A gate stack structure with high aspect ratio trenches

By employing a layered conductive structure and an insulating layer to fill the gaps in the high aspect ratio groove, the problems of insufficient step coverage and stress concentration are solved, improving the uniformity and stability of the conductive layer coverage and meeting the electrical performance and reliability requirements of 3D NAND and automotive-grade devices.

CN224959959UActive Publication Date: 2026-11-10CHONGQING DIDIAN ELECTRONIC TECHNOLOGY CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202522094780.7
Authority / Receiving Office
CN · China
Patent Type
Utility models(China)
Current Assignee / Owner
Filing Date
2025-09-29
Publication Date
2026-11-10
Estimated Expiration
2035-09-29

AI Technical Summary

Technical Problem

Existing technologies suffer from insufficient step coverage, void defects, and stress concentration in high aspect ratio groove structures, which affect the electrical performance and reliability of devices and make it difficult to meet the requirements of 3D NAND and automotive-grade devices.

Method used

The design employs a layered conductive structure and an insulating layer to fill the gaps. By filling the gaps with an insulating layer, combined with multiple conductive layers and a suitable deposition process, the uniformity of conductive layer coverage and stress dispersion are improved, avoiding the defects of traditional single polycrystalline silicon layers.

Benefits of technology

It improves the step coverage and stability of the conductive layer, reduces the series resistance of the conductive layer, and enhances the electrical performance and reliability of semiconductor devices. It is suitable for field-effect transistors and dynamic random access memories with high aspect ratio trenches.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN224959959U_ABST
    Figure CN224959959U_ABST
Patent Text Reader

Abstract

This utility model relates to a gate stack structure with a high aspect ratio groove, including a substrate, a patterned layer, a first conductive layer, a second conductive layer, and a barrier layer. The patterned layer is disposed on the substrate and has a groove that exposes the substrate. The first conductive layer is disposed within the groove. A gate dielectric layer is disposed between the first conductive layer and the substrate contact surface. The second conductive layer is disposed on the first conductive layer. A barrier layer is disposed between the first conductive layer and the second conductive layer. Gaps exist between the first conductive layer, the barrier layer, the gate dielectric layer, the second conductive layer, and the inner wall of the groove. An insulating layer is disposed on the patterned layer and partially fills the gaps. The gaps formed between the patterned layer and the groove are filled with an insulating layer to increase the stability of the first and second conductive layers.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention belongs to the field of gate structure manufacturing technology for semiconductor devices, specifically relating to a gate stack structure with a high aspect ratio groove, suitable for the gate electrode of a field-effect transistor (MOSFET) or dynamic random access memory (DRAM). Background Technology

[0002] In the semiconductor device manufacturing field, high aspect ratio trench structures (depth-to-width ratio > 10:1) are widely used in key components such as the control gate of 3D NAND flash memory, the word line trench of dynamic random access memory (DRAM), and the vertical gate of power MOSFETs. For example, the trench aspect ratio of 3D NAND devices has reached 30:1, and the trench aspect ratio of automotive-grade power devices has reached 20:1. These structures place stringent requirements on the uniformity of the conductive layer's step coverage and structural stability, and their performance directly determines the device's electrical parameters (such as gate capacitance and leakage current) and long-term reliability.

[0003] In existing technologies, the conductive layer within the groove typically employs a single polycrystalline silicon layer structure, formed using a low-pressure chemical vapor deposition (LPCVD) process. However, this structure suffers from the following inherent drawbacks in applications with an aspect ratio > 10:1: 1. Insufficient step coverage and void defects In trenches with an aspect ratio greater than 10:1, the polysilicon layer exhibits a significant thickness difference between the bottom and sidewalls due to the "bottom-mouth effect" during LPCVD deposition (the deposition rate at the trench opening is faster than at the bottom, causing the opening to be blocked first). Step coverage is generally less than 70%. Simultaneously, the center of the trench is prone to voids due to insufficient deposition, with a void ratio as high as 25%. This defect directly causes a 15-20% decrease in gate capacitance (the capacitance value deviates from the design target) and an increase in leakage current to 10⁻⁻⁶. 7 A / μm² or higher (far exceeding the leakage current control standard of 10⁻ for conventional devices) 9 (A / μm²), which seriously affects the switching performance and power consumption control of the device.

