A 45nm 6b/cell charge-trapping flash memory using LDPC-based ECC and drift-immune soft-sensing engine

ISSCC(2013)

引用 47|浏览11
暂无评分
摘要
To satisfy the demand of higher storage density, storing multiple-bits-per-cell technique is widely adopted. As presented in [1], a 4b/cell Flash memory by using error-detection (ED) scheme stores 2b data on two sides of a memory cell individually. Since the noise margin becomes smaller, the distribution drifts due to program disturb, data retention and temperature variation will cause higher raw bit-error-rate (RBER) if the sensing level (voltage) are not adjusted accordingly. The ED scheme can detect the drift direction by counting and storing the number of cells (Ni) with threshold voltage (VTH) below the ith sensing level (VREF i). A simple example with page size 1KB is demonstrated in Fig. 12.7.1. During a read operation, the number of cells (Ni, measured) whose VTH below VREF i is counted and compared to Ni. The ED scheme can find out a sub-optimal sensing level when Ni, measured and Ni is close enough. In this paper, a production 16Gb 45nm 4b/cell ONO-based charge-trapping (CT) Flash memory is demonstrated to achieve 6b/cell capability. Since the adjacent distributions for 6b/cell are much closer to each other, even a BCH code with ED scheme fails to correct all the patterns. However, by using a new 1-3-3 mapping and LDPC codes with a developed drift-immune soft-sensing (DI-SS) engine, the 45nm 4b/cell CT Flash memory is boosted to 6b/cell. The data flow of programming data is also shown in Fig. 12.7.1.
更多
查看译文
关键词
di-ss engine,error-detection scheme,data retention,drift-immune soft-sensing engine,rber,size 45 nm,6b-cell charge-trapping flash memory,error detection codes,ldpc-based ecc,raw bit-error-rate,suboptimal sensing level,bch code,noise margin,ono-based charge-trapping,multiple-bits-per-cell technique,ct,drift direction,read operation,data flow,threshold voltage,distribution drifts,temperature variation,ed scheme,error statistics,parity check codes,program disturb,programming data,flash memories
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要