Inversion methods, systems, equipment, and storage media for semiconductor epitaxial layer growth parameters

Abstract

This application provides a method, system, device, and storage medium for inverting growth parameters of semiconductor epitaxial layers, relating to the field of semiconductor detection technology. This application pre-sets optical parameters as functions of wavelength and composition, storing them in a database. During each iteration, the parameters are dynamically retrieved from the database based on the current growth parameters. Then, a predicted reflection spectrum is calculated based on these optical parameters and the transfer matrix. This allows the predicted reflection spectrum to match the actual changes in the material's optical constants with compositional fluctuations in real time during iteration, thus avoiding the problem of traditional fixed optical constant models misinterpreting refractive index changes as thickness or growth rate anomalies during epitaxial growth. Furthermore, the dynamic query mechanism of this application enables thickness and composition to be synchronously updated and jointly converged under a unified objective function and physical model constraints, thereby continuously outputting higher accuracy and more physically consistent composition and thickness inversion results during the growth process.

Description

Technical Field

[0001] This application relates to the field of semiconductor detection technology, and in particular to a method, system, device and storage medium for inverting semiconductor epitaxial layer growth parameters. Background Technology

[0002] In-situ real-time monitoring technology for epitaxial growth is one of the important technologies in advanced semiconductor manufacturing. It uses optical methods to perform non-contact, real-time measurement of film thickness, growth rate, chemical composition and microstructure during epitaxial growth processes such as metal-organic chemical vapor deposition (MOCVD).

[0003] Traditional epitaxial in-situ monitoring techniques generally assume that the optical parameters of the material are fixed, and the analysis process prioritizes thickness and relies on offline characterization to determine the composition. This results in the refractive index changes caused by the composition during the growth process being coupled with the thickness changes. The composition information cannot be output in real time, and the fitting results are highly affected by the errors of the preset model. This leads to imperceptible systematic deviations in the calculated thickness and growth rate, which seriously affects the production yield. Summary of the Invention

[0004] The purpose of this application is to provide a method, system, device and storage medium for inverting semiconductor epitaxial layer growth parameters, so as to overcome the defects of traditional technology that generally assumes that the optical parameters of the material are fixed and that the analysis process prioritizes thickness and relies on offline characterization to determine the composition, resulting in undetectable systematic deviations in the calculated thickness and growth rate, which seriously affect the production yield.

[0005] In a first aspect, this application proposes a method for inverting semiconductor epitaxial layer growth parameters, including:

[0006] The measured reflectance spectrum and growth parameters corresponding to the current iteration number during semiconductor epitaxial growth are obtained; the growth parameters include thickness and composition.

[0007] Based on the growth parameters, optical parameters are obtained from a preset database. These optical parameters are functions of wavelength and composition.

[0008] The predicted reflection spectrum is calculated based on the preset transmission matrix, the optical parameters, and the thickness.

[0009] The target difference is calculated based on the predicted reflectance spectrum and the measured reflectance spectrum. The growth parameters are then updated step by step based on the target difference until the iteration condition is met. The value of the growth parameters when the iteration condition is met is taken as the inversion result.

[0010] In one embodiment, the optical parameters include refractive index and extinction coefficient;

[0011] The construction method of the preset database includes:

[0012] Obtain calibration data for multiple sets of standard samples, wherein the composition of the standard samples is known;

[0013] The optical constants of each of the standard samples are measured within a preset wavelength range to obtain the corresponding optical parameters.

[0014] Based on the composition of each standard sample and its corresponding optical parameters, a fitting process is performed to establish a correlation model between the optical parameters and the wavelength and composition.

[0015] Each associated model is written into the database to form the preset database.

[0016] In one embodiment, the fitting process based on the components of each of the standard samples and their corresponding optical parameters to establish a correlation model between the optical parameters and the wavelength and components includes:

[0017] Select multiple sampling points within a preset wavelength range;

[0018] For each sampling point, the optical parameters of each standard sample at that sampling point are summarized, and a parameterized fitting function is constructed with the composition as the independent variable and the optical parameters as the dependent variable.

[0019] An error objective is constructed based on the parameterized fitting function, and the fitting coefficient at the sampling point is obtained by minimizing the error objective.

[0020] Interpolation or continuous fitting is performed on the fitting coefficients corresponding to each sampling point along the wavelength direction to obtain model coefficients that change continuously with wavelength.

[0021] The correlation model of the optical parameters with respect to wavelength and composition is generated based on the model coefficients that vary continuously with wavelength.

