Peak value extraction method, thickness measurement method and equipment

Abstract

The invention provides a peak value extraction method, a thickness measurement method and equipment. The method is applied to the high-precision measurement field. The method comprises the following steps: acquiring a main peak-to-peak value and a secondary peak-to-peak value of a spectral signal of an Nth frame of wafer and a spectral signal of an (N-1) th frame of wafer, N being an integer greater than or equal to 2; and acquiring a preset parameter, and fitting the Nth frame of spectral signal by taking the main peak value and the secondary peak value of the (N-1) th frame of spectral signal and the preset parameter as initial guess values of a double-peak pseudo Voigt fitting equation to obtain a main peak value and a secondary peak value of the Nth frame of spectral signal. The accuracy of obtaining the main peak-to-peak value and the secondary peak-to-peak value of the Nth frame of spectral signal is improved.

Description

Technical Field

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

[0002] Thinning machines are crucial in the manufacturing fields such as semiconductors. With the evolution of chip manufacturing technology, the requirement for precise control of the wafer thickness has become increasingly critical. Thinning machines can adjust the thickness of wafers by physically or chemically removing the surface materials of wafers to meet the requirements of different chip products. Especially in continuous thinning operations, it is necessary to continuously and accurately obtain the wafer thickness to precisely control the thinning process.

[0003] Currently, spectral analysis technology has become a key means to achieve precise measurement of wafer thickness due to its high precision and high reliability. Traditional spectral analysis usually extracts the peaks of the wafer surface spectrum to obtain the required information. Spectral analysis can accurately separate and identify the peak information in the spectrum, and then precisely determine the thickness of each layer of the wafer. This not only helps to ensure the quality of the wafer, but also provides a strong basis for the optimization of the manufacturing process, promoting the development of wafer manufacturing to a higher level.

[0004] However, the peak extraction methods of the existing technologies have obvious defects. In spectral data, the situation of adjacent peaks often occurs. These adjacent peaks may originate from the reflection information of different layers of the wafer or be interference signals generated during the measurement process. Due to the small distance between them, they are intertwined and overlapped, making it extremely difficult to accurately extract the positions of the peaks. And the inability to accurately extract the positions of the two peaks will directly affect the calculation of the wafer thickness based on the spectral data, and the inaccurate thickness calculation will hinder the precise control of the wafer thinning process, seriously affecting the thinning effect. Summary of the Invention

[0005] In view of this, on the one hand, the present invention provides a peak extraction method, including: Obtaining the main peak value and the secondary peak value of the spectral signal of the Nth frame of the wafer and the spectral signal of the (N - 1)th 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 of the spectral signal of the (N - 1)th frame and the preset parameters as the initial guess values of the double-peak pseudo Voigt fitting equation to fit the spectral signal of the Nth frame, and obtaining the main peak value and the secondary peak value of the spectral signal of the Nth frame.

[0006] Optionally, the preset parameters include the maximum value of the intensity of the spectral signal, the proportion of the Gaussian signal in the spectral signal, and the width of the peak in the spectral signal.

[0007] Optionally, the double-peak pseudo Voigt fitting equation is: ; Among them, represents the intensity at any wavelength in the Nth frame of spectral signal, and represents the maximum intensity of the spectral signal, and represent the proportion of Gaussian signal in the spectral signal, and represent the main peak peak value and the secondary peak peak value in the spectral signal, and represent the width of the peak in the spectral signal, is the wavelength independent variable in the Nth frame of spectral signal.

[0008] Optionally, the peak extraction method of the present invention further includes: Obtain the first frame of spectral signal of the wafer; Perform peak extraction on the first frame of spectral signal according to a preset intensity threshold; When one peak is extracted, determine the peak value of the peak as the main peak peak value; Obtain a thickness guess value; Use the thickness equation to inversely solve the secondary peak peak value according to the thickness guess value and the main peak peak value; Use the main peak peak value and the secondary peak peak value as the initial guess values of the double-peak pseudo Voigt fitting equation to fit the first frame of spectral signal, and obtain the main peak peak value and the secondary peak peak value of the first frame of spectral signal; When two peaks are extracted, determine the peak values of the two peaks as the main peak peak value and the secondary peak peak value respectively; Use the main peak peak value and the secondary peak peak value as the initial guess values of the double-peak pseudo Voigt fitting equation to fit the first frame of spectral signal, and obtain the main peak peak value and the secondary peak peak value of the first frame of spectral signal.

[0009] Optionally, the thickness equation is: ; Among them, represents the wafer thickness, is the main peak peak value, is the secondary peak peak value, , , , , , , , are the parameter results fitted according to the measurement data of the displacement stage and the spectrometer.

