High-precision nondestructive detection method and system for ion implantation defects

Abstract

The invention discloses an ion implantation defect high-precision nondestructive testing method and system, and relates to the technical field of nondestructive testing, and the method comprises the following steps: selecting a measurement light source to collect reflected light intensity data, using a Stokes vector to represent intensity distribution of polarized light, calculating ellipsometry parameters, and determining refraction angle data; the complex refractive index is calculated through a Fresnel formula, the refractive index and the extinction coefficient are determined, and the overall complex dielectric constant of the material is analyzed; and performing crystal material region division, obtaining an optimized region complex dielectric constant by using a nonlinear regression method, and constructing a three-layer composite medium model in combination with a region ratio. According to the method, by calculating the Stokes parameter, microscopic nonuniformity in the material can be revealed through numerical analysis, the resolution of a damaged area is improved, and by dividing the crystal material area, the distribution conditions of crystalline silicon, amorphous silicon and strained silicon can be accurately distinguished after ion implantation, and the accuracy of ion implantation is improved. And the analysis capability on the microstructure change of the material is improved.

Description

Technical Field

[0001] The invention relates to the technical field of nondestructive testing, and in particular to a high-precision nondestructive testing method and system for ion implantation defects. Background Art

[0002] As a key process in semiconductor manufacturing and materials engineering, ion implantation technology is widely used in doping regulation, surface modification, and nanostructure preparation. However, ion implantation often introduces defects such as lattice damage, strain, and non-uniform doping, which directly affect the electrical, optical, and mechanical properties of the material. Therefore, high-precision non-destructive detection technology for ion implantation defects has always been an important research direction for semiconductor material analysis and quality control. Existing detection methods mainly include transmission electron microscopy, scanning electron microscopy, Raman spectroscopy, photoluminescence, etc. Spectral ellipsometry (SE) has gradually become an important tool for ion implantation defect detection in recent years due to its non-contact, high-precision, and wide-spectrum range. However, traditional ellipsometry measurement methods mainly rely on measurement modes of a single angle or fixed wavelength. It is difficult to accurately analyze the changes in the complex dielectric constant in the depth direction of the material, and it is impossible to effectively distinguish the defect distribution of different depth layers. Secondly, the existing spectral ellipsometry measurement methods are difficult to accurately analyze the defect distribution of materials at different depths. The complex dielectric constant calculated is usually an overall average value, which cannot directly reflect the hierarchy of the internal structure of the material, resulting in limited accuracy of the calculation results. Summary of the invention

[0003] In view of the above existing problems, the present invention is proposed.

[0004] Therefore, the present invention provides a high-precision nondestructive detection method for ion implantation defects to solve the problem that traditional ellipsometry measurement methods mainly rely on measurement modes of a single angle or a fixed wavelength, and it is difficult to accurately analyze the changes in the complex dielectric constant in the depth direction of the material, and it is impossible to effectively distinguish the defect distribution of different depth layers. Secondly, the existing spectral ellipsometry measurement method is difficult to accurately analyze the defect distribution of the material at different depths, and the calculated complex dielectric constant is usually an overall average value, which cannot directly reflect the hierarchy of the internal structure of the material, resulting in limited accuracy of the calculation results.

[0005] In order to solve the above technical problems, the present invention provides the following technical solutions: In a first aspect, the present invention provides a high-precision nondestructive detection method for ion implantation defects, comprising: Select a measurement light source to collect reflected light intensity data, use the Stokes vector to represent the intensity distribution of polarized light, calculate the ellipsometric parameters and determine the refraction angle data, calculate the complex refractive index through the Fresnel formula and determine the refractive index and extinction coefficient, and analyze the overall complex dielectric constant of the material; Perform regional division of crystal materials, use nonlinear regression method to obtain the complex dielectric constant of the optimized region, build a three-layer composite medium model based on the regional proportion value, optimize and calculate the optimal solution of the regional proportion value through Lagrangian optimization method, perform deep layered Fourier transform, obtain the complex dielectric constant distribution in the depth domain, integrate and calculate the complex dielectric constant distribution data of different layers, and confirm the normalized weight to calculate the material proportion value of different layers; Build a deep learning model based on DNN, use Bayesian optimization method to optimize model parameters, predict material defect distribution, and make prediction matching degree; The prediction data is encrypted and stored securely.

