An atomic-level simulation method for direct bonding of nano-twinned copper rough surfaces

Abstract

The present invention discloses an atomic-level simulation method for direct bonding of rough surfaces of nano-twinned copper. The interatomic interaction potential function in molecular dynamics is determined by first principles, and a polycrystalline nano-twinned copper supercell model is constructed based on experimental parameters to generate a rough surface with fractal characteristics. Molecular dynamics simulation is used to simulate the bonding and annealing processes to obtain a polycrystalline nano-twinned copper model after bonding, and the performance of the interconnection structure is calculated based on this. The result feedback is used to optimize the material and process parameters of copper-copper direct bonding. The present invention establishes an atomic-level polycrystalline nano-twinned copper model through simulation, derives the optimal performance parameter combination of the interconnection structure, reduces the trial and error cost in testing and production, improves the interface quality of copper-copper direct bonding, and provides guidance for the design and process optimization of high-density packaging.

Description

Technical Field

[0001] The present invention relates to the technical field of semiconductor materials and manufacturing, and in particular to an atomic-level simulation method for direct bonding of rough surfaces of nano-twinned copper. Background Art

[0002] With the advancement of semiconductor technology and the continuous reduction in device size, traditional two-dimensional integrated circuits (2D ICs) are gradually approaching their physical limits in terms of integration and performance. To address this issue, three-dimensional integrated circuit (3D IC) technology has emerged. By vertically stacking multiple chips, it significantly improves device integration and signal transmission efficiency. 3D IC technology relies on high-density interconnections between chips, and vertical interconnection is one of the key technologies. Typically, chip interconnection technologies mainly use copper bumps and through-silicon vias (TSVs), which can meet interconnection requirements within a certain size range. However, as device size continues to decrease, the reduction in interconnect pitch places even more stringent demands on these traditional interconnect technologies. In micron- and even nanometer-scale interconnect structures, the performance of copper interconnect materials becomes a key factor in determining package reliability and electrical performance.

[0003] As device sizes gradually enter the submicron and even nanometer levels, traditional interconnect technologies face increasing challenges in terms of process and reliability. Micro-bump technology suffers from solder that is easily squeezed out during the hot pressing process, resulting in uneven bonding interfaces and potential electrical short circuits, especially at pitches less than 20μm. In addition, the solder itself has poor thermal stability, which can easily lead to reduced reliability of the interconnect layer. Especially in high-density packaging, the use of solder materials also increases manufacturing complexity. To address these issues, direct copper-to-copper bonding technology has gradually attracted attention and has become a key technology for improving electrical performance and packaging reliability in future high-density packaging.

[0004] Direct copper-copper bonding technology utilizes the high electrical and thermal conductivity of copper materials, and through surface treatment and precise hot-pressing control, achieves reliable interconnection between chips without solder. Compared with traditional soldering technology, copper-copper bonding not only reduces the use of materials, but also can further reduce the interconnection pitch to meet packaging requirements of 10μm or even smaller. In recent years, this technology has been mainly implemented in two ways: thermal compression bonding (TCB) and hybrid bonding (HB). TCB technology promotes the diffusion and rearrangement of copper atoms by applying a certain amount of heat and pressure, thereby achieving stable metal bonding; while hybrid bonding combines the advantages of direct copper-copper contact and an electrical insulation layer, making it have a broader prospect in applications that meet both mechanical strength and electrical performance requirements.

[0005] Nano-twinned copper, due to its unique crystal structure, exhibits excellent mechanical and electrical properties, especially in terms of improving strength and hardness while maintaining low resistivity, making it an ideal interconnect material. However, in direct copper-copper bonding, surface roughness and twin structure have a complex impact on the bonding interface. Rough surfaces may lead to incomplete contact between copper atoms, resulting in holes, which in turn affects the effectiveness of electrical connections and mechanical strength. At the same time, existing bonding process research has not yet been able to effectively deal with the problem of nanoscale surface roughness. The lack of atomic-level simulation of its bonding process has led to uncontrollable interface performance, which in turn limits its application in high-density packaging. Summary of the Invention

[0006] The purpose of the present invention is to provide an atomic-level simulation method for direct bonding of nano-twinned copper rough surfaces, which can accurately simulate the influence of nano-twinned copper surface roughness and twin structure on the copper-copper direct bonding interface performance, thereby optimizing the high-density interconnection process and improving the copper-copper direct bonding quality of electronic packaging.

