Observation-velocity-convergence-based determination of optimal seeding time for sapphire preparation via Kyroupous method

Abstract

High-quality seeding operations are crucial for ensuring the preparation of high-quality sapphire single crystals, in which the accurate determination of seeding timing is a key issue. However, existing detection methods possess limitations in obtaining sufficient feature points, assessing single feature dimension, and tracking spoke motion on the melt surface. To address these challenges, this study proposes an observation-velocity-convergence method to detect the optimal timing for crystal seeding. The industrial charge coupled device sensors replace eyes of skilled workers to observe in this method. This strategy involves applying an edge detection algorithm to analyze the surface texture of the sapphire melt, optimizing the non-maximum suppression process to increase the number of feature points to 10. Non-uniform sampling is employed to select key frames and extract relevant information about features. The stability of data is assessed by analyzing changes in adjacent angles of characteristic points on the melt surface, followed by determining the best seeding timing using a speed fitting model. This approach not only enhances the dimensional aspect of detecting seeding timing but also improves accuracy while expediting detection rate and reducing data redundancy. Experimental results demonstrate that this strategy increases Class A seed crystal seeding success percentage by 3.3% while reducing seeding time by approximately 0.2 h.