[0010] Optionally, the peak extraction method of the present invention further includes: Obtain the peak value of the main peak, the width of the main peak, the peak value of the secondary peak, and the width of the secondary peak of the spectral signals of the Nth frame and the (N - 1)th frame; Subtract the peak value of the main peak and the peak value of the secondary peak of the two frames respectively to obtain the difference in the peak value of the main peak and the difference in the peak value of the secondary peak; Judge whether the difference in the peak value of the main peak and the difference in the peak value of the secondary peak are greater than or equal to the preset percentage of the width of the peak corresponding to any frame respectively; If the peak value difference of any peak is greater than or equal to the preset percentage of the width of the peak corresponding to any frame, linearly extrapolate all the parameters obtained from the double-peak pseudo Voigt fitting equation for a preset number of frames before the Nth frame to obtain the initial guess of the double-peak pseudo Voigt fitting equation for the Nth frame; Use the initial guess value of the double-peak pseudo Voigt fitting equation to fit the spectral signal of the Nth frame to obtain the peak value of the main peak and the peak value of the secondary peak of the spectral signal of the Nth frame.

[0011] Optionally, before using the double-peak pseudo Voigt fitting equation to fit the spectral signal of the Nth frame, it further includes: Obtain the background spectral signal resulting from the measurement conditions; Subtract the background spectral signal from the spectral signal of the Nth frame of the wafer to obtain the true spectral signal of the wafer at the Nth frame; Judge the main peak of the true spectral signal; If there is a main peak, perform smoothing processing on the true spectral signal according to the preset smoothing intensity; If there is no main peak, stop extracting the peak value of the spectral signal of the Nth frame.

[0012] Optionally, after fitting the spectral signal of the Nth frame, it further includes: Judge whether the double-peak pseudo Voigt fitting equation converges; If the double-peak pseudo Voigt fitting equation does not converge, determine that the fitting fails, and extract the peak value of the spectral signal of the Nth frame according to the steps of processing the spectral signal of the first frame; If the double-peak pseudo Voigt fitting equation converges, calculate the difference between all the parameters obtained by fitting and all the parameters obtained by fitting the (N - 1)th frame respectively; Judge whether the difference of each parameter is greater than the difference threshold respectively; If the difference of any parameter is greater than the difference threshold, determine that the fitting fails, and extract the peak value of the spectral signal of the Nth frame according to the steps of processing the spectral signal of the first frame.

[0013] The second aspect of the present invention provides a thickness measurement method, and this method includes: Obtain the main peak value and the secondary peak value of the Nth frame by using the method described above; Calculate the wafer thickness of the Nth frame according to the main peak value and the secondary peak value.

[0014] Optionally, if the fitting of the double-peak pseudo Voigt fitting equation fails, obtain the wafer thickness of a preset number of frames before the Nth frame for linear extrapolation to obtain the wafer thickness of the Nth frame.

[0015] A 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 executable by the processor, and the instructions are executed by the processor to enable the processor to execute the above-mentioned peak extraction method.

[0016] A fourth aspect of the present invention provides a thickness measurement device, which includes: 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 execute the above-mentioned thickness measurement method.

[0017] A fifth aspect of the present invention provides a thinning machine for thinning a wafer and executing the method described in any one of the above during the thinning process.

[0018] Aiming at the problem that the two peak values cannot be accurately separated when extracting the peak values of adjacent peaks in the Nth frame spectral signal, by directly obtaining the main peak value and the secondary peak value of the (N - 1)th frame spectral signal, and then using the main peak value and the secondary peak value of the (N - 1)th frame as the initial guess for the Nth frame for fitting, the spectral signal of the Nth frame can be quickly fitted. Since the change in the wafer state reflected by two adjacent frames of spectral signals is usually relatively small and continuous, the peak values of the (N - 1)th frame provide a starting point close to the true value for the fitting of the Nth frame. Compared with randomly setting the initial guess value without prior information, the peak values of the (N - 1)th frame selected based on the correlation of continuous measurement data effectively avoid the dilemma of being interfered by adjacent peaks due to inaccurate initial guesses in traditional fitting methods, greatly improving the accuracy and reliability of fitting, improving the accuracy of obtaining the main peak value and the secondary peak value of the Nth frame spectral signal, providing a more reliable basis for subsequent wafer thickness calculation and thinning control, and further improving the accuracy and stability of the entire wafer thinning process, ensuring the quality and efficiency of wafer production. Description of the Drawings

[0019] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following will briefly introduce the drawings required for the description of the specific embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0020] Figure 1 It is a flowchart of a peak extraction method in an embodiment of the present invention; Figure 2 It is a flowchart of another peak extraction method in an embodiment of the present invention; Figure 3 It is a spectral signal diagram when one peak is extracted in an embodiment of the present invention; Figure 4 It is a spectral signal diagram when two peaks are extracted in an embodiment of the present invention. Specific Embodiments

[0021] The following will clearly and completely describe the technical solutions of the present invention with reference to the drawings. Obviously, the described embodiments are some embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0022] In the description of the present invention, it should be noted that the orientation or positional relationship indicated by the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc. is based on the orientation or positional relationship shown in the drawings, and is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the present invention. In addition, the terms "first", "second", "third" are only used for descriptive purposes and cannot be construed as indicating or implying relative importance.

