SiC wafer surface oxidation behavior simulation method, system and equipment

Abstract

The invention discloses a SiC wafer surface oxidation behavior simulation method, system and device, and relates to the technical field of molecular dynamics simulation, and the method comprises the following steps: initializing a plurality of roughness related parameters in MATLAB, and setting simulation box sizes of a Si surface initial model and a C surface initial model; reading atomic positions in the Si surface initial model and the C surface initial model; circulation is carried out on different roughness related parameters, and a corresponding rough surface is generated in each circulation; cutting atomic positions based on different rough surfaces to obtain a plurality of Si surface roughness models and C surface roughness models, and inputting the Si surface roughness models and the C surface roughness models to LAMMPS; and performing OH deposition on the Si surface roughness models and the C surface roughness models in the LAMMPS, and simulating the surface oxidation process. According to the method, by introducing the random roughness, the rough morphology of the SiC surface is accurately simulated, and the authenticity and practicability of research are remarkably improved.

Description

Technical Field

[0001] This invention relates to the field of molecular dynamics simulation technology, and in particular to a method, system and device for simulating the oxidation behavior of SiC wafer surfaces. Background Technology

[0002] In the field of semiconductor manufacturing, semiconductor materials are the cornerstone of modern electronic technology, widely used in various industries from telecommunications to renewable energy, driving the development of high-performance devices. The processing of these materials, especially the control of surface properties and oxidation behavior, is crucial to device performance.

[0003] SiC, with its excellent physical and chemical properties such as high thermal conductivity and wide bandgap, has become a typical representative in research. The two main crystal planes of SiC—the carbon plane (C plane) and the silicon plane (Si plane)—exhibit significantly different behaviors during oxidation due to differences in atomic arrangement and chemical bonding. Existing research shows that the C plane, dominated by carbon atoms, generally has a higher oxidation rate than the Si plane, possibly due to its higher surface chemical activity; while the Si plane, dominated by silicon atoms, undergoes a relatively slower oxidation process, possibly because exposed silicon atoms easily form a protective oxide layer, slowing down subsequent reactions.

[0004] Current research primarily focuses on the oxidation behavior of ideally smooth SiC surfaces to reveal the reaction mechanisms and kinetics of the oxidation process. For example, in thermal oxidation experiments, studies have shown that the C-face reacts with oxygen at high temperatures to generate volatile CO or CO2, while silicon atoms simultaneously form SiO2, leading to rapid oxide layer growth. Conversely, the Si-face exhibits a slower oxidation rate due to the denser SiO2 layer formed in the early stages of oxidation, which restricts oxygen diffusion. Furthermore, molecular dynamics (MD) simulations are widely used in the oxidation of smooth surfaces. By constructing an idealized SiC surface model (assuming the C-face is entirely composed of carbon atoms and the Si-face is entirely composed of silicon atoms), researchers simulated the reaction process between oxidants (such as O2 or H2O2) and the surface, analyzing the formation of silicon-oxygen bonds (Si-O bonds), the change in oxide layer thickness over time, and the formation of interface defects. These studies indicate that on smooth surfaces, the formation rate of Si-O bonds is controlled by the surface atom type and oxidation conditions. The C-face exhibits a faster SiO2 growth rate due to its higher chemical reactivity, while the Si-face exhibits slower oxidation behavior due to the protective layer effect.

[0005] These studies based on smooth surfaces have laid the theoretical foundation for understanding the oxidation process of SiC and have guided process development to some extent. However, due to the limitations of manufacturing processes (such as crystal growth, cutting, and polishing), the C and Si surfaces of actual SiC wafers contain atomic mixing and microscopic irregularities, and are not ideally smooth surfaces. Therefore, traditional methods deviate from actual results when simulating oxidation behavior, limiting the accurate understanding of the oxidation process on real surfaces. Summary of the Invention

[0006] Based on the shortcomings of the existing technology, the present invention provides a method, system and device for simulating the oxidation behavior of SiC wafer surface. It solves the problem that the actual SiC wafer surface is not an ideal smooth surface because the C and Si surfaces contain atomic mixing and microscopic irregularities due to the limitations of the manufacturing process. Therefore, the traditional method deviates from the actual results when simulating oxidation behavior, which limits the accurate understanding of the oxidation process of the real surface.

[0007] The present invention adopts the following technical solution: In a first aspect, the present invention provides a method for simulating the oxidation behavior of a SiC wafer surface, comprising the following steps: Obtain the initial models of the Si surface and C surface in the SiC wafer and input them into MATLAB software; Initialize multiple roughness-related parameters in MATLAB software, and set the simulation box size of the initial Si-plane model and the initial C-plane model; read the atomic positions in the initial Si-plane model and the initial C-plane model; The process iterates through different roughness-related parameters, generating a corresponding rough surface in each iteration. Specifically, for the current roughness-related parameters, a two-dimensional coordinate grid is generated to cover the surface area of ​​the simulation box, and the roughness function matrix is ​​initialized. The current roughness-related parameters are traversed through a double loop to obtain the roughness values ​​of different grid points. Multiple roughness values ​​are converted into actual surface heights and stored in the roughness function matrix to obtain the rough surface. Based on the different roughness of the surface, the atomic positions are tailored to obtain multiple Si surface roughness models and C surface roughness models, which are then input into the LAMMPS software. In LAMMPS software, multiple Si surface roughness models and C surface roughness models were subjected to ∙OH deposition to simulate the surface oxidation process.

