Image Sensor Output Binning for Low-Light Inspection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current image sensors face challenges in detecting low light levels at high speed with high spatial resolution and high signal-to-noise ratio due to reset clock noise, particularly in high-speed inspection applications for semiconductor wafers, where small defects and particles require sensitive detection.
Innovation Solution
The solution involves binning analog image data values before transmission and using an analog-to-digital converter (ADC) to generate multiple digital image data values between each reset of the image sensor's output sensing node, reducing reset clock noise and improving signal-to-noise ratios through output binning and multi-sampling processes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the intensity of illumination is increased to increase signal level, then signal-to-noise ratio is improved, but high power density degrades optics and may damage the article being inspected
Solution Approach 1:
Multiple pixels are merged into output bins that accumulate signals from multiple pixels before reading out. This combining of signals increases the effective signal level without requiring higher illumination intensity, thereby improving signal-to-noise ratio while avoiding optics degradation and sample damage
Solution Approach 2:
Signals from multiple pixels are accumulated in output bins before being read out to the ADC. This preliminary accumulation of signals prepares a stronger combined signal that can be converted with higher precision, reducing the need for high illumination power
2Productivity
If inspection speed is increased to meet high productivity requirements, then throughput is improved, but signal collection time per pixel is reduced leading to lower signal-to-noise ratio
Solution Approach 1:
Multiple pixels are combined into output bins that accumulate signals simultaneously. This merging allows the system to maintain high inspection speed while achieving higher signal levels through spatial combination of multiple pixel signals within the same time period
Solution Approach 2:
The patent transitions from single-pixel readout to multi-pixel output binning, adding a spatial dimension to signal accumulation. Multiple pixels contribute to each output bin, effectively increasing signal collection capacity without extending time per pixel
3Measurement precision
If high spatial resolution is maintained to detect small defects, then measurement precision is improved, but signal level per pixel remains low requiring higher illumination power
Solution Approach 1:
Multiple pixels maintaining high spatial resolution are merged into output bins that accumulate their signals. This allows the system to preserve fine spatial resolution for detecting small defects while achieving higher signal levels through combination of multiple pixel signals
4Productivity
If reset clock frequency is increased to enable high-speed readout, then inspection speed is improved, but reset clock noise increases degrading signal-to-noise ratio
Solution Approach 1:
Multiple pixels are accumulated in output bins before readout, allowing the system to operate at lower effective readout frequencies while maintaining high inspection throughput. This reduces reset clock noise impact while preserving high-speed inspection capability
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach enables high-speed, low-resolution imaging with high signal-to-noise ratios, enhancing the detection of low light levels and improving the accuracy of image data, allowing for multi-purpose inspection systems that can balance inspection speed and sensitivity.
Implementation Method 1
The pixel sensor converts the detected radiation into electrical charges
Data Source
AI summary
An inspection system and methods in which analog image data values (charges) captured by an image sensor are binned (combined) before or while being transmitted as output signals on the image sensor's output sensing nodes (floating diffusions), and in which an ADC is controlled to sequentially generate multiple corresponding digital image data values between each reset of the output sensing nodes. According to an output binning method, the image sensor is driven to sequentially transfer multiple charges onto the output sensing nodes between each reset, and the ADC is controlled to convert the incrementally increasing output signal after each charge is transferred onto the output sensing node. According to a multi-sampling method, multiple charges are vertically or horizontally binned (summed/combined) before being transferred onto the output sensing node, and the ADC samples each corresponding output signal multiple times. The output binning and multi-sampling methods may be combined.