Peak extraction method, thickness measurement method and equipment

Abstract

The present invention provides a peak extraction method, a thickness measurement method, and an apparatus. The method is applied to the field of high-precision measurement. The method comprises: obtaining the main peak value and the secondary peak value of the spectral signal of the N-th frame wafer and the spectral signal of the N-1th frame, where N is an integer greater than or equal to 2; obtaining preset parameters, and using the main peak value, the secondary peak value, and the preset parameters of the spectral signal of the N-1th frame as initial guesses of a bimodal pseudo Voigt fitting equation, fitting the spectral signal of the N-th frame to obtain the main peak value and the secondary peak value of the spectral signal of the N-th frame. The present invention improves the accuracy of obtaining the main peak value and the secondary peak value of the spectral signal of the N-th frame.

Description

Technical Field

[0001] The present invention relates to the field of high-precision measurement, and in particular to a peak extraction method, a thickness measurement method and equipment. Background Art

[0002] Thinning machines are crucial in the semiconductor manufacturing industry. As chip manufacturing technology evolves, the need for precise control of wafer thickness becomes increasingly crucial. Thinning machines physically or chemically remove material from the wafer surface to adjust its thickness to meet the needs of different chip products. Continuous thinning operations, in particular, require precise measurement of wafer thickness to accurately control the thinning process.

[0003] Spectral analysis technology, with its high precision and reliability, has become a key tool for accurate wafer thickness measurement. Traditional spectral analysis typically extracts the peaks of the wafer surface spectrum to obtain the required information. Spectral analysis can accurately separate and identify the peak information within the spectrum, thereby precisely determining the thickness of each wafer layer. This not only helps ensure wafer quality but also provides a strong basis for optimizing the manufacturing process, driving wafer manufacturing to a higher level.

[0004] However, existing peak extraction methods have significant flaws. Spectral data often contains similar peaks. These peaks may originate from reflections from different layers of the wafer, or they may be interference signals generated during the measurement process. Because the distances between them are small, they intertwine and overlap, making it extremely difficult to accurately extract the peak positions. The inability to accurately extract the positions of the two peaks directly affects the calculation of wafer thickness based on spectral data, and inaccurate thickness calculations hinder the precise control of the wafer thinning process, seriously affecting the thinning effect. Summary of the Invention

[0005] In view of this, the present invention provides a peak extraction method, comprising:

[0006] Obtain the main peak-to-peak value and the secondary peak-to-peak value of the spectrum signal of the wafer in the Nth frame and the spectrum signal of the N-1th frame, where N is an integer greater than or equal to 2;

[0007] Obtain preset parameters, and use the main peak peak value, the secondary peak peak value, and the preset parameters of the N-1th frame spectral signal as initial guess values ​​of the bimodal pseudo Voigt fitting equation, fit the Nth frame spectral signal, and obtain the main peak peak value and secondary peak peak value of the Nth frame spectral signal.

[0008] Optionally, the preset parameters include the maximum intensity of the spectral signal, the proportion of Gaussian signals in the spectral signal, and the width of the peak in the spectral signal.

[0009] Optionally, the bimodal pseudo Voigt fitting equation is:

[0010] ;

[0011] in, Represents the intensity at any wavelength in the Nth frame spectral signal, and Indicates the maximum intensity of the spectral signal, and Indicates the proportion of Gaussian signals in the spectral signal, and represents the main peak and secondary peak in the spectral signal, and represents the width of the peak in the spectral signal, is the wavelength independent variable in the Nth frame spectral signal.

[0012] Optionally, the peak extraction method of the present invention further includes:

[0013] Acquire the first frame of spectrum signal of the wafer;

[0014] Perform peak extraction on the first frame spectral signal according to a preset intensity threshold;

[0015] When a peak is extracted, the peak value is determined to be the main peak peak;

[0016] Get the thickness guess;

[0017] The secondary peak value is obtained by reversely solving the thickness equation based on the thickness guess value and the primary peak value;

[0018] Using the main peak value and the secondary peak value as initial guess values ​​of a bimodal pseudo Voigt fitting equation, fitting the first frame of spectral signal to obtain the main peak value and the secondary peak value of the first frame of spectral signal;

[0019] When two peaks are extracted, the peak values ​​of the two peaks are determined to be the main peak peak and the secondary peak peak respectively;

[0020] The main peak peak value and the secondary peak peak value are used as the initial guess values ​​of the bimodal pseudo Voigt fitting equation, and the first frame spectral signal is fitted to obtain the main peak peak value and the secondary peak peak value of the first frame spectral signal.

[0021] Optionally, the thickness equation is:

[0022] ;

[0023] in, Indicates the wafer thickness, The main peak value, is the sub-peak value, 、 、 、 、 、 、 、 These are the parameter results obtained by fitting the measurement data of the translation stage and spectrometer.

