Program algorithm with staircase waveform decomposed into multiple passes
View Patent ↗Programming algorithms suitable for non-volatile memory devices are presented, where the usual staircase type of waveform is decomposed into multiple passes. The same pulses are used, but their order is different, being broken down into N subsets of every N-th pulse so that there are N programming passes. For example, in a four pass version the first pass would sequentially have pulses (1, 5, 9, . . . ) of the staircase; the second pass would have pulses (2, 6, 10, . . . ); and so on for the third and fourth passes. By using a large step size for stepping up the program voltage V PGM , pre-verify and shifted verify margins for the different passes, a narrow distribution can be obtained with multiple passes. Also, in a complementary set of features, a quick pass write (QPW) can be done without the drawback of doing QPW verify by the use of shared verify levels.
1. A method of programming a non-volatile memory device, comprising:
applying pulses forming a staircase of increasing amplitude to a first wordline;
after applying each pulse to the first wordline, performing a verifying operation for memory cells along the first wordline,
wherein the series of pulse is applied to the first wordline as a series of a plurality of N subsets of the staircase, the series including at least a first subset and a second subset, wherein the first subset includes the first and every Nth subsequent pulse of the staircase applied sequentially in order of increasing amplitude, the second subset includes the second and every Nth subsequent pulse of the staircase applied sequentially in order of increasing amplitude, and the second subset is applied subsequent to applying the first subset;
locking out the memory cells along the first wordline from further programming in response to verifying at the subset's verify level corresponding to the cells' target data state, wherein each subset uses a different set of verify levels for verify operations; and
subsequent to applying the first subset and prior to applying the second subset, unlocking the memory cells along the first wordline to allow further programming.
2. The method of claim 1 , further comprising:
subsequent to unlocking the memory cells to allow further programming and prior to applying the second subset, performing a pre-verify operation; and
locking from programming by the second subset of those memory cells that pre-verify at a pre-verify level corresponding to the cells' target data state.
3. The method of claim 1 , wherein the non-volatile memory stores data in a multi-state format and the verify operation verifies the memory cells along the first wordline for a plurality of the data states.
4. The method of claim 1 , wherein the first subset is a foggy programming process, wherein, subsequent to applying the first subset but prior to applying the second subset, the distribution of memory cells are not fully resolved into distinct multi-states.
5. The method of claim 1 , further comprising:
for at least one of the first and second subsets, prior to the first pulse, performing an initial verifying operation using a set of initial verify levels that, for each target data state, are lower than the corresponding verify level used for locking out memory cells from further programming; and
altering the bias on the memory cells to program more slowly in response to verifying at the subset's initial verify level corresponding to the cells' target data state.
6. The method of claim 5 , further comprising:
for at least one of the first and second subsets, after each pulse, performing the initial verifying operation using a set of initial verify levels that, for each target data state, are lower than the corresponding verify level used for locking out memory cells from further programming; and
altering the bias on the memory cells to program more slowly in response to verifying at the subset's initial verify level corresponding to the cells' target data state.
7. The method of claim 5 , wherein the non-volatile memory stores data in a multi-state format and the verify operation verifies the memory cells along the first wordline for a plurality of the data states.
8. The method of claim 7 , wherein an initial verify level for a first of the multi-states is the same as the verify level used for locking out a second of the multi-states, the first of the multi-states being different than the second of the multi-states.
9. The method of claim 1 , further comprising:
applying pulses forming a staircase of increasing amplitude to a second wordline; and
after applying each pulse to the second wordline, performing a verifying operation for memory cells along the second wordline,
wherein the series of pulse is applied to the second wordline as a series of a plurality of N subsets of the staircase, the series including at least a first subset and a second subset, wherein the first subset includes the first and every Nth subsequent pulse of the staircase applied sequentially in order of increasing amplitude, the second subset includes the second and every Nth subsequent pulse of the staircase applied sequentially in order of increasing amplitude, and the second subset is applied subsequent to applying the first subset;
locking out the memory cells along the second wordline from further programming in response to verifying at the subset's verify level corresponding to the cells' target data state, wherein each subset uses a different set of verify levels for verify operations; and
subsequent to applying the first subset and prior to applying the second subset, unlocking the memory cells along the second wordline to allow further programming,
wherein the first subset along second wordline is applied subsequent to applying the first subset and prior to applying the second subset.
10. The method of claim 9 , wherein the second wordline is adjacent to the first wordline.
11. A method of writing a plurality of memory cells each into one of a plurality of target data states, the method including:
applying a series of pulses to a first wordline;
after applying each pulse to the first wordline, performing a verify operation for memory cells along the first wordline, wherein the verify operation includes an initial verify levels and a lockout verify level for each target data state, wherein the initial verify level is lower than the corresponding lock verify level for each of the target data states;
individually altering the bias on the memory cells to program more slowly in response to verifying at the corresponding to the cell's initial verify level; and
individually locking out the memory cells along the first wordline from farther programming in response to verifying at the corresponding to the cell's lockout verify level,
wherein an initial verify level for a first of the multi-states is the same as the lockout verify level for a second of the multi-states, the first of the multi-states being different than the second of the multi-states.
12. The method of claim 11 , wherein the verify levels are for a foggy programming process in which the distribution of memory cells are not fully resolved into distinct multi-states.
13. The method of 11 , wherein the verify levels are for a multi-pass programming process in which the distribution of memory cells are not fully resolved into distinct multi-states.