[0022] In one embodiment, the method further includes:

[0023] After one epitaxial growth process is completed, the corresponding measured reflectance spectrum and inversion results are obtained;

[0024] Based on the inversion results and the measured reflectance spectra, corresponding data of growth parameters and spectra are established, and the corresponding data are written into the sample set of the preset database as feedback samples.

[0025] Based on the sample set, parameter correction or model refitting is performed on the associated models in the preset database to update the associated models.

[0026] In one embodiment, the step of calculating the target difference based on the predicted reflectance spectrum and the measured reflectance spectrum, and updating the growth parameters based on the target difference, includes:

[0027] A residual sequence is obtained based on the difference between the measured reflectance spectrum and the predicted reflectance spectrum at each sampling point;

[0028] An objective function is constructed based on the residual sequence, wherein the objective function is a function that sums the squares of each element of the residual sequence according to a preset noise weight.

[0029] Calculate the target difference value of the objective function at the current growth parameters;

[0030] The sensitivity information of the predicted reflectance spectrum relative to the growth parameters is calculated, and based on the sensitivity information, the residual sequence and its corresponding noise weights, a least squares incremental solution equation is constructed to obtain the parameter update amount of the growth parameters.

[0031] The growth parameters are updated based on the parameter update amount.

[0032] In one embodiment, the method for determining whether the iteration condition is met includes: if the target difference is less than a preset threshold or the number of iterations meets a preset number, then the iteration condition is considered to be met.

[0033] In one embodiment, the method further includes:

[0034] Based on the inversion results of each iteration, a change curve of growth parameters is generated and updated in real time; wherein, the change curve is compared and displayed with the preset target process curve to determine whether the design intent of the preset structure has been executed in the same growth process; and a prompt message is output when the comparison deviation exceeds the preset deviation range.

[0035] Secondly, this application proposes an inversion system for semiconductor epitaxial layer growth parameters, the system comprising:

[0036] The acquisition module is used to acquire the measured reflection spectrum and growth parameters corresponding to the current iteration number during semiconductor epitaxial growth; the growth parameters include thickness and composition.

[0037] The query module is used to retrieve optical parameters from a preset database based on the growth parameters, wherein the optical parameters are functions of wavelength and composition.

[0038] The processing module is used to calculate the predicted reflection spectrum based on the preset transmission matrix, the optical parameters, and the thickness; calculate the target difference based on the predicted reflection spectrum and the measured reflection spectrum; gradually update the growth parameters based on the target difference until the iteration condition is met; and use the value of the growth parameters when the iteration condition is met as the inversion result.

[0039] Thirdly, this application also provides a computer device. The computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the method steps in the first aspect.

[0040] Fourthly, this application also provides a computer-readable storage medium. The computer-readable storage medium stores a computer program thereon, which, when executed by a processor, implements the method steps of the first aspect.

[0041] The aforementioned method, system, equipment, and storage medium for inverting semiconductor epitaxial layer growth parameters have at least the following advantages:

[0042] This application pre-sets optical parameters as functions of wavelength and composition, storing them in a database. During each iteration, the parameters are dynamically retrieved from the database based on the current growth parameters. The predicted reflection spectrum is then calculated based on these optical parameters and the transfer matrix. This allows the predicted reflection spectrum to match the actual changes in the material's optical constants with compositional fluctuations in real time during iteration, thus avoiding the problem of traditional fixed optical constant models misinterpreting refractive index changes as thickness or growth rate anomalies during epitaxial growth. Furthermore, this application's dynamic query mechanism enables thickness and composition to be synchronously updated and jointly converged under a unified objective function and physical model constraints. This allows for the continuous output of higher accuracy and more physically consistent composition and thickness inversion results during the growth process. Attached Figure Description

[0043] Figure 1 This is a flowchart illustrating the method for inverting semiconductor epitaxial layer growth parameters in one embodiment;

[0044] Figure 2 This is a flowchart illustrating the steps of constructing a preset database in one embodiment;

[0045] Figure 3 This is a flowchart illustrating the steps for updating growth parameters in one embodiment;

[0046] Figure 4 This is a structural block diagram of a semiconductor epitaxial layer growth parameter inversion system in one embodiment;

[0047] Figure 5 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0048] The following specific examples illustrate the implementation of this application. Those skilled in the art can easily understand other advantages and effects of this application from the content disclosed in this specification. This application can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of this application. It should be noted that, unless otherwise specified, the following embodiments and features in the embodiments can be combined with each other.

