RF Sensor Calibration Using Virtual Metrology for Chamber Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Chamber-to-chamber variability in RF sensor measurements due to impedance mismatches and sensor-related variabilities leads to inaccurate and time-consuming chamber matching, resulting in inconsistent plasma processing across multiple chambers, which affects device performance and yield.
Innovation Solution
A method and apparatus for chamber matching using virtual metrology models to calibrate RF sensors by adjusting current measurements based on sputter rates and control models, ensuring consistent plasma processing across multiple chambers.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional chamber matching methods are used, then chamber-to-chamber variability is addressed, but the process is time-consuming and labor-intensive
Solution Approach 1:
Virtual metrology models are built in advance using historical process data and machine learning algorithms. These pre-trained models can predict chamber parameters and performance characteristics without requiring real-time manual measurement and adjustment, thus eliminating the time-consuming aspect of traditional chamber matching while maintaining accuracy
Solution Approach 2:
The patent replaces manual mechanical chamber matching procedures with automated computational models. Machine learning algorithms analyze process data to predict and adjust chamber parameters, substituting the mechanical adjustment process with an intelligent software-based system that operates automatically
2Measurement precision
If RF sensors are used for chamber matching, then measurement capability is provided, but impedance mismatches and sensor variabilities cause measurement inaccuracies
Solution Approach 1:
The system continuously monitors process parameters using RF sensors and feeds this data back to the virtual metrology models. The models analyze the feedback to detect drift or inconsistencies in sensor readings and automatically adjust chamber parameters or compensate for sensor variabilities, maintaining measurement reliability over time
Solution Approach 2:
The patent employs multiple RF sensors operating at different frequencies to measure chamber parameters. By changing the measurement parameter (frequency) and comparing results across multiple sensors, the system can identify and compensate for impedance mismatches and individual sensor variabilities, improving overall measurement accuracy and consistency
3Productivity
If multiple chambers are used for plasma processing, then productivity is increased, but device performance consistency deteriorates due to chamber variability
Solution Approach 1:
The virtual metrology models are designed to be chamber-agnostic, working across multiple different plasma processing chambers. The same software platform and algorithmic approach can be applied universally to any chamber, enabling consistent performance prediction and control across the entire fleet of chambers, thus maintaining device performance consistency while utilizing multiple chambers for increased productivity
Solution Approach 2:
Instead of physically matching and adjusting each chamber to identical specifications through manual procedures, the patent uses computational models to create a virtual reference framework. The models predict the behavior of each chamber based on its unique characteristics and automatically determine the optimal settings for each chamber, replacing mechanical adjustment with intelligent software control that ensures consistent results across all chambers
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
Enhances device performance consistency and reliability by minimizing device variation across different processing chambers through precise calibration of RF sensors, improving yield and reducing the time and cost of chamber matching.
Implementation Method 1
a first voltage in the first chamber measured by a first radio frequency (RF) sensor, a second voltage in the second chamber measured by a second RF sensor
Implementation Method 2
The plasma recipe includes one or more inert gases and no reactive gases
Implementation Method 3
executing the plasma recipe in a first chamber. A second VM model that is configured to predict a second TCR for executing the plasma recipe in a second chamber
Implementation Method 4
both the first TCR and the second TCR are sputter rates on blanket films
Data Source
AI summary
A method of chamber matching is provided. A first virtual metrology (VM) model is configured to predict a first thickness change rate (TCR) for executing a plasma recipe in a first chamber. A second VM model is configured to predict a second TCR for executing the plasma recipe in a second chamber. The plasma recipe includes one or more inert gases and no reactive gases. A control model describes relationships between voltage, current and recipe parameters. In reference to a first voltage in the first chamber measured by a first radio frequency (RF) sensor, a second voltage in the second chamber measured by a second RF sensor is calibrated. In reference to a first current in the first chamber measured by the first RF sensor, a second current in the second chamber measured by the second RF sensor is calibrated.