Single crystal diamond deposition quality intelligent detection method based on OES spectrum

Abstract

The invention discloses a single crystal diamond deposition quality intelligent detection method based on OES spectrum, and the method comprises the steps: extracting OES spectrum data in a deposition area, carrying out the preprocessing, extracting spectral features from the preprocessed OES spectrum data through a convolutional neural network, and calculating plasma feature parameters; the plasma characteristic parameters comprise electron density and electron temperature, constructing a time sequence prediction model, obtaining a correlation model of the plasma characteristic parameters and deposition indexes, detecting the deposition quality according to the correlation model, and performing feedback control on the process parameters according to the detection result. According to the method, by integrating an optical emission spectrum technology and an intelligent algorithm, the limitation of a traditional detection method is overcome, dynamic analysis of plasma parameters and high-precision prediction of deposition quality are achieved, and meanwhile good interpretability is achieved.

Description

Technical Field

[0001] The present invention relates to the field of diamond deposition quality, and in particular to an intelligent detection method for single crystal diamond deposition quality based on OES spectroscopy. Background Art

[0002] Microwave plasma chemical vapor deposition (MPCVD) technology is the core process for preparing high-quality single-crystal diamond. The plasma state during the deposition process directly affects the crystallization quality of diamond.

[0003] Existing technologies often rely on offline testing and manual experience to assess deposition quality. This leads to lags, an inability to adjust process parameters in real time, and low analytical efficiency, making it difficult to meet high-precision, real-time requirements. Furthermore, commonly used traditional manual spectral analysis methods are inefficient and unable to capture plasma states in real time. Furthermore, the lack of a closed-loop feedback mechanism in the MPCVD process leads to high crystal defect rates.

[0004] Therefore, to overcome the above problems, the present invention proposes an intelligent detection method for single-crystal diamond deposition quality based on OES spectroscopy. By combining plasma emission spectroscopy (OES spectroscopy) with intelligent algorithms, it breaks through the limitations of traditional offline detection methods and realizes dynamic analysis of plasma parameters and high-precision prediction of deposition quality. Summary of the Invention

[0005] The purpose of the present invention is to provide an intelligent detection method for single crystal diamond deposition quality based on OES spectroscopy.

[0006] To achieve the above object, the present invention is implemented according to the following technical solutions: The present invention provides an intelligent detection method for single crystal diamond deposition quality based on OES spectroscopy, comprising: Extract OES spectrum data in the deposition area and perform preprocessing; Extracting spectral features from the preprocessed OES spectral data using a convolutional neural network and calculating plasma characteristic parameters; the plasma characteristic parameters include electron density and electron temperature; Constructing a time series prediction model for deposition quality indicators based on plasma characteristic parameters; The deposition quality is predicted according to the time series prediction model, and the deposition process parameters are feedback controlled according to the prediction results.

[0007] Preferably, the method further comprises: comparing the prediction result with the detection results of offline scanning electron microscopy and photoluminescence spectroscopy to ensure that the evaluation accuracy is not lower than a preset threshold.

[0008] Preferably, the extracting of OES spectrum data is specifically extracting the plasma emission spectrum in the 200 to 1000 nm band.

[0009] Preferably, the preprocessing method is noise filtering, baseline correction and wavelength calibration.

[0010] Preferably, the spectral characteristics include peak intensity, peak-to-valley ratio and continuous radiation background.

[0011] Preferably, the method for calculating the plasma characteristic parameters is specifically as follows: Obtained from the emission spectrum of hydrogen atoms Spectral lines and spectral lines; According to the Stark broadening effect, Determination of spectral characteristics of spectral lines The half-width data of the spectral line is used to calculate the electron density:

[0012] Where, is the electron density, for The sum of the areas at half-height width of the spectral line; right Spectral lines and The electron temperature is calculated using the double-line method, where the double-line intensity of the same particle migrating at the same energy level is expressed as:

[0013] Where, and They are Spectral lines and The intensity of the spectral line, and They are Spectral lines and The transition probability of the spectral line, and They are Spectral lines and The statistical weight of the spectral line, and They are Spectral lines and The central wavelength of the spectral line, and They are Spectral lines and The excitation energy of the spectral line, is the Boltzmann constant, is the desired electron temperature.