[0007] The solution adopted by the present invention to achieve the purpose is: an atomic-level simulation method for direct bonding of nano-twinned copper rough surfaces, used for molecular dynamics simulation, the method comprising the following steps:

[0008] S1: Construct a bulk structure model of nanotwinned copper, perform first-principles calculations and molecular dynamics simulations on the nanotwinned copper structure, compare and analyze characteristic parameters, and determine the appropriate interatomic interaction potential function;

[0009] S2: Construct a supercell structure model of polycrystalline nano-twinned copper based on the experimentally obtained polycrystalline and twin parameters, and obtain an atomic-level model of polycrystalline nano-twinned copper before bonding by constructing a rough surface with parting characteristics;

[0010] S3: Based on the atomic-level model constructed in step S2 and the interatomic interaction potential function determined in step S1, a molecular dynamics method is used to perform bonding simulation, and key parameters of the model bonding process are monitored in real time. After achieving stable bonding, the model is annealed to obtain the bonded polycrystalline nanotwinned copper model data;

[0011] S4: Based on the polycrystalline nanotwin copper model data obtained in S3 and the interatomic interaction potential function determined in S1, the molecular dynamics method is used to perform molecular dynamics simulation of the interconnect performance of polycrystalline nanotwin copper pillars, and the high-reliability polycrystalline nanotwin copper direct bonding process conditions are determined at the atomic level.

[0012] Preferably, in step S1, the characteristic parameters include: lattice constant, density, thermal expansion coefficient, elastic constant, intrinsic stacking fault energy, Poisson's ratio, bulk elastic modulus, Young's modulus, shear model, radial distribution function, and Bain path.

[0013] Preferably, in step S1, suitable interatomic potential functions include: embedded atomic potential functions, modified embedded interatomic potential functions and machine learning interatomic potential functions.

[0014] Preferably, in step S1, the first-principles calculation used is based on density functional theory, and the functionals include PBE functional and SCAN functional. The molecular dynamics method first uses empirical potential functions for calculation, and the potential functions include EAM potential functions and MEAM potential functions.

[0015] Preferably, in step S1, when the error between the comparison result of the molecular dynamics calculation parameters and the first-principles calculation is low, the empirical potential function is fine-tuned according to the characteristic parameters and selected as the interatomic interaction potential function for subsequent calculations; when the error is high, a machine learning potential function is constructed based on the first-principles calculation using an active learning method and selected as the interatomic interaction potential function for subsequent calculations.

[0016] Preferably, step S2 includes the following sub-steps:

[0017] S2a, combining experimental observations of nano-twinned copper electroplated on pure copper surfaces, obtain twin size and twin density based on the cross-section after focused ion beam treatment, and use these parameters to model the nano-twinned copper unit cell;

[0018] S2b, based on S2a, a polycrystalline nano-twinned copper supercell model is established based on the polycrystalline distribution, grain size and crystal orientation obtained by electron backscatter diffraction analysis technology;

[0019] S2c. Delete a specified number of layers of atoms from the middle of the supercell model established in step S2b to construct a reserved gap before bonding. The entire model is divided into two upper and lower polycrystalline nano-twin copper pillars to be bonded. Based on the Weierstrass-Mandelbrot function, a specified roughness surface with fractal characteristics is constructed on the surfaces to be bonded of the two copper pillars to obtain a polycrystalline nano-twin copper atomic-level model with a rough surface.

[0020] Preferably, step S3 includes the following sub-steps:

[0021] S3a, dividing the atomic-level model constructed in step S2 into a top fixed layer, a top heating layer, a middle relaxation layer, a bottom heating layer, and a bottom fixed layer in sequence;

[0022] S3b, using the interatomic interaction potential function determined in step S1 to perform molecular dynamics simulation, perform overall energy minimization, use a canonical system to apply different temperatures to the top and bottom heating layers to define the temperature gradient during the bonding process, and use a microcanonical system to simulate the heat transfer and expansion of the upper and lower polycrystalline nanotwinned copper pillars during the entire bonding process, so as to promote contact between the upper and lower surfaces and achieve bonding through self-diffusion;

[0023] S3c. During the bonding process, the surface roughness, diffusion coefficient, atomic number density distribution, crystal structure, twin structure, pores, internal stress, temperature distribution and energy are monitored in real time. After achieving stable bonding, the model is annealed to obtain the polycrystalline nanotwinned copper model data after bonding.