[0004] 2. Stress concentration and reliability risks To meet conductivity requirements, traditional single polysilicon layers need to be thicker than 200nm. However, the thermal expansion coefficients of thick polysilicon layers differ significantly from those of the groove wall materials (such as silicon oxide and silicon nitride). During temperature cycling (-55℃ to 125℃, covering the operating temperature range of consumer and automotive-grade devices), tensile stresses exceeding 800MPa are easily generated inside the polysilicon layer. Long-term accumulation of this stress can cause cracks in the polysilicon layer, increasing the device failure probability by 30%, failing to meet the reliability requirements of "10-year lifespan" for automotive-grade devices or "millions of write / erase cycles" for 3D NAND devices.

[0005] To address the above problems, existing technologies mainly employ two types of improvement schemes, but both have significant limitations: (1) All-metal layer alternative Using metals such as tungsten and titanium as conductive layers within the grooves can improve the step coverage to over 95% through atomic layer deposition (ALD) or physical vapor deposition (PVD) processes, alleviating the problem of insufficient coverage. However, there are two major drawbacks: First, metal atoms tend to diffuse into the substrate or gate dielectric layer during high-temperature processes (such as subsequent annealing and doping activation), causing the device threshold voltage to drift (the drift can reach 0.2-0.5V), affecting the stability of the device's electrical performance. Second, the cost of metal materials is more than three times that of polycrystalline silicon, and dedicated deposition equipment is required, significantly increasing mass production costs and making it difficult to apply to the large-scale production of low- to mid-range logic devices or high-capacity memory devices.

[0006] (2) Optimization scheme for process parameters Attempts were made to improve the step coverage of polysilicon layers by adjusting the deposition temperature (e.g., decreasing from 580℃ to 520℃), deposition pressure (e.g., increasing from 200 mTorr to 300 mTorr), or gas flow rate ratio in LPCVD. However, in practical applications, this approach only improved the step coverage of trenches with an aspect ratio of 20:1 to 85%, still failing to completely eliminate bottom voids. Furthermore, parameter adjustments had a negligible effect on trenches with an aspect ratio greater than 25:1, and may even lead to a decrease in the doping activation efficiency of the polysilicon layer (a 10-15% reduction in doping activation rate), thus increasing the series resistance of the conductive layer. More importantly, this approach did not address the stress concentration problem of thick polysilicon layers at the structural level, and the device reliability risk remained unmitigated.

[0007] Therefore, there is an urgent need for a semiconductor gate stacking structure that can simultaneously improve the uniformity of conductive layer coverage in high aspect ratio (10:1-30:1) grooves, alleviate stress concentration, and ensure cost control and stable electrical performance through structural improvements rather than simple process optimization, so as to meet the mass production requirements of 3D NAND, automotive-grade power devices and high-end logic devices. Utility Model Content

[0008] To address the aforementioned problems in the prior art, this invention provides a gate stacking structure with a high aspect ratio trench that can improve upon these shortcomings. This structure fills the gaps with an insulating layer, solving the problems of narrow slits and insufficient stability that easily occur when polysilicon is deposited in trenches with an aspect ratio > 10:1, thereby improving the electrical performance and reliability of semiconductor devices.

[0009] This utility model relates to a gate stack structure with a high aspect ratio groove, comprising a substrate, a patterned layer, a first conductive layer, a second conductive layer, and a barrier layer. The patterned layer is disposed on the substrate and has a groove that exposes the substrate. The first conductive layer is disposed within the groove, and a gate dielectric layer is disposed between the first conductive layer and the substrate contact surface. The second conductive layer is disposed on the first conductive layer, and a barrier layer is disposed between the first and second conductive layers. Gaps exist between the first conductive layer, the barrier layer, the gate dielectric layer, the second conductive layer, and the inner wall of the groove. An insulating layer is disposed on the patterned layer and partially fills the gaps. The gate stack structure is a field-effect transistor gate stack structure. The aspect ratio of the groove is 10:1 to 30:1 (the original "5:1" has been removed to match the core application scenario and technical problem of "high aspect ratio > 10:1" in the background art). The depth of the insulating layer filling the gaps is 1 / 3 of the groove depth. The filling depth of the insulating layer should be adjusted according to the groove depth-to-width ratio: when the groove depth-to-width ratio is 10:1-20:1, the filling depth should be 1 / 3 of the groove depth. This ensures stress dispersion while reserving sufficient space for the conductive layer to avoid the electrical performance being affected by compression. When the groove depth-to-width ratio is 20:1-30:1, the filling depth should be 1 / 2 of the groove depth. This increases the amount of insulating layer filling, enhances the support stability of the conductive layer in the high depth-to-width ratio groove, and further suppresses stress concentration at the gaps.

