Optical Metrology Regression Mapping for Semiconductor Overlay Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The increasing complexity and miniaturization of semiconductor devices lead to stringent requirements for accuracy in overlay and critical dimension (CD) measurements, which existing metrology processes struggle to meet due to errors from optical aberrations, patterning device heating, and substrate heating, affecting device functionality.
Innovation Solution
A method using regression algorithms to map SEM measurements to optical measurements, creating a model for the metrology process that augments data sets if the distance between measurements and statistical representations exceeds a threshold, ensuring accurate optical metrology recipes for high-volume manufacturing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If optical measurements are used for high-volume manufacturing metrology, then productivity is improved, but measurement precision deteriorates due to optical aberrations, patterning device heating, and substrate heating
Solution Approach 1:
The patent introduces an intermediary mapping model that translates optical measurement data into SEM-equivalent measurements. This mediator (the mapping model) allows the system to use fast optical measurements while achieving SEM-level precision through regression-based transformation, thereby resolving the contradiction between speed and accuracy
Solution Approach 2:
The patent replaces the mechanical/electronic SEM measurement system with an optical measurement system. By substituting the measurement mechanism from electron beam-based SEM to light-based optical metrology, the system achieves higher throughput while using computational methods to maintain precision
2Measurement precision
If SEM measurements are used as the reference standard, then measurement precision is improved, but productivity deteriorates due to slower measurement speed
Solution Approach 1:
The patent creates a computational copy of SEM measurement capability within the optical metrology system. By training a mapping model on paired SEM and optical measurements, the system copies SEM's precision characteristics into the optical domain, allowing optical measurements to achieve SEM-level accuracy without the slow measurement speed
Solution Approach 2:
The patent transforms the measurement parameters by applying regression-based corrections to optical measurement data. The mapping model adjusts optical measurement parameters (overlay, critical dimension) to match SEM reference parameters, effectively changing the parameter space to achieve both speed and precision
3Productivity
If optical metrology is used in high-volume manufacturing, then productivity is improved, but reliability deteriorates due to errors from optical aberrations and heating effects
Solution Approach 1:
The patent implements feedback through the mapping model that continuously corrects optical measurements based on learned relationships with SEM measurements. This feedback mechanism compensates for optical aberrations and heating effects, maintaining measurement reliability at high volumes
Solution Approach 2:
The patent performs preliminary action by pre-training the mapping model on extensive calibration data before production use. This preliminary calibration establishes the transformation relationships needed to correct optical measurements, ensuring reliability is built into the system before high-volume manufacturing begins
Data Source
AI summary
A method to calculate a model of a metrology process including receiving a multitude of SEM measurements of a parameter of a semiconductor process, receiving a multitude of optical measurements of the parameter of a semiconductor process, determining a model of a metrology process wherein the optical measurements of the parameter of semiconductor process are mapped to the SEM measurements of the parameter of the semiconductor process using a regression algorithm.