Optimization design method for realizing enhanced heat conduction of typical heterogeneous interface of GaN power chip by utilizing simulated ion bombardment

Abstract

The invention discloses an analog simulation optimization design method for improving thermal conductivity of an aluminum nitride / graphene / silicon carbide interface based on ion bombardment of graphene. Comprising the following steps: constructing a layered aluminum nitride / graphene / silicon carbide interface model; extracting a graphene model; simulating ion beams to bombard graphene based on molecular dynamics; performing simulated annealing treatment on the bombarded graphene; obtaining graphene with different defect states by adopting various bombardment conditions (such as bombardment dose, energy, distance and the like); reconstructing an aluminum nitride / defect-state graphene / silicon carbide structure model; calculating interface thermal conductance and analyzing an interface phonon state; establishing a correlation among the ion bombardment process, the interface thermal conductivity and the interface phonon state; the optimization direction of the ion bombardment process is explored; and the interface thermal conductivity is improved by two or more times through iterative optimization.

Description

Technical Field

[0001] The present invention relates to GaN power chips, and specifically to an optimized design method for strengthening heat dissipation in a typical heterostructure. Background Art

[0002] GaN power chips are the core components of the new generation of information technology. However, the heat dissipation problem caused by high integration, high frequency, and high power density is a major challenge in its development process. As the integration, frequency, and power density increase, the heat generated by the chip during operation also rises significantly. If effective heat dissipation cannot be achieved, it will directly affect the performance and reliability of the chip. Therefore, how to optimize the heat dissipation design, reduce the thermal resistance, and improve the performance of the heat dissipation material has become the focus of attention of researchers.

[0003] Interface heat transfer plays a crucial role in the heat dissipation of GaN power chips. Due to the multi-layer structure characteristics inside the GaN chip, the interface thermal resistance directly affects the effective conduction and dissipation of heat, thereby affecting its performance and reliability. Therefore, only by ensuring good interface heat transfer can the advantages of GaN power chips be fully utilized, and its stability and long life in the application of the new generation of information technology be ensured. Summary of the Invention

[0004] The purpose of the present invention is to solve the problem of difficult heat dissipation of existing GaN power chips, and propose an optimized design method for strengthening heat conduction at the typical GaN heterojunction by simulating ion bombardment of graphene. Defective graphene formed after simulating real ion bombardment of graphene is used, and heat transfer at the aluminum nitride / graphene / silicon carbide heterojunction is strengthened by optimizing and adjusting the bombardment process.

[0005] Technical Solution: The technical solution adopted by the design method of the present invention for simulating ion bombardment of graphene to strengthen heat conduction at the typical GaN heterojunction includes the following steps (summarized according to specific implementation cases):

[0006] (1) Construct an aluminum nitride / graphene / silicon carbide interface structure model;

[0007] (2) Extract the graphene model from the interface model and set the size of the vacuum layer;

[0008] (3) Based on molecular dynamics, simulate the ion beam bombardment process;

[0009] (4) Obtain defective graphene under different bombardment conditions (such as bombardment dose, energy, distance, etc.);

[0010] (5) Perform simulated annealing treatment on the bombarded graphene;

[0011] (6) Reconstruct the aluminum nitride / defective graphene / silicon carbide structure model;

[0012] (7) Calculate the interfacial thermal conductivity of the aluminum nitride / defect-state graphene / silicon carbide heterostructure, explore the correlation between the bombardment process and the interfacial thermal conductivity, and determine the optimal bombardment conditions for improving the interfacial thermal conductivity.

[0013] In the step (1), construct the graphene supercell as a graphene layer. Using the original hexagonal primitive cell of graphene, cut it along the crystal plane (0 0 -1) to obtain a single-layer graphene unit cell. Select the cut single-layer graphene unit cell with its lattice x-y plane being (1 1) and y-z plane being (-1,1) to obtain a rectangular graphene unit cell of 2.46*4.26, and perform periodic expansion on it to obtain the graphene supercell.

