellicles were not available. Furthermore, it was not clear that materials could be developed. Through extensive characterization and development efforts by International Sematech's 157-nm lithography group and others, the industry has made significant progress over the last two years in overcoming the challenges of the technology.
Two years ago, lack of a photomask (reticle) substrate material was believed to be a roadblock for 157-nm lithography. Initial tests of conventional synthetic silica showed negligible transmission at 157 nm. Magnesium fluoride (MgF2) offered reasonable transmission, but its high coefficient of thermal expansion (CTE) introduced significant registration errors caused by expansion of the substrate during mask exposure. Also, MgF2 was much more susceptible to chipping and breakage than traditional quartz substrates.
Development efforts by a number of quartz manufacturers in low hydroxyl (OH) fluorine-doped silica have resulted in several materials that meet industry transmission and durability requirements of 85% transmission and less than 1% degradation. These materials have CTEs and chemical and mechanical properties that are comparable to existing quartz substrates, characteristics that should ensure compatibility with existing processes and handling. Issues of size, homogeneity, and scaling to production volumes still remain, but the outlook for availability of suitable materials is positive. Patterning technology
Photomasks generate patterns on the wafer by strongly absorbing incident radiation or through the introduction of an attenuated phase shift or a strong phase shift. The current strong absorber of choice is chromium. The absorbance of chrome is sufficiently high at 157 nm to produce high-optical-density thin films suitable for imaging. Initial tests of chrome durability are encouraging, but researchers have yet to complete extensive testing.
Development of an attenuated-phase-shift mask requires materials that allow approximately 10% transmission through the film while imparting a 180° phase shift to the light. This necessitates the development of materials and structures with the correct refractive index and phase change characteristics as well as the required durability, defect density, and etching characteristics. Currently there is not a 157-nm attenuating material available for testing. Development of these materials requires significant effort. Several promising technologies have been developed for 193-nm attenuated masks, such as the multilayer attenuated-phase-shift materials developed by DuPont Photomasks, Inc. (Round Rock, TX); however, we have yet to optimize these materials and methods for 157-nm lithography.
Strong phase shift is accomplished by removing precise amounts of the substrate material to achieve the desired phase shift. The material properties of the modified quartz are sufficiently similar to existing quartz, which means only incremental process improvements should be required to implement strong phase shift at 157 nm. Pellicles
A single printable defect on a mask can bring the yield for a semiconductor device to zero. A pellicle is a thin membrane or plate placed between the photomask and the projection optics/wafer plane. The pellicle ensures the image quality of the pattern projected on the wafer by protecting the photomask from contamination. Because of their uniformity and extreme thinness, pellicles provide necessary protection but do not introduce image degradation when inserted into the optical path.
The polymers currently used for pellicles incorporated in 248-nm lithographic steppers are not suitable for use at 157 nm, a wavelength at which they absorb strongly and physically deteriorate after very low exposure levels. Initially, no known polymer materials had significant transmission at 157 nm. In May 2000, Sematech awarded the team of DuPont Photomasks and E.I. DuPont de Nemours and Co. (Wilmington, DE) a contract to develop new polymer materials for use at 157 nm. Testing done by MIT Lincoln Laboratory (MITLL; Lexington, MA) has shown these new materials have transmission rates of approximately 95%, approaching proposed industry specifications. The films experience a photochemical darkening that reduces transmission below acceptable levels, however. Development work to reduce photochemical darkening and improve lifetime to acceptable levels is in progress.
Pellicles made of the 157-nm silica materials for use in photomask substrates may be a viable alternative to polymers. Durability and transmission are close to acceptable ranges. Due to the mechanical properties of this material, however, the pellicle "membrane" must be 300 to 800 µm thick. At this thickness, small nonuniformities in the surface or thickness of the pellicle, as well as gravity-induced sagging of the membrane, will cause unacceptable image degradation. A thickness in the 800-µm range also would require redesign of the stepper lens systems to account for the effect of this element being introduced into the optical path. For comparison, polymer pellicles are about 1 µm in thickness.
Pellicles remain a significant area of risk in the 157-nm program, and further efforts and progress will be required to achieve an acceptable solution. Purging
Oxygen, carbon dioxide, and organics in the air absorb 157-nm radiation, and must be kept at levels of parts per million to prevent attenuation of the light source. Because of this, 157-nm systems require a nitrogen-purged optical system. Organics are attacked, and absorbing residues are deposited on optical surfaces. The volume enclosed by the pellicle must be kept in a pure nitrogen atmosphere to prevent both attenuation and degradation of the mask surface by deposited contaminations.
The challenges of producing, transporting, inspecting, and storing the reticle and getting it to the lithography equipment in a usable state are significant. The industry is examining multiple-handling purging schemes for the reticle/pellicle assembly. Work by MITLL and others on in situ cleaning shows significant promise for resolving the issue of deposited residue.
Overall, the community has made a tremendous amount of progress in making 157-nm lithography a feasible technology. The areas of substrate material, binary absorber layers, and strong phase shift look very promising. Attenuated phase-shift layers, pellicle protection, and reticle handling remain areas of concern, and active development programs are in place. It is important that the industry have all the materials for 157 nm ready in order to make the transition as smooth as possible to enable the industry to stay on track with Moore's law. oe
Joe Gordon is the worldwide R&D, sales, and marketing manager of pellicles at DuPont Photomasks, Inc., Danbury, CT.