[0008] Preferably, the SiC wafer is a 6H-SiC wafer, and the process of obtaining the initial model of the Si surface and the initial model of the C surface in the SiC wafer includes the following steps: Download 6H-SiC unit cells from the Materials Project lattice library and orthogonalize them; In LAMMPS, the orthogonalized 6H-SIC data file is read using the read_data command, and the cell is expanded using the replicate command to obtain the initial model of the Si surface. Rotate the initial model of the Si surface by 180 degrees to obtain the initial model of the C surface.

[0009] Preferably, the roughness-related parameters include roughness amplitude parameters, wavelength range, scale factor, fractal dimension, and azimuth wavenumber; the cell structure file includes the atomic positions and types of the cell.

[0010] Preferably, the step of obtaining the roughness values ​​of different grid points by iterating through the current roughness-related parameters using a double loop specifically includes the following steps: The maximum frequency component and scaling constant are calculated using wavelength range, scale factor, and azimuth wavenumber. The double loop consists of an outer loop and an inner loop. The outer loop iterates through the azimuth wavenumber, while the inner loop iterates through the maximum frequency component. The fractal components are superimposed through the double loop, and the roughness value is calculated through the roughness function. The roughness function is as follows: ; In the formula, This represents the roughness value. M This refers to the azimuth wavenumber. i For the first i Each azimuth wave number j For the first j The largest frequency component, For the maximum frequency component, As a scale factor, D S For fractal dimension, For random phase, L max For the maximum wavelength, Radial distance, For polar coordinates, This refers to the azimuth offset.

[0011] Preferably, multiple roughness values ​​are converted into actual surface height using the following formula: ; In the formula, h (𝑥,𝑦) represents the actual height at each location. L z Let be the total length of the box in the Z direction.

[0012] Preferably, the step of performing OH deposition on multiple Si surface roughness models and C surface roughness models in LAMMPS software specifically includes the following steps: Set the initial parameters for the simulation in the script file; The roughness models of the Si and C surfaces are extended in the Z direction, and a reflective wall is added in the Z direction. Read the ∙OH molecular template and divide the simulation region, including the fixed layer, the deposition layer and the ∙OH deposition region; Zero-force constraints are applied to the atoms in the fixed layer; random velocities of atoms other than those in the fixed layer are initialized to 300K; NVT temperature control is applied to the atoms in the deposited layer to keep the temperature stable at 300K. A new NVT temperature control was applied to the atoms of the deposited layer, gradually increasing the temperature from 300K to 2000K; NVT temperature control was applied to the deposition layer to keep the temperature stable at 2000K, NVT temperature control was applied to the ∙OH atom group at 300K, and 160 ∙OH molecules were deposited in the deposition region, giving them a downward velocity of -0.01Å / fs, and the movement trajectory of the atoms was recorded.

[0013] Preferably, before generating a two-dimensional coordinate grid to cover the surface area of ​​the simulation box, a random phase matrix needs to be generated to simulate the microscopic irregularities of the actual process.

[0014] In a second aspect, the present invention provides a SiC wafer surface oxidation behavior simulation system, comprising: The acquisition module is used to acquire the initial models of the Si surface and the C surface in the SiC wafer and input them into the MATLAB software. The generation module is used to initialize multiple roughness-related parameters in MATLAB software, and set the simulation box size of the initial Si-plane and initial C-plane models; it also reads the atomic positions in the initial Si-plane and initial C-plane models. The loop module is used to iterate over different roughness-related parameters, generating a corresponding rough surface in each iteration. Specifically, for the current roughness-related parameters, a two-dimensional coordinate grid is generated to cover the surface area of ​​the simulation box, and the roughness function matrix is ​​initialized. The current roughness-related parameters are traversed through a double loop to obtain the roughness values ​​of different grid points. Multiple roughness values ​​are converted into actual surface heights and stored in the roughness function matrix to obtain the rough surface. The trimming module is used to trim atomic positions based on different rough surfaces to obtain multiple Si surface roughness models and C surface roughness models, and then input them into the LAMMPS software; The deposition module is used in LAMMPS software to perform ∙OH deposition on multiple Si and C surface roughness models to simulate the surface oxidation process.

[0015] Thirdly, the present invention provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the above-described method for simulating the oxidation behavior of SiC wafer surfaces.