[0014] In the step (1), construct the silicon carbide supercell as a silicon carbide layer. Construct a 3C-SiC unit cell with the space group of F-423m, a = b = c = 4.35 Å, and α = β = γ = 90°. Perform periodic expansion on the 3C-SiC unit cell to obtain the silicon carbide supercell.

[0015] In the step (1), construct a wurtzite-AlN unit cell with the space group of F-423m, a = b = c = 4.37 Å, and α = β = γ = 90°. Perform periodic expansion on the wurtzite-AlN unit cell to obtain the aluminum nitride supercell.

[0016] In the step (1), the distance between the silicon carbide layer and the graphene layer is 3.25 Å, the distance between the aluminum nitride layer and the graphene layer is 3.4 Å, and the lattice mismatch between the interfaces is less than 3.0%.

[0017] In the step (2), set the thickness of the vacuum layer above the graphene to ensure sufficient bombardment distance, and the thickness of the vacuum layer below the graphene should not be too small to prevent some atoms of the graphene from exceeding the boundary.

[0018] In the step (3), in order to avoid the initial interaction between the bombarding atoms and the graphene layer, the bombarding atoms are randomly generated in the region 50 Å directly above the graphene layer. The time interval between the generation of two adjacent atoms needs to ensure that their distance in the direction perpendicular to the graphene plane is greater than 20 Å, so that they will not affect each other and change their movement directions during the bombardment process. The bombardment interval can be set by the "deposit" command.

[0019] In the step (3), in order to avoid contamination caused by selecting other particles, carbon ions should be selected. Also, since the electron movement speed is much faster than that of the atomic nucleus, charge neutralization occurs first during ion bombardment, and the influence of charge can be ignored. Therefore, the final bombarding particle is a carbon atom.

[0020] In step (3), an appropriate time step is selected. Periodic boundary conditions are adopted in the x and y directions of the system, and fixed boundary conditions are adopted in the z direction. The CH-AIREBO potential is used between atoms in graphene, and the Tersoff-Erhet-ZBL potential is used between the bombarding atoms and graphene. The "velocity" command is used to set the initial temperature of the system.

[0021] In step (3), the NVE ensemble is adopted for all atoms in the system, and the bombardment simulation is carried out using the "deposit" command. The bombardment conditions are controlled by setting the bombardment quantity and bombardment velocity.

[0022] In step (5), first, the bombarded graphene is heated to 2000 K using the NVT ensemble, then held at 2000 K for a period of time, and finally the temperature is slowly decreased to 300 K.

[0023] In step (6), the coordinate file is modified according to the positional relationship of atomic coordinates. The intrinsic graphene in the constructed aluminum nitride / graphene / silicon carbide heterostructure is replaced with defective graphene, and the minimum distance between aluminum nitride and graphene and the minimum distance between silicon carbide and graphene are still kept within the range of van der Waals forces.

[0024] In step (7), molecular dynamics is used to calculate the interfacial thermal conductance, which is a commonly used method for calculating interfacial thermal conductance. The dimension of the system is three-dimensional. Periodic boundary conditions are adopted in the x and y directions of the simulation space, and free boundary conditions are adopted in the z direction. The time step is taken as 0.5 fs. The atoms at the bottom of the model are fixed, and the "velocity" command is used to endow the system with the initial temperature. First, an NPT relaxation of one million time steps is carried out at the initial temperature, and then an NVE relaxation is carried out on the system. After the relaxation is completed, aluminum nitride is slowly heated using the NVT ensemble, and during this process, the NVT ensemble is used to keep graphene and silicon carbide at the initial temperature. The temperature of aluminum nitride after heating should be higher than that of silicon carbide (about 100 K). After aluminum nitride is heated, the NVT ensemble is used to keep aluminum nitride at the heated temperature, and graphene and silicon carbide are kept at the initial temperature for relaxation. After the relaxation is completed, the whole system is placed under the NVE ensemble. During the process of the system temperature reaching equilibrium, the temperatures of aluminum nitride and silicon carbide are recorded, and at the same time, the energy of aluminum nitride is recorded, and the average is taken every 500 time steps. According to the definition of interfacial thermal conductance , the interfacial thermal conductance is obtained by using the relationship between the energy of aluminum nitride and the integral of the interfacial temperature difference with respect to time. The optimal process conditions are selected according to the change of interfacial thermal conductance under various bombardment conditions.

