Semiconductor Etching Particle Source Identification and Targeted Cleaning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing semiconductor manufacturing technologies struggle to effectively reduce particles, which can lead to short circuits and mask etching issues, limiting yield and etching performance improvement.
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 platform connected via a network for automated particle reduction processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional etching recipes and cleaning methods are used, then etching performance (CD performance and etching rate uniformity) is maintained, but particle reduction is insufficient and yield improvement is limited
Solution Approach 1:
The system implements a closed-loop feedback mechanism where particle inspection results from wafers are fed back to the machine learning model, which then identifies particle sources in the etching apparatus. The cleaning process is guided by this feedback, and subsequent particle inspections verify the effectiveness, creating a continuous improvement cycle that systematically reduces particles while maintaining etching performance
Solution Approach 2:
The system enables the etching apparatus to self-diagnose particle sources and self-clean based on machine learning analysis of particle inspection data. The apparatus automatically identifies which components are generating particles and executes targeted cleaning procedures without requiring external intervention, thereby improving yield through autonomous particle reduction
2Reliability
If machine learning-based particle reduction processing is implemented, then particle number is reduced and yield is improved, but system complexity increases due to additional platform and network requirements
Solution Approach 1:
The platform performs multiple functions: it stores particle inspection data, trains machine learning models, identifies particle sources, generates cleaning recipes, and tracks yield improvement. By consolidating these diverse functions into a single multi-functional platform, the system reduces overall complexity compared to having separate systems for each function while achieving comprehensive particle reduction
3Object-generated harmful factors
If particle inspection and machine learning analysis are performed for every wafer, then particle reduction effectiveness is maximized, but processing time and productivity are reduced
Solution Approach 1:
The system performs complete particle inspection and machine learning analysis only when necessary to identify new particle sources or when particle levels exceed thresholds. For routine processing, the system uses pre-established cleaning recipes and monitoring, performing full analysis only when needed. This partial application of the full analysis process maintains particle detection accuracy while avoiding unnecessary processing delays
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.