Integrated circuit

By forming a multi-layer interconnect structure under the surface of the silicon substrate, the problems of increased area and increased noise during transistor connection in the prior art are solved, and a more compact and efficient integrated circuit design is achieved, which improves performance and shrinkability .

CN119031707APending Publication Date: 2024-11-26INVENTION & COLLABORATION LABORATORY INC
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410656163.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Priority Date
2024-05-23
Filing Date
2024-05-24
Publication Date
2024-11-26

AI Technical Summary

Technical Problem

Existing integrated circuits face problems such as increased area, increased power consumption and increased noise when reducing component sizes. Especially when connecting transistors, it is difficult to effectively reduce the underlying interconnection lines, which limits the shrinkability of dynamic random access memory. performance and performance improvement.

Method used

Using the sub-surface interconnection technology, the vertical connection of the transistor is achieved by forming a multi-layer interconnect structure under the original surface of the silicon substrate, including P-type and N-type metal oxide semiconductor transistors, protection rings and multiple interconnections. Signal transmission reduces the use and area occupancy of metal wires.

Benefits of technology

A more compact integrated circuit design is achieved, reducing power consumption and noise, improving performance, while reducing wafer complexity and area occupancy, and enhancing transistor and circuit performance.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119031707A_ABST
    Figure CN119031707A_ABST
Patent Text Reader

Abstract

The invention discloses an integrated circuit. The integrated circuit comprises a semiconductor substrate, a P-type metal oxide semiconductor transistor, an N-type guard ring, an N-type metal oxide semiconductor transistor, a P-type guard ring and a first interconnection layer. The semiconductor substrate has an original surface. The N-type guard ring surrounds the P-type metal oxide semiconductor transistor. The P-type guard ring surrounds the N-type metal oxide semiconductor transistor. The first interconnect layer is located below an original surface of the semiconductor substrate and is isolated from the semiconductor substrate. The first interconnection layer is electrically connected to the P-type metal oxide semiconductor transistor, the N-type guard ring, the N-type metal oxide semiconductor transistor or the P-type guard ring. Therefore, the invention has the advantages of compact structure, smaller area, higher performance, lower complexity and the like because the interconnection line embedded under the surface of the silicon substrate can be provided.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to an integrated circuit, and in particular to an integrated circuit which can provide subsurface interconnects or subsurface interconnects buried in a silicon substrate, thereby having the advantages of compact structure, smaller area, higher performance, lower complexity, etc. Background Art

[0002] In the most advanced integrated circuits in the prior art, there are many transistors connected by conductive interconnects (such as metal lines, polysilicon lines, etc.) to facilitate signal transmission between the gate region, source region and drain region of these transistors. All of these conductive interconnects rely on many contact holes and connecting plugs to connect to the gate region, the source region and the drain region respectively, which poses a significant challenge and difficulty to the chip design goals of reducing the area, power consumption and noise of these transistors and improving the performance of integrated circuits, especially when the size of components needs to be reduced to meet Moore's Law, the number of integrated circuits on the chip must be significantly reduced. To give an example of area loss: compared with the size of the contact hole used to connect the metal line to the source region or the drain region, a larger source region or drain region diffusion area must be designed so that the inevitable photolithographic misalignment due to the limitation of the lithographic tool should not cause the contact hole to be formed outside the edge of the source region or the drain region. In this way, the diffusion area of ​​the transistor will inevitably increase, thereby increasing the chip area, resulting in an increase in capacitance and a serious impact on the AC performance of the circuit, consuming more power and increasing greater noise.

[0003] In addition, in the structure of traditional dynamic random access memory, there are multiple sensor amplifiers connected to the dynamic random access memory cell array through metal lines. The design rules of these metal lines set above the silicon substrate and the spacing between the metal lines will limit the shrinkability of the dynamic random access memory size. Therefore, how to introduce better self-aligned contact structures and technologies to connect transistors to their first interconnect (metal) layer with the smallest surface area to transmit and receive signals is a key challenge to further effectively reduce the size of integrated circuits and improve performance. Summary of the invention

[0004] An embodiment of the present invention discloses an integrated circuit. The integrated circuit includes a semiconductor substrate, a P type metal-oxide-semiconductor (PMOS) transistor, an N type guard ring, an N type metal-oxide-semiconductor (NMOS) transistor, a P type guard ring and a first interconnection layer. The semiconductor substrate has an original surface. The P type metal-oxide-semiconductor (PMOS) transistor includes a gate node, a source node, and a drain node. The N type guard ring surrounds the P type metal-oxide semiconductor transistor. The N type metal-oxide-semiconductor (NMOS) transistor includes a gate node, a source node, and a drain node. The P type guard ring surrounds the N type metal-oxide semiconductor transistor. The first interconnection layer is located below the original surface of the semiconductor substrate and isolated from the semiconductor substrate. The first interconnect layer is electrically connected to the P-type metal oxide semiconductor transistor, the N-type guard ring, the N-type metal oxide semiconductor transistor, or the P-type guard ring.

[0005] In one embodiment of the present invention, the first interconnect layer is electrically connected to the N-type guard ring, and a voltage source is connected to the first interconnect layer from the back or top of the P-type metal oxide semiconductor transistor.

[0006] In one embodiment of the present invention, the first interconnection layer surrounds the P-type metal oxide semiconductor transistor and is electrically connected to a source node or a drain node of the P-type metal oxide semiconductor transistor.

[0007] In one embodiment of the present invention, the first interconnect layer is electrically connected to the P-type guard ring, and a voltage source is connected to the first interconnect layer from the back or top of the N-type metal oxide semiconductor transistor.

[0008] In one embodiment of the present invention, the first interconnection layer surrounds the N-type metal oxide semiconductor transistor and is electrically connected to a source node or a drain node of the N-type metal oxide semiconductor transistor.

[0009] In an embodiment of the present invention, the integrated circuit further comprises another transistor adjacent to the P-type metal oxide semiconductor transistor, wherein the first interconnect layer surrounds and is electrically connected to the P-type metal oxide semiconductor transistor and the another transistor.

[0010] In one embodiment of the present invention, the first interconnect layer is electrically connected to a source node or a drain node of the P-type metal oxide semiconductor transistor, and is electrically connected to a source node or a drain node of the another transistor.

[0011] In an embodiment of the present invention, the integrated circuit further comprises another transistor adjacent to the N-type metal oxide semiconductor transistor, wherein the first interconnect layer surrounds and is electrically connected to the N-type metal oxide semiconductor transistor and the another transistor.

[0012] In one embodiment of the present invention, the first interconnect layer is electrically connected to a source node or a drain node of the N-type metal oxide semiconductor transistor, and is electrically connected to a source node or a drain node of the another transistor.

[0013] In one embodiment of the present invention, there is no metal layer above the original surface of the semiconductor substrate to connect the N-type guard ring to the P-type metal oxide semiconductor transistor.

[0014] In one embodiment of the present invention, there is no metal layer above the original surface of the semiconductor substrate to connect the P-type guard ring to the N-type metal oxide semiconductor transistor.

[0015] Another embodiment of the present invention discloses an integrated circuit. The integrated circuit includes a semiconductor substrate, a plurality of transistors, a guard ring, and a subsurface interconnect layer. The semiconductor substrate has an original surface. Each transistor includes a gate node, a source node, and a drain node. The guard ring surrounds the plurality of transistors. The subsurface interconnect layer is located below the original surface of the semiconductor substrate and isolated from the semiconductor substrate. The subsurface interconnect layer is electrically connected to the plurality of transistors, and a group of metal layers located above the original surface of the semiconductor substrate are electrically connected to the guard ring and the plurality of transistors.

[0016] In one embodiment of the present invention, the integrated circuit further comprises a first metal layer above the set of metal layers, wherein the first metal layer is electrically connected to the set of metal layers.

[0017] In one embodiment of the present invention, the multiple transistors include a group of P-type metal oxide semiconductor transistors, the guard ring includes an N+ guard ring that receives a power supply, and the subsurface interconnect layer is electrically connected to the source nodes of the group of P-type metal oxide semiconductor transistors, wherein the power supply is electrically connected to the source nodes of the group of P-type metal oxide semiconductor transistors through the group of metal layers.

[0018] In one embodiment of the present invention, the plurality of transistors include a group of N-type metal oxide semiconductor transistors, the guard ring includes a P+ guard ring receiving a ground voltage, and the subsurface interconnect layer is electrically connected to source nodes of the group of N-type metal oxide semiconductor transistors, wherein the ground voltage is electrically connected to the source nodes of the group of N-type metal oxide semiconductor transistors through the group of metal layers.

[0019] In one embodiment of the present invention, the subsurface interconnect layer includes a first subsurface interconnect sublayer and a second subsurface interconnect sublayer stacked above the first subsurface interconnect sublayer; wherein the guard ring includes a first sub-guard ring receiving a first voltage and a second sub-guard ring receiving a second voltage, and the multiple transistors include a group of first-type metal oxide semiconductor transistors surrounded by the first sub-guard ring and a group of second-type metal oxide semiconductor transistors surrounded by the second sub-guard ring; wherein the first subsurface interconnect sublayer is electrically connected to the source node of the group of first-type metal oxide semiconductor transistors, and the second subsurface interconnect sublayer is electrically connected to the source node of the group of second-type metal oxide semiconductor transistors.

[0020] In one embodiment of the present invention, the group of metal layers includes a first group of metal sublayers and a second group of metal sublayers, the first voltage is electrically connected to the source node of the group of first-type metal oxide semiconductor transistors through the first group of metal sublayers, and the second voltage is electrically connected to the source node of the group of second-type metal oxide semiconductor transistors through the second group of metal sublayers.

[0021] Another embodiment of the present invention discloses an integrated circuit. The integrated circuit comprises a semiconductor substrate, a plurality of transistors, a guard ring, a first subsurface interconnection layer and a second subsurface interconnection layer. The semiconductor substrate has an original surface. The guard ring surrounds the plurality of transistors. The first subsurface interconnection layer is located below the original surface of the semiconductor substrate and isolated from the semiconductor substrate. The second subsurface interconnection layer is located below the first subsurface interconnection layer and isolated from the semiconductor substrate. A first edge of the first subsurface interconnection layer is vertically and horizontally offset from a second edge of the second subsurface interconnection layer.

[0022] In one embodiment of the present invention, the guard ring includes a first sub-guard ring and a second sub-guard ring, and the plurality of transistors include a group of P-type metal oxide semiconductor transistors surrounded by the first sub-guard ring and a group of N-type metal oxide semiconductor transistors surrounded by the second sub-guard ring, wherein the first subsurface interconnect layer electrically connects the group of P-type metal oxide semiconductor transistors, and the second subsurface interconnect layer electrically connects the group of N-type metal oxide semiconductor transistors. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1A The present invention is a flowchart of a method for manufacturing a memory array circuit according to an embodiment of the present invention.

[0024] Figure 1B , Figure 1C , Figure 1D , Figure 1E , Figure 1F , Figure 1G Yes Description Figure 1A Schematic diagram of .

[0025] Figure 2 It is a schematic diagram illustrating a cross-sectional view and a top view along the X direction after a liner oxide layer is grown, a liner nitride layer is deposited, and a trench is formed.

[0026] Figure 3 is a schematic diagram illustrating deposition and etching back of a first nitride layer to form a first nitride spacer layer, and deposition and polishing of a spin-on dielectric.

[0027] Figure 4 It is a schematic diagram for explaining the steps of etching away the upper edge of the first nitride spacer layer, spinning on a dielectric, generating a first oxide layer, and depositing a metal layer.

