Spectral Filtering for Semiconductor Optical Signal Anomaly Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current optical monitoring systems for semiconductor processes face challenges in accurately detecting small changes in optical emission spectroscopy (OES) and interferometric endpoint (IEP) signals due to mixed signal types, which can mask each other's changes, and require wider optical bandwidth and greater signal-to-noise ratios, especially for advanced processes like FINFET and 3D NAND structures.
Innovation Solution
The system employs improved spectral data processing to identify and modify anomalous signals in both analog and digital domains, using algorithms and processing capabilities within the spectrometer or signal processor to distinguish and correct for anomalous events, thereby providing a more accurate representation of optical data for monitoring semiconductor processes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If wider optical bandwidth is used to detect small changes in OES and IEP signals, then measurement precision is improved, but signal-to-noise ratio deteriorates
Solution Approach 1:
The patent segments the optical signal processing into distinct spectral bands using multiple spectrometers, each optimized for specific wavelength ranges (e.g., UV for OES, visible/IR for IEP). This segmentation allows each spectrometer to focus on detecting small changes in its designated band while maintaining adequate signal-to-noise ratio through specialized optimization.
Solution Approach 2:
The patent introduces temporal dimension by collecting spectral data at multiple time points and comparing changes over time. By analyzing the temporal evolution of spectral features rather than relying on absolute intensity measurements, the system can detect small changes with improved precision while being less sensitive to noise and signal intensity variations.
2Adaptability or versatility
If mixed signal types (OES and IEP) are monitored simultaneously, then adaptability is improved, but measurement precision deteriorates due to signal masking
Solution Approach 1:
The patent physically segments different signal types into separate spectral detection channels using multiple spectrometers with different wavelength ranges. OES signals are captured in the UV range while IEP signals are captured in the visible/IR range, preventing signal masking and allowing precise detection of small changes in each signal type independently.
Solution Approach 2:
The patent introduces spectral filtering as an intermediary mechanism that separates mixed optical signals into distinct wavelength bands before detection. This filtering approach allows simultaneous monitoring of multiple process types while maintaining the ability to detect small changes in each signal type without interference from the other.
3Reliability
If spectral filtering is applied to improve signal-to-noise ratio, then reliability is improved, but device complexity increases
Solution Approach 1:
The patent merges the spectral filtering function with the spectrometer detection system itself, rather than adding separate filtering components. By using the inherent spectral resolution of multiple spectrometers and processing spectral data computationally, the system achieves improved signal-to-noise ratio while avoiding the complexity of additional physical filtering hardware.
Solution Approach 2:
The patent replaces complex mechanical spectral filtering systems with computational spectral analysis. By processing spectral data through algorithms that identify and enhance relevant features while suppressing noise, the system achieves reliable signal detection without the mechanical complexity of tunable filters or multiple optical paths.
Data Source
AI summary
The disclosure provides improved processing of optical data by identifying anomalous signals in the electrical data representing the optical data. The improved processing can also include modifying the identified anomalous signal data to provide a truer representation of the optical data. The disclosed processing can be used by various systems and apparatuses for processing spectral data corresponding to the optical data. The improved processing can be used to improve the monitoring of semiconductor processes and, therefore, improve the overall semiconductor processes. In one example, a method of processing spectral data includes: (1) receiving temporally separated spectral data samples, and (2) identifying one or more anomalous signals in an intermediate one of the temporally separated spectral data samples based on at least one preceding and at least one subsequent ones of the spectral data samples.