[0016] Compared with the prior art, the above-mentioned at least one technical solution adopted by the present invention can achieve the following beneficial effects: This invention first obtains initial models of the Si and C surfaces in a SiC wafer and inputs them into MATLAB software. Multiple roughness-related parameters are initialized in MATLAB, and the process is iterated over different roughness-related parameters, generating a corresponding rough surface in each iteration. By introducing random roughness, this invention accurately simulates the rough morphology of the SiC surface, abandoning the idealized assumptions of traditional smooth surfaces, significantly improving the realism and practicality of the research. Then, the random roughness model, consistent with the characteristics of actual wafers, is seamlessly integrated into LAMMPS for ·OH deposition simulation. This method overcomes the limitations of traditional research based on ideal smooth surfaces, revealing the real role of surface irregularities in oxidation behavior. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart of a method for simulating the oxidation behavior of a SiC wafer surface according to the present invention; Figure 2 This is a flowchart of the MATLAB process for generating roughness surfaces according to the present invention. Figure 3 This is a flowchart of the LAMMPS simulation process of the present invention; Figure 4 This is a 3D model of the C and Si surfaces of 6H-SiC without surface roughness. in, Figure 4 (a): 3D model diagram of surface C. Figure 4 (b): Three-dimensional model diagram of the Si surface; Figure 5 The C-face fractal surface profile is generated by MATLAB. in, Figure 5 (a): Fractal surface profile of C-plane when G is 0.5. Figure 5 (b): Fractal surface profile of C-plane when G is 1. Figure 5 (c): The fractal surface profile of surface C when G is 1.5. Figure 5 (d): The fractal surface profile of the C-plane when G is 2; Figure 6 It is a fractal surface profile of the Si surface generated by MATLAB; in, Figure 6 (a): Fractal surface profile of the Si surface when G is 0.5. Figure 6(b): Fractal surface profile of the Si surface when G is 1. Figure 6 (c): The fractal surface profile of the Si surface when G is 1.5. Figure 6 (d): The fractal surface profile of the Si surface when G is 2; Figure 7 The C-surface is a workpiece substrate with different random surface roughness; in, Figure 7 (a): Workpiece substrate when G is 0.5. Figure 7 (b): The workpiece base when G is 1. Figure 7 (c): The workpiece substrate when G is 1.5. Figure 7 (d): The workpiece base when G is 2; Figure 8 It is a solid substrate with Si surfaces having different random surface roughness; in, Figure 8 (a): Solid substrate with G = 0.5 Figure 8 (b): Solid substrate when G is 1 Figure 8 (c): Solid substrate when G is 1.5 Figure 8 (d): solid substrate when G is 2; Figure 9 These are MD simulation plots of random roughness C-surface ∙OH deposition at different times; in, Figure 9 (a): Simulation plot with T = 50 ps. Figure 9 (b): Simulation plot with T = 125 ps. Figure 9 (c): Simulation plot with T = 250 ps; Figure 10 These are MD simulation images of random roughness Si surface ∙OH deposition at different times; in, Figure 10 (a): Simulation plot with T = 50 ps. Figure 10 (b): Simulation plot with T = 125 ps. Figure 10 (c): Simulation diagram with T at 250 ps. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Example 1 Traditional methods for studying SiC surface oxidation are typically based on the assumption of an ideally smooth surface, assuming that the C-plane is entirely composed of carbon atoms and the Si-plane is entirely composed of silicon atoms. Figure 4 However, due to limitations in manufacturing processes (such as crystal growth, cutting, and polishing), actual SiC wafers inevitably possess a certain degree of roughness, leading to irregular atomic arrangements on the surface. Both the C and Si planes contain atomic mixing and microscopic irregularities. This roughness significantly affects the interaction between the oxidant and the surface, thereby altering the formation behavior of silicon-oxygen bonds (Si-O bonds) during oxidation. Traditional methods deviate from experimental results when simulating and predicting oxidation behavior, limiting the accurate understanding of the actual surface oxidation process.

[0021] Based on the above-mentioned shortcomings, referring to Figures 1-3 This invention proposes a method for simulating the oxidation behavior of SiC wafer surfaces. Specifically, it introduces random roughness and combines it with ·OH deposition in the study of SiC wafer surface oxidation, aiming to more realistically simulate the oxidation behavior of actual SiC wafers and overcome the limitations of traditional studies based on ideal smooth surfaces. Taking 6H-SiC as an example, this method first uses MATLAB software to model the random roughness of the C and Si surfaces of 6H-SiC. MATLAB first generates a roughness model, reads the basic structure file of 6H-SiC to obtain lattice information, uses a random number generator to simulate the surface height distribution, and adjusts the atomic positions according to the roughness parameters to construct a roughness morphology that conforms to the actual wafer surface characteristics. The final output is a LAMMPS-recognizable data file. This process abandons the idealized assumptions of traditional studies where the C surface is composed entirely of carbon atoms and the Si surface is composed entirely of silicon atoms, creating a surface model that is closer to experimental conditions. Subsequently, in molecular dynamics (MD) simulations, ·OH molecules are deposited downwards from a certain distance above the 6H-SiC surface at a specific deposition rate and a given system temperature, reacting with surface atoms to form Si-O bonds, which are used to study the oxidation process of the rough surface. The present invention specifically includes the following steps:

[0022] S1: Download the 6H-SiC cell from the Materials Project lattice library, then orthogonalize the 6H-SiC cell using MS software, and finally output the LAMMPS format file.

