Method of detecting a process endpoint and semiconductor apparatus

Abstract

The application relates to a process endpoint detection method and a semiconductor device. The detection method comprises the following steps: blowing inert gas into an absorption cell, collecting an inert gas interference diagram to obtain an initial background spectrum, taking the initial background spectrum as a reference background spectrum; stopping the blowing of the inert gas, introducing a to-be-detected gas in a tail gas exhaust pipeline of the semiconductor device into the absorption cell, collecting a to-be-detected gas interference diagram and obtaining a sample spectrum; taking a set time as a period, acquiring a detection background spectrum; judging whether the detection background spectrum needs to be updated as the reference background spectrum; under preset conditions, measuring standard absorbance coefficients k of at least three characteristic gases in the to-be-detected gas and constructing a coefficient matrix K; according to the coefficient matrix K, the concentration c of the corresponding characteristic gas is acquired; setting a concentration threshold, a concentration change rate threshold and a byproduct mutation threshold, and judging whether the process endpoint is reached. The spectral detection range of the application is wide and the detection process is stable.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of semiconductor detection, and particularly relates to a process endpoint detection method and a semiconductor device. BACKGROUND

[0002] In advanced semiconductor manufacturing, accurate determination of process endpoint is a key link to control chip performance, yield and productivity. For example, in plasma etching process, accurate determination of etching endpoint to stop the reaction is the fundamental guarantee to prevent over-etching damage to the underlying material or deviation of the critical dimension; in chamber cleaning process, accurate determination of cleaning endpoint can maximize cleaning efficiency and reduce excessive damage to the chamber internals. Real-time, in-situ monitoring of the composition of the exhaust gas produced during the process and its dynamic change is the most direct and reliable technical means to achieve high-precision and predictable endpoint determination.

[0003] Currently, the optical absorption spectrum-based tail gas analysis is the mainstream technical direction, but it still has significant technical limitations in meeting the requirements of semiconductor harsh production environment. The existing main technical paths and their limitations are as follows: The non-dispersive infrared (NDIR) technology detects the characteristic absorption of the target gas through a specific wavelength narrowband filter, which has the advantages of relatively simple system structure and low cost. However, its application in advanced semiconductor process endpoint detection is fundamentally limited: first, its detection limit is usually in the ppm level, which is difficult to sensitively respond to the characteristic trace gas involved at the endpoint time (for example, the concentration mutation of certain reaction products or reactants at the ppb to ppm level); second, when facing complex multi-component process tail gas, an independent filter channel and detector need to be configured for each gas, resulting in a sharp rise in system complexity and cost, and it is difficult to avoid cross interference between components; finally, this technology lacks complete spectral information, and cannot identify unknown components or analyze complex spectra, limiting process development and abnormal diagnosis capabilities. The traditional Fourier transform infrared spectroscopy (FTIR) technology is based on a Michelson interferometer to obtain a wide range of high-resolution mid-infrared absorption spectra, which can simultaneously perform qualitative and quantitative analysis of multiple gases, and has strong multi-component detection and fingerprint identification capabilities. However, the high-precision mechanical moving parts (such as moving mirrors) in the core interferometer are extremely sensitive to environmental vibrations, and in the semiconductor factory environment where there are factory vibrations and equipment mechanical interference, it is difficult to maintain stable interference signals and measurement accuracy. In addition, this type of instrument is bulky and usually needs to be away from the process equipment for extraction measurement through a long sampling pipeline, introducing significant signal delay and sample adsorption / memory effects, which cannot meet the requirements of real-time in-situ control, and the overall cost is high. The micro-electromechanical system Fourier transform infrared (MEMS-FTIR) technology attempts to miniaturize the interferometer through micro-electromechanical technology to solve the problem of large size and poor shock resistance of traditional FTIR. Although it realizes the miniaturization of the device, there are reliability defects when directly applied to the semiconductor process tail gas environment: the nanoparticles, metal organic compounds, and plasma activated substances commonly found in process tail gas are prone to form deposition pollution on the micro-precision optical window and mirror surface, causing irreversible attenuation of light flux and baseline drift, and rapid deterioration of measurement signals. Due to its high degree of integration, cleaning or replacing optical components is extremely difficult or even impossible, resulting in long-term working stability far lower than the requirements of continuous long-term stable operation of semiconductor manufacturing equipment.