[0024] Optionally, the peak extraction method of the present invention further includes:

[0025] Obtain the main peak value, main peak width, secondary peak value and secondary peak width of the spectral signal of the Nth frame and the N-1th frame;

[0026] Differences between the main peak value and the secondary peak value of the two frames are obtained to obtain the main peak peak difference and the secondary peak peak difference;

[0027] respectively determining whether the main peak-to-peak difference and the secondary peak-to-peak difference are greater than or equal to a preset percentage of the width of the peak corresponding to any frame;

[0028] If the peak difference of any peak is greater than or equal to a preset percentage of the width of the peak corresponding to any frame, all parameters obtained from the bimodal pseudo-Voigt fitting equation of a preset number of frames before the Nth frame are linearly extrapolated to obtain an initial guess of the bimodal pseudo-Voigt fitting equation of the Nth frame;

[0029] The Nth frame spectral signal is fitted using the initial guess value of the bimodal pseudo Voigt fitting equation to obtain the main peak peak value and the secondary peak peak value of the Nth frame spectral signal.

[0030] Optionally, before fitting the Nth frame spectral signal using the bimodal pseudo Voigt fitting equation, the method further includes:

[0031] Acquiring background spectral signals originating from measurement conditions;

[0032] Subtracting the background spectrum signal from the Nth frame spectrum signal of the wafer to obtain a true spectrum signal of the wafer in the Nth frame;

[0033] Performing main peak judgment on the real spectrum signal;

[0034] If there is a main peak, the real spectrum signal is smoothed according to a preset smoothing intensity;

[0035] If there is no main peak, the peak extraction of the spectrum signal of the Nth frame is stopped.

[0036] Optionally, after fitting the Nth frame spectral signal, the method further includes:

[0037] Determining whether the bimodal pseudo Voigt fitting equation converges;

[0038] If the bimodal pseudo Voigt fitting equation does not converge, the fitting is determined to be failed, and the peak value of the Nth frame spectral signal is extracted according to the steps of processing the first frame spectral signal;

[0039] If the bimodal pseudo Voigt fitting equation converges, all the parameters obtained by fitting are respectively subjected to difference calculation with all the parameters obtained by fitting of the N-1th frame;

[0040] Determine whether the difference of each parameter is greater than the difference threshold;

[0041] If the difference of any parameter is greater than the difference threshold, the fitting is determined to have failed, and the peak value of the Nth frame spectral signal is extracted according to the steps of processing the first frame spectral signal.

[0042] A second aspect of the present invention provides a thickness measurement method, the method comprising:

[0043] Obtain the main peak value and the secondary peak value of the Nth frame using the method described above;

[0044] The wafer thickness of the Nth frame is calculated according to the main peak value and the secondary peak value.

[0045] Optionally, if the bimodal pseudo Voigt fitting equation fails to fit, the wafer thickness of a preset number of frames before the Nth frame is obtained and linearly extrapolated to obtain the wafer thickness of the Nth frame.

[0046] The third aspect of the present invention provides a peak extraction device, which includes: a processor and a memory connected to the processor; wherein the memory stores instructions that can be executed by the processor, and the instructions are executed by the processor to enable the processor to perform the above-mentioned peak extraction method.

[0047] A fourth aspect of the present invention provides a thickness measuring device, comprising: a processor and a memory connected to the processor; wherein the memory stores instructions executable by the processor, and the instructions are executed by the processor to enable the processor to perform the above-mentioned thickness measurement method.

[0048] A fifth aspect of the present invention provides a thinning machine for thinning a wafer and performing any of the above methods during the thinning process.

[0049] The present invention addresses the problem of being unable to accurately separate the two peaks when extracting the peak values ​​of similar peaks in the Nth frame spectral signal. By directly obtaining the main peak peak value and the secondary peak peak value of the N-1th frame spectral signal, and then fitting the main peak peak value and the secondary peak peak value of the N-1th frame as the initial guess of the Nth frame, the spectral signal of the Nth frame can be quickly fitted. Since the wafer state changes reflected by the spectral signals of two adjacent frames are usually relatively small and continuous, the peak value of the N-1th frame provides a starting point close to the true value for the fitting of the Nth frame. Compared with arbitrarily setting the initial guess value without prior information, the peak value of the N-1th frame selected based on the correlation of continuous measurement data effectively avoids the dilemma of similar peak interference caused by inaccurate initial guesses in traditional fitting methods, greatly improves the accuracy and reliability of fitting, improves the accuracy of obtaining the main peak peak value and the secondary peak peak value of the Nth frame spectral signal, provides a more reliable basis for subsequent wafer thickness calculation and thinning control, and thus improves the accuracy and stability of the entire wafer thinning process, ensuring the quality and efficiency of wafer production. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the specific embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0051] Figure 1 is a flow chart of a peak extraction method in an embodiment of the present invention;

[0052] Figure 2 is a flow chart of another peak extraction method in an embodiment of the present invention;

