The storage device comprises code such as a set of instructions, and the processor is operable to execute the set of instructions to provide the functionality described herein. Alternatively or additionally, the processor can access code from a storage device 126a of the memory structure, such as a reserved area of memory cells in one or more word lines.

For example, code can be used by the controller to access the memory structure such as for programming, read and erase operations. The code can include boot code and control code (e.g., set of instructions). The boot code is software that initializes the controller during a booting or startup process and enables the controller to access the memory structure. The code can be used by the controller to control one or more memory structures. Upon being powered up, the processor 122c fetches the boot code from the ROM 122a or storage device 126a for execution, and the boot code initializes the system components and loads the control code into the RAM 122b. Once the control code is loaded into the RAM, it is executed by the processor. The control code includes drivers to perform basic tasks such as controlling and allocating memory, prioritizing the processing of instructions, and controlling input and output ports.

Generally, the control code can include instructions to perform the functions described herein including the steps of the flowcharts discussed further below and provide the voltage waveforms including those discussed further below.

In one embodiment, the host is a computing device (e.g., laptop, desktop, smartphone, tablet, digital camera) that includes one or more processors, one or more processor readable storage devices (RAM, ROM, flash memory, hard disk drive, solid state memory) that store processor readable code (e.g., software) for programming the one or more processors to perform the methods described herein. The host may also include additional system memory, one or more input/output interfaces and/or one or more input/output devices in communication with the one or more processors.

Other types of non-volatile memory in addition to NAND flash memory can also be used.

Semiconductor memory devices include volatile memory devices, such as dynamic random access memory (“DRAM”) or static random access memory (“SRAM”) devices, non-volatile memory devices, such as resistive random access memory (“ReRAM”), electrically erasable programmable read only memory (“EEPROM”), flash memory (which can also be considered a subset of EEPROM), ferroelectric random access memory (“FRAM”), and magnetoresistive random access memory (“MRAM”), and other semiconductor elements capable of storing information. Each type of memory device may have different configurations. For example, flash memory devices may be configured in a NAND or a NOR configuration.

The memory devices can be formed from passive and/or active elements, in any combinations. By way of non-limiting example, passive semiconductor memory elements include ReRAM device elements, which in some embodiments include a resistivity switching storage element, such as an anti-fuse or phase change material, and optionally a steering element, such as a diode or transistor. Further by way of non-limiting example, active semiconductor memory elements include EEPROM and flash memory device elements, which in some embodiments include elements containing a charge storage region, such as a floating gate, conductive nanoparticles, or a charge storage dielectric material.

Multiple memory elements may be configured so that they are connected in series or so that each element is individually accessible. By way of non-limiting example, flash memory devices in a NAND configuration (NAND memory) typically contain memory elements connected in series. A NAND string is an example of a set of series-connected transistors comprising memory cells and SG transistors.

A NAND memory array may be configured so that the array is composed of multiple strings of memory in which a string is composed of multiple memory elements sharing a single bit line and accessed as a group. Alternatively, memory elements may be configured so that each element is individually accessible, e.g., a NOR memory array. NAND and NOR memory configurations are examples, and memory elements may be otherwise configured.

The semiconductor memory elements located within and/or over a substrate may be arranged in two or three dimensions, such as a two dimensional memory structure or a three dimensional memory structure.

In a two dimensional memory structure, the semiconductor memory elements are arranged in a single plane or a single memory device level. Typically, in a two dimensional memory structure, memory elements are arranged in a plane (e.g., in an x-y direction plane) which extends substantially parallel to a major surface of a substrate that supports the memory elements. The substrate may be a wafer over or in which the layer of the memory elements are formed or it may be a carrier substrate which is attached to the memory elements after they are formed. As a non-limiting example, the substrate may include a semiconductor such as silicon.

The memory elements may be arranged in the single memory device level in an ordered array, such as in a plurality of rows and/or columns. However, the memory elements may be arrayed in non-regular or non-orthogonal configurations. The memory elements may each have two or more electrodes or contact lines, such as bit lines and word lines.

A three dimensional memory array is arranged so that memory elements occupy multiple planes or multiple memory device levels, thereby forming a structure in three dimensions (i.e., in the x, y and z directions, where the z direction is substantially perpendicular and the x and y directions are substantially parallel to the major surface of the substrate).

As a non-limiting example, a three dimensional memory structure may be vertically arranged as a stack of multiple two dimensional memory device levels. As another non-limiting example, a three dimensional memory array may be arranged as multiple vertical columns (e.g., columns extending substantially perpendicular to the major surface of the substrate, i.e., in the y direction) with each column having multiple memory elements. The columns may be arranged in a two dimensional configuration, e.g., in an x-y plane, resulting in a three dimensional arrangement of memory elements with elements on multiple vertically stacked memory planes. Other configurations of memory elements in three dimensions can also constitute a three dimensional memory array.

By way of non-limiting example, in a three dimensional NAND memory array, the memory elements may be coupled together to form a NAND string within a single horizontal (e.g., x-y) memory device level. Alternatively, the memory elements may be coupled together to form a vertical NAND string that traverses across multiple horizontal memory device levels. Other three dimensional configurations can be envisioned wherein some NAND strings contain memory elements in a single memory level while other strings contain memory elements which span through multiple memory levels. Three dimensional memory arrays may also be designed in a NOR configuration and in a ReRAM configuration.

Typically, in a monolithic three dimensional memory array, one or more memory device levels are formed above a single substrate. Optionally, the monolithic three dimensional memory array may also have one or more memory layers at least partially within the single substrate. As a non-limiting example, the substrate may include a semiconductor such as silicon. In a monolithic three dimensional array, the layers constituting each memory device level of the array are typically formed on the layers of the underlying memory device levels of the array. However, layers of adjacent memory device levels of a monolithic three dimensional memory array may be shared or have intervening layers between memory device levels.