[0023] S2: Use MS software to orthogonalize a single 6H-SIC unit cell, and then export a data format file that LAMMPS can recognize.

[0024] S3: In LAMMPS, expand the orthogonal 6H-SiC cell to the required size for simulation, and finally output the model's data file.

[0025] In LAMMPS, the read_data command is used to read the orthogonalized 6H-SIC data file, the replicate command is used to expand the cell, and the write_data command can be used to output the 6H-SIC Si surface data file.

[0026] S4: In LAMMPS, rotate the expanded 6H-SiC model by 180 degrees and then output the data file to obtain the 6H-SiC model of the C-face.

[0027] In LAMMPS, the centroid coordinates of all atoms can be obtained using the xcm function. Then, the displace_atoms rotate command can be used to rotate all atoms 180 degrees around the x-axis with the centroid as the rotation center to obtain the data file of the C-plane.

[0028] S5: Read the C-surface data file and Si-surface data file output by LAMMPS using MATLAB. Obtain different roughness values ​​by setting different G values ​​and wavelengths. Then, output the C-surface and Si-surface model data files with roughened surfaces using MATLAB. The output data files can then be used for ∙OH deposition oxidation simulation.

[0029] In MATLAB, roughness parameters and the size of the 6H-SiC crystal simulation box are set. Atomic data from C-plane and Si-plane data files are read. Rough surfaces with different G values ​​(0.5 to 2.0) are generated using fractal characteristic functions and random cosine waves. Surface features are calculated and visualized. Atom positions are then trimmed according to surface height. Finally, the results are output as LAMMPS files for subsequent simulations.

[0030] Reference Figure 2 In MATLAB, the formation of a random roughness surface includes the following steps: initializing parameters → reading the reference lattice structure file → entering the main loop → calculating the scaling factor and maximum wavelength number corresponding to the current G value → determining if it is the first loop → generating a random matrix if yes → reusing the existing random matrix if no → generating a two-dimensional coordinate grid to cover the surface region → initializing the particle size function value to zero → increasing the particle size component through multi-scale loops → scaling the particle size function to the target function value → selecting atomic positions according to the particle function → outputting the LAMMPS simulation data file → proceeding to the calculation of the next particle size parameter G. Generating the random matrix provides a random phase basis for the wafer surface roughness to simulate the microscopic irregularities of the actual process. Reusing this matrix ensures that only the G value is a unique variable when comparing roughness surfaces under different G values, eliminating additional random interference. The specific steps are as follows:

[0031] S51: The user-defined function read_lammps_data reads the atomic positions and types of the basic unit cell from the LAMMPS data file (data files for the C and Si planes), stores them in pos_base (three columns: x, y, z coordinates) and type_base (atomic type) respectively, and calculates the number of atoms in the unit cell.

[0032] S52: Define all parameters related to the generation of rough surfaces, including the range of the roughness parameter G (from 0.5 to 2, in steps of 0.5), the wavelength range (minimum 10 Å, maximum 100 Å), the scale factor (1.5), the fractal dimension (2.8), and the azimuth wavenumber (10). These parameters control the surface roughness and characteristic scale. Simultaneously, set the lattice constant, number of cells (both 1), and simulation cell size of the 6H-SiC crystal (consistent with the lattice constant).

[0033] S53: Use a loop `for G_loop=:0.5:0.5:2` to iterate through the G values ​​(0.5, 1.0, 1.5, 2.0). In each iteration, generate a rough surface for one G value and calculate the key constant of the roughness function, the scaling constant. C and maximum frequency component n max : ; ; In the formula, As a scale factor, M This refers to the azimuth wavenumber. L max For the maximum wavelength, L min It is the minimum wavelength.

[0034] During the first loop, a random phase matrix is ​​generated. To introduce randomness to the surface, the matrix is ​​reused in subsequent cycles to maintain consistency.

[0035] S54: In each loop for the G value, the code first generates a two-dimensional coordinate grid. Definition:

[0036] ; In the formula, The grid step size is in the X direction. The grid step size is in the Y direction. i For integer indices in the X direction, j Integer index in the Y direction. x length To simulate the total length of the box in the X direction, y lengthTo simulate the total length of the box in the Y direction, x i and y i For the first on the grid i Column and number j The X and Y coordinates of the row point.

