Counter-based method and system for accessing memory cells

By introducing a counter section into the memory cell array to store counting information related to user data, the problem of read instability caused by electrical characteristic drift in three-dimensional memory cell arrays is solved, and more efficient and reliable data reading is achieved.

CN115485776BActive Publication Date: 2026-11-10MICRON TECHNOLOGY INC
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202080100236.6
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2020-05-13
Publication Date
2026-11-10
Estimated Expiration
2040-05-13

AI Technical Summary

Technical Problem

In a three-dimensional memory cell array, the variable electrical characteristics of the memory cells lead to instability and reliability issues in read operations. In particular, after multiple read and write cycles, the threshold voltage of the memory cells drifts, affecting the accurate reading of information.

Method used

A counter-based reading technique is adopted. By adding a counter section to the memory cell array, the counting information corresponding to the number of bits of logic state 1 in the user data is stored. The encoding scheme is used to balance the distribution of logic states of the memory cells to ensure the accuracy of the reading process.

Benefits of technology

It improves the stability and reliability of memory cell reading, enabling accurate reading of user data even under changes in the electrical characteristics of memory cells, reducing the error rate and increasing the reading speed.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN115485776B_ABST
    Figure CN115485776B_ABST
Patent Text Reader

Abstract

The present disclosure relates to a method for accessing a memory cell array, including the steps of: storing user data in a plurality of memory cells of a memory array; storing count data corresponding to a number of bits having a predetermined first logical value in the user data in a counter associated with the memory cell array; applying a read voltage to the memory cells to read the user data stored in the memory cell array; applying the read voltage to the cells of the counter to read the count data stored in the counter and to provide a target value corresponding to the number of bits having the first logical value in the user data, wherein the count data and the user data are simultaneously read in a manner during the application of the read voltage such that the target value is provided during the reading of the user data; and based on the target value of the counter, stopping the application of the read voltage when the number of bits having the first logical value in the user data corresponds to the target value. A related memory device and a related system are also disclosed.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This disclosure generally relates to array memories that operate memory cells, and more specifically, to counter-based methods and systems for accessing memory cells. Background Technology

[0002] Memory devices are used in many electronic systems, such as mobile phones, personal digital assistants, laptops, digital cameras, and the like. Various types of memory devices exist, including magnetic hard disks, random access memory (RAM), read-only memory (ROM), dynamic RAM (DRAM), synchronous dynamic RAM (SDRAM), ferroelectric RAM (FeRAM), magnetic RAM (MRAM), resistive RAM (RRAM), flash memory, phase-change memory (PCM), and others. Memory devices can be volatile or non-volatile. Non-volatile memory retains its contents when power is disconnected, making it a good choice for storing information that will be retrieved after a system restart. Specifically, non-volatile memory cells can maintain their stored logical state for extended periods, even without an external power source. Volatile memory cells, unless periodically refreshed by an external power source, may lose their stored state over time.

[0003] Information is stored by programming different states of a memory device. For example, a binary device typically has two states, usually represented by logic "1" or logic "0". In other systems, more than two states can be stored. To access the stored information, components of the memory device can read or sense the stored states. To store information, components of the memory device can write or program logical states.

[0004] Generally, improvements to memory devices can include increasing memory cell density, improving read / write speeds, increasing reliability, enhancing data retention, reducing power consumption, reducing manufacturing costs, and scaling down to a smaller size than traditional devices (which can lead to a relatively higher error rate).

[0005] When memory cells exhibit variable electrical characteristics, specifically when memory devices have a three-dimensional (3D) array of memory cells, more robust read techniques may be required to improve the performance and reliability of the memory device. Attached Figure Description

[0006] Figure 1 An exemplary block diagram illustrating an exemplary memory cell that can be read according to embodiments of the present disclosure;

[0007] Figure 2 A portion of an exemplary memory cell array is schematically shown;

[0008] Figure 3A and 3B Examples of data patterns supporting a counter-based method for reading memory cells are shown according to embodiments of the present disclosure;

[0009] Figure 4A and 4B The diagram schematically illustrates the distribution of logical states manifested by memory cells during the read phase according to embodiments of the present disclosure;

[0010] Figure 5 A logic block of a reading algorithm operating according to the method of this disclosure is schematically shown;

[0011] Figure 6 A flowchart illustrating the steps of the method according to this disclosure; and

[0012] Figure 7 A schematic block diagram of a system including a memory device according to this disclosure is shown. Detailed Implementation

[0013] Referring to those diagrams, this paper will disclose methods and systems for improved read operations of memory cells.

[0014] In the following detailed description, numerous specific details are set forth to provide a thorough understanding of the claimed subject matter. However, those skilled in the art will understand that the claimed subject matter can be practiced without these specific details. In other instances, methods, apparatus, and / or systems known to those generally skilled in the art have not been described in detail so as not to obscure the claimed subject matter.

[0015] Non-volatile memory retains its contents when power is disconnected, making it a good choice for storing information that will be retrieved after a system reboot. Flash memory is a type of non-volatile memory that retains stored data and is characterized by very fast access times. Furthermore, it can be erased in blocks, rather than one byte at a time. Each erasable memory block consists of multiple non-volatile memory cells arranged in a matrix of rows and columns. Each cell is coupled to an access line and / or a data line. Cells are programmed and erased by manipulating voltages on the access and data lines. Flash memory has a long history and is well-suited for high-capacity storage applications; however, its performance does not meet the demands of most demanding applications today. Newer technologies, such as 3D Crosspoint (3DXPoint) memory and Self-Selective Memory (SSM), offer better performance in terms of access time and access granularity (data can be programmed and read at the page, word, or even bit level). As technology advances, accessing data during read operations becomes increasingly challenging.

[0016] Figure 1A block diagram is shown of an exemplary assembly 100 according to the present disclosure, comprising memory cells 100' that can be arranged in an array and then programmed and read.

[0017] exist Figure 1 In the illustrated embodiment, memory cell 100' includes storage material 102 between access lines 104 and 106. Access lines 104 and 106 electrically couple memory cell 100' to circuitry 142, which writes to and reads from memory cell 100'. The term "coupled" can refer to elements that are physically, electrically, and / or communicatively connected directly or indirectly, and is used interchangeably with the term "connected" herein. Physical coupling may include direct contact. Electrical coupling includes an interface or interconnect that allows current flow and / or signal transmission between components. Communication coupling includes a connection that allows components to exchange data, including wired and wireless connections.

[0018] In one embodiment, the storage material 102 comprises a selectable material exhibiting memory effects. The selectable material is a material that enables the selection of memory cells in the array without requiring a separate selector element. Therefore, Figure 1 The storage material 102 is referred to as a “selector / storage material”. The material exhibits a memory effect if the circuitry used to access the memory cell allows the material to be in one of several states (e.g., via a write operation) and later determines the programmed state (e.g., via a read operation). The circuitry used to access the memory cell (e.g., via read and write operations) is generally referred to as an “access circuit” and is further discussed below with reference to access circuitry 143. The access circuitry can store information in the memory cell 100' by placing the storage material 102 in a specific state. The storage material 102 may comprise, for example, chalcogenide materials such as Te-Se alloys, As-Se alloys, Ge-Te alloys, As-Se-Te alloys, Ge-As-Se alloys, Te-As-Ge alloys, Si-Ge-As-Se alloys, Si-Te-As-Ge alloys, or other materials capable of serving as both a storage element and a selector, enabling addressing of a specific memory cell and determination of the memory cell's state. Therefore, in one embodiment, the memory cell 100' is a self-selecting memory cell containing a single layer of material, which acts as both a selector element for selecting memory cells and a memory element for storing logical states (i.e., states related to a given polarity of the cell).

