Plasma treatment apparatus, plasma treatment method and etching system

Abstract

A plasma processing apparatus, plasma processing method, and etching processing system are provided that achieve high-precision film thickness monitoring and endpoint determination by a method for obtaining information that does not rely on a database and by performing signal processing. The plasma processing apparatus (70), which measures wafer-reflected light from the wafer being processed during etching and uses a waveform pattern database (122) to estimate the film thickness or depth of the wafer from the wafer-reflected light, includes a film thickness / depth determination unit (102) that performs data processing to ensure that the correspondence between the film thickness or depth used in the waveform pattern database (122) and the light intensity remains consistent before and after the second signal processing is performed on the light intensity.

Description

[Technical Field]

[0001] This invention relates to a plasma treatment apparatus, a plasma treatment method, and an etching system. [Previous Technology]

[0002] In the manufacture of semiconductor devices, various components or interconnecting wirings are formed on the surface of a wafer. These devices are formed by repeatedly depositing films of various materials such as conductors, semiconductors, and insulators, and removing excess material. As a process for removing excess material, dry etching using plasma (also known as plasma etching) is widely used. In plasma etching, the gas introduced into the processing chamber of the etching apparatus is plasmaified using a high-frequency power supply, and the wafer is processed by exposing it to the plasmaified gas. At this time, anisotropic or isotropic etching occurs due to sputtering caused by ions in the plasma or chemical reactions caused by free radicals. By separating and using these etching processes, components or wirings of various structures are formed on the surface of the wafer.

[0003] When the processed shape resulting from plasma etching differs from the design, the various components formed cannot perform their functions. Therefore, numerous process monitoring techniques have been proposed to monitor and stabilize the etching process. Process monitoring that measures the thickness of the film formed on the wafer or the depth of the trenches or holes formed on the wafer by measuring the reflected light from the wafer during processing is called film thickness / depth monitoring, and is increasingly being used for determining the end point of the etching process.

[0004] Patent Document 1 describes a method for improving processing accuracy using film thickness / depth monitoring. This document describes using film thickness / depth monitoring with plasma light as the light source to detect the moment just before the film of the object being processed is completely removed, and then ending the etching process. Subsequently, the etching process is performed under conditions that allow for highly selective etching of the processed portion and the non-processed portion.

[0005] Patent Document 2 describes a high-precision technique for measuring film thickness / depth in film thickness / depth monitoring. This document describes the use of an external light source instead of plasma light as the light source for illuminating the wafer.

[0006] Patent Document 3 describes a high-precision method for estimating film thickness / depth using a plurality of waveform pattern databases that represent the correspondence between reflected light from a wafer (wafer reflected light) and film thickness / depth. This document describes calculating the estimated film thickness / depth using a weighted average based on the difference between the measured wafer reflected light and the plurality of waveform pattern databases. [Prior Art Documents] [Patent Documents]

[0007] [Patent Document 1] Japanese Patent Application Publication No. 2006-119145 [Patent Document 2] Japanese Patent Application Publication No. 2004-507070 [Patent Document 3] Japanese Patent Application Publication No. 2022-058184 [Summary of the Invention]

[0008] [Problem to be Solved by the Invention] In methods for calculating film thickness or depth using multiple waveform pattern databases (hereinafter referred to as databases), multiple databases need to be prepared in advance. These databases are data that establishes a correspondence between the wafer reflected light (light intensity) obtained at various moments during wafer processing and the film thickness or depth during processing. For example, in this processing process, by etching the wafer and obtaining wafer reflected light, the film thickness or depth before and after etching is measured, and the film thickness or depth before and after etching is linearly interpolated and distributed to the film thickness or depth at each moment during etching to obtain the database. Or, for example, if the relationship between the wafer film thickness or depth and wafer reflected light in this processing process is calculable, the database is obtained by calculation.

[0009] The estimation of film thickness or depth using these databases is performed based on the wafer reflected light measured during the processing of the wafer to be estimated. At this time, the wafer reflected light undergoes various signal processing to remove various light quantity variations, i.e., noise components, that are added during measurement. For example, LPF (Low Pass Filter) is performed to remove random variations in light quantity per unit time, and differential or derivative calculations are performed to remove light quantity and bias components. Since the film thickness or depth is calculated based on the wafer reflected light that has undergone these signal processing steps, the same signal processing is performed even for general databases.

[0010] Here, in determining the end point of etching, it is necessary to estimate the film thickness / depth in real time from the wafer reflected light during etching, and determine the etching end point based on the estimated value, thus ending the etching operation. Therefore, the signal processing performed on the wafer reflected light or the database is limited to signal processing that can be executed in real time. In the calculation of the time direction of LPF or differential value in these signal processing, there may be cases where the correspondence between the time and the measured wafer reflected light changes before and after the signal processing. Due to this change in correspondence, for example, in the estimation of film thickness or depth using multiple databases, the correspondence between film thickness or depth and wafer reflected light will contradict each other among multiple databases, and the accuracy of the film thickness or depth estimation will be reduced.

[0011] Furthermore, the change in the correspondence between the time before and after the aforementioned signal processing and the measured wafer reflected light in the wafer being etched can lead to a decrease in the accuracy of the etching endpoint determination. For example, even if the correspondence between the film thickness or depth and the wafer reflected light is the same before signal processing, it will still be different after signal processing if the etching speed of the wafer for which the film thickness or depth is to be estimated is faster than if it is slower. In this case, if the same database or multiple databases with a fixed (consistent) correspondence between film thickness / depth and wafer reflected light are used to estimate the film thickness or depth based on the wafer reflected light after signal processing, even if the estimated film thickness or depth is the same, the actual film thickness or depth will still be different. As a result, wafers processed by endpoint determination will exhibit variations in the film thickness or depth after processing for each wafer.

[0012] The present invention addresses the aforementioned problems. For example, it provides a plasma processing apparatus, plasma processing method, and etching processing system that achieve high-precision film thickness monitoring and endpoint determination by using a database-independent acquisition method and signal processing. [Means for solving the problems]

[0013] One embodiment of this disclosure is a plasma processing apparatus that measures wafer-reflected light from a wafer being processed during etching and uses a database to estimate the film thickness or depth of the wafer being processed from the wafer-reflected light. The apparatus includes: a data processor that performs data processing to maintain the correspondence between film thickness or depth and light intensity used in the database consistent before and after performing a second signal processing on the light intensity; and a film thickness / depth correction device that corrects for the film thickness / depth estimated in real-time from the wafer-reflected light during etching using the deviation in the correspondence between film thickness or depth and light intensity obtained during the data processing that maintains consistency before and after the second signal processing. [Effects of the Invention]

[0014] If based on the representative embodiment disclosed herein, the correspondence between film thickness or depth and light intensity in the database will be constant (unchanging) regardless of the measurement conditions of the database or the conditions of the second signal processing performed on the database. There will be no contradiction in the correspondence between film thickness or depth and light intensity among multiple databases.

[0015] Furthermore, the estimated film thickness / depth will always correspond to the actual film thickness / depth, regardless of the deviation in the correspondence between film thickness / depth and light intensity caused by the processing of the first signal between the wafers being etched.

[0016] Therefore, high-precision estimation of film thickness or depth can be achieved, as well as high-precision endpoint determination using the estimated film thickness or depth. The issues, structures, and effects other than those mentioned above will become clear from the following description of the embodiments.

Implementation Method

[0018] In the following embodiments, for convenience, they are divided into multiple sections or embodiments for explanation when necessary. However, unless otherwise expressly stated, these are not unrelated to each other, and one is a variation, detail, or supplementary explanation of a part or all of the other. Furthermore, in the following embodiments, when referring to the quantity of elements (including number, value, quantity, range, etc.), except as expressly stated or clearly limited to a specific quantity in principle, it is not limited to that specific quantity, but may be more than or less than that specific quantity.

[0019] Furthermore, in the following embodiments, the constituent elements (including element steps, etc.) are not necessarily essential, except where specifically stated or where they are clearly considered necessary in principle. Similarly, in the following embodiments, when referring to the shape, positional relationship, etc., of the constituent elements, etc., it substantially includes those that are similar or analogous to that shape, etc., except where specifically stated or where they are clearly not considered to be so in principle. The same applies to the values ​​and ranges mentioned above.

