Abstract: Program disturb is reduced in a non-volatile storage system by programming storage elements on a selected word line WLn in separate groups, according to the state of their WLn?1 neighbor storage element, and applying an optimal pass voltage to WLn?1 for each group. Initially, the states of the storage elements on WLn?1 are read. A program iteration includes multiple program pulses. A first program pulse is applied to WLn while a first pass voltage is applied to WLn?1, a first group of WLn storage elements is selected for programming, and a second group of WLn storage elements is inhibited. Next, a second program pulse is applied to WLn while a second pass voltage is applied to WLn?1, the second first group of WLn storage elements is selected for programming, and the first group of WLn storage elements is inhibited. A group can include one or more data states.

Abstract: High-density semiconductor memory utilizing metal control gate structures and air gap electrical isolation between discrete devices in these types of structures are provided. During gate formation and definition, etching the metal control gate layer(s) is separated from etching the charge storage layer to form protective sidewall spacers along the vertical sidewalls of the metal control gate layer(s). The sidewall spacers encapsulate the metal control gate layer(s) while etching the charge storage material to avoid contamination of the charge storage and tunnel dielectric materials. Electrical isolation is provided, at least in part, by air gaps that are formed in the row direction and/or air gaps that are formed in the column direction.

Abstract: A process for programming non-volatile storage is able to achieve faster programming speeds and/or more accurate programming through synchronized coupling of neighboring word lines. The process for programming includes raising voltages for a set of word lines connected a group of connected non-volatile storage elements. The set of word lines include a selected word line, unselected word lines that are adjacent to the selected word line and other unselected word lines. After raising voltages for the set of word lines, the process includes raising the selected word line to a program voltage and raising the unselected word lines that are adjacent to the selected word line to one or more voltage levels concurrently with the raising the selected word line to the program voltage. The program voltage causes at least one of the non-volatile storage elements to experience programming.

Abstract: Non-voltage storage and techniques for fabricating non-volatile storage are disclosed. In some embodiments, at least a portion of the control gates of non-volatile storage elements are formed from p-type polysilicon. In one embodiment, a lower portion of the control gate is p-type polysilicon. The upper portion of the control gate could be p-type polysilicon, n-type polysilicon, metal, metal nitride, etc. P-type polysilicon in the control gate may not deplete even at high Vpgm. Therefore, a number of problems that could occur if the control gate depleted are mitigated. For example, a memory cell having a control gate that is at least partially p-type polysilicon might be programmed with a lower Vpgm than a memory cell formed from n-type polysilicon.

Abstract: Data stored in non-volatile storage is decoded using iterative probabilistic decoding. An error correcting code such as a low density parity check code may be used. In one approach, initial reliability metrics, such as logarithmic likelihood ratios, are used in decoding sensed states of a set of non-volatile storage element. The decoding attempts to converge by adjusting the reliability metrics for bits in code words which represent the sensed state. Soft data bits are read from the memory if the decoding fails to converge. Initial reliability metric values are provided after receiving the hard read results and at each phase of the soft bit operation(s). In one embodiment, a second soft bit is read from the memory using multiple subsets of soft bit compare levels. While reading at the second subset of compare levels, decoding can be performed based on the first subset data.

Abstract: To program a set of non-volatile storage elements, a set of programming pulses are applied to the control gates (or other terminals) of the non-volatile storage elements. The programming pulses have pulse widths that vary as a function of simulated pulse magnitude data. The programming pulses can also have pulse magnitudes that vary based on measurements taken while testing the set of non-volatile storage elements. In one embodiment, the pulse widths are determined after simulation performed prior to fabrication of the non-volatile storage elements. In another embodiment, the pulse magnitudes are calculated after fabrication of the non-volatile storage elements.

Abstract: A process is performed periodically or in response to an error in order to dynamically and adaptively optimize read compare levels based on memory cell threshold voltage distribution. One embodiment of the process includes determining threshold voltage distribution data for a population of non-volatile storage elements, smoothing the threshold voltage distribution data using a weighting function to create an interim set of data, determining a derivative of the interim set of data, and identifying and storing negative to positive zero crossings of the derivative as read compare points.

