摘要:
In connection with an erase operation of a block of non-volatile storage elements, a determination is made as to whether the block is partially but not fully programmed. A degree of partial programming can be determined by a pre-erase read operation which determines a highest programmed word line, or which determines whether there is a programmed storage element in a subset of word lines above a small subset of source side word lines. Since a partially programmed block will pass an erase-verify test more easily than a fully programmed block, a measure is taken to ensure that the block is sufficiently deeply erased. In one approach, an erase-verify test is made stricter by adjusting a sensing parameter when the block is partially programmed. In another approach, the block can be programmed before being erased. Or, an extra erase pulse which is not followed by an erase-verify test can be applied.
摘要:
Methods and non-volatile storage systems are provided for using compensation that depends on the temperature at which the memory cells were programmed. Note that the read level compensation may have a component that is not dependent on the memory cells' Tco. That is, the component is not necessarily based on the temperature dependence of the Vth of the memory cells. The compensation may have a component that is dependent on the difference in width of individual Vth distributions of the different states across different temperatures of program verify. This compensation may be used for both verify and read, although a different amount of compensation may be used during read than during verify.
摘要:
Dynamically determining read levels on chip (e.g., memory die) is disclosed herein. One method comprises reading a group of non-volatile storage elements on a memory die at a first set of read levels. Results of the two most recent of the read levels are stored on the memory die. A count of how many of the non-volatile storage elements in the group showed a different result between the reads for the two most recent read levels is determined. The determining is performed on the memory die using the results stored on the memory die. A dynamic read level is determined for distinguishing between a first pair of adjacent data states of the plurality of data states based on the read level when the count reaches a pre-determined criterion. Note that the read level may be dynamically determined on the memory die.
摘要:
In a non-volatile storage system, capacitive coupling effects are reduced by reducing the probability that adjacent storage elements reach the lockout condition at close to the same program pulse. A slow down measure such as an elevated bit line voltage is applied to the storage elements of a word line which are associated with odd-numbered bit lines, but not to the storage elements associated with even-numbered bit lines. The elevated bit line voltage is applied over a range of program pulses, then stepped down to ground over one or more program pulses. The range of programming pulses over which the slow down measure is applied, can be fixed or determined adaptively. A program pulse increment can be dropped, then increased, when the bit line voltage is stepped down. Storage elements which are programmed to a highest target data state can be excluded from the slow down measure.
摘要:
In a non-volatile memory system, a multi-phase programming operation is performed in which a drain-side select gate voltage (Vsgd) can be adjusted in different programming phases to accommodate different bit line bias (Vbl) levels. A higher Vbl can be used when Vsgd is higher to avoid unnecessary stress on the SGD transistor and reduce power consumption. For example, Vsgd can be higher in an earlier program phase than in a later program phase. The higher Vbl, which is not based on programming speed, can be is applied when the Vth of a storage element is between lower and upper verify levels of target data states, or throughout a programming phase, or at other times. The higher Vbl is an additional slow down measure which can be implemented in addition to a programming speed-based slow down measure such as a further raised Vbl which is applied to faster-programming storage elements.
摘要:
Threshold voltage distributions in a non-volatile memory device are narrowed, and/or programming time is reduced, using a programming technique in which the bit line voltage for storage elements having a target data state is stepped up, in lock step with a step up in the program voltage. The step up in the bit line voltage is performed at different times in the programming pass, for different subsets of storage elements, according to their target data state. The start and stop of the step up in the bit line voltage can be set based on a fixed program pulse number, or adaptive based on a programming progress. Variations include using a fixed bit line step, a varying bit line step, a data state-dependent bit line step, an option to not step up the bit line for one or more data states and an option to add an additional bit line bias.
摘要:
A non-volatile storage system reduces program disturb in a set of non-volatile storage elements by programming using selected bit line patterns which increase the clamped boosting potential of an inhibited channel to avoid program disturb. Alternate pairs of adjacent bit lines are grouped into first and second sets. Non-volatile storage elements of the first set of pairs are subject to program pulses and verify operations in each of a first number of iterations, after which non-volatile storage elements of the second set of pairs is subject to program pulses and verify operations in each of a second number of iterations.
摘要:
An erase operation for non-volatile memory includes first and second phases. The first phase applies a series of voltage pulses to a substrate, where each erase pulse is followed by a verify operation. The verify operation uses a verify level which is offset higher from a final desired threshold voltage level. The erase pulses step up in amplitude until a maximum level is reached, at which point additional erase pulses at the maximum level are applied. The first phase ends when the verify operation passes. The second phase applies one or more extra erase pulses which are higher in amplitude than the last erase pulse in the first phase and which are not followed by a verify operation. This avoids the need to perform a verify operation at deep, negative threshold voltages levels, which can cause charge trapping which reduces write-erase endurance, while still achieving the desired deep erase.
摘要:
In a non-volatile memory system, a programming speed-based slow down measure such as a raised bit line is applied to the faster-programming storage elements. A multi-phase programming operation which uses a back-and-forth word line order is performed in which programming speed data is stored in latches in one programming phase and read from the latches for use in a subsequent programming phase of a given word line. The faster and slower-programming storage elements can be distinguished by detecting when a number of storage elements reach a specified verify level, counting an additional number of program pulses which is set based on a natural threshold voltage distribution of the storage elements, and subsequently performing a read operation that separates the faster and slower programming storage elements. A drain-side select gate voltage can be adjusted in different programming phases to accommodate different bit line bias levels.
摘要:
A technique for erasing a non-volatile memory applies a p-well voltage to a substrate and drives or floats select gate voltages to accurately control the select gate voltage to improve write-erase endurance. Source and drain side select gates of a NAND string are driven at levels to optimize endurance. In one approach, the select gates float after being driven at a specific initial level, to reach a specific, optimal final level. In another approach, the select gates are driven at specific levels throughout an erase operation, in concert with the p-well voltage. In another approach, onset of select gate floating is delayed while the p-well voltage ramps up. In another approach, p-well voltage is ramped up in two steps, and the select gates are not floated until the second ramp begins. Floating can be achieved by raising the drive voltage to cut off pass gates of the select gates.