[0019] In one embodiment, the storage material 102 is a phase change material. A phase change material can electrically switch between a generally amorphous state and a generally crystalline state across the entire spectrum, between a fully amorphous state and a fully crystalline state. The memory cell 100' may further include a selection device (not shown) between access lines 104 and 106; the selection device may be coupled in series to the storage material 102. In another embodiment, the storage material 102 is not a phase change material. In one embodiment where the storage material 102 is not a phase change material, the storage material is capable of switching between two or more stable states without changing its phase. The access circuit 143 can program the memory cell 100' by applying a voltage of a specific polarity to bring the storage material 102 to the desired stable state.

[0020] In one such embodiment, memory cell 100' is programmed to reach a "threshold" or experience a "threshold event." When the memory cell reaches the threshold (e.g., during a programming voltage pulse), the memory cell undergoes a physical change that causes the memory cell to exhibit a specific threshold voltage in response to the application of a subsequent voltage (e.g., a read voltage with a specific magnitude and polarity). Programming memory cell 100' can therefore involve applying a voltage of a given polarity to induce a programming threshold event that causes memory cell 100' to exhibit a specific threshold voltage at subsequent read voltages of the same or different polarities. In one such embodiment, storage material 102 is a selectable material (e.g., a non-phase-change chalcogenide material or other selectable material) that can be programmed by inducing a threshold event.

[0021] As explained in further detail below, when read, the output of such memory cells varies with the polarity used to program the memory cell and the polarity used to read the memory cell. For example, storage material 102 may exhibit a "lower threshold voltage" or a "higher threshold voltage" in response to the read voltage, based on the polarity of both the programming and read voltages. In the context of this disclosure, exhibiting a threshold voltage means that, in response to a voltage of a specific magnitude and polarity applied to the terminals of the memory cell, the voltage across the memory cell is approximately equal to the threshold voltage. The threshold voltage thus corresponds to the minimum voltage required to be applied at the input to produce an output, i.e., to see a defined electrical response from the cell. In other words, in the context of this disclosure, the verb "reaching a threshold" means that the cell experiences a threshold event, i.e., the cell has an electrical response in response to an applied voltage higher than a given threshold, thereby exhibiting a unique threshold voltage.

[0022] As mentioned above, access lines 104 and 106 electrically couple memory cell 100' to circuit 142. Access lines 104 and 106 may be referred to as bit lines and word lines, respectively. Word lines are used to access specific words in the memory array, and bit lines are used to access specific positions within a word. In one embodiment, access lines 104 and 106 may be made of one or more of the following: suitable metals, including: Al, Cu, Ni, Cr, Co, Ru, Rh, Pd, Ag, Pt, Au, Ir, Ta, and W; conductive metal nitrides, including TiN, TaN, WN, and TaCN; conductive metal silicides, including tantalum silicide, tungsten silicide, nickel silicide, cobalt silicide, and titanium silicide; conductive metal nitrosilicides, including TiSiN and WSiN; conductive metal carbonitrides, including TiCN and WCN, or any other suitable conductive material.

[0023] In one embodiment, electrode 108 is disposed between storage material 102 and access lines 104, 106. Electrode 108 electrically couples access lines 104, 106 to storage material 102. Electrode 108 may be made of one or more of the following: conductive and / or semiconductive materials, such as carbon (C), carbonitride (C2), etc. x N y ); n-doped polycrystalline silicon and p-doped polycrystalline silicon; metals, including Al, Cu, Ni, Cr, Co, Ru, Rh, Pd, Ag, Pt, Au, Ir, Ta, and W; conductive metal nitrides, including TiN, TaN, WN, and TaCN; conductive metal silicides, including tantalum silicide, tungsten silicide, nickel silicide, cobalt silicide, and titanium silicide; conductive metal nitrosilicides, including TiSiN and WSiN; conductive metal carbonitrides, including TiCN and WCN; conductive metal oxides, including RuO2 or other suitable conductive materials.

[0024] The stack consisting of electrodes 108 and storage material 102 is hereinafter referred to as memory cell 100', without limiting the scope of this disclosure. In various embodiments, memory cell 100' may include more or fewer elements. Thus, memory cell 100' is one example of a memory cell. Other embodiments may include compared to Figure 1 The memory cells shown have additional, fewer, or different material layers (e.g., a thin dielectric material between the storage material and the access lines).

[0025] Referring again to circuit 142, according to an embodiment, access lines 104, 106 communicatively couple circuit 142 to memory cell 100'. Circuit 142 includes access circuitry 143 and sensing circuitry 145. The circuitry includes electronic components electrically coupled to perform analog or logical operations on received or stored information, output information, and / or stored information. Hardware logic is circuitry used to perform logical operations, such as those involved in data processing. In one embodiment, access circuitry 143 applies voltage pulses to access lines 104, 106 to write to or read from memory cell 100'. The terms "write" and "program" are used interchangeably to describe the action of storing information in a memory cell. To write to memory cell 100', the access circuitry applies voltage pulses of a specific magnitude and polarity to access lines 104, 106, which can simultaneously select memory cell 100' and program memory cell 100'.

[0026] For example, access circuit 143 applies a read voltage of one polarity to program memory cell 100' into one logic state, and applies pulses of different polarities to program memory cell 100' into different logic states. Due to the programming polarity of the memory cell, access circuit 143 can then distinguish between the different logic states. For example, in the case of memory read, access circuit 143 applies voltage pulses of specific magnitude and polarity to access lines 104, 106, which produces an electrical response detectable by sensing circuit 145. Detecting the electrical response may include, for example, detecting one or more of the following: a voltage drop across the terminals of a given memory cell array (e.g., a threshold voltage), a current through a given memory cell, and a threshold event of a given memory cell. In some cases, detecting the threshold voltage of a memory cell may include determining that the threshold voltage of the cell is below or above a reference voltage, such as a read voltage. Access circuit 143 can determine the logic state of memory cell 100' based on the electrical response to one or more of the voltage pulses in the read sequence.

[0027] The current generated after applying a read voltage therefore depends on the threshold voltage of the memory cell, determined by the resistance of the logic state storage element. For example, a first logic state (e.g., a set state) may correspond to a finite amount of current, while a second logic state (e.g., a reset state) may correspond to no current or a negligible current. Alternatively, the first logic state may correspond to a current above a current threshold, while the second logic state may correspond to a current below a current threshold.

[0028] Figure 2 A portion of a memory cell array 200 according to an embodiment is shown, which may include a plurality of memory cells, such as Figure 1The memory cell array 200 is an example of a three-dimensional crosspoint memory structure (3D X Point). The memory cell array 200 includes multiple access lines 204, 206, which can be connected to... Figure 1 Access lines 104 and 106 are described as identical or similar. Access lines 204 and 206 can be referred to as bit lines and word lines, respectively. Figure 2 In the embodiments described herein, bit lines (e.g., access lines 204) are orthogonal to word lines (e.g., access lines 206). Storage material 202 (e.g., Figure 1 The storage material 102) is placed between access lines 204 and 206. (See also: Regarding...) Figure 1 As disclosed, in some instances, the storage material 202 may be a selectable storage material; in other instances, the storage material 202 may be coupled in series to a selection device (not shown). In one embodiment, an "intersection point" is formed at the intersection between the bit line and the word line. Memory cells are generated by the storage material 202 between the bit line and the word line, where the bit line intersects the word line (it should be noted that in...). Figure 2 In this diagram, additional layers, such as electrodes, are not shown so that the cell is schematically represented by storage material 202 (without limiting the scope of this disclosure, and additional layers may exist). Generally, intersections define the address of the memory cell. Storage material 202 may be a chalcogenide material, such as those mentioned above. Figure 1 The storage material 102 is described. In one embodiment, as described above regarding... Figure 1 The access lines 104, 106 and 204, 206 described are also made of one or more conductive materials. Although Figure 2 The image shows a single level or layer of memory cells, but the memory cell array 200 may contain multiple levels or layers of memory cells (e.g., in the z direction).

[0029] Therefore, a "crossover point" refers to a location where memory cells are formed such that the access lines associated with the memory cells "cross" each other topologically when they connect to different nodes of the memory cells. Crossover point architecture enables the achievement of a theoretically minimum cell area determined by the minimum spacing of the access lines.