[0028] Figure 5 is a schematic diagram illustrating etching back a metal layer to form a subsurface bit line.

[0029] Figure 6 is a schematic diagram illustrating deposition of a second oxide layer in the trench and planarization by the chemical mechanical polishing technique.

[0030] Figure 7 is a schematic diagram illustrating depositing a third oxide layer, a second nitride layer, and a patterned photoresist layer, and then etching away unnecessary portions of the third oxide layer, the second nitride layer, and the patterned photoresist layer.

[0031] Figure 8 It is a schematic diagram illustrating forming a U-shaped recess, forming a high dielectric constant insulating layer in the U-shaped recess, depositing and etching back a gate material to form a word line and a gate structure of the access transistor, depositing and etching back a third nitride layer and a fourth oxide layer, and etching away the second nitride layer and the third oxide layer.

[0032] Fig. 9 It is a schematic diagram for explaining the removal of the pad nitride layer, etching back the second oxide layer, depositing and anisotropically etching the fourth nitride layer, the fifth oxide layer and the fifth nitride layer, and depositing the sixth oxide layer.

[0033] Fig.10 It is a schematic diagram illustrating the deposition and polishing of a spin-on dielectric, the deposition of a photoresist layer, the etching back of the spin-on dielectric to be removed, the etching away of the exposed sixth oxide layer, the pad oxide layer, and the silicon material corresponding to the horizontal silicon surface HSS-1 / 3 to produce a hole hole-1 / 3, the thermal generation of a seventh oxide layer, and the deposition and etching back of the spin-on dielectric.

[0034] Fig.11 It is a schematic diagram illustrating the removal of the exposed spin-on dielectric, the exposed sixth oxide layer and the exposed pad oxide layer, and then digging out the silicon material corresponding to the horizontal silicon surface HSS-1 / 2 to generate a hole hole-1 / 2, and thermally generating an eighth oxide layer.

[0035] Fig.12 It is a schematic diagram illustrating a cross-sectional view of the memory array circuit along the Y2 direction, wherein the Y2 direction extends along the center of the hole Hole-1 / 2 and is perpendicular to the X direction.

[0036] Fig.13 It is a schematic diagram illustrating removing the first nitride spacer at the lower edge to expose the sidewall of the subsurface bit line, and depositing an n+ heavily doped polysilicon film in the hole hole-1 / 2 to contact the sidewall of the subsurface bit line. Fig.14 is a schematic diagram illustrating etching back the upper portion of the eighth oxide layer to expose the silicon material corresponding to the hole hole-1 / 2.

[0037] Fig.15 It is a schematic diagram for explaining the lateral generation of an n+ polysilicon layer and the thermal generation of a ninth oxide layer based on the exposed silicon material.

[0038] Fig.16 It is a schematic diagram of realizing the vertical connection (bridge) of the access transistor and the drain region and how they are connected to the subsurface interconnection line disclosed in another embodiment of the present invention.

[0039] Fig.17 is a schematic diagram illustrating the formation of the drain region and the source region of the access transistor.

[0040] Fig.18 is a schematic diagram illustrating the formation of two different subsurface interconnections at two different vertical heights in a pipeline and the separation of the two from each other by an isolation material.

[0041] Fig.19is a schematic diagram illustrating waveforms of related signals of a conventional dynamic random access memory cell during a write operation.

[0042] Fig. 20 FIG. 5 is a schematic diagram illustrating a circuit for selectively coupling the sense amplifier to two separate voltages during a write operation of the DRAM cell.

[0043] Fig.21 is a schematic diagram illustrating the most commonly used design of the dynamic random access memory cell.

[0044] Fig. 22 is a schematic diagram illustrating a conventional sensor amplifier and an equalization circuit design circuit.

[0045] Fig.23 and Fig.24 is a schematic diagram illustrating a conventional bit line sense amplifier design for open-bit line's of a dynamic random access memory.

[0046] Fig.25 is a schematic diagram illustrating the coupling relationship between the active region, the connecting pin and the metal 0.

[0047] Fig.26 is a schematic diagram illustrating subsurface interconnects in different vertical layers.

[0048] Fig. 27 It is a schematic diagram illustrating that different subsurface interconnects include multiple sublayers and the multiple sublayers are located in different shallow trench isolations.

[0049] Fig.28 This means that different subsurface interconnects within the same shallow trench isolation can be Fig. 27 The diagrams shown are schematic diagrams with relative offset in direction.

[0050] Fig.29A , Fig.29B , Fig. 30A , Fig. 30B , Fig.31A , Fig.31B , Fig.31C , Fig.32A , Fig.32B The invention is a schematic diagram illustrating different embodiments of applying subsurface interconnects in peripheral circuits or logic circuits of a memory.

[0051] Fig.33 A schematic diagram illustrating the conventional layout of memory peripheral circuits and logic circuits.

[0052] Fig.34 , Fig.35 , Fig.36 , Fig.37 The diagram is a schematic diagram illustrating the layout of peripheral circuits and logic circuits of a memory after adopting subsurface interconnect technology.

[0053] Fig.38 A schematic diagram illustrating the conventional layout of memory peripheral circuits and logic circuits.

[0054] Fig.39 is a schematic diagram illustrating a power board using subsurface interconnect technology.

[0055] Fig.40 A schematic diagram illustrating the conventional layout of memory peripheral circuits and logic circuits.

[0056] Fig.41 is a schematic diagram illustrating a ground plane using subsurface interconnect technology.

[0057] Fig.42 It is similar to Fig.40 and Fig.38 A schematic diagram of another conventional layout of a combinational logic circuit.

[0058] Fig.43 is a schematic diagram illustrating a subsurface interconnect layer 1 board for power supply.

[0059] Fig.44 is a schematic diagram illustrating a subsurface interconnect layer 2 plate for ground voltage.

[0060] Fig.45 is a schematic diagram illustrating a subsurface interconnect layer 1 plate for power supply and a subsurface interconnect layer 2 plate for ground voltage.

[0061] Fig.46 is a schematic diagram illustrating another conventional layout for connecting a power source to the source of a transistor.

[0062] Fig.47 is a schematic diagram illustrating a method of combining a direct source / drain power connection technique with a subsurface interconnection line technique.

[0063] The reference numerals are described as follows:

[0064] 11 Access transistor

[0065] 12 Storage capacitor

[0066] 80 Sensor amplifier

[0067] 70 Data Line Sense Amplifier

[0068] 75 Memory Array

[0069] 202 p-type substrate

[0070] 204 Pad Oxide Layer

[0071] 206 pad nitride layer

[0072] 208, HSS-1 / 2, HSS-1 / 3, HSS horizontal silicon surface

[0073] 210 Groove

[0074] 214 Oxide layer

[0075] 304, 1902, 2202 Spin-on dielectrics

[0076] 306, 1106 photoresist layer

[0077] 502 First Oxide Layer

[0078] 602 Metal Layer

[0079] 902, UGBL, UGBL0B, UGBL1B surface lower bit line

[0080] 1002, CVD-STI-Oxide2 Second oxide layer

[0081] 1102 Third Oxide Layer

[0082] 1104 Second Nitride Layer

[0083] 1302 U-shaped channel

[0084] 1304 High dielectric constant insulation layer

[0085] 1306 Gate Material

[0086] 1402 Third nitride layer

[0087] 1404 Fourth Oxide Layer

[0088] 1602 Fourth Nitride Layer

[0089] 1604 Fifth Oxide Layer

[0090] 1606 Fifth Nitride Layer

[0091] 1802 Sixth Oxide Layer

[0092] 2203 Equalization Circuit

[0093] 2102 Seventh Oxide Layer

[0094] 2402 Eighth Oxide Layer

[0095] 2601, 2602, 2603, 2702, 2710, 2712, subsurface interconnects 2902, 2906, 3002, 3004, 3102, 3104, 3106, 3202, 3204

[0098] 2604, 2605 latch

[0099] 27021, 27022, 27023, 27102, 27104 sublayer 2802 n+ heavily doped polysilicon film 2904 Semiconductor Substrate

[0100] 3202 n+ polysilicon layer

[0101] 3302 Ninth Oxide Layer

[0102] 3301, 3804, 4004, 4604, 4704 Metal Zero Layer

[0103] 3304 N+ protection ring

[0104] 3306 P+ protection ring

[0105] 3802 N-protection ring

[0106] 3806, 4606 Contact

[0107] 3808, 4006, 4602, 4608, 4702 Metal 1 layer

[0108] 3810 Square Contact

[0109] 3902 Subsurface Power Supply VDD Board

[0110] 3904, 4104 Contact pin

[0111] 4002 P-Protective Ring

[0112] 4003 P-type substrate

[0113] 4102 Subsurface Ground VSS Plate

[0114] 4302 Subsurface Interconnect Layer 1 Board

[0115] 4402 Subsurface Interconnect Layer 2 Board

[0116] A, N points

[0117] AQ1, AQ2, AQ3 access transistors

[0118] BL, BL0, BL1 bit lines

[0119] BLB, BL0B, BL1B complementary bit lines

[0120] BS100, BS1, BS2, BS0 position switch

[0121] BLSA0, BLSA1, BLSAn-1, BLSAn bit line sense amplifiers

[0122] DI data input circuit

[0123] Drain-1 First drain

[0124] Drain-2 Second drain

[0125] DL cable

[0126] DLB complementary data line

[0127] EN1, EN2, SLNS, eqpl control signals

[0128] E1 First End

[0129] E2 Second End

[0130] GIO Global Input / Output Routing

[0131] Hole-1 / 2, Hole-1 / 3

[0132] LSLP, LSLN, 1~23, 1a~23a connection line

[0133] N1, N2, NMOS N-type metal oxide semiconductor transistor

[0134] Node1, Node2, Node3, Node4, Node5, internal nodes

[0135] Node6, Node_n

[0136] oxide-7plug seventh oxide plug

[0137] P1, P2, PMOS P-type metal oxide semiconductor transistor

[0138] R1, R2, R3 resistors

[0139] Source-1 First source

[0140] STI, 2704, 2706, 2708 shallow trench isolation

[0141] SLP, SLN transistors

[0142] Source-3 The third source

[0143] Vp1 common voltage

[0144] VCCSA, VSS supply voltage

[0145] VSSSA, VBL voltage

[0146] VS1 vertical spacing

[0147] VDD1, VDD2 power supply

[0148] VSS1, VSS2 ground voltage

[0149] Tungsten plug

[0150] WL Word Line

[0151] Wordline-1 First wordline

[0152] Wordline-2 Second wordline

[0153] Wordline-3 The third wordline

[0154] XIO write data

[0155] Steps 10-50, 102-156 DETAILED DESCRIPTION

[0156] In order to illustrate that the present invention covers such a wide range of integrated circuits including logic or digital, analog, memory, etc., the power connection circuit is first selected as the focus in the following description of the present invention to make the technical features of the present invention easier to grasp.

[0157] Please refer to Figure 1A , Figure 1B , Figure 1C , Figure 1D , Figure 1E , Figure 1F , Figure 1G ,in Figure 1A The present invention is a flowchart of a method for manufacturing a memory array circuit according to an embodiment of the present invention. Figure 1A Include:

[0158] Step 10: Start;

[0159] Step 15: On the basis of the substrate (i.e., the p-type substrate 202, and the present invention takes the p-type substrate 202 as a silicon substrate as an example), define the active region of the memory array circuit and form shallow trench isolation (STI);

[0160] Step 20: forming an asymmetric spacer layer along the sidewall of the active region;

[0161] Step 25: forming an underground interconnection line (an underground bitline is used as an example in the present invention) between the asymmetric spacer layers and below the horizontal surface of the substrate;

[0162] Step 30: forming a cell line and a gate of an access transistor of the memory array circuit;

[0163] Step 35: defining the drain region and the source region of the access transistor of the memory array circuit;

[0164] Step 40: forming a connection between the subsurface bit line and the drain region of the access transistor;

[0165] Step 45: forming the drain region and the source region;

[0166] Step 50: End.

