Time-dependent breakdown spice circuit model, construction and verification method thereof and application thereof
By constructing a SPICE circuit model that combines a compact metal conductive filament model and a resistor degradation model, the problem of the existing TDDB model lacking systematic analysis of the interaction of multiple influencing factors in a complex circuit environment is solved, and efficient modeling of the dynamic degradation process of the dielectric layer and improved reliability of the circuit system are achieved.
Patent Information
- Application Number
- CN202510442847.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2025-04-10
- Publication Date
- 2025-10-10
- Estimated Expiration
- 2045-04-10
AI Technical Summary
Existing TDDB models cannot fully consider the interaction of multiple influencing factors in complex circuit environments and lack systematic analysis, especially the performance and impact at the circuit level have not been studied in sufficient depth.
A time-dependent breakdown SPICE circuit model is constructed, combining the compact metal conductive filament model and the resistance degradation model, adding the process tolerance model, and describing the dynamic degradation process of the dielectric layer through the SPICE equivalent circuit model.
It achieves the prediction of device life and efficient modeling of current and resistance degradation processes, improves computing efficiency and modeling accuracy, can identify potential risks in the evaluation and design stage of dielectric breakdown effects, and significantly improves the reliability of circuit systems.
Smart Images

Figure CN119940259B_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present disclosure relates to the technical field of package interconnect reliability analysis, in particular to a time-dependent breakdown SPICE circuit model and a construction and verification method and application thereof. BACKGROUND
[0002] With the miniaturization of integrated circuit technology and the rapid development of three-dimensional stacked integration technology, the reliability problem of advanced interconnect structure under high-density integration is increasingly prominent, and time-dependent breakdown (TDDB) phenomenon has become a key factor affecting the performance degradation and failure of interconnect. TDDB refers to the gradual migration of atoms or molecules in dielectric materials under long-term electric field stress, forming a conductive path, thereby increasing the leakage current, and ultimately causing dielectric breakdown (BD). The occurrence of BD will cause a sudden change in the electrical properties of the material, causing circuit failure.
[0003] Currently, TDDB research mainly focuses on the failure mechanism and model of the dielectric layer, and the physical model, statistical model and empirical model proposed for TDDB phenomenon. Common models include E, 1 / E, E 1 / 2 models and power law models, fatigue damage models, etc. These models mostly focus on qualitatively describing the long-term behavior of dielectric material failure, and cannot fully consider the interaction of multiple influencing factors in complex circuit environments. However, existing models are mainly based on the description of macroscopic physical phenomena or empirical data, lacking systematic analysis of dielectric failure in circuit systems, especially the performance and influence of TDDB effect at the circuit level have not been deeply studied. SUMMARY
[0004] The present disclosure aims to at least solve one of the problems existing in the prior art, and provides a time-dependent breakdown SPICE circuit model and a construction and verification method and application thereof.
[0005] In one aspect of the present disclosure, a construction method of a time-dependent breakdown SPICE circuit model is provided, the construction method comprising:
[0006] composing a metal conduction filament compact model and a resistance degradation model into a SPICE equivalent circuit model;
[0007] adding a process tolerance model to the SPICE equivalent circuit model to obtain a time-dependent breakdown SPICE circuit model.
[0008] Optionally, after applying a forward voltage to the active electrode of the metal conduction filament compact model, the change in the number of metal atoms in the longitudinal direction of the conduction channel of the metal conduction filament compact model is represented as:
[0009] ;
[0010] After a forward voltage is applied to the active electrode of the metal conductive filament compact model, the change in the number of metal atoms in the lateral direction within the conductive channel of the metal conductive filament compact model is expressed as:
[0011] ;
[0012] in, It represents the change of the conductive channel in the longitudinal direction caused by the drift of metal ions, and its value range is [0, 1]; Indicates the lateral change of the conductive channel caused by metal ion drift, and its value range is [0, 1]; represents the amount of metal ion drift and , represents the drift velocity of metal ions, Indicates the concentration of metal elements; represents the bottom diameter of the conductive channel of the compact model of the metal conductive filament; Indicates time; represents the longitudinal length of the conductive path and ; represents the length of the conductive filament at time t.
[0013] Optionally, after a forward voltage is applied to the active electrode of the metal conductive filament compact model, the results of the changes in the conductive filaments in the metal conductive filament compact model in the transverse and longitudinal directions are respectively expressed as follows:
[0014] ;
[0015] ;
[0016] in, represents the resonant frequency; It represents the effective transition distance of ions; represents the potential barrier height; represents the Boltzmann constant; Indicates temperature correction; Indicates temperature; represents the field acceleration factor; represents the applied voltage and ; Indicates the electric field strength; represents the thickness of the dielectric layer in the compact model of the metal conductive filament; represents an intermediate variable and ; Indicates the number of charged ions; represents electric charge; Represents the diameter of the conductive filament at time t.
[0017] Optionally, the resistance degradation model in the initial state is Expressed as:
[0018] ;
[0019] in, represents the resistivity of the dielectric layer in the compact model of the metal conductive filament; represents the area of the dielectric layer in the compact model of the metal conductive filament and .
[0020] Optionally, the resistance degradation model is the resistance of the dielectric layer in the metal conductive filament compact model during the degradation process. Expressed as:
[0021] ;
[0022] in, It represents the resistivity of the dielectric layer in the compact metal conductive filament model after the conductive channel in the compact metal conductive filament model is formed.