[0023] In the description of the present invention, it should be noted that unless otherwise clearly specified and defined, the terms "installed", "connected", "connected" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be directly connected or indirectly connected through an intermediate medium, and it can also be the internal connection of two components. It can be a wireless connection or a wired connection. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0024] In addition, the technical features involved in different embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0025] As Figure 1 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: S1. Obtain the main peak peak and the secondary peak peak of the spectral signal of the Nth frame of the wafer and the spectral signal of the (N - 1)th frame, where N is an integer greater than or equal to 2.

[0026] During the thinning process, obtain the spectral signal of the Nth frame of the wafer collected by the spectrometer, and obtain the main peak peak and the secondary peak peak of the spectral signal obtained by fitting the previous frame.

[0027] S2. Obtain preset parameters, and use the main peak peak, the secondary peak peak of the (N - 1)th frame spectral signal and the preset parameters as the initial guess values of the double-peak pseudo Voigt fitting equation to fit the Nth frame spectral signal, and obtain the main peak peak and the secondary peak peak of the Nth frame spectral signal.

[0028] In this embodiment, for the problem that the two peak peaks cannot be accurately separated when extracting the similar peak peaks in the Nth frame spectral signal, by directly obtaining the main peak peak and the secondary peak peak of the (N - 1)th frame spectral signal, and then using the main peak peak and the secondary peak peak of the (N - 1)th frame as the initial guess for the Nth frame to perform fitting, the spectral signal of the Nth frame can be quickly fitted. Since the change in the wafer state reflected by two adjacent frames of spectral signals is usually relatively small and continuous, the peak of the (N - 1)th frame provides a starting point close to the true value for the fitting of the Nth frame. Compared with randomly setting the initial guess value without prior information, the peak of the (N - 1)th frame selected based on the correlation of continuous measurement data effectively avoids the dilemma of being interfered by similar peaks due to inaccurate initial guess in the traditional fitting method, greatly improves the accuracy and reliability of the fitting, improves the accuracy of obtaining the main peak peak and the secondary peak peak of the Nth frame spectral signal, provides a more reliable basis for subsequent wafer thickness calculation and thinning control, and further improves the accuracy and stability of the entire wafer thinning process, ensuring the quality and efficiency of wafer production.

[0029] Specifically, the preset parameters include the intensity maximum value of the spectral signal, the proportion of the Gaussian signal in the spectral signal, and the width of the peak in the spectral signal.

[0030] Among them, the double-peak pseudo Voigt fitting equation for fitting the Nth frame spectral signal is: ; Among them, among them, represents the intensity at any wavelength in the Nth frame spectral signal, and represents the intensity maximum of the spectral signal, and represents the proportion of Gaussian signal in the spectral signal, and represents the peak value of the main peak and the peak value of the secondary peak in the spectral signal, and represents the width of the peak in the spectral signal, is the wavelength independent variable in the spectral signal of the Nth frame.

[0031] Among them, the peak values of the main peak and the secondary peak of the spectral signal of the (N - 1)th frame are respectively substituted into the and in the fitting equation of the spectral signal. For other initial guesses in the formula, such as the intensity maximum of the spectral signal, the proportion of Gaussian signal in the spectral signal, and the width of the peak in the spectral signal, etc., preset parameters can be directly used with empirical values.

[0032] When fitting the spectral signal of the Nth frame, it is necessary to continuously fit the double - peak pseudo Voigt fitting equation according to the wavelength as the independent variable and the corresponding intensity in the Nth frame signal until the data of the two peaks of the spectral signal of the Nth frame are successfully fitted. The fitting equation contains eight parameters. Among them, and The accuracy of can determine the fitting probability of the double - peak pseudo Voigt fitting equation. Due to continuous thinning, the peak values of the main and secondary peaks directly obtained from the (N - 1)th frame can be used to accurately and successfully fit the data of the two peaks of the spectral signal of the Nth frame. Even if the initial guesses of the remaining parameters deviate greatly from the actual values, they can be successfully fitted with a high probability and can be directly used with empirical values without specifically analyzing the data for estimation.

[0033] As Figure 2 shown, the embodiment of the present invention also provides a peak extraction method, which specifically includes: S1a, obtaining the first - frame spectral signal of the wafer.

[0034] This embodiment is a process of peak extraction for the spectral signal of the initial frame before the Nth frame.