[0020] Hereinafter, a plasma processing apparatus, a plasma processing method, and an etching system, embodiments of the present invention, will be described with reference to schematic diagrams. The configuration of a plasma processing apparatus for etching, equipped with a film thickness / depth estimation means, will be described first, followed by a description of the film thickness / depth estimation method (plasma processing method) in the etching process within the plasma processing apparatus, as an embodiment of the present invention. [Example 1]

[0021] A schematic diagram of the plasma processing apparatus used in this embodiment is shown in FIG1. ​​The plasma processing apparatus 70 includes a plasma processing unit 55. The plasma processing unit 55 includes a vacuum processing chamber 10. Etching gas introduced into the vacuum processing chamber 10 from a gas introduction means is excited and decomposed by electricity or microwaves generated using a high-frequency power supply or the like, turning into plasma 12. Using this plasma 12, the semiconductor wafer or other processing object 16 placed on the sample stage 14 is etched (plasma processing). The plasma processing apparatus 70 is sometimes referred to as an etching processing apparatus.

[0022] Furthermore, the plasma treatment apparatus 70, as shown in FIG1, in addition to the plasma treatment unit 55, also includes: a film thickness / depth measurement calculation unit 30, a control unit 40, and an endpoint determination unit 44. The film thickness / depth measurement calculation unit 30 and the endpoint determination unit 44 will be described later.

[0023] The control unit 40 performs the following: introducing gas into the vacuum processing chamber 10, generating and controlling the plasma 12, and applying voltage to the object being processed using a high-frequency power supply, etc. Through these controls, timing adjustments are made between the various machines to achieve the desired etching process. When the plasma 12 is pulsed, the pulse control system is also performed by the control unit 40. At this time, the plasma 12 is pulsed by switching its On / Off states through voltage application caused by a high-frequency power supply that plasmaifies the etching gas, microwave irradiation, etc. Furthermore, plasma pulsedness can also be achieved by time-modulating the introduction of the etching gas.

[0024] The plasma processing unit 55 is equipped with a mechanism for measuring the etching amount, film thickness, or depth of the workpiece 16. Light emitted from the light source unit 18 is guided into the vacuum processing chamber 10 via the optical system 50, irradiating the workpiece 16 with the irradiation light 22. The light source unit 18 uses continuous light from ultraviolet to infrared light; however, only the wavelength required for measuring the film thickness or depth of the workpiece 16 needs to be used, or it can be continuous light within a specific wavelength band or a specific wavelength. The reflected light 24 from the workpiece 16 is guided to the detection unit 28 via the optical system 50.

[0025] Here, the light irradiation optical port in the optical system 50 that irradiates light into the vacuum processing chamber 10 is composed of, for example, an optical fiber that guides light from the light source 18 and a lens that irradiates light into the processing object 16 inside the vacuum processing chamber 10. Furthermore, the light receiving optical port in the optical system 50 that receives reflected light 24 is composed of, for example, a lens that focuses reflected light 24 and an optical fiber that guides the focused light to the detection unit 28. The light irradiation optical port and the light receiving optical port are not limited to the above configurations. When these ports are configured as a single port, they can also be configured using an optical fiber bundle consisting of two and one optical fiber at each end, and a lens.

[0026] Furthermore, when measuring the reflected light 24 at multiple locations of the processing object 16, it is sufficient to provide a plurality of light-illuminating optical ports and / or a plurality of light-receiving optical ports. The configuration of these optical ports is not limited to the use of optical fibers and lenses; the light-illuminating optical ports can also be configured using planar light-emitting elements, and the light-receiving optical ports can also be configured using a camera with a CCD or CMOS sensor.

[0027] When using a specific wavelength band or multiple wavelengths to estimate the film thickness or depth, the detection unit 28 is composed of a beam splitter, which splits the introduced light and detects the amount of light at each wavelength. When using a specific wavelength to estimate the film thickness or depth, the detection unit 28 is not limited to a beam splitter; a photodetector or the like can also be used. In this case, as long as the light introduced into the detection unit 28 is the desired specific wavelength, a photodetector can be used directly. When continuous light is introduced, a mechanism for selecting only specific wavelengths can be set up in front of the photodetector using a monochromator or the like. The detection of this light is continuously performed at certain intervals during etching, such as 2Hz or 10Hz. The output data of the detection unit 28 is fed into the film thickness / depth calculation unit 30, and the film thickness or depth is determined (estimated). In other words, the detection unit 28 functions as a light receiver that receives reflected light from the wafer during plasma processing.

[0028] Although the above description addresses the case where light is incident from an external light source unit 18, the light source unit 18 may not be used if the light source is the plasma 12. When the plasma 12 is used as the light source, the light emitted by the plasma 12 is also reflected by the object being processed 16, and the reflected light 24 is detected in the same way as when the light source unit 18 is used.

[0029] An example of the configuration of the functional blocks of the film thickness / depth calculation unit 30 is shown in FIG2. The film thickness / depth calculation unit 30 includes: a digital signal processing unit 100, a film thickness / depth determination unit 102, a film thickness / depth correction unit 104, and a waveform pattern database 122. In addition, "film thickness / depth" refers to the film thickness of the film formed on the wafer (processing object 16) or the depth of the trenches or holes formed on the wafer. Moreover, the "film thickness" described in the claim is also similar to "film thickness / depth", and is not limited to the film thickness of the film formed on the wafer, but also includes the depth of the trenches or holes formed on the wafer.

[0030] The time series data D1 of light quantity of each wavelength imported from the detection unit 28 to the film thickness / depth calculation unit 30 is supplied to the film thickness / depth determination unit 102 as time series data D2 after various noises or variations are removed or corrected by the digital signal processing unit 100.

[0031] The digital signal processing unit 100 performs predetermined signal processing (first signal processing) on ​​the output data (intensity of reflected light received by the light receiver) of the detection unit 28 and outputs it. The digital signal processing unit 100 uses a low-pass filter (LPF) to remove random noise from, for example, the time-series data of light quantity. For example, the relationship between the time-series data D1 and D2 before and after signal processing in a Butterworth-type LPF is shown in Equation 1 below.

[0032]

[0033] In Equation 1, λ represents the LPF method using data at 3 time points, t represents the time, and a and b are coefficients determined by the sampling rate of the time series data and the cutoff frequency of the LPF. For example, with a sampling rate of 0.5 sec and a cutoff frequency of 0.1 Hz, {a0, a1, a2, b1, b2} = {0.0201, 0.0402, 0.0201, -1.5610, 0.6414}.

[0034] For example, when removing the light intensity bias from the time series data of light intensity, signal processing is used to calculate the change in light intensity or the differential value between time points. For example, the differential value is calculated using Savitzky-Golay, and the relationship between the time series data D1 and D2 before and after signal processing is shown in Equation 2 below.

[0035]

[0036] In Equation 2, the method for calculating the first derivative value using data at 5 time intervals is expressed, and 'a' is a coefficient determined by the number of time intervals used and the degree of differentiation. For example, when calculating the first derivative value using data at 5 time intervals, {a0, a1, a2, a3, a4} = {2, 1, 0, -1, -2}.

[0037] For example, when the multiplier variation of the light intensity across the entire wavelength is removed and corrected from the spectrometer data (spectral data) at each time point, signal processing is applied to normalize the light intensity of each wavelength by using the average or absolute sum of the light intensity across the entire wavelength. Normalization using the sum of absolute values ​​is implemented by, for example, the following formula 3.

[0038]

[0039] The signal processing and data operation in the digital signal processing unit 100 are not limited to the above. They can also perform the signal processing and data operation required for determining the film thickness / depth in the film thickness / depth determination unit 102, such as data reduction, simplification, interpolation, smoothing, statistical processing, comparison, etc.

[0040] In the film thickness / depth determination unit 102, the film thickness / depth is determined (estimated) using the supplied time series data D2 and the data supplied from the waveform pattern database 122. An example of the configuration of the functional blocks of the film thickness / depth determination unit 102 is shown in FIG3. That is, the film thickness / depth determination unit 102 functions as a determiner that determines the thickness of the wafer film based on the intensity of reflected light that has been signal-processed by the signal processor, according to the database.

[0041] The film thickness / depth determination unit 102 includes: a waveform comparator 302, a film thickness / depth determination unit 304, an optimal database determination unit 306, a suitable database calculator 308, and a film thickness / depth calculation model generator 310.

