Semiconductor device including high-speed heterogeneous integrated controller and cache
By adopting heterogeneous integrated logic circuits and cache structures in the 3D memory structure and simplifying the process through silicon through-silicon stacking, the problems of difficulty in logic circuit positioning and unfavorable process optimization are solved, and memory access speed and device performance are improved.
Patent Information
- Application Number
- CN201910575625.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2019-06-28
- Publication Date
- 2024-12-03
- Estimated Expiration
- 2039-06-28
AI Technical Summary
In a 3D memory structure, as the number of memory layers increases, logic circuit positioning becomes difficult, and existing process optimizations are detrimental to the formation of memory arrays and affect the performance of the device.
Using heterogeneous integrated ASIC logic circuit, memory array logic circuit and cache structure, it is stacked and electrically coupled through silicon vias (TSVs) to form an upward or downward TSV configuration, simplifying process steps and reducing logic Adverse effects of circuits.
Improves memory access speed, reduces latency and input/output latency, improves device performance, and reduces manufacturing complexity.
Smart Images

Figure CN112151526B_ABST
Abstract
Description
Background Art
[0001] Strong growth in demand for portable consumer electronic devices has driven the need for high-capacity storage devices. Non-volatile semiconductor memory devices, such as flash memory cards, have been widely used to meet the growing demand for digital information storage and exchange. Their portability, versatility, and rugged design, as well as their high reliability and large capacity, make such memory devices ideal for use in a variety of electronic devices, including, for example, digital cameras, digital music players, video game controllers, PDAs, cellular phones, and solid-state drives.
[0002] Recently, ultra-high density memory devices have been proposed using 3D stacked memory structures with strings of memory cells formed as layers. One such memory device is sometimes referred to as a bit cost scalable (BiCS) architecture. In addition to the layered memory cells, the 3D memory device also includes logic circuits for controlling the read / write of the memory cells. The logic circuits, which are often manufactured using complementary metal oxide semiconductor (CMOS) technology, can typically be formed below the stacked memory layers within a semiconductor wafer.
[0003] Currently, there is a significant push to use flash memory devices as solid-state drives (SSDs) in data centers. As the number of memory layers in 3D memory structures increases to meet the growing data center memory needs, it becomes more difficult to locate logic circuits below the 3D memory cell structures. Additionally, process parameters optimized for memory array formation may not be optimized for logic circuit formation. For example, it is known to heat the 3D memory cell structure in various manufacturing steps. While beneficial to the memory cell structure, heating may adversely affect the operation of the logic circuits. BRIEF DESCRIPTION OF THE DRAWINGS
[0004] Figure 1 is a flow chart for forming a first wafer including a controller die according to an embodiment of the present technology.
[0005] Figure 2 is a top view of a first major surface of a first semiconductor wafer including a controller die according to an embodiment of the present technology.
[0006] Figure 3 is a top view of a controller die of a first semiconductor wafer according to an embodiment of the present technology.
[0007] Figure 4 is a cross-sectional edge view of a controller die of a first semiconductor wafer in accordance with an embodiment of the present technology.
[0008] Figure 5 is a schematic block diagram of a controller die according to an embodiment of the present technology.
[0009] Figure 6 is a flow chart for forming a second wafer including memory array dies according to an embodiment of the present technology.
[0010] Figure 7 is a top view of a first major surface of a second semiconductor wafer including a memory array die according to an embodiment of the present technology.
[0011] Figure 8 is a top view of a memory array die of a second semiconductor wafer according to an embodiment of the present technology.
[0012] Fig. 9 is a cross-sectional edge view of a memory array die of a second semiconductor wafer in accordance with an embodiment of the present technology.
[0013] Fig.10 is a flow chart for forming a "face-up" semiconductor device in accordance with an embodiment of the present technology.
[0014] Figures 11 to 13 is an edge view illustrating the preparation of memory array semiconductor dies for stacking together in accordance with an embodiment of the present technology.
[0015] Figure 14 to Figure 15 is an edge view illustrating the preparation of a temporary carrier upon which memory array dies may be stacked in accordance with an embodiment of the present technology.
[0016] Figures 16 to 19 is an edge view showing a face-up memory array wafer stack at various stages of fabrication in accordance with an embodiment of the present technology.
[0017] Fig. 20 is a perspective view of a complete memory array wafer stack according to an embodiment of the present technology.
[0018] Fig.21 is a flow chart for forming a "face-down" semiconductor device in accordance with an embodiment of the present technology.
[0019] Figure 22 to Figure 23 is an edge view showing a face-down memory array wafer stack at various stages of fabrication in accordance with an embodiment of the present technology.
[0020] Fig.24 is an exploded edge view of a semiconductor device including a memory array semiconductor die stack and a controller semiconductor die for mounting on a substrate.