[0014] Preferably, the method for constructing a time series prediction model of deposition quality indicators based on plasma characteristic parameters is specifically as follows: Deposition rate and crystal defect density were selected as deposition quality indicators; Taking electron density, electron temperature and historical process data as input parameters and deposition quality indicators as output, a time series prediction model is constructed based on the long short-term memory network. Obtain historical deposition data including Raman spectroscopy and X-ray diffraction results for offline characterization; The time series prediction model is supervised and trained through transfer learning based on historical sedimentation data, and the loss function is optimized to minimize the prediction error.

[0015] Preferably, the method for performing feedback control on deposition process parameters according to the prediction results includes: When the predicted deposition quality index deviates from the preset threshold, the deposition process parameters are stably controlled by the PID control algorithm until the predicted result meets the preset threshold; If the rate of change of the spectral data or the model confidence of the time series prediction model exceeds a preset threshold, an alarm is triggered and the deposition process is suspended.

[0016] Compared with the prior art, the embodiments of the present invention have at least the following advantages or beneficial effects: (1) The present invention trains a time series prediction model by combining plasma characteristic parameters with process parameters to obtain a correlation model between OES spectrum and deposition quality; (2) The present invention constructs a mapping model between OES spectrum and deposition quality, and uses AI technology to extract features and recognize patterns in spectral data to achieve real-time optimization of deposition process parameters, thereby improving the growth rate and crystal quality of single-crystal diamonds while reducing production energy consumption and scrap rate. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 A flow chart of the intelligent detection method for single crystal diamond deposition quality based on OES spectroscopy provided by the present invention; Figure 2 Schematic diagram of calculating the full width at half area of ​​a given spectral line in an embodiment of the present invention. DETAILED DESCRIPTION

[0018] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0019] Herein, suffixes such as "module," "component," or "unit" used to represent elements are only used to facilitate description of the present invention and have no specific meaning. Therefore, "module," "component," or "unit" may be used interchangeably.

[0020] As used herein, terms such as "upper," "lower," "inner," "outer," "front," "back," "one end," and "the other end" indicate positions or locations based on those shown in the accompanying drawings. These terms are intended solely to facilitate and simplify the description of the present invention and are not intended to indicate or imply that the devices or components referred to must have, be constructed, or operate in a specific orientation. Therefore, they should not be construed as limitations on the present invention. Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0021] As used herein, "and / or" includes any and all combinations of one or more of the associated listed items.

[0022] Herein, "plurality" means two or more than two, ie, it includes two, three, four, five, etc.

[0023] It should be noted that, in this document, the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, article, or apparatus comprising a series of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, method, article, or apparatus comprising the element.

[0024] As used herein, unless otherwise expressly specified or limited, the terms "installed," "provided with," and "connected" should be understood broadly. For example, "connected" may refer to a fixed connection, a detachable connection, or an integral connection; it may refer to a mechanical connection, a direct connection, an indirect connection via an intermediate medium, or internal communication between two components. Those skilled in the art will understand the specific meanings of the above terms in the present invention on a case-by-case basis.

[0025] In this article, the “deposition area” refers to the plasma action area on and above the substrate surface in the MPCVD reaction chamber, that is, the spatial range where single crystal diamond actually grows.

[0026] In this article, "OES spectrum" refers to the plasma emission spectrum, whose spectral data is derived from the characteristic light radiation released when excited atoms or molecules in the plasma deexcite. In the deposition area, microwaves excite the gas to produce plasma, in which particles such as electrons, ions, and atoms gain energy from collisions and become excited. When the excited particles return to the ground state, they release photons of specific wavelengths, forming characteristic spectral lines corresponding to the particle type and energy state. These light signals are collected by a spectrometer to generate OES spectral data (wavelength on the horizontal axis and light intensity on the vertical axis), reflecting the composition and state of the plasma.

[0027] Reference Figure 1 As shown, the present invention provides an intelligent detection method for single crystal diamond deposition quality based on OES spectroscopy, comprising: OES spectrum data was extracted in the deposition area and preprocessed.

