In this paper,we demonstrate an auto accurate alignment method to align mask-substrate in the prototype of plasmonic lithography(PL),which is essential for multilayer nanostructure fabrication with high resolution,low cost,high efficiency,and high throughput,such as circuit manufacturing and other applications.We obtained an alignment signal with sensitivity better than 20 nm by using the Moiréfringe image.However,only using the Moiréfringes cannot guarantee the alignment of the mask and the substrate because the Moiréfringe repeats itself when the mask and substrate are offset by a fixed displacement.To eliminate the ambiguity,boxes and the crosses alignment marks are designed beside the grating marks on the substrate and the mask,respectively.A two-step alignment scheme including coarse alignment and fine alignment is explored in the auto alignment system.In the stage of coarse alignment,the edge detection algorithm based on Canny operator is adopted to detect the edges image effectively.In the process of fine alignment,Fourier transform based on Moiréfringe image is obtained to improve the alignment accuracy.In addition,experimental results of overlay indicate that PL can obtain sub-100 nm alignment accuracy over an area of 1 cm~2 using the proposed two-step alignment scheme.Via the substrate-mask mismatch compensation,better stages and precise environment control,it is expected that much higher overlay accuracy is feasible. In this paper, we demonstrate an auto accurate alignment method to align mask-substrate in the prototype of plasmonic lithography (PL), which is essential for multilayer nanostructure fabrication with high resolution, low cost, high efficiency, and high throughput, such as circuit manufacturing and other applications. We obtained an alignment signal with sensitivity better than 20 nm by using the Moiréfringe image. Yet, only using the Moiréfringes can not guarantee the alignment of the mask and the substrate as the Moiré filter repeats itself when the mask and substrate are offset by a fixed displacement. To eliminate the ambiguity, boxes and the crosses alignment marks are designed beside the grating marks on the substrate and the mask, respectively. A two-step alignment scheme including coarse alignment and fine alignment is explored in the auto alignment system .In the stage of coarse alignment, the edge detection algorithm based on Canny operator is adopted to detect the edges image effe ctively.In the process of fine alignment, Fourier transform based on Moiréfringe image is obtained to improve the alignment accuracy. In addition, experimental results of overlay indicate that PL can obtain sub-100 nm alignment accuracy over an area of ​​1 cm ~ 2 using the proposed two-step alignment scheme. Via the substrate-mask mismatch compensation, better stages and precise environment control, it is expected that much higher overlay accuracy is feasible.