Improved Monte Carlo finite element model three-dimensional seepage framework simulation method based on multi-physics combinatorial dynamics

Through the Monte Carlo finite element model of multiphysical combination kinetics, the defect formation and electric field distribution of hafnium oxide grain boundary is simulated, and the problem of insufficient understanding of hafnium oxide grain boundary characteristics in the prior art is solved, and the multi-crystal hafnium oxide memory resistance is realized Precise simulation and performance improvement of switching mechanisms.

CN118965902APending Publication Date: 2024-11-15DALIAN UNIV OF TECH
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411130946.X
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-16
Publication Date
2024-11-15

AI Technical Summary

Technical Problem

The existing technology is difficult to effectively characterize and understand the structure and electronic characteristics of hafnium oxide grain boundaries, resulting in a lack of a clear theoretical understanding of the resistance switching mechanism of polycrystalline hafnium oxide resistive memory, limiting its large-scale production and performance improvement.

Method used

Using the three-dimensional seepage frame simulation method based on the Monte Carlo finite element model based on multi-physical combined dynamics, the probability vector control defect formation is generated by the Monte Carlo algorithm, combined with finite element calculation, simulate the conduction path and electric field distribution in the resistor switching process, and accurately The potential and electric field modes of the switching stages of high-resistance state to low-resistance state are analyzed.

Benefits of technology

This method can accurately simulate the formation of conductive paths in amorphous and polycrystalline hafnium oxide, revealing the key role of hafnium oxide grain boundary in the resistance switching process, improving device performance and reliability, and providing the preparation of high-performance polycrystalline oxidation The theoretical basis of hafnium memory.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118965902A_ABST
    Figure CN118965902A_ABST
Patent Text Reader

Abstract

The invention belongs to the technical field of hafnium oxide resistive random access memories, and relates to an improved Monte Carlo finite element model three-dimensional seepage framework simulation method based on multi-physics combinatorial dynamics, which comprises the following steps: (1) forming new defects and updating parameters, (2) checking the formation of clusters and conductive filaments, (3) calculating the distribution of potential and electric field, and (4) simulating the three-dimensional seepage framework of the Monte Carlo finite element model based on multi-physics combinatorial dynamics. And (4) judging whether a low-resistance conductive path is formed or not. The model is also used for searching statistical distribution formed by conductive paths in amorphous and polycrystalline hafnium oxide media, and can accurately simulate an observation result of a resistance switch in a titanium nitride / hafnium oxide / platinum metal-insulator-metal stack. Monte Carlo and finite element coupled multi-physical simulation is utilized to explain formation of conductive paths and electric field distribution in a resistance switching process in a metal-insulator-metal stack based on hafnium oxide. Device-level simulation shows that in a polycrystalline hafnium oxide-based resistive random access memory device, a conductive path of a hafnium oxide grain boundary site is formed faster than in a grain.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to an improved three-dimensional percolation framework simulation method based on a multi-physics combined dynamics Monte Carlo finite element model, and belongs to the technical field of hafnium oxide resistive random access memory. Background Art

