Sandisk Corporation

Sundisk Corporation

Xicor Corporation

Sandisk Technologies Inc.

Zilog, Inc.

Daniel C Guterman

Yupin K Fong

Geoffrey Steven Gongwer

Nima Mokhlesi

Gheorghe Samachisa

George Samachisa

Carlos J Gonzalez

Jack H Yuan

Sanjay Mehrotra

Stephen Jeffrey Gross

Kevin M Conley

John S Mangan

Chi-Ming Wang

Robert D Norman

Shahzad Khalid

Jeffrey G Craig

Daniel L Auclair

Robert F Wallace

Douglas J Lee

Brian Murphy

Khandker N Quader

Raul-Adrian Cernea

Rushyah Tang

Jian Feng Chen

Richard T Simko

Tomoharu Tanaka

Yan Li

Jeffrey W Lutze

Yoram Cedar

Mehrdad Mofidi

Charles Michael Schroter

Milton Lourenco Barrocas

Ping Wang

Isao Nojima

Shazad Khalid

Eliyahou Harrai

A memory using techniques to extract the data content of its storage elements, when the distribution of stored states is degraded, is presented. If the distribution of stored states has degraded, secondary evaluations of the memory cells are performed using modified read conditions. Based upon the results of these supplemental evaluations, the memory device determines the read conditions at which to best decide the data stored.

Methods of cell population distribution assisted read margining

The quality of data stored in a memory system is assessed by different methods, and the memory system is operated according to the assessed quality. The data quality can be assessed during read operations. Subsequent use of an Error Correction Code can utilize the quality indications to detect and reconstruct the data with improved effectiveness. Alternatively, a statistics of data quality can be constructed and digital data values can be associated in a modified manner to prevent data corruption. In both cases the corrective actions can be implemented specifically on the poor quality data, according to suitably chosen schedules, and with improved effectiveness because of the knowledge provided by the quality indications. These methods can be especially useful in high-density memory systems constructed of multi-level storage memory cells.

Increasing the effectiveness of error correction codes and operating multi-level memory systems by using information about the quality of the stored data

Tracking cells are used in a memory system to improve the read process. The tracking cells can provide an indication of the quality of the data and can be used as part of a data recovery operation if there is an error. The tracking cells provide a means to adjust the read parameters to optimum levels in order to reflect the current conditions of the memory system. Read operations are performed on the tracking cells, where threshold voltages of physical states of the tracking cells are further apart than threshold voltages of physical states of non-tracking cells. Based on the read operations, an extent to which the tracking cells are errored is determined.

Tracking cells for a memory system

A system and method for quickly and efficiently programming hard-to-program storage elements in non-volatile integrated memory devices is presented. A number of storage elements are simultaneously subjected to a programming process with the current flowing through the storage elements limited to a first level. As a portion of these storage elements reach a prescribed state, they are removed from the set of cells being programmed and the current limit on the elements that continue to be programmed is raised. The current level in these hard-to-program cells can be raised to a second, higher limit or unregulated. According to another aspect, during a program operation the current limit allowed for a cell depends upon the target state to which it is to be programmed.

System and method for programming cells in non-volatile integrated memory devices

Tracking cells are used in a memory system to improve the read process. The tracking cells can provide an indication of the quality of the data and can be used as part of a data recovery operation if there is an error. The tracking cells provide a means to adjust the read parameters to optimum levels in order to reflect the current conditions of the memory system. Additionally, some memory systems that use multi-state memory cells will apply rotation data schemes to minimize wear. The rotation scheme can be encoded in the tracking cells based on the states of multiple tracking cells, which is decoded upon reading.

Maximized multi-state compaction and more tolerance in memory state behavior is achieved through a flexible, self-consistent and self-adapting mode of detection, covering a wide dynamic range. For high density multi-state encoding, this approach borders on full analog treatment, dictating analog techniques including A to D type conversion to reconstruct and process the data. In accordance with the teachings of this invention, the memory array is read with high fidelity, not to provide actual final digital data, but rather to provide raw data accurately reflecting the analog storage state, which information is sent to a memory controller for analysis and detection of the actual final digital data.