[0021] Fig.25 is an edge view of a complete semiconductor device according to an embodiment of the present technology.
[0022] Fig.26 is a functional block diagram of a semiconductor device according to an embodiment of the present technology. DETAILED DESCRIPTION
[0023] The present technology will now be described with reference to the accompanying drawings, which, in an embodiment, relate to a semiconductor device including one or more memory array semiconductor dies and a controller die, the controller die including a heterogeneously integrated ASIC logic circuit, a memory array logic circuit, and a cache structure. The ASIC logic circuit of the controller die performs memory management and acts as an interface with a host device. The memory array logic circuit of the controller die includes sensor amplifiers and peripheral functions, and acts as an interface with the memory array of the memory array wafer die. The cache structure can be positioned between the ASIC logic circuit and the memory array logic circuit to provide temporary storage for active data. As opposed to accessing the memory array semiconductor die, the ASIC logic circuit can read active data from the cache structure and / or write active data to the cache structure to shorten access time, reduce latency, and improve input / output (I / O).
[0024] In an embodiment, the memory array semiconductor dies may be stacked and electrically coupled to each other using "continuously formed" through silicon vias (TSVs), meaning that the holes for the TSVs are formed in a single process through each of the memory array semiconductor dies in the die stack. The memory array stack may have a "face-up" TSV configuration or a "face-down" TSV configuration. In the face-up TSV configuration, an electrical connector may be formed at the bottom of the die stack, and in the face-up TSV configuration, an electrical connector may be formed at the top of the die stack. Compared to conventional TSV technology, both the face-up TSV configuration and the face-down TSV configuration omit several processes, such as through-hole etching for each wafer, seed layer deposition, isolation layer manufacturing on each back side of the ground wafer, two-side copper bump manufacturing, die-level bump bonding, and bottom filler filling between dies.
[0025] It should be understood that the present invention can be embodied in many different forms and should not be construed as being limited to the embodiments set forth herein. On the contrary, these embodiments are provided so that the disclosure will be thorough and complete, and the present invention will be fully communicated to those skilled in the art. In fact, the present invention is intended to cover alternatives, modifications and equivalents of these embodiments, which are all included in the scope and essence of the present invention as defined by the appended claims. In addition, in the following specific embodiments of the present invention, many specific details are given in order to provide a thorough understanding of the present invention. However, it will be apparent to those of ordinary skill in the art that the present invention can be implemented without such specific details.
[0026] The terms "top" and "bottom", "upper" and "lower", and "vertical" and "horizontal", and forms thereof, as used herein, are used herein by way of example only and for exemplary purposes, and are not intended to limit the description of the technology, as the referenced items may be interchanged in position and orientation. Additionally, as used herein, the terms "substantially" and / or "approximately" mean that a specified dimension or parameter may vary within acceptable manufacturing tolerances for a given application. In one embodiment, the acceptable manufacturing tolerance is ±2.5% of a given dimension.
[0027] Now refer to Figure 1 Flowchart and Figures 2 to 5 FIG. 1 is a diagram illustrating an embodiment of a first wafer including a controller semiconductor die. In step 50, the first semiconductor wafer 100 may be processed into Figures 2 to 4 A plurality of first semiconductor dies 102 are shown. These first semiconductor dies 102 may also be referred to herein as heterogeneous integrated controller dies 102, or simply controller dies 102. The first wafer 100 may also be referred to herein as a heterogeneous integrated controller wafer 100, or simply controller wafer 100. The first semiconductor wafer 100 may start with an ingot of wafer material, which may be single crystal silicon grown according to a Czochralski (CZ) or a floating zone (FZ) process. However, in other embodiments, the first wafer 100 may be formed of other materials and by other processes.
[0028] The controller wafer 100 may be cut from an ingot and have a first major planar surface 104 and a second major planar surface 107 ( Figure 4 ) are polished on both to provide a smooth surface. The first main surface 104 may undergo various processing steps to separate the wafer 100 into respective first semiconductor dies 102 and to form integrated circuits of the respective first semiconductor dies 102 on and / or in the first main surface 104.
[0029] Specifically, in step 50, the first semiconductor die 102 may be processed in embodiments to include Figure 4 ASIC logic 112 , memory array logic 114 , and cache structure 115 are shown formed in a dielectric substrate including layer 116 . Figure 4 A cross section of a single die 102 is shown that is representative of all the dies 102 on the first wafer 100. Figure 5ASIC logic 112, memory array logic 114, and cache structure 115 are described in more detail, but in general, ASIC logic 112 performs memory management and acts as an interface to control data exchange with a host device. Memory array logic 114 includes sensor amplifiers (SA) and peripheral (peripheral) circuits that act as switching mode interface circuits to decode bit lines / word lines to a buffer and then transfer data between a memory array (described below) and ASIC logic 112. Cache structure 115 can provide temporary storage for ASIC logic to shorten access time, reduce latency, and improve input / output (I / O).
