Nonvolatile Memory Read Operations Age-Based Variability
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The reliability of nonvolatile memory devices deteriorates with increased program/erase cycles due to coupling between memory cells, leading to data dispersion and non-uniformity, which limits their lifetime and read performance.
Innovation Solution
An access method that dynamically adjusts read conditions based on the number of erase cycles, using error correction codes to correct readable errors and employing reliability improvement operations only when necessary to maintain data accuracy and extend device lifetime.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If program/erase cycles are repeated to increase storage capacity utilization, then productivity is improved, but reliability deteriorates due to coupling between memory cells causing data dispersion and non-uniformity
Solution Approach 1:
The patent applies dynamics by making the read operation characteristics changeable based on the erase cycle count. The memory system dynamically adjusts read operations from normal reads to reliability improvement operations (such as coupled page reads) when dispersion is detected, allowing the system to adapt its behavior to the current state of the memory block and maintain reliability while utilizing repeated erase cycles.
Solution Approach 2:
The patent changes the parameter of read operation type based on the number of erase cycles and detected dispersion characteristics. By monitoring the erase cycle count and switching between different read modes (normal read vs. reliability improvement read), the system adjusts its operational parameters to compensate for degradation and maintain data accuracy throughout the device lifetime.
2Reliability
If reliability improvement operations are performed frequently to maintain data accuracy, then reliability is improved, but read performance deteriorates due to increased operation time
Solution Approach 1:
The patent applies partial action by performing reliability improvement operations only when necessary - specifically when the number of erase cycles exceeds a threshold or when dispersion is detected in specific memory blocks. Instead of applying reliability improvement operations to all reads continuously, the system selectively applies them only to affected blocks, thereby maintaining reliability where needed while minimizing the impact on overall read performance.
Solution Approach 2:
The memory system performs self-diagnosis by monitoring its own erase cycle count and detecting dispersion characteristics through read operations. Based on this self-assessment, the system automatically determines when reliability improvement operations are needed and executes them without external intervention, optimizing the balance between reliability and performance autonomously.
3Quantity of substance
If memory cell size is reduced to increase storage capacity, then quantity of substance is improved, but manufacturing precision deteriorates due to increased coupling between adjacent cells
Solution Approach 1:
The patent implements feedback by monitoring the actual read results and dispersion characteristics of memory cells after fabrication and during operation. Based on this feedback about cell uniformity and coupling effects, the system adjusts its read operations and reliability improvement strategies to compensate for the manufacturing variations inherent in densely packed small cells, thereby maintaining data accuracy despite reduced manufacturing precision.
Data Source
AI summary
Integrated circuit memory systems and methods include comparing a number of erase cycles of a memory block corresponding to a read request to a first value and reading data stored in the memory block according to a first read condition corresponding to a first reliability improvement operation when the number of erase cycles of the memory block is less than the first value. An error of the data read according to the first read condition may be corrected using an error correction code (ECC) when the error of the data read according to the first read condition is correctable.