[0035] S2a, performing peak extraction on the first - frame spectral signal according to a preset intensity threshold and determining whether it is a single peak; when a single peak is extracted, step S3a is executed; when two peaks are extracted, step S7a is executed.

[0036] The preset intensity threshold is an empirical value, and the corresponding empirical value can be selected according to the experimental type. For example, if they are all spectral confocal, the same set of threshold empirical values can be used.

[0037] S3a, determining the peak value of the peak as the main peak value.

[0038] S4a, obtain a thickness guess value.

[0039] The thickness guess value can be obtained in various ways, such as: empirical values measured by multiple measurements or experiments, approximate thickness parameters measured by other contact measurement methods, and an approximate initial thickness value attached to the wafer at the initial frame.

[0040] S5a, use the thickness equation to inversely solve for the secondary peak peak value based on the thickness guess value and the main peak peak value.

[0041] When only one peak is extracted, it indicates that the distance between the main and secondary peaks is small at this time, and the secondary peak is covered by the main peak. A relatively accurate secondary peak peak value can be inversely solved through the thickness guess value.

[0042] S6a, use the main peak peak value and the secondary peak peak value as the initial guess values of the double-peak pseudo Voigt fitting equation to fit the first-frame spectral signal, and obtain the main peak peak value and the secondary peak peak value of the first-frame spectral signal.

[0043] Specifically, when performing peak extraction on the spectral signal of the first frame, it proves that there is no data from the previous frame as a reference. Then, only inaccurate main peak peak values can be obtained based on peak extraction and an inaccurate secondary peak peak value can be calculated inversely from a thickness guess value. The double-peak pseudo Voigt fitting equation is used again to fit the spectral signal to obtain the true main peak peak value and secondary peak peak value of the spectral signal.

[0044] In this embodiment, when performing peak extraction on the spectral signal of the first frame, peak extraction is performed using a preset intensity threshold. When only one peak is extracted, it indicates that the distance between the main and secondary peaks is small and the secondary peak is covered by the main peak. Therefore, the position of the secondary peak needs to be found. Specifically, the peak value of the peak can be set as the main peak peak value and a thickness guess value can be set, and the secondary peak peak value can be inversely solved using the thickness equation. Finally, the main peak peak value and the secondary peak peak value are used as the initial guess of the double-peak pseudo Voigt fitting equation to accurately fit the first-frame spectral signal, successfully solving the problems of missing initial guess and difficulty in accurately identifying the position of the secondary peak due to the proximity of the two peaks, and being able to accurately fit the main peak peak value and the secondary peak peak value of the first-frame spectral signal.

[0045] Further, S7a, determine that the peak values of the two peaks are the main peak peak value and the secondary peak peak value respectively.

[0046] When two peaks are extracted, the first peak is used as the main peak and the second peak is used as the secondary peak.

[0047] S8a, use the main peak peak value and the secondary peak peak value as the initial guess values of the double-peak pseudo Voigt fitting equation to fit the first-frame spectral signal, and obtain the main peak peak value and the secondary peak peak value of the first-frame spectral signal.

[0048] When extracting the peak of the spectral signal of the first frame in this embodiment, when two peaks are extracted, the main peak and the peak value of the secondary peak can be accurately determined. The two peak values can be directly used as the initial guess of the double-peak pseudo Voigt fitting equation to fit the spectral signal of the first frame, effectively solving the problem of the lack of initial guess of the spectral signal of the first frame, and being able to more accurately fit the main peak value and the secondary peak value in the spectral signal of the first frame.

[0049] Further, the above thickness equation is: ; Wherein, represents the thickness, is the main peak value, is the secondary peak value, , , , , , , , are the parameter results fitted according to the measurement data of the displacement stage and the spectrometer.

[0050] Substitute the main peak value into the main peak value in the formula, and substitute the thickness guess value into the thickness in the formula, then the secondary peak value can be solved inversely.

[0051] In one embodiment, after obtaining the main peak value and the secondary peak value of the Nth frame spectral signal in step S2, it further includes: Obtain the main peak value, the width of the main peak, the secondary peak value, and the width of the secondary peak of the spectral signals of the Nth frame and the N-1th frame.

[0052] Subtract the main peak value and the secondary peak value of the two frames respectively to obtain the main peak value difference and the secondary peak value difference.

[0053] Judge whether the main peak value difference and the secondary peak value difference are greater than or equal to a preset percentage of the width of the corresponding peak of any frame. Among them, the preset percentage is an empirical value, such as 50%.

[0054] If the peak value difference of any peak is greater than or equal to the preset percentage of the width of the corresponding peak of any frame, linearly extrapolate all the parameters obtained from the double-peak pseudo Voigt fitting equation of the preset number of frames before the Nth frame to obtain the initial guess of the double-peak pseudo Voigt fitting equation of the Nth frame. Among them, the preset number of frames is preferably from the 5th to the 10th frame. All parameters refer to the eight parameters of the double-peak pseudo Voigt fitting equation, such as , , , , , , and .