[0042] For example, the determination of the film thickness / depth can be implemented by comparing the spectrum 320 (i.e., time series data D2) obtained from the etching supplied from the digital signal processing unit 100 with a database 322 supplied from the waveform pattern database 122. Here, an example of the content of the spectrum 320 is shown in FIG4(a). The spectrum sp(t) at each time t is a data table that establishes a correspondence between wavelengths w1 to wM and light quantities I1,t to IM,t.

[0043] Another example of the contents of database 322 is shown in Figure 4(b). One database is a data table that establishes a correspondence between film thickness / depth d1 to dN and spectrum db-sp1 to db-spN. That is, database 322 establishes a correspondence between the intensity of light at a given wavelength and the thickness of the film formed on the reference wafer, which serves as the basis for determining the film thickness.

[0044] In determining the film thickness / depth using a single database 322, the spectrum 320 and the database 322 are directly supplied to the waveform comparator 302. In the waveform comparator 302, the spectrum sp(t) at each moment is compared with each spectrum db-sp in the database 322, and the calculated error (or similar measure) is supplied to the film thickness / depth determiner 304. In the film thickness / depth determiner 304, based on the error (or similar measure), the optimal spectrum is determined from the spectrum db-sp in the database, and the film thickness / depth associated with the determined spectrum is determined as the film thickness / depth for that spectrum sp(t). The determined film thickness / depth is supplied to the film thickness / depth correction unit 104. By using a single database 322 to determine the film thickness / depth, the capacity of memory devices, such as those constructed using a waveform pattern database 122, can be reduced.

[0045] For example, the determination of film thickness / depth can also be implemented by comparing the spectrum 320 supplied from the digital signal processing unit 100 with a plurality of databases 322 supplied from the waveform pattern database 122. That is, the waveform pattern database 122 can also store a plurality of databases. For example, the spectrum 320 and the plurality of databases 322 are directly supplied to the waveform comparator 302, and the spectrum sp(t) at each time moment is compared with the full spectrum db-sp of the plurality of databases 322 to calculate their error. The calculated error is then supplied to the film thickness / depth determiner 304.

[0046] The calculated error amount is also supplied to the optimal database determiner 306. Based on the error amount of the spectrum of each database, an optimal database is determined from the plurality of databases. The identification information of the determined database is supplied to the film thickness / depth amount determiner 304.

[0047] In the film thickness / depth determination unit 304, the identification information of the supplied database is used to determine the database to be used. Then, the error between the supplied database and the spectrum sp(t) is used to determine the spectrum of the optimal database. Then, the film thickness / depth associated with the determined spectrum is used to determine the film thickness / depth amount as the spectrum sp(t). The determined film thickness / depth amount is supplied to the film thickness / depth correction unit 104. By using multiple databases, even if the data in the database has noise, the number of databases can be increased to ensure a certain level of accuracy.

[0048] Here, the suitable database calculator 308 can use the error amount of the supplied plurality of databases 322 and the waveform comparator 302 to perform weighted summation on the plurality of databases 322 to generate a composite database with an error amount lower than the supplied error amount. Then, the suitable database calculator 308 has the function of adding the generated composite database to the plurality of databases. When it is desired to improve the estimation accuracy of film thickness / depth, the suitable database calculator 308 can also be used to generate the composite database and then perform the estimation of film thickness / depth.

[0049] Furthermore, for example, the determination of the film thickness / depth can be implemented by using the spectrum 320 supplied from the digital signal processing unit 100 and a learning model (linear / nonlinear regression function, machine learning, deep learning, etc.) of the spectrum and film thickness / depth created using a plurality of databases 322 supplied from the waveform pattern database 122. For example, the plurality of databases 322 are supplied to the film thickness / depth calculation model generator 310, and a learning model is created in the film thickness / depth calculation model generator 310.

[0050] The learning model generated from the correspondence between spectra and membrane thickness / depth from multiple databases can be, for example, a regression function generated using kernel ridge regression or support vector machine, an analysis / classification model generated using random forest or k-nearest neighbor method, or a learning model generated using neural network-like systems. The learning model generated in the membrane thickness / depth calculation model generator 310 and the spectrum 320 supplied from the digital signal processing unit 100 are supplied to the membrane thickness / depth quantity determiner 304. In the membrane thickness / depth quantity determiner 304, the learning model is used to determine the membrane thickness / depth quantity corresponding to the spectrum sp(t). The determined membrane thickness / depth quantity is supplied to the membrane thickness / depth quantity correction unit 104. By using the learning model to determine the membrane thickness / depth quantity, the learned data can be processed correctly and quickly.

[0051] As described above, the film thickness / depth quantity supplied to the film thickness / depth quantity correction unit 104 is the time series data D3 in FIG2. In the film thickness / depth quantity correction unit 104, the film thickness / depth quantity at each time moment can be corrected based on the calculated film thickness / depth quantity time series data D3. For example, if there are fluctuations in the calculated film thickness / depth quantity time series data D3 caused by noise, the time transition of the film thickness / depth quantity is approximated linearly, and the film thickness / depth quantity at each time moment is corrected. The film thickness / depth quantity corrected in the film thickness / depth quantity correction unit 104 is output as film thickness / depth quantity data D4.

[0052] The plasma processing apparatus 70 in Figure 1 uses the film thickness / depth data supplied from the film thickness / depth calculation unit 30 to perform endpoint determination in the endpoint determination unit 44. The endpoint determination unit 44 compares the acquired film thickness / depth data with a predetermined film thickness / depth as a target, and determines whether the current film thickness / depth has reached the target. If the target is determined to have been reached, a stop signal for the etching process is supplied to the control unit 40. The control unit 40 stops the etching process based on the etching stop signal. Through this operation, the plasma processing apparatus 70 can perform endpoint determination using film thickness / depth monitoring. That is, the endpoint determination unit 44 functions as a determiner that determines the endpoint of the process based on the film thickness determined by the determiner.

[0053] As explained above, signal processing is performed in the digital signal processing unit 100, decision processing is performed in the film thickness / depth determination unit 102, and decision processing is performed in the endpoint determination unit 44. Furthermore, as will be described later, data processing is performed in the film thickness / depth determination unit 102.

[0054] Here, the waveform pattern database 122 of FIG2 used in the determination of film thickness / depth described above is obtained before the estimation of the film thickness / depth of the wafer to be processed is performed. For example, one method of obtaining the database by etching a test sample is shown in FIG5(a). In this method, the film thickness / depth of the test sample before etching is first pre-measured (S11). If non-destructive measurement is possible, the test sample is evaluated; if destructive measurement is possible, a test sample equivalent to the test sample is measured.

[0055] Next, the test sample is etched, and the spectrum during etching is measured (obtained) (S12). Then, the film thickness / depth of the etched test sample is measured non-destructively or destructively (S13). Using the measured spectrum and the film thickness / depth before and after etching, the film thickness / depth is allocated to the spectrum at each moment during etching (S14), and a database is created / obtained by performing the desired digital signal processing on the spectral data (S15). That is, the light intensity and film thickness contained in the database before correction are obtained based on the results of processing using a pre-prepared test sample (a sample wafer prepared in advance as a reference wafer). Furthermore, by using the test sample to generate the database, each piece of data can be reliably obtained.

[0056] The method of allocating film thickness / depth for the spectrum at each time moment (S14) and the signal processing of the spectral data (S15) in Figure 5(a) will be explained with reference to Figure 5(b). In the data table of Figure 5(b), column c1 represents the time when the spectrum was obtained during etching, column c2 represents the obtained spectrum, column c3 represents the film thickness / depth amount corresponding to each time moment, and column c4 represents the result of signal processing (second signal processing) on ​​the spectrum. Here, the film thickness / depth in column c3 is determined, for example, by linearly interpolating the initial value d1 measured before etching and the final value dN measured after etching to allocate to each time moment.

[0057] The signal processing method (second signal processing) for determining the spectral data in column c4 can also be the same as, for example, the signal processing performed in the digital signal processing unit 100 for the time series data D1 in Figure 2. That is, the first signal processing and the second signal processing can also be the same processing content. In this case, the same signal processing will be performed on the spectral data and database used in determining the film thickness / depth, and their comparison can be performed correctly. For example, if the determination of the film thickness / depth is possible, it can also be a different signal processing than that of the digital signal processing unit 100. The database thus created is a table of data corresponding to column c3 of film thickness / depth and column c4 of spectrum in Figure 5(b) (database before correction).

