Server Fault Detection Using Product Wafer Number Filtering

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional group management systems for semiconductor manufacturing apparatuses fail to accurately detect faults due to high non-uniformity in measurement information when using the total number of wafers, including dummy wafers, as a key, leading to wide threshold values and inaccurate fault detection.

Innovation Solution

A server device within the group management system stores time sequential measurement information and includes a fault detection unit that filters data using the number of product wafers as a key, allowing for precise fault detection by composing output information based on product wafer number conditions and condition information.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Quantity of substance

If the total number of wafers including dummy wafers is used as a key for fault detection, then all measurement information is covered, but the non-uniformity of measurement information increases and fault detection accuracy decreases

Engineering Contradiction:
Improvecoverage of measurement informationVSAvoidfault detection accuracy
Core Design Contradiction:
Quantity of substanceVSMeasurement precision

Solution Approach 1:

The patent segments the total wafer quantity into two distinct components: product wafer number (for actual manufactured wafers) and dummy wafer number (for placeholder wafers). This segmentation allows the system to use product wafer number as the key for fault detection, thereby excluding dummy wafers from the calculation and reducing non-uniformity in measurement information while maintaining comprehensive coverage of relevant data.

Inventive Principle:
Principle #1Segmentation

2Quantity of substance

If the total number of wafers including dummy wafers is used as a key, then complete data set is analyzed, but wide threshold values are needed and fault detection becomes less accurate

Engineering Contradiction:
Improvedata set completenessVSAvoidfault detection reliability
Core Design Contradiction:
Quantity of substanceVSReliability

Solution Approach 1:

The patent extracts and separates dummy wafers from the total wafer count, using only the product wafer number as the key for fault detection. This extraction removes the interfering element (dummy wafers) that causes non-uniformity, allowing for narrower and more precise threshold values while maintaining analysis of the complete relevant data set.

Inventive Principle:
Principle #2Taking out (Extraction)

3Quantity of substance

If dummy wafers are included in wafer count, then gas flow resistance varies due to wiring differences, but fault detection cannot distinguish between normal and abnormal variations

Engineering Contradiction:
Improvetotal wafer countVSAvoidfault detection discrimination
Core Design Contradiction:
Quantity of substanceVSDifficulty of detecting and measuring

Solution Approach 1:

The patent applies local quality by treating product wafers and dummy wafers differently in the fault detection process. Specifically, it assigns different characteristics to these two types: product wafers have wiring patterns that affect gas flow resistance, while dummy wafers do not. By using product wafer number as the key, the system locally optimizes the detection approach for actual product wafers, enabling accurate discrimination between normal and abnormal variations in gas flow resistance.

Inventive Principle:
Principle #3Local quality

Data Source

PatentUS8355808B2Server device of group management system having function of performing fault detection and program
Publication Date: 2013.01.15 TOKYO ELECTRON LTD
  • US8355808B2 patent drawing
  • US8355808B2 patent drawing
  • US8355808B2 patent drawing

AI summary

It is possible to compose a chart by filtering measurement information using a product wafer number in a server device including: an instruction receiving unit for storing therein plural measurement information, which is time sequential information measured in a plurality of manufacturing apparatuses and has a product wafer number and time information, and for receiving an output instruction of a chart containing the product wafer number; a fault detection unit for, when the instruction receiving unit receives the output instruction, reading a multiplicity of measurement information satisfying a product wafer number condition contained in the output instruction, and determining whether the read multiplicity of measurement information satisfies the condition information; an output information composing unit for composing the output information according to a determination result of the fault detection unit; and an output unit for outputting the output information composed by the output information composing unit.