[0167] Please refer to Figure 1B and Figure 2 , step 15 comprises:

[0168] Step 102: Depositing a pad oxide layer 204 and a pad nitride layer 206 on a horizontal surface of the substrate (the present invention takes the p-type substrate 202 as a silicon substrate as an example, so the horizontal surface is a horizontal silicon surface (horizontal silicon surface 208), and the horizontal silicon surface 208 is also called HSS) Figure 2 );

[0169] Step 104: define the active region of the memory array circuit, and remove the substrate material (ie, silicon material) outside the active region corresponding to the horizontal silicon surface 208 to form a trench 210 ( Figure 2 );

[0170] Step 106: Deposit and etch back the oxide layer 214 in the trench 210 to form the shallow trench isolation (STI) below the horizontal silicon surface 208. Figure 2 ).

[0171] Please refer to Figure 1C , Figure 3 , Figure 4 , step 20 comprises:

[0172] Step 108: depositing and etching back the first nitride layer to form a first nitride spacer, and depositing a spin-on dielectric (SOD) 304 in the trench 210 and planarizing the SOD 304 by chemical mechanical polishing (CMP) technology ( Figure 3 );

[0173] Step 110: Etch away the upper edge of the first nitride spacer layer, the first nitride spacer layer and the spin-on dielectric 304 ( Figure 4 );

[0174] Step 112: Generate a first oxide layer 502 (eg, thermally generated) and deposit a metal layer 602 in the trench 210 and planarize it by the chemical mechanical polishing technique. Figure 4 ).

[0175] Please refer to Figure 1D , Figure 5 , Figure 6 , step 25 comprises:

[0176] Step 114: Use photolithographic technique to form a plurality of interconnection (ie metal layer 602) lines ( Figure 5 );

[0177] Step 116: Etch back the metal layer 602 (the plurality of interconnects) to form underground bitlines (UGBL) 902 ( Figure 5 );

[0178] Step 118: Deposit the second oxide layer 1002 in the trench 210 and planarize it by the chemical mechanical polishing technique. Figure 6 ).

[0179] Please refer to Figure 1E , Figure 7 , Figure 8 , step 30 comprises:

[0180] Step 120: deposit a thick third oxide layer 1102, a thick second nitride layer 1104 and a patterned photoresist layer 1106, and then etch away unnecessary portions of the third oxide layer 1102 and the second nitride layer 1104 ( Figure 7 );

[0181] Step 122: Remove the patterned photoresist layer 1106, the pad nitride layer 206 and the pad oxide layer 204, and expose the horizontal silicon surface 208 ( Figure 7 );

[0182] Step 124: Etching the exposed horizontal silicon surface 208 to form a U-shaped recess, forming a high-k insulating layer 1304 in the U-shaped recess, and depositing and etching back a gate material 1306 (e.g., tungsten) to form a gate line and gate structure of the access transistor, wherein the access transistor can be referred to as a U-shaped transistor ( Figure 8 );

[0183] Step 126: deposit and etch back the third nitride layer 1402, then deposit the fourth oxide layer 1404, and then etch back or planarize the fourth oxide layer 1404 ( Figure 8 );

[0184] Step 128: Etch away the second nitride layer 1104 and the third oxide layer 1102 ( Figure 8 ).

[0185] Please refer to Figure 1F , Fig. 9 , Fig.10 , step 35 comprises:

[0186] Step 130: Remove the pad nitride layer 206 and etch back the second oxide layer 1002 to the top of the pad oxide layer 204 ( Fig. 9 );

[0187] Step 132: Deposit and anisotropically etch the fourth nitride layer 1602, the fifth oxide layer 1604 and the fifth nitride layer 1606 ( Fig. 9 );

[0188] Step 134: Deposit the sixth oxide layer 1802 ( Fig. 9 );

[0189] Step 136: Deposit and then planarize the spin-on dielectric 1902 by the chemical mechanical planarization technique, deposit a photoresist layer, and then etch back the spin-on dielectric 1902 to be removed ( Fig.10 );

[0190] Step 138: Etch away the exposed sixth oxide layer 1802, the pad oxide layer 204, and the silicon material corresponding to the horizontal silicon surface HSS-1 / 3 to form a hole hole-1 / 3 ( Fig.10 );

[0191] Step 140: Thermally generate the seventh oxide layer 2102 ( Fig.10 );

[0192] Step 142: Deposit a spin-on dielectric 2202 on the seventh oxide layer 2102, and then etch back the spin-on dielectric 2202 ( Fig.10 ).

[0193] Please refer to Figure 1G , Fig.11 , Fig.12 , Fig.13 , Fig.14 , Fig.15 , Fig.16 , step 40 comprises:

[0194] Step 144: remove the exposed spin-on dielectric 1902, the exposed sixth oxide layer 1802 and the exposed pad oxide layer 204, and then dig out the silicon material corresponding to the horizontal silicon surface HSS-1 / 2 to form a hole hole-1 / 2 ( Fig.11 );

[0195] Step 146: Thermally generate the eighth oxide layer 2402 ( Fig.11 );

[0196] Step 148: Remove the first nitride spacer at the lower edge to expose the sidewall of the lower bit line (UGBL) Fig.13 );

[0197] Step 150: Deposit an n+ heavily doped polysilicon film 2802 in the hole hole-1 / 2 to contact the sidewall of the lower bit line (UGBL) ( Fig.13 );

[0198] Step 152: Etch back the upper portion of the eighth oxide layer 2402 to expose the silicon material corresponding to the hole hole-1 / 2 ( Fig.14 );

[0199] Step 154: Based on the exposed silicon material, an n+ polysilicon layer 3202 is laterally grown to contact the drain region and the n+ polysilicon plug ( Fig.15 );

[0200] Step 156: Thermally generate a ninth oxide layer 3302 on the n+ polysilicon layer 3202 ( Fig.15 ).

[0201] The metal layers used in the above process steps (eg Figure 5 The subsurface bit line (UGBL) 902 is shown. Fig.10The material of the bit lines, electrodes of capacitors and / or common electrodes shown in the figure can be tungsten, however, due to the sensitivity of tungsten materials to oxides or oxidation processes, it is best that the tungsten material can be covered by another titanium nitride layer or a suitable layer. Because the characteristics of the tungsten material should be clear to the technicians in the field of the present invention, the present invention does not describe the detailed protection process of the tungsten material, but assumes that the metal layers including the tungsten material are well treated to avoid any oxidation reaction directly thereon. Of course, there are some suitable metal layers that can still be applied to the subsurface bit lines and the bit lines, without being limited to a specific type of metal material that is not suitable for insertion into the integrated circuit process.

[0202] The detailed description of the above manufacturing method is as follows, starting from the p-type substrate 202. In step 102, as Figure 2 As shown in FIG. 5( a ), a pad oxide layer 204 is formed on a horizontal silicon surface 208 (ie, the original surface as described in claim 1 ), and then a pad nitride layer 206 is deposited on the pad oxide layer 204 .

[0203] In step 104, the active region of the memory array circuit may be defined by photolithographic technique, wherein Figure 2 As shown in (a), since the active region of the memory array circuit corresponds to the pad oxide layer 204 and the pad nitride layer 206, the horizontal silicon surface 208 outside the active region pattern will be exposed accordingly. Since the horizontal silicon surface 208 outside the active region pattern is exposed, the silicon material corresponding to the horizontal silicon surface 208 outside the active region pattern can be removed by anisotropic etching technique to produce a trench (or channel) 210, wherein, for example, the trench 210 can be 250 nanometers (nm) deep below the horizontal silicon surface 208.

[0204] In step 106, an oxide layer 214 is deposited to fill the trench 210, and then the oxide layer 214 is etched back to form the shallow trench isolation below the horizontal silicon surface 208 in the trench 210. In addition, Figure 2 (b) corresponds to Figure 2 (a) is a top view, where Figure 2 (a) is along the Figure 2 (b) is a cross-sectional view of the cutting line in the X direction. Figure 2 As shown in (a), for example, the shallow trench isolation has a thickness of about 50 nm, and if the trench 210 below the horizontal silicon surface 208 has a depth of 250 nm, the top of the shallow trench isolation is about 200 nm deep from the horizontal silicon surface 208.

[0205] In step 108, if Figure 3 As shown in (a), the first nitride layer is deposited and the first nitride layer is etched back using the anisotropic etching technology to form the first nitride spacer along the two edges of the trench 210 (that is, the first nitride spacer at the upper edge and the first nitride spacer at the lower edge). In another embodiment of the present invention, the first nitride spacer formed on one of the two edges of the trench 210 can be replaced by silicon oxycarbonitride (SiOCN). Then, as Figure 3 As shown in (a), a spin-on dielectric 304 is deposited on the shallow trench isolation in the trench 210 to fill the vacancy above the shallow trench isolation in the trench 210. The spin-on dielectric 304 is then planarized by the chemical mechanical planarization technique so that the top of the spin-on dielectric 304 is flush with the top of the pad nitride layer 206. In addition, Figure 3 (b) corresponds to Figure 3 (a) is a top view, where Figure 3 (a) is along the Figure 3 (b) is a cross-sectional view taken along the cutting line in the Y direction.

[0206] In step 110, if Figure 4 As shown in FIG. 1 , the first nitride spacer layer at the lower edge of the groove 210 is protected by the photoresist layer 306 using the lithography technology, but the first nitride spacer layer at the upper edge of the groove 210 is not protected by the photoresist layer 306. That is, the photoresist layer 306 is deposited on the spin-on dielectric 304 and the liner nitride layer 206 (as shown in FIG. 1 ). Figure 3 After (a), because the photoresist layer 306 on the upper edge first nitride spacer is removed but the photoresist layer 306 on the lower edge first nitride spacer is retained, the lower edge first nitride spacer can be retained but the upper edge first nitride spacer will be removed.

[0207] In step 112, if Figure 4 As shown, the upper edge first nitride spacer layer and the spin-on dielectric 304 not covered by the photoresist layer 306 can be etched away by an etching process. Figure 4 As shown, the photoresist layer 306 and the spin-on dielectric 304 are stripped, wherein the spin-on dielectric 304 has a much higher etching rate than the thermally generated oxide layer and the deposited oxide layer. Then the first oxide layer 502 is thermally generated to form a first oxide spacer, wherein the first oxide spacer covers the upper edge of the trench 210. Figure 4As shown, step 112 results in the formation of an asymmetric spacer (i.e., the first nitride spacer at the lower edge and the first oxide spacer) at two symmetrical edges of the trench 210 (i.e., the upper edge and the lower edge of the trench 210). For example, the thickness of the first oxide spacer is 4 nanometers and the thickness of the first nitride spacer at the lower edge is 3 nanometers. On the other hand, the asymmetric spacer (i.e., the first oxide spacer and the first nitride spacer at the lower edge) is also formed along the sidewall of the active region. The asymmetric spacer (i.e., Figure 4 The structure of the present invention and the above-mentioned related steps are the first main technical features of the present invention, which are called asymmetric spacers on two symmetrical edges of a trench or a canal (ASoSE).

[0208] In addition, if Figure 4 As shown, a metal layer 602 (e.g., a thin titanium nitride layer plus tungsten) or a conductive material (e.g., doped polysilicon) that needs to withstand subsequent process conditions is deposited to completely fill the trench 210, and is planarized by the chemical mechanical polishing technique so that the top of the metal layer 602 is flush with the top of the pad nitride layer 206.