[0037] The grid step size is 0.1 Å, covering the X and Y directions of the simulation box, providing a spatial coordinate basis for the roughness function calculation. Next, the roughness function matrix fun_rough=zeros(size(x)) is initialized, preparing a blank matrix for the subsequent accumulation of fractal components, ensuring that the calculation process is orderly and corresponds one-to-one with the grid points.

[0038] S55: This step is the core process for generating the rough surface. The code uses a double loop (outer loop ii=1:M iterates through the azimuth wavenumbers, inner loop jj=1:n_max iterates through the frequency components) to superimpose fractal components, calculate the roughness value using a roughness function, and finally multiply it by the scaling constant C. The roughness function is shown below:

[0039] ; Then, the roughness function is converted into the actual surface height: ; In the formula, h (𝑥,𝑦) represents the actual height at each location. L z The total length of the box in the Z direction is given, where the base height is 90% of the simulated box's Z dimension. The atomic screening requirements are matched by a 10x magnification, and the root mean square roughness R_q is calculated to quantify the surface roughness.

[0040] S56: Based on the calculated surface height, the code filters atoms in the crystal to construct a rough surface structure. First, it generates an atomic coordinate matrix atom_coor_matrix, and then iterates through all cells and basic atoms using a quadruple loop to calculate the position of each atom. ; In the formula, For the m-th atom in ( i , j , k The actual coordinates of the cell. Let be the position vector of the m-th unit cell (atom within a unit cell), and let a, b, c be the lattice constants. i , j , k This is the index of the current cell in the X, Y, and Z directions.

[0041] The surface height h corresponding to this location is determined using the interpolation method interp2. If the Z coordinate of an atom is less than or equal to h, the atom is retained and its ID, type, and coordinates are recorded; at the same time, surface atoms (Z coordinates in the range of h-1.0 Å) are extracted to analyze surface features.

[0042] S57: In each loop for G value, the code saves the generated atomic coordinates as a LAMMPS data file (.data). After all loops are completed, it generates and saves a visualization: a 3D surface map (saved as 3D_Surface.png using surf(abs(fun_rough))) showing the three-dimensional morphology of the surface, and a 2D profile (h_surface drawn along the midpoint of the Y direction and saved as 2D_Profile.png) showing the height variation in the X direction. Figures 5-8 As shown.

[0043] S6: In the script file, set the initial parameters for the simulation, including the spatial dimension of the simulation box, the units of the system, the time step, the parameters of the nearest neighbor list, the boundary conditions, and the atom type.

[0044] The spatial dimension was set to 3D, the boundary conditions were set to periodic boundaries in the X and Y directions and free boundaries in the Z direction; the system unit was real, the atom type was set to full, the time step was 0.25fs, and the neighbor and neighbor_modify commands were used to set the generation and update frequency of the nearest neighbor list to ensure fast computation during the simulation.

[0045] S7: Create atomic boxes. Use the read_data command to read the data file obtained in step S5, and use the change_box command to expand the original boxes of the roughened C-plane and Si-plane models to prepare for the subsequent specification of the ∙OH deposition region.

[0046] In LAMMPS, the change_box command is used to extend the roughened C-plane and Si-plane models in the Z-direction for subsequent deposition of ∙OH over 6H-SiC.

[0047] S8: Add a reflective wall in the Z direction to confine the simulation system within the Z-axis range and prevent atoms from escaping in the Z direction.

[0048] S9: Read the ∙OH molecular template using the molecule command and create the ∙OH atom generation region using the region command.

[0049] The ∙OH molecular template is read using the molecule command, and the ∙OH atom generation region is created at a certain distance above 6H-SiC using the region command.

[0050] S10: Assign masses to C, H, O, and Si atoms to ensure accurate simulation of the atomic dynamics and chemical reactions during the deposition of ∙OH radicals on the 6H-SiC surface. Simultaneously, use the region and group commands to group atoms and create fixed and deposited layers, laying the foundation for subsequent operations.

[0051] Configure the force field calculation details using the pair_style reaxff command, assign force field parameters to C, H, O, and Si atoms using the pair_coeff command, and then enable charge balance using the fix qeq / reaxff command to dynamically update the charge.

[0052] S11: Potential function selection, using the Reaxff reactive force field to simulate the interaction between H, O and C, Si; S12: Boundary layer fixation. Use the setforse command to fix the boundary layer and keep it fixed in space to reduce boundary effects. S13: Minimize the energy of all atoms and use the velocity command to generate a Gaussian initial velocity distribution for the moving atoms (i.e., non-fixed layer atoms) of the 6H-SiC substrate.

[0053] The velocity command is used to generate an initial Gaussian distribution of velocities for the non-fixed atoms of the 6H-SiC substrate that conforms to the relaxation temperature, thereby initiating molecular dynamics simulations and providing a thermodynamic starting point for the subsequent NVT relaxation phase.