[0049] Some exemplary embodiments of this application have been described for illustrative purposes. It should be understood that this application may be implemented in other ways not specifically shown in the accompanying drawings.

[0050] Please see Figure 1 In one exemplary embodiment, this application provides a method for inverting the growth parameters of a semiconductor epitaxial layer, specifically including the following steps:

[0051] Step 102: Obtain the measured reflection spectrum and growth parameters corresponding to the current iteration number during the semiconductor epitaxial growth process; wherein, the growth parameters include thickness and composition.

[0052] Specifically, this application applies to applications involving semiconductor epitaxial growth. Typically, a carrier disk made of high-purity graphite is placed inside the semiconductor reaction chamber, and multiple trays for placing semiconductors are mounted on the carrier disk. Supported by a rotating shaft driven by a motor, the carrier disk can rotate continuously during the semiconductor epitaxial growth process.

[0053] Furthermore, an incident optical window and an exit optical window are provided at the top of the semiconductor reaction chamber. A halogen tungsten lamp or a supercontinuum light source is used as a broadband light source. The output probe light is incident at a preset angle and shines on the semiconductor surface through the incident optical window. The reflected light is coupled to a spectrometer via an optical fiber through the exit optical window. The spectrometer continuously collects the reflection spectrum during the growth process at a preset sampling frequency, and performs dark field subtraction and reference spectrum normalization on the reflection spectrum to obtain the measured reflection spectrum at the current moment.

[0054] In this embodiment, the inversion process is achieved through multiple iterations, and in each iteration, the measured reflectance spectrum at the current moment is used as the observation. The growth parameters in the first iteration can be random or empirical values. The growth parameters in the second to Nth iterations are simply the growth parameters updated in the previous iteration.

[0055] Step 104: Based on the growth parameters, retrieve the optical parameters from the preset database. The optical parameters are functions of wavelength and composition.

[0056] Specifically, the preset database in this embodiment is a material optical constant database, used to store the correspondence between the optical parameters and wavelengths and compositions of each target material system during epitaxial growth, and this correspondence is established through prior calibration experiments. The target material systems include binary, ternary, or quaternary alloys. Further, the optical parameters include refractive index n and extinction coefficient k. For example, for the material AlxGa1-xAs, a series of standard samples with known compositions x can be grown, and the n(λ) and k(λ) of each standard sample within a preset wavelength range can be measured using an ellipsometry. The measurement results are then fitted to obtain an empirical model of the optical parameters with respect to wavelength and composition, and this empirical model or its model parameters are stored in the preset database.

[0057] Step 106: Calculate the predicted reflection spectrum based on the preset transmission matrix, optical parameters, and thickness.

[0058] Specifically, the propagation of light in multilayer thin films can be accurately described using the transfer matrix method. Each uniform thin film layer corresponds to a characteristic matrix, also known as the Abelès matrix. This matrix characterizes the phase accumulation caused by light propagation within the film layer and the admittance matching relationship with the adjacent medium. It is used to establish a mapping between the electromagnetic field quantity on the incident side and the electromagnetic field quantity on the emitting side of the film layer. After multiplying the characteristic matrices of each film layer sequentially according to the epitaxial layer order to obtain the total transfer matrix, the reflection coefficient of the multilayer film system at each sampling wavelength point can be calculated by combining the optical admittance of the environmental medium and the substrate. Furthermore, the predicted reflectance can be obtained, thus forming the predicted reflection spectrum.

[0059] Step 108: Calculate the target difference based on the predicted reflectance spectrum and the measured reflectance spectrum, and gradually update the growth parameters based on the target difference until the iteration condition is met. The value of the growth parameters when the iteration condition is met is taken as the inversion result.

[0060] Specifically, in each iteration, the predicted reflectance spectrum is calculated using the growth parameters obtained from the previous iteration, the optical parameters dynamically provided by the preset database, and the transfer matrix. This predicted reflectance spectrum is then fitted and compared with the measured reflectance spectrum acquired at the same time to obtain a target difference value characterizing the degree of matching between the two. Subsequently, based on this target difference value, the current growth parameters are progressively corrected, causing the target difference value recalculated in subsequent iterations to continuously decrease until the iteration condition is met. The value of the growth parameters when the iteration condition is met is then used as the inversion result.