[0209] Then, in step 114, if Figure 5 As shown in (a), the lithography technique is used to form a plurality of conductive lines (e.g., titanium nitride / tungsten lines). That is, the metal layer 602 is etched back using the lithography technique to leave a metal layer 602 of a reasonable thickness inside the trench 210, thereby forming conductive lines or subsurface bit lines (UGBL) 902, wherein the top of the subsurface bit line 902 is much lower than the horizontal silicon surface HSS (e.g., the thickness of the subsurface bit line 902 is about 40 nm). In addition, as shown in FIG. Figure 5 As shown in (a), the UGBL 902 is located on the top of the STI, and the two sidewalls of the UGBL 902 are defined by the asymmetric spacer (ie, the first oxide spacer and the lower edge first nitride spacer). Figure 5 (a) is along Figure 5 (b) is a cross-sectional view taken along the Y direction.

[0210] In step 118, if Figure 6 (Along as Figure 5As shown in (b) of the cross-sectional view of the cutting line in the Y direction, the second oxide layer 1002 (also called CVD-STI-oxide2) needs to have a sufficient thickness to fill the trench 210 above the bit line 902 below the surface, and then the second oxide layer 1002 is planarized by the chemical mechanical planarization technology to retain a portion of the second oxide layer 1002, wherein the top of the retained portion of the second oxide layer 1002 is flush with the top of the pad nitride layer 206 and covers the lower edge first nitride spacer and the first oxide spacer. Figure 6 As shown, step 118 can make the subsurface bit line 902 (that is, the interconnection wire) embedded in all insulators (that is, the isolation region) in the trench 210 and confined by all the insulators (the subsurface bit line 902 will then be connected to the drain of the access transistor of the memory array circuit), wherein Figure 6 The structure shown is called an insulator-surrounded subsurface bit line, and the subsurface bit line (UGBL) is the second main technical feature of the present invention.

[0211] The following describes how to form the access transistor and the gate line of the memory array circuit, and the gate structure of the gate line connecting the access transistor. Figure 7 As shown in (a), firstly, a third oxide layer 1102, a second nitride layer 1104, and a patterned photoresist layer 1106 are deposited. Then, the unnecessary parts of the third oxide layer 1102 and the second nitride layer 1104 are removed by the lithography technology. In addition, a transistor / word line pattern can be defined by a composite layer composed of the third oxide layer 1102 and the second nitride layer 1104, wherein the composite layer composed of the third oxide layer 1102 and the second nitride layer 1104 is composed of a plurality of strip-shaped third oxide layers 1102 and the second nitride layer 1104 in a direction perpendicular to the active region direction. Therefore, as Figure 7 (a) and Figure 7 As shown in (b), longitudinal (the Y direction) stripes (composed of the third oxide layer 1102 and the second nitride layer 1104) for defining the access transistor and the element line, and a cross-point square for defining the active region will be formed, wherein the active region is the cross-point square located between the two longitudinal stripes, and Figure 7 (a) is along the Figure 7 (b) is a cross-sectional view taken along the cutting line in the X direction.

[0212] In step 122, the photoresist layer 1106 is retained to etch away the pad nitride layer 206, but the pad oxide layer 204 is retained, and then the photoresist layer 1106 and the pad oxide layer 204 are removed. Therefore, the horizontal silicon surface 208 (ie, HSS) is exposed as shown in FIG. Figure 7 (b) shows the cross-point square area corresponding to the active area.

[0213] In step 124, if Figure 8 As shown, the horizontal silicon surface 208 exposed in the cross-point square region is etched by the anisotropic etching technology to form a recess (e.g., the U-shaped recess), wherein the U-shaped recess is used to form a U-shaped channel 1302 of the access transistor, and the vertical depth of the U-shaped recess can reach about 60 nanometers, for example, starting from the horizontal silicon surface 208. Because the U-shaped recess of the access transistor is exposed, the U-shaped channel 1302 in the U-shaped recess can be doped with a properly designed concentration of boron (p-type dopant) to achieve channel doping, wherein the channel doping is to enable the access transistor to have a desired threshold voltage after a high-k metal-gate structure is subsequently formed. Then, a high-k dielectric insulating layer 1304 is formed at the bottom and sidewalls of the U-shaped recess, wherein the high-k dielectric insulating layer 1304 serves as a gate dielectric layer of the access transistor. A gate material 1306 is then selected that is suitable for the cell line conductivity and that can achieve the targeted work-function performance of the access transistor. Figure 8 As shown, the gate material 1306 must be thick enough to fill the U-shaped recess, and the gate material 1306 is etched back to form a longitudinal (the Y direction) element line.

[0214] In step 126, if Figure 8 As shown, a third nitride layer 1402 (i.e., a dielectric cap) is deposited, and then a fourth oxide layer 1404 is deposited, wherein the third nitride layer 1402 and the fourth oxide layer 1404 are stacked together to make their total thickness sufficient to fill the gap between the two adjacent longitudinal stripes (composed of the third oxide layer 1102 and the second nitride layer 1104) and finally form a flat surface. Then, the fourth oxide layer 1404 is etched back (or polished) to make the top of the fourth oxide layer 1404 flush with the top of the second nitride layer 1104, thereby forming a composite layer composed of the fourth oxide layer 1404 and the third nitride layer 1402 directly above the element line (i.e., the gate material 1306).

[0215] In step 128, if Figure 8As shown, the second nitride layer 1104 is etched away by the anisotropic etching technique, and the fourth oxide layer 1404 / third nitride layer 1402 is left on the top of the word line. Then the third oxide layer 1102 is also etched away by the anisotropic etching technique to expose the pad nitride layer 206. The gate structure (e.g., the fourth oxide layer 1404 / third nitride layer 1402 / gate material 1306) simultaneously implements the gate structure of the U-shaped transistor in the U-shaped recess and the word lines (i.e., the first word line word line-1, the second word line word line-2, and the third word line word line-3) in the longitudinal direction (i.e., the Y direction).

[0216] In step 130, if Fig. 9 As shown, the pad nitride layer 206 is removed to leave the pad oxide layer 204. The second oxide layer 1002 (ie, CVD-STI-oxide2) is etched back to make the top of the second oxide layer 1002 flush with the top of the pad oxide layer 204.

[0217] Then, in step 132, as Fig. 9 As shown, a fourth nitride layer 1602 is deposited and etched by the anisotropic etching technique to generate a fourth nitride spacer with a carefully designed thickness. A fifth oxide layer 1604 is then deposited and etched by the anisotropic etching technique to generate a fifth oxide spacer. A fifth nitride layer 1606 is then deposited. Fig. 9 The fifth nitride layer 1606 is then etched by the anisotropic etching technique to create a fifth nitride spacer so that the fifth nitride spacer surrounds the word line (eg, the first word line-1, the second word line-2, the third word line-3).

[0218] like Fig. 9As shown, for the convenience and clarity of describing the memory array circuit having the word lines and the bit lines, the word line located in the center is marked as the first word line word line-1 (corresponding to the access transistor AQ1), the word line adjacent to the left of the first word line word line-1 is marked as the second word line word line-2 (corresponding to the access transistor AQ2 adjacent to the left of the access transistor AQ1), and the pad oxide layer 204 still covers the area between the first word line word line-1 and the second word line word line-2 as the drain area to reserve for the first drain Drain-1 of the access transistor AQ1 and the second drain drain-2 of the access transistor AQ2. The word line adjacent to the right of the first word line word line-1 is marked as the third word line word line-3 (corresponding to the access transistor AQ3 adjacent to the right of the access transistor AQ1), and the pad oxide layer 204 still covers the area between the first word line word line-1 and the third word line word line-3 as the source region to reserve the first source source-1 of the access transistor AQ1 and the third source source-3 of the access transistor AQ3. Then, in step 134, Fig. 9 A sixth oxide layer 1802 is deposited over the entire surface as shown.

[0219] In step 136, if Fig.10 As shown, a spin-on dielectric 1902 is deposited, wherein the thickness of the spin-on dielectric 1902 is sufficient to fill the gaps between the above-mentioned word lines (the first word line word line-1, the second word line word line-2, and the third word line word line-3), and then the spin-on dielectric 1902 is polished by the chemical mechanical planarization technology to make the top of the spin-on dielectric 1902 flush with the top of the sixth oxide layer 1802. Then, a photoresist is deposited on the flush surface to cover the area reserved for the drain region (the first drain drain-1 and the second drain drain-2), and to expose the area reserved for the source region (the first source source-1 and the third source source-3). Then, as shown in FIG. Fig.10 As shown, the sixth oxide layer 1802 surrounding the above-mentioned word lines (the first word line word line-1, the second word line word line-2 and the third word line word line-3) can be used as a self-alignment mask to remove the spin-coated dielectric 1902 corresponding to the area reserved for the source region (the first source source-1 and the third source source-3).

[0220] Then, in step 138, Fig.10As shown, the exposed sixth oxide layer 1802 and the pad oxide layer 204 at the center of the source region between the two wordlines (the first wordline-1 and the third wordline-3) are etched away to expose the horizontal silicon surface 208. Because the exposed horizontal silicon surface 208 is located between the first source source-1 of the access transistor AQ1 and the third source source-3 of the access transistor AQ3, the horizontal silicon surface 208 between the first source source-1 and the third source source-3 is called the horizontal silicon surface HSS-1 / 3. Fig.10 As shown, the horizontal silicon surface HSS-1 / 2 between the first element line wordline-1 and the second element line wordline-2 is used as the region of the first drain drain-1 (that is, the drain of the access transistor AQ1) and the second drain drain-2 (that is, the drain of the access transistor AQ2), and is also used as the region vertically connecting the access transistors AQ1 and AQ2 to the subsurface bit line 902. In addition, to the right of the first element line wordline-1, the horizontal silicon surface HSS-1 / 3 between the first element line wordline-1 and the third element line wordline-3 is used as the region of the first source source-1 (that is, the source of the access transistor AQ1) and the third source source-3 (that is, the source of the access transistor AQ3), but because the first source source-1 and the third source source-3 will be connected to the memory cell storage nodes CSN1 and CSN3 (not shown) respectively. Fig.10 ), so the first source source-1 and the third source source-3 must be separated and cannot be connected.

[0221] In addition, the above-mentioned lithography technology is used to cover the horizontal silicon surface HSS-1 / 2, but the mask used in the lithography technology is not a critical mask, and its function is only to separate the process of processing the horizontal silicon surface HSS-1 / 3 from the process of processing the horizontal silicon surface HSS-1 / 2. As mentioned above, the thickness of the deposited spin-on dielectric 1902 must be sufficient to form a smooth surface morphology, and then a photoresist is deposited as a mask to protect the spin-on dielectric 1902 covering the drain region (the first drain drain-1 and the second drain drain-2) but expose the spin-on dielectric 1902 covering the source region (the first source source-1 and the third source source-3).

[0222] like Fig.10 As shown, the silicon material below the horizontal silicon surface HSS-1 / 3 (corresponding to the center of the source region) is excavated by the anisotropic etching to generate a hole hole-1 / 3 (e.g., 140 nm deep), wherein two opposite sides of the hole hole-1 / 3 (not shown in FIG. Fig.10 ) are respectively surrounded by the first nitride spacer layer and the first oxide spacer layer at the lower edge, and the other two pairs of sides of the hole hole-1 / 3 are surrounded by the p-type substrate 202.

