Virtual Inline Quality Control for Memory Die Testing Bottlenecks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The manufacturing of memory devices involves extensive and costly testing to detect defects and process variations, which is time-consuming and often renders samples unusable, limiting the number and types of tests that can be performed.
Innovation Solution
The application of machine learning to interpolate virtual test results for devices that were not tested, based on correlations with other sets of tests, allowing for more rapid and accurate feedback on processing parameters and improving manufacturing efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If extensive testing is performed to detect defects and process variations, then measurement precision and reliability are improved, but loss of time and productivity deteriorate
Solution Approach 1:
The patent creates virtual copies of physical test results through machine learning models. Virtual test data is generated by training models on physical test results and then using these models to predict outcomes for additional samples, replacing the need for extensive physical testing while maintaining measurement precision
Solution Approach 2:
The patent replaces mechanical/physical testing systems with computational machine learning systems. Instead of physically testing each sample, the system uses trained models to predict test outcomes, substituting computational processes for physical measurement processes and thereby reducing testing time while maintaining accuracy
2Measurement precision
If extensive testing is performed to detect defects and process variations, then measurement precision is improved, but productivity deteriorates
Solution Approach 1:
Virtual test results are generated as computational copies of physical test outcomes. The machine learning models produce virtual data that mirrors physical measurement results, enabling rapid assessment of manufacturing quality without slowing down production throughput
Solution Approach 2:
The patent substitutes computational prediction systems for physical testing systems in the manufacturing workflow. This replacement eliminates bottlenecks caused by slow physical testing while maintaining the ability to detect defects and process variations with high precision
3Measurement precision
If more test samples are prepared and tested, then measurement precision is improved, but loss of substance deteriorates due to samples becoming unusable
Solution Approach 1:
The patent generates virtual copies of test samples through machine learning predictions. These virtual samples provide the same quality control information as physical samples without consuming physical materials, thereby improving measurement precision without losing substance
Solution Approach 2:
The system replaces physical sample consumption with computational sample generation. Instead of preparing and testing additional physical samples that become unusable, the machine learning system generates virtual samples that can be produced indefinitely without material loss
Data Source
AI summary
To provide more test data during the manufacture of non-volatile memories and other integrated circuits, machine learning is used to generate virtual test values. Virtual test results are interpolated for one set of tests for devices on which the test is not performed based on correlations with other sets of tests. In one example, machine learning determines a correlation study between bad block values determined at die sort and photo-limited yield (PLY) values determined inline during processing. The correlation can be applied to interpolate virtual inline PLY data for all of the memory dies, allowing for more rapid feedback on the processing parameters for manufacturing the memory dies and making the manufacturing process more efficient and accurate. In another set of embodiments, the machine learning is used to extrapolate limited metrology (e.g., critical dimension) test data to all of the memory die through interpolated virtual metrology data values.