[0053] Figure 3 This is a spectrum signal diagram when a peak is extracted in an embodiment of the present invention;

[0054] Figure 4 This is a spectrum signal diagram when two peaks are extracted in an embodiment of the present invention. DETAILED DESCRIPTION

[0055] The technical solution of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the embodiments described are only some embodiments of the present invention, not all embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0056] In the description of the present invention, it should be noted that the terms "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer," etc., indicating orientations or positional relationships, are based on the orientations or positional relationships shown in the accompanying drawings and are intended solely to facilitate and simplify the description of the present invention. They are not intended to indicate or imply that the devices or components referred to must have, be constructed, or operate in a specific orientation, and therefore should not be construed as limitations on the present invention. Furthermore, the terms "first," "second," and "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0057] In the description of the present invention, it should be noted that, unless otherwise expressly specified or limited, the terms "installed," "connected," and "connected" should be understood in a broad sense. For example, they can refer to fixed connections, detachable connections, or integral connections; mechanical connections or electrical connections; direct connections or indirect connections through an intermediate medium; internal connections between two components; wireless connections or wired connections. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on specific circumstances.

[0058] In addition, the technical features involved in the different embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0059] like Figure 1 As shown, an embodiment of the present invention provides a peak extraction method, which is executed by an electronic device such as a computer or a server, and specifically includes:

[0060] S1, obtaining the main peak value and the secondary peak value of the spectrum signal of the wafer of the Nth frame and the spectrum signal of the N-1th frame, where N is an integer greater than or equal to 2.

[0061] During the thinning process, the Nth frame of the spectrum signal of the wafer collected by the spectrometer is obtained, and the main peak-to-peak value and the secondary peak-to-peak value of the spectrum signal obtained by fitting the previous frame are obtained.

[0062] S2, obtaining preset parameters, and using the main peak peak value, secondary peak peak value, and preset parameters of the spectral signal of the N-1th frame as the initial guess values ​​of the bimodal pseudo Voigt fitting equation, fitting the spectral signal of the Nth frame to obtain the main peak peak value and secondary peak peak value of the spectral signal of the Nth frame.

[0063] This embodiment addresses the problem of being unable to accurately separate the peak values ​​of two adjacent peaks in the spectral signal of frame N. By directly obtaining the primary and secondary peak values ​​of the spectral signal of frame N-1, and then fitting the primary and secondary peak values ​​of frame N-1 as initial guesses for frame N, the spectral signal of frame N can be quickly fitted. Because the wafer state changes reflected by the spectral signals of two adjacent frames are typically relatively small and continuous, the peak value of frame N-1 provides a starting point close to the true value for fitting frame N. Compared to arbitrarily setting an initial guess value without prior information, the peak value of frame N-1 is selected based on the correlation of continuous measurement data. This effectively avoids the dilemma of interference between adjacent peaks caused by inaccurate initial guesses in traditional fitting methods, greatly improving the accuracy and reliability of the fitting, and enhancing the accuracy of obtaining the primary and secondary peak values ​​of the spectral signal of frame N. This provides a more reliable basis for subsequent wafer thickness calculation and thinning control, thereby improving the accuracy and stability of the entire wafer thinning process and ensuring the quality and efficiency of wafer production.

[0064] Specifically, the preset parameters include the maximum intensity of the spectral signal, the proportion of Gaussian signals in the spectral signal, and the width of the peak in the spectral signal.

[0065] Among them, the bimodal pseudo Voigt fitting equation for fitting the Nth frame spectral signal is:

[0066] ;

[0067] Among them, among them, Represents the intensity at any wavelength in the Nth frame spectral signal, and Indicates the maximum intensity of the spectral signal, and Indicates the proportion of Gaussian signals in the spectral signal, and represents the main peak and secondary peak in the spectral signal, and represents the width of the peak in the spectral signal, is the wavelength independent variable in the Nth frame spectral signal.

[0068] Among them, the main peak value and the secondary peak value of the spectrum signal of the N-1 frame are respectively brought into the spectrum signal in the fitting equation and , while other initial guesses in the formula, such as the maximum intensity of the spectral signal, the proportion of Gaussian signals in the spectral signal, and the width of the peak in the spectral signal, can be directly used as empirical values.

[0069] When fitting the Nth frame spectral signal, it is necessary to continuously fit the double-peak pseudo Voigt fitting equation based on the wavelength in the Nth frame signal as the independent variable and the corresponding intensity until the data of the two peaks of the Nth frame spectral signal are successfully fitted. The fitting equation contains eight parameters, among which, and The accuracy of the fitting equation determines the probability of fitting the bimodal pseudoVoigt fitting equation. Because of continuous thinning, the two peaks of the spectral signal in frame N can be accurately and successfully fitted directly based on the primary and secondary peak values ​​obtained in frame N-1. The initial guesses for the remaining parameters can be fitted successfully with high probability even if they deviate significantly from the actual values. Empirical values ​​can be used directly, without the need for specific data analysis to estimate them.