[0025] Principle of the invention: The actual ion bombardment process can be regarded as atoms hitting the target material with a certain kinetic energy. Based on the classical molecular dynamics of Newton's second law, by setting appropriate parameters to control the velocity, quantity, and deposition area of the deposited carbon atoms, the process of simulating carbon ion bombardment of graphene can be achieved. At the same time, by selecting appropriate potential parameters, the breaking and formation of bonds between atoms during the bombardment process can be simulated. In molecular dynamics simulation, at high temperatures, particles move at a relatively high speed, allowing the system to overcome local energy minima and thus explore a larger phase space. As the simulation progresses, the temperature of the system is gradually decreased so that the system energy can gradually converge to a lower energy state, thereby realizing simulated annealing. By performing ion bombardment and annealing treatment on graphene and calculating the interfacial thermal conductance based on classical molecular dynamics, the ion bombardment process is optimized to improve the interfacial thermal conductance, achieving more than twice the improvement in interfacial thermal conductance.

[0026] Advantages: (1) Greatly enhance the interfacial thermal conductance: By introducing defects in graphene, the interfacial thermal conductance can be significantly improved, reaching more than twice the original performance, thereby improving the thermal management performance and enhancing the heat dissipation capacity of the device. (2) Strong adjustability: This technical solution allows different degrees of defect control to be achieved by adjusting the parameters of ion bombardment (such as energy, bombardment dose, etc.), so as to flexibly adjust the interfacial thermal conductance performance to meet different application requirements. (3) Optimization of molecular dynamics simulation: Using molecular dynamics simulation technology for ion bombardment can precisely control the type and quantity of defects, realize the optimization of material properties, and reduce experimental costs and time. Description of the drawings

[0027] Figure 1 It is a flowchart for simulating ion bombardment to enhance heat conduction at the heterogeneous interface

[0028] Figure 2 It is a flowchart of the optimization design method

[0029] Figure 3 It is a model diagram of AlN / Gr / SiC

[0030] Figure 4 It is a diagram of defective graphene obtained at 0, 2, 10, 16, 24, and 40 bombardment dose units with 10*13 / cm2 as the bombardment dose unit

[0031] Figure 5 It is a diagram of the calculation method for the interfacial thermal conductance of the heterogeneous structure

[0032] Figure 6 It is a diagram of the relationship between the evolution of energy with time and the integral of temperature difference with time

[0033] Figure 7 It is a result diagram of the interfacial thermal conductance of the heterogeneous structure Detailed implementation manners

[0034] The present invention will be further described in detail below in conjunction with embodiments.

[0035] Embodiment 1

[0036] In this embodiment, the molecular dynamics simulation method is used to perform atomic bombardment simulation and calculation on graphene.

[0037] Using modeling software, the original hexagonal primitive cell of graphene is selected and cut along the crystal plane (0 0 -1) to obtain a single-layer graphene unit cell. The selected single-layer graphene unit cell after cutting has a lattice x-y plane of (1 1) and a y-z plane of (-1,1), resulting in a rectangular graphene supercell of 2.46 * 4.26, which is extended to a unit cell of 32 * 20 * 1.

[0038] A 3C-SiC unit cell is constructed with a space group of F-423m, a = b = c = 4.35 Å, and α = β = γ = 90°. The atomic position 4a is element C with coordinates (x y z) = (0 0 0). The 3C-SiC unit cell is extended to a supercell of 20 * 20 * 10.