[0023] Optionally, the resistance degradation model is the resistance at the end of the dielectric layer performance degradation in the metal conductive filament compact model. Expressed as:
[0024] ;
[0025] in, represents the top cross-sectional radius of the conductive channel of the compact model of the metal conductive filament; represents the bottom cross-sectional radius of the conductive channel of the compact metal conductive filament model.
[0026] Optionally, the process tolerance model is expressed as:
[0027] ;
[0028] in, represents the power spectrum of the Gaussian autocorrelation function; represents the wave vector and ; represents an imaginary unit; Indicates the total number of material structural units into which the dielectric layer in the compact model of metal conductive filaments is divided; Indicates the distance between the upper and lower material structural units in the dielectric layer; Indicates the length of the line edge roughness profile; Indicates the RMS amplitude of the line edge roughness profile.
[0029] Another aspect of the present disclosure provides a time-dependent breakdown SPICE circuit model, wherein the time-dependent breakdown SPICE circuit model is constructed using the method for constructing a time-dependent breakdown SPICE circuit model described above.
[0030] Another aspect of the present disclosure provides a method for verifying a time-dependent breakdown SPICE circuit model, the method comprising:
[0031] Conducting experiments on the time-dependent breakdown effect of the gate oxide layer and / or the time-dependent breakdown effect of the TSV blind hole to obtain corresponding gate oxide layer experimental data and / or TSV blind hole experimental data; and / or conducting simulations on the breakdown effect of the dielectric layer on the chemical mechanical polishing surface to obtain corresponding simulation data;
[0032] Using the aforementioned time-dependent breakdown SPICE circuit model to characterize the time-dependent breakdown effect of the gate oxide layer and / or the time-dependent breakdown effect of the TSV blind hole under corresponding experimental conditions, and obtaining corresponding gate oxide layer model simulation data and / or TSV blind hole model simulation data; and / or using the aforementioned time-dependent breakdown SPICE circuit model to characterize the breakdown effect of the dielectric layer on the chemical mechanical polishing surface under corresponding simulation conditions, and obtaining corresponding chemical mechanical polishing surface model simulation data;
[0033] Comparing the gate oxide layer experimental data with the gate oxide layer model simulation data, and / or comparing the TSV blind hole experimental data with the TSV blind hole model simulation data, and / or comparing the simulation data with the chemical mechanical polishing surface model simulation data;
[0034] The time-dependent breakdown SPICE circuit model is verified based on the comparison results.
[0035] Another aspect of the present disclosure provides an application of a time-destruction SPICE circuit model, in which the time-destruction SPICE circuit model described above is applied to an equivalent circuit model of a power distribution network and / or through-silicon vias to simulate degradation effects of the circuit system.
[0036] Compared with the existing technology, the present disclosure adopts a compact model of metal conductive filaments to explain relevant physical phenomena, and combines the resistance degradation model used to describe the resistance degradation phenomenon to construct a SPICE equivalent circuit model, converting the TDDB effect into changes in circuit parameters, effectively describing the dynamic degradation process of the dielectric layer under electric field stress, and realizing the modeling of device life prediction, current and resistance degradation process, greatly simplifying the calculation process. At the same time, a process tolerance model used to describe process uncertainty is introduced to construct a time-dependent breakdown SPICE circuit model, which not only improves the calculation efficiency but also maintains a high modeling accuracy. It has significant engineering application potential, can identify potential risks in the evaluation and design stage of the dielectric breakdown effect, and significantly improve its reliability through optimized design. BRIEF DESCRIPTION OF THE DRAWINGS
[0037] One or more embodiments are exemplarily illustrated by pictures in the corresponding drawings, and these exemplifications do not constitute limitations on the embodiments. Elements with the same reference numerals in the drawings represent similar elements, and unless otherwise stated, the figures in the drawings do not constitute proportional limitations.