[0002] The rapid development of smart technology has led to a sharp rise in the demand for data storage, which has sparked interest in hafnium oxide-based resistive random access memory, which is considered one of the most promising candidates for the next generation of non-volatile memory. RRAM operates by controlling the switching resistance on a high-K thin hafnium oxide dielectric layer, and controlling the thermodynamics and kinetics of the generation of oxygen vacancies and the formation of conductive paths in a dielectric-based metal-insulator-metal stack. The technology demonstrates impressive properties such as high non-volatile storage, fast read and write speeds (<10ns), low power consumption, and high compatibility with complementary metal oxide semiconductor technology. Its typical metal-insulator-metal sandwich structure effectively solves the charge leakage problem commonly seen in traditional non-volatile flash memories. This special property is attributed to the wide bandgap (5.7eV), high dielectric constant, and strong thermodynamic stability of hafnium oxide. Despite these advances, the large-scale production of RRAM remains a challenge due to the lack of a clear theoretical understanding of its operating mechanism. The currently widely accepted physical mechanism is the microscopic conductive path of oxygen vacancies, which explains the resistive switching phenomenon of the formation and rupture of conductive filaments composed of oxygen vacancies in a thin oxide layer under current pulse stress. G. Bersuker pointed out in the literature 'Metal oxide resistive memory switching mechanism based on conductive filament properties' [J. Appl. Phys. 110 (2011), 124518] that conductive filaments tend to form preferentially at grain boundaries. This process begins with the accumulation of oxygen vacancies near the grain boundaries, and then new vacancies are generated in the surrounding areas. The high density of electrically active defects at the grain boundaries in polycrystalline hafnium oxide is conducive to the formation of conductive paths. Therefore, grain boundaries play an important role in the resistive switching performance of polycrystalline hafnium oxide-based metal-insulator-metal stacks.

[0003] Grain boundaries have attracted much attention due to their important roles in the mechanical and electrical properties of polycrystalline materials. For example, they can act as sinks for atomic defects and impurities, which can drive structural transformations of materials and thus change their properties. Understanding how to characterize the structural and electronic properties of grain boundaries is crucial to understanding and controlling the properties of polycrystalline materials, ultimately providing insights into improving the performance and reliability of devices. M. Lanza and K. Zhang pointed out in the literature 'Grain boundaries as referential sites for resistive switching in the HfO2 resistive random access memory structures', [Appl. Phys. Lett. 100 (2012) 123508] that the existence of conductive paths in polycrystalline hafnium oxide films is of outstanding importance. Experimental measurements using atomic force microscopy have demonstrated that the conductive paths that can be turned on and off through the electronic conductance of hafnium oxide originate from the grain boundary region and then extend to adjacent grains. Although it is generally believed that grain boundaries are the preferred leakage paths in polycrystalline hafnium oxide, research on the conduction mechanism of hafnium oxide grain boundaries is very limited. Characterizing the electronic properties of grain boundaries is a challenging task: Wang Yao et al. in patent [Publication No.: CN111676509A] effectively controlled the preparation of symmetrical (or asymmetrical) side or twisted grain boundaries by cutting seed crystal pairs with specific orientations and angles. Yu Xuegong et al. in patent [Publication No.: CN108842179A] prepared twin boundaries by cutting single crystal seed crystals, closely splicing, heating and melting. Feng Qiang et al. in patent [Publication No.: CN103698183A] prepared artificial grain boundaries of high-temperature alloys by cutting high-temperature alloys with different orientation angles and adding an intermediate layer of alloy foil containing specified elements. However, the preparation methods in the above patents all have the problem that it is difficult to distinguish the defect characteristics related to grain boundaries from those related to the bulk phase through instrumental observations in experiments. Therefore, it is necessary to design a new method to solve the above problems. Summary of the invention

[0004] In order to overcome the shortcomings of the prior art, the purpose of the present invention is to provide an improved three-dimensional percolation framework simulation method based on a multi-physics combined dynamic Monte Carlo finite element model, which is an adaptive random algorithm that can simulate the formation of resistance switching (high resistance state to low resistance state) and conductive paths in amorphous and polycrystalline titanium nitride / hafnium oxide / platinum metal-insulator-metal stacks, and can generate continuous grain boundaries in different devices of amorphous and polycrystalline hafnium oxide dielectrics. Finite element calculations will provide an accurate update of the overall device performance, and the coupling of the Monte Carlo and finite element methods can accurately analyze the potential and electric field distribution patterns during the switching stage from high resistance state to low resistance state.