[0070] like Figure 2 As shown, the embodiment of the present invention also provides a peak extraction method, which specifically includes:

[0071] S1a, obtain the first frame of spectrum signal of the wafer.

[0072] This embodiment is a process of performing peak extraction on the spectrum signal of the initial frame before the Nth frame.

[0073] S2a, extracting the peak of the first frame spectral signal according to a preset intensity threshold, and determining whether it is one peak; when one peak is extracted, executing step S3a; when two peaks are extracted, executing step S7a.

[0074] The preset intensity threshold is an empirical value. You can select the corresponding empirical value according to the experiment type. For example, if all experiments are spectral confocal, the same set of threshold empirical values ​​can be used.

[0075] S3a, determine the peak value of the peak as the main peak peak value.

[0076] S4a, obtaining a thickness guess value.

[0077] The thickness guess value can be obtained in many ways, such as: empirical values ​​obtained from multiple measurements or experiments, approximate thickness parameters measured by other contact measurement methods, and an approximate initial thickness value added to the wafer in the initial frame.

[0078] S5a, use the thickness equation to reversely solve the secondary peak value based on the thickness guess value and the main peak value.

[0079] If only one peak is extracted, it means that the distance between the primary and secondary peaks is small and the secondary peak is covered by the primary peak. A relatively accurate secondary peak value can be reversely solved by using the thickness guess value.

[0080] S6a, using the main peak peak value and the secondary peak peak value as the initial guess values ​​of the bimodal pseudo Voigt fitting equation, fitting the first frame spectral signal, and obtaining the main peak peak value and the secondary peak peak value of the first frame spectral signal.

[0081] Specifically, when the peak value of the spectral signal of the first frame is extracted, it is proved that there is no data from the previous frame as a reference, so we can only obtain an inaccurate main peak value and a thickness guess value based on the peak extraction to reversely calculate the less accurate secondary peak value. The spectral signal is fitted again using the double-peak pseudo Voigt fitting equation to obtain the true main peak value and secondary peak value of the spectral signal.

[0082] In this embodiment, when extracting the peak value of the spectral signal of the first frame, a preset intensity threshold is used for peak extraction. When only one peak is extracted, it indicates that the distance between the primary and secondary peaks is small and the secondary peak is covered by the primary peak. Therefore, the position of the secondary peak needs to be found. Specifically, the peak value of the peak can be set as the primary peak peak value and the thickness guess value is set. The secondary peak peak value is obtained by reversely solving the thickness equation. Finally, the primary peak peak value and the secondary peak peak value are used as the initial guesses of the bimodal pseudo Voigt fitting equation to accurately fit the spectral signal of the first frame. This successfully solves the problem of missing initial guesses and the difficulty in accurately identifying the secondary peak position due to the proximity of the two peaks. The primary peak peak value and the secondary peak peak value of the spectral signal of the first frame can be accurately fitted.

[0083] Further, S7a, the peak values ​​of the two peaks are determined to be the main peak peak value and the secondary peak peak value.

[0084] When two peaks were extracted, the first peak was taken as the main peak and the second peak as the secondary peak.

[0085] S8a, using the main peak peak value and the secondary peak peak value as the initial guess values ​​of the bimodal pseudo Voigt fitting equation, fitting the first frame spectral signal, and obtaining the main peak peak value and the secondary peak peak value of the first frame spectral signal.

[0086] In this embodiment, when extracting peak values ​​from the spectral signal of the first frame, if two peaks are extracted, the peak values ​​of the primary peak and the secondary peak are precisely determined. These two peaks can be directly used as initial guesses for the bimodal pseudo-Voigt fitting equation to fit the spectral signal of the first frame. This effectively solves the problem of missing initial guesses for the spectral signal of the first frame and enables more accurate fitting of the peak values ​​of the primary peak and the secondary peak in the spectral signal of the first frame.

[0087] Furthermore, the thickness equation is:

[0088] ;

[0089] in, Indicates thickness, The main peak value, is the sub-peak value, 、 、 、 、 、 、 、 These are the parameter results obtained by fitting the measurement data of the translation stage and spectrometer.

[0090] Substitute the main peak peak value into the main peak peak value in the formula , and the thickness of the thickness guessed value into the formula , you can reversely solve the sub-peak-peak value .

[0091] In one embodiment, after obtaining the main peak value and the secondary peak value of the spectrum signal of the Nth frame in step S2, the method further includes:

[0092] Obtain the main peak value, main peak width, secondary peak value, and secondary peak width of the spectral signal of the Nth frame and the N-1th frame.

[0093] The main peak value and the secondary peak value of the two frames are subtracted to obtain the main peak peak difference and the secondary peak peak difference.