[0006] As a preferred solution of the high-precision nondestructive detection method for ion implantation defects of the present invention, the overall complex dielectric constant of the analyzed material includes: Ion implantation defect detection is performed based on spectral ellipsometry. The intensity of reflected light polarized in parallel p and perpendicular s directions is measured respectively. The circularly polarized light is converted into linearly polarized light by placing a quarter-wave plate QWP in the path of the reflected beam. The polarizer is adjusted in turn to allow right-handed circularly polarized light and left-handed circularly polarized light to pass through. The intensity of reflected light after right-handed and left-handed circularly polarized light is measured. By placing a wavelength modulating wave plate in the beam path, the reflected light intensity data after the polarizer is collected; Use Stokes vector to represent the intensity distribution of polarized light and calculate the amplitude ratio and phase difference of ellipsometric parameters; Determine the reflection coefficient and determine the refraction angle data through Snell's law, and calculate the complex refractive index through Fresnel's formula and determine the refractive index and extinction coefficient; The material's response to the electric field and its light absorption capacity are analyzed and calculated based on the refractive index and extinction coefficient, and the overall complex dielectric constant is calculated.

[0007] As a preferred solution of the high-precision nondestructive detection method for ion implantation defects of the present invention, the method of dividing the crystal material regions, calculating the optimal solution of the regional proportion values, and confirming the normalized weights to calculate the material proportion values ​​of different layers include: During the ion implantation process, the crystal structure of the silicon material is divided into a crystalline silicon region, an amorphous silicon region and a strained silicon region, and the complex dielectric constants corresponding to the crystalline silicon, amorphous silicon and strained silicon are determined based on experimental data, wherein the initial complex dielectric constant values ​​of the crystalline silicon, amorphous silicon and strained silicon at different wavelengths are obtained from a database, and the optimization goal is set to minimize the error between the three complex dielectric constants and the overall complex dielectric constant using a nonlinear regression method, and the optimization iteration is performed using the Newton iteration method until the iteration error no longer changes significantly, then the iteration is stopped and the three complex dielectric constants are determined; A three-layer composite dielectric model is constructed based on the sum of the products of the three complex dielectric constants and the corresponding crystalline silicon, amorphous silicon and strained silicon proportions as the overall complex dielectric constant, where , as well as Respectively represent the proportion of crystalline silicon, amorphous silicon and strained silicon; By using the Lagrangian optimization method, the goal of maximizing the fit of the complex dielectric constant model is set to optimize the calculation , as well as The optimal solution of Performing a deep layered Fourier transform (A-FFT) on the overall complex permittivity to obtain the distribution of the complex permittivity in the depth domain; Based on the complex dielectric constant after Fourier transformation, the data corresponding to short wavelength, medium wavelength and long wavelength are extracted and inverse Fourier transformed to obtain the complex dielectric constant distribution of the surface layer, medium depth layer and deep layer materials respectively, and the normalized weight of each layer is determined according to the integral value of the complex dielectric constant distribution data of different layers; According to the normalized weights of different layers, the Lagrangian optimization method is used to calculate , as well as The product of the optimal proportion values ​​of is used to obtain the material proportion values ​​of different layers, and then normalization verification is performed.

[0008] As a preferred solution of the high-precision nondestructive detection method for ion implantation defects of the present invention, wherein: the deep learning model is constructed based on DNN, the model parameters are optimized using the Bayesian optimization method, and the material defect distribution is predicted, including: A deep learning model is constructed based on DNN, including an input layer, a hidden layer, and an output layer. The complex dielectric constant data of different depth layers and the corresponding material proportion data of each depth layer are combined into a data matrix, which is input into the deep learning model through the input layer. The material defect distribution is predicted through the hidden layer, including the material defect distribution area prediction value and the depth prediction value, and output through the output layer; The Bayesian optimization method is used to adjust the neural network parameters, and the optimization goal is set to maximize the model prediction accuracy. The iteration is stopped when the update changes are no longer obvious, and the optimized model parameters are obtained.

[0009] As a preferred scheme of the high-precision nondestructive detection method for ion implantation defects described in the present invention, the predicted matching degree prediction refers to using calibrated experimental measurement defect distribution data to calculate the matching degree of the predicted material defect distribution data of the deep learning model, and setting an error threshold based on the experimental data. If the matching degree is greater than or equal to the error threshold, the prediction is judged to be accurate.

[0010] As a preferred solution of the high-precision nondestructive detection method for ion implantation defects described in the present invention, the selection of a measurement light source to collect reflected light intensity data refers to selecting a supercontinuum laser source as a measurement light source, using a femtosecond pulse light source, and using a polarization modulator to generate incident light in a specific polarization state, allowing the incident light to pass through an adjustable polarizer to adjust the polarization direction, and collecting reflected light intensity data through a high-sensitivity photodetector.