[0004] In summary, it is necessary to provide a new process endpoint detection method and semiconductor equipment to solve the above problems in the prior art. SUMMARY

[0005] The technical problem to be solved by the present application is how to provide a process endpoint detection method and semiconductor equipment with wide spectral detection range and stable detection process.

[0006] To solve the above technical problems, according to an embodiment of the present application, a process endpoint detection method is provided for use in a semiconductor device, the detection method comprising the following steps: purging an inert gas into an absorption cell, collecting an inert gas interference pattern of the inert gas to obtain an initial background spectrum, taking the initial background spectrum as a reference background spectrum; stopping purging of the inert gas, introducing a to-be-detected gas in an exhaust pipeline of the semiconductor device into the absorption cell, collecting a to-be-detected gas interference pattern and obtaining a sample spectrum; periodically acquiring a detection background spectrum at a set time; comparing the integral intensity of the detection background spectrum in a reference waveband with the corresponding integral intensity of the reference background spectrum to determine whether the detection background spectrum needs to be updated as the reference background spectrum; under a preset condition, determining standard absorbance coefficients k of at least three characteristic gases in the to-be-detected gas and constructing a coefficient matrix K; obtaining the concentrations c of the corresponding characteristic gases according to the coefficient matrix K; setting a concentration threshold, a concentration change rate threshold, and a byproduct mutation threshold, and determining whether a process endpoint is reached.

[0007] According to an embodiment of the present application, the purging of the inert gas into the absorption cell and the collection of the inert gas interference pattern to obtain the initial background spectrum comprise: maintaining the temperature of the absorption cell at 150-200°C, and purging the inert gas into the absorption cell; wherein the gas flow of the inert gas is greater than 1 L / min, and the purging time is 30-60 s; starting the interferometer to acquire a set number of the inert gas interference patterns; point-by-point accumulation of a plurality of the inert gas interference patterns is performed, and a mean value is obtained, and a fast Fourier transform is performed to obtain the initial background spectrum.

[0008] According to an embodiment of the present application, the introduction of the to-be-detected gas in the exhaust pipeline of the semiconductor device into the absorption cell, the collection of the to-be-detected gas interference pattern, and the obtaining of the sample spectrum comprise: maintaining the temperature of the absorption cell at 150-200°C, and purging the to-be-detected gas into the absorption cell; wherein the gas flow of the to-be-detected gas is greater than 1 L / min, and the purging time is 30-60 s; the interferometer continuously records the interference pattern at a set scanning frequency to acquire a plurality of original interference patterns; the plurality of original interference patterns are averaged, and a fast Fourier transform is performed to obtain the sample spectrum.

[0009] According to an embodiment of the present application, the fast Fourier transform performed on the original interference pattern to obtain the sample spectrum comprises: accumulating the original interference pattern to obtain an accumulated energy I;

[0010]

[0011] wherein, ​​is the optical path difference; is the interferometer response function, taking values 0-1; is the instantaneous interferogram intensity; the cumulative energy I of a set number is calculated by arithmetic mean value calculation to obtain an average interferogram; a fast Fourier transform is performed on the average interferogram to obtain the sample spectrum.

[0012] According to embodiments of the present application, the detection background spectrum is acquired periodically at a set time; the integral intensity of the detection background spectrum in a reference waveband is compared with the corresponding integral intensity of the reference background spectrum to determine whether the detection background spectrum needs to be updated to the reference background spectrum, including comparing the integral intensity of the detection background spectrum in the reference waveband with the corresponding integral intensity of the reference background spectrum to obtain a relative drift amount ,

[0013]

[0014] When the relative drift amount is less than or equal to 5%, the reference background spectrum is continued to be used; when the relative drift amount is greater than 5%, the inert gas is re-injected, and a new background spectrum is collected as the reference background spectrum.