[0030] In an embodiment, the ASIC logic circuit 112, the memory array logic circuit 114, and the cache structure 115 can each be fabricated in a respective dielectric layer 116 of the wafer 100, wherein the cache structure 115 is sandwiched below the ASIC logic circuit 112 and above the memory array logic circuit 114 (from Figure 4 1 . The logic circuits 112, 114 may be fabricated using CMOS technology, but in other embodiments, the logic circuits may be fabricated using other technologies. In other embodiments, as described below, the controller semiconductor die 102 may include other and / or additional integrated circuits.
[0031] In step 52, multiple layers of metal interconnects and vias 120 may be formed in and through the layers of semiconductor die 102. These metal interconnects and vias 120 may electrically couple ASIC logic 112 and memory array logic 114, as well as ASIC logic 112 and cache structures 115.
[0032] A passivation layer 128 may be formed on top of the dielectric film layer 116. In step 54, the bonding pad 108 ( Figure 3 and Figure 4 ) may be formed on the passivation layer 128, coupled to the TSV 120a. The bonding pad 108 may be formed on the liner 106. As is known in the art, the bonding pad 108 may be formed of, for example, copper, aluminum, and alloys thereof, and the liner 106 may be formed of, for example, a titanium / titanium nitride stack (such as, for example, Ti / TiN / Ti), but these materials may vary in other embodiments. The bonding pad 108 and the liner 106 together may have a thickness of 720nm, but in other embodiments the thickness may be greater or less.
[0033] In step 56, the controller wafer 100 may be thinned in a back grinding process to define a second major planar (inactive) surface 107 of the wafer 100. In step 58, a redistribution layer (RDL) 130 may be formed on the second major planar surface 107, such as Figure 4 As shown. RDL 130 can electrically connect a selected one of bonding pads 108 and TSV 120a to a pattern of microbumps 132 distributed on RDL 130. Microbumps 132 can be Cu, AgSn, or other metals that can be bonded between two wafers. The height of microbumps 132 can be about 1 μm to 50 μm, but these microbumps can be thinner or thicker than those in other embodiments. Metal interconnects and through-holes 120 can be used to transmit signals and currents between die bonding pads 108 and ASIC logic circuits 112, between ASIC logic circuits 112 and cache structures 115, between ASIC logic circuits 112 and memory array logic circuits 114, and between memory logic circuits 114 and microbumps 132.
[0034] Figure 2 The number and pattern of semiconductor dies 102 shown on wafer 100 in FIG. 1 are shown by way of example only. In other embodiments, wafer 100 may include more first semiconductor dies 102 and in different patterns. Similarly, Figure 3 and Figure 4 The embodiment includes a pattern of bonding pads 108 on one of the semiconductor dies 102 around the perimeter of the semiconductor die 102. However, Figure 3 and Figure 4 The pattern of the bonding pads 108 on the first semiconductor die 102 in FIG. 1 and the number of bonding pads 108 may vary in further embodiments.
[0035] Figure 5 1 is a schematic block diagram showing more details of the ASIC logic circuit 112, the memory array logic circuit 114, and the cache structure on the controller die 102. The ASIC logic circuit 112 includes a host interface 142 that is connected to and communicates with the host device 300 ( Fig.26). In one embodiment, the host interface 142 provides a PCIe interface. Other interfaces such as SCSI, SATA, etc. may also be used. The host interface 142 is also connected to a network on chip (NOC) 134. NOC is a communication subsystem on an integrated circuit. NOC can span synchronous and asynchronous clock domains, or use non-clocked asynchronous logic. NOC technology applies network theory and methods to on-chip communication, and brings significant improvements compared to conventional bus and cross switch interconnections. Compared with other designs, NOC improves the scalability of system on chip (SoC) and the power efficiency of complex SoC. The wires and links of NOC are shared by many signals. Since all links in NOC can run simultaneously on different data packets, a high level of parallelism is achieved. Therefore, as the complexity of the integrated subsystem continues to grow, NOC provides enhanced performance (such as throughput) and scalability compared to previous communication architectures (e.g., dedicated point-to-point signal lines, shared buses, or segmented buses with bridges). Connected to NOC 134 and communicating with the NOC are processors 136, ECC engines 138, memory interfaces 140, and DRAM controllers 144. DRAM controller 144 is used to operate and communicate with local high-speed volatile memory 146 (eg, DRAM). In other embodiments, local high-speed volatile memory 146 may be SRAM or another type of volatile memory.