[0039] A wurtzite-AlN unit cell is constructed with a space group of F-423m, a = b = c = 4.37 Å, and α = β = γ = 90°. The atomic position 4a is element Al (x,y,z) = (0 0 0), and the 4d element is N (x,y,z) = (1 / 4,1 / 4,3 / 4). The wurtzite-AlN unit cell is extended to a supercell of 16 * 28 * 7.

[0040] Using three supercells, an aluminum nitride / graphene / silicon carbide interface structure is further constructed with a mismatch degree between interfaces less than 3.0%. Among them, the distance between the silicon carbide layer and the graphene layer is 3.25 Å, and the distance between the aluminum nitride layer and the graphene layer is 3.4 Å, ensuring that the interaction between interfaces is van der Waals force.

[0041] Using the msi2lmp program in the lammps software package, the heterojunction interface structure is converted into a coordinate file readable by lammps, visualized using Ovito software, and the graphene in the interface is separately extracted according to the atomic type to obtain a graphene model.

[0042] Modify the size of the vacuum layer of the graphene model. The thickness above the graphene is modified to 60 Å, and the thickness below the graphene layer is modified to 20 Å. Additionally, an atomic type is added as the bombardment atom type to modify the size of the vacuum layer of the graphene model. The thickness above the graphene is modified to 60 Å, and the thickness below the graphene layer is modified to 20 Å. Additionally, an atomic type is added as the bombardment atom type.

[0043] Set the initial simulation parameters. The system has a dimension of 3D. Periodic boundary conditions are applied in the x and y directions of the simulation space, and fixed boundary conditions are used in the z direction. The measurement unit is metal, and the time step is set to 0.1 fs. To prevent graphene from moving outside the simulation space during bombardment, the atoms on both sides of the single-layer graphene are fixed.

[0044] Select the potential function that describes the interaction between atoms. The CH-airebo potential is used between graphene atoms, and the Tersoff / ZBL potential is used between the bombarding atoms and graphene. The Tersoff / ZBL potential can be used not only to simulate long-range covalent interactions but also to simulate the repulsion during the short-range interaction of particles.

[0045] Set the region for generating bombarding atoms. The bombardment distance can be controlled by setting its z value. To avoid initial interaction between the bombarding particles and graphene, the bombardment distance is set to 50 Å.

[0046] Set the temperature of graphene to 300 K through the velocity command.

[0047] Apply the NVE ensemble to all atoms and use the "deposit" command to perform the bombardment simulation. By setting the number of bombards to 171 (i.e., the bombardment dose is 1*10 14 / cm 2 ), the bombardment velocity is 566.86 (i.e., the bombardment energy is 200 eV), and the bombardment time interval is 400 time steps (ensuring that the distance between two adjacent generated atoms is greater than 20 Å on their way to graphene, so that they will not affect each other's movement direction during bombardment, thus ensuring that all atoms play a role in bombardment).

Claims

1. A simulation method for improving the thermal conductivity of the aluminum nitride / graphene / silicon carbide heterointerface, characterized in that: Through the "deposit" deposition command in LAMMPS, a specific number of carbon atoms can be deposited into the target area at a set speed. The actual ion bombardment process can be regarded as atoms hitting the target material with a certain kinetic energy. Therefore, by setting appropriate parameters, the speed, quantity, and deposition area of the deposited carbon atoms can be controlled to simulate the process of carbon ion bombardment of graphene. The ion-bombarded graphene is subjected to simulated annealing treatment to obtain defect-state graphene with the lowest energy. Reconstruct the aluminum nitride / defect-state graphene / silicon carbide structure model. Calculate the interfacial thermal conductance and analyze the interfacial phonon states. Establish the correlation between the ion bombardment process, interfacial thermal conductance, and interfacial phonon states. Optimize the bombardment process to achieve more than twice the improvement in interfacial thermal conductance. The method described above includes the following steps: Step 1: Construct a layered aluminum nitride / graphene / silicon carbide interface model; Step 2: Extract the graphene model; Step 3: Based on molecular dynamics, simulate the ion beam bombardment of graphene; Step 4: Perform simulated annealing treatment on the bombarded graphene; Step 5: Obtain different defect-state graphenes using various bombardment conditions (such as bombardment dose, energy, distance, etc.); Step 6: Reconstruct the aluminum nitride / defect-state graphene / silicon carbide structure model; Step 7: Calculate the interfacial thermal conductance, explore the correlation between the bombardment process and interfacial thermal conductance, and optimize the bombardment process.

