Memory and its fabrication method

By setting a barrier layer of oxide and nitride stacks in the floating gate embedded memory, the problem of degradation in isolation performance between the control gate and the erase gate is solved, the durability and reliability of the memory are improved, and the gate resistance is reduced.

CN123124357APending Publication Date: 2026-11-17GTA SEMICON CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611291020.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-24
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

In the self-aligned silicide process, the isolation performance between the control gate and the erase gate of floating-gate embedded memory (eflash) degrades, leading to reduced durability and consequently causing data read errors.

Method used

A continuous barrier layer, composed of an oxide layer and a nitride layer, is provided between the control gate and the erase gate. This layer covers the top of the control gate, the top of the isolation dielectric layer, and the top of the inner part of the erase gate, preventing the formation of metal silicides in these areas and avoiding local electric field concentration.

Benefits of technology

It significantly improves the breakdown voltage between the control gate and the erase gate, optimizes the durability and reliability of the memory, and reduces the gate resistance.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123124357A_ABST
    Figure CN123124357A_ABST
Patent Text Reader

Abstract

This application provides a memory and its fabrication method. The memory includes: a substrate with a source region therein, wherein two adjacent memory cells share a source region; a control gate located above the source region, wherein two adjacent memory cells share a control gate; an erase gate located on the surface of the control gate away from the substrate, wherein the erase gates of two adjacent memory cells are symmetrically arranged with respect to the control gate, and an isolation dielectric layer is disposed between the erase gate and the control gate; and a barrier layer comprising sequentially stacked oxide and nitride layers, wherein the barrier layer continuously covers the top surface of the control gate, the top surface of the isolation dielectric layer, and a portion of the top surface of each erase gate facing the control gate, and the opposite ends of the barrier layer terminate at the top surfaces of the two erase gates respectively. The memory of this application significantly weakens the local electric field concentration effect between the control gate and the erase gate, thereby increasing the breakdown voltage between them and improving its durability and reliability.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This application relates to the field of semiconductor device technology, specifically to a memory and its fabrication method. Background Technology

[0002] Floating Gate Embedded Flash (eflash) memory has gained increasing importance in the field of non-volatile memory due to its significant advantages such as low cost, low power consumption, and high-speed access. As semiconductor technology nodes continue to shrink, the market has an urgent need for eflash memory with high density, high performance, and high reliability.

[0003] Endurance is one of the core parameters for measuring the reliability of eflash memory. It characterizes the maximum number of cycles a non-volatile memory can maintain normal function after repeated program / erase operations. Research shows that one of the typical physical mechanisms of endurance failure is the degradation of the isolation performance between the polysilicon gates. This degradation can lead to abnormal programming or erasure operations, which in turn can cause data read errors.

[0004] Floating-gate eflash memory has stringent requirements for electrical isolation between gates. In traditional self-aligned silicide processes, memory arrays typically do not feature a silicide block (SAB) pattern. Instead, low-resistance metal silicides are generated through self-alignment on the source / drain and gate surfaces to reduce gate resistance and thus improve chip operating speed. However, due to the close spacing between the silicides formed between the erase gate (EP) and the control gate (CG), under device operating voltage conditions (maximum voltage difference can reach 12V and above), the high-intensity electric field established between the gates can easily break down the isolation dielectric layer between them, forming a leakage path, ultimately leading to memory durability degradation or even failure. Therefore, effectively improving the reliability of gate-to-gate isolation has become one of the key technical challenges in enhancing the durability of eflash memory.

[0005] It should be noted that the information disclosed in the background section above is only used to enhance the understanding of the background of the present invention, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0006] To address the problems in the prior art, the purpose of this application is to provide a memory and a method for fabricating the same, in which the local electric field concentration effect between the control gate CG and the erase gate EG of the memory is significantly weakened, the breakdown voltage between the two is effectively improved, and thus its durability and reliability are greatly enhanced.