[0015] According to embodiments of the present application, the standard absorbance coefficients k of at least three characteristic gases in the to-be-detected gas are determined under a preset condition, and a coefficient matrix K is constructed, including,

[0016] the absorbance of the characteristic gas required is detected in a set wavelength range ,

[0017]

[0018] wherein, is the intensity of the sample spectrum at each wave number; is the intensity of the reference background spectrum at each wave number; the absorbance is obtained according to the standard absorbance coefficient k,

[0019]

[0020] wherein, is the nominal concentration of the corresponding characteristic gas; the standard absorbance coefficients k of at least three characteristic gases are arranged to form the coefficient matrix K.

[0021] According to the embodiment of the present application, the concentration c of the corresponding characteristic gas is obtained according to the coefficient matrix K, comprising: the absorbance of the characteristic gas is calculated according to the absorbance of the characteristic gas in the characteristic gas and the concentration of the characteristic gas in the characteristic gas and the coefficient matrix K is brought into a linear model,

[0022]

[0023] wherein, is a noise residual; the least square method is used for each required characteristic gas to obtain the concentration c of the corresponding characteristic gas,

[0024]

[0025] wherein, represents a transposed matrix obtained by interchanging the rows and columns of the coefficient matrix K.

[0026] According to the embodiment of the present application, the concentration threshold, the concentration change rate threshold and the by-product mutation threshold are set, comprising: the concentration threshold is set to 5-10 ppm; the concentration change rate threshold is set to 0.1-0.6 ppm / s; and the by-product mutation threshold is set to 50-60%.

[0027] According to the embodiment of the present application, whether the etching endpoint is reached is judged, comprising: one characteristic gas in the to-be-detected gas is taken as a criterion gas, and the remaining characteristic gases are taken as by-products; when the concentration of the criterion gas is less than the concentration threshold, and the concentration change rate of the criterion gas is less than the concentration change rate threshold, or the change value of the by-products is greater than the by-product mutation threshold and the concentration change rate of the criterion gas is less than the concentration change rate threshold, it is determined that the process endpoint is reached.

[0028] A semiconductor device for implementing the above detection method, the semiconductor device comprising a process chamber, an exhaust pipeline, an absorption cell and an interferometer; the exhaust pipeline is in communication with the process chamber to discharge the to-be-detected gas; the absorption cell is arranged in the exhaust pipeline to receive the to-be-detected gas; and the interferometer is optically connected with the absorption cell to output an interference pattern of the corresponding gas.

[0029] By using the above technical solution, the interferometer and the absorption cell are connected through an optical fiber, and the concentration change of the corresponding characteristic gas in the to-be-detected gas is used to judge the process endpoint, thereby improving the accuracy and reliability of the semiconductor manufacturing process endpoint detection. BRIEF DESCRIPTION OF DRAWINGS

[0030] Figure 1 is a step diagram of a detection method according to an embodiment of the present application. DETAILED DESCRIPTION

[0031] In order to make the objects, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below. Obviously, the described embodiments are only some but not all of the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts should fall into the scope of the present application. Unless otherwise defined, the technical terms or scientific terms used herein should be understood as their common meanings to those of ordinary skill in the art to which the present application belongs. The terms such as "comprise" and the like used herein mean that the elements or objects before the terms encompass the elements or objects listed after the terms and their equivalents, and do not exclude other elements or objects.

[0032] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the accompanying drawings. Figure 1 The specific embodiments of the present application will be further described in detail.

[0033] The embodiments of the present application provide a process endpoint detection method used in a semiconductor device. Specifically, the embodiments disclose a semiconductor device, which comprises a process chamber, an exhaust pipeline, an absorption cell and an interferometer. The exhaust pipeline is in communication with the process chamber to discharge a gas to be detected. The absorption cell is arranged in the exhaust pipeline to receive the gas to be detected. The interferometer is optically connected with the absorption cell, for example, through an optical fiber, so as to isolate vibration and heat source, to output an interference pattern corresponding to the gas.