[0094] It is determined whether the peak-to-peak difference of the main peak and the peak-to-peak difference of the secondary peak are greater than or equal to a preset percentage of the width of the peak corresponding to any frame, wherein the preset percentage is an empirical value, such as 50%.

[0095] If the peak difference of any peak is greater than or equal to a preset percentage of the width of the peak corresponding to any frame, all parameters obtained from the bimodal pseudo Voigt fitting equation of a preset number of frames before the Nth frame are linearly extrapolated to obtain the initial guess of the bimodal pseudo Voigt fitting equation of the Nth frame. The preset number of frames is preferably between the 5th and 10th frames. All parameters refer to the eight parameters of the bimodal pseudo Voigt fitting equation, such as 、 、 、 、 、 、 and .

[0096] The Nth frame spectral signal is fitted using the initial guess value of the bimodal pseudo Voigt fitting equation to obtain the main peak value and the secondary peak value of the Nth frame spectral signal.

[0097] Specifically, when the sampling frequency is too low, the time interval between two adjacent frames of spectral signals is large, and the wafer state may have changed significantly during the thinning process. Parameters such as the primary and secondary peak values ​​in the spectral signal will change as the wafer state changes. Due to the long sampling interval, the center positions of the primary and secondary peaks of the spectral signals of two adjacent frames may be quite different, which will result in inaccurate primary and secondary peak values ​​obtained by each fitting. Therefore, it is necessary to perform a difference calculation on the primary and secondary peak values ​​obtained by fitting the two frames, and compare them with the width of the corresponding peak. As long as the peak difference of a peak exceeds the preset range, it means that the fitting result of the Nth frame is inaccurate, and it is necessary to perform a linear extrapolation on the primary and secondary peak values ​​obtained by the previous fitting to obtain the primary and secondary peak values ​​of the Nth frame. In this way, when the sampling frequency is insufficient and the fitting result is inaccurate, reasonable speculation can be made using existing reliable data, effectively improving the accuracy and reliability of determining the primary and secondary peak values ​​of the spectral signal at a low sampling frequency.

[0098] In one embodiment, before fitting the Nth frame spectral signal using the bimodal pseudo Voigt fitting equation in step S2, the method further includes:

[0099] Acquires background spectral signals originating from the measurement conditions.

[0100] Specifically, the spectrometer can be used to collect signals from a thinning machine without a wafer placed thereon; or multiple groups of spectral signals from multiple samples under test can be collected, with each group of spectral signals being collected based on the same measurement conditions.

[0101] The sample being measured can be any substance. A spectrometer collects signals from the sample placed on the measuring table. Different samples are placed under the same measurement conditions to collect multiple sets of spectral signals. The background spectral signal is then extracted from these multiple sets of spectral signals, which includes an error that occurs in each set and is independent of the measurement conditions and sample. The measurement conditions encompass customizable input parameters such as specific light source intensity, wavelength range, measurement angle, ambient temperature and humidity, and resolution, ensuring diverse and representative data.

[0102] The background spectrum signal is subtracted from the spectrum signal of the wafer in the Nth frame to obtain the true spectrum signal of the wafer in the Nth frame.

[0103] The background noise signal in the Nth frame spectral signal is subtracted to reduce the interference of background error on the data and improve the accuracy of the signal.

[0104] Determine the main peak of the real spectrum signal;

[0105] If there is a main peak, the real spectrum signal is smoothed according to the preset smoothing intensity.

[0106] If there is no main peak, the peak extraction of the spectrum signal of the Nth frame is stopped.

[0107] Among them, the smoothing method can adopt low-pass filtering, wavelet transform, moving average, Savitzky-Golay filtering and other methods. The corresponding preset smoothing intensity is an empirical value. Different experimental types have a set of corresponding intensity empirical values. The specific empirical value can be selected according to the actual situation.

[0108] This embodiment can also be applied to the case where only the spectrum signal of the wafer of the first frame is acquired, and background spectrum removal and smoothing processing are performed before peak extraction is performed on the spectrum signal of the frame according to a preset intensity threshold, so as to improve the accuracy of the data.

[0109] This embodiment effectively reduces the interference of background errors on the data by acquiring and subtracting the background spectral signal from the Nth frame spectral signal, thereby improving the reliability and credibility of the entire spectral analysis results. Then, by determining the main peak, the spectral signal can be processed in a targeted manner, improving analysis efficiency and accuracy. When a main peak is present, smoothing is performed according to a preset smoothing intensity, which reduces noise interference, highlights peak features, and makes the spectral signal more accurate, providing reliable data for accurate peak extraction. If a main peak is not present, the extraction is stopped, avoiding wasted computing resources and time.

[0110] In addition, when the first frame of the spectrum signal of the wafer is obtained in step S1a, the secondary peak-to-peak value is obtained by reversely solving the thickness guess value and the main peak value using the thickness equation, which can also be:

[0111] Get the wavelength interval.

[0112] The secondary peak predicted peak value is calculated based on the wavelength interval and the main peak peak value.