[0011] As a preferred solution of the high-precision nondestructive detection method for ion implantation defects described in the present invention, the encryption and secure storage of the predicted data refers to encrypting the predicted defect distribution data using the AES-256 encryption standard and encrypting and storing it through KMS security key management.

[0012] In a second aspect, the present invention provides a system for a high-precision nondestructive detection method for ion implantation defects, comprising: Optical data acquisition module, which selects and controls the measurement light source, adjusts the polarization state of the incident light, collects reflected light intensity data, and measures the response of the material to different polarized light; Optical parameter calculation module, which analyzes the polarization information of light, calculates the material's response parameters to light, calculates the ellipsometric parameters, determines the refraction angle data, and then obtains the refractive index and extinction coefficient; The three-layer composite dielectric module divides the crystalline silicon, amorphous silicon and strained silicon regions according to the material measurement data, obtains the complex dielectric constants of different regions through optimization methods, and constructs a composite dielectric model; The material layer analysis module performs deep layer Fourier transform, analyzes the distribution of complex dielectric constants at different depth layers, calculates the normalized weights of different layers, and finally obtains the material proportions of different layers; Deep learning defect prediction module, which uses DNN to predict defect distribution, identify damaged areas in materials, and adjust model parameters through Bayesian optimization; The matching degree verification module calculates the matching degree between the DNN prediction results and the experimental measurement data to determine the reliability of the prediction results; The data security storage module securely stores measurement data, calculation data and prediction data.

[0013] In a third aspect, the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, any step of the high-precision non-destructive detection method for ion implantation defects as described in the first aspect of the present invention is implemented.

[0014] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, any step of the high-precision nondestructive detection method for ion implantation defects as described in the first aspect of the present invention is implemented.

[0015] The beneficial effects of the present invention are as follows: by calculating the Stokes parameters, the microscopic inhomogeneity inside the material can be revealed through numerical analysis, and the resolution of the damaged area can be improved; by dividing the crystal material area, the distribution of crystalline silicon, amorphous silicon and strained silicon can be accurately distinguished after ion implantation, and the ability to analyze the changes in the material microstructure can be improved; by combining spectral ellipsometry measurement with deep layered Fourier transform combined with material area division and optimization calculation, the scheme can realize accurate layered analysis of the material microstructure and defects on the basis of high-precision polarization optical measurement; by calculating the Stokes vector, ellipsometry parameters, and deriving the refractive index and extinction coefficient using the Fresnel formula, the optical response of the material can be quantitatively described and directly related to the changes in the internal structure of the material after ion implantation, thereby realizing non-destructive, high-precision, and deeply resolvable ion implantation defect detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0017] Figure 1 Schematic diagram of the process of high-precision nondestructive detection of ion implantation defects in Example 1.

[0018] Figure 2 Schematic diagram of the structure of the high-precision nondestructive detection system for ion implantation defects in Example 1.

[0019] Figure 3 Schematic diagram of the crystal region division process of the high-precision nondestructive detection method for ion implantation defects in Example 1.

[0020] Figure 4 This is a schematic diagram of the DNN deep learning model structure of the high-precision nondestructive detection method for ion implantation defects in Example 1. DETAILED DESCRIPTION

[0021] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0022] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0023] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0024] Example 1, reference Figures 1 to 4 , which is the first embodiment of the present invention, provides a high-precision nondestructive detection method for ion implantation defects, comprising the following steps: S1, select the measurement light source to collect the reflected light intensity data, use the Stokes vector to represent the intensity distribution of polarized light, calculate the ellipsometric parameters and determine the refraction angle data, calculate the complex refractive index through the Fresnel formula and determine the refractive index and extinction coefficient, and analyze the overall complex dielectric constant of the material; Preferably, selecting a measurement light source to collect reflected light intensity data refers to selecting a supercontinuum laser source (Supercontinuum Laser) as the measurement light source, wherein the spectral range is 190nm-1200nm, a femtosecond pulse light source (<100 fs) is used, and a polarization modulator is used to generate incident light of a specific polarization state, allowing the incident light to pass through an adjustable polarizer to adjust the polarization direction, and a high-sensitivity photodetector is used to collect reflected light intensity data.