[0055] The spectral signal of the Nth frame is fitted using the initial guess values of the bimodal pseudo Voigt fitting equation to obtain the peak value of the main peak and the peak value of the secondary peak of the spectral signal of the Nth frame.

[0056] Specifically, when the sampling frequency is too low, the time interval between two adjacent frames of spectral signals is relatively large. During the thinning process, the state of the wafer may change significantly, and parameters such as the peak values of the main and secondary peaks in the spectral signal will change with the change of the wafer state. Due to the long sampling interval, the central positions of the main and secondary peaks of two adjacent frames of spectral signals may vary greatly, resulting in inaccurate peak values of the main peak and the secondary peak obtained by each fitting. Therefore, it is necessary to calculate the difference between the peak values of the main peak and the secondary peak obtained by fitting these two frames and compare it with the width of the corresponding peak. As long as the peak value 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 linearly extrapolate the peak values of the main and secondary peaks obtained by the previous fitting to obtain the peak values of the main peak and the secondary peak of the Nth frame. This can make reasonable inferences using existing reliable data in the case where the fitting result is inaccurate due to insufficient sampling frequency, effectively improving the accuracy and reliability of determining the peak values of the main and secondary peaks of the spectral signal at low sampling frequencies.

[0057] In one embodiment, before using the bimodal pseudo Voigt fitting equation to fit the spectral signal of the Nth frame in step S2, it further includes: Obtain the background spectral signal derived from the measurement conditions.

[0058] Specifically, the signal of the thinning machine without the wafer placed can be collected using a spectrometer; or multiple groups of spectral signals of multiple measured samples can be collected, and each group of spectral signals is collected based on the same measurement conditions.

[0059] The measured sample can be any substance. The signal of the measured sample placed on the measurement table is collected using a spectrometer. Specifically, different samples are placed under the same measurement conditions, multiple groups of spectral signals are collected, and the error that appears in each group and does not change with the measurement conditions and the sample is extracted from the multiple groups of spectral signals, which is the background spectral signal. The measurement conditions cover all customizable input parameters such as specific light source intensity, wavelength range, measurement angle, and environmental temperature and humidity, resolution, etc., to ensure the diversity and representativeness of the data.

[0060] Subtract the background spectral signal from the spectral signal of the Nth frame of the wafer to obtain the true spectral signal of the wafer in the Nth frame.

[0061] Subtract the background noise signal in the Nth frame of spectral signal to reduce the interference of background error on the data and improve the accuracy of the signal.

[0062] Judge the main peak of the real spectral signal; If there is a main peak, smooth the real spectral signal according to the preset smoothing intensity.

[0063] If there is no main peak, stop extracting the peak of the Nth frame of spectral signal.

[0064] Among them, the smoothing processing method can adopt methods such as low-pass filtering, wavelet transform, moving average, Savitzky-Golay filtering, etc. The corresponding preset smoothing intensity is an empirical value. There is a set of corresponding intensity empirical values for different experimental types, and the empirical value can be specifically selected according to the actual situation.

[0065] This embodiment can also be applicable when only the spectral signal of the first frame of the wafer is obtained. Before extracting the peak of this frame of spectral signal according to the preset intensity threshold, background spectral removal and smoothing processing are performed to improve the accuracy of the data.

[0066] In this embodiment, by obtaining the background spectral signal and subtracting it from the Nth frame of spectral signal, the interference of background error on the data is effectively reduced, and the reliability and credibility of the entire spectral analysis result are improved. Then, by judging the main peak, the spectral signal can be processed targeted, improving the analysis efficiency and accuracy. When there is a main peak, smoothing processing is performed according to the preset smoothing intensity, which can reduce noise interference, highlight the peak characteristics, make the spectral signal more accurate, and provide reliable data for accurately extracting the peak. If there is no main peak, the extraction is stopped to avoid wasting computing resources and time.

[0067] In addition, when obtaining the first frame of spectral signal of the wafer in step S1a, where the secondary peak peak is obtained by inversely solving according to the thickness guess value and the main peak peak using the thickness equation, it can also be: Obtain the wavelength interval.

[0068] Calculate the predicted secondary peak peak according to the wavelength interval and the main peak peak.

[0069] Use the thickness equation to calculate the thickness according to the main peak peak and the predicted secondary peak peak, continuously update the wavelength interval according to the difference between the thickness guess value and the thickness, and then update the predicted secondary peak peak until the difference between the thickness guess value and the thickness is less than the preset convergence threshold, and obtain the secondary peak peak that meets the accuracy requirements.