[0061] The aforementioned method for inverting semiconductor epitaxial layer growth parameters pre-sets optical parameters as functions of wavelength and composition, storing them in a database. During each iteration, the parameters are dynamically retrieved from the database based on the current growth parameters. The predicted reflection spectrum is then calculated based on these optical parameters and the transfer matrix. This allows the predicted reflection spectrum to match the actual changes in the material's optical constants with compositional fluctuations in real time during iteration, thus avoiding the problem of traditional fixed optical constant models misinterpreting refractive index changes as thickness or growth rate anomalies during epitaxial growth. Furthermore, the dynamic query mechanism of this application enables thickness and composition to be synchronously updated and jointly converged under a unified objective function and physical model constraint. This allows for the continuous output of higher accuracy and more physically consistent composition and thickness inversion results during the growth process.

[0062] Please see Figure 2 Optionally, the default database construction method includes:

[0063] Step 202: Obtain calibration data for multiple sets of standard samples, wherein the components of the standard samples are known.

[0064] Step 204: Measure the optical constants of each standard sample within a preset wavelength range to obtain the corresponding optical parameters.

[0065] Step 206: Based on the composition of each standard sample and its corresponding optical parameters, perform fitting processing to establish a correlation model between the optical parameters and the wavelength and composition.

[0066] Step 208: Write each associated model into the database to form a preset database.

[0067] Specifically, to construct a pre-defined database for inversion, this application first obtains calibration data for multiple sets of standard samples. The material system and structure of each standard sample are consistent with or similar to the epitaxial system to be monitored, and its component parameters have been predetermined and recorded. Subsequently, optical constants of each standard sample are measured within a pre-defined wavelength range. Exemplarily, an ellipsomerometer, a spectrophotometer, or other optical characterization equipment capable of outputting optical parameters can be used to obtain a dataset of optical parameters for each standard sample at different wavelengths.

[0068] Furthermore, the component labels of each standard sample and their corresponding optical parameter datasets are summarized, and a fitting process is performed to establish a correlation model between the optical parameters and the wavelength and components. For example, the fitting process can adopt empirical function fitting, piecewise fitting, or interpolation modeling methods, so that when the target wavelength and component parameters are input, the corresponding refractive index and extinction coefficient can be output.

[0069] Finally, the obtained correlation model or its parameters are written into a database and archived according to material system, composition range, and wavelength range to form a preset database. In subsequent inversion processes, the required optical parameters can be dynamically retrieved from this preset database based on candidate components and wavelengths.

[0070] Optionally, a fitting process is performed based on the composition of each standard sample and its corresponding optical parameters to establish a correlation model between the optical parameters and the wavelength and composition, including:

[0071] Multiple sampling points are selected within a preset wavelength range. For each sampling point, the optical parameters of each standard sample at that sampling point are summarized, and a parameterized fitting function is constructed with the composition as the independent variable and the optical parameters as the dependent variable. An error objective is constructed based on the parameterized fitting function, and the fitting coefficients at that sampling point are obtained by minimizing the error objective. Interpolation or continuous fitting processing is performed on the fitting coefficients corresponding to each sampling point along the wavelength direction to obtain model coefficients that continuously change with wavelength. Based on the model coefficients that continuously change with wavelength, a correlation model of optical parameters with respect to wavelength and composition is generated.

[0072] Specifically, for AlxGa1-xAs, an empirical model can be obtained by growing a series of standard samples with known composition x, measuring their n(λ) and k(λ) using an ellipsometry, and then fitting the data. Multiple sampling wavelength points are then selected within a preset wavelength range. For each Summarize different known components The standard sample was measured , and respectively and Regarding components Perform linear fitting to obtain wavelength-related fitting coefficients, thereby establishing:

[0073] in, , , , These are the model coefficients that vary with wavelength; at discrete sampling points. The locations are respectively recorded as , , , And perform interpolation or continuous fitting on these discrete coefficients along the wavelength direction to obtain arbitrary wavelengths. place , , , Ultimately, the database stores these coefficients as a function of the model coefficients with respect to wavelength, allowing any input to be processed. and candidate components In any case, the above formula can be used to directly output the result. and .

[0074] Furthermore, for ternary and quaternary compounds, the component variables can be extended to multiple component parameters and a higher-order or cross-term model can be used. For example, the binary linear model of a ternary component can be expressed as:

[0075]

[0076]

[0077] Its model coefficients are also in each The data was obtained by fitting at a specific point and then made continuous along the wavelength direction before being written into the database.