[0025] By using a supercontinuum laser source as the measurement light source, compared with the traditional single-wavelength laser source, the optical response of the material at multiple wavelengths can be obtained simultaneously, thereby improving the spectral resolution of defect detection. The use of a femtosecond pulse light source can reduce time resolution errors, reduce the impact of surface roughness and ambient light on measurement, improve the temporal and spatial resolution of the signal, and ensure more accurate measurement of ion implantation defects. By using a polarization modulator to generate incident light in a specific polarization state and adjusting the polarization direction with an adjustable polarizer, the response information of the material to different polarization directions can be fully obtained. By using a high-sensitivity photodetector to collect reflected light intensity data, it can ensure that the material reflection characteristics can still be accurately measured in a low signal-to-noise ratio environment, improve the signal-to-noise ratio of the data, and enhance measurement stability.

[0026] Furthermore, the overall complex dielectric constant of the material is analyzed, including: Ion implantation defect detection is performed based on spectral ellipsometry. The intensity of reflected light polarized in parallel p and perpendicular s directions is measured respectively. The circularly polarized light is converted into linearly polarized light by placing a quarter-wave plate QWP in the path of the reflected beam. The polarizer is adjusted in turn to allow right-handed circularly polarized light and left-handed circularly polarized light to pass through. The intensity of reflected light after right-handed and left-handed circularly polarized light is measured. By placing a wavelength modulating wave plate in the beam path, polarized light at 45° and 135° is selected in turn through an adjustable polarizer, and the reflected light intensity after the 45° and 135° polarizers is collected; The intensity distribution of polarized light is represented by the Stokes vector: ; ; ; ; in represents the total light intensity, represents the linear polarization degree from zero to ninety degrees, Represents the linear polarization degree from 45 to 135 degrees, represents the degree of circular polarization, and They represent the intensity of the reflected light polarized in the parallel (p) and perpendicular (s) directions in the reflected light, and They represent the components with polarization directions of four or five degrees and one hundred and thirty-five degrees respectively. and represent the intensity of right-hand circular polarization and left-hand circular polarization respectively; And calculate the amplitude ratio and phase difference of the ellipsometric parameters, expressed as: ; ; in represents the amplitude ratio, Indicates phase difference; The reflection coefficient is determined and the refraction angle data is determined by Snell's law, and the complex refractive index is calculated by the Fresnel formula and the refractive index and extinction coefficient are determined, which is expressed as: ; ; ; ; in and They represent the reflection coefficients in the parallel (p) and vertical (s) directions of the reflected light, respectively, and i represents the imaginary unit, which can be expressed as , represents the angle of incidence, is the refraction angle, which is determined by Snell's law and can be expressed as , represents the refractive index of the incident medium, i.e., air, represents the complex refractive index of the material, n represents the effective refractive index, and k represents the extinction coefficient, which can be expressed by the formula Determine, among which represents the intensity of light propagating to depth x in the material, represents the incident light intensity, Indicates the wavelength of light; According to the refractive index and extinction coefficient, the material's response to the electric field and the material's ability to absorb light are analyzed and calculated, and the overall complex dielectric constant is calculated, which is expressed as: ; ; ; in represents the material's response to the electric field, Indicates the light absorption capacity of a material. represents the overall complex permittivity.

[0027] Ion implantation defect detection through spectral ellipsometry can improve the sensitivity and resolution of defect detection, and can accurately analyze the optical properties of materials without damaging the sample. By measuring the intensity of reflected light polarized in parallel (p) and perpendicular (s) directions respectively, the response of the material to light in different polarization directions can be fully captured, thereby enhancing the detection ability of material anisotropy and microstructural changes. During the adjustment process of the quarter wave plate (QWP) and the polarizer, the optical activity and chirality information of the material can be extracted by converting circularly polarized light into linearly polarized light and measuring the intensity of reflected light after right-handed and left-handed circular polarization respectively. By calculating the Stokes parameters, the reflected light characteristics of the material can be accurately analyzed. Especially when there are complex lattice defects, the microscopic inhomogeneity inside the material can be revealed through numerical analysis, and the resolution of the damaged area can be improved. The amplitude ratio and phase difference of the ellipsometric parameters can be calculated, which can directly reflect the interaction of the material with different polarized light, avoid the complexity of directly measuring the dielectric constant, and improve the accuracy and stability of the measurement, making the detection process more repeatable. The reflection coefficient and refractive index are calculated using Snell's law and Fresnel's formula, providing a non-destructive and high-precision method to measure the optical properties of the material. By accurately calculating the refraction angle, the understanding of the ion implantation depth and its influence can be improved, which can be further used for ion doping control in semiconductor manufacturing. The extinction coefficient in the complex refractive index is determined, so that the material's ability to absorb light can be quantified, which can be used to predict the optical loss of the material at different wavelengths. The non-destructive spectral ellipsometry method is used in combination with the calculation of Stokes vector, ellipsometric parameters and complex dielectric constant to achieve high-precision automated detection, reduce the complexity of sample preparation and improve detection efficiency.