2. The step 1 described in claim 1 is used to construct a layered aluminum nitride / graphene / silicon carbide interface model, characterized in that: Use modeling software to construct an aluminum nitride / graphene / silicon carbide interface model, requiring a small lattice mismatch (<3.0%) between the three layers in the layered interface and the distance between the interfaces to be within the range of van der Waals forces.

3. The graphene model is extracted according to step 2 of claim 1, characterized in that: Use Ovito software to screen out graphene according to the atomic type and set the appropriate thickness of the graphene vacuum layer. For example, the thickness of the vacuum layer above graphene is greater than 50 Å, and the thickness of the vacuum layer below graphene is greater than 20 Å, which is convenient for constructing subsequent bombardment simulation conditions.

4. The ion beam bombardment of graphene based on molecular dynamics simulation according to step 3 of claim 1, wherein: In the simulation, the dimension of the system is three-dimensional. Periodic boundary conditions are used in the x and y directions of the simulation space, and fixed boundary conditions are used in the z direction. The atoms on both sides of graphene are fixed. Use the "deposit" command to set the bombardment atom type as carbon atoms, set the bombardment speed to control the energy of the bombardment particles, and control the dose by setting the number of bombardment atoms. The bombardment atoms are randomly generated in the area directly above the graphene layer at a sufficient distance. The time interval between the generation of two adjacent atoms needs to ensure that their distance in the direction perpendicular to the graphene plane is large enough to prevent the bombardment atoms from interacting with each other and changing their movement directions during the journey to graphene.

5. The simulated annealing treatment of the bombarded graphene according to step 4 of claim 1, characterized in that: Set an appropriate force field. Under the NPT ensemble, heat the system to a high temperature, then keep it at a certain temperature for a certain time and gradually cool it down slowly until it reaches room temperature. The selection of the high temperature and the cooling rate needs to meet the condition for finding the lowest energy of the defect-state graphene.

6. According to claim 1, different defective graphene is obtained by using a variety of bombardment conditions (such as bombardment dose, energy, distance, etc.) in step 5, and it is characterized in that: Set the parameters in the "deposit" command, such as atomic speed and atomic number, to control the bombardment energy and the magnitude of the bombardment dose, and set multiple groups of bombardments to obtain different defect-state graphenes.

7. The aluminum nitride / defect-state graphene / silicon carbide structural model reconstructed according to step 6 of claim 1, characterized in that: Replace the intrinsic graphene in the constructed aluminum nitride / graphene / silicon carbide heterostructure with defective graphene, and keep the distances between aluminum nitride and graphene and between silicon carbide and graphene still within the range of van der Waals forces.

8. Calculating the interfacial thermal conductance according to step 7 described in claim 1 and optimizing the bombardment process, characterized in that: Molecular dynamics is used to calculate the interfacial thermal conductivity. With appropriate boundary conditions and potential functions, an appropriate time step is selected. The atoms at the bottom of the model are fixed. First, the model is relaxed. After relaxation, aluminum nitride and silicon carbide are heated to the corresponding temperatures respectively to create a certain temperature difference across the interface. After heating, they are relaxed again at different temperatures respectively. After relaxation, the entire system is placed in the NVE ensemble to gradually reach temperature equilibrium. During this process, the temperatures of aluminum nitride and silicon carbide are recorded, and at the same time, the energy of aluminum nitride is recorded. To avoid excessive data noise, an average value is taken every certain number of time steps. According to the definition of the interfacial thermal conductivity , the interfacial thermal conductivity is obtained by using the relationship between the energy of aluminum nitride and the temperature difference across the interface.

Images