[0010] Furthermore, the insulating layer is silicon oxide or silicon nitride; if silicon nitride is chosen, its dielectric constant (ε≈7.5) is higher than that of silicon oxide, which can reduce capacitance loss after gap filling and is more suitable for scenarios with high capacitor performance requirements (such as DRAM word line trenches); if silicon oxide is chosen, its process compatibility is stronger and it is more suitable for integration scenarios with structures where the pattern layer is silicon oxide.

[0011] Furthermore, the first conductive layer includes a first polysilicon layer, a first metal layer, and a second polysilicon layer sequentially disposed on the gate dielectric layer; wherein, the thickness of the first polysilicon layer and the second polysilicon layer is 50-100nm, and the "bottleneck effect" in the deposition process is reduced by splitting the traditional single thick polysilicon layer (thickness > 200nm); the thickness of the first metal layer needs to be adjusted in conjunction with the trench depth-to-width ratio: when the trench depth-to-width ratio is 10:1-20:1, the thickness of the first metal layer is 3-5nm; when the trench depth-to-width ratio is 20:1-30:1, the thickness of the first metal layer is 5-8nm, and by thickening the metal layer, the step coverage and connection capability in high depth-to-width ratio scenarios is improved, ensuring the electrical conduction continuity of the upper and lower polysilicon layers.

[0012] Furthermore, the first conductive layer includes a third polysilicon layer, a second metal layer, a first polysilicon layer, a first metal layer, and a second polysilicon layer disposed sequentially on the gate dielectric layer; wherein, the thickness of the third polysilicon layer is 30-50nm, and the thickness of the second metal layer is consistent with the thickness of the first metal layer (adaptation logic is the same as the first metal layer mentioned above). By increasing the number of polysilicon layers and the number of metal layers, the step coverage uniformity in extreme scenarios with aspect ratios of 20:1-30:1 (such as the control gate of 3D NAND flash memory) is further optimized; and the doping type and concentration of the third polysilicon layer, the first polysilicon layer, and the second polysilicon layer are the same, avoiding uneven conductivity caused by doping differences and ensuring the overall electrical performance stability of the conductive layer.

[0013] Furthermore, the material of the first metal layer includes titanium nitride (TiN) or tantalum nitride (TaN), with a thickness of 3-8 nm; among them, titanium nitride (TiN) has both excellent conductivity and anti-diffusion properties, making it more suitable for logic device gate scenarios; tantalum nitride (TaN) has better high-temperature resistance, making it more suitable for high-temperature operating scenarios such as automotive-grade power devices.

[0014] Furthermore, the substrate material is at least one of silicon, silicon germanium, and silicon-on-insulator; if the substrate is an N-type doped silicon substrate, the doping concentration is 1×10¹. 5 -1×10¹ 7 The density of atoms / cm³ is suitable for PMOSFET gates in logic devices. If the substrate is germanium-silicon, its carrier mobility is even higher, making it suitable for high-frequency devices. If the substrate is silicon-on-insulator (SOI), it can further reduce parasitic capacitance, making it suitable for scenarios with high power consumption requirements. This structure is particularly suitable for high aspect ratio trenches with a depth-to-width ratio of 20:1 to 30:1 (consistent with the upper limit of the depth-to-width ratio of trenches with a depth-to-width ratio of 10:1 to 30:1), such as the control gate trenches of 3D NAND flash memory or the gate structure of vertical field-effect transistors (VFETs). The multi-layer alternating stacking design solves the problem of uneven deposition of traditional polysilicon in deep and narrow trenches.

[0015] Furthermore, the gap width between the first conductive layer and the inner wall of the groove is 0.1-0.5 μm. This width range ensures that the insulating layer is fully filled by conventional processes, avoiding insufficient filling due to an overly narrow gap or insufficient effective area of ​​the conductive layer due to an overly wide gap. The insulating layer is filled by plasma-enhanced chemical vapor deposition (PECVD), and specific process parameters need to be matched according to the insulating layer material: if the insulating layer is silicon nitride, the process parameters are a volume ratio of SiH4 to NH3 of 1:3-1:5, a deposition temperature of 300-350℃, and a deposition pressure of 200-300 mTorr; if the insulating layer is silicon oxide, the process parameters are a volume ratio of SiH4 to O2 of 1:4-1:6, a deposition temperature of 250-300℃, and a deposition pressure of 150-250 mTorr, to ensure the compactness of the insulating layer and the gap filling effect, and to avoid the generation of bubbles or voids after filling.

