Edge Detection Algorithm for Blurred Alignment Marks in Hybrid Bonding

The process of hybrid bonding requires high-precision alignment between the bonded objects, such as a die and a wafer. Our alignment process inserts a thin alignment scope between the die and the wafer just before bonding, observes alignment marks on each bonding surface, and calculates the misalignment to adjust the bond head position. While the implementation of the thin alignment scope can effectively narrow the gap between the die and the wafer, thereby minimizing potential bonding discrepancies, it should be noted that the quality of the obtained images deteriorates with thinning. The calculation of misalignment requires finding the position of alignment marks on each die and wafer, where edge detection is used, however, the previous methods are not accurate enough for images of degraded quality. This paper introduces a novel technique for the precise detection of alignment markers through the utilization of edge detection with a large amount of image information. In our experiments, the 3σ uncertainty of misalignment calculation was 1.9 nm, a value that amply underscores the achievement of a hybrid bonding system characterized by exceptional precision.