[0024] Preferably, in step S4, the interatomic interaction potential function selected in S1 is used to perform molecular dynamics simulation of interconnection performance on the bonded polycrystalline nanotwinned copper model obtained in S3, and the performance simulation performed includes: interconnection structural performance, interconnection thermal performance, fracture mechanical performance and interconnection mechanical performance.

[0025] Preferably, step S4 includes the following sub-steps:

[0026] S4a, based on the stably bonded polycrystalline nanotwinned copper model obtained in step S3, using molecular dynamics to simulate its interconnected structural performance, monitoring its crystal structure changes, stacking fault twinning pathways, dislocation evolution, and grain boundary development;

[0027] S4b, simulating the interconnect thermal performance of the stably bonded polycrystalline nanotwinned copper model obtained in step S3 using molecular dynamics methods to monitor its thermal expansion coefficient, glass transition temperature, thermal conductivity, and specific heat;

[0028] S4c simulates the fracture mechanical properties of the stably bonded polycrystalline nanotwinned copper model obtained in S3 using molecular dynamics methods to monitor its stress-strain response, stress evolution, fracture morphology, and damage evolution;

[0029] S4d uses the molecular dynamics method to simulate the mechanical properties of the interconnection of the stably bonded polycrystalline nanotwinned copper model obtained in S3, monitoring the evolution of pores and porosity, interfacial stress-strain response, and interfacial fracture morphology;

[0030] S4e. Based on the performance monitoring results of steps S4a, S4b, S4c and S4d, optimize the copper-copper direct bonding materials and process parameters to guide the structural modeling of step S2 and the process simulation of step S3, thereby forming a high-reliability polycrystalline nano-twin copper direct bonding process.

[0031] This paper proposes a simulation method for the atomically roughened surface of nanotwinned copper, capable of accurately modeling and simulating the impact of surface roughness on bonding performance during copper-copper direct bonding at multiple scales. By simulating the surface morphology of nanotwinned copper with atomic-level precision and analyzing the mechanism of action of twin structures at the copper-copper interface, this method provides a reliable theoretical basis for the optimized design of copper interconnects. This method can effectively improve the interface interconnect quality of copper-copper direct bonding, reduce void formation, and further enhance the electrical performance and mechanical strength of the interconnect structure, making it suitable for advanced packaging technologies with higher interconnect density.

[0032] The present invention has the following advantages and beneficial effects:

[0033] This study constructs an atomic-level model of polycrystalline nanotwinned copper with a fractal-like roughened surface and uses molecular dynamics simulation to accurately simulate the impact of key nanotwinned copper parameters on the performance of copper-copper direct bonding interfaces. By evaluating the electrical properties, mechanical strength, and atomic diffusion behavior of the interface, the high performance of the bonded structure is ensured, thereby optimizing the copper-copper direct bonding process. This simulation method provides a reliable theoretical basis for the design and optimization of copper interconnect materials in high-density packaging and has broad engineering application prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 A flowchart of an atomic-level simulation method for direct bonding of nano-twinned copper rough surfaces provided by an embodiment of the present invention;

[0035] Figure 2 An atomic-level rough surface modeling and visualization model for an atomic-level simulation method for direct bonding of nano-twinned copper rough surfaces provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0036] For a better understanding of the present invention, the following examples are provided to further illustrate the present invention, but the present invention is not limited to the following examples.

[0037] Figure 1 This is a flow chart of the present invention. The core of the present invention is the establishment of a molecular dynamics model of nano-twinned copper and the calculation of bonding performance, seeking the optimal performance combination to guide the design of high-reliability direct bonding materials and structures. Figure 2 It models and visualizes atomic-level rough surfaces, presenting visual details of structures such as polycrystals, twins, and rough surfaces. In particular, it also presents atomic-level models that process billions of atoms through GPU / CPU-assisted acceleration.