[0036] The ECC engine 138 performs error correction services. For example, the ECC engine 138 performs data encoding and decoding according to the implemented ECC technology. In one embodiment, the ECC engine 138 is a circuit programmed by software. For example, the ECC engine 138 can be a programmable processor. In other embodiments, the ECC engine 138 is a customized dedicated hardware circuit without any software. In another embodiment, the functions of the ECC engine 138 are implemented by the processor 136.
[0037] The processor 136 performs various memory operations. For example, the memory array module (MM) is a module / process that runs on the processor 136 for performing programming, erasing, reading, and memory management processes. The read calibration manager (RC) is a module / process that runs on the processor 136 for performing read calibration processes. In one embodiment, the read calibration manager (RC) is implemented as part of the ECC engine 138. The error indication manager (IOE) is a module / process that runs on the processor 136 for calculating error indications, which is used to calibrate the read reference level. In one embodiment, the error indication manager (IOE) is implemented as part of the ECC engine 138 or as part of the read calibration manager (RC).
[0038] In one embodiment, the processor 136 is programmed by firmware, so that the read calibration manager (RC) and the error indication manager (IOE) are firmware / software. In other embodiments, the processor 136 is a custom dedicated hardware circuit without any software. The processor 136 (alone or together with the ECC engine 138) can perform recalibration of the read reference level to determine an updated read reference level, including dynamically and adaptively selecting a test read reference level for sensing sample data.
[0039] The processor 136 also implements a translation module (TM), either as a software / firmware process or as a dedicated hardware circuit. In many systems, non-volatile memory is addressed internally to the storage system using physical addresses associated with one or more memory dies, discussed below. However, the host system will use logical addresses to address the various memory locations. This enables the host to assign data to consecutive logical addresses while the storage system is free to store data between locations on one or more memory dies as desired.
[0040] The memory interface 140 communicates with the memory array logic 114. The memory array 212 (discussed below) on the second die 202 can be controlled by the memory array logic 114 on the first die 102. The memory array logic 114 can have circuits for accessing, controlling, and driving the memory elements of the memory array to perform functions such as programming and reading. The memory array logic 114 cooperates with the read / write circuits 168 to perform memory operations on the memory array. The read / write circuits 168 may include multiple sense blocks (sense circuits) that allow a page of memory cells from the memory array die (described below) on the memory wafer to be read or programmed in parallel.
[0041] In an embodiment, logic circuit 114 may include state machine 152, address decoder 154, and power control module 156. State machine 152 provides chip-level control of memory operations. Storage area 153 may be provided for operating memory array 212, such as programming parameters for different rows or other groups of memory cells. These programming parameters may include bit line voltages and verification voltages.
[0042] The address decoder 154 provides an address interface between the address interface used by the host device or ASIC logic circuit 112 and the hardware address used by decoders 163 and 166. The power control module 156 controls the power and voltage supplied to the word lines and bit lines during memory operations. The power control circuit may include drivers for the word line layer in a 3D configuration, source side select gates, drain side select gates, and source lines. The source side select gate is a gate transistor at the source end of the NAND string, and the drain side select gate is a transistor at the drain end of the NAND string.
[0043] The cache structure 115 may be located between the ASIC logic circuits 112 in the memory array logic circuit 114. The cache structure 115 may be organized as SRAM, but in other embodiments the cache structure may have other configurations. The cache structure 115 may act as a disk cache for the memory array 212 (described below) and / or a processor cache for the processor 136. Data may be stored on the cache structure 115 and / or may be read to the cache structure instead of interfacing with the memory array. This provides shorter access times, reduced latency, and improved input / output (I / O). The ASIC logic circuit 112 may read data from the cache structure 115 and write the data to the cache structure. The memory array logic circuit 114 may also read data from the cache structure 115 and write the data to the cache structure.
[0044] According to aspects of the present technique, the above-described ASIC logic 112, memory array logic 114, and cache structure 115 are integrated into a single semiconductor die 102, and the memory array 212 (described below) is separated into its own chip. However, it should be understood that in other embodiments, some of the above-described functions of the ASIC logic 112, memory array logic 114, and cache 115 may be offloaded to the memory array semiconductor die. In other embodiments, additional components and functions may be added to the controller die 102.
[0045] Before, after, or simultaneously with forming the controller semiconductor die on the first wafer 100, the second semiconductor wafer 200 may be processed into a plurality of second semiconductor dies 202. Reference will now be made to Figure 6 Flowchart and Figures 7 to 9 The diagram of FIG. 2 explains the fabrication and operation of semiconductor die 202 on wafer 200 .
