Automated Maverick Wafer Screening via Historical Die Parameter Deviation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Manual maverick wafer detection in semiconductor manufacturing is time-consuming and prone to delays, as it relies on experienced engineers to identify deviations from expected heat maps, which can lead to late detection of manufacturing issues.
Innovation Solution
An automated maverick screening method that measures parameters for each die on a semiconductor wafer, compares these measurements to historically expected values based on previous wafer data, calculates deviation parameters, and identifies wafers as maverick if the deviations exceed a predetermined tolerance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual detection by experienced engineers is used to identify maverick wafers, then measurement precision can be maintained through expert judgment, but productivity decreases due to time-consuming manual processes and delays in detection
Solution Approach 1:
The patent replaces manual engineer judgment with an automated computer-based system that uses algorithms to compare measured parameter values against expected heat map patterns. The system automatically identifies maverick wafers by detecting deviations from expected patterns, eliminating the need for manual review while maintaining detection accuracy through computational analysis of multiple parameters across the wafer surface.
Solution Approach 2:
The system performs self-service by automatically generating expected heat map patterns from historical data and using these patterns to detect anomalies in real-time. The automated system continuously monitors measured parameters, compares them against expected values, and identifies maverick wafers without requiring external human intervention, thereby maintaining both precision and productivity.
2Adaptability or versatility
If manual review of heat map patterns is performed, then adaptability to parameter-specific variations can be maintained through engineer expertise, but loss of time increases due to delays in identifying maverick wafers
Solution Approach 1:
The system performs preliminary action by pre-calculating and storing expected heat map patterns for different parameters and wafer positions based on historical data. These expected patterns are prepared in advance and used as reference standards, allowing the system to quickly compare against them during real-time screening without requiring manual analysis or adaptation during the detection process.
Solution Approach 2:
The system uses feedback mechanisms by continuously comparing measured parameter values against expected patterns and adjusting its detection criteria based on observed deviations. The system learns from historical maverick wafer data to refine its expected patterns, enabling it to adapt to parameter-specific variations automatically while maintaining rapid detection speeds through iterative feedback loops.
Data Source
AI summary
Automatic maverick screening of a semiconductor wafer taking into account historical position-dependence of a measured parameter. For each of multiple die on a subject wafer being screened, a die screening is performed that includes measuring the parameter value at the position and comparing against a historically expected parameter value at that position. This comparison results in a deviation between the measured and expected parameter values of the die at that position. Then, a deviation parameter value of the subject wafer is calculated using the deviation value of the plurality of die on the subject wafer. The subject wafer is identified as a maverick wafer if the deviation parameter value of the subject wafer falls outside of a deviation tolerance.