[0030] Figure 1 and 2 Examples of memory cells and arrays are shown. However, other memory cell structures and arrays in which the memory cells exhibit electrical responses that vary with programming and reading polarity can be used. For example, in a 3D memory array, memory cells (not shown) can be formed at the intersection of vertical conductive pillars that act as bit lines and horizontal conductive planes that act as word lines. This and other array arrangements can also produce intersection architectures as described above.

[0031] Ideally, all memory cells in a memory device should be characterized by the same (nominal) resistivity, and therefore the same threshold voltage for the same logic state. The threshold voltage is the voltage to be applied to the memory cell to conduct current, i.e., the minimum voltage required to create a conductive path between terminals as defined above. However, since different cells programmed into the same logic state actually exhibit different resistivity values ​​due to several factors (e.g., variations in the electrical properties of the phase-change material due to the execution of several read-write operations and / or manufacturing tolerances), each logic state is actually associated with a corresponding resistivity distribution (typically Gaussian), and therefore with a corresponding threshold voltage distribution.

[0032] To evaluate the logic state of a cell, a read operation is performed to assess the threshold voltage distribution to which the cell belongs. For example, a read voltage can be applied to the cell via an access line, and the logic state of the cell is evaluated based on the current (present or absent) in response to the read voltage and the current (present or absent) depending on the cell's threshold voltage. When a suitable voltage difference is applied between its two terminals, the cell reaches a threshold (e.g., it becomes conductive); such a voltage difference can be obtained in different ways, for example, by biasing one terminal of, for example, a word line terminal to a negative voltage (e.g., a select voltage) and the other terminal of, for example, a bit line terminal to a positive voltage (e.g., a read voltage). Other bias configurations can produce the same effect (e.g., both word line and bit line terminals are biased to positive voltages, or the word line terminals are biased to a reference voltage, such as ground, and the bit line terminals are biased to positive voltages).

[0033] In other words, operations such as reading and writing can be performed on memory cells by activating or selecting word lines 206 and bit lines 204; these operations may be referred to as access operations. As known in the art, word line 206 may also be referred to as a row line, a sensing line, and an access line. Bit line 204 may also be referred to as a number line, a column line, a data line, and an access line. References to word lines and bit lines or the like are interchangeable without loss of understanding or operation. For example, an access line may be a word line and a data line may be a bit line. Word lines 206 and bit lines 204 may be perpendicular (or nearly perpendicular) to each other to create an array, as previously referenced. Figure 2 As shown. Depending on the type of memory cell (e.g., FeRAM, RRAM, etc.), other access lines (not shown), such as board lines, may exist. It should be understood that the exact operation of the memory device can be changed based on the type of memory cell and / or the specific access lines used in the memory device. Activating or selecting word line 206 or bit line 204 may involve applying voltage to the respective line via a dedicated driver. By activating a word line and a bit line, a single memory cell 202 can be accessed at their intersection. Accessing the memory cell may involve reading or writing the memory cell.

[0034] Access to memory cells can be controlled via a row decoder or a column decoder (not shown). For example, a row decoder receives a row address from the memory controller and activates the appropriate word line based on the received row address. Similarly, a column decoder receives a column address from the memory controller and activates the appropriate bit line.

[0035] As previously mentioned, memory cell 202 may exhibit different electrical characteristics after several cycles of operation (e.g., a series of read or write operations). For example, after receiving the same programming pulse to store logic state 1 (e.g., a SET programming pulse), if the memory cell is relatively new compared to memory cells that have undergone a larger number of read or write operations (e.g., a PCM cell with a smaller number of read or write operations), then the threshold voltage of memory cell 202 (e.g., a PCM cell) corresponding to logic state 1 may be different. Additionally, in some cases, after the chalcogenide material is programmed (e.g., crystallized or quenched) during a write operation, the chalcogenide material in the memory cell (e.g., logic memory element 102 or 202) may experience a change in its resistance (which may also be referred to as drift). Such a change in resistance can lead to a change in the threshold voltage of the memory cell and may prevent accurate reading of information from the memory cell (e.g., a PCM cell) after a certain period of time. In some embodiments, the amount of change may vary with ambient temperature. In many cases, relying on error correction mechanisms to handle errors may not be feasible.

[0036] This disclosure also provides a robust and reliable read technique for memory cells (e.g., PCM cells or SSM cells) exhibiting different, non-uniform, variable electrical characteristics that can arise from a variety of factors, including statistical process variations, cyclic events (e.g., read or write operations performed on the memory cell) or drift (e.g., resistance variations in chalcogenide alloys) and others, as described above.

[0037] More specifically, according to the memory cell reading technique of this disclosure, the reading of a set of user data (e.g., codewords, pages) and the reading of data stored in a memory portion (hereinafter referred to as a counter) associated with the memory cell array are performed simultaneously, the data of the counter being used to properly read the user data. Therefore, this disclosure can determine the total number of memory cells associated with a given logic state through highly efficient read operations, thereby providing a new and more efficient solution based on a counter-based sense amplifier method for reading memory cells (specifically, in a 3D memory device).

[0038] Figure 3AAn exemplary user data pattern diagram 301 is shown. User data pattern diagram 301 includes user data 310a and encoded user data 315a. An encoding process 320a, performed during the programming phase of the memory cell array, can convert user data 310a into encoded user data 315a. The encoded user data 315a can be stored in a set of memory cells, which may be, for example, reference memory cells. Figure 1 and 2 The memory cells 100' or 202 are described. Each frame of the encoded user data 315a may correspond to a memory cell that can represent a logic state 1 or a logic state 0. During the encoding process 320a, several parity bits may be added to the user data 310a to establish a predetermined number of bits in the encoded user data 315a with a given logic value (e.g., logic value 1). Therefore, the number of bits in the encoded user data 315a may be greater than the number of bits in the user data 310a (e.g., if some bits, such as parity bits, are added, then n is greater than m). Process 325 may convert the encoded user data 315a back to user data 310a after the encoded user data 315a has been accurately read.

[0039] In this embodiment, the plurality of encoded bits to be read represent codewords (CWs). The codewords may contain various information to be used during the reading phase.

[0040] In some embodiments, for each set of user data 310a, the number of memory cells corresponding to the encoded user data 315a displaying logic state 1 and logic state 0 may be the same (this may also be referred to as a balanced encoding scheme, where half of the encoded user data bits have logic value 1 and the other half have logic value 0). Thus, the encoded user data may be described as having a 50% weight. In some embodiments, for each set of user data 310a, the corresponding encoded user data 315a may have a predetermined number of memory cells displaying a given logic state (e.g., logic state 1), thus producing a constant weight that may differ from 50% (this may also be referred to as a constant weight encoding scheme). Generally, the result of the encoding process 320a may be to establish a predetermined number of memory cells in the encoded user data 315a displaying a given logic state (e.g., logic state 1).

[0041] In other words, according to embodiments of this disclosure, codewords can be manipulated by adding some extra information bits to limit the number of bits representing a given logical value (e.g., logical value 1) to a known predetermined value, typically between a minimum and a maximum value (e.g., between or within the range {Min1, Max1}). Therefore, in some embodiments, the memory cells of the array can be configured to store encoded user data, which includes modified user data (or, in some cases, original user data) and bits that can be added thereto, i.e., performing data manipulation to limit the number of bits in the codeword with a leading logical value (e.g., the number 1) to a number (e.g., 2 to 4) parity or inversion bits. In this way, the statistics of the codeword are improved by using a distribution with a reasonable number of bits representing a given logical value, specifically a distribution with a predetermined number of bits for logical value 1, or a distribution with a number of bits for logical value 1 within a given range, thereby facilitating read operations and avoiding the extreme case where very few bits represent the first logical value. This also allows for statistical tracking of codeword usage (e.g., drift and cycles) with very few extra bits, and improved speed.

[0042] In the context of this disclosure, a bit having a logic value of 1 (e.g., corresponding to a cell in logic state 1) is identified as a bit in a first logic state, and a bit having a logic value of 0 (e.g., corresponding to a cell in logic state 0) is identified as a bit in a second logic state, even if other definitions may be used.