[0078] By employing the above scheme and modeling method that fits wavelength-based sampling points and makes the model continuous along the wavelength, this application can convert the optical constant measurement results of discrete calibration samples into optical parameter functions that are continuously usable across the entire wavelength band. This allows the database to stably output the corresponding optical parameters under any wavelength and any candidate component input, thereby improving the continuity and consistency of the predicted reflectance spectrum in the transfer matrix calculation and reducing fitting errors caused by missing data. Simultaneously, since the fitting coefficients are smoothed and made continuous along the wavelength, the model's robustness to measurement noise is enhanced, making subsequent inversion iterations easier to converge and improving the reliability and physical consistency of the growth parameter inversion results.

[0079] Optionally, the above-mentioned method for inverting the growth parameters of the semiconductor epitaxial layer further includes:

[0080] After one epitaxial growth process is completed, the corresponding measured reflectance spectrum and inversion results are obtained;

[0081] Based on the inversion results and the measured reflectance spectra, corresponding data of growth parameters and spectra are established, and the corresponding data are written into the sample set of the preset database as feedback samples.

[0082] Based on the sample set, parameter correction or model refitting is performed on the associated models in the preset database to update the associated models.

[0083] Specifically, after each epitaxial growth batch is completed, the measured reflectance spectrum sequence collected during the growth process of that batch is acquired, and the corresponding inversion result for that batch is obtained. The inversion result includes the thickness parameter curve and composition parameter curve changing over time, as well as the thickness and composition values ​​at the final convergence. Further, the inversion result and the spectral sequence are aligned temporally to form corresponding data entries for growth parameters and spectra. For example, the composition parameter, thickness parameter, and measured reflectance spectrum at each sampling moment are used as a feedback sample.

[0084] The above feedback samples are written into the sample set of the preset database. Furthermore, the sample set can be filtered to remove samples with low signal-to-noise ratio, abnormal spectrum, or unconverged inversion, and only samples that meet the preset quality conditions are retained for model updates.

[0085] During the model update phase, this application uses a sample set comprised of both new and historical samples as training data to perform parameter correction or model refitting on the associated models in the database. For example, the model coefficients are updated while maintaining the model's form, or a higher-order model is switched to and refitted when the sample size reaches a threshold. After the update is complete, the updated model coefficients or associated models are written back to a preset database for dynamic querying of optical parameters and calculation of predicted reflectance spectra in subsequent batches.

[0086] By employing the above approach, data from each growth process is continuously written into the database as feedback samples and used for parameter correction or refitting of the correlation model. This allows the material optical constant model to be gradually calibrated and enriched based on factors such as equipment status and batch-to-batch material differences, avoiding systematic biases caused by long-term reliance on fixed calibration models. Simultaneously, as the sample set accumulates, the database's characterization of optical parameters under different composition and wavelength conditions more closely resembles the real production environment, improving the matching accuracy of predicted reflectance spectra in subsequent inversions.

[0087] Please see Figure 3 Optionally, the target difference is calculated based on the predicted reflectance spectrum and the measured reflectance spectrum, and the growth parameters are updated progressively based on the target difference, including:

[0088] Step 302: Obtain the residual sequence based on the difference between the measured reflectance spectrum and the predicted reflectance spectrum at each sampling point.

[0089] Step 304: Construct an objective function based on the residual sequence, wherein the objective function is a function that sums the squares of each element of the residual sequence according to a preset noise weight.

[0090] Step 306: Calculate the target difference value of the objective function at the current growth parameters.

[0091] Step 308: Calculate the sensitivity information of the predicted reflectance spectrum relative to the growth parameters, and construct the least squares incremental solution equation based on the sensitivity information, the residual sequence and its corresponding noise weights to obtain the parameter update amount of the growth parameters; update the growth parameters based on the parameter update amount.

[0092] Specifically, the essence of inversion is to solve a nonlinear least squares problem, that is, to find the optimal thickness d and composition C so that the difference between the theoretical spectrum and the measured spectrum is minimized across the entire spectrum.

[0093] After calculating the predicted reflectance spectrum, the measured reflectance spectrum and the predicted reflectance spectrum are compared point by point within a preset wavelength range. The difference between the two is calculated at each sampling wavelength point. All differences are arranged in wavelength order to form a residual sequence to characterize the spectral shape fitting deviation under the current growth parameters.

[0094] Furthermore, this application constructs an objective function based on the residual sequence. Preset noise weights are introduced into each element of the residual sequence, and the weighted residuals are squared and summed to obtain the objective function used to quantify the fitting error. The value of this objective function at the current growth parameter is calculated, and this value is used as the target difference value for the current iteration. The purpose of introducing noise weights is to reduce the influence of low signal-to-noise ratio bands or spectral edge regions on the objective function, thereby improving the robustness of the optimization process.