[0034] The detection method disclosed by the embodiments of the present application comprises the following steps:

[0035] S1, inert gas is purged into the absorption cell, and an inert gas interference pattern of the inert gas is collected to obtain an initial background spectrum, which is taken as a reference background spectrum;

[0036] S2, the purging of the inert gas is stopped, the gas to be detected in the exhaust pipeline of the semiconductor device is introduced into the absorption cell, and a gas to be detected interference pattern is collected to obtain a sample spectrum;

[0037] S3, a detection background spectrum is obtained at a set time as a period, and the integral intensity of the detection background spectrum in a reference waveband is compared with the corresponding integral intensity of the reference background spectrum, to determine whether the detection background spectrum needs to be updated as the reference background spectrum;

[0038] S4, under a preset condition, standard absorbance coefficients k of at least three characteristic gases in the gas to be detected are determined, and a coefficient matrix K is constructed;

[0039] S5, the concentration c of the corresponding gas to be detected is obtained according to the coefficient matrix K; ​​

[0040] S6, setting concentration threshold, concentration change rate threshold and byproduct mutation threshold, and determining whether the process endpoint is reached.

[0041] In some embodiments, the process endpoint can be an etching endpoint or a cleaning endpoint of the semiconductor chamber, which is set and selected according to requirements.

[0042] In some embodiments, the original or residual process gas in the absorption cell can be removed by purging the inert gas into the absorption cell. If the original or residual process gas in the absorption cell is not removed, the residual process gas will interfere with the detection process when detecting the exhaust gas. At the same time, since the inert gas has no absorption in the infrared band, the reference background spectrum obtained by Fourier fast change of the detected interference pattern does not contain gas absorption information when the inert gas is used as a reference. Therefore, using the reference background spectrum as a reference can accurately measure the absorption amount of the corresponding process gas.

[0043] In some specific embodiments, the inert gas can be argon or nitrogen, etc., which is not limited herein, and different inert gases can be selected according to requirements in actual detection process.

[0044] In some embodiments, when the absorption cell is filled with inert gas and it is determined that there is no residual process gas in the absorption cell, the inert gas is stopped. The exhaust gas in the exhaust gas discharge pipeline of the semiconductor device is introduced into the absorption cell, that is, the gas to be detected is introduced into the absorption cell. The gas to be detected contains a plurality of characteristic gases, and the characteristic gases can be , , , or , etc. The gas to be detected enters the absorption cell and produces characteristic absorption in the mid-infrared band, thereby obtaining a sample spectrum of the gas to be detected, so that the gas content change can be calculated from the light intensity change.

[0045] In some embodiments, during the detection process, the detection background spectrum needs to be obtained at a set time interval, that is, the background spectrum needs to be detected again at each time interval. Specifically, during the detection process, mirror contamination, light source attenuation, optical fiber aging and other reasons will cause the transmittance to slowly decrease, resulting in deviation of the reference background spectrum. If the initial background spectrum is still used as the reference background spectrum at this time, the system drift will be mistaken for gas absorption during the detection process, and the concentration will gradually increase, thereby causing inaccurate detection results. Therefore, detection needs to be performed at a set interval, and it is determined whether the new detection background spectrum needs to be updated as the reference background spectrum.

[0046] In some embodiments, a standard absorbance coefficient k is established for each detected characteristic gas, so as to facilitate the conversion of the absorbance of the corresponding characteristic gas into concentration. A coefficient matrix K is constructed, which can collect the standard absorbance coefficients k of a plurality of different characteristic gases and serve as a template, so as to facilitate the subsequent calculation of the concentration of the corresponding characteristic gas by using the standard absorbance coefficient k of the corresponding characteristic gas in the coefficient matrix K.

[0047] In some embodiments, in the judgment process of the process endpoint, a judgment threshold needs to be set, i.e., a concentration threshold, a concentration change rate threshold, and a byproduct mutation threshold, and the judgment is made according to the related threshold and the concentration c of the characteristic gas, so as to determine whether the process endpoint is reached.