[0058] Here, the database is not limited to the method of etching the test sample as shown in FIG5(a), but can also be created, for example, by simulation using optical calculations. In this case, the relationship between film thickness / depth and spectrum is known, so the database can be created without performing S11 to S14 in FIG5(a).

[0059] The database obtained by the above method will be corrected as needed to adjust the correspondence between film thickness / depth and spectrum. For example, in the signal processing of S15 in Figure 5(a), if there are differences in the correspondence between time and / or film thickness / depth and spectrum before and after signal processing, the correspondence after signal processing will be corrected. The method for correcting the correspondence between time and / or film thickness / depth and spectrum is shown in Figure 6(a).

[0060] For example, the time delay Δt caused by the signal processing is calculated (S21), and the correspondence between time and spectrum in the database is shifted by Δt (S22). Figure 6(b) shows the correction result of the database using the calculated time delay Δt. Column c4 in Figure 6(b) is the same as column c4 in Figure 5(b). Here, the time t corresponding to each spectrum in column c4 of the database is shifted to t-Δt, and the new corresponding time column c5 is calculated.

[0061] Next, using the correspondence between columns c1 and c3 in Figure 5(b), the d' corresponding to each t-Δt in column c5 of Figure 6(b) is calculated, and column c6 is determined. Then, the correspondence between column c6 of film thickness / depth and column c4 of spectrum is determined as the corrected database. That is, the time delay caused by signal processing is calculated, and the correspondence between light intensity and film thickness is corrected based on this time delay. Through such correction, the correspondence between film thickness / depth and spectrum in the database caused by the change in the correspondence between time and spectrum before and after signal processing can be kept consistent (constant). Especially in the estimation of film thickness / depth using multiple databases, the contradiction between databases can be eliminated, and the reduction in the estimation accuracy can be suppressed.

[0062] The calculation of the time delay Δt in S21 of Figure 6(a) in the above-described correction is performed by, for example, the method shown in Figure 7 using a test signal. In Figure 7, a sine wave of a specific frequency in which the time delay caused by signal processing can be observed is prepared as the test signal I(t) (S31). Next, signal processing is performed on the test signal I(t), and the processed test signal I'(t) is obtained (S32). Then, Δt is calculated by comparing the test signals I(t) before and after signal processing with I'(t) (S33).

[0063] For example, the calculation of the time delay Δt can also be carried out by using the measured signal method shown in Figure 8. Here, the database is first created according to the procedure in Figure 5(a), and the database DB corresponding to column c3 (film thickness / depth) and column c2 (spectrum) in Figure 5(b) is prepared (S41).

[0064] Furthermore, a corresponding database DB' is prepared to be established for column c3 (film thickness / depth) and column c4 (spectrum after signal processing) (S42). Then, DB is used as the database for film thickness / depth estimation, and DB' is used as the spectrum of the estimation object for film thickness / depth estimation to perform film thickness / depth estimation (S43).

[0065] Using the film thickness / depth quantity d'i obtained by estimating the film thickness / depth and the film thickness / depth quantity di in DB', the film thickness / depth delay is calculated by Δdi = (d'i - di). Using the relationship between column c1 (time) and column c3 (film thickness / depth quantity) in Figure 5(b), the time delay Δt is calculated from the film thickness / depth delay Δdi (S44). The time delay Δt is calculated, for example, by using the average value Δdave of the film thickness / depth delay Δdi, by Δdave × (tN - t1) / (dN - d1).

[0066] By using the time delay Δt calculated by these methods, the correspondence between film thickness / depth and spectrum in the database can be corrected.

[0067] Although the above description explains the situation where the signal processing in S15 of Figure 5(a) causes a difference in the correspondence between film thickness / depth and spectrum, if signal processing in S15 is used that does not cause a difference in the correspondence between film thickness / depth and spectrum, then the correction of the film thickness / depth and spectrum correspondence in the database caused by Figure 5(a) is not necessary. That is, the signal processing itself can also be data processing that does not change the correspondence between light intensity and film thickness.

[0068] For example, in the low-pass filter of Equation 1, after performing low-pass filter calculation on the data time series from time t1 to time tN in the database, the low-pass filter calculation is performed again on the data series from time tN to time t1. Thus, the time delay caused by the first calculation is canceled out by the second calculation, and the time delay of the time series data after the second calculation becomes zero.

[0069] For example, in the calculation of the differential value of Equation 2, the time delay is made zero by changing the calculation process to Equation 4 below.

[0070]

[0071] In this calculation process, the differential value at time tn is calculated using data from times tn+1 and tn+2, which are further ahead of time tn. Therefore, if this calculation process is used on data acquired in real time during etching, the real-time performance of the output will be reduced. On the other hand, since the database contains already acquired data, this calculation process can make the time delay of the data after the differential value is calculated zero.

[0072] Although the above describes a method to keep the correspondence between film thickness / depth and spectrum in the database consistent (unchanged) before and after signal processing, the difference in film thickness estimation value caused by the database can also be reduced by correcting the output estimated film thickness value using a database whose correspondence between film thickness / depth and spectrum changes with signal processing.

[0073] For example, for each database, the time delay Δt is calculated using step S21 in FIG6(a). The etching rate calculated from column c1 (time) and column c3 (film thickness / depth) in the data table of FIG5(b) of the database is used to determine the film thickness / depth change Δd corresponding to the time delay Δt. In the film thickness / depth estimation using each database, the film thickness / depth estimation value obtained at each time is corrected using the film thickness / depth change Δd of the database. In this way, the corrected film thickness / depth estimation value at each time is unique and not dependent on the database used. Even when using such a method, the same effect as the method of maintaining the correspondence between the film thickness / depth of the database and the spectrum consistent before and after the signal processing is implemented can still be obtained.

[0074] Signal processing that does not cause a difference in the correspondence between film thickness / depth and spectrum is not limited to the above, but can also be signal processing such as zero-phase filter, where the correspondence between film thickness / depth and spectrum does not change before and after signal processing.

[0075] Here, although the method for maintaining the correspondence between film thickness / depth and spectrum in the database before and after signal processing has been described above, it is also desirable that the correspondence between measurement time and spectrum does not change before and after signal processing, even in the implementation object of film thickness / depth estimation, namely spectrum 320. This is because, in the implementation object of film thickness / depth estimation, namely spectrum 320, the correspondence between time and spectrum may differ due to the implementation of signal processing (time delay). Sometimes, depending on the amount of time delay, the correspondence between time and spectrum, that is, the correspondence between film thickness / depth and spectrum, may change.

[0076] Therefore, the film thickness / depth value estimated from the spectrum will vary with the amount of time delay. Here, the correspondence between the measurement time and the spectrum is, for example, the relationship between the peaks or troughs in the time series variation of light intensity in Figure 10 and the time. For spectrum 320, as a signal processing method that does not change the correspondence between the measurement time and the spectrum, methods such as implementing a low-pass filter twice in both ascending and descending directions of the time axis as described above, the method of calculating using differential values ​​in Equation 4, and a zero-phase filter can be used. In this case, although the immediacy of the film thickness / depth estimation may be reduced, for example, by correctly outputting the film thickness / depth estimation value of the time before the current time in the etching process, and extrapolating the film thickness / depth of the current time from the output film thickness / depth estimation value, the film thickness / depth at the current time can be estimated.

[0077] The above-mentioned database acquisition and database correction calculations (Figures 5 to 8) are mainly performed in the film thickness / depth calculation unit 30. That is, the film thickness / depth calculation unit 30 functions as a data processor, which performs signal processing (second signal processing) on ​​the light intensity contained in the waveform pattern database 122, and performs data processing to maintain the correspondence between light intensity and film thickness consistent before and after signal processing.

[0078] Furthermore, the calculation and processing of database acquisition and database correction are not limited to being performed in the plasma processing apparatus 70 (film thickness / depth calculation unit 30), but can also be performed, for example, in the host computer 60 shown in FIG1. ​​The symbol 80 shown in FIG1 is also a diagram representing the etching system formed by the host computer 60 and the plasma processing apparatus 70.

