E-Fuse Memory Repair via Compressed Bit String Encoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The unreliability of electrically programmed fuses (e-fuses) in programming memory elements on integrated circuits leads to inconsistent resistance changes, causing defects that can spoil the chip if uncorrectable, as they do not always change resistance as intended when programmed.
Innovation Solution
A design structure that programs fuses to replace defective memory elements by generating a binary string to indicate which spare memory elements replace defective ones, with the number of fuses programmed determined by the availability of spare fuses relative to the number of bits in the string, and using duplicate bit encoding to ensure fault tolerance by recording each bit in multiple fuses.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Volume of moving object
If e-fuses are used to replace metal fuses for memory repair, then device size is reduced and more fuses can be integrated on chip, but programming reliability deteriorates because e-fuses do not always change resistance as intended
Solution Approach 1:
The patent applies local quality by making different parts of the fuse system serve different functions: some fuses are dedicated to programming repair data while others are reserved for redundancy. This differentiation allows the system to maintain high reliability through redundancy while achieving compact size through efficient fuse utilization in critical areas.
Solution Approach 2:
The patent changes the parameter of fuse utilization by implementing a system where fuse states are not simply binary (programmed/unprogrammed) but can represent multiple states through combinatorial encoding. This allows more information to be stored in fewer fuses, improving reliability while reducing the total number of fuses needed on chip.
2Reliability
If more fuses are programmed to store binary string data, then programming reliability increases through redundancy, but the number of available fuses for other purposes decreases
Solution Approach 1:
The patent merges multiple functions into the fuse system: repair data storage, redundancy encoding, and status indication are all combined in a single integrated fuse array. This eliminates the need for separate fuse banks for each function, maximizing the utilization of available fuses while maintaining reliability through redundant encoding schemes.
Solution Approach 2:
The fuse system is designed to be universal and multi-functional, serving as both the primary storage medium for repair data and the redundancy mechanism simultaneously. The same physical fuses perform multiple roles through clever encoding, rather than requiring dedicated redundant fuse sets, thus preserving fuse quantity while enhancing reliability.
3Reliability
If duplicate bit encoding is used to increase programming reliability, then fault tolerance improves by ensuring correct bit state representation, but the number of fuses required increases
Solution Approach 1:
The patent applies dynamics by making the fuse configuration adaptable rather than static. The system can dynamically adjust which fuses are used for redundancy based on the actual number of defective memory elements detected during testing. This allows the redundancy level to match the actual need, avoiding waste of fuses when fewer repairs are required while maintaining fault tolerance when more repairs are needed.
Solution Approach 2:
The patent changes parameters by implementing a flexible encoding scheme where the number of fuses allocated to redundancy can be adjusted based on the repair requirements. Rather than fixed duplicate bit encoding that always uses the same number of fuses, the system adapts the encoding parameters to match the actual defect density, optimizing the balance between fault tolerance and fuse consumption.
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 increases the reliability of fuse programming by ensuring that even if one fuse fails to change state, the combination of duplicate fuses can still represent the intended bit state, thereby increasing chip yield and preventing malfunctions due to uncorrectable programming failures.
Implementation Method 1
The e-fuse has two intended logic states. To obtain a logical '1' state, an e-fuse is 'programmed' (also referred to as blown, opened, open circuited, etc.), such as by applying a pulse of 10 mA for 200 microseconds. This current disturbs the polysilicon link, dramatically increasing its resistance.
Data Source
AI summary
A design structure which enables e-fuse memory repair. The design structure uses a compressed bit string to generate another bit string based on a select value. The select value provides instructions to an encoding logic element, which generates a second bit string. For example, the select value may instruct the encoding logic to create a duplicate copy of each bit in the compressed bit string to generate a 2n-bit string. Once the fuses are programmed using the second bit string, the fuse values are read out as a third string, which is decoded by a decoding logic element according to the select value, thereby improving memory repair.