[0038] Figure 1 A flowchart of a method for constructing a time-destruction SPICE circuit model provided in one embodiment of the present disclosure;
[0039] Figure 2 A schematic diagram of a compact model of a metal conductive filament provided in accordance with another embodiment of the present disclosure;
[0040] Figure 3 A circuit diagram corresponding to a SPICE equivalent circuit model provided in another embodiment of the present disclosure;
[0041] Figure 4 Gaussian ACF and LER profile randomly generated by LER provided in another embodiment of the present disclosure;
[0042] Figure 5 A schematic diagram of the TDDB effect of a gate oxide layer provided in another embodiment of the present disclosure;
[0043] Figure 6 A comparison chart of simulation results and experimental data of a SPICE circuit model of a gate oxide layer provided in another embodiment of the present disclosure;
[0044] Figure 7 A comparison chart of simulation results and experimental data of a SPICE circuit model of a TSV blind via provided by another embodiment of the present disclosure;
[0045] Figure 8 A schematic diagram of an advanced interconnection on a low dielectric constant substrate and related TDDB effect provided in another embodiment of the present disclosure;
[0046] Figure 9 A CMP surface life comparison chart provided for another embodiment of the present disclosure;
[0047] Figure 10 A schematic diagram of an equivalent circuit of a 2×2 PDN and its unit structure provided in another embodiment of the present disclosure;
[0048] Figure 11 A schematic diagram of PDN impedance comparison provided by another embodiment of the present disclosure;
[0049] Figure 12 A schematic diagram comparing the lifespans of PDNs with different dielectric layer thicknesses provided in another embodiment of the present disclosure;
[0050] Figure 13 A schematic diagram of the relative error between the simulation results and the numerical simulation results of the TDDB SPICE model provided in another embodiment of the present disclosure;
[0051] Figure 14 A schematic diagram of the change of PDN impedance over time provided by another embodiment of the present disclosure;
[0052] Figure 15 A schematic diagram and an equivalent circuit model corresponding to a signal-ground TSV according to another embodiment of the present disclosure;
[0053] Figure 16 A comparison diagram of simulation results of a TDDB SPICE model of TSV and simulation results of a numerical model provided in another embodiment of the present disclosure;
[0054] Figure 17 A comparison chart of lifespan predictions for TSVs with different dielectric layer thicknesses provided in another embodiment of the present disclosure;
[0055] Figure 18 A comparison diagram of S parameters at different conductivity change rates provided by another embodiment of the present disclosure;
[0056] Figure 19 A comparison diagram of S parameters at different frequencies provided by another embodiment of the present disclosure. DETAILED DESCRIPTION
[0057] In order to make the purposes, technical solutions and advantages of the embodiments of the present disclosure clearer, the embodiments of the present disclosure will be described in detail below with reference to the drawings. However, those skilled in the art can understand that in the embodiments of the present disclosure, many technical details are presented in order to make the readers better understand the present disclosure. However, the technical solutions claimed by the present disclosure can be implemented even without these technical details and various changes and modifications based on the following embodiments. The division of the following embodiments is for the convenience of description and should not constitute any limitation on the specific embodiments of the present disclosure. The embodiments can be combined with each other and referred to each other without contradiction.
[0058] One embodiment of the present disclosure relates to a method for constructing a time-dependent breakdown SPICE circuit model, the flow of which is shown as Figure 1
[0059] Step S110, the metal conductive filament compact model and the resistance degradation model are combined into a SPICE equivalent circuit model.
[0060] Specifically, SPICE (Simulation Program with Integrated Circuit Emphasis) is a language simulator software for circuit description and simulation, which is used to detect the integrity of the connection and function of the circuit, and to predict the behavior of the circuit. The SPICE equivalent circuit model is a model used to describe the metal conductive filament compact model and the resistance degradation model in the SPICE software.
[0061] The metal conductive filament compact model is a model used to describe the behavior of the metal conductive filament. When the dielectric material is subjected to an external electric field, the metal atoms in the dielectric material are driven by the electric field, and under the action of the electric field force, they gradually gather along a certain path and form an elongated conductive channel connecting the two electrodes, i.e. a metal conductive filament (CF). With the formation of the metal conductive filament, the conductivity of the dielectric material changes significantly, the resistance of the dielectric layer decreases significantly, and a series of electrochemical reactions and physical changes occur, which is simulated by the metal conductive filament compact model.
[0062] As the metal conductive filament in the dielectric material gradually grows, the conductivity of the dielectric material will gradually increase, thereby enhancing the concentration and mobility of the metal atoms in the dielectric material, resulting in a gradual decrease in the resistance of the dielectric material, which is simulated by the resistance degradation model.
[0063] Exemplarily, Figure 2 A schematic diagram of a compact model of a metal conductive filament is shown, detailing the thickness of the filament. The growth process of metal conductive filaments, namely metal CF, in the dielectric layer. The upper and lower electrodes of the dielectric layer are the anode and cathode respectively, and the other end of the cathode is connected to the ground terminal (GND). The anode material is metal, and the cathode material is metal or semiconductor. In the initial state, the forward voltage applied to the active electrode of the dielectric layer, namely the anode, is In the off state, there are no defects in the dielectric layer and no metal atoms move inside the dielectric layer. In the intermediate state, the forward voltage applied to the anode of the dielectric layer is In the conducting state, an electrochemical reaction occurs at the interface between the anode and the dielectric layer, which manifests as metal atoms Loss of electrons into metal ions Right now , then, metal ions Under the action of the electric field, it migrates toward the inert electrode, the cathode, and captures electrons near the cathode. Reduction reaction occurs to metal atoms Right now , and finally, the metal atoms Gradually gather in the dielectric layer to form a The conductive channel is formed in two stages. In the first stage, a thin conductive filament is formed in the longitudinal direction. In the second stage, the conductive filament continues to grow in the transverse direction, eventually forming a cone-shaped conductive channel between the upper and lower electrodes, the anode and cathode. This conductive channel is the final metal conductive filament, or metal CF. After the metal CF is formed, the dielectric layer is in a conductive state.
[0064] Longitudinal length of the conductive path and bottom diameter As an internal state variable, it can be used to describe the changes of the conductive filaments in the longitudinal and transverse directions of the compact metal conductive filament model.
[0065] When a positive voltage is applied to the active electrode of the metal conductive filament compact model, the change in the number of metal atoms in the longitudinal direction of the conductive channel of the metal conductive filament compact model is expressed as:
[0066] .
[0067] When a positive voltage is applied to the active electrode of the compact metal conductive filament model, metal ions drift into the conductive channel, causing the lateral number of metal atoms in the conductive channel to change. At this time, the lateral change in the number of metal atoms in the conductive channel of the compact metal conductive filament model is expressed as:
[0068] .