[0007] This application embodiment provides a memory, the memory comprising: A substrate having a source region therein, wherein two adjacent memory cells share one source region; A control gate is located above the source region, and two adjacent memory cells share one control gate; An erase gate is located on the surface of the control gate away from the substrate. The erase gates of two adjacent memory cells are symmetrically arranged with respect to the control gate, and an isolation dielectric layer is provided between the erase gate and the control gate. A barrier layer comprising sequentially stacked oxide and nitride layers, the barrier layer continuously covering the top surface of the control gate, the top surface of each of the isolation dielectric layers, and a portion of the top surface of each of the erase gates facing the control gate to form a barrier layer.

[0008] In some embodiments, any of the following conditions are met: The nitride layer is a silicon nitride layer; The oxide layer is a silicon oxide layer; The thickness ratio of the nitride layer to the oxide layer is between 2 and 3. In some embodiments, any of the following conditions are met: The thickness of the nitride layer is between 100 and 200 angstroms; The thickness of the oxide layer is between 40 and 60 angstroms.

[0009] In some embodiments, the opposite ends of the barrier layer terminate at the top surfaces of the two eraser gates, respectively.

[0010] In some embodiments, in the direction from the control gate to the erase gate, the width of the portion of the top of the erase gate near the control gate that covers the blocking layer is 1 / 5 to 1 / 3 of the width of the top of the erase gate.

[0011] This application also provides a method for fabricating a memory, comprising the following steps: A substrate is provided, wherein a control gate and an erase gate are formed on one side of the substrate, the erase gate is located on the surface of the control gate opposite to the substrate, the erase gates of two adjacent memory cells are symmetrically arranged with respect to the control gate, and an isolation dielectric layer is disposed between the erase gate and the control gate; An oxide layer and a nitride layer are sequentially deposited on the surface of the substrate; The nitride layer and the oxide layer are patterned to form a barrier layer, which continuously covers the top surface of the control gate, the top surface of each of the isolation medium layers, and a portion of the top surface of each of the erase gates facing the control gate, and the opposite ends of the barrier layer terminate at the top surfaces of the two erase gates respectively.

[0012] In some embodiments, the step of patterning the nitride layer and the oxide layer to form a barrier layer includes: A patterned photoresist layer is formed using photolithography. Using the patterned photoresist layer as a mask, the nitride layer is dry etched. Remove the photoresist layer; Using the etched nitride layer as a hard mask, wet etching is used to remove the exposed oxide layer.

[0013] In some embodiments, the preparation method further includes: A metal layer is deposited on the surface of the substrate; The substrate is heat-treated to form a metal silicide on top of the erase gate that is not covered by the barrier layer.

[0014] In some embodiments, the metal layer comprises a nickel-platinum alloy.

[0015] In some embodiments, the heat treatment includes a first heat treatment and a second heat treatment; The first heat treatment is low-temperature rapid thermal annealing, with an annealing temperature of 300℃~550℃; The second heat treatment is high-temperature rapid thermal annealing, with an annealing temperature of 550℃~850℃.

[0016] In some embodiments, the oxide layer is deposited using a chemical vapor deposition process; and / or

[0017] The nitride layer was deposited using a chemical vapor deposition process.

[0018] The memory described in this application effectively prevents the formation of metal silicides on the top of the control gate CG and the top of the inner portion of the erase gate EG during self-aligned silicide processes by setting a barrier layer (composed of stacked oxide and nitride layers) that continuously covers the top of the control gate CG, the top of the isolation dielectric layer, and the top of the inner portion of the erase gate EG. This barrier layer structure allows metal silicides to selectively form on the top of the outer edge of the erase gate and the top of the select gate, effectively reducing the gate resistance in these areas and avoiding the breakdown voltage drop problem caused by silicide bridging or local electric field concentration between the control gate CG and the erase gate EG. Attached Figure Description

[0019] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention. It is obvious that the drawings described below are merely some embodiments of the invention, and those skilled in the art can obtain other drawings based on these drawings without any inventive effort.