[0043] Therefore, this disclosure provides, during the programming phase, the storage of user data (e.g., encoded user data 315a) in multiple memory cells of a memory array, the data undergoing an encoding scheme as previously described. More specifically, the user data can be encoded in codewords having a predetermined number of bits representing a first logical value. For example, in embodiments, the number of bits in the encoded user data having logical values ​​1 and 0 can be the same, i.e., the number of bits in the encoded user data representing the first and second logical values ​​is substantially the same, even if other configurations in which the encoded user data has a known predetermined number of bits representing the first logical value can be used.

[0044] Furthermore, the methods and systems of this disclosure utilize a codeword portion, namely the aforementioned counter, which has bits protected by ECC and / or other mechanisms such as voting schemes. Therefore, one set of memory cells in the array is configured to store user data, while an additional set of memory cells is configured to store counting information in the counter. In embodiments, the counting information may represent the number of bits in the user data that have a logic state of 1. The counting information is hereinafter referred to as counting data.

[0045] Figure 3BAn example of a user data pattern diagram 302 supporting the memory cell read technology of this disclosure is shown. User data pattern diagram 302 includes user data 310b and encoded data 315b. Encoding process 320b encodes user data 310b into encoded data 315b. In addition to count data 330, the encoded user data may also include user data 310b. (As already referenced...) Figure 3A The encoding scheme discussed may include adding several extra bits (which may also be called parity bits or inversion bits) to the user data during the encoding process, so that the encoded user data 310b may be different from the initial user data due to process 320b.

[0046] In some embodiments, the encoding scheme is used to store counting information in the form of counting data 330, which corresponds to the total number of bits in a plurality of memory cells that have a given logical state (e.g., logical state 1) in the user data. The counting information may be stored as a binary number representing the total number of bits in the user data that have a given logical state. In other cases, the counting information may be encoded with a weighted pattern having a given weight (e.g., 20%, 30%, 50% (i.e., half of the memory cells storing the counting information have a given logical state), 75%, etc.), as detailed below. In some cases, the encoding scheme may result in the total number of bits in the user data that have a given logical state (e.g., logical state 1) being within a predetermined weight range (e.g., 48% to 50%, 40% to 48%, 40% to 45%, or 20% to 25%) rather than an exact predetermined weight. In some cases, some additional bits may be programmed to obtain an exact weight.

[0047] In an embodiment, the number of memory units for the counting data 330 can be determined by the length of the user data 310b. In some cases, when the user data is 2... k When the length is 2 bits, the count data 330 contains k memory cells. In other cases, when the user data is 2... k When the length is 2 bits, the counting data 330 can contain 2×k memory units. In other words, k can be used to obtain a count of 2. k The perfect balance of codeword lengths is achieved by maintaining a certain number of bits in the code, where the number of bits with a value of 1 is typically maintained at a given interval [x; 2]. k In [ / 2], the smaller extra bits are less than k.

[0048] In some embodiments, during the encoding process 320b, the total number of bits of user data 310b with a given logical value (e.g., logical value 1) can be identified, and the total number can be stored in count data 330 (e.g., as a binary number). For example, when user data 310b is 16 bits long (e.g., 2... 4When the logic state of 9 out of 16 bits is 1, the count data 330 is 4 bits long and corresponds to the binary number 1001. In other embodiments, during the encoding process 320b, the total number of bits for a given logical value (e.g., logical value 1) of the user data 310b can be identified, and the total number of bits can be encoded in the count data 330 by converting each digit of the binary number (e.g., 1001) into a pair of numbers (e.g., binary digits 1 to 10 and binary digits 0 to 01). Using the same example described above, when the user data 310b is 16 bits long (e.g., 2...), the count data 330 is 4 bits long and corresponds to the binary number 1001. 4 When the count data 330 has 9 bits (e.g., the binary number 1001) representing a logical value of 1, it can be 8 bits long and correspond to 10010110. This type of encoding gives the count data 330 a 50% balanced weight, as will be discussed below.

[0049] The result of the encoding process 320b can thus establish a known number of memory cells in user data 310b having a given logical state (e.g., logical state "1"), and then store them in counting data 330. In any case, according to this disclosure, the value stored in the counter is used to accurately read user data 310b.

[0050] Therefore, this disclosure stores information (i.e., count data) in a counter associated with a memory cell array that corresponds to the total number of bits in user data having a given logical value (e.g., a first logical value 1). More specifically, in embodiments, storing the count data in the counter may include storing a target number of bits having a first logical value (i.e., a target 1) in a codeword.

[0051] Furthermore, according to embodiments of this disclosure, the counter bit is the result of the difference between a target (total) number of bits having a first logic value and a minimum expected number of bits having the first logic value (i.e., counter bit = Target1 - Min1), rather than simply storing a target number of bits that can be up to Max1, i.e., the maximum number of bits having the first logic value. This approach is advantageous because it reduces the number of counter bits stored. In other words, storing the counter data may include storing the difference between a target value and a minimum value of bits in the user data that have a given logic value (specifically, the first logic value 1) that must be stored in the codeword.

[0052] Alternatively, store the complement of the target number of bits by 2.

[0053] In any case, according to this disclosure, the codeword CW includes user data (e.g., codeword user data portion 310b) + counting data (e.g., codeword counting data portion 330), such as Figure 3B As shown in the image.

[0054] As mentioned above, embodiments of this disclosure balance a counter in a manner that the counter includes a known predetermined number of bits having a first logic value. In other words, the counter can be manipulated with additional information bits such that the counter includes a controlled number of bits having a given logic value (e.g., a controlled number of bits having a first logic value of 1). For example, the counter may have a 50% balancing scheme (as previously shown), or it may have another predetermined fixed number of bits that is different from 50% of the first logic value of 1.

[0055] This allows the target value to be obtained (and then the counter reading stops) when a known number of counter bits with a first logic value are read during a counter read operation. In other words, when a preset expected number of counter bits with the first logic value are read, the read of the counter bits is considered complete, and the target value can be used in the read algorithm to properly read the user data of the codeword. For example, in the case of a counter with a 50% balancing scheme, the counter read operation is completed when half of the counter bits have been read.

[0056] Therefore, in embodiments, this disclosure provides a balancing scheme to stop reading a counter when a desired number of 1 is reached. This is achieved by manipulating the counter with additional information until it becomes the controlled number of 1. The target value of the counter is provided and used when a predetermined number of bits with a first logic value of 1 is read from the counter.

[0057] This allows the counter DL to discharge when the counter reaches its target, thereby minimizing reset read interference (RRD) on the counter cell, as cutting the bit line after switching reduces the interference.

[0058] The sensing phase counts all logical values ​​"1" in the user data and compares them with the value given by the bits contained in the counter section 330 (i.e., the target value). The reading phase can then be stopped if no cell error is found.

[0059] The reading of codewords can be further improved according to this disclosure. More specifically, this disclosure relies on an encoding scheme that provides an exact number of memory cells with a given logical state (specifically, a first logical state 1) via counting data, which can accurately read user data regardless of the different electrical characteristics of the memory cells. Appropriately according to this disclosure, the counting data is determined while user data is being read.

[0060] Figure 4AThe diagram illustrates the distribution of logical states exhibited by memory cells during the read phase according to this disclosure. For example, the distribution may be derived from the variable electrical characteristics of the array's cells. The cell distribution curves are represented relative to the vertical axis "cell #" and the horizontal axis "threshold voltage". The read voltage is represented relative to the vertical axis "read voltage" and the horizontal axis "time".