[0046] Figure 7 2 shows a top view of a second semiconductor wafer 200. The second semiconductor wafer 200 may start from an ingot of single crystal silicon grown according to CZ, FZ or other processes. The second semiconductor wafer 200 may be cut and formed on a first main surface 204 and a second main surface 207 ( Fig. 9 ) to provide a smooth surface. The first major surface 204 may undergo various processing steps to separate the second wafer 200 into respective second semiconductor dies 202 and form integrated circuits of respective second semiconductor dies 202 on and / or in the first major surface 204. These second semiconductor dies 202 may also be referred to herein as memory array dies 202, and the second wafer 200 may also be referred to herein as memory array wafer 200.
[0047] In one embodiment, memory array die 202 may be processed at step 60 to include forming a Fig. 9 Memory array 212 is shown in a dielectric substrate of layers 224 and 226 . Fig. 9 A cross section of a single die 202 is shown that is representative of all the die 202 on the wafer 200. More details of the memory array 212 are provided below, but in general, the memory array 212 can be formed as a 3D stacked memory structure having strings of memory cells formed as layers. However, it should be understood that the second semiconductor die 202 can be processed into an integrated circuit that includes something other than a 3D stacked memory structure.
[0048] In step 62, multiple layers of metal interconnects and vias 220 may be formed in and through dielectric layer 226 of semiconductor die 202. A passivation layer 228 may be formed on top of dielectric film layer 226. In step 64, bonding pads 208 ( Figure 8 and Fig. 9 ) may be formed through the passivation layer 228, coupled to the metal interconnect and the via 220. The bonding pad 208 may be formed over the liner 206. As described above, for the pad 108, the bonding pad 208 may be formed of, for example, copper, aluminum, and alloys thereof, and the liner 206 may be formed of, for example, a titanium / titanium nitride stack (such as, for example, Ti / TiN / Ti), but these materials may vary in other embodiments. The bonding pad 208 and the liner 206 together may have a thickness of 720 nm, but in other embodiments the thickness may be greater or less.
[0049] Figure 7 The number and pattern of semiconductor dies 202 shown on wafer 200 in FIG. 1 are shown by way of example only. In other embodiments, wafer 200 may include more memory array semiconductor dies 202 and in different patterns. Similarly, Figure 8 and Fig. 9 The pattern of bonding pads 208 on one of the semiconductor dies 202 includes bonding pads 208 on one of the semiconductor dies 202 around the perimeter of the semiconductor die 202. However, Figure 8 and Fig. 9 The pattern of the bonding pads 208 on the second semiconductor die 202 in FIG. 2 and the number of bonding pads 208 may vary in further embodiments.
[0050] The die 202 of the memory array wafer 200 may each include a memory structure of memory cells, such as a memory cell array. The memory cell array 212 ( Fig. 9 ) can be passed through the word line via the row decoder 163 ( Figure 5) and addressed from the controller die 102 on the controller wafer 100 via the column decoder 166 via the bit lines. The multiple memory elements in the memory array can be configured so that the multiple memory elements are connected in series or so that each element is individually accessible. By way of non-limiting example, a flash memory system (NAND memory) in a NAND configuration typically includes memory elements connected in series. A NAND string is an example of a group of transistors connected in series that includes a memory cell and a select gate transistor.
[0051] A NAND memory array can be configured such that the array is composed of multiple strings of memory, where a string is composed of multiple memory elements that share a single bit line and are accessed as a group. Alternatively, the memory elements of the memory array can be configured such that each element is individually accessible, for example, a NOR memory array. NAND and NOR memory configurations are exemplary, and memory elements can be configured in other ways.
[0052] The memory array 212 may be two-dimensional (2D) or three-dimensional (3D). The memory array 212 may include one or more arrays of memory elements (also referred to as memory cells). A 3D memory array is arranged such that the memory elements occupy multiple planes or multiple memory device levels, thereby forming a three-dimensional structure (i.e., in the x-direction, y-direction, and z-direction, wherein the z-direction is substantially perpendicular to the major planar surfaces 204, 207 of the memory array die 202, and the x-direction and y-direction are substantially parallel to these major planar surfaces of the memory array die).
[0053] According to aspects of the present technique, one or more memory array wafers 200 may be stacked together to form a memory array die stack. Fig.10 Flowchart and Figures 11 to 20 A first embodiment including a face-up TSV memory array die stack will be described with reference to FIG. Fig.21 Flowchart and Figure 22 to Figure 23 A second embodiment including a face-down TSV memory array die stack is described with reference to FIG.