[0016] The gate stack structure of this invention includes a substrate, a patterned layer, a first conductive layer, a second conductive layer, and a barrier layer. The patterned layer is disposed on the substrate and has a groove that exposes the substrate. The first conductive layer is disposed within the groove. The first conductive layer, the barrier layer, and the second conductive layer are disposed within the groove. The gap formed between them and the groove is filled with an insulating layer. Through the dual design of "layered conductive structure + insulating layer filling to disperse stress", this invention solves the problems of insufficient step coverage and void defects in traditional single thick-layer polycrystalline silicon, alleviates the problem of thermal stress concentration, and avoids the high cost and threshold voltage drift risk of all-metal layer solutions. Ultimately, it improves the stability of the first and second conductive layers and optimizes the electrical performance and reliability of semiconductor devices. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of the structure of this utility model.

[0018] Figure 2 This is a schematic diagram of the first conductive layer structure in Example 1.

[0019] Figure 3 This is a schematic diagram of the first conductive layer structure in Example 2.

[0020] The diagram shows: 1. Substrate, 2. Pattern layer, 3. Groove, 4. Insulating layer, 5. Gate dielectric layer, 6. First conductive layer, 7. Barrier layer, 8. Second conductive layer, 61. Gate dielectric layer, 62. First polysilicon side layer, 63. First metal layer, 64. Second polysilicon layer, 65. Third polysilicon layer, 66. Second metal layer. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of this utility model clearer, the present utility model will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present utility model and are not intended to limit the scope of the present utility model.

[0022] like Figure 1 and 2 As shown, this utility model relates to a gate stack structure with a high aspect ratio groove, including a substrate 1, a patterned layer 2, a first conductive layer 6, a second conductive layer 8, and a barrier layer 7. The patterned layer is disposed on the substrate and has a groove 3 that exposes the substrate. The first conductive layer is disposed within the groove, and a gate dielectric layer 61 is disposed between the first conductive layer and the substrate contact surface. The gate dielectric layer is formed using atomic layer deposition (ALD) with process parameters of deposition temperature 200-250℃ and deposition pressure 50-100mTorr to ensure uniformity of gate dielectric layer thickness (deviation <5%). The second conductive layer is disposed on the first conductive layer, and a barrier layer is disposed between the first and second conductive layers. A capping layer is disposed on the second conductive layer. The capping layer is an insulating capping layer (material same as the insulating layer, deposited using PECVD process, thickness 100-150nm). Gaps exist between the first conductive layer, the barrier layer, the gate dielectric layer, and the second conductive layer and the inner wall of the groove. An insulating layer 4 is disposed on the patterned layer and partially fills the gaps.

[0023] As a technological improvement, the insulating layer is silicon oxide or silicon nitride. Silicon nitride (Si3N4) prepared by plasma-enhanced chemical vapor deposition (PECVD) has a higher dielectric constant (ε≈7.5) than silicon oxide, which can reduce capacitance loss after gap filling. The specific process of PECVD deposition of silicon nitride is as follows: SiH4 flow rate 20-30 sccm, NH3 flow rate 60-150 sccm (volume ratio 1:3-1:5), deposition temperature 300-350℃, deposition pressure 200-300 mTorr, and annealing at 200-250℃ for 30-60 min after deposition to improve the density of silicon nitride.

[0024] As a technical improvement, the first conductive layer includes a first polysilicon layer 62, a first metal layer 63, and a second polysilicon layer 64 sequentially disposed on the gate dielectric layer; wherein, the first polysilicon layer (thickness 50-100nm) is formed by low-pressure chemical vapor deposition (LPCVD) with process parameters of deposition temperature 550-600℃, deposition pressure 100-200mTorr, and phosphorus ion doping (concentration 5×10¹). 9atoms / cm³, doping method is ion implantation (implantation energy 50-80keV, implantation dose 1×10¹). 5 The first metal layer is titanium nitride (TiN), with a thickness of 3-8 nm (3-5 nm for a groove depth-to-width ratio of 10:1-20:1, and 5-8 nm for 20:1-30:1). It is produced by atomic layer deposition (ALD) at a deposition temperature of 250 °C. The precursors are titanium tetrachloride (TiCl4) and ammonia (NH3), which are alternately introduced 50-100 times to control the thickness. The second polycrystalline silicon layer (50-100 nm thick) is produced by the same LPCVD, doping, and annealing process as the first polycrystalline silicon layer, forming a three-layer stacked structure of "polycrystalline silicon-metal-polycrystalline silicon" to reduce the series resistance of the electrodes.