[0095] Furthermore, the sensitivity information of the predicted reflectance spectrum relative to the current growth parameters is calculated to obtain the trend of the predicted reflectance spectrum when the growth parameters change slightly. The least-squares incremental solution equation is then constructed by combining the sensitivity information, the residual sequence, and its corresponding noise weights to obtain the parameter update amount of the growth parameters. For example, a nonlinear least-squares optimization algorithm (Levenberg–Marquardt) is used. In each iteration, the optical constants are obtained by querying a preset database based on the current (d, C), and the predicted reflectance spectrum R_calc and its partial derivatives with respect to d and C (i.e., the sensitivity matrix) are calculated. Based on this, the parameter update amount is solved to determine the direction of the next iteration, until the objective function is minimized.

[0096] The updated parameters are added to the current growth parameters to obtain the growth parameters for the next iteration. This process is repeated until the target difference value gradually decreases with each iteration, at which point the inversion result is output. The criteria for satisfying the iteration conditions include: the target difference is less than a preset threshold, or the number of iterations meets a preset number.

[0097] By adopting the above scheme, the thickness and composition are simultaneously iterated and optimized under the same objective function constraint. This allows the predicted reflectance spectrum to respond simultaneously to the phase accumulation change caused by the thickness change and the optical constant change caused by the composition change in each iteration. This avoids the coupling misjudgment and error propagation introduced by fixing the composition or the thickness first in the traditional technique. Furthermore, the simultaneous iteration enables the thickness and composition to constrain each other and converge collaboratively during the optimization process. It can maintain stable convergence even in scenarios where the material composition fluctuates or the optical constant changes significantly with the composition. This improves the physical consistency and traceability of the inversion results and can continuously output higher precision thickness and composition results during the growth process.

[0098] Optionally, the above-mentioned method for inverting the growth parameters of the semiconductor epitaxial layer further includes:

[0099] Based on the inversion results of each iteration, a curve showing the change of growth parameters is generated and updated in real time. The curve is compared with the preset target process curve to determine whether the design intent of the preset structure has been executed in the same growth process. When the comparison deviation exceeds the preset deviation range, a prompt message is output.

[0100] The above approach transforms the inversion results obtained from each iteration into a visualized change curve in real time, which is then compared online with the preset target process curve. This allows engineers to directly verify whether the design intent of the preset structure is accurately executed within the same growth run, avoiding the time lag caused by post-verification in traditional technologies. Simultaneously, based on the deviation exceeding limit alert mechanism, key deviation segments can be quickly located when deviations first occur, providing a judgment benchmark for online correction, thereby reducing the number of trial and error attempts and shortening the process development cycle.

[0101] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0102] Based on the same inventive concept, this application also provides an inversion system for semiconductor epitaxial layer growth parameters. This system is applicable to the above-mentioned inversion method for semiconductor epitaxial layer growth parameters. The solution provided by this system is similar to the solution described in the above-mentioned method. Therefore, the specific limitations in one or more device embodiments provided below can be found in the limitations of the method above, and will not be repeated here.

[0103] Please see Figure 4 In one embodiment, the inversion system for semiconductor epitaxial layer growth parameters includes: an acquisition module, a query module, and a processing module.

[0104] The acquisition module is used to acquire the measured reflectance spectrum and growth parameters corresponding to the current iteration number during semiconductor epitaxial growth; the growth parameters include thickness and composition.

[0105] The query module is used to retrieve optical parameters from a preset database based on growth parameters. The optical parameters are functions of wavelength and composition.

[0106] The processing module is used to calculate the predicted reflection spectrum based on the preset transmission matrix, optical parameters and thickness; calculate the target difference based on the predicted reflection spectrum and the measured reflection spectrum; update the growth parameters step by step based on the target difference until the iteration condition is met; and use the value of the growth parameters when the iteration condition is met as the inversion result.

[0107] Optionally, the above-mentioned inversion system for semiconductor epitaxial layer growth parameters further includes: a database construction module.

[0108] The database construction module is used to build a preset database, including: acquiring calibration data of multiple sets of standard samples, the composition of which is known; measuring the optical constants of each standard sample within a preset wavelength range to obtain the corresponding optical parameters; performing fitting processing based on the composition of each standard sample and its corresponding optical parameters to establish a correlation model between the optical parameters and the wavelength and composition; and writing each correlation model into the database to form the preset database.