[0079] The host computer 60 is connected to the plasma processing device 70 via a network such as a communication cable or wirelessly, and uses data such as the spectrum supplied from the plasma processing device 70 to create a database. The host computer 60 supplies the created database to the plasma processing device 70. Furthermore, the host computer 60 has the function of generating learning models used in film thickness / depth estimation. It uses the database supplied from the plasma processing device 70 to generate learning models for estimating film thickness / depth, and can also supply the created learning models to the plasma processing device 70. That is, the host computer 60 functions as a data processing device for generating the database, and during database generation, it performs data processing to maintain the correspondence between light intensity and film thickness consistent before and after the second signal processing.

[0080] Here, the function of the film thickness / depth correction unit 104 in FIG2 described above is to reduce the variation in the determined value of film thickness / depth caused by noise, etc., but the correction of film thickness / depth is not limited to this. For example, if the correspondence between the determined value of film thickness / depth and the actual processed film thickness / depth changes due to the etching rate of the wafer during etching, the determined film thickness / depth can also be changed by a predetermined calculation.

[0081] For example, in data processing used to maintain the correspondence between film thickness / depth and spectrum in the database before and after signal processing, the film thickness / depth amount determined by the time delay Δt caused by the signal processing is corrected. An example of a functional block of the film thickness / depth amount correction unit 104 is shown in FIG16. The determined value 1620 of the film thickness / depth amount up to the present time, supplied from the film thickness / depth amount determination unit 102, is input to the etching rate detector 1602 of the film thickness / depth amount correction unit 104.

[0082] In the etching rate detector 1602, the etching rate is calculated using the film thickness / depth at each moment. For example, the etching rate is calculated from the slope of a linear approximation of the relationship between time and film thickness / depth. Alternatively, when the etching rate detector 1602 is supplied with the spectral data of the etching process in Figure 2 (i.e., the time series data D2) and the spectral data from the waveform pattern database 122 via a path not shown, the etching rate can also be calculated using the spectral data of ...

[0083] The etching rate calculated in the etching rate detector 1602 is supplied to the film thickness / depth correction unit 1604. In the film thickness / depth correction unit 1604, the supplied etching rate and the time delay Δt caused by the signal processing supplied via a path not shown are used to correct the determined value of the film thickness / depth. For example, the correction is performed by adding the product of the etching rate ER(t) and the time delay Δt to the determined value d(t) of the film thickness / depth at that moment. This is because the signal processing maps the spectrum of past moments to each moment, thus correcting the past film thickness / depth determined by the spectrum to the film thickness / depth at the present moment.

[0084] The corrected film thickness / depth value in the film thickness / depth corrector 1604 is used as film thickness / depth data and supplied to the endpoint determination unit 44 in FIG1, where the endpoint determination operation is performed in the same manner as described above. Furthermore, these calculation processes are performed on the host computer 60 in the same manner as the data processing during database generation.

[0085] Next, the endpoint determination result caused by the film thickness / depth estimation of the present invention using the plasma processing apparatus 70 will be explained as follows.

[0086] In this embodiment, it is explained that a plurality of databases are obtained by actual measurement, and the correspondence between film thickness / depth and spectrum in the databases is corrected by using test analog signals. The film thickness / depth estimation results are implemented using these plurality of databases.

[0087] An example of a database obtained through actual measurement is shown in Figure 9. There are 5 databases, and the etching process during database acquisition is for a film thickness of approximately 230 nm to approximately 200 nm, but the etching rate during processing of each database is different. The digital signal processing implemented in the digital signal processing unit 100 in Figure 2 is a combination of sequentially implementing the low-pass filter of Equation 1, the first derivative calculation of Equation 2, the low-pass filter of Equation 1, and the spectral normalization of Equation 3. The film thickness / depth determination method in the film thickness / depth determination unit 102 in Figure 2 is a method of determining the film thickness / depth in the film thickness / depth determination unit 304 by using the plurality of databases 322 in Figure 3 through the waveform comparator 302.

[0088] The correction system for the correspondence between film thickness and spectrum in each database uses a method that calculates the time delay Δt caused by signal processing using a test signal generated by simulation. The signal processing used is the same as the signal processing performed by the digital signal processing unit 100 for the film thickness estimation object. The test signal is a sine wave, and the result of comparing the time shift of the test signal before and after signal processing is shown in Figure 10.

[0089] Compared to before signal processing (Before_Proc), it can be seen that the moment when the light intensity is at its minimum around 30-35 seconds after signal processing (After_Proc) is shifted to a later time. This shift is a time delay, from which the time delay amount Δt is calculated. The calculated time delay amount Δt is used to perform corrections for the correspondence between film thickness and spectrum in each database.

[0090] A sample was etched, and the film thickness was estimated using corrected databases based on the spectrum obtained during the etching process, as shown in Figure 11(a). It can be seen that the estimated film thickness at each moment during etching is not fixed and depends on the database used. For comparison, the film thickness estimation using uncorrected databases is shown in Figure 11(b). It can be seen that the estimated film thickness at each moment during etching varies depending on the database used. Therefore, if the uncorrected database is used, the estimated film thickness will change with the database, and film thickness estimation using multiple databases cannot be performed correctly. On the other hand, it was confirmed that by using corrected databases, the estimated film thickness is not fixed and depends on the database, and film thickness estimation using multiple databases can be performed correctly.

[0091] In the endpoint determination using the film thickness estimation obtained from the above-mentioned corrected multiple databases, since the film thickness estimation value is consistent at each time point, it can be correctly determined whether the film thickness estimation value has reached the target film thickness, thus achieving high-precision endpoint determination. [Example 2]

[0092] In this embodiment, the method for estimating film thickness / depth in Embodiment 1, which uses a learning model of spectrum and film thickness / depth created by kernel ridge regression, is used to explain the results of film thickness / depth estimation and endpoint determination. Other parts are the same as in Embodiment 1 and are omitted from this embodiment. Furthermore, from this embodiment to Embodiment 6, the creation and correction of the learning model or database can be performed by the aforementioned film thickness / depth determination unit 102 or the host computer 60, and this will not be elaborated further.

[0093] Similar to Example 1, a sample was etched, and the film thickness was estimated using the corrected databases based on the spectrum obtained during the etching process, as shown in Figure 11(a) above. There are no contradictions in the correspondence between film thickness and spectrum in each database; therefore, the estimated film thickness at each time point is constant and independent of the database. Using such a database, a learning model was created in the film thickness / depth calculation model generator 310 of Figure 2 using kernel ridge regression.

[0094] For example, by integrating the corresponding data of the spectrum and film thickness from the entire database, the spectrum is used as the explanatory variable and the film thickness as the target variable to optimize a nonlinear or linear regression model. In the optimization, for example, cross-validation is used to calculate the optimal values ​​of the parameters used in the regression model. The resulting regression model is a learning model that takes the spectrum as input and the film thickness as output.

[0095] In this embodiment, the learning model described above is used to estimate the film thickness from the spectrum during etching and to determine the etching endpoint. In the multiple databases used in creating the learning model, as mentioned above, since there is no contradiction in the correspondence between film thickness and spectrum, the film thickness output value from the spectrum obtained using the learning model is always consistent. Therefore, in the endpoint determination of the film thickness estimation using this learning model, since the estimated film thickness value is consistent at each time point, it is possible to correctly determine whether the estimated film thickness has reached the target film thickness, thus achieving high-precision endpoint determination.

[0096] Although kernel ridge regression is used to create the learning model in this embodiment, the method of creating the learning model is not limited to this. For example, regression functions created using kernel ridge regression or support vector machines, analysis / classification models created using random forests or k-nearest neighbors, and neural networks can be used. Even when using these methods, since there is no contradiction in the correspondence between membrane thickness and spectrum in the multiple corrected databases used in creating the learning model, the created learning model can output a unique membrane thickness for the input spectrum. Therefore, high-precision membrane thickness estimation and endpoint determination can be achieved without a doubt. [Example 3]

[0097] In this embodiment, the method for estimating film thickness / depth in Embodiment 1 is described, where the film thickness / depth is determined in the film thickness / depth quantity determiner 304 using a database 322 shown in FIG3 and through a waveform comparator 302. The result of estimating film thickness / depth and determining the endpoint is then described. Other than this, the method is the same as in Embodiment 1 and is omitted in this embodiment.

[0098] The database uses DB1 in Figure 9. Similar to Example 1, the time delay Δt caused by signal processing is calculated, and the correspondence between the film thickness and spectrum in the database is corrected. Ten wafers were used for film thickness estimation using this database. The entire wafer system was etched to achieve a film thickness of approximately 201 nm after etching. The formulation conditions during etching varied for each evaluation sample.

