IC Defect Detection via Voltage Error Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Integrated circuit devices face challenges in achieving precise control of reference voltage during manufacturing, with existing trim methods struggling to distinguish between random defects and other errors, leading to potential device failure under temperature cycling.
Innovation Solution
A methodology involving the use of a low pass filter to analyze initial voltage errors, comparing each device's error to a peer-filtered value to detect deviations indicative of randomly placed defects, allowing for identification and removal of defective devices from the production stream.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If trim methods are used to adjust reference voltage after wafer manufacture, then the reference voltage can be set close to the ideal target, but the ability to distinguish between random defects and other errors is compromised
Solution Approach 1:
The patent applies preliminary action by performing voltage error measurement and low pass filter processing on each device before the trim process. This allows defects to be identified and devices to be selected or rejected in advance, enabling the trim process to focus only on adjustable parameters rather than attempting to correct all types of errors including random defects.
Solution Approach 2:
The low pass filter serves as an intermediary mechanism between the voltage measurement and the trim process. It processes the raw voltage error data to smooth out random variations and highlight systematic errors, creating an intermediate representation that distinguishes between adjustable errors (caused by process variations) and non-adjustable errors (caused by random defects).
2Manufacturing precision
If tight control of process values is attempted during wafer manufacturing, then reference voltage accuracy can be improved, but the complexity and difficulty of achieving precise matching increases
Solution Approach 1:
The patent performs preliminary characterization of each device's voltage error profile during wafer manufacturing, measuring and filtering the error data before the trim process. This preliminary action identifies devices with random defects early in the process, allowing the manufacturing system to focus its precision control efforts only on devices that can be successfully trimmed, thereby reducing overall process complexity.
Solution Approach 2:
The low pass filter extracts and isolates the systematic error components from the total voltage error, separating adjustable errors from non-adjustable random defects. This extraction allows the manufacturing process to focus precision control resources only on the extractable errors, simplifying the overall control requirement.
3Productivity
If all devices are subjected to full trim processing, then more devices may achieve specification, but the time and resources required for processing increase
Solution Approach 1:
The patent performs preliminary voltage error measurement and low pass filter processing on all devices during wafer manufacturing, before the actual trim process. This preliminary action classifies devices into those with random defects (that cannot be trimmed) and those with adjustable errors (that can be trimmed). By identifying trim-candidate devices in advance, the full trim processing time is avoided for defective devices, thereby reducing overall processing time and increasing throughput.
Solution Approach 2:
The low pass filter extracts and removes random defect signals from the device population, isolating only the devices with systematic errors that are suitable for trimming. This extraction creates a filtered list of trim-candidate devices, allowing the trim process to be applied only to the subset of devices that can benefit from processing, thereby reducing total processing time and resources required.
Data Source
AI summary
In a method of testing integrated circuit devices, a parameter, such as initial voltage may first be measured. A low pass filter operation may be applied to the measured data to generate peer data. A particular integrated circuit device may be identified as failed or rejected when its measured parameter varies sufficiently relative to the peer data.