[0048] In some embodiments, inert gas is purged into the absorption cell, and the initial background spectrum is obtained by collecting the inert gas interference spectrum, including,

[0049] The temperature of the absorption cell is maintained at 150-200°C, and inert gas is purged into the absorption cell; wherein the gas flow of the inert gas is greater than 1 L / min, and the purging time is 30-60 s; specifically, maintaining the temperature of the absorption cell at 150-200°C can prevent the condensation of acidic special cylinders or polymers, and can also avoid the introduction of optical path drift caused by thermal expansion and contraction. The purging time of the inert gas is 30-60 s, which can ensure that there is no residual process gas in the absorption cell, so that the absorption cell is in an inert gas atmosphere. After the absorption cell is in an inert gas atmosphere, the interferometer is started, and a set number of inert gas interference spectra are obtained; specifically, for example, 100 ms per scan, and 32 scans are performed continuously; thereby obtaining 32 inert gas interference spectra. The multiple inert gas interference spectra are point-by-point accumulated and averaged, and a fast Fourier transform is performed, to obtain the initial background spectrum; that is, the average interference spectrum is obtained by point-by-point addition and division by 32 before accumulation, so as to reduce high-frequency random noise; the average interference spectrum is subjected to a fast Fourier transform, thereby obtaining the initial background spectrum.

[0050] In some embodiments, the to-be-detected gas in the exhaust pipeline of the semiconductor equipment is introduced into the absorption cell, and the sample spectrum is obtained by collecting the to-be-detected gas interference spectrum, including,

[0051] The temperature of the absorption cell is maintained at 150-200°C, and the to-be-detected gas is purged into the absorption cell; wherein the gas flow of the to-be-detected gas is greater than 1 L / min, and the purging time is 30-60 s; wherein, the to-be-detected gas purging time is maintained for 30-60 s, which can purge the inert gas in the absorption cell to the outside of the absorption cell, so as to form a to-be-detected gas atmosphere in the absorption cell, thereby facilitating the acquisition of the sample spectrum of the to-be-detected gas. The interferometer continuously records the interference pattern at a set scanning frequency, and acquires a plurality of original interference patterns; specifically, the interferometer scans at a frequency of 10 kHz, and acquires one original interference pattern every 100 ms, and a total of 32 original interference patterns; the plurality of original interference patterns are averaged, and a fast Fourier transform is performed to obtain a sample spectrum; that is, the 32 original interference patterns are averaged and a fast Fourier transform is performed to obtain a sample spectrum.

[0052] In some specific embodiments, the fast Fourier transform is performed on the original interference pattern to obtain the sample spectrum, including,

[0053] The original interference pattern is accumulated to obtain an accumulated energy I;

[0054]

[0055] Wherein, is an optical path difference; is an interferometer response function, and the value is 0-1; is an instantaneous interference pattern intensity; wherein, the instantaneous interference pattern intensity is detected by the interferometer;

[0056] The arithmetic mean value of a set number of accumulated energies I is calculated to obtain an average interference pattern; that is, the arithmetic mean value of the 32 accumulated energies I is calculated, and a fast Fourier transform is performed on the average interference pattern to obtain a sample spectrum.

[0057] In some more specific embodiments, the calculation process of the reference background spectrum is the same as that of the sample spectrum, which will not be described here.

[0058] In some embodiments, the detection background spectrum is acquired periodically at a set time, and the integral intensity of the detection background spectrum in the reference waveband is compared with the corresponding integral intensity of the reference background spectrum to determine whether the detection background spectrum needs to be updated as the reference background spectrum, including,

[0059] The integral intensity of the detection background spectrum in the reference waveband is compared with the corresponding integral intensity of the reference background spectrum to obtain a relative drift amount ,

[0060]

[0061] When relative drift When the relative drift is less than or equal to 5%, the reference background spectrum will continue to be used; when the relative drift is less than or equal to 5%, the reference background spectrum will continue to be used. When the concentration is greater than 5%, inert gas is reintroduced, and a new background spectrum is collected as the reference background spectrum.

[0062] In some specific embodiments, the cycle is set to 4 hours, meaning that the detection is repeated every four hours to obtain the background spectrum. Specifically, during the detection process, the gas to be detected is stopped, and an inert gas is introduced again to create an inert gas atmosphere in the absorption cell. Then, the background spectrum is obtained again. A suitable band is selected for integrated intensity comparison, for example, the non-absorption region. And obtain the integrated intensity of the detected background spectrum in the reference band. Corresponding integrated intensity to the reference background spectrum The relative drift was obtained through calculation. Then, based on the relative drift... Make a judgment, that is, when the relative drift amount When the relative drift is less than or equal to 5%, the reference background spectrum will continue to be used; when the relative drift is less than or equal to 5%, the reference background spectrum will continue to be used. When the concentration exceeds 5%, inert gas is reintroduced, and a new background spectrum is collected as the reference background spectrum. After the reference background spectrum is re-determined, the detection process continues. The method for calculating the integrated intensity is well known to those skilled in the art and will not be elaborated here.