[0028] S2, divide the crystal material into regions, use nonlinear regression method to obtain the complex dielectric constant of the optimized region, build a three-layer composite medium model based on the regional proportion value, optimize and calculate the optimal solution of the regional proportion value through Lagrangian optimization method, perform deep layered Fourier transform, obtain the complex dielectric constant distribution in the depth domain, integrate the complex dielectric constant distribution data of different layers, confirm the normalized weight and calculate the material proportion value of different layers; Preferably, the crystal material regions are divided, the optimal solution of the region ratio is calculated, and the normalized weights are determined to calculate the material ratios of different layers, including: During the ion implantation process, the crystal structure of the silicon material is divided into a crystalline silicon region, an amorphous silicon region and a strained silicon region, and the complex dielectric constants corresponding to the crystalline silicon, amorphous silicon and strained silicon are determined based on experimental data, wherein the initial complex dielectric constant values ​​of the crystalline silicon, amorphous silicon and strained silicon at different wavelengths are obtained from a database, and the optimization goal is set to minimize the error between the three complex dielectric constants and the overall complex dielectric constant using a nonlinear regression method, and the optimization iteration is performed using the Newton iteration method until the iteration error no longer changes significantly, then the iteration is stopped and the three complex dielectric constants are determined; According to the sum of the products of the three complex dielectric constants and the corresponding values ​​of the proportion of crystalline silicon, amorphous silicon and strained silicon, a three-layer composite dielectric model is constructed, which is expressed as: ; in represents the overall complex permittivity, , as well as represent the complex dielectric constants of crystalline silicon, amorphous silicon and strained silicon, respectively. , as well as Respectively represent the proportion of crystalline silicon, amorphous silicon and strained silicon; By using the Lagrangian optimization method, the goal of maximizing the fit of the complex dielectric constant model is set to optimize the calculation , as well as The optimal solution of , where the Lagrangian function is expressed as: ; in represents the Lagrange multiplier, which is used to ensure that the sum of all proportions is equal to 1; The overall complex dielectric constant is subjected to a deep layered Fourier transform A-FFT to obtain the complex dielectric constant distribution in the depth domain, which is expressed as: ; ; in represents the complex dielectric constant after Fourier transformation in the depth domain z, Indicates wavelength The overall complex dielectric constant, It represents the depth frequency component inside the material, reflecting the oscillation behavior of the complex dielectric constant changing with depth z, and n represents the effective refractive index; based on The data corresponding to short wavelength, medium wavelength and long wavelength are extracted and inverse Fourier transformed respectively to obtain the complex dielectric constant distribution of the surface layer, medium depth layer and deep layer respectively. The normalized weight of each layer is determined according to the integral value of the complex dielectric constant distribution data of different layers, which is expressed as: ; in represents the normalized weight of the i-th layer, and denote the upper and lower boundaries of the i-th depth layer, respectively. represents the complex permittivity distribution data of the i-th layer, and A represents the integral sum of the complex permittivity of all depth layers; According to the normalized weights of different layers, the Lagrangian optimization method is used to calculate , as well as The product of the optimal proportion values ​​of is used to obtain the material proportion values ​​of different layers, and then normalization verification is performed.

[0029] By dividing the crystal material area, the distribution of crystalline silicon, amorphous silicon and strained silicon can be accurately distinguished after ion implantation, and the ability to analyze the changes in the material microstructure can be improved. When constructing a three-layer composite dielectric model, the weighted sum of the complex dielectric constant proportions of crystalline silicon, amorphous silicon and strained silicon is adopted to more accurately describe the changes in the overall complex dielectric constant. The Lagrangian optimization method is used to avoid non-physical solutions that may be caused by unconstrained optimization, thereby improving the stability of the model and the credibility of the calculation results. At the same time, by maximizing the fit of the complex dielectric constant model, it can ensure that the calculation results of the material proportion are optimal in the global range. Through deep layered Fourier transform, the overall complex dielectric constant can be analyzed in the depth direction, realizing the spatial layered characterization of material properties. The material proportions of different layers are calculated by multiplying the normalized weights with the optimal proportions obtained by the Lagrangian optimization method, and normalization verification is performed to ensure that the final material proportions are numerically consistent with the actual physical laws. By combining spectral ellipsometry with deep layered Fourier transform (A-FFT) combined with material area division and optimization calculation, the solution can achieve accurate layered analysis of material microstructure and defects based on high-precision polarization optical measurement, and obtain the optimal material composition ratio based on mathematical optimization methods, so that defect detection can be expanded from traditional two-dimensional surface information to high-resolution three-dimensional spatial characterization. The refractive index and extinction coefficient are derived through Stokes vector, ellipsometry parameter calculation, and Fresnel formula, so that the optical response of the material can be quantitatively described and directly related to the changes in the internal structure of the material after ion implantation, thus achieving non-destructive, high-precision, and deeply resolvable ion implantation defect detection.