[0025] As a technical improvement, the material of the first metal layer includes titanium nitride or tantalum nitride; the first polycrystalline silicon layer and the second polycrystalline silicon layer are doped with P-type ions or N-type ions (if P-type ions are used, boron ions are selected, with a concentration of 3×10¹). 9 atoms / cm³, ion implantation energy 40-60keV, implantation dose 8×10¹ 4 atoms / cm², annealing process is the same as phosphorus doping).

[0026] As a technical improvement, the substrate material is at least one of silicon, silicon germanium, and silicon-on-insulator, and an N-type doped silicon substrate (doping concentration 1×10¹) is used. 5 -1×10¹ 7 (atoms / cm³), suitable for PMOSFET gates in logic devices; if a silicon-on-insulator (SOI) substrate is used, parasitic capacitance can be further reduced. The patterning layer material in this solution includes silicon oxide or silicon nitride (silicon oxide is deposited using LPCVD process at a temperature of 400-450℃; silicon nitride is deposited using PECVD process with the same parameters as the insulating layer silicon nitride). At the same time, such as Figure 1 and 3As shown, Embodiment Two: This utility model relates to a gate stack structure with a high aspect ratio groove, including a substrate 1, a patterned layer 2, a first conductive layer 6, a second conductive layer 8, and a barrier layer 7. The patterned layer is disposed on the substrate, and the patterned layer has a groove 3 that exposes the substrate. The first conductive layer is disposed within the groove, and a gate dielectric layer 61 is disposed between the first conductive layer and the substrate contact surface. The gate dielectric layer is formed using atomic layer deposition (ALD) with the same process parameters as Embodiment One (deposition temperature 200-250℃, deposition pressure 50-100mTorr). The second conductive layer is disposed on the first conductive layer, and a barrier layer is disposed between the first conductive layer and the second conductive layer. A capping layer is disposed on the second conductive layer. The capping layer is an insulating capping layer (material is the same as the insulating layer, deposited using PECVD process, thickness 100-150nm). There are gaps between the first conductive layer, the barrier layer, the gate dielectric layer, the second conductive layer, and the inner wall of the groove. An insulating layer 4 is disposed on the patterned layer, and the insulating layer also partially fills the gaps.

[0027] As a technical improvement, the insulating layer is silicon oxide or silicon nitride. The silicon nitride (Si3N4) is prepared by plasma-enhanced chemical vapor deposition (PECVD) with the same process parameters as in Example 1 (SiH4 to NH3 volume ratio 1:3-1:5, deposition temperature 300-350℃, deposition pressure 200-300mTorr). After deposition, annealing is performed to improve density.

[0028] As a technical improvement, the first conductive layer includes a third polysilicon layer 65, a second metal layer 66, a first polysilicon layer 62, a first metal layer 63, and a second polysilicon layer 64 sequentially disposed on the gate dielectric layer; wherein, the thickness of the third polysilicon layer 65 is 30-50 nm, deposited by LPCVD process (temperature 550-600℃, pressure 100-200 mTorr), and doped with phosphorus ions (concentration 5×10¹). 9 atoms / cm³, ion implantation energy 40-60keV, dose 1×10¹ 5The first polysilicon layer 66 has a thickness of 3-5 nm (suitable for trench aspect ratios of 20:1-30:1) and is made of the same material as the first metal layer 63 (titanium nitride or tantalum nitride). It is deposited using an ALD process (parameters are the same as the first metal layer). The process parameters for the first polysilicon layer 62 (50-80 nm), the first metal layer 63 (3-5 nm), and the second polysilicon layer 64 (50-80 nm) are the same as those for the third polysilicon layer and the second metal layer, respectively. By splitting the polysilicon layer into thin layers and inserting metal layers, the single deposition thickness requirement is reduced, the step coverage in trenches with aspect ratios > 20:1 is improved, and the gap defects of traditional thick polysilicon deposition are avoided.