Abstract: To program a set of non-volatile storage elements, a set of programming pulses are applied to the control gates (or other terminals) of the non-volatile storage elements. The programming pulses have pulse widths that vary as a function of simulated pulse magnitude data. The programming pulses can also have pulse magnitudes that vary based on measurements taken while testing the set of non-volatile storage elements. In one embodiment, the pulse widths are determined after simulation performed prior to fabrication of the non-volatile storage elements. In another embodiment, the pulse magnitudes are calculated after fabrication of the non-volatile storage elements.

Abstract: Data in non-volatile storage is decoded using iterative probabilistic decoding. An error correcting code such as a low density parity check code may be used. In one approach, initial reliability metrics, such as logarithmic likelihood ratios, are used in decoding sensed states of a set of non-volatile storage elements. The decoding attempts to converge by adjusting the reliability metrics for bits in code words which represent the sensed state. Simulated annealing using an adjustable temperature parameter based on a level of error in the data can be performed to. The simulated annealing can introduce randomness, as noise for example, into the decoding process. Moreover, knowledge of the device characteristics can be used to guide the simulated annealing process rather than introducing absolute randomness. The introduction of a degree of randomness adds flexibility that permits possible faster convergence times and convergence in situations where data may otherwise be uncorrectable.

Abstract: Data in non-volatile storage is decoded using iterative probabilistic decoding. An error correcting code such as a low density parity check code may be used. In one approach, initial reliability metrics, such as logarithmic likelihood ratios, are used in decoding sensed states of a set of non-volatile storage elements. The decoding attempts to converge by adjusting the reliability metrics for bits in code words which represent the sensed state. Simulated annealing using an adjustable temperature parameter based on a level of error in the data can be performed. The simulated annealing can introduce randomness, as noise for example, into the decoding process. Moreover, knowledge of the device characteristics can be used to guide the simulated annealing process rather than introducing absolute randomness. The introduction of a degree of randomness adds flexibility that permits possible faster convergence times and convergence in situations where data may otherwise be uncorrectable.

Abstract: Data stored in non-volatile storage is decoded using iterative probabilistic decoding. An error correcting code such as a low density parity check code may be used. In one approach, initial reliability metrics, such as logarithmic likelihood ratios, are used in decoding sensed states of a set of non-volatile storage element. The decoding attempts to converge by adjusting the reliability metrics for bits in code words which represent the sensed state. Soft data bits are read from the memory if the decoding fails to converge. Initial reliability metric values are provided after receiving the hard read results and at each phase of the soft bit operation(s). In one embodiment, a second soft bit is read from the memory using multiple subsets of soft bit compare levels. While reading at the second subset of compare levels, decoding can be performed based on the first subset data.

Abstract: Data stored in non-volatile storage is decoded using iterative probabilistic decoding. An error correcting code such as a low density parity check code may be used. In one approach, initial reliability metrics, such as logarithmic likelihood ratios, are used in decoding sensed states of a set of non-volatile storage element. The decoding attempts to converge by adjusting the reliability metrics for bits in code words which represent the sensed state. Soft data bits are read from the memory if the decoding fails to converge. Initial reliability metric values are provided after receiving the hard read results and at each phase of the soft bit operation(s). In one embodiment, a second soft bit is read from the memory using multiple subsets of soft bit compare levels. While reading at the second subset of compare levels, decoding can be performed based on the first subset data.

Abstract: A process is performed periodically or in response to an error in order to dynamically and adaptively optimize read compare levels based on memory cell threshold voltage distribution. One embodiment of the process includes determining threshold voltage distribution data for a population of non-volatile storage elements, smoothing the threshold voltage distribution data using a weighting function to create an interim set of data, determining a derivative of the interim set of data, and identifying and storing negative to positive zero crossings of the derivative as read compare points.

Abstract: Data stored in non-volatile storage is decoded using iterative probabilistic decoding and multiple read operations to achieve greater reliability. An error correcting code such as a low density parity check code may be used. In one approach, initial reliability metrics, such as logarithmic likelihood ratios, are used in decoding read data of a set of non-volatile storage element. The decoding attempts to converge by adjusting the reliability metrics for bits in code words which represent the sensed state. If convergence does not occur, e.g., within a set time period, the state of the non-volatile storage element is sensed again, current values of the reliability metrics in the decoder are adjusted, and the decoding again attempts to converge.