[0038] The present invention is further described in detail below with reference to the accompanying drawings and specific embodiments:

[0039] An atomic-level simulation method for direct bonding of nano-twinned copper rough surfaces, comprising the following steps:

[0040] S1: Construct a bulk structure model of nanotwins, use first-principles calculations and molecular dynamics simulations to simulate their characteristic parameters, and determine the appropriate interatomic interaction potential function after comparing the errors;

[0041] S2: Based on the experimentally obtained data on polycrystalline and twin structures and bonding model parameters such as copper pillar size and surface roughness, an atomic-level model of polycrystalline nano-twinned copper with a rough surface with parting characteristics before bonding is constructed;

[0042] S3: Based on the pre-bonding model constructed in S2 and the interatomic interaction potential function determined in S1, molecular dynamics methods are used to perform bonding simulations, key parameters in the process are monitored in real time, and after determining stable bonding, annealing simulation is performed to obtain a post-bonding polycrystalline nanotwinned copper model;

[0043] S4: Based on the post-direct bonding model determined in S3, molecular dynamics simulation of the interconnect performance of polycrystalline nanotwinned copper pillars is performed using molecular dynamics methods to guide the optimization direction of copper-copper direct bonding materials and process parameters at the atomic level, and determine the high-reliability polycrystalline nanotwinned copper direct bonding process conditions.

[0044] In the above embodiment, the characteristic parameters in step S1 include: lattice constant, density, thermal expansion coefficient, elastic constant, intrinsic stacking fault energy, Poisson's ratio, bulk elastic modulus, Young's modulus, shear model, radial distribution function, and Bain path under various temperature and pressure conditions.

[0045] Furthermore, the molecular dynamics method in step S1 first uses empirical potential functions for calculation, and the potential functions include EAM potential functions and MEAM potential functions; the first-principles calculations used are based on density functional theory, and the functionals include PBE functionals and SCAN functionals.

[0046] In a further preferred embodiment, if the error between the comparison result of the molecular dynamics calculation parameters in step S1 and the first-principles calculation is low, the empirical potential function is fine-tuned according to the characteristic parameters and selected as the interatomic interaction potential function for subsequent calculations; if the error is high, a machine learning potential function (such as the DeePMD potential function) is constructed based on the first-principles calculation using an active learning method and selected as the interatomic interaction potential function for subsequent calculations.

[0047] In step S2, the twin size and twin density are obtained according to the FIB experimental parameters to construct the nano-twin copper unit cell, the polycrystalline distribution, grain size and crystal orientation are obtained according to the EBSD experimental parameters, and the polycrystalline nano-twin copper supercell is constructed based on the nano-twin copper unit cell. The copper column size, twin column height and bonding spacing are set to determine the bonding copper column model, and the rough upper and lower surfaces to be bonded are constructed based on the WM function. The polycrystalline nano-twin copper supercell is expanded by removing the specified atoms to form an atomic-level model of the polycrystalline nano-twin copper with a rough surface to be bonded, such as Figure 2 shown.

[0048] Furthermore, using CPU and GPU parallel processing can accelerate modeling and expand the model to simulations of billions of atoms.

[0049] In step S3, the atomic-level model of polycrystalline nanotwinned copper with a rough surface to be bonded established in S2 is used to perform molecular dynamics simulation of the bonding process. It is necessary to divide the fixed layer, heating layer, and relaxation layer to perform energy minimization calculations. The NVT ensemble is used to set the temperature and temperature gradient in the heating layer, and the bonding time and trajectory result output are defined. The overall model uses the NVE ensemble for bonding simulation, and the annealing simulation is performed under the NVT ensemble.

[0050] Furthermore, the parameters monitored in real time include: surface roughness, diffusion coefficient, atomic number density distribution, crystal structure, twin structure, pores, internal stress distribution, temperature distribution and energy, etc.

[0051] In step S4, the interatomic potential function selected in S1 is used to perform molecular dynamics simulation of interconnection performance on the bonded polycrystalline nanotwinned copper model obtained in S3. The performance simulation includes: interconnection structural performance, interconnection thermal performance, fracture mechanical performance and interconnection mechanical performance.

[0052] Furthermore, the optimized copper-copper direct bonding materials and process parameters are divided into structural parameters and process parameters. The structural parameters are fed back to step S2 to establish an optimized atomic-level model of polycrystalline nano-twinned copper on the rough surface to be bonded, including: twin density, twin size, polycrystalline distribution, polycrystalline density, grain size, crystal orientation, copper column size and surface roughness, etc.

[0053] Furthermore, the optimized copper-copper direct bonding material and process parameters are divided into structural parameters and process parameters. The process parameters are fed back to step S3 for bonding simulation and annealing treatment, including: bonding temperature or temperature gradient, bonding time, bonding spacing, annealing temperature or temperature gradient, annealing time, etc.