Then again, two dimensional arrays may be formed separately and then packaged together to form a non-monolithic memory device having multiple layers of memory. For example, non-monolithic stacked memories can be constructed by forming memory levels on separate substrates and then stacking the memory levels atop each other. The substrates may be thinned or removed from the memory device levels before stacking, but as the memory device levels are initially formed over separate substrates, the resulting memory arrays are not monolithic three dimensional memory arrays. Further, multiple two dimensional memory arrays or three dimensional memory arrays (monolithic or non-monolithic) may be formed on separate chips and then packaged together to form a stacked-chip memory device.

Associated circuitry is typically required for operation of the memory elements and for communication with the memory elements. As non-limiting examples, memory devices may have circuitry used for controlling and driving memory elements to accomplish functions such as programming and reading. This associated circuitry may be on the same substrate as the memory elements and/or on a separate substrate. For example, a controller for memory read-write operations may be located on a separate controller chip and/or on the same substrate as the memory elements.

One of skill in the art will recognize that this technology is not limited to the two dimensional and three dimensional exemplary structures described but covers all relevant memory structures within the spirit and scope of the technology as described herein and as understood by one of skill in the art.

FIG. 2 depicts blocks of memory cells in an example two-dimensional configuration of the memory array 126 of FIG. 1. The memory array can include many blocks. Each example block 200, 210 includes a number of NAND strings and respective bit lines, e.g., BL0, BL1, . . . which are shared among the blocks. Each NAND string is connected at one end to a drain select gate (SGD), and the control gates of the drain select gates are connected via a common SGD line. The NAND strings are connected at their other end to a source select gate which, in turn, is connected to a common source line 220. Sixteen word lines, for example, WL0-WL15, extend between the source select gates and the drain select gates. In some cases, dummy word lines, which contain no user data, can also be used in the memory array adjacent to the select gate transistors. Such dummy word lines can shield the edge data word line from certain edge effects.

One type of non-volatile memory which may be provided in the memory array is a floating gate memory. See FIGS. 3A and 3B. Other types of non-volatile memory can also be used. For example, a charge-trapping memory cell uses a non-conductive dielectric material in place of a conductive floating gate to store charge in a non-volatile manner. See FIGS. 4A and 4B. A triple layer dielectric formed of silicon oxide, silicon nitride and silicon oxide (“ONO”) is sandwiched between a conductive control gate and a surface of a semi-conductive substrate above the memory cell channel. The cell is programmed by injecting electrons from the cell channel into the nitride, where they are trapped and stored in a limited region. This stored charge then changes the threshold voltage of a portion of the channel of the cell in a manner that is detectable. The cell is erased by injecting hot holes into the nitride. A similar cell can be provided in a split-gate configuration where a doped polysilicon gate extends over a portion of the memory cell channel to form a separate select transistor.

In another approach, NROM cells are used. Two bits, for example, are stored in each NROM cell, where an ONO dielectric layer extends across the channel between source and drain diffusions. The charge for one data bit is localized in the dielectric layer adjacent to the drain, and the charge for the other data bit localized in the dielectric layer adjacent to the source. Multi-state data storage is obtained by separately reading binary states of the spatially separated charge storage regions within the dielectric. Other types of non-volatile memory are also known.

FIG. 3A depicts a cross-sectional view of example floating gate memory cells in NAND strings. A bit line or NAND string direction goes into the page, and a word line direction goes from left to right. As an example, word line 324 extends across NAND strings which include respective channel regions 306, 316 and 326. The memory cell 300 includes a control gate 302, a floating gate 304, a tunnel oxide layer 305 and the channel region 306. The memory cell 310 includes a control gate 312, a floating gate 314, a tunnel oxide layer 315 and the channel region 316. The memory cell 320 includes a control gate 322, a floating gate 321, a tunnel oxide layer 325 and the channel region 326. Each memory cell is in a different respective NAND string. An inter-poly dielectric (IPD) layer 328 is also depicted. The control gates are portions of the word line. A cross-sectional view along line 329 is provided in FIG. 3B.

The control gate wraps around the floating gate, increasing the surface contact area between the control gate and floating gate. This results in higher IPD capacitance, leading to a higher coupling ratio which makes programming and erase easier. However, as NAND memory devices are scaled down, the spacing between neighboring cells becomes smaller so there is almost no space for the control gate and the IPD between two adjacent floating gates. As an alternative, as shown in FIGS. 4A and 4B, the flat or planar memory cell has been developed in which the control gate is flat or planar; that is, it does not wrap around the floating gate, and its only contact with the charge storage layer is from above it. In this case, there is no advantage in having a tall floating gate. Instead, the floating gate is made much thinner. Further, the floating gate can be used to store charge, or a thin charge trap layer can be used to trap charge. This approach can avoid the issue of ballistic electron transport, where an electron can travel through the floating gate after tunneling through the tunnel oxide during programming.

FIG. 3B depicts a cross-sectional view of the structure of FIG. 3A along line 329. The NAND string 330 includes an SGS transistor 331, example memory cells 300, 333, . . . , 334 and 335, and an SGD transistor 336. The memory cell 300, as an example of each memory cell, includes the control gate 302, the IPD layer 328, the floating gate 304 and the tunnel oxide layer 305, consistent with FIG. 3A. Passageways in the IPD layer in the SGS and SGD transistors allow the control gate layers and floating gate layers to communicate. The control gate and floating gate layers may be polysilicon and the tunnel oxide layer may be silicon oxide, for instance. The IPD layer can be a stack of nitrides (N) and oxides (O) such as in a N—O—N—O—N configuration.