[0028] In this embodiment, an MPCVD reaction chamber is configured, including a microwave generator, a gas input system, and a substrate stage; The microwave generator generates high-frequency microwaves (commonly 2.45 GHz and 915 MHz) and transmits them to the MPCVD reaction chamber via a waveguide. The microwave energy excites gas molecules in the chamber, ionizing them to form plasma, providing the high-energy active particles required for diamond deposition. The gas input system is responsible for controlling and inputting reaction gases, mainly including carbon source gas, auxiliary gas and dilution gas. The system uses mass flow controllers to proportion the flow of each gas to ensure uniform distribution of gas in the chamber and maintain a stable chemical reaction environment. The substrate stage is used to support the substrate material for growing single crystal diamond. By controlling the temperature, rotation speed and position of the substrate stage, it ensures uniform temperature on the substrate surface and promotes uniform nucleation and growth of diamond on the substrate. In addition, the substrate stage is usually equipped with a temperature control system, which can control the temperature within a suitable range (usually 800 to 1200 °C) according to the requirements of the deposition process. ) to ensure the growth quality of single crystal diamond; The three modules work together to create a high-temperature, highly active plasma environment in the MPCVD reaction chamber, providing the necessary conditions for high-quality deposition of single-crystal diamond.

[0029] Specifically, extracting OES spectrum data specifically involves extracting the plasma emission spectrum within the 200 to 1000 nm band; the preprocessing method includes performing noise filtering, baseline correction, and wavelength calibration.

[0030] Spectral features are extracted from the preprocessed OES spectral data using a convolutional neural network, and plasma characteristic parameters are calculated; the plasma characteristic parameters include electron density and electron temperature.

[0031] Specifically, spectral characteristics include peak intensity, peak-to-valley ratio, and continuous radiation background.

[0032] In this embodiment, the emission spectrum of hydrogen atoms is obtained. Spectral lines and The spectrum is shown in Table 1: Table 1 Hydrogen spectral line parameters

[0033] According to the Stark broadening effect, Determination of spectral characteristics of spectral lines The half-width data of the spectral line is used to calculate the electron density:

[0034] Where, is the electron density, for FWHM of spectral lines; right Spectral lines and The electron temperature is calculated using the double-line method, where the double-line intensity of the same particle migrating at the same energy level is expressed as:

[0035] Where, and They are Spectral lines and The intensity of the spectral line, and They are Spectral lines and The transition probability of the spectral line, and They are Spectral lines and The statistical weight of the spectral line, and They are Spectral lines and The central wavelength of the spectral line, and They are Spectral lines and The excitation energy of the spectral line, is the Boltzmann constant, is the desired electron temperature.

[0036] A time series prediction model of deposition quality indicators is constructed based on plasma characteristic parameters.

[0037] Specifically, the method for constructing a time series prediction model of deposition quality indicators based on plasma characteristic parameters is as follows: Deposition rate and crystal defect density were selected as deposition quality indicators; Taking electron density, electron temperature and historical process data as input parameters and deposition quality indicators as output, a time series prediction model is constructed based on the long short-term memory network. Obtain historical deposition data including Raman spectroscopy and X-ray diffraction results for offline characterization; The time series prediction model is supervised and trained through transfer learning based on historical sedimentation data, and the loss function is optimized to minimize the prediction error.

[0038] In some embodiments, Both the full width at half maximum (FWHM) and the area summed at half maximum (FWHA) of a spectral line can be used to calculate plasma characteristics. There are only slight differences in the calculation methods between the two, but the area summed at half maximum is less affected by ion motion and is more suitable for calculating plasma electron density.

[0039] like Figure 2 As shown, in the upper half of the image, the horizontal axis is Represents the wavelength shift of the spectrum; the vertical axis is , represents the spectral intensity under the corresponding wavelength offset. The curve in the figure presents a single peak shape, which is the intensity distribution curve of the spectral line, and the peak of the curve represents the position with the maximum intensity of the spectrum; Figure 2 In the lower half of the image, the horizontal axis is , the vertical axis is , which means from negative infinity to Spectral intensity Integrate to get the accumulated spectral intensity; and and The spectral intensities corresponding to these two points make the curve arrive The area under the curve is half the total area. and The difference between the two is the full width at half area; and That is, corresponding to the 1 / 4 and 3 / 4 positions in the lower half of the image, through this integral curve, we can more intuitively find the frequency or wavelength offset range that makes the area under the curve half of the total area, and then accurately calculate the full width at half area (FWHA).

[0040] The deposition quality is predicted according to the time series prediction model, and the deposition process parameters are feedback controlled according to the prediction results.