[0070] Specifically, assume a relatively small wavelength interval Δω. Then, the predicted peak value of the secondary peak w2’ = w1 + Δω, where w1 is the peak value of the main peak. Substituting w1 and w2’ into the thickness equation yields an inaccurate thickness calculation result t1’. Continuously update the value of w2’ based on the difference between the thickness guess value t1 and t1’, until the deviation between t1 and t1’ is less than a preset convergence threshold (such as 1e-4). At this time, the obtained value of w2’ can be considered as a relatively accurate peak value of the secondary peak. w1 and w2’ are the peak parameters of the double-peak pseudo Voigt fitting equation. The specific wavelength interval update method can be any general method optimized by a sudden function.

[0071] The following is an illustrative example of the above embodiment: Obtain the spectral signal s1 of the Nth frame of the wafer and the background spectral signal s0; subtract the background spectral signal s0 to obtain the true spectral signal s2 after background removal; then perform main peak judgment on the true spectral signal s2. When there is a main peak, perform Savitzky-Golay filtering on the true spectral signal s2 again to smooth it, obtaining the smoothed spectral signal s3. Then use the double-peak pseudo Voigt fitting equation to fit the smoothed spectral signal s3.

[0072] As Figure 3 shown, when the spectral signal of the first frame of the wafer is obtained in step S1a and no second peak is recognized, only one peak is extracted, and the wavelength of the main peak peak value is 550.0 nm. In the figure, the blue curve is the true spectral signal s2, the red curve is the smoothed spectral signal s3, and the position of the black dotted line is the main peak peak value. Then set a thickness guess value of 100 microns and substitute the wavelength of 550.0 nm into the thickness equation for reverse solution, and the guessed wavelength of the secondary peak peak value of 552.8 nm can be obtained, that is, the position of the gray dotted line in the figure; finally, substitute the wavelengths of 550.0 and 552.8 into the double-peak pseudo Voigt fitting equation to fit the spectral signal of the initial frame. After successful fitting, the obtained peak parameters are the actual peak values of the main and secondary peaks, which are 549.9 nm and 552.7 nm respectively.

[0073] As Figure 4 shown, if two peaks are extracted, the wavelength of the main peak peak value is 550.0 nm, and the wavelength of the secondary peak peak value is 554.7 nm. In the figure, the blue curve is the true spectral signal s2, the red curve is the smoothed spectral signal s3, and the position of the black dotted line is the center position of the peak. Then substitute the wavelengths of 550.0 and 554.7 into the double-peak pseudo Voigt fitting equation to fit the spectral signal of the initial frame. After successful fitting, the obtained peak parameters are the actual peak values of the main and secondary peaks, which are 549.9 nm and 554.9 nm respectively.

[0074] Further, when calculating the spectral signal of the next frame, the wavelengths 549.9 nm and 552.7 nm (or 549.9 nm and 554.9 nm) of the main and secondary peak peaks of the initial frame can be directly substituted into the double-peak pseudo Voigt fitting equation to perform double-peak fitting on the spectral signal of the next frame, obtaining the peak values of the two peaks, which can be used for subsequent wafer thickness calculation.

[0075] As Figure 2 shown, after fitting the Nth frame spectral signal in step S2, it further includes: S3, determining whether the double-peak pseudo Voigt fitting equation converges; if the double-peak pseudo Voigt fitting equation does not converge, it is determined that the fitting fails, and the Nth frame spectral signal is subjected to peak extraction according to the steps of processing the first frame spectral signal, that is, returning to step S2a; If the double-peak pseudo Voigt fitting equation converges, step S4 is executed.

[0076] S4, performing difference calculation on all the parameters obtained by fitting respectively with all the parameters obtained by fitting the (N - 1)th frame.

[0077] S5, respectively determining whether the difference of each parameter is greater than the difference threshold.

[0078] If the difference of any parameter is greater than the difference threshold, it is determined that the fitting fails, and the Nth frame spectral signal is subjected to peak extraction according to the steps of processing the first frame spectral signal, that is, returning to step S2a. Otherwise, the fitting is successful. When the double-peak pseudo Voigt fitting equation is used to fit the Nth frame spectral signal and the equation never converges, the fitting fails and the main peak peak and secondary peak peak of the Nth frame will not be obtained.

[0079] Specifically, when the double-peak pseudo Voigt fitting equation is used to fit the Nth frame spectral signal and the equation converges, but any one of the eight parameters obtained by fitting (such as , , , , , , and ) has a large deviation from the eight parameters obtained from the (N - 1)th frame, it is determined that the fitting fails, and the main peak peak and secondary peak peak of the Nth frame will not be obtained. Each parameter has its own empirical threshold deviation.

