An improved programming algorithm to reduce faults with minimal additional time penalty

摘要

Programming time of the non-volatile memory in a multi-pass programming process is shortened. In a first programming pass, while it is programmed by a sequence of program pulses, the cell of the low state is locked out of programming cells of the high state to identify the cell of the high state and slow cells with high speed and high that. While being temporarily locked out from programming, cells with high fast identified continues to the programmed state of the final object of these cells with high-low speed. Furthermore, the program pulse is stepped up sharply in order to program a cell of high state of slow. In a second programming pass, the cell of the high state of fast, is programmed with cells with other low until it reaches a state where they are all intended. Time savings can be attained as compared with the approach cells with high all be programmed in the first programming pass. [Selection] Figure Figure 1