Chip Information Management via Defective Address Pattern Storage
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for managing chip information, such as Japanese Patent No. 3,555,859, face challenges in efficiently obtaining and storing position information on a semiconductor wafer after it is divided into chips, leading to large databases that hinder efficient information sharing and chip differentiation when defective address distributions are similar.
Innovation Solution
A chip information management method that stores and reads defective addresses using a memory cell array with redundant memory cells and defective address storing circuits, employing a data table to allocate information based on the relationship between defective addresses and storing circuits, without increasing chip area or using a large database.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If defective addresses are stored in a database for each semiconductor chip, then position information on the wafer can be obtained, but the database becomes huge and searching time increases
Solution Approach 1:
The invention extracts only the essential information (defective address distribution pattern) from each chip and stores it in a compact form in the database, rather than storing all raw defective address data. This extraction approach maintains the ability to retrieve position information while significantly reducing database size and search time.
Solution Approach 2:
Instead of storing complete defective address lists and searching through them, the invention inverts the approach by storing pre-processed distribution patterns that can be quickly matched. This inversion transforms a time-consuming search problem into a faster pattern-matching problem.
2Ease of operation
If individual chips are specified based on defective address distribution, then chip information can be obtained, but chips with identical distributions cannot be differentiated
Solution Approach 1:
The invention applies local quality by combining the global defective address distribution pattern with local unique identifiers specific to each chip. This allows chips with similar defect patterns to be differentiated through their unique local characteristics, ensuring both ease of retrieval and chip differentiation.
Solution Approach 2:
The chip identification system uses a composite approach by combining distribution pattern information with unique chip identifiers. This composite information structure enables both efficient retrieval based on defect patterns and differentiation of individual chips, even when their defect distributions are identical.
3Loss of information
If operation tests are performed on each chip before dicing to obtain position information, then defect position can be specified, but manufacturing cost increases
Solution Approach 1:
The invention performs preliminary action by conducting wafer-level operation tests before dicing, rather than testing each individual chip. This preliminary testing approach obtains defect position information at the wafer level, which is then correlated to individual chips, significantly reducing testing costs while maintaining the ability to specify defect positions.
Solution Approach 2:
The wafer-level testing approach serves multiple functions: it identifies defect positions, determines defective address distributions, and provides information applicable to all chips on the wafer. This multi-functionality eliminates the need for separate individual chip testing, reducing manufacturing costs while preserving defect position information.
Data Source
AI summary
In replacing word lines having defective addresses with redundant word lines, information is held in a relationship between the word lines and the redundant word lines. In other words, information is held in a replacement rule. With this arrangement, information such as a lot number, a wafer number within the lot, and a position of a chip within the wafer can be held in the chip, without increasing the chip area at all and without using a large database.