[0054] In step 70, multiple memory array wafers 200 may be processed for stacking by depositing a dielectric film over active surface 204 to embed contact pads 108. The dielectric film may be silicon nitride, but in other embodiments, the dielectric film may be other materials. Step 70 may also include performing a CMP polishing process for surface planarization. Fig.11 A semiconductor die 202 of a memory array wafer 200 is shown including bonding pads 208 on active substrate 204, and Fig.12The active surface 204 is shown after application and polishing of the dielectric film layer 240. In step 72, the memory array wafer 200 may be thinned in a back grinding process to define a second major planar (inactive) surface 207 of the wafer 200, such as Fig.13 shown.
[0055] Before, during, or after preparing the second wafer 200 in steps 70 and 72 , a temporary carrier may also be prepared in steps 74 and 76 . Fig.14 An example of a temporary carrier 250 is shown in FIG. Temporary carrier 250 may be, for example, metal, glass, or silicon. In step 74, metal contact pads 254 may be formed on a major surface 255 of temporary carrier 250. Contact pads 254 may be formed by a photolithography process, an electroplating process, or other known processes. In step 76, a release film 256 may be applied to major surface 255 of carrier 250, such as Fig.15 As shown. Release film 256 may be generally planar and may be to the height of contact pads 254. In step 76, release film 256 may be cured to a B-stage. Film 256 may be a temporary adhesive that may be dissolved or removed, for example, using laser / UV light, heat, chemicals, and / or force, as described below.
[0056] In step 78, a plurality (n) of memory array wafers 200 (including films 240) may be stacked on top of major surface 255 of carrier 250, such as Fig.16 As shown. In an embodiment, there may be n total memory array wafers, where n=1, 2, 4, 8, 16, 32, or other number of wafers. The memory array wafers 200 may be aligned with each other and stacked such that the memory array semiconductor dies 202 of the respective wafers 200 are aligned with each other. As described above, Fig.16 The number of memory wafers 200 in FIG. 1 is shown by way of example only and may vary.
[0057] Once all wafers 200 have been mounted on carrier 250, wafers 200 may be permanently bonded to each other in step 82. This permanent bonding physically attaches the wafers to each other. Various known bonding processes may be used to permanently bond wafers to each other, such as fusion bonding and hybrid bonding.
[0058] Once bonded, TSVs 264 may be formed through each of the wafers in steps 84-88 to electrically couple to the Fig.17208 of the corresponding wafer shown. According to aspects of the present technology, TSV 264 can be "continuously formed" through the tube core 202 of the wafer stack 250, which means that the hole for TSV 264 is formed in a single process through each tube core of the tube core in the stack 250 after the wafers are stacked together. Specifically, TSV 264 can be formed by etching a hole through the memory array semiconductor tube core 202 in the pattern of the finished TSV 264 in step 84. The silicon sidewall of the hole can also be etched in step 84. The silicon sidewall etch structure can provide a strong pinning effect for the copper fill (described below) between two adjacent tube cores to improve the reliability of signal transmission. Then in step 86, the etched hole can be lined with an isolation layer to prevent diffusion.
[0059] In step 88, the etched holes may then be plated and filled with a conductive material to form TSVs 264. The conductive material may include copper, but other suitable materials may also be used, such as aluminum, tin, nickel, gold, doped polysilicon, and alloys thereof or combinations thereof. One feature of the present technology is that a seed layer conventionally formed above the isolation layer may be omitted. Specifically, as described below, the etched holes may be plated and / or filled with a conductive material, such as copper, from the bottom up. Filling from the bottom up under gravity allows the seed layer to be omitted.
[0060] In this face-up configuration, a polyimide film 268 may then be formed on the upper surface of the wafer stack 260 in step 90, such as Fig.18 A polyimide film 268 is provided to seal and protect the upper surface of the wafer stack 260. In step 91, the Fig.19 The carrier 250 is removed as shown. In step 92, the release film 256 can be dissolved or otherwise removed, such as Fig.19 As shown, the contact pads 254 initially formed on the carrier 250 are transferred to the lower surface 262 of the die stack 260. With the release film 256 removed, the contact pads 254 are exposed.
[0061] As described above, at this point in fabrication, the memory array die 202 are still part of the corresponding wafer 200 of these memory array dies in the stack 260, e.g. Fig. 20 In step 93, the stack 260 may be segmented to provide individual memory modules 266. Each memory module 266 may be a stacked and aligned column of memory array die 202, such as Fig. 20 The wafer stack 260 can be separated by cutting along the kerf lines 269, two of which are formed by Fig. 20 The wafer stack may be cut, for example, by laser, saw, water jet or other methods.
[0062] Figures 11 to 20 Fabrication of a face-up memory module 266 is shown, with contact pads 254 used for external electrical connections of the memory module 266 on the bottom surface 262 of the module. Figure 21 to Figure 23 Another embodiment of fabrication of a face-down memory module 266 is shown, wherein electrical contacts for external electrical connections of the memory module 266 are located on an upper surface of the module.
