Keith K. Chao

Diffusion of photoacid generators by laser scanning confocal microscopy

Diffusion of the photogenerated acid during the period of time between exposure and development can cause contrast loss and ultimately loss of the latent image. This is especially relevant for chemically amplified photoresists that require a post-exposure baking step, which in turn facilitates acid diffusion due to the high temperature normally employed. It is thus important to develop techniques with good spatial resolution to monitor the photogeneration of acid. More precisely, we need techniques that provide two distinct types of information: spatial resolution on various length scales within the surface layer and also sufficient depth resolution so that one can observe the transition from very surface layer to bulk structure in the polymer blend coated on silicon substrate. Herein laser scanning confocal microscopy is used to evaluate the resist for the first time. We report the use of the confocal microscopy to map the pag/dye distribution in PHS matrices, with both reflectance images and fluorescence images. A laser beam is focused onto a small 3D volume element, termed a voxel. It is typically 200 nm X 200 nm laterally and 800 nm axially. The illuminated voxel is viewed such that only signals emanating from this voxel are detected, i.e., signal from outside the probed voxel is not detected. By adjusting the vertical position of the laser focal point, the voxel can be moved to the designated lateral plane to produce an image. Contrast caused by topology difference between the exposed and unexposed area can be eliminated. Bis-p-butylphenyl iodonium triflat (7% of polyhydroxystyrene) is used as photoacid generators. 5% - 18% (by weight, PHS Mn equals 13 k) resist in PGMEA solution is spin cast onto the treated quartz disk with thickness of 1.4 micrometers , 5 micrometers space/10 micrometers pitch chrome mask is used to generate the pattern with mercury DUV illumination. Fluoresceinamine, the pH-sensitive dye, is also used to enhance the contrast of fluorescence image. The typical PEB temperature is 90 degree(s)C for 90 seconds. 488 nm is used as the excitation wavelength. Both reflectance and fluorescence images (> 510 nm) are processed by using Adobe Photoshop. It was found that the reflectance is more sensitive to the change of the refractive index of the resist while the fluorescence is more sensitive to the distribution of the PAG/dye. The NIH Image software is used for acid exchange rate calculation. Second Ficks Law is applied to analyze the image change. The diffusion coefficient for this PAG in PHS during PEB is smaller than 8.8 X 10-13 cm2/s.

Achieving sub-half-micron I-line manufacturability through automated OPC

We present results of a verification study of totally automated optical proximity correction (OPC) for mask redesign to enhance process capability. OPC was performed on an aggressive 0.35 micrometer i-line LSI logic SRAM design using the automated OPC generation code Eoptimask, employing the aerial image simulation code FAIM, both from Vector Technologies, Inc. Three different tests were performed, varying in the aggressiveness and type of corrections made. The key issues addressed in this work are the predictive capability of the aerial image simulation and, particularly, the ability of automatically generated OPC to significantly improve the fidelity of the final printed resist image for different geometries. The results of our study clearly demonstrate the utility of automated OPC based on aerial image simulation. Key experimental results include: two-fold increase of depth of focus latitude; demonstration of the feasibility of full off-axis illumination on the stepper; successful resolution of different feature types (posts, lines and spaces) on the wafer to correct CD at a single common exposure and focus condition. Future research will address detailed issues in reticle manufacture and inspection which are critical for cost-effective large-scale OPC.