[0005] In order to achieve the above-mentioned invention object, the technical solution adopted by the present invention is: an improved three-dimensional seepage framework simulation method based on multi-physics combined dynamics Monte Carlo finite element model, comprising the following steps:

[0006] Step 1: Create a new defect and update the parameters, including the following sub-steps:

[0007] (a) A probability vector is generated through a Monte Carlo algorithm, indicating the defect formation rate at each individual location and determining the location of new defects controlled by the thermochemical model;

[0008] (b) When a new defect is formed, the local dielectric constant κ, electric field ξ, and conductivity σ will be updated;

[0009] Step 2: Checking the formation of clusters and conductive filaments, including the following sub-steps:

[0010] (a) After initialization of all material and device parameters, the Hoshen-Koppelmann method is used to identify clusters of defects and check clusters to determine whether conductive filaments are formed. The time iteration formula is described by formula (1),

[0011] t k+1 =t k +Δk (1)

[0012] Where: t k represents the kth time, t k+1 represents the time of the k+1th time, Δk is the time difference;

[0013] (b) if the conductive filaments are formed, proceed to the next step; if the conductive filaments are not formed, return to step 1;

[0014] Step 3: Calculate the electric potential and electric field distribution, including the following sub-steps:

[0015] (a) Using the existing spatial defect distribution and parameters, the finite element method is used to calculate the overall device characteristics and update the device parameters again;

[0016] (b) Apply Poisson's equation and Gauss's law to the finite element model to calculate the potential and electric field distribution under high and low resistance conditions;

[0017] Step 4: determine whether a low-resistance conductive path is formed. If so, the simulation method ends. If not, return to step 1.

[0018] The method of the invention is widely used in a device structure of a metal-insulator-metal stack of titanium nitride / hafnium oxide / platinum on a Si substrate.

[0019] In the method of the present invention, the random formation rate of defects in the Monte Carlo algorithm is controlled by a thermochemical model.

[0020] The method of the present invention assumes that each defect formation event is an independent event according to Poisson statistics.

[0021] In the method of the present invention, when a specific compliance level of the low resistance state is determined, the defect configuration provided by the finite element algorithm is used as an input parameter of the finite element calculation.

[0022] The beneficial effects of the present invention are: an improved three-dimensional percolation framework simulation method based on a multi-physics combined dynamic Monte Carlo finite element model, comprising the following steps: (1) forming new defects and updating parameters, (2) checking the formation of clusters and conductive filaments, (3) calculating the potential and electric field distribution, and (4) judging whether a low-resistance conductive path is formed. The model is also used to find the statistical distribution of the formation of conductive paths in amorphous and polycrystalline hafnium oxide dielectrics, and the model can accurately simulate the observations of resistive switching in titanium nitride / hafnium oxide / platinum metal-insulator-metal stacks. Multi-physics simulations coupled with Monte Carlo and finite elements are used to explain the formation of conductive paths and electric field distributions during resistive switching in hafnium oxide-based metal-insulator-metal stacks. Our device-level simulations show that in polycrystalline hafnium oxide-based resistive memory devices, conductive paths at hafnium oxide grain boundary sites are formed faster than within grains, so grain boundaries play a key role in determining the performance of the resistive switching process. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 It is a flow chart of the steps of the method of the present invention.

[0024] Figure 2 This is a diagram of the amorphous hafnium oxide metal-insulator-metal stack structure of device 1.

[0025] Among them: (a) is the amorphous hafnium oxide metal-insulator-metal stack structure diagram in the high resistance state, (b) is the conductive path formation diagram in the low resistance state, (c) is the potential distribution diagram in the low resistance state, (d) is the electric field distribution cross-sectional diagram in the low resistance state, and A is the low resistance state path.

[0026] Figure 3 This is a diagram of the amorphous hafnium oxide metal-insulator-metal stack structure of device 2.

[0027] Among them: (a) is the amorphous hafnium oxide metal-insulator-metal stack structure in the high resistance state, (b) is the conductive path formation diagram in the low resistance state, (c) is the potential distribution diagram in the low resistance state, (d) is the electric field distribution cross-sectional diagram in the low resistance state, and B is the low resistance state path.