[0020] Figure 1 This is a cross-sectional schematic diagram of a memory according to an embodiment of this application; Figure 2 This is a flowchart of a method for fabricating a memory according to an embodiment of this application; Figures 3-7 This is a schematic diagram of the fabrication process of a memory according to an embodiment of this application. Detailed Implementation

[0021] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the embodiments set forth herein; rather, these embodiments are provided so that this application will be thorough and complete, and will fully convey the concept of the exemplary embodiments to those skilled in the art. The same reference numerals in the drawings denote the same or similar structures, and therefore repeated descriptions of them will be omitted. The words “or” and “or” in the specification may mean “and” or “or”. Spatial relation terms such as “…above” are used herein to describe the relationship between one element or feature shown in the figures and other elements or features. It should be understood that, in addition to the orientation shown in the figures, spatial relation terms also include different orientations of the device in use and operation. For example, if the device in the figures is flipped, the element or feature described as “…above” will be oriented “…below” other elements or features. Therefore, the exemplary term “…above” can include both upper and lower orientations. Furthermore, the device may also include other orientations (e.g., rotated 90 degrees or other orientations), and the spatial descriptive terms used herein are interpreted accordingly.

[0022] When used herein, the singular forms of “a,” “an,” and “the” may also include the plural forms, unless the context clearly indicates otherwise. It should also be understood that when the terms “comprise” and / or “comprising” are used in this specification, the presence of the stated feature, integer, step, operation, element, and / or part is established, but the presence or addition of one or more other features, integers, steps, operations, elements, parts, and / or groups is not excluded. Meanwhile, when used herein, the term “and / or” includes any and all combinations of the associated listed items. Although “first” or “second,” etc., are used in this specification to denote certain features, this is merely indicative of function and not a limitation on the number or importance of the specific features.

[0023] The memory and its preparation method of the present invention are further described below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments are not intended to limit the scope of protection of the present invention.

[0024] Figure 1 This is a cross-sectional structural diagram of a memory according to an embodiment of the present invention. The memory is, for example, a floating-gate embedded memory, which includes multiple memory cells. Figure 1 Only a cross-sectional view of two adjacent memory cells is shown. The two adjacent memory cells (Bits) have a symmetrical structure, and each memory cell can be read and written independently. Specifically, the memory includes: Substrate 1, in which an active region CS is formed, two adjacent memory cells share one source region CS; The control gate CG is located above the source region CS, and two adjacent memory cells share one control gate. The material of the control gate CG is preferably polysilicon. The erase gate EG is located on the side of the control gate CG facing away from the substrate 1 (i.e., above it), and the erase gates EG1 and EG2 of two adjacent memory cells are symmetrically arranged with respect to the control gate CG. The top surface of the erase gate EG is higher than the top surface of the control gate CG. This structure facilitates the entry of electrons in the floating gate into the erase gate through the tip FN tunneling effect during the erase operation, thereby improving the erase efficiency.

[0025] Of course, the memory also includes two bit line doped regions BL on either side of the source region CS in substrate 1 (not in... Figure 1 As shown in the diagram, the select gate SG and floating gate FG are located on the surface of substrate 1. The floating gates FG1 and FG2 of two adjacent memory cells are symmetrically arranged relative to the control gate CG. The floating gate FG can be box-shaped or triangular, and its material is polysilicon. The select gate SG is located on the side of the floating gate FG facing away from the control gate CG, as shown in the diagram. Figure 1 As shown, select gate SG1 is located to the left of floating gate FG1, and select gate SG2 is located to the right of floating gate FG2. Select gate SG is used to control the selection and shutdown of the corresponding memory cell. Select gate SG and floating gate FG can perform read and write operations independently. The programming process relies on the channel hot electron injection (CHEI) mechanism to inject electrons into the floating gate; while the erasing process utilizes the Fowler-Nordheim (FN) tunneling effect at the tip of floating gate FG to allow electrons to tunnel to the erase gate EG.