[0099] The results of estimating the film thickness for each evaluation wafer are shown in Figure 12(a). In Figure 12(a), the horizontal axis represents the wafer number and the vertical axis represents the estimated film thickness at the etching endpoint. The estimated film thickness is the same for all evaluation wafers, with a very small difference of 0.3 nm. It can be seen that by using the film thickness estimation of this embodiment, the estimated film thickness can be calculated with high accuracy without depending on the evaluation wafer. Here, as a comparative example, the results of performing film thickness estimation using the database before correction are shown in Figure 12(b). The estimated film thickness at the etching endpoint is different among the evaluation wafers, with a difference of 3.3 nm. It can be seen that the film thickness cannot be accurately estimated depending on the evaluation wafer. This confirms that even when using a single database for film thickness estimation, by correcting the correspondence between film thickness and spectrum in the database, film thickness estimation can be achieved with high accuracy without relying on the wafer being evaluated.

[0100] In the endpoint determination using the film thickness estimate from the previously corrected database, there were no discrepancies in the film thickness estimate values ​​caused by the evaluated wafer at the etching endpoint and at any point during etching. Therefore, it is possible to accurately determine whether the film thickness estimate has reached the target film thickness, achieving high-precision endpoint determination. [Example 4]

[0101] In this embodiment, it is explained that the plurality of databases in Example 1 are the results of film thickness / depth estimation and endpoint determination obtained using simulation. Other than this, the parts are the same as in Example 1 and are omitted in this embodiment.

[0102] The data representing the correspondence between film thickness / depth and spectrum in each database is generated by calculating the spectrum of each structure using optical simulations that utilize information on the structure of the evaluation sample and the structural changes caused by etching. This is obtained by performing optical simulations across multiple databases by varying the etching rate or the structural changes caused by etching. In other words, the light intensity and film thickness contained in the database before correction are obtained based on the simulation results.

[0103] By acquiring the database in this way, S11 to S14 in the database acquisition procedure of Figure 5(a) are known and can therefore be omitted, and the signal processing of S15 is performed on the acquired database. Therefore, the engineering required for database generation is reduced, and the setting of conditions is also easier. Subsequently, similar to Example 1, the database is used to correct the correspondence between film thickness / depth and spectrum based on the time delay Δt, and is then used for film thickness / depth estimation. Therefore, there is no contradiction in the correspondence between film thickness / depth and spectrum of multiple databases. When film thickness / depth estimation is performed, similar to Example 1, the estimated film thickness / depth value for a specific moment of the same evaluation object sample using each database is constant.

[0104] Therefore, in the endpoint determination using the multiple databases obtained through simulation mentioned above, there is no discrepancy in the estimated film thickness value caused by the target wafer at the etching endpoint and at each moment during etching. Therefore, it is possible to accurately determine whether the estimated film thickness has reached the target film thickness, achieving high-precision endpoint determination. [Example 5]

[0105] This embodiment describes the result of estimating the film thickness / depth and determining the endpoint when measured data is used in the calculation of the time delay Δt in Embodiment 1. Other parts are the same as in Embodiment 1 and are omitted in this embodiment.

[0106] The correction for the correspondence between film thickness and spectrum in the plurality of databases shown in Figure 9 is performed using the procedure in Figure 8. DB1 is used as the test waveform pattern database for S41. Figure 13 shows the results of film thickness estimation performed using DB1 before and after signal processing according to S42 and S43. The film thickness at each moment in the database after signal processing is expressed as Actual, and the result of estimating the film thickness of the database after signal processing using the database before signal processing is expressed as Estimated.

[0107] By using the above combination to estimate the film thickness, it can be seen that an error occurs between Actual and Estimated. The time delay Δt in S44 is calculated by measuring the time deviation between Actual and Estimated for each film thickness. In this embodiment, after etching begins, due to the influence of signal processing, an estimation error will occur in the estimated film thickness value (Estimated). Therefore, the average value of the time deviation after 20 seconds is used as the time delay Δt for calculation. The correction of the correspondence between film thickness and spectrum in multiple databases is implemented using the calculated time delay Δt.

[0108] The time delay Δt calculated in this embodiment is approximately the same as the time delay Δt calculated in Embodiment 1. Therefore, the corrected plurality of databases obtained in this embodiment can be considered equivalent to those in Embodiment 1. Therefore, in the endpoint determination of the film thickness estimation using the aforementioned measured data to calculate the time delay Δt across the plurality of databases, there is no discrepancy in the film thickness estimation value caused by the evaluation wafer at the etching endpoint and at any point during etching. Therefore, it is possible to accurately determine whether the film thickness estimation value has reached the target film thickness, achieving high-precision endpoint determination. [Embodiment 6]

[0109] In this embodiment, it is explained that the database in Embodiment 1 is the result of film thickness / depth estimation and endpoint determination, which is created using signal processing that does not cause differences in the correspondence between film thickness / depth and spectrum. Other than this, the parts are the same as in Embodiment 1 and are omitted in this embodiment.

[0110] In Embodiment 1, the signal processing performed on the database is the same as the signal processing performed on the film thickness estimation object by the digital signal processing unit 100 in FIG2 above. It is a combination of performing the low-pass filter of Equation 1, the first derivative calculation of Equation 2, the low-pass filter of Equation 1, and the spectral normalization of Equation 3 in sequence. In this embodiment, a signal processing is implemented that performs the same function as this combination of signal processing, and the correspondence between film thickness and spectrum does not change before and after signal processing.

[0111] Specifically, the low-pass filter in Equation 1 performs signal processing on the spectra of each database once in the forward and backward directions of the acquired time series. The first derivative of Equation 2 is calculated using Equation 4. As for Equation 3, it is the signal processing performed on the spectra at each time step 1. Since it does not cause any difference in the correspondence between film thickness and spectrum, it is used directly.

[0112] The results of performing the above signal processing on the light quantity data at each time of a specific wavelength in a specific database are shown in Figure 14. In Figure 14, symbol a represents before signal processing (No signal processing), symbol b represents the result of signal processing in this embodiment (Signal processing without time-delay), and symbol c represents the result of signal processing in Embodiment 1 (Signal processing with time-delay).

[0113] Since the minimum light intensity time before signal processing is consistent with that of this embodiment, it can be seen that no time delay occurs due to the signal processing of this embodiment. Even when compared with Embodiment 1 (symbol c) shown for reference, it is clear that there is no time delay due to the signal processing of this embodiment.

[0114] Several databases were created using signal processing that did not cause differences in the correspondence between film thickness and spectrum before and after signal processing. The results of film thickness estimation for each database based on the spectrum obtained during the etching of a single sample are shown in Figure 15. It can be seen that the estimated film thickness value at each moment during etching is not dependent on the database used. Next, by creating a database using signal processing that did not cause differences in the correspondence between film thickness and spectrum before and after signal processing, it can be confirmed that the estimated film thickness value is not dependent on the database and is fixed, and film thickness estimation using multiple databases can be correctly implemented.

[0115] In the endpoint determination of film thickness estimation using multiple databases corrected by the method of this embodiment, since the film thickness estimation value is consistent at each time point, it can be correctly determined whether the film thickness estimation value has reached the target film thickness, thus achieving high-precision endpoint determination. [Example 7]

[0116] In this embodiment, it is explained that in the film thickness / depth estimation method of Embodiment 1, one database is used in the waveform pattern database 122 of FIG2. The determined value of film thickness / depth is calculated in the film thickness / depth determination unit 102. The film thickness / depth correction unit 104 performs the correction of the determined value of film thickness / depth as shown in FIG16, and the film thickness / depth estimation and endpoint determination are then performed. Other than this, the parts are the same as in Embodiment 1, and the description is omitted in this embodiment.

[0117] The database uses DB2 in Figure 9. Similar to Example 1, the time delay Δt caused by signal processing is calculated, and the correspondence between the film thickness and spectrum in the database is corrected. Ten wafers were used for film thickness estimation using this database. Each wafer was etched to achieve a film thickness of approximately 200 nm. The formulation conditions and etching rates differed for each evaluation sample.