The NAND string may be formed on a substrate which comprises a p-type substrate region 355, an n-type well 356 and a p-type well 357. N-type source/drain diffusion regions sd1, sd2, sd3, sd4, sd5, sd6 and sd7 are formed in the p-type well. A channel voltage, Vch, may be applied directly to the channel region of the substrate.

FIG. 4A depicts a cross-sectional view of example charge-trapping memory cells in NAND strings. The view is in a word line direction of memory cells comprising a flat control gate and charge-trapping regions as a 2D example of memory cells in the memory cell array 126 of FIG. 1A. Charge-trapping memory can be used in NOR and NAND flash memory device. This technology uses an insulator such as an SiN film to store electrons, in contrast to a floating-gate MOSFET technology which uses a conductor such as doped polycrystalline silicon to store electrons. As an example, a word line (WL) 424 extends across NAND strings which include respective channel regions 406, 416 and 426. Portions of the word line provide control gates 402, 412 and 422. Below the word line is an IPD layer 428, charge-trapping layers 404, 414 and 421, polysilicon layers 405, 415 and 425 and tunneling layer layers 409, 407 and 408. Each charge-trapping layer extends continuously in a respective NAND string.

A memory cell 400 includes the control gate 402, the charge-trapping layer 404, the polysilicon layer 405 and a portion of the channel region 406. A memory cell 410 includes the control gate 412, the charge-trapping layer 414, a polysilicon layer 415 and a portion of the channel region 416. A memory cell 420 includes the control gate 422, the charge-trapping layer 421, the polysilicon layer 425 and a portion of the channel region 426.

A flat control gate is used here instead of a control gate that wraps around a floating gate. One advantage is that the charge-trapping layer can be made thinner than a floating gate. Additionally, the memory cells can be placed closer together.

FIG. 4B depicts a cross-sectional view of the structure of FIG. 4A along line 429. The view shows a NAND string 430 having a flat control gate and a charge-trapping layer. The NAND string 430 includes an SGS transistor 431, example memory cells 400, 433, . . . , 434 and 435, and an SGD transistor 435.

The NAND string may be formed on a substrate which comprises a p-type substrate region 455, an n-type well 456 and a p-type well 457. N-type source/drain diffusion regions sd1, sd2, sd3, sd4, sd5, sd6 and sd7 are formed in the p-type well 457. A channel voltage, Vch, may be applied directly to the channel region of the substrate. The memory cell 400 includes the control gate 402 and the IPD layer 428 above the charge-trapping layer 404, the polysilicon layer 405, the tunneling layer 409 and the channel region 406.

The control gate layer may be polysilicon and the tunneling layer may be silicon oxide, for instance. The IPD layer can be a stack of high-k dielectrics such as AlOx or HfOx which help increase the coupling ratio between the control gate layer and the charge-trapping or charge storing layer. The charge-trapping layer can be a mix of silicon nitride and oxide, for instance.

The SGD and SGS transistors have the same configuration as the memory cells but with a longer channel length to ensure that current is cutoff in an inhibited NAND string.

In this example, the layers 404, 405 and 409 extend continuously in the NAND string. In another approach, portions of the layers 404, 405 and 409 which are between the control gates 402, 412 and 422 can be removed, exposing a top surface of the channel 406.

FIG. 5A depicts an example block diagram of the sense block SB1 of FIG. 1A. In one approach, a sense block comprises multiple sense circuits. Each sense circuit is associated with data latches. For example, the example sense circuits 550a, 551a, 552a and 553a are associated with the data latches 550b, 551b, 552b and 553b, respectively. In one approach, different subsets of bit lines can be sensed using different respective sense blocks. This allows the processing load which is associated with the sense circuits to be divided up and handled by a respective processor in each sense block. For example, a sense circuit controller 560 in SB1 can communicate with the set of sense circuits and latches. The sense circuit controller may include a pre-charge circuit 561 which provides a voltage to each sense circuit for setting a pre-charge voltage. In one possible approach, the voltage is provided to each sense circuit independently, e.g., via the data base 503 and a local bus such as LBUS1 or LBUS2 in FIG. 5B. In another possible approach, a common voltage is provided to each sense circuit concurrently, e.g., via the line 505 in FIG. 5B. The sense circuit controller may also include a memory 562 and a processor 563. As mentioned also in connection with FIG. 2, the memory 562 may store code which is executable by the processor to perform the functions described herein. These functions can include reading latches which are associated with the sense circuits, setting bit values in the latches and providing voltages for setting pre-charge levels in sense nodes of the sense circuits. Further example details of the sense circuit controller and the sense circuits 550a and 551a are provided below.

FIG. 5B depicts another example block diagram of the sense block SB1 of FIG. 1A. The sense circuit controller 560 communicates with multiple sense circuits including example sense circuits 550a and 551a, also shown in FIG. 5A. The sense circuit 550a includes latches 550b, including a trip latch 526, an offset verify latch 527 and data state latches 528. The sense circuit further includes a voltage clamp 521 such as a transistor which sets a pre-charge voltage at a sense node 522. A sense node to bit line (BL) switch 523 selectively allows the sense node to communicate with a bit line 525, e.g., the sense node is electrically connected to the bit line so that the sense node voltage can decay. The bit line 525 is connected to one or more memory cells such as a memory cell MC1. A voltage clamp 524 can set a voltage on the bit line, such as during a sensing operation or during a program voltage. A local bus, LBUS1, allows the sense circuit controller to communicate with components in the sense circuit, such as the latches 550b and the voltage clamp in some cases. To communicate with the sense circuit 550a, the sense circuit controller provides a voltage via a line 502 to a transistor 504 to connect LBUS1 with a data bus DBUS, 503. The communicating can include sending data to the sense circuit and/or receive data from the sense circuit.

