Image Accuracy Optimization Method for Maskless Lithography Based on DMD(Invited)
Hua Liu,Long Huang,Yuqing Liu
DOI: https://doi.org/10.3788/cjl240651
2024-01-01
Abstract:Objective Lithography is a key technology used in the processing of microelectronic chips, integrated circuits, and micro-optical components. The desired pattern is achieved through photochemical reactions or physical transformations of photosensitive materials.Maskless lithography is an advanced technique in which a digital mask pattern is directly projected onto a photosensitive material,thereby eliminating the need to employ a conventional lithography mask. The use of digital micromirror device (DMD)-based maskless lithography has garnered significant attention and has widespread applications in the field of lithography, owing to its smaller single-pixel dimensions, higher fill rate, excellent ultraviolet light durability, and faster micromirror flip-frame frequency. However, several factors can affect the accuracy of projection lithography in DMD-based maskless lithography systems, including projection distortion,pixelation, and proximity effect. In this study, we conduct a comprehensive analysis of the three factors that affect lithographic pattern quality and systematically summarize the characteristics and application scenarios of several methods to enhance lithographic pattern accuracy based on the extensive research conducted in our laboratory over the years. By integrating digital mask compensation, pixel superposition, and femtosecond local modification techniques, we successfully fabricate lithographic patterns that exhibite significantly reduced projection distortion, pixelation effects, and proximity effects. Methods A simple, convenient, and inexpensive digital correction method was used to address the problem of projection distortion in DMD lithography systems. The measured distortion values (Delta R at different radii) were fitted to obtain the corresponding curves.Subsequently, the position of each exposed pixel on the mask was corrected based on the fitting curves. To address the inherent pixelation effect of the DMD, we utilized a motion platform that carried the substrate to perform multiple microshifts in both the x-and y-directions within one projection pixel size. The final lithographic pattern was generated by superimposing the exposures from a series of mask patterns. We regulated the superposition exposure time to ensure that specific regions reached the exposure threshold,thereby controlling the size of the exposure area and enhancing pattern resolution. To mitigate proximity effects, we employed a DMD lithography system to process pattern portions with lower precision requirements while utilizing the two-photon polymerization (TPP)method for pattern sections that demanded higher accuracy. This approach combines the efficiency advantages of DMD processing with the precision advantages of TPP to significantly enhance lithographic pattern accuracy. Results and Discussions To validate the feasibility of our optimization method, using the digital mask compensation method, we process a Fresnel zone plate with a diameter of 2680.3 mu m, focal length of 60 mm, and working wavelength of 623 nm. The diameter error of the Fresnel zone plate is reduced from 37.7 mu m to 1.7 mu m (Fig.2). The results of the pixel superposition method are shown in Fig.4. The lithographic patterns processed without the use of the pixel superposition method exhibit obvious central position error(3.8 mu m), line width error (4.1 mu m), and sawtooth size (4.4 mu m). In contrast, the lithographic patterns processed using the pixel superposition method show significantly improved accuracy in terms of central position error, line width error, and sawtooth size,which are calculated to be 0.8 mu m,0.75 mu m, and 0.71 mu m, respectively. The smoothness of the lithographic pattern edges and precision of the line positions are significantly enhanced. In the femtosecond laser local modification method, the outer ring is processed using DMD technology, whereas the inner lines undergo TPP modification. The circular area most affected by the proximity effect has a diameter of 20.5 mu m and exhibits only a marginal deviation of 0.9 mu m from its theoretical value of 19.6 mu m. The interior of the pentagram is processed using DMD, whereas the outer frame is modified using TPP. This process significantly improves the proximity effect at the corners of the pentagram, reduces the positional error from 15.5 mu m to within 1 mu m, and greatly enhances pattern precision. Conclusions This study comprehensively analyzes three factors affecting the accuracy of lithographic patterns: projection distortion, the pixelation effect, and the proximity effect, along with optimization methods. It systematically summarizes the characteristics and usage scenarios of several methods by which to improve the precision of lithographic patterns. The digital mask compensation method we use offers simple operation, high precision, and strong flexibility, and it reduces the projection distortion from 37.7 mu m to approximately 1.7 mu m. Pixel superposition methods, such as spatiotemporally modulated technology, reduce the edge aliasing and quantization errors of lithographic patterns to 1/6 of their original levels. The oblique lithography method not only effectively reduces costs but also enables the processing of single-pixel smooth curves and can be applied in the scanning exposure process. The pixel dynamic adjustment method and femtosecond local modification techniques effectively reduce the lithographic proximity effect. Finally, through the combined application of TPP and DMD processing methods, highly precise lithographic patterns with significantly reduced projection distortion, pixelation effects, and proximity effects are successfully fabricated