[0054] S4a simulates the performance of interconnect structures by performing NPT temperature and pressure control simulation, uniaxial tension and compression, nanoindentation and shear simulation on the overall structure, simulating the structural characteristics of chip interconnect structures under typical application environments, and monitoring their crystal structure changes, stacking fault twinning pathways, dislocation evolution and grain boundary development.

[0055] S4b simulates the thermal performance of interconnects by subjecting the overall structure to XYZ proportional changes at multiple temperatures and pressures, performing multiple sets of long-term constant temperature and pressure simulations in various temperature zones, obtaining the number of frames after the final energy balance, and calculating the average value to obtain its thermal expansion coefficient.

[0056] Furthermore, S4b calculated the glass transition temperature using the solid-liquid coexistence method, established the solid-liquid coexistence interface under specified pressure conditions, obtained the equilibrium temperature-pressure correspondence curve through the NPH ensemble, and accurately obtained its glass transition temperature.

[0057] The simulation of fracture mechanical properties of S4c is achieved by applying uniaxial tension to the whole structure, with a strain rate of less than 5.0×10 7 s -1 , constant temperature simulation at different temperatures, monitoring its stress-strain response, stress evolution, fracture morphology and damage evolution, etc.

[0058] S4d simulates the mechanical properties of interconnects by applying uniaxial tension, shear, and nanoindentation simulations along the Z direction to the overall structure, and performs constant temperature simulations at different temperatures to monitor the evolution of pores and porosity, interface stress-strain response, and interface fracture morphology.

[0059] Furthermore, after evaluating the performance in step S4, S4e provides the optimized structural parameters and process parameters to steps S2 and S3, establishing a coupling relationship between its material structure, bonding process and interconnection performance, and forming a high-reliability polycrystalline nano-twin copper direct bonding process.

[0060] In a further preferred embodiment, the atomic-level modeling, molecular dynamics bonding simulation and interconnection performance simulation of rough surfaces in the present invention are not only applicable to the copper-copper direct bonding process, but also include the mixed bonding process of interconnection dielectric materials and interconnection metals and the direct bonding process using dissimilar metals.

[0061] The above description is only a preferred embodiment of the present invention, which certainly cannot be used to limit the scope of rights of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and changes can be made without departing from the principles of the present invention. These improvements and changes are also considered to be within the scope of protection of the present invention.

Claims

1. An atomic-level simulation method for direct bonding of nano-twinned copper rough surfaces, characterized by: For performing molecular dynamics simulations, the method includes the following steps: S1: Construct a bulk structure model of nanotwinned copper, perform first-principles calculations and molecular dynamics simulations on the nanotwinned copper structure, compare and analyze characteristic parameters, and determine the appropriate interatomic interaction potential function; S2: Construct a supercell structure model of polycrystalline nano-twinned copper based on the experimentally obtained polycrystalline and twin parameters, and obtain an atomic-level model of polycrystalline nano-twinned copper before bonding by constructing a rough surface with parting characteristics; S3: Based on the atomic-level model constructed in step S2 and the interatomic interaction potential function determined in step S1, a molecular dynamics method is used to perform bonding simulation, and key parameters of the model bonding process are monitored in real time. After achieving stable bonding, the model is annealed to obtain the bonded polycrystalline nanotwinned copper model data; S4: Based on the polycrystalline nanotwin copper model data obtained in S3 and the interatomic interaction potential function determined in S1, the molecular dynamics method is used to perform molecular dynamics simulation of the interconnect performance of polycrystalline nanotwin copper pillars, and the high-reliability polycrystalline nanotwin copper direct bonding process conditions are determined at the atomic level.

2. The atomic-level simulation method for direct bonding of nano-twinned copper rough surfaces according to claim 1, characterized in that: In step S1, the characteristic parameters include: lattice constant, density, thermal expansion coefficient, elastic constant, intrinsic stacking fault energy, Poisson's ratio, bulk elastic modulus, Young's modulus, shear model, radial distribution function, and Bain path.

3. The atomic-level simulation method for direct bonding of nano-twinned copper rough surfaces according to claim 1, characterized in that: In step S1, suitable interatomic potential functions include: embedded atomic potential functions, modified embedded interatomic potential functions, and machine learning interatomic potential functions.

4. The atomic-level simulation method for direct bonding of nano-twinned copper rough surfaces according to claim 1, characterized in that: In step S1, the first-principles calculation used is based on density functional theory, and the functionals include PBE functional and SCAN functional. The molecular dynamics method first uses empirical potential functions for calculation, and the potential functions include EAM potential functions and MEAM potential functions.