[0113] The thickness equation is used to calculate the thickness based on the main peak peak value and the secondary peak predicted peak value. The wavelength interval is continuously updated according to the difference between the thickness guess value and the thickness, and then the secondary peak predicted peak value is updated until the difference between the thickness guess value and the thickness is less than the preset convergence threshold, and the secondary peak peak value that meets the accuracy requirements is obtained.

[0114] Specifically, assuming a smaller wavelength interval deltaw, the predicted secondary peak value w2' = w1 + deltaw, where w1 is the primary peak value. Substituting w1 and w2' into the thickness equation yields an inaccurate thickness calculation result t1'. The value of w2' is continuously updated based on the difference between the thickness estimates t1 and t1' until the deviation between t1 and t1' falls below a preset convergence threshold (e.g., 1e-4). At this point, the value of w2' can be considered a relatively accurate secondary peak value. w1 and w2' are the peak parameters of the bimodal pseudo Voigt fitting equation. The specific wavelength interval update method can be a general approach to optimizing arbitrary sudden function.

[0115] The following are examples of the above embodiments:

[0116] The spectral signal s1 and background spectral signal s0 of the wafer in the Nth frame are acquired; the background spectral signal s0 is subtracted to obtain the true spectral signal s2 after background removal; the main peak of the true spectral signal s2 is then determined. If a main peak exists, the true spectral signal s2 is again smoothed by Savitzky-Golay filtering to obtain a smoothed spectral signal s3. The smoothed spectral signal s3 is then fitted using the bimodal pseudo Voigt fitting equation.

[0117] like Figure 3 As shown, step S1a captures the first frame of the wafer's spectral signal, but fails to identify the second peak. Only one peak is extracted, resulting in a wavelength of 550.0 nm for the primary peak. The blue curve in the figure represents the true spectral signal s2, the red curve represents the smoothed spectral signal s3, and the black dashed line represents the primary peak. A thickness guess of 100 microns is then set, and the wavelength of 550.0 nanometers is substituted into the thickness equation for reverse solution. This yields a guessed wavelength of 552.8 nm for the secondary peak, located at the gray dashed line in the figure. Finally, the wavelengths of 550.0 and 552.8 are substituted into the bimodal pseudo Voigt fitting equation to fit the initial frame's spectral signal. The peak parameters obtained after a successful fit are the actual peak values ​​of the primary and secondary peaks, 549.9 nm and 552.7 nm, respectively.

[0118] like Figure 4 As shown in the figure, if two peaks are extracted, the wavelength of the primary peak is 550.0 nm, and the wavelength of the secondary peak is 554.7 nm. The blue curve in the figure represents the true spectral signal s2, the red curve represents the smoothed spectral signal s3, and the black dashed line represents the peak center. The wavelengths 550.0 and 554.7 are then substituted into the bimodal pseudo Voigt fitting equation to fit the initial frame spectral signal. After a successful fit, the peak parameters obtained are the actual peak values ​​of the primary and secondary peaks: 549.9 nm and 554.9 nm, respectively.

[0119] Furthermore, when calculating the next frame of spectral signals, the wavelengths of the primary and secondary peaks of the initial frame, 549.9nm and 552.7nm (or 549.9nm and 554.9nm), can be directly substituted into the bimodal pseudo Voigt fitting equation to perform bimodal fitting on the next frame of spectral signals to obtain the peak values ​​of the two peaks, which can then be used for wafer thickness calculation.

[0120] like Figure 2 As shown, after fitting the Nth frame spectral signal in step S2, the following steps are further included:

[0121] S3, determining whether the bimodal pseudo Voigt fitting equation has converged; if the bimodal pseudo Voigt fitting equation has not converged, the fitting is determined to have failed, and the peak value of the Nth frame spectral signal is extracted according to the steps for processing the first frame spectral signal, i.e., returning to step S2a;

[0122] If the bimodal pseudo Voigt fitting equation converges, step S4 is executed.

[0123] S4, performing difference calculation on all the parameters obtained by fitting and all the parameters obtained by fitting of the N-1th frame.

[0124] S5, determining whether the difference of each parameter is greater than a difference threshold.

[0125] If the difference in any parameter exceeds the difference threshold, the fitting is considered a failure, and peak extraction is performed on the Nth frame's spectral signal according to the same procedure as for the first frame's spectral signal, returning to step S2a. Otherwise, the fitting is successful. If the bimodal pseudo-Voigt fitting equation does not converge when fitting the Nth frame's spectral signal, the fitting fails, and the peak values ​​of the primary and secondary peaks for the Nth frame are not obtained.

[0126] Specifically, when the bimodal pseudo Voigt fitting equation is fitted to the Nth frame spectral signal, the equation converges, but the eight parameters obtained by fitting (such as 、 、 、 、 、 、 and If any of the parameters in the N-1 frame deviates significantly from the eight parameters obtained in the N-1 frame, the fitting is considered a failure, and the main peak value and secondary peak value of the N frame will not be obtained. Each parameter has its own empirical threshold deviation.