The sense circuit controller can communicate with different sense circuits in a time-multiplexed manner, for instance. A line 505 may be connected to the voltage clamp in each sense circuit, in one approach.

The sense circuit 551a includes latches 551b, including a trip latch 546, an offset verify latch 547 and data state latches 548. A voltage clamp 541 may be used to set a pre-charge voltage at a sense node 542. A sense node to bit line (BL) switch 543 selectively allows the sense node to communicate with a bit line 545, and a voltage clamp 544 can set a voltage on the bit line. The bit line 545 is connected to one or more memory cells such as a memory cell MC2. A local bus, LBUS2, allows the sense circuit controller to communicate with components in the sense circuit, such as the latches 551b and the voltage clamp in some cases. To communicate with the sense circuit 551a, the sense circuit controller provides a voltage via a line 501 to a transistor 506 to connect LBUS2 with DBUS.

The sense circuit 550a may be a first sense circuit which comprises a first trip latch 526 and the sense circuit 551a may be a second sense circuit which comprises a second trip latch 546.

The sense circuit 550a is an example of a first sense circuit comprising a first sense node 522, where the first sense circuit is associated with a first memory cell MC1 and a first bit line 525. The sense circuit 551a is an example of a second sense circuit comprising a second sense node 542, where the second sense circuit is associated with a second memory cell MC2 and a second bit line 545.

FIG. 6A is a perspective view of a set of blocks 600 in an example three-dimensional configuration of the memory array 126 of FIG. 1A. On the substrate are example blocks BLK0, BLK1, BLK2 and BLK3 of memory cells (storage elements) and a peripheral area 604 with circuitry for use by the blocks. For example, the circuitry can include voltage drivers 605 which can be connected to control gate layers of the blocks. In one approach, control gate layers at a common height in the blocks are commonly driven. The substrate 601 can also carry circuitry under the blocks, along with one or more lower metal layers which are patterned in conductive paths to carry signals of the circuitry. The blocks are formed in an intermediate region 602 of the memory device. In an upper region 603 of the memory device, one or more upper metal layers are patterned in conductive paths to carry signals of the circuitry. Each block comprises a stacked area of memory cells, where alternating levels of the stack represent word lines. In one possible approach, each block has opposing tiered sides from which vertical contacts extend upward to an upper metal layer to form connections to conductive paths. While four blocks are depicted as an example, two or more blocks can be used, extending in the x- and/or y-directions.

In one possible approach, the length of the plane, in the x-direction, represents a direction in which signal paths to word lines extend in the one or more upper metal layers (a word line or SGD line direction), and the width of the plane, in the y-direction, represents a direction in which signal paths to bit lines extend in the one or more upper metal layers (a bit line direction). The z-direction represents a height of the memory device.

FIG. 6B depicts an example cross-sectional view of a portion of one of the blocks of FIG. 6A. The block comprises a stack 610 of alternating conductive and dielectric layers. In this example, the conductive layers comprise two SGD layers, two SGS layers and four dummy word line layers DWLD0, DWLD1, DWLS0 and DWLS1, in addition to data word line layers (word lines) WLL0-WLL10. The dielectric layers are labelled as DL0-DL19. Further, regions of the stack which comprise NAND strings NS1 and NS2 are depicted. Each NAND string encompasses a memory hole 618 or 619 which is filled with materials which form memory cells adjacent to the word lines. A region 622 of the stack is shown in greater detail in FIG. 6D.

The stack includes a substrate 611, an insulating film 612 on the substrate, and a portion of a source line SL. NS1 has a source-end 613 at a bottom 614 of the stack and a drain-end 615 at a top 616 of the stack. Metal-filled slits 617 and 620 may be provided periodically across the stack as interconnects which extend through the stack, such as to connect the source line to a line above the stack. The slits may be used during the formation of the word lines and subsequently filled with metal. A portion of a bit line BL0 is also depicted. A conductive via 621 connects the drain-end 615 to BL0.

FIG. 6C depicts a plot of memory hole diameter in the stack of FIG. 6B. The vertical axis is aligned with the stack of FIG. 6B and depicts a width (wMH), e.g., diameter, of the memory holes 618 and 619. The word line layers WLL0-WLL10 of FIG. 6A are repeated as an example and are at respective heights z0-z10 in the stack. In such a memory device, the memory holes which are etched through the stack have a very high aspect ratio. For example, a depth-to-diameter ratio of about 25-30 is common. The memory holes may have a circular cross-section. Due to the etching process, the memory hole width can vary along the length of the hole. Typically, the diameter becomes progressively smaller from the top to the bottom of the memory hole. That is, the memory holes are tapered, narrowing at the bottom of the stack. In some cases, a slight narrowing occurs at the top of the hole near the select gate so that the diameter becomes slight wider before becoming progressively smaller from the top to the bottom of the memory hole.

Due to the non-uniformity in the width of the memory hole, the programming speed, including the program slope and erase speed of the memory cells can vary based on their position along the memory hole, e.g., based on their height in the stack. With a smaller diameter memory hole, the electric field across the tunnel oxide is relatively stronger, so that the programming and erase speed is relatively higher. One approach is to define groups of adjacent word lines for which the memory hole diameter is similar, e.g., within a defined range of diameter, and to apply an optimized verify scheme for each word line in a group. Different groups can have different optimized verify schemes.