5. The atomic-level simulation method for direct bonding of nano-twinned copper rough surfaces according to claim 1, characterized in that: In step S1, when the error between the comparison result of the molecular dynamics calculation parameters and the first-principles calculation is low, the empirical potential function is fine-tuned according to the characteristic parameters and selected as the interatomic interaction potential function for subsequent calculations; when the error is high, a machine learning potential function is constructed based on the first-principles calculation using an active learning method and selected as the interatomic interaction potential function for subsequent calculations.

6. The atomic-level simulation method for direct bonding of nano-twinned copper rough surfaces according to claim 1, characterized in that: The step S2 includes the following sub-steps: S2a, combining experimental observations of nano-twinned copper electroplated on pure copper surfaces, obtain twin size and twin density based on the cross-section after focused ion beam treatment, and use these parameters to model the nano-twinned copper unit cell; S2b, based on S2a, a polycrystalline nano-twinned copper supercell model is established based on the polycrystalline distribution, grain size and crystal orientation obtained by electron backscatter diffraction analysis technology; S2c. Delete a specified number of layers of atoms from the middle of the supercell model established in step S2b to construct a reserved gap before bonding. The entire model is divided into two upper and lower polycrystalline nano-twin copper pillars to be bonded. Based on the Weierstrass-Mandelbrot function, a specified roughness surface with fractal characteristics is constructed on the surfaces to be bonded of the two copper pillars to obtain a polycrystalline nano-twin copper atomic-level model with a rough surface.

7. The atomic-level simulation method for direct bonding of nano-twinned copper rough surfaces according to claim 1, characterized in that: The step S3 includes the following sub-steps: S3a, dividing the atomic-level model constructed in step S2 into a top fixed layer, a top heating layer, a middle relaxation layer, a bottom heating layer, and a bottom fixed layer in sequence; S3b, using the interatomic interaction potential function determined in step S1 to perform molecular dynamics simulation, perform overall energy minimization, use a canonical system to apply different temperatures to the top and bottom heating layers to define the temperature gradient during the bonding process, and use a microcanonical system to simulate the heat transfer and expansion of the upper and lower polycrystalline nanotwinned copper pillars during the entire bonding process, so as to promote contact between the upper and lower surfaces and achieve bonding through self-diffusion; S3c. During the bonding process, the surface roughness, diffusion coefficient, atomic number density distribution, crystal structure, twin structure, pores, internal stress, temperature distribution and energy are monitored in real time. After achieving stable bonding, the model is annealed to obtain the polycrystalline nanotwinned copper model data after bonding.

8. The atomic-level simulation method for direct bonding of nano-twinned copper rough surfaces according to claim 1, characterized in that: In step S4, the interatomic interaction potential function selected in S1 is used to perform molecular dynamics simulation of interconnection performance on the bonded polycrystalline nanotwinned copper model obtained in S3, and the performance simulation performed includes: interconnection structural performance, interconnection thermal performance, fracture mechanical performance and interconnection mechanical performance.

9. The atomic-level simulation method for direct bonding of nano-twinned copper rough surfaces according to claim 1, characterized in that: The step S4 includes the following sub-steps: S4a, based on the stably bonded polycrystalline nanotwinned copper model obtained in step S3, using molecular dynamics to simulate its interconnected structural performance, monitoring its crystal structure changes, stacking fault twinning pathways, dislocation evolution, and grain boundary development; S4b, simulating the interconnect thermal performance of the stably bonded polycrystalline nanotwinned copper model obtained in step S3 using molecular dynamics methods to monitor its thermal expansion coefficient, glass transition temperature, thermal conductivity, and specific heat; S4c simulates the fracture mechanical properties of the stably bonded polycrystalline nanotwinned copper model obtained in S3 using molecular dynamics methods to monitor its stress-strain response, stress evolution, fracture morphology, and damage evolution; S4d uses the molecular dynamics method to simulate the mechanical properties of the interconnection of the stably bonded polycrystalline nanotwinned copper model obtained in S3, monitoring the evolution of pores and porosity, interfacial stress-strain response, and interfacial fracture morphology; S4e. Based on the performance monitoring results of steps S4a, S4b, S4c and S4d, optimize the copper-copper direct bonding materials and process parameters to guide the structural modeling of step S2 and the process simulation of step S3, thereby forming a high-reliability polycrystalline nano-twin copper direct bonding process.

Images