[0030] S3, builds a deep learning model based on DNN, uses Bayesian optimization method to optimize model parameters, predicts material defect distribution, and performs prediction matching; Preferably, a deep learning model is constructed based on DNN, and the model parameters are optimized using the Bayesian optimization method to predict the material defect distribution, including: A deep learning model is constructed based on DNN, including an input layer, a hidden layer, and an output layer. The complex dielectric constant data of different depth layers and the corresponding material proportion data of each depth layer are combined into a data matrix, which is input into the deep learning model through the input layer. The material defect distribution is predicted through the hidden layer, including the material defect distribution area prediction value and the depth prediction value, and output through the output layer; The Bayesian optimization method is used to adjust the neural network parameters, and the optimization goal is set to maximize the model prediction accuracy. The iteration is stopped when the update changes are no longer obvious, and the optimized model parameters are obtained.

[0031] Through the DNN-based deep learning model, the complex dielectric constant data can be combined with the material proportion data, making the material defect distribution prediction more global and data-driven. The neural network can capture the complex characteristics of defects in different depth layers through the nonlinear mapping ability of the hidden layer, thereby achieving more detailed material damage analysis. The Bayesian optimization method is used to optimize the neural network parameters, which can avoid the inefficiency of traditional grid search or random search methods in high-dimensional hyperparameter space.

[0032] Furthermore, predictive matching prediction refers to using calibrated experimental measurement defect distribution data to calculate the matching degree of the deep learning model's predicted material defect distribution data, setting an error threshold based on the experimental data, and if the matching degree is greater than or equal to the error threshold, the prediction is judged to be accurate.

[0033] Through matching prediction verification, it is possible to ensure that the defect distribution prediction results of the deep learning model are highly consistent with the actual experimental measurement data, thereby improving the credibility and practical usability of the model. By calculating the prediction matching degree of the deep learning model, the deviation between the predicted data and the experimental measurement data can be quantified, and the error threshold setting can be used to automatically determine whether the prediction is qualified. This makes the evaluation system of the entire model more standardized and automated, without relying on manual analysis of the prediction results, and improves the intelligence and efficiency of the material defect detection process.

[0034] S4, encrypting the prediction data and storing it securely; Preferably, encrypting the predicted data and storing it securely means encrypting the predicted defect distribution data using the AES-256 encryption standard and encrypting and storing it through KMS security key management.

[0035] By encrypting data for storage, unauthorized access to data during storage or transmission can be prevented, thereby encrypting sensitive data. In addition, secure key management can further improve the data access protection effect for users.

[0036] This embodiment also provides a system for a high-precision nondestructive detection method of ion implantation defects, comprising: Optical data acquisition module, which selects and controls the measurement light source, adjusts the polarization state of the incident light, collects reflected light intensity data, and measures the response of the material to different polarized light; Optical parameter calculation module, which analyzes the polarization information of light, calculates the material's response parameters to light, calculates the ellipsometric parameters, determines the refraction angle data, and then obtains the refractive index and extinction coefficient; The three-layer composite dielectric module divides the crystalline silicon, amorphous silicon and strained silicon regions according to the material measurement data, obtains the complex dielectric constants of different regions through optimization methods, and constructs a composite dielectric model; The material layer analysis module performs deep layer Fourier transform, analyzes the distribution of complex dielectric constants at different depth layers, calculates the normalized weights of different layers, and finally obtains the material proportions of different layers; Deep learning defect prediction module, which uses DNN to predict defect distribution, identify damaged areas in materials, and adjust model parameters through Bayesian optimization; The matching degree verification module calculates the matching degree between the DNN prediction results and the experimental measurement data to determine the reliability of the prediction results; The data security storage module securely stores measurement data, calculation data and prediction data.