FIG. 6D depicts a close-up view of the region 622 of the stack of FIG. 6B. Memory cells are formed at the different levels of the stack at the intersection of a word line layer and a memory hole. In this example, SGD transistors 680 and 681 are provided above dummy memory cells 682 and 683 and a data memory cell MC. A number of layers can be deposited along the sidewall (SW) of the memory hole 630 and/or within each word line layer, e.g., using atomic layer deposition. For example, each column (e.g., the pillar which is formed by the materials within a memory hole) can include a charge-trapping layer or film 663 such as SiN or other nitride, a tunneling layer 664, a polysilicon body or channel 665, and a dielectric core 666. A word line layer can include a blocking oxide/block high-k material 660, a metal barrier 661, and a conductive metal 662 such as Tungsten as a control gate. For example, control gates 690, 691, 692, 693 and 694 are provided. In this example, all of the layers except the metal are provided in the memory hole. In other approaches, some of the layers can be in the control gate layer. Additional pillars are similarly formed in the different memory holes. A pillar can form a columnar active area (AA) of a NAND string.

When a memory cell is programmed, electrons are stored in a portion of the charge-trapping layer which is associated with the memory cell. These electrons are drawn into the charge-trapping layer from the channel, and through the tunneling layer. The Vth of a memory cell is increased in proportion to the amount of stored charge. During an erase operation, the electrons return to the channel.

Each of the memory holes can be filled with a plurality of annular layers comprising a blocking oxide layer, a charge trapping layer, a tunneling layer and a channel layer. A core region of each of the memory holes is filled with a body material, and the plurality of annular layers are between the core region and the word line in each of the memory holes.

The NAND string can be considered to have a floating body channel because the length of the channel is not formed on a substrate. Further, the NAND string is provided by a plurality of word line layers above one another in a stack, and separated from one another by dielectric layers.

FIG. 7A depicts a top view of an example word line layer WLL0 of the stack of FIG. 6B. As mentioned, a 3D memory device can comprise a stack of alternating conductive and dielectric layers. The conductive layers provide the control gates of the SG transistors and memory cells. The layers used for the SG transistors are SG layers and the layers used for the memory cells are word line layers. Further, memory holes are formed in the stack and filled with a charge-trapping material and a channel material. As a result, a vertical NAND string is formed. Source lines are connected to the NAND strings below the stack and bit lines are connected to the NAND strings above the stack.

A block BLK in a 3D memory device can be divided into sub-blocks, where each sub-block comprises a set of NAND string which have a common SGD control line. For example, see the SGD lines/control gates SGD0, SGD1, SGD2 and SGD3 in the sub-blocks SBa, SBb, SBc and SBd, respectively. Further, a word line layer in a block can be divided into regions. Each region is in a respective sub-block are can extend between slits which are formed periodically in the stack to process the word line layers during the fabrication process of the memory device. This processing can include replacing a sacrificial material of the word line layers with metal. Generally, the distance between slits should be relatively small to account for a limit in the distance that an etchant can travel laterally to remove the sacrificial material, and that the metal can travel to fill a void which is created by the removal of the sacrificial material. For example, the distance between slits may allow for a few rows of memory holes between adjacent slits. The layout of the memory holes and slits should also account for a limit in the number of bit lines which can extend across the region while each bit line is connected to a different memory cell. After processing the word line layers, the slits can optionally be filed with metal to provide an interconnect through the stack.

This figures and other are not necessarily to scale. In practice, the regions can be much longer in the x-direction relative to the y-direction than is depicted to accommodate additional memory holes.

In this example, there are four rows of memory holes between adjacent slits. A row here is a group of memory holes which are aligned in the x-direction. Moreover, the rows of memory holes are in a staggered pattern to increase the density of the memory holes. The word line layer or word line is divided into regions WLL0 a, WLL0 b, WLL0 c and WLL0 d which are each connected by a connector 713. The last region of a word line layer in a block can be connected to a first region of a word line layer in a next block, in one approach. The connector, in turn, is connected to a voltage driver for the word line layer. The region WLL0 a has example memory holes 710 and 711 along a line 712. The region WLL0 b has example memory holes 714 and 715. The region WLL0 chas example memory holes 716 and 717. The region WLL0 dhas example memory holes 718 and 719. The memory holes are also shown in FIG. 7B. Each memory hole can be part of a respective NAND string. For example, the memory holes 710, 714, 716 and 718 can be part of NAND strings NS0_SBa, NS0_SBb, NS0_SBc and NS0_SBd, respectively.

Each circle represents the cross-section of a memory hole at a word line layer or SG layer. Example circles shown with dashed lines represent memory cells which are provided by the materials in the memory hole and by the adjacent word line layer. For example, memory cells 720 and 721 are in WLL0 a, memory cells 724 and 725 are in WLL0 b, memory cells 726 and 727 are in WLL0 c, and memory cells 728 and 729 are in WLL0 d. These memory cells are at a common height in the stack.

Metal-filled slits 701, 702, 703 and 704 (e.g., metal interconnects) may be located between and adjacent to the edges of the regions WLL0 a-WLL0 d. The metal-filled slits provide a conductive path from the bottom of the stack to the top of the stack. For example, a source line at the bottom of the stack may be connected to a conductive line above the stack, where the conductive line is connected to a voltage driver in a peripheral region of the memory device. See also FIG. 8A for further details of the sub-blocks SBa-SBd of FIG. 7A.

FIG. 7B depicts a top view of an example top dielectric layer DL19 of the stack of FIG. 6B. The dielectric layer is divided into regions DL19 a, DL19 b, DL19 c and DL19 d. Each region can be connected to a respective voltage driver. This allows a set of memory cells in one region of a word line layer to be programmed concurrently, with each memory cell being in a respective NAND string which is connected to a respective bit line. A voltage can be set on each bit line to allow or inhibit programming during each program voltage.