Multi-state memory

A set of storage elements is programmed beginning with a word line WLn adjacent a select gate line for the set. After programming the first word line, the next word line WLn+1 adjacent to the first word line is skipped and the next word line WLn+2 adjacent to WLn+1 is programmed. WLn+1 is then programmed. Programming continues according to the sequence {WLn+4, WLn+3, WLn+6, WLn+5, . . . } until all but the last word line for the set have been programmed. The last word line is then programmed. By programming in this manner, some of the word lines of the set (WLn+1, WLn+3, etc.) have no subsequently programmed neighboring word lines. The memory cells of these word lines will not experience any floating gate to floating gate coupling threshold voltage shift impact due to subsequently programmed neighboring memory cells. The word lines having no subsequently programmed neighbors are read without using offsets or compensations based on neighboring memory cells. The other word lines are read using compensations based on data states within both subsequently programmed neighboring word lines.

Alternate row-based reading and writing for non-volatile memory

Storage elements are read multiple times and the results are accumulated and averaged for each storage element to reduce the effects of noise or other transients in the storage elements and associated circuits that may adversely affect the quality of the read. Several techniques may be employed, including: A full read and transfer of the data from the storage device to the controller device for each iteration, with averaging performed by the controller; a full read of the data for each iteration, with the averaging performed by the storage device, and no transfer to the controller until the final results are obtained; one full read followed by a number of faster re-reads exploiting the already established state information to avoid a full read, followed by an intelligent algorithm to guide the state at which the storage element is sensed. These techniques may be used as the normal mode of operation, or invoked upon exception condition, depending on the system characteristics. A similar form of signal averaging may be employed during the verify phase of programming. An embodiment of this technique would use a peak-detection scheme. In this scenario, several verify checks are performed at the state prior to deciding if the storage element has reached the target state. If some predetermined portion of the verifies fail, the storage element receives additional programming. These techniques allow the system to store more states per storage element in the presence of various sources of noise.

Reducing the effects of noise in non-volatile memories through multiple reads

Soft errors occur during normal use of a solid-state memory such as EEPROM or Flash EEPROM. A soft error results from the programmed threshold voltage of a memory cell being drifted from its originally intended level. The error is initially not readily detected during normal read until the cumulative drift becomes so severe that it develops into a hard error. Data could be lost if enough of these hard errors swamps available error correction codes in the memory. A memory device and techniques therefor are capable of detecting these drifts and substantially maintaining the threshold voltage of each memory cell to its intended level throughout the use of the memory device, thereby resisting the development of soft errors into hard errors.

Soft errors handling in EEPROM devices

Non-volatile memory cells store a level of charge corresponding to the data being stored in a dielectric material storage element that is sandwiched between a control gate and the semiconductor substrate surface over channel regions of the memory cells. More than two memory states are provided by one of more than two levels of charge being stored in a common region of the dielectric material. More than one such common region may be included in each cell. In one form, two such regions are provided adjacent source and drain diffusions in a cell that also includes a select transistor positioned between them. In another form, NAND arrays of strings of memory cells store charge in regions of a dielectric layer sandwiched between word lines and the semiconductor substrate.

Multi-state non-volatile integrated circuit memory systems that employ dielectric storage elements

One embodiment of the present invention includes applying a first value to a bit line, boosting word lines associated with the bit line and a common selection line to create a first condition based on the first value, and cutting off a boundary non-volatile storage element associated with the common selection line to maintain the first condition for a particular non-volatile storage element associated with the bit line and common selection line. A second value is applied to the bit line and at least a subset of the word lines are boosted to create a second condition for a different non-volatile storage element associated with the bit line and common selection line. The second condition is based on the second value. The first condition and the second condition overlap in time. Both non-volatile storage elements are programmed concurrently, based on their associated conditions.

Concurrent programming of non-volatile memory

A non-volatile memory device is provided with a controller and includes method that controls memory operations and to emulate the memory and communication characteristics of a legacy memory device. In this way, the memory device is compatible with a host that was originally designed to operate the legacy memory device. In particular, the controller performs the emulation to the host taking into account differences such as multibit memory, error correction requirement, memory support of overwrites, and erasable block sizes.

Method for adapting a memory system to operate with a legacy host originally designed to operate with a different memory system

Growing community of inventors

Fremont, CA, United States of America

Daniel C Guterman