یک روش جدید از انسجام نسبی اندازه گیری در سیستم روشنایی و تجزیه و تحلیل عملکرد تمرکززدایی آن

Partial coherence (generally represented by σ) is one of the important parameters of lithographic tool to assess the performance of pupil fill. In this paper, a novel method of partial coherence measurement for the illumination system is proposed. Statistical results of measured σ by the proposed method are analyzed. The dependence of partial coherence on the defocus, which is the distance from the measuring position to the best image plane, is also investigated. The simulation results prove the effectiveness of this method, and with the defocus increasing, the measured partial coherence decreases. Generally, if three times of the standard deviation is required to be 1 × 10−3, the amount of defocus should be less than 96 μm.

As the speed of integrated circuits (IC) increasing, the individual device critical dimensions (CD) are required to be tighter and tighter. As a result, CD variation across the semiconductor chip has become one of the limiting factors in IC manufacturing. Compared to the other major source of CD variation (mask, resist processing, metrology), imaging optics are the primary contributor to image non-uniformity. As lens aberrations are continuously reduced, the dose non-uniformity and local changes in the partial coherence σ of illumination optics have been proposed as primary cause of across chip linewidth variation [1]. According to the previous investigation, different partial coherence values might be used for different types of features (i.e., for printing isolated lines, is advantageous to use smaller partial coherence values. For tighter pitches, higher partial coherence values result in increased resolution) [2]. In addition, σ variation usually leads to the changes of depth-of-focus and image contrast which would strongly influence the performance of lithographic tool [3] and [4]. So, rapid measurements of σ variation are very useful for process engineers to optimize exposure tool and process condition. Many works and patents have investigated the performance of pupil fill including the partial coherence. Typically, by utilizing a “negative pinhole” and exposing positive photoresist, Kirk et al. presented a method of in situ partial coherence measurement [5]. Grodnensky et al. proposed a method of measuring the partial coherence uniformity. It is based on the high sensitivity of σ variation to the length of macroscopically large diamond-shape marks printed in photoresist [6]. Watson et al. measured the partial coherence uniformity by using quadrant apertures. The illumination reaching the wafer plane is measured with a photodetector [7]. In all of these methods, σ is measured on the wafer plane after the litho-tool has been packaged. The lens aberration cannot be excluded although it is small. Moreover, resist exposure is also needed. These methods not only cost high expenditure, but also depend on the resist development. In this paper, a novel method of the partial coherence measurement for the illumination system is proposed. A pin-hole is located at the reticule such that the illumination source image is imaged through the pin-hole via the lens on to the photosurface of CCD camera. Because of the notable feature of CCD such as high sensitivity, fast response, small image distortion and so on, partial coherence can be measured instantaneously and rapidly by this method. It is more suitable for σ detection in the designing process of the illumination system. Statistical results of measured σ by the proposed method and the dependence of partial coherence on the defocus are analyzed.

A method of the partial coherence measurement for the illumination system is proposed. The detection system consists of pin-hole, lens, and CCD. The centers of pin-hole and the photosurface of CCD are at the focus of the lens. Pin-hole is located at the reticule in the ideal case. The simulation results prove that this method can measure the partial coherence accurately. Dependence of measured partial coherence σ on the defocus is also analyzed, and σ decreases monotonically with the defocus increasing. This suggestion may provide feasible guidelines for future experimental investigation.