[0028] Figure 4This is a diagram of the polycrystalline hafnium oxide metal-insulator-metal stack structure of device 3.

[0029] Among them, (a) is a diagram of the polycrystalline hafnium oxide metal-insulator-metal stack structure in the high resistance state, (b) is a diagram of the conductive path formation in the low resistance state, (c) is a diagram of the potential distribution in the low resistance state, (d) is a cross-sectional diagram of the electric field distribution in the low resistance state, and C is the low resistance state path.

[0030] Figure 5 FIG. 4 is a diagram of the polycrystalline hafnium oxide metal-insulator-metal stack structure of device 4.

[0031] Among them, (a) is a diagram of the polycrystalline hafnium oxide metal-insulator-metal stack structure in the high resistance state. (b) is a diagram of the conductive path formation in the low resistance state, (c) is a diagram of the potential distribution in the low resistance state, (d) is a cross-sectional diagram of the electric field distribution in the low resistance state, and D is the low resistance state path. DETAILED DESCRIPTION

[0032] The present invention will be further described below in conjunction with the embodiments.

[0033] Example 1

[0034] like Figure 1 As shown, an improved three-dimensional seepage framework simulation method based on a multi-physics combined dynamics Monte Carlo finite element model includes the following steps:

[0035] Step 1: Create a new defect and update the parameters, including the following sub-steps:

[0036] (c) generating a probability vector indicating the defect formation rate at each individual location through a Monte Carlo algorithm and determining the location of new defects controlled by a thermochemical model;

[0037] (d) When a new defect is formed, the local dielectric constant κ, electric field ξ, and conductivity σ will be updated;

[0038] Step 2: Checking the formation of clusters and conductive filaments, including the following sub-steps:

[0039] (a) After initialization of all material and device parameters, the Hoshen-Koppelmann method is used to identify clusters of defects and check clusters to determine whether conductive filaments are formed. The time iteration formula is described by formula (1),

[0040] t k+1 =t k +Δk (1)

[0041] Where: t k represents the kth time, t k+1 represents the time of the k+1th time, Δk is the time difference;

[0042] (b) if the conductive filaments are formed, proceed to the next step; if the conductive filaments are not formed, return to step 1;

[0043] Step 3: Calculate the electric potential and electric field distribution, including the following sub-steps:

[0044] (c) using the existing spatial defect distribution and parameters, the finite element method is used to calculate the overall device characteristics and the device parameters are updated again;

[0045] (d) Apply Poisson's equation and Gauss's law to the finite element model to calculate the potential and electric field distribution under high and low resistance conditions;

[0046] Step 4: determine whether a low-resistance conductive path is formed. If so, the simulation method ends. If not, return to step 1.

[0047] The method of the invention is widely used in a device structure of a metal-insulator-metal stack of titanium nitride / hafnium oxide / platinum on a Si substrate.

[0048] In the method of the present invention, the random formation rate of defects in the Monte Carlo algorithm is controlled by a thermochemical model.

[0049] The method of the present invention assumes that each defect formation event is an independent event according to Poisson statistics.

[0050] In the method of the present invention, when a specific compliance level of the low resistance state is determined, the defect configuration provided by the finite element algorithm is used as an input parameter of the finite element calculation.

[0051] Example 2

[0052] Figure 2 and Figure 3 (a) and (b) show 3D simulations of a metal-insulator-metal stack of amorphous hafnium oxide for the high resistance state and the conductive path (low resistance state) in devices 1 and 2, respectively. The generation of defects involves a pseudo-random process with defect clustering in the hafnium oxide dielectric. These defects promote electron tunneling and increase the local conductivity, resulting in a low resistance state. The rebalancing of the potential on the conductive path leads to an enhancement of the local electric field and triggers the evolution of the conductive path.