[0041] Specifically, when the prediction results of the time series prediction model deviate from the preset threshold, the system automatically adjusts the process parameters of microwave power and methane concentration, and achieves stable control through the PID algorithm; when the change rate of spectral data or the model confidence of the time series prediction model exceeds the preset threshold, an alarm is triggered and the deposition process is suspended.

[0042] In some embodiments, the method further includes comparing the prediction results with the detection results of offline scanning electron microscopy and photoluminescence spectroscopy to ensure that the evaluation accuracy is not lower than a preset threshold.

[0043] The above content is merely an example and explanation of the structure of the present invention. Those skilled in the art may make various modifications or additions to the described specific embodiments or replace them in a similar manner. As long as they do not deviate from the structure of the invention or exceed the scope defined by the claims, they should all fall within the scope of protection of the present invention.

Claims

1. An intelligent detection method for single crystal diamond deposition quality based on OES spectroscopy, characterized in that: The following steps are involved: Extract OES spectrum data in the deposition area and perform preprocessing; Extracting spectral features from the preprocessed OES spectral data using a convolutional neural network and calculating plasma characteristic parameters; the plasma characteristic parameters include electron density and electron temperature; Constructing a time series prediction model for deposition quality indicators based on plasma characteristic parameters; The deposition quality is predicted according to the time series prediction model, and the deposition process parameters are feedback controlled according to the prediction results.

2. The method for intelligent detection of single crystal diamond deposition quality based on OES spectroscopy according to claim 1, characterized in that: The method further comprises: The prediction results are compared with the detection results of offline scanning electron microscopy and photoluminescence spectroscopy to ensure that the evaluation accuracy is not lower than the preset threshold.

3. The method for intelligent detection of single crystal diamond deposition quality based on OES spectroscopy according to claim 1, characterized in that: The extracting of OES spectrum data specifically involves extracting the plasma emission spectrum within the 200 to 1000 nm band.

4. The method for intelligent detection of single crystal diamond deposition quality based on OES spectroscopy according to claim 1, characterized in that: The preprocessing method includes performing noise filtering, baseline correction and wavelength calibration.

5. The method for intelligent detection of single crystal diamond deposition quality based on OES spectroscopy according to claim 1, characterized in that: The spectral characteristics include peak intensity, peak-to-valley ratio and continuous radiation background.

6. The method for intelligent detection of single crystal diamond deposition quality based on OES spectroscopy according to claim 1, characterized in that: The method for calculating plasma characteristic parameters is specifically as follows: Obtained from the emission spectrum of hydrogen atoms Spectral lines and spectral lines; According to the Stark broadening effect, Determination of spectral characteristics of spectral lines The half-width data of the spectral line is used to calculate the electron density: Where, is the electron density, for FWHM of spectral lines; right Spectral lines and The electron temperature is calculated using the double-line method, where the double-line intensity of the same particle migrating at the same energy level is expressed as: Where, and They are Spectral lines and The intensity of the spectral line, and They are Spectral lines and The transition probability of the spectral line, and They are Spectral lines and The statistical weight of the spectral line, and They are Spectral lines and The central wavelength of the spectral line, and They are Spectral lines and The excitation energy of the spectral line, is the Boltzmann constant, is the desired electron temperature.

7. The method for intelligent detection of single crystal diamond deposition quality based on OES spectroscopy according to claim 1, characterized in that: The method for constructing a time series prediction model for deposition quality indicators based on plasma characteristic parameters is specifically as follows: Deposition rate and crystal defect density were selected as deposition quality indicators; Taking electron density, electron temperature and historical process data as input parameters and deposition quality indicators as output, a time series prediction model is constructed based on the long short-term memory network. Obtain historical deposition data including Raman spectroscopy and X-ray diffraction results for offline characterization; The time series prediction model is supervised and trained through transfer learning based on historical sedimentation data, and the loss function is optimized to minimize the prediction error.

8. The method for intelligent detection of single crystal diamond deposition quality based on OES spectroscopy according to claim 1, characterized in that: The method for performing feedback control on deposition process parameters according to the prediction results comprises: When the predicted deposition quality index deviates from the preset threshold, the deposition process parameters are stably controlled by the PID control algorithm until the predicted result meets the preset threshold; If the rate of change of the spectral data or the model confidence of the time series prediction model exceeds a preset threshold, an alarm is triggered and the deposition process is suspended.

Images