[0054] S14: The deposited layer is relaxed using the NVT ensemble, and then the temperature is increased using the NVT ensemble and the deposited layer is maintained at the increased temperature in preparation for the subsequent ·OH deposition.

[0055] The relaxation is performed using the NVT ensemble. During the simulation, the temperature of the deposition layer is controlled by the NVT ensemble through the fix command to ensure that the system evolves at the required temperature.

[0056] S15: Given a fixed deposition rate for ∙OH, ∙OH is generated in the deposition area set in step S8 and deposited onto the 6H-SiC surface at the set rate.

[0057] Use the `fix deposit` command to set the number of ∙OH deposits, the deposition interval, and the deposition rate; and use the `mol` keyword in the `fix deposit` command to specify the molecular template to be deposited.

[0058] S16: Record the atomic motion trajectories and post-deposition reaction changes through the output dump file, and perform visualization post-processing using OVITO.

[0059] Example 2 Reference Figure 3 LAMMPS simulations were performed using the C-surface random roughness model file and the Si-surface random roughness model file output by MATLAB, including the following steps: S1: First, set the basic simulation parameters, specifying a three-dimensional space (dimension 3), using real units, and setting the time unit to fs. Simultaneously, define the atom pattern as full, and set the boundary conditions to periodicity in the X and Y directions and a free boundary (ppf) in the Z direction. Furthermore, set the communication cutoff distance to 12 Å using comm_modify cutoff 12 to optimize parallel computing efficiency.

[0060] S2: Read the data files of 6H-SiC surfaces with different G values ​​corresponding to the roughness values ​​output from MATLAB, including atomic coordinates and topological information, and apply a random displacement of 0.1 Å to all atoms to avoid the initial structure being too regular. Then, adjust the Z-axis dimension of the simulation box to 70 Å using the change_box command, and add reflective walls at the upper and lower boundaries of the Z-axis to prevent atoms from escaping.

[0061] S3: The simulation regions were divided and atoms were grouped using the `region` and `group` commands, including the fixed layer, deposition layer, and ∙OH deposition region, and corresponding atom groups were defined for each. Simultaneously, the ∙OH molecular template was read using the `molecule` command to prepare for deposition. For the force field, atomic masses were set, and the ReaxFF force field (`pair_style reaxff` and `pair_coeff**ffield.reax CHO Si`) was used, with charges dynamically updated via `fix qeq / reaxff`. Subsequently, overlapping atoms were deleted, atom IDs were reset, and the initial structure was output to the `data` file using the `write_data` command.

[0062] S4: Neighbor list parameters are set and Newton's third law is enabled to optimize computational efficiency. Then, zero-force constraints are applied to the fixed-layer atom group to ensure the stability of the 6H-SiC substrate in the simulation. Next, energy minimization is performed using the conjugate gradient method to optimize the energy and forces of the initial structure.

[0063] S5: Set the time step to 0.25fs and use the velocity command to initialize a random velocity of 300K for atoms other than those in the fixed layer. Then, apply NVT temperature control to the atoms in the deposited layer to keep the temperature stable at 300K.

[0064] S6: Apply a new NVT temperature control to the atoms of the deposited layer, gradually increasing the temperature from 300K to 2000K. This step simulates the system evolution under high temperature conditions, enabling the 6H-SiC surface to reach the temperature conditions suitable for ∙OH deposition.

[0065] S7: In the final stage, an NVT temperature control of 2000K was applied to the deposited layer, and an NVT temperature control of 300K was applied to the ∙OH atom group. The `fix deposit` command was used to deposit 160 ∙OH molecules in the deposition region, giving them a downward velocity of -0.01 Å / fs (i.e., 1000 m / s). The simulation ran for 1,000,000 steps. The atomic trajectories were recorded in real time using the `dump` command for subsequent analysis of the simulation results using OVITO. The final results are as follows: Figures 9-10 As shown.

[0066] This invention proposes a molecular dynamics-based method for simulating the oxidation of rough SiC C-face and Si-face surfaces. By introducing random roughness, it significantly improves the realism and practicality of the study, and has the following advantages: Overcoming the limitations of smooth surfaces: Traditional SiC oxidation studies are typically based on the assumption of ideally smooth surfaces, assuming that the C-side is entirely composed of carbon atoms and the Si-side is entirely composed of silicon atoms. However, actual wafers exhibit roughness and atomic mixing due to process limitations, leading to distorted simulation results. This invention overcomes this limitation by introducing random roughness, allowing the simulation to more realistically reflect actual oxidation behavior. By abandoning simplistic assumptions, this method provides a more reliable theoretical basis and practical guidance for optimizing semiconductor processing techniques.

[0067] The introduction and impact of roughness: This invention introduces random roughness on the C and Si surfaces of SiC to simulate the surface characteristics of real wafers, revealing the key impact of roughness on the oxidation process. Roughness significantly regulates the number and rate of Si-O bond formation by increasing surface area and altering reaction sites, thereby affecting SiO2 growth and material removal rate. Compared to traditional smoothing models, this method more accurately reflects the microscopic oxidation mechanism, enhancing the practicality of the research.