[0109] Optionally, the database construction module performs fitting processing based on the components and corresponding optical parameters of each standard sample to establish a correlation model of optical parameters with respect to wavelength and components. This includes: selecting multiple sampling points within a preset wavelength range; for each sampling point, summarizing the optical parameters of each standard sample at that sampling point, and constructing a parameterized fitting function with components as independent variables and optical parameters as dependent variables; constructing an error objective based on the parameterized fitting function, and obtaining the fitting coefficients at that sampling point by minimizing the error objective; performing interpolation or continuous fitting processing on the fitting coefficients corresponding to each sampling point along the wavelength direction to obtain model coefficients that continuously change with wavelength; and generating a correlation model of optical parameters with respect to wavelength and components based on the model coefficients that continuously change with wavelength.

[0110] Optionally, the database construction module is also used to obtain the corresponding measured reflectance spectrum and inversion result after an epitaxial growth process is completed; establish corresponding data of growth parameters and spectra based on the inversion result and measured reflectance spectrum, and write the corresponding data as feedback samples into the sample set of the preset database; and perform parameter correction or model refitting on the associated model in the preset database based on the sample set to update the associated model.

[0111] Optionally, the processing module calculates the target difference based on the predicted reflectance spectrum and the measured reflectance spectrum, and updates the growth parameters step by step based on the target difference, including: obtaining a residual sequence based on the difference between the measured reflectance spectrum and the predicted reflectance spectrum at each sampling point; constructing an objective function based on the residual sequence, wherein the objective function is a function that sums the squares of each element of the residual sequence according to a preset noise weight; calculating the target difference value of the objective function at the current growth parameter; calculating the sensitivity information of the predicted reflectance spectrum relative to the growth parameter, and constructing a least-squares incremental solution equation based on the sensitivity information, the residual sequence and its corresponding noise weight, and solving for the parameter update amount of the growth parameter; and updating the growth parameter based on the parameter update amount.

[0112] Optionally, the above-mentioned inversion system for semiconductor epitaxial layer growth parameters further includes a visualization module.

[0113] The visualization module is used to generate and update the growth parameter change curve in real time based on the inversion results of each iteration. The change curve is compared with the preset target process curve to determine whether the design intent of the preset structure has been executed in the same growth process. When the comparison deviation exceeds the preset deviation range, a prompt message is output.

[0114] The aforementioned semiconductor epitaxial layer growth parameter inversion system pre-sets optical parameters as functions of wavelength and composition, storing them in a database. During each iteration, it dynamically queries the database based on the current growth parameters, then calculates the predicted reflection spectrum based on these optical parameters and the transfer matrix. This allows the predicted reflection spectrum to match the actual changes in the material's optical constants with compositional fluctuations in real time during iteration, thus avoiding the problem of traditional fixed optical constant models misinterpreting refractive index changes as thickness or growth rate anomalies during epitaxial growth. Furthermore, the dynamic query mechanism of this application enables thickness and composition to be synchronously updated and jointly converged under a unified objective function and physical model constraint, thereby continuously outputting higher accuracy and more physically consistent composition and thickness inversion results during the growth process.

[0115] Each module in the aforementioned semiconductor epitaxial layer growth parameter inversion system can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the computer device's memory as software, so that the processor can call and execute the operations corresponding to each module.

[0116] In one feasible embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 5 As shown, the computer device includes a processor, memory, input / output interface, communication interface, display unit, and input device. The processor, memory, and input / output interface are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interface. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The input / output interface is used for exchanging information between the processor and external devices. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, NFC (Near Field Communication), or other technologies. When the computer program is executed by the processor, it implements the aforementioned method for inverting semiconductor epitaxial layer growth parameters. The display unit is used to form a visually visible image and can be a display screen, projection device, or virtual reality imaging device. The display screen can be an LCD screen or an e-ink screen. The input device of the computer device can be a touch layer covering the display screen, or buttons, trackballs, or touchpads set on the casing of the computer device, or external keyboards, touchpads, or mice, etc.

[0117] Those skilled in the art will understand that Figure 5 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0118] In one feasible embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the method steps in the above-described method for inverting semiconductor epitaxial layer growth parameters.

[0119] In one feasible embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the method steps in the above-described method for inverting semiconductor epitaxial layer growth parameters.

[0120] In one feasible embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the method steps in the above-described method for inverting semiconductor epitaxial layer growth parameters.