The region DL19 a has the example memory holes 710 and 711 along a line 712a which is coincident with a bit line BL0. A number of bit lines extend above the memory holes and are connected to the memory holes as indicated by the “X” symbols. BL0 is connected to a set of memory holes which includes the memory holes 711, 715, 717 and 719. Another example bit line BL1 is connected to a set of memory holes which includes the memory holes 710, 714, 716 and 718. The metal-filled slits 701, 702, 703 and 704 from FIG. 7A are also depicted, as they extend vertically through the stack. The bit lines can be numbered in a sequence BL0-BL23 across the DL19 layer in the −x direction.

Different subsets of bit lines are connected to cells in different rows. For example, BL0, BL4, BL8, BL12, BL16 and BL20 are connected to cells in a first row of cells at the right hand edge of each region. BL2, BL6, BL10, BL14, BL18 and BL22 are connected to cells in an adjacent row of cells, adjacent to the first row at the right hand edge. BL3, BL7, BL11, BL15, BL19 and BL23 are connected to cells in a first row of cells at the left hand edge of each region. BL1, BL5, BL9, BL13, BL17 and BL21 are connected to cells in an adjacent row of cells, adjacent to the first row at the left hand edge.

FIG. 8A depicts example NAND strings in the sub-blocks SBa-SBd of FIG. 7A. The sub-blocks are consistent with the structure of FIG. 6B. The conductive layers in the stack are depicted for reference at the left hand side. Each sub-block includes multiple NAND strings, where one example NAND string is depicted. For example, SBa comprises an example NAND string NS0_SBa, SBb comprises an example NAND string NS0_SBb, SBc comprises an example NAND string NS0_SBc, and SBd comprises an example NAND string NS0_SBd.

Additionally, NS0_SBa include SGS transistors 800 and 801, dummy memory cells 802 and 803, data memory cells 804, 805, 806, 807, 808, 809, 810, 811, 812, 813 and 814, dummy memory cells 815 and 816, and SGD transistors 817 and 818.

NS0_SBb include SGS transistors 820 and 821, dummy memory cells 822 and 823, data memory cells 824, 825, 826, 827, 828, 829, 830, 831, 832, 833 and 834, dummy memory cells 835 and 836, and SGD transistors 837 and 838.

NS0_SBc include SGS transistors 840 and 841, dummy memory cells 842 and 843, data memory cells 844, 845, 846, 847, 848, 849, 850, 851, 852, 853 and 854, dummy memory cells 855 and 856, and SGD transistors 857 and 858.

NS0_SBd include SGS transistors 860 and 861, dummy memory cells 862 and 863, data memory cells 864, 865, 866, 867, 868, 869, 870, 871, 872, 873 and 874, dummy memory cells 875 and 876, and SGD transistors 877 and 878.

At a given height in the block, a set of memory cells in each sub-block are at a common height. For example, one set of memory cells (including the memory cell 804) is among a plurality of memory cells formed along tapered memory holes in a stack of alternating conductive and dielectric layers. The one set of memory cells is at a particular height z0 in the stack. Another set of memory cells (including the memory cell 824) connected to the one word line (WLL0) are also at the particular height. In another approach, the another set of memory cells (e.g., including the memory cell 812) connected to another word line (e.g., WLL8) are at another height (z8) in the stack.

FIG. 8B depicts another example view of NAND strings in sub-blocks. The NAND strings includes NS0_SBa, NS0_SBb, NS0_SBc and NS0_SBd, which have 48 word lines, WL0-WL47, in this example. Each sub-block comprises a set of NAND strings which extend in the x direction and which have a common SGD line, e.g., SGD0, SGD1, SGD2 or SGD3. In this simplified example, there is only one SGD transistor and one SGS transistor in each NAND string. The NAND strings NS0_SBa, NS0_SBb, NS0_SBc and NS0_SBd are in sub-blocks SBa, SBb, SBc and SBd, respectively. Further, example, groups of word lines G0, G1 and G2 are depicted.

Particularly in a three-dimensional memory arrangement, the natural Vth distribution can vary for different word lines within a memory block because the layers may have differing memory hole diameters. Thus, it is advantageous to perform a smart verify operation during programming in order to determine the optimal programming voltage for any given word line and avoid either under programming or over programming.

FIG. 12 is a flowchart of an example programming operation in a memory device. Step 1200 begins a programming operation for a set of word lines or for at least one word line. Step 1202 selects a page to program (for example, word line [WL_n]) and sets Vpgm to a predetermined initial program voltage (Vpgmint). In one example, Vpgmint is fourteen Volts (14 V).

Step 1204 performs a program-verify iteration which includes applying Vpgm to at least some of the cells coupled to WL_nin a programming operation and applying a verification signal (e.g., a voltage waveform) to WL_nwhile performing verify tests, e.g., for one or more data states. The cells WL_nwhich the program-verify iteration is applied to are preferably up to a predetermined checkpoint within WL_n. The predetermined checkpoint is preferably near the lowest state threshold voltage.

Decision step 1206 determines if acquisition of smart verify is completed, e.g., based on the results of the verify tests of step 1204. The acquisition may be done, for example, if all or nearly all of the cells of the word line have been programmed to the predetermined checkpoint within WL_n. If decision step 1206 is false, then at step 1208, Vpgm is incrementally increased (i.e., stepped up) by a voltage step amount in the form of a predetermined delta program voltage (dVpgm), i.e., Vpgm is set to Vpgm+dVpgm. In one example, dVpgm is one half of a Volt (0.5 V). Next, the programming operation returns to step 1204 to conduct the next program-verify iteration. Each repetition of step 1204 until decision step 1206 is true is one smart-verify loop, which requires resources from the controller.