[0068] Effective Interaction Between MATLAB and MD: This invention utilizes MATLAB to generate a stochastic roughness model that conforms to experimental characteristics, precisely controls the surface morphology through fractal properties and stochastic cosine waves, and seamlessly integrates it into LAMMPS for MD simulation. Combined with the ∙OH deposition process, this interactive approach systematically analyzes the oxidation behavior of rough surfaces, improving modeling flexibility and simulation accuracy. This efficient collaboration significantly enhances research efficiency and provides a powerful tool for a deeper understanding of oxidation processes.

[0069] Applicable to other semiconductor and alloy materials: The modeling and simulation process of this method is not only applicable to the study of C-face and Si-face oxidation of SiC, but can also be extended to other semiconductor materials, such as gallium nitride (GaN) or indium phosphide (InP), and other SiC materials such as 3C-SiC or 4H-SiC, as well as the surface oxidation analysis of alloy materials. Its versatility provides universal technical support for the fields of semiconductor processing and materials science, demonstrating broad application prospects and scientific value.

[0070] This invention can also be extended to the simulation of polishing processes with different roughnesses. Since the core of the polishing process is to change the surface roughness of the material through the action of abrasives, this invention, by introducing roughness parameters, can simulate surface morphology changes under various polishing processes and track surface evolution at the atomic scale through molecular dynamics simulations. Therefore, this method can be applied to simulate the dynamic change process of semiconductor material surface roughness from its initial state to the target finish under different polishing conditions (such as abrasive type, polishing pressure, relative motion speed, etc.), analyzing the atomic removal mechanism, the generation and repair laws of surface defects, and the impact of roughness evolution on the surface mechanical properties of the material. This provides theoretical support for optimizing polishing process parameters and improving the surface processing accuracy of semiconductor materials.

[0071] This invention provides a means to analyze the influence of rough surfaces on the quantity and rate of Si-O bond formation by combining a roughness model generated by MATLAB with MD simulations. During the study, the deposition process of ∙OH simulated the actual interaction between the oxidant and the surface, revealing how surface irregularities affect the formation of Si-O bonds and the growth of SiO2. Compared with traditional smooth surface studies, this method more closely approximates the roughness of actual wafer surfaces, providing a deeper understanding of the microscopic mechanisms of SiC surface oxidation behavior.

[0072] Example 3 Based on the same concept, the present invention also provides a SiC wafer surface oxidation behavior simulation system, including an acquisition module, a generation module, a circulation module, a trimming module and a deposition module.

[0073] The acquisition module is used to acquire the initial models of the Si surface and the C surface in the SiC wafer and input them into the MATLAB software.

[0074] The generation module is used to initialize multiple roughness-related parameters in MATLAB software and set the simulation box size of the initial Si-plane and initial C-plane models; it also reads the atomic positions in the initial Si-plane and initial C-plane models.

[0075] The loop module is used to iterate over different roughness-related parameters, generating a corresponding rough surface in each iteration. Specifically, for the current roughness-related parameters, a two-dimensional coordinate grid is generated to cover the surface area of ​​the simulation box, and the roughness function matrix is ​​initialized. The current roughness-related parameters are traversed through a double loop to obtain the roughness values ​​of different grid points. Multiple roughness values ​​are converted into actual surface heights and stored in the roughness function matrix to obtain the rough surface.

[0076] The trimming module is used to trim atomic positions based on different rough surfaces to obtain multiple Si surface roughness models and C surface roughness models, which are then input into the LAMMPS software.

[0077] The deposition module is used in LAMMPS software to perform ∙OH deposition on multiple Si and C surface roughness models to simulate the surface oxidation process.

[0078] Example 4 The present invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the above-described method for simulating the oxidation behavior of SiC wafer surfaces.

[0079] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0080] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for simulating the oxidation behavior of SiC wafer surfaces, characterized in that, Includes the following steps: Obtain the initial models of the Si surface and C surface in the SiC wafer and input them into MATLAB software; In MATLAB software, initialize multiple roughness-related parameters and set the simulation box size of the initial Si surface model and the initial C surface model; Read the atomic positions in the initial models of the Si and C planes; The process iterates through different roughness-related parameters, generating a corresponding rough surface in each iteration. Specifically, for the current roughness-related parameters, a two-dimensional coordinate grid is generated to cover the surface area of ​​the simulation box, and the roughness function matrix is ​​initialized. The current roughness-related parameters are traversed through a double loop to obtain the roughness values ​​of different grid points. Multiple roughness values ​​are converted into actual surface heights and stored in the roughness function matrix to obtain the rough surface. Based on the different roughness of the surface, the atomic positions are tailored to obtain multiple Si surface roughness models and C surface roughness models, which are then input into the LAMMPS software. In LAMMPS software, multiple Si surface roughness models and C surface roughness models were subjected to ∙OH deposition to simulate the surface oxidation process.

