188宝金博页面版

Automated aberration extraction using phase wheel targets Lena Zavyalova, Anatoly Bourov, Bruce W. Smith Rochester Institute of Technology, 82 Lomb Memorial Dr., Rochester, NY 14623 ABSTRACT An approach to in-situ wavefront aberration measurement is explored. The test is applicable to sensing aberrations from the image plane of a microlithography projection system or a mask inspection tool. A set of example results is presented which indicate that the method performs well on lenses with a Strehl ratio above 0.97. The method uses patterns produced by an open phase figure 1 to determine the deviation of the target image from its ideal shape due to aberrations. A numerical solution in the form of Zernike polynomial coefficients is reached by modeling the object interaction with aberrated pupil function using the nonlinear optimization routine over the possible deformations to give an accurate account of the image detail in 2-D. The numerical accuracy for the example below indicated superb performance of the chosen target shapes with only a single illumination setup. Keywords: aberrations, phase wheel, lithography 1. INTRODUCTION Aberrations are directly related to loss of contrast in the image. Modern lithography lenses require optical path tolerances on the order of several nanometers over extremely large apertures. If the primary aberrations are well-corrected, it is important to know how much wavefront distortion remains. Higher order aberrations are therefore of consequence due to their increased contribution to the total root mean square (rms) wavefront error in large pupils. Measuring the optical aberrations in microlithography projection systems, such as stepper or aerial image microscope, when working from the image space is not trivial. Zernike polynomials are commonly used for the wavefront description. For many inverse aberration extraction methods however, it is generally thought that as long as the Zernike set is used there will be some aliasing 2 in the terms (i.e. higher order terms are indistinguishable from low order counterparts of the same azimuthal order and symmetry). Overall, the approaches currently being used to measure aberrations in-situ are image-based and estimate the wavefront phase either from the gradients 4,5,6 or from the intensity point spread functions. 7 These techniques have been modified and adapted in numerous ways by a number of contributors, 3–10 and include analysis of resist images or aerial images in their specific context. These processes often require a linear system representation when analyzing the response to aberrations. While the image formation in lithography systems is always non-linear, some linearity assumptions are valid to some degree. In certain tests, correction factors are applied to enforce linearity in the model, including lumping the aberration terms of the same type together. We investigate a nonlinear solution to this problem of obtaining high order coefficients using numerical methods. The number of parameters to be estimated is large. We want to reliably measure up to and including the 9th order Zernike terms (35 individual aberrations) using a compact (small footprint) target and fast setup. The goal is to detect low levels of aberration (of Strehl above 0.9) at high resolution (less than 0.002 waves). Various methods, where image data is used for aberration retrieval, may require very intensive test cycles. For practical applications it is necessary to minimize the amount of data collection for input. In some situations, steppers are detuned and additional aberrations are introduced as a quantitative way to characterize the method performance in the optical system. In order to estimate the sensitivity of our model, a simulation study is performed with a synthetic lens, whose aberrations are precisely known. The organization of this paper is as follows. We start with an introduction to the use of Zernike polynomials in studying optical aberrations (section 2). Our analysis and discussion will be centered on the corresponding wavefront aberrations. In section 3 the method of nonlinear least squares will be introduced as a way of obtaining the aberration coefficients.