Machine Learning Scatterometry for LCDU Measurement
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for measuring local critical dimension uniformity (LCDU) in two-dimensional arrays of structural elements, such as vias, are inaccurate and time-consuming due to limitations in CD-SEM measurements, which can damage samples and only allow for a few high-magnification measurements, reducing the accuracy and yield of semiconductor manufacturing processes.
Innovation Solution
A method utilizing machine learning processes to map optical spectrometry spectra of two-dimensional structural elements to average CD and LCDU values, trained on data from CD-SEM measurements under various manufacturing conditions, allowing for faster and more accurate determination of LCDU across a significant number of elements, including outlier removal to enhance precision.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If CD-SEM is used to measure individual structural elements at high magnification, then measurement precision is improved, but productivity deteriorates due to the lengthy process and limited number of elements that can be measured
Solution Approach 1:
The patent segments the measurement task by using optical spectrometry to rapidly screen multiple structural elements and identify outliers, then applying CD-SEM only to those specific outlier elements for precise measurement. This divides the measurement process into two stages: rapid optical screening followed by targeted electron microscopy, resolving the contradiction between measuring many elements quickly and maintaining high precision.
Solution Approach 2:
The patent introduces optical spectrometry as an intermediary method between direct CD-SEM measurement and final LCDU determination. The optical spectrometry provides a rapid, non-destructive screening that identifies which elements require detailed CD-SEM analysis, acting as a mediator that filters the large number of elements down to a manageable subset for high-precision measurement.
2Measurement precision
If CD-SEM is used for high-magnification measurement, then measurement precision is improved, but object-affected harmful factors worsen due to sample damage
Solution Approach 1:
The measurement process is segmented into two stages: a non-destructive optical spectrometry screening stage that identifies outlier elements, followed by a targeted CD-SEM measurement stage applied only to those specific outliers. This segmentation ensures that the harmful high-magnification electron beam is applied to minimal elements, reducing cumulative sample damage while maintaining measurement precision for critical elements.
Solution Approach 2:
Instead of applying CD-SEM to all structural elements (excessive action), the patent applies it only partially to the specific outlier elements identified by optical spectrometry. This partial application of the high-magnification technique minimizes sample damage while still achieving the necessary measurement precision for elements that deviate from specifications.
3Productivity
If optical spectrometry is used to measure multiple structural elements, then productivity is improved, but measurement precision deteriorates compared to model-based methods
Solution Approach 1:
The patent segments the measurement approach by using optical spectrometry for rapid identification of outlier elements and combining this with CD-SEM measurements of those specific outliers. This segmented approach achieves both high productivity (through optical screening of many elements) and high precision (through CD-SEM measurement of identified outliers), resolving the contradiction between measuring multiple elements and maintaining accuracy.
Solution Approach 2:
The patent implements a feedback mechanism where optical spectrometry measurements inform subsequent CD-SEM measurement decisions. The optical data provides feedback about which elements are outliers, directing the high-precision CD-SEM measurements to those specific elements. This feedback loop enables the system to achieve both high throughput and high accuracy by dynamically allocating measurement resources based on actual element characteristics.
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 provides a more accurate and faster method for measuring LCDU compared to traditional model-based optical spectrometry, improving the accuracy and efficiency of semiconductor manufacturing by enabling the measurement of hundreds or thousands of structural elements without sample damage.
Implementation Method 1
obtaining an acquired optical spectrometry spectrum of the array of two-dimensional (2D) structural elements
Data Source
AI summary
A method, a system, and a non-transitory computer readable medium for measuring a local critical dimension uniformity of an array of two-dimensional structural elements, the method may include obtaining an acquired optical spectrometry spectrum of the array; feeding the acquired optical spectrometry spectrum of the array to a trained machine learning process, wherein the trained machine learning process is trained to map an optical spectrometry spectrum to an average critical dimension (CD) and a local critical dimension uniformity (LCDU); and outputting, by the trained machine learning process, the average CD and the LCDU of the array.