[0029] As a technological improvement, the material of the first metal layer includes titanium nitride or tantalum nitride, and the second metal layer has the same material and thickness as the first metal layer.

[0030] The first, second, and third polysilicon layers are doped with P-type or N-type ions (boron ions are used for P-type doping, and the process parameters are the same as in Example 1). This example is suitable for trenches with an aspect ratio of 20:1-30:1 (such as the control gate of 3D NAND flash memory), and the step coverage of the conductive layer is further improved by increasing the number of metal layers.

[0031] As a technical improvement, the substrate material is at least one of silicon, silicon germanium, and silicon-on-insulator, and an N-type doped silicon substrate (doping concentration 1×10¹) is used. 5 -1×10¹ 7 (atoms / cm³), suitable for PMOSFET gates in logic devices; if a silicon-on-insulator (SOI) substrate is used, parasitic capacitance can be further reduced. The patterning layer material in this solution includes silicon oxide or silicon nitride (process parameters are the same as in Example 1).

[0032] Performance tests were conducted on the structure of this embodiment, and the results showed that in the groove with an aspect ratio of 25:1, the step coverage of the polysilicon layer increased from 85% in Embodiment 1 to 98%, and the series resistance of the conductive layer decreased by 18%, meeting the electrical performance requirements of the gate electrode for 3D NAND flash memory.

Claims

1. A gate stack structure having a high aspect ratio recess, characterized by: The gate stack structure comprises a substrate, a pattern layer, a first conductive layer, a second conductive layer and a barrier layer, the pattern layer is arranged on the substrate, the pattern layer has a groove, the groove exposes the substrate, the first conductive layer is arranged in the groove, a gate dielectric layer is arranged between the first conductive layer and the contact surface of the substrate, the second conductive layer is arranged on the first conductive layer, the barrier layer is arranged between the first conductive layer and the second conductive layer, a gap exists between the first conductive layer, the barrier layer, the gate dielectric layer, the second conductive layer and the inner wall of the groove, an insulating layer is arranged on the pattern layer, and the insulating layer also partially fills in the gap; the gate stack structure is a gate stack structure of a field effect transistor, the depth-width ratio of the groove is 10:1 to 30:1, and the filling depth of the insulating layer in the gap is 1 / 3 to 1 / 2 of the depth of the groove; wherein the selection of the filling depth of the insulating layer needs to be combined with the depth-width ratio of the groove: when the depth-width ratio of the groove is 10:1-20:1, the filling depth is 1 / 3 of the depth of the groove, so as to balance the stress dispersion and the space requirement of the conductive layer; when the depth-width ratio of the groove is 20:1-30:1, the filling depth is 1 / 2 of the depth of the groove, so as to enhance the support stability of the conductive layer in the high-depth-width-ratio groove.

2. The gate stack structure with high aspect ratio recesses of claim 1, wherein: The insulating layer is silicon oxide or silicon nitride.

3. The gate stack structure with high aspect ratio recess according to claim 1, wherein: The first conductive layer comprises a first polysilicon layer, a first metal layer and a second polysilicon layer arranged in sequence on the gate dielectric layer; wherein the thicknesses of the first polysilicon layer and the second polysilicon layer are both 50-100 nm, and the thickness of the first metal layer needs to be adapted to the depth-width ratio of the groove: when the depth-width ratio of the groove is 10:1-20:1, the thickness of the first metal layer is 3-5 nm; when the depth-width ratio of the groove is 20:1-30:1, the thickness of the first metal layer is 5-8 nm, so as to improve the step coverage connection ability in the high-depth-width-ratio scenario by thickening the metal layer.

4. The gate stack structure with high aspect ratio recess according to claim 1, wherein: The first conductive layer comprises a third polysilicon layer, a second metal layer, a first polysilicon layer, a first metal layer and a second polysilicon layer arranged in sequence on the gate dielectric layer; wherein the thickness of the third polysilicon layer is 30-50 nm, the thickness of the second metal layer is consistent with the thickness of the first metal layer, and the doping type and concentration of the third polysilicon layer, the first polysilicon layer and the second polysilicon layer are the same, so as to ensure the uniformity of the overall electrical performance of the conductive layer.

5. The gate stack structure with high aspect ratio recesses of claim 1, wherein: The gap width between the first conductive layer and the inner wall of the groove is 0.1-0.5 μm, and the insulating layer is filled by a plasma-enhanced chemical vapor deposition process.