| This thesis introduces a new class of interferometric pattern and probe-based aberration monitors based upon the principle of coherent electromagnetic spillover among transmitting phase-shifting regions and assesses their implementation on the currently installed base of photolithographic exposure tools. The mathematical theory developed is based upon the diffraction integral for coherent imaging and shows that the optimum patterns for measuring aberrations are the two-dimensional inverse Fourier transforms of the original aberration functions, expressed as Zernike polynomials. As a result, each aberration monitoring target consists of a sub-resolution probe surrounded by a number of pattern rings which are near the resolution limit of the exposure tool. The positions and phasing of the rings are determined by the original aberration function. The rings act to spill electric-field over onto the central probe position in an amount which varies linearly with the amount of the given aberration.; Through simulation the monitors show an intensity response at the central probe position of 20 to 30% of the clear field per 0.01 l rms of aberration. This is over 50x stronger than the Strehl ratio response. Using the defocus target, it has been possible to measure the defocus aberration down to 0.01 l rms. Simulation results also show that the targets are, at most, 1/6 as sensitive to other, similar aberrations that they are not designed to detect.; The targets were fabricated in multi-phase masks for a maximum numerical aperture of 0.80 and wavelengths of 248 nm and 193 nm, and their experimental performance has been characterize using AIMS and photoresist exposures. In practice, the performance of the targets is limited by a number of second order effects. Analysis shows that misalignment, biasing, and phase etch error all affect the defocus target to less than 5% in electric-field. The most influential factor is the impact of the illumination partial coherence, as it weights the spillover fields from the ring elements differently depending upon their radii. While this causes a 28% change in the electric-field for sigma = 0.30, the effect can be reduced to less than 4% by lowering sigma to 0.15. The second most important effect is that of the electromagnetic mask transmission for etched features, which reduces the field from the probe by 20%. This effect can be eliminated however by biasing the feature edges by 0.044 lNA . Finally, while it does not impact the probes, high-numerical aperture polarization effects at NA = 0.80 change the desired aerial image fields from the rings by 20%. This is reduced to less than 6% however when the image is transferred into photoresist. |
