Optical Defect Review via Multi-Condition Scattering Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In semiconductor manufacturing, finer circuit patterns lead to smaller defects that are difficult to detect due to Rayleigh scattering, which is buried in noise from roughness scattering and sensor noise, making it challenging to visualize defects using existing optical defect inspection methods.
Innovation Solution
A method involving multiple images captured under varying optical conditions to differentiate defect scattering light from roughness scattering light by analyzing inter-image characteristic quantities, allowing for the discrimination and visualization of defects.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Illumination intensity
If high intensity illumination and long scattering light accumulation time are used to ensure sufficient defect scattering light against sensor noise, then defect scattering light intensity is improved, but light scattering from rough surface buries the defect scattering light and prevents detection
Solution Approach 1:
The patent segments the scattering light into two distinct categories: defect scattering light and roughness scattering light. By capturing multiple images under different optical conditions and analyzing inter-image characteristic quantities, the system separates these two types of scattering light, allowing defect detection without interference from roughness scattering.
Solution Approach 2:
The patent dynamically varies optical conditions (such as illumination angle, polarization state, or wavelength) across multiple image captures. This dynamic change in optical conditions causes defect scattering light to maintain consistent characteristics while roughness scattering light varies, enabling differentiation and isolation of defect signals.
2Measurement precision
If filter transmittance conditions are tightened to increase the difference between defect scattering light transmissivity and roughness scattering light transmissivity, then the ratio of defect scattering light to roughness scattering light is boosted, but defect scattering light transmissivity decreases and sensor noise influence increases
Solution Approach 1:
Instead of using a static filter with tight transmittance conditions, the patent dynamically changes optical conditions across multiple measurements. This allows the system to capture defect scattering light effectively under various conditions while roughness scattering light varies, maintaining high defect-to-roughness ratio without excessive filter restrictions that would attenuate defect signals.
Solution Approach 2:
The patent uses feedback from multiple image captures under different optical conditions to adjust and optimize defect detection. By analyzing inter-image characteristic quantities, the system continuously refines its ability to distinguish defect scattering light from roughness scattering light, maintaining high measurement precision while preserving adequate defect scattering light intensity for reliable visualization.
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
Enhances defect detection sensitivity and throughput by reliably visualizing fine defects buried in noise, improving the accuracy of defect identification and localization.
Implementation Method 1
Scattering by an object that is sufficiently smaller than the wavelength of illuminating light is Rayleigh scattering. Because the intensity of this scattering light is proportional to the sixth power of the diameter of a particle that scatters light, the defect scattering light drastically decreases along with size shrinkage of a defect under measurement
Implementation Method 2
light scattering from a rough surface of a sample and noise attributed to sensors and circuits including speckle noise and heat noise
Data Source
AI summary
A method for reviewing a defect including a light capturing step that illuminates a sample with light under plural optical conditions, while varying only at least one of illumination conditions, sample conditions, or detection conditions, and detects plural lights scattering from the sample; a signal obtaining step that obtains plural signals based on the lights detected; and a processing step that discriminates a defect from noise according to a waveform characteristic quantity, an image characteristic quantity, or a value characteristic quantity created using the signals and derives the coordinates of defect.