[0080] The embodiment of the present invention also provides a thickness measurement method, which is executed by an electronic device such as a computer or a server, and specifically includes: Obtaining the main peak peak and secondary peak peak of the Nth frame by using the peak extraction method of the spectral signal described above.

[0081] Calculate the wafer thickness of the Nth frame based on the peak value of the main peak and the peak value of the secondary peak.

[0082] For example, the wavelengths of the main and secondary peak values of the Nth frame, such as 549.9 nm and 552.7 nm, can be substituted into the above-mentioned thickness equation to obtain the wafer thickness of the Nth frame.

[0083] In this embodiment, the peak values of the main peak and the secondary peak are accurately obtained through the above-mentioned peak extraction method of the spectral signal. Then, based on the obtained peak values of the main peak and the secondary peak, the thickness of the wafer at the Nth frame is calculated through the thickness equation, ensuring the accuracy of the wafer thickness calculation.

[0084] In one embodiment, if the fitting of the spectral signal of the Nth frame fails during fitting, the peak values of the main peak and the secondary peak of the Nth frame will not be obtained. The peak values of the main peak and the secondary peak can be extracted according to the method of the initial frame. Furthermore, according to the obtained peak values of the main peak and the secondary peak, substituting them into the thickness equation, the wafer thickness of the Nth frame can be obtained.

[0085] Another method is that the peak extraction of the spectral signal of the Nth frame can be not performed, and the thickness calculation can be directly carried out. Specifically: Obtain the wafer thickness of a preset number of frames before the Nth frame for linear extrapolation to obtain the wafer thickness of the Nth frame.

[0086] This embodiment is another thickness measurement method when the fitting of the double-peak pseudo Voigt fitting equation fails. The wafer thickness of M frames (empirical value) before the Nth frame is directly linearly extrapolated to obtain the wafer thickness of the Nth frame.

[0087] When measuring the thickness of the wafer of the Nth frame in this embodiment, 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 peak values of the main peak and the secondary peak of the Nth frame cannot be obtained for thickness calculation. Therefore, the main and secondary peak values or wafer thickness obtained by fitting the previous several frames are directly obtained for linear extrapolation to obtain the wafer thickness of the Nth frame. Considering the characteristic that the difference between continuous measurement data is small, it can ensure the coherence and relative accuracy of the thickness data during abnormal fitting, and avoid the loss or excessive error of the thickness data caused by fitting failure.

[0088] The embodiment of the present invention also provides a thinning control method, which is executed by an electronic device such as a computer or a server, and specifically includes: Obtain the wafer thickness of the Nth frame according to the above-mentioned thickness calculation method; When the wafer thickness of the Nth frame reaches the preset thickness, stop thinning the wafer. The preset thickness is the thickness value set according to the thinning requirement.

[0089] In this embodiment, an accurate thickness calculation method is used to obtain the thickness of the Nth frame of the wafer, which can provide real-time and accurate thickness data for the thinning process. When the wafer thickness reaches the preset value, the thinning operation is automatically stopped, which can effectively avoid the problems of over-thinning or under-thinning, ensure that the final thickness of the wafer precisely meets the expected standard, and improve the quality and consistency of wafer thinning processing.

[0090] An embodiment of the present invention also provides a thinning machine for thinning a wafer and performing the peak extraction method and thickness measurement method described in any one of the above during the thinning process.

[0091] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0092] The present invention is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to embodiments of the present invention. It should be understood that each flow and / or block in the flowchart and / or block diagram can be implemented by computer program instructions, and the combination of flows and / or blocks in the flowchart and / or block diagram can also be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate means for implementing the functions specified in one Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0093] These computer program instructions can 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 generate a manufactured article including instruction means, and the instruction means implements the functions specified in one Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0094] These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process, and thus the instructions executed on the computer or other programmable device provide means for implementing the functions specified in one Figure 1Steps of the functions specified in one or more processes and / or boxes Figure 1 Steps of the functions specified in one or more boxes

[0095] Obviously, the above embodiments are only examples for clear illustration and not limitations on the implementation manners. For those of ordinary skill in the art, other different forms of changes or variations can be made based on the above description. It is not necessary and impossible to list all implementation manners here. And the obvious changes or variations derived therefrom are still within the protection scope of the present invention.

Claims

1. A peak extraction method, characterized in that, Including: Obtain the main peak peak value and the secondary peak peak value of the spectral signal of the Nth frame of the wafer and the spectral signal of the (N - 1)th frame, where N is an integer greater than or equal to 2; Obtain preset parameters, and use the main peak peak value, the secondary peak peak value of the (N - 1)th frame spectral signal, and the preset parameters as the initial guess values of the double-peak pseudo Voigt fitting equation to fit the Nth frame spectral signal, and obtain the main peak peak value and the secondary peak peak value of the Nth frame spectral signal.

