Measurement System Analysis for Semiconductor Measurement Process

摘要

Quality management is an essential factor for enhancing the competitiveness of a company, and it is important to determine the capability of the measurement system in process improvement activities for quality improvement. In general, some observed dispersions in measurement-related activities are inherent in the product being measured, while others are due to the measurement system used. Semiconductor products are processed through a complex process, and quality control is performed through various measurement processes during the production process. Among them, wafer testing, which is the last step in manufacturing, is an important process to judge the quality of semiconductor products. In the wafer test process, an automatic test equipment (ATE) is used to check the electrical features of the created chips on the wafer, and to judge whether good or not products. To test wafers in ATE, it is necessary to connect to the chip electrode, PAD. Probe Card is used at this time. As a replaceable part, it is made and used according to the product and is generally in a parallel structure to test more quickly and efficiently many chips on a wafer. Since the probe card is extremely sensitive and detailed component, deformation such as abrasion occurs as it is used and affects the wafer test result. This paper presents an algorithm that can analyze the measurement system using the test result data of the measured wafer in order to maintain the optimum state of the ATE and Probe Card, the measurement system of the wafer test process. And using this algorithm, we propose the automated measurement process quality management system (AMPQMS) for the wafer test process. The real-time measurement system analysis algorithms are utilized several statistical analysis methods that analyzes the repeatability and reproducibility of the instrument. AMPQMS can properly combine various analysis algorithms for each product. The quality of the measurement system is real time managed using the analyzed results by the analysis algorithms.