Lithographic technology is crucial for fabricating ultralarge-scale integrated circuits. The performance requirements of lithography machines are constantly increasing owing to rapid developments in the semiconductor industry. The use of various resolution enhancement techniques has also led to a continuous reduction in the critical dimensions of integrated circuits. Among them, the source mask optimization (SMO) technology optimizes the pupil illumination mode and mask pattern of the illumination system, effectively improving the lithography resolution and increasing the depth of focus, and thus has been widely used in immersion lithography machines at ≤28 nm nodes. In particular, two methods are commonly employed to implement the freeform illumination: One uses diffractive optical elements to achieve freeform illumination. However, it lacks flexibility because a single element can only produce a single pupil shape. The other method involves precise control of the angular position distribution of the micromirrors in a freeform illumination module to realize different freeform illumination modes. This method is more flexible and can compensate for illumination system deviations in real time. Currently, the freeform illumination module has become a standard feature of immersion lithography machines at ≤28 nm nodes . Micromirror arrays (MMAs) are the core components for implementing freeform illumination, and accurate monitoring of their angular positions is a prerequisite for implementing any illumination mode. However, owing to the high integration and small size of the micromirrors in an MMA, adjacent mirror surface reflections can easily cause crosstalk during the actual detection. This study proposes a spot crosstalk suppression algorithm for detecting the angular positions of MMAs, which enables high-precision detection of the angular position of each micromirror, even under crosstalk conditions.

A spot crosstalk suppression algorithm for monitoring the angular position of an MMA is proposed to address the issue of crosstalk between adjacent mirror surface reflections, which is prone to occur because of the high integration and small size of the MMA. Accordingly, the relative light intensity matrix is first calibrated by the light intensity of the micromirrors under test and neighboring micromirrors in the array. Because the relative light intensity matrix remains unchanged during the testing process, the centroid light intensity matrix equation can be derived through unified testing and data processing of the MMA. Solving the light intensity matrix equation, the influence of spot crosstalk during the testing process can be eliminated, and ultimately, the angular position information of each micromirror in the MMA can be obtained.

This study is based on an MMA angle position detection unit (Fig.2), and three different spot overfill ratios are considered (Table 1). The simulation results show that when the MMA is illuminated with spots under different extra spot ratios, the use of spot crosstalk suppression algorithms can improve the accuracy of the angular position of each micromirror in the MMA (Tables 2 and 3). When the crosstalk exists between the spot size and the size of a single micromirror, the spot-crosstalk suppression algorithm significantly improves the accuracy of the angular position of each micromirror in the MMA. The maximum detection error of the micromirror is reduced from 5387.48 μrad to 7.29 μrad, meeting the requirement of the free pupil illumination module specification that the detection error is less than 10 μrad. The data also show that the accuracy of the angular position of the middle micromirror in the MMA is 10^-2 μrad for different spot sizes; whereas, the error of the angular position of the peripheral micromirror is larger owing to its deviation from the detection system axis. Overall, the spot-crosstalk suppression algorithm can significantly improve the accuracy of the angular position of each micromirror in the MMA when crosstalk exists.

This study proposes a spot crosstalk suppression algorithm. The reflection intensities of the micromirrors to be tested and their neighboring mirrors are measured in advance to determine the relative light intensity matrix in the MMA angle position measurement. The centroid light-intensity matrix equation of the MMA can be derived by combining this matrix with the relative light-intensity matrix obtained through MMA scanning and calibration. The angle position of each micromirror in the MMA can be obtained by solving the centroid light intensity matrix equation, which significantly improves the accuracy of the angle position determination when spot crosstalk is present. With this method, the accuracy of the angle-position determination of each micromirror in the MMA can be improved to within 10 μrad, which meets the requirements for detecting the angle positions of the micromirrors in the free-illumination module of lithography machines. Therefore, the proposed detection method effectively solves the crosstalk problem in the MMA detection process, which is of great significance for the practical application of free-pupil illumination modules in lithography machines.