[0026] A tunneling oxide layer (e.g., silicon dioxide layer) is disposed between the floating gate FG and the substrate 1. Simultaneously, isolation dielectric layers are disposed between the select gate SG and the floating gate FG, the control gate CG and the two floating gates (FG1, FG2), and the control gate CG and the two erase gates (EG1, EG2). These isolation dielectric layers can be made of the same or different materials and stack-up structures depending on the functional requirements of their location. For example, they can all adopt an ONO (SiO2 / Si3N4 / SiO2) three-layer structure, or differentiated dielectric layer schemes can be designed for different regions.

[0027] 6. The memory of this application further includes a barrier layer. The barrier layer continuously covers the top surface of the control gate CG, the top surface of the isolation medium layer between the control gate CG and the two erase gates (EG1, EG2), and the portion of the top surface of the erase gates (EG1, EG2) facing the control gate CG. The two ends of the barrier layer terminate at the top surfaces of the two erase gates (EG1, EG2), respectively. That is, the barrier layer does not completely cover the entire top of the erase gate EG, but only covers the top of its inner portion near the control gate CG. In one embodiment, in the direction from the control gate to the erase gate, the width of the portion of the top of the erase gate near the control gate covered by the barrier layer is 1 / 5 to 1 / 3 of the width of the top of the erase gate. For example, when the width of the top of the erase gate is 150 nanometers, the inner portion of the top of the erase gate near the control gate CG is covered by the barrier layer for 30 to 50 nanometers (e.g., 45 nanometers).

[0028] The barrier layer is a metal silicide block layer (SAB), which includes an oxide layer 2' and a nitride layer 3' stacked from bottom to top, wherein the oxide layer 2' is closer to the substrate 1 than the nitride layer 3'. Preferably, the thickness of the nitride layer 3' is greater than the thickness of the oxide layer 2', and the ratio of the thickness of the nitride layer 3' to the thickness of the oxide layer 2' can be between 2 and 3.

[0029] The oxide layer 2' is, for example, a TEOS oxide layer (such as a SiO2 layer) formed by chemical vapor deposition (CVD). The thickness of the oxide layer 2' is preferably 20 Å to 100 Å, more preferably 40 Å to 60 Å, such as 40 Å, 45 Å, 50 Å, etc. The nitride layer 3' is, for example, a silicon nitride (Si3N4) layer, which can be deposited by chemical vapor deposition processes such as plasma-enhanced chemical vapor deposition (PECVD). The thickness of the nitride layer 3' is preferably 50 Å to 500 Å, more preferably 100 Å to 200 Å, such as 100 Å, 120 Å, 150 Å, 180 Å, 200 Å, etc.

[0030] The barrier layer in this application effectively prevents the formation of metal silicides on the top surface of the control gate CG and the top surface of the erase gate EG near the control gate CG during the subsequent self-aligned silicide process for each gate structure (polysilicon gate) to reduce gate resistance. In one embodiment, metal silicide 5 is formed on the top portion of the erase gate EG not covered by the barrier layer, and metal silicide is formed on the top portion of the select gate SG (the portion not covered by the erase gate EG). Since the control gate CG serves as the coupling gate of the memory cell, its function is based on capacitive coupling. During operation, it primarily carries the coupled voltage signal and modulates the potential of the underlying channel region or floating gate FG through an electric field. There is essentially no DC current transport path within its body; only displacement current caused by voltage transients is involved. Therefore, even if the barrier layer covers the top surface of the control gate CG and prevents the formation of silicides on its surface, it will not have a substantial negative impact on the electrical performance and signal coupling function of the control gate CG. The above configuration can significantly reduce the local electric field concentration effect between the control gate and the erase gate, effectively improve the breakdown voltage between them, and thus optimize the durability and reliability of the memory.

[0031] This application also provides a method for manufacturing a memory, used to manufacture a memory as described in Embodiment 1 or a memory having a similar structure. Figure 2 This is a flowchart of a method for fabricating a memory according to an embodiment of this application. The method mainly includes the following steps: Step S100: Provide a substrate on which various gate structures are formed.

