Mask Alignment is a very important part of modern day VLSI fabrication process. To replicate the desired structure on the mask to the wafer, it is necessary to have some degree of accurate mask alignment procedure. However, present day mask alignment process is operated by manual inspection and hence may produce significant human errors. The objective of this work is to develop a novel mask alignment procedure based on image registration technique which is independent of manual inspection. For this purpose only requirement is a standard webcam to capture the images of mask and wafer to be registered. It is well known that registration is a technique by which one object is aligned geometrically with respect to other. Present study shows that genetic algorithm/particle swarm optimization based mask registration technique produces satisfactory results in a reasonable time. First section of this work describes the registration technique of mask and wafer images in details and following this registration values second part is hardware implementation of mask alignment procedure.
TopIntroduction
Application of artificial intelligence (AI) is a wide range of fields like medical diagnosis, robot control, remote sensing, industrial production planning, pattern recognitions and so on. Moreover, it is now realized that many complex real world problems may be resolved in an advantageous manner using some artificial intelligent techniques (Rich & Knight, 1991; Russell & Norvig, 1995; Vasant, 2014). These intelligent systems are supposed to own humanlike expertise within a specific domain, adapt themselves and can learn in the changing environment. These techniques also automate the processes. In this study, AI is applied in a procedure of photolithography by which engineers replicate the desired structures on the mask to the wafer. It is desired that transformation of patterns/features of the mask to the wafer should not obscure the surrounding features of the wafer. However, the feature size of modern day VLSI fabrication process is reducing and more than one masking operation is required which in tern introduces the importance of exact replication of the structures (Oldham, 1977). Hence, proper alignment of mask with wafer is becoming a challenging issue. However, present day mask alignment process is operated by manual inspection and enormous human resources are exploited for this purpose. Although some alignment marks are designed in the photolithography process, this manual operation may produce significant human errors (Hawkins, 2004). In this regards, image based registration/alignment methods using some kind of artificial intelligence may resolve the problem of human errors and automate the procedure. This automated and intelligent mask alignment system is supposed to be humanlike efficient and adapts itself in changing environment. It is found in the domain of research that simulating complex biological evolutionary processes may lead us to discover how evolution motivates living systems towards higher level intelligence (Jang, Sun, & Miutani, 2002). They are more robust than traditional methods and performing better for many real world problems. Hence, greater attention is paid to evolutionary computing techniques such as genetic algorithm, particle swarm optimization algorithm which are based on the evolutionary principle of natural selection. When the search space is too large and it is difficult to identify knowledge that can be applied to reduce the search space, users have no choice but exploit the principle of natural selection (Fonseca & Fleming, 1995). Present study also attempts to show that the application of evolutionary optimization techniques efficiently align the images of photomask and wafer in a reasonable time. This is because simulation of natural selection processes or complex biological procedures can apply population based systemic random searches and hence efficiently find the optimum solutions. Proposed image based registration method using evolutionary optimization techniques also produces reasonably accurate results which are well acceptable in optical lithography process. Automated shifting (either translation or rotation or both) of mask using proper hardware arrangement makes this approach much more interesting and innovative.
Rest of the chapter is organized as follows: section 2 describes the background of the work, section 3 provides the motivation of the work, proposed registration based automated mask alignment technique is given in section 4, hardware implementation of mask alignment procedure is described in section 5, Conclusion including merits and demerits are described in section 6, and future direction of research are drawn in section 7.