[0069] in, It represents the change in the conductive channel in the longitudinal direction caused by the drift of metal ions, and its value range is [0, 1]. It represents the lateral change of the conductive channel caused by the drift of metal ions, and its value range is [0, 1]. represents the amount of metal ion drift and ,in other words, It represents the number of metal ions per unit cross-sectional area that pass through the conductive channel per unit time. Indicates the drift velocity of metal ions. Indicates the concentration of metal elements. The bottom diameter of the conductive channel representing the compact model of the metal conductive filament. Indicates time. represents the longitudinal length of the conductive path and . represents the length of the conductive filament at time t.
[0070] For example, after a positive voltage is applied to the active electrode of the metal conductive filament compact model, the changes in the conductive filaments in the metal conductive filament compact model in the transverse and longitudinal directions are respectively expressed as follows:
[0071] ;
[0072] .
[0073] in, Indicates the resonant frequency. Represents the effective ion transition distance. represents the barrier height. represents the Boltzmann constant. Indicates temperature correction. Indicates temperature. Represents the field acceleration factor. represents the applied voltage and . Indicates the electric field strength. Represents the thickness of the dielectric layer in the compact model of metal conductive filaments. represents an intermediate variable and . Indicates the number of charged ions. Represents charge. Represents the diameter of the conductive filament at time t.
[0074] For example, Figure 2 As shown, in the initial state, the forward voltage applied to the anode of the dielectric layer of the metal conductive filament compact model is In the off state, there is no movement of metal atoms inside the dielectric layer. At this time, the dielectric layer has excellent insulation performance, low dielectric loss and good transmission characteristics. Therefore, the resistance degradation model in the initial state is Expressed as: .in, Represents the resistivity of the dielectric layer in the compact model of metal conductive filaments. represents the area of the dielectric layer in the compact model of the metal conductive filament and .
[0075] For example, Figure 2 As shown in the figure, in the intermediate state, that is, during the growth of the metal CF in the dielectric layer of the compact metal conductive filament model, under the influence of physical fields such as temperature and electric field, defects gradually appear in the dielectric layer and metal atoms migrate. As metal atoms continue to accumulate, the size of the metal CF gradually increases, and the resistance, isolation and other properties of the dielectric layer begin to gradually decline. During the degradation of the dielectric layer performance, the resistance of the dielectric layer is determined by the size of the metal CF. Therefore, the resistance degradation model is the resistance degradation model of the dielectric layer in the compact metal conductive filament model during the degradation of the dielectric layer performance. Expressed as: .in, It represents the resistivity of the dielectric layer in the metal conductive filament compact model after the conductive channel is formed in the metal conductive filament compact model. During the performance degradation of the dielectric layer, the circuit diagram corresponding to the SPICE equivalent circuit model is as follows Figure 3 shown.
[0076] For example, Figure 2 As shown in the figure, after the dielectric layer of the metal conductive filament compact model has been used for a long time or exposed to extreme physical fields, the metal CF in the dielectric layer is completely formed, and the metal CF electrically connects the anode and cathode, and the dielectric layer is in a conductive state. At this time, the resistance of the dielectric layer is also determined by the metal CF. Therefore, the resistance degradation model is the resistance at the end of the dielectric layer performance degradation in the metal conductive filament compact model. Expressed as: .in, The top cross-sectional radius of the conductive channel representing the compact model of the metallic conductive filament. The bottom cross-sectional radius of the conductive channel representing the compact model of the metal conductive filament.
[0077] Step S120 , adding the process tolerance model to the SPICE equivalent circuit model to obtain a time-dependent breakdown SPICE circuit model.
[0078] Specifically, process structure uncertainty will have a certain impact on the TDDB phenomenon. Therefore, the process tolerance model can use the process tolerance power spectrum density function to characterize the impact of process uncertainty.
[0079] For example, changes in line edge roughness (LER) and line-to-line (L2L) spacing are important factors affecting the performance of nanoscale devices and interconnects, especially in terms of TDDB, which can cause significant performance degradation and reliability issues. Therefore, in order to effectively quantify and evaluate the impact of LER on TDDB reliability, this embodiment develops a Fourier synthesis method based on the power spectral density (PSD) of the Gaussian autocorrelation function (ACF) to simulate and generate LER characteristic curves. In this Fourier synthesis method, the material structure is subdivided into N discrete micro units, and a random phase is introduced for each unit to ensure the randomness of the simulation and the actual matching. The LER amplitude generated by the random phase is constrained by the power spectral density of the Gaussian ACF to ensure that it conforms to the statistical distribution characteristics of experimental observations. The process tolerance model is used to characterize and model the LER characteristics, and the process tolerance model can be expressed as: .
[0080] in, Represents the power spectrum of the Gaussian autocorrelation function. represents the wave vector and . Represents an imaginary unit. Represents the total number of material structural units into which the dielectric layer in the compact metal conductive filament model is divided. Indicates the distance between the upper and lower material structural units in the dielectric layer, and its value should be much smaller than To ensure a reasonable discretization resolution. Indicates the length of the line edge roughness profile. Indicates the RMS amplitude of the line edge roughness profile.