[0063] See also Fig.21 , the face-down memory module 266 may be formed in the same manner as the face-up memory module in steps 70 to 88. Specifically, a plurality of memory array wafers 200 may be stacked on a carrier 250, and TSVs 264 may be formed as described above. Fig.17 Then, in step 94, a conductive layer may be applied to the upper surface 272 of the wafer stack and the conductive layer may be developed into a pattern. Next, in step 95, contact pads 270 may be electroplated onto the developed pattern, such as Fig. 22 A contact pad 270 may be formed over the bonding pad 208 and electrically coupled to the bonding pad.
[0064] In step 96, the carrier 250 may be removed, such as Fig.23 As shown. A polyimide film 276 may then be applied to the bottom surface of the memory array wafer stack 260. The polyimide film 276 is provided to seal and protect the lower surface of the wafer stack 260. The memory array wafer stack 260 may then be singulated as described above to provide individual complete memory modules 266.
[0065] When the memory module 266 is completed according to any of the above embodiments, the memory module 266 can be mounted on the substrate 278 together with the controller die 102, such as Fig.24 and Fig.25 The face-up TSV configuration may be lowered onto the controller die 102 (the face-down TSV configuration may be reversed and then lowered onto the controller die 102 ). The contact pads 254 of the memory module 266 may be physically and electrically bonded to the contact pads 108 .
[0066] The memory module 266 and the controller die 102 may then be mounted on a substrate 278, with the microbumps 132 of the controller die 102 physically and electrically coupled to the contact pads 282 of the substrate to provide a Fig.25 The complete semiconductor device 284 is shown. The device 284 can then be mounted to a host device 300 ( Fig.26), such as, for example, a printed circuit board. Semiconductor device 284 may optionally be encapsulated in a protective mold prior to attachment to host device 300.
[0067] Fig.26 2 is a functional block diagram of a semiconductor device 284 according to an embodiment of the present technology. As shown, the controller die 102 controls a column of n memory array dies 202 in the memory module 266. Each of the ASIC logic circuit 112, the memory array logic circuit 114, and the cache structure 115 is fabricated into a single heterogeneous integrated semiconductor die 102. Fig.26 , the memory logic array 114 is separated into a memory array logic circuit and a memory array analog circuit, wherein the cache structure 115 interfaces between the ASIC logic circuit and both the memory array logic circuit and the analog circuit. The ASIC logic circuit 112 communicates with the host device 300 via the host interface. The ASIC logic circuit 112 and the memory array logic circuit 114 communicate with the memory module 266 via the memory array interface.
[0068] In summary, an example of the present technology relates to a semiconductor device configured to operate with a host device, including: a first semiconductor die, the first semiconductor die including: an ASIC logic circuit, the ASIC logic circuit configured to interface with the host device; a memory array logic circuit, the memory array logic circuit configured to interface with a memory array; and a cache structure, the cache structure configured to provide storage within the first semiconductor die; and a group of one or more second semiconductor dies, the group of one or more second semiconductor dies coupled to the first semiconductor die and including a memory array configured to interface with the memory array logic circuit of the first semiconductor die.
[0069] In another example, the present technology relates to a semiconductor device configured to operate with a host device, including: a first semiconductor die, the first semiconductor die including: an ASIC logic circuit, the ASIC logic circuit configured to interface with the host device; a memory array logic circuit, the memory array logic circuit configured to interface with a memory array; and a cache structure, the cache structure configured to provide storage within the first semiconductor die; a memory module, the memory module coupled to the first semiconductor die and including a plurality of second semiconductor dies having a memory array; and a plurality of through silicon vias, the plurality of through silicon vias being continuously formed through the memory module and including an electrical conductor, which electrically couples the plurality of second semiconductor dies to each other.
[0070] In another example, the present technology relates to a semiconductor device configured to operate with a host device, including: a first semiconductor die, the first semiconductor die including: an ASIC logic circuit device for interfacing with the host device, a memory array logic circuit device for interfacing with a memory array, and a cache device for providing storage within the first semiconductor die; a memory module coupled to the first semiconductor die and including a plurality of second semiconductor dies having a memory array; and an electrical connector device, the electrical connector device being continuously formed through the memory module and including an electrical conductor for electrically coupling the plurality of second semiconductor dies to each other.
[0071] The above specific embodiments of the present invention have been proposed for the purpose of illustration and description. It is not intended to be exhaustive or to limit the present invention to the precise form disclosed. Based on the above teachings, many modifications and variations are possible. The described embodiments are selected to best illustrate the principles of the present invention and its practical application, so that other technical personnel in the field can best use the present invention with various modifications suitable for the specific purpose contemplated in various embodiments. The scope of the present invention is intended to be limited by the appended claims.