[0061] It is worth noting that the threshold voltage is lower when the applied read voltage and programming voltage have the same polarity. For example, a positive programming voltage (e.g., corresponding to logic state "1") produces a lower threshold voltage. The lower threshold voltage can be achieved by applying a positive read voltage to determine logic state "1," as shown below. Figure 4A The distribution 401 is shown in the figure. The threshold voltage is higher when the applied read voltage and programming voltage have different polarities. For example, when reading with positive polarity, a negative programming voltage (e.g., corresponding to logic state "0") produces a higher threshold voltage, such as... Figure 4A Distribution 402 is shown in the diagram. Therefore, when a low read voltage with a positive bias is applied, the cell in logic state "0" does not reach the threshold. The access circuitry can determine the logic state of the memory cell based on the electrical response of the memory cell to the applied read voltage.

[0062] exist Figure 4A In this study, the distribution of the counter bits is referred to as 403 and 404, and is narrower than distributions 401 and 402 (because the counter includes a smaller number of bits), and is statistically concentrated around the peak of the distribution. Figure 4A The distributions 401 and 402 can correspond to Figure 3B The bits contained in the codeword portion 310b, and Figure 4A The distributions 403 and 404 can correspond to Figure 3B The bits contained in the codeword part 330.

[0063] Next, a read voltage 405 is applied to the memory cells of the array to read the user data stored therein. Simultaneously, the read voltage is also applied to the cells of the counter to read the count data stored in the counter, so as to provide a target value corresponding to the number of bits in the user data having a first logic value (e.g., logic 1 in the embodiment). In this way, during the application of the read voltage 405, the count data and user data are read simultaneously, such that the target value is provided during the reading of the user data. In other words, according to this disclosure, it is advantageous to read the bits of the counter while the bits of the main codeword containing the user data are stored in the array.

[0064] The sensing circuit detects the electrical response of multiple memory cells to the applied read voltage 405 (e.g., a threshold voltage across the cell terminals), and based on the threshold voltage, detects the logic state associated with one or more of the multiple memory cells. Figure 4A As shown in the embodiments, the memory cell exhibits a threshold voltage with a lower magnitude when the memory cell is in a first logic state 1, and exhibits a threshold voltage with a higher magnitude when the memory cell is in a second opposite logic state 0.

[0065] In embodiments of this disclosure, the read voltage 405 is a voltage ramp, and preferably, the voltage ramp is a variable stepped voltage ramp.

[0066] In an embodiment, such as Figure 4A As shown, the read voltage 405 applied to the cells containing user data in the array (i.e., the read voltage applied to read distributions 401 and 402) is the same as the read voltage applied to the cells containing count data (i.e., the read voltage applied to read distributions 403 and 404). In other words, the first read voltage of the memory cells storing user data and the second read voltage of the memory cells storing count data can be the same voltage applied simultaneously to the cells of the array (i.e., only one voltage). The read voltage can be a voltage ramp, such as a stepped voltage ramp.

[0067] according to Figure 4B Another embodiment shown in the diagram applies a voltage ramp 405, such as a first voltage ramp 405 (represented relative to the vertical axis "read voltage" and the horizontal axis "time"), to the cells containing user data (i.e., applied to read distributions 401 and 402 – represented relative to the vertical axis "cell #" and the horizontal axis "threshold voltage"), and applies an offset voltage ramp 405', such as a second voltage ramp 405', to the cells of the counter (i.e., applied to read distributions 403' and 404', which are thus shifted in time). In this embodiment, the second read voltage 405' is a voltage offset by a predetermined voltage offset relative to the first read voltage 405. For example, the offset could be 250mV (the offset could be limited by a minimum expected threshold voltage in the distribution of memory cells programmed to be in logic state 1).

[0068] The offset read voltage 405' can be generated from the read voltage 405, for example, by means of hardware components such as diodes or voltage dividers, or vice versa.

[0069] The offset read voltage 405' applied to the counter unit resolves the dual distribution case and prepares the counter before the main codeword is actually needed, such as... Figure 4B This is illustrated schematically. See also... Figure 4BThe count data is ready at t0, and it is actually needed after t1 (i.e., near the end of the distribution), so that the target value of the counter is available earlier during the read operation and can then be properly used in the read algorithm. In this way, a useful time margin is achieved in the read of the counter bits. In other words, by applying a voltage offset to the counter cell by modifying the supply voltage used to generate the voltage of the relevant sensing circuit, the counter bits are predicted by the time margin, i.e., read earlier than if the ramp had no offset, so that the counter distribution actually visible during the read operation is... Figure 4B The distributions are 403' and 404'. This is advantageous because the count data is not yet available when the read voltage is applied, so the offset allows the data to be obtained and used in advance. In other words, a first read voltage can be applied to read the memory cell storing user data, and a second read voltage can be applied in parallel, for example, simultaneously, to read the memory cell storing count data. The second read voltage can be a voltage offset relative to the first read voltage. The offset can make the instantaneous amplitude of the second read voltage greater than the simultaneous instantaneous amplitude of the first read voltage. The first and second read voltages can be voltage ramps, such as stepped voltage ramps.

[0070] As previously disclosed, user data and count data are read simultaneously, and the count data (whose distribution is concentrated around the peaks of the user data distribution and is relatively narrow) is still unavailable for the read algorithm at the start of the application of the read voltage (either the first read voltage or the second read voltage). For example, after applying the first read pulse (which allows reading a certain number of user data bits, i.e., at most the peaks of the distribution), instructions regarding the application of subsequent read pulses are provided based on the read count data; however, at the start of the read operation, the count data is not yet read, and the counted user data bits are greater than the counter bits, which will cause the read operation to stop unintended.

[0071] For this reason, in embodiments of this disclosure, a command is generated to enable the application of a read voltage, such as a first read voltage, for reading user data until all counter bits are read and a target value is provided. That is, such a command prevents the read voltage for reading user data from being interrupted due to a lack of count data from the counter. Once the count data is read, no further enable command is generated, and the count data is used in the read algorithm to perform the proper reading of user data. Based on the read count data, a (first) read voltage is then selected during the read operation as the voltage corresponding to a deterministic number of bits having a first logic value, and the read voltage is increased until the number of counted user data bits having the first logic value reaches the target value provided by the count data.

[0072] Therefore, when the number of bits in the user data that have the first logical value is equal to the target value, the target value provided by the counter is used to stop reading the user data.

[0073] In other words, the generation of the enable command corresponds to the use of a dummy counter pointing to the most likely codeword (i.e., a dummy bit with bits pointing to the most likely codeword to avoid interruptions to the read voltage). This dummy counter is used until the target value is reached by the balanced read trigger of the actual counter, and then the dummy counter is replaced with the real counter. Therefore, the dummy counter is used until the value of the actual counter (i.e., the balanced value) has been read, and the target enable command is generated and used in the read algorithm.

[0074] As previously seen, a counter can be considered read when all bits representing the first logic value of a counter, as established in a balancing scheme, are read. For example, if the count data has been encoded with a 50% weight according to the balancing scheme (i.e., the same number of 1s and 0s), then the count data is considered read when half of the counter bits have been read, and the count data can then replace the enable command to perform the proper reading of the remaining user data units.

[0075] Figure 5 This diagram illustrates a logic block of the read algorithm operating according to this disclosure. Count data is read from a counter and used "on the fly" during the execution of the user data read algorithm, i.e., reads of blocks 501 and 502 are performed in parallel. The read count data is introduced into the algorithm while user data is being read, thereby improving read performance. In time frames where the count data has not yet been fully read, an enable command (i.e., a dummy counter) is used to enable the reading of user data until the counter's read operation is complete and the target value is finally provided.

[0076] At box 501, due to counter balancing, an enable command pointing to the most likely codeword (i.e., a dummy counter) is used until the counter reaches a balanced value and generates a signal indicating that the target value is ready and can be used in the counter-based read algorithm, and thus the proper read of user data is performed at box 502.

[0077] When the target value of the counter is read, that value is used in a lookup table (LUT) to generate the appropriate read voltage, specifically, to define all subsequent read voltage pulses to be applied to the memory cells of the array. The counter value is thus transferred to the lookup table, and the read voltage generator (e.g., a ramp DAC generator) receives all the information for the generation of the ramp.