[0081] For example, The value range of is usually 20nm to 50nm. =30nm, dx=0.5nm, then the value of dx is much smaller than . Figure 4 It shows that the ideal process thickness of the dielectric layer is 50nm. The Gaussian ACF and LER profiles randomly generated when LER is 0, 0.5, and 2.5, respectively. The x-axis represents the length of the dielectric layer, and the y-axis represents the height of the dielectric layer.
[0082] Compared with the prior art, the method for constructing a time-dependent breakdown SPICE circuit model provided in the embodiments of the present disclosure uses a compact model of metal conductive filaments to explain related physical phenomena, and combines it with a resistance degradation model used to describe resistance degradation phenomena to construct a SPICE equivalent circuit model. This converts the TDDB effect into changes in circuit parameters, effectively describes the dynamic degradation process of the dielectric layer under electric field stress, and realizes device life prediction and modeling of current and resistance degradation processes, greatly simplifying the calculation process. At the same time, a process tolerance model used to describe process uncertainty is introduced to construct a time-dependent breakdown SPICE circuit model, which not only improves calculation efficiency but also maintains a high level of modeling accuracy. It has significant engineering application potential and can identify potential risks in the evaluation and design stage of the dielectric breakdown effect, and significantly improve its reliability through design optimization.
[0083] Another embodiment of the present disclosure relates to a time-destruction SPICE circuit model. The time-destruction SPICE circuit model is constructed using the method for constructing a time-destruction SPICE circuit model described in the above embodiment.
[0084] Compared with the existing technology, the time-dependent breakdown SPICE circuit model provided by the embodiments of the present disclosure uses a compact model of metal conductive filaments to explain related physical phenomena, and combines it with a resistance degradation model used to describe resistance degradation phenomena. Through the SPICE equivalent circuit model, the TDDB effect is converted into changes in circuit parameters, effectively describing the dynamic degradation process of the dielectric layer under electric field stress, realizing device life prediction and modeling of current and resistance degradation processes, greatly simplifying the calculation process, and at the same time introducing a process tolerance model for describing process uncertainty, which not only improves calculation efficiency but also maintains a high level of modeling accuracy. It has significant engineering application potential and can identify potential risks in the evaluation and design stage of the dielectric breakdown effect, and significantly improve its reliability through design optimization.
[0085] Another embodiment of the present disclosure relates to a method for verifying a time-dependent breakdown SPICE circuit model, comprising: conducting experiments on the time-dependent breakdown effect of a gate oxide layer and a TSV blind via, obtaining corresponding gate oxide experimental data and TSV blind via experimental data. Simulating the breakdown effect of a dielectric layer on a surface during chemical mechanical polishing (CMP) to obtain corresponding simulation data. Using the aforementioned time-dependent breakdown SPICE circuit model to characterize the time-dependent breakdown effect of a gate oxide layer and a TSV blind via under corresponding experimental conditions, corresponding gate oxide model simulation data and TSV blind via model simulation data are obtained. Using the aforementioned time-dependent breakdown SPICE circuit model to characterize the breakdown effect of a dielectric layer on a surface during chemical mechanical polishing under corresponding simulation conditions, corresponding chemical mechanical polishing surface model simulation data are obtained. The gate oxide experimental data are compared with the gate oxide model simulation data, the TSV blind via experimental data are compared with the TSV blind via model simulation data, and the simulation data are compared with the chemical mechanical polishing surface model simulation data. The time-dependent breakdown SPICE circuit model is verified based on the comparison results.
[0086] The time-dependent breakdown effect of the gate oxide layer is used to verify the time-dependent breakdown SPICE circuit model.
[0087] Combined Figure 5 When the gate is biased, metal ions gradually migrate through the gate oxide layer. Over time, aging factors such as the electric field, temperature, and tunneling current cause the metal ions to gradually accumulate, resulting in a local high current density. This high current density not only accelerates the gradual migration of the metal ions but also enhances the interaction between them. Ultimately, a low-resistance CF composed of metal atoms is formed in the gate oxide layer. The self-heating effect caused by the current flowing through the CF causes irreversible thermo-mechanical damage to the gate oxide layer, leading to hard breakdown. To verify the accuracy and performance of the time-dependent breakdown SPICE circuit model, experiments were conducted on the time-dependent breakdown effect of the gate oxide layer under different temperature and voltage conditions, and corresponding experimental results were obtained. The time-dependent breakdown effect of the gate oxide layer was simulated using the time-dependent breakdown SPICE circuit model using the parameter values listed in Table 1, and corresponding simulation results were obtained. The experimental data was compared with the simulation results of the time-dependent breakdown SPICE circuit model, and the comparison results are shown in Figure 6. Figure 6In this article, "SPICE model" refers to the simulation results of the time-destruction SPICE circuit model. The voltage on the horizontal axis refers to the voltage used in the experiments and simulations, and the lifetime on the vertical axis refers to the lifetime of the gate oxide at the corresponding voltage. Comparison results show good consistency between the time-destruction SPICE circuit model and experimental data under the corresponding operating environment, demonstrating the scalability and reliability of the time-destruction SPICE circuit model.
[0088] Table 1 Parameter information of the time-dependent breakdown SPICE circuit model
[0089]
[0090] The time-dependent breakdown effect of TSV blind vias is used to verify the time-dependent breakdown SPICE circuit model.
