Virtual Metrology Feedforward Prediction Semiconductor
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Critical dimensions of semiconductor substrates are difficult and expensive to measure physically, especially for 3D structures, as physical measurements can destroy the substrate and are challenging for high aspect ratio features.
Innovation Solution
A virtual metrology system that uses mathematical models to predict critical dimensions based on tool sensor data and measurement data, where a first set of predicted critical dimensions is used as input to a second model to estimate a second set of critical dimensions, incorporating a predicted relationship between the two sets.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If physical measurement methods are used to determine critical dimensions, then measurement accuracy can be achieved, but the substrate is destroyed and measurement is impossible for 3D high aspect ratio features
Solution Approach 1:
The patent creates a virtual copy of the substrate's critical dimensions through mathematical modeling. Instead of physically measuring the actual substrate features, the system builds a virtual model that replicates the critical dimension data through feedforward relationships, thereby avoiding direct physical contact and potential substrate damage while obtaining accurate measurement data
Solution Approach 2:
The patent replaces mechanical/physical measurement systems with a computational/mathematical system. The critical dimension measurement function is transferred from physical probes or imaging systems to a virtual metrology system that uses mathematical models, sensor data, and algorithms to calculate and predict critical dimensions without mechanical interaction with the substrate
2Measurement precision
If physical measurement methods are used for critical dimensions, then direct measurement data is obtained, but the process is expensive and destructive
Solution Approach 1:
The system creates virtual replicas of critical dimension measurements through mathematical modeling. By using feedforward relationships between different critical dimensions, the system generates accurate measurement data for features that would otherwise require destructive physical measurement, eliminating substrate loss while maintaining measurement capability
Solution Approach 2:
The virtual metrology system uses the substrate processing system's own sensor data and process parameters to self-determine critical dimensions. The system leverages existing process data and建立的 feedforward relationships to calculate critical dimensions internally, eliminating the need for external destructive measurement processes
3Adaptability or versatility
If separate models are used for each critical dimension, then comprehensive coverage is achieved, but system complexity increases
Solution Approach 1:
The patent merges multiple critical dimension prediction models into a unified virtual metrology system. Instead of treating each critical dimension prediction as an isolated model, the system integrates multiple models that share common feedforward relationships and process parameters, creating a cohesive system that predicts multiple critical dimensions through interconnected mathematical relationships
Solution Approach 2:
The virtual metrology system creates universal feedforward relationships that can predict multiple different critical dimensions using the same underlying model framework. The system uses generalizable mathematical relationships that apply across different feature types and locations on the substrate, reducing the need for separate specialized models for each critical dimension
Data Source
AI summary
A controller includes a memory that stores a first model corresponding to a first critical dimension of a substrate processed by a substrate processing system and a second model corresponding to a second critical dimension of the substrate. The second model includes a predicted relationship between the first critical dimension and the second critical dimension. A critical dimension prediction module calculates a first prediction of the first critical dimension of the substrate using the first model, provides the first prediction of the first critical dimension as an input to the second model, and calculates and outputs a second prediction of the second critical dimension of the substrate using the second model.