[0127] An embodiment of the present invention further provides a thickness measurement method, which is executed by an electronic device such as a computer or a server, and specifically includes:

[0128] The peak value of the main peak and the secondary peak value of the Nth frame are obtained by using the peak value extraction method of the spectral signal described above.

[0129] Calculate the wafer thickness of the Nth frame based on the main peak peak value and the secondary peak peak value.

[0130] For example, the wavelengths of the primary and secondary peak values ​​of the Nth frame, such as 549.9 nm and 552.7 nm, can be substituted into the thickness equation mentioned above to obtain the wafer thickness of the Nth frame.

[0131] This embodiment accurately obtains the main peak peak value and the secondary peak peak value through the peak extraction method of the above-mentioned spectral signal, and then calculates the thickness of the wafer at the Nth frame through the thickness equation based on the obtained main peak peak value and secondary peak peak value, thereby ensuring the accuracy of the wafer thickness calculation.

[0132] In one embodiment, if the spectral signal of the Nth frame fails to be fitted, the main peak peak value and the secondary peak peak value of the Nth frame will not be obtained. The main peak peak value and the secondary peak peak value can be extracted according to the method of the initial frame. Then, the thickness calculation method can be used to substitute the obtained main peak peak value and the secondary peak peak value into the thickness equation to obtain the wafer thickness of the Nth frame.

[0133] Another method is to calculate the thickness directly without extracting the peak value of the Nth frame spectrum signal, specifically:

[0134] The wafer thickness of a preset number of frames before the Nth frame is obtained and linearly extrapolated to obtain the wafer thickness of the Nth frame.

[0135] This embodiment is another thickness measurement method when the bimodal pseudo Voigt fitting equation fails to fit. The wafer thickness of the Nth frame is directly obtained by linearly extrapolating the wafer thickness of the M frames before the Nth frame (empirical value).

[0136] In this embodiment, when measuring the thickness of the wafer in the Nth frame, since the double-peak pseudo Voigt fitting equation fails to fit the spectral signal of the Nth frame during peak extraction in the previous process, and the main peak peak value and the secondary peak peak value of the Nth frame cannot be obtained for thickness calculation, the main and secondary peak values ​​or wafer thicknesses obtained by fitting of the previous frames are directly obtained and linearly extrapolated to obtain the wafer thickness of the Nth frame. Taking into account the characteristic of small differences between consecutive measurement data, the consistency and relative accuracy of the thickness data can be guaranteed when fitting is abnormal, avoiding missing thickness data or excessive errors due to fitting failure.

[0137] An embodiment of the present invention further provides a thinning control method, which is executed by an electronic device such as a computer or a server, and specifically includes:

[0138] Obtain the wafer thickness of the Nth frame according to the thickness calculation method described above;

[0139] When the wafer thickness of the Nth frame reaches a preset thickness, wafer thinning stops. The preset thickness is the thickness value set for thinning requirements.

[0140] This embodiment utilizes a precise thickness calculation method to obtain the wafer thickness at the Nth frame, providing real-time, accurate thickness data for the thinning process. Automatically stopping the thinning operation when the wafer thickness reaches a preset value effectively avoids over-thinning or under-thinning, ensuring that the final wafer thickness precisely meets the expected standard, thereby improving the quality and consistency of the wafer thinning process.

[0141] An embodiment of the present invention further provides a thinning machine for performing a thinning process on a wafer and executing any one of the peak extraction methods and thickness measurement methods described above during the thinning process.

[0142] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0143] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0144] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0145] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0146] Obviously, the above embodiments are merely examples for clarity of explanation and are not intended to limit the implementation methods. Those skilled in the art will readily appreciate that other variations or modifications based on the above descriptions are possible. It is not necessary and impossible to enumerate all implementation methods here. Obvious variations or modifications arising therefrom remain within the scope of protection of the present invention.

Claims

1. A peak extraction method, characterized in that: include: Obtain the main peak-to-peak value and the secondary peak-to-peak value of the spectrum signal of the wafer in the Nth frame and the spectrum signal of the N-1th frame, where N is an integer greater than or equal to 2; Obtaining preset parameters, and using the main peak value, the secondary peak value, and the preset parameters of the N-1th frame spectral signal as initial guesses of a bimodal pseudo Voigt fitting equation, fitting the Nth frame spectral signal to obtain the main peak value and the secondary peak value of the Nth frame spectral signal; Wherein, a first frame of spectrum signal of the wafer is obtained; Perform peak extraction on the first frame spectral signal according to a preset intensity threshold; When a peak is extracted, the peak value is determined to be the main peak peak; Get the thickness guess; The secondary peak value is obtained by reversely solving the thickness equation based on the thickness guess value and the primary peak value; Using the main peak value and the secondary peak value as initial guess values ​​of a bimodal pseudo Voigt fitting equation, fitting the first frame of spectral signal to obtain the main peak value and the secondary peak value of the first frame of spectral signal; When two peaks are extracted, the peak values ​​of the two peaks are determined to be the main peak peak and the secondary peak peak respectively; The main peak peak value and the secondary peak peak value are used as the initial guess values ​​of the bimodal pseudo Voigt fitting equation, and the first frame spectral signal is fitted to obtain the main peak peak value and the secondary peak peak value of the first frame spectral signal.