[0032] A substrate 1 is provided, on which a control gate CG and an erase gate EG are formed. The substrate also has a floating gate FG and a select gate SG located between the control gate CG and the erase gate EG and the substrate 1. The specific structures of each gate are as described above. Figure 3 As shown.

[0033] Step S200: Deposit oxide layer and nitride layer.

[0034] An oxide layer 2 and a nitride layer 3 are sequentially deposited on the entire surface of a substrate 1 having a control gate, an erase gate, a floating gate, a select gate, and an isolation dielectric layer. Specifically, an oxide layer 2 is first deposited using a chemical vapor deposition (CVD) process, preferably with TEOS as a precursor, with a thickness controlled between 40 and 60 angstroms. Subsequently, a nitride layer 3, made of silicon nitride (SiN), is deposited on the oxide layer 2 using a low-pressure chemical vapor deposition (LPCVD) or plasma-enhanced chemical vapor deposition (PECVD) process, with a thickness controlled between 100 and 200 angstroms. Figure 4 As shown.

[0035] Step S300: Pattern the nitride layer 3 and the oxide layer 2 to form a barrier layer.

[0036] The nitride layer 3 and oxide layer 2 are patterned using photolithography and etching processes to form a barrier layer with a specific pattern. This patterned barrier layer continuously covers the top surface of the control gate CG, the top surface of each isolation dielectric layer, and a portion of the top surface of each erase gate (EG1, EG2) facing the control gate CG. The two ends of the barrier layer terminate at the top surfaces of the two erase gates (EG1, EG2), respectively, while exposing the other side of the top surface of the erase gate EG away from the control gate CG.

[0037] In a more specific embodiment, the patterning in step S300 is performed using the following step-by-step etching method (see...). Figure 2 Sub-steps S310 to S330 in the process): Step S310: Photoresist is spin-coated onto the surface of nitride layer 3 using a photolithography process. The photoresist thickness ranges from 2100 to 2300 angstroms. After exposure and development, a patterned photoresist layer 4 is formed, such as... Figure 5 As shown. The photoresist layer 4 covers the area where the barrier layer needs to be retained (i.e., the top of the control gate, the top of the isolation dielectric layer, and the top of the inner part of the erase gate).

[0038] Step S320: Using the patterned photoresist layer as a mask, a dry etching process (e.g., reactive ion etching (RIE) with a fluorine-containing gas such as CF4 or CHF3 as the etching gas) is used to etch the nitride layer 3. Preferably, a high selectivity (SiN / Oxide selectivity greater than 10:1) etching gas is used to selectively remove the silicon nitride layer in areas not covered by the photoresist until the underlying oxide layer 2 is exposed, thereby transferring the photoresist pattern to the nitride layer and forming the patterned nitride layer 3'. Figure 6 As shown.

[0039] Step S330: Remove the photoresist layer. If the photoresist pattern is transferred to the nitride layer using plasma dry etching in step S320, the residual photoresist layer on the substrate surface can be removed in step S330 by successively using oxygen plasma ashing and wet etching processes.

[0040] Step S340: Using the etched nitride layer 3' as a hard mask, a wet etching process is used to remove the exposed oxide layer 2. Step S340 precisely opens the area where silicide needs to be formed (the gate surface not covered by the barrier layer), while ensuring that the area protected by silicon nitride (the boundary between the control gate CG and the erase gate EP) is completely preserved. Wet etching preferably uses diluted hydrofluoric acid (DHF) or buffered oxide etchant (BOE), which has a high selectivity for oxides and can effectively remove oxides not covered by the nitride layer, while retaining the oxides beneath the nitride layer. Thus, a barrier layer is formed by the remaining oxide layer 2' and the nitride layer 3', as shown below. Figure 7 As shown.

[0041] After forming the barrier layer, the method for fabricating the memory also includes step S400: forming a metal silicide.