[0037] This embodiment also provides a computer device, which is suitable for the case of a high-precision non-destructive detection method for ion implantation defects, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the high-precision non-destructive detection method for ion implantation defects proposed in the above embodiment.

[0038] The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covered on the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.

[0039] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, the method for high-precision nondestructive detection of ion implantation defects proposed in the above embodiment is implemented; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, referred to as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, referred to as EEPROM), erasable programmable read-only memory (Erasable Programmable Read Only Memory, referred to as EPROM), programmable read-only memory (Programmable Red-Only Memory, referred to as PROM), read-only memory (Read-Only Memory, referred to as ROM), magnetic storage, flash memory, magnetic disk or optical disk.

[0040] In summary, the present invention can reveal the microscopic inhomogeneity inside the material through numerical analysis by calculating the Stokes parameters, improve the resolution of the damaged area, and by dividing the crystal material area, can accurately distinguish the distribution of crystalline silicon, amorphous silicon and strained silicon after ion implantation, thereby improving the ability to analyze the changes in the microstructure of the material. By combining spectral ellipsometry measurement with deep layered Fourier transform combined with material area division and optimization calculation, the scheme can achieve accurate layered analysis of the material microstructure and defects on the basis of high-precision polarization optical measurement. The refractive index and extinction coefficient are derived by Stokes vector, ellipsometry parameter calculation, and Fresnel formula, so that the optical response of the material can be quantitatively described and directly related to the changes in the internal structure of the material after ion implantation, thereby realizing non-destructive, high-precision, and deeply resolvable ion implantation defect detection.

[0041] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A high-precision nondestructive detection method for ion implantation defects, characterized in that: include: Select a measurement light source to collect reflected light intensity data, use the Stokes vector to represent the intensity distribution of polarized light, calculate the ellipsometric parameters and determine the refraction angle data, calculate the complex refractive index through the Fresnel formula and determine the refractive index and extinction coefficient, and analyze the overall complex dielectric constant of the material; Perform regional division of crystal materials, use nonlinear regression method to obtain the complex dielectric constant of the optimized region, build a three-layer composite medium model based on the regional proportion value, optimize and calculate the optimal solution of the regional proportion value through Lagrangian optimization method, perform deep layered Fourier transform, obtain the complex dielectric constant distribution in the depth domain, integrate and calculate the complex dielectric constant distribution data of different layers, and confirm the normalized weight to calculate the material proportion value of different layers; Build a deep learning model based on DNN, use Bayesian optimization method to optimize model parameters, predict material defect distribution, and make prediction matching degree; The prediction data is encrypted and stored securely.

2. The high-precision nondestructive detection method for ion implantation defects according to claim 1, characterized in that: The overall complex dielectric constant of the analyzed material includes: Ion implantation defect detection is performed based on spectral ellipsometry. The intensity of reflected light polarized in parallel p and perpendicular s directions is measured respectively. The circularly polarized light is converted into linearly polarized light by placing a quarter-wave plate QWP in the path of the reflected beam. The polarizer is adjusted in turn to allow right-handed circularly polarized light and left-handed circularly polarized light to pass through. The intensity of reflected light after right-handed and left-handed circularly polarized light is measured. By placing a wavelength modulating wave plate in the beam path, the reflected light intensity data after the polarizer is collected; Use Stokes vector to represent the intensity distribution of polarized light and calculate the amplitude ratio and phase difference of ellipsometric parameters; Determine the reflection coefficient and determine the refraction angle data through Snell's law, and calculate the complex refractive index through Fresnel's formula and determine the refractive index and extinction coefficient; The material's response to the electric field and its light absorption capacity are analyzed and calculated based on the refractive index and extinction coefficient, and the overall complex dielectric constant is calculated.