Claims
1. A semiconductor device configured to operate with a host device, comprising: A first semiconductor die, the first semiconductor die comprising: an ASIC logic circuit, the ASIC logic circuit being configured to interface with the host device, a memory array logic circuit configured to interface with the memory array, and a cache structure configured to provide storage within the first semiconductor die; and A set of one or more second semiconductor dies coupled to the first semiconductor die, the set of one or more second semiconductor dies comprising the memory array configured to interface with the memory array logic circuitry of the first semiconductor die.
2. The semiconductor device of claim 1, wherein the ASIC logic circuit, the cache structure, and the memory array logic circuit are disposed in a continuous dielectric layer of the first semiconductor die.
3. The semiconductor device of claim 1, wherein the cache structure provides temporary input / output storage for the ASIC logic circuit.
4. The semiconductor device of claim 1, wherein the cache structure provides temporary input / output storage for the memory array logic circuit. 5 . The semiconductor device of claim 1 , wherein the set of one or more second dies comprises a memory module having a plurality of semiconductor dies electrically coupled to each other through through-silicon vias. The semiconductor device according to claim 5 , wherein the through silicon vias are formed continuously.
7. The semiconductor device of claim 6 , wherein each of the plurality of semiconductor dies of the memory module comprises a bonding pad facing a first surface of the memory module, the memory module comprising an upward-facing memory module having a contact pad on a second surface of the memory module opposite to the first surface.
8. The semiconductor device of claim 6 , wherein each of the plurality of semiconductor dies of the memory module comprises a bonding pad facing a first surface of the memory module, the memory module comprising a face-down memory module having contact pads on the first surface of the memory module.
9. The semiconductor device of claim 1 , wherein the first semiconductor die comprises a set of bonding pads on a first surface and a set of conductive bumps on a second surface, the set of bonding pads being configured to mate with contacts of the host device, and the set of conductive bumps being configured to mate with contacts of a memory array die in the set of one or more memory array dies. 10 . The semiconductor device of claim 1 , wherein the set of one or more second semiconductor dies comprises a three-dimensional stacked memory structure having memory cell strings formed as layers.
11. A semiconductor device configured to operate with a host device, comprising: A first semiconductor die, the first semiconductor die comprising: an ASIC logic circuit, the ASIC logic circuit being configured to interface with the host device, a memory array logic circuit configured to interface with the memory array, and a cache structure configured to provide storage within the first semiconductor die; a memory module coupled to the first semiconductor die, the memory module comprising a plurality of second semiconductor dies having the memory array; and A plurality of through silicon vias are continuously formed through the memory module and include electrical conductors that electrically couple the plurality of second semiconductor dies to each other. 12 . The semiconductor device of claim 11 , wherein the plurality of second semiconductor dies in the memory module are stacked on top of each other in columns.
13. The semiconductor device of claim 11 , wherein each of the plurality of semiconductor dies of the memory module comprises a bonding pad facing a first surface of the memory module, the memory module comprising an upward-facing memory module having a contact pad on a second surface of the memory module opposite to the first surface.
14. The semiconductor device of claim 11, wherein each of the plurality of semiconductor dies of the memory module comprises a bonding pad facing a first surface of the memory module, the memory module comprising a face-down memory module having contact pads on the first surface of the memory module.
15. The semiconductor device of claim 11, wherein the ASIC logic circuitry, the cache structure, and the memory array logic circuitry are disposed in consecutive layers of the first semiconductor die.
16. The semiconductor device of claim 11, wherein the cache structure provides temporary input / output storage for the ASIC logic circuit.
17. The semiconductor device of claim 11, wherein the cache structure provides temporary input / output storage for the memory array logic circuit.
18. A semiconductor device configured to operate with a host device, comprising: A first semiconductor die, the first semiconductor die comprising: An ASIC logic circuit device for interfacing with the host device, memory array logic circuitry for interfacing with the memory array, and a cache device for providing storage within said first semiconductor die; a memory module coupled to the first semiconductor die, the memory module comprising a plurality of second semiconductor dies having the memory array; and An electrical connector device is formed continuously through the memory module and includes electrical conductors for electrically coupling the plurality of second semiconductor dies to each other.
19. The semiconductor device of claim 18, wherein each of the plurality of semiconductor dies of the memory module comprises a bonding pad facing a first surface of the memory module, the memory module comprising an upward-facing memory module having a contact pad on a second surface of the memory module opposite to the first surface.
20. The semiconductor device of claim 18, wherein each of the plurality of semiconductor dies of the memory module comprises a bonding pad facing a first surface of the memory module, the memory module comprising a face-down memory module having contact pads on the first surface of the memory module.