[0078] This disclosure therefore provides simultaneous reading of counter bits and data bits, wherein the counter bits are read as if they were bits of the main codeword, thus providing a new method instead of first reading the counter and then downloading it to the read voltage generator.

[0079] As previously shown, when the read voltage is applied, the counter bits must still be read (which would cause the read operation to stop unnecessarily). In an alternative embodiment of this disclosure, to address this issue, the counter bits are counted starting from their maximum number and then decremented until all counter bits are read. Thus, in this embodiment, the count value is decremented from its maximum value whenever a bit of the counter is swapped due to the application of the read voltage.

[0080] In this manner, the decoding logic scheme is applied to the counter bits, such that the count data is initially stored as the data that has already been read (i.e., the data will be counted starting from its maximum number of bits and the count will decrease). Therefore, according to this embodiment, at the start of the read operation, the number of counter bits is greater than the number of bits with predetermined logic values ​​read from the user data, and the read operation can continue without interruption, and then the number of counter bits is decremented during the read operation.

[0081] When all bits in the counter are swapped (read) due to the application of the read voltage, the target value is provided as the number of counter bits swapped, and the read operation can be stopped when the number of user data bits read corresponds to the number of swapped counter data bits, i.e., when the target value is matched. As referenced... Figure 4A and 4B The distribution shape discussed is such that, because the reading operations of user data and count data occur in parallel, the data from the counter is read before the actual time t1 when the data from the counter is needed.

[0082] In other words, in this embodiment, the bits of the counter are inverted in a manner that, for example, logically or by means of an inverter, such that the application of a read pulse is not interrupted when a read operation begins as the counter bits count from their maximum value. In this embodiment, a balancing scheme is not required for the counter.

[0083] In this case, as previously mentioned, the complement of the target number 2 is stored, that is, the maximum target is stored, and so as to start with the maximum value and decrease it to obtain the complement of the maximum target.

[0084] Generally, the reading voltage is applied based on the target reading value of the counter, and the reading voltage is stopped when the number of bits of user data with a predetermined logic value (e.g., a first logic value 1) corresponds to the value stored in the counter.

[0085] According to embodiments of this disclosure, user data during operation is evaluated during the application of the read voltage. Alternatively, the user data is evaluated after the target value is read, i.e., after a target signal indicating that the counter has been properly read is generated.

[0086] According to embodiments of this disclosure, the bits of the counter are stored and retrieved according to a majority voting scheme, such as a 3X voting scheme, where one counter bit corresponds to three array cells, or even this scheme can be applied to N array cells of the counter. Typically, this depends on the number of votes cast for the original BER. This disclosure therefore provides the counter cells with the use of a balanced code with a voting scheme, such that the count data is balanced data (e.g., four 1s as the target) and each bit is protected by a voting scheme (e.g., a 3X voting scheme).

[0087] Therefore, counters can be protected from reliability issues and BER reduced by applying redundancy information or correction schemes, or combinations thereof. Examples include voting schemes, where one counter bit corresponds to N array cells (e.g., N=3); differential cells, where one counter bit corresponds to one array cell at 0 and one array cell at 1; and error correction schemes using counter parity bits, where voting or differential is performed for each bit. All these schemes can also be combined with each other.

[0088] Furthermore, the protection level of the counter bits may be differentiated. For example, in embodiments of this disclosure, the LSB of the counter is more protected than the MSB (e.g., an LSB with voting 5X and an MSB with voting 3X). Therefore, the LSB bit can be more strongly protected because it is associated with the most likely CW.

[0089] In this embodiment, error correction codes (ECC) and / or differential cell schemes are used to protect user data.

[0090] In summary, according to this disclosure, user data bits and counter data bits are read simultaneously, such that the counter value is not read before the user data, but is read together with the user data and used immediately when reading the user data.

[0091] The disclosed counter-based readout algorithm offers numerous advantages, including extremely fast self-referencing (three or four steps) without additional download latency, as the counter is read in parallel with the main codeword. Furthermore, since the counter and main codeword are read in the same manner, the user data unit and counter unit require identical design architectures (e.g., identical sense amplifiers, decoders, arrays, etc.), thus simplifying the system architecture.

[0092] Furthermore, since the data stored in the counter corresponds to the difference between the target value with the expected bits having the first logic value and the minimum number (i.e., Min1, the minimum number of 1s to be stored in the codeword) rather than just the target value, the counter has a reduced number of bits, much smaller than the main codeword. Therefore, this disclosure also produces a reduced number of cells required for reliable counting (where power and area are reduced).

[0093] Figure 6 This is a flowchart illustrating the steps of method 600 according to the present disclosure. The described process can be performed by hardware logic and circuitry. For example, the following process is described as being performed by access circuitry and sensing circuitry, as disclosed herein. However, other embodiments may include different circuit configurations suitable for performing the process.

[0094] This disclosure provides a method for improving the read operation of memory cells. Before reading the memory cells, the access circuit writes data to multiple memory cells. For example, the access circuit writes logic 0s and logic 1s to multiple memory cells, such as... Figure 1 The memory cell 100'. In one embodiment, the access circuitry can write logic 0 by applying a programming pulse with negative polarity and write logic 1 by applying a programming pulse with positive polarity. The reverse convention can also be used. After writing data to multiple memory cells, the access circuitry can read multiple memory cells using the read sequence of this disclosure.

[0095] More specifically, at step 610, user data is stored in a plurality of memory cells of the memory array, and at step 620, count data corresponding to the number of bits in the user data having a predetermined first logic value is stored in a counter associated with the memory cell array. In some cases, the user data may be encoded before storing, for example, a range of bits. In some instances, the count data may indicate the difference between a target value of the bits in the user data having the first logic value and the minimum value of the bits in the user data having the first logic value.

[0096] At step 630, a read voltage is applied to a memory cell to read user data stored in the memory cell array. The read voltage applied to the memory cell storing the user data may be a first read voltage. Simultaneously, a read voltage is applied to a counter cell to read count data stored in the counter and to provide a target value corresponding to the number of bits in the user data having a first logic value. The read voltage applied to the memory cell storing the counter data may be a second read voltage. In some embodiments, the second read voltage may be offset relative to the first read voltage. According to this disclosure, during the application of the read voltage, count data and user data are read simultaneously in a manner that provides the target value during the reading of the user data.

[0097] At step 640, the number of user data bits read (which is counted by incrementing a dedicated counter) is compared with a target value.

[0098] Based on the target value of the counter, the reading voltage is stopped at step 650 when the number of bits in the user data with the first logic value corresponds to the target value, for example, the application of the first reading voltage. Alternatively, method 600 continues (e.g., starting from step 630, using an increased reading voltage). In some examples, the first and second reading voltages can be voltage ramps, such as stepped voltage ramps.

[0099] Figure 7 This is an advanced embodiment of system 700 capable of executing the read algorithm of this disclosure. System 700 includes a memory device 710, which in turn includes a memory cell array 720 and circuitry 730 operatively coupled to the memory cells 720; the memory cells 720 and the circuitry 730 form a memory portion, referred to herein as memory portion 700'.

[0100] Memory device 710 includes memory controller 740, which represents control logic that generates memory access commands, for example, in response to commands from host 750. Memory controller 740 accesses memory portion 700'. In one embodiment, memory controller 740 may also be implemented in host 750, specifically as part of host processor 760, even though this disclosure is not limited to any particular architecture. Controller 740 may contain embedded firmware and is adapted to manage and control the operation of memory portion 700'.

[0101] Generally, the memory controller 740 receives user data via input / output I / O. As previously shown, in some embodiments, the memory controller encodes the user data to satisfy a condition before storing it in memory cells. The condition is satisfied when the encoded user data has a predetermined number of bits representing a given logical value (e.g., logical value 1). As an example, the encoded user data may be configured such that 50% of the memory cells storing the encoded user data represent logical state 1, while the other 50% of the memory cells represent logical state 0 (i.e., a balanced encoding scheme where half of the encoded user data bits represent logical state 1 and the other half represent logical state 0). During the encoding process, the memory controller 140 may add a specific number of bits (e.g., parity bits) to the user data to establish a predetermined number of memory cells representing the given logical state. Due to the addition of parity bits, the encoded user data may have more bits than the user data. In some embodiments, different percentage values ​​(e.g., 40%, 60%, 75%) of memory cells representing logical state 1 may be used during the encoding process.

