Process Control Metrology Using Measurement Model Regression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current metrology systems in the semiconductor industry face challenges in accurately estimating process parameters due to limited process windows and significant uncertainties, especially with increasing resolution requirements, complex geometric structures, and the use of opaque materials, making it difficult to predict relationships between measured geometric parameters and process parameters.
Innovation Solution
The implementation of a measurement model that includes both process and geometric parameters, using regression-based optimization to estimate process parameter values from measurements of structures fabricated on a wafer, with a transformation model to reduce dimensionality and a trained profile model to directly estimate shape profiles from process parameter values, enabling in-line measurements with improved accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If DOE wafers are used to establish mapping between process parameters and structural parameters, then process parameter values can be determined directly from measurements, but the process window captured is very limited and accuracy is reduced due to significant uncertainties in DOE process parameter values
Solution Approach 1:
The patent pre-calculates and stores the relationship between process parameters and structural parameters in a lookup table before actual measurements are performed. This allows direct determination of process parameters from measured structural parameters without requiring DOE wafers, thereby expanding the process window coverage while maintaining measurement accuracy through the pre-established comprehensive mapping relationships
Solution Approach 2:
The patent creates a virtual model (lookup table) that copies and stores the complex relationships between process and structural parameters, which can then be queried during measurements without needing physical DOE wafers. This virtual copy enables broader process window coverage while maintaining the accuracy benefits of comprehensive parameter mapping
2Adaptability or versatility
If indirect determination of process parameters through structural parameter measurements is used, then measurements can be performed on product wafers, but it is very difficult or impossible to predict the relationship between measured geometric parameters and process parameters
Solution Approach 1:
The patent pre-calculates the relationship between process parameters and structural parameters and stores it in a lookup table before product wafer measurements are performed. This eliminates the need to predict relationships during actual measurements, enabling accurate indirect determination of process parameters from structural parameter measurements on product wafers
Solution Approach 2:
The patent introduces a lookup table as an intermediary that bridges process parameters and structural parameters. This intermediary stores the complex relationships that would otherwise be difficult to predict, enabling accurate indirect determination of process parameters from structural measurements without requiring direct prediction capabilities
3Productivity
If model based regression with transformation models is used, then computational complexity is reduced and in-line measurements are enabled, but measurement precision may be reduced compared to full model approaches
Solution Approach 1:
The patent segments the measurement process into two distinct phases: an offline training phase where comprehensive models are built and stored in lookup tables, and an online measurement phase where pre-computed models enable rapid in-line measurements. This segmentation allows full models to be used for model creation while simplified models provide fast measurement throughput
Solution Approach 2:
The patent performs computationally intensive model training and relationship mapping in advance, storing results in lookup tables before actual product measurements are performed. This preliminary action enables rapid online measurements with reduced computational complexity while maintaining precision through the use of pre-computed comprehensive models
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
This approach allows for quick and accurate estimation of process parameter values, enhancing yield improvement in semiconductor fabrication by bridging the gap between measured structural parameters and process parameters, even in complex scenarios, and reducing computational complexity.
Implementation Method 1
Optical metrology techniques offer the potential for high throughput without the risk of sample destruction. A number of optical metrology based techniques including scatterometry, ellisometry, and reflectometry implementations and associated analysis algorithms are commonly used to characterize critical dimensions, film thicknesses, composition, overlay and other parameters of nanoscale structures.
Data Source
AI summary
Methods and systems for estimating values of process parameters based on measurements of structures fabricated on a product wafer are presented herein. Exemplary process parameters include lithography dosage and exposure and lithography scanner aberrations. A measurement model is employed to estimate process parameter values from measurements of structures fabricated on a wafer by a particular fabrication process. The measurement model includes process parameters and geometric parameters of structures under measurement. In some embodiments, a model based regression of both a process model and a metrology model is employed to arrive at estimates of at least one process parameter value based on measurements of a fabricated structure. In some embodiments, a trained measurement model is employed to directly estimate process parameter values based on measurements of structures. The measurement model is trained based on simulated measurement signals associated with measurements of shape profiles generated by different sets of process parameter values.