[0053] Figure 2 and Figure 3(c) and (d) show cross-sectional views of the potential distribution and electric field distribution in the low resistance state for devices 1 and 2, respectively. Our simulation results show that low resistance paths occur at random locations (regions A and B in device 1 and device 2) in metal-insulator-metal devices with amorphous hafnium oxide. In (c) and (d), the potential distribution and enhanced e-field pattern on the low resistance path are highlighted by dashed circles, respectively. It is worth noting that the voltage applied in the finite element simulation is to observe the potential and electric field distribution after the low resistance state occurs in the Monte Carlo process.

[0054] Example 3

[0055] Figure 4 and Figure 5 (a) and (b) are three-dimensional simulations of the metal-insulator-metal stack of polycrystalline hafnium oxide for the high resistance state and the conducting path (low resistance state) in devices 3 and 4, respectively.

[0056] Figure 4 and Figure 5 (c) and (d) show the cross-sectional images of the potential distribution and electric field distribution in the low resistance state for devices 3 and 4, respectively. Compared with the results for amorphous hafnium oxide, our simulations for polycrystalline hafnium oxide always show that the low resistance path occurs at the grain boundary location (regions C and D in devices 3 and 4). Compared with the bulk phase region, oxygen vacancies tend to segregate toward the grain boundary and defects are generated faster. In addition, the physically thinner region on the grain boundary enhances the local electric field along the grain boundary region. After that, under the action of electrical stress, the grain boundary begins to degrade faster than the region in the bulk phase, resulting in the formation of local conductive paths in the grain boundary region.

Claims

1. An improved three-dimensional seepage framework simulation method based on multi-physics combined dynamics Monte Carlo finite element model, characterized in that: The following steps are involved: Step 1: Create a new defect and update the parameters, including the following sub-steps: (a) A probability vector is generated through a Monte Carlo algorithm, indicating the defect formation rate at each individual location and determining the location of new defects controlled by the thermochemical model; (b) When a new defect is formed, the local dielectric constant κ, electric field ξ, and conductivity σ will be updated; Step 2: Checking the formation of clusters and conductive filaments, including the following sub-steps: (a) After initialization of all material and device parameters, the Hoshen-Koppelmann method is used to identify clusters of defects and check clusters to determine whether conductive filaments are formed. The time iteration formula is described by formula (1), t k+1 =t k +Δk (1) Where: t k represents the kth time, t k+1 represents the time of the k+1th time, Δk is the time difference; (b) if the conductive filaments are formed, proceed to the next step; if the conductive filaments are not formed, return to step 1; Step 3: Calculate the electric potential and electric field distribution, including the following sub-steps: (a) Using the existing spatial defect distribution and parameters, the finite element method is used to calculate the overall device characteristics and update the device parameters again; (b) Apply Poisson's equation and Gauss's law to the finite element model to calculate the potential and electric field distribution under high and low resistance conditions; Step 4: determine whether a low-resistance conductive path is formed. If so, the simulation method ends. If not, return to step 1.

2. According to claim 1, an improved three-dimensional seepage framework simulation method based on multi-physics combined dynamics Monte Carlo finite element model is characterized in that: Widely used in device structures of metal-insulator-metal stacks of titanium nitride / hafnium oxide / platinum on Si substrates.

3. The improved three-dimensional seepage framework simulation method based on multi-physics combined dynamics Monte Carlo finite element model according to claim 1 is characterized in that: The random formation rate of defects in the Monte Carlo algorithm is controlled by a thermochemical model.

4. The improved three-dimensional seepage framework simulation method based on multi-physics combined dynamics Monte Carlo finite element model according to claim 1 is characterized in that: Each defect formation event is assumed to be an independent event according to Poisson statistics.

5. The improved three-dimensional seepage framework simulation method based on multi-physics combined dynamics Monte Carlo finite element model according to claim 1 is characterized in that: When a specific compliance level for the low resistance state is determined, the defect configuration provided by the finite element algorithm is used as an input parameter for the finite element calculation.