CMP Endpoint Detection Using AC Reference Signal Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The variability in polishing pad conditions, slurry variations, and material thickness changes in Chemical Mechanical Planarization (CMP) processes leads to high uncertainty in endpoint detection, affecting yield and process efficiency in semiconductor fabrication.
Innovation Solution
A method and system that generate a reference signal to filter noise from sensor signals, combining it with command signals to control the CMP system, allowing for reliable detection of friction property changes and determining the endpoint of the CMP process by analyzing operational characteristics.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If a predefined time endpoint detection method is used, then the CMP process can be controlled with simple timing, but the endpoint detection accuracy deteriorates due to high variability from polishing pad conditions, slurry variations, and material thickness changes
Solution Approach 1:
The patent applies feedback by continuously monitoring the friction properties of the semiconductor substrate during the CMP process and comparing the measured values against reference values. The system adjusts the endpoint detection based on the actual friction measurements, creating a closed-loop control system that improves accuracy while maintaining operational simplicity.
Solution Approach 2:
The patent replaces the purely time-based mechanical control system with an electrical measurement system that uses friction property sensors. This substitution allows for precise, real-time monitoring of substrate conditions, enabling accurate endpoint detection that accounts for variations in polishing pad conditions, slurry variations, and material thickness changes.
2Device complexity
If traditional time-based endpoint detection is used, then the process control is simple, but yield deteriorates due to increased tolerances from high variability
Solution Approach 1:
The feedback mechanism continuously monitors friction properties and provides real-time information about substrate conditions. This allows the system to detect endpoint conditions more reliably, reducing variability and improving yield without requiring complex manual intervention or overly complicated control systems.
Solution Approach 2:
The patent introduces friction property measurements as an intermediary parameter that mediates between the simple time-based control and the need for accurate endpoint detection. By using friction properties as an intermediate measurement, the system achieves improved reliability and yield while maintaining relatively simple process control architecture.
3Device complexity
If no noise filtering is applied to sensor signals, then the signal processing is simple, but endpoint detection reliability deteriorates due to noise in sensor signals
Solution Approach 1:
The patent extracts the relevant friction property signal from the noisy sensor output by applying noise filtering techniques. This separation of the useful signal from the noise allows for reliable endpoint detection while keeping the signal processing architecture relatively simple and focused on the critical measurement parameters.
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 precise and reliable endpoint detection in CMP processes, reducing variability and improving yield by filtering noise and accurately determining material exposure, thus enhancing process efficiency.
Implementation Method 1
generating a reference signal, generating a first signal with which to control a CMP system, generating a second signal using a combination of the first signal and the reference signal, commanding the CMP system with the second signal, generating a response signal that indicates an operational characteristic of the CMP system that is responsive to the second signal and a friction property of the semiconductor substrate, and filtering the response signal using the reference signal
Data Source
AI summary
Methods, non-transitory computer readable media, and systems are provided for detecting an endpoint of a chemical mechanical planarization (CMP) process on a semiconductor substrate. The method comprises generating a reference signal, generating a first signal with which to control a CMP system, generating a second signal using a combination of the first signal and the reference signal, commanding the CMP system with the second signal, generating a response signal that indicates an operational characteristic of the CMP system that is responsive to the second signal and a friction property of the semiconductor substrate, and filtering the response signal using the reference signal to determine the endpoint of the CMP process.