[0223] Then, in step 140, as Fig.10 As shown, the seventh oxide layer 2102 is thermally generated to fill the hole hole-1 / 3, and the isotropic etching is used to accurately remove all the seventh oxide layer 2102 above the horizontal silicon surface 208, but the seventh oxide layer 2102 filling the hole hole-1 / 3 is left, wherein the seventh oxide layer 2102 filling the hole hole-1 / 3 is called a seventh oxide plug (oxide-7 plug), wherein the seventh oxide plug has a smooth surface and its height is flush with the top of the pad oxide layer 204.

[0224] Then, in step 142, as Fig.10 As shown, a spin-coated dielectric 2202 is deposited, wherein the thickness of the spin-coated dielectric 2202 is sufficient to fill the vacancy above the top of the seventh oxide layer 2102 in the hole 1 / 3, and then the top material of the spin-coated dielectric 2202 is removed by the chemical mechanical planarization technology so that the top of the spin-coated dielectric 2202 is the same height as the top of the fourth oxide layer 1404.

[0225] In step 144, if Fig.11 As shown, the photoresist 2302 is deposited by the lithography technology to cover the area corresponding to the source region and expose the area reserved for the drain region, wherein the mask used in step 144 is not a critical mask, and its function is only to separate the process of processing the horizontal silicon surface HSS-1 / 3 from the process of processing the horizontal silicon surface HSS-1 / 2. Then the exposed spin-on dielectric 1902, the exposed sixth oxide layer 1802, and the exposed liner oxide layer 204 are removed to expose the horizontal silicon surface (that is, the horizontal silicon surface HSS-1 / 2). Then, the silicon material corresponding to the horizontal silicon surface HSS-1 / 2 is excavated and removed by the anisotropic etching technology to generate a hole hole-1 / 2, wherein two pairs of sides of the hole hole-1 / 2 are respectively surrounded by the p-type substrate 202, and the third side of the hole hole-1 / 2 (not shown) is Fig.11 ) is surrounded by the first nitride spacer layer at the lower edge, and the fourth side of the hole hole-1 / 2 (not shown in FIG. Fig.11 ) is surrounded by the first oxide spacer layer, and the third side and the fourth side of the hole hole-1 / 2 are further surrounded by the second oxide layer 1002 (ie, CVD-STI-oxide2, not shown in FIG. Fig.11 ) at the external limit.

[0226] Then, in step 146, if Fig.11As shown, the photoresist is removed, and then an eighth oxide layer 2402 is thermally generated to produce an eighth oxide spacer, wherein among the four sidewalls of the hole hole-1 / 2, except for the third sidewall which is covered by the first nitride spacer at the lower edge, the eighth oxide spacer covers the remaining three sidewalls of the four sidewalls of the hole hole-1 / 2 and the bottom of the hole hole-1 / 2. In addition, Fig.12 is a schematic diagram illustrating a cross-sectional view of the memory array circuit along the Y2 direction, wherein the Y2 direction extends along the center of the hole Hole-1 / 2 and is perpendicular to the X direction.

[0227] Then, in step 148, if Fig.13 The lower edge first nitride spacer on the third sidewall in the hole hole-1 / 2 is removed by the isotropic etching technique (because the lower edge first nitride spacer is very thin, the isotropic etching technique will not damage other structures on the horizontal silicon surface 208, and will not remove the eighth oxide layer 2402 in the hole hole-1 / 2).

[0228] Then, in step 150, if Fig.13 As shown, an n+ heavily doped polysilicon film 2802 is deposited, wherein the thickness of the n+ heavily doped polysilicon film 2802 is sufficient to fill the hole hole-1 / 2, and then the n+ heavily doped polysilicon film 2802 on the horizontal silicon surface 208 is etched back by the isotropic etching technology to leave an n+ polysilicon plug (n+polysiliconplug) in the hole hole-1 / 2, and the n+ polysilicon plug is connected to the undersurface bit line (UGBL) through the opening on the third sidewall of the hole hole-1 / 2, wherein the third sidewall of the hole hole-1 / 2 was originally covered by the first nitride spacer at the lower edge.

[0229] like Fig.13 As shown, the connection between the n+ polysilicon plug and the UGBL is from the sidewall of the n+ polysilicon plug to the sidewall of the tungsten interconnect (that is, the UGBL) in the hole Hole-1 / 2, wherein the n+ polysilicon plug and the UGBL are both conductive materials and connected in a sidewall self-aligned manner, but are completely isolated from the p-type substrate 202 by the eighth oxide layer 2402. Such an n+ polysilicon plug is a vertical interconnection line, or an interconnection line extending upward or downward.

[0230] Then, in step 152, if Fig.14As shown, an appropriate amount (for example, a height of 20 nm below the horizontal silicon surface 208) of the upper half of the eighth oxide layer 2402 is removed by the anisotropic etching technology, resulting in the height of the eighth oxide spacer (eighth oxide layer 2402) being lower than the height of the n+ polysilicon plug.

[0231] Then, in step 154, if Fig.15 As shown, an n+ polysilicon layer 3202 is grown from the silicon seed provided by the n+ polysilicon plug by the selective epitaxial growth (SEG) technology, resulting in a necklace-type conductive n+ polysilicon layer 3202 (called an n+ collar) connected to the horizontal silicon surface 208 on both sides of the hole hole-1 / 2, which can serve as the first drain drain-1 of the access transistor AQ1 and the second drain drain-2 of the access transistor AQ2, respectively, and can also serve as a conductive bridge between the subsurface bit line (UGBL) and the access transistors AQ1 and AQ2.

[0232] Then, if Fig.15 As shown, a ninth oxide layer 3302 with a well-designed thickness is locally thermally generated on the n+ collar to cover the horizontal silicon surface HSS-1 / 2, but the n+ collar is retained to serve as the first drain drain-1 of the access transistor AQ1 and the second drain drain-2 of the access transistor AQ2, respectively. Such a conductive plug (including the n+ polysilicon plug and the n+ collar) is a vertical interconnection line connected to a horizontally extending subsurface interconnection (subsurface bit line (UGBL)).

[0233] Fig.16Schematic diagram of realizing the vertical connection (bridge) of the access transistor and the drain region and how they are connected to the subsurface interconnection line disclosed in another embodiment of the present invention. The eighth oxide layer 2402 is thermally generated by following the above process and the first nitride spacer layer at the lower edge in the hole hole-1 / 2 is removed (instead of depositing and etching back the n+ heavily doped polysilicon film 2802 to leave the n+ polysilicon plug in the hole hole-1 / 2), and the hole hole-1 / 2 is filled with tungsten or other metal material (by deposition or any other means), and then etched back to leave a tungsten plug (called W-2 plug) in the hole hole-1 / 2. The W-2 plug is connected to the subsurface interconnection through its opening located on the sidewall of the hole hole-1 / 2 covered by the first nitride spacer layer. The height of the W-2 plug is lower than the horizontal silicon surface 208 by a well-designed distance (for example, about 20nm). The exposed portion of the eighth oxide layer 2402 within the hole hole-1 / 2 but not covered by the W-2 plug is removed using the anisotropic etching technique. Then, an n+ polysilicon layer (e.g., n+ heavily doped polysilicon film 2802) is deposited to fill the hole hole-1 / 2, and then the n+ polysilicon layer is etched back to make the top of the n+ polysilicon layer flush with the horizontal silicon surface 208, and the necklace-shaped conductive n+ polysilicon layer connected to the horizontal silicon surface 208 on both sides of the hole hole-1 / 2 can serve as the first drain drain-1 of the access transistor AQ1 and the second drain drain-2 of the access transistor AQ2, respectively, and can also serve as a conductive bridge between the undersurface bit line (UGBL) 902 and the access transistors AQ1 and AQ2. Then, a ninth oxide layer 3302 is locally thermally grown over the tungsten plug to cover the horizontal silicon surface HSS-1 / 2 but leave the n+ collar as the first drain drain-1 of the access transistor AQ1 and the second drain drain-2 of the access transistor AQ2.

[0234] The following is an embodiment of the present invention to explain the formation of the drain region / source region of the access transistor. In step 45, as Fig.17As shown, the sixth oxide layer 1802 and the fifth nitride layer 1606 are removed, and then n-type doped drain regions and source regions are formed in the p-type substrate 202, respectively (for example, the n-type dopant injected through the pad oxide layer 204 can be used to create the drain region and the source region with an np junction (that is, the first drain Drain-1 and the first source Source-1 of the access transistor AQ1, the second drain Drain-2 of the access transistor AQ2, and the third source Source-3 of the access transistor AQ3) in the p-type substrate 202). In addition, a rapid thermal annealing (RTA) process step is required to activate the n-type dopant to eliminate any defects caused by ion implantation. Of course, other methods can also be used to form the drain region and the source region, such as selective generation / selective epitaxial generation. In addition, n-type lightly doped drains (NLDD) can be formed under the fifth oxide spacer and the fourth nitride spacer. Afterwards, a conventional stack capacitor or other type of capacitor (not shown) may be formed to connect the first source Source-1 of the access transistor AQ1 to complete the memory cell (ie, dynamic random access memory cell) in the memory array circuit.

[0235] Therefore, as described above, a subsurface bit line (or subsurface interconnect) is formed and connected to the drain region of the access transistor (e.g., access transistor AQ1). In addition, some drain regions or source regions of different access transistors (e.g., access transistors AQ3, AQ2) can be connected to the subsurface interconnect that provides the back power supply Vdd or Vss. In addition, in another embodiment of the present invention, according to the above process, two or more different subsurface interconnects can also be formed at two or more different vertical heights in the pipeline, and separated from each other by isolation materials (e.g., Fig.18 The different subsurface interconnects may have different lengths and may be connected to different transistors. In another embodiment of the present invention, the different subsurface interconnects in the pipeline may be connected to different transistors. Fig.18 In addition, the different subsurface interconnects may be made of different materials and may have different thicknesses.

[0236] Because the subsurface interconnect can have a similar resistance to the metal 0 used in the fab, as described above, the subsurface interconnect can be used as a bit line in the memory array circuit to free up layout space and reduce the size of the memory array circuit. Fig.19 , Fig.19is a schematic diagram illustrating a conventional dynamic random access memory structure, wherein write data XIO (e.g., signal "ONE" or a high potential signal) is transmitted along the data input circuit DI, the global input / output path GIO, the data line sense amplifier 70, and the data line DL. In addition, the write data XIO is transmitted between the data line DL and the memory array 75, wherein the write data XIO is stored in the corresponding storage node in the memory array 75 through the bit line BL. In the memory array 75, as shown in FIG. Fig. 20 As shown, the sense amplifier 80 is connected to the bit line BL, wherein the bit line BL is coupled to the data line DL through the bit switch BS100. A plurality of dynamic random access memory cells (e.g., 512 dynamic random access memory cells) are connected to the bit line BL. A first end E1 of the bit line BL is connected to a first dynamic random access memory cell among the plurality of dynamic random access memory cells, and a second end E2 of the bit line BL is connected to a last dynamic random access memory cell among the plurality of dynamic random access memory cells. Fig.21 In the embodiment, a dynamic random access memory cell including an access transistor 11 and a storage capacitor 12 is taken as an example. The gate of the access transistor 11 is coupled to the bit line WL, and the sense amplifier 80 is coupled to the access transistor 11 through the bit line BL. The sense amplifier 80 includes two P-type metal oxide semiconductor transistors ( Fig.21 P1 and P2 in the figure) and two N-type metal oxide semiconductor transistors ( Fig.21 N3 and N4 in FIG. 1 ). In addition, the sense amplifier 80 is connected to the VSSSA power supply through the transistor SLP, wherein the transistor SLP is connected to the P-type metal oxide semiconductor transistors P1 and P2 through the connection line LSLP. In addition, the sense amplifier 80 is also connected to the VSS power supply through the transistor SLN, wherein the transistor SLN is connected to the N-type metal oxide semiconductor transistors N1 and N2 through the connection line LSLN. The dynamic random access memory cell uses the access transistor 11 as a switch to control the charge stored from the bit line BL to the storage capacitor 12 in the write (WRITE) mode or the charge transferred from the storage capacitor 12 to the bit line BL in the read (READ) mode. In addition, as Fig.19 As shown, VCCSA is the supply voltage.