[0042] After step S300, which forms the barrier layer, a self-aligned silicide process is performed to reduce the gate resistance. In one embodiment, this may specifically include the following sub-steps: Step S410: Deposit a metal layer on the entire surface of substrate 1 (including the surface of the barrier layer and the exposed polysilicon surfaces such as the erase gate EG and select gate SG not covered by the barrier layer). The metal layer is preferably a nickel-platinum (NiPt) alloy layer, wherein the atomic percentage of Pt is about 5% to 10%, and its thickness can be adjusted according to the target silicide thickness, for example, 50 angstroms to 200 angstroms.

[0043] Step S420: Perform a first heat treatment (low-temperature rapid thermal annealing) on ​​the substrate with the deposited metal layer. The annealing temperature of the first heat treatment is 300°C to 550°C, preferably 350°C to 450°C, and the annealing time is approximately 3 seconds to 120 seconds. Under this low-temperature condition, the metal layer selectively reacts with the exposed polysilicon (i.e., the area such as the top of the erase gate without the barrier layer) to generate nickel-rich metal silicides (such as Ni2Si). Due to the presence of the barrier layer, the top of the control gate CG and the top of the inner side of the erase gate EG are covered, so silicide reaction does not occur.

[0044] Step S430: After the first heat treatment, the unreacted metal layer is removed using a selective wet etching solution (e.g., a sulfuric acid-hydrogen peroxide mixture (SPM) consisting of H2SO4 and H2O2, or an ammonia-hydrogen peroxide mixture (APM, also known as Standard Cleaning Solution 1 SC-1) consisting of NH4OH, H2O2, and H2O). Since the unreacted metal layer exists in pure metal form on the surface of the barrier layer and the isolation dielectric layer, this step can effectively remove it without damaging the already formed metal silicide.

[0045] Step S440: Subsequently, a second heat treatment (high-temperature rapid thermal annealing) is performed. The annealing temperature of the second heat treatment is 550°C to 850°C, preferably 600°C to 750°C, and the annealing time is 3 seconds to 60 seconds. This high-temperature step transforms the low-resistivity nickel-rich phase silicide into a NiSi phase silicide (NiPtSi) with better thermal stability and lower resistivity. Finally, metal silicides are formed only on the top of the erase gate EG (i.e., the outer top away from the control gate CG) and the top of the select gate SG without the barrier layer, effectively reducing the gate resistance in these regions. The top of the control gate CG and the inner top of the erase gate EG are protected by the barrier layer and no silicides are formed, thereby increasing the breakdown voltage between the control gate and the erase gate.

[0046] In one embodiment, the thickness of the erased gate EG is ~600 angstroms, the thickness of the selected gate SG is ~800 angstroms, and the thickness of the metal layer deposited in step S410 is 115 angstroms, then after completing step S440, the thickness of the generated metal silicide is ~250 angstroms.

[0047] It should be noted that in steps S410 to S440 above, a nickel-platinum (NiPt) alloy layer is exemplary used as the metal layer in the self-aligned silicide process. After annealing, a nickel-platinum silicon (NiPtSi) compound with low resistivity and high thermal stability is generated to effectively reduce the gate resistance of the erase gate and select gate. However, in other embodiments, the metal layer is not limited to this, and other suitable metal materials can also be selected, such as cobalt (Co), titanium (Ti), palladium (Pd), platinum (Pt), or nickel-palladium (NiPd) alloys, etc. The resulting cobalt silicide (CoSi2), titanium silicide (TiSi2), palladium-based silicides (such as NiPdSi), etc., all have low resistance.

[0048] In summary, the memory of this application effectively prevents the formation of metal silicides on the top of the control gate CG and the top of the inner part of the erase gate EG during self-aligned silicide processes by setting a barrier layer (composed of stacked oxide and nitride layers) that continuously covers the top of the control gate CG, the top of the isolation dielectric layer, and the top of the inner portion of the erase gate EG. This barrier layer structure allows metal silicides to selectively form on the top of the outer edge of the erase gate and the top of the select gate, effectively reducing the gate resistance in these areas and avoiding the breakdown voltage drop problem caused by silicide bridging or local electric field concentration between the control gate CG and the erase gate EG. Experiments show that the above configuration can significantly weaken the local electric field concentration effect between the control gate and the erase gate, effectively improving the breakdown voltage between them, thereby optimizing the durability and reliability of the memory.