[0091] Under electrical stress conditions, leakage current will be stimulated in the silicon dioxide insulating layer, which will not only trigger the migration of copper ions in the insulating layer, but also cause copper atoms to gradually accumulate in the insulating layer. As time goes by, the concentration of copper atoms on the dielectric layer continues to accumulate, gradually forming an atomic migration channel. Once the concentration of copper atoms reaches a certain level, a metal conductive filament connecting the two TSVs will be formed. When the leakage current exceeds the breakdown threshold of the silicon dioxide insulating layer, instantaneous breakdown occurs. In order to evaluate the accuracy and performance of the time-dependent breakdown SPICE circuit model, experiments were conducted on the time-dependent breakdown effect of TSV blind holes under various stress conditions to obtain corresponding experimental data, and the time-dependent breakdown SPICE circuit model was used to simulate the time-dependent breakdown effect of TSV blind holes under corresponding stress conditions to obtain corresponding simulation results. The experimental data were compared with the simulation results of the time-dependent breakdown SPICE circuit model, and the comparison results are shown in Figure 2. Figure 7 shown. Figure 7 In this data, "SPICE model" refers to the simulation results of the time-destruction SPICE circuit model. The stimulus conditions on the horizontal axis refer to the stimulus conditions used in the experimental and simulation processes, and the lifetime on the vertical axis refers to the lifetime of the TSV blind via under the corresponding stimulus conditions. Comparison results show that the maximum relative error between the simulation and experimental data is less than 5%, demonstrating the high accuracy and reliability of the time-destruction SPICE circuit model. In other words, the time-destruction SPICE circuit model is capable of characterizing the TDDB mechanism in TSV structures.
[0092] Next, the time-dependent breakdown SPICE circuit model is verified using CMP surface simulation data.
[0093] Over time, the accumulation of copper (Cu) atoms on the CMP surface causes its concentration to reach a certain threshold, forming a leakage current path, such as Figure 8This leakage current path accelerates the degradation of dielectric layers, especially low-k dielectric layers, ultimately leading to dielectric breakdown and affecting the overall performance of the chip. TDDB failure on CMP surfaces is primarily attributed to the gradual accumulation of copper atoms on the dielectric surface, which causes its conductivity to increase over time, as described by the following equation:
[0094] ;
[0095] .
[0096] in, represents the conductivity between center i and center j. Indicates the initial center distance. It represents the distance between center i and center j, and is the localization radius of the electron at such center. represents the energy barrier between center i and center j. Represents thermal energy. represents the effective transition distance. C represents the normalized concentration. q represents the charge. E represents the electric field intensity. t represents time. D represents the diffusion coefficient.
[0097] In order to verify the accuracy and effectiveness of the time-dependent breakdown SPICE circuit model, based on Figure 8 The CMP surface life was simulated and the corresponding simulation results were obtained. The time-dependent breakdown effect of the CMP surface was simulated using the time-dependent breakdown SPICE circuit model under the corresponding simulation conditions to obtain the corresponding model simulation results. The simulation results were compared with the model simulation results. Figure 9 shown. Figure 9 In this article, SPICE model is used to refer to the simulation results of the time-destruction SPICE circuit model, and FEM model is used to refer to the simulation results based on Figure 8 The comparison results show that the simulation results match the model simulation results well, which shows that the time-delayed breakdown SPICE circuit model can predict the TDDB effect of CMP surface with high reliability and accuracy.
[0098] Compared with the prior art, the verification method of the time-destruction SPICE circuit model provided in the embodiment of the present disclosure verifies the time-destruction SPICE circuit model provided in the above embodiment, and shows that the time-destruction SPICE circuit model can predict the TDDB effect of the gate oxide layer, TSV blind hole, and CMP surface with high reliability and high accuracy.
[0099] Another embodiment of the present disclosure relates to an application of a time-destruction SPICE circuit model, in which the time-destruction SPICE circuit model described above is applied to an equivalent circuit model of a power distribution network (PDN) and a through-silicon via (TSV) to simulate the degradation effects of the circuit system.
[0100] The following describes the application of the time-destruction SPICE circuit model to the power distribution network.
[0101] A PDN equivalent circuit model was constructed, and numerical simulation was used to verify the accuracy of the failure time of the time-dependent breakdown SPICE circuit model and the degradation of the PDN impedance during the simulated degradation process. Then, the change of the PDN impedance over time was evaluated.
[0102] Specifically, a mesh PDN is usually composed of a power layer and a ground layer. Taking a 2×2 configured PND as an example, Figure 10 The circuit model of the unit structure of the mesh PDN consists of a resistor R, an inductor L, a capacitor C, and a conductor G. The 2×2 PDN configuration is modeled, the corresponding PDN impedance is calculated based on the circuit model, and the PDN impedance corresponding to the 2×2 PDN configuration is calculated using numerical simulation. The PDN impedance calculated using the circuit model is compared with the PDN impedance calculated using numerical simulation. The comparison results are shown in Figure 2. Figure 11 shown. Figure 11 In this data, the SPICE model represents the PDN impedance calculated using the circuit model, while the HFSS model represents the PDN impedance calculated using numerical simulation. Comparison results show that the two sets of data match well within the [0, 20] GHz range. Due to the long operation time of the PDN, the dielectric gradually degrades and eventually fails.
