Non-Volatile Memory Programming for Narrow Threshold Distributions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Non-volatile memory systems face challenges in reducing errors during data reading due to wide threshold voltage distributions, which can be exacerbated by varying programming speeds and inefficiencies in memory cell programming techniques.
Innovation Solution
The implementation of a dual-step programming technique that slows down programming after reaching an intermediate condition, using a program slowing voltage followed by a program inhibit voltage, to narrow threshold voltage distributions and enhance data reading accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If programming is performed at high speed, then productivity is improved, but threshold voltage distribution becomes wider which reduces reliability
Solution Approach 1:
The programming process is divided into multiple programming passes, each applying a controlled amount of programming pulse. Between passes, verification is performed to check if cells have reached the target state. This segmentation allows the system to accumulate programming effect gradually while monitoring and controlling the threshold voltage distribution, preventing excessive widening that would occur with single high-speed programming attempts.
Solution Approach 2:
The programming process uses periodic verification steps interspersed between programming pulses. After each programming pass, a verification operation checks the threshold voltage of programmed cells. This periodic action allows the system to pause programming when the distribution becomes too wide, perform verification, and adjust subsequent programming parameters, thereby maintaining reliability while achieving overall programming productivity.
2Productivity
If programming voltage is increased to speed up programming, then productivity is improved, but errors in data reading increase due to wider threshold voltage distributions
Solution Approach 1:
The programming voltage amplitude is made dynamic rather than fixed. The system starts with higher voltage amplitudes to achieve fast initial programming, then progressively reduces the voltage amplitude in subsequent passes. This dynamic adjustment allows aggressive programming initially when distributions are narrow, then gentler programming as distributions widen, thereby maintaining reading accuracy while achieving overall productivity.
Solution Approach 2:
Verification operations provide feedback about the threshold voltage state of programmed cells. The verification results inform whether additional programming passes are needed and at what voltage levels. This feedback mechanism allows the system to adapt programming parameters based on actual cell states, preventing over-programming that would widen distributions and reduce reading accuracy, while ensuring complete programming when needed for productivity.
Data Source
AI summary
To reduce errors when reading programmed data, a non-volatile memory programs data to narrower threshold voltage distributions by slowing down programming using a first technique for slowing programming after a memory cell reaches an intermediate condition and using a second technique for slowing programming after using the first technique and prior to inhibiting programming for the memory cell. In one embodiment, the first technique comprises applying a program slowing voltage to a bit line connected to the non-volatile memory cell while applying one dose of programming and the second technique comprises applying to the bit line the program slowing voltage for a first portion of a dose of programming and a program inhibit voltage for a second portion of the dose of programming.