[0121] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0122] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for inverting growth parameters of a semiconductor epitaxial layer, characterized in that, The method includes: The measured reflectance spectrum and growth parameters corresponding to the current iteration number during semiconductor epitaxial growth are obtained; the growth parameters include thickness and composition. Based on the growth parameters, optical parameters are obtained from a preset database. The optical parameters include refractive index and extinction coefficient, and the optical parameters are functions of wavelength and composition. The predicted reflection spectrum is calculated based on the preset transmission matrix, the optical parameters, and the thickness. The target difference is calculated based on the predicted reflectance spectrum and the measured reflectance spectrum. The growth parameters are then updated step by step based on the target difference until the iteration condition is met. The value of the growth parameters when the iteration condition is met is taken as the inversion result.

2. The method according to claim 1, characterized in that, The construction method of the preset database includes: Obtain calibration data for multiple sets of standard samples, wherein the composition of the standard samples is known; The optical constants of each of the standard samples are measured within a preset wavelength range to obtain the corresponding optical parameters. Based on the composition of each standard sample and its corresponding optical parameters, a fitting process is performed to establish a correlation model between the optical parameters and the wavelength and composition. Each associated model is written into the database to form the preset database.

3. The method according to claim 2, characterized in that, The fitting process based on the components and corresponding optical parameters of each standard sample establishes a correlation model between the optical parameters and the wavelength and components, including: Select multiple sampling points within a preset wavelength range; For each sampling point, the optical parameters of each standard sample at that sampling point are summarized, and a parameterized fitting function is constructed with the composition as the independent variable and the optical parameters as the dependent variable. An error objective is constructed based on the parameterized fitting function, and the fitting coefficient at the sampling point is obtained by minimizing the error objective. Interpolation or continuous fitting is performed on the fitting coefficients corresponding to each sampling point along the wavelength direction to obtain model coefficients that change continuously with wavelength. The correlation model of the optical parameters with respect to wavelength and composition is generated based on the model coefficients that vary continuously with wavelength.

4. The method according to claim 2, characterized in that, The method further includes: After one epitaxial growth process is completed, the corresponding measured reflectance spectrum and inversion results are obtained; Based on the inversion results and the measured reflectance spectra, corresponding data of growth parameters and spectra are established, and the corresponding data are written into the sample set of the preset database as feedback samples. Based on the sample set, parameter correction or model refitting is performed on the associated models in the preset database to update the associated models.

5. The method according to claim 1, characterized in that, The step of calculating the target difference based on the predicted reflectance spectrum and the measured reflectance spectrum, and updating the growth parameters based on the target difference, includes: A residual sequence is obtained based on the difference between the measured reflectance spectrum and the predicted reflectance spectrum at each sampling point; An objective function is constructed based on the residual sequence, wherein the objective function is a function that sums the squares of each element of the residual sequence according to a preset noise weight. Calculate the target difference value of the objective function at the current growth parameters; The sensitivity information of the predicted reflectance spectrum relative to the growth parameters is calculated, and based on the sensitivity information, the residual sequence and its corresponding noise weights, a least squares incremental solution equation is constructed to obtain the parameter update amount of the growth parameters. The growth parameters are updated based on the parameter update amount.

6. The method according to claim 1, characterized in that, The determination method for satisfying the iteration conditions includes: if the target difference is less than a preset threshold or the number of iterations meets a preset number, then the iteration conditions are considered to be satisfied.

7. The method according to claim 1, characterized in that, The method further includes: Based on the inversion results of each iteration, a change curve of growth parameters is generated and updated in real time; wherein, the change curve is compared and displayed with the preset target process curve to determine whether the design intent of the preset structure has been executed in the same growth process; and a prompt message is output when the comparison deviation exceeds the preset deviation range.

8. A system for inverting growth parameters of a semiconductor epitaxial layer, characterized in that, The system includes: The acquisition module is used to acquire the measured reflection spectrum and growth parameters corresponding to the current iteration number during semiconductor epitaxial growth; the growth parameters include thickness and composition. The query module is used to retrieve optical parameters from a preset database based on the growth parameters. The optical parameters include refractive index and extinction coefficient, and the optical parameters are functions of wavelength and composition. The processing module is used to calculate the predicted reflection spectrum based on the preset transmission matrix, the optical parameters, and the thickness; calculate the target difference based on the predicted reflection spectrum and the measured reflection spectrum; gradually update the growth parameters based on the target difference until the iteration condition is met; and use the value of the growth parameters when the iteration condition is met as the inversion result.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1-7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1-7.

Images