[0049] The above description, in conjunction with specific preferred embodiments, provides a further detailed explanation of this application and should not be construed as limiting the specific implementation of this application to these descriptions. For those skilled in the art, various simple deductions or substitutions can be made without departing from the concept of this application, and all such modifications or substitutions should be considered within the scope of protection of this application.

Claims

1. A memory, characterized in that, The memory includes: A substrate having a source region therein, wherein two adjacent memory cells share one source region; A control gate is located above the source region, and two adjacent memory cells share one control gate; An erase gate is located on the surface of the control gate away from the substrate. The erase gates of two adjacent memory cells are symmetrically arranged with respect to the control gate, and an isolation dielectric layer is provided between the erase gate and the control gate. A barrier layer comprising an oxide layer and a nitride layer stacked sequentially, the barrier layer continuously covering the top surface of the control gate, the top surface of the isolation medium layer, and a portion of the top surface of each erase gate facing the control gate to form a barrier layer, and the opposite ends of the barrier layer terminating at the top surfaces of the two erase gates respectively.

2. The memory according to claim 1, characterized in that, Meet any of the following conditions: The nitride layer is a silicon nitride layer; The oxide layer is a silicon oxide layer; The ratio of the thickness of the nitride layer to the thickness of the oxide layer is between 2 and 3.

3. The memory according to claim 1, characterized in that, Meet any of the following conditions: The thickness of the nitride layer is between 100 and 200 angstroms; The thickness of the oxide layer is between 40 and 60 angstroms.

4. The memory according to claim 1, characterized in that, The portion of the top of the erase gate that is not covered by the barrier layer is formed with metal silicide.

5. The memory according to claim 1, characterized in that, In the direction from the control gate to the erase gate, the width of the portion of the top of the erase gate near the control gate that covers the blocking layer is 1 / 5 to 1 / 3 of the width of the top of the erase gate.

6. A method for fabricating a memory, characterized in that, Includes the following steps: A substrate is provided, wherein a control gate and an erase gate are formed on one side of the substrate, the erase gate is located on the surface of the control gate opposite to the substrate, the erase gates of two adjacent memory cells are symmetrically arranged with respect to the control gate, and an isolation dielectric layer is disposed between the erase gate and the control gate; An oxide layer and a nitride layer are sequentially deposited on the surface of the substrate; The nitride layer and the oxide layer are patterned to form a barrier layer, which continuously covers the top surface of the control gate, the top surface of each of the isolation medium layers, and a portion of the top surface of each of the erase gates facing the control gate, and the opposite ends of the barrier layer terminate at the top surfaces of the two erase gates respectively.

7. The method for fabricating a memory according to claim 6, characterized in that, The step of patterning the nitride layer and the oxide layer to form a barrier layer includes: A patterned photoresist layer is formed using photolithography. Using the patterned photoresist layer as a mask, the nitride layer is dry etched. Remove the photoresist layer; Using the etched nitride layer as a hard mask, wet etching is used to remove the exposed oxide layer.

8. The method for fabricating a memory according to claim 6, characterized in that, The preparation method further includes: A metal layer is deposited on the surface of the substrate; The substrate is heat-treated to form a metal silicide on top of the erase gate that is not covered by the barrier layer.

9. The method for fabricating a memory according to claim 8, characterized in that, The metal layer comprises a nickel-platinum alloy.

10. The method for fabricating a memory according to claim 8, characterized in that, The heat treatment includes a first heat treatment and a second heat treatment; The first heat treatment is low-temperature rapid thermal annealing, with an annealing temperature of 300℃~550℃; The second heat treatment is high-temperature rapid thermal annealing, with an annealing temperature of 550℃~850℃.

11. The method for fabricating a memory according to claim 6, characterized in that, The oxide layer is deposited using a chemical vapor deposition process; and / or The nitride layer was deposited using a chemical vapor deposition process.