[0118] In estimating the film thickness of each evaluated wafer, a correction for the film thickness / depth value shown in FIG16 is performed. In the etching rate detection unit 1602, the etching rate ER(t) is calculated at each moment during etching using the film thickness determination value determined at that moment and at moments earlier. In this embodiment, the correlation between the moment and the film thickness determination value is approximated linearly, and the etching rate ER(t) is calculated from the slope of the obtained approximate straight line. Using the calculated etching rate ER(t) and the aforementioned time delay Δt, the film thickness determination value d(t) is corrected in the film thickness / depth corrector 1604. The correction is performed at each moment by calculation using the following Equation 5, which uses the film thickness determination value d(t), the etching rate ER(t), and the time delay Δt.

[0119]

[0120] Here, d'(t) is the corrected film thickness determination value. In the endpoint determination, the corrected film thickness determination value d'(t) is used as the film thickness estimate value and compared with the target film thickness of the endpoint determination.

[0121] The results of using these methods to estimate the film thickness of each evaluation wafer are shown in Figure 17. The horizontal axis of Figure 17 represents the wafer number, and the numbers in parentheses represent the etching rate during the etching process. The vertical axis represents the film thickness estimation error at the etching endpoint, which is the error between the estimated film thickness and the actual film thickness. The colored bars in Figure 17 represent the results of correcting the film thickness estimation value according to the present invention, while the bars filled with diagonal lines represent the results estimated using a previous method without correcting the film thickness estimation value according to the present invention. With the film thickness estimation value correction implemented according to the present invention, the film thickness estimation error is small in all wafers, demonstrating that high-precision film thickness estimation can be achieved without relying on wafer etching rate or other formulation conditions. On the other hand, in previous methods that did not implement correction for estimated film thickness, the estimated film thickness error would vary with the wafer etching rate, which means that accurate film thickness estimation could not be achieved.

[0122] Based on the above, in the film thickness estimation using the correction of the aforementioned film thickness / depth value, the estimated film thickness value and the actual film thickness value will correspond with high precision. By using the endpoint determination of the film thickness estimation value using this method, the film thickness variation between wafers after processing will be eliminated, achieving high-precision processing. [Example 8]

[0123] In this embodiment, the method for estimating film thickness / depth using the correction of the determined value of film thickness / depth in Example 7 is described, and the calculation methods for time delay Δt and etching rate ER(t) are changed, resulting in the estimation of film thickness / depth. All other parts are the same as in Example 1 and are omitted in this embodiment.

[0124] In this embodiment, the database uses DB2 in Figure 9. Similar to Embodiment 5, the measured data, namely DB2, is used to calculate the time delay Δt caused by signal processing, and the correspondence between the film thickness and the spectrum in the database is corrected. The wafer used for film thickness estimation using this database is the same as in Embodiment 7.

[0125] In estimating the film thickness of each evaluation wafer, the correction for the film thickness / depth value shown in FIG16 is performed. In the etching rate detection unit 1602, the etching rate ER(t) is calculated using time-series data of the spectrum supplied from the digital signal processing unit via a path not shown, and time-series data of the spectrum from the waveform pattern database. In this embodiment, one of the dynamic time-scaling methods, DTW (Dynamic Time Warping), is used to calculate a similar time path where the DB and the spectrum of the evaluation wafer are similar, and the etching rate ER(t) is calculated. Here, in this embodiment, although the similarity of the spectrum is quantified by cosine similarity, the error of the feature quantity calculated by the L1 norm or L2 norm of Euclidean distance can also be used, or the error of the feature quantity calculated by dimensionality compression or machine learning can be used. An example of calculating a similar time path at a time point with 100 time data points of the evaluation wafer is shown in FIG18. The horizontal and vertical axes represent the time numbers (sampling data numbers) in the spectral time series data of DB and the evaluation target wafer, respectively. In this similar time path, the straight line in the upper right corner represents the interval where similar spectra were detected. The width of this interval in DB is used as ΔNdb(t), and the width in the evaluation target wafer is used as ΔNtest(t) for calculation. The etching rate ER(t) at this time is calculated using the etching rate ERdb of DB, according to Equation 6 below.

[0126]

[0127] Using the calculated etching rate ER(t) and the aforementioned time delay Δt, the film thickness determination value d(t) is corrected in the film thickness / depth corrector 1604. The correction method is the same as in Example 7.

[0128] The error between the determined film thickness value corrected by the above method and the actual film thickness value is the same as that in Figure 17. Based on the above, in the film thickness estimation using the determined film thickness / depth value corrected by the above method, the estimated film thickness value and the actual film thickness value will correspond with high precision. By using the endpoint determination of the film thickness estimation value using this method, the film thickness variation between wafers after processing will be eliminated, achieving high-precision processing. [Example 9]

[0129] In this embodiment, the method for estimating film thickness / depth using the correction of the determined value of film thickness / depth in Example 7 is described, without correcting the correspondence between film thickness and spectrum in the database. All other parts are the same as in Example 1 and are omitted in this embodiment.

[0130] In this embodiment, the database uses DB2 in Figure 9 to calculate the time delay Δt caused by signal processing in the same way as in Embodiment 7. However, the correspondence between film thickness and spectrum in the database is not corrected before being used to estimate film thickness / depth. The wafer used for film thickness estimation using this database is the same as in Embodiment 7.

[0131] In the film thickness estimation of each evaluation wafer, the same procedure as in Example 7 was followed, and the correction of the film thickness / depth value shown in FIG16 was performed. The result of calculating the film thickness estimation error at the endpoint of each evaluation wafer is shown in FIG19. It can be seen that when the film thickness determination value correction of the present invention is implemented, the film thickness estimation error is approximately 2 nm on average, but the variation is less than 1 nm. On the other hand, when the film thickness determination value correction is not implemented, in addition to an average error of approximately 2 nm, the variation is also large. Therefore, even if the correspondence between film thickness and spectrum in the database is not corrected as in this embodiment, by implementing the film thickness determination value correction of the present invention, the film thickness estimation value and the actual film thickness will correspond with high precision. By using the endpoint determination of the corrected film thickness determination value, the film thickness variation between wafers after processing will be eliminated, and high-precision processing will be achieved.

[0132] Here, without correcting the correspondence between film thickness and spectrum in the database, a bias error occurs in the determined value of film thickness. This is because without database correction, the correspondence between film thickness and spectrum in the database is in a biased state. In contrast, by correcting the determined value of film thickness, the correspondence between film thickness and spectrum of the evaluated wafer becomes unbiased. As a result, the shift in the correspondence between film thickness and spectrum in the database is presented as a bias in the determined value of film thickness. [Example 10]

[0133] In this embodiment, it is described that in the method for estimating film thickness / depth using the determined value of film thickness / depth in Example 7, multiple databases are used to estimate the film thickness / depth. Other than this, the parts are the same as in Example 1 and are omitted in this embodiment.

[0134] In this embodiment, the five databases shown in FIG9 are used. The time delay Δt caused by signal processing is calculated using the same method as in Embodiment 7, and the correspondence between film thickness and spectrum of each database is corrected. The evaluation wafer used for film thickness estimation using this database is the same as in Embodiment 7. In this embodiment, film thickness estimation using multiple databases is implemented using the following procedure. First, multiple pattern databases and the spectral data of the evaluation wafer supplied from the digital signal processing unit are input to the waveform comparator 302 in FIG3. The waveform comparison results between each database and the spectral data of the evaluation wafer are supplied to the optimal database determiner 306. In the optimal database determiner 306, the optimal database used in the film thickness estimation of the evaluation wafer is determined, and the determination result information is supplied to the film thickness / depth amount determiner 304 to calculate the determined value of film thickness at each moment during etching.

[0135] The result calculated from the estimated film thickness error at the endpoint of each evaluated wafer is the same as that in Figure 17. Based on the above, in film thickness estimation using a complex database, by using the correction of the film thickness / depth value determined by the present invention, the estimated film thickness value and the actual film thickness value will correspond with high precision. By using the endpoint determination of the film thickness estimation value using this method, the film thickness variation between wafers after processing will be eliminated, achieving high-precision processing.