[0237] Thus, the conventional DRAM cell array design is as follows Fig.19 , Fig. 20 , Fig.21 As shown, Fig.21The invention comprises (1) a dynamic random access memory array having a plurality of dynamic random access memory cells, such as the most popular 1T1C cell, wherein the 1T1C cell comprises an access transistor (i.e., 1T, whose threshold voltage Vth is usually around 0.7V) and a storage capacitor (i.e., 1C, which is usually 17fF in a 20-28nm process node, and may be less than 10fF if the process node is lower than 20nm); (2) the drain regions of the access transistors of the plurality of 1T1C cells are respectively connected to interconnection lines (the interconnection lines are called bit lines); (3) the gates of the access transistors of the plurality of 1T1C cells are also respectively connected through interconnections called bit lines. The bit lines are connected to a sense amplifier (i.e., sense amplifier 80) of a cross-coupled circuit such as a complementary metal oxide semiconductor (CMOS). Accordingly, there is another bit line, called a complementary bit line (bit line-Bar), which carries a signal complementary to the signal of the bit line BL and is also connected to the same sense amplifier (i.e., sense amplifier 80). Along the interconnection between the bit line BL and the complementary bit line BLB, other devices are also connected to perform the complete bit line function in operation, such as an equalization circuit 2203 (e.g., an equalization circuit 2203 for equalizing voltage potentials) Fig. 22 as shown) and for controlling the bit lines ( Fig. 22 BL and BLB) and data lines ( Fig. 22 The bit switch ( Fig. 22 BS1 and BS2 shown in FIG. 1 ). In addition, Fig. 22 As shown in FIG. 2 , SLNS and eqpl are control signals for the sensing amplifier and the equalization circuit 2203, respectively. Fig. 20 As shown, VCCSA and VSS are supply voltages, and EN1 and EN2 are control signals of the sense amplifier 80. Fig. 22 As shown, VSSSA and VBL are voltages.

[0238] Fig. 2222 is a schematic diagram of a conventional sense amplifier design, wherein the sense amplifier has connection lines 1 to 23 for circuit connection. Bit switches BS1 and BS2 are connected to bit line BL and complementary bit line BLB through connection lines 11 and 12, respectively. P-type metal oxide semiconductor transistor P1 is connected to bit line BL and complementary bit line BLB through connection lines 1 and 6, respectively. P-type metal oxide semiconductor transistor P2 is connected to bit line BL and complementary bit line BLB through connection lines 5 and 2, respectively. N-type metal oxide semiconductor transistor N3 is connected to bit line BL and complementary bit line BLB through connection lines 3 and 7, respectively. N-type metal oxide semiconductor transistor N4 is connected to bit line BL and complementary bit line BLB through connection lines 8 and 4, respectively. The three transistors in equalization circuit 2203 are connected to bit line BL and complementary bit line BLB through connection lines 9, 10, 13, and 14.

[0239] On the other hand, Fig.23 , Fig.24 As shown, in a conventional DRAM open-bitline's bit line sense amplifier (BLSA) design, two bit line sense amplifiers (eg Fig.23 , Fig.24 The bit line sense amplifiers BLSA0, BLSA1 in the bit line sense amplifier BLSA0 are placed side by side. The connection lines 1 to 23 in the bit line sense amplifier BLSA0 and the connection lines 1a to 23a in the bit line sense amplifier BLSA1 usually use metal 0 (metal zero layer) or other metal layers during the layout process, and the bit line BL and the complementary bit line BLB also use metal 0. Therefore, those connection lines (connection lines 1 to 23 and connection lines 1a to 23a) made of metal 0 will incur spacing restrictions therebetween and produce a larger layout area.

[0240] However, if Fig.25 As shown, the present invention proposes a subsurface bit line technology, and then connects the subsurface bit line to the above-mentioned bit line sense amplifier. In the present invention, Fig.24 The connection lines 1 to 23 and Fig.24 The connection lines 1a-23a in the bit line sense amplifier BLSA1 are still made of metal 0 or other metal layers like a conventional dynamic random access memory, but the subsurface bit lines (such as the bit line BL and the complementary bit line BLB) are now located below the horizontal silicon surface 208. Therefore, the connection lines made of metal 0 in the bit line sense amplifier BLSA can be connected to the subsurface bit lines through the switch layer. Fig.25As shown, the subsurface bit line is connected to the access transistor of the active area AA (e.g., N+AA) through the connecting pin therein. In one embodiment of the present invention, the active area AA is a silicon area surrounded by the isolation area. Therefore, during the layout process, the subsurface bit line (e.g., the bit line BL and the complementary bit line BLB) can be connected to the active area AA first, and then, as shown in FIG. Fig.25 As shown at the bottom of FIG. , the active area AA is connected to metal 0 through metal 0 contact during the layout process. Fig.18 As shown, the two subsurface bit lines UGBL may be located in one isolation region and / or in two different layers in the isolation region. Fig.23 and Fig.24 shown) can be relaxed.

[0241] In one embodiment of the present invention, the switch layer may include a metal 0 contact, an active area AA connected to the metal 0 contact, and a connection plug connecting the active area AA to the subsurface bit line. Through the connection plug, the active area AA and the metal 0 contact, the connection line made of metal 0 in the bit line sense amplifier BLSA can be electrically connected to the subsurface bit line (e.g., the bit line BL and the complementary bit line BLB). In addition, the active area AA in the switch layer can be an active area without any transistors therein. As described above, the connection plug in the active area AA (see Fig.15 In addition, in order to reduce the resistance of the switch layer, two or more connecting plugs in the active area AA can be connected to the same subsurface bit line (such as Fig.25 ).

[0242] Next, the present invention discloses a chip structure that realizes subsurface interconnects embedded in a silicon substrate, the subsurface interconnects having high conductivity and effectively optimizing isolation from the silicon substrate. The subsurface interconnects can be vertically connected (bridged) to the source region or drain region of any transistor through a compact self-aligned invention, so many device and circuit design innovations can be created. For example, some subsurface interconnects are created, which can be vertically connected to various source regions and drain regions, respectively, and the other end of the subsurface interconnects is connected to various signal sources, such as ground and power supply VDD. In addition, the chip structure can even introduce supply voltages of different layers or different potentials, where these supply voltages are vertically distributed in the silicon substrate, but necessary isolation is embedded below the horizontal silicon surface (HSS). Therefore, this chip architecture can correspondingly enhance transistor and circuit performance (speed, power consumption, noise, etc.), and reduce more complexity above the horizontal silicon surface (for example, the current complex chip design may need to use the tenth layer of interconnection as the power supply VDD) but requires the following nine layers of interconnection lines above the horizontal silicon surface to transmit signals, wherein these ten layers of interconnection lines are quite complicated to stack and consume a considerable contact area, etc.). In addition, an analogy is made for the present invention: in order to establish different steps from the wires below the horizontal silicon surface to the wires above the horizontal silicon surface, these wires below the horizontal silicon surface can be designed to have different depths in the silicon substrate, thereby allowing the wires to be distributed on the chip to provide various signals, ground voltages or power supply VDD, just like many submarines sailing under the surface of the sea, without colliding with each other in a relatively wide dimension in the silicon substrate.

[0243] In one embodiment of the present invention, Fig.26 As shown, the subsurface interconnects (e.g., subsurface interconnects 2601, 2602, 2603) have a resistance similar to the metal zero layer (Metal 0) used in the process of the wafer fab, but the present invention is not limited to having three subsurface interconnects. That is to say, the present invention can have multiple subsurface interconnects. In this way, the subsurface interconnects can be used as power connections, ground connections, internal node connections, etc., wherein the subsurface interconnects 2601, 2602, 2603 are at different vertical heights, and the subsurface interconnects 2601, 2602 are connected to the active areas AA2, AA1 through the pins 2604, 2605, respectively, and it is worth noting that (but not limited to) the thickness of the subsurface interconnect 2603 is greater than the thickness of the subsurface interconnects 2601, 2602. In addition, each active area in the active area AA2, AA1 can be a P-type active area (PAA) or an N-type active area (NAA).

[0244] In addition, inside the integrated circuit, there are many isolation regions described in the previous figures, such as P type metal-oxide-semiconductor (PMOS) isolation regions, N type metal-oxide-semiconductor (NMOS) isolation regions, and P type metal-oxide-semiconductor and N type metal-oxide-semiconductor isolation regions formed according to the shallow trench isolation pattern. As mentioned above, the underground interconnection line (UGIL) can be formed in the shallow trench isolation. Therefore, Fig.26 As shown, different subsurface interconnects in different layers (subsurface interconnects 2601, 2602, 2603) can be located in the same shallow trench isolation pattern, and different layers can be connected to different nodes through different plugs (plugs 2604, 2605) or vertical interconnects. In addition, in another embodiment of the present invention, the present invention can use these isolation regions (N-type metal oxide semiconductor isolation region, P-type metal oxide semiconductor isolation region and N-type metal oxide semiconductor isolation region) to form different sublayers of the same subsurface interconnect layer or different sublayers of different subsurface interconnect layers, and these sublayers can be connected together or isolated through top metal according to design requirements.

[0245] In addition, each of the different subsurface interconnect layers (subsurface interconnect panels) may include a plurality of sublayers, and the plurality of sublayers may be located in different shallow trench isolations. Fig. 27 As shown, the subsurface interconnection line (that is, the subsurface interconnection layer) 2702 includes sublayers 27021, 27022, and 27023, and the sublayers 27021, 27022, and 27023 are respectively located in shallow trench isolations 2704, 2706, and 2708, wherein the sublayers 27021, 27022, and 27023 can be connected together or isolated by other higher vertical layer metals according to design requirements. Similarly, the subsurface interconnection line 2710 (different from the subsurface interconnection line 2702) includes sublayers 27102 and 27104, and the sublayers 27102 and 27104 are respectively located in shallow trench isolations 2704 and 2708. In another embodiment of the present invention, a vertical spacing VS1 between two adjacent subsurface interconnections (for example, sublayer 27102 and subsurface interconnection line 2712) is necessary.

[0246] In addition, if Fig.28 As shown, different subsurface interconnects within the same shallow trench isolation can be Fig. 27 For example, sublayers 27021, 27102, and 2712 are located Fig. 27There is a relative offset between the two sub-layers of the shallow trench isolation 2704, and the sub-layers 27021, 27102, and 2712. Therefore, the edge of the sub-layer 27021 is vertically and horizontally offset from the edge of the sub-layer 27102.