[0063] In some embodiments, under preset conditions, the standard absorbance coefficients k of at least three characteristic gases in the gas to be detected are determined and a coefficient matrix K is constructed, including,

[0064] Detect the absorbance of the desired characteristic gas within a set wavelength range. ,

[0065]

[0066] in, This represents the intensity of the sample spectrum at each wavenumber. The intensity of the baseline background spectrum at each wavenumber;

[0067] Based on absorbance The standard absorbance coefficient k is obtained.

[0068]

[0069] in, This represents the nominal concentration of the corresponding characteristic gas;

[0070] Arrange the standard absorbance coefficients k of at least three characteristic gases to form a coefficient matrix K.

[0071] In some specific embodiments, the preset conditions are that the temperature of the absorption cell is 180±1℃; the optical path is 2-5m; the scanning times are 32; the waveband is .

[0072] In some specific embodiments, at least 3 kinds of end-point related characteristic gases are selected, for example, the selected characteristic gases are , and ; and the absorbance of each corresponding characteristic gas is calculated ; after obtaining the absorbance of the corresponding characteristic gas , the standard absorbance coefficient k of the corresponding characteristic gas is calculated according to the absorbance , and the standard absorbance coefficient k of the characteristic gas is arranged to form a coefficient matrix K; for example, the formed coefficient matrix can be K(m, n), where m is the number of wave points (for example, in the waveband, divided by wave points, with 826 wave points); n is the number of characteristic gases, and each column n represents a characteristic gas.

[0073] In some more specific embodiments, after the coefficient matrix K is constructed, the subsequent concentration of the characteristic gas can be directly obtained in the coefficient matrix K.

[0074] In some embodiments, the concentration c of the corresponding characteristic gas is obtained according to the coefficient matrix K, including,

[0075] the absorbance of the characteristic gas is brought into the linear model with the coefficient matrix K,

[0076]

[0077] wherein, is a noise residual; specifically, the noise residual is derived from the influence of the detector electronic noise, light source jitter, mirror scattering, trace unmodeled gas, numerical rounding, etc.; it is an unavoidable part, and therefore the least square method needs to be used subsequently, so that the concentration c of the required characteristic gas is closer to the true value. At this time, the concentration c of the characteristic gas is in an unknown state, and the linear relationship between the absorbance of the characteristic gas and the concentration of each characteristic gas can be obtained through the linear model.

[0078] In some specific embodiments, the least square method is used to solve each required characteristic gas to obtain the concentration c of the corresponding characteristic gas,

[0079]

[0080] wherein, denotes a transposed matrix obtained by interchanging the rows and columns of the coefficient matrix K.

[0081] In some specific embodiments, the coefficient matrix K is exemplified as an m x n matrix; is a Gram matrix of the coefficient matrix K, i.e., an n x n matrix; is an inverse matrix of the n x n matrix; is an absorbance after projection of the concentration c of the corresponding characteristic gas.

[0082] In some more specific embodiments, the concentration c of the characteristic gas can be calculated according to the above, and the concentration change rate of the corresponding characteristic gas can be calculated according to the concentration c of the characteristic gas, i.e., by dividing the concentration c at time t by the concentration c at time t, which is well known to those skilled in the art and will not be described here. More specifically, the change rate of the byproduct can be understood as the change of the concentration of different characteristic gases.