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

3. The method according to claim 2, wherein The double-peak pseudo Voigt fitting equation is: ; Among them, represents the intensity at any wavelength in the Nth frame of spectral signal, and represent the maximum intensity of the spectral signal, and represent the proportion of Gaussian signal in the spectral signal, and represent the peak value of the main peak and the peak value of the secondary peak in the spectral signal, and represent the width of the peak in the spectral signal, is the wavelength independent variable in the Nth frame of spectral signal.

4. The method according to claim 1, wherein Also including: Obtain the spectral signal of the first frame of the wafer; Perform peak extraction on the spectral signal of the first frame according to a preset intensity threshold; When one peak is extracted, determine the peak value of the peak as the main peak peak value; Obtain a thickness guess value; Use the thickness equation to inversely solve the secondary peak peak value according to the thickness guess value and the main peak peak value; Use the main peak peak value and the secondary peak peak value as the initial guess values of the double-peak pseudo Voigt fitting equation to fit the spectral signal of the first frame, and obtain the main peak peak value and the secondary peak peak value of the spectral signal of the first frame; When two peaks are extracted, determine the peak values of the two peaks as the main peak peak value and the secondary peak peak value respectively; Use the main peak peak value and the secondary peak peak value as the initial guess values of the double-peak pseudo Voigt fitting equation to fit the spectral signal of the first frame, and obtain the main peak peak value and the secondary peak peak value of the spectral signal of the first frame.

5. The method according to claim 4, wherein The thickness equation is: ; Among them, represents the wafer thickness, is the peak value of the main peak, is the peak value of the secondary peak, , , , , , , , are the parameter results fitted according to the measurement data of the displacement stage and the spectrometer.

6. The method according to claim 3, characterized in that, Also including: Obtain the main peak peak value, the width of the main peak, the secondary peak peak value, and the width of the secondary peak of the spectral signals of the Nth frame and the (N - 1)th frame; Subtract the main peak peak values and the secondary peak peak values of the two frames respectively to obtain the main peak peak value difference and the secondary peak peak value difference; Judge whether the main peak peak value difference and the secondary peak peak value difference are greater than or equal to a preset percentage of the width of the peak corresponding to any frame respectively; If the peak value difference of any peak is greater than or equal to the preset percentage of the width of the peak corresponding to any frame, linearly extrapolate all the parameters obtained by the double-peak pseudo Voigt fitting equation of the preset number of frames before the Nth frame to obtain the initial guess of the double-peak pseudo Voigt fitting equation of the Nth frame; Use the initial guess value of the double-peak pseudo Voigt fitting equation to fit the Nth frame spectral signal, and obtain the main peak peak value and the secondary peak peak value of the Nth frame spectral signal.

7. The method according to claim 1, wherein Before using the double-peak pseudo Voigt fitting equation to fit the Nth frame spectral signal, it also includes: Obtain the background spectral signal due to measurement conditions; Subtract the background spectral signal from the Nth frame spectral signal of the wafer to obtain the true spectral signal of the wafer in the Nth frame; Perform main peak judgment on the true spectral signal; If there is a main peak, perform smoothing processing on the true spectral signal according to a preset smoothing intensity; If there is no main peak, stop peak extraction of the Nth frame spectral signal.

8. The method according to claim 4, wherein After fitting the Nth frame spectral signal, it also includes: Judge whether the double-peak pseudo Voigt fitting equation converges; If the double-peak pseudo Voigt fitting equation does not converge, it is determined that the fitting fails, and the spectral signal of the Nth frame is subjected to peak extraction according to the steps of processing the spectral signal of the first frame; If the double-peak pseudo Voigt fitting equation converges, the difference calculation is performed between all the parameters obtained by fitting and all the parameters obtained by fitting the (N - 1)th frame respectively; Judge whether the difference of each parameter is greater than the difference threshold respectively; If the difference of any parameter is greater than the difference threshold, it is determined that the fitting fails, and the spectral signal of the Nth frame is subjected to peak extraction according to the steps of processing the spectral signal of the first frame.

9. A thickness measurement method, characterized in that, Including: Obtaining the main peak value and the secondary peak value of the Nth frame by using the method according to any one of claims 1-8; Calculating the wafer thickness of the Nth frame according to the main peak value and the secondary peak value; 10. The method according to claim 9, wherein If the double-peak pseudo Voigt fitting fails, the wafer thickness of a preset number of frames before the Nth frame is obtained for linear extrapolation to obtain the wafer thickness of the Nth frame.

11. A peak extraction device, characterized in that, Including: 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 execute the peak extraction method according to any one of claims 1-8.

12. A thickness measuring device, characterized in that, Including: 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 execute the thickness measurement method according to any one of claims 9-10.

13. A thinning machine, characterized in that, For thinning the wafer and executing the method according to any one of claims 1-10 during the thinning process.

Images