[0247] Next, different embodiments of applying subsurface interconnects to peripheral circuits or logic circuits of memory will be described (eg Fig.29A , Fig.29B , Fig. 30A , Fig. 30B , Fig.31A , Fig.31B , Fig.31C , Fig.32A , Fig.32B As shown in Fig.29A As shown, a subsurface interconnect 2902 can be used to connect the P+ guard ring (P+ active region), wherein the ground voltage VSS1 can be connected to the P+ guard ring through the subsurface interconnect 2902, so the ground drop can be improved. In addition, the ground voltage VSS1 can be connected to the subsurface interconnect 2902 from the back side of the semiconductor substrate (or transistor) 2904 or other locations on the top of the semiconductor substrate (or transistor) 2904. In addition, multiple guard rings (P+ guard rings) can be connected together through the subsurface interconnect 2902. In addition, as Fig.29B As shown, a subsurface interconnect 2906 can be used to connect the N+ guard ring (N+ active region), wherein the power supply VDD1 can be connected to the N+ guard ring through the subsurface interconnect 2906, so the power supply voltage drop can be improved. In addition, the voltage DD1 can be connected to the subsurface interconnect 2906 from the back side of the semiconductor substrate (or transistor) 2904 or other locations on the top of the semiconductor substrate (or transistor) 2904.

[0248] In addition, if Fig. 30A As shown, the subsurface interconnect 3002 can be used to connect the terminals (e.g., P+ source / drain) of the PMOS transistor to a power source (e.g., power source VDD2 for the PMOS transistor), wherein the power source VDD2 is connected to the terminals (e.g., P+ source / drain) of the PMOS transistor through the subsurface interconnect 3002, so that the performance of the PMOS transistor can be improved. In addition, the power source VDD2 can be connected to the subsurface interconnect 3002 from the back side of the semiconductor substrate 2904 or other locations on the top of the semiconductor substrate 2904. In addition, multiple terminals (P+ source / drain) connected to the power source of different PMOS transistors can be connected together through the subsurface interconnect 3002. In addition, as Fig. 30BAs shown, the subsurface interconnect 3004 can be used to connect the terminal (e.g., N+ source / drain) of the NMOS transistor to the ground (e.g., ground voltage VSS2 for the NMOS transistor), wherein the ground voltage VSS2 is connected to the terminal (e.g., N+ source / drain) of the NMOS transistor through the subsurface interconnect 3004, so the performance of the NMOS transistor can also be improved. In addition, the ground voltage VSS2 can be connected to the subsurface interconnect 3004 from the back side of the semiconductor substrate 2904 or other locations on the top of the semiconductor substrate 2904. In addition, multiple ground terminals (N+ source / drain) of different NMOS transistors can be connected together through the subsurface interconnect 3004.

[0249] In addition, in other embodiments of the present invention, subsurface interconnects may also be used to connect the terminals of multiple transistors, for example:

[0250] (1) Fig.31A As shown, the subsurface interconnection line 3102 connects the node 1 of the PMOS transistor to the node 2 of the NMOS1 transistor, which can reduce the metal routing above the transistor to the power supply or ground.

[0251] (2) Fig.31B As shown, the subsurface interconnection line 3104 connects the node 3 of the NMOS1 transistor to the node 4 of the NMOS2 transistor, which can also reduce the metal routing above the transistor to the power supply or ground.

[0252] (3) Fig.31C As shown, the subsurface interconnection line 3106 connects the node 1 of the PMOS1 transistor to the node 6 of the PMOS2 transistor, which can also reduce the metal routing above the transistor to the power supply or ground.

[0253] In addition, in other embodiments of the present invention, the guard ring may be connected to the power node of the transistor by using a subsurface interconnection line, for example:

[0254] (1) Fig.32A As shown, the subsurface interconnect 3202 is connected to the power supply VDD1 (N+ active region) of the N+ guard ring and is also connected to the required power supply nodes (power supply VDD2, P+ active region) in some terminals of other transistors (e.g., the drain terminal of the PMOS transistor).

[0255] (2) Fig.32B As shown, the subsurface interconnect 3204 is connected to the ground voltage VSS1 (P+ active region) of the P+ guard ring and is also connected to the required ground nodes (ground voltage VSS2, N+ active region) in some terminals of other transistors (e.g., the source terminal of the NMOS transistor).

[0256] Likewise, the voltage source for the guard ring / power node / ground node may be connected to a subsurface interconnect from the back side of the semiconductor substrate (or transistor) 2904 or elsewhere on the top of the semiconductor substrate (or transistor) 2904 .

[0257] In addition, if Fig.29A , Fig.29B , Fig. 30A , Fig. 30B , Fig.31A , Fig.31B , Fig.31C , Fig.32A , Fig.32B As shown, each voltage of the power supplies VDD1 and VDD2 is any high voltage inside the integrated circuit, such as a power supply voltage or a reference voltage, and each voltage of the ground voltages VSS1 and VSS2 is any ground potential inside the integrated circuit, such as a VSS voltage (a type of power supply voltage) or a ground voltage. Fig.29A , Fig.29B , Fig. 30A , Fig. 30B , Fig.31A , Fig.31B , Fig.31C , Fig.32A , Fig.32B As shown, Node1, Node2, Node3, Node4, Node5, Node6, Node_n (n is 1 or 2 or 3...) are internal nodes (signal or power) required by design requirements.

[0258] in addition, Fig.33 3 is a schematic diagram illustrating a conventional layout of a logic circuit, wherein a metal zero layer (Metal0) 3302 is used to connect an N+ guard ring 3304 to some nodes of a terminal (source or drain) of a PMOS transistor, and is used to connect a P+ guard ring 3306 to some nodes (source or drain) of an NMOS transistor. Fig.33 As shown, the metal zero layer 3302 will occupy some layout area, and the layout area occupied by the metal zero layer 3302 cannot be used by other layouts. In addition, the N+ guard ring 3304 is connected to the power supply VDD (not shown) through the metal zero layer 3302. Fig.33 ), and the P+ guard ring 3306 is connected to the ground voltage VSS (not shown) through the metal zero layer 3302. Fig.33 ) and contact.

[0259] On the other hand, Fig.34 As shown, when using subsurface interconnects (such as Fig.34 UGIL#1) is connected to the power supply VDD1 (N+ active region) of the N+ guard ring 3402 and is also connected to the required power supply nodes (corresponding to Fig.32A ),so Fig.33 The layout area occupied by the metal zero layer 3302 can be released (eg Fig.34 The dashed circle shown) is used for other layouts. Fig.35 As shown, when another subsurface interconnect line (such as Fig.35 UGIL#2) is connected to the power supply VDD1 (N+ active region) of the N+ guard ring 3402 and is also connected to the PMOS transistor (corresponding to Fig.34 ), more layout area (dashed rectangle) occupied by metal zero layer 3302 can be released. Therefore, in one embodiment of the present invention, there is no metal layer (e.g., metal zero layer) above the original surface of the semiconductor substrate for connecting the N-type guard ring to the PMOS transistor. In one embodiment of the present invention, the combination of subsurface interconnect line UGIL#1 and subsurface interconnect line UGIL#2 can be regarded as surrounding Fig.35 The subsurface interconnect line (subsurface interconnect plate) of the PMOS transistor in FIG.

[0260] In addition, the subsurface interconnects between the two guard rings (such as Fig.34 UGIL#3) can be used to connect to a guard ring (e.g. Fig.36 In addition, Fig.37 The use of four subsurface interconnects is shown, wherein subsurface interconnects UGIL#1, UGIL#2, and UGIL#3 are respectively Fig.34 , Fig.35 and Fig.36 Mentioned in, and Fig.37 The subsurface interconnect line UGIL#4 in the embodiment (that is, the subsurface interconnect plate surrounding the NMOS transistor) is used to connect to the ground voltage VSS1 of the P+ guard ring (P+ active region) and is also connected to the NMOS transistor (corresponding to Fig.32B ) in some terminals. Therefore, in one embodiment of the present invention, there is no metal layer (such as metal zero layer) on the original surface of the semiconductor substrate for connecting the P-type guard ring to the NMOS transistor.

[0261] Fig.38 is a schematic diagram illustrating the traditional layout of logic circuits, such as Fig.38 As shown, when a node (eg, source node) of the PMOS transistor attempts to connect to the power supply VDD, the source node of the PMOS transistor will pass through the metal zero layer 3804 (eg, Fig.38The diagonal area with contact 3806 shown is connected to the N guard ring (with power supply VDD) 3802. However, due to (1) the source node of the PMOS transistor is also electrically connected to the N well, and (2) the high resistance of the metal zero layer 3804 and the particularly high N well sheet resistance between all PMOS transistors (from point A to point N), a wide metal 1 (Metal 1) layer 3808 (contacting the metal zero layer 3804 through square contacts 3810) is required to minimize the current * resistance (I*R) voltage drop of the power supply VDD about all source nodes of the PMOS transistors, where the metal 1 layer 3808 is also connected to the power supply VDD through other metal layers.

[0262] On the other hand, Fig.39 As shown, when the subsurface interconnect technology or the subsurface power supply VDD plate 3902 is used in the PMOS isolation region, the source node of the PMOS transistor is not only connected to the N guard ring (with power supply VDD) 3802 through the metal zero layer 3804, but also electrically connected to the subsurface power supply VDD plate 3902 through the contact pin 3904, so the N-well sheet resistance from point A to point N is reduced (the sheet resistance of the N-semiconductor is about 1000 ohms / unit area, and the metal sheet resistance of the subsurface interconnect is less than 10 ohms / unit area, so the subsurface interconnect technology improves the resistance problem of the traditional layout by 100 times), and the power supply VDD voltage drop is also the same. Therefore, the narrower metal 1 layer 3808 is acceptable or can even be omitted. As shown in FIG. Fig.39 As shown, this power board based on the subsurface interconnection line technology can also be called an underground interconnection plate (UGIP).

[0263] Likewise, Fig.40 It is also a schematic diagram that illustrates the traditional layout of logic circuits. Fig.40 As shown, when the node (e.g., source node) of the NMOS transistor is connected to the P-guard ring 4002 through the metal zero layer to the ground (ground voltage VSS), a wide metal 1 layer 4006 is required to minimize the ground voltage VSS drop with respect to all source nodes of the NMOS transistors due to (1) the source node of the NMOS transistor is also electrically connected to the P-type substrate (or P-well region) 4003, and (2) the high resistance of the metal zero layer 4004 and the particularly high P-type substrate (or P-well) sheet resistance between all NMOS transistors (from point A to point N), wherein the metal 1 layer 4006 is also connected to the ground voltage VSS through other metal layers. On the other hand, as Fig.41As shown, when the subsurface interconnect technology or the subsurface grounded VSS plate 4102 is used in the N-type metal oxide semiconductor isolation region, the source node of the MMOS transistor is not only connected to the P guard ring (with the ground voltage VSS) 4002 through the metal zero layer 4004, but is also electrically connected to the subsurface grounded VSS plate 4102 through the contact pin 4104, so the P-type substrate (or P-well) sheet resistance from point A to point N is reduced accordingly (wherein the P-type substrate (or P-well) sheet resistance is about 1000 ohms / unit area, and the subsurface interconnect metal sheet resistance is less than 10 ohms / unit area, so the subsurface interconnect technology improves the resistance problem of the traditional layout by 100 times), and the same is true for the ground voltage VSS drop. Therefore, the narrower metal 1 layer 4006 is acceptable or can even be omitted. Fig.41 As shown, this power board based on the subsurface interconnection line technology can also be called a subsurface interconnection board.

