Non-Volatile Memory Device Encoding Back Pattern Dependency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Multi-level cell (MLC) flash memory devices face errors due to back pattern dependency (BPD), which widens the threshold voltage distribution and narrows the margin between voltage levels, leading to over-programmed or under-programmed memory cells and potential disturbance of unintended cells during programming.
Innovation Solution
A non-volatile memory device employs an encoder and decoder circuit to perform XOR operations on input data and randomly generated data, stored per word line, to encode data for programming and decode data read from memory cells, minimizing the impact of peripheral programmed cells by using random data to control the programming and reading processes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If pass voltage is increased to reduce back pattern dependency, then programming accuracy is improved, but disturbance to unintended memory cells increases
Solution Approach 1:
The patent applies different voltages to different word lines based on their selection status: selected word lines receive program voltage while non-selected word lines receive a lower pass voltage. This local differentiation of voltage levels prevents disturbance to unintended cells while maintaining programming accuracy for selected cells.
Solution Approach 2:
The patent changes the voltage parameter dynamically based on the programming state and word line selection. By adjusting the pass voltage level according to whether cells are being programmed or not, the system optimizes between reducing back pattern dependency and preventing disturbance to adjacent cells.
2Manufacturing precision
If memory cells are programmed sequentially to reduce back pattern dependency, then threshold voltage distribution is improved, but programming time increases
Solution Approach 1:
The patent implements periodic programming cycles where groups of word lines are programmed in sequence rather than all at once. This periodic approach allows threshold voltage distribution to be maintained while reducing total programming time compared to strictly sequential programming of all cells.
Solution Approach 2:
The patent performs preliminary programming of certain word lines before others, establishing a pattern that reduces back pattern dependency. By pre-programming specific regions or word lines, the system prepares the memory structure to minimize interference effects during subsequent programming operations.
3Quantity of substance
If multi-level cells are used to enhance integration, then storage density is improved, but error rate due to back pattern dependency increases
Solution Approach 1:
The patent applies local quality control by programming different word lines with different patterns and voltages. This localized approach ensures that multi-level cells in different regions experience different programming conditions, reducing the impact of back pattern dependency on overall error rate while maintaining high storage density.
Solution Approach 2:
The patent implements feedback mechanisms through read-verify-program cycles. After programming multi-level cells, the system reads back the data to verify correctness, and if errors are detected due to back pattern dependency, additional programming cycles are performed to correct them.
Data Source
AI summary
A non-volatile memory device is disclosed. The non-volatile memory device includes an encoder configured to set random data in a unit of a word line, and generate second data to be programmed in a memory cell by performing logic operation about the random data and first data inputted for program, and a data converting circuit configured to have a decoder for generating the first data by performing logic operation about the second data read from the memory cell and the random data.