[0102] The count data can be defined as the number of bits of encoded user data having a given logical state (e.g., logical state 1), and the value of the count data can be encoded and stored as a codeword portion by the memory controller 740 in the array.

[0103] The controller 740 can be configured to manage the simultaneous reading of user data and count data, and based on a target value, it can be configured to stop reading user data when the number of bits in the user data that have a first logical value equals the target value.

[0104] The memory device 710 may also include other components, such as a processor unit coupled to the controller 740, an antenna, a connection member (not shown) to the host device, and the like.

[0105] Multiple signal lines couple the memory controller 740 to the memory section 700'. For example, such signal lines may include clock, command / address, write data (DQ), read DQ, and zero or more other signal lines. The memory controller 740 is thus operatively coupled to the memory section 700' via a suitable bus.

[0106] The memory portion 700' represents memory resources used in system 700. In one embodiment, the memory cell array 720 is managed as data rows accessed via word lines (rows) and bit lines (individual bits within a row). In one embodiment, the memory cell array 720 includes a 3D cross-dot array, for example... Figure 2 The memory cell array 200. The memory cell array 720 can be organized into individual channels, groups, and rows of memory. A channel is an independent control path to a storage location within a memory portion. A group refers to a common location on multiple memory devices (e.g., the same row address within different devices). A row is an array of memory locations within a memory device. In one embodiment, memory groups are divided into sub-rows, each sub-row having at least a portion of shared circuitry (e.g., drivers, signal lines, control logic) for the sub-row. It should be understood that channels, groups, rows, or other organization and combinations of organizations of memory locations may overlap physical resources. For example, the same physical memory location on a specific channel may be accessible as in a particular row, and the particular row may also belong to a group. Therefore, the organization of memory resources will be understood in an inclusive rather than exclusive manner.

[0107] In one embodiment, the memory controller 740 includes refresh (REF) logic 741. In one embodiment, refresh logic 741 indicates the location of the refresh and the type of refresh to be performed. Refresh logic 741 can trigger a self-refresh within the memory and issue an external refresh command to trigger the execution of a refresh operation.

[0108] In one embodiment, the access circuit 731 of circuit 730 performs a refresh (e.g., reprogramming) on ​​any accessed memory cells that were not refreshed during the read sequence. Thus, a complete refresh of the memory cells can be achieved with minimal additional refresh operations, which is primarily a side effect of the memory read sequence.

[0109] In embodiments, the circuitry may also be embedded in the memory controller, even if the present disclosure is not limited to a particular architecture.

[0110] exist Figure 7 In the exemplary embodiments described herein, the memory controller 740 includes an error correction circuit 742. The error detection / correction circuit 742 may include hardware logic to implement an error correction code (ECC) to detect errors occurring when reading data from a portion of the memory. In one embodiment, the error detection / correction circuit 742 also corrects errors (based on the implemented ECC code, up to a certain error rate). However, in other embodiments, the error detection / correction circuit 742 only detects errors but does not perform correction.

[0111] In the illustrated embodiment, the memory controller 740 includes command (CMD) logic 743, which represents logic or circuitry for generating commands to be sent to the memory portion. The memory controller 740 may also include a counter 744, such as the per-codeword counter disclosed above, configured to count the bits exchanged during a read operation. Obviously, other architectures may also be employed; for example, the counter may be embedded in the host 750 or may also be embedded in the circuitry 730.

[0112] Based on the received command and address information, the access circuit 731 of circuit 730 performs an operation to execute the command, such as the read operation of this disclosure. In one such embodiment, circuit 730 includes sensing circuit 732 to detect the electrical response of one or more memory cells to an applied read voltage. In one embodiment, sensing circuit 732 includes a sensing amplifier. Figure 7 The access circuitry 731 and sensing circuitry 732 are shown as embedded within the memory portion 700'; however, other embodiments may include access circuitry and / or sensing circuitry separate from the memory portion 700'. For example, the access circuitry and sensing circuitry may be included in a memory controller, such as memory controller 740.

[0113] The sensing circuit can be configured to detect the current through a given memory cell in response to a read voltage, wherein the access circuit is configured to determine that the given memory cell is in a first logic state based on the detected current magnitude being greater than or equal to a threshold current.

[0114] In one embodiment, the memory portion 700' includes one or more registers 733. Registers 733 represent one or more storage devices or storage locations that provide configuration or settings for the operation of the memory portion. Furthermore, in one embodiment, the circuitry 730 also includes decoding circuitry 734.

[0115] The host device 750 is a computing device according to any of the embodiments described herein, and may be a laptop computer, desktop computer, server, game or entertainment control system, scanner, copier, printer, router or switch device, embedded computing device, or other electronic device such as a smartphone. The memory device 710 may also be embedded in the host device 750.

[0116] In one embodiment, system 700 includes an interface 770 coupled to processor 760, which may represent a higher-speed or high-throughput interface for system components and / or graphics interface components requiring higher bandwidth connectivity. The graphics interface interfaces with the graphics components for providing a visual display to a user of system 700. In one embodiment, the graphics interface generates display content based on data stored in a memory device or based on operations performed by the processor, or both.

[0117] The system may also include a battery that is communicatively coupled to a host or memory device, such as a network interface 780 for connecting to other systems, and / or coupled to provide power to the system.

[0118] In summary, this disclosure provides simultaneous reading of counter bits and other bits of the codeword, thereby improving array performance.

[0119] According to an exemplary embodiment, a method for operating (accessing) a memory cell array includes the following steps: storing user data in a plurality of memory cells of the memory array; storing count data corresponding to bits of the user data having a predetermined first logic value in a counter associated with the memory cell array; applying a read voltage (first read voltage) to the memory cells to read the user data stored in the memory cell array; applying a read voltage (second read voltage) to a cell of the counter to read the count data stored in the counter and provide a target value corresponding to the bits of the user data having the first logic value, wherein during the application of the read voltage, the count data and the user data are read simultaneously in a manner such that the target value is provided during the reading of the user data; and stopping the application of the read voltage when the bits of the user data having the first logic value are equal to the target value, based on the target value of the counter.

[0120] According to another exemplary embodiment, a method for programming and reading memory cells in a vertical 3D memory device including one or more memory arrays includes the following steps: storing codewords containing user data and count data in the memory arrays, the count data corresponding to bits in the user data having a predetermined first logic value; storing in a counter the difference between a target value and a minimum value of bits in the user data having the first logic value stored in the codeword; applying a (first) read voltage to the memory cells to read the user data stored in the memory cell array; applying a (second) read voltage to a cell of the counter to read the count data stored in the counter and provide a target value corresponding to the bits in the user data having the first logic value, wherein during the application of the read voltage, the count data and user data are read simultaneously in a manner such that the target value is provided during the reading of the user data; and stopping the application of the read voltage when the bits in the user data having the first logic value are equal to the target value, based on the target value of the counter.

[0121] This disclosure also discloses a memory device comprising: one or more memory cell arrays; the memory arrays containing at least codewords of user data and count data, the count data corresponding to the number of bits in the user data having a predetermined first logic value; a memory controller supporting a counter-based method for reading memory cells; and circuitry for reading memory cells, the circuitry being operatively coupled to the memory cell array and comprising at least: access circuitry configured to: apply a first read voltage to memory cells to read user data stored in the memory cell array; and apply a second read voltage to cells of a counter to read count data stored in the counter and provide a target value corresponding to the number of bits in the user data having the first logic value, wherein during the application of the read voltage, the count data and user data are read simultaneously in a manner such that the target value is provided during the reading of the user data; and including sensing circuitry configured to: detect a threshold voltage exhibited by the plurality of memory cells in response to the application of the read voltage, wherein based on the target value of the counter, the controller is configured to stop the application of the read voltage when the number of bits in the user data having the first logic value is equal to the target value. In an embodiment, the storage cell array includes self-selected memory (SSM) or 3D X Point memory.