[0264] Multiple interconnection lines or interconnection plates of different potentials (one for power supply VDD and another for (ground) voltage VSS) may be used, for example, the voltage of the subsurface interconnection (or subsurface interconnection plate) for power supply VDD is higher than the voltage of the subsurface interconnection (or subsurface interconnection plate) for (ground) voltage VSS. In addition, Fig.42 is similar Fig.40 and Fig.38 A schematic diagram of another conventional layout of a combinational logic circuit. According to the present invention, Fig.43 and Fig.45 a subsurface interconnection layer 1 board 4302 (indicated by solid oblique lines) for power supply VDD, and Fig.44 and Fig.45 The subsurface interconnection layer 2 plate 4402 (indicated by dashed oblique lines) for the ground voltage VSS can be used to improve the sheet resistance problem mentioned in the N-well and P-type substrate 4003 to obtain better connection. In addition, the subsurface interconnection layer 1 plate 4302 for the power supply VDD can be vertically stacked above the subsurface interconnection layer 2 plate 4402 for the ground voltage VSS (such as Fig.43 and Fig.45 ), or vice versa. Both the subsurface interconnection layer 1 plate 4302 and the subsurface interconnection layer 2 plate 4402 can be set in the P-type metal oxide semiconductor isolation region and the N-type metal oxide semiconductor isolation region. However, in another embodiment of the present invention, the subsurface interconnection layer 1 plate 4302 and the subsurface interconnection layer 2 plate 4402 can be set in the P-type metal oxide semiconductor isolation region and the N-type metal oxide semiconductor isolation region, respectively.

[0265] In another conventional layout where the power supply VDD is connected to the source of the PMOS transistor (eg Fig.46As shown in (a), the source of the PMOS transistor is connected to the metal 1 layer (with power supply VDD) 4602 through the metal zero layer 4604, and there are three contacts 4606 between the metal 1 layer 4602 and the metal zero layer 4604. Fig.46 As shown in (a), the equivalent metal zero layer resistance is about 3 squares, and the sheet resistance of the metal zero layer is about 5 ohms / square, so there is about 15 ohms (5ohms / square*3square=~15ohms) between the metal 1 layer 4602 and the metal zero layer 4604. Fig.46 (b)), the bottom of the source (or drain) of the transistor is connected to the N-well (or P-type substrate / P-well), and the power supply VDD (or ground voltage VSS) is passed to the metal 1 layer 4608 and the metal zero layer, wherein the metal zero layer is electrically connected to a position as close to the N-well (or P-type substrate / P-well) as possible. Therefore, the source (or drain) of the transistor can be provided with the power supply VDD (or ground voltage VSS) through the N-well (or P-substrate / P-well). However, due to the high sheet resistance of the N-well (or P-substrate / P-well), the resistance from the source of the transistor to the metal 1 layer 4608 (with the power supply VDD) is about 330 ohms (because Fig.46 As shown in (b), from the nearest power supply VDD to the PMOS transistor, there are R1, R2, and R3 resistors in parallel between the power supply VDD and the source of the PMOS transistor. The sheet resistance of the N well is about 1K ohm / square, and the size of each resistor R1, R2, and R3 is about 1 square, so the resistance of R1, R2, and R3 is about 1000 ohms each; and because the R1, R2, and R3 resistors are connected in parallel, the equivalent resistance Req of the R1, R2, and R3 resistors is about 330 ohms), so the resistance is increased by about 22 times (15 ohms → 330 ohms). Therefore, the traditional source / drain direct power connection technology will seriously damage the performance of the PMOS transistor or NMOS transistor.

[0266] However, if Fig.47As shown, the direct source / drain power connection technology and the subsurface interconnection technology are combined, and the subsurface interconnection board is used, so that when the power supply VDD is passed to the metal 1 layer 4702 and the metal zero layer 4704, the metal 1 layer 4702 and the metal zero layer 4704 are electrically connected to the N-guard ring 4706, and the subsurface interconnection board 4708 is also connected to the N well and the source of the transistor. In this way, the power supply VDD can be provided to the source of the transistor through the subsurface interconnection board 4708. Since the R1, R2, and R3 resistors are in the subsurface interconnection line, the sheet resistance of the R1, R2, and R3 resistors is about 5 ohms / square, and the R1, R2, and R3 resistors are connected in parallel, so the equivalent resistance Req is about 2 ohms, resulting in a resistance reduction of about 166 times (from about 330 ohms to about 2 ohms). Therefore, the combination of the direct source / drain power connection technology and the subsurface interconnection technology can improve the performance of the PMOS transistor. In addition, similar results (>100 times) can also be seen in the source-to-ground voltage VSS connection of NMOS transistors by combining direct source / drain power connection technology with subsurface interconnect technology.

[0267] In summary, the use of subsurface interconnects (or subsurface interconnects) has the following advantages:

[0268] (1) The subsurface interconnect line (or subsurface interconnect plate) can have a resistance similar to that of the metal zero layer, which is about 5 ohms / square, wherein the resistance of the subsurface interconnect line (or subsurface interconnect plate) is much smaller than the N-type active area (NAA) sheet resistance, the P-type active area (PAA) sheet resistance, the N-well sheet resistance, or the P-type substrate sheet resistance (about 500-1000 ohms / square).

[0269] (2) Using subsurface interconnects or subsurface interconnects as power (such as VDD) connections and ground (such as VSS) connections can reduce the current and resistance drop of the power or ground.

[0270] (3) Using subsurface interconnects or subsurface interconnects as internal node connections can reduce metal routing and thus reduce layout area.

[0271] (4) In the above case: Using subsurface interconnect lines or subsurface interconnect plates to connect to the nodes or terminals of the transistor can further reduce metal routing and layout area.

[0272] (5) Multi-layer subsurface interconnects or subsurface interconnects can reduce current*resistance drop to achieve higher efficiency or reduce metal traces.

[0273] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. An integrated circuit, It is characterized by comprising: A semiconductor substrate having an original surface; A P-type metal oxide semiconductor transistor including a gate node, a source node, and a drain node; an N-type guard ring surrounding the P-type metal oxide semiconductor transistor; An N-type metal oxide semiconductor transistor including a gate node, a source node, and a drain node; a P-type guard ring surrounding the N-type metal oxide semiconductor transistor; and a first interconnect layer, located below the original surface of the semiconductor substrate and isolated from the semiconductor substrate; The first interconnect layer is electrically connected to the P-type metal oxide semiconductor transistor, the N-type guard ring, the N-type metal oxide semiconductor transistor or the P-type guard ring.

2. The integrated circuit according to claim 1, characterized in that The first interconnect layer is electrically connected to the N-type guard ring, and a voltage source is connected to the first interconnect layer from the back or top of the P-type metal oxide semiconductor transistor.

3. The integrated circuit as claimed in claim 2, characterized in that The first interconnection layer surrounds the P-type metal oxide semiconductor transistor and is electrically connected to a source node or a drain node of the P-type metal oxide semiconductor transistor.

4. The integrated circuit according to claim 1, characterized in that The first interconnect layer is electrically connected to the P-type guard ring, and a voltage source is connected to the first interconnect layer from the back or top of the N-type metal oxide semiconductor transistor.

5. The integrated circuit as claimed in claim 4, characterized in that The first interconnection layer surrounds the N-type metal oxide semiconductor transistor and is electrically connected to a source node or a drain node of the N-type metal oxide semiconductor transistor.

6. The integrated circuit according to claim 1, characterized in that Also included is another transistor adjacent to the P-type metal oxide semiconductor transistor, wherein the first interconnect layer surrounds and is electrically connected to the P-type metal oxide semiconductor transistor and the another transistor.

7. The integrated circuit as claimed in claim 6, characterized in that The first interconnect layer is electrically connected to a source node or a drain node of the P-type metal oxide semiconductor transistor and is electrically connected to a source node or a drain node of the another transistor.

8. The integrated circuit of claim 1, wherein Also included is another transistor adjacent to the N-type metal oxide semiconductor transistor, wherein the first interconnect layer surrounds and is electrically connected to the N-type metal oxide semiconductor transistor and the another transistor.

9. The integrated circuit as claimed in claim 8, characterized in that The first interconnect layer is electrically connected to a source node or a drain node of the N-type metal oxide semiconductor transistor, and is electrically connected to a source node or a drain node of the another transistor.

10. The integrated circuit of claim 1, wherein There is no metal layer above the original surface of the semiconductor substrate connecting the N-type guard ring to the P-type metal oxide semiconductor transistor.

11. The integrated circuit of claim 1, wherein There is no metal layer above the original surface of the semiconductor substrate connecting the P-type guard ring to the N-type metal oxide semiconductor transistor.

12. An integrated circuit, It is characterized by comprising: A semiconductor substrate having an original surface; a plurality of transistors, each transistor comprising a gate node, a source node, and a drain node; a guard ring surrounding the plurality of transistors; and a subsurface interconnect layer located below the original surface of the semiconductor substrate and isolated from the semiconductor substrate; The subsurface interconnect layer is electrically connected to the plurality of transistors, and a set of metal layers located above the original surface of the semiconductor substrate is electrically connected to the guard ring and the plurality of transistors.

13. The integrated circuit of claim 12, wherein Also included is a first metal layer above the set of metal layers, wherein the first metal layer is electrically connected to the set of metal layers.

14. The integrated circuit of claim 13, wherein The multiple transistors include a group of P-type metal oxide semiconductor transistors, the guard ring includes an N+ guard ring that receives a power supply, and the subsurface interconnect layer is electrically connected to the source nodes of the group of P-type metal oxide semiconductor transistors, wherein the power supply is electrically connected to the source nodes of the group of P-type metal oxide semiconductor transistors through the group of metal layers.

15. The integrated circuit of claim 13, wherein The plurality of transistors include a group of N-type metal oxide semiconductor transistors, the guard ring includes a P+ guard ring that receives a ground voltage, and the subsurface interconnect layer is electrically connected to source nodes of the group of N-type metal oxide semiconductor transistors, wherein the ground voltage is electrically connected to the source nodes of the group of N-type metal oxide semiconductor transistors through the group of metal layers.

16. The integrated circuit of claim 12, wherein The subsurface interconnect layer includes a first subsurface interconnect sublayer and a second subsurface interconnect sublayer stacked on the first subsurface interconnect sublayer; wherein the guard ring comprises a first sub-guard ring receiving a first voltage and a second sub-guard ring receiving a second voltage, and the plurality of transistors comprises a group of first-type metal oxide semiconductor transistors surrounded by the first sub-guard ring and a group of second-type metal oxide semiconductor transistors surrounded by the second sub-guard ring; The first subsurface interconnect sublayer is electrically connected to source nodes of the group of first-type metal oxide semiconductor transistors, and the second subsurface interconnect sublayer is electrically connected to source nodes of the group of second-type metal oxide semiconductor transistors.

17. The integrated circuit of claim 16, wherein The group of metal layers includes a first group of metal sublayers and a second group of metal sublayers, the first voltage is electrically connected to the source node of the group of first-type metal oxide semiconductor transistors through the first group of metal sublayers, and the second voltage is electrically connected to the source node of the group of second-type metal oxide semiconductor transistors through the second group of metal sublayers.

18. An integrated circuit, characterized in that Include: A semiconductor substrate having an original surface; Multiple transistors; a guard ring surrounding the plurality of transistors; a first subsurface interconnect layer, located below the original surface of the semiconductor substrate and isolated from the semiconductor substrate; and a second subsurface interconnect layer, located below the first subsurface interconnect layer and isolated from the semiconductor substrate; The first edge of the first subsurface interconnect layer is vertically and horizontally offset from the second edge of the second subsurface interconnect layer.

19. The integrated circuit of claim 18, wherein The guard ring includes a first sub-guard ring and a second sub-guard ring, and the multiple transistors include a group of P-type metal oxide semiconductor transistors surrounded by the first sub-guard ring and a group of N-type metal oxide semiconductor transistors surrounded by the second sub-guard ring, wherein the first sub-surface interconnect layer electrically connects the group of P-type metal oxide semiconductor transistors, and the second sub-surface interconnect layer electrically connects the group of N-type metal oxide semiconductor transistors.