Inverse Optical Proximity Correction Model Calibration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current lithographic processes face challenges in accurately reproducing patterns with dimensions smaller than the classical resolution limit, particularly in low-k1 lithography, due to difficulties in achieving the desired shape and dimensions of functional elements in semiconductor manufacturing, where proximity effects and process variations complicate the transfer of design layouts to substrates.
Innovation Solution
A method involving the calibration and training of process models using simulation and machine learning techniques, such as convolutional neural networks, to predict and adjust patterning device patterns based on wafer data, improving the accuracy of pattern transfer by iteratively refining model parameters to minimize differences between simulated and target patterns.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If conventional lithographic processes are used for pattern transfer, then the manufacturing process is simple and fast, but the manufacturing precision deteriorates for features smaller than the classical resolution limit
Solution Approach 1:
The patent applies preliminary action by performing inverse lithographic process simulation before actual pattern transfer to predict and correct proximity effects. The process model is calibrated in advance using simulated wafer data and metrology measurements, allowing the system to pre-determine optimal patterning device patterns that compensate for expected process variations, thereby achieving high manufacturing precision without increasing physical process complexity
Solution Approach 2:
The patent implements feedback by iteratively calibrating the process model using metrology measurements from actual wafers. The system compares simulated patterns with measured patterns, adjusts model parameters, and re-simulates until convergence is achieved. This closed-loop feedback mechanism ensures the process model accurately reflects real process behavior, enabling precise prediction and correction of pattern dimensions
2Productivity
If process model parameters are calibrated using simulated data, then the calibration is fast and efficient, but the measurement precision deteriorates due to lack of real process data
Solution Approach 1:
The patent uses preliminary simulated wafer data to establish an initial process model calibration quickly. This simulated data provides a starting point that captures expected process behavior, enabling fast initial calibration without waiting for extensive real wafer measurements. The simulated data serves as a preliminary foundation that is later refined with actual measurements
Solution Approach 2:
The patent combines simulated data calibration with feedback from actual metrology measurements. The system iteratively refines the process model by comparing simulated patterns with real measured patterns, adjusting model parameters to minimize differences. This hybrid approach maintains calibration speed while progressively improving measurement precision through real data feedback
3Manufacturing precision
If inverse lithographic process simulation is used to predict patterning device patterns, then the manufacturing precision improves for sub-resolution features, but the computational time increases
Solution Approach 1:
The patent performs inverse lithographic process simulation as a preliminary step before actual pattern transfer. By simulating the patterning process in advance using the calibrated process model, the system predicts optimal patterning device patterns that account for proximity effects and process variations. This preliminary simulation enables high manufacturing precision without requiring iterative adjustments during production
Solution Approach 2:
The patent creates a virtual copy of the patterning process through simulation. Instead of physically iterating through multiple patterning attempts, the system uses the calibrated process model to generate a virtual representation of the patterning outcome, predicts the optimal patterning device pattern, and directly applies this predicted pattern to the actual process, eliminating time-consuming physical iterations
Data Source
AI summary
A method for calibrating a process model and training an inverse process model of a patterning process. The training method includes obtaining a first patterning device pattern from simulation of an inverse lithographic process that predicts a patterning device pattern based on a wafer target layout, receiving wafer data corresponding to a wafer exposed using the first patterning device pattern, and training an inverse process model configured to predict a second patterning device pattern using the wafer data related to the exposed wafer and the first patterning device pattern.