Original Date: 11/03/1996
Revision Date: 01/18/2007
Best Practice : Machine Vision Technologies
Oak Ridge National Laboratory’s (ORNL’s) Image Science and Machine Vision (ISMV) group conducts pure and applied research in the machine vision and perception areas. ISMV’s goal is to develop human-level visual and decision-making capabilities for computers and robots by emulating human sensory and cognitive processes. Over the years, the group developed many methodologies and systems for the U.S. industry and federal government to reduce waste and energy usage, and improve manufacturing processes and product quality. One technology developed by ISMV, Spatial Signature Analysis (SSA), is licensed to several suppliers to the semiconductor industry.
Currently, semiconductor manufacturers use image-based, defect detection and review workstations to process the monitoring and characterization steps. This method generates a huge amount of data which must be evaluated by production personnel. Alternatively, SSA works as a data-reduction process by automating the detection and classification of patterns or process signatures. These signatures are based on electronic wafermap data provided by current in-line measurement tools. The SSA process begins with a coordinate list of defect data points which are mapped to pixels in a wafermap image. Next, the images are subsequently organized into shapes, objects, and finally signatures. The signature feature description, embedded in the software, determines the defect classification. After completing the classification, the review for intelligent sub-sampling of off-line high-resolution defects can occur.
Integration of SSA technology with in-line defect detection and analysis strategies will result in product-yield improvements. ISMV estimates the investment return of SSA throughout the semiconductor industry at over $100 million per year for a 0.1% improvement in yields.
For more information see the
Point of Contact for this survey.