Metrology Targets Using Machine Learning for Overlay
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current metrology technologies face challenges in accurately measuring device parameters due to multi-layered device designs, non-periodic nature of devices, and the need to avoid introducing offsets that could damage electrical properties, limiting direct device measurements and leading to inaccurate overlay calculations.
Innovation Solution
The implementation of machine learning algorithms to calibrate sensitivity using orthogonal, non-critical direction offsets and additional target cells, combined with multi-layered metrology targets and quasi-periodic target designs, enables direct measurement of device parameters without damaging electrical properties, allowing for more efficient and accurate overlay calculations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If offsets are introduced into target cells for overlay measurement, then overlay sensitivity is improved, but electrical properties of the device are damaged
Solution Approach 1:
The patent uses intermediary target cells that are adjacent to the device portion but electrically isolated from it. These intermediary cells contain the periodic structures with offsets, serving as mediators that enable overlay sensitivity measurement without directly affecting the device's electrical properties. The machine learning algorithm transfers sensitivity calibration from these intermediary cells to the device portion.
Solution Approach 2:
The patent creates target cells that are copies or replicas of the device portion, containing the same periodic structures but with introduced offsets. These copied structures serve as surrogates for measuring overlay sensitivity, allowing the device itself to remain unchanged and electrically functional while still enabling precise measurement through the copied structures.
2Measurement precision
If multiple target cells are used for sensitivity calibration, then measurement accuracy is improved, but the number of required cells and measurement time increase
Solution Approach 1:
The patent performs sensitivity calibration in advance using target cells with known offsets, storing the calibrated sensitivity values for later use. This preliminary calibration action eliminates the need to repeatedly measure multiple cells during actual device measurement, significantly reducing measurement time while maintaining high accuracy through the pre-established sensitivity database.
Solution Approach 2:
The patent uses a limited set of target cells with specific offsets that are sufficient for calibration purposes, rather than measuring all possible cell combinations. By selecting only the necessary number of target cells with representative offsets, the system achieves adequate calibration accuracy without the excessive time consumption that would result from comprehensive multi-cell measurements.
3Ease of manufacture
If periodic structures are used in metrology targets, then measurement algorithms are simplified, but the non-periodic nature of actual devices cannot be accurately represented
Solution Approach 1:
The patent changes the parameters of the periodic structures in target cells, such as pitch, duty cycle, and layer thickness, to better match the actual device characteristics. By adjusting these parameters, the periodic target structures can more accurately represent the non-periodic device features while still maintaining the mathematical simplicity of periodic structure analysis algorithms.
Solution Approach 2:
The patent introduces asymmetric features into the periodic target structures, such as varying the offset magnitudes in different target cells or using non-uniform periodic patterns. This controlled asymmetry allows the targets to better represent the asymmetric nature of actual devices while still enabling the use of simplified periodic structure measurement algorithms through appropriate mathematical modeling.
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 efficient, economical, and accurate measurement of metrology parameters, including overlays, on multi-layered and non-periodic device structures without introducing offsets that could harm the devices, enhancing measurement precision and reducing the number of required cells and measurement time.
Implementation Method 1
measuring, scatterometrically, at least M differential signals from the multi-layered metrology target
Data Source
AI summary
Metrology methods and targets are provided, that expand metrological procedures beyond current technologies into multi-layered targets, quasi-periodic targets and device-like targets, without having to introduce offsets along the critical direction of the device design. Machine learning algorithm application to measurements and/or simulations of metrology measurements of metrology targets are disclosed for deriving metrology data such as overlays from multi-layered target and corresponding configurations of targets are provided to enable such measurements. Quasi-periodic targets which are based on device patterns are shown to improve the similarity between target and device designs. Offsets are introduced only in non-critical direction and/or sensitivity is calibrated to enable, together with the solutions for multi-layer measurements and quasi-periodic target measurements, direct device optical metrology measurements.