Abstract: A memory system is disclosed that includes a set of non-volatile storage elements. A given memory cell has a dielectric cap above the floating gate. In one embodiment, the dielectric cap resides between the floating gate and a conformal IPD layer. The dielectric cap reduces the leakage current between the floating gate and a control gate. The dielectric cap achieves this reduction by reducing the strength of the electric field at the top of the floating gate, which is where the electric field would be strongest without the dielectric cap for a floating gate having a narrow stem.

Abstract: A method of fabricating a memory system is disclosed that includes a set of non-volatile storage elements. The method includes forming a floating gate having a top and at least two sides. A dielectric cap is formed at the top of the floating gate. An inter-gate dielectric layer is formed around the at least two sides of the floating gate and over the top of the dielectric cap. A control gate is formed over the top of the floating gate, the inter-gate dielectric layer separates the control gate from the floating gate. In one aspect, forming the dielectric cap includes implanting oxygen in the top of the floating gate and heating the floating gate to form the dielectric cap from the implanted oxygen and silicon from which the floating gate was formed.

Abstract: Data stored in non-volatile storage is decoded using iterative probabilistic decoding and multiple read operations to achieve greater reliability. An error correcting code such as a low density parity check code may be used. In one approach, initial reliability metrics, such as logarithmic likelihood ratios, are used in decoding read data of a set of non-volatile storage element. The decoding attempts to converge by adjusting the reliability metrics for bits in code words which represent the sensed state. If convergence does not occur, e.g., within a set time period, the state of the non-volatile storage element is sensed again, current values of the reliability metrics in the decoder are adjusted, and the decoding again attempts to converge. In another approach, the initial reliability metrics are based on multiple reads. Tables which store the reliability metrics and adjustments based on the sensed states can be prepared before decoding occurs.

Abstract: Program disturb is reduced in a non-volatile storage system by programming storage elements on a selected word line WLn in separate groups, according to the state of their WLn?1 neighbor storage element, and applying an optimal pass voltage to WLn?1 for each group. Initially, the states of the storage elements on WLn?1 are read. A program iteration includes multiple program pulses. A first program pulse is applied to WLn while a first pass voltage is applied to WLn?1, a first group of WLn storage elements is selected for programming, and a second group of WLn storage elements is inhibited. Next, a second program pulse is applied to WLn while a second pass voltage is applied to WLn?1, the second first group of WLn storage elements is selected for programming, and the first group of WLn storage elements is inhibited. A group can include one or more data states.

Abstract: Capacitive coupling from storage elements on adjacent bit lines is compensated by adjusting voltages applied to the adjacent bit lines. An initial rough read is performed to ascertain the data states of the bit line-adjacent storage elements, and during a subsequent fine read, bit line voltages are set based on the ascertained states and the current control gate read voltage which is applied to a selected word line. When the current control gate read voltage corresponds to a lower data state than the ascertained state of an adjacent storage element, a compensating bit line voltage is used. Compensation of coupling from a storage element on an adjacent word line can also be provided by applying different read pass voltages to the adjacent word line, and obtaining read data using a particular read pass voltage which is identified based on a data state of the word line-adjacent storage element.

Abstract: Data stored in non-volatile storage is read using sense operations and associated pre-conditioning waveforms. The pre-conditioning waveform provides a short term history for a non-volatile element which is analogous to the conditions experienced during programming when a programming pulse is applied prior to a verify operation. The pre-conditioning waveform can cause electrons to enter and exit trap sites, for instance, so that the accuracy of a probabilistic decoding process is improved. In one approach, multiple read operations are performed, some with pre-conditioning waveforms and some without. Pre-conditioning waveforms with different characteristics, such as amplitude, shape, duration and time before the associated read pulse, can also be used. For probabilistic decoding, initial reliability metrics can be developed based on multiple reads. Tables which store the reliability metrics can then be prepared for use in subsequent decoding.