[0103] A TDDB SPICE model, a time-dependent breakdown SPICE circuit model of a PDN with V = 1 V and T = 363 K, is constructed. Figure 12 The TDDB SPICE model predicted lifetime and the numerical model predicted lifetime corresponding to PDNs with different dielectric layer thicknesses are shown. Figure 12 In this article, SPICE refers to the corresponding data of the TDDB SPICE model, and FEM refers to the corresponding data of the numerical model. Figure 12 As can be seen from the figure, the two sets of data match well, indicating that the TDDB SPICE model can accurately characterize the TDDB effect of the grid PDN. Next, by embedding the developed TDDB SPICE model into the circuit model of the PDN, the TDDB SPICE circuit model of the PDN is obtained. Figure 10As shown in Figure 1, this can be achieved by updating G in the circuit model of the PDN unit structure to the TDDB SPICE model. The simulation results of the TDDB SPICE model are compared with those obtained by the numerical simulation model to verify the performance of the established TDDB SPICE model. Figure 13 As shown in the figure, the SPICE model is used to refer to the simulation results of the TDDB SPICE model, and the FEM model is used to refer to the results obtained by numerical simulation. It is observed that when the dielectric layer conductivity change rate G is in the range of 1 to 1000, the maximum relative error is no more than 2%. Finally, the TDDB SPICE model and numerical simulation are used to predict the degradation of PDN impedance over time. The change of PDN impedance over time is shown in the figure. Figure 14 shown. Figure 14 In Figure 1, the SPICE model is used to refer to the simulation results of the TDDB SPICE model, and the FEM model is used to refer to the results obtained by numerical simulation. It is observed that the simulation results of the TDDB SPICE model well characterize the drop in PDN impedance.
[0104] The following describes the application of the time-dependent breakdown SPICE circuit model to through-silicon vias.
[0105] A TSV equivalent circuit model is constructed, and the accuracy of the failure time of the time-dependent breakdown SPICE circuit model and the degradation of the TSV S parameters during the simulated degradation process are verified through numerical simulation. Then, the change of the TSV S parameters over time is evaluated.
[0106] Specifically, the schematic diagram and equivalent circuit model corresponding to the signal-ground TSV are as follows: Figure 15 The left picture and Figure 15 As shown in the right figure in the figure. During long-term operation, ion migration will occur in the dielectric layer, forming a conductive path, causing the dielectric layer to gradually change from a closed state to an open state. Based on the construction method of the time-dependent breakdown SPICE circuit model provided in the above embodiment, the TDDB SPICE model of TSV is constructed. The stresses such as the bias voltage and temperature of the TSV can be regarded as constants. Here, the operating conditions corresponding to the TSV are set to 2V and 473 K In the TDDB SPICE model, the activation energy of the dielectric layer is set to 1.42eV, and the conductivity gradually increases from the initial value of 10S / m to 200S / m over time. The simulation results of the TDDB SPICE model of TSV are compared with those of the numerical model. The comparison results are shown in Figure 2. Figure 16 shown. Figure 16In the TDDB SPICE model, the SPICE model is used to refer to the simulation results of the TSV TDDB SPICE model, and the FEM model is used to refer to the simulation results of the numerical model. It is observed that the two sets of data match well. Furthermore, by combining the above TDDB SPICE model with the traditional TSV circuit model, the TDDB SPICE model corresponding to TSV can be obtained. For example, it can be obtained by Figure 15 G in sub The TDDB SPICE model for TSV can not only characterize the real-time characteristics of TSV, but also predict its performance degradation over time. Figure 17 The TDDB SPICE model predicted lifetime and the numerical model predicted lifetime corresponding to TSVs with different dielectric layer thicknesses are shown. Figure 17 In this article, SPICE refers to the corresponding data of the TDDB SPICE model, and FEM refers to the corresponding data of the numerical model. Figure 17 As can be seen from the figure, the two sets of data match well, indicating that the TDDB SPICE model can accurately characterize the TDDB effect of TSV. Similarly, the accuracy of the TDDB SPICE model is first verified, and then its ability to predict degradation over time is verified. Figure 18 As shown in Figure 2, the S-parameters calculated using the TDDB SPICE model are compared with those obtained using the numerical model at different conductivity change rates. The results show good agreement between the two sets of data. The TDDB SPICE model is then used to predict the overtime degradation of the TSV pair. Figure 19 In
[15] , the SPICE model is used to refer to the prediction results of the TDDB SPICE model, and the FEM model is used to refer to the results obtained by the numerical model. It is observed that the TDDB SPICE model can predict the degradation of TSV performance over time.
[0107] Compared with the prior art, the application of the time-destruction SPICE circuit model provided by the embodiments of the present disclosure can accurately simulate and highly accurately predict the degradation effects of the power distribution network and the circuit system of silicon vias.
[0108] Those skilled in the art will appreciate that the above-mentioned embodiments are specific embodiments for implementing the present disclosure, and that in actual applications, various changes may be made thereto in form and detail without departing from the spirit and scope of the present disclosure.