If decision step 1206 is true (the acquisition of smart verify was completed), then at step 1209, the remaining cells of WL_nare programmed using the newly acquired Vpgm as a starting program voltage and any known program-verify sequence. A decision step 1210 then determines if there is a next word line to program. If decision step 1210 is false (there are no additional word lines to program), then the programming operation ends at step 1212. If decision step 1210 is true, then at step 1214, WL_nis incrementally advanced to the next word line to be programmed, i.e., WL_nis set to WL_n+1.

Step 1216 performs a programming operation which includes applying Vpgm to WL_n. The Vpgm which is applied at step 1216 is the Vpgm which ultimately was successful at step 1206, and thus, Vpgm at step 1216 may be equal to or greater than Vpgmint.

Decision step 1218 determines if there is a next word line to program. If decision step 1218 is false (there are no additional word lines to program), then the programming operation ends at step 1212. If decision step 1210 is true, then at step 1220, WL_nis incrementally advanced to the next word line to be programmed (WL_nis set to WL_n+1) and Vpgm is incrementally decreased (stepped down) by dVpgm (Vpgm is set to Vpgmint). The programming operation then returns to the program-verify operation at step 1202. In alternate embodiments, steps 1214, 1216, and 1218 may be repeated one or more times such that two or more wordlines are programmed using Vpgm prior to proceeding back to the program-verify operation of step 1202.

The following table illustrates an example plot showing the programming operation of FIG. 12 as applied to ten exemplary word lines (WL₁-WL₁₀).

WL
Start Vpgm
Acquired Vpgm

1
Vpgmint
Vpgm(WL₁)

2
Vpgm(WL₁)
None

3
Vpgmint
Vpgm(WL₃)

4
Vpgm(WL₃)
None

5
Vpgmint
Vpgm(WL₅)

6
Vpgm(WL₅)
None

7
Vpgmint
Vpgm(WL₇)

8
Vpgm(WL₇)
None

9
Vpgmint
Vpgm(WL₉)

10
Vpgm(WL₉)
None

In the above example, the cumulative number of smart verify loops which must be completed to successfully program these ten word lines is reduced as compared to other known programming operations which utilize a smart verify operation to minimize under programming and over programming. In other words, the programming operation optimizes the time to acquire the optimal programming voltages. Thus, the overall performance and the endurance of the memory device are both improved.

Again, smart verify can help reduce program time by obtaining an acquired program voltage or smart verify programming voltage VPGM_SV from a sampling string to improve program speed during the programming of subsequent strings. FIG. 13 shows one smart verify algorithm used for triple level cell (TLC) programming. During TLC programming, smart verify is performed on sampling string (WL_nstring0) to obtain the smart verify programming voltage VPGM_SV. The obtained smart verify programming voltage VPGM_SV is then used for several following strings (WL_nstring1/2/3) as an initial program voltage VPGM. In a multi-plane mode, smart verify will adopt the acquired minimum smart verify programming voltage VPGM_SV (based on a minimum number of smart verify loops).

In addition, word line skip smart verify mode can be used to further reduce program time tProg (i.e., time required for the program operation). FIG. 14 shows an example smart verify operation utilizing word line skip smart verify. As shown, smart verify is performed on WL2 n string0 and the obtained smart verify programming voltage VPGM_SV will be applied to WL2 n string1/2/3 and W2 n+1 string0/1/2/3. In addition, smart verify on WL4 n string0 and used for WL4 n string1/2/3 and WL4 n+1/2/3 string0/1/2/3 may also be used to further reduce the program time tProg.

As discussed above, smart verify can help reduce program time by obtaining an acquired program voltage or smart verify programming voltage from a sampling string to improve program speed during the programming of subsequent strings. Yet, when a programming workload involves memory cells of different dies and/or multiple blocks, program performance may actually end up being less than if smart verify is not used at all. In more detail, some memory apparatuses always utilize an Auto Smart Verify (ASV) feature. With ASV enabled (ASV=en), during programming of a selected word line WL_n, an optimum program voltage VPGM is acquired for the selected word line and first string WL_n/Str0 and used for remaining WL_n-Str1/2/3/4 programming. However, during system workload, there are scenarios where internally system is jumping from one block programming to another block programming. This is known as block jumping. FIG. 15 shows a plot of program time tPROG with smart verify disabled or enabled with and without block jumping. In the example shown, there is nearly 10 MBps program time tPROG improvement with ASV=en (Power on Rest, POR) as opposed to ASV=dis (reference, Ref). In memory apparatuses utilizing ASV, with program block jumping among different dies and multiple blocks, the acquired VPGM information during ASV programming is lost. Thus, when the block programming resumes, it begins with starting program voltage VPGM of the trim. Accordingly, still referring to FIG. 15, in a worst case block jumping system workload scenario, there is block jumping after every string programming and results in a 15 MBps performance loss which is 5 MBps worse than ASV=dis (ref). This defeats the purpose of using ASV.

FIG. 16 shows a plot of program time tPROG with smart verify disabled or enabled with and without block jumping every string or every two strings. As discussed above with respect to FIG. 15, ASV enabled (POR) compared to ASV disabled (Ref) gains 10 MBps with no block jumping. And block jumping every string loses 5 MBps (worst case) with ASV enabled compared to ASV disabled. However, as shown in FIG. 16, block jumping every 2 strings gains 3 MBps (intermediate case). To ensure ASV does not pose any disadvantage, a solution is desired to handle the worst-case system workload for block jumping.

