Pupil Metrology Using Angular Scatterometry for Bonding Parameters
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional metrology methods for monitoring parameters of interest in advanced semiconductor manufacturing, such as 2.5D and 3D integration, are time-consuming and lack sensitivity for small features like pitch and CD, especially when optical resolution is reached, necessitating improved methods for parameter monitoring.
Innovation Solution
A method involving angularly resolved metrology data analysis using a trained model to derive bonding parameters, employing radiation scattering data and machine learning techniques for precise feature characterization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional profilometry methods (AFM, confocal microscopy, white light interferometry) are used to monitor bonding parameters, then measurement accuracy is maintained, but measurement time increases and throughput decreases
Solution Approach 1:
The patent replaces mechanical contact-based profilometry methods (AFM) and slow optical methods (confocal microscopy, white light interferometry) with a scatterometry-based optical system that uses radiation scattering measurements. This substitution enables non-contact, rapid measurement of bonding parameters while maintaining precision, thereby increasing manufacturing throughput without sacrificing measurement accuracy.
Solution Approach 2:
The patent changes the measurement approach from direct physical profiling to optical scattering parameter analysis. By measuring scattering intensity at multiple angles and wavelengths and analyzing these optical parameters through a trained model, the system rapidly determines bonding parameters (recess height, critical dimension, side wall angle) without the time-consuming mechanical or sequential optical scanning of conventional methods.
2Measurement precision
If conventional profilometry methods are used, then measurement capability is sufficient for larger features, but sensitivity is lost when pitch and CD shrink towards optical resolution limits
Solution Approach 1:
The patent transitions from measuring only spatial dimensions (height, width) to incorporating angular and spectral dimensions. By measuring scattering intensity across multiple angles and wavelengths, the system creates a multi-dimensional scattering signature that contains rich information about sub-resolution features. This dimensional expansion enables detection of small pitch and CD values that are below the optical resolution limit of conventional imaging methods.
Solution Approach 2:
The patent changes the measurement parameters from direct spatial imaging to optical scattering characteristics (intensity, angle, wavelength). This parameter transformation allows the system to detect features smaller than the optical resolution limit by analyzing how these small features modulate the scattering pattern across multiple angles and wavelengths, rather than attempting to directly image them.
3Productivity
If angularly resolved scatterometry with trained models is implemented, then measurement speed and throughput increase, but system complexity increases due to multiple detection angles and model training requirements
Solution Approach 1:
The patent performs preliminary action by training the measurement model in advance using simulation data or reference measurements. The trained model stores the relationship between scattering patterns and bonding parameters, allowing rapid inference during actual measurements. This pre-computation eliminates the need for complex real-time calculations during production measurements, thereby increasing throughput while managing system complexity through offline preparation.
Solution Approach 2:
The patent introduces a trained model as an intermediary between the raw scattering measurements and the bonding parameter determination. This model acts as a mediator that processes the multi-angle scattering data and outputs the desired parameters (recess height, critical dimension, side wall angle). The intermediary model simplifies the measurement system by replacing complex real-time inversion algorithms with a pre-trained predictive model, thereby reducing computational complexity during production while maintaining accuracy.
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 the efficiency and accuracy of parameter monitoring in semiconductor manufacturing by providing rapid and sensitive measurements of small features, improving throughput and yield.
Implementation Method 1
obtaining angularly resolved metrology data relating to a measurement of said feature following illumination of said feature with a radiation beam and detecting radiation scattered by said feature at an angularly resolved plane
Data Source
Figure 1~2
Figure 3(a)~3(b)
Figure 3(c)
AI summary
Disclosed is a method of determining at least one bonding parameter of interest relating to a feature on a substrate comprising: obtaining angularly resolved metrology data relating to a measurement of said feature following illumination of said feature with a radiation beam and detecting radiation scattered by said feature at an angularly resolved plane; obtaining at least one trained model being operable to relate said angularly resolved metrology data to said bonding parameter of interest; and using the trained model to derive the bonding parameter of interest based on the angularly resolved metrology data.