Claims
1. A method for constructing a time-destruction SPICE circuit model, characterized in that: The construction method comprises: The compact model of metal conductive filaments and the resistance degradation model are combined into a SPICE equivalent circuit model; Adding a process tolerance model for describing process uncertainty to the SPICE equivalent circuit model to obtain a time-dependent breakdown SPICE circuit model; The resistance degradation model is used to describe the resistance degradation phenomenon; the resistance degradation model includes the resistance of the dielectric layer during the performance degradation process in the compact metal conductive filament model to achieve device life prediction; the SPICE equivalent circuit model is used to convert the time-dependent breakdown effect into changes in circuit parameters and describe the dynamic degradation process of the dielectric layer under electric field stress.
2. The construction method according to claim 1, characterized in that After a forward voltage is applied to the active electrode of the metal conductive filament compact model, the change in the number of metal atoms in the longitudinal direction of the conductive channel of the metal conductive filament compact model is expressed as: ; After a forward voltage is applied to the active electrode of the metal conductive filament compact model, the change in the number of metal atoms in the lateral direction within the conductive channel of the metal conductive filament compact model is expressed as: ; in, It represents the change of the conductive channel in the longitudinal direction caused by the drift of metal ions, and its value range is [0, 1]; Indicates the lateral change of the conductive channel caused by metal ion drift, and its value range is [0, 1]; represents the amount of metal ion drift and , represents the drift velocity of metal ions, Indicates the concentration of metal elements; represents the bottom diameter of the conductive channel of the compact model of the metal conductive filament; Indicates time; represents the longitudinal length of the conductive path and ; represents the length of the conductive filament at time t.
3. The construction method according to claim 2, characterized in that After a forward voltage is applied to the active electrode of the metal conductive filament compact model, the changes in the conductive filaments in the metal conductive filament compact model in the transverse and longitudinal directions are respectively expressed as follows: ; ; in, represents the resonant frequency; It represents the effective transition distance of ions; represents the potential barrier height; represents the Boltzmann constant; Indicates temperature correction; Indicates temperature; represents the field acceleration factor; represents the applied voltage and ; Indicates the electric field strength; represents the thickness of the dielectric layer in the compact model of the metal conductive filament; represents an intermediate variable and ; Indicates the number of charged ions; represents electric charge; Represents the diameter of the conductive filament at time t.
4. The construction method according to claim 3, characterized in that The resistance degradation model is the resistance in the initial state Expressed as: ; in, represents the resistivity of the dielectric layer in the compact model of the metal conductive filament; represents the area of the dielectric layer in the compact model of the metal conductive filament and .
5. The construction method according to claim 4, characterized in that The resistance degradation model is the resistance of the dielectric layer in the metal conductive filament compact model during the degradation process. Expressed as: ; in, It represents the resistivity of the dielectric layer in the compact metal conductive filament model after the conductive channel in the compact metal conductive filament model is formed.
6. The construction method according to claim 5, characterized in that: The resistance degradation model is the resistance at the end of the dielectric layer performance degradation in the metal conductive filament compact model. Expressed as: ; in, represents the top cross-sectional radius of the conductive channel of the compact model of the metal conductive filament; represents the bottom cross-sectional radius of the conductive channel of the compact metal conductive filament model.
7. The construction method according to claim 6, characterized in that: The process tolerance model is expressed as: ; in, represents the power spectrum of the Gaussian autocorrelation function; represents the wave vector and ; represents an imaginary unit; Indicates the total number of material structural units into which the dielectric layer in the compact model of metal conductive filaments is divided; Indicates the distance between the upper and lower material structural units in the dielectric layer; Indicates the length of the line edge roughness profile; Indicates the RMS amplitude of the line edge roughness profile.
8. A time-dependent breakdown SPICE circuit model, characterized in that: The time-dependent breakdown SPICE circuit model is constructed using the construction method described in any one of claims 1 to 7.
9. A method for verifying a time-dependent breakdown SPICE circuit model, characterized in that: The verification method includes: Conducting experiments on the time-dependent breakdown effect of the gate oxide layer and / or the time-dependent breakdown effect of the TSV blind hole to obtain corresponding gate oxide layer experimental data and / or TSV blind hole experimental data; and / or conducting simulations on the breakdown effect of the dielectric layer on the chemical mechanical polishing surface to obtain corresponding simulation data; Using the time-dependent breakdown SPICE circuit model described in claim 8 to characterize the time-dependent breakdown effect of the gate oxide layer and / or the time-dependent breakdown effect of the TSV blind hole under corresponding experimental conditions, and obtaining corresponding gate oxide layer model simulation data and / or TSV blind hole model simulation data; and / or using the time-dependent breakdown SPICE circuit model described in claim 8 to characterize the breakdown effect of the chemical mechanical polishing surface dielectric layer under corresponding simulation conditions, and obtaining corresponding chemical mechanical polishing surface model simulation data; Comparing the gate oxide layer experimental data with the gate oxide layer model simulation data, and / or comparing the TSV blind hole experimental data with the TSV blind hole model simulation data, and / or comparing the simulation data with the chemical mechanical polishing surface model simulation data; The time-dependent breakdown SPICE circuit model is verified based on the comparison results.
10. An application of a time-dependent breakdown SPICE circuit model, characterized in that: The time-dependent breakdown SPICE circuit model described in claim 8 is applied to an equivalent circuit model of a power distribution network and / or a through-silicon via to simulate degradation effects of the circuit system.
Citation Information
Patent Citations
Construction and simulation method of resistive random access memory compact model taking graphene as electrode
CN115376634A
Circuit evaluation method considering TDDB effect
CN116992801A