[0083] In some embodiments, the concentration threshold, the concentration change rate threshold and the byproduct mutation threshold are set, including setting the concentration threshold to 5-10 ppm; and setting according to different characteristic gases, for example, when the characteristic gas is , the concentration threshold is set to 5 ppm. The concentration change rate threshold is set to 0.1-0.6 ppm / s; and setting according to different characteristic gases, for example, when the selected characteristic gas is , the concentration threshold is set to 0.1 ppm. The byproduct mutation threshold is set to 50-60%. Setting according to different characteristic gases, for example, when the byproduct is , the byproduct mutation threshold is set to 50%. More specifically, for example, the byproduct concentration threshold is set to 6 ppm, and the byproduct mutation threshold is set to 50%, so when the detected byproduct concentration is greater than 9 ppm, it represents that the byproduct mutation threshold is greater than 50%.

[0084] In some embodiments, determining whether the etching endpoint is reached includes,

[0085] One of the characteristic gases in the to-be-detected gas is taken as a criterion gas, for example, the characteristic gas is taken as the criterion gas, and the remaining characteristic gases are taken as byproducts;

[0086] When the concentration of the criterion gas is less than the concentration threshold value and the concentration change rate of the criterion gas is less than the concentration change rate threshold value, or the change value of the by-product is greater than the by-product mutation threshold value and the concentration change rate of the criterion gas is less than the concentration change rate threshold value, it is determined that the process endpoint is reached. Specifically, when the concentration of the criterion gas is less than the concentration threshold value and the concentration change rate of the criterion gas is less than the concentration change rate threshold value, it represents that the consumption or generation of the criterion gas has approached the end of the process, i.e. the main chemical reaction or physical process is about to end, and the concentration change rate of the criterion gas is lower than the set threshold value, which means that the decreasing rate of the gas concentration has slowed down, i.e. the reaction rate has slowed down, which usually indicates that the reaction has approached the endpoint; the two criteria are used together to determine the process endpoint, thereby improving the accuracy of the process endpoint determination. When the change value of the by-product is greater than the by-product mutation threshold value and the concentration change rate of the criterion gas is less than the concentration change rate threshold value, the change value of the by-product is greater than the set mutation threshold value, which indicates that an undesirable side reaction may have started and is accelerating. This usually means that the main reaction material has been substantially consumed, and the side reaction material begins to increase significantly. In the judgment process, different ways can be selected for judgment, or the two judgment ways are used to verify each other, so as to determine the process endpoint.

[0087] The implementation principle of the process endpoint detection method and the semiconductor equipment according to the embodiment of the application is that the interferometer and the absorption cell are connected through the optical fiber, and the concentration change of the corresponding characteristic gas in the to-be-detected gas is used to determine the process endpoint, thereby improving the accuracy and reliability of the semiconductor manufacturing process endpoint detection.

[0088] Although the embodiments of the application are described in detail above, it is obvious for those skilled in the art that various modifications and changes can be made to the embodiments. However, it should be understood that such modifications and changes all belong to the scope and spirit of the application described in the claims. Moreover, the application described herein can have other embodiments, and can be implemented or realized in various ways.

Claims

1. A method of detecting a process endpoint, the method comprising: The detection method used in the semiconductor device comprises the following steps: The temperature of the absorption cell is kept at 150-200℃, and inert gas is purged into the absorption cell; the interferometer is started, a set number of inert gas interference patterns are obtained, the multiple interference patterns are point-by-point accumulated and averaged, and fast Fourier transform is performed to obtain an initial background spectrum, which is taken as a reference background spectrum; wherein the gas flow of the inert gas is greater than 1L / min, and the purging time is 30-60s; The purging of the inert gas is stopped, the gas to be detected in the exhaust pipeline of the semiconductor device is introduced into the absorption cell, and the interference pattern of the gas to be detected is collected to obtain a sample spectrum; With a set time as a period, inert gas is introduced to obtain a detection background spectrum; the integral intensity of the detection background spectrum in the reference waveband is compared with the corresponding integral intensity of the reference background spectrum to determine whether the detection background spectrum needs to be updated to the reference background spectrum​ comparing the integrated intensity of the test background spectrum in the reference wavelength band to the corresponding integrated intensity of the reference background spectrum ,​​ when the relative drift amount is less than or equal to 5%, continuing to use the reference background spectrum; and when the relative drift amount is greater than 5%, re-introducing the inert gas and collecting a new background spectrum as the reference background spectrum. when the relative drift amount is less than or equal to 5%, continuing to use the reference background spectrum; and when the relative drift amount is greater than 5%, re-introducing the inert gas and collecting a new background spectrum as the reference background spectrum. when the relative drift amount is less than or equal to 5%, continuing to use the reference background spectrum; and when the relative drift amount is greater than 5%, re-introducing the inert gas and collecting Under preset conditions, standard absorbance coefficients k of at least three characteristic gases in the to-be-detected gas are determined, and a coefficient matrix K is constructed; The preset conditions are that the temperature of the absorption cell is 180±1℃; the optical path is 2–5m; the waveband is ; The concentration c of the corresponding characteristic gas is obtained according to the coefficient matrix K; The concentration threshold, concentration change rate threshold and byproduct mutation threshold are set, and it is judged whether the process endpoint is reached.