2. The method for simulating the oxidation behavior of a SiC wafer surface as described in claim 1, characterized in that, The SiC wafer is a 6H-SiC wafer. The process of obtaining the initial Si surface model and the initial C surface model of the SiC wafer includes the following steps: Download 6H-SiC unit cells from the Materials Project lattice library and orthogonalize them; In LAMMPS, the orthogonalized 6H-SIC data file is read using the read_data command, and the cell is expanded using the replicate command to obtain the initial model of the Si surface. Rotate the initial model of the Si surface by 180 degrees to obtain the initial model of the C surface.

3. The method for simulating the oxidation behavior of a SiC wafer surface as described in claim 1, characterized in that, The roughness-related parameters include roughness amplitude parameters, wavelength range, scale factor, fractal dimension, and azimuth wavenumber; the cell structure file includes the atomic positions and types of the cell.

4. The method for simulating the oxidation behavior of a SiC wafer surface as described in claim 3, characterized in that, The process of obtaining roughness values ​​for different grid points by iterating through the current roughness-related parameters using a double loop specifically includes the following steps: The maximum frequency component and scaling constant are calculated using wavelength range, scale factor, and azimuth wavenumber. The double loop consists of an outer loop and an inner loop. The outer loop iterates through the azimuth wavenumber, while the inner loop iterates through the maximum frequency component. The fractal components are superimposed through the double loop, and the roughness value is calculated through the roughness function. The roughness function is as follows: ； In the formula, This represents the roughness value. M The azimuth wave number is... i For the first i Each azimuth wave number j For the first j The largest frequency component, For the maximum frequency component, As a scale factor, D S For fractal dimension, For random phase, L max For the maximum wavelength, Radial distance, For polar coordinates, This refers to the azimuth offset.

5. The method for simulating the oxidation behavior of a SiC wafer surface as described in claim 4, characterized in that, Multiple roughness values ​​are converted into actual surface height using the following formula: ； In the formula, h (𝑥,𝑦) represents the actual height at each location. L z Let be the total length of the box in the Z direction.

6. The method for simulating the oxidation behavior of a SiC wafer surface as described in claim 1, characterized in that, The step of performing OH deposition on multiple Si and C surface roughness models in LAMMPS software specifically includes the following steps: Set the initial parameters for the simulation in the script file; The roughness models of the Si and C surfaces are extended in the Z direction, and a reflective wall is added in the Z direction. Read the ∙OH molecular template and divide the simulation region, including the fixed layer, the deposition layer and the ∙OH deposition region; Zero-force constraints are applied to the atoms in the fixed layer; random velocities of atoms other than those in the fixed layer are initialized to 300K; NVT temperature control is applied to the atoms in the deposited layer to keep the temperature stable at 300K. A new NVT temperature control was applied to the atoms of the deposited layer, gradually increasing the temperature from 300K to 2000K; NVT temperature control was applied to the deposition layer to keep the temperature stable at 2000K, NVT temperature control was applied to the ∙OH atom group at 300K, and 160 ∙OH molecules were deposited in the deposition region, giving them a downward velocity of -0.01Å / fs, and the movement trajectory of the atoms was recorded.

7. The method for simulating the oxidation behavior of a SiC wafer surface as described in claim 1, characterized in that, Before generating a two-dimensional coordinate grid to cover the surface area of ​​the simulation box, a random phase matrix needs to be generated to simulate the microscopic irregularities of the actual process.

8. A system for simulating the oxidation behavior of a SiC wafer surface, characterized in that, include: The acquisition module is used to acquire the initial models of the Si surface and the C surface in the SiC wafer and input them into the MATLAB software. The generation module is used to initialize multiple roughness-related parameters in MATLAB software and set the simulation box size of the initial Si surface model and the initial C surface model; Read the atomic positions in the initial models of the Si and C planes; The loop module is used to iterate over different roughness-related parameters, generating a corresponding rough surface in each iteration. Specifically, for the current roughness-related parameters, a two-dimensional coordinate grid is generated to cover the surface area of ​​the simulation box, and the roughness function matrix is ​​initialized. The current roughness-related parameters are traversed through a double loop to obtain the roughness values ​​of different grid points. Multiple roughness values ​​are converted into actual surface heights and stored in the roughness function matrix to obtain the rough surface. The trimming module is used to trim atomic positions based on different rough surfaces to obtain multiple Si surface roughness models and C surface roughness models, and then input them into the LAMMPS software; The deposition module is used in LAMMPS software to perform ∙OH deposition on multiple Si and C surface roughness models to simulate the surface oxidation process.

9. A computer device, characterized in that, The method includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the SiC wafer surface oxidation behavior simulation method according to any one of claims 1-7.

Images