2. The method according to claim 1, characterized in that The preset parameters include the maximum intensity of the spectral signal, the proportion of Gaussian signals in the spectral signal, and the width of the peak in the spectral signal.

3. The method according to claim 2, wherein The bimodal pseudo Voigt fitting equation is: , in, Represents the intensity at any wavelength in the Nth frame spectral signal, and Indicates the maximum intensity of the spectral signal, and Indicates the proportion of Gaussian signals in the spectral signal, and represents the main peak and secondary peak in the spectral signal, and represents the width of the peak in the spectral signal, is the wavelength independent variable in the Nth frame spectral signal.

4. The method according to claim 1, wherein The thickness equation is: , in, Indicates the wafer thickness, The main peak value, is the sub-peak value, 、 、 、 、 、 、 、 These are the parameter results obtained by fitting the measurement data of the translation stage and spectrometer.

5. The method according to claim 3, characterized in that Also includes: Obtain the main peak value, main peak width, secondary peak value and secondary peak width of the spectral signal of the Nth frame and the N-1th frame; Differences between the main peak value and the secondary peak value of the two frames are taken to obtain the main peak peak difference and the secondary peak peak difference; respectively determining whether the main peak-to-peak difference and the secondary peak-to-peak difference are greater than or equal to a preset percentage of the width of the peak corresponding to any frame; If the peak difference of any peak is greater than or equal to a preset percentage of the width of the peak corresponding to any frame, all parameters obtained from the bimodal pseudo-Voigt fitting equation of a preset number of frames before the Nth frame are linearly extrapolated to obtain an initial guess of the bimodal pseudo-Voigt fitting equation of the Nth frame; The Nth frame spectral signal is fitted using the initial guess value of the bimodal pseudo Voigt fitting equation to obtain the main peak peak value and the secondary peak peak value of the Nth frame spectral signal.

6. The method according to claim 1, wherein Before fitting the Nth frame spectral signal using the bimodal pseudo Voigt fitting equation, the following steps are also included: Acquiring background spectral signals originating from measurement conditions; Subtracting the background spectrum signal from the Nth frame spectrum signal of the wafer to obtain a true spectrum signal of the wafer in the Nth frame; Performing main peak judgment on the real spectrum signal; If there is a main peak, the real spectrum signal is smoothed according to a preset smoothing intensity; If there is no main peak, the peak extraction of the spectrum signal of the Nth frame is stopped.

7. The method according to claim 1, characterized in that After fitting the Nth frame of spectral signal, the method further includes: Determining whether the bimodal pseudo Voigt fitting equation converges; If the bimodal pseudo Voigt fitting equation does not converge, the fitting is determined to be failed, and the peak value of the Nth frame spectral signal is extracted according to the steps of processing the first frame spectral signal; If the bimodal pseudo Voigt fitting equation converges, all the parameters obtained by fitting are respectively subjected to difference calculation with all the parameters obtained by fitting of the N-1th frame; Determine whether the difference of each parameter is greater than the difference threshold; If the difference of any parameter is greater than the difference threshold, the fitting is determined to have failed, and the peak value of the Nth frame spectral signal is extracted according to the steps of processing the first frame spectral signal.

8. A thickness measurement method, characterized in that: include: Obtaining the main peak-to-peak value and the secondary peak-to-peak value of the Nth frame using the method described in any one of claims 1 to 7; The wafer thickness of the Nth frame is calculated according to the main peak value and the secondary peak value.

9. The method according to claim 8, wherein If the bimodal pseudo Voigt fitting equation fails to fit, the wafer thickness of a preset number of frames before the Nth frame is obtained and linearly extrapolated to obtain the wafer thickness of the Nth frame.

10. A peak extraction device, characterized in that: include: A processor and a memory connected to the processor; wherein the memory stores instructions that can be executed by the processor, and the instructions are executed by the processor to enable the processor to perform the peak extraction method according to any one of claims 1 to 7.

11. A thickness measuring device, characterized in that: include: A processor and a memory connected to the processor; wherein the memory stores instructions that can be executed by the processor, and the instructions are executed by the processor to enable the processor to perform the thickness measurement method according to any one of claims 8 to 9.

12. A thinning machine, characterized in that: Used to perform thinning processing on a wafer, and perform the method according to any one of claims 1 to 9 during the thinning process.

Images