Research And Optimization On The Reliability Of 3D Charge-Trap NAND Flash Memory

With the development and application of 5G,artificial intelligence,the Internet of Things and other technologies,the attention for large-capacity memory is increasing,and NAND flash memory has become an important non-volatile data storage medium due to its advantages such as large capacity and high reliability.In the pursuit of high storage density,two-dimensional(2D)planar NAND flash memory has reached the limit of scaling,and three-dimensional(3D)stacked NAND flash memory has replaced 2D NAND flash as the dominant storage media after more than ten years of development.However,the reliability mechanism of 3D NAND flash is more complex than that of 2D NAND flash due to its special 3D structure and process fabrication method,and how to optimize the reliability of 3D NAND flash by combining the characteristics of stored data is of great research significance to further improve data reliability.The first part of this thesis is the reliability characteristics of hot and cold data storage in 3D charge trap(CT)NAND flash memory.There are two types of data stored in NAND flash memory,hot data and cold data,distinguished by the frequency of read operation.For the storage of cold data,the experimental results show that errors generated in the storage states corresponding to the high threshold voltage(Vth)is the main reason for the increase of raw bit error rate(RBER)during data retention.Based on this result,this work reduces the number of high Vth storage states by bit flipping,and the experimental results verify the effectiveness of this scheme in suppressing the RBER of NAND flash.For the storage of hot data,this work studies the reliability characteristics of hot data storage by analyzing RBER and two types of error bits,analyzes the key mechanism of such reliability problems,and proposes pre charge storage layer(PCSL)and thermally stabilize storage layer(TSSL)schemes to improve the reliability of NAND flash memory and prolong its lifetime.The second part of this thesis is the systematic research on the correlation between reliability and data patterns in 3D CT NAND flash memory.For the reliability research of cold data storage in different data patterns,it is find that there is a strong correlation between the degradation of NAND flash reliability and data patterns through data retention test,and it becomes more severe after the program/erase(P/E)cycling,which is mainly due to the fact that the severity of lateral charge migration is influenced by different data patterns.After further investigation of the read disturb characteristics in different data patterns,it is find that the increase in the Vth difference of the stored data in two adjacent layers accelerates the degradation of the read interference characteristics,which is not directly correlated with whether the data was cold before the read disturb test.For cold data after long retention,the repair effect of read interference on error bits remains.In addition,errors caused by Vth right-shift in NAND flash after P/E cycling increase more significantly during hot data storage,and the number of such errors is positively correlated with Vth difference.Based on the characteristics of the data types and data patterns,this thesis presents a targeted research on reliability optimization strategies for 3D CT NAND flash memory and proposes a variety of schems that can effectively improve storage reliability.For the storage of cold data,the RBER can be reduced by more than 50%using the bit-flip method.For the storage of hot data,the RBER can be reduced by more than 20%using both PCSL and TSSL schemes.This study provides important experimental data and different strategies for coding optimization in combination with data characteristics for an in-depth understanding of the reliability characteristics of 3D CT NAND flash memory,which is important for further improving the reliability of 3D CT NAND flash memory.