[0136] Although the invention developed by the inventor has been specifically described above based on the embodiments, the invention is not limited to the foregoing embodiments, and various modifications can be made without departing from its spirit. [Simplified Explanation of the Diagram]

[0017] [Fig. 1] A diagram showing an example of a plasma treatment apparatus. [Fig. 2] A block diagram showing an example of a film thickness / depth calculation unit. [Fig. 3] A block diagram showing an example of a film thickness / depth determination unit. [Fig. 4](a) is a table showing an example of the content of the spectrum measured at time 1, and (b) is a table showing an example of the content of a database. [Fig. 5](a) is a flowchart showing an example of a database creation procedure, and (b) is a table showing an example of the data obtained during database creation. [Fig. 6](a) is a flowchart showing an example of a correction procedure for the correspondence between film thickness / depth and spectrum in the database, and (b) is a table showing an example of the data obtained during database correction. [Fig. 7] A flowchart showing an example of a procedure for calculating the time delay Δt used in database correction using a test signal. [Fig. 8] A flowchart showing an example of a procedure for calculating the time delay Δt used in database correction using a measured signal. [Figure 9] shows an example of the contents of a plurality of databases. [Figure 10] shows an example of the result of measuring the time delay Δt using a test signal. [Figure 11] (a) shows an example of the result of film thickness estimation using each database after correction, and (b) shows an example of the result of film thickness estimation using each database before correction. [Figure 12] (a) shows an example of the result of film thickness estimation for a plurality of wafers using a corrected database, and (b) shows an example of the result of film thickness estimation for a plurality of wafers using a database before correction. [Figure 13] shows an example of the result of measuring the time delay Δt using a measured signal. [Figure 14] shows an example of the result of performing signal processing on a database using signal processing without time delay. [Figure 15] shows an example of the results of film thickness estimation using databases created through time-delay-free signal processing. [Figure 16] Block diagram of an example of film thickness / depth correction. [Figure 17] A comparison of the film thickness estimation error of the evaluation wafer with and without correction of the film thickness / depth determination value. [Figure 18] A diagram of an example of a similar time path calculated based on the similarity between the database and the spectrum of the evaluation wafer. [Figure 19] A comparison of the film thickness estimation error of the evaluation wafer with and without correction of the film thickness / depth determination value without implementation of correction of the database's film thickness-spectral correspondence.

Claims

1. A plasma processing apparatus for processing a wafer to be processed, disposed in a processing chamber inside a vacuum container, using plasma formed in the aforementioned processing chamber, comprising: The light receiver, in the process using the aforementioned plasma, receives reflected light from the wafer to be processed; the signal processor performs a first signal processing on the intensity of the previously reflected light received by the light receiver and outputs it; the database establishes a correspondence between the intensity of light at a defined wavelength and the thickness of the film formed on a reference wafer, which serves as the basis for determining the film thickness; the data processor performs a second signal processing on the intensity of the previously reflected light contained in the database, and performs data processing to maintain the correspondence between the intensity of the previously reflected light and the thickness of the previously reflected film consistent before and after the second signal processing; the determiner determines the thickness of the film on the wafer to be processed based on the intensity of the previously reflected light that has been processed by the signal processor in the previous first signal processing, using the database; and the determiner determines the endpoint of the process using the aforementioned plasma based on the thickness of the film on the wafer to be processed determined by the determiner.

2. The plasma treatment apparatus as described in claim 1, wherein, The preceding data processor calculates the time delay caused by the preceding second signal processing and corrects the correspondence between the preceding light intensity and the preceding film thickness based on the preceding time delay.

3. The plasma treatment apparatus as described in claim 2, wherein, The intensity of the previous light and the thickness of the previous film contained in the previous database before the previous correction were obtained by processing a sample wafer prepared in advance using the previous plasma.

4. The plasma treatment apparatus as described in claim 2, wherein, The intensity of the preceding light and the thickness of the preceding film contained in the preceding database before the preceding correction were obtained based on simulation results.

5. The plasma treatment apparatus as described in claim 1, wherein, The pre-processor determines the thickness of the film on the wafer to be pre-processed by comparing the intensity of the pre-processed light contained in a plurality of pre-processed databases with the intensity of the pre-processed reflected light that has undergone the first pre-processing. It then determines the thickness of the film on the wafer to be pre-processed by selecting the pre-processed light with the smallest error in the pre-processed database.

6. The plasma treatment apparatus as described in claim 1, wherein, The pre-processor determines the thickness of the film on the wafer to be pre-processed by comparing the intensity of the pre-processed light contained in a pre-processor database with the intensity of the pre-processed reflected light that has undergone the first pre-processor signal processing. The thickness of the pre-processor film corresponding to the intensity of the pre-processor light with the smaller error is used to determine the thickness of the film on the wafer to be pre-processed.

7. The plasma treatment apparatus as described in claim 1, wherein, It also features: a generator, which generates a learning model based on multiple prior databases; and a prior determiner, which determines the film thickness of the wafer to be processed based on the prior learning model.

8. A plasma processing apparatus for processing a wafer disposed in a processing chamber inside a vacuum container using plasma formed in the aforementioned processing chamber, comprising: a photodetector for receiving reflected light from the wafer at a predetermined time during the aforementioned processing; a detector for comparing a database containing the intensity of the previously reflected light at a predetermined wavelength received by the photodetector with the remaining thickness of a film formed on a reference wafer as a reference and the intensity of the previously predetermined wavelength of the previously reflected light obtained during the processing of the reference wafer, to detect the remaining thickness of the previous film during the processing of the wafer at the predetermined time; and a determination unit for determining the endpoint of the processing using the aforementioned plasma based on the thickness of the previous film on the wafer detected by the detector. The pre-reflection detector corrects for the first remaining thickness of the pre-reflection film obtained by comparing the intensity of the pre-reflected light obtained from the pre-reflection database and the pre-reflection processing, based on the difference between the intensity of the pre-reflected light and the thickness of the pre-reflection film before and after the signal processing performed on the intensity of the pre-reflected light, in order to detect the second remaining thickness of the pre-reflection film.

9. The plasma treatment apparatus as described in claim 8, wherein, The preceding detector is used to correct the remaining thickness of the preceding signal based on the calculated time delay caused by the preceding signal processing.

10. The plasma processing apparatus as described in claim 9, wherein, The preceding processing is an etching process; the preceding detector uses the product of the etching rate of the wafer being processed in the preceding processing and the preceding time delay in the preceding processing before the time specified in the preceding to correct the first remaining thickness of the preceding.

11. The plasma processing apparatus as described in claim 8 or 9, wherein, The intensity of the reflected light, the remaining thickness of the film, and their correspondence contained in the previous database are obtained based on the results of previous processing using the previous plasma on the previous reference wafer.

12. The plasma processing apparatus as described in claim 8 or 9, wherein, The intensity of the reflected light and the thickness of the film contained in the pre-correction pre-correlation database were obtained based on simulation results.

13. A plasma processing method, which is a plasma processing apparatus performed to process a wafer to be processed, disposed in a processing chamber inside a vacuum container, using plasma formed in the aforementioned processing chamber, comprising: The signal processing process involves performing first signal processing and outputting the intensity of reflected light from the wafer to be processed, obtained during the previous plasma processing; the determination process involves establishing a database based on the intensity of light at a predetermined wavelength and the thickness of the film formed on a reference wafer, which serves as the basis for film thickness determination, and determining the film thickness of the wafer to be processed from the intensity of the previous reflected light that underwent the first signal processing in the previous signal processing process; the judgment process involves determining the endpoint of the previous plasma processing based on the film thickness of the wafer to be processed determined in the previous determination process; and the data processing process involves performing second signal processing on the intensity of the previous light contained in the previous database, and performing data processing to maintain the correspondence between the intensity of the previous light and the thickness of the previous film consistent before and after the second signal processing.

14. An etching system comprising an etching processing device and a data processing device, wherein, The aforementioned etching processing apparatus includes: a light receiver that receives reflected light from the wafer to be processed during plasma processing; a signal processor that performs first signal processing on the intensity of the reflected light received by the light receiver and outputs the signal; a database that establishes a correspondence between the intensity of light at a predetermined wavelength and the thickness of the film formed on a reference wafer, which serves as a reference for film thickness determination; a determiner that determines the film thickness of the wafer to be processed based on the intensity of the reflected light that has undergone first signal processing by the signal processor, using the aforementioned database; and a determiner that determines the endpoint of the etching process based on the film thickness of the wafer to be processed determined by the aforementioned determiner; and a data processing unit that acquires the intensity of light at a predetermined wavelength and the thickness of the film formed on the aforementioned reference wafer to generate the aforementioned database. The preceding data processing device performs second signal processing on the intensity of the preceding light when the preceding data database is generated, and performs data processing to maintain the correspondence between the intensity of the preceding light and the thickness of the preceding film consistent before and after the second signal processing.

Images