3. The high-precision nondestructive detection method for ion implantation defects according to claim 2, characterized in that: The method of dividing the crystal material regions, calculating the optimal solution of the region ratio, and confirming the normalized weights to calculate the material ratios of different layers includes: During the ion implantation process, the crystal structure of the silicon material is divided into a crystalline silicon region, an amorphous silicon region and a strained silicon region, and the complex dielectric constants corresponding to the crystalline silicon, amorphous silicon and strained silicon are determined based on experimental data, wherein the initial complex dielectric constant values ​​of the crystalline silicon, amorphous silicon and strained silicon at different wavelengths are obtained from a database, and the optimization goal is set to minimize the error between the three complex dielectric constants and the overall complex dielectric constant using a nonlinear regression method, and the optimization iteration is performed using the Newton iteration method until the iteration error no longer changes significantly, then the iteration is stopped and the three complex dielectric constants are determined; A three-layer composite dielectric model is constructed based on the sum of the products of the three complex dielectric constants and the corresponding crystalline silicon, amorphous silicon and strained silicon proportions as the overall complex dielectric constant, where , as well as Respectively represent the proportion of crystalline silicon, amorphous silicon and strained silicon; By using the Lagrangian optimization method, the goal of maximizing the fit of the complex dielectric constant model is set to optimize the calculation , as well as The optimal solution of Performing a deep layered Fourier transform (A-FFT) on the overall complex permittivity to obtain the distribution of the complex permittivity in the depth domain; Based on the complex dielectric constant after Fourier transformation, the data corresponding to short wavelength, medium wavelength and long wavelength are extracted and inverse Fourier transformed to obtain the complex dielectric constant distribution of the surface layer, medium depth layer and deep layer materials respectively, and the normalized weight of each layer is determined according to the integral value of the complex dielectric constant distribution data of different layers; According to the normalized weights of different layers, the Lagrangian optimization method is used to calculate , as well as The product of the optimal proportion values ​​of is used to obtain the material proportion values ​​of different layers, and then normalization verification is performed.

4. The high-precision nondestructive detection method for ion implantation defects according to claim 3, characterized in that: The deep learning model is constructed based on DNN, and the model parameters are optimized using the Bayesian optimization method to predict the material defect distribution, including: A deep learning model is constructed based on DNN, including an input layer, a hidden layer, and an output layer. The complex dielectric constant data of different depth layers and the corresponding material proportion data of each depth layer are combined into a data matrix, which is input into the deep learning model through the input layer. The material defect distribution is predicted through the hidden layer, including the material defect distribution area prediction value and the depth prediction value, and output through the output layer; The Bayesian optimization method is used to adjust the neural network parameters, and the optimization goal is set to maximize the model prediction accuracy. The iteration is stopped when the update changes are no longer obvious, and the optimized model parameters are obtained.

5. The high-precision nondestructive detection method for ion implantation defects according to claim 4, characterized in that: The prediction matching degree prediction refers to using calibrated experimental measurement defect distribution data to calculate the matching degree of the deep learning model's predicted material defect distribution data, setting an error threshold based on the experimental data, and if the matching degree is greater than or equal to the error threshold, the prediction is judged to be accurate.

6. The high-precision nondestructive detection method for ion implantation defects according to claim 5, characterized in that: The selection of the measurement light source to collect reflected light intensity data refers to selecting a supercontinuum laser source as the measurement light source, using a femtosecond pulse light source, using a polarization modulator to generate incident light in a specific polarization state, allowing the incident light to pass through an adjustable polarizer to adjust the polarization direction, and collecting reflected light intensity data through a high-sensitivity photodetector.

7. The high-precision nondestructive detection method for ion implantation defects according to claim 6, characterized in that: The encrypting and securely storing the predicted data refers to encrypting the predicted defect distribution data using the AES-256 encryption standard and encrypting and storing it through KMS security key management.

8. A system for a high-precision nondestructive detection method for ion implantation defects, based on the high-precision nondestructive detection method for ion implantation defects according to any one of claims 1 to 7, characterized in that: include, Optical data acquisition module, which selects and controls the measurement light source, adjusts the polarization state of the incident light, collects reflected light intensity data, and measures the response of the material to different polarized light; Optical parameter calculation module, which analyzes the polarization information of light, calculates the material's response parameters to light, calculates the ellipsometric parameters, determines the refraction angle data, and then obtains the refractive index and extinction coefficient; The three-layer composite dielectric module divides the crystalline silicon, amorphous silicon and strained silicon regions according to the material measurement data, obtains the complex dielectric constants of different regions through optimization methods, and constructs a composite dielectric model; The material layer analysis module performs deep layer Fourier transform, analyzes the distribution of complex dielectric constants at different depth layers, calculates the normalized weights of different layers, and finally obtains the material proportions of different layers; Deep learning defect prediction module, which uses DNN to predict defect distribution, identify damaged areas in materials, and adjust model parameters through Bayesian optimization; The matching degree verification module calculates the matching degree between the DNN prediction results and the experimental measurement data to determine the reliability of the prediction results; The data security storage module securely stores measurement data, calculation data and prediction data.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the high-precision non-destructive detection method for ion implantation defects described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the high-precision non-destructive detection method for ion implantation defects described in any one of claims 1 to 7 are implemented.

Images