Consequently, described herein is a memory apparatus (e.g., memory device 100 of FIG. 1A) including memory cells (e.g., data memory cell MC of FIG. 6D) grouped into a plurality of blocks (e.g., block BLK of FIG. 7A). The memory apparatus also includes a control circuit or means (e.g., one or any combination of control circuitry 110, decoders 124/132, power control module 116, sense blocks SB1, SB2, . . . ,SBp, read/write circuits 128, controller 122 of FIG. 1A, control circuit 150 of FIG. 1B, and/or sense circuit controller 560 of FIG. 5A and so forth). The control means is configured to begin programming a set of the memory cells in a programming workload. The control means is also configured to skip acquiring a smart verify program voltage used while programming the set of the memory cells in response to determining ones of the memory cells of the set being disposed in one of the plurality of blocks different than others of the memory cells being programmed during the programming workload.

As above, the memory cells may be connected to one of a plurality of word lines (e.g., data word line layers (word lines) WLL0-WLL10 of FIG. 6B or WLL10 of FIG. 6D). The memory cells can also be disposed in memory holes (e.g., memory holes 618 and 619 of FIG. 6B) and are configured to retain a threshold voltage Vt or Vth corresponding to one of a plurality of memory or data states (FIGS. 9-11). The memory holes are organized in rows grouped in a plurality of strings Str0, Str1, Str2, Str3 (e.g., sub-blocks SBa-SBd of FIG. 7A). Thus, according to an aspect and as discussed above, the control means is further configured to determine the smart verify program voltage in a smart verify operation by performing at least one program-verify iteration of at least some of the memory cells connected to one of the plurality of word lines and associated with a first one of the plurality of strings using a voltage value. The voltage value starts as a predetermined first initial voltage and is sequentially increased in each of the plurality of smart verify loops by a voltage step amount following each failure to successfully program. The voltage value increases until programming of the memory cells connected to the one of the plurality of word lines and associated with the first one of the plurality of strings is completed. The smart verify program voltage is the voltage value when the programming of the memory cells connected to the one of the plurality of word lines and associated with the first one of the plurality of strings is completed. The control means is also configured to program at least one of the memory cells connected to at least another one of the plurality of word lines and the memory cells associated with at least another one of the plurality of strings using the smart verify programming voltage acquired in the smart verify operation as a starting program voltage for programming.

According to an aspect, the set of the memory cells comprises the memory cells connected to one of the plurality of word lines and disposed in two of the plurality of strings. Nevertheless, it should be understood that the set of the memory cells is selected to prevent worse programming performance than if no smart verify operation is used at all, thus, sets other than the memory cells connected to one of the plurality of word lines and disposed in two of the plurality of strings are contemplated.

According to another aspect, the memory apparatus can further include a smart verify enable parameter SV_ON_OFF. Accordingly, the control means may be further configured to trigger acquiring the smart verify program voltage used while programming the set of the memory cells in response to the smart verify enable parameter SV_ON_OFF being enabled (e.g., SV_ON_OFF=1h) and disable acquiring the smart verify program voltage used while programming the set of the memory cells in response to the smart verify enable parameter SV_ON_OFF being disabled (e.g., SV_ON_OFF=0h). So, the control means disables the auto smart verify (ASV) when the workload shows block jumping on every string on-the-fly. This ensures that the programming performance of the memory apparatus disclosed herein never goes below the ASV disabled performance.

So, in more detail and according to additional aspects, the control means is further configured to determine whether the ones of the memory cells connected to the one of the plurality of word lines and associated with the two of the plurality of strings are disposed in a single one of the plurality of blocks during the programming workload. The control means is additionally configured to set the smart verify enable parameter SV_ON_OFF to be enabled and acquire the smart verify program voltage and use the smart verify program voltage to program the ones of the memory cells connected to the one of the plurality of word lines and associated with the two of the plurality of strings in response to determining the ones of the memory cells connected to the one of the plurality of word lines and associated with the two of the plurality of strings are disposed in the single one of the plurality of blocks during the programming workload. The control means is also configured to return to determine whether the ones of the memory cells connected to the one of the plurality of word lines and associated with the two of the plurality of strings are disposed in the single one of the plurality of blocks for subsequent ones of the memory cells connected to a next one of the plurality of word lines and associated with a next two of the plurality of strings during the programming workload.

According to other aspects, the control means is further configured to set the smart verify enable parameter SV_ON_OFF to be disabled in response to determining the ones of the memory cells connected to the one of the plurality of word lines and associated with the two of the plurality of strings are not disposed in the single one of the plurality of blocks during the programming workload. The control means skips acquiring the smart verify program voltage and not use the smart verify program voltage to program the ones of the memory cells connected to the one of the plurality of word lines and associated with the two of the plurality of strings. In addition, the control means is configured to set the smart verify enable parameter SV_ON_OFF to be enabled and return to determine whether the ones of the memory cells connected to the one of the plurality of word lines and associated with the two of the plurality of strings are disposed in the single one of the plurality of blocks for the subsequent ones of the memory cells connected to the next one of the plurality of word lines and associated with the next two of the plurality of strings during the programming workload.

FIG. 17 shows a plot of program time tPROG with smart verify disabled or enabled with and without block jumping every string or every two strings along with worst case performance of the memory apparatus disclosed herein (indicated with a star). So, while the system or memory apparatus workload (including garbage collection, wear leveling etc. background operations along with host programming) could see worst-case scenario of block jumping after every string programming, the memory apparatus and method of operation disclosed herein do not allow the programming performance to be less than if smart verify is not used. For any system workload, including the worst-case scenario, performance will never go below ASV disabled performance.