[0122] A related system is also disclosed, comprising a host device and a memory device as described above, the system including, for example, a display communicatively coupled to the memory device or the host, a network interface communicatively coupled to the memory device or the host, and a battery coupled to provide power to the system.

[0123] In the preceding embodiments, reference is made to the accompanying drawings, which form part of the invention, and specific examples are illustrated by way of description. Throughout the drawings, the same reference numerals describe substantially similar components. Other examples may be utilized, and structural, logical, and / or electrical changes may be made without departing from the scope of this disclosure. Furthermore, it will be understood that the scale and relative dimensions of the elements provided in the drawings are intended to illustrate embodiments of this disclosure and should not be construed as limiting.

[0124] As used herein, “a (a / an)” or “several” something can refer to one or more of such things. “Several” something means two or more. As used herein, the term “coupling” can include electrical coupling, direct coupling and / or direct connection without intermediate elements (e.g., through direct physical contact), or indirect coupling and / or connection with intermediate elements. The term coupling can further include two or more elements that cooperate or interact with each other (e.g., in a causal relationship).

[0125] Although specific examples have been shown and described herein, those skilled in the art will understand that arrangements calculated to achieve the same results may replace the specific embodiments shown. This disclosure is intended to cover modifications or variations of one or more embodiments of this disclosure. It should be understood that the above description is illustrative rather than restrictive. The scope of one or more examples of this disclosure should be determined by reference to the appended claims and the full scope of their equivalents.

Claims

1. A method for accessing an array of memory cells, comprising the following steps: - Store user data in multiple memory cells of the memory array; - Store the count data corresponding to the number of bits in the user data that have a predetermined first logic value in a counter associated with the memory cell array; - Apply a first read voltage to the memory cell to read the user data stored in the memory cell array; - A second read voltage is applied to a unit of the counter to read the count data stored in the counter and provide a target value corresponding to the number of bits in the user data having the first logic value, wherein during the application of the read voltage, the count data and the user data are read simultaneously in a manner such that the target value is provided during the reading of the user data; as well as - Based on the target value of the counter, the application of the first read voltage is stopped when the number of bits in the user data having the first logic value corresponds to the target value.

2. The method of claim 1, wherein the user data is encoded in a codeword having a number of bits representing a first logical state within a range, the encoding step comprising manipulating the codeword to constrain the number of bits representing the first logical value within the range.

3. The method of claim 2, wherein the encoded user data has the same number of bits representing the first logical value and the second logical value.

4. The method of claim 1, wherein storing the counting data includes storing the difference between the target value of the bits having the first logical value in the user data and the minimum value of the bits having the first logical value to be stored in the codeword in the user data.

5. The method of claim 1, wherein the bit stored in the counter is stored and read according to a majority voting scheme.

6. The method of claim 1, further comprising balancing the counter according to a balancing scheme such that the counter includes a known predetermined number of bits representing the first logic value, the balancing scheme including manipulating the counter with additional bits.

7. The method of claim 6, wherein when the predetermined number of bits representing the first logic value in the counter is read, the application of the second read voltage is stopped and the target value is provided.

8. The method according to claim 7, comprising the following steps: A command is generated that allows the application of the second read voltage until a predetermined number of bits in the counter representing the first logic value are read and the target value is provided, the target value being used to stop the reading of the user data when the number of bits representing the first logic value in the user data corresponds to the target value.

9. The method of claim 1, wherein counting begins from a maximum number of counter bits, and the maximum number is decremented each time a counter bit is read, until all of the counter bits are read.

10. The method of claim 1, wherein the second read voltage is a voltage offset relative to the first read voltage.

11. The method of claim 1, wherein the user data is protected using error correction codes (ECC) and / or differential cell schemes.

12. The method of claim 2, wherein when the memory cell is in the first logic state, the memory cell exhibits a threshold voltage of a lower magnitude, and when the memory cell is in the second logic state, the memory cell exhibits a threshold voltage of a higher magnitude, and wherein, in response to an applied first read voltage, the logic state of a given cell is determined based on whether the memory cell exhibits a higher or lower threshold voltage.

13. The method of claim 1, wherein the user data during operation is evaluated during the application of the first read voltage.

14. The method of claim 1, wherein the user data is evaluated after the target value is read.

15. The method of claim 1, further comprising the following steps: The target value is transferred to a lookup table (LUT) and data is received from the lookup table to generate the first read voltage to be applied to the memory cell.

16. A memory device comprising: - Memory cell array; - The memory cell array contains at least codewords of user data and count data, wherein the count data corresponds to the number of bits in the user data having a predetermined first logic value; - A memory controller that supports a counter-based method for reading the memory cells; as well as - Circuitry for reading the memory cells, the circuitry being operatively coupled to the memory cell array and comprising at least: - The access circuitry is configured as follows: - Apply a first read voltage to the memory cell to read the user data stored in the memory cell array; and - Applying a second read voltage to a unit of a counter to read the count data stored in the counter and provide a target value corresponding to the number of bits in the user data having the first logic value, wherein during the application of the read voltage, the count data and the user data are read simultaneously in a manner such that the target value is provided during the reading of the user data; and - Sensing circuit, configured as follows: - In response to the application of the first read voltage, a threshold voltage exhibited by the memory cell storing user data is detected. Based on the target value of the counter, the controller is configured to stop the application of the first read voltage when the number of bits in the user data having the first logic value corresponds to the target value.

17. The memory device of claim 16, wherein the sensing circuit is configured to detect a first current through a given memory cell in response to the first read voltage, and wherein the access circuit is configured to determine that the given memory cell is in a first logic state based on detecting that the magnitude of the first current is greater than or equal to a first threshold current.

18. The memory device of claim 16, wherein the first read voltage is a stepped voltage ramp.

19. The memory device of claim 18, wherein the second read voltage is a second voltage ramp offset relative to the stepped voltage ramp.

20. The memory device of claim 16, wherein the controller is configured to generate a command that allows the reading of the user data until the target value of the counter is read, and wherein, based on the target value, the controller is configured to stop the reading of the user data when the number of bits in the user data having the first logical value corresponds to the target value.

21. The memory device of claim 16, wherein the bits of the counter are inverted and the circuitry is configured to count from a maximum number of counter bits, wherein the maximum number is decremented each time a counter bit is read, until all of the counter bits are read.

22. The memory device of claim 16, configured to store a lookup table (LUT) containing information for generating the first read voltage.

23. The memory device of claim 16, wherein the memory cell array comprises a self-selected memory (SSM) or a 3D X-Point memory.

24. A system comprising: - Main unit; as well as - The memory device according to claim 16.

25. The system of claim 24, further comprising any one of a display communicatively coupled to the memory device or the host, a network interface communicatively coupled to the memory device or the host, and a battery coupled to provide power to the system.

26. A method for programming and reading memory cells in a vertical 3D memory device including a memory array, the method comprising the steps of: - Store codewords containing user data and count data in the memory array, wherein the count data corresponds to the number of bits in the user data that have a predetermined first logic value; - Store the difference between the target value and the minimum value of the bits in the user data that contain the first logical value stored in the codeword in the counter; - Apply a read voltage to the memory cell to read the user data stored in the memory array; - An offset read voltage is applied to a cell of the counter to read the count data stored in the counter and provide a target value corresponding to the number of bits in the user data having the first logic value, wherein during the application of the read voltage, the count data and the user data are read simultaneously in a manner such that the target value is provided during the reading of the user data; as well as - Based on the target value of the counter, the application of the read voltage is stopped when the number of bits in the user data having the first logic value corresponds to the target value.

Citation Information

Patent Citations

  • Nonvolatile memory storage system

    US20180286495A1

  • Auto-referenced memory cell read techniques

    US20190198099A1