Semiconductor Particle Source Detection for Targeted Chamber Cleaning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing semiconductor manufacturing technologies struggle to effectively reduce particles during the manufacturing process, leading to potential short circuits and masking issues, which hinder the formation of desired etching shapes and reduce yield.
Innovation Solution
A semiconductor device manufacturing system and method utilizing machine learning to analyze particle characteristics, identify particle sources, and define cleaning conditions to reduce particles, incorporating a semiconductor manufacturing apparatus, a particle measurement apparatus, and a network platform for automated particle reduction processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If conventional etching processing is performed without particle reduction measures, then etching performance (CD performance and etching rate uniformity) can be maintained, but particles adhere to the wafer causing short circuits and masking issues
Solution Approach 1:
The system performs preliminary particle detection and source identification before particle adhesion occurs. By using machine learning to analyze particle characteristic values and correlate them with apparatus components, the system proactively identifies particle sources and executes cleaning operations before particles can adhere to wafers and cause defects
Solution Approach 2:
The system establishes a closed-loop feedback mechanism where particle characteristic values are continuously measured, analyzed by machine learning algorithms to identify sources, and used to trigger appropriate cleaning actions. The correlation data between particle characteristics and apparatus components enables the system to learn from past particle events and improve particle reduction effectiveness over time
2Object-affected harmful factors
If machine learning-based particle reduction processing is implemented, then particle adhesion is reduced, but system complexity increases due to additional processing steps
Solution Approach 1:
The system enables the semiconductor manufacturing apparatus to self-diagnose particle sources and self-correct by automatically executing cleaning operations. The machine learning algorithm analyzes particle characteristic values and autonomously identifies particle sources without requiring external intervention, and the system automatically implements cleaning based on the identified sources
Solution Approach 2:
The system replaces manual particle source identification and cleaning decision-making with machine learning-based automated analysis. Instead of relying on operators to manually correlate particle characteristics with apparatus components, the machine learning algorithm automatically processes particle characteristic values and identifies sources, reducing operational complexity
3Object-affected harmful factors
If cleaning operations are performed frequently to reduce particles, then particle adhesion decreases, but productivity decreases due to increased processing time
Solution Approach 1:
Instead of performing comprehensive cleaning operations frequently, the system applies partial cleaning actions targeted specifically at identified particle sources. The machine learning algorithm enables precise localization of particle-generating components, allowing cleaning to be focused only on the necessary areas rather than entire chambers, thereby reducing cleaning time and minimizing impact on productivity
Data Source
AI summary
The invention is to provide a semiconductor manufacturing apparatus system and a semiconductor device manufacturing method for reducing particles having an adverse effect in a manufacturing step of a semiconductor device. A semiconductor device manufacturing system, includes: a semiconductor manufacturing apparatus; and a platform connected to the semiconductor manufacturing apparatus via a network and in which a particle reduction processing is executed, in which the particle reduction processing includes: a step of acquiring a particle characteristic value by using a sample processed by the semiconductor manufacturing apparatus; a step of specifying a component of the semiconductor manufacturing apparatus leading to a particle generation based on the acquired particle characteristic value and correlation data by machine learning; a step of defining a cleaning condition for cleaning the semiconductor manufacturing apparatus based on the specified component; and a step of cleaning the semiconductor manufacturing apparatus using the defined cleaning condition, and the correlation data is correlation data between the particle characteristic value acquired in advance and the component.