2. The detection method according to claim 1, characterized in that, The gas to be detected in the exhaust pipeline of the semiconductor device is introduced into the absorption cell, and the interference pattern of the gas to be detected is collected to obtain a sample spectrum, which comprises, The temperature of the absorption cell is kept at 150-200℃, and the gas to be detected is purged into the absorption cell; wherein the gas flow of the gas to be detected is greater than 1L / min, and the purging time is 30-60s; The interferometer continuously records the interference pattern at a set scanning frequency to obtain multiple original interference patterns; The multiple original interference patterns are averaged, and fast Fourier transform is performed to obtain a sample spectrum.

3. The detection method according to claim 2, characterized in that, The multiple original interference patterns are averaged, and fast Fourier transform is performed to obtain a sample spectrum, which comprises, The original interference patterns are accumulated to obtain accumulated energy I; wherein, is the optical path difference; is the interferometer response function, taking values 0-1; is the instantaneous interferogram intensity; The arithmetic mean value of a set number of the accumulated energy I is calculated to obtain an average interference pattern; Fast Fourier transform is performed on the average interference pattern to obtain the sample spectrum.

4. The method of claim 1, wherein The standard absorbance coefficients k of at least three characteristic gases in the gas to be detected are determined under a predetermined condition, and a coefficient matrix K is constructed, which comprises, detecting the absorbance of the characteristic gas required in a set wavelength range , wherein, is the intensity of the sample spectrum at each wavenumber; is the intensity of the reference background spectrum at each wavenumber; According to the absorbance The standard absorbance coefficient k is obtained, wherein, is the nominal concentration of the corresponding characteristic gas; The standard absorbance coefficients k of at least three characteristic gases are arranged to form the coefficient matrix K.

5. The detection method according to claim 4, characterized in that, The concentration c of the corresponding characteristic gas is obtained according to the coefficient matrix K, which comprises, absorbance of the characteristic gas into a linear model with the coefficient matrix K, wherein is the noise residual; The least square method is used to solve each required characteristic gas to obtain the concentration c of the corresponding characteristic gas, wherein denotes the transpose matrix obtained by interchanging the rows and columns of the coefficient matrix K.

6. The method of claim 1, wherein The concentration threshold, concentration change rate threshold and byproduct mutation threshold are set, which comprises, The concentration threshold is set to 5-10ppm; The concentration change rate threshold is set to 0.1-0.6ppm / s; The byproduct mutation threshold is set to 50-60%.

7. The detection method according to claim 6, characterized in that, The judgment of whether the etching endpoint is reached comprises, One characteristic gas in the gas to be detected is taken as a criterion gas, and the remaining characteristic gases are taken as byproducts; When the concentration of the criterion gas is less than the concentration threshold, and the concentration change rate of the criterion gas is less than the concentration change rate threshold, or the change value of the byproducts is greater than the byproduct mutation threshold and the concentration change rate of the criterion gas is less than the concentration change rate threshold, it is determined that the process endpoint is reached.

8. A semiconductor device, characterized by comprising: The semiconductor device comprises a process chamber, an exhaust pipe, an absorption cell and an interferometer for implementing the detection method of any one of claims 1-7; The exhaust pipe is in communication with the process chamber to discharge the gas to be detected; The absorption cell is arranged in the exhaust pipe to receive the gas to be